s3:events: Call all ready fd event handlers on each iteration of the main loop

Previously, only one fd handler was being called per main message loop
in all smbd child processes.

In the case where multiple fds are available for reading the fd
corresponding to the event closest to the beginning of the event list
would be run.  Obviously this is arbitrary and could cause unfairness.

Usually, the first event fd is the network socket, meaning heavy load
of client requests can starve out other fd events such as oplock
or notify upcalls from the kernel.

In this patch, I have changed the behavior of run_events() to unset
any fd that it has already called a handler function, as well
as decrement the number of fds that were returned from select().
This allows the caller of run_events() to iterate it, until all
available fds have been handled.

I then changed the main loop in smbd child processes to iterate
run_events().  This way, all available fds are handled on each wake
of select, while still checking for timed or signalled events between
each handler function call.  I also added an explicit check for
EINTR from select(), which previously was masked by the fact that
run_events() would handle any signal event before the return code
was checked.

This required a signature change to run_events() but all other callers
should have no change in their behavior.  I also fixed a bug in
run_events() where it could be called with a selrtn value of -1,
doing unecessary looping through the fd_event list when no fds were
available.

Also, remove the temporary echo handler hack, as all fds should be
treated fairly now.

5 files changed, 38 insertions, 33 deletions

diff --git a/source3/include/event.h b/source3/include/event.h
index 493bc15925..6b41b1d81b 100644
--- a/source3/include/event.h
+++ b/source3/include/event.h
@@ -31,7 +31,7 @@ bool event_add_to_select_args(struct event_context *event_ctx,
 			      fd_set *read_fds, fd_set *write_fds,
 			      struct timeval *timeout, int *maxfd);
 bool run_events(struct event_context *event_ctx,
-		int selrtn, fd_set *read_fds, fd_set *write_fds);
+		int *selrtn, fd_set *read_fds, fd_set *write_fds);
 struct timeval *get_timed_events_timeout(struct event_context *event_ctx,
 					 struct timeval *to_ret);
 void dump_event_list(struct event_context *event_ctx);
diff --git a/source3/lib/events.c b/source3/lib/events.c
index 70a0d7c302..0e127f02dc 100644
--- a/source3/lib/events.c
+++ b/source3/lib/events.c
@@ -67,7 +67,7 @@ bool event_add_to_select_args(struct tevent_context *ev,
 }
 
 bool run_events(struct tevent_context *ev,
-		int selrtn, fd_set *read_fds, fd_set *write_fds)
+		int *selrtn, fd_set *read_fds, fd_set *write_fds)
 {
 	struct tevent_fd *fde;
 	struct timeval now;
@@ -112,7 +112,7 @@ bool run_events(struct tevent_context *ev,
 		return true;
 	}
 
-	if (selrtn == 0) {
+	if (*selrtn <= 0) {
 		/*
 		 * No fd ready
 		 */
@@ -122,8 +122,16 @@ bool run_events(struct tevent_context *ev,
 	for (fde = ev->fd_events; fde; fde = fde->next) {
 		uint16 flags = 0;
 
-		if (FD_ISSET(fde->fd, read_fds)) flags |= EVENT_FD_READ;
-		if (FD_ISSET(fde->fd, write_fds)) flags |= EVENT_FD_WRITE;
+		if (FD_ISSET(fde->fd, read_fds)) {
+			flags |= EVENT_FD_READ;
+			FD_CLR(fde->fd, read_fds);
+			(*selrtn)--;
+		}
+		if (FD_ISSET(fde->fd, write_fds)) {
+			flags |= EVENT_FD_WRITE;
+			FD_CLR(fde->fd, write_fds);
+			(*selrtn)--;
+		}
 
 		if (flags & fde->flags) {
 			fde->handler(ev, fde, flags, fde->private_data);
@@ -162,7 +170,7 @@ static int s3_event_loop_once(struct tevent_context *ev, const char *location)
 	struct timeval now, to;
 	fd_set r_fds, w_fds;
 	int maxfd = 0;
-	int ret;
+	int ret = 0;
 
 	FD_ZERO(&r_fds);
 	FD_ZERO(&w_fds);
@@ -170,7 +178,7 @@ static int s3_event_loop_once(struct tevent_context *ev, const char *location)
 	to.tv_sec = 9999;	/* Max timeout */
 	to.tv_usec = 0;
 
-	if (run_events(ev, 0, NULL, NULL)) {
+	if (run_events(ev, &ret, NULL, NULL)) {
 		return 0;
 	}
 
@@ -189,7 +197,7 @@ static int s3_event_loop_once(struct tevent_context *ev, const char *location)
 		return -1;
 	}
 
-	run_events(ev, ret, &r_fds, &w_fds);
+	run_events(ev, &ret, &r_fds, &w_fds);
 	return 0;
 }
 
diff --git a/source3/nmbd/nmbd_packets.c b/source3/nmbd/nmbd_packets.c
index 5d5a67bf62..c80bac46b5 100644
--- a/source3/nmbd/nmbd_packets.c
+++ b/source3/nmbd/nmbd_packets.c
@@ -1840,7 +1840,7 @@ bool listen_for_packets(bool run_election)
 
 	fd_set r_fds;
 	fd_set w_fds;
-	int selrtn;
+	int selrtn = 0;
 	struct timeval timeout;
 #ifndef SYNC_DNS
 	int dns_fd;
@@ -1867,7 +1867,7 @@ bool listen_for_packets(bool run_election)
 #endif
 
