]> git.ipfire.org Git - thirdparty/tor.git/log
thirdparty/tor.git
15 years agoDon't crash when accountingmax is set in non-server Tors
Nick Mathewson [Fri, 3 Dec 2010 18:37:13 +0000 (13:37 -0500)] 
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.

15 years agoFix a harmless off-by-one error in counting controller argument lengths
Nick Mathewson [Thu, 2 Dec 2010 18:19:21 +0000 (13:19 -0500)] 
Fix a harmless off-by-one error in counting controller argument lengths

Bugfix on 0.1.1.1-alpha; found by boboper.

15 years agoSpecified grammars for orconn-status and entry-guards for Tor versions 0.1.2.2-alpha...
Poet (Tim Sally) [Wed, 1 Dec 2010 17:19:26 +0000 (11:19 -0600)] 
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.

15 years agoCorrect information about support for guards being called helper nodes.
Poet (Tim Sally) [Wed, 1 Dec 2010 11:26:05 +0000 (05:26 -0600)] 
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.

15 years agoCorrect grammars to reflect that VERBOSE_NAMES is part of the protocol.
Poet (Tim Sally) [Wed, 1 Dec 2010 11:25:17 +0000 (05:25 -0600)] 
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.

15 years agoSeveral changes to the way tokens describing servers are documented.
Poet (Tim Sally) [Wed, 1 Dec 2010 11:24:33 +0000 (05:24 -0600)] 
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.

15 years agoClarify description of FEATURES in control-spec.
Poet (Tim Sally) [Wed, 1 Dec 2010 11:21:34 +0000 (05:21 -0600)] 
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.

15 years agoMerge branch 'maint-0.2.1' into maint-0.2.2
Roger Dingledine [Wed, 1 Dec 2010 05:11:27 +0000 (00:11 -0500)] 
Merge branch 'maint-0.2.1' into maint-0.2.2

Conflicts:

doc/Makefile.am

15 years agoadd 0.2.1.27 blurb and changelog to release notes
Roger Dingledine [Wed, 1 Dec 2010 05:07:03 +0000 (00:07 -0500)] 
add 0.2.1.27 blurb and changelog to release notes

15 years agofix changes file for 2210
Nick Mathewson [Wed, 1 Dec 2010 00:19:36 +0000 (19:19 -0500)] 
fix changes file for 2210

15 years agoRevise comment on 2210 a little; clean up n_streams/num_streams confusion
Nick Mathewson [Mon, 29 Nov 2010 20:59:59 +0000 (15:59 -0500)] 
Revise comment on 2210 a little; clean up n_streams/num_streams confusion

Also add a changes file

15 years agoAdd wrappers function for libc random()
Nick Mathewson [Mon, 29 Nov 2010 20:53:33 +0000 (15:53 -0500)] 
Add wrappers function for libc random()

On windows, it's called something different.

15 years agoFix whitespace in patch for 2210 and backport to 0.2.2
Nick Mathewson [Mon, 29 Nov 2010 20:53:12 +0000 (15:53 -0500)] 
Fix whitespace in patch for 2210 and backport to 0.2.2

15 years agoImprove fairness when activating streams in circuit_resume_edge_reading_helper
Mashael AlSabah [Mon, 29 Nov 2010 20:34:21 +0000 (15:34 -0500)] 
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.

15 years agoMerge branch 'exitstats' into maint-0.2.2
Nick Mathewson [Mon, 29 Nov 2010 20:28:22 +0000 (15:28 -0500)] 
Merge branch 'exitstats' into maint-0.2.2

15 years agocomment karsten's bug2196 patch a little
Nick Mathewson [Mon, 29 Nov 2010 20:27:54 +0000 (15:27 -0500)] 
comment karsten's bug2196 patch a little

15 years agoMerge remote branch 'rransom/fix2195-fix' into maint-0.2.2
Nick Mathewson [Wed, 24 Nov 2010 22:02:22 +0000 (17:02 -0500)] 
Merge remote branch 'rransom/fix2195-fix' into maint-0.2.2

15 years agoAvoid dereferencing NULL if a bridge fails to build an ei descriptor.
Robert Ransom [Wed, 24 Nov 2010 15:50:29 +0000 (07:50 -0800)] 
Avoid dereferencing NULL if a bridge fails to build an ei descriptor.

