682460 - Leak when forcing layout during mutation event for table attribute

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Jesse Ruderman]

Attachment: testcase

###!!! ASSERTION: potential leak, plus waste of work: 'mTableInheritedAttributes == TABLE_ATTRS_DIRTY', file content/html/content/src/nsHTMLTableElement.cpp, line 1228

+ trace-refcnt reports a leaked "nsMappedAttributes" object

Comment 1

[Jesse Ruderman]

Attachment: stack trace

Comment 2

[Olli Pettay [:smaug][[e-mail]]]

Using nsHTMLTableElement::SetAttr is just buggy

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

I think this is a regression from bug 211636

Comment 4

[Olli Pettay [:smaug][[e-mail]]]

Yes, bug 211636 should have used Before/AfterSetAttr, not override SetAttr.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Olli, are you fixing this?  Or is Bernd?

Requesting tracking for the leak regression.

Comment 6

[Bernd]

I don't feel comfortable with this, see https://bugzilla.mozilla.org/show_bug.cgi?id=211636#c9 
>I am insecure about the changes in SetAttr and UnsetAttr

Comment 7

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Attachment: Patch v1

Something like this? (Untested)

Comment 8

[Mozilla RelEng Bot]

Try run for edb5c7975703 is complete.
Detailed breakdown of the results available here:
    http://tbpl.allizom.org/?tree=Try&usebuildbot=1&rev=edb5c7975703
Results (out of 229 total builds):
    success: 212
    warnings: 15
    failure: 2
Builds available at http://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/[e-mail]-edb5c7975703

Comment 9

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Attachment: Patch for checkin

As suggested by Bernd, fixed bug 211636 comment 16 as well. Ed, if you don't mind? (Passed try: <http://tbpl.allizom.org/?tree=Try&usebuildbot=1&rev=edb5c7975703>)

Comment 10

[Ed Morley [:emorley]]

http://hg.mozilla.org/integration/mozilla-inbound/rev/f1cf848b1ff2

Comment 11

[Ed Morley [:emorley]]

http://hg.mozilla.org/mozilla-central/rev/f1cf848b1ff2