CrashKill/2013-04-22
From MozillaWiki
Contents
General
Socorro
Flash
- It looks like Flash 11.7 is slightly better in terms of hangs as 11.6 (and 11.5), possibly back to 11.4 levels at least.
- Lower crash and hang level on 21 Beta are holding up, no regressions of our patch known so far.
Desktop
Release
- bug 812683 and bug 819028 (D3D11) are down by significant degrees, the cause is unclear.
- bug 851934 (Outlook.com) is down somewhat as well over the weekend, missing activity in reaction to a user helping us! Action needed!
- bug 836263 (Crossrider) also decreased in volume somewhat early last week, no activity on implementing a fix on our side.
Beta
- bug 863714 (nsCSSScanner::Next) - new in 21.0b3, now #14, seems to be another Radeon spike
- bug 829954 (mozilla::gfx::AlphaBoxBlur::Blur) has been rising in b3 as well, just #30 though.
- bug 837370 (js::types::TypeCompartment::addPendingRecompile) definitely looks fixed in 21.0b3
- bug 852915 (mozilla::MediaDecoderReader::DecodeToFirstAudioData) - patch to disable on Win7 without SP1 landed on trunk, uplifts needed.
Aurora
- bug 856670 (js::RemapWrapper) is still #2 though fixed on trunk and we still have no candidate for uplift, and no investigation happening.
Trunk
- bug 863646 (nsContentUtils::GetCurrentJSContext) was large but is fixed now, ultimately by backing out the offending part of bug 860438 as the other fix caused orange.
- Multiple IonMonkey crashes started appearing/spiking with the 4/20 build:
- bug 864033 (MarkInternal) is now the #1 by far, now assigned, investigation going on.
- bug 864037 (js::InvokeKernel, multiple signatures) is also high, not assigned yet, but investigation starting.
- bug 864125 (js::ion::DoTypeMonitorFallback, multiple signatures) is high as well, could be a dupe of one of the others
- bug 821733 (meta - js::CompartmentChecker::fail) rising might also be the same cause.
- A number of other signatures that flared up don't have bugs but might also be the same thing(s), e.g. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] - we might need to see what remains when the others are fixed.
Mobile
- Tegra crashes bug 802827 and bug 782223 are rising over the weekend (the latter might just be the signature switching due to some kind of symbols we got)
- bug 847834 (org.mozilla.gecko.BrowserApp.onActivityResult) has been rising sharply on Aurora and Nightly since 4/18 and is #1 and #4 on the respective channels now.
- bug 861796 (LibHandle::~LibHandle) and bug 860466 (mozilla::gl::GLContextEGL::GetSharedHandleDetails) are both fixed after spiking sharply on Nightly, but need some time to get off by-crash-date stats.
- Some of the IonMonkey trunk signatures also show up on Android, as one would expect.
- Some Baseline Compiler issues on Armv6 are still unsolved, see bug 863685.
- STRs are listed for bug 760394
- bug 852467 is blocking bug 834243 - crash in mozilla::layers::LayerManagerOGL::Initialize with abort message: "We need a context on Android"
B2G
- Work week last week in Spain
- lots of fixs, including bluetooth and other things
- MMS stuff landed last week, some ril incompat/crashes expected until we get RIL/Gecko to be compatible again.
- ckoehler + ted working w/ Geeksphone to get symbols and have them uploadead: