Programmer 7500/MAX_PROGRAMMERS.iso/PROGRAMS/UTILS/HAMRADIO/TCP90112.ZIP/TCP90112.TXT

Text File | 1990-09-20 | 12.1 KB | 359 lines

TCP-Group Digest            Sat, 18 Aug 90       Volume 90 : Issue 112

Today's Topics:
                         Compiling with TC++
                     is it really quiet (2 msgs)
                            LAN Technology
                      Mailbox bug in NOS 900810
                  NOS Net/Rom code broken? (2 msgs)
                    NOS remote listener stack size
               RSPF implementation available  (2 msgs)
                             subscribe..
                             TC C++ & NOS
                     Upgrade PA0GRI's NOS (0812)

Send Replies or notes for publication to: <[e-mail]>
Send requests of an administrative nature (addition to, deletion from the
distribution list, et al) to: <[e-mail]>

Archives of past issues of the TCP-Group Digest are available
(by FTP only) from UCSD.Edu in directory "mailarchives".
----------------------------------------------------------------------

Date: Fri, 17 Aug 90 16:57 GMT
From: <[e-mail]>
Subject: Compiling with TC++
To: [e-mail]

Hi

I compiled plain vanilla NOS 900730 last night under TC++, after
making the patches that were previously posted here (changing .h
files and change to library routine for setjmp). The compile was
fine, and the executable works as expected, except that there
are a bunch of messages about attempting to free invalid pointers
the first few times that I try to do things like ftp telnet or
ping. Has anyone else found this? What's the cause, and is there
a fix?

After a while all the extraneous messages go away, and everything appears
to work as it should.

73 -- Doc

SPAN:   ORION::DEVANS
BITNET: devans@cololasp
Internet: [e-mail]
Snail:  Radiophysics, Inc., 5475 Western Ave., Boulder, Colorado 80301 USA
Analogue switching network: (303)-447-9524

Non-work-related may also go to:
Packet (AX.25): NQ0I @ KE0XA
Packet (TCP/IP): [e-mail] [44.20.0.3]


                              ..._._

------------------------------

Date: Fri, 17 Aug 90 11:38:16 GMT
From: toth!dave (David B. Toth)
Subject: is it really quiet
To: [e-mail]

I have n't seen any tcp-group mail for over a week ...
wonder if my feed is broken, or if it is really this quiet ...
still seeing info-hams, for what that is worth ...
73, Dave

ve3gyq @ ve3gyq.on.can.na

ria.ccs.uwo.ca!toth!dave

------------------------------

Date: Fri, 17 Aug 90 11:21:36 EDT
From: [e-mail] (D. Crompton)
Subject: is it really quiet
To: [e-mail], toth![e-mail]

Everyone is watching CNN!

------------------------------

Date: Fri, 17 Aug 90 22:51:56 -0700
From: Mike Chepponis <[e-mail]>
Subject: LAN Technology
To: [e-mail]

The LAN Technology August 90 edition has a bright yellow starburst on a blue
background of a front cover, with the text "SPECIAL NOS SECTION" - yazoo!

I scrambled to the index, figuring that Phil has done one hell of a lot of PR
work lately!

But, alas and alack, they talk about Network Operating Systems, meaning things
like Microsoft's LAN Manager 2.0 and NetWare...

Oh well, maybe next month!

-Mike

------------------------------

Date: Fri, 17 Aug 90 13:54:09 edt
From: [e-mail] (Phil R. Karn)
Subject: Mailbox bug in NOS 900810
To: [e-mail]

I received the following bug report from Dave Singer. Rather than take
the actual fix he sent, I followed his other suggestion. The code on thumper
has been updated.

