[RFC PATCH v4 0/4] vsock: update tools and error handling

[RFC PATCH v4 0/4] vsock: update tools and error handling

[Arseniy Krasnov]

Patchset consists of two parts:

1) Kernel patch
One patch from Bobby Eshleman. I took single patch from Bobby:
https://lore.kernel.org/lkml/d81818b868216c774613dd03641fcfe63cc55a45
.1660362668.git.bobby.eshleman@bytedance.com/ and use only part for
af_vsock.c, as VMCI and Hyper-V parts were rejected.

I used it, because for SOCK_SEQPACKET big messages handling was broken -
ENOMEM was returned instead of EMSGSIZE. And anyway, current logic which
always replaces any error code returned by transport to ENOMEM looks
strange for me also(for example in EMSGSIZE case it was changed to
ENOMEM).

2) Tool patches
Since there is work on several significant updates for vsock(virtio/
vsock especially): skbuff, DGRAM, zerocopy rx/tx, so I think that this
patchset will be useful.

This patchset updates vsock tests and tools a little bit. First of all
it updates test suite: two new tests are added. One test is reworked
message bound test. Now it is more complex. Instead of sending 1 byte
messages with one MSG_EOR bit, it sends messages of random length(one
half of messages are smaller than page size, second half are bigger)
with random number of MSG_EOR bits set. Receiver also don't know total
number of messages. Message bounds control is maintained by hash sum
of messages length calculation. Second test is for SOCK_SEQPACKET - it
tries to send message with length more than allowed. I think both tests
will be useful for DGRAM support also.

Third thing that this patchset adds is small utility to test vsock
performance for both rx and tx. I think this util could be useful as
'iperf', because:
1) It is small comparing to 'iperf()', so it very easy to add new
   mode or feature to it(especially vsock specific).
2) It is located in kernel source tree, so it could be updated by the
   same patchset which changes related kernel functionality in vsock.

I used this util very often to check performance of my rx zerocopy
support(this tool has rx zerocopy support, but not in this patchset).

Patchset was rebased and tested on skbuff v5 patch from Bobby Eshleman:
https://lore.kernel.org/netdev/20221202173520.10428-1-bobby.eshleman@bytedance.com/

Changelog:
 v3 -> v4:
 - Kernel patch: update commit message by adding error case description
 - Message bounds test:
   - Typo fix: s/contols/controls
   - Fix error output on 'setsockopt()'s
 - vsock_perf:
   - Add 'vsock_perf' target to 'all' in Makefile
   - Fix error output on 'setsockopt()'s
   - Swap sender/receiver roles: now sender does 'connect()' and sends
     data, while receiver accepts connection.
   - Update arguments names: s/mb/bytes, s/so_rcvlowat/rcvlowat
   - Update usage output and description in README

 v2 -> v3:
 - Patches for VMCI and Hyper-V were removed from patchset(commented by
   Vishnu Dasa and Dexuan Cui)
 - In message bounds test hash is computed from data buffer with random
   content(in v2 it was size only). This approach controls both data
   integrity and message bounds.
 - vsock_perf:
   - grammar fixes
   - only long parameters supported(instead of only short)

 v1 -> v2:
 - Three new patches from Bobby Eshleman to kernel part
 - Message bounds test: some refactoring and add comment to describe
   hashing purpose
 - Big message test: check 'errno' for EMSGSIZE and  move new test to
   the end of tests array
 - vsock_perf:
   - update README file
   - add simple usage example to commit message
   - update '-h' (help) output
   - use 'stdout' for output instead of 'stderr'
   - use 'strtol' instead of 'atoi'

Bobby Eshleman(1):
 vsock: return errors other than -ENOMEM to socket

Arseniy Krasnov(3):
 test/vsock: rework message bound test
 test/vsock: add big message test
 test/vsock: vsock_perf utility

 net/vmw_vsock/af_vsock.c         |   3 +-
 tools/testing/vsock/Makefile     |   3 +-
 tools/testing/vsock/README       |  34 +++
 tools/testing/vsock/control.c    |  28 +++
 tools/testing/vsock/control.h    |   2 +
 tools/testing/vsock/util.c       |  13 ++
 tools/testing/vsock/util.h       |   1 +
 tools/testing/vsock/vsock_perf.c | 434 +++++++++++++++++++++++++++++++++++++++
 tools/testing/vsock/vsock_test.c | 197 ++++++++++++++++--
 9 files changed, 698 insertions(+), 17 deletions(-)
-- 
2.25.1

[RFC PATCH v4 1/4] vsock: return errors other than -ENOMEM to socket

[Arseniy Krasnov]

From: Bobby Eshleman <bobby.eshleman@bytedance.com>

This removes behaviour, where error code returned from any transport
was always switched to ENOMEM. For example when user tries to send too
big message via SEQPACKET socket, transport layers return EMSGSIZE, but
this error code will be replaced to ENOMEM and returned to user.

Signed-off-by: Bobby Eshleman <bobby.eshleman@bytedance.com>
Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
---
 net/vmw_vsock/af_vsock.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
index 884eca7f6743..61ddab664c33 100644
--- a/net/vmw_vsock/af_vsock.c
+++ b/net/vmw_vsock/af_vsock.c
@@ -1862,8 +1862,9 @@ static int vsock_connectible_sendmsg(struct socket *sock, struct msghdr *msg,
 			written = transport->stream_enqueue(vsk,
 					msg, len - total_written);
 		}
+
 		if (written < 0) {
-			err = -ENOMEM;
+			err = written;
 			goto out_err;
 		}
 
-- 
2.25.1

[RFC PATCH v4 2/4] test/vsock: rework message bounds test

[Arseniy Krasnov]

This updates message bound test making it more complex. Instead of
sending 1 bytes messages with one MSG_EOR bit, it sends messages of
random length(one half of messages are smaller than page size, second
half are bigger) with random number of MSG_EOR bits set. Receiver
also don't know total number of messages.

Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
---
 tools/testing/vsock/control.c    |  28 +++++++
 tools/testing/vsock/control.h    |   2 +
 tools/testing/vsock/util.c       |  13 ++++
 tools/testing/vsock/util.h       |   1 +
 tools/testing/vsock/vsock_test.c | 128 +++++++++++++++++++++++++++----
 5 files changed, 157 insertions(+), 15 deletions(-)

diff --git a/tools/testing/vsock/control.c b/tools/testing/vsock/control.c
index 4874872fc5a3..d2deb4b15b94 100644
--- a/tools/testing/vsock/control.c
+++ b/tools/testing/vsock/control.c
@@ -141,6 +141,34 @@ void control_writeln(const char *str)
 	timeout_end();
 }
 
+void control_writeulong(unsigned long value)
+{
+	char str[32];
+
+	if (snprintf(str, sizeof(str), "%lu", value) >= sizeof(str)) {
+		perror("snprintf");
+		exit(EXIT_FAILURE);
+	}
+
+	control_writeln(str);
+}
+
+unsigned long control_readulong(void)
+{
+	unsigned long value;
+	char *str;
+
+	str = control_readln();
+
+	if (!str)
+		exit(EXIT_FAILURE);
+
+	value = strtoul(str, NULL, 10);
+	free(str);
+
+	return value;
+}
+
 /* Return the next line from the control socket (without the trailing newline).
  *
  * The program terminates if a timeout occurs.
diff --git a/tools/testing/vsock/control.h b/tools/testing/vsock/control.h
index 51814b4f9ac1..c1f77fdb2c7a 100644
--- a/tools/testing/vsock/control.h
+++ b/tools/testing/vsock/control.h
@@ -9,7 +9,9 @@ void control_init(const char *control_host, const char *control_port,
 void control_cleanup(void);
 void control_writeln(const char *str);
 char *control_readln(void);
+unsigned long control_readulong(void);
 void control_expectln(const char *str);
 bool control_cmpln(char *line, const char *str, bool fail);
+void control_writeulong(unsigned long value);
 
 #endif /* CONTROL_H */
diff --git a/tools/testing/vsock/util.c b/tools/testing/vsock/util.c
index 2acbb7703c6a..01b636d3039a 100644
--- a/tools/testing/vsock/util.c
+++ b/tools/testing/vsock/util.c
@@ -395,3 +395,16 @@ void skip_test(struct test_case *test_cases, size_t test_cases_len,
 
 	test_cases[test_id].skip = true;
 }
+
+unsigned long hash_djb2(const void *data, size_t len)
+{
+	unsigned long hash = 5381;
+	int i = 0;
+
+	while (i < len) {
+		hash = ((hash << 5) + hash) + ((unsigned char *)data)[i];
+		i++;
+	}
+
+	return hash;
+}
diff --git a/tools/testing/vsock/util.h b/tools/testing/vsock/util.h
index a3375ad2fb7f..fb99208a95ea 100644
--- a/tools/testing/vsock/util.h
+++ b/tools/testing/vsock/util.h
@@ -49,4 +49,5 @@ void run_tests(const struct test_case *test_cases,
 void list_tests(const struct test_case *test_cases);
 void skip_test(struct test_case *test_cases, size_t test_cases_len,
 	       const char *test_id_str);
+unsigned long hash_djb2(const void *data, size_t len);
 #endif /* UTIL_H */
diff --git a/tools/testing/vsock/vsock_test.c b/tools/testing/vsock/vsock_test.c
index bb6d691cb30d..26c38ad9d07b 100644
--- a/tools/testing/vsock/vsock_test.c
+++ b/tools/testing/vsock/vsock_test.c
@@ -284,10 +284,14 @@ static void test_stream_msg_peek_server(const struct test_opts *opts)
 	close(fd);
 }
 
-#define MESSAGES_CNT 7
-#define MSG_EOR_IDX (MESSAGES_CNT / 2)
+#define SOCK_BUF_SIZE (2 * 1024 * 1024)
+#define MAX_MSG_SIZE (32 * 1024)
+
 static void test_seqpacket_msg_bounds_client(const struct test_opts *opts)
 {
+	unsigned long curr_hash;
+	int page_size;
+	int msg_count;
 	int fd;
 
 	fd = vsock_seqpacket_connect(opts->peer_cid, 1234);
@@ -296,18 +300,79 @@ static void test_seqpacket_msg_bounds_client(const struct test_opts *opts)
 		exit(EXIT_FAILURE);
 	}
 
-	/* Send several messages, one with MSG_EOR flag */
-	for (int i = 0; i < MESSAGES_CNT; i++)
-		send_byte(fd, 1, (i == MSG_EOR_IDX) ? MSG_EOR : 0);
+	/* Wait, until receiver sets buffer size. */
+	control_expectln("SRVREADY");
+
+	curr_hash = 0;
+	page_size = getpagesize();
+	msg_count = SOCK_BUF_SIZE / MAX_MSG_SIZE;
+
+	for (int i = 0; i < msg_count; i++) {
+		ssize_t send_size;
+		size_t buf_size;
+		int flags;
+		void *buf;
+
+		/* Use "small" buffers and "big" buffers. */
+		if (i & 1)
+			buf_size = page_size +
+					(rand() % (MAX_MSG_SIZE - page_size));
+		else
+			buf_size = 1 + (rand() % page_size);
+
+		buf = malloc(buf_size);
+
+		if (!buf) {
+			perror("malloc");
+			exit(EXIT_FAILURE);
+		}
+
+		memset(buf, rand() & 0xff, buf_size);
+		/* Set at least one MSG_EOR + some random. */
+		if (i == (msg_count / 2) || (rand() & 1)) {
+			flags = MSG_EOR;
+			curr_hash++;
+		} else {
+			flags = 0;
+		}
+
+		send_size = send(fd, buf, buf_size, flags);
+
+		if (send_size < 0) {
+			perror("send");
+			exit(EXIT_FAILURE);
+		}
+
+		if (send_size != buf_size) {
+			fprintf(stderr, "Invalid send size\n");
+			exit(EXIT_FAILURE);
+		}
+
+		/*
+		 * Hash sum is computed at both client and server in
+		 * the same way:
+		 * H += hash('message data')
+		 * Such hash "controls" both data integrity and message
+		 * bounds. After data exchange, both sums are compared
+		 * using control socket, and if message bounds wasn't
+		 * broken - two values must be equal.
+		 */
+		curr_hash += hash_djb2(buf, buf_size);
+		free(buf);
+	}
 
 	control_writeln("SENDDONE");
+	control_writeulong(curr_hash);
 	close(fd);
 }
 
 static void test_seqpacket_msg_bounds_server(const struct test_opts *opts)
 {
+	unsigned long sock_buf_size;
+	unsigned long remote_hash;
+	unsigned long curr_hash;
 	int fd;
-	char buf[16];
+	char buf[MAX_MSG_SIZE];
 	struct msghdr msg = {0};
 	struct iovec iov = {0};
 
