Messages in this thread | | | Date | Sat, 18 Nov 2023 11:35:30 +0000 | Subject | Re: [PATCH v4 4/7] media: qcom: camss: Move VFE power-domain specifics into vfe.c | From | Bryan O'Donoghue <> |
| |
On 18/11/2023 00:32, Konrad Dybcio wrote: >> + /* count the # of VFEs which have flagged power-domain */ > [...] > > Personal peeve, but this comment seems a bit excessive
Well, to me this code "does stuff" that isn't terribly obvious. It took a while to understand the Gordian knot so its worthwhile documenting it inline until we can use named pds to make it all go away forever.
> >> + for (vfepd_num = i = 0; i < camss->vfe_total_num; i++) { >> + if (res->vfe_res[i].has_pd) >> + vfepd_num++; >> + } >> >> - camss->genpd_link = devm_kmalloc_array(dev, camss->genpd_num, >> - sizeof(*camss->genpd_link), >> - GFP_KERNEL); >> - if (!camss->genpd_link) >> - return -ENOMEM; >> + /* >> + * If the number of power-domains is greater than the number of VFEs >> + * then the additional power-domain is for the entire CAMSS block the >> + * 'top' power-domain. > the last 3 words seem out of place
> >> + */ >> + if (camss->genpd_num <= vfepd_num) >> + return 0; > if (!(camss->genpd_num > vfepd_num)) > > would probably be easier to follow given your comment above
Sure, if its easier to read/understand for you, then it is probably so for others.
I will V5 this.
--- bod
| |