Reported by an anonymous commenter on Trac.

15 years agoReport only the top 10 ports in exit-port stats.
Karsten Loesing [Tue, 23 Nov 2010 20:09:12 +0000 (21:09 +0100)] 
Report only the top 10 ports in exit-port stats.

15 years agoFix compilation with mingw and OpenSSL 0.9.8m+
mingw-san [Mon, 26 Jul 2010 19:05:11 +0000 (15:05 -0400)] 
Fix compilation with mingw and OpenSSL 0.9.8m+

15 years agoUse S_CASE for ehostunreach, not E_CASE. Partial backport of 69deb22f. Fixes 0.2...
Nick Mathewson [Tue, 23 Nov 2010 17:46:41 +0000 (12:46 -0500)] 
Use S_CASE for ehostunreach, not E_CASE. Partial backport of 69deb22f. Fixes 0.2.1 compilation on windows

15 years agostop shipping doc/img and doc/website in the tarball
Roger Dingledine [Tue, 23 Nov 2010 05:03:50 +0000 (00:03 -0500)] 
stop shipping doc/img and doc/website in the tarball

15 years agopick a more accurate release date tor-0.2.2.19-alpha
Roger Dingledine [Tue, 23 Nov 2010 03:44:20 +0000 (22:44 -0500)] 
pick a more accurate release date

15 years agobump to 0.2.2.19-alpha
Roger Dingledine [Sun, 21 Nov 2010 23:00:12 +0000 (18:00 -0500)] 
bump to 0.2.2.19-alpha

15 years agoprepare for our emergency openssl fix release
Roger Dingledine [Sun, 21 Nov 2010 21:59:42 +0000 (16:59 -0500)] 
prepare for our emergency openssl fix release

15 years agoMerge remote branch 'origin/maint-0.2.1' into maint-0.2.2
Nick Mathewson [Sun, 21 Nov 2010 19:33:11 +0000 (14:33 -0500)] 
Merge remote branch 'origin/maint-0.2.1' into maint-0.2.2

15 years agoMerge branch 'fix2204' into maint-0.2.1
Nick Mathewson [Sun, 21 Nov 2010 19:28:38 +0000 (14:28 -0500)] 
Merge branch 'fix2204' into maint-0.2.1

15 years agoDo not set the hostname TLS extension server-side; only client-side
Nick Mathewson [Sun, 21 Nov 2010 03:21:50 +0000 (22:21 -0500)] 
Do not set the hostname TLS extension server-side; only client-side

This may fix bug 2204, and resolve the incompatibility with openssl
0.9.8p/1.0.0b.

15 years agoFix a unit test broken by fix for 2195
Nick Mathewson [Fri, 19 Nov 2010 21:50:14 +0000 (16:50 -0500)] 
Fix a unit test broken by fix for 2195

15 years agoMerge branch 'fix2183', remote branch 'rransom/fix2195-v2' into maint-0.2.2
Nick Mathewson [Fri, 19 Nov 2010 21:48:13 +0000 (16:48 -0500)] 
Merge branch 'fix2183', remote branch 'rransom/fix2195-v2' into maint-0.2.2

15 years agoAdd changes file for bug2195.
Robert Ransom [Fri, 19 Nov 2010 21:46:09 +0000 (13:46 -0800)] 
Add changes file for bug2195.

15 years agoFix logic error in router_dump_router_to_string.
Robert Ransom [Fri, 19 Nov 2010 21:23:45 +0000 (13:23 -0800)] 
Fix logic error in router_dump_router_to_string.

Spotted by Nick Mathewson.