Phil


---------- Begin Forwarded Message ----------
To: [e-mail], [e-mail]
Subject: Mailbox bug in NOS 900810
Date: Thu, 16 Aug 90 10:03:16 -0800
From: "David Singer" <[e-mail]>

I found that an AX.25 connection to the NOS mailbox wouldn't take
unless the remote station sent 80 bytes of garbage (as opposed to the
normal NOS requirement of one packet to be thrown away).  The problem
seems to be the call to recvline in mailbox; recvline wants to see a
newline, but the AX.25 station sends a carriage return.  My fix may not
be the best (I think changing the socket to ASCII earlier might be
better), but it got me back on the air -- here it is:

*** mailbox.c   Thu Aug 16 09:48:02 1990
--- mailbox.old Thu Aug 16 09:53:32 1990
***************
*** 131,137 ****
  void *p;
  {
          int s,type;
-         char junk[80];

          freeargs(argc,argv);    /* Args are not used */
          if (Axi_sock != -1)
--- 131,136 ----
***************
*** 153,159 ****
                  /* Eat the line that triggered the connection
                   * and then start the mailbox
                   */
!                 mbxrecvline(s,junk,80,-1);
                  newproc("mbox",2048,mbx_incom,s,(void *)&type,NULL);
          }
          close_s(Axi_sock);
--- 152,158 ----
                  /* Eat the line that triggered the connection
                   * and then start the mailbox
                   */
!                 recvline(s,NULLCHAR,80);
                  newproc("mbox",2048,mbx_incom,s,(void *)&type,NULL);
          }
          close_s(Axi_sock);

-- David Singer, N6TFX   ([e-mail], SINGER at ALMADEN,
                          [e-mail], or N6TFX @ N6LDL.#NOCAL.CA.USA)
----------- End Forwarded Message -----------

------------------------------

Date: Sat, 18 Aug 90 11:43:18 +0200
From: [e-mail] (Ari Husa OH8NUP)
Subject: NOS Net/Rom code broken?
To: [e-mail]

A quick try with the latest NOS version produced an unexpected

    free: WARNING! invalid pointer (0x71f50010) pc = 0x4daf 1c7 proc cmdintrp

during autoexec.net with my freshly compiled NOS trying to

    netrom bcnodes ax0,

where ax0 represents an asy interface to a KISS TNC. The phenomenon
was repeatable at net> prompt as well.

What did I do wrong? (Or, rather, left to be done) I don't assume this
is a bug?

Compiler was TC 2.0 with TASM 1.0, makefile and config.h unmodified
from thumper.bellcore.com src.zip package on a 386 PC.

    Luru

------------------------------

Date: Sat, 18 Aug 90 06:40:33 EDT
From: [e-mail] (Jon Bloom)
Subject: NOS Net/Rom code broken?
To: [e-mail]

I've had a report from one of the locals who uses NET/ROM that it is
now broken.  I don't yet have more details than that he's getting the
WARNING! messages.  In any case, you aren't the only one.

Jon

------------------------------

Date: Fri, 17 Aug 90 19:39:03 edt
From: [e-mail] (Phil R. Karn)
Subject: NOS remote listener stack size
To: [e-mail]

Jon,

When you had your stack overflow in the remote server, had you been
tracing packets?

Phil

------------------------------

Date: Fri, 17 Aug 90 19:57:43 +0200
From: [e-mail]
Subject: RSPF implementation available
To: [e-mail]

I have polished my implementation of RSPF a little bit. It is now
supposed to work with the latest versions of NOS. The implementation
follows version 2.1 of Fred Goldsteins specification.

The Radio Shortest Path First protocol works more or less in the
following way. Each node that runs RSPF tries to figure out which
other nodes are its neighbours. They do so by listening for
Router-to-Router-Hello (RRH) messages sent by other RSPF nodes. When
such a RRH message is received, NOS will try to figure out if the link
is bi-directional by pinging the other station.
All neighbour nodes might not be running RSPF, but they are discovered
by monitoring changes in the ARP table.
The quality of the AX.25 link is assessed by comparing the number of
sent and received AX.25 frames on the channel from the particular node.
The routing table broadcasts contain not only a list of the
originating nodes neighbours, but a list of everybody elses neighbouring
nodes as well. Using this information, it is possible to calculate the
shortest path to any station on the network.
The protocol is described in detail in the RSPF 2.1 specification.

