Stefan Winter wrote:
Well, this morning's GIT doesn't quite agree. Sending a TERM to the server running -X:
Arg. I may have to back-port a bunch of fixes from the master branch.
gdb during termination time shows that there's something wrong inside free'ing eap types, it seems:
Double free, probably.
Program received signal SIGSEGV, Segmentation fault. 0x00007ffff65a5634 in free () from /lib64/libc.so.6 (gdb) bt #0 0x00007ffff65a5634 in free () from /lib64/libc.so.6 #1 0x000000000040e2c0 in cf_section_parse_free (cs=0x675240, base=0x8209c0) at conffile.c:344 #2 0x00007ffff56e0d31 in eaptype_free (i=0x8247f0) at mem.c:253 #3 0x00007ffff56de877 in eap_detach (instance=0x823ad0) at rlm_eap.c:69
I'll take a look. I spent some effort in the master branch to clean it up. It now exits cleanly when using "radiusd -M". I'm wary of making changes to the v2.x.x branch, but bug fixes are definitely on the list. Alan DeKok.