Re: [PATCH 1/2] irq/irq_sim: provide irq_sim_fire_edge()

From: Bartosz Golaszewski
Date: Mon Dec 03 2018 - 05:23:55 EST


niedz., 2 gru 2018 o 23:20 Bartosz Golaszewski <brgl@xxxxxxxx> napisaÅ(a):
>
> niedz., 2 gru 2018 o 22:56 Uwe Kleine-KÃnig
> <u.kleine-koenig@xxxxxxxxxxxxxx> napisaÅ(a):
> >
> > Hello,
> >
> > On Thu, Nov 29, 2018 at 07:14:45PM +0100, Bartosz Golaszewski wrote:
> > > We're getting too much into details of how to handle simulated
> > > interrupts and we can continue discussing it, but meanwhile I'd like
> > > to address a different thing:
> > >
> > > Thomas, Linus: after commit fa38869b0161 ("gpiolib: Don't support irq
> > > sharing for userspace") some libgpiod tests are failing because we can
> > > no longer depend on reading the value of a dummy GPIO after detecting
> > > an interrupt to know the edge of the interrupt. While these interrupts
> > > are triggered from debugfs and debugfs is not required to maintain
> > > compatibility, I thing having a working test suite for the GPIO
> > > subsystem and uAPI is worth applying these two patches and also the
> > > previous one[1].
> > >
> > > Can we have them applied for 4.20 or are there any objections?
> >
> > Just for the record: I objected the patch, Bartosz agrees to discuss
> > further and but because this is too much detail the patch should now be
> > applied anyhow to fix the test suite of an external project. This seems
> > wrong to me.
> >
>
> Just to look at it from a different perspective: we have a project
> whose tests rely on a behavior that was changed by Uwe's patch. While
> the patch is fine, we need to find a correct way of testing the GPIO
> user API. This may take a long time. In order to not break the tests
> of an external project in 4.20 I propose to patch the interrupt
> simulator (a component only used for testing) for now and to revisit
> it later without time pressure.
>
> Best regards,
> Bartosz

In fact after re-reading this conversation I'm still not sure what
your objection is exactly. You're proposing a solution that may well
be nicely engineered but it's specific to your gpio-simulator.
Meanwhile I'm trying to provide a more generalized API for more
testing modules to use.

Why exactly would you not merge this fix?

Bart