Hello,
I use a rtlinux-3.1/kernel 2.2.19 based system. Some time ago,
I stated that from time to time, a newly started process dies randomly with
a segmentation fault. I first of all though about bad hardware
and about my own software being responsible. But I stated that if a process
runs for let's say a minute, it also runs very stable for a month...
After many hours of investigations, the problem looks like follows:
The target that shows the problem is a STPC Industrial based system.
The CPU core is a Cyrix 486. I tested also with a Celeron based system,
where the error does not occur. To test, I run five times concurrently
the following bash script:
while true; do cat /proc/meminfo > /dev/null; done
and the program 'top' with 1s interval. It does not depend on the cat,
it also fails with other programs. It can last a minute or even an hour
until cat or bash causes a segmentation fault.
As my kernel is patched, I though that this could be the problem.
I fetched a clean 2.2.19 kernel and the problem was gone. Then, I applied
the rtlinux-3.1 patch (I also tried rtlinux-3.0), and the segmentation
faults were back. I also tested with rtlinux-2.2, and with my patched kernel and
turned of rtlinux (I disabled CONFIG_RTLINUX and CONFIG_RTL in arch/i386/config.in)
but there is (or seems to be) no problem.
Well, I know that this sounds very very strange... I know that this is no proof that
rtlinux is responsible, it could also be a CPU core error, still another hardware
or linux kernel problem, or something else. But maybe someone has a hint for me
or can tell me how to go on, as I'm really stuck at the moment. I tried to understand
and check the rtlinux kernel patch but this is not so easy.
As it is a process start-up problem, I could imagine that it is related to the MMU...
Thanks,
Stephan
-- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
--
For more information on Real-Time Linux see:
http://www.rtlinux.org/
