424766 - [FIX]Left part of the page isn't repainted properly

Status: VERIFIED FIXED

(Core :: Web Painting, defect)

Description

[Mats Palmgren (inactive)]

STEPS TO REPRODUCE
1. load the URL (you should see a login page, enable cookies if you don't)

Try Reload a few times if the problem does not occur at once.

ACTUAL RESULT
Left part of the page isn't repainted properly, see attached screenshot.
Covering the window fixes it.

PLATFORMS AND BUILDS TESTED
Bug occurs in Firefox 2008032304 on MacOSX 10.5.2
Bug occurs in Firefox 2008032204 on Linux
Could not reproduce the problem using Firefox 2.0.0.11 on MacOSX 10.5.2
and Firefox 2.0.0.12 on Linux

Comment 1

[Mats Palmgren (inactive)]

Attachment: Screenshot

Comment 2

[Mats Palmgren (inactive)]

Regression window *appears* to be: 2008-02-06-04 -- 2008-02-07-04
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=2008-02-06+03%3A00&maxdate=2008-02-07+05%3A00&cvsroot=%2Fcvsroot
Possible culprit in that range: bug 414298

Comment 3

[Boris Zbarsky [:bzbarsky]]

I would love a 100% reproducible testcase.  Given one, I can try to fix this, but I just don't have time until late May to try to chase this down if it'll take significant debugging time.

Comment 4

[José Jeria]

Firefox crashes every time that the url is visited... Mac OS X, latest trunk build, fresh profile.

Comment 5

[Mats Palmgren (inactive)]

(the [Silverlight plugin] crash on the same URL is bug 424692)

Comment 6

[Boris Zbarsky [:bzbarsky]]

Mats, does the patch in bug 424465 fix this?

Comment 7

[Mats Palmgren (inactive)]

Nope, still reproducible with that patch in Linux and MacOSX debug builds.

Comment 8

[Boris Zbarsky [:bzbarsky]]

Hmm.  I managed to reproduce this exactly once on Linux...  I'm really going to need help with either making this more reproducible or smaller... :(

Comment 9

[José Jeria]

Attachment: Testcase 1

Not exactly a small testcase, but at least it shows the issue more often than the URL.

You need to do a shift + reload to see the bug. Bug is visible 80% of them times doing this.

What causes the table to jump a bit is the image title of the "LDDR_westlaw_banner.jpg" image. Remove the title and it wont jump again.

I hope this testcase is enough/OK. If not I'll try to minimize it a bit more.

Comment 10

[José Jeria]

Attachment: Testcase 2

Smaller testcase. Shift + reload to trigger issue.

Comment 11

[Boris Zbarsky [:bzbarsky]]

Attachment: Reproducible testcase: load and wait 100ms. There should be no red.

José, thanks!  That helped a lot.

This testcase should show the problem every time, independently of network conditions.  Looking into it now.

Comment 12

[Boris Zbarsky [:bzbarsky]]

Attachment: Fix [Checkin: Comment 17]

I could've sworn this code already looked like this... :(

Comment 13

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 312667 [details] [diff] [review]
Fix
[Checkin: Comment 17]

Requesting approval.  Simple thinko fix.  Quite safe.

Comment 14

[Robert Sayre]

Comment on attachment 312667 [details] [diff] [review]
Fix
[Checkin: Comment 17]

has a reftest, I assume

Comment 15

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

Reftests don't catch invalidation bugs. We don't have a test framework that can catch them, unfortunately. One is planned as part of post-1.9 work.

Comment 16

[Robert Sayre]

ok

Comment 17

[Boris Zbarsky [:bzbarsky]]

Checked in.

Comment 18

[José Jeria]

Verified fixed with: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9pre) Gecko/2008040204 Minefield/3.0pre

Comment 19

[Jonathan Griffin (:jgriffin)]

Attachment: invalidate reftest

Comment 20

[Serge Gautherie (:sgautherie)]

Comment on attachment 384715 [details] [diff] [review]
invalidate reftest


http://hg.mozilla.org/mozilla-central/rev/544c0a55c402

Comment 21

[Serge Gautherie (:sgautherie)]

You may want to the test in 1.9.1 too...

Comment 22

[Dão Gottwald [:dao]]

http://hg.mozilla.org/releases/mozilla-1.9.1/rev/004f430f1c70

Comment 23

[Serge Gautherie (:sgautherie)]

(In reply to comment #22)
> http://hg.mozilla.org/releases/mozilla-1.9.1/rev/004f430f1c70

Please, back out asap: this is timing out on both FF3.5 and SM2.0!

Comment 24

[Dão Gottwald [:dao]]

(In reply to comment #22)
> http://hg.mozilla.org/releases/mozilla-1.9.1/rev/004f430f1c70

backed out as well, because of this:

REFTEST TEST-UNEXPECTED-FAIL | file:///builds/moz2_slave/mozilla-1.9.1-macosx-unittest/build/layout/reftests/bugs/424766-1.html | timed out waiting for reftest-wait to be removed (after onload fired)

Comment 25

[Boris Zbarsky [:bzbarsky]]

Hmm.  I thought we had MozReftestInvalidate on branch... roc, any idea what's up there?

Comment 26

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

I don't think we landed it on branch. It landed on trunk early this year, after we branched for 1.9.1.