Description of problem: on my desktop, in.authd is taking all the cpu time it can get: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 5047 ident 25 0 50304 1008 48m R 95.4 0.0 83:33.61 in.authd No idea why. strace and ltrace dont seem to show it doing anything attacking gdb and getting a bt shows: #0 0x0000002a956c54cb in malloc_consolidate () from /lib64/tls/libc.so.6 #1 0x0000002a956c670f in _int_malloc () from /lib64/tls/libc.so.6 #2 0x0000002a956c8975 in malloc () from /lib64/tls/libc.so.6 #3 0x0000002a956b698a in __fopen_internal () from /lib64/tls/libc.so.6 #4 0x0000000000402c53 in get_info (out=0x7fbffff760, in={lport = 36922, rport = 25}, pathname=0x4056b7 "/proc/net/tcp6") at authd.c:543 #5 0x000000000040419e in main (argc=5, argv=0x7fbffffc98) at authd.c:800 #6 0x0000002a95673fda in __libc_start_main () from /lib64/tls/libc.so.6 #7 0x0000000000401f0a in _start () and then frames 8-119 dont show debug info... as far as I know, aside from irc to devserv, I dont know of anything that would be using ident Version-Release number of selected component (if applicable): authd-1.3.4-1.fc3 How reproducible: No idea
frame 4 shows that the remote port is 25, looks like a MTA is querying it. Are you still able to reproduce this?
looks like it, it started doing it overnight, so its currently doing it on my workstation
this seems to be working on sane compilers/environments now.
Created attachment 116304 [details] Patch to fix free of already freed object.
I see this also, and discovered that it is due to an attempt to free a freed object. This was verified by setting MALLOC_CHECK_=2 in the environment and analysing the resultant core dump.
The bug is still not fixed in Fedora 15.