Bug 30761 - White textures on people in sauerbraten
Summary: White textures on people in sauerbraten
Status: RESOLVED FIXED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/Gallium/r300 (show other bugs)
Version: 7.9
Hardware: x86 (IA32) All
: medium normal
Assignee: Default DRI bug account
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-10-10 20:04 UTC by Jeff Cook
Modified: 2010-11-01 10:07 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
image of white texture on character model (204.53 KB, image/jpeg)
2010-10-10 20:07 UTC, Jeff Cook
Details
console output with RADEON_DEBUG=fp,noopt (780.38 KB, text/plain)
2010-10-10 20:35 UTC, Jeff Cook
Details
console output with RADEON_DEBUG=fp (624.28 KB, text/plain)
2010-10-10 20:36 UTC, Jeff Cook
Details

Description Jeff Cook 2010-10-10 20:04:14 UTC
Have MacBook Pro 2,2 with Radeon Mobility X1600 using ati-dri-gallium-git from ArchLinux AUR, just built a bit earlier tonight. Have Mesa 7.9 and xorg 1.9.

Character models in sauerbraten are usually white and show no textures. I have attached a screenshot of this effect. This game that I just joined is the first game where this effect is not universal -- you can see some other characters appropriately textured, in all other games/maps I have joined _everyone_ has been whited out. This has been this way for at least the last couple of weeks or so.

This is my first bug report for an open-source video driver so please let me know if I am doing it wrong.
Comment 1 Jeff Cook 2010-10-10 20:07:18 UTC
Created attachment 39335 [details]
image of white texture on character model

I tried adding this on the original comment but it seems to have been eaten. Perhaps it was too big as a PNG. I have therefore attached a much smaller jpeg. Apologies if this gets double-posted.
Comment 2 Jeff Cook 2010-10-10 20:19:17 UTC
tstellar in IRC suggested running with RADEON_DEBUG=noopt. This appears to fix the issue.
Comment 3 Jeff Cook 2010-10-10 20:35:30 UTC
Created attachment 39336 [details]
console output with RADEON_DEBUG=fp,noopt

tstellar in IRC suggests that I attach the output of the program with RADEON_DEBUG=fp,noopt.
Comment 4 Jeff Cook 2010-10-10 20:36:21 UTC
Created attachment 39337 [details]
console output with RADEON_DEBUG=fp

tstellar in IRC suggests that I attach the output of the program with RADEON_DEBUG=fp.
Comment 5 Jeff Cook 2010-10-10 20:38:43 UTC
We should note that after a few runs the white model problem becomes less and less frequent. Just now while I was trying to grab these logs, I couldn't get any more white models to display, so I rebooted and now they produce again. Perhaps it is a card initialization variable that is resilient to flipping or something? I know nothing about driver development so please excuse all stupid suggestions I make. :)
Comment 6 Tom Stellard 2010-10-10 22:36:51 UTC
(In reply to comment #5)
> We should note that after a few runs the white model problem becomes less and
> less frequent. Just now while I was trying to grab these logs, I couldn't get
> any more white models to display, so I rebooted and now they produce again.
> Perhaps it is a card initialization variable that is resilient to flipping or
> something? I know nothing about driver development so please excuse all stupid
> suggestions I make. :)

Do all the characters normally look the same?  It is possible that there only a few characters that aren't being rendered correctly and they don't show up every time.  Is the output of RADEON_DEBUG=fp the same no matter how the characters are rendered?
Comment 7 Jeff Cook 2010-10-11 03:46:48 UTC
(In reply to comment #6)
> Do all the characters normally look the same?  It is possible that there only a
> few characters that aren't being rendered correctly and they don't show up
> every time.  Is the output of RADEON_DEBUG=fp the same no matter how the
> characters are rendered?

Well, I'm not sure if it's always the same model that renders incorrectly or not. It seems to fluctuate to me. I know earlier that almost everyone was all white; as I have tried more often, it seems that only a handful are now white/untextured. It may have something to do with the map that is played.

How should I check for fp similarities? Just run diff? I don't really know how to read that output, I was hoping someone here did. Without noopt, I usually get white characters so it is difficult to tell. My laptop is not available immediately atm so I want to clarify before I begin testing. :)
Comment 8 Tom Stellard 2010-10-12 00:15:04 UTC
(In reply to comment #7)
> 
> How should I check for fp similarities? Just run diff? I don't really know how
> to read that output, I was hoping someone here did. Without noopt, I usually
> get white characters so it is difficult to tell. My laptop is not available
> immediately atm so I want to clarify before I begin testing. :)

I would just use diff.
Comment 9 Tom Stellard 2010-10-27 23:47:37 UTC
Can you try the latest version of mesa from git?
Comment 10 Jeff Cook 2010-10-31 21:14:34 UTC
I didn't play very long but I did not observe the whited-out textures on models any more. This appears to be fixed. Sorry I did not provide the requested details earlier; only my laptop runs an ati card and I don't use it very often.

If it's not too much trouble, can you provide an overview of the fix that is suspected to have corrected this bug, just for posterity? I'd appreciate it. :)

Thanks again.
Comment 11 Tom Stellard 2010-11-01 10:07:59 UTC
(In reply to comment #10)
> I didn't play very long but I did not observe the whited-out textures on models
> any more. This appears to be fixed. Sorry I did not provide the requested
> details earlier; only my laptop runs an ati card and I don't use it very often.
> 
> If it's not too much trouble, can you provide an overview of the fix that is
> suspected to have corrected this bug, just for posterity? I'd appreciate it. :)
> 
> Thanks again.


You could always bisect it to see which commit fixed your problem, but my guess is that it was one of these:
a15cf3cd0b21d593033a3abd2b1788de292001bd
aa43176ebd26227947b07221f4e475c52bd7a76b
9d2ab6cb00e72fd8b53d0f97578758504b49ee23


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.