NetBSD/vax

Post by Rod Smallwood
Has anybody successfuly installed NetBSD on a VAX.

I've run NetBSD on a VAX.

Post by Rod Smallwood
If which version on which VAX

In my case, it was relatively old NetBSD - probably in the 1.x range,
likely an old version of my 1.4T - on a MicroVAX-II (KA630 CPU).

Indeed, I can _almost_ run NetBSD 1.4T on my KA630 emulator.

/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML ***@rodents-montreal.org
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Anders Magnusson

2020-03-17 07:41:20 UTC

Morning,

Post by Rod Smallwood
With the upcoming demise of VMS/VAX I thought I might give NetBSD/vax a try.
Downloading a bootable image and burning it onto a CD was not a problem.
On the 3100 with attached RRD42 and SCSI drive RZ26l the CD duly booted
ito the NetBSD install menu.
All of versions 7,8 and 9. failed after partitioning at the point where
the system is copied to the hard disk.
Has anybody successfuly installed NetBSD on a VAX. If which version on
which VAX

I've installed (at least -8, if not -9) on my 4000/90. About 2 years
ago I fixed a bug which
prevented 3100/38 to use the SCSI disk, but that fix should be in -9.

What error do you get? It is supposed to work, so I may have to fire up
an old machine if it's not working.

-- R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Johnny Billquist

2020-03-17 09:47:28 UTC

Post by Anders Magnusson
Morning,

I can't remember myself, but I think there are some issues with the
standard installation tools.

The system can be installed, and get to run, but it probably require
some manual work.

Also, NetBSD/vax cannot build itself. But I assume Ragge already knows
this. And there might be issues building a lot of other software as
well. At this time, I'd say it is somewhat questionable how useful the
VAX port is for a random user. :-(

But yes, I have (more or less) current running on a 4000/90.

Johnny

--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: ***@softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Martin Husemann

2020-03-17 10:50:50 UTC

The system can be installed, and get to run, but it probably require some
manual work.

I think fixed all low memory issues in the install media some years
ago, but of course new issues might have come up and went unnoticed.
Back then booting from CD and installing worked on a 3100 with 24 MB
(my only currently available test machine has 128 MB and I have no spare
SCSI disks any more to test installations - but of course we can test in
SIMH).

Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Erno Palonheimo

2020-03-17 11:03:19 UTC

The system can be installed, and get to run, but it probably require some
manual work.

FWIW, I had 8.0 running (slowly) on a MicroVAX 3100m30 with 24 megabytes
of memory. I netbooted the installer and it worked just fine. I might
reinstall NetBSD on the machine in coming weeks and can report how it
goes.

-e

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

David Cooper

2020-03-17 23:46:08 UTC

CD drives are finicky on the 3100 MicroVAXen in my experience. I generally
use a linux box and dd the cd image to a scsi drive and install from there.

-----Original Message-----
From: port-vax-***@NetBSD.org <port-vax-***@NetBSD.org> On Behalf Of
Erno Palonheimo
Sent: Tuesday, March 17, 2020 4:03 AM
To: Martin Husemann <***@duskware.de>
Cc: Johnny Billquist <***@update.uu.se>; Anders Magnusson
<***@tethuvudet.se>; Rod Smallwood <***@btinternet.com>;
port-***@netbsd.org
Subject: Re: NetBSD/vax

Post by Johnny Billquist
The system can be installed, and get to run, but it probably require
some manual work.

I think fixed all low memory issues in the install media some years
ago, but of course new issues might have come up and went unnoticed.
Back then booting from CD and installing worked on a 3100 with 24 MB
(my only currently available test machine has 128 MB and I have no
spare SCSI disks any more to test installations - but of course we can
test in SIMH).

FWIW, I had 8.0 running (slowly) on a MicroVAX 3100m30 with 24 megabytes of
memory. I netbooted the installer and it worked just fine. I might reinstall
NetBSD on the machine in coming weeks and can report how it goes.

-e

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Erno Palonheimo

2020-03-19 06:41:27 UTC

Post by Johnny Billquist
The system can be installed, and get to run, but it probably
require some
manual work.

So yesterday I did this. Installer ran its course fine but the system
would not boot - kernel hangs right after bootloader messages, before
the kernel itself prints anything out. Then I remembered this same
thing happened with 8.0 and I actually fixed it, but apparently my
patch was never incorporated and thus I need to build a custom kernel
for it to actually boot. Back then, Ragge asked me to send him a diff
so he could commit it, which I did, but looks like it was forgotten.

Here's a link to a diff against 8.0 that will probably apply just fine
to 9.0.

https://esp.iki.fi/netbsd-8.0.lcg0.lkkbd.patch

-e

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Johnny Billquist

2020-03-17 23:49:26 UTC

I've installed from CD. The only requirement is that it can do 512 byte
blocks, which does put a slight limit on choice of drives, but otherwise
I haven't had much trouble.

But I usually have been installing over the network, which I find much
more convenient.

Get mopd running, tftpd, bootpd (or dhcpd), and nfs, and you're all set.

Johnny

CD drives are finicky on the 3100 MicroVAXen in my experience. I generally
use a linux box and dd the cd image to a scsi drive and install from there.
-----Original Message-----
Erno Palonheimo
Sent: Tuesday, March 17, 2020 4:03 AM
Subject: Re: NetBSD/vax

Post by Johnny Billquist
The system can be installed, and get to run, but it probably require
some manual work.

I think fixed all low memory issues in the install media some years
ago, but of course new issues might have come up and went unnoticed.
Back then booting from CD and installing worked on a 3100 with 24 MB
(my only currently available test machine has 128 MB and I have no
spare SCSI disks any more to test installations - but of course we can
test in SIMH).

sadness

2020-03-18 22:45:00 UTC

Post by Rod Smallwood
Has anybody successfuly installed NetBSD on a VAX. If which version on
which VAX
Rod Smallwood

I've just completed an installation of NetBSD 9.0 on a VaxStation
4000/96 with 80Mb of memory, booted from cdrom and got through the
installation process without a single problem.

On the other hand trying to boot NetBSD 9.0 on a VaxStation 3100m30
results in a kernel crash when probing the scsi bus, so maybe there is
some problem with ncr 5380 scsi driver on this system ?

I will try on an VaxStation 3100m76 in the next few days to see how the
installation will go on a system with less memory (24Mb on this one) and
the ncr 5380 scsi subsystem.

--
Giuliano Paolo Brunetti
http://www.flickr.com/photos/***@N02/sets/
Italian hobbyist DECnet network - http://decnet.ipv7.net

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Martin Husemann

2020-03-19 05:54:21 UTC

Post by sadness
On the other hand trying to boot NetBSD 9.0 on a VaxStation 3100m30
results in a kernel crash when probing the scsi bus, so maybe there is
some problem with ncr 5380 scsi driver on this system ?

Any details of the crash?

Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

sadness

2020-03-21 21:56:27 UTC

Any details of the crash?

That's a bootlog of NetBSD 9.0 on the 3100m30 with scsi disk attached on
the bus, the panic happen every time there's a scsi disk on the bus,
maker model or size of the disk doesn't matter, this machine will run
without problem with OpenVMS or Ultrix so I don't think there's any
underlying hardware problem, I will try some older release of NetBSD in
the next few days

sh dev

VMS/VMB ULTRIX ADDR DEVTYP NUMBYTES RM/FX WP DEVNAM
------- ------ -------- ------ -------- ----- -- ------
ESA0 SE0 08-00-2B-0F-BD-76

DKA0 RZ0 A/0/0/00 DISK 525 MB FX EMPIRE_
DKA400 RZ4 A/4/0/00 RODISK 341 MB RM CD-ROM
...HostID.... A/6 INITR

-DKA400

Post by sadness
NetBSD/vax boot [1.12 (Fri Feb 14 00:06:28 UTC 2020)] <<
Press any key to abort autoboot

getdisklabel: no disk label
nfs_open: must mount first.
open netbsd.vax: No such file or directory

Post by Martin Husemann
boot netbsd

getdisklabel: no disk label
nfs_open: must mount first.
3293856+205208[238480+226939]=0x3c821c
[ 1.0000000] Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002,
2003, 2004, 2005,
[ 1.0000000] 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014,
2015, 2016, 2017,
[ 1.0000000] 2018, 2019, 2020 The NetBSD Foundation, Inc. All
rights reserved.
[ 1.0000000] Copyright (c) 1982, 1986, 1989, 1991, 1993
[ 1.0000000] The Regents of the University of California. All
rights reserved.

