Commit | Line | Data |
---|---|---|
2874c5fd | 1 | /* SPDX-License-Identifier: GPL-2.0-or-later */ |
1da177e4 LT |
2 | /* |
3 | * INET An implementation of the TCP/IP protocol suite for the LINUX | |
4 | * operating system. INET is implemented using the BSD Socket | |
5 | * interface as the means of communication with the user level. | |
6 | * | |
7 | * Definitions for the TCP protocol. | |
8 | * | |
9 | * Version: @(#)tcp.h 1.0.2 04/28/93 | |
10 | * | |
11 | * Author: Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> | |
1da177e4 LT |
12 | */ |
13 | #ifndef _LINUX_TCP_H | |
14 | #define _LINUX_TCP_H | |
15 | ||
1da177e4 | 16 | |
1da177e4 | 17 | #include <linux/skbuff.h> |
64033892 | 18 | #include <linux/win_minmax.h> |
1da177e4 | 19 | #include <net/sock.h> |
463c84b9 | 20 | #include <net/inet_connection_sock.h> |
8feaf0c0 | 21 | #include <net/inet_timewait_sock.h> |
607ca46e | 22 | #include <uapi/linux/tcp.h> |
1da177e4 | 23 | |
aa8223c7 ACM |
24 | static inline struct tcphdr *tcp_hdr(const struct sk_buff *skb) |
25 | { | |
9c70220b | 26 | return (struct tcphdr *)skb_transport_header(skb); |
aa8223c7 ACM |
27 | } |
28 | ||
d9b3fca2 CG |
29 | static inline unsigned int __tcp_hdrlen(const struct tcphdr *th) |
30 | { | |
31 | return th->doff * 4; | |
32 | } | |
33 | ||
ab6a5bb6 ACM |
34 | static inline unsigned int tcp_hdrlen(const struct sk_buff *skb) |
35 | { | |
d9b3fca2 | 36 | return __tcp_hdrlen(tcp_hdr(skb)); |
ab6a5bb6 ACM |
37 | } |
38 | ||
6a674e9c JG |
39 | static inline struct tcphdr *inner_tcp_hdr(const struct sk_buff *skb) |
40 | { | |
41 | return (struct tcphdr *)skb_inner_transport_header(skb); | |
42 | } | |
43 | ||
44 | static inline unsigned int inner_tcp_hdrlen(const struct sk_buff *skb) | |
45 | { | |
46 | return inner_tcp_hdr(skb)->doff * 4; | |
47 | } | |
48 | ||
504148fe ED |
49 | /** |
50 | * skb_tcp_all_headers - Returns size of all headers for a TCP packet | |
51 | * @skb: buffer | |
52 | * | |
53 | * Used in TX path, for a packet known to be a TCP one. | |
54 | * | |
55 | * if (skb_is_gso(skb)) { | |
56 | * int hlen = skb_tcp_all_headers(skb); | |
57 | * ... | |
58 | */ | |
59 | static inline int skb_tcp_all_headers(const struct sk_buff *skb) | |
60 | { | |
61 | return skb_transport_offset(skb) + tcp_hdrlen(skb); | |
62 | } | |
63 | ||
64 | /** | |
65 | * skb_inner_tcp_all_headers - Returns size of all headers for an encap TCP packet | |
66 | * @skb: buffer | |
67 | * | |
68 | * Used in TX path, for a packet known to be a TCP one. | |
69 | * | |
70 | * if (skb_is_gso(skb) && skb->encapsulation) { | |
71 | * int hlen = skb_inner_tcp_all_headers(skb); | |
72 | * ... | |
73 | */ | |
74 | static inline int skb_inner_tcp_all_headers(const struct sk_buff *skb) | |
75 | { | |
76 | return skb_inner_transport_offset(skb) + inner_tcp_hdrlen(skb); | |
77 | } | |
78 | ||
ab6a5bb6 ACM |
79 | static inline unsigned int tcp_optlen(const struct sk_buff *skb) |
80 | { | |
aa8223c7 | 81 | return (tcp_hdr(skb)->doff - 5) * 4; |
ab6a5bb6 ACM |
82 | } |
83 | ||
2100c8d2 YC |
84 | /* TCP Fast Open */ |
85 | #define TCP_FASTOPEN_COOKIE_MIN 4 /* Min Fast Open Cookie size in bytes */ | |
86 | #define TCP_FASTOPEN_COOKIE_MAX 16 /* Max Fast Open Cookie size in bytes */ | |
10467163 | 87 | #define TCP_FASTOPEN_COOKIE_SIZE 8 /* the size employed by this impl. */ |
2100c8d2 YC |
88 | |
89 | /* TCP Fast Open Cookie as stored in memory */ | |
