Re: [PATCH 2/4] media: qcom: iris: Add Long-Term Reference support for encoder

[Wangao Wang]

On 2025/11/27 18:41, Bryan O'Donoghue wrote:
> > @@ -693,6 +693,31 @@ iris_hfi_gen1_packet_session_set_property(struct 
> > hfi_session_set_property_pkt *p
> >           packet->shdr.hdr.size += sizeof(u32) + sizeof(*intra_refresh);
> >           break;
> >       }
> > +    case HFI_PROPERTY_PARAM_VENC_LTRMODE: {
> > +        struct hfi_ltr_mode *in = pdata, *ltr_mode = prop_data;
> > +
> > +        ltr_mode->mode = in->mode;
> > +        ltr_mode->count = in->count;
> > +        ltr_mode->trust_mode = in->trust_mode;
> > +        packet->shdr.hdr.size += sizeof(u32) + sizeof(*ltr_mode);
>
> This looks like an unconditional pattern in this code
>
> packet->shdr.hdr.size += sizeof(u32) for each one of the switch cases.
>
> Could you at least add a patch to the end of this series to tidy this up 
> - either always incrmenting by sizeof(u32) before the swich - or making 
> a macro or static inline that packs the header appropriately.
>
> Also what is the sizeof(u32) representative of ? "Some field in the 
> header" - get that but right now its just a literal unnamed dword.

A patch will be added in v2 to optimize the code structure here.
--
Best Regards,
Wangao