To activate RSPF, do the following.

Set the broadcast address for the ax0 interface:
    ifconfig ax0 44.255.255.255
Configure ax0 as an RSPF interface with quality 8 and a horizon of 32 (hops).
    rspf interface ax0 8 32
Set the interval between RRH messages.
    rspf rrhtimer 900
Define how long it takes until an idle link is suspected to be bad.
    rspf suspecttimer 2000
Set the interval between routing updates.
    rspf timer 900

I haven't really had the possibility to test this latest version of
RSPF, so it would be nice with some feedback. I am especially
interested in knowing if the implementation scales, if it is stable,
etc. RSPF is a rather complex protocol and the implementation is about
twice as large as RIP, so there may of course be bugs.

The code is available with anonymous ftp from sics.se. The filename is
archive/packet/ka9q/nos/rspf.arc. It will be necessary to change the
definition of a pair of functions in NOS from static to gloabl. You
will notice this when compiling the code.

Anders

------------------------------

Date: Sat, 18 Aug 90 11:38:22 +0200
From: [e-mail]
Subject: RSPF implementation available
To: [e-mail]

A small correction: a correct command to set the broadcast address
for interface "ax0" is
    ifconfig ax0 broadcast 44.255.255.255
This automatically creates a routing entry for 44.255.255.255 in the
routing table. If you intend to use RSPF on more than one interface,
the interfaces must be given different broadcast addresses, or the
routing entries will overwrite each other. Exactly how the broadcast
address looks like is of minor concern to NOS, since any IP datagram
sent inside an AX.25 frame for "QST" is regarded as a broadcast
datagram.

Anders

------------------------------

Date: Fri, 17 Aug 90 12:43:23 EDT
From: [e-mail] (Prabhakar Somu)
Subject: subscribe..
To: [e-mail]

Please add my name to the mailing list.
Thanks,
/somu    [e-mail]
--
  ()     __   _ _ _   _    __
  /\    / ') ' ) ) ) ' )  /
 /  )  /  /   / / /   /  /
/__/_ (__/   / ' (_  (__/

------------------------------

Date: Fri, 17 Aug 90 12:50:07 PDT
From: Ron Henderson <[e-mail]>
Subject: TC C++ & NOS
To: [e-mail]

>I compiled plain vanilla NOS 900730 last night under TC++, after
>making the patches that were previously posted here (changing .h
>files and change to library routine for setjmp). The compile was
>fine, and the executable works as expected, except that there
>are a bunch of messages about attempting to free invalid pointers
>the first few times that I try to do things like ftp telnet or
>ping. Has anyone else found this? What's the cause, and is there
>a fix?

I didn't use version '730 compiled by TC++, so I can't comment on it.  I
am, however, using '810 compiled by TC++, with my changes to global.h and
the setjmp module.  I have experienced no problems when using ftp or
telnet.  My use of the PC version is limited to telnet and ftp (at home
I use the version I've ported to the HP Portable Plus, which is another
subject entirely) via the 3c501 packet driver, so I can't say it's
'clean', just that I have had no problems.

>73 -- Doc


--
Ron Henderson  WA7TAS
[e-mail]

------------------------------

Date: Fri, 17 Aug 90  11:49:34 CDT
From: [e-mail]
Subject: Upgrade PA0GRI's NOS (0812)
To: [e-mail]

Yes, i found it on 0812 and the timestamp is 0812 so i call it 0812...
I have integraded Phils changes and added PPP to it as well.
It is undeffed as when building it into NOS something gers over 64k and link abo
So , it is not functional. But undeffing takes it all away..
Split sreen still works an rlogin also works with phils new interface.
Userdoc is running behind now (sorry).
Guess phil wil move nose0812 (.exe) and noss0812.zip (sources) to the pa0gri
dir on thumper.bellcore.com.
Regards, gerard.

------------------------------

End of TCP-Group Digest
******************************