[PATCH 2/2] drm/amdgpu: log the user queue object PRT mapping

[Christian König]

On 5/7/25 09:05, Liang, Prike wrote:
> [Public]
> 
>> -----Original Message-----
>> From: Koenig, Christian <Christian.Koenig at amd.com>
>> Sent: Tuesday, May 6, 2025 10:09 PM
>> To: Liang, Prike <Prike.Liang at amd.com>; amd-gfx at lists.freedesktop.org
>> Cc: Deucher, Alexander <Alexander.Deucher at amd.com>
>> Subject: Re: [PATCH 2/2] drm/amdgpu: log the user queue object PRT mapping
>>
>> On 5/6/25 15:34, Liang, Prike wrote:
>>> [Public]
>>>
>>> Any feedback?
>>
>>
>> Sorry totally missed that. We already have a trace point for all page table updates.
>>
>> If you want to only see the PRT updates you can potentially enable that as a filter.
>>
>> I can try to come up with the necessary commands for that if you want.
> Yes, we can check the PRT mappings by checking the trace point at amdgpu_vm_set_ptes() or amdgpu_vm_update_ptes(), but it requires further parsing the PTE flag and PTE update owner.
> Anyway, it's ok to confirm the user queue PRT mapping by checking the trace point, and I will drop the patch sets.

The trace points have build in filter functionality and you can also use the RMV tool to visualize what is going on in the address space: https://gpuopen.com/rmv/

Regards,
Christian.

> 
> Thanks,
> Prike
>>
>> Regards,
>> Christian.
>>
>>>
>>> Regards,
>>>       Prike
>>>
>>>> -----Original Message-----
>>>> From: Liang, Prike
>>>> Sent: Wednesday, April 30, 2025 10:51 AM
>>>> To: amd-gfx at lists.freedesktop.org
>>>> Cc: Deucher, Alexander <Alexander.Deucher at amd.com>; Koenig, Christian
>>>> <Christian.Koenig at amd.com>
>>>> Subject: RE: [PATCH 2/2] drm/amdgpu: log the user queue object PRT
>>>> mapping
>>>>
>>>> Soft ping for this series.
>>>>
>>>> Regards,
>>>>       Prike
>>>>
>>>>> -----Original Message-----
>>>>> From: Liang, Prike <Prike.Liang at amd.com>
>>>>> Sent: Monday, April 28, 2025 2:29 PM
>>>>> To: amd-gfx at lists.freedesktop.org
>>>>> Cc: Deucher, Alexander <Alexander.Deucher at amd.com>; Koenig,
>>>>> Christian <Christian.Koenig at amd.com>; Liang, Prike
>>>>> <Prike.Liang at amd.com>
>>>>> Subject: [PATCH 2/2] drm/amdgpu: log the user queue object PRT
>>>>> mapping
>>>>>
>>>>> This patch will log the user queue buffer PRT mapping, and this will
>>>>> help on checking the user queue PRT functional status.
>>>>>
>>>>> Signed-off-by: Prike Liang <Prike.Liang at amd.com>
>>>>> ---
>>>>>  drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c | 4 ++++
>>>>>  1 file changed, 4 insertions(+)
>>>>>
>>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
>>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
>>>>> index 30022123b0bf..7992a97f8745 100644
>>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
>>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
>>>>> @@ -903,6 +903,10 @@ int amdgpu_vm_ptes_update(struct
>>>>> amdgpu_vm_update_params *params,
>>>>>                                                 upd_flags,
>>>>>                                                 vm->task_info ?
>>>>> vm->task_info-
>>>>>> tgid : 0,
>>>>>
>>>>> vm->immediate.fence_context);
>>>>> +                   if (upd_flags | AMDGPU_PTE_PRT_FLAG(adev)) {
>>>>> +                           if (params->vm->is_userq_context)
>>>>> +                                   dev_dbg_ratelimited(adev->dev,
>>>>> + "The kgd
>>>>> user queue PRT map is functional\n");
>>>>> +                   }
>>>>>                     amdgpu_vm_pte_update_flags(params,
>>>>> to_amdgpu_bo_vm(pt),
>>>>>                                                cursor.level, pe_start, dst,
>>>>>                                                nptes, incr,
>>>>> upd_flags);
>>>>> --
>>>>> 2.34.1
>>>
>