nihilist/tor
1
0
Fork 0
mirror of https://gitlab.torproject.org/tpo/core/tor.git synced 2024-11-11 13:43:47 +01:00
Code Issues Packages Projects Releases Wiki Activity
d9ceab5bc3
tor/src/or/config.c

7114 lines
236 KiB
C
Raw Normal View History

Happy new year! svn:r5949
2006-02-09 06:46:49 +01:00
/* Copyright (c) 2001 Matej Pfajfar.
* Copyright (c) 2001-2004, Roger Dingledine.
clean up copyrights, and assign 2007 copyrights to The Tor Project, Inc svn:r12786
2007-12-12 22:09:01 +01:00
* Copyright (c) 2004-2006, Roger Dingledine, Nick Mathewson.
Bump copyright statements to 2011
2011-01-03 17:50:39 +01:00
* Copyright (c) 2007-2011, The Tor Project, Inc. */
Implemented link padding and receiver token buckets Each socket reads at most 'bandwidth' bytes per second sustained, but can handle bursts of up to 10*bandwidth bytes. Cells are now sent out at evenly-spaced intervals, with padding sent out otherwise. Set Linkpadding=0 in the rc file to send cells as soon as they're available (and to never send padding cells). Added license/copyrights statements at the top of most files. router->min and router->max have been merged into a single 'bandwidth' value. We should make the routerinfo_t reflect this (want to do that, Mat?) As the bandwidth increases, and we want to stop sleeping more and more frequently to send a single cell, cpu usage goes up. At 128kB/s we're pretty much calling poll with a timeout of 1ms or even 0ms. The current code takes a timeout of 0-9ms and makes it 10ms. prepare_for_poll() handles everything that should have happened in the past, so as long as our buffers don't get too full in that 10ms, we're ok. Speaking of too full, if you run three servers at 100kB/s with -l debug, it spends too much time printing debugging messages to be able to keep up with the cells. The outbuf ultimately fills up and it kills that connection. If you run with -l err, it works fine up through 500kB/s and probably beyond. Down the road we'll want to teach it to recognize when an outbuf is getting full, and back off. svn:r50
2002-07-16 03:12:15 +02:00
/* See LICENSE for licensing information */
flesh out the source file descriptions for doxygen svn:r4404
2005-06-11 07:31:17 +02:00
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
/**
doxygen uses backslashes, not forward-slashes svn:r3033
2004-11-30 07:17:35 +01:00
* \file config.c
* \brief Code to parse and interpret configuration files.
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
**/
r12349@catbus: nickm | 2007-04-11 09:18:15 -0400 Add code to shrink the cell memory pool by discarding empty chunks that have been empty for the last 60 seconds. Also, instead of having test.c duplicate declarations for exposed functions, put them inside #ifdef foo_PRIVATE blocks in the headers. This prevents bugs where test.c gets out of sync. svn:r9944
2007-04-11 15:18:25 +02:00
#define CONFIG_PRIVATE
Initial revision svn:r2
2002-06-27 00:45:49 +02:00
#include "or.h"
Create circuitbuild.h
2010-07-22 01:21:00 +02:00
#include "circuitbuild.h"
Create circuitlist.h
2010-07-22 09:46:23 +02:00
#include "circuitlist.h"
Create config.h
2010-07-22 10:22:51 +02:00
#include "config.h"
Create connection.h
2010-07-22 10:32:52 +02:00
#include "connection.h"
Create connection_edge.h
2010-07-22 10:43:02 +02:00
#include "connection_edge.h"
reinit per-conn token buckets on config or consensus change
2010-08-15 10:01:42 +02:00
#include "connection_or.h"
Create control.h
2010-07-22 11:35:09 +02:00
#include "control.h"
Create cpuworker.h
2010-07-22 11:40:39 +02:00
#include "cpuworker.h"
Create dirserv.h
2010-07-22 12:09:49 +02:00
#include "dirserv.h"
Create dirvote.h
2010-07-22 12:19:28 +02:00
#include "dirvote.h"
Create dns.h
2010-07-22 12:24:25 +02:00
#include "dns.h"
Create geoip.h
2010-07-21 14:38:52 +02:00
#include "geoip.h"
Create hibernate.h
2010-07-22 12:30:46 +02:00
#include "hibernate.h"
Create main.h
2010-07-23 19:58:06 +02:00
#include "main.h"
Create networkstatus.h
2010-07-23 20:18:55 +02:00
#include "networkstatus.h"
Create policies.h
2010-07-23 20:51:25 +02:00
#include "policies.h"
Create relay.h
2010-07-23 21:53:11 +02:00
#include "relay.h"
Create rendclient.h
2010-07-22 00:13:51 +02:00
#include "rendclient.h"
Create rendservice.h
2010-07-22 00:30:17 +02:00
#include "rendservice.h"
Create rephist.h
2010-07-23 22:57:20 +02:00
#include "rephist.h"
Create router.h
2010-07-21 16:17:10 +02:00
#include "router.h"
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
#include "util.h"
Create routerlist.h
2010-07-21 17:08:11 +02:00
#include "routerlist.h"
Updated #includes etc. to use transports.[ch].
2011-07-18 02:33:31 +02:00
#include "transports.h"
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifdef _WIN32
On windows, default datadir is APPDATA/tor, and default config file is APPDATA/tor/torrc. (APPDATA is usually somedrive:/Documents and Settings/.../Application Data/) svn:r2265
2004-08-18 05:42:55 +02:00
#include <shlobj.h>
#endif
cell.c is now obsolete svn:r133
2002-10-03 04:17:41 +02:00
Tweak disable_debugger_attachment a little Don't warn when we have no implementation of this function (since it's on-by-default); reformat the changes entry; fix an overlong line.
2011-11-25 05:39:44 +01:00
#include "procmon.h"
/* From main.c */
extern int quiet_level;
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
/** Enumeration of types which option values can take */
typedef enum config_type_t {
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
CONFIG_TYPE_STRING = 0, /**< An arbitrary string. */
Clean up a macro/cpp interaction related to bug 707. Also, add a new "filename" config value type which is currently just a synonym for string, but which might be subject to expansion later. svn:r15305
2008-06-16 20:09:53 +02:00
CONFIG_TYPE_FILENAME, /**< A filename: some prefixes get expanded. */
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
CONFIG_TYPE_UINT, /**< A non-negative integer less than MAX_INT */
"(Socks|Control|etc)Port auto" now tells Tor to open an arbitrary port This is the major part of the implementation for trac issue 3076.
2011-05-02 21:05:10 +02:00
CONFIG_TYPE_PORT, /**< A port from 1...65535, 0 for "not set", or
* "auto". */
Implement weekly/monthly/daily accounting svn:r2936
2004-11-22 22:56:51 +01:00
CONFIG_TYPE_INTERVAL, /**< A number of seconds, with optional units*/
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
CONFIG_TYPE_MSEC_INTERVAL,/**< A number of milliseconds, with optional
* units */
Implement weekly/monthly/daily accounting svn:r2936
2004-11-22 22:56:51 +01:00
CONFIG_TYPE_MEMUNIT, /**< A number of bytes, with optional units*/
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
CONFIG_TYPE_DOUBLE, /**< A floating-point value */
CONFIG_TYPE_BOOL, /**< A boolean value, expressed as 0 or 1. */
Add a new configuration type, "AUTOBOOL", to handle 1/0/auto types We only have one of these now, but I'm about to add a few more. Yes, I have already thought of the "Filenotfoundian logic" joke.
2010-11-08 19:34:40 +01:00
CONFIG_TYPE_AUTOBOOL, /**< A boolean+auto value, expressed 0 for false,
* 1 for true, and -1 for auto */
Spell-check Tor.
2009-05-27 23:55:51 +02:00
CONFIG_TYPE_ISOTIME, /**< An ISO-formatted time relative to GMT. */
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
CONFIG_TYPE_CSV, /**< A list of strings, separated by commas and
* optional whitespace. */
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
CONFIG_TYPE_LINELIST, /**< Uninterpreted config lines */
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
CONFIG_TYPE_LINELIST_S, /**< Uninterpreted, context-sensitive config lines,
* mixed with other keywords. */
CONFIG_TYPE_LINELIST_V, /**< Catch-all "virtual" option to summarize
* context-sensitive config lines when fetching.
*/
r17188@tombo: nickm | 2008-07-18 14:35:18 -0400 Add new ExcludeExitNodes option. Also add a new routerset type to handle Exclude[Exit]Nodes. It is optimized for O(1) membership tests, so as to make choosing a random router run in O(N_routers) time instead of in O(N_routers*N_Excluded_Routers). svn:r16061
2008-07-18 20:36:32 +02:00
CONFIG_TYPE_ROUTERSET, /**< A list of router names, addrs, and fps,
* parsed into a routerset_t. */
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
CONFIG_TYPE_OBSOLETE, /**< Obsolete (ignored) option. */
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
} config_type_t;
clean up config.c so it doesn't expose as much svn:r430
2003-09-08 07:16:18 +02:00
without braces, what will hold up the code? svn:r5209
2005-10-06 07:08:00 +02:00
/** An abbreviation for a configuration option allowed on the command line. */
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
typedef struct config_abbrev_t {
Tricksy compiler warnings! We hates them, hates them forever, my precious! svn:r2615
2004-10-27 23:14:11 +02:00
const char *abbreviated;
const char *full;
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
int commandline_only;
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
int warn;
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
} config_abbrev_t;
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
/* Handy macro for declaring "In the config file or on the command line,
* you can abbreviate <b>tok</b>s as <b>tok</b>". */
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
#define PLURAL(tok) { #tok, #tok "s", 0, 0 }
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
Document most undocumented variables. svn:r17754
2008-12-23 18:56:31 +01:00
/** A list of abbreviations and aliases to map command-line options, obsolete
* option names, or alternative option names, to their current values. */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
static config_abbrev_t _option_abbrevs[] = {
Allow authorities to baddir/badexit/invalid/reject nodes by cc Implements ticket #4207
2012-01-13 18:28:32 +01:00
PLURAL(AuthDirBadDirCC),
PLURAL(AuthDirBadExitCC),
PLURAL(AuthDirInvalidCC),
PLURAL(AuthDirRejectCC),
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
PLURAL(ExitNode),
EntryNodess is not the plural of EntryNodes svn:r2728
2004-11-09 07:09:06 +01:00
PLURAL(EntryNode),
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
PLURAL(ExcludeNode),
PLURAL(FirewallPort),
Annotate circuits w/ whether they aim to contain high uptime nodes and/or high capacity nodes. When building circuits, choose appropriate nodes. New config option LongLivedPorts to indicate application streams that will want high uptime circuits. When attaching a stream to a circuit, pay attention to its requirements. This means that every single node in an intro rend circuit, not just the last one, will have a minimum uptime. Boost the min uptime from an hour to 24 hours. svn:r3339
2005-01-12 05:58:23 +01:00
PLURAL(LongLivedPort),
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
PLURAL(HiddenServiceNode),
PLURAL(HiddenServiceExcludeNode),
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
PLURAL(NumCPU),
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
PLURAL(RendNode),
PLURAL(RendExcludeNode),
allow StrictEntryNode and StrictExitNode to be singular svn:r2922
2004-11-21 00:16:03 +01:00
PLURAL(StrictEntryNode),
PLURAL(StrictExitNode),
Switch to a StrictNodes config option. This is step one of handling ExcludedNodes better. This first step is just to make EntryNodes and ExitNodes do what they did before.
2009-09-18 04:45:54 +02:00
PLURAL(StrictNode),
Add a missing struct initializer field svn:r5264
2005-10-18 01:00:08 +02:00
{ "l", "Log", 1, 0},
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
{ "AllowUnverifiedNodes", "AllowInvalidNodes", 0, 0},
r12946@catbus: nickm | 2007-05-25 10:46:50 -0400 Add an AutomapHostsOnResolve option. It seems to work. svn:r10324
2007-05-25 16:48:16 +02:00
{ "AutomapHostSuffixes", "AutomapHostsSuffixes", 0, 0},
{ "AutomapHostOnResolve", "AutomapHostsOnResolve", 0, 0},
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
{ "BandwidthRateBytes", "BandwidthRate", 0, 0},
{ "BandwidthBurstBytes", "BandwidthBurst", 0, 0},
{ "DirFetchPostPeriod", "StatusFetchPeriod", 0, 0},
{ "MaxConn", "ConnLimit", 0, 1},
stop discouraging the old *BindAddress config options. we will just confuse our users when they upgrade. svn:r5354
2005-11-11 18:04:14 +01:00
{ "ORBindAddress", "ORListenAddress", 0, 0},
{ "DirBindAddress", "DirListenAddress", 0, 0},
{ "SocksBindAddress", "SocksListenAddress", 0, 0},
entry nodes are now entry guards. this is our last easy chance for a wholesale change. heave ho. svn:r5782
2006-01-10 23:42:44 +01:00
{ "UseHelperNodes", "UseEntryGuards", 0, 0},
{ "NumHelperNodes", "NumEntryGuards", 0, 0},
{ "UseEntryNodes", "UseEntryGuards", 0, 0},
{ "NumEntryNodes", "NumEntryGuards", 0, 0},
r8874@Kushana: nickm | 2006-09-21 15:22:27 -0400 Rename and document SearchDomains and ResolvConf options; warn if ServerDNSResolvConfFile is given but eventdns isnt enabled. svn:r8439
2006-09-21 23:48:16 +02:00
{ "ResolvConf", "ServerDNSResolvConfFile", 0, 1},
{ "SearchDomains", "ServerDNSSearchDomains", 0, 1},
Switch to a StrictNodes config option. This is step one of handling ExcludedNodes better. This first step is just to make EntryNodes and ExitNodes do what they did before.
2009-09-18 04:45:54 +02:00
{ "ServerDNSAllowBrokenResolvConf", "ServerDNSAllowBrokenConfig", 0, 0},
Allow people to say PreferTunnelledDirConns rather than PreferTunneledDirConns, for those alternate-spellers out there. svn:r12466
2007-11-11 02:06:16 +01:00
{ "PreferTunnelledDirConns", "PreferTunneledDirConns", 0, 0},
Make getinfo ns/purpose/bridge actually work Also, dump our bridge router status entries to disk every 30 minutes. svn:r12871
2007-12-19 05:58:58 +01:00
{ "BridgeAuthoritativeDirectory", "BridgeAuthoritativeDir", 0, 0},
Make "HashedControlPassword" an alias for "__HashedControlSessionPassword" if it appears on the commandline. should help with bug 586. svn:r13683
2008-02-23 00:20:28 +01:00
{ "HashedControlPassword", "__HashedControlSessionPassword", 1, 0},
Switch to a StrictNodes config option. This is step one of handling ExcludedNodes better. This first step is just to make EntryNodes and ExitNodes do what they did before.
2009-09-18 04:45:54 +02:00
{ "StrictEntryNodes", "StrictNodes", 0, 1},
{ "StrictExitNodes", "StrictNodes", 0, 1},
helper nodes are dead. long live entry nodes. (config options EntryNodes and StrictEntryNodes still not implemented.) svn:r5673
2005-12-28 10:07:31 +01:00
{ NULL, NULL, 0, 0},
};
Document most undocumented variables. svn:r17754
2008-12-23 18:56:31 +01:00
/** A list of state-file "abbreviations," for compatibility. */
helper nodes are dead. long live entry nodes. (config options EntryNodes and StrictEntryNodes still not implemented.) svn:r5673
2005-12-28 10:07:31 +01:00
static config_abbrev_t _state_abbrevs[] = {
Add descriptions for state variables; rename AccountingBytesRead{In}Interval; remove extraneous and unused struct member. svn:r5739
2006-01-06 16:43:03 +01:00
{ "AccountingBytesReadInterval", "AccountingBytesReadInInterval", 0, 0 },
entry nodes are now entry guards. this is our last easy chance for a wholesale change. heave ho. svn:r5782
2006-01-10 23:42:44 +01:00
{ "HelperNode", "EntryGuard", 0, 0 },
{ "HelperNodeDownSince", "EntryGuardDownSince", 0, 0 },
{ "HelperNodeUnlistedSince", "EntryGuardUnlistedSince", 0, 0 },
{ "EntryNode", "EntryGuard", 0, 0 },
{ "EntryNodeDownSince", "EntryGuardDownSince", 0, 0 },
{ "EntryNodeUnlistedSince", "EntryGuardUnlistedSince", 0, 0 },
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
{ NULL, NULL, 0, 0},
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
};
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
#undef PLURAL
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
lay the groundwork for a default value for each config option. tolerate null exitnodes, entrynodes, etc config options. svn:r2655
2004-11-03 11:08:44 +01:00
/** A variable allowed in the configuration file or on the command line. */
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
typedef struct config_var_t {
lay the groundwork for a default value for each config option. tolerate null exitnodes, entrynodes, etc config options. svn:r2655
2004-11-03 11:08:44 +01:00
const char *name; /**< The full keyword (case insensitive). */
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
config_type_t type; /**< How to interpret the type and turn it into a
* value. */
lay the groundwork for a default value for each config option. tolerate null exitnodes, entrynodes, etc config options. svn:r2655
2004-11-03 11:08:44 +01:00
off_t var_offset; /**< Offset of the corresponding member of or_options_t. */
const char *initvalue; /**< String (or null) describing initial value. */
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
} config_var_t;
Document new config code svn:r2596
2004-10-27 05:08:04 +02:00
/** An entry for config_vars: "The option <b>name</b> has type
* CONFIG_TYPE_<b>conftype</b>, and corresponds to
* or_options_t.<b>member</b>"
*/
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
#define VAR(name,conftype,member,initvalue) \
{ name, CONFIG_TYPE_ ## conftype, STRUCT_OFFSET(or_options_t, member), \
Add descriptions for state variables; rename AccountingBytesRead{In}Interval; remove extraneous and unused struct member. svn:r5739
2006-01-06 16:43:03 +01:00
initvalue }
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
/** As VAR, but the option name and member name are the same. */
#define V(member,conftype,initvalue) \
VAR(#member, conftype, member, initvalue)
Document new config code svn:r2596
2004-10-27 05:08:04 +02:00
/** An entry for config_vars: "The option <b>name</b> is obsolete." */
Add descriptions for state variables; rename AccountingBytesRead{In}Interval; remove extraneous and unused struct member. svn:r5739
2006-01-06 16:43:03 +01:00
#define OBSOLETE(name) { name, CONFIG_TYPE_OBSOLETE, 0, NULL }
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
Document new config code svn:r2596
2004-10-27 05:08:04 +02:00
/** Array of configuration options. Until we disallow nonstandard
* abbreviations, order is significant, since the first matching option will
* be chosen first.
*/
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
static config_var_t _option_vars[] = {
r11658@Kushana: nickm | 2006-12-20 15:58:44 -0500 Remove long-deprecated log and accounting options. svn:r9164
2006-12-20 22:02:10 +01:00
OBSOLETE("AccountingMaxKB"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(AccountingMax, MEMUNIT, "0 bytes"),
V(AccountingStart, STRING, NULL),
V(Address, STRING, NULL),
Disable .exit notation unless AllowDotExit is 1.
2009-08-08 01:26:41 +02:00
V(AllowDotExit, BOOL, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(AllowInvalidNodes, CSV, "middle,rendezvous"),
V(AllowNonRFC953Hostnames, BOOL, "0"),
Proposal 152 implementation from Josh Albrecht, with tweaks. svn:r16983
2008-09-26 20:58:45 +02:00
V(AllowSingleHopCircuits, BOOL, "0"),
V(AllowSingleHopExits, BOOL, "0"),
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
V(AlternateBridgeAuthority, LINELIST, NULL),
V(AlternateDirAuthority, LINELIST, NULL),
V(AlternateHSAuthority, LINELIST, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(AssumeReachable, BOOL, "0"),
New config options AuthDirBadDir and AuthDirListBadDirs for authorities to mark certain relays as "bad directories" in the networkstatus documents. Also supports the "!baddir" directive in the approved-routers file. svn:r12754
2007-12-10 17:49:54 +01:00
V(AuthDirBadDir, LINELIST, NULL),
fix crash bug in original feature4207 branch PLURAL() assumes that the plural is the canonical name for the option, so now it is.
2012-01-17 03:20:46 +01:00
V(AuthDirBadDirCCs, CSV, ""),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(AuthDirBadExit, LINELIST, NULL),
fix crash bug in original feature4207 branch PLURAL() assumes that the plural is the canonical name for the option, so now it is.
2012-01-17 03:20:46 +01:00
V(AuthDirBadExitCCs, CSV, ""),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(AuthDirInvalid, LINELIST, NULL),
fix crash bug in original feature4207 branch PLURAL() assumes that the plural is the canonical name for the option, so now it is.
2012-01-17 03:20:46 +01:00
V(AuthDirInvalidCCs, CSV, ""),
raise AuthDirFastGuarantee from 20KB to 100KB This patch reverts part of 39ceda7e05d7 (where it used to be 100KB).
2011-11-17 04:08:01 +01:00
V(AuthDirFastGuarantee, MEMUNIT, "100 KB"),
parameterize bw cutoffs to guarantee Fast and Guard flags Now it will be easier for researchers to simulate Tor networks with different values. Resolves ticket 4484.
2011-11-17 02:55:33 +01:00
V(AuthDirGuardBWGuarantee, MEMUNIT, "250 KB"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(AuthDirReject, LINELIST, NULL),
fix crash bug in original feature4207 branch PLURAL() assumes that the plural is the canonical name for the option, so now it is.
2012-01-17 03:20:46 +01:00
V(AuthDirRejectCCs, CSV, ""),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(AuthDirRejectUnlisted, BOOL, "0"),
New config options AuthDirBadDir and AuthDirListBadDirs for authorities to mark certain relays as "bad directories" in the networkstatus documents. Also supports the "!baddir" directive in the approved-routers file. svn:r12754
2007-12-10 17:49:54 +01:00
V(AuthDirListBadDirs, BOOL, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(AuthDirListBadExits, BOOL, "0"),
r17548@catbus: nickm | 2008-01-10 11:08:12 -0500 Make proposal-109 behavior optional. svn:r13090
2008-01-10 17:08:47 +01:00
V(AuthDirMaxServersPerAddr, UINT, "2"),
V(AuthDirMaxServersPerAuthAddr,UINT, "5"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
VAR("AuthoritativeDirectory", BOOL, AuthoritativeDir, "0"),
V(AutomapHostsOnResolve, BOOL, "0"),
V(AutomapHostsSuffixes, CSV, ".onion,.exit"),
V(AvoidDiskWrites, BOOL, "0"),
Raise the default BandwidthRate/BandwidthBurst to 5MB/10MB, to accommodate the growing number of servers that use the default and are reaching it. svn:r12364
2007-11-03 23:08:52 +01:00
V(BandwidthBurst, MEMUNIT, "10 MB"),
V(BandwidthRate, MEMUNIT, "5 MB"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(BridgeAuthoritativeDir, BOOL, "0"),
VAR("Bridge", LINELIST, Bridges, NULL),
start working on serving bridge status from the dirport, for bridge communities. svn:r12927
2007-12-22 11:54:21 +01:00
V(BridgePassword, STRING, NULL),
r15530@tombo: nickm | 2007-12-17 16:54:03 -0500 First wodge of geoip code so bridges can figure out which countries are blocking them. svn:r12845
2007-12-17 23:44:11 +01:00
V(BridgeRecordUsageByCountry, BOOL, "1"),
Stop being so aggressive about fetching dir info if your DirPort is on but your ORPort is off. Add a new config option BridgeRelay that specifies you want to be a bridge relay. Right now the only difference is that it makes you answer begin_dir requests, and it makes you cache dir info, even if your DirPort isn't on. Refactor directory_caches_dir_info() into some more functions. svn:r12668
2007-12-04 19:35:03 +01:00
V(BridgeRelay, BOOL, "0"),
If configured, write cell statistics to disk periodically.
2009-07-05 19:53:25 +02:00
V(CellStatistics, BOOL, "0"),
Bug 1296: Add option+logic to disable CBT learning. There are now four ways that CBT can be disabled: 1. Network-wide, with the cbtdisabled consensus param. 2. Via config, with "LearnCircuitBuildTimeout 0" 3. Via config, with "AuthoritativeDirectory 1" 4. Via a state file write failure.
2010-05-08 20:54:29 +02:00
V(LearnCircuitBuildTimeout, BOOL, "1"),
Use our variable directly for timeout. Using CircuitBuildTimeout is prone to issues with SIGHUP, etc. Also, shuffle the circuit build times array after loading it in so that newer measurements don't replace chunks of similarly timed measurements.
2009-09-02 05:27:43 +02:00
V(CircuitBuildTimeout, INTERVAL, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(CircuitIdleTimeout, INTERVAL, "1 hour"),
New config option "CircuitStreamTimeout" New config option "CircuitStreamTimeout" to override our internal timeout schedule for how many seconds until we detach a stream from a circuit and try a new circuit. If your network is particularly slow, you might want to set this to a number like 60.
2009-11-22 05:36:36 +01:00
V(CircuitStreamTimeout, INTERVAL, "0"),
Change interface for configuring cell ewma algorithm. The rule is now: take the value from the CircuitPriorityHalflife config option if it is set. If it zero, disable the cell_ewma algorithm. If it is set, use it to calculate the scaling factor. If it is not set, look for a CircPriorityHalflifeMsec parameter in the consensus networkstatus. If *that* is zero, then disable the cell_ewma algorithm; if it is set, use it to calculate the scaling factor. If it is not set at all, disable the algorithm.
2009-12-15 19:53:53 +01:00
V(CircuitPriorityHalflife, DOUBLE, "-100.0"), /*negative:'Use default'*/
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(ClientDNSRejectInternalAddresses, BOOL,"1"),
V(ClientOnly, BOOL, "0"),
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
V(ClientRejectInternalAddresses, BOOL, "1"),
Style and grammar tweaks on 2841 branch
2011-07-03 05:23:07 +02:00
V(ClientTransportPlugin, LINELIST, NULL),
ConsensusParams config option lists key=value params finishes the authority-operator interface side of proposal 167.
2009-09-15 10:40:08 +02:00
V(ConsensusParams, STRING, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(ConnLimit, UINT, "1000"),
Call the new config option ConnDirectionStatistics.
2010-08-15 14:58:35 +02:00
V(ConnDirectionStatistics, BOOL, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(ConstrainedSockets, BOOL, "0"),
V(ConstrainedSockSize, MEMUNIT, "8192"),
V(ContactInfo, STRING, NULL),
V(ControlListenAddress, LINELIST, NULL),
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
V(ControlPort, LINELIST, NULL),
Add a ControlPortFileGroupWritable option
2011-05-13 01:17:48 +02:00
V(ControlPortFileGroupReadable,BOOL, "0"),
Write automatically-chosen control ports to a file.
2011-05-09 18:13:37 +02:00
V(ControlPortWriteToFile, FILENAME, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(ControlSocket, LINELIST, NULL),
minor cleanups while reviewing 3216
2011-05-30 21:31:06 +02:00
V(ControlSocketsGroupWritable, BOOL, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(CookieAuthentication, BOOL, "0"),
V(CookieAuthFileGroupReadable, BOOL, "0"),
V(CookieAuthFile, STRING, NULL),
Patch from cagara: Add a CountPrivateBandwidth flag
2011-04-05 20:50:32 +02:00
V(CountPrivateBandwidth, BOOL, "0"),
Clean up a macro/cpp interaction related to bug 707. Also, add a new "filename" config value type which is currently just a synonym for string, but which might be subject to expansion later. svn:r15305
2008-06-16 20:09:53 +02:00
V(DataDirectory, FILENAME, NULL),
r11658@Kushana: nickm | 2006-12-20 15:58:44 -0500 Remove long-deprecated log and accounting options. svn:r9164
2006-12-20 22:02:10 +01:00
OBSOLETE("DebugLogFile"),
New 'DisableNetwork' option to prevent Tor from using the network Some controllers want this so they can mess with Tor's configuration for a while via the control port before actually letting Tor out of the house. We do this with a new DisableNetwork option, that prevents Tor from making any outbound connections or binding any non-control listeners. Additionally, it shuts down the same functionality as shuts down when we are hibernating, plus the code that launches directory downloads. To make sure I didn't miss anything, I added a clause straight to connection_connect, so that we won't even try to open an outbound socket when the network is disabled. In my testing, I made this an assert, but since I probably missed something, I've turned it into a BUG warning for testing.
2011-11-28 21:44:10 +01:00
V(DisableNetwork, BOOL, "0"),
Give DirAllowPrivateAddress an explicit default By convention, we say whether each bool's default is 0 or 1 Fixes 4536; found by "troll_un"
2011-12-02 22:04:18 +01:00
V(DirAllowPrivateAddresses, BOOL, "0"),
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
V(TestingAuthDirTimeToLearnReachability, INTERVAL, "30 minutes"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(DirListenAddress, LINELIST, NULL),
make DirFetchPeriod and StatusFetchPeriod truly obsolete. svn:r6373
2006-04-10 22:16:46 +02:00
OBSOLETE("DirFetchPeriod"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(DirPolicy, LINELIST, NULL),
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
V(DirPort, LINELIST, NULL),
Mark DirPortFrontPage as a FILENAME rather than a STRING. Right now this has no effect. svn:r17520
2008-12-08 19:00:34 +01:00
V(DirPortFrontPage, FILENAME, NULL),
Make DirPostPeriod config option obsolete. Force new descriptor upload every 18 hours, not 24, to avoid races. Change rate-limiting on new descriptors due to bandwidth changes, from 45 minutes to 20 minutes. svn:r4814
2005-08-22 23:53:12 +02:00
OBSOLETE("DirPostPeriod"),
Change the way how directories that are configured with --enable-geoip-stats write geoip stats to disk. - Write geoip stats to disk every 24 hours, not every hour. - Remove configuration options and define reasonable defaults. - Clear history of client requests every 24 hours (which wasn't done at all before).
2009-05-27 17:52:46 +02:00
OBSOLETE("DirRecordUsageByCountry"),
OBSOLETE("DirRecordUsageGranularity"),
OBSOLETE("DirRecordUsageRetainIPs"),
OBSOLETE("DirRecordUsageSaveInterval"),
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
V(DirReqStatistics, BOOL, "1"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
VAR("DirServer", LINELIST, DirServers, NULL),
Implement DisableAllSwap to avoid putting secret info in page files. This commit implements a new config option: 'DisableAllSwap' This option probably only works properly when Tor is started as root. We added two new functions: tor_mlockall() and tor_set_max_memlock(). tor_mlockall() attempts to mlock() all current and all future memory pages. For tor_mlockall() to work properly we set the process rlimits for memory to RLIM_INFINITY (and beyond) inside of tor_set_max_memlock(). We behave differently from mlockall() by only allowing tor_mlockall() to be called one single time. All other calls will result in a return code of 1. It is not possible to change DisableAllSwap while running. A sample configuration item was added to the torrc.complete.in config file. A new item in the man page for DisableAllSwap was added. Thanks to Moxie Marlinspike and Chris Palmer for their feedback on this patch. Please note that we make no guarantees about the quality of your OS and its mlock/mlockall implementation. It is possible that this will do nothing at all. It is also possible that you can ulimit the mlock properties of a given user such that root is not required. This has not been extensively tested and is unsupported. I have included some comments for possible ways we can handle this on win32.
2009-10-22 06:21:57 +02:00
V(DisableAllSwap, BOOL, "0"),
New configuration option: DisableDebuggerAttachment If set to 1, Tor will attempt to prevent basic debugging attachment attempts by other processes. (Default: 1) Supports Mac OS X and Gnu/Linux. Sebastian provided useful feedback and refactoring suggestions. Signed-off-by: Jacob Appelbaum <jacob@appelbaum.net>
2011-05-30 17:06:51 +02:00
V(DisableDebuggerAttachment, BOOL, "1"),
Add a new option to enable/disable IOCP support
2010-09-28 20:01:45 +02:00
V(DisableIOCP, BOOL, "1"),
s/DynamicPrimes/DynamicDHGroups/g
2011-11-25 01:00:14 +01:00
V(DynamicDHGroups, BOOL, "1"),
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
V(DNSPort, LINELIST, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(DNSListenAddress, LINELIST, NULL),
V(DownloadExtraInfo, BOOL, "0"),
V(EnforceDistinctSubnets, BOOL, "1"),
Add country-code support to configured node lists to implement the ever-popular "no exits in Monaco" feature (ExcludeExitNodes {MC}). Also allow country codes and IP ranges in ExitNodes. (EntryNodes needs more work.) Based on code by Robert Hogan. Needs more testing. svn:r16966
2008-09-25 22:21:35 +02:00
V(EntryNodes, ROUTERSET, NULL),
If configured, write entry-node statistics to disk periodically.
2009-07-05 20:48:16 +02:00
V(EntryStatistics, BOOL, "0"),
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
V(TestingEstimatedDescriptorPropagationTime, INTERVAL, "10 minutes"),
r17188@tombo: nickm | 2008-07-18 14:35:18 -0400 Add new ExcludeExitNodes option. Also add a new routerset type to handle Exclude[Exit]Nodes. It is optimized for O(1) membership tests, so as to make choosing a random router run in O(N_routers) time instead of in O(N_routers*N_Excluded_Routers). svn:r16061
2008-07-18 20:36:32 +02:00
V(ExcludeNodes, ROUTERSET, NULL),
V(ExcludeExitNodes, ROUTERSET, NULL),
Proposal 152 implementation from Josh Albrecht, with tweaks. svn:r16983
2008-09-26 20:58:45 +02:00
V(ExcludeSingleHopRelays, BOOL, "1"),
Add country-code support to configured node lists to implement the ever-popular "no exits in Monaco" feature (ExcludeExitNodes {MC}). Also allow country codes and IP ranges in ExitNodes. (EntryNodes needs more work.) Based on code by Robert Hogan. Needs more testing. svn:r16966
2008-09-25 22:21:35 +02:00
V(ExitNodes, ROUTERSET, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(ExitPolicy, LINELIST, NULL),
V(ExitPolicyRejectPrivate, BOOL, "1"),
If configured, write per-port exit statistics to disk periodically. [Original patch series from Karsten, revised and squashed by Nick]
2009-06-24 19:51:45 +02:00
V(ExitPortStatistics, BOOL, "0"),
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
V(ExtraInfoStatistics, BOOL, "1"),
Change interface for configuring cell ewma algorithm. The rule is now: take the value from the CircuitPriorityHalflife config option if it is set. If it zero, disable the cell_ewma algorithm. If it is set, use it to calculate the scaling factor. If it is not set, look for a CircPriorityHalflifeMsec parameter in the consensus networkstatus. If *that* is zero, then disable the cell_ewma algorithm; if it is set, use it to calculate the scaling factor. If it is not set at all, disable the algorithm.
2009-12-15 19:53:53 +01:00
Port Tor to work on Windows CE Most of the changes here are switches to use APIs available on Windows CE. The most pervasive change is that Windows CE only provides the wide-character ("FooW") variants of most of the windows function, and doesn't support the older ASCII verions at all. This patch will require use of the wcecompat library to get working versions of the posix-style fd-based file IO functions. [commit message by nickm]
2010-05-21 04:53:39 +02:00
#if defined (WINCE)
V(FallbackNetworkstatusFile, FILENAME, "fallback-consensus"),
#else
Clean up a macro/cpp interaction related to bug 707. Also, add a new "filename" config value type which is currently just a synonym for string, but which might be subject to expansion later. svn:r15305
2008-06-16 20:09:53 +02:00
V(FallbackNetworkstatusFile, FILENAME,
r16252@catbus: nickm | 2007-10-29 11:13:03 -0400 Change symbol from DATADIR to SHARE_DATADIR, since DATADIR conflicts with objidl.h in mingw. svn:r12268
2007-10-29 16:15:49 +01:00
SHARE_DATADIR PATH_SEPARATOR "tor" PATH_SEPARATOR "fallback-consensus"),
Port Tor to work on Windows CE Most of the changes here are switches to use APIs available on Windows CE. The most pervasive change is that Windows CE only provides the wide-character ("FooW") variants of most of the windows function, and doesn't support the older ASCII verions at all. This patch will require use of the wcecompat library to get working versions of the posix-style fd-based file IO functions. [commit message by nickm]
2010-05-21 04:53:39 +02:00
#endif
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(FascistFirewall, BOOL, "0"),
V(FirewallPorts, CSV, ""),
V(FastFirstHopPK, BOOL, "1"),
new config option FetchDirInfoEarly for tup/ioerror and the tordnsel svn:r12896
2007-12-21 07:08:00 +01:00
V(FetchDirInfoEarly, BOOL, "0"),
Add support for a new option: FetchDirInfoExtraEarly This new option will allow clients to download the newest fresh consensus much sooner than they normally would do so, even if they previously set FetchDirInfoEarly. This includes a proper ChangeLog entry and an updated man page.
2009-07-12 03:43:33 +02:00
V(FetchDirInfoExtraEarly, BOOL, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(FetchServerDescriptors, BOOL, "1"),
V(FetchHidServDescriptors, BOOL, "1"),
V(FetchUselessDescriptors, BOOL, "0"),
Only authorities should automatically download v2 networkstatus documents Clients and relays haven't used them since early 0.2.0.x. The only remaining use by authorities learning about new relays ahead of scedule; see proposal 147 for what we intend to do about that. We're leaving in an option (FetchV2Networkstatus) to manually fetch v2 networkstatuses, because apparently dnsel and maybe bwauth want them. This fixes bug 3022.
2011-04-28 16:05:32 +02:00
V(FetchV2Networkstatus, BOOL, "0"),
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifdef _WIN32
Clean up a macro/cpp interaction related to bug 707. Also, add a new "filename" config value type which is currently just a synonym for string, but which might be subject to expansion later. svn:r15305
2008-06-16 20:09:53 +02:00
V(GeoIPFile, FILENAME, "<default>"),
On win32, default to looking for the geoip file in the same directory as torrc. This is a dumb hack; it should turn into a general mechanism. svn:r14796
2008-05-29 03:22:30 +02:00
#else
Clean up a macro/cpp interaction related to bug 707. Also, add a new "filename" config value type which is currently just a synonym for string, but which might be subject to expansion later. svn:r15305
2008-06-16 20:09:53 +02:00
V(GeoIPFile, FILENAME,
SHARE_DATADIR PATH_SEPARATOR "tor" PATH_SEPARATOR "geoip"),
On win32, default to looking for the geoip file in the same directory as torrc. This is a dumb hack; it should turn into a general mechanism. svn:r14796
2008-05-29 03:22:30 +02:00
#endif
Obsolete GiveGuardFlagTo_CVE_2011_2768_VulnerableRelays Closes ticket 4572.
2012-04-11 16:59:11 +02:00
OBSOLETE("GiveGuardFlagTo_CVE_2011_2768_VulnerableRelays"),
Patch from Jacob Appelbaum and me to make User option more robust, properly set supplementary groups, deprecated the Group option, and log more information on credential switching svn:r17200
2008-11-07 03:06:12 +01:00
OBSOLETE("Group"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(HardwareAccel, BOOL, "0"),
Initial heartbeat subsystem commit. Sets: * Documentation * Logging domain * Configuration option * Scheduled event * Makefile It also creates status.c and the log_heartbeat() function. All code was written by Sebastian Hahn. Commit message was written by me (George Kadianakis).
2010-12-01 02:24:03 +01:00
V(HeartbeatPeriod, INTERVAL, "6 hours"),
Add support for dynamic OpenSSL hardware crypto acceleration engines.
2009-05-24 01:42:44 +02:00
V(AccelName, STRING, NULL),
V(AccelDir, FILENAME, NULL),
Allow multiple HashedControlPassword config lines, to support multiple controller passwords. svn:r12732
2007-12-09 05:59:27 +01:00
V(HashedControlPassword, LINELIST, NULL),
Directory mirrors store and serve v2 hidden service descriptors by default. svn:r16858
2008-09-11 22:06:04 +02:00
V(HidServDirectoryV2, BOOL, "1"),
Alphebetize config variables svn:r4714
2005-08-05 03:35:43 +02:00
VAR("HiddenServiceDir", LINELIST_S, RendConfigLines, NULL),
r17309@aud-055: nickm | 2008-07-23 16:05:43 +0200 Patch from Christian Wilms: remove (HiddenService|Rend)(Exclude)?Nodes options. They never worked properly, and nobody seems to be using them. Resolves bug 754. svn:r16144
2008-07-23 16:07:32 +02:00
OBSOLETE("HiddenServiceExcludeNodes"),
OBSOLETE("HiddenServiceNodes"),
Alphebetize config variables svn:r4714
2005-08-05 03:35:43 +02:00
VAR("HiddenServiceOptions",LINELIST_V, RendConfigLines, NULL),
VAR("HiddenServicePort", LINELIST_S, RendConfigLines, NULL),
r16300@catbus: nickm | 2007-10-31 15:36:41 -0400 Next patch from Karsten: rename some macros, tunnel dir connections, generate (and upload) multiple descriptors as appropriate. svn:r12299
2007-10-31 21:48:06 +01:00
VAR("HiddenServiceVersion",LINELIST_S, RendConfigLines, NULL),
The first of Karsten's proposal 121 patches: configure and maintain client authorization data. Tweaked a bit: see comments on or-dev. svn:r16475
2008-08-08 16:36:11 +02:00
VAR("HiddenServiceAuthorizeClient",LINELIST_S,RendConfigLines, NULL),
Next patch from Karsten: client-side configuration stuff for proposal 121. svn:r16510
2008-08-12 18:12:26 +02:00
V(HidServAuth, LINELIST, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(HSAuthoritativeDir, BOOL, "0"),
Remove v0 hidden service statistics code. The HSAuthorityRecordStats option was used to track statistics of overall hidden service usage on the version 0 hidden service authorities. With the version 2 hidden service directories being deployed and version 0 descriptors being phased out, these statistics are not as useful anymore. Goodbye, you fine piece of software; my first major code contribution to Tor.
2009-10-28 07:47:47 +01:00
OBSOLETE("HSAuthorityRecordStats"),
Add an option to close 'almost-connected' HS client circs on timeout
2011-12-24 09:46:37 +01:00
V(CloseHSClientCircuitsImmediatelyOnTimeout, BOOL, "0"),
Add an option to close HS service-side rend circs on timeout
2011-12-24 13:55:20 +01:00
V(CloseHSServiceRendCircuitsImmediatelyOnTimeout, BOOL, "0"),
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
V(HTTPProxy, STRING, NULL),
V(HTTPProxyAuthenticator, STRING, NULL),
V(HTTPSProxy, STRING, NULL),
V(HTTPSProxyAuthenticator, STRING, NULL),
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
VAR("ServerTransportPlugin", LINELIST, ServerTransportPlugin, NULL),
Enable Tor to connect through SOCKS 4/5 proxies Added a sanity check in config.c and a check in directory.c directory_initiate_command_rend() to catch any direct connection attempts when a socks proxy is configured.
2009-06-19 01:59:18 +02:00
V(Socks4Proxy, STRING, NULL),
V(Socks5Proxy, STRING, NULL),
V(Socks5ProxyUsername, STRING, NULL),
V(Socks5ProxyPassword, STRING, NULL),
get rid of IgnoreVersion and the forced auto-exit mis-feature svn:r4746
2005-08-08 23:52:57 +02:00
OBSOLETE("IgnoreVersion"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(KeepalivePeriod, INTERVAL, "5 minutes"),
VAR("Log", LINELIST, Logs, NULL),
Add a torrc option to report log domains
2011-01-25 21:53:15 +01:00
V(LogMessageDomains, BOOL, "0"),
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
OBSOLETE("LinkPadding"),
r11658@Kushana: nickm | 2006-12-20 15:58:44 -0500 Remove long-deprecated log and accounting options. svn:r9164
2006-12-20 22:02:10 +01:00
OBSOLETE("LogLevel"),
OBSOLETE("LogFile"),
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
V(LogTimeGranularity, MSEC_INTERVAL, "1 second"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(LongLivedPorts, CSV,
Fix overwide line in config.c
2011-06-20 21:29:22 +02:00
"21,22,706,1863,5050,5190,5222,5223,6523,6667,6697,8300"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
VAR("MapAddress", LINELIST, AddressMap, NULL),
V(MaxAdvertisedBandwidth, MEMUNIT, "1 GB"),
V(MaxCircuitDirtiness, INTERVAL, "10 minutes"),
Add an option to limit the number of non-open client circuits. This is mainly meant as a way to keep clients from accidentally DOSing themselves by (e.g.) enabling IsolateDestAddr or IsolateDestPort on a port that they use for HTTP.
2011-07-07 20:54:54 +02:00
V(MaxClientCircuitsPending, UINT, "32"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(MaxOnionsPending, UINT, "100"),
clean up some confusing logs. make MonthlyAccountingStart truly obsolete now. svn:r4782
2005-08-15 11:17:47 +02:00
OBSOLETE("MonthlyAccountingStart"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(MyFamily, STRING, NULL),
V(NewCircuitPeriod, INTERVAL, "30 seconds"),
Now that directory admins assure me they have adjested teir configs, it is safe to make authorities be non-versioning, non-naming by default. svn:r5230
2005-10-08 07:47:58 +02:00
VAR("NamingAuthoritativeDirectory",BOOL, NamingAuthoritativeDir, "0"),
Change Natd into NATD in our options. Breaking this out of the last commit because this might be more controversial.
2010-10-30 06:08:47 +02:00
V(NATDListenAddress, LINELIST, NULL),
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
V(NATDPort, LINELIST, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(Nickname, STRING, NULL),
Add option to not warn when getting an IP instead of hostname
2010-06-03 12:52:34 +02:00
V(WarnUnsafeSocks, BOOL, "1"),
Break NoPublish support
2010-10-30 06:38:50 +02:00
OBSOLETE("NoPublish"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
VAR("NodeFamily", LINELIST, NodeFamilies, NULL),
Merge remote branch 'origin/maint-0.2.2' Conflicts: src/or/config.c src/or/cpuworker.c
2010-11-15 20:14:13 +01:00
V(NumCPUs, UINT, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(NumEntryGuards, UINT, "3"),
V(ORListenAddress, LINELIST, NULL),
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
V(ORPort, LINELIST, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(OutboundBindAddress, STRING, NULL),
r11726@catbus: nickm | 2007-02-08 16:04:53 -0500 Resolve some XXXX012 items: - Remove PathlenCoinWeight: if we want it again, we can add it back in. - Ditto with RelayBandwidth*. - Decide to leave in the "hey, you didn't set end_reason!" BUG log message, but stop telling people to bug me personally. - Postpone strengthening assert_connection_ok(): it's important, but it's also a good way to introduce weird bugs. - Move some expensive consistency checking from dns_free_all() into assert_cache_ok(). svn:r9533
2007-02-08 23:07:56 +01:00
OBSOLETE("PathlenCoinWeight"),
add config options to override. somebody should add man page entries.
2009-12-30 05:13:03 +01:00
V(PerConnBWBurst, MEMUNIT, "0"),
V(PerConnBWRate, MEMUNIT, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(PidFile, STRING, NULL),
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
V(TestingTorNetwork, BOOL, "0"),
Add an OptimisticData option to control client-side optimistic data
2011-07-20 16:38:00 +02:00
V(OptimisticData, AUTOBOOL, "auto"),
Start tor-fw-helper in the background, and log whatever it outputs
2010-06-16 20:47:06 +02:00
V(PortForwarding, BOOL, "0"),
V(PortForwardingHelper, FILENAME, "tor-fw-helper"),
forward-port recent changes svn:r13935
2008-03-10 08:50:09 +01:00
V(PreferTunneledDirConns, BOOL, "1"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(ProtocolWarnings, BOOL, "0"),
Make PublishServerDescriptor default to 1, so the default doesn't have to change as we invent new directory protocol versions. svn:r12922
2007-12-22 09:27:42 +01:00
V(PublishServerDescriptor, CSV, "1"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(PublishHidServDescriptors, BOOL, "1"),
V(ReachableAddresses, LINELIST, NULL),
V(ReachableDirAddresses, LINELIST, NULL),
V(ReachableORAddresses, LINELIST, NULL),
V(RecommendedVersions, LINELIST, NULL),
V(RecommendedClientVersions, LINELIST, NULL),
V(RecommendedServerVersions, LINELIST, NULL),
Remove RedirectExit feature; it has been deprecated since 0.2.0.3-alpha svn:r17663
2008-12-18 00:02:04 +01:00
OBSOLETE("RedirectExit"),
Add a new configuration type, "AUTOBOOL", to handle 1/0/auto types We only have one of these now, but I'm about to add a few more. Yes, I have already thought of the "Filenotfoundian logic" joke.
2010-11-08 19:34:40 +01:00
V(RefuseUnknownExits, AUTOBOOL, "auto"),
New config options WarnPlaintextPorts and RejectPlaintextPorts so Tor can warn and/or refuse connections to ports commonly used with vulnerable-plaintext protocols. We still need to figure out some good defaults for them. svn:r13198
2008-01-20 06:54:15 +01:00
V(RejectPlaintextPorts, CSV, ""),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(RelayBandwidthBurst, MEMUNIT, "0"),
V(RelayBandwidthRate, MEMUNIT, "0"),
r17309@aud-055: nickm | 2008-07-23 16:05:43 +0200 Patch from Christian Wilms: remove (HiddenService|Rend)(Exclude)?Nodes options. They never worked properly, and nobody seems to be using them. Resolves bug 754. svn:r16144
2008-07-23 16:07:32 +02:00
OBSOLETE("RendExcludeNodes"),
OBSOLETE("RendNodes"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(RendPostPeriod, INTERVAL, "1 hour"),
V(RephistTrackTime, INTERVAL, "24 hours"),
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
OBSOLETE("RouterFile"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(RunAsDaemon, BOOL, "0"),
Comment out the (unused) RunTesting option The code that makes use of the RunTesting option is #if 0, so setting this option has no effect. Mark the option as obsolete for now, so that Tor doesn't list it as an available option erroneously.
2010-10-30 00:03:31 +02:00
// V(RunTesting, BOOL, "0"),
OBSOLETE("RunTesting"), // currently unused
Allow SafeLogging to exclude client related information
2009-09-28 15:08:32 +02:00
V(SafeLogging, STRING, "1"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(SafeSocks, BOOL, "0"),
Rename ServerDNSAllowBrokenResolvConf to ServerDNSAllowBrokenConfig. (Many users have no idea what a resolv.conf is, and shouldn't be forced to learn. The old option will keep working for now.) Also, document it. svn:r17661
2008-12-17 23:58:14 +01:00
V(ServerDNSAllowBrokenConfig, BOOL, "1"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(ServerDNSAllowNonRFC953Hostnames, BOOL,"0"),
V(ServerDNSDetectHijacking, BOOL, "1"),
Implement the 0x20-hack to make DNS poisoning harder against us, especially when resolving large names. Add a cfg option to disable it, since apparently 3/10 of a percent of servers get it wrong. svn:r17171
2008-10-29 20:20:02 +01:00
V(ServerDNSRandomizeCase, BOOL, "1"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(ServerDNSResolvConfFile, STRING, NULL),
V(ServerDNSSearchDomains, BOOL, "0"),
V(ServerDNSTestAddresses, CSV,
r11723@Kushana: nickm | 2006-12-28 13:52:48 -0500 Fix bug 364: check for whether popular hostnames (curently google, yahoo, mit, and slashdot) are getting wildcarded. If they are, we are probably behind a DNS server that is useless: change our exit policy to reject *:*. svn:r9199
2006-12-28 22:29:11 +01:00
"www.google.com,www.mit.edu,www.yahoo.com,www.slashdot.org"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(ShutdownWaitLength, INTERVAL, "30 seconds"),
V(SocksListenAddress, LINELIST, NULL),
V(SocksPolicy, LINELIST, NULL),
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
V(SocksPort, LINELIST, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(SocksTimeout, INTERVAL, "2 minutes"),
make DirFetchPeriod and StatusFetchPeriod truly obsolete. svn:r6373
2006-04-10 22:16:46 +02:00
OBSOLETE("StatusFetchPeriod"),
Switch to a StrictNodes config option. This is step one of handling ExcludedNodes better. This first step is just to make EntryNodes and ExitNodes do what they did before.
2009-09-18 04:45:54 +02:00
V(StrictNodes, BOOL, "0"),
r11658@Kushana: nickm | 2006-12-20 15:58:44 -0500 Remove long-deprecated log and accounting options. svn:r9164
2006-12-20 22:02:10 +01:00
OBSOLETE("SysLog"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(TestSocks, BOOL, "0"),
Take out the TestVia config option, since it was a workaround for a bug that was fixed in Tor 0.1.1.21. svn:r16409
2008-08-05 02:12:05 +02:00
OBSOLETE("TestVia"),
refill our token buckets 10 times/sec, not 100 refilling often is good, but refilling often has unclear side effects on a) cpu load, and b) making sure every cell, ever, is sent out one at a time
2011-09-28 21:38:36 +02:00
V(TokenBucketRefillInterval, MSEC_INTERVAL, "100 msec"),
Add Tor2webMode configuration option
2011-06-01 16:03:01 +02:00
V(Tor2webMode, BOOL, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(TrackHostExits, CSV, NULL),
V(TrackHostExitsExpire, INTERVAL, "30 minutes"),
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
OBSOLETE("TrafficShaping"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(TransListenAddress, LINELIST, NULL),
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
V(TransPort, LINELIST, NULL),
forward-port recent changes svn:r13935
2008-03-10 08:50:09 +01:00
V(TunnelDirConns, BOOL, "1"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(UpdateBridgesFromAuthority, BOOL, "0"),
Revert "Add an "auto" option to UseBridges" This reverts commit 507c1257a4d9c629fefc2adbad8db73607749734.
2011-06-17 22:45:53 +02:00
V(UseBridges, BOOL, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(UseEntryGuards, BOOL, "1"),
Turn on microdescriptors for clients
2011-07-11 22:54:43 +02:00
V(UseMicrodescriptors, AUTOBOOL, "auto"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(User, STRING, NULL),
New UserspaceIOCPBuffers option to set SO_{SND,RCV}BUF to zero When running with IOCP, we are in theory able to use userspace- allocated buffers to avoid filling up the stingy amount of kernel space allocated for sockets buffers. The bufferevent_async implementation in Libevent provides this ability, in theory. (There are likely to be remaining bugs). This patch adds a new option that, when using IOCP bufferevents, sets each socket's send and receive buffers to 0, so that we should use this ability. When all the bugs are worked out here, if we are right about bug 98, this might solve or mitigate bug 98. This option is experimental and will likely require lots of testing and debugging.
2011-11-18 23:43:03 +01:00
V(UserspaceIOCPBuffers, BOOL, "0"),
Make authdirs smarter. Reject/invalidate by key and IP. Remember that not every authdir has to be a naming authdir. svn:r5074
2005-09-15 16:39:05 +02:00
VAR("V1AuthoritativeDirectory",BOOL, V1AuthoritativeDir, "0"),
New config option V2AuthoritativeDirectory that all directory authorities should set. This will let future authorities choose not to serve V2 directory information. Also, go through and revamp all the authdir_mode stuff so it tries to do the right thing if you're an auth but not a V1 or V2 auth. svn:r10092
2007-05-02 11:12:04 +02:00
VAR("V2AuthoritativeDirectory",BOOL, V2AuthoritativeDir, "0"),
r12898@catbus: nickm | 2007-05-22 13:11:04 -0400 More v3 directory code: have authorities load certificates; have everybody store certificates to disk and load them; provide a way to configure v3 authorities. svn:r10293
2007-05-22 19:58:25 +02:00
VAR("V3AuthoritativeDirectory",BOOL, V3AuthoritativeDir, "0"),
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
V(TestingV3AuthInitialVotingInterval, INTERVAL, "30 minutes"),
V(TestingV3AuthInitialVoteDelay, INTERVAL, "5 minutes"),
V(TestingV3AuthInitialDistDelay, INTERVAL, "5 minutes"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(V3AuthVotingInterval, INTERVAL, "1 hour"),
V(V3AuthVoteDelay, INTERVAL, "5 minutes"),
V(V3AuthDistDelay, INTERVAL, "5 minutes"),
V(V3AuthNIntervalsValid, UINT, "3"),
r19690@catbus: nickm | 2008-05-11 22:13:31 -0400 Implement a proposal to let a directory authority migrate its identity key without ceasing to sign consensuses. svn:r14584
2008-05-12 04:14:01 +02:00
V(V3AuthUseLegacyKey, BOOL, "0"),
Implement measured bw parsing + unit tests.
2009-07-31 06:33:53 +02:00
V(V3BandwidthsFile, FILENAME, NULL),
Now that directory admins assure me they have adjested teir configs, it is safe to make authorities be non-versioning, non-naming by default. svn:r5230
2005-10-08 07:47:58 +02:00
VAR("VersioningAuthoritativeDirectory",BOOL,VersioningAuthoritativeDir, "0"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(VirtualAddrNetwork, STRING, "127.192.0.0/10"),
New config options WarnPlaintextPorts and RejectPlaintextPorts so Tor can warn and/or refuse connections to ports commonly used with vulnerable-plaintext protocols. We still need to figure out some good defaults for them. svn:r13198
2008-01-20 06:54:15 +01:00
V(WarnPlaintextPorts, CSV, "23,109,110,143"),
Add a testing-only option to use bufferevent_openssl as a filter We need filtering bufferevent_openssl so that we can wrap around IOCP bufferevents on Windows. This patch adds a temporary option to turn on filtering mode, so that we can test it out on non-IOCP systems to make sure it hasn't got any surprising bugs. It also fixes some allocation/teardown errors in using bufferevent_openssl as a filter.
2010-11-09 21:36:27 +01:00
V(_UseFilteringSSLBufferevents, BOOL, "0"),
Add new internal-use-only option for controllers to use to prevent SIGHUP from reloading the configuration. Fixes bug 856. svn:r17567
2008-12-10 23:17:02 +01:00
VAR("__ReloadTorrcOnSIGHUP", BOOL, ReloadTorrcOnSIGHUP, "1"),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
VAR("__AllDirActionsPrivate", BOOL, AllDirActionsPrivate, "0"),
VAR("__DisablePredictedCircuits",BOOL,DisablePredictedCircuits, "0"),
VAR("__LeaveStreamsUnattached",BOOL, LeaveStreamsUnattached, "0"),
r14236@tombo: nickm | 2008-02-17 13:44:55 -0500 Partial fix for bug 586: Add an ephemeral __HashedControlSessionPassword. svn:r13543
2008-02-17 19:45:07 +01:00
VAR("__HashedControlSessionPassword", LINELIST, HashedControlSessionPassword,
NULL),
Implement __OwningControllerProcess option Implements part of feature 3049.
2011-05-15 17:23:04 +02:00
VAR("__OwningControllerProcess",STRING,OwningControllerProcess, NULL),
Increase default required uptime for HSDirs to 25 hours
2011-05-10 12:35:45 +02:00
V(MinUptimeHidServDirectoryV2, INTERVAL, "25 hours"),
Add a VoteOnHidServDirectoriesV2 configuration option
2011-05-10 11:06:07 +02:00
V(VoteOnHidServDirectoriesV2, BOOL, "1"),
Tweak the bug2250 fix slightly Rename the _UsingTestingTorNetwork hidden option to _UsingTestNetworkDefaults (thanks Nick for the suggestion) and added a changes file.
2011-03-01 10:05:18 +01:00
V(_UsingTestNetworkDefaults, BOOL, "0"),
Favor quiet circuits when choosing which order to relay cells in. Each circuit is ranked in terms of how many cells from it have been relayed recently, using a time-weighted average. This patch has been tested this on a private Tor network on PlanetLab, and gotten improvements of 12-35% in time it takes to fetch a small web page while there's a simultaneous large data transfer going on simultaneously. [Commit msg by nickm based on mail from Ian Goldberg.]
2009-12-10 17:12:42 +01:00
Add descriptions for state variables; rename AccountingBytesRead{In}Interval; remove extraneous and unused struct member. svn:r5739
2006-01-06 16:43:03 +01:00
{ NULL, CONFIG_TYPE_OBSOLETE, 0, NULL }
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
};
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
Document most undocumented variables. svn:r17754
2008-12-23 18:56:31 +01:00
/** Override default values with these if the user sets the TestingTorNetwork
* option. */
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
static const config_var_t testing_tor_network_defaults[] = {
Rename ServerDNSAllowBrokenResolvConf to ServerDNSAllowBrokenConfig. (Many users have no idea what a resolv.conf is, and shouldn't be forced to learn. The old option will keep working for now.) Also, document it. svn:r17661
2008-12-17 23:58:14 +01:00
V(ServerDNSAllowBrokenConfig, BOOL, "1"),
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
V(DirAllowPrivateAddresses, BOOL, "1"),
V(EnforceDistinctSubnets, BOOL, "0"),
V(AssumeReachable, BOOL, "1"),
V(AuthDirMaxServersPerAddr, UINT, "0"),
V(AuthDirMaxServersPerAuthAddr,UINT, "0"),
V(ClientDNSRejectInternalAddresses, BOOL,"0"),
Add an option to disable the block-private-addresses feature Suggested by rransom. Probably necessary for testing network mode.
2011-01-26 17:35:24 +01:00
V(ClientRejectInternalAddresses, BOOL, "0"),
Patch from cagara: Add a CountPrivateBandwidth flag
2011-04-05 20:50:32 +02:00
V(CountPrivateBandwidth, BOOL, "1"),
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
V(ExitPolicyRejectPrivate, BOOL, "0"),
V(V3AuthVotingInterval, INTERVAL, "5 minutes"),
V(V3AuthVoteDelay, INTERVAL, "20 seconds"),
V(V3AuthDistDelay, INTERVAL, "20 seconds"),
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
V(TestingV3AuthInitialVotingInterval, INTERVAL, "5 minutes"),
V(TestingV3AuthInitialVoteDelay, INTERVAL, "20 seconds"),
V(TestingV3AuthInitialDistDelay, INTERVAL, "20 seconds"),
V(TestingAuthDirTimeToLearnReachability, INTERVAL, "0 minutes"),
V(TestingEstimatedDescriptorPropagationTime, INTERVAL, "0 minutes"),
Remove delay to become HSDir in privnets
2010-10-26 18:37:57 +02:00
V(MinUptimeHidServDirectoryV2, INTERVAL, "0 minutes"),
Tweak the bug2250 fix slightly Rename the _UsingTestingTorNetwork hidden option to _UsingTestNetworkDefaults (thanks Nick for the suggestion) and added a changes file.
2011-03-01 10:05:18 +01:00
V(_UsingTestNetworkDefaults, BOOL, "1"),
Patch from cagara: Add a CountPrivateBandwidth flag
2011-04-05 20:50:32 +02:00
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
{ NULL, CONFIG_TYPE_OBSOLETE, 0, NULL }
};
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
#undef VAR
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
#define VAR(name,conftype,member,initvalue) \
{ name, CONFIG_TYPE_ ## conftype, STRUCT_OFFSET(or_state_t, member), \
Add descriptions for state variables; rename AccountingBytesRead{In}Interval; remove extraneous and unused struct member. svn:r5739
2006-01-06 16:43:03 +01:00
initvalue }
Document most undocumented variables. svn:r17754
2008-12-23 18:56:31 +01:00
/** Array of "state" variables saved to the ~/.tor/state file. */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
static config_var_t _state_vars[] = {
Write initial documentation for the contents of the state file Fixes bug 2987. There is still some information to go, but now we have a place to put it.
2012-03-28 10:08:07 +02:00
/* Remember to document these in state-contents.txt ! */
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(AccountingBytesReadInInterval, MEMUNIT, NULL),
V(AccountingBytesWrittenInInterval, MEMUNIT, NULL),
V(AccountingExpectedUsage, MEMUNIT, NULL),
V(AccountingIntervalStart, ISOTIME, NULL),
V(AccountingSecondsActive, INTERVAL, NULL),
Base our expected bw accounting usage on time before soft limit Previously, we were also considering the time spent in soft-hibernation. If this was a long time, we would wind up underestimating our bandwidth by a lot, and skewing our wakeup time towards the start of the accounting interval. This patch also makes us store a few more fields in the state file, including the time at which we entered soft hibernation. Fixes bug 1789. Bugfix on 0.0.9pre5.
2010-09-03 20:29:17 +02:00
V(AccountingSecondsToReachSoftLimit,INTERVAL, NULL),
V(AccountingSoftLimitHitAt, ISOTIME, NULL),
V(AccountingBytesAtSoftLimit, MEMUNIT, NULL),
cleanups on r11258 svn:r11263
2007-08-24 10:01:47 +02:00
entry nodes are now entry guards. this is our last easy chance for a wholesale change. heave ho. svn:r5782
2006-01-10 23:42:44 +01:00
VAR("EntryGuard", LINELIST_S, EntryGuards, NULL),
VAR("EntryGuardDownSince", LINELIST_S, EntryGuards, NULL),
VAR("EntryGuardUnlistedSince", LINELIST_S, EntryGuards, NULL),
r14826@catbus: nickm | 2007-08-29 13:19:55 -0400 Add a line to the state file for each guard to let us know which version added the guard. If the line is absent, assume the guard was added by whatever version of Tor last wrote the state file. Remove guards if the version that added them was using a bad guard selection algorithm. (Previously, we removed guards if the version that wrote the file was using a bad guard selection algorithm, even if the guards themselves were chosen by a good version.) svn:r11298
2007-08-29 19:22:00 +02:00
VAR("EntryGuardAddedBy", LINELIST_S, EntryGuards, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(EntryGuards, LINELIST_V, NULL),
Alphebetize config variables svn:r4714
2005-08-05 03:35:43 +02:00
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
VAR("TransportProxy", LINELIST_S, TransportProxies, NULL),
V(TransportProxies, LINELIST_V, NULL),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(BWHistoryReadEnds, ISOTIME, NULL),
V(BWHistoryReadInterval, UINT, "900"),
V(BWHistoryReadValues, CSV, ""),
Add Maxima lists to bandwidth state. Right now, Tor routers don't save the maxima values from the bw_history_t between sessions. That's no good, since we use those values to determine bandwidth. This code adds a new BWHist.*Maximum set of values to the state file. If they're not present, we estimate them by taking the observed total bandwidth and dividing it by the period length, which provides a lower bound. This should fix bug 1863. I'm calling it a feature.
2011-01-10 20:11:22 +01:00
V(BWHistoryReadMaxima, CSV, ""),
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(BWHistoryWriteEnds, ISOTIME, NULL),
V(BWHistoryWriteInterval, UINT, "900"),
V(BWHistoryWriteValues, CSV, ""),
Add Maxima lists to bandwidth state. Right now, Tor routers don't save the maxima values from the bw_history_t between sessions. That's no good, since we use those values to determine bandwidth. This code adds a new BWHist.*Maximum set of values to the state file. If they're not present, we estimate them by taking the observed total bandwidth and dividing it by the period length, which provides a lower bound. This should fix bug 1863. I'm calling it a feature.
2011-01-10 20:11:22 +01:00
V(BWHistoryWriteMaxima, CSV, ""),
Count bytes we spend on answering directory requests.
2010-07-07 14:55:42 +02:00
V(BWHistoryDirReadEnds, ISOTIME, NULL),
V(BWHistoryDirReadInterval, UINT, "900"),
V(BWHistoryDirReadValues, CSV, ""),
Add Maxima lists to bandwidth state. Right now, Tor routers don't save the maxima values from the bw_history_t between sessions. That's no good, since we use those values to determine bandwidth. This code adds a new BWHist.*Maximum set of values to the state file. If they're not present, we estimate them by taking the observed total bandwidth and dividing it by the period length, which provides a lower bound. This should fix bug 1863. I'm calling it a feature.
2011-01-10 20:11:22 +01:00
V(BWHistoryDirReadMaxima, CSV, ""),
Count bytes we spend on answering directory requests.
2010-07-07 14:55:42 +02:00
V(BWHistoryDirWriteEnds, ISOTIME, NULL),
V(BWHistoryDirWriteInterval, UINT, "900"),
V(BWHistoryDirWriteValues, CSV, ""),
Add Maxima lists to bandwidth state. Right now, Tor routers don't save the maxima values from the bw_history_t between sessions. That's no good, since we use those values to determine bandwidth. This code adds a new BWHist.*Maximum set of values to the state file. If they're not present, we estimate them by taking the observed total bandwidth and dividing it by the period length, which provides a lower bound. This should fix bug 1863. I'm calling it a feature.
2011-01-10 20:11:22 +01:00
V(BWHistoryDirWriteMaxima, CSV, ""),
Keep bandwidth history accross restarts/crashes svn:r5637
2005-12-24 00:56:42 +01:00
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(TorVersion, STRING, NULL),
add a TorVersion state string. not yet set to anything. svn:r5687
2005-12-31 09:16:07 +01:00
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
V(LastRotatedOnionKey, ISOTIME, NULL),
V(LastWritten, ISOTIME, NULL),
helper nodes are dead. long live entry nodes. (config options EntryNodes and StrictEntryNodes still not implemented.) svn:r5673
2005-12-28 10:07:31 +01:00
Fix issues found by arma in review.
