Closed Bug 1656068 Opened 4 years ago Closed 4 years ago

SUMMARY: ThreadSanitizer: data race /builds/worker/checkouts/gecko/media/webrtc/trunk/webrtc/modules/audio_processing/aec/echo_cancellation.cc:131:43 in webrtc::WebRtcAec_Create()

Tracking

()

Status:

RESOLVED FIXED

Milestone:

83 Branch

Tracking Flags:

Tracking

Status

firefox81

---

wontfix

firefox82

---

wontfix

firefox83

---

fixed

People

(Reporter: intermittent-bug-filer, Assigned: padenot)

References

Details

Attachments

(7 files)

stacks for the accesses 4 years ago Paul Adenot (:padenot) 12.40 KB, text/plain		Details
stacks.txt 4 years ago Paul Adenot (:padenot) 12.41 KB, text/plain		Details
Bug 1656068 - Do the processing for audio input data on the graph runner thread and not the audio callback thread. r?pehrsons 4 years ago Paul Adenot (:padenot) 47 bytes, text/x-phabricator-request		Details \| Review
Detailed Crash Information 4 years ago Christian Holler (:decoder) 42.95 KB, text/plain		Details
Bug 1656068 - Make instance_count in echo_cancellation.h atomic, it can be read and written concurrently. r?pehrsons 4 years ago Paul Adenot (:padenot) 47 bytes, text/x-phabricator-request		Details \| Review
Bug 1656068 - Backout first patch in 1656068 for breaking bluetooth audio input on Linux/Pulse. r?pehrsons 4 years ago Paul Adenot (:padenot) 47 bytes, text/x-phabricator-request		Details \| Review
Bug 1656068 - Add temporary suppression. r?bwc 4 years ago Christian Holler (:decoder) 47 bytes, text/x-phabricator-request		Details \| Review

Treeherder Bug Filer

Reporter

Description

•

4 years ago

treeherder

Filed by: bcampen [at] mozilla.com
Parsed log: https://treeherder.mozilla.org/logviewer.html#?job_id=311427427&repo=try
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/Tj6fDsfMQXWQioRT1H82lw/runs/0/artifacts/public/logs/live_backing.log

Byron Campen [:bwc]

Updated

•

4 years ago

Blocks: mochitest_media_tsan

Byron Campen [:bwc]

Comment 1

•

4 years ago

Looks like multiple threads are accessing this:

https://searchfox.org/mozilla-central/rev/cffd9b5302b6b6f51533d895a785b48ff418aec1/media/webrtc/trunk/webrtc/modules/audio_processing/aec/echo_cancellation.cc#102

We did not manage to restore the other stack, and searchfox is having trouble figuring out where this variable is accessed for some reason, possibly because we're accessing a static member variable with "->"? Doing a text search, I don't see anything using this besides WebRtcAec_Create:

https://searchfox.org/mozilla-central/search?q=instance_count&case=true&path=media%2Fwebrtc%2Ftrunk%2Fwebrtc

Looks like more than one thread is calling this.

Byron Campen [:bwc]

Comment 2

•

4 years ago

Current code in Chromium looks similar.

https://source.chromium.org/chromiumos/chromiumos/codesearch/+/master:src/third_party/webrtc-apm/modules/audio_processing/aec/echo_cancellation.h;l=111?originalUrl=https:%2F%2Fcs.chromium.org%2F

https://source.chromium.org/chromiumos/chromiumos/codesearch/+/master:src/third_party/webrtc-apm/modules/audio_processing/aec/echo_cancellation.cc;l=122?originalUrl=https:%2F%2Fcs.chromium.org%2F

Paul Adenot (:padenot)

Assignee

Updated

•

4 years ago

Assignee: nobody → padenot

Paul Adenot (:padenot)

Assignee

Comment 3

•

4 years ago

•

Edited

Attached file stacks for the accesses — Details

The stacks for this one.

Comment hidden (obsolete)

Paul Adenot (:padenot)

Assignee

Comment 5

•

4 years ago

The real cause for this is that we're not doing input processing in the graph runner thread, but on the audioipc/audio callback thread. The problem is that we run control messages from the graph runner thread, so we reconfigure the aec from the graph runner thread.

This is all synchronized via the graph runner lock. It would be best however to do everything in the graph runner thread.

