On Sun, Feb 20, 2000 at 07:54:33PM +0100, Guest section DW wrote:
> On Sun, Feb 20, 2000 at 06:21:57PM +0100, Ralf Baechle wrote:
>
> > > hwclock tries to use /dev/rtc to read and set the hardware clock.
> > >
> > > [Of course the idea is that if the kernel provides this access
> > > then a userspace program need not try to access the hardware itself.
> > > Today there are still machines where "hwclock --directisa" works
> > > but "hwclock" fails, so the kernel code for rtc is not perfect yet.]
> >
> > I'd actually like to see --directisa support to be removed from hwclock or
> > at least to be a #ifdef __i386__. The last time I had to hack hwclock for
> > the n+1th MIPS port it was much more portable than it's predecessor clock
> > but the direct port accesses still make it somewhat messy.
> >
> > What is the problem with /dev/rtc on those machines where hwclock fails?
>
> It fails :-)
>
> The symptoms vary. Sometimes it just doesnt work.
> Sometimes it causes a kernel hang.
Well, after a kernel crash, you'll "only" have to wait for fsck to finish.
But it is no fun at all to come up in 2055 (or so) because hwclock's
fallback method (which is to access the ISA bus, I think) fails (no RTC..).
After this, "make" will fail, accounting systems report very strange times
and so on... That's really not nice;( I think for these reasons, the
RTC drivers should be fixed. Some architectures use different RTCs. In this
case, I think that a module is the best solution since you can supply some
options;) But unfortunately module support (at least for Alpha?) seems
to be falling off the desk right now;(
MfG, JBG
-- Fehler eingestehen, Größe zeigen: Nehmt die Rechtschreibreform zurück!!! keyID=0x8399E1BB fingerprint=250D 3BCF 7127 0D8C A444 A961 1DBD 5E75 8399 E1BB
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Wed Feb 23 2000 - 21:00:25 EST