Re: [PATCH v3 07/13] media: controls: Validate H264 stateless controls

[Ezequiel Garcia]

On Fri, 2020-11-20 at 10:30 +0100, Hans Verkuil wrote:
> On 18/11/2020 19:46, Ezequiel Garcia wrote:
> > Check that all the fields that correspond or are related
> > to a H264 specification syntax element have legal values.
> > 
> > Signed-off-by: Ezequiel Garcia <ezequiel@xxxxxxxxxxxxx>
> > ---
> >  drivers/media/v4l2-core/v4l2-ctrls.c | 83 ++++++++++++++++++++++++++++
> >  1 file changed, 83 insertions(+)
> > 
[..]
> >  	case V4L2_CTRL_TYPE_H264_DECODE_PARAMS:
> >  		p_h264_dec_params = p;
> >  
> > +		if (p_h264_dec_params->nal_ref_idc > 3)
> > +			return -EINVAL;
> >  		for (i = 0; i < V4L2_H264_NUM_DPB_ENTRIES; i++) {
> >  			struct v4l2_h264_dpb_entry *dpb_entry =
> >  				&p_h264_dec_params->dpb[i];
> > 
> 
> General question: I don't see anything in std_init_compound() for these
> controls. Is initializing these compound controls to 0 enough to make them
> pass std_validate_compound()? It probably is, otherwise you'd see errors
> in the compliance test, I guess.
> guess.
> 

Indeed. You can see all the checks are for fields to not
exceed some maximum value.

This is common in H264/HEVC: you'll see  spread of _minusN
syntax. This is so to make zero-valued syntax common,
which in turns creates more redundancy, and make headers
more compressed.

Thanks,
Ezequiel