[PATCH iproute2 0/3] ss: add AF_VSOCK support

[PATCH iproute2 0/3] ss: add AF_VSOCK support

[Stefan Hajnoczi]

This patch series adds AF_VSOCK support to ss(8).  AF_VSOCK is a host<->guest
communications channel supported by VMware, KVM (virtio-vsock), and Hyper-V.

To dump AF_VSOCK sockets:

  $ ss --vsock

The vsock_diag.ko kernel module for Linux was posted in "[PATCH 0/5] VSOCK: add
sock_diag interface".  That patch series also adds the
<linux/vm_sockets_diag.h> header file that this patch series relies on.  The
Linux patches are also available here:

  https://github.com/stefanha/linux/commits/vsock-diag

Stefan Hajnoczi (3):
  ss: allow AF_FAMILY constants >32
  include: add <linux/vm_sockets_diag.h>
  ss: add AF_VSOCK support

 include/linux/vm_sockets_diag.h |  33 ++++++
 misc/ss.c                       | 238 +++++++++++++++++++++++++++++++++++-----
 man/man8/ss.8                   |   8 +-
 3 files changed, 249 insertions(+), 30 deletions(-)
 create mode 100644 include/linux/vm_sockets_diag.h

-- 
2.13.6

[PATCH iproute2 1/3] ss: allow AF_FAMILY constants >32

[Stefan Hajnoczi]

Linux has more than 32 address families defined in <bits/socket.h>.  Use
a 64-bit type so all of them can be represented in the filter->families
bitmask.

It's easy to introduce bugs when using (1 << AF_FAMILY) because the
value is 32-bit.  This can produce incorrect results from bitmask
operations so introduce the FAMILY_MASK() macro to eliminate these bugs.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 misc/ss.c | 54 ++++++++++++++++++++++++++++--------------------------
 1 file changed, 28 insertions(+), 26 deletions(-)

diff --git a/misc/ss.c b/misc/ss.c
index dd8dfaa4..12a31c90 100644
--- a/misc/ss.c
+++ b/misc/ss.c
@@ -170,55 +170,57 @@ enum {
 struct filter {
 	int dbs;
 	int states;
-	int families;
+	__u64 families;
 	struct ssfilter *f;
 	bool kill;
 };
 
+#define FAMILY_MASK(family) ((__u64)1 << (family))
+
 static const struct filter default_dbs[MAX_DB] = {
 	[TCP_DB] = {
 		.states   = SS_CONN,
-		.families = (1 << AF_INET) | (1 << AF_INET6),
+		.families = FAMILY_MASK(AF_INET) | FAMILY_MASK(AF_INET6),
 	},
 	[DCCP_DB] = {
 		.states   = SS_CONN,
-		.families = (1 << AF_INET) | (1 << AF_INET6),
+		.families = FAMILY_MASK(AF_INET) | FAMILY_MASK(AF_INET6),
 	},
 	[UDP_DB] = {
 		.states   = (1 << SS_ESTABLISHED),
-		.families = (1 << AF_INET) | (1 << AF_INET6),
+		.families = FAMILY_MASK(AF_INET) | FAMILY_MASK(AF_INET6),
 	},
 	[RAW_DB] = {
 		.states   = (1 << SS_ESTABLISHED),
-		.families = (1 << AF_INET) | (1 << AF_INET6),
+		.families = FAMILY_MASK(AF_INET) | FAMILY_MASK(AF_INET6),
 	},
 	[UNIX_DG_DB] = {
 		.states   = (1 << SS_CLOSE),
-		.families = (1 << AF_UNIX),
+		.families = FAMILY_MASK(AF_UNIX),
 	},
 	[UNIX_ST_DB] = {
 		.states   = SS_CONN,
-		.families = (1 << AF_UNIX),
+		.families = FAMILY_MASK(AF_UNIX),
 	},
 	[UNIX_SQ_DB] = {
 		.states   = SS_CONN,
-		.families = (1 << AF_UNIX),
+		.families = FAMILY_MASK(AF_UNIX),
 	},
 	[PACKET_DG_DB] = {
 		.states   = (1 << SS_CLOSE),
-		.families = (1 << AF_PACKET),
+		.families = FAMILY_MASK(AF_PACKET),
 	},
 	[PACKET_R_DB] = {
 		.states   = (1 << SS_CLOSE),
-		.families = (1 << AF_PACKET),
+		.families = FAMILY_MASK(AF_PACKET),
 	},
 	[NETLINK_DB] = {
 		.states   = (1 << SS_CLOSE),
-		.families = (1 << AF_NETLINK),
+		.families = FAMILY_MASK(AF_NETLINK),
 	},
 	[SCTP_DB] = {
 		.states   = SS_CONN,
-		.families = (1 << AF_INET) | (1 << AF_INET6),
+		.families = FAMILY_MASK(AF_INET) | FAMILY_MASK(AF_INET6),
 	},
 };
 
