Linux-Misc Digest #325, Volume #19 Sat, 6 Mar 99 03:13:08 EST
Contents:
Re: undefined symbol: __register_frame_info (Paul Kimoto)
Re: Microkernels are an abstraction inversion (steve mcadams)
Re: egcs 1.1.1 i386.rpm where ? (Micha� Kuratczyk)
Re: as86: Command not found (Micha� Kuratczyk)
Linux Install auf Laptop via NFS (Andreas Raum)
Re: Can Linux use 36-bit Xeon addressing? (Johan Kullstam)
Re: Can Linux use 36-bit Xeon addressing? (Johan Kullstam)
Re: OS with a seamless object model (Craig Kelley)
Red Hat 5.2 ugrade problem
Re: Question about ZIP Disks with Linux (Robert Heller)
Upgrading KDE1.1-0.1 to KDE1.1-3 (rhino)
Re: FreeBSD vs LINUX (Christopher Browne)
Re: Directory colours in RedHat 5.2 (kernel)
Re: Adjust time drift? (Bill Unruh)
Re: diald vs. pppd (Bill Unruh)
Re: Disks partitions ("Charles Sullivan")
Deleting inode from ext2fs ([EMAIL PROTECTED])
insmod: unresolved symbols ([EMAIL PROTECTED])
Re: Increase Num Procs (Roy Stogner)
Re: Exporting Windows filesystem for Linux... (Sparkzz)
whoops. fdisk and partition numbering woes. ([EMAIL PROTECTED])
Re: Newsreaders and Star Office (David Brown)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Paul Kimoto)
Subject: Re: undefined symbol: __register_frame_info
Date: 6 Mar 1999 00:52:03 -0500
Reply-To: [EMAIL PROTECTED]
In article <[EMAIL PROTECTED]>,
Daniel Kollar wrote:
> I get this error when using imlib applications.
> I.e. gqmpeg, windowmaker:setstyle, ...?
>
> Current configuration:
> imlib-1.9.4
> gtk+-1.2.0
> glib-1.2.0
> gcc-2.7.2.3
This is a symptom of moving binaries between systems with
differently-compiled libraries, e.g., taking a binary compiled
with egcs on a system with egcs-compiled glibc and trying to run
it on a system with gcc-compiled glibc. (Not easy to describe!)
(If this particular situation is yours -- although perhaps other
configuration combinations may cause the __register_frame_info
problem -- you need to ask the supplier of your binary executable
to compile with gcc, or to install an egcs-compiled libc on your
own system. Note that libc is crucial to the functioning of your
system, and replacing it is not for the faint-hearted.)
--
Paul Kimoto <[EMAIL PROTECTED]>
------------------------------
From: [EMAIL PROTECTED] (steve mcadams)
Subject: Re: Microkernels are an abstraction inversion
Date: Sat, 06 Mar 1999 05:55:28 GMT
Reply-To: [EMAIL PROTECTED]
[Posted & mailed, snipped, quoted is ">"]
[EMAIL PROTECTED] (Christopher Browne) wrote:
>OS/400 is pretty different. It looks like it is one of those fabled
>OSes actually written in C++.
Not surprising. Did you realize that IBM had a high-level
system-programming language before C was available, and used it to
write OS/360 in the early/mid '60s? As I heard it, Bell Labs (or
maybe it was ATT) came to IBM and asked to buy it, and IBM decided to
play keepsies and wouldn't release it as a product; C was the result.
At least that's the way I heard the story from the old-timers when I
worked at IBM. I know for a fact that I saw the source code for
OS/360 in about 1968 and much of it was written in that language; it's
still around inside IBM in it's Nth-evolved incarnation.
So when you talk about "fabled" OSes actually written in C++, and when
I remember what they were doing to the Nth-evolved incarnation of
their 30-year old systems programming language when I left the company
5 years ago, it's real hard not to ROTF.
____________________________________________________________________________
"Always enforce your assumptions." -steve, http://www.codetools.com/showcase
------------------------------
From: [EMAIL PROTECTED] (Micha� Kuratczyk)
Subject: Re: egcs 1.1.1 i386.rpm where ?
Reply-To: [EMAIL PROTECTED]
Date: Sat, 06 Mar 1999 05:56:22 GMT
Jacek M. Holeczek wrote:
>Is there somewhere a place where I can get the egcs 1.1.1 i386.rpm ( I
>can only find egcs-1.1b ) ?
ftp.ps.pl as far as I remember /pub/linux/PLD
--
Micha� Kuratczyk
------------------------------
From: [EMAIL PROTECTED] (Micha� Kuratczyk)
Subject: Re: as86: Command not found
Reply-To: [EMAIL PROTECTED]
Date: Sat, 06 Mar 1999 05:56:22 GMT
Michael Robbins wrote:
> as86: Command not found
>In which library can I find this command.
bin86
>I already have gcc-2.8.1 installed.
egcs is better to compile 2.2.x kernels.
--
Micha� Kuratczyk
------------------------------
From: Andreas Raum <[EMAIL PROTECTED]>
Subject: Linux Install auf Laptop via NFS
Date: Fri, 05 Mar 1999 18:05:12 +0000
Kann mir jemand erkl�ren wie ich SUSE linux auf einen Laptop
(Netzwerk ist eingerichtet) installieren kann.
Ich habe auf dem NT-Server via Omni NFS einen Server
eingerichtet und das CD-Rom freigegeben.
Wenn ich mit der Boot Disk Starte zeigt mir OmniNFS auch
dass der laptop das Share gelockt hat- ich komme
jedoch nicht ueber den Screen Linux installieren
nicht hinaus ( es erscheint ganz kurz eine gruen Message)
ich kann jedoch nix mounten etc. - Hat jemand eine idee
Thanx im voraus
AndY
------------------------------
Crossposted-To: comp.os.linux,comp.os.linux.hardware
Subject: Re: Can Linux use 36-bit Xeon addressing?
From: Johan Kullstam <[EMAIL PROTECTED]>
Date: 05 Mar 1999 19:18:47 -0500
John Burton <[EMAIL PROTECTED]> writes:
> Johan Kullstam wrote:
> >
> > John Burton <[EMAIL PROTECTED]> writes:
> >
> > > There are multiple reasons for and against going with an Alpha or PPC
> > > vs. Intel... on of which is *all* the other hardware is Intel x86 based
> > > and having *binary* compatibility is important.
> >
> > who cares about binary compatibility? just recompile!
>
> Ummm...try providing support to multiple scientists developing code on
> their personal workstations (x86 based), running multiple jobs on
> multiple machines, writing data to the same location (typically a
> partition on their personal workstation), using the generated (binary)
> data as input to other jobs on other machines...add to that the problems
> of maintaining some type of version control over the software being
> developed...so now add to that simply *recompiling* *all* the code that
> has been developed, and make sure that the version of code (and
> compilers, and libraries and utilities, and...) are the same on both the
> x86 *and* the alpha / ppc machines... And on top of all that, the data
> produce MUST BE CORRECT!
if you want *identical* results, most of the problem is the
bletcherous 80 bit float mode of the x86. are you using gcc/g77? you
are aware that they spill the 80 bit registers into 64 temporaries.
thus small changes in the code or fiddling with optimization switches
can cause spillage in different places and hence small changes in the
results.
intel should simply be avoided for any serious numerical work.
that said, your numerical algorithms should be designed to handle
variations in precision. if you have to bet that hard on the
precision, your algorithm is ill-conditioned, i.e., broken.
if you set your intel chips to do honest to god 64 bit ieee floating
point, it will agree with the alpha. and since both are little-endian
machines, binary data transfer isn't that bad either.
> So, while in concept having an alpha or ppc machine on the LAN with x86
> machines is good idea, in practice there are a lot of minor problems and
> it introduces a lot of places for bugs to enter the processing
> setup...
just using x86 all on its own introduces a lot of problems.
and if you use the 36 bit address space, the new code will hardly be
compatible with the other x86 stuff. therefore, you still lose.
> > if it's some proprietary offering, chances are it would never be
> > offered in 32+32 bit large mode. even if linux were to support it,
> > it'd be a bitch to port.
> >
> Like I said, I don't care about the 36 bit address space, just the
> bandwidth issue (described below) and getting the most bang for the
> buck...
i thought the whole damned point of this whole thread was the 36 bit
address space!!!! is the subject completely meaningless? what is it
you are *really* trying to say?
sorry for the rant but i am sorry intel arches really lose and there's
no fixing that.
--
J o h a n K u l l s t a m
[[EMAIL PROTECTED]]
Don't Fear the Penguin!
------------------------------
Crossposted-To: comp.os.linux,comp.os.linux.hardware
Subject: Re: Can Linux use 36-bit Xeon addressing?
From: Johan Kullstam <[EMAIL PROTECTED]>
Date: 05 Mar 1999 19:25:33 -0500
Robert Krawitz <[EMAIL PROTECTED]> writes:
> 2 GB RAM is a satisfactory virtual address space for a single process
> for most purposes, but 1 or 2 GB RAM is not a satisfactory upper limit
> on RAM today.
but these are not `most purposes'. the big ram user will almost
certainly need a shitload of ram for *one* process. otherwise, you'd
just buy more machines and run one of the big (but not humongous)
processes on each.
e.g., a finite-element analysis requiring 7GB of memory for the matrix
math. or a database application juggling a giant lump in memory.
--
J o h a n K u l l s t a m
[[EMAIL PROTECTED]]
Don't Fear the Penguin!
------------------------------
Crossposted-To: gnu.misc.discuss,comp.os.linux.advocacy
Subject: Re: OS with a seamless object model
From: Craig Kelley <[EMAIL PROTECTED]>
Date: 05 Mar 1999 23:17:59 -0700
[EMAIL PROTECTED] (Roger Espel Llima) writes:
> In article <[EMAIL PROTECTED]>,
> Craig Kelley <[EMAIL PROTECTED]> wrote:
>
> >Just because you rename a file to 'persistent object' doesn't change
> >the fact that a meaningful token must name it and that that token
> >cannot be used by others.
>
> That's a bit like saying that all variables must be global, because no
> other thing could possibly exist...
True, but even the best scoping rules can generate collisions (if
nothing else, in the compiling stage). If you are allowed to store
persistent data, then that must have a token so that you can retrieve
that data later on. If two things (for lack of a better term) decide
to use the same token then something must happen.
Under UNIX, the POSIX library has functions to deal with this issue
(man mktemp).
It all depends on what you decide to do with a collision. It is true
that UNIX gives the programmer a lot more control over this than other
operating systems may, but that is not always a bad thing.
--
The wheel is turning but the hamster is dead.
Craig Kelley -- [EMAIL PROTECTED]
http://www.isu.edu/~kellcrai finger [EMAIL PROTECTED] for PGP block
------------------------------
Date: Thu, 4 Mar 1999 20:36:20 -0500
From: <[EMAIL PROTECTED]>
Subject: Red Hat 5.2 ugrade problem
I just upgraded from Red Hat 5.1 to 5.2 but now
get the following error message when I try to
start a particular app;
error in loading shared libraries /usr/lib/libtcl8.0.so: undefined symbol: stat
The app worked fine before the upgrade. Any ideas
on what needs to be done?
--Greg
------------------------------
Subject: Re: Question about ZIP Disks with Linux
From: [EMAIL PROTECTED] (Robert Heller)
Date: 4 Mar 1999 20:36:49 -0500
Anatol Quabach <[EMAIL PROTECTED]>,
In a message on Wed, 03 Mar 1999 23:22:51 +0100, wrote :
AQ>
AQ> [EMAIL PROTECTED] wrote:
AQ> >
AQ> > Does anyone know if and when I finally get my ZIP disk drive to work
AQ> > with Linux, if I can then access my DOS formatted ZIP disks?
AQ> >
AQ> > Or will I have to format them with a Unix/Linux Partition thus making
AQ> > it non-cross platformable?
AQ>
AQ> No. Linux reads fat/vfat, no matter wether on a harddrive, a
AQ> floppy or a zip disk. When compiling your kernel, include
AQ> dos/vfat support or build is as a module.
