On 2022-07-18 02:57, Christian König wrote:
Am 15.07.22 um 22:04 schrieb Ruijing Dong:
From VCN4, AMDGPU_HW_IP_VCN_UNIFIED is used to support
both encoding and decoding jobs, it re-uses the same
queue number of AMDGPU_HW_IP_VCN_ENC.
link:
https://gitlab.freedesktop.org/mesa/drm/-/merge_requests/245/commits
Signed-off-by: Ruijing Dong <ruijing.dong@xxxxxxx>
---
include/uapi/drm/amdgpu_drm.h | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/include/uapi/drm/amdgpu_drm.h
b/include/uapi/drm/amdgpu_drm.h
index 18d3246d636e..e268cd3cdb12 100644
--- a/include/uapi/drm/amdgpu_drm.h
+++ b/include/uapi/drm/amdgpu_drm.h
@@ -560,6 +560,12 @@ struct drm_amdgpu_gem_va {
#define AMDGPU_HW_IP_UVD_ENC 5
#define AMDGPU_HW_IP_VCN_DEC 6
#define AMDGPU_HW_IP_VCN_ENC 7
+/**
Please don't use "/**" here, that is badly formated for a kerneldoc
comment.
+ * From VCN4, AMDGPU_HW_IP_VCN_UNIFIED is used to support
+ * both encoding and decoding jobs, it re-uses the same
+ * queue number of AMDGPU_HW_IP_VCN_ENC.
+ */
+#define AMDGPU_HW_IP_VCN_UNIFIED AMDGPU_HW_IP_VCN_ENC
I'm still in doubt that adding another define with the same value as
AMDGPU_HW_IP_VCN_ENC is a good idea.
Hi Christian,
From VCN4, there is no VCN dec and enc ring type any more, the
decode/encode will go through the unified queue, so using
AMDGPU_HW_IP_VCN_ENC is no longer accurate . Keeping
AMDGPU_HW_IP_VCN_ENC type is for legacy HW, and the new
AMDGPU_HW_IP_VCN_UNIFIED just happen to use the same HW ring as legacy
encode ring, so reuse the value, and that is the whole idea.
Thanks,
Leo
Instead we should just add the comment to AMDGPU_HW_IP_VCN_ENC.
Regards,
Christian.
#define AMDGPU_HW_IP_VCN_JPEG 8
#define AMDGPU_HW_IP_NUM 9
