Re: [PATCH V5 05/12] ACPI: check and enumerate CMOS RTC devices explicitly

["Rafael J. Wysocki" <rjw@rjwysocki.net>]

On Monday, April 28, 2014 12:17:49 AM Rafael J. Wysocki wrote:
> On Tuesday, April 08, 2014 12:06:52 AM Zhang Rui wrote:
> > For CMOS RTC devices, the acpi pnp scan handler does not work because
> > there is already a cmos rtc scan handler installed, thus we need to
> > check those devices and enumerate them to PNP bus explicitly.
> > Plus, the cmos rtc scan handler needs to return 1 so that it will not
> > be enumerated to platform bus.
> > 
> > Note: the CMOS RTC device id is not removed from ACPI pnp scan handler
> >       id list, thus, if cmos rtc scan handler is compiled out, the ACPI
> >       pnp scan handler will be attached to the CMOS RTC devices instead,
> >       to prevent these devices from being created to platform bus.
> > 
> > Signed-off-by: Zhang Rui <rui.zhang@intel.com>
> > ---
> >  drivers/acpi/acpi_cmos_rtc.c |    2 +-
> >  drivers/acpi/acpi_pnp.c      |   22 +++++++++++++++++++++-
> >  2 files changed, 22 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/acpi/acpi_cmos_rtc.c b/drivers/acpi/acpi_cmos_rtc.c
> > index 961b45d..2da8660 100644
> > --- a/drivers/acpi/acpi_cmos_rtc.c
> > +++ b/drivers/acpi/acpi_cmos_rtc.c
> > @@ -68,7 +68,7 @@ static int acpi_install_cmos_rtc_space_handler(struct acpi_device *adev,
> >  		return -ENODEV;
> >  	}
> >  
> > -	return 0;
> > +	return 1;
> >  }
> >  
> >  static void acpi_remove_cmos_rtc_space_handler(struct acpi_device *adev)
> > diff --git a/drivers/acpi/acpi_pnp.c b/drivers/acpi/acpi_pnp.c
> > index 8cf7d9d..387b150 100644
> > --- a/drivers/acpi/acpi_pnp.c
> > +++ b/drivers/acpi/acpi_pnp.c
> > @@ -349,9 +349,29 @@ static struct acpi_scan_handler acpi_pnp_handler = {
> >  	.attach = acpi_pnp_scan_handler_attach,
> >  };
> >  
> > +/*
> > + * For CMOS RTC devices, the acpi pnp spcan handler does not work because
> > + * there is already a cmos rtc scan handler installed, thus we need to
> > + * check those devices and enumerate them to PNP bus explicitly.
> > + */
> > +static int is_cmos_rtc_device(struct acpi_device *adev)
> > +{
> > +	struct acpi_device_id ids[] = {
> > +		{ "PNP0B00" },
> > +		{ "PNP0B01" },
> > +		{ "PNP0B02" },
> > +		{""},
> > +	};
> > +	return !acpi_match_device_ids(adev, ids);
> > +}
> > +
> >  bool acpi_is_pnp_device(struct acpi_device *device)
> >  {
> > -	return device->handler == &acpi_pnp_handler;
> > +	if (device->handler == &acpi_pnp_handler)
> > +		return true;
> > +	if (is_cmos_rtc_device(device))
> > +		return true;
> > +	return false;
> 
> What about doing
> 
> 	return device->handler == &acpi_pnp_handler || is_cmos_rtc_device(device);
> 
> instead?
> 
> >  }
> >  EXPORT_SYMBOL_GPL(acpi_is_pnp_device);

Moreover, it looks like this patch should be folded into [2/12] to avoid
breaking bisection.

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.