Letters to the Editor

Readers sound off.
Data is Our Life

This was our first buyer's guide and we made some mistakes, but we learned from them and plan to have an even better issue next time. One of the things we are most concerned about is data gathering methods. For this issue, other than the sunsite listings, we printed only what was sent to us. If you did not send in updated information, we would not have updated it for you. Red Hat obviously did send in updated information. If you did send in updated information, then I apologize for the table not getting updated. Actually, in either case, I apologize. Next time, we'll include a check for the latest distributions in our procedure. We do know what the current distributions are.

File Locking Services

Mr. Kraft's comments in the March 1997 issue of Linux Journal, regarding Linux's lack of network file locking services, are dead on the mark; however, I would now like to make it publicly known that there is an ongoing development effort to provide a lockd and statd for Linux.

This effort is currently combined with an effort, led by Olaf Kirch, to revise major portions of the Linux NFS implementation. A kernel-space lockd, written by Olaf, and a user-space statd, initially written by me and then significantly modified by Olaf, are currently part of Olaf's NFS development distribution “snapshots”.

A developers' mailing list exists for people who wish to contribute to, or participate in the alpha/beta testing of, this development effort. The list address is lockd-statd@linux.nrao.edu, and the list's subscription address is majordomo@linux.nrao.edu. Current snapshots of the linux-nfs development code can be retrieved from the following anonymous FTP directories:

ftp.mathematik.th-darmstadt.de:/pub/linux/okir/dontuse/ linux.nrao.edu:/pub/people/linux/okir/dontuse/

There are currently plans to publish an introduction to network file locking, together with a description of the Linux implementation, in an upcoming issue of Linux Journal. In addition, I will be giving a short presentation on this subject at the April 1997 Linux Expo in Raleigh, North Carolina. —Jeff Uphoff juphoff@nrao.edu

Security Issues

If you are going to do a security article, get it right. People get cgi and suid programs wrong on their own without your printing an article that contains serious errors. A good article on cgi security would have been just what is needed. Unfortunately this wasn't it.

Let's take this:


If I run this handy provided example by doing:

cp /bin/hash ./home
ln -s suidxi—program ./foo
export IFS
I get a shell as the person it is setuid to.

Why? Because the system runs the command through the shell, and the shell uses IFS as its “white space” definition.

This is basic setuid security stuff.

The procmail-based example at least does use a magic cookie to stop fake mails. It has other bugs; notably, it forgets sendmail may deliver multiple mails in parallel using data, but then I guess it makes it plain it's just trying to show the trick, not do it right. —Alan Cox alan@cymru.net

On-Line Linux Users Group

Hi. I have been a longtime reader of LJ and it has been a great help to me, and I am sure that applies to many in the Linux Community! Now, my friends on the Net and I have also done something as a contribution to Linux which I thought would be interesting to you and helpful to your readers. We have created an On-Line Linux Users Group for people interested in learning more about Linux, providing help to other Linuxers, and promoting Linux:

http://www.linuxware.com/ —Peter Lazecky peter@linuxware.com