[Intel-gfx] linux-next: Tree for Aug 13 [ screen corruption in graphical mode ]

[Sedat Dilek]

On Tue, Aug 13, 2013 at 4:45 PM, Sedat Dilek <sedat.dilek at gmail.com> wrote:
> On Tue, Aug 13, 2013 at 4:07 PM, Sedat Dilek <sedat.dilek at gmail.com> wrote:
>> On Tue, Aug 13, 2013 at 11:57 AM, Sedat Dilek <sedat.dilek at gmail.com> wrote:
>>> On Tue, Aug 13, 2013 at 11:52 AM, Chris Wilson <chris at chris-wilson.co.uk> wrote:
>>>> On Tue, Aug 13, 2013 at 11:47:19AM +0200, Sedat Dilek wrote:
>>>>> On Tue, Aug 13, 2013 at 11:39 AM, Chris Wilson <chris at chris-wilson.co.uk> wrote:
>>>>> > On Tue, Aug 13, 2013 at 11:35:53AM +0200, Sedat Dilek wrote:
>>>>> >> After a logout from my "BROKEN" Unity-2D session - the login-screen
>>>>> >> for LightDM seems to be OK.
>>>>> >> Then entering my Unity-2D desktop is OK - no screen corruptions.
>>>>> >
>>>>> > What hardware and display do you have?
>>>>>
>>>>> It's a Samsung ultrabook with SandyBridge CPU.
>>>>>
>>>>> [   333.291] (--) intel(0): Integrated Graphics Chipset: Intel(R) HD
>>>>> Graphics 3000
>>>>
>>>> using LVDS.
>>>>
>>>>> - Sedat -
>>>>>
>>>>> P.S.: I switched to intel-ddx v2-21-14-35-g5840bf in the meantime.
>>>>
>>>> Did that make a difference? It shouldn't if the error is occuring before
>>>> X even starts...
>>>
>>> NO, was just confused not seeing "GT2" (HD-3000 was new to me) in my
>>> Xorg.log :-).
>>>
>>> As said logging out of Unity-2D and entering LightDM greeter - screen is fine.
>>> Starting again a Unity-2D session - no screen corruption, too.
>>>
>>> - Sedat -
>>
>> Some more testing:
>>
>> [1] With my X stack:
>>
>> FIRST BAD: next-20130812
>> LAST GOOD: next-20130809
>>
>> [2] With Ubuntu's X stack:
>>
>> next-20130813 is OK (Xorg.log attached)
>>
>
> drm-intel-nightly is also BAD with my X stack (with Ubuntu's X stack
> no problems).
>

I have bisected the issue on Linux v3.11-rc5 + drm-intel-nightly:

5456fe3882812aba251886e36fe55bfefb8e8829 is the first bad commit
commit 5456fe3882812aba251886e36fe55bfefb8e8829
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date:   Thu Aug 8 14:41:07 2013 +0100

    drm/i915: Allocate LLC ringbuffers from stolen

    As stolen objects now behave identically (wrt to default LLC cacheing)
    as their normal system counterparts, we no longer have to differentiate
    our usage for ringbuffers. So allocate them from stolen on SNB+ as well.

    Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
    Reviewed-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
    Signed-off-by: Daniel Vetter <daniel.vetter at ffwll.ch>

:040000 040000 de063a052f39095f4d2f51b49caef9f827df41e8
1c819aa5501a9fcc9912a5c7c037c71b9b9e9a6b M      drivers

See also attached files!

- Sedat -
-------------- next part --------------
git bisect start
# good: [d4e4ab86bcba5a72779c43dc1459f71fea3d89c8] Linux 3.11-rc5
git bisect good d4e4ab86bcba5a72779c43dc1459f71fea3d89c8
# bad: [5b79c8eb1b61d499ca70cc3e4e2ced7822209876] Merge branch 'drm-intel-nightly' into 3.11.0-rc5-iniza-small
git bisect bad 5b79c8eb1b61d499ca70cc3e4e2ced7822209876
# good: [36f2d1f151215c48d902947d64b86dc5ab277e19] drm/i915: rip out legacy encoder->mode_set callback
git bisect good 36f2d1f151215c48d902947d64b86dc5ab277e19
# good: [c35426d2bc25b242ee2a9a7a1d62634be1e86bb0] drm/i915: Split plane watermark parameters into a separate struct
git bisect good c35426d2bc25b242ee2a9a7a1d62634be1e86bb0
# good: [32c913e4369ce7bd1d16a9b6983f7b8975c13f5a] Merge tag 'drm-intel-next-2013-07-26-fixed' of git://people.freedesktop.org/~danvet/drm-intel into drm-next
git bisect good 32c913e4369ce7bd1d16a9b6983f7b8975c13f5a
# good: [d46f1c3f1372e3a72fab97c60480aa4a1084387f] drm/i915: Allow the GPU to cache stolen memory
git bisect good d46f1c3f1372e3a72fab97c60480aa4a1084387f
# bad: [b8a1868b10bb4fe7fb7d283da5d56064b1a189f4] drm/i915: Allow the user to set bo into the DISPLAY cache domain
git bisect bad b8a1868b10bb4fe7fb7d283da5d56064b1a189f4
# bad: [a698a20e8cf9a946b0a491cb58ff96c0b4332d08] i915: Fix SDVO potentially turning off randomly
git bisect bad a698a20e8cf9a946b0a491cb58ff96c0b4332d08
# bad: [8d2d9e6517d95d2bfa35dfa650330cf0b591d7e1] drm/i915: Only do a chipset flush after a clflush
git bisect bad 8d2d9e6517d95d2bfa35dfa650330cf0b591d7e1
# bad: [5456fe3882812aba251886e36fe55bfefb8e8829] drm/i915: Allocate LLC ringbuffers from stolen
git bisect bad 5456fe3882812aba251886e36fe55bfefb8e8829
# first bad commit: [5456fe3882812aba251886e36fe55bfefb8e8829] drm/i915: Allocate LLC ringbuffers from stolen
-------------- next part --------------
commit 5456fe3882812aba251886e36fe55bfefb8e8829
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date:   Thu Aug 8 14:41:07 2013 +0100

    drm/i915: Allocate LLC ringbuffers from stolen
    
    As stolen objects now behave identically (wrt to default LLC cacheing)
    as their normal system counterparts, we no longer have to differentiate
    our usage for ringbuffers. So allocate them from stolen on SNB+ as well.
    
    Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
    Reviewed-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
    Signed-off-by: Daniel Vetter <daniel.vetter at ffwll.ch>

 drivers/gpu/drm/i915/intel_ringbuffer.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
-------------- next part --------------
commit 5456fe3882812aba251886e36fe55bfefb8e8829
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date:   Thu Aug 8 14:41:07 2013 +0100

    drm/i915: Allocate LLC ringbuffers from stolen
    
    As stolen objects now behave identically (wrt to default LLC cacheing)
    as their normal system counterparts, we no longer have to differentiate
    our usage for ringbuffers. So allocate them from stolen on SNB+ as well.
    
    Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
    Reviewed-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
    Signed-off-by: Daniel Vetter <daniel.vetter at ffwll.ch>