[Devel] [PATCH 2/3] C/R: Fix storing IPv6 addresses and handle the "ipv6only" socket flag

[Dan Smith]

The first item is a result of sockaddr_in6 being larger than the base
sockaddr structure, thus not being long enough to reserve enough space in
the checkpoint header.

The second comes into play when things (like sshd) bind to INADDR6_ANY,
set the "ipv6only" socket flag and then bind an IPv4 socket to the same
port.

Signed-off-by: Dan Smith <danms at us.ibm.com>
---
 include/linux/checkpoint_hdr.h |   13 +++++++++++--
 net/ipv4/checkpoint.c          |   34 ++++++++++++++++++++--------------
 net/ipv6/af_inet6.c            |    2 ++
 3 files changed, 33 insertions(+), 16 deletions(-)

diff --git a/include/linux/checkpoint_hdr.h b/include/linux/checkpoint_hdr.h
index 90e4934..e771758 100644
--- a/include/linux/checkpoint_hdr.h
+++ b/include/linux/checkpoint_hdr.h
@@ -758,12 +758,21 @@ struct ckpt_hdr_socket_inet {
 		struct in6_addr saddr;
 		struct in6_addr rcv_saddr;
 		struct in6_addr daddr;
+		__u8 ipv6only;
 	} inet6 __attribute__ ((aligned(8)));
 
 	__u32 laddr_len;
 	__u32 raddr_len;
-	struct sockaddr_in laddr;
-	struct sockaddr_in raddr;
+	union {
+		struct sockaddr laddr;
+		struct sockaddr_in laddr4;
+		struct sockaddr_in6 laddr6;
+	};
+	union {
+		struct sockaddr raddr;
+		struct sockaddr_in raddr4;
+		struct sockaddr_in6 raddr6;
+	};
 } __attribute__((aligned(8)));
 
 struct ckpt_hdr_file_socket {
diff --git a/net/ipv4/checkpoint.c b/net/ipv4/checkpoint.c
index b4024e7..57b185d 100644
--- a/net/ipv4/checkpoint.c
+++ b/net/ipv4/checkpoint.c
@@ -190,12 +190,14 @@ static int sock_inet_restore_connection(struct sock *sk,
 	struct inet_sock *inet = inet_sk(sk);
 	int tcp_gso = sk->sk_family == AF_INET ? SKB_GSO_TCPV4 : SKB_GSO_TCPV6;
 
-	inet->inet_daddr = hh->raddr.sin_addr.s_addr;
-	inet->inet_saddr = hh->laddr.sin_addr.s_addr;
-	inet->inet_rcv_saddr = inet->inet_saddr;
+	if (sk->sk_family == AF_INET) {
+		inet->inet_daddr = hh->raddr4.sin_addr.s_addr;
+		inet->inet_saddr = hh->laddr4.sin_addr.s_addr;
+		inet->inet_rcv_saddr = inet->inet_saddr;
 
-	inet->inet_dport = hh->raddr.sin_port;
-	inet->inet_sport = hh->laddr.sin_port;
+		inet->inet_dport = hh->raddr4.sin_port;
+		inet->inet_sport = hh->laddr4.sin_port;
+	}
 
 	if (sk->sk_protocol == IPPROTO_TCP)
 		sk->sk_gso_type = tcp_gso;
@@ -266,6 +268,7 @@ static int sock_inet_cptrst(struct ckpt_ctx *ctx,
 			ipv6_addr_copy(&inet6->rcv_saddr, &hh->inet6.rcv_saddr);
 			ipv6_addr_copy(&inet6->daddr, &hh->inet6.daddr);
 		}
+		CKPT_COPY(op, hh->inet6.ipv6only, inet6->ipv6only);
 	}
 
 	return ret;
@@ -281,8 +284,8 @@ int inet_checkpoint(struct ckpt_ctx *ctx, struct socket *sock)
 		return -EINVAL;
 
 	ret = ckpt_sock_getnames(ctx, sock,
-				(struct sockaddr *)&in->laddr, &in->laddr_len,
-				(struct sockaddr *)&in->raddr, &in->raddr_len);
+				&in->laddr, &in->laddr_len,
+				&in->raddr, &in->raddr_len);
 	if (ret)
 		goto out;
 
@@ -387,19 +390,19 @@ static int inet_precheck(struct socket *sock, struct ckpt_hdr_socket_inet *in)
 	__u8 nonagle_mask = TCP_NAGLE_OFF | TCP_NAGLE_CORK | TCP_NAGLE_PUSH;
 	__u8 ecn_mask = TCP_ECN_OK | TCP_ECN_QUEUE_CWR | TCP_ECN_DEMAND_CWR;
 
-	if ((htons(in->laddr.sin_port) < PROT_SOCK) &&
+	if ((htons(in->laddr4.sin_port) < PROT_SOCK) &&
 	    !capable(CAP_NET_BIND_SERVICE)) {
 		ckpt_debug("unable to bind to port %hu\n",
-			   htons(in->laddr.sin_port));
+			   htons(in->laddr4.sin_port));
 		return -EINVAL;
 	}
 
-	if (in->laddr_len > sizeof(struct sockaddr_in)) {
+	if (in->laddr_len > sizeof(in->laddr6)) {
 		ckpt_debug("laddr_len is too big\n");
 		return -EINVAL;
 	}
 
-	if (in->raddr_len > sizeof(struct sockaddr_in)) {
+	if (in->raddr_len > sizeof(in->laddr6)) {
 		ckpt_debug("raddr_len is too big\n");
 		return -EINVAL;
 	}
@@ -498,13 +501,16 @@ int inet_restore(struct ckpt_ctx *ctx,
 	    ((h->sock.state == TCP_CLOSE) && (in->laddr_len > 0))) {
 		sock->sk->sk_reuse = 2;
 		inet_sk(sock->sk)->freebind = 1;
-		ret = sock->ops->bind(sock,
-				      (struct sockaddr *)&in->laddr,
-				      in->laddr_len);
+		ret = sock->ops->bind(sock, &in->laddr, in->laddr_len);
 		ckpt_debug("inet bind: %i\n", ret);
 		if (ret < 0)
 			goto out;
 
+		if (in->inet6.ipv6only) {
+			struct ipv6_pinfo *np = inet6_sk(sock->sk);
+			np->ipv6only = 1;
+		}
+
 		if (h->sock.state == TCP_LISTEN) {
 			ret = sock->ops->listen(sock, h->sock.backlog);
 			ckpt_debug("inet listen: %i\n", ret);
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index 12e69d3..9acb55a 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -523,6 +523,8 @@ const struct proto_ops inet6_stream_ops = {
 	.mmap		   = sock_no_mmap,
 	.sendpage	   = tcp_sendpage,
 	.splice_read	   = tcp_splice_read,
+	.checkpoint	   = inet_checkpoint,
+	.restore	   = inet_restore,
 #ifdef CONFIG_COMPAT
 	.compat_setsockopt = compat_sock_common_setsockopt,
 	.compat_getsockopt = compat_sock_common_getsockopt,
-- 
1.6.2.5

_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers