Subject: Linux-Development Digest #574
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:     Wed, 23 Mar 94 14:13:09 EST

Linux-Development Digest #574, Volume #1         Wed, 23 Mar 94 14:13:09 EST

Contents:
  crash hosed my superblock...can it be saved? (Jerod Tufte)
  Re: 'format' program (Sun style) (Keith Medcalf)
  Re: IPX compliancy? (Alan Cox)
  PPP defroute problem - hosts, rc.inet1 config? (Eric Kimminau)
  Kernel don't recognizes HP PC LAN Ethernet Card ! (Thomas Heiling)
  Hanging CSLIP with ftp (Nick Vargish)
  Re: Real-Time Linux and a/d device drivers (Matthew Donadio)
  Re: <stddef.h> in libc-4.5.21? (Dwayne Springfield)
  plotter works with 0.99pl1[4,5], but not 1.0 (Denis Endisch)
  bug? or don't I get something? (Filip M Gieszczykiewicz)

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

From: jet5@pyrite.SOM.CWRU.Edu (Jerod Tufte)
Subject: crash hosed my superblock...can it be saved?
Date: 23 Mar 1994 10:40:15 GMT

I had my first crash in many weeks a few hours ago, and I can't seem to
be able to do anything to recover.  I was shutting down X, when the
system just suddenly seized.  I rebooted to get this message:
extra data not valid  Current error sd801 sense Key Illegal request
Additional sense indicates Logical block address out of range

and when I booted with an old recovery disk, I got:
in2000: 01 80 4b 36 00
Dumping sense buffer: 112 0 5 0 0 0 0 10 0 0
host 0 id 0 lun 0 return code 28000000
Sense class 7 sense error 0 extended sense 5
scsidisk I/O error dev 0801, sector 2
EXT2-fs unable to read superblock.

My system is a dx2/66 VESA m/b w/ IN2000 scsi and S3-805VLB video.
someone please tell me there's a tool that can help me recover some/all
of my files.  thanks.

Jerod Tufte
jet5@po.cwru.edu
--
jet5@po.cwru.edu       Case Western Reserve University   Computer Engineering
jet@b62528.student.cwru.edu  <<< finger me for PGP2.3a public key

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

From: kmedcalf@zealot.uucp (Keith Medcalf)
Subject: Re: 'format' program (Sun style)
Date: 23 Mar 1994 06:40:23 GMT
Reply-To: kmedcalf%zealot@io.org

W. Tait Cyrus (cyrus@jemez.eece.unm.edu) wrote:

: Last week I purchased a new SCSI disk and after only 6 days it quit
: working.  Fortunately I had made a backup off all my stuff the day
: before it died so I didn't loose a lot of data.

If the SCSI disk is properly designed, and properly implements the SCSI
spec, all that is required is to set the appropriate control pages, issue a
mode select, the issue a format unit with the certification bit set (which
is SUPPOSED to be the default).  Many non-SCSI compliant drive manufacturers
(particularly of low-cost low-reliability low-stability drives) do not
implement the SCSI spec correctly (most notable is Maxtor).

If you set the control mode pages, do a mode select and a default format
unit, and the drive develops an error within 2 years, you should send it
back to the manufacturer for replacement as it has a manufacturing or design
defect.

Manufacturers who create high quality drives will take them back for free
re-conditioning under warranty on a yearly basis (CDC, Fujitsu), and will
last for years of continuous heavy service with no problems.

Other drives (not to mention Maxtor again, but they make the most unreliable
and feature-free drives you can waste your money on) will last barely any
time at all before developing errors.

This is my opinion and experience only, and your mileage may, of course,
vary.  You will find however that a bottom-of-the-line Fujitsu drive will
operate and last much longer than a top-of-the-line Maxtor, and that the
Maxtor does not implement most of the features in SCSI that are most useful.

I have a CDC (real CDC, not ScuzGate) drive that is now coming up on 8 years
of continuous heavy service, almost 24 hours a day, 366 days a year, with
nary a hitch.  I have yet to see any maxtor drive last more than a year
under the same load ...

