From: Jens Axboe <axboe@fb.com>
Date: Thu, 9 Oct 2014 17:55:16 +0000 (-0600)
Subject: engines/net: add socket buffer window size setting
X-Git-Tag: fio-2.1.14~80
X-Git-Url: https://git.kernel.dk/?a=commitdiff_plain;h=1008602c5af0b098b45020a0ac63408d979a41b0;p=fio.git

engines/net: add socket buffer window size setting

Signed-off-by: Jens Axboe <axboe@fb.com>
---

diff --git a/HOWTO b/HOWTO
index 2cecbbba..f78e47c1 100644
--- a/HOWTO
+++ b/HOWTO
@@ -1672,6 +1672,7 @@ that defines them is selected.
 [net] listen	For TCP network connections, tell fio to listen for incoming
 		connections rather than initiating an outgoing connection. The
 		hostname must be omitted if this option is used.
+
 [net] pingpong	Normaly a network writer will just continue writing data, and
 		a network reader will just consume packages. If pingpong=1
 		is set, a writer will send its normal payload to the reader,
@@ -1684,6 +1685,8 @@ that defines them is selected.
 		single reader when multiple readers are listening to the same
 		address.
 
+[net] window_size	Set the desired socket buffer size for the connection.
+
 [e4defrag] donorname=str
 	        File will be used as a block donor(swap extents between files)
 [e4defrag] inplace=int
diff --git a/configure b/configure
index e3ec2523..3e3c9785 100755
--- a/configure
+++ b/configure
@@ -1074,6 +1074,25 @@ if compile_prog "" "" "TCP_NODELAY"; then
 fi
 echo "TCP_NODELAY                   $tcp_nodelay"
 
+##########################################
+# Check whether we have SO_SNDBUF
+window_size="no"
+cat > $TMPC << EOF
+#include <stdio.h>
+#include <sys/types.h>
+#include <sys/socket.h>
+#include <netinet/tcp.h>
+int main(int argc, char **argv)
+{
+  setsockopt(0, SOL_SOCKET, SO_SNDBUF, NULL, 0);
+  setsockopt(0, SOL_SOCKET, SO_RCVBUF, NULL, 0);
+}
+EOF
+if compile_prog "" "" "SO_SNDBUF"; then
+  window_size="yes"
+fi
+echo "Net engine window_size        $window_size"
+
 ##########################################
 # Check whether we have RLIMIT_MEMLOCK
 rlimit_memlock="no"
@@ -1429,6 +1448,9 @@ fi
 if test "$tcp_nodelay" = "yes" ; then
   output_sym "CONFIG_TCP_NODELAY"
 fi
+if test "$window_size" = "yes" ; then
+  output_sym "CONFIG_NET_WINDOWSIZE"
+fi
 if test "$rlimit_memlock" = "yes" ; then
   output_sym "CONFIG_RLIMIT_MEMLOCK"
 fi
diff --git a/engines/net.c b/engines/net.c
index 61bdfdd1..ac5a93c8 100644
--- a/engines/net.c
+++ b/engines/net.c
@@ -39,6 +39,7 @@ struct netio_options {
 	unsigned int pingpong;
 	unsigned int nodelay;
 	unsigned int ttl;
+	unsigned int window_size;
 	char *intfc;
 };
 
@@ -165,6 +166,18 @@ static struct fio_option options[] = {
 		.category = FIO_OPT_C_ENGINE,
 		.group	= FIO_OPT_G_NETIO,
 	},
+#ifdef CONFIG_NET_WINDOWSIZE
+	{
+		.name	= "window_size",
+		.lname	= "Window Size",
+		.type	= FIO_OPT_INT,
+		.off1	= offsetof(struct netio_options, window_size),
+		.minval	= 0,
+		.help	= "Set socket buffer window size",
+		.category = FIO_OPT_C_ENGINE,
+		.group	= FIO_OPT_G_NETIO,
+	},
+#endif
 	{
 		.name	= NULL,
 	},
@@ -185,6 +198,41 @@ static inline int is_ipv6(struct netio_options *o)
 	return o->proto == FIO_TYPE_UDP_V6 || o->proto == FIO_TYPE_TCP_V6;
 }
 
+static int set_window_size(struct thread_data *td, int fd)
+{
+#ifdef CONFIG_NET_WINDOWSIZE
+	struct netio_options *o = td->eo;
+	unsigned int wss;
+	int snd, rcv, ret;
+
+	if (!o->window_size)
+		return 0;
+
+	rcv = o->listen || o->pingpong;
+	snd = !o->listen || o->pingpong;
+	wss = o->window_size;
+	ret = 0;
+
+	if (rcv) {
+		ret = setsockopt(fd, SOL_SOCKET, SO_RCVBUF, (void *) &wss,
+					sizeof(wss));
+		if (ret < 0)
+			td_verror(td, errno, "rcvbuf window size");
+	}
+	if (snd && !ret) {
+		ret = setsockopt(fd, SOL_SOCKET, SO_SNDBUF, (void *) &wss,
+					sizeof(wss));
+		if (ret < 0)
+			td_verror(td, errno, "sndbuf window size");
+	}
+
+	return ret;
+#else
+	td_verror(td, -EINVAL, "setsockopt window size");
+	return -1;
+#endif
+}
+
 /*
  * Return -1 for error and 'nr events' for a positive number
  * of events
@@ -603,6 +651,11 @@ static int fio_netio_connect(struct thread_data *td, struct fio_file *f)
 	}
 #endif
 
+	if (set_window_size(td, f->fd)) {
+		close(f->fd);
+		return 1;
+	}
+
 	if (is_udp(o)) {
 		if (!fio_netio_is_multicast(td->o.filename))
 			return 0;
@@ -1044,6 +1097,11 @@ static int fio_netio_setup_listen_inet(struct thread_data *td, short port)
 	}
 #endif
 
+	if (set_window_size(td, fd)) {
+		close(fd);
+		return 1;
+	}
+
 	if (td->o.filename) {
 		if (!is_udp(o) || !fio_netio_is_multicast(td->o.filename)) {
 			log_err("fio: hostname not valid for non-multicast inbound network IO\n");
diff --git a/fio.1 b/fio.1
index 91c30748..dc727d36 100644
--- a/fio.1
+++ b/fio.1
@@ -1525,6 +1525,9 @@ completion latency measures how long it took for the other end to receive and
 send back. For UDP multicast traffic pingpong=1 should only be set for a single
 reader when multiple readers are listening to the same address.
 .TP
+.BI (net, window_size) \fR=\fPint
+Set the desired socket buffer size for the connection.
+.TP
 .BI (e4defrag,donorname) \fR=\fPstr
 File will be used as a block donor (swap extents between files)
 .TP