Skip to contentSkip to conversation

mptcp: disallow setting the MPTCP ULP from userspace

PATCH Superseded Davide Caratti 4 participants 11 messages Oct 8, 2019 – Oct 9, 2019

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Davide Caratti

author
#0 lore raw

cover mptcp: disallow setting the MPTCP ULP from userspace

with this series we make

setscokopt(..., IPPROTO_TCP, TCP_ULP, "mptcp", sizeof("mptcp"));

return -1 and set errno to -EOPNOPSUPPORT.

this allows attaching MPTCP ULP only to a TCP socket that has been created through sock_create_kern().

- Patch 1/2 lets subflow_init() fail with -EOPNOTSUPPORT in case users try to setsockopt(..., TCP_ULP) on a TCP socket that has been created by a userspace program. I suggest squashing this patch in "mptcp:
Associate MPTCP context with TCP socket".
- Patch 2/2 includes kselftest code. I suggest squashing it with "mptcp:
add basic kselftest for mptcp".*** SUBJECT HERE ***

Davide Caratti (2):
mptcp: subflow: disallow setsockopt(... TCP_ULP, "mptcp")
selftests: test setsockopt(..., TCP_ULP, "mptcp", ...)

 net/mptcp/subflow.c                           |  7 +++
 .../selftests/net/mptcp/mptcp_connect.c       | 49 +++++++++++++++++--
 .../selftests/net/mptcp/mptcp_connect.sh      | 19 +++++++
 3 files changed, 72 insertions(+), 3 deletions(-)
Signature · 1 line
2.21.0
author
#1 lore raw

PATCH 1/2 mptcp: subflow: disallow setsockopt(... TCP_ULP, "mptcp")

the MPTCP ULP can be attached to a TCP socket only if the socket has been created with sock_create_kern(): ensure that the value of sk->sk_kern_sock is 1 in the MPTCP subflow init() function, and let it return -EOPNOTSUPP otherwise.

Signed-off-by Davide Caratti <dcaratti(a)redhat.com>
---
 net/mptcp/subflow.c | 7 +++++++
 1 file changed, 7 insertions(+)
net/mptcp/subflow.c +7 -0
index ebe93525b0aa..b40d927e71ca 100644
@@ -435,6 +435,13 @@ static int subflow_ulp_init(struct sock *sk)
 	struct tcp_sock *tp = tcp_sk(sk);
 	int err = 0;
 
+	/* disallow attaching ULP to a socket unless it has been
+	 * created with sock_create_kern()
+	 */
+	if (!sk->sk_kern_sock) {
+		err = -EOPNOTSUPP;
+		goto out;
+	}
 	ctx = subflow_create_ctx(sk, sk->sk_socket, GFP_KERNEL);
 	if (!ctx) {
 		err = -ENOMEM;
Signature · 1 line
2.21.0
#2 lore raw

[MPTCP] Re: [PATCH 1/2] mptcp: subflow: disallow setsockopt(... TCP_ULP, "mptcp")

Hi Davide,

Quoted 15 lines · Davide Caratti

the MPTCP ULP can be attached to a TCP socket only if the socket has been created with sock_create_kern(): ensure that the value of sk->sk_kern_sock is 1 in the MPTCP subflow init() function, and let it return -EOPNOTSUPP otherwise.

Signed-off-by Davide Caratti <dcaratti(a)redhat.com>
---
 net/mptcp/subflow.c | 7 +++++++
 1 file changed, 7 insertions(+)
net/mptcp/subflow.c +0 -0
index ebe93525b0aa..b40d927e71ca 100644

net/mptcp/subflow.c static int subflow_ulp_init(struct sock *sk) · line 444

Quoted 10 lines
net/mptcp/subflow.c +7 -0
@@ -435,6 +435,13 @@ static int subflow_ulp_init(struct sock *sk)
 	struct tcp_sock *tp = tcp_sk(sk);
 	int err = 0;
 
+	/* disallow attaching ULP to a socket unless it has been
+	 * created with sock_create_kern()
+	 */
+	if (!sk->sk_kern_sock) {
+		err = -EOPNOTSUPP;
+		goto out;
+	}

That's good to do that only in MPTCP code!

A very small detail if you have to do a v2: may you add a new line here to separate your new code from the following?

If you don't need a v2 and OK with you, I can add a new line by applying the new patch.

net/mptcp/subflow.c static int subflow_ulp_init(struct sock *sk) · line 447

Quoted 4 lines
net/mptcp/subflow.c +0 -0
 	ctx = subflow_create_ctx(sk, sk->sk_socket, GFP_KERNEL);
 	if (!ctx) {
 		err = -ENOMEM;

Cheers,
Matt

Signature · 5 lines
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium
#3 lore raw

[MPTCP] Re: [PATCH 1/2] mptcp: subflow: disallow setsockopt(... TCP_ULP, "mptcp")

In reply to Davide Caratti
Quoted 5 lines · Davide Caratti

the MPTCP ULP can be attached to a TCP socket only if the socket has been created with sock_create_kern(): ensure that the value of sk->sk_kern_sock is 1 in the MPTCP subflow init() function, and let it return -EOPNOTSUPP otherwise.

Looks good to me (with Matthieu's whitespace suggestion). Nice that this method doesn't require ULP infrastructure changes!

Mat

Quoted 29 lines
Signed-off-by Davide Caratti <dcaratti(a)redhat.com>

---
net/mptcp/subflow.c | 7 +++++++

1 file changed, 7 insertions(+)
net/mptcp/subflow.c +7 -0
index ebe93525b0aa..b40d927e71ca 100644
@@ -435,6 +435,13 @@ static int subflow_ulp_init(struct sock *sk)
 	struct tcp_sock *tp = tcp_sk(sk);
 	int err = 0;
 
+	/* disallow attaching ULP to a socket unless it has been
+	 * created with sock_create_kern()
+	 */
+	if (!sk->sk_kern_sock) {
+		err = -EOPNOTSUPP;
+		goto out;
+	}
 	ctx = subflow_create_ctx(sk, sk->sk_socket, GFP_KERNEL);
 	if (!ctx) {
 		err = -ENOMEM;
Signature · 4 lines
2.21.0
_______________________________________________
mptcp mailing list -- mptcp(a)lists.01.org
To unsubscribe send an email to mptcp-leave(a)lists.01.org

--
Mat Martineau
Intel

author
#4 lore raw

PATCH 2/2 selftests: test setsockopt(..., TCP_ULP, "mptcp", ...)

In reply to Davide Caratti

add a selftest to verify that the kernel forbids attaching MPTCP subflows to userspace TCP sockets with setsockopt(..., IPPROTO_TCP, TCP_ULP, ...).

Signed-off-by Davide Caratti <dcaratti(a)redhat.com>
---
 .../selftests/net/mptcp/mptcp_connect.c       | 49 +++++++++++++++++--
 .../selftests/net/mptcp/mptcp_connect.sh      | 19 +++++++
 2 files changed, 65 insertions(+), 3 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_connect.c +46 -3
index cac71f0ac8f8..12a722e0c527 100644
@@ -18,7 +18,8 @@
 
 #include <netdb.h>
 #include <netinet/in.h>
-#include <netinet/tcp.h>
+
+#include <linux/tcp.h>
 
 extern int optind;
 
@@ -32,11 +33,12 @@ static int  poll_timeout;
 static const char *cfg_host;
 static const char *cfg_port	= "12000";
 static int cfg_sock_proto	= IPPROTO_MPTCP;
+static bool tcpulp_audit;
 
 
 static void die_usage(void)
 {
-	fprintf(stderr, "Usage: mptcp_connect [-s MPTCP|TCP] [-p port] "
+	fprintf(stderr, "Usage: mptcp_connect [-u] [-s MPTCP|TCP] [-p port] "
 		"[ -l ] [ -t timeout ] connect_address\n");
 	exit(1);
 }
@@ -114,6 +116,41 @@ static int sock_listen_mptcp(const char * const listenaddr,
 	return sock;
 }
 