15 years agoAdd comments to some of the bug2183 fix code
Nick Mathewson [Fri, 19 Nov 2010 21:13:11 +0000 (16:13 -0500)] 
Add comments to some of the bug2183 fix code

15 years agoGenerate a router descriptor even if generating an extra-info descriptor fails.
Robert Ransom [Mon, 15 Nov 2010 09:36:08 +0000 (01:36 -0800)] 
Generate a router descriptor even if generating an extra-info descriptor fails.

Fixes bug #2195.

15 years agoDo not emit an extra-info-digest descriptor line if the digest is zero.
Robert Ransom [Mon, 15 Nov 2010 06:17:32 +0000 (22:17 -0800)] 
Do not emit an extra-info-digest descriptor line if the digest is zero.

15 years agoTweak the bugfix for 2183 a bit more.
Karsten Loesing [Wed, 17 Nov 2010 09:26:34 +0000 (10:26 +0100)] 
Tweak the bugfix for 2183 a bit more.

15 years agonew development version
Roger Dingledine [Tue, 16 Nov 2010 19:45:40 +0000 (14:45 -0500)] 
new development version

15 years agofinal touchups tor-0.2.2.18-alpha
Roger Dingledine [Tue, 16 Nov 2010 06:12:32 +0000 (01:12 -0500)] 
final touchups

15 years agobump to 0.2.2.18-alpha
Roger Dingledine [Tue, 16 Nov 2010 05:20:49 +0000 (00:20 -0500)] 
bump to 0.2.2.18-alpha

15 years agowarn more about AllowSingleHopExits
Roger Dingledine [Tue, 16 Nov 2010 05:13:25 +0000 (00:13 -0500)] 
warn more about AllowSingleHopExits

15 years agomerge in more changelog entries
Roger Dingledine [Tue, 16 Nov 2010 04:54:20 +0000 (23:54 -0500)] 
merge in more changelog entries

15 years agoMerge commit 'nickm/1776_redux_v1' into maint-0.2.2
Roger Dingledine [Tue, 16 Nov 2010 04:29:09 +0000 (23:29 -0500)] 
Merge commit 'nickm/1776_redux_v1' into maint-0.2.2

15 years agoClean up my 1776 fix a bit
Nick Mathewson [Tue, 16 Nov 2010 00:38:19 +0000 (19:38 -0500)] 
Clean up my 1776 fix a bit

Sebastian notes (and I think correctly) that one of our ||s should
have been an &&, which simplifies a boolean expression to decide
whether to replace bridges.  I'm also refactoring out the negation at
the start of the expression, to make it more readable.

15 years agoMove controller event for socks warning into log_unsafe_socks_warning
Nick Mathewson [Mon, 15 Nov 2010 20:41:21 +0000 (15:41 -0500)] 
Move controller event for socks warning into log_unsafe_socks_warning

15 years agoMerge branch 'bug2000_nm_022' into maint-0.2.2
Nick Mathewson [Mon, 15 Nov 2010 20:31:40 +0000 (15:31 -0500)] 
Merge branch 'bug2000_nm_022' into maint-0.2.2

15 years agochanges entry for nopublish removal in 5040c855d
Nick Mathewson [Mon, 15 Nov 2010 19:29:53 +0000 (14:29 -0500)] 
changes entry for nopublish removal in 5040c855d

15 years agoMake the coducmentation for SingleHop stuff slightly more dire
Nick Mathewson [Mon, 15 Nov 2010 19:09:32 +0000 (14:09 -0500)] 
Make the coducmentation for SingleHop stuff slightly more dire

15 years agoMerge remote branch 'sebastian/manpagefixups' into maint-0.2.2
Nick Mathewson [Mon, 15 Nov 2010 19:07:00 +0000 (14:07 -0500)] 
Merge remote branch 'sebastian/manpagefixups' into maint-0.2.2

15 years agoRate-limit unsafe socks warning
Sebastian Hahn [Mon, 4 Oct 2010 11:31:58 +0000 (13:31 +0200)] 
Rate-limit unsafe socks warning

Pick 5 seconds as the limit. 5 seconds is a compromise here between
making sure the user notices that the bad behaviour is (still) happening
and not spamming their log too much needlessly (the log message is
pretty long). We also keep warning every time if safesocks is
specified, because then the user presumably wants to hear about every
blocked instance.

(This is based on the original patch by Sebastian, then backported to
0.2.2 and with warnings split into their own function.)

15 years agoDon't use log_err for non-criticial warnings.
Karsten Loesing [Mon, 15 Nov 2010 12:39:53 +0000 (13:39 +0100)] 
Don't use log_err for non-criticial warnings.

15 years agoTry harder not to exceed the 50 KB extra-info descriptor limit.
Karsten Loesing [Sat, 13 Nov 2010 21:25:19 +0000 (22:25 +0100)] 
Try harder not to exceed the 50 KB extra-info descriptor limit.

Our checks that we don't exceed the 50 KB size limit of extra-info
descriptors apparently failed. This patch fixes these checks and reserves
another 250 bytes for appending the signature. Fixes bug 2183.

15 years agoAvoid perma-blocking the controller on bug in shrink_freelist
Nick Mathewson [Fri, 12 Nov 2010 18:05:58 +0000 (13:05 -0500)] 
Avoid perma-blocking the controller on bug in shrink_freelist

In all likelihood, this bug would make Tor assert, but if it doesn't,
let's not have two bugs.

15 years agoAdd changes file for bug1125
Nick Mathewson [Fri, 12 Nov 2010 18:03:18 +0000 (13:03 -0500)] 
Add changes file for bug1125

15 years agoMerge remote branch 'rransom/bug1125-022' into maint-0.2.2
Nick Mathewson [Fri, 12 Nov 2010 18:03:00 +0000 (13:03 -0500)] 
Merge remote branch 'rransom/bug1125-022' into maint-0.2.2

15 years agoMerge remote branch 'origin/maint-0.2.1' into maint-0.2.2
Nick Mathewson [Fri, 12 Nov 2010 18:01:41 +0000 (13:01 -0500)] 
Merge remote branch 'origin/maint-0.2.1' into maint-0.2.2

15 years agoAdd changes file for bug1125
Nick Mathewson [Fri, 12 Nov 2010 17:59:42 +0000 (12:59 -0500)] 
Add changes file for bug1125

15 years agoDisable logging to control port connections in buf_shrink_freelists.
Robert Ransom [Fri, 12 Nov 2010 11:07:09 +0000 (03:07 -0800)] 
Disable logging to control port connections in buf_shrink_freelists.

If buf_shrink_freelists calls log_warn for some reason, we don't want the log
call itself to throw buf_shrink_freelists further off the rails.

15 years agoMove the original log_info call out of the core of buf_shrink_freelists.
Robert Ransom [Fri, 12 Nov 2010 11:04:07 +0000 (03:04 -0800)] 
Move the original log_info call out of the core of buf_shrink_freelists.

Sending a log message to a control port can cause Tor to allocate a buffer,
thereby changing the length of the freelist behind buf_shrink_freelists's back,
thereby causing an assertion to fail.

Fixes bug #1125.

15 years agoDisable logging to control port connections in buf_shrink_freelists.
Robert Ransom [Fri, 12 Nov 2010 08:46:26 +0000 (00:46 -0800)] 
Disable logging to control port connections in buf_shrink_freelists.

If buf_shrink_freelists calls log_warn for some reason, we don't want the log
call itself to throw buf_shrink_freelists further off the rails.

15 years agoMove the original log_info call out of the core of buf_shrink_freelists.
Robert Ransom [Fri, 12 Nov 2010 08:21:03 +0000 (00:21 -0800)] 
Move the original log_info call out of the core of buf_shrink_freelists.

Sending a log message to a control port can cause Tor to allocate a buffer,
thereby changing the length of the freelist behind buf_shrink_freelists's back,
thereby causing an assertion to fail.

