[PATCH 1/7] pwm: Add pwm core driver

[Arun MURTHY]

Hi Trilok,

> Hi Arun,
> 
> On Fri, Oct 1, 2010 at 8:55 AM, Arun MURTHY
> <arun.murthy at stericsson.com> wrote:
> > Hi Trilok,
> >
> >> Hi Arun,
> >>
> >> Adding Bill Gatliff (anyway, CC list already crowded)
> >>
> >> On Wed, Sep 29, 2010 at 10:19 AM, Arun MURTHY
> >> <arun.murthy at stericsson.com> wrote:
> >> >> Arun MURTHY wrote:
> >> >> >>>> Shouldn't PWM_DEVICES select HAVE_PWM?
> >> >> >>>
> >> >> >>> No not required, the entire concept is to remove HAVE_PWM and
> >> use
> >> >> >> PWM_CORE.
> >>
> >> There is already nice and clean framework written by Bill for PWM,
> if
> >> you grep the LKML and linux-embedded mailing list archive then you
> >> will get his patches, and it seems that he had promised to send the
> >> updated version few week back, but not heard from him (may be
> because
> >> he was travelling as per FB status).
> >>
> >> Please evaluate that framework too.
> >>
> > Thanks for this information, I did search in linux-embedded mailing
> list
> > archive. Below are my views on that patch set.
> > Many of the functions that has been defined in pwm core driver
> > written by Bill Gatliff is not being used by the most of the pwm
> drivers
> > except Atmel PWM driver. I rather felt the pwm core driver was an
> attempt
> > made to generalize the Atmel pwm driver.
> > And moreover this was posted long back somewhere in the beginning of
> this
> > year i.e Feb and the thread is dead thereafter.
> >
> > This patch has been submitted focusing all the existing pwm drivers
> and
> > only these are the functions that are being used by pwm drivers.
> > This patch set also included patch to align all the existing pwm
> driver
> > with the pwm core driver.
> > So it is an attempt to generalize most of the pwm drivers and
> > conclude with a pwm core driver.
> 
> I don't agree that Bill had only atmel drivers view. The PWM framework
> was discussed in-depth and at that time reviewers also requested once
> to provide more example drivers using these drivers, someone said "we
> atleast need three drivers as rule of thumb". Let's wait until Bill
> reviews your framework, I am sure we don't need to end up the same
> problems faced by Bill while designing that framework in your code
> too.
> 
You can have a look at the pwm_config_nosleep(),pwm_set_polarity(),
pwm_synchronize(),pwm_unsynchronize(), pwm_set_handler() etc.
These are not being used by the exsting pwm drivers except Atmel pwm.
I mean not the functions but the functionality.
PWM is a simple device and most of its clients are controlling intensity
of backlight, leds, vibrator etc.
I don't think these complex functionality are required.
And moreover it also refers to GPIO pins, in that case it comes under
a different classification. The one that I have suggested is a generic
pwm core driver.
You can have a look at the existing pwm drivers in drivers/mfd/twl6030-pwm.c,
arch/arm/plat-samsung/pwm.c, arch/arm/plat-mxc/pwm.c, arch/arm/plat-pxa/pwm.c,
arch/mips/jz4740/pwm.c.
None of these include the function provided the patch " [PWM PATCH 1/5] API
to consolidate PWM devices behind a common user and kernel interface "
except pwm_enable, pwm_config, pwm_disable.
I have focused on all these and come up with this design.
And moreover Bill's patch set for pwm core driver, becomes incompatible with
pwm based backlight and led driver(drivers/leds/leds-pwm.c,
drivers/video/backlight/pwm_bl.c) and drivers/input/misc/pwm-beeper.c.

I don't mind waiting for Bill's review on my patch, but he is not active
since Feb 2010.

Thanks and Regards,
Arun R Murthy
-------------