Subject: Linux-Development Digest #944
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:     Sun, 24 Jul 94 18:13:05 EDT

Linux-Development Digest #944, Volume #1         Sun, 24 Jul 94 18:13:05 EDT

Contents:
  Re: X11R6 pl3 for Linux beta release (Pascal Haible)
  Re: Xfree86: increase pallate? (Harry Langenbacher)
  Re: GOTO haters ..Re: Linux Performance Enhance ? (the MOST SIGNIFICANT bit)
  Re: gcc and Linux (compiling)... (Reinhard Schiedermeier)
  Re: link() system call (Jim Balter)
  Re: procps-0.95 broken -- problem is in linux/sched.h (Sander van Malssen)
  ADAPTEC 2740 (Jose Miguel M.Sieuve Afonso)
  Big bug in 1.1.24 networking? (Michael Nelson)
  Re: Linux Performance Enhance ? (Erik Blass)
  Re: Wanted: The Application which can read *.DBF (David Dave Nuttall)
  Re: QMAGIC is nice, but impure ZMAGIC is smaller (Brandon S. Allbery)
  Re: Fix for problems with sound when system is idle. (Rob Janssen)
  Re: New kernel message at boot time? (Matthew Dillon)
  1.1.34 swap cache ?! (The Fantasy Adonis)
  Re: Anyone else lust for threads? (Patrick Schaaf)
  Re: Linux Performance Enhance ? (Jim Balter)
  tftpd, 16MB filesize limit ? (Robert Kaempf)
  Re: ARP broken ?! (H.P.Heidinger)
  Re: New kernel message at boot time? (Rene COUGNENC)
  Linux Programmer Needed ASAP (Pete Kruckenberg)

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

From: haible@izfm.uni-stuttgart.de (Pascal Haible)
Subject: Re: X11R6 pl3 for Linux beta release
Date: 24 Jul 1994 12:11:45 GMT

In article <30kpc5$2eo@adam.barrnet.net>,
Adam J. Richter <adam@adam.barrnet.net> wrote:
>       This is a beta release of X11R6 pl3 by Yggdrasil Computing,
>Inc., not by Xfree86, Inc. (so don't bombard them with bug reports or

              XFree86[tm] - please notice the upper case 'F'.

>blame them for our mistakes). 
[...]
>       The contents of the release is as follows:
>               
>XF86_8514.gz   X server for IBM 8514 compatibles
>XF86_Mach32.gz X server for ATI Mach32 cards
>XF86_Mach8.gz  X server for ATI Mach8 cards
>XF86_Mono.gz   X server for Hercules monochrome cards

The XFree86[tm] hercules server was not yet done when X11R6 was released.
If Yggdrasil didn't put several days of work based on deep knowledge
of the mfb library in this server, the only thing you will see from this
server is a core dump.

If this is true, the mentioning of this server should be moved to
the 'TOTALLY UNTESTED' section (and labeled with 'CAN'T WORK').

Pascal Haible, speaking for himself, not for XFree86[tm].
-- 
cand. mach. Pascal Haible haible@IZFM.Uni-Stuttgart.DE

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

From: harry@brain.jpl.nasa.gov (Harry Langenbacher)
Subject: Re: Xfree86: increase pallate?
Date: 22 Jul 1994 22:01:28 GMT

In article <CtBquL.DHy@sscnet.ucla.edu> johnsonp@nicco.sscnet.ucla.edu writes:
>In article <30ne0u$6gm@nic.umass.edu>,
>Christopher M. May <cmay@titan.ucs.umass.edu> wrote:
>>I heard that millions of colors would arrive with XFree 3.1
>Perhaps you are joking, but someone out there will believe you.  
>XFree86[tm] 3.1 will not increase the colors available.

well it depends on the server. not all hardware supports true color
display, and even if it does, the associated server may or may not
support a true-color visual. Right ???


                Harry
--
It gives us those _nice bright_ colors ... Gives us the _greens_ of summers
Makes you think all the world's a _sunny_day_, oh yeah !!
I got a _Nikon_ camera ... I love to take a _phot_o_graph_
So Momma, don't taaaaake my Kodachrome away           -- Paul Simon
-- 
H Langenbacher harry@brain.jpl.nasa.gov (818)354-9513
Concurrent Processing Devices Group, Jet Propulsion Lab
4800 Oak Grove Dr, Pasadena California 91109 USA

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

From: xmsb@borland.com (the MOST SIGNIFICANT bit)
Subject: Re: GOTO haters ..Re: Linux Performance Enhance ?
Date: Thu, 21 Jul 1994 06:13:26 GMT

In article <cairnss.774748222@ucsu.colorado.edu>,
cairnss <cairnss@ucsu.Colorado.EDU> wrote:
>What about the simple relationship between GOTO and your machine's
>assembly language?
>
>Maybe it's the FORTRAN in me but I prefer 
>       start:
>       if (cond)
>          {
>            stat;
>            goto start;
>          }
>to the "top-down" invention 
>       while (cond) {
>         stat;
>       }
>
>IF I was a compiler I would have much less difficulty
>generating efficient code from the first example.
>
    if you were a compiler, and didn't generate the same
    code for both of those code samples, I'd use a different
    compiler.

    if you think it's clearer, that's your opinion.  Since
    I don't need to read and/or maintain your code, it's hard
    to care. :)

--
Maurice S. Barnum               ==    I speak for me, not my employer.
xmsb@genghis.borland.com        ||    "There is no confusion like the
mosigbit@deeptht.armory.com     ||    confusion of a simple mind."
mbarnum@nyx.cs.du.edu           ==       -- F. Scott Fitzgerald

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

From: Reinhard.Schiedermeier@deejai.mch.sni.de (Reinhard Schiedermeier)
Subject: Re: gcc and Linux (compiling)...
Date: 21 Jul 1994 06:17:34 GMT

Dwight M Evers (evers@plains.NoDak.edu) wrote:
: Dwight M Evers (evers@plains.NoDak.edu) wrote:
: :     I was wondering if when you ( I in this case ) compile a program 
: : under Linux if the statement:
: : #include    <stdio.h>
:   main()
: ^^^^^^^^^ Sorry, sorry, sorry...I typed the post by had and read off a 
: listing. I DID include this in my program.
: : {
: :     value1 = 10;
: :     value2 = 20;
: :     sum = value1 + value2;

: :     printf(" the sum of %i and %i is equal to %i",value1,value2,sum);
: :     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
: : }
: :             is valid. I keep getting run errors. how do you compile 
: : and use gcc under Linux? I looked in the FAQ and I did what was told. Not 
: : a thing happens. I works under all other compilers I use.

What is "%i"? I always use "%d" to output integers.

--Schiedi

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

From: jqb@netcom.com (Jim Balter)
Subject: Re: link() system call
Date: Thu, 21 Jul 1994 06:50:18 GMT

In article <30dq7q$o8f@frigate.doc.ic.ac.uk>,
A N Burton <anb@doc.ic.ac.uk> wrote:
>Would anyone know if any of the file systems available under linux allow
>'root' to create _HARD_ links to directories?

For reasons shrouded in (pre-symbolic link) history, UN*X directories are not
allowed any links other than ".", the ".."s of its children, and one link from
its (unique) parent.  We took advantage of this fact when writing "Norton
Change Directory", and I was amused to note that the trick was independently
invented for GNU find (the trick is that the link count - 2 gives the number
of subdirectories, so when traversing a directory tree you can stop stat'ing
entries once you reach the count; for a directory like bin, with many entries
but no subdirectories, the speed advantage can be quite significant).
-- 
<J Q B>

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

From: svm@kozmix.hacktic.nl (Sander van Malssen)
Subject: Re: procps-0.95 broken -- problem is in linux/sched.h
Reply-To: svm@kozmix.hacktic.nl
Date: Sun, 24 Jul 1994 13:44:45 GMT

nickkral@po.EECS.Berkeley.EDU (Nick Kralevich) <nickkral@po.EECS.Berkeley.EDU> wrote:
> However, when I tried to compile it, I came up with these error 
> messages:
> 
> In file included from /usr/include/linux/fs.h:15,
>                  from /usr/include/linux/sched.h:86,
>                  from testfile.c:1:
> /usr/include/linux/net.h:104: parse error before `select_table'

[etc.]