2009-09-07 05:43:02 +02:00
V(TotalBuildTimes, UINT, NULL),
tiny change i found in my other sandbox
2010-08-11 07:48:29 +02:00
V(CircuitBuildAbandonedCount, UINT, "0"),
space/indent cleanups, plus point out three bugs
2009-09-07 05:14:13 +02:00
VAR("CircuitBuildTimeBin", LINELIST_S, BuildtimeHistogram, NULL),
VAR("BuildtimeHistogram", LINELIST_V, BuildtimeHistogram, NULL),
Add descriptions for state variables; rename AccountingBytesRead{In}Interval; remove extraneous and unused struct member. svn:r5739
2006-01-06 16:43:03 +01:00
{ NULL, CONFIG_TYPE_OBSOLETE, 0, NULL }
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
};
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
#undef VAR
r14231@Kushana: nickm | 2007-08-27 14:56:14 -0400 Most configuration options have the same names in or.h and in torrc. Make the macros reflect this, so that it is easier to fit them onto a line, and so that mismatched options stand out more svn:r11290
2007-08-27 20:56:20 +02:00
#undef V
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
#undef OBSOLETE
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Represents an English description of a configuration variable; used when
* generating configuration file comments. */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
typedef struct config_var_description_t {
const char *name;
const char *description;
} config_var_description_t;
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
/** Type of a callback to validate whether a given configuration is
doc pedant svn:r9634
2007-02-24 08:50:38 +01:00
* well-formed and consistent. See options_trial_assign() for documentation
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
* of arguments. */
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
typedef int (*validate_fn_t)(void*,void*,int,char**);
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
Make doxygen marginally happier svn:r5208
2005-10-06 06:33:40 +02:00
/** Information on the keys, value types, key-to-struct-member mappings,
* variable descriptions, validation functions, and abbreviations for a
* configuration or storage format. */
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
typedef struct {
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
size_t size; /**< Size of the struct that everything gets parsed into. */
uint32_t magic; /**< Required 'magic value' to make sure we have a struct
* of the right type. */
doc pedant svn:r9634
2007-02-24 08:50:38 +01:00
off_t magic_offset; /**< Offset of the magic value within the struct. */
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
config_abbrev_t *abbrevs; /**< List of abbreviations that we expand when
doc pedant svn:r9634
2007-02-24 08:50:38 +01:00
* parsing this format. */
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
config_var_t *vars; /**< List of variables we recognize, their default
* values, and where we stick them in the structure. */
validate_fn_t validate_fn; /**< Function to validate config. */
resolve too-long-lines svn:r6001
2006-02-13 09:22:57 +01:00
/** If present, extra is a LINELIST variable for unrecognized
* lines. Otherwise, unrecognized lines are an error. */
config_var_t *extra;
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
} config_format_t;
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
/** Macro: assert that <b>cfg</b> has the right magic field for format
* <b>fmt</b>. */
r13477@catbus: nickm | 2007-06-17 14:22:03 -0400 Sun CC likes to give warnings for the do { } while(0) construction for making statement-like macros. Define STMT_BEGIN/STMT_END macros that do the right thing, and use them everywhere. svn:r10645
2007-06-17 20:22:39 +02:00
#define CHECK(fmt, cfg) STMT_BEGIN \
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
tor_assert(fmt && cfg); \
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
tor_assert((fmt)->magic == \
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
*(uint32_t*)STRUCT_VAR_P(cfg,fmt->magic_offset)); \
r13477@catbus: nickm | 2007-06-17 14:22:03 -0400 Sun CC likes to give warnings for the do { } while(0) construction for making statement-like macros. Define STMT_BEGIN/STMT_END macros that do the right thing, and use them everywhere. svn:r10645
2007-06-17 20:22:39 +02:00
STMT_END
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifdef _WIN32
blind-forward-port that, while i'm at it svn:r14929
2008-06-04 09:08:05 +02:00
static char *get_windows_conf_root(void);
#endif
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
static void config_line_append(config_line_t **lst,
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
const char *key, const char *val);
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
static void option_clear(const config_format_t *fmt, or_options_t *options,
const config_var_t *var);
static void option_reset(const config_format_t *fmt, or_options_t *options,
const config_var_t *var, int use_defaults);
static void config_free(const config_format_t *fmt, void *options);
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
static int config_lines_eq(config_line_t *a, config_line_t *b);
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
static int option_is_same(const config_format_t *fmt,
const or_options_t *o1, const or_options_t *o2,
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
const char *name);
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
static or_options_t *options_dup(const config_format_t *fmt,
const or_options_t *old);
static int options_validate(or_options_t *old_options,
or_options_t *options,
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
int from_setconf, char **msg);
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
static int options_act_reversible(const or_options_t *old_options, char **msg);
static int options_act(const or_options_t *old_options);
static int options_transition_allowed(const or_options_t *old,
const or_options_t *new,
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
char **msg);
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
static int options_transition_affects_workers(
const or_options_t *old_options, const or_options_t *new_options);
static int options_transition_affects_descriptor(
const or_options_t *old_options, const or_options_t *new_options);
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
static int check_nickname_list(const char *lst, const char *name, char **msg);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
static int parse_bridge_line(const char *line, int validate_only);
Various trivial changes. * Improved function documentation. * Renamed find_bridge_transport_by_addrport() to find_transport_by_bridge_addrport(). * Sanitized log severities we use. * Ran check-spaces.
2011-06-12 16:41:32 +02:00
static int parse_client_transport_line(const char *line, int validate_only);
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
static int parse_server_transport_line(const char *line, int validate_only);
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
static int parse_dir_server_line(const char *line,
Automated rename from authority_type_t to dirinfo_type_t We were already overloading this type to mean "a directory that can serve us X" in addition to "a directory that is an authority for X."
2010-11-08 20:27:36 +01:00
dirinfo_type_t required_type,
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
int validate_only);
(Unused) backend logic for stream isolation This patch adds fields to track how streams should be isolated, and ensures that those fields are set correctly. It also adds fields to track what streams can go on a circuit, and adds functions to see whether a streams can go on a circuit and update the circuit accordingly. Those functions aren't yet called.
2011-07-06 22:03:47 +02:00
static void port_cfg_free(port_cfg_t *port);
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
static int parse_ports(const or_options_t *options, int validate_only,
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
char **msg_out, int *n_ports_out);
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
static int check_server_ports(const smartlist_t *ports,
const or_options_t *options);
make datadirectory work again svn:r2724
2004-11-09 06:35:49 +01:00
static int validate_data_directory(or_options_t *options);
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
static int write_configuration_file(const char *fname,
const or_options_t *options);
static config_line_t *get_assigned_option(const config_format_t *fmt,
const void *options, const char *key,
int escape_val);
static void config_init(const config_format_t *fmt, void *options);
or_state_validate() needs an extra arg too or it will crash svn:r5772
2006-01-10 21:38:33 +01:00
static int or_state_validate(or_state_t *old_options, or_state_t *options,
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
int from_setconf, char **msg);
r15512@catbus: nickm | 2007-10-02 16:27:43 -0400 Make some functions static; remove some dead code. svn:r11750
2007-10-02 22:35:23 +02:00
static int or_state_load(void);
static int options_init_logs(or_options_t *options, int validate_only);
clean up config.c so it doesn't expose as much svn:r430
2003-09-08 07:16:18 +02:00
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
static uint64_t config_parse_memunit(const char *s, int *ok);
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
static int config_parse_msec_interval(const char *s, int *ok);
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
static int config_parse_interval(const char *s, int *ok);
Add a new option to enable/disable IOCP support
2010-09-28 20:01:45 +02:00
static void init_libevent(const or_options_t *options);
Warn when binding to a public address for socks. This warning needs to be improved. Also, make options_validate() only warn when an option is changed to a bad value.. (Previously, it warned when an option was bad and any option was changed.) svn:r5532
2005-12-08 19:56:32 +01:00
static int opt_streq(const char *s1, const char *s2);
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
/** Magic value for or_options_t. */
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
#define OR_OPTIONS_MAGIC 9090909
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
/** Configuration format for or_options_t. */
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
static config_format_t options_format = {
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
sizeof(or_options_t),
OR_OPTIONS_MAGIC,
STRUCT_OFFSET(or_options_t, _magic),
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
_option_abbrevs,
_option_vars,
(validate_fn_t)options_validate,
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
NULL
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
};
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
/** Magic value for or_state_t. */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
#define OR_STATE_MAGIC 0x57A73f57
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
/** "Extra" variable in the state that receives lines we can't parse. This
* lets us preserve options from versions of Tor newer than us. */
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
static config_var_t state_extra_var = {
"__extra", CONFIG_TYPE_LINELIST, STRUCT_OFFSET(or_state_t, ExtraLines), NULL
};
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
/** Configuration format for or_state_t. */
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
static const config_format_t state_format = {
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
sizeof(or_state_t),
OR_STATE_MAGIC,
STRUCT_OFFSET(or_state_t, _magic),
helper nodes are dead. long live entry nodes. (config options EntryNodes and StrictEntryNodes still not implemented.) svn:r5673
2005-12-28 10:07:31 +01:00
_state_abbrevs,
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
_state_vars,
(validate_fn_t)or_state_validate,
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
&state_extra_var,
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
};
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
/*
* Functions to read and write the global options pointer.
*/
/** Command-line and config-file options. */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
static or_options_t *global_options = NULL;
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
/** DOCDOC */
static or_options_t *global_default_options = NULL;
some fixes so far svn:r2880
2004-11-15 04:53:03 +01:00
/** Name of most recently read torrc file. */
Add a config-file GETINFO entry; fix a minor memory leak on some SAVECONF calls. svn:r4761
2005-08-10 20:05:20 +02:00
static char *torrc_fname = NULL;
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
/** DOCDOC */
static char *torrc_defaults_fname;
save and validate TorVersion in the state file. svn:r5694
2006-01-02 05:14:52 +01:00
/** Persistent serialized state. */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
static or_state_t *global_state = NULL;
Store options we got from commandline seperately from bare argv/argc svn:r13945
2008-03-10 13:41:40 +01:00
/** Configuration Options set by command line. */
static config_line_t *global_cmdline_options = NULL;
cleanups on r17500 svn:r17501
2008-12-07 02:34:45 +01:00
/** Contents of most recently read DirPortFrontPage file. */
New DirPortFrontPage option that takes an html file and publishes it as "/" on the DirPort. Now relay operators can provide a disclaimer without needin to set up a separate webserver. There's a sample disclaimer in contrib/tor-exit-notice.html. svn:r17500
2008-12-07 02:21:19 +01:00
static char *global_dirfrontpagecontents = NULL;
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
/** List of port_cfg_t for all configured ports. */
static smartlist_t *configured_ports = NULL;
New DirPortFrontPage option that takes an html file and publishes it as "/" on the DirPort. Now relay operators can provide a disclaimer without needin to set up a separate webserver. There's a sample disclaimer in contrib/tor-exit-notice.html. svn:r17500
2008-12-07 02:21:19 +01:00
/** Return the contents of our frontpage string, or NULL if not configured. */
const char *
get_dirportfrontpage(void)
{
return global_dirfrontpagecontents;
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Allocate an empty configuration object of a given format type. */
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
static void *
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_alloc(const config_format_t *fmt)
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
{
fix funny-looking assignment that crashes unit tests svn:r7027
2006-08-11 09:41:21 +02:00
void *opts = tor_malloc_zero(fmt->size);
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
*(uint32_t*)STRUCT_VAR_P(opts, fmt->magic_offset) = fmt->magic;
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
CHECK(fmt, opts);
return opts;
}
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
/** Return the currently configured options. */
or_options_t *
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
get_options_mutable(void)
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
{
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
tor_assert(global_options);
return global_options;
}
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
/** Returns the currently configured options */
const or_options_t *
get_options(void)
{
return get_options_mutable();
}
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
/** Change the current global options to contain <b>new_val</b> instead of
* their current value; take action based on the new value; free the old value
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
* as necessary. Returns 0 on success, -1 on failure.
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
*/
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
int
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
set_options(or_options_t *new_val, char **msg)
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
{
Emits CONF_CHANGED events whenever Tor's configuration values change.
2011-08-08 01:38:53 +02:00
int i;
smartlist_t *elements;
config_line_t *line;
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
or_options_t *old_options = global_options;
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
global_options = new_val;
stop freeing the new global options list whenever we update it. svn:r4853
2005-08-26 22:26:20 +02:00
/* Note that we pass the *old* options below, for comparison. It
* pulls the new options directly out of global_options. */
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
if (options_act_reversible(old_options, msg)<0) {
tor_assert(*msg);
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
global_options = old_options;
return -1;
}
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
if (options_act(old_options) < 0) { /* acting on the options failed. die. */
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
log_err(LD_BUG,
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
"Acting on config options left us in a broken state. Dying.");
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
exit(1);
}
Emits CONF_CHANGED events whenever Tor's configuration values change.
2011-08-08 01:38:53 +02:00
/* Issues a CONF_CHANGED event to notify controller of the change. If Tor is
* just starting up then the old_options will be undefined. */
if (old_options) {
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
elements = smartlist_new();
Emits CONF_CHANGED events whenever Tor's configuration values change.
2011-08-08 01:38:53 +02:00
for (i=0; options_format.vars[i].name; ++i) {
Remove a needless config_find_option
2011-08-17 16:46:22 +02:00
const config_var_t *var = &options_format.vars[i];
const char *var_name = var->name;
Don't compare LINELIST_S and OBSOLETE config options.
2011-08-16 03:28:38 +02:00
if (var->type == CONFIG_TYPE_LINELIST_S ||
var->type == CONFIG_TYPE_OBSOLETE) {
continue;
}
Remove a needless config_find_option
2011-08-17 16:46:22 +02:00
if (!option_is_same(&options_format, new_val, old_options, var_name)) {
line = get_assigned_option(&options_format, new_val, var_name, 1);
Emits CONF_CHANGED events whenever Tor's configuration values change.
2011-08-08 01:38:53 +02:00
if (line) {
for (; line; line = line->next) {
Refactor to do CONF_CHANGED event formatting inside control.c
2011-08-08 16:19:06 +02:00
smartlist_add(elements, line->key);
smartlist_add(elements, line->value);
Emits CONF_CHANGED events whenever Tor's configuration values change.
2011-08-08 01:38:53 +02:00
}
} else {
Fix a const warning in bug1692 code
2011-08-10 21:06:11 +02:00
smartlist_add(elements, (char*)options_format.vars[i].name);
Refactor to do CONF_CHANGED event formatting inside control.c
2011-08-08 16:19:06 +02:00
smartlist_add(elements, NULL);
Emits CONF_CHANGED events whenever Tor's configuration values change.
2011-08-08 01:38:53 +02:00
}
}
}
Refactor to do CONF_CHANGED event formatting inside control.c
2011-08-08 16:19:06 +02:00
control_event_conf_changed(elements);
Emits CONF_CHANGED events whenever Tor's configuration values change.
2011-08-08 01:38:53 +02:00
smartlist_free(elements);
}
Now that FOO_free(NULL) always works, remove checks before calling it.
2009-12-12 08:07:59 +01:00
config_free(&options_format, old_options);
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
return 0;
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
}
Add the first 8 bytes of the git commit digest to our versions. Note that unlike subversion revision numbers, it isn't meaningful to compare these for anything but equality. We define a sort-order anyway, in case one of these accidentally slips into a recommended-versions list.
2009-06-12 19:38:37 +02:00
extern const char tor_git_revision[]; /* from tor_main.c */
r12852@catbus: nickm | 2007-05-22 11:00:27 -0400 Use svn revisions consistently throughout all log messages. svn:r10291
2007-05-22 17:48:46 +02:00
Document most undocumented variables. svn:r17754
2008-12-23 18:56:31 +01:00
/** The version of this Tor process, as parsed. */
Convert instances of tor_snprintf+strdup into tor_asprintf These were found by looking for tor_snprintf() instances that were followed closely by tor_strdup(), though I probably converted some other snprintfs as well.
2012-01-11 20:02:59 +01:00
static char *the_tor_version = NULL;
Do not publish the "git-XXX" tag in server descriptors Instead, allow packagers to put a 'TOR_BUILD_TAG' field in the server descriptor to indicate a platform-specific value, if they need to. (According to weasel, this was his use for the git- tag previously.) This is part of 2988
2012-05-12 00:06:12 +02:00
static char *the_short_tor_version = NULL;
free another string, and the buffer freelists, on exit. svn:r10851
2007-07-17 11:26:45 +02:00
Document most undocumented variables. svn:r17754
2008-12-23 18:56:31 +01:00
/** Return the current Tor version. */
r12852@catbus: nickm | 2007-05-22 11:00:27 -0400 Use svn revisions consistently throughout all log messages. svn:r10291
2007-05-22 17:48:46 +02:00
const char *
get_version(void)
{
Convert instances of tor_snprintf+strdup into tor_asprintf These were found by looking for tor_snprintf() instances that were followed closely by tor_strdup(), though I probably converted some other snprintfs as well.
2012-01-11 20:02:59 +01:00
if (the_tor_version == NULL) {
Add the first 8 bytes of the git commit digest to our versions. Note that unlike subversion revision numbers, it isn't meaningful to compare these for anything but equality. We define a sort-order anyway, in case one of these accidentally slips into a recommended-versions list.
2009-06-12 19:38:37 +02:00
if (strlen(tor_git_revision)) {
Do not publish the "git-XXX" tag in server descriptors Instead, allow packagers to put a 'TOR_BUILD_TAG' field in the server descriptor to indicate a platform-specific value, if they need to. (According to weasel, this was his use for the git- tag previously.) This is part of 2988
2012-05-12 00:06:12 +02:00
tor_asprintf(&the_tor_version, "%s (git-%s)", get_short_version(),
tor_git_revision);
r12852@catbus: nickm | 2007-05-22 11:00:27 -0400 Use svn revisions consistently throughout all log messages. svn:r10291
2007-05-22 17:48:46 +02:00
} else {
Do not publish the "git-XXX" tag in server descriptors Instead, allow packagers to put a 'TOR_BUILD_TAG' field in the server descriptor to indicate a platform-specific value, if they need to. (According to weasel, this was his use for the git- tag previously.) This is part of 2988
2012-05-12 00:06:12 +02:00
the_tor_version = tor_strdup(get_short_version());
r12852@catbus: nickm | 2007-05-22 11:00:27 -0400 Use svn revisions consistently throughout all log messages. svn:r10291
2007-05-22 17:48:46 +02:00
}
}
Convert instances of tor_snprintf+strdup into tor_asprintf These were found by looking for tor_snprintf() instances that were followed closely by tor_strdup(), though I probably converted some other snprintfs as well.
2012-01-11 20:02:59 +01:00
return the_tor_version;
r12852@catbus: nickm | 2007-05-22 11:00:27 -0400 Use svn revisions consistently throughout all log messages. svn:r10291
2007-05-22 17:48:46 +02:00
}
Do not publish the "git-XXX" tag in server descriptors Instead, allow packagers to put a 'TOR_BUILD_TAG' field in the server descriptor to indicate a platform-specific value, if they need to. (According to weasel, this was his use for the git- tag previously.) This is part of 2988
2012-05-12 00:06:12 +02:00
/** Return the current Tor version, without any git tag. */
const char *
get_short_version(void)
{
if (the_short_tor_version == NULL) {
#ifdef TOR_BUILD_TAG
tor_asprintf(&the_short_tor_version, "%s (%s)", VERSION, TOR_BUILD_TAG);
#else
the_short_tor_version = tor_strdup(VERSION);
#endif
}
return the_short_tor_version;
}
Add country-code support to configured node lists to implement the ever-popular "no exits in Monaco" feature (ExcludeExitNodes {MC}). Also allow country codes and IP ranges in ExitNodes. (EntryNodes needs more work.) Based on code by Robert Hogan. Needs more testing. svn:r16966
2008-09-25 22:21:35 +02:00
/** Release additional memory allocated in options
*/
static void
or_options_free(or_options_t *options)
{
*_free functions now accept NULL Some *_free functions threw asserts when passed NULL. Now all of them accept NULL as input and perform no action when called that way. This gains us consistence for our free functions, and allows some code simplifications where an explicit null check is no longer necessary.
2009-09-28 16:37:01 +02:00
if (!options)
return;
Now that FOO_free(NULL) always works, remove checks before calling it.
2009-12-12 08:07:59 +01:00
routerset_free(options->_ExcludeExitNodesUnion);
Implement node-based router family code Also, make the NodeFamily option into a list of routersets. This lets us git rid of router_in_nickname_list (or whatever it was called) without porting it to work with nodes, and also lets people specify country codes and IP ranges in NodeFamily
2010-10-01 00:25:25 +02:00
if (options->NodeFamilySets) {
SMARTLIST_FOREACH(options->NodeFamilySets, routerset_t *,
rs, routerset_free(rs));
smartlist_free(options->NodeFamilySets);
}
put a _ before or_options_t elements that aren't configurable it's fine with me if we change the current convention, but we should actually decide to change it if we want to.
2012-04-01 21:59:38 +02:00
tor_free(options->_BridgePassword_AuthDigest);
Add country-code support to configured node lists to implement the ever-popular "no exits in Monaco" feature (ExcludeExitNodes {MC}). Also allow country codes and IP ranges in ExitNodes. (EntryNodes needs more work.) Based on code by Robert Hogan. Needs more testing. svn:r16966
2008-09-25 22:21:35 +02:00
config_free(&options_format, options);
}
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Release all memory and resources held by global configuration structures.
*/
Free even more things on shutdown. Temporarily move tor_free_all out from #ifdef so it gets tested more. svn:r3614
2005-02-11 02:26:47 +01:00
void
config_free_all(void)
{
Now that FOO_free(NULL) always works, remove checks before calling it.
2009-12-12 08:07:59 +01:00
or_options_free(global_options);
global_options = NULL;
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
or_options_free(global_default_options);
global_default_options = NULL;
Now that FOO_free(NULL) always works, remove checks before calling it.
2009-12-12 08:07:59 +01:00
config_free(&state_format, global_state);
global_state = NULL;
config_free_lines(global_cmdline_options);
global_cmdline_options = NULL;
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (configured_ports) {
SMARTLIST_FOREACH(configured_ports,
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
port_cfg_t *, p, tor_free(p));
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
smartlist_free(configured_ports);
configured_ports = NULL;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
}
Add a config-file GETINFO entry; fix a minor memory leak on some SAVECONF calls. svn:r4761
2005-08-10 20:05:20 +02:00
tor_free(torrc_fname);
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
tor_free(torrc_defaults_fname);
Convert instances of tor_snprintf+strdup into tor_asprintf These were found by looking for tor_snprintf() instances that were followed closely by tor_strdup(), though I probably converted some other snprintfs as well.
2012-01-11 20:02:59 +01:00
tor_free(the_tor_version);
New DirPortFrontPage option that takes an html file and publishes it as "/" on the DirPort. Now relay operators can provide a disclaimer without needin to set up a separate webserver. There's a sample disclaimer in contrib/tor-exit-notice.html. svn:r17500
2008-12-07 02:21:19 +01:00
tor_free(global_dirfrontpagecontents);
Free even more things on shutdown. Temporarily move tor_free_all out from #ifdef so it gets tested more. svn:r3614
2005-02-11 02:26:47 +01:00
}
Refactor the safe_str_*() API to make more sense. The new rule is: safe_str_X() means "this string is a piece of X information; make it safe to log." safe_str() on its own means "this string is a piece of who-knows-what; make it safe to log".
2009-12-15 23:23:36 +01:00
/** Make <b>address</b> -- a piece of information related to our operation as
* a client -- safe to log according to the settings in options->SafeLogging,
* and return it.
*
* (We return "[scrubbed]" if SafeLogging is "1", and address otherwise.)
first iteration of scrubbing sensitive strings from logs. also generally clean up log messages. svn:r4174
2005-05-03 12:04:08 +02:00
*/
const char *
Refactor the safe_str_*() API to make more sense. The new rule is: safe_str_X() means "this string is a piece of X information; make it safe to log." safe_str() on its own means "this string is a piece of who-knows-what; make it safe to log".
2009-12-15 23:23:36 +01:00
safe_str_client(const char *address)
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
{
r14018@Kushana: nickm | 2007-08-14 15:39:35 -0400 Resolve XXXX020s in config.c svn:r11108
2007-08-14 22:19:46 +02:00
tor_assert(address);
Cache the parsed value of SafeLogging as an enum.
2009-12-12 07:12:47 +01:00
if (get_options()->_SafeLogging == SAFELOG_SCRUB_ALL)
Allow SafeLogging to exclude client related information
2009-09-28 15:08:32 +02:00
return "[scrubbed]";
else
return address;
}
Refactor the safe_str_*() API to make more sense. The new rule is: safe_str_X() means "this string is a piece of X information; make it safe to log." safe_str() on its own means "this string is a piece of who-knows-what; make it safe to log".
2009-12-15 23:23:36 +01:00
/** Make <b>address</b> -- a piece of information of unspecified sensitivity
* -- safe to log according to the settings in options->SafeLogging, and
* return it.
*
* (We return "[scrubbed]" if SafeLogging is anything besides "0", and address
* otherwise.)
first iteration of scrubbing sensitive strings from logs. also generally clean up log messages. svn:r4174
2005-05-03 12:04:08 +02:00
*/
const char *
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
safe_str(const char *address)
{
r14018@Kushana: nickm | 2007-08-14 15:39:35 -0400 Resolve XXXX020s in config.c svn:r11108
2007-08-14 22:19:46 +02:00
tor_assert(address);
Cache the parsed value of SafeLogging as an enum.
2009-12-12 07:12:47 +01:00
if (get_options()->_SafeLogging != SAFELOG_SCRUB_NONE)
first iteration of scrubbing sensitive strings from logs. also generally clean up log messages. svn:r4174
2005-05-03 12:04:08 +02:00
return "[scrubbed]";
else
return address;
}
Refactor the safe_str_*() API to make more sense. The new rule is: safe_str_X() means "this string is a piece of X information; make it safe to log." safe_str() on its own means "this string is a piece of who-knows-what; make it safe to log".
2009-12-15 23:23:36 +01:00
/** Equivalent to escaped(safe_str_client(address)). See reentrancy note on
r8873@Kushana: nickm | 2006-09-21 14:38:22 -0400 Fix a bug: Remember, each call to escaped() replaces the value returned from the last call to escaped(). svn:r8438
2006-09-21 23:48:11 +02:00
* escaped(): don't use this outside the main thread, or twice in the same
* log statement. */
Add some functions to escape values from the network before sending them to the log. Use them everywhere except for routerinfo->plaftorm, routerinfo->contact_info, and rend*.c. (need sleep now) svn:r6087
2006-03-05 10:50:26 +01:00
const char *
Refactor the safe_str_*() API to make more sense. The new rule is: safe_str_X() means "this string is a piece of X information; make it safe to log." safe_str() on its own means "this string is a piece of who-knows-what; make it safe to log".
2009-12-15 23:23:36 +01:00
escaped_safe_str_client(const char *address)
Add some functions to escape values from the network before sending them to the log. Use them everywhere except for routerinfo->plaftorm, routerinfo->contact_info, and rend*.c. (need sleep now) svn:r6087
2006-03-05 10:50:26 +01:00
{
Cache the parsed value of SafeLogging as an enum.
2009-12-12 07:12:47 +01:00
if (get_options()->_SafeLogging == SAFELOG_SCRUB_ALL)
Allow SafeLogging to exclude client related information
2009-09-28 15:08:32 +02:00
return "[scrubbed]";
else
return escaped(address);
}
add 6697 (ircs) as longlivedport. svn:r8496
2006-09-25 07:24:43 +02:00
/** Equivalent to escaped(safe_str(address)). See reentrancy note on
r8873@Kushana: nickm | 2006-09-21 14:38:22 -0400 Fix a bug: Remember, each call to escaped() replaces the value returned from the last call to escaped(). svn:r8438
2006-09-21 23:48:11 +02:00
* escaped(): don't use this outside the main thread, or twice in the same
* log statement. */
Add some functions to escape values from the network before sending them to the log. Use them everywhere except for routerinfo->plaftorm, routerinfo->contact_info, and rend*.c. (need sleep now) svn:r6087
2006-03-05 10:50:26 +01:00
const char *
escaped_safe_str(const char *address)
{
Cache the parsed value of SafeLogging as an enum.
2009-12-12 07:12:47 +01:00
if (get_options()->_SafeLogging != SAFELOG_SCRUB_NONE)
Add some functions to escape values from the network before sending them to the log. Use them everywhere except for routerinfo->plaftorm, routerinfo->contact_info, and rend*.c. (need sleep now) svn:r6087
2006-03-05 10:50:26 +01:00
return "[scrubbed]";
else
return escaped(address);
}
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
/** Add the default directory authorities directly into the trusted dir list,
* but only add them insofar as they share bits with <b>type</b>. */
Leave options->DirServers alone -- if the user didn't specify any, just add the default ones directly to the trusted dirserver list. This fixes a bug where people running controllers would setconf or the equivalent, and Tor would start yelling at them about setting their own DirServer lines. svn:r5418
2005-11-18 03:47:09 +01:00
static void
Automated rename from authority_type_t to dirinfo_type_t We were already overloading this type to mean "a directory that can serve us X" in addition to "a directory that is an authority for X."
2010-11-08 20:27:36 +01:00
add_default_trusted_dir_authorities(dirinfo_type_t type)
Leave options->DirServers alone -- if the user didn't specify any, just add the default ones directly to the trusted dirserver list. This fixes a bug where people running controllers would setconf or the equivalent, and Tor would start yelling at them about setting their own DirServer lines. svn:r5418
2005-11-18 03:47:09 +01:00
{
make it so there is one place to edit for new dirservers, not two. svn:r6186
2006-03-18 23:37:27 +01:00
int i;
Leave options->DirServers alone -- if the user didn't specify any, just add the default ones directly to the trusted dirserver list. This fixes a bug where people running controllers would setconf or the equivalent, and Tor would start yelling at them about setting their own DirServer lines. svn:r5418
2005-11-18 03:47:09 +01:00
const char *dirservers[] = {
rotate keys for moria1 and gabelmoo
2010-01-19 20:12:39 +01:00
"moria1 orport=9101 no-v2 "
"v3ident=D586D18309DED4CD6D57C18FDB97EFA96D330566 "
"128.31.0.39:9131 9695 DFC3 5FFE B861 329B 9F1A B04C 4639 7020 CE31",
r19725@catbus: nickm | 2008-05-13 08:47:18 -0400 Forward-port: update authority keys affected by Debian OpenSSL bug (See CVE-2008-0166 or http://lists.debian.org/debian-security-announce/2008/msg00152.html ) svn:r14603
2008-05-13 14:47:27 +02:00
"tor26 v1 orport=443 v3ident=14C131DFC5C6F93646BE72FA1401C02A8DF2E8B4 "
add a v3 authority svn:r11542
2007-09-20 23:36:56 +02:00
"86.59.21.38:80 847B 1F85 0344 D787 6491 A548 92F9 0493 4E4E B85D",
forward-port r14421 svn:r14424
2008-04-23 02:31:20 +02:00
"dizum orport=443 v3ident=E8A9C45EDE6D711294FADF8E7951F4DE6CA56B58 "
"194.109.206.212:80 7EA6 EAD6 FD83 083C 538F 4403 8BBF A077 587D D755",
Move Tonga to an alternate IP address
2009-10-01 04:35:05 +02:00
"Tonga orport=443 bridge no-v2 82.94.251.203:80 "
Set up Tonga as the default bridge directory authority. svn:r11276
2007-08-26 00:02:55 +02:00
"4A0C CD2D DC79 9508 3D73 F5D6 6710 0C8A 5831 F16D",
Wrap long line; strlen("ides")<strlen("turtles").
2012-04-05 03:05:19 +02:00
"turtles orport=9090 no-v2 "
"v3ident=27B6B5996C426270A5C95488AA5BCEB6BCC86956 "
ides has become turtles, and gotten a new IP address As per ticket 5569
2012-04-05 01:51:04 +02:00
"76.73.17.194:9030 F397 038A DC51 3361 35E7 B80B D99C A384 4360 292B",
Change gabelmoo's IP address and ports.
2010-12-16 13:10:19 +01:00
"gabelmoo orport=443 no-v2 "
rotate keys for moria1 and gabelmoo
2010-01-19 20:12:39 +01:00
"v3ident=ED03BB616EB2F60BEC80151114BB25CEF515B226 "
Change gabelmoo's IP address and ports.
2010-12-16 13:10:19 +01:00
"212.112.245.170:80 F204 4413 DAC2 E02E 3D6B CF47 35A1 9BCA 1DE9 7281",
Set up dannenberg (run by CCC) as the sixth v3 directory authority. svn:r13231
2008-01-22 23:38:38 +01:00
"dannenberg orport=443 no-v2 "
"v3ident=585769C78764D58426B8B52B6651A5A71137189A "
new dannenberg address; make moria2's demise official.
2010-02-12 20:31:08 +01:00
"193.23.244.244:80 7BE6 83E6 5D48 1413 21C5 ED92 F075 C553 64AC 7123",
Set up urras as the seventh v3 directory authority.
2009-08-10 07:32:51 +02:00
"urras orport=80 no-v2 v3ident=80550987E1D626E3EBA5E5E75A458DE0626D088C "
"208.83.223.34:443 0AD3 FA88 4D18 F89E EA2D 89C0 1937 9E0E 7FD9 4417",
Add maatuska as eighth v3 directory authority.
2010-06-10 16:56:08 +02:00
"maatuska orport=80 no-v2 "
"v3ident=49015F787433103580E3B66A1707A00E60F2D15B "
new ip address for maatuska
2012-02-29 19:22:41 +01:00
"171.25.193.9:443 BD6A 8292 55CB 08E6 6FBE 7D37 4836 3586 E46B 3810",
make it so there is one place to edit for new dirservers, not two. svn:r6186
2006-03-18 23:37:27 +01:00
NULL
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
};
r17611@catbus: nickm | 2008-01-14 13:44:16 -0500 add some missing checks for failing return values. svn:r13130
2008-01-14 20:00:23 +01:00
for (i=0; dirservers[i]; i++) {
if (parse_dir_server_line(dirservers[i], type, 0)<0) {
log_err(LD_BUG, "Couldn't parse internal dirserver line %s",
dirservers[i]);
}
}
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
}
/** Look at all the config options for using alternate directory
* authorities, and make sure none of them are broken. Also, warn the
* user if we changed any dangerous ones.
*/
static int
validate_dir_authorities(or_options_t *options, or_options_t *old_options)
{
config_line_t *cl;
if (options->DirServers &&
(options->AlternateDirAuthority || options->AlternateBridgeAuthority ||
options->AlternateHSAuthority)) {
log_warn(LD_CONFIG,
"You cannot set both DirServers and Alternate*Authority.");
return -1;
}
/* do we want to complain to the user about being partitionable? */
if ((options->DirServers &&
(!old_options ||
!config_lines_eq(options->DirServers, old_options->DirServers))) ||
(options->AlternateDirAuthority &&
(!old_options ||
!config_lines_eq(options->AlternateDirAuthority,
old_options->AlternateDirAuthority)))) {
log_warn(LD_CONFIG,
"You have used DirServer or AlternateDirAuthority to "
"specify alternate directory authorities in "
"your configuration. This is potentially dangerous: it can "
"make you look different from all other Tor users, and hurt "
"your anonymity. Even if you've specified the same "
"authorities as Tor uses by default, the defaults could "
"change in the future. Be sure you know what you're doing.");
}
/* Now go through the four ways you can configure an alternate
* set of directory authorities, and make sure none are broken. */
for (cl = options->DirServers; cl; cl = cl->next)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
if (parse_dir_server_line(cl->value, NO_DIRINFO, 1)<0)
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
return -1;
for (cl = options->AlternateBridgeAuthority; cl; cl = cl->next)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
if (parse_dir_server_line(cl->value, NO_DIRINFO, 1)<0)
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
return -1;
for (cl = options->AlternateDirAuthority; cl; cl = cl->next)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
if (parse_dir_server_line(cl->value, NO_DIRINFO, 1)<0)
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
return -1;
for (cl = options->AlternateHSAuthority; cl; cl = cl->next)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
if (parse_dir_server_line(cl->value, NO_DIRINFO, 1)<0)
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
return -1;
return 0;
}
/** Look at all the config options and assign new dir authorities
* as appropriate.
*/
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
consider_adding_dir_authorities(const or_options_t *options,
const or_options_t *old_options)
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
{
config_line_t *cl;
int need_to_update =
but don't disable all the default v1 authorities quite yet svn:r12778
2007-12-12 05:46:16 +01:00
!smartlist_len(router_get_trusted_dir_servers()) || !old_options ||
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
!config_lines_eq(options->DirServers, old_options->DirServers) ||
!config_lines_eq(options->AlternateBridgeAuthority,
old_options->AlternateBridgeAuthority) ||
!config_lines_eq(options->AlternateDirAuthority,
old_options->AlternateDirAuthority) ||
!config_lines_eq(options->AlternateHSAuthority,
old_options->AlternateHSAuthority);
if (!need_to_update)
return 0; /* all done */
/* Start from a clean slate. */
clear_trusted_dir_servers();
if (!options->DirServers) {
/* then we may want some of the defaults */
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
dirinfo_type_t type = NO_DIRINFO;
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
if (!options->AlternateBridgeAuthority)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
type |= BRIDGE_DIRINFO;
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
if (!options->AlternateDirAuthority)
Code to make clients fetch and use microdescriptors for circuit building To turn this on, set UseMicrodescriptors to "1" (or "auto" if you want it on-if-you're-a-client). It should go auto-by-default once 0.2.3.1-alpha is released. Because of our node logic, directory caches will never use microdescriptors when they have the right routerinfo available.
2010-11-08 20:21:32 +01:00
type |= V1_DIRINFO | V2_DIRINFO | V3_DIRINFO | EXTRAINFO_DIRINFO |
MICRODESC_DIRINFO;
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
if (!options->AlternateHSAuthority)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
type |= HIDSERV_DIRINFO;
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
add_default_trusted_dir_authorities(type);
}
for (cl = options->DirServers; cl; cl = cl->next)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
if (parse_dir_server_line(cl->value, NO_DIRINFO, 0)<0)
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
return -1;
for (cl = options->AlternateBridgeAuthority; cl; cl = cl->next)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
if (parse_dir_server_line(cl->value, NO_DIRINFO, 0)<0)
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
return -1;
for (cl = options->AlternateDirAuthority; cl; cl = cl->next)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
if (parse_dir_server_line(cl->value, NO_DIRINFO, 0)<0)
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
return -1;
for (cl = options->AlternateHSAuthority; cl; cl = cl->next)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
if (parse_dir_server_line(cl->value, NO_DIRINFO, 0)<0)
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
return -1;
return 0;
Leave options->DirServers alone -- if the user didn't specify any, just add the default ones directly to the trusted dirserver list. This fixes a bug where people running controllers would setconf or the equivalent, and Tor would start yelling at them about setting their own DirServer lines. svn:r5418
2005-11-18 03:47:09 +01:00
}
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
/** Fetch the active option list, and take actions based on it. All of the
* things we do should survive being done repeatedly. If present,
* <b>old_options</b> contains the previous value of the options.
*
* Return 0 if all goes well, return -1 if things went badly.
*/
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
options_act_reversible(const or_options_t *old_options, char **msg)
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
{
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
smartlist_t *new_listeners = smartlist_new();
smartlist_t *replaced_listeners = smartlist_new();
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
static int libevent_initialized = 0;
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
or_options_t *options = get_options_mutable();
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
int running_tor = options->command == CMD_RUN_TOR;
int set_conn_limit = 0;
int r = -1;
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
int logs_marked = 0;
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
r13186@catbus: nickm | 2007-06-03 19:00:20 -0400 Bind ports before setuid/setgid. svn:r10473
2007-06-04 01:00:26 +02:00
/* Daemonize _first_, since we only want to open most of this stuff in
r14227@Kushana: nickm | 2007-08-27 11:33:28 -0400 Add a new ClientDNSRejectInternalAddresses option (default: on) to refuse to believe that any address can map to or from an internal address. This blocks some kinds of potential browser-based attacks, especially on hosts using DNSPort. Also clarify behavior in some comments. Backport candiate? svn:r11287
2007-08-27 17:33:58 +02:00
* the subprocess. Libevent bases can't be reliably inherited across
* processes. */
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
if (running_tor && options->RunAsDaemon) {
/* No need to roll back, since you can't change the value. */
start_daemon();
}
r13266@catbus: nickm | 2007-06-05 16:38:08 -0400 Patch from peter palfrader: control interface via unix domain socket svn:r10504
2007-06-05 22:54:49 +02:00
#ifndef HAVE_SYS_UN_H
Clean up the 2972 implementation a little
2011-04-23 02:57:53 +02:00
if (options->ControlSocket || options->ControlSocketsGroupWritable) {
*msg = tor_strdup("Unix domain sockets (ControlSocket) not supported "
"on this OS/with this build.");
goto rollback;
}
#else
if (options->ControlSocketsGroupWritable && !options->ControlSocket) {
*msg = tor_strdup("Setting ControlSocketGroupWritable without setting"
"a ControlSocket makes no sense.");
r13266@catbus: nickm | 2007-06-05 16:38:08 -0400 Patch from peter palfrader: control interface via unix domain socket svn:r10504
2007-06-05 22:54:49 +02:00
goto rollback;
}
#endif
r13189@catbus: nickm | 2007-06-03 19:08:01 -0400 Oops. Only bind ports and start libevent if we are actually running Tor. svn:r10474
2007-06-04 01:08:07 +02:00
if (running_tor) {
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
int n_ports=0;
r13189@catbus: nickm | 2007-06-03 19:08:01 -0400 Oops. Only bind ports and start libevent if we are actually running Tor. svn:r10474
2007-06-04 01:08:07 +02:00
/* We need to set the connection limit before we can open the listeners. */
r18269@catbus: nickm | 2008-02-20 17:28:24 -0500 Apply patch from Sebastian Hahn: stop imposing an arbitrary maximum on the number of file descriptors used for busy servers. Bug reported by Olaf Selke. svn:r13626
2008-02-20 23:28:26 +01:00
if (set_max_file_descriptors((unsigned)options->ConnLimit,
&options->_ConnLimit) < 0) {
r13189@catbus: nickm | 2007-06-03 19:08:01 -0400 Oops. Only bind ports and start libevent if we are actually running Tor. svn:r10474
2007-06-04 01:08:07 +02:00
*msg = tor_strdup("Problem with ConnLimit value. See logs for details.");
goto rollback;
}
set_conn_limit = 1;
r13186@catbus: nickm | 2007-06-03 19:00:20 -0400 Bind ports before setuid/setgid. svn:r10473
2007-06-04 01:00:26 +02:00
r13189@catbus: nickm | 2007-06-03 19:08:01 -0400 Oops. Only bind ports and start libevent if we are actually running Tor. svn:r10474
2007-06-04 01:08:07 +02:00
/* Set up libevent. (We need to do this before we can register the
* listeners as listeners.) */
if (running_tor && !libevent_initialized) {
Add a new option to enable/disable IOCP support
2010-09-28 20:01:45 +02:00
init_libevent(options);
r13189@catbus: nickm | 2007-06-03 19:08:01 -0400 Oops. Only bind ports and start libevent if we are actually running Tor. svn:r10474
2007-06-04 01:08:07 +02:00
libevent_initialized = 1;
}
r13186@catbus: nickm | 2007-06-03 19:00:20 -0400 Bind ports before setuid/setgid. svn:r10473
2007-06-04 01:00:26 +02:00
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
/* Adjust the port configuration so we can launch listeners. */
if (parse_ports(options, 0, msg, &n_ports)) {
Fix memory leak in options_act_reversible: fix Coverity CID 486,487
2011-10-06 18:54:34 +02:00
if (!*msg)
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
*msg = tor_strdup("Unexpected problem parsing port config");
Fix memory leak in options_act_reversible: fix Coverity CID 486,487
2011-10-06 18:54:34 +02:00
goto rollback;
}
Oops: we need to call parse_client_ports with validate_only==0 at least once
2011-07-20 19:35:02 +02:00
Consider hibernation before dropping privs Without this patch, Tor wasn't sure whether it would be hibernating or not, so it postponed opening listeners until after the privs had been dropped. This doesn't work so well for low ports. Bug was introduced in the fix for bug 2003. Fixes bug 4217, reported by Zax and katmagic. Thanks!
2011-10-11 02:40:06 +02:00
/* Set the hibernation state appropriately.*/
consider_hibernation(time(NULL));
r13189@catbus: nickm | 2007-06-03 19:08:01 -0400 Oops. Only bind ports and start libevent if we are actually running Tor. svn:r10474
2007-06-04 01:08:07 +02:00
/* Launch the listeners. (We do this before we setuid, so we can bind to
New 'DisableNetwork' option to prevent Tor from using the network Some controllers want this so they can mess with Tor's configuration for a while via the control port before actually letting Tor out of the house. We do this with a new DisableNetwork option, that prevents Tor from making any outbound connections or binding any non-control listeners. Additionally, it shuts down the same functionality as shuts down when we are hibernating, plus the code that launches directory downloads. To make sure I didn't miss anything, I added a clause straight to connection_connect, so that we won't even try to open an outbound socket when the network is disabled. In my testing, I made this an assert, but since I probably missed something, I've turned it into a BUG warning for testing.
2011-11-28 21:44:10 +01:00
* ports under 1024.) We don't want to rebind if we're hibernating. If
* networking is disabled, this will close all but the control listeners,
* but disable those. */
Bug #919: Don't rebind ports if we receive SIGHUP while hibernating.
2010-08-01 04:35:20 +02:00
if (!we_are_hibernating()) {
if (retry_all_listeners(replaced_listeners, new_listeners) < 0) {
*msg = tor_strdup("Failed to bind one of the listener ports.");
goto rollback;
}
r13189@catbus: nickm | 2007-06-03 19:08:01 -0400 Oops. Only bind ports and start libevent if we are actually running Tor. svn:r10474
2007-06-04 01:08:07 +02:00
}
New 'DisableNetwork' option to prevent Tor from using the network Some controllers want this so they can mess with Tor's configuration for a while via the control port before actually letting Tor out of the house. We do this with a new DisableNetwork option, that prevents Tor from making any outbound connections or binding any non-control listeners. Additionally, it shuts down the same functionality as shuts down when we are hibernating, plus the code that launches directory downloads. To make sure I didn't miss anything, I added a clause straight to connection_connect, so that we won't even try to open an outbound socket when the network is disabled. In my testing, I made this an assert, but since I probably missed something, I've turned it into a BUG warning for testing.
2011-11-28 21:44:10 +01:00
if (options->DisableNetwork) {
/* Aggressively close non-controller stuff, NOW */
log a notice when disablenetwork is set
2011-11-28 22:07:13 +01:00
log_notice(LD_NET, "DisableNetwork is set. Tor will not make or accept "
"non-control network connections. Shutting down all existing "
"connections.");
New 'DisableNetwork' option to prevent Tor from using the network Some controllers want this so they can mess with Tor's configuration for a while via the control port before actually letting Tor out of the house. We do this with a new DisableNetwork option, that prevents Tor from making any outbound connections or binding any non-control listeners. Additionally, it shuts down the same functionality as shuts down when we are hibernating, plus the code that launches directory downloads. To make sure I didn't miss anything, I added a clause straight to connection_connect, so that we won't even try to open an outbound socket when the network is disabled. In my testing, I made this an assert, but since I probably missed something, I've turned it into a BUG warning for testing.
2011-11-28 21:44:10 +01:00
connection_mark_all_noncontrol_connections();
}
r13186@catbus: nickm | 2007-06-03 19:00:20 -0400 Bind ports before setuid/setgid. svn:r10473
2007-06-04 01:00:26 +02:00
}
r17664@tombo: nickm | 2008-08-06 12:32:09 -0400 Patch from Christopher Davis: open /dev/pf before dropping privileges. Fixes bug 782. Backport candidate. svn:r16450
2008-08-06 18:32:17 +02:00
#if defined(HAVE_NET_IF_H) && defined(HAVE_NET_PFVAR_H)
/* Open /dev/pf before dropping privileges. */
if (options->TransPort) {
if (get_pf_socket() < 0) {
*msg = tor_strdup("Unable to open /dev/pf for transparent proxy.");
goto rollback;
}
}
#endif
Implement DisableAllSwap to avoid putting secret info in page files. This commit implements a new config option: 'DisableAllSwap' This option probably only works properly when Tor is started as root. We added two new functions: tor_mlockall() and tor_set_max_memlock(). tor_mlockall() attempts to mlock() all current and all future memory pages. For tor_mlockall() to work properly we set the process rlimits for memory to RLIM_INFINITY (and beyond) inside of tor_set_max_memlock(). We behave differently from mlockall() by only allowing tor_mlockall() to be called one single time. All other calls will result in a return code of 1. It is not possible to change DisableAllSwap while running. A sample configuration item was added to the torrc.complete.in config file. A new item in the man page for DisableAllSwap was added. Thanks to Moxie Marlinspike and Chris Palmer for their feedback on this patch. Please note that we make no guarantees about the quality of your OS and its mlock/mlockall implementation. It is possible that this will do nothing at all. It is also possible that you can ulimit the mlock properties of a given user such that root is not required. This has not been extensively tested and is unsupported. I have included some comments for possible ways we can handle this on win32.
2009-10-22 06:21:57 +02:00
/* Attempt to lock all current and future memory with mlockall() only once */
if (options->DisableAllSwap) {
if (tor_mlockall() == -1) {
*msg = tor_strdup("DisableAllSwap failure. Do you have proper "
"permissions?");
goto done;
}
}
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
/* Setuid/setgid as appropriate */
Patch from Jacob Appelbaum and me to make User option more robust, properly set supplementary groups, deprecated the Group option, and log more information on credential switching svn:r17200
2008-11-07 03:06:12 +01:00
if (options->User) {
if (switch_id(options->User) != 0) {
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
/* No need to roll back, since you can't change the value. */
Patch from Jacob Appelbaum and me to make User option more robust, properly set supplementary groups, deprecated the Group option, and log more information on credential switching svn:r17200
2008-11-07 03:06:12 +01:00
*msg = tor_strdup("Problem with User value. See logs for details.");
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
goto done;
}
}
/* Ensure data directory is private; create if possible. */
r17274@catbus: nickm | 2007-12-20 01:14:09 -0500 Don't try to create the datadir when we are only verifying the configuration or hashing a password. Resolves bug 540. svn:r12884
2007-12-20 07:15:09 +01:00
if (check_private_dir(options->DataDirectory,
Make ControlSocketsGroupWritable work with User. Original message from bug3393: check_private_dir() to ensure that ControlSocketsGroupWritable is safe to use. Unfortunately, check_private_dir() only checks against the currently running user… which can be root until privileges are dropped to the user and group configured by the User config option. The attached patch fixes the issue by adding a new effective_user argument to check_private_dir() and updating the callers. It might not be the best way to fix the issue, but it did in my tests. (Code by lunar; changelog by nickm)
2011-06-14 18:18:32 +02:00
running_tor ? CPD_CREATE : CPD_CHECK,
options->User)<0) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
"Couldn't access/create private data directory \"%s\"",
options->DataDirectory);
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
goto done;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
/* No need to roll back, since you can't change the value. */
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
}
Don't write ControlPorts to disk till after we setuid and make datadir Fix for bug 3747; bugfix on 0.2.2.26-beta.
2011-08-17 17:22:16 +02:00
/* Write control ports to disk as appropriate */
control_ports_write_to_file();
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
if (directory_caches_v2_dir_info(options)) {
Convert instances of tor_snprintf+strdup into tor_asprintf These were found by looking for tor_snprintf() instances that were followed closely by tor_strdup(), though I probably converted some other snprintfs as well.
2012-01-11 20:02:59 +01:00
char *fn = NULL;
tor_asprintf(&fn, "%s"PATH_SEPARATOR"cached-status",
If we have correct permissions on $datadir, we complain to stdout and fail to start. But dangerous permissions on $datadir/cached-status/ would cause us to open a log and complain there. Now complain to stdout and fail to start in both cases. Fixes bug 820, reported by seeess. svn:r16998
2008-09-29 12:09:05 +02:00
options->DataDirectory);
Make ControlSocketsGroupWritable work with User. Original message from bug3393: check_private_dir() to ensure that ControlSocketsGroupWritable is safe to use. Unfortunately, check_private_dir() only checks against the currently running user… which can be root until privileges are dropped to the user and group configured by the User config option. The attached patch fixes the issue by adding a new effective_user argument to check_private_dir() and updating the callers. It might not be the best way to fix the issue, but it did in my tests. (Code by lunar; changelog by nickm)
2011-06-14 18:18:32 +02:00
if (check_private_dir(fn, running_tor ? CPD_CREATE : CPD_CHECK,
options->User) < 0) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
If we have correct permissions on $datadir, we complain to stdout and fail to start. But dangerous permissions on $datadir/cached-status/ would cause us to open a log and complain there. Now complain to stdout and fail to start in both cases. Fixes bug 820, reported by seeess. svn:r16998
2008-09-29 12:09:05 +02:00
"Couldn't access/create private data directory \"%s\"", fn);
tor_free(fn);
goto done;
}
tor_free(fn);
}
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
/* Bail out at this point if we're not going to be a client or server:
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
* we don't run Tor itself. */
r13189@catbus: nickm | 2007-06-03 19:08:01 -0400 Oops. Only bind ports and start libevent if we are actually running Tor. svn:r10474
2007-06-04 01:08:07 +02:00
if (!running_tor)
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
goto commit;
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
mark_logs_temp(); /* Close current logs once new logs are open. */
logs_marked = 1;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
if (options_init_logs(options, 0)<0) { /* Configure the log(s) */
*msg = tor_strdup("Failed to init Log options. See logs for details.");
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
goto rollback;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
}
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
commit:
r = 0;
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
if (logs_marked) {
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
log_severity_list_t *severity =
tor_malloc_zero(sizeof(log_severity_list_t));
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
close_temp_logs();
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
add_callback_log(severity, control_event_logmsg);
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
control_adjust_event_log_severity();
r18639@catbus: nickm | 2008-03-07 20:11:48 -0500 Change semantics of add-a-log functions to copy severity setup: that is way less error-prone. Fix up config.c to act accordingly. svn:r13888
2008-03-08 02:11:54 +01:00
tor_free(severity);
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
}
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
SMARTLIST_FOREACH(replaced_listeners, connection_t *, conn,
{
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_notice(LD_NET, "Closing old %s on %s:%d",
conn_type_to_string(conn->type), conn->address, conn->port);
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
connection_close_immediate(conn);
connection_mark_for_close(conn);
});
goto done;
rollback:
r = -1;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
tor_assert(*msg);
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
if (logs_marked) {
rollback_log_changes();
control_adjust_event_log_severity();
}
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
if (set_conn_limit && old_options)
r18269@catbus: nickm | 2008-02-20 17:28:24 -0500 Apply patch from Sebastian Hahn: stop imposing an arbitrary maximum on the number of file descriptors used for busy servers. Bug reported by Olaf Selke. svn:r13626
2008-02-20 23:28:26 +01:00
set_max_file_descriptors((unsigned)old_options->ConnLimit,
&options->_ConnLimit);
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
SMARTLIST_FOREACH(new_listeners, connection_t *, conn,
{
r9025@Kushana: nickm | 2006-09-29 18:33:13 -0400 Differentiate more duplicated log entries svn:r8542
2006-09-30 00:33:40 +02:00
log_notice(LD_NET, "Closing partially-constructed listener %s on %s:%d",
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
conn_type_to_string(conn->type), conn->address, conn->port);
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
connection_close_immediate(conn);
connection_mark_for_close(conn);
});
done:
smartlist_free(new_listeners);
smartlist_free(replaced_listeners);
return r;
}
Resolve many DOCDOCs. svn:r17662
2008-12-17 23:58:20 +01:00
/** If we need to have a GEOIP ip-to-country map to run with our configured
* options, return 1 and set *<b>reason_out</b> to a description of why. */
Widen the conditions under which we whine about not having a geoip file to include "a country code was configured in a node list." svn:r16968
2008-09-25 23:06:32 +02:00
int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
options_need_geoip_info(const or_options_t *options, const char **reason_out)
Widen the conditions under which we whine about not having a geoip file to include "a country code was configured in a node list." svn:r16968
2008-09-25 23:06:32 +02:00
{
int bridge_usage =
options->BridgeRelay && options->BridgeRecordUsageByCountry;
int routerset_usage =
routerset_needs_geoip(options->EntryNodes) ||
routerset_needs_geoip(options->ExitNodes) ||
routerset_needs_geoip(options->ExcludeExitNodes) ||
routerset_needs_geoip(options->ExcludeNodes);
if (routerset_usage && reason_out) {
*reason_out = "We've been configured to use (or avoid) nodes in certain "
fix a spelling mistake in config.c for "contries"
2009-05-22 07:19:55 +02:00
"countries, and we need GEOIP information to figure out which ones they "
Widen the conditions under which we whine about not having a geoip file to include "a country code was configured in a node list." svn:r16968
2008-09-25 23:06:32 +02:00
"are.";
} else if (bridge_usage && reason_out) {
*reason_out = "We've been configured to see which countries can access "
"us as a bridge, and we need GEOIP information to tell which countries "
"clients are in.";
}
return bridge_usage || routerset_usage;
}
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
/** Return the bandwidthrate that we are going to report to the authorities
* based on the config options. */
Fix a signed/unsigned compile warning in 0.2.1.19
2009-07-29 00:34:35 +02:00
uint32_t
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
get_effective_bwrate(const or_options_t *options)
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
{
Fix signed/unsigned comparison warnings in get_effective_bw(rate|burst)
2009-07-30 15:15:07 +02:00
uint64_t bw = options->BandwidthRate;
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
if (bw > options->MaxAdvertisedBandwidth)
Fix signed/unsigned comparison warnings in get_effective_bw(rate|burst)
2009-07-30 15:15:07 +02:00
bw = options->MaxAdvertisedBandwidth;
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
if (options->RelayBandwidthRate > 0 && bw > options->RelayBandwidthRate)
Fix signed/unsigned comparison warnings in get_effective_bw(rate|burst)
2009-07-30 15:15:07 +02:00
bw = options->RelayBandwidthRate;
Cleaner fix for get_effective_bw(rate|burst), with comment on why it is ok.
2009-07-30 16:14:12 +02:00
/* ensure_bandwidth_cap() makes sure that this cast can't overflow. */
return (uint32_t)bw;
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
}
/** Return the bandwidthburst that we are going to report to the authorities
* based on the config options. */
Fix a signed/unsigned compile warning in 0.2.1.19
2009-07-29 00:34:35 +02:00
uint32_t
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
get_effective_bwburst(const or_options_t *options)
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
{
Fix signed/unsigned comparison warnings in get_effective_bw(rate|burst)
2009-07-30 15:15:07 +02:00
uint64_t bw = options->BandwidthBurst;
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
if (options->RelayBandwidthBurst > 0 && bw > options->RelayBandwidthBurst)
Fix signed/unsigned comparison warnings in get_effective_bw(rate|burst)
2009-07-30 15:15:07 +02:00
bw = options->RelayBandwidthBurst;
Cleaner fix for get_effective_bw(rate|burst), with comment on why it is ok.
2009-07-30 16:14:12 +02:00
/* ensure_bandwidth_cap() makes sure that this cast can't overflow. */
return (uint32_t)bw;
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
}
Simply initialize TLS context if DynamicDHGroups change. We used to do init_keys() if DynamicDHGroups changed after a HUP, so that the dynamic DH modulus was stored on the disk. Since we are now doing dynamic DH modulus storing in crypto.c, we can simply initialize the TLS context and be good with it. Introduce a new function router_initialize_tls_context() which initializes the TLS context and use it appropriately.
2011-11-24 06:40:02 +01:00
/** Return True if any changes from <b>old_options</b> to
* <b>new_options</b> needs us to refresh our TLS context. */
static int
options_transition_requires_fresh_tls_context(const or_options_t *old_options,
const or_options_t *new_options)
{
tor_assert(new_options);
if (!old_options)
return 0;
if ((old_options->DynamicDHGroups != new_options->DynamicDHGroups)) {
return 1;
}
return 0;
}
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
/** Fetch the active option list, and take actions based on it. All of the
* things we do should survive being done repeatedly. If present,
* <b>old_options</b> contains the previous value of the options.
require BandwidthRate to be at least 10kB/s svn:r3232
2004-12-31 22:47:54 +01:00
*
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
* Return 0 if all goes well, return -1 if it's time to die.
require BandwidthRate to be at least 10kB/s svn:r3232
2004-12-31 22:47:54 +01:00
*
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
* Note: We haven't moved all the "act on new configuration" logic
require BandwidthRate to be at least 10kB/s svn:r3232
2004-12-31 22:47:54 +01:00
* here yet. Some is still in do_hup() and other places.
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
*/
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
options_act(const or_options_t *old_options)
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
{
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *cl;
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
or_options_t *options = get_options_mutable();
Make unit tests (and others) run without launching listeners, creating subdirectories, and so on. svn:r4876
2005-08-27 01:22:27 +02:00
int running_tor = options->command == CMD_RUN_TOR;
[Forward-port ]Test and document last patch. svn:r6400
2006-04-18 05:51:18 +02:00
char *msg;
Reinit keys at the start of options_act(). Previously we did this nearer to the end (in the old_options && transition_affects_workers() block). But other stuff cares about keys being consistent with options... particularly anything which tries to access a key, which can die in assert_identity_keys_ok(). Fixes bug 3228; bugfix on 0.2.2.18-alpha.
2011-05-23 22:38:35 +02:00
const int transition_affects_workers =
old_options && options_transition_affects_workers(old_options, options);
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
Disallow disabling DisableDebuggerAttachment on runnning Tor Also, have tor_disable_debugger_attach() return a tristate of success/failure/don't-know-how , and only log appropriately.
2011-12-08 09:19:09 +01:00
/* disable ptrace and later, other basic debugging techniques */
{
/* Remember if we already disabled debugger attachment */
static int disabled_debugger_attach = 0;
/* Remember if we already warned about being configured not to disable
* debugger attachment */
static int warned_debugger_attach = 0;
Never disable debugger attachment for the unit tests
2012-03-22 12:50:44 +01:00
/* Don't disable debugger attachment when we're running the unit tests. */
if (options->DisableDebuggerAttachment && !disabled_debugger_attach &&
running_tor) {
Disallow disabling DisableDebuggerAttachment on runnning Tor Also, have tor_disable_debugger_attach() return a tristate of success/failure/don't-know-how , and only log appropriately.
2011-12-08 09:19:09 +01:00
int ok = tor_disable_debugger_attach();
if (warned_debugger_attach && ok == 1) {
log_notice(LD_CONFIG, "Disabled attaching debuggers for unprivileged "
"users.");
}
disabled_debugger_attach = (ok == 1);
} else if (!options->DisableDebuggerAttachment &&
!warned_debugger_attach) {
log_notice(LD_CONFIG, "Not disabling debugger attaching for "
"unprivileged users.");
warned_debugger_attach = 1;
}
New configuration option: DisableDebuggerAttachment If set to 1, Tor will attempt to prevent basic debugging attachment attempts by other processes. (Default: 1) Supports Mac OS X and Gnu/Linux. Sebastian provided useful feedback and refactoring suggestions. Signed-off-by: Jacob Appelbaum <jacob@appelbaum.net>
2011-05-30 17:06:51 +02:00
}
Move disable-debugger-attachment fn to compat where it belongs. Fix whitespace
2011-11-25 05:45:47 +01:00
Add a lockfile to the Tor data directory to avoid situations where two Tors start with the same datadir, or where a --list-fingerprints races with a server to create keys, or such. svn:r16722
2008-09-01 22:06:26 +02:00
if (running_tor && !have_lockfile()) {
if (try_locking(options, 1) < 0)
return -1;
}
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
if (consider_adding_dir_authorities(options, old_options) < 0)
return -1;
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
Warn loudly on startup and SIGHUP if Tor is built for a non-anonymous mode
2011-11-24 01:46:38 +01:00
#ifdef NON_ANONYMOUS_MODE_ENABLED
log(LOG_WARN, LD_GENERAL, "This copy of Tor was compiled to run in a "
"non-anonymous mode. It will provide NO ANONYMITY.");
#endif
Add a compile-time #define to control whether Tor runs in 'tor2web mode' The Tor2webMode torrc option is still required to run a Tor client in 'tor2web mode', but now it can't be turned on at runtime in a normal build of Tor. (And a tor2web build of Tor can't be used as a normal Tor client, so we don't have to worry as much about someone distributing packages with this particular pistol accessible to normal users.)
2011-11-24 01:58:15 +01:00
#ifdef ENABLE_TOR2WEB_MODE
if (!options->Tor2webMode) {
log_err(LD_CONFIG, "This copy of Tor was compiled to run in "
"'tor2web mode'. It can only be run with the Tor2webMode torrc "
"option enabled.");
return -1;
}
#else
if (options->Tor2webMode) {
log_err(LD_CONFIG, "This copy of Tor was not compiled to run in "
"'tor2web mode'. It cannot be run with the Tor2webMode torrc "
Add some more documentation
2011-11-24 10:28:38 +01:00
"option enabled. To enable Tor2webMode recompile with the "
"--enable-tor2webmode option.");
Add a compile-time #define to control whether Tor runs in 'tor2web mode' The Tor2webMode torrc option is still required to run a Tor client in 'tor2web mode', but now it can't be turned on at runtime in a normal build of Tor. (And a tor2web build of Tor can't be used as a normal Tor client, so we don't have to worry as much about someone distributing packages with this particular pistol accessible to normal users.)
2011-11-24 01:58:15 +01:00
return -1;
}
#endif
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
if (options->Bridges) {
Preserve bridge download status across SETCONF, HUP This code changes it so that we don't remove bridges immediately when we start re-parsing our configuration. Instead, we mark them all, and remove all the marked ones after re-parsing our bridge lines. As we add a bridge, we see if it's already in the list. If so, we just unmark it. This new behavior will lose the property we used to have that bridges were in bridge_list in the same order in which they appeared in the torrc. I took a quick look through the code, and I'm pretty sure we didn't actually depend on that anywhere. This is for bug 3019; it's a fix on 0.2.0.3-alpha.
2011-05-15 18:46:26 +02:00
mark_bridge_list();
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
for (cl = options->Bridges; cl; cl = cl->next) {
if (parse_bridge_line(cl->value, 0)<0) {
polish r9726-r9903 svn:r10182
2007-05-13 11:25:06 +02:00
log_warn(LD_BUG,
"Previously validated Bridge line could not be added!");
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
return -1;
}
}
Preserve bridge download status across SETCONF, HUP This code changes it so that we don't remove bridges immediately when we start re-parsing our configuration. Instead, we mark them all, and remove all the marked ones after re-parsing our bridge lines. As we add a bridge, we see if it's already in the list. If so, we just unmark it. This new behavior will lose the property we used to have that bridges were in bridge_list in the same order in which they appeared in the torrc. I took a quick look through the code, and I'm pretty sure we didn't actually depend on that anywhere. This is for bug 3019; it's a fix on 0.2.0.3-alpha.
2011-05-15 18:46:26 +02:00
sweep_bridge_list();
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
}
Make unit tests (and others) run without launching listeners, creating subdirectories, and so on. svn:r4876
2005-08-27 01:22:27 +02:00
if (running_tor && rend_config_services(options, 0)<0) {
polish r9726-r9903 svn:r10182
2007-05-13 11:25:06 +02:00
log_warn(LD_BUG,
r12074@catbus: nickm | 2007-03-04 15:11:43 -0500 Make all LD_BUG log messsages get prefixed with "Bug: ". Remove manually-generated "Bug: "s from log-messages. (Apparently, we remembered to add them about 40% of the time.) svn:r9733
2007-03-04 21:11:46 +01:00
"Previously validated hidden services line could not be added!");
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
return -1;
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
}
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
Next patch from Karsten: client-side configuration stuff for proposal 121. svn:r16510
2008-08-12 18:12:26 +02:00
if (running_tor && rend_parse_service_authorization(options, 0) < 0) {
log_warn(LD_BUG, "Previously validated client authorization for "
"hidden services could not be added!");
return -1;
}
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
/* Load state */
minor cleanup in options_act() we set running_tor to options->command == CMD_RUN_TOR once and used that in all but one place. Now we use running_tor in that place also. svn:r13819
2008-03-03 13:48:13 +01:00
if (! global_state && running_tor) {
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
if (or_state_load())
return -1;
r14425@Kushana: nickm | 2007-09-13 13:00:57 -0400 Do not load state when options->command is not RUN_TOR. (Resolves bug 499; backport candidate) svn:r11437
2007-09-13 19:01:08 +02:00
rep_hist_load_mtbf_data(time(NULL));
r14729@catbus: nickm | 2007-08-20 11:58:02 -0400 Trigger load and save of MTBF data. svn:r11219
2007-08-20 17:59:31 +02:00
}
Prepare circuitbuild.[ch] and config.[ch] for SIGHUPs. * Create mark/sweep functions for transports. * Create a transport_resolve_conflicts() function that tries to resolve conflicts when registering transports.
2011-09-11 20:28:47 +02:00
mark_transport_list();
pt_prepare_proxy_list_for_config_read();
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
if (options->ClientTransportPlugin) {
for (cl = options->ClientTransportPlugin; cl; cl = cl->next) {
if (parse_client_transport_line(cl->value, 0)<0) {
log_warn(LD_BUG,
"Previously validated ClientTransportPlugin line "
"could not be added!");
return -1;
}
}
}
if (options->ServerTransportPlugin) {
for (cl = options->ServerTransportPlugin; cl; cl = cl->next) {
if (parse_server_transport_line(cl->value, 0)<0) {
log_warn(LD_BUG,
"Previously validated ServerTransportPlugin line "
"could not be added!");
return -1;
}
}
}
Prepare circuitbuild.[ch] and config.[ch] for SIGHUPs. * Create mark/sweep functions for transports. * Create a transport_resolve_conflicts() function that tries to resolve conflicts when registering transports.
2011-09-11 20:28:47 +02:00
sweep_transport_list();
sweep_proxy_list();
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
Fix crash with "tor --list-fingerprint" (reported by seeess). svn:r9328
2007-01-11 00:48:24 +01:00
/* Bail out at this point if we're not going to be a client or server:
* we want to not fork, and to log stuff to stderr. */
only clear the bridge list if options->Bridges is set. this doesn't matter yet so we aren't adding them in other ways yet. svn:r12730
2007-12-09 04:38:57 +01:00
if (!running_tor)
Fix crash with "tor --list-fingerprint" (reported by seeess). svn:r9328
2007-01-11 00:48:24 +01:00
return 0;
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
/* Finish backgrounding the process */
simplify options_act()
2009-05-24 23:01:30 +02:00
if (options->RunAsDaemon) {
Resolve many XXXs and all DOCDOCs svn:r2755
2004-11-09 19:22:17 +01:00
/* We may be calling this for the n'th time (on SIGHUP), but it's safe. */
bugfix: chdir to your datadirectory at the *end* of the daemonize process, not the beginning. this was a problem because the first time you run tor, if your datadir isn't there, and you have runasdaemon set to 1, it will try to chdir to it before it tries to create it. oops. svn:r4033
2005-04-06 21:07:38 +02:00
finish_daemon(options->DataDirectory);
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
}
Make DynamicPrimes SIGHUP-able. Instead of passing the DynamicPrimes configuration option to crypto_global_init(), generate and set a new TLS DH prime when we read the torrc.
2011-11-23 23:39:46 +01:00
/* If needed, generate a new TLS DH prime according to the current torrc. */
Simplify DH prime generation logic some. This is just refactoring work here. The old logic was kind of convoluted, especially after the bug 5572 fix. We don't actually need to distinguish so many cases here. Dropping detection of the "!old_options || !old_options->DynamicDHGroups" case is fine because that's the same that we'd do for clients. Also add a changes file for bug 5572.
2012-04-08 01:07:53 +02:00
if (server_mode(options) && options->DynamicDHGroups) {
char *keydir = get_datadir_fname("keys");
if (check_private_dir(keydir, CPD_CREATE, options->User)) {
tor_free(keydir);
return -1;
}
tor_free(keydir);
if (!old_options || !old_options->DynamicDHGroups) {
char *fname = get_datadir_fname2("keys", "dynamic_dh_params");
crypto_set_tls_dh_prime(fname);
tor_free(fname);
Make DynamicPrimes SIGHUP-able. Instead of passing the DynamicPrimes configuration option to crypto_global_init(), generate and set a new TLS DH prime when we read the torrc.
2011-11-23 23:39:46 +01:00
}
Only bother with dynamic DH moduli if we are a bridge.
2011-11-25 17:44:43 +01:00
} else { /* clients don't need a dynamic DH prime. */
crypto_set_tls_dh_prime(NULL);
Make DynamicPrimes SIGHUP-able. Instead of passing the DynamicPrimes configuration option to crypto_global_init(), generate and set a new TLS DH prime when we read the torrc.
2011-11-23 23:39:46 +01:00
}
stop asserting at boot The patch for 3228 made us try to run init_keys() before we had loaded our state file, resulting in an assert inside init_keys. We had moved it too early in the function. Now it's later in the function, but still above the accounting calls.
2011-05-31 05:50:37 +02:00
/* We want to reinit keys as needed before we do much of anything else:
keys are important, and other things can depend on them. */
if (transition_affects_workers ||
(options->V3AuthoritativeDir && (!old_options ||
!old_options->V3AuthoritativeDir))) {
if (init_keys() < 0) {
log_warn(LD_BUG,"Error initializing keys; exiting");
return -1;
}
Simply initialize TLS context if DynamicDHGroups change. We used to do init_keys() if DynamicDHGroups changed after a HUP, so that the dynamic DH modulus was stored on the disk. Since we are now doing dynamic DH modulus storing in crypto.c, we can simply initialize the TLS context and be good with it. Introduce a new function router_initialize_tls_context() which initializes the TLS context and use it appropriately.
2011-11-24 06:40:02 +01:00
} else if (old_options &&
options_transition_requires_fresh_tls_context(old_options,
options)) {
if (router_initialize_tls_context() < 0) {
log_warn(LD_BUG,"Error initializing TLS context.");
return -1;
}
stop asserting at boot The patch for 3228 made us try to run init_keys() before we had loaded our state file, resulting in an assert inside init_keys. We had moved it too early in the function. Now it's later in the function, but still above the accounting calls.
2011-05-31 05:50:37 +02:00
}
Spell-check Tor.
2009-05-27 23:55:51 +02:00
/* Write our PID to the PID file. If we do not have write permissions we
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
* will log a warning */
simplify options_act()
2009-05-24 23:01:30 +02:00
if (options->PidFile)
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
write_pidfile(options->PidFile);
Add a new AddressMap directive to rewrite incoming socks addresses. Add a new TrackHostExits directive to trigger addressmaps for certain incoming socks addresses, for sites that break when your exit keeps changing. Redo the client-side dns cache so it's just an addressmap too. svn:r3641
2005-02-22 01:53:08 +01:00
/* Register addressmap directives */
config_register_addressmaps(options);
[forward-port] Implement an option, VirtualAddrMask, to set which addresses get handed out in response to mapaddress requests. Needs testing and docs! svn:r6398
2006-04-18 05:36:28 +02:00
parse_virtual_addr_network(options->VirtualAddrNetwork, 0, &msg);
Add a new AddressMap directive to rewrite incoming socks addresses. Add a new TrackHostExits directive to trigger addressmaps for certain incoming socks addresses, for sites that break when your exit keeps changing. Redo the client-side dns cache so it's just an addressmap too. svn:r3641
2005-02-22 01:53:08 +01:00
Rename exit_policy to addr_policy, since it gets used for SOCKS and directory connections too. Make all policies get validated in options_validate, and make SOCKS/directory policies get set in options_act. svn:r2819
2004-11-12 20:39:13 +01:00
/* Update address policies. */
r17910@catbus: nickm | 2008-02-05 15:36:29 -0500 Check for correctness of AuthDir* options in options_validate; check for possible bugs where options_validate() is happy but parse_policies_from_options() is sad. svn:r13384
2008-02-05 22:39:32 +01:00
if (policies_parse_from_options(options) < 0) {
/* This should be impossible, but let's be sure. */
log_warn(LD_BUG,"Error parsing already-validated policy options.");
return -1;
}
Rename exit_policy to addr_policy, since it gets used for SOCKS and directory connections too. Make all policies get validated in options_validate, and make SOCKS/directory policies get set in options_act. svn:r2819
2004-11-12 20:39:13 +01:00
backport candidate: - If we require CookieAuthentication but we fail to write the cookie file, we would warn but not exit, and end up in a state where no controller could authenticate. Now we exit. - If we require CookieAuthentication, stop generating a new cookie every time we change any piece of our config. svn:r11117
2007-08-15 17:26:14 +02:00
if (init_cookie_authentication(options->CookieAuthentication) < 0) {
log_warn(LD_CONFIG,"Error creating cookie authentication file.");
return -1;
}
Resolve a bunch of FIXME items; mark a lot more for attention; ask for clarification on some. Turn all XXXX008 ("showstopper for 0.0.8 release") items into XXXX009 or XXXX, since plainly they were not showstoppers for 0.0.8. Add/clean some docs. svn:r2808
2004-11-12 17:39:03 +01:00
Exit immediately if we can't monitor our owning controller process tor_process_monitor_new can't currently return NULL, but if it ever can, we want that to be an explicitly fatal error, without relying on the fact that monitor_owning_controller_process's chain of caller will exit if it fails.
2011-05-16 19:25:59 +02:00
monitor_owning_controller_process(options->OwningControllerProcess);
Implement __OwningControllerProcess option Implements part of feature 3049.
2011-05-15 17:23:04 +02:00
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
/* reload keys as needed for rendezvous services. */
if (rend_service_load_keys()<0) {
polish r9726-r9903 svn:r10182
2007-05-13 11:25:06 +02:00
log_warn(LD_GENERAL,"Error loading rendezvous service keys");
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
return -1;
}
Resolve some hibernating-related XXXX009s. svn:r2875
2004-11-14 23:21:23 +01:00
/* Set up accounting */
Implement weekly/monthly/daily accounting svn:r2936
2004-11-22 22:56:51 +01:00
if (accounting_parse_options(options, 0)<0) {
polish r9726-r9903 svn:r10182
2007-05-13 11:25:06 +02:00
log_warn(LD_CONFIG,"Error in accounting options");
Implement weekly/monthly/daily accounting svn:r2936
2004-11-22 22:56:51 +01:00
return -1;
}
fix a bug in configuring accounting in options_act() svn:r2881
2004-11-15 05:01:31 +01:00
if (accounting_is_enabled(options))
Resolve some hibernating-related XXXX009s. svn:r2875
2004-11-14 23:21:23 +01:00
configure_accounting(time(NULL));
Convert bufferevents to use rate-limiting. This requires the latest Git version of Libevent as of 24 March 2010. In the future, we'll just say it requires Libevent 2.0.5-alpha or later. Since Libevent doesn't yet support hierarchical rate limit groups, there isn't yet support for tracking relayed-bytes separately when using the bufferevent system. If a future version does add support for hierarchical buckets, we can add that back in.
2010-02-22 19:59:34 +01:00
#ifdef USE_BUFFEREVENTS
/* If we're using the bufferevents implementation and our rate limits
* changed, we need to tell the rate-limiting system about it. */
if (!old_options ||
old_options->BandwidthRate != options->BandwidthRate ||
old_options->BandwidthBurst != options->BandwidthBurst ||
old_options->RelayBandwidthRate != options->RelayBandwidthRate ||
old_options->RelayBandwidthBurst != options->RelayBandwidthBurst)
connection_bucket_init();
#endif
Optimize cell-ewma circuit priority algorithm. There are two big changes here: - We store active circuits in a priority queue for each or_conn, rather than doing a linear search over all the active circuits before we send each cell. - Rather than multiplying every circuit's cell-ewma by a decay factor every time we send a cell (thus normalizing the value of a current cell to 1.0 and a past cell to alpha^t), we instead only scale down the cell-ewma every tick (ten seconds atm), normalizing so that a cell sent at the start of the tick has value 1.0).
2009-12-12 06:49:48 +01:00
/* Change the cell EWMA settings */
Change interface for configuring cell ewma algorithm. The rule is now: take the value from the CircuitPriorityHalflife config option if it is set. If it zero, disable the cell_ewma algorithm. If it is set, use it to calculate the scaling factor. If it is not set, look for a CircPriorityHalflifeMsec parameter in the consensus networkstatus. If *that* is zero, then disable the cell_ewma algorithm; if it is set, use it to calculate the scaling factor. If it is not set at all, disable the algorithm.
2009-12-15 19:53:53 +01:00
cell_ewma_set_scale_factor(options, networkstatus_get_latest_consensus());
Optimize cell-ewma circuit priority algorithm. There are two big changes here: - We store active circuits in a priority queue for each or_conn, rather than doing a linear search over all the active circuits before we send each cell. - Rather than multiplying every circuit's cell-ewma by a decay factor every time we send a cell (thus normalizing the value of a current cell to 1.0 and a past cell to alpha^t), we instead only scale down the cell-ewma every tick (ten seconds atm), normalizing so that a cell sent at the start of the tick has value 1.0).
2009-12-12 06:49:48 +01:00
Do not use strcmp() to compare an http authenticator to its expected value This fixes a side-channel attack on the (fortunately unused!) BridgePassword option for bridge authorities. Fix for bug 5543; bugfix on 0.2.0.14-alpha.
2012-04-01 04:51:28 +02:00
/* Update the BridgePassword's hashed version as needed. We store this as a
* digest so that we can do side-channel-proof comparisons on it.
*/
if (options->BridgePassword) {
char *http_authenticator;
http_authenticator = alloc_http_authenticator(options->BridgePassword);
if (!http_authenticator) {
log_warn(LD_BUG, "Unable to allocate HTTP authenticator. Not setting "
"BridgePassword.");
return -1;
}
put a _ before or_options_t elements that aren't configurable it's fine with me if we change the current convention, but we should actually decide to change it if we want to.
2012-04-01 21:59:38 +02:00
options->_BridgePassword_AuthDigest = tor_malloc(DIGEST256_LEN);
crypto_digest256(options->_BridgePassword_AuthDigest,
Do not use strcmp() to compare an http authenticator to its expected value This fixes a side-channel attack on the (fortunately unused!) BridgePassword option for bridge authorities. Fix for bug 5543; bugfix on 0.2.0.14-alpha.
2012-04-01 04:51:28 +02:00
http_authenticator, strlen(http_authenticator),
DIGEST_SHA256);
tor_free(http_authenticator);
}
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
/* Check for transitions that need action. */
if (old_options) {
When TrackExitHosts changes, remove all no-longer-valid mappings This bug couldn't happen when TrackExitHosts changed in torrc, since the SIGHUP to reload the torrc would clear out all the transient addressmap entries before. But if you used SETCONF to change TrackExitHosts, old entries would be left alone: that's a bug, and so this is a bugfix on Tor 0.1.0.1-rc.
2011-05-13 22:22:58 +02:00
int revise_trackexithosts = 0;
Handle transitions in Automap*, VirtualAddrNetwork correctly Previously, if they changed in torrc during a SIGHUP, all was well, since we would just clear all transient entries from the addrmap thanks to bug 1345. But if you changed them from the controller, Tor would leave old mappings in place. The VirtualAddrNetwork bug has been here since 0.1.1.19-rc; the AutomapHosts* bug has been here since 0.2.0.1-alpha.
2011-05-13 22:59:31 +02:00
int revise_automap_entries = 0;
Abandon circs if the user changes Exclude*Nodes If ExcludeNodes or ExcludeExitNodes changes on a config reload, mark and discard all our origin circuits.
2009-09-20 03:03:49 +02:00
if ((options->UseEntryGuards && !old_options->UseEntryGuards) ||
discard circuits when we change our bridge configuration otherwise we might reuse circuits from the previous configuration, which could be bad depending on the user's situation
2011-05-18 02:51:47 +02:00
options->UseBridges != old_options->UseBridges ||
(options->UseBridges &&
!config_lines_eq(options->Bridges, old_options->Bridges)) ||
Simplify calls to routerset_equal The routerset_equal function explicitly handles NULL inputs, so there's no need to check inputs for NULL before calling it. Also fix a bug in routerset_equal where a non-NULL routerset with no entries didn't get counted as equal to a NULL routerset. This was untriggerable, I think, but potentially annoying down the road.
2011-04-04 01:13:36 +02:00
!routerset_equal(old_options->ExcludeNodes,options->ExcludeNodes) ||
!routerset_equal(old_options->ExcludeExitNodes,
options->ExcludeExitNodes) ||
!routerset_equal(old_options->EntryNodes, options->EntryNodes) ||
!routerset_equal(old_options->ExitNodes, options->ExitNodes) ||
Be more willing to use an unsuitable circuit for exit. Specifically, there are two cases: a) are we willing to start a new circuit at a node not in your ExitNodes config option, and b) are we willing to make use of a circuit that's already established but has an unsuitable exit. Now we discard all your circuits when you set ExitNodes, so the only way you could end up with an exit circuit that ends at an unsuitable place is if we explicitly ran out of exit nodes, StrictNodes was 0, and we built this circuit to solve a stream that needs solving. Fixes bug in dc322931, which would ignore the just-built circuit because it has an unsuitable exit.
2009-09-20 03:52:57 +02:00
options->StrictNodes != old_options->StrictNodes) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_info(LD_CIRC,
discard circuits when we change our bridge configuration otherwise we might reuse circuits from the previous configuration, which could be bad depending on the user's situation
2011-05-18 02:51:47 +02:00
"Changed to using entry guards or bridges, or changed "
"preferred or excluded node lists. "
"Abandoning previous circuits.");
On sighup, if usehelpernodes changed to 1, use new circuits. svn:r5203
2005-10-06 01:02:40 +02:00
circuit_mark_all_unused_circs();
circuit_expire_all_dirty_circs();
When TrackExitHosts changes, remove all no-longer-valid mappings This bug couldn't happen when TrackExitHosts changed in torrc, since the SIGHUP to reload the torrc would clear out all the transient addressmap entries before. But if you used SETCONF to change TrackExitHosts, old entries would be left alone: that's a bug, and so this is a bugfix on Tor 0.1.0.1-rc.
2011-05-13 22:22:58 +02:00
revise_trackexithosts = 1;
On sighup, if usehelpernodes changed to 1, use new circuits. svn:r5203
2005-10-06 01:02:40 +02:00
}
When TrackExitHosts changes, remove all no-longer-valid mappings This bug couldn't happen when TrackExitHosts changed in torrc, since the SIGHUP to reload the torrc would clear out all the transient addressmap entries before. But if you used SETCONF to change TrackExitHosts, old entries would be left alone: that's a bug, and so this is a bugfix on Tor 0.1.0.1-rc.
2011-05-13 22:22:58 +02:00
if (!smartlist_strings_eq(old_options->TrackHostExits,
options->TrackHostExits))
revise_trackexithosts = 1;
if (revise_trackexithosts)
When there is a transition in permitted nodes, apply it to trackexithosts map IOW, if we were using TrackExitHosts, and we added an excluded node or removed a node from exitnodes, we wouldn't actually remove the mapping that points us at the new node. Also, note with an XXX022 comment a place that I think we are looking at the wrong string.
2011-04-04 01:43:47 +02:00
addressmap_clear_excluded_trackexithosts(options);
When TrackExitHosts changes, remove all no-longer-valid mappings This bug couldn't happen when TrackExitHosts changed in torrc, since the SIGHUP to reload the torrc would clear out all the transient addressmap entries before. But if you used SETCONF to change TrackExitHosts, old entries would be left alone: that's a bug, and so this is a bugfix on Tor 0.1.0.1-rc.
2011-05-13 22:22:58 +02:00
Fixup whitespace issues from 3122 commit
2011-05-16 02:12:01 +02:00
if (!options->AutomapHostsOnResolve) {
if (old_options->AutomapHostsOnResolve)
revise_automap_entries = 1;
} else {
Handle transitions in Automap*, VirtualAddrNetwork correctly Previously, if they changed in torrc during a SIGHUP, all was well, since we would just clear all transient entries from the addrmap thanks to bug 1345. But if you changed them from the controller, Tor would leave old mappings in place. The VirtualAddrNetwork bug has been here since 0.1.1.19-rc; the AutomapHosts* bug has been here since 0.2.0.1-alpha.
2011-05-13 22:59:31 +02:00
if (!smartlist_strings_eq(old_options->AutomapHostsSuffixes,
options->AutomapHostsSuffixes))
revise_automap_entries = 1;
else if (!opt_streq(old_options->VirtualAddrNetwork,
options->VirtualAddrNetwork))
revise_automap_entries = 1;
On sighup, if usehelpernodes changed to 1, use new circuits. svn:r5203
2005-10-06 01:02:40 +02:00
}
Handle transitions in Automap*, VirtualAddrNetwork correctly Previously, if they changed in torrc during a SIGHUP, all was well, since we would just clear all transient entries from the addrmap thanks to bug 1345. But if you changed them from the controller, Tor would leave old mappings in place. The VirtualAddrNetwork bug has been here since 0.1.1.19-rc; the AutomapHosts* bug has been here since 0.2.0.1-alpha.
2011-05-13 22:59:31 +02:00
if (revise_automap_entries)
addressmap_clear_invalid_automaps(options);
Only delay counting bridge users if we were a relay before
2010-07-08 14:29:04 +02:00
/* How long should we delay counting bridge stats after becoming a bridge?
* We use this so we don't count people who used our bridge thinking it is
* a relay. If you change this, don't forget to change the log message
delay geoip stats even more for relays-turned-bridges
2010-08-18 22:12:44 +02:00
* below. It's 4 hours (the time it takes to stop being used by clients)
* plus some extra time for clock skew. */
#define RELAY_BRIDGE_STATS_DELAY (6 * 60 * 60)
Only delay counting bridge users if we were a relay before
2010-07-08 14:29:04 +02:00
When our bridge status changes, forget old geoip data. This fixes bug 932, where all of our usage totals for users when we were a regular server would get included in the extrainfo document.
2009-05-12 21:35:53 +02:00
if (! bool_eq(options->BridgeRelay, old_options->BridgeRelay)) {
Only delay counting bridge users if we were a relay before
2010-07-08 14:29:04 +02:00
int was_relay = 0;
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
if (options->BridgeRelay) {
Only delay counting bridge users if we were a relay before
2010-07-08 14:29:04 +02:00
time_t int_start = time(NULL);
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (config_lines_eq(old_options->ORPort, options->ORPort)) {
Only delay counting bridge users if we were a relay before
2010-07-08 14:29:04 +02:00
int_start += RELAY_BRIDGE_STATS_DELAY;
was_relay = 1;
}
geoip_bridge_stats_init(int_start);
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
log_info(LD_CONFIG, "We are acting as a bridge now. Starting new "
delay geoip stats even more for relays-turned-bridges
2010-08-18 22:12:44 +02:00
"GeoIP stats interval%s.", was_relay ? " in 6 "
Only delay counting bridge users if we were a relay before
2010-07-08 14:29:04 +02:00
"hours from now" : "");
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
} else {
geoip_bridge_stats_term();
log_info(LD_GENERAL, "We are no longer acting as a bridge. "
"Forgetting GeoIP stats.");
}
When our bridge status changes, forget old geoip data. This fixes bug 932, where all of our usage totals for users when we were a regular server would get included in the extrainfo document.
2009-05-12 21:35:53 +02:00
}
Reinit keys at the start of options_act(). Previously we did this nearer to the end (in the old_options && transition_affects_workers() block). But other stuff cares about keys being consistent with options... particularly anything which tries to access a key, which can die in assert_identity_keys_ok(). Fixes bug 3228; bugfix on 0.2.2.18-alpha.
2011-05-23 22:38:35 +02:00
if (transition_affects_workers) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_info(LD_GENERAL,
"Worker-related options changed. Rotating workers.");
Fix assert for relay/bridge state change When we added support for separate client tls certs on bridges in a2bb0bfdd5 we forgot to correctly initialize this when changing from relay to bridge or vice versa while Tor is running. Fix that by always initializing keys when the state changes. Fixes bug 2433.
2011-01-25 15:28:58 +01:00
when you tried to set orport through the controller, it would trigger an assert because you hadn't made your onion keys yet you wanted to launch cpu workers. now init keys first, reset uptime, and tell the user that we're starting the reachability testing. svn:r5835
2006-01-17 05:01:32 +01:00
if (server_mode(options) && !server_mode(old_options)) {
r11724@Kushana: nickm | 2006-12-28 14:22:35 -0500 Refactor and unify my-ip-addr-changed logic. Make change in IP address or in nameservers reset and relaunch DNS hijacking tests. svn:r9200
2006-12-28 22:29:20 +01:00
ip_address_changed(0);
Rename has_completed_circuit to can_complete_circuit Also redocument it. Related to #1362.
2010-09-22 07:52:57 +02:00
if (can_complete_circuit || !any_predicted_circuits(time(NULL)))
when you tried to set orport through the controller, it would trigger an assert because you hadn't made your onion keys yet you wanted to launch cpu workers. now init keys first, reset uptime, and tell the user that we're starting the reachability testing. svn:r5835
2006-01-17 05:01:32 +01:00
inform_testing_reachability();
}
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
cpuworkers_rotate();
r12122@Kushana: nickm | 2007-02-02 10:41:39 -0500 Fail when we are unable to parse the nameserver configuration. svn:r9475
2007-02-02 19:58:04 +01:00
if (dns_reset())
return -1;
r12080@catbus: nickm | 2007-03-04 21:40:55 -0500 Remove dnsworkers and related code. there goes another 550 lines of code. svn:r9736
2007-03-05 03:40:58 +01:00
} else {
r12122@Kushana: nickm | 2007-02-02 10:41:39 -0500 Fail when we are unable to parse the nameserver configuration. svn:r9475
2007-02-02 19:58:04 +01:00
if (dns_reset())
return -1;
r8608@Kushana: nickm | 2006-08-27 16:57:47 -0400 Make it possible to change nameserver options while Tor is running. svn:r8255
2006-08-28 05:15:55 +02:00
}
r14017@Kushana: nickm | 2007-08-14 15:13:14 -0400 Enable more code for v3 voting timing information. Also, spec clarification. svn:r11107
2007-08-14 22:19:40 +02:00
reinit per-conn token buckets on config or consensus change
2010-08-15 10:01:42 +02:00
if (options->PerConnBWRate != old_options->PerConnBWRate ||
options->PerConnBWBurst != old_options->PerConnBWBurst)
connection_or_update_token_buckets(get_connection_array(), options);
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
}
r15530@tombo: nickm | 2007-12-17 16:54:03 -0500 First wodge of geoip code so bridges can figure out which countries are blocking them. svn:r12845
2007-12-17 23:44:11 +01:00
/* Maybe load geoip file */
r17231@catbus: nickm | 2007-12-18 16:21:55 -0500 Document and clean-up geoip code; give it some unit tests. svn:r12856
2007-12-18 22:27:08 +01:00
if (options->GeoIPFile &&
((!old_options || !opt_streq(old_options->GeoIPFile, options->GeoIPFile))
r15530@tombo: nickm | 2007-12-17 16:54:03 -0500 First wodge of geoip code so bridges can figure out which countries are blocking them. svn:r12845
2007-12-17 23:44:11 +01:00
|| !geoip_is_loaded())) {
Downlgrade tweak, and answer lots of XXX021s. No actual code fixes in this patch. svn:r17686
2008-12-18 17:11:24 +01:00
/* XXXX Don't use this "<default>" junk; make our filename options
On win32, default to looking for the geoip file in the same directory as torrc. This is a dumb hack; it should turn into a general mechanism. svn:r14796
2008-05-29 03:22:30 +02:00
* understand prefixes somehow. -NM */
Triage the XXX022 and XXX021 comments remaining in the code Remove some, postpone others, leave some alone. Now the only remaining XXX022s are ones that seem important to fix or investigate.
2011-03-25 21:01:16 +01:00
/* XXXX023 Reload GeoIPFile on SIGHUP. -NM */
On win32, default to looking for the geoip file in the same directory as torrc. This is a dumb hack; it should turn into a general mechanism. svn:r14796
2008-05-29 03:22:30 +02:00
char *actual_fname = tor_strdup(options->GeoIPFile);
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifdef _WIN32
On win32, default to looking for the geoip file in the same directory as torrc. This is a dumb hack; it should turn into a general mechanism. svn:r14796
2008-05-29 03:22:30 +02:00
if (!strcmp(actual_fname, "<default>")) {
const char *conf_root = get_windows_conf_root();
tor_free(actual_fname);
Convert instances of tor_malloc+tor_snprintf into tor_asprintf These were found by looking for tor_snprintf() instances that were preceeded closely by tor_malloc(), though I probably converted some more snprintfs as well. (In every case, make sure that the length variable (if any) is removed, renamed, or lowered, so that anything else that might have assumed a longer buffer doesn't exist.)
2012-01-11 20:02:59 +01:00
tor_asprintf(&actual_fname, "%s\\geoip", conf_root);
On win32, default to looking for the geoip file in the same directory as torrc. This is a dumb hack; it should turn into a general mechanism. svn:r14796
2008-05-29 03:22:30 +02:00
}
#endif
forward-port r14934 svn:r14936
2008-06-04 11:02:25 +02:00
geoip_load_file(actual_fname, options);
On win32, default to looking for the geoip file in the same directory as torrc. This is a dumb hack; it should turn into a general mechanism. svn:r14796
2008-05-29 03:22:30 +02:00
tor_free(actual_fname);
r15530@tombo: nickm | 2007-12-17 16:54:03 -0500 First wodge of geoip code so bridges can figure out which countries are blocking them. svn:r12845
2007-12-17 23:44:11 +01:00
}
Some tweaks to statistics. Changes to directory request statistics: - Rename GEOIP statistics to DIRREQ statistics, because they now include more than only GeoIP-based statistics, whereas other statistics are GeoIP-dependent, too. - Rename output file from geoip-stats to dirreq-stats. - Add new config option DirReqStatistics that is required to measure directory request statistics. - Clean up ChangeLog. Also ensure that entry guards statistics have access to a local GeoIP database.
2009-07-15 16:32:40 +02:00
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
if (options->CellStatistics || options->DirReqStatistics ||
Add metric on uni/bidirectional connection usage.
2010-08-02 15:06:14 +02:00
options->EntryStatistics || options->ExitPortStatistics ||
Add new stats type: descriptor fetch stats This is used for the bridge authority currently, to get a better intuition on how many descriptors are actually fetched from it and how many fetches happen in total. Implements ticket 4200.
2011-10-07 02:39:48 +02:00
options->ConnDirectionStatistics ||
options->BridgeAuthoritativeDir) {
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
time_t now = time(NULL);
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
int print_notice = 0;
Don't log about stats when running as a client without geoip Completely disable stats if we aren't running as a relay. We won't collect any anyway, so setting up the infrastructure for them and logging about them is wrong. This also removes a confusing log message that clients without a geoip db would have seen. Fixes bug 4353.
2011-11-19 00:50:03 +01:00
/* If we aren't acting as a server, we can't collect stats anyway. */
if (!server_mode(options)) {
options->CellStatistics = 0;
options->DirReqStatistics = 0;
options->EntryStatistics = 0;
options->ExitPortStatistics = 0;
}
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
if ((!old_options || !old_options->CellStatistics) &&
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
options->CellStatistics) {
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
rep_hist_buffer_stats_init(now);
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
print_notice = 1;
}
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
if ((!old_options || !old_options->DirReqStatistics) &&
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
options->DirReqStatistics) {
if (geoip_is_loaded()) {
geoip_dirreq_stats_init(now);
print_notice = 1;
} else {
Disable stats requiring geoip info if we have none In other parts of the code we will otherwise attempt to collect these statistics, and that will lead to crashes.
2010-12-22 08:34:41 +01:00
options->DirReqStatistics = 0;
Don't tell Tor client users about missing geoip They don't need the geoip file for stats, so a missing geoipfile is not a big issue. Also make the log message a bit friendlier. Fixes bug 2496.
2011-02-06 00:25:33 +01:00
/* Don't warn Tor clients, they don't use statistics */
if (options->ORPort)
log_notice(LD_CONFIG, "Configured to measure directory request "
"statistics, but no GeoIP database found. "
"Please specify a GeoIP database using the "
"GeoIPFile option.");
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
}
}
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
if ((!old_options || !old_options->EntryStatistics) &&
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
options->EntryStatistics && !should_record_bridge_info(options)) {
if (geoip_is_loaded()) {
geoip_entry_stats_init(now);
print_notice = 1;
} else {
Disable stats requiring geoip info if we have none In other parts of the code we will otherwise attempt to collect these statistics, and that will lead to crashes.
2010-12-22 08:34:41 +01:00
options->EntryStatistics = 0;
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
log_notice(LD_CONFIG, "Configured to measure entry node "
Don't tell Tor client users about missing geoip They don't need the geoip file for stats, so a missing geoipfile is not a big issue. Also make the log message a bit friendlier. Fixes bug 2496.
2011-02-06 00:25:33 +01:00
"statistics, but no GeoIP database found. "
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
"Please specify a GeoIP database using the "
Don't tell Tor client users about missing geoip They don't need the geoip file for stats, so a missing geoipfile is not a big issue. Also make the log message a bit friendlier. Fixes bug 2496.
2011-02-06 00:25:33 +01:00
"GeoIPFile option.");
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
}
}
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
if ((!old_options || !old_options->ExitPortStatistics) &&
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
options->ExitPortStatistics) {
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
rep_hist_exit_stats_init(now);
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
print_notice = 1;
}
Call the new config option ConnDirectionStatistics.
2010-08-15 14:58:35 +02:00
if ((!old_options || !old_options->ConnDirectionStatistics) &&
Merge remote branch 'karsten/dirreq-stats-default' Conflicts: src/or/config.c
2010-12-21 22:01:00 +01:00
options->ConnDirectionStatistics) {
Add metric on uni/bidirectional connection usage.
2010-08-02 15:06:14 +02:00
rep_hist_conn_stats_init(now);
Merge remote branch 'karsten/dirreq-stats-default' Conflicts: src/or/config.c
2010-12-21 22:01:00 +01:00
}
Don't initialize desc stats for non-bridgedirauth nodes Also make sure that calling rep_hist_note_desc_served() while stats aren't initialized just returns. Bug spotted by SwissTorHelp. Thanks!
2011-10-24 12:51:00 +02:00
if ((!old_options || !old_options->BridgeAuthoritativeDir) &&
options->BridgeAuthoritativeDir) {
Add new stats type: descriptor fetch stats This is used for the bridge authority currently, to get a better intuition on how many descriptors are actually fetched from it and how many fetches happen in total. Implements ticket 4200.
2011-10-07 02:39:48 +02:00
rep_hist_desc_stats_init(now);
print_notice = 1;
}
Turn on directory request statistics by default. Change the default values for collecting directory request statistics and inlcuding them in extra-info descriptors to 1. Don't break if we are configured to collect directory request or entry statistics and don't have a GeoIP database. Instead, print out a notice and skip initializing the affected statistics code.
2010-11-09 14:18:00 +01:00
if (print_notice)
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
log_notice(LD_CONFIG, "Configured to measure statistics. Look for "
"the *-stats files that will first be written to the "
"data directory in 24 hours from now.");
}
if (old_options && old_options->CellStatistics &&
!options->CellStatistics)
rep_hist_buffer_stats_term();
if (old_options && old_options->DirReqStatistics &&
!options->DirReqStatistics)
geoip_dirreq_stats_term();
if (old_options && old_options->EntryStatistics &&
!options->EntryStatistics)
geoip_entry_stats_term();
if (old_options && old_options->ExitPortStatistics &&
!options->ExitPortStatistics)
rep_hist_exit_stats_term();
Call the new config option ConnDirectionStatistics.
2010-08-15 14:58:35 +02:00
if (old_options && old_options->ConnDirectionStatistics &&
!options->ConnDirectionStatistics)
Add metric on uni/bidirectional connection usage.
2010-08-02 15:06:14 +02:00
rep_hist_conn_stats_term();
Add new stats type: descriptor fetch stats This is used for the bridge authority currently, to get a better intuition on how many descriptors are actually fetched from it and how many fetches happen in total. Implements ticket 4200.
2011-10-07 02:39:48 +02:00
if (old_options && old_options->BridgeAuthoritativeDir &&
!options->BridgeAuthoritativeDir)
rep_hist_desc_stats_term();
Allow enabling or disabling *Statistics while Tor is running. With this patch we stop scheduling when we should write statistics using a single timestamp in run_scheduled_events(). Instead, we remember when a statistics interval starts separately for each statistic type in geoip.c and rephist.c. Every time run_scheduled_events() tries to write stats to disk, it learns when it should schedule the next such attempt. This patch also enables all statistics to be stopped and restarted at a later time. This patch comes with a few refactorings, some of which were not easily doable without the patch.
2010-06-21 18:52:46 +02:00
Finish implementing config's EntryNodes and StrictEntryNodes option. The logging is way verbose for now. svn:r5683
2005-12-31 07:32:57 +01:00
/* Check if we need to parse and add the EntryNodes config option. */
bugfix: only try to use the EntryNodes config option if it is set. svn:r5701
2006-01-02 06:14:21 +01:00
if (options->EntryNodes &&
(!old_options ||
If EntryNodes and ExcludeNodes overlap, obey ExcludeNodes.
2009-10-18 00:54:20 +02:00
!routerset_equal(old_options->EntryNodes,options->EntryNodes) ||
!routerset_equal(old_options->ExcludeNodes,options->ExcludeNodes)))
Finish implementing config's EntryNodes and StrictEntryNodes option. The logging is way verbose for now. svn:r5683
2005-12-31 07:32:57 +01:00
entry_nodes_should_be_added();
Track whether descriptor is dirty/uploaded. When any options are set, mark it dirty. Once a minute, regenerate and upload the server descriptor if it is dirty. svn:r2832
2004-11-13 17:53:48 +01:00
/* Since our options changed, we might need to regenerate and upload our
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
* server descriptor.
Track whether descriptor is dirty/uploaded. When any options are set, mark it dirty. Once a minute, regenerate and upload the server descriptor if it is dirty. svn:r2832
2004-11-13 17:53:48 +01:00
*/
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
if (!old_options ||
options_transition_affects_descriptor(old_options, options))
log the reason for publishing a new relay descriptor now we have a better chance of hunting down the root cause of bug 1810.
2011-05-20 05:36:20 +02:00
mark_my_descriptor_dirty("config change");
Track whether descriptor is dirty/uploaded. When any options are set, mark it dirty. Once a minute, regenerate and upload the server descriptor if it is dirty. svn:r2832
2004-11-13 17:53:48 +01:00
r16032@catbus: nickm | 2007-10-22 11:56:53 -0400 When our directory status or our v3 authority status changes, reschedule operations as appropriate. (Fixes some xxxx020 items) svn:r12102
2007-10-22 18:32:04 +02:00
/* We may need to reschedule some directory stuff if our status changed. */
r16051@catbus: nickm | 2007-10-22 18:53:53 -0400 fix a dumb bug in r12102. svn:r12111
2007-10-23 00:54:28 +02:00
if (old_options) {
if (authdir_mode_v3(options) && !authdir_mode_v3(old_options))
dirvote_recalculate_timing(options, time(NULL));
new config option FetchDirInfoEarly for tup/ioerror and the tordnsel svn:r12896
2007-12-21 07:08:00 +01:00
if (!bool_eq(directory_fetches_dir_info_early(options),
directory_fetches_dir_info_early(old_options)) ||
!bool_eq(directory_fetches_dir_info_later(options),
directory_fetches_dir_info_later(old_options))) {
r16051@catbus: nickm | 2007-10-22 18:53:53 -0400 fix a dumb bug in r12102. svn:r12111
2007-10-23 00:54:28 +02:00
/* Make sure update_router_have_min_dir_info gets called. */
router_dir_info_changed();
/* We might need to download a new consensus status later or sooner than
* we had expected. */
update_consensus_networkstatus_fetch_time(time(NULL));
}
r16032@catbus: nickm | 2007-10-22 11:56:53 -0400 When our directory status or our v3 authority status changes, reschedule operations as appropriate. (Fixes some xxxx020 items) svn:r12102
2007-10-22 18:32:04 +02:00
}
Spell-check Tor.
2009-05-27 23:55:51 +02:00
/* Load the webpage we're going to serve every time someone asks for '/' on
New DirPortFrontPage option that takes an html file and publishes it as "/" on the DirPort. Now relay operators can provide a disclaimer without needin to set up a separate webserver. There's a sample disclaimer in contrib/tor-exit-notice.html. svn:r17500
2008-12-07 02:21:19 +01:00
our DirPort. */
tor_free(global_dirfrontpagecontents);
if (options->DirPortFrontPage) {
global_dirfrontpagecontents =
read_file_to_str(options->DirPortFrontPage, 0, NULL);
Small whitespace fix to properly format if () {} statement. svn:r17503
2008-12-07 02:51:56 +01:00
if (!global_dirfrontpagecontents) {
This helps return a better error message when the file supplied to DirPortFrontPage is missing. svn:r17502
2008-12-07 02:48:30 +01:00
log_warn(LD_CONFIG,
"DirPortFrontPage file '%s' not found. Continuing anyway.",
options->DirPortFrontPage);
}
New DirPortFrontPage option that takes an html file and publishes it as "/" on the DirPort. Now relay operators can provide a disclaimer without needin to set up a separate webserver. There's a sample disclaimer in contrib/tor-exit-notice.html. svn:r17500
2008-12-07 02:21:19 +01:00
}
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
return 0;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
}
/*
* Functions to parse config options
*/
Document new config code svn:r2596
2004-10-27 05:08:04 +02:00
/** If <b>option</b> is an official abbreviation for a longer option,
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
* return the longer option. Otherwise return <b>option</b>.
* If <b>command_line</b> is set, apply all abbreviations. Otherwise, only
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
* apply abbreviations that work for the config file and the command line.
* If <b>warn_obsolete</b> is set, warn about deprecated names. */
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
static const char *
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
expand_abbrev(const config_format_t *fmt, const char *option, int command_line,
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
int warn_obsolete)
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
{
int i;
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
if (! fmt->abbrevs)
return option;
for (i=0; fmt->abbrevs[i].abbreviated; ++i) {
Spell-check Tor.
2009-05-27 23:55:51 +02:00
/* Abbreviations are case insensitive. */
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
if (!strcasecmp(option,fmt->abbrevs[i].abbreviated) &&
(command_line || !fmt->abbrevs[i].commandline_only)) {
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
if (warn_obsolete && fmt->abbrevs[i].warn) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG,
"The configuration option '%s' is deprecated; "
"use '%s' instead.",
fmt->abbrevs[i].abbreviated,
fmt->abbrevs[i].full);
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
}
Let our config abbreviations rewrite more than once
2009-09-18 02:32:42 +02:00
/* Keep going through the list in case we want to rewrite it more.
* (We could imagine recursing here, but I don't want to get the
* user into an infinite loop if we craft our list wrong.) */
option = fmt->abbrevs[i].full;
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
}
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
}
return option;
}
Implement RedirectExit. svn:r2550
2004-10-17 03:57:34 +02:00
bugfix: if you gave tor an odd number of command-line arguments, we were silently ignoring the last one. now we complain and fail. svn:r4988
2005-09-10 06:40:27 +02:00
/** Helper: Read a list of configuration options from the command line.
* If successful, put them in *<b>result</b> and return 0, and return
* -1 and leave *<b>result</b> alone. */
static int
config_get_commandlines(int argc, char **argv, config_line_t **result)
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
{
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *front = NULL;
config_line_t **new = &front;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
char *s;
int i = 1;
bugfix: if you gave tor an odd number of command-line arguments, we were silently ignoring the last one. now we complain and fail. svn:r4988
2005-09-10 06:40:27 +02:00
while (i < argc) {
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
unsigned command = CONFIG_LINE_NORMAL;
int want_arg = 1;
Implement two flavors of authentication for control connections: one for trusted FS, one for untrusted FS. svn:r2664
2004-11-03 20:49:03 +01:00
if (!strcmp(argv[i],"-f") ||
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
!strcmp(argv[i],"--defaults-torrc") ||
Implement two flavors of authentication for control connections: one for trusted FS, one for untrusted FS. svn:r2664
2004-11-03 20:49:03 +01:00
!strcmp(argv[i],"--hash-password")) {
i += 2; /* command-line option with argument. ignore them. */
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
continue;
make tor --verify-config closer to working svn:r6137
2006-03-12 21:46:00 +01:00
} else if (!strcmp(argv[i],"--list-fingerprint") ||
r11666@catbus: nickm | 2007-02-06 13:17:24 -0500 Implement an --ignore-missing-torrc option svn:r9501
2007-02-06 19:36:21 +01:00
!strcmp(argv[i],"--verify-config") ||
r16322@catbus: nickm | 2007-11-01 00:26:15 -0400 Implement --quiet. svn:r12306
2007-11-01 05:38:43 +01:00
!strcmp(argv[i],"--ignore-missing-torrc") ||
Add --hush switch. New --hush command-line option similar to --quiet. While --quiet disables all logging to the console on startup, --hush limits the output to messages of warning and error severity. svn:r14222
2008-03-27 18:25:49 +01:00
!strcmp(argv[i],"--quiet") ||
!strcmp(argv[i],"--hush")) {
tor --list-fingerprint to print fingerprint and exit svn:r2627
2004-10-30 21:18:37 +02:00
i += 1; /* command-line option. ignore it. */
Implement two flavors of authentication for control connections: one for trusted FS, one for untrusted FS. svn:r2664
2004-11-03 20:49:03 +01:00
continue;
patch from edmanm to make "-nt-service" command-line work too svn:r9349
2007-01-15 09:52:01 +01:00
} else if (!strcmp(argv[i],"--nt-service") ||
!strcmp(argv[i],"-nt-service")) {
forward-port NT service patch svn:r3565
2005-02-05 22:40:25 +01:00
i += 1;
continue;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
}
r16322@catbus: nickm | 2007-11-01 00:26:15 -0400 Implement --quiet. svn:r12306
2007-11-01 05:38:43 +01:00
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
*new = tor_malloc_zero(sizeof(config_line_t));
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
s = argv[i];
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
Disallow command line keywords with more than two dashes as prefix. This might help fix cid 422, where coverity fails to notice that argv strings are null-escaped.
2009-10-27 17:50:24 +01:00
/* Each keyword may be prefixed with one or two dashes. */
if (*s == '-')
s++;
if (*s == '-')
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
s++;
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
/* Figure out the command, if any. */
if (*s == '+') {
s++;
command = CONFIG_LINE_APPEND;
} else if (*s == '/') {
s++;
command = CONFIG_LINE_CLEAR;
/* A 'clear' command has no argument. */
want_arg = 0;
}
if (want_arg && i == argc-1) {
log_warn(LD_CONFIG,"Command-line option '%s' with no value. Failing.",
argv[i]);
config_free_lines(front);
return -1;
}
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
(*new)->key = tor_strdup(expand_abbrev(&options_format, s, 1, 1));
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
(*new)->value = want_arg ? tor_strdup(argv[i+1]) : tor_strdup("");
(*new)->command = command;
Do not reverse command-line configuration options; Use new minimal-listener-close code; Add code to canonicalize configuration names. svn:r4529
2005-07-11 19:35:36 +02:00
(*new)->next = NULL;
Spell-check Tor.
2009-05-27 23:55:51 +02:00
log(LOG_DEBUG, LD_CONFIG, "command line: parsed keyword '%s', value '%s'",
Do not reverse command-line configuration options; Use new minimal-listener-close code; Add code to canonicalize configuration names. svn:r4529
2005-07-11 19:35:36 +02:00
(*new)->key, (*new)->value);
new = &((*new)->next);
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
i += want_arg ? 2 : 1;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
}
bugfix: if you gave tor an odd number of command-line arguments, we were silently ignoring the last one. now we complain and fail. svn:r4988
2005-09-10 06:40:27 +02:00
*result = front;
return 0;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
}
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
/** Helper: allocate a new configuration option mapping 'key' to 'val',
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
* append it to *<b>lst</b>. */
static void
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_append(config_line_t **lst,
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
const char *key,
const char *val)
Add code to configure hidden services, parse configuration, generate keys and service IDs, and store/load them from disk svn:r1410
2004-03-31 23:35:23 +02:00
{
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *newline;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
Make linelists always overridden by the command line This starts an effort to refactor torrc handling code to make it easier to live with. It makes it possible to override exit policies from the command line, and possible to override (rather than append to) socksport lists from the command line. It'll be necessary to make a "base" torrc implementation work at all.
2011-11-28 00:11:50 +01:00
newline = tor_malloc_zero(sizeof(config_line_t));
Add code to configure hidden services, parse configuration, generate keys and service IDs, and store/load them from disk svn:r1410
2004-03-31 23:35:23 +02:00
newline->key = tor_strdup(key);
newline->value = tor_strdup(val);
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
newline->next = NULL;
while (*lst)
lst = &((*lst)->next);
(*lst) = newline;
Add code to configure hidden services, parse configuration, generate keys and service IDs, and store/load them from disk svn:r1410
2004-03-31 23:35:23 +02:00
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
/** Helper: parse the config string and strdup into key/value
* strings. Set *result to the list, or NULL if parsing the string
Stop using separate defaults for no-config-file and empty-config-file svn:r2329
2004-09-03 00:08:36 +02:00
* failed. Return 0 on success, -1 on failure. Warn and ignore any
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
* misformatted lines.
*
* If <b>extended</b> is set, then treat keys beginning with / and with + as
* indicating "clear" and "append" respectively. */
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
int
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
config_get_lines(const char *string, config_line_t **result, int extended)
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
{
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *list = NULL, **next;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
char *k, *v;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
next = &list;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
do {
Fix numerous memory leaks: some were almost impossible to trigger, and some almost inevitable. svn:r16779
2008-09-05 22:52:15 +02:00
k = v = NULL;
r15786@tombo: nickm | 2008-01-02 01:11:51 -0500 Push the strdups used for parsing configuration lines into parse_line_from_string(). This will make it easier to parse more complex value formats, which in turn will help fix bug 557 svn:r13020
2008-01-02 07:59:12 +01:00
string = parse_config_line_from_str(string, &k, &v);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
if (!string) {
config_free_lines(list);
Fix numerous memory leaks: some were almost impossible to trigger, and some almost inevitable. svn:r16779
2008-09-05 22:52:15 +02:00
tor_free(k);
tor_free(v);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
return -1;
}
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
if (k && v) {
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
unsigned command = CONFIG_LINE_NORMAL;
if (extended) {
if (k[0] == '+') {
char *k_new = tor_strdup(k+1);
tor_free(k);
k = k_new;
command = CONFIG_LINE_APPEND;
} else if (k[0] == '/') {
char *k_new = tor_strdup(k+1);
tor_free(k);
k = k_new;
tor_free(v);
v = tor_strdup("");
command = CONFIG_LINE_CLEAR;
}
}
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
/* This list can get long, so we keep a pointer to the end of it
other trivial tweaks svn:r12614
2007-11-30 19:53:14 +01:00
* rather than using config_line_append over and over and getting
* n^2 performance. */
Make linelists always overridden by the command line This starts an effort to refactor torrc handling code to make it easier to live with. It makes it possible to override exit policies from the command line, and possible to override (rather than append to) socksport lists from the command line. It'll be necessary to make a "base" torrc implementation work at all.
2011-11-28 00:11:50 +01:00
*next = tor_malloc_zero(sizeof(config_line_t));
r15786@tombo: nickm | 2008-01-02 01:11:51 -0500 Push the strdups used for parsing configuration lines into parse_line_from_string(). This will make it easier to parse more complex value formats, which in turn will help fix bug 557 svn:r13020
2008-01-02 07:59:12 +01:00
(*next)->key = k;
(*next)->value = v;
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
(*next)->next = NULL;
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
(*next)->command = command;
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
next = &((*next)->next);
r15786@tombo: nickm | 2008-01-02 01:11:51 -0500 Push the strdups used for parsing configuration lines into parse_line_from_string(). This will make it easier to parse more complex value formats, which in turn will help fix bug 557 svn:r13020
2008-01-02 07:59:12 +01:00
} else {
tor_free(k);
tor_free(v);
Stop reversing and re-reversing config lines: keep them in order all the time. Removes needless confusion and fixes bug #50. svn:r3305
2005-01-05 07:05:37 +01:00
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
} while (*string);
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
*result = list;
make free_options() and init_options() use the big array. also remove obsolete config_assign_defaults(). svn:r2672
2004-11-04 09:26:34 +01:00
return 0;
Initial revision svn:r2
2002-06-27 00:45:49 +02:00
}
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
/**
* Free all the configuration lines on the linked list <b>front</b>.
*/
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
void
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_free_lines(config_line_t *front)
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
{
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *tmp;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
while (front) {
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
tmp = front;
front = tmp->next;
prefer tor_free to free plus complain more loudly when we fail to parse a dir we just fetched svn:r2401
2004-09-29 08:52:36 +02:00
tor_free(tmp->key);
tor_free(tmp->value);
tor_free(tmp);
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
}
}
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
/** As config_find_option, but return a non-const pointer. */
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
static config_var_t *
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_find_option_mutable(config_format_t *fmt, const char *key)
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
{
int i;
fix an edge case in parsing config options (thanks weasel) svn:r3486
2005-02-01 01:05:57 +01:00
size_t keylen = strlen(key);
Re-normalize whitespace; make "make check-spaces" happy svn:r3566
2005-02-05 22:42:46 +01:00
if (!keylen)
Spell-check Tor.
2009-05-27 23:55:51 +02:00
return NULL; /* if they say "--" on the command line, it's not an option */
Document new config code svn:r2596
2004-10-27 05:08:04 +02:00
/* First, check for an exact (case-insensitive) match */
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
for (i=0; fmt->vars[i].name; ++i) {
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
if (!strcasecmp(key, fmt->vars[i].name)) {
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
return &fmt->vars[i];
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
}
Document new config code svn:r2596
2004-10-27 05:08:04 +02:00
}
/* If none, check for an abbreviated match */
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
for (i=0; fmt->vars[i].name; ++i) {
if (!strncasecmp(key, fmt->vars[i].name, keylen)) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG, "The abbreviation '%s' is deprecated. "
"Please use '%s' instead",
key, fmt->vars[i].name);
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
return &fmt->vars[i];
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
}
}
fix bug reported by gozu: if we get a linelist or linelist_s config option from the torrc and it has no value, warn and skip rather than silently resetting it to its default. svn:r6125
2006-03-11 19:40:33 +01:00
/* Okay, unrecognized option */
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
return NULL;
}
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
/** If <b>key</b> is a configuration option, return the corresponding const
* config_var_t. Otherwise, if <b>key</b> is a non-standard abbreviation,
* warn, and return the corresponding const config_var_t. Otherwise return
* NULL.
*/
static const config_var_t *
config_find_option(const config_format_t *fmt, const char *key)
{
return config_find_option_mutable((config_format_t*)fmt, key);
}
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
/** Return the number of option entries in <b>fmt</b>. */
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_count_options(const config_format_t *fmt)
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
{
int i;
for (i=0; fmt->vars[i].name; ++i)
;
return i;
}
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
/*
* Functions to assign config options.
*/
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
/** <b>c</b>-\>key is known to be a real key. Update <b>options</b>
* with <b>c</b>-\>value and return 0, or return -1 if bad value.
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
*
* Called from config_assign_line() and option_reset().
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
*/
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_assign_value(const config_format_t *fmt, or_options_t *options,
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
config_line_t *c, char **msg)
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
{
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
int i, ok;
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *var;
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
void *lvalue;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
CHECK(fmt, options);
var = config_find_option(fmt, c->key);
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
tor_assert(var);
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
lvalue = STRUCT_VAR_P(options, var->var_offset);
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
Normalize space: add one between every control keyword and control clause. svn:r3003
2004-11-28 10:05:49 +01:00
switch (var->type) {
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
"(Socks|Control|etc)Port auto" now tells Tor to open an arbitrary port This is the major part of the implementation for trac issue 3076.
2011-05-02 21:05:10 +02:00
case CONFIG_TYPE_PORT:
if (!strcasecmp(c->value, "auto")) {
*(int *)lvalue = CFG_AUTO_PORT;
break;
}
/* fall through */
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
case CONFIG_TYPE_UINT:
"(Socks|Control|etc)Port auto" now tells Tor to open an arbitrary port This is the major part of the implementation for trac issue 3076.
2011-05-02 21:05:10 +02:00
i = (int)tor_parse_long(c->value, 10, 0,
var->type==CONFIG_TYPE_PORT ? 65535 : INT_MAX,
&ok, NULL);
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
if (!ok) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
"Int keyword '%s %s' is malformed or out of bounds.",
c->key, c->value);
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
return -1;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
}
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
*(int *)lvalue = i;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
break;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
case CONFIG_TYPE_INTERVAL: {
i = config_parse_interval(c->value, &ok);
if (!ok) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
"Interval '%s %s' is malformed or out of bounds.",
c->key, c->value);
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
return -1;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
}
*(int *)lvalue = i;
break;
}
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
case CONFIG_TYPE_MSEC_INTERVAL: {
i = config_parse_msec_interval(c->value, &ok);
if (!ok) {
tor_asprintf(msg,
"Msec interval '%s %s' is malformed or out of bounds.",
c->key, c->value);
return -1;
}
*(int *)lvalue = i;
break;
}
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
case CONFIG_TYPE_MEMUNIT: {
uint64_t u64 = config_parse_memunit(c->value, &ok);
if (!ok) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
"Value '%s %s' is malformed or out of bounds.",
c->key, c->value);
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
return -1;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
}
*(uint64_t *)lvalue = u64;
break;
}
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
case CONFIG_TYPE_BOOL:
r14388@tombo: nickm | 2008-02-21 22:44:28 -0500 More 64-to-32 fixes. svn:r13672
2008-02-22 04:44:36 +01:00
i = (int)tor_parse_long(c->value, 10, 0, 1, &ok, NULL);
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
if (!ok) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
"Boolean '%s %s' expects 0 or 1.",
c->key, c->value);
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
return -1;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
}
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
*(int *)lvalue = i;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
break;
Add a new configuration type, "AUTOBOOL", to handle 1/0/auto types We only have one of these now, but I'm about to add a few more. Yes, I have already thought of the "Filenotfoundian logic" joke.
2010-11-08 19:34:40 +01:00
case CONFIG_TYPE_AUTOBOOL:
if (!strcmp(c->value, "auto"))
*(int *)lvalue = -1;
else if (!strcmp(c->value, "0"))
*(int *)lvalue = 0;
else if (!strcmp(c->value, "1"))
*(int *)lvalue = 1;
else {
tor_asprintf(msg, "Boolean '%s %s' expects 0, 1, or 'auto'.",
c->key, c->value);
return -1;
}
break;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
case CONFIG_TYPE_STRING:
Clean up a macro/cpp interaction related to bug 707. Also, add a new "filename" config value type which is currently just a synonym for string, but which might be subject to expansion later. svn:r15305
2008-06-16 20:09:53 +02:00
case CONFIG_TYPE_FILENAME:
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
tor_free(*(char **)lvalue);
*(char **)lvalue = tor_strdup(c->value);
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
break;
case CONFIG_TYPE_DOUBLE:
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
*(double *)lvalue = atof(c->value);
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
break;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
case CONFIG_TYPE_ISOTIME:
if (parse_iso_time(c->value, (time_t *)lvalue)) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
"Invalid time '%s' for keyword '%s'", c->value, c->key);
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
return -1;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
}
break;
r17188@tombo: nickm | 2008-07-18 14:35:18 -0400 Add new ExcludeExitNodes option. Also add a new routerset type to handle Exclude[Exit]Nodes. It is optimized for O(1) membership tests, so as to make choosing a random router run in O(N_routers) time instead of in O(N_routers*N_Excluded_Routers). svn:r16061
2008-07-18 20:36:32 +02:00
case CONFIG_TYPE_ROUTERSET:
if (*(routerset_t**)lvalue) {
routerset_free(*(routerset_t**)lvalue);
}
*(routerset_t**)lvalue = routerset_new();
if (routerset_parse(*(routerset_t**)lvalue, c->value, c->key)<0) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg, "Invalid exit list '%s' for option '%s'",
r17188@tombo: nickm | 2008-07-18 14:35:18 -0400 Add new ExcludeExitNodes option. Also add a new routerset type to handle Exclude[Exit]Nodes. It is optimized for O(1) membership tests, so as to make choosing a random router run in O(N_routers) time instead of in O(N_routers*N_Excluded_Routers). svn:r16061
2008-07-18 20:36:32 +02:00
c->value, c->key);
return -1;
}
break;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
case CONFIG_TYPE_CSV:
and take the bang out svn:r5042
2005-09-14 04:13:35 +02:00
if (*(smartlist_t**)lvalue) {
ok, ok, csv's get replaced even when we assign them from the torrc. svn:r5041
2005-09-14 04:12:29 +02:00
SMARTLIST_FOREACH(*(smartlist_t**)lvalue, char *, cp, tor_free(cp));
smartlist_clear(*(smartlist_t**)lvalue);
} else {
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
*(smartlist_t**)lvalue = smartlist_new();
Remove broken and undocumented wart where multiple CSV config lines are appended rather than replaced. Resolves bug #49 svn:r3301
2005-01-05 03:51:04 +01:00
}
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
smartlist_split_string(*(smartlist_t**)lvalue, c->value, ",",
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, 0);
break;
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
case CONFIG_TYPE_LINELIST:
case CONFIG_TYPE_LINELIST_S:
Make linelists always overridden by the command line This starts an effort to refactor torrc handling code to make it easier to live with. It makes it possible to override exit policies from the command line, and possible to override (rather than append to) socksport lists from the command line. It'll be necessary to make a "base" torrc implementation work at all.
2011-11-28 00:11:50 +01:00
{
config_line_t *lastval = *(config_line_t**)lvalue;
if (lastval && lastval->fragile) {
if (c->command != CONFIG_LINE_APPEND) {
config_free_lines(lastval);
*(config_line_t**)lvalue = NULL;
} else {
lastval->fragile = 0;
}
}
config_line_append((config_line_t**)lvalue, c->key, c->value);
}
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
break;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
case CONFIG_TYPE_OBSOLETE:
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG, "Skipping obsolete configuration option '%s'", c->key);
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
break;
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
case CONFIG_TYPE_LINELIST_V:
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
"You may not provide a value for virtual option '%s'", c->key);
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
return -1;
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
default:
tor_assert(0);
break;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
}
Re-write configuration option lookup code: use a table, not a big if( || || || ||). Deprecate unoffical abbrevs and abbrevs not on the command line. svn:r2594
2004-10-27 04:30:28 +02:00
return 0;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
}
Make linelists always overridden by the command line This starts an effort to refactor torrc handling code to make it easier to live with. It makes it possible to override exit policies from the command line, and possible to override (rather than append to) socksport lists from the command line. It'll be necessary to make a "base" torrc implementation work at all.
2011-11-28 00:11:50 +01:00
/** Mark every linelist in <b>options<b> "fragile", so that fresh assignments
* to it will replace old ones. */
static void
config_mark_lists_fragile(const config_format_t *fmt, or_options_t *options)
{
int i;
tor_assert(fmt);
tor_assert(options);
for (i = 0; fmt->vars[i].name; ++i) {
const config_var_t *var = &fmt->vars[i];
config_line_t *list;
if (var->type != CONFIG_TYPE_LINELIST &&
var->type != CONFIG_TYPE_LINELIST_V)
continue;
list = *(config_line_t **)STRUCT_VAR_P(options, var->var_offset);
if (list)
list->fragile = 1;
}
}
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
/** If <b>c</b> is a syntactically valid configuration line, update
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
* <b>options</b> with its value and return 0. Otherwise return -1 for bad
* key, -2 for bad value.
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
*
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
* If <b>clear_first</b> is set, clear the value first. Then if
* <b>use_defaults</b> is set, set the value to the default.
*
* Called from config_assign().
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
*/
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_assign_line(const config_format_t *fmt, or_options_t *options,
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
config_line_t *c, int use_defaults,
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
int clear_first, bitarray_t *options_seen, char **msg)
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
{
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *var;
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
CHECK(fmt, options);
var = config_find_option(fmt, c->key);
if (!var) {
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
if (fmt->extra) {
void *lvalue = STRUCT_VAR_P(options, fmt->extra->var_offset);
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_info(LD_CONFIG,
"Found unrecognized option '%s'; saving it.", c->key);
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
config_line_append((config_line_t**)lvalue, c->key, c->value);
return 0;
} else {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
"Unknown option '%s'. Failing.", c->key);
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
return -1;
}
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
}
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
/* Put keyword into canonical case. */
if (strcmp(var->name, c->key)) {
tor_free(c->key);
c->key = tor_strdup(var->name);
}
fix bug reported by gozu: if we get a linelist or linelist_s config option from the torrc and it has no value, warn and skip rather than silently resetting it to its default. svn:r6125
2006-03-11 19:40:33 +01:00
if (!strlen(c->value)) {
/* reset or clear it, then return */
if (!clear_first) {
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
if ((var->type == CONFIG_TYPE_LINELIST ||
var->type == CONFIG_TYPE_LINELIST_S) &&
c->command != CONFIG_LINE_CLEAR) {
Spell-check Tor.
2009-05-27 23:55:51 +02:00
/* We got an empty linelist from the torrc or command line.
fix bug reported by gozu: if we get a linelist or linelist_s config option from the torrc and it has no value, warn and skip rather than silently resetting it to its default. svn:r6125
2006-03-11 19:40:33 +01:00
As a special case, call this an error. Warn and ignore. */
log_warn(LD_CONFIG,
"Linelist option '%s' has no value. Skipping.", c->key);
} else { /* not already cleared */
option_reset(fmt, options, var, use_defaults);
}
}
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
return 0;
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
} else if (c->command == CONFIG_LINE_CLEAR && !clear_first) {
option_reset(fmt, options, var, use_defaults);
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
}
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
if (options_seen && (var->type != CONFIG_TYPE_LINELIST &&
var->type != CONFIG_TYPE_LINELIST_S)) {
/* We're tracking which options we've seen, and this option is not
* supposed to occur more than once. */
Fix a compile warning on OS X 10.6 Also update the changes file to contain a note on which bug was fixed by this.
2010-08-02 18:44:00 +02:00
int var_index = (int)(var - fmt->vars);
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
if (bitarray_is_set(options_seen, var_index)) {
log_warn(LD_CONFIG, "Option '%s' used more than once; all but the last "
"value will be ignored.", var->name);
}
bitarray_set(options_seen, var_index);
}
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
if (config_assign_value(fmt, options, c, msg) < 0)
break out a new function config_assign_value() that just assigns the value it's given, and doesn't do any of their weird clearing or resetting stuff. use that when we're trying to clear or reset values. svn:r4984
2005-09-10 04:42:32 +02:00
return -2;
return 0;
}
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
/** Restore the option named <b>key</b> in options to its default value.
* Called from config_assign(). */
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
static void
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_reset_line(const config_format_t *fmt, or_options_t *options,
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
const char *key, int use_defaults)
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
{
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *var;
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
CHECK(fmt, options);
var = config_find_option(fmt, key);
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
if (!var)
return; /* give error on next pass. */
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
option_reset(fmt, options, var, use_defaults);
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
}
Separate is-recognized-option from get-option-value, since NULL is ambiguous and returning "" misrepresents. svn:r2731
2004-11-09 07:40:32 +01:00
/** Return true iff key is a valid configuration option. */
int
Rename cross-format config.c stuff to config_* and or_options_t-specific stuff to option[s]_* svn:r4641
2005-07-23 03:58:05 +02:00
option_is_recognized(const char *key)
Separate is-recognized-option from get-option-value, since NULL is ambiguous and returning "" misrepresents. svn:r2731
2004-11-09 07:40:32 +01:00
{
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *var = config_find_option(&options_format, key);
Separate is-recognized-option from get-option-value, since NULL is ambiguous and returning "" misrepresents. svn:r2731
2004-11-09 07:40:32 +01:00
return (var != NULL);
}
r17623@catbus: nickm | 2008-01-15 00:22:04 -0500 bulletproof option_get_canonical_name() svn:r13135
2008-01-15 06:57:14 +01:00
/** Return the canonical name of a configuration option, or NULL
* if no such option exists. */
Do not reverse command-line configuration options; Use new minimal-listener-close code; Add code to canonicalize configuration names. svn:r4529
2005-07-11 19:35:36 +02:00
const char *
Rename cross-format config.c stuff to config_* and or_options_t-specific stuff to option[s]_* svn:r4641
2005-07-23 03:58:05 +02:00
option_get_canonical_name(const char *key)
Do not reverse command-line configuration options; Use new minimal-listener-close code; Add code to canonicalize configuration names. svn:r4529
2005-07-11 19:35:36 +02:00
{
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *var = config_find_option(&options_format, key);
r17623@catbus: nickm | 2008-01-15 00:22:04 -0500 bulletproof option_get_canonical_name() svn:r13135
2008-01-15 06:57:14 +01:00
return var ? var->name : NULL;
Do not reverse command-line configuration options; Use new minimal-listener-close code; Add code to canonicalize configuration names. svn:r4529
2005-07-11 19:35:36 +02:00
}
Spell-check Tor.
2009-05-27 23:55:51 +02:00
/** Return a canonical list of the options assigned for key.
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
*/
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
option_get_assignment(const or_options_t *options, const char *key)
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
{
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
return get_assigned_option(&options_format, options, key, 1);
}
/** Return true iff value needs to be quoted and escaped to be used in
* a configuration file. */
static int
config_value_needs_escape(const char *value)
{
if (*value == '\"')
return 1;
while (*value) {
switch (*value)
{
case '\r':
case '\n':
case '#':
/* Note: quotes and backspaces need special handling when we are using
* quotes, not otherwise, so they don't trigger escaping on their
* own. */
return 1;
default:
if (!TOR_ISPRINT(*value))
return 1;
}
++value;
}
return 0;
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
}
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
/** Return a newly allocated deep copy of the lines in <b>inp</b>. */
Start implementing the server side of the new directory protocol. Probably very buggy, since I can't actually run an authdir. Features - Generate and publish new network-status format - Code to cache and re-serve network-status objects generated by others. - Publish individual descriptors (by fingerprint, by "all", and by "tell me yours.") [Still needs compression logic] - Publish client and server recommended versions seprately. - Add digest of descriptor to routerinfo_t, so we can track them better, and length, so we can server them more easily. Cleanups - Unify code to sign directory-like things - Make resolve_my_address() able to tell you which name it wound up resolving. - Unify code to store and serve directory-like things so it all uses cached_dir_t. - Unify code to set the value of cached_dir_t objects. svn:r4835
2005-08-25 22:33:17 +02:00
static config_line_t *
config_lines_dup(const config_line_t *inp)
{
config_line_t *result = NULL;
config_line_t **next_out = &result;
while (inp) {
Make linelists always overridden by the command line This starts an effort to refactor torrc handling code to make it easier to live with. It makes it possible to override exit policies from the command line, and possible to override (rather than append to) socksport lists from the command line. It'll be necessary to make a "base" torrc implementation work at all.
2011-11-28 00:11:50 +01:00
*next_out = tor_malloc_zero(sizeof(config_line_t));
Start implementing the server side of the new directory protocol. Probably very buggy, since I can't actually run an authdir. Features - Generate and publish new network-status format - Code to cache and re-serve network-status objects generated by others. - Publish individual descriptors (by fingerprint, by "all", and by "tell me yours.") [Still needs compression logic] - Publish client and server recommended versions seprately. - Add digest of descriptor to routerinfo_t, so we can track them better, and length, so we can server them more easily. Cleanups - Unify code to sign directory-like things - Make resolve_my_address() able to tell you which name it wound up resolving. - Unify code to store and serve directory-like things so it all uses cached_dir_t. - Unify code to set the value of cached_dir_t objects. svn:r4835
2005-08-25 22:33:17 +02:00
(*next_out)->key = tor_strdup(inp->key);
(*next_out)->value = tor_strdup(inp->value);
inp = inp->next;
next_out = &((*next_out)->next);
}
(*next_out) = NULL;
return result;
}
r11822@catbus: nickm | 2007-02-16 12:49:20 -0500 Tech config.c about libevent 1.3, and resolve all docdoc elements in config.c svn:r9591
2007-02-16 19:12:47 +01:00
/** Return newly allocated line or lines corresponding to <b>key</b> in the
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
* configuration <b>options</b>. If <b>escape_val</b> is true and a
* value needs to be quoted before it's put in a config file, quote and
* escape that value. Return NULL if no such key exists. */
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
static config_line_t *
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
get_assigned_option(const config_format_t *fmt, const void *options,
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
const char *key, int escape_val)
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
{
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *var;
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
const void *value;
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *result;
Separate is-recognized-option from get-option-value, since NULL is ambiguous and returning "" misrepresents. svn:r2731
2004-11-09 07:40:32 +01:00
tor_assert(options && key);
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
CHECK(fmt, options);
var = config_find_option(fmt, key);
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
if (!var) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG, "Unknown option '%s'. Failing.", key);
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
return NULL;
}
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
value = STRUCT_VAR_P(options, var->var_offset);
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
result = tor_malloc_zero(sizeof(config_line_t));
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
result->key = tor_strdup(var->name);
Normalize space: add one between every control keyword and control clause. svn:r3003
2004-11-28 10:05:49 +01:00
switch (var->type)
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
{
case CONFIG_TYPE_STRING:
Clean up a macro/cpp interaction related to bug 707. Also, add a new "filename" config value type which is currently just a synonym for string, but which might be subject to expansion later. svn:r15305
2008-06-16 20:09:53 +02:00
case CONFIG_TYPE_FILENAME:
Separate is-recognized-option from get-option-value, since NULL is ambiguous and returning "" misrepresents. svn:r2731
2004-11-09 07:40:32 +01:00
if (*(char**)value) {
result->value = tor_strdup(*(char**)value);
} else {
tor_free(result->key);
tor_free(result);
return NULL;
}
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
break;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
case CONFIG_TYPE_ISOTIME:
if (*(time_t*)value) {
result->value = tor_malloc(ISO_TIME_LEN+1);
format_iso_time(result->value, *(time_t*)value);
} else {
tor_free(result->key);
tor_free(result);
}
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
escape_val = 0; /* Can't need escape. */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
break;
"(Socks|Control|etc)Port auto" now tells Tor to open an arbitrary port This is the major part of the implementation for trac issue 3076.
