----- Original Message -----
From: Harald Hanche-Olsen <[EMAIL PROTECTED]>
> | Do you mean I should trace qmail-queue? If so, how do I do that when
> | qmail-queue is invoked from qmail-inject?
>
> You seem to be on a Solaris system. Then use truss -f to trace the
> children. (You can also use -o filename to redirect the trace
> output.) Since qmail-queue is a setuid program you have to be root to
> do this, but then I believe it will work.
Thanks for the advice, Harald! Ok, here's what I get now using -f flag:
(first what seems to be "normal" trace output, and then around line 100)
....
2761: close(6) = 0
2761: fcntl(3, F_GETFL, 0x00000000) = 2
2761: ioctl(3, I_FIND, "sockmod") = 0
2761: close(0) = 0
2761: fcntl(3, F_DUPFD, 0x00000000) = 0
2761: close(3) = 0
2761: fcntl(5, F_GETFL, 0x00000000) = 2
2761: ioctl(5, I_FIND, "sockmod") = 0
2761: close(1) = 0
2761: fcntl(5, F_DUPFD, 0x00000001) = 1
2761: close(5) = 0
2761: execve("bin/qmail-queue", 0x0002F1DC, 0xEFFFF91C) argc = 1
2761: *** SUID: ruid/euid/suid = 0 / 1129 / 1129 ***
2761: open("/dev/zero", O_RDONLY) Err#13 EACCES
2761: open("/dev/zero", O_RDONLY) Err#13 EACCES
...
(*thousands* of these EACCES error messages, and then the tail)
...
2761: open("/dev/zero", O_RDONLY) Err#13 EACCES
2761: Incurred fault #6, FLTBOUNDS %pc = 0xEF7E6D8C
2761: siginfo: SIGSEGV SEGV_MAPERR addr=0xEF7FFFE8
2761: Received signal #11, SIGSEGV [default]
2761: siginfo: SIGSEGV SEGV_MAPERR addr=0xEF7FFFE8
2760: Received signal #18, SIGCLD, in wait() [default]
2760: siginfo: SIGCLD CLD_KILLED pid=2761 status=0x000B
2760: wait() = 2761 [0x000B]
qmail-inject: fatal: qmail-queue was killed
2760: write(2, " q m a i l - i n j e c t".., 44) = 44
2760: _exit(111)
2761: *** process killed ***