Joe Vieira wrote:
I've been trying to pin down a rather elusive segfault for over 2 months now. and i finally got it to happen inside of gdb.
this is freeradius 1.1.6, on rhel5 x86-64
if this problem is fixed in 2.0 or 1.1.7 please let me know.
Since we have no idea what the problem is, the answer is likely "no". ...
Thread 1 (Thread 46912543318400 (LWP 8450)): #0 0x00002aaaac1666d5 in _int_malloc () from /lib64/libc.so.6 No symbol table info available.
If malloc() is core dumping, then something else is going wrong. i.e. some other part of the server is over-writing memory. ..
#11 0x00002aaaae1d77da in tls_handshake_recv (ssn=0x555558718240) at tls.c:173 err = <value optimized out>
For debugging, you should probably build the server without using -O flags. That way, all of the information will be available to gdb. I would try 2.0. Large amounts of code have been re-written or updated. It may not be perfect, but there are good odds that this problem won't re-appear. Alan DeKok.