2011-05-02 21:05:10 +02:00
case CONFIG_TYPE_PORT:
if (*(int*)value == CFG_AUTO_PORT) {
result->value = tor_strdup("auto");
escape_val = 0;
break;
}
minor cleanups while reviewing 3216
2011-05-30 21:31:06 +02:00
/* fall through */
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
case CONFIG_TYPE_INTERVAL:
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
case CONFIG_TYPE_MSEC_INTERVAL:
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
case CONFIG_TYPE_UINT:
Resolve many XXXs and all DOCDOCs svn:r2755
2004-11-09 19:22:17 +01:00
/* This means every or_options_t uint or bool element
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
* needs to be an int. Not, say, a uint16_t or char. */
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(&result->value, "%d", *(int*)value);
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
escape_val = 0; /* Can't need escape. */
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
break;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
case CONFIG_TYPE_MEMUNIT:
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(&result->value, U64_FORMAT,
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
U64_PRINTF_ARG(*(uint64_t*)value));
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
escape_val = 0; /* Can't need escape. */
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
break;
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
case CONFIG_TYPE_DOUBLE:
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(&result->value, "%f", *(double*)value);
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
escape_val = 0; /* Can't need escape. */
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
break;
Add a new configuration type, "AUTOBOOL", to handle 1/0/auto types We only have one of these now, but I'm about to add a few more. Yes, I have already thought of the "Filenotfoundian logic" joke.
2010-11-08 19:34:40 +01:00
case CONFIG_TYPE_AUTOBOOL:
if (*(int*)value == -1) {
result->value = tor_strdup("auto");
escape_val = 0;
break;
}
/* fall through */
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
case CONFIG_TYPE_BOOL:
result->value = tor_strdup(*(int*)value ? "1" : "0");
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
escape_val = 0; /* Can't need escape. */
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
break;
r17188@tombo: nickm | 2008-07-18 14:35:18 -0400 Add new ExcludeExitNodes option. Also add a new routerset type to handle Exclude[Exit]Nodes. It is optimized for O(1) membership tests, so as to make choosing a random router run in O(N_routers) time instead of in O(N_routers*N_Excluded_Routers). svn:r16061
2008-07-18 20:36:32 +02:00
case CONFIG_TYPE_ROUTERSET:
result->value = routerset_to_string(*(routerset_t**)value);
break;
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
case CONFIG_TYPE_CSV:
fix some dereferences svn:r2726
2004-11-09 06:59:22 +01:00
if (*(smartlist_t**)value)
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
result->value =
smartlist_join_strings(*(smartlist_t**)value, ",", 0, NULL);
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
else
result->value = tor_strdup("");
break;
During options_dup, don't copy LINELIST_S or OBSOLETE options. And recognize obsolete options now in config_get_assigned_option(). svn:r2729
2004-11-09 07:18:17 +01:00
case CONFIG_TYPE_OBSOLETE:
vidalia asks us for the value of the Group config option. but now it's obsolete. which causes us to inform the user every time, even though the user can't do anything about it other than get confused. now it's an info-level log by default. svn:r17206
2008-11-07 05:38:58 +01:00
log_fn(LOG_PROTOCOL_WARN, LD_CONFIG,
"You asked me for the value of an obsolete config option '%s'.",
key);
During options_dup, don't copy LINELIST_S or OBSOLETE options. And recognize obsolete options now in config_get_assigned_option(). svn:r2729
2004-11-09 07:18:17 +01:00
tor_free(result->key);
tor_free(result);
return NULL;
r7324@Kushana: nickm | 2006-08-10 23:23:15 -0700 Add more warnings to the list of those we tolerate. Start using GCC attributes more, for better error checking and better code generation. svn:r7020
2006-08-11 09:09:17 +02:00
case CONFIG_TYPE_LINELIST_S:
log_warn(LD_CONFIG,
"Can't return context-sensitive '%s' on its own", key);
tor_free(result->key);
tor_free(result);
return NULL;
case CONFIG_TYPE_LINELIST:
case CONFIG_TYPE_LINELIST_V:
tor_free(result->key);
tor_free(result);
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
result = config_lines_dup(*(const config_line_t**)value);
break;
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
default:
tor_free(result->key);
tor_free(result);
r12074@catbus: nickm | 2007-03-04 15:11:43 -0500 Make all LD_BUG log messsages get prefixed with "Bug: ". Remove manually-generated "Bug: "s from log-messages. (Apparently, we remembered to add them about 40% of the time.) svn:r9733
2007-03-04 21:11:46 +01:00
log_warn(LD_BUG,"Unknown type %d for known key '%s'",
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
var->type, key);
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
return NULL;
}
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
if (escape_val) {
config_line_t *line;
for (line = result; line; line = line->next) {
if (line->value && config_value_needs_escape(line->value)) {
char *newval = esc_for_log(line->value);
tor_free(line->value);
line->value = newval;
}
}
}
Add function to generate/copy the config_lines, given the name of an option svn:r2660
2004-11-03 19:29:29 +01:00
return result;
}
make free_options() and init_options() use the big array. also remove obsolete config_assign_defaults(). svn:r2672
2004-11-04 09:26:34 +01:00
/** Iterate through the linked list of requested options <b>list</b>.
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
* For each item, convert as appropriate and assign to <b>options</b>.
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
* If an item is unrecognized, set *msg and return -1 immediately,
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
* else return 0 for success.
*
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
* If <b>clear_first</b>, interpret config options as replacing (not
* extending) their previous values. If <b>clear_first</b> is set,
* then <b>use_defaults</b> to decide if you set to defaults after
* clearing, or make the value 0 or NULL.
*
* Here are the use cases:
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
* 1. A non-empty AllowInvalid line in your torrc. Appends to current
ok, ok, csv's get replaced even when we assign them from the torrc. svn:r5041
2005-09-14 04:12:29 +02:00
* if linelist, replaces current if csv.
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
* 2. An empty AllowInvalid line in your torrc. Should clear it.
* 3. "RESETCONF AllowInvalid" sets it to default.
* 4. "SETCONF AllowInvalid" makes it NULL.
* 5. "SETCONF AllowInvalid=foo" clears it and sets it to "foo".
if you give a config option in the torrc or the commandline with no value, and reset is false, then it clears it entirely. svn:r4926
2005-09-08 08:55:53 +02:00
*
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
* Use_defaults Clear_first
* 0 0 "append"
* 1 0 undefined, don't use
* 0 1 "set to null first"
* 1 1 "set to defaults first"
if you give a config option in the torrc or the commandline with no value, and reset is false, then it clears it entirely. svn:r4926
2005-09-08 08:55:53 +02:00
* Return 0 on success, -1 on bad key, -2 on bad value.
fix bug reported by gozu: if we get a linelist or linelist_s config option from the torrc and it has no value, warn and skip rather than silently resetting it to its default. svn:r6125
2006-03-11 19:40:33 +01:00
*
* As an additional special case, if a LINELIST config option has
* no value and clear_first is 0, then warn and ignore it.
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
*/
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
/*
There are three call cases for config_assign() currently.
Case one: Torrc entry
options_init_from_torrc() calls config_assign(0, 0)
stop the call duplication inside config_assign() svn:r5045
2005-09-14 04:35:06 +02:00
calls config_assign_line(0, 0).
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
if value is empty, calls option_reset(0) and returns.
calls config_assign_value(), appends.
Case two: setconf
options_trial_assign() calls config_assign(0, 1)
calls config_reset_line(0)
calls option_reset(0)
calls option_clear().
stop the call duplication inside config_assign() svn:r5045
2005-09-14 04:35:06 +02:00
calls config_assign_line(0, 1).
if value is empty, returns.
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
calls config_assign_value(), appends.
Case three: resetconf
options_trial_assign() calls config_assign(1, 1)
calls config_reset_line(1)
calls option_reset(1)
calls option_clear().
calls config_assign_value(default)
stop the call duplication inside config_assign() svn:r5045
2005-09-14 04:35:06 +02:00
calls config_assign_line(1, 1).
returns.
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
*/
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_assign(const config_format_t *fmt, void *options, config_line_t *list,
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
int use_defaults, int clear_first, char **msg)
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
{
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *p;
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
bitarray_t *options_seen;
const int n_options = config_count_options(fmt);
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
CHECK(fmt, options);
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
/* pass 1: normalize keys */
for (p = list; p; p = p->next) {
Add ability to warn when using abbrev mechanism to deprecate option names svn:r5259
2005-10-17 05:06:00 +02:00
const char *full = expand_abbrev(fmt, p->key, 0, 1);
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
if (strcmp(full,p->key)) {
tor_free(p->key);
p->key = tor_strdup(full);
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
}
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
}
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
/* pass 2: if we're reading from a resetting source, clear all
* mentioned config options, and maybe set to their defaults. */
if (clear_first) {
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
for (p = list; p; p = p->next)
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
config_reset_line(fmt, options, p->key, use_defaults);
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
}
Add ability for some-but-not-all abbrevs to work in config file. Add a bunch of singular/plural abbrevs suggested by arma svn:r2612
2004-10-27 19:37:01 +02:00
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
options_seen = bitarray_init_zero(n_options);
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
/* pass 3: assign. */
while (list) {
Make set-option functions distinguish between bad keys and bad values, in case the caller cares svn:r2707
2004-11-07 23:59:30 +01:00
int r;
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
if ((r=config_assign_line(fmt, options, list, use_defaults,
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
clear_first, options_seen, msg))) {
bitarray_free(options_seen);
Make set-option functions distinguish between bad keys and bad values, in case the caller cares svn:r2707
2004-11-07 23:59:30 +01:00
return r;
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
}
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
list = list->next;
remove trailing whitespace svn:r951
2003-12-17 22:09:31 +01:00
}
Warn when encounter the same (non-list) option twice in the same place It's okay to get (say) a SocksPort line in the torrc, and then a SocksPort on the command line to override it, and then a SocksPort via a controller to override *that*. But if there are two occurrences of SocksPort in the torrc, or on the command line, or in a single SETCONF command, then the user is likely confused. Our old code would not help unconfuse the user, but would instead silently ignore all but the last occurrence. This patch changes the behavior so that if the some option is passed more than once to any torrc, command line, or SETCONF (each of which coincidentally corresponds to a call to config_assign()), and the option is not a type that allows multiple occurrences (LINELIST or LINELIST_X), then we can warn the user. This closes trac entry 1384.
2010-07-31 19:16:48 +02:00
bitarray_free(options_seen);
Make linelists always overridden by the command line This starts an effort to refactor torrc handling code to make it easier to live with. It makes it possible to override exit policies from the command line, and possible to override (rather than append to) socksport lists from the command line. It'll be necessary to make a "base" torrc implementation work at all.
2011-11-28 00:11:50 +01:00
/** Now we're done assigning a group of options to the configuration.
* Subsequent group assignments should _replace_ linelists, not extend
* them. */
config_mark_lists_fragile(fmt, options);
make config_assign fail rather than warn for unrecognized config options svn:r1194
2004-03-02 06:00:50 +01:00
return 0;
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
}
Change interface to config_trial_assign even more; have it use get_options() instead svn:r2754
2004-11-09 18:15:17 +01:00
/** Try assigning <b>list</b> to the global options. You do this by duping
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
* options, assigning list to the new one, then validating it. If it's
don't check permissions on our datadirectory until we switch uid/gid svn:r2759
2004-11-10 01:11:37 +01:00
* ok, then throw out the old one and stick with the new one. Else,
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
* revert to old and return failure. Return SETOPT_OK on success, or
* a setopt_err_t on failure.
Fix a couple of things that make GCC complain with verbose warnings on. Also, fix some whitespace. svn:r6317
2006-04-08 07:43:52 +02:00
*
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
* If not success, point *<b>msg</b> to a newly allocated string describing
* what went wrong.
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
*/
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
setopt_err_t
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
options_trial_assign(config_line_t *list, int use_defaults,
int clear_first, char **msg)
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
{
Make set-option functions distinguish between bad keys and bad values, in case the caller cares svn:r2707
2004-11-07 23:59:30 +01:00
int r;
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
or_options_t *trial_options = options_dup(&options_format, get_options());
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
if ((r=config_assign(&options_format, trial_options,
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
list, use_defaults, clear_first, msg)) < 0) {
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
config_free(&options_format, trial_options);
Make set-option functions distinguish between bad keys and bad values, in case the caller cares svn:r2707
2004-11-07 23:59:30 +01:00
return r;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
}
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
if (options_validate(get_options_mutable(), trial_options, 1, msg) < 0) {
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
config_free(&options_format, trial_options);
Downlgrade tweak, and answer lots of XXX021s. No actual code fixes in this patch. svn:r17686
2008-12-18 17:11:24 +01:00
return SETOPT_ERR_PARSE; /*XXX make this a separate return value. */
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
}
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
if (options_transition_allowed(get_options(), trial_options, msg) < 0) {
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
config_free(&options_format, trial_options);
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
return SETOPT_ERR_TRANSITION;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
}
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
if (set_options(trial_options, msg)<0) {
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
config_free(&options_format, trial_options);
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
return SETOPT_ERR_SETTING;
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
}
/* we liked it. put it in place. */
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
return SETOPT_OK;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
}
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
/** Reset config option <b>var</b> to 0, 0.0, NULL, or the equivalent.
* Called from option_reset() and config_free(). */
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
static void
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
option_clear(const config_format_t *fmt, or_options_t *options,
Replace a "const const" with a "const" Looks like this squeaked in while I was doing a search-and-replace to constify things. Coverity CID 483.
2011-07-01 17:11:35 +02:00
const config_var_t *var)
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
{
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
void *lvalue = STRUCT_VAR_P(options, var->var_offset);
Add a new warning to our "warn a lot" list: unused parameters. This means we have to explicitly "use" unuseds, but it can catch bugs. (It caught two coding mistakes so far.) svn:r6532
2006-06-05 00:42:13 +02:00
(void)fmt; /* unused */
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
switch (var->type) {
case CONFIG_TYPE_STRING:
Clean up a macro/cpp interaction related to bug 707. Also, add a new "filename" config value type which is currently just a synonym for string, but which might be subject to expansion later. svn:r15305
2008-06-16 20:09:53 +02:00
case CONFIG_TYPE_FILENAME:
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
tor_free(*(char**)lvalue);
break;
case CONFIG_TYPE_DOUBLE:
*(double*)lvalue = 0.0;
break;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
case CONFIG_TYPE_ISOTIME:
*(time_t*)lvalue = 0;
Add missing break statements for Coverity CIDs #406,407. The code for these was super-wrong, but will only break things when we reset an option on a platform where sizeof(time_t) is different from sizeof(int).
2009-10-27 02:35:26 +01:00
break;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
case CONFIG_TYPE_INTERVAL:
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
case CONFIG_TYPE_MSEC_INTERVAL:
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
case CONFIG_TYPE_UINT:
"(Socks|Control|etc)Port auto" now tells Tor to open an arbitrary port This is the major part of the implementation for trac issue 3076.
2011-05-02 21:05:10 +02:00
case CONFIG_TYPE_PORT:
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
case CONFIG_TYPE_BOOL:
*(int*)lvalue = 0;
break;
Add a new configuration type, "AUTOBOOL", to handle 1/0/auto types We only have one of these now, but I'm about to add a few more. Yes, I have already thought of the "Filenotfoundian logic" joke.
2010-11-08 19:34:40 +01:00
case CONFIG_TYPE_AUTOBOOL:
*(int*)lvalue = -1;
break;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
case CONFIG_TYPE_MEMUNIT:
*(uint64_t*)lvalue = 0;
break;
r17188@tombo: nickm | 2008-07-18 14:35:18 -0400 Add new ExcludeExitNodes option. Also add a new routerset type to handle Exclude[Exit]Nodes. It is optimized for O(1) membership tests, so as to make choosing a random router run in O(N_routers) time instead of in O(N_routers*N_Excluded_Routers). svn:r16061
2008-07-18 20:36:32 +02:00
case CONFIG_TYPE_ROUTERSET:
if (*(routerset_t**)lvalue) {
routerset_free(*(routerset_t**)lvalue);
*(routerset_t**)lvalue = NULL;
}
Add missing break statements for Coverity CIDs #406,407. The code for these was super-wrong, but will only break things when we reset an option on a platform where sizeof(time_t) is different from sizeof(int).
2009-10-27 02:35:26 +01:00
break;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
case CONFIG_TYPE_CSV:
if (*(smartlist_t**)lvalue) {
SMARTLIST_FOREACH(*(smartlist_t **)lvalue, char *, cp, tor_free(cp));
smartlist_free(*(smartlist_t **)lvalue);
*(smartlist_t **)lvalue = NULL;
}
break;
case CONFIG_TYPE_LINELIST:
case CONFIG_TYPE_LINELIST_S:
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_free_lines(*(config_line_t **)lvalue);
*(config_line_t **)lvalue = NULL;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
break;
case CONFIG_TYPE_LINELIST_V:
/* handled by linelist_s. */
break;
case CONFIG_TYPE_OBSOLETE:
break;
}
it really does help with the functions end. svn:r4927
2005-09-08 08:59:58 +02:00
}
if you give a config option in the torrc or the commandline with no value, and reset is false, then it clears it entirely. svn:r4926
2005-09-08 08:55:53 +02:00
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
/** Clear the option indexed by <b>var</b> in <b>options</b>. Then if
* <b>use_defaults</b>, set it to its default value.
* Called by config_init() and option_reset_line() and option_assign_line(). */
if you give a config option in the torrc or the commandline with no value, and reset is false, then it clears it entirely. svn:r4926
2005-09-08 08:55:53 +02:00
static void
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
option_reset(const config_format_t *fmt, or_options_t *options,
const config_var_t *var, int use_defaults)
if you give a config option in the torrc or the commandline with no value, and reset is false, then it clears it entirely. svn:r4926
2005-09-08 08:55:53 +02:00
{
config_line_t *c;
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
char *msg = NULL;
if you give a config option in the torrc or the commandline with no value, and reset is false, then it clears it entirely. svn:r4926
2005-09-08 08:55:53 +02:00
CHECK(fmt, options);
option_clear(fmt, options, var); /* clear it first */
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
if (!use_defaults)
return; /* all done */
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
if (var->initvalue) {
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
c = tor_malloc_zero(sizeof(config_line_t));
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
c->key = tor_strdup(var->name);
c->value = tor_strdup(var->initvalue);
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
if (config_assign_value(fmt, options, c, &msg) < 0) {
log_warn(LD_BUG, "Failed to assign default: %s", msg);
tor_free(msg); /* if this happens it's a bug */
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
config_free_lines(c);
}
}
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
/** Print a usage message for tor. */
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
static void
print_usage(void)
{
a first cut at a print_usage() svn:r2887
2004-11-15 08:29:27 +01:00
printf(
correct copyright statement svn:r13370
2008-02-04 17:54:14 +01:00
"Copyright (c) 2001-2004, Roger Dingledine\n"
"Copyright (c) 2004-2006, Roger Dingledine, Nick Mathewson\n"
Bump copyright statements to 2011
2011-01-03 17:50:39 +01:00
"Copyright (c) 2007-2011, The Tor Project, Inc.\n\n"
a first cut at a print_usage() svn:r2887
2004-11-15 08:29:27 +01:00
"tor -f <torrc> [args]\n"
patch from karsten to clean up documentation and to integrate more fixes into rend-spec.txt. svn:r12715
2007-12-07 22:27:58 +01:00
"See man page for options, or https://www.torproject.org/ for "
"documentation.\n");
a skeletal print_usage() function svn:r634
2003-10-20 03:19:54 +02:00
}
r11651@Kushana: nickm | 2006-12-20 12:05:04 -0500 Add a maintainer script and a new make target "make check-docs" to get a quick dump of which options are undocumented where, and which documentation refers to nonexistent options. svn:r9160
2006-12-20 18:05:48 +01:00
/** Print all non-obsolete torrc options. */
static void
list_torrc_options(void)
{
int i;
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
smartlist_t *lines = smartlist_new();
r11651@Kushana: nickm | 2006-12-20 12:05:04 -0500 Add a maintainer script and a new make target "make check-docs" to get a quick dump of which options are undocumented where, and which documentation refers to nonexistent options. svn:r9160
2006-12-20 18:05:48 +01:00
for (i = 0; _option_vars[i].name; ++i) {
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *var = &_option_vars[i];
r11651@Kushana: nickm | 2006-12-20 12:05:04 -0500 Add a maintainer script and a new make target "make check-docs" to get a quick dump of which options are undocumented where, and which documentation refers to nonexistent options. svn:r9160
2006-12-20 18:05:48 +01:00
if (var->type == CONFIG_TYPE_OBSOLETE ||
var->type == CONFIG_TYPE_LINELIST_V)
continue;
printf("%s\n", var->name);
}
r17639@catbus: nickm | 2008-01-15 19:09:21 -0500 Fix some hard to trigger but nonetheless real memory leaks spotted by an anonymous contributor. Needs review. Partial backport candidate. svn:r13147
2008-01-16 06:27:19 +01:00
smartlist_free(lines);
r11651@Kushana: nickm | 2006-12-20 12:05:04 -0500 Add a maintainer script and a new make target "make check-docs" to get a quick dump of which options are undocumented where, and which documentation refers to nonexistent options. svn:r9160
2006-12-20 18:05:48 +01:00
}
r8846@totoro: nickm | 2006-10-02 16:59:57 -0400 Move is_local_IP to config.c; have it check for same-/24; make it used only for reachability (not for banwidth, because that is probably not what we want). Fixes an XXX. svn:r8578
2006-10-02 23:00:35 +02:00
/** Last value actually set by resolve_my_address. */
static uint32_t last_resolved_addr = 0;
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
/**
absolutely refuse to let people pick internal IP addresses if they're using the default dirservers. we're getting a big pile of a dozen or so servers that have picked private IP addresses despite the warning (presumably they don't even know they have logs) svn:r4354
2005-06-08 22:32:22 +02:00
* Based on <b>options-\>Address</b>, guess our public IP address and put it
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
* (in host order) into *<b>addr_out</b>. If <b>hostname_out</b> is provided,
* set *<b>hostname_out</b> to a new string holding the hostname we used to
* get the address. Return 0 if all is well, or -1 if we can't find a suitable
Start implementing the server side of the new directory protocol. Probably very buggy, since I can't actually run an authdir. Features - Generate and publish new network-status format - Code to cache and re-serve network-status objects generated by others. - Publish individual descriptors (by fingerprint, by "all", and by "tell me yours.") [Still needs compression logic] - Publish client and server recommended versions seprately. - Add digest of descriptor to routerinfo_t, so we can track them better, and length, so we can server them more easily. Cleanups - Unify code to sign directory-like things - Make resolve_my_address() able to tell you which name it wound up resolving. - Unify code to store and serve directory-like things so it all uses cached_dir_t. - Unify code to set the value of cached_dir_t objects. svn:r4835
2005-08-25 22:33:17 +02:00
* public IP address.
Make relays handle an address suggestion from a directory server giving an IPv6 address. last_guessed_ip becomes a tor_addr_t. Most parts of router_new_address_suggestion() learns about IPv6 (resolve_my_address() is still IPv4 only).
2012-03-27 15:00:34 +02:00
* XXXX ipv6
Doxygenate config.c svn:r1834
2004-05-10 06:48:13 +02:00
*/
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
resolve_my_address(int warn_severity, const or_options_t *options,
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
uint32_t *addr_out, char **hostname_out)
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
{
prevent servers from publishing an internal IP as their address svn:r1272
2004-03-14 19:12:59 +01:00
struct in_addr in;
commit my annotations while i was hunting down the host order bug
2010-03-05 22:04:01 +01:00
uint32_t addr; /* host order */
o Handle servers with dynamic IP addresses: don't replace options->Address with the resolved one at startup. o detect our address right before we make a routerinfo each time. o external IP vs bind-IP. Already done, just use options->Address. o OutboundBindAddress config option, to bind to a specific IP address for outgoing connect()s. svn:r2241
2004-08-16 13:43:18 +02:00
char hostname[256];
make it warn about internal IPs not only if we had to guess, but also if they used a hostname rather than an IP svn:r1285
2004-03-15 05:57:24 +01:00
int explicit_ip=1;
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
int explicit_hostname=1;
r11824@Kushana: nickm | 2007-01-03 17:15:28 -0500 control-spec: upcase arguments in status events; note unimplemented events individually r11825@Kushana: nickm | 2007-01-03 17:41:43 -0500 Implement EXTERNAL IP server status event. r11826@Kushana: nickm | 2007-01-03 17:47:10 -0500 Implement BAD_SERVER_DESCRIPTOR server status event. r11827@Kushana: nickm | 2007-01-03 18:01:56 -0500 Implement SOCKS_UNKNOWN_PROTOCOL and DANGEROUS_SOCKS client events. r11828@Kushana: nickm | 2007-01-03 18:23:22 -0500 Implement BUG controller events. Also, flush ERR-level status events just like ERR-level log messages. r11829@Kushana: nickm | 2007-01-03 23:37:27 -0500 Yet more status events: CLOCK_SKEW, GOOD/ACCEPTED_SERVER_DESCRIPTOR, {CHECKING_}REACHABILITY_{SUCCEEDED|FAILED} r11833@Kushana: nickm | 2007-01-05 16:56:37 -0500 Note some unimplementedness in control-spec.txt svn:r9279
2007-01-06 06:42:31 +01:00
int from_interface=0;
Tweak the fmt_addr32 code Clarify documentation, rename a local, and fix a memory leak.
2010-10-15 23:07:27 +02:00
char *addr_string = NULL;
Patch from goodell for task 131: Do not die on internal address if we are not publishing. svn:r4186
2005-05-09 06:31:00 +02:00
const char *address = options->Address;
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
int notice_severity = warn_severity <= LOG_NOTICE ?
LOG_NOTICE : warn_severity;
prevent servers from publishing an internal IP as their address svn:r1272
2004-03-14 19:12:59 +01:00
Start implementing the server side of the new directory protocol. Probably very buggy, since I can't actually run an authdir. Features - Generate and publish new network-status format - Code to cache and re-serve network-status objects generated by others. - Publish individual descriptors (by fingerprint, by "all", and by "tell me yours.") [Still needs compression logic] - Publish client and server recommended versions seprately. - Add digest of descriptor to routerinfo_t, so we can track them better, and length, so we can server them more easily. Cleanups - Unify code to sign directory-like things - Make resolve_my_address() able to tell you which name it wound up resolving. - Unify code to store and serve directory-like things so it all uses cached_dir_t. - Unify code to set the value of cached_dir_t objects. svn:r4835
2005-08-25 22:33:17 +02:00
tor_assert(addr_out);
o Handle servers with dynamic IP addresses: don't replace options->Address with the resolved one at startup. o detect our address right before we make a routerinfo each time. o external IP vs bind-IP. Already done, just use options->Address. o OutboundBindAddress config option, to bind to a specific IP address for outgoing connect()s. svn:r2241
2004-08-16 13:43:18 +02:00
workaround for user error: some people were putting "Address " in their torrc, and they had a buggy resolver that resolved " " to 0.0.0.0. Oops. svn:r4150
2005-04-29 20:52:05 +02:00
if (address && *address) {
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
strlcpy(hostname, address, sizeof(hostname));
o Handle servers with dynamic IP addresses: don't replace options->Address with the resolved one at startup. o detect our address right before we make a routerinfo each time. o external IP vs bind-IP. Already done, just use options->Address. o OutboundBindAddress config option, to bind to a specific IP address for outgoing connect()s. svn:r2241
2004-08-16 13:43:18 +02:00
} else { /* then we need to guess our address */
make it warn about internal IPs not only if we had to guess, but also if they used a hostname rather than an IP svn:r1285
2004-03-15 05:57:24 +01:00
explicit_ip = 0; /* it's implicit */
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
explicit_hostname = 0; /* it's implicit */
prevent servers from publishing an internal IP as their address svn:r1272
2004-03-14 19:12:59 +01:00
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
if (gethostname(hostname, sizeof(hostname)) < 0) {
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(warn_severity, LD_NET,"Error obtaining local hostname");
prevent servers from publishing an internal IP as their address svn:r1272
2004-03-14 19:12:59 +01:00
return -1;
}
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_debug(LD_CONFIG,"Guessed local host name as '%s'",hostname);
prevent servers from publishing an internal IP as their address svn:r1272
2004-03-14 19:12:59 +01:00
}
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
/* now we know hostname. resolve it and keep only the IP address */
prevent servers from publishing an internal IP as their address svn:r1272
2004-03-14 19:12:59 +01:00
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
if (tor_inet_aton(hostname, &in) == 0) {
make it warn about internal IPs not only if we had to guess, but also if they used a hostname rather than an IP svn:r1285
2004-03-15 05:57:24 +01:00
/* then we have to resolve it */
explicit_ip = 0;
commit my annotations while i was hunting down the host order bug
2010-03-05 22:04:01 +01:00
if (tor_lookup_hostname(hostname, &addr)) { /* failed to resolve */
uint32_t interface_ip; /* host order */
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
if (explicit_hostname) {
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(warn_severity, LD_CONFIG,
"Could not resolve local Address '%s'. Failing.", hostname);
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
return -1;
}
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(notice_severity, LD_CONFIG,
"Could not resolve guessed local hostname '%s'. "
"Trying something else.", hostname);
parameterize the loudness of get_interface_address() svn:r8358
2006-09-09 21:20:27 +02:00
if (get_interface_address(warn_severity, &interface_ip)) {
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(warn_severity, LD_CONFIG,
"Could not get local interface IP address. Failing.");
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
return -1;
}
r11824@Kushana: nickm | 2007-01-03 17:15:28 -0500 control-spec: upcase arguments in status events; note unimplemented events individually r11825@Kushana: nickm | 2007-01-03 17:41:43 -0500 Implement EXTERNAL IP server status event. r11826@Kushana: nickm | 2007-01-03 17:47:10 -0500 Implement BAD_SERVER_DESCRIPTOR server status event. r11827@Kushana: nickm | 2007-01-03 18:01:56 -0500 Implement SOCKS_UNKNOWN_PROTOCOL and DANGEROUS_SOCKS client events. r11828@Kushana: nickm | 2007-01-03 18:23:22 -0500 Implement BUG controller events. Also, flush ERR-level status events just like ERR-level log messages. r11829@Kushana: nickm | 2007-01-03 23:37:27 -0500 Yet more status events: CLOCK_SKEW, GOOD/ACCEPTED_SERVER_DESCRIPTOR, {CHECKING_}REACHABILITY_{SUCCEEDED|FAILED} r11833@Kushana: nickm | 2007-01-05 16:56:37 -0500 Note some unimplementedness in control-spec.txt svn:r9279
2007-01-06 06:42:31 +01:00
from_interface = 1;
refactor all these tor_inet_ntoa idioms but don't refactor the ones that look messy
2010-10-02 03:31:09 +02:00
addr = interface_ip;
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(notice_severity, LD_CONFIG, "Learned IP address '%s' for "
refactor all these tor_inet_ntoa idioms but don't refactor the ones that look messy
2010-10-02 03:31:09 +02:00
"local interface. Using that.", fmt_addr32(addr));
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
strlcpy(hostname, "<guessed from interfaces>", sizeof(hostname));
commit my annotations while i was hunting down the host order bug
2010-03-05 22:04:01 +01:00
} else { /* resolved hostname into addr */
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
if (!explicit_hostname &&
refactor all these tor_inet_ntoa idioms but don't refactor the ones that look messy
2010-10-02 03:31:09 +02:00
is_internal_IP(addr, 0)) {
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
uint32_t interface_ip;
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(notice_severity, LD_CONFIG, "Guessed local hostname '%s' "
refactor all these tor_inet_ntoa idioms but don't refactor the ones that look messy
2010-10-02 03:31:09 +02:00
"resolves to a private IP address (%s). Trying something "
"else.", hostname, fmt_addr32(addr));
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
parameterize the loudness of get_interface_address() svn:r8358
2006-09-09 21:20:27 +02:00
if (get_interface_address(warn_severity, &interface_ip)) {
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(warn_severity, LD_CONFIG,
"Could not get local interface IP address. Too bad.");
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
} else if (is_internal_IP(interface_ip, 0)) {
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(notice_severity, LD_CONFIG,
"Interface IP address '%s' is a private address too. "
refactor all these tor_inet_ntoa idioms but don't refactor the ones that look messy
2010-10-02 03:31:09 +02:00
"Ignoring.", fmt_addr32(interface_ip));
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
} else {
r11824@Kushana: nickm | 2007-01-03 17:15:28 -0500 control-spec: upcase arguments in status events; note unimplemented events individually r11825@Kushana: nickm | 2007-01-03 17:41:43 -0500 Implement EXTERNAL IP server status event. r11826@Kushana: nickm | 2007-01-03 17:47:10 -0500 Implement BAD_SERVER_DESCRIPTOR server status event. r11827@Kushana: nickm | 2007-01-03 18:01:56 -0500 Implement SOCKS_UNKNOWN_PROTOCOL and DANGEROUS_SOCKS client events. r11828@Kushana: nickm | 2007-01-03 18:23:22 -0500 Implement BUG controller events. Also, flush ERR-level status events just like ERR-level log messages. r11829@Kushana: nickm | 2007-01-03 23:37:27 -0500 Yet more status events: CLOCK_SKEW, GOOD/ACCEPTED_SERVER_DESCRIPTOR, {CHECKING_}REACHABILITY_{SUCCEEDED|FAILED} r11833@Kushana: nickm | 2007-01-05 16:56:37 -0500 Note some unimplementedness in control-spec.txt svn:r9279
2007-01-06 06:42:31 +01:00
from_interface = 1;
refactor all these tor_inet_ntoa idioms but don't refactor the ones that look messy
2010-10-02 03:31:09 +02:00
addr = interface_ip;
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(notice_severity, LD_CONFIG,
"Learned IP address '%s' for local interface."
refactor all these tor_inet_ntoa idioms but don't refactor the ones that look messy
2010-10-02 03:31:09 +02:00
" Using that.", fmt_addr32(addr));
When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. svn:r6158
2006-03-14 23:51:15 +01:00
strlcpy(hostname, "<guessed from interfaces>", sizeof(hostname));
}
}
make it warn about internal IPs not only if we had to guess, but also if they used a hostname rather than an IP svn:r1285
2004-03-15 05:57:24 +01:00
}
Fix bug in tor_parse_string when Address is given as a dotted quad. Found by Ian Goldberg and Mashael AlSabah. Bugfix on 0.2.3.0-alpha dev.
2010-11-01 20:52:43 +01:00
} else {
addr = ntohl(in.s_addr); /* set addr so that addr_string is not
* illformed */
prevent servers from publishing an internal IP as their address svn:r1272
2004-03-14 19:12:59 +01:00
}
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
Tweak the fmt_addr32 code Clarify documentation, rename a local, and fix a memory leak.
2010-10-15 23:07:27 +02:00
addr_string = tor_dup_ip(addr);
Merge branch 'maint-0.2.2' Conflicts: src/or/config.c
2010-11-11 18:19:37 +01:00
if (is_internal_IP(addr, 0)) {
absolutely refuse to let people pick internal IP addresses if they're using the default dirservers. we're getting a big pile of a dozen or so servers that have picked private IP addresses despite the warning (presumably they don't even know they have logs) svn:r4354
2005-06-08 22:32:22 +02:00
/* make sure we're ok with publishing an internal IP */
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
if (!options->DirServers && !options->AlternateDirAuthority) {
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
/* if they are using the default dirservers, disallow internal IPs
* always. */
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(warn_severity, LD_CONFIG,
"Address '%s' resolves to private IP address '%s'. "
"Tor servers that use the default DirServers must have public "
Tweak the fmt_addr32 code Clarify documentation, rename a local, and fix a memory leak.
2010-10-15 23:07:27 +02:00
"IP addresses.", hostname, addr_string);
tor_free(addr_string);
Patch from goodell for task 131: Do not die on internal address if we are not publishing. svn:r4186
2005-05-09 06:31:00 +02:00
return -1;
absolutely refuse to let people pick internal IP addresses if they're using the default dirservers. we're getting a big pile of a dozen or so servers that have picked private IP addresses despite the warning (presumably they don't even know they have logs) svn:r4354
2005-06-08 22:32:22 +02:00
}
if (!explicit_ip) {
/* even if they've set their own dirservers, require an explicit IP if
* they're using an internal address. */
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
log_fn(warn_severity, LD_CONFIG, "Address '%s' resolves to private "
"IP address '%s'. Please set the Address config option to be "
Tweak the fmt_addr32 code Clarify documentation, rename a local, and fix a memory leak.
2010-10-15 23:07:27 +02:00
"the IP address you want to use.", hostname, addr_string);
tor_free(addr_string);
absolutely refuse to let people pick internal IP addresses if they're using the default dirservers. we're getting a big pile of a dozen or so servers that have picked private IP addresses despite the warning (presumably they don't even know they have logs) svn:r4354
2005-06-08 22:32:22 +02:00
return -1;
}
prevent servers from publishing an internal IP as their address svn:r1272
2004-03-14 19:12:59 +01:00
}
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
refactor all these tor_inet_ntoa idioms but don't refactor the ones that look messy
2010-10-02 03:31:09 +02:00
log_debug(LD_CONFIG, "Resolved Address to '%s'.", fmt_addr32(addr));
*addr_out = addr;
r8846@totoro: nickm | 2006-10-02 16:59:57 -0400 Move is_local_IP to config.c; have it check for same-/24; make it used only for reachability (not for banwidth, because that is probably not what we want). Fixes an XXX. svn:r8578
2006-10-02 23:00:35 +02:00
if (last_resolved_addr && last_resolved_addr != *addr_out) {
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
/* Leave this as a notice, regardless of the requested severity,
* at least until dynamic IP address support becomes bulletproof. */
if we think our ip address just changed, log what we thought it changed *to*. svn:r12551
2007-11-22 02:38:54 +01:00
log_notice(LD_NET,
"Your IP address seems to have changed to %s. Updating.",
Tweak the fmt_addr32 code Clarify documentation, rename a local, and fix a memory leak.
2010-10-15 23:07:27 +02:00
addr_string);
r11724@Kushana: nickm | 2006-12-28 14:22:35 -0500 Refactor and unify my-ip-addr-changed logic. Make change in IP address or in nameservers reset and relaunch DNS hijacking tests. svn:r9200
2006-12-28 22:29:20 +01:00
ip_address_changed(0);
Checkpoint in-progress fixes: Add 'testing' circuit purpose, for reachability testing. Notice when our IP changes, and reset stats. Try to pull down a directory via Tor to see if our DirPort is working. Try to extend a circuit back to us to see if our ORPort is working. Only publish a descriptor if they're both reachable. These mostly work, and I'd better get them in before I cause conflicts. svn:r3703
2005-02-27 10:47:01 +01:00
}
r11824@Kushana: nickm | 2007-01-03 17:15:28 -0500 control-spec: upcase arguments in status events; note unimplemented events individually r11825@Kushana: nickm | 2007-01-03 17:41:43 -0500 Implement EXTERNAL IP server status event. r11826@Kushana: nickm | 2007-01-03 17:47:10 -0500 Implement BAD_SERVER_DESCRIPTOR server status event. r11827@Kushana: nickm | 2007-01-03 18:01:56 -0500 Implement SOCKS_UNKNOWN_PROTOCOL and DANGEROUS_SOCKS client events. r11828@Kushana: nickm | 2007-01-03 18:23:22 -0500 Implement BUG controller events. Also, flush ERR-level status events just like ERR-level log messages. r11829@Kushana: nickm | 2007-01-03 23:37:27 -0500 Yet more status events: CLOCK_SKEW, GOOD/ACCEPTED_SERVER_DESCRIPTOR, {CHECKING_}REACHABILITY_{SUCCEEDED|FAILED} r11833@Kushana: nickm | 2007-01-05 16:56:37 -0500 Note some unimplementedness in control-spec.txt svn:r9279
2007-01-06 06:42:31 +01:00
if (last_resolved_addr != *addr_out) {
const char *method;
const char *h = hostname;
if (explicit_ip) {
method = "CONFIGURED";
h = NULL;
} else if (explicit_hostname) {
method = "RESOLVED";
} else if (from_interface) {
method = "INTERFACE";
h = NULL;
} else {
method = "GETHOSTNAME";
}
control_event_server_status(LOG_NOTICE,
"EXTERNAL_ADDRESS ADDRESS=%s METHOD=%s %s%s",
Tweak the fmt_addr32 code Clarify documentation, rename a local, and fix a memory leak.
2010-10-15 23:07:27 +02:00
addr_string, method, h?"HOSTNAME=":"", h);
r11824@Kushana: nickm | 2007-01-03 17:15:28 -0500 control-spec: upcase arguments in status events; note unimplemented events individually r11825@Kushana: nickm | 2007-01-03 17:41:43 -0500 Implement EXTERNAL IP server status event. r11826@Kushana: nickm | 2007-01-03 17:47:10 -0500 Implement BAD_SERVER_DESCRIPTOR server status event. r11827@Kushana: nickm | 2007-01-03 18:01:56 -0500 Implement SOCKS_UNKNOWN_PROTOCOL and DANGEROUS_SOCKS client events. r11828@Kushana: nickm | 2007-01-03 18:23:22 -0500 Implement BUG controller events. Also, flush ERR-level status events just like ERR-level log messages. r11829@Kushana: nickm | 2007-01-03 23:37:27 -0500 Yet more status events: CLOCK_SKEW, GOOD/ACCEPTED_SERVER_DESCRIPTOR, {CHECKING_}REACHABILITY_{SUCCEEDED|FAILED} r11833@Kushana: nickm | 2007-01-05 16:56:37 -0500 Note some unimplementedness in control-spec.txt svn:r9279
2007-01-06 06:42:31 +01:00
}
r8846@totoro: nickm | 2006-10-02 16:59:57 -0400 Move is_local_IP to config.c; have it check for same-/24; make it used only for reachability (not for banwidth, because that is probably not what we want). Fixes an XXX. svn:r8578
2006-10-02 23:00:35 +02:00
last_resolved_addr = *addr_out;
Start implementing the server side of the new directory protocol. Probably very buggy, since I can't actually run an authdir. Features - Generate and publish new network-status format - Code to cache and re-serve network-status objects generated by others. - Publish individual descriptors (by fingerprint, by "all", and by "tell me yours.") [Still needs compression logic] - Publish client and server recommended versions seprately. - Add digest of descriptor to routerinfo_t, so we can track them better, and length, so we can server them more easily. Cleanups - Unify code to sign directory-like things - Make resolve_my_address() able to tell you which name it wound up resolving. - Unify code to store and serve directory-like things so it all uses cached_dir_t. - Unify code to set the value of cached_dir_t objects. svn:r4835
2005-08-25 22:33:17 +02:00
if (hostname_out)
*hostname_out = tor_strdup(hostname);
Tweak the fmt_addr32 code Clarify documentation, rename a local, and fix a memory leak.
2010-10-15 23:07:27 +02:00
tor_free(addr_string);
prevent servers from publishing an internal IP as their address svn:r1272
2004-03-14 19:12:59 +01:00
return 0;
}
r17641@31-33-44: nickm | 2008-08-05 16:07:53 -0400 Initial conversion of uint32_t addr to tor_addr_t addr in connection_t and related types. Most of the Tor wire formats using these new types are in, but the code to generate and use it is not. This is a big patch. Let me know what it breaks for you. svn:r16435
2008-08-05 22:08:19 +02:00
/** Return true iff <b>addr</b> is judged to be on the same network as us, or
* on a private network.
r8846@totoro: nickm | 2006-10-02 16:59:57 -0400 Move is_local_IP to config.c; have it check for same-/24; make it used only for reachability (not for banwidth, because that is probably not what we want). Fixes an XXX. svn:r8578
2006-10-02 23:00:35 +02:00
*/
int
r17641@31-33-44: nickm | 2008-08-05 16:07:53 -0400 Initial conversion of uint32_t addr to tor_addr_t addr in connection_t and related types. Most of the Tor wire formats using these new types are in, but the code to generate and use it is not. This is a big patch. Let me know what it breaks for you. svn:r16435
2008-08-05 22:08:19 +02:00
is_local_addr(const tor_addr_t *addr)
r8846@totoro: nickm | 2006-10-02 16:59:57 -0400 Move is_local_IP to config.c; have it check for same-/24; make it used only for reachability (not for banwidth, because that is probably not what we want). Fixes an XXX. svn:r8578
2006-10-02 23:00:35 +02:00
{
r17641@31-33-44: nickm | 2008-08-05 16:07:53 -0400 Initial conversion of uint32_t addr to tor_addr_t addr in connection_t and related types. Most of the Tor wire formats using these new types are in, but the code to generate and use it is not. This is a big patch. Let me know what it breaks for you. svn:r16435
2008-08-05 22:08:19 +02:00
if (tor_addr_is_internal(addr, 0))
r8846@totoro: nickm | 2006-10-02 16:59:57 -0400 Move is_local_IP to config.c; have it check for same-/24; make it used only for reachability (not for banwidth, because that is probably not what we want). Fixes an XXX. svn:r8578
2006-10-02 23:00:35 +02:00
return 1;
the other half of the is_local_IP patch svn:r8581
2006-10-03 07:45:58 +02:00
/* Check whether ip is on the same /24 as we are. */
if (get_options()->EnforceDistinctSubnets == 0)
return 0;
r17641@31-33-44: nickm | 2008-08-05 16:07:53 -0400 Initial conversion of uint32_t addr to tor_addr_t addr in connection_t and related types. Most of the Tor wire formats using these new types are in, but the code to generate and use it is not. This is a big patch. Let me know what it breaks for you. svn:r16435
2008-08-05 22:08:19 +02:00
if (tor_addr_family(addr) == AF_INET) {
Triage the XXX022 and XXX021 comments remaining in the code Remove some, postpone others, leave some alone. Now the only remaining XXX022s are ones that seem important to fix or investigate.
2011-03-25 21:01:16 +01:00
/*XXXX023 IP6 what corresponds to an /24? */
r17641@31-33-44: nickm | 2008-08-05 16:07:53 -0400 Initial conversion of uint32_t addr to tor_addr_t addr in connection_t and related types. Most of the Tor wire formats using these new types are in, but the code to generate and use it is not. This is a big patch. Let me know what it breaks for you. svn:r16435
2008-08-05 22:08:19 +02:00
uint32_t ip = tor_addr_to_ipv4h(addr);
/* It's possible that this next check will hit before the first time
* resolve_my_address actually succeeds. (For clients, it is likely that
* resolve_my_address will never be called at all). In those cases,
* last_resolved_addr will be 0, and so checking to see whether ip is on
* the same /24 as last_resolved_addr will be the same as checking whether
* it was on net 0, which is already done by is_internal_IP.
*/
Fix compile warnings on Snow Leopard Big thanks to nickm and arma for helping me with this!
2009-09-01 05:23:47 +02:00
if ((last_resolved_addr & (uint32_t)0xffffff00ul)
== (ip & (uint32_t)0xffffff00ul))
r17641@31-33-44: nickm | 2008-08-05 16:07:53 -0400 Initial conversion of uint32_t addr to tor_addr_t addr in connection_t and related types. Most of the Tor wire formats using these new types are in, but the code to generate and use it is not. This is a big patch. Let me know what it breaks for you. svn:r16435
2008-08-05 22:08:19 +02:00
return 1;
}
r8846@totoro: nickm | 2006-10-02 16:59:57 -0400 Move is_local_IP to config.c; have it check for same-/24; make it used only for reachability (not for banwidth, because that is probably not what we want). Fixes an XXX. svn:r8578
2006-10-02 23:00:35 +02:00
return 0;
}
polish r9933-r9994 svn:r10335
2007-05-25 21:41:31 +02:00
/** Release storage held by <b>options</b>. */
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
static void
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_free(const config_format_t *fmt, void *options)
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
{
make free_options() and init_options() use the big array. also remove obsolete config_assign_defaults(). svn:r2672
2004-11-04 09:26:34 +01:00
int i;
*_free functions now accept NULL Some *_free functions threw asserts when passed NULL. Now all of them accept NULL as input and perform no action when called that way. This gains us consistence for our free functions, and allows some code simplifications where an explicit null check is no longer necessary.
2009-09-28 16:37:01 +02:00
if (!options)
return;
tor_assert(fmt);
salvage another 300 bytes per hup svn:r3106
2004-12-07 10:11:38 +01:00
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
for (i=0; fmt->vars[i].name; ++i)
option_clear(fmt, options, &(fmt->vars[i]));
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
if (fmt->extra) {
config_line_t **linep = STRUCT_VAR_P(options, fmt->extra->var_offset);
config_free_lines(*linep);
*linep = NULL;
}
salvage another 300 bytes per hup svn:r3106
2004-12-07 10:11:38 +01:00
tor_free(options);
move closer to being able to reload config on HUP rename APPort to SocksPort introduce new tor_free() macro svn:r642
2003-10-21 11:48:17 +02:00
}
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
/** Return true iff a and b contain identical keys and values in identical
* order. */
static int
config_lines_eq(config_line_t *a, config_line_t *b)
{
while (a && b) {
if (strcasecmp(a->key, b->key) || strcmp(a->value, b->value))
return 0;
a = a->next;
b = b->next;
}
if (a || b)
return 0;
return 1;
}
If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". svn:r12923
2007-12-22 10:04:46 +01:00
/** Return true iff the option <b>name</b> has the same value in <b>o1</b>
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
* and <b>o2</b>. Must not be called for LINELIST_S or OBSOLETE options.
*/
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
option_is_same(const config_format_t *fmt,
const or_options_t *o1, const or_options_t *o2,
const char *name)
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
{
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *c1, *c2;
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
int r = 1;
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
CHECK(fmt, o1);
CHECK(fmt, o2);
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
c1 = get_assigned_option(fmt, o1, name, 0);
c2 = get_assigned_option(fmt, o2, name, 0);
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
r = config_lines_eq(c1, c2);
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
config_free_lines(c1);
config_free_lines(c2);
return r;
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
/** Copy storage held by <b>old</b> into a new or_options_t and return it. */
static or_options_t *
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
options_dup(const config_format_t *fmt, const or_options_t *old)
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
{
fix some dereferences svn:r2726
2004-11-09 06:59:22 +01:00
or_options_t *newopts;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
int i;
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *line;
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
newopts = config_alloc(fmt);
for (i=0; fmt->vars[i].name; ++i) {
if (fmt->vars[i].type == CONFIG_TYPE_LINELIST_S)
During options_dup, don't copy LINELIST_S or OBSOLETE options. And recognize obsolete options now in config_get_assigned_option(). svn:r2729
2004-11-09 07:18:17 +01:00
continue;
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
if (fmt->vars[i].type == CONFIG_TYPE_OBSOLETE)
During options_dup, don't copy LINELIST_S or OBSOLETE options. And recognize obsolete options now in config_get_assigned_option(). svn:r2729
2004-11-09 07:18:17 +01:00
continue;
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
line = get_assigned_option(fmt, old, fmt->vars[i].name, 0);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
if (line) {
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
char *msg = NULL;
if (config_assign(fmt, newopts, line, 0, 0, &msg) < 0) {
r12074@catbus: nickm | 2007-03-04 15:11:43 -0500 Make all LD_BUG log messsages get prefixed with "Bug: ". Remove manually-generated "Bug: "s from log-messages. (Apparently, we remembered to add them about 40% of the time.) svn:r9733
2007-03-04 21:11:46 +01:00
log_err(LD_BUG, "Config_get_assigned_option() generated "
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
"something we couldn't config_assign(): %s", msg);
tor_free(msg);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
tor_assert(0);
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
}
config_free_lines(line);
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
}
fix some dereferences svn:r2726
2004-11-09 06:59:22 +01:00
return newopts;
Support code for resettable options, and option sets. Still needs validate-and-then-replace logic svn:r2679
2004-11-04 23:31:50 +01:00
}
Fix segfault in unit tests svn:r4760
2005-08-09 20:49:43 +02:00
/** Return a new empty or_options_t. Used for testing. */
or_options_t *
options_new(void)
{
return config_alloc(&options_format);
}
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
/** Set <b>options</b> to hold reasonable defaults for most options.
* Each option defaults to zero. */
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
void
options_init(or_options_t *options)
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
{
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
config_init(&options_format, options);
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
}
other cleanups that have been sitting in my sandbox svn:r13649
2008-02-21 10:01:32 +01:00
/** Set all vars in the configuration object <b>options</b> to their default
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
* values. */
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
static void
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_init(const config_format_t *fmt, void *options)
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
{
make free_options() and init_options() use the big array. also remove obsolete config_assign_defaults(). svn:r2672
2004-11-04 09:26:34 +01:00
int i;
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *var;
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
CHECK(fmt, options);
make free_options() and init_options() use the big array. also remove obsolete config_assign_defaults(). svn:r2672
2004-11-04 09:26:34 +01:00
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
for (i=0; fmt->vars[i].name; ++i) {
var = &fmt->vars[i];
Normalize space: add one between every control keyword and control clause. svn:r3003
2004-11-28 10:05:49 +01:00
if (!var->initvalue)
make free_options() and init_options() use the big array. also remove obsolete config_assign_defaults(). svn:r2672
2004-11-04 09:26:34 +01:00
continue; /* defaults to NULL or 0 */
checkpoint: clean up and document the three ways to call config_assign() and reduce code duplication in config_free() and option_is_same(). svn:r5040
2005-09-14 04:07:35 +02:00
option_reset(fmt, options, var, 1);
make free_options() and init_options() use the big array. also remove obsolete config_assign_defaults(). svn:r2672
2004-11-04 09:26:34 +01:00
}
move closer to being able to reload config on HUP rename APPort to SocksPort introduce new tor_free() macro svn:r642
2003-10-21 11:48:17 +02:00
}
r11629@catbus: nickm | 2007-02-02 15:06:17 -0500 Removing the last DOCDOC comment hurt so much that I had to use Doxygen to identify undocumented macros and comments, and add 150 more DOCDOCs to point out where they were. Oops. Hey, kids! Fixing some of these could be your first Tor patch! svn:r9477
2007-02-02 21:06:43 +01:00
/** Allocate and return a new string holding the written-out values of the vars
r11598@Kushana: nickm | 2006-12-15 15:59:00 -0500 Add internal documentation for a bunch of configuration options. We should do something to keep this list, the canonical list, the tor.1 list, and the torrc.complete list in sync. svn:r9133
2006-12-15 22:27:19 +01:00
* in 'options'. If 'minimal', do not write out any default-valued vars.
* Else, if comment_defaults, write default values as comments.
*/
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
static char *
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
config_dump(const config_format_t *fmt, const void *default_options,
const void *options, int minimal,
r11598@Kushana: nickm | 2006-12-15 15:59:00 -0500 Add internal documentation for a bunch of configuration options. We should do something to keep this list, the canonical list, the tor.1 list, and the torrc.complete list in sync. svn:r9133
2006-12-15 22:27:19 +01:00
int comment_defaults)
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
{
smartlist_t *elements;
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
const or_options_t *defaults = default_options;
void *defaults_tmp = NULL;
Fix another memleak. I love you, valgrind! svn:r5794
2006-01-11 05:09:40 +01:00
config_line_t *line, *assigned;
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
char *result;
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
int i;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
char *msg = NULL;
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
if (defaults == NULL) {
defaults = defaults_tmp = config_alloc(fmt);
config_init(fmt, defaults_tmp);
}
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
/* XXX use a 1 here so we don't add a new log line while dumping */
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
if (default_options == NULL) {
if (fmt->validate_fn(NULL, defaults_tmp, 1, &msg) < 0) {
log_err(LD_BUG, "Failed to validate default config.");
tor_free(msg);
tor_assert(0);
}
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
}
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
elements = smartlist_new();
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
for (i=0; fmt->vars[i].name; ++i) {
r11598@Kushana: nickm | 2006-12-15 15:59:00 -0500 Add internal documentation for a bunch of configuration options. We should do something to keep this list, the canonical list, the tor.1 list, and the torrc.complete list in sync. svn:r9133
2006-12-15 22:27:19 +01:00
int comment_option = 0;
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
if (fmt->vars[i].type == CONFIG_TYPE_OBSOLETE ||
fmt->vars[i].type == CONFIG_TYPE_LINELIST_S)
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
continue;
Implement the common case of ATTACHSTREAM. svn:r3751
2005-03-12 05:22:01 +01:00
/* Don't save 'hidden' control variables. */
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
if (!strcmpstart(fmt->vars[i].name, "__"))
Implement the common case of ATTACHSTREAM. svn:r3751
2005-03-12 05:22:01 +01:00
continue;
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
if (minimal && option_is_same(fmt, options, defaults, fmt->vars[i].name))
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
continue;
r11598@Kushana: nickm | 2006-12-15 15:59:00 -0500 Add internal documentation for a bunch of configuration options. We should do something to keep this list, the canonical list, the tor.1 list, and the torrc.complete list in sync. svn:r9133
2006-12-15 22:27:19 +01:00
else if (comment_defaults &&
option_is_same(fmt, options, defaults, fmt->vars[i].name))
comment_option = 1;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
r17499@catbus: nickm | 2008-01-07 13:39:46 -0500 Bugfix on fix for 557: Make values containing special characters work right with getconf, setconf, and saveconf. Document this in control-spec.txt svn:r13056
2008-01-07 19:54:55 +01:00
line = assigned = get_assigned_option(fmt, options, fmt->vars[i].name, 1);
Fix bug 240: dont dump descriptions when not dumping actual config lines. svn:r5804
2006-01-11 20:43:14 +01:00
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
for (; line; line = line->next) {
Try to use smartlist_add_asprintf consistently (To ensure correctness, in every case, make sure that the temporary variable is deleted, renamed, or lowered in scope, so we can't have any bugs related to accidentally relying on the no-longer-filled variable.)
2012-01-11 19:44:10 +01:00
smartlist_add_asprintf(elements, "%s%s %s\n",
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
comment_option ? "# " : "",
line->key, line->value);
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
}
Fix another memleak. I love you, valgrind! svn:r5794
2006-01-11 05:09:40 +01:00
config_free_lines(assigned);
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
}
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
if (fmt->extra) {
line = *(config_line_t**)STRUCT_VAR_P(options, fmt->extra->var_offset);
for (; line; line = line->next) {
Try to use smartlist_add_asprintf consistently (To ensure correctness, in every case, make sure that the temporary variable is deleted, renamed, or lowered in scope, so we can't have any bugs related to accidentally relying on the no-longer-filled variable.)
2012-01-11 19:44:10 +01:00
smartlist_add_asprintf(elements, "%s %s\n", line->key, line->value);
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
}
}
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
result = smartlist_join_strings(elements, "", 0, NULL);
SMARTLIST_FOREACH(elements, char *, cp, tor_free(cp));
smartlist_free(elements);
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
if (defaults_tmp)
config_free(fmt, defaults_tmp);
Write most of the implementation for SAVECONF. svn:r2820
2004-11-12 21:09:54 +01:00
return result;
}
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
/** Return a string containing a possible configuration file that would give
* the configuration in <b>options</b>. If <b>minimal</b> is true, do not
* include options that are the same as Tor's defaults.
*/
New controller command "getinfo config-text" It returns the contents that Tor would write if you send it a SAVECONF command, so the controller can write the file to disk itself.
2009-12-14 01:21:06 +01:00
char *
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
options_dump(const or_options_t *options, int minimal)
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
{
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
return config_dump(&options_format, global_default_options,
options, minimal, 0);
Parameterize configuration parsing functions svn:r4635
2005-07-22 22:37:42 +02:00
}
r11629@catbus: nickm | 2007-02-02 15:06:17 -0500 Removing the last DOCDOC comment hurt so much that I had to use Doxygen to identify undocumented macros and comments, and add 150 more DOCDOCs to point out where they were. Oops. Hey, kids! Fixing some of these could be your first Tor patch! svn:r9477
2007-02-02 21:06:43 +01:00
/** Return 0 if every element of sl is a string holding a decimal
a few typo fixes svn:r4738
2005-08-07 23:58:23 +02:00
* representation of a port number, or if sl is NULL.
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
* Otherwise set *msg and return -1. */
Annotate circuits w/ whether they aim to contain high uptime nodes and/or high capacity nodes. When building circuits, choose appropriate nodes. New config option LongLivedPorts to indicate application streams that will want high uptime circuits. When attaching a stream to a circuit, pay attention to its requirements. This means that every single node in an intro rend circuit, not just the last one, will have a minimum uptime. Boost the min uptime from an hour to 24 hours. svn:r3339
2005-01-12 05:58:23 +01:00
static int
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
validate_ports_csv(smartlist_t *sl, const char *name, char **msg)
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
{
Annotate circuits w/ whether they aim to contain high uptime nodes and/or high capacity nodes. When building circuits, choose appropriate nodes. New config option LongLivedPorts to indicate application streams that will want high uptime circuits. When attaching a stream to a circuit, pay attention to its requirements. This means that every single node in an intro rend circuit, not just the last one, will have a minimum uptime. Boost the min uptime from an hour to 24 hours. svn:r3339
2005-01-12 05:58:23 +01:00
int i;
tor_assert(name);
Re-normalize whitespace; make "make check-spaces" happy svn:r3566
2005-02-05 22:42:46 +01:00
if (!sl)
Annotate circuits w/ whether they aim to contain high uptime nodes and/or high capacity nodes. When building circuits, choose appropriate nodes. New config option LongLivedPorts to indicate application streams that will want high uptime circuits. When attaching a stream to a circuit, pay attention to its requirements. This means that every single node in an intro rend circuit, not just the last one, will have a minimum uptime. Boost the min uptime from an hour to 24 hours. svn:r3339
2005-01-12 05:58:23 +01:00
return 0;
SMARTLIST_FOREACH(sl, const char *, cp,
{
i = atoi(cp);
if (i < 1 || i > 65535) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg, "Port '%s' out of range in %s", cp, name);
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
return -1;
Annotate circuits w/ whether they aim to contain high uptime nodes and/or high capacity nodes. When building circuits, choose appropriate nodes. New config option LongLivedPorts to indicate application streams that will want high uptime circuits. When attaching a stream to a circuit, pay attention to its requirements. This means that every single node in an intro rend circuit, not just the last one, will have a minimum uptime. Boost the min uptime from an hour to 24 hours. svn:r3339
2005-01-12 05:58:23 +01:00
}
});
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
return 0;
Annotate circuits w/ whether they aim to contain high uptime nodes and/or high capacity nodes. When building circuits, choose appropriate nodes. New config option LongLivedPorts to indicate application streams that will want high uptime circuits. When attaching a stream to a circuit, pay attention to its requirements. This means that every single node in an intro rend circuit, not just the last one, will have a minimum uptime. Boost the min uptime from an hour to 24 hours. svn:r3339
2005-01-12 05:58:23 +01:00
}
if you're using relaybandwidthrate and relaybandwidthburst, make sure that's reflected in your router descriptor. svn:r10114
2007-05-04 11:20:13 +02:00
/** If <b>value</b> exceeds ROUTER_MAX_DECLARED_BANDWIDTH, write
* a complaint into *<b>msg</b> using string <b>desc</b>, and return -1.
* Else return 0.
*/
static int
r13090@catbus: nickm | 2007-05-30 03:17:57 -0400 If the user makes a torrc that exceeds the bandwidth cap by one byte, let them have it. svn:r10411
2007-05-30 09:18:00 +02:00
ensure_bandwidth_cap(uint64_t *value, const char *desc, char **msg)
if you're using relaybandwidthrate and relaybandwidthburst, make sure that's reflected in your router descriptor. svn:r10114
2007-05-04 11:20:13 +02:00
{
r13103@catbus: nickm | 2007-05-31 14:40:18 -0400 Fix patch r10411; fix from robert hogan. svn:r10422
2007-05-31 20:48:31 +02:00
if (*value > ROUTER_MAX_DECLARED_BANDWIDTH) {
r13090@catbus: nickm | 2007-05-30 03:17:57 -0400 If the user makes a torrc that exceeds the bandwidth cap by one byte, let them have it. svn:r10411
2007-05-30 09:18:00 +02:00
/* This handles an understandable special case where somebody says "2gb"
* whereas our actual maximum is 2gb-1 (INT_MAX) */
--*value;
}
if (*value > ROUTER_MAX_DECLARED_BANDWIDTH) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg, "%s ("U64_FORMAT") must be at most %d",
desc, U64_PRINTF_ARG(*value),
ROUTER_MAX_DECLARED_BANDWIDTH);
if you're using relaybandwidthrate and relaybandwidthburst, make sure that's reflected in your router descriptor. svn:r10114
2007-05-04 11:20:13 +02:00
return -1;
}
return 0;
}
If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". svn:r12923
2007-12-22 10:04:46 +01:00
/** Parse an authority type from <b>options</b>-\>PublishServerDescriptor
* and write it to <b>options</b>-\>_PublishServerDescriptor. Treat "1"
* as "v2,v3" unless BridgeRelay is 1, in which case treat it as "bridge".
* Treat "0" as "".
Correctly check for bad options to the "PublishServerDescriptor" config option. Bugfix on 0.2.0.1-alpha; reported by Matt Edman. svn:r11968
2007-10-16 05:26:39 +02:00
* Return 0 on success or -1 if not a recognized authority type (in which
If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". svn:r12923
2007-12-22 10:04:46 +01:00
* case the value of _PublishServerDescriptor is undefined). */
Change the PublishServerDescriptor config option from a boolean into a string: "v1", "v2", bridge", "". Continue to support "0" and "1". svn:r10136
2007-05-08 11:09:26 +02:00
static int
If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". svn:r12923
2007-12-22 10:04:46 +01:00
compute_publishserverdescriptor(or_options_t *options)
Change the PublishServerDescriptor config option from a boolean into a string: "v1", "v2", bridge", "". Continue to support "0" and "1". svn:r10136
2007-05-08 11:09:26 +02:00
{
If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". svn:r12923
2007-12-22 10:04:46 +01:00
smartlist_t *list = options->PublishServerDescriptor;
Automated rename from authority_type_t to dirinfo_type_t We were already overloading this type to mean "a directory that can serve us X" in addition to "a directory that is an authority for X."
2010-11-08 20:27:36 +01:00
dirinfo_type_t *auth = &options->_PublishServerDescriptor;
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
*auth = NO_DIRINFO;
writing instructions before you release means you can make sure your instructions don't seg fault svn:r10984
2007-07-30 10:42:47 +02:00
if (!list) /* empty list, answer is none */
return 0;
r12697@catbus: nickm | 2007-05-09 00:15:40 -0400 Change authority_type_t to a set of flags; use it more consistently. svn:r10144
2007-05-09 06:15:46 +02:00
SMARTLIST_FOREACH(list, const char *, string, {
if (!strcasecmp(string, "v1"))
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
*auth |= V1_DIRINFO;
If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". svn:r12923
2007-12-22 10:04:46 +01:00
else if (!strcmp(string, "1"))
if (options->BridgeRelay)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
*auth |= BRIDGE_DIRINFO;
If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". svn:r12923
2007-12-22 10:04:46 +01:00
else
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
*auth |= V2_DIRINFO | V3_DIRINFO;
r12697@catbus: nickm | 2007-05-09 00:15:40 -0400 Change authority_type_t to a set of flags; use it more consistently. svn:r10144
2007-05-09 06:15:46 +02:00
else if (!strcasecmp(string, "v2"))
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
*auth |= V2_DIRINFO;
r12898@catbus: nickm | 2007-05-22 13:11:04 -0400 More v3 directory code: have authorities load certificates; have everybody store certificates to disk and load them; provide a way to configure v3 authorities. svn:r10293
2007-05-22 19:58:25 +02:00
else if (!strcasecmp(string, "v3"))
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
*auth |= V3_DIRINFO;
r12697@catbus: nickm | 2007-05-09 00:15:40 -0400 Change authority_type_t to a set of flags; use it more consistently. svn:r10144
2007-05-09 06:15:46 +02:00
else if (!strcasecmp(string, "bridge"))
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
*auth |= BRIDGE_DIRINFO;
r12697@catbus: nickm | 2007-05-09 00:15:40 -0400 Change authority_type_t to a set of flags; use it more consistently. svn:r10144
2007-05-09 06:15:46 +02:00
else if (!strcasecmp(string, "hidserv"))
Ignore and warn about "PublishServerDescriptor hidserv" Fixes #2408.
2011-01-18 12:54:56 +01:00
log_warn(LD_CONFIG,
"PublishServerDescriptor hidserv is invalid. See "
"PublishHidServDescriptors.");
If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". svn:r12923
2007-12-22 10:04:46 +01:00
else if (!strcasecmp(string, "") || !strcmp(string, "0"))
r12697@catbus: nickm | 2007-05-09 00:15:40 -0400 Change authority_type_t to a set of flags; use it more consistently. svn:r10144
2007-05-09 06:15:46 +02:00
/* no authority */;
else
Correctly check for bad options to the "PublishServerDescriptor" config option. Bugfix on 0.2.0.1-alpha; reported by Matt Edman. svn:r11968
2007-10-16 05:26:39 +02:00
return -1;
r12697@catbus: nickm | 2007-05-09 00:15:40 -0400 Change authority_type_t to a set of flags; use it more consistently. svn:r10144
2007-05-09 06:15:46 +02:00
});
Change the PublishServerDescriptor config option from a boolean into a string: "v1", "v2", bridge", "". Continue to support "0" and "1". svn:r10136
2007-05-08 11:09:26 +02:00
return 0;
}
Cleanup on time-relaqted constants. New conventions: 1) Surround all constants by (parens), whether we'll be using them in a denominator or not. 2) Express all time periods as products (24*60*60), not as multiplied-out constants (86400). 3) Comments like "(60*60) /* one hour */" are as pointless as comments like "c = a + b; /* set c to the sum of a and b */". Remove them. 4) All time periods should be #defined constants, not given inline. 5) All time periods should have doxygen comments. 6) All time periods, unless specified, are in seconds. It's not necessary to say so. To summarize, the old (lack of) style would allow: #define FOO_RETRY_INTERVAL 60*60 /* one hour (seconds) */ next_try = now + 3600; The new style is: /** How often do we reattempt foo? */ #define FOO_RETRY_INTERVAL (60*60) next_try = now + RETRY_INTERVAL; svn:r6142
2006-03-12 23:48:18 +01:00
/** Lowest allowable value for RendPostPeriod; if this is too low, hidden
* services can overload the directory system. */
r12012@catbus: nickm | 2007-02-28 18:25:18 -0500 Twiddle constants wrt uploading hidden service descriptors. svn:r9694
2007-03-01 01:41:28 +01:00
#define MIN_REND_POST_PERIOD (10*60)
Cleanup on time-relaqted constants. New conventions: 1) Surround all constants by (parens), whether we'll be using them in a denominator or not. 2) Express all time periods as products (24*60*60), not as multiplied-out constants (86400). 3) Comments like "(60*60) /* one hour */" are as pointless as comments like "c = a + b; /* set c to the sum of a and b */". Remove them. 4) All time periods should be #defined constants, not given inline. 5) All time periods should have doxygen comments. 6) All time periods, unless specified, are in seconds. It's not necessary to say so. To summarize, the old (lack of) style would allow: #define FOO_RETRY_INTERVAL 60*60 /* one hour (seconds) */ next_try = now + 3600; The new style is: /** How often do we reattempt foo? */ #define FOO_RETRY_INTERVAL (60*60) next_try = now + RETRY_INTERVAL; svn:r6142
2006-03-12 23:48:18 +01:00
r11598@Kushana: nickm | 2006-12-15 15:59:00 -0500 Add internal documentation for a bunch of configuration options. We should do something to keep this list, the canonical list, the tor.1 list, and the torrc.complete list in sync. svn:r9133
2006-12-15 22:27:19 +01:00
/** Highest allowable value for RendPostPeriod. */
Cleanup on time-relaqted constants. New conventions: 1) Surround all constants by (parens), whether we'll be using them in a denominator or not. 2) Express all time periods as products (24*60*60), not as multiplied-out constants (86400). 3) Comments like "(60*60) /* one hour */" are as pointless as comments like "c = a + b; /* set c to the sum of a and b */". Remove them. 4) All time periods should be #defined constants, not given inline. 5) All time periods should have doxygen comments. 6) All time periods, unless specified, are in seconds. It's not necessary to say so. To summarize, the old (lack of) style would allow: #define FOO_RETRY_INTERVAL 60*60 /* one hour (seconds) */ next_try = now + 3600; The new style is: /** How often do we reattempt foo? */ #define FOO_RETRY_INTERVAL (60*60) next_try = now + RETRY_INTERVAL; svn:r6142
2006-03-12 23:48:18 +01:00
#define MAX_DIR_PERIOD (MIN_ONION_KEY_LIFETIME/2)
Clip the MaxCircuitDirtiness config option to a minimum of 10 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.0.1-rc. Patch by Sebastian. Clip the CircuitBuildTimeout to a minimum of 30 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.1.17-rc. Patch by Sebastian. svn:r17657
2008-12-17 23:32:17 +01:00
/** Lowest allowable value for MaxCircuitDirtiness; if this is too low, Tor
* will generate too many circuits and potentially overload the network. */
#define MIN_MAX_CIRCUIT_DIRTINESS 10
add a minimum for CircuitStreamTimeout, plus a man page plus some other unrelated touchups that have been sitting in my sandbox
2009-11-22 13:15:30 +01:00
/** Lowest allowable value for CircuitStreamTimeout; if this is too low, Tor
* will generate too many circuits and potentially overload the network. */
#define MIN_CIRCUIT_STREAM_TIMEOUT 10
Initial heartbeat subsystem commit. Sets: * Documentation * Logging domain * Configuration option * Scheduled event * Makefile It also creates status.c and the log_heartbeat() function. All code was written by Sebastian Hahn. Commit message was written by me (George Kadianakis).
2010-12-01 02:24:03 +01:00
/** Lowest allowable value for HeartbeatPeriod; if this is too low, we might
* expose more information than we're comfortable with. */
#define MIN_HEARTBEAT_PERIOD (30*60)
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Return 0 if every setting in <b>options</b> is reasonable, and a
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
* permissible transition from <b>old_options</b>. Else return -1.
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
* Should have no side effects, except for normalizing the contents of
* <b>options</b>.
hack to address bug 238. needs a better fix, but not for 0.1.1.11. svn:r5771
2006-01-10 21:27:47 +01:00
*
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
* On error, tor_strdup an error explanation into *<b>msg</b>.
*
hack to address bug 238. needs a better fix, but not for 0.1.1.11. svn:r5771
2006-01-10 21:27:47 +01:00
* XXX
* If <b>from_setconf</b>, we were called by the controller, and our
* Log line should stay empty. If it's 0, then give us a default log
* if there are no logs defined.
Warn when binding to a public address for socks. This warning needs to be improved. Also, make options_validate() only warn when an option is changed to a bad value.. (Previously, it warned when an option was bad and any option was changed.) svn:r5532
2005-12-08 19:56:32 +01:00
*/
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
static int
hack to address bug 238. needs a better fix, but not for 0.1.1.11. svn:r5771
2006-01-10 21:27:47 +01:00
options_validate(or_options_t *old_options, or_options_t *options,
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
int from_setconf, char **msg)
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
{
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
int i;
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *cl;
more cleanups, including a shiny new XXX012 svn:r9250
2007-01-03 11:30:26 +01:00
const char *uname = get_uname();
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
int n_ports=0;
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
#define REJECT(arg) \
r13477@catbus: nickm | 2007-06-17 14:22:03 -0400 Sun CC likes to give warnings for the do { } while(0) construction for making statement-like macros. Define STMT_BEGIN/STMT_END macros that do the right thing, and use them everywhere. svn:r10645
2007-06-17 20:22:39 +02:00
STMT_BEGIN *msg = tor_strdup(arg); return -1; STMT_END
#define COMPLAIN(arg) STMT_BEGIN log(LOG_WARN, LD_CONFIG, arg); STMT_END
disallow changing from ORPort==0 to ORPort>0 on hup svn:r1172
2004-02-29 00:30:41 +01:00
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
tor_assert(msg);
*msg = NULL;
Warn if running obsolete windows as a server. svn:r5961
2006-02-11 22:26:40 +01:00
if (server_mode(options) &&
(!strcmpstart(uname, "Windows 95") ||
!strcmpstart(uname, "Windows 98") ||
!strcmpstart(uname, "Windows Me"))) {
log(LOG_WARN, LD_CONFIG, "Tor is running as a server, but you are "
Retain unrecognized options in state file, so that we can be forward-compatible. svn:r5985
2006-02-12 23:28:30 +01:00
"running %s; this probably won't work. See "
fix comments and other typos
2009-06-20 09:21:52 +02:00
"https://wiki.torproject.org/TheOnionRouter/TorFAQ#ServerOS "
more cleanups, including a shiny new XXX012 svn:r9250
2007-01-03 11:30:26 +01:00
"for details.", uname);
Warn if running obsolete windows as a server. svn:r5961
2006-02-11 22:26:40 +01:00
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (parse_ports(options, 1, msg, &n_ports) < 0)
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
return -1;
Warn when binding to a public address for socks. This warning needs to be improved. Also, make options_validate() only warn when an option is changed to a bad value.. (Previously, it warned when an option was bad and any option was changed.) svn:r5532
2005-12-08 19:56:32 +01:00
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
if (validate_data_directory(options)<0)
REJECT("Invalid DataDirectory");
Normalize DataDirectory in options_validate, making SIGHUP survivable. svn:r2722
2004-11-09 06:26:49 +01:00
abstract ORPort/SocksPort checks into server_mode(), proxy_mode(), clique_mode(), etc. Dont change underlying comments. svn:r2054
2004-07-18 23:47:04 +02:00
if (options->Nickname == NULL) {
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
if (server_mode(options)) {
r8776@totoro: nickm | 2006-09-29 00:50:46 -0400 Reserve the nickname "Unnamed" for routers that can't pick a hostname; any router can call itself Unnamed; directory servers will never allocate Unnamed to any particular router; clients won't believe that any router is the canonical Unnamed. svn:r8529
2006-09-29 06:51:28 +02:00
options->Nickname = tor_strdup(UNNAMED_ROUTER_NICKNAME);
only pick a default nickname if you're a server svn:r2134
2004-08-04 04:15:22 +02:00
}
abstract ORPort/SocksPort checks into server_mode(), proxy_mode(), clique_mode(), etc. Dont change underlying comments. svn:r2054
2004-07-18 23:47:04 +02:00
} else {
hey, we have a function for that svn:r4781
2005-08-15 10:51:20 +02:00
if (!is_legal_nickname(options->Nickname)) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
"Nickname '%s' is wrong length or contains illegal characters.",
options->Nickname);
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
return -1;
warn and fail if server chose a nickname with illegal characters svn:r1318
2004-03-20 21:28:53 +01:00
}
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
}
whine at you if you're a server and you don't set your contactinfo svn:r4178
2005-05-04 22:45:38 +02:00
if (server_mode(options) && !options->ContactInfo)
In my private little universe, terminals are still 80 columns. Impose a 160-character-per-line limit; this will creep down. svn:r5548
2005-12-09 06:37:26 +01:00
log(LOG_NOTICE, LD_CONFIG, "Your ContactInfo config option is not set. "
"Please consider setting it, so we can contact you if your server is "
"misconfigured or something else goes wrong.");
whine at you if you're a server and you don't set your contactinfo svn:r4178
2005-05-04 22:45:38 +02:00
a first go at fixing a bug matt edman reported: when he tries to "setconf log" via the controller, it is friendly and gives him a log to stdout, even though he didn't ask for one. svn:r5562
2005-12-11 11:01:21 +01:00
/* Special case on first boot if no Log options are given. */
Make --quiet and --hush apply to default logs, not only initial logs Fixes bug 3550; bugfix on 0.2.0.10-alpha (where --quiet was introduced).
2011-08-04 20:23:51 +02:00
if (!options->Logs && !options->RunAsDaemon && !from_setconf) {
if (quiet_level == 0)
config_line_append(&options->Logs, "Log", "notice stdout");
else if (quiet_level == 1)
config_line_append(&options->Logs, "Log", "warn stdout");
}
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
Rename cross-format config.c stuff to config_* and or_options_t-specific stuff to option[s]_* svn:r4641
2005-07-23 03:58:05 +02:00
if (options_init_logs(options, 1)<0) /* Validate the log(s) */
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
REJECT("Failed to validate Log options. See logs for details.");
Validate log options during options_validate(). Reject unrecognized log types like "Log notice magic-twanger". Also, make sure User and Group stay the same, and do not crash when somebody tries to change PidFile to NULL. svn:r2778
2004-11-10 04:40:30 +01:00
Allow servers with no hostname or IP address to learn their IP address by asking the directory authorities. This code only kicks in when you would normally have exited with a "no address" error. This design is flawed, though, since the X-Your-Address-Is header is not authenticated, and doing it this way introduces too many new attacks. The right answer is to give IP address hints inside the HELLO cell; much of this code can be reused when we switch. svn:r6774
2006-07-17 08:35:06 +02:00
if (authdir_mode(options)) {
o Handle servers with dynamic IP addresses: don't replace options->Address with the resolved one at startup. o detect our address right before we make a routerinfo each time. o external IP vs bind-IP. Already done, just use options->Address. o OutboundBindAddress config option, to bind to a specific IP address for outgoing connect()s. svn:r2241
2004-08-16 13:43:18 +02:00
/* confirm that our address isn't broken, so we can complain now */
uint32_t tmp;
parameterize the loudness of resolve_my_address(), and call things IP addresses, not IPs. svn:r6764
2006-07-15 22:26:05 +02:00
if (resolve_my_address(LOG_WARN, options, &tmp, NULL) < 0)
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
REJECT("Failed to resolve/guess local address. See logs for details.");
Store options->Address as IP, not hostname And figure it out while reading config, not every time you rebuild the descriptor svn:r1226
2004-03-04 02:53:56 +01:00
}
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifndef _WIN32
Refuse to use RunAsDaemon when torrc is a relative path. Fixes bug 229. svn:r5767
2006-01-10 21:00:20 +01:00
if (options->RunAsDaemon && torrc_fname && path_is_relative(torrc_fname))
REJECT("Can't use a relative path to torrc when RunAsDaemon is set.");
#endif
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
/* XXXX require that the only port not be DirPort? */
/* XXXX require that at least one port be listened-upon. */
if (n_ports == 0 && !options->RendConfigLines)
r17469@catbus: nickm | 2008-01-05 20:14:07 -0500 Fix bug 579: Count DNSPort and hidden services when checking whether Tor is going to do anything. Change "no configured ports" from fatal to warning. svn:r13036
2008-01-06 02:14:11 +01:00
log(LOG_WARN, LD_CONFIG,
Change Natd into NATD in our options. Breaking this out of the last commit because this might be more controversial.
2010-10-30 06:08:47 +02:00
"SocksPort, TransPort, NATDPort, DNSPort, and ORPort are all "
r17469@catbus: nickm | 2008-01-05 20:14:07 -0500 Fix bug 579: Count DNSPort and hidden services when checking whether Tor is going to do anything. Change "no configured ports" from fatal to warning. svn:r13036
2008-01-06 02:14:11 +01:00
"undefined, and there aren't any hidden services configured. "
"Tor will still run, but probably won't do anything.");
client now survives going offline better fix badness in usage() if neither socksport nor orrport is defined, quit obsolete connection_flush_buf() svn:r780
2003-11-10 09:06:55 +01:00
r7299@Kushana: nickm | 2006-08-10 01:08:58 -0700 Patch from Tup to add support for transparent AP connections: this basically bundles the functionality of trans-proxy-tor into the tor mainline. Now hosts with compliant pf/netfilter implementations can redirect TCP connections straight to Tor without diverting through SOCKS. svn:r7007
2006-08-10 11:01:37 +02:00
#ifndef USE_TRANSPARENT
if (options->TransPort || options->TransListenAddress)
REJECT("TransPort and TransListenAddress are disabled in this build.");
#endif
Move around check for TokenBucketRefillInterval; disallow changes to it
2011-09-08 03:02:49 +02:00
if (options->TokenBucketRefillInterval <= 0
|| options->TokenBucketRefillInterval > 1000) {
REJECT("TokenBucketRefillInterval must be between 1 and 1000 inclusive.");
}
r17188@tombo: nickm | 2008-07-18 14:35:18 -0400 Add new ExcludeExitNodes option. Also add a new routerset type to handle Exclude[Exit]Nodes. It is optimized for O(1) membership tests, so as to make choosing a random router run in O(N_routers) time instead of in O(N_routers*N_Excluded_Routers). svn:r16061
2008-07-18 20:36:32 +02:00
if (options->ExcludeExitNodes || options->ExcludeNodes) {
options->_ExcludeExitNodesUnion = routerset_new();
routerset_union(options->_ExcludeExitNodesUnion,options->ExcludeExitNodes);
routerset_union(options->_ExcludeExitNodesUnion,options->ExcludeNodes);
}
Implement node-based router family code Also, make the NodeFamily option into a list of routersets. This lets us git rid of router_in_nickname_list (or whatever it was called) without porting it to work with nodes, and also lets people specify country codes and IP ranges in NodeFamily
2010-10-01 00:25:25 +02:00
if (options->NodeFamilies) {
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
options->NodeFamilySets = smartlist_new();
Implement node-based router family code Also, make the NodeFamily option into a list of routersets. This lets us git rid of router_in_nickname_list (or whatever it was called) without porting it to work with nodes, and also lets people specify country codes and IP ranges in NodeFamily
2010-10-01 00:25:25 +02:00
for (cl = options->NodeFamilies; cl; cl = cl->next) {
routerset_t *rs = routerset_new();
if (routerset_parse(rs, cl->value, cl->key) == 0) {
smartlist_add(options->NodeFamilySets, rs);
} else {
routerset_free(rs);
}
}
}
Switch to a StrictNodes config option. This is step one of handling ExcludedNodes better. This first step is just to make EntryNodes and ExitNodes do what they did before.
2009-09-18 04:45:54 +02:00
if (options->ExcludeNodes && options->StrictNodes) {
COMPLAIN("You have asked to exclude certain relays from all positions "
"in your circuits. Expect hidden services and other Tor "
"features to be broken in unpredictable ways.");
}
Start implementing the server side of the new directory protocol. Probably very buggy, since I can't actually run an authdir. Features - Generate and publish new network-status format - Code to cache and re-serve network-status objects generated by others. - Publish individual descriptors (by fingerprint, by "all", and by "tell me yours.") [Still needs compression logic] - Publish client and server recommended versions seprately. - Add digest of descriptor to routerinfo_t, so we can track them better, and length, so we can server them more easily. Cleanups - Unify code to sign directory-like things - Make resolve_my_address() able to tell you which name it wound up resolving. - Unify code to store and serve directory-like things so it all uses cached_dir_t. - Unify code to set the value of cached_dir_t objects. svn:r4835
2005-08-25 22:33:17 +02:00
if (options->AuthoritativeDir) {
commit sebastian's patch to not require contactinfo when testingtornetwork is set svn:r19257
2009-04-10 09:02:19 +02:00
if (!options->ContactInfo && !options->TestingTorNetwork)
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
REJECT("Authoritative directory servers must set ContactInfo");
Actually, only v1 auth dir servers need to define recommendedversions. But versioning dirservers do need to cause recommendclientversions and recommendedserverversions to somehow be non-empty. svn:r6181
2006-03-18 00:25:40 +01:00
if (options->V1AuthoritativeDir && !options->RecommendedVersions)
Spell-check Tor.
2009-05-27 23:55:51 +02:00
REJECT("V1 authoritative dir servers must set RecommendedVersions.");
Start implementing the server side of the new directory protocol. Probably very buggy, since I can't actually run an authdir. Features - Generate and publish new network-status format - Code to cache and re-serve network-status objects generated by others. - Publish individual descriptors (by fingerprint, by "all", and by "tell me yours.") [Still needs compression logic] - Publish client and server recommended versions seprately. - Add digest of descriptor to routerinfo_t, so we can track them better, and length, so we can server them more easily. Cleanups - Unify code to sign directory-like things - Make resolve_my_address() able to tell you which name it wound up resolving. - Unify code to store and serve directory-like things so it all uses cached_dir_t. - Unify code to set the value of cached_dir_t objects. svn:r4835
2005-08-25 22:33:17 +02:00
if (!options->RecommendedClientVersions)
options->RecommendedClientVersions =
config_lines_dup(options->RecommendedVersions);
if (!options->RecommendedServerVersions)
options->RecommendedServerVersions =
config_lines_dup(options->RecommendedVersions);
Actually, only v1 auth dir servers need to define recommendedversions. But versioning dirservers do need to cause recommendclientversions and recommendedserverversions to somehow be non-empty. svn:r6181
2006-03-18 00:25:40 +01:00
if (options->VersioningAuthoritativeDir &&
(!options->RecommendedClientVersions ||
!options->RecommendedServerVersions))
Fixes to spelling fixes. Thanks, Roger!
2009-05-28 17:54:56 +02:00
REJECT("Versioning authoritative dir servers must set "
"Recommended*Versions.");
entry nodes are now entry guards. this is our last easy chance for a wholesale change. heave ho. svn:r5782
2006-01-10 23:42:44 +01:00
if (options->UseEntryGuards) {
right now we don't support entryguards on auth dirservers, but that doesn't mean we need to remind them every time they start up. svn:r6179
2006-03-18 00:19:51 +01:00
log_info(LD_CONFIG, "Authoritative directory servers can't set "
"UseEntryGuards. Disabling.");
entry nodes are now entry guards. this is our last easy chance for a wholesale change. heave ho. svn:r5782
2006-01-10 23:42:44 +01:00
options->UseEntryGuards = 0;
prevent auth dir servers from enabling entry nodes. svn:r5702
2006-01-02 09:40:58 +01:00
}
Directory authorities should only automatically download Extra Info documents if they're v1, v2, or v3 authorities. svn:r12898
2007-12-21 07:33:02 +01:00
if (!options->DownloadExtraInfo && authdir_mode_any_main(options)) {
r12981@Kushana: nickm | 2007-05-18 14:12:19 -0400 First cut at code to download extra-info docs. Also note a bad bug in directory.c (look for the string BUG BUG BUG). svn:r10209
2007-05-18 23:19:19 +02:00
log_info(LD_CONFIG, "Authoritative directories always try to download "
"extra-info documents. Setting DownloadExtraInfo.");
options->DownloadExtraInfo = 1;
}
r14018@Kushana: nickm | 2007-08-14 15:39:35 -0400 Resolve XXXX020s in config.c svn:r11108
2007-08-14 22:19:46 +02:00
if (!(options->BridgeAuthoritativeDir || options->HSAuthoritativeDir ||
options->V1AuthoritativeDir || options->V2AuthoritativeDir ||
options->V3AuthoritativeDir))
REJECT("AuthoritativeDir is set, but none of "
Don't trigger an assert if we start a directory authority with a private IP address (like 127.0.0.1). svn:r13371
2008-02-04 17:58:50 +01:00
"(Bridge/HS/V1/V2/V3)AuthoritativeDir is set.");
try loading the bandwidth measurement file on startup too, in case it's broken.
2009-07-06 04:47:36 +02:00
/* If we have a v3bandwidthsfile and it's broken, complain on startup */
if (options->V3BandwidthsFile && !old_options) {
dirserv_read_measured_bandwidths(options->V3BandwidthsFile, NULL);
}
actually, if recommendedversions is required for dirservers, then require it. svn:r1253
2004-03-10 08:44:31 +01:00
}
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
if (options->AuthoritativeDir && !options->DirPort)
REJECT("Running as authoritative directory, but no DirPort set.");
Implement several 008pre1 items: needs more testing svn:r1981
2004-06-21 06:37:27 +02:00
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
if (options->AuthoritativeDir && !options->ORPort)
REJECT("Running as authoritative directory, but no ORPort set.");
checkpoint: start working on bandwidth tracking and letting clients evaluate whether they're suitable servers svn:r2037
2004-07-13 09:42:20 +02:00
fix bug 235 svn:r5768
2006-01-10 21:06:24 +01:00
if (options->AuthoritativeDir && options->ClientOnly)
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
REJECT("Running as authoritative directory, but ClientOnly also set.");
checkpoint: start working on bandwidth tracking and letting clients evaluate whether they're suitable servers svn:r2037
2004-07-13 09:42:20 +02:00
Add support for a new option: FetchDirInfoExtraEarly This new option will allow clients to download the newest fresh consensus much sooner than they normally would do so, even if they previously set FetchDirInfoEarly. This includes a proper ChangeLog entry and an updated man page.
2009-07-12 03:43:33 +02:00
if (options->FetchDirInfoExtraEarly && !options->FetchDirInfoEarly)
REJECT("FetchDirInfoExtraEarly requires that you also set "
"FetchDirInfoEarly");
Don't allow v0 HS auths to act as clients A v0 HS authority stores v0 HS descriptors in the same descriptor cache that its HS client functionality uses. Thus, if the HS authority operator clears its client HS descriptor cache, ALL v0 HS descriptors will be lost. That would be bad.
2011-04-21 00:20:10 +02:00
if (options->HSAuthoritativeDir && proxy_mode(options))
REJECT("Running as authoritative v0 HS directory, but also configured "
"as a client.");
Fix some more obscure compiler warnings svn:r3758
2005-03-14 04:28:46 +01:00
if (options->ConnLimit <= 0) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
"ConnLimit must be greater than 0, but was set to %d",
Fix some more obscure compiler warnings svn:r3758
2005-03-14 04:28:46 +01:00
options->ConnLimit);
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
return -1;
Fix some more obscure compiler warnings svn:r3758
2005-03-14 04:28:46 +01:00
}
Add an option to limit the number of non-open client circuits. This is mainly meant as a way to keep clients from accidentally DOSing themselves by (e.g.) enabling IsolateDestAddr or IsolateDestPort on a port that they use for HTTP.
2011-07-07 20:54:54 +02:00
if (options->MaxClientCircuitsPending <= 0 ||
options->MaxClientCircuitsPending > MAX_MAX_CLIENT_CIRCUITS_PENDING) {
tor_asprintf(msg,
"MaxClientCircuitsPending must be between 1 and %d, but "
"was set to %d", MAX_MAX_CLIENT_CIRCUITS_PENDING,
options->MaxClientCircuitsPending);
return -1;
}
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
if (validate_ports_csv(options->FirewallPorts, "FirewallPorts", msg) < 0)
return -1;
Annotate circuits w/ whether they aim to contain high uptime nodes and/or high capacity nodes. When building circuits, choose appropriate nodes. New config option LongLivedPorts to indicate application streams that will want high uptime circuits. When attaching a stream to a circuit, pay attention to its requirements. This means that every single node in an intro rend circuit, not just the last one, will have a minimum uptime. Boost the min uptime from an hour to 24 hours. svn:r3339
2005-01-12 05:58:23 +01:00
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
if (validate_ports_csv(options->LongLivedPorts, "LongLivedPorts", msg) < 0)
return -1;
Generalize FirewallPorts to FirewallIPs so I can run happily when locked in net 18. svn:r4737
2005-08-07 23:24:00 +02:00
New config options WarnPlaintextPorts and RejectPlaintextPorts so Tor can warn and/or refuse connections to ports commonly used with vulnerable-plaintext protocols. We still need to figure out some good defaults for them. svn:r13198
2008-01-20 06:54:15 +01:00
if (validate_ports_csv(options->RejectPlaintextPorts,
"RejectPlaintextPorts", msg) < 0)
return -1;
if (validate_ports_csv(options->WarnPlaintextPorts,
"WarnPlaintextPorts", msg) < 0)
return -1;
whenever we hupped or did a controller setconf, we were prepending another reachableaddresses *:80,*:443 if fascistfirewall was set, and we were appending another reject *:* regardless. svn:r5560
2005-12-11 10:18:25 +01:00
if (options->FascistFirewall && !options->ReachableAddresses) {
Fix a crash if you enable FascistFirewall but not FirewallPorts. Reported by Frediano Ziglio. svn:r6746
2006-07-08 19:38:46 +02:00
if (options->FirewallPorts && smartlist_len(options->FirewallPorts)) {
Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses svn:r6009
2006-02-13 22:17:20 +01:00
/* We already have firewall ports set, so migrate them to
clean up the Reachable*Addresses changes svn:r6041
2006-02-19 09:31:47 +01:00
* ReachableAddresses, which will set ReachableORAddresses and
* ReachableDirAddresses if they aren't set explicitly. */
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
smartlist_t *instead = smartlist_new();
Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses svn:r6009
2006-02-13 22:17:20 +01:00
config_line_t *new_line = tor_malloc_zero(sizeof(config_line_t));
new_line->key = tor_strdup("ReachableAddresses");
/* If we're configured with the old format, we need to prepend some
* open ports. */
SMARTLIST_FOREACH(options->FirewallPorts, const char *, portno,
Replace (Fascist)Firewall* with a new ReachableAddresses option that understands address policies. svn:r4751
2005-08-08 23:58:48 +02:00
{
int p = atoi(portno);
if (p<0) continue;
Convert instances of tor_snprintf+strdup into tor_asprintf These were found by looking for tor_snprintf() instances that were followed closely by tor_strdup(), though I probably converted some other snprintfs as well.
2012-01-11 20:02:59 +01:00
smartlist_add_asprintf(instead, "*:%d", p);
Replace (Fascist)Firewall* with a new ReachableAddresses option that understands address policies. svn:r4751
2005-08-08 23:58:48 +02:00
});
Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses svn:r6009
2006-02-13 22:17:20 +01:00
new_line->value = smartlist_join_strings(instead,",",0,NULL);
/* These have been deprecated since 0.1.1.5-alpha-cvs */
clean up the Reachable*Addresses changes svn:r6041
2006-02-19 09:31:47 +01:00
log(LOG_NOTICE, LD_CONFIG,
"Converting FascistFirewall and FirewallPorts "
Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses svn:r6009
2006-02-13 22:17:20 +01:00
"config options to new format: \"ReachableAddresses %s\"",
new_line->value);
options->ReachableAddresses = new_line;
SMARTLIST_FOREACH(instead, char *, cp, tor_free(cp));
smartlist_free(instead);
} else {
clean up the Reachable*Addresses changes svn:r6041
2006-02-19 09:31:47 +01:00
/* We do not have FirewallPorts set, so add 80 to
* ReachableDirAddresses, and 443 to ReachableORAddresses. */
Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses svn:r6009
2006-02-13 22:17:20 +01:00
if (!options->ReachableDirAddresses) {
config_line_t *new_line = tor_malloc_zero(sizeof(config_line_t));
new_line->key = tor_strdup("ReachableDirAddresses");
new_line->value = tor_strdup("*:80");
options->ReachableDirAddresses = new_line;
log(LOG_NOTICE, LD_CONFIG, "Converting FascistFirewall config option "
"to new format: \"ReachableDirAddresses *:80\"");
}
if (!options->ReachableORAddresses) {
config_line_t *new_line = tor_malloc_zero(sizeof(config_line_t));
new_line->key = tor_strdup("ReachableORAddresses");
new_line->value = tor_strdup("*:443");
options->ReachableORAddresses = new_line;
log(LOG_NOTICE, LD_CONFIG, "Converting FascistFirewall config option "
"to new format: \"ReachableORAddresses *:443\"");
}
}
Generalize FirewallPorts to FirewallIPs so I can run happily when locked in net 18. svn:r4737
2005-08-07 23:24:00 +02:00
}
clean up the Reachable*Addresses changes svn:r6041
2006-02-19 09:31:47 +01:00
for (i=0; i<3; i++) {
config_line_t **linep =
Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses svn:r6009
2006-02-13 22:17:20 +01:00
(i==0) ? &options->ReachableAddresses :
clean up the Reachable*Addresses changes svn:r6041
2006-02-19 09:31:47 +01:00
(i==1) ? &options->ReachableORAddresses :
&options->ReachableDirAddresses;
Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses svn:r6009
2006-02-13 22:17:20 +01:00
if (!*linep)
continue;
Replace (Fascist)Firewall* with a new ReachableAddresses option that understands address policies. svn:r4751
2005-08-08 23:58:48 +02:00
/* We need to end with a reject *:*, not an implicit accept *:* */
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
for (;;) {
whenever we hupped or did a controller setconf, we were prepending another reachableaddresses *:80,*:443 if fascistfirewall was set, and we were appending another reject *:* regardless. svn:r5560
2005-12-11 10:18:25 +01:00
if (!strcmp((*linep)->value, "reject *:*")) /* already there */
break;
Replace (Fascist)Firewall* with a new ReachableAddresses option that understands address policies. svn:r4751
2005-08-08 23:58:48 +02:00
linep = &((*linep)->next);
whenever we hupped or did a controller setconf, we were prepending another reachableaddresses *:80,*:443 if fascistfirewall was set, and we were appending another reject *:* regardless. svn:r5560
2005-12-11 10:18:25 +01:00
if (!*linep) {
*linep = tor_malloc_zero(sizeof(config_line_t));
clean up the Reachable*Addresses changes svn:r6041
2006-02-19 09:31:47 +01:00
(*linep)->key = tor_strdup(
(i==0) ? "ReachableAddresses" :
(i==1) ? "ReachableORAddresses" :
"ReachableDirAddresses");
whenever we hupped or did a controller setconf, we were prepending another reachableaddresses *:80,*:443 if fascistfirewall was set, and we were appending another reject *:* regardless. svn:r5560
2005-12-11 10:18:25 +01:00
(*linep)->value = tor_strdup("reject *:*");
break;
}
Replace (Fascist)Firewall* with a new ReachableAddresses option that understands address policies. svn:r4751
2005-08-08 23:58:48 +02:00
}
}
Annotate circuits w/ whether they aim to contain high uptime nodes and/or high capacity nodes. When building circuits, choose appropriate nodes. New config option LongLivedPorts to indicate application streams that will want high uptime circuits. When attaching a stream to a circuit, pay attention to its requirements. This means that every single node in an intro rend circuit, not just the last one, will have a minimum uptime. Boost the min uptime from an hour to 24 hours. svn:r3339
2005-01-12 05:58:23 +01:00
Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses svn:r6009
2006-02-13 22:17:20 +01:00
if ((options->ReachableAddresses ||
options->ReachableORAddresses ||
options->ReachableDirAddresses) &&
server_mode(options))
fix bug 235 svn:r5768
2006-01-10 21:06:24 +01:00
REJECT("Servers must be able to freely connect to the rest "
Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses svn:r6009
2006-02-13 22:17:20 +01:00
"of the Internet, so they must not set Reachable*Addresses "
fix bug 235 svn:r5768
2006-01-10 21:06:24 +01:00
"or FascistFirewall.");
Refuse to start if both ORPort and UseBridges are set. Bugfix on 0.2.0.x. svn:r12182
2007-10-25 06:40:27 +02:00
if (options->UseBridges &&
server_mode(options))
REJECT("Servers must be able to freely connect to the rest "
"of the Internet, so they must not set UseBridges.");
make formal a constraint that's been true a while now
2011-03-11 11:39:26 +01:00
/* If both of these are set, we'll end up with funny behavior where we
* demand enough entrynodes be up and running else we won't build
* circuits, yet we never actually use them. */
if (options->UseBridges && options->EntryNodes)
REJECT("You cannot set both UseBridges and EntryNodes.");
Fix bug2571: warn on EntryNodes set and UseEntryGuards disabled
2011-12-22 16:31:52 +01:00
if (options->EntryNodes && !options->UseEntryGuards)
log_warn(LD_CONFIG, "EntryNodes is set, but UseEntryGuards is disabled. "
"EntryNodes will be ignored.");
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
options->_AllowInvalid = 0;
if (options->AllowInvalidNodes) {
SMARTLIST_FOREACH(options->AllowInvalidNodes, const char *, cp, {
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
if (!strcasecmp(cp, "entry"))
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
options->_AllowInvalid |= ALLOW_INVALID_ENTRY;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
else if (!strcasecmp(cp, "exit"))
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
options->_AllowInvalid |= ALLOW_INVALID_EXIT;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
else if (!strcasecmp(cp, "middle"))
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
options->_AllowInvalid |= ALLOW_INVALID_MIDDLE;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
else if (!strcasecmp(cp, "introduction"))
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
options->_AllowInvalid |= ALLOW_INVALID_INTRODUCTION;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
else if (!strcasecmp(cp, "rendezvous"))
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
options->_AllowInvalid |= ALLOW_INVALID_RENDEZVOUS;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
else {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
Finish the transition from the word 'verified' to the words 'named' and 'valid'. svn:r6188
2006-03-19 02:21:59 +01:00
"Unrecognized value '%s' in AllowInvalidNodes", cp);
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
return -1;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
}
});
}
Cache the parsed value of SafeLogging as an enum.
2009-12-12 07:12:47 +01:00
if (!options->SafeLogging ||
!strcasecmp(options->SafeLogging, "0")) {
options->_SafeLogging = SAFELOG_SCRUB_NONE;
} else if (!strcasecmp(options->SafeLogging, "relay")) {
options->_SafeLogging = SAFELOG_SCRUB_RELAY;
} else if (!strcasecmp(options->SafeLogging, "1")) {
options->_SafeLogging = SAFELOG_SCRUB_ALL;
} else {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
Allow SafeLogging to exclude client related information
2009-09-28 15:08:32 +02:00
"Unrecognized value '%s' in SafeLogging",
Cache the parsed value of SafeLogging as an enum.
2009-12-12 07:12:47 +01:00
escaped(options->SafeLogging));
Allow SafeLogging to exclude client related information
2009-09-28 15:08:32 +02:00
return -1;
}
If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". svn:r12923
2007-12-22 10:04:46 +01:00
if (compute_publishserverdescriptor(options) < 0) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg, "Unrecognized value in PublishServerDescriptor");
Change the PublishServerDescriptor config option from a boolean into a string: "v1", "v2", bridge", "". Continue to support "0" and "1". svn:r10136
2007-05-08 11:09:26 +02:00
return -1;
}
Prevent bridges from publishing router descriptors. Bridges are not supposed to publish router descriptors to the directory authorities. It defeats the point of bridges when they are included in the public relay directory. This patch puts out a warning and exits when the node is configured as a bridge and to publish v1, v2, or v3 descriptors at the same time. Also fixes part of bug 932.
2009-05-13 18:14:29 +02:00
if ((options->BridgeRelay
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
|| options->_PublishServerDescriptor & BRIDGE_DIRINFO)
Use | with flags, not +.
2009-05-13 22:43:51 +02:00
&& (options->_PublishServerDescriptor
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
& (V1_DIRINFO|V2_DIRINFO|V3_DIRINFO))) {
Prevent bridges from publishing router descriptors. Bridges are not supposed to publish router descriptors to the directory authorities. It defeats the point of bridges when they are included in the public relay directory. This patch puts out a warning and exits when the node is configured as a bridge and to publish v1, v2, or v3 descriptors at the same time. Also fixes part of bug 932.
2009-05-13 18:14:29 +02:00
REJECT("Bridges are not supposed to publish router descriptors to the "
"directory authorities. Please correct your "
"PublishServerDescriptor line.");
}
Disable DirPort when BridgeRelay is set
2010-11-22 19:12:48 +01:00
if (options->BridgeRelay && options->DirPort) {
log_warn(LD_CONFIG, "Can't set a DirPort on a bridge relay; disabling "
"DirPort");
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
config_free_lines(options->DirPort);
options->DirPort = NULL;
Disable DirPort when BridgeRelay is set
2010-11-22 19:12:48 +01:00
}
karsten's second refactoring patch svn:r12607
2007-11-29 16:25:04 +01:00
if (options->MinUptimeHidServDirectoryV2 < 0) {
log_warn(LD_CONFIG, "MinUptimeHidServDirectoryV2 option must be at "
r16262@catbus: nickm | 2007-10-29 13:21:35 -0400 Patch from Karsten: Code to act as (and use) v2 hidden service directories. svn:r12272
2007-10-29 20:10:42 +01:00
"least 0 seconds. Changing to 0.");
karsten's second refactoring patch svn:r12607
2007-11-29 16:25:04 +01:00
options->MinUptimeHidServDirectoryV2 = 0;
r16262@catbus: nickm | 2007-10-29 13:21:35 -0400 Patch from Karsten: Code to act as (and use) v2 hidden service directories. svn:r12272
2007-10-29 20:10:42 +01:00
}
If anybody set DirFetchPostPeriod, give them StatuFetchPeriod instead. Impose minima and maxima for all *Period options; impose even tighter maxima for fetching if we are a caching dirserver. Clip rather than rejecting. arma: are these good? svn:r3024
2004-11-30 03:23:51 +01:00
if (options->RendPostPeriod < MIN_REND_POST_PERIOD) {
add a minimum for CircuitStreamTimeout, plus a man page plus some other unrelated touchups that have been sitting in my sandbox
2009-11-22 13:15:30 +01:00
log_warn(LD_CONFIG, "RendPostPeriod option is too short; "
"raising to %d seconds.", MIN_REND_POST_PERIOD);
If anybody set DirFetchPostPeriod, give them StatuFetchPeriod instead. Impose minima and maxima for all *Period options; impose even tighter maxima for fetching if we are a caching dirserver. Clip rather than rejecting. arma: are these good? svn:r3024
2004-11-30 03:23:51 +01:00
options->RendPostPeriod = MIN_REND_POST_PERIOD;
Break DirFetchPostPeriod into: - DirFetchPeriod for fetching full directory, - StatusFetchPeriod for fetching running-routers, - DirPostPeriod for posting server descriptor, - RendPostPeriod for posting hidden service descriptors. Also make sure the hidden service descriptors are at a random offset from each other, to hinder linkability. svn:r2889
2004-11-15 10:05:54 +01:00
}
If anybody set DirFetchPostPeriod, give them StatuFetchPeriod instead. Impose minima and maxima for all *Period options; impose even tighter maxima for fetching if we are a caching dirserver. Clip rather than rejecting. arma: are these good? svn:r3024
2004-11-30 03:23:51 +01:00
if (options->RendPostPeriod > MAX_DIR_PERIOD) {
add a minimum for CircuitStreamTimeout, plus a man page plus some other unrelated touchups that have been sitting in my sandbox
2009-11-22 13:15:30 +01:00
log_warn(LD_CONFIG, "RendPostPeriod is too large; clipping to %ds.",
MAX_DIR_PERIOD);
If anybody set DirFetchPostPeriod, give them StatuFetchPeriod instead. Impose minima and maxima for all *Period options; impose even tighter maxima for fetching if we are a caching dirserver. Clip rather than rejecting. arma: are these good? svn:r3024
2004-11-30 03:23:51 +01:00
options->RendPostPeriod = MAX_DIR_PERIOD;
Merge flagday into main branch. svn:r1683
2004-04-25 00:17:50 +02:00
}
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
Turn off LearnCircuitBuildTimeout when tor2web mode is on
2011-06-14 10:40:02 +02:00
if (options->Tor2webMode && options->LearnCircuitBuildTimeout) {
/* LearnCircuitBuildTimeout and Tor2webMode are incompatible in
* two ways:
*
* - LearnCircuitBuildTimeout results in a low CBT, which
* Tor2webMode's use of one-hop rendezvous circuits lowers
* much further, producing *far* too many timeouts.
*
* - The adaptive CBT code does not update its timeout estimate
* using build times for single-hop circuits.
*
* If we fix both of these issues someday, we should test
* Tor2webMode with LearnCircuitBuildTimeout on again. */
log_notice(LD_CONFIG,"Tor2webMode is enabled; turning "
"LearnCircuitBuildTimeout off.");
options->LearnCircuitBuildTimeout = 0;
}
Clip the MaxCircuitDirtiness config option to a minimum of 10 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.0.1-rc. Patch by Sebastian. Clip the CircuitBuildTimeout to a minimum of 30 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.1.17-rc. Patch by Sebastian. svn:r17657
2008-12-17 23:32:17 +01:00
if (options->MaxCircuitDirtiness < MIN_MAX_CIRCUIT_DIRTINESS) {
add a minimum for CircuitStreamTimeout, plus a man page plus some other unrelated touchups that have been sitting in my sandbox
2009-11-22 13:15:30 +01:00
log_warn(LD_CONFIG, "MaxCircuitDirtiness option is too short; "
"raising to %d seconds.", MIN_MAX_CIRCUIT_DIRTINESS);
Clip the MaxCircuitDirtiness config option to a minimum of 10 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.0.1-rc. Patch by Sebastian. Clip the CircuitBuildTimeout to a minimum of 30 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.1.17-rc. Patch by Sebastian. svn:r17657
2008-12-17 23:32:17 +01:00
options->MaxCircuitDirtiness = MIN_MAX_CIRCUIT_DIRTINESS;
}
add a minimum for CircuitStreamTimeout, plus a man page plus some other unrelated touchups that have been sitting in my sandbox
2009-11-22 13:15:30 +01:00
if (options->CircuitStreamTimeout &&
options->CircuitStreamTimeout < MIN_CIRCUIT_STREAM_TIMEOUT) {
log_warn(LD_CONFIG, "CircuitStreamTimeout option is too short; "
"raising to %d seconds.", MIN_CIRCUIT_STREAM_TIMEOUT);
options->CircuitStreamTimeout = MIN_CIRCUIT_STREAM_TIMEOUT;
}
Implement more heartbeat message stuff. (This squashes multiple commits: * Adds uptime monitoring support. * Adds circuit counting code. * Trivially tweaks the documentation. * Trivial run_scheduled_events() code tweaking. * Adds a status.h to export functions. * Added bandwidth monitoring code. * Added consensus presense detection code. * Restricts the precision of the bandwidth output. * Various fixes. * Fixed style and spacing problems. * Tidied up src/or/Makefile.am * Couple of minor fixes on status.c functions. * 'Implemented' client heartbeat support )
2010-12-01 02:32:42 +01:00
Initial heartbeat subsystem commit. Sets: * Documentation * Logging domain * Configuration option * Scheduled event * Makefile It also creates status.c and the log_heartbeat() function. All code was written by Sebastian Hahn. Commit message was written by me (George Kadianakis).
2010-12-01 02:24:03 +01:00
if (options->HeartbeatPeriod &&
options->HeartbeatPeriod < MIN_HEARTBEAT_PERIOD) {
log_warn(LD_CONFIG, "HeartbeatPeriod option is too short; "
"raising to %d seconds.", MIN_HEARTBEAT_PERIOD);
options->HeartbeatPeriod = MIN_HEARTBEAT_PERIOD;
}
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
if (options->KeepalivePeriod < 1)
REJECT("KeepalivePeriod option must be positive.");
remove popt dependency, consolidate config stuff reformat parts of onion.c svn:r136
2002-11-23 07:49:01 +01:00
r13090@catbus: nickm | 2007-05-30 03:17:57 -0400 If the user makes a torrc that exceeds the bandwidth cap by one byte, let them have it. svn:r10411
2007-05-30 09:18:00 +02:00
if (ensure_bandwidth_cap(&options->BandwidthRate,
if you're using relaybandwidthrate and relaybandwidthburst, make sure that's reflected in your router descriptor. svn:r10114
2007-05-04 11:20:13 +02:00
"BandwidthRate", msg) < 0)
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
return -1;
r13090@catbus: nickm | 2007-05-30 03:17:57 -0400 If the user makes a torrc that exceeds the bandwidth cap by one byte, let them have it. svn:r10411
2007-05-30 09:18:00 +02:00
if (ensure_bandwidth_cap(&options->BandwidthBurst,
if you're using relaybandwidthrate and relaybandwidthburst, make sure that's reflected in your router descriptor. svn:r10114
2007-05-04 11:20:13 +02:00
"BandwidthBurst", msg) < 0)
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
return -1;
r13090@catbus: nickm | 2007-05-30 03:17:57 -0400 If the user makes a torrc that exceeds the bandwidth cap by one byte, let them have it. svn:r10411
2007-05-30 09:18:00 +02:00
if (ensure_bandwidth_cap(&options->MaxAdvertisedBandwidth,
if you're using relaybandwidthrate and relaybandwidthburst, make sure that's reflected in your router descriptor. svn:r10114
2007-05-04 11:20:13 +02:00
"MaxAdvertisedBandwidth", msg) < 0)
return -1;
r13090@catbus: nickm | 2007-05-30 03:17:57 -0400 If the user makes a torrc that exceeds the bandwidth cap by one byte, let them have it. svn:r10411
2007-05-30 09:18:00 +02:00
if (ensure_bandwidth_cap(&options->RelayBandwidthRate,
if you're using relaybandwidthrate and relaybandwidthburst, make sure that's reflected in your router descriptor. svn:r10114
2007-05-04 11:20:13 +02:00
"RelayBandwidthRate", msg) < 0)
return -1;
r13090@catbus: nickm | 2007-05-30 03:17:57 -0400 If the user makes a torrc that exceeds the bandwidth cap by one byte, let them have it. svn:r10411
2007-05-30 09:18:00 +02:00
if (ensure_bandwidth_cap(&options->RelayBandwidthBurst,
if you're using relaybandwidthrate and relaybandwidthburst, make sure that's reflected in your router descriptor. svn:r10114
2007-05-04 11:20:13 +02:00
"RelayBandwidthBurst", msg) < 0)
return -1;
Restrict PerConnBWRate|Burst to INT32_MAX, update manpage All other bandwidthrate settings are restricted to INT32_MAX, but this check was forgotten for PerConnBWRate and PerConnBWBurst. Also update the manpage to reflect the fact that specifying a bandwidth in terabytes does not make sense, because that value will be too large.
2010-02-25 12:25:57 +01:00
if (ensure_bandwidth_cap(&options->PerConnBWRate,
"PerConnBWRate", msg) < 0)
return -1;
if (ensure_bandwidth_cap(&options->PerConnBWBurst,
"PerConnBWBurst", msg) < 0)
return -1;
parameterize bw cutoffs to guarantee Fast and Guard flags Now it will be easier for researchers to simulate Tor networks with different values. Resolves ticket 4484.
2011-11-17 02:55:33 +01:00
if (ensure_bandwidth_cap(&options->AuthDirFastGuarantee,
"AuthDirFastGuarantee", msg) < 0)
return -1;
if (ensure_bandwidth_cap(&options->AuthDirGuardBWGuarantee,
"AuthDirGuardBWGuarantee", msg) < 0)
return -1;
if you're using relaybandwidthrate and relaybandwidthburst, make sure that's reflected in your router descriptor. svn:r10114
2007-05-04 11:20:13 +02:00
move the clause above the "if bw is too low" check
2011-02-08 05:22:45 +01:00
if (options->RelayBandwidthRate && !options->RelayBandwidthBurst)
options->RelayBandwidthBurst = options->RelayBandwidthRate;
if (options->RelayBandwidthBurst && !options->RelayBandwidthRate)
options->RelayBandwidthRate = options->RelayBandwidthBurst;
r11944@catbus: nickm | 2007-02-25 14:43:18 -0500 Add a lower-bound on MaxAdvertisedBandwidth. svn:r9652
2007-02-25 20:43:23 +01:00
if (server_mode(options)) {
Raise the minimum bandwidth to be a relay from 20000 bytes to 20480 bytes (aka 20KB/s), to match our documentation. Also update directory authorities so they always assign the Fast flag to relays with 20KB/s of capacity. Now people running relays won't suddenly find themselves not seeing any use, if the network gets faster on average. svn:r19305
2009-04-12 09:56:58 +02:00
if (options->BandwidthRate < ROUTER_REQUIRED_MIN_BANDWIDTH) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
r11944@catbus: nickm | 2007-02-25 14:43:18 -0500 Add a lower-bound on MaxAdvertisedBandwidth. svn:r9652
2007-02-25 20:43:23 +01:00
"BandwidthRate is set to %d bytes/second. "
"For servers, it must be at least %d.",
(int)options->BandwidthRate,
Raise the minimum bandwidth to be a relay from 20000 bytes to 20480 bytes (aka 20KB/s), to match our documentation. Also update directory authorities so they always assign the Fast flag to relays with 20KB/s of capacity. Now people running relays won't suddenly find themselves not seeing any use, if the network gets faster on average. svn:r19305
2009-04-12 09:56:58 +02:00
ROUTER_REQUIRED_MIN_BANDWIDTH);
r11944@catbus: nickm | 2007-02-25 14:43:18 -0500 Add a lower-bound on MaxAdvertisedBandwidth. svn:r9652
2007-02-25 20:43:23 +01:00
return -1;
} else if (options->MaxAdvertisedBandwidth <
Raise the minimum bandwidth to be a relay from 20000 bytes to 20480 bytes (aka 20KB/s), to match our documentation. Also update directory authorities so they always assign the Fast flag to relays with 20KB/s of capacity. Now people running relays won't suddenly find themselves not seeing any use, if the network gets faster on average. svn:r19305
2009-04-12 09:56:58 +02:00
ROUTER_REQUIRED_MIN_BANDWIDTH/2) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
r11944@catbus: nickm | 2007-02-25 14:43:18 -0500 Add a lower-bound on MaxAdvertisedBandwidth. svn:r9652
2007-02-25 20:43:23 +01:00
"MaxAdvertisedBandwidth is set to %d bytes/second. "
"For servers, it must be at least %d.",
(int)options->MaxAdvertisedBandwidth,
Raise the minimum bandwidth to be a relay from 20000 bytes to 20480 bytes (aka 20KB/s), to match our documentation. Also update directory authorities so they always assign the Fast flag to relays with 20KB/s of capacity. Now people running relays won't suddenly find themselves not seeing any use, if the network gets faster on average. svn:r19305
2009-04-12 09:56:58 +02:00
ROUTER_REQUIRED_MIN_BANDWIDTH/2);
r11944@catbus: nickm | 2007-02-25 14:43:18 -0500 Add a lower-bound on MaxAdvertisedBandwidth. svn:r9652
2007-02-25 20:43:23 +01:00
return -1;
}
Add a separate set of token buckets for relayed traffic. Right now that's just defined as answers to directory requests. svn:r9881
2007-03-20 03:55:31 +01:00
if (options->RelayBandwidthRate &&
options->RelayBandwidthRate < ROUTER_REQUIRED_MIN_BANDWIDTH) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
Add a separate set of token buckets for relayed traffic. Right now that's just defined as answers to directory requests. svn:r9881
2007-03-20 03:55:31 +01:00
"RelayBandwidthRate is set to %d bytes/second. "
"For servers, it must be at least %d.",
(int)options->RelayBandwidthRate,
ROUTER_REQUIRED_MIN_BANDWIDTH);
return -1;
}
allow bandwidthburst to be smaller, but whine if it's small. svn:r3364
2005-01-18 20:34:22 +01:00
}
r11944@catbus: nickm | 2007-02-25 14:43:18 -0500 Add a lower-bound on MaxAdvertisedBandwidth. svn:r9652
2007-02-25 20:43:23 +01:00
cleanups, bump to 0.2.0.12-alpha svn:r12516
2007-11-17 03:22:56 +01:00
if (options->RelayBandwidthRate > options->RelayBandwidthBurst)
REJECT("RelayBandwidthBurst must be at least equal "
"to RelayBandwidthRate.");
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
if (options->BandwidthRate > options->BandwidthBurst)
REJECT("BandwidthBurst must be at least equal to BandwidthRate.");
various fixes svn:r2712
2004-11-09 02:24:10 +01:00
forward-port recent changes svn:r13935
2008-03-10 08:50:09 +01:00
/* if they set relaybandwidth* really high but left bandwidth*
* at the default, raise the defaults. */
if (options->RelayBandwidthRate > options->BandwidthRate)
options->BandwidthRate = options->RelayBandwidthRate;
if (options->RelayBandwidthBurst > options->BandwidthBurst)
options->BandwidthBurst = options->RelayBandwidthBurst;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
if (accounting_parse_options(options, 1)<0)
REJECT("Failed to parse accounting options. See logs for details.");
First attempt at hibernation code. It needs more work around the XXXXs, but first I want arma to review the basic approach while I meditate on how to fix them. svn:r2649
2004-11-03 00:47:32 +01:00
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
if (options->HTTPProxy) { /* parse it now */
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(options->HTTPProxy,
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
&options->HTTPProxyAddr, &options->HTTPProxyPort) < 0)
REJECT("HTTPProxy failed to parse or resolve. Please fix.");
if (options->HTTPProxyPort == 0) { /* give it a default */
options->HTTPProxyPort = 80;
parse HttpProxy address in config fix a potential confusion in fetch_from_buf_http() make all our int config options non-negative better bounds checking on options that are ports svn:r2456
2004-10-12 22:22:09 +02:00
}
}
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
if (options->HTTPProxyAuthenticator) {
Fix bug2752 : 48-char HTTPProxyAuthenticator limitation Bumped the char maximum to 512 for HTTPProxyAuthenticator & HTTPSProxyAuthenticator. Now stripping all '\n' after base64 encoding in alloc_http_authenticator.
2011-05-16 22:09:35 +02:00
if (strlen(options->HTTPProxyAuthenticator) >= 512)
REJECT("HTTPProxyAuthenticator is too long (>= 512 chars).");
add HttpProxyAuthenticator config option too svn:r4272
2005-05-20 10:51:45 +02:00
}
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
if (options->HTTPSProxy) { /* parse it now */
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(options->HTTPSProxy,
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
&options->HTTPSProxyAddr, &options->HTTPSProxyPort) <0)
REJECT("HTTPSProxy failed to parse or resolve. Please fix.");
if (options->HTTPSProxyPort == 0) { /* give it a default */
options->HTTPSProxyPort = 443;
add support for CONNECTing through https proxies. not sure if it works. i don't have an https proxy. svn:r3682
2005-02-24 11:56:55 +01:00
}
}
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
if (options->HTTPSProxyAuthenticator) {
Fix bug2752 : 48-char HTTPProxyAuthenticator limitation Bumped the char maximum to 512 for HTTPProxyAuthenticator & HTTPSProxyAuthenticator. Now stripping all '\n' after base64 encoding in alloc_http_authenticator.
2011-05-16 22:09:35 +02:00
if (strlen(options->HTTPSProxyAuthenticator) >= 512)
REJECT("HTTPSProxyAuthenticator is too long (>= 512 chars).");
Add basic HttpsProxyAuthenticator support, based on patch from Adam Langley. svn:r4115
2005-04-26 20:33:33 +02:00
}
Enable Tor to connect through SOCKS 4/5 proxies Added a sanity check in config.c and a check in directory.c directory_initiate_command_rend() to catch any direct connection attempts when a socks proxy is configured.
2009-06-19 01:59:18 +02:00
if (options->Socks4Proxy) { /* parse it now */
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(options->Socks4Proxy,
Enable Tor to connect through SOCKS 4/5 proxies Added a sanity check in config.c and a check in directory.c directory_initiate_command_rend() to catch any direct connection attempts when a socks proxy is configured.
2009-06-19 01:59:18 +02:00
&options->Socks4ProxyAddr,
&options->Socks4ProxyPort) <0)
REJECT("Socks4Proxy failed to parse or resolve. Please fix.");
if (options->Socks4ProxyPort == 0) { /* give it a default */
options->Socks4ProxyPort = 1080;
}
}
if (options->Socks5Proxy) { /* parse it now */
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(options->Socks5Proxy,
Change proxy-address type to tor_addr_t to allow ipv6 proxies.
2009-06-19 18:48:00 +02:00
&options->Socks5ProxyAddr,
&options->Socks5ProxyPort) <0)
Enable Tor to connect through SOCKS 4/5 proxies Added a sanity check in config.c and a check in directory.c directory_initiate_command_rend() to catch any direct connection attempts when a socks proxy is configured.
2009-06-19 01:59:18 +02:00
REJECT("Socks5Proxy failed to parse or resolve. Please fix.");
if (options->Socks5ProxyPort == 0) { /* give it a default */
options->Socks5ProxyPort = 1080;
}
}
Various small tweaks around config.c and or.h
2011-06-21 18:49:04 +02:00
/* Check if more than one proxy type has been enabled. */
if (!!options->Socks4Proxy + !!options->Socks5Proxy +
!!options->HTTPSProxy + !!options->ClientTransportPlugin > 1)
Style and grammar tweaks on 2841 branch
2011-07-03 05:23:07 +02:00
REJECT("You have configured more than one proxy type. "
We now warn the user if a proxy server is not up when we try to connect with it.
2011-06-14 02:51:59 +02:00
"(Socks4Proxy|Socks5Proxy|HTTPSProxy|ClientTransportPlugin)");
A changelog entry and a bit more documentation for socks-client
2009-08-26 17:34:45 +02:00
Enable Tor to connect through SOCKS 4/5 proxies Added a sanity check in config.c and a check in directory.c directory_initiate_command_rend() to catch any direct connection attempts when a socks proxy is configured.
2009-06-19 01:59:18 +02:00
if (options->Socks5ProxyUsername) {
size_t len;
len = strlen(options->Socks5ProxyUsername);
if (len < 1 || len > 255)
REJECT("Socks5ProxyUsername must be between 1 and 255 characters.");
if (!options->Socks5ProxyPassword)
REJECT("Socks5ProxyPassword must be included with Socks5ProxyUsername.");
len = strlen(options->Socks5ProxyPassword);
if (len < 1 || len > 255)
REJECT("Socks5ProxyPassword must be between 1 and 255 characters.");
} else if (options->Socks5ProxyPassword)
REJECT("Socks5ProxyPassword must be included with Socks5ProxyUsername.");
Resolve a bunch of FIXME items; mark a lot more for attention; ask for clarification on some. Turn all XXXX008 ("showstopper for 0.0.8 release") items into XXXX009 or XXXX, since plainly they were not showstoppers for 0.0.8. Add/clean some docs. svn:r2808
2004-11-12 17:39:03 +01:00
if (options->HashedControlPassword) {
Allow multiple HashedControlPassword config lines, to support multiple controller passwords. svn:r12732
2007-12-09 05:59:27 +01:00
smartlist_t *sl = decode_hashed_passwords(options->HashedControlPassword);
if (!sl) {
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
REJECT("Bad HashedControlPassword: wrong length or bad encoding");
Allow multiple HashedControlPassword config lines, to support multiple controller passwords. svn:r12732
2007-12-09 05:59:27 +01:00
} else {
SMARTLIST_FOREACH(sl, char*, cp, tor_free(cp));
smartlist_free(sl);
}
Document remaining configuration options; add a ccheck for redundant authentication mechanisms in options_validate; add TODO items for post-009 hibernation features svn:r2833
2004-11-13 18:17:04 +01:00
}
r14236@tombo: nickm | 2008-02-17 13:44:55 -0500 Partial fix for bug 586: Add an ephemeral __HashedControlSessionPassword. svn:r13543
2008-02-17 19:45:07 +01:00
if (options->HashedControlSessionPassword) {
smartlist_t *sl = decode_hashed_passwords(
options->HashedControlSessionPassword);
if (!sl) {
REJECT("Bad HashedControlSessionPassword: wrong length or bad encoding");
} else {
SMARTLIST_FOREACH(sl, char*, cp, tor_free(cp));
smartlist_free(sl);
}
}
Implement __OwningControllerProcess option Implements part of feature 3049.
2011-05-15 17:23:04 +02:00
if (options->OwningControllerProcess) {
const char *validate_pspec_msg = NULL;
if (tor_validate_process_specifier(options->OwningControllerProcess,
&validate_pspec_msg)) {
tor_asprintf(msg, "Bad OwningControllerProcess: %s",
validate_pspec_msg);
return -1;
}
}
r13919@Kushana: nickm | 2007-08-02 10:58:31 -0700 Warn about unsafe ControlPort configurations. svn:r11038
2007-08-02 23:03:40 +02:00
if (options->ControlPort && !options->HashedControlPassword &&
r14236@tombo: nickm | 2008-02-17 13:44:55 -0500 Partial fix for bug 586: Add an ephemeral __HashedControlSessionPassword. svn:r13543
2008-02-17 19:45:07 +01:00
!options->HashedControlSessionPassword &&
r13919@Kushana: nickm | 2007-08-02 10:58:31 -0700 Warn about unsafe ControlPort configurations. svn:r11038
2007-08-02 23:03:40 +02:00
!options->CookieAuthentication) {
log_warn(LD_CONFIG, "ControlPort is open, but no authentication method "
"has been configured. This means that any program on your "
"computer can reconfigure your Tor. That's bad! You should "
"upgrade your Tor controller as soon as possible.");
}
Warn for bad combination of cookie options Setting CookieAuthFileGroupReadable but without setting CookieAuthFile makes no sense, because unix directory permissions for the data directory prevent the group from accessing the file anyways.
2010-08-20 02:42:17 +02:00
if (options->CookieAuthFileGroupReadable && !options->CookieAuthFile) {
Clean up a warning a bit
2011-04-29 17:14:53 +02:00
log_warn(LD_CONFIG, "CookieAuthFileGroupReadable is set, but will have "
"no effect: you must specify an explicit CookieAuthFile to "
"have it group-readable.");
Warn for bad combination of cookie options Setting CookieAuthFileGroupReadable but without setting CookieAuthFile makes no sense, because unix directory permissions for the data directory prevent the group from accessing the file anyways.
2010-08-20 02:42:17 +02:00
}
entry nodes are now entry guards. this is our last easy chance for a wholesale change. heave ho. svn:r5782
2006-01-10 23:42:44 +01:00
if (options->UseEntryGuards && ! options->NumEntryGuards)
REJECT("Cannot enable UseEntryGuards with NumEntryGuards set to 0");
Add getinfo logic for accounting; add options for helper nodes svn:r4632
2005-07-22 16:55:09 +02:00
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
if (check_nickname_list(options->MyFamily, "MyFamily", msg))
return -1;
Implement user-declared node families. svn:r2534
2004-10-15 22:52:09 +02:00
for (cl = options->NodeFamilies; cl; cl = cl->next) {
Implement node-based router family code Also, make the NodeFamily option into a list of routersets. This lets us git rid of router_in_nickname_list (or whatever it was called) without porting it to work with nodes, and also lets people specify country codes and IP ranges in NodeFamily
2010-10-01 00:25:25 +02:00
routerset_t *rs = routerset_new();
if (routerset_parse(rs, cl->value, cl->key)) {
routerset_free(rs);
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
return -1;
Implement node-based router family code Also, make the NodeFamily option into a list of routersets. This lets us git rid of router_in_nickname_list (or whatever it was called) without porting it to work with nodes, and also lets people specify country codes and IP ranges in NodeFamily
2010-10-01 00:25:25 +02:00
}
routerset_free(rs);
Implement user-declared node families. svn:r2534
2004-10-15 22:52:09 +02:00
}
Refactor and consolidate addr/exit policies into a new policies.c. Fix some minor bugs and memory leaks along the way. svn:r6246
2006-03-27 04:25:34 +02:00
if (validate_addr_policies(options, msg) < 0)
return -1;
Rename exit_policy to addr_policy, since it gets used for SOCKS and directory connections too. Make all policies get validated in options_validate, and make SOCKS/directory policies get set in options_act. svn:r2819
2004-11-12 20:39:13 +01:00
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
if (validate_dir_authorities(options, old_options) < 0)
REJECT("Directory authority line did not parse. See logs for details.");
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
Revert "Add an "auto" option to UseBridges" This reverts commit 507c1257a4d9c629fefc2adbad8db73607749734.
2011-06-17 22:45:53 +02:00
if (options->UseBridges && !options->Bridges)
REJECT("If you set UseBridges, you must specify at least one bridge.");
More work towards making bridge users able to connect via bridges: - demand options->Bridges and options->TunnelDirConns if options->UseBridges is set. - after directory fetches, accept descriptors that aren't referenced by our networkstatuses, *if* they're for a configured bridge. - delay directory fetching until we have at least one bridge descriptor. - learn how to build a one-hop circuit when we have neither routerinfo nor routerstatus for our destination. - teach directory connections how to pick a bridge as the destination directory when doing non-anonymous fetches. - tolerate directory commands for which the dir_port is 0. - remember descriptors when the requested_resource was "authority", rather than just ignoring them. - put bridges on our entry_guards list once we have a descriptor for them. When UseBridges is set, only pick entry guards that are bridges. Else vice versa. svn:r10571
2007-06-12 11:17:23 +02:00
if (options->UseBridges && !options->TunnelDirConns)
Revert "Add an "auto" option to UseBridges" This reverts commit 507c1257a4d9c629fefc2adbad8db73607749734.
2011-06-17 22:45:53 +02:00
REJECT("If you set UseBridges, you must set TunnelDirConns.");
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
for (cl = options->Bridges; cl; cl = cl->next) {
if (parse_bridge_line(cl->value, 1)<0)
REJECT("Bridge line did not parse. See logs for details.");
}
Style and grammar tweaks on 2841 branch
2011-07-03 05:23:07 +02:00
for (cl = options->ClientTransportPlugin; cl; cl = cl->next) {
if (parse_client_transport_line(cl->value, 1)<0)
REJECT("Transport line did not parse. See logs for details.");
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
}
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
for (cl = options->ServerTransportPlugin; cl; cl = cl->next) {
if (parse_server_transport_line(cl->value, 1)<0)
REJECT("Server transport line did not parse. See logs for details.");
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
}
r13773@catbus: nickm | 2007-07-16 11:58:25 -0400 Initial "constrained socket buffers" patch from coderman. needs tweaking. svn:r10842
2007-07-16 18:23:34 +02:00
if (options->ConstrainedSockets) {
/* If the user wants to constrain socket buffer use, make sure the desired
* limit is between MIN|MAX_TCPSOCK_BUFFER in k increments. */
r13774@catbus: nickm | 2007-07-16 12:23:28 -0400 Tweaks on constrained socket buffers patch from coderman: Add a changelog; rename some variables; fix some long lines and whitespace; make ConstrainedSockSize a memunit; pass setsockopt a void. svn:r10843
2007-07-16 18:23:36 +02:00
if (options->ConstrainedSockSize < MIN_CONSTRAINED_TCP_BUFFER ||
options->ConstrainedSockSize > MAX_CONSTRAINED_TCP_BUFFER ||
options->ConstrainedSockSize % 1024) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
r13773@catbus: nickm | 2007-07-16 11:58:25 -0400 Initial "constrained socket buffers" patch from coderman. needs tweaking. svn:r10842
2007-07-16 18:23:34 +02:00
"ConstrainedSockSize is invalid. Must be a value between %d and %d "
"in 1024 byte increments.",
r13774@catbus: nickm | 2007-07-16 12:23:28 -0400 Tweaks on constrained socket buffers patch from coderman: Add a changelog; rename some variables; fix some long lines and whitespace; make ConstrainedSockSize a memunit; pass setsockopt a void. svn:r10843
2007-07-16 18:23:36 +02:00
MIN_CONSTRAINED_TCP_BUFFER, MAX_CONSTRAINED_TCP_BUFFER);
r13773@catbus: nickm | 2007-07-16 11:58:25 -0400 Initial "constrained socket buffers" patch from coderman. needs tweaking. svn:r10842
2007-07-16 18:23:34 +02:00
return -1;
}
if (options->DirPort) {
/* Providing cached directory entries while system TCP buffers are scarce
* will exacerbate the socket errors. Suggest that this be disabled. */
COMPLAIN("You have requested constrained socket buffers while also "
"serving directory entries via DirPort. It is strongly "
"suggested that you disable serving directory requests when "
"system TCP buffer resources are scarce.");
}
}
r14017@Kushana: nickm | 2007-08-14 15:13:14 -0400 Enable more code for v3 voting timing information. Also, spec clarification. svn:r11107
2007-08-14 22:19:40 +02:00
if (options->V3AuthVoteDelay + options->V3AuthDistDelay >=
options->V3AuthVotingInterval/2) {
New config option ServerDNSAllowBrokenResolvConf to start a relay even when the local resolv.conf file is missing, broken, or contains only unusable nameservers. Now I can run a local network on my laptop when I'm on an airplane. svn:r13402
2008-02-06 13:46:17 +01:00
REJECT("V3AuthVoteDelay plus V3AuthDistDelay must be less than half "
r14017@Kushana: nickm | 2007-08-14 15:13:14 -0400 Enable more code for v3 voting timing information. Also, spec clarification. svn:r11107
2007-08-14 22:19:40 +02:00
"V3AuthVotingInterval");
}
r14032@Kushana: nickm | 2007-08-15 11:35:16 -0400 Add unit tests for median functions; enforce sensible ranges for intervals and delays. svn:r11121
2007-08-15 17:38:58 +02:00
if (options->V3AuthVoteDelay < MIN_VOTE_SECONDS)
REJECT("V3AuthVoteDelay is way too low.");
if (options->V3AuthDistDelay < MIN_DIST_SECONDS)
REJECT("V3AuthDistDelay is way too low.");
if (options->V3AuthNIntervalsValid < 2)
r14017@Kushana: nickm | 2007-08-14 15:13:14 -0400 Enable more code for v3 voting timing information. Also, spec clarification. svn:r11107
2007-08-14 22:19:40 +02:00
REJECT("V3AuthNIntervalsValid must be at least 2.");
r14885@Kushana: nickm | 2007-10-11 10:36:16 -0400 Document minimal values for voting times. Use a macro instead of a magic number. Remove an "enforce this" xxxx020 that was already enforced. svn:r11868
2007-10-11 18:06:42 +02:00
if (options->V3AuthVotingInterval < MIN_VOTE_INTERVAL) {
r14017@Kushana: nickm | 2007-08-14 15:13:14 -0400 Enable more code for v3 voting timing information. Also, spec clarification. svn:r11107
2007-08-14 22:19:40 +02:00
REJECT("V3AuthVotingInterval is insanely low.");
} else if (options->V3AuthVotingInterval > 24*60*60) {
REJECT("V3AuthVotingInterval is insanely high.");
} else if (((24*60*60) % options->V3AuthVotingInterval) != 0) {
COMPLAIN("V3AuthVotingInterval does not divide evenly into 24 hours.");
}
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
if (rend_config_services(options, 1) < 0)
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
REJECT("Failed to configure rendezvous options. See logs for details.");
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
Next patch from Karsten: client-side configuration stuff for proposal 121. svn:r16510
2008-08-12 18:12:26 +02:00
/* Parse client-side authorization for hidden services. */
if (rend_parse_service_authorization(options, 1) < 0)
REJECT("Failed to configure client authorization for hidden services. "
"See logs for details.");
[Forward-port ]Test and document last patch. svn:r6400
2006-04-18 05:51:18 +02:00
if (parse_virtual_addr_network(options->VirtualAddrNetwork, 1, NULL)<0)
[forward-port] Implement an option, VirtualAddrMask, to set which addresses get handed out in response to mapaddress requests. Needs testing and docs! svn:r6398
2006-04-18 05:36:28 +02:00
return -1;
checkpoint as we add PreferTunneledDirConns config option svn:r9249
2007-01-03 07:33:03 +01:00
if (options->PreferTunneledDirConns && !options->TunnelDirConns)
REJECT("Must set TunnelDirConns if PreferTunneledDirConns is set.");
Enable Tor to connect through SOCKS 4/5 proxies Added a sanity check in config.c and a check in directory.c directory_initiate_command_rend() to catch any direct connection attempts when a socks proxy is configured.