@@ -317,25 +382,57 @@ static void test_seqpacket_msg_bounds_server(const struct test_opts *opts)
 		exit(EXIT_FAILURE);
 	}
 
+	sock_buf_size = SOCK_BUF_SIZE;
+
+	if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_MAX_SIZE,
+		       &sock_buf_size, sizeof(sock_buf_size))) {
+		perror("setsockopt(SO_VM_SOCKETS_BUFFER_MAX_SIZE)");
+		exit(EXIT_FAILURE);
+	}
+
+	if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_SIZE,
+		       &sock_buf_size, sizeof(sock_buf_size))) {
+		perror("setsockopt(SO_VM_SOCKETS_BUFFER_SIZE)");
+		exit(EXIT_FAILURE);
+	}
+
+	/* Ready to receive data. */
+	control_writeln("SRVREADY");
+	/* Wait, until peer sends whole data. */
 	control_expectln("SENDDONE");
 	iov.iov_base = buf;
 	iov.iov_len = sizeof(buf);
 	msg.msg_iov = &iov;
 	msg.msg_iovlen = 1;
 
-	for (int i = 0; i < MESSAGES_CNT; i++) {
-		if (recvmsg(fd, &msg, 0) != 1) {
-			perror("message bound violated");
-			exit(EXIT_FAILURE);
-		}
+	curr_hash = 0;
 
-		if ((i == MSG_EOR_IDX) ^ !!(msg.msg_flags & MSG_EOR)) {
-			perror("MSG_EOR");
+	while (1) {
+		ssize_t recv_size;
+
+		recv_size = recvmsg(fd, &msg, 0);
+
+		if (!recv_size)
+			break;
+
+		if (recv_size < 0) {
+			perror("recvmsg");
 			exit(EXIT_FAILURE);
 		}
+
+		if (msg.msg_flags & MSG_EOR)
+			curr_hash++;
+
+		curr_hash += hash_djb2(msg.msg_iov[0].iov_base, recv_size);
 	}
 
 	close(fd);
+	remote_hash = control_readulong();
+
+	if (curr_hash != remote_hash) {
+		fprintf(stderr, "Message bounds broken\n");
+		exit(EXIT_FAILURE);
+	}
 }
 
 #define MESSAGE_TRUNC_SZ 32
@@ -427,7 +524,7 @@ static void test_seqpacket_timeout_client(const struct test_opts *opts)
 	tv.tv_usec = 0;
 
 	if (setsockopt(fd, SOL_SOCKET, SO_RCVTIMEO, (void *)&tv, sizeof(tv)) == -1) {
-		perror("setsockopt 'SO_RCVTIMEO'");
+		perror("setsockopt(SO_RCVTIMEO)");
 		exit(EXIT_FAILURE);
 	}
 
@@ -644,7 +741,7 @@ static void test_stream_poll_rcvlowat_client(const struct test_opts *opts)
 
 	if (setsockopt(fd, SOL_SOCKET, SO_RCVLOWAT,
 		       &lowat_val, sizeof(lowat_val))) {
-		perror("setsockopt");
+		perror("setsockopt(SO_RCVLOWAT)");
 		exit(EXIT_FAILURE);
 	}
 
@@ -837,6 +934,7 @@ int main(int argc, char **argv)
 		.peer_cid = VMADDR_CID_ANY,
 	};
 
+	srand(time(NULL));
 	init_signals();
 
 	for (;;) {
-- 
2.25.1

[RFC PATCH v4 3/4] test/vsock: add big message test

[Arseniy Krasnov]

This adds test for sending message, bigger than peer's buffer size.
For SOCK_SEQPACKET socket it must fail, as this type of socket has
message size limit.

Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
---
 tools/testing/vsock/vsock_test.c | 69 ++++++++++++++++++++++++++++++++
 1 file changed, 69 insertions(+)

diff --git a/tools/testing/vsock/vsock_test.c b/tools/testing/vsock/vsock_test.c
index 26c38ad9d07b..67e9f9df3a8c 100644
--- a/tools/testing/vsock/vsock_test.c
+++ b/tools/testing/vsock/vsock_test.c
@@ -569,6 +569,70 @@ static void test_seqpacket_timeout_server(const struct test_opts *opts)
 	close(fd);
 }
 
+static void test_seqpacket_bigmsg_client(const struct test_opts *opts)
+{
+	unsigned long sock_buf_size;
+	ssize_t send_size;
+	socklen_t len;
+	void *data;
+	int fd;
+
+	len = sizeof(sock_buf_size);
+
+	fd = vsock_seqpacket_connect(opts->peer_cid, 1234);
+	if (fd < 0) {
+		perror("connect");
+		exit(EXIT_FAILURE);
+	}
+
+	if (getsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_SIZE,
+		       &sock_buf_size, &len)) {
+		perror("getsockopt");
+		exit(EXIT_FAILURE);
+	}
+
+	sock_buf_size++;
+
+	data = malloc(sock_buf_size);
+	if (!data) {
+		perror("malloc");
+		exit(EXIT_FAILURE);
+	}
+
+	send_size = send(fd, data, sock_buf_size, 0);
+	if (send_size != -1) {
+		fprintf(stderr, "expected 'send(2)' failure, got %zi\n",
+			send_size);
+		exit(EXIT_FAILURE);
+	}
+
+	if (errno != EMSGSIZE) {
+		fprintf(stderr, "expected EMSGSIZE in 'errno', got %i\n",
+			errno);
+		exit(EXIT_FAILURE);
+	}
+
+	control_writeln("CLISENT");
+
+	free(data);
+	close(fd);
+}
+
+static void test_seqpacket_bigmsg_server(const struct test_opts *opts)
+{
+	int fd;
+
+	fd = vsock_seqpacket_accept(VMADDR_CID_ANY, 1234, NULL);
+	if (fd < 0) {
+		perror("accept");
+		exit(EXIT_FAILURE);
+	}
+
+	control_expectln("CLISENT");
+
+	close(fd);
+}
+
 #define BUF_PATTERN_1 'a'
 #define BUF_PATTERN_2 'b'
 
@@ -851,6 +915,11 @@ static struct test_case test_cases[] = {
 		.run_client = test_stream_poll_rcvlowat_client,
 		.run_server = test_stream_poll_rcvlowat_server,
 	},
+	{
+		.name = "SOCK_SEQPACKET big message",
+		.run_client = test_seqpacket_bigmsg_client,
+		.run_server = test_seqpacket_bigmsg_server,
+	},
 	{},
 };
 
-- 
2.25.1

[RFC PATCH v4 4/4] test/vsock: vsock_perf utility

[Arseniy Krasnov]

This adds utility to check vsock rx/tx performance.

Usage as sender:
./vsock_perf --sender <cid> --port <port> --bytes <bytes to send)
Usage as receiver:
./vsock_perf --port <port> --rcvlowat <SO_RCVLOWAT>

Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
---
 tools/testing/vsock/Makefile     |   3 +-
 tools/testing/vsock/README       |  34 +++
 tools/testing/vsock/vsock_perf.c | 434 +++++++++++++++++++++++++++++++
 3 files changed, 470 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/vsock/vsock_perf.c

diff --git a/tools/testing/vsock/Makefile b/tools/testing/vsock/Makefile
index f8293c6910c9..43a254f0e14d 100644
--- a/tools/testing/vsock/Makefile
+++ b/tools/testing/vsock/Makefile
@@ -1,8 +1,9 @@
 # SPDX-License-Identifier: GPL-2.0-only
-all: test
+all: test vsock_perf
 test: vsock_test vsock_diag_test
 vsock_test: vsock_test.o timeout.o control.o util.o
 vsock_diag_test: vsock_diag_test.o timeout.o control.o util.o
+vsock_perf: vsock_perf.o
 
 CFLAGS += -g -O2 -Werror -Wall -I. -I../../include -I../../../usr/include -Wno-pointer-sign -fno-strict-overflow -fno-strict-aliasing -fno-common -MMD -U_FORTIFY_SOURCE -D_GNU_SOURCE
 .PHONY: all test clean
diff --git a/tools/testing/vsock/README b/tools/testing/vsock/README
index 4d5045e7d2c3..e6f6735bba05 100644
--- a/tools/testing/vsock/README
+++ b/tools/testing/vsock/README
@@ -35,3 +35,37 @@ Invoke test binaries in both directions as follows:
                        --control-port=$GUEST_IP \
                        --control-port=1234 \
                        --peer-cid=3
+
+vsock_perf utility
+-------------------
+'vsock_perf' is a simple tool to measure vsock performance. It works in
+sender/receiver modes: sender connect to peer at the specified port and
+starts data transmission to the receiver. After data processing is done,
+it prints several metrics(see below).
+
+Usage:
+# run as sender
+# connect to CID 2, port 1234, send 1G of data, tx buf size is 1M
+./vsock_perf --sender 2 --port 1234 --bytes 1G --buf-size 1M
+
+Output:
+tx performance: A Gb/s
+
+Output explanation:
+A is calculated as "number of bytes to send" / "time in tx loop"
+
+# run as receiver
+# listen port 1234, rx buf size is 1M, socket buf size is 1G, SO_RCVLOWAT is 64K
+./vsock_perf --port 1234 --buf-size 1M --vsk-size 1G --rcvlowat 64K
+
+Output:
+rx performance: A Gb/s
+total in 'read()': B sec
+POLLIN wakeups: C
+average in 'read()': D ns
+
+Output explanation:
+A is calculated as "number of received bytes" / "time in rx loop".
+B is time, spent in 'read()' system call(excluding 'poll()')
+C is number of 'poll()' wake ups with POLLIN bit set.
+D is B / C, e.g. average amount of time, spent in single 'read()'.
diff --git a/tools/testing/vsock/vsock_perf.c b/tools/testing/vsock/vsock_perf.c
new file mode 100644
index 000000000000..248fc3285d5c
--- /dev/null
+++ b/tools/testing/vsock/vsock_perf.c
@@ -0,0 +1,434 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * vsock_perf - benchmark utility for vsock.
+ *
+ * Copyright (C) 2022 SberDevices.
+ *
+ * Author: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
+ */
+#include <getopt.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <string.h>
+#include <errno.h>
+#include <unistd.h>
+#include <time.h>
+#include <stdint.h>
+#include <poll.h>
+#include <sys/socket.h>
+#include <linux/vm_sockets.h>
+
+#define DEFAULT_BUF_SIZE_BYTES	(128 * 1024)
+#define DEFAULT_TO_SEND_BYTES	(64 * 1024)
+#define DEFAULT_VSOCK_BUF_BYTES (256 * 1024)
+#define DEFAULT_RCVLOWAT_BYTES	1
+#define DEFAULT_PORT		1234
+
+#define BYTES_PER_GB		(1024 * 1024 * 1024ULL)
+#define NSEC_PER_SEC		(1000000000ULL)
+
+static unsigned int port = DEFAULT_PORT;
+static unsigned long buf_size_bytes = DEFAULT_BUF_SIZE_BYTES;
+static unsigned long vsock_buf_bytes = DEFAULT_VSOCK_BUF_BYTES;
+
+static inline time_t current_nsec(void)
+{
+	struct timespec ts;
+
+	if (clock_gettime(CLOCK_REALTIME, &ts)) {
+		perror("clock_gettime");
+		exit(EXIT_FAILURE);
+	}
+
+	return (ts.tv_sec * NSEC_PER_SEC) + ts.tv_nsec;
+}
+
+/* From lib/cmdline.c. */
+static unsigned long memparse(const char *ptr)
+{
+	char *endptr;
+
+	unsigned long long ret = strtoull(ptr, &endptr, 0);
+
+	switch (*endptr) {
+	case 'E':
+	case 'e':
+		ret <<= 10;
+	case 'P':
+	case 'p':
+		ret <<= 10;
+	case 'T':
+	case 't':
+		ret <<= 10;
+	case 'G':
+	case 'g':
+		ret <<= 10;
+	case 'M':
+	case 'm':
+		ret <<= 10;
+	case 'K':
+	case 'k':
+		ret <<= 10;
+		endptr++;
+	default:
+		break;
+	}
+
+	return ret;
+}
+
+static void vsock_increase_buf_size(int fd)
+{
+	if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_MAX_SIZE,
+		       &vsock_buf_bytes, sizeof(vsock_buf_bytes))) {
+		perror("setsockopt(SO_VM_SOCKETS_BUFFER_MAX_SIZE)");
+		exit(EXIT_FAILURE);
+	}
+
+	if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_SIZE,
+		       &vsock_buf_bytes, sizeof(vsock_buf_bytes))) {
+		perror("setsockopt(SO_VM_SOCKETS_BUFFER_SIZE)");
+		exit(EXIT_FAILURE);
+	}
+}
+
+static int vsock_connect(unsigned int cid, unsigned int port)
+{
+	union {
+		struct sockaddr sa;
+		struct sockaddr_vm svm;
+	} addr = {
+		.svm = {
+			.svm_family = AF_VSOCK,
+			.svm_port = port,
+			.svm_cid = cid,
+		},
+	};
+	int fd;
+
+	fd = socket(AF_VSOCK, SOCK_STREAM, 0);
+
+	if (fd < 0) {
+		perror("socket");
+		return -1;
+	}
+
+	if (connect(fd, &addr.sa, sizeof(addr.svm)) < 0) {
+		perror("connect");
+		close(fd);
+		return -1;
+	}
+
+	return fd;
+}
+
+static float get_gbps(unsigned long bytes, time_t ns_delta)
+{
+	return ((float)bytes / BYTES_PER_GB) /
+	       ((float)ns_delta / NSEC_PER_SEC);
+}
+
+static void run_receiver(unsigned long rcvlowat_bytes)
+{
+	unsigned int read_cnt;
+	time_t rx_begin_ns;
+	time_t in_read_ns;
+	size_t total_recv;
+	int client_fd;
+	char *data;
+	int fd;
+	union {
+		struct sockaddr sa;
+		struct sockaddr_vm svm;
+	} addr = {
+		.svm = {
+			.svm_family = AF_VSOCK,
+			.svm_port = port,
+			.svm_cid = VMADDR_CID_ANY,
+		},
+	};
+	union {
+		struct sockaddr sa;
+		struct sockaddr_vm svm;
+	} clientaddr;
+
+	socklen_t clientaddr_len = sizeof(clientaddr.svm);
+
+	printf("Run as receiver\n");
+	printf("Listen port %u\n", port);
+	printf("RX buffer %lu bytes\n", buf_size_bytes);
+	printf("vsock buffer %lu bytes\n", vsock_buf_bytes);
+	printf("SO_RCVLOWAT %lu bytes\n", rcvlowat_bytes);
+
+	fd = socket(AF_VSOCK, SOCK_STREAM, 0);
+
+	if (fd < 0) {
+		perror("socket");
+		exit(EXIT_FAILURE);
+	}
+
+	if (bind(fd, &addr.sa, sizeof(addr.svm)) < 0) {
+		perror("bind");
+		exit(EXIT_FAILURE);
+	}
+
+	if (listen(fd, 1) < 0) {
+		perror("listen");
+		exit(EXIT_FAILURE);
+	}
+
+	client_fd = accept(fd, &clientaddr.sa, &clientaddr_len);
+
+	if (client_fd < 0) {
+		perror("accept");
+		exit(EXIT_FAILURE);
+	}
+
+	vsock_increase_buf_size(client_fd);
+
+	if (setsockopt(client_fd, SOL_SOCKET, SO_RCVLOWAT,
+		       &rcvlowat_bytes,
+		       sizeof(rcvlowat_bytes))) {
+		perror("setsockopt(SO_RCVLOWAT)");
+		exit(EXIT_FAILURE);
+	}
+
+	data = malloc(buf_size_bytes);
+
+	if (!data) {
+		printf("malloc failed\n");
+		exit(EXIT_FAILURE);
+	}
+
+	read_cnt = 0;
+	in_read_ns = 0;
+	total_recv = 0;
+	rx_begin_ns = current_nsec();
+
+	while (1) {
+		struct pollfd fds = { 0 };
+
+		fds.fd = client_fd;
+		fds.events = POLLIN | POLLERR |
+			     POLLHUP | POLLRDHUP;
+
+		if (poll(&fds, 1, -1) < 0) {
+			perror("poll");
+			exit(EXIT_FAILURE);
+		}
+
+		if (fds.revents & POLLERR) {
+			printf("'poll()' error\n");
+			exit(EXIT_FAILURE);
+		}
+
+		if (fds.revents & POLLIN) {
+			ssize_t bytes_read;
+			time_t t;
+
+			t = current_nsec();
+			bytes_read = read(fds.fd, data, buf_size_bytes);
+			in_read_ns += (current_nsec() - t);
+			read_cnt++;
+
+			if (!bytes_read)
+				break;
+
+			if (bytes_read < 0) {
+				perror("read");
+				exit(EXIT_FAILURE);
+			}
+
+			total_recv += bytes_read;
+		}
+
+		if (fds.revents & (POLLHUP | POLLRDHUP))
+			break;
+	}
+
+	printf("rx performance: %f Gb/s\n",
+	       get_gbps(total_recv, current_nsec() - rx_begin_ns));
+	printf("total time in 'read()': %f sec\n", (float)in_read_ns / NSEC_PER_SEC);
+	printf("POLLIN wakeups: %i\n", read_cnt);
+	printf("average time in 'read()': %f ns\n", (float)in_read_ns / read_cnt);
+
+	free(data);
+	close(client_fd);
+	close(fd);
+}
+
+static void run_sender(int peer_cid, unsigned long to_send_bytes)
+{
+	time_t tx_begin_ns;
+	size_t total_send;
+	void *data;
+	int fd;
+
+	printf("Run as sender\n");
+	printf("Connect to %i:%u\n", peer_cid, port);
+	printf("Send %lu bytes\n", to_send_bytes);
+	printf("TX buffer %lu bytes\n", buf_size_bytes);
+
+	fd = vsock_connect(peer_cid, port);
+
+	if (fd < 0)
+		exit(EXIT_FAILURE);
+
+	data = malloc(buf_size_bytes);
+
+	if (!data) {
+		printf("malloc failed\n");
+		exit(EXIT_FAILURE);
+	}
+
+	memset(data, 0, buf_size_bytes);
+	total_send = 0;
+	tx_begin_ns = current_nsec();
+
+	while (total_send < to_send_bytes) {
+		ssize_t sent;
+
+		sent = write(fd, data, buf_size_bytes);
+
+		if (sent <= 0) {
+			perror("write");
+			exit(EXIT_FAILURE);
+		}
+
+		total_send += sent;
+	}
+
+	printf("tx performance: %f Gb/s\n",
+	       get_gbps(total_send, current_nsec() - tx_begin_ns));
+
+	close(fd);
+	free(data);
+}
+
+static const char optstring[] = "";
+static const struct option longopts[] = {
+	{
+		.name = "help",
+		.has_arg = no_argument,
+		.val = 'H',
+	},
+	{
+		.name = "sender",
+		.has_arg = required_argument,
+		.val = 'S',
+	},
+	{
+		.name = "port",
+		.has_arg = required_argument,
+		.val = 'P',
+	},
+	{
+		.name = "bytes",
+		.has_arg = required_argument,
+		.val = 'M',
+	},
+	{
+		.name = "buf-size",
+		.has_arg = required_argument,
+		.val = 'B',
+	},
+	{
+		.name = "vsk-size",
+		.has_arg = required_argument,
+		.val = 'V',
+	},
+	{
+		.name = "rcvlowat",
+		.has_arg = required_argument,
+		.val = 'R',
+	},
+	{},
+};
+
+static void usage(void)
+{
+	printf("Usage: ./vsock_perf [--help] [options]\n"
+	       "\n"
+	       "This is benchmarking utility, to test vsock performance.\n"
+	       "It runs in two modes: sender or receiver. In sender mode, it\n"
+	       "connects to the specified CID and starts data transmission.\n"
+	       "\n"
+	       "Options:\n"
+	       "  --help			This message\n"
+	       "  --sender   <cid>		Sender mode (receiver default)\n"
+	       "                                <cid> of the receiver to connect to\n"
+	       "  --port     <port>		Port (default %d)\n"
+	       "  --bytes    <bytes>KMG		Bytes to send (default %d)\n"
+	       "  --buf-size <bytes>KMG		Data buffer size (default %d). In sender mode\n"
+	       "                                it is the buffer size, passed to 'write()'. In\n"
+	       "                                receiver mode it is the buffer size passed to 'read()'.\n"
+	       "  --vsk-size <bytes>KMG		Socket buffer size (default %d)\n"
+	       "  --rcvlowat <bytes>KMG		SO_RCVLOWAT value (default %d)\n"
+	       "\n", DEFAULT_PORT, DEFAULT_TO_SEND_BYTES,
+	       DEFAULT_BUF_SIZE_BYTES, DEFAULT_VSOCK_BUF_BYTES,
+	       DEFAULT_RCVLOWAT_BYTES);
+	exit(EXIT_FAILURE);
+}
+
+static long strtolx(const char *arg)
+{
+	long value;
+	char *end;
+
+	value = strtol(arg, &end, 10);
+
+	if (end != arg + strlen(arg))
+		usage();
+
+	return value;
+}
+
+int main(int argc, char **argv)
+{
+	unsigned long to_send_bytes = DEFAULT_TO_SEND_BYTES;
+	unsigned long rcvlowat_bytes = DEFAULT_RCVLOWAT_BYTES;
+	int peer_cid = -1;
+	bool sender = false;
+
+	while (1) {
+		int opt = getopt_long(argc, argv, optstring, longopts, NULL);
+
+		if (opt == -1)
+			break;
+
+		switch (opt) {
+		case 'V': /* Peer buffer size. */
+			vsock_buf_bytes = memparse(optarg);
+			break;
+		case 'R': /* SO_RCVLOWAT value. */
+			rcvlowat_bytes = memparse(optarg);
+			break;
+		case 'P': /* Port to connect to. */
+			port = strtolx(optarg);
+			break;
+		case 'M': /* Bytes to send. */
+			to_send_bytes = memparse(optarg);
+			break;
+		case 'B': /* Size of rx/tx buffer. */
+			buf_size_bytes = memparse(optarg);
+			break;
+		case 'S': /* Sender mode. CID to connect to. */
+			peer_cid = strtolx(optarg);
+			sender = true;
+			break;
+		case 'H': /* Help. */
+			usage();
+			break;
+		default:
+			usage();
+		}
+	}
+
+	if (!sender)
+		run_receiver(rcvlowat_bytes);
+	else
+		run_sender(peer_cid, to_send_bytes);
+
+	return 0;
+}
-- 
2.25.1

Re: [RFC PATCH v4 1/4] vsock: return errors other than -ENOMEM to socket

[Stefano Garzarella]

On Tue, Dec 06, 2022 at 08:49:19PM +0000, Arseniy Krasnov wrote:
>From: Bobby Eshleman <bobby.eshleman@bytedance.com>
>
>This removes behaviour, where error code returned from any transport
>was always switched to ENOMEM. For example when user tries to send too
>big message via SEQPACKET socket, transport layers return EMSGSIZE, but
>this error code will be replaced to ENOMEM and returned to user.

Just a minor thing here, I would write
"this error code was always replaced with ENOMEM and returned to user"

To make it clearer that it was the previous behavior.

