[PATCH net v4 1/3] vsock/bpf: Fix EINTR connect() racing sockmap update

[Michal Luczaj]

Signal delivery during connect() may result in a disconnect of an already
TCP_ESTABLISHED socket. Problem is that such established socket might have
been placed in a sockmap before the connection was closed. We end up with a
SS_UNCONNECTED vsock in a sockmap. And this, combined with the ability to
reassign (unconnected) vsock's transport to NULL, breaks the sockmap
contract. As manifested by WARN_ON_ONCE.

connect
  / state = SS_CONNECTED /
                                sock_map_update_elem
  if signal_pending
    state = SS_UNCONNECTED

connect
  transport = NULL
                                vsock_bpf_recvmsg
                                  WARN_ON_ONCE(!vsk->transport)

Ensure the socket does not stay in sockmap.

WARNING: CPU: 8 PID: 1228 at net/vmw_vsock/vsock_bpf.c:97 
vsock_bpf_recvmsg+0xb43/0xe00
CPU: 8 UID: 0 PID: 1228 Comm: a.out Not tainted 6.14.0-rc5+
RIP: 0010:vsock_bpf_recvmsg+0xb43/0xe00
 sock_recvmsg+0x1b2/0x220
 __sys_recvfrom+0x190/0x270
 __x64_sys_recvfrom+0xdc/0x1b0
 do_syscall_64+0x93/0x1b0
 entry_SYSCALL_64_after_hwframe+0x76/0x7e

Fixes: 634f1a7110b4 ("vsock: support sockmap")
Reviewed-by: Stefano Garzarella <sgarz...@redhat.com>
Signed-off-by: Michal Luczaj <m...@rbox.co>
---
 net/vmw_vsock/af_vsock.c  | 10 +++++++++-
 net/vmw_vsock/vsock_bpf.c |  1 +
 2 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
index 
7e3db87ae4333cf63327ec105ca99253569bb9fe..81b1b8e9c946a646778367ab78ca180cef75ef72
 100644
--- a/net/vmw_vsock/af_vsock.c
+++ b/net/vmw_vsock/af_vsock.c
@@ -1579,7 +1579,15 @@ static int vsock_connect(struct socket *sock, struct 
sockaddr *addr,
 
                if (signal_pending(current)) {
                        err = sock_intr_errno(timeout);
-                       sk->sk_state = sk->sk_state == TCP_ESTABLISHED ? 
TCP_CLOSING : TCP_CLOSE;
+                       if (sk->sk_state == TCP_ESTABLISHED) {
+                               /* Might have raced with a sockmap update. */
+                               if (sk->sk_prot->unhash)
+                                       sk->sk_prot->unhash(sk);
+
+                               sk->sk_state = TCP_CLOSING;
+                       } else {
+                               sk->sk_state = TCP_CLOSE;
+                       }
                        sock->state = SS_UNCONNECTED;
                        vsock_transport_cancel_pkt(vsk);
                        vsock_remove_connected(vsk);
diff --git a/net/vmw_vsock/vsock_bpf.c b/net/vmw_vsock/vsock_bpf.c
index 
07b96d56f3a577af71021b1b8132743554996c4f..c68fdaf09046b68254dac3ea70ffbe73dfa45cef
 100644
--- a/net/vmw_vsock/vsock_bpf.c
+++ b/net/vmw_vsock/vsock_bpf.c
@@ -127,6 +127,7 @@ static void vsock_bpf_rebuild_protos(struct proto *prot, 
const struct proto *bas
 {
        *prot        = *base;
        prot->close  = sock_map_close;
+       prot->unhash = sock_map_unhash;
        prot->recvmsg = vsock_bpf_recvmsg;
        prot->sock_is_readable = sk_msg_is_readable;
 }

-- 
2.48.1