Random IDE Filesystem Corruption W/ New Triton MB

Random IDE Filesystem Corruption W/ New Triton MB

Post by Lou Sortm » Thu, 28 Dec 1995 04:00:00

I decided to add an IDE drive that I have to my disk farm and the files (and
filesystem level stuff) on the disk become corrupt over time; seemingly at
random.  This is consistent with the problem experienced by many people with
certain PCI IDE controllers, but I have been lead to believe that the EIDE
controllers on Triton based motherboards are immune.  I have used this same
drive before with a 386 (and a paddleboard controller) and did not have this

My kernel is 1.3.35.  I haven't done anything unusual to it.  I did configure
it to use the "new" IDE driver, if that is important.

My drive is a Conner 170MB IDE, which is a few years old, but has been idle for
a non-trivial portion of that time.  It passes surface tests (in DOS) just fine
(with the same controller).

My Motherboard is a STEELBLUE Triton board (with a P100) which I bought from

I am not having random signal 11 problems, even when doing a kernel compile.
Of course, when I run a program which has been corrupted, I get a signal 11
every time.

Any idas?



1. Random filesystem corruption - dual boot only

I currently have a dual boot system with Win95 as the first partition and the
Linux / partition second.

Whenever the Win95 partition fills up the next time I boot up Linux the /
partition is badly damaged.  I suspect that Win95 expects to see partition
boundries on certain multiples of cylinders and perhaps it is walking off the
end of /dev/hda1 when it fills up.

This is pure speculation and a fix (apart from the obvious) could be to have a
buffer partition os say 100Meg doing nothing as /dev/hda2.

I had the same problem when using an adaptec scsi card and have no problem with
the same distribution (Suse 5.3) on other Linux only boxes.

Has anyone else observed this and or is there a more elegent fix than my idea
above?. Am I spouting total rubbish (possible) or is Win95 stupid enough to
ignore partition boundries?

Mark Baker

Remove REMOVE to reply by mail.
"One World, One Web, One Program" - Microsoft slogan
"Ein Volk, Ein Reich, Ein Fuehrer" - Adolf Hitler

2. Paradox Front End, Linux based SQL backend?

3. HIMEM.SYS reports "unreliable memory at 0817c70h" with new Triton MB

4. kde b4.1 error : can't resolve symbol '__fxstat'

5. triton IDE busmastering on Data Expert triton based motherboard

6. Looking for web systems

7. New Motherboard Problem Triton II PCI IDE not probed correctly.

8. Help creating boot disk for RedHat4.0

9. Random thoughts re: RH 6.0 (Kernel 2.2.x), IDE, filesystem failures...

10. New triton PCI/IDE motherboard controller

11. [2.5.63] New IDE changes cause data corruption on PIIX4 AND additional problems

12. Q: which of Intel Zappa/Intel Triton mb?

13. Triton MB and invisible CD