90 | struct tcp_fastopen_cookie { | |
438ac880 | 91 | __le64 val[DIV_ROUND_UP(TCP_FASTOPEN_COOKIE_MAX, sizeof(u64))]; |
2100c8d2 | 92 | s8 len; |
7f9b838b | 93 | bool exp; /* In RFC6994 experimental option format */ |
2100c8d2 YC |
94 | }; |
95 | ||
1da177e4 | 96 | /* This defines a selective acknowledgement block. */ |
269bd27e AV |
97 | struct tcp_sack_block_wire { |
98 | __be32 start_seq; | |
99 | __be32 end_seq; | |
100 | }; | |
101 | ||
1da177e4 | 102 | struct tcp_sack_block { |
3a137d20 ACM |
103 | u32 start_seq; |
104 | u32 end_seq; | |
1da177e4 LT |
105 | }; |
106 | ||
ab56222a VS |
107 | /*These are used to set the sack_ok field in struct tcp_options_received */ |
108 | #define TCP_SACK_SEEN (1 << 0) /*1 = peer is SACK capable, */ | |
ab56222a VS |
109 | #define TCP_DSACK_SEEN (1 << 2) /*1 = DSACK was received from peer*/ |
110 | ||
1da177e4 LT |
111 | struct tcp_options_received { |
112 | /* PAWS/RTTM data */ | |
cca9bab1 | 113 | int ts_recent_stamp;/* Time we stored ts_recent (for aging) */ |
3a137d20 ACM |
114 | u32 ts_recent; /* Time stamp to echo next */ |
115 | u32 rcv_tsval; /* Time stamp value */ | |
116 | u32 rcv_tsecr; /* Time stamp echo reply */ | |
117 | u16 saw_tstamp : 1, /* Saw TIMESTAMP on last packet */ | |
1da177e4 LT |
118 | tstamp_ok : 1, /* TIMESTAMP seen on SYN packet */ |
119 | dsack : 1, /* D-SACK is scheduled */ | |
120 | wscale_ok : 1, /* Wscale seen on SYN packet */ | |
60e2a778 UB |
121 | sack_ok : 3, /* SACK seen on SYN packet */ |
122 | smc_ok : 1, /* SMC seen on SYN packet */ | |
1da177e4 LT |
123 | snd_wscale : 4, /* Window scaling received from sender */ |
124 | rcv_wscale : 4; /* Window scaling to send to receiver */ | |
7656d684 MKL |
125 | u8 saw_unknown:1, /* Received unknown option */ |
126 | unused:7; | |
3a137d20 | 127 | u8 num_sacks; /* Number of SACK blocks */ |
435cf559 | 128 | u16 user_mss; /* mss requested by user in ioctl */ |
3a137d20 | 129 | u16 mss_clamp; /* Maximal mss, negotiated at connection setup */ |
1da177e4 LT |
130 | }; |
131 | ||
519855c5 WAS |
132 | static inline void tcp_clear_options(struct tcp_options_received *rx_opt) |
133 | { | |
435cf559 WAS |
134 | rx_opt->tstamp_ok = rx_opt->sack_ok = 0; |
135 | rx_opt->wscale_ok = rx_opt->snd_wscale = 0; | |
60e2a778 UB |
136 | #if IS_ENABLED(CONFIG_SMC) |
137 | rx_opt->smc_ok = 0; | |
138 | #endif | |
519855c5 WAS |
139 | } |
140 | ||
4389dded | 141 | /* This is the max number of SACKS that we'll generate and process. It's safe |
435cf559 | 142 | * to increase this, although since: |
4389dded AL |
143 | * size = TCPOLEN_SACK_BASE_ALIGNED (4) + n * TCPOLEN_SACK_PERBLOCK (8) |
144 | * only four options will fit in a standard TCP header */ | |
145 | #define TCP_NUM_SACKS 4 | |
146 | ||
435cf559 WAS |
147 | struct tcp_request_sock_ops; |
148 | ||
2e6599cb | 149 | struct tcp_request_sock { |
cfb6eeb4 | 150 | struct inet_request_sock req; |
b2e4b3de | 151 | const struct tcp_request_sock_ops *af_specific; |
9a568de4 | 152 | u64 snt_synack; /* first SYNACK sent time */ |
9439ce00 | 153 | bool tfo_listener; |
cec37a6e | 154 | bool is_mptcp; |
90bf4513 PA |
155 | #if IS_ENABLED(CONFIG_MPTCP) |
156 | bool drop_req; | |
157 | #endif | |
58d607d3 | 158 | u32 txhash; |
435cf559 WAS |
159 | u32 rcv_isn; |
160 | u32 snt_isn; | |
95a22cae | 161 | u32 ts_off; |
a9b2c06d | 162 | u32 last_oow_ack_time; /* last SYNACK */ |
10467163 JC |
163 | u32 rcv_nxt; /* the ack # by SYNACK. For |
164 | * FastOpen it's the seq# | |
165 | * after data-in-SYN. | |
166 | */ | |
e9b12edc | 167 | u8 syn_tos; |
2e6599cb ACM |
168 | }; |
169 | ||
60236fdd | 170 | static inline struct tcp_request_sock *tcp_rsk(const struct request_sock *req) |
2e6599cb ACM |
171 | { |
172 | return (struct tcp_request_sock *)req; | |
173 | } | |
174 | ||
dfa2f048 ED |
175 | #define TCP_RMEM_TO_WIN_SCALE 8 |
176 | ||
1da177e4 | 177 | struct tcp_sock { |
463c84b9 ACM |
178 | /* inet_connection_sock has to be the first member of tcp_sock */ |
179 | struct inet_connection_sock inet_conn; | |
2ff52f28 | 180 | u16 tcp_header_len; /* Bytes of tcp header to send */ |
605ad7f1 | 181 | u16 gso_segs; /* Max number of segs per GSO packet */ |
1da177e4 | 182 | |
31770e34 FW |
183 | /* |
184 | * Header prediction flags | |
185 | * 0x5?10 << 16 + snd_wnd in net byte order | |
186 | */ | |
187 | __be32 pred_flags; | |
188 | ||
1da177e4 LT |
189 | /* |
190 | * RFC793 variables by their proper names. This means you can | |
191 | * read the code and the spec side by side (and laugh ...) | |
192 | * See RFC793 and RFC1122. The RFC writes these in capitals. | |
193 | */ | |
bdd1f9ed ED |
194 | u64 bytes_received; /* RFC4898 tcpEStatsAppHCThruOctetsReceived |
195 | * sum(delta(rcv_nxt)), or how many bytes | |
196 | * were acked. | |
197 | */ | |
2efd055c MRL |
198 | u32 segs_in; /* RFC4898 tcpEStatsPerfSegsIn |
199 | * total number of segments in. | |
200 | */ | |
a44d6eac MKL |
201 | u32 data_segs_in; /* RFC4898 tcpEStatsPerfDataSegsIn |
202 | * total number of data segments in. | |
203 | */ | |
3a137d20 | 204 | u32 rcv_nxt; /* What we want to receive next */ |
54287cc1 ED |
205 | u32 copied_seq; /* Head of yet unread data */ |
206 | u32 rcv_wup; /* rcv_nxt on last window update sent */ | |
3a137d20 | 207 | u32 snd_nxt; /* Next sequence we send */ |
2efd055c MRL |
208 | u32 segs_out; /* RFC4898 tcpEStatsPerfSegsOut |
209 | * The total number of segments sent. | |
210 | */ | |
a44d6eac MKL |
211 | u32 data_segs_out; /* RFC4898 tcpEStatsPerfDataSegsOut |
212 | * total number of data segments sent. | |
213 | */ | |
ba113c3a WW |
214 | u64 bytes_sent; /* RFC4898 tcpEStatsPerfHCDataOctetsOut |
215 | * total number of data bytes sent. | |
216 | */ | |
0df48c26 ED |
217 | u64 bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked |
218 | * sum(delta(snd_una)), or how many bytes | |
219 | * were acked. | |
220 | */ | |
7e10b655 WW |
221 | u32 dsack_dups; /* RFC4898 tcpEStatsStackDSACKDups |
222 | * total number of DSACK blocks received | |
223 | */ | |
3a137d20 ACM |
224 | u32 snd_una; /* First byte we want an ack for */ |
225 | u32 snd_sml; /* Last byte of the most recently transmitted small packet */ | |
226 | u32 rcv_tstamp; /* timestamp of last received ACK (for keepalives) */ | |
227 | u32 lsndtime; /* timestamp of last sent data packet (for restart window) */ | |
f2b2c582 | 228 | u32 last_oow_ack_time; /* timestamp of last out-of-window ACK */ |
86de5921 | 229 | u32 compressed_ack_rcv_nxt; |
1da177e4 | 230 | |
ceaa1fef AV |
231 | u32 tsoffset; /* timestamp offset */ |
232 | ||
46d3ceab | 233 | struct list_head tsq_node; /* anchor in tsq_tasklet.head list */ |
e2080072 | 234 | struct list_head tsorted_sent_queue; /* time-sorted sent but un-SACKed skbs */ |
46d3ceab | 235 | |
3a137d20 ACM |
236 | u32 snd_wl1; /* Sequence for window update */ |
237 | u32 snd_wnd; /* The window we expect to receive */ | |
238 | u32 max_window; /* Maximal window ever seen from peer */ | |
239 | u32 mss_cache; /* Cached effective mss, not including SACKS */ | |
1da177e4 | 240 | |
3a137d20 ACM |
241 | u32 window_clamp; /* Maximal window to advertise */ |
242 | u32 rcv_ssthresh; /* Current window clamp */ | |
dfa2f048 | 243 | u8 scaling_ratio; /* see tcp_win_from_space() */ |
659a8ad5 YC |
244 | /* Information of the most recently (s)acked skb */ |
245 | struct tcp_rack { | |
9a568de4 | 246 | u64 mstamp; /* (Re)sent time of the skb */ |
deed7be7 | 247 | u32 rtt_us; /* Associated RTT */ |
1d0833df | 248 | u32 end_seq; /* Ending TCP sequence of the skb */ |
1f255691 PJ |
249 | u32 last_delivered; /* tp->delivered at last reo_wnd adj */ |
250 | u8 reo_wnd_steps; /* Allowed reordering window */ | |
251 | #define TCP_RACK_RECOVERY_THRESH 16 | |
252 | u8 reo_wnd_persist:5, /* No. of recovery since last adj */ | |
253 | dsack_seen:1, /* Whether DSACK seen after last adj */ | |
7ec65372 | 254 | advanced:1; /* mstamp advanced since last lost marking */ |
659a8ad5 | 255 | } rack; |
b79eeeb9 | 256 | u16 advmss; /* Advertised MSS */ |
5d9f4262 | 257 | u8 compressed_ack; |
76be93fc YC |
258 | u8 dup_ack_counter:2, |
259 | tlp_retrans:1, /* TLP is a retransmission */ | |
260 | unused:5; | |
05b055e8 FY |
261 | u32 chrono_start; /* Start time in jiffies of a TCP chrono */ |
262 | u32 chrono_stat[3]; /* Time in jiffies for chrono_stat stats */ | |
263 | u8 chrono_type:2, /* current chronograph type */ | |
264 | rate_app_limited:1, /* rate_{delivered,interval_us} limited? */ | |
19f6d3f3 | 265 | fastopen_connect:1, /* FASTOPEN_CONNECT sockopt */ |
71c02379 | 266 | fastopen_no_cookie:1, /* Allow send/recv SYN+data without a cookie */ |
d4761754 | 267 | is_sack_reneg:1, /* in recovery from loss with SACK reneg? */ |
48027478 | 268 | fastopen_client_fail:2; /* reason why fastopen failed */ |
36e31b0a AP |
269 | u8 nonagle : 4,/* Disable Nagle algorithm? */ |
270 | thin_lto : 1,/* Use linear timeouts for thin streams */ | |
b75eba76 | 271 | recvmsg_inq : 1,/* Indicate # of bytes in queue upon recvmsg */ |
e33099f9 YC |
272 | repair : 1, |
273 | frto : 1;/* F-RTO (RFC5682) activated in CA_Loss */ | |
ee995283 | 274 | u8 repair_queue; |
267cf9fa MKL |
275 | u8 save_syn:2, /* Save headers of SYN packet */ |
276 | syn_data:1, /* SYN includes data */ | |
6f73601e | 277 | syn_fastopen:1, /* SYN includes Fast Open option */ |
2646c831 | 278 | syn_fastopen_exp:1,/* SYN includes Fast Open exp. option */ |
cf1ef3f0 | 279 | syn_fastopen_ch:1, /* Active TFO re-enabling probe */ |
ca8a2263 | 280 | syn_data_acked:1,/* data in SYN is acked by SYN-ACK */ |
267cf9fa | 281 | is_cwnd_limited:1;/* forward progress limited by snd_cwnd? */ |
76be93fc | 282 | u32 tlp_high_seq; /* snd_nxt at the time of TLP */ |
1da177e4 | 283 | |
a842fe14 | 284 | u32 tcp_tx_delay; /* delay (in usec) added to TX packets */ |
9799ccb0 | 285 | u64 tcp_wstamp_ns; /* departure time for next sent data packet */ |
5f6188a8 | 286 | u64 tcp_clock_cache; /* cache last tcp_clock_ns() (see tcp_mstamp_refresh()) */ |
9799ccb0 | 287 | |
1da177e4 | 288 | /* RTT measurement */ |
9a568de4 | 289 | u64 tcp_mstamp; /* most recent packet received/sent */ |
740b0f18 ED |
290 | u32 srtt_us; /* smoothed round trip time << 3 in usecs */ |
291 | u32 mdev_us; /* medium deviation */ | |
