Bug 3944 - GdkPixbuf rendering with fbCompositeGeneral and shadowfb in CVS HEAD
Summary: GdkPixbuf rendering with fbCompositeGeneral and shadowfb in CVS HEAD
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Server/General (show other bugs)
Version: git
Hardware: x86 (IA32) Linux (All)
: high normal
Assignee: Xorg Project Team
QA Contact:
URL:
Whiteboard:
Keywords:
: 4152 5108 5152 (view as bug list)
Depends on:
Blocks: 1690
  Show dependency treegraph
 
Reported: 2005-08-02 04:53 UTC by Søren Sandmann Pedersen
Modified: 2005-12-08 11:29 UTC (History)
4 users (show)

See Also:
i915 platform:
i915 features:


Attachments
Screenshot showing the bug (32.53 KB, image/png)
2005-09-22 11:08 UTC, Søren Sandmann Pedersen
no flags Details
A patch (620 bytes, patch)
2005-12-08 06:30 UTC, Søren Sandmann Pedersen
no flags Details | Splinter Review
New patch (963 bytes, patch)
2005-12-08 06:38 UTC, Søren Sandmann Pedersen
no flags Details | Splinter Review

Description Søren Sandmann Pedersen 2005-08-02 04:53:01 UTC
With the new scanline based fbCompositeGeneral and a packed 24 bit shadowFB,
GdkPixbuf rendering is broken as the soon-to-be-attached screenshot shows.

GdkPixbufs are non-premultiplied RGBA images. To make RENDER deal with this
format, gtk+ renders gdkpixbufs with the image itself as the image, ie.:

   RGBA-image IN RGBA-image OVER destination

where the IN-operation makes sure the image that is composited OVER the
destination, will actually be premultiplied.

I have verified that the bug was introduced with the commit of the scanline
based fbCompositeGeneral.
Comment 1 Søren Sandmann Pedersen 2005-08-02 04:53:32 UTC
Created attachment 3205 [details] [review]
proposed fix
Comment 2 Søren Sandmann Pedersen 2005-09-22 11:08:22 UTC
Created attachment 3366 [details]
Screenshot showing the bug

Reattching the screenshot since the old one was apparently lost in the disk
crash
Comment 3 Søren Sandmann Pedersen 2005-11-30 06:25:56 UTC
Lots of people are hitting this, so I'm making it block 6.9/7.0
Comment 4 Adam Jackson 2005-11-30 15:56:51 UTC
*** Bug 5108 has been marked as a duplicate of this bug. ***
Comment 5 Adam Jackson 2005-11-30 15:58:06 UTC
*** Bug 4152 has been marked as a duplicate of this bug. ***
Comment 6 Adam Jackson 2005-11-30 15:59:17 UTC
*** Bug 5152 has been marked as a duplicate of this bug. ***
Comment 7 Søren Sandmann Pedersen 2005-12-08 06:30:49 UTC
Created attachment 4030 [details] [review]
A patch

Note: This patch is untested, but I am pretty confident in it anyway, because
the bug it fixes

- would have symptoms like those reported
- would only happen on 24 bit packed framebuffers
- was introduced with the scanline based compositing.

all of which are consistent with what people have reported.

[If I get the server to compile on my system, I'll test it].
Comment 8 Søren Sandmann Pedersen 2005-12-08 06:38:19 UTC
Created attachment 4031 [details] [review]
New patch

There is another instance of the same bug.
Comment 9 Søren Sandmann Pedersen 2005-12-09 03:51:34 UTC
Chris Aillon happened to have a system displaying this bug, and the patch does
indeed fix it.
Comment 10 Adam Jackson 2005-12-09 06:29:11 UTC
that makes you my new hero.

applied, thanks!


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.