[Intel-gfx] nasty suspend resume problem with sandybridge

[Oleksij Rempel (Alexey Fisher)]

Am 13.02.2012 18:06, schrieb Daniel Vetter:
> On Mon, Feb 13, 2012 at 17:53, Oleksij Rempel (fishor)
> <bug-track at fisher-privat.net> wrote:
>> Am 13.02.2012 17:41, schrieb Daniel Vetter:
>>> On Mon, Feb 13, 2012 at 11:44:00AM +0100, Oleksij Rempel (Alexey Fisher) wrote:
>>>> -----BEGIN PGP SIGNED MESSAGE-----
>>>> Hash: SHA1
>>>>
>>>> Am 13.02.2012 11:18, schrieb Daniel Vetter:
>>>>> On Sat, Feb 04, 2012 at 07:17:01PM +0100, Oleksij Rempel (fishor)
>>>>> wrote:
>>>>>> i do not know if this is really graphic related problem, so i
>>>>>> need your help. even excluding some thing will help.
>>>>>>
>>>>>> Please take look at this bug:
>>>>>> https://bugzilla.kernel.org/show_bug.cgi?id=42728
>>>>>
>>>>> Judging from your ehci workaround I think your suspend/resume issue
>>>>> is not i915 related. You mention though somewhere that you also
>>>>> experience memory corruption issues after hibernat (assuming I've
>>>>> read the comments correctly). We are aware of memory corruptions
>>>>> issues due to i915.ko when hibernating, unfortunately no one has
>>>>> yet root-caused this. See e.g.
>>>>>
>>>>> https://bugs.freedesktop.org/show_bug.cgi?id=35648
>>>>
>>>> thx, for the response.
>>>>
>>>> unfortunately this memory corruption appears not in S4, but after
>>>> filed S3.
>>>> Looks like some part entered S3 state and never come back, even after
>>>> poweroff. I think it, because in some cases power-led blinking after
>>>> poweroff - which indicate S3 state. Are there any thing to check to
>>>> see what is misconfigured?
>>>>
>>>> I agree, that i915 is not the reason of this problem, but i915 make it
>>>> impossible to get oops trace! Any chance to get it work?
>>>
>>> As already mentioned by Jesse in the bugzilla, just blacklist it. We don't
>>> really support nomodeset (because for things like suspend/resume it's
>>> actually impossible to properly support).
>>> -Daniel
>>
>> I did response on the bugzilla - blacklisting the module do not help. i
>> get just black screen, no oops trace. Are there any way to get trace
>> with modeset?
> 
> Well, if you've checked that none of the i915, drm and intel-agp
> modules are loaded with lsmod there really shouldn't be any difference
> cause by modeset or nomodeset. I.e. if that's the case, your black
> screen is cause by a suspend/resume failure caused by another driver.
> -Daniel

ufff... again not understood.

>>> i know this issue isn't fault of intel graphic <<<

with my older intel laptop i was able to get oops trace in such case.
With new intel hardware i get blank screen with cursor in left top corner.

>>> Instead of cursor i expect oops trace! <<<
Is it possible? If not please say so. If yes, then is is a bug.

-- 
Regards,
Alexey