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 
unix_msg_wait_data(struct sock * sk,struct sk_psock * psock,long timeo)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 
__unix_recvmsg(struct sock * sk,struct msghdr * msg,size_t len,int flags)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 
unix_bpf_recvmsg(struct sock * sk,struct msghdr * msg,size_t len,int flags,int * addr_len)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 	psock = sk_psock_get(sk);
58 	if (unlikely(!psock))
59 		return __unix_recvmsg(sk, msg, len, flags);
60 
61 	mutex_lock(&u->iolock);
62 	if (!skb_queue_empty(&sk->sk_receive_queue) &&
63 	    sk_psock_queue_empty(psock)) {
64 		mutex_unlock(&u->iolock);
65 		sk_psock_put(sk, psock);
66 		return __unix_recvmsg(sk, msg, len, flags);
67 	}
68 
69 msg_bytes_ready:
70 	copied = sk_msg_recvmsg(sk, psock, msg, len, flags);
71 	if (!copied) {
72 		long timeo;
73 		int data;
74 
75 		timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
76 		data = unix_msg_wait_data(sk, psock, timeo);
77 		if (data) {
78 			if (!sk_psock_queue_empty(psock))
79 				goto msg_bytes_ready;
80 			mutex_unlock(&u->iolock);
81 			sk_psock_put(sk, psock);
82 			return __unix_recvmsg(sk, msg, len, flags);
83 		}
84 		copied = -EAGAIN;
85 	}
86 	mutex_unlock(&u->iolock);
87 	sk_psock_put(sk, psock);
88 	return copied;
89 }
90 
91 static struct proto *unix_dgram_prot_saved __read_mostly;
92 static DEFINE_SPINLOCK(unix_dgram_prot_lock);
93 static struct proto unix_dgram_bpf_prot;
94 
95 static struct proto *unix_stream_prot_saved __read_mostly;
96 static DEFINE_SPINLOCK(unix_stream_prot_lock);
97 static struct proto unix_stream_bpf_prot;
98 
unix_dgram_bpf_rebuild_protos(struct proto * prot,const struct proto * base)99 static void unix_dgram_bpf_rebuild_protos(struct proto *prot, const struct proto *base)
100 {
101 	*prot        = *base;
102 	prot->close  = sock_map_close;
103 	prot->recvmsg = unix_bpf_recvmsg;
104 	prot->sock_is_readable = sk_msg_is_readable;
105 }
106 
unix_stream_bpf_rebuild_protos(struct proto * prot,const struct proto * base)107 static void unix_stream_bpf_rebuild_protos(struct proto *prot,
108 					   const struct proto *base)
109 {
110 	*prot        = *base;
111 	prot->close  = sock_map_close;
112 	prot->recvmsg = unix_bpf_recvmsg;
113 	prot->sock_is_readable = sk_msg_is_readable;
114 	prot->unhash  = sock_map_unhash;
115 }
116 
unix_dgram_bpf_check_needs_rebuild(struct proto * ops)117 static void unix_dgram_bpf_check_needs_rebuild(struct proto *ops)
118 {
119 	if (unlikely(ops != smp_load_acquire(&unix_dgram_prot_saved))) {
120 		spin_lock_bh(&unix_dgram_prot_lock);
121 		if (likely(ops != unix_dgram_prot_saved)) {
122 			unix_dgram_bpf_rebuild_protos(&unix_dgram_bpf_prot, ops);
123 			smp_store_release(&unix_dgram_prot_saved, ops);
124 		}
125 		spin_unlock_bh(&unix_dgram_prot_lock);
126 	}
127 }
128 
unix_stream_bpf_check_needs_rebuild(struct proto * ops)129 static void unix_stream_bpf_check_needs_rebuild(struct proto *ops)
130 {
131 	if (unlikely(ops != smp_load_acquire(&unix_stream_prot_saved))) {
132 		spin_lock_bh(&unix_stream_prot_lock);
133 		if (likely(ops != unix_stream_prot_saved)) {
134 			unix_stream_bpf_rebuild_protos(&unix_stream_bpf_prot, ops);
135 			smp_store_release(&unix_stream_prot_saved, ops);
136 		}
137 		spin_unlock_bh(&unix_stream_prot_lock);
138 	}
139 }
140 
unix_dgram_bpf_update_proto(struct sock * sk,struct sk_psock * psock,bool restore)141 int unix_dgram_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore)
142 {
143 	if (sk->sk_type != SOCK_DGRAM)
144 		return -EOPNOTSUPP;
145 
146 	if (restore) {
147 		sk->sk_write_space = psock->saved_write_space;
148 		WRITE_ONCE(sk->sk_prot, psock->sk_proto);
149 		return 0;
150 	}
151 
152 	unix_dgram_bpf_check_needs_rebuild(psock->sk_proto);
153 	WRITE_ONCE(sk->sk_prot, &unix_dgram_bpf_prot);
154 	return 0;
155 }
156 
unix_stream_bpf_update_proto(struct sock * sk,struct sk_psock * psock,bool restore)157 int unix_stream_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore)
158 {
159 	if (restore) {
160 		sk->sk_write_space = psock->saved_write_space;
161 		WRITE_ONCE(sk->sk_prot, psock->sk_proto);
162 		return 0;
163 	}
164 
165 	unix_stream_bpf_check_needs_rebuild(psock->sk_proto);
166 	WRITE_ONCE(sk->sk_prot, &unix_stream_bpf_prot);
167 	return 0;
168 }
169 
unix_bpf_build_proto(void)170 void __init unix_bpf_build_proto(void)
171 {
172 	unix_dgram_bpf_rebuild_protos(&unix_dgram_bpf_prot, &unix_dgram_proto);
173 	unix_stream_bpf_rebuild_protos(&unix_stream_bpf_prot, &unix_stream_proto);
174 
175 }
176