[Intel-gfx] [PATCH] drm/i915: Update write_domains on active list after flush.
Adam Lantos
hege at playma.org
Sun Feb 7 19:45:31 CET 2010
On Sun, Feb 7, 2010 at 6:55 PM, Daniel Vetter <daniel at ffwll.ch> wrote:
> On Sun, Feb 07, 2010 at 06:39:49PM +0100, Adam Lantos wrote:
>> This patch seems to fix the BUG_ON in my case. However, I still have
>> lots of these:
>
> Good, thanks for testing.
>
>> [drm:i915_gem_object_pin_and_relocate] *ERROR* Failure to install fence: -28
>> [drm:i915_gem_object_pin_and_relocate] *ERROR* Failure to install fence: -28
>> [drm:i915_gem_do_execbuffer] *ERROR* Failed to pin buffer 5 of 7,
>> total 8519680 bytes: -28
>> [drm:i915_gem_do_execbuffer] *ERROR* 369 objects [7 pinned], 91172864
>> object bytes [11747328 pinned], 11780096/260308992 gtt bytes
>>
>> and also visual corruption in the 'present windows' plugin (which was
>> working fine w/ 2.6.32.y)
>
> That's to be expected. Userspace is submitting batchbuffers the hw can't
> handle (that's all the "Failure to install fence" noise in your dmesg
> about, there are simply not enough fences available, -28 == -ENOSPC).
>
> Furthermore there's a bug in the kernel (patch has not yet hit -stable,
> iirc) that artificially limits the available fences from 8 to 5. Just
> cherry pick
>
> drm/i915: Don't reserve compatibility fence regs in KMS mode.
Well, that took care of the dmesg fencing noise, but the visual
corruption didn't stop (see attached screenshot - it should sport a
fancy wallpaper, and a chromium window, but it flickers like hell
instead and sometimes parts of the screen become black and diagonal
lines appear from the left-upper edge).
Is this a known regression from 2.6.32.y to 2.6.33-rc, or should I
file a new fd.o issue?
> from drm-intel-next branch.
> -Daniel
> --
> Daniel Vetter
> Mail: daniel at ffwll.ch
> Mobile: +41 (0)79 365 57 48
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: visual-corruption-2.6.33-rc7-3.png
Type: image/png
Size: 45005 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20100207/2039c6ea/attachment.png>
More information about the Intel-gfx
mailing list