Hi, all.
We've got a 4 CPU SparcServer 1000E that we've just started running
Linux on. Everything is looking good, except we can't get it to boot
SMP kernels.
After fiddling around in the source code and adding some debug
messages, it looks like the "slave" cpus hang somewhere in
smp4d_callin. CPU 0 finds 3 other processors, and tries to start
them. CPU 1 prints debug messages up until the first
local_flush_cache_all() in smp4d_callin
(<URL:http://lxr.linux.no/source/arch/sparc/kernel/sun4d_smp.c#L99>).
CPU 0 waits some, reports CPU 1 as stuck, and proceeds to kick off CPU
2. CPU 2 gets just as far as CPU 1, then everything hangs. CPU 0
never reports it as stuck, nor does it try to start CPU 3.
At this point, everything is dead. With a stock kernel I have to hard
reset the machine, with the debug messages I get a watchdog reset
after a while. I'm not sure what causes this difference, but it might
just be something that got tickled somewhere, I suppose.
The lights at the back are fixed at
First mainboard: GG-----YYY
Second mainboard: GGYYYYYYYY
(That's "G"reen, "Y"ellow and "-" for off.)
This is with several kernels, but the last we tried was 2.2.14-pre13
from the CVS tree at cvs.on.openprojects.net. gcc is 2.95.2 19991109.
/proc/cpuinfo in UP mode contains this:
cpu : Texas Instruments, Inc. - SuperSparc 50
fpu : SuperSparc on-chip FPU
promlib : Version 3 Revision 2
prom : 2.31
type : sun4d
ncpus probed : 4
ncpus active : 1
BogoMips : 59.80
MMU type : TI Viking/MXCC
invall : 0
invmm : 0
invrnge : 0
invpg : 0
contexts : 65536
The box ran Solaris/SMP happily just a few days ago, and it'd be a
shame if we had to give up Linux/SMP on it. We are, however, at a bit
of a loss at the moment. I'm not eager to try to untangle the
local_cache_flush_all stuff, as it looks rather involved. Does anyone
have any bright ideas on what to try or where to investigate next?
TIA,
Per Kristian
-
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of the message to [EMAIL PROTECTED]