And if you install the HFS module, Linux will just as happily read Mac
formatted CD-ROMS, floppies, harddrive partitions, or Zip/Jaz disks.
If you want *efficient* storage of data only meant for Linux use, you
*can* re-partition Zip disks and put a e2fs file system on them.
AQ>
AQ> --
AQ> Anatol Quabach <[EMAIL PROTECTED]>
AQ>
--
\/
Robert Heller ||InterNet: [EMAIL PROTECTED]
http://vis-www.cs.umass.edu/~heller || [EMAIL PROTECTED]
http://www.deepsoft.com /\FidoNet: 1:321/153
------------------------------
From: rhino <[EMAIL PROTECTED]>
Subject: Upgrading KDE1.1-0.1 to KDE1.1-3
Date: Sat, 6 Mar 1999 01:32:14 -0500
Hi all,
I'm a KDE 1.1-0.1 user and Linux newbie trying to install the latest KDE
(1.1-3) RPMS on top of my existing system. I downloaded all the 5 RPMS
kde-installer
kdesupport
kdelibs
kdebase
kdeapplications
And tried to start it off by doing rpm -Uvh kde-installer-1.1-3rh5x.i386.rpm
but it keeps giving me a series of errors:
/usr/doc/KDE-1.1/installation-guide/installation-guide-1.html conflicts with file from
kdesupport-1.1-0.1alpha1rh5x
/usr/doc/KDE-1.1/installation-guide/installation-guide-2.html conflicts with file from
kdesupport-1.1-0.1alpha1rh5x
/usr/doc/KDE-1.1/installation-guide/installation-guide-3.html conflicts with file from
kdesupport-1.1-0.1alpha1rh5x
/usr/doc/KDE-1.1/installation-guide/installation-guide-4.html conflicts with file from
kdesupport-1.1-0.1alpha1rh5x
/usr/doc/KDE-1.1/installation-guide/installation-guide-5.html conflicts with file from
kdesupport-1.1-0.1alpha1rh5x
and so on....
I tried removing the entire /usr/doc/KDE-1.1 directory but this error still
occurs. I installed my original KDE 1.1-0.1 from a lot of RPMs on the Red Hat
5.2 CD-ROM. Do I have to completely uninstall these before I start the new KDE
upgrade? How do I do that? Please advise.. thanks very much!
rhino
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Crossposted-To: comp.unix.bsd.freebsd.misc
Subject: Re: FreeBSD vs LINUX
Reply-To: [EMAIL PROTECTED]
Date: Fri, 05 Mar 1999 01:01:31 GMT
On 3 Mar 1999 15:07:39 -0800, david parsons <[EMAIL PROTECTED]> wrote:
> ____
> david parsons \bi/ If you're going to do this, lemme know: the FreeBSD
> \/ kernel is about as different from linux 2.0 as Linux
> 2.2 is.
And is the fact of the difference a bad thing?
--
"Unlike people, guns don't have Y2K problems..."
[EMAIL PROTECTED] <http://www.hex.net/~cbbrowne/lsf.html>
------------------------------
From: kernel <[EMAIL PROTECTED]>
Subject: Re: Directory colours in RedHat 5.2
Date: Thu, 04 Mar 1999 19:09:34 -0600
regoltd wrote:
>
> I have recently installed RedHat 5.2 and am new to Linux. I have previously
> tried Slackware and liked the way the directories and other files were in
> colours and the normal text type files were white. I prefer the RedHat to
> the Slackware because it is easier for me to use until I get use to Linux.
> Can I do this with RedHat 5.2 and how do I do this.
> M. Rego
there is a mini-howto about this particular issue...
--
...Saure really turns out to be an adept at the difficult art of
papryomancy,
the ability to prophesy through contemplating the way people roll
reefers -
the shape, the licking pattern, the wrinkles and folds or absence
thereof
in the paper. "You will soon be in love," sez Saure, "see, this line
here."
"It's long, isn't it? Does that mean --" "Length is usually intensity.
Not time."
-- Thomas Pynchon, _Gravity's Rainbow_
------------------------------
From: [EMAIL PROTECTED] (Bill Unruh)
Subject: Re: Adjust time drift?
Date: 5 Mar 1999 02:14:42 GMT
In <[EMAIL PROTECTED]> John Hasler <[EMAIL PROTECTED]> writes:
>Stef writes:
>> Thanks for your suggestion. But since my machines are connected to the
>> internet permanently, and also run permanently, I went for xntpd
>Chrony works fine for this. The main thing chrony lacks that xntp3 has is
>support for exotic hardware.
And what it has that xntp does not is control of the real time clock, so
that if you switch off your machine, when it comes on again it has a
good approximation of the real time, even if the RTC is out. Since your
RTC could be hours out, xntp would refuse to fix the wrong time, while
chrony would have predicted it and compensated.
------------------------------
From: [EMAIL PROTECTED] (Bill Unruh)
Subject: Re: diald vs. pppd
Date: 5 Mar 1999 02:16:24 GMT
In <[EMAIL PROTECTED]> Conway Yee <[EMAIL PROTECTED]>
writes:
>I was reading the documentation for pppd. It appears as if it is
>capable of bringing up a ppp connection on demand. Then the question
>becomes. What are the essential differences between diald and pppd?
Until very recently the kernels did not have support for demand
dialing.
------------------------------
From: "Charles Sullivan" <[EMAIL PROTECTED]>
Subject: Re: Disks partitions
Date: Thu, 4 Mar 1999 21:09:28 -0500
With fdisk as I recall, you just start creating logical partitions and the
extended partition will be created automatically. I assume cfdisk works
the same way.
[EMAIL PROTECTED] wrote in message
<7bm1on$9td$[EMAIL PROTECTED]>...
>Hi there!
>I have a 6 GB SCSI disk which I want to partition the following way:
>/ - 300 MB
>swap - 200 MB
>/usr - 1000 MB
>/var - 500 MB
>/var/www - 1500 MB
>/tmp - 500 MB
>/var/mail - 1000 MB
>/var/logs - 1000 MB
>
>The problem is that I can only create 3 primary partitions with cfdisk.
>There is an option that creates a logical partion, but how can I split
>a logical partion into several partions like we do in DOS?
>
>Many thanks.
>
>
>-----------== Posted via Deja News, The Discussion Network ==----------
>http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
------------------------------
From: [EMAIL PROTECTED]
Subject: Deleting inode from ext2fs
Date: Thu, 04 Mar 1999 22:04:46 GMT
Hello,
PLEASE EMAIL ME ANY RESPONSES - ITS HARD TO KEEP UP WITH NEWSGROUP VOLUMES.
Had a disk crash, corrupted part of the file system on /usr and now I'm left
with ugly inodes such as:
[root@marble lib]# pwd
/usr/lib
[root@marble lib]# ls -al findutils
p--S-wsrwT 1 7710 7710 0 Jan 5 1986 findutils
[root@marble lib]# lsattr findutils
lsattr 1.12, 9-Jul-98 for EXT2 FS 0.5b, 95/08/09
lsattr: Invalid argument While reading flags on findutils
All chmod's I've thrown at that barf so I am wondering if I can possibly
manually delete that somehow. Is there an ext2fs debugging tool of some sort?
I'm treading on somewhat uncharted territory here as far as editing an ext2fs
filesystem.
The one suggestion I've had so far - mke2fs :(
Any ideas? Can't use any of the packages that lost files in the corruption -
ie. findutils, perl.
This sucks.
Thanks for any help.
Kevin
============= Posted via Deja News, The Discussion Network ============
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: linux.redhat.misc
Subject: insmod: unresolved symbols
Date: Thu, 04 Mar 1999 22:48:37 GMT
Hi there,
I have RH5.2 running straight out of the box (i.e. I didn't tweak the kernel
or anything else). Now I am trying to recompile and load the scsi tape driver
st with DEBUG turned on.
Here is how I compiled st.c:
gcc -DMODULE -D__KERNEL__ -I/usr/src/linux/drivers/scsi -Wall -c st.c
It worked fine. But when I insmod st.o (even after depmod -a returned
nothing), I got the following errors:
st.o: unresolved symbol bad_user_access_length
st.o: unresolved symbol suser
st.o: unresolved symbol __constant_memcpy
st.o: unresolved symbol __constant_c_memset
st.o: unresolved symbol down
st.o: unresolved symbol __memcpy
st.o: unresolved symbol up
st.o: unresolved symbol __constant_c_and_count_memset
st.o: unresolved symbol __memset_generic
Something is wrong here. Browsing the usenet, I notice someone mentioned the
problem with incompatible versions. How can I find out whether I am running
into the problem or not?
Thanks in advance.
Huayong
============= Posted via Deja News, The Discussion Network ============
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
------------------------------
From: [EMAIL PROTECTED] (Roy Stogner)
Subject: Re: Increase Num Procs
Date: 6 Mar 1999 07:19:54 GMT
On Thu, 04 Mar 1999 08:33:08 -0800, Darren Ehmke wrote:
> Somehow, the first post did not seem to make it, if this is a
> duplicate please ignore.
>
>Question: How can I increase the number of processes that can run
>upon my linux system? I reached a max of 283, where I receive fork
>error if I add more. I need to run around 500 processes for a test
>on our network. The system is slakware 5.2 with 96 Megs of Ram.
>Thanks in advance.
You'll need to increase NR_TASKS in
/usr/src/linux/include/linux/tasks.h and recompile.
---
Roy Stogner
------------------------------
From: [EMAIL PROTECTED] (Sparkzz)
Subject: Re: Exporting Windows filesystem for Linux...
Date: 6 Mar 1999 07:25:52 GMT
You may be able to get by with smbfs, and Samba.
.
.
....Ken
------------------------------
From: [EMAIL PROTECTED]
Subject: whoops. fdisk and partition numbering woes.
Date: Thu, 04 Mar 1999 23:57:33 GMT
My tale of woe.
My computer was partitioned thusly:
1 OS/2 boot manager
2 DOS C
3 DOS D
4 Extended partition
5 NT
6 Linux swap
7 Linux
Late at night I removed the NT partition using
the Linux fdisk command and then added a partition
back using the same cylinder numbers (don't ask).
You may have guessed by now what happened next.
I did not realize that the partition numbers would
change in this fashion:
4 Extended partition
5 Linux swap
6 Linux
7 (was) NT
to make an ugly story short, I have to boot to
a floppy to access my linux partition, and mounting
it root still results in kernel panic.
My question is this
Can I change the fdisk numbering scheme to put things
back in order? Is this the time to back up and play
with a sector editor? advice? ridicule? easier way out?
README.fdisk explains A LOT but did not explain how to
recover from being a complete idiot.
Tune in next week when I rm / *
Any help would be like a soothing balm on the open wound
of my own stupidity.
------------------------------
From: David Brown <[EMAIL PROTECTED]>
Subject: Re: Newsreaders and Star Office
Date: Sat, 06 Mar 1999 07:33:53 GMT
Reply-To: [EMAIL PROTECTED]
> I've found the Star Office news reader to be very touchy.
This is becoming my experience with the whole suite - It's also the only
Linux program I've found slower than my windoze equivelent. I'll try WP.
> There's slrn. And you could always run a news server like
> leafnode or inn and connect any newsreader you want to it
> through the loopback interface.
Can you give me any pointers on how to get started with either slrn or
leafnode. I think I'm missing something basic - nothing happens when I try
ro run them (from KDE) - perhaps I have to edit some config files ? (I do
have other news and mail readers working fine.)
I don't really understand how to get suck working either.
And, where do I get slrnpull ?
Sorry for all the questions - I'm getting there, honest.
> Virtual Access ....Sorry, never heard of it... What's it do?
Off-line mail and news reader. Keeps both news AND mail messages correctly
threaded and handles the lot in one pass. At the moment this is the killer
app keeping me with windows !
David Brown
------------------------------
** 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.misc) 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-Misc Digest
******************************
