Skip to contentSkip to conversation

Add 'dump_subflow' test in selftests

PATCH In review Gang Yan 3 participants 12 messages Apr 25, 2025 – Apr 29, 2025

Subsystems (3)
lore
Share & tools
lore

Gang Yan

author
#0 lore raw

v3 cover Add 'dump_subflow' test in selftests

The 'subflow_get_info_size' is called only in 'inet_sk_attr_size', and the only calltrace is that 'tcp_diag_dump_one'->'inet_diag_dump_one_icsk' ->'inet_sk_attr_size'.

So It's necessary to add 'get_subflow_info' in mptcp_diag.c to cover this function. Given a specific TCP connection, it is possible to judge if it is a MPTCP subflow by 'INET_ULP_INFO_MPTCP' and print 'subflow_info'.

------
Changelog
v2:
- Fix the CI compiling error.
v3:
- Set idaig_ext when 'r' is declared.
- Declare iovlen to 0, and adjust the increment logic.
- Squash patch 3 into patch 4.
- Rename 'subflow_addr' to 'subflow_addrs'
- Remove the comment and declare [ad]ddr on the same line.
- Set id.idiag_cookie[0/1] when r is declared.
- Remove the 'id.idiag_if = 0' because the default value is 0.
- Squash patch 6 into patch 5.
- Add the example of output in commit message.
- Adjust the input of 'mptcp_diag -s' to "saddr:sport daddr:dport" to save a few lines in diag.sh
- In diag.sh, use "mptcp_diag -s '$subflow_line'" instead of the old version, and call 'ss' only once to ensure the safety.
- Fix some error/check/warning problems in CI.
- Put the link of issue into the commit message of patch 5.

Gang Yan (5):
selftests: mptcp: add struct params in mptcp_diag
selftests: mptcp: refactor send_query parameters for code clarity
selftests: mptcp: refactor NLMSG handling with 'proto'
selftests: mptcp: add helpers to get subflow_info
selftests: mptcp: add chk_sublfow in diag.sh

 tools/testing/selftests/net/mptcp/diag.sh     |  33 +++
 .../testing/selftests/net/mptcp/mptcp_diag.c  | 228 +++++++++++++++---
 2 files changed, 229 insertions(+), 32 deletions(-)
Signature · 1 line
2.43.0
author
#1 lore raw

PATCH v3 1/5 selftests: mptcp: add struct params in mptcp_diag

This patch adds a struct named 'params' to save 'target_token' and other future parameters. This structure facilitates future function expansions.

---
 tools/testing/selftests/net/mptcp/mptcp_diag.c | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_diag.c +11 -5
index 37d5015ad08c..ea7cb1128044 100644
@@ -19,6 +19,10 @@
 #define IPPROTO_MPTCP 262
 #endif
 
+struct params {
+	__u32 target_token;
+};
+
 struct mptcp_info {
 	__u8	mptcpi_subflows;
 	__u8	mptcpi_add_addr_signal;
@@ -237,7 +241,7 @@ static void get_mptcpinfo(__u32 token)
 	close(fd);
 }
 
-static void parse_opts(int argc, char **argv, __u32 *target_token)
+static void parse_opts(int argc, char **argv, struct params *p)
 {
 	int c;
 
@@ -250,7 +254,7 @@ static void parse_opts(int argc, char **argv, __u32 *target_token)
 			die_usage(0);
 			break;
 		case 't':
-			sscanf(optarg, "%x", target_token);
+			sscanf(optarg, "%x", &p->target_token);
 			break;
 		default:
 			die_usage(1);
@@ -261,10 +265,12 @@ static void parse_opts(int argc, char **argv, __u32 *target_token)
 
 int main(int argc, char *argv[])
 {
-	__u32 target_token;
+	struct params p = { 0 };
+
+	parse_opts(argc, argv, &p);
 
-	parse_opts(argc, argv, &target_token);
-	get_mptcpinfo(target_token);
+	if (p.target_token)
+		get_mptcpinfo(p.target_token);
 
 	return 0;
 }
Signature · 1 line
2.43.0
author
#2 lore raw

PATCH v3 2/5 selftests: mptcp: refactor send_query parameters for code clarity

In reply to Gang Yan

This patch use 'inet_diag_req_v2' instead of 'token' as parameters of send_query, and construct the req in 'get_mptcpinfo'. This modification can enhance the enhancing clarity of the code, and prepare for the dump_subflow_info.

---
 .../testing/selftests/net/mptcp/mptcp_diag.c  | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_diag.c +10 -9
index ea7cb1128044..c9533883149b 100644
@@ -62,7 +62,7 @@ static void die_usage(int r)
 	exit(r);
 }
 
-static void send_query(int fd, __u32 token)
+static void send_query(int fd, struct inet_diag_req_v2 *r)
 {
 	struct sockaddr_nl nladdr = {
 		.nl_family = AF_NETLINK
@@ -76,19 +76,13 @@ static void send_query(int fd, __u32 token)
 			.nlmsg_type = SOCK_DIAG_BY_FAMILY,
 			.nlmsg_flags = NLM_F_REQUEST
 		},
-		.r = {
-			.sdiag_family = AF_INET,
-			/* Real proto is set via INET_DIAG_REQ_PROTOCOL */
-			.sdiag_protocol = IPPROTO_TCP,
-			.id.idiag_cookie[0] = token,
-		}
+		.r = *r
 	};
 	struct rtattr rta_proto;
 	struct iovec iov[6];
 	int iovlen = 1;
 	__u32 proto;
 
