On Wed, Nov 06, 2024 at 10:18:01AM +0000, Matt Coster wrote: > On 05/11/2024 18:13, Conor Dooley wrote: > > On Tue, Nov 05, 2024 at 06:05:54PM +0000, Conor Dooley wrote: > >> On Tue, Nov 05, 2024 at 03:58:09PM +0000, Matt Coster wrote: > >>> The single existing GPU (AXE-1-16M) only requires a single power domain. > >>> Subsequent patches will add support for BXS-4-64 MC1, which has two power > >>> domains. Add infrastructure now to allow for this. > >>> > >>> Signed-off-by: Matt Coster <matt.coster@xxxxxxxxxx> > >>> --- > >>> .../devicetree/bindings/gpu/img,powervr-rogue.yaml | 29 +++++++++++++++++++++- > >>> 1 file changed, 28 insertions(+), 1 deletion(-) > >>> > >>> diff --git a/Documentation/devicetree/bindings/gpu/img,powervr-rogue.yaml b/Documentation/devicetree/bindings/gpu/img,powervr-rogue.yaml > >>> index 6924831d3e9dd9b2b052ca8f9d7228ff25526532..55f422be1bc5b7564e3e81f24c4b93857f3e12fe 100644 > >>> --- a/Documentation/devicetree/bindings/gpu/img,powervr-rogue.yaml > >>> +++ b/Documentation/devicetree/bindings/gpu/img,powervr-rogue.yaml > >>> @@ -49,7 +49,16 @@ properties: > >>> maxItems: 1 > >>> > >>> power-domains: > >>> - maxItems: 1 > >>> + minItems: 1 > >>> + maxItems: 2 > >>> + > >>> + power-domain-names: > >>> + oneOf: > >>> + - items: > >>> + - const: a > >>> + - items: > >>> + - const: a > >>> + - const: b > > > > Additionally, a & b? Are those actually the names for the power domains? > > Sadly yes, Rogue has power domains that are literally just A, B, etc. I > wouldn't believe me either; the attached image is taken directly from > the documentation for BXS-4-64. heh, nice... Also - if you have the oneOf stuff here for the same reason as the locks, the same logic with min/maxItems applies here. > > >>> > >>> required: > >>> - compatible > >>> @@ -57,10 +66,27 @@ required: > >>> - clocks > >>> - clock-names > >>> - interrupts > >>> + - power-domains > >>> + - power-domain-names > >> > >> A new required property is an ABI break. Please explain why this is > >> acceptable in your commit message. > > Strictly it's only necessary for multi-domain GPUs, or perhaps in > instances where the SoC power controller already enforces the > dependencies between power domains. In reality, I think it was simply an > oversight not to enfore this requirement in the first place. We have > very, very few cores that require <2 power domains so names are always > required if domains are enumerated in dt. > > Would you prefer we drop the requirement for "power-domains" and gate > the requirement for "power-domain-names" behind >2 domains, or just > explain the change properly and make the ABI break now while only one > core is supported? I dunno, depends really on whether or not it is possible to have power domain "a" and domain "c" in a rogue gpu. If "a" and "b" is all it will ever be, the order is fixed by the binding and you can do genpd_dev_pm_attach_by_id() instead of genpd_dev_pm_attach_by_name().
Attachment:
signature.asc
Description: PGP signature
