tcp_bpf_sendmsg() keeps msg_tx across sk_stream_wait_memory(), which drops and reacquires the socket lock. Its error path used the current value of psock->cork to decide whether msg_tx named the stack-local temporary message. Two senders can therefore interleave as follows: Thread A Thread B msg_tx = psock->cork sk_msg_alloc() fails sk_stream_wait_memory() releases the socket lock acquires the socket lock completes the cork psock->cork = NULL frees the cork reacquires the socket lock msg_tx != psock->cork sk_msg_free(msg_tx) The stale cork is mistaken for the local temporary message and freed again. KASAN reported: BUG: KASAN: slab-use-after-free in sk_msg_free+0x49/0x50 Read of size 4 at addr ffff88810c908800 by task poc/90 Call Trace: sk_msg_free+0x49/0x50 tcp_bpf_sendmsg+0x14f5/0x1cc0 __sys_sendto+0x32c/0x3a0 __x64_sys_sendto+0xdb/0x1b0 Allocated by task 89: __kasan_kmalloc+0x8f/0xa0 tcp_bpf_sendmsg+0x16b3/0x1cc0 Freed by task 91: __kasan_slab_free+0x43/0x70 kfree+0x131/0x3c0 tcp_bpf_sendmsg+0xec3/0x1cc0 The same unclear ownership also leaves several error paths inconsistent. A failed cork allocation leaves cork_bytes armed, a subsequent temporary message can escape without releasing its pages, iterator errors are returned as success, and a failed redirect of a corked message discards data while reporting it as sent. Re-evaluating an existing cork can also copy the object onto itself. Make temporary ownership explicit by freeing only &tmp at the common exit. Reset cork_bytes when allocating the persistent cork fails, propagate iterator errors, preserve redirect failures and clear the copied count when corked data is discarded. Skip the copy when the message is already the persistent cork. Fixes: 604326b41a6f ("bpf, sockmap: convert to generic sk_msg interface") Cc: stable@vger.kernel.org Suggested-by: Emil Tsalapatis Signed-off-by: Chengfeng Ye --- Changes in v2: - Address Sashiko review findings for stale cork state, iterator-error propagation, cork self-copy, and masked redirect failures. - Keep the original stale-cork use-after-free fix. Sashiko: https://sashiko.dev/#/patchset/20260719161630.2901208-1-nicoyip.dev%40gmail.com Link: https://lore.kernel.org/netdev/20260719161630.2901208-1-nicoyip.dev@gmail.com/ [v1] net/ipv4/tcp_bpf.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) diff --git a/net/ipv4/tcp_bpf.c b/net/ipv4/tcp_bpf.c index 8e905b50dead..0594ee4013b0 100644 --- a/net/ipv4/tcp_bpf.c +++ b/net/ipv4/tcp_bpf.c @@ -443,12 +443,14 @@ static int tcp_bpf_send_verdict(struct sock *sk, struct sk_psock *psock, psock->cork = kzalloc_obj(*psock->cork, GFP_ATOMIC | __GFP_NOWARN); if (!psock->cork) { + psock->cork_bytes = 0; sk_msg_free(sk, msg); *copied = 0; return -ENOMEM; } } - memcpy(psock->cork, msg, sizeof(*msg)); + if (psock->cork != msg) + memcpy(psock->cork, msg, sizeof(*msg)); return 0; } @@ -495,14 +497,15 @@ static int tcp_bpf_send_verdict(struct sock *sk, struct sk_psock *psock, if (unlikely(ret < 0)) { int free = sk_msg_free(sk, msg); - if (!cork) + if (cork) + *copied = 0; + else *copied -= free; } if (cork) { sk_msg_free(sk, msg); kfree(msg); msg = NULL; - ret = 0; } break; case __SK_DROP: @@ -579,6 +582,7 @@ static int tcp_bpf_sendmsg(struct sock *sk, struct msghdr *msg, size_t size) copy); if (ret < 0) { sk_msg_trim(sk, msg_tx, osize); + err = ret; goto out_err; } @@ -603,13 +607,12 @@ static int tcp_bpf_sendmsg(struct sock *sk, struct msghdr *msg, size_t size) set_bit(SOCK_NOSPACE, &sk->sk_socket->flags); wait_for_memory: err = sk_stream_wait_memory(sk, &timeo); - if (err) { - if (msg_tx && msg_tx != psock->cork) - sk_msg_free(sk, msg_tx); + if (err) goto out_err; - } } out_err: + if (msg_tx == &tmp) + sk_msg_free(sk, msg_tx); if (err < 0) err = sk_stream_error(sk, msg->msg_flags, err); release_sock(sk); -- 2.43.0