2009-06-19 01:59:18 +02:00
if ((options->Socks4Proxy || options->Socks5Proxy) &&
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
!options->HTTPProxy && !options->PreferTunneledDirConns)
Enable Tor to connect through SOCKS 4/5 proxies Added a sanity check in config.c and a check in directory.c directory_initiate_command_rend() to catch any direct connection attempts when a socks proxy is configured.
2009-06-19 01:59:18 +02:00
REJECT("When Socks4Proxy or Socks5Proxy is configured, "
"PreferTunneledDirConns and TunnelDirConns must both be "
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
"set to 1, or HTTPProxy must be configured.");
Enable Tor to connect through SOCKS 4/5 proxies Added a sanity check in config.c and a check in directory.c directory_initiate_command_rend() to catch any direct connection attempts when a socks proxy is configured.
2009-06-19 01:59:18 +02:00
r12946@catbus: nickm | 2007-05-25 10:46:50 -0400 Add an AutomapHostsOnResolve option. It seems to work. svn:r10324
2007-05-25 16:48:16 +02:00
if (options->AutomapHostsSuffixes) {
SMARTLIST_FOREACH(options->AutomapHostsSuffixes, char *, suf,
{
size_t len = strlen(suf);
if (len && suf[len-1] == '.')
suf[len-1] = '\0';
});
}
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
if (options->TestingTorNetwork && !options->DirServers) {
REJECT("TestingTorNetwork may only be configured in combination with "
"a non-default set of DirServers.");
}
warn more about AllowSingleHopExits
2010-11-16 06:13:25 +01:00
if (options->AllowSingleHopExits && !options->DirServers) {
COMPLAIN("You have set AllowSingleHopExits; now your relay will allow "
"others to make one-hop exits. However, since by default most "
"clients avoid relays that set this option, most clients will "
"ignore you.");
}
Triage the XXX022 and XXX021 comments remaining in the code Remove some, postpone others, leave some alone. Now the only remaining XXX022s are ones that seem important to fix or investigate.
2011-03-25 21:01:16 +01:00
/*XXXX023 checking for defaults manually like this is a bit fragile.*/
Comments and doc tweaks on Karsten's testing-dir-networks patch svn:r15254
2008-06-14 18:11:37 +02:00
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
/* Keep changes to hard-coded values synchronous to man page and default
* values table. */
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
if (options->TestingV3AuthInitialVotingInterval != 30*60 &&
Tweak the bug2250 fix slightly Rename the _UsingTestingTorNetwork hidden option to _UsingTestNetworkDefaults (thanks Nick for the suggestion) and added a changes file.
2011-03-01 10:05:18 +01:00
!options->TestingTorNetwork && !options->_UsingTestNetworkDefaults) {
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
REJECT("TestingV3AuthInitialVotingInterval may only be changed in testing "
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
"Tor networks!");
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
} else if (options->TestingV3AuthInitialVotingInterval < MIN_VOTE_INTERVAL) {
REJECT("TestingV3AuthInitialVotingInterval is insanely low.");
} else if (((30*60) % options->TestingV3AuthInitialVotingInterval) != 0) {
REJECT("TestingV3AuthInitialVotingInterval does not divide evenly into "
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
"30 minutes.");
}
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
if (options->TestingV3AuthInitialVoteDelay != 5*60 &&
Tweak the bug2250 fix slightly Rename the _UsingTestingTorNetwork hidden option to _UsingTestNetworkDefaults (thanks Nick for the suggestion) and added a changes file.
2011-03-01 10:05:18 +01:00
!options->TestingTorNetwork && !options->_UsingTestNetworkDefaults) {
Fix GETINFO config-text for private networks In private networks, the defaults for some options are changed. This means that in options_validate(), where we're testing that the defaults are what we think they are, we fail. Use a workaround by setting a hidden configuration option _UsingTestingTorNetwork when we have altered the configuration this way, so that options_validate() can do the right thing. Fixes bug 2250, bugfix on 0.2.1.2-alpha (the version introducing private network options).
2011-02-26 09:06:09 +01:00
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
REJECT("TestingV3AuthInitialVoteDelay may only be changed in testing "
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
"Tor networks!");
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
} else if (options->TestingV3AuthInitialVoteDelay < MIN_VOTE_SECONDS) {
REJECT("TestingV3AuthInitialVoteDelay is way too low.");
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
}
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
if (options->TestingV3AuthInitialDistDelay != 5*60 &&
Tweak the bug2250 fix slightly Rename the _UsingTestingTorNetwork hidden option to _UsingTestNetworkDefaults (thanks Nick for the suggestion) and added a changes file.
2011-03-01 10:05:18 +01:00
!options->TestingTorNetwork && !options->_UsingTestNetworkDefaults) {
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
REJECT("TestingV3AuthInitialDistDelay may only be changed in testing "
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
"Tor networks!");
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
} else if (options->TestingV3AuthInitialDistDelay < MIN_DIST_SECONDS) {
REJECT("TestingV3AuthInitialDistDelay is way too low.");
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
}
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
if (options->TestingV3AuthInitialVoteDelay +
options->TestingV3AuthInitialDistDelay >=
options->TestingV3AuthInitialVotingInterval/2) {
REJECT("TestingV3AuthInitialVoteDelay plus TestingV3AuthInitialDistDelay "
"must be less than half TestingV3AuthInitialVotingInterval");
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
}
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
if (options->TestingAuthDirTimeToLearnReachability != 30*60 &&
Tweak the bug2250 fix slightly Rename the _UsingTestingTorNetwork hidden option to _UsingTestNetworkDefaults (thanks Nick for the suggestion) and added a changes file.
2011-03-01 10:05:18 +01:00
!options->TestingTorNetwork && !options->_UsingTestNetworkDefaults) {
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
REJECT("TestingAuthDirTimeToLearnReachability may only be changed in "
"testing Tor networks!");
} else if (options->TestingAuthDirTimeToLearnReachability < 0) {
REJECT("TestingAuthDirTimeToLearnReachability must be non-negative.");
} else if (options->TestingAuthDirTimeToLearnReachability > 2*60*60) {
COMPLAIN("TestingAuthDirTimeToLearnReachability is insanely high.");
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
}
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
if (options->TestingEstimatedDescriptorPropagationTime != 10*60 &&
Tweak the bug2250 fix slightly Rename the _UsingTestingTorNetwork hidden option to _UsingTestNetworkDefaults (thanks Nick for the suggestion) and added a changes file.
2011-03-01 10:05:18 +01:00
!options->TestingTorNetwork && !options->_UsingTestNetworkDefaults) {
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
REJECT("TestingEstimatedDescriptorPropagationTime may only be changed in "
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
"testing Tor networks!");
Added prefixes to testing-network-only configuration options. svn:r15375
2008-06-20 19:03:13 +02:00
} else if (options->TestingEstimatedDescriptorPropagationTime < 0) {
REJECT("TestingEstimatedDescriptorPropagationTime must be non-negative.");
} else if (options->TestingEstimatedDescriptorPropagationTime > 60*60) {
COMPLAIN("TestingEstimatedDescriptorPropagationTime is insanely high.");
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
}
if (options->TestingTorNetwork) {
log_warn(LD_CONFIG, "TestingTorNetwork is set. This will make your node "
"almost unusable in the public Tor network, and is "
"therefore only advised if you are building a "
"testing Tor network!");
}
Add support for dynamic OpenSSL hardware crypto acceleration engines.
2009-05-24 01:42:44 +02:00
if (options->AccelName && !options->HardwareAccel)
options->HardwareAccel = 1;
if (options->AccelDir && !options->AccelName)
REJECT("Can't use hardware crypto accelerator dir without engine name.");
Bug #1107: Complain if PublishServerDescriptor receives 0 or 1 in a list 0 or 1 should only be passed as the sole argument. Warn for now, reject in 0.2.3.x.
2010-08-01 00:06:40 +02:00
if (options->PublishServerDescriptor)
SMARTLIST_FOREACH(options->PublishServerDescriptor, const char *, pubdes, {
if (!strcmp(pubdes, "1") || !strcmp(pubdes, "0"))
if (smartlist_len(options->PublishServerDescriptor) > 1) {
COMPLAIN("You have passed a list of multiple arguments to the "
"PublishServerDescriptor option that includes 0 or 1. "
"0 or 1 should only be used as the sole argument. "
"This configuration will be rejected in a future release.");
break;
}
});
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (options->BridgeRelay == 1 && ! options->ORPort)
REJECT("BridgeRelay is 1, ORPort is not set. This is an invalid "
"combination.");
Bug #928: Disallow BridgeRelay 1 and ORPort 0 configuration
2010-08-01 01:42:29 +02:00
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
return 0;
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
#undef REJECT
#undef COMPLAIN
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
}
Validate log options during options_validate(). Reject unrecognized log types like "Log notice magic-twanger". Also, make sure User and Group stay the same, and do not crash when somebody tries to change PidFile to NULL. svn:r2778
2004-11-10 04:40:30 +01:00
/** Helper: return true iff s1 and s2 are both NULL, or both non-NULL
* equal strings. */
static int
opt_streq(const char *s1, const char *s2)
{
Turn streq_opt into a generic strcmp_opt.
2011-07-19 08:36:11 +02:00
return 0 == strcmp_opt(s1, s2);
Validate log options during options_validate(). Reject unrecognized log types like "Log notice magic-twanger". Also, make sure User and Group stay the same, and do not crash when somebody tries to change PidFile to NULL. svn:r2778
2004-11-10 04:40:30 +01:00
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
/** Check if any of the previous options have changed but aren't allowed to. */
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
options_transition_allowed(const or_options_t *old,
const or_options_t *new_val,
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
char **msg)
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
{
Normalize space: add one between every control keyword and control clause. svn:r3003
2004-11-28 10:05:49 +01:00
if (!old)
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
return 0;
Validate log options during options_validate(). Reject unrecognized log types like "Log notice magic-twanger". Also, make sure User and Group stay the same, and do not crash when somebody tries to change PidFile to NULL. svn:r2778
2004-11-10 04:40:30 +01:00
if (!opt_streq(old->PidFile, new_val->PidFile)) {
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
*msg = tor_strdup("PidFile is not allowed to change.");
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
return -1;
}
first iteration of scrubbing sensitive strings from logs. also generally clean up log messages. svn:r4174
2005-05-03 12:04:08 +02:00
if (old->RunAsDaemon != new_val->RunAsDaemon) {
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
*msg = tor_strdup("While Tor is running, changing RunAsDaemon "
"is not allowed.");
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
return -1;
}
once normalized, DataDirectory is always defined svn:r2734
2004-11-09 08:20:21 +01:00
if (strcmp(old->DataDirectory,new_val->DataDirectory)!=0) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg,
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
"While Tor is running, changing DataDirectory "
"(\"%s\"->\"%s\") is not allowed.",
old->DataDirectory, new_val->DataDirectory);
Stop DataDirectory from changing; also stop using new as an identifier? svn:r2717
2004-11-09 05:46:24 +01:00
return -1;
}
Validate log options during options_validate(). Reject unrecognized log types like "Log notice magic-twanger". Also, make sure User and Group stay the same, and do not crash when somebody tries to change PidFile to NULL. svn:r2778
2004-11-10 04:40:30 +01:00
if (!opt_streq(old->User, new_val->User)) {
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
*msg = tor_strdup("While Tor is running, changing User is not allowed.");
Validate log options during options_validate(). Reject unrecognized log types like "Log notice magic-twanger". Also, make sure User and Group stay the same, and do not crash when somebody tries to change PidFile to NULL. svn:r2778
2004-11-10 04:40:30 +01:00
return -1;
}
Add support for dynamic OpenSSL hardware crypto acceleration engines.
2009-05-24 01:42:44 +02:00
if ((old->HardwareAccel != new_val->HardwareAccel)
Fix-ups for dynamic OpenSSL engine patch. Include a changelog, and don't try to compare strings with !=.
2009-05-31 19:36:18 +02:00
|| !opt_streq(old->AccelName, new_val->AccelName)
|| !opt_streq(old->AccelDir, new_val->AccelDir)) {
Add support for dynamic OpenSSL hardware crypto acceleration engines.
2009-05-24 01:42:44 +02:00
*msg = tor_strdup("While Tor is running, changing OpenSSL hardware "
"acceleration engine is not allowed.");
Load hardware acceleration options when/where available. Can anybody test this? svn:r4467
2005-06-20 20:56:35 +02:00
return -1;
}
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
if (old->TestingTorNetwork != new_val->TestingTorNetwork) {
*msg = tor_strdup("While Tor is running, changing TestingTorNetwork "
"is not allowed.");
Write all statistics to disk exactly every 24 hours.
2009-08-19 15:41:12 +02:00
return -1;
}
Implement DisableAllSwap to avoid putting secret info in page files. This commit implements a new config option: 'DisableAllSwap' This option probably only works properly when Tor is started as root. We added two new functions: tor_mlockall() and tor_set_max_memlock(). tor_mlockall() attempts to mlock() all current and all future memory pages. For tor_mlockall() to work properly we set the process rlimits for memory to RLIM_INFINITY (and beyond) inside of tor_set_max_memlock(). We behave differently from mlockall() by only allowing tor_mlockall() to be called one single time. All other calls will result in a return code of 1. It is not possible to change DisableAllSwap while running. A sample configuration item was added to the torrc.complete.in config file. A new item in the man page for DisableAllSwap was added. Thanks to Moxie Marlinspike and Chris Palmer for their feedback on this patch. Please note that we make no guarantees about the quality of your OS and its mlock/mlockall implementation. It is possible that this will do nothing at all. It is also possible that you can ulimit the mlock properties of a given user such that root is not required. This has not been extensively tested and is unsupported. I have included some comments for possible ways we can handle this on win32.
2009-10-22 06:21:57 +02:00
if (old->DisableAllSwap != new_val->DisableAllSwap) {
*msg = tor_strdup("While Tor is running, changing DisableAllSwap "
"is not allowed.");
Disable run-time changes to DisableIOCP: They do not work
2011-08-19 23:07:54 +02:00
return -1;
}
Move around check for TokenBucketRefillInterval; disallow changes to it
2011-09-08 03:02:49 +02:00
if (old->TokenBucketRefillInterval != new_val->TokenBucketRefillInterval) {
*msg = tor_strdup("While Tor is running, changing TokenBucketRefill"
"Interval is not allowed");
return -1;
}
Disable run-time changes to DisableIOCP: They do not work
2011-08-19 23:07:54 +02:00
if (old->DisableIOCP != new_val->DisableIOCP) {
*msg = tor_strdup("While Tor is running, changing DisableIOCP "
"is not allowed.");
Disallow disabling DisableDebuggerAttachment on runnning Tor Also, have tor_disable_debugger_attach() return a tristate of success/failure/don't-know-how , and only log appropriately.
2011-12-08 09:19:09 +01:00
return -1;
}
if (old->DisableDebuggerAttachment &&
!new_val->DisableDebuggerAttachment) {
*msg = tor_strdup("While Tor is running, disabling "
"DisableDebuggerAttachment is not allowed.");
Implement DisableAllSwap to avoid putting secret info in page files. This commit implements a new config option: 'DisableAllSwap' This option probably only works properly when Tor is started as root. We added two new functions: tor_mlockall() and tor_set_max_memlock(). tor_mlockall() attempts to mlock() all current and all future memory pages. For tor_mlockall() to work properly we set the process rlimits for memory to RLIM_INFINITY (and beyond) inside of tor_set_max_memlock(). We behave differently from mlockall() by only allowing tor_mlockall() to be called one single time. All other calls will result in a return code of 1. It is not possible to change DisableAllSwap while running. A sample configuration item was added to the torrc.complete.in config file. A new item in the man page for DisableAllSwap was added. Thanks to Moxie Marlinspike and Chris Palmer for their feedback on this patch. Please note that we make no guarantees about the quality of your OS and its mlock/mlockall implementation. It is possible that this will do nothing at all. It is also possible that you can ulimit the mlock properties of a given user such that root is not required. This has not been extensively tested and is unsupported. I have included some comments for possible ways we can handle this on win32.
2009-10-22 06:21:57 +02:00
return -1;
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
return 0;
}
fix function headers from the descriptor-frequency commit svn:r4809
2005-08-22 08:45:18 +02:00
/** Return 1 if any change from <b>old_options</b> to <b>new_options</b>
Spell-check Tor.
2009-05-27 23:55:51 +02:00
* will require us to rotate the CPU and DNS workers; else return 0. */
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
options_transition_affects_workers(const or_options_t *old_options,
const or_options_t *new_options)
rotate dns and cpu workers if the controller changes options that will affect them. svn:r4787
2005-08-16 01:46:18 +02:00
{
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
if (!opt_streq(old_options->DataDirectory, new_options->DataDirectory) ||
Synx manpage and source wrt option capitalization We had a spelling discrepancy between the manpage and the source code for some option. Resolve these in favor of the manpage, because it makes more sense (for example, HTTP should be capitalized).
2010-10-29 19:41:24 +02:00
old_options->NumCPUs != new_options->NumCPUs ||
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
!config_lines_eq(old_options->ORPort, new_options->ORPort) ||
r8874@Kushana: nickm | 2006-09-21 15:22:27 -0400 Rename and document SearchDomains and ResolvConf options; warn if ServerDNSResolvConfFile is given but eventdns isnt enabled. svn:r8439
2006-09-21 23:48:16 +02:00
old_options->ServerDNSSearchDomains !=
new_options->ServerDNSSearchDomains ||
Don't recreate descriptor on sighup We used to regenerate our descriptor whenever we'd get a sighup. This was caused by a bug in options_transition_affects_workers() that would return true even if the options were exactly the same. Down the call path we'd call init_keys(), which made us make a new descriptor which the authorities would reject, and the node would subsequently fall out of the consensus. This patch fixes only the first part of this bug: options_transition_affects_workers() behaves correctly now. The second part still wants a fix.
2011-05-21 19:10:11 +02:00
old_options->_SafeLogging != new_options->_SafeLogging ||
If we start a server with ClientOnly 1, then set ClientOnly to 0 and hup, stop triggering an assert based on an empty onion_key. svn:r9540
2007-02-09 01:22:43 +01:00
old_options->ClientOnly != new_options->ClientOnly ||
Fix assert for relay/bridge state change When we added support for separate client tls certs on bridges in a2bb0bfdd5 we forgot to correctly initialize this when changing from relay to bridge or vice versa while Tor is running. Fix that by always initializing keys when the state changes. Fixes bug 2433.
2011-01-25 15:28:58 +01:00
public_server_mode(old_options) != public_server_mode(new_options) ||
Add a torrc option to report log domains
2011-01-25 21:53:15 +01:00
!config_lines_eq(old_options->Logs, new_options->Logs) ||
old_options->LogMessageDomains != new_options->LogMessageDomains)
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
return 1;
/* Check whether log options match. */
/* Nothing that changed matters. */
return 0;
}
fix function headers from the descriptor-frequency commit svn:r4809
2005-08-22 08:45:18 +02:00
/** Return 1 if any change from <b>old_options</b> to <b>new_options</b>
* will require us to generate a new descriptor; else return 0. */
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
options_transition_affects_descriptor(const or_options_t *old_options,
const or_options_t *new_options)
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
{
Downlgrade tweak, and answer lots of XXX021s. No actual code fixes in this patch. svn:r17686
2008-12-18 17:11:24 +01:00
/* XXX We can be smarter here. If your DirPort isn't being
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
* published and you just turned it off, no need to republish. Etc. */
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
if (!opt_streq(old_options->DataDirectory, new_options->DataDirectory) ||
!opt_streq(old_options->Nickname,new_options->Nickname) ||
!opt_streq(old_options->Address,new_options->Address) ||
!config_lines_eq(old_options->ExitPolicy,new_options->ExitPolicy) ||
cleanups, bump to 0.2.0.12-alpha svn:r12516
2007-11-17 03:22:56 +01:00
old_options->ExitPolicyRejectPrivate !=
new_options->ExitPolicyRejectPrivate ||
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
!config_lines_eq(old_options->ORPort, new_options->ORPort) ||
!config_lines_eq(old_options->DirPort, new_options->DirPort) ||
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
old_options->ClientOnly != new_options->ClientOnly ||
New 'DisableNetwork' option to prevent Tor from using the network Some controllers want this so they can mess with Tor's configuration for a while via the control port before actually letting Tor out of the house. We do this with a new DisableNetwork option, that prevents Tor from making any outbound connections or binding any non-control listeners. Additionally, it shuts down the same functionality as shuts down when we are hibernating, plus the code that launches directory downloads. To make sure I didn't miss anything, I added a clause straight to connection_connect, so that we won't even try to open an outbound socket when the network is disabled. In my testing, I made this an assert, but since I probably missed something, I've turned it into a BUG warning for testing.
2011-11-28 21:44:10 +01:00
old_options->DisableNetwork != new_options->DisableNetwork ||
Change the PublishServerDescriptor config option from a boolean into a string: "v1", "v2", bridge", "". Continue to support "0" and "1". svn:r10136
2007-05-08 11:09:26 +02:00
old_options->_PublishServerDescriptor !=
new_options->_PublishServerDescriptor ||
Changing MaxAdvertisedBW may not need a republish Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian.
2009-07-07 18:04:00 +02:00
get_effective_bwrate(old_options) != get_effective_bwrate(new_options) ||
get_effective_bwburst(old_options) !=
get_effective_bwburst(new_options) ||
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
!opt_streq(old_options->ContactInfo, new_options->ContactInfo) ||
!opt_streq(old_options->MyFamily, new_options->MyFamily) ||
!opt_streq(old_options->AccountingStart, new_options->AccountingStart) ||
Have transitions in public_server_mode count as affects_descriptor Previously, we'd get a new descriptor for free when public_server_mode() changed, since it would count as affects_workers, which would call init_keys(), which would make us regenerate a new descriptor. But now that we fixed bug 3263, init_keys() is no longer necessarily a new descriptor, and so we need to make sure that public_server_mode() counts as a descriptor transition.
2011-07-07 17:05:06 +02:00
old_options->AccountingMax != new_options->AccountingMax ||
public_server_mode(old_options) != public_server_mode(new_options))
Make set_options a little smarter: have options_act handle transitions on its own, and only dirty our descriptor when we really want to. svn:r4802
2005-08-22 02:18:45 +02:00
return 1;
rotate dns and cpu workers if the controller changes options that will affect them. svn:r4787
2005-08-16 01:46:18 +02:00
return 0;
}
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifdef _WIN32
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
/** Return the directory on windows where we expect to find our application
* data. */
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
static char *
get_windows_conf_root(void)
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
{
static int is_set = 0;
static char path[MAX_PATH+1];
Make the windows build succeed with or without -DUNICODE enabled. This should keep WinCE working (unicode always-on) and get Win98 working again (unicode never-on). There are two places where we explicitly use ASCII-only APIs, still: in ntmain.c and in the unit tests. This patch also fixes a bug in windoes tor_listdir that would cause the first file to be listed an arbitrary number of times that was also introduced with WinCE support. Should fix bug 1797.
2010-08-20 18:30:25 +02:00
TCHAR tpath[MAX_PATH] = {0};
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
LPITEMIDLIST idl;
IMalloc *m;
HRESULT result;
if (is_set)
return path;
Spell-check strings and comments svn:r3052
2004-12-01 04:48:14 +01:00
/* Find X:\documents and settings\username\application data\ .
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
* We would use SHGetSpecialFolder path, but that wasn't added until IE4.
*/
This patch changes the default location where config and data files are stored when the --enable-local-appdata option is configured. This changes the Windows path from %APPDATA% to a host local %USERPROFILE%\Local Settings\Application Data\ path (aka, LOCAL_APPDATA). Patch from coderman. svn:r18122
2009-01-16 00:07:11 +01:00
#ifdef ENABLE_LOCAL_APPDATA
#define APPDATA_PATH CSIDL_LOCAL_APPDATA
#else
#define APPDATA_PATH CSIDL_APPDATA
#endif
if (!SUCCEEDED(SHGetSpecialFolderLocation(NULL, APPDATA_PATH, &idl))) {
Port Tor to work on Windows CE Most of the changes here are switches to use APIs available on Windows CE. The most pervasive change is that Windows CE only provides the wide-character ("FooW") variants of most of the windows function, and doesn't support the older ASCII verions at all. This patch will require use of the wcecompat library to get working versions of the posix-style fd-based file IO functions. [commit message by nickm]
2010-05-21 04:53:39 +02:00
getcwd(path,MAX_PATH);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
is_set = 1;
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG,
"I couldn't find your application data folder: are you "
"running an ancient version of Windows 95? Defaulting to \"%s\"",
path);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
return path;
}
/* Convert the path from an "ID List" (whatever that is!) to a path. */
Make the windows build succeed with or without -DUNICODE enabled. This should keep WinCE working (unicode always-on) and get Win98 working again (unicode never-on). There are two places where we explicitly use ASCII-only APIs, still: in ntmain.c and in the unit tests. This patch also fixes a bug in windoes tor_listdir that would cause the first file to be listed an arbitrary number of times that was also introduced with WinCE support. Should fix bug 1797.
2010-08-20 18:30:25 +02:00
result = SHGetPathFromIDList(idl, tpath);
#ifdef UNICODE
wcstombs(path,tpath,MAX_PATH);
#else
strlcpy(path,tpath,sizeof(path));
#endif
Port Tor to work on Windows CE Most of the changes here are switches to use APIs available on Windows CE. The most pervasive change is that Windows CE only provides the wide-character ("FooW") variants of most of the windows function, and doesn't support the older ASCII verions at all. This patch will require use of the wcecompat library to get working versions of the posix-style fd-based file IO functions. [commit message by nickm]
2010-05-21 04:53:39 +02:00
Answer question from Roger.
2010-06-15 00:32:44 +02:00
/* Now we need to free the memory that the path-idl was stored in. In
* typical Windows fashion, we can't just call 'free()' on it. */
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
SHGetMalloc(&m);
if (m) {
m->lpVtbl->Free(m, idl);
m->lpVtbl->Release(m);
}
if (!SUCCEEDED(result)) {
return NULL;
}
strlcat(path,"\\tor",MAX_PATH);
is_set = 1;
return path;
}
#endif
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
/** Return the default location for our torrc file.
* DOCDOC defaults_file */
Add a config-file GETINFO entry; fix a minor memory leak on some SAVECONF calls. svn:r4761
2005-08-10 20:05:20 +02:00
static const char *
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
get_default_conf_file(int defaults_file)
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
{
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifdef _WIN32
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
if (defaults_file) {
static char defaults_path[MAX_PATH+1];
tor_snprintf(defaults_path, MAX_PATH, "%s\\torrc-defaults",
get_windows_conf_root());
return defaults_path;
} else {
static char path[MAX_PATH+1];
tor_snprintf(path, MAX_PATH, "%s\\torrc",
get_windows_conf_root());
return path;
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
#else
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
return defaults_file ? CONFDIR "/torrc-defaults" : CONFDIR "/torrc";
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
#endif
}
Now NodeFamily and MyFamily config options allow spaces in identity fingerprints, so it's easier to paste them in. Suggested by Lucky Green. svn:r17021
2008-10-01 05:41:33 +02:00
/** Verify whether lst is a string containing valid-looking comma-separated
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
* nicknames, or NULL. Return 0 on success. Warn and return -1 on failure.
*/
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
static int
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
check_nickname_list(const char *lst, const char *name, char **msg)
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
{
int r = 0;
smartlist_t *sl;
if (!lst)
return 0;
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
sl = smartlist_new();
Now NodeFamily and MyFamily config options allow spaces in identity fingerprints, so it's easier to paste them in. Suggested by Lucky Green. svn:r17021
2008-10-01 05:41:33 +02:00
smartlist_split_string(sl, lst, ",",
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK|SPLIT_STRIP_SPACE, 0);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
SMARTLIST_FOREACH(sl, const char *, s,
{
if (!is_legal_nickname_or_hexdigest(s)) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(msg, "Invalid nickname '%s' in %s line", s, name);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
r = -1;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
break;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
}
});
SMARTLIST_FOREACH(sl, char *, s, tor_free(s));
smartlist_free(sl);
return r;
}
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
/** Learn config file name from command line arguments, or use the default,
* DOCDOC defaults_file */
r18723@catbus: nickm | 2008-03-11 00:25:30 -0400 Fix bug spotted by mwenge: a server_event should not be a sever_event. Also, fix compile errors in config.c and control.c with --enable-gcc-warnings. svn:r13957
2008-03-11 05:30:14 +01:00
static char *
options_init_from_torrc(): move code that looks for torrc into its own function Part of options_init_from_torrc()'s job was looking for -f flags (to specify an alternate config file) on the command line, complaining if more than one is given or the given does not exist. If none is given then use the compiled-in default location, accepting if it does not exist. This logic has been moved into its own function in an attemped to make options_init_from_torrc() easier to deal with. svn:r13942
2008-03-10 13:41:29 +01:00
find_torrc_filename(int argc, char **argv,
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
int defaults_file,
options_init_from_torrc(): move code that looks for torrc into its own function Part of options_init_from_torrc()'s job was looking for -f flags (to specify an alternate config file) on the command line, complaining if more than one is given or the given does not exist. If none is given then use the compiled-in default location, accepting if it does not exist. This logic has been moved into its own function in an attemped to make options_init_from_torrc() easier to deal with. svn:r13942
2008-03-10 13:41:29 +01:00
int *using_default_torrc, int *ignore_missing_torrc)
{
char *fname=NULL;
int i;
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
const char *fname_opt = defaults_file ? "--defaults-torrc" : "-f";
const char *ignore_opt = defaults_file ? NULL : "--ignore-missing-torrc";
if (defaults_file)
*ignore_missing_torrc = 1;
options_init_from_torrc(): move code that looks for torrc into its own function Part of options_init_from_torrc()'s job was looking for -f flags (to specify an alternate config file) on the command line, complaining if more than one is given or the given does not exist. If none is given then use the compiled-in default location, accepting if it does not exist. This logic has been moved into its own function in an attemped to make options_init_from_torrc() easier to deal with. svn:r13942
2008-03-10 13:41:29 +01:00
for (i = 1; i < argc; ++i) {
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
if (i < argc-1 && !strcmp(argv[i],fname_opt)) {
options_init_from_torrc(): move code that looks for torrc into its own function Part of options_init_from_torrc()'s job was looking for -f flags (to specify an alternate config file) on the command line, complaining if more than one is given or the given does not exist. If none is given then use the compiled-in default location, accepting if it does not exist. This logic has been moved into its own function in an attemped to make options_init_from_torrc() easier to deal with. svn:r13942
2008-03-10 13:41:29 +01:00
if (fname) {
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
log(LOG_WARN, LD_CONFIG, "Duplicate %s options on command line.",
fname_opt);
options_init_from_torrc(): move code that looks for torrc into its own function Part of options_init_from_torrc()'s job was looking for -f flags (to specify an alternate config file) on the command line, complaining if more than one is given or the given does not exist. If none is given then use the compiled-in default location, accepting if it does not exist. This logic has been moved into its own function in an attemped to make options_init_from_torrc() easier to deal with. svn:r13942
2008-03-10 13:41:29 +01:00
tor_free(fname);
}
fname = expand_filename(argv[i+1]);
Prepend cwd for relative config file paths. Modifies filenames which do not start with '/' or '.' on non-Windows platforms; uses _fullpath on Windows.
2011-12-04 17:33:20 +01:00
{
char *absfname;
absfname = make_path_absolute(fname);
tor_free(fname);
fname = absfname;
}
options_init_from_torrc(): move code that looks for torrc into its own function Part of options_init_from_torrc()'s job was looking for -f flags (to specify an alternate config file) on the command line, complaining if more than one is given or the given does not exist. If none is given then use the compiled-in default location, accepting if it does not exist. This logic has been moved into its own function in an attemped to make options_init_from_torrc() easier to deal with. svn:r13942
2008-03-10 13:41:29 +01:00
*using_default_torrc = 0;
++i;
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
} else if (ignore_opt && !strcmp(argv[i],ignore_opt)) {
options_init_from_torrc(): move code that looks for torrc into its own function Part of options_init_from_torrc()'s job was looking for -f flags (to specify an alternate config file) on the command line, complaining if more than one is given or the given does not exist. If none is given then use the compiled-in default location, accepting if it does not exist. This logic has been moved into its own function in an attemped to make options_init_from_torrc() easier to deal with. svn:r13942
2008-03-10 13:41:29 +01:00
*ignore_missing_torrc = 1;
}
}
if (*using_default_torrc) {
/* didn't find one, try CONFDIR */
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
const char *dflt = get_default_conf_file(defaults_file);
options_init_from_torrc(): move code that looks for torrc into its own function Part of options_init_from_torrc()'s job was looking for -f flags (to specify an alternate config file) on the command line, complaining if more than one is given or the given does not exist. If none is given then use the compiled-in default location, accepting if it does not exist. This logic has been moved into its own function in an attemped to make options_init_from_torrc() easier to deal with. svn:r13942
2008-03-10 13:41:29 +01:00
if (dflt && file_status(dflt) == FN_FILE) {
fname = tor_strdup(dflt);
} else {
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifndef _WIN32
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
char *fn = NULL;
if (!defaults_file)
fn = expand_filename("~/.torrc");
options_init_from_torrc(): move code that looks for torrc into its own function Part of options_init_from_torrc()'s job was looking for -f flags (to specify an alternate config file) on the command line, complaining if more than one is given or the given does not exist. If none is given then use the compiled-in default location, accepting if it does not exist. This logic has been moved into its own function in an attemped to make options_init_from_torrc() easier to deal with. svn:r13942
2008-03-10 13:41:29 +01:00
if (fn && file_status(fn) == FN_FILE) {
fname = fn;
} else {
tor_free(fn);
fname = tor_strdup(dflt);
}
#else
fname = tor_strdup(dflt);
#endif
}
}
return fname;
}
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
/** Load torrc from disk, setting torrc_fname if successful.
* DOCDOC defaults_file */
r18723@catbus: nickm | 2008-03-11 00:25:30 -0400 Fix bug spotted by mwenge: a server_event should not be a sever_event. Also, fix compile errors in config.c and control.c with --enable-gcc-warnings. svn:r13957
2008-03-11 05:30:14 +01:00
static char *
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
load_torrc_from_disk(int argc, char **argv, int defaults_file)
options_init_from_torrc(): move code that loads torrc into its own function move code that loads torrc from disk and sets torrc_fname into its own function svn:r13944
2008-03-10 13:41:36 +01:00
{
char *fname=NULL;
char *cf = NULL;
int using_default_torrc = 1;
int ignore_missing_torrc = 0;
Set torrc_fname in load_torrc_from_disk Bugfix on commit 230422b955e1708f27f42cdd25e8b21a33fdd3dd, not yet in any release. Fixes bug #4604; reported by koolfy.
2011-11-30 02:11:49 +01:00
char **fname_var = defaults_file ? &torrc_defaults_fname : &torrc_fname;
options_init_from_torrc(): move code that loads torrc into its own function move code that loads torrc from disk and sets torrc_fname into its own function svn:r13944
2008-03-10 13:41:36 +01:00
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
fname = find_torrc_filename(argc, argv, defaults_file,
options_init_from_torrc(): move code that loads torrc into its own function move code that loads torrc from disk and sets torrc_fname into its own function svn:r13944
2008-03-10 13:41:36 +01:00
&using_default_torrc, &ignore_missing_torrc);
tor_assert(fname);
log(LOG_DEBUG, LD_CONFIG, "Opening config file \"%s\"", fname);
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
tor_free(*fname_var);
*fname_var = fname;
options_init_from_torrc(): move code that loads torrc into its own function move code that loads torrc from disk and sets torrc_fname into its own function svn:r13944
2008-03-10 13:41:36 +01:00
/* Open config file */
if (file_status(fname) != FN_FILE ||
!(cf = read_file_to_str(fname,0,NULL))) {
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
if (using_default_torrc == 1 || ignore_missing_torrc) {
if (!defaults_file)
log(LOG_NOTICE, LD_CONFIG, "Configuration file \"%s\" not present, "
"using reasonable defaults.", fname);
options_init_from_torrc(): move code that loads torrc into its own function move code that loads torrc from disk and sets torrc_fname into its own function svn:r13944
2008-03-10 13:41:36 +01:00
tor_free(fname); /* sets fname to NULL */
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
*fname_var = NULL;
options_init_from_torrc(): move code that loads torrc into its own function move code that loads torrc from disk and sets torrc_fname into its own function svn:r13944
2008-03-10 13:41:36 +01:00
cf = tor_strdup("");
} else {
log(LOG_WARN, LD_CONFIG,
"Unable to open configuration file \"%s\".", fname);
goto err;
}
Log which config file we read It is often not entirely clear what options Tor was built with, so it might not be immediately obvious which config file Tor is using when it found one. Log the config file at startup.
2011-01-27 15:31:34 +01:00
} else {
log(LOG_NOTICE, LD_CONFIG, "Read configuration file \"%s\".", fname);
options_init_from_torrc(): move code that loads torrc into its own function move code that loads torrc from disk and sets torrc_fname into its own function svn:r13944
2008-03-10 13:41:36 +01:00
}
return cf;
err:
tor_free(fname);
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
*fname_var = NULL;
options_init_from_torrc(): move code that loads torrc into its own function move code that loads torrc from disk and sets torrc_fname into its own function svn:r13944
2008-03-10 13:41:36 +01:00
return NULL;
}
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
/** Read a configuration file into <b>options</b>, finding the configuration
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
* file location based on the command line. After loading the file
* call options_init_from_string() to load the config.
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
* Return 0 if success, -1 if failure. */
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
int
Rename cross-format config.c stuff to config_* and or_options_t-specific stuff to option[s]_* svn:r4641
2005-07-23 03:58:05 +02:00
options_init_from_torrc(int argc, char **argv)
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
{
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
char *cf=NULL, *cf_defaults=NULL;
Fix a memory leak in error path of my default-torrc code Fixes Coverity CID # 500
2011-11-30 18:16:39 +01:00
int i, command;
int retval = -1;
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
static char **backup_argv;
static int backup_argc;
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
char *command_arg = NULL;
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
char *errmsg=NULL;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
Spell-check Tor.
2009-05-27 23:55:51 +02:00
if (argv) { /* first time we're called. save command line args */
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
backup_argv = argv;
backup_argc = argc;
} else { /* we're reloading. need to clean up old options first. */
argv = backup_argv;
argc = backup_argc;
}
if (argc > 1 && (!strcmp(argv[1], "-h") || !strcmp(argv[1],"--help"))) {
print_usage();
exit(0);
}
r11651@Kushana: nickm | 2006-12-20 12:05:04 -0500 Add a maintainer script and a new make target "make check-docs" to get a quick dump of which options are undocumented where, and which documentation refers to nonexistent options. svn:r9160
2006-12-20 18:05:48 +01:00
if (argc > 1 && !strcmp(argv[1], "--list-torrc-options")) {
/* For documenting validating whether we've documented everything. */
list_torrc_options();
exit(0);
}
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
if (argc > 1 && (!strcmp(argv[1],"--version"))) {
r12852@catbus: nickm | 2007-05-22 11:00:27 -0400 Use svn revisions consistently throughout all log messages. svn:r10291
2007-05-22 17:48:46 +02:00
printf("Tor version %s.\n",get_version());
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
exit(0);
}
Command-line option to dump SHA1 digests of all source files. Now, when you call tor --digests, it dumps the SHA1 digest of each source file that Tor was built with. We support both 'sha1sum' and 'openssl sha1'. If the user is building from a tarball and they haven't edited anything, they don't need any program that calculates SHA1. If they _have_ modified a file but they don't have a program to calculate SHA1, we try to build so we do not output digests.
2009-01-04 03:43:05 +01:00
if (argc > 1 && (!strcmp(argv[1],"--digests"))) {
printf("Tor version %s.\n",get_version());
printf("%s", libor_get_digests());
Include the *_sha1.i files in their own *_codedigest.c files. This way we do not need to rebuild util.c and/or config.c whenever any unrelated source file in src/common or src/or has changed.
2009-05-08 18:35:36 +02:00
printf("%s", tor_get_digests());
Command-line option to dump SHA1 digests of all source files. Now, when you call tor --digests, it dumps the SHA1 digest of each source file that Tor was built with. We support both 'sha1sum' and 'openssl sha1'. If the user is building from a tarball and they haven't edited anything, they don't need any program that calculates SHA1. If they _have_ modified a file but they don't have a program to calculate SHA1, we try to build so we do not output digests.
2009-01-04 03:43:05 +01:00
exit(0);
}
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
Store options we got from commandline seperately from bare argv/argc svn:r13945
2008-03-10 13:41:40 +01:00
/* Go through command-line variables */
if (!global_cmdline_options) {
/* Or we could redo the list every time we pass this place.
* It does not really matter */
if (config_get_commandlines(argc, argv, &global_cmdline_options) < 0) {
goto err;
}
}
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
command = CMD_RUN_TOR;
options_init_from_torrc(): Split argv processing into two parts Split the argv processing loop into two poarts, one that deals with figuring out which conffile to use, and the other that figures out which "command" (hash fingerprint, verify config, list fpr, run tor) the user asked for. There is a third part further down that imports command line args into the config but that is not touched. svn:r13941
2008-03-10 13:41:26 +01:00
for (i = 1; i < argc; ++i) {
if (!strcmp(argv[i],"--list-fingerprint")) {
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
command = CMD_LIST_FINGERPRINT;
options_init_from_torrc(): Split argv processing into two parts Split the argv processing loop into two poarts, one that deals with figuring out which conffile to use, and the other that figures out which "command" (hash fingerprint, verify config, list fpr, run tor) the user asked for. There is a third part further down that imports command line args into the config but that is not touched. svn:r13941
2008-03-10 13:41:26 +01:00
} else if (!strcmp(argv[i],"--hash-password")) {
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
command = CMD_HASH_PASSWORD;
command_arg = tor_strdup( (i < argc-1) ? argv[i+1] : "");
options_init_from_torrc(): Split argv processing into two parts Split the argv processing loop into two poarts, one that deals with figuring out which conffile to use, and the other that figures out which "command" (hash fingerprint, verify config, list fpr, run tor) the user asked for. There is a third part further down that imports command line args into the config but that is not touched. svn:r13941
2008-03-10 13:41:26 +01:00
++i;
} else if (!strcmp(argv[i],"--verify-config")) {
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
command = CMD_VERIFY_CONFIG;
options_init_from_torrc(): Split argv processing into two parts Split the argv processing loop into two poarts, one that deals with figuring out which conffile to use, and the other that figures out which "command" (hash fingerprint, verify config, list fpr, run tor) the user asked for. There is a third part further down that imports command line args into the config but that is not touched. svn:r13941
2008-03-10 13:41:26 +01:00
}
}
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
r19091@catbus: nickm | 2008-03-27 12:42:57 -0400 Stop reading torrc when all we want to do is --hash-password. Fix based on patch from Sebastian Hahn. Backport candidate. svn:r14214
2008-03-27 17:46:34 +01:00
if (command == CMD_HASH_PASSWORD) {
cf = tor_strdup("");
} else {
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
cf_defaults = load_torrc_from_disk(argc, argv, 1);
cf = load_torrc_from_disk(argc, argv, 0);
r19091@catbus: nickm | 2008-03-27 12:42:57 -0400 Stop reading torrc when all we want to do is --hash-password. Fix based on patch from Sebastian Hahn. Backport candidate. svn:r14214
2008-03-27 17:46:34 +01:00
if (!cf)
goto err;
}
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
appease "make check-spaces"
2011-11-30 18:09:53 +01:00
retval = options_init_from_string(cf_defaults, cf, command, command_arg,
&errmsg);
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
err:
Fix a memory leak in error path of my default-torrc code Fixes Coverity CID # 500
2011-11-30 18:16:39 +01:00
tor_free(cf);
tor_free(cf_defaults);
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
if (errmsg) {
log(LOG_WARN,LD_CONFIG,"%s", errmsg);
tor_free(errmsg);
}
Fix a memory leak in error path of my default-torrc code Fixes Coverity CID # 500
2011-11-30 18:16:39 +01:00
return retval < 0 ? -1 : 0;
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
}
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
/** Load the options from the configuration in <b>cf</b>, validate
* them for consistency and take actions based on them.
*
* Return 0 if success, negative on error:
* * -1 for general errors.
* * -2 for failure to parse/validate,
* * -3 for transition not allowed
* * -4 for error while setting the new options
*/
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
setopt_err_t
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
options_init_from_string(const char *cf_defaults, const char *cf,
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
int command, const char *command_arg,
char **msg)
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
{
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
or_options_t *oldoptions, *newoptions, *newdefaultoptions=NULL;
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
config_line_t *cl;
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
int retval, i;
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
setopt_err_t err = SETOPT_ERR_MISC;
r18753@catbus: nickm | 2008-03-11 14:56:39 -0400 Make some assert()s into tor_assert()s. Make some tor_assert()s called from logging into assert()s, and document why. svn:r13977
2008-03-11 19:56:41 +01:00
tor_assert(msg);
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
oldoptions = global_options; /* get_options unfortunately asserts if
this is the first time we run*/
newoptions = tor_malloc_zero(sizeof(or_options_t));
newoptions->_magic = OR_OPTIONS_MAGIC;
options_init(newoptions);
newoptions->command = command;
newoptions->command_arg = command_arg;
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
for (i = 0; i < 2; ++i) {
const char *body = i==0 ? cf_defaults : cf;
if (!body)
continue;
/* get config lines, assign them */
retval = config_get_lines(body, &cl, 1);
if (retval < 0) {
err = SETOPT_ERR_PARSE;
goto err;
}
retval = config_assign(&options_format, newoptions, cl, 0, 0, msg);
config_free_lines(cl);
if (retval < 0) {
err = SETOPT_ERR_PARSE;
goto err;
}
if (i==0)
newdefaultoptions = options_dup(&options_format, newoptions);
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
}
options_init_from_torrc(): tread non-existing torrc like empty torrc. Tread the case of a non-existing conffile, when allowed, exactly like the one with an empty torrc. svn:r13943
2008-03-10 13:41:33 +01:00
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
/* Go through command-line variables too */
Store options we got from commandline seperately from bare argv/argc svn:r13945
2008-03-10 13:41:40 +01:00
retval = config_assign(&options_format, newoptions,
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
global_cmdline_options, 0, 0, msg);
if (retval < 0) {
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
err = SETOPT_ERR_PARSE;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
goto err;
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
}
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
/* If this is a testing network configuration, change defaults
* for a list of dependent config options, re-initialize newoptions
* with the new defaults, and assign all options to it second time. */
if (newoptions->TestingTorNetwork) {
Downlgrade tweak, and answer lots of XXX021s. No actual code fixes in this patch. svn:r17686
2008-12-18 17:11:24 +01:00
/* XXXX this is a bit of a kludge. perhaps there's a better way to do
Comments and doc tweaks on Karsten's testing-dir-networks patch svn:r15254
2008-06-14 18:11:37 +02:00
* this? We could, for example, make the parsing algorithm do two passes
* over the configuration. If it finds any "suite" options like
* TestingTorNetwork, it could change the defaults before its second pass.
* Not urgent so long as this seems to work, but at any sign of trouble,
* let's clean it up. -NM */
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
/* Change defaults. */
int i;
for (i = 0; testing_tor_network_defaults[i].name; ++i) {
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *new_var = &testing_tor_network_defaults[i];
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
config_var_t *old_var =
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_find_option_mutable(&options_format, new_var->name);
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
tor_assert(new_var);
tor_assert(old_var);
old_var->initvalue = new_var->initvalue;
}
/* Clear newoptions and re-initialize them with new defaults. */
config_free(&options_format, newoptions);
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
config_free(&options_format, newdefaultoptions);
newdefaultoptions = NULL;
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
newoptions = tor_malloc_zero(sizeof(or_options_t));
newoptions->_magic = OR_OPTIONS_MAGIC;
options_init(newoptions);
newoptions->command = command;
newoptions->command_arg = command_arg;
/* Assign all options a second time. */
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
for (i = 0; i < 2; ++i) {
const char *body = i==0 ? cf_defaults : cf;
if (!body)
continue;
/* get config lines, assign them */
retval = config_get_lines(body, &cl, 1);
if (retval < 0) {
err = SETOPT_ERR_PARSE;
goto err;
}
retval = config_assign(&options_format, newoptions, cl, 0, 0, msg);
config_free_lines(cl);
if (retval < 0) {
err = SETOPT_ERR_PARSE;
goto err;
}
if (i==0)
newdefaultoptions = options_dup(&options_format, newoptions);
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
}
Stop discarding command-line arguments when TestingTorNetwork is set Discovered by Kevin Bauer. Fixes bug 5373; bugfix on 0.2.3.9-alpha, where task 4552 added support for two layers of torrc files.
2012-03-14 07:40:04 +01:00
/* Assign command-line variables a second time too */
retval = config_assign(&options_format, newoptions,
global_cmdline_options, 0, 0, msg);
if (retval < 0) {
err = SETOPT_ERR_PARSE;
goto err;
}
Patch from Karsten to implement proposal 135 ("Testing Tor Networks"). svn:r15253
2008-06-14 18:01:29 +02:00
}
Make an options_act() where we do all the things a new options set needs. Still needs more work. svn:r2716
2004-11-09 05:28:18 +01:00
/* Validate newoptions */
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
if (options_validate(oldoptions, newoptions, 0, msg) < 0) {
Downlgrade tweak, and answer lots of XXX021s. No actual code fixes in this patch. svn:r17686
2008-12-18 17:11:24 +01:00
err = SETOPT_ERR_PARSE; /*XXX make this a separate return value.*/
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
goto err;
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
}
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
if (options_transition_allowed(oldoptions, newoptions, msg) < 0) {
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
err = SETOPT_ERR_TRANSITION;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
goto err;
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
}
break out validate_options leave options.FirewallPorts NULL if it's NULL svn:r2673
2004-11-04 11:23:30 +01:00
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
if (set_options(newoptions, msg)) {
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
err = SETOPT_ERR_SETTING;
Add new config.c function to set options that can fail, and roll back if they do. This should solve the setconf-an-impossible-port bug. svn:r5046
2005-09-14 04:36:29 +02:00
goto err; /* frees and replaces old options */
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
}
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
config_free(&options_format, global_default_options);
global_default_options = newdefaultoptions;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
r18787@catbus: nickm | 2008-03-13 11:11:52 -0400 Make set-option functions return sensible error codes from an enum, not mysterious negative integers svn:r14004
2008-03-13 16:11:56 +01:00
return SETOPT_OK;
options_init_from_torrc(): split off options_init_from_string() svn:r13946
2008-03-10 13:41:44 +01:00
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
err:
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
config_free(&options_format, newoptions);
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
config_free(&options_format, newdefaultoptions);
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
if (*msg) {
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
char *old_msg = *msg;
tor_asprintf(msg, "Failed to parse/validate config: %s", old_msg);
tor_free(old_msg);
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
}
different exit codes for options_init_from_string() Change options_init_from_string() so that it returns different exit codes in the error case, depending on what went wrong. Also push the responsibility to log the error to the caller. svn:r13947
2008-03-10 13:41:49 +01:00
return err;
Added getoptions() and made minor adjustment to poptReadDefaultOptions() svn:r25
2002-07-03 18:31:22 +02:00
}
Add a config-file GETINFO entry; fix a minor memory leak on some SAVECONF calls. svn:r4761
2005-08-10 20:05:20 +02:00
/** Return the location for our configuration file.
*/
const char *
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
get_torrc_fname(int defaults_fname)
Add a config-file GETINFO entry; fix a minor memory leak on some SAVECONF calls. svn:r4761
2005-08-10 20:05:20 +02:00
{
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
const char *fname = defaults_fname ? torrc_defaults_fname : torrc_fname;
if (fname)
return fname;
Add a config-file GETINFO entry; fix a minor memory leak on some SAVECONF calls. svn:r4761
2005-08-10 20:05:20 +02:00
else
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
return get_default_conf_file(defaults_fname);
Add a config-file GETINFO entry; fix a minor memory leak on some SAVECONF calls. svn:r4761
2005-08-10 20:05:20 +02:00
}
Spell-check Tor.
2009-05-27 23:55:51 +02:00
/** Adjust the address map based on the MapAddress elements in the
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
* configuration <b>options</b>
*/
bug1666 - Pass-through support for SOCKS5 authentication If a SOCKS5 client insists on authentication, allow it to negotiate a connection with Tor's SOCKS server successfully. Any credentials the client provides are ignored. This allows Tor to work with SOCKS5 clients that can only support 'authenticated' connections. Also add a bunch of basic unit tests for SOCKS4/4a/5 support in buffers.c.
2010-08-03 23:28:55 +02:00
void
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
config_register_addressmaps(const or_options_t *options)
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
{
Add a new AddressMap directive to rewrite incoming socks addresses. Add a new TrackHostExits directive to trigger addressmaps for certain incoming socks addresses, for sites that break when your exit keeps changing. Redo the client-side dns cache so it's just an addressmap too. svn:r3641
2005-02-22 01:53:08 +01:00
smartlist_t *elts;
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *opt;
Add a new AddressMap directive to rewrite incoming socks addresses. Add a new TrackHostExits directive to trigger addressmaps for certain incoming socks addresses, for sites that break when your exit keeps changing. Redo the client-side dns cache so it's just an addressmap too. svn:r3641
2005-02-22 01:53:08 +01:00
char *from, *to;
Get address map resetting implemented. svn:r3745
2005-03-11 22:39:39 +01:00
addressmap_clear_configured();
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
elts = smartlist_new();
Add a new AddressMap directive to rewrite incoming socks addresses. Add a new TrackHostExits directive to trigger addressmaps for certain incoming socks addresses, for sites that break when your exit keeps changing. Redo the client-side dns cache so it's just an addressmap too. svn:r3641
2005-02-22 01:53:08 +01:00
for (opt = options->AddressMap; opt; opt = opt->next) {
Refactor addressmap_match_superdomains and representation of wildcards In this new representation for wildcarded addresses, there are no longer any 'magic addresses': rather, "a.b c.d", "*.a.b c.d" and "*.a.b *.c.d" are all represented by a mapping from "a.b" to "c.d". we now distinguish them by setting bits in the addressmap_entry_t structure, where src_wildcard is set if the source address had a wildcard, and dst_wildcard is set if the target address had a wildcard. This lets the case where "*.a.b *.c.d" or "*.a.b c.d" remap the address "a.b" get handled trivially, and lets us simplify and improve the addressmap_match_superdomains implementation: we can now have it run in O(parts of address) rather than O(entries in addressmap).
2011-09-08 17:54:24 +02:00
int from_wildcard = 0, to_wildcard = 0;
Add a new AddressMap directive to rewrite incoming socks addresses. Add a new TrackHostExits directive to trigger addressmaps for certain incoming socks addresses, for sites that break when your exit keeps changing. Redo the client-side dns cache so it's just an addressmap too. svn:r3641
2005-02-22 01:53:08 +01:00
smartlist_split_string(elts, opt->value, NULL,
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, 2);
Address nickm's comments at https://trac.torproject.org/projects/tor/ticket/933#comment:8 1. Only allow '*.' in MapAddress expressions. Ignore '*ample.com' and '.example.com'. This has resulted in a slight refactoring of config_register_addressmaps. 2. Add some more detail to the man page entry for AddressMap. 3. Fix initialization of a pointer to NULL rather than 0. 4. Update the unit tests to cater for the changes in 1 and test more explicitly for recursive mapping.
2010-12-27 18:35:16 +01:00
if (smartlist_len(elts) < 2) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG,"MapAddress '%s' has too few arguments. Ignoring.",
opt->value);
Address nickm's comments at https://trac.torproject.org/projects/tor/ticket/933#comment:8 1. Only allow '*.' in MapAddress expressions. Ignore '*ample.com' and '.example.com'. This has resulted in a slight refactoring of config_register_addressmaps. 2. Add some more detail to the man page entry for AddressMap. 3. Fix initialization of a pointer to NULL rather than 0. 4. Update the unit tests to cater for the changes in 1 and test more explicitly for recursive mapping.
2010-12-27 18:35:16 +01:00
goto cleanup;
Add a new AddressMap directive to rewrite incoming socks addresses. Add a new TrackHostExits directive to trigger addressmaps for certain incoming socks addresses, for sites that break when your exit keeps changing. Redo the client-side dns cache so it's just an addressmap too. svn:r3641
2005-02-22 01:53:08 +01:00
}
Address nickm's comments at https://trac.torproject.org/projects/tor/ticket/933#comment:8 1. Only allow '*.' in MapAddress expressions. Ignore '*ample.com' and '.example.com'. This has resulted in a slight refactoring of config_register_addressmaps. 2. Add some more detail to the man page entry for AddressMap. 3. Fix initialization of a pointer to NULL rather than 0. 4. Update the unit tests to cater for the changes in 1 and test more explicitly for recursive mapping.
2010-12-27 18:35:16 +01:00
from = smartlist_get(elts,0);
to = smartlist_get(elts,1);
if (to[0] == '.' || from[0] == '.') {
log_warn(LD_CONFIG,"MapAddress '%s' is ambiguous - address starts with a"
"'.'. Ignoring.",opt->value);
goto cleanup;
}
Forbid remapping of * It might be nice to support this someday, but for now it would fail with an infinite remap cycle. (If I say "remap * *.foo.exit", then example.com -> example.com.foo.exit -> example.com.foo.exit.foo.exit -> example.com.foo.exit.foo.exit.foo.exit -> ...)
2011-09-08 18:04:34 +02:00
if (!strcmp(to, "*") || !strcmp(from, "*")) {
log_warn(LD_CONFIG,"MapAddress '%s' is unsupported - can't remap from "
"or to *. Ignoring.",opt->value);
goto cleanup;
}
Refactor addressmap_match_superdomains and representation of wildcards In this new representation for wildcarded addresses, there are no longer any 'magic addresses': rather, "a.b c.d", "*.a.b c.d" and "*.a.b *.c.d" are all represented by a mapping from "a.b" to "c.d". we now distinguish them by setting bits in the addressmap_entry_t structure, where src_wildcard is set if the source address had a wildcard, and dst_wildcard is set if the target address had a wildcard. This lets the case where "*.a.b *.c.d" or "*.a.b c.d" remap the address "a.b" get handled trivially, and lets us simplify and improve the addressmap_match_superdomains implementation: we can now have it run in O(parts of address) rather than O(entries in addressmap).
2011-09-08 17:54:24 +02:00
/* Detect asterisks in expressions of type: '*.example.com' */
if (!strncmp(from,"*.",2)) {
from += 2;
from_wildcard = 1;
Add a new AddressMap directive to rewrite incoming socks addresses. Add a new TrackHostExits directive to trigger addressmaps for certain incoming socks addresses, for sites that break when your exit keeps changing. Redo the client-side dns cache so it's just an addressmap too. svn:r3641
2005-02-22 01:53:08 +01:00
}
Refactor addressmap_match_superdomains and representation of wildcards In this new representation for wildcarded addresses, there are no longer any 'magic addresses': rather, "a.b c.d", "*.a.b c.d" and "*.a.b *.c.d" are all represented by a mapping from "a.b" to "c.d". we now distinguish them by setting bits in the addressmap_entry_t structure, where src_wildcard is set if the source address had a wildcard, and dst_wildcard is set if the target address had a wildcard. This lets the case where "*.a.b *.c.d" or "*.a.b c.d" remap the address "a.b" get handled trivially, and lets us simplify and improve the addressmap_match_superdomains implementation: we can now have it run in O(parts of address) rather than O(entries in addressmap).
2011-09-08 17:54:24 +02:00
if (!strncmp(to,"*.",2)) {
to += 2;
to_wildcard = 1;
}
Address nickm's comments at https://trac.torproject.org/projects/tor/ticket/933#comment:8 1. Only allow '*.' in MapAddress expressions. Ignore '*ample.com' and '.example.com'. This has resulted in a slight refactoring of config_register_addressmaps. 2. Add some more detail to the man page entry for AddressMap. 3. Fix initialization of a pointer to NULL rather than 0. 4. Update the unit tests to cater for the changes in 1 and test more explicitly for recursive mapping.
2010-12-27 18:35:16 +01:00
Refactor addressmap_match_superdomains and representation of wildcards In this new representation for wildcarded addresses, there are no longer any 'magic addresses': rather, "a.b c.d", "*.a.b c.d" and "*.a.b *.c.d" are all represented by a mapping from "a.b" to "c.d". we now distinguish them by setting bits in the addressmap_entry_t structure, where src_wildcard is set if the source address had a wildcard, and dst_wildcard is set if the target address had a wildcard. This lets the case where "*.a.b *.c.d" or "*.a.b c.d" remap the address "a.b" get handled trivially, and lets us simplify and improve the addressmap_match_superdomains implementation: we can now have it run in O(parts of address) rather than O(entries in addressmap).
2011-09-08 17:54:24 +02:00
if (to_wildcard && !from_wildcard) {
Address nickm's comments at https://trac.torproject.org/projects/tor/ticket/933#comment:8 1. Only allow '*.' in MapAddress expressions. Ignore '*ample.com' and '.example.com'. This has resulted in a slight refactoring of config_register_addressmaps. 2. Add some more detail to the man page entry for AddressMap. 3. Fix initialization of a pointer to NULL rather than 0. 4. Update the unit tests to cater for the changes in 1 and test more explicitly for recursive mapping.
2010-12-27 18:35:16 +01:00
log_warn(LD_CONFIG,
"Skipping invalid argument '%s' to MapAddress: "
"can only use wildcard (i.e. '*.') if 'from' address "
Refactor addressmap_match_superdomains and representation of wildcards In this new representation for wildcarded addresses, there are no longer any 'magic addresses': rather, "a.b c.d", "*.a.b c.d" and "*.a.b *.c.d" are all represented by a mapping from "a.b" to "c.d". we now distinguish them by setting bits in the addressmap_entry_t structure, where src_wildcard is set if the source address had a wildcard, and dst_wildcard is set if the target address had a wildcard. This lets the case where "*.a.b *.c.d" or "*.a.b c.d" remap the address "a.b" get handled trivially, and lets us simplify and improve the addressmap_match_superdomains implementation: we can now have it run in O(parts of address) rather than O(entries in addressmap).
2011-09-08 17:54:24 +02:00
"uses wildcard also", opt->value);
Address nickm's comments at https://trac.torproject.org/projects/tor/ticket/933#comment:8 1. Only allow '*.' in MapAddress expressions. Ignore '*ample.com' and '.example.com'. This has resulted in a slight refactoring of config_register_addressmaps. 2. Add some more detail to the man page entry for AddressMap. 3. Fix initialization of a pointer to NULL rather than 0. 4. Update the unit tests to cater for the changes in 1 and test more explicitly for recursive mapping.
2010-12-27 18:35:16 +01:00
goto cleanup;
}
if (address_is_invalid_destination(to, 1)) {
log_warn(LD_CONFIG,
Refactor addressmap_match_superdomains and representation of wildcards In this new representation for wildcarded addresses, there are no longer any 'magic addresses': rather, "a.b c.d", "*.a.b c.d" and "*.a.b *.c.d" are all represented by a mapping from "a.b" to "c.d". we now distinguish them by setting bits in the addressmap_entry_t structure, where src_wildcard is set if the source address had a wildcard, and dst_wildcard is set if the target address had a wildcard. This lets the case where "*.a.b *.c.d" or "*.a.b c.d" remap the address "a.b" get handled trivially, and lets us simplify and improve the addressmap_match_superdomains implementation: we can now have it run in O(parts of address) rather than O(entries in addressmap).
2011-09-08 17:54:24 +02:00
"Skipping invalid argument '%s' to MapAddress", opt->value);
Address nickm's comments at https://trac.torproject.org/projects/tor/ticket/933#comment:8 1. Only allow '*.' in MapAddress expressions. Ignore '*ample.com' and '.example.com'. This has resulted in a slight refactoring of config_register_addressmaps. 2. Add some more detail to the man page entry for AddressMap. 3. Fix initialization of a pointer to NULL rather than 0. 4. Update the unit tests to cater for the changes in 1 and test more explicitly for recursive mapping.
2010-12-27 18:35:16 +01:00
goto cleanup;
}
Refactor addressmap_match_superdomains and representation of wildcards In this new representation for wildcarded addresses, there are no longer any 'magic addresses': rather, "a.b c.d", "*.a.b c.d" and "*.a.b *.c.d" are all represented by a mapping from "a.b" to "c.d". we now distinguish them by setting bits in the addressmap_entry_t structure, where src_wildcard is set if the source address had a wildcard, and dst_wildcard is set if the target address had a wildcard. This lets the case where "*.a.b *.c.d" or "*.a.b c.d" remap the address "a.b" get handled trivially, and lets us simplify and improve the addressmap_match_superdomains implementation: we can now have it run in O(parts of address) rather than O(entries in addressmap).
2011-09-08 17:54:24 +02:00
addressmap_register(from, tor_strdup(to), 0, ADDRMAPSRC_TORRC,
from_wildcard, to_wildcard);
Address nickm's comments at https://trac.torproject.org/projects/tor/ticket/933#comment:8 1. Only allow '*.' in MapAddress expressions. Ignore '*ample.com' and '.example.com'. This has resulted in a slight refactoring of config_register_addressmaps. 2. Add some more detail to the man page entry for AddressMap. 3. Fix initialization of a pointer to NULL rather than 0. 4. Update the unit tests to cater for the changes in 1 and test more explicitly for recursive mapping.
2010-12-27 18:35:16 +01:00
if (smartlist_len(elts) > 2)
log_warn(LD_CONFIG,"Ignoring extra arguments to MapAddress.");
cleanup:
Add a new AddressMap directive to rewrite incoming socks addresses. Add a new TrackHostExits directive to trigger addressmaps for certain incoming socks addresses, for sites that break when your exit keeps changing. Redo the client-side dns cache so it's just an addressmap too. svn:r3641
2005-02-22 01:53:08 +01:00
SMARTLIST_FOREACH(elts, char*, cp, tor_free(cp));
smartlist_clear(elts);
}
smartlist_free(elts);
}
Allow multiple logfiles at different severity ranges svn:r1899
2004-05-19 22:07:08 +02:00
/**
* Initialize the logs based on the configuration file.
*/
r15512@catbus: nickm | 2007-10-02 16:27:43 -0400 Make some functions static; remove some dead code. svn:r11750
2007-10-02 22:35:23 +02:00
static int
Rename cross-format config.c stuff to config_* and or_options_t-specific stuff to option[s]_* svn:r4641
2005-07-23 03:58:05 +02:00
options_init_logs(or_options_t *options, int validate_only)
Allow multiple logfiles at different severity ranges svn:r1899
2004-05-19 22:07:08 +02:00
{
Be consistent about preferring foo* to struct foo* svn:r4637
2005-07-22 23:12:10 +02:00
config_line_t *opt;
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
int ok;
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
smartlist_t *elts;
ignore RunAsDaemon more thoroughly when we're running on windows. svn:r6471
2006-05-23 09:04:55 +02:00
int daemon =
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifdef _WIN32
ignore RunAsDaemon more thoroughly when we're running on windows. svn:r6471
2006-05-23 09:04:55 +02:00
0;
#else
options->RunAsDaemon;
#endif
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
Fix log granularity based on Nick's comments. Instead of rejecting a value that doesn't divide into 1 second, round to the nearest divisor of 1 second and warn. Document that the option only controls the granularity written by Tor to a file or console log. It does not (for example) "batch up" log messages to affect times logged by a controller, times attached to syslog messages, or the mtime fields on log files.
2010-11-08 16:37:20 +01:00
if (options->LogTimeGranularity <= 0) {
log_warn(LD_CONFIG, "Log time granularity '%d' has to be positive.",
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
options->LogTimeGranularity);
return -1;
Fix log granularity based on Nick's comments. Instead of rejecting a value that doesn't divide into 1 second, round to the nearest divisor of 1 second and warn. Document that the option only controls the granularity written by Tor to a file or console log. It does not (for example) "batch up" log messages to affect times logged by a controller, times attached to syslog messages, or the mtime fields on log files.
2010-11-08 16:37:20 +01:00
} else if (1000 % options->LogTimeGranularity != 0 &&
options->LogTimeGranularity % 1000 != 0) {
int granularity = options->LogTimeGranularity;
if (granularity < 40) {
do granularity++;
while (1000 % granularity != 0);
} else if (granularity < 1000) {
granularity = 1000 / granularity;
while (1000 % granularity != 0)
granularity--;
granularity = 1000 / granularity;
} else {
granularity = 1000 * ((granularity / 1000) + 1);
}
log_warn(LD_CONFIG, "Log time granularity '%d' has to be either a "
"divisor or a multiple of 1 second. Changing to "
"'%d'.",
options->LogTimeGranularity, granularity);
Make LogTimeGranularity respect validate_only
2010-11-08 18:40:26 +01:00
if (!validate_only)
set_log_time_granularity(granularity);
Fix log granularity based on Nick's comments. Instead of rejecting a value that doesn't divide into 1 second, round to the nearest divisor of 1 second and warn. Document that the option only controls the granularity written by Tor to a file or console log. It does not (for example) "batch up" log messages to affect times logged by a controller, times attached to syslog messages, or the mtime fields on log files.