[ 1.0000000] NetBSD 9.0 (GENERIC) #0: Fri Feb 14 00:06:28 UTC 2020
[ 1.0000000]
***@mkrepro.NetBSD.org:/usr/src/sys/arch/vax/compile/GENERIC
[ 1.0000000] VAXstation 3100/m{30,40}
[ 1.0000000] total memory = 16264 KB
[ 1.0000000] avail memory = 11444 KB
[ 1.0000000] mainbus0 (root)
[ 1.0000000] cpu0 at mainbus0: KA420, CVAX, 1KB L1 cache, 64KB L2 cache
[ 1.0000000] vsbus0 at mainbus0
[ 1.0000000] vsbus0: interrupt mask 8
[ 1.0000000] le0 at vsbus0 csr 0x200e0000 vec 120 ipl 17 maskbit 5 buf
0x4ab000-0x4bafff
[ 1.0000000] le0: address 08:00:2b:0f:bd:76
[ 1.0000000] le0: 32 receive buffers, 8 transmit buffers
[ 1.0000000] dz0 at vsbus0 csr 0x200a0000 vec 304 ipl 17 maskbit 6
[ 1.0000000] dz0: 4 lines
[ 1.0000000] lkkbd0 at dz0
[ 1.0000000] lkkbd0: no keyboard
[ 1.0000000] wskbd0 at lkkbd0 mux 1
[ 1.0000000] lkms0 at dz0
[ 1.0000000] wsmouse0 at lkms0 mux 0
[ 1.0000000] hdc0 at vsbus0 csr 0x200c0000 vec 774 ipl 17 maskbit 0
[ 1.0000000] si0 at vsbus0 csr 0x200c0080 vec 770 ipl 17 maskbit 1
[ 1.0000000] si0: NCR5380, SCSI ID 6
[ 1.0000000] scsibus0 at si0: 8 targets, 8 luns per target
[ 1.0000000] smg0 at vsbus0 csr 0x200f0000 vec 104 ipl 17 maskbit 3
[ 1.0000000] smg0: could not find 8x15 font
[ 1.0600080] scsibus0: waiting 2 seconds for devices to settle...
[ 3.0900080] r0=00000000 r1=80f1c800 r2=80f1c800 r3=80f1c974
r4=00004000 r5=0000004c r6=80f1a020 r7=80f1ccd8
[ 3.0900080] r8=00000000 r9=00000000 r10=80f1a020 r11=00000000
[ 3.0900080] ap=82f33cc4 fp=82f33ca8 sp=00000000 pc=80007780
[ 3.0900080] panic: SEGV in kernel mode: pc 0x80007780 addr 0x44
[ 3.0900080] cpu0: Begin traceback...
[ 3.0900080] panic: SEGV in kernel mode: pc 0x80007780 addr 0x44
[ 3.0900080] Stack traceback :
[ 3.0900080] Process is executing in user space.
[ 3.0900080] cpu0: End traceback...
Stopped in pid 0.16 (system) at netbsd:vpanic+0x171: pushl $0

Martin Husemann

2020-03-22 07:38:55 UTC

Post by sadness
[ 1.0600080] scsibus0: waiting 2 seconds for devices to settle...
[ 3.0900080] r0=00000000 r1=80f1c800 r2=80f1c800 r3=80f1c974
r4=00004000 r5=0000004c r6=80f1a020 r7=80f1ccd8
[ 3.0900080] r8=00000000 r9=00000000 r10=80f1a020 r11=00000000
[ 3.0900080] ap=82f33cc4 fp=82f33ca8 sp=00000000 pc=80007780
[ 3.0900080] panic: SEGV in kernel mode: pc 0x80007780 addr 0x44

This is a NULL deref in:

(gdb) list *(0x80007780)
0x80007780 is in si_dma_alloc (../../../../arch/vax/vsa/ncr.c:278).
273 found:
274 dh = &sc->ncr_dma[i];
275 dh->dh_flags = SIDH_BUSY;
276 dh->dh_addr = ncr_sc->sc_dataptr;
277 dh->dh_len = xlen;
278 dh->dh_proc = xs->bp->b_proc;
279
280 /* Remember dest buffer parameters */
281 if (xs->xs_control & XS_CTL_DATA_OUT)
282 dh->dh_flags |= SIDH_OUT;

Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Anders Magnusson

2020-03-22 09:50:45 UTC

(gdb) list *(0x80007780)
0x80007780 is in si_dma_alloc (../../../../arch/vax/vsa/ncr.c:278).
274 dh = &sc->ncr_dma[i];
275 dh->dh_flags = SIDH_BUSY;
276 dh->dh_addr = ncr_sc->sc_dataptr;
277 dh->dh_len = xlen;
278 dh->dh_proc = xs->bp->b_proc;
279
280 /* Remember dest buffer parameters */
281 if (xs->xs_control & XS_CTL_DATA_OUT)
282 dh->dh_flags |= SIDH_OUT;

A quick look at this says that the scsipi_xfer struct do not have any
filled in buf struct, so it crashes trying to extract the active process.
The proc struct is needed on vax to extract the address if doing DMA to
userspace (since it needs to copy to/from a DMA buffer to/from the
destination).

This struct is not needed if the kernel itself is doing DMA, so without
deeper investigation I would assume the fix to be something like:

- Check if ncr_sc->sc_dataptr is in user space, if so assign
dh->dh_proc.   Also add a panic if xs->bp is not set.

This means; replace line 278 with:
    if (((vaddr_t)ncr_sc->sc_dataptr & KERNBASE) == 0) {
        if (xs->bp == NULL)
            panic("si_dma_alloc");
        dh->dh_proc = xs->bp->b_proc;
    }

-- Ragge

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

sadness

2020-03-18 22:52:13 UTC

Post by Anders Magnusson
I've installed (at least -8, if not -9) on my 4000/90. About 2 years
ago I fixed a bug which
prevented 3100/38 to use the SCSI disk, but that fix should be in -9.

Not sure if related but on a 3100/30 (which use the same scsi controller
as the 3100/38) with NetBSD 9 I'm seeing a kernel crash when probing the
scsi subsystem, and the interesting thing is that the crash happen only
when there's scsi disk on the bus, with only the cdrom I don't get any
problem but with a disk attached (tried several drive with different
configuration of the bus) the kernel crash during the probe.
Can you point out at the patch for that bug to see if it's included in -9 ?

David Cooper

2020-03-19 17:08:27 UTC

I know that those models of VAXen don't like SCSI drives over 2GB because of
the set of SCSI commands they understand.

-----Original Message-----
From: port-vax-***@NetBSD.org <port-vax-***@NetBSD.org> On Behalf Of
Erno Palonheimo
Sent: Wednesday, March 18, 2020 11:41 PM
To: Martin Husemann <***@duskware.de>
Cc: Johnny Billquist <***@update.uu.se>; Anders Magnusson
<***@tethuvudet.se>; Rod Smallwood <***@btinternet.com>;
port-***@netbsd.org
Subject: Re: NetBSD/vax

Post by Johnny Billquist
The system can be installed, and get to run, but it probably require
some manual work.

I think fixed all low memory issues in the install media some years
ago, but of course new issues might have come up and went unnoticed.
Back then booting from CD and installing worked on a 3100 with 24 MB
(my only currently available test machine has 128 MB and I have no
spare SCSI disks any more to test installations - but of course we
can test in SIMH).

So yesterday I did this. Installer ran its course fine but the system would
not boot - kernel hangs right after bootloader messages, before the kernel
itself prints anything out. Then I remembered this same thing happened with
8.0 and I actually fixed it, but apparently my patch was never incorporated
and thus I need to build a custom kernel for it to actually boot. Back then,
Ragge asked me to send him a diff so he could commit it, which I did, but
looks like it was forgotten.

Here's a link to a diff against 8.0 that will probably apply just fine to
9.0.

https://esp.iki.fi/netbsd-8.0.lcg0.lkkbd.patch

-e

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

David Brownlee

2020-03-19 18:03:28 UTC

Post by David Cooper
I know that those models of VAXen don't like SCSI drives over 2GB because of
the set of SCSI commands they understand.

To clarify (in case it helps anyone :) providing only the first 2GB of
disk is used everything should be fine in all cases.

More than 2GB should be usable once the kernel has loaded and its
driver is being used - so keeping the boot blocks and partition from
which the kernel is loaded within the first 2GB is sufficient
(typically this just means the root partition, but you can do wacky
things like have a small boot partition under 2GB and then a huge root
partition for everything else - unix doesn't stop you from doing dumb
things as that would also stop you from doing clever things :)

David

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Johnny Billquist

2020-03-19 19:44:10 UTC

Post by David Cooper
I know that those models of VAXen don't like SCSI drives over 2GB because of
the set of SCSI commands they understand.

Actually, they work perfectly fine with large SCSI drives. It is only
the driver in the ROM that have a problem, and that is only relevant at
boot time.

So if you have something like NetBSD, just make the root partition less
than 2GB, and you're all set to go. Once the OS have started, you are
using the driver in the OS instead, which do not have that limitation.
And with NetBSD, booting happens only from the first partition, so if
that partition isn't bigger, then the boot image being read in will be
below the limitation of the driver in the ROM, and then you're fine.

Johnny

Post by David Cooper
-----Original Message-----
Erno Palonheimo
Sent: Wednesday, March 18, 2020 11:41 PM
Subject: Re: NetBSD/vax

Post by Johnny Billquist
The system can be installed, and get to run, but it probably require
some manual work.

I think fixed all low memory issues in the install media some years
ago, but of course new issues might have come up and went unnoticed.
Back then booting from CD and installing worked on a 3100 with 24 MB
(my only currently available test machine has 128 MB and I have no
spare SCSI disks any more to test installations - but of course we
can test in SIMH).

So yesterday I did this. Installer ran its course fine but the system would
not boot - kernel hangs right after bootloader messages, before the kernel
itself prints anything out. Then I remembered this same thing happened with
8.0 and I actually fixed it, but apparently my patch was never incorporated
and thus I need to build a custom kernel for it to actually boot. Back then,
Ragge asked me to send him a diff so he could commit it, which I did, but
looks like it was forgotten.
Here's a link to a diff against 8.0 that will probably apply just fine to
9.0.
https://esp.iki.fi/netbsd-8.0.lcg0.lkkbd.patch
-e

Anders Magnusson

2020-03-20 09:47:07 UTC

The system can be installed, and get to run, but it probably require some
manual work.

