2.6.35-rc6+: i915: Bisected regression

2.6.35-rc6+: i915: Bisected regression

[Thomas Meyer]

This commit makes suspend to ram unreliable for me:

commit d1b851fc0d105caa6b6e3e7c92d2987dfb52cbe0
Author: Zou Nan hai <nanhai.zou@intel.com>
Date:   Fri May 21 09:08:57 2010 +0800

    drm/i915: implement BSD ring buffer V2
    
    The BSD (bit stream decoder) ring is used for accessing the BSD engine
    which decodes video bitstream for H.264 and VC1 on G45+.  It is
    asynchronous with the render ring and has access to separate parts of
    the GPU from it, though the render cache is coherent between the two.
    
    Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
    Signed-off-by: Xiang Hai hao <haihao.xiang@intel.com>
    Signed-off-by: Eric Anholt <eric@anholt.net>

git revert didn't work on 2.6.35-rc6 so I just changed the HAS_BSD thing to:

#define HAS_BSD(dev)            (0)

with this change applied suspend to ram is back to the reliability of 2.6.34.y

with kind regards
thomas

Re: 2.6.35-rc6+: i915: Bisected regression

[Thomas Meyer]

Am 01.08.2010 um 13:42 schrieb Stefan Richter <stefanr@s5r6.in-berlin.de>:

> Thomas Meyer wrote:
>> This commit makes suspend to ram unreliable for me:
>> 
>> commit d1b851fc0d105caa6b6e3e7c92d2987dfb52cbe0
>> Author: Zou Nan hai <nanhai.zou@intel.com>
>> Date:   Fri May 21 09:08:57 2010 +0800
>> 
>>    drm/i915: implement BSD ring buffer V2
>> 
>>    The BSD (bit stream decoder) ring is used for accessing the BSD engine
>>    which decodes video bitstream for H.264 and VC1 on G45+.  It is
>>    asynchronous with the render ring and has access to separate parts of
>>    the GPU from it, though the render cache is coherent between the two.
>> 
>>    Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
>>    Signed-off-by: Xiang Hai hao <haihao.xiang@intel.com>
>>    Signed-off-by: Eric Anholt <eric@anholt.net>
>> 
>> git revert didn't work on 2.6.35-rc6 so I just changed the HAS_BSD thing to:
>> 
>> #define HAS_BSD(dev)            (0)
>> 
>> with this change applied suspend to ram is back to the reliability of 2.6.34.y
>> 
>> with kind regards
>> thomas
> 
> Added CCs.
> 
> Thomas, on what hardware did you observe the regression?

Is a Intel mobile express gm45 chipset.

>  Is it suspend
> or resume that is unreliable?

suspend never finishes/hangs.

>  What are the particular symptoms --- does
> not suspend, crashes during suspend, crashes during resume...?

Does not suspend, backlight is on.

> -- 
> Stefan Richter
> -=====-==-=- =--- ----=
> http://arcgraph.de/sr/

Re: 2.6.35-rc6+: i915: Bisected regression

[Stefan Richter]

Thomas Meyer wrote:
> This commit makes suspend to ram unreliable for me:
> 
> commit d1b851fc0d105caa6b6e3e7c92d2987dfb52cbe0
> Author: Zou Nan hai <nanhai.zou@intel.com>
> Date:   Fri May 21 09:08:57 2010 +0800
> 
>     drm/i915: implement BSD ring buffer V2
>     
>     The BSD (bit stream decoder) ring is used for accessing the BSD engine
>     which decodes video bitstream for H.264 and VC1 on G45+.  It is
>     asynchronous with the render ring and has access to separate parts of
>     the GPU from it, though the render cache is coherent between the two.
>     
>     Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
>     Signed-off-by: Xiang Hai hao <haihao.xiang@intel.com>
>     Signed-off-by: Eric Anholt <eric@anholt.net>
> 
> git revert didn't work on 2.6.35-rc6 so I just changed the HAS_BSD thing to:
> 
> #define HAS_BSD(dev)            (0)
> 
> with this change applied suspend to ram is back to the reliability of 2.6.34.y
> 
> with kind regards
> thomas

Added CCs.

Thomas, on what hardware did you observe the regression?  Is it suspend
or resume that is unreliable?  What are the particular symptoms --- does
not suspend, crashes during suspend, crashes during resume...?
-- 
Stefan Richter
-=====-==-=- =--- ----=
http://arcgraph.de/sr/

Re: 2.6.35-rc6+: i915: Bisected regression

[Paul Rolland]

Hello Thomas,

On Sun, 1 Aug 2010 12:51:36 +0200
Thomas Meyer <thomas@m3y3r.de> wrote:

> Am 01.08.2010 um 13:42 schrieb Stefan Richter <stefanr@s5r6.in-berlin.de>:
> 
> > Thomas, on what hardware did you observe the regression?
> 
> Is a Intel mobile express gm45 chipset.

My machine has :
00:02.0 0300: 8086:2a42 (rev 07)
00:02.1 0380: 8086:2a43 (rev 07)

00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)
00:02.1 Display controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)

(Dell Vostro 1520, BIOS Revision A04)

> >  Is it suspend
> > or resume that is unreliable?
> suspend never finishes/hangs.
and is randomly suffering from exactly the same problems...

So, I've applied your hint (in i915_drv.h), rebuilt a kernel, booted, and
I've now successfully managed to go thru several suspend/resume cycles
without having the machine hanging while suspend !

Will check in the long term, but so far, it's a win !

Thx,
Paul

Re: 2.6.35-rc6+: i915: Bisected regression

[Roberto Oppedisano]

  On 01/08/2010 15:14, Paul Rolland wrote:
> My machine has :
> 00:02.0 0300: 8086:2a42 (rev 07)
> 00:02.1 0380: 8086:2a43 (rev 07)
>
> 00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)
> 00:02.1 Display controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)
>
> (Dell Vostro 1520, BIOS Revision A04)
>
>>>   Is it suspend
>>> or resume that is unreliable?
>> suspend never finishes/hangs.
> and is randomly suffering from exactly the same problems...
>
> So, I've applied your hint (in i915_drv.h), rebuilt a kernel, booted, and
> I've now successfully managed to go thru several suspend/resume cycles
> without having the machine hanging while suspend !
I have to add a me too (HP 6730B, same PCI ids).
I was suffering from the same problem which seem completely gone with 
the suggested hint.

R

Re: 2.6.35-rc6+: i915: Bisected regression

[Shuang He]

  On 2010-8-1 18:51, Thomas Meyer wrote:
> Am 01.08.2010 um 13:42 schrieb Stefan Richter<stefanr@s5r6.in-berlin.de>:
>
>> Thomas Meyer wrote:
>>> This commit makes suspend to ram unreliable for me:
>>>
>>> commit d1b851fc0d105caa6b6e3e7c92d2987dfb52cbe0
>>> Author: Zou Nan hai<nanhai.zou@intel.com>
>>> Date:   Fri May 21 09:08:57 2010 +0800
>>>
>>>     drm/i915: implement BSD ring buffer V2
>>>
>>>     The BSD (bit stream decoder) ring is used for accessing the BSD engine
>>>     which decodes video bitstream for H.264 and VC1 on G45+.  It is
>>>     asynchronous with the render ring and has access to separate parts of
>>>     the GPU from it, though the render cache is coherent between the two.
>>>
>>>     Signed-off-by: Zou Nan hai<nanhai.zou@intel.com>
>>>     Signed-off-by: Xiang Hai hao<haihao.xiang@intel.com>
>>>     Signed-off-by: Eric Anholt<eric@anholt.net>
>>>
>>> git revert didn't work on 2.6.35-rc6 so I just changed the HAS_BSD thing to:
>>>
>>> #define HAS_BSD(dev)            (0)
>>>
>>> with this change applied suspend to ram is back to the reliability of 2.6.34.y
>>>
>>> with kind regards
>>> thomas
>> Added CCs.
>>
>> Thomas, on what hardware did you observe the regression?
> Is a Intel mobile express gm45 chipset.
>
>>   Is it suspend
>> or resume that is unreliable?
> suspend never finishes/hangs.
>
>>   What are the particular symptoms --- does
>> not suspend, crashes during suspend, crashes during resume...?
> Does not suspend, backlight is on.