Paul Adenot (:padenot)

Assignee

Comment 6

•

4 years ago

Attached file Bug 1656068 - Do the processing for audio input data on the graph runner thread and not the audio callback thread. r?pehrsons — Details

Comment hidden (Intermittent Failures Robot)

Phabricator Automation

Updated

•

4 years ago

Attachment #9167032 - Attachment description: Bug 1656068 - Do the processing for audio input data on the graph runner thread and not the audio callback thread. r?karlt → Bug 1656068 - Do the processing for audio input data on the graph runner thread and not the audio callback thread. r?pehrsons

Pulsebot

Comment 8

•

4 years ago

Pushed by padenot@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/e8dd0a07098b
Do the processing for audio input data on the graph runner thread and not the audio callback thread. r=pehrsons

Andreea Pavel [:apavel]

Comment 9

•

4 years ago

bugherder

https://hg.mozilla.org/mozilla-central/rev/e8dd0a07098b

Status: NEW → RESOLVED

Closed: 4 years ago

status-firefox81: --- → fixed

Resolution: --- → FIXED

Target Milestone: --- → 81 Branch

Karl Tomlinson (:karlt)

Comment 10

•

4 years ago

Read of size 4 at 0x7fa2380759b0 by thread T77 (mutexes: write M405459446912339208, write M414185342989784144, write M414466817966494856):
  #0 webrtc::WebRtcAec_Create() /builds/worker/checkouts/gecko/media/webrtc/trunk/webrtc/modules/audio_processing/aec/echo_cancellation.cc:131:43 (libxul.so+0x5df666e)
  #1 Canceller /builds/worker/checkouts/gecko/media/webrtc/trunk/webrtc/modules/audio_processing/echo_cancellation_impl.cc:83:14 (libxul.so+0x5e06f68)
  #2 webrtc::EchoCancellationImpl::Initialize(int, unsigned long, unsigned long, unsigned long) /builds/worker/checkouts/gecko/media/webrtc/trunk/webrtc/modules/audio_processing/echo_cancellation_impl.cc:433:32 (libxul.so+0x5e06f68)
  #3 webrtc::EchoCancellationImpl::Enable(bool) /builds/worker/checkouts/gecko/media/webrtc/trunk/webrtc/modules/audio_processing/echo_cancellation_impl.cc:220:5 (libxul.so+0x5e06d6f)
  #4 mozilla::AudioInputProcessing::UpdateAECSettings(bool, bool, webrtc::EchoCancellation::SuppressionLevel, webrtc::EchoControlMobile::RoutingMode) /builds/worker/checkouts/gecko/dom/media/webrtc/MediaEngineWebRTCAudio.cpp:707:5 (libxul.so+0x40f7e22)
  #5 mozilla::MediaEngineWebRTCMicrophoneSource::UpdateAECSettings(bool, bool, webrtc::EchoCancellation::SuppressionLevel, webrtc::EchoControlMobile::RoutingMode)::$_10::operator()() const::Message::Run() /builds/worker/checkouts/gecko/dom/media/webrtc/MediaEngineWebRTCAudio.cpp:189:31 (libxul.so+0x410ac26)
  #6 mozilla::MediaTrackGraphImpl::RunMessagesInQueue() /builds/worker/checkouts/gecko/dom/media/MediaTrackGraph.cpp:1151:20 (libxul.so+0x3d652f8)
  #7 mozilla::MediaTrackGraphImpl::OneIterationImpl(long, long, mozilla::AudioMixer*) /builds/worker/checkouts/gecko/dom/media/MediaTrackGraph.cpp:1394:3 (libxul.so+0x3d672d7)
  #8 mozilla::GraphRunner::Run() /builds/worker/checkouts/gecko/dom/media/GraphRunner.cpp:114:32 (libxul.so+0x3c00f20)
  #9 nsThread::ProcessNextEvent(bool, bool*) /builds/worker/checkouts/gecko/xpcom/threads/nsThread.cpp:1234:14 (libxul.so+0xb23bdb)
  #10 NS_ProcessNextEvent(nsIThread*, bool) /builds/worker/checkouts/gecko/xpcom/threads/nsThreadUtils.cpp:513:10 (libxul.so+0xb28dd2)
  #11 mozilla::ipc::MessagePumpForNonMainThreads::Run(base::MessagePump::Delegate*) /builds/worker/checkouts/gecko/ipc/glue/MessagePump.cpp:302:20 (libxul.so+0x13f586e)
  #12 RunInternal /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:334:10 (libxul.so+0x138429c)
  #13 RunHandler /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:327:3 (libxul.so+0x138429c)
  #14 MessageLoop::Run() /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:309:3 (libxul.so+0x138429c)
  #15 nsThread::ThreadFunc(void*) /builds/worker/checkouts/gecko/xpcom/threads/nsThread.cpp:447:10 (libxul.so+0xb1fc65)
  #16 _pt_root /builds/worker/checkouts/gecko/nsprpub/pr/src/pthreads/ptthread.c:201:5 (libnspr4.so+0x44daf)
Previous write of size 4 at 0x7fa2380759b0 by thread T27:
  [failed to restore the stack]
Location is global 'webrtc::Aec::instance_count' of size 4 at 0x7fa2380759b0 (libxul.so+0x00000b50a9b0)
Mutex M405459446912339208 is already destroyed.
Mutex M414185342989784144 is already destroyed.
Mutex M414466817966494856 is already destroyed.
Thread T77 'GraphRunner' (tid=2755, running) created by main thread at:
  #0 pthread_create /builds/worker/fetches/llvm-project/llvm/projects/compiler-rt/lib/tsan/rtl/tsan_interceptors.cc:967:3 (firefox+0x584eb)
  #1 _PR_CreateThread /builds/worker/checkouts/gecko/nsprpub/pr/src/pthreads/ptthread.c:458:14 (libnspr4.so+0x3bfb3)
  #2 PR_CreateThread /builds/worker/checkouts/gecko/nsprpub/pr/src/pthreads/ptthread.c:533:12 (libnspr4.so+0x308e2)
  #3 nsThread::Init(nsTSubstring<char> const&) /builds/worker/checkouts/gecko/xpcom/threads/nsThread.cpp:659:8 (libxul.so+0xb2122e)
  #4 nsThreadManager::NewNamedThread(nsTSubstring<char> const&, unsigned int, nsIThread**) /builds/worker/checkouts/gecko/xpcom/threads/nsThreadManager.cpp:629:12 (libxul.so+0xb284ca)
  #5 NS_NewNamedThread(nsTSubstring<char> const&, nsIThread**, already_AddRefed<nsIRunnable>, unsigned int) /builds/worker/checkouts/gecko/xpcom/threads/nsThreadUtils.cpp:161:57 (libxul.so+0xb2e958)
  #6 NS_NewNamedThread<12> /builds/worker/workspace/obj-build/dist/include/nsThreadUtils.h:85:10 (libxul.so+0x3c006a4)
  #7 mozilla::GraphRunner::Create(mozilla::MediaTrackGraphImpl*) /builds/worker/checkouts/gecko/dom/media/GraphRunner.cpp:37:7 (libxul.so+0x3c006a4)
  #8 mozilla::MediaTrackGraphImpl::MediaTrackGraphImpl(mozilla::MediaTrackGraph::GraphDriverType, mozilla::MediaTrackGraph::GraphRunType, int, unsigned int, void const*, mozilla::AbstractThread*) /builds/worker/checkouts/gecko/dom/media/MediaTrackGraph.cpp:3007:26 (libxul.so+0x3d6eee6)
  #9 mozilla::MediaTrackGraph::GetInstance(mozilla::MediaTrackGraph::GraphDriverType, nsPIDOMWindowInner*, int, void const*) /builds/worker/checkouts/gecko/dom/media/MediaTrackGraph.cpp:3146:17 (libxul.so+0x3d6f6ce)
  #10 mozilla::GetUserMediaStreamRunnable::Run() /builds/worker/checkouts/gecko/dom/media/MediaManager.cpp:1199:28 (libxul.so+0x3d36f8e)
  #11 mozilla::RunnableTask::Run() /builds/worker/checkouts/gecko/xpcom/threads/TaskController.cpp:242:16 (libxul.so+0xb0d077)
  #12 mozilla::TaskController::DoExecuteNextTaskOnlyMainThreadInternal(mozilla::detail::BaseAutoLock<mozilla::Mutex&> const&) /builds/worker/checkouts/gecko/xpcom/threads/TaskController.cpp:512:26 (libxul.so+0xb0a8b5)
  #13 mozilla::TaskController::ExecuteNextTaskOnlyMainThreadInternal(mozilla::detail::BaseAutoLock<mozilla::Mutex&> const&) /builds/worker/checkouts/gecko/xpcom/threads/TaskController.cpp:371:15 (libxul.so+0xb09546)
  #14 mozilla::TaskController::ProcessPendingMTTask(bool) /builds/worker/checkouts/gecko/xpcom/threads/TaskController.cpp:168:36 (libxul.so+0xb097e4)
  #15 operator() /builds/worker/checkouts/gecko/xpcom/threads/TaskController.cpp:83:37 (libxul.so+0xb10cc4)
  #16 mozilla::detail::RunnableFunction<mozilla::TaskController::InitializeInternal()::$_4>::Run() /builds/worker/workspace/obj-build/dist/include/nsThreadUtils.h:577:5 (libxul.so+0xb10cc4)
  #17 nsThread::ProcessNextEvent(bool, bool*) /builds/worker/checkouts/gecko/xpcom/threads/nsThread.cpp:1234:14 (libxul.so+0xb23970)
  #18 NS_ProcessNextEvent(nsIThread*, bool) /builds/worker/checkouts/gecko/xpcom/threads/nsThreadUtils.cpp:513:10 (libxul.so+0xb28dd2)
  #19 mozilla::ipc::MessagePump::Run(base::MessagePump::Delegate*) /builds/worker/checkouts/gecko/ipc/glue/MessagePump.cpp:87:21 (libxul.so+0x13f4d69)
  #20 mozilla::ipc::MessagePumpForChildProcess::Run(base::MessagePump::Delegate*) /builds/worker/checkouts/gecko/ipc/glue/MessagePump.cpp:270:30 (libxul.so+0x13f577b)
  #21 RunInternal /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:334:10 (libxul.so+0x138429c)
  #22 RunHandler /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:327:3 (libxul.so+0x138429c)
  #23 MessageLoop::Run() /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:309:3 (libxul.so+0x138429c)
  #24 nsBaseAppShell::Run() /builds/worker/checkouts/gecko/widget/nsBaseAppShell.cpp:137:27 (libxul.so+0x4925273)
  #25 XRE_RunAppShell() /builds/worker/checkouts/gecko/toolkit/xre/nsEmbedFunctions.cpp:913:20 (libxul.so+0x6560a99)
  #26 mozilla::ipc::MessagePumpForChildProcess::Run(base::MessagePump::Delegate*) /builds/worker/checkouts/gecko/ipc/glue/MessagePump.cpp:237:9 (libxul.so+0x13f572a)
  #27 RunInternal /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:334:10 (libxul.so+0x138429c)
  #28 RunHandler /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:327:3 (libxul.so+0x138429c)
  #29 MessageLoop::Run() /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:309:3 (libxul.so+0x138429c)
  #30 XRE_InitChildProcess(int, char**, XREChildData const*) /builds/worker/checkouts/gecko/toolkit/xre/nsEmbedFunctions.cpp:744:34 (libxul.so+0x6560819)
  #31 mozilla::BootstrapImpl::XRE_InitChildProcess(int, char**, XREChildData const*) /builds/worker/checkouts/gecko/toolkit/xre/Bootstrap.cpp:67:12 (libxul.so+0x6568922)
  #32 content_process_main /builds/worker/checkouts/gecko/browser/app/../../ipc/contentproc/plugin-container.cpp:56:28 (firefox+0xca0d7)
  #33 main /builds/worker/checkouts/gecko/browser/app/nsBrowserApp.cpp:303:18 (firefox+0xca0d7)
Thread T27 'AudioIPC0' (tid=2116, running) created by thread T24 at:
  #0 pthread_create /builds/worker/fetches/llvm-project/llvm/projects/compiler-rt/lib/tsan/rtl/tsan_interceptors.cc:967:3 (firefox+0x584eb)
  #1 std::sys::unix::thread::Thread::new::h77964362e8ddd9a4 /rustc/dd67187965e136bff1ed05e035293441c60f0790/src/libstd/sys/unix/thread.rs:68:19 (libxul.so+0x7e00b41)
  #2 cubeb_backend::capi::capi_init::hc5b21a181ddfb099 /builds/worker/checkouts/gecko/third_party/rust/cubeb-backend/src/capi.rs:68:25 (libxul.so+0x757145f)
