From:     Digestifier <Linux-Admin-Request@senator-bedfellow.mit.edu>
To:       Linux-Admin@senator-bedfellow.mit.edu
Reply-To: Linux-Admin@senator-bedfellow.mit.edu
Date:     Sun, 2 Oct 94 18:14:20 EDT
Subject:  Linux-Admin Digest #135

Linux-Admin Digest #135, Volume #2                Sun, 2 Oct 94 18:14:20 EDT

Contents:
  Re: Fast Scsi-2 COMPAQ (Mark Stockton)
  Re: linux+slip+bootp. How? (Jon Peatfield)
  Re: <Q> Can Linux Mount a Mac Floppy (Steve Kneizys)
  Re: PPP vs SLIP? (Mike Battersby)
  Re: LILO & BusLogic 445s (Marc de Courville)
  FTPs PCTCP and Linux lock. (Brian C. Huffman)
  Re: keymap question (Benjamin John Walter)
  handshaking with dip? (Benjamin John Walter)
  Re: Telnet & ftp freeze! (System Administrator)
  Re: Term problems (Patrick Reijnen)
  Lynx under Linux (Mubashir Cheema)
  Slip autoanswer (Gioel M Molinari)
  Re: ftp freeze problems (Matti Aarnio)

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

From: marks@schooner.sys.hou.compaq.com (Mark Stockton)
Subject: Re: Fast Scsi-2 COMPAQ
Reply-To: marks@schooner.sys.hou.compaq.com
Date: Fri, 30 Sep 1994 15:04:22 GMT

TlingitMan (tlingitman@aol.com) wrote:
: In article <785@bjohns.win.net>, bret@bjohns.win.net (Bret A. Johnson)
: writes:

: >I am tryingto install Linux Slackware Pro. ver. 2.0 on a Compaq
: server. 

: Bad News Bret.... Compaq doesn't like to tell people developing free
: software what bits to tweak on their hardware.  I've been down this road
: on several items.

Not exactly true.  For components using technology that we don't
license from someone else, we will give out information to developers.
The QVision controller is a good example.  Technical reference guides
are available and I've sent a set out to a Linux developer.

: You know that nice 32-bit ethernet Netflex NIC they sell?  Linux doesn't
: like it or even try to recognize it.

The microcode that the driver must download to the board is licensed
from Texas Instruments.  You can try purchasing a license for the
microcode, but I don't think that you'll want to pay their price.  None
of the existing drivers will work with this board.

: Motherboard SCSI controller?  Same damn story.

Which one?  The NEC C710, C94 or the AMD?

: Memory past 16MB?  Good luck - please let me know if you figure that one
: out. 

I've posted the solution to this before.  Since we don't officially 
support Linux, you probably aren't going to find anyone on the support
lines who can help with this.  There are several of us who happen to use 
Linux and we're happy to share what we know about Linux and Compaq 
hardware.  Please keep in mind, though, that Compaq does not test
or support Linux.

: So I run an Adaptec 2740, SMC Ultra-16 ethernet, and 16MB RAM on my
: Pentium server.  

Until the C710 support is working (see the kernel source), and since
you probably don't have a license for the TI microcode, the only thing
I can help you with is the memory problem.  

First, using your EISA configuration diskettes, make sure that your
RAM is in linear mode vs. "Compaq Compatible" mode.  Then edit
/usr/src/linux/init/main.c and hardwire the memory_end variable
to the amount of your RAM.  For example, if you have 32MB, then
make this change:

was:  memory_end = (1<<20) + (EXT_MEM_K<<10);
now:  memory_end = 32*1024*1024;

Finally, compile a new kernel and reboot.

: Tlingitman@aol.com

MarkS
--
Mark Stockton
marks@schooner.sys.hou.compaq.com

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

From: J.S.Peatfield@amtp.cam.ac.uk (Jon Peatfield)
Crossposted-To: comp.os.linux.development
Subject: Re: linux+slip+bootp. How?
Date: 02 Oct 1994 17:10:18 GMT

> tried it and couldn't do it, think that was because the slip
> connection doesn't have an ethernet address (ie in the form
> xx:xx:xx:xx:xx:xx). the reason i was trying it was to get a computer
> to telnet in, the computer was local so I tried it with plip which
> does have an ethernet address type setup but still couldn't get it
> working. I did end up getting it working using rarp tho and maybe that
> would work over slip???

The lack of a MAC address shouldn't stop it working assuming that the
bootp server "knows" who you are by some other means (e.g. it knows
what physical port you are on or who you logged in as.)

However in all versions of bootpc I've released so far it didn't work,
'cos I always put the ARPHRD_ETHER type in the htype bootp field.

On NET3 (1.1.14+ I think) based kernels you can extract the correct
hardware type from any interface using SIOCGIFHWADDR so it should just
work correctly with that fix to the bootpc code.

However I've not got a SLIP (or PPP etc) line to test this on so I'm
on rather shakey ground where this is concerned.

Could someone with a SLIP line and a new(ish) kernel run the bootpc
(0.31) code thus:

  bootpc --dev sl0 --verbose --debug

andmail me all the output?

-- Jon
--
Jon Peatfield, Computer Officer, the DAMTP, University of Cambridge
Telephone: (+44 223) 3-37852     Mail: J.S.Peatfield@amtp.cam.ac.uk

      Friends don't let friends use PP.     PP: Just say NO.


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

Subject: Re: <Q> Can Linux Mount a Mac Floppy
From: STEVO@acad.ursinus.edu (Steve Kneizys)
Date: 2 Oct 94 12:56:53 EST

Tim Bass (Network Systems Engineer) (bass@cais2.cais.com) wrote:


: Got some ASCII text files on my Powerbook... Would like to write them
: to the PB floppy and then mount the floppy on my linux box and
: read the ascii text (and do some other stuff).  I haven't seen this
: in any FAQ or the The Linux Bible.   Any clues for the clueless ;-)

Well, my solution is, instead of making Linux understand the Mac
format why not make the Mac understand a format Linux does!  I have
to exchange data between Mac, PC, VAX, Unix, and use FTP for most of
it.  But when I do flopppy exchange, I use a DOS formatted floppy...
the Dayna product DOSMOUNTER for Mac lets me mount Dos floppies on the
Mac, write ascii files, then mount on Unix, etc.

I heard a rumor there was a shareware program that mounted DOS disks
on Macs, but never looked into it.  But if you want to go this roundabout
way then I'd think some Mac folks could help more than I have.

Good luck!

Steve... 

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

From: mike@starbug.apana.org.au (Mike Battersby)
Subject: Re: PPP vs SLIP?
Date: 30 Sep 1994 14:28:02 +1000

macgyver@MCS.COM (MacGyver) writes:
>Currently, I'm happily churning away, using [C]SLIP on my machine.  Recently,
>someone told me that PPP is more efficient and much better than using
>SLIP.  Is this true?  Does anyone have a rough idea or maybe even some
>numbers showing which is better of the two to use?  If PPP is better,
>where can I get the necessary software to run it?  Is it as trivial to
>configure as SLIP was?  (With SLIP it was merely changing a sample
>script slightly).

This is some (hopefully correct) info on SLIP and PPP from a technical
point of view.  I don't know which is better in terms of Linux implementation,
or if there is any difference.

TCP: Most of the traffic you'll be copying with is TCP frames.  These are
     20 bytes of header, optional options (grin) and the data.

IP: Your basic IP frame is 20 bytes of header, some (usually 0) bytes of
    IP options, and some bytes of data.

