Hi Mansur, I tested the patch on kernel 5.4 on an Acer Chromebook Spin 513 based on the Qualcomm Snapdragon 7c platform. VP8 test vectors [0] were able to be decoded and decoded picture MD5's matched the vpxdec reference decoder. Prior to this patch, three files failed to decode. I didn't see any regressions with VP9 Profile 0 (8-bit) test vectors [1] using the same decode and checksum test methods. [0] https://chromium.googlesource.com/webm/vp8-test-vectors [1] https://www.webmproject.org/vp9/levels/#test-bitstreams Tested-by: Nathan Hebert <nhebert@xxxxxxxxxxxx> Best regards, Nathan Hebert On Wed, Aug 25, 2021 at 1:44 AM Mansur Alisha Shaik <mansur@xxxxxxxxxxxxxx> wrote: > > Video driver maintains an internal context for the output buffer size. > During S_fmt() on capture plane, the output buffer size is not updated > in driver context. As a result, during buf_prepare(), the size of the > vb2_plane and internal size of the buffer, as maintained by the driver, > does not match. This leads to buf_prepare() failure. > Update the instance context for the output buffer size during s_fmt(). > > Signed-off-by: Mansur Alisha Shaik <mansur@xxxxxxxxxxxxxx> > --- > drivers/media/platform/qcom/venus/vdec.c | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/drivers/media/platform/qcom/venus/vdec.c b/drivers/media/platform/qcom/venus/vdec.c > index 198e47eb63f4..c129b061a325 100644 > --- a/drivers/media/platform/qcom/venus/vdec.c > +++ b/drivers/media/platform/qcom/venus/vdec.c > @@ -332,8 +332,11 @@ static int vdec_s_fmt(struct file *file, void *fh, struct v4l2_format *f) > > if (f->type == V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE) > inst->fmt_out = fmt; > - else if (f->type == V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE) > + else if (f->type == V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE) { > inst->fmt_cap = fmt; > + inst->output2_buf_size = > + venus_helper_get_framesz(pixfmt_cap, orig_pixmp.width, orig_pixmp.height); > + } > > return 0; > } > -- > QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member > of Code Aurora Forum, hosted by The Linux Foundation >
