(inactive) Jim Chen [:jchen] [:darchons] Reporter
	Description • 7 years ago

This bug was filed from the Socorro interface and is 
report bp-ef120168-9a59-49e7-a243-42a672170323.
=============================================================

Windows crash that seemed to have started suddenly with the 2017-03-21 Aurora build, but doesn't appear in Nightly builds. I didn't see any strong suspects in the regression range [1]. Any ideas?

[1] https://hg.mozilla.org/releases/mozilla-aurora/pushloghtml?fromchange=6a72a11138b9b41a02455a3cb32a8994a7f41934&tochange=0496e366a8f4d1393a8caba1db5f6d1f49da468d

	Marcia Knous [:marcia]
	Comment 1 • 7 years ago

This is the top crash for Aurora in the last 7 days: https://crash-stats.mozilla.com/topcrashers/?product=Firefox&version=54.0a2&days=7

	Nicholas Nethercote [inactive]
	Comment 2 • 7 years ago

This is the #6 Windows topcrash in Aurora 20170331004006.

	Gerry Chang [:gchang]
	Comment 3 • 7 years ago

Hi :kanru,
Can you help find someone to look at this? This is the topcrash in Aurora.

	[:philipp]
	Comment 4 • 7 years ago

[Tracking Requested - why for this release]: this is now also the topcrash in early data for 53.0b10 where the signatures are accounting for 13% of all browser crashes.
timing-wise, this would correlate with the uplift of bug 1342433.

	Julien Cristau [:jcristau]
	Comment 5 • 7 years ago

[Tracking Requested - why for this release]: bug 1342433 was also uplifted to esr52

	alexander :surkov (:asurkov) Assignee
	Comment 6 • 7 years ago

The bug is on my list, and actually a part of 'broken accessible tree' problem we deal with currently. I try to hunt it down with varied success, but not yet caught all issues in the code.

	Liz Henry (:lizzard) (relman/hg->git project)
	Comment 7 • 7 years ago

1000 crashes in one day of beta is probably release-blocking. It's an order of magnitude higher than other top crashes on beta. 

How about backing out the work from bug 1342433? What do you think? 
n-i to jesup and davidb for advice.

	alexander :surkov (:asurkov) Assignee
	Comment 8 • 7 years ago

(In reply to Liz Henry (:lizzard) (needinfo? me) from comment #7)
> 1000 crashes in one day of beta is probably release-blocking. It's an order
> of magnitude higher than other top crashes on beta. 
> 
> How about backing out the work from bug 1342433? What do you think? 

maybe it's worth to try, however it should be a good idea to keep tracking of the related signatures bug 1321384, bug 1309686 and bug 1318645 for possible correlation.

	Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)
	Comment 9 • 7 years ago

IF we decide to back it out from release53, we will also need to back it out from esr52. Julien FYI.

	[:philipp]
	Comment 10 • 7 years ago

(In reply to alexander :surkov from comment #8)
> maybe it's worth to try, however it should be a good idea to keep tracking
> of the related signatures bug 1321384, bug 1309686 and bug 1318645 for
> possible correlation.

in beta 9 all those signatures (including the one in this bug) accounted for ~0.4% of browser crashes.

while data from beta 10 is still quite early and not totally conclusive, it looks like signatures from bug 1321384 & bug 1309686 are no longer present - but the crashing volume here is obviously much worse than before.

	David Bolter [:davidb] (NeedInfo me for attention)
	Comment 11 • 7 years ago

(In reply to alexander :surkov from comment #8)
> (In reply to Liz Henry (:lizzard) (needinfo? me) from comment #7)
> > 1000 crashes in one day of beta is probably release-blocking. It's an order
> > of magnitude higher than other top crashes on beta. 
> > 
> > How about backing out the work from bug 1342433? What do you think? 
> 
> maybe it's worth to try, however it should be a good idea to keep tracking
> of the related signatures bug 1321384, bug 1309686 and bug 1318645 for
> possible correlation.

Based on volume analysis from philipp in comment 10, let's backout 1342433. Who generally does that for beta these days?

	Liz Henry (:lizzard) (relman/hg->git project)
	Comment 12 • 7 years ago

Ryan can do the backout for bug 1342433. I don't mind if we leave it on trunk (since 55 is going to remain in nightly for another 6 weeks), up to Alexander for that one

	alexander :surkov (:asurkov) Assignee
	Comment 13 • 7 years ago

(In reply to Liz Henry (:lizzard) (needinfo? me) from comment #12)
> Ryan can do the backout for bug 1342433. I don't mind if we leave it on
> trunk (since 55 is going to remain in nightly for another 6 weeks), up to
> Alexander for that one

it is hard to argue against data, but this is surprising with me the bug 1342433 spikes the crashes, since it's supposed to be a good thing for the accessible tree correctness overall, so it's definitely worth to keep it on nightly.

	Ryan VanderMeulen [:RyanVM]
	Comment 14 • 7 years ago

Backed out bug 1342433 from everywhere but mozilla-central.

	alexander :surkov (:asurkov) Assignee
	Comment 15 • 7 years ago

fixed by bug 1363027 (landed 25 may in Firefox 54), the last crash is dated by 25 May (https://crash-stats.mozilla.com/signature/?product=Firefox&signature=mozilla%3A%3Aa11y%3A%3AAccessible%3A%3AHasGenericType&date=%3E%3D2017-05-23T16%3A39%3A00.000Z&date=%3C2017-05-30T16%3A39%3A00.000Z&_columns=date&_columns=product&_columns=version&_columns=build_id&_columns=platform&_columns=reason&_columns=address&_columns=install_time&_sort=-build_id&_sort=version&_sort=-date&page=1).