Re: [PATCH v3 1/4] dt-bindings: arm: qcom-soc: simplify SoC-matching patterns

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, 1 Nov 2024 at 11:23, Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote:
>
> On 01/11/2024 09:52, Dmitry Baryshkov wrote:
> > On Fri, Nov 01, 2024 at 09:37:23AM +0100, Krzysztof Kozlowski wrote:
> >> On 01/11/2024 08:47, Dmitry Baryshkov wrote:
> >>> On Fri, Nov 01, 2024 at 08:26:04AM +0100, Krzysztof Kozlowski wrote:
> >>>> On Fri, Nov 01, 2024 at 02:49:22AM +0200, Dmitry Baryshkov wrote:
> >>>>> The patterns for individual SoC families grew up to be pretty complex,
> >>>>> containing lots of special cases and optional suffixes. Split them per
> >>>>> the suffix to make it easier to extend SoC patterns.
> >>>>
> >>>> This is doing something quite different - split is not important here.
> >>>> Instead you narrow the patterns significantly and disallow things like
> >>>> msm8994pro, sc8280p or sc8280px, and allow things like sa5200p.
> >>>
> >>> Just for the sake of correctness, msm8994pro is still allowed, if I'm
> >>> not mistaken.
> >>>
> >>>> I don't see here much of pattern simplifying - dropping (pro)? really
> >>>> makes little difference.
> >>>
> >>> Patterns are simplified by being explicit. E.g. in the previous
> >>> iteration I completely didn't notice the intersection of the |p that I
> >>> have added with the existing [a-z][a-z]? pattern. If you think that
> >>> sa5200p should be disallowed, I can tune the numeric part of the
> >>> pattern. And sc8280p / sc8280px should not be allowed in the first
> >>> place, such platforms don't exist.
> >>
> >> I am fine with this, but extend the commit msg with some good rationale.
> >> Have in mind that the point of this pattern was *not* to validate SoCs
> >> names. sa5200p is fine, sc8180p is fine and all others are fine, sc8280z
> >> as well, because we do not want to grow this pattern with every new model.
> >>
> >> The only, single point of this entire binding is to disallow incorrect
> >> order of block names in compatible. Not validate the SoC names. If you
> >> need narrower patterns to achieve that objective, sure. If you need
> >> narrower patterns to validate SoC names, then nope.
> >
> > I need narrower patterns to simplify adding new SoCs.
> > Another option is to define a mega-pattern like
> > qcom,(msm|sm|sd[am]|.....)[0-9]+[a-z]*-.* . Frankly speaking I'm fine
> > with that approach too.
>
> I do not see how narrower patterns, changing:
> "^qcom,(sa|sc)8[0-9]+[a-z][a-z]?-.*$"
> into
> pattern: "^qcom,sa[0-9]+p-.*$"
>
> instead of
> pattern: "^qcom,sa[0-9]+[a-z]-.*$"
> is needed for that. It's true that 'p' is simpler than '[a-z]' but if
> this results in new commit next time we have sa8995r, then benefit is lost.

Ok, let's drop it, I will just update the qcom-soc.yaml to handle
sar21330p via the "^qcom,(sa|sar|sc)8[0-9]+[a-z][a-z]?-.*$"

>
> Best regards,
> Krzysztof
>


-- 
With best wishes
Dmitry




[Index of Archives]     [Device Tree Compilter]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux PCI Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Yosemite Backpacking]


  Powered by Linux