On Sat, Mar 01, 2025 at 09:41:14PM +0800, damon wrote:
> On 25/02/28 06:22PM, Conor Dooley wrote:
> > On Fri, Feb 28, 2025 at 08:40:23PM +0800, Zixian Zeng wrote:
> > > diff --git a/arch/riscv/boot/dts/sophgo/sg2042.dtsi b/arch/riscv/boot/dts/sophgo/sg2042.dtsi
> > > index e62ac51ac55abd922b5ef796ba8c2196383850c4..500645147b1f8ed0a08ad3cafb38ea79cf57d737 100644
> > > --- a/arch/riscv/boot/dts/sophgo/sg2042.dtsi
> > > +++ b/arch/riscv/boot/dts/sophgo/sg2042.dtsi
> > > @@ -545,5 +545,33 @@ sd: mmc@704002b000 {
> > > "timer";
> > > status = "disabled";
> > > };
> > > +
> > > + spi0: spi@7040004000 {
> > > + compatible = "snps,dw-apb-ssi";
> >
> > I thought were were dropping the use of "snps,dw-abp-ssi" in isolation,
> > and starting to require soc-specific compatibles now.
> >
> That’s better, I will do it in next revision.
> BTW, <https://lore.kernel.org/linux-riscv/20240703-garbage-explicit-bd95f8deb716@wendy>
> has similar situation,
> those patches only create binding but not implement the compatible
> property of "thead,th1520-spi” in driver.
> When kernel matching, it will find out there’s no compatible of
> "thead,th1520-spi” so fallback to "snps,dw-apb-ssi”
> May I take you a few minutes for reason behind it? My understanding is
> that using soc-specific compatible is better
> for future maintaining, If the device actually comes with some unknown
> quirks not present in the generic IP.
Yup, that's pretty much it. Those quirks could be some sort of bug etc,
or they could be an extra feature. Sometimes the specific compatible is
only used to enforce constraints on what properties are allowed, but
that's more of a thing for the more configurable IPs like the dwmac.
> And all we need to do for fixing is implementing the soc-specific
> compatible property.
Correct. You just need to add a specific compatible to the binding and
to the dts, the driver will work just fine with the "fallback" dw one.
Attachment:
signature.asc
Description: PGP signature
