drm/amdgpu: Add AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES

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
v4: put old fence AFTER acquiring the scheduled fence.

Signed-off-by: Andrey Grodzovsky <[e-mail]>
Suggested-by: Christian Koenig <[e-mail]>
Reviewed-by: Christian König <[e-mail]>
Signed-off-by: Alex Deucher <[e-mail]>

drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c

···
 		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:
···
 
 		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);
+			struct dma_fence *old = fence;
+
+			fence = dma_fence_get(&s_fence->scheduled);
+			dma_fence_put(old);
+		}
+
 		if (IS_ERR(fence)) {
 			r = PTR_ERR(fence);
 			amdgpu_ctx_put(ctx);
···
 
 		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;

drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c

···
  * - 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;

include/uapi/drm/amdgpu_drm.h

···
 #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;