+static bool sock_test_tcpulp(const char * const remoteaddr,
+			     const char * const port)
+{
+
+	struct addrinfo hints = {
+		.ai_protocol = IPPROTO_TCP,
+		.ai_socktype = SOCK_STREAM,
+	};
+	struct addrinfo *a, *addr;
+	int sock = -1, ret = 0;
+	bool test_pass = false;
+
+	hints.ai_family = AF_INET;
+
+	xgetaddrinfo(remoteaddr, port, &hints, &addr);
+	for (a = addr; a; a = a->ai_next) {
+		sock = socket(a->ai_family, a->ai_socktype, IPPROTO_TCP);
+		if (sock < 0) {
+			perror("socket");
+			continue;
+		}
+
+		ret = setsockopt(sock, IPPROTO_TCP, TCP_ULP, "mptcp",
+				 sizeof("mptcp"));
+		close(sock);
+		if (ret == -1) {
+			test_pass = true;
+			break;
+		}
+
+		perror("setsockopt(TCP_ULP)");
+	}
+	return test_pass;
+}
+
 static int sock_connect_mptcp(const char * const remoteaddr,
 			      const char * const port, int proto)
 {
@@ -362,7 +399,7 @@ static void parse_opts(int argc, char **argv)
 {
 	int c;
 
-	while ((c = getopt(argc, argv, "lp:s:ht:")) != -1) {
+	while ((c = getopt(argc, argv, "lp:s:hut:")) != -1) {
 		switch (c) {
 		case 'l':
 			listen_mode = true;
@@ -376,6 +413,9 @@ static void parse_opts(int argc, char **argv)
 		case 'h':
 			die_usage();
 			break;
+		case 'u':
+			tcpulp_audit = true;
+			break;
 		case 't':
 			poll_timeout = atoi(optarg) * 1000;
 			if (poll_timeout <= 0)
@@ -395,6 +435,9 @@ int main(int argc, char *argv[])
 
 	parse_opts(argc, argv);
 
+	if (tcpulp_audit)
+		return sock_test_tcpulp(cfg_host, cfg_port) ? 0 : 1;
+
 	if (listen_mode) {
 		int fd = sock_listen_mptcp(cfg_host, cfg_port);
 
tools/testing/selftests/net/mptcp/mptcp_connect.sh +19 -0
index 7d312bd9ac77..25f1a181c73e 100755
@@ -134,6 +134,23 @@ check_mptcp_disabled()
 	return 0
 }
 
+check_mptcp_ulp_setsockopt()
+{
+	local t="ns_ulp" retval=
+
+	ip netns add ${t} || exit $ksft_skip
+	if ! ip netns exec ${t} ./mptcp_connect -u -p 10000 -s TCP 127.0.0.1 >/dev/null 2>&1; then
+		printf "setsockopt(..., TCP_ULP, \"mptcp\", ...) allowed\t[ FAIL ]\n"
+		retval=1
+		ret=$retval
+	else
+		printf "setsockopt(..., TCP_ULP, \"mptcp\", ...) blocked\t[ OK ]\n"
+		retval=0
+	fi
+	ip netns del ${t}
+	return $retval
+}
+
 do_ping()
 {
 	listener_ns="$1"
@@ -271,6 +288,8 @@ make_file "$sin" "server"
 
 check_mptcp_disabled
 
+check_mptcp_ulp_setsockopt
+
 for sender in 1 2 3 4;do
 	do_ping ns1 ns$sender 10.0.1.1
 
Signature · 1 line
2.21.0
#5 lore raw

[MPTCP] Re: [PATCH 2/2] selftests: test setsockopt(..., TCP_ULP, "mptcp", ...)

Hi Davide,

Quoted 2 lines · Davide Caratti

add a selftest to verify that the kernel forbids attaching MPTCP subflows to userspace TCP sockets with setsockopt(..., IPPROTO_TCP, TCP_ULP, ...).

Good idea to have a test for that!

Quoted 10 lines
Signed-off-by Davide Caratti <dcaratti(a)redhat.com>
---
 .../selftests/net/mptcp/mptcp_connect.c       | 49 +++++++++++++++++--
 .../selftests/net/mptcp/mptcp_connect.sh      | 19 +++++++
 2 files changed, 65 insertions(+), 3 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_connect.c +0 -0
index cac71f0ac8f8..12a722e0c527 100644

[...]