-	req.r.idiag_ext |= (1 << (INET_DIAG_INFO - 1));
 	proto = IPPROTO_MPTCP;
 	rta_proto.rta_type = INET_DIAG_REQ_PROTOCOL;
 	rta_proto.rta_len = RTA_LENGTH(sizeof(proto));
@@ -229,13 +223,20 @@ static void recv_nlmsg(int fd)
 
 static void get_mptcpinfo(__u32 token)
 {
+	struct inet_diag_req_v2 r = {
+		.sdiag_family           = AF_INET,
+		/* Real proto is set via INET_DIAG_REQ_PROTOCOL */
+		.sdiag_protocol         = IPPROTO_TCP,
+		.idiag_ext = 1 << (INET_DIAG_INFO - 1),
+		.id.idiag_cookie[0]     = token,
+	};
 	int fd;
 
 	fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_SOCK_DIAG);
 	if (fd < 0)
 		die_perror("Netlink socket");
 
-	send_query(fd, token);
+	send_query(fd, &r);
 	recv_nlmsg(fd);
 
 	close(fd);
Signature · 1 line
2.43.0
maintainer
#3 lore raw

Re: [PATCH mptcp-next v3 2/5] selftests: mptcp: refactor send_query parameters for code clarity

Hi Gang,

tools/testing/selftests/net/mptcp/mptcp_diag.c

Quoted 16 lines · Gang Yan

This patch use 'inet_diag_req_v2' instead of 'token' as parameters of send_query, and construct the req in 'get_mptcpinfo'. This modification can enhance the enhancing clarity of the code, and prepare for the dump_subflow_info.

---
 .../testing/selftests/net/mptcp/mptcp_diag.c  | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_diag.c +0 -0
index ea7cb1128044..c9533883149b 100644

(...

tools/testing/selftests/net/mptcp/mptcp_diag.c static void recv_nlmsg(int fd) · line 230

Quoted 9 lines
tools/testing/selftests/net/mptcp/mptcp_diag.c +5 -0
@@ -229,13 +223,20 @@ static void recv_nlmsg(int fd)
 
 static void get_mptcpinfo(__u32 token)
 {
+	struct inet_diag_req_v2 r = {
+		.sdiag_family           = AF_INET,
+		/* Real proto is set via INET_DIAG_REQ_PROTOCOL */
+		.sdiag_protocol         = IPPROTO_TCP,
+		.idiag_ext = 1 << (INET_DIAG_INFO - 1),

detail: I guess you might want to keep all the equals aligned, no?

tools/testing/selftests/net/mptcp/mptcp_diag.c static void recv_nlmsg(int fd) · line 242

Quoted 13 lines
tools/testing/selftests/net/mptcp/mptcp_diag.c +3 -1
+		.id.idiag_cookie[0]     = token,
+	};
 	int fd;
 
 	fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_SOCK_DIAG);
 	if (fd < 0)
 		die_perror("Netlink socket");
 
-	send_query(fd, token);
+	send_query(fd, &r);
 	recv_nlmsg(fd);
 
 	close(fd);

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.
author
#4 lore raw

PATCH v3 3/5 selftests: mptcp: refactor NLMSG handling with 'proto'

In reply to Gang Yan

This patch introduces the '__u32 proto' variable to the 'send_query' and 'recv_nlmsg' functions for further extending function.

In the 'send_query' function, the inclusion of this variable makes the structure clearer and more readable.

In the 'recv_nlmsg' function, the '__u32 proto' variable ensures that the 'diag_info' field remains unmodified when processing IPPROTO_TCP data, thereby preventing unintended transformation into 'mptcp_info' format.

While at it, increment iovlen directly when an item is added to simplify this portion of the code and improve its readaility.

---
 .../testing/selftests/net/mptcp/mptcp_diag.c  | 38 ++++++++++---------
 1 file changed, 20 insertions(+), 18 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_diag.c +20 -18
index c9533883149b..e19b21631641 100644
@@ -62,7 +62,7 @@ static void die_usage(int r)
 	exit(r);
 }
 
-static void send_query(int fd, struct inet_diag_req_v2 *r)
+static void send_query(int fd, struct inet_diag_req_v2 *r, __u32 proto)
 {
 	struct sockaddr_nl nladdr = {
 		.nl_family = AF_NETLINK
@@ -80,21 +80,22 @@ static void send_query(int fd, struct inet_diag_req_v2 *r)
 	};
 	struct rtattr rta_proto;
 	struct iovec iov[6];
-	int iovlen = 1;
-	__u32 proto;
+	int iovlen = 0;
 
-	proto = IPPROTO_MPTCP;
-	rta_proto.rta_type = INET_DIAG_REQ_PROTOCOL;
-	rta_proto.rta_len = RTA_LENGTH(sizeof(proto));
-
-	iov[0] = (struct iovec) {
+	iov[iovlen++] = (struct iovec) {
 		.iov_base = &req,
 		.iov_len = sizeof(req)
 	};
-	iov[iovlen] = (struct iovec){ &rta_proto, sizeof(rta_proto)};
-	iov[iovlen + 1] = (struct iovec){ &proto, sizeof(proto)};
-	req.nlh.nlmsg_len += RTA_LENGTH(sizeof(proto));
-	iovlen += 2;
+
+	if (proto == IPPROTO_MPTCP) {
+		rta_proto.rta_type = INET_DIAG_REQ_PROTOCOL;
+		rta_proto.rta_len = RTA_LENGTH(sizeof(proto));
+
+		iov[iovlen++] = (struct iovec){ &rta_proto, sizeof(rta_proto)};
+		iov[iovlen++] = (struct iovec){ &proto, sizeof(proto)};
+		req.nlh.nlmsg_len += RTA_LENGTH(sizeof(proto));
+	}
+
 	struct msghdr msg = {
 		.msg_name = &nladdr,
 		.msg_namelen = sizeof(nladdr),
@@ -158,7 +159,7 @@ static void print_info_msg(struct mptcp_info *info)
 	printf("bytes_acked:      %llu\n", info->mptcpi_bytes_acked);
 }
 
-static void parse_nlmsg(struct nlmsghdr *nlh)
+static void parse_nlmsg(struct nlmsghdr *nlh, __u32 proto)
 {
 	struct inet_diag_msg *r = NLMSG_DATA(nlh);
 	struct rtattr *tb[INET_DIAG_MAX + 1];
@@ -167,7 +168,7 @@ static void parse_nlmsg(struct nlmsghdr *nlh)
 			   nlh->nlmsg_len - NLMSG_LENGTH(sizeof(*r)),
 			   NLA_F_NESTED);
 
-	if (tb[INET_DIAG_INFO]) {
+	if (proto == IPPROTO_MPTCP && tb[INET_DIAG_INFO]) {
 		int len = RTA_PAYLOAD(tb[INET_DIAG_INFO]);
 		struct mptcp_info *info;
 
@@ -183,7 +184,7 @@ static void parse_nlmsg(struct nlmsghdr *nlh)
 	}
 }
 
-static void recv_nlmsg(int fd)
+static void recv_nlmsg(int fd, __u32 proto)
 {
 	char rcv_buff[8192];
 	struct nlmsghdr *nlh = (struct nlmsghdr *)rcv_buff;
@@ -216,7 +217,7 @@ static void recv_nlmsg(int fd)
 			       -(err->error), strerror(-(err->error)));
 			break;
 		}
-		parse_nlmsg(nlh);
+		parse_nlmsg(nlh, proto);
 		nlh = NLMSG_NEXT(nlh, len);
 	}
 }
@@ -230,14 +231,15 @@ static void get_mptcpinfo(__u32 token)
 		.idiag_ext = 1 << (INET_DIAG_INFO - 1),
 		.id.idiag_cookie[0]     = token,
 	};
+	__u32 proto = IPPROTO_MPTCP;
 	int fd;
 
 	fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_SOCK_DIAG);
 	if (fd < 0)
 		die_perror("Netlink socket");
 
-	send_query(fd, &r);
-	recv_nlmsg(fd);
+	send_query(fd, &r, proto);
+	recv_nlmsg(fd, proto);
 
 	close(fd);
 }
Signature · 1 line
2.43.0
author
#5 lore raw

PATCH v3 4/5 selftests: mptcp: add helpers to get subflow_info

In reply to Gang Yan

This patch adds 'get_subflow_info' in 'mptcp_diag', which can check whether a TCP connection is an MPTCP subflow based on the "INET_ULP_INFO_MPTCP" with tcp_diag method.

The helper 'print_subflow_info' in 'mptcp_diag' can print the subflow_filed of an MPTCP subflow for further checking the 'subflow_info' through inet_diag method.

The example of the whole output should be:

'''
127.0.0.1:10000 -> 127.0.0.1:38984
It's a mptcp subflow, the subflow info:
flags:Mec token:0000(id:0)/4278e77e(id:0) seq:9288466187236176036 \
sfseq:1 ssnoff:2317083055 maplen:215
'''

---
 .../testing/selftests/net/mptcp/mptcp_diag.c  | 159 +++++++++++++++++-
 1 file changed, 157 insertions(+), 2 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_diag.c +157 -2
index e19b21631641..13848a25938d 100644
@@ -8,6 +8,7 @@
 #include <sys/socket.h>
 #include <netinet/in.h>
 #include <linux/tcp.h>
+#include <arpa/inet.h>
 
 #include <unistd.h>
 #include <stdlib.h>
@@ -19,8 +20,13 @@
 #define IPPROTO_MPTCP 262
 #endif
 
+#define parse_rtattr_nested(tb, max, rta) \
+	(parse_rtattr_flags((tb), (max), RTA_DATA(rta), RTA_PAYLOAD(rta), \
+			    NLA_F_NESTED))
+
 struct params {
 	__u32 target_token;
+	char subflow_addrs[1024];
 };
 
 struct mptcp_info {
@@ -50,6 +56,37 @@ struct mptcp_info {
 	__u32	mptcpi_last_ack_recv;
 };
 
+enum {
+	MPTCP_SUBFLOW_ATTR_UNSPEC,
+	MPTCP_SUBFLOW_ATTR_TOKEN_REM,
+	MPTCP_SUBFLOW_ATTR_TOKEN_LOC,
+	MPTCP_SUBFLOW_ATTR_RELWRITE_SEQ,
+	MPTCP_SUBFLOW_ATTR_MAP_SEQ,
+	MPTCP_SUBFLOW_ATTR_MAP_SFSEQ,
+	MPTCP_SUBFLOW_ATTR_SSN_OFFSET,
+	MPTCP_SUBFLOW_ATTR_MAP_DATALEN,
+	MPTCP_SUBFLOW_ATTR_FLAGS,
+	MPTCP_SUBFLOW_ATTR_ID_REM,
+	MPTCP_SUBFLOW_ATTR_ID_LOC,
+	MPTCP_SUBFLOW_ATTR_PAD,
+
+	__MPTCP_SUBFLOW_ATTR_MAX
+};
+
+#define MPTCP_SUBFLOW_ATTR_MAX (__MPTCP_SUBFLOW_ATTR_MAX - 1)
+
+#define MPTCP_SUBFLOW_FLAG_MCAP_REM		_BITUL(0)
+#define MPTCP_SUBFLOW_FLAG_MCAP_LOC		_BITUL(1)
+#define MPTCP_SUBFLOW_FLAG_JOIN_REM		_BITUL(2)
+#define MPTCP_SUBFLOW_FLAG_JOIN_LOC		_BITUL(3)
+#define MPTCP_SUBFLOW_FLAG_BKUP_REM		_BITUL(4)
+#define MPTCP_SUBFLOW_FLAG_BKUP_LOC		_BITUL(5)
+#define MPTCP_SUBFLOW_FLAG_FULLY_ESTABLISHED	_BITUL(6)
+#define MPTCP_SUBFLOW_FLAG_CONNECTED		_BITUL(7)
+#define MPTCP_SUBFLOW_FLAG_MAPVALID		_BITUL(8)
+
+#define rta_getattr(type, value)		(*(type *)RTA_DATA(value))
+
 static void die_perror(const char *msg)
 {
 	perror(msg);
@@ -58,7 +95,9 @@ static void die_perror(const char *msg)
 
 static void die_usage(int r)
 {
-	fprintf(stderr, "Usage: mptcp_diag -t\n");
+	fprintf(stderr, "Usage:\n"
+			"mptcp_diag -t <token>\n"
+			"mptcp_diag -s \"<saddr>:<sport> <daddr>:<dport>\"\n");
 	exit(r);
 }
 
@@ -159,6 +198,66 @@ static void print_info_msg(struct mptcp_info *info)
 	printf("bytes_acked:      %llu\n", info->mptcpi_bytes_acked);
 }
 
+/*
+ * 'print_subflow_info' is from 'mptcp_subflow_info'
+ * which is a function in 'misc/ss.c' of iproute2.
+ */
+static void print_subflow_info(struct rtattr *tb[])
+{
+	u_int32_t flags = 0;
+
+	printf("It's a mptcp subflow, the subflow info:\n");
+	if (tb[MPTCP_SUBFLOW_ATTR_FLAGS]) {
+		char caps[32 + 1] = { 0 }, *cap = &caps[0];
+
+		flags = rta_getattr(__u32, tb[MPTCP_SUBFLOW_ATTR_FLAGS]);
+
+		if (flags & MPTCP_SUBFLOW_FLAG_MCAP_REM)
+			*cap++ = 'M';
+		if (flags & MPTCP_SUBFLOW_FLAG_MCAP_LOC)
+			*cap++ = 'm';
+		if (flags & MPTCP_SUBFLOW_FLAG_JOIN_REM)
+			*cap++ = 'J';
+		if (flags & MPTCP_SUBFLOW_FLAG_JOIN_LOC)
+			*cap++ = 'j';
+		if (flags & MPTCP_SUBFLOW_FLAG_BKUP_REM)
+			*cap++ = 'B';
+		if (flags & MPTCP_SUBFLOW_FLAG_BKUP_LOC)
+			*cap++ = 'b';
+		if (flags & MPTCP_SUBFLOW_FLAG_FULLY_ESTABLISHED)
+			*cap++ = 'e';
+		if (flags & MPTCP_SUBFLOW_FLAG_CONNECTED)
+			*cap++ = 'c';
+		if (flags & MPTCP_SUBFLOW_FLAG_MAPVALID)
+			*cap++ = 'v';
+
+		if (flags)
+			printf(" flags:%s", caps);
+	}
+	if (tb[MPTCP_SUBFLOW_ATTR_TOKEN_REM] &&
+	    tb[MPTCP_SUBFLOW_ATTR_TOKEN_LOC] &&
+	    tb[MPTCP_SUBFLOW_ATTR_ID_REM] &&
+	    tb[MPTCP_SUBFLOW_ATTR_ID_LOC])
+		printf(" token:%04x(id:%u)/%04x(id:%u)",
+		       rta_getattr(__u32, tb[MPTCP_SUBFLOW_ATTR_TOKEN_REM]),
+		       rta_getattr(__u8, tb[MPTCP_SUBFLOW_ATTR_ID_REM]),
+		       rta_getattr(__u32, tb[MPTCP_SUBFLOW_ATTR_TOKEN_LOC]),
+		       rta_getattr(__u8, tb[MPTCP_SUBFLOW_ATTR_ID_LOC]));
+	if (tb[MPTCP_SUBFLOW_ATTR_MAP_SEQ])
+		printf(" seq:%llu",
+		       rta_getattr(__u64, tb[MPTCP_SUBFLOW_ATTR_MAP_SEQ]));
+	if (tb[MPTCP_SUBFLOW_ATTR_MAP_SFSEQ])
+		printf(" sfseq:%u",
+		       rta_getattr(__u32, tb[MPTCP_SUBFLOW_ATTR_MAP_SFSEQ]));
+	if (tb[MPTCP_SUBFLOW_ATTR_SSN_OFFSET])
+		printf(" ssnoff:%u",
+		       rta_getattr(__u32, tb[MPTCP_SUBFLOW_ATTR_SSN_OFFSET]));
+	if (tb[MPTCP_SUBFLOW_ATTR_MAP_DATALEN])
+		printf(" maplen:%u",
+		       rta_getattr(__u32, tb[MPTCP_SUBFLOW_ATTR_MAP_DATALEN]));
+	printf("\n");
+}
+
 static void parse_nlmsg(struct nlmsghdr *nlh, __u32 proto)
 {
 	struct inet_diag_msg *r = NLMSG_DATA(nlh);
@@ -182,6 +281,22 @@ static void parse_nlmsg(struct nlmsghdr *nlh, __u32 proto)
 		}
 		print_info_msg(info);
 	}
+	if (proto == IPPROTO_TCP && tb[INET_DIAG_ULP_INFO]) {
+		struct rtattr *ulpinfo[INET_ULP_INFO_MAX + 1] = { 0 };
+
+		parse_rtattr_nested(ulpinfo, INET_ULP_INFO_MAX,
+				    tb[INET_DIAG_ULP_INFO]);
+
+		if (ulpinfo[INET_ULP_INFO_MPTCP]) {
+			struct rtattr *sfinfo[MPTCP_SUBFLOW_ATTR_MAX + 1] = { 0 };
+
+			parse_rtattr_nested(sfinfo, MPTCP_SUBFLOW_ATTR_MAX,
+					    ulpinfo[INET_ULP_INFO_MPTCP]);
+			print_subflow_info(sfinfo);
+		} else {
+			printf("It's a normal TCP!\n");
+		}
+	}
 }
 
