14 Aug
2020
14 Aug
'20
9:06 p.m.
Gentlepeople,
I just found a small but nasty bug in the DECnet/E NML (network management listener).
When it is asked for information that comes back in a stream of messages, it may hit a
transmit queue full condition. That's more likely if the requesting node does ACK
delaying, as PyDECnet does, but it can also happen if the network latency is significant.
For that status, the transmit is supposed to be retries, but due to a one line error in
the code it was not. The attached patch file fixes that. Feed it to ONLPAT, with
"file to patch" the NML image, which normally would be DECNET$:NCP.TSK.
Tested on V10.1; the patch may well work on older versions too.
If you have NML enabled on your HECnet system, the network mapper will probably connect
and ask for a bunch of things. Those requests are likely to exercise this bug, so for
such systems I urge you to install the patch.
paul
14 Aug
14 Aug
11:22 p.m.
On 2020-08-14 21:06, Paul Koning wrote:
Gentlepeople,
I just found a small but nasty bug in the DECnet/E NML (network management listener).
When it is asked for information that comes back in a stream of messages, it may hit a
transmit queue full condition. That's more likely if the requesting node does ACK
delaying, as PyDECnet does, but it can also happen if the network latency is significant.
For that status, the transmit is supposed to be retries, but due to a one line error in
the code it was not. The attached patch file fixes that. Feed it to ONLPAT, with
"file to patch" the NML image, which normally would be DECNET$:NCP.TSK.
Tested on V10.1; the patch may well work on older versions too.
If you have NML enabled on your HECnet system, the network mapper will probably connect
and ask for a bunch of things. Those requests are likely to exercise this bug, so for
such systems I urge you to install the patch.
Awesome, Paul.
I love that RSTS/E is getting some love.
I seem to remember you've posted a fix or two before. You really should
start collecting them and making them available in some organized form
somewhere. So that people can start knowing that if they want something
RSTS/E related, that's the place to go and check.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
15 Aug
15 Aug
5:06 p.m.
On 8/14/20 5:22 PM, Johnny Billquist wrote:
Awesome, Paul.
I love that RSTS/E is getting some love.
Agreed, loudly!
I seem to remember you've posted a fix or two
before. You really should
start collecting them and making them available in some organized form
somewhere. So that people can start knowing that if they want something
RSTS/E related, that's the place to go and check.
That would be great. I myself don't have as much time as I'd like to
hack on PDP-11s. Every few weeks or months I come back to RSTS or RSX
etc and then have to try to remember what happened on various mailing
lists and go find the stuff I want to work with. Having a central
repository for current, useful things like Paul's recent patches would
be great.
Maybe Brett Bump should put up a page for this sort of stuff on rsts.org?
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
7:44 p.m.
On Aug 14, 2020, at 5:22 PM, Johnny Billquist <bqt
at softjar.se> wrote:
On 2020-08-14 21:06, Paul Koning wrote:
Yes, I've created a number of them over the years. Good suggestion, I'll try to
make the collection and find a home for it.
paul
Gentlepeople,
I just found a small but nasty bug in the DECnet/E NML (network management listener).
When it is asked for information that comes back in a stream of messages, it may hit a
transmit queue full condition. That's more likely if the requesting node does ACK
delaying, as PyDECnet does, but it can also happen if the network latency is significant.
For that status, the transmit is supposed to be retries, but due to a one line error in
the code it was not. The attached patch file fixes that. Feed it to ONLPAT, with
"file to patch" the NML image, which normally would be DECNET$:NCP.TSK.
Tested on V10.1; the patch may well work on older versions too.
If you have NML enabled on your HECnet system, the network mapper will probably connect
and ask for a bunch of things. Those requests are likely to exercise this bug, so for
such systems I urge you to install the patch.
Awesome, Paul.
I love that RSTS/E is getting some love.
I seem to remember you've posted a fix or two before. You really should start
collecting them and making them available in some organized form somewhere. So that people
can start knowing that if they want something RSTS/E related, that's the place to go
and check.
Johnny
16 Aug
16 Aug
1:23 a.m.
On 2020-08-15 19:44, Paul Koning wrote:
I have a ton of patches for RSX, but have never figured out a good way
of distributing them and making it easy for people to apply them.
If you come up with any good ideas, I might copy it. ;-)
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
On Aug 14, 2020, at 5:22 PM, Johnny Billquist
<bqt at softjar.se> wrote:
On 2020-08-14 21:06, Paul Koning wrote:
Yes, I've created a number of them over the years. Good suggestion, I'll try to
make the collection and find a home for it. Gentlepeople,
I just found a small but nasty bug in the DECnet/E NML (network management listener).
When it is asked for information that comes back in a stream of messages, it may hit a
transmit queue full condition. That's more likely if the requesting node does ACK
delaying, as PyDECnet does, but it can also happen if the network latency is significant.
For that status, the transmit is supposed to be retries, but due to a one line error in
the code it was not. The attached patch file fixes that. Feed it to ONLPAT, with
"file to patch" the NML image, which normally would be DECNET$:NCP.TSK.
Tested on V10.1; the patch may well work on older versions too.
If you have NML enabled on your HECnet system, the network mapper will probably connect
and ask for a bunch of things. Those requests are likely to exercise this bug, so for
such systems I urge you to install the patch.
Awesome, Paul.
I love that RSTS/E is getting some love.
I seem to remember you've posted a fix or two before. You really should start
collecting them and making them available in some organized form somewhere. So that people
can start knowing that if they want something RSTS/E related, that's the place to go
and check.
Johnny
1621
days inactive
1622
days old
4 comments
3 participants
participants (3)
-
bqt＠softjar.se -
mcguire＠neurotica.com -
paulkoning＠comcast.net