Closed Bug 1079621 Opened 11 years ago Closed 11 years ago

Intermittent test_conformance__textures__tex-image-and-sub-image-2d-with-video.html | Assertion count 1 is greater than expected range (ASSERTION: Couldn't pass frame to decoder: 'Error', file content/media/fmp4/apple/AppleVTDecoder.cpp, line 243)

Categories

(Core :: Audio/Video, defect)

34 Branch
x86_64
macOS
defect
Not set
normal

Tracking

()

RESOLVED FIXED
mozilla36
Tracking Status
firefox33 --- unaffected
firefox34 --- wontfix
firefox35 --- fixed
firefox36 --- fixed
firefox-esr31 --- unaffected

People

(Reporter: RyanVM, Assigned: jya)

Details

(Keywords: assertion, intermittent-failure)

Attachments

(4 files, 2 obsolete files)

https://tbpl.mozilla.org/php/getParsedLog.php?id=48649907&tree=Mozilla-Central Rev5 MacOSX Mountain Lion 10.8 mozilla-central debug test mochitest-2 on 2014-09-22 21:08:12 PDT for push 603dad513d7c slave: talos-mtnlion-r5-023 21:25:29 INFO - [Parent 1291] WARNING: VideoToolbox decoder returned no data: file /builds/slave/m-cen-osx64-d-0000000000000000/build/content/media/fmp4/apple/AppleVTDecoder.cpp, line 165 21:25:29 INFO - [Parent 1291] ###!!! ASSERTION: Couldn't pass frame to decoder: 'Error', file /builds/slave/m-cen-osx64-d-0000000000000000/build/content/media/fmp4/apple/AppleVTDecoder.cpp, line 243 21:25:30 INFO - JavaScript error: jar:file:///builds/slave/talos-slave/test/build/application/FirefoxNightlyDebug.app/Contents/MacOS/omni.ja!/components/nsHandlerService.js, line 891: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIProperties.get] 21:25:45 INFO - JavaScript error: jar:file:///builds/slave/talos-slave/test/build/application/FirefoxNightlyDebug.app/Contents/MacOS/omni.ja!/components/nsHandlerService.js, line 891: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIProperties.get] 21:25:56 INFO - nsRunnableMethodImpl<tag_nsresult (mozilla::AppleVTDecoder::*)(mp4_demuxer::MP4Sample*), nsAutoPtr<mp4_demuxer::MP4Sample>, true>::Run() [xpcom/glue/nsThreadUtils.h:363] 21:25:56 INFO - mozilla::MediaTaskQueue::Runner::Run() [mfbt/RefPtr.h:301] 21:25:56 INFO - nsThreadPool::Run() [xpcom/glue/nsCOMPtr.h:520] 21:25:56 INFO - _ZThn8_N12nsThreadPool3RunEv [obj-firefox/xpcom/threads/Unified_cpp_xpcom_threads0.cpp:234] 21:25:56 INFO - nsThread::ProcessNextEvent(bool, bool*) [xpcom/threads/nsThread.cpp:823] 21:25:56 INFO - NS_ProcessNextEvent(nsIThread*, bool) [xpcom/glue/nsThreadUtils.cpp:265] 21:25:56 INFO - mozilla::ipc::MessagePumpForNonMainThreads::Run(base::MessagePump::Delegate*) [ipc/glue/MessagePump.cpp:339] 21:25:56 INFO - MessageLoop::Run() [ipc/chromium/src/base/message_loop.cc:509] 21:25:56 INFO - nsThread::ThreadFunc(void*) [xpcom/threads/nsThread.cpp:359] 21:25:57 INFO - _pt_root [nsprpub/pr/src/pthreads/ptthread.c:215] 21:25:57 INFO - libsystem_c.dylib + 0x14782 21:25:57 INFO - OpenGL version detected: 210 21:25:57 INFO - OpenGL vendor: Intel Inc. 21:25:57 INFO - OpenGL renderer: Intel HD Graphics 3000 OpenGL Engine 21:25:57 INFO - [Parent 1291] WARNING: We can't support non-premultiplied alpha for video!: '(aSurfaceFlags & SFE_PREFER_NO_PREMULTIPLY_ALPHA) == 0', file /builds/slave/m-cen-osx64-d-0000000000000000/build/layout/base/nsLayoutUtils.cpp, line 5901 21:25:57 INFO - [Parent 1291] WARNING: We can't support non-premultiplied alpha for video!: '(aSurfaceFlags & SFE_PREFER_NO_PREMULTIPLY_ALPHA) == 0', file /builds/slave/m-cen-osx64-d-0000000000000000/build/layout/base/nsLayoutUtils.cpp, line 5901 21:25:57 INFO - [Parent 1291] WARNING: We can't support non-premultiplied alpha for video!: '(aSurfaceFlags & SFE_PREFER_NO_PREMULTIPLY_ALPHA) == 0', file /builds/slave/m-cen-osx64-d-0000000000000000/build/layout/base/nsLayoutUtils.cpp, line 5901 21:25:57 INFO - [Parent 1291] WARNING: We can't support non-premultiplied alpha for video!: '(aSurfaceFlags & SFE_PREFER_NO_PREMULTIPLY_ALPHA) == 0', file /builds/slave/m-cen-osx64-d-0000000000000000/build/layout/base/nsLayoutUtils.cpp, line 5901 21:25:57 INFO - 1040 INFO TEST-OK | /tests/dom/canvas/test/webgl-conformance/_wrappers/test_conformance__textures__tex-image-and-sub-image-2d-with-video.html | took 8638ms 21:25:57 INFO - [Parent 1291] WARNING: Failed to unlock the wakelock.: '!rv.Failed()', file /builds/slave/m-cen-osx64-d-0000000000000000/build/content/html/content/src/HTMLVideoElement.cpp, line 245 21:25:57 INFO - ++DOMWINDOW == 32 (0x125db4c00) [pid = 1291] [serial = 2682] [outer = 0x124483400] 21:25:57 INFO - 1041 INFO TEST-UNEXPECTED-ERROR | /tests/dom/canvas/test/webgl-conformance/_wrappers/test_conformance__textures__tex-image-and-sub-image-2d-with-video.html | Assertion count 1 is greater than expected range 0-0 assertions.
Assignee: nobody → jyavenard
Version: Trunk → 34 Branch
should I still apply this fix for the 34/beta branch ?
Flags: needinfo?(ryanvm)
Sure :)
Flags: needinfo?(ryanvm)
Backported for 34 branch
Attachment #8504431 - Flags: review?(kinetik)
Attachment #8504431 - Flags: review?(kinetik) → review+
Comment on attachment 8504431 [details] [diff] [review] Return error instead of asserting Approval Request Comment [Feature/regressing bug #]:1043696 [User impact if declined]:will fail mochitest and may cause crashes [Describe test coverage new/current, TBPL]:code has been in use in nightly for a month [Risks and why]: can't foresee any. [String/UUID change made/needed]: none
Attachment #8504431 - Flags: approval-mozilla-beta?
Comment on attachment 8504431 [details] [diff] [review] Return error instead of asserting Beta+
Attachment #8504431 - Flags: approval-mozilla-beta? → approval-mozilla-beta+
Status: NEW → RESOLVED
Closed: 11 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla35
Wait a sec, this was the spinoff from bug 1069005. The beta patch here was for that one, no?
Target Milestone: mozilla35 → ---
Status: RESOLVED → REOPENED
Flags: needinfo?(jyavenard)
Resolution: FIXED → ---
the patch here is for beta (well, it was for aurora, but that one became beta in the mean time)... bug 1069005 was for nightly, but it got fixed as a side effect of resolving bug 1066369, which couldn't be applied without rewrite on aurora (now beta) Hope that makes sense... So it's all good :)
Flags: needinfo?(jyavenard)
But the patch here fixed the "Couldn't pass frame to decoder: 'rv == noErr'" assert bug 1066369 was filed for, not the "Couldn't pass frame to decoder: 'Error'" assert (on trunk no less) that this was spun off for.
s/1066369/1069005
I see. They are both fixed with this patch... 3 potential assertions were removed... Why it asserted may have been for different reasons, but the end result was the same: it was a simple error and should have been treated as such.
It's confusing because bug 1069005 comment 3 (what this bug was filed for) happened on trunk after bug 1066369 landed.
(In reply to Ryan VanderMeulen [:RyanVM UTC-4] from comment #12) > It's confusing because bug 1069005 comment 3 (what this bug was filed for) > happened on trunk after bug 1066369 landed. I'm tracking the revision on that file. I don't even see what could could have ever created that assertion... unless NS_ERROR cause an assertion as far as the test bot are concerned. Ah yes it does in debug mode... Then yes, you're right, that patch applies to bug 1066369 and that should still be uplifted to beta IMHO... Why the Apple decoder fails at random from time to time is still unexplained... but the error is usually recovered after the next packet and isn't a problem per say... not much we can do about it
I think there's enough hits here to prove the point now that this is still an ongoing issue independent of bug 1066369.
Flags: needinfo?(jyavenard)
Make non-fatal errors warnings only.
Attachment #8517783 - Flags: review?(giles)
Comment on attachment 8517783 [details] [diff] [review] Change non-fatal errors to warnings, fatal ones to errors Review of attachment 8517783 [details] [diff] [review]: ----------------------------------------------------------------- The theory is that NS_ERROR() triggers an assertion in debug builds, which tinderbox reports as a failure. NS_WARNING calls NS_DebugBreak, but the lower severity skips invoking the debugger. Please put that in the body of your commit message and fix the summary. The whole point of the patch is to _not_ make the fatal ones errors. I still have reservations about papering over the issue like this, but I suppose it's better to fix the intermittent test failures. r=me with the commit message fixed.
Attachment #8517783 - Flags: review?(giles) → review+
Commit message should be like: Bug 1079621: Change non-fatal errors to warnings. r=rillian This resolves intermittent failures in continuous integration. NS_ERROR() triggers an assertion in debug builds, which tinderbox reports as a failure. NS_WARNING calls NS_DebugBreak, but with the lower severity level it skips invoking the debugger.
Carrying r+
Attachment #8517783 - Attachment is obsolete: true
Flags: needinfo?(jyavenard)
Keywords: checkin-needed
Status: REOPENED → RESOLVED
Closed: 11 years ago11 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla36
Can we land this on Aurora/Beta please? Seems pretty low-risk :)
Flags: needinfo?(jyavenard)
Aurora patch
Attachment #8521075 - Flags: review?(giles)
Attachment #8504431 - Attachment is obsolete: true
rename patch file. Simple backport to already r+ patch
Attachment #8504431 - Attachment is obsolete: false
Flags: needinfo?(jyavenard)
Attachment #8521075 - Attachment is obsolete: true
Attachment #8521075 - Flags: review?(giles)
Backport for beta
Comment on attachment 8521077 [details] [diff] [review] Aurora: Change non-fatal errors to warnings Approval Request Comment [Feature/regressing bug #]:1043696 [User impact if declined]:will fail mochitest and may cause crashes [Describe test coverage new/current, TBPL]:code has been in use in nightly for a month [Risks and why]: can't foresee any. [String/UUID change made/needed]: none
Attachment #8521077 - Flags: approval-mozilla-aurora?
Comment on attachment 8521082 [details] [diff] [review] Beta: Change non-fatal errors to warnings Approval Request Comment [Feature/regressing bug #]:1043696 [User impact if declined]:will fail mochitest and may cause crashes [Describe test coverage new/current, TBPL]:code has been in use in nightly for a month [Risks and why]: can't foresee any. [String/UUID change made/needed]: none
Attachment #8521082 - Flags: approval-mozilla-beta?
Comment on attachment 8521082 [details] [diff] [review] Beta: Change non-fatal errors to warnings Beta's crash rate is in a good range for release. Without knowing that this causes frequent or serious crashes I suggest that even though this is a simple change that we ship it in 35. Please renom if you really think this fix should ship in 34.
Attachment #8521082 - Flags: approval-mozilla-beta? → approval-mozilla-beta-
Attachment #8521077 - Flags: approval-mozilla-aurora? → approval-mozilla-aurora+
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: