1) Change AsyncSocket's SignalReadEvent and SignalWriteEvent's thread mode to multi_threaded_local as they can be accessed from different threads.
2) Protect NATServer::TransEntry::whitelist.
3) Protect PhysicalSocket:error_.
Detail failures can be seen from issue 2080, comment #5.
TBR=fischman@webrtc.org
RISK=P1
TEST=try bots and tsanv2
BUG=2080
Review URL: https://webrtc-codereview.appspot.com/2669005
git-svn-id: http://webrtc.googlecode.com/svn/trunk@5026 4adac7df-926f-26a2-2b94-8c16560cd09d
1) init start_time_ns_ before the file_read_thread_ is started to avoid data racing as the start_time_ns_ will also be read by the file_read_thread_.
2) add CriticalSection to protect |finished_| that is accessed by FileReadThread and the main thread
Also remove the suppression for filemediaengine.cc, which has already been fixed in other cl.
TBR=henrike@webrtc.org
TEST=try bots and manual tsan v2 test
BUG=2078
Review URL: https://webrtc-codereview.appspot.com/2509004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@5018 4adac7df-926f-26a2-2b94-8c16560cd09d
Update Thread documentation, fix all subclasses that had a problem.
This is to avoid a data racing between the destructor modifying the vtable, and
Thread::PreRun calling virtual method Run at the same time.
For example:
[ RUN ] FileMediaEngineTest.TestGetCapabilities
==================
WARNING: ThreadSanitizer: data race on vptr (ctor/dtor vs virtual call) (pid=2967)
Read of size 8 at 0x7d480000bd00 by thread T1:
#0 talk_base::Thread::PreRun(void*) /mnt/data/b/build/slave/Linux_Tsan_v2/build/src/out/Release/../../talk/base/thread.cc:353 (libjingle_media_unittest+0x000000234da8)
Previous write of size 8 at 0x7d480000bd00 by main thread:
#0 talk_base::Thread::~Thread() /mnt/data/b/build/slave/Linux_Tsan_v2/build/src/out/Release/../../talk/base/thread.cc:158 (libjingle_media_unittest+0x00000023478c)
#1 ~RtpSenderReceiver /mnt/data/b/build/slave/Linux_Tsan_v2/build/src/out/Release/../../talk/media/base/filemediaengine.cc:122 (libjingle_media_unittest+0x0000001b551f)
...
RISK=P2
TESTED=try bots and tsan
BUG=2078,2080
R=fischman@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2428004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4999 4adac7df-926f-26a2-2b94-8c16560cd09d
Add suppressions for libjingle tests so they pass under TSan v2.
Disable the following tests for TSan v2 (only) since they're failing:
* StunServerTest.TestGood
* JsepPeerConnectionP2PTestClient.*
See build logs at:
http://build.chromium.org/p/client.webrtc.fyi/builders/Linux%20Tsan%20v2/
for more details.
BUG=1205,2078,2079,2080,2517
TEST=Ran a successful run of each test locally on Linux using:
GYP_DEFINES='tsan=1 linux_use_tcmalloc=0 release_extra_cflags="-gline-tables-only"' gclient runhooks
ninja -C out/Release
For each test, run standing in trunk/:
TSAN_OPTIONS="suppressions=tools/valgrind-webrtc/tsan_v2/suppressions.txt print_suppressions=1 report_signal_unsafe=0 report_thread_leaks=0 history_size=7" out/Release/[libjingle_testname]
R=wu@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2411004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4977 4adac7df-926f-26a2-2b94-8c16560cd09d
In https://codereview.chromium.org/26190002 Chromium
started a cleanup of their wrapper script, by adding
the --target flag and start passing build dir similar
to how the bots are setting it (i.e. pass out and not
out/Release).
This CL adds --target support for our wrapper script,
without changing the existing behavior (I'll do a
larger update at a later stage).
BUG=none
TEST=Ran the following successfully:
tools/valgrind-webrtc/webrtc_tests.sh --build_dir out/Release --target Release --test test_support_unittests
tools/valgrind-webrtc/webrtc_tests.sh --build_dir out --target Release --test test_support_unittests
tools/valgrind-webrtc/webrtc_tests.sh --target Release --test out/Release/test_support_unittests
TBR=phoglund@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2396004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4957 4adac7df-926f-26a2-2b94-8c16560cd09d
a different thread.
One example is the _playWarning can be changed in AudioDeviceLinuxALSA::Init, which is called on the application's thread. At the same time it can be read via PlayoutWarning() on the VoE's process_thread.
RISK=P2
TESTED=try bots and tsan test:
tools/valgrind-webrtc/webrtc_tests.sh --tool=tsan -t out/Debug/libjingle_peerconnection_unittest --gtest_filter=PeerConnectionFactoryTestInternal.CreatePCUsingInternalModules
BUG=1205
R=xians@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2315004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4866 4adac7df-926f-26a2-2b94-8c16560cd09d
Implements the ExternalVideoDecoder interface for VideoReceiveStream.
Also adds a FakeDecoder used in tests, removing the overhead of running
the EngineTest tests with VP8 under Memcheck/TSan, allowing us to enable
them under Memcheck/TSan as well.
BUG=2346,2312
R=stefan@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2172004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4702 4adac7df-926f-26a2-2b94-8c16560cd09d
I mistakenly ommitted the checks when logging.h was ported from
libjingle to webrtc. This caused a significant CPU cost for logs which
were later filtered out anyway.
Verified with LS_VERBOSE logging in neteq4, running:
$ out/Release/modules_unittests \
--gtest_filter=NetEqDecodingTest.TestBitExactness \
--gtest_repeat=50 > time.txt
$ grep "case ran" time.txt | grep "[0-9]* ms" -o | sort
Results on a MacBook Retina, averaged over 5 runs:
Verbose logs disabled: 666 ms
Exisiting implementation, verbose logs enabled: 944 ms (1.42x)
New implementation, verbose logs enabled: 673 ms (1.01x)
BUG=2314
R=henrik.lundin@webrtc.org, henrike@webrtc.org, kjellander@webrtc.org, turaj@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2160005
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4682 4adac7df-926f-26a2-2b94-8c16560cd09d
This test goes into an eternal loop sometimes when run under
TSan v2, therefore I'm disabling it until this is resolved.
Doing that, the TSan v2 bot in our new WebRTC FYI waterfall at
http://build.chromium.org/p/client.webrtc.fyi/waterfall
will hopefully go green.
The define used only exists for TSan v2, so v1 should be
unaffected.
The suppression added in http://review.webrtc.org/2126005/
turns out actually has no effect, so I'm removing it again.
It's probably useless since TSan v2 obviously alters the code's
behavior, causing these errors. Since the errors are not TSan v2
failures, the suppression has no effect.
BUG=2259
TEST=test passing locally when run as described in 2259 (since
no tests are run).
R=tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2115005
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4639 4adac7df-926f-26a2-2b94-8c16560cd09d
This test goes into an eternal loop sometimes when run under
TSan v2, therefore I'm disabling it until this is resolved.
Doing that, our new WebRTC FYI waterfall with a TSan v2 bot will hopefully go green when deployed (see https://codereview.chromium.org/23231004)
The define used only exists for TSan v2, so v1 should be
unaffected.
BUG=2259
TEST=test passing locally when run as described in 2259 (since
no tests are run).
R=tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2126005
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4635 4adac7df-926f-26a2-2b94-8c16560cd09d
