Subject: Linux-Development Digest #963
From: Digestifier <Linux-Development-Request@senator-bedfellow.MIT.EDU>
To: Linux-Development@senator-bedfellow.MIT.EDU
Reply-To: Linux-Development@senator-bedfellow.MIT.EDU
Date:     Fri, 29 Jul 94 22:13:05 EDT

Linux-Development Digest #963, Volume #1         Fri, 29 Jul 94 22:13:05 EDT

Contents:
  Re: Hayes ESP suport (Brian R. Curnow, Jr.)
  Debugger breakpoints FIXED! (Darren Hiebert)
  1.1.36 make problem: 'NULL' undeclared (Andreas Joppich)
  Re: Realtime? (Matthias Urlichs)
  Truecolor support for XFr (Riku Saikkonen)
  Re: Found "talk" bug, fixing. (Nick Holloway)
  How do I find out what floppy drives are hooked up? (Salvatore Valente)
  Re: Voice Mail cards. (Vassili Leonov)
  Q: Net Diagram &/or SNMP App for X11 (Dane Jasper)
  Free Network Management System -- port to linux? (Jude George)
  Re: SLIP, CSLIP, PPP and modems (Bart Kindt)
  Re: Linux Performance Enhance ? (Starcon SysAdmin)

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

Reply-To: brianc@curnow.win.net (Brian R. Curnow, Jr.)
From: brianc@curnow.win.net (Brian R. Curnow, Jr.)
Date: Fri, 29 Jul 1994 09:30:44 GMT
Subject: Re: Hayes ESP suport

I'm just monitoring this newsgroup since I'm thinking about trying
Linux out, but I was checking out the same thing recently.  I
contacted Hayes via their BBS (800) US-HAYES, and the tech folks
can provide you source/docs that do a fair job of describing the
advanced features if you are interested in writing a driver for
<cool-new-OS here>.

I would also assume that the normal 16550 UART part of the board
would work without special support.

Brian

  In article <3C3QB3G@kmk.rhein-main.de>, Kai Kretschmann
(kai@kmk.rhein-main.de) writes: >Hello,
>
>did  anyone already try  an  Hayes ESP  serial card  under  Linux?  It
>should  be hardware compatible to   normal UART's with some functional
>restrictions.  Full   features might   only   be available  by special
>drivers, any experiences?
>-- 
>Kai Kretschmann,
>  >>>   FidoNet:  2:2461/312, 21:491/1161, 16:100/6006   <<<
>  >>>   Internet: kai@fix.kmk.rhein-main.de              <<<
>  >>>   FAX/BBS:  +49-6172-305379                        <<<
>


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

From: darren@hunan.rastek.com (Darren Hiebert)
Subject: Debugger breakpoints FIXED!
Date: Fri, 29 Jul 1994 20:05:13 GMT

fox@graphics.cs.nyu.edu (David Fox) writes:

>] I just noticed that gdb won't stop at breakpoints under version
>] 1.1.32.  I rebooted into 1.0.9 and they worked fine.
>
>I'd be interested to hear if anyone besides me has experienced
>this.  By the way, 1.1.23 works for me too, and 1.1.34 does
>not, but I can't speak about the ones between 23 and 32.

Though I had reported the same problem yesterday, I got 1.1.37 today and the
problem had disappeared. Hurry for quick turnarounds on bug fixes!! Open
development proves itself again.

-- 
============================================================================
Darren Hiebert                  "If I have not seen further, it is because
darren@rastek.com                   giants were standing on my shoulders."
============================================================================

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

From: aj@z2-db11.ms.DeTeMobil.de (Andreas Joppich)
Subject: 1.1.36 make problem: 'NULL' undeclared
Date: 26 Jul 1994 17:33:46 GMT
Reply-To: aj@ms.DeTeMobil.de


Today i got the new 1.1.36 version. After patching the 1.1.35 and
running the various commands below I get the following error-message:

make mrproper ; make config
make dep ; make clean ; make

 
[..lots of stuff deleted..]

