Re: [PATCH 0/6] ACPI: EC: Updates related to initialization

["Rafael J. Wysocki" <rafael@kernel.org>]

On Tue, Mar 3, 2020 at 8:29 AM Jian-Hong Pan <jian-hong@endlessm.com> wrote:
>
> Rafael J. Wysocki <rafael@kernel.org> 於 2020年3月2日 週一 下午7:45寫道：
> >
> > On Mon, Mar 2, 2020 at 11:38 AM Rafael J. Wysocki <rafael@kernel.org> wrote:
> > >
> > > On Mon, Mar 2, 2020 at 6:54 AM Jian-Hong Pan <jian-hong@endlessm.com> wrote:
> > > >
> > > > Daniel Drake <drake@endlessm.com> 於 2020年2月28日 週五 下午5:43寫道：
> > > > >
> > > > > On Thu, Feb 27, 2020 at 10:25 PM Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> > > > > > The purpose of this series of update of the ACPI EC driver is to make its
> > > > > > initialization more straightforward.
> > > > > >
> > > > > > They fix a couple of issues, clean up some things, remove redundant code etc.
> > > > > >
> > > > > > Please refer to the changelogs of individual patches for details.
> > > > > >
> > > > > > For easier access, the series is available in the git branch at
> > > > > >
> > > > > >  git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git \
> > > > > >  acpi-ec-work
> > > > > >
> > > > > > on top of 5.6-rc3.
> > > > >
> > > > > Jian-Hong, can you please test this on Asus UX434DA?
> > > > > Check if the screen brightness hotkeys are still working after these changes.
> > > >
> > > > Hi Rafael,
> > > >
> > > > Thanks for your patches, but we found an issue:
> > > > The laptops like ASUS UX434DA's screen brightness hotkeys work before
> > > > this patch series.  However, the hotkeys are failed with the patch
> > > > "ACPI: EC: Unify handling of event handler installation failures".
> > >
> > > So I have modified the series to avoid the change that can possibly break this.
> > >
> > > Can you please pull the new series from
> > >
> > >  git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git \
> > >  acpi-ec-work
> > >
> > > (same branch) and retest?
> >
> > Note that the current top-most commit in that branch is
> >
> > 0957d98f50da ACPI: EC: Consolidate event handler installation code
>
> I tested the commits in acpi-ec-work branch whose last commit is
> 0957d98f50da ("ACPI: EC: Consolidate event handler installation
> code").  The screen brightness hotkeys are still failed with
> 0957d98f50da ("ACPI: EC: Consolidate event handler installation
> code").
>
> I tweak and add some debug messages:
>
> diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
> index 85f1fe8e208a..3887f427283c 100644
> --- a/drivers/acpi/ec.c
> +++ b/drivers/acpi/ec.c
> @@ -1443,23 +1443,27 @@ static bool install_gpe_event_handler(struct
> acpi_ec *ec)
>         return true;
>  }
>
> -static bool install_gpio_irq_event_handler(struct acpi_ec *ec,
> +static int install_gpio_irq_event_handler(struct acpi_ec *ec,
>                                            struct acpi_device *device)
>  {
>         int irq, ret;
>
>         /* ACPI reduced hardware platforms use a GpioInt specified in _CRS. */
>         irq = acpi_dev_gpio_irq_get(device, 0);
> -       if (irq < 0)
> -               return false;
> +       if (irq < 0) {
> +               pr_err("%s: acpi_dev_gpio_irq_get returns %d\n", __func__, irq);
> +               return irq;
> +       }
>
>         ret = request_irq(irq, acpi_ec_irq_handler, IRQF_SHARED, "ACPI EC", ec);
> -       if (ret < 0)
> -               return false;
> +       if (ret < 0) {
> +               pr_err("%s: request_irq returns %d\n", __func__, ret);
> +               return ret;
> +       }
>
>         ec->irq = irq;
>
> -       return true;
> +       return 0;
>  }
>
>  /**
> @@ -1517,9 +1521,11 @@ static int ec_install_handlers(struct acpi_ec
> *ec, struct acpi_device *device)
>                          * fatal, because the EC can be polled for events.
>                          */
>                 } else {
> -                       ready = install_gpio_irq_event_handler(ec, device);
> -                       if (!ready)
> -                               return -ENXIO;
> +                       pr_err("%s: install_gpio_irq_event_handler\n",
> __func__);
> +                       int ret = install_gpio_irq_event_handler(ec, device);
> +                       if (ret)
> +                               return ret;
> +                       ready = true;
>                 }
>                 if (ready) {
>                         set_bit(EC_FLAGS_EVENT_HANDLER_INSTALLED, &ec->flags);
>
> The dmesg shows:
>
> [    0.121117] ACPI: EC: ec_install_handlers: install_gpio_irq_event_handler
> [    0.121133] ACPI: EC: install_gpio_irq_event_handler:
> acpi_dev_gpio_irq_get returns -517
>
> Originally, ec_install_handlers() will return the returned value from
> install_gpio_irq_event_handler() from acpi_dev_gpio_irq_get(), which
> is -EPROBE_DEFER, instead of -ENXIO.  However, ec_install_handlers()
> returns -ENXIO directly if install_gpio_irq_event_handler() returns
> false in patch ("ACPI: EC: Consolidate event handler installation
> code").  Here needs some modification.

Thanks, I forgot about the -EPROBE_DEFER case.