Anyway, the patch LGTM:

Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>

>
>Signed-off-by: Bobby Eshleman <bobby.eshleman@bytedance.com>
>Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>---
> net/vmw_vsock/af_vsock.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
>diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
>index 884eca7f6743..61ddab664c33 100644
>--- a/net/vmw_vsock/af_vsock.c
>+++ b/net/vmw_vsock/af_vsock.c
>@@ -1862,8 +1862,9 @@ static int vsock_connectible_sendmsg(struct socket *sock, struct msghdr *msg,
> 			written = transport->stream_enqueue(vsk,
> 					msg, len - total_written);
> 		}
>+
> 		if (written < 0) {
>-			err = -ENOMEM;
>+			err = written;
> 			goto out_err;
> 		}
>
>-- 
>2.25.1

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [RFC PATCH v4 1/4] vsock: return errors other than -ENOMEM to socket

[Stefano Garzarella]

On Tue, Dec 06, 2022 at 08:49:19PM +0000, Arseniy Krasnov wrote:
>From: Bobby Eshleman <bobby.eshleman@bytedance.com>
>
>This removes behaviour, where error code returned from any transport
>was always switched to ENOMEM. For example when user tries to send too
>big message via SEQPACKET socket, transport layers return EMSGSIZE, but
>this error code will be replaced to ENOMEM and returned to user.

Just a minor thing here, I would write
"this error code was always replaced with ENOMEM and returned to user"

To make it clearer that it was the previous behavior.

Anyway, the patch LGTM:

Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>

>
>Signed-off-by: Bobby Eshleman <bobby.eshleman@bytedance.com>
>Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>---
> net/vmw_vsock/af_vsock.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
>diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
>index 884eca7f6743..61ddab664c33 100644
>--- a/net/vmw_vsock/af_vsock.c
>+++ b/net/vmw_vsock/af_vsock.c
>@@ -1862,8 +1862,9 @@ static int vsock_connectible_sendmsg(struct socket *sock, struct msghdr *msg,
> 			written = transport->stream_enqueue(vsk,
> 					msg, len - total_written);
> 		}
>+
> 		if (written < 0) {
>-			err = -ENOMEM;
>+			err = written;
> 			goto out_err;
> 		}
>
>-- 
>2.25.1

Re: [RFC PATCH v4 2/4] test/vsock: rework message bounds test

[Stefano Garzarella]

On Tue, Dec 06, 2022 at 08:50:55PM +0000, Arseniy Krasnov wrote:
>This updates message bound test making it more complex. Instead of
>sending 1 bytes messages with one MSG_EOR bit, it sends messages of
>random length(one half of messages are smaller than page size, second
>half are bigger) with random number of MSG_EOR bits set. Receiver
>also don't know total number of messages.
>
>Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>---
> tools/testing/vsock/control.c    |  28 +++++++
> tools/testing/vsock/control.h    |   2 +
> tools/testing/vsock/util.c       |  13 ++++
> tools/testing/vsock/util.h       |   1 +
> tools/testing/vsock/vsock_test.c | 128 +++++++++++++++++++++++++++----
> 5 files changed, 157 insertions(+), 15 deletions(-)

Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [RFC PATCH v4 2/4] test/vsock: rework message bounds test

[Stefano Garzarella]

On Tue, Dec 06, 2022 at 08:50:55PM +0000, Arseniy Krasnov wrote:
>This updates message bound test making it more complex. Instead of
>sending 1 bytes messages with one MSG_EOR bit, it sends messages of
>random length(one half of messages are smaller than page size, second
>half are bigger) with random number of MSG_EOR bits set. Receiver
>also don't know total number of messages.
>
>Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>---
> tools/testing/vsock/control.c    |  28 +++++++
> tools/testing/vsock/control.h    |   2 +
> tools/testing/vsock/util.c       |  13 ++++
> tools/testing/vsock/util.h       |   1 +
> tools/testing/vsock/vsock_test.c | 128 +++++++++++++++++++++++++++----
> 5 files changed, 157 insertions(+), 15 deletions(-)

Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>

Re: [RFC PATCH v4 4/4] test/vsock: vsock_perf utility

[Stefano Garzarella]

On Tue, Dec 06, 2022 at 08:54:28PM +0000, Arseniy Krasnov wrote:
>This adds utility to check vsock rx/tx performance.
>
>Usage as sender:
>./vsock_perf --sender <cid> --port <port> --bytes <bytes to send)

Little typo "<bytes to send>". I mean replace `)` with `>`

>Usage as receiver:
>./vsock_perf --port <port> --rcvlowat <SO_RCVLOWAT>
>
>Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>---
> tools/testing/vsock/Makefile     |   3 +-
> tools/testing/vsock/README       |  34 +++
> tools/testing/vsock/vsock_perf.c | 434 +++++++++++++++++++++++++++++++
> 3 files changed, 470 insertions(+), 1 deletion(-)
> create mode 100644 tools/testing/vsock/vsock_perf.c
>
>diff --git a/tools/testing/vsock/Makefile b/tools/testing/vsock/Makefile
>index f8293c6910c9..43a254f0e14d 100644
>--- a/tools/testing/vsock/Makefile
>+++ b/tools/testing/vsock/Makefile
>@@ -1,8 +1,9 @@
> # SPDX-License-Identifier: GPL-2.0-only
>-all: test
>+all: test vsock_perf
> test: vsock_test vsock_diag_test
> vsock_test: vsock_test.o timeout.o control.o util.o
> vsock_diag_test: vsock_diag_test.o timeout.o control.o util.o
>+vsock_perf: vsock_perf.o
>
> CFLAGS += -g -O2 -Werror -Wall -I. -I../../include -I../../../usr/include -Wno-pointer-sign -fno-strict-overflow -fno-strict-aliasing -fno-common -MMD -U_FORTIFY_SOURCE -D_GNU_SOURCE
> .PHONY: all test clean
>diff --git a/tools/testing/vsock/README b/tools/testing/vsock/README
>index 4d5045e7d2c3..e6f6735bba05 100644
>--- a/tools/testing/vsock/README
>+++ b/tools/testing/vsock/README
>@@ -35,3 +35,37 @@ Invoke test binaries in both directions as follows:
>                        --control-port=$GUEST_IP \
>                        --control-port=1234 \
>                        --peer-cid=3
>+
>+vsock_perf utility
>+-------------------
>+'vsock_perf' is a simple tool to measure vsock performance. It works in
>+sender/receiver modes: sender connect to peer at the specified port and
>+starts data transmission to the receiver. After data processing is done,
>+it prints several metrics(see below).
>+
>+Usage:
>+# run as sender
>+# connect to CID 2, port 1234, send 1G of data, tx buf size is 1M
>+./vsock_perf --sender 2 --port 1234 --bytes 1G --buf-size 1M
>+
>+Output:
>+tx performance: A Gb/s
>+
>+Output explanation:
>+A is calculated as "number of bytes to send" / "time in tx loop"
>+
>+# run as receiver
>+# listen port 1234, rx buf size is 1M, socket buf size is 1G, SO_RCVLOWAT is 64K
>+./vsock_perf --port 1234 --buf-size 1M --vsk-size 1G --rcvlowat 64K
>+
>+Output:
>+rx performance: A Gb/s
>+total in 'read()': B sec
>+POLLIN wakeups: C
>+average in 'read()': D ns
>+
>+Output explanation:
>+A is calculated as "number of received bytes" / "time in rx loop".
>+B is time, spent in 'read()' system call(excluding 'poll()')
>+C is number of 'poll()' wake ups with POLLIN bit set.
>+D is B / C, e.g. average amount of time, spent in single 'read()'.
>diff --git a/tools/testing/vsock/vsock_perf.c b/tools/testing/vsock/vsock_perf.c
>new file mode 100644
>index 000000000000..248fc3285d5c
>--- /dev/null
>+++ b/tools/testing/vsock/vsock_perf.c
>@@ -0,0 +1,434 @@
>+// SPDX-License-Identifier: GPL-2.0-only
>+/*
>+ * vsock_perf - benchmark utility for vsock.
>+ *
>+ * Copyright (C) 2022 SberDevices.
>+ *
>+ * Author: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>+ */
>+#include <getopt.h>
>+#include <stdio.h>
>+#include <stdlib.h>
>+#include <stdbool.h>
>+#include <string.h>
>+#include <errno.h>
>+#include <unistd.h>
>+#include <time.h>
>+#include <stdint.h>
>+#include <poll.h>
>+#include <sys/socket.h>
>+#include <linux/vm_sockets.h>
>+
>+#define DEFAULT_BUF_SIZE_BYTES	(128 * 1024)
>+#define DEFAULT_TO_SEND_BYTES	(64 * 1024)
>+#define DEFAULT_VSOCK_BUF_BYTES (256 * 1024)
>+#define DEFAULT_RCVLOWAT_BYTES	1
>+#define DEFAULT_PORT		1234
>+
>+#define BYTES_PER_GB		(1024 * 1024 * 1024ULL)
>+#define NSEC_PER_SEC		(1000000000ULL)
>+
>+static unsigned int port = DEFAULT_PORT;
>+static unsigned long buf_size_bytes = DEFAULT_BUF_SIZE_BYTES;
>+static unsigned long vsock_buf_bytes = DEFAULT_VSOCK_BUF_BYTES;
>+
>+static inline time_t current_nsec(void)
>+{
>+	struct timespec ts;
>+
>+	if (clock_gettime(CLOCK_REALTIME, &ts)) {
>+		perror("clock_gettime");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	return (ts.tv_sec * NSEC_PER_SEC) + ts.tv_nsec;
>+}
>+
>+/* From lib/cmdline.c. */
>+static unsigned long memparse(const char *ptr)
>+{
>+	char *endptr;
>+
>+	unsigned long long ret = strtoull(ptr, &endptr, 0);
>+
>+	switch (*endptr) {
>+	case 'E':
>+	case 'e':
>+		ret <<= 10;
>+	case 'P':
>+	case 'p':
>+		ret <<= 10;
>+	case 'T':
>+	case 't':
>+		ret <<= 10;
>+	case 'G':
>+	case 'g':
>+		ret <<= 10;
>+	case 'M':
>+	case 'm':
>+		ret <<= 10;
>+	case 'K':
>+	case 'k':
>+		ret <<= 10;
>+		endptr++;
>+	default:
>+		break;
>+	}
>+
>+	return ret;
>+}
>+
>+static void vsock_increase_buf_size(int fd)
>+{
>+	if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_MAX_SIZE,
>+		       &vsock_buf_bytes, sizeof(vsock_buf_bytes))) {
>+		perror("setsockopt(SO_VM_SOCKETS_BUFFER_MAX_SIZE)");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_SIZE,
>+		       &vsock_buf_bytes, sizeof(vsock_buf_bytes))) {
>+		perror("setsockopt(SO_VM_SOCKETS_BUFFER_SIZE)");
>+		exit(EXIT_FAILURE);
>+	}
>+}
>+
>+static int vsock_connect(unsigned int cid, unsigned int port)
>+{
>+	union {
>+		struct sockaddr sa;
>+		struct sockaddr_vm svm;
>+	} addr = {
>+		.svm = {
>+			.svm_family = AF_VSOCK,
>+			.svm_port = port,
>+			.svm_cid = cid,
>+		},
>+	};
>+	int fd;
>+
>+	fd = socket(AF_VSOCK, SOCK_STREAM, 0);
>+
>+	if (fd < 0) {
>+		perror("socket");
>+		return -1;
>+	}
>+
>+	if (connect(fd, &addr.sa, sizeof(addr.svm)) < 0) {
>+		perror("connect");
>+		close(fd);
>+		return -1;
>+	}
>+
>+	return fd;
>+}
>+
>+static float get_gbps(unsigned long bytes, time_t ns_delta)
>+{
>+	return ((float)bytes / BYTES_PER_GB) /
>+	       ((float)ns_delta / NSEC_PER_SEC);
>+}
>+
>+static void run_receiver(unsigned long rcvlowat_bytes)
>+{
>+	unsigned int read_cnt;
>+	time_t rx_begin_ns;
>+	time_t in_read_ns;
>+	size_t total_recv;
>+	int client_fd;
>+	char *data;
>+	int fd;
>+	union {
>+		struct sockaddr sa;
>+		struct sockaddr_vm svm;
>+	} addr = {
>+		.svm = {
>+			.svm_family = AF_VSOCK,
>+			.svm_port = port,
>+			.svm_cid = VMADDR_CID_ANY,
>+		},
>+	};
>+	union {
>+		struct sockaddr sa;
>+		struct sockaddr_vm svm;
>+	} clientaddr;
>+
>+	socklen_t clientaddr_len = sizeof(clientaddr.svm);
>+
>+	printf("Run as receiver\n");
>+	printf("Listen port %u\n", port);
>+	printf("RX buffer %lu bytes\n", buf_size_bytes);
>+	printf("vsock buffer %lu bytes\n", vsock_buf_bytes);
>+	printf("SO_RCVLOWAT %lu bytes\n", rcvlowat_bytes);
>+
>+	fd = socket(AF_VSOCK, SOCK_STREAM, 0);
>+
>+	if (fd < 0) {
>+		perror("socket");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	if (bind(fd, &addr.sa, sizeof(addr.svm)) < 0) {
>+		perror("bind");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	if (listen(fd, 1) < 0) {
>+		perror("listen");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	client_fd = accept(fd, &clientaddr.sa, &clientaddr_len);
>+
>+	if (client_fd < 0) {
>+		perror("accept");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	vsock_increase_buf_size(client_fd);
>+
>+	if (setsockopt(client_fd, SOL_SOCKET, SO_RCVLOWAT,
>+		       &rcvlowat_bytes,
>+		       sizeof(rcvlowat_bytes))) {
>+		perror("setsockopt(SO_RCVLOWAT)");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	data = malloc(buf_size_bytes);
>+
>+	if (!data) {
>+		printf("malloc failed\n");

