drm/nouveau/ttm: explicity sync with kernel channel before moving buffer
The GEM code handles this currently, but that'll be removed.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c
index 7a52fb8..03e9a31 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo.c
@@ -560,28 +560,6 @@
}
-/* GPU-assisted copy using NV_MEMORY_TO_MEMORY_FORMAT, can access
- * TTM_PL_{VRAM,TT} directly.
- */
-
-static int
-nouveau_bo_move_accel_cleanup(struct nouveau_channel *chan,
- struct nouveau_bo *nvbo, bool evict,
- bool no_wait_gpu, struct ttm_mem_reg *new_mem)
-{
- struct nouveau_fence *fence = NULL;
- int ret;
-
- ret = nouveau_fence_new(chan, false, &fence);
- if (ret)
- return ret;
-
- ret = ttm_bo_move_accel_cleanup(&nvbo->bo, fence, evict,
- no_wait_gpu, new_mem);
- nouveau_fence_unref(&fence);
- return ret;
-}
-
static int
nve0_bo_move_init(struct nouveau_channel *chan, u32 handle)
{
@@ -967,6 +945,7 @@
{
struct nouveau_drm *drm = nouveau_bdev(bo->bdev);
struct nouveau_channel *chan = drm->ttm.chan;
+ struct nouveau_fence *fence;
int ret;
/* create temporary vmas for the transfer and attach them to the
@@ -980,14 +959,20 @@
}
mutex_lock_nested(&chan->cli->mutex, SINGLE_DEPTH_NESTING);
-
- ret = drm->ttm.move(chan, bo, &bo->mem, new_mem);
+ ret = nouveau_fence_sync(bo->sync_obj, chan);
if (ret == 0) {
- struct nouveau_bo *nvbo = nouveau_bo(bo);
- ret = nouveau_bo_move_accel_cleanup(chan, nvbo, evict,
- no_wait_gpu, new_mem);
+ ret = drm->ttm.move(chan, bo, &bo->mem, new_mem);
+ if (ret == 0) {
+ ret = nouveau_fence_new(chan, false, &fence);
+ if (ret == 0) {
+ ret = ttm_bo_move_accel_cleanup(bo, fence,
+ evict,
+ no_wait_gpu,
+ new_mem);
+ nouveau_fence_unref(&fence);
+ }
+ }
}
-
mutex_unlock(&chan->cli->mutex);
return ret;
}