lkml.org 
[lkml]   [2020]   [Oct]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] pwm: sysfs: Set class on pwm devices
On Thu, Oct 01, 2020 at 01:24:49PM +0200, Greg Kroah-Hartman wrote:
> On Thu, Oct 01, 2020 at 11:05:31AM +0200, Lars Poeschel wrote:
> > On Wed, Sep 30, 2020 at 11:41:46AM +0200, Uwe Kleine-König wrote:
> > > Hello,
> > >
> > > I added Greg Kroah-Hartman who I discussed this with via irc a bit to
> > > Cc:.
> > >
> > > On Wed, Sep 30, 2020 at 11:20:56AM +0200, Lars Poeschel wrote:
> > > > thank you for your review!
> > > >
> > > > On Wed, Sep 30, 2020 at 08:57:26AM +0200, Uwe Kleine-König wrote:
> > > > > On Tue, Sep 29, 2020 at 02:19:53PM +0200, poeschel@lemonage.de wrote:
> > > > > > From: Lars Poeschel <poeschel@lemonage.de>
> > > > > >
> > > > > > This adds a class to exported pwm devices.
> > > > > > Exporting a pwm through sysfs did not yield udev events. The
> > > > >
> > > > > I wonder what is your use-case here. This for sure also has a place to
> > > > > be mentioned in the commit log. I suspect there is a better way to
> > > > > accomplish you way.
> > > >
> > > > Use-case is to be able to use a pwm from a non-root userspace process.
> > > > I use udev rules to adjust permissions.
> > >
> > > Hmm, how do you trigger the export? Without being aware of all the
> > > details in the sysfs code I would expect that the exported stuff is
> > > available instantly once the write used to export the PWM is completed.
> > > So changing the permissions can be done directly after triggering the
> > > export in the same process.
> >
> > The export is triggered through the userspace process itself. Why can it
> > do this ? Because there is another udev rule, that changes permissions
> > when a pwmchip appears.
> > Then I'd like to have the second udev rule, that changes permissions on
> > the freshly exported pwm. The userspace process can't do this.
> > You are right I could propably do everything from within udev: If a
> > pwmchip appears, export certain pwms and right away change their
> > permissions. It does not also not feel right. It'd require knowledge
> > from the userspace application to be mapped to udev.
>
> The way the kernel code is now, yes, you will not have any way to
> trigger it by userspace as the kernel is creating a "raw" struct device
> that isn't assigned to anything. That is what needs to be fixed here.
>
> > > Out of interest: What do you use the pwm for? Isn't there a suitable
> > > kernel driver that can do the required stuff? Compared to the kernel-API
> > > the sysfs interface isn't atomic. Is this an annoyance?
> >
> > Use-case is generating a voltage from the pwm. This voltage is used to
> > signal different states and does not change very often. This is
> > absolutely not annoying that this is not atomic. We just change the duty
> > cycle on the fly. Everything else is configured one time at startup.
> > I'd call what I need pwm-dac. I could not find a ready to use driver.
> > Maybe I could misuse some kernel driver for this. Maybe I could use
> > pwm-led or pwm-brightness or pwm-fan. Propably pwm-regulator could work,
> > there is even a userspace facing part for this, but this is not
> > devicetree ready.
> > ...and the worst, please don't blame me: The application is java, so
> > ioctl is a problem.
>
> I thought java could do ioctls, otherwise how would it ever be able to
> talk to serial ports?
>
> Anyway, this needs to be fixed in the kernel...

If atomicity was a problem, we could potentially add a mechanism to the
sysfs interface to enable that. I don't see a good way of doing that in
a single file, since that works against how sysfs is designed. But one
thing I could imagine is adding a file ("lock", or whatever you want to
call it) that you can use for atomic fencing:

$ echo 1 > lock # locks the hardware state
$ echo 100 > period
$ echo 50 > duty_cycle
$ echo 0 > lock # flushes state to hardware

But it sounds like that's not even a big issue.

However, given the use-case description, it sounds to me like
pwm-regulator would be a better candidate to solve this, because it's
purpose is literally to generate a voltage using a PWM. There is a
device tree binding for pwm-regulator, so this should work there as
well.

Lars, what exactly are the problems that you're running into when trying
to use pwm-regulator with device tree?

Thierry
[unhandled content-type:application/pgp-signature]
\
 
 \ /
  Last update: 2020-10-05 11:31    [W:0.111 / U:2.528 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site