Created attachment 93791 [details]
Cogs shadow corruption
When enabling shadows on Cogs, the game looks like the attached screenshot, shadows become corrupt.
GPU: HD2600 Mobility
OS: Ubuntu 13.10
Running the latest builds of Mesa thanks to the oibaf repository.
Disabling shadows allows me to play the game fine.
Created attachment 93792 [details]
Cogs shadow corruption
Is this a regression? If so, can you bisect? Also, please attach your xorg log and glxinfo and dmesg output.
No idea if it's a regression. The first time I played this game on Linux was ~2 weeks ago and the issue has been there since that time.
Shadows are on by default, and this is a pretty old game. My guess is it is a regression or it would have been noticed previously.
I'm not sure what bisect is or how to perform one. I'm also not sure where the xorg log is. Is that constantly logging? Should I delete the file and launch the game to create a small one if so?
glxinfo and dmesg are too large, how do I dump them to a file?
Just redirect the output to a file:
glxinfo > glxinfo.txt
dmesg > dmesg.txt
The xorg log location is usually /var/log/Xorg.0.log. Don't worry about deleting it. It gets overwritten each time you start X.
Google for "git bisect howto" to see how to use git to bisect a regression. If you are not comfortable with that, can you try older driver releases and see if you can narrow down which one caused the problem?
Created attachment 93797 [details]
Created attachment 93798 [details]
I purged the PPA which downgraded me to mesa 9.2.1 (I think). The issue is still there.
Try to set this environment variable:
I launched Cogs via Steam and used the following "launch option", hopefully that's correct.
That actually fixed the shadows, they all look normal.
Another game seems to be affected by this issue, "Anomaly Warzone Earth". Disabling HyperZ for that game also solves the shadows...
..but it reveals some other graphical anomalies and crashing issues probably more suited for another ticket.
Probably the same bug happens for L4D2:
Re-assigning back to r600g
(In reply to comment #12)
> Re-assigning back to r600g
Why r600g? Is HyperZ not enabled on others?
(In reply to comment #13)
> (In reply to comment #12)
> > Re-assigning back to r600g
> Why r600g? Is HyperZ not enabled on others?
There should be a separate bug for hyperz issues with other drivers.
http://www.mesa3d.org/relnotes/10.0.5.html stated this bug is fixed :).
Is it really fixed for r600 driver with enabled hyperz or it is fixed because hyperz is disabled by default?
I also asking this because i don't have this bug in radeonsi :).
You can try with HyperZ by passing R600_DEBUG=hyperz and running it again.
Most likely it's fixed by disabling HyperZ.
I haven't tested since I made the ticket. I assumed if there was going to be a change that fixed it that someone would respond here to inform me.
Is this fixed with current Mesa git?
I'll update my oibaf laptop and try get back to you as soon as possible.
Oh I should ask this now. Is HyperZ enable now or do I need to change the Steam launch options?
You need to set the environment variable:
No. Still broken using latest Oibaf.
(In reply to comment #22)
> No. Still broken using latest Oibaf.
Latest oibaf ppa packages still does not contain these fixes :) You need to wait for newer packages, one more update :)
Updated Oibaf and tried again. This time it does appear to be fixed! :-)
(In reply to comment #24)
> Updated Oibaf and tried again. This time it does appear to be fixed! :-)
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct.