Re: Panics going multiuser with NetBSD 8, 9, current

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

On Thu, 7 Oct 2021 at 17:28, John Klos <john%ziaspace.com@localhost> wrote:
>
> Hi,
>
> I now have some new data...
>
> > CSMk3 and CSPPC have the same product id. So as long as we don't run a PPC
> > function test the only way to differentiate them was the serial number,
> > which starts with an 'F' for the MK3.
> >
> > You serial number seems to be empty. Or maybe it is somewhere else, due to a
> > Flash ROM update? dev/p5bus.c assumes to find it at 0xf00010.
> >
> > But this is just a side note. It has definitely nothing to do with your
> > problems. Our boards are still treated the same by the kernel.
>
> Indeed. I had flashed the latest 44.71 firmware from http://phase5.a1k.org/
>
>
> >> FYI - in NetBSD 9 and, I think, current, my A1200 can't write to the CF
> >> card that it could write to just fine with NetBSD 8, but I think that's
> >> a separate issue.
> >
> > Probably. But tell us what happens. An error, a panic?
>
> This is a separate issue which may've been fixed recently by Rin:
> http://mail-index.netbsd.org/source-changes/2021/10/05/msg132733.html
>
> The A1200 would have errors on any attempt to write. Similar, likely
> related IDE issues exist in mac68k (Quadra 630), hpcsh, hpcarm, et cetera.
> I hope to test this soon.
>
> > I seem to remember that there have been some effort to make the CF port work
> > with 16-bit network cards over the last years...
>
> Doesn't it? I remember a Zaurus with a CF wifi card.
>
> >> ...and I get the same panic when booting via NFS...
> >
> > Wow... this is really strange. Would be nice to have a second CSPPC/Mk3
> > A4000 for comparison. I may try to test with mine (which also has a CSPPC)
> > but it is defective - the Zorro bus no longer works correctly.
>
> The new data is that I've fixed an A3000 I've had for ages and installed
> the Cyberstorm Mk3 in to it.
>
> I can boot NetBSD just fine! Well, ignoring the complexities of the boot
> process, it's fine.
>
> >> What if this has something to do with a crazy date or something else
> >> stored in the battery backed RAM?
> >
> > As far as I know NetBSD will always use the last time from the file system,
> > after it displayed "WARNING: CHECK AND RESET THE DATE!".
>
> I connected a battery to the motherboard of the A4000, but no attempts to
> set the date / time in AmigaOS can be saved - "use" is fine, which sets
> the system time for the current boot, but "save" is greyed out, and every
> program I can find says there's no RTC at all.
>
> Since the motherboard was damaged by a leaky battery, this by itself isn't
> surprising, but the fact that this appears to be causing panics is
> surprising.
>
> Even a kernel without a34kbbc0 and a2kbbc0 panics on the A4000.
>
> Next, I'm going to try to repair the RTC...

Curiouser and  curiouser...!

More random thoughts
- What happens if you chroot into a small tmpfs with some binaries
- Does the same kernel without a34kbbc0 work fine on the A3000
- Does current panic the same

David