Re: [PATCH] drm/xe: Use drm_sched_priority enum instead of 0

[Nirmoy Das <nirmoy.das@intel.com>]

On 4/12/2024 2:48 PM, Matthew Auld wrote:
> On 12/04/2024 13:03, Nirmoy Das wrote:
>> Hi Matt,
>>
>>
>> Shouldn't this be DRM_SCHED_PRIORITY_NORMAL ?
>
> Looks like this was already sent: 
> https://lore.kernel.org/intel-xe/ZgJhpTlz6OfqiaQg@DUT025-TGLU.fm.intel.com/
Ah I missed it. This need to be documented so someone else doesn't 
stumble upon again.
>
>>
>> Regards,
>>
>> Nirmoy
>>
>> On 4/12/2024 1:34 PM, Nirmoy Das wrote:
>>> Use enum drm_sched_priority, DRM_SCHED_PRIORITY_KERNEL
>>> instead of 0.
>>>
>>> Cc: Matthew Auld<matthew.auld@intel.com>
>>> Cc: Matthew Brost<matthew.brost@intel.com>
>>> Signed-off-by: Nirmoy Das<nirmoy.das@intel.com>
>>> ---
>>>   drivers/gpu/drm/xe/xe_execlist.c      | 3 ++-
>>>   drivers/gpu/drm/xe/xe_gpu_scheduler.h | 2 +-
>>>   2 files changed, 3 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/xe/xe_execlist.c 
>>> b/drivers/gpu/drm/xe/xe_execlist.c
>>> index dece2785933c..a8c83b36289f 100644
>>> --- a/drivers/gpu/drm/xe/xe_execlist.c
>>> +++ b/drivers/gpu/drm/xe/xe_execlist.c
>>> @@ -341,7 +341,8 @@ static int execlist_exec_queue_init(struct 
>>> xe_exec_queue *q)
>>>           goto err_free;
>>>         sched = &exl->sched;
>>> -    err = drm_sched_entity_init(&exl->entity, 0, &sched, 1, NULL);
>>> +    err = drm_sched_entity_init(&exl->entity, 
>>> DRM_SCHED_PRIORITY_KERNEL,
>>> +                    &sched, 1, NULL);
>>>       if (err)
>>>           goto err_sched;
>>>   diff --git a/drivers/gpu/drm/xe/xe_gpu_scheduler.h 
>>> b/drivers/gpu/drm/xe/xe_gpu_scheduler.h
>>> index 10c6bb9c9386..1f712f4fc76a 100644
>>> --- a/drivers/gpu/drm/xe/xe_gpu_scheduler.h
>>> +++ b/drivers/gpu/drm/xe/xe_gpu_scheduler.h
>>> @@ -63,7 +63,7 @@ static inline int
>>>   xe_sched_entity_init(struct xe_sched_entity *entity,
>>>                struct xe_gpu_scheduler *sched)
>>>   {
>>> -    return drm_sched_entity_init(entity, 0,
>>> +    return drm_sched_entity_init(entity, DRM_SCHED_PRIORITY_KERNEL,
>>>                        (struct drm_gpu_scheduler **)&sched,
>>>                        1, NULL);
>>>   }