1450226 - Crash in js::gcstats::Statistics::beginSlice

Status: RESOLVED WORKSFORME

(Core :: JavaScript: GC, defect)

Description

[Calixte Denizet (:calixte)]

This bug was filed from the Socorro interface and is
report bp-22ef851b-48a8-4545-b79a-7bee60180329.
=============================================================

Top 10 frames of crashing thread:

0 libxul.so js::gcstats::Statistics::beginSlice mozglue/misc/TimeStamp.h:67
1 libxul.so Sprite_D16_S32::blitRect gfx/skia/skia/src/core/SkColorData.h:577
2 libxul.so js::gc::GCRuntime::gcCycle js/src/gc/Statistics.h:406
3 libxul.so Sprite_D16_S32::blitRect gfx/skia/skia/src/core/SkColorData.h:577
4 libxul.so js::gc::GCRuntime::collect js/src/gc/GC.cpp:7538
5 libxul.so Sprite_D16_S32::blitRect gfx/skia/skia/src/core/SkColorData.h:577
6 libxul.so js::gc::GCRuntime::startGC js/src/gc/GC.cpp
7 libxul.so Sprite_D16_S32::blitRect gfx/skia/skia/src/core/SkColorData.h:577
8 libxul.so wcsrtombs 
9 libxul.so nsJSContext::GarbageCollectNow dom/base/nsJSEnvironment.cpp:1242

=============================================================

There are 455 crashes (from 324 installations) in 60.0b7 with buildid 20180326164103.
:overholt, could you investigate please ?

Comment 1

[Calixte Denizet (:calixte)]

[Tracking Requested - why for this release]:
The volume is pretty high on FennecAndroid and it's a regression (no crash before 60b7 on android).

Comment 2

[Andrew Overholt [:overholt]]

Is it possible to get a regression window?

I see the spike started around March 29th ... Steve, any ideas? I see you landed bug 1449066 and Jan landed bug 1448589 (this is just me looking for "GC" at https://hg.mozilla.org/mozilla-central/shortlog).

Comment 3

[Ryan VanderMeulen [:RyanVM]]

Note that this was a spike on Beta, however.

Comment 4

[Ryan VanderMeulen [:RyanVM]]

Here's the changelog between b5 and b7:
https://hg.mozilla.org/releases/mozilla-beta/pushloghtml?fromchange=FENNEC_60_0b5_RELEASE&tochange=FENNEC_60_0b7_RELEASE

Comment 5

[Andrew Overholt [:overholt]]

Thanks, Ryan. Nothing stands out to me in that pushlog.

Comment 6

[Jan de Mooij [:jandem]]

These crashes are all SIGILL (illegal instruction). We should look at crash dumps to see what instruction this is exactly.

Comment 7

[Julien Cristau [:jcristau]]

No crashes in 60.0b9 or 60.0b11, this was all 60.0b7.

Comment 8

[Julien Cristau [:jcristau]]

I'm gonna clear the blocking flag and hope this doesn't come back :/