We're seeing exim (4.50, compiled from ports on FreeBSD
5.21) occasionally crash with signal 11 during message
submission from a PHP script. I haven't been able to
reproduce the problem in controlled conditions (we have a
periodic script which sends mail to a bunch of recipients
which provokes it). The same problem occured with the
previous exim version.
I do have a stack trace from the crash, though. This was
obtained by adding a signal handler for SIGSEGV which
sleeps to give an opportunity to attach a debugger. This
one came from a process doing
exim -Mc 1DDfbC-000Pc3-AY
I haven't had a chance to look at the code in any detail
yet, but it looks like the crash is occuring after exim
has negotiated TLS with a remote SMTP server.
Has anyone seen anything like this before?
#0 0x284f00cf in nanosleep () from /lib/libc.so.5
#1 0x284e420f in sleep () from /lib/libc.so.5
#2 0x080b8fb2 in crash_signal_handler ()
#3 <signal handler called>
#4 0x080a87f6 in string_vformat (buffer=0xbfbfe4f0 "Ñ\016", buflen=0,
format=0x80f4e96 "%s %s\r\n", ap=0x80efe85 "EHLO") at string.c:1113
#5 0x080a42c4 in smtp_write_command (outblock=0xbfbfe4f0, noflush=0,
format=0xf0 <Error reading address 0xf0: Bad address>) at smtp_out.c:315
#6 0x080cdad9 in smtp_deliver (addrlist=0x811f388, host=0x811f868,
host_af=240, port=25,
interface=0xf0 <Error reading address 0xf0: Bad address>,
tblock=0x811cca8, copy_host=0, message_defer=0xbfbfe5dc, suppress_tls=0)
at smtp.c:1011
#7 0x080cff49 in smtp_transport_entry (tblock=0x811cca8, addrlist=0x811f388)
at smtp.c:2456
#8 0x08063339 in do_remote_deliveries (fallback=0) at deliver.c:3754
#9 0x080661bb in deliver_message (id=0xbfbfed69 "1DDfbC-000Pc3-AY", forced=0,
give_up=0) at deliver.c:5779
#10 0x0806ef0e in main (argc=3, cargv=0xbfbfec78) at exim.c:3767
#11 0x08056de2 in _start ()
--
``What does it mean? It means I never have to work again.''
(Don McLean, on `American Pie', attrib.)
