Commit Graph

204 Commits

Author SHA1 Message Date
Sebastian Hahn
265e40b481 Raise libevent dependency to 2.0.10-stable or newer
Only some very ancient distributions don't ship with Libevent 2 anymore,
even the oldest supported Ubuntu LTS version has it. This allows us to
get rid of a lot of compat code.
2016-07-04 12:40:09 +02:00
Nick Mathewson
57699de005 Update the copyright year. 2016-02-27 18:48:19 +01:00
Sebastian Hahn
32220d38c0 Ensure worker threads actually exit when it is time
This includes a small refactoring to use a new enum (workqueue_reply_t)
for the return values instead of just ints.
2015-08-21 10:36:53 -04:00
Nick Mathewson
d4c1716263 add an explanatory comment 2015-03-12 11:28:18 -04:00
Nick Mathewson
648af0438d Also, add an assertion to assign_onionskin_to_cpuworker 2015-03-12 11:15:50 -04:00
Nick Mathewson
66c8180207 Fix crash bug when calling cpuworkers_rotate_keyinfo on a client.
Fixes bug 15245; bugfix on 0.2.6.3-alpha. Thanks to anonym for reporting!
2015-03-12 11:14:39 -04:00
Nick Mathewson
5d2a23397a Fix a few coverity "Use after NULL check" warnings
Also remove the unit test mocks that allowed get_options() to be
NULL; that's an invariant violation for get_options().
2015-02-16 15:40:15 -05:00
Sebastian Hahn
3bcdb26267 Call cpu_init if we change to being a relay
The issue is that we use the cpuworker system with relays only, so if we
start up as a client and transition to being a relay later, we'll be
sad.

This fixes bug 14901; not in any released version of Tor.
2015-02-15 11:49:19 +01:00
Sebastian Hahn
9667b2b88a Add some comments to or_circuit_t.workqueue_entry
These were suggested by dgoulet, thanks!
2015-02-09 16:13:08 +01:00
Sebastian Hahn
7337510090 Avoid use-after-free of circ belonging to cancelled job
This fixes a bug where we decide to free the circuit because it isn't on
any workqueue anymore, and then the job finishes and the circuit gets
freed again.

Fixes bug #14815, not in any released version of Tor.
2015-02-09 16:12:47 +01:00
Nick Mathewson
2274221557 Fix a work-counting bug introduced by the workqueue merge
David Goulet finds that when he runs a busy relay for a while with the
latest version of the git code, the number of onionskins handled
slowly dwindles to zero, with total_pending_tasks wedged at its
maximum value.

I conjecture this is because the total_pending_tasks variable isn't
decremented when we successfully cancel a job.  Fixed that.

Fixes bug 14741; bugfix not on any released version of tor.
2015-02-05 12:17:08 -05:00
Nick Mathewson
a598d0f575 Bail early in cpuworker_onion_handshake_replyfn if the circuit is marked 2015-01-26 10:19:07 -05:00
Nick Mathewson
034e2788f8 whitespace fixes 2015-01-23 11:18:28 -05:00
David Goulet
84f5cb749d Fix: remove whitespace and update a comment in cpuworker.c
Signed-off-by: David Goulet <dgoulet@ev0ke.net>
2015-01-21 14:31:02 -05:00
Nick Mathewson
a52e549124 Update workqueue implementation to use a single queue for the work
Previously I used one queue per worker; now I use one queue for
everyone.  The "broadcast" code is gone, replaced with an idempotent
'update' operation.
2015-01-15 11:05:22 -05:00
Nick Mathewson
fb5ebfb507 Avoid chan/circ linear lookups for requests
The solution I took is to not free a circuit with a pending
uncancellable work item, but rather to set its magic number to a
sentinel value.  When we get a work item, we check whether the circuit
has that magic sentinel, and if so, we free it rather than processing
the reply.
2015-01-14 11:31:12 -05:00
Nick Mathewson
6c9c54e7fa Remove if (1) indentation in cpuworker.c
To avoid having diffs turn out too big, I had replaced some unneeded
ifs and fors with if (1), so that the indentation would still work out
right.  Now I might as well clean those up.
2015-01-14 11:28:26 -05:00
Nick Mathewson
1e896214e7 Refactor cpuworker to use workqueue/threadpool code. 2015-01-14 11:23:34 -05:00
Nick Mathewson
f54e54b0b4 Bump copyright dates to 2015, in case someday this matters. 2015-01-02 14:27:39 -05:00
Nick Mathewson
bbd8d07167 Apply new calloc coccinelle patch 2014-11-02 11:56:02 -05:00
Nick Mathewson
fcdcb377a4 Add another year to our copyright dates.
Because in 95 years, we or our successors will surely care about
enforcing the BSD license terms on this code.  Right?
2014-10-28 15:30:16 -04:00
Nick Mathewson
2bfd92d0d1 Apply coccinelle script to replace malloc(a*b)->calloc(a,b) 2014-08-13 10:39:56 -04:00
Nick Mathewson
58f4200789 Thread support is now required
Long ago we supported systems where there was no support for
threads, or where the threading library was broken. We shouldn't
have do that any more: on every OS that matters, threads exist, and
the OS supports running threads across multiple CPUs.

