Reason for revert:
Causes a new TSan race warning. Will reland after fixing. Note this is the same race as will be fixed by https://codereview.webrtc.org/2876273002/.
Original issue's description:
> Fixing potential AsyncInvoker deadlock that occurs for "reentrant" invocations.
>
> The deadlock occurs if the AsyncInvoker is destroyed on thread A while
> a task on thread B is running, which AsyncInvokes a task back on thread
> A.
>
> This was causing pending_invocations_ to end up negative, because
> an AsyncClosure that's never added to a thread's message queue (due to
> the "destroying_" flag) caused the count to be decremented but not
> incremented.
>
> BUG=None
>
> Review-Url: https://codereview.webrtc.org/2885143006
> Cr-Commit-Position: refs/heads/master@{#18225}
> Committed: ef37ca5fb3TBR=nisse@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=None
Review-Url: https://codereview.webrtc.org/2904543003
Cr-Commit-Position: refs/heads/master@{#18248}
The deadlock occurs if the AsyncInvoker is destroyed on thread A while
a task on thread B is running, which AsyncInvokes a task back on thread
A.
This was causing pending_invocations_ to end up negative, because
an AsyncClosure that's never added to a thread's message queue (due to
the "destroying_" flag) caused the count to be decremented but not
incremented.
BUG=None
Review-Url: https://codereview.webrtc.org/2885143006
Cr-Commit-Position: refs/heads/master@{#18225}
The documentation for AsyncInvoker states that it owns the lifetime of
calls, and when its destructor is called, all in-flight calls are
cancelled or finish executing. The "cancelled" part is working, but if
a call is in the middle of executing, the destructor does *not* wait.
This is fixed by keeping a count of pending invocations, which is
decremented when a call is either cleared from a message queue or
finishes executing.
BUG=webrtc:3914, webrtc:3911
Review-Url: https://codereview.webrtc.org/2694723004
Cr-Commit-Position: refs/heads/master@{#16811}
Every message will now be traced with the location from which it was
posted, including function name, file and line number.
This CL also writes a normal LOG message when the dispatch took more
than a certain amount of time (currently 50ms).
This logging should help us identify messages that are taking
longer than expected to be dispatched.
R=pthatcher@webrtc.org, tommi@webrtc.org
Review URL: https://codereview.webrtc.org/2019423006 .
Cr-Commit-Position: refs/heads/master@{#13104}
But keep #including scoped_ptr.h in .h files, so as not to break
WebRTC users who expect those .h files to give them rtc::scoped_ptr.
BUG=webrtc:5520
Review-Url: https://codereview.webrtc.org/1937693002
Cr-Commit-Position: refs/heads/master@{#12581}
This propagated into various other places. Also had to #include headers that
were implicitly pulled by "scoped_ptr.h".
BUG=webrtc:5520
Review URL: https://codereview.webrtc.org/1920043002
Cr-Commit-Position: refs/heads/master@{#12501}
This CL fixes a race where for Thread objects the parent MessageQueue
constructor registers the object in the MessageQueueManager even though
the Thread is not constructed completely yet. Same happens during
destruction.
BUG=webrtc:1225
Review URL: https://codereview.webrtc.org/1666863002
Cr-Commit-Position: refs/heads/master@{#11497}
Macro incorrectly displays DISABLED_ON_ANDROID in test names for
parameterized tests under --gtest_list_tests, causing tests to be
disabled on all platforms since they contain the DISABLED_ prefix rather
than their expanded variants.
This expands the macro variants to inline if they're disabled or not,
and removes building some tests under configurations where they should
fail, instead of building them but disabling them by default.
The change also removes gtest_disable.h as an unused include from many
other files.
BUG=webrtc:5387, webrtc:5400
R=kjellander@webrtc.org, phoglund@webrtc.orgTBR=henrik.lundin@webrtc.org
Review URL: https://codereview.webrtc.org/1547343002 .
Cr-Commit-Position: refs/heads/master@{#11150}
* Move PlatformThread to rtc::.
* Remove ::CreateThread factory method.
* Make non-scoped_ptr from a lot of invocations.
* Make Start/Stop void.
* Remove rtc::Thread priorities, which were unused and would collide.
* Add ::IsRunning() to PlatformThread.
BUG=
R=tommi@webrtc.org
Review URL: https://codereview.webrtc.org/1476453002 .
Cr-Commit-Position: refs/heads/master@{#10812}
The flag used in thread_unittest.cc:FunctorB is subject to a (mostly
harmless) data race. In a tsan build, reproduce using
out/Release/rtc_unittests --gtest_filter=AsyncInvokeTest.FireAndForget
There are additional tsan warnings, not all deterministic, when
running all the rtc_unittets: Some data races related to destructors,
and a locking-order-inversion warning. Hence applying this patch does
not make the unit tests tsan-clean.
I should also add that this is my very first cl, so I'm not at all
familiar with the process.
Review URL: https://codereview.webrtc.org/1439613004
Cr-Commit-Position: refs/heads/master@{#10645}
Previously if thread A->Send is called on thread B, B->ReceiveSends will be called, which enables an arbitrary thread to invoke calls on B while B is wait for A->Send to return. This caused mutliple problems like issue 3559, 3579.
The fix is to limit B->ReceiveSends to only process requests from A.
Also disallow the worker thread invoking other threads.
BUG=3559
R=juberti@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/15089004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7290 4adac7df-926f-26a2-2b94-8c16560cd09d
The synchronization access is unnecessary for rtc::Thread::WrapCurrent (called from JingleThreadWrapper) since JingleThreadWrapper never calls rtc::Thread::Stop or rtc::Thread::Join. Failing to get the access caused crashes in Chrome since rtc::Thread::Current will be NULL when rtc::Thread::WrapCurrent fails.
rtc::ThreadManager::WrapCurrentThread still requires the synchronization access, since I am not sure if the callers (e.g. the plugin) depends on it.
BUG=crbug/413853
R=juberti@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/30429004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7224 4adac7df-926f-26a2-2b94-8c16560cd09d
- Thread::Release() wrote a local variable on the calling thread but read it on
another thread, with no synchronization. Happily it has no non-test callers
so deleting it instead of trying to fix it (see bug for details).
- Thread::started_ similarly was racily being written to; replaced with a
running_ Event, and hid the accessor except for tests & legacy callers,
with a note about why it's a bad idea.
webrtc/base patched with:
git diff origin --relative=talk/base | patch -p1 -dwebrtc/base
followed by manual merge of 3 thunks that ran afoul of naming differences
between talk/base and webrtc/base.
BUG=3388
R=andrew@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/14589005
git-svn-id: http://webrtc.googlecode.com/svn/trunk@6236 4adac7df-926f-26a2-2b94-8c16560cd09d
