[10/10] amdgpu: list each entry safely for sw semaphore when submit ib

Commit Message

Marek Olšák Jan. 12, 2016, 9:23 p.m. UTC

From: Junwei Zhang <[e-mail]>

Signed-off-by: Junwei Zhang <[e-mail]>
Reviewed-by: Michel Dänzer <[e-mail]>
Reviewed-by: David Zhou <[e-mail]>
Reviewed-by: Christian König <[e-mail]>
---
 amdgpu/amdgpu_cs.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Patch

diff --git a/amdgpu/amdgpu_cs.c b/amdgpu/amdgpu_cs.c
index 5c7a3a3..82fa805 100644
--- a/amdgpu/amdgpu_cs.c
+++ b/amdgpu/amdgpu_cs.c
@@ -179,7 +179,7 @@  static int amdgpu_cs_submit_one(amdgpu_context_handle context,
 	struct drm_amdgpu_cs_chunk_dep *dependencies = NULL;
 	struct drm_amdgpu_cs_chunk_dep *sem_dependencies = NULL;
 	struct list_head *sem_list;
-	amdgpu_semaphore_handle sem;
+	amdgpu_semaphore_handle sem, tmp;
 	uint32_t i, size, sem_count = 0;
 	bool user_fence;
 	int r = 0;
@@ -282,7 +282,7 @@  static int amdgpu_cs_submit_one(amdgpu_context_handle context,
 			goto error_unlock;
 		}
 		sem_count = 0;
-		LIST_FOR_EACH_ENTRY(sem, sem_list, list) {
+		LIST_FOR_EACH_ENTRY_SAFE(sem, tmp, sem_list, list) {
 			struct amdgpu_cs_fence *info = &sem->signal_fence;
 			struct drm_amdgpu_cs_chunk_dep *dep = &sem_dependencies[sem_count++];
 			dep->ip_type = info->ip_type;