Created attachment 71229 [details] Screenshot of UI Problem With 3.0 Mesa 9.1-devel (git-122dfc5) i get the attached garbled screen output when starting Linux native Team Fortress 2 Beta client. I tried all of the visual settings under advanced options but the problem persists. If you need more input feel free to ask.
Created attachment 71230 [details] 32bit glxinfo
Is this a regression or has it always had rendering problems? Does it render ok with mesa 9.0.x or 8.0.x? If so, can you bisect?
Only tried it with mesa from git but i can give 9.0 a shot. Though as far as I know the native client is not officially released yet, so I don't know if bisecting will help.
1eedebc65b02130ef7a27062a1ed67972a317a08 is the first bad commit commit 1eedebc65b02130ef7a27062a1ed67972a317a08 Author: Marek Olšák <maraeo@gmail.com> Date: Thu Nov 1 02:00:37 2012 +0100 r600g: re-enable handling of DISCARD_RANGE, improving performance It seems to work for me now. Even the graphics corruption is gone. This also boosts performance in Reaction Quake. Reverting this commit on current master (e721a76) does fix the garbled UI on my rv770.
I have the garbled issue, but if I try to play the game it soon crashes. Do you all have this additional issue as well?
(In reply to comment #5) > I have the garbled issue, but if I try to play the game it soon crashes. Do > you all have this additional issue as well? Nope, I can play with the garbled issue and it doesn't crash.
*** Bug 60224 has been marked as a duplicate of this bug. ***
Do the piglit transform feedback tests pass?
All 241 tests with this command pass: piglit-run.py -t transform_feedback /usr/share/piglit/tests/all.tests
Created attachment 74780 [details] Full results of quick-driver piglit tests
I just tried steam and tf2 on linux, and I didn't see any rendering error. The only abnormal thing was this line: This system DOES NOT support the OpenGL extension GL_ARB_uniform_buffer. HD6850, OpenGL version string: 3.0 Mesa 9.2-devel (git-b681ed6)
Could you please attach an apitrace file showing the issue?
Apitrace done.. It's about 300MB so I uploaded it to GoogleDrive. https://docs.google.com/file/d/0B9PdLrdrtm1wRW9VdVJiOTl5YVU/edit?usp=sharing
Fixed by Mesa commit aac81387447152b5bda99150414d636a003adc5. Closing.
I can still reproduce the garbled ui on my rv770 Mesa f6b40ddd2d92fd82ccf7ce695d4ed026ef7cfd11 Kernel 3.8.0
the patches in bug 50655 may help.
Created attachment 75334 [details] [review] possible fix Does this patch help?
Nope it doesn't help :/
Created attachment 75374 [details] [review] possible fix How about this one?
It doesn't help either.
I've tested Mesa f6b40ddd2d92fd82ccf7ce695d4ed026ef7cfd11 with the patch set "r6xx flushing rework and enable CP DMA (v2)" from http://lists.freedesktop.org/archives/mesa-dev/2013-February/035081.html and it doesn't fix the issue. Note that older kernels (eg 3.5) which don't have CP DMA are affected by this issue too. I could reproduce the garbled UI on Counter-Strike: Source too. Reverting 1eedebc65b02130ef7a27062a1ed67972a317a08 fixes the issue on CSS.
(In reply to comment #21) > I could reproduce the garbled UI on Counter-Strike: Source too. > Reverting 1eedebc65b02130ef7a27062a1ed67972a317a08 fixes the issue on CSS. Me too. I have a RV770 and reverting 1eedebc65b02130ef7a27062a1ed67972a317a08 fixes the issue on CSS.
Mesa 9.2-devel (git-4deefd9) showed the problem here with TF2, I don't have CSS. Added patches: r600g: emit a ps partial flush after CP DMA r600g: enable CP DMA on r6xx (v3) and the problem persisted. Then reverting commit 1eedebc65b02130ef7a27062a1ed67972a317a08 fixed the issue. Mesa 9.0.3 works. HD3850 (RV670) AGP, kernel 3.8.0 First comment so I want to thank the devs for their work. :)
(In reply to comment #23) > Mesa 9.2-devel (git-4deefd9) showed the problem here with TF2, I don't have > CSS. > Added patches: > r600g: emit a ps partial flush after CP DMA > r600g: enable CP DMA on r6xx (v3) > and the problem persisted. > Then reverting commit 1eedebc65b02130ef7a27062a1ed67972a317a08 fixed the > issue. > Mesa 9.0.3 works. > HD3850 (RV670) AGP, kernel 3.8.0 > > First comment so I want to thank the devs for their work. :) Nice to hear from you! Could you test this patch http://lists.freedesktop.org/archives/mesa-dev/2013-February/035347.html It helps here on rv770 and kernel 3.8.0
> Could you test this patch > http://lists.freedesktop.org/archives/mesa-dev/2013-February/035347.html Patch 3/5 doesn't help for me. Should I try out the other patches in that series? (nee ATI RV670 [Radeon HD 3870], Git kernel (pre 3.9))
(In reply to comment #25) > > Could you test this patch > > http://lists.freedesktop.org/archives/mesa-dev/2013-February/035347.html > > Patch 3/5 doesn't help for me. Should I try out the other patches in that > series? (nee ATI RV670 [Radeon HD 3870], Git kernel (pre 3.9)) The only patch that's relevant for r6xx is 2/5. Does that help? You might need to apply 1/5 first.
> The only patch that's relevant for r6xx is 2/5. Does that help? That's just turning that off (same as a revert specific to r600 though). Works perfectly.
Andreas Boll requested a branch of all my latest Mesa patches for people to test it. It's here: git://people.freedesktop.org/~mareko/mesa master
(In reply to comment #28) > Andreas Boll requested a branch of all my latest Mesa patches for people to > test it. It's here: > > git://people.freedesktop.org/~mareko/mesa master This branch fixes the garbled screen and improves the performance (~ 120 - 160%) here with tf2 and css on my rv770. Please test it!
(In reply to comment #28) > Andreas Boll requested a branch of all my latest Mesa patches for people to > test it. It's here: > > git://people.freedesktop.org/~mareko/mesa master Awesome! With my RV770 all is rendering correctly now in CS:S! Moreover, I'm always at 60FPS (vsync certainly enabled) with all graphical options enabled (even MSAA 2x) (1680*1050)! Note that I'm using 3.7.9-201.fc18.x86_64.
I tested master Mesa 9.2-devel (git-a40ba43) and TF2 works with my RV670.
Tested and works for me!
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.