On Sun, 10 Aug 2003, Vadim Vygonets wrote:
> It happens with Exim 4.20 and 4.12, but not with 3.36 from the
> OpenBSD ports collection. OpenBSD 3.3 on sparc64, default Exim
> configuration file.
Hello Vadim,
As I don't have access to such a system, I can't do any direct debugging
of this. You may have to insert additional debugging statements to find
out exactly where the crash is happening.
> 18:29:45 13168 ---1 Get 8026488 64 ../src/os.c 649
> 18:29:45 13168 Actual local interface address is 127.0.0.1 (lo0)
> 18:29:45 13168 ---1 Get 8026552 64 ../src/os.c 649
> 18:29:45 13168 Actual local interface address is 10.0.0.200 (hme0)
That message is in the code that discovers the list of running
interfaces on the system. Exim is trying to discover whether any of the
IP addresses it has found from the MX record is the local host. So that
narrows it down a bit. The failure is either in the file src/os.c, in
the loop for finding local interfaces (the problem is with the next
interface), or, if there are no more interfaces, back in src/host.c just
after calling the host_scan_for_local_hosts() function around line 2345.
> 18:29:45 1159 child 13168 ended: status=0xa
>
> Am I correct in assuming that status 0xA means the child process
> received SIGBUS?
That's right for Solaris on SPARC; presumably the same for OpenBSD.
> When Exim is compiled, gcc gives the following warning several
> times:
>
> store.c: In function `store_get_3':
> store.c:170: warning: cast from pointer to integer of different size
Line 170 is a line where it is printing debugging output (the addresses
of memory blocks). This should not affect what Exim is doing. Maybe I
should change those to longs.
I can't suggest anything other than inserting more debugging around the
points I suggest above, to try to narrow things down some more.
Philip
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
Get the Exim 4 book: http://www.uit.co.uk/exim-book
