xref: /openbmc/linux/net/unix/unix_bpf.c (revision e9b7b8b3)
1 // SPDX-License-Identifier: GPL-2.0
2 /* Copyright (c) 2021 Cong Wang <cong.wang@bytedance.com> */
3 
4 #include <linux/skmsg.h>
5 #include <linux/bpf.h>
6 #include <net/sock.h>
7 #include <net/af_unix.h>
8 
9 #define unix_sk_has_data(__sk, __psock)					\
10 		({	!skb_queue_empty(&__sk->sk_receive_queue) ||	\
11 			!skb_queue_empty(&__psock->ingress_skb) ||	\
12 			!list_empty(&__psock->ingress_msg);		\
13 		})
14 
15 static int unix_msg_wait_data(struct sock *sk, struct sk_psock *psock,
16 			      long timeo)
17 {
18 	DEFINE_WAIT_FUNC(wait, woken_wake_function);
19 	struct unix_sock *u = unix_sk(sk);
20 	int ret = 0;
21 
22 	if (sk->sk_shutdown & RCV_SHUTDOWN)
23 		return 1;
24 
25 	if (!timeo)
26 		return ret;
27 
28 	add_wait_queue(sk_sleep(sk), &wait);
29 	sk_set_bit(SOCKWQ_ASYNC_WAITDATA, sk);
30 	if (!unix_sk_has_data(sk, psock)) {
31 		mutex_unlock(&u->iolock);
32 		wait_woken(&wait, TASK_INTERRUPTIBLE, timeo);
33 		mutex_lock(&u->iolock);
34 		ret = unix_sk_has_data(sk, psock);
35 	}
36 	sk_clear_bit(SOCKWQ_ASYNC_WAITDATA, sk);
37 	remove_wait_queue(sk_sleep(sk), &wait);
38 	return ret;
39 }
40 
41 static int __unix_recvmsg(struct sock *sk, struct msghdr *msg,
42 			  size_t len, int flags)
43 {
44 	if (sk->sk_type == SOCK_DGRAM)
45 		return __unix_dgram_recvmsg(sk, msg, len, flags);
46 	else
47 		return __unix_stream_recvmsg(sk, msg, len, flags);
48 }
49 
50 static int unix_bpf_recvmsg(struct sock *sk, struct msghdr *msg,
51 			    size_t len, int flags, int *addr_len)
52 {
53 	struct unix_sock *u = unix_sk(sk);
54 	struct sk_psock *psock;
55 	int copied;
56 
57 	if (!len)
58 		return 0;
59 
60 	psock = sk_psock_get(sk);
61 	if (unlikely(!psock))
62 		return __unix_recvmsg(sk, msg, len, flags);
63 
64 	mutex_lock(&u->iolock);
65 	if (!skb_queue_empty(&sk->sk_receive_queue) &&
66 	    sk_psock_queue_empty(psock)) {
67 		mutex_unlock(&u->iolock);
68 		sk_psock_put(sk, psock);
69 		return __unix_recvmsg(sk, msg, len, flags);
70 	}
71 
72 msg_bytes_ready:
73 	copied = sk_msg_recvmsg(sk, psock, msg, len, flags);
74 	if (!copied) {
75 		long timeo;
76 		int data;
77 
78 		timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
79 		data = unix_msg_wait_data(sk, psock, timeo);
80 		if (data) {
81 			if (!sk_psock_queue_empty(psock))
82 				goto msg_bytes_ready;
83 			mutex_unlock(&u->iolock);
84 			sk_psock_put(sk, psock);
85 			return __unix_recvmsg(sk, msg, len, flags);
86 		}
87 		copied = -EAGAIN;
88 	}
89 	mutex_unlock(&u->iolock);
90 	sk_psock_put(sk, psock);
91 	return copied;
92 }
93 
94 static struct proto *unix_dgram_prot_saved __read_mostly;
95 static DEFINE_SPINLOCK(unix_dgram_prot_lock);
96 static struct proto unix_dgram_bpf_prot;
97 
98 static struct proto *unix_stream_prot_saved __read_mostly;
99 static DEFINE_SPINLOCK(unix_stream_prot_lock);
100 static struct proto unix_stream_bpf_prot;
101 
102 static void unix_dgram_bpf_rebuild_protos(struct proto *prot, const struct proto *base)
103 {
104 	*prot        = *base;
105 	prot->close  = sock_map_close;
106 	prot->recvmsg = unix_bpf_recvmsg;
