Re: [PATCH] Documentation: dt: mtd: replace "nor-jedec" binding with "jedec,spi-nor"

[Brian Norris <computersforpeace@gmail.com>]

On Thu, May 21, 2015 at 10:01:05AM +0200, Rafał Miłecki wrote:
> On 21 May 2015 at 09:25, Brian Norris <computersforpeace@gmail.com> wrote:
> > (trim CC a bit, as this is no longer a DT binding question)
> >
> > On Thu, May 21, 2015 at 09:12:25AM +0200, Rafał Miłecki wrote:
> >> On 20 May 2015 at 23:35, Brian Norris <computersforpeace@gmail.com> wrote:
> >> > On Tue, May 19, 2015 at 09:27:50AM +0200, Rafał Miłecki wrote:
> >> >> On 19 May 2015 at 03:34, Brian Norris <computersforpeace@gmail.com> wrote:
> >> >> > So how about the following patch? It seems like we'll need to be able to
> >> >> > ignore useless 'modalias' values in cases like this:
> >> >> >
> >> >> >         // modalias = "shinynewdevice"
> >> >> >         compatible = "myvendor,shinynewdevice", "jedec,spi-nor";
> >> >> >
> >> >> > and also if somebody leaves off the entire shinynewdevice string:
> >> >> >
> >> >> >         // modalias = "spi-nor"
> >> >> >         compatible = "jedec,spi-nor";
> >> >> >
> >> >> > So we rework the spi-nor library to not reject "bad" names, and just
> >> >> > fall back to autodetection, and we add the .of_match_table to properly
> >> >> > catch all "jedec,spi-nor".
> >> >>
> >> >> That's nice but what about platforms using platform data instead of
> >> >> DT? I would like to use some kind of "spi-nor" (with some prefix
> >> >> *maybe*) for them too.
> >> >
> >> > For platform devices, you might as well just use the name of the driver,
> >> > which is 'm25p80'. Isn't that how most platform devices are matched with
> >> > drivers?
> >>
> >> Yes and I think it's ugly because it keeps causing the warning about
> >> read flash model not matching specified one (m25p80).
> >
> > Sure, I agree.
> >
> >> Are you
> >> seriously not going to allow platform stuff *clearly* request flash
> >> model detection (JEDEC RDID OP)? Just because they don't use DT?
> >
> > No, this isn't about "allowing" anything. It's just that my primary
> > concern was to get the DT binding straightened out properly. Linus'
> > current tree now has the proper binding, but the m25p80.c code doesn't
> > quite bind properly. It will work if "jedec,spi-nor" is the first
> > entry in the compatible property (and so it becomes the 'modalias', but
> > not second, third, etc. So my patch fixes that properly.
> >
> > Now, the secondary concern is that you want platform devices to specify
> > something generic, and that doesn't yield a "found X, expected Y"
> > message. I'm perfectly fine with fixing that too, if you have a patch
> > for it. What do you propose?
> 
> Maybe I wasn't clear enough. I was going to start using struct
> flash_platform_data with
> .type = "spi-nor",
> but your proposed patch removes support for such name.

Ah, OK. So that's the part I was overlooking.

> While I like matching DT *clearly* against the whole "jedec,spi-nor"
> string (really, I'm all for it), I'm confused what I should use for
> platform stuff now. I don't have any proposal as my initial plan was
> exactly to use this "spi-nor".
> I guess I don't want to re-add support for "spi-nor" (as you just
> proposed to remove it),

I wasn't really trying to remove "spi-nor", that was mostly a side
effect.

> so I think I have to bounce the question: what
> alternative do you propose?

I think your comments suggest that I shouldn't be removing "spi-nor"
from m25p_ids[] nor from this block:

	if (data && data->type)
		flash_name = data->type;
	else if (!strcmp(spi->modalias, "spi-nor"))
		flash_name = NULL; /* auto-detect */
	else
		flash_name = spi->modalias;

So it stays in both m25p_ids[] and .of_match_table.

I suppose that can work. It then allows people to do weird stuff like:

	compatible = "idontknowwhatimdoing,spi-nor";

in their device tree. But other than that, there's not much downside I don't
think.

Brian