Commit | Line | Data |
---|---|---|
2e6599cb ACM |
1 | /* |
2 | * NET Generic infrastructure for Network protocols. | |
3 | * | |
4 | * Definitions for request_sock | |
5 | * | |
6 | * Authors: Arnaldo Carvalho de Melo <acme@conectiva.com.br> | |
7 | * | |
8 | * From code originally in include/net/tcp.h | |
9 | * | |
10 | * This program is free software; you can redistribute it and/or | |
11 | * modify it under the terms of the GNU General Public License | |
12 | * as published by the Free Software Foundation; either version | |
13 | * 2 of the License, or (at your option) any later version. | |
14 | */ | |
15 | #ifndef _REQUEST_SOCK_H | |
16 | #define _REQUEST_SOCK_H | |
17 | ||
18 | #include <linux/slab.h> | |
0e87506f | 19 | #include <linux/spinlock.h> |
2e6599cb | 20 | #include <linux/types.h> |
547b792c | 21 | #include <linux/bug.h> |
0e87506f | 22 | |
2e6599cb ACM |
23 | #include <net/sock.h> |
24 | ||
60236fdd | 25 | struct request_sock; |
2e6599cb ACM |
26 | struct sk_buff; |
27 | struct dst_entry; | |
28 | struct proto; | |
29 | ||
e6b4d113 WAS |
30 | /* empty to "strongly type" an otherwise void parameter. |
31 | */ | |
32 | struct request_values { | |
33 | }; | |
34 | ||
60236fdd | 35 | struct request_sock_ops { |
2e6599cb | 36 | int family; |
2e6599cb | 37 | int obj_size; |
e18b890b | 38 | struct kmem_cache *slab; |
7e56b5d6 | 39 | char *slab_name; |
2e6599cb | 40 | int (*rtx_syn_ack)(struct sock *sk, |
e6b4d113 WAS |
41 | struct request_sock *req, |
42 | struct request_values *rvp); | |
6edafaaf | 43 | void (*send_ack)(struct sock *sk, struct sk_buff *skb, |
60236fdd | 44 | struct request_sock *req); |
cfb6eeb4 YH |
45 | void (*send_reset)(struct sock *sk, |
46 | struct sk_buff *skb); | |
60236fdd | 47 | void (*destructor)(struct request_sock *req); |
72659ecc OP |
48 | void (*syn_ack_timeout)(struct sock *sk, |
49 | struct request_sock *req); | |
2e6599cb ACM |
50 | }; |
51 | ||
e6c022a4 ED |
52 | extern int inet_rtx_syn_ack(struct sock *parent, struct request_sock *req); |
53 | ||
60236fdd | 54 | /* struct request_sock - mini sock to represent a connection request |
2e6599cb | 55 | */ |
60236fdd ACM |
56 | struct request_sock { |
57 | struct request_sock *dl_next; /* Must be first member! */ | |
2e6599cb | 58 | u16 mss; |
e6c022a4 ED |
59 | u8 num_retrans; /* number of retransmits */ |
60 | u8 cookie_ts:1; /* syncookie: encode tcpopts in timestamp */ | |
61 | u8 num_timeout:7; /* number of timeouts */ | |
2e6599cb ACM |
62 | /* The following two fields can be easily recomputed I think -AK */ |
63 | u32 window_clamp; /* window clamp at creation time */ | |
64 | u32 rcv_wnd; /* rcv_wnd offered first time */ | |
65 | u32 ts_recent; | |
66 | unsigned long expires; | |
72a3effa | 67 | const struct request_sock_ops *rsk_ops; |
2e6599cb | 68 | struct sock *sk; |
4237c75c | 69 | u32 secid; |
6b877699 | 70 | u32 peer_secid; |
2e6599cb ACM |
71 | }; |
72 | ||
72a3effa | 73 | static inline struct request_sock *reqsk_alloc(const struct request_sock_ops *ops) |
2e6599cb | 74 | { |
54e6ecb2 | 75 | struct request_sock *req = kmem_cache_alloc(ops->slab, GFP_ATOMIC); |
2e6599cb ACM |
76 | |
77 | if (req != NULL) | |
60236fdd | 78 | req->rsk_ops = ops; |
2e6599cb ACM |
79 | |
80 | return req; | |
81 | } | |
82 | ||
60236fdd | 83 | static inline void __reqsk_free(struct request_sock *req) |
2e6599cb | 84 | { |
60236fdd | 85 | kmem_cache_free(req->rsk_ops->slab, req); |
2e6599cb ACM |
86 | } |
87 | ||
60236fdd | 88 | static inline void reqsk_free(struct request_sock *req) |
2e6599cb | 89 | { |
60236fdd ACM |
90 | req->rsk_ops->destructor(req); |
91 | __reqsk_free(req); | |
2e6599cb ACM |
92 | } |
93 | ||
0e87506f ACM |
94 | extern int sysctl_max_syn_backlog; |
95 | ||
2ad69c55 | 96 | /** struct listen_sock - listen state |
0e87506f ACM |
97 | * |
98 | * @max_qlen_log - log_2 of maximal queued SYNs/REQUESTs | |
99 | */ | |
2ad69c55 | 100 | struct listen_sock { |
0e87506f | 101 | u8 max_qlen_log; |
946cedcc ED |
102 | u8 synflood_warned; |
103 | /* 2 bytes hole, try to use */ | |
0e87506f ACM |
104 | int qlen; |
105 | int qlen_young; | |
106 | int clock_hand; | |
107 | u32 hash_rnd; | |
83e3609e | 108 | u32 nr_table_entries; |
0e87506f ACM |
109 | struct request_sock *syn_table[0]; |
110 | }; | |
111 | ||
10467163 JC |
112 | /* |
113 | * For a TCP Fast Open listener - | |
114 | * lock - protects the access to all the reqsk, which is co-owned by | |
115 | * the listener and the child socket. | |
116 | * qlen - pending TFO requests (still in TCP_SYN_RECV). | |
117 | * max_qlen - max TFO reqs allowed before TFO is disabled. | |
118 | * | |
119 | * XXX (TFO) - ideally these fields can be made as part of "listen_sock" | |
120 | * structure above. But there is some implementation difficulty due to | |
121 | * listen_sock being part of request_sock_queue hence will be freed when | |
122 | * a listener is stopped. But TFO related fields may continue to be | |
123 | * accessed even after a listener is closed, until its sk_refcnt drops | |
124 | * to 0 implying no more outstanding TFO reqs. One solution is to keep | |
125 | * listen_opt around until sk_refcnt drops to 0. But there is some other | |
126 | * complexity that needs to be resolved. E.g., a listener can be disabled | |
127 | * temporarily through shutdown()->tcp_disconnect(), and re-enabled later. | |
128 | */ | |
129 | struct fastopen_queue { | |
130 | struct request_sock *rskq_rst_head; /* Keep track of past TFO */ | |
131 | struct request_sock *rskq_rst_tail; /* requests that caused RST. | |
132 | * This is part of the defense | |
133 | * against spoofing attack. | |
134 | */ | |
135 | spinlock_t lock; | |
136 | int qlen; /* # of pending (TCP_SYN_RECV) reqs */ | |
137 | int max_qlen; /* != 0 iff TFO is currently enabled */ | |
138 | }; | |
139 | ||
0e87506f ACM |
140 | /** struct request_sock_queue - queue of request_socks |
141 | * | |
142 | * @rskq_accept_head - FIFO head of established children | |
143 | * @rskq_accept_tail - FIFO tail of established children | |
295f7324 | 144 | * @rskq_defer_accept - User waits for some data after accept() |
0e87506f ACM |
145 | * @syn_wait_lock - serializer |
146 | * | |
147 | * %syn_wait_lock is necessary only to avoid proc interface having to grab the main | |
148 | * lock sock while browsing the listening hash (otherwise it's deadlock prone). | |
149 | * | |
150 | * This lock is acquired in read mode only from listening_get_next() seq_file | |
151 | * op and it's acquired in write mode _only_ from code that is actively | |
152 | * changing rskq_accept_head. All readers that are holding the master sock lock | |
153 | * don't need to grab this lock in read mode too as rskq_accept_head. writes | |
154 | * are always protected from the main sock lock. | |
155 | */ | |
156 | struct request_sock_queue { | |
157 | struct request_sock *rskq_accept_head; | |
158 | struct request_sock *rskq_accept_tail; | |
159 | rwlock_t syn_wait_lock; | |
ec0a1966 DM |
160 | u8 rskq_defer_accept; |
161 | /* 3 bytes hole, try to pack */ | |
2ad69c55 | 162 | struct listen_sock *listen_opt; |
10467163 JC |
163 | struct fastopen_queue *fastopenq; /* This is non-NULL iff TFO has been |
164 | * enabled on this listener. Check | |
165 | * max_qlen != 0 in fastopen_queue | |
166 | * to determine if TFO is enabled | |
167 | * right at this moment. | |
168 | */ | |
0e87506f ACM |
169 | }; |
170 | ||
171 | extern int reqsk_queue_alloc(struct request_sock_queue *queue, | |
72a3effa | 172 | unsigned int nr_table_entries); |
0e87506f | 173 | |
dab6ba36 | 174 | extern void __reqsk_queue_destroy(struct request_sock_queue *queue); |
83e3609e | 175 | extern void reqsk_queue_destroy(struct request_sock_queue *queue); |
10467163 JC |
176 | extern void reqsk_fastopen_remove(struct sock *sk, |
177 | struct request_sock *req, bool reset); | |
83e3609e | 178 | |
0e87506f ACM |
179 | static inline struct request_sock * |
180 | reqsk_queue_yank_acceptq(struct request_sock_queue *queue) | |
181 | { | |
182 | struct request_sock *req = queue->rskq_accept_head; | |
183 | ||
3eb4801d | 184 | queue->rskq_accept_head = NULL; |
0e87506f ACM |
185 | return req; |
186 | } | |
187 | ||
188 | static inline int reqsk_queue_empty(struct request_sock_queue *queue) | |
189 | { | |
190 | return queue->rskq_accept_head == NULL; | |
191 | } | |
192 | ||
193 | static inline void reqsk_queue_unlink(struct request_sock_queue *queue, | |
194 | struct request_sock *req, | |
195 | struct request_sock **prev_req) | |
196 | { | |
197 | write_lock(&queue->syn_wait_lock); | |
198 | *prev_req = req->dl_next; | |
199 | write_unlock(&queue->syn_wait_lock); | |
200 | } | |
201 | ||
202 | static inline void reqsk_queue_add(struct request_sock_queue *queue, | |
203 | struct request_sock *req, | |
204 | struct sock *parent, | |
205 | struct sock *child) | |
206 | { | |
207 | req->sk = child; | |
208 | sk_acceptq_added(parent); | |
209 | ||
210 | if (queue->rskq_accept_head == NULL) | |
211 | queue->rskq_accept_head = req; | |
212 | else | |
213 | queue->rskq_accept_tail->dl_next = req; | |
214 | ||
215 | queue->rskq_accept_tail = req; | |
216 | req->dl_next = NULL; | |
217 | } | |
218 | ||
219 | static inline struct request_sock *reqsk_queue_remove(struct request_sock_queue *queue) | |
220 | { | |
221 | struct request_sock *req = queue->rskq_accept_head; | |
222 | ||
547b792c | 223 | WARN_ON(req == NULL); |
0e87506f ACM |
224 | |
225 | queue->rskq_accept_head = req->dl_next; | |
226 | if (queue->rskq_accept_head == NULL) | |
227 | queue->rskq_accept_tail = NULL; | |
228 | ||
229 | return req; | |
230 | } | |
231 | ||
0e87506f ACM |
232 | static inline int reqsk_queue_removed(struct request_sock_queue *queue, |
233 | struct request_sock *req) | |
234 | { | |
2ad69c55 | 235 | struct listen_sock *lopt = queue->listen_opt; |
0e87506f | 236 | |
e6c022a4 | 237 | if (req->num_timeout == 0) |
0e87506f ACM |
238 | --lopt->qlen_young; |
239 | ||
240 | return --lopt->qlen; | |
241 | } | |
242 | ||
243 | static inline int reqsk_queue_added(struct request_sock_queue *queue) | |
244 | { | |
2ad69c55 | 245 | struct listen_sock *lopt = queue->listen_opt; |
0e87506f ACM |
246 | const int prev_qlen = lopt->qlen; |
247 | ||
248 | lopt->qlen_young++; | |
249 | lopt->qlen++; | |
250 | return prev_qlen; | |
251 | } | |
252 | ||
463c84b9 | 253 | static inline int reqsk_queue_len(const struct request_sock_queue *queue) |
0e87506f ACM |
254 | { |
255 | return queue->listen_opt != NULL ? queue->listen_opt->qlen : 0; | |
256 | } | |
257 | ||
463c84b9 | 258 | static inline int reqsk_queue_len_young(const struct request_sock_queue *queue) |
0e87506f ACM |
259 | { |
260 | return queue->listen_opt->qlen_young; | |
261 | } | |
262 | ||
463c84b9 | 263 | static inline int reqsk_queue_is_full(const struct request_sock_queue *queue) |
0e87506f ACM |
264 | { |
265 | return queue->listen_opt->qlen >> queue->listen_opt->max_qlen_log; | |
266 | } | |
267 | ||
268 | static inline void reqsk_queue_hash_req(struct request_sock_queue *queue, | |
269 | u32 hash, struct request_sock *req, | |
8129765a | 270 | unsigned long timeout) |
0e87506f | 271 | { |
2ad69c55 | 272 | struct listen_sock *lopt = queue->listen_opt; |
0e87506f ACM |
273 | |
274 | req->expires = jiffies + timeout; | |
e6c022a4 ED |
275 | req->num_retrans = 0; |
276 | req->num_timeout = 0; | |
0e87506f ACM |
277 | req->sk = NULL; |
278 | req->dl_next = lopt->syn_table[hash]; | |
279 | ||
280 | write_lock(&queue->syn_wait_lock); | |
281 | lopt->syn_table[hash] = req; | |
282 | write_unlock(&queue->syn_wait_lock); | |
283 | } | |
284 | ||
2e6599cb | 285 | #endif /* _REQUEST_SOCK_H */ |