107 	prot->sock_is_readable = sk_msg_is_readable;
108 }
109 
110 static void unix_stream_bpf_rebuild_protos(struct proto *prot,
111 					   const struct proto *base)
112 {
113 	*prot        = *base;
114 	prot->close  = sock_map_close;
115 	prot->recvmsg = unix_bpf_recvmsg;
116 	prot->sock_is_readable = sk_msg_is_readable;
117 	prot->unhash  = sock_map_unhash;
118 }
119 
120 static void unix_dgram_bpf_check_needs_rebuild(struct proto *ops)
121 {
122 	if (unlikely(ops != smp_load_acquire(&unix_dgram_prot_saved))) {
123 		spin_lock_bh(&unix_dgram_prot_lock);
124 		if (likely(ops != unix_dgram_prot_saved)) {
125 			unix_dgram_bpf_rebuild_protos(&unix_dgram_bpf_prot, ops);
126 			smp_store_release(&unix_dgram_prot_saved, ops);
127 		}
128 		spin_unlock_bh(&unix_dgram_prot_lock);
129 	}
130 }
131 
132 static void unix_stream_bpf_check_needs_rebuild(struct proto *ops)
133 {
134 	if (unlikely(ops != smp_load_acquire(&unix_stream_prot_saved))) {
135 		spin_lock_bh(&unix_stream_prot_lock);
136 		if (likely(ops != unix_stream_prot_saved)) {
137 			unix_stream_bpf_rebuild_protos(&unix_stream_bpf_prot, ops);
138 			smp_store_release(&unix_stream_prot_saved, ops);
139 		}
140 		spin_unlock_bh(&unix_stream_prot_lock);
141 	}
142 }
143 
144 int unix_dgram_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore)
145 {
146 	if (sk->sk_type != SOCK_DGRAM)
147 		return -EOPNOTSUPP;
148 
149 	if (restore) {
150 		sk->sk_write_space = psock->saved_write_space;
151 		sock_replace_proto(sk, psock->sk_proto);
152 		return 0;
153 	}
154 
155 	unix_dgram_bpf_check_needs_rebuild(psock->sk_proto);
156 	sock_replace_proto(sk, &unix_dgram_bpf_prot);
157 	return 0;
158 }
159 
160 int unix_stream_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore)
161 {
162 	struct sock *sk_pair;
163 
164 	/* Restore does not decrement the sk_pair reference yet because we must
165 	 * keep the a reference to the socket until after an RCU grace period
166 	 * and any pending sends have completed.
167 	 */
168 	if (restore) {
169 		sk->sk_write_space = psock->saved_write_space;
170 		sock_replace_proto(sk, psock->sk_proto);
171 		return 0;
172 	}
173 
174 	/* psock_update_sk_prot can be called multiple times if psock is
175 	 * added to multiple maps and/or slots in the same map. There is
176 	 * also an edge case where replacing a psock with itself can trigger
177 	 * an extra psock_update_sk_prot during the insert process. So it
178 	 * must be safe to do multiple calls. Here we need to ensure we don't
179 	 * increment the refcnt through sock_hold many times. There will only
180 	 * be a single matching destroy operation.
181 	 */
182 	if (!psock->sk_pair) {
183 		sk_pair = unix_peer(sk);
184 		sock_hold(sk_pair);
185 		psock->sk_pair = sk_pair;
186 	}
187 
188 	unix_stream_bpf_check_needs_rebuild(psock->sk_proto);
189 	sock_replace_proto(sk, &unix_stream_bpf_prot);
190 	return 0;
191 }
192 
193 void __init unix_bpf_build_proto(void)
194 {
195 	unix_dgram_bpf_rebuild_protos(&unix_dgram_bpf_prot, &unix_dgram_proto);
196 	unix_stream_bpf_rebuild_protos(&unix_stream_bpf_prot, &unix_stream_proto);
197 
198 }
199