On Fri, 2003-07-11 at 15:49, Shlomi Fish wrote: > Here's the output from the program on my machine: > > Father > Sweet child in time: 8826 > you'll see the light > > It seems the exec is happening. I'm using kernel 2.4.21-0.18mdk with > glibc-2.3.1-10mdk on a Mandrake 9.1 system running on a P3 Machine.
This is very troubling. The exact same program gets stuck on my machine right after "Sweet child in time", and I get no further printings. This means that neither the execlp() nor the perror() after it were called. The process table looks pretty weird, too: alexsh 12056 0.2 0.1 3624 396 pts/0 S 10:15 0:00 ./thfork alexsh 12057 0.0 0.0 0 0 pts/0 Z 10:15 0:00 [thfork] <defunct> alexsh 12059 0.0 0.1 3624 424 pts/0 S 10:15 0:00 ./thfork It seems as if the forked process is a zombie, but the thread that was created has a higher PID than the process? Even though it was created before? Or can the thread be a zombie somehow? And there's a gap of one number between the PID of the zombie and the next PID of the program. (That's consistent between runs.) This means that something was started there and exited. What could that be? I tried to put a sleep 1000 instead of the echo to see if that's what was running, but the process table remains the same. Trying to attach with gdb to either one of the live processes yields the following error: Can't attach LWP 12027: Operation not permitted I'm really stumped here. Does anybody have any idea what's going on? By the way, I'm running Debian unstable, kernel 2.4.21 (vanilla, no patches), glibc 2.3.1, gcc 3.3 and 2.95 (tried both). I've also tried this on a machine with kernel 2.4.20. -- Alex Shnitman <[EMAIL PROTECTED]> http://www.hectic.net/ UIN 188956 PGP 0xEC5D619D / E1 F2 7B 6C A0 31 80 28 63 B8 02 BA 65 C7 8B BA ================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
