From: dihu anny.hu@linux.alibaba.com
[ Upstream commit 487082fb7bd2a32b66927d2b22e3a81b072b44f0 ]
When user application calls read() with MSG_PEEK flag to read data of bpf sockmap socket, kernel panic happens at __tcp_bpf_recvmsg+0x12c/0x350. sk_msg is not removed from ingress_msg queue after read out under MSG_PEEK flag is set. Because it's not judged whether sk_msg is the last msg of ingress_msg queue, the next sk_msg may be the head of ingress_msg queue, whose memory address of sg page is invalid. So it's necessary to add check codes to prevent this problem.
[20759.125457] BUG: kernel NULL pointer dereference, address: 0000000000000008 [20759.132118] CPU: 53 PID: 51378 Comm: envoy Tainted: G E 5.4.32 #1 [20759.140890] Hardware name: Inspur SA5212M4/YZMB-00370-109, BIOS 4.1.12 06/18/2017 [20759.149734] RIP: 0010:copy_page_to_iter+0xad/0x300 [20759.270877] __tcp_bpf_recvmsg+0x12c/0x350 [20759.276099] tcp_bpf_recvmsg+0x113/0x370 [20759.281137] inet_recvmsg+0x55/0xc0 [20759.285734] __sys_recvfrom+0xc8/0x130 [20759.290566] ? __audit_syscall_entry+0x103/0x130 [20759.296227] ? syscall_trace_enter+0x1d2/0x2d0 [20759.301700] ? __audit_syscall_exit+0x1e4/0x290 [20759.307235] __x64_sys_recvfrom+0x24/0x30 [20759.312226] do_syscall_64+0x55/0x1b0 [20759.316852] entry_SYSCALL_64_after_hwframe+0x44/0xa9
Signed-off-by: dihu anny.hu@linux.alibaba.com Signed-off-by: Alexei Starovoitov ast@kernel.org Acked-by: John Fastabend john.fastabend@gmail.com Acked-by: Jakub Sitnicki jakub@cloudflare.com Link: https://lore.kernel.org/bpf/20200605084625.9783-1-anny.hu@linux.alibaba.com Signed-off-by: Sasha Levin sashal@kernel.org --- net/ipv4/tcp_bpf.c | 3 +++ 1 file changed, 3 insertions(+)
diff --git a/net/ipv4/tcp_bpf.c b/net/ipv4/tcp_bpf.c index 69b0254083904..ad9f382027311 100644 --- a/net/ipv4/tcp_bpf.c +++ b/net/ipv4/tcp_bpf.c @@ -96,6 +96,9 @@ int __tcp_bpf_recvmsg(struct sock *sk, struct sk_psock *psock, } while (i != msg_rx->sg.end);
if (unlikely(peek)) { + if (msg_rx == list_last_entry(&psock->ingress_msg, + struct sk_msg, list)) + break; msg_rx = list_next_entry(msg_rx, list); continue; }