[PATCH v2] drm/amdgpu: skip the VMID 0 flush for VRAM clear-on-release

[PATCH v2] drm/amdgpu: skip the VMID 0 flush for VRAM clear-on-release

[Arunpravin Paneer Selvam]

Clear-on-release only runs on VRAM, which amdgpu_ttm_map_buffer() reaches
via its direct MC address without programming a GART window, yet the wipe
still forces a VMID 0 flush. On GFX11 (e.g. Navi33) that spurious SDMA
flush can wedge the engine; only flush when a GART window is actually used.

v2: Let amdgpu_ttm_map_buffer() return whether the VMID 0 flush is needed,
    and drive the clear and copy paths from that. (Christian)

Fixes: a68c7eaa7a8f ("drm/amdgpu: Enable clear page functionality")
Closes: https://gitlab.freedesktop.org/drm/amd/-/work_items/5413
Cc: stable@vger.kernel.org
Cc: Christian König <christian.koenig@amd.com>
Signed-off-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 6c07cee8e8777..e0bdd95c64de8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -190,6 +190,8 @@ amdgpu_ttm_job_submit(struct amdgpu_device *adev, struct amdgpu_ttm_buffer_entit
  * @tmz: if we should setup a TMZ enabled mapping
  * @size: in number of bytes to map, out number of bytes mapped
  * @addr: resulting address inside the MC address space
+ * @vm_needs_flush: out, set true if a GART window was programmed (VMID 0 flush
+ *		    needed) or false for a direct address; may be NULL
  *
  * Setup one of the GART windows to access a specific piece of memory or return
  * the physical address for local memory.
@@ -199,7 +201,8 @@ static int amdgpu_ttm_map_buffer(struct amdgpu_ttm_buffer_entity *entity,
 				 struct ttm_resource *mem,
 				 struct amdgpu_res_cursor *mm_cur,
 				 unsigned int window,
-				 bool tmz, uint64_t *size, uint64_t *addr)
+				 bool tmz, uint64_t *size, uint64_t *addr,
+				 bool *vm_needs_flush)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
 	unsigned int offset, num_pages, num_dw, num_bytes;
@@ -220,9 +223,14 @@ static int amdgpu_ttm_map_buffer(struct amdgpu_ttm_buffer_entity *entity,
 	if (!tmz && mem->start != AMDGPU_BO_INVALID_OFFSET) {
 		*addr = amdgpu_ttm_domain_start(adev, mem->mem_type) +
 			mm_cur->start;
+		if (vm_needs_flush)
+			*vm_needs_flush = false;
 		return 0;
 	}
 
+	/* A GART window is programmed below, so its VMID 0 TLB needs a flush */
+	if (vm_needs_flush)
+		*vm_needs_flush = true;
 
 	/*
 	 * If start begins at an offset inside the page, then adjust the size
@@ -322,6 +330,7 @@ static int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
 	while (src_mm.remaining) {
 		uint64_t from, to, cur_size, tiling_flags;
 		uint32_t num_type, data_format, max_com, write_compress_disable;
+		bool src_vm_flush, dst_vm_flush;
 		struct dma_fence *next;
 
 		/* Never copy more than 256MiB at once to avoid a timeout */
@@ -329,12 +338,12 @@ static int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
 
 		/* Map src to window 0 and dst to window 1. */
 		r = amdgpu_ttm_map_buffer(entity, src->bo, src->mem, &src_mm,
-					  0, tmz, &cur_size, &from);
+					  0, tmz, &cur_size, &from, &src_vm_flush);
 		if (r)
 			goto error;
 
 		r = amdgpu_ttm_map_buffer(entity, dst->bo, dst->mem, &dst_mm,
-					  1, tmz, &cur_size, &to);
+					  1, tmz, &cur_size, &to, &dst_vm_flush);
 		if (r)
 			goto error;
 
@@ -362,7 +371,7 @@ static int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
 		}
 
 		r = amdgpu_copy_buffer(adev, entity, from, to, cur_size, resv,
-				       &next, true, copy_flags);
+				       &next, src_vm_flush || dst_vm_flush, copy_flags);
 		if (r)
 			goto error;
 
@@ -2578,6 +2587,7 @@ int amdgpu_ttm_clear_buffer(struct amdgpu_ttm_buffer_entity *entity,
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
 	struct dma_fence *fence = NULL;
 	struct amdgpu_res_cursor dst;
+	bool vm_needs_flush = false;
 	int r;
 
 	if (!entity)
@@ -2599,13 +2609,13 @@ int amdgpu_ttm_clear_buffer(struct amdgpu_ttm_buffer_entity *entity,
 		cur_size = min(dst.size, 256ULL << 20);
 
 		r = amdgpu_ttm_map_buffer(entity, &bo->tbo, bo->tbo.resource, &dst,
-					  0, false, &cur_size, &to);
+					  0, false, &cur_size, &to, &vm_needs_flush);
 		if (r)
 			goto error;
 
 		r = amdgpu_ttm_fill_mem(adev, entity,
 					0, to, cur_size, resv,
-					&next, true, k_job_id);
+					&next, vm_needs_flush, k_job_id);
 		if (r)
 			goto error;
 
-- 
2.34.1

Re: [PATCH v2] drm/amdgpu: skip the VMID 0 flush for VRAM clear-on-release

[Timur Kristóf]

On Monday, August 31, 2026 2:28:41 PM Central European Summer Time Arunpravin 
Paneer Selvam wrote:
> Clear-on-release only runs on VRAM, which amdgpu_ttm_map_buffer() reaches
> via its direct MC address without programming a GART window, yet the wipe
> still forces a VMID 0 flush. On GFX11 (e.g. Navi33) that spurious SDMA
> flush can wedge the engine; only flush when a GART window is actually used.
> 
> v2: Let amdgpu_ttm_map_buffer() return whether the VMID 0 flush is needed,
>     and drive the clear and copy paths from that. (Christian)
> 
> Fixes: a68c7eaa7a8f ("drm/amdgpu: Enable clear page functionality")
> Closes: https://gitlab.freedesktop.org/drm/amd/-/work_items/5413
> Cc: stable@vger.kernel.org
> Cc: Christian König <christian.koenig@amd.com>
> Signed-off-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>

Looks good, thank you!

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>

> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 22 ++++++++++++++++------
>  1 file changed, 16 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index
> 6c07cee8e8777..e0bdd95c64de8 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> @@ -190,6 +190,8 @@ amdgpu_ttm_job_submit(struct amdgpu_device *adev, struct
> amdgpu_ttm_buffer_entit * @tmz: if we should setup a TMZ enabled mapping
>   * @size: in number of bytes to map, out number of bytes mapped
>   * @addr: resulting address inside the MC address space
> + * @vm_needs_flush: out, set true if a GART window was programmed (VMID 0
> flush + *		    needed) or false for a direct address; may be NULL
>   *
>   * Setup one of the GART windows to access a specific piece of memory or
> return * the physical address for local memory.
> @@ -199,7 +201,8 @@ static int amdgpu_ttm_map_buffer(struct
> amdgpu_ttm_buffer_entity *entity, struct ttm_resource *mem,
>  				 struct amdgpu_res_cursor 
*mm_cur,
>  				 unsigned int window,
> -				 bool tmz, uint64_t *size, 
uint64_t *addr)
> +				 bool tmz, uint64_t *size, 
uint64_t *addr,
> +				 bool *vm_needs_flush)
>  {
>  	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
>  	unsigned int offset, num_pages, num_dw, num_bytes;
> @@ -220,9 +223,14 @@ static int amdgpu_ttm_map_buffer(struct
> amdgpu_ttm_buffer_entity *entity, if (!tmz && mem->start !=
> AMDGPU_BO_INVALID_OFFSET) {
>  		*addr = amdgpu_ttm_domain_start(adev, mem->mem_type) +
>  			mm_cur->start;
> +		if (vm_needs_flush)
> +			*vm_needs_flush = false;
>  		return 0;
>  	}
> 
> +	/* A GART window is programmed below, so its VMID 0 TLB needs a 
flush */
> +	if (vm_needs_flush)
> +		*vm_needs_flush = true;
> 
>  	/*
>  	 * If start begins at an offset inside the page, then adjust the 
size
> @@ -322,6 +330,7 @@ static int amdgpu_ttm_copy_mem_to_mem(struct
> amdgpu_device *adev, while (src_mm.remaining) {
>  		uint64_t from, to, cur_size, tiling_flags;
>  		uint32_t num_type, data_format, max_com, 
write_compress_disable;
> +		bool src_vm_flush, dst_vm_flush;
>  		struct dma_fence *next;
> 
>  		/* Never copy more than 256MiB at once to avoid a 
timeout */
> @@ -329,12 +338,12 @@ static int amdgpu_ttm_copy_mem_to_mem(struct
> amdgpu_device *adev,
> 
>  		/* Map src to window 0 and dst to window 1. */
>  		r = amdgpu_ttm_map_buffer(entity, src->bo, src->mem, 
&src_mm,
> -					  0, tmz, &cur_size, 
&from);
> +					  0, tmz, &cur_size, 
&from, &src_vm_flush);
>  		if (r)
>  			goto error;
> 
>  		r = amdgpu_ttm_map_buffer(entity, dst->bo, dst->mem, 
&dst_mm,
> -					  1, tmz, &cur_size, 
&to);
> +					  1, tmz, &cur_size, 
&to, &dst_vm_flush);
>  		if (r)
>  			goto error;
> 
> @@ -362,7 +371,7 @@ static int amdgpu_ttm_copy_mem_to_mem(struct
> amdgpu_device *adev, }
> 
>  		r = amdgpu_copy_buffer(adev, entity, from, to, cur_size, 
resv,
> -				       &next, true, copy_flags);
> +				       &next, src_vm_flush || 
dst_vm_flush, copy_flags);
>  		if (r)
>  			goto error;
> 
> @@ -2578,6 +2587,7 @@ int amdgpu_ttm_clear_buffer(struct
> amdgpu_ttm_buffer_entity *entity, struct amdgpu_device *adev =
> amdgpu_ttm_adev(bo->tbo.bdev);
>  	struct dma_fence *fence = NULL;
>  	struct amdgpu_res_cursor dst;
> +	bool vm_needs_flush = false;
>  	int r;
> 
>  	if (!entity)
> @@ -2599,13 +2609,13 @@ int amdgpu_ttm_clear_buffer(struct
> amdgpu_ttm_buffer_entity *entity, cur_size = min(dst.size, 256ULL << 20);
> 
>  		r = amdgpu_ttm_map_buffer(entity, &bo->tbo, bo-
>tbo.resource, &dst,
> -					  0, false, &cur_size, 
&to);
> +					  0, false, &cur_size, 
&to, &vm_needs_flush);
>  		if (r)
>  			goto error;
> 
>  		r = amdgpu_ttm_fill_mem(adev, entity,
>  					0, to, cur_size, resv,
> -					&next, true, 
k_job_id);
> +					&next, vm_needs_flush, 
k_job_id);
>  		if (r)
>  			goto error;

Re: [PATCH v2] drm/amdgpu: skip the VMID 0 flush for VRAM clear-on-release

[Christian König]

On 8/31/26 14:28, Arunpravin Paneer Selvam wrote:
> Clear-on-release only runs on VRAM, which amdgpu_ttm_map_buffer() reaches
> via its direct MC address without programming a GART window, yet the wipe
> still forces a VMID 0 flush. On GFX11 (e.g. Navi33) that spurious SDMA
> flush can wedge the engine; only flush when a GART window is actually used.
> 
> v2: Let amdgpu_ttm_map_buffer() return whether the VMID 0 flush is needed,
>     and drive the clear and copy paths from that. (Christian)
> 
> Fixes: a68c7eaa7a8f ("drm/amdgpu: Enable clear page functionality")
> Closes: https://gitlab.freedesktop.org/drm/amd/-/work_items/5413
> Cc: stable@vger.kernel.org
> Cc: Christian König <christian.koenig@amd.com>
> Signed-off-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 22 ++++++++++++++++------
>  1 file changed, 16 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> index 6c07cee8e8777..e0bdd95c64de8 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> @@ -190,6 +190,8 @@ amdgpu_ttm_job_submit(struct amdgpu_device *adev, struct amdgpu_ttm_buffer_entit
>   * @tmz: if we should setup a TMZ enabled mapping
>   * @size: in number of bytes to map, out number of bytes mapped
>   * @addr: resulting address inside the MC address space
> + * @vm_needs_flush: out, set true if a GART window was programmed (VMID 0 flush
> + *		    needed) or false for a direct address; may be NULL

Does that really needs to be NULL for any use case? I would just make it mandatory for simplicity.

Apart from that looks good to me,
Christian.

>   *
>   * Setup one of the GART windows to access a specific piece of memory or return
>   * the physical address for local memory.
> @@ -199,7 +201,8 @@ static int amdgpu_ttm_map_buffer(struct amdgpu_ttm_buffer_entity *entity,
>  				 struct ttm_resource *mem,
>  				 struct amdgpu_res_cursor *mm_cur,
>  				 unsigned int window,
> -				 bool tmz, uint64_t *size, uint64_t *addr)
> +				 bool tmz, uint64_t *size, uint64_t *addr,
> +				 bool *vm_needs_flush)
>  {
>  	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
>  	unsigned int offset, num_pages, num_dw, num_bytes;
> @@ -220,9 +223,14 @@ static int amdgpu_ttm_map_buffer(struct amdgpu_ttm_buffer_entity *entity,
>  	if (!tmz && mem->start != AMDGPU_BO_INVALID_OFFSET) {
>  		*addr = amdgpu_ttm_domain_start(adev, mem->mem_type) +
>  			mm_cur->start;
> +		if (vm_needs_flush)
> +			*vm_needs_flush = false;
>  		return 0;
>  	}
>  
> +	/* A GART window is programmed below, so its VMID 0 TLB needs a flush */
> +	if (vm_needs_flush)
> +		*vm_needs_flush = true;
>  
>  	/*
>  	 * If start begins at an offset inside the page, then adjust the size
> @@ -322,6 +330,7 @@ static int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
>  	while (src_mm.remaining) {
>  		uint64_t from, to, cur_size, tiling_flags;
>  		uint32_t num_type, data_format, max_com, write_compress_disable;
> +		bool src_vm_flush, dst_vm_flush;
>  		struct dma_fence *next;
>  
>  		/* Never copy more than 256MiB at once to avoid a timeout */
> @@ -329,12 +338,12 @@ static int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
>  
>  		/* Map src to window 0 and dst to window 1. */
>  		r = amdgpu_ttm_map_buffer(entity, src->bo, src->mem, &src_mm,
> -					  0, tmz, &cur_size, &from);
> +					  0, tmz, &cur_size, &from, &src_vm_flush);
>  		if (r)
>  			goto error;
>  
>  		r = amdgpu_ttm_map_buffer(entity, dst->bo, dst->mem, &dst_mm,
> -					  1, tmz, &cur_size, &to);
> +					  1, tmz, &cur_size, &to, &dst_vm_flush);
>  		if (r)
>  			goto error;
>  
> @@ -362,7 +371,7 @@ static int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
>  		}
>  
>  		r = amdgpu_copy_buffer(adev, entity, from, to, cur_size, resv,
> -				       &next, true, copy_flags);
> +				       &next, src_vm_flush || dst_vm_flush, copy_flags);
>  		if (r)
>  			goto error;
>  
> @@ -2578,6 +2587,7 @@ int amdgpu_ttm_clear_buffer(struct amdgpu_ttm_buffer_entity *entity,
>  	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
>  	struct dma_fence *fence = NULL;
>  	struct amdgpu_res_cursor dst;
> +	bool vm_needs_flush = false;
>  	int r;
>  
>  	if (!entity)
> @@ -2599,13 +2609,13 @@ int amdgpu_ttm_clear_buffer(struct amdgpu_ttm_buffer_entity *entity,
>  		cur_size = min(dst.size, 256ULL << 20);
>  
>  		r = amdgpu_ttm_map_buffer(entity, &bo->tbo, bo->tbo.resource, &dst,
> -					  0, false, &cur_size, &to);
> +					  0, false, &cur_size, &to, &vm_needs_flush);
>  		if (r)
>  			goto error;
>  
>  		r = amdgpu_ttm_fill_mem(adev, entity,
>  					0, to, cur_size, resv,
> -					&next, true, k_job_id);
> +					&next, vm_needs_flush, k_job_id);
>  		if (r)
>  			goto error;
>