SLIP: The SLIP encapsulation adds an extra 2 bytes per frame, as well as
      escaping and 0xC0 and 0xDB characters in the frame.  Hopefully these
      are fairly rare.  It has a simple encapsulation protocol, and nothing
      else.  Things to note here are:
        - there is no negotiation of options between ends, no possibility of
          exchanging IP addresses for dynamic lines, and no SLIP checksum
          on the frame (not a big problem with error correcting modems).

CSLIP: Adds compression to the TCP and IP headers, reducing the (transmitted)
       header size from 40 bytes to 4 or 5 bytes.  This is a big win for
       interactive traffic, where if you push a key you get 1 byte of data
       sent with 40 bytes of header[1].

PPP: As well as simply providing an encapsulation method for IP frames, PPP
     also provides parameter negotiation between the two ends.  PPP 
     nominally adds 7 bytes of overhead for each frame, but the two ends can
     (and usually do) negotiate this down to 5 bytes.  PPP contains a CRC
     for each frame, and can do header compression just like CSLIP (this is
     another negotiable option).  PPP can also operate with protocols other
     than IP, though most (all?) people won't use it for such.  One thing
     to note is that by default PPP will escape all characters less than
     0x20.  This is a major lose, so if you're going to run PPP you need to
     set the asyncmap to escape as few as possible.

So, the advantages of PPP:

        - Dynamic negotiation of parameters, such as IP addresses.
        - Header compression.
        - CRC on each frame (not a big win for newer modems)

Disadvantages of PPP:

        - 3 bytes of extra overhead.

Why you would use:

SLIP:   You wouldn't, not if you could help it.  Demand at least CSLIP!

CSLIP:  You have an error correcting modem, a static IP address and you're
        a speed freak who'll miss the extra 3 bytes.  Because your
        provider doesn't offer PPP.

PPP:    Because you can. Because you need to send something other than
        IP over the link.  Because it's more convenient.

cheers

 - Mike

p.s.  Much of this info from W. Richard Stevens' book, TCP/IP Illustrated.

[1] Okay, so this isn't the whole deal, but it illustrates the point.

-- 
  Mike Battersby   <mike@starbug.apana.org.au>   <mib@molly.cs.monash.edu.au>
          CompSci honours student, Linux user and part time beggar.
  "You know you're really somebody in the software world when Richard
   Stallman complains about you having a gratuitous patent." - Anon.

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

From: courvill@garfield.enst.fr (Marc de Courville)
Crossposted-To: ucb.os.linux
Subject: Re: LILO & BusLogic 445s
Date: 2 Oct 1994 17:23:25 GMT

Hi,

I had the same problem this week-end! And now it is solved :-)
d) It is due to something else.... 
Well it is not quite oubvious to explain.
As you turned on the Buslogic 445s jumper for larger than 1Gb disks I assume 
that you did a translation on the head/cyl in order to have cyl<1024.
What I did is quite simple. In /etc/disktab you can specify the
geometry of your SCSI disk. Then specify here the translated geometry.
Now install lilo on your hard disk. And before booting modify /etc/disktab
and enter the geometry detected by the lilo fdisk (with cyl>1024).
It should work now... (I hope)

Marc 


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

From: sheep@news.udel.edu (Brian C. Huffman)
Subject: FTPs PCTCP and Linux lock.
Date: 2 Oct 1994 17:38:40 GMT
Reply-To: sheep@strauss.udel.edu

        I have Slackware's 2.0 distribution of Linux, and I am having a problem
logging on using FTP Software's PCTCP.  It tends to lock the telnet session
(not the linux) and I have to quit and start again.  If I log into a unix
machine first and then telnet to the linux, it works fine.  Anyone have
any suggestions?

