[Wanted] Dual ROSS RT626@200 MHz

[Joël BERTRAND]

Hello,

I'm looking for one (or several) dual cpu module(s) for an old SS20
that runs NetBSD (HEAD). One of its modules is defective (TOD register
fault).

I use this old workstation to port NetBSD on Leon4 ITX development
card. If someone has this kind of Mbus module, please contact me.

Thanks in advance,

JB

[BERTRAND Joël]

I have a 150Mhz module laying around somewhere... I needs to be repasted
but I think it works other than that. I have dual SM-81s in my SS20 now
so..... there isn't alot of point in a single 150Mhz hypersparc on my
end since I don't have 2 of them.
Have you considered that this might just be a kernel bug? I think I
remember there being timing issues on the SMP kernels did that ever get
completely resolved? Also does that leon card plug into something or is
it standalone just curious... you said ITX but it sounds like it is PCI
or something in which case it might be interesting to put it in a
Sparc64 server that has PCI-X.

Thanks a lot for your answer.

I have a lot of SS20 with 2*SM71, 2****@200 and my main workstation
with 4****@200 to develop on sparc32 as I have some trouble to run
NetBSD on Leon4-ITX system.

Target board is this one :
http://www.gaisler.com/index.php/products/boards/gr-leon4-itx

It natively runs an old linux release. I have installed a [dr]ecent
debian release but debian has now dropped sparc32 support. Thus, I'm
trying to port NetBSD on this board. But, I have to check that all
required programs (http://www.rpl2.net) can run without modifications on
NetBSD/sparc. Each build process takes with -j4 several hours to
complete :-(

On regular SS20, SMP kernel seems to run like a charm. Current
installation was made from daily binaries (maybe with a 5.99.56 kernel
?) and I've only seen a few "clock backward" message. This SS20 contains
: two 73GB disks (in raid1), 4 and now 2 ***@200MHz, 448 MB, a 4MB
VRAM and one HME ethernet interface (and of course some extra fans ;-)
). It is stable, more stable than it was under Solaris 9.

Best regards,

JKB

[BERTRAND Joël]

I have a 150Mhz module laying around somewhere... I needs to be repasted
but I think it works other than that. I have dual SM-81s in my SS20 now
so..... there isn't alot of point in a single 150Mhz hypersparc on my
end since I don't have 2 of them.
Have you considered that this might just be a kernel bug?

I have forgotten. CPU is invalidated by post when diag-switch? is set
(TOD register test failed). If diag-switch? isn't set, sometimes three
or four processors were detected but system is not stable. By
bissection, I have found faulty module and done some test with only this
module. I'm sure it's really faulty.

Regards,

JKB

[Martin Husemann]

You could also use some newer ultra sparc and run a 32bit kernel and
sparc userland on it for developement, but it might cause some minor
differences if the program's configure detects a v9 cpu.

E.g. I have a Ultra 2 with two 300 MHz cpus and 2 GB ram, which is pretty
nice for natively compiling sparc stuff. (And it is even more stable than
my dual SS20, even under heavy load.)

Martin

[Manuel Bouyer]

Post by Martin Husemann
You could also use some newer ultra sparc and run a 32bit kernel and
sparc userland on it for developement, but it might cause some minor
differences if the program's configure detects a v9 cpu.
E.g. I have a Ultra 2 with two 300 MHz cpus and 2 GB ram, which is pretty
nice for natively compiling sparc stuff. (And it is even more stable than
my dual SS20, even under heavy load.)

FYI, the packages in ftp.netbsd.org/pub/pkgsrc/packages/NetBSD/sparc
are built on 2 blade 100.

[Joël BERTRAND]

Post by Manuel Bouyer

Post by Martin Husemann
You could also use some newer ultra sparc and run a 32bit kernel and
sparc userland on it for developement, but it might cause some minor
differences if the program's configure detects a v9 cpu.
E.g. I have a Ultra 2 with two 300 MHz cpus and 2 GB ram, which is pretty
nice for natively compiling sparc stuff. (And it is even more stable than
my dual SS20, even under heavy load.)

FYI, the packages in ftp.netbsd.org/pub/pkgsrc/packages/NetBSD/sparc
are built on 2 blade 100.

Hello,

I have installed a Blade2k/SMP, but I'm not sure that this workstation
is stable enough to develop. If I haven't seen for a long time spurious
reboots I've seen with 6.1 kernel, I cannot use X server. I have tried
XVR500 and Creator3D (ffb2+) framebuffer without success. startx is a
new function to do a quick an dirty reboot (without panic or usable
message) :-(

Best regards,

JKB

[Martin Husemann]

Post by Joël BERTRAND
I have installed a Blade2k/SMP, but I'm not sure that this
workstation is stable enough to develop. If I haven't seen for a long time
spurious reboots I've seen with 6.1 kernel, I cannot use X server. I have
tried XVR500 and Creator3D (ffb2+) framebuffer without success. startx is a
new function to do a quick an dirty reboot (without panic or usable
message) :-(

You should run -current on that, it is way better.

I don't know about XVR500, but XVR100 and Creator3D work just fine for
me - if you can help with some debugging off-list, we should be able to
sort out any crashes you have.

Martin

[Joël BERTRAND]

Post by Martin Husemann

Post by Joël BERTRAND
I have installed a Blade2k/SMP, but I'm not sure that this
workstation is stable enough to develop. If I haven't seen for a long time
spurious reboots I've seen with 6.1 kernel, I cannot use X server. I have
tried XVR500 and Creator3D (ffb2+) framebuffer without success. startx is a
new function to do a quick an dirty reboot (without panic or usable
message) :-(

You should run -current on that, it is way better.

I have installed -current.

Post by Martin Husemann
I don't know about XVR500, but XVR100 and Creator3D work just fine for
me - if you can help with some debugging off-list, we should be able to
sort out any crashes you have.

Sure. But at this time, I'm unable to post a BR as my blade immediatly
reboots and erases screen. I have tried to check if some information
were sent over serial line without success.

Regards,

JKB

[Eduardo Horvath]

Post by Joël BERTRAND
Sure. But at this time, I'm unable to post a BR as my blade immediatly
reboots and erases screen. I have tried to check if some information were sent
over serial line without success.

You won't get serial port output unless the serial port is the console.

Try setting the OBP "intput-device" and "output-device" to "serial" and
see if you can fire up X. Unlike other machines, SPARCs should do
at least some initialization of display devices even if they are not the
console, so depending on the kernel driver. If you can do that then
both kernel and firmware output should go to the serial port.

Another thing to look at is turning up the "verbosity" and turning on
"diag-switch?" If the machine is taking a watchdog reset or otherwise
dying in the firmware, this might turn on some output on the serial port
even though the console is the display.

Finally, if nothing else works, you can fiddle with the kenel to swap the
kernel cn_tab back to the serial port.

Eduardo

[Jim MacKenzie]

-----Original Message-----
From: port-sparc-***@NetBSD.org [mailto:port-sparc-***@NetBSD.org] On
Behalf Of BERTRAND Joël
Sent: March-04-14 2:26 PM
To: chase rayfield
Cc: port-***@NetBSD.org
Subject: Re: [Wanted] Dual ROSS ***@200 MHz

I have a lot of SS20 with 2*SM71, 2****@200 and my main
workstation with 4****@200 to develop on sparc32 as I have some trouble to
run NetBSD on Leon4-ITX system.
===

Not sure if it's useful, but I have an SS20 (still running the latest 4.0
NetBSD release, because of the SMP problem) that has a pair of RT620/625 180
MHz ROSS CPUs. It's powered down but ran as recently as a few months ago,
and I don't mind firing it up if anyone needs me to test something on it.

Have an SS10 with a pair of RT620/625 90 MHz CPUs in it too. RT number as
reported by /var/log/messages, anyway.

Jim

[Mouse]

[not sure of attributions - the quoting is unusual - but I think these
two quotes are from different people.]

Post by BERTRAND Joël
NetBSD on Leon4-ITX system.
Not sure if it's useful, but I have an SS20 (still running the latest
4.0 NetBSD release, because of the SMP problem) that has a pair of
RT620/625 180 MHz ROSS CPUs. [...]

I use SS20s as my favourite screen-&-keyboard machines, largely so I
can use Sun type-3 keyboards (my second favourite keyboard, and my
first favourite of those I have) and 24bpp colour (via cg14s).

They're all uniprocessor - I'm running pre-MP NetBSD on them. Here are
boot-time CPU reports for the ones I can easily check:

cpu0 at mainbus0: TMS390Z50 v0 or TMS390Z55 @ 75 MHz, on-chip FPU
cpu0: physical 20K instruction (64 b/l), 16K data (32 b/l), 1024K external (32 b/l): cache enabled

cpu0 at mainbus0: RT620/625 @ 125 MHz, on-chip FPU
cpu0: 256K byte write-back, 64 bytes/line, sw flush: cache enabled

cpu0 at mainbus0: TMS390Z50 v0 or TMS390Z55 @ 75 MHz, on-chip FPU
cpu0: physical 20K instruction (64 b/l), 16K data (32 b/l), 1024K external (32 b/l): cache enabled

(the first and third are different machines, but apparently with
identical CPUs).

I have a few spare Mbus cards, but I doubt any; of them are dual-CPU.
Joël, if you don't find anything elsewhere, I can dig them out and
check, but note also I'm in Canada, so if, as your address implies,
you're in France, getting them to you might cost more than it's worth.

/~\ 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