/usr/src/linux/include/linux/timer.h: In function `init_timer':
In file included from psaux.c:28:
/usr/src/linux/include/linux/timer.h:85: `NULL' undeclared (first use this function)
/usr/src/linux/include/linux/timer.h:85: (Each undeclared identifier is reported only once
/usr/src/linux/include/linux/timer.h:85: for each function it appears in.)
make[2]: *** [psaux.o] Error 1
make[2]: Leaving directory `/usr/src/linux/drivers/char'
make[1]: *** [driversubdirs] Error 1
make[1]: Leaving directory `/usr/src/linux/drivers'
make: *** [linuxsubdirs] Error 1


Am I doing something wrong or is it a bug ? 
-- 
_______________________________________________________________
Andreas Joppich                              aj@ms.DeTeMobil.de
DeTeMobil GmbH                        
D-48153 Muenster                         Phone +49-251-977-2943              
Germany                                  Fax   +49-251-977-2949
===============================================================
   The above statements are my privat and personal opinions 
       and not represantive for the DeTeMobil company !  

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

From: urlichs@smurf.noris.de (Matthias Urlichs)
Subject: Re: Realtime?
Date: 29 Jul 1994 22:16:09 +0200

In comp.os.linux.development, article <JOHNSONM.94Jul28232913@calypso-2.oit.unc.edu>,
  johnsonm@calypso-2.oit.unc.edu (Michael K. Johnson) writes:
> 
> However, as long as you don't need that kind of support (if you don't,
> please don't throw the word "realtime" around), Linux is not a bad
> platform for gathering data.

Don't forget that one big contributor to high response times is the time it
takes to page a program in from disk. Surprise, Linux has a process-specific
flag to prevent swapping. It's currently used for /sbin/init, but should be
usable for other processes as soon as somebody adds a system call to
actually set the flag. ;-)

-- 
"William Safire would have a cow, but somehow that doesn't disturb me."
        -- Evan Hunt (evanh@sco.com)
-- 
Matthias Urlichs        \ XLink-POP N|rnberg  | EMail: urlichs@smurf.noris.de
Schleiermacherstra_e 12  \  Unix+Linux+Mac    | Phone: ...please use email.
90491 N|rnberg (Germany)  \   Consulting+Networking+Programming+etc'ing     42

Click <A HREF="http://smurf.noris.de/~urlichs/finger">here</A>.

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

Subject: Truecolor support for XFr
From: riku.saikkonen@compart.fi (Riku Saikkonen)
Date: Sat, 30 Jul 94 00:36:00 +0200

>I've been wondering if there is any effort being made to add truecolor
>and/or hicolor support to XFree86.  I have a truecolor capable card,
>and it really makes me puke to have to boot DOS and run Windoze in
>order to get 24-bit color.  Can anyone enlighten me?

XFree86 doesn't support it (yet), but svgalib does. That means that most
non-X graphics programs will work in 24-bit...

-=- Rjs -=- riku.saikkonen@compart.fi - IRC: Rjs
"It is useless to meet revenge with revenge: it will heal nothing."
 - J.R.R. Tolkien


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

From: Nick.Holloway@alfie.demon.co.uk (Nick Holloway)
Subject: Re: Found "talk" bug, fixing.
Date: Tue, 26 Jul 1994 08:43:33 +0000

In <30rsk3$ajh@usenet.srv.cis.pitt.edu> Doug DeJulio wrote:
>                                              Given a remote address,
> what's the quickest way to get the address of the local interface that
> traffic to that address goes through?

This is something I need a solution to also.  I am trying to use
chat2 with Perl, and that does a bind to the address returned by
gethostbyname(`hostname`).  Same problem.

In my case, it returns the address of the dummy interface :-(  This is
of course non-routable.  It is by no means a Linux specific problem,
since you may have a firewall machine where the internal network is
non-routable (from the global point of view).

>                                        I know I can do it by leafing
> through /proc/net/route, but is there some more elegant or portable
> way?

