On Sat 2020-07-11 23:01:11, Ondřej Jirman wrote:
> Hello Pavel,
> 
> On Sat, Jul 11, 2020 at 12:04:09PM +0200, Pavel Machek wrote:
> > Hi!
> > 
> > > Some LED controllers may come with an internal HW triggering mechanism
> > > for the LED and an ability to switch between user control of the LED,
> > > or the internal control. One such example is AXP20X PMIC, that allows
> > > wither for user control of the LED, or for internal control based on
> > > the state of the battery charger.
> > > 
> > > Add support for registering per-LED device trigger.
> > > 
> > > Names of private triggers need to be globally unique, but may clash
> > > with other private triggers. This is enforced during trigger
> > > registration. Developers can register private triggers just like
> > > the normal triggers, by setting private_led to a classdev
> > > of the LED the trigger is associated with.
> > 
> > What about this? Should address Marek's concerns about resource use...
> 
> What concerns? Marek's concerns seem to be about case where we register
> a trigger for (each led * self-working configuration) which I admit
> can be quite a lot of triggers if there are many functions. But that's
> not my proposal.
> 
> My proposal is to only register on trigger per LED at most. So on my
> system that's 1 extra trigger and on Marek's system that'd be 48 new
> triggers. Neither seems like a meaningful problem from resource
> use perspective.

So.. 48 triggers on Marek's systems means I'll not apply your patch.

Please take a look at my version, it is as simple and avoids that
problem.

If it works for you, you can submit it properly and I'll likely accept
it.

Best regards,
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html