[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Problems with hwclock on Mac IIvx.



On Tue, Nov 19, 2002 at 11:09:45PM -0500, Ray Knight wrote:
> On Tue, 2002-11-19 at 09:37, Chris Tillman wrote:
> > On Tue, Nov 19, 2002 at 01:16:15AM -0500, Ray Knight wrote:
> > > I just did a clean install of woody on my Mac IIvx with a DayStar Turbo
> > > 040 accelerator.  This system has been running potato for over a year
> > > (not continuously due to power outages) with no problems.  The install
> > > went fine, but the system locked up after the first reboot at the first
> > > call to hwclock.  I rebooted the install and commented out the calls to
> > > hwclock in rcS.d.  The system now boots and runs fine.  Entering hwclock
> > > at the prompt locks the system hard.  The potato version of hwclock did
> > > not have this problem.  I'm running the same kernel version as I was
> > > with potato.  Does anyone know what changed in hwclock between Debian
> > > 2.2 and 3.0?
> > > 
> > > Ray
> > > 
> > 
> > There's some info in 
> > 
> > lists.debian.org/debian-68k/2002/debian-68k-200209/msg00083.html
> > 
> 
> Which is a not found URL?  I've already searched the list archives and
> the only mention of hwclock problems is the known problem with the
> Quadra 950.  This is a IIvx.  The problem does not exist with potato and
> the exact same kernel.  Therefore I believe it is a change in the
> hwclock code.

Oops, that's what I get for retyping URLs. s/200209/200208/

Well, that does make sense if it's the same kernel. 

But, here's what I was looking at. It sounded possibly relevant, maybe not.


   #index top up prev next
     _________________________________________________________________________________________________________________

   [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

                                                      Re: clock problems
     _________________________________________________________________________________________________________________

     * To: "R.A." <funkaster@directo.cl>
     * Subject: Re: clock problems
     * From: flar@pants.nu (Brad Boyer)
     * Date: Thu, 15 Aug 2002 14:11:41 -0700
     * Cc: debian-68k@lists.debian.org
     * In-reply-to: <p05100300b981c78a613e@[192.168.0.3]>
     * References: <p05100300b981c78a613e@[192.168.0.3]>
     * User-agent: Mutt/1.3.28i
     _________________________________________________________________________________________________________________

On Thu, Aug 15, 2002 at 05:11:13PM -0400, R.A. wrote:
> I'm having problems with the clock: I'm experiencing an offset of
> about 5-10 minutes per day. My comp is a Quadra 950 with debian woody
> (3.0r0). At install time I disabled the hwclock scripts, becouse
> otherwise it won't boot. Does this has to do with the problem? (it
> doesn't matter if I leave it on or off). Maybe it's the battery...
> any ideas?

>From my understanding, the 68k Macs have a habit of losing timer
interrupts. This causes the clock to lose a lot of time. I just
run ntpd to keep mine on the right time. Of course, that requires
a permanent connection via network to something with a more
stable clock...

> PS: where can I find the source code of hwclock? I would like to take
> a look at it, to see if I can fix it...

I believe the problem is in the kernel. I seem to recall that
the clock routines in the kernel don't support every model. Try
looking at the function mac_hwclk in arch/m68k/mac/config.c in
2.2 kernels, or in arch/m68k/mac/misc.c in 2.4 kernels. It looks
like it should treat a Q950 like the Mac IIfx for clock routines,
which may or may not be correct. It decides what to do based on
the ADB controller... It does work for reading the clock on my
IIfx with a recent kernel. Writing the clock breaks everything.

        Brad Boyer
        flar@allandria.com


-- 
*----------------------------------------------------------------*
|            Chris Tillman        tillman@voicetrak.com          |
|                  To HAVE, GIVE all TO all (ACIM)               |
*----------------------------------------------------------------*



Reply to: