On Wed, 12 Sep 2018, Radu Pirea wrote:

On Wed, 2018-09-12 at 14:12 +0100, Lee Jones wrote:

quoted

On Wed, 12 Sep 2018, Alexandre Belloni wrote:

quoted

On 12/09/2018 12:43:52+0100, Lee Jones wrote:

quoted

quoted

quoted

But ... we can't have it both ways.  *Either* it's a true
MFD, in
which case it can/should have 2 separate compatible strings
which can
be specified directly from the DT.  *Or* it's not an MFD.  In
the
latter case, which I think we're all agreeing on (else we'd
have 2
compatible strings), MFD is not the place to handle this (my
original
point).

If that is what bothers you, then let's move it out of mfd.

As I've already mentioned.  I don't just want it moved out of MFD
and
shoved somewhere else.  My aim is to fix this properly.

If it is out of MFD, then I'm not sure why you would care too much
about
it as you won't be maintaining that code. And I still this what was
done
was correct but I'm open to test what you suggest.

I care for the kernel in general, not just the areas I'm responsible
for.  I guess I'm just that kinda guy! ;)

Well, Lee, like you, I think this driver should not be a MFD driver,
but Alex has a good point of view. 

quoted

quoted

quoted

quoted

quoted

So ... this is a USART device which can do SPI, right?

My current thinking is that; as this is a USART device first
&
foremost, the USART should be probed in the first instance
regardless,
then if SPI mode is specified it (the USART driver) registers
the SPI
platform driver (as MFD does currently) and exits gracefully,
allowing
the SPI driver to take over.

Spanner in the works: is it physically possible to change the
mode at
run-time?  :s

Yes it is possible but on Linux that will not happen without
probing
the drivers again.

Not sure I understand what you mean.

I was just commenting on changing the mode at runtime.

Oh I see.  My question was relating to whether the H/W is physically
capable of changing modes on-the-fly, rather than how Linux would
handle that.  If this is something we'd wish to support, then it
would
have to be a single driver, which is why I was asking.  By separating
the drivers this way, we are blocking that as a
possibility.  Although
I guess the OP has already thought about that and made the decision
not to support it.

Is possible to change modes on-the-fly, but you have no reason to do
that. On the PCB you will have a SPI slave or a serial console :)
Anyway, the current form of the driver, and through this I want to say
"this ugly hack", allows the user to switch from serial to SPI mode by
adding only one property to the device tree node of USART. If the
driver were in his first form, a simple SPI driver, how you will make a
dtsi file for an IP like this? You will add two nodes for the same IP
in dtsi and will take care to enable correct node in dts?
I think this driver is only a tradeoff between having an ugly hack in
kernel or having an messy device tree.

quoted

quoted

quoted

I'm suggesting that you use the same platform_* interfaces MFD
uses to
register the SPI driver if SPI mode has been selected.  Only do
so
from the appropriate driver i.e. USART.

Yeah, I understood that but I didn't comment because I'm not sure
this
will work yet.

Other drivers already do this.

Can you give me an example please?

Sorry for the delay, I have been on vacation.

Grep for 'platform_device_add' in drivers/

I am open to suggestions.

Sorry for that acked-by. There was a lot of "reviewed-by", "acked-by",
etc in a single version and I messed up :).

-- 
Lee Jones [???]
Linaro Services Technical Lead
Linaro.org ? Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog