Re: [PATCH 2/2] firmware: arm_scmi: Add compatibility checks for shmem node

From: Etienne Carriere <etienne.carriere@linaro.org>
To: Sudeep Holla <sudeep.holla@arm.com>
Cc: "moderated list:ARM/FREESCALE IMX / MXC ARM ARCHITECTURE"
	<linux-arm-kernel@lists.infradead.org>,
	 Cristian Marussi <cristian.marussi@arm.com>,
	Kevin Hilman <khilman@baylibre.com>,
	 Neil Armstrong <narmstrong@baylibre.com>,
	Jerome Brunet <jbrunet@baylibre.com>,
	Florian Fainelli <f.fainelli@gmail.com>,
	Jim Quinlan <jim2101024@gmail.com>
Subject: Re: [PATCH 2/2] firmware: arm_scmi: Add compatibility checks for shmem node
Date: Wed, 2 Jun 2021 09:44:40 +0200	[thread overview]
Message-ID: <CAN5uoS8T=D+dJ+nCpxr-6HJJRAphr1LgZq4DBab65CWT=ciuXg@mail.gmail.com> (raw)
In-Reply-To: <20210602073653.x4bon6jbiat2jnqv@bogus>

On Wed, 2 Jun 2021 at 09:37, Sudeep Holla <sudeep.holla@arm.com> wrote:
>
> On Wed, Jun 02, 2021 at 09:33:03AM +0200, Etienne Carriere wrote:
> > Hello Sudeep,
> >
> >
> > On Wed, 2 Jun 2021 at 00:51, Sudeep Holla <sudeep.holla@arm.com> wrote:
> > >
> > > The shared memory node used for communication between the firmware and
> > > the OS should be compatible with "arm,scmi-shmem". Add the check for the
> > > same while parsing the node before fetching the memory regions.
> > >
> > > Cc: Cristian Marussi <cristian.marussi@arm.com>
> > > Cc: Florian Fainelli <f.fainelli@gmail.com>
> > > Cc: Jim Quinlan <jim2101024@gmail.com>
> > > Cc: Etienne Carriere <etienne.carriere@linaro.org>
> > > Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
> > > ---
> > >  drivers/firmware/arm_scmi/mailbox.c | 3 +++
> > >  drivers/firmware/arm_scmi/smc.c     | 3 +++
> > >  2 files changed, 6 insertions(+)
> > >
> > > diff --git a/drivers/firmware/arm_scmi/mailbox.c b/drivers/firmware/arm_scmi/mailbox.c
> > > index 4626404be541..e3dcb58314ae 100644
> > > --- a/drivers/firmware/arm_scmi/mailbox.c
> > > +++ b/drivers/firmware/arm_scmi/mailbox.c
> > > @@ -69,6 +69,9 @@ static int mailbox_chan_setup(struct scmi_chan_info *cinfo, struct device *dev,
> > >                 return -ENOMEM;
> > >
> > >         shmem = of_parse_phandle(cdev->of_node, "shmem", idx);
> > > +       if (!of_device_is_compatible(shmem, "arm,scmi-shmem"))
> > > +               return -ENXIO;
> >
> > Before this change, one could use another type of memory node, like "mmio-sram".
> > Is there a strong reason to enforce use of "arm,scmi-shmem" nodes?
> >
>
> No that is for the entire SRAM which still holds and generic on-chip SRAM
> driver will take care of that, this is only for the subsections that is
> reserved for the scp shmem. The binding has been always there, just the
> missing check. When I move to yaml, I realised that and hence the
> addition of check.

Ok, I understand. True the binding was there but only in the DTS
examples snipped.
This constraint on the compatible property of the shmem node should be
clearly stated in the yaml I think.

br,
etienne

>
> --
> Regards,
> Sudeep

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel