RM FILE.TXT -> segmentation fault -> Major Crash

[sigurður Einarsson]

Hi all, here is some reading about an interesting error which leads to major crash. So if you are up to it, please read on....

Status/Symptoms before crash:

SYMPTOM 1. When I try to delete some files with ???rm file.txt??? it results to a segmentation fault. I chmod the file to writable, try to delete again and succeed.

SYMPTOM 2. I have made this little program ???LittleProg.c??? which allocates, deallocates and displays a shared memory segment. After I delete the executable ???LittleProg???, the object file ???LittleProg.o??? and compile the program it works fine. Then I run ???ls ???l??? and after that, the little program gives me ???Illegal instruction??? message. ???LittleProg??? will not work again until I repeat the deleting of the two files and re-compilation of the program.

In hope that reboot will help, I reeboot but then the following problems occur in the same order as they appear here.

1. Mounting Proc filesystems failed
2. Cannot open font file latarcyrheb-sun16
3. Mount USB filesystem failed
4. Could???nt open /proc/partitions: no such file or directory. Is /Proc mounted?
5. Press to start normally: -> ( I press ) ->
Unmounting filesystems
/etc/rc.d/rc.sysinit: line 269: 135 segmentation fault mount ???n ???o remount, ro /
-> ( the computer reboots itself with the same error messages. )

Before I rebooted I checked if ???/proc??? and its whole contents existed and it did. I checked because this has happened 2 times before.

At this point, I must install another hard disk drive in the computer, install a fresh OS(Red Hat 8.0) on it, mount the crashed disk to it and copy the data to the new one.

This process is very tireing. So if anyone has a theory or a suggestion for me it would be highly appreciated.

Keywords for what is being developed on the computer are: static library, dynamic library, JNI, RMI, shared memory and semaphores.

With best regards, Siggi.

[Manuel Wolfshant]

1. check the hardware. especially memory, coolers and hard disk (including hard disk temperature)
2. verify your hard disk. I mean, smartctl and all related stuff
3. verify that the machine has not been tampered (like in rootkit)
4. verify that the kernel image + modules (and other packages) you are using are all OK and updated to latest errata.
5. skip RH 8 if you can. either stick to 7.3 or switch to 9.

[Michael Steele_2]

That you can install another O/S and reboot at this point, confirms an O/S problem and not HW.

Somewhere you are writting to the O/S disk and corrupting it. But disk and physical memory are not physcially linked and segmentation fault is an indication of an illegal access to a segment of memory not mapped for your access, probably system memory.

So the two should be unrelated.

If you're not able to reboot because /proc is unmountable then ramdisk or /dev or the kernel have become corrupt.
If you're able to reload the old kernel then this would confirm corruption of the kernel.

It confuses me that you have to use a new disk. You can't fdisk and reformat?

[Michael Steele_2]

Check for BIOS with the motherboard manufacturer. Its the bios that maps out the disk geometry, but I can't see a relation to your problem unless an upgrade is needed. Your motherboard has been stable all along, right?

[sigurður Einarsson]

The crashed HD is not crashed in the sense that nothing can be retrieved from it. I can mount it to another HD and copy my data, and that is what I do.
Afterwards I can reformat the crashed disk and do whatever.

Brg, Siggi

[sigurður Einarsson]

The crashed HD is not crashed in the sense that nothing can be retrieved from it. I can mount it to another HD and copy my data, and that is what I do.
Afterwards I can reformat the crashed disk and do whatever.

Brg, Siggi