This resolves tickets 9495 and 12439.  It's a prerequisite to making
our workqueue code work better, since sensible workqueue
implementations don't split across multiple processes.
2014-06-20 10:20:10 -04:00
Nick Mathewson
b6c8a14bf3 Merge remote-tracking branch 'public/bug4345a_024' 2014-05-01 12:13:07 -04:00
Nick Mathewson
78bc814c04 scan-build: in cpuworker, initialize tv_start
scan-build doesn't realize that a request can't be timed at the end
unless it's timed at the start, and so it's not possible for us to
be subtracting start from end without start being set.
Nevertheless, let's not confuse it.
2014-04-18 21:12:45 -04:00
Nick Mathewson
6ad7f3417c Renamed "onionskins_completed" to "onionskins_assigned"
This improves the accuracy of the function/variable names.
2014-03-27 22:57:53 -04:00
Nick Mathewson
85b46d57bc Check spawn_func() return value
If we don't, we can wind up with a wedged cpuworker, and write to it
for ages and ages.

Found by skruffy.  This was a bug in 2dda97e8fd, a.k.a. svn
revision 402. It's been there since we have been using cpuworkers.
2014-01-17 12:04:53 -05:00
Roger Dingledine
2c877d2da4 collect and log statistics about onionskins received/processed
we skip onionskins that came from non-relays, so we're less likely to
run into privacy troubles.

starts to implement ticket 9658.
2013-09-05 01:44:52 -04:00
Nick Mathewson
63b67577d6 Check return values from fcntl and setsockopt
(Based on a patch from flupzor; bug #8206)
2013-03-18 14:28:38 -04:00
Nick Mathewson
339df5df08 Fix 8447: use %u to format circid_t.
Now that circid_t is 4 bytes long, the default integer promotions will
leave it alone when sizeof(int) == 4, which will leave us formatting an
unsigned as an int.  That's technically undefined behavior.

Fixes bug 8447 on bfffc1f0fc.  Bug not
in any released Tor.
2013-03-10 19:52:06 -04:00
Nick Mathewson
d6634001c9 Merge remote-tracking branch 'public/wide_circ_ids'
Conflicts:
	src/or/channel.h
	src/or/connection_or.c
	src/or/cpuworker.c
2013-02-15 16:23:43 -05:00
Nick Mathewson
719940df2b Fix a nigh-impossible overflow in cpuworker.c
When we compute the estimated microseconds we need to handle our
pending onionskins, we could (in principle) overflow a uint32_t if
we ever had 4 million pending onionskins before we had any data
about how onionskins take.  Nevertheless, let's compute it properly.

Fixes bug 8210; bugfix on 0.2.4.10. Found by coverity; this is CID
980651.
2013-02-11 16:46:38 -05:00
Nick Mathewson
8cdd8b8353 Fix numerous problems with Tor's weak RNG.
We need a weak RNG in a couple of places where the strong RNG is
both needless and too slow.  We had been using the weak RNG from our
platform's libc implementation, but that was problematic (because
many platforms have exceptionally horrible weak RNGs -- like, ones
that only return values between 0 and SHORT_MAX) and because we were
using it in a way that was wrong for LCG-based weak RNGs.  (We were
counting on the low bits of the LCG output to be as random as the
high ones, which isn't true.)

