Subject: Linux-Development Digest #782
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:     Tue, 31 May 94 23:13:07 EDT

Linux-Development Digest #782, Volume #1         Tue, 31 May 94 23:13:07 EDT

Contents:
  Networking and Linux (Scott S Critchley)
  HP 27252A Ethernet Card doesn't work! (Thomas Heiling)
  Re: SLIP problems with kernel 1.1.15 (Lou Sortman)
  Re: Logitech Scanner (Re: Twain Scanner) (Karl Eichwalder)
  Re: 1.1.15 breaks SCSI (Ken Hoover)
  Re: Serial drivers in 1.1.16 (autoconfig for IRQ's) (njs@bcrvm1.vnet.ibm.com)
  Re: ARP broken with kernel 1.1.13 (Alan Cox)
  Re: Writing a new file system, help (Jonathan Magid)
  Re: Dynamic linking with shared libraries. (Eric Youngdale)
  Re: 1.1.15 breaks SCSI (Eric Youngdale)
  Re: Serial drivers in 1.1.16 (autoconfig for IRQ's) (Mikael Nordqvist)
  Re: kernel 1.1.14 (&13) + xiafs segfaults rpc.nfsd (Rene COUGNENC)

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

From: scottc@alpha2.csd.uwm.edu (Scott S Critchley)
Subject: Networking and Linux
Date: 31 May 1994 15:43:50 GMT

Hello there... its Howard Feiges here...

A while ago I set up a linux partition on my 486 and had that up and
running for a while...   It was quite fun and interesting but I really
didn't need it all that much so I got rid of it so I could have more
drive space for my system....  (so I have been able to set up a linux
system in the past, even though I am a bit sketchy on the details, it 
really wasn't anywhere near as hard as I thought it was going to be...)

anyway, I know that you can network a bunch of computers together
using linux... but I really have no idea how to go about doing that...
I guess you could do it with a bunch of x86 computers and ethernet/lan
boards and concentrators and the like....

but the other day, the thought just occurred to me....  Couldn't it be
possible to hook up nodes to a linux-based server just through plain
everyday old serial connections....  I mean, I know that serial would
be slower, but if all you want to do is text processing then 19,200
baud would be plenty fast enough..  And more than that, but couldn't
you then use ANY computer that had a serial connection (like, for
example an Apple //GS or //e with a serial board, or a Mac plus or a 
crappy old 286 or XT or whatever) and terminal software....  The reason
why I am asking this is because I have several old apple //s sitting around
collecting dust, and the thought of using those apple //s as network nodes
is quite appealing (I mean, to access my 486 though an apple //e in a
completely different room would be quite a feat)...  Another possibility
could be school systems with apple //s.  The apple // is a worthless
computer now-a-days but hook them up to a 486 and you could teach
a programming class, or perhaps access the internet or whatever, all 
though an old apple //.  But It could only happen if you could use 
serial (try getting an ethernet board for an apple //e)

anyway, the thought just occurred to me... I have no idea how possible
it really is (It seems to me, that with linux, anything is possible).. 
but I am more concerned about the hardware end (can you use a concentrator
of sorts to bring together serial connections from several different
computers, and combine them so they can go into a single computer, either
through a serial port or even an ethernet board or whatever...  I am 
JUST learning about setting up networks (we have a novell system at
one of our facilities and we have ethernet boards in our 486's here at
our office, even though we don't have a server or novell software to run
on them).   I just thought that through serial connections, you could
cheaply put together a nice network of computers, using old, outdated,
existing hardware....  There are thousands of worthless computers out
there, that could be worth having if you could network them....

for now, I am not interested in the details or how practical it is
to have such a network.  All I want to know for now is if its possible,
and worry about the implementation later..   does the hardware
exist where you could network computers using serial interfaces...   Can
you 'concentrate' serial so you can channel the data all into a single
computer, or perhaps can you set up a 'token ring' system where you
could have two serial connections, one incoming and one outgoing, from
each computer in the network hooked up in a circle....   Just what
is possible???   I have at home right now a 486 dx2/66 computer and 
an apple //e, an apple //GS, and a macintosh plus.  I am a programmer
so I could actually write the software needed if it doesn't exist
already....  but as you can probably tell, I am no hardware expert....
for now, I just want to make lemondade from my lemons, thats all, and
perhaps learn a thing or two while doing it...

anyway, please let me know...  You can email me at the following
address:

scottc@alpha2.csd.uwm.edu

thanx much...