--
Keith Medcalf                | Internet: kmedcalf@zealot.uucp kmedcalf@io.org 
Keith Medcalf and Associates |  UUCP:  ... uunet.ca!io.org!zealot!kmedcalf
Database Systems Engineer    |  Anonymous UUCP:  (416) 465-4370  in: nuucp
Toronto/Canada (416)465-9578 |  Fidonet:  1:250/750 - IBMNet:  40:6482/303

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

From: iiitac@uk.ac.swan.pyr (Alan Cox)
Subject: Re: IPX compliancy?
Date: Tue, 22 Mar 1994 14:55:03 GMT

In article <1994Mar14.185508.46244@ucl.ac.uk> zceed04@ucl.ac.uk (Mr Ivan Alastair Beveridge) writes:
>Basically, I was wondering if anyone has made Linux compliant with Netware
>at all. As I do not really know much about protocols, I cannot really ask
>much more than this.

There is a beta test IPX layer for Linux, but no netware support. Novell
guards its netware details with lawyers and complex licensing agreements
involving thousands of dollars. So forget it - Linux does Lan manager and NFS


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

From: ekimmina@pms709.pms.ford.com (Eric Kimminau)
Subject: PPP defroute problem - hosts, rc.inet1 config?
Date: 22 Mar 1994 15:42:00 GMT

I have just completed upgrading to LINUX 1.0, GCC 2.5.8 & LIBC4.5.21.

PPP was working beautifully on 99.14 before the upgrade. I have a
static hostname and IP address on every connect, so I want to hardcode
my system files to use it. I am still unable to get a stable chat
script to work consistantly, but manually connecting via kermit will
work with an additional step. Here lies my problem:

Now, after connectiong to my ppp server and escaping back to the
kermit prompt and running pppd, I now have to manually add the default
gw route to get things working. Under 99.14, that route was added
automagically. I have tried with and without the defaultroute
command.

Could someone who has ppp working in that way under 1.0 send me a
couple things:

#1 A copy of your /etc/hosts with the significant portions (routeer,
net, localhost, etc. 

#2 A copy of your rc.inet1 or wherever you happen to configure lo
and/or ppp0 with your hostname, ip, default raoute, net and gateway.

#3 A copy of your chat (or kermit) startup script if you have it working.

Thanks in advance!

-- 
Eric Kimminau                       Workstation Systems Department
313-322-3431                        Product & Manufacturing Systems
ekimmina@pms709.pms.ford.com        Ford Motor Co.
Planning and Implementation         "Not an official Ford Spokesperson"
TIP#111

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

From: tom@wpzd07.pzlc.uni-wuerzburg.de (Thomas Heiling)
Subject: Kernel don't recognizes HP PC LAN Ethernet Card !
Date: 22 Mar 1994 15:57:30 GMT

Hello all!
Finally i got a new machine and tried to install
Linux. Everything fine, some fiddling with XFree 2.1,
and no the Problem:
I have a HP PC LAN Adapter / 16 TL+ (aka HP 27252A )
and compiled the Kernel ( Version 1.0 + patch 1 + 2 )
with CONFIG HP PCLAN. But after compiling, the Kernel
don't recognizes my Card.
Here is my Machine:
HP Vectra 486/66 MHZ, 8 MB,
integrated VGA Card Localbus with 512 KB
HP 27252A Ethernet Card /* This Card is mentioned in the
Ethernet-HOWTO as one of the best Cards, but there are no
hints / tips , why the Kernel don't recognizes it */

What i have done this weekend :
HP has a nice program  for setting up the card,
it is called hplanset.
I have done this :
Set the IO to 300-31fh
        IRQ to 10
        Station Adress is 080009-530D4C
compiled with CONFIG HP PCLAN -> no chance
compiled with CONFIG_LANCE -> no chance
compiled with every available Ethernet card -> no chance
Tried it with LILO : ether=10,0x300,0,0,eth0 -> no chance
Copied a kernel from this machine ( This is a 386/25 MHz
HP Vectra with only 4 MB and a HP LAN Card, the older model
with only 8 bit ) to the new machine, but it fails again.