 static void recv_nlmsg(int fd, __u32 proto)
@@ -244,6 +359,39 @@ static void get_mptcpinfo(__u32 token)
 	close(fd);
 }
 
+static void get_subflow_info(char *subflow_addrs)
+{
+	struct inet_diag_req_v2 r = {
+		.sdiag_family           = AF_INET,
+		.sdiag_protocol         = IPPROTO_TCP,
+		.id.idiag_cookie[0]	= INET_DIAG_NOCOOKIE,
+		.id.idiag_cookie[1]	= INET_DIAG_NOCOOKIE,
+	};
+	char saddr[64], daddr[64];
+	int sport, dport;
+	int ret;
+	int fd;
+
+	ret = sscanf(subflow_addrs, "%[^:]:%d %[^:]:%d", saddr, &sport, daddr, &dport);
+	if (ret != 4)
+		die_perror("IP PORT Pairs has style problems!");
+
+	printf("%s:%d -> %s:%d\n", saddr, sport, daddr, dport);
+
+	fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_SOCK_DIAG);
+	if (fd < 0)
+		die_perror("Netlink socket");
+
+	r.id.idiag_sport = htons(sport);
+	r.id.idiag_dport = htons(dport);
+
+	inet_pton(AF_INET, saddr, &r.id.idiag_src);
+	inet_pton(AF_INET, daddr, &r.id.idiag_dst);
+	r.idiag_ext |= (1 << (INET_DIAG_INFO - 1));
+	send_query(fd, &r, IPPROTO_TCP);
+	recv_nlmsg(fd, IPPROTO_TCP);
+}
+
 static void parse_opts(int argc, char **argv, struct params *p)
 {
 	int c;
@@ -251,7 +399,7 @@ static void parse_opts(int argc, char **argv, struct params *p)
 	if (argc < 2)
 		die_usage(1);
 
-	while ((c = getopt(argc, argv, "ht:")) != -1) {
+	while ((c = getopt(argc, argv, "ht:s:")) != -1) {
 		switch (c) {
 		case 'h':
 			die_usage(0);
@@ -259,6 +407,10 @@ static void parse_opts(int argc, char **argv, struct params *p)
 		case 't':
 			sscanf(optarg, "%x", &p->target_token);
 			break;
+		case 's':
+			snprintf(p->subflow_addrs, strlen(optarg) + 1,
+				 "%s", optarg);
+			break;
 		default:
 			die_usage(1);
 			break;
@@ -275,6 +427,9 @@ int main(int argc, char *argv[])
 	if (p.target_token)
 		get_mptcpinfo(p.target_token);
 
+	if (strlen(p.subflow_addrs) != 0)
+		get_subflow_info(p.subflow_addrs);
+
 	return 0;
 }
 
Signature · 1 line
2.43.0
maintainer
#6 lore raw

Re: [PATCH mptcp-next v3 4/5] selftests: mptcp: add helpers to get subflow_info

Hi Gang,

Quoted 28 lines · Gang Yan

This patch adds 'get_subflow_info' in 'mptcp_diag', which can check whether a TCP connection is an MPTCP subflow based on the "INET_ULP_INFO_MPTCP" with tcp_diag method.

The helper 'print_subflow_info' in 'mptcp_diag' can print the subflow_filed of an MPTCP subflow for further checking the 'subflow_info' through inet_diag method.

The example of the whole output should be:

'''
127.0.0.1:10000 -> 127.0.0.1:38984
It's a mptcp subflow, the subflow info:
flags:Mec token:0000(id:0)/4278e77e(id:0) seq:9288466187236176036 \
sfseq:1 ssnoff:2317083055 maplen:215
'''

---
 .../testing/selftests/net/mptcp/mptcp_diag.c  | 159 +++++++++++++++++-
 1 file changed, 157 insertions(+), 2 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_diag.c +0 -0
index e19b21631641..13848a25938d 100644

(...)

Quoted 23 lines
tools/testing/selftests/net/mptcp/mptcp_diag.c +19 -0
@@ -244,6 +359,39 @@ static void get_mptcpinfo(__u32 token)
 	close(fd);
 }
 
+static void get_subflow_info(char *subflow_addrs)
+{
+	struct inet_diag_req_v2 r = {
+		.sdiag_family           = AF_INET,
+		.sdiag_protocol         = IPPROTO_TCP,
+		.id.idiag_cookie[0]	= INET_DIAG_NOCOOKIE,
+		.id.idiag_cookie[1]	= INET_DIAG_NOCOOKIE,
+	};
+	char saddr[64], daddr[64];
+	int sport, dport;
+	int ret;
+	int fd;
+
+	ret = sscanf(subflow_addrs, "%[^:]:%d %[^:]:%d", saddr, &sport, daddr, &dport);
+	if (ret != 4)
+		die_perror("IP PORT Pairs has style problems!");
+
+	printf("%s:%d -> %s:%d\n", saddr, sport, daddr, dport);
+

tools/testing/selftests/net/mptcp/mptcp_diag.c static void get_mptcpinfo(__u32 token) · line 390

Quoted 10 lines
tools/testing/selftests/net/mptcp/mptcp_diag.c +10 -0
+	fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_SOCK_DIAG);
+	if (fd < 0)
+		die_perror("Netlink socket");
+
+	r.id.idiag_sport = htons(sport);
+	r.id.idiag_dport = htons(dport);
+
+	inet_pton(AF_INET, saddr, &r.id.idiag_src);
+	inet_pton(AF_INET, daddr, &r.id.idiag_dst);
+	r.idiag_ext |= (1 << (INET_DIAG_INFO - 1));

This could be declared above as well I suppose.

Quoted 17 lines
tools/testing/selftests/net/mptcp/mptcp_diag.c +5 -1
+	send_query(fd, &r, IPPROTO_TCP);
+	recv_nlmsg(fd, IPPROTO_TCP);
+}
+
 static void parse_opts(int argc, char **argv, struct params *p)
 {
 	int c;
@@ -251,7 +399,7 @@ static void parse_opts(int argc, char **argv, struct params *p)
 	if (argc < 2)
 		die_usage(1);
 
-	while ((c = getopt(argc, argv, "ht:")) != -1) {
+	while ((c = getopt(argc, argv, "ht:s:")) != -1) {
 		switch (c) {
 		case 'h':
 			die_usage(0);

tools/testing/selftests/net/mptcp/mptcp_diag.c static void parse_opts(int argc, char **argv, struct params *p) · line 411

Quoted 5 lines
tools/testing/selftests/net/mptcp/mptcp_diag.c +2 -0
@@ -259,6 +407,10 @@ static void parse_opts(int argc, char **argv, struct params *p)
 		case 't':
 			sscanf(optarg, "%x", &p->target_token);
 			break;
+		case 's':
+			snprintf(p->subflow_addrs, strlen(optarg) + 1,

The max size (2nd argument) should be linked to the destination buffer size, not the source one.

Note that it would make more sense to use "strncpy()" here.

Quoted 6 lines
tools/testing/selftests/net/mptcp/mptcp_diag.c +2 -0
+				 "%s", optarg);
+			break;
 		default:
 			die_usage(1);
 			break;

tools/testing/selftests/net/mptcp/mptcp_diag.c int main(int argc, char *argv[]) · line 435

Quoted 9 lines
tools/testing/selftests/net/mptcp/mptcp_diag.c +3 -0
@@ -275,6 +427,9 @@ int main(int argc, char *argv[])
 	if (p.target_token)
 		get_mptcpinfo(p.target_token);
 
+	if (strlen(p.subflow_addrs) != 0)
+		get_subflow_info(p.subflow_addrs);
+
 	return 0;
 }
 

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.
author
#7 lore raw

PATCH v3 5/5 selftests: mptcp: add chk_sublfow in diag.sh

In reply to Gang Yan

This patch aims to add chk_dump_subflow in diag.sh. The subflow's info can be obtained through "ss -tin", then use the 'mptcp_diag' to verify the token in subflow_info.