Howard Feiges
TPK

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

From: tom@wpzd07.pzlc.uni-wuerzburg.de (Thomas Heiling)
Subject: HP 27252A Ethernet Card doesn't work!
Date: 31 May 1994 16:01:50 GMT

Hello !
Has anyone a driver for this Card ? Unfortunately I have this card
ordered, because it was rated as a very good Card in the
Ethernet-HOWTO.
But this Information is WRONG!
Linux can use *only* the older Cards with the HPPCLAN.COM Packet driver,
not the newer HPPCLANP.COM Packet drivers.
After mailing to Mr. Becker some time ago, he says that he would
perhaps look after this driver .
I have tried to write it myself, but my knowledge was far to low to
do this. I haven't  find the specs and the HP-manual is useless for
this theme.

So if anyone can help me Much thanks
Bye Thomas
--
===================================================
Thomas Heiling Pharmacist & Doctorate at 
Pharmazeutisches Institut Uni Wuerzburg - Germany 
Email phar006@rzbox.uni-wuerzburg.de (HP-UX)
      tom@wpzd07.pzlc.uni-wuerzburg.de (Linux)
or phar006@vax.rz.uni-wuerzburg.de ( VAX )
===================================================

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

From: lou@tfnet.ils.unc.edu (Lou Sortman)
Crossposted-To: comp.os.linux.help
Subject: Re: SLIP problems with kernel 1.1.15
Date: 31 May 1994 16:20:03 GMT

Any idea why I cannot even ping my loopback with kernel 1.1.16?
It just hangs there.  It doesn't even print the heading line.
It works fine with 1.1.12.

Maybe I have something set up marginally or need new user programs?

--
lou@tfnet.ils.unc.edu (Lou Sortman)       for(i=0; i<3; i++) puts(
                                                  "Janet!    Dr. Scott!   \n"
                                                  "Janet!    Brad!        \n"
Whoever dies with the most LEGO wins.             "Rocky!    <Uh!>        \n");

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

From: karl@pertron.central.de (Karl Eichwalder)
Subject: Re: Logitech Scanner (Re: Twain Scanner)
Reply-To: keichwa@gwdg.de (Karl Eichwalder)
Date: Mon, 30 May 1994 23:35:15 GMT

Karl Eichwalder (karl@pertron.central.de) wrote:

> Also for the ScanMan256 ?  logi32.c complains here all the time:

>         logi32.c: cannot register major number

To be a little bit more precise:  The first time, when I try to load logi32.o
the driver says:

module `logi32' (3 pages @ 0x01818000) created
initializing module `logi32', 8928 (0x22e0) bytes
  init entry @ 0x0181a0d4, cleanup entry @ 0x0181a1c4
LOGITECH handheld scanner driver  version 0.1a
Scanner-interface is incompatible.
logi32.c: scanner interface not found

And when I try it again:

module `logi32' (3 pages @ 0x01818000) created
initializing module `logi32', 8928 (0x22e0) bytes
  init entry @ 0x0181a0d4, cleanup entry @ 0x0181a1c4
LOGITECH handheld scanner driver  version 0.1a
logi32.c: cannot register major number

I am running 1.1.16; IRQ 5, DMA 3, Address 280 : these settings work under
DOS/Windows.

Another question: Is there any OCR oder grafik scanner software?

--
                        | keichwa@gwdg.de
                        | karl@pertron.central.de
Karl Eichwalder         | 2:2437/209

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

From: ken@ch201c.ed.psu.edu (Ken Hoover)
Subject: Re: 1.1.15 breaks SCSI
Date: 31 May 1994 17:04:58 GMT

Rene COUGNENC (rene@renux.frmug.fr.net) wrote:
: [...]
: Configuring Adaptec at IO:330, IRQ 11, DMA priority 5
: scsi0 : Adaptec 1542
: scsi : 1 hosts.
:   Vendor: FUJITSU   Model: M2622F-512        Rev: 0404
:   Type:   Direct-Access                      ANSI SCSI revision: 01 CCS
: Detected scsi disk sda at scsi0, id 0, lun 0
: [...]

  My system gives an almost identical error message, but with a totally
Different background.
============================= begin ==================================
Calibrating delay loop.. ok - 33.22 BogoMips
scsi0 : Seagate ST-01/ST-02
scsi : 1 hosts.
  Vendor:           Model:                   Rev:     
  Type:   Direct-Access                      ANSI SCSI revision: 00
Detected scsi disk sda at scsi0, id 1, lun 0
  Vendor: TEXEL     Model: CD-ROM DM-XX28    Rev: 3.06
  Type:   CD-ROM                             ANSI SCSI revision: 02
Detected scsi CD-ROM sr0 at scsi0, id 5, lun 0
scsi : detected 1 CD-ROM drive total.
Scd sectorsize = 2048 bytes
================================== end =================================

  Booting proceeds normally after that, and the system runs OK except that
I cannot access a CDROM.

  I have no SCSI hard disks in my system, only the single TEXEL CDrom
at SCSI ID 5 which seems to be properly detected.  However, I cannot mount
from /dev/sr0 (fails with a "isofs_read_super: bread failed, dev 0xb00 
iso_blknum 16" followed by mount's generic error msg).

  System info:
- Intel 486dx2/66, 16m ram, AMI BIOS
- Seagate 450m HD (IDE - works great)
- 3Com 3c509 Ethernet adapter  (IRQ 10)
- Future Domain FD 850M SCSI controller  (IRQ 5)
- PLEXTOR 3028 internal CD-ROM drive (identified in ROM as TEXEL)
        ( Texel recently changed its name to Plextor and hasn't updated their
          ROMs yet, it appears )

  This last worked under 1.1.13, but works fine under DOS.

--
Kenneth J. Hoover                | "... at .7c, a bit is about 40 meters long."
ITSS Supervisor of Systems & Ops | -- antonyc@linuxftp.caltech.edu (on usenet)
Penn State College of Education  | ##### Run Linux!  It's good for you! #####
=====  ken@ch201c.ed.psu.edu ** ken@psuedvax.ed.psu.edu ** kjh6@psu.edu =====

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

From: njs@bcrvm1.vnet.ibm.com
Subject: Re: Serial drivers in 1.1.16 (autoconfig for IRQ's)
Date: Tue, 31 May 1994 15:26:30 GMT
Reply-To: njs@watson.ibm.com

In <Cqn086.2yF@pe1chl.ampr.org>, rob@pe1chl.ampr.org (Rob Janssen) writes:
>In <CqMp9p.EzF@scifi.uucp> njs@scifi.uucp (Nick Simicich) writes:

>>I have a serial card with four serial ports on it.  These are straight
>>16550's, so they each need an interrupt, and are plugged on their own
>>interrupt.  

>Now, you should finally learn about setserial.  It has been around for
>a long time, and was there to replace hacking in serial.c
>With the new tty drivers, that possibility was removed.
>
>So: get the "setserial" program (current version is 2.10) and use it
>instead.

Thanks for the advice.   I got a copy of 'setserial', and stuck in the 
rc.serial file, appropriately modified, early on in the boot process, 
just after the mounts (after learning that it wouldn't work, later in 
the boot process because of selection].  If this is the way that it 
is to be done, fine, but I still feel, somehow, that the old way was 
better.  I do suspect that the 'setserial', 'tunelp', and so forth 
approaches are more general, but I was never shy about recompiling 
the kernel with the options I wanted.

Now if only I could get SL/IP to work reliably with a recent kernel :-).
--
Nick Simicich - njs@bcrvm1.vnet.ibm.com - njs@watson.ibm.com
                - njs%scifi.uucp@uunet.uu.net

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

From: iiitac@uk.ac.swan.pyr (Alan Cox)
Subject: Re: ARP broken with kernel 1.1.13
Date: Tue, 31 May 1994 16:53:53 GMT

In article <HCJ.94May31004231@gull.uncc.edu> hcj@gull.uncc.edu (Harry C. Johnson) writes:
>I just loaded version 1.1.13 of the kernel on the system I am using as
>a slip server, the machine that is called in to.  It seems however
>that arp is broken with this version of the kernel.  Execution of arp
>gives the error:
>
>/bin/arp: /dev/arp device not present 
>
You need an up to date set of net tools (net032d) that uses /proc/net/arp
and you should find that with the current 1.1.17 kernel both it and arp -s
work fine. All the /dev support for networking has been removed to avoid
wasting space and duplication of /proc

Alan


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

From: jem@bittyblue.oit.unc.edu (Jonathan Magid)
Subject: Re: Writing a new file system, help
Date: 31 May 1994 22:40:21 GMT

In article <CqosBF.FyE@sys.uea.ac.uk>,
Ben Frank <u9219765@sys.uea.ac.uk> wrote:
>Hi,
>   I'm trying to write a new file system (Third year porject) I seem to
>remember that someone was writing a "HowTo write a file system" document,
>I can't find it anywhere, nor any reference to it. Does it exist, if not is
>there anything that would be useful to look at ? I have read/am reading all
>the stuff from the LDP

An account of the proc filesystem  and the VFS layer in the kernel can
be found in Michael K. Johnson's book, "The Linux Kernel Hacker Guide".

It is available on sunsite.unc.edu /pub/Linux/docs/LDP/kernel-hackers-guide.

jem.


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

From: ericy@cais.com (Eric Youngdale)
Subject: Re: Dynamic linking with shared libraries.
Date: 31 May 1994 22:31:55 GMT

In article <1994May30.031952.17721@midway.uchicago.edu>,
owen taylor <otaylor@midway.uchicago.edu> wrote:
>There are several possible solutions to this problem.  The cleanest,
>most useful to have, and definitely the hardest-to-implement solution
>would be to switch to a different executable format that supports the
>easy creation and linking of shared objects.  (I understand that ELF
>(or is it COFF?) fits this description.)  However, this would mean a
>major departure from current practice, and, (I think), the support for
>ELF shared libraries is not yet present in the the GNU tools.  For
>these reasons this does not seem practical in the short term, although
>it might be desirable in the long term.

        ELF is the one you would need.  COFF shared libraries are even 
worse than the current linux shared libraries.  The GNU tools do not 
support ELF shared libraries yet, but I have patches to support linking 
to shared libraries that should soon appear in the development snapshots 
at cygnus (once the gas/binutils-2.4 stuff settles down a little).  This 
code is not well tested, and you cannot as of yet generate a shared library
but this is enough for most end users.  Hopefully given a little bit of time
it should become stable, and eventually the support required to actually 
generate a shared library will be added.

>In contrast to the above approach, the approach I have taken is very
>elementary.  In essence, what I have done is modify the linker (the
>standard Linux linker based on the GNU linker v1.9) to optionally
>write out shared library information in combination with a relocatable
>object, creating the extra necessary relocations within this
>information.  I then modified dld to read this information and call
>ld.so.  This is definitely a bit of a hack/kludge, which is especially
>true of the modifications to GNU ld, which in the 1.9 form is a
>somewhat of an ungainly monster. With the addition of the Linux shared
>library additions + my patches, ughh!  The newer versions (current
>2.4) seem much cleaner, but are, unfortunately without Linux shared
>library support (a project for someone? It probably wouldn't offer,
>however, too many tangible benefits other than beauty.)

        This would indeed be a good project for someone.  At some point in
the future, we will try and transition to ELF shared libraries and having
ELF as the native binary format, and when this happens it will be handy to
have support for the current shared scheme.  I could do it, I suppose, but
it is currently low priority and I already have a pretty long list on my
TODO, so I will not get to it for the forseeable future (t>6 months). 

>I would also be interested in any comments about the general approach
>or information about any existing solutions.
>
>                              Owen Taylor
>                              o-taylor@uchicago.edu

-Eric

-- 
"The woods are lovely, dark and deep.  But I have promises to keep,
And lines to code before I sleep, And lines to code before I sleep."

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

From: ericy@cais.com (Eric Youngdale)
Subject: Re: 1.1.15 breaks SCSI
Date: 31 May 1994 22:38:53 GMT

In article <2s5ra8$8a8@uuneo.neosoft.com>,
John Lellis <lellis@dmccorp.com> wrote:
>Rene COUGNENC (rene@renux.frmug.fr.net) wrote:
>
>: The kernel 1.1.15 does not work at all with my Adaptec 1540B.
>: I get "Unable to reset SCSI host 0, probably a SCSI bus hang."
>
>: And I can't boot.
>
>Funny, works fine with my Adaptec 1542CF.  Are you sure it patched correctly?

        Sigh.  I go away for the weekend, and all hell breaks loose.

        In pl15, there were some changes to the handling of reset/abort.  
This was mainly intended to clean up the one last area of the mid-level 
code that was not really working right, and the main idea was that the 
low-level code can now report a variety of possible responses to the 
mid-level driver so that abort/reset cleanup and handling can be more exact.

        In addition, the 1542 driver was modified to attempt to do a bus 
device reset when we reached a point that the device completely failed to 
respond.  In some cases this appears to hang the bus or worse.  Switching 
back to the old case will be relatively easy, and will be my project for 
this evening.  I will of course look for bugs, but offhand I suspect that
some device goes insane when you attempt the bus device reset, and this
hangs the bus.  I will also look over the Ultrastor, as it appears as if 
something there is not quite right for some people.

-Eric

-- 
"The woods are lovely, dark and deep.  But I have promises to keep,
And lines to code before I sleep, And lines to code before I sleep."

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

From: d91mn@efd.lth.se (Mikael Nordqvist)
Subject: Re: Serial drivers in 1.1.16 (autoconfig for IRQ's)
Date: 31 May 1994 18:18:01 GMT

>>>As a remedy, use the setserial program in your /etc/rc*
>
>>It seems that setserial doesn't do the job under some circumstances (like
>>mine). When I use setserial to change irqs it reports them as changed, but
>>the device (a mouse) doesn't work. If I enable the autoconfig in the kernel
>>it works flawlessly without any need for setserial.
>
>You must run the setserial before you start the mouse application.
>(e.g. selection)

I did...

/Mikael
-- 
Mikael Nordqvist, student    | d91mn@efd.lth.se | What is GNU Hurd? A whole
Lund Institute of Technology | mech@df.lth.se   | bunch of GNUs running around.

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

From: rene@renux.frmug.fr.net (Rene COUGNENC)
Subject: Re: kernel 1.1.14 (&13) + xiafs segfaults rpc.nfsd
Date: 31 May 1994 21:36:00 GMT
Reply-To: cougnenc@blaise.ibp.fr (Rene COUGNENC)

I wrote:

> I have found one binary working fine, coming from  Slackware. But
> we are unable to compile a working version from nfs-server 2.0 with
> the current compiler and libs..


Guess what ?

This rpc.nfsd binary does not work any more with kernel 1.1.17 !
Signal 11 and bye... But an old binary (not working on 1.1.13->1.1.16)
seems to work !

(All of them still working fine with all kernels on the 486 box)
 
--
 linux linux linux linux -[ cougnenc@renux.frmug.fr.net ]- linux linux linux 

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


** 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
******************************