And now i have absolutly no idea, what to do next.
Any help would be very nice.
Thanks Thomas



--
===================================================
Thomas Heiling Pharmacist & Doctorate at 
Pharmazeutisches Institut Uni Wuerzburg - Germany 
Email phar006@wrzx12.rz.uni-wuerzburg.de (HP-UX)
      tom@wpzd07.pzlc.uni-wuerzburg.de (Linux)
or phar006@vax.rz.uni-wuerzburg.de ( VAX )
===================================================

--
===================================================
Thomas Heiling Pharmacist & Doctorate at 
Pharmazeutisches Institut Uni Wuerzburg - Germany 
Email phar006@wrzx12.rz.uni-wuerzburg.de (HP-UX)
      tom@wpzd07.pzlc.uni-wuerzburg.de (Linux)
or phar006@vax.rz.uni-wuerzburg.de ( VAX )
===================================================

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

From: vargish@yx.sura.net (Nick Vargish)
Crossposted-To: comp.os.linux.misc
Subject: Hanging CSLIP with ftp
Date: 23 Mar 1994 14:38:41 GMT

Folks,

Here's the scoop... All I have to do to hang my CSLIP connection is
try to ftp an uncompressed xpilot map across the connection. Once I
compress the map (which I will include here), the transfer goes
smoothly.

I suspect it has something to do with long strings of identical
characters and the compression algorithm(s), but I'm basically taking
an educated guess (no facts or code-grazing to back me up).

I thought someone might be interested in this, assuming it's not some
boneheaded mistake on my part.

And by the way, I meant _hung_, so don't try this unless you're
willing to restart your CSLIP connection...

Nick Vargish

p.s. This is the map I was talking about, gzipped and uuencoded. It
should work gzipped, and fail ungzipped...

============================================ CUT HERE
begin 644 13.map.gz
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E#S2K"C3V0&-/'31^[7!HH*F;QL/W4?@15GB%W!OP"[TWED\  #ZV
 
end
============================================ CUT HERE


-- 
 ----------------------     ----------------------     -----------------------
 |.   Nick Vargish   .|     |.     SURAnet      .|     |. O: (301) 982-4600 .|
 |. vargish@sura.net .|     |. systems engineer .|     |. rust never sleeps .|
 ----------------------     ----------------------     -----------------------

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

From: donadio@mxd120.rh.psu.edu (Matthew Donadio)
Subject: Re: Real-Time Linux and a/d device drivers
Date: 22 Mar 1994 16:42:57 GMT

Kevin Brown (kevin@frobozz.sccsi.com) wrote:
: There may be something I'm overlooking, of course (which wouldn't be
: surprising), but I'm somewhat convinced that between the above and some
: kind of semaphore implementation, you should be able to do pretty much
: everything you'd expect out of a "real" threads implementation.

You can emulate a lot with fork(), but in my experience the
applications I have written using pthreads() seem to behave better.
Mutexes can be done with semaphores, but conditions (as in
pthread_condition) get a little hairy when you have a master process
controlling a bunch of slave processes.  Plus, pthreads lets you give
up the processor with pthread_yeild() which can make threaded
applicaitons more effieient.

--
Beaker aka Matt Donadio   | Life is short,     ---   __ o    __~o    __ o
donadio@mxd120.rh.psu.edu |    ride like    ----    _`\<,   _`\<,   _`\<,
--- Penn State Cycling ---|      the wind.    ---  ( )/( ) ( )/( ) ( )/( )

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

From: dwaynes@netcom.com (Dwayne Springfield)
Subject: Re: <stddef.h> in libc-4.5.21?
Date: 20 Mar 94 18:54:06 GMT

Joseph Toman (toman@darkwing.uoregon.edu) wrote:
: Hi, I am trying to compile various source code packages for "Lee-noocks" :)
: and I don't seem to have the ANSI C standard include file <stddef.h>.  It is
: neither in Slackware 1.1.2 nor in libc-4.5.21 on tsx-11.  Where can I find it?

: Thanks, Johannes
sometimes becuase of the gcc installation that file gets moved to 
a relatively obscure spot in the directory hierarchy....
if you have not tried using the find command to locate the file
(find / -name stddef.h -print) then you just might find the file
in a directory path resembling but not necessarily exactly like --->
/usr/lib/gcc-lib/i486-linux/2.5.7/include/stddef.h
Like I said... obscure.  You will of course adjust OS and version numbers 
according to your own installation :)

hope this helps.

l8r,
dwayne


--
+------------------------------+-------------------------------------------+
| Dwayne Springfield           | Time flies like an arrow...               |
| dwaynes@netcom.com           | Fruit flies like a banana.                |
| dwayne.springfield@octel.com |            Groucho (Karlo's Brother) Marx |
+------------------------------+-------------------------------------------+

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

From: dendisch@cat..physics.uwo.ca (Denis Endisch)
Subject: plotter works with 0.99pl1[4,5], but not 1.0
Date: 22 Mar 94 21:07:49 GMT

Hi there

I have a plotter (HP ColorPro) attached to /dev/cua1.
It worked fine with plain 0.99pl14 and 0.99pl15, but with 1.0
it seem to loose characters from time to time. Sometimes the plots
are ok, but most of the time the plotter starts going crazy in
the middle of the plot, obviosly getting some wrong vector values.

For the older kernels (pl14,pl15) I have NOT used the kernel patch 
mentioned in the Printing-HOWTO (Dec 1 93, included at the end of 
this posting) for setting up a serial printer at that time, since 
it worked fine without it.

In the current Printing-HOWTO (Feb 22 94) the patch is not mentioned any 
more. Also the relevant lines in /usr/src/linux/kernel/chr_drv/tty_ioctl.c 
have changed, so that the older patch can not be applied anyway. It 
looks as if some kind of bug was implemented by this recent changes.

Does anybody has similar problems or a idea what's going wrong?
Or a idea for solving the problem? 
My current solution is not to use 1.0 :-(

Thanks Denis

====================================================================
Some more information:
I have the plotter running with lpd connected to a dummy device, as
it is described in the Printing-HOWTO. The setting for the serial
port are made at boottime with 'stty'. I have set up these settings
with the 0.99pl14 kernel and have nothing changed since then.

====================================================================
My stty settings are:
cat:~$ stty -a < /dev/cua1
speed 9600 baud; rows 0; columns 0; line = 0;
intr = ^C; quit = ^\; erase = ^?; kill = ^U; eof = ^D; eol = <undef>;
eol2 = <undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R; werase = ^W;
lnext = ^V; flush = ^O; min = 1; time = 0;
-parenb -parodd cs8 hupcl -cstopb cread -clocal crtscts
-ignbrk -brkint ignpar -parmrk -inpck -istrip -inlcr -igncr -icrnl ixon -ixoff
-iuclc -ixany -imaxbel
-opost -olcuc -ocrnl -onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab3 bs0 vt0
ff0
-isig -icanon -iexten -echo -echoe -echok -echonl -noflsh -xcase -tostop
-echoprt -echoctl -echoke

====================================================================
From the older Printing-HOWTO (1 Dec 93) about setting up a serial printer:
...
First, there is a kernel patch you need to make. This applies
to  pre-net-2 lpd  as well  (actually, the  use of  this should
probably just  be commented  out of  the lpd  source code). The
change is for  at least patchlevel 10,  but probably all others
as well.

in /usr/src/linux/kernel/chr_drv/tty_ioctl.c, around line
390, you will see:

        case TIOCEXCL:
        return -EINVAL; /* not implemented */

Change this to return 0,  recompile the kernel, and you'll be
  all set.
...
====================================================================

I can give more info about system configuration, if this is important.

--
Denis Endisch                     Phone:  (519) 661 - 2111x6413
Department of Physics             Fax:    (519) 661 - 2033
Univ. of Western Ontario          Email:  dendisch@uwo.ca
London, Ontario, Canada N6A 3K7

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

From: filip@alpha.smi.med.pitt.edu (Filip M Gieszczykiewicz)
Subject: bug? or don't I get something?
Date: 23 Mar 94 16:45:10 GMT


        Greetings. I ran a butchered version of iozone 1.6 (old) with
        mods to allow for concurrent execution (to test just to what
        extent IDE sucks in multi-tasking applications - oh, btw, it
        sucks the chrome off the bumpers... more later)

        Anyway, for this little test to be "fair", I had to shoot 
        Linux buffers in the foot (by allocating most of my memory
        with an "eatmem" program). I left about 1.5MB for the
        multiple instances (up to 20) of the 23K executable.

        What I noticed was that linux was groping the swap even 
        though there was ~1MB in the buffers. After a few [dozen]
        trials, it was REALLY eating swap like there was no tomorrow!

        I'd like to know something before I go any further: What does
        linux prefer to do a) write/read directly to disk or b) write/
        read to buffers which are really swap [sigh]. Also, can it
        tell the difference? Further, is there anything that can be
        done to help this?

