mptcp: sockopt: info: stop early if no buffer

Up to recently, it has been recommended to use getsockopt(MPTCP_INFO) on
an 'accept'ed socket, for a server app to check if the client requested
to use MPTCP.

In this case, the userspace app is only interested by the returned value
of the getsocktop() call, and can then give 0 for the option length, and
NULL for the buffer address. An easy optimisation is then to stop early,
and avoid filling a local buffer -- which now requires two different
locks -- if it is not needed.

Note that userspace apps should use getsockopt(SO_PROTOCOL) in such case
instead: it looks less like a workaround, and it works with any kernel
versions, while the MPTCP_INFO method requires kernels >= v5.16.

Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>

1 files changed, 6 insertions, 0 deletions

diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index 1fea43f5b6f3de..b0d1dc4df0c1a7 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -960,6 +960,12 @@ static int mptcp_getsockopt_info(struct mptcp_sock *msk, char __user *optval, in
 	if (get_user(len, optlen))
 		return -EFAULT;
 
+	/* Opti when used to check if a fallback to TCP happened on an 'accept'
+	 * socket. Userspace apps should use getsockopt(SO_PROTOCOL) instead.
+	 */
+	if (len == 0)
+		return 0;
+
 	len = min_t(unsigned int, len, sizeof(struct mptcp_info));
 
 	mptcp_diag_fill_info(msk, &m_info);