292 | u32 mdev_max_us; /* maximal mdev for the last rtt period */ | |
293 | u32 rttvar_us; /* smoothed mdev_max */ | |
3a137d20 | 294 | u32 rtt_seq; /* sequence number to update rttvar */ |
64033892 | 295 | struct minmax rtt_min; |
3a137d20 ACM |
296 | |
297 | u32 packets_out; /* Packets which are "in flight" */ | |
3a137d20 | 298 | u32 retrans_out; /* Retransmitted packets out */ |
ca8a2263 | 299 | u32 max_packets_out; /* max packets_out in last window */ |
f4ce91ce | 300 | u32 cwnd_usage_seq; /* right edge of cwnd usage tracking flight */ |
b79eeeb9 IJ |
301 | |
302 | u16 urg_data; /* Saved octet of OOB data and control flags */ | |
b79eeeb9 | 303 | u8 ecn_flags; /* ECN status bits. */ |
dca145ff ED |
304 | u8 keepalive_probes; /* num of allowed keep alive probes */ |
305 | u32 reordering; /* Packet reordering metric. */ | |
7ec65372 | 306 | u32 reord_seen; /* number of data packet reordering events */ |
33f5f57e IJ |
307 | u32 snd_up; /* Urgent pointer */ |
308 | ||
1da177e4 LT |
309 | /* |
310 | * Options received (usually on last packet, some only on SYN packets). | |
311 | */ | |
1da177e4 LT |
312 | struct tcp_options_received rx_opt; |
313 | ||
314 | /* | |
315 | * Slow start and congestion control (see also Nagle, and Karn & Partridge) | |
316 | */ | |
3a137d20 ACM |
317 | u32 snd_ssthresh; /* Slow start size threshold */ |
318 | u32 snd_cwnd; /* Sending congestion window */ | |
f78a1b38 | 319 | u32 snd_cwnd_cnt; /* Linear increase counter */ |
e0ef57cc | 320 | u32 snd_cwnd_clamp; /* Do not allow snd_cwnd to grow above this */ |
3a137d20 ACM |
321 | u32 snd_cwnd_used; |
322 | u32 snd_cwnd_stamp; | |
4faf7839 | 323 | u32 prior_cwnd; /* cwnd right before starting loss recovery */ |
a262f0cd ND |
324 | u32 prr_delivered; /* Number of newly delivered packets to |
325 | * receiver in Recovery. */ | |
326 | u32 prr_out; /* Total number of pkts sent during Recovery. */ | |
ddf1af6f | 327 | u32 delivered; /* Total data packets delivered incl. rexmits */ |
e21db6f6 | 328 | u32 delivered_ce; /* Like the above but only ECE marked packets */ |
0682e690 | 329 | u32 lost; /* Total data packets lost incl. rexmits */ |
d7722e85 | 330 | u32 app_limited; /* limited until "delivered" reaches this val */ |
9a568de4 ED |
331 | u64 first_tx_mstamp; /* start of window send phase */ |
332 | u64 delivered_mstamp; /* time we reached "delivered" */ | |
eb8329e0 YC |
333 | u32 rate_delivered; /* saved rate sample: packets delivered */ |
334 | u32 rate_interval_us; /* saved rate sample: time elapsed */ | |
1da177e4 | 335 | |
3a137d20 | 336 | u32 rcv_wnd; /* Current receiver window */ |
3a137d20 | 337 | u32 write_seq; /* Tail(+1) of data held in tcp send buffer */ |
c9bee3b7 | 338 | u32 notsent_lowat; /* TCP_NOTSENT_LOWAT */ |
3a137d20 | 339 | u32 pushed_seq; /* Last pushed seq, required to talk to windows */ |
b79eeeb9 IJ |
340 | u32 lost_out; /* Lost packets */ |
341 | u32 sacked_out; /* SACK'd packets */ | |
1da177e4 | 342 | |
218af599 | 343 | struct hrtimer pacing_timer; |
5d9f4262 | 344 | struct hrtimer compressed_ack_timer; |
218af599 | 345 | |
b79eeeb9 IJ |
346 | /* from STCP, retrans queue hinting */ |
347 | struct sk_buff* lost_skb_hint; | |
b79eeeb9 | 348 | struct sk_buff *retransmit_skb_hint; |
b79eeeb9 | 349 | |
9f5afeae YW |
350 | /* OOO segments go in this rbtree. Socket lock must be held. */ |
351 | struct rb_root out_of_order_queue; | |
352 | struct sk_buff *ooo_last_skb; /* cache rb_last(out_of_order_queue) */ | |
b79eeeb9 | 353 | |
c0a788c4 | 354 | /* SACKs data, these 2 need to be together (see tcp_options_write) */ |
1da177e4 LT |
355 | struct tcp_sack_block duplicate_sack[1]; /* D-SACK block */ |
356 | struct tcp_sack_block selective_acks[4]; /* The SACKS themselves*/ | |
357 | ||
fd6dad61 | 358 | struct tcp_sack_block recv_sack_cache[4]; |
6a438bbe | 359 | |
ecb97192 NC |
360 | struct sk_buff *highest_sack; /* skb just after the highest |
361 | * skb with SACKed bit set | |
a47e5a98 IJ |
362 | * (validity guaranteed only if |
363 | * sacked_out > 0) | |
364 | */ | |
d738cd8f | 365 | |
6a438bbe | 366 | int lost_cnt_hint; |
6a438bbe | 367 | |
4b749440 | 368 | u32 prior_ssthresh; /* ssthresh saved at recovery start */ |
3a137d20 | 369 | u32 high_seq; /* snd_nxt at onset of congestion */ |
1da177e4 | 370 | |
3a137d20 | 371 | u32 retrans_stamp; /* Timestamp of the last retransmit, |
1da177e4 LT |
372 | * also used in SYN-SENT to remember stamp of |
373 | * the first SYN. */ | |
989e04c5 | 374 | u32 undo_marker; /* snd_una upon a new recovery episode. */ |
1da177e4 | 375 | int undo_retrans; /* number of undoable retransmissions. */ |
fb31c9b9 WW |
376 | u64 bytes_retrans; /* RFC4898 tcpEStatsPerfOctetsRetrans |
377 | * Total data bytes retransmitted | |
378 | */ | |
b79eeeb9 | 379 | u32 total_retrans; /* Total retransmits for entire connection */ |
3868ab0f A |
380 | u32 rto_stamp; /* Start time (ms) of last CA_Loss recovery */ |
381 | u16 total_rto; /* Total number of RTO timeouts, including | |
382 | * SYN/SYN-ACK and recurring timeouts. | |
383 | */ | |
384 | u16 total_rto_recoveries; /* Total number of RTO recoveries, | |
385 | * including any unfinished recovery. | |
386 | */ | |
387 | u32 total_rto_time; /* ms spent in (completed) RTO recoveries. */ | |
b79eeeb9 | 388 | |
3a137d20 | 389 | u32 urg_seq; /* Seq of received urgent pointer */ |
1da177e4 LT |
390 | unsigned int keepalive_time; /* time before keep alive takes place */ |
391 | unsigned int keepalive_intvl; /* time interval between keep alive probes */ | |
1da177e4 | 392 | |
a0f82f64 | 393 | int linger2; |
1da177e4 | 394 | |
b13d8807 LB |
395 | |
396 | /* Sock_ops bpf program related variables */ | |
397 | #ifdef CONFIG_BPF | |
398 | u8 bpf_sock_ops_cb_flags; /* Control calling BPF programs | |
399 | * values defined in uapi/linux/tcp.h | |
400 | */ | |
061ff040 MKL |
401 | u8 bpf_chg_cc_inprogress:1; /* In the middle of |
402 | * bpf_setsockopt(TCP_CONGESTION), | |
403 | * it is to avoid the bpf_tcp_cc->init() | |
404 | * to recur itself by calling | |
405 | * bpf_setsockopt(TCP_CONGESTION, "itself"). | |
406 | */ | |
b13d8807 LB |
407 | #define BPF_SOCK_OPS_TEST_FLAG(TP, ARG) (TP->bpf_sock_ops_cb_flags & ARG) |
408 | #else | |
409 | #define BPF_SOCK_OPS_TEST_FLAG(TP, ARG) 0 | |
410 | #endif | |
411 | ||
32efcc06 AK |
412 | u16 timeout_rehash; /* Timeout-triggered rehash attempts */ |
413 | ||
f9af2dbb TH |
414 | u32 rcv_ooopack; /* Received out-of-order packets, for tcpinfo */ |
415 | ||
1da177e4 | 416 | /* Receiver side RTT estimation */ |
3f6c65d6 | 417 | u32 rcv_rtt_last_tsecr; |
1da177e4 | 418 | struct { |
9a568de4 ED |
419 | u32 rtt_us; |
420 | u32 seq; | |
421 | u64 time; | |
1da177e4 LT |
422 | } rcv_rtt_est; |
423 | ||
424 | /* Receiver queue space */ | |
425 | struct { | |
607065ba | 426 | u32 space; |
9a568de4 ED |
427 | u32 seq; |
428 | u64 time; | |
1da177e4 | 429 | } rcvq_space; |
0e7b1368 JH |
430 | |
431 | /* TCP-specific MTU probe information. */ | |
432 | struct { | |
3a137d20 ACM |
433 | u32 probe_seq_start; |
434 | u32 probe_seq_end; | |
0e7b1368 | 435 | } mtu_probe; |
29c1c446 | 436 | u32 plb_rehash; /* PLB-triggered rehash attempts */ |
563d34d0 ED |
437 | u32 mtu_info; /* We received an ICMP_FRAG_NEEDED / ICMPV6_PKT_TOOBIG |
438 | * while socket was owned by user. | |
439 | */ | |
2303f994 PK |
440 | #if IS_ENABLED(CONFIG_MPTCP) |
441 | bool is_mptcp; | |
442 | #endif | |
267cf9fa | 443 | #if IS_ENABLED(CONFIG_SMC) |
48b6190a | 444 | bool (*smc_hs_congested)(const struct sock *sk); |
267cf9fa MKL |
445 | bool syn_smc; /* SYN includes SMC */ |
446 | #endif | |
cfb6eeb4 YH |
447 | |
448 | #ifdef CONFIG_TCP_MD5SIG | |
449 | /* TCP AF-Specific parts; only used by MD5 Signature support so far */ | |
b2e4b3de | 450 | const struct tcp_sock_af_ops *af_specific; |
cfb6eeb4 | 451 | |
b2e4b3de | 452 | /* TCP MD5 Signature Option information */ |
a8afca03 | 453 | struct tcp_md5sig_info __rcu *md5sig_info; |
cfb6eeb4 | 454 | #endif |
435cf559 | 455 | |
10467163 JC |
456 | /* TCP fastopen related information */ |
457 | struct tcp_fastopen_request *fastopen_req; | |
458 | /* fastopen_rsk points to request_sock that resulted in this big | |
459 | * socket. Used to retransmit SYNACKs etc. | |
460 | */ | |
d983ea6f | 461 | struct request_sock __rcu *fastopen_rsk; |
70a217f1 | 462 | struct saved_syn *saved_syn; |
1da177e4 LT |
463 | }; |
464 | ||
40fc3423 | 465 | enum tsq_enum { |
46d3ceab ED |
466 | TSQ_THROTTLED, |
467 | TSQ_QUEUED, | |
6f458dfb ED |
468 | TCP_TSQ_DEFERRED, /* tcp_tasklet_func() found socket was owned */ |
469 | TCP_WRITE_TIMER_DEFERRED, /* tcp_write_timer() found socket was owned */ | |
470 | TCP_DELACK_TIMER_DEFERRED, /* tcp_delack_timer() found socket was owned */ | |
563d34d0 ED |
471 | TCP_MTU_REDUCED_DEFERRED, /* tcp_v{4|6}_err() could not call |
472 | * tcp_v{4|6}_mtu_reduced() | |
473 | */ | |
133c4c0d | 474 | TCP_ACK_DEFERRED, /* TX pure ack is deferred */ |
46d3ceab ED |
475 | }; |
476 | ||
40fc3423 | 477 | enum tsq_flags { |
133c4c0d ED |
478 | TSQF_THROTTLED = BIT(TSQ_THROTTLED), |
479 | TSQF_QUEUED = BIT(TSQ_QUEUED), | |
480 | TCPF_TSQ_DEFERRED = BIT(TCP_TSQ_DEFERRED), | |
481 | TCPF_WRITE_TIMER_DEFERRED = BIT(TCP_WRITE_TIMER_DEFERRED), | |
482 | TCPF_DELACK_TIMER_DEFERRED = BIT(TCP_DELACK_TIMER_DEFERRED), | |
483 | TCPF_MTU_REDUCED_DEFERRED = BIT(TCP_MTU_REDUCED_DEFERRED), | |
484 | TCPF_ACK_DEFERRED = BIT(TCP_ACK_DEFERRED), | |
40fc3423 ED |
485 | }; |
486 | ||
e9d9da91 ED |
487 | #define tcp_sk(ptr) container_of_const(ptr, struct tcp_sock, inet_conn.icsk_inet.sk) |
488 | ||
489 | /* Variant of tcp_sk() upgrading a const sock to a read/write tcp socket. | |
490 | * Used in context of (lockless) tcp listeners. | |
491 | */ | |
492 | #define tcp_sk_rw(ptr) container_of(ptr, struct tcp_sock, inet_conn.icsk_inet.sk) | |
1da177e4 | 493 | |
8feaf0c0 ACM |
494 | struct tcp_timewait_sock { |
495 | struct inet_timewait_sock tw_sk; | |
d475f090 ED |
496 | #define tw_rcv_nxt tw_sk.__tw_common.skc_tw_rcv_nxt |
497 | #define tw_snd_nxt tw_sk.__tw_common.skc_tw_snd_nxt | |
3a137d20 | 498 | u32 tw_rcv_wnd; |
ceaa1fef | 499 | u32 tw_ts_offset; |
3a137d20 | 500 | u32 tw_ts_recent; |
4fb17a60 NC |
501 | |