This is an error message, so we can use fprintf(stderr, ...)

>+		exit(EXIT_FAILURE);
>+	}
>+
>+	read_cnt = 0;
>+	in_read_ns = 0;
>+	total_recv = 0;
>+	rx_begin_ns = current_nsec();
>+
>+	while (1) {
>+		struct pollfd fds = { 0 };
>+
>+		fds.fd = client_fd;
>+		fds.events = POLLIN | POLLERR |
>+			     POLLHUP | POLLRDHUP;
>+
>+		if (poll(&fds, 1, -1) < 0) {
>+			perror("poll");
>+			exit(EXIT_FAILURE);
>+		}
>+
>+		if (fds.revents & POLLERR) {
>+			printf("'poll()' error\n");

Ditto.

>+			exit(EXIT_FAILURE);
>+		}
>+
>+		if (fds.revents & POLLIN) {
>+			ssize_t bytes_read;
>+			time_t t;
>+
>+			t = current_nsec();
>+			bytes_read = read(fds.fd, data, buf_size_bytes);
>+			in_read_ns += (current_nsec() - t);
>+			read_cnt++;
>+
>+			if (!bytes_read)
>+				break;
>+
>+			if (bytes_read < 0) {
>+				perror("read");
>+				exit(EXIT_FAILURE);
>+			}
>+
>+			total_recv += bytes_read;
>+		}
>+
>+		if (fds.revents & (POLLHUP | POLLRDHUP))
>+			break;
>+	}
>+
>+	printf("rx performance: %f Gb/s\n",
>+	       get_gbps(total_recv, current_nsec() - rx_begin_ns));
>+	printf("total time in 'read()': %f sec\n", (float)in_read_ns / NSEC_PER_SEC);
>+	printf("POLLIN wakeups: %i\n", read_cnt);
>+	printf("average time in 'read()': %f ns\n", (float)in_read_ns / read_cnt);
>+
>+	free(data);
>+	close(client_fd);
>+	close(fd);
>+}
>+
>+static void run_sender(int peer_cid, unsigned long to_send_bytes)
>+{
>+	time_t tx_begin_ns;
>+	size_t total_send;
>+	void *data;
>+	int fd;
>+
>+	printf("Run as sender\n");
>+	printf("Connect to %i:%u\n", peer_cid, port);
>+	printf("Send %lu bytes\n", to_send_bytes);
>+	printf("TX buffer %lu bytes\n", buf_size_bytes);
>+
>+	fd = vsock_connect(peer_cid, port);
>+
>+	if (fd < 0)
>+		exit(EXIT_FAILURE);
>+
>+	data = malloc(buf_size_bytes);
>+
>+	if (!data) {
>+		printf("malloc failed\n");

Ditto.

>+		exit(EXIT_FAILURE);
>+	}
>+
>+	memset(data, 0, buf_size_bytes);
>+	total_send = 0;
>+	tx_begin_ns = current_nsec();
>+
>+	while (total_send < to_send_bytes) {
>+		ssize_t sent;
>+
>+		sent = write(fd, data, buf_size_bytes);
>+
>+		if (sent <= 0) {
>+			perror("write");
>+			exit(EXIT_FAILURE);
>+		}
>+
>+		total_send += sent;
>+	}

I don't know if it would be useful, but maybe we can also print the 
total number of bytes and the time spent (the same in the receiver as 
well).

Here I would add

         tx_end_ns = current_nsec();

Then use it later.

>+
>+	printf("tx performance: %f Gb/s\n",
>+	       get_gbps(total_send, current_nsec() - tx_begin_ns));
>+

Did you compared the performance with iperf-vsock and uperf?

I see an order of magnitude difference but I did not have time to check 
all the parameters to make a fair comparison. It would be helpful if you 
can compare and report it.

Thanks,
Stefano

>+	close(fd);
>+	free(data);
>+}
>+
>+static const char optstring[] = "";
>+static const struct option longopts[] = {
>+	{
>+		.name = "help",
>+		.has_arg = no_argument,
>+		.val = 'H',
>+	},
>+	{
>+		.name = "sender",
>+		.has_arg = required_argument,
>+		.val = 'S',
>+	},
>+	{
>+		.name = "port",
>+		.has_arg = required_argument,
>+		.val = 'P',
>+	},
>+	{
>+		.name = "bytes",
>+		.has_arg = required_argument,
>+		.val = 'M',
>+	},
>+	{
>+		.name = "buf-size",
>+		.has_arg = required_argument,
>+		.val = 'B',
>+	},
>+	{
>+		.name = "vsk-size",
>+		.has_arg = required_argument,
>+		.val = 'V',
>+	},
>+	{
>+		.name = "rcvlowat",
>+		.has_arg = required_argument,
>+		.val = 'R',
>+	},
>+	{},
>+};
>+
>+static void usage(void)
>+{
>+	printf("Usage: ./vsock_perf [--help] [options]\n"
>+	       "\n"
>+	       "This is benchmarking utility, to test vsock performance.\n"
>+	       "It runs in two modes: sender or receiver. In sender mode, it\n"
>+	       "connects to the specified CID and starts data transmission.\n"
>+	       "\n"
>+	       "Options:\n"
>+	       "  --help			This message\n"
>+	       "  --sender   <cid>		Sender mode (receiver default)\n"
>+	       "                                <cid> of the receiver to connect to\n"
>+	       "  --port     <port>		Port (default %d)\n"
>+	       "  --bytes    <bytes>KMG		Bytes to send (default %d)\n"
>+	       "  --buf-size <bytes>KMG		Data buffer size (default %d). In sender mode\n"
>+	       "                                it is the buffer size, passed to 'write()'. In\n"
>+	       "                                receiver mode it is the buffer size passed to 'read()'.\n"
>+	       "  --vsk-size <bytes>KMG		Socket buffer size (default %d)\n"
>+	       "  --rcvlowat <bytes>KMG		SO_RCVLOWAT value (default %d)\n"
>+	       "\n", DEFAULT_PORT, DEFAULT_TO_SEND_BYTES,
>+	       DEFAULT_BUF_SIZE_BYTES, DEFAULT_VSOCK_BUF_BYTES,
>+	       DEFAULT_RCVLOWAT_BYTES);
>+	exit(EXIT_FAILURE);
>+}
>+
>+static long strtolx(const char *arg)
>+{
>+	long value;
>+	char *end;
>+
>+	value = strtol(arg, &end, 10);
>+
>+	if (end != arg + strlen(arg))
>+		usage();
>+
>+	return value;
>+}
>+
>+int main(int argc, char **argv)
>+{
>+	unsigned long to_send_bytes = DEFAULT_TO_SEND_BYTES;
>+	unsigned long rcvlowat_bytes = DEFAULT_RCVLOWAT_BYTES;
>+	int peer_cid = -1;
>+	bool sender = false;
>+
>+	while (1) {
>+		int opt = getopt_long(argc, argv, optstring, longopts, NULL);
>+
>+		if (opt == -1)
>+			break;
>+
>+		switch (opt) {
>+		case 'V': /* Peer buffer size. */
>+			vsock_buf_bytes = memparse(optarg);
>+			break;
>+		case 'R': /* SO_RCVLOWAT value. */
>+			rcvlowat_bytes = memparse(optarg);
>+			break;
>+		case 'P': /* Port to connect to. */
>+			port = strtolx(optarg);
>+			break;
>+		case 'M': /* Bytes to send. */
>+			to_send_bytes = memparse(optarg);
>+			break;
>+		case 'B': /* Size of rx/tx buffer. */
>+			buf_size_bytes = memparse(optarg);
>+			break;
>+		case 'S': /* Sender mode. CID to connect to. */
>+			peer_cid = strtolx(optarg);
>+			sender = true;
>+			break;
>+		case 'H': /* Help. */
>+			usage();
>+			break;
>+		default:
>+			usage();
>+		}
>+	}
>+
>+	if (!sender)
>+		run_receiver(rcvlowat_bytes);
>+	else
>+		run_sender(peer_cid, to_send_bytes);
>+
>+	return 0;
>+}
>-- 
>2.25.1

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [RFC PATCH v4 4/4] test/vsock: vsock_perf utility

[Stefano Garzarella]

On Tue, Dec 06, 2022 at 08:54:28PM +0000, Arseniy Krasnov wrote:
>This adds utility to check vsock rx/tx performance.
>
>Usage as sender:
>./vsock_perf --sender <cid> --port <port> --bytes <bytes to send)

