Add 'show_econnreset' TCP socket option

An ECONNRESET is a socket error which tells us that a TCP peer has sent
an RST. The RST indicates that they have aborted the connection and
that the payload we have received should not be considered complete. Up
until now, the implementation of TCP in inet_drv.c has hidden the
receipt of the RST from the user, treating it as though it was just
a FIN terminating the read side of the socket.

There are many cases where user code needs to be able to distinguish
between a socket that was closed normally and one that was aborted.
Setting the option {show_econnreset, true} enables the user to receive
ECONNRESET errors on both active and passive sockets.

A connected socket returned from gen_tcp:accept/1 will inherit the
show_econnreset setting of the listening socket.

By default this option is set to {show_econnreset, false}.

Note that this patch only enables the reporting of ECONNRESET when
the socket is being read from. It does not report ECONNRESET (or
EPIPE) when the user tries to write to a connection when an RST
has already been received. Currently the TCP implementation in
inet_drv.c hides all such send errors from the user in favour
of returning {error, close}. A separate patch will be needed to
enable the reporting of such errors.

2 files changed, 48 insertions, 2 deletions

diff --git a/erts/emulator/drivers/common/inet_drv.c b/erts/emulator/drivers/common/inet_drv.c
index e001f31932..b5903f0c07 100644
--- a/erts/emulator/drivers/common/inet_drv.c
+++ b/erts/emulator/drivers/common/inet_drv.c
@@ -835,6 +835,7 @@ static int my_strncasecmp(const char *s1, const char *s2, size_t n)
 #define TCP_ADDF_PENDING_SHUT_RDWR 32 /* Call shutdown(sock, SHUT_RDWR) when queue empties */
 #define TCP_ADDF_PENDING_SHUTDOWN \
 		(TCP_ADDF_PENDING_SHUT_WR | TCP_ADDF_PENDING_SHUT_RDWR)
+#define TCP_ADDF_SHOW_ECONNRESET   64 /* Tell user about incoming RST */
 
 /* *_REQ_* replies */
 #define INET_REP_ERROR       0
@@ -879,6 +880,7 @@ static int my_strncasecmp(const char *s1, const char *s2, size_t n)
 #define INET_LOPT_MSGQ_HIWTRMRK     36  /* set local msgq high watermark */
 #define INET_LOPT_MSGQ_LOWTRMRK     37  /* set local msgq low watermark */
 #define INET_LOPT_NETNS             38  /* Network namespace pathname */
+#define INET_LOPT_TCP_SHOW_ECONNRESET 39  /* tell user about incoming RST */
 /* SCTP options: a separate range, from 100: */
 #define SCTP_OPT_RTOINFO		100
 #define SCTP_OPT_ASSOCINFO		101
@@ -6255,6 +6257,16 @@ static int inet_set_opts(inet_descriptor* desc, char* ptr, int len)
 	    continue;
 #endif
 
+	case INET_LOPT_TCP_SHOW_ECONNRESET:
+	    if (desc->sprotocol == IPPROTO_TCP) {
+		tcp_descriptor* tdesc = (tcp_descriptor*) desc;
+		if (ival)
+		    tdesc->tcp_add_flags |= TCP_ADDF_SHOW_ECONNRESET;
+		else
+		    tdesc->tcp_add_flags &= ~TCP_ADDF_SHOW_ECONNRESET;
+	    }
+	    continue;
+
 	case INET_OPT_REUSEADDR: 
 #ifdef __WIN32__
 	    continue;  /* Bjorn says */
@@ -7234,6 +7246,17 @@ static ErlDrvSSizeT inet_fill_opts(inet_descriptor* desc,
 	    continue;
 #endif
 
+	case INET_LOPT_TCP_SHOW_ECONNRESET:
+	    if (desc->sprotocol == IPPROTO_TCP) {
+		tcp_descriptor* tdesc = (tcp_descriptor*) desc;
+		*ptr++ = opt;
+		ival = !!(tdesc->tcp_add_flags & TCP_ADDF_SHOW_ECONNRESET);
+		put_int32(ival, ptr);
+	    } else {
+		TRUNCATE_TO(0,ptr);
+	    }
+	    continue;
+
 	case INET_OPT_PRIORITY:
 #ifdef SO_PRIORITY
 	    type = SO_PRIORITY;
@@ -9077,6 +9100,11 @@ static tcp_descriptor* tcp_inet_copy(tcp_descriptor* desc,SOCKET s,
     copy_desc->low           = desc->low;
     copy_desc->send_timeout  = desc->send_timeout;
     copy_desc->send_timeout_close = desc->send_timeout_close;
+
+    if (desc->tcp_add_flags & TCP_ADDF_SHOW_ECONNRESET)
+	copy_desc->tcp_add_flags |= TCP_ADDF_SHOW_ECONNRESET;
+    else
+	copy_desc->tcp_add_flags &= ~TCP_ADDF_SHOW_ECONNRESET;
     
     /* The new port will be linked and connected to the original caller */
     port = driver_create_port(port, owner, "tcp_inet", (ErlDrvData) copy_desc);
@@ -10014,7 +10042,10 @@ static int tcp_recv(tcp_descriptor* desc, int request_len)
 	int err = sock_errno();
 	if (err == ECONNRESET) {
 	    DEBUGF((" => detected close (connreset)\r\n"));
-	    return tcp_recv_closed(desc);
+	    if (desc->tcp_add_flags & TCP_ADDF_SHOW_ECONNRESET)
+		return tcp_recv_error(desc, err);
+	    else
+		return tcp_recv_closed(desc);
 	}
 	if (err == ERRNO_BLOCK) {
 	    DEBUGF((" => would block\r\n"));
@@ -10226,7 +10257,19 @@ static void tcp_inet_event(ErlDrvData e, ErlDrvEvent event)
     if (netEv.lNetworkEvents & FD_CLOSE) {
 	/* error in err = netEv.iErrorCode[FD_CLOSE_BIT] */
 	DEBUGF(("Detected close in %s, line %d\r\n", __FILE__, __LINE__));
-	tcp_recv_closed(desc);
+	if (desc->tcp_add_flags & TCP_ADDF_SHOW_ECONNRESET) {
+	    err = netEv.iErrorCode[FD_CLOSE_BIT];
+	    if (err == ECONNRESET)
+		tcp_recv_error(desc, err);
+	    else if (err == ECONNABORTED && IS_CONNECTED(INETP(desc))) {
+		/* translate this error to ECONNRESET */
+		tcp_recv_error(desc, ECONNRESET);
+	    }
+	    else
+		tcp_recv_closed(desc);
+	}
+	else
+	    tcp_recv_closed(desc);
     }
     DEBUGF(("tcp_inet_event(%ld) }\r\n", (long)desc->inet.port));
     return;
diff --git a/erts/preloaded/src/prim_inet.erl b/erts/preloaded/src/prim_inet.erl
index 622e1be869..fb0f67aa8a 100644
--- a/erts/preloaded/src/prim_inet.erl
+++ b/erts/preloaded/src/prim_inet.erl
@@ -1140,6 +1140,7 @@ enc_opt(delay_send)      -> ?INET_LOPT_TCP_DELAY_SEND;
 enc_opt(packet_size)     -> ?INET_LOPT_PACKET_SIZE;
 enc_opt(read_packets)    -> ?INET_LOPT_READ_PACKETS;
 enc_opt(netns)           -> ?INET_LOPT_NETNS;
+enc_opt(show_econnreset) -> ?INET_LOPT_TCP_SHOW_ECONNRESET;
 enc_opt(raw)             -> ?INET_OPT_RAW;
 % Names of SCTP opts:
 enc_opt(sctp_rtoinfo)	 	   -> ?SCTP_OPT_RTOINFO;
@@ -1197,6 +1198,7 @@ dec_opt(?INET_LOPT_TCP_DELAY_SEND)   -> delay_send;
 dec_opt(?INET_LOPT_PACKET_SIZE)      -> packet_size;
 dec_opt(?INET_LOPT_READ_PACKETS)     -> read_packets;
 dec_opt(?INET_LOPT_NETNS)           -> netns;
+dec_opt(?INET_LOPT_TCP_SHOW_ECONNRESET) -> show_econnreset;
 dec_opt(?INET_OPT_RAW)              -> raw;
 dec_opt(I) when is_integer(I)     -> undefined.
 
@@ -1296,6 +1298,7 @@ type_opt_1(delay_send)      -> bool;
 type_opt_1(packet_size)     -> uint;
 type_opt_1(read_packets)    -> uint;
 type_opt_1(netns)           -> binary;
+type_opt_1(show_econnreset) -> bool;
 %% 
 %% SCTP options (to be set). If the type is a record type, the corresponding
 %% record signature is returned, otherwise, an "elementary" type tag