[PATCH v2 bpf] sockmap: Fix sk_psock_drop() race vs sock_map_{unhash,close,destroy}().

[PATCH v2 bpf] sockmap: Fix sk_psock_drop() race vs sock_map_{unhash,close,destroy}().

[Kuniyuki Iwashima]

syzbot reported a splat in sock_map_destroy() [0], where psock was
NULL even though sk->sk_prot still pointed to tcp_bpf_prots[][].

The stack trace shows how badly the path was excercised, see
inet_release() calls tcp_close(), not sock_map_close() yet, but
finally reaching sock_map_destroy().

The root cause is a lack of synchronisation.

Even if sk_psock_get() fails to bump psock->refcnt, it does not
guarantee that sk_psock_drop() has finished, and thus sk->sk_prot
might not have been restored to the original one.

Commit 4b4647add7d3 ("sock_map: avoid race between sock_map_close
and sk_psock_put") attempted to address this, but it was insufficient
for two reasons.

It did not cover sock_map_unhash() and sock_map_destroy(), and
it missed the corner case where sk_psock() is NULL.

On non-x86 platforms, sk_psock_restore_proto(sk, psock) and
rcu_assign_sk_user_data(sk, NULL) can be reordered because there
is no address dependency between sk->sk_prot and sk->sk_user_data.

sk_psock_get() returning NULL implies nothing about sk->sk_prot.

Let's simply retry sk_psock_get() in the unlikely case.

Note that we cannot avoid loop even if we added memory barrier
in sk_psock_drop() and sock_map_psock_get_checked().

[0]:
WARNING: CPU: 1 PID: 8459 at net/core/sock_map.c:1667 sock_map_destroy+0x28b/0x2b0 net/core/sock_map.c:1667
Modules linked in:
CPU: 1 UID: 0 PID: 8459 Comm: syz.0.1109 Not tainted syzkaller #0 PREEMPT_{RT,(full)}
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 07/12/2025
RIP: 0010:sock_map_destroy+0x28b/0x2b0 net/core/sock_map.c:1667
Code: 8b 36 49 83 c6 38 4c 89 f0 48 c1 e8 03 42 80 3c 38 00 74 08 4c 89 f7 e8 93 62 22 f9 4d 8b 3e e9 79 ff ff ff e8 a6 2b c3 f8 90 <0f> 0b 90 eb 9c e8 9b 2b c3 f8 4c 89 e7 be 03 00 00 00 e8 0e 4e bc
RSP: 0018:ffffc9000d067be8 EFLAGS: 00010293
RAX: ffffffff88fb30aa RBX: ffff888024832000 RCX: ffff888024283b80
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 0000000000000001 R08: 0000000000000000 R09: 0000000000000000
R10: dffffc0000000000 R11: ffffed100862e946 R12: dffffc0000000000
R13: ffff888024832000 R14: ffffffff995b2208 R15: ffffffff88fb2e20
FS:  0000555579a7d500(0000) GS:ffff8881269c2000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00002000000048c0 CR3: 000000003713a000 CR4: 00000000003526f0
Call Trace:
 <TASK>
 inet_csk_destroy_sock+0x166/0x3a0 net/ipv4/inet_connection_sock.c:1294
 __tcp_close+0xcc1/0xfd0 net/ipv4/tcp.c:3262
 tcp_close+0x28/0x110 net/ipv4/tcp.c:3274
 inet_release+0x144/0x190 net/ipv4/af_inet.c:435
 __sock_release net/socket.c:649 [inline]
 sock_close+0xc0/0x240 net/socket.c:1439
 __fput+0x45b/0xa80 fs/file_table.c:468
 task_work_run+0x1d4/0x260 kernel/task_work.c:227
 resume_user_mode_work include/linux/resume_user_mode.h:50 [inline]
 exit_to_user_mode_loop+0xec/0x110 kernel/entry/common.c:43
 exit_to_user_mode_prepare include/linux/irq-entry-common.h:225 [inline]
 syscall_exit_to_user_mode_work include/linux/entry-common.h:175 [inline]
 syscall_exit_to_user_mode include/linux/entry-common.h:210 [inline]
 do_syscall_64+0x2bd/0x3b0 arch/x86/entry/syscall_64.c:100
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0033:0x7f265847ebe9
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:00007ffd158dfbd8 EFLAGS: 00000246 ORIG_RAX: 00000000000001b4
RAX: 0000000000000000 RBX: 000000000002ddb0 RCX: 00007f265847ebe9
RDX: 0000000000000000 RSI: 000000000000001e RDI: 0000000000000003
RBP: 00007f26586a7da0 R08: 0000000000000001 R09: 0000000e158dfecf
R10: 0000001b30a20000 R11: 0000000000000246 R12: 00007f26586a5fac
R13: 00007f26586a5fa0 R14: ffffffffffffffff R15: 00007ffd158dfcf0
 </TASK>

Fixes: 1aa12bdf1bfb ("bpf: sockmap, add sock close() hook to remove socks")
Fixes: b05545e15e1f ("bpf: sockmap, fix transition through disconnect without close")
Fixes: d8616ee2affc ("bpf, sockmap: Fix sk->sk_forward_alloc warn_on in sk_stream_kill_queues")
Reported-by: syzbot+b0842d38af58376d1fdc@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/bpf/69cec5ef.050a0220.2dbe29.0009.GAE@google.com/
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
---
v2: Fix small race window in no-loop-approach.
v1: https://lore.kernel.org/bpf/20260402203754.280844-1-kuniyu@google.com/
---
 net/core/sock_map.c | 39 +++++++++++++++++++++++++--------------
 1 file changed, 25 insertions(+), 14 deletions(-)

diff --git a/net/core/sock_map.c b/net/core/sock_map.c
index 02a68be3002a..99e3789492a0 100644
--- a/net/core/sock_map.c
+++ b/net/core/sock_map.c
@@ -1630,18 +1630,23 @@ void sock_map_unhash(struct sock *sk)
 	void (*saved_unhash)(struct sock *sk);
 	struct sk_psock *psock;
 
+retry:
 	rcu_read_lock();
 	psock = sk_psock(sk);
 	if (unlikely(!psock)) {
 		rcu_read_unlock();
 		saved_unhash = READ_ONCE(sk->sk_prot)->unhash;
+		if (unlikely(saved_unhash == sock_map_unhash))
+			goto retry;
 	} else {
 		saved_unhash = psock->saved_unhash;
 		sock_map_remove_links(sk, psock);
 		rcu_read_unlock();
+
+		if (WARN_ON_ONCE(saved_unhash == sock_map_unhash))
+			return;
 	}
-	if (WARN_ON_ONCE(saved_unhash == sock_map_unhash))
-		return;
+
 	if (saved_unhash)
 		saved_unhash(sk);
 }
@@ -1652,20 +1657,25 @@ void sock_map_destroy(struct sock *sk)
 	void (*saved_destroy)(struct sock *sk);
 	struct sk_psock *psock;
 
+retry:
 	rcu_read_lock();
 	psock = sk_psock_get(sk);
 	if (unlikely(!psock)) {
 		rcu_read_unlock();
 		saved_destroy = READ_ONCE(sk->sk_prot)->destroy;
+		if (unlikely(saved_destroy == sock_map_destroy))
+			goto retry;
 	} else {
 		saved_destroy = psock->saved_destroy;
 		sock_map_remove_links(sk, psock);
 		rcu_read_unlock();
 		sk_psock_stop(psock);
 		sk_psock_put(sk, psock);
+
+		if (WARN_ON_ONCE(saved_destroy == sock_map_destroy))
+			return;
 	}
-	if (WARN_ON_ONCE(saved_destroy == sock_map_destroy))
-		return;
+
 	if (saved_destroy)
 		saved_destroy(sk);
 }
@@ -1676,32 +1686,33 @@ void sock_map_close(struct sock *sk, long timeout)
 	void (*saved_close)(struct sock *sk, long timeout);
 	struct sk_psock *psock;
 
+retry:
 	lock_sock(sk);
 	rcu_read_lock();
-	psock = sk_psock(sk);
+	psock = sk_psock_get(sk);
 	if (likely(psock)) {
 		saved_close = psock->saved_close;
 		sock_map_remove_links(sk, psock);
-		psock = sk_psock_get(sk);
-		if (unlikely(!psock))
-			goto no_psock;
 		rcu_read_unlock();
 		sk_psock_stop(psock);
 		release_sock(sk);
 		cancel_delayed_work_sync(&psock->work);
 		sk_psock_put(sk, psock);
+
+		/* Make sure we do not recurse. This is a bug.
+		 * Leak the socket instead of crashing on a stack overflow.
+		 */
+		if (WARN_ON_ONCE(saved_close == sock_map_close))
+			return;
 	} else {
 		saved_close = READ_ONCE(sk->sk_prot)->close;
-no_psock:
 		rcu_read_unlock();
 		release_sock(sk);
+
+		if (unlikely(saved_close == sock_map_close))
+			goto retry;
 	}
 
