1422929 - Add custom get_error_search_term() handling for "Assertion failure: ..." error lines

Status: NEW

(Tree Management :: Treeherder, enhancement, P3)

Description

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

• The common assertion failures often hit random tests because the asserting code is deep in the backend, e.g. JavaScript garbage collection, graphic failures e.g. reserving memory or related to timing and painting.
    • If there is any intermittent failure filed for a test file, a general intermittent bug “Intermittent Assertion failure: foo at bar.cpp” won’t be suggested.
    • Suggest these general intermittent assertion failures unconditionally.

Comment 1

[Ed Morley [:emorley]]

Do you have some example logs / log lines and resultant failure panel entries that demonstrate the issue? :-)

Comment 2

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

From the sheriffs etherpad, we had those recently:
    Assertion failure: !aState.GetIsCurrentlyDecoded(), at FrameAnimator.cpp:471 https://bugzilla.mozilla.org/show_bug.cgi?id=1389988

    Assertion failure: aVsyncTimestamp <= TimeStamp::Now() https://bugzilla.mozilla.org/show_bug.cgi?id=1367182

    Assertion failure: [GFX1]: BlitImage(SurfaceTextureImage) not implemented., at Logging.h:520 https://bugzilla.mozilla.org/show_bug.cgi?id=1401090

    Assertion failure: false (Ran out of memory while building cycle collector graph), at nsCycleCollector.cpp:930 https://bugzilla.mozilla.org/show_bug.cgi?id=1392228

    Assertion failure: firstTimeStamp == mTimers[0]->Timeout(), at TimerThread.cpp:663 https://bugzilla.mozilla.org/show_bug.cgi?id=1402717

    Assertion failure: IsIdle(oldState) || IsRead(oldState), at PLDHashTable.h:118 https://bugzilla.mozilla.org/show_bug.cgi?id=1337253

    Assertion failure: js::CheckGrayMarkingState(mJSRuntime), at CycleCollectedJSRuntime.cpp:1207 https://bugzilla.mozilla.org/show_bug.cgi?id=1399866

    Assertion failure: [GFX1]: Failed to create DrawTarget: https://bugzilla.mozilla.org/show_bug.cgi?id=1401228

    Assertion failure: [GFX1]: Texture deallocated too late during shutdown, at gfx/2d/Logging.h:518 https://bugzilla.mozilla.org/show_bug.cgi?id=1414825

    Assertion failure: [GFX1]: Unexpected BufferProvider over-production., at gfx/2d/Logging.h:521 https://bugzilla.mozilla.org/show_bug.cgi?id=1418120

    Assertion failure: mLength > 0 (|First()| called on an empty string), at nsTSubstring.cpp:755 https://bugzilla.mozilla.org/show_bug.cgi?id=1383150

    Assertion failure: !mThreadIsShuttingDown (Huh?!), at LazyIdleThread.cpp:278 https://bugzilla.mozilla.org/show_bug.cgi?id=1392551

    Assertion failure: ok (Inconsistent time data), at js/src/gc/Statistics.cpp:830 https://bugzilla.mozilla.org/show_bug.cgi?id=1400153

    Assertion failure: OnGraphThreadOrNotRunning(), at MediaStreamGraph.h:1372 https://bugzilla.mozilla.org/show_bug.cgi?id=1409683

    Assertion failure: PermissionAvaliable(prin, aType), at extensions/cookie/nsPermissionManager.cpp:2273 https://bugzilla.mozilla.org/show_bug.cgi?id=1376106

    Assertion failure (fixed in Gecko 56): removedCount <= gUnusedAtomCount, at nsAtomTable.cpp:432: https://bugzilla.mozilla.org/show_bug.cgi?id=1367374

    Assertion failure: removedCount == gUnusedAtomCount: https://bugzilla.mozilla.org/show_bug.cgi?id=1389826

    Assertion failure: work_queue_.empty(), in MessageLoop::DeletePendingTasks() in message_loop.cc: https://bugzilla.mozilla.org/show_bug.cgi?id=1355196

    Assertion failure: zone->isGCScheduled(), atjsgc.cpp:7046 https://bugzilla.mozilla.org/show_bug.cgi?id=1405980

Comment 3

[Ed Morley [:emorley]]

Ahhh I understand now.

The request here is to add custom bug suggestion support for Assertions (like we have for say crash stack traces or leakcheck failures), that uses a custom regex and search logic, here:
https://github.com/mozilla/treeherder/blob/8b8394134e9407eff596866f9750cd0f6c22396e/treeherder/model/error_summary.py#L99-L118

I agree this would be useful. Would you be up for a PR? -:-)