On Sun, Apr 21, 2019 at 10:55:07PM +0200, Mark Kettenis wrote:
> > Date: Sun, 21 Apr 2019 21:25:25 +0200
> > From: Marcus Glocker <[email protected]>
> >
> > I found a second thermal sensor INT3403 in my X1 which didn't attach
> > to acpitz(4) because its type returns AML_OBJTYPE_DEVICE instead of
> > AML_OBJTYPE_THERMZONE.
> >
> > When matching it explicitly it attaches to acpitz1 and seems to work
> > fine (I did some load tests to see it scaling up and down):
> >
> > hw.sensors.acpitz0.temp0=54.00 degC (zone temperature)
> > -> hw.sensors.acpitz1.temp0=41.00 degC (zone temperature)
> >
> > Can we add acpi_matchhids() for such sensors?
>
> This INT3403 device is part of Intel's DPTF and I don't think it makes
> sense to have acpitz(4) handle this. Looking at the relevant Linux
> code the DPTF stuff is more complicated.
I see.
> Maybe we should add a separate driver for this?
Makes sense.
I'll have a glance over what is around about DPTF, but probably I'll
run out of time soon again ...
> > Index: sys/dev/acpi/acpitz.c
> > ===================================================================
> > RCS file: /cvs/src/sys/dev/acpi/acpitz.c,v
> > retrieving revision 1.54
> > diff -u -p -r1.54 acpitz.c
> > --- sys/dev/acpi/acpitz.c 29 Jun 2018 17:39:18 -0000 1.54
> > +++ sys/dev/acpi/acpitz.c 21 Apr 2019 19:12:10 -0000
> > @@ -76,6 +76,11 @@ struct cfdriver acpitz_cd = {
> > NULL, "acpitz", DV_DULL
> > };
> >
> > +const char *acpitz_hids[] = {
> > + "INT3403",
> > + NULL
> > +};
> > +
> > void acpitz_init_perf(void *);
> > void acpitz_setperf(int);
> > void acpitz_monitor(struct acpitz_softc *);
> > @@ -171,6 +176,13 @@ acpitz_match(struct device *parent, void
> > {
> > struct acpi_attach_args *aa = aux;
> > struct cfdata *cf = match;
> > +
> > + /*
> > + * Some thermal sensors don't match to AML_OBJTYPE_THERMZONE.
> > + * First try to match them explicitly therefore.
> > + */
> > + if (acpi_matchhids(aa, acpitz_hids, cf->cf_driver->cd_name))
> > + return (1);
> >
> > /* sanity */
> > if (aa->aaa_name == NULL ||