You might want to try compiling with -D__KERNEL__ added to CFLAGS.

Cheers,
Sander
-- 
Sander van Malssen
svm@kozmix.hacktic.nl

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

From: breeders@skull (Jose Miguel M.Sieuve Afonso)
Subject: ADAPTEC 2740
Date: 24 Jul 1994 14:27:28 GMT

Is someone working in the support of the ADAPTEC Twinchannel SCSI 2740 card ? 
--
Joao Alves <opjoao@cc.fc.ul.pt>

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

From: mikenel@netcom.com (Michael Nelson)
Subject: Big bug in 1.1.24 networking?
Date: Sun, 24 Jul 1994 15:01:51 GMT

I know that 1.1.24 is old, but this problem was also exhibited by 1.0.9...

If I "tcpspray" a Linux 1.1.24 box from another machine about 5-10 times
consecutively (i.e. I have 10 tcpspray's on one machine to this Linux box
going on at the same time), the eth0 interface on the 1.1.24 box dies. If I
do ifconfig eth0 down, then up, everything starts working again
(networking-wise).

3Com EtherLink III (3C509)
486/33 with 8 megs

Any ideas?

mikenel@netcom.com

-- Mike
-- 

____________________________________________________________________________
Michael Nelson                                            mikenel@netcom.com
Windows NT / Linux / Windows 3.1                         MailPro Development
 PGP 2.3 Public Key - use finger or ftp: netcom.com:/pub/mikenel/pubkey.asc

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

From: root@i486.gondor.sub.org (Erik Blass)
Subject: Re: Linux Performance Enhance ?
Date: Thu, 21 Jul 1994 18:50:36 GMT

Jos Horsmeier (jos@and.nl) wrote:
: In article <774545127snz@genesis.demon.co.uk> fred@genesis.demon.co.uk writes:
: |In article <30b8vg$9ds@brachio.zrz.TU-Berlin.DE>
: |           goldt@math.tu-berlin.de "Sven Goldt" writes:

: |>Could you please discuss these C features (Duffy's Device) in
: |>comp.lang.c instead ? This is going off-topic.

: |Duff's device is certainly off-topic for comp.lang.c.

: Not that it matter much actually, but I beg to differ here. Duff's
: device is an application of a very sneaky loophole in the syntax
: of the C language with very interesting semantical consequences ...
: I think I like that thingie ...
Yes...yes very interesting, indeed.

But I'm (who posted the "original" question) more interested in
sending Linus this enhancement - ugly or not. So, if you haven't
already mailed the code to Linus, please do it. 

Also my original question was:

is it somehow possible to swap out a single process by command ?
"swap_out <PID>" ?

If not, why not write the swap_out programm ?! :-)

Greetings,
        Erik

-- 
Erik Blass|Internet erik@i486.gondor.sub.org|Z-Netz erik@stargate.zer
Ich habe Dinge gesehen, die ihr Menschen niemals glauben wuerdet. Gigantische
Schiffe die brannten an der Schulter des Orion. Und C-Beams - glitzernd in der
Dunkelheit nahe dem Tannhaeuser Tor. All diese Momente werden verloren sein in
der Zeit...so wie Traenen im Regen.       Zeit zu sterben...       [Roy Batty]
Hi! I'm a .signature virus!  Add me to your .signature and join in the fun!

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

From: dnuttall@cup.portal.com (David Dave Nuttall)
Subject: Re: Wanted: The Application which can read *.DBF
Date: Sun, 24 Jul 94 07:37:16 PDT

Regarding *.DBF files in Linux, I was told on 7/22/94 that a
commercial development product is now available for Linux.
The most interesting characteristic from my perspective is that
it allows CA-Clipper CODE to be compiled/run under Linux.  To
do that, it clearly has to have the capability of reading/writing
to *.DBF files.

I was told for a limited time, a single user copy is $200
and unlimited user development system is $500.

Interested and/or familiar developers can contact me off line.

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

From: bsa@kf8nh.wariat.org (Brandon S. Allbery)
Subject: Re: QMAGIC is nice, but impure ZMAGIC is smaller
Date: Sat, 23 Jul 1994 03:08:03 GMT

In article <30pgh6$b4c@sun.cais.com>, ericy@cais.cais.com (Eric Youngdale) says:
+---------------
| >executables, so far I haven't been able to generate an impure QMAGIC
| >Am I missing something ?
| 
|       Yes.  There is no such thing as impure ZMAGIC.  The format is 
| called NMAGIC as I recall.  Just as there is no such thing as impure 
+------------->8

Actually, it's OMAGIC.  NMAGIC is pure non-demand-paged (does anyone still use
that?).

++Brandon
-- 
Brandon S. Allbery         kf8nh@44.70.4.88               bsa@kf8nh.wariat.org
Friends don't let friends load Windows NT (tnx Sun)    A Linux iBCS2 developer

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

From: rob@pe1chl.ampr.org (Rob Janssen)
Subject: Re: Fix for problems with sound when system is idle.
Reply-To: pe1chl@rabo.nl
Date: Sun, 24 Jul 1994 11:36:48 GMT

In <30t6i7$rgj@backdoor.ethernet> traveler@backdoor.hacktic.nl (Oliver Daudey) writes:

>Some time ago my sound suddenly stopped working...  As soon as my system
>was idle, the sound would get scrambled.  I tried all kinds of things, like
>changing IRQ/DMA/address on my SB-Pro, without success...  The weird thing
>was that it worked perfectly on a friend's system with the same kernel, so
>I expected a hardware problem.  This was even more strange because it went
>away by booting an old kernel, which seemed to indicate a software problem..

>Just recently, it suddenly came to me that sys_idle() had been changed to
>halt the CPU when the kernel is idle.  I decided to give it a try and removed
>the hlt-instruction.  After that, everything suddenly worked again.  No more
>need to start a CPU-intensive program just to be able to play some MOD's. :-)

>I wonder if more people experienced this problem, as I didn't read anything
>about it in the Linux-groups.  It seems to be a hardware problem, triggered
>by software...  Maybe my motherboard handles the halts incorrectly.  I have
>a Highscreen (one of this "exclusive" Colani-designs  :-)  ).  The mother-
>board is a 4386-VC-V, equipped with a 486 DX/2-66, 8 Megs of RAM and 256K
>cache...  Chipset is VIA VT82C480...  The soundcard is a SB-Pro, which is
>reported at bootup as "snd2 <SoundBlaster Pro 3.2> at 0x220 irq 7 drq 1"...

I would suggest making sure that nothing else is using irq 7...
Normally this IRQ is used for the printer port.  Maybe your printer port
permanently drives this line, and therefore the IRQ from the soundblaster
does not get through reliably?
That would at least explain why it does not wakeup the CPU from the halted
state...

Rob
-- 
=========================================================================
| Rob Janssen                | AMPRnet:   rob@pe1chl.ampr.org           |
| e-mail: pe1chl@rabo.nl     | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU     |
=========================================================================

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

From: dillon@apollo.west.oic.com (Matthew Dillon)
Subject: Re: New kernel message at boot time?
Date: 22 Jul 1994 23:10:21 -0700

In article <30pf62$i0m@klaava.Helsinki.FI> torvalds@cc.Helsinki.FI (Linus Torvalds) writes:
:In article <30p3ks$cfm@crl5.crl.com>, Bogdan Urma <bogdan@crl.com> wrote:
:>   With the new kernel 1.1.33 I'm getting a new kernel message at boot time.
:..
:make using the remaining memory more efficient (4 bytes per physical
:page, to be exact: you have a 16MB machine?)
:
:With 1.1.33 (and now 1.1.34 which cleans up some more and fixes ptrace,
:for example), the kernel remembers which pages have been swapped out
:even after thay are swapped in again so that the next time they are
:swapped out they don't even need to be written to disk if they haven't
:changed in the meantime.  With a 4MB machine this should result in
:rather noticeable speedups of up to 30% when compiling the kernel, for
:example. 

    Wait, I'm confused... isn't the swap space deallocated when the page
    is paged in again so the total VM is physical+swap?  Is it leaving
    the swap space allocated or is it deallocating but remembering it,
    then invalidating the cache if somebody else allocates that swap
    page ?

    i.e. process A: swapout page X
         process A: swapin page X
         process B: swapout page Y to disk block previously allocated
                    by X.
         process A: swapout page X - must rewrite it to disk.

                verses

         process A: swapout page X
         process A: swapin page X, disk block deallocated.
         process B: swapout page Y to different disk block (have a moving 
                    index to allocate swap blocks so we do not reuse
                    deallocated blocks quickly?)
         process A: swapout page X - if not dirty, no need to re-write
                    to swap.

                verses

         process A: swapout page X
         process A: swapin page X, disk block NOT deallocated.
         process B: swapout page Y to different disk block
         process A: swapout page X - if not dirty, no need to re-write
                    to swap. 

    The reason I'm asking is that if the swap space on disk is not
    'deallocated' when the page is swapped in, then a machine with
    32MB of ram and 32MB of swap will *not* have 64MB of VM available.
    If it is 'deallocated', then that same configuration *will* have
    64MB of VM available.

    Also, I'm curious... I noticed this stuff in the patches :-)

                                                -Matt

-- 

    Matthew Dillon              dillon@apollo.west.oic.com
    1005 Apollo Way             ham: KC6LVW (no mail drop)
    Incline Village, NV. 89451  Obvious Implementations Corporation
    USA                         Sandel-Avery Engineering
    [always include a portion of the original email in any response!]


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

From: jdli@csie.nctu.edu.tw (The Fantasy Adonis)
Subject: 1.1.34 swap cache ?!
Date: 24 Jul 1994 17:36:19 GMT
Reply-To: jdli@csie.nctu.edu.tw


        I am using 1.1.34 now without any error/problem.  Thanks.
        
        But I have a question, why it seems that my swap useage
        
        will only increase and won't decrease ?!  Won't the swap
        
        be released after it was swap-out from HD ?
        
        Thanks for your answer ........
--

        E-mail : jdli@csie.nctu.edu.tw      u8117084@cc.nctu.edu.tw

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

From: bof@wg.saar.de (Patrick Schaaf)
Subject: Re: Anyone else lust for threads?
Date: Sun, 24 Jul 1994 18:55:38 GMT

rob@pe1chl.ampr.org (Rob Janssen) writes:

>It seems like it is time to re-post the clone document again :-)
>Maybe this time someone picks up the challenge and uses the info to
>really implement something?

*grin*

I already mailed it to David; perhaps he'll use it for something.

