DQDEV support (DNS in ITS) by bictorv · Pull Request #2073 · PDP-10/its

bictorv · 2021-09-07T18:18:36Z

This is a cleanup of the stuff I did back in 2005, previously "documented" at http://its.victor.se/wiki/dqdev. I added support in SUPDUP, and changed the default DNS server to 1.1.1.1.

To make this work, I think the only thing you might need is

iptables -I PREROUTING -t nat -s $YOUR_KLH10_ITS_IP -p udp --dport 53 -j DNAT --to-destination $YOUR_DNS_RESOLVER

It might not be needed if your ITS has a "real" IP address (rather than a tun address). For Chaosnet support, you need a DNS server which knows about DNS.Chaosnet.NET and allows recursion (for your requests).

Before merging, it needs some more/better documentation, and perhaps some testing. (Some of the fixes in DQDEV I admit I haven't understood now - there are some odd comments, but "it seems to work".)

How do I make this a "draft" pull request, or is that not necessary?

with some minor touchups.

but HSTSIX implementation should go in RESOLV

since that was already implemented.

larsbrinkhoff · 2021-09-07T18:44:00Z

Thanks!

It's not really necessary to convert to a draft but it could be nice to signal the intent. There's a small line saying "Still in progress? Convert to draft" in the right sidebar on this page.

and use it in SUPDUP

for simplicity etc

…qdev-support

But for now, use only the HSTSIX of NETWRK (HOSTS3) since it's faster and hard to tell if it failed.

which decides whether to use DQ: (DQXDEV) or DOMAIN: (DQDEV). Default DQ:. Support that in NAME, SUPDUP and COMSAT.

and rename DODQ to DODNS for a little less confusion.

src/sysen2/name.558

larsbrinkhoff · 2021-09-09T05:56:01Z

I think it's polite to ping @cstacy and explain the DQDEV code is being hacked here.

larsbrinkhoff · 2021-09-09T06:24:46Z

Reviewing the changes to NAME took me on a tour in time to see the ups and downs of ITS.

SYSENG; NAME 166        1975-11-11
itshst: 206
        306

SYSENG; NAME 208        1976-10-09
hsits:  .byte 9 ? 106 ? 206 ? 306 ? 354 ? 0

SYSEN2; NAME 482        1983-02-07
        ; List of ITS hosts
hsits:  irps x,,ai ml mc dm

SYSEN2; NAME 493        1983-03-02
        ; List of ITS hosts
hsits:  irps x,,ml mc dm

SYSEN2; NAME 516        1983-12-01
        ; Too-short list of ITS hosts
hsits:  irps x,,ml mc

SYSEN2; NAME 522        1984-04-01
        ; Too-short list of ITS hosts
hsits:  irps x,,mc

SYSEN2; NAME 551        1986-03-01
        ; Too-short list of ITS hosts
hsits:  irps x,,[mc,ai]

SYSEN2; NAME 554        1986-04-02
hsits:  irps x,,[AI,MX,MC]
        sixbit /x/

doc/networking.md

larsbrinkhoff · 2021-09-09T06:39:30Z

src/sysnet/resolv.35

 	CAIL B,1000		; If any of high 3 bits were set,
 	 JRST CVH3N3		;  it must be a HOSTS3 strange-fmt addr.
 	JUMPN B,CVH3N2		; If not zero, then must assume HOSTS2 fmt.
+;; Noone uses Arpanet anymore - but some use Chaosnet.


I WANT TO USE ARPANET!

larsbrinkhoff · 2021-09-09T06:41:19Z

doc/networking.md

-Check out this [external guide](https://its.victor.se/wiki/dqdev)
+To make ITS use DNS like a modern netizen, you need to do the following:
+
+1. Compile the handler for the DOMAIN: device, which interfaces to DNS.


Explain about DQ, DQX, and DOMAIN?

Yes, somewhere. I'm not sure where to document what really, this file seems to be "user friendly" while explaining DQ/DQXDEV and DOMAIN/DQDEV might be more low-level info?

Info can go in INFO. :-)

larsbrinkhoff · 2021-09-09T06:42:32Z

src/sysnet/supdup.325

 	MOVEI A,JCLBUF
 	PUSHJ P,NETWRK"HSTLOOK	;GET HOST NUMBER INTO A, NETWORK NUMBER INTO TT
+ifn DODNS,[
+	 ;; NETWRK failed, try DNS


Why can't NETWRK"HSTLOOK do DNS under the covers?

It would make sense, except it would (under the hood) make many programs larger and more complex/slower. Consider e.g. PEEK which is in a sense a real-time program, and might need to resolve addresses quite a few times per time unit in some circumstances.

The other way around might be more reasonable: let RESOLV use NETWRK under the hood?

But it's really unsustainable to have "all these programs" do both, in a manual fashion, reproducing the same code.

I see your point about size/speed considerations. I don't know what PEEK does with hostnames. Would it make sense for programs to say $$DNS==0 or 1 before .INSRTing NETWRK to control whether to use DNS or not? It seems to me HSTLOOK is the natural entry point for any program that want to look up a host name, and it might be possible to extend this API in a way to make all callers happy.

And then again, NETWRK is not only about name/address resolution, and RESOLV is not only about name<->address but also other DNS things. Hmm.

Ho hum, reading the code is a way to learn. DQDEV already uses NETWRK if asked for it (e.g. domain:HOSTS3;ch;a;no.nocrew.org rather than domain:QUERY;ch;a;no.nocrew.org), so making RESOLV first ask for HOSTS3 and then DNS data should be easy - not efficient, but easy. ;-)

larsbrinkhoff · 2021-09-09T07:48:19Z

/T would be ok. Or people can also just edit the file.

also spelling error fix

Need to develop HOSTNM replacement in resolv or telnet.

Needed after PR PDP-10#2061, where the address is read from the IMP rather than hardcoded.

bictorv · 2021-09-16T06:02:37Z

I'm still working on this - there is plenty still to do - but I find it a bit convenient to have the sum of all the commits visible in a PR. (Maybe there is another way?)

larsbrinkhoff · 2021-09-16T06:59:56Z

I think it's perfectly fine to have an evolving work in progress as a draft pull request.

(plus typo fix)

larsbrinkhoff · 2022-01-12T16:17:50Z

Is this still in (intermittent) progress?

bictorv · 2022-01-12T16:54:11Z

It is, indeed, but as you say, intermittently. Will find the time!

bictorv added 5 commits September 6, 2021 12:24

Code from PI

7b506ed

Fixes from 2005

05717eb

with some minor touchups.

And make it work

9d9f129

Fix for SUPDUP

39e8858

but HSTSIX implementation should go in RESOLV

Remove comments

a473770

since that was already implemented.

bictorv marked this pull request as draft September 8, 2021 05:49

bictorv added 11 commits September 8, 2021 08:07

Merge branch 'PDP-10:master' into dqdev-support

994b1be

Implement HSTSIX in RESOLV

1105b33

and use it in SUPDUP

Default DODQ==0

96f3524

for simplicity etc

Modernize Chaos address domain

50800b2

Test program for DQDEV

11e86ca

Merge branch 'dqdev-support' of https://github.com/bictorv/its into d…

e5433fc

…qdev-support

Try HOSTS3 first, then DNS

3c3d3ec

Try HOSTS3 first, then DNS

2087a62

But for now, use only the HSTSIX of NETWRK (HOSTS3) since it's faster and hard to tell if it failed.

Let RESOLV be configurable using $$DQDQ

be2f0f2

which decides whether to use DQ: (DQXDEV) or DOMAIN: (DQDEV). Default DQ:. Support that in NAME, SUPDUP and COMSAT.

Some documentation

f00488b

and rename DODQ to DODNS for a little less confusion.

Update networking.md

08965d7

larsbrinkhoff reviewed Sep 9, 2021

View reviewed changes

src/sysen2/name.558 Outdated Show resolved Hide resolved

larsbrinkhoff reviewed Sep 9, 2021

View reviewed changes

doc/networking.md Show resolved Hide resolved

larsbrinkhoff reviewed Sep 9, 2021

View reviewed changes

bictorv added 3 commits September 9, 2021 09:05

Minor fixes.

8aa11a7

Finish parameters section

0c40562

Update networking.md

0c653ef

JCL compatibility in strcpy

c585527

also spelling error fix

bictorv added 2 commits September 9, 2021 14:10

Partial DNS support

0e83319

Need to develop HOSTNM replacement in resolv or telnet.

Use NETHST in OWNHST

93c7af9

Needed after PR PDP-10#2061, where the address is read from the IMP rather than hardcoded.

Handle RENMWO better

3a719b9

(plus typo fix)

larsbrinkhoff force-pushed the master branch from 01f5706 to e479dca Compare March 7, 2023 14:22

larsbrinkhoff force-pushed the master branch from 15e616b to 4db8d27 Compare October 7, 2023 06:45

Conversation

bictorv commented Sep 7, 2021

Uh oh!

larsbrinkhoff commented Sep 7, 2021

Uh oh!

Uh oh!

larsbrinkhoff commented Sep 9, 2021

Uh oh!

larsbrinkhoff commented Sep 9, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

larsbrinkhoff commented Sep 9, 2021 via email

Uh oh!

bictorv commented Sep 16, 2021

Uh oh!

larsbrinkhoff commented Sep 16, 2021

Uh oh!

larsbrinkhoff commented Jan 12, 2022

Uh oh!

bictorv commented Jan 12, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

larsbrinkhoff commented Sep 9, 2021 •

edited

Loading