CrashKill/2014-02-10: Difference between revisions
Jump to navigation
Jump to search
m (→Trunk) |
|||
| Line 31: | Line 31: | ||
* Bas came up with a patch in {{bug|805406}} (PushClipsToDT) but unfortunately it doesn't look like it helped so far. | * Bas came up with a patch in {{bug|805406}} (PushClipsToDT) but unfortunately it doesn't look like it helped so far. | ||
* Confirmed (New) an old bug {{bug|677043}} as the signatures involved have spiked with recent builds. | * Confirmed (New) an old bug {{bug|677043}} as the signatures involved have spiked with recent builds. | ||
* Filed {{bug|970366}} have linked to a few other very similar crashes that also recently (starting 20140207) spiked in Nightly builds. We'll see if each signature should be broken out separately. At a high level glance, I suspect {{bug|960873} is responsible for all of the related crashes. | * Filed {{bug|970366}} have linked to a few other very similar crashes that also recently (starting 20140207) spiked in Nightly builds. We'll see if each signature should be broken out separately. At a high level glance, I suspect {{bug|960873}} is responsible for all of the related crashes. | ||
==Metro== | ==Metro== | ||
Revision as of 18:01, 10 February 2014
General
Socorro
- Released 73 last week
- Finally got ADI data from Metrics - a problem with syncing logs on the LB. They have now asked if we can pull data instead of having it pushed, which we have wanted for a long time. Plans in progress.
- Web Eng work week will be week of March 24 in Mountain View
Flash
- We didn't have symbols for Flash 12.0.0.44 (urgent security fix of last week), but after contacting Adobe and their Asian people coming back from holidays, they should be uploaded now.
- bug 970396 - It appeared like we would have large spikes of new signatures in Flash crashes last week, but this was only due to the missing symbols.
Desktop
- Discussion: bug 934509 - what process improvements should we make so that this doesn't get lost?
Release
- bug 934509 was the #1 topcrash early on and is staying at #3, patch is in -central and approved up to -beta, bug 969923 even has a testcase now. This was introduced in 27 by bug 918345 and apparently we missed catching it until release. :(
- bug 856670 and bug 969441 were topcrashes #1/#2 at some points last weeks but haven't happened in the last few days with any significant volume. They probably should be investigated and fixed still, they might flare up again at any point.
- bug 897870 (abort in Pickle::Pickle) is #8 and should also be investigated.
- None of the above really call for a point-release though, probably are good for 28, but riding along a point-release coming otherwise would be nice.
Beta
- Beta 28 looks even worse than 27 in overall rates in the first days, but that seems to be mostly one bugs for which we already have a patch (just need uplift, see below).
- Beta explosiveness report shows some regressions between 27 and 28 on Beta, main ones being bug 965914 (has patch, needs uplift), bug 952321, and bug 963316. There might be a few more we should take a look at.
Aurora
- bug 965914 - Top crash #1 in Aurora, and at the top of the explosive reports. Needs uplift to Aurora.
- bug 970362 - Showing up in the top 10, also in the explosive reports. Possibly F-secure related. Reached out to one user to find out more.
Trunk
- Bas came up with a patch in bug 805406 (PushClipsToDT) but unfortunately it doesn't look like it helped so far.
- Confirmed (New) an old bug bug 677043 as the signatures involved have spiked with recent builds.
- Filed bug 970366 have linked to a few other very similar crashes that also recently (starting 20140207) spiked in Nightly builds. We'll see if each signature should be broken out separately. At a high level glance, I suspect bug 960873 is responsible for all of the related crashes.
Metro
- Very low volume across branches. A little surprising. Metro mode could be more discoverable.
Android
- In overall crash rates, it looks like 28 beta is going to be better than 27 beta for Android.
B2G
General
- need symbols from ZTE for OpenC on 1.3, Alcatel for OT-995, flatfish
- Symbols missing on 1.3 : inari, buri, ZTE OPENC, geeksphone, Alcatel OT-995, flatfish, rockchip rk30sdk, , Sony R800i, Sony Xperia S, LGE AOSP on Mako, WonderMedia WM-8880-MID
- Symbol missing on 1.4 : dev builds that hasn't had symbols pushed. Chances are these are builds that dev made themselves and flashed with drivers that may be old. ie they haven't repulled the OEM blob out from the latest blob.
- looks like we might be missing symbols from releng builds for 1.4: https://crash-stats.mozilla.com/report/index/fad1690d-9f81-4f9d-b19a-212f22140210
- not enough data for 1.2 and below.
1.4
- bug 970403 - crash in android_atomic_add
- filed bug for nhirata to follow up and see if he can repro
- bug 945335 - crash in audioTrack_callBack_pullFromBuffQueue
- happens in 1.3
- bug 964328 - crash in mozilla::layers::DeprecatedContentClientRemoteBuffer::CreateAndAllocateDeprecatedTextureClient(mozilla::RefPtr<mozilla::layers::DeprecatedTextureClient>&)
- fall back is only in buri at this moment in time I think? Sotaro made a comment
1.3
- bug 871574 - crash in mozilla::dom::indexedDB::PIndexedDBRequestChild::OnMessageReceived
- happens on 1.4 as well
- bug 970418 - crash in __libc_android_abort | __android_log_assert | android::OMXCodec::signalBufferReturned
- inari only?