On 4/8/19 2:32 PM, Philipp Zabel wrote: > v4l2-compliance sets colorimetry on the output queue and then verifies > that querying colorimetry on the capture queue returns the same > configuration. For this to work, the encoder must allow setting context > colorimetry parameters on the output queue. > > Signed-off-by: Philipp Zabel <p.zabel@xxxxxxxxxxxxxx> > --- > drivers/media/platform/coda/coda-common.c | 10 +++++----- > 1 file changed, 5 insertions(+), 5 deletions(-) > > diff --git a/drivers/media/platform/coda/coda-common.c b/drivers/media/platform/coda/coda-common.c > index 2966eb1c4d2d..7b89dbae938d 100644 > --- a/drivers/media/platform/coda/coda-common.c > +++ b/drivers/media/platform/coda/coda-common.c > @@ -819,6 +819,11 @@ static int coda_s_fmt_vid_out(struct file *file, void *priv, > if (ret) > return ret; > > + ctx->colorspace = f->fmt.pix.colorspace; > + ctx->xfer_func = f->fmt.pix.xfer_func; > + ctx->ycbcr_enc = f->fmt.pix.ycbcr_enc; > + ctx->quantization = f->fmt.pix.quantization; > + > if (ctx->inst_type != CODA_INST_DECODER) > return 0; > > @@ -831,11 +836,6 @@ static int coda_s_fmt_vid_out(struct file *file, void *priv, > } > ctx->codec = codec; > > - ctx->colorspace = f->fmt.pix.colorspace; > - ctx->xfer_func = f->fmt.pix.xfer_func; > - ctx->ycbcr_enc = f->fmt.pix.ycbcr_enc; > - ctx->quantization = f->fmt.pix.quantization; > - > dst_vq = v4l2_m2m_get_vq(ctx->fh.m2m_ctx, V4L2_BUF_TYPE_VIDEO_CAPTURE); > if (!dst_vq) > return -EINVAL; > Isn't the colorimetry information encoded in the stream's metadata? So should it be set in an encoder register so it ends up in the right place in the bitstream? And for decoders it should then be read back from a register. Just curious how this would work for this codec and if it is even possible. For now this patch is OK, but I think more work is needed. Regards, Hans
