Commit Graph

12651 Commits

Author SHA1 Message Date
Nick Mathewson
462185d180 Add a proposal-ideas document for crypto migration. 2010-12-13 23:39:54 -05:00
Nick Mathewson
2118028c50 start reformatting and editing the pluggable-transport proposal 2010-12-11 04:18:15 -05:00
Nick Mathewson
1fb3a60f54 convert draft pluggable transport proposal to plaintext 2010-12-10 14:34:26 -05:00
Nick Mathewson
4e9f9a4ee8 Merge remote branch 'origin/maint-0.2.2' 2010-12-07 11:37:14 -05:00
Nick Mathewson
f924fbf19f Merge branch 'bug2081_followup_022' into maint-0.2.2 2010-12-07 11:35:49 -05:00
Nick Mathewson
5efe6f04c6 Reject relay versions older than 0.2.0.26-rc
This was the first version to cache the correct directory information.

Fixes bug 2156.
2010-12-07 11:35:32 -05:00
Nick Mathewson
8f31a9da43 revise bug2081_followup changelog 2010-12-07 11:27:51 -05:00
Nick Mathewson
ed0eeed835 Merge branch 'isolated-streams'
Conflicts:
	doc/spec/proposals/000-index.txt
2010-12-07 11:02:10 -05:00
Nick Mathewson
a1e46c5393 Revise proposal 171 from start to finish
The big semantic change is to make the IsolateFoo options exist on a
per-client-port basis.
2010-12-07 11:00:10 -05:00
Nick Mathewson
3fc43debfb Merge remote branch 'origin/maint-0.2.2' 2010-12-06 12:07:37 -05:00
Nick Mathewson
dc2f10bd81 Fix a bug in calculating wakeup time on 64-bit machines.
If you had TIME_MAX > INT_MAX, and your "time_to_exhaust_bw =
accountingmax/expected_bandwidth_usage * 60" calculation managed to
overflow INT_MAX, then your time_to_consider value could underflow and
wind up being rediculously low or high.  "Low" was no problem;
negative values got caught by the (time_to_consider <= 0) check.
"High", however, would get you a wakeup time somewhere in the distant
future.

The fix is to check for time_to_exhaust_bw overflowing INT_MAX, not
TIME_MAX: We don't allow any accounting interval longer than a month,
so if time_to_exhaust_bw is significantly larger than 31*24*60*60, we
can just clip it.

This is a bugfix on 0.0.9pre6, when accounting was first introduced.
It fixes bug 2146, unless there are other causes there too.  The fix
is from boboper.  (I tweaked it slightly by removing an assignment
that boboper marked as dead, and lowering a variable that no longer
needed to be function-scoped.)
2010-12-06 12:01:32 -05:00
Nick Mathewson
bbadf07d16 Merge remote branch 'origin/maint-0.2.2' 2010-12-06 11:37:37 -05:00
Nick Mathewson
feffbce814 Add a missing ! to directory_fetches_from_authorities
The old logic would have us fetch from authorities if we were refusing
unknown exits and our exit policy was reject*.  Instead, we want to
fetch from authorities if we're refusing unknown exits and our exit
policy is _NOT_ reject*.

Fixed by boboper.  Fixes more of 2097.  Bugfix on 0.2.2.16-alpha.
2010-12-06 11:36:01 -05:00
Nick Mathewson
36d68d2301 Fetch missing certs to authenticate microdesc consensuses as needed
Fixes bug 2253; fix by rransom
2010-12-06 11:26:45 -05:00
Nick Mathewson
0893793f01 Merge remote branch 'origin/maint-0.2.2' 2010-12-03 13:45:10 -05:00
Nick Mathewson
c0f1517d87 Don't crash when accountingmax is set in non-server Tors
We use a hash of the identity key to seed a prng to tell when an
accounting period should end.  But thanks to the bug998 changes,
clients no longer have server-identity keys to use as a long-term seed
in accounting calculations.  In any case, their identity keys (as used
in TLS) were never never fixed.  So we can just set the wakeup time
from a random seed instead there.  Still open is whether everybody
should be random.

This patch fixes bug 2235, which was introduced in 0.2.2.18-alpha.

Diagnosed with help from boboper on irc.
2010-12-03 13:37:13 -05:00
Karsten Loesing
0bd884cd5b Make connection direction statistics compile again. 2010-12-03 17:37:13 +01:00
Karsten Loesing
8db10c6d2f Write bidirectional connection stats in single line.
There's no need to have a separate line conn-stats-end for the end date
and interval length with only a single line conn-bi-direct following.
2010-12-03 16:47:53 +01:00
Karsten Loesing
91fec693e0 Refactor conn stats and add unit tests. 2010-12-03 16:47:53 +01:00
Karsten Loesing
076a688d76 Add ChangeLog entry and man page entry. 2010-12-03 16:47:52 +01:00
Karsten Loesing
c3c6a6ed34 Document BIDI_* constants. 2010-12-03 16:47:52 +01:00
Karsten Loesing
f56529b4b2 Call the new config option ConnDirectionStatistics. 2010-12-03 16:47:52 +01:00
Karsten Loesing
5dfdf075ac Add metric on uni/bidirectional connection usage. 2010-12-03 16:47:51 +01:00
Nick Mathewson
520182579a Merge remote branch 'origin/maint-0.2.2' 2010-12-02 13:21:09 -05:00
Nick Mathewson
ee8f451bf1 Fix a harmless off-by-one error in counting controller argument lengths
Bugfix on 0.1.1.1-alpha; found by boboper.
2010-12-02 13:19:21 -05:00
Nick Mathewson
7ce8f5c621 Merge remote branch 'origin/maint-0.2.2' 2010-12-01 12:45:38 -05:00
Poet (Tim Sally)
31b69027d3 Specified grammars for orconn-status and entry-guards for Tor versions 0.1.2.2-alpha through 0.2.2.1-alpha with feature VERBOSE_NAMES turned off. 2010-12-01 12:44:48 -05:00
Poet (Tim Sally)
45c302efe0 Correct information about support for guards being called helper nodes.
The spec stated that support for the helper-nodes command would be removed
in 0.1.3.x, however support for this command is still in Tor. Updated the spec
to reflect this and added a node that the command is deprecated.
2010-12-01 12:44:48 -05:00
Poet (Tim Sally)
87158368dc Correct grammars to reflect that VERBOSE_NAMES is part of the protocol.
Several updates to grammars for events and GETINFO results.  All relate
to the fact that LongName has replaced ServerID since 0.2.2.1-alpha. See
documentation of VERBOSE_NAMES for more information. The following
grammars were changed:
  * orconn-status GETINFO result
  * entry-guards GETINFO result
  * Path general token
  * OR Connection status changed event
  * New descriptors available event
In all cases a note was added about when the old grammar applies.
2010-12-01 12:44:48 -05:00
Poet (Tim Sally)
608bad6e32 Several changes to the way tokens describing servers are documented.
(1) Made the wording of the comments consistant with token names.
Digest/Fingerprint and Name/Nickname were being used interchangeably.
Better to just use Fingerprint and Nickname becuase they are the names
of the tokens.

(2) Places the tokens currently in use before the tokens used in older
versions.  ServerSpec should be documented before ServerID.

(3) Added a note to the comments about ServerID that cross reference
the VERBOSE_FEATURE, allowing users to see when and why ServerID was
replaced with LongName.
2010-12-01 12:44:48 -05:00
Poet (Tim Sally)
cff180f8b0 Clarify description of FEATURES in control-spec.
(1) On by default is a bad way to describe features. Rather, they
are always on and should be viewed as a part of the control
protocol. Updated the wording in USEFEATURE to reflect this.

(2) Made descriptions of Tor versions consistant across all
features. There is the version in which a feature was introduced and
the version in which it became part of the protocol.

(3) Reworded the description of the VERBOSE_NAMES feature. The
previous wording describes the way things used to be first. Better to
lead with the current state of things and then describe how it differs
from old versions.
2010-12-01 12:44:48 -05:00
Steven Murdoch
d5127ebdd8 Fix connecting the stdin of tor-fw-helper to /dev/null
This wasn't working due to the parameters of dup2 being in the wrong order.
As a result, tor-fw-helper was inheriting the stdin of Tor.
2010-12-01 12:22:21 -05:00
Nick Mathewson
9908404f01 Merge remote branch 'sjmurdoch/cloexec' 2010-12-01 11:42:34 -05:00
Steven Murdoch
a961521a86 Check that FD_CLOEXEC is set before using it
I don't know if any platforms we care about don't have FD_CLOEXEC,
but this is what we do elsewhere
2010-12-01 15:43:17 +00:00
Steven Murdoch
367794ce02 Open connection to DNS resolvers with CLOEXEC flag set
NB: this will now register the socket with the socket accounting code
2010-12-01 15:40:11 +00:00
Steven Murdoch
786abbd54c Open log files with CLOEXEC flag set 2010-12-01 15:38:18 +00:00
Roger Dingledine
fa1a419622 Merge branch 'maint-0.2.2' 2010-12-01 00:12:33 -05:00
Roger Dingledine
b259c08318 Merge branch 'maint-0.2.1' into maint-0.2.2
Conflicts:

	doc/Makefile.am
2010-12-01 00:11:27 -05:00
Roger Dingledine
b071217d1f add 0.2.1.27 blurb and changelog to release notes 2010-12-01 00:07:03 -05:00
Nick Mathewson
3ed7505dc5 Merge remote branch 'origin/maint-0.2.2'
Conflicts:
	src/or/relay.c
2010-11-30 19:23:40 -05:00
Nick Mathewson
ad87d6172b fix changes file for 2210 2010-11-30 19:19:36 -05:00
Nick Mathewson
8fa4450fde Do not invoke tls_renegotiated_cb for non-bufferevent connections too early.
This is not the most beautiful fix for this problem, but it is the simplest.

Bugfix for 2205.  Thanks to Sebastian and Mashael for finding the
bug, and boboper/cypherpunks for figuring out why it was happening
and how to fix it, and for writing a few fixes.
2010-11-30 17:55:27 -05:00
Roger Dingledine
9f51e26715 fix memory leak introduced in 26e89742
found via valgrind
2010-11-30 16:52:19 -05:00
Nick Mathewson
25b0fd8868 Revise comment on 2210 a little; clean up n_streams/num_streams confusion
Also add a changes file
2010-11-29 16:07:27 -05:00
Nick Mathewson
89e97bdf94 Add wrappers function for libc random()
On windows, it's called something different.
2010-11-29 16:00:47 -05:00
Nick Mathewson
0eafe23ff3 Fix whitespace in patch for 2210 and backport to 0.2.2 2010-11-29 16:00:43 -05:00
Mashael AlSabah
12fa6e23cb Improve fairness when activating streams in circuit_resume_edge_reading_helper
The reason the "streams problem" occurs is due to the complicated
interaction between Tor's congestion control and libevent. At some point
during the experiment, the circuit window is exhausted, which blocks all
edge streams. When a circuit level sendme is received at Exit, it
resumes edge reading by looping over linked list of edge streams, and
calling connection_start_reading() to inform libevent to resume reading.
When the streams are activated again, Tor gets the chance to service the
first three streams activated before the circuit window is exhausted
again, which causes all streams to be blocked again. As an experiment,
we reversed the order in which the streams are activated, and indeed the
first three streams, rather than the last three, got service, while the
others starved.

 Our solution is to change the order in which streams are activated. We
choose a random edge connection from the linked list, and then we
activate streams starting from that chosen stream. When we reach the end
of the list, then we continue from the head of the list until our chosen
stream (treating the linked list as a circular linked list). It would
probably be better to actually remember which streams have received
service recently, but this way is simple and effective.
2010-11-29 15:34:21 -05:00
Nick Mathewson
19019f26b0 Merge remote branch 'origin/maint-0.2.2' 2010-11-29 15:30:17 -05:00
Nick Mathewson
a5174b092e Merge branch 'exitstats' into maint-0.2.2 2010-11-29 15:28:22 -05:00
Nick Mathewson
a8a8e08220 comment karsten's bug2196 patch a little 2010-11-29 15:27:54 -05:00