CrashKill/2014-04-14
From MozillaWiki
Contents
General
Previous Action Items
- KaiRo/liz to evaluate options for public stability list [KaiRo: Liz was off last week, will connect with her this week]
- [DONE] juanb to file remaining Seer crash from comment #10 [ this is bug 994344 ]
- [DONE] Anthony to dig into Aurora regression range for bug 740325
- Naoki still trying to get a reply from GP, KaiRo possibly trying to contact them as well. [KaiRo: held out due to new findings that this seems to be a seccomp violation.]
- bsmedberg to follow up with Jason on JS crashes without action in 29 (Beta)
- Anthony to follow up with bsmedberg on impact of GGC crashes
- [DONE] KaiRo to bring the __fnNCDESTROY crash up in the channel meeting for a decision by relman. [relman said "yes", David has questions in the bug, relman to follow up]
Socorro
Reported bug 995991 Sort is broken for columns in the Reports tab. (lizzard)
- Landed middleware changes for Bug 915246 - Graphs of crashes/ADU for individual signatures. Waiting on UI
- some UI fixes, skiplist changes, and supersearch is now exposed in the public API
- API rate limiting bug fixed that was causing rate limits to be lower than they should be for everyone
- Details: https://bugzilla.mozilla.org/buglist.cgi?bug_id=890946%2C945894%2C988604%2C984562%2C977406%2C915246%2C971706%2C984940%2C989055%2C991730%2C992347%2C992410%2C915246%2C981854%2C993786%2C994221%2C994913&list_id=9895683
Flash
It looks like F284861780, F_1466881225, and F344568511 (there may be more at lower volume) are OSX-specific Flash signatures that went from 0 to significant on 2014-04-07/08 (last Mon/Tue). Did Adobe ship something bad for Mac?
Desktop
- Looks like we're still missing some MS symbols (kernel32.dll, kernelbase.dll) from recent Patch Tuesday.
Beta
- Ongoing regression tracking:
- bug 987248 (main part of Seer crashes) has landed for 29.0b7 and looks fixed!
- bug 976181 (nsContentUtils::IsCallerChrome, dupe of bug 913138) landed on 29.0b7 and the signature is gone from the top 300 as well.
- bug 976536 (JIT, also bug 988961) has landed for 29.0b7 but apparently didn't help, Till and Jan are discussing that and following up.
- bug 970362 F-secure is (hopefully) looking at it, meanwhile David created a blocklisting patch for those old versions on XP, which landed on trunk and has approvals, waiting for uplift.
- bug 856796 (Yarr, correlated to Avast) - looks like new Avast versions are not triggering it any more, but we also have a workaround patch that just landed on trunk, uplifts hopefully coming soon.
- bug 994344 (remaining Seer issue) has landed on trunk, uplift hopefully coming soon.
- bug 974656 - top #2 on Fx29, linux, needs investigation
- bug 988961 - top #5 on Fx29 this week, also needs traction
- juanb filed bug 994335 (nsXULPrototypeDocument::GetCompilationGlobal) which is #6 and a startup crash, but not exactly new. We probably should have it investigated, though.
Aurora
- bug 740325 (_cairo_surface_snapshot_copy_on_write) is still the major issue, bug 991767 is supposed to fix it but was backed out on Aurora. Android might have the same crash as well.
- bug 994335 - startup crash in nsXULPrototypeDocument::GetCompilationGlobal() also mentioned above (beta). this startup crash has risen to #12 on Fx30 topcrash list.
- bug 990082 - crash in nsWindowGfx::CreateIcon(imgIContainer*, bool, unsigned int, unsigned int, nsIntSize, HICON__**) - asked dev to request uplift to Aurora for the fix that has eliminated this on Nightly.
Trunk
- tracking 83 crash bugs for Firefox 31. 16 are topcrashers. https://wiki.mozilla.org/Releases/Firefox_31/Test_Plan#Stability
- bugs marked GGC: https://bugzilla.mozilla.org/buglist.cgi?list_id=9894954&resolution=---&resolution=DUPLICATE&status_whiteboard_type=allwordssubstr&query_format=advanced&status_whiteboard=%5BGGC%5D
- bug 995982 2 crash signatures, looks like a pdf.js bug, still needs regression window. A couple of outlying crashes on Firefox 28 but a big spike in early april for 31. (lizzard)
Android
- bug 812881 is in for b8 and should fix the largest offenders on 29 atm
- Fixes the # 1,3 and 5 crashes on release
- Might fix bug 983211 #6
- Looked for the #1 signature with a build date > 2014-04-11 and no crashes were found will keep an eye on it
B2G
- TCL says they have uploaded new symbols but they might not be used correctly, do we need to unpack them or something?
- lots of missing symbols for various devices
1.5
- bug 983976 - crash [@ libq3dtools_adreno200.so@0x335fa ] when launching Usage app after app's initial setup
- what Kairo stated earlier
- bug 990908 - [RTSP] Video app crash at android::OMXCodec::read when opening RTSP streaming
- still needs work
1.4
- bug 989408 - crash in mozilla::layers::PLayerTransactionChild::SendPLayerConstructor(mozilla::layers::PLayerChild*)
- happens on 1.3; needs further QA investigation; bug 993933 is a possible dup
- bug 992313 - Crash in ContentChild::RecvNotifyIdleObserver while running stability scripts
- looking for dev.
- bug 994903 - Camera app crashes during stability test in mozilla::layers::GrallocImage::~GrallocImage (
- being patched
- bug 993736 - [crash] AtomicBaseIncDec<unsigned int> | RefCounted<mozilla::layers::TextureChild> | TextureChild::ActorDestroy | PColorPickerChild::DestroySubtree
- bug 993947 may be a dup of this bug
1.3/1.3t
- bug 976656 - crash in mozilla::gfx::DrawTargetCairo::CreateSourceSurfaceFromNativeSurface(mozilla::gfx::NativeSurface const&) const
- bug 970007 - [tarako]monkey test crash at libxul.so!BufferUnrotate
- bug 991550 - [Tarako][Music]Switch between Music and Songs categories get crash
Notes
Upcoming PTO
- bsmedberg out next Monday for Easter
- ashughes out next week
- lonnen on the road for 3 weeks after next week, will find someone else to attend this meeting for Socorro but will be reachable during the week for questions.
Action Items
- [Carryover] KaiRo/liz to evaluate options for public stability list
- Naoki still trying to get a reply from GP, KaiRo possibly trying to contact them as well with new info on seccomp, etc.
- [DONE] KaiRo to check if Mac Flash crashes really increased [it did, filed bug 996637]
- bsmedberg to follow up on getting investigation for bug 994335