[PATCH v2 2/2] firmware: arm_scmi: Implement arm,poll-transport property

[Cristian Marussi]

On Mon, Jan 12, 2026 at 12:06:11PM +0000, Sudeep Holla wrote:
> On Wed, Dec 31, 2025 at 10:29:19PM +0100, Marek Vasut wrote:
> > Implement new property arm,poll-transport, which sets all SCMI operation into
> > poll mode. This is meant to work around uncooperative SCP implementations,
> > which do not generate completion interrupts. This applies to mbox/shmem based
> > implementations.
> > 
> > With this property set, such implementations which do not generate interrupts
> > can be interacted with, until they are fixed to generate interrupts properly.
> > 
> > Signed-off-by: Marek Vasut <marek.vasut+renesas at mailbox.org>
> > ---
> > Cc: Conor Dooley <conor+dt at kernel.org>
> > Cc: Cristian Marussi <cristian.marussi at arm.com>
> > Cc: Florian Fainelli <florian.fainelli at broadcom.com>
> > Cc: Krzysztof Kozlowski <krzk+dt at kernel.org>
> > Cc: Rob Herring <robh at kernel.org>
> > Cc: Sudeep Holla <sudeep.holla at arm.com>
> > Cc: arm-scmi at vger.kernel.org
> > Cc: devicetree at vger.kernel.org
> > Cc: linux-arm-kernel at lists.infradead.org
> > Cc: linux-renesas-soc at vger.kernel.org
> > ---
> > V2: Drop no IRQ handling from SMC transport and update commit message
> > ---
> >  drivers/firmware/arm_scmi/common.h | 4 ++++
> >  drivers/firmware/arm_scmi/driver.c | 4 ++++
> >  2 files changed, 8 insertions(+)
> > 
> > diff --git a/drivers/firmware/arm_scmi/common.h b/drivers/firmware/arm_scmi/common.h
> > index 7c35c95fddbaf..7c9617d080a02 100644
> > --- a/drivers/firmware/arm_scmi/common.h
> > +++ b/drivers/firmware/arm_scmi/common.h
> > @@ -235,6 +235,9 @@ struct scmi_transport_ops {
> >   *		      to have an execution latency lesser-equal to the threshold
> >   *		      should be considered for atomic mode operation: such
> >   *		      decision is finally left up to the SCMI drivers.
> > + * @no_completion_irq: Flag to indicate that this transport has no completion
> > + *		       interrupt and has to be polled. This is similar to the
> > + *		       force_polling below, except this is set via DT property.
> >   * @force_polling: Flag to force this whole transport to use SCMI core polling
> >   *		   mechanism instead of completion interrupts even if available.
> >   * @sync_cmds_completed_on_ret: Flag to indicate that the transport assures
> > @@ -254,6 +257,7 @@ struct scmi_desc {
> >  	int max_msg;
> >  	int max_msg_size;
> >  	unsigned int atomic_threshold;
> > +	bool no_completion_irq;
> >  	const bool force_polling;
> 
> My preference would be to reuse `force_polling` for this. We need to drop
> const but that should be OK. Anyways I would like to know if Cristian thinks
> otherwise for any reasons I might be missing to see.

I would rather keep the 2 things separate since force_polling is more of
a brutal low level debug/test facility and, even though it basically
produces the same result as the new @no_completion_irq, if we remove it
and unify it in a single boolean that can be overriden from the DT we end
up in a situation in which we cannot anymore easily force_polling by
switching the flag in the code since it could be overridden by a
conflicting 'arm,poll-transport' DT setup. (and you have to patch DT
for testing)

So if we have one single underlying boolean (e.g. 'poll') and by any chance
we end up with a DT containing:

	arm,poll-transport = false

we cannot anymore override the condition by forcing in the code

	poll = true,

since it would be switfly overridden by the DT prop.

Also semantically force_polling express much more the situation.

Anyway...I may be overthinking or missing something.

Cheers,
Cristian