Fixes bug #1125.

15 years agoMerge remote branch 'sebastian/remove-osx-expert-package' into maint-0.2.2
Nick Mathewson [Thu, 11 Nov 2010 18:53:53 +0000 (13:53 -0500)] 
Merge remote branch 'sebastian/remove-osx-expert-package' into maint-0.2.2

15 years agoMerge branch 'maint-0.2.1' into maint-0.2.2
Roger Dingledine [Thu, 11 Nov 2010 17:12:17 +0000 (12:12 -0500)] 
Merge branch 'maint-0.2.1' into maint-0.2.2

15 years agolet unpublished bridges learn their ip address too
Roger Dingledine [Thu, 11 Nov 2010 16:26:42 +0000 (11:26 -0500)] 
let unpublished bridges learn their ip address too

15 years agoMerge remote branch 'origin/maint-0.2.1' into maint-0.2.2
Nick Mathewson [Wed, 10 Nov 2010 21:06:43 +0000 (16:06 -0500)] 
Merge remote branch 'origin/maint-0.2.1' into maint-0.2.2

15 years agoEnforce multiplicity rules when parsing annotations.
Nick Mathewson [Tue, 2 Nov 2010 15:49:58 +0000 (11:49 -0400)] 
Enforce multiplicity rules when parsing annotations.

We would never actually enforce multiplicity rules when parsing
annotations, since the counts array never got entries added to it for
annotations in the token list that got added by earlier calls to
tokenize_string.

Found by piebeer.

15 years agoFix a bug where seting allow_annotations==0 only ignores annotations, but does not...
Nick Mathewson [Tue, 2 Nov 2010 15:45:14 +0000 (11:45 -0400)] 
Fix a bug where seting allow_annotations==0 only ignores annotations, but does not block them

15 years agoBulletproof the routerlist manipulation functions to handle reinserting the same...
Nick Mathewson [Wed, 10 Nov 2010 19:55:00 +0000 (14:55 -0500)] 
Bulletproof the routerlist manipulation functions to handle reinserting the same descriptor

15 years agoDocument hs dht min uptime change in privnets
Sebastian Hahn [Mon, 8 Nov 2010 07:34:03 +0000 (08:34 +0100)] 
Document hs dht min uptime change in privnets

In f87c6f100d2faf8d we made it so that MinUptimeHidServDirectoryV2
defaults to 0 if TestingTorNetwork is set. Add a manpage entry and a
changes file.

15 years agoDocument the AllowSingleHopExits option
Sebastian Hahn [Thu, 4 Nov 2010 02:10:42 +0000 (03:10 +0100)] 
Document the AllowSingleHopExits option

15 years agoImprove documentation for RelayBandwdith* options
Sebastian Hahn [Fri, 22 Oct 2010 21:37:54 +0000 (23:37 +0200)] 
Improve documentation for RelayBandwdith* options

Jesse Adelman noticed that we don't currently describe what
happens very well.

15 years agoAdd V3BandwidthsFile to the manpage
Sebastian Hahn [Sat, 30 Oct 2010 09:04:21 +0000 (11:04 +0200)] 
Add V3BandwidthsFile to the manpage

15 years agoBreak NoPublish support
Sebastian Hahn [Sat, 30 Oct 2010 04:38:50 +0000 (06:38 +0200)] 
Break NoPublish support

15 years agoChange Natd into NATD in our options.
Sebastian Hahn [Sat, 30 Oct 2010 04:08:47 +0000 (06:08 +0200)] 
Change Natd into NATD in our options.

Breaking this out of the last commit because this might be more
controversial.

15 years agoSynx manpage and source wrt option capitalization
Sebastian Hahn [Fri, 29 Oct 2010 17:41:24 +0000 (19:41 +0200)] 
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).

15 years agoComment out the (unused) RunTesting option
Sebastian Hahn [Fri, 29 Oct 2010 22:03:31 +0000 (00:03 +0200)] 
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.

15 years agoFix a typo in the manpage
Sebastian Hahn [Fri, 29 Oct 2010 22:03:10 +0000 (00:03 +0200)] 
Fix a typo in the manpage

15 years agoRemove the torrc.complete file.
Sebastian Hahn [Fri, 29 Oct 2010 21:04:21 +0000 (23:04 +0200)] 
Remove the torrc.complete file.

We haven't been keeping it up to date, and the manpage exists as a
replacement for users who want an overview of all available options.

15 years agoChange checkOptionDocs.pl to be case sensitive
Sebastian Hahn [Fri, 29 Oct 2010 22:19:59 +0000 (00:19 +0200)] 
Change checkOptionDocs.pl to be case sensitive

15 years agoRemove everything related to os x expert package
Sebastian Hahn [Mon, 11 Oct 2010 16:47:14 +0000 (18:47 +0200)] 
Remove everything related to os x expert package

We decided to no longer ship expert packages for OS X because they're a
lot of trouble to keep maintained and confuse users. For those who want
a tor on OS X without Vidalia, macports is a fine option. Alternatively,
building from source is easy, too.

The polipo stuff that is still required for the Vidalia bundle build can
now be found in the torbrowser repository,
git://git.torproject.org/torbrowser.git.

15 years agofold the changes files into a changelog
Roger Dingledine [Wed, 10 Nov 2010 02:50:30 +0000 (21:50 -0500)] 
fold the changes files into a changelog

15 years agomove to the november 1 maxmind geoip db
Roger Dingledine [Sat, 6 Nov 2010 21:41:36 +0000 (17:41 -0400)] 
move to the november 1 maxmind geoip db

15 years agoFix the assert in bug 1776
Nick Mathewson [Tue, 2 Nov 2010 15:20:09 +0000 (11:20 -0400)] 
Fix the assert in bug 1776

In the case where old_router == NULL but sdmap has an entry for the
router, we can currently safely infer that the old_router was not a
bridge.  Add an assert to ensure that this remains true, and fix the
logic not to die with the tor_assert(old_router) call.

15 years agoRemove delay to become HSDir in privnets
Sebastian Hahn [Tue, 26 Oct 2010 16:37:57 +0000 (18:37 +0200)] 
Remove delay to become HSDir in privnets

15 years agoProperly refcount client_identity_key
Sebastian Hahn [Tue, 26 Oct 2010 16:22:04 +0000 (18:22 +0200)] 
Properly refcount client_identity_key

In a2bb0bf we started using a separate client identity key. When we are
in "public server mode" (that means not a bridge) we will use the same
key. Reusing the key without doing the proper refcounting leads to a
segfault on cleanup during shutdown. Fix that.

Also introduce an assert that triggers if our refcount falls below 0.
That should never happen.

15 years agoFix a whitespace error
Nick Mathewson [Thu, 21 Oct 2010 20:12:04 +0000 (16:12 -0400)] 
Fix a whitespace error

15 years agoMerge branch 'bug988-nm' into maint-0.2.2
Nick Mathewson [Thu, 21 Oct 2010 20:11:02 +0000 (16:11 -0400)] 
Merge branch 'bug988-nm' into maint-0.2.2

15 years agoAdd some asserts to get_{tlsclient|server}_identity_key
Nick Mathewson [Thu, 21 Oct 2010 17:54:12 +0000 (13:54 -0400)] 
Add some asserts to get_{tlsclient|server}_identity_key

We now require that:
  - Only actual servers should ever call get_server_identity_key
  - If you're being a client or bridge, the client and server keys should
    differ.
  - If you're being a public relay, the client and server keys
    should be the same.

15 years agoRename get_client_identity_key to get_tlsclient_identity_key
Nick Mathewson [Thu, 21 Oct 2010 17:53:31 +0000 (13:53 -0400)] 
Rename get_client_identity_key to get_tlsclient_identity_key

15 years agoFix a remaining bug in Robert's bug1859 fix.
Nick Mathewson [Thu, 21 Oct 2010 15:08:15 +0000 (11:08 -0400)] 
Fix a remaining bug in Robert's bug1859 fix.

