Re: [PATCH v7 3/6] media: qcom: iris: Add B frames support for encoder

[Wangao Wang]

On 2026/5/13 19:00, Dmitry Baryshkov wrote:
> > +int iris_set_intra_period(struct iris_inst *inst, enum platform_inst_fw_cap_type cap_id)
> > +{
> > +	const struct iris_hfi_session_ops *hfi_ops = inst->hfi_session_ops;
> > +	u32 gop_size = inst->fw_caps[GOP_SIZE].value;
> > +	u32 b_frame = inst->fw_caps[B_FRAME].value;
> > +	u32 hfi_id = inst->fw_caps[cap_id].hfi_id;
> > +	struct hfi_intra_period intra_period;
> > +
> > +	if (!gop_size || b_frame >= gop_size)
> > +		return -EINVAL;
>
> The same comment. Maybe I misunderstand something, please correct me if
> I'm wrong. The definition of the GOP_SIZE capability allows 0 as a valid
> value. Here you are declining it. Why?
>
> > +

GOP_SIZE
- Value 0: Behavior depends on the firmware implementation.In most 
cases, the firmware will treat it as an invalid value or fall back to a 
default value.
- Value 1: Explicitly means all I-frames.

The gen1 firmware does not support passing GOP size directly — it is 
determined by the number of P-frames and B-frames between I-frames. If 
the client passes a value of 0, it will cause the subsequent gop_size - 
1 calculation to produce a negative value.

--
Best Regards,
Wangao