[rescue] What to do with the SS20

Sandwich Maker adh at an.bradford.ma.us
Sun Jun 22 19:37:08 CDT 2014

" From: Steven M Jones <smj+rescue at crash.com>
" On 06/22/2014 06:21 AM, Sandwich Maker wrote:
" > from memory, the reason for 4.1.4 was
" > integrating the ross kernel fixes as sun started selling ross-equipped
" > ss20s, and i believe it supports smp on rosses.
" >
" > it also supports smp on ti cpus, afaik something sun never advertised.
" > i personally tested and verified this.
" Sun introduced multiprocessor support in SunOS 4.1.2, IIRC, partly in
" response to Solbourne (IMHO). The SPARCserver 600MP models were out and
" 4.1.2 supported them well before the first harrowingly unstable Solaris
" 2 release, so I don't think the MP support was a secret.

yes, but it didn't run on ss20s or ross cpus.  i'm curious if it runs
on the sparc powerup enhanced sun4c, or only proper sun-branded cpus.

" Also, I believe all SunOS 4 releases can only have one CPU executing
" kernel code at a time. I'm not sure if technically you have to call that
" asymmetric multiprocessing, rather than SMP.

now that you mention this, i think you're right.  it wasn't a problem
for ${WORK[1995]}, as we were doing complex fpga layouts and sims, and
2 cpus meant the engineers - and the systems - could do useful work on
one cpu while the big heavy sim app hogged the other one.

dredged out of 2-decade-old memories, solbourne dug into the kernel
code and implemented fine-grained spinlocks, but sun basically just
put one giant spinlock around the entire kernel.  since solbourne
licensed sun's codebase, why couldn't sun just build on their work?
Andrew Hay                                  the genius nature
internet rambler                            is to see what all have seen
adh at an.bradford.ma.us                       and think what none thought

More information about the rescue mailing list