Lack of MTX_CMDID_NULL causes topaz fence to timeout

this timeout results in reset the hardware for every
jpeg frame encode. vaSyncSurface will not complete until
the timeout occurs, since it is still waiting for
MTX_CMID_NULL. This shows up as slow performance
(we have to wait for timeout).

Issue: ANDROID-1633

Signed-off-by: naeem.m.afzal@intel.com
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

diff --git a/drivers/staging/mrst/imgv/pnw_topaz.c b/drivers/staging/mrst/imgv/pnw_topaz.c
index b646fd7..5ada7b4 100644 (file)
--- a/drivers/staging/mrst/imgv/pnw_topaz.c
+++ b/drivers/staging/mrst/imgv/pnw_topaz.c
@@ -521,6 +521,7 @@ pnw_topaz_send(struct drm_device *dev, void *cmd,
                case MTX_CMDID_RESET_ENCODE:
                case MTX_CMDID_ISSUEBUFF:
                case MTX_CMDID_SETUP:
+               case MTX_CMDID_NULL:
                        cur_cmd_header->seq = topaz_priv->topaz_cmd_count++;
                        cur_cmd_header->enable_interrupt = 0;
                        cur_cmd_size = 4;