This would be a Linux specific solution -- I'd prefer something neater.

-- 
Nick Holloway |  `O O'  | Home: Nick.Holloway@alfie.demon.co.uk
[aka `Alfie'] | // ^ \\ | Work: nickh@parallax.demon.co.uk

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

From: svalente@athena.mit.edu (Salvatore Valente)
Subject: How do I find out what floppy drives are hooked up?
Date: 30 Jul 1994 00:18:56 GMT


Hello.

I'm trying to port to Linux a little program that gives a bit of
information about the hardware setup.  One of the options to this
program displays a list of floppy drives.  Currently, what I'm doing
is this: run "dmesg", looking for the line of the form:
"Floppy drive(s): fd0: 1.44M fd1: 1.2M"
and parse it into a list of disk drives.

This actually works suprisingly well.

But, it's ugly.  So, I'm trying to figure out: once Linux autodetects
the drives, does it store this information anywhere so it's available
through a regular syscall, or maybe through /proc or something?

Thanks.

-- 
/*                 Sal Valente   <svalente@athena.mit.edu>               */
/*  All opinions stated here are shared by my school, my employer, my    */
/*  friends, my family, a large group of people I've never met, the      */
/*  entire cast of "Taxi," President Clinton, the Pope, Elvis, and you.  */

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

From: vassili@cs.sunysb.edu (Vassili Leonov)
Subject: Re: Voice Mail cards.
Date: 29 Jul 1994 18:44:51 GMT

Frank Lofaro (ftlofaro@unlv.edu) wrote:

: And it is our decision if we want to (and do) reverse-engineer anything.

: And then release source code ourselves. We would not be under
: non-disclosure.
I believe there are always resonable vendors that give you the
programming details of their hardware without non-disclosure. The whole
idea of non-dislosure hardware is aliend to me. 
You see - we live here under Linux in the realm of the free software.
It should be a corresponding realm of an open hardware. I believe that
you can manufacture open hardware and still make a profit. You don't
have to spend much on the advertising and market penetration in the
first place. 
Speakin of voice boards - I have a non-disclosure for the BT114 board -
and let me disclose that fact that I don't think that writing a driver
for them is much fun. Interface is quite a mess. The board is very good.
Maybe they can be explained why it's nice to have a driver for Linux.
On the other hand having DSP technology handy it might be a worthwhile
to design a special Linux DSP OPEN card - which can do voice, music, 
modem, fax, packet-radio and spread spectrum. It's very feasible to have
all this functions in one card. Even if it's manufacturing will be about
$300 - the ability to design a software for it (I mean DSP code) in an
open fascion would lead to a great product - and much cheaper then the
above combined.

Vassili.

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

From: dane@nermal.santarosa.edu (Dane Jasper)
Crossposted-To: comp.os.linux.admin,comp.os.linux.help,comp.protocols.snmp,comp.unix.admin,comp.unix.aix,comp.unix.pc-clone.32bit,comp.unix.questions
Subject: Q: Net Diagram &/or SNMP App for X11
Date: 27 Jul 1994 14:41:19 GMT

I am looking for a free application to aid in diagramming and/or managing
networks.

The platforms would be Linux and AIX, so source code is required.  The
application should run under X11, and be reasonable portable so that I can
build it under both systems. 

In the most simple sense, a 'flowchart' type application that would allow me
to create objects and tie them together would be acceptable.  It would allow
me to design objects and their connection points, and then connect them with
colored lines (red for fiber, blue for thin-net, etc).  It should be able to
handle images of any dimensions, and scrolling around the `network map'
should not be too cumbersome.  Printing to a postscript file should be
possible. 

In a perfect world, this application would also allow me to put in IP
addresses for devices on the network, and it would dynamicly report status
on each device.  `Dumb' devices would be either green or red depending upon
ping status, and smart SNMP manageable devices would have MIB information
available as well.  Automatic discovery would be another nice feature.  =) 

I know about the many commercial applications to accomplish these tasks,
particularly the SNMP management; things like HP's Openview.  One of these
applications would be perfect, except for the first two requirements: That
it be free, and that it include source so that I can build under Linux. 
There is no Openview for Linux (yet).  

Thanks very much for any pointers you may be able to offer.  Private email
is appreciated (because I notice it right away), and I will summarize to
this thread if I come up with a solid solution.  Follow-ups are encouraged
if the information is of general interest.  Please note the large newsgroups
line, and make sure your follow-up is relevant to most or all of the groups. 

Dane Jasper
Network Tech
Santa Rosa JC
<dane@nermal.santarosa.edu>

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

Crossposted-To: comp.os.linux.misc
From: jude@wk106.nas.nasa.gov (Jude George)
Subject: Free Network Management System -- port to linux?
Date: Sat, 30 Jul 1994 00:23:00 GMT

Would someone like to port this to linux?

A free package called HNMS, the NAS Hierarchical Network Management
System, is now available.  The software may be ftp'd from ftp.netcom.com
in the directory pub/heyjude.  (We're looking for other sites
to pick this up, as netcom is pretty bogged down -- offers are welcome).

For sample screen snapshots, try the following in your favorite
WWW viewer:
        ftp://ftp.netcom.com/pub/heyjude/SCREEN1.gif
        ftp://ftp.netcom.com/pub/heyjude/SCREEN2.gif

============================================================================

                                HNMS v2.0g2

        This is the first public release of HNMS, the NAS Hierarchical
        Network Management System.  HNMS may be used to monitor status
        and generate traffic statistics for a large, routed IP network.

        Graphical displays are provided for the X11 Window System and
        make use of the Motif widget set.  HNMS is unique is that it
        can be used to graphically display routing information.  The
        GUI provides compact representations of LANs, in which the status
        of every subnet, host, and IP address can be displayed in a small
        area.  Netmask misconfigurations can be picked out immediately.
        It also correctly displays various WAN architectures, such as
        those which include multiple IP addresses per interface
        (or vice versa), or subnets which spread across multiple links.
        A custom version of HNMS was used to generate the live,
        three-dimensional representation of the cross-country ATM network
        at Supercomputing '93 in Portland, Oregon.

        Data collection is handled via SNMP, ICMP, and direct layer 2
        monitoring.  Distribution of network management information is
        done via HNMP, a new protocol which builds upon the simple,
        stateless client-server model used by SNMP.  HNMP defines
        network objects, binds SNMP variables to them, and facilitates
        higher level management operations on them.  The ASN.1
        specification of HNMP is included in this package.

        Please note that there is also a release of HNMS that we are
        distributing through COSMIC, NASA's software technology
        transfer organization.  COSMIC may (or may not) provide
        support for that distribution, but please do not ask them
        for any kind of support concerning this one.  They will not
        be able to help you.  This version of HNMS has been made
        available to you, by special dispensation, as FREE SOFTWARE
        under the GNU public licence ("copyleft").  It is distributed
        as is, with NO WARRANTY for its fitness for any purpose.
        If it is redistributed, you may not charge for it or any
        derivative work.  Please see the file "COPYING" for details
        about all this.

        Full source code is provided.  We encourage the Internet
        community to experiment with, build upon, use, port, learn from,
        add modules to, write scripts for, and expand this software
        package.  There are some areas in which it can be improved,
        especially in the areas of generality and efficiency.
        We would like to see standardized displays for various types
        of networks and monitored objects, as well as customized displays
        for specific network hardware, and have them be freely available
        to everyone.

        The authors do not have the resources to maintain this software
        outside of NAS.  Although we may continue to add enhancements,
        we are hoping that other(s) will take on the role of incorporating
        bug fixes and extensions to the code, and sending out new releases.

        We do welcome any suggestions or bug reports that you may
        have -- but there's no guarantee that we will be able to
        help you, or even have time to respond.  We are working full
        time (and then some) on other projects.  To reach us via
        email, mail to hnmsdev@nas.nasa.gov.

        This software has not been extensively tested.  Neither the authors,
        nor NASA, nor anybody at all takes any responsibility for any damage
        that HNMS may cause, directly or indirectly, to your network, your
        users, or anything.

        That said, it works pretty well for us.

        Jude George             Leslie Schlecht
        jude@nas.nasa.gov       schlecht@nas.nasa.gov

        Any opinions expressed above are a figment of your imagination,
        and should not be taken to reflect the views of NAS, NASA, CSC,
        or anybody else.

============================================================================
        TO BUILD, follow the instructions in the top-level Makefile.

======> Be forewarned that you will need to obtain the Motif libraries,
======> ISODE=7.0 (not 8.0) and tcpdump if they do not already exist
        on your system.

        FOR COMPLETE INSTRUCTIONS on installing and using HNMS,
        please read the file docs.ps.

============================================================================
Here are some quick-start instructions for the HNMS daemon, the graphical
user interface, and the textual user interface.  The binaries are called
hnmsd, hnms, and hnmstool.

INSTALLATION

1. Untar and build the HNMS distribution.  Sources, docs, and auxiliary
   files are included.  This version of HNMS has only been tested on
   the Silicon Graphics Iris under IRIX 4.0.5.  It should work with little
   modification on SparcStations running SunOS 4.x.

1. Choose a directory to be the hnms "home", and copy the binaries
   and the background/ directory to that directory.

2. Set the $HNMS_HOME environment variable to be that directory.  You
   may want to put this in your .login.

3. Make sure tcpdump is in your path.  It's usually located in
   /usr/local/etc.  The HNMS IO module uses tcpdump to discover new
   IP addresses.

4. Make the hnmsd (and tcpdump, if necessary) setuid root.  The HNMS
   IO module, contained within hnmsd, needs root to access the ICMP
   socket.

5. Set the HNMS_PROMISCUOUS environment variable.  We are just using
   this as a flag to tell the IO module that it's okay use promiscuous
   Ethernet monitoring.

6. Start hnmsd.

RUNTIME -- GUI

The GUI binary is called "hnms".  If you are running it on a machine
different from the server, set the HNMS_SERVER environment variable
to have the server machine's name.

Start the hnms program.  Once the GUI's main window appears, choose
"Open" from the "HNMP" menu at the top center of the window.  A new
window titled "Server" will pop up.  In this window, type "public"
in the Community field, and the hostname of the server machine in the
Selection field.  Then click the Connect button. 

At this point, you can start building diagrams.  Choose "New -->
Custom Status View" from the HNMS menu on the main window.
An "EDIT" window will pop up.  Type an asterisk in the Selection
field, then click Add, then click the check mark at the top.
You will see a display of all the hosts that the server currently
knows about ("Processor" objects), along with their IP addresses
("Ipaddr" objects") and the subnets they are connected to ("Subnet
objects).  If you have just recently started up hnmsd, the objects
will appear magenta while hnmsd is ramping up.  This process
takes three minutes.  Afterwards they will change to green.
When a reachable object becomes unreachable, it changes to yellow,
then red.

Feel free to experiment with the rest of the menu items.  "Site" and
"WAN" diagrams may be built in a manner similar to the "Custom".
For WAN diagrams to be displayed properly, the sysLocation field of
your hosts should be in the format <any-string> <longitude> <latitude>.
For example, "NASA Ames Research Center -122.5 37.1".

RUNTIME -- PLAINTEXT

You may also run hnmstool.  This program takes commands on stdin and
outputs results on stdout, so it can be used effectively from within
shell scripts.  For a demonstration, we will run it interactively.

The hnmstool does not give a prompt.  After starting it, wait a minute
or two for it to get the list of HNMS objects from the server.
for the HNMS daemon to pick up a good assortment of objects from the
network.  Then type list.  You should see a list of objects scroll down
the screen.  Only hosts with SNMP agents responding to "public" will be
shown.  To pick up other hosts, type hint <hostname> <community-name>.
For example:

hint snafu.nas.nasa.gov foobar

The next time you type list, you should see                     

processor:snafu

among the objects.  To get data about an object, type
subscribe "<object>" <variable>@<interval>.  For example:

subscribe "processor:snafu" sysUpTime.0@15

This will give you a report of the value of the sysUpTime.0 variable
every 15 seconds, if the variable changes.  Since sysUpTime.0 constantly
changes, you will get an update on stdout every 15 seconds.

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

From: bart@dunedin.es.co.nz (Bart Kindt)
Subject: Re: SLIP, CSLIP, PPP and modems
Date: Sat, 30 Jul 1994 18:27:32 GMT

In article <316njp$hov@cronkite.ocis.temple.edu> yuriev@astro.ocis.temple.edu (Starcon SysAdmin) writes:
>From: yuriev@astro.ocis.temple.edu (Starcon SysAdmin)
>Subject: SLIP, CSLIP, PPP and modems
>Date: 27 Jul 1994 22:37:45 GMT
>Summary: IMO, do not use efficiently most of modems
>Keywords: SLIP, CSLIP, PPP, Modems

>Hi!

>        Okay. Here is my point. I have tested kernels 1.1.33, 1.0.9 with
>Zoom 14400, DataSystems 1440, Bocas, etc in SLIP mode. It appears that
>when two hosts connected via SLIP, CSLIP, PPP, and ftp from each other
>total output of connection  is still about 1.7K/second. But all these
>modems can use bi-directional mode! All of these modems give approximately
>3.4K/sec under BiModem, HSLink, Hydra, Jaguar/SpeedLink protocol. 
>        If there's someone who is willing to try to implement full
>bi-directional mode, I would like to hear from him/her. All code of my
>Jaguar/SpeedLink could be used here. The only thing which is a bit foggy
>to me is how to detect when to switch from single-direction mode to
>bi-directional mode.

>        Alex Yuriev

Huh?  I thought the lower throughput was caused by the IP header / tail 
overhead. Modems are normally always "bi-directional", e.g full duplex. 
SLIP does not change any settings on the modem at all.

Bart.   
       



=================================================
Bart Kindt (ZL4FOX/PA2FOX), Dunedin, New Zealand.
=================================================

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

From: yuriev@astro.ocis.temple.edu (Starcon SysAdmin)
Crossposted-To: comp.lang.c
Subject: Re: Linux Performance Enhance ?
Date: 26 Jul 1994 17:18:42 GMT

Lawrence Kirby (fred@genesis.demon.co.uk) wrote:
: In article <30n4po$79i@gazette.esd.sgi.com>
:            erik@westworld.esd.sgi.com "Erik Fortune" writes:
: >Um, if you find any features of C that *can't* be implemented in machine
: >code, I'm sure we'd all love to hear about it...
: I'm considering Duffs device as a source code construct. Most C fearures
: have no direct machine code counterpart, for loops, structs etc. have to

1) All C/C++/Whatever-other-language instructions end up in machine code
2) If that C/C++/Whatever-other-language code works, then machine code
generated by compiler works too
3) Compiler generates this *working* code because someone described how to
translate this C/C++/Whatever-other-language code into machine code
4) It follows from (3) that that someone did write that
C/C++/Whatever-other-languiage code in machine code form. 
5) It follows from (4) that that thereis nothin that can be done in
C/C++/Whatever other language and can't be done in machine code.

Best wishes,
Alex Yuriev
ACCU

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


** FOR YOUR REFERENCE **

The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:

    Internet: Linux-Development-Request@NEWS-DIGESTS.MIT.EDU

You can send mail to the entire list (and comp.os.linux.development) via:

    Internet: Linux-Development@NEWS-DIGESTS.MIT.EDU

Linux may be obtained via one of these FTP sites:
    nic.funet.fi				pub/OS/Linux
    tsx-11.mit.edu				pub/linux
    sunsite.unc.edu				pub/Linux

End of Linux-Development Digest
******************************