When intro->extend_info is created for an introduction point, it
only starts out with a nickname, not necessarily an identity digest.
Thus, doing router_get_by_digest isn't necessarily safe.

15 years agoMerge remote branch 'hoganrobert/bug1859' into maint-0.2.2
Nick Mathewson [Thu, 21 Oct 2010 15:01:12 +0000 (11:01 -0400)] 
Merge remote branch 'hoganrobert/bug1859' into maint-0.2.2

15 years agoMerge remote branch 'sebastian/relay_early_rend' into maint-0.2.2
Nick Mathewson [Thu, 21 Oct 2010 14:49:44 +0000 (10:49 -0400)] 
Merge remote branch 'sebastian/relay_early_rend' into maint-0.2.2

15 years agoFix a logic error in 98aee84. Found by boboper
Nick Mathewson [Wed, 20 Oct 2010 18:40:09 +0000 (14:40 -0400)] 
Fix a logic error in 98aee84. Found by boboper

15 years agoAdd a ! to directory_caches_dir_info() to fix a logic error
Nick Mathewson [Wed, 20 Oct 2010 17:49:38 +0000 (13:49 -0400)] 
Add a ! to directory_caches_dir_info() to fix a logic error

We want to fetch directory info more aggressively if we need it to
refuseunknownexits.  Thus, we'll want it if our exit policy is _NOT_
reject *.

15 years agoMerge branch 'bug2097' into maint-0.2.2
Nick Mathewson [Wed, 20 Oct 2010 17:45:53 +0000 (13:45 -0400)] 
Merge branch 'bug2097' into maint-0.2.2

15 years agoFix a read of a freed pointer while in set_current_consensus
Nick Mathewson [Wed, 20 Oct 2010 16:34:02 +0000 (12:34 -0400)] 
Fix a read of a freed pointer while in set_current_consensus

Found by rransom while working on issue #988.  Bugfix on
0.2.2.17-alpha.  Fixes bug 2097.

15 years agoRemove redundant -Wpointer-sign CFLAG
Sebastian Hahn [Wed, 20 Oct 2010 11:01:30 +0000 (13:01 +0200)] 
Remove redundant -Wpointer-sign CFLAG

-Wpointer-sign is implied with -Wall, which we use when building with
--enable-gcc-warnings.

15 years agoRemove redundant -Wformat -Wformat-security CFLAGS
Sebastian Hahn [Wed, 20 Oct 2010 10:47:09 +0000 (12:47 +0200)] 
Remove redundant -Wformat -Wformat-security CFLAGS

When configuring with --enable-gcc-warnings, we use -Wformat=2 which
automatically enables the available -Wformat switches, so adding them
again in the --enable-gcc-hardening case doesn't make sense..

15 years agoUse ssp-buffer-size param when hardening
Sebastian Hahn [Tue, 19 Oct 2010 17:14:58 +0000 (19:14 +0200)] 
Use ssp-buffer-size param when hardening

We used to enable ssp-buffer-size=1 only when building with
--enable-gcc-warnings. That would result in warnings (and no
protection for small arrays) when building with
--enable-gcc-hardening without enabling warnings, too. Fixes bug
2031.

Also remove an XXX: We now allow to build with -fstack-protector
by using --enable-gcc-hardening.

15 years agoSend relay_early cells in rend circs
Sebastian Hahn [Mon, 18 Oct 2010 02:23:31 +0000 (04:23 +0200)] 
Send relay_early cells in rend circs

There are no relay left that run version 0.2.1.3 through 0.2.1.18, so
changing this behaviour should be safe now.

15 years agoIssues with router_get_by_nickname() (3)
Robert Hogan [Sun, 17 Oct 2010 14:12:25 +0000 (15:12 +0100)] 
Issues with router_get_by_nickname() (3)

Add changes file