382262 - "ASSERTION: Event listener manager hash not empty at shutdown!" with -moz-column-count and float

Status: RESOLVED WORKSFORME

(Core :: Layout, defect)

Description

[Jesse Ruderman]

Loading the testcase (in Mac trunk debug) and then quitting triggers:

###!!! ASSERTION: Event listener manager hash not empty at shutdown!: 'sEventListenerManagersHash.entryCount == 0', file mozilla/content/base/src/nsContentUtils.cpp, line 694

Comment 1

[Jesse Ruderman]

Attachment: testcase

Comment 2

[Boris Zbarsky [:bzbarsky]]

Chances are we're leaking (content nodes via frames?)

Comment 3

[Boris Zbarsky [:bzbarsky]]

Yeah, we seem to be leaking via the frames created by nsBlockFrame::SplitPlaceholder.

Perhaps we should be asserting somewhere that mOverflowPlaceholders has been emptied?  And see when that assert fails?

Comment 4

[Jesse Ruderman]

CCing Eli, who loves floats and -moz-column.

Comment 5

[Eli Friedman]

I'd love to fix this, but I can't reproduce :)

Comment 6

[Jesse Ruderman]

Can you reproduce if you add the assertion bz suggested in comment 3?

Comment 7

[Eli Friedman]

I have no idea what assertion to add; we already have the following:

nsBlockReflowState::~nsBlockReflowState()
{
  NS_ASSERTION(mOverflowPlaceholders.IsEmpty(),
               "Leaking overflow placeholder frames");

Comment 8

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

I can't reproduce this either.

If we were leaking frames we'd get the "Some frame destructors not called" assertion, wouldn't we?

Comment 9

[Jesse Ruderman]

I can't reproduce this any more.

Comment 10

[Jesse Ruderman]

Crashtest checked in.