bpf, sockmap: Fix an infinite loop error when len is 0 in tcp_bpf_recvmsg_parser()
Subsystems (2)
PATCH 1/1 bpf, sockmap: Fix an infinite loop error when len is 0 in tcp_bpf_recvmsg_parser()
From: Liu Jian <[email protected]>
commit d900f3d20cc3169ce42ec72acc850e662a4d4db2 upstream.
When the buffer length of the recvmsg system call is 0, we got the flollowing soft lockup problem:
watchdog: BUG: soft lockup - CPU#3 stuck for 27s! [a.out:6149] CPU: 3 PID: 6149 Comm: a.out Kdump: loaded Not tainted 6.2.0+ #30 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.15.0-1 04/01/2014 RIP: 0010:remove_wait_queue+0xb/0xc0 Code: 5e 41 5f c3 cc cc cc cc 0f 1f 80 00 00 00 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3 0f 1e fa 0f 1f 44 00 00 41 57 <41> 56 41 55 41 54 55 48 89 fd 53 48 89 f3 4c 8d 6b 18 4c 8d 73 20 RSP: 0018:ffff88811b5978b8 EFLAGS: 00000246 RAX: 0000000000000000 RBX: ffff88811a7d3780 RCX: ffffffffb7a4d768 RDX: dffffc0000000000 RSI: ffff88811b597908 RDI: ffff888115408040 RBP: 1ffff110236b2f1b R08: 0000000000000000 R09: ffff88811a7d37e7 R10: ffffed10234fa6fc R11: 0000000000000001 R12: ffff88811179b800 R13: 0000000000000001 R14: ffff88811a7d38a8 R15: ffff88811a7d37e0 FS: 00007f6fb5398740(0000) GS:ffff888237180000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000020000000 CR3: 000000010b6ba002 CR4: 0000000000370ee0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: <TASK> tcp_msg_wait_data+0x279/0x2f0 tcp_bpf_recvmsg_parser+0x3c6/0x490 inet_recvmsg+0x280/0x290 sock_recvmsg+0xfc/0x120 ____sys_recvmsg+0x160/0x3d0 ___sys_recvmsg+0xf0/0x180 __sys_recvmsg+0xea/0x1a0 do_syscall_64+0x3f/0x90 entry_SYSCALL_64_after_hwframe+0x72/0xdc
The logic in tcp_bpf_recvmsg_parser is as follows:
msg_bytes_ready:
copied = sk_msg_recvmsg(sk, psock, msg, len, flags);
if (!copied) {
wait data;
goto msg_bytes_ready;
}In this case, "copied" always is 0, the infinite loop occurs.
According to the Linux system call man page, 0 should be returned in this case. Therefore, in tcp_bpf_recvmsg_parser(), if the length is 0, directly return. Also modify several other functions with the same problem.
[ kovalev: bp to fix CVE-2023-53133; applied only to tcp_bpf_recvmsg as the
older kernel lacks tcp_bpf_recvmsg_parser, udp_bpf_recvmsg and unix_bpf_recvmsg (see upstream commits c5d2177a72a1, 1f5be6b3b063 and 9825d866ce0d) ]
--- net/ipv4/tcp_bpf.c | 3 +++ 1 file changed, 3 insertions(+)
net/ipv4/tcp_bpf.c +3 -0
@@ -277,6 +277,9 @@ static int tcp_bpf_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,if (unlikely (flags &MSG_ERRQUEUE ))return inet_recv_error (sk, msg, len, addr_len); +if (!len) +return 0 ; + psock =sk_psock_get (sk);if (unlikely (!psock))return tcp_recvmsg (sk, msg, len, nonblock, flags, addr_len);
Signature · 1 line
2.50.1
[email protected]
AI reviewSashiko AI review found 1 potential issue(s): - [High] Unconditional early return for `len == 0` in `tcp_bpf_recvmsg()` breaks UAPI semantics by masking socket errors, returning 0 instead of…
Sashiko AI review found 1 potential issue(s):
- [High] Unconditional early return for len == 0 in tcp_bpf_recvmsg() breaks UAPI semantics by masking socket errors, returning 0 instead of -EAGAIN for non-blocking sockets, and bypassing state checks. --
commit 17cc1ee5a157715faa482ac2a4f167f9022cb333
Author: Liu Jian <[email protected]>
bpf, sockmap: Fix an infinite loop error when len is 0 in tcp_bpf_recvmsg_parser()
This commit addresses a soft lockup issue where a 0-length recvmsg causes an infinite loop in tcp_bpf_recvmsg_parser(). It fixes the issue by returning 0 early when the length is 0.
net/ipv4/tcp_bpf.c static int tcp_bpf_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, · line 281
Quoted 8 lines
net/ipv4/tcp_bpf.c +2
-0@@ -277,6 +277,9 @@ static int tcp_bpf_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,if (unlikely (flags &MSG_ERRQUEUE ))return inet_recv_error (sk, msg, len, addr_len); +if (!len) +return 0 ;
Does this unconditional early return bypass necessary socket state checks?
In the standard Linux TCP implementation, a 0-length receive is not an immediate no-op. For instance, if the socket is non-blocking and empty, tcp_recvmsg() normally returns -EAGAIN. If the socket has a pending error, it returns that error, and if the socket is in TCP_LISTEN, it returns -ENOTCONN.
By unconditionally returning 0 here, a non-blocking recv(0) on an empty socket will return 0 instead of -EAGAIN. Since a 0 return value typically indicates EOF (connection closed) to userspace applications, could this cause applications to incorrectly assume the connection was closed and drop it?
Would it be safer to delegate 0-length reads to the standard TCP implementation by returning tcp_recvmsg() here instead, ensuring that the existing UAPI error semantics are preserved?
Quoted 4 lines
net/ipv4/tcp_bpf.c +1
-0+ psock =sk_psock_get (sk);if (unlikely (!psock))return tcp_recvmsg (sk, msg, len, nonblock, flags, addr_len);
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1