Yes, correct! I put it aside because I wanted to test something, but then
I forgot it. I'll take a look at it again this weekend.

-- R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

David Brownlee

2020-03-21 14:21:18 UTC

On Tue, 17 Mar 2020 at 04:10, Rod Smallwood

Do you recall what errors were shown? Was it relating to a disk
timeout or similar?

I definitely installed NetBSD-6 onto a 4000/90A - had it running as a
web server & NFS server for a diskless Dreamcast (
http://netbsd0.blogspot.com/2012/06/retrocomputing-with-vamp-stack-vax.html
)

Would you be willing to try testing NetBSD6 & maybe even 5 to see if
they work on your hardware? If one does we could try to bisect to
determine when the problem was introduced.

Also as another test would you be willing to try formatting and
installing by hand? You should be able to drop out of the installer to
a shell and then something like (from memory)

# dhcpcd
(auto configure network if not already done)
# disklabel -i sd0
(interactive disklabel tool - for testing you could just create a
single partition "a" for up to 2GB)
# newfs sd0a
# mount -o async /dev/sd0a /mnt
# cd /mnt
# ftp ftp://ftp.netbsd.org/pub/NetBSD/NetBSD-9.0/vax/binary/sets/
mget *
quit
# for f in *tgz ; do
tar xzpf $f
done
# installboot /dev/sd0a /usr/mdec/bootxx_ffsv1 /

David

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Martin Husemann

2020-03-22 09:56:40 UTC

- Check if ncr_sc->sc_dataptr is in user space, if so assign dh->dh_proc.
Also add a panic if xs->bp is not set.
    if (((vaddr_t)ncr_sc->sc_dataptr & KERNBASE) == 0) {
        if (xs->bp == NULL)
            panic("si_dma_alloc");
        dh->dh_proc = xs->bp->b_proc;
    }

Souds good - are you able to test with the soure change or should I compile
a test kernel for you? Or do you need a full ISO?

Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Anders Magnusson

2020-03-22 10:03:04 UTC

Souds good - are you able to test with the soure change or should I compile
a test kernel for you? Or do you need a full ISO?

I haven't any 5380 machine running, so somewhat difficult for me to test it.
I'll also be away today (with my family) so I don't know when/if I can
test it.

Do you have a KA42 with SCSI disk running? In that case:

1) verify that a -current install kernel crashes
2) recompile a new kernel with the same options, but patch applied.
3) testboot it (don't have to be from CD).

-- R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Martin Husemann

2020-03-22 10:07:14 UTC

Post by Anders Magnusson
I haven't any 5380 machine running, so somewhat difficult for me to test it.
I'll also be away today (with my family) so I don't know when/if I can test
it.

Sorry, that wasn't meant to you but the original reporter - I don't have
any 5380 VAX either for local testing.

Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Anders Magnusson

2020-03-22 10:08:21 UTC

Post by Anders Magnusson
I haven't any 5380 machine running, so somewhat difficult for me to test it.
I'll also be away today (with my family) so I don't know when/if I can test
it.

Sorry, that wasn't meant to you but the original reporter - I don't have
any 5380 VAX either for local testing.

:-)

-- R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Martin Husemann

2020-03-22 14:11:31 UTC

Post by Anders Magnusson
1) verify that a -current install kernel crashes
2) recompile a new kernel with the same options, but patch applied.
3) testboot it (don't have to be from CD).

The kernel compiles OK on the NetBSD 9.0? With CURRENT or STABLE?

The patch is pretty obvious correct (and needed), and I doubt there
are any changes in -current.

You can get test binaries easily from:

for current:
https://nycdn.netbsd.org/pub/NetBSD-daily/HEAD/
for stable:
https://nycdn.netbsd.org/pub/NetBSD-daily/netbsd-9/

(both would be interesting to test but very likely will crash the same as 9.0)

If you need a kernel or install ISO (or whatever) for testing with the patch,
let me know. If you can compile it yourself, that is fine too of course.

Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

sadness

2020-03-22 16:21:45 UTC

Souds good - are you able to test with the soure change or should I compile
a test kernel for you? Or do you need a full ISO?

I have a fast system already setup for cross-compiling netbsd and I can
netboot the vaxstation easily, I will try the patch in a few hours,
maybe less, right now I'm also trying to resurrect the power supply of
my 3100m76 so I can test it on another system with the ncr 5380 controller
Will report ASAP.

emanuel stiebler

2020-03-22 14:39:18 UTC

Post by Martin Husemann
If you need a kernel or install ISO (or whatever) for testing with the patch,
let me know. If you can compile it yourself, that is fine too of course.
Martin

I'll go to my storage, and will try to find a '30 to test ...

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

sadness

2020-03-22 17:41:57 UTC

Souds good - are you able to test with the soure change or should I compile
a test kernel for you? Or do you need a full ISO?

I need to do more testing but with this patch the -9 kernel does probe
the bus without problem, all the device (including the disk) is
recognized. At this point I will try to install the system and see if
there's any more problem.
I was also able to repair the m76 psu so I will do some testing on that
system too, it will probably happen in the next few days or maybe the
next few hours if I'm forced to lock down my shop.

Here's the bootlog with the patch applied:

[ 1.0000000] Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002,
2003, 2004, 2005,
[ 1.0000000] 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014,
2015, 2016, 2017,
[ 1.0000000] 2018, 2019, 2020 The NetBSD Foundation, Inc. All
rights reserved.
[ 1.0000000] Copyright (c) 1982, 1986, 1989, 1991, 1993
[ 1.0000000] The Regents of the University of California. All
rights reserved.

[ 1.0000000] NetBSD 9.0 (SCSIPATCH) #0: Sun Mar 22 16:41:30 GMT 2020
[ 1.0000000]
***@proliant.anarchy.net:/usr/obj/sys/arch/vax/compile/SCSIPATCH
[ 1.0000000] VAXstation 3100/m{30,40}
[ 1.0000000] total memory = 16264 KB
[ 1.0000000] avail memory = 11444 KB
[ 1.0000000] mainbus0 (root)
[ 1.0000000] cpu0 at mainbus0: KA420, CVAX, 1KB L1 cache, 64KB L2 cache
[ 1.0000000] vsbus0 at mainbus0
[ 1.0000000] vsbus0: interrupt mask 8
[ 1.0000000] le0 at vsbus0 csr 0x200e0000 vec 120 ipl 17 maskbit 5 buf
0x4ab000-0x4bafff
[ 1.0000000] le0: address 08:00:2b:0f:bd:76
[ 1.0000000] le0: 32 receive buffers, 8 transmit buffers
[ 1.0000000] dz0 at vsbus0 csr 0x200a0000 vec 304 ipl 17 maskbit 6
[ 1.0000000] dz0: 4 lines
[ 1.0000000] lkkbd0 at dz0
[ 1.0000000] lkkbd0: no keyboard
[ 1.0000000] wskbd0 at lkkbd0 mux 1
[ 1.0000000] lkms0 at dz0
[ 1.0000000] wsmouse0 at lkms0 mux 0
[ 1.0000000] hdc0 at vsbus0 csr 0x200c0000 vec 774 ipl 17 maskbit 0
[ 1.0000000] si0 at vsbus0 csr 0x200c0080 vec 770 ipl 17 maskbit 1
[ 1.0000000] si0: NCR5380, SCSI ID 6
[ 1.0000000] scsibus0 at si0: 8 targets, 8 luns per target
[ 1.0000000] smg0 at vsbus0 csr 0x200f0000 vec 104 ipl 17 maskbit 3
[ 1.0000000] smg0: could not find 8x15 font
[ 1.0500080] scsibus0: waiting 2 seconds for devices to settle...
[ 3.0900080] sd0 at scsibus0 target 0 lun 0: <QUANTUM, EMPIRE_540S,
1230> disk fixed
[ 3.1500080] sd0: 500 MB, 2874 cyl, 4 head, 89 sec, 512 bytes/sect x
1025901 sectors
[ 3.1800080] sd0: async, 8-bit transfers
[ 3.2300080] sd1 at scsibus0 target 1 lun 0: <SEAGATE, ST1480, 4511>
disk fixe
d
[ 3.2900080] sd1: 406 MB, 1476 cyl, 9 head, 62 sec, 512 bytes/sect x
832527 sectors
[ 3.3200080] sd1: async, 8-bit transfers
[ 3.3800080] cd0 at scsibus0 target 4 lun 0: <TOSHIBA, CD-ROM
XM-5301TA, 0925> cdrom removable
[ 3.4100080] cd0: async, 8-bit transfers
[ 3.5700080] label: 6e6f6974
[ 3.6100080] label: 6e6f6974
[ 3.6900080] WARNING: using Ultrix partition information
[ 3.7500080] WARNING: using Ultrix partition information
[ 3.7700080] WARNING: 1 error while detecting hardware; check system log.
[ 3.7800080] boot device: le0
[ 3.8000080] root on le0
[ 3.8200080] nfs_boot: trying DHCP/BOOTP
[ 6.9400080] nfs_boot: DHCP next-server: 192.168.0.40
[ 6.9600080] nfs_boot: my_name=irata.anarchy.net
[ 6.9700080] nfs_boot: my_domain=anarchy.net
[ 6.9800080] nfs_boot: my_addr=192.168.0.43
[ 6.9900080] nfs_boot: my_mask=255.255.255.0
[ 7.0000080] nfs_boot: gateway=192.168.0.25
[ 13.2300080] root on 192.168.0.40:/export/vax/irata/root
[ 13.2400080] root file system type: nfs
[ 13.2600080] kern.module.path=/stand/vax/9.0/modules

sadness

2020-03-22 17:50:19 UTC

Post by Anders Magnusson
1) verify that a -current install kernel crashes
2) recompile a new kernel with the same options, but patch applied.
3) testboot it (don't have to be from CD).

