On 11/7/19 12:45 PM, Philipp Zabel wrote:
> Encoders that implement output side G/S_PARM and ENUM_FRAMEINTERVALS
> to allow setting the nominal bitstream frame rate cause compliance
> failures:
>
> fail: v4l2-test-formats.cpp(1330): ret && node->has_frmintervals
> test VIDIOC_G/S_PARM: FAIL
>
> because v4l2-compliance is missing capture side G/S_PARM. Skip the
> frameinterval check and make sure that capture side G_PARM fails.
It is completely undocumented that G/S_PARM should be valid only for the OUTPUT
type.
My opinion is that for M2M devices G/S_PARM should be valid for both CAPTURE
and OUTPUT. The framerate is a property of the m2m device itself, and both
queues share this information.
vidioc_enum_frameintervals is also shared between both queues.
For the stateful encoder spec I was planning to make this explicit: i.e. G/S_PARM
can be called for both capture and output buf types. And it should be mentioned
in the G/S_PARM documentation as well.
I think it makes sense, and I also think that that means that this test doesn't
need to be changed (but the coda driver does).
Regards,
Hans
>
> Signed-off-by: Philipp Zabel <p.zabel@xxxxxxxxxxxxxx>
> ---
> utils/v4l2-compliance/v4l2-test-formats.cpp | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/utils/v4l2-compliance/v4l2-test-formats.cpp b/utils/v4l2-compliance/v4l2-test-formats.cpp
> index b87e59c8be69..7336a823a269 100644
> --- a/utils/v4l2-compliance/v4l2-test-formats.cpp
> +++ b/utils/v4l2-compliance/v4l2-test-formats.cpp
> @@ -1283,6 +1283,11 @@ static int testParmType(struct node *node, unsigned type)
> switch (type) {
> case V4L2_BUF_TYPE_VIDEO_CAPTURE:
> case V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE:
> + if (IS_ENCODER(node)) {
> + fail_on_test(ret == 0);
> + break;
> + }
> + /* fallthrough */
> case V4L2_BUF_TYPE_VIDEO_OUTPUT:
> case V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE:
> if (type && (node->g_caps() & buftype2cap[type]))
>