  #3 audioipc_client_init /builds/worker/checkouts/gecko/media/audioipc/client/src/lib.rs:83:5 (libxul.so+0x757145f)
  #4 cubeb_backend::capi::capi_init::hc5b21a181ddfb099 /builds/worker/checkouts/gecko/third_party/rust/cubeb-backend/src/capi.rs:68:25 (libxul.so+0x757145f)
  #5 audioipc_client_init /builds/worker/checkouts/gecko/media/audioipc/client/src/lib.rs:83:5 (libxul.so+0x757145f)
  #6 mozilla::CubebUtils::GetCubebContextUnlocked() /builds/worker/checkouts/gecko/dom/media/CubebUtils.cpp:498:9 (libxul.so+0x3b95387)
  #7 mozilla::CubebUtils::GetCubebContext() /builds/worker/checkouts/gecko/dom/media/CubebUtils.cpp:298:10 (libxul.so+0x3b95028)
  #8 mozilla::CubebDeviceEnumerator::CubebDeviceEnumerator() /builds/worker/checkouts/gecko/dom/media/webrtc/CubebDeviceEnumerator.cpp:62:9 (libxul.so+0x40e47a2)
  #9 mozilla::CubebDeviceEnumerator::GetInstance() /builds/worker/checkouts/gecko/dom/media/webrtc/CubebDeviceEnumerator.cpp:30:21 (libxul.so+0x40e4430)
  #10 GetEnumerator /builds/worker/checkouts/gecko/dom/media/webrtc/MediaEngineWebRTC.cpp:32:10 (libxul.so+0x40e6056)
  #11 mozilla::MediaEngineWebRTC::MediaEngineWebRTC(mozilla::MediaEnginePrefs&) /builds/worker/checkouts/gecko/dom/media/webrtc/MediaEngineWebRTC.cpp:45:7 (libxul.so+0x40e6056)
  #12 mozilla::MediaManager::GetBackend() /builds/worker/checkouts/gecko/dom/media/MediaManager.cpp:3340:20 (libxul.so+0x3c72561)
  #13 operator() /builds/worker/checkouts/gecko/dom/media/MediaManager.cpp:1802:30 (libxul.so+0x3d2d5c6)
  #14 mozilla::media::LambdaTask<mozilla::MediaManager::EnumerateRawDevices(unsigned long, mozilla::dom::MediaSourceEnum, mozilla::dom::MediaSourceEnum, mozilla::MediaSinkEnum, mozilla::MediaManager::DeviceEnumerationType, mozilla::MediaManager::DeviceEnumerationType, bool, RefPtr<mozilla::media::Refcountable<nsTArray<RefPtr<mozilla::MediaDevice> > > > const&)::$_67>::Run() /builds/worker/workspace/obj-build/dist/include/mozilla/media/MediaTaskUtils.h:32:5 (libxul.so+0x3d2d5c6)
  #15 mozilla::TaskQueue::Runner::Run() /builds/worker/checkouts/gecko/xpcom/threads/TaskQueue.cpp:158:20 (libxul.so+0xb12f22)
  #16 nsThreadPool::Run() /builds/worker/checkouts/gecko/xpcom/threads/nsThreadPool.cpp:299:14 (libxul.so+0xb2b286)
  #17 nsThread::ProcessNextEvent(bool, bool*) /builds/worker/checkouts/gecko/xpcom/threads/nsThread.cpp:1234:14 (libxul.so+0xb23bdb)
  #18 NS_ProcessNextEvent(nsIThread*, bool) /builds/worker/checkouts/gecko/xpcom/threads/nsThreadUtils.cpp:513:10 (libxul.so+0xb28dd2)
  #19 mozilla::ipc::MessagePumpForNonMainThreads::Run(base::MessagePump::Delegate*) /builds/worker/checkouts/gecko/ipc/glue/MessagePump.cpp:302:20 (libxul.so+0x13f586e)
  #20 RunInternal /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:334:10 (libxul.so+0x138429c)
  #21 RunHandler /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:327:3 (libxul.so+0x138429c)
  #22 MessageLoop::Run() /builds/worker/checkouts/gecko/ipc/chromium/src/base/message_loop.cc:309:3 (libxul.so+0x138429c)
  #23 nsThread::ThreadFunc(void*) /builds/worker/checkouts/gecko/xpcom/threads/nsThread.cpp:447:10 (libxul.so+0xb1fc65)
  #24 _pt_root /builds/worker/checkouts/gecko/nsprpub/pr/src/pthreads/ptthread.c:201:5 (libnspr4.so+0x44daf)

Paul Adenot (:padenot)

Assignee

Comment 11

•

4 years ago

This is still a problem according to :decoder. I'll have another go at it.

Status: RESOLVED → REOPENED

Resolution: FIXED → ---

Christian Holler (:decoder)

Comment 12

•

4 years ago

This is an automated crash issue comment:

Summary: ThreadSanitizer: data race [@ webrtc::WebRtcAec_Create] vs. [@ webrtc::WebRtcAec_Create]
Build version: try revision 795a87bdf14e22f9a68ebe5fdcce1042dedb4465

For detailed crash information, see attachment.

Christian Holler (:decoder)

Comment 13

•

4 years ago

Attached file Detailed Crash Information — Details

admin

Comment 14

•

4 years ago

I'm unable to use my mic without sever artifacts and crackling.

I used an online voice recorder and mozregression to pinpoint the issue to this commit, as it stands the microphone issues mean I cannot use firefox.

2020-09-29T11:03:19.431000: INFO : Narrowed integration regression window from [d1735ab1, f3a3edb8] (3 builds) to [d1735ab1, e8dd0a07] (2 builds) (~1 steps left)
2020-09-29T11:03:19.450000: DEBUG : Starting merge handling...
2020-09-29T11:03:19.451000: DEBUG : Using url: https://hg.mozilla.org/integration/autoland/json-pushes?changeset=e8dd0a07098ba86648f246b41528d5c86a52c676&full=1
2020-09-29T11:03:19.451000: DEBUG : redo: attempt 1/3
2020-09-29T11:03:19.452000: DEBUG : redo: retry: calling _default_get with args: ('https://hg.mozilla.org/integration/autoland/json-pushes?changeset=e8dd0a07098ba86648f246b41528d5c86a52c676&full=1',), kwargs: {}, attempt #1
2020-09-29T11:03:19.457000: DEBUG : urllib3.connectionpool: Resetting dropped connection: hg.mozilla.org
2020-09-29T11:03:20.992000: DEBUG : urllib3.connectionpool: https://hg.mozilla.org:443 "GET /integration/autoland/json-pushes?changeset=e8dd0a07098ba86648f246b41528d5c86a52c676&full=1 HTTP/1.1" 200 None
2020-09-29T11:03:21.051000: DEBUG : Found commit message:
Bug 1656068 - Do the processing for audio input data on the graph runner thread and not the audio callback thread. r=pehrsons

Differential Revision: https://phabricator.services.mozilla.com/D85404

2020-09-29T11:03:21.052000: DEBUG : Did not find a branch, checking all integration branches
2020-09-29T11:03:21.057000: INFO : The bisection is done.
2020-09-29T11:03:21.059000: INFO : Stopped

Paul Adenot (:padenot)

Assignee

Comment 15

•

4 years ago

(In reply to admin from comment #14)

I'm unable to use my mic without sever artifacts and crackling.

I used an online voice recorder and mozregression to pinpoint the issue to this commit, as it stands the microphone issues mean I cannot use firefox.

2020-09-29T11:03:19.431000: INFO : Narrowed integration regression window from [d1735ab1, f3a3edb8] (3 builds) to [d1735ab1, e8dd0a07] (2 builds) (~1 steps left)
2020-09-29T11:03:19.450000: DEBUG : Starting merge handling...
2020-09-29T11:03:19.451000: DEBUG : Using url: https://hg.mozilla.org/integration/autoland/json-pushes?changeset=e8dd0a07098ba86648f246b41528d5c86a52c676&full=1
2020-09-29T11:03:19.451000: DEBUG : redo: attempt 1/3
2020-09-29T11:03:19.452000: DEBUG : redo: retry: calling _default_get with args: ('https://hg.mozilla.org/integration/autoland/json-pushes?changeset=e8dd0a07098ba86648f246b41528d5c86a52c676&full=1',), kwargs: {}, attempt #1
2020-09-29T11:03:19.457000: DEBUG : urllib3.connectionpool: Resetting dropped connection: hg.mozilla.org
2020-09-29T11:03:20.992000: DEBUG : urllib3.connectionpool: https://hg.mozilla.org:443 "GET /integration/autoland/json-pushes?changeset=e8dd0a07098ba86648f246b41528d5c86a52c676&full=1 HTTP/1.1" 200 None
2020-09-29T11:03:21.051000: DEBUG : Found commit message:
Bug 1656068 - Do the processing for audio input data on the graph runner thread and not the audio callback thread. r=pehrsons

Differential Revision: https://phabricator.services.mozilla.com/D85404

2020-09-29T11:03:21.052000: DEBUG : Did not find a branch, checking all integration branches
2020-09-29T11:03:21.057000: INFO : The bisection is done.
2020-09-29T11:03:21.059000: INFO : Stopped

Please attach a copy of the raw data from about:support.

