[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <qdrecp36jp34rrfwrrse4t7q626smyfddgzeozfjaovhemwoxa@nkg5ihejtzsr>
Date: Tue, 18 Jul 2023 13:24:27 +0200
From: "mripard@...nel.org" <mripard@...nel.org>
To: Krzysztof Kozlowski <krzysztof.kozlowski@...aro.org>
Cc: Frank Binns <Frank.Binns@...tec.com>,
"conor@...nel.org" <conor@...nel.org>,
Sarah Walker <Sarah.Walker@...tec.com>,
"luben.tuikov@....com" <luben.tuikov@....com>,
"christian.koenig@....com" <christian.koenig@....com>,
"krzysztof.kozlowski+dt@...aro.org"
<krzysztof.kozlowski+dt@...aro.org>,
"tzimmermann@...e.de" <tzimmermann@...e.de>,
"matthew.brost@...el.com" <matthew.brost@...el.com>,
"daniel@...ll.ch" <daniel@...ll.ch>,
"hns@...delico.com" <hns@...delico.com>,
"maarten.lankhorst@...ux.intel.com"
<maarten.lankhorst@...ux.intel.com>,
"boris.brezillon@...labora.com" <boris.brezillon@...labora.com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"dri-devel@...ts.freedesktop.org" <dri-devel@...ts.freedesktop.org>,
"afd@...com" <afd@...com>, "dakr@...hat.com" <dakr@...hat.com>,
"devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
"robh+dt@...nel.org" <robh+dt@...nel.org>,
"airlied@...il.com" <airlied@...il.com>,
"conor+dt@...nel.org" <conor+dt@...nel.org>,
Donald Robson <Donald.Robson@...tec.com>,
"faith.ekstrand@...labora.com" <faith.ekstrand@...labora.com>
Subject: Re: [PATCH v4 02/17] dt-bindings: gpu: Add Imagination Technologies
PowerVR GPU
Hi,
To expand on what Krzysztof said
On Tue, Jul 18, 2023 at 01:10:14PM +0200, Krzysztof Kozlowski wrote:
> On 18/07/2023 13:08, Frank Binns wrote:
> >> And this
> >> items:
> >> - const: gpu
> >> can just be
> >> const: gpu
> >>
> >> Although, if there is only one interrupt this is probably not
> >> particularly helpful. Are there other implementations of this IP that
> >> have more interrupts?
> >
> > No, all our current GPUs just have a single interrupt. I assume it's more future
> > proof to keep the name in case that ever changes?
>
> Why do you need name in the first place? If there is single entry, the
> name is pointless, especially if it repeats the name of the IP block.
Generally speaking, if there's only one resource (interrupt, clock, etc)
we don't put any discriminant.
If you need to extend it later on for some reason, then you'll add an
interrupt-names property and you can either require it for that new GPU
or whatever, or fallback on the nameless on if no name was found.
> > As in, by having the name now
> > we can make it a required property, which I guess we won't be able to do at some
> > later point.
>
> Why even making it required?
There's no issue with introducing a new property later on if a GPU needs
it. Then, you can either make it required only for that particular
generation, or you can have some fallback case.
Both are fine and easy to do.
Maxime
Download attachment "signature.asc" of type "application/pgp-signature" (229 bytes)
Powered by blists - more mailing lists