Re: [PATCH i-g-t v2 1/2] drm-uapi/amdgpu: Add queue priority and secure flags definitions

["Christian König" <christian.koenig@amd.com>]

Am 17.04.25 um 18:37 schrieb Kamil Konieczny:
> Hi Jesse.zhang,
> On 2025-04-17 at 16:52:46 +0800, Jesse.zhang@amd.com wrote:
>> Add UAPI definitions for queue priority levels (NORMAL_LOW, LOW, etc.)
>> and secure queue flag in amdgpu_drm.h. This matches the libdrm changes
>> in commit fdf384d4b546 ("amdgpu: add priority and secure flags for user queues").
> Please write from what tree comes this commit or
> give lore.kernel.org link.
>
> I tried to find out this on current drm-tip and failed,
> am I missing something?

This is a patch for the IGT test cases and not the Linux kernel.

It has i-g-t in the subject line and is send out on the igt-dev mailing list.

Regards,
Christian.

>
> Regards,
> Kamil
>
>> v2: In general, UAPI changes should be part of a different change and only uapi should be part of it (Sunil)
>>
>> Signed-off-by: Jesse.Zhang <Jesse.zhang@amd.com>
>> ---
>>  include/drm-uapi/amdgpu_drm.h | 12 +++++++++++-
>>  1 file changed, 11 insertions(+), 1 deletion(-)
>>
>> diff --git a/include/drm-uapi/amdgpu_drm.h b/include/drm-uapi/amdgpu_drm.h
>> index 8191d0bd0..3f4813879 100644
>> --- a/include/drm-uapi/amdgpu_drm.h
>> +++ b/include/drm-uapi/amdgpu_drm.h
>> @@ -329,6 +329,16 @@ union drm_amdgpu_ctx {
>>  #define AMDGPU_USERQ_OP_CREATE	1
>>  #define AMDGPU_USERQ_OP_FREE	2
>>  
>> +/* queue priority levels */
>> +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_MASK  0x3
>> +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_SHIFT 0
>> +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_NORMAL_LOW 0
>> +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_LOW 1
>> +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_NORMAL_HIGH 2
>> +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_HIGH 3 /* admin only */
>> +/* for queues that need access to protected content */
>> +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE  (1 << 2)
>> +
>>  /*
>>   * This structure is a container to pass input configuration
>>   * info for all supported userqueue related operations.
>> @@ -355,7 +365,7 @@ struct drm_amdgpu_userq_in {
>>  	 * and doorbell_offset in the doorbell bo.
>>  	 */
>>  	__u32   doorbell_offset;
>> -	__u32 _pad;
>> +	__u32 flags;
>>  	/**
>>  	 * @queue_va: Virtual address of the GPU memory which holds the queue
>>  	 * object. The queue holds the workload packets.
>> -- 
>> 2.25.1
>>