memcached crash intermittently, segfault from within notify_io_complete

Bug reporter has run into this crash intermittently for some time. Previously, no core file was available so some debugger scripting was used to try to identify the crash. It has identified where the problem is coming from:

Thread 7 (Thread 1105209664 (LWP 12012)):
#0 0x00002aaaaba72440 in pivot_root () from /lib64/libc.so.6
#1 <signal handler called>
#2 0x00000000004129e9 in notify_io_complete (cookie=0x8, status=ENGINE_SUCCESS) at thread.c:517
#3 0x00002aaaabf6d440 in TapConnMap::notifyIOThreadMain (this=0x641cf0, engine=0x641760, serverApi=0x624f00) at /usr/lib/gcc/x86_64-redhat-linux/4.1.2/../../../../include/c++/4.1.2/bits/stl_function.h:519
#4 0x00002aaaabf413e7 in EventuallyPersistentEngine::notifyTapIoThread (this=0x641760) at ep_engine.cc:2538
#5 0x00002aaaabf415f9 in EvpNotifyTapIo (arg=0x8) at ep_engine.cc:783
#6 0x00002aaaab78b2f7 in start_thread () from /lib64/libpthread.so.0
#7 0x00002aaaaba71e3d in clone () from /lib64/libc.so.6

note that pivot_root() is just an unused function as a jumping off point to script where the stack trace should be generated from. The signal handled is likely a SEGV.