Bug 73227

Summary: REGRESSION(r101172): It made fast/dom/clone-node-style.html assert
Product: WebKit Reporter: Zhenyao Mo <zmo>
Component: DOMAssignee: Andreas Kling <kling>
Status: RESOLVED FIXED    
Severity: Critical CC: ap, dbeam, dglazkov, kling, koivisto, macpherson, ossy, vsevik, webkit.review.bot
Priority: P1    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Bug Depends on:    
Bug Blocks: 73392, 73121    
Attachments:
Description Flags
QtWebKit gdb backtrack
none
Proposed patch none

Zhenyao Mo
Reported 2011-11-28 10:25:28 PST
It's unclear when this started to happen. Dimitri, can you help triage?
Attachments
QtWebKit gdb backtrack (12.80 KB, text/plain)
2011-11-29 08:29 PST, Csaba Osztrogonác 		no flags
Details
Proposed patch (5.06 KB, patch)
2011-11-29 10:11 PST, Andreas Kling 		no flags
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
Zhenyao Mo
Comment 1 2011-11-28 10:29:33 PST
Added the test to test_expectations.txt in r101264
Csaba Osztrogonác
Comment 2 2011-11-29 07:22:15 PST
It is a regression caused by r101172. I got it with manual bisecting. Andreas, could you check it, please?
Andreas Kling
Comment 3 2011-11-29 07:25:17 PST
(In reply to comment #2) > It is a regression caused by r101172. I got it with manual bisecting. > Andreas, could you check it, please? On it! Suspect it's the same issue as bug 73213, great that it broke a test :)
Csaba Osztrogonác
Comment 4 2011-11-29 07:27:17 PST
Unfortunately it is a flakey crash. But you can easily reprodeuce with old-run-webkit-tests --debug fast/dom/clone-node-style.html --iterations 100
Csaba Osztrogonác
Comment 5 2011-11-29 08:29:08 PST
Created attachment 116975 [details] QtWebKit gdb backtrack I reproduced this bug inside GBD with the following method: $ gdb WebKitBuild/Debug/bin/DumpRenderTree (gdb) run - <ctrl+v> <ctrl+v> <ctrl+v> <ctrl+v> <ctrl+v> ... (fast/dom/clone-node-style.html was on my clipboard) The crash backtrace is attached.
Csaba Osztrogonác
Comment 6 2011-11-29 09:11:50 PST
I added this test to the Qt Skipped list: http://trac.webkit.org/changeset/101389 Please unskip it with the proper fix. Thanks in advance.
Andreas Kling
Comment 7 2011-11-29 10:09:08 PST
*** Bug 73213 has been marked as a duplicate of this bug. ***
Andreas Kling
Comment 8 2011-11-29 10:11:00 PST
Created attachment 116992 [details] Proposed patch
Dan Beam
Comment 9 2011-11-29 21:24:18 PST
bug 73392 may just be a straight dupe of this, but let's just make it block it for now (the proposed patch fixed the issue we're having in Chromium, btw).
Antti Koivisto
Comment 10 2011-11-30 04:56:55 PST
Comment on attachment 116992 [details] Proposed patch View in context: https://bugs.webkit.org/attachment.cgi?id=116992&action=review r=me, with comments > Source/WebCore/css/CSSMutableStyleDeclaration.h:-79 > - CSSMutableStyleDeclaration& operator=(const CSSMutableStyleDeclaration&); You probably want to hide this. > Source/WebCore/css/CSSMutableStyleDeclaration.h:80 > + void copyPropertiesAndStrictnessFrom(const CSSMutableStyleDeclaration&); As discussed, it might be clearer to just have copyProperties() and use the existing accessors for transfering the strictness.
Andreas Kling
Comment 11 2011-11-30 05:14:04 PST
Comment on attachment 116992 [details] Proposed patch As discussed with Antti, I'm cq+ing this as is so we get rid of the regression. I'm currently in transit and will make a follow-up patch with the requested adjustments later today.
WebKit Review Bot
Comment 12 2011-11-30 06:07:45 PST
Comment on attachment 116992 [details] Proposed patch Clearing flags on attachment: 116992 Committed r101502: <http://trac.webkit.org/changeset/101502>
WebKit Review Bot
Comment 13 2011-11-30 06:07:52 PST
All reviewed patches have been landed. Closing bug.
Note You need to log in before you can comment on or make changes to this bug.