I can confirm that a -current install kernel crash in the same ways of
-9, I can also confirm that an -8 kernel doesn't have that problem.

With the patch applied the problem goes away, the kernel is able to
probe the bus even with scsi disk attached to it

I will try to install the system to see if there's any more problem with
the scsi subsystem or the installer on this 3100m30.

Martin Husemann

2020-03-22 17:56:56 UTC

Post by sadness
I need to do more testing but with this patch the -9 kernel does probe
the bus without problem, all the device (including the disk) is
recognized.

Very cool - Anders, can you commit + request pullup please?

Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Anders Magnusson

2020-03-22 20:41:34 UTC

Post by sadness
I need to do more testing but with this patch the -9 kernel does probe
the bus without problem, all the device (including the disk) is
recognized.

Very cool - Anders, can you commit + request pullup please?

Yep, done + done! :-)

Thanks for your help everyone!

-- R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Martin Husemann

2020-03-23 10:07:39 UTC

Post by sadness
I need to do more testing but with this patch the -9 kernel does probe
the bus without problem, all the device (including the disk) is
recognized.

Very cool - Anders, can you commit + request pullup please?

Yep, done + done! :-)

and pullup processed!

Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Anders Magnusson

2020-03-23 17:52:38 UTC

Post by sadness
I need to do more testing but with this patch the -9 kernel does probe
the bus without problem, all the device (including the disk) is
recognized.

Very cool - Anders, can you commit + request pullup please?

Yep, done + done! :-)

and pullup processed!

Great! Fast bugfixes! :-)

-- R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Dave McGuire

2020-03-22 17:50:14 UTC

Post by sadness
I was also able to repair the m76 psu so I will do some testing on that
system too, it will probably happen in the next few days or maybe the
next few hours if I'm forced to lock down my shop.

Out of curiosity, what was the failure mode of that power supply?

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

sadness

2020-03-22 20:37:55 UTC

Out of curiosity, what was the failure mode of that power supply?

In my power supply the problem was caused by failed capacitor, there was
four nichicon 1800uf 25v caps that has leaked all over the board and
corroded some of the solder joint in the pcb, every other caps in the
psu was fine except this four, replaced them, cleaned the pcb, redone
the damaged solder joint and the power supply is now working as is
should, before the repair it was totaly dead, no voltage on any rail, no
power led, not even the classic ticking sound of a switching psu in trouble.

sadness

2020-03-22 20:53:36 UTC

Post by sadness
I have a fast system already setup for cross-compiling netbsd and I can
netboot the vaxstation easily, I will try the patch in a few hours,
maybe less, right now I'm also trying to resurrect the power supply of
my 3100m76 so I can test it on another system with the ncr 5380 controller
Will report ASAP.

Got the vaxstation 3100m76 up and running, with a standard netbsd-9 or
-current kernel it has the same kernel panic when probing the scsi bus,
with the patch everything is fine and not only the system will boot
without problem but was able to complete an installation of netbsd-9,
standard installation with sysinst, no manual intervention required, the
system is working good.

I'm tempeted to try on my microvax 3100m30 (ka45) and also on a
vaxstation 2000 (ka410), to see if this two system is also affected by
that bug.

Anders Magnusson

2020-03-22 21:08:12 UTC

Great! Thanks for your help!

Post by sadness
I'm tempeted to try on my microvax 3100m30 (ka45) and also on a
vaxstation 2000 (ka410), to see if this two system is also affected by
that bug.

The m30 is affected, but the VS2k have a different SCSI chip so it is
not affected.

-- R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Dave McGuire

2020-03-22 21:50:48 UTC

Great! Thanks for your help!

Post by sadness
I'm tempeted to try on my microvax 3100m30 (ka45) and also on a
vaxstation 2000 (ka410), to see if this two system is also affected by
that bug.

The m30 is affected, but the VS2k have a different SCSI chip so it is
not affected.

Huh? The VS2000/MV2000 have an NCR 5380.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Maciej W. Rozycki

2020-03-22 22:48:20 UTC

Post by Dave McGuire
Out of curiosity, what was the failure mode of that power supply?

