Re: [PATCH 1/2] drm/amdgpu: store userq_managers in a list in adev

["Paneer Selvam, Arunpravin" <arunpravin.paneerselvam@amd.com>]

Hi Alex,

I requested the tester to reproduce the suspend/resume hang issue with 
this patch series.
I will update as soon as I get the results.

Thanks,
Arun.

On 3/20/2025 10:22 PM, Alex Deucher wrote:
> So we can iterate across them when we need to manage
> all user queues.
>
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu.h           |  3 +++
>   drivers/gpu/drm/amd/amdgpu/amdgpu_device.c    |  3 +++
>   drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c | 15 ++++++++++++++-
>   drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h |  1 +
>   4 files changed, 21 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> index 4f770a362048a..28cfa600b798f 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> @@ -1228,6 +1228,9 @@ struct amdgpu_device {
>   	 * in KFD: VRAM or GTT.
>   	 */
>   	bool                            apu_prefer_gtt;
> +
> +	struct list_head		userq_mgr_list;
> +	struct mutex                    userq_mutex;
>   };
>   
>   static inline uint32_t amdgpu_ip_version(const struct amdgpu_device *adev,
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index 0396ac30c2a4f..526c5aa32825a 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -4299,6 +4299,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
>   	mutex_init(&adev->gfx.kfd_sch_mutex);
>   	mutex_init(&adev->gfx.workload_profile_mutex);
>   	mutex_init(&adev->vcn.workload_profile_mutex);
> +	mutex_init(&adev->userq_mutex);
>   
>   	amdgpu_device_init_apu_flags(adev);
>   
> @@ -4326,6 +4327,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
>   
>   	INIT_LIST_HEAD(&adev->pm.od_kobj_list);
>   
> +	INIT_LIST_HEAD(&adev->userq_mgr_list);
> +
>   	INIT_DELAYED_WORK(&adev->delayed_init_work,
>   			  amdgpu_device_delayed_init_work_handler);
>   	INIT_DELAYED_WORK(&adev->gfx.gfx_off_delay_work,
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
> index a02614cbda36e..b89bfad52abd5 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
> @@ -365,6 +365,9 @@ amdgpu_userqueue_create(struct drm_file *filp, union drm_amdgpu_userq *args)
>   		goto unlock;
>   	}
>   	args->out.queue_id = qid;
> +	mutex_lock(&adev->userq_mutex);
> +	list_add(&uq_mgr->list, &adev->userq_mgr_list);
> +	mutex_unlock(&adev->userq_mutex);
>   
>   unlock:
>   	mutex_unlock(&uq_mgr->userq_mutex);
> @@ -661,14 +664,24 @@ int amdgpu_userq_mgr_init(struct amdgpu_userq_mgr *userq_mgr, struct amdgpu_devi
>   
>   void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr)
>   {
> -	uint32_t queue_id;
> +	struct amdgpu_device *adev = userq_mgr->adev;
>   	struct amdgpu_usermode_queue *queue;
> +	struct amdgpu_userq_mgr *uqm, *tmp;
> +	uint32_t queue_id;
>   
>   	cancel_delayed_work(&userq_mgr->resume_work);
>   
>   	mutex_lock(&userq_mgr->userq_mutex);
>   	idr_for_each_entry(&userq_mgr->userq_idr, queue, queue_id)
>   		amdgpu_userqueue_cleanup(userq_mgr, queue, queue_id);
> +	mutex_lock(&adev->userq_mutex);
> +	list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) {
> +		if (uqm == userq_mgr) {
> +			list_del(&uqm->list);
> +			break;
> +		}
> +	}
> +	mutex_unlock(&adev->userq_mutex);
>   	idr_destroy(&userq_mgr->userq_idr);
>   	mutex_unlock(&userq_mgr->userq_mutex);
>   	mutex_destroy(&userq_mgr->userq_mutex);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h
> index 0f358f77f2d9b..ec1a4ca6f6321 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h
> @@ -76,6 +76,7 @@ struct amdgpu_userq_mgr {
>   	struct mutex			userq_mutex;
>   	struct amdgpu_device		*adev;
>   	struct delayed_work		resume_work;
> +	struct list_head		list;
>   };
>   
>   struct amdgpu_db_info {