 	/* Process a signal and timer events now... */
-	if (run_events(nmbd_event_context(), 0, NULL, NULL)) {
+	if (run_events(nmbd_event_context(), &selrtn, NULL, NULL)) {
 		return False;
 	}
 
@@ -1889,7 +1889,7 @@ bool listen_for_packets(bool run_election)
 
 	selrtn = sys_select(maxfd+1,&r_fds,&w_fds,NULL,&timeout);
 
-	if (run_events(nmbd_event_context(), selrtn, &r_fds, &w_fds)) {
+	if (run_events(nmbd_event_context(), &selrtn, &r_fds, &w_fds)) {
 		return False;
 	}
 
diff --git a/source3/smbd/process.c b/source3/smbd/process.c
index a484dfd3f2..8d36dd3c89 100644
--- a/source3/smbd/process.c
+++ b/source3/smbd/process.c
@@ -949,7 +949,7 @@ void smbd_setup_sig_hup_handler(struct tevent_context *ev,
 static NTSTATUS smbd_server_connection_loop_once(struct smbd_server_connection *conn)
 {
 	fd_set r_fds, w_fds;
-	int selrtn;
+	int selrtn = 0;
 	struct timeval to;
 	int maxfd = 0;
 
@@ -977,7 +977,7 @@ static NTSTATUS smbd_server_connection_loop_once(struct smbd_server_connection *
 	}
 
 	/* Process a signal and timed events now... */
-	if (run_events(smbd_event_context(), 0, NULL, NULL)) {
+	if (run_events(smbd_event_context(), &selrtn, NULL, NULL)) {
 		return NT_STATUS_RETRY;
 	}
 
@@ -994,26 +994,23 @@ static NTSTATUS smbd_server_connection_loop_once(struct smbd_server_connection *
 
 	/* Check if error */
 	if (selrtn == -1) {
-		/* something is wrong. Maybe the socket is dead? */
-		return map_nt_error_from_unix(errno);
+		if (errno == EINTR)
+			return NT_STATUS_RETRY;
+		else
+			/* Maybe the socket is dead? */
+			return map_nt_error_from_unix(errno);
 	}
 
-        if ((conn->smb1.echo_handler.trusted_fd != -1)
-	    && FD_ISSET(conn->sock, &r_fds)
-	    && FD_ISSET(conn->smb1.echo_handler.trusted_fd, &r_fds)) {
-		/*
-		 * Prefer to read pending requests from the echo handler. To
-		 * quote Jeremy (da70f8ab1): This is a hack of monstrous
-		 * proportions...
-		 */
-		FD_CLR(conn->sock, &r_fds);
-        }
-
-	if (run_events(smbd_event_context(), selrtn, &r_fds, &w_fds)) {
-		return NT_STATUS_RETRY;
-	}
+	/* Process events until all available fds have been handled.
+	 * This allows for fair round-robin handling of all available fds
+	 * on each select() wakeup, while still maintaining responsiveness
+	 * by re-checking for signal and timed events between the handling
+	 * of each ready fd. */
+	do {
+		run_events(smbd_event_context(), &selrtn, &r_fds, &w_fds);
+	} while (selrtn > 0);
 
-	/* Did we timeout ? */
+	/* Processed all fds or timed out */
 	if (selrtn == 0) {
 		return NT_STATUS_RETRY;
 	}
diff --git a/source3/winbindd/winbindd_dual.c b/source3/winbindd/winbindd_dual.c
index a4d8b8ac41..5dfd123b7c 100644
--- a/source3/winbindd/winbindd_dual.c
+++ b/source3/winbindd/winbindd_dual.c
@@ -1374,7 +1374,7 @@ static bool fork_domain_child(struct winbindd_child *child)
 
 	while (1) {
 
-		int ret;
+		int ret = 0;
 		fd_set r_fds;
 		fd_set w_fds;
 		int maxfd;
@@ -1386,7 +1386,7 @@ static bool fork_domain_child(struct winbindd_child *child)
 		int iov_count;
 		NTSTATUS status;
 
-		if (run_events(winbind_event_context(), 0, NULL, NULL)) {
+		if (run_events(winbind_event_context(), &ret, NULL, NULL)) {
 			TALLOC_FREE(frame);
 			continue;
 		}
@@ -1424,7 +1424,7 @@ static bool fork_domain_child(struct winbindd_child *child)
 
 		ret = sys_select(maxfd + 1, &r_fds, &w_fds, NULL, tp);
 
-		if (run_events(winbind_event_context(), ret, &r_fds, &w_fds)) {
+		if (run_events(winbind_event_context(), &ret, &r_fds, &w_fds)) {
 			/* We got a signal - continue. */
 			TALLOC_FREE(frame);
 			continue;