This patch adds a separate type for a weak RNG, adds an LCG
implementation for it, and uses that exclusively where we had been
using the platform weak RNG.
2013-02-08 16:28:05 -05:00
Andrea Shepard
dfbd19df41 Merge branch 'time_based_onionqueue_v2' of ssh://git-rw.torproject.org/nickm/tor 2013-01-24 08:10:12 -08:00
Nick Mathewson
4da083db3b Update the copyright date to 201. 2013-01-16 01:54:56 -05:00
Nick Mathewson
677d18278e Better handling (I think) for onionskin timing w jumpy clocks
The fix: Instead of clipping huge/negative times, ignore them as
probably invalid.
2013-01-03 13:26:59 -05:00
Nick Mathewson
30e139389b Record and report the overhead of how we handle onionskins. 2013-01-03 13:20:20 -05:00
Nick Mathewson
b0b3c14c11 Eliminate MaxOnionsPending; replace it with MaxOnionQueueDelay
The right way to set "MaxOnionsPending" was to adjust it until the
processing delay was appropriate.  So instead, let's measure how long
it takes to process onionskins (sampling them once we have a big
number), and then limit the queue based on its expected time to
finish.

This change is extra-necessary for ntor, since there is no longer a
reasonable way to set MaxOnionsPending without knowing what mix of
onionskins you'll get.

This patch also reserves 1/3 of the onionskin spots for ntor
handshakes, on the theory that TAP handshakes shouldn't be allowed to
starve their speedier cousins.  We can change this later if need be.

Resolves 7291.
2013-01-03 13:03:41 -05:00
Nick Mathewson
94cb7bd24d Complete all DOCDOC entries from the ntor branch 2013-01-03 11:29:48 -05:00
Nick Mathewson
115e8fe9a5 Use created_cell_format where appropriate 2013-01-03 11:29:47 -05:00
Nick Mathewson
2802ccaeb6 Teach cpuworker and others about create_cell_t and friends
The unit of work sent to a cpuworker is now a create_cell_t; its
response is now a created_cell_t.  Several of the things that call or
get called by this chain of logic now take create_cell_t or
created_cell_t too.

Since all cpuworkers are forked or spawned by Tor, they don't need a
stable wire protocol, so we can just send structs.  This saves us some
insanity, and helps p
2013-01-03 11:29:46 -05:00
Nick Mathewson
f58d4dfcd6 Massive refactoring of the various handshake types
The three handshake types are now accessed from a unified interface;
their state is abstracted from the rest of the cpath state, and so on.
2013-01-03 11:29:46 -05:00
Nick Mathewson
f7e590df05 Split onion.[ch] into onion{,_fast,_tap}.[ch]
I'm going to want a generic "onionskin" type and set of wrappers, and
for that, it will be helpful to isolate the different circuit creation
handshakes.  Now the original handshake is in onion_tap.[ch], the
CREATE_FAST handshake is in onion_fast.[ch], and onion.[ch] now
handles the onion queue.

This commit does nothing but move code and adjust header files.
2013-01-02 14:11:14 -05:00
Nick Mathewson
bfffc1f0fc Allow a v4 link protocol for 4-byte circuit IDs.
Implements proposal 214.

Needs testing.
2012-11-06 21:23:46 -05:00
Andrea Shepard
e709fe320a Use U64_FORMAT/U64_PRINTF_ARG rather than %lu for channel_t 2012-10-08 19:48:06 -07:00
Andrea Shepard
77dac97354 Use channel_t in cpuworker.c
Note: this is a squashed commit; see branch bug6465_rebased_v2 of user/andrea/tor.git for full history of the following 2 commits:

Use channel_t in cpuworker.c
Fix bug in channel_t usage in cpuworker.c that was killing relaying on channel_t-ized Tor.  The tags passed to the worker now have a channel ID, not a connection ID.
2012-10-08 03:06:09 -07:00
Nick Mathewson
7faf115dff Change all SMARTLIST_FOREACH loops of >=10 lines to use BEGIN/END
The SMARTLIST_FOREACH macro is more convenient than BEGIN/END when
you have a nice short loop body, but using it for long bodies makes
your preprocessor tell the compiler that all the code is on the same
line.  That causes grief, since compiler warnings and debugger lines
will all refer to that one line.

So, here's a new style rule: SMARTLIST_FOREACH blocks need to be
short.
2012-07-17 10:34:08 -04:00
Nick Mathewson
0fa107a6aa Update copyright dates to 2012; add a few missing copyright statements 2012-06-04 20:58:17 -04:00
Nick Mathewson
26e789fbfd 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 15:53:30 -05:00
Nick Mathewson
6e8c2a3e46 Use SOCKET_OK macros in even more places
Add a TOR_INVALID_SOCKET macro to wrap -1/INVALID_SOCKET.

Partial work for bug4533.
2012-01-17 16:35:07 -05:00