> On Wed, Jan 08, 2003 at 03:40:15PM +0100, ragge@ludd.luth.se wrote:
> > I just tried to compile a kernel on a similar machine (4/670, 4 CPUs,
> > 128MB RAM) but it do not manage to start the other cpus. Boot message:
> [...]
> > cpu0 at mainbus0: mid 8: CY7C601/605 (v.c) @ 40 MHz, RT602 or WTL3171 FPU
> > cpu0: 64K byte write-back, 32 bytes/line, sw flush: cache enabled
> > cpu1 at mainbus0: mid 9: CY7C601/605 (v.c) @ 40 MHz, RT602 or WTL3171 FPU
> > cpu1: 64K byte write-back, 32 bytes/line, sw flush: cache enabled
> > cpu2 at mainbus0: mid 10: CY7C601/605 (v.c) @ 40 MHz, RT602 or WTL3171 FPU
> > cpu2: 64K byte write-back, 32 bytes/line, sw flush: cache enabled
> > cpu3 at mainbus0: mid 11: CY7C601/605 (v.c) @ 40 MHz, RT602 or WTL3171 FPU
> > cpu3: 64K byte write-back, 32 bytes/line, sw flush: cache enabled
> [...]
> > cpu0: booting secondary processors: cpu1 cpu2 cpu3
> > xcall(cpu1,0xf0261c98): couldn't ping cpus: cpu0 cpu2 cpu3
> > xcall(cpu1,0xf025bce0): couldn't ping cpus: cpu0 cpu2 cpu3
> 
> That's what I'm seeing on my 670mp with 4 CPUs (looks like we have
> identical machines). I was wondering if it's a hw issue with some of
> the 601/605 modules; certain versions were known to run SunOS 4.1.x,
> but not Solaris. I know that this applies to my machine, it never
> ran Solaris for that reason back when it was in service at the U.
> as a fileserver.
> 
I don't think Solaris was officially supported on any 605 CPUs.
But there are also sub-revisions of the CY7C605, and SunOS4
had workarounds for of them:

 * If we are running on a vers=0xE (A.3) Cy7C605,
 * we must not use copyback mode.
 * If we are running on a vers=0xD (A.5) Cy7C605,
 * we must not use ASI-based flushes.

Whatever this means, I'm no expert on the Sparc architecture :-)

-- Ragge