Re: [PATCH RESEND] pinctrl: intel: Turn Baytrail support to tristate

[Takashi Iwai <tiwai@suse.de>]

On Thu, 26 Jan 2017 10:19:31 +0100,
Mika Westerberg wrote:
> 
> On Thu, Jan 26, 2017 at 09:55:36AM +0100, Linus Walleij wrote:
> > On Wed, Jan 25, 2017 at 4:26 PM, Jean Delvare <jdelvare@suse.de> wrote:
> > 
> > > The pinctrl-baytrail driver builds just fine as a module so give
> > > users this option.
> > >
> > > Signed-off-by: Jean Delvare <jdelvare@suse.de>
> > > Cc: Mika Westerberg <mika.westerberg@linux.intel.com>
> > > Cc: Heikki Krogerus <heikki.krogerus@linux.intel.com>
> > > Cc: Linus Walleij <linus.walleij@linaro.org>
> > > ---
> > > This was discussed almost one year ago, with no clear conclusion, but
> > > also no evidence that the driver can't be built as a module. Is there
> > > any way to push this forward?
> > 
> > I see ACKs for this patch, but in my git I also have:
> > 
> > commit 360943a8d26265825025b88da32961bd9ad4f7c6
> > pinctrl: baytrail: make it explicitly non-modular
> > 
> > Acked by Mika.
> 
> Heh, yeah we even removed possibility to unbind the driver with that
> commit. Totally forgot that one.
> 
> > So which one is it going to be?
> 
> Good question. I'm fine with both but I would really like to get some
> confirmation that turning the driver to module actually does not break
> anything.

I guess it would break things on some machines if the module loading
order isn't setup properly.  For example, it's known that
pinctrl-cherrytrail breaks MMC or others if it's loaded too lately.
On distros, we often work around it by a specific module loading order
in initrd.

But this doesn't mean that the modularization itself is wrong.  It's
merely a setup issue.


thanks,

Takashi