Re: [PATCH] rtc: rtc-stmp3xxx: detect power failure on backup power domain

[Michael Thalmeier <michael.thalmeier@hale.at>]

On Thu, May 18, 2017 at 05:56:31PM +0200, Alexandre Belloni wrote:
> Hi,
> 
> On 18/05/2017 at 16:45:21 +0200, Michael Thalmeier wrote:
> > To detect when the backup power domain has lost power a software defined bit
> > is set in one of the general purpose persistent registers when writing a new
> > time into the rtc.
> > When reading the time this bit is checked to determine if a power fail has
> > happened since the last time the rtc time was set.
> > 
> 
> I'm kind of concerned that other people may want to use those register
> for something else but I don't currently have anything better to
> suggest.
> 
> The other concern is that when updating the kernel, this will make amm
> the rtc report that the time is invalid until the next update. So this
> should be disabled by default.

How would you think this is best to disable by default? With a config
option or with a module parameter?
Either way would also allow to make the bit mask in the PERSISTENT2
register configurable, so one can use whichever bit is still unused by
some other applications.

> 
> > When we detect a power fail we return -ENODATA.
> > 
> 
> All the other drivers return -EINVAL in that case.
> 
> > Signed-off-by: Michael Thalmeier <michael.thalmeier@hale.at>
> > ---
> >  drivers/rtc/rtc-stmp3xxx.c | 9 +++++++++
> >  1 file changed, 9 insertions(+)
> > 
> > diff --git a/drivers/rtc/rtc-stmp3xxx.c b/drivers/rtc/rtc-stmp3xxx.c
> > index d578e40..51330ec 100644
> > --- a/drivers/rtc/rtc-stmp3xxx.c
> > +++ b/drivers/rtc/rtc-stmp3xxx.c
> > @@ -62,6 +62,9 @@
> >  /* missing bitmask in headers */
> >  #define STMP3XXX_RTC_PERSISTENT1_FORCE_UPDATER	0x80000000
> >  
> > +#define STMP3XXX_RTC_PERSISTENT2		0x80
> > +#define STMP3XXX_RTC_PERSISTENT2_VALID_TIME	0x01
> > +
> >  struct stmp3xxx_rtc_data {
> >  	struct rtc_device *rtc;
> >  	void __iomem *io;
> > @@ -160,6 +163,10 @@ static int stmp3xxx_rtc_gettime(struct device *dev, struct rtc_time *rtc_tm)
> >  	if (ret)
> >  		return ret;
> >  
> > +	if (!(readl(rtc_data->io + STMP3XXX_RTC_PERSISTENT2) &
> > +				STMP3XXX_RTC_PERSISTENT2_VALID_TIME))
> > +		return -ENODATA;
> > +
> >  	rtc_time_to_tm(readl(rtc_data->io + STMP3XXX_RTC_SECONDS), rtc_tm);
> >  	return 0;
> >  }
> > @@ -169,6 +176,8 @@ static int stmp3xxx_rtc_set_mmss(struct device *dev, unsigned long t)
> >  	struct stmp3xxx_rtc_data *rtc_data = dev_get_drvdata(dev);
> >  
> >  	writel(t, rtc_data->io + STMP3XXX_RTC_SECONDS);
> > +	writel(STMP3XXX_RTC_PERSISTENT2_VALID_TIME,
> > +		rtc_data->io + STMP3XXX_RTC_PERSISTENT2 + STMP_OFFSET_REG_SET);
> >  	return stmp3xxx_wait_time(rtc_data);
> >  }
> >  
> > -- 
> > 2.9.2
> > 
> 
> -- 
> Alexandre Belloni, Free Electrons
> Embedded Linux and Kernel engineering
> http://free-electrons.com

-- 

Michael Thalmeier
(Entwicklung)
HALE electronic GmbH
Eugen-Müller-Straße 18, 5020 Salzburg, Austria
Tel: +43 (662) 439011 0
Fax: +43 (662) 439011 9
michael.thalmeier@hale.at
Firmenbuchnummer: FN 66801m HG Salzburg