Looking further, I found core files on the mupdate master, from mupdate,
and suspect these are correlated with the downstream crashes.
In this case the crash is here:
#0 0x00000036ed647e2c in vfprintf () from /lib64/libc.so.6
#1 0x00000036ed6e4f10 in __vsyslog_chk () from /lib64/libc.so.6
#2 0x00000036ed6e5140 in syslog () from /lib64/libc.so.6
#3 0x000000000040d703 in mupdate_connect (server=0xe45e70
"mupdate-internal",
port=0x0, handle=0x7ff4e88a5dc8, cbs=0x7ff4e0001d10)
at mupdate-client.c:133
#4 0x000000000040d437 in mupdate_client_start (rock=0x0)
at mupdate-slave.c:309
#5 0x00000036eda07a51 in start_thread () from /lib64/libpthread.so.0
#6 0x00000036ed6e89ad in clone () from /lib64/libc.so.6
Line 133 in mupdate-client.c matches this:
h->conn = backend_connect(NULL, server, &mupdate_protocol,
"", cbs, &status);
Again, conn is null at this point (see below). I suspect these crashes
cause those in lmtpd.
I'm not sure what else to offer here ... a patch not to dereference conn
if it's null?
(gdb) print *h
$2 = {conn = 0x0, tagn = 0, tag = {s = 0x0, len = 0, alloc = 0, flags = 0},
cmd = {s = 0x0, len = 0, alloc = 0, flags = 0}, arg1 = {s = 0x0, len
= 0,
alloc = 0, flags = 0}, arg2 = {s = 0x0, len = 0, alloc = 0, flags =
0},
arg3 = {s = 0x0, len = 0, alloc = 0, flags = 0},
mailbox_buf = '\000' <repeats 1023 times>,
server_buf = '\000' <repeats 1023 times>, acl_buf = 0x0, acl_buf_len
= 0,
mailboxdata_buf = {mailbox = 0x0, server = 0x0, acl = 0x0, t = ACTIVE},
saslcompleted = 0}
Thanks,
g