During the last reposting, somebody noted that the do_clone() code
didn't work without -fomit-frame-pointer. Here's a version
that should always work. Rob, please use this the next time :)

Patrick

int do_clone(unsigned long esp, unsigned long flags, int (*func)(void))
{ int ret;

  __asm__ (
           /* register setting (see input/output constraints):
            * %ebx is esp
            * %ecx is flags
            * %edx is func
            */
           "movl $" DEREF_STR(__NR_clone) ", %%eax\n\t"
           "int $0x80\n\t"
           /* error? */
           "jnc 1f\n\t"
           /* error. */
           "movl %%eax, _errno\n\t"
           "movl $-1, %0\n\t"
           "jmp 3f\n\t"
           "1:\n\t"
           "testl %0, %0\n\t"
           "jne 3f\n\t"
           /* the clone */
           "call *%3\n\t"
           "pushl $0\n\t"
           "call _exit\n\t"
           /* not reached */
           "1:\n\t"
           "jmp 1b\n\t"
           /* the parent */
           "3:\n\t"
           : "=a" (ret)
           : "b" (esp),
             "c" (flags),
             "d" (func)
  );
  return ret;
}

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

From: jqb@netcom.com (Jim Balter)
Subject: Re: Linux Performance Enhance ?
Date: Fri, 22 Jul 1994 14:14:47 GMT

In article <30821p$ji9@news.u.washington.edu>,
Tim Smith <tzs@u.washington.edu> wrote:
>In article <CsvDLF.tv@pell.com>, Orc <orc@pell.com> wrote:
>>   You want to put a Duffs Device into the kernel?
>>
>>
>>   Urk.
>
>Every kernel should have at least one cute, ugly, elegant, disgusting,
>wonderful, nauseating thing in it.  While nothing can really compare
>to the place deep in the Unix scheduler circa V7 where there was
>code like this:
>
>       /* you are not expected to understand this */
>       if ( setjmp(foo) && ! setjmp(bar) )

Actually I think it was aretu and retu, which were sort of like setjmp
but with insane semantics (as I recall, retu would restore the stack but
not the pc, so the code for the "setjmp != 0" case would follow the
"longjmp" call, but would have the wrong locals on the stack).


-- 
<J Q B>

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

From: kaempf@irx1 (Robert Kaempf)
Subject: tftpd, 16MB filesize limit ?
Date: 22 Jul 1994 14:36:33 GMT

Hi ! 

Today I started to expriment with tftpd on Linux.

I tried to transfer one 28MB file via tftp. 
But net-2 tftpd fails after about 15.9 MB of data. 

Any suggestions ? 

--
Robert Kaempf, THD, Hochschulrechenzentrum, Petersenstr. 30, D-64287 Darmstadt
E-Mail:  kaempf@hrz.th-darmstadt.de    Tel. +49 (0)6151 16-5588    Fax 16-3050

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

From: hph@hphbbs.E.open.DE (H.P.Heidinger)
Subject: Re: ARP broken ?!
Reply-To: hph@hphbbs.ruhr.de
Date: Sat, 16 Jul 1994 19:48:26 GMT

In article <1994Jul6.082426.3693@uk.ac.swan.pyr>,
Alan Cox <iiitac@uk.ac.swan.pyr> wrote:
# In article <1994Jul1.092248.4210@dagoba.priconet.de> strauss@dagoba.priconet.de (Frank Strauss) writes:
# >The simplest workaround is to set a static arp entry on the server by
# >"arp -s client cl:ie:nt:hw:ad:dr". This works for me, though it's not
# >the best way. Perhaps, we should take a look at the bootpd.
# 
# Yes bootpd has a bug, it fails to set the address family of the arp entry it
  ^^^^^^^^^^^^^^^^^^^^
# adds with SIOCSARP.

  It's mystical -- the very same 'bootpd' (from the Slackware 1.2 dist.)
  works perfectly with kernel 1.0. Other people reported that the
  problem arose with the 1.1.18 kernel. [and is still in it; I just
  installed 1.1.27 ... and ... Cherio! Here we go: 'ioctl(SIOCGARP) Protocol
  family not supported' is spit on the sys-logger]

  Under 1.1.18 I recompiled the NetNews-sources for BootpD-V2.2 ...
  ... but: "The same procedure as every year, James!"

