1632170 - Crash in [@ mozilla::PendingAnimationTracker::TriggerPendingAnimationsOnNextTick]

Status: RESOLVED WORKSFORME

(Core :: DOM: Animation, defect, P3)

Description

[Simon Giesecke [:sg] [he/him]]

This bug is for crash report bp-5bcc7466-17fa-4d20-9b06-4958f0200417.

Top 10 frames of crashing thread:

0 xul.dll mozilla::PendingAnimationTracker::TriggerPendingAnimationsOnNextTick dom/animation/PendingAnimationTracker.cpp:78
1 xul.dll TriggerPendingAnimations layout/painting/nsDisplayList.cpp:2767
2 xul.dll nsDisplayList::PaintRoot layout/painting/nsDisplayList.cpp:3114
3 xul.dll static nsLayoutUtils::PaintFrame layout/base/nsLayoutUtils.cpp:4114
4 xul.dll mozilla::PresShell::Paint layout/base/PresShell.cpp:6159
5 xul.dll nsViewManager::ProcessPendingUpdatesPaint view/nsViewManager.cpp:460
6 xul.dll nsViewManager::ProcessPendingUpdatesForView view/nsViewManager.cpp:395
7 xul.dll nsViewManager::ProcessPendingUpdates view/nsViewManager.cpp:1018
8 xul.dll nsRefreshDriver::Tick layout/base/nsRefreshDriver.cpp:2204
9 xul.dll mozilla::RefreshDriverTimer::TickRefreshDrivers layout/base/nsRefreshDriver.cpp:351

There have been a few Nightly crashes starting with build id 20200417100143. Multiple reports from soundcloud.com URLs.

Comment 1

[Brian Birtles (:birtles, Feb 26~ parental leave)]

I wonder what changes recently to trigger these. PendingAnimationTracker itself doesn't appear to have changed. Did something change with the way we handle hashtables or timestamps or animations or painting?

Boris, I'm trying to work out what priority to give this. Are you aware of any changes from just under 1 week ago that might affect this?

Comment 2

[Boris Chiou [:boris]]

At least there is no update for animation, based on the build id 20200417100143 and the pushlog. Hiro, are there any suspect patches for printing that might be related to this call stack?

Comment 3

[Hiroyuki Ikezoe (:hiro)]

I did give up checking the commits around the push log since the crash also happened on Firefox 72. e.g. bp-7fd9f544-036e-4239-ace0-051fa0200130

Though it's not related to this crash, we call TriggerPendingAnimationsOnNextTick when the pres shell hasn't yet generated OR the pres shell is being suppressed. But to me, we should skip calling the function when the pres shell is being destroyed. I will file a bug for this case.

Hey Simon, do you have the permission to read the URL field in crash reports? Can you see some suspicious URLs we can reproduce the crash? I am seeing some of crashes have the same install time, so I am assuming they visited the same URL. Thanks!

Comment 4

[Simon Giesecke [:sg] [he/him]]

As I wrote in the description, most crash reports that have a URL are from soundcloud.com, e.g. https://soundcloud.com/search with different search terms. There's only a single URL that is from a different domain. Maybe soundcloud.com changed something recently, which causes these crashes?

Comment 5

[Simon Giesecke [:sg] [he/him]]

Interestingly, Bug 1632161 seems also be spiking (at a low volume) on soundcloud.com very recently.

Comment 6

[Hiroyuki Ikezoe (:hiro)]

Thank you Simon. I had missed the URL in the comment. I am sorry.

(In reply to Simon Giesecke [:sg] [he/him] from comment #4)

As I wrote in the description, most crash reports that have a URL are from soundcloud.com, e.g. https://soundcloud.com/search with different search terms. There's only a single URL that is from a different domain. Maybe soundcloud.com changed something recently, which causes these crashes?

Yeah, maybe.

From https://crash-stats.mozilla.org/report/index/2793e74d-0404-4db5-b86f-9fc400200424#tab-correlation

(100.0% in signature vs 00.06% overall) address = 0x90
(100.0% in signature vs 14.77% overall) reason = EXCEPTION_ACCESS_VIOLATION_READ
(100.0% in signature vs 17.53% overall) Module "MP3DMOD.DLL" = true
(100.0% in signature vs 18.11% overall) Module "msdmo.dll" = true
(50.00% in signature vs 03.86% overall) platform_pretty_version = Windows 8.1
(42.86% in signature vs 03.75% overall) Module "netapi32.dll" = true
(42.86% in signature vs 03.98% overall) Module "wkscli.dll" = true

MP3DMOD.DLL affected this crash somehow?

(In reply to Simon Giesecke [:sg] [he/him] from comment #5)

Interestingly, Bug 1632161 seems also be spiking (at a low volume) on soundcloud.com very recently.

Looks like the same root cause..

Comment 7

[Brian Birtles (:birtles, Feb 26~ parental leave)]

I've got no idea what's going on here. As with bug 1632161 it's possible there is corruption coming from somewhere else. I'm not aware of anything particularly concerning about this or seeing anything immediately actionable, so setting to P3 for now.

Comment 8

[BugBot [:suhaib / :marco/ :calixte]]

The severity field is not set for this bug.
:birtles, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 9

[BugBot [:suhaib / :marco/ :calixte]]

Closing because no crashes reported for 12 weeks.