sctp: Fetch inet6_sk() after setting ->pinet6 in sctp_clone_sock().

syzbot reported the lockdep splat below. [0]

sctp_clone_sock() sets the child socket's ipv6_mc_list to NULL,
but somehow sock_release() in an error path finally acquires
lock_sock() in ipv6_sock_mc_close().

The root cause is that sctp_clone_sock() fetches inet6_sk(newsk)
before setting newinet->pinet6, meaning that the parent's
ipv6_mc_list was actually cleared.

Also, sctp_v6_copy_ip_options() uses inet6_sk() but is called
before newinet->pinet6 is set.

Let's use inet6_sk() only after setting newinet->pinet6.

[0]:
WARNING: possible recursive locking detected
syzkaller #0 Not tainted

syz.0.17/5996 is trying to acquire lock:
ffff888031af4c60 (sk_lock-AF_INET6){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1700 [inline]
ffff888031af4c60 (sk_lock-AF_INET6){+.+.}-{0:0}, at: ipv6_sock_mc_close+0xd3/0x140 net/ipv6/mcast.c:348

but task is already holding lock:
ffff888031af4320 (sk_lock-AF_INET6){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1700 [inline]
ffff888031af4320 (sk_lock-AF_INET6){+.+.}-{0:0}, at: sctp_getsockopt+0x135/0xb60 net/sctp/socket.c:8131

other info that might help us debug this:
 Possible unsafe locking scenario:

       CPU0
       ----
  lock(sk_lock-AF_INET6);
  lock(sk_lock-AF_INET6);

 *** DEADLOCK ***

 May be due to missing lock nesting notation

1 lock held by syz.0.17/5996:
 #0: ffff888031af4320 (sk_lock-AF_INET6){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1700 [inline]
 #0: ffff888031af4320 (sk_lock-AF_INET6){+.+.}-{0:0}, at: sctp_getsockopt+0x135/0xb60 net/sctp/socket.c:8131

stack backtrace:
CPU: 0 UID: 0 PID: 5996 Comm: syz.0.17 Not tainted syzkaller #0 PREEMPT(full)
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/25/2025
Call Trace:
 <TASK>
 dump_stack_lvl+0x189/0x250 lib/dump_stack.c:120
 print_deadlock_bug+0x279/0x290 kernel/locking/lockdep.c:3041
 check_deadlock kernel/locking/lockdep.c:3093 [inline]
 validate_chain kernel/locking/lockdep.c:3895 [inline]
 __lock_acquire+0x2540/0x2cf0 kernel/locking/lockdep.c:5237
 lock_acquire+0x117/0x340 kernel/locking/lockdep.c:5868
 lock_sock_nested+0x48/0x100 net/core/sock.c:3780
 lock_sock include/net/sock.h:1700 [inline]
 ipv6_sock_mc_close+0xd3/0x140 net/ipv6/mcast.c:348
 inet6_release+0x47/0x70 net/ipv6/af_inet6.c:482
 __sock_release net/socket.c:653 [inline]
 sock_release+0x85/0x150 net/socket.c:681
 sctp_getsockopt_peeloff_common+0x56b/0x770 net/sctp/socket.c:5732
 sctp_getsockopt_peeloff_flags+0x13b/0x230 net/sctp/socket.c:5801
 sctp_getsockopt+0x3ab/0xb60 net/sctp/socket.c:8151
 do_sock_getsockopt+0x2b4/0x3d0 net/socket.c:2399
 __sys_getsockopt net/socket.c:2428 [inline]
 __do_sys_getsockopt net/socket.c:2435 [inline]
 __se_sys_getsockopt net/socket.c:2432 [inline]
 __x64_sys_getsockopt+0x1a5/0x250 net/socket.c:2432
 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
 do_syscall_64+0xfa/0xf80 arch/x86/entry/syscall_64.c:94
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0033:0x7f8f8c38f749
Code: ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 a8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007ffcfdade018 EFLAGS: 00000246 ORIG_RAX: 0000000000000037
RAX: ffffffffffffffda RBX: 00007f8f8c5e5fa0 RCX: 00007f8f8c38f749
RDX: 000000000000007a RSI: 0000000000000084 RDI: 0000000000000003
RBP: 00007f8f8c413f91 R08: 0000200000000040 R09: 0000000000000000
R10: 0000200000000340 R11: 0000000000000246 R12: 0000000000000000
R13: 00007f8f8c5e5fa0 R14: 00007f8f8c5e5fa0 R15: 0000000000000005
 </TASK>

Fixes: 16942cf4d3e31 ("sctp: Use sk_clone() in sctp_accept().")
Reported-by: syzbot+c59e6bb54e7620495725@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/netdev/6936d112.a70a0220.38f243.00a7.GAE@google.com/
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20251210081206.1141086-2-kuniyu@google.com
Acked-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

diff --git a/net/sctp/socket.c b/net/sctp/socket.c
index d808096f5ab17a6b5aeaecfb33fda4a67e92e3d8..2493a5b1fa3ca7484b2ae4301b6dbf3b7a8233fb 100644 (file)
--- a/net/sctp/socket.c
+++ b/net/sctp/socket.c
@@ -4863,8 +4863,6 @@ static struct sock *sctp_clone_sock(struct sock *sk,
 
        newsp->pf->to_sk_daddr(&asoc->peer.primary_addr, newsk);
        newinet->inet_dport = htons(asoc->peer.port);
-
-       newsp->pf->copy_ip_options(sk, newsk);
        atomic_set(&newinet->inet_id, get_random_u16());
 
        inet_set_bit(MC_LOOP, newsk);
@@ -4874,17 +4872,20 @@ static struct sock *sctp_clone_sock(struct sock *sk,
 
 #if IS_ENABLED(CONFIG_IPV6)
        if (sk->sk_family == AF_INET6) {
-               struct ipv6_pinfo *newnp = inet6_sk(newsk);
+               struct ipv6_pinfo *newnp;
 
                newinet->pinet6 = &((struct sctp6_sock *)newsk)->inet6;
                newinet->ipv6_fl_list = NULL;
 
+               newnp = inet6_sk(newsk);
                memcpy(newnp, inet6_sk(sk), sizeof(struct ipv6_pinfo));
                newnp->ipv6_mc_list = NULL;
                newnp->ipv6_ac_list = NULL;
        }
 #endif
 
+       newsp->pf->copy_ip_options(sk, newsk);
+
        newsp->do_auto_asconf = 0;
        skb_queue_head_init(&newsp->pd_lobby);