NB this is the usual failure mode of the H7821 PSU used in these systems,
which has capacitors suffering from a known issue with the quaternary
ammonium salt system used in the electrolyte. It causes the seal to fail
eventually and the affected capacitor to leak wreaking havoc. Thankfully
the H7821 is mounted such that the capacitors are upside down in normal
operation making the gravity prevent the electrolyte from interfering with
the seal. Consequently leaks I believe only happen if a machine using an
H7821 is stored in a different orientation, such as on its side, and not
while in use.

The capacitors I have seen fail in this PSU were actually Chemi-Con SXF
line ones (1800µF/25V as you say) and according to my notes there ought to
be five rather than four actually. I would highly recommend chasing the
fifth one and replacing it too.

Other reputable makes and lines of capacitors of the time also suffered
from the said problem, e.g. I saw Nichicon PL parts leak in the H7826 PSU
used with some DECstation 5000 machines.

Maciej

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Andy Farkas

2020-03-22 22:14:00 UTC

Post by sadness
In my power supply the problem was caused by failed capacitor

Something similar to what happened to my BA23 PSU?

https://imgur.com/Lfqsg8h

-andyf

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Dave McGuire

2020-03-23 00:08:35 UTC

Post by Andy Farkas

Post by sadness
In my power supply the problem was caused by failed capacitor

Something similar to what happened to my BA23 PSU?
https://imgur.com/Lfqsg8h

Ah, the infamous Rifa safety capacitors.

I have a (very, very) large stock of replacements for these. Many
thousands. I am slowly gearing up to sell them, but my to-do list is
long and there's no help so setting up a web store just hasn't happened yet.

BA23 owners beware; this WILL happen to every BA23.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Dave McGuire

2020-03-23 05:10:06 UTC

Out of curiosity, what was the failure mode of that power supply?

Nice recovery. Thanks for the info. We have a number of machines
that use the H7821 power supply in storage at LSSM. We've begun
construction of a new exhibit floor on which some of these machines will
be shown off, so I've made a note of your PSU's failure mode to keep
handy when we start digging into those machines.

That is, if we ever get out of COVID19 lockdown. Sigh.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Dave McGuire

2020-03-23 05:16:46 UTC

Post by Maciej W. Rozycki
NB this is the usual failure mode of the H7821 PSU used in these systems,
which has capacitors suffering from a known issue with the quaternary
ammonium salt system used in the electrolyte. It causes the seal to fail
eventually and the affected capacitor to leak wreaking havoc. Thankfully
the H7821 is mounted such that the capacitors are upside down in normal
operation making the gravity prevent the electrolyte from interfering with
the seal. Consequently leaks I believe only happen if a machine using an
H7821 is stored in a different orientation, such as on its side, and not
while in use.
The capacitors I have seen fail in this PSU were actually Chemi-Con SXF
line ones (1800µF/25V as you say) and according to my notes there ought to
be five rather than four actually. I would highly recommend chasing the
fifth one and replacing it too.
Other reputable makes and lines of capacitors of the time also suffered
from the said problem, e.g. I saw Nichicon PL parts leak in the H7826 PSU
used with some DECstation 5000 machines.

Excellent information Maciej, I've made a note of this too. Thank you
for the explanation.

Of course now I remember that I've been storing probably 25
VAXstations and DECstations of that era up on their sides for twenty
years. Crap!

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Anders Magnusson

2020-03-23 07:26:29 UTC

Great! Thanks for your help!

Post by sadness
I'm tempeted to try on my microvax 3100m30 (ka45) and also on a
vaxstation 2000 (ka410), to see if this two system is also affected by
that bug.

The m30 is affected, but the VS2k have a different SCSI chip so it is
not affected.

Huh? The VS2000/MV2000 have an NCR 5380.

D'oh! Of course it has! I'll blame the lack of coffee. Sorry for the
misinformation.

-- R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Dave McGuire

2020-03-23 16:19:18 UTC

Post by Anders Magnusson
The m30 is affected, but the VS2k have a different SCSI chip so it is
not affected.

Huh? The VS2000/MV2000 have an NCR 5380.

D'oh! Of course it has! I'll blame the lack of coffee. Sorry for the
misinformation.

Too much blood in your caffeine stream Ragge? ;)

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

Anders Magnusson

2020-03-23 17:55:56 UTC

Post by Anders Magnusson
The m30 is affected, but the VS2k have a different SCSI chip so it is
not affected.

Huh? The VS2000/MV2000 have an NCR 5380.

D'oh! Of course it has! I'll blame the lack of coffee. Sorry for the
misinformation.

Too much blood in your caffeine stream Ragge? ;)

Obviously yes :-)

-- R

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

sadness

2020-03-28 22:52:51 UTC