Bug 86877 - Must set/reset pixel unpack alignment to 1 during texSubImage2D
Summary: Must set/reset pixel unpack alignment to 1 during texSubImage2D
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: WebGL (show other bugs)
Version: 528+ (Nightly build)
Hardware: All All
: P2 Normal
Assignee: Kenneth Russell
URL:
Keywords:
Depends on:
Blocks: 87042 87136
  Show dependency tree / graph
 
Reported: 2012-05-18 11:22 PDT by Kenneth Russell
Modified: 2012-05-22 09:26 PDT (History)
9 users (show)

See Also:

Attachments
Patch (63.76 KB, patch)
2012-05-21 14:27 PDT, Kenneth Russell 		no flags Details | Formatted Diff | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Kenneth Russell 2012-05-18 11:22:22 PDT 
Most of WebGLRenderingContext's texImage2D and texSubImage2D code paths, in particular those taking DOM elements and ImageData, have to pack the texture data into a temporary buffer before uploading. This temporary buffer is tightly packed, so the UNPACK_ALIGNMENT pixel store parameter has to be set to 1 during the upload. This was done for texImage2D a long time ago but is missing in the texSubImage2D code paths.
Comment 1 Kenneth Russell 2012-05-21 14:27:00 PDT 
Created attachment 143095 [details]
Patch
Comment 2 James Robinson 2012-05-21 14:32:24 PDT 
Comment on attachment 143095 [details]
Patch

R=me.  C++ changes look fine, I haven't carefully looked at the layout tests but since they are imports I'm assuming they are fine.
Comment 3 WebKit Review Bot 2012-05-21 23:57:21 PDT 
Comment on attachment 143095 [details]
Patch

Clearing flags on attachment: 143095

Committed r117918: <http://trac.webkit.org/changeset/117918>
Comment 4 WebKit Review Bot 2012-05-21 23:57:31 PDT 
All reviewed patches have been landed.  Closing bug.
Comment 5 Brady Eidson 2012-05-22 08:56:28 PDT 
(In reply to comment #2)
> (From update of attachment 143095 [details])
> R=me.  C++ changes look fine, I haven't carefully looked at the layout tests but since they are imports I'm assuming they are fine.

They weren't.  This started 4 failures on Mac bots

http://build.webkit.org/results/Lion%20Release%20(Tests)/r117918%20(8526)/results.html