1461204 - Crash in mozilla::layers::DIGroup::ComputeGeometryChange (needs test)

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, defect, P1)

Description

[Darkspirit]

Seen on Socorro. This has been overlooked.

Apparently bug 1455488 didn't completely fix it.

https://crash-stats.mozilla.com/search/?signature=~mozilla%3A%3Alayers%3A%3ADIGroup%3A%3AComputeGeometryChange&app_notes=~WR%2B&product=Firefox&date=%3E%3D2018-04-26T19%3A46%3A00.000Z&date=%3C2018-05-13T19%3A46%3A10.000Z&_sort=-date&_facets=signature&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#crash-reports

bp-035d7c3b-a789-4d73-acdb-754890180513 build 2018-05-12_100055 Win10
> MOZ_RELEASE_ASSERT(rect.IsEqualEdges(aData->mRect))

Comment 1

[Darkspirit]

bp-d995bdac-4917-4301-8b16-09c3b0180511
> Just listening to music on Spotify when this page crashed.

Comment 2

[Kartikaya Gupta (email:[e-mail])]

This is the #3 WR-specific topcrash over the last 7 days.

Comment 3

[Jeff Muizelaar [:jrmuizel]]

I can reproduce this logging into mogo.ca

Comment 4

[Ryan VanderMeulen [:RyanVM]]

Insta-crash for me on creditkarma.com when I log into my account.

Comment 5

[Kartikaya Gupta (email:[e-mail])]

This is now the #1 WR-specific topcrash, since the previous topcrashes are now fixed.

Comment 6

[Thomas Daede [:TD-Linux]]

This is reproducible without a login just by visiting velosmedia.com and scrolling down.

Comment 7

[Aria Beingessner [:Gankra]]

ah thanks!

Comment 8

[Aria Beingessner [:Gankra]]

Attachment: Bug 1461204 - check the residual offset for following groups.

This was missed when the residual offset was added; it's correctly checked elsewhere

Review commit: https://reviewboard.mozilla.org/r/253580/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/253580/

Comment 9

[Aria Beingessner [:Gankra]]

hooray, easy fix once found. With some local testing this *appears* to fix:

https://bugzilla.mozilla.org/show_bug.cgi?id=1459760

and

https://bugzilla.mozilla.org/show_bug.cgi?id=1465948#c8

but not

https://bugzilla.mozilla.org/show_bug.cgi?id=1465948 itself

Comment 10

[Kartikaya Gupta (email:[e-mail])]

Comment on attachment 8988319 [details]
Bug 1461204 - check the residual offset for following groups.

https://reviewboard.mozilla.org/r/253580/#review260174

Can you add a test?

Comment 11

[Aria Beingessner [:Gankra]]

Comment on attachment 8988319 [details]
Bug 1461204 - check the residual offset for following groups.

https://reviewboard.mozilla.org/r/253580/#review260174

it's gonna take me a bit to reduce down the page into a crash test; ok with landing without it in the meantime to fix the crash for our users?

Comment 12

[Darkspirit]

(In reply to Alexis Beingessner [:Gankro] from comment #9)
> hooray

tick...tock... checkin-needed :D

Comment 13

[Aria Beingessner [:Gankra]]

waiting for kats' reply

Comment 14

[Kartikaya Gupta (email:[e-mail])]

Yeah that's fine. Let's keep this bug open until the test is in.

Comment 15

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/db455160668d
check the residual offset for following groups. r=kats

Comment 16

[Cosmin Sabou [:CosminS]]

https://hg.mozilla.org/mozilla-central/rev/db455160668d

Comment 17

[Kartikaya Gupta (email:[e-mail])]

(In reply to Alexis Beingessner [:Gankro] from comment #9)
> hooray, easy fix once found. With some local testing this *appears* to fix:
> 
> https://bugzilla.mozilla.org/show_bug.cgi?id=1459760
> 
> and
> 
> https://bugzilla.mozilla.org/show_bug.cgi?id=1465948#c8
> 

I'm still seeing this on the latest macOS nightly which has the fix.

Comment 18

[Kartikaya Gupta (email:[e-mail])]

The fix seems to have worked - I don't see any crash reports past the 20180627222831 buildid. Any ETA on the test? Let's get that in and close out this bug.

Comment 19

[Darkspirit]

Three new reports from two Win10 users:
bp-7a4b0d43-eeee-4605-8645-6df410180706
bp-e58570d4-1a8d-46a5-8cfb-7c7200180705
bp-1beaf54f-81fc-463d-a0a9-9053a0180705

Comment 20

[Jeff Muizelaar [:jrmuizel]]

I thought I could reproduce this, but can't seem to anymore.

Comment 21

[Kartikaya Gupta (email:[e-mail])]

Likely the instance you could reproduce was fixed by the patch in comment 16, then. There's probably another root cause lurking somewhere.

Comment 22

[Jeff Muizelaar [:jrmuizel]]

I was able to reproduce this on https://brightdb.com/elmeurope18/#frame3897

Comment 23

[Darkspirit]

(In reply to Jeff Muizelaar [:jrmuizel] from comment #22)
> I was able to reproduce this on https://brightdb.com/elmeurope18/#frame3897

It's "cold" here: bp-3a7ee1d0-01bb-4787-a198-55b870180711

Comment 24

[Aria Beingessner [:Gankra]]

i've wasted too much time trying to reduce this to something reasonable; gonna close without test.