https://bugs.freedesktop.org/show_bug.cgi?id=67016
Priority: medium Bug ID: 67016 Assignee: dri-devel@lists.freedesktop.org Summary: Lockup on piglit test vs-textureSize-compare with AMD 6950 Severity: normal Classification: Unclassified OS: Linux (All) Reporter: g02maran@gmail.com Hardware: x86-64 (AMD64) Status: NEW Version: DRI CVS Component: DRM/Radeon Product: DRI
I'm running kernel http://cgit.freedesktop.org/~agd5f/linux/?h=drm-fixes-3.11 with latest commit 444bddc4b9b3313a562cd3ba40f780fb82570f7d and mesa master with latest commit e4fdf1b008ce29c5b5a52985c586b61f35d31e4c
When I run spec/glsl-1.30/execution/vs-textureSize-compare my system locks up. I can't reboot or ssh into it, I have the power cycle the machine to reset it.
Nothing is printed to dmesg before it hangs. I have tried both with and without radeon.dpm=1, no difference.
I did a bisect and it identified this commit: http://cgit.freedesktop.org/~agd5f/linux/commit/?h=drm-fixes-3.11&id=1c0...
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #1 from Martin Andersson g02maran@gmail.com --- Perhaps mesa needs to be updated after this commit. I have tried to change mesa to align to 32768 instead of 4096. With some changes I can make the test pass a few times before it hangs. It does not hard lock everytime with the changes, so I managed to get a dmesg. Don't know how useful it is though because of my, probably incorrect, changes.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #2 from Alex Deucher agd5f@yahoo.com --- Does the test work reliably with that kernel commit reverted? You don't need to adjust the alignment of anything in mesa. I don't see how that commit would cause any regressions. It just adjusts the alignment of VM page table blocks in the kernel driver. Even if cayman doesn't need 32K VM page table alignment, over-aligning shouldn't hurt.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #3 from Martin Andersson g02maran@gmail.com --- (In reply to comment #2)
Does the test work reliably with that kernel commit reverted? You don't need to adjust the alignment of anything in mesa. I don't see how that commit would cause any regressions. It just adjusts the alignment of VM page table blocks in the kernel driver. Even if cayman doesn't need 32K VM page table alignment, over-aligning shouldn't hurt.
I ran the test 500 times without fail with RADEON_VA=0 with that kernel commit.
I also ran the test 500 times without fail with that kernel commit reverted and RADEON_VA=1.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #4 from Martin Andersson g02maran@gmail.com --- Created attachment 82599 --> https://bugs.freedesktop.org/attachment.cgi?id=82599&action=edit dmesg
I managed to get a dmesg without my patches.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #5 from Alex Deucher agd5f@yahoo.com --- Do you still get the issue with dpm disabled?
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #6 from Michel Dänzer michel@daenzer.net --- That commit causes problems (VM faults, lockups) on my Cape Verde card as well.
I suspect other kernel code needs to be adjusted for the increased buffer sizes, or something like that.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #7 from Martin Andersson g02maran@gmail.com --- (In reply to comment #5)
Do you still get the issue with dpm disabled?
yes
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #8 from Alex Deucher agd5f@yahoo.com --- Ok. I'll go ahead and revert it for now.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #9 from Martin Andersson g02maran@gmail.com --- Created attachment 82616 --> https://bugs.freedesktop.org/attachment.cgi?id=82616&action=edit possible fix
This patch fixes the issue for me.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #10 from Alex Deucher agd5f@yahoo.com --- I think that patch is correct. We have to align the PTE block size as well.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #11 from Martin Andersson g02maran@gmail.com --- At least it works for me, I have run a complete piglit test(quick.tests) with that patch without issues, with dpm enabled.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #12 from Christian König deathsimple@vodafone.de --- (In reply to comment #9)
Created attachment 82616 [details] [review] possible fix
This patch fixes the issue for me.
The patch itself is correct, but the problem is it shouldn't be necessary!
When the increment is incorrect we should just make allot of small page directory updates instead of one big update, not as efficient but should work also...
Could you try to setting the increment to something like 0xffffffff and so disable the accumulation of updates and see if that still doesn't work?
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #13 from Alex Deucher agd5f@yahoo.com --- Created attachment 82622 --> https://bugs.freedesktop.org/attachment.cgi?id=82622&action=edit only align the pt base to 32k
Does this patch help? Only the page table base address should need the 32k alignment.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #14 from Martin Andersson g02maran@gmail.com --- (In reply to comment #13)
Created attachment 82622 [details] [review] only align the pt base to 32k
Does this patch help? Only the page table base address should need the 32k alignment.
Yes that patch fixes the problem.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #15 from Christian König deathsimple@vodafone.de --- (In reply to comment #14)
(In reply to comment #13)
Created attachment 82622 [details] [review] [review] only align the pt base to 32k
Does this patch help? Only the page table base address should need the 32k alignment.
Yes that patch fixes the problem.
Even if the problem is fixed for now can you please make the test I suggested?
We are still having some problems with the virtual memory support for NI and it would be nice if we can narrow thos down a bit more.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
--- Comment #16 from Martin Andersson g02maran@gmail.com --- (In reply to comment #15)
(In reply to comment #14)
(In reply to comment #13)
Created attachment 82622 [details] [review] [review] [review] only align the pt base to 32k
Does this patch help? Only the page table base address should need the 32k alignment.
Yes that patch fixes the problem.
Even if the problem is fixed for now can you please make the test I suggested?
We are still having some problems with the virtual memory support for NI and it would be nice if we can narrow thos down a bit more.
I reset the branch to drm-fixes-3.11 and set incr to 0xffffffff. The computer booted fine but when I ran vs-textureSize-compare the computer locked up.
https://bugs.freedesktop.org/show_bug.cgi?id=67016
Alex Deucher agd5f@yahoo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |serkan@hosca.com
--- Comment #17 from Alex Deucher agd5f@yahoo.com --- *** Bug 67102 has been marked as a duplicate of this bug. ***
https://bugs.freedesktop.org/show_bug.cgi?id=67016
Marek Olšák maraeo@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Product|DRI |Mesa Version|DRI CVS |unspecified Component|DRM/Radeon |Drivers/Gallium/r600
https://bugs.freedesktop.org/show_bug.cgi?id=67016
GitLab Migration User gitlab-migration@fdo.invalid changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution|--- |MOVED
--- Comment #18 from GitLab Migration User gitlab-migration@fdo.invalid --- -- GitLab Migration Automatic Message --
This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.
You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/mesa/mesa/issues/454.
dri-devel@lists.freedesktop.org
-
bugzilla-daemon＠freedesktop.org