• Subscribe to Tech tips from Linux Journal's Weekly News Notes

• bin/cue - iso files
• Segmentation Faults -- At odd times.
• editing shell scripts
• 2.2.20 kernel?
• Linux on Sun Sparc??
• Routing Mail
• Command to read CMOS from running Linux system
• linux ftp problem
• Self extracting shell script
• Cannot Format Network Drive
• Wu-FTP and Linux Newbie A source of confusion.
• Root Password
• RH7.1 switch to KDE login as default
• 3d linux

bin/cue - iso files

Question From: "M.Mahtaney" <mahtaney from pobox.com>

hi james

i came across your website while looking for some help on file compression/conversion

i downloaded an archive in winrar format - which then decrompressed to bin/cue files

i'm not sure what to do with that now - i've converted it to an iso file, but i need to basically open it up to gain access to the application that was downloaded.

i'm a little lost on how to do that - can you help?

thanks, mgm.

Hi MGM,

Have you tried to mount the iso image and copy the files out? There maybe another way but this worked for me.

mount -o loop -t is09660 file.iso /mnt
ls /mnt

should then give a list of the files in the iso image.

Hope this helps.

Kind regards Andrew Higgs

Segmentation Faults -- At odd times.

Question From: "Jon Coulter" <jon from alverno.org>

I've recently upgraded the ram on my linux box in an attempt to wean some of the memory problems that have been occurring

What were the previous problems. Just running out of memory? Or other things.

, and now get "Segmentation Fault"'s on almost every program about 80% of the time (if I keep trying and keep trying to run the program, it eventually doesn't Segmentation Fault for one time... sometimes then segmentation faulting later in its execution). My question is: Is there anything I can do about this?!? But seriously, is there any chance that a kernel upgrade would help fix this (my current kernel is 2.4.2)? I just don't understand why a memory upgrade would cause problems like this. I've run memtest86 test on the ram to ensure that it wasn't corrupted at all.

Frequent segfaults at random times usually indicate a hardware problem, as you suspected. A buggy kernel or libraries can throw segfaults left and right too, but if you didn't install any software at the time the faults started occurring, that rules that out.

Does /etc/lilo.conf have a "MEM=128M" number or something like that in it? Some BIOSes require this attribute to be included, other's don't. But if it's specified, it must be correct. Especially, it must not be larger than the amount of RAM you have: otherwise it will segfault all over the place and you'll be lucky to even get a login prompt. Remember to run lilo after changing lilo.conf.

Try taking out the memory and reseating it. Also check the disks: are the cables all tight and the partitions in order? Since memory gets swapped to disk if swap is enabled, it's possible for disk problems to masquerade as memory problems.

If you put back your old memory, does the problem go away? Or if you're mixing old and new memory, what happens if you remove the old memory? Are the two types of memory the same speed? Hopefully the faster memory will cycle down to the slower memory, but maybe the slower memory isn't catching up. What happens if you move each memory block to a different socket?

I'm running 2.4.0. There are always certain evil kernels that must be avoided, but I don't remember any such warnings for the 2.4 kernels. Whenever I upgrade, I ask around about the latest two stable kernels, and use whichever one people say they've had better experiences with.

Good luck. Random segfaults can be a very irritating thing.

More From: Ben Okopnik (The Answer Gang)

Take a look at the Sig 11 FAQ: <http://www.bitwizard.nl/sig11/>;.

2c tip : editing shell scripts

For shell scripts it's a common error to accidentally include a space after the line-continuation character. This is interpreted as a literal space with no continuation; probably not what you want. Since spaces are invisible, it can also be a hard error to spot. If you use emacs, you can flag such shell script errors using fontification. Add this code to your ~/.emacs file:

;;; This is a neat trick that makes bad shell script continuation
;;; marks, e.g. \ with trailing spaces, glow bright red:
(if (eq window-system 'x)
  (progn
    (set-face-background 'font-lock-warning-face "red")
    (set-face-foreground 'font-lock-warning-face "white")
    (font-lock-add-keywords 'sh-mode
      '(("\\\\[ \t]+$" . font-lock-warning-face)))
  )
)

Matt Willis

More From: Dan Wilder (The Answer Gang)

And in vi,

:set list

reveals all trailing spaces, as well as any other normally non-printing characters, such as those troublesome leading tabs required in some lines of makefiles, and any mixture of tab-indented lines with space-indented lines you might have introduced into Python scripts.

:set nolist

nullifies the "list" setting.

More From: Ben Okopnik (The Answer Gang)

Good tip, Matt! I'll add a bit to that: if you're using "vi", enter

:set nu list

in command mode to number all lines (useful when errors are reported), show all tabs as "^I", and end-of-lines as "$". Extra spaces become obvious.

In editors such as "mcedit", where selected lines are highlighted, start at the top of the document, begin the selection, and arrow (or page) down. The highlight will show any extra spaces at line-ends.

2.2.20 kernel?

Question From: Ben Okopnik

Where can I get one? I've looked at kernel.org, and they only go up to .19...

Ben

Glad you asked, Ben, now I have an excuse to cough up a 2c Tip.

Cool. Thanks!

The two places to look for "the rest of it" if you really want a bleeding edge kernel are:

For Linus' tree -

http://www.kernel.org/pub/linux/kernel/testing

Nope; just 2.4.x stuff.

For Alan's tree -

http://www.kernel.org/pub/linux/kernel/people/alan

Alan keeps subdirectories in there which should help you figure things out. In the moment I type this, his 2.4.6 patch is at ac5 and his 2.2.20pre is at 7.

As of press time, 2.4.7 patch is at ac3 and 2.2.20pre is at 8. -- Heather

These are of course patches that you apply to the standard source after you unpack the tarball.

For a last tidbit, the web page at http://www.kernel.org keeps track of the current Linus tree, listing both the released version number and the latest file in the /testing area.

Linux on Sun Sparc??

Question From: Danie Robberts <DanieR from PQAfrica.co.za>

Is there a way to install this. I can see the Sparc64 architecture under /usr/src/linux/arch

Booting from the slackware 8 cd does not work, and it seems as if Slackware's Web site is "unavailable"

Please, any pointers!

Cheers Danie

Sure.

The Linux Weekly News just wrote that Slackware is no longer supporting their Sparc distro. There's a sourceforge project to take over the code: http://sourceforge.net/projects/splack

Entirely possible that some decent community interest would boost it nicely, but if so, why'd Slackware have to let go of it... dunno...

Maybe you want to try another distro that builds for Sparc - RH's sparc build isn't well regarded amongst a few non-Intel types I know locally... I think in part because they get to Sparcs last. But you've lots of choices anyway. The search engines (keywords: sparc distribution) reveal that Mandrake and SuSE both have Sparc distros, Rock Linux (a distro that builds from sources - http://www.rocklinux.org) builds successfully on it, and of course there's my personal favorite, Debian.

The Sparc-HOWTO adds Caldera OpenLinux and TurboLinux to the mix. I know that Caldera is really pushing to support the enterprise but that points to openlinux.org and appears to be an unhappy weblink - I only did a spot check, but didn't see "Sparc" on Caldera's own pages. One may safely presume that someone who wants Caldera's style of enterprise level support for Linux will be getting Sun maintenance contracts for their high end Sparcs, right?

Anyways the reasons that Debian is my personal favorite in this regard are:

1. It's the only distro that I've tried on a Sparc
2. It worked just as smoothly as on a PC
3. apt-get install <package name here>

But you don't have to take my word for it There's a bunch of folks at http://www.ultralinux.org

who pay attention to how well things work on ultrasparcs, and they've helpfully gathered a bunch of pointers into the arcvhives and subscribe methods for the lists maintained by the major vendors in this regard. So check out what people who really use Sparcs have been saying about 'em! Using your model number as a search key may narrow things down, too.

Routing Mail

Question From: Danie Robberts (DanieR from PQAfrica.co.za)

Hi,

I am trying to set my environment up so that I can use Star Office on my Slackware 8 Laptop, and send mail via our Exchange Servers. The problem is that the IIS Department has only allowed my Suna Workstation to Send mail to their List server. I think it is limited by IP @.

Is there a way to configure a type of mail routing agent, so that my sending adress in S/O will be the IP@ of my Sun Workstation, and the receiving will be the exchange server (At the moment I can only receive e-mail)

Here is the IP Setup:

 List Server:	196.10.24.112
 Exchange 	196.10.24.33
 Laptop:		192.168.102.241
 Sun:		192.168.102.44

Thanx
Danie

Not directly, while your Sun workstation is online. You'd have to give your Linux workstation the Sun's IP number. Two hosts with the same IP on the same network is a recipe for trouble!

If you've full control of your Sun, you could arrange to use it as your email relay host, for an indirect solution.

It looks like StarOffice can be configured to use two different mail servers, one for outgoing, one for incoming. Set the outgoing server to the Sun, and the incoming to the exchange server. Configure the MTA on the Sun (sendmail, I expect) to relay for you, and you're all set.

To test the Sun,

telnet sun 25
helo linux.hostname
mail from: <you@your.domain>
rcpt to: <somebody@reasonable.domain>
data
some data
.
quit

"linux.hostname" is the hostname of your linux system "you@your.domain" is your return email address "somebody@reasonable.domain" is some reasonable recipient

If you don't get a refusal, and the email goes through, the Sun is set up to relay.

Command to read CMOS from running Linux system

Question From: Paul Kellogg (pkellogg from avaya.com)

To "The Answer Gang":

I am looking for a program I can run on my Linux machine that will display the CMOS settings and CPU information. I would like to avoid rebooting to display the information. Have you heard of a program that will work for this? And if so, do you know where I can get it? If not, do you have some suggestions for where to start if I wanted to write such a program? I am working with a RedHat 6.2 system on a intel platform.

Thanks - Paul.

Hi Paul,

Try `cat /proc/cpuinfo` for information regarding the CPU. I am not to sure about the CMOS info. It depends what you are looking for?

Look at the files in /proc and see if you can find what you are looking for.

I hope this will help in some small way.

Kind regards

linux ftp problem

Question From: Brad Webster (webster_brad from hotmail.com)

ok heres the deal, im having a heck of a time with the ftp client on my linux server. im running red hat 6 and i can connect fine, but then it will not accept the username and password for any of the users. the worst part is the same username and passwords will connect through telnet? any suggestions would be very appriciated

bard webster

Hello,

First off, I shall begin by saying, please in future send your e-mail in plain-text format and NOT in HTML. Poor Heather will have a hell of a time trying to extract the important information from the e-mail

It's easier than quoted printable, but, yeah, it's a pain. -- Heather

So, to your problem. There are a number of things that you can check. Firstly, when using FTP, do you make use of the hidden files ".netrc", in each of your users home directory? In that file, you can store the ftp machine, and the username and password of the user. Typically, it would look like the following:

machine ftp.server.sch.uk   login   usernametom  password xxxxxxx

The reason why your users can log in via telnet, is that telnet uses different protocols. I would also suggest using a program such as "ssh", which is more secure than telnet, but I digress.

Regards,

Thomas Adam

Self extracting shell script

Question From: Albert J. Evans (evans.albert from mayo.edu)

Hi,

I'm looking through some older Linux Gazzette articles and noticed "fuzzybear" had put out a location to download his "self extracting w/make" shell script. He was responding to your inquiry about this script. The URL he listed is now defunct. Did you by chance get a copy of it, or know his current URL?

You know, I searched the past issues of LG... and I must admit that I couldn't find my own article mentioning that. I'm pretty sure that the URL I gave is the same as it is now - I don't rememberr changing it at all - and I've just tested it (with Lynx.) It's still good. Anyway, here it is -

http://www.geocities.com/ben-fuzzybear/sfx-0.9.4.tgz

The reason I haven't been particularly hyped about this thing is that, the more time passes, the less I think of it as a good idea. Given that one of the classes I teach for a living is Unix (Solaris) security, it's actually my job to believe that it's a bad idea. Just like many other things, it's perfectly fine as long as you trust the source of the data - but things rapidly skid downhill as soon as that comes into question. Sure, in the Wind*ws world, people blithely exchange "Setup.exe" files; they also suffer from innumerable viruses, etc.

Think well before you use this tool. Obviously - and if it isn't obvious, even downloading it might be a bad idea - never-ever-*ever* run a self-extracting archive, or any executable that is not 100% trustworthy, as root.

...at this point Ben drew a huge ASCII-art skull and crossbones...

You have been warned.

Ben Okopnik

Cannot Format Network Drive

I'm not quite sure if you can help me out with this situation. Okay here's the lay down I have 1 IDE Hardrive 35 gigs About a month ago I added the program System Comander so I could put Windows 98, Windows 2000, and Linux on my system. I did it successfully! So now about a week ago I wanted everything off, cause I bought a serperate PC for Linux. This is how I wipe everything off of my drive. I used a Debug to wipe everything off and used fdisk to try to get rid off all of the partititions... but it still didn't get all of the partitions off (looking in fdisk). How I got rid of all of them was doing the commands

lock c:
fdisk /mbr

Than going back to fdisk and deleted the remaining partitions. This did work... no longer is there any drives stated in fdisk. So than I created 2 drive c: and d: both about 15 gigs or more But now this is were I run into my problem.... After rebooting I tried to format those drives a it's giving the error message CANNOT FORMAT A NETWORK DRIVE???? I know this is a lot more information than you prbably needed, but I thought I'd better say everything so that you know the whole story... Can you help me? Do you know of anything I could try?