502 | /* The time we sent the last out-of-window ACK: */ | |
503 | u32 tw_last_oow_ack_time; | |
504 | ||
cca9bab1 | 505 | int tw_ts_recent_stamp; |
a842fe14 | 506 | u32 tw_tx_delay; |
cfb6eeb4 | 507 | #ifdef CONFIG_TCP_MD5SIG |
2397849b | 508 | struct tcp_md5sig_key *tw_md5_key; |
cfb6eeb4 | 509 | #endif |
8feaf0c0 ACM |
510 | }; |
511 | ||
512 | static inline struct tcp_timewait_sock *tcp_twsk(const struct sock *sk) | |
513 | { | |
514 | return (struct tcp_timewait_sock *)sk; | |
515 | } | |
516 | ||
10467163 JC |
517 | static inline bool tcp_passive_fastopen(const struct sock *sk) |
518 | { | |
d983ea6f ED |
519 | return sk->sk_state == TCP_SYN_RECV && |
520 | rcu_access_pointer(tcp_sk(sk)->fastopen_rsk) != NULL; | |
10467163 JC |
521 | } |
522 | ||
0536fcc0 | 523 | static inline void fastopen_queue_tune(struct sock *sk, int backlog) |
10467163 | 524 | { |
0536fcc0 | 525 | struct request_sock_queue *queue = &inet_csk(sk)->icsk_accept_queue; |
dbf650b6 | 526 | int somaxconn = READ_ONCE(sock_net(sk)->core.sysctl_somaxconn); |
0536fcc0 | 527 | |
70f360dd | 528 | WRITE_ONCE(queue->fastopenq.max_qlen, min_t(unsigned int, backlog, somaxconn)); |
10467163 JC |
529 | } |
530 | ||
805c4bc0 ED |
531 | static inline void tcp_move_syn(struct tcp_sock *tp, |
532 | struct request_sock *req) | |
533 | { | |
534 | tp->saved_syn = req->saved_syn; | |
535 | req->saved_syn = NULL; | |
536 | } | |
537 | ||
cd8ae852 ED |
538 | static inline void tcp_saved_syn_free(struct tcp_sock *tp) |
539 | { | |
540 | kfree(tp->saved_syn); | |
541 | tp->saved_syn = NULL; | |
542 | } | |
543 | ||
70a217f1 MKL |
544 | static inline u32 tcp_saved_syn_len(const struct saved_syn *saved_syn) |
545 | { | |
267cf9fa MKL |
546 | return saved_syn->mac_hdrlen + saved_syn->network_hdrlen + |
547 | saved_syn->tcp_hdrlen; | |
70a217f1 MKL |
548 | } |
549 | ||
48040793 | 550 | struct sk_buff *tcp_get_timestamping_opt_stats(const struct sock *sk, |
e7ed11ee YS |
551 | const struct sk_buff *orig_skb, |
552 | const struct sk_buff *ack_skb); | |
1c885808 | 553 | |
3541f9e8 ED |
554 | static inline u16 tcp_mss_clamp(const struct tcp_sock *tp, u16 mss) |
555 | { | |
556 | /* We use READ_ONCE() here because socket might not be locked. | |
557 | * This happens for listeners. | |
558 | */ | |
559 | u16 user_mss = READ_ONCE(tp->rx_opt.user_mss); | |
560 | ||
561 | return (user_mss && user_mss < mss) ? user_mss : mss; | |
562 | } | |
3b4929f6 ED |
563 | |
564 | int tcp_skb_shift(struct sk_buff *to, struct sk_buff *from, int pcount, | |
565 | int shiftlen); | |
566 | ||
6fadaa56 | 567 | void __tcp_sock_set_cork(struct sock *sk, bool on); |
db10538a | 568 | void tcp_sock_set_cork(struct sock *sk, bool on); |
480aeb96 | 569 | int tcp_sock_set_keepcnt(struct sock *sk, int val); |
aad4a0a9 | 570 | int tcp_sock_set_keepidle_locked(struct sock *sk, int val); |
71c48eb8 | 571 | int tcp_sock_set_keepidle(struct sock *sk, int val); |
d41ecaac | 572 | int tcp_sock_set_keepintvl(struct sock *sk, int val); |
6fadaa56 | 573 | void __tcp_sock_set_nodelay(struct sock *sk, bool on); |
12abc5ee | 574 | void tcp_sock_set_nodelay(struct sock *sk); |
ddd061b8 | 575 | void tcp_sock_set_quickack(struct sock *sk, int val); |
557eadfc | 576 | int tcp_sock_set_syncnt(struct sock *sk, int val); |
d58f2e15 | 577 | int tcp_sock_set_user_timeout(struct sock *sk, int val); |
db10538a | 578 | |
1da177e4 | 579 | #endif /* _LINUX_TCP_H */ |