Could you help report a bug for this issue by following 
http://intellinuxgraphics.org/how_to_report_bug.html

Thanks
     --Shuang

>> -- 
>> Stefan Richter
>> -=====-==-=- =--- ----=
>> http://arcgraph.de/sr/
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: 2.6.35-rc6+: i915: Bisected regression

[David John]

On 08/05/2010 11:13 AM, Shuang He wrote:

> Could you help report a bug for this issue by following
> http://intellinuxgraphics.org/how_to_report_bug.html
> 

I have the same issue on a GM45 and Thomas's hack works in fixing it. It
is still present in 2.6.35.

Regards,
David.

Re: 2.6.35-rc6+: i915: Bisected regression

[Thomas Meyer]

On Sunday 01 August 2010 17:17:54 Roberto Oppedisano wrote:
>   On 01/08/2010 15:14, Paul Rolland wrote:
> > My machine has :
> > 00:02.0 0300: 8086:2a42 (rev 07)
> > 00:02.1 0380: 8086:2a43 (rev 07)
> >
> > 00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)
> > 00:02.1 Display controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)
> >
> > (Dell Vostro 1520, BIOS Revision A04)
> >
> >>>   Is it suspend
> >>> or resume that is unreliable?
> >> suspend never finishes/hangs.

Seems to be fixed in 2.6.36-rc2. Probably by merge commit 
4238a417a91643e1162a98770288f630e37f0484.

Paul, Roberto: Could you please confirm this on your hardware?

kind regards
thomas

Re: 2.6.35-rc6+: i915: Bisected regression

[Roberto Oppedisano]

2010/8/24, Thomas Meyer <thomas@m3y3r.de>:
> On Sunday 01 August 2010 17:17:54 Roberto Oppedisano wrote:
>>   On 01/08/2010 15:14, Paul Rolland wrote:
>> > My machine has :
>> > 00:02.0 0300: 8086:2a42 (rev 07)
>> > 00:02.1 0380: 8086:2a43 (rev 07)
>> >
>> > 00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series
>> > Chipset Integrated Graphics Controller (rev 07)
>> > 00:02.1 Display controller: Intel Corporation Mobile 4 Series Chipset
>> > Integrated Graphics Controller (rev 07)
>> >
>> > (Dell Vostro 1520, BIOS Revision A04)
>> >
>> >>>   Is it suspend
>> >>> or resume that is unreliable?
>> >> suspend never finishes/hangs.
>
> Seems to be fixed in 2.6.36-rc2. Probably by merge commit
> 4238a417a91643e1162a98770288f630e37f0484.
>
> Paul, Roberto: Could you please confirm this on your hardware?

The problem seems fixed for me too.

Re: 2.6.35-rc6+: i915: Bisected regression

[Paul Rolland]

Hello Thomas,

On Tue, 24 Aug 2010 19:36:18 +0200
Thomas Meyer <thomas@m3y3r.de> wrote:

> On Sunday 01 August 2010 17:17:54 Roberto Oppedisano wrote:
> >   On 01/08/2010 15:14, Paul Rolland wrote:
> > > My machine has :
> > > 00:02.0 0300: 8086:2a42 (rev 07)
> > > 00:02.1 0380: 8086:2a43 (rev 07)
> > >
> > > 00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series
> > > Chipset Integrated Graphics Controller (rev 07) 00:02.1 Display
> > > controller: Intel Corporation Mobile 4 Series Chipset Integrated
> > > Graphics Controller (rev 07)
> > >
> > > (Dell Vostro 1520, BIOS Revision A04)
> > >
> > >>>   Is it suspend
> > >>> or resume that is unreliable?
> > >> suspend never finishes/hangs.
> 
> Seems to be fixed in 2.6.36-rc2. Probably by merge commit 
> 4238a417a91643e1162a98770288f630e37f0484.
> 
> Paul, Roberto: Could you please confirm this on your hardware?

I've just installed 2.6.36-rc2 on my machine, and I've tested a few
"Sleep", and so far, no problem any more...

So, I can't help you on the "is this the exact commit ?", but it is fixed
(or at least it looks as if ;)

Regards,
Paul