Little typo "<bytes to send>". I mean replace `)` with `>`

>Usage as receiver:
>./vsock_perf --port <port> --rcvlowat <SO_RCVLOWAT>
>
>Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>---
> tools/testing/vsock/Makefile     |   3 +-
> tools/testing/vsock/README       |  34 +++
> tools/testing/vsock/vsock_perf.c | 434 +++++++++++++++++++++++++++++++
> 3 files changed, 470 insertions(+), 1 deletion(-)
> create mode 100644 tools/testing/vsock/vsock_perf.c
>
>diff --git a/tools/testing/vsock/Makefile b/tools/testing/vsock/Makefile
>index f8293c6910c9..43a254f0e14d 100644
>--- a/tools/testing/vsock/Makefile
>+++ b/tools/testing/vsock/Makefile
>@@ -1,8 +1,9 @@
> # SPDX-License-Identifier: GPL-2.0-only
>-all: test
>+all: test vsock_perf
> test: vsock_test vsock_diag_test
> vsock_test: vsock_test.o timeout.o control.o util.o
> vsock_diag_test: vsock_diag_test.o timeout.o control.o util.o
>+vsock_perf: vsock_perf.o
>
> CFLAGS += -g -O2 -Werror -Wall -I. -I../../include -I../../../usr/include -Wno-pointer-sign -fno-strict-overflow -fno-strict-aliasing -fno-common -MMD -U_FORTIFY_SOURCE -D_GNU_SOURCE
> .PHONY: all test clean
>diff --git a/tools/testing/vsock/README b/tools/testing/vsock/README
>index 4d5045e7d2c3..e6f6735bba05 100644
>--- a/tools/testing/vsock/README
>+++ b/tools/testing/vsock/README
>@@ -35,3 +35,37 @@ Invoke test binaries in both directions as follows:
>                        --control-port=$GUEST_IP \
>                        --control-port=1234 \
>                        --peer-cid=3
>+
>+vsock_perf utility
>+-------------------
>+'vsock_perf' is a simple tool to measure vsock performance. It works in
>+sender/receiver modes: sender connect to peer at the specified port and
>+starts data transmission to the receiver. After data processing is done,
>+it prints several metrics(see below).
>+
>+Usage:
>+# run as sender
>+# connect to CID 2, port 1234, send 1G of data, tx buf size is 1M
>+./vsock_perf --sender 2 --port 1234 --bytes 1G --buf-size 1M
>+
>+Output:
>+tx performance: A Gb/s
>+
>+Output explanation:
>+A is calculated as "number of bytes to send" / "time in tx loop"
>+
>+# run as receiver
>+# listen port 1234, rx buf size is 1M, socket buf size is 1G, SO_RCVLOWAT is 64K
>+./vsock_perf --port 1234 --buf-size 1M --vsk-size 1G --rcvlowat 64K
>+
>+Output:
>+rx performance: A Gb/s
>+total in 'read()': B sec
>+POLLIN wakeups: C
>+average in 'read()': D ns
>+
>+Output explanation:
>+A is calculated as "number of received bytes" / "time in rx loop".
>+B is time, spent in 'read()' system call(excluding 'poll()')
>+C is number of 'poll()' wake ups with POLLIN bit set.
>+D is B / C, e.g. average amount of time, spent in single 'read()'.
>diff --git a/tools/testing/vsock/vsock_perf.c b/tools/testing/vsock/vsock_perf.c
>new file mode 100644
>index 000000000000..248fc3285d5c
>--- /dev/null
>+++ b/tools/testing/vsock/vsock_perf.c
>@@ -0,0 +1,434 @@
>+// SPDX-License-Identifier: GPL-2.0-only
>+/*
>+ * vsock_perf - benchmark utility for vsock.
>+ *
>+ * Copyright (C) 2022 SberDevices.
>+ *
>+ * Author: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>+ */
>+#include <getopt.h>
>+#include <stdio.h>
>+#include <stdlib.h>
>+#include <stdbool.h>
>+#include <string.h>
>+#include <errno.h>
>+#include <unistd.h>
>+#include <time.h>
>+#include <stdint.h>
>+#include <poll.h>
>+#include <sys/socket.h>
>+#include <linux/vm_sockets.h>
>+
>+#define DEFAULT_BUF_SIZE_BYTES	(128 * 1024)
>+#define DEFAULT_TO_SEND_BYTES	(64 * 1024)
>+#define DEFAULT_VSOCK_BUF_BYTES (256 * 1024)
>+#define DEFAULT_RCVLOWAT_BYTES	1
>+#define DEFAULT_PORT		1234
>+
>+#define BYTES_PER_GB		(1024 * 1024 * 1024ULL)
>+#define NSEC_PER_SEC		(1000000000ULL)
>+
>+static unsigned int port = DEFAULT_PORT;
>+static unsigned long buf_size_bytes = DEFAULT_BUF_SIZE_BYTES;
>+static unsigned long vsock_buf_bytes = DEFAULT_VSOCK_BUF_BYTES;
>+
>+static inline time_t current_nsec(void)
>+{
>+	struct timespec ts;
>+
>+	if (clock_gettime(CLOCK_REALTIME, &ts)) {
>+		perror("clock_gettime");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	return (ts.tv_sec * NSEC_PER_SEC) + ts.tv_nsec;
>+}
>+
>+/* From lib/cmdline.c. */
>+static unsigned long memparse(const char *ptr)
>+{
>+	char *endptr;
>+
>+	unsigned long long ret = strtoull(ptr, &endptr, 0);
>+
>+	switch (*endptr) {
>+	case 'E':
>+	case 'e':
>+		ret <<= 10;
>+	case 'P':
>+	case 'p':
>+		ret <<= 10;
>+	case 'T':
>+	case 't':
>+		ret <<= 10;
>+	case 'G':
>+	case 'g':
>+		ret <<= 10;
>+	case 'M':
>+	case 'm':
>+		ret <<= 10;
>+	case 'K':
>+	case 'k':
>+		ret <<= 10;
>+		endptr++;
>+	default:
>+		break;
>+	}
>+
>+	return ret;
>+}
>+
>+static void vsock_increase_buf_size(int fd)
>+{
>+	if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_MAX_SIZE,
>+		       &vsock_buf_bytes, sizeof(vsock_buf_bytes))) {
>+		perror("setsockopt(SO_VM_SOCKETS_BUFFER_MAX_SIZE)");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_SIZE,
>+		       &vsock_buf_bytes, sizeof(vsock_buf_bytes))) {
>+		perror("setsockopt(SO_VM_SOCKETS_BUFFER_SIZE)");
>+		exit(EXIT_FAILURE);
>+	}
>+}
>+
>+static int vsock_connect(unsigned int cid, unsigned int port)
>+{
>+	union {
>+		struct sockaddr sa;
>+		struct sockaddr_vm svm;
>+	} addr = {
>+		.svm = {
>+			.svm_family = AF_VSOCK,
>+			.svm_port = port,
>+			.svm_cid = cid,
>+		},
>+	};
>+	int fd;
>+
>+	fd = socket(AF_VSOCK, SOCK_STREAM, 0);
>+
>+	if (fd < 0) {
>+		perror("socket");
>+		return -1;
>+	}
>+
>+	if (connect(fd, &addr.sa, sizeof(addr.svm)) < 0) {
>+		perror("connect");
>+		close(fd);
>+		return -1;
>+	}
>+
>+	return fd;
>+}
>+
>+static float get_gbps(unsigned long bytes, time_t ns_delta)
>+{
>+	return ((float)bytes / BYTES_PER_GB) /
>+	       ((float)ns_delta / NSEC_PER_SEC);
>+}
>+
>+static void run_receiver(unsigned long rcvlowat_bytes)
>+{
>+	unsigned int read_cnt;
>+	time_t rx_begin_ns;
>+	time_t in_read_ns;
>+	size_t total_recv;
>+	int client_fd;
>+	char *data;
>+	int fd;
>+	union {
>+		struct sockaddr sa;
>+		struct sockaddr_vm svm;
>+	} addr = {
>+		.svm = {
>+			.svm_family = AF_VSOCK,
>+			.svm_port = port,
>+			.svm_cid = VMADDR_CID_ANY,
>+		},
>+	};
>+	union {
>+		struct sockaddr sa;
>+		struct sockaddr_vm svm;
>+	} clientaddr;
>+
>+	socklen_t clientaddr_len = sizeof(clientaddr.svm);
>+
>+	printf("Run as receiver\n");
>+	printf("Listen port %u\n", port);
>+	printf("RX buffer %lu bytes\n", buf_size_bytes);
>+	printf("vsock buffer %lu bytes\n", vsock_buf_bytes);
>+	printf("SO_RCVLOWAT %lu bytes\n", rcvlowat_bytes);
>+
>+	fd = socket(AF_VSOCK, SOCK_STREAM, 0);
>+
>+	if (fd < 0) {
>+		perror("socket");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	if (bind(fd, &addr.sa, sizeof(addr.svm)) < 0) {
>+		perror("bind");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	if (listen(fd, 1) < 0) {
>+		perror("listen");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	client_fd = accept(fd, &clientaddr.sa, &clientaddr_len);
>+
>+	if (client_fd < 0) {
>+		perror("accept");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	vsock_increase_buf_size(client_fd);
>+
>+	if (setsockopt(client_fd, SOL_SOCKET, SO_RCVLOWAT,
>+		       &rcvlowat_bytes,
>+		       sizeof(rcvlowat_bytes))) {
>+		perror("setsockopt(SO_RCVLOWAT)");
>+		exit(EXIT_FAILURE);
>+	}
>+
>+	data = malloc(buf_size_bytes);
>+
>+	if (!data) {
>+		printf("malloc failed\n");

This is an error message, so we can use fprintf(stderr, ...)

>+		exit(EXIT_FAILURE);
>+	}
>+
>+	read_cnt = 0;
>+	in_read_ns = 0;
>+	total_recv = 0;
>+	rx_begin_ns = current_nsec();
>+
>+	while (1) {
>+		struct pollfd fds = { 0 };
>+
>+		fds.fd = client_fd;
>+		fds.events = POLLIN | POLLERR |
>+			     POLLHUP | POLLRDHUP;
>+
>+		if (poll(&fds, 1, -1) < 0) {
>+			perror("poll");
>+			exit(EXIT_FAILURE);
>+		}
>+
>+		if (fds.revents & POLLERR) {
>+			printf("'poll()' error\n");

Ditto.

>+			exit(EXIT_FAILURE);
>+		}
>+
>+		if (fds.revents & POLLIN) {
>+			ssize_t bytes_read;
>+			time_t t;
>+
>+			t = current_nsec();
>+			bytes_read = read(fds.fd, data, buf_size_bytes);
>+			in_read_ns += (current_nsec() - t);
>+			read_cnt++;
>+
>+			if (!bytes_read)
>+				break;
>+
>+			if (bytes_read < 0) {
>+				perror("read");
>+				exit(EXIT_FAILURE);
>+			}
>+
>+			total_recv += bytes_read;
>+		}
>+
>+		if (fds.revents & (POLLHUP | POLLRDHUP))
>+			break;
>+	}
>+
>+	printf("rx performance: %f Gb/s\n",
>+	       get_gbps(total_recv, current_nsec() - rx_begin_ns));
>+	printf("total time in 'read()': %f sec\n", (float)in_read_ns / NSEC_PER_SEC);
>+	printf("POLLIN wakeups: %i\n", read_cnt);
>+	printf("average time in 'read()': %f ns\n", (float)in_read_ns / read_cnt);
>+
>+	free(data);
>+	close(client_fd);
>+	close(fd);
>+}
>+
>+static void run_sender(int peer_cid, unsigned long to_send_bytes)
>+{
>+	time_t tx_begin_ns;
>+	size_t total_send;
>+	void *data;
>+	int fd;
>+
>+	printf("Run as sender\n");
>+	printf("Connect to %i:%u\n", peer_cid, port);
>+	printf("Send %lu bytes\n", to_send_bytes);
>+	printf("TX buffer %lu bytes\n", buf_size_bytes);
>+
>+	fd = vsock_connect(peer_cid, port);
>+
>+	if (fd < 0)
>+		exit(EXIT_FAILURE);
>+
>+	data = malloc(buf_size_bytes);
>+
>+	if (!data) {
>+		printf("malloc failed\n");

Ditto.

>+		exit(EXIT_FAILURE);
>+	}
>+
>+	memset(data, 0, buf_size_bytes);
>+	total_send = 0;
>+	tx_begin_ns = current_nsec();
>+
>+	while (total_send < to_send_bytes) {
>+		ssize_t sent;
>+
>+		sent = write(fd, data, buf_size_bytes);
>+
>+		if (sent <= 0) {
>+			perror("write");
>+			exit(EXIT_FAILURE);
>+		}
>+
>+		total_send += sent;
>+	}

I don't know if it would be useful, but maybe we can also print the 
total number of bytes and the time spent (the same in the receiver as 
well).

Here I would add

         tx_end_ns = current_nsec();

Then use it later.

>+
>+	printf("tx performance: %f Gb/s\n",
>+	       get_gbps(total_send, current_nsec() - tx_begin_ns));
>+

Did you compared the performance with iperf-vsock and uperf?

I see an order of magnitude difference but I did not have time to check 
all the parameters to make a fair comparison. It would be helpful if you 
can compare and report it.

Thanks,
Stefano

>+	close(fd);
>+	free(data);
>+}
>+
>+static const char optstring[] = "";
>+static const struct option longopts[] = {
>+	{
>+		.name = "help",
>+		.has_arg = no_argument,
>+		.val = 'H',
>+	},
>+	{
>+		.name = "sender",
>+		.has_arg = required_argument,
>+		.val = 'S',
>+	},
>+	{
>+		.name = "port",
>+		.has_arg = required_argument,
>+		.val = 'P',
>+	},
>+	{
>+		.name = "bytes",
>+		.has_arg = required_argument,
>+		.val = 'M',
>+	},
>+	{
>+		.name = "buf-size",
>+		.has_arg = required_argument,
>+		.val = 'B',
>+	},
>+	{
>+		.name = "vsk-size",
>+		.has_arg = required_argument,
>+		.val = 'V',
>+	},
>+	{
>+		.name = "rcvlowat",
>+		.has_arg = required_argument,
>+		.val = 'R',
>+	},
>+	{},
>+};
>+
>+static void usage(void)
>+{
>+	printf("Usage: ./vsock_perf [--help] [options]\n"
>+	       "\n"
>+	       "This is benchmarking utility, to test vsock performance.\n"
>+	       "It runs in two modes: sender or receiver. In sender mode, it\n"
>+	       "connects to the specified CID and starts data transmission.\n"
>+	       "\n"
>+	       "Options:\n"
>+	       "  --help			This message\n"
>+	       "  --sender   <cid>		Sender mode (receiver default)\n"
>+	       "                                <cid> of the receiver to connect to\n"
>+	       "  --port     <port>		Port (default %d)\n"
>+	       "  --bytes    <bytes>KMG		Bytes to send (default %d)\n"
>+	       "  --buf-size <bytes>KMG		Data buffer size (default %d). In sender mode\n"
>+	       "                                it is the buffer size, passed to 'write()'. In\n"
>+	       "                                receiver mode it is the buffer size passed to 'read()'.\n"
>+	       "  --vsk-size <bytes>KMG		Socket buffer size (default %d)\n"
>+	       "  --rcvlowat <bytes>KMG		SO_RCVLOWAT value (default %d)\n"
>+	       "\n", DEFAULT_PORT, DEFAULT_TO_SEND_BYTES,
>+	       DEFAULT_BUF_SIZE_BYTES, DEFAULT_VSOCK_BUF_BYTES,
>+	       DEFAULT_RCVLOWAT_BYTES);
>+	exit(EXIT_FAILURE);
>+}
>+
>+static long strtolx(const char *arg)
>+{
>+	long value;
>+	char *end;
>+
>+	value = strtol(arg, &end, 10);
>+
>+	if (end != arg + strlen(arg))
>+		usage();
>+
>+	return value;
>+}
>+
>+int main(int argc, char **argv)
>+{
>+	unsigned long to_send_bytes = DEFAULT_TO_SEND_BYTES;
>+	unsigned long rcvlowat_bytes = DEFAULT_RCVLOWAT_BYTES;
>+	int peer_cid = -1;
>+	bool sender = false;
>+
>+	while (1) {
>+		int opt = getopt_long(argc, argv, optstring, longopts, NULL);
>+
>+		if (opt == -1)
>+			break;
>+
>+		switch (opt) {
>+		case 'V': /* Peer buffer size. */
>+			vsock_buf_bytes = memparse(optarg);
>+			break;
>+		case 'R': /* SO_RCVLOWAT value. */
>+			rcvlowat_bytes = memparse(optarg);
>+			break;
>+		case 'P': /* Port to connect to. */
>+			port = strtolx(optarg);
>+			break;
>+		case 'M': /* Bytes to send. */
>+			to_send_bytes = memparse(optarg);
>+			break;
>+		case 'B': /* Size of rx/tx buffer. */
>+			buf_size_bytes = memparse(optarg);
>+			break;
>+		case 'S': /* Sender mode. CID to connect to. */
>+			peer_cid = strtolx(optarg);
>+			sender = true;
>+			break;
>+		case 'H': /* Help. */
>+			usage();
>+			break;
>+		default:
>+			usage();
>+		}
>+	}
>+
>+	if (!sender)
>+		run_receiver(rcvlowat_bytes);
>+	else
>+		run_sender(peer_cid, to_send_bytes);
>+
>+	return 0;
>+}
>-- 
>2.25.1

Re: [RFC PATCH v4 4/4] test/vsock: vsock_perf utility

[Arseniy Krasnov]

On 13.12.2022 14:17, Stefano Garzarella wrote:
> On Tue, Dec 06, 2022 at 08:54:28PM +0000, Arseniy Krasnov wrote:
>> This adds utility to check vsock rx/tx performance.
>>
>> Usage as sender:
>> ./vsock_perf --sender <cid> --port <port> --bytes <bytes to send)
> 
> Little typo "<bytes to send>". I mean replace `)` with `>`
> 
>> Usage as receiver:
>> ./vsock_perf --port <port> --rcvlowat <SO_RCVLOWAT>
>>
>> Signed-off-by: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>> ---
>> tools/testing/vsock/Makefile     |   3 +-
>> tools/testing/vsock/README       |  34 +++
>> tools/testing/vsock/vsock_perf.c | 434 +++++++++++++++++++++++++++++++
>> 3 files changed, 470 insertions(+), 1 deletion(-)
>> create mode 100644 tools/testing/vsock/vsock_perf.c
>>
>> diff --git a/tools/testing/vsock/Makefile b/tools/testing/vsock/Makefile
>> index f8293c6910c9..43a254f0e14d 100644
>> --- a/tools/testing/vsock/Makefile
>> +++ b/tools/testing/vsock/Makefile
>> @@ -1,8 +1,9 @@
>> # SPDX-License-Identifier: GPL-2.0-only
>> -all: test
>> +all: test vsock_perf
>> test: vsock_test vsock_diag_test
>> vsock_test: vsock_test.o timeout.o control.o util.o
>> vsock_diag_test: vsock_diag_test.o timeout.o control.o util.o
>> +vsock_perf: vsock_perf.o
>>
>> CFLAGS += -g -O2 -Werror -Wall -I. -I../../include -I../../../usr/include -Wno-pointer-sign -fno-strict-overflow -fno-strict-aliasing -fno-common -MMD -U_FORTIFY_SOURCE -D_GNU_SOURCE
>> .PHONY: all test clean
>> diff --git a/tools/testing/vsock/README b/tools/testing/vsock/README
>> index 4d5045e7d2c3..e6f6735bba05 100644
>> --- a/tools/testing/vsock/README
>> +++ b/tools/testing/vsock/README
>> @@ -35,3 +35,37 @@ Invoke test binaries in both directions as follows:
>>                        --control-port=$GUEST_IP \
>>                        --control-port=1234 \
>>                        --peer-cid=3
>> +
>> +vsock_perf utility
>> +-------------------
>> +'vsock_perf' is a simple tool to measure vsock performance. It works in
>> +sender/receiver modes: sender connect to peer at the specified port and
>> +starts data transmission to the receiver. After data processing is done,
>> +it prints several metrics(see below).
>> +
>> +Usage:
>> +# run as sender
>> +# connect to CID 2, port 1234, send 1G of data, tx buf size is 1M
>> +./vsock_perf --sender 2 --port 1234 --bytes 1G --buf-size 1M
>> +
>> +Output:
>> +tx performance: A Gb/s
>> +
>> +Output explanation:
>> +A is calculated as "number of bytes to send" / "time in tx loop"
>> +
>> +# run as receiver
>> +# listen port 1234, rx buf size is 1M, socket buf size is 1G, SO_RCVLOWAT is 64K
>> +./vsock_perf --port 1234 --buf-size 1M --vsk-size 1G --rcvlowat 64K
>> +
>> +Output:
>> +rx performance: A Gb/s
>> +total in 'read()': B sec
>> +POLLIN wakeups: C
>> +average in 'read()': D ns
>> +
>> +Output explanation:
>> +A is calculated as "number of received bytes" / "time in rx loop".
>> +B is time, spent in 'read()' system call(excluding 'poll()')
>> +C is number of 'poll()' wake ups with POLLIN bit set.
>> +D is B / C, e.g. average amount of time, spent in single 'read()'.
>> diff --git a/tools/testing/vsock/vsock_perf.c b/tools/testing/vsock/vsock_perf.c
>> new file mode 100644
>> index 000000000000..248fc3285d5c
>> --- /dev/null
>> +++ b/tools/testing/vsock/vsock_perf.c
>> @@ -0,0 +1,434 @@
>> +// SPDX-License-Identifier: GPL-2.0-only
>> +/*
>> + * vsock_perf - benchmark utility for vsock.
>> + *
>> + * Copyright (C) 2022 SberDevices.
>> + *
>> + * Author: Arseniy Krasnov <AVKrasnov@sberdevices.ru>
>> + */
>> +#include <getopt.h>
>> +#include <stdio.h>
>> +#include <stdlib.h>
>> +#include <stdbool.h>
>> +#include <string.h>
>> +#include <errno.h>
>> +#include <unistd.h>
>> +#include <time.h>
>> +#include <stdint.h>
>> +#include <poll.h>
>> +#include <sys/socket.h>
>> +#include <linux/vm_sockets.h>
>> +
>> +#define DEFAULT_BUF_SIZE_BYTES    (128 * 1024)
>> +#define DEFAULT_TO_SEND_BYTES    (64 * 1024)
>> +#define DEFAULT_VSOCK_BUF_BYTES (256 * 1024)
>> +#define DEFAULT_RCVLOWAT_BYTES    1
>> +#define DEFAULT_PORT        1234
>> +
>> +#define BYTES_PER_GB        (1024 * 1024 * 1024ULL)
>> +#define NSEC_PER_SEC        (1000000000ULL)
>> +
>> +static unsigned int port = DEFAULT_PORT;
>> +static unsigned long buf_size_bytes = DEFAULT_BUF_SIZE_BYTES;
>> +static unsigned long vsock_buf_bytes = DEFAULT_VSOCK_BUF_BYTES;
>> +
>> +static inline time_t current_nsec(void)
>> +{
>> +    struct timespec ts;
>> +
>> +    if (clock_gettime(CLOCK_REALTIME, &ts)) {
>> +        perror("clock_gettime");
>> +        exit(EXIT_FAILURE);
>> +    }
>> +
>> +    return (ts.tv_sec * NSEC_PER_SEC) + ts.tv_nsec;
>> +}
>> +
>> +/* From lib/cmdline.c. */
>> +static unsigned long memparse(const char *ptr)
>> +{
>> +    char *endptr;
>> +
>> +    unsigned long long ret = strtoull(ptr, &endptr, 0);
>> +
>> +    switch (*endptr) {
>> +    case 'E':
>> +    case 'e':
>> +        ret <<= 10;
>> +    case 'P':
>> +    case 'p':
>> +        ret <<= 10;
>> +    case 'T':
>> +    case 't':
>> +        ret <<= 10;
>> +    case 'G':
>> +    case 'g':
>> +        ret <<= 10;
>> +    case 'M':
>> +    case 'm':
>> +        ret <<= 10;
>> +    case 'K':
>> +    case 'k':
>> +        ret <<= 10;
>> +        endptr++;
>> +    default:
>> +        break;
>> +    }
>> +
>> +    return ret;
>> +}
>> +
>> +static void vsock_increase_buf_size(int fd)
>> +{
>> +    if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_MAX_SIZE,
>> +               &vsock_buf_bytes, sizeof(vsock_buf_bytes))) {
>> +        perror("setsockopt(SO_VM_SOCKETS_BUFFER_MAX_SIZE)");
>> +        exit(EXIT_FAILURE);
>> +    }
>> +
>> +    if (setsockopt(fd, AF_VSOCK, SO_VM_SOCKETS_BUFFER_SIZE,
>> +               &vsock_buf_bytes, sizeof(vsock_buf_bytes))) {
>> +        perror("setsockopt(SO_VM_SOCKETS_BUFFER_SIZE)");
>> +        exit(EXIT_FAILURE);
>> +    }
>> +}
>> +
>> +static int vsock_connect(unsigned int cid, unsigned int port)
>> +{
>> +    union {
>> +        struct sockaddr sa;
>> +        struct sockaddr_vm svm;
>> +    } addr = {
>> +        .svm = {
>> +            .svm_family = AF_VSOCK,
>> +            .svm_port = port,
>> +            .svm_cid = cid,
>> +        },
>> +    };
>> +    int fd;
>> +
>> +    fd = socket(AF_VSOCK, SOCK_STREAM, 0);
>> +
>> +    if (fd < 0) {
>> +        perror("socket");
>> +        return -1;
>> +    }
>> +
>> +    if (connect(fd, &addr.sa, sizeof(addr.svm)) < 0) {
>> +        perror("connect");
>> +        close(fd);
>> +        return -1;
>> +    }
>> +
>> +    return fd;
>> +}
>> +
>> +static float get_gbps(unsigned long bytes, time_t ns_delta)
>> +{
>> +    return ((float)bytes / BYTES_PER_GB) /
>> +           ((float)ns_delta / NSEC_PER_SEC);
>> +}
>> +
>> +static void run_receiver(unsigned long rcvlowat_bytes)
>> +{
>> +    unsigned int read_cnt;
>> +    time_t rx_begin_ns;
>> +    time_t in_read_ns;
>> +    size_t total_recv;
>> +    int client_fd;
>> +    char *data;
>> +    int fd;
>> +    union {
>> +        struct sockaddr sa;
>> +        struct sockaddr_vm svm;
>> +    } addr = {
>> +        .svm = {
>> +            .svm_family = AF_VSOCK,
>> +            .svm_port = port,
>> +            .svm_cid = VMADDR_CID_ANY,
>> +        },
>> +    };
>> +    union {
>> +        struct sockaddr sa;
>> +        struct sockaddr_vm svm;
>> +    } clientaddr;
>> +
>> +    socklen_t clientaddr_len = sizeof(clientaddr.svm);
>> +
>> +    printf("Run as receiver\n");
>> +    printf("Listen port %u\n", port);
>> +    printf("RX buffer %lu bytes\n", buf_size_bytes);
>> +    printf("vsock buffer %lu bytes\n", vsock_buf_bytes);
>> +    printf("SO_RCVLOWAT %lu bytes\n", rcvlowat_bytes);
>> +
>> +    fd = socket(AF_VSOCK, SOCK_STREAM, 0);
>> +
>> +    if (fd < 0) {
>> +        perror("socket");
>> +        exit(EXIT_FAILURE);
>> +    }
>> +
>> +    if (bind(fd, &addr.sa, sizeof(addr.svm)) < 0) {
>> +        perror("bind");
>> +        exit(EXIT_FAILURE);
>> +    }
>> +
>> +    if (listen(fd, 1) < 0) {
>> +        perror("listen");
>> +        exit(EXIT_FAILURE);
>> +    }
>> +
>> +    client_fd = accept(fd, &clientaddr.sa, &clientaddr_len);
>> +
>> +    if (client_fd < 0) {
>> +        perror("accept");
>> +        exit(EXIT_FAILURE);
>> +    }
>> +
>> +    vsock_increase_buf_size(client_fd);
>> +
>> +    if (setsockopt(client_fd, SOL_SOCKET, SO_RCVLOWAT,
>> +               &rcvlowat_bytes,
>> +               sizeof(rcvlowat_bytes))) {
>> +        perror("setsockopt(SO_RCVLOWAT)");
>> +        exit(EXIT_FAILURE);
>> +    }
>> +
>> +    data = malloc(buf_size_bytes);
>> +
>> +    if (!data) {
>> +        printf("malloc failed\n");
> 
> This is an error message, so we can use fprintf(stderr, ...)
> 
>> +        exit(EXIT_FAILURE);
>> +    }
>> +
>> +    read_cnt = 0;
>> +    in_read_ns = 0;
>> +    total_recv = 0;
>> +    rx_begin_ns = current_nsec();
>> +
>> +    while (1) {
>> +        struct pollfd fds = { 0 };
>> +
>> +        fds.fd = client_fd;
>> +        fds.events = POLLIN | POLLERR |
>> +                 POLLHUP | POLLRDHUP;
>> +
>> +        if (poll(&fds, 1, -1) < 0) {
>> +            perror("poll");
>> +            exit(EXIT_FAILURE);
>> +        }
>> +
>> +        if (fds.revents & POLLERR) {
>> +            printf("'poll()' error\n");
> 
> Ditto.
> 
>> +            exit(EXIT_FAILURE);
>> +        }
>> +
>> +        if (fds.revents & POLLIN) {
>> +            ssize_t bytes_read;
>> +            time_t t;
>> +
>> +            t = current_nsec();
>> +            bytes_read = read(fds.fd, data, buf_size_bytes);
>> +            in_read_ns += (current_nsec() - t);
>> +            read_cnt++;
>> +
>> +            if (!bytes_read)
>> +                break;
>> +
>> +            if (bytes_read < 0) {
>> +                perror("read");
>> +                exit(EXIT_FAILURE);
>> +            }
>> +
>> +            total_recv += bytes_read;
>> +        }
>> +
>> +        if (fds.revents & (POLLHUP | POLLRDHUP))
>> +            break;
>> +    }
>> +
>> +    printf("rx performance: %f Gb/s\n",
>> +           get_gbps(total_recv, current_nsec() - rx_begin_ns));
>> +    printf("total time in 'read()': %f sec\n", (float)in_read_ns / NSEC_PER_SEC);
>> +    printf("POLLIN wakeups: %i\n", read_cnt);
>> +    printf("average time in 'read()': %f ns\n", (float)in_read_ns / read_cnt);
>> +
>> +    free(data);
>> +    close(client_fd);
>> +    close(fd);
>> +}
>> +
>> +static void run_sender(int peer_cid, unsigned long to_send_bytes)
>> +{
>> +    time_t tx_begin_ns;
>> +    size_t total_send;
>> +    void *data;
>> +    int fd;
>> +
>> +    printf("Run as sender\n");
>> +    printf("Connect to %i:%u\n", peer_cid, port);
>> +    printf("Send %lu bytes\n", to_send_bytes);
>> +    printf("TX buffer %lu bytes\n", buf_size_bytes);
>> +
>> +    fd = vsock_connect(peer_cid, port);
>> +
>> +    if (fd < 0)
>> +        exit(EXIT_FAILURE);
>> +
>> +    data = malloc(buf_size_bytes);
>> +
>> +    if (!data) {
>> +        printf("malloc failed\n");
> 
> Ditto.
> 
>> +        exit(EXIT_FAILURE);
>> +    }
>> +
>> +    memset(data, 0, buf_size_bytes);
>> +    total_send = 0;
>> +    tx_begin_ns = current_nsec();
>> +
>> +    while (total_send < to_send_bytes) {
>> +        ssize_t sent;
>> +
>> +        sent = write(fd, data, buf_size_bytes);
>> +
>> +        if (sent <= 0) {
>> +            perror("write");
>> +            exit(EXIT_FAILURE);
>> +        }
>> +
>> +        total_send += sent;
>> +    }
> 
> I don't know if it would be useful, but maybe we can also print the total number of bytes and the time spent (the same in the receiver as well).
> 
> Here I would add
> 
>         tx_end_ns = current_nsec();
> 
> Then use it later.
> 
>> +
>> +    printf("tx performance: %f Gb/s\n",
>> +           get_gbps(total_send, current_nsec() - tx_begin_ns));
>> +
> 
> Did you compared the performance with iperf-vsock and uperf?
> 
> I see an order of magnitude difference but I did not have time to check all the parameters to make a fair comparison. It would be helpful if you can compare and report it.
Thanks for review,

