Linux-Development-Sys Digest #203, Volume #6 Sat, 2 Jan 99 13:16:17 EST
Contents:
Re: Poor NFS performance from Solaris to Linux (Steve Linton)
Backing up Processes (Klaus Niederkrueger)
Re: lp0 on fire in 2.1.131 (Jonathan Buzzard)
Digital joystick support in 2.1.x ("D. Stimits")
Re: Programming CDROM (Peter Samuelson)
Re: lp0 on fire in 2.1.131 (James Youngman)
Re: What about "Linux.. the home game"?? (a consumer version) (David T. Blake)
Re: Support for new (cheap) wireless networks? ("Daniel A. Jablonski")
Re: Registry for Linux - Bad idea (Caspian Maclean)
Re: Electric Fence 100x slower in 2.1.130 than 2.0.35? (jwk)
Re: CD Drivers (Carlos Vidal)
Re: Kernel-2.2-pre2 (Paul Martin)
Re: Electric Fence 100x slower in 2.1.130 than 2.0.35? ("T.E.Dickey")
Re: Counting page faults ("Matej Artač")
Re: Registry for Linux - Why? (Craig Kelley)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Steve Linton)
Subject: Re: Poor NFS performance from Solaris to Linux
Date: Sat, 02 Jan 1999 07:38:58 GMT
[EMAIL PROTECTED] (Pete Zaitcev) writes:
>James Hewitt <[EMAIL PROTECTED]> writes:
>>This is an old problem, but it doesn't seem to have been fixed yet.
>>NFS data transfer speed between a server running Solaris 2.6 and a
>>PC running Linux (Slackware 3.6 with kernel 2.1.130) is 5 to 8
>>times slower than between two Solaris 2.6 machines. This is true
>>even using udp and NFS version 2 between the Solaris machines.
>Traditionally it is attributed to bad default transfer size.
>Did you mount with rsize=8192,wsize=8192?
The full story is the following:
1. NFS specifiies that a write transaction cannot complete until the
data is physically on the disk
2. This is utterly inappropriate for modern network speeds
3. Sun get round this using a client-side daemon to hide the latency
4. Linux gets round this by ignoring the NFS spec on the server side
5 Hence Sun server and Linux client gets the worst of all possible
worlds.
The problem can be ameliorated by increasing the size of the NFS
transfers, as suggested. Try also 4K and 16K sizes.
On Solaris 2.6 it may be possible to set an option to the NFS server
to make it work asynchronously.
Runnign Linux on the Sun file server is another option.
Steve
------------------------------
From: Klaus Niederkrueger <[EMAIL PROTECTED]>
Subject: Backing up Processes
Reply-To: [EMAIL PROTECTED]
Date: Sat, 02 Jan 1999 07:39:48 GMT
Hi,
I had last week the following problem:
I wanted to reboot my machine but had a calculation running for 2 weeks.
It does not matter anymore, but still I was wondering if it would have
been possible to store the running process to disk, reboot and then
restart this process again.
If it is not possible, I think that a real cool thing would be to make
the /proc filesystem even more compatible to normal directories by
allowing e.g. "chown" on the processes, so that you could change the
owner of processes or "cp" files from /proc to the real harddisk to copy
it later back to /proc and having it running like before copying it. If
some kernel-hacker is bored, maybe he could try to implement those
things. I have no idea about kernel-programming, but I like the idea
much.
By the way I would like to know, what the "procesor-type" for
kernel-compilation is -- I am using an AMD-K6 and trying to install a
2.0.xx kernel. Just Pentium or PentiumPro(686)??
Thanks
Klaus
------------------------------
From: [EMAIL PROTECTED] (Jonathan Buzzard)
Subject: Re: lp0 on fire in 2.1.131
Date: Sat, 02 Jan 1999 07:39:22 GMT
In article <[EMAIL PROTECTED]>,
Lars Clausen <[EMAIL PROTECTED]> writes:
> On 13 Dec 1998, bill davidsen outgrape:
[SNIP]
> The 'lp0 on fire' error message exists in most Unixes (except where
> over-eager serious people have forced it out), just like the tunafish note
> in the man entry for tunefs. I think it dates back to Bell Labs, but I'm
> not sure.
>
My understanding, and I can't remember where I got the idea, is that back
in the early days of Unix, a printer was failing to acknowledge it's
readyness to recieve data. On investigation (ie. going to the room in which
the printer was housed), it turned out the printer was on fire. Accordingly
the source to the printer daemons was modified to reflect this newly
discovered error condition.
JAB.
--
Jonathan A. Buzzard Email: [EMAIL PROTECTED]
Northumberland, United Kingdom. Tel: +44(0)1661-832195
------------------------------
From: "D. Stimits" <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
Subject: Digital joystick support in 2.1.x
Date: Sat, 02 Jan 1999 07:40:32 GMT
Despite having module support for digital joysticks (in my case wingman extreme), I've
been unable
to use the joystick. I'm wondering if anyone out there has made a joystick
configuration or testing
program that has worked for digital sticks, especially the wingman extreme?
Thanks!
D. Stimits
------------------------------
From: [EMAIL PROTECTED] (Peter Samuelson)
Subject: Re: Programming CDROM
Reply-To: Peter Samuelson <[EMAIL PROTECTED]>
Date: Sat, 02 Jan 1999 07:45:20 GMT
[Gyorgy Krajcsovits <[EMAIL PROTECTED]>]
> What I did was to write a simple (about 45 lines) Tcl/Tk script that
> has a button for (mount/umount and eject).
I think what you want is the supermount patch by Stephen Tweedie. I
haven't tried it, but the way it's supposed to work is that the kernel
mounts something removable (floppy/cd/zip/etc) but lets you change
media out from under it, and automatically does the unmounting and
remounting when necessary. Also avoids delayed writes, for obvious
reasons. The idea is to make removable media behave as it does under
MS-DOS. Find supermount at:
http://sunsite.unc.edu/pub/Linux/kernel/patches/diskdrives/
IIRC supermount doesn't poll but checks the media when it's actually
accessed. (Sort of like automount in this respect.)
--
Peter Samuelson
<sampo.creighton.edu!psamuels>
------------------------------
From: James Youngman <[EMAIL PROTECTED]>
Subject: Re: lp0 on fire in 2.1.131
Date: Sat, 02 Jan 1999 07:46:40 GMT
[EMAIL PROTECTED] (bill davidsen) writes:
> I had a chance to show someone Linux, and they had three computers and
> four printers, all of which work under Win95 and SCO OpenServer. Not
> only did no combination of printer and system work, the "lp0 on fire"
> message created the impression that Linux was an in joke. Having code
> written by people in their teens is a strength of free software, but
> perhaps massages like lp0 on fire could at least be expanded to include
> the info that it means the kernel wasn't able to determine what the
> error was (if any).
This is a traditional feature of Unix, it isn't peculiar to Linux.
--
ACTUALLY reachable as @free-lunch.demon.(whitehouse)co.uk:james+usenet
------------------------------
From: [EMAIL PROTECTED] (David T. Blake)
Subject: Re: What about "Linux.. the home game"?? (a consumer version)
Date: Sat, 02 Jan 1999 07:39:25 GMT
Jeff Johnson <[EMAIL PROTECTED]> writes:
> We already have a high-end, stable, performance operating
>system/environment with Linux. Why not strip it down and come up with a
>single-user, GUI only environment version of Linux strictly for the home
>use, non-computer professional. I mean hell, behind the GUI desktop does
>it matter whether its DOS32 or a Linux kernel running the show?
Didn't this come up last week with the leanux thread ?
What does GUI only add to GUI over text config ?
What does single user add to multi-user ?
Have you seen what a VAR box installed with KDE and xdm (run level 5)
looks like ?
The issue is that this product already exists and can be had pretty
easily.
--
Dave Blake
[EMAIL PROTECTED]
------------------------------
From: "Daniel A. Jablonski" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.networking
Subject: Re: Support for new (cheap) wireless networks?
Date: Sat, 02 Jan 1999 06:48:42 GMT
==============E344A8F8F34AC27A3D0E9ED0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Rupa Schomaker wrote:
Harris Semiconductor's latest-greatest product is the PRISM chip - a
wireless networking chip, and I know that a linux driver exisits. Check
out: wwwnt.semi.harris.com/prism/prism.htm . There are links for
contact information there, where you could probably find out more.
Good luck.
-DJ
> I've seen drivers for the older more expensive (and potentially less
> supported) networking solutions from Digital and IBM (and others). Is
> anyone writing drivers to support the newer and cheaper 2.4Ghz
> wireless networking products that are being introduced? For example,
> Diamond has what looks like a nice product in the $100/node range.
>
> I would love to ditch the network cable on my laptop, but can't do it
> until my linux server supports one of the wireless network cards.
>
> So, anyone working on these? Alpha drivers available? :)
>
> --
>
> -rupa
--
=============================
Daniel A. Jablonski
[EMAIL PROTECTED]
=============================
==============E344A8F8F34AC27A3D0E9ED0
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit
<HTML>
Rupa Schomaker wrote:
<P>Harris Semiconductor's latest-greatest product is the PRISM chip - a
wireless networking chip, and I know that a linux driver exisits.
Check out: <A
HREF="http://wwwnt.semi.harris.com/prism/prism.htm">wwwnt.semi.harris.com/prism/prism.htm</A>
. There are links for contact information there, where you could
probably find out more.
<P>Good luck.
<P> -DJ
<BLOCKQUOTE TYPE=CITE>I've seen drivers for the older more expensive (and
potentially less
<BR>supported) networking solutions from Digital and IBM (and others).
Is
<BR>anyone writing drivers to support the newer and cheaper 2.4Ghz
<BR>wireless networking products that are being introduced? For example,
<BR>Diamond has what looks like a nice product in the $100/node range.
<P>I would love to ditch the network cable on my laptop, but can't do it
<BR>until my linux server supports one of the wireless network cards.
<P>So, anyone working on these? Alpha drivers available? :)
<P>--
<P>-rupa</BLOCKQUOTE>
<PRE>--
=============================
Daniel A. Jablonski
[EMAIL PROTECTED]
=============================</PRE>
</HTML>
==============E344A8F8F34AC27A3D0E9ED0==
------------------------------
From: [EMAIL PROTECTED] (Caspian Maclean)
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.setup
Subject: Re: Registry for Linux - Bad idea
Date: 2 Jan 1999 19:35:35 +0800
In article <76c9mk$ma5$[EMAIL PROTECTED]>,
Christopher Browne <[EMAIL PROTECTED]> wrote:
>I don't have a URL handy (I ought to); there has been a port/simulation
>of part of the Hurd "translator" that is used to mount processes as if
>they were filesystems *to Linux.*
>It is *probably* linked in somewhere around
><http://www.fig.org/gord/hurd/hurd.html>.
>Gord, whose site that is, was the writer of the Linux-based translator.
>From his sig on the debian-hurd mailing list:
[Unfortunately, www.fig.org is broken. Please stay tuned for details.]
------------------------------
From: [EMAIL PROTECTED] (jwk)
Subject: Re: Electric Fence 100x slower in 2.1.130 than 2.0.35?
Date: Sat, 02 Jan 1999 16:02:42 GMT
Reply-To: [EMAIL PROTECTED]
On Sat, 02 Jan 1999 07:47:19 GMT, Stefan Monnier
<[EMAIL PROTECTED]>
wrote:
>>>>>> "Dan" == Dan Kegel <[EMAIL PROTECTED]> writes:
>> The difference is probably Electric Fence; if I compile
>> a non-debugging build of the program (without electic fence),
>> it finishes in a few minutes, like normal.
>
>If you end up running your program with Electric Fence turned on all the time,
>then you're probably better off staying away from C and switch to another
>language where such checks are *much* cheaper (because mostly done at compile
>time). Any language apart from C, Pascal, C++ and Fortran should fill the
>bill: faster in this context, easier to write, ...
>Automatic memory management goes a long way already.
>
>
> Stefan
there were some messages on the kernel list saying that for electric
fence, you'd better use an ac* kernel (from Alan Cox), because IIRC
those had AVL-memory-tree support in them. I don't know what it is,
but you could try it!
Jurriaan
------------------------------
From: Carlos Vidal <[EMAIL PROTECTED]>
Subject: Re: CD Drivers
Date: Mon, 28 Dec 1998 08:52:09 +0000
Claudia wrote:
> Hi,
> I have a Creative CD820E.1v788200 8X CD-ROM
> Where can i get the drivers to use it on linux
It's an IDE CD-ROM so it should work out of the box. Ifyou cannot access
it look at:
/usr/src/linux/Documentation/cdrom/ide-cd
If you don't have Linux installed yet you can find this file
in some of the several Linux files.
--
Carlos Vidal
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Paul Martin)
Subject: Re: Kernel-2.2-pre2
Date: 2 Jan 1999 15:14:46 GMT
In article <[EMAIL PROTECTED]>, Ulrich Küttler wrote:
>Compiling the new kernel I got:
>
>gcc -D__KERNEL__ -I/usr/src/linux-2.2.0-pre2/include -Wall
>-Wstrict-prototypes -O2 -fomit-frame-pointer -pipe -fno-strength-reduce
>-m486 -malign-loops=2 -malign-jumps=2 -malign-functions=2 -DCPU=686 -c
>-o init/main.o init/main.c
>/usr/src/linux-2.2.0-pre2/include/asm/bugs.h: In function
>`check_config':
>In file included from init/main.c:27:
>/usr/src/linux-2.2.0-pre2/include/asm/bugs.h:354: `smp_found_config'
>undeclared
>(first use this function)
>/usr/src/linux-2.2.0-pre2/include/asm/bugs.h:354: (Each undeclared
>identifier is reported only once
>/usr/src/linux-2.2.0-pre2/include/asm/bugs.h:354: for each function it
>appears in.)
>make: *** [init/main.o] Error 1
>
>It seems to me there is something like '#include <asm/smp.h>' and
>'#ifdef __SMP__' missing in 'asm/bugs.h'. That's just a guess.
>By the way: I tried to compile without SMP.
include/asm-i386/bugs.h
Find the line containing smp_found_config. Add an #ifdef __SMP__ / #endif
around that.
--- include/asm-i386/bugs.h.old Thu Dec 31 05:58:13 1998
+++ include/asm-i386/bugs.h Sat Jan 2 14:43:43 1999
@@ -351,8 +351,10 @@
/* If we were told we had a good APIC for SMP, we'd better be a PPro */
#ifdef CONFIG_GOOD_APIC
+#ifdef __SMP__
if (smp_found_config && boot_cpu_data.x86 <= 5)
panic("Kernel compiled for PPro+, assumes local APIC without
read-before-write bug");
+#endif
#endif
}
(Apologies for the broken patch, but my newsreader won't let me post
articles with lines longer than 80 characters. The lines with the panic()
call should be a single line.
--
Paul Martin <[EMAIL PROTECTED]>
at home, swap dash to dot to email.
------------------------------
From: "T.E.Dickey" <[EMAIL PROTECTED]>
Subject: Re: Electric Fence 100x slower in 2.1.130 than 2.0.35?
Date: 2 Jan 1999 15:13:22 GMT
jwk <[EMAIL PROTECTED]> wrote:
> there were some messages on the kernel list saying that for electric
> fence, you'd better use an ac* kernel (from Alan Cox), because IIRC
> those had AVL-memory-tree support in them. I don't know what it is,
> but you could try it!
AVL trees are binary-trees. Presumably someone's got a slow memory
manager that searches linked lists - linear searches are slow.
--
Thomas E. Dickey
[EMAIL PROTECTED]
http://www.clark.net/pub/dickey
------------------------------
From: "Matej Artač" <[EMAIL PROTECTED]>
Subject: Re: Counting page faults
Date: Sat, 2 Jan 1999 18:09:08 +0100
Hello again!
The information about the number of page faults is indeed in each process'
stat file.
About the allocation of really big blocks, I am still confused though. To
test and demonstrate this, I have written a small program that allocates an
array with an increasing size to see how big the array would be before
malloc returns NULL. It got to 50MB, which is still less than 64MB. The
funny thing is, I put another malloc in the same loop, and got two 50MB
arrays. I suppose this would do for what I'm trying to do (which is allocate
a block that would not fit into physical memory, then write to and read from
various locations of this array and count page faults - it's for a faculty
project, by the way...), but it's strange anyway.
The program (dumprlimits.c and malloctest.c) also dumps the values acquired
by rlimits, and the output is in output.txt file. I have also put Linux's
ulimit output in the ulimitdump.txt for the refference. I hope you find all
these files attached to this message. My Linux kernel is 2.0.36, which means
that memory can be overcommitted, right? (In Windows, committed pages can
get paged to disk by the system as well - is it the same in Linux?).
Thanks.
Matej
begin 666 files.zip
M4$L#!!0````(``>4(B8Y1O1 =0$``&\#```,````;6%L;&]C=&5S="YC?5)=
M3\(P%'WOKSA",!TN,(@Q480'7GB ^>:3&C.W#IJ4C;0=&@G_W:[K8)&/9%FV
M>\ZYO>>>MGD6BR)A>%8Z$?RKMYJ0=K/&\[)$^ETDQ7HC!5]SK7HQNGW"?C23
M&;8Y3YH@]4:$D'@5272UC#*E^"^CY>M30WED1X *55IB#%K]>.M(B#RFP\#H
M`?.(/%NBU,VGAJ8PF6 0C)I :(&*<8K.:C1LH.;A*:C#;\8(/%,"U$;R3*?4
MS.2CU4EFTY;ONMAYF%#LH RO*<-:&9Y1SJ\IY[5R?E2>)5J>I5A#DNG"!&'0
M$<B>$)O(.N(9-2I$<AG[;N/F>_OVT<P@Y4PDIRD,JA0:G.%%4AVM(=P'CP]N
MR_]N!. \M-ZS5IUPDF-G3:22,6J/<:Z.A:&K*-P=^QO"^;GKK5P:N<+W^%YQ
MP4"K(\I(7EX7"P^WMW"G'FK'-=<&%I$R;HLX9DJEA8#M'&F>9T_H)* =Y1F/
M)B$?C?M?-MK_`5!+`P04````" `-E"(FZ/K:$LX!``#Y! ``#0```&1U;7!R
M;&EM:71S+F.-4VUKVS 0_F[P?SA<.NS5[?9U2S,(7@*F>2E)]F$O(7C2>16U
M)2/)I5WH?]_)3F!.J9O#6%AZ[KG'SYW.A&1%S1&NS9/Y8$6)5W=??.^LLZW1
MJ%JSXZ-:"F/Y,=YRH6C/\[T')3CPNJQT(4IA31CYW@[ ]P#:'6C>`[<AI 4J
M\W,#0]@MI^DL76^3VV\QN'" 3NP!DU7Z8QSW`+Z.UJ.XCV&U'B4W?0S)8CGN
M95BN5OTBY[?+1=)78KZ8I%-7Y#7 ;#R;+I*;Y\8H=I=I>$]6&:M;MR @H\"U
MCEP$@TQ);H+X!5L;P2Q[%&5=0BX*-.(OOHXLLT?@F<W@;9BQ&;L_`<>4QJ;R
M"5CZ1\&1YL*@/0$NZ_(W:E Y5%HQ\@=[3.@FJ IEZT=_1J'8/?)+(2]++)5^
M@HQSUPDP5<8P:!O4$^XX5QI"-^V2>O=Q0,LU?*+EXB)RQ[N6X@_:]HJ$[E+(
M30SOFL]H7Z+21)&'P;GY#*S6VMDTA""&_6#(S0$I<@B;U"M'N"4P#(?-9&W3
M^22=I^OOT4'V@;6630;RX,""A<%CU#FG>EWJ: !=?3$XWT@9G?QGP@M9#>ID
M6;_D&\((T)%&]$Z `S[['CV^]P]02P,$% ````@`$Y0B)O?#+[_H````*0(`
M``H```!/5510550N5%A4E9#):L0P#(;O>0H=6^A XG@<=Z"7]MJ!7OH`JJV
M&2_!"TS[]'6F#;-!EYOUH\_ZI*>75\C&$1@/B53P.FU E1C)9WB XJUQ)I.^
M`X?[TZ#9XMZXXF TEI+YH%.L:X=^X)UD?.&.23,'&C/")?7#L+E.&=7N"I*]
ME**5"_)='@`5(AWTKB#6\8%+)HY^YZ-BW4C/K8GR?S5]<6\4(8PPQ: H)3J[
M*%N+A:O/"R),Y+\.^@MB@]J17AF_<N1"? ?4NDHG2!.JO]HVSYCJ>D7-EF.Q
M@+;^B]D$OX$UZ\70BQ9N^/WV\;;Y!%!+`P04````" `3E"(F_KOZB:4````_
M`0``#@```'5L:6UI=&1U;7 N='AT;9#-#H,@$(3O/,4>VYN@^/,XB-N&"$)8
M;&J?OFJT)&WG.CO?3%;[B' S%H',"^'26Z]'N@*(HFNXY&Q020'A_?#'?DFX
M^3!/UCB3<&#_XKMX4;6RJ9E33W#H?%R^*)E!2>GQIV-5RSO!=)@A&;=ZA-I/
MPZ<A`[:.F3!"B%XC$=)Q(F3-@@GG0LD%9'[+?,!I?\!Y?VK+/4Q,L[+G^+Q,
M\$*65<G>4$L!`A0`% ````@`!Y0B)CE&]$!U`0``;P,```P``````````0`@
M`+:!`````&UA;&QO8W1E<W0N8U!+`0(4`!0````(``V4(B;H^MH2S@$``/D$
M```-``````````$`( "V@9\!``!D=6UP<FQI;6ET<RYC4$L!`A0`% ````@`
M$Y0B)O?#+[_H````*0(```H``````````0`@`+:!F ,``$]55%!55"Y46%10
M2P$"% `4````" `3E"(F_KOZB:4````_`0``#@`````````!`" `MH&H! ``
D=6QI;6ET9'5M<"YT>'102P4&``````0`! #I````>04`````
`
end
------------------------------
From: [EMAIL PROTECTED] (Craig Kelley)
Subject: Re: Registry for Linux - Why?
Date: 2 Jan 1999 09:11:04 -0700
In article <[EMAIL PROTECTED]>,
George MacDonald <[EMAIL PROTECTED]> wrote:
->For that matter why limit it to just Linux or Unix? Some applications span
->multiple operating systems. For example lets say I am using Netscape on
->a number of different platforms, wouldn't it be nice if I could just set
->some configurable parameters and have the apps look for configuration/app
->information on other machines?
Already done. I use Netscape from Windows and Linux machines and they
both look in my home drive (one via samba) for the TEXT FILES which
set it up.
Works like a charm.
I still haven't seen any benefits of a 'registry' that are not already
present with the current system.
--
The wheel is turning but the hamster is dead.
Craig Kelley -- [EMAIL PROTECTED]
http://www.isu.edu/~kellcrai finger [EMAIL PROTECTED] for PGP block
------------------------------
** 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: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.development.system) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Development-System Digest
******************************
