[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20130528110011.GC11547@mithrandir>
Date: Tue, 28 May 2013 13:00:12 +0200
From: Thierry Reding <thierry.reding@...il.com>
To: Arnd Bergmann <arnd@...db.de>
Cc: H Hartley Sweeten <hartleys@...ionengravers.com>,
Linux Kernel <linux-kernel@...r.kernel.org>,
"gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
Ryan Mallon <rmallon@...il.com>,
"mcrapet@...il.com" <mcrapet@...il.com>,
Lars Poeschel <poeschel@...onage.de>
Subject: Re: [PATCH 00/14] misc/ep93xx_pwm: cleanup driver for conversion to
PWM framework
On Mon, May 27, 2013 at 07:12:08PM +0200, Arnd Bergmann wrote:
> On Monday 27 May 2013, H Hartley Sweeten wrote:
> > Ryan Mallon has also provided a Reviewed-by for this series.
> >
> > Will you be the one that merges this? I would like it to be in linux-next
> > before I convert it to the PWM framework.
>
> While Greg and I are both maintainers for drivers/misc, he is the one who
> actually has a git tree for it, so he would merge it.
>
> However, I think it would be better to just merge it all through the pwm
> tree. Your current series is good, and with my Ack I see no problem to
> just do the conversion to pwm on top and send a pull request for all of
> it to Thierry.
While I have no objection to taking the patches through the PWM tree, I
prefer to pick up patches from my inbox. I should probably get used to
taking pull requests to make my life easier, but there was never a need
so far. So if you'll be pushing the series to a tree anyway I can take a
look at it and see if I feel comfortable pulling, but I'd still want to
see the full series as patches for easier review.
> > Also, I have a question about the conversion.
> >
> > If I strip the sysfs support out of this driver the conversion is quite simple.
> > But, my use for this driver requires user space control of the PWM.
> >
> > Should I:
> > 1) convert the driver to the PWM framework and leave the sysfs stuff in it
> > 2) work out a generic sysfs support for the PWM framework and then
> > convert the driver
> > 3) other...
> >
> > I've been looking at 2) by doing something like how gpiolib does it. Do
> > you think that would be acceptable?
>
> That would be for Thierry to decide. It does sound better to me than the 1)
> and I don't have a better idea for 3).
>
> I wonder how the arbitration between in-kernel and user-space consumers
> of the pwm lines would work though.
I've added Lars Poeschel on Cc, who's done some work on a sysfs
interface for the PWM subsystem already. It's undergone some review
already[0] and I think he's working on a v2 now.
Thierry
[0]: http://marc.info/?l=linux-kernel&m=136499756101273&w=2
Content of type "application/pgp-signature" skipped
Powered by blists - more mailing lists