-	/* Make sure we do not recurse. This is a bug.
-	 * Leak the socket instead of crashing on a stack overflow.
-	 */
-	if (WARN_ON_ONCE(saved_close == sock_map_close))
-		return;
 	saved_close(sk, timeout);
 }
 EXPORT_SYMBOL_GPL(sock_map_close);
-- 
2.54.0.rc1.555.g9c883467ad-goog

Re: [PATCH v2 bpf] sockmap: Fix sk_psock_drop() race vs sock_map_{unhash,close,destroy}().

[sashiko-bot]

Sashiko AI review found 1 potential issue(s):
- [High] The newly introduced unbounded `goto retry;` loops in `sock_map_unhash()`, `sock_map_destroy()`, and `sock_map_close()` can cause a hard lockup (deadlock) due to softirq preemption or priority inversion.
--

commit 08b24da207c576eef44ad9d88356dbd6600ab733
Author: Kuniyuki Iwashima <kuniyu@google.com>
sockmap: Fix sk_psock_drop() race vs sock_map_{unhash,close,destroy}().

This commit fixes a race condition where sk_psock_drop() might not have
finished restoring sk->sk_prot even if sk_psock_get() fails to bump the
refcount. It adds a retry loop to sock_map_unhash(), sock_map_destroy(),
and sock_map_close() when sk_psock_get() returns NULL but the protocol
hooks haven't been restored yet.

Closes: https://lore.kernel.org/bpf/69cec5ef.050a0220.2dbe29.0009.GAE@google.com/

> diff --git a/net/core/sock_map.c b/net/core/sock_map.c
> index 02a68be3002a2..99e3789492a09 100644
> --- a/net/core/sock_map.c
> +++ b/net/core/sock_map.c
[ ... ]
> @@ -1652,20 +1657,25 @@ void sock_map_destroy(struct sock *sk)
>  	void (*saved_destroy)(struct sock *sk);
>  	struct sk_psock *psock;
>  
> +retry:
>  	rcu_read_lock();
>  	psock = sk_psock_get(sk);
>  	if (unlikely(!psock)) {
>  		rcu_read_unlock();
>  		saved_destroy = READ_ONCE(sk->sk_prot)->destroy;
> +		if (unlikely(saved_destroy == sock_map_destroy))
> +			goto retry;

Can this unbounded retry loop cause a hard lockup due to softirq preemption
or priority inversion?

If sk_psock_put() is called from process context with bottom-halves enabled
(such as from sock_map_close() after release_sock()), an interrupt could
preempt the thread immediately after refcount_dec_and_test(&psock->refcnt)
drops to 0 but before sk_psock_drop() disables BH.

If a network interrupt fires on the same CPU, NET_RX_SOFTIRQ may process a
packet (such as an RST) that triggers socket destruction via:
tcp_done() -> inet_csk_destroy_sock() -> sk->sk_prot->destroy()

Since sk_psock_drop() has not yet restored the protocol, sk->sk_prot->destroy
is still sock_map_destroy().

When sock_map_destroy() calls sk_psock_get(), it returns NULL because the
refcount is exactly 0. The code then falls into the !psock branch, sees that
sk->sk_prot->destroy is still sock_map_destroy(), and jumps to retry.

Because the softirq spins infinitely in this tight loop and never yields the
CPU, the preempted process context can never execute sk_psock_drop(),
resulting in a permanent hard lockup.

A similar priority inversion deadlock could also occur on PREEMPT_RT if the
thread calling sk_psock_drop() is preempted by a higher-priority task.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260420194846.1089595-1-kuniyu@google.com?part=1

Re: [PATCH v2 bpf] sockmap: Fix sk_psock_drop() race vs sock_map_{unhash,close,destroy}().

[Jiayuan Chen]

On 4/21/26 9:13 AM, sashiko-bot@kernel.org wrote:
> Sashiko AI review found 1 potential issue(s):
> - [High] The newly introduced unbounded `goto retry;` loops in `sock_map_unhash()`, `sock_map_destroy()`, and `sock_map_close()` can cause a hard lockup (deadlock) due to softirq preemption or priority inversion.
> --
>
> commit 08b24da207c576eef44ad9d88356dbd6600ab733
> Author: Kuniyuki Iwashima <kuniyu@google.com>
> sockmap: Fix sk_psock_drop() race vs sock_map_{unhash,close,destroy}().
>
> This commit fixes a race condition where sk_psock_drop() might not have
> finished restoring sk->sk_prot even if sk_psock_get() fails to bump the
> refcount. It adds a retry loop to sock_map_unhash(), sock_map_destroy(),
> and sock_map_close() when sk_psock_get() returns NULL but the protocol
> hooks haven't been restored yet.
>
> Closes: https://lore.kernel.org/bpf/69cec5ef.050a0220.2dbe29.0009.GAE@google.com/
>
>> diff --git a/net/core/sock_map.c b/net/core/sock_map.c
>> index 02a68be3002a2..99e3789492a09 100644
>> --- a/net/core/sock_map.c
>> +++ b/net/core/sock_map.c
> [ ... ]
>> @@ -1652,20 +1657,25 @@ void sock_map_destroy(struct sock *sk)
>>   	void (*saved_destroy)(struct sock *sk);
>>   	struct sk_psock *psock;
>>   
>> +retry:
>>   	rcu_read_lock();
>>   	psock = sk_psock_get(sk);
>>   	if (unlikely(!psock)) {
>>   		rcu_read_unlock();
>>   		saved_destroy = READ_ONCE(sk->sk_prot)->destroy;
>> +		if (unlikely(saved_destroy == sock_map_destroy))
>> +			goto retry;
> Can this unbounded retry loop cause a hard lockup due to softirq preemption
> or priority inversion?
>
> If sk_psock_put() is called from process context with bottom-halves enabled
> (such as from sock_map_close() after release_sock()), an interrupt could
> preempt the thread immediately after refcount_dec_and_test(&psock->refcnt)
> drops to 0 but before sk_psock_drop() disables BH.
>
> If a network interrupt fires on the same CPU, NET_RX_SOFTIRQ may process a
> packet (such as an RST) that triggers socket destruction via:
> tcp_done() -> inet_csk_destroy_sock() -> sk->sk_prot->destroy()
>
> Since sk_psock_drop() has not yet restored the protocol, sk->sk_prot->destroy
> is still sock_map_destroy().
>
> When sock_map_destroy() calls sk_psock_get(), it returns NULL because the
> refcount is exactly 0. The code then falls into the !psock branch, sees that
> sk->sk_prot->destroy is still sock_map_destroy(), and jumps to retry.
>
> Because the softirq spins infinitely in this tight loop and never yields the
> CPU, the preempted process context can never execute sk_psock_drop(),
> resulting in a permanent hard lockup.


sock_map_close(sk)
|___ sk_psock_put(sk, psock)   <- refcnt-hits-0 window lives here
|___ saved_close == tcp_close
       |__tcp_close
             |____ sock_orphan   <- SOCK_DEAD set here
             |____(later) inet_csk_destroy_sock

At the exact instant the refcnt can be observed at 0 with
sk_prot not yet restored, SOCK_DEAD is guaranteed not to be set.

So I think AI is not correct.

> A similar priority inversion deadlock could also occur on PREEMPT_RT if the
> thread calling sk_psock_drop() is preempted by a higher-priority task.
>

Re: [PATCH v2 bpf] sockmap: Fix sk_psock_drop() race vs sock_map_{unhash,close,destroy}().

[Jiayuan Chen]

On 4/21/26 3:48 AM, Kuniyuki Iwashima wrote:
> syzbot reported a splat in sock_map_destroy() [0], where psock was
> NULL even though sk->sk_prot still pointed to tcp_bpf_prots[][].
>
> The stack trace shows how badly the path was excercised, see
> inet_release() calls tcp_close(), not sock_map_close() yet, but
> finally reaching sock_map_destroy().
>
> The root cause is a lack of synchronisation.
>
> Even if sk_psock_get() fails to bump psock->refcnt, it does not
> guarantee that sk_psock_drop() has finished, and thus sk->sk_prot
> might not have been restored to the original one.
>
> Commit 4b4647add7d3 ("sock_map: avoid race between sock_map_close
> and sk_psock_put") attempted to address this, but it was insufficient
> for two reasons.
>
> It did not cover sock_map_unhash() and sock_map_destroy(), and
> it missed the corner case where sk_psock() is NULL.
>
> On non-x86 platforms, sk_psock_restore_proto(sk, psock) and
> rcu_assign_sk_user_data(sk, NULL) can be reordered because there
> is no address dependency between sk->sk_prot and sk->sk_user_data.
>
> sk_psock_get() returning NULL implies nothing about sk->sk_prot.
>
> Let's simply retry sk_psock_get() in the unlikely case.
>
> Note that we cannot avoid loop even if we added memory barrier
> in sk_psock_drop() and sock_map_psock_get_checked().
>
> [0]:
> WARNING: CPU: 1 PID: 8459 at net/core/sock_map.c:1667 sock_map_destroy+0x28b/0x2b0 net/core/sock_map.c:1667
> Modules linked in:
> CPU: 1 UID: 0 PID: 8459 Comm: syz.0.1109 Not tainted syzkaller #0 PREEMPT_{RT,(full)}
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 07/12/2025
> RIP: 0010:sock_map_destroy+0x28b/0x2b0 net/core/sock_map.c:1667
> Code: 8b 36 49 83 c6 38 4c 89 f0 48 c1 e8 03 42 80 3c 38 00 74 08 4c 89 f7 e8 93 62 22 f9 4d 8b 3e e9 79 ff ff ff e8 a6 2b c3 f8 90 <0f> 0b 90 eb 9c e8 9b 2b c3 f8 4c 89 e7 be 03 00 00 00 e8 0e 4e bc
> RSP: 0018:ffffc9000d067be8 EFLAGS: 00010293
> RAX: ffffffff88fb30aa RBX: ffff888024832000 RCX: ffff888024283b80
> RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
> RBP: 0000000000000001 R08: 0000000000000000 R09: 0000000000000000
> R10: dffffc0000000000 R11: ffffed100862e946 R12: dffffc0000000000
> R13: ffff888024832000 R14: ffffffff995b2208 R15: ffffffff88fb2e20
> FS:  0000555579a7d500(0000) GS:ffff8881269c2000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 00002000000048c0 CR3: 000000003713a000 CR4: 00000000003526f0
> Call Trace:
>   <TASK>
>   inet_csk_destroy_sock+0x166/0x3a0 net/ipv4/inet_connection_sock.c:1294
>   __tcp_close+0xcc1/0xfd0 net/ipv4/tcp.c:3262
>   tcp_close+0x28/0x110 net/ipv4/tcp.c:3274
>   inet_release+0x144/0x190 net/ipv4/af_inet.c:435
>   __sock_release net/socket.c:649 [inline]
>   sock_close+0xc0/0x240 net/socket.c:1439
>   __fput+0x45b/0xa80 fs/file_table.c:468
>   task_work_run+0x1d4/0x260 kernel/task_work.c:227
>   resume_user_mode_work include/linux/resume_user_mode.h:50 [inline]
>   exit_to_user_mode_loop+0xec/0x110 kernel/entry/common.c:43
>   exit_to_user_mode_prepare include/linux/irq-entry-common.h:225 [inline]
>   syscall_exit_to_user_mode_work include/linux/entry-common.h:175 [inline]
>   syscall_exit_to_user_mode include/linux/entry-common.h:210 [inline]
>   do_syscall_64+0x2bd/0x3b0 arch/x86/entry/syscall_64.c:100
>   entry_SYSCALL_64_after_hwframe+0x77/0x7f
> RIP: 0033:0x7f265847ebe9
> 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:00007ffd158dfbd8 EFLAGS: 00000246 ORIG_RAX: 00000000000001b4
> RAX: 0000000000000000 RBX: 000000000002ddb0 RCX: 00007f265847ebe9
> RDX: 0000000000000000 RSI: 000000000000001e RDI: 0000000000000003
> RBP: 00007f26586a7da0 R08: 0000000000000001 R09: 0000000e158dfecf
> R10: 0000001b30a20000 R11: 0000000000000246 R12: 00007f26586a5fac
> R13: 00007f26586a5fa0 R14: ffffffffffffffff R15: 00007ffd158dfcf0
>   </TASK>
>
> Fixes: 1aa12bdf1bfb ("bpf: sockmap, add sock close() hook to remove socks")
> Fixes: b05545e15e1f ("bpf: sockmap, fix transition through disconnect without close")
> Fixes: d8616ee2affc ("bpf, sockmap: Fix sk->sk_forward_alloc warn_on in sk_stream_kill_queues")
> Reported-by: syzbot+b0842d38af58376d1fdc@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/bpf/69cec5ef.050a0220.2dbe29.0009.GAE@google.com/
> Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>


Reviewed-by: Jiayuan Chen <jiayuan.chen@linux.dev>

Thanks,

LGTM.  I reproduced and tested it and the patch fixed the problem.