@@ -258,14 +260,14 @@ static void filter_db_set(struct filter *f, int db)
 static void filter_af_set(struct filter *f, int af)
 {
 	f->states	   |= default_afs[af].states;
-	f->families	   |= 1 << af;
+	f->families	   |= FAMILY_MASK(af);
 	do_default	    = 0;
 	preferred_family    = af;
 }
 
 static int filter_af_get(struct filter *f, int af)
 {
-	return f->families & (1 << af);
+	return !!(f->families & FAMILY_MASK(af));
 }
 
 static void filter_default_dbs(struct filter *f)
@@ -302,7 +304,7 @@ static void filter_merge_defaults(struct filter *f)
 			f->families |= default_dbs[db].families;
 	}
 	for (af = 0; af < AF_MAX; af++) {
-		if (!(f->families & (1 << af)))
+		if (!(f->families & FAMILY_MASK(af)))
 			continue;
 
 		if (!(default_afs[af].dbs & f->dbs))
@@ -2608,7 +2610,7 @@ static int show_one_inet_sock(const struct sockaddr_nl *addr,
 	struct inet_diag_msg *r = NLMSG_DATA(h);
 	struct sockstat s = {};
 
-	if (!(diag_arg->f->families & (1 << r->idiag_family)))
+	if (!(diag_arg->f->families & FAMILY_MASK(r->idiag_family)))
 		return 0;
 
 	parse_diag_msg(h, &s);
@@ -2802,7 +2804,7 @@ static int tcp_show(struct filter *f)
 		return -1;
 	}
 
-	if (f->families & (1<<AF_INET)) {
+	if (f->families & FAMILY_MASK(AF_INET)) {
 		if ((fp = net_tcp_open()) == NULL)
 			goto outerr;
 
@@ -2812,7 +2814,7 @@ static int tcp_show(struct filter *f)
 		fclose(fp);
 	}
 
-	if ((f->families & (1<<AF_INET6)) &&
+	if ((f->families & FAMILY_MASK(AF_INET6)) &&
 	    (fp = net_tcp6_open()) != NULL) {
 		setbuffer(fp, buf, bufsize);
 		if (generic_record_read(fp, tcp_show_line, f, AF_INET6))
@@ -2911,7 +2913,7 @@ static int udp_show(struct filter *f)
 	    && inet_show_netlink(f, NULL, IPPROTO_UDP) == 0)
 		return 0;
 
-	if (f->families&(1<<AF_INET)) {
+	if (f->families&FAMILY_MASK(AF_INET)) {
 		if ((fp = net_udp_open()) == NULL)
 			goto outerr;
 		if (generic_record_read(fp, dgram_show_line, f, AF_INET))
@@ -2919,7 +2921,7 @@ static int udp_show(struct filter *f)
 		fclose(fp);
 	}
 
-	if ((f->families&(1<<AF_INET6)) &&
+	if ((f->families&FAMILY_MASK(AF_INET6)) &&
 	    (fp = net_udp6_open()) != NULL) {
 		if (generic_record_read(fp, dgram_show_line, f, AF_INET6))
 			goto outerr;
@@ -2951,7 +2953,7 @@ static int raw_show(struct filter *f)
 	    inet_show_netlink(f, NULL, IPPROTO_RAW) == 0)
 		return 0;
 
-	if (f->families&(1<<AF_INET)) {
+	if (f->families&FAMILY_MASK(AF_INET)) {
 		if ((fp = net_raw_open()) == NULL)
 			goto outerr;
 		if (generic_record_read(fp, dgram_show_line, f, AF_INET))
@@ -2959,7 +2961,7 @@ static int raw_show(struct filter *f)
 		fclose(fp);
 	}
 
