1 files changed, 16 insertions, 7 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index d371000a5727..de5cb056c9ad 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -109,6 +109,7 @@ static int amdgpu_cs_p1_ib(struct amdgpu_cs_parser *p,
 		return r;
 
 	++(num_ibs[r]);
+	p->gang_leader_idx = r;
 	return 0;
 }
 
@@ -300,7 +301,7 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 		if (ret)
 			goto free_all_kdata;
 	}
-	p->gang_leader = p->jobs[p->gang_size - 1];
+	p->gang_leader = p->jobs[p->gang_leader_idx];
 
 	if (p->ctx->vram_lost_counter != p->gang_leader->vram_lost_counter) {
 		ret = -ECANCELED;
@@ -1195,16 +1196,18 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
 			return r;
 	}
 
-	for (i = 0; i < p->gang_size - 1; ++i) {
+	for (i = 0; i < p->gang_size; ++i) {
+		if (p->jobs[i] == leader)
+			continue;
+
 		r = amdgpu_sync_clone(&leader->sync, &p->jobs[i]->sync);
 		if (r)
 			return r;
 	}
 
-	r = amdgpu_ctx_wait_prev_fence(p->ctx, p->entities[p->gang_size - 1]);
+	r = amdgpu_ctx_wait_prev_fence(p->ctx, p->entities[p->gang_leader_idx]);
 	if (r && r != -ERESTARTSYS)
 		DRM_ERROR("amdgpu_ctx_wait_prev_fence failed.\n");
-
 	return r;
 }
 
@@ -1238,9 +1241,12 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	for (i = 0; i < p->gang_size; ++i)
 		drm_sched_job_arm(&p->jobs[i]->base);
 
-	for (i = 0; i < (p->gang_size - 1); ++i) {
+	for (i = 0; i < p->gang_size; ++i) {
 		struct dma_fence *fence;
 
+		if (p->jobs[i] == leader)
+			continue;
+
 		fence = &p->jobs[i]->base.s_fence->scheduled;
 		r = amdgpu_sync_fence(&leader->sync, fence);
 		if (r)
@@ -1276,7 +1282,10 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	list_for_each_entry(e, &p->validated, tv.head) {
 
 		/* Everybody except for the gang leader uses READ */
-		for (i = 0; i < (p->gang_size - 1); ++i) {
+		for (i = 0; i < p->gang_size; ++i) {
+			if (p->jobs[i] == leader)
+				continue;
+
 			dma_resv_add_fence(e->tv.bo->base.resv,
 					   &p->jobs[i]->base.s_fence->finished,
 					   DMA_RESV_USAGE_READ);
@@ -1286,7 +1295,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 		e->tv.num_shared = 0;
 	}
 
-	seq = amdgpu_ctx_add_fence(p->ctx, p->entities[p->gang_size - 1],
+	seq = amdgpu_ctx_add_fence(p->ctx, p->entities[p->gang_leader_idx],
 				   p->fence);
 	amdgpu_cs_post_dependencies(p);