Looks as if I can reply to my own message! Maybe this exchange might help someone else. It seems that EPOLLIN events with no data are inevitable and have to be accepted. So in a way, it is a quirk of OpenSSL that typical example code results in the no data case being interpreted as an error. I've concluded that if the analysis of the return gives SYSCALL error but there is nothing to be had from ERR_get_error, then the case should be ignored, and not logged as an error. But I'd still be happy to receive a second opinion! -- View this message in context: http://openssl.6102.n7.nabble.com/Extra-EPOLLIN-event-at-end-of-SSL-connection-tp63384p63394.html Sent from the OpenSSL - User mailing list archive at Nabble.com.
