]> git.ipfire.org Git - thirdparty/tor.git/log
thirdparty/tor.git
14 years agoFix compile warning on windows
Sebastian Hahn [Thu, 5 May 2011 00:21:09 +0000 (02:21 +0200)] 
Fix compile warning on windows

14 years agoChange who calls microdesc_cache_rebuild().
Nick Mathewson [Tue, 3 May 2011 21:28:28 +0000 (17:28 -0400)] 
Change who calls microdesc_cache_rebuild().

Previously we ensured that it would get called periodically by doing
it from inside the code that added microdescriptors.  That won't work
though: it would interfere with our code that tried to read microdescs
from disk initially.  Instead, we should consider rebuilding the cache
periodically, and on startup.

14 years agoRebuild the microdesc cache when a sufficient number of bytes are dropped
Nick Mathewson [Tue, 3 May 2011 21:03:49 +0000 (17:03 -0400)] 
Rebuild the microdesc cache when a sufficient number of bytes are dropped

Previously on 0.2.2, we'd never clean the cache.  Now that we can
clean it, we want to add a condition to rebuild it: that should happen
whenever we have dropped enough microdescriptors that we could save a
lot of space.

No changes file, since 0.2.3 doesn't need one and 0.2.2 already has some
changes files for the backport of the microdesc_clean_cahce() function.

14 years agoBackport microdesc_cache_clean to 0.2.2
Nick Mathewson [Tue, 3 May 2011 20:45:15 +0000 (16:45 -0400)] 
Backport microdesc_cache_clean to 0.2.2

Otherwise we have no way to keep authorities' microdesc caches in 0.2.2
from growing without bound.

14 years agoFix a check for when to rebuild the microdesc cache. (Backport from 0.2.3.
Nick Mathewson [Tue, 3 May 2011 20:29:39 +0000 (16:29 -0400)] 
Fix a check for when to rebuild the microdesc cache. (Backport from 0.2.3.

14 years agoAdd missing code to set cache->journal_len when reading microdesc journal
Nick Mathewson [Tue, 3 May 2011 20:22:31 +0000 (16:22 -0400)] 
Add missing code to set cache->journal_len when reading microdesc journal

This could be one reason that authorities' journals would grow without
bound; related to bug 2230. Bugfix on 0.2.2.6-alpha.  Fix by
"cypherpunks".

14 years agoAdd a couple of notes to doc/HACKING based on 0.2.2.25-alpha process
Nick Mathewson [Mon, 2 May 2011 18:45:44 +0000 (14:45 -0400)] 
Add a couple of notes to doc/HACKING based on 0.2.2.25-alpha process

14 years agoClean up a warning a bit
Nick Mathewson [Fri, 29 Apr 2011 15:14:53 +0000 (11:14 -0400)] 
Clean up a warning a bit

14 years agoadd another heuristic for changes stanzas
Roger Dingledine [Fri, 29 Apr 2011 05:35:21 +0000 (01:35 -0400)] 
add another heuristic for changes stanzas

14 years agoUpdate hacking file with terse notes on formatting changelog
Nick Mathewson [Fri, 29 Apr 2011 03:44:48 +0000 (23:44 -0400)] 
Update hacking file with terse notes on formatting changelog

14 years agoMerge remote-tracking branch 'arma/bug3012' into maint-0.2.2
Nick Mathewson [Fri, 29 Apr 2011 01:15:14 +0000 (21:15 -0400)] 
Merge remote-tracking branch 'arma/bug3012' into maint-0.2.2

14 years agorelays checkpoint their state file twice a day
Roger Dingledine [Fri, 29 Apr 2011 01:06:25 +0000 (21:06 -0400)] 
relays checkpoint their state file twice a day

14 years agoMerge remote-tracking branch 'arma/bug3039' into maint-0.2.2
Nick Mathewson [Fri, 29 Apr 2011 00:53:44 +0000 (20:53 -0400)] 
Merge remote-tracking branch 'arma/bug3039' into maint-0.2.2

14 years agostop putting wacky values into state->lastwritten
Roger Dingledine [Fri, 29 Apr 2011 00:40:15 +0000 (20:40 -0400)] 
stop putting wacky values into state->lastwritten

14 years agoClarify comment to say which version fixed 2722
Nick Mathewson [Fri, 29 Apr 2011 00:38:15 +0000 (20:38 -0400)] 
Clarify comment to say which version fixed 2722

14 years agoMerge remote-tracking branches 'rransom/bug2722' and 'rransom/bug2722b' into maint...
Nick Mathewson [Fri, 29 Apr 2011 00:36:38 +0000 (20:36 -0400)] 
Merge remote-tracking branches 'rransom/bug2722' and 'rransom/bug2722b' into maint-0.2.2

14 years agoMerge remote-tracking branch 'sebastian/bug3020' into maint-0.2.2
Nick Mathewson [Fri, 29 Apr 2011 00:03:57 +0000 (20:03 -0400)] 
Merge remote-tracking branch 'sebastian/bug3020' into maint-0.2.2

14 years agofix a function comment
Roger Dingledine [Thu, 28 Apr 2011 23:19:04 +0000 (19:19 -0400)] 
fix a function comment

14 years agoCorrectly check elapsed time in last hibernation period
Sebastian Hahn [Thu, 28 Apr 2011 23:18:32 +0000 (01:18 +0200)] 
Correctly check elapsed time in last hibernation period

Fix bug 3020.

14 years agoMerge branch 'bug3000-022-part2-squashed' into maint-0.2.2
Nick Mathewson [Thu, 28 Apr 2011 22:13:17 +0000 (18:13 -0400)] 
Merge branch 'bug3000-022-part2-squashed' into maint-0.2.2

14 years agoAvoid false positives from proxy_mode()
Nick Mathewson [Thu, 28 Apr 2011 21:45:41 +0000 (17:45 -0400)] 
Avoid false positives from proxy_mode()

Previously it would erroneously return true if ListenAddr was set for
a client port, even if that port itself was 0.  This would give false
positives, which were not previously harmful... but which were about
to become.

14 years agoFix bug 1930
Robert Ransom [Tue, 26 Apr 2011 09:21:25 +0000 (02:21 -0700)] 
Fix bug 1930

14 years agoIgnore SIGNAL NEWNYM on relay-only Tor instances
Robert Ransom [Thu, 21 Apr 2011 00:01:41 +0000 (17:01 -0700)] 
Ignore SIGNAL NEWNYM on relay-only Tor instances

14 years agoDon't allow v0 HS auths to act as clients
Robert Ransom [Wed, 20 Apr 2011 22:20:10 +0000 (15:20 -0700)] 
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.

14 years agoCorrect the logic from f14754fbd for tor_gmtime_r
John Brooks [Tue, 4 Jan 2011 04:36:09 +0000 (21:36 -0700)] 
Correct the logic from f14754fbd for tor_gmtime_r

14 years agoDetect and handle NULL returns from (gm/local)time_r
Nick Mathewson [Tue, 7 Dec 2010 19:37:50 +0000 (14:37 -0500)] 
Detect and handle NULL returns from (gm/local)time_r

These functions can return NULL for otherwise-valid values of
time_t.  Notably, the glibc gmtime manpage says it can return NULL
if the year if greater than INT_MAX, and the windows MSDN gmtime
page says it can return NULL for negative time_t values.

Also, our formatting code is not guaranteed to correctly handle
years after 9999 CE.

This patch tries to correct this by detecting NULL values from
gmtime/localtime_r, and trying to clip them to a reasonable end of
the scale.  If they are in the middle of the scale, we call it a
downright error.

Arguably, it's a bug to get out-of-bounds dates like this to begin
with.  But we've had bugs of this kind in the past, and warning when
we see a bug is much kinder than doing a NULL-pointer dereference.

Boboper found this one too.

14 years agoMerge remote-tracking branch 'sebastian/bug2497' into maint-0.2.2
Nick Mathewson [Thu, 28 Apr 2011 20:05:34 +0000 (16:05 -0400)] 
Merge remote-tracking branch 'sebastian/bug2497' into maint-0.2.2

14 years agoClean up merge of bug3k_021
Robert Ransom [Thu, 28 Apr 2011 17:37:35 +0000 (10:37 -0700)] 
Clean up merge of bug3k_021

14 years agoMerge branch 'bug3k_021' into bug3k_022
Sebastian Hahn [Thu, 28 Apr 2011 17:00:34 +0000 (19:00 +0200)] 
Merge branch 'bug3k_021' into bug3k_022

Conflicts:
src/or/or.h
src/or/rendclient.c

14 years agoFix a failure case of connection_ap_handshake_attach_circuit()
Sebastian Hahn [Wed, 27 Apr 2011 23:48:25 +0000 (01:48 +0200)] 
Fix a failure case of connection_ap_handshake_attach_circuit()

tor_fragile_assert() might be a no-op, so we have to return something
here to indicate failure to the caller.

