Re: [PATCH] pcmcia: at91_cf: switch to using gpiod API

From: Alexandre Belloni <alexandre.belloni@bootlin.com>
To: Arnd Bergmann <arnd@arndb.de>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>,
	linux-kernel@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org,
	Claudiu Beznea <claudiu.beznea@microchip.com>
Subject: Re: [PATCH] pcmcia: at91_cf: switch to using gpiod API
Date: Sat, 24 Sep 2022 13:42:37 +0200	[thread overview]
Message-ID: <Yy7tLRcMaKeet9An@mail.local> (raw)
In-Reply-To: <68c63077-848b-45f5-8aca-ed995391f2b6@www.fastmail.com>

On 24/09/2022 10:33:29+0200, Arnd Bergmann wrote:
> On Sat, Sep 24, 2022, at 8:04 AM, Dmitry Torokhov wrote:
> > This patch switches the driver to use newer gpiod API instead of legacy
> > gpio API. This moves us closer to the goal of stopping exporting
> > OF-specific APIs of gpiolib.
> >
> > While at it, stop using module-global for regmap.
> >
> > Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> 
> This looks good to me overall. Three comments:
> 
> > @@ -63,7 +62,7 @@ struct at91_cf_socket {
> > 
> >  static inline int at91_cf_present(struct at91_cf_socket *cf)
> >  {
> > -	return !gpio_get_value(cf->board->det_pin);
> > +	return gpiod_get_value(cf->board->det_pin);
> >  }
> 
> a) The change in polarity looks wrong here, I can't really tell
> from the patch. If this is intentional, maybe explain it in
> the changelog. With that addressed (either way)
> 
> Reviewed-by: Arnd Bergmann <arnd@arndb.de>
> 
> 
> b) In case you are doing more patches like this one at the moment,
> note that I'm in the process of removing all unused board files
> for arch/arm/, which will in turn make a lot of drivers unused.
> I should be able to provide a branch soon, which can be used to
> identify drivers that don't have DT support any more and don't
> have any board files. Rather than converting them to gpio
> descriptors, we can probably just remove those drivers.
> 
> c) I'm not sure about the state of the at91_cf driver. Apparently
> we used to have three drivers for the same hardware (pcmcia,
> pata and ide), and only the pcmcia driver remained in the tree
> after drivers/ide/ was removed and pata_at91 did not get converted
> to DT. I think in the long run we will remove the pcmcia layer,
> so if you are actually trying to use this hardware, we may want to
> revive the pata variant and drop this one instead.
> There is no dts file in tree that actually declares either of them,
> so chances are that nobody is actually using the CF slot on at91
> any more.
> 

I'm pretty sure it is broken since eaa9a21dd14b ("pcmcia: at91_cf: Use
syscon to configure the MC/smc") as this change has never been tested.

-- 
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com