[Intel-gfx] [PATCH RFC] drm/i915: reduce cursor size for GEN5 hardware

[Martin Peres]

On 07/02/17 15:22, Uwe Kleine-König wrote:
> Hello,
>
> On 02/01/2017 03:37 PM, Ville Syrjälä wrote:
>> On Wed, Feb 01, 2017 at 01:41:08PM +0100, Maarten Lankhorst wrote:
>>> Op 31-01-17 om 20:13 schreef Uwe Kleine-König:
>>>> On Tue, Jan 31, 2017 at 10:03:26AM +0100, Maarten Lankhorst wrote:
>>>>> Op 31-01-17 om 09:09 schreef Uwe Kleine-König:
>>>>> Just curious, does this help?
>>>>>
>>>>> diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
>>>>> index ae2c0bb4b2e8..13de4c526ca6 100644
>>>>> --- a/drivers/gpu/drm/i915/intel_pm.c
>>>>> +++ b/drivers/gpu/drm/i915/intel_pm.c
>>>>> @@ -1838,7 +1838,7 @@ static uint32_t ilk_compute_cur_wm(const struct intel_crtc_state *cstate,
>>>>>  	 * this is necessary to avoid flickering.
>>>>>  	 */
>>>>>  	int cpp = 4;
>>>>> -	int width = pstate->base.visible ? pstate->base.crtc_w : 64;
>>>>> +	int width = 256;
>>>>>
>>>>>  	if (!cstate->base.active)
>>>>>  		return 0;
>>>>>
>>>> On a kernel with this patch applied I cannot reproduce the flickering. I
>>>> keep that kernel running but expect that it also fixes the crash.
>>>
>>> Ok that's good news.
>>>
>>> Maybe ville or matt can comment whether this patch is the right fix?
>>
>> Well, it's just extending the hack even further. The right fix would be
>> to fix the wm programming sequence to respect the frame boundaries
>> correctly (ie. my old vblank based wm stuff).
>
> so I wonder how this goes forward. The situation seems to be well
> understood, but other than testing patches I don't know what to do (and
> there is currently no patch to test).
>
> Best regards
> Uwe
>

The way I understand this is that no-one wants to restrict the 
capabilities exposed by the kernel and would like a proper fix for this. 
However, I agree with Uwe, given the low priority status of Gen5 (people 
would rather work on hw that is used by a lot of people), we should 
probably accept the patch proposed by Maarten as it fixes someone's 
workflow and does not regress anything meaningful.

The proper fix for watermark computation can be worked on as time 
permits, later on.

Again, I would like to thanks Uwe for pushing hard for this, we are 
definitely not active-enough on this issue, flashing screens should be a 
big NO-NO, yet we seem to be OK with it :s

Martin