713610 - "ASSERTION: 2nd pass should never guess wrong" in TextOverflow::ExamineLineFrames

Status: VERIFIED FIXED

Description

[Jesse Ruderman]

Created attachment 584389 [details]
testcase

###!!! ASSERTION: 2nd pass should never guess wrong: 'pass == 0', file /builds/slave/m-cen-osx64-dbg/build/layout/generic/TextOverflow.cpp, line 540

Might be a regression from bug 690187.

Comment 1

[Jesse Ruderman]

Created attachment 584390 [details]
stack trace

Comment 2

[Bernd]

the try server run https://tbpl.mozilla.org/?tree=Try&rev=cab7d4e1219f
 will not execute the test 
"try: -b do -p all -u reftest -t none"

-u crashtest would do

Comment 3

[Mats Palmgren (:mats)]

Thanks Bernd.

Comment 4

[Mats Palmgren (:mats)]

Created attachment 584552 [details] [diff] [review]
fix

We should guess 'false' for a side that is inactive.
(the rendering should be correct so there's no need to fix this on Aurora)

Comment 5

[Mats Palmgren (:mats)]

Created attachment 584553 [details] [diff] [review]
regression test

Comment 6

[Mats Palmgren (:mats)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/18218243232a
https://hg.mozilla.org/integration/mozilla-inbound/rev/ec08627769aa

Comment 7

[Kyle Huey [:khuey] (Exited; not receiving bugmail, email if necessary)]

https://hg.mozilla.org/mozilla-central/rev/18218243232a
https://hg.mozilla.org/mozilla-central/rev/ec08627769aa

Comment 8

[Paul Silaghi, QA [:pauly]]

I'm not able to reproduce the problem. Jesse, is there any chance to remember on what FF did you catch this?

Comment 9

[Jesse Ruderman]

I think it was just a matter of loading the testcase in a debug build of Firefox.

Comment 10

[Paul Silaghi, QA [:pauly]]

I was able to see the crash on 2011-12-26-mozilla-central-debug build.
Verified fixed on FF 12 2012-04-04-mozilla-beta-debug on Win 7/64, Ubuntu 11.10 and Mac OS X 10.6.