Well, since you've removed Linux, your options are pretty thin. It sounds like your 'fdisk' screwed up somewhere, or you have a virus (I remember, a looong time ago, of some that did that.) One of the easiest ways to handle it would be to download Tom's rootboot <http://www.toms.net/rb/>;, run it, and use the Linux fdisk that comes with it (you should read the 'fdisk' man page unless you're very familiar with it.) It can rewrite your partitions and assign the correct types to them; after that, you should be able to boot to DOS and do whatever you need.

Ben Okopnik

Wu-FTP and Linux Newbie

Question From: seboulva (seboulva from gmx.de)

Sorry about this Simple Question, but I am a Linux Newbie and I want, no I must Update our Wu- FTP Server.It Comes with Red Hat 6.3

I Downloaded the 3 *.patch Files . And want to Apply it with the Patch Command.

Do you have the source code for wu-ftp and did you compile it yourself last time? patch files are usually for source code, and they will change a specific version to a specific other version. Are you sure that the patch files math your source code version? If it's RedHats patch files for their 6.3 distribution that should match. But most probably it matches only the source code from their 6.3 CD's.

If all this matches, you should be able to compile the basic wu-tfp as it comes with 6.3. Then you can try to patch the source and recompile.

Depending on the patch files (especially with which path they are generated) you would do something like:

cd wu-ftp-source
patch -p 1 < ./path/to/patchfile

the -p x optioon strips off x directory levels from the path in the patch file. If you look in the ascii patch-files you will see file names with path -- they have to match where your source is.

'patch --help' says among other very usful thing:

  --dry-run  Do not actually change any files; just print what would happen.

so this is the option for testing until it looks as if it will apply cleanly. Then remove the dry-run and do it for real.

recompile wu-ftp and then install it in the system. You will have to restart it of course.

K.-H.

Root Password

Question From: Adam Wilhite (a.m.wilhite from larc.nasa.gov)

Somehow the root password was lost or doesn't work anymore for one of the computers I administer. I went through your steps to reboot with init=/bin/sh... My problem is when I try to mount /usr. It says it can't find /usr. I would really appreciate your help.

thanks,
adam wilhite

Hi,

Your error message about not being able to find /usr, suggests to me that "/usr" is not your mount point. Take a look in the file "/etc/fstab", which is the filesystem table.

Locate the entry which points to "/usr". Your mount points are usually in the second field, i.e. after all the "/dev/hda1" entries.

When you have found it, type in the correct value, using:

mount /path/to/mountpoint

Failing that, try issuing the command:

mount -a

that will tell your computer, to (try and) mount all the entries within the file "/etc/fstab". You can then check what has been mounted, by typing in the following:

mount

HTH,

Thomas Adam

RH7.1 switch to KDE login as default

Questions From: Larry Sanders (lsanders from hsa-env.com) and Jim (anonymous)

Having completed the installation of Red Hat 7.1 with both GNOME and KDE, the default graphical login is GNOME. How is this changed to the KDE default login for the system? -- Larry

I love and use both the KDE and Gnome desktops. I have a strong aesthetic preference, though, for the KDE login manager.

My RedHat 7.1 install has resisted all efforts to switch it from its default Gnome login manager to the KDE one. This isn't a big deal but where is this managed and how do I make the change? -- Best Regards, Jim

Try running the program "switchdesk". It might be on the GNOME menu, but you can run it from a command-line quicker. It will let you switch default desktops and even different desktops for different displays (although that tends to be a bit problematic).

-- Regards,
Faber Fedor

3d linux

Hello!

This week I found some interesting programs that can help (a little bit) to Philippe CABAL.

K-3D

http://freshmeat.net/redir/k-3d/4976/url_tgz/k3d-0.1.29.0-src.tar.gz

It is frontend(Opensource, GPL) to render engines that are available to linux and win:

BMRT,

RDC

(ftp://ftp.dotcsw.com/rdc32lnx.tar.gz ->demo)

and 3Delight

(http://www.3delight.com - I didn't try it)

BTW: K-3D is one of the best applications I have seen on linux. Also no other application need so much time and memory resources for compiling like K-3D

http://www.dotcsw.com have perfect page with links. There is lof of rendering staff (also many that are just expected)

I hope this help.

Zdeno

This page edited and maintained by the Editors of Linux Gazette Copyright © 2001
Published in issue 69 of Linux Gazette August 2001

HTML script maintained by Heather Stern of Starshine Technical Services, http://www.starshine.org/