[olug] Linux mess
mpeterson at charles.omhcoxmail.com
Wed Dec 19 01:48:28 UTC 2001
Has anyone ever written a C program that messed up the 2.4 kernel so bad that it would not reboot and would not let you access the last file system that you were running the program on?
I have a disk I/O benchmark program that I have been working with over the years that I modified this weekend so that it displays the results to the screen and logs the results to a file and it performs a large number of runs so that it takes hours to run on a 233 MHz system.
I was running Redhat 7.2 this weekend with 128 MB RAM on a 233 MHz system with ext3 files systems and RAID1 also running.
In the program I have a character array that write the benchmark lines and I made an error in upgrading the program and was locked out of the home file system and could log into the system as root but it would not reboot with the reboot or ctrl-alt-Del command. It would not shutdown -h now either. I reset the system and if I ran the program for a short run I would get a Segmentation Fault but it would run to completion before the Fault occurred.
The compile caught part of the errors but there was enough problems that were in the final executable that the kernel was trashed so to speak.
When I was learning C I used to do these type of things to DOS 3.3 and had to power off the system and could not do a Ctrl-ALT-DEL.
Just wondering if anyone has had Linux appear to do the same thing with other applications.
I had heard about the Memory hang-up problem in the 2.4 kernel but this was not it.
I have the code of the working program and the bad one if anyone wants to do Disk I/O benchmarking or want to try to mess up a 2.5 kernel and see if it protects itself or can protect itself any better.
More information about the OLUG