Comment #27 on issue 99 by dorma...@rydia.net: Memcached 1.4.2 server segmentation fault
http://code.google.com/p/memcached/issues/detail?id=99
That doesn't really help me no... As I said above, that particular case (line 5510) is entirely new code that the test doesn't even run. That's for a specific case when a client using the TAP protocol connects and subscribes to a data stream.
The update_event() logic hasn't changed at all, which is where the backtrace would always fail.
I'm curious as to what the results of the first patch (with the asserts) are for anyone who's actually having this failure.