Re: 10.0 - Installer newfs segfaults on SS2

[David Brownlee <abs%absd.org@localhost>]

On Sun, 8 Sept 2024 at 20:24, vom513 <vom513%gmail.com@localhost> wrote:
>
> > On Sep 6, 2024, at 12:14 PM, David Brownlee <abs%absd.org@localhost> wrote:
> >
> > On Fri, 6 Sept 2024 at 00:01, vom513 <vom513%gmail.com@localhost> wrote:
> >>
> >>> On Sep 2, 2024, at 9:13 AM, David Brownlee <abs%absd.org@localhost> wrote:
> >>>
> >>> Random likely non relevant thoughts.
> >>>
> >>> Does adding -d to the newfs command run without issues? (I'd expect
> >>> so, but probably worth a quick check).
> >>> Could test newfs on a smaller partition not starting at 0 (not
> >>> expecting any different behaviour)
> >>> Could test newfs_msdos instead (again, not expecting anything
> >>> different... but maybe).
> >>>
> >>> (A small voice in the back of my mind is saying "floating point", but
> >>> I think we're probably best off all ignoring it :)
> >>
> >> I should have mentioned this earlier… :/
> >>
> >> So this machine was purchased so I could extract the Weitek PowerUP to put in my IPX.  After that was successful, I figured “hey I’ll put the old 40mhz Weitek back in this SS2 and see if I can get it running nicely…”
> >>
> >> The cache is an AT&T chip, the FPU a TI.  Are there issues mixing and matching CPU/FPU/Cache in this platform ?
> >>
> >> I have another SS2 that I should have this weekend that looks to have LSI CPU + Cache and TI FPU.  So there’s always a total mainboard swap, but if there is any value in swapping these 3 chips around into a kosher arrangement please let me know.
> >
> > OK, so there could potentially be "other issues" here :)
> >
> > How much free space do you have on your bootable install image, can you run
> >
> > dd if=/dev/zero of=temp bs=100k count=1
> > newfs -F temp
> >
> > Out of curiosity, does the system run fine if you run it diskless
> > across NFS, or alternatively if you swap in an already installed disk
> > from the IPX or similar?
>
> Thanks again for the reply and ideas.
>
> I finally got the other SS2 this weekend and it was in much better condition than I had thought.  I swapped the mainboard, PROM, rejiggered some RAM, and modified the original NVRAM with a glitchwrks top PCB/battery.
>
> I’m happy to report newfs didn’t segfault and it’s installing normally as I type this.
>
> Another possibly interesting data point.  With the old mainboard, when I would tweak partitions to give it more swap, it seemed to let me put the partitions “right next to / touching each other”.
>
> This time with the new mainboard, it offset my swap by ~ 8mb ?  Basically I have been making a: 7168mb and b: (swap) 512mb.  Before I am pretty sure it was letting me land swap starting at 7169 but this time it padded that to 7177 I think.

Sun PROM (and by extension NetBSD/sparc) disklabels only support
partitions aligned on cylinder boundaries, though as long as they end
on a cylinder boundeary they can be contiguous. That could imply some
odd PROM difference or (handwaves).

Would be very interested to see a dmesg from both boards with the same
hardware attached :)

> So while I did change a few variables at once (I know I know not the best troubleshooting…) - I really do believe the old mainboard / CPU (although this was happy in the IPX) / FPU / Cache are ill.  I’m going to keep this old machine around and if/when I ever have enough SS2 flavored parts to troubleshoot some more I will so that I can salvage as much as possible from it.

Definitely an interesting project to work out exactly what is not
quite right with the original board. Good to hear you have a working
SS2 system now!

David