2010-11-08 16:37:20 +01:00
} else {
Make LogTimeGranularity respect validate_only
2010-11-08 18:40:26 +01:00
if (!validate_only)
set_log_time_granularity(options->LogTimeGranularity);
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
}
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
ok = 1;
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
elts = smartlist_new();
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
for (opt = options->Logs; opt; opt = opt->next) {
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
log_severity_list_t *severity;
const char *cfg = opt->value;
severity = tor_malloc_zero(sizeof(log_severity_list_t));
if (parse_log_severity_config(&cfg, severity) < 0) {
log_warn(LD_CONFIG, "Couldn't parse log levels in Log option 'Log %s'",
opt->value);
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
ok = 0; goto cleanup;
}
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
smartlist_split_string(elts, cfg, NULL,
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, 2);
if (smartlist_len(elts) == 0)
smartlist_add(elts, tor_strdup("stdout"));
if (smartlist_len(elts) == 1 &&
(!strcasecmp(smartlist_get(elts,0), "stdout") ||
!strcasecmp(smartlist_get(elts,0), "stderr"))) {
int err = smartlist_len(elts) &&
!strcasecmp(smartlist_get(elts,0), "stderr");
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
if (!validate_only) {
if (daemon) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG,
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
"Can't log to %s with RunAsDaemon set; skipping stdout",
err?"stderr":"stdout");
Keep running if we ask for an impossible stdout log; just warn. svn:r5964
2006-02-11 22:56:03 +01:00
} else {
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
add_stream_log(severity, err?"<stderr>":"<stdout>",
Change logging code to use fds instead of stdio. Fixes bug 861, and probably makes logging slightly faster. Not a backport candidate: bug 861 is too obscure and harmless. svn:r17456
2008-12-03 00:36:58 +01:00
fileno(err?stderr:stdout));
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
}
}
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
goto cleanup;
}
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
if (smartlist_len(elts) == 1 &&
!strcasecmp(smartlist_get(elts,0), "syslog")) {
#ifdef HAVE_SYSLOG_H
if (!validate_only) {
add_syslog_log(severity);
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
}
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
#else
log_warn(LD_CONFIG, "Syslog is not supported on this system. Sorry.");
#endif
goto cleanup;
}
if (smartlist_len(elts) == 2 &&
!strcasecmp(smartlist_get(elts,0), "file")) {
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
if (!validate_only) {
Make the Log configuration option expand ~
2011-04-22 16:06:52 +02:00
char *fname = expand_filename(smartlist_get(elts, 1));
if (add_file_log(severity, fname) < 0) {
when add_file_log() fails, tell us why. svn:r14277
2008-04-01 23:05:31 +02:00
log_warn(LD_CONFIG, "Couldn't open file for 'Log %s': %s",
opt->value, strerror(errno));
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
ok = 0;
add missing warning svn:r5806
2006-01-11 21:09:37 +01:00
}
Make the Log configuration option expand ~
2011-04-22 16:06:52 +02:00
tor_free(fname);
Fix bug 230: add a rollback function to reverse all changes since the last mark_logs_temp(), and move log initialization into the two-phase part of option setting. svn:r5803
2006-01-11 20:40:14 +01:00
}
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
goto cleanup;
}
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
log_warn(LD_CONFIG, "Bad syntax on file Log option 'Log %s'",
opt->value);
ok = 0; goto cleanup;
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
cleanup:
SMARTLIST_FOREACH(elts, char*, cp, tor_free(cp));
smartlist_clear(elts);
r18630@catbus: nickm | 2008-03-05 17:31:33 -0500 Implement domain-selection for logging. Source is documented; needs documentation in manpage (maybe). For now, see doxygen comment on parse_log_severity_config in log.c svn:r13875
2008-03-05 23:31:39 +01:00
tor_free(severity);
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
}
smartlist_free(elts);
Add a torrc option to report log domains
2011-01-25 21:53:15 +01:00
if (ok && !validate_only)
logs_set_domain_logging(options->LogMessageDomains);
Use new log format; convert old format to new format. svn:r2684
2004-11-05 06:50:35 +01:00
return ok?0:-1;
}
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
/** Read the contents of a Bridge line from <b>line</b>. Return 0
* if the line is well-formed, and -1 if it isn't. If
* <b>validate_only</b> is 0, and the line is well-formed, then add
* the bridge described in the line to our internal bridge list. */
static int
parse_bridge_line(const char *line, int validate_only)
{
smartlist_t *items = NULL;
int r;
r17641@31-33-44: nickm | 2008-08-05 16:07:53 -0400 Initial conversion of uint32_t addr to tor_addr_t addr in connection_t and related types. Most of the Tor wire formats using these new types are in, but the code to generate and use it is not. This is a big patch. Let me know what it breaks for you. svn:r16435
2008-08-05 22:08:19 +02:00
char *addrport=NULL, *fingerprint=NULL;
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
char *transport_name=NULL;
char *field1=NULL;
r17641@31-33-44: nickm | 2008-08-05 16:07:53 -0400 Initial conversion of uint32_t addr to tor_addr_t addr in connection_t and related types. Most of the Tor wire formats using these new types are in, but the code to generate and use it is not. This is a big patch. Let me know what it breaks for you. svn:r16435
2008-08-05 22:08:19 +02:00
tor_addr_t addr;
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
uint16_t port = 0;
char digest[DIGEST_LEN];
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
items = smartlist_new();
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
smartlist_split_string(items, line, NULL,
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, -1);
if (smartlist_len(items) < 1) {
log_warn(LD_CONFIG, "Too few arguments to Bridge line.");
goto err;
}
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
/* field1 is either a transport name or addrport */
field1 = smartlist_get(items, 0);
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
smartlist_del_keeporder(items, 0);
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
Future-proof and user-proof parse_bridge_line
2011-07-03 05:38:00 +02:00
if (!(strstr(field1, ".") || strstr(field1, ":"))) {
/* new-style bridge line */
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
transport_name = field1;
Future-proof and user-proof parse_bridge_line
2011-07-03 05:38:00 +02:00
if (smartlist_len(items) < 1) {
log_warn(LD_CONFIG, "Too few items to Bridge line.");
goto err;
}
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
addrport = smartlist_get(items, 0);
smartlist_del_keeporder(items, 0);
Style and grammar tweaks on 2841 branch
2011-07-03 05:23:07 +02:00
} else {
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
addrport = field1;
Style and grammar tweaks on 2841 branch
2011-07-03 05:23:07 +02:00
}
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(addrport, &addr, &port)<0) {
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
log_warn(LD_CONFIG, "Error parsing Bridge address '%s'", addrport);
goto err;
}
if (!port) {
If the bridge config line doesn't specify a port, assume 443. This makes bridge lines a bit smaller and easier for users to understand. Also, remove a duplicate changelog entry from the past. svn:r19260
2009-04-11 02:16:05 +02:00
log_info(LD_CONFIG,
"Bridge address '%s' has no port; using default port 443.",
addrport);
port = 443;
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
}
if (smartlist_len(items)) {
fingerprint = smartlist_join_strings(items, "", 0, NULL);
if (strlen(fingerprint) != HEX_DIGEST_LEN) {
log_warn(LD_CONFIG, "Key digest for Bridge is wrong length.");
goto err;
}
if (base16_decode(digest, DIGEST_LEN, fingerprint, HEX_DIGEST_LEN)<0) {
log_warn(LD_CONFIG, "Unable to decode Bridge key digest.");
goto err;
}
}
if (!validate_only) {
Style and grammar tweaks on 2841 branch
2011-07-03 05:23:07 +02:00
log_debug(LD_DIR, "Bridge at %s:%d (transport: %s) (%s)",
fmt_addr(&addr), (int)port,
transport_name ? transport_name : "no transport",
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
fingerprint ? fingerprint : "no key listed");
Style and grammar tweaks on 2841 branch
2011-07-03 05:23:07 +02:00
bridge_add_from_config(&addr, port,
fingerprint ? digest : NULL, transport_name);
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
}
r = 0;
goto done;
Style and grammar tweaks on 2841 branch
2011-07-03 05:23:07 +02:00
err:
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
r = -1;
Style and grammar tweaks on 2841 branch
2011-07-03 05:23:07 +02:00
done:
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
SMARTLIST_FOREACH(items, char*, s, tor_free(s));
smartlist_free(items);
tor_free(addrport);
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
tor_free(transport_name);
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
tor_free(fingerprint);
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
return r;
}
Various trivial changes. * Improved function documentation. * Renamed find_bridge_transport_by_addrport() to find_transport_by_bridge_addrport(). * Sanitized log severities we use. * Ran check-spaces.
2011-06-12 16:41:32 +02:00
/** Read the contents of a ClientTransportPlugin line from
* <b>line</b>. Return 0 if the line is well-formed, and -1 if it
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
* isn't.
*
* If <b>validate_only</b> is 0, and the line is well-formed:
* - If it's an external proxy line, add the transport described in the line to
* our internal transport list.
* - If it's a managed proxy line, launch the managed proxy. */
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
static int
Various trivial changes. * Improved function documentation. * Renamed find_bridge_transport_by_addrport() to find_transport_by_bridge_addrport(). * Sanitized log severities we use. * Ran check-spaces.
2011-06-12 16:41:32 +02:00
parse_client_transport_line(const char *line, int validate_only)
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
{
smartlist_t *items = NULL;
int r;
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
char *field2=NULL;
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
const char *transports=NULL;
smartlist_t *transport_list=NULL;
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
char *addrport=NULL;
tor_addr_t addr;
Various trivial changes. * Improved function documentation. * Renamed find_bridge_transport_by_addrport() to find_transport_by_bridge_addrport(). * Sanitized log severities we use. * Ran check-spaces.
2011-06-12 16:41:32 +02:00
uint16_t port = 0;
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
int socks_ver=PROXY_NONE;
/* managed proxy options */
int is_managed=0;
char **proxy_argv=NULL;
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
char **tmp=NULL;
int proxy_argc,i;
int line_length;
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
items = smartlist_new();
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
smartlist_split_string(items, line, NULL,
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, -1);
Various trivial changes. * Improved function documentation. * Renamed find_bridge_transport_by_addrport() to find_transport_by_bridge_addrport(). * Sanitized log severities we use. * Ran check-spaces.
2011-06-12 16:41:32 +02:00
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
line_length = smartlist_len(items);
if (line_length < 3) {
Trivial code tweaks and documentation updates.
2011-06-28 05:43:40 +02:00
log_warn(LD_CONFIG, "Too few arguments on ClientTransportPlugin line.");
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
goto err;
}
Various trivial changes. * Improved function documentation. * Renamed find_bridge_transport_by_addrport() to find_transport_by_bridge_addrport(). * Sanitized log severities we use. * Ran check-spaces.
2011-06-12 16:41:32 +02:00
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
/* Get the first line element, split it to commas into
transport_list (in case it's multiple transports) and validate
the transport names. */
transports = smartlist_get(items, 0);
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
transport_list = smartlist_new();
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
smartlist_split_string(transport_list, transports, ",",
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, 0);
SMARTLIST_FOREACH_BEGIN(transport_list, const char *, transport_name) {
if (!string_is_C_identifier(transport_name)) {
log_warn(LD_CONFIG, "Transport name is not a C identifier (%s).",
transport_name);
goto err;
}
} SMARTLIST_FOREACH_END(transport_name);
Various trivial changes. * Improved function documentation. * Renamed find_bridge_transport_by_addrport() to find_transport_by_bridge_addrport(). * Sanitized log severities we use. * Ran check-spaces.
2011-06-12 16:41:32 +02:00
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
/* field2 is either a SOCKS version or "exec" */
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
field2 = smartlist_get(items, 1);
Various trivial changes. * Improved function documentation. * Renamed find_bridge_transport_by_addrport() to find_transport_by_bridge_addrport(). * Sanitized log severities we use. * Ran check-spaces.
2011-06-12 16:41:32 +02:00
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
if (!strcmp(field2,"socks4")) {
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
socks_ver = PROXY_SOCKS4;
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
} else if (!strcmp(field2,"socks5")) {
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
socks_ver = PROXY_SOCKS5;
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
} else if (!strcmp(field2,"exec")) {
is_managed=1;
} else {
log_warn(LD_CONFIG, "Strange ClientTransportPlugin field '%s'.",
field2);
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
goto err;
}
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
if (is_managed) { /* managed */
if (!validate_only) { /* if we are not just validating, use the
rest of the line as the argv of the proxy
to be launched */
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
proxy_argc = line_length-2;
tor_assert(proxy_argc > 0);
proxy_argv = tor_malloc_zero(sizeof(char*)*(proxy_argc+1));
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
tmp = proxy_argv;
Spawn multiple protocols using a single managed proxy. If multiple torrc transport lines have the same argv, tor instructs a single managed proxy to launch multiple protocols.
2011-08-12 21:33:05 +02:00
for (i=0;i<proxy_argc;i++) { /* store arguments */
*tmp++ = smartlist_get(items, 2);
smartlist_del_keeporder(items, 2);
}
Fix comments: There is no such thing as a NUL pointer
2012-05-07 16:57:59 +02:00
*tmp = NULL; /*terminated with NULL, just like execve() likes it*/
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
Spawn multiple protocols using a single managed proxy. If multiple torrc transport lines have the same argv, tor instructs a single managed proxy to launch multiple protocols.
2011-08-12 21:33:05 +02:00
/* kickstart the thing */
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
pt_kickstart_client_proxy(transport_list, proxy_argv);
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
}
} else { /* external */
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
if (smartlist_len(transport_list) != 1) {
log_warn(LD_CONFIG, "You can't have an external proxy with "
"more than one transports.");
goto err;
}
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
addrport = smartlist_get(items, 2);
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(addrport, &addr, &port)<0) {
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
log_warn(LD_CONFIG, "Error parsing transport "
"address '%s'", addrport);
goto err;
}
if (!port) {
log_warn(LD_CONFIG,
"Transport address '%s' has no port.", addrport);
goto err;
}
if (!validate_only) {
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
transport_add_from_config(&addr, port, smartlist_get(transport_list, 0),
socks_ver);
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
log_info(LD_DIR, "Transport '%s' found at %s:%d",
transports, fmt_addr(&addr), (int)port);
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
}
}
r = 0;
goto done;
err:
r = -1;
done:
SMARTLIST_FOREACH(items, char*, s, tor_free(s));
smartlist_free(items);
Fix a couple of pluggable transport bugs Fix coverity complaints 490, 491 and 492. Especially the one in parse_client_transport_line() could've been a remotely triggerable segfault, I think.
2011-10-26 16:49:24 +02:00
if (transport_list) {
SMARTLIST_FOREACH(transport_list, char*, s, tor_free(s));
smartlist_free(transport_list);
}
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
return r;
}
/** Read the contents of a ServerTransportPlugin line from
* <b>line</b>. Return 0 if the line is well-formed, and -1 if it
* isn't.
* If <b>validate_only</b> is 0, the line is well-formed, and it's a
* managed proxy line, launch the managed proxy. */
static int
parse_server_transport_line(const char *line, int validate_only)
{
smartlist_t *items = NULL;
int r;
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
const char *transports=NULL;
smartlist_t *transport_list=NULL;
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
char *type=NULL;
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
char *addrport=NULL;
tor_addr_t addr;
uint16_t port = 0;
/* managed proxy options */
int is_managed=0;
char **proxy_argv=NULL;
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
char **tmp=NULL;
int proxy_argc,i;
int line_length;
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
items = smartlist_new();
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
smartlist_split_string(items, line, NULL,
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, -1);
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
line_length = smartlist_len(items);
if (line_length < 3) {
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
log_warn(LD_CONFIG, "Too few arguments on ServerTransportPlugin line.");
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
goto err;
}
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
/* Get the first line element, split it to commas into
transport_list (in case it's multiple transports) and validate
the transport names. */
transports = smartlist_get(items, 0);
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
transport_list = smartlist_new();
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
smartlist_split_string(transport_list, transports, ",",
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, 0);
SMARTLIST_FOREACH_BEGIN(transport_list, const char *, transport_name) {
if (!string_is_C_identifier(transport_name)) {
log_warn(LD_CONFIG, "Transport name is not a C identifier (%s).",
transport_name);
goto err;
}
} SMARTLIST_FOREACH_END(transport_name);
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
type = smartlist_get(items, 1);
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
if (!strcmp(type, "exec")) {
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
is_managed=1;
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
} else if (!strcmp(type, "proxy")) {
is_managed=0;
} else {
log_warn(LD_CONFIG, "Strange ServerTransportPlugin type '%s'", type);
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
goto err;
}
Various trivial changes. * Improved function documentation. * Renamed find_bridge_transport_by_addrport() to find_transport_by_bridge_addrport(). * Sanitized log severities we use. * Ran check-spaces.
2011-06-12 16:41:32 +02:00
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
if (is_managed) { /* managed */
if (!validate_only) {
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
proxy_argc = line_length-2;
tor_assert(proxy_argc > 0);
proxy_argv = tor_malloc_zero(sizeof(char*)*(proxy_argc+1));
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
tmp = proxy_argv;
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
Spawn multiple protocols using a single managed proxy. If multiple torrc transport lines have the same argv, tor instructs a single managed proxy to launch multiple protocols.
2011-08-12 21:33:05 +02:00
for (i=0;i<proxy_argc;i++) { /* store arguments */
*tmp++ = smartlist_get(items, 2);
smartlist_del_keeporder(items, 2);
}
Fix comments: There is no such thing as a NUL pointer
2012-05-07 16:57:59 +02:00
*tmp = NULL; /*terminated with NULL, just like execve() likes it*/
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
Spawn multiple protocols using a single managed proxy. If multiple torrc transport lines have the same argv, tor instructs a single managed proxy to launch multiple protocols.
2011-08-12 21:33:05 +02:00
/* kickstart the thing */
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
pt_kickstart_server_proxy(transport_list, proxy_argv);
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
}
} else { /* external */
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
if (smartlist_len(transport_list) != 1) {
log_warn(LD_CONFIG, "You can't have an external proxy with "
"more than one transports.");
goto err;
}
Fixes small bugs.
2011-06-14 16:00:55 +02:00
Let's be smarter while parsing {Client,Server}TransportPlugin lines.
2011-07-18 16:42:31 +02:00
addrport = smartlist_get(items, 2);
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(addrport, &addr, &port)<0) {
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
log_warn(LD_CONFIG, "Error parsing transport "
"address '%s'", addrport);
Fixes small bugs.
2011-06-14 16:00:55 +02:00
goto err;
Changed a couple of 180 spec stuff according to #3578. * Restored "proxy" in external ServerTransportPlugin lines. * Changed the extended OR port and ORPort env. vars to addr:port.
2011-07-14 01:03:35 +02:00
}
if (!port) {
log_warn(LD_CONFIG,
"Transport address '%s' has no port.", addrport);
goto err;
}
if (!validate_only) {
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
log_info(LD_DIR, "Server transport '%s' at %s:%d.",
transports, fmt_addr(&addr), (int)port);
Add support for managed {Client,Server}TransportPlugin parsing.
2011-07-13 18:58:11 +02:00
}
ClientTransportPlugin parsing done.
2011-06-11 17:08:31 +02:00
}
r = 0;
goto done;
err:
r = -1;
done:
SMARTLIST_FOREACH(items, char*, s, tor_free(s));
smartlist_free(items);
Fix a NULL pointer dereference in parse_server_transport_line().
2011-10-26 17:13:56 +02:00
if (transport_list) {
SMARTLIST_FOREACH(transport_list, char*, s, tor_free(s));
smartlist_free(transport_list);
}
Support multiple transports in a single transport line. Support multiple comma-separated transpotrs in a single {Client,Server}TransportPlugin line.
2011-10-07 14:13:41 +02:00
Interim commit: new config options Bridge and UseBridges. It is becoming increasingly clear to me that bridges should be a special case of entry guards, not a whole separate pile of nearly identical functions. svn:r10141
2007-05-08 13:28:05 +02:00
return r;
}
Don't trigger an assert if we start a directory authority with a private IP address (like 127.0.0.1). svn:r13371
2008-02-04 17:58:50 +01:00
/** Read the contents of a DirServer line from <b>line</b>. If
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
* <b>validate_only</b> is 0, and the line is well-formed, and it
* shares any bits with <b>required_type</b> or <b>required_type</b>
* is 0, then add the dirserver described in the line (minus whatever
Don't trigger an assert if we start a directory authority with a private IP address (like 127.0.0.1). svn:r13371
2008-02-04 17:58:50 +01:00
* bits it's missing) as a valid authority. Return 0 on success,
* or -1 if the line isn't well-formed or if we can't add it. */
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
static int
Automated rename from authority_type_t to dirinfo_type_t We were already overloading this type to mean "a directory that can serve us X" in addition to "a directory that is an authority for X."
2010-11-08 20:27:36 +01:00
parse_dir_server_line(const char *line, dirinfo_type_t required_type,
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
int validate_only)
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
{
smartlist_t *items = NULL;
r8957@totoro: nickm | 2006-10-08 22:35:17 -0400 The otherwise regrettable MIPSpro C compiler warns about values set but never used, and about mixing enums and ints; these are good warnings, and so should be fixed. This removes some dead code and some potential bugs. Thanks to pnx. svn:r8664
2006-10-09 04:35:51 +02:00
int r;
fix minor memory leak in config svn:r5318
2005-10-25 20:06:29 +02:00
char *addrport=NULL, *address=NULL, *nickname=NULL, *fingerprint=NULL;
r11676@Kushana: nickm | 2006-12-23 20:42:17 -0500 Add an orport option to dirserver lines so that clients can tell where to connect to open an encrypted tunnel to a dirserver even before they have its descriptor. svn:r9171
2006-12-24 03:45:27 +01:00
uint16_t dir_port = 0, or_port = 0;
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
char digest[DIGEST_LEN];
r12898@catbus: nickm | 2007-05-22 13:11:04 -0400 More v3 directory code: have authorities load certificates; have everybody store certificates to disk and load them; provide a way to configure v3 authorities. svn:r10293
2007-05-22 19:58:25 +02:00
char v3_digest[DIGEST_LEN];
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
dirinfo_type_t type = V2_DIRINFO;
r12697@catbus: nickm | 2007-05-09 00:15:40 -0400 Change authority_type_t to a set of flags; use it more consistently. svn:r10144
2007-05-09 06:15:46 +02:00
int is_not_hidserv_authority = 0, is_not_v2_authority = 0;
More stuff for new directories. - Distinguish v1 authorities (all currently trusted directories) from v2 authorities (all trusted directories). - Add configuration option for which dirs are v1 authories. - Add configuration option for whether to be a v1 authority. - Make trusted dirserver selection functions take options to choose which functionality we need. - Remove option when getting directory cache to see whether they support running-routers; they all do now. Replace it with one to see whether caches support v2 stuff. - Parse, cache, and serve network-status objects properly. - Serve compressed groups of router descriptors. The compression logic here could be more memory-efficient. - svn:r4911
2005-09-07 18:42:53 +02:00
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
items = smartlist_new();
link nicknames to trusted directory servers; log these nicknames when mentioning servers; also, when we get a naming conflict; log which nicknames or keys are supposed to bind. svn:r5184
2005-10-04 23:21:09 +02:00
smartlist_split_string(items, line, NULL,
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, -1);
r9023@Kushana: nickm | 2006-09-29 17:27:24 -0400 Make distinct all non-bug messages at notice or higher that appear 3 or more times. svn:r8541
2006-09-30 00:33:34 +02:00
if (smartlist_len(items) < 1) {
log_warn(LD_CONFIG, "No arguments on DirServer line.");
Efficiency hack: call tor_fix_source_file late, not early. Add "BUG" domain. Domains are now bitmasks... just in case. Make some err msgs non-general. svn:r5309
2005-10-25 09:05:03 +02:00
goto err;
link nicknames to trusted directory servers; log these nicknames when mentioning servers; also, when we get a naming conflict; log which nicknames or keys are supposed to bind. svn:r5184
2005-10-04 23:21:09 +02:00
}
More stuff for new directories. - Distinguish v1 authorities (all currently trusted directories) from v2 authorities (all trusted directories). - Add configuration option for which dirs are v1 authories. - Add configuration option for whether to be a v1 authority. - Make trusted dirserver selection functions take options to choose which functionality we need. - Remove option when getting directory cache to see whether they support running-routers; they all do now. Replace it with one to see whether caches support v2 stuff. - Parse, cache, and serve network-status objects properly. - Serve compressed groups of router descriptors. The compression logic here could be more memory-efficient. - svn:r4911
2005-09-07 18:42:53 +02:00
link nicknames to trusted directory servers; log these nicknames when mentioning servers; also, when we get a naming conflict; log which nicknames or keys are supposed to bind. svn:r5184
2005-10-04 23:21:09 +02:00
if (is_legal_nickname(smartlist_get(items, 0))) {
nickname = smartlist_get(items, 0);
smartlist_del_keeporder(items, 0);
}
r8977@Kushana: nickm | 2006-09-28 19:56:41 -0400 Make "is a v1 authority", "is a v2 authority", and "is a hidden service authority" into separate flags so we can eventually migrate more trust away from moria. svn:r8523
2006-09-29 01:57:59 +02:00
while (smartlist_len(items)) {
char *flag = smartlist_get(items, 0);
if (TOR_ISDIGIT(flag[0]))
break;
if (!strcasecmp(flag, "v1")) {
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
type |= (V1_DIRINFO | HIDSERV_DIRINFO);
r8977@Kushana: nickm | 2006-09-28 19:56:41 -0400 Make "is a v1 authority", "is a v2 authority", and "is a hidden service authority" into separate flags so we can eventually migrate more trust away from moria. svn:r8523
2006-09-29 01:57:59 +02:00
} else if (!strcasecmp(flag, "hs")) {
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
type |= HIDSERV_DIRINFO;
r8977@Kushana: nickm | 2006-09-28 19:56:41 -0400 Make "is a v1 authority", "is a v2 authority", and "is a hidden service authority" into separate flags so we can eventually migrate more trust away from moria. svn:r8523
2006-09-29 01:57:59 +02:00
} else if (!strcasecmp(flag, "no-hs")) {
is_not_hidserv_authority = 1;
add a 'bridge' flag for dirserver config entries svn:r10128
2007-05-07 10:26:50 +02:00
} else if (!strcasecmp(flag, "bridge")) {
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
type |= BRIDGE_DIRINFO;
r8977@Kushana: nickm | 2006-09-28 19:56:41 -0400 Make "is a v1 authority", "is a v2 authority", and "is a hidden service authority" into separate flags so we can eventually migrate more trust away from moria. svn:r8523
2006-09-29 01:57:59 +02:00
} else if (!strcasecmp(flag, "no-v2")) {
r12697@catbus: nickm | 2007-05-09 00:15:40 -0400 Change authority_type_t to a set of flags; use it more consistently. svn:r10144
2007-05-09 06:15:46 +02:00
is_not_v2_authority = 1;
r11676@Kushana: nickm | 2006-12-23 20:42:17 -0500 Add an orport option to dirserver lines so that clients can tell where to connect to open an encrypted tunnel to a dirserver even before they have its descriptor. svn:r9171
2006-12-24 03:45:27 +01:00
} else if (!strcasecmpstart(flag, "orport=")) {
int ok;
when we added orport= to the dirserver line, we started freeing the string in the middle. this was probably xiando's crash bug. svn:r9265
2007-01-05 02:23:34 +01:00
char *portstring = flag + strlen("orport=");
or_port = (uint16_t) tor_parse_long(portstring, 10, 1, 65535, &ok, NULL);
r11676@Kushana: nickm | 2006-12-23 20:42:17 -0500 Add an orport option to dirserver lines so that clients can tell where to connect to open an encrypted tunnel to a dirserver even before they have its descriptor. svn:r9171
2006-12-24 03:45:27 +01:00
if (!ok)
when we added orport= to the dirserver line, we started freeing the string in the middle. this was probably xiando's crash bug. svn:r9265
2007-01-05 02:23:34 +01:00
log_warn(LD_CONFIG, "Invalid orport '%s' on DirServer line.",
portstring);
r12898@catbus: nickm | 2007-05-22 13:11:04 -0400 More v3 directory code: have authorities load certificates; have everybody store certificates to disk and load them; provide a way to configure v3 authorities. svn:r10293
2007-05-22 19:58:25 +02:00
} else if (!strcasecmpstart(flag, "v3ident=")) {
char *idstr = flag + strlen("v3ident=");
if (strlen(idstr) != HEX_DIGEST_LEN ||
base16_decode(v3_digest, DIGEST_LEN, idstr, HEX_DIGEST_LEN)<0) {
log_warn(LD_CONFIG, "Bad v3 identity digest '%s' on DirServer line",
flag);
} else {
Code to make clients fetch and use microdescriptors for circuit building To turn this on, set UseMicrodescriptors to "1" (or "auto" if you want it on-if-you're-a-client). It should go auto-by-default once 0.2.3.1-alpha is released. Because of our node logic, directory caches will never use microdescriptors when they have the right routerinfo available.
2010-11-08 20:21:32 +01:00
type |= V3_DIRINFO|EXTRAINFO_DIRINFO|MICRODESC_DIRINFO;
r12898@catbus: nickm | 2007-05-22 13:11:04 -0400 More v3 directory code: have authorities load certificates; have everybody store certificates to disk and load them; provide a way to configure v3 authorities. svn:r10293
2007-05-22 19:58:25 +02:00
}
r8977@Kushana: nickm | 2006-09-28 19:56:41 -0400 Make "is a v1 authority", "is a v2 authority", and "is a hidden service authority" into separate flags so we can eventually migrate more trust away from moria. svn:r8523
2006-09-29 01:57:59 +02:00
} else {
log_warn(LD_CONFIG, "Unrecognized flag '%s' on DirServer line",
flag);
}
tor_free(flag);
link nicknames to trusted directory servers; log these nicknames when mentioning servers; also, when we get a naming conflict; log which nicknames or keys are supposed to bind. svn:r5184
2005-10-04 23:21:09 +02:00
smartlist_del_keeporder(items, 0);
More stuff for new directories. - Distinguish v1 authorities (all currently trusted directories) from v2 authorities (all trusted directories). - Add configuration option for which dirs are v1 authories. - Add configuration option for whether to be a v1 authority. - Make trusted dirserver selection functions take options to choose which functionality we need. - Remove option when getting directory cache to see whether they support running-routers; they all do now. Replace it with one to see whether caches support v2 stuff. - Parse, cache, and serve network-status objects properly. - Serve compressed groups of router descriptors. The compression logic here could be more memory-efficient. - svn:r4911
2005-09-07 18:42:53 +02:00
}
r8977@Kushana: nickm | 2006-09-28 19:56:41 -0400 Make "is a v1 authority", "is a v2 authority", and "is a hidden service authority" into separate flags so we can eventually migrate more trust away from moria. svn:r8523
2006-09-29 01:57:59 +02:00
if (is_not_hidserv_authority)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
type &= ~HIDSERV_DIRINFO;
r12697@catbus: nickm | 2007-05-09 00:15:40 -0400 Change authority_type_t to a set of flags; use it more consistently. svn:r10144
2007-05-09 06:15:46 +02:00
if (is_not_v2_authority)
Replace _AUTHORITY enum values with _DIRINFO values (automted)
2010-11-08 20:35:02 +01:00
type &= ~V2_DIRINFO;
r8977@Kushana: nickm | 2006-09-28 19:56:41 -0400 Make "is a v1 authority", "is a v2 authority", and "is a hidden service authority" into separate flags so we can eventually migrate more trust away from moria. svn:r8523
2006-09-29 01:57:59 +02:00
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
if (smartlist_len(items) < 2) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG, "Too few arguments to DirServer line.");
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
goto err;
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
}
addrport = smartlist_get(items, 0);
Avoid a double-free when parsing malformed DirServer lines. svn:r9264
2007-01-05 02:12:10 +01:00
smartlist_del_keeporder(items, 0);
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (addr_port_lookup(LOG_WARN, addrport, &address, NULL, &dir_port)<0) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG, "Error parsing DirServer address '%s'", addrport);
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
goto err;
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
}
r11676@Kushana: nickm | 2006-12-23 20:42:17 -0500 Add an orport option to dirserver lines so that clients can tell where to connect to open an encrypted tunnel to a dirserver even before they have its descriptor. svn:r9171
2006-12-24 03:45:27 +01:00
if (!dir_port) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG, "Missing port in DirServer address '%s'",addrport);
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
goto err;
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
}
link nicknames to trusted directory servers; log these nicknames when mentioning servers; also, when we get a naming conflict; log which nicknames or keys are supposed to bind. svn:r5184
2005-10-04 23:21:09 +02:00
fingerprint = smartlist_join_strings(items, "", 0, NULL);
if (strlen(fingerprint) != HEX_DIGEST_LEN) {
Be clearer on the various roles for auth dir types. Bridge authorities no longer write bridge descriptors to their cached-routers file -- this gets complex because of extrainfo documents. svn:r10545
2007-06-09 09:05:19 +02:00
log_warn(LD_CONFIG, "Key digest for DirServer is wrong length %d.",
(int)strlen(fingerprint));
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
goto err;
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
}
backport candidate: Refuse to start with certain directory authority keys, and encourage people using them to stop. svn:r11171
2007-08-19 04:51:54 +02:00
if (!strcmp(fingerprint, "E623F7625FBE0C87820F11EC5F6D5377ED816294")) {
put a note so the distant future developers can simplify tor svn:r12691
2007-12-06 12:19:00 +01:00
/* a known bad fingerprint. refuse to use it. We can remove this
* clause once Tor 0.1.2.17 is obsolete. */
backport candidate: Refuse to start with certain directory authority keys, and encourage people using them to stop. svn:r11171
2007-08-19 04:51:54 +02:00
log_warn(LD_CONFIG, "Dangerous dirserver line. To correct, erase your "
"torrc file (%s), or reinstall Tor and use the default torrc.",
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
get_torrc_fname(0));
backport candidate: Refuse to start with certain directory authority keys, and encourage people using them to stop. svn:r11171
2007-08-19 04:51:54 +02:00
goto err;
}
link nicknames to trusted directory servers; log these nicknames when mentioning servers; also, when we get a naming conflict; log which nicknames or keys are supposed to bind. svn:r5184
2005-10-04 23:21:09 +02:00
if (base16_decode(digest, DIGEST_LEN, fingerprint, HEX_DIGEST_LEN)<0) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG, "Unable to decode DirServer key digest.");
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
goto err;
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
}
Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities, rather than the all-or-nothing replacement that DirServer offers. svn:r12777
2007-12-12 05:38:54 +01:00
if (!validate_only && (!required_type || required_type & type)) {
if (required_type)
type &= required_type; /* pare down what we think of them as an
* authority for. */
log_debug(LD_DIR, "Trusted %d dirserver at %s:%d (%s)", (int)type,
address, (int)dir_port, (char*)smartlist_get(items,0));
Don't trigger an assert if we start a directory authority with a private IP address (like 127.0.0.1). svn:r13371
2008-02-04 17:58:50 +01:00
if (!add_trusted_dir_server(nickname, address, dir_port, or_port,
digest, v3_digest, type))
goto err;
Separate validate from activate. Document undocumented stuff in config.c. svn:r2711
2004-11-09 00:12:40 +01:00
}
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
r = 0;
goto done;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
err:
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
r = -1;
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
done:
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
SMARTLIST_FOREACH(items, char*, s, tor_free(s));
smartlist_free(items);
fix minor memory leak in config svn:r5318
2005-10-25 20:06:29 +02:00
tor_free(addrport);
Make options no longer a global variable. Now we can try setting an option but back out if it fails to parse, or if it's disallowed (e.g. changing RunAsDaemon from 1 to 0). Use parse_line_from_str rather than parse_line_from_file. svn:r2692
2004-11-06 06:18:11 +01:00
tor_free(address);
link nicknames to trusted directory servers; log these nicknames when mentioning servers; also, when we get a naming conflict; log which nicknames or keys are supposed to bind. svn:r5184
2005-10-04 23:21:09 +02:00
tor_free(nickname);
tor_free(fingerprint);
Add an implementation of trusted_dir_server logic to switch towards simply remembering an addr/port/keyid for each trusted dir server svn:r2443
2004-10-12 17:55:20 +02:00
return r;
}
(Unused) backend logic for stream isolation This patch adds fields to track how streams should be isolated, and ensures that those fields are set correctly. It also adds fields to track what streams can go on a circuit, and adds functions to see whether a streams can go on a circuit and update the circuit accordingly. Those functions aren't yet called.
2011-07-06 22:03:47 +02:00
/** Free all storage held in <b>port</b> */
static void
port_cfg_free(port_cfg_t *port)
{
tor_free(port);
}
Fix a bogus warning
2011-09-14 09:05:03 +02:00
/** Warn for every port in <b>ports</b> that is on a publicly routable
* address. */
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
static void
warn_nonlocal_client_ports(const smartlist_t *ports, const char *portname)
{
SMARTLIST_FOREACH_BEGIN(ports, const port_cfg_t *, port) {
Check whether a client port is a Unix socket before using its IP addr Bugfix on commit c1ac0695d5bc64b555c345e4be87b18bab3ae56b, not yet in any release. Fixes bug 4091; bug reported by SwissTorHelp.
2011-10-20 12:17:23 +02:00
if (port->is_unix_addr) {
/* Unix sockets aren't accessible over a network. */
} else if (!tor_addr_is_internal(&port->addr, 1)) {
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
log_warn(LD_CONFIG, "You specified a public address for %sPort. "
"Other people on the Internet might find your computer and "
"use it as an open proxy. Please don't allow this unless you "
"have a good reason.", portname);
reinstate a notice for the non-loopback socksport case Thanks to prop171, it's no longer a crazy thing to do, but you should make sure that you really meant it!
2011-10-07 22:34:21 +02:00
} else if (!tor_addr_is_loopback(&port->addr)) {
Fix a bunch of whitespace errors
2011-10-11 17:30:01 +02:00
log_notice(LD_CONFIG, "You configured a non-loopback address for "
"%sPort. This allows everybody on your local network to use "
"your machine as a proxy. Make sure this is what you wanted.",
reinstate a notice for the non-loopback socksport case Thanks to prop171, it's no longer a crazy thing to do, but you should make sure that you really meant it!
2011-10-07 22:34:21 +02:00
portname);
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
}
} SMARTLIST_FOREACH_END(port);
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
/** DOCDOC */
static void
warn_nonlocal_controller_ports(smartlist_t *ports, unsigned forbid)
{
int warned = 0;
SMARTLIST_FOREACH_BEGIN(ports, port_cfg_t *, port) {
if (port->type != CONN_TYPE_CONTROL_LISTENER)
continue;
if (port->is_unix_addr)
continue;
if (!tor_addr_is_loopback(&port->addr)) {
if (forbid) {
if (!warned)
log_warn(LD_CONFIG,
"You have a ControlPort set to accept "
"unauthenticated connections from a non-local address. "
"This means that programs not running on your computer "
"can reconfigure your Tor, without even having to guess a "
"password. That's so bad that I'm closing your ControlPort "
"for you. If you need to control your Tor remotely, try "
"enabling authentication and using a tool like stunnel or "
"ssh to encrypt remote access.");
warned = 1;
port_cfg_free(port);
SMARTLIST_DEL_CURRENT(ports, port);
} else {
log_warn(LD_CONFIG, "You have a ControlPort set to accept "
"connections from a non-local address. This means that "
"programs not running on your computer can reconfigure your "
"Tor. That's pretty bad, since the controller "
"protocol isn't encrypted! Maybe you should just listen on "
"127.0.0.1 and use a tool like stunnel or ssh to encrypt "
"remote connections to your control port.");
return; /* No point in checking the rest */
}
}
} SMARTLIST_FOREACH_END(port);
}
#define CL_PORT_NO_OPTIONS (1u<<0)
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
#define CL_PORT_WARN_NONLOCAL (1u<<1)
#define CL_PORT_ALLOW_EXTRA_LISTENADDR (1u<<2)
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
#define CL_PORT_SERVER_OPTIONS (1u<<3)
#define CL_PORT_FORBID_NONLOCAL (1u<<4)
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
/**
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
* Parse port configuration for a single port type.
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
*
* Read entries of the "FooPort" type from the list <b>ports</b>, and
* entries of the "FooListenAddress" type from the list
* <b>listenaddrs</b>. Two syntaxes are supported: a legacy syntax
* where FooPort is at most a single entry containing a port number and
* where FooListenAddress has any number of address:port combinations;
* and a new syntax where there are no FooListenAddress entries and
* where FooPort can have any number of entries of the format
* "[Address:][Port] IsolationOptions".
*
* In log messages, describe the port type as <b>portname</b>.
*
* If no address is specified, default to <b>defaultaddr</b>. If no
* FooPort is given, default to defaultport (if 0, there is no default).
*
* If CL_PORT_NO_OPTIONS is set in <b>flags</b>, do not allow stream
* isolation options in the FooPort entries.
*
* If CL_PORT_WARN_NONLOCAL is set in <b>flags</b>, warn if any of the
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
* ports are not on a local address. If CL_PORT_FORBID_NONLOCAL is set,
* this is a contrl port with no password set: don't even allow it.
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
*
* Unless CL_PORT_ALLOW_EXTRA_LISTENADDR is set in <b>flags</b>, warn
* if FooListenAddress is set but FooPort is 0.
*
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
* If CL_PORT_SERVER_OPTIONS is set in <b>flags</b>, do not allow stream
* isolation options in the FooPort entries; instead allow the
* server-port option set.
*
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
* On success, if <b>out</b> is given, add a new port_cfg_t entry to
* <b>out</b> for every port that the client should listen on. Return 0
* on success, -1 on failure.
*/
static int
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
parse_port_config(smartlist_t *out,
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
const config_line_t *ports,
const config_line_t *listenaddrs,
const char *portname,
int listener_type,
const char *defaultaddr,
int defaultport,
unsigned flags)
{
smartlist_t *elts;
int retval = -1;
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
const unsigned is_control = (listener_type == CONN_TYPE_CONTROL_LISTENER);
const unsigned allow_no_options = flags & CL_PORT_NO_OPTIONS;
const unsigned use_server_options = flags & CL_PORT_SERVER_OPTIONS;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
const unsigned warn_nonlocal = flags & CL_PORT_WARN_NONLOCAL;
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
const unsigned forbid_nonlocal = flags & CL_PORT_FORBID_NONLOCAL;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
const unsigned allow_spurious_listenaddr =
flags & CL_PORT_ALLOW_EXTRA_LISTENADDR;
/* FooListenAddress is deprecated; let's make it work like it used to work,
* though. */
if (listenaddrs) {
fix whitespace (two-space indent)
2011-09-11 07:33:04 +02:00
int mainport = defaultport;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
fix whitespace (two-space indent)
2011-09-11 07:33:04 +02:00
if (ports && ports->next) {
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
log_warn(LD_CONFIG, "%sListenAddress can't be used when there are "
"multiple %sPort lines", portname, portname);
return -1;
fix whitespace (two-space indent)
2011-09-11 07:33:04 +02:00
} else if (ports) {
if (!strcmp(ports->value, "auto")) {
mainport = CFG_AUTO_PORT;
} else {
int ok;
mainport = (int)tor_parse_long(ports->value, 10, 0, 65535, &ok, NULL);
if (!ok) {
log_warn(LD_CONFIG, "%sListenAddress can only be used with a single "
"%sPort with value \"auto\" or 1-65535.",
portname, portname);
return -1;
}
}
}
if (mainport == 0) {
if (allow_spurious_listenaddr)
return 1;
log_warn(LD_CONFIG, "%sPort must be defined if %sListenAddress is used",
portname, portname);
return -1;
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (use_server_options && out) {
/* Add a no_listen port. */
port_cfg_t *cfg = tor_malloc_zero(sizeof(port_cfg_t));
cfg->type = listener_type;
cfg->port = mainport;
tor_addr_make_unspec(&cfg->addr); /* Server ports default to 0.0.0.0 */
cfg->no_listen = 1;
cfg->ipv4_only = 1;
smartlist_add(out, cfg);
}
fix whitespace (two-space indent)
2011-09-11 07:33:04 +02:00
for (; listenaddrs; listenaddrs = listenaddrs->next) {
tor_addr_t addr;
uint16_t port = 0;
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(listenaddrs->value, &addr, &port) < 0) {
fix whitespace (two-space indent)
2011-09-11 07:33:04 +02:00
log_warn(LD_CONFIG, "Unable to parse %sListenAddress '%s'",
portname, listenaddrs->value);
return -1;
}
if (out) {
port_cfg_t *cfg = tor_malloc_zero(sizeof(port_cfg_t));
cfg->type = listener_type;
cfg->port = port ? port : mainport;
tor_addr_copy(&cfg->addr, &addr);
cfg->session_group = SESSION_GROUP_UNSET;
cfg->isolation_flags = ISO_DEFAULT;
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
cfg->no_advertise = 1;
fix whitespace (two-space indent)
2011-09-11 07:33:04 +02:00
smartlist_add(out, cfg);
}
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (warn_nonlocal && out) {
if (is_control)
warn_nonlocal_controller_ports(out, forbid_nonlocal);
else
warn_nonlocal_client_ports(out, portname);
}
fix whitespace (two-space indent)
2011-09-11 07:33:04 +02:00
return 0;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
} /* end if (listenaddrs) */
/* No ListenAddress lines. If there's no FooPort, then maybe make a default
* one. */
if (! ports) {
if (defaultport && out) {
port_cfg_t *cfg = tor_malloc_zero(sizeof(port_cfg_t));
cfg->type = listener_type;
cfg->port = defaultport;
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
tor_addr_parse(&cfg->addr, defaultaddr);
Implement sensible isolation for tunneled directory conns One-hop dirconn streams all share a session group, and get the ISO_SESSIONGRP flag: they may share circuits with each other and nothing else. Anonymized dirconn streams get a new internal-use-only ISO_STREAM flag: they may not share circuits with anything, including each other.
2011-07-08 21:54:30 +02:00
cfg->session_group = SESSION_GROUP_UNSET;
(Unused) backend logic for stream isolation This patch adds fields to track how streams should be isolated, and ensures that those fields are set correctly. It also adds fields to track what streams can go on a circuit, and adds functions to see whether a streams can go on a circuit and update the circuit accordingly. Those functions aren't yet called.
2011-07-06 22:03:47 +02:00
cfg->isolation_flags = ISO_DEFAULT;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
smartlist_add(out, cfg);
}
return 0;
}
/* At last we can actually parse the FooPort lines. The syntax is:
* [Addr:](Port|auto) [Options].*/
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
elts = smartlist_new();
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
for (; ports; ports = ports->next) {
tor_addr_t addr;
int port;
Implement sensible isolation for tunneled directory conns One-hop dirconn streams all share a session group, and get the ISO_SESSIONGRP flag: they may share circuits with each other and nothing else. Anonymized dirconn streams get a new internal-use-only ISO_STREAM flag: they may not share circuits with anything, including each other.
2011-07-08 21:54:30 +02:00
int sessiongroup = SESSION_GROUP_UNSET;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
unsigned isolation = ISO_DEFAULT;
char *addrport;
uint16_t ptmp=0;
int ok;
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
int no_listen = 0, no_advertise = 0, all_addrs = 0,
ipv4_only = 0, ipv6_only = 0;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
smartlist_split_string(elts, ports->value, NULL,
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, 0);
if (smartlist_len(elts) == 0) {
log_warn(LD_CONFIG, "Invalid %sPort line with no value", portname);
goto err;
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (allow_no_options && smartlist_len(elts) > 1) {
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
log_warn(LD_CONFIG, "Too many options on %sPort line", portname);
goto err;
}
/* Now parse the addr/port value */
addrport = smartlist_get(elts, 0);
if (!strcmp(addrport, "auto")) {
port = CFG_AUTO_PORT;
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
tor_addr_parse(&addr, defaultaddr);
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
} else if (!strcasecmpend(addrport, ":auto")) {
char *addrtmp = tor_strndup(addrport, strlen(addrport)-5);
port = CFG_AUTO_PORT;
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(addrtmp, &addr, &ptmp)<0 || ptmp) {
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
log_warn(LD_CONFIG, "Invalid address '%s' for %sPort",
escaped(addrport), portname);
tor_free(addrtmp);
goto err;
}
} else {
Check for port config before addr config in client port options Otherwise, we'll fail, since "9050" looks like a perfectly fine address.
2011-07-20 19:40:09 +02:00
/* Try parsing integer port before address, because, who knows?
"9050" might be a valid address. */
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
port = (int) tor_parse_long(addrport, 10, 0, 65535, &ok, NULL);
Check for port config before addr config in client port options Otherwise, we'll fail, since "9050" looks like a perfectly fine address.
2011-07-20 19:40:09 +02:00
if (ok) {
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
tor_addr_parse(&addr, defaultaddr);
} else if (tor_addr_port_lookup(addrport, &addr, &ptmp) == 0) {
Check for port config before addr config in client port options Otherwise, we'll fail, since "9050" looks like a perfectly fine address.
2011-07-20 19:40:09 +02:00
if (ptmp == 0) {
log_warn(LD_CONFIG, "%sPort line has address but no port", portname);
goto err;
}
port = ptmp;
} else {
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
log_warn(LD_CONFIG, "Couldn't parse address '%s' for %sPort",
escaped(addrport), portname);
goto err;
}
}
/* Now parse the rest of the options, if any. */
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (use_server_options) {
/* This is a server port; parse advertising options */
SMARTLIST_FOREACH_BEGIN(elts, char *, elt) {
if (elt_sl_idx == 0)
continue; /* Skip addr:port */
if (!strcasecmp(elt, "NoAdvertise")) {
no_advertise = 1;
} else if (!strcasecmp(elt, "NoListen")) {
no_listen = 1;
#if 0
/* not implemented yet. */
} else if (!strcasecmp(elt, "AllAddrs")) {
all_addrs = 1;
#endif
} else if (!strcasecmp(elt, "IPv4Only")) {
ipv4_only = 1;
} else if (!strcasecmp(elt, "IPv6Only")) {
ipv6_only = 1;
} else {
log_warn(LD_CONFIG, "Unrecognized %sPort option '%s'",
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
portname, escaped(elt));
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
} SMARTLIST_FOREACH_END(elt);
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (no_advertise && no_listen) {
log_warn(LD_CONFIG, "Tried to set both NoListen and NoAdvertise "
"on %sPort line '%s'",
portname, escaped(ports->value));
goto err;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (ipv4_only && ipv6_only) {
log_warn(LD_CONFIG, "Tried to set both IPv4Only and IPv6Only "
"on %sPort line '%s'",
portname, escaped(ports->value));
goto err;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (ipv4_only && tor_addr_family(&addr) == AF_INET6) {
log_warn(LD_CONFIG, "Could not interpret %sPort address as IPv6",
portname);
goto err;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (ipv6_only && tor_addr_family(&addr) == AF_INET) {
log_warn(LD_CONFIG, "Could not interpret %sPort address as IPv4",
portname);
goto err;
}
} else {
/* This is a client port; parse isolation options */
SMARTLIST_FOREACH_BEGIN(elts, char *, elt) {
int no = 0, isoflag = 0;
const char *elt_orig = elt;
if (elt_sl_idx == 0)
continue; /* Skip addr:port */
if (!strcasecmpstart(elt, "SessionGroup=")) {
int group = (int)tor_parse_long(elt+strlen("SessionGroup="),
10, 0, INT_MAX, &ok, NULL);
if (!ok) {
log_warn(LD_CONFIG, "Invalid %sPort option '%s'",
portname, escaped(elt));
goto err;
}
if (sessiongroup >= 0) {
log_warn(LD_CONFIG, "Multiple SessionGroup options on %sPort",
portname);
goto err;
}
sessiongroup = group;
continue;
}
if (!strcasecmpstart(elt, "No")) {
no = 1;
elt += 2;
}
if (!strcasecmpend(elt, "s"))
elt[strlen(elt)-1] = '\0'; /* kill plurals. */
if (!strcasecmp(elt, "IsolateDestPort")) {
isoflag = ISO_DESTPORT;
} else if (!strcasecmp(elt, "IsolateDestAddr")) {
isoflag = ISO_DESTADDR;
} else if (!strcasecmp(elt, "IsolateSOCKSAuth")) {
isoflag = ISO_SOCKSAUTH;
} else if (!strcasecmp(elt, "IsolateClientProtocol")) {
isoflag = ISO_CLIENTPROTO;
} else if (!strcasecmp(elt, "IsolateClientAddr")) {
isoflag = ISO_CLIENTADDR;
} else {
log_warn(LD_CONFIG, "Unrecognized %sPort option '%s'",
portname, escaped(elt_orig));
}
if (no) {
isolation &= ~isoflag;
} else {
isolation |= isoflag;
}
} SMARTLIST_FOREACH_END(elt);
}
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
Make FooPort 0 correctly disable the FooPort. Since the prop171 stuff, it had instead made Tor bind port 0, and re-bind it differently all the time. Resolves bug3704; not in any released version.
2011-08-09 20:52:22 +02:00
if (out && port) {
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
port_cfg_t *cfg = tor_malloc_zero(sizeof(port_cfg_t));
tor_addr_copy(&cfg->addr, &addr);
Reorder initialisation of port_cfg to match order of members in struct.
2012-03-19 04:57:19 +01:00
cfg->port = port;
cfg->type = listener_type;
(Unused) backend logic for stream isolation This patch adds fields to track how streams should be isolated, and ensures that those fields are set correctly. It also adds fields to track what streams can go on a circuit, and adds functions to see whether a streams can go on a circuit and update the circuit accordingly. Those functions aren't yet called.
2011-07-06 22:03:47 +02:00
cfg->isolation_flags = isolation;
Reorder initialisation of port_cfg to match order of members in struct.
2012-03-19 04:57:19 +01:00
cfg->session_group = sessiongroup;
Fix cut'n'paste bug (#5151).
2012-03-19 04:55:17 +01:00
cfg->no_advertise = no_advertise;
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
cfg->no_listen = no_listen;
cfg->all_addrs = all_addrs;
cfg->ipv4_only = ipv4_only;
cfg->ipv6_only = ipv6_only;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
smartlist_add(out, cfg);
}
SMARTLIST_FOREACH(elts, char *, cp, tor_free(cp));
smartlist_clear(elts);
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (warn_nonlocal && out) {
if (is_control)
warn_nonlocal_controller_ports(out, forbid_nonlocal);
else
warn_nonlocal_client_ports(out, portname);
}
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
retval = 0;
err:
SMARTLIST_FOREACH(elts, char *, cp, tor_free(cp));
smartlist_free(elts);
return retval;
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
/** DOCDOC */
static int
parse_socket_config(smartlist_t *out, const config_line_t *cfg,
int listener_type)
{
if (!out)
return 0;
for ( ; cfg; cfg = cfg->next) {
size_t len = strlen(cfg->value);
port_cfg_t *port = tor_malloc_zero(sizeof(port_cfg_t) + len + 1);
port->is_unix_addr = 1;
memcpy(port->unix_addr, cfg->value, len+1);
port->type = listener_type;
smartlist_add(out, port);
}
return 0;
}
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
/** Parse all client port types (Socks, DNS, Trans, NATD) from
* <b>options</b>. On success, set *<b>n_ports_out</b> to the number of
* ports that are listed and return 0. On failure, set *<b>msg</b> to a
* description of the problem and return -1.
*
* If <b>validate_only</b> is false, set configured_client_ports to the
* new list of ports parsed from <b>options</b>.
**/
static int
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
parse_ports(const or_options_t *options, int validate_only,
char **msg, int *n_ports_out)
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
{
smartlist_t *ports;
int retval = -1;
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
ports = smartlist_new();
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
*n_ports_out = 0;
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (parse_port_config(ports,
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
options->SocksPort, options->SocksListenAddress,
"Socks", CONN_TYPE_AP_LISTENER,
"127.0.0.1", 9050,
CL_PORT_WARN_NONLOCAL|CL_PORT_ALLOW_EXTRA_LISTENADDR) < 0) {
*msg = tor_strdup("Invalid SocksPort/SocksListenAddress configuration");
goto err;
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (parse_port_config(ports,
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
options->DNSPort, options->DNSListenAddress,
"DNS", CONN_TYPE_AP_DNS_LISTENER,
"127.0.0.1", 0,
CL_PORT_WARN_NONLOCAL) < 0) {
*msg = tor_strdup("Invalid DNSPort/DNSListenAddress configuration");
goto err;
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (parse_port_config(ports,
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
options->TransPort, options->TransListenAddress,
"Trans", CONN_TYPE_AP_TRANS_LISTENER,
"127.0.0.1", 0,
CL_PORT_WARN_NONLOCAL) < 0) {
*msg = tor_strdup("Invalid TransPort/TransListenAddress configuration");
goto err;
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (parse_port_config(ports,
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
options->NATDPort, options->NATDListenAddress,
"NATD", CONN_TYPE_AP_NATD_LISTENER,
"127.0.0.1", 0,
CL_PORT_WARN_NONLOCAL) < 0) {
*msg = tor_strdup("Invalid NatdPort/NatdListenAddress configuration");
goto err;
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
{
unsigned control_port_flags = CL_PORT_NO_OPTIONS | CL_PORT_WARN_NONLOCAL;
const int any_passwords = (options->HashedControlPassword ||
options->HashedControlSessionPassword ||
options->CookieAuthentication);
if (! any_passwords)
control_port_flags |= CL_PORT_FORBID_NONLOCAL;
if (parse_port_config(ports,
options->ControlPort, options->ControlListenAddress,
"Control", CONN_TYPE_CONTROL_LISTENER,
"127.0.0.1", 0,
control_port_flags) < 0) {
*msg = tor_strdup("Invalid ControlPort/ControlListenAddress "
"configuration");
goto err;
}
if (parse_socket_config(ports,
options->ControlSocket,
CONN_TYPE_CONTROL_LISTENER) < 0) {
*msg = tor_strdup("Invalid ControlSocket configuration");
goto err;
}
}
if (! options->ClientOnly) {
if (parse_port_config(ports,
options->ORPort, options->ORListenAddress,
"OR", CONN_TYPE_OR_LISTENER,
"0.0.0.0", 0,
CL_PORT_SERVER_OPTIONS) < 0) {
*msg = tor_strdup("Invalid ORPort/ORListenAddress configuration");
goto err;
}
if (parse_port_config(ports,
options->DirPort, options->DirListenAddress,
"Dir", CONN_TYPE_DIR_LISTENER,
"0.0.0.0", 0,
CL_PORT_SERVER_OPTIONS) < 0) {
*msg = tor_strdup("Invalid DirPort/DirListenAddress configuration");
goto err;
}
}
if (check_server_ports(ports, options) < 0) {
*msg = tor_strdup("Misconfigured server ports");
goto err;
}
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
*n_ports_out = smartlist_len(ports);
if (!validate_only) {
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (configured_ports) {
SMARTLIST_FOREACH(configured_ports,
(Unused) backend logic for stream isolation This patch adds fields to track how streams should be isolated, and ensures that those fields are set correctly. It also adds fields to track what streams can go on a circuit, and adds functions to see whether a streams can go on a circuit and update the circuit accordingly. Those functions aren't yet called.
2011-07-06 22:03:47 +02:00
port_cfg_t *, p, port_cfg_free(p));
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
smartlist_free(configured_ports);
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
configured_ports = ports;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
ports = NULL; /* prevent free below. */
}
retval = 0;
err:
if (ports) {
(Unused) backend logic for stream isolation This patch adds fields to track how streams should be isolated, and ensures that those fields are set correctly. It also adds fields to track what streams can go on a circuit, and adds functions to see whether a streams can go on a circuit and update the circuit accordingly. Those functions aren't yet called.
2011-07-06 22:03:47 +02:00
SMARTLIST_FOREACH(ports, port_cfg_t *, p, port_cfg_free(p));
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
smartlist_free(ports);
}
return retval;
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
/** DOCDOC */
static int
check_server_ports(const smartlist_t *ports,
const or_options_t *options)
{
int n_orport_advertised = 0;
int n_orport_advertised_ipv4 = 0;
int n_orport_listeners = 0;
int n_dirport_advertised = 0;
int n_dirport_listeners = 0;
int n_low_port = 0;
int r = 0;
SMARTLIST_FOREACH_BEGIN(ports, const port_cfg_t *, port) {
if (port->type == CONN_TYPE_DIR_LISTENER) {
if (! port->no_advertise)
++n_dirport_advertised;
if (! port->no_listen)
++n_dirport_listeners;
} else if (port->type == CONN_TYPE_OR_LISTENER) {
if (! port->no_advertise) {
++n_orport_advertised;
if (tor_addr_family(&port->addr) == AF_INET ||
(tor_addr_family(&port->addr) == AF_UNSPEC && !port->ipv6_only))
++n_orport_advertised_ipv4;
}
if (! port->no_listen)
++n_orport_listeners;
} else {
continue;
}
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifndef _WIN32
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (!port->no_advertise && port->port < 1024)
++n_low_port;
#endif
} SMARTLIST_FOREACH_END(port);
if (n_orport_advertised && !n_orport_listeners) {
log_warn(LD_CONFIG, "We are advertising an ORPort, but not actually "
"listening on one.");
r = -1;
}
if (n_dirport_advertised && !n_dirport_listeners) {
log_warn(LD_CONFIG, "We are advertising a DirPort, but not actually "
"listening on one.");
r = -1;
}
if (n_dirport_advertised > 1) {
log_warn(LD_CONFIG, "Can't advertise more than one DirPort.");
r = -1;
}
if (n_orport_advertised && !n_orport_advertised_ipv4 &&
!options->BridgeRelay) {
log_warn(LD_CONFIG, "Configured non-bridge only to listen on an IPv6 "
"address.");
r = -1;
}
if (n_low_port && options->AccountingMax) {
log(LOG_WARN, LD_CONFIG,
"You have set AccountingMax to use hibernation. You have also "
"chosen a low DirPort or OrPort. This combination can make Tor stop "
"working when it tries to re-attach the port after a period of "
"hibernation. Please choose a different port or turn off "
"hibernation unless you know this combination will work on your "
"platform.");
}
return r;
}
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
/** Return a list of port_cfg_t for client ports parsed from the
* options. */
const smartlist_t *
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
get_configured_ports(void)
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
{
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
if (!configured_ports)
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
configured_ports = smartlist_new();
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
return configured_ports;
}
Pass OR address to PT proxy, even with IPv6 or ORListenAddress. Introduce get_first_listener_addrport_for_pt() which returns a string containing the addrport of the first listener we could find. Use it to form the TOR_PT_ORPORT managed proxy protocol line.
2012-03-31 14:04:58 +02:00
/** Return an <address>:<port> string representation of the address
* where the first <b>listener_type</b> listener waits for
* connections. Return NULL if we couldn't find a listener. The
* string is allocated on the heap and it's the responsibility of the
* caller to free it after use.
*
* This function is meant to be used by the pluggable transport proxy
Fix issues found by nickm. * Document fmt_addr_impl() and friends. * Parenthesize macro arguments. * Rename get_first_listener_addrport_for_pt() to get_first_listener_addrport_string(). * Handle port_cfg_t with no_listen. * Handle failure of router_get_active_listener_port_by_type(). * Add an XXX to router_get_active_listener_port_by_type().
2012-04-12 22:42:37 +02:00
* spawning code, please make sure that it fits your purposes before
* using it. */
Pass OR address to PT proxy, even with IPv6 or ORListenAddress. Introduce get_first_listener_addrport_for_pt() which returns a string containing the addrport of the first listener we could find. Use it to form the TOR_PT_ORPORT managed proxy protocol line.
2012-03-31 14:04:58 +02:00
char *
Fix issues found by nickm. * Document fmt_addr_impl() and friends. * Parenthesize macro arguments. * Rename get_first_listener_addrport_for_pt() to get_first_listener_addrport_string(). * Handle port_cfg_t with no_listen. * Handle failure of router_get_active_listener_port_by_type(). * Add an XXX to router_get_active_listener_port_by_type().
2012-04-12 22:42:37 +02:00
get_first_listener_addrport_string(int listener_type)
Pass OR address to PT proxy, even with IPv6 or ORListenAddress. Introduce get_first_listener_addrport_for_pt() which returns a string containing the addrport of the first listener we could find. Use it to form the TOR_PT_ORPORT managed proxy protocol line.
2012-03-31 14:04:58 +02:00
{
static const char *ipv4_localhost = "127.0.0.1";
static const char *ipv6_localhost = "[::1]";
const char *address;
uint16_t port;
char *string = NULL;
if (!configured_ports)
return NULL;
SMARTLIST_FOREACH_BEGIN(configured_ports, const port_cfg_t *, cfg) {
Fix issues found by nickm. * Document fmt_addr_impl() and friends. * Parenthesize macro arguments. * Rename get_first_listener_addrport_for_pt() to get_first_listener_addrport_string(). * Handle port_cfg_t with no_listen. * Handle failure of router_get_active_listener_port_by_type(). * Add an XXX to router_get_active_listener_port_by_type().
2012-04-12 22:42:37 +02:00
if (cfg->no_listen)
continue;
Pass OR address to PT proxy, even with IPv6 or ORListenAddress. Introduce get_first_listener_addrport_for_pt() which returns a string containing the addrport of the first listener we could find. Use it to form the TOR_PT_ORPORT managed proxy protocol line.
2012-03-31 14:04:58 +02:00
if (cfg->type == listener_type &&
tor_addr_family(&cfg->addr) != AF_UNSPEC) {
/* We found the first listener of the type we are interested in! */
/* If a listener is listening on INADDR_ANY, assume that it's
also listening on 127.0.0.1, and point the transport proxy
there: */
if (tor_addr_is_null(&cfg->addr))
address = tor_addr_is_v4(&cfg->addr) ? ipv4_localhost : ipv6_localhost;
else
address = fmt_and_decorate_addr(&cfg->addr);
/* If a listener is configured with port 'auto', we are forced
to iterate all listener connections and find out in which
port it ended up listening: */
Fix issues found by nickm. * Document fmt_addr_impl() and friends. * Parenthesize macro arguments. * Rename get_first_listener_addrport_for_pt() to get_first_listener_addrport_string(). * Handle port_cfg_t with no_listen. * Handle failure of router_get_active_listener_port_by_type(). * Add an XXX to router_get_active_listener_port_by_type().
2012-04-12 22:42:37 +02:00
if (cfg->port == CFG_AUTO_PORT) {
Pass OR address to PT proxy, even with IPv6 or ORListenAddress. Introduce get_first_listener_addrport_for_pt() which returns a string containing the addrport of the first listener we could find. Use it to form the TOR_PT_ORPORT managed proxy protocol line.
2012-03-31 14:04:58 +02:00
port = router_get_active_listener_port_by_type(listener_type);
Fix issues found by nickm. * Document fmt_addr_impl() and friends. * Parenthesize macro arguments. * Rename get_first_listener_addrport_for_pt() to get_first_listener_addrport_string(). * Handle port_cfg_t with no_listen. * Handle failure of router_get_active_listener_port_by_type(). * Add an XXX to router_get_active_listener_port_by_type().
2012-04-12 22:42:37 +02:00
if (!port)
return NULL;
} else {
Pass OR address to PT proxy, even with IPv6 or ORListenAddress. Introduce get_first_listener_addrport_for_pt() which returns a string containing the addrport of the first listener we could find. Use it to form the TOR_PT_ORPORT managed proxy protocol line.
2012-03-31 14:04:58 +02:00
port = cfg->port;
Fix issues found by nickm. * Document fmt_addr_impl() and friends. * Parenthesize macro arguments. * Rename get_first_listener_addrport_for_pt() to get_first_listener_addrport_string(). * Handle port_cfg_t with no_listen. * Handle failure of router_get_active_listener_port_by_type(). * Add an XXX to router_get_active_listener_port_by_type().
2012-04-12 22:42:37 +02:00
}
Pass OR address to PT proxy, even with IPv6 or ORListenAddress. Introduce get_first_listener_addrport_for_pt() which returns a string containing the addrport of the first listener we could find. Use it to form the TOR_PT_ORPORT managed proxy protocol line.
2012-03-31 14:04:58 +02:00
tor_asprintf(&string, "%s:%u", address, port);
return string;
}
} SMARTLIST_FOREACH_END(cfg);
return NULL;
}
Turn get_first_advertised_v4_port_by_type() into get_first_advertised_port_by_type_af().
2011-11-24 17:49:31 +01:00
/** Return the first advertised port of type <b>listener_type</b> in
<b>address_family</b>. */
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
int
Turn get_first_advertised_v4_port_by_type() into get_first_advertised_port_by_type_af().
2011-11-24 17:49:31 +01:00
get_first_advertised_port_by_type_af(int listener_type, int address_family)
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
{
if (!configured_ports)
return 0;
SMARTLIST_FOREACH_BEGIN(configured_ports, const port_cfg_t *, cfg) {
if (cfg->type == listener_type &&
!cfg->no_advertise &&
Turn get_first_advertised_v4_port_by_type() into get_first_advertised_port_by_type_af().
2011-11-24 17:49:31 +01:00
(tor_addr_family(&cfg->addr) == address_family ||
Whitespace changes.
2011-11-29 11:00:43 +01:00
tor_addr_family(&cfg->addr) == AF_UNSPEC)) {
Turn get_first_advertised_v4_port_by_type() into get_first_advertised_port_by_type_af().
2011-11-24 17:49:31 +01:00
if (tor_addr_family(&cfg->addr) != AF_UNSPEC ||
Whitespace changes.
2011-11-29 11:00:43 +01:00
(address_family == AF_INET && !cfg->ipv6_only) ||
(address_family == AF_INET6 && !cfg->ipv4_only)) {
return cfg->port;
Turn get_first_advertised_v4_port_by_type() into get_first_advertised_port_by_type_af().
2011-11-24 17:49:31 +01:00
}
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
}
} SMARTLIST_FOREACH_END(cfg);
return 0;
Parse prop171 options; refactor listener/port option code Proposal 171 gives us a new syntax for parsing client port options. You can now have as many FooPort options as you want (for Foo in Socks, Trans, DNS, NATD), and they can have address:port arguments, and you can specify the level of isolation on those ports. Additionally, this patch refactors the client port parsing logic to use a new type, port_cfg_t. Previously, ports to be bound were half-parsed in config.c, and later re-parsed in connection.c when we're about to bind them. Now, parsing a port means converting it into a port_cfg_t, and binding it uses only a port_cfg_t, without needing to parse the user-provided strings at all. We should do a related refactoring on other port types. For control ports, that'll be easy enough. For ORPort and DirPort, we'll want to do this when we solve proposal 118 (letting servers bind to and advertise multiple ports). This implements tickets 3514 and 3515.
2011-06-30 20:01:02 +02:00
}
don't check permissions on our datadirectory until we switch uid/gid svn:r2759
2004-11-10 01:11:37 +01:00
/** Adjust the value of options->DataDirectory, or fill it in if it's
Resolve many XXXs and all DOCDOCs svn:r2755
2004-11-09 19:22:17 +01:00
* absent. Return 0 on success, -1 on failure. */
Normalize DataDirectory in options_validate, making SIGHUP survivable. svn:r2722
2004-11-09 06:26:49 +01:00
static int
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
normalize_data_directory(or_options_t *options)
{
Use the standard _WIN32, not the Torism MS_WINDOWS or deprecated WIN32 This commit is completely mechanical; I used this perl script to make it: #!/usr/bin/perl -w -i.bak -p if (/^\s*\#/) { s/MS_WINDOWS/_WIN32/g; s/\bWIN32\b/_WIN32/g; }
2012-01-31 16:59:42 +01:00
#ifdef _WIN32
Stop using the wrong DataDirectory when we're validating. Also validate/normalize the DataDirectory better. svn:r2732
2004-11-09 08:05:53 +01:00
char *p;
if (options->DataDirectory)
return 0; /* all set */
p = tor_malloc(MAX_PATH);
strlcpy(p,get_windows_conf_root(),MAX_PATH);
options->DataDirectory = p;
return 0;
On windows, default datadir is APPDATA/tor, and default config file is APPDATA/tor/torrc. (APPDATA is usually somedrive:/Documents and Settings/.../Application Data/) svn:r2265
2004-08-18 05:42:55 +02:00
#else
Stop using the wrong DataDirectory when we're validating. Also validate/normalize the DataDirectory better. svn:r2732
2004-11-09 08:05:53 +01:00
const char *d = options->DataDirectory;
Normalize space: add one between every control keyword and control clause. svn:r3003
2004-11-28 10:05:49 +01:00
if (!d)
Implement several 008pre1 items: needs more testing svn:r1981
2004-06-21 06:37:27 +02:00
d = "~/.tor";
more int to size_t conversions, fixing one or more amd64 bugs plus a whitespace patch on config.c from vicman svn:r2482
2004-10-14 04:47:09 +02:00
Stop using the wrong DataDirectory when we're validating. Also validate/normalize the DataDirectory better. svn:r2732
2004-11-09 08:05:53 +01:00
if (strncmp(d,"~/",2) == 0) {
Normalize DataDirectory in options_validate, making SIGHUP survivable. svn:r2722
2004-11-09 06:26:49 +01:00
char *fn = expand_filename(d);
if (!fn) {
polish r9726-r9903 svn:r10182
2007-05-13 11:25:06 +02:00
log_warn(LD_CONFIG,"Failed to expand filename \"%s\".", d);
Stop using the wrong DataDirectory when we're validating. Also validate/normalize the DataDirectory better. svn:r2732
2004-11-09 08:05:53 +01:00
return -1;
Normalize DataDirectory in options_validate, making SIGHUP survivable. svn:r2722
2004-11-09 06:26:49 +01:00
}
Resolve many XXXs and all DOCDOCs svn:r2755
2004-11-09 19:22:17 +01:00
if (!options->DataDirectory && !strcmp(fn,"/.tor")) {
/* If our homedir is /, we probably don't want to use it. */
r8851@totoro: nickm | 2006-10-02 18:13:27 -0400 Remove/clarify some XXXs for no longer being accurate; for begin things we do not indend to fix; for already being parts of big todo issues (like "/* XXX ipv6 */"); etc. Also fix some spaces. svn:r8580
2006-10-03 00:13:42 +02:00
/* Default to LOCALSTATEDIR/tor which is probably closer to what we
* want. */
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG,
"Default DataDirectory is \"~/.tor\". This expands to "
r12474@Kushana: nickm | 2007-03-06 16:10:05 -0500 We have a PATH_SEPARATOR macro. How about we use it? svn:r9782
2007-03-09 22:39:30 +01:00
"\"%s\", which is probably not what you want. Using "
"\"%s"PATH_SEPARATOR"tor\" instead", fn, LOCALSTATEDIR);
If DataDir is ~/.tor, and that expands to /.tor, then default to LOCALSTATEDIR/tor instead. svn:r4170
2005-05-03 02:36:57 +02:00
tor_free(fn);
r12474@Kushana: nickm | 2007-03-06 16:10:05 -0500 We have a PATH_SEPARATOR macro. How about we use it? svn:r9782
2007-03-09 22:39:30 +01:00
fn = tor_strdup(LOCALSTATEDIR PATH_SEPARATOR "tor");
Resolve many XXXs and all DOCDOCs svn:r2755
2004-11-09 19:22:17 +01:00
}
Normalize DataDirectory in options_validate, making SIGHUP survivable. svn:r2722
2004-11-09 06:26:49 +01:00
tor_free(options->DataDirectory);
options->DataDirectory = fn;
}
return 0;
Stop using the wrong DataDirectory when we're validating. Also validate/normalize the DataDirectory better. svn:r2732
2004-11-09 08:05:53 +01:00
#endif
}
Resolve many XXXs and all DOCDOCs svn:r2755
2004-11-09 19:22:17 +01:00
/** Check and normalize the value of options->DataDirectory; return 0 if it
Fix a typo
2009-12-19 21:08:58 +01:00
* is sane, -1 otherwise. */
Stop using the wrong DataDirectory when we're validating. Also validate/normalize the DataDirectory better. svn:r2732
2004-11-09 08:05:53 +01:00
static int
Docment or add DOCDOC comments to undocumented functions in src/or. Make function definition format uniform. svn:r4411
2005-06-11 20:52:12 +02:00
validate_data_directory(or_options_t *options)
{
once normalized, DataDirectory is always defined svn:r2734
2004-11-09 08:20:21 +01:00
if (normalize_data_directory(options) < 0)
Stop using the wrong DataDirectory when we're validating. Also validate/normalize the DataDirectory better. svn:r2732
2004-11-09 08:05:53 +01:00
return -1;
tor_assert(options->DataDirectory);
if (strlen(options->DataDirectory) > (512-128)) {
polish r9726-r9903 svn:r10182
2007-05-13 11:25:06 +02:00
log_warn(LD_CONFIG, "DataDirectory is too long.");
Stop using the wrong DataDirectory when we're validating. Also validate/normalize the DataDirectory better. svn:r2732
2004-11-09 08:05:53 +01:00
return -1;
}
return 0;
Implement several 008pre1 items: needs more testing svn:r1981
2004-06-21 06:37:27 +02:00
}
and some comments svn:r5656
2005-12-27 03:48:02 +01:00
/** This string must remain the same forevermore. It is how we
* recognize that the torrc file doesn't need to be backed up. */
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended. svn:r5582
2005-12-14 21:40:40 +01:00
#define GENERATED_FILE_PREFIX "# This file was generated by Tor; " \
"if you edit it, comments will not be preserved"
and some comments svn:r5656
2005-12-27 03:48:02 +01:00
/** This string can change; it tries to give the reader an idea
* that editing this file by hand is not a good plan. */
try a better string at the top of torrc's autogenerated torrc. svn:r6569
2006-06-09 04:20:42 +02:00
#define GENERATED_FILE_COMMENT "# The old torrc file was renamed " \
"to torrc.orig.1 or similar, and Tor will ignore it"
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
/** Save a configuration file for the configuration in <b>options</b>
* into the file <b>fname</b>. If the file already exists, and
* doesn't begin with GENERATED_FILE_PREFIX, rename it. Otherwise
* replace it. Return 0 on success, -1 on failure. */
static int
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
write_configuration_file(const char *fname, const or_options_t *options)
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
{
some fixes so far svn:r2880
2004-11-15 04:53:03 +01:00
char *old_val=NULL, *new_val=NULL, *new_conf=NULL;
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
int rename_old = 0, r;
Refactor some code and add some asserts based on scanner results. svn:r16783
2008-09-05 23:19:53 +02:00
tor_assert(fname);
switch (file_status(fname)) {
case FN_FILE:
old_val = read_file_to_str(fname, 0, NULL);
Fix crash when read_file_to_string() fails in SAVECONF The new behavior is to try to rename the old file if there is one there that we can't read. In all likelihood, that will fail too, but at least we tried, and at least it won't crash.
2011-05-12 04:05:41 +02:00
if (!old_val || strcmpstart(old_val, GENERATED_FILE_PREFIX)) {
Refactor some code and add some asserts based on scanner results. svn:r16783
2008-09-05 23:19:53 +02:00
rename_old = 1;
}
tor_free(old_val);
break;
case FN_NOENT:
break;
case FN_ERROR:
case FN_DIR:
default:
log_warn(LD_CONFIG,
"Config file \"%s\" is not a file? Failing.", fname);
return -1;
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
}
Rename cross-format config.c stuff to config_* and or_options_t-specific stuff to option[s]_* svn:r4641
2005-07-23 03:58:05 +02:00
if (!(new_conf = options_dump(options, 1))) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_BUG, "Couldn't get configuration string");
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
goto err;
}
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(&new_val, "%s\n%s\n\n%s",
and compiling would be good too svn:r5657
2005-12-27 03:48:35 +01:00
GENERATED_FILE_PREFIX, GENERATED_FILE_COMMENT, new_conf);
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
if (rename_old) {
int i = 1;
Convert instances of tor_malloc+tor_snprintf into tor_asprintf These were found by looking for tor_snprintf() instances that were preceeded closely by tor_malloc(), though I probably converted some more snprintfs as well. (In every case, make sure that the length variable (if any) is removed, renamed, or lowered, so that anything else that might have assumed a longer buffer doesn't exist.)
2012-01-11 20:02:59 +01:00
char *fn_tmp = NULL;
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
while (1) {
Convert instances of tor_malloc+tor_snprintf into tor_asprintf These were found by looking for tor_snprintf() instances that were preceeded closely by tor_malloc(), though I probably converted some more snprintfs as well. (In every case, make sure that the length variable (if any) is removed, renamed, or lowered, so that anything else that might have assumed a longer buffer doesn't exist.)
2012-01-11 20:02:59 +01:00
tor_asprintf(&fn_tmp, "%s.orig.%d", fname, i);
some fixes so far svn:r2880
2004-11-15 04:53:03 +01:00
if (file_status(fn_tmp) == FN_NOENT)
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
break;
Convert instances of tor_malloc+tor_snprintf into tor_asprintf These were found by looking for tor_snprintf() instances that were preceeded closely by tor_malloc(), though I probably converted some more snprintfs as well. (In every case, make sure that the length variable (if any) is removed, renamed, or lowered, so that anything else that might have assumed a longer buffer doesn't exist.)
2012-01-11 20:02:59 +01:00
tor_free(fn_tmp);
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
++i;
}
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_notice(LD_CONFIG, "Renaming old configuration file to \"%s\"", fn_tmp);
try to address bug 257: if rename() fails during saveconf, tell the controller. svn:r6049
2006-02-20 00:12:26 +01:00
if (rename(fname, fn_tmp) < 0) {
r8851@totoro: nickm | 2006-10-02 18:13:27 -0400 Remove/clarify some XXXs for no longer being accurate; for begin things we do not indend to fix; for already being parts of big todo issues (like "/* XXX ipv6 */"); etc. Also fix some spaces. svn:r8580
2006-10-03 00:13:42 +02:00
log_warn(LD_FS,
more minor cleanups svn:r8630
2006-10-07 08:28:50 +02:00
"Couldn't rename configuration file \"%s\" to \"%s\": %s",
fname, fn_tmp, strerror(errno));
try to address bug 257: if rename() fails during saveconf, tell the controller. svn:r6049
2006-02-20 00:12:26 +01:00
tor_free(fn_tmp);
goto err;
}
Remove assumptions that maximum path length is anything like 1024. From comments by Chris Palmer svn:r3652
2005-02-22 07:38:39 +01:00
tor_free(fn_tmp);
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
}
the other half of fixing bug 257. catch an error in more places. svn:r6050
2006-02-20 02:06:27 +01:00
if (write_str_to_file(fname, new_val, 0) < 0)
goto err;
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
r = 0;
goto done;
err:
r = -1;
done:
tor_free(new_val);
tor_free(new_conf);
return r;
}
/**
* Save the current configuration file value to disk. Return 0 on
* success, -1 on failure.
**/
int
Rename cross-format config.c stuff to config_* and or_options_t-specific stuff to option[s]_* svn:r4641
2005-07-23 03:58:05 +02:00
options_save_current(void)
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
{
Remove some redundant code in options_save_current() get_torrc_fname() does the same thing we did in this code, so let's replace it.
2010-02-22 12:21:58 +01:00
/* This fails if we can't write to our configuration file.
*
* If we try falling back to datadirectory or something, we have a better
* chance of saving the configuration, but a better chance of doing
* something the user never expected. */
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
return write_configuration_file(get_torrc_fname(0), get_options());
Add simpler option_is_same, as suggested by arma. Finish implementing SAVECONF svn:r2868
2004-11-14 21:51:28 +01:00
}
Make doxygen marginally happier svn:r5208
2005-10-06 06:33:40 +02:00
/** Mapping from a unit name to a multiplier for converting that unit into a
Document a few more members.
2011-03-16 23:10:15 +01:00
* base unit. Used by config_parse_unit. */
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
struct unit_table_t {
Document a few more members.
2011-03-16 23:10:15 +01:00
const char *unit; /**< The name of the unit */
uint64_t multiplier; /**< How many of the base unit appear in this unit */
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
};
Document most undocumented variables. svn:r17754
2008-12-23 18:56:31 +01:00
/** Table to map the names of memory units to the number of bytes they
* contain. */
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
static struct unit_table_t memory_units[] = {
Bugfix: when no units are given, default to seconds or bytes. svn:r2960
2004-11-23 11:52:27 +01:00
{ "", 1 },
Implement weekly/monthly/daily accounting svn:r2936
2004-11-22 22:56:51 +01:00
{ "b", 1<< 0 },
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
{ "byte", 1<< 0 },
{ "bytes", 1<< 0 },
{ "kb", 1<<10 },
fix some irix compile complaints; make "kbytes" work as a memory unit svn:r12936
2007-12-23 05:22:55 +01:00
{ "kbyte", 1<<10 },
{ "kbytes", 1<<10 },
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
{ "kilobyte", 1<<10 },
{ "kilobytes", 1<<10 },
{ "m", 1<<20 },
{ "mb", 1<<20 },
fix some irix compile complaints; make "kbytes" work as a memory unit svn:r12936
2007-12-23 05:22:55 +01:00
{ "mbyte", 1<<20 },
{ "mbytes", 1<<20 },
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
{ "megabyte", 1<<20 },
{ "megabytes", 1<<20 },
{ "gb", 1<<30 },
fix some irix compile complaints; make "kbytes" work as a memory unit svn:r12936
2007-12-23 05:22:55 +01:00
{ "gbyte", 1<<30 },
{ "gbytes", 1<<30 },
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
{ "gigabyte", 1<<30 },
{ "gigabytes", 1<<30 },
{ "tb", U64_LITERAL(1)<<40 },
{ "terabyte", U64_LITERAL(1)<<40 },
{ "terabytes", U64_LITERAL(1)<<40 },
{ NULL, 0 },
};
Document most undocumented variables. svn:r17754
2008-12-23 18:56:31 +01:00
/** Table to map the names of time units to the number of seconds they
* contain. */
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
static struct unit_table_t time_units[] = {
Bugfix: when no units are given, default to seconds or bytes. svn:r2960
2004-11-23 11:52:27 +01:00
{ "", 1 },
fix some typos; move to 0.0.9rc1 svn:r2951
2004-11-23 08:37:25 +01:00
{ "second", 1 },
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
{ "seconds", 1 },
{ "minute", 60 },
{ "minutes", 60 },
{ "hour", 60*60 },
{ "hours", 60*60 },
{ "day", 24*60*60 },
{ "days", 24*60*60 },
{ "week", 7*24*60*60 },
{ "weeks", 7*24*60*60 },
{ NULL, 0 },
};
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
/** Table to map the names of time units to the number of milliseconds
* they contain. */
static struct unit_table_t time_msec_units[] = {
{ "", 1 },
{ "msec", 1 },
{ "millisecond", 1 },
{ "milliseconds", 1 },
{ "second", 1000 },
{ "seconds", 1000 },
{ "minute", 60*1000 },
{ "minutes", 60*1000 },
{ "hour", 60*60*1000 },
{ "hours", 60*60*1000 },
{ "day", 24*60*60*1000 },
{ "days", 24*60*60*1000 },
{ "week", 7*24*60*60*1000 },
{ "weeks", 7*24*60*60*1000 },
{ NULL, 0 },
};
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
/** Parse a string <b>val</b> containing a number, zero or more
* spaces, and an optional unit string. If the unit appears in the
* table <b>u</b>, then multiply the number by the unit multiplier.
* On success, set *<b>ok</b> to 1 and return this product.
* Otherwise, set *<b>ok</b> to 0.
*/
static uint64_t
config_parse_units(const char *val, struct unit_table_t *u, int *ok)
{
Allow interval and memunit cfg variables to be set to fractions.
2009-07-15 16:02:49 +02:00
uint64_t v = 0;
double d = 0;
int use_float = 0;
Revise parsing of time and memory units to handle spaces. When we added support for fractional units (like 1.5 MB) I broke support for giving units with no space (like 2MB). This patch should fix that. It also adds a propoer tor_parse_double(). Fix for bug 1076. Bugfix on 0.2.2.1-alpha.
2009-08-31 06:18:55 +02:00
char *cp;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
tor_assert(ok);
Revise parsing of time and memory units to handle spaces. When we added support for fractional units (like 1.5 MB) I broke support for giving units with no space (like 2MB). This patch should fix that. It also adds a propoer tor_parse_double(). Fix for bug 1076. Bugfix on 0.2.2.1-alpha.
2009-08-31 06:18:55 +02:00
v = tor_parse_uint64(val, 10, 0, UINT64_MAX, ok, &cp);
if (!*ok || (cp && *cp == '.')) {
d = tor_parse_double(val, 0, UINT64_MAX, ok, &cp);
if (!*ok)
Allow interval and memunit cfg variables to be set to fractions.
2009-07-15 16:02:49 +02:00
goto done;
use_float = 1;
}
Revise parsing of time and memory units to handle spaces. When we added support for fractional units (like 1.5 MB) I broke support for giving units with no space (like 2MB). This patch should fix that. It also adds a propoer tor_parse_double(). Fix for bug 1076. Bugfix on 0.2.2.1-alpha.
2009-08-31 06:18:55 +02:00
if (!cp) {
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
*ok = 1;
Allow interval and memunit cfg variables to be set to fractions.
2009-07-15 16:02:49 +02:00
v = use_float ? DBL_TO_U64(d) : v;
goto done;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
}
Allow interval and memunit cfg variables to be set to fractions.
2009-07-15 16:02:49 +02:00
Revise parsing of time and memory units to handle spaces. When we added support for fractional units (like 1.5 MB) I broke support for giving units with no space (like 2MB). This patch should fix that. It also adds a propoer tor_parse_double(). Fix for bug 1076. Bugfix on 0.2.2.1-alpha.
2009-08-31 06:18:55 +02:00
cp = (char*) eat_whitespace(cp);
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
for ( ;u->unit;++u) {
Revise parsing of time and memory units to handle spaces. When we added support for fractional units (like 1.5 MB) I broke support for giving units with no space (like 2MB). This patch should fix that. It also adds a propoer tor_parse_double(). Fix for bug 1076. Bugfix on 0.2.2.1-alpha.
2009-08-31 06:18:55 +02:00
if (!strcasecmp(u->unit, cp)) {
Allow interval and memunit cfg variables to be set to fractions.
2009-07-15 16:02:49 +02:00
if (use_float)
v = u->multiplier * d;
else
v *= u->multiplier;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
*ok = 1;
Allow interval and memunit cfg variables to be set to fractions.
2009-07-15 16:02:49 +02:00
goto done;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
}
}
Revise parsing of time and memory units to handle spaces. When we added support for fractional units (like 1.5 MB) I broke support for giving units with no space (like 2MB). This patch should fix that. It also adds a propoer tor_parse_double(). Fix for bug 1076. Bugfix on 0.2.2.1-alpha.
2009-08-31 06:18:55 +02:00
log_warn(LD_CONFIG, "Unknown unit '%s'.", cp);
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
*ok = 0;
Allow interval and memunit cfg variables to be set to fractions.
2009-07-15 16:02:49 +02:00
done:
if (*ok)
return v;
else
return 0;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
}
/** Parse a string in the format "number unit", where unit is a unit of
* information (byte, KB, M, etc). On success, set *<b>ok</b> to true
* and return the number of bytes specified. Otherwise, set
* *<b>ok</b> to false and return 0. */
static uint64_t
Reformat inconsistent function declarations. svn:r5160
2005-09-30 03:09:52 +02:00
config_parse_memunit(const char *s, int *ok)
{
Revise parsing of time and memory units to handle spaces. When we added support for fractional units (like 1.5 MB) I broke support for giving units with no space (like 2MB). This patch should fix that. It also adds a propoer tor_parse_double(). Fix for bug 1076. Bugfix on 0.2.2.1-alpha.
2009-08-31 06:18:55 +02:00
uint64_t u = config_parse_units(s, memory_units, ok);
return u;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
}
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
/** Parse a string in the format "number unit", where unit is a unit of
* time in milliseconds. On success, set *<b>ok</b> to true and return
* the number of milliseconds in the provided interval. Otherwise, set
* *<b>ok</b> to 0 and return -1. */
static int
config_parse_msec_interval(const char *s, int *ok)
{
uint64_t r;
r = config_parse_units(s, time_msec_units, ok);
if (!ok)
return -1;
if (r > INT_MAX) {
log_warn(LD_CONFIG, "Msec interval '%s' is too long", s);
*ok = 0;
return -1;
}
return (int)r;
}
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
/** Parse a string in the format "number unit", where unit is a unit of time.
* On success, set *<b>ok</b> to true and return the number of seconds in
* the provided interval. Otherwise, set *<b>ok</b> to 0 and return -1.
*/
static int
Reformat inconsistent function declarations. svn:r5160
2005-09-30 03:09:52 +02:00
config_parse_interval(const char *s, int *ok)
{
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
uint64_t r;
r = config_parse_units(s, time_units, ok);
if (!ok)
return -1;
if (r > INT_MAX) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_CONFIG, "Interval '%s' is too long", s);
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
*ok = 0;
return -1;
}
Compile cleanly on windows; prevent some insane bandwidth cases (e.g., "BandwidthBurst 1000 TB" from occuring. svn:r2941
2004-11-22 23:24:10 +01:00
return (int)r;
Add "MEMUNIT" and "INTERVAL" types to configuration. Also tweak Accounting setup. More docs needed svn:r2911
2004-11-20 01:37:00 +01:00
}
Autodetect the number of CPUs when possible if NumCPUs==0 This is needed for IOCP, since telling the IOCP backend about all your CPUs is a good idea. It'll also come in handy with asn's multithreaded crypto stuff, and for people who run servers without reading the manual.
2010-09-28 20:36:28 +02:00
/** Return the number of cpus configured in <b>options</b>. If we are
* told to auto-detect the number of cpus, return the auto-detected number. */
int
get_num_cpus(const or_options_t *options)
{
Merge remote branch 'origin/maint-0.2.2' Conflicts: src/or/config.c src/or/cpuworker.c
2010-11-15 20:14:13 +01:00
if (options->NumCPUs == 0) {
Autodetect the number of CPUs when possible if NumCPUs==0 This is needed for IOCP, since telling the IOCP backend about all your CPUs is a good idea. It'll also come in handy with asn's multithreaded crypto stuff, and for people who run servers without reading the manual.
2010-09-28 20:36:28 +02:00
int n = compute_num_cpus();
return (n >= 1) ? n : 1;
} else {
Merge remote branch 'origin/maint-0.2.2' Conflicts: src/or/config.c src/or/cpuworker.c
2010-11-15 20:14:13 +01:00
return options->NumCPUs;
Autodetect the number of CPUs when possible if NumCPUs==0 This is needed for IOCP, since telling the IOCP backend about all your CPUs is a good idea. It'll also come in handy with asn's multithreaded crypto stuff, and for people who run servers without reading the manual.
2010-09-28 20:36:28 +02:00
}
}
add patch for 143 and code to check for known-bad/slow libevent versions. Needs more testing. svn:r4344
2005-06-08 15:53:19 +02:00
/**
* Initialize the libevent library.
*/
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
static void
Add a new option to enable/disable IOCP support
2010-09-28 20:01:45 +02:00
init_libevent(const or_options_t *options)
add patch for 143 and code to check for known-bad/slow libevent versions. Needs more testing. svn:r4344
2005-06-08 15:53:19 +02:00
{
Move the Libvent setup logic into compat_libevent from config. This has been some pretty ugly and voodoo-laden code. I've tried to clean it up a bit, but more work probably remains.
2009-06-04 09:16:26 +02:00
const char *badness=NULL;
Add a new option to enable/disable IOCP support
2010-09-28 20:01:45 +02:00
tor_libevent_cfg cfg;
tor_assert(options);
Move the Libvent setup logic into compat_libevent from config. This has been some pretty ugly and voodoo-laden code. I've tried to clean it up a bit, but more work probably remains.
2009-06-04 09:16:26 +02:00
add patch for 143 and code to check for known-bad/slow libevent versions. Needs more testing. svn:r4344
2005-06-08 15:53:19 +02:00
configure_libevent_logging();
/* If the kernel complains that some method (say, epoll) doesn't
* exist, we don't care about it, since libevent will cope.
*/
suppress_libevent_log_msg("Function not implemented");
Check that Libevent header version matches Libevent library version. Unfortunately, old Libevents don't _put_ a version in their headers, so this can get a little tricky. Fortunately, the only binary-compatibility issue we care about is the size of struct event. Even more fortunately, Libevent 2.0 will let us keep binary compatiblity forever by letting us decouple ourselves from the structs, if we like. svn:r18014
2009-01-07 22:05:02 +01:00
Move the Libvent setup logic into compat_libevent from config. This has been some pretty ugly and voodoo-laden code. I've tried to clean it up a bit, but more work probably remains.
2009-06-04 09:16:26 +02:00
tor_check_libevent_header_compatibility();
Check that Libevent header version matches Libevent library version. Unfortunately, old Libevents don't _put_ a version in their headers, so this can get a little tricky. Fortunately, the only binary-compatibility issue we care about is the size of struct event. Even more fortunately, Libevent 2.0 will let us keep binary compatiblity forever by letting us decouple ourselves from the structs, if we like. svn:r18014
2009-01-07 22:05:02 +01:00
Add a new option to enable/disable IOCP support
2010-09-28 20:01:45 +02:00
memset(&cfg, 0, sizeof(cfg));
cfg.disable_iocp = options->DisableIOCP;
Autodetect the number of CPUs when possible if NumCPUs==0 This is needed for IOCP, since telling the IOCP backend about all your CPUs is a good idea. It'll also come in handy with asn's multithreaded crypto stuff, and for people who run servers without reading the manual.
2010-09-28 20:36:28 +02:00
cfg.num_cpus = get_num_cpus(options);
Make bufferevents work with TokenBucketRefillInterval
2011-09-08 04:00:48 +02:00
cfg.msec_per_tick = options->TokenBucketRefillInterval;
Add a new option to enable/disable IOCP support
2010-09-28 20:01:45 +02:00
tor_libevent_initialize(&cfg);
Update Tor to use Libevent 2.0 APIs when available. This patch adds a new compat_libevent.[ch] set of files, and moves our Libevent compatibility and utilitity functions there. We build them into a separate .a so that nothing else in src/commmon depends on Libevent (partially fixing bug 507). Also, do not use our own built-in evdns copy when we have Libevent 2.0, whose evdns is finally good enough (thus fixing Bug 920).
2009-06-04 07:05:23 +02:00
add patch for 143 and code to check for known-bad/slow libevent versions. Needs more testing. svn:r4344
2005-06-08 15:53:19 +02:00
suppress_libevent_log_msg(NULL);
Experimentally re-enable kqueue on OSX when using libevent 1.1b or later. Log when we are doing this, so we can diagnose it when it fails. svn:r7004
2006-08-10 10:00:13 +02:00
Move the Libvent setup logic into compat_libevent from config. This has been some pretty ugly and voodoo-laden code. I've tried to clean it up a bit, but more work probably remains.
2009-06-04 09:16:26 +02:00
tor_check_libevent_version(tor_libevent_get_method(),
Initial hacking for proposal 186. This code handles the new ORPort options, and incidentally makes all remaining port types use the new port configuration systems. There are some rough edges! It doesn't do well in the case where your Address says one thing but you say to Advertise another ORPort. It doesn't handle AllAddrs. It doesn't actually advertise anything besides the first listed advertised IPv4 ORPort and DirPort. It doesn't do port forwarding to them either. It's not tested either, it needs more documentation, and it probably forgets to put the milk back in the refrigerator.
2011-11-08 22:10:38 +01:00
get_options()->ORPort != NULL,
Move the Libvent setup logic into compat_libevent from config. This has been some pretty ugly and voodoo-laden code. I've tried to clean it up a bit, but more work probably remains.
2009-06-04 09:16:26 +02:00
&badness);
r11872@Kushana: nickm | 2007-01-06 02:14:12 -0500 Implement a control status event for bad libevent version/method combos. Warn that libevent <1.1 with select() is needlessly slow. Reply to comment. svn:r9284
2007-01-06 08:34:02 +01:00
if (badness) {
Move the Libvent setup logic into compat_libevent from config. This has been some pretty ugly and voodoo-laden code. I've tried to clean it up a bit, but more work probably remains.
2009-06-04 09:16:26 +02:00
const char *v = tor_libevent_get_version_str();
const char *m = tor_libevent_get_method();
r11872@Kushana: nickm | 2007-01-06 02:14:12 -0500 Implement a control status event for bad libevent version/method combos. Warn that libevent <1.1 with select() is needlessly slow. Reply to comment. svn:r9284
2007-01-06 08:34:02 +01:00
control_event_general_status(LOG_WARN,
"BAD_LIBEVENT VERSION=%s METHOD=%s BADNESS=%s RECOVERED=NO",
v, m, badness);
add patch for 143 and code to check for known-bad/slow libevent versions. Needs more testing. svn:r4344
2005-06-08 15:53:19 +02:00
}
}
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Return the persistent state struct for this Tor. */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
or_state_t *
get_or_state(void)
{
Fix crash with "tor --list-fingerprint" (reported by seeess). svn:r9328
2007-01-11 00:48:24 +01:00
tor_assert(global_state);
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
return global_state;
}
r15877@catbus: nickm | 2007-10-17 12:54:56 -0400 Make unverified-consensus get removed when it is accepted or rejected. Make a new get_datadir_fname*() set of functions to eliminate the common code of "get the options, get the datadir, append some stuff". svn:r12000
2007-10-17 18:55:44 +02:00
/** Return a newly allocated string holding a filename relative to the data
* directory. If <b>sub1</b> is present, it is the first path component after
* the data directory. If <b>sub2</b> is also present, it is the second path
* component after the data directory. If <b>suffix</b> is present, it
* is appended to the filename.
*
* Examples:
* get_datadir_fname2_suffix("a", NULL, NULL) -> $DATADIR/a
* get_datadir_fname2_suffix("a", NULL, ".tmp") -> $DATADIR/a.tmp
* get_datadir_fname2_suffix("a", "b", ".tmp") -> $DATADIR/a/b/.tmp
* get_datadir_fname2_suffix("a", "b", NULL) -> $DATADIR/a/b
*
* Note: Consider using the get_datadir_fname* macros in or.h.
*/
char *
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
options_get_datadir_fname2_suffix(const or_options_t *options,
Add a lockfile to the Tor data directory to avoid situations where two Tors start with the same datadir, or where a --list-fingerprints races with a server to create keys, or such. svn:r16722
2008-09-01 22:06:26 +02:00
const char *sub1, const char *sub2,
const char *suffix)
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
{
r15877@catbus: nickm | 2007-10-17 12:54:56 -0400 Make unverified-consensus get removed when it is accepted or rejected. Make a new get_datadir_fname*() set of functions to eliminate the common code of "get the options, get the datadir, append some stuff". svn:r12000
2007-10-17 18:55:44 +02:00
char *fname = NULL;
size_t len;
tor_assert(options);
tor_assert(options->DataDirectory);
tor_assert(sub1 || !sub2); /* If sub2 is present, sub1 must be present. */
len = strlen(options->DataDirectory);
if (sub1) {
len += strlen(sub1)+1;
if (sub2)
len += strlen(sub2)+1;
}
if (suffix)
len += strlen(suffix);
len++;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
fname = tor_malloc(len);
r15877@catbus: nickm | 2007-10-17 12:54:56 -0400 Make unverified-consensus get removed when it is accepted or rejected. Make a new get_datadir_fname*() set of functions to eliminate the common code of "get the options, get the datadir, append some stuff". svn:r12000
2007-10-17 18:55:44 +02:00
if (sub1) {
if (sub2) {
tor_snprintf(fname, len, "%s"PATH_SEPARATOR"%s"PATH_SEPARATOR"%s",
options->DataDirectory, sub1, sub2);
} else {
tor_snprintf(fname, len, "%s"PATH_SEPARATOR"%s",
options->DataDirectory, sub1);
}
} else {
strlcpy(fname, options->DataDirectory, len);
}
if (suffix)
strlcat(fname, suffix, len);
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
return fname;
}
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
/** Return true if <b>line</b> is a valid state TransportProxy line.
* Return false otherwise. */
static int
Constification.
2011-09-11 23:35:00 +02:00
state_transport_line_is_valid(const char *line)
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
{
smartlist_t *items = NULL;
char *addrport=NULL;
tor_addr_t addr;
uint16_t port = 0;
int r;
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
items = smartlist_new();
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
smartlist_split_string(items, line, NULL,
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, -1);
if (smartlist_len(items) != 2) {
log_warn(LD_CONFIG, "state: Not enough arguments in TransportProxy line.");
goto err;
}
addrport = smartlist_get(items, 1);
Fix names of functions that convert strings to addrs Now let's have "lookup" indicate that there can be a hostname resolution, and "parse" indicate that there wasn't. Previously, we had one "lookup" function that did resolution; four "parse" functions, half of which did resolution; and a "from_str()" function that didn't do resolution. That's confusing and error-prone! The code changes in this commit are exactly the result of this perl script, run under "perl -p -i.bak" : s/tor_addr_port_parse/tor_addr_port_lookup/g; s/parse_addr_port(?=[^_])/addr_port_lookup/g; s/tor_addr_from_str/tor_addr_parse/g; This patch leaves aton and pton alone: their naming convention and behavior is is determined by the sockets API. More renaming may be needed.
2011-10-11 17:21:31 +02:00
if (tor_addr_port_lookup(addrport, &addr, &port) < 0) {
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
log_warn(LD_CONFIG, "state: Could not parse addrport.");
goto err;
}
if (!port) {
log_warn(LD_CONFIG, "state: Transport line did not contain port.");
goto err;
}
r = 1;
goto done;
err:
r = 0;
done:
SMARTLIST_FOREACH(items, char*, s, tor_free(s));
smartlist_free(items);
return r;
}
/** Return 0 if all TransportProxy lines in <b>state</b> are well
* formed. Otherwise, return -1. */
static int
validate_transports_in_state(or_state_t *state)
{
int broken = 0;
Improve how we access or_state_t. * Use get_or_state()->VirtualOption instead of relying on config_find_option(), STRUCT_VAR_P and voodoo.
2011-09-11 20:57:01 +02:00
config_line_t *line;
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
Improve how we access or_state_t. * Use get_or_state()->VirtualOption instead of relying on config_find_option(), STRUCT_VAR_P and voodoo.
2011-09-11 20:57:01 +02:00
for (line = state->TransportProxies ; line ; line = line->next) {
tor_assert(!strcmp(line->key, "TransportProxy"));
Fix a couple of pluggable transport bugs Fix coverity complaints 490, 491 and 492. Especially the one in parse_client_transport_line() could've been a remotely triggerable segfault, I think.
2011-10-26 16:49:24 +02:00
if (!state_transport_line_is_valid(line->value))
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
broken = 1;
}
if (broken)
log_warn(LD_CONFIG, "state: State file seems to be broken.");
return 0;
}
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Return 0 if every setting in <b>state</b> is reasonable, and a
* permissible transition from <b>old_state</b>. Else warn and return -1.
* Should have no side effects, except for normalizing the contents of
* <b>state</b>.
*/
or_state_validate() needs an extra arg too or it will crash svn:r5772
2006-01-10 21:38:33 +01:00
/* XXX from_setconf is here because of bug 238 */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
static int
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
or_state_validate(or_state_t *old_state, or_state_t *state,
int from_setconf, char **msg)
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
{
Add a new warning to our "warn a lot" list: unused parameters. This means we have to explicitly "use" unuseds, but it can catch bugs. (It caught two coding mistakes so far.) svn:r6532
2006-06-05 00:42:13 +02:00
/* We don't use these; only options do. Still, we need to match that
* signature. */
(void) from_setconf;
(void) old_state;
cleanups on r11258 svn:r11263
2007-08-24 10:01:47 +02:00
revert the recommended-guard-version thing. it did not do what we want, which is to expire old guards *every* time somebody moves from an old version to the new one. also, refine which version numbers count as 'new enough'. svn:r11272
2007-08-25 22:34:13 +02:00
if (entry_guards_parse_state(state, 0, msg)<0)
save and validate TorVersion in the state file. svn:r5694
2006-01-02 05:14:52 +01:00
return -1;
revert the recommended-guard-version thing. it did not do what we want, which is to expire old guards *every* time somebody moves from an old version to the new one. also, refine which version numbers count as 'new enough'. svn:r11272
2007-08-25 22:34:13 +02:00
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
if (validate_transports_in_state(state)<0)
return -1;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
return 0;
}
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Replace the current persistent state with <b>new_state</b> */
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
static int
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
or_state_set(or_state_t *new_state)
{
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
char *err = NULL;
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
int ret = 0;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
tor_assert(new_state);
Now that FOO_free(NULL) always works, remove checks before calling it.
2009-12-12 08:07:59 +01:00
config_free(&state_format, global_state);
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
global_state = new_state;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
if (entry_guards_parse_state(global_state, 1, &err)<0) {
log_warn(LD_GENERAL,"%s",err);
tor_free(err);
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
ret = -1;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
}
if (rep_hist_load_state(global_state, &err)<0) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_GENERAL,"Unparseable bandwidth history state: %s",err);
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
tor_free(err);
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
ret = -1;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
}
Refactor circuit_build_times_parse_state Remove the msg parameter to pass an error message out. This wasn't needed and made it harder to detect a memory leak.
2010-08-15 14:22:32 +02:00
if (circuit_build_times_parse_state(&circ_times, global_state) < 0) {
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
ret = -1;
}
return ret;
}
/**
* Save a broken state file to a backup location.
*/
static void
or_state_save_broken(char *fname)
{
int i;
file_status_t status;
Convert instances of tor_malloc+tor_snprintf into tor_asprintf These were found by looking for tor_snprintf() instances that were preceeded closely by tor_malloc(), though I probably converted some more snprintfs as well. (In every case, make sure that the length variable (if any) is removed, renamed, or lowered, so that anything else that might have assumed a longer buffer doesn't exist.)
2012-01-11 20:02:59 +01:00
char *fname2 = NULL;
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
for (i = 0; i < 100; ++i) {
Convert instances of tor_malloc+tor_snprintf into tor_asprintf These were found by looking for tor_snprintf() instances that were preceeded closely by tor_malloc(), though I probably converted some more snprintfs as well. (In every case, make sure that the length variable (if any) is removed, renamed, or lowered, so that anything else that might have assumed a longer buffer doesn't exist.)
2012-01-11 20:02:59 +01:00
tor_asprintf(&fname2, "%s.%d", fname, i);
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
status = file_status(fname2);
if (status == FN_NOENT)
break;
Convert instances of tor_malloc+tor_snprintf into tor_asprintf These were found by looking for tor_snprintf() instances that were preceeded closely by tor_malloc(), though I probably converted some more snprintfs as well. (In every case, make sure that the length variable (if any) is removed, renamed, or lowered, so that anything else that might have assumed a longer buffer doesn't exist.)
2012-01-11 20:02:59 +01:00
tor_free(fname2);
Clean up Fallon's partially complete GSoC project. The code actually isn't that bad. It's a shame she didn't finish. Using it as the base for this feature.
2009-08-26 02:13:12 +02:00
}
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
if (i == 100) {
log_warn(LD_BUG, "Unable to parse state in \"%s\"; too many saved bad "
"state files to move aside. Discarding the old state file.",
fname);
unlink(fname);
} else {
log_warn(LD_BUG, "Unable to parse state in \"%s\". Moving it aside "
"to \"%s\". This could be a bug in Tor; please tell "
"the developers.", fname, fname2);
if (rename(fname, fname2) < 0) {
log_warn(LD_BUG, "Weirdly, I couldn't even move the state aside. The "
"OS gave an error of %s", strerror(errno));
}
}
tor_free(fname2);
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
}
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Reload the persistent state from disk, generating a new state as needed.
* Return 0 on success, less than 0 on failure.
*/
r15512@catbus: nickm | 2007-10-02 16:27:43 -0400 Make some functions static; remove some dead code. svn:r11750
2007-10-02 22:35:23 +02:00
static int
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
or_state_load(void)
{
or_state_t *new_state = NULL;
char *contents = NULL, *fname;
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
char *errmsg = NULL;
r8572@Kushana: nickm | 2006-08-25 16:35:49 -0400 Fix for bug 308: When we have a state file we cannot parse, tell the user, and move it aside. svn:r8224
2006-08-25 23:01:56 +02:00
int r = -1, badstate = 0;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
r15877@catbus: nickm | 2007-10-17 12:54:56 -0400 Make unverified-consensus get removed when it is accepted or rejected. Make a new get_datadir_fname*() set of functions to eliminate the common code of "get the options, get the datadir, append some stuff". svn:r12000
2007-10-17 18:55:44 +02:00
fname = get_datadir_fname("state");
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
switch (file_status(fname)) {
case FN_FILE:
r9274@Kushana: nickm | 2006-10-19 16:16:58 -0400 Add unit tests for tor_mmap_file(); make tor_mmap_t.size always be the size of the file (not the size of the mapping); add an extra argument to read_file_to_str() so it can return the size of the result string. svn:r8762
2006-10-20 01:05:02 +02:00
if (!(contents = read_file_to_str(fname, 0, NULL))) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_FS, "Unable to read state file \"%s\"", fname);
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
goto done;
}
break;
case FN_NOENT:
break;
r7324@Kushana: nickm | 2006-08-10 23:23:15 -0700 Add more warnings to the list of those we tolerate. Start using GCC attributes more, for better error checking and better code generation. svn:r7020
2006-08-11 09:09:17 +02:00
case FN_ERROR:
case FN_DIR:
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
default:
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_warn(LD_GENERAL,"State file \"%s\" is not a file? Failing.", fname);
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
goto done;
}
new_state = tor_malloc_zero(sizeof(or_state_t));
new_state->_magic = OR_STATE_MAGIC;
config_init(&state_format, new_state);
if (contents) {
config_line_t *lines=NULL;
int assign_retval;
Add the ability to append and clear linelist options from cmdline This will be important for getting stuff to work right across zones.
2011-11-28 03:32:51 +01:00
if (config_get_lines(contents, &lines, 0)<0)
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
goto done;
also send syntax and parse errors back to the controller. svn:r6242
2006-03-26 10:09:19 +02:00
assign_retval = config_assign(&state_format, new_state,
lines, 0, 0, &errmsg);
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
config_free_lines(lines);
if (assign_retval<0)
r8572@Kushana: nickm | 2006-08-25 16:35:49 -0400 Fix for bug 308: When we have a state file we cannot parse, tell the user, and move it aside. svn:r8224
2006-08-25 23:01:56 +02:00
badstate = 1;
if (errmsg) {
log_warn(LD_GENERAL, "%s", errmsg);
tor_free(errmsg);
}
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
}
r8572@Kushana: nickm | 2006-08-25 16:35:49 -0400 Fix for bug 308: When we have a state file we cannot parse, tell the user, and move it aside. svn:r8224
2006-08-25 23:01:56 +02:00
if (!badstate && or_state_validate(NULL, new_state, 1, &errmsg) < 0)
badstate = 1;
if (errmsg) {
log_warn(LD_GENERAL, "%s", errmsg);
tor_free(errmsg);
When the controller's *setconf commands fail, collect an error message in a string and hand it back. This starts to resolve bug 275. svn:r6241
2006-03-26 08:51:26 +02:00
}
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
r8572@Kushana: nickm | 2006-08-25 16:35:49 -0400 Fix for bug 308: When we have a state file we cannot parse, tell the user, and move it aside. svn:r8224
2006-08-25 23:01:56 +02:00
if (badstate && !contents) {
fix a log level -- err is for things that kill tor, warn is for things that tor can recover from. also, avoid situations where people who don't read their logs accumulate ten thousand useless files in their datadir. svn:r8227
2006-08-26 06:48:50 +02:00
log_warn(LD_BUG, "Uh oh. We couldn't even validate our own default state."
" This is a bug in Tor.");
r8572@Kushana: nickm | 2006-08-25 16:35:49 -0400 Fix for bug 308: When we have a state file we cannot parse, tell the user, and move it aside. svn:r8224
2006-08-25 23:01:56 +02:00
goto done;
} else if (badstate && contents) {
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
or_state_save_broken(fname);
r8572@Kushana: nickm | 2006-08-25 16:35:49 -0400 Fix for bug 308: When we have a state file we cannot parse, tell the user, and move it aside. svn:r8224
2006-08-25 23:01:56 +02:00
tor_free(contents);
config_free(&state_format, new_state);
new_state = tor_malloc_zero(sizeof(or_state_t));
new_state->_magic = OR_STATE_MAGIC;
config_init(&state_format, new_state);
} else if (contents) {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_info(LD_GENERAL, "Loaded state from \"%s\"", fname);
r8572@Kushana: nickm | 2006-08-25 16:35:49 -0400 Fix for bug 308: When we have a state file we cannot parse, tell the user, and move it aside. svn:r8224
2006-08-25 23:01:56 +02:00
} else {
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_info(LD_GENERAL, "Initialized state");
r8572@Kushana: nickm | 2006-08-25 16:35:49 -0400 Fix for bug 308: When we have a state file we cannot parse, tell the user, and move it aside. svn:r8224
2006-08-25 23:01:56 +02:00
}
More gracefully handle corrupt state files. Save a backup if we get odd circuitbuildtimes and other state info. In the case of circuit build times, we no longer assert, and reset our state.
2010-07-06 21:08:13 +02:00
if (or_state_set(new_state) == -1) {
or_state_save_broken(fname);
}
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
new_state = NULL;
if (!contents) {
r11468@Kushana: nickm | 2006-12-07 14:56:57 -0500 Revise logic used to flush state to disk. Now, we try to batch non-urgent changes so that we do not do too many writes, and we save very-non-urgent changes every once in a rare while, and we never save more than once per second. svn:r9047
2006-12-07 21:11:30 +01:00
global_state->next_write = 0;
or_state_save(time(NULL));
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
}
r = 0;
r8572@Kushana: nickm | 2006-08-25 16:35:49 -0400 Fix for bug 308: When we have a state file we cannot parse, tell the user, and move it aside. svn:r8224
2006-08-25 23:01:56 +02:00
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
done:
tor_free(fname);
tor_free(contents);
if (new_state)
config_free(&state_format, new_state);
return r;
}
stop putting wacky values into state->lastwritten
2011-04-29 02:40:15 +02:00
/** Did the last time we tried to write the state file fail? If so, we
* should consider disabling such features as preemptive circuit generation
* to compute circuit-build-time. */
static int last_state_file_write_failed = 0;
/** Return whether the state file failed to write last time we tried. */
int
did_last_state_file_write_fail(void)
{
return last_state_file_write_failed;
}
Wait 60 minutes before retrying failed state save; bug2346
2011-01-10 22:44:42 +01:00
/** If writing the state to disk fails, try again after this many seconds. */
#define STATE_WRITE_RETRY_INTERVAL 3600
relays checkpoint their state file twice a day
2011-04-29 03:06:25 +02:00
/** If we're a relay, how often should we checkpoint our state file even
* if nothing else dirties it? This will checkpoint ongoing stats like
* bandwidth used, per-country user stats, etc. */
#define STATE_RELAY_CHECKPOINT_INTERVAL (12*60*60)
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Write the persistent state to disk. Return 0 for success, <0 on failure. */
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
int
r11468@Kushana: nickm | 2006-12-07 14:56:57 -0500 Revise logic used to flush state to disk. Now, we try to batch non-urgent changes so that we do not do too many writes, and we save very-non-urgent changes every once in a rare while, and we never save more than once per second. svn:r9047
2006-12-07 21:11:30 +01:00
or_state_save(time_t now)
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
{
char *state, *contents;
char tbuf[ISO_TIME_LEN+1];
char *fname;
save and validate TorVersion in the state file. svn:r5694
2006-01-02 05:14:52 +01:00
tor_assert(global_state);
r11468@Kushana: nickm | 2006-12-07 14:56:57 -0500 Revise logic used to flush state to disk. Now, we try to batch non-urgent changes so that we do not do too many writes, and we save very-non-urgent changes every once in a rare while, and we never save more than once per second. svn:r9047
2006-12-07 21:11:30 +01:00
if (global_state->next_write > now)
return 0;
/* Call everything else that might dirty the state even more, in order
* to avoid redundant writes. */
entry nodes are now entry guards. this is our last easy chance for a wholesale change. heave ho. svn:r5782
2006-01-10 23:42:44 +01:00
entry_guards_update_state(global_state);
Keep bandwidth history accross restarts/crashes svn:r5637
2005-12-24 00:56:42 +01:00
rep_hist_update_state(global_state);
Use our variable directly for timeout. Using CircuitBuildTimeout is prone to issues with SIGHUP, etc. Also, shuffle the circuit build times array after loading it in so that newer measurements don't replace chunks of similarly timed measurements.
2009-09-02 05:27:43 +02:00
circuit_build_times_update_state(&circ_times, global_state);
fix xiando's bug -- we were reporting stuff about hibernation even though it wasn't enabled. svn:r9055
2006-12-09 03:55:40 +01:00
if (accounting_is_enabled(get_options()))
accounting_run_housekeeping(now);
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
stop putting wacky values into state->lastwritten
2011-04-29 02:40:15 +02:00
global_state->LastWritten = now;
save and validate TorVersion in the state file. svn:r5694
2006-01-02 05:14:52 +01:00
tor_free(global_state->TorVersion);
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(&global_state->TorVersion, "Tor %s", get_version());
r12852@catbus: nickm | 2007-05-22 11:00:27 -0400 Use svn revisions consistently throughout all log messages. svn:r10291
2007-05-22 17:48:46 +02:00
Support for a defaults torrc file. This will mainly help distributors by giving a way to set system or package defaults that a user can override, and that a later package can replace. No promises about the particular future location or semantics for this: we will probably want to tweak it some before 0.2.3.x-rc The file is searched for in CONFDIR/torrc-defaults , which can be overridden with the "--defaults-torrc" option on the command line.
2011-11-28 04:25:52 +01:00
state = config_dump(&state_format, NULL, global_state, 1, 0);
stop putting wacky values into state->lastwritten
2011-04-29 02:40:15 +02:00
format_local_iso_time(tbuf, now);
Tweak users of snprintf to use asprintf where appropriate
2010-03-01 03:46:50 +01:00
tor_asprintf(&contents,
clarify our use of local time vs GMT svn:r9109
2006-12-15 00:39:14 +01:00
"# Tor state file last generated on %s local time\n"
"# Other times below are in GMT\n"
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
"# You *do not* need to edit this file.\n\n%s",
tbuf, state);
tor_free(state);
r15877@catbus: nickm | 2007-10-17 12:54:56 -0400 Make unverified-consensus get removed when it is accepted or rejected. Make a new get_datadir_fname*() set of functions to eliminate the common code of "get the options, get the datadir, append some stuff". svn:r12000
2007-10-17 18:55:44 +02:00
fname = get_datadir_fname("state");
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
if (write_str_to_file(fname, contents, 0)<0) {
Fix a heap overflow found by debuger, and make it harder to make that mistake again Our public key functions assumed that they were always writing into a large enough buffer. In one case, they weren't. (Incorporates fixes from sebastian)
2011-01-13 20:36:41 +01:00
log_warn(LD_FS, "Unable to write state to file \"%s\"; "
"will try again later", fname);
stop putting wacky values into state->lastwritten
2011-04-29 02:40:15 +02:00
last_state_file_write_failed = 1;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
tor_free(fname);
tor_free(contents);
Explain bug2346 fix better based on suggestions from arma
2011-01-12 18:37:42 +01:00
/* Try again after STATE_WRITE_RETRY_INTERVAL (or sooner, if the state
* changes sooner). */
Wait 60 minutes before retrying failed state save; bug2346
2011-01-10 22:44:42 +01:00
global_state->next_write = now + STATE_WRITE_RETRY_INTERVAL;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
return -1;
}
Bug 1296: Add option+logic to disable CBT learning. There are now four ways that CBT can be disabled: 1. Network-wide, with the cbtdisabled consensus param. 2. Via config, with "LearnCircuitBuildTimeout 0" 3. Via config, with "AuthoritativeDirectory 1" 4. Via a state file write failure.
2010-05-08 20:54:29 +02:00
stop putting wacky values into state->lastwritten
2011-04-29 02:40:15 +02:00
last_state_file_write_failed = 0;
more log conversions. whee. svn:r6003
2006-02-13 10:02:35 +01:00
log_info(LD_GENERAL, "Saved state to \"%s\"", fname);
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
tor_free(fname);
tor_free(contents);
relays checkpoint their state file twice a day
2011-04-29 03:06:25 +02:00
if (server_mode(get_options()))
global_state->next_write = now + STATE_RELAY_CHECKPOINT_INTERVAL;
else
global_state->next_write = TIME_MAX;
Possibly broken implementation of persistant state; handles helper nodes; does not handle accounting info yet. svn:r4680
2005-07-28 21:01:48 +02:00
return 0;
}
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
/** Return the config line for transport <b>transport</b> in the current state.
* Return NULL if there is no config line for <b>transport</b>. */
static config_line_t *
get_transport_in_state_by_name(const char *transport)
{
Improve how we access or_state_t. * Use get_or_state()->VirtualOption instead of relying on config_find_option(), STRUCT_VAR_P and voodoo.
2011-09-11 20:57:01 +02:00
or_state_t *or_state = get_or_state();
config_line_t *line;
Fix bug in get_transport_in_state_by_name() when using strcmpstart(). We now split the state lines into smartlists and compare the token properly. Not that efficient but it's surely correct.
2011-09-11 21:22:37 +02:00
config_line_t *ret = NULL;
smartlist_t *items = NULL;
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
Improve how we access or_state_t. * Use get_or_state()->VirtualOption instead of relying on config_find_option(), STRUCT_VAR_P and voodoo.
2011-09-11 20:57:01 +02:00
for (line = or_state->TransportProxies ; line ; line = line->next) {
tor_assert(!strcmp(line->key, "TransportProxy"));
Fix bug in get_transport_in_state_by_name() when using strcmpstart(). We now split the state lines into smartlists and compare the token properly. Not that efficient but it's surely correct.
2011-09-11 21:22:37 +02:00
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
items = smartlist_new();
Fix bug in get_transport_in_state_by_name() when using strcmpstart(). We now split the state lines into smartlists and compare the token properly. Not that efficient but it's surely correct.
2011-09-11 21:22:37 +02:00
smartlist_split_string(items, line->value, NULL,
SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, -1);
if (smartlist_len(items) != 2) /* broken state */
goto done;
if (!strcmp(smartlist_get(items, 0), transport)) {
ret = line;
goto done;
}
SMARTLIST_FOREACH(items, char*, s, tor_free(s));
smartlist_free(items);
items = NULL;
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
}
Fix bug in get_transport_in_state_by_name() when using strcmpstart(). We now split the state lines into smartlists and compare the token properly. Not that efficient but it's surely correct.
2011-09-11 21:22:37 +02:00
done:
if (items) {
SMARTLIST_FOREACH(items, char*, s, tor_free(s));
smartlist_free(items);
}
return ret;
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
}
/** Return string containing the address:port part of the
Spawn multiple protocols using a single managed proxy. If multiple torrc transport lines have the same argv, tor instructs a single managed proxy to launch multiple protocols.
2011-08-12 21:33:05 +02:00
* TransportProxy <b>line</b> for transport <b>transport</b>.
* If the line is corrupted, return NULL. */
Trivial fixes around the code. * C90-fy. * Remove ASN comments. * Don't smartlist_clear() before smartlist_free(). * Plug a mem. leak.
2011-09-11 23:33:31 +02:00
static const char *
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
get_transport_bindaddr(const char *line, const char *transport)
{
Be more defensive in get_transport_bindaddr(). Make sure that lines in get_transport_bindaddr() begin with the name of the transport and a space.
2011-09-11 21:33:02 +02:00
char *line_tmp = NULL;
if (strlen(line) < strlen(transport) + 2) {
goto broken_state;
} else {
/* line should start with the name of the transport and a space.
(for example, "obfs2 127.0.0.1:47245") */
tor_asprintf(&line_tmp, "%s ", transport);
if (strcmpstart(line, line_tmp))
goto broken_state;
tor_free(line_tmp);
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
return (line+strlen(transport)+1);
Be more defensive in get_transport_bindaddr(). Make sure that lines in get_transport_bindaddr() begin with the name of the transport and a space.
2011-09-11 21:33:02 +02:00
}
broken_state:
tor_free(line_tmp);
return NULL;
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
}
Server transports should be instructed to bind on INADDR_ANY by default.
2011-12-18 13:21:58 +01:00
/** Return a string containing the address:port that a proxy transport
* should bind on. The string is stored on the heap and must be freed
* by the caller of this function. */
char *
Improve names of some pluggable transport-related functions.
2012-01-13 15:44:30 +01:00
get_stored_bindaddr_for_server_transport(const char *transport)
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
{
Server transports should be instructed to bind on INADDR_ANY by default.
2011-12-18 13:21:58 +01:00
char *default_addrport = NULL;
const char *stored_bindaddr = NULL;
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
config_line_t *line = get_transport_in_state_by_name(transport);
Server transports should be instructed to bind on INADDR_ANY by default.
2011-12-18 13:21:58 +01:00
if (!line) /* Found no references in state for this transport. */
goto no_bindaddr_found;
stored_bindaddr = get_transport_bindaddr(line->value, transport);
if (stored_bindaddr) /* found stored bindaddr in state file. */
return tor_strdup(stored_bindaddr);
no_bindaddr_found:
/** If we didn't find references for this pluggable transport in the
state file, we should instruct the pluggable transport proxy to
listen on INADDR_ANY on a random ephemeral port. */
tor_asprintf(&default_addrport, "%s:%s", fmt_addr32(INADDR_ANY), "0");
return default_addrport;
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
}
Spawn multiple protocols using a single managed proxy. If multiple torrc transport lines have the same argv, tor instructs a single managed proxy to launch multiple protocols.
2011-08-12 21:33:05 +02:00
/** Save <b>transport</b> listening on <b>addr</b>:<b>port</b> to
state */
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
void
save_transport_to_state(const char *transport,
Improve the code a tad. * Use strcmpstart() instead of strcmp(x,y,strlen(y)). * Warn the user if the managed proxy failed to launch. * Improve function documentation. * Use smartlist_len() instead of n_unconfigured_proxies. * Split managed_proxy_destroy() to managed_proxy_destroy() and managed_proxy_destroy_with_transports(). * Constification.
2011-08-15 17:26:03 +02:00
const tor_addr_t *addr, uint16_t port)
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
{
or_state_t *state = get_or_state();
char *transport_addrport=NULL;
/** find where to write on the state */
config_line_t **next, *line;
/* see if this transport is already stored in state */
config_line_t *transport_line =
get_transport_in_state_by_name(transport);
Improve wording in some comments and log messages.
2011-09-23 17:50:56 +02:00
if (transport_line) { /* if transport already exists in state... */
const char *prev_bindaddr = /* get its addrport... */
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
get_transport_bindaddr(transport_line->value, transport);
tor_asprintf(&transport_addrport, "%s:%d", fmt_addr(addr), (int)port);
/* if transport in state has the same address as this one, life is good */
if (!strcmp(prev_bindaddr, transport_addrport)) {
Put some sense into our logging. Transform our logging severities to something more sensible. Remove sneaky printf()s.
2011-09-12 00:10:07 +02:00
log_info(LD_CONFIG, "Transport seems to have spawned on its usual "
Make check-spaces happy.
2011-09-11 23:34:36 +02:00
"address:port.");
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
goto done;
Improve wording in some comments and log messages.
2011-09-23 17:50:56 +02:00
} else { /* if addrport in state is different than the one we got */
Put some sense into our logging. Transform our logging severities to something more sensible. Remove sneaky printf()s.
2011-09-12 00:10:07 +02:00
log_info(LD_CONFIG, "Transport seems to have spawned on different "
Make check-spaces happy.
2011-09-11 23:34:36 +02:00
"address:port. Let's update the state file with the new "
"address:port");
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
tor_free(transport_line->value); /* free the old line */
tor_asprintf(&transport_line->value, "%s %s:%d", transport,
fmt_addr(addr),
(int) port); /* replace old addrport line with new line */
}
} else { /* never seen this one before; save it in state for next time */
Put some sense into our logging. Transform our logging severities to something more sensible. Remove sneaky printf()s.
2011-09-12 00:10:07 +02:00
log_info(LD_CONFIG, "It's the first time we see this transport. "
Server transport proxies should bind on the same port each time, if possible.
2011-08-07 18:05:40 +02:00
"Let's save its address:port");
next = &state->TransportProxies;
/* find the last TransportProxy line in the state and point 'next'
right after it */
line = state->TransportProxies;
while (line) {
next = &(line->next);
line = line->next;
}
/* allocate space for the new line and fill it in */
*next = line = tor_malloc_zero(sizeof(config_line_t));
line->key = tor_strdup("TransportProxy");
tor_asprintf(&line->value, "%s %s:%d", transport,
fmt_addr(addr), (int) port);
next = &(line->next);
}
if (!get_options()->AvoidDiskWrites)
or_state_mark_dirty(state, 0);
done:
tor_free(transport_addrport);
}
r15750@catbus: nickm | 2007-10-13 20:06:47 -0400 Eventually delete the obsolete cached-routers and cached-routers.new files, so they don't sit around on disk forever. svn:r11918
2007-10-14 02:13:06 +02:00
/** Given a file name check to see whether the file exists but has not been
* modified for a very long time. If so, remove it. */
void
remove_file_if_very_old(const char *fname, time_t now)
{
#define VERY_OLD_FILE_AGE (28*24*60*60)
struct stat st;
if (stat(fname, &st)==0 && st.st_mtime < now-VERY_OLD_FILE_AGE) {
char buf[ISO_TIME_LEN+1];
format_local_iso_time(buf, st.st_mtime);
log_notice(LD_GENERAL, "Obsolete file %s hasn't been modified since %s. "
"Removing it.", fname, buf);
unlink(fname);
}
}
Comments: cleanups and additions. svn:r6174
2006-03-17 06:50:41 +01:00
/** Helper to implement GETINFO functions about configuration variables (not
* their values). Given a "config/names" question, set *<b>answer</b> to a
* new string describing the supported configuration variables and their
* types. */
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
int
r11479@Kushana: nickm | 2006-12-07 23:38:54 -0500 Refactor GETINFO into a table-driven dispatch, as suggested by arma. My brain hurts. svn:r9052
2006-12-08 05:39:13 +01:00
getinfo_helper_config(control_connection_t *conn,
Make the controller act more usefully when GETINFO fails Right now it says "552 internal error" because there's no way for getinfo_helper_*() countries to specify an error message. This patch changes the getinfo_helper_*() interface, and makes most of the getinfo helpers give useful error messages in response to failures. This should prevent recurrences of bug 1699, where a missing GeoIPFile line in the torrc made GETINFO ip-to-county/* fail in a "not obvious how to fix" way.
2010-07-18 17:05:58 +02:00
const char *question, char **answer,
const char **errmsg)
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
{
r11479@Kushana: nickm | 2006-12-07 23:38:54 -0500 Refactor GETINFO into a table-driven dispatch, as suggested by arma. My brain hurts. svn:r9052
2006-12-08 05:39:13 +01:00
(void) conn;
Make the controller act more usefully when GETINFO fails Right now it says "552 internal error" because there's no way for getinfo_helper_*() countries to specify an error message. This patch changes the getinfo_helper_*() interface, and makes most of the getinfo helpers give useful error messages in response to failures. This should prevent recurrences of bug 1699, where a missing GeoIPFile line in the torrc made GETINFO ip-to-county/* fail in a "not obvious how to fix" way.
2010-07-18 17:05:58 +02:00
(void) errmsg;
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
if (!strcmp(question, "config/names")) {
Rename nonconformant identifiers. Fixes bug 4893. These changes are pure mechanical, and were generated with this perl script: /usr/bin/perl -w -i.bak -p s/crypto_pk_env_t/crypto_pk_t/g; s/crypto_dh_env_t/crypto_dh_t/g; s/crypto_cipher_env_t/crypto_cipher_t/g; s/crypto_digest_env_t/crypto_digest_t/g; s/aes_free_cipher/aes_cipher_free/g; s/crypto_free_cipher_env/crypto_cipher_free/g; s/crypto_free_digest_env/crypto_digest_free/g; s/crypto_free_pk_env/crypto_pk_free/g; s/_crypto_dh_env_get_dh/_crypto_dh_get_dh/g; s/_crypto_new_pk_env_rsa/_crypto_new_pk_from_rsa/g; s/_crypto_pk_env_get_evp_pkey/_crypto_pk_get_evp_pkey/g; s/_crypto_pk_env_get_rsa/_crypto_pk_get_rsa/g; s/crypto_new_cipher_env/crypto_cipher_new/g; s/crypto_new_digest_env/crypto_digest_new/g; s/crypto_new_digest256_env/crypto_digest256_new/g; s/crypto_new_pk_env/crypto_pk_new/g; s/crypto_create_crypto_env/crypto_cipher_new/g; s/connection_create_listener/connection_listener_new/g; s/smartlist_create/smartlist_new/g; s/transport_create/transport_new/g;
2012-01-18 21:53:30 +01:00
smartlist_t *sl = smartlist_new();
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
int i;
for (i = 0; _option_vars[i].name; ++i) {
Make the get_options() return const This lets us make a lot of other stuff const, allows the compiler to generate (slightly) better code, and will make me get slightly fewer patches from folks who stick mutable stuff into or_options_t. const: because not every input is an output!
2011-06-14 19:01:38 +02:00
const config_var_t *var = &_option_vars[i];
Remove online config descriptions. They weren't in sync with reality nor manpage, and only useful to a human who could simply have checked the manpage.
2009-12-19 21:25:25 +01:00
const char *type;
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
switch (var->type) {
Replace (Fascist)Firewall* with a new ReachableAddresses option that understands address policies. svn:r4751
2005-08-08 23:58:48 +02:00
case CONFIG_TYPE_STRING: type = "String"; break;
Clean up a macro/cpp interaction related to bug 707. Also, add a new "filename" config value type which is currently just a synonym for string, but which might be subject to expansion later. svn:r15305
2008-06-16 20:09:53 +02:00
case CONFIG_TYPE_FILENAME: type = "Filename"; break;
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
case CONFIG_TYPE_UINT: type = "Integer"; break;
"(Socks|Control|etc)Port auto" now tells Tor to open an arbitrary port This is the major part of the implementation for trac issue 3076.
2011-05-02 21:05:10 +02:00
case CONFIG_TYPE_PORT: type = "Port"; break;
Replace (Fascist)Firewall* with a new ReachableAddresses option that understands address policies. svn:r4751
2005-08-08 23:58:48 +02:00
case CONFIG_TYPE_INTERVAL: type = "TimeInterval"; break;
Make logging resolution configurable. Implements enhancement 1668.
2010-08-27 08:13:54 +02:00
case CONFIG_TYPE_MSEC_INTERVAL: type = "TimeMsecInterval"; break;
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
case CONFIG_TYPE_MEMUNIT: type = "DataSize"; break;
case CONFIG_TYPE_DOUBLE: type = "Float"; break;
case CONFIG_TYPE_BOOL: type = "Boolean"; break;
Add a new configuration type, "AUTOBOOL", to handle 1/0/auto types We only have one of these now, but I'm about to add a few more. Yes, I have already thought of the "Filenotfoundian logic" joke.
2010-11-08 19:34:40 +01:00
case CONFIG_TYPE_AUTOBOOL: type = "Boolean+Auto"; break;
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
case CONFIG_TYPE_ISOTIME: type = "Time"; break;
r17188@tombo: nickm | 2008-07-18 14:35:18 -0400 Add new ExcludeExitNodes option. Also add a new routerset type to handle Exclude[Exit]Nodes. It is optimized for O(1) membership tests, so as to make choosing a random router run in O(N_routers) time instead of in O(N_routers*N_Excluded_Routers). svn:r16061
2008-07-18 20:36:32 +02:00
case CONFIG_TYPE_ROUTERSET: type = "RouterList"; break;
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
case CONFIG_TYPE_CSV: type = "CommaList"; break;
case CONFIG_TYPE_LINELIST: type = "LineList"; break;
case CONFIG_TYPE_LINELIST_S: type = "Dependant"; break;
case CONFIG_TYPE_LINELIST_V: type = "Virtual"; break;
default:
case CONFIG_TYPE_OBSOLETE:
type = NULL; break;
}
if (!type)
continue;
Try to use smartlist_add_asprintf consistently (To ensure correctness, in every case, make sure that the temporary variable is deleted, renamed, or lowered in scope, so we can't have any bugs related to accidentally relying on the no-longer-filled variable.)
2012-01-11 19:44:10 +01:00
smartlist_add_asprintf(sl, "%s %s\n",var->name,type);
Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. svn:r4694
2005-08-04 21:56:41 +02:00
}
*answer = smartlist_join_strings(sl, "", 0, NULL);
SMARTLIST_FOREACH(sl, char *, c, tor_free(c));
smartlist_free(sl);
}
return 0;
}
Reference in New Issue Copy Permalink