]> git.itanic.dy.fi Git - linux-stable/commitdiff
vsock: use ns_capable_noaudit() on socket create
authorJeff Vander Stoep <jeffv@google.com>
Fri, 23 Oct 2020 14:37:57 +0000 (16:37 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Tue, 10 Nov 2020 11:37:30 +0000 (12:37 +0100)
[ Upstream commit af545bb5ee53f5261db631db2ac4cde54038bdaf ]

During __vsock_create() CAP_NET_ADMIN is used to determine if the
vsock_sock->trusted should be set to true. This value is used later
for determing if a remote connection should be allowed to connect
to a restricted VM. Unfortunately, if the caller doesn't have
CAP_NET_ADMIN, an audit message such as an selinux denial is
generated even if the caller does not want a trusted socket.

Logging errors on success is confusing. To avoid this, switch the
capable(CAP_NET_ADMIN) check to the noaudit version.

Reported-by: Roman Kiryanov <rkir@google.com>
https://android-review.googlesource.com/c/device/generic/goldfish/+/1468545/
Signed-off-by: Jeff Vander Stoep <jeffv@google.com>
Reviewed-by: James Morris <jamorris@linux.microsoft.com>
Link: https://lore.kernel.org/r/20201023143757.377574-1-jeffv@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
net/vmw_vsock/af_vsock.c

index 7bd6c8199ca678a2021ec9b4c7abe11e7121b77f..3a074a03d38208eea53088e249136cbc3a97a17a 100644 (file)
@@ -621,7 +621,7 @@ struct sock *__vsock_create(struct net *net,
                vsk->owner = get_cred(psk->owner);
                vsk->connect_timeout = psk->connect_timeout;
        } else {
-               vsk->trusted = capable(CAP_NET_ADMIN);
+               vsk->trusted = ns_capable_noaudit(&init_user_ns, CAP_NET_ADMIN);
                vsk->owner = get_current_cred();
                vsk->connect_timeout = VSOCK_DEFAULT_CONNECT_TIMEOUT;
        }