On 11/11/13 01:39 PM, Julien Cristau wrote: > "Unreviewed" and "testers said it doesn't work" aren't exactly the > same. Which is it?
A bit of both. Chris Wilson provided two patches. After applying the first patch, one tester said the problem was fixed. Chris then came up with a second patch which he thought would clear up the root cause of the problem. Two testers reported the second patch did not fix the issue, so Chris needs to review both patches to figure out what's wrong with his attempt to address the root cause. I guess we could try the first patch against the version in Debian, but before doing that, I would want to know why Chris went on to try to solve it a different way (which apparently failed). Here's the first patch which reportedly works: https://bugs.freedesktop.org/attachment.cgi?id=75820 Here are the comments between Chris and the testers regarding both patches: Chris Wilson 2013-03-03 10:05:29 UTC Created attachment 75820 [details] [review] Fix up fence counts My belief is that the error in fence counting is magnified through the clear_relocs() function, and if true this patch should fix up the leak. aux...@gmail.com 2013-05-03 13:39:13 UTC Glyph corruption has been absent for the past two days after I applied the patch, even after suspend to RAM. This seems to be fixed, thanks. [reply] [-] Comment 19 Chris Wilson 2013-05-08 16:45:15 UTC Created attachment 79029 [details] [review] Avoid overcounting fences for self-relocs I think this is the root cause of the miscounting issue. Please test without the other patch applied. [reply] [-] Comment 20 aux...@gmail.com 2013-05-08 20:05:26 UTC Empty windows and the message (EE) intel(0): Failed to submit batch buffer, expect rendering corruption: Resource deadlock avoided. just appeared with the second patch applied. [reply] [-] Comment 21 M.R 2013-05-11 18:20:42 UTC Second patch didn't fix it for me either. Atom N550 Desktop: GNOME 3.8 (Fedora 19) Patch applied on libdrm commit 040f6b015e Ben
signature.asc
Description: OpenPGP digital signature
