[PATCH] drm/amdgpu: do not modify ring before doing ring validation

Christian König christian.koenig at amd.com
Tue Feb 25 11:08:32 UTC 2020


Am 25.02.20 um 12:03 schrieb Nirmoy:
>
> On 2/25/20 11:52 AM, Christian König wrote:
>> Am 25.02.20 um 11:39 schrieb Nirmoy Das:
>>> changing ring->sched.ready should be done only if the ring is 
>>> initialized
>>
>> I don't think that this makes much difference. When the rings are 
>> freed the hardware and software stack needs to be disabled quite some 
>> time before.
>
> Yes it doesn't make any difference. I wanted amdgpu_ring_fini() to 
> look bit cleaner.

Ok, you should note that in the subject and commit message. Something 
like "cleanup amdgpu_ring_fini" / "Check the prerequisites before the 
actual operation".

Otherwise if that is not clearly state somebody might think that this is 
a bug fix and pick it up for backporting and we probably don't want that.

Regards,
Christian.

>
>
> Regards,
>
> Nirmoy
>
>>
>> Christian.
>>
>>>
>>> Signed-off-by: Nirmoy Das <nirmoy.das at amd.com>
>>> ---
>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c | 3 ++-
>>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c 
>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
>>> index 539be138260e..18e11b0fdc3e 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
>>> @@ -344,12 +344,13 @@ int amdgpu_ring_init(struct amdgpu_device 
>>> *adev, struct amdgpu_ring *ring,
>>>    */
>>>   void amdgpu_ring_fini(struct amdgpu_ring *ring)
>>>   {
>>> -    ring->sched.ready = false;
>>>         /* Not to finish a ring which is not initialized */
>>>       if (!(ring->adev) || !(ring->adev->rings[ring->idx]))
>>>           return;
>>>   +    ring->sched.ready = false;
>>> +
>>>       amdgpu_device_wb_free(ring->adev, ring->rptr_offs);
>>>       amdgpu_device_wb_free(ring->adev, ring->wptr_offs);
>>



More information about the amd-gfx mailing list