Jesse Barnes ha scritto:
> On Tue, 27 Oct 2009 22:21:39 +0100
> Riccardo Magliocchetti <riccardo.magliocchetti@gmail.com> wrote:
> 
>> Jesse,
>>
>> Jesse Barnes ha scritto:
>>> On Sun, 25 Oct 2009 21:57:32 +0100
>>> Riccardo Magliocchetti <riccardo.magliocchetti@gmail.com> wrote:
>>>
>>>> Hello,
>>>>
>>>> i lost video output, screen stays black, after suspend after
>>>> commit c1c7af60892070e4b82ad63bbfb95ae745056de0 "drm/i915: force
>>>> mode set at lid open time".
>>>>
>>>> I'm attaching my dmesg with 2.6.32-rc5 (actually
>>>> 964fe080d94db82a3268443e9b9ece4c60246414), my bisect log and my
>>>> config.
>>>>
>>>> git bisect response:
>>>>
>>>> c1c7af60892070e4b82ad63bbfb95ae745056de0 is the first bad commit
>>>> commit c1c7af60892070e4b82ad63bbfb95ae745056de0
>>>> Author: Jesse Barnes <jbarnes@virtuousgeek.org>
>>>> Date:   Thu Sep 10 15:28:03 2009 -0700
>>>>
>>>>      drm/i915: force mode set at lid open time
>>>>
>> [snip]
>>> I wonder if this is the real problem...  This patch introduced an
>>> issue that was fixed by 06891e27a9b5dba5268bb80e41a283f51335afe7
>>> that could cause a blank screen on lid open or resume, depending on
>>> how lid events were delivered.  Should be easy enough to test
>>> though, you can remove the whole
>>>
>>> +       if (acpi_lid_open())
>>> +               drm_helper_resume_force_mode(dev);
>>>
>>> block and see if that also helps (in the current tree there's a lock
>>> taken around the force_mode call, you'll need to remove those lines
>>> too for testing).
>> So i've tried separately both this and the patch with the fix for 
>> thinkpad (i have a samsung p460aa01 btw) with latest git and they
>> have not fixed the issue.
>>
>> +#if 0
>>   	if (acpi_lid_open() && !dev_priv->suspended) {
>>   		mutex_lock(&dev->mode_config.mutex);
>>   		drm_helper_resume_force_mode(dev);
>>   		mutex_unlock(&dev->mode_config.mutex);
>>   	}
>> +#endif
> 
> So just removing the above lines doesn't fix things, but reverting the
> whole patch (c1c7af60892070e4b82ad63bbfb95ae745056de0) does?  If so the
> only other thing I can think of is
> 2c907b72db4dd4e8af6dccb6e0ac122d78627b8d, which is in Linus's tree now,
> but may not be in the drm-intel tree you're testing?

Sorry for taking so long to respond. Anyway latest Linus' git master 
with reverted c1c7af60892070e4b82ad63bbfb95ae745056de0 , per attached 
patch, does not fix the issue for me. Please note that just reverting 
the commit does not compile because of an added user of acpi_lid_open() 
which requires acpi/button.h which has been added in 
7121413f2accf14cf05b38539fb7a8be77543370. Will try to revert that too.

thanks,
riccardo