14 years agoFix a bug introduced by purging rend_cache on NEWNYM
Robert Ransom [Mon, 25 Apr 2011 13:38:35 +0000 (06:38 -0700)] 
Fix a bug introduced by purging rend_cache on NEWNYM

If the user sent a SIGNAL NEWNYM command after we fetched a rendezvous
descriptor, while we were building the introduction-point circuit, we
would give up entirely on trying to connect to the hidden service.

Original patch by rransom slightly edited to go into 0.2.1

14 years agoAllow rend_client_send_introduction to fail transiently
Robert Ransom [Mon, 25 Apr 2011 15:36:02 +0000 (08:36 -0700)] 
Allow rend_client_send_introduction to fail transiently

i.e. without closing the AP connection.

14 years agoForget all rendezvous client state on SIGNAL NEWNYM
Robert Ransom [Wed, 20 Apr 2011 09:27:58 +0000 (02:27 -0700)] 
Forget all rendezvous client state on SIGNAL NEWNYM

14 years agoMerge remote-tracking branch 'origin/maint-0.2.1' into maint-0.2.2
Nick Mathewson [Thu, 28 Apr 2011 02:14:28 +0000 (22:14 -0400)] 
Merge remote-tracking branch 'origin/maint-0.2.1' into maint-0.2.2

Conflicts:
src/common/torint.h

14 years agoI guess that had a bug number: add it to the changes file.
Nick Mathewson [Wed, 27 Apr 2011 21:26:28 +0000 (17:26 -0400)] 
I guess that had a bug number: add it to the changes file.

14 years agoFix clear_trackhostexits_mapping() to actually work as advertised
Nick Mathewson [Wed, 27 Apr 2011 21:21:41 +0000 (17:21 -0400)] 
Fix clear_trackhostexits_mapping() to actually work as advertised

Previously, it would remove every trackhostexits-derived mapping
*from* xyz.<exitname>.exit; it was supposed to remove every
trackhostexits-derived mapping *to* xyz.<exitname>.exit.

Bugfix on 0.2.0.20-rc: fixes an XXX020 added while staring at bug-1090
issues.

14 years agomake make check-spaces happier.
Roger Dingledine [Wed, 27 Apr 2011 19:18:34 +0000 (15:18 -0400)] 
make make check-spaces happier.

(still not happy.)

14 years agoFix double-free bug in microdesc parser
cypherpunks [Wed, 27 Apr 2011 18:10:56 +0000 (11:10 -0700)] 
Fix double-free bug in microdesc parser

14 years agominor tweaks to manpage
Roger Dingledine [Wed, 27 Apr 2011 18:04:50 +0000 (14:04 -0400)] 
minor tweaks to manpage

14 years agoMerge remote-tracking branch 'arma/bug1090-part1-squashed' into bug1090-part1-squashed
Nick Mathewson [Wed, 27 Apr 2011 17:45:28 +0000 (13:45 -0400)] 
Merge remote-tracking branch 'arma/bug1090-part1-squashed' into bug1090-part1-squashed

14 years agoMerge branch 'bug1090-manpage-squashed' into bug1090-part1-squashed
Nick Mathewson [Wed, 27 Apr 2011 17:44:55 +0000 (13:44 -0400)] 
Merge branch 'bug1090-manpage-squashed' into bug1090-part1-squashed

14 years agoexplain an argument in a function comment
Roger Dingledine [Wed, 27 Apr 2011 17:43:11 +0000 (13:43 -0400)] 
explain an argument in a function comment

14 years agoRevise the manpage to contain the actual intended *Nodes behavior
Nick Mathewson [Thu, 10 Mar 2011 23:25:51 +0000 (18:25 -0500)] 
Revise the manpage to contain the actual intended *Nodes behavior

This is a squashed version of my former desired_nodes_behavior branch
that we used to specify the intended results wrt bug 1090.

14 years agoAdd a big changelog entry for bug 1090 fixes
Nick Mathewson [Wed, 27 Apr 2011 17:33:38 +0000 (13:33 -0400)] 
Add a big changelog entry for bug 1090 fixes

14 years agoInstead of checking whether we have unremoved intro points, check for usable ones
Nick Mathewson [Wed, 20 Apr 2011 20:49:41 +0000 (16:49 -0400)] 
Instead of checking whether we have unremoved intro points, check for usable ones

14 years agobetter user-facing warnings for unexpected last hops
Roger Dingledine [Wed, 27 Apr 2011 02:48:00 +0000 (22:48 -0400)] 
better user-facing warnings for unexpected last hops

these still aren't perfect, but we won't know how to correct them until
we start experiencing surprised users.

14 years agowarn if we launch too many circuits for a given stream
Roger Dingledine [Wed, 27 Apr 2011 02:18:01 +0000 (22:18 -0400)] 
warn if we launch too many circuits for a given stream

14 years agorevert most of ef81649d2fc
Roger Dingledine [Tue, 26 Apr 2011 23:55:34 +0000 (19:55 -0400)] 
revert most of ef81649d2fc

Now we believe it to be the case that we never build a circuit for our
stream that has an unsuitable exit, so we'll never need to use such
a circuit. The risk is that we have some code that builds the circuit,
but now we refuse to use it, meaning we just build a bazillion circuits
and ignore them all.

14 years agoAdd a circuit_purpose_to_string() function, and use it
Nick Mathewson [Fri, 8 Apr 2011 17:27:25 +0000 (13:27 -0400)] 
Add a circuit_purpose_to_string() function, and use it

We had a circuit_purpose_to_controller_string() function, but it was
pretty coarse-grained and didn't try to be human-readable.

14 years agoFix a check-spaces complaint
Sebastian Hahn [Fri, 8 Apr 2011 02:16:26 +0000 (04:16 +0200)] 
Fix a check-spaces complaint

14 years agoExplain the "using anyway" log message better
Sebastian Hahn [Fri, 8 Apr 2011 02:01:23 +0000 (04:01 +0200)] 
Explain the "using anyway" log message better

Also add a request to report any cases where we are not honoring
StrictNodes to the developers: These should now all be bugs.

14 years agoFix a log msg
Sebastian Hahn [Fri, 8 Apr 2011 02:03:50 +0000 (04:03 +0200)] 
Fix a log msg

14 years agoCheck transition of circuit purpose from INTRO->GENERAL if nodes are constrained
Nick Mathewson [Mon, 4 Apr 2011 00:06:31 +0000 (20:06 -0400)] 
Check transition of circuit purpose from INTRO->GENERAL if nodes are constrained

This looked at first like another fun way around our node selection
logic: if we had introduction circuits, and we wound up building too
many, we would turn extras into general-purpose circuits.  But when we
did so, we wouldn't necessarily check whether the general-purpose
circuits conformed to our node constraints.  For example, the last
node could totally be in ExcludedExitNodes and we wouldn't have cared...

...except that the circuit should already be internal, so it won't get user
streams attached to it, so the transition should generally be allowed.
Add an assert to make sure we're right about this, and have it not
check whether ExitNodes is set, since that's irrelevant to internal
circuits.

14 years agoWhen cannibalizing a circuit, make sure it has no ExcludeNodes on it
Nick Mathewson [Sun, 3 Apr 2011 23:58:28 +0000 (19:58 -0400)] 
When cannibalizing a circuit, make sure it has no ExcludeNodes on it

This could happen if StrictNodes was 0 and we were forced to pick an
excluded node as the last hop of the circuit.

14 years agoWhen there is a transition in permitted nodes, apply it to trackexithosts map
Nick Mathewson [Sun, 3 Apr 2011 23:43:47 +0000 (19:43 -0400)] 
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.

14 years agoSimplify calls to routerset_equal
Nick Mathewson [Sun, 3 Apr 2011 23:13:36 +0000 (19:13 -0400)] 
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.

14 years agoIf we have chosen an exit that shares a family with all bridges, fail the circuit
Nick Mathewson [Sun, 3 Apr 2011 22:20:19 +0000 (18:20 -0400)] 
If we have chosen an exit that shares a family with all bridges, fail the circuit

We could probably do something smarter here, but the situation is
unusual enough that it's okay to just fail the circuit.

14 years agoDo not try to download descriptors for bridges in ExcludeNodes.
Nick Mathewson [Sun, 3 Apr 2011 22:12:26 +0000 (18:12 -0400)] 
Do not try to download descriptors for bridges in ExcludeNodes.

14 years agoCorrect the behavior of .exit with ExcludeNodes, StrictNodes, etc.
Nick Mathewson [Sun, 3 Apr 2011 21:08:59 +0000 (17:08 -0400)] 
Correct the behavior of .exit with ExcludeNodes, StrictNodes, etc.

ExcludeExitNodes foo now means that foo.exit doesn't work.  If
StrictNodes is set, then ExcludeNodes foo also overrides foo.exit.

foo.exit , however, still works even if foo is not listed in ExitNodes.

14 years agoNote another place that we need to fix a 1090 issue.
Nick Mathewson [Sun, 3 Apr 2011 21:08:29 +0000 (17:08 -0400)] 
Note another place that we need to fix a 1090 issue.

14 years agoDo not automatically ignore Fast/Stable for exits when ExitNodes is set
Nick Mathewson [Mon, 28 Mar 2011 21:29:59 +0000 (17:29 -0400)] 
Do not automatically ignore Fast/Stable for exits when ExitNodes is set

This once maybe made sense when ExitNodes meant "Here are 3 exits;
use them all", but now it more typically means "Here are 3
countries; exit from there."  Using non-Fast/Stable exits created a
potential partitioning opportunity and an annoying stability
problem.

(Don't worry about the case where all of our ExitNodes are non-Fast
or non-Stable: we handle that later in the function by retrying with
need_capacity and need_uptime set to 0.)

14 years agoNote a slightly less likely way to violate ExcludeNodes
Nick Mathewson [Mon, 28 Mar 2011 20:51:00 +0000 (16:51 -0400)] 
Note a slightly less likely way to violate ExcludeNodes

14 years agoNote that circuit purpose changing can violate ExcludeNodes
Nick Mathewson [Mon, 28 Mar 2011 20:44:40 +0000 (16:44 -0400)] 
Note that circuit purpose changing can violate ExcludeNodes

14 years agoIf we're excluded, and StrictNodes is set, do not do self-tests.
Nick Mathewson [Mon, 28 Mar 2011 18:14:45 +0000 (14:14 -0400)] 
If we're excluded, and StrictNodes is set, do not do self-tests.

14 years agoAdd an XXX022-1090 to note consider_exit_fmily b0rkenness
Nick Mathewson [Fri, 25 Mar 2011 19:02:37 +0000 (15:02 -0400)] 
Add an XXX022-1090 to note consider_exit_fmily b0rkenness

14 years agothree more cases where maybe we want to exclude
Roger Dingledine [Fri, 11 Mar 2011 11:19:15 +0000 (06:19 -0500)] 
three more cases where maybe we want to exclude

14 years agomake formal a constraint that's been true a while now
Roger Dingledine [Fri, 11 Mar 2011 10:39:26 +0000 (05:39 -0500)] 
make formal a constraint that's been true a while now

14 years agorefuse moria1.exit if moria1 is excluded
Roger Dingledine [Fri, 11 Mar 2011 10:29:28 +0000 (05:29 -0500)] 
refuse moria1.exit if moria1 is excluded

add a note reminding us to do this for foo.moria1.exit if we decide to.

14 years agonote another case where strictnodes is considered for exits
Roger Dingledine [Fri, 11 Mar 2011 09:36:24 +0000 (04:36 -0500)] 
note another case where strictnodes is considered for exits

14 years agoIf ExitNodes and Exclude{Exit}Nodes overlap, obey Exclude{Exit}Nodes.
Roger Dingledine [Fri, 11 Mar 2011 09:35:08 +0000 (04:35 -0500)] 
If ExitNodes and Exclude{Exit}Nodes overlap, obey Exclude{Exit}Nodes.

Also, ExitNodes are always strict.

14 years agothe new entrynodes behavior is always strict
Roger Dingledine [Fri, 11 Mar 2011 09:21:25 +0000 (04:21 -0500)] 
the new entrynodes behavior is always strict

14 years agodon't exit enclave to excluded relays
Roger Dingledine [Fri, 11 Mar 2011 08:57:01 +0000 (03:57 -0500)] 
don't exit enclave to excluded relays

14 years agoslight tweak on circuit_conforms_to_options
Roger Dingledine [Fri, 11 Mar 2011 08:22:53 +0000 (03:22 -0500)] 
slight tweak on circuit_conforms_to_options

this function really needs to get a total rewrite (or die)

For now, use #if 0 to disable it.

14 years agohandle excludenodes for dir fetch/post
Roger Dingledine [Fri, 11 Mar 2011 08:09:24 +0000 (03:09 -0500)] 
handle excludenodes for dir fetch/post

If we're picking a random directory node, never pick an excluded one.
But if we've chosen a specific one (or all), allow it unless strictnodes
is set (in which case warn so the user knows it's their fault).

When warning that we won't connect to a strictly excluded node,
log what it was we were trying to do at that node.

When ExcludeNodes is set but StrictNodes is not set, we only use
non-excluded nodes if we can, but fall back to using excluded nodes
if none of those nodes is usable.

14 years agoIf EntryNodes and ExcludeNodes overlap, obey ExcludeNodes.
Roger Dingledine [Sat, 17 Oct 2009 22:54:20 +0000 (18:54 -0400)] 
If EntryNodes and ExcludeNodes overlap, obey ExcludeNodes.

14 years agorefuse excluded hidserv nodes if strictnodes
Roger Dingledine [Sat, 17 Oct 2009 22:52:18 +0000 (18:52 -0400)] 
refuse excluded hidserv nodes if strictnodes

Make hidden services more flaky for people who set both ExcludeNodes
and StrictNodes. Not recommended, especially for hidden service operators.

14 years agoExpose a new process_signal(uintptr_t), not signal_callback()
Nick Mathewson [Tue, 26 Apr 2011 19:20:08 +0000 (15:20 -0400)] 
Expose a new process_signal(uintptr_t), not signal_callback()

This is a tweak to the bug2917 fix.  Basically, if we want to simulate
a signal arriving in the controller, we shouldn't have to pretend that
we're Libevent, or depend on how Tor sets up its Libevent callbacks.

14 years agoMerge remote-tracking branch 'sebastian/bug2917' into maint-0.2.2
Nick Mathewson [Tue, 26 Apr 2011 19:17:03 +0000 (15:17 -0400)] 
Merge remote-tracking branch 'sebastian/bug2917' into maint-0.2.2

14 years agoMerge remote-tracking branch 'public/bug2332' into maint-0.2.2
Nick Mathewson [Tue, 26 Apr 2011 17:10:54 +0000 (13:10 -0400)] 
Merge remote-tracking branch 'public/bug2332' into maint-0.2.2

14 years agoMake SIZE_T_CEILING unsigned; add a signed SSIZE_T_CEILING
Nick Mathewson [Tue, 26 Apr 2011 17:00:46 +0000 (13:00 -0400)] 
Make SIZE_T_CEILING unsigned; add a signed SSIZE_T_CEILING

None of the comparisons were _broken_ previously, but avoiding
signed/unsigned comparisons makes everybody happier.

Fixes bug2475.

14 years agoDowngrade notice to info when downloading a cert.
Nick Mathewson [Tue, 26 Apr 2011 16:46:07 +0000 (12:46 -0400)] 
Downgrade notice to info when downloading a cert.

14 years agofix some comments before they create conflicts
Roger Dingledine [Tue, 26 Apr 2011 15:28:49 +0000 (11:28 -0400)] 
fix some comments before they create conflicts

14 years agoFix compilation in last patch
Nick Mathewson [Tue, 26 Apr 2011 15:14:46 +0000 (11:14 -0400)] 
Fix compilation in last patch

14 years agoMerge remote-tracking branch 'sebastian/bug2704' into maint-0.2.2
Nick Mathewson [Tue, 26 Apr 2011 15:07:48 +0000 (11:07 -0400)] 
Merge remote-tracking branch 'sebastian/bug2704' into maint-0.2.2

14 years agoFix more of bug 2704
Sebastian Hahn [Tue, 26 Apr 2011 13:33:08 +0000 (15:33 +0200)] 
Fix more of bug 2704

The last entry of the *Maxima values in the state file was inflated by a
factor of NUM_SECS_ROLLING_MEASURE (currently 10). This could lead to
a wrong maximum value propagating through the state file history.

14 years agoDon't leak the local hostname in relay nicknames
Sebastian Hahn [Tue, 26 Apr 2011 03:07:37 +0000 (05:07 +0200)] 
Don't leak the local hostname in relay nicknames

Fixes bug 2979, reported by tagnaq.

14 years agoMerge commit '91aa6f08bcf0acbdfa038aaffe73e327ddd87c67' into maint-0.2.2
Nick Mathewson [Mon, 25 Apr 2011 23:03:15 +0000 (19:03 -0400)] 
Merge commit '91aa6f08bcf0acbdfa038aaffe73e327ddd87c67' into maint-0.2.2

14 years agoMake the Log configuration option expand ~
Sebastian Hahn [Fri, 22 Apr 2011 14:06:52 +0000 (16:06 +0200)] 
Make the Log configuration option expand ~

14 years agoMerge remote-tracking branch 'origin/maint-0.2.1' into maint-0.2.2
Nick Mathewson [Thu, 21 Apr 2011 17:40:00 +0000 (13:40 -0400)] 
Merge remote-tracking branch 'origin/maint-0.2.1' into maint-0.2.2

14 years agoFix a bug in removing DNSPort requests from their circular list
Nick Mathewson [Thu, 21 Apr 2011 17:39:00 +0000 (13:39 -0400)] 
Fix a bug in removing DNSPort requests from their circular list

Under heavy load, this could result in an assertion failure.  Fix for
bug 2933; bugfix on 0.2.0.10-alpha.

14 years agoPrevent hugely inflated observed bandwidth values
Sebastian Hahn [Tue, 19 Apr 2011 14:00:41 +0000 (16:00 +0200)] 
Prevent hugely inflated observed bandwidth values

When reading the bw history from the state file, we'd add the 900-second
value as traffic that occured during one second. Fix that by adding the
average value to each second.

This bug was present since 0.2.0.5-alpha, but was hidden until
0.2.23-alpha when we started using the saved values.

14 years agoDon't sometimes undercount bw average
Sebastian Hahn [Tue, 19 Apr 2011 15:12:27 +0000 (17:12 +0200)] 
Don't sometimes undercount bw average

This fixes the first part of bug 2704. It should occur only rarely when
no bw maxima are known. Bugfix on 0.2.2.23-alpha. Fixes bug 2704.

14 years agoStandardize our printf code on %d, not %i.
Nick Mathewson [Tue, 19 Apr 2011 16:40:26 +0000 (12:40 -0400)] 
Standardize our printf code on %d, not %i.

14 years agoMerge remote-tracking branch 'rransom/bug2948' into maint-0.2.2
Nick Mathewson [Tue, 19 Apr 2011 16:36:49 +0000 (12:36 -0400)] 
Merge remote-tracking branch 'rransom/bug2948' into maint-0.2.2

14 years agoMerge remote-tracking branch 'origin/maint-0.2.1' into maint-0.2.2
Nick Mathewson [Tue, 19 Apr 2011 16:31:42 +0000 (12:31 -0400)] 
Merge remote-tracking branch 'origin/maint-0.2.1' into maint-0.2.2

14 years agoMerge remote-tracking branch 'rransom/bug2750-v3' into maint-0.2.1
Nick Mathewson [Tue, 19 Apr 2011 16:30:50 +0000 (12:30 -0400)] 
Merge remote-tracking branch 'rransom/bug2750-v3' into maint-0.2.1

14 years agoMerge remote-tracking branch 'origin/maint-0.2.1' into maint-0.2.2
Nick Mathewson [Tue, 19 Apr 2011 16:30:11 +0000 (12:30 -0400)] 
Merge remote-tracking branch 'origin/maint-0.2.1' into maint-0.2.2

14 years agoOuch: correctly tabify the micro-revision.i target in Makefile.am (0.2.1 only)
Nick Mathewson [Tue, 19 Apr 2011 16:29:20 +0000 (12:29 -0400)] 
Ouch: correctly tabify the micro-revision.i target in Makefile.am (0.2.1 only)

14 years agoDon't report empty bw-history lines in extrainfo
Sebastian Hahn [Sat, 16 Apr 2011 14:01:36 +0000 (16:01 +0200)] 
Don't report empty bw-history lines in extrainfo

Some tor relays would report lines like these in their extrainfo
documents:
dirreq-write-history 2011-03-14 16:46:44 (900 s)

This was confusing to some people who look at the stats. It would happen
whenever a relay first starts up, or when a relay has dirport disabled.
Change this so that lines without actual bw entries are omitted.
Implements ticket 2497.

14 years agoCorrect HS descriptor length check
Nick Mathewson [Mon, 18 Apr 2011 20:53:13 +0000 (13:53 -0700)] 
Correct HS descriptor length check

Fixes bug 2948.