Re: [XenPPC] continuous reprogram_timer[00] messages

[Eric V Van hensbergen <ericvanhensbergen@xxxxxxxxxx>]

Jimi Xenidis <jimix@xxxxxxxxxxxxxx> wrote on 03/31/2006 09:35:30 AM:
> > Jimi, this fix works for me.  I simply cvs updated the run/gpul
> > directory of my mambo.
>
> Be aware that this "fix" change to the timebase frequency makes the
> Linux function calibrate_delay() take a minutes, if fact I'm still
> waiting for it as I write this email.
> so you may be stuck at:
>    Memory: 49532k/65536k available (4112k kernel code, 15936k
> reserved, 1480k data, 451k bss, 224k init)
> in the console output for a very long time.
> I assume the systemsim-linux patch has a workaround for this.
> Eric?
>

No, all the bogus boot stuff was taken out.  Bogus Halt doesn't help
you here because its too soon. The right thing to do is set the timebase
divider properly (in mambo, not just the devtree) - then you can set it
to whatever you want (1/8, 1/64, whatever) -- that should resolve the
problem.  I didn't do it in the tcl becuase I was hoping Peterson was
going to set the default to be the right thing from book 4 (1/8) instead
of 1/1.

You need to add a:
myconf config processor timebase_frequency 1/8 # (for the book 4 default)
-or-
myconf config processor timebase_frequency 1/60 # (to make it look like a
G5)
-or-
myconf config processor timebase_frequency 1/512 # (to make it look like
jimi's old maple configuration)

This will handle setting up mambo and openfirmware properly.

             -eric


_______________________________________________
Xen-ppc-devel mailing list
Xen-ppc-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ppc-devel