348729 - [FIX]Crash [@ nsRuleNode::GetParentData] with :first-letter anonymous content and removing styles

Status: VERIFIED FIXED

(Core :: Layout, defect, P1)

Description

[Martijn Wargers (dead)]

See upcoming testcase, which crashes for me on load. Usually, it crashes for me the first time. If it doesn't, try reloading a few times.

Talkback ID: TB22114083K
nsRuleNode::GetParentData   nsStyleContext::GetStyleData   nsRuleNode::WalkRuleTree   nsRuleNode::GetVisibilityData  

This is a regression.
It doesn't crash in a 2004-10-29 build, it crashes in a 2005-05-06 build.

Comment 1

[Martijn Wargers (dead)]

Attachment: testcase

Comment 2

[Martijn Wargers (dead)]

Attachment: original file

Comment 3

[Martijn Wargers (dead)]

So I get a regression window of 2004-10-30 and 2004-10-31:
http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=all&branch=HEAD&branchtype=match&dir=&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2004-10-30+08&maxdate=2004-10-31+08&cvsroot=%2Fcvsroot
A regression from bug 264914, somehow?

Comment 4

[Boris Zbarsky [:bzbarsky]]

Attachment: Fix

Yeah, this is a regression from bug 264914.  The problem is that when we remove the <tfoot>'s abs pos frame we also remove the placeholder.  So we hit that first hunk in this patch, which clobbers parentFrame to be the parent of the placeholder.  Then we use parentFrame to construct the frame constructor state for recovering the letter frames.  And the patch for bug 264914 made us use the frame constructor state to determine the float parent, so we put the float on totally the wrong float list, and things break.

The first hunk of the patch makes us not clobber parentFrame and is enough to fix this bug; the second hunk just avoids an extraneous call to GetFloatContainingBlock.

Comment 5

[Boris Zbarsky [:bzbarsky]]

We should get this in on branches too.

Comment 6

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Comment on attachment 233798 [details] [diff] [review]
Fix

looks branch-good to me

Comment 7

[Boris Zbarsky [:bzbarsky]]

Fixed on trunk.

Comment 8

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 233798 [details] [diff] [review]
Fix

a=dbaron on behalf of drivers.  Please land on MOZILLA_1_8_BRANCH and add the fixed1.8.1 keyword once you have done so.

Comment 9

[Daniel Veditz [:dveditz]]

Comment on attachment 233798 [details] [diff] [review]
Fix

approved for 1.8.0 branch, a=dveditz for drivers

Comment 10

[Boris Zbarsky [:bzbarsky]]

Attachment: Branch version of patch

Comment 11

[Boris Zbarsky [:bzbarsky]]

Fixed for 1.8.1 and 1.8.0.7.

Comment 12

[Bob Clary [:bc] (inactive)]

https://bugzilla.mozilla.org/attachment.cgi?id=233782
ff2b2 debug/nightly windows/linux no crash

###!!! ASSERTION: out of bounds: 'PRInt32(aIndex) >= 0 && aIndex <= length', file /work/mozilla/builds/ff/2.0/mozilla/layout/base/nsChildIterator.h, line 133
Break: at file /work/mozilla/builds/ff/2.0/mozilla/layout/base/nsChildIterator.h, line 133
###!!! ASSERTION: Float frame has wrong parent: 'floatFrame->GetParent() == mBlock', file /work/mozilla/builds/ff/2.0/mozilla/layout/generic/nsBlockReflowState.cpp, line 835
Break: at file /work/mozilla/builds/ff/2.0/mozilla/layout/generic/nsBlockReflowState.cpp, line 835

verified fixed 1.8

Comment 13

[alice nodelman [:alice] [:anode]]

https://bugzilla.mozilla.org/attachment.cgi?id=233782&action=view should not crash browser.

Mozilla/5.0 (Macintosh; U; Intel Mac OS X; en-US; rv:1.8.0.7pre) Gecko/20060821 Firefox/1.5.0.7pre

verified 1.8.0.7

Comment 14

[Daniel Veditz [:dveditz]]

Given the regression window this is not a problem on the 1.7/aviary branches, right?

Comment 15

[Bob Clary [:bc] (inactive)]

crash test landed
http://hg.mozilla.org/mozilla-central/rev/f35038f6935a