mptcp: introduce MPTCP_FULL_INFO getsockopt

Some user-space applications want to monitor the subflows utilization.

Dumping the per subflow tcp_info is not enough, as the PM could close
and re-create the subflows under-the-hood, fooling the accounting.
Even checking the src/dst addresses used by each subflow could not
be enough, because new subflows could re-use the same address/port of
the just closed one.

This patch introduces a new socket option, allow dumping all the relevant
information all-at-once (everything, everywhere...), in a consistent
manner.

Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/388
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

include/uapi/linux/mptcp.h

@@ -249,9 +249,33 @@ struct mptcp_subflow_addrs {
 	};
 };
 
+struct mptcp_subflow_info {
+	__u32				id;
+	struct mptcp_subflow_addrs	addrs;
+};
+
+struct mptcp_full_info {
+	__u32		size_tcpinfo_kernel;	/* must be 0, set by kernel */
+	__u32		size_tcpinfo_user;
+	__u32		size_sfinfo_kernel;	/* must be 0, set by kernel */
+	__u32		size_sfinfo_user;
+	__u32		num_subflows;		/* must be 0, set by kernel (real subflow count) */
+	__u32		size_arrays_user;	/* max subflows that userspace is interested in;
+						 * the buffers at subflow_info/tcp_info
+						 * are respectively at least:
+						 *  size_arrays * size_sfinfo_user
+						 *  size_arrays * size_tcpinfo_user
+						 * bytes wide
+						 */
+	__aligned_u64		subflow_info;
+	__aligned_u64		tcp_info;
+	struct mptcp_info	mptcp_info;
+};
+
 /* MPTCP socket options */
 #define MPTCP_INFO		1
 #define MPTCP_TCPINFO		2
 #define MPTCP_SUBFLOW_ADDRS	3
+#define MPTCP_FULL_INFO		4
 
 #endif /* _UAPI_MPTCP_H */

net/mptcp/sockopt.c

@@ -15,6 +15,7 @@
 #include "protocol.h"
 
 #define MIN_INFO_OPTLEN_SIZE		16
+#define MIN_FULL_INFO_OPTLEN_SIZE	40
 
 static struct sock *__mptcp_tcp_fallback(struct mptcp_sock *msk)
 {
@@ -981,7 +982,8 @@ static int mptcp_put_subflow_data(struct mptcp_subflow_data *sfd,
 }
 
 static int mptcp_get_subflow_data(struct mptcp_subflow_data *sfd,
-				  char __user *optval, int __user *optlen)
+				  char __user *optval,
+				  int __user *optlen)
 {
 	int len, copylen;
 
@@ -1162,6 +1164,125 @@ static int mptcp_getsockopt_subflow_addrs(struct mptcp_sock *msk, char __user *o
 	return 0;
 }
 
+static int mptcp_get_full_info(struct mptcp_full_info *mfi,
+			       char __user *optval,
+			       int __user *optlen)
+{
+	int len;
+
+	BUILD_BUG_ON(offsetof(struct mptcp_full_info, mptcp_info) !=
+		     MIN_FULL_INFO_OPTLEN_SIZE);
+
+	if (get_user(len, optlen))
+		return -EFAULT;
+
+	if (len < MIN_FULL_INFO_OPTLEN_SIZE)
+		return -EINVAL;
+
+	memset(mfi, 0, sizeof(*mfi));
+	if (copy_from_user(mfi, optval, MIN_FULL_INFO_OPTLEN_SIZE))
+		return -EFAULT;
+
+	if (mfi->size_tcpinfo_kernel ||
+	    mfi->size_sfinfo_kernel ||
+	    mfi->num_subflows)
+		return -EINVAL;
+
+	if (mfi->size_sfinfo_user > INT_MAX ||
+	    mfi->size_tcpinfo_user > INT_MAX)
+		return -EINVAL;
+
+	return len - MIN_FULL_INFO_OPTLEN_SIZE;
+}
+
+static int mptcp_put_full_info(struct mptcp_full_info *mfi,
+			       char __user *optval,
+			       u32 copylen,
+			       int __user *optlen)
+{
+	copylen += MIN_FULL_INFO_OPTLEN_SIZE;
+	if (put_user(copylen, optlen))
+		return -EFAULT;
+
+	if (copy_to_user(optval, mfi, copylen))
+		return -EFAULT;
+	return 0;
+}
+
+static int mptcp_getsockopt_full_info(struct mptcp_sock *msk, char __user *optval,
+				      int __user *optlen)
+{
+	unsigned int sfcount = 0, copylen = 0;
+	struct mptcp_subflow_context *subflow;
+	struct sock *sk = (struct sock *)msk;
+	void __user *tcpinfoptr, *sfinfoptr;
+	struct mptcp_full_info mfi;
+	int len;
+
+	len = mptcp_get_full_info(&mfi, optval, optlen);
+	if (len < 0)
+		return len;
+
+	/* don't bother filling the mptcp info if there is not enough
+	 * user-space-provided storage
+	 */
+	if (len > 0) {
+		mptcp_diag_fill_info(msk, &mfi.mptcp_info);
+		copylen += min_t(unsigned int, len, sizeof(struct mptcp_info));
+	}
+
+	mfi.size_tcpinfo_kernel = sizeof(struct tcp_info);
+	mfi.size_tcpinfo_user = min_t(unsigned int, mfi.size_tcpinfo_user,
+				      sizeof(struct tcp_info));
+	sfinfoptr = u64_to_user_ptr(mfi.subflow_info);
+	mfi.size_sfinfo_kernel = sizeof(struct mptcp_subflow_info);
+	mfi.size_sfinfo_user = min_t(unsigned int, mfi.size_sfinfo_user,
+				     sizeof(struct mptcp_subflow_info));
+	tcpinfoptr = u64_to_user_ptr(mfi.tcp_info);
+
+	lock_sock(sk);
+	mptcp_for_each_subflow(msk, subflow) {
+		struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
+		struct mptcp_subflow_info sfinfo;
+		struct tcp_info tcp_info;
+
+		if (sfcount++ >= mfi.size_arrays_user)
+			continue;
+
+		/* fetch addr/tcp_info only if the user space buffers
+		 * are wide enough
+		 */
+		memset(&sfinfo, 0, sizeof(sfinfo));
+		sfinfo.id = subflow->subflow_id;
+		if (mfi.size_sfinfo_user >
+		    offsetof(struct mptcp_subflow_info, addrs))
+			mptcp_get_sub_addrs(ssk, &sfinfo.addrs);
+		if (copy_to_user(sfinfoptr, &sfinfo, mfi.size_sfinfo_user))
+			goto fail_release;
+
+		if (mfi.size_tcpinfo_user) {
+			tcp_get_info(ssk, &tcp_info);
+			if (copy_to_user(tcpinfoptr, &tcp_info,
+					 mfi.size_tcpinfo_user))
+				goto fail_release;
+		}
+
+		tcpinfoptr += mfi.size_tcpinfo_user;
+		sfinfoptr += mfi.size_sfinfo_user;
+	}
+	release_sock(sk);
+
+	mfi.num_subflows = sfcount;
+	if (mptcp_put_full_info(&mfi, optval, copylen, optlen))
+		return -EFAULT;
+
+	return 0;
+
+fail_release:
+	release_sock(sk);
+	return -EFAULT;
+}
+
 static int mptcp_put_int_option(struct mptcp_sock *msk, char __user *optval,
 				int __user *optlen, int val)
 {
@@ -1235,6 +1356,8 @@ static int mptcp_getsockopt_sol_mptcp(struct mptcp_sock *msk, int optname,
 	switch (optname) {
 	case MPTCP_INFO:
 		return mptcp_getsockopt_info(msk, optval, optlen);
+	case MPTCP_FULL_INFO:
+		return mptcp_getsockopt_full_info(msk, optval, optlen);
 	case MPTCP_TCPINFO:
 		return mptcp_getsockopt_tcpinfo(msk, optval, optlen);
 	case MPTCP_SUBFLOW_ADDRS: