AMD-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH v3] drm/amdgpu: Add AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES
@ 2019-02-01 21:41 Andrey Grodzovsky
       [not found] ` <1549057286-31201-1-git-send-email-andrey.grodzovsky-5C7GfCeVMHo@public.gmane.org>
  0 siblings, 1 reply; 3+ messages in thread
From: Andrey Grodzovsky @ 2019-02-01 21:41 UTC (permalink / raw)
  To: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW
  Cc: Andrey Grodzovsky, Marek.Olsak-5C7GfCeVMHo,
	christian.koenig-5C7GfCeVMHo

New chunk for dependency on start of job's execution instead on
the end. This is used for GPU deadlock prevention when
userspace uses mid-IB fences to wait for mid-IB work on other rings.

v2: Fix typo in AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES
v3: Bump KMS version

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Suggested-by: Christian Koenig <Christian.Koenig@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c  | 12 +++++++++++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c |  3 ++-
 include/uapi/drm/amdgpu_drm.h           |  1 +
 3 files changed, 14 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 1c49b82..3f21eca 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -214,6 +214,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs
 		case AMDGPU_CHUNK_ID_DEPENDENCIES:
 		case AMDGPU_CHUNK_ID_SYNCOBJ_IN:
 		case AMDGPU_CHUNK_ID_SYNCOBJ_OUT:
+		case AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES:
 			break;
 
 		default:
@@ -1090,6 +1091,14 @@ static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
 
 		fence = amdgpu_ctx_get_fence(ctx, entity,
 					     deps[i].handle);
+
+		if (chunk->chunk_id == AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES) {
+			struct drm_sched_fence *s_fence = to_drm_sched_fence(fence);
+
+			dma_fence_put(fence);
+			fence = dma_fence_get(&s_fence->scheduled);
+		}
+
 		if (IS_ERR(fence)) {
 			r = PTR_ERR(fence);
 			amdgpu_ctx_put(ctx);
@@ -1177,7 +1186,8 @@ static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
 
 		chunk = &p->chunks[i];
 
-		if (chunk->chunk_id == AMDGPU_CHUNK_ID_DEPENDENCIES) {
+		if (chunk->chunk_id == AMDGPU_CHUNK_ID_DEPENDENCIES ||
+		    chunk->chunk_id == AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES) {
 			r = amdgpu_cs_process_fence_dep(p, chunk);
 			if (r)
 				return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index c806f98..1158a6f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -71,9 +71,10 @@
  * - 3.25.0 - Add support for sensor query info (stable pstate sclk/mclk).
  * - 3.26.0 - GFX9: Process AMDGPU_IB_FLAG_TC_WB_NOT_INVALIDATE.
  * - 3.27.0 - Add new chunk to to AMDGPU_CS to enable BO_LIST creation.
+ * - 3.28.0 - Add AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES
  */
 #define KMS_DRIVER_MAJOR	3
-#define KMS_DRIVER_MINOR	27
+#define KMS_DRIVER_MINOR	28
 #define KMS_DRIVER_PATCHLEVEL	0
 
 int amdgpu_vram_limit = 0;
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
index faaad04..43d03a2 100644
--- a/include/uapi/drm/amdgpu_drm.h
+++ b/include/uapi/drm/amdgpu_drm.h
@@ -526,6 +526,7 @@ struct drm_amdgpu_gem_va {
 #define AMDGPU_CHUNK_ID_SYNCOBJ_IN      0x04
 #define AMDGPU_CHUNK_ID_SYNCOBJ_OUT     0x05
 #define AMDGPU_CHUNK_ID_BO_HANDLES      0x06
+#define AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES	0x07
 
 struct drm_amdgpu_cs_chunk {
 	__u32		chunk_id;
-- 
2.7.4

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH v3] drm/amdgpu: Add AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES
       [not found] ` <1549057286-31201-1-git-send-email-andrey.grodzovsky-5C7GfCeVMHo@public.gmane.org>
@ 2019-02-03 15:51   ` Koenig, Christian
       [not found]     ` <2d32f38c-5ea7-d967-2a5d-bdc11fbb5e7b-5C7GfCeVMHo@public.gmane.org>
  0 siblings, 1 reply; 3+ messages in thread
From: Koenig, Christian @ 2019-02-03 15:51 UTC (permalink / raw)
  To: Grodzovsky, Andrey,
	amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
  Cc: Olsak, Marek

Am 01.02.19 um 22:41 schrieb Andrey Grodzovsky:
> New chunk for dependency on start of job's execution instead on
> the end. This is used for GPU deadlock prevention when
> userspace uses mid-IB fences to wait for mid-IB work on other rings.
>
> v2: Fix typo in AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES
> v3: Bump KMS version
>
> Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
> Suggested-by: Christian Koenig <Christian.Koenig@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c  | 12 +++++++++++-
>   drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c |  3 ++-
>   include/uapi/drm/amdgpu_drm.h           |  1 +
>   3 files changed, 14 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> index 1c49b82..3f21eca 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> @@ -214,6 +214,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs
>   		case AMDGPU_CHUNK_ID_DEPENDENCIES:
>   		case AMDGPU_CHUNK_ID_SYNCOBJ_IN:
>   		case AMDGPU_CHUNK_ID_SYNCOBJ_OUT:
> +		case AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES:
>   			break;
>   
>   		default:
> @@ -1090,6 +1091,14 @@ static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
>   
>   		fence = amdgpu_ctx_get_fence(ctx, entity,
>   					     deps[i].handle);
> +
> +		if (chunk->chunk_id == AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES) {
> +			struct drm_sched_fence *s_fence = to_drm_sched_fence(fence);
> +
> +			dma_fence_put(fence);
> +			fence = dma_fence_get(&s_fence->scheduled);

You need to change the order here or otherwise the dma_fence_put could 
destroy the fence we want.

Apart from that looks really good to me,
Christian.

> +		}
> +
>   		if (IS_ERR(fence)) {
>   			r = PTR_ERR(fence);
>   			amdgpu_ctx_put(ctx);
> @@ -1177,7 +1186,8 @@ static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
>   
>   		chunk = &p->chunks[i];
>   
> -		if (chunk->chunk_id == AMDGPU_CHUNK_ID_DEPENDENCIES) {
> +		if (chunk->chunk_id == AMDGPU_CHUNK_ID_DEPENDENCIES ||
> +		    chunk->chunk_id == AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES) {
>   			r = amdgpu_cs_process_fence_dep(p, chunk);
>   			if (r)
>   				return r;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> index c806f98..1158a6f 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> @@ -71,9 +71,10 @@
>    * - 3.25.0 - Add support for sensor query info (stable pstate sclk/mclk).
>    * - 3.26.0 - GFX9: Process AMDGPU_IB_FLAG_TC_WB_NOT_INVALIDATE.
>    * - 3.27.0 - Add new chunk to to AMDGPU_CS to enable BO_LIST creation.
> + * - 3.28.0 - Add AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES
>    */
>   #define KMS_DRIVER_MAJOR	3
> -#define KMS_DRIVER_MINOR	27
> +#define KMS_DRIVER_MINOR	28
>   #define KMS_DRIVER_PATCHLEVEL	0
>   
>   int amdgpu_vram_limit = 0;
> diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
> index faaad04..43d03a2 100644
> --- a/include/uapi/drm/amdgpu_drm.h
> +++ b/include/uapi/drm/amdgpu_drm.h
> @@ -526,6 +526,7 @@ struct drm_amdgpu_gem_va {
>   #define AMDGPU_CHUNK_ID_SYNCOBJ_IN      0x04
>   #define AMDGPU_CHUNK_ID_SYNCOBJ_OUT     0x05
>   #define AMDGPU_CHUNK_ID_BO_HANDLES      0x06
> +#define AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES	0x07
>   
>   struct drm_amdgpu_cs_chunk {
>   	__u32		chunk_id;

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH v3] drm/amdgpu: Add AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES
       [not found]     ` <2d32f38c-5ea7-d967-2a5d-bdc11fbb5e7b-5C7GfCeVMHo@public.gmane.org>
@ 2019-02-04 15:45       ` Grodzovsky, Andrey
  0 siblings, 0 replies; 3+ messages in thread
From: Grodzovsky, Andrey @ 2019-02-04 15:45 UTC (permalink / raw)
  To: Koenig, Christian,
	amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
  Cc: Olsak, Marek


[-- Attachment #1.1: Type: text/plain, Size: 775 bytes --]

Thanks, I will update, add your RB and push.

Andrey

On 2/3/19 10:51 AM, Koenig, Christian wrote:

@@ -1090,6 +1091,14 @@ static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,

                fence = amdgpu_ctx_get_fence(ctx, entity,
                                             deps[i].handle);
+
+               if (chunk->chunk_id == AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES) {
+                       struct drm_sched_fence *s_fence = to_drm_sched_fence(fence);
+
+                       dma_fence_put(fence);
+                       fence = dma_fence_get(&s_fence->scheduled);


You need to change the order here or otherwise the dma_fence_put could
destroy the fence we want.

Apart from that looks really good to me,
Christian.



[-- Attachment #1.2: Type: text/html, Size: 1148 bytes --]

[-- Attachment #2: Type: text/plain, Size: 154 bytes --]

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2019-02-04 15:45 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2019-02-01 21:41 [PATCH v3] drm/amdgpu: Add AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES Andrey Grodzovsky
     [not found] ` <1549057286-31201-1-git-send-email-andrey.grodzovsky-5C7GfCeVMHo@public.gmane.org>
2019-02-03 15:51   ` Koenig, Christian
     [not found]     ` <2d32f38c-5ea7-d967-2a5d-bdc11fbb5e7b-5C7GfCeVMHo@public.gmane.org>
2019-02-04 15:45       ` Grodzovsky, Andrey

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox