292078 - 100% CPU utilization when opening page with TABLE and lots of DIVs that have style overflow:auto

Status: RESOLVED FIXED

(Core :: Layout: Tables, defect)

Description

[Adam Sawicki]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.7) Gecko/20050414 Firefox/1.0.3
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.7) Gecko/20050414 Firefox/1.0.3

I met this bug at my PhpBB2 forum at shoutbox. My friend entered there a BBCode
that generated several DIVs, one inside each other. That caused the web page
with the shoutbox inside a iframe hung up the browser.

I isolated the problem and prepared a short example that causes this bug. It
looks like it happens when the browser is going to render several DIVs (but only
if they are inside a cell of a table!) that have style:

PADDING: 5px;
WIDTH: 90%;
OVERFLOW: auto;

On the other browsers (MSIE, Opera) it works fine. On Mozilla (also on the older
versions) it hangs up too.

Reproducible: Always

Steps to Reproduce:
1. Install the Firefox :)
2. Open a web page that contains this code.
3. Kill Firefox process :) - you can't do anything other.
Actual Results:  
Firefox utilizes 100% CPU and never ends - it just loops infinitely with all the
tabs and windows, not reacting on any input from mouse or keyboard and even not
redrawing its window.

Expected Results:  
Render this page properly.

Firefox 1.0.3, Windows XP, no viruses, no adware/spyware, no additional toolbars
or themes, only small flashblock and livehttpheaders plug-ins instaled in FF.

Comment 1

[Ben Fowler]

Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.8b2) Gecko/20050427
Firefox/1.0+

No crash for me, but the page took a long time to finish.

You might want to remove the attributes to the <TD> open tag, and you seem 
to have an unmatched close </span> tag.

Comment 2

[Steve England [:stevee]]

Dupe of bug 54542?

Comment 3

[Steve England [:stevee]]

Attachment: Reporters Testcase From URL

Comment 4

[Steve England [:stevee]]

Attachment: Testcase that validates

Comment 5

[Steve England [:stevee]]

Attachment: Testcase that is valid HTML 4.0 Transitiona

Comment 6

[Steve England [:stevee]]

Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8b2) Gecko/20050427
Firefox/1.0+
I see the described bug too.

Comment 7

[Radek 'sysKin' Czyz]

Memory usage grows (linearly with time) by 80mb while browser is chewing these
24 divs. That's over 3mb per div, impressive ;)

Not really a memleak since the memory goes back to original level once the
testcase is displayed.

Comment 8

[Bernd]

> Dupe of bug 54542?
No, this bug as a clearly structured testcase, while 54542 is merely a dumping
ground for rant and votes ;-)

Comment 9

[Bernd]

This looks much more like bug 260191. I guess the problem is that xul frames
reflow is pretty inefficient, especially when asked for MEW or maximumwidth when
they are in a table cell. A frame reflow log
(http://www.mozilla.org/newlayout/doc/frame_reflow_debug.html) should be good
enough to verify this. For this a reduced number of divs will probably do the job.
This should improve with bug 240276 when the div with overflow:auto will not use
the xul frame reflow. 

Comment 10

[Radek 'sysKin' Czyz]

Hmm if you add a text to the last div, you can resize it and this will cause
more reflows. This reflow is indeed very very slow, but takes about 1 second an
my computer and no extra memory.
Therefore I'd say it's not reflow taking inital 20 seconds and 80mb. Not
directly, at least.

Comment 11

[Steve England [:stevee]]

Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8b2) Gecko/20050429
Firefox/1.0+
I no longer see the bug. Could this have been fixed by bug 240276?

Comment 12

[Radek 'sysKin' Czyz]

(In reply to comment #11)
> Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8b2) Gecko/20050429
> Firefox/1.0+
> I no longer see the bug. Could this have been fixed by bug 240276?

Confirmed WFM with the same build. No slowdown at load and no slowdown at
reflows. Yay \o/