Quoted 13 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +9 -0
@@ -114,6 +116,41 @@ static int sock_listen_mptcp(const char * const listenaddr,
 	return sock;
 }
 
+static bool sock_test_tcpulp(const char * const remoteaddr,
+			     const char * const port)
+{
+
+	struct addrinfo hints = {
+		.ai_protocol = IPPROTO_TCP,
+		.ai_socktype = SOCK_STREAM,
+	};
+	struct addrinfo *a, *addr;

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 137

Quoted 10 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +10 -0
+	int sock = -1, ret = 0;
+	bool test_pass = false;
+
+	hints.ai_family = AF_INET;
+
+	xgetaddrinfo(remoteaddr, port, &hints, &addr);
+	for (a = addr; a; a = a->ai_next) {
+		sock = socket(a->ai_family, a->ai_socktype, IPPROTO_TCP);
+		if (sock < 0) {
+			perror("socket");

Why could we have an error here? It is just to avoid error message in the output for something we don't consider as an error.

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 144

Quoted 7 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +7 -0
+			continue;
+		}
+
+		ret = setsockopt(sock, IPPROTO_TCP, TCP_ULP, "mptcp",
+				 sizeof("mptcp"));
+		close(sock);
+		if (ret == -1) {

Should we check errno as well? Just to be sure we are not "blocked" for another reason.

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 156

Quoted 12 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +9 -0
+			test_pass = true;
+			break;
+		}
+
+		perror("setsockopt(TCP_ULP)");
+	}
+	return test_pass;
+}
+
 static int sock_connect_mptcp(const char * const remoteaddr,
 			      const char * const port, int proto)
 {

Cheers,
Matt

Signature · 5 lines
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium
#6 lore raw

[MPTCP] Re: [PATCH 2/2] selftests: test setsockopt(..., TCP_ULP, "mptcp", ...)

Quoted 39 lines · Matthieu Baerts

Hi Davide,

Quoted 2 lines · Davide Caratti

add a selftest to verify that the kernel forbids attaching MPTCP subflows to userspace TCP sockets with setsockopt(..., IPPROTO_TCP, TCP_ULP, ...).

Good idea to have a test for that!

Quoted 10 lines
Signed-off-by Davide Caratti <dcaratti(a)redhat.com>
---
 .../selftests/net/mptcp/mptcp_connect.c       | 49 +++++++++++++++++--
 .../selftests/net/mptcp/mptcp_connect.sh      | 19 +++++++
 2 files changed, 65 insertions(+), 3 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_connect.c +0 -0
index cac71f0ac8f8..12a722e0c527 100644

[...]

Quoted 18 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +14 -0
@@ -114,6 +116,41 @@ static int sock_listen_mptcp(const char * const listenaddr,
 	return sock;
 }
 
+static bool sock_test_tcpulp(const char * const remoteaddr,
+			     const char * const port)
+{
+
+	struct addrinfo hints = {
+		.ai_protocol = IPPROTO_TCP,
+		.ai_socktype = SOCK_STREAM,
+	};
+	struct addrinfo *a, *addr;
+	int sock = -1, ret = 0;
+	bool test_pass = false;
+
+	hints.ai_family = AF_INET;
+
Quoted 8 lines

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 137

Quoted 5 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +5 -0
+	xgetaddrinfo(remoteaddr, port, &hints, &addr);
+	for (a = addr; a; a = a->ai_next) {
+		sock = socket(a->ai_family, a->ai_socktype, IPPROTO_TCP);
+		if (sock < 0) {
+			perror("socket");

Why could we have an error here? It is just to avoid error message in the output for something we don't consider as an error.

It can fail e.g. due to memory pressure, or ulimits. Should never happen in normal circumstances, but I think it's better to catch this up, as this patch does.

Quoted 10 lines

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 144

Quoted 7 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +7 -0
+			continue;
+		}
+
+		ret = setsockopt(sock, IPPROTO_TCP, TCP_ULP, "mptcp",
+				 sizeof("mptcp"));
+		close(sock);
+		if (ret == -1) {

Should we check errno as well? Just to be sure we are not "blocked" for another reason.

I agree additionally checking errno == -EOPNOTSUPP would be better.

Otherwise LGTM.

In the longer run we may consider moving this test and ev. diag, MIB, etc to a separate, functional-test binary/script, but I think this is good for now.

Cheers,

Paolo

#7 lore raw

[MPTCP] Re: [PATCH 2/2] selftests: test setsockopt(..., TCP_ULP, "mptcp", ...)

Hi Paolo,

Quoted 49 lines · Paolo Abeni
Quoted 39 lines · Matthieu Baerts

Hi Davide,

Quoted 2 lines · Davide Caratti

add a selftest to verify that the kernel forbids attaching MPTCP subflows to userspace TCP sockets with setsockopt(..., IPPROTO_TCP, TCP_ULP, ...).

Good idea to have a test for that!

Quoted 10 lines
Signed-off-by Davide Caratti <dcaratti(a)redhat.com>
---
 .../selftests/net/mptcp/mptcp_connect.c       | 49 +++++++++++++++++--
 .../selftests/net/mptcp/mptcp_connect.sh      | 19 +++++++
 2 files changed, 65 insertions(+), 3 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_connect.c +0 -0
index cac71f0ac8f8..12a722e0c527 100644

[...]

Quoted 18 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +14 -0
@@ -114,6 +116,41 @@ static int sock_listen_mptcp(const char * const listenaddr,
 	return sock;
 }
 
+static bool sock_test_tcpulp(const char * const remoteaddr,
+			     const char * const port)
+{
+
+	struct addrinfo hints = {
+		.ai_protocol = IPPROTO_TCP,
+		.ai_socktype = SOCK_STREAM,
+	};
+	struct addrinfo *a, *addr;
+	int sock = -1, ret = 0;
+	bool test_pass = false;
+
+	hints.ai_family = AF_INET;
+
Quoted 8 lines

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 137

Quoted 5 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +5 -0
+	xgetaddrinfo(remoteaddr, port, &hints, &addr);
+	for (a = addr; a; a = a->ai_next) {
+		sock = socket(a->ai_family, a->ai_socktype, IPPROTO_TCP);
+		if (sock < 0) {
+			perror("socket");

Why could we have an error here? It is just to avoid error message in the output for something we don't consider as an error.

Quoted 3 lines

It can fail e.g. due to memory pressure, or ulimits. Should never happen in normal circumstances, but I think it's better to catch this up, as this patch does.

Sorry, I should have added my comment at the next line. My question was
also: should we not "break" here in case of error?
But if should never fail, I guess we can still go to the next address. As long as we don't have "useless" error messages each time we execute the tests, I am then fine with both "break" and "continue" here.

Quoted 11 lines
Quoted 10 lines

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 144

Quoted 7 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +7 -0
+			continue;
+		}
+
+		ret = setsockopt(sock, IPPROTO_TCP, TCP_ULP, "mptcp",
+				 sizeof("mptcp"));
+		close(sock);
+		if (ret == -1) {

Should we check errno as well? Just to be sure we are not "blocked" for another reason.

Quoted 7 lines

I agree additionally checking errno == -EOPNOTSUPP would be better.

Otherwise LGTM.

In the longer run we may consider moving this test and ev. diag, MIB, etc to a separate, functional-test binary/script, but I think this is good for now.

Good idea, I was thinking about the same thing yesterday. I can have a look at this in the future if needed.

Cheers,
Matt

Signature · 5 lines
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium
author
#8 lore raw

[MPTCP] Re: [PATCH 2/2] selftests: test setsockopt(..., TCP_ULP, "mptcp", ...)

Quoted 9 lines · Matthieu Baerts

Hi Paolo,

Quoted 6 lines · Paolo Abeni
Quoted 5 lines · Matthieu Baerts

Hi Davide,

Quoted 2 lines · Davide Caratti

add a selftest to verify that the kernel forbids attaching MPTCP subflows to userspace TCP sockets with setsockopt(..., IPPROTO_TCP, TCP_ULP, ...).

[...]

Quoted 13 lines
Quoted 9 lines
Quoted 8 lines

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 137

Quoted 5 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +5 -0
+	xgetaddrinfo(remoteaddr, port, &hints, &addr);
+	for (a = addr; a; a = a->ai_next) {
+		sock = socket(a->ai_family, a->ai_socktype, IPPROTO_TCP);
+		if (sock < 0) {
+			perror("socket");

Why could we have an error here? It is just to avoid error message in the output for something we don't consider as an error.

Quoted 3 lines

It can fail e.g. due to memory pressure, or ulimits. Should never happen in normal circumstances, but I think it's better to catch this up, as this patch does.

Quoted 5 lines

Sorry, I should have added my comment at the next line. My question was
also: should we not "break" here in case of error?
But if should never fail, I guess we can still go to the next address. As long as we don't have "useless" error messages each time we execute the tests, I am then fine with both "break" and "continue" here.

speaking of this, I see that I redirected the standard error of the test to /dev/null. This makes both "perror()" useless. Unless you object, I will remove redirections from the shellscript.

