subdaemons: remove gossipd fd from per-peer daemons.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

channeld/channeld.c

@@ -45,9 +45,9 @@
 #include <wire/peer_wire.h>
 #include <wire/wire_sync.h>
 
-/* stdin == requests, 3 == peer, 4 = gossip, 5 = HSM */
+/* stdin == requests, 3 == peer, 4 = HSM */
 #define MASTER_FD STDIN_FILENO
-#define HSM_FD 5
+#define HSM_FD 4
 
 struct peer {
 	struct per_peer_state *pps;
@@ -2132,13 +2132,12 @@ static void peer_in(struct peer *peer, const u8 *msg)
 {
 	enum peer_wire type = fromwire_peektype(msg);
 
-	if (handle_peer_gossip_or_error(peer->pps, &peer->channel_id, msg))
+	if (handle_peer_error(peer->pps, &peer->channel_id, msg))
 		return;
 
 	/* Must get funding_locked before almost anything. */
 	if (!peer->funding_locked[REMOTE]) {
 		if (type != WIRE_FUNDING_LOCKED
-		    && type != WIRE_PONG
 		    && type != WIRE_SHUTDOWN
 		    /* We expect these for v2 !! */
 		    && type != WIRE_TX_SIGNATURES
@@ -2215,7 +2214,7 @@ static void peer_in(struct peer *peer, const u8 *msg)
 		handle_unexpected_reestablish(peer, msg);
 		return;
 
-	/* These are all swallowed by handle_peer_gossip_or_error */
+	/* These are all swallowed by connectd */
 	case WIRE_CHANNEL_ANNOUNCEMENT:
 	case WIRE_CHANNEL_UPDATE:
 	case WIRE_NODE_ANNOUNCEMENT:
@@ -2798,7 +2797,7 @@ skip_tlvs:
 	do {
 		clean_tmpctx();
 		msg = peer_read(tmpctx, peer->pps);
-	} while (handle_peer_gossip_or_error(peer->pps, &peer->channel_id, msg) ||
+	} while (handle_peer_error(peer->pps, &peer->channel_id, msg) ||
 		 capture_premature_msg(&premature_msgs, msg));
 
 got_reestablish:
@@ -3752,9 +3751,9 @@ static void init_channel(struct peer *peer)
 			       tal_dup(peer, struct penalty_base, &pbases[i]));
 	tal_free(pbases);
 
-	/* stdin == requests, 3 == peer, 4 = gossip */
+	/* stdin == requests, 3 == peer */
 	peer->pps = new_per_peer_state(peer);
-	per_peer_state_set_fds(peer->pps, 3, 4);
+	per_peer_state_set_fd(peer->pps, 3);
 
 	status_debug("init %s: remote_per_commit = %s, old_remote_per_commit = %s"
 		     " next_idx_local = %"PRIu64
@@ -3895,11 +3894,10 @@ int main(int argc, char *argv[])
 	FD_ZERO(&fds_in);
 	FD_SET(MASTER_FD, &fds_in);
 	FD_SET(peer->pps->peer_fd, &fds_in);
-	FD_SET(peer->pps->gossip_fd, &fds_in);
 
 	FD_ZERO(&fds_out);
 	FD_SET(peer->pps->peer_fd, &fds_out);
-	nfds = peer->pps->gossip_fd+1;
+	nfds = peer->pps->peer_fd+1;
 
 	while (!shutdown_complete(peer)) {
 		struct timemono first;
@@ -3958,13 +3956,6 @@ int main(int argc, char *argv[])
 			/* This could take forever, but who cares? */
 			msg = peer_read(tmpctx, peer->pps);
 			peer_in(peer, msg);
-		} else if (FD_ISSET(peer->pps->gossip_fd, &rfds)) {
-			msg = wire_sync_read(tmpctx, peer->pps->gossip_fd);
-			/* Gossipd hangs up on us to kill us when a new
-			 * connection comes in. */
-			if (!msg)
-				peer_failed_connection_lost();
-			handle_gossip_msg(peer->pps, take(msg));
 		}
 	}
 

closingd/closingd.c

@@ -29,9 +29,9 @@
 #include <wire/peer_wire.h>
 #include <wire/wire_sync.h>
 
-/* stdin == requests, 3 == peer, 4 = gossip, 5 = hsmd */
+/* stdin == requests, 3 == peer, 4 = hsmd */
 #define REQ_FD STDIN_FILENO
-#define HSM_FD 5
+#define HSM_FD 4
 
 static void notify(enum log_level level, const char *fmt, ...)
 {
@@ -117,15 +117,10 @@ static u8 *closing_read_peer_msg(const tal_t *ctx,
 {
 	for (;;) {
 		u8 *msg;
-		bool from_gossipd;
 
 		clean_tmpctx();
-		msg = peer_or_gossip_sync_read(ctx, pps, &from_gossipd);
+		msg = peer_read(ctx, pps);
-		if (from_gossipd) {
+		if (!handle_peer_error(pps, channel_id, msg))
-			handle_gossip_msg(pps, take(msg));
-			continue;
-		}
-		if (!handle_peer_gossip_or_error(pps, channel_id, msg))
 			return msg;
 	}
 }
@@ -892,9 +887,9 @@ int main(int argc, char *argv[])
 				    &wrong_funding))
 		master_badmsg(WIRE_CLOSINGD_INIT, msg);
 
-	/* stdin == requests, 3 == peer, 4 = gossip, 5 = hsmd */
+	/* stdin == requests, 3 == peer, 4 = hsmd */
 	pps = notleak(new_per_peer_state(ctx));
-	per_peer_state_set_fds(pps, 3, 4);
+	per_peer_state_set_fd(pps, 3);
 
 	funding_wscript = bitcoin_redeem_2of2(ctx,
 					      &funding_pubkey[LOCAL],

common/peer_failed.c

@@ -19,7 +19,6 @@ peer_fatal_continue(const u8 *msg TAKES, const struct per_peer_state *pps)
  	status_send(msg);
 
 	status_send_fd(pps->peer_fd);
-	status_send_fd(pps->gossip_fd);
 	exit(0x80 | (reason & 0xFF));
 }
 

common/per_peer_state.c

@@ -12,39 +12,25 @@ static void destroy_per_peer_state(struct per_peer_state *pps)
 {
 	if (pps->peer_fd != -1)
 		close(pps->peer_fd);
-	if (pps->gossip_fd != -1)
-		close(pps->gossip_fd);
 }
 
 struct per_peer_state *new_per_peer_state(const tal_t *ctx)
 {
 	struct per_peer_state *pps = tal(ctx, struct per_peer_state);
 
-	pps->peer_fd = pps->gossip_fd = -1;
+	pps->peer_fd = -1;
 	tal_add_destructor(pps, destroy_per_peer_state);
 	return pps;
 }
 
-void per_peer_state_set_fds(struct per_peer_state *pps,
+void per_peer_state_set_fd(struct per_peer_state *pps, int peer_fd)
-			    int peer_fd, int gossip_fd)
 {
 	assert(pps->peer_fd == -1);
-	assert(pps->gossip_fd == -1);
 	pps->peer_fd = peer_fd;
-	pps->gossip_fd = gossip_fd;
-}
-
-void per_peer_state_set_fds_arr(struct per_peer_state *pps, const int *fds)
-{
-	/* We expect 2 fds. */
-	assert(tal_count(fds) == 2);
-	per_peer_state_set_fds(pps, fds[0], fds[1]);
 }
 
 void per_peer_state_fdpass_send(int fd, const struct per_peer_state *pps)
 {
 	assert(pps->peer_fd != -1);
-	assert(pps->gossip_fd != -1);
 	fdpass_send(fd, pps->peer_fd);
-	fdpass_send(fd, pps->gossip_fd);
 }

common/per_peer_state.h

@@ -9,19 +9,15 @@
 /* Things we hand between daemons to talk to peers. */
 struct per_peer_state {
 	/* If not -1, closed on freeing */
-	int peer_fd, gossip_fd;
+	int peer_fd;
 };
 
 /* Allocate a new per-peer state and add destructor to close fds if set;
- * sets fds to -1. */
+ * sets peer_fd to -1. */
 struct per_peer_state *new_per_peer_state(const tal_t *ctx);
 
 /* Initialize the fds (must be -1 previous) */
-void per_peer_state_set_fds(struct per_peer_state *pps,
+void per_peer_state_set_fd(struct per_peer_state *pps, int peer_fd);
-			    int peer_fd, int gossip_fd);
-
-/* Array version of above: tal_count(fds) must be 2 */
-void per_peer_state_set_fds_arr(struct per_peer_state *pps, const int *fds);
 
 void per_peer_state_fdpass_send(int fd, const struct per_peer_state *pps);
 

common/read_peer_msg.c

@@ -12,39 +12,6 @@
 #include <wire/peer_wire.h>
 #include <wire/wire_sync.h>
 
-u8 *peer_or_gossip_sync_read(const tal_t *ctx,
-			     struct per_peer_state *pps,
-			     bool *from_gossipd)
-{
-	fd_set readfds;
-	u8 *msg;
-
-	FD_ZERO(&readfds);
-	FD_SET(pps->peer_fd, &readfds);
-	FD_SET(pps->gossip_fd, &readfds);
-
-	if (select(pps->peer_fd > pps->gossip_fd
-		   ? pps->peer_fd + 1 : pps->gossip_fd + 1,
-		   &readfds, NULL, NULL, NULL) <= 0) {
-		status_failed(STATUS_FAIL_GOSSIP_IO,
-			      "select failed?: %s", strerror(errno));
-	}
-
-	if (FD_ISSET(pps->peer_fd, &readfds)) {
-		msg = peer_read(ctx, pps);
-		*from_gossipd = false;
-		return msg;
-	}
-
-	msg = wire_sync_read(ctx, pps->gossip_fd);
-	if (!msg)
-		status_failed(STATUS_FAIL_GOSSIP_IO,
-			      "Error reading gossip msg: %s",
-			      strerror(errno));
-	*from_gossipd = true;
-	return msg;
-}
-
 bool is_peer_error(const tal_t *ctx, const u8 *msg,
 		   const struct channel_id *channel_id,
 		   char **desc, bool *warning)
@@ -94,70 +61,23 @@ bool is_wrong_channel(const u8 *msg, const struct channel_id *expected,
 	return !channel_id_eq(expected, actual);
 }
 
-void handle_gossip_msg(struct per_peer_state *pps, const u8 *msg TAKES)
+bool handle_peer_error(struct per_peer_state *pps,
-{
-	u8 *gossip;
-
-	/* It's a raw gossip msg: this copies or takes() */
-	gossip = tal_dup_talarr(tmpctx, u8, msg);
-
-	/* Gossipd can send us gossip messages, OR warnings */
-	if (fromwire_peektype(gossip) == WIRE_WARNING) {
-		peer_write(pps, gossip);
-		peer_failed_connection_lost();
-	} else {
-		peer_write(pps, gossip);
-	}
-}
-
-bool handle_peer_gossip_or_error(struct per_peer_state *pps,
 		       const struct channel_id *channel_id,
 		       const u8 *msg TAKES)
 {
 	char *err;
 	bool warning;
-	u8 *pong;
-
-#if DEVELOPER
-	/* Any odd-typed unknown message is handled by the caller, so if we
-	 * find one here it's an error. */
-	assert(!is_unknown_msg_discardable(msg));
-#else
-	/* BOLT #1:
-	 *
-	 * A receiving node:
-	 *   - upon receiving a message of _odd_, unknown type:
-	 *     - MUST ignore the received message.
-	 */
-	if (is_unknown_msg_discardable(msg))
-		goto handled;
-#endif
-
-	if (check_ping_make_pong(NULL, msg, &pong)) {
-		if (pong)
-			peer_write(pps, take(pong));
-		return true;
-	} else if (is_msg_for_gossipd(msg)) {
-		wire_sync_write(pps->gossip_fd, msg);
-		/* wire_sync_write takes, so don't take again. */
-		return true;
-	}
-
 	if (is_peer_error(tmpctx, msg, channel_id, &err, &warning)) {
 		/* Ignore unknown channel errors. */
-		if (!err)
+		if (!err) {
-			goto handled;
-
-		/* We hang up when a warning is received. */
-		peer_failed_received_errmsg(pps, err, channel_id, warning);
-
-		goto handled;
-	}
-
-	return false;
-
-handled:
 			if (taken(msg))
 				tal_free(msg);
 			return true;
 		}
+
+		/* We hang up when a warning is received. */
+		peer_failed_received_errmsg(pps, err, channel_id, warning);
+	}
+
+	return false;
+}