        Also, my system is as follows: Linux 1.0 kernel (debugging on,
        SCSI compiled in (not used)), xiafs FS on a 200MB Maxtor IDE
        (17 ms), 486DX 2/66 VLB, 16MB ram, 13MB swap (partition), 
        256K cache, 11MHz ISA buss, 33MHz VLBuss. (note that eatmem
        program "ate" 11MB)

        I include a list of "free" after it started looking fishy:

=================chop=with=axe================chop=with=axe=====================
             total       used       free     shared    buffers
Mem:         13748      13668         80       1268       1536
Swap:        13368       3096      10272

             total       used       free     shared    buffers
Mem:         13748      13432        316        976       1864
Swap:        13368       3508       9860

             total       used       free     shared    buffers
Mem:         13748      13668         80       1336       2212
Swap:        13368       3768       9600

             total       used       free     shared    buffers
Mem:         13748      13668         80       1624       2480
Swap:        13368       4488       8880

             total       used       free     shared    buffers
Mem:         13748      13164        584        912       2832
Swap:        13368       4608       8760

             total       used       free     shared    buffers
Mem:         13748      13668         80       1408       3048
Swap:        13368       4728       8640

             total       used       free     shared    buffers
Mem:         13748      13400        348       1000       3180
Swap:        13368       4856       8512
             total       used       free     shared    buffers
Mem:         13748      13384        364       1000       3684
Swap:        13368       5376       7992

             total       used       free     shared    buffers
Mem:         13748      13384        364       1000       3752
Swap:        13368       5444       7924

             total       used       free     shared    buffers
Mem:         13748      13624        124       1008       4012
Swap:        13368       5464       7904

             total       used       free     shared    buffers
Mem:         13748      13584        164        836       5420
Swap:        13368       6640       6728

             total       used       free     shared    buffers
Mem:         13748      13668         80       1320       4852
Swap:        13368       6712       6656
=================chop=with=axe================chop=with=axe=====================

        By this time ~5 hours & maybe 30-40 trials had been run. Right
        now, as I write this, "free" says:

             total       used       free     shared    buffers
Mem:         13748      13012        736       1056       4860
Swap:        13368       6740       6628

        And, let me exit the eatmem,

             total       used       free     shared    buffers
Mem:         13748       7268       6480       1044       4864
Swap:        13368       1120      12248

        Well, the swap looks better but... why was I loosing memory
        above... Is that a "fact of life", is it a bug, is it a leak,
        is it paranoia, what is it?

        Also, why does linux keep buffers @ 4MB even though it [should]
        know that they are really swap. Or is that the problem... it
        doesn't know the buffers are swap... eh?

        BTW, I case anyone is wondering what kind of #'s I got from
        my benchmarks... If you're planning to get an IDE, you 
        DEFINITELY want to read this (hopefully you'll change your
        mind):

        Well, the best (or worst, depending on your point of view)
        test was 20 instances of iozone writing/reading a 1MB file.

RESULTS:

        20 : concurrent jobs
         1 : MB (file size)
     10425 : bytes/second READ
     89767 : bytes/second WRITE

        [sniff..sniff..Whhhaaaaaa....]

        Right behind (or in front :-) was 10 instances of 2MB iozone.

RESULTS:

        10 : concurrent jobs
         2 : MB (file size)
     20274 : bytes/second READ
     76145 : bytes/second WRITE

        It got better with less instances... 1 instance of 20MB was

RESULTS:

         1 : concurrent jobs
        20 : MB (file size)
    302009 : bytes/second READ
    552318 : bytes/second WRITE

        which just goes to prove that while IDE is "fast" in a single
        task environment like MS-DOG or, to a lesser extent, MS-WINDOZE,
        it really reeks for anything that has 2 or more processes writing
        to the disk. And, God forbid, you have 20 users on line...

        This is my BEFORE segment... I'll rerun all these tests after
        I get my BT445S & 1.8GB :-) and post the AFTER...

        I can upload the kludged iozone to sunsite so others can have
        a go at it (if you don't believe me). A more complete version
        of this article will be posted to comp.periphs.scsi to stop
        silly flame-fest on IDEvsSCSI speed...

        Take care (sorry this is so long and I hope the first item can
                   solved. Thanks!)
--
+-->Filip "I'll buy a vowel" Gieszczykiewicz | E-mail: filip@alpha.med.pitt.edu
| ftp to bode.ee.ualberta.ca and get OLD electonics faq from /pub/cookbook/faq
| Checkout: ftp bode.ee.ualberta.ca and get file /pub/cookbook/contents. Neat.
| Making money with CS and spending it on EE, robotics, windsurfing, & dreams.

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


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