Enable NCP; make Arpanet great again

[larsbrinkhoff]

• Turn on NCPP.
• Add transparent conversion between short and long leader messages to IMP-host interface.
• Add IMP-host interface to the PDP-10 emulator(s).
• Check if ITS will talk Arpanet to an IMP, fix if not. (Can NCP and TCP/IP coexist?)
• Make sure Apanet clients and servers work: check https://github.com/PDP-10/its/issues?q=label%3Aarpanet

[larsbrinkhoff]

FYI @Quantx

[larsbrinkhoff]

Turning on NCPP, I noticed SYSTEM;CORE uses DMI in conditionals. That needs to be updated to use KAIMP or DMIMP.

EDIT: Adding NCP to the already big KA ITS makes it necessary to move up SALV even higher.

[larsbrinkhoff]

Maybe the KS10 can't do Arpanet since the new IMP code doesn't have NCP. The KA10 and KL10 use the "IMPOLD WTHNCP" code.

[larsbrinkhoff]

Various updates to the KA10 and H316 are emulators are needed. With things fixed, I can attempt to make a TELNET connection back to the same ITS host.

*telnet^K!
User TELNET.753
TELNET>NCP protocol (not TCP)
TELNET>Open connection to 6
 Trying...  Host dead due to random network lossage
TELNET>
NET: RST TIMEOUT HST= 6  18:01:31
NET: CTL LNK RFNM TIMEOUT HST 6  18:01:38
NET: RRP W-O RST HST 6  18:01:38

What's going on here is that before making the TELNET connection, ITS sends a RST (reset) message to the peer. This is probably because ITS hasn't had any contact with the peer since booting. However, there's a timeout waiting for the RRP (reset reply) acknowledgement. The RRP is sent, but somewhat later. Then there's a complaint that there was an RRP received but no RST in progress.

[larsbrinkhoff]

CC @charlesUnixPro, @SpareTimeGizmos

[larsbrinkhoff]

I made a handwritten note, as is the time-honored tradition.

[larsbrinkhoff]

Note to self: check if this works for incoming Arpanet connections. The IMPUS (Arpanet) address is unconditionally commented out and replaced with IMPUS3 (IP address).

IMPRFQ: CAIL E,NETSRS   ;IF < 1000, START JOB "NETRFC"
         JRST IMRFQ5
        MOVE T,IMPHTN(H)
;       CAIE T,<IMPUS_-6>+<IMPUS&77>_9
        CAME T,[IMPUS3-NW%ARP]  ; Compare with our own host (minus net #)
         SKIPL NETUSW
          CAIA
           JRST IMREFU  ;REFUSE CONNECTION
        HRROI T,NTRFCL
        PUSHJ P,NUJBST  ;LOAD SERVER

[larsbrinkhoff]

See also #2349 "ITS code duplicated between NET and NCP." There are assembler errors if NCPP=1 and INETP=0.

[larsbrinkhoff]

Work in progress on these branches:

• ITS updates - https://github.com/PDP-10/its/tree/lars/ncp
• H316 conversion between old/new 1822 leaders - https://github.com/larsbrinkhoff/open-simh/tree/lars/convert-1822
• PDP-10 host-IMP interface - https://github.com/larsbrinkhoff/ka10-simh/tree/lars/ncp
• Fixing the Linux NCP - https://github.com/larsbrinkhoff/linux-ncp/tree/lars/ncp

[larsbrinkhoff]

Note to self: look into #1475
On DM a network demon is summoned from SYS;ATSIGN RFCnnn. All other ITSes start a network server from DEVICE;LBSIGN RFCnnn.

[larsbrinkhoff]

I'm testing ITS together with the "Linux NCP" (need a better name?) I was "pinging" from the Linux host to ITS, i.e. sending an ECO (echo) and expecting an ERP (echo reply) back. I see the UDP packets arriving safely on the ITS side, but I wasn't seeing any replies. I checked the code, and indeed ITS is supposed to reply. After a few reruns and several pings later, I checked the logs by chance and did see one ERP in the IMP and Linux logs. So similar to the very late RRP (reset reply) I mentioned above, there seems to be some delay going on. I put breakpoints on the IMP interrupt handlers and didn't see anything. Strangely, when I TELNET from ITS to itself, I do hit the breakpoints. The ITS messages are about 24 bits longer, for some unknown reason. The Host-Host protocol specification says there is an "M₃" field after an NCP message, with 0 or more bits of padding. I figured 0 would be ok, but maybe not.

[larsbrinkhoff]

Indeed there was no interrupt raised for incoming data, because I had missed to handle a corner case. Now that I do, ITS will reply to the ECO messages from "Linux NCP".

Testing with two ITS hosts on the network, I get a reaction to a TELNET connection: LOGIN RFC027 0 HST000. This means there was a "request for connection" on socket 27 octal = 23 decimal, which is the correct number for "new TELNET". HST000 doesn't look right, though.

Still, clearly a step forward. The hosts successfully exchange RST (reset) and RRP (reset reply) messages.

[larsbrinkhoff]

I rebuilt NETRFC with DEMONP=0 and made the LBSIGN links for RFC027 (telnet). This seems to have done the trick. I made a prototype TELNET client to tickle ITS, and I'm getting some data back.

[larsbrinkhoff]

Various updates to my NCP and TELNET client later, I get this session. It's not perfect, e.g. no echo handling or cbreak mode yet, but it shows promise. -cn 70 means client new protocol to host 70 = octal 106.

$NCP=ncp006 ../linux-ncp/src/telnet -cn 70
TELNET to host 106.
Unknown ITS PDP-10
It's a lovely day to be a turist!


KA ITS.1653. DDT.1548.
TTY 53
You're all alone, Fair share = 79%
Welcome to ITS!

For brief information, type ?
For a list of colon commands, type :? and press Enter.
For the full info system, type :INFO and Enter.

Happy hacking!
lars^F
lars^F
KA   LARS  
FREE BLOCKS #2=32 #3=916 #0=650 #1=829 
  2   BACKUP 23     2   11/30/2016 03:56:40
  0   BUBUNV 78     1   11/30/2016 03:56:40
  1   LARS   LOGIN  1 ! 11/23/2024 13:41:32
  0   LITES  79     1   11/30/2016 03:56:40
  3   LORENZ 68     1   11/30/2016 03:56:40
  0   MINSKY TRON   1   11/30/2016 03:56:40
  1   MUNCH  340    1   11/30/2016 03:56:40
  1   MUNCH  FASL   1   11/21/2024 16:31:57
  1   MUNCH  LISP   1 1/30/2016 03:56:40
  2   MUNCH  UNFASL 1   11/21/2024 16:31:57
  3   QRCODE 8      6   11/30/2016 03:56:40
  2   TS     LORENZ 1   11/21/2024 16:21:13
  2   TS     MINSKY 1   11/21/2024 16:21:10
  2   TS     MUNCH  1   11/21/2024 16:21:06
  3   TS     TVBROT 1   11/21/2024 16:21:16
  2   TT2500 57     1   11/30/2016 03:56:40
  2   TVBROT 209    1   11/30/2016 03:56:40
  1   UTNAM  20     1   11/30/2016 03:56:40
  3   XOROSH 2      1   11/30/2016 03:56:40
  0   _MI OUTPUT 0 ! 12/18/2024 17:37:53
*
1/    LOGIN? ^[^[u
$$u
KA ITS 1653  Console 53 Free. 20:45:45