-	if ((f->families&(1<<AF_INET6)) &&
+	if ((f->families&FAMILY_MASK(AF_INET6)) &&
 	    (fp = net_raw6_open()) != NULL) {
 		if (generic_record_read(fp, dgram_show_line, f, AF_INET6))
 			goto outerr;
@@ -3703,13 +3705,13 @@ static int handle_follow_request(struct filter *f)
 	int groups = 0;
 	struct rtnl_handle rth;
 
-	if (f->families & (1 << AF_INET) && f->dbs & (1 << TCP_DB))
+	if (f->families & FAMILY_MASK(AF_INET) && f->dbs & (1 << TCP_DB))
 		groups |= 1 << (SKNLGRP_INET_TCP_DESTROY - 1);
-	if (f->families & (1 << AF_INET) && f->dbs & (1 << UDP_DB))
+	if (f->families & FAMILY_MASK(AF_INET) && f->dbs & (1 << UDP_DB))
 		groups |= 1 << (SKNLGRP_INET_UDP_DESTROY - 1);
-	if (f->families & (1 << AF_INET6) && f->dbs & (1 << TCP_DB))
+	if (f->families & FAMILY_MASK(AF_INET6) && f->dbs & (1 << TCP_DB))
 		groups |= 1 << (SKNLGRP_INET6_TCP_DESTROY - 1);
-	if (f->families & (1 << AF_INET6) && f->dbs & (1 << UDP_DB))
+	if (f->families & FAMILY_MASK(AF_INET6) && f->dbs & (1 << UDP_DB))
 		groups |= 1 << (SKNLGRP_INET6_UDP_DESTROY - 1);
 
 	if (groups == 0)
-- 
2.13.6

[PATCH iproute2 2/3] include: add <linux/vm_sockets_diag.h>

[Stefan Hajnoczi]

This new Linux header file defines the sock_diag interface used by
AF_VSOCK.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/linux/vm_sockets_diag.h | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 include/linux/vm_sockets_diag.h

diff --git a/include/linux/vm_sockets_diag.h b/include/linux/vm_sockets_diag.h
new file mode 100644
index 00000000..14cd7dc5
--- /dev/null
+++ b/include/linux/vm_sockets_diag.h
@@ -0,0 +1,33 @@
+/* AF_VSOCK sock_diag(7) interface for querying open sockets */
+
+#ifndef _UAPI__VM_SOCKETS_DIAG_H__
+#define _UAPI__VM_SOCKETS_DIAG_H__
+
+#include <linux/types.h>
+
+/* Request */
+struct vsock_diag_req {
+	__u8	sdiag_family;	/* must be AF_VSOCK */
+	__u8	sdiag_protocol;	/* must be 0 */
+	__u16	pad;		/* must be 0 */
+	__u32	vdiag_states;	/* query bitmap (e.g. 1 << TCP_LISTEN) */
+	__u32	vdiag_ino;	/* must be 0 (reserved) */
+	__u32	vdiag_show;	/* must be 0 (reserved) */
+	__u32	vdiag_cookie[2];
+};
+
+/* Response */
+struct vsock_diag_msg {
+	__u8	vdiag_family;	/* AF_VSOCK */
+	__u8	vdiag_type;	/* SOCK_STREAM or SOCK_DGRAM */
+	__u8	vdiag_state;	/* sk_state (e.g. TCP_LISTEN) */
+	__u8	vdiag_shutdown; /* local RCV_SHUTDOWN | SEND_SHUTDOWN */
+	__u32   vdiag_src_cid;
+	__u32   vdiag_src_port;
+	__u32   vdiag_dst_cid;
+	__u32   vdiag_dst_port;
+	__u32	vdiag_ino;
+	__u32	vdiag_cookie[2];
+};
+
+#endif /* _UAPI__VM_SOCKETS_DIAG_H__ */
-- 
2.13.6

[PATCH iproute2 3/3] ss: add AF_VSOCK support

[Stefan Hajnoczi]

The AF_VSOCK address family is a host<->guest communications channel
supported by VMware, KVM, and Hyper-V.  Initial VMware support was
released in Linux 3.9 in 2013 and transports for other hypervisors were
added later.

AF_VSOCK addresses are <u32 cid, u32 port> tuples.  The 32-bit cid
integer is comparable to an IP address.  AF_VSOCK ports work like
TCP/UDP ports.

Both SOCK_STREAM and SOCK_DGRAM socket types are available.

This patch adds AF_VSOCK support to ss(8) so that sockets can be
observed.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 misc/ss.c     | 184 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
 man/man8/ss.8 |   8 ++-
 2 files changed, 188 insertions(+), 4 deletions(-)

diff --git a/misc/ss.c b/misc/ss.c
index 12a31c90..164356a0 100644
--- a/misc/ss.c
+++ b/misc/ss.c
@@ -44,6 +44,7 @@
 #include <linux/packet_diag.h>
 #include <linux/netlink_diag.h>
 #include <linux/sctp.h>
+#include <linux/vm_sockets_diag.h>
 
 #define MAGIC_SEQ 123456
 
@@ -126,6 +127,8 @@ enum {
 	PACKET_R_DB,
 	NETLINK_DB,
 	SCTP_DB,
+	VSOCK_ST_DB,
+	VSOCK_DG_DB,
 	MAX_DB
 };
 
@@ -134,6 +137,7 @@ enum {
 #define ALL_DB ((1<<MAX_DB)-1)
 #define INET_L4_DBM ((1<<TCP_DB)|(1<<UDP_DB)|(1<<DCCP_DB)|(1<<SCTP_DB))
 #define INET_DBM (INET_L4_DBM | (1<<RAW_DB))
+#define VSOCK_DBM ((1<<VSOCK_ST_DB)|(1<<VSOCK_DG_DB))
 
 enum {
 	SS_UNKNOWN,
@@ -222,6 +226,14 @@ static const struct filter default_dbs[MAX_DB] = {
 		.states   = SS_CONN,
 		.families = FAMILY_MASK(AF_INET) | FAMILY_MASK(AF_INET6),
 	},
+	[VSOCK_ST_DB] = {
+		.states   = SS_CONN,
+		.families = FAMILY_MASK(AF_VSOCK),
+	},
+	[VSOCK_DG_DB] = {
+		.states   = SS_CONN,
+		.families = FAMILY_MASK(AF_VSOCK),
+	},
 };
 
 static const struct filter default_afs[AF_MAX] = {
@@ -245,6 +257,10 @@ static const struct filter default_afs[AF_MAX] = {
 		.dbs    = (1 << NETLINK_DB),
 		.states = (1 << SS_CLOSE),
 	},
+	[AF_VSOCK] = {
+		.dbs    = VSOCK_DBM,
+		.states = SS_CONN,
+	},
 };
 
 static int do_default = 1;
@@ -283,6 +299,8 @@ static void filter_default_dbs(struct filter *f)
 	filter_db_set(f, PACKET_DG_DB);
 	filter_db_set(f, NETLINK_DB);
 	filter_db_set(f, SCTP_DB);
+	filter_db_set(f, VSOCK_ST_DB);
+	filter_db_set(f, VSOCK_DG_DB);
 }
 
 static void filter_states_set(struct filter *f, int states)
@@ -791,6 +809,18 @@ static const char *proto_name(int protocol)
 	return "???";
 }
 
+static const char *vsock_netid_name(int type)
+{
+	switch (type) {
+	case SOCK_STREAM:
+		return "v_str";
+	case SOCK_DGRAM:
+		return "v_dgr";
+	default:
+		return "???";
+	}
+}
+
 static void sock_state_print(struct sockstat *s)
 {
 	const char *sock_name;
@@ -823,6 +853,9 @@ static void sock_state_print(struct sockstat *s)
 	case AF_NETLINK:
 		sock_name = "nl";
 		break;
+	case AF_VSOCK:
+		sock_name = vsock_netid_name(s->type);
+		break;
 	default:
 		sock_name = "unknown";
 	}
@@ -1149,6 +1182,8 @@ static int run_ssfilter(struct ssfilter *f, struct sockstat *s)
 			return s->lport == 0 && s->local.data[0] == 0;
 		if (s->local.family == AF_NETLINK)
 			return s->lport < 0;
+		if (s->local.family == AF_VSOCK)
+			return s->lport > 1023;
 
 		return is_ephemeral(s->lport);
 	}
@@ -1524,6 +1559,15 @@ void *parse_devcond(char *name)
 	return res;
 }
 
+static void vsock_set_inet_prefix(inet_prefix *a, __u32 cid)
+{
+	*a = (inet_prefix){
+		.bytelen = sizeof(cid),
+		.family = AF_VSOCK,
+	};
+	memcpy(a->data, &cid, sizeof(cid));
+}
+
 void *parse_hostcond(char *addr, bool is_port)
 {
 	char *port = NULL;
@@ -1598,6 +1642,37 @@ void *parse_hostcond(char *addr, bool is_port)
 		goto out;
 	}
 
+	if (fam == AF_VSOCK || strncmp(addr, "vsock:", 6) == 0) {
+		__u32 cid = ~(__u32)0;
+
+		a.addr.family = AF_VSOCK;
+		if (strncmp(addr, "vsock:", 6) == 0)
+			addr += 6;
+
+		if (is_port)
+			port = addr;
+		else {
+			port = strchr(addr, ':');
+			if (port) {
+				*port = '\0';
+				port++;
+			}
+		}
+
+		if (port && strcmp(port, "*") &&
+		    get_u32((__u32 *)&a.port, port, 0))
+			return NULL;
+
+		if (addr[0] && strcmp(addr, "*")) {
+			a.addr.bitlen = 32;
+			if (get_u32(&cid, addr, 0))
+				return NULL;
+		}
+		vsock_set_inet_prefix(&a.addr, cid);
+		fam = AF_VSOCK;
+		goto out;
+	}
+
 	if (fam == AF_INET || !strncmp(addr, "inet:", 5)) {
 		fam = AF_INET;
 		if (!strncmp(addr, "inet:", 5))
@@ -3674,6 +3749,88 @@ static int netlink_show(struct filter *f)
 	return 0;
 }
 
+static bool vsock_type_skip(struct sockstat *s, struct filter *f)
+{
+	if (s->type == SOCK_STREAM && !(f->dbs & (1 << VSOCK_ST_DB)))
+		return true;
+	if (s->type == SOCK_DGRAM && !(f->dbs & (1 << VSOCK_DG_DB)))
+		return true;
+	return false;
+}
+
+static void vsock_addr_print(inet_prefix *a, __u32 port)
+{
+	char cid_str[sizeof("4294967295")];
+	char port_str[sizeof("4294967295")];
+	__u32 cid;
+
+	memcpy(&cid, a->data, sizeof(cid));
+
+	if (cid == ~(__u32)0)
+		snprintf(cid_str, sizeof(cid_str), "*");
+	else
+		snprintf(cid_str, sizeof(cid_str), "%u", cid);
+
+	if (port == ~(__u32)0)
+		snprintf(port_str, sizeof(port_str), "*");
+	else
+		snprintf(port_str, sizeof(port_str), "%u", port);
+
+	sock_addr_print(cid_str, ":", port_str, NULL);
+}
+
+static void vsock_stats_print(struct sockstat *s, struct filter *f)
+{
+	sock_state_print(s);
+
+	vsock_addr_print(&s->local, s->lport);
+	vsock_addr_print(&s->remote, s->rport);
+
+	proc_ctx_print(s);
+
+	printf("\n");
+}
+
+static int vsock_show_sock(const struct sockaddr_nl *addr,
+			   struct nlmsghdr *nlh, void *arg)
+{
+	struct filter *f = (struct filter *)arg;
+	struct vsock_diag_msg *r = NLMSG_DATA(nlh);
+	struct sockstat stat = {
+		.type = r->vdiag_type,
+		.lport = r->vdiag_src_port,
+		.rport = r->vdiag_dst_port,
+		.state = r->vdiag_state,
+		.ino = r->vdiag_ino,
+	};
+
+	vsock_set_inet_prefix(&stat.local, r->vdiag_src_cid);
+	vsock_set_inet_prefix(&stat.remote, r->vdiag_dst_cid);
+
+	if (vsock_type_skip(&stat, f))
+		return 0;
+
+	if (f->f && run_ssfilter(f->f, &stat) == 0)
+		return 0;
+
+	vsock_stats_print(&stat, f);
+
+	return 0;
+}
+
+static int vsock_show(struct filter *f)
+{
+	DIAG_REQUEST(req, struct vsock_diag_req r);
+
+	if (!filter_af_get(f, AF_VSOCK))
+		return 0;
+
+	req.r.sdiag_family = AF_VSOCK;
+	req.r.vdiag_states = f->states;
+
+	return handle_netlink_request(f, &req.nlh, sizeof(req), vsock_show_sock);
+}
+
 struct sock_diag_msg {
 	__u8 sdiag_family;
 };
@@ -3694,6 +3851,8 @@ static int generic_show_sock(const struct sockaddr_nl *addr,
 		return packet_show_sock(addr, nlh, arg);
 	case AF_NETLINK:
 		return netlink_show_sock(addr, nlh, arg);
+	case AF_VSOCK:
+		return vsock_show_sock(addr, nlh, arg);
 	default:
 		return -1;
 	}
@@ -3921,14 +4080,15 @@ static void _usage(FILE *dest)
 "   -d, --dccp          display only DCCP sockets\n"
 "   -w, --raw           display only RAW sockets\n"
 "   -x, --unix          display only Unix domain sockets\n"
+"       --vsock         display only vsock sockets\n"
 "   -f, --family=FAMILY display sockets of type FAMILY\n"
-"       FAMILY := {inet|inet6|link|unix|netlink|help}\n"
+"       FAMILY := {inet|inet6|link|unix|netlink|vsock|help}\n"
 "\n"
 "   -K, --kill          forcibly close sockets, display what was closed\n"
 "   -H, --no-header     Suppress header line\n"
 "\n"
 "   -A, --query=QUERY, --socket=QUERY\n"
-"       QUERY := {all|inet|tcp|udp|raw|unix|unix_dgram|unix_stream|unix_seqpacket|packet|netlink}[,QUERY]\n"
+"       QUERY := {all|inet|tcp|udp|raw|unix|unix_dgram|unix_stream|unix_seqpacket|packet|netlink|vsock_stream|vsock_dgram}[,QUERY]\n"
 "\n"
 "   -D, --diag=FILE     Dump raw information about TCP sockets to FILE\n"
 "   -F, --filter=FILE   read filter information from FILE\n"
@@ -4001,6 +4161,9 @@ static int scan_state(const char *state)
 	exit(-1);
 }
 
