Re: [PATCH V2 1/2] DT: mailbox: add binding doc for the ARM SMC mailbox

[Andre Przywara]

On Mon, 3 Jun 2019 17:56:51 +0100
Sudeep Holla <sudeep.holla@xxxxxxx> wrote:

Hi,

> On Mon, Jun 03, 2019 at 09:22:16AM -0700, Florian Fainelli wrote:
> > On 6/3/19 1:30 AM, peng.fan@xxxxxxx wrote:  
> > > From: Peng Fan <peng.fan@xxxxxxx>
> > >
> > > The ARM SMC mailbox binding describes a firmware interface to trigger
> > > actions in software layers running in the EL2 or EL3 exception levels.
> > > The term "ARM" here relates to the SMC instruction as part of the ARM
> > > instruction set, not as a standard endorsed by ARM Ltd.
> > >
> > > Signed-off-by: Peng Fan <peng.fan@xxxxxxx>
> > > ---
> > >
> > > V2:
> > > Introduce interrupts as a property.
> > >
> > > V1:
> > > arm,func-ids is still kept as an optional property, because there is no
> > > defined SMC funciton id passed from SCMI. So in my test, I still use
> > > arm,func-ids for ARM SIP service.
> > >
> > >  .../devicetree/bindings/mailbox/arm-smc.txt        | 101 +++++++++++++++++++++
> > >  1 file changed, 101 insertions(+)
> > >  create mode 100644 Documentation/devicetree/bindings/mailbox/arm-smc.txt
> > >
> > > diff --git a/Documentation/devicetree/bindings/mailbox/arm-smc.txt b/Documentation/devicetree/bindings/mailbox/arm-smc.txt
> > > new file mode 100644
> > > index 000000000000..401887118c09
> > > --- /dev/null
> > > +++ b/Documentation/devicetree/bindings/mailbox/arm-smc.txt
> > > @@ -0,0 +1,101 @@  
> 
> [...]
> 
> > > +Optional properties:
> > > +- arm,func-ids		An array of 32-bit values specifying the function
> > > +			IDs used by each mailbox channel. Those function IDs
> > > +			follow the ARM SMC calling convention standard [1].
> > > +			There is one identifier per channel and the number
> > > +			of supported channels is determined by the length
> > > +			of this array.
> > > +- interrupts		SPI interrupts may be listed for notification,
> > > +			each channel should use a dedicated interrupt
> > > +			line.  
> >
> > I would not go about defining a specific kind of interrupt, since SPI is
> > a GIC terminology, this firmware interface could be used in premise with
> > any parent interrupt controller, for which the concept of a SPI/PPI/SGI
> > may not be relevant.
> >  
> 
> While I agree the binding document may not contain specifics, I still
> don't see how to use SGI with this. Also note it's generally reserved
> for OS future use(IPC) and using this for other than IPC may be bit
> challenging IMO. It opens up lots of questions.

Well, a PPI might be possible to use, although it's somewhat dodgy to hijack the GIC's (re-)distributor from EL3 to write to GICD_ISPENDR<n>. Need to ask Marc about his feelings towards this. But it's definitely possible from a hypervisor to inject arbitrary interrupts into a guest.

But more importantly: is there any actual reason this needs to be a GIC interrupt? If I understand the code correctly, this could just be any interrupt, including one of an interrupt combiner or a GPIO chip. So why not just use the standard wording of: "exactly one interrupt specifier for each channel"?

By the way: Shouldn't new bindings use the YAML format instead?

Cheers,
Andre.