[PATCH net] net/smc: hash socket only after full initialisation in smc_sk_init()

smc_sk_init() calls sk->sk_prot->hash(sk) before several fields are
fully initialised: clcsock_release_lock, the saved clcsk_* callbacks,
use_fallback/fallback_rsn, and conn.close_work.  Once hash() returns the
socket is visible to concurrent hash walkers, which can then observe
uninitialised state.

Move hash(sk) to the end of smc_sk_init() so the socket is published
only after it is fully constructed.

Fixes: d0e35656d834 ("net/smc: refactoring initialization of smc sock")
Reviewed-by: Hidayath Khan <redacted>
Reviewed-by: Sidraya Jayagond <sidraya@linux.ibm.com>
Signed-off-by: Mahanta Jambigi <mjambigi@linux.ibm.com>
---
Found by syzbot while testing a diag dump patch that drops the hash
read lock before calling mutex_lock(&smc->clcsock_release_lock).  That
change made the uninitialised mutex reachable for the first time and
triggered DEBUG_LOCKS_WARN_ON(lock->magic != lock) immediately.

syzbot report:
https://ci.syzbot.org/series/ab0d12fd-741a-4a0f-9f49-e52b84e94f4a

 net/smc/af_smc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c
index 00403175b740..5e7560417ec7 100644
--- a/net/smc/af_smc.c
+++ b/net/smc/af_smc.c

@@ -409,13 +409,13 @@ void smc_sk_init(struct net *net, struct sock *sk, int protocol)
 				      "sk_lock-AF_SMC", &smc_key);
 	spin_lock_init(&smc->accept_q_lock);
 	spin_lock_init(&smc->conn.send_lock);
-	sk->sk_prot->hash(sk);
 	mutex_init(&smc->clcsock_release_lock);
 	smc_init_saved_callbacks(smc);
 	smc->limit_smc_hs = net->smc.limit_smc_hs;
 	smc->use_fallback = false; /* assume rdma capability first */
 	smc->fallback_rsn = 0;
 	smc_close_init(smc);
+	sk->sk_prot->hash(sk);
 }
 
 static struct sock *smc_sock_alloc(struct net *net, struct socket *sock,

-- 
2.50.1