+/* Values 'v' and 'V' are already used so a non-character is used */
+#define OPT_VSOCK 256
+
 static const struct option long_opts[] = {
 	{ "numeric", 0, 0, 'n' },
 	{ "resolve", 0, 0, 'r' },
@@ -4017,6 +4180,7 @@ static const struct option long_opts[] = {
 	{ "udp", 0, 0, 'u' },
 	{ "raw", 0, 0, 'w' },
 	{ "unix", 0, 0, 'x' },
+	{ "vsock", 0, 0, OPT_VSOCK },
 	{ "all", 0, 0, 'a' },
 	{ "listening", 0, 0, 'l' },
 	{ "ipv4", 0, 0, '4' },
@@ -4102,6 +4266,9 @@ int main(int argc, char *argv[])
 		case 'x':
 			filter_af_set(&current_filter, AF_UNIX);
 			break;
+		case OPT_VSOCK:
+			filter_af_set(&current_filter, AF_VSOCK);
+			break;
 		case 'a':
 			state_filter = SS_ALL;
 			break;
@@ -4128,6 +4295,8 @@ int main(int argc, char *argv[])
 				filter_af_set(&current_filter, AF_UNIX);
 			else if (strcmp(optarg, "netlink") == 0)
 				filter_af_set(&current_filter, AF_NETLINK);
+			else if (strcmp(optarg, "vsock") == 0)
+				filter_af_set(&current_filter, AF_VSOCK);
 			else if (strcmp(optarg, "help") == 0)
 				help();
 			else {
@@ -4193,6 +4362,15 @@ int main(int argc, char *argv[])
 					filter_db_set(&current_filter, PACKET_DG_DB);
 				} else if (strcmp(p, "netlink") == 0) {
 					filter_db_set(&current_filter, NETLINK_DB);
+				} else if (strcmp(p, "vsock") == 0) {
+					filter_db_set(&current_filter, VSOCK_ST_DB);
+					filter_db_set(&current_filter, VSOCK_DG_DB);
+				} else if (strcmp(p, "vsock_stream") == 0 ||
+					   strcmp(p, "v_str") == 0) {
+					filter_db_set(&current_filter, VSOCK_ST_DB);
+				} else if (strcmp(p, "vsock_dgram") == 0 ||
+					   strcmp(p, "v_dgr") == 0) {
+					filter_db_set(&current_filter, VSOCK_DG_DB);
 				} else {
 					fprintf(stderr, "ss: \"%s\" is illegal socket table id\n", p);
 					usage();
@@ -4408,6 +4586,8 @@ int main(int argc, char *argv[])
 		dccp_show(&current_filter);
 	if (current_filter.dbs & (1<<SCTP_DB))
 		sctp_show(&current_filter);
+	if (current_filter.dbs & VSOCK_DBM)
+		vsock_show(&current_filter);
 
 	if (show_users || show_proc_ctx || show_sock_ctx)
 		user_ent_destroy();
diff --git a/man/man8/ss.8 b/man/man8/ss.8
index 3bec97f0..3af509e9 100644
--- a/man/man8/ss.8
+++ b/man/man8/ss.8
@@ -125,14 +125,18 @@ Display Unix domain sockets (alias for -f unix).
 .B \-S, \-\-sctp
 Display SCTP sockets.
 .TP
+.B \-\-vsock
+Display vsock sockets (alias for -f vsock).
+.TP
 .B \-f FAMILY, \-\-family=FAMILY
 Display sockets of type FAMILY.
-Currently the following families are supported: unix, inet, inet6, link, netlink.
+Currently the following families are supported: unix, inet, inet6, link, netlink, vsock.
 .TP
 .B \-A QUERY, \-\-query=QUERY, \-\-socket=QUERY
 List of socket tables to dump, separated by commas. The following identifiers
 are understood: all, inet, tcp, udp, raw, unix, packet, netlink, unix_dgram,
-unix_stream, unix_seqpacket, packet_raw, packet_dgram, dccp, sctp.
+unix_stream, unix_seqpacket, packet_raw, packet_dgram, dccp, sctp,
+vsock_stream, vsock_dgram.
 .TP
 .B \-D FILE, \-\-diag=FILE
 Do not display anything, just dump raw information about TCP sockets to FILE after applying filters. If FILE is - stdout is used.
-- 
2.13.6

Re: [PATCH iproute2 2/3] include: add <linux/vm_sockets_diag.h>

[Stephen Hemminger]

On Tue,  3 Oct 2017 13:57:43 -0400
Stefan Hajnoczi <stefanha@redhat.com> wrote:

> This new Linux header file defines the sock_diag interface used by
> AF_VSOCK.
> 
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> ---
>  include/linux/vm_sockets_diag.h | 33 +++++++++++++++++++++++++++++++++
>  1 file changed, 33 insertions(+)
>  create mode 100644 include/linux/vm_sockets_diag.h
> 

All header files in linux must be headers that come from 'make install_headers'
in kernel. I don't see vm_sockets_diag.h even in net-next.

Re: [PATCH iproute2 1/3] ss: allow AF_FAMILY constants >32

[Stephen Hemminger]

On Tue,  3 Oct 2017 13:57:42 -0400
Stefan Hajnoczi <stefanha@redhat.com> wrote:

> Linux has more than 32 address families defined in <bits/socket.h>.  Use
> a 64-bit type so all of them can be represented in the filter->families
> bitmask.
> 
> It's easy to introduce bugs when using (1 << AF_FAMILY) because the
> value is 32-bit.  This can produce incorrect results from bitmask
> operations so introduce the FAMILY_MASK() macro to eliminate these bugs.
> 
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> ---
>  misc/ss.c | 54 ++++++++++++++++++++++++++++--------------------------
>  1 file changed, 28 insertions(+), 26 deletions(-)
> 
> diff --git a/misc/ss.c b/misc/ss.c
> index dd8dfaa4..12a31c90 100644
> --- a/misc/ss.c
> +++ b/misc/ss.c
> @@ -170,55 +170,57 @@ enum {
>  struct filter {
>  	int dbs;
>  	int states;
> -	int families;
> +	__u64 families;

Since this isn't a value that is coming from kernel. It should be uint64_t
rather than __u64.

Re: [PATCH iproute2 1/3] ss: allow AF_FAMILY constants >32

[Stefan Hajnoczi]

On Tue, Oct 03, 2017 at 11:26:53AM -0700, Stephen Hemminger wrote:
> On Tue,  3 Oct 2017 13:57:42 -0400
> Stefan Hajnoczi <stefanha@redhat.com> wrote:
> 
> > Linux has more than 32 address families defined in <bits/socket.h>.  Use
> > a 64-bit type so all of them can be represented in the filter->families
> > bitmask.
> > 
> > It's easy to introduce bugs when using (1 << AF_FAMILY) because the
> > value is 32-bit.  This can produce incorrect results from bitmask
> > operations so introduce the FAMILY_MASK() macro to eliminate these bugs.
> > 
> > Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> > ---
> >  misc/ss.c | 54 ++++++++++++++++++++++++++++--------------------------
> >  1 file changed, 28 insertions(+), 26 deletions(-)
> > 
> > diff --git a/misc/ss.c b/misc/ss.c
> > index dd8dfaa4..12a31c90 100644
> > --- a/misc/ss.c
> > +++ b/misc/ss.c
> > @@ -170,55 +170,57 @@ enum {
> >  struct filter {
> >  	int dbs;
> >  	int states;
> > -	int families;
> > +	__u64 families;
> 
> Since this isn't a value that is coming from kernel. It should be uint64_t
> rather than __u64.

Okay, will fix in v2.

RE: [PATCH iproute2 1/3] ss: allow AF_FAMILY constants >32

[David Laight]

From: Stefan Hajnoczi
> Sent: 04 October 2017 16:01
...
> > > --- a/misc/ss.c
> > > +++ b/misc/ss.c
> > > @@ -170,55 +170,57 @@ enum {
> > >  struct filter {
> > >  	int dbs;
> > >  	int states;
> > > -	int families;
> > > +	__u64 families;
> >
> > Since this isn't a value that is coming from kernel. It should be uint64_t
> > rather than __u64.
> 
> Okay, will fix in v2.

But that looks like a kernel structure, not one exposed to userspace.
So surely __u64 is correct.

	David

Re: [PATCH iproute2 2/3] include: add <linux/vm_sockets_diag.h>

[Stefan Hajnoczi]

On Tue, Oct 03, 2017 at 11:24:55AM -0700, Stephen Hemminger wrote:
> On Tue,  3 Oct 2017 13:57:43 -0400
> Stefan Hajnoczi <stefanha@redhat.com> wrote:
> 
> > This new Linux header file defines the sock_diag interface used by
> > AF_VSOCK.
> > 
> > Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> > ---
> >  include/linux/vm_sockets_diag.h | 33 +++++++++++++++++++++++++++++++++
> >  1 file changed, 33 insertions(+)
> >  create mode 100644 include/linux/vm_sockets_diag.h
> > 
> 
> All header files in linux must be headers that come from 'make install_headers'
> in kernel. I don't see vm_sockets_diag.h even in net-next.

This series depends on "[PATCH 0/5] VSOCK: add sock_diag interface".
Once that has been merged in Linux the <linux/vm_sockets_diag.h> header
will be available.

I sent the ss(8) patch at the same time and referenced the Linux patch
series in the cover letter so they can be reviewed/tested together.

Thanks,
Stefan

Re: [PATCH iproute2 1/3] ss: allow AF_FAMILY constants >32

[Stephen Hemminger]

On Wed, 4 Oct 2017 15:03:09 +0000
David Laight <David.Laight@ACULAB.COM> wrote:

> From: Stefan Hajnoczi
> > Sent: 04 October 2017 16:01  
> ...
> > > > --- a/misc/ss.c
> > > > +++ b/misc/ss.c
> > > > @@ -170,55 +170,57 @@ enum {
> > > >  struct filter {
> > > >  	int dbs;
> > > >  	int states;
> > > > -	int families;
> > > > +	__u64 families;  
> > >
> > > Since this isn't a value that is coming from kernel. It should be uint64_t
> > > rather than __u64.  
> > 
> > Okay, will fix in v2.  
> 
> But that looks like a kernel structure, not one exposed to userspace.
> So surely __u64 is correct.
> 
> 	David
> 

families is a bit field which is built from data, not from kernel.