Yeah, it was those articles that i spotted earlier. I'm being a bit of an idiot though, stumbling on how to recompile the damn kernel for bridge support.
On Mon, Dec 10, 2012 at 10:24 AM, Jordi Guillaumes i Pons <jg at jordi.guillaumes.name> wrote:
El 10/12/2012, a les 11:21, Tony Blews <tonyblews at gmail.com> va escriure:
The bridge is running on a different Pi, but what i'm trying to do is get it up and going on the same one as the image.
The problem is that they can't talk to each other. I know there is a workaround, but I can't work it out.
Oh, I understand.
I use vde to overcome that problem. Perhaps these posts will be of help:
http://ancientbits.blogspot.com.es/2012_09_01_archive.htmlhttp://ancientbits.blogspot.com.es/2012/06/simh-39-using-vde-for-fun-and.ht…
Right now I'm running 5 simh simulators, a KLH10 instance and the bridge at one Pi. :)
Jordi Guillaumes i Pons
jg at jordi.guillaumes.name
HECnet: BITXOV::JGUILLAUMES
El 10/12/2012, a les 11:21, Tony Blews <tonyblews at gmail.com> va escriure:
The bridge is running on a different Pi, but what i'm trying to do is get it up and going on the same one as the image.
The problem is that they can't talk to each other. I know there is a workaround, but I can't work it out.
Oh, I understand.
I use vde to overcome that problem. Perhaps these posts will be of help:
http://ancientbits.blogspot.com.es/2012_09_01_archive.htmlhttp://ancientbits.blogspot.com.es/2012/06/simh-39-using-vde-for-fun-and.ht…
Right now I'm running 5 simh simulators, a KLH10 instance and the bridge at one Pi. :)
Jordi Guillaumes i Pons
jg at jordi.guillaumes.name
HECnet: BITXOV::JGUILLAUMES
The bridge is running on a different Pi, but what i'm trying to do is get it up and going on the same one as the image.
The problem is that they can't talk to each other. I know there is a workaround, but I can't work it out.
On Mon, Dec 10, 2012 at 10:17 AM, Jordi Guillaumes i Pons <jg at jordi.guillaumes.name> wrote:
El 10/12/2012, a les 10:26, Tony Blews <tonyblews at gmail.com> va escriure:
What I'd like to do is run the bridge program on the same raspberry that the SIMH image runs on, but I have no idea how to do this. I seem to recall a post about doing it recently, but I can't recall how to view the list archives.
Running the bridge in the Pi is a non-issue. You just have to recompile it and you are set.
As for running DECnet, there is no decnet support in the raspbian kernel. I have tried to build myself a kernel with decnet support (cross-compiling it from another machine) and my first try was a complete failure (the new kernel crashes during boot). I will give it another try...
On the other hand, since I am running KLH10 in the same machine the main interface is hijacked by a virtual TOPS-20 machine, so I won't be able to run native linux decnet anyway... :)
Jordi Guillaumes i Pons
jg at jordi.guillaumes.name
HECnet: BITXOV::JGUILLAUMES
El 10/12/2012, a les 10:26, Tony Blews <tonyblews at gmail.com> va escriure:
What I'd like to do is run the bridge program on the same raspberry that the SIMH image runs on, but I have no idea how to do this. I seem to recall a post about doing it recently, but I can't recall how to view the list archives.
Running the bridge in the Pi is a non-issue. You just have to recompile it and you are set.
As for running DECnet, there is no decnet support in the raspbian kernel. I have tried to build myself a kernel with decnet support (cross-compiling it from another machine) and my first try was a complete failure (the new kernel crashes during boot). I will give it another try...
On the other hand, since I am running KLH10 in the same machine the main interface is hijacked by a virtual TOPS-20 machine, so I won't be able to run native linux decnet anyway... :)
Jordi Guillaumes i Pons
jg at jordi.guillaumes.name
HECnet: BITXOV::JGUILLAUMES
Hi,
Despite me not doing anything different apart from rebooting my machine, TARDIS is back up.
This has spurred me into action.
What I'd like to do is run the bridge program on the same raspberry that the SIMH image runs on, but I have no idea how to do this. I seem to recall a post about doing it recently, but I can't recall how to view the list archives.
Ideally I'd like to be the bridge, the TARDIS SIMH image and Linus DECnet on the same machine. Is this possible?
Tony.
Jordi Guillaumes i Pons
Barcelona - Catalunya - Europa
El 04/12/2012, a les 22:55, "Boyanich, Alastair" <Alastair.Boyanich at au.fujitsu.com> va escriure:
I've not messed around with TOPS-10/20 and only VMS on simh, but there
you need specify which O/S is running inside the VM and specify it with
something like:
set cpu idle=VMS
Well, KLH10 os a different beast. You have to patch the guest OS to do a write to an " idler" device so the simulator knows the guest is not doing anything and it can enter a wait state until next timer tick or interrups happens. This patch is included in the Panda monitor and it works as expected...
... Unless you happen to be using a "high performance" host system (he, and the Pi seems to fit that class lol...). Then you have to build the simulator to use a synchronous implementation for the timers and device interrupts (see https://groups.google.com/forum/m/?fromgroups#!msg/alt.sys.pdp10/vHcQ49tjMa…) and the idler simply does not work at all. Unfortunately it seem there is no fix for this problem (see https://groups.google.com/forum/m/?hl=en&fromgroups#!msg/alt.sys.pdp10/xndb…) so I will have to live with it. Not a big problem in the raspi anyway...
(Apologies for the double post... The bus just caught a bump and my finger moved to the send button when I was completing the message :))
Jordi Guillaumes i Pons
Barcelona - Catalunya - Europa
El 04/12/2012, a les 22:55, "Boyanich, Alastair" <Alastair.Boyanich at au.fujitsu.com> va escriure:
I've not messed around with TOPS-10/20 and only VMS on simh, but there
you need specify which O/S is running inside the VM and specify it with
something like:
set cpu idle=VMS
Well, KLH10 os a different beast. You have to patch the guest OS to do a write to an " idler" device so the simulator knows the guest is not doing anything and it can enter a wait state until next timer tick or interrups happens. This patch is included in the Panda monitor and it works as expected...
... Unless you happen to be using a "high performance" host system (he, and the Pi seems to fit that class lol...). Then you have to build the simulator to use a synchronous implementation for the timers and device interrupts (see https://groups.google.com/forum/m/?fromgroups#!msg/alt.sys.pdp10/vHcQ49tjMa…) and the idler simply does not work at all. Unfortunately it seem there is no fix for this problem (see https://groups.google.com/forum/m/?hl=en&fromgroups#!msg/alt.sys.pdp10/xndb…) so I will have to liv
On Dec 4, 2012, at 4:55 PM, Boyanich, Alastair wrote:
Hi Jordi,
I've not messed around with TOPS-10/20 and only VMS on simh, but there
you need specify which O/S is running inside the VM and specify it with
something like:
set cpu idle=VMS
If I get it wrong or omit it, then it will sit there wasting a huge
amount of cycles on my alpha. Off the top of my head, I suspect simh is
actually looking for instruction patterns for the idle loop and
detecting based upon that, which would differ from O/S to O/S. Perhaps
TOPS-10/20 requires something similar?
Yes, that's what SIMH does. Depending on the processor type, this code may be rather fragile. On a PDP-11, most (though not all) idle loops have a WAIT in them somewhere. On VAX, there is no such thing, and the code looks for "scheduler idle loop" patterns. It's pretty ugly and your particular OS may not work well. For example, NetBSD 6 does not match any of the VAX idle patterns, and in fact it's not clear any idle handling can be done sanely there without help from the NetBSD code...
The first step would be to make sure the idle checker is set for your OS. If that still doesn't work, you may need to read the emulation code to see what the checker actually does, to determine if it needs to be updated to handle the OS you're running.
paul
Hi Jordi,
I've not messed around with TOPS-10/20 and only VMS on simh, but there
you need specify which O/S is running inside the VM and specify it with
something like:
set cpu idle=VMS
If I get it wrong or omit it, then it will sit there wasting a huge
amount of cycles on my alpha. Off the top of my head, I suspect simh is
actually looking for instruction patterns for the idle loop and
detecting based upon that, which would differ from O/S to O/S. Perhaps
TOPS-10/20 requires something similar?
Al.
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE]
On Behalf Of Jordi Guillaumes i Pons
Sent: Monday, 3 December 2012 8:25 PM
To: hecnet at Update.UU.SE
Subject: [HECnet] KLH10 on Raspberri Pi / Nodelist update
Hello,
Please add to the HECnet nodelist:
7.79 BITXT0 TOPS-20(PANDA) - KLH10 - Raspbery Pi
I am running this simulator in the same Pi which currently holds 5
more VMs. I
will see if it keeps stable, and if it does I will run it permanently.
Issues to solve
- KLH10 does not idle in the Pi. I'll check the idler device code to
see if I can
guess what is happening.
- I tried to give KLH10 its own bridge device, plugged into the rest
of my
network via VDE switches and plugs, but it does not work. I have lots
of IPBLP
bugchecks and IP networking is unusable (weird enough, DECNET and LAT
seem to work). It works fine if I use the main host interface in
non-exclusive
mode (but then I don't get LAT).
- KLH10 frontend can't load *.EXE files.
- TOPS-10 does not run well. I used a x86 KLH to dump the load code in
SAV
format and then I can boot TOPS-10, but it hungs after a while.
On the other hand, I'm thinking about building a blinkenlights device
using
the GPIO pins of the Rasp-pi basing it on the actual parallel port
blinkenlights
panel driven by the "lites" device... :) It could be another nice
hobby project...
Jordi Guillaumes i Pons
jg at jordi.guillaumes.name
HECnet: BITXOV::JGUILLAUMES
