Hi,
i don't know, if my problem is SMP related, but it's definitly strange
and related to kernel 2.2.* (now i run 2.2.10, but the same happens
with 2.2.5, 2.0.37 works)
I run linux on a Pentium Pro system. binutils 2.9.1.0.25
(2.9.4.0.5 doesn't work either), and egcs-19990629 (a prerelease of gcc-2.95).
My problem is the following: if i run in tcsh:
while 1
g77 -o t t.f ; ./t
end
with t.f:
WRITE(*,*) 'HALLO FORTRAN'
END
in my home directory (a nfs-mounted dir from a Solaris system)
i get an irregular queue of HALLO FORTRAN and Segmentation Faults (from
running t not from the compiler!).
If i do it locally on my disk, it seems to work all the time (but i think
i had Segfaults here, too, but they are now not reproducible).
The corresponding binaries differ. I append two binaries (which actually
run both) as gzipped files. By xxd i found out, that they differ by a
intermediate 0 Byte. I don't know, where this comes from...
An executable, which segfault, segfaults all the time. So it's not
linux' executable loader, which produces a problem.
if i run the command g77 -o t t.f and g77 -o t1 t.f, then t and t1
differ. If i do the same locally, they are identically.
Do i have a SMP problem? (SMP on 2.0.37 works flawlessly).
Or a problem with nfs? Or a hardware problem? or a Solaris-nfs problem?
or an binutils problem? or a egcs problem? or should i look for a psychatrist?
Please tell me, what i could do to find out!
Thanks in advance,
Martin.
PS: Does anybody else see such things?
PPS: My machine runs 24h/day since about 3-4years whith very few reboots
every several months.
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
