On 2025-05-27 21:55, Lang Yu wrote:
> For simulation C models that don't run CP FW where adev->mes.sched_version
> is not populated correctly. This causes NULL dereference in
> amdgpu_amdkfd_free_gtt_mem(dev->adev, (void **)&pqn->q->wptr_bo_gart)
> and warning on unpinned BO in amdgpu_bo_gpu_offset(q->properties.wptr_bo).
>
> Compared with adding version check here and there,
> always map wptr BO to GART simplifies things.
>
> v2: Add NULL check in amdgpu_amdkfd_free_gtt_mem.(Philip)
>
> Signed-off-by: Lang Yu <lang...@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehl...@amd.com>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 3 +++
> .../amd/amdkfd/kfd_process_queue_manager.c | 23 ++++++++-----------
> 2 files changed, 13 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
> index 4cec3a873995..d8ac4b1051a8 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
> @@ -368,6 +368,9 @@ void amdgpu_amdkfd_free_gtt_mem(struct amdgpu_device
> *adev, void **mem_obj)
> {
> struct amdgpu_bo **bo = (struct amdgpu_bo **) mem_obj;
>
> + if (!bo || !*bo)
> + return;
> +
> (void)amdgpu_bo_reserve(*bo, true);
> amdgpu_bo_kunmap(*bo);
> amdgpu_bo_unpin(*bo);
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
> b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
> index 6d5fa57d4a23..c643e0ccec52 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
> @@ -279,20 +279,17 @@ static int init_user_queue(struct process_queue_manager
> *pqm,
> /* Starting with GFX11, wptr BOs must be mapped to GART for MES
> to determine work
> * on unmapped queues for usermode queue oversubscription (no
> aggregated doorbell)
> */
> - if (((dev->adev->mes.sched_version &
> AMDGPU_MES_API_VERSION_MASK)
> - >> AMDGPU_MES_API_VERSION_SHIFT) >= 2) {
> - if (dev->adev !=
> amdgpu_ttm_adev(q_properties->wptr_bo->tbo.bdev)) {
> - pr_err("Queue memory allocated to wrong
> device\n");
> - retval = -EINVAL;
> - goto free_gang_ctx_bo;
> - }
> + if (dev->adev !=
> amdgpu_ttm_adev(q_properties->wptr_bo->tbo.bdev)) {
> + pr_err("Queue memory allocated to wrong device\n");
> + retval = -EINVAL;
> + goto free_gang_ctx_bo;
> + }
>
> - retval =
> amdgpu_amdkfd_map_gtt_bo_to_gart(q_properties->wptr_bo,
> -
> &(*q)->wptr_bo_gart);
> - if (retval) {
> - pr_err("Failed to map wptr bo to GART\n");
> - goto free_gang_ctx_bo;
> - }
> + retval = amdgpu_amdkfd_map_gtt_bo_to_gart(q_properties->wptr_bo,
> + &(*q)->wptr_bo_gart);
> + if (retval) {
> + pr_err("Failed to map wptr bo to GART\n");
> + goto free_gang_ctx_bo;
> }
> }
>
