Commit Graph

28210 Commits

Author SHA1 Message Date
Nick Mathewson
03b96882de Rename token_bucket_t to token_bucket_rw_t.
This is a simple search-and-replace to rename the token bucket type
to indicate that it contains both a read and a write bucket, bundled
with their configuration.  It's preliminary to refactoring the
bucket type.
2018-04-13 10:54:26 -04:00
Nick Mathewson
b152d62cee Merge branch 'token_bucket_refactor_squashed' 2018-04-13 10:47:24 -04:00
Nick Mathewson
2fe499eb3f Changes files for post-loop events (25374) 2018-04-13 10:44:15 -04:00
Nick Mathewson
62f4d5a265 Add a unit test for post-loop events
This test works by having two post-loop events activate one another
in a tight loop.  If the "post-loop" mechanism didn't work, this
would be enough to starve all other events.
2018-04-13 10:44:15 -04:00
Nick Mathewson
320bd2b3a5 Move connection_ap_attach_pending(0) into a postloop event
This is a second motivating case for our postloop event logic.
2018-04-13 10:44:15 -04:00
Nick Mathewson
5719dfb48f Move the "activate linked connections" logic to a postloop event.
A linked connection_t is one that gets its I/O, not from the
network, but from another connection_t.  When such a connection has
something to write, we want the corresponding connection to run its
read callback ... but not immediately, to avoid infinite recursion
and/or event loop starvation.

Previously we handled this case by activating the read events
outside the event loop.  Now we use the "postloop event" logic.
This lets us simplify do_main_loop_once() a little.
2018-04-13 10:44:15 -04:00
Nick Mathewson
c5a3e2ca44 Generic mechaism for "post-loop" callbacks
We've been labeling some events as happening "outside the event
loop", to avoid Libevent starvation.  This patch provides a cleaner
mechanism to avoid that starvation.

For background, the problem here is that Libevent only scans for new
events once it has run all its active callbacks.  So if the
callbacks keep activating new callbacks, they could potentially
starve Libevent indefinitely and keep it from ever checking for
timed, socket, or signal events.

To solve this, we add the ability to label some events as
"post-loop".  The rule for a "post-loop" event is that any events
_it_ activates can only be run after libevent has re-scanned for new
events at least once.
2018-04-13 10:44:15 -04:00
Nick Mathewson
ad57b1279a Disable load_geoip_file() tests on windows
See bug #25787 for discussion; we should have a better fix here.
2018-04-13 10:42:19 -04:00
Nick Mathewson
787bafc0f9 Increase tolerances for imprecise time. 2018-04-13 10:41:15 -04:00
Nick Mathewson
3f514fe3b1 Accept small hops backward in the monotonic timer. 2018-04-13 10:41:15 -04:00
Nick Mathewson
c63761a0a6 changes file for token_bucket refactoring 2018-04-13 10:41:14 -04:00
Nick Mathewson
12f58f2f87 Remove a bunch of int casts; make clang happier. 2018-04-13 10:41:14 -04:00
Nick Mathewson
6be994fa71 Ensure that global buckets are updated on configuration change 2018-04-13 10:41:14 -04:00
Nick Mathewson
a38fd9bc5b Replace the global buckets with token_bucket_t 2018-04-13 10:41:14 -04:00
Nick Mathewson
9fced56ef1 Refactor or_connection token buckets to use token_bucket_t 2018-04-13 10:41:14 -04:00
Nick Mathewson
8a85239746 Add a helper function to decrement read and write at the same time 2018-04-13 10:41:14 -04:00
Nick Mathewson
c376200f6a Add a new token-bucket backend abstraction, with tests
This differs from our previous token bucket abstraction in a few
ways:

  1) It is an abstraction, and not a collection of fields.
  2) It is meant to be used with monotonic timestamps, which should
     produce better results than calling gettimeofday over and over.
2018-04-13 10:41:14 -04:00
Nick Mathewson
d8ef9a2d1e Expose a function that computes stamp units from msec.
(It turns out we can't just expose STAMP_TICKS_PER_SECOND, since
Apple doesn't have that.)
2018-04-13 10:41:08 -04:00
Nick Mathewson
2d6914e391 Refine extend_info_for_node's "enough info" check once again.
In d1874b4339, we adjusted this check so that we insist on
using routerinfos for bridges.  That's almost correct... but if we
have a bridge that is also a regular relay, then we should use
insist on its routerinfo when connecting to it as a bridge
(directly), and be willing to use its microdescriptor when
connecting to it elsewhere in our circuits.

This bug is a likely cause of some (all?) of the (exit_ei == NULL)
failures we've been seeing.

Fixes bug 25691; bugfix on 0.3.3.4-alpha
2018-04-12 16:56:29 -04:00
Nick Mathewson
d3b9b5a3dd Remove windows log_from_handle as unused.
This function was only used by PortForwardingHelper, which was
removed in 9df110cd72.  Its presence caused warnings on windows.
2018-04-12 12:38:46 -04:00
Nick Mathewson
f0887e30dd Merge branch 'maint-0.3.3' 2018-04-12 12:31:41 -04:00
Nick Mathewson
46795a7be6 Attempt to fix 32-bit clang builds, which broke with 31508a0abc
When size_t is 32 bits, the unit tests can't fit anything more than
4GB-1 into a size_t.

Additionally, tt_int_op() uses "long" -- we need tt_u64_op() to
safely test uint64_t values for equality.

Bug caused by tests for #24782 fix; not in any released Tor.
2018-04-12 12:30:36 -04:00
Nick Mathewson
467c882baa Merge branch 'maint-0.3.3' 2018-04-12 12:25:51 -04:00
Nick Mathewson
4aaa4215e7 Attempt to fix 32-bit builds, which broke with 31508a0abc
When size_t is 32 bits, doing "size_t ram; if (ram > 8GB) { ... }"
produces a compile-time warning.

Bug caused by #24782 fix; not in any released Tor.
2018-04-12 12:25:09 -04:00
Nick Mathewson
037fb0c804 Merge branch 'maint-0.3.3' 2018-04-12 11:14:42 -04:00
Alexander Færøy
31508a0abc Use less memory for MaxMemInQueues for machines with more than 8 GB of RAM.
This patch changes the algorithm of compute_real_max_mem_in_queues() to
use 0.4 * RAM iff the system has more than or equal to 8 GB of RAM, but
will continue to use the old value of 0.75 * RAM if the system have less
than * GB of RAM available.

This patch also adds tests for compute_real_max_mem_in_queues().

See: https://bugs.torproject.org/24782
2018-04-12 11:14:16 -04:00
Alexander Færøy
5633a63379 Use STATIC for compute_real_max_mem_in_queues
This patch makes compute_real_max_mem_in_queues use the STATIC macro,
which allows us to test the function.

See: https://bugs.torproject.org/24782
2018-04-12 10:51:48 -04:00
Alexander Færøy
bd42367a1e Make get_total_system_memory mockable.
This patch makes get_total_system_memory mockable, which allows us to
alter the return value of the function in tests.

See: https://bugs.torproject.org/24782
2018-04-12 10:51:45 -04:00
Nick Mathewson
a51630cc9a Merge branch 'maint-0.3.3' 2018-04-11 15:38:00 -04:00
Nick Mathewson
0803d79f55 Merge branch 'bug25581_033_v2_asn_squashed' into maint-0.3.3 2018-04-11 15:37:56 -04:00
Nick Mathewson
8b8630a501 Rename HSLayer{2,3}Nodes to start without an underscore.
The old single-underscore names remain as a deprecated synonym.

Fixes bug 25581; bugfix on 0.3.3.1-alpha.
2018-04-11 15:37:49 -04:00
Nick Mathewson
0c8f901ee7 Merge branch 'maint-0.3.3' 2018-04-11 10:48:46 -04:00
Mike Perry
f9ba0c6546 Bug 24989: Count client hsdir gets towards MaxClientCircuitsPending.
We removed this by breaking them out from general in #13837.
2018-04-11 10:47:06 -04:00
Nick Mathewson
6bdfaa8b24 Merge remote-tracking branch 'isis-github/bug25425_squashed2' 2018-04-10 15:32:26 -04:00
Nick Mathewson
6e467a7a34 Merge remote-tracking branch 'isis-github/bug25409' 2018-04-10 15:27:09 -04:00
Isis Lovecruft
4178642bf8
changes: Add changes file for 25425. 2018-04-10 19:26:46 +00:00
Isis Lovecruft
550037f677
changes: Add changes file for 25409. 2018-04-10 19:21:10 +00:00
Isis Lovecruft
65d6b66e99
config: Obsolete PortForwarding and PortForwardingHelper options.
* FIXES part of #25409: https://bugs.torproject.org/25409
2018-04-10 19:08:59 +00:00
Nick Mathewson
d3ac47b415 Merge branch 'maint-0.3.1' into maint-0.3.2 2018-04-10 14:26:32 -04:00
Nick Mathewson
0b1a054d68 Merge branch 'maint-0.3.2' into maint-0.3.3 2018-04-10 14:26:32 -04:00
Nick Mathewson
ef16a11b90 Merge branch 'maint-0.3.3' 2018-04-10 14:26:32 -04:00
Nick Mathewson
db6902c235 Merge branch 'maint-0.2.9' into maint-0.3.1 2018-04-10 14:26:29 -04:00
Nick Mathewson
10a1969ca3 Merge remote-tracking branch 'ahf-github/bugs/24854_029_2' into maint-0.2.9 2018-04-10 14:25:57 -04:00
Nick Mathewson
16f08de0fd Remove TestingEnableTbEmptyEvent
This option was used for shadow testing previously, but is no longer
used for anything.  It interferes with refactoring our token buckets.
2018-04-10 12:16:21 -04:00
Isis Lovecruft
3ee7a8d3a5
tests: Make tt_finished() macro for tests without tt_*_op() calls. 2018-04-09 19:32:47 +00:00
Isis Lovecruft
c2c5b13e5d
test: Add testing module and some unittests for bridges.c.
This roughly doubles our test coverage of the bridges.c module.

 * ADD new testing module, .../src/test/test_bridges.c.
 * CHANGE a few function declarations from `static` to `STATIC`.
 * CHANGE one function in transports.c, transport_get_by_name(), to be
   mockable.
 * CLOSES #25425: https://bugs.torproject.org/25425
2018-04-09 19:32:46 +00:00
Nick Mathewson
e0809ec5f5 Prefer 32-bit implementation for timing wheels on 32-bit systems.
This might make our timing-wheel code a tiny bit faster there.

Closes ticket 24688.
2018-04-09 15:21:10 -04:00
David Goulet
395fa0258d compat: Fix unchecked return value from event_del()
Explicitly tell the compiler we don't care about it.

Coverity CID 1434156

Signed-off-by: David Goulet <dgoulet@torproject.org>
2018-04-09 14:12:45 -04:00
Nick Mathewson
e58555135a Add a comment explaining why we do a certain redundant check
Closes ticket 25291.
2018-04-09 12:58:17 -04:00
Alexander Færøy
1295044dc8 Lift the list of default directory servers into their own file.
This patch lifts the list of default directory authorities from config.c
into their own auth_dirs.inc file, which is then included in config.c
using the C preprocessor.

Patch by beastr0.

See: https://bugs.torproject.org/24854
2018-04-09 16:00:26 +02:00