                                        Brian

--
        +------------------------------+
        +---- Brian C. Huffman --------+------------------------------+
        |  University of Delaware      |                              |   
        |   206 New Castle Hall        |   sheep@bitbytes.clark.net   |
        |    Newark, DE  19717         |    sheep@chopin.udel.edu     |
        |     (302)/837-8567           |                              |
        +------------------------------+------------------------------+


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

From: ben@tsunami.demon.co.uk (Benjamin John Walter)
Subject: Re: keymap question
Date: Thu, 29 Sep 1994 07:32:58 +0000

I have this line in my /etc/rc.d/rc.keymap:

/usr/bin/loadkeys /usr/lib/kbd/keytables/uk.map

peace, Ben



-- 
   __   _                
  / /  (_)__  __ ____  __
 / /__/ / _ \/ // /\ \/ /  . . .  t h e   c h o i c e   o f   a
/____/_/_//_/\_,_/ /_/\_\              G N U   g e n e r a t i o n . . .

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

From: ben@tsunami.demon.co.uk (Benjamin John Walter)
Subject: handshaking with dip?
Date: Thu, 29 Sep 1994 15:24:33 +0000

Does dip use RTS/CTS handshaking by itself?  Or must I first run a
command like 'stty crtscts < /dev/cua0'?  Does that work?

peace,  Ben
-- 
   __   _                
  / /  (_)__  __ ____  __
 / /__/ / _ \/ // /\ \/ /  . . .  t h e   c h o i c e   o f   a
/____/_/_//_/\_,_/ /_/\_\              G N U   g e n e r a t i o n . . .

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

From: root@jaguar.tigerden.com (System Administrator)
Crossposted-To: comp.os.linux.help,comp.os.linux.misc,comp.os.linux.development
Subject: Re: Telnet & ftp freeze!
Date: 2 Oct 1994 18:21:24 GMT

Trevor Lampre (trevor@xanax.apana.org.au) wrote:

[Text describing and lamenting problem my myself and others deleted.]

: Many have. I have posted twice myself about it and seen at least 5 other 
: posts not including this thread. I have never seen a response and my emails
: to other posters has never been answered. It's pissing me off that nobody
: seems to know the answer or have a fix. I've been patching my kernel up
: to 1.1.51 (I think it got worse at .51) as well as rebuilding my daemons.

: As the admin of a public access system it is of great concern to me, I've
: had sendmail die for about 2 days before I noticed as well as the other
: problems described. I spend more time now checking/killing/rebooting
: my network stuff than I do giving more value to my users. I might just
: switch to *BSD, at least the network code works.

Thank WHATEVER that others are seeing this problem!  And thanks to 
        Trevor Lampre (trevor@xanax.apna.ort.au)
        Michael Haardt ((michael)u31b3hs@pool.informatik.rwth-aachen.de)
        Thomas E Zerucha (zerucha@shell.portal.com)
and     Steve Kneizys (STEVO@acad.ursinus.edu)
for confirming what we've been seeing!  I suggest we keep this thread 
open and fill it with additional information until the problem gets the 
attention it needs.  I'm not a programmer, much less a kernel hacker, so 
I can only voice frustration with the situation.

Some additional information gleened from observations:

First, the original problem as I originally mentioned it:  

We are running slip to our internet provider, and intermittantly
experience telnet lockups during logins.  The system either 1) refuses
connections 2) accepts the connection, but just sits 3) provides a login
prompt, takes input, and never gives the password prompt (ususally
creating a login zombie in the process). 

Additional information/trends noticed:

If the lockup occurs, allowing the telnet session with the locked
connection to sit while starting another is *always* successful.  It
*appears* that a particular ttyp# gets buggered somehow, and forcing the
system to seek another one will get you in.  I.E.  We've had *tons* of
complaints about ttyp1 and ttyp4 lately (although I've seen the problem
also on ttyp3 in the past). 

In the event 'refused connections' are the symptom to those telnetting in
over the SLIP connection, logging in by adding an x-term *on the console*
that grabs the offending ttyp port will suddenly allow SLIP telnet
accesses. 

I thought that once a user was successfully logged in, everything was
fine.  However, I have had complaints of 'gradual slowing' or 'gradual
slowing then locking' from a couple of users.  I intiially dismissed this
as 'net problems', but after hearing Michael Haardt's comment, I'm
beginning to think that's what's happening to us as well.  I also suspect
that other 'general' user complaints about our 'slowness' at times would
turn out to be the same thing as well. 

I have been experimenting with MTU sizes with ifconfig, but have no feel
for whether or not this has any effect.  I *have* noticed that MTU gets
reset to 1500 by *something* some random time after I've changed it (note
this is without system reboots). 

Here's a sample of what we have:

yiffy:~# ifconfig
lo        Link encap Local Loopback  
          inet addr 127.0.0.1  Bcast 127.255.255.255  Mask 255.255.255.0
          UP BROADCAST LOOPBACK RUNNING  MTU 2000  Metric 0
          RX packets 0 errors 0 dropped 0 overrun 0
          TX packets 39754 errors 7 dropped 0 overrun 0

sl0       Link encap AMPR AX.25  HWaddr 
          inet addr 198.30.162.1  P-t-P 199.18.108.11  Mask 255.255.255.0
          UP POINTOPOINT RUNNING  MTU 1500  Metric 0
          RX packets 1583360 errors 0 dropped 0 overrun 0
          TX packets 1995660 errors 0 dropped 16514 overrun 0

eth0      Link encap 10Mbps Ethernet  HWaddr 00:20:AF:16:4C:3E
          inet addr 198.30.162.1  Bcast 198.30.162.255  Mask 255.255.255.0
          UP BROADCAST RUNNING  MTU 1500  Metric 0
          RX packets 293959 errors 0 dropped 0 overrun 0
          TX packets 285447 errors 0 dropped 0 overrun 0

Note that a few moments prior to running this, I had set sl0 MTU to 2000, 
and confirmed that the value was accepted.  Now it is 1500 again without 
any action on my part.

One last observartion.  When we first started with kernel 0.99.15, telnet
sessions were locking up when large amounts of data was to be sent *out*. 
That is, if someone did a large directory listing or other function with
lots of output, their session would hang.  The send buffer information in
'netstat' showed several thousand characters waiting to be sent, and the
session would be effectively frozen.  This problem was acknoledged by
others at the time, but as in this case, no answers were provided.  The
problem went away when moving to the 1.0 kernel.  So *something* was done
bye *someone* for that one. 

I'm new to all this, and don't know all the avenues to pursue.  I'd 
appreciate any help in getting this problem hilighted and information 
flowing to the *someone* who understands how the net interfaces really 
work and who can really and *finally* *fix* it!  How do we proceed?


George Nemeyer       (root@tigerden.com)
System Administrator
Tigerden.com

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

From: patrickr@cs.kun.nl (Patrick Reijnen)
Subject: Re: Term problems
Date: Sun, 2 Oct 1994 16:44:19 GMT

In <Cwt3tL.K6E@murdoch.acc.Virginia.EDU> bryan@elvis.phys.virginia.edu (Bryan Wright) writes:

>Hi Folks,

>       I'm having a tough time setting up 'term'.  When I run linecheck,
>I get the following output (stderr) on the local and remote machines:

[ lots of stuff deleted ]

>Term 1.08

[ stuff deleted ]

>       Does anyone have any suggestions?

Yup, what about getting a new term version to start with. 1.08 is really old !!!!

We are at version 2.1.4 at this moment. check sunsite.unc.edu:/pub/Linux/apps/comm/term


>                               Thanks in advance,
>                               Bryan

>--
>===============================================================================
>Bryan Wright                |"If you take cranberries and stew them like 
>Physics Department          | applesauce, they taste much more like prunes 
>University of Virginia      | than rhubarb does."  --  Groucho 
>Charlottesville, VA  22901  |                  
>(804) 924-6814              |         bryan@sphinx.phys.virginia.edu
>===============================================================================

Patrick Reijnen


--
************************* Patrick Reijnen *************************
* Department of Computer Science, Catholic University of Nijmegen * 
* Email:  patrickr@{sci,cs}.kun.nl                                *
* WWW:    http://{atlas,zeus}.cs.kun.nl:4080/homepage.html        *

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

From: cheema@earth.sparco.com (Mubashir Cheema)
Crossposted-To: comp.infosystems.www.misc,comp.infosystems.www.providers,comp.os.linux.help,comp.os.linux.misc
Subject: Lynx under Linux
Date: 2 Oct 1994 18:28:09 GMT


I have noticed that when I run Lynx under Linux all available
options get highlighted instead of the ones I move my 
cursor to. 

We have setup a guest account which by default runs lynx when
a user telnets into our machine and logs in as guest. A new user
who has never used lynx before gets confused. We have received
mail from people who said they couldn't use it. 

The guest account automatically recognizes the terminal type of
user logging in. If for some reason it can't, it prompts the user
for that information. I do not suspect that the guest account is
not detecting the terminal type correctly, since I see the same 
behaviour when I run lynx on my machine under xterm, vt100 etc. on
this machine. 

If you have telnet access, open a telnet session to sparco.com and 
login as guest to see what I mean.       

Any help will be greatly appreciated.      

Mubashir Cheema            - new, expanded .sig -     cheema@sparco.com 

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

From: Gioel M Molinari <gm3h+@andrew.cmu.edu>
Subject: Slip autoanswer
Date: Fri, 30 Sep 1994 21:38:25 -0400

Hey y'all I am trying to set up a Slip server for people to dial into...
I think I am just about set to go except one thing :

    How do I get my modem to autoanswer the phone when it detects a ring ??

Thanx


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

From: mea@utu.fi (Matti Aarnio)
Subject: Re: ftp freeze problems
Date: Sat, 1 Oct 1994 22:51:32 GMT

There is some problem in NCSA-Telnet's FTP with some versions of
Linux TCP code, thus I would like to know a bit more about these
troubled DOS machines:

dtran@emelnitz.ucla.edu (Daniel Tran) writes:
>In article <36f5lb$3ab@hobbes.cc.uga.edu> taylor@pollux.cs.uga.edu (john taylor) writes:
>>Craig Tavener (craig@chem.chem.wits.ac.za) wrote:
>>I've got linux (Slackware2.0) recently installed on a 468-66.  It has an 
>>ethernet card and is networked to Novell and and UNIX.  When ftp/telnetting 
>>from linux to elsewhere things generally work well.  However, when trying to 
>>ftp into the linux machine (most notably from the novell network) the 
>>session frequently freezes.  Pressing cntl-C returns the ftp prompt, but the 
>>last action ends up being truncated.  A good example of this is a file 
>>transfer.  All packet but the last one get through.  Then it freezes and the 
>>last packet it lost.
>>
>>Does anyone have any idea what is going on here?
>>
>>I have the same problem. When I ftp from my novell server to my linux box. 
>>The entire file will transfer except the last 1 or 2K. How can this be fixed?
>>
>>John

>I do not have that problem at all.  My workstation sits on the Novell network, 
>I am constantly telneting and ftping to my linux box w/o any problems.
>I'm running kernel 1.1.35

>Daniel Tran - dtran@emelnitz.ucla.edu

        What TCP/IP stacks does each of you have/use on their
        DOS machines ?

        What is your Linux kernel configuration regarding following:
                PC/TCP compability
                Disable NAGLE


        Some successes have been obtained with:
                PC/TCP compability      y
                Disable NAGLE           n
        (I haven't tested this FTP problem with PC/TCP compab Y, though..)


        /Matti Aarnio   <mea@utu.fi>

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


** 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-Admin-Request@NEWS-DIGESTS.MIT.EDU

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

    Internet: Linux-Admin@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-Admin Digest
******************************