Paul Adenot (:padenot)

Assignee

Comment 16

•

4 years ago

(In reply to Christian Holler (:decoder) from comment #12)

This is an automated crash issue comment:

Summary: ThreadSanitizer: data race [@ webrtc::WebRtcAec_Create] vs. [@
webrtc::WebRtcAec_Create]
Build version: try revision 795a87bdf14e22f9a68ebe5fdcce1042dedb4465

For detailed crash information, see attachment.

Indeed, it's possible to have two of those running at once, and this instance_count is a global flag, it needs to be atomic.

Paul Adenot (:padenot)

Assignee

Comment 17

•

4 years ago

Attached file Bug 1656068 - Make instance_count in echo_cancellation.h atomic, it can be read and written concurrently. r?pehrsons — Details

Pulsebot

Comment 18

•

4 years ago

Pushed by padenot@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/b2678e1eff71
Make instance_count in echo_cancellation.h atomic, it can be read and written concurrently. r=pehrsons

Atila Butkovits

Comment 19

•

4 years ago

bugherder

https://hg.mozilla.org/mozilla-central/rev/b2678e1eff71

Status: REOPENED → RESOLVED

Closed: 4 years ago → 4 years ago

status-firefox83: --- → fixed

Resolution: --- → FIXED

BugBot [:suhaib / :marco/ :calixte]

Comment 20

•

4 years ago

Change the status for beta to have the same as nightly and release.
For more information, please visit auto_nag documentation.

status-firefox82: --- → fixed

Andreas Pehrson [:pehrsons]

Updated

•

4 years ago

Regressions: 1667464

Paul Adenot (:padenot)

Assignee

Comment 21

•

4 years ago

We need to back this first patch out asap, everywhere, and then we can figure out something better.

Paul Adenot (:padenot)

Assignee

Comment 22

•

4 years ago

Attached file Bug 1656068 - Backout first patch in 1656068 for breaking bluetooth audio input on Linux/Pulse. r?pehrsons — Details

Phabricator Automation

Updated

•

4 years ago

Attachment #9179288 - Attachment description: Bug 1656068 - Backout first patch in 1656068 for breaking bluetooth audio input. r?pehrsons → Bug 1656068 - Backout first patch in 1656068 for breaking bluetooth audio input on Linux/Pulse. r?pehrsons

Paul Adenot (:padenot)

Assignee

Comment 23

•

4 years ago

The reason for this breakage is that pulse calls the cubeb data_callback with very small buffers, that are filled entirely from the scratch buffer. The audio input processing doesn't run, then. This happens only when using bluetooth devices because in handsfree, it's running at 16kHz or 8kHz, and so the number of frames is small, for the same duration.

Pulsebot

Comment 24

•

4 years ago

Pushed by padenot@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/7b50f2418f3d
Backout first patch in 1656068 for breaking bluetooth audio input on Linux/Pulse. r=pehrsons

Julien Cristau [:jcristau]

Comment 25

•

4 years ago

Beta backout:
https://hg.mozilla.org/releases/mozilla-beta/rev/28a6086b486a

Status: RESOLVED → REOPENED

status-firefox81: fixed → wontfix

status-firefox82: fixed → wontfix

status-firefox83: fixed → affected

Resolution: FIXED → ---

Target Milestone: 81 Branch → ---

Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)

Comment 26

•

4 years ago

Backout merged to central: https://hg.mozilla.org/mozilla-central/rev/7b50f2418f3d

Christian Holler (:decoder)

Comment 27

•

4 years ago

Attached file Bug 1656068 - Add temporary suppression. r?bwc — Details

Depends on D93509

Pulsebot

Comment 28

•

4 years ago

Pushed by choller@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/00a0f95a6140
Add temporary suppression. r=bwc

Narcis Beleuzu [:NarcisB]

Comment 29

•

4 years ago

bugherder

https://hg.mozilla.org/mozilla-central/rev/00a0f95a6140

Status: REOPENED → RESOLVED

Closed: 4 years ago → 4 years ago

status-firefox83: affected → fixed

Resolution: --- → FIXED

Target Milestone: --- → 83 Branch

You need to log in before you can comment on or make changes to this bug.