Bug 60402 - [gen4] SNA driver has again some artifacts (cf 55500)
Summary: [gen4] SNA driver has again some artifacts (cf 55500)
Status: RESOLVED WORKSFORME
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: git
Hardware: Other All
: medium normal
Assignee: Chris Wilson
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2013-02-07 08:58 UTC by Zdenek Kabelac
Modified: 2013-02-26 14:58 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
Artifacts in terminal tab. (6.14 KB, image/png)
2013-02-07 08:58 UTC, Zdenek Kabelac
no flags Details
gedit tabs gm45 sna 2.21.2 (5.49 KB, image/png)
2013-02-11 07:55 UTC, Edward Sheldrake
no flags Details

Description Zdenek Kabelac 2013-02-07 08:58:11 UTC
Created attachment 74324 [details]
Artifacts in terminal tab.

This BZ looks a bit similar to already fixed one Bug 50477.

However this time I do not have an easy way to 'trigger' this rendering error.
It usually happens with  gnome-terminal - and I'm not sure now if I've seen it with some other app.  As seen as application get some window focus event and is redrawn - it disappears.

I'm attaching small PNG cut of gnome-terminal tab.

I'm currently running xf86 commit 9c80a0337ec12b6baab5aab380503e672e925677
(from Feb 5) - but I'm seeing this problem already for a few weeks.

But bisecting would be quite nontrivial - until I'd discover a decent easy to reliable trigger it.

Just for completeness - hw T61, Fedora Rawhide, 
xorg-x11-server-Xorg-1.13.99.901-4.20130109.fc19.x86_64
gnome-terminal-3.7.2-2.fc19.x86_64
Comment 1 Chris Wilson 2013-02-07 09:14:22 UTC
Actually that looks like one of the rendering artifacts of bug 55500. But it could conceivably be a bug in the trapezoid mask generation, however that would be deterministic.

I'll leave this open for a few days in case you can narrow down the cause/effect, but for the time being I'm going to consider it as just another gen4 misrendering bug.
Comment 2 Edward Sheldrake 2013-02-11 07:55:33 UTC
Created attachment 74586 [details]
gedit tabs gm45 sna 2.21.2

The lines in GTK3 tabs have become noticeably worse in 2.21.2 (with GM45 SNA). With 2.21.0 (and earlier), the problem was generally limited to only a single line in one tab, which usually cleared up as things get re-drawn.

With 2.21.2, there are lines in all tabs, and they don't go away as you click on different tabs, the lines just change.
Comment 3 Chris Wilson 2013-02-13 18:40:32 UTC
If you have the opportunity, can you quickly check the corruption remains if compiled with -O0?
Comment 4 Edward Sheldrake 2013-02-13 19:15:41 UTC
The corruption is still present after rebuilding the driver with -O0.

I noticed FORCE_NONRECTILINEAR_SPANS is still defined in gen4_render.c but it doesn't seem to be used anywhere (removed in c69b4389abc324533a9a311c17a667bf8a1e1673) - restoring the bit of code that used it got rid of the corruption.
Comment 5 Chris Wilson 2013-02-13 20:56:00 UTC
Ta, my fault. I was trying to rework all the gen at the same time, my intention for gen4 was to simply mark it as PREFER_GPU_SPANS==false and have that fall out. Will push the fixed up workaround.
Comment 6 Chris Wilson 2013-02-13 22:10:28 UTC
With commit 86e025a1842f3c2f319676818b1f4624cd94ebc5
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Wed Feb 13 22:05:34 2013 +0000

    sna/gen4: Restore nonrectiinear spans workaround
    
    Instead of relying on the macro, the intention was simply to use the
    prefer_gpu hint. However, I dropped it whilst refactoring ideas from
    later generations. So restore both the debug control to force spans as
    well as the intended workaround.
    
    Reported-by: Edward Sheldrake <ejsheldrake@gmail.com>
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

are we back to the less regular flickering corruption?
Comment 7 Edward Sheldrake 2013-02-14 10:38:33 UTC
GTK3 tabs are now fine with current git, better than 2.21.0.

I've never experienced any flickering corruption like in bug 55500 but then I don't use chromium.
Comment 8 Chris Wilson 2013-02-14 12:25:13 UTC
Closing this, as I suspect we've reduced this back to bug 55500 now.


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.