Port-sparc archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

(was: Re: Net-installing SS20 & NFS woes (?))



Le lun. 10 août 2020 à 13:52, Julian Coleman <jdc%coris.org.uk@localhost> a écrit :
> I'm fairly sure that we don't support SMP on those

O ye of little faith :-) "Of course it runs NetBSD" :-)
(well, _boots_ at least, and it hasn't crashed yet).

The MBus page claims it was working once
(<http://mbus.sunhelp.org/systems/sun/ss20.htm#11>).
After switching the J1410 jumper for 40 MHz mbus operation, the normal
netbsd kernel booted just fine. So I tried the MP kernel...

```
bash-5.0$ dmesg | grep cpu
[     1.000000] cpu0 at mainbus0: mid 8: Cypress,CY605 @ 40 MHz, RT602
or WTL3171 FPU
[     1.000000] cpu0: 64K byte write-back, 32 bytes/line, sw flush:
cache enabled
[     1.000000] cpu1 at mainbus0: mid 9: Cypress,CY605 @ 40 MHz, RT602
or WTL3171 FPU
[     1.000000] cpu1: 64K byte write-back, 32 bytes/line, sw flush:
cache enabled
[     1.000000] cpu2 at mainbus0: mid 10: Cypress,CY605 @ 40 MHz,
RT602 or WTL3171 FPU
[     1.000000] cpu2: 64K byte write-back, 32 bytes/line, sw flush:
cache enabled
[     1.000000] cpu3 at mainbus0: mid 11: Cypress,CY605 @ 40 MHz,
RT602 or WTL3171 FPU
[     1.000000] cpu3: 64K byte write-back, 32 bytes/line, sw flush:
cache enabled
[     1.000006] cpu0: booting secondary processors: cpu1 cpu2 cpu3
bash-5.0$ cpuctl list
Num  HwId Unbound LWPs Interrupts Last change              #Intr
---- ---- ------------ ---------- ------------------------ -----
0    0    online       intr       Mon Aug 10 16:52:39 2020 0
1    1    online       intr       Mon Aug 10 16:52:39 2020 0
2    2    online       intr       Mon Aug 10 16:52:39 2020 0
3    3    online       intr       Mon Aug 10 16:52:39 2020 0
bash-5.0$ for X in $(seq 0 3); do cpuctl identify $X; done
cpu0: mid 8: Cypress,CY605 @ 40 MHz, RT602 or WTL3171 FPU
cpu0: 64K byte write-back, 32 bytes/line, sw flush: cache enabled
cpu0: SPARC v7
cpu1: mid 9: Cypress,CY605 @ 40 MHz, RT602 or WTL3171 FPU
cpu1: 64K byte write-back, 32 bytes/line, sw flush: cache enabled
cpu1: SPARC v7
cpu2: mid 10: Cypress,CY605 @ 40 MHz, RT602 or WTL3171 FPU
cpu2: 64K byte write-back, 32 bytes/line, sw flush: cache enabled
cpu2: SPARC v7
cpu3: mid 11: Cypress,CY605 @ 40 MHz, RT602 or WTL3171 FPU
cpu3: 64K byte write-back, 32 bytes/line, sw flush: cache enabled
cpu3: SPARC v7
```

I did not run extensive tests, but I was able to start X11 (with
windowmaker and rxvt as a terminal, no less) while `makemandb` was
burning a CPU (which it is still doing, 30 minutes of CPU time
later...). It took a while for X to start... I also successfully
installed a binary package during X11 startup.

The current disk is too small to try some pkgsrc building, and those
CPUs are *slow*, but as far as I can tell they are still working :-)
And you definitely need 4 of them to do anything, otherwise even a
pair of SM41 is probably a better deal...

So once more - kudos to everyone involved to have a kernel that still
works on such an obsolete (and rare?) piece of kit!

I'll probably switch back to the SM61 soon, but if there's some
specific test(s) to run to check the SMP functionality, let me know
(... on a standard kernel or prebuilt kernel and without pkgsrc, slow
machine with little disk space sorry).

Cordially,

-- 
Romain Dolbeau


Home | Main Index | Thread Index | Old Index