Quoted 11 lines
Quoted 11 lines
Quoted 10 lines

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 144

Quoted 7 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +7 -0
+			continue;
+		}
+
+		ret = setsockopt(sock, IPPROTO_TCP, TCP_ULP, "mptcp",
+				 sizeof("mptcp"));
+		close(sock);
+		if (ret == -1) {

Should we check errno as well? Just to be sure we are not "blocked" for another reason.

Quoted 1 line
Quoted 1 line

I agree additionally checking errno == -EOPNOTSUPP would be better.

sure, will do that in v2.

Signature · 1 line
davide
#9 lore raw

[MPTCP] Re: [PATCH 2/2] selftests: test setsockopt(..., TCP_ULP, "mptcp", ...)

Hi Davide,

Quoted 4 lines · Davide Caratti
Quoted 3 lines · Matthieu Baerts
Quoted 2 lines · Paolo Abeni
Quoted 1 line · Matthieu Baerts

On 08/10/2019 16:43, Davide Caratti wrote:

[...]

Quoted 19 lines
Quoted 13 lines
Quoted 9 lines
Quoted 8 lines

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 137

Quoted 5 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +5 -0
+	xgetaddrinfo(remoteaddr, port, &hints, &addr);
+	for (a = addr; a; a = a->ai_next) {
+		sock = socket(a->ai_family, a->ai_socktype, IPPROTO_TCP);
+		if (sock < 0) {
+			perror("socket");

Why could we have an error here? It is just to avoid error message in the output for something we don't consider as an error.

Quoted 3 lines

It can fail e.g. due to memory pressure, or ulimits. Should never happen in normal circumstances, but I think it's better to catch this up, as this patch does.

Quoted 5 lines

Sorry, I should have added my comment at the next line. My question was
also: should we not "break" here in case of error?
But if should never fail, I guess we can still go to the next address. As long as we don't have "useless" error messages each time we execute the tests, I am then fine with both "break" and "continue" here.

Quoted 3 lines

speaking of this, I see that I redirected the standard error of the test to /dev/null. This makes both "perror()" useless. Unless you object, I will remove redirections from the shellscript.

Good catch and good idea!

Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 10 lines

tools/testing/selftests/net/mptcp/mptcp_connect.c static int sock_listen_mptcp(const char * const listenaddr, · line 144

Quoted 7 lines
tools/testing/selftests/net/mptcp/mptcp_connect.c +7 -0
+			continue;
+		}
+
+		ret = setsockopt(sock, IPPROTO_TCP, TCP_ULP, "mptcp",
+				 sizeof("mptcp"));
+		close(sock);
+		if (ret == -1) {

Should we check errno as well? Just to be sure we are not "blocked" for another reason.

Quoted 3 lines
Quoted 1 line
Quoted 1 line

I agree additionally checking errno == -EOPNOTSUPP would be better.

sure, will do that in v2.

Thank you!

Cheers,
Matt

Signature · 5 lines
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium
#10 lore raw
In reply to Davide Caratti

Hi Davide,

Quoted 16 lines · Davide Caratti

with this series we make

setscokopt(..., IPPROTO_TCP, TCP_ULP, "mptcp", sizeof("mptcp"));

return -1 and set errno to -EOPNOPSUPPORT.

this allows attaching MPTCP ULP only to a TCP socket that has been created through sock_create_kern().

- Patch 1/2 lets subflow_init() fail with -EOPNOTSUPPORT in case users try to setsockopt(..., TCP_ULP) on a TCP socket that has been created by a userspace program. I suggest squashing this patch in "mptcp:
Associate MPTCP context with TCP socket".
- Patch 2/2 includes kselftest code. I suggest squashing it with "mptcp:
add basic kselftest for mptcp".*** SUBJECT HERE ***

Quoted 3 lines

Davide Caratti (2):
mptcp: subflow: disallow setsockopt(... TCP_ULP, "mptcp")
selftests: test setsockopt(..., TCP_ULP, "mptcp", ...)

Thank you for these two patches!

I just have a few questions, see my answers in my two other emails, nothing blocking!

Cheers,
Matt

Signature · 5 lines
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium