Bug 62243 - Only draw portions of tiles in contentRect
Summary: Only draw portions of tiles in contentRect
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: John Bauman
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-06-07 15:43 PDT by John Bauman
Modified: 2011-06-08 14:16 PDT (History)
3 users (show)

See Also:


Attachments
Patch (2.13 KB, patch)
2011-06-07 15:57 PDT, John Bauman
no flags Details | Formatted Diff | Diff
Patch (8.11 KB, patch)
2011-06-07 17:52 PDT, John Bauman
no flags Details | Formatted Diff | Diff
Patch (4.81 KB, patch)
2011-06-07 18:13 PDT, John Bauman
no flags Details | Formatted Diff | Diff
Correct for offset top left corner, add rebaseline requirement (6.14 KB, patch)
2011-06-07 19:17 PDT, John Bauman
no flags Details | Formatted Diff | Diff
Hopefully up to date (6.10 KB, patch)
2011-06-08 01:06 PDT, John Bauman
jamesr: review+
Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description John Bauman 2011-06-07 15:43:10 PDT
Only draw portions of tiles in contentRect
Comment 1 John Bauman 2011-06-07 15:57:46 PDT
Created attachment 96321 [details]
Patch
Comment 2 John Bauman 2011-06-07 15:59:50 PDT
This patch seems to work, although I'm not certain it's logically correct. This bug is the root cause of http://code.google.com/p/chromium/issues/detail?id=84901 .
Comment 3 James Robinson 2011-06-07 16:03:11 PDT
Comment on attachment 96321 [details]
Patch

Can you make a reduced test case, please?
Comment 4 James Robinson 2011-06-07 16:03:49 PDT
Comment on attachment 96321 [details]
Patch

Even if this patch is correct (I haven't thought about it in detail) we'll need a regression test.
Comment 5 John Bauman 2011-06-07 17:52:44 PDT
Created attachment 96346 [details]
Patch
Comment 6 James Robinson 2011-06-07 17:57:51 PDT
Comment on attachment 96346 [details]
Patch

The two -expected.pngs are identical, so just put one copy of it in LayoutTests/compositing/repaint/ next to the test itself.

Also in the test body put the incantation:
if (window.layoutTestController)
  layoutTestController.dumpAsText(true);

to get it to only output the pixel results and not output any render tree dump.
Comment 7 John Bauman 2011-06-07 17:59:46 PDT
Oh, I was so shocked that DRT gave the same results on two different platforms that I didn't know how to respond.
Comment 8 Adrienne Walker 2011-06-07 18:13:50 PDT
Thanks for fixing this.  This looks good to me too.

Does this change not cause any other layout test failures when compositing? I've seen no end of tiny filtering changes (theoretically due to floating point math) when quad sizes have changed in the past.
Comment 9 John Bauman 2011-06-07 18:13:56 PDT
Created attachment 96350 [details]
Patch
Comment 10 John Bauman 2011-06-07 18:21:06 PDT
Hmm, there do seem to be some unexpected failures with this patch. I'll have to look some more and see if they're really unexpected on this computer.
Comment 11 John Bauman 2011-06-07 18:30:20 PDT
Comment on attachment 96350 [details]
Patch

seems to fail some layout tests.
Comment 12 John Bauman 2011-06-07 19:17:21 PDT
Created attachment 96355 [details]
Correct for offset top left corner, add rebaseline requirement
Comment 13 James Robinson 2011-06-07 19:43:11 PDT
Comment on attachment 96355 [details]
Correct for offset top left corner, add rebaseline requirement

R=me, enne can you take a look and flip cq+ if you're happy?
Comment 14 WebKit Review Bot 2011-06-07 21:33:49 PDT
Comment on attachment 96355 [details]
Correct for offset top left corner, add rebaseline requirement

Rejecting attachment 96355 [details] from commit-queue.

Failed to run "['./Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '--bot-id=ec2-cq-01', '--port..." exit_code: 2

Last 500 characters of output:
t/shrink-layer.html
patching file LayoutTests/platform/chromium/test_expectations.txt
Hunk #1 FAILED at 4009.
1 out of 1 hunk FAILED -- saving rejects to file LayoutTests/platform/chromium/test_expectations.txt.rej
patching file Source/WebCore/ChangeLog
Hunk #1 succeeded at 1 with fuzz 3.
patching file Source/WebCore/platform/graphics/chromium/LayerTilerChromium.cpp

Failed to run "[u'/mnt/git/webkit-commit-queue/Tools/Scripts/svn-apply', u'--reviewer', u'James Robinson', u'--f..." exit_code: 1

Full output: http://queues.webkit.org/results/8810030
Comment 15 John Bauman 2011-06-08 01:06:00 PDT
Created attachment 96395 [details]
Hopefully up to date
Comment 16 James Robinson 2011-06-08 14:07:36 PDT
Comment on attachment 96395 [details]
Hopefully up to date

Will land by hand
Comment 17 James Robinson 2011-06-08 14:16:52 PDT
Committed r88386: <http://trac.webkit.org/changeset/88386>