Subject: Re: --db_more-- in recent sparc64 kernel
To: Andrey Petrov <petrov@netbsd.org>
From: john heasley <heas@shrubbery.net>
List: port-sparc64
Date: 07/17/2001 12:12:04
Tue, Jul 17, 2001 at 12:00:16PM -0700, Andrey Petrov:
> On Tue, Jul 17, 2001 at 11:32:54AM +0200, Martin Husemann wrote:
> > > Hum..... magic didn't happen, I moved toolchain, but kernel
> > > still crashes. 
> > 
> > You are not by chance running an older libc on the machine where it works?
> 
> Thanks for suggestion, libc seems fine, the latest I got is
> it doesn't matter on which machine I build kernel and what toolchain
> I used, something is different in machines themselves:
> 1) ultra2 with 2 cpu and serial console, kernel boots and works
> 2) ultra2 with 1 cpu and display, kernel crashes at the very beginning
> 
> 	Andrey

i have
1) ultra2 w/ 2x300mhz, 1g dram, serial console, 1/2001 toolchain tarball
2) ultra2 w/ 2x300mhz, 256m dram, serial console, dwarf toolchain tarball

DDB enabled build on 2) produces the --db_more-- kernel when booted on
either macheen and 1) produces functional kernel when booted on either.  in
case that was not clear from before.

i have not had a chance to look closer at the trace.  will try to asap.

> > 
> > I'm currently running a real current kernel, but leave userland and libc
> > behind, because when I last tried the new memcpy/memset stuff made the
> > toolchain go berserk.
> > 
> > And, before anyone asks: no, I did not try it very recently - so it might well
> > be fixed now.
> > 
> > 
> > Martin