---
 tools/testing/selftests/net/mptcp/diag.sh | 33 +++++++++++++++++++++++
 1 file changed, 33 insertions(+)
tools/testing/selftests/net/mptcp/diag.sh +33 -0
index e7a75341f0f3..621b405e519f 100755
@@ -225,6 +225,38 @@ chk_dump_one()
 	fi
 }
 
+chk_dump_subflow()
+{
+	local inet_diag_token
+	local subflow_line
+	local ss_output
+	local ss_token
+	local msg
+
+	ss_output=$(ss -tniN $ns)
+
+	subflow_line=$(echo "$ss_output" | \
+		       grep -m1 -Eo '[0-9.]+:[0-9].+ +[0-9.]+:[0-9.]+')
+
+	ss_token=$(echo "$ss_output" | \
+		   grep -m1 -Eo 'token:[^ ]+')
+
+	inet_diag_token=$(ip netns exec $ns ./mptcp_diag -s "$subflow_line" | \
+			  grep -Eo 'token:[^ ]+')
+
+	msg="....chk dump_subflow"
+
+	mptcp_lib_print_title "$msg"
+	if [ -n "$ss_token" ] && [ "$ss_token" == "$inet_diag_token" ]; then
+		mptcp_lib_pr_ok
+		mptcp_lib_result_pass "${msg}"
+	else
+		mptcp_lib_pr_fail "expected $ss_token found $inet_diag_token"
+		mptcp_lib_result_fail "${msg}"
+		ret=${KSFT_FAIL}
+	fi
+}
+
 msk_info_get_value()
 {
 	local port="${1}"
@@ -316,6 +348,7 @@ chk_msk_fallback_nr 0 "....chk no fallback"
 chk_msk_inuse 2
 chk_msk_cestab 2
 chk_dump_one
+chk_dump_subflow
 flush_pids
 
 chk_msk_inuse 0 "2->0"
Signature · 1 line
2.43.0
maintainer
#8 lore raw

Re: [PATCH mptcp-next v3 5/5] selftests: mptcp: add chk_sublfow in diag.sh

Hi Gang,

Quoted 6 lines · Gang Yan

This patch aims to add chk_dump_subflow in diag.sh. The subflow's info can be obtained through "ss -tin", then use the 'mptcp_diag' to verify the token in subflow_info.

There should not be any empty lines here: "Closes:" should be with the other tags without any empty lines in between.

Quoted 20 lines
---
 tools/testing/selftests/net/mptcp/diag.sh | 33 +++++++++++++++++++++++
 1 file changed, 33 insertions(+)
tools/testing/selftests/net/mptcp/diag.sh +5 -0
index e7a75341f0f3..621b405e519f 100755
@@ -225,6 +225,38 @@ chk_dump_one()
 	fi
 }
 
+chk_dump_subflow()
+{
+	local inet_diag_token
+	local subflow_line
+	local ss_output

tools/testing/selftests/net/mptcp/diag.sh chk_dump_one() · line 242

Quoted 10 lines
tools/testing/selftests/net/mptcp/diag.sh +10 -0
+	local ss_token
+	local msg
+
+	ss_output=$(ss -tniN $ns)
+
+	subflow_line=$(echo "$ss_output" | \
+		       grep -m1 -Eo '[0-9.]+:[0-9].+ +[0-9.]+:[0-9.]+')
+
+	ss_token=$(echo "$ss_output" | \
+		   grep -m1 -Eo 'token:[^ ]+')

(This could go to the previous line)

Quoted 21 lines
tools/testing/selftests/net/mptcp/diag.sh +17 -0
+
+	inet_diag_token=$(ip netns exec $ns ./mptcp_diag -s "$subflow_line" | \
+			  grep -Eo 'token:[^ ]+')
+
+	msg="....chk dump_subflow"
+
+	mptcp_lib_print_title "$msg"
+	if [ -n "$ss_token" ] && [ "$ss_token" == "$inet_diag_token" ]; then
+		mptcp_lib_pr_ok
+		mptcp_lib_result_pass "${msg}"
+	else
+		mptcp_lib_pr_fail "expected $ss_token found $inet_diag_token"
+		mptcp_lib_result_fail "${msg}"
+		ret=${KSFT_FAIL}
+	fi
+}
+
 msk_info_get_value()
 {
 	local port="${1}"

tools/testing/selftests/net/mptcp/diag.sh chk_msk_fallback_nr 0 "....chk no fallback" · line 354

Quoted 7 lines
tools/testing/selftests/net/mptcp/diag.sh +1 -0
@@ -316,6 +348,7 @@ chk_msk_fallback_nr 0 "....chk no fallback"
 chk_msk_inuse 2
 chk_msk_cestab 2
 chk_dump_one
+chk_dump_subflow
 flush_pids
 
 chk_msk_inuse 0 "2->0"

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.
#9 lore raw
In reply to Gang Yan

Hi Gang,

Thank you for your modifications, that's great!

