On Wed 25 Oct, Chris Gilbert wrote:
> On Wed, 25 Oct 2000, Kjetil B. Thomassen wrote:
> > There has been a lot of talk about installing NetBSD/arm32 1.5 lately,
> 
> Yep, I'm hoping to get some activity on this front again.

That's excellent.

> > and this prompted me to check on the status of two of my PRs:
> > port-arm32/8786
> >   [open doc-bug] Prep file for RISC OS is not included with Install
> >                  documentation
> > port-arm32/8787
> >   [open doc-bug] Prep file for RISC OS needs to be updated
> >
> > There does not seem to have been taken any action on either of them, and
> > that is very disappointing.
> 
> They're not the only PR's open awaiting something to happen (some are over a 
> year old!)

I can imagine.

> > I think that both of these should be fixed before 1.5 is released, and
> > especially the former.
> 
> Yes, but it's a matter of time on someone's part.

OK, I can't do much about the former, but I can do something about the
latter, if I can find some time to do it.

> > I hope to be able to test NetBSD/1.5_BETA on my Risc PC 600 with Power-tec
> > SCSI-2, and two Acorn SCSI cards if I can get the original ROMs for the
> > latter in time.
> 
> Excellent :)

As expected, the Acorn SCSI card with a PowerROM lead the Risc PC into
a hang, so I had to put podule1.disable in the boot loader extra
options.

The kernel I used is the one that Chris made for the Epsom show.

> > Also, PR 9226 also is still at open.
> 
> hmmm.  Is this any better in 1.5 alpha or beta?

Unfortunately I have been unable to test that as the kernel panics on a
dead wd0 IDE drive when sysinst is trying to scan the drives after
having asked me whether I want to continue with the install to a new
drive (the first option in sysinst).

The details are:
Sysinst install to harddrive
  - "Shall we continue?" Yes
     => wdc0:0:0: lost interrupt
        Type: ata tc_bcount: 512 tc_skip: 0
        Non-emulated page fault with intr_depth > 0
        Data abort: 'Translation fault (page)' status=007 ...
        Stopped at inq_entry+0xa0 ldr r2,[r7,r9,lsl#2]
        db> trace
        _mi_switch
        _ltsleep
        _biowait
        _filecore_label_read
        _readdisklabel
        _wdgetdisklabel
        _wdopen
        _spec_open
        _vn_open
        _sys_open
        _syscall

I managed to get the networking up and running on my Etherlan 600 (ne0),
and stuck the dmesg (/kern/msgbuf) at:
http://www.thomassen.priv.no/riscpc.dmesg.15alpha

More information can be found at:
http://home.eunet.no/~kjetilbt/MyRiscPC/

I would like to congratulate the people who have fixed the problem with
ne0 that required it to be configured in RISC OS before booting NetBSD.
That is no longer necessary. Well done!

A couple of other problems I have seen:
1) I am using a custom made monitor definition file (MDF) for my HP
   A1097C monitor. This only supports 1280x1024 in 70 Hz. I started the
   boot loader and defined the screen mode to X640 Y480 C256 F60 as my
   other monitor (see above) supports that. When I booted and switched
   to this monitor, that did not help at all. A shift-boot had to be
   done to solve it, and then it worked fine.
2) Sometimes the computer hangs a long time before the colour stripes
   appear.
3) After a ping from sysinst, the screen is not updated properly, and I
   therefore am unable to see that it was successful.

> > My problem is that my time is very limited, but I will test any kernel
> > that people throw at me.
> >
> > I also discovered a problem with (I think) 1.4.x that it just panics when
> > I try to run the install kernel on my Risc PC with my current Power-tec
> > primary drive formatted to 1.5GB for RISC OS. This is with RISC OS 4.02.

This could be the same problem as the one reported above, as I suspect
that sysinst has been changed to do the scanning of wd drives at a later
stage nowadays.

> > Gary Partis of Partis Computing (http://www.partis.org.uk/) has agreed to
> > help anyone in the NetBSD team asking for information, so I think that
> > someone with kernel knowledge should contact him
> > (mailto:gary@partis.org.uk) to get the details needed.
> 
> Excellent, quite a few people at the acorn show were willing to give info and 
> hardware as well.

Partis Computing also supports the Morley cards, so we may get some
valuable information about those from them. Also, all the cards with
PowerROM should be within reach with information from there.

> > There are also the problems with PowerROMs being identified as Power-tec
> > cards, but there does not seem to be a PR on that. I could submit one if
> > no-one else picks up that ball.
> 
> The problem is that the power-roms break the required info, and there's no 
> way to detect the underlying card (although looking at the podulebus 
> detection code there's a way to force a podule to be something by giving it a 
> specific manufacturer's and product id code...

There are ways to detect this, and some information has been sent to
David Brownlee and Chris, so maybe we can get something working now.
If other people are interested, send me a mail, and you'll get it.


I am afraid that I am unable to get any further now, but if I want to
shut down that drive that is causing me problems, I have to disconnect
all my peripherals in order to be able to open my Risc PC. I don't have
the time right now to do that, but maybe on Sunday or the next couple of
weeks.

Well, anyway, this is how far I got this time.

A newer kernel could solve this problem, and if someone builds a newer
install kernel and get it to me, I will test it.

TIA!

Kjetil B.
http://www.thomassen.priv.no/
mailto:kjetil@thomassen.priv.no