Details
-
Bug
-
Resolution: Fixed
-
1.6.0 beta4
-
None
-
Operating System: All
Platform: All
Description
While looking for the culprit to another bug, I saw that memcached had seemingly core'ed while it was doing top-keys stuff while handling a binary protocol incr. Here was the stack on the offending thread...
-----------------
Thread 6 Crashed:
0 libSystem.B.dylib 0x00007fff8421601e __semwait_signal_nocancel + 10
1 libSystem.B.dylib 0x00007fff84215f20 nanosleep$NOCANCEL + 129
2 libSystem.B.dylib 0x00007fff84272902 usleep$NOCANCEL + 57
3 libSystem.B.dylib 0x00007fff84291eb8 abort + 93
4 libSystem.B.dylib 0x00007fff8427eef0 __pthread_markcancel + 0
5 memcached 0x00000001000124d4 genhash_string_hash + 100
6 memcached 0x0000000100012764 genhash_find_entry + 36
7 memcached 0x0000000100012829 genhash_find + 9
8 memcached 0x0000000100010811 topkeys_item_get_or_create + 49
9 memcached 0x0000000100009546 complete_incr_bin + 1510
10 memcached 0x000000010000d70f complete_nread_binary + 767
11 memcached 0x000000010000e90a conn_nread + 970
12 memcached 0x0000000100002626 event_handler + 166
13 libevent-1.4.2.dylib 0x0000000100029031 event_base_loop + 1217
14 memcached 0x0000000100010411 worker_libevent + 65
15 libSystem.B.dylib 0x00007fff841dc8b6 _pthread_start + 331
16 libSystem.B.dylib 0x00007fff841dc769 thread_start + 13