Sounds interestring, i'll check it and post results in cv of the next version(along with minor changes in patchset).

Thank You
> 
> Thanks,
> Stefano
> 
>> +    close(fd);
>> +    free(data);
>> +}
>> +
>> +static const char optstring[] = "";
>> +static const struct option longopts[] = {
>> +    {
>> +        .name = "help",
>> +        .has_arg = no_argument,
>> +        .val = 'H',
>> +    },
>> +    {
>> +        .name = "sender",
>> +        .has_arg = required_argument,
>> +        .val = 'S',
>> +    },
>> +    {
>> +        .name = "port",
>> +        .has_arg = required_argument,
>> +        .val = 'P',
>> +    },
>> +    {
>> +        .name = "bytes",
>> +        .has_arg = required_argument,
>> +        .val = 'M',
>> +    },
>> +    {
>> +        .name = "buf-size",
>> +        .has_arg = required_argument,
>> +        .val = 'B',
>> +    },
>> +    {
>> +        .name = "vsk-size",
>> +        .has_arg = required_argument,
>> +        .val = 'V',
>> +    },
>> +    {
>> +        .name = "rcvlowat",
>> +        .has_arg = required_argument,
>> +        .val = 'R',
>> +    },
>> +    {},
>> +};
>> +
>> +static void usage(void)
>> +{
>> +    printf("Usage: ./vsock_perf [--help] [options]\n"
>> +           "\n"
>> +           "This is benchmarking utility, to test vsock performance.\n"
>> +           "It runs in two modes: sender or receiver. In sender mode, it\n"
>> +           "connects to the specified CID and starts data transmission.\n"
>> +           "\n"
>> +           "Options:\n"
>> +           "  --help            This message\n"
>> +           "  --sender   <cid>        Sender mode (receiver default)\n"
>> +           "                                <cid> of the receiver to connect to\n"
>> +           "  --port     <port>        Port (default %d)\n"
>> +           "  --bytes    <bytes>KMG        Bytes to send (default %d)\n"
>> +           "  --buf-size <bytes>KMG        Data buffer size (default %d). In sender mode\n"
>> +           "                                it is the buffer size, passed to 'write()'. In\n"
>> +           "                                receiver mode it is the buffer size passed to 'read()'.\n"
>> +           "  --vsk-size <bytes>KMG        Socket buffer size (default %d)\n"
>> +           "  --rcvlowat <bytes>KMG        SO_RCVLOWAT value (default %d)\n"
>> +           "\n", DEFAULT_PORT, DEFAULT_TO_SEND_BYTES,
>> +           DEFAULT_BUF_SIZE_BYTES, DEFAULT_VSOCK_BUF_BYTES,
>> +           DEFAULT_RCVLOWAT_BYTES);
>> +    exit(EXIT_FAILURE);
>> +}
>> +
>> +static long strtolx(const char *arg)
>> +{
>> +    long value;
>> +    char *end;
>> +
>> +    value = strtol(arg, &end, 10);
>> +
>> +    if (end != arg + strlen(arg))
>> +        usage();
>> +
>> +    return value;
>> +}
>> +
>> +int main(int argc, char **argv)
>> +{
>> +    unsigned long to_send_bytes = DEFAULT_TO_SEND_BYTES;
>> +    unsigned long rcvlowat_bytes = DEFAULT_RCVLOWAT_BYTES;
>> +    int peer_cid = -1;
>> +    bool sender = false;
>> +
>> +    while (1) {
>> +        int opt = getopt_long(argc, argv, optstring, longopts, NULL);
>> +
>> +        if (opt == -1)
>> +            break;
>> +
>> +        switch (opt) {
>> +        case 'V': /* Peer buffer size. */
>> +            vsock_buf_bytes = memparse(optarg);
>> +            break;
>> +        case 'R': /* SO_RCVLOWAT value. */
>> +            rcvlowat_bytes = memparse(optarg);
>> +            break;
>> +        case 'P': /* Port to connect to. */
>> +            port = strtolx(optarg);
>> +            break;
>> +        case 'M': /* Bytes to send. */
>> +            to_send_bytes = memparse(optarg);
>> +            break;
>> +        case 'B': /* Size of rx/tx buffer. */
>> +            buf_size_bytes = memparse(optarg);
>> +            break;
>> +        case 'S': /* Sender mode. CID to connect to. */
>> +            peer_cid = strtolx(optarg);
>> +            sender = true;
>> +            break;
>> +        case 'H': /* Help. */
>> +            usage();
>> +            break;
>> +        default:
>> +            usage();
>> +        }
>> +    }
>> +
>> +    if (!sender)
>> +        run_receiver(rcvlowat_bytes);
>> +    else
>> +        run_sender(peer_cid, to_send_bytes);
>> +
>> +    return 0;
>> +}
>> -- 
>> 2.25.1
>