common/read_peer_msg.h

@@ -8,25 +8,6 @@ struct crypto_state;
 struct channel_id;
 struct per_peer_state;
 
-/**
- * peer_or_gossip_sync_read - read a peer message, or maybe a gossip msg.
- * @ctx: context to allocate return packet from.
- * @pps: the per-peer peer state and fds
- * @from_gossipd: true if the msg was from gossipd, otherwise false.
- *
- * Will call peer_failed_connection_lost() or
- * status_failed(STATUS_FAIL_GOSSIP_IO) or return a message.
- *
- * Usually, you should call handle_gossip_msg if *@from_gossipd is
- * true, otherwise if is_peer_error() handle the error, otherwise if
- * is_msg_for_gossipd() then send to gossipd, otherwise if is
- * is_wrong_channel() send that as a reply.  Otherwise it should be
- * a valid message.
- */
-u8 *peer_or_gossip_sync_read(const tal_t *ctx,
-			     struct per_peer_state *pps,
-			     bool *from_gossipd);
-
 /**
  * is_peer_error - if it's an error, describe if it applies to this channel.
  * @ctx: context to allocate return from.
@@ -55,26 +36,15 @@ bool is_wrong_channel(const u8 *msg, const struct channel_id *expected,
 
 
 /**
- * handle_peer_gossip_or_error - simple handler for all the above cases.
+ * handle_peer_error - simple handler for errors
  * @pps: per-peer state.
  * @channel_id: the channel id of the current channel.
  * @msg: the peer message (only taken if returns true).
  *
- * This returns true if it handled the packet: a gossip packet (forwarded
+ * This returns true if it handled the packet.
- * to gossipd), or an error packet (causes peer_failed_received_errmsg or
- * ignored), or a ping (may reply with pong).
  */
-bool handle_peer_gossip_or_error(struct per_peer_state *pps,
+bool handle_peer_error(struct per_peer_state *pps,
 		       const struct channel_id *channel_id,
 		       const u8 *msg TAKES);
 
-/**
- * handle_timestamp_filter - deal with timestamp filter requests.
- * @pps: per-peer state.
- * @msg: the peer message (only taken if returns true).
- */
-bool handle_timestamp_filter(struct per_peer_state *pps, const u8 *msg TAKES);
-
-/* We got this message from gossipd: forward/quit as it asks. */
-void handle_gossip_msg(struct per_peer_state *pps, const u8 *msg TAKES);
 #endif /* LIGHTNING_COMMON_READ_PEER_MSG_H */

connectd/connectd.c

@@ -336,6 +336,8 @@ static struct io_plan *peer_reconnected(struct io_conn *conn,
 /*~ When we free a peer, we remove it from the daemon's hashtable */
 static void destroy_peer(struct peer *peer, struct daemon *daemon)
 {
+	if (peer->gossip_fd >= 0)
+		close(peer->gossip_fd);
 	peer_htable_del(&daemon->peers, peer);
 }
 
@@ -395,7 +397,7 @@ struct io_plan *peer_connected(struct io_conn *conn,
 	struct peer *peer;
 	int unsup;
 	size_t depender, missing;
-	int subd_fd, gossip_fd;
+	int subd_fd;
 
 	peer = peer_htable_get(&daemon->peers, id);
 	if (peer)
@@ -454,8 +456,8 @@ struct io_plan *peer_connected(struct io_conn *conn,
 		return io_close(conn);
 
 	/* If gossipd can't give us a file descriptor, we give up connecting. */
-	gossip_fd = get_gossipfd(daemon, id, their_features);
+	peer->gossip_fd = get_gossipfd(daemon, id, their_features);
-	if (gossip_fd < 0) {
+	if (peer->gossip_fd < 0) {
 		close(subd_fd);
 		return tal_free(peer);
 	}
@@ -469,10 +471,9 @@ struct io_plan *peer_connected(struct io_conn *conn,
 
 	/*~ daemon_conn is a message queue for inter-daemon communication: we
 	 * queue up the `connect_peer_connected` message to tell lightningd
-	 * we have connected, and give the peer and gossip fds. */
+	 * we have connected, and give the peer fd. */
 	daemon_conn_send(daemon->master, take(msg));
 	daemon_conn_send_fd(daemon->master, subd_fd);
-	daemon_conn_send_fd(daemon->master, gossip_fd);
 
 	/*~ Now we set up this connection to read/write from subd */
 	return multiplex_peer_setup(conn, peer);
@@ -1894,20 +1895,19 @@ static void peer_final_msg(struct io_conn *conn,
 	struct peer *peer;
 	struct node_id id;
 	u8 *finalmsg;
+	int peer_fd;
 
 	if (!fromwire_connectd_peer_final_msg(tmpctx, msg, &id, &finalmsg))
 		master_badmsg(WIRE_CONNECTD_PEER_FINAL_MSG, msg);
 
-	/* Get the peer_fd and gossip_fd for this peer: we don't need them. */
+	/* Get the peer_fd for this peer: we don't need it though! */
 	io_fd_block(io_conn_fd(conn), true);
-	for (size_t i = 0; i < 2; i++) {
+	peer_fd = fdpass_recv(io_conn_fd(conn));
-		int fd = fdpass_recv(io_conn_fd(conn));
+	if (peer_fd == -1)
-		if (fd == -1)
 		status_failed(STATUS_FAIL_MASTER_IO,
-				      "Getting fd %zu after peer_final_msg: %s",
+			      "Getting peer fd after peer_final_msg: %s",
-				      i, strerror(errno));
+			      strerror(errno));
-		close(fd);
+	close(peer_fd);
-	}
 	io_fd_block(io_conn_fd(conn), false);
 
 	/* This can happen if peer hung up on us. */

connectd/connectd.h

@@ -81,6 +81,9 @@ struct peer {
 	/* Random ping timer, to detect dead connections. */
 	struct oneshot *ping_timer;
 
+	/* FIXME: remove! */
+	int gossip_fd;
+
 #if DEVELOPER
 	bool dev_read_enabled;
 	/* If non-NULL, this counts down; 0 means disable */

connectd/connectd_wire.csv

@@ -60,7 +60,7 @@ msgdata,connectd_connect_failed,failreason,wirestring,
 msgdata,connectd_connect_failed,seconds_to_delay,u32,
 msgdata,connectd_connect_failed,addrhint,?wireaddr_internal,
 
-# Connectd -> master: we got a peer. Three fds: peer, gossip and gossip_store
+# Connectd -> master: we got a peer. Plus fd for peer daemon
 msgtype,connectd_peer_connected,2002
 msgdata,connectd_peer_connected,id,node_id,
 msgdata,connectd_peer_connected,addr,wireaddr_internal,
@@ -72,7 +72,7 @@ msgdata,connectd_peer_connected,features,u8,flen
 msgtype,connectd_peer_disconnected,2015
 msgdata,connectd_peer_disconnected,id,node_id,
 
-# master -> connectd: give message to peer and disconnect.  Three fds: peer, gossip and gossip_store
+# master -> connectd: give message to peer and disconnect.  Plus fd for peer
 msgtype,connectd_peer_final_msg,2003
 msgdata,connectd_peer_final_msg,id,node_id,
 msgdata,connectd_peer_final_msg,len,u16,

lightningd/channel_control.c

@@ -506,9 +506,9 @@ static unsigned channel_msg(struct subd *sd, const u8 *msg, const int *fds)
 		peer_got_shutdown(sd->channel, msg);
 		break;
 	case WIRE_CHANNELD_SHUTDOWN_COMPLETE:
-		/* We expect 2 fds. */
+		/* We expect 1 fd. */
 		if (!fds)
-			return 2;
+			return 1;
 		peer_start_closingd_after_shutdown(sd->channel, msg, fds);
 		break;
 	case WIRE_CHANNELD_FAIL_FALLEN_BEHIND:
@@ -603,7 +603,6 @@ void peer_start_channeld(struct channel *channel,
 					   channel_errmsg,
 					   channel_set_billboard,
 					   take(&peer_fd->fd),
-					   take(&peer_fd->gossip_fd),
 					   take(&hsmfd), NULL));
 
 	if (!channel->owner) {

lightningd/closing_control.c

@@ -341,8 +341,7 @@ static unsigned closing_msg(struct subd *sd, const u8 *msg, const int *fds UNUSE
 	return 0;
 }
 
-void peer_start_closingd(struct channel *channel,
+void peer_start_closingd(struct channel *channel, struct peer_fd *peer_fd)
-			 struct peer_fd *peer_fd)
 {
 	u8 *initmsg;
 	u32 min_feerate, feerate, *max_feerate;
@@ -372,7 +371,6 @@ void peer_start_closingd(struct channel *channel,
 					   channel_errmsg,
 					   channel_set_billboard,
 					   take(&peer_fd->fd),
-					   take(&peer_fd->gossip_fd),
 					   take(&hsmfd),
 					   NULL));
 

lightningd/connect_control.c

@@ -438,9 +438,9 @@ static unsigned connectd_msg(struct subd *connectd, const u8 *msg, const int *fd
 		break;
 
 	case WIRE_CONNECTD_PEER_CONNECTED:
-		if (tal_count(fds) != 2)
+		if (tal_count(fds) != 1)
-			return 2;
+			return 1;
-		peer_connected(connectd->ld, msg, fds[0], fds[1]);
+		peer_connected(connectd->ld, msg, fds[0]);
 		break;
 
 	case WIRE_CONNECTD_CONNECT_FAILED:

lightningd/dual_open_control.c

@@ -2994,16 +2994,16 @@ static unsigned int dual_opend_msg(struct subd *dualopend,
 			handle_dry_run_finished(dualopend, msg);
 			return 0;
 		case WIRE_DUALOPEND_CHANNEL_LOCKED:
-			if (tal_count(fds) != 2)
+			if (tal_count(fds) != 1)
-				return 2;
+				return 1;
 			handle_channel_locked(dualopend, fds, msg);
 			return 0;
 		case WIRE_DUALOPEND_GOT_SHUTDOWN:
 			handle_peer_wants_to_close(dualopend, msg);
 			return 0;
 		case WIRE_DUALOPEND_SHUTDOWN_COMPLETE:
-			if (tal_count(fds) != 2)
+			if (tal_count(fds) != 1)
-				return 2;
+				return 1;
 			handle_channel_closed(dualopend, fds, msg);
 			return 0;
 		case WIRE_DUALOPEND_FAIL_FALLEN_BEHIND:
@@ -3230,7 +3230,6 @@ static void start_fresh_dualopend(struct peer *peer,
 					  channel_errmsg,
 					  channel_set_billboard,
 					  take(&peer_fd->fd),
-					  take(&peer_fd->gossip_fd),
 					  take(&hsmfd), NULL);
 
 	if (!channel->owner) {
@@ -3297,7 +3296,6 @@ void peer_restart_dualopend(struct peer *peer,
 					   channel_errmsg,
 					   channel_set_billboard,
 					   take(&peer_fd->fd),
-					   take(&peer_fd->gossip_fd),
 					   take(&hsmfd), NULL));
 	if (!channel->owner) {
 		log_broken(channel->log, "Could not subdaemon channel: %s",

lightningd/opening_common.c

@@ -195,9 +195,8 @@ void handle_reestablish(struct lightningd *ld,
 			      take(towire_connectd_peer_final_msg(NULL, peer_id,
 								  err)));
 		subd_send_fd(ld->connectd, peer_fd->fd);
-		subd_send_fd(ld->connectd, peer_fd->gossip_fd);
+		/* Don't close this fd! */
-		/* Don't close those fds! */
+		peer_fd->fd = -1;
-		peer_fd->fd = peer_fd->gossip_fd = -1;
 	}
 }
 

lightningd/opening_control.c

@@ -838,8 +838,8 @@ static unsigned int openingd_msg(struct subd *openingd,
 			tal_free(openingd);
 			return 0;
 		}
-		if (tal_count(fds) != 2)
+		if (tal_count(fds) != 1)
-			return 2;
+			return 1;
 		opening_funder_finished(openingd, msg, fds, uc->fc);
 		return 0;
 	case WIRE_OPENINGD_FUNDER_START_REPLY:
@@ -862,8 +862,8 @@ static unsigned int openingd_msg(struct subd *openingd,
 		return 0;
 
 	case WIRE_OPENINGD_FUNDEE:
-		if (tal_count(fds) != 2)
+		if (tal_count(fds) != 1)
-			return 2;
+			return 1;
 		opening_fundee_finished(openingd, msg, fds, uc);
 		return 0;
 
@@ -872,8 +872,8 @@ static unsigned int openingd_msg(struct subd *openingd,
 		return 0;
 
 	case WIRE_OPENINGD_GOT_REESTABLISH:
-		if (tal_count(fds) != 2)
+		if (tal_count(fds) != 1)
-			return 2;
+			return 1;
 		opening_got_reestablish(openingd, msg, fds, uc);
 		return 0;
 
@@ -919,7 +919,6 @@ void peer_start_openingd(struct peer *peer, struct peer_fd *peer_fd)
 					opend_channel_errmsg,
 					opend_channel_set_billboard,
 					take(&peer_fd->fd),
-					take(&peer_fd->gossip_fd),
 					take(&hsmfd), NULL);
 	if (!uc->open_daemon) {
 		uncommitted_channel_disconnect(uc, LOG_BROKEN,

lightningd/peer_control.c

@@ -1054,9 +1054,8 @@ send_error:
 		      take(towire_connectd_peer_final_msg(NULL, &peer->id,
 							  error)));
 	subd_send_fd(ld->connectd, payload->peer_fd->fd);
-	subd_send_fd(ld->connectd, payload->peer_fd->gossip_fd);
+	/* Don't close the fd! */
-	/* Don't close those fds! */
+	payload->peer_fd->fd = -1;
-	payload->peer_fd->fd = payload->peer_fd->gossip_fd = -1;
 }
 
 static bool
@@ -1112,8 +1111,7 @@ REGISTER_PLUGIN_HOOK(peer_connected,
 
 /* Connectd tells us a peer has connected: it never hands us duplicates, since
  * it holds them until we say peer_died. */
-void peer_connected(struct lightningd *ld, const u8 *msg,
+void peer_connected(struct lightningd *ld, const u8 *msg, int peer_fd)
-		    int peer_fd, int gossip_fd)
 {
 	struct node_id id;
 	u8 *their_features;
@@ -1130,7 +1128,7 @@ void peer_connected(struct lightningd *ld, const u8 *msg,
 		fatal("Connectd gave bad CONNECT_PEER_CONNECTED message %s",
 		      tal_hex(msg, msg));
 
-	hook_payload->peer_fd = new_peer_fd(hook_payload, peer_fd, gossip_fd);
+	hook_payload->peer_fd = new_peer_fd(hook_payload, peer_fd);
 
 	/* If we're already dealing with this peer, hand off to correct
 	 * subdaemon.  Otherwise, we'll hand to openingd to wait there. */

lightningd/peer_control.h

@@ -64,8 +64,7 @@ struct peer *peer_from_json(struct lightningd *ld,
 			    const char *buffer,
 			    const jsmntok_t *peeridtok);
 
-void peer_connected(struct lightningd *ld, const u8 *msg,
+void peer_connected(struct lightningd *ld, const u8 *msg, int peer_fd);
-		    int peer_fd, int gossip_fd);
 
 /* Could be configurable. */
 #define OUR_CHANNEL_FLAGS CHANNEL_FLAGS_ANNOUNCE_CHANNEL

lightningd/peer_fd.c

@@ -7,23 +7,20 @@ static void destroy_peer_fd(struct peer_fd *peer_fd)
 {
 	if (peer_fd->fd != -1)
 		close(peer_fd->fd);
-	if (peer_fd->gossip_fd != -1)
-		close(peer_fd->gossip_fd);
 }
 
-struct peer_fd *new_peer_fd(const tal_t *ctx, int peer_fdnum, int gossip_fd)
+struct peer_fd *new_peer_fd(const tal_t *ctx, int peer_fdnum)
 {
 	struct peer_fd *peer_fd = tal(ctx, struct peer_fd);
 
 	peer_fd->fd = peer_fdnum;
-	peer_fd->gossip_fd = gossip_fd;
 	tal_add_destructor(peer_fd, destroy_peer_fd);
 	return peer_fd;
 }
 
-struct peer_fd *new_peer_fd_arr(const tal_t *ctx, const int *fds)
+struct peer_fd *new_peer_fd_arr(const tal_t *ctx, const int *fd)
 {
-	/* We expect 2 fds. */
+	/* We expect 1 fd. */
-	assert(tal_count(fds) == 2);
+	assert(tal_count(fd) == 1);
-	return new_peer_fd(ctx, fds[0], fds[1]);
+	return new_peer_fd(ctx, fd[0]);
 }

lightningd/peer_fd.h

@@ -3,19 +3,16 @@
 #include "config.h"
 #include <ccan/tal/tal.h>
 
-/* This name is a little preemptive: it still contains the gossip_fd
+/* Tal wrapper for fd connecting subd to connectd */
- * for now! */
 struct peer_fd {
 	/* If not -1, closed on freeing */
 	int fd;
-	int gossip_fd;
 };
 
-/* Allocate a new per-peer state and add destructor to close fds if set;
+/* Allocate a new per-peer state and add destructor to close fd if set. */
- * sets fds to -1. */
+struct peer_fd *new_peer_fd(const tal_t *ctx, int peer_fd);
-struct peer_fd *new_peer_fd(const tal_t *ctx, int peer_fd, int gossip_fd);
 
-/* Array version of above: tal_count(fds) must be 2 */
+/* Array version of above: tal_count(fds) must be 1 */
-struct peer_fd *new_peer_fd_arr(const tal_t *ctx, const int *fds);
+struct peer_fd *new_peer_fd_arr(const tal_t *ctx, const int *fd);
 
 #endif /* LIGHTNING_LIGHTNINGD_PEER_FD_H */

lightningd/subd.c

@@ -414,7 +414,7 @@ static bool log_status_fail(struct subd *sd, const u8 *msg)
 	return true;
 }
 
-static bool handle_peer_error(struct subd *sd, const u8 *msg, int fds[2])
+static bool handle_peer_error(struct subd *sd, const u8 *msg, int fds[1])
 {
 	void *channel = sd->channel;
 	struct channel_id channel_id;
@@ -533,11 +533,11 @@ static struct io_plan *sd_msg_read(struct io_conn *conn, struct subd *sd)
 	if (sd->channel) {
 		switch ((enum peer_status_wire)type) {
 		case WIRE_STATUS_PEER_ERROR:
-			/* We expect 2 fds after this */
+			/* We expect 1 fd after this */
 			if (!sd->fds_in) {
 				/* Don't free msg_in: we go around again. */
 				tal_steal(sd, sd->msg_in);
-				plan = sd_collect_fds(conn, sd, 2);
+				plan = sd_collect_fds(conn, sd, 1);
 				goto out;
 			}
 			if (!handle_peer_error(sd, sd->msg_in, sd->fds_in))

lightningd/test/run-find_my_abspath.c

@@ -168,7 +168,7 @@ struct log *new_log(const tal_t *ctx UNNEEDED, struct log_book *record UNNEEDED,
 struct log_book *new_log_book(struct lightningd *ld UNNEEDED, size_t max_mem UNNEEDED)
 { fprintf(stderr, "new_log_book called!\n"); abort(); }
 /* Generated stub for new_peer_fd_arr */
-struct peer_fd *new_peer_fd_arr(const tal_t *ctx UNNEEDED, const int *fds UNNEEDED)
+struct peer_fd *new_peer_fd_arr(const tal_t *ctx UNNEEDED, const int *fd UNNEEDED)
 { fprintf(stderr, "new_peer_fd_arr called!\n"); abort(); }
 /* Generated stub for new_topology */
 struct chain_topology *new_topology(struct lightningd *ld UNNEEDED, struct log *log UNNEEDED)

lightningd/test/run-invoice-select-inchan.c

@@ -437,7 +437,7 @@ struct height_states *new_height_states(const tal_t *ctx UNNEEDED,
 					const u32 *blockheight UNNEEDED)
 { fprintf(stderr, "new_height_states called!\n"); abort(); }
 /* Generated stub for new_peer_fd */
-struct peer_fd *new_peer_fd(const tal_t *ctx UNNEEDED, int peer_fd UNNEEDED, int gossip_fd UNNEEDED)
+struct peer_fd *new_peer_fd(const tal_t *ctx UNNEEDED, int peer_fd UNNEEDED)
 { fprintf(stderr, "new_peer_fd called!\n"); abort(); }
 /* Generated stub for new_reltimer_ */
 struct oneshot *new_reltimer_(struct timers *timers UNNEEDED,

lightningd/test/run-shuffle_fds.c

@@ -109,7 +109,7 @@ struct log *new_log(const tal_t *ctx UNNEEDED, struct log_book *record UNNEEDED,
 		    const char *fmt UNNEEDED, ...)
 { fprintf(stderr, "new_log called!\n"); abort(); }
 /* Generated stub for new_peer_fd_arr */
-struct peer_fd *new_peer_fd_arr(const tal_t *ctx UNNEEDED, const int *fds UNNEEDED)
+struct peer_fd *new_peer_fd_arr(const tal_t *ctx UNNEEDED, const int *fd UNNEEDED)
 { fprintf(stderr, "new_peer_fd_arr called!\n"); abort(); }
 /* Generated stub for subdaemon_path */
 const char *subdaemon_path(const tal_t *ctx UNNEEDED, const struct lightningd *ld UNNEEDED, const char *name UNNEEDED)

openingd/dualopend.c

@@ -44,9 +44,9 @@
 #include <unistd.h>
 #include <wire/wire_sync.h>
 
-/* stdin == lightningd, 3 == peer, 4 == gossipd, 5 = hsmd */
+/* stdin == lightningd, 3 == peer, 4 = hsmd */
 #define REQ_FD STDIN_FILENO
-#define HSM_FD 5
+#define HSM_FD 4
 
 /* tx_add_input, tx_add_output, tx_rm_input, tx_rm_output */
 #define NUM_TX_MSGS (TX_RM_OUTPUT + 1)
@@ -1164,7 +1164,6 @@ static u8 *opening_negotiate_msg(const tal_t *ctx, struct state *state)
 	 * form, but we use it in a very limited way. */
 	for (;;) {
 		u8 *msg;
-		bool from_gossipd;
 		char *err;
 		bool warning;
 		struct channel_id actual;
@@ -1175,20 +1174,7 @@ static u8 *opening_negotiate_msg(const tal_t *ctx, struct state *state)
 		clean_tmpctx();
 
 		/* This helper routine polls both the peer and gossipd. */
-		msg = peer_or_gossip_sync_read(ctx, state->pps, &from_gossipd);
+		msg = peer_read(ctx, state->pps);
-
-		/* Use standard helper for gossip msgs (forwards, if it's an
-		 * error, exits). */
-		if (from_gossipd) {
-			handle_gossip_msg(state->pps, take(msg));
-			continue;
-		}
-
-		/* Some messages go straight to gossipd. */
-		if (is_msg_for_gossipd(msg)) {
-			wire_sync_write(state->pps->gossip_fd, take(msg));
-			continue;
-		}
 
 		/* BOLT #1:
 		 *
@@ -3440,19 +3426,6 @@ static u8 *handle_funding_depth(struct state *state, u8 *msg)
 	return NULL;
 }
 
-/*~ If we see the gossip_fd readable, we read a whole message.  Sure, we might
- * block, but we trust gossipd. */
-static void handle_gossip_in(struct state *state)
-{
-	u8 *msg = wire_sync_read(NULL, state->pps->gossip_fd);
-
-	if (!msg)
-		status_failed(STATUS_FAIL_GOSSIP_IO,
-			      "Reading gossip: %s", strerror(errno));
-
-	handle_gossip_msg(state->pps, take(msg));
-}
-
 /* BOLT #2:
  *
  * A receiving node:
@@ -3553,7 +3526,7 @@ static void do_reconnect_dance(struct state *state)
 	do {
 		clean_tmpctx();
 		msg = peer_read(tmpctx, state->pps);
-	} while (handle_peer_gossip_or_error(state->pps,
+	} while (handle_peer_error(state->pps,
 				   &state->channel_id,
 				   msg));
 
@@ -3750,9 +3723,8 @@ static u8 *handle_peer_in(struct state *state)
 		break;
 	}
 
-	/* Handles standard cases, and legal unknown ones. */
+	/* Handles errors. */
-	if (handle_peer_gossip_or_error(state->pps,
+	if (handle_peer_error(state->pps, &state->channel_id, msg))
-					&state->channel_id, msg))
 		return NULL;
 
 	peer_write(state->pps,
@@ -3775,7 +3747,7 @@ int main(int argc, char *argv[])
 {
 	common_setup(argv[0]);
 
-	struct pollfd pollfd[3];
+	struct pollfd pollfd[2];
 	struct state *state = tal(NULL, struct state);
 	struct secret *none;
 	struct fee_states *fee_states;
@@ -3901,9 +3873,9 @@ int main(int argc, char *argv[])
 
 
 
-	/* 3 == peer, 4 == gossipd, 5 = hsmd */
+	/* 3 == peer, 4 = hsmd */
 	state->pps = new_per_peer_state(state);
-	per_peer_state_set_fds(state->pps, 3, 4);
+	per_peer_state_set_fd(state->pps, 3);
 
 	/*~ We need an initial per-commitment point whether we're funding or
 	 * they are, and lightningd has reserved a unique dbid for us already,
@@ -3927,10 +3899,8 @@ int main(int argc, char *argv[])
 	/*~ We manually run a little poll() loop here.  With only three fds */
 	pollfd[0].fd = REQ_FD;
 	pollfd[0].events = POLLIN;
-	pollfd[1].fd = state->pps->gossip_fd;
+	pollfd[1].fd = state->pps->peer_fd;
 	pollfd[1].events = POLLIN;
-	pollfd[2].fd = state->pps->peer_fd;
-	pollfd[2].events = POLLIN;
 
 	/* Do reconnect, if need be */
 	if (state->channel) {
@@ -3946,7 +3916,7 @@ int main(int argc, char *argv[])
 		/*~ If we get a signal which aborts the poll() call, valgrind
 		 * complains about revents being uninitialized.  I'm not sure
 		 * that's correct, but it's easy to be sure. */
-		pollfd[0].revents = pollfd[1].revents = pollfd[2].revents = 0;
+		pollfd[0].revents = pollfd[1].revents = 0;
 
 		poll(pollfd, ARRAY_SIZE(pollfd), -1);
 		/* Subtle: handle_master_in can do its own poll loop, so
@@ -3955,11 +3925,8 @@ int main(int argc, char *argv[])
 		if (pollfd[0].revents & POLLIN)
 			msg = handle_master_in(state);
 		/* Second priority: messages from peer. */
-		else if (pollfd[2].revents & POLLIN)
-			msg = handle_peer_in(state);
-		/* Last priority: chit-chat from gossipd. */
 		else if (pollfd[1].revents & POLLIN)
-			handle_gossip_in(state);
+			msg = handle_peer_in(state);
 
 		/* If we've shutdown, we're done */
 		if (shutdown_complete(state))

openingd/openingd.c

@@ -35,9 +35,9 @@
 #include <wire/peer_wire.h>
 #include <wire/wire_sync.h>
 
-/* stdin == lightningd, 3 == peer, 4 == gossipd, 5 = hsmd */
+/* stdin == lightningd, 3 == peer, 4 = hsmd */
 #define REQ_FD STDIN_FILENO
-#define HSM_FD 5
+#define HSM_FD 4
 
 #if DEVELOPER
 /* If --dev-force-tmp-channel-id is set, it ends up here */
@@ -184,7 +184,6 @@ static u8 *opening_negotiate_msg(const tal_t *ctx, struct state *state,
 	 * form, but we use it in a very limited way. */
 	for (;;) {
 		u8 *msg;
-		bool from_gossipd;
 		char *err;
 		bool warning;
 		struct channel_id actual;
@@ -194,20 +193,7 @@ static u8 *opening_negotiate_msg(const tal_t *ctx, struct state *state,
 		clean_tmpctx();
 
 		/* This helper routine polls both the peer and gossipd. */
-		msg = peer_or_gossip_sync_read(ctx, state->pps, &from_gossipd);
+		msg = peer_read(ctx, state->pps);
-
-		/* Use standard helper for gossip msgs (forwards, if it's an
-		 * error, exits). */
-		if (from_gossipd) {
-			handle_gossip_msg(state->pps, take(msg));
-			continue;
-		}
-
-		/* Some messages go straight to gossipd. */
-		if (is_msg_for_gossipd(msg)) {
-			wire_sync_write(state->pps->gossip_fd, take(msg));
-			continue;
-		}
 
