Re: [PATCH v3 01/13] spi: dt-bindings: allow spi-max-frequency to specify a frequency pair

From: Miquel Raynal <miquel.raynal@bootlin.com>
To: Conor Dooley <conor@kernel.org>
Cc: Santhosh Kumar K <s-k6@ti.com>,
	 broonie@kernel.org,  robh@kernel.org, krzk+dt@kernel.org,
	 conor+dt@kernel.org,  richard@nod.at, vigneshr@ti.com,
	 pratyush@kernel.org,  mwalle@kernel.org,
	takahiro.kuwano@infineon.com,  linux-spi@vger.kernel.org,
	devicetree@vger.kernel.org,  linux-kernel@vger.kernel.org,
	linux-mtd@lists.infradead.org,  praneeth@ti.com,
	 u-kumar1@ti.com, a-dutta@ti.com
Subject: Re: [PATCH v3 01/13] spi: dt-bindings: allow spi-max-frequency to specify a frequency pair
Date: Fri, 05 Jun 2026 08:32:56 +0200	[thread overview]
Message-ID: <87a4t9wl7r.fsf@bootlin.com> (raw)
In-Reply-To: <20260604-shrunk-acclaim-25368d7ad63e@spud> (Conor Dooley's message of "Thu, 4 Jun 2026 09:36:49 +0100")

On 04/06/2026 at 09:36:49 +01, Conor Dooley <conor@kernel.org> wrote:

> On Thu, Jun 04, 2026 at 09:14:16AM +0200, Miquel Raynal wrote:
>> Hi Conor,
>> 
>> >> >> > Right, and this I guess is what scuppers letting the controller driver
>> >> >> > sort the configuration out itself and leaving the property as-is.
>> >> >> > It could be that the speed in spi-max-frequency is lower than the "base
>> >> >> > speed" of the controller but because of board routing or device
>> >> >> > capability that the tuned mode is still required, right?
>> >> >> 
>> >> >> I do not actually expect any tuned mode/frequency to be mandatory.
>> >> >
>> >> > I think you misunderstood my use of "required", I meant that the new
>> >> > property/information was needed in the scenario I described, not that it
>> >> > should be a required property in a binding.
>> >> 
>> >> Yes I misunderstood the term indeed. However I still fail to catch what
>> >> you meant here, I'm sorry. Would you mind rephrasing?
>> >
>> > I was talking about a scenario where you want to use the tuned mode to
>> > achieve the maximum rate because of the device and/or board configuration,
>> > but the rate is below the point where the controller would need tuning.
>> > Say the controller needs tuning above 8 Hz but the conditions require
>> > tuning to achieve more than 5 Hz. In this example, if the device can do
>> > 6 Hz, spi-max-frequency (in the current form) would be set to 6 Hz, and
>> > the controller would not enable the tuned mode, leading to problems
>> > because the inflection point determined from the controller compatible
>> > of 8 Hz would not have been reached.
>> 
>> I don't think this is a real situation. If the "conditions", as you say
>> (ie. PCB routing, mostly) require tuning above 5, then spi-max-frequency
>> should be 5.
>
> Then tuning mode would never be used.

Well, this is exactly what we propose in this series, a way to indicate
two maximum frequencies, one that just works (like before) and a higher
frequency that is only reachable after an extra tuning procedure.

> Remember, this is a theoretical world
> where spi-max-frequency would contain the tuned frequency and the
> controller was using compatible-specific speed thresholds to determine
> if tuning was required.

That is not what we are proposing. I don't think indicating the "after
tuning" frequency in a property that has long been used for an always
reachable frequency is wise. Hence either the use of an array (the
second entry could contain a higher frequency) or a secondary
spi-max-frequency-whatever property.

Thanks,
Miquèl