From bac6abe18d28373e0d2d0666c411020404197337 Mon Sep 17 00:00:00 2001 From: Matt Caswell Date: Mon, 4 Sep 2017 11:20:27 +0100 Subject: [PATCH 1/1] Allow an endpoint to read the alert data before closing the socket If an alert gets sent and then we close the connection immediately with data still in the input buffer then a TCP-RST gets sent. Some OSs immediately abandon data in their input buffer if a TCP-RST is received - meaning the alert data itself gets ditched. Sending a TCP-FIN before the TCP-RST seems to avoid this. This was causing test failures in MSYS2 builds. Reviewed-by: Rich Salz Reviewed-by: Ben Kaduk (Merged from https://github.com/openssl/openssl/pull/4333) --- apps/s_socket.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/apps/s_socket.c b/apps/s_socket.c index 3bdb587395..2f3e90bbf9 100644 --- a/apps/s_socket.c +++ b/apps/s_socket.c @@ -221,6 +221,23 @@ int do_server(int *accept_sock, const char *host, const char *port, break; } i = (*cb)(sock, type, protocol, context); + /* + * If we ended with an alert being sent, but still with data in the + * network buffer to be read, then calling BIO_closesocket() will + * result in a TCP-RST being sent. On some platforms (notably + * Windows) then this will result in the peer immediately abandoning + * the connection including any buffered alert data before it has + * had a chance to be read. Shutting down the sending side first, + * and then closing the socket sends TCP-FIN first followed by + * TCP-RST. This seems to allow the peer to read the alert data. + */ +#ifdef _WIN32 +# ifdef SD_SEND + shutdown(sock, SD_SEND); +# endif +#elif defined(SHUT_WR) + shutdown(sock, SHUT_WR); +#endif BIO_closesocket(sock); } else { i = (*cb)(asock, type, protocol, context); -- 2.34.1