 		/* BOLT #1:
 		 *
@@ -1259,9 +1245,8 @@ static u8 *handle_peer_in(struct state *state)
 	if (t == WIRE_OPEN_CHANNEL)
 		return fundee_channel(state, msg);
 
-	/* Handles standard cases, and legal unknown ones. */
+	/* Handles error cases. */
-	if (handle_peer_gossip_or_error(state->pps,
+	if (handle_peer_error(state->pps, &state->channel_id, msg))
-					&state->channel_id, msg))
 		return NULL;
 
 	extracted = extract_channel_id(msg, &channel_id);
@@ -1285,19 +1270,6 @@ static u8 *handle_peer_in(struct state *state)
 	peer_failed_connection_lost();
 }
 
-/*~ If we see the gossip_fd readable, we read a whole message.  Sure, we might
- * block, but we trust gossipd. */
-static void handle_gossip_in(struct state *state)
-{
-	u8 *msg = wire_sync_read(NULL, state->pps->gossip_fd);
-
-	if (!msg)
-		status_failed(STATUS_FAIL_GOSSIP_IO,
-			      "Reading gossip: %s", strerror(errno));
-
-	handle_gossip_msg(state->pps, take(msg));
-}
-
 /* Memory leak detection is DEVELOPER-only because we go to great lengths to
  * record the backtrace when allocations occur: without that, the leak
  * detection tends to be useless for diagnosing where the leak came from, but
@@ -1393,7 +1365,7 @@ int main(int argc, char *argv[])
 	setup_locale();
 
 	u8 *msg;
-	struct pollfd pollfd[3];
+	struct pollfd pollfd[2];
 	struct state *state = tal(NULL, struct state);
 	struct secret *none;
 	struct channel_id *force_tmp_channel_id;
@@ -1424,9 +1396,9 @@ int main(int argc, char *argv[])
 	dev_force_tmp_channel_id = force_tmp_channel_id;
 #endif
 
-	/* 3 == peer, 4 == gossipd, 5 = hsmd */
+	/* 3 == peer, 4 = hsmd */
 	state->pps = new_per_peer_state(state);
-	per_peer_state_set_fds(state->pps, 3, 4);
+	per_peer_state_set_fd(state->pps, 3);
 
 	/*~ Initially we're not associated with a channel, but
 	 * handle_peer_gossip_or_error compares this. */
@@ -1463,10 +1435,8 @@ int main(int argc, char *argv[])
 	/*~ We manually run a little poll() loop here.  With only three fds */
 	pollfd[0].fd = REQ_FD;
 	pollfd[0].events = POLLIN;
-	pollfd[1].fd = state->pps->gossip_fd;
+	pollfd[1].fd = state->pps->peer_fd;
 	pollfd[1].events = POLLIN;
-	pollfd[2].fd = state->pps->peer_fd;
-	pollfd[2].events = POLLIN;
 
 	/* We exit when we get a conclusion to write to lightningd: either
 	 * opening_funder_reply or opening_fundee. */
@@ -1475,7 +1445,7 @@ int main(int argc, char *argv[])
 		/*~ If we get a signal which aborts the poll() call, valgrind
 		 * complains about revents being uninitialized.  I'm not sure
 		 * that's correct, but it's easy to be sure. */
-		pollfd[0].revents = pollfd[1].revents = pollfd[2].revents = 0;
+		pollfd[0].revents = pollfd[1].revents = 0;
 
 		poll(pollfd, ARRAY_SIZE(pollfd), -1);
 		/* Subtle: handle_master_in can do its own poll loop, so
@@ -1484,22 +1454,19 @@ int main(int argc, char *argv[])
 		if (pollfd[0].revents & POLLIN)
 			msg = handle_master_in(state);
 		/* Second priority: messages from peer. */
-		else if (pollfd[2].revents & POLLIN)
-			msg = handle_peer_in(state);
-		/* Last priority: chit-chat from gossipd. */
 		else if (pollfd[1].revents & POLLIN)
-			handle_gossip_in(state);
+			msg = handle_peer_in(state);
 
 		/* Since we're the top-level event loop, we clean up */
 		clean_tmpctx();
 	}
 
-	/*~ Write message and hand back the peer fd and gossipd fd.  This also
+	/*~ Write message and hand back the peer fd.  This also means that if
-	 * means that if the peer or gossipd wrote us any messages we didn't
+	 * the peer wrote us any messages we didn't read yet, it will simply
-	 * read yet, it will simply be read by the next daemon. */
+	 * be read by the next daemon. */
 	wire_sync_write(REQ_FD, msg);
 	per_peer_state_fdpass_send(REQ_FD, state->pps);
-	status_debug("Sent %s with fds",
+	status_debug("Sent %s with fd",
 		     openingd_wire_name(fromwire_peektype(msg)));
 
 	/* This frees the entire tal tree. */

wallet/test/run-wallet.c

@@ -442,7 +442,7 @@ struct chain_coin_mvt *new_coin_wallet_deposit(const tal_t *ctx UNNEEDED,
 
 { fprintf(stderr, "new_coin_wallet_deposit called!\n"); abort(); }
 /* Generated stub for new_peer_fd */
-struct peer_fd *new_peer_fd(const tal_t *ctx UNNEEDED, int peer_fd UNNEEDED, int gossip_fd UNNEEDED)
+struct peer_fd *new_peer_fd(const tal_t *ctx UNNEEDED, int peer_fd UNNEEDED)
 { fprintf(stderr, "new_peer_fd called!\n"); abort(); }
 /* Generated stub for notify_chain_mvt */
 void notify_chain_mvt(struct lightningd *ld UNNEEDED, const struct chain_coin_mvt *mvt UNNEEDED)