Our CI did some validations and here is its report:

- KVM Validation: normal: Success! ✅
- KVM Validation: debug: Success! ✅
- KVM Validation: btf-normal (only bpftest_all): Success! ✅
- KVM Validation: btf-debug (only bpftest_all): Success! ✅
- Task: https://github.com/multipath-tcp/mptcp_net-next/actions/runs/14663915269

Initiator: Patchew Applier
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/0a861e272da4
Patchwork: https://patchwork.kernel.org/project/mptcp/list/?series=956852

If there are some issues, you can reproduce them using the same environment as the one used by the CI thanks to a docker image, e.g.:

    $ cd [kernel source code]
    $ docker run -v "${PWD}:${PWD}:rw" -w "${PWD}" --privileged --rm -it \
        --pull always mptcp/mptcp-upstream-virtme-docker:latest \
        auto-normal

For more details:

https://github.com/multipath-tcp/mptcp-upstream-virtme-docker

Please note that despite all the efforts that have been already done to have a stable tests suite when executed on a public CI like here, it is possible some reported issues are not due to your modifications. Still, do not hesitate to help us improve that ;-)

Cheers,
MPTCP GH Action bot
Bot operated by Matthieu Baerts (NGI0 Core)

maintainer Reviewed-by on the v3 series
#10 lore raw
In reply to Gang Yan

Hi Gang,

Quoted 22 lines · Gang Yan

The 'subflow_get_info_size' is called only in 'inet_sk_attr_size', and the only calltrace is that 'tcp_diag_dump_one'->'inet_diag_dump_one_icsk' ->'inet_sk_attr_size'.

So It's necessary to add 'get_subflow_info' in mptcp_diag.c to cover this function. Given a specific TCP connection, it is possible to judge if it is a MPTCP subflow by 'INET_ULP_INFO_MPTCP' and print 'subflow_info'.

------
Changelog
v2:
- Fix the CI compiling error.
v3:
- Set idaig_ext when 'r' is declared.
- Declare iovlen to 0, and adjust the increment logic.
- Squash patch 3 into patch 4.
- Rename 'subflow_addr' to 'subflow_addrs'
- Remove the comment and declare [ad]ddr on the same line.
- Set id.idiag_cookie[0/1] when r is declared.
- Remove the 'id.idiag_if = 0' because the default value is 0.

Quoted 8 lines

- Squash patch 6 into patch 5.
- Add the example of output in commit message.
- Adjust the input of 'mptcp_diag -s' to "saddr:sport daddr:dport" to save a few lines in diag.sh
- In diag.sh, use "mptcp_diag -s '$subflow_line'" instead of the old version, and call 'ss' only once to ensure the safety.
- Fix some error/check/warning problems in CI.
- Put the link of issue into the commit message of patch 5.

Thank you for the new version. It looks good to me, just some details that I can include when applying the patches.

Quoted 8 lines

Gang Yan (5):
selftests: mptcp: add struct params in mptcp_diag
selftests: mptcp: refactor send_query parameters for code clarity
selftests: mptcp: refactor NLMSG handling with 'proto'
selftests: mptcp: add helpers to get subflow_info
selftests: mptcp: add chk_sublfow in diag.sh

Quoted 4 lines
 tools/testing/selftests/net/mptcp/diag.sh     |  33 +++
 .../testing/selftests/net/mptcp/mptcp_diag.c  | 228 +++++++++++++++---
 2 files changed, 229 insertions(+), 32 deletions(-)

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.
maintainer
#11 lore raw
Quoted 34 lines · Matthieu Baerts

Hi Gang,

Quoted 22 lines · Gang Yan

The 'subflow_get_info_size' is called only in 'inet_sk_attr_size', and the only calltrace is that 'tcp_diag_dump_one'->'inet_diag_dump_one_icsk' ->'inet_sk_attr_size'.

So It's necessary to add 'get_subflow_info' in mptcp_diag.c to cover this function. Given a specific TCP connection, it is possible to judge if it is a MPTCP subflow by 'INET_ULP_INFO_MPTCP' and print 'subflow_info'.

------
Changelog
v2:
- Fix the CI compiling error.
v3:
- Set idaig_ext when 'r' is declared.
- Declare iovlen to 0, and adjust the increment logic.
- Squash patch 3 into patch 4.
- Rename 'subflow_addr' to 'subflow_addrs'
- Remove the comment and declare [ad]ddr on the same line.
- Set id.idiag_cookie[0/1] when r is declared.
- Remove the 'id.idiag_if = 0' because the default value is 0.

Quoted 8 lines

- Squash patch 6 into patch 5.
- Add the example of output in commit message.
- Adjust the input of 'mptcp_diag -s' to "saddr:sport daddr:dport" to save a few lines in diag.sh
- In diag.sh, use "mptcp_diag -s '$subflow_line'" instead of the old version, and call 'ss' only once to ensure the safety.
- Fix some error/check/warning problems in CI.
- Put the link of issue into the commit message of patch 5.

Quoted 4 lines

Thank you for the new version. It looks good to me, just some details that I can include when applying the patches.

Now in our tree:

New patches for t/upstream:
- f41228c0ba3b: selftests: mptcp: add struct params in mptcp_diag
- c92368e75550: selftests: mptcp: refactor send_query parameters for code clarity
- c4857ae914a9: selftests: mptcp: refactor NLMSG handling with 'proto'
- 3f3180dba6f5: selftests: mptcp: add helpers to get subflow_info
- c46a1689f413: selftests: mptcp: add chk_sublfow in diag.sh
- Results: 59ef7029a603..e17ea3d6bbc4 (export)

Tests are now in progress:

- export:
https://github.com/multipath-tcp/mptcp_net-next/commit/563df787b6317c2cd27fad121b1ccdabc8e0372f/checks

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.