On Sunday 01 April 2012, Pedro de Oliveira wrote:
> By reading the DS1687 datasheet I managed to fix it. A diff follows:
>
> # cvs diff -u sys/dev/ic/ds1687reg.h sys/arch/sgi/dev/dsrtc.c
> Index: sys/dev/ic/ds1687reg.h
> ===================================================================
> RCS file: /cvs/src/sys/dev/ic/ds1687reg.h,v
> retrieving revision 1.2
> diff -u -r1.2 ds1687reg.h
> --- sys/dev/ic/ds1687reg.h 31 Mar 2008 07:14:00 -0000 1.2
> +++ sys/dev/ic/ds1687reg.h 1 Apr 2012 00:59:52 -0000
> @@ -62,3 +62,5 @@
> #define DS1687_EXT_CTRL 0x4a /* Extended control
> register. */ #define DS1687_KICKSTART 0x01 /* Kickstart flag. */
>
> +#define DS1687_EXT_CTRL_B 0x4b /* Extended control register B. */
> +#define DS1687_ALRM_ENABLE 0x02 /* Bit 1 - WIE */
> Index: sys/arch/sgi/dev/dsrtc.c
> ===================================================================
> RCS file: /cvs/src/sys/arch/sgi/dev/dsrtc.c,v
> retrieving revision 1.11
> diff -u -r1.11 dsrtc.c
> --- sys/arch/sgi/dev/dsrtc.c 7 Nov 2009 14:49:01 -0000 1.11
> +++ sys/arch/sgi/dev/dsrtc.c 1 Apr 2012 00:59:52 -0000
> @@ -315,7 +315,7 @@
> ds1687_set(void *v, struct tod_time *ct)
> {
> struct dsrtc_softc *sc = v;
> - int year, century, ctrl, dm;
> + int year, century, ctrl, dm, alarm;
>
> century = ct->year / 100 + 19;
> year = ct->year % 100;
> @@ -342,6 +342,16 @@
> (*sc->write)(sc, DS1687_MONTH, tobcd(ct->mon, dm));
> (*sc->write)(sc, DS1687_YEAR, tobcd(year, dm));
> (*sc->write)(sc, DS1687_CENTURY, tobcd(century, dm));
> +
> + /* Reset the RTC alarm. */
> + (*sc->write)(sc, DS1687_SEC_ALRM, 0);
> + (*sc->write)(sc, DS1687_MIN_ALRM, 0);
> + (*sc->write)(sc, DS1687_HOUR_ALRM, 0);
> + (*sc->write)(sc, DS1687_DATE_ALRM, 0);
> +
> + /* Disable the alarm wakeup. */
> + alarm = (*sc->read)(sc, DS1687_EXT_CTRL_B);
> + (*sc->write)(sc, DS1687_EXT_CTRL_B, alarm & ~DS1687_ALARM_ENABLE);
Nice - I figured this was the underlying cause. It should be sufficient just
to disable the alarm without touching the alarm time registers. However, I am
curious as to how the alarm got enabled in the first place (Irix?) and if so,
should OpenBSD be touching/undoing it?
> /* Enable updates. */
> (*sc->write)(sc, DS1687_CTRL_B, ctrl);
> #
>
> On Sun, Mar 25, 2012 at 9:00 PM, Miod Vallat <m...@online.fr> wrote:
> >> But a few days ago I installed OpenBSD/sgi 5.0 (also tried the snapshot,
> >> but it crashed) and every day since I installed it, every day at
> >> midnight the machine automaticly powers on.
> >>
> >> Does this happen to anyone else? How can I disable this?
> >
> > This has happened to me a few years ago on a low-end R5000SC@180 O2, but
> > eventually stopped after leaving the machine unplugged for a while
> > (read: months). There is probably something slightly wrong in the way
> > the Dallas clock registers are updated by OpenBSD, but I have not been
> > able to find it so far.
> >
> > Miod
--
"Reason is not automatic. Those who deny it cannot be conquered by it.
Do not count on them. Leave them alone." -- Ayn Rand
