newer saw that, the last 20 years

NOTA: cp is /usr/bin/cp (no alias) PATH is correct.
i own . .. and all files, perms are OK.
reboot && fsck change nothing.
history|od -c reveals nothing strange
any idea ?
i am on solaris7, ultra1, use (on cmd line) csh
tia

/4tg1/war1disk/neuforge/post-src/bin>!cp
cp ../qp ra
/4tg1/war1disk/neuforge/post-src/bin>Illegal Instruction - core dumped

/4tg1/war1disk/neuforge/post-src/bin>!cp
cp ../qp ra
/4tg1/war1disk/neuforge/post-src/bin>Segmentation Fault - core dumped

/4tg1/war1disk/neuforge/post-src/bin>file ../qp
../qp: ELF 32-bit MSB executable SPARC Version 1, dynamically linked, not stripped

/4tg1/war1disk/neuforge/post-src/bin>ls -l ../qp
-rwxrwxr-x 1 otto otto 9008 Sep 15 16:55 ../qp

succesive cp are ok

Do a
file core
to be sure you know where the core file came from. Is this actually repeatable? If so, do a
truss !c
instead of just
!c
If that dumps core, post the last 20 lines of so of the truss.

I'd say one of three things:

1. Your machine has been owned and something like libc.so.1 is messed up.
2. You've got a bad memory SIMM. Check /var/adm/messages.
3. You've got a bad disk block. Check /var/adm/messages.

I'd vote for a bad or loose SIMM first. When you reboot, enable a full memory test at the boot interpreter (STOP-A right after the machine starts booting or do an init 0 from a running system).

i cannot reproduce it, it happens today first time
file core give me totally the wrong result, sure not 'cp'
i will try truss (every time i issue a cp )
/var/adm/* && dmesg are clean, don't really believe an HW-probl
thanks 4 help, i'll investigate