Re: [PATCH v2 07/11] gpiolib: rework the locking mechanism for lineevent kfifo

[Andy Shevchenko]

On Thu, Dec 5, 2019 at 11:31 AM Bartosz Golaszewski
<bgolaszewski@xxxxxxxxxxxx> wrote:
> År., 4 gru 2019 o 23:25 Andy Shevchenko <andy.shevchenko@xxxxxxxxx> napisaÅ(a):
> > On Wed, Dec 4, 2019 at 6:01 PM Bartosz Golaszewski <brgl@xxxxxxxx> wrote:

> > > +       spin_lock(&le->wait.lock);
> > >         if (!kfifo_is_empty(&le->events))
> > >                 events = EPOLLIN | EPOLLRDNORM;
> > > +       spin_unlock(&le->wait.lock);
> >
> > Sound like a candidate to have kfifo_is_empty_spinlocked().
>
> Yeah, I noticed but I thought I'd just add it later separately - it's
> always easier to merge self-contained series.

...and easier to forget about.
But it's up to you :-)

> > >         struct lineevent_state *le = filep->private_data;
> > > -       unsigned int copied;
> > > +       struct gpioevent_data event;
> > >         int ret;
> >
> > > +       if (count < sizeof(event))
> > >                 return -EINVAL;
> >
> > This still has an issue with compatible syscalls. See patch I have
> > sent recently.
> > I dunno how you see is the better way: a) apply mine and rebase your
> > series, or b) otherwise.
> > I can do b) if you think it shouldn't be backported.
> >
>
> Looking at your patch it seems to me it's best to rebase yours on top
> of this one - where I simply do copy_to_user() we can add a special
> case for 32-bit user-space. I can try to do this myself for v3 if you
> agree.

Yea, I'm fine with it.

-- 
With Best Regards,
Andy Shevchenko