  So I grabbed kernel 1.1.21. Made a 'make clean' first on the sources and
  started a fresh re-compile ... Bang: Compiler error: 'ATF_INUSE not
  found', since a line got lost in /usr/include/linux/if_arp.h:

        #define ATF_INUSE       0x01

  Well, I fixed this (under PL21 and now under PL27 again). Recompiling
  'bootpd' still lead to the ioctl-error with ARP-cache ioctls ... under
  kernel 1.1.27, though!

  Just to be sure, I then mounted the /support-area from my primary
  LINUX box at the secondary LINUX box [that with the distributed 1.0
  kernel], "unclamped" the `bootps' in inetd.conf and then started the
  new booptd from the NFS-mounted disk by hand ...
  Result: the boot-client loaded perfectly with the new bootpd as with the
  one which was delivered with Slackware. This points out definitly, that
  bootpd is NOT the culprit ... the ioctl-system in the kernel is ... 

  I hope this helps to fix the problem. If the author of the code shows up, 
  I am willing to do the integrity checks ... 

# 
# Alan
# 

Regards, Peter
-- 
#--------- H.P. Heidinger, Steeler Str. 121, 45138 Essen /Germany ----------#
# Call : +49-201-287433 (data/FAX)  # E-Mail : hph@hphbbs.ruhr.de           #
# A supercomputer is a machine, that runs an endless loop in just 2 seconds #

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

From: rene@renux.frmug.fr.net (Rene COUGNENC)
Subject: Re: New kernel message at boot time?
Date: 24 Jul 1994 00:11:10 GMT
Reply-To: cougnenc@blaise.ibp.fr (Rene COUGNENC)

Ce brave Linus Torvalds ecrit:

> With 1.1.33 (and now 1.1.34 which cleans up some more and fixes ptrace,
> for example), the kernel remembers which pages have been swapped out
> even after thay are swapped in again so that the next time they are
> swapped out they don't even need to be written to disk if they haven't

Since my mail and news are down for the week end, it may be fixed when
I write this article, but if not... As of patch 34, my syslog file
is filled by:

Jul 23 06:23:00 renux vmunix: swap_duplicate: trying to duplicate unused page
Jul 23 06:23:00 renux vmunix: swap_duplicate: trying to duplicate unused page
Jul 23 06:25:00 renux last message repeated 6 times
Jul 23 06:35:00 renux last message repeated 28 times

No problem with patch33...

--
 linux linux linux linux -[ cougnenc@renux.frmug.fr.net ]- linux linux linux 

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

From: pete@dswi.com (Pete Kruckenberg)
Subject: Linux Programmer Needed ASAP
Date: 24 Jul 1994 14:35:00 -0600

I am posting this at the request of Dale Dhillon of SDL.
(sdl@world.std.com).

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

SDL is looking for a Linux kernel programmer to help them develop a
Linux driver for their N2 interface board. This board is a dual-port,
co-processed v.35 interface for ISA-based computers. It can handle
aggregate speeds up to 12Mbps. Typical uses would be to connect to
dedicated digital circuits (56kb, T1) through a DSU/CSU.

SDL currently has BSDI and SCO drivers for this board. They are
committed to develop a Linux driver as quickly as possible. However,
they have limited resources to do so at this time. Using their own
resources, a driver would take months or even years to develop.

So, in the interest of accelerating the development, SDL is offering
to provide (permanently) their equipment to an experienced Linux
kernel programmer in exchange for developing the driver.  The driver
would be released under the GNU Public License (and hopefully make it
into the kernel code someday).  They will provide source code from
their other drivers, as well as any other information needed to do the
development.

It would be preferrable if you live near them (Boston area), but they
are anxious to get this development done and will talk to anyone who
is qualified to do this work. If you're interested, please contact
Dale Dhillon at (508) 238-4490. If you'd like to get more information
on-line, you can FTP to ftp.std.com directory /pub/sdl. They have
driver code, technical information, and info-sheets for their products
on-lin.

If you have any further questions, please call Dale at the number
above, or contact him at sdl@world.std.com.

Pete Kruckenberg
pete@dswi.com






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


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