Blame view
net/rds/tcp_listen.c
9.77 KB
70041088e RDS: Add TCP tran... |
1 |
/* |
eee2fa6ab rds: Changing IP ... |
2 |
* Copyright (c) 2006, 2018 Oracle and/or its affiliates. All rights reserved. |
70041088e RDS: Add TCP tran... |
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 |
* * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU * General Public License (GPL) Version 2, available from the file * COPYING in the main directory of this source tree, or the * OpenIB.org BSD license below: * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * * - Redistributions of source code must retain the above * copyright notice, this list of conditions and the following * disclaimer. * * - Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials * provided with the distribution. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. * */ #include <linux/kernel.h> |
5a0e3ad6a include cleanup: ... |
34 |
#include <linux/gfp.h> |
70041088e RDS: Add TCP tran... |
35 36 37 38 39 |
#include <linux/in.h> #include <net/tcp.h> #include "rds.h" #include "tcp.h" |
480aeb963 tcp: add tcp_sock... |
40 |
void rds_tcp_keepalive(struct socket *sock) |
f711a6ae0 net/rds: RDS-TCP:... |
41 42 43 44 |
{ /* values below based on xs_udp_default_timeout */ int keepidle = 5; /* send a probe 'keepidle' secs after last data */ int keepcnt = 5; /* number of unack'ed probes before declaring dead */ |
f711a6ae0 net/rds: RDS-TCP:... |
45 |
|
ce3d9544c net: add sock_set... |
46 |
sock_set_keepalive(sock->sk); |
480aeb963 tcp: add tcp_sock... |
47 |
tcp_sock_set_keepcnt(sock->sk, keepcnt); |
71c48eb81 tcp: add tcp_sock... |
48 |
tcp_sock_set_keepidle(sock->sk, keepidle); |
f711a6ae0 net/rds: RDS-TCP:... |
49 50 51 |
/* KEEPINTVL is the interval between successive probes. We follow * the model in xs_tcp_finish_connecting() and re-use keepidle. */ |
d41ecaac9 tcp: add tcp_sock... |
52 |
tcp_sock_set_keepintvl(sock->sk, keepidle); |
f711a6ae0 net/rds: RDS-TCP:... |
53 |
} |
5916e2c15 RDS: TCP: Enable ... |
54 55 56 57 58 59 60 61 62 |
/* rds_tcp_accept_one_path(): if accepting on cp_index > 0, make sure the * client's ipaddr < server's ipaddr. Otherwise, close the accepted * socket and force a reconneect from smaller -> larger ip addr. The reason * we special case cp_index 0 is to allow the rds probe ping itself to itself * get through efficiently. * Since reconnects are only initiated from the node with the numerically * smaller ip address, we recycle conns in RDS_CONN_ERROR on the passive side * by moving them to CONNECTING in this function. */ |
bb7897631 RDS: mark few int... |
63 |
static |
5916e2c15 RDS: TCP: Enable ... |
64 65 66 |
struct rds_tcp_connection *rds_tcp_accept_one_path(struct rds_connection *conn) { int i; |
1a0e100fb RDS: TCP: Force e... |
67 |
int npaths = max_t(int, 1, conn->c_npaths); |
5916e2c15 RDS: TCP: Enable ... |
68 |
|
1a0e100fb RDS: TCP: Force e... |
69 |
/* for mprds, all paths MUST be initiated by the peer |
5916e2c15 RDS: TCP: Enable ... |
70 71 |
* with the smaller address. */ |
eee2fa6ab rds: Changing IP ... |
72 |
if (rds_addr_cmp(&conn->c_faddr, &conn->c_laddr) >= 0) { |
1a0e100fb RDS: TCP: Force e... |
73 74 75 76 77 78 |
/* Make sure we initiate at least one path if this * has not already been done; rds_start_mprds() will * take care of additional paths, if necessary. */ if (npaths == 1) rds_conn_path_connect_if_down(&conn->c_path[0]); |
5916e2c15 RDS: TCP: Enable ... |
79 |
return NULL; |
1a0e100fb RDS: TCP: Force e... |
80 |
} |
5916e2c15 RDS: TCP: Enable ... |
81 |
|
117d15bbf RDS: TCP: start m... |
82 |
for (i = 0; i < npaths; i++) { |
5916e2c15 RDS: TCP: Enable ... |
83 84 85 86 87 88 89 90 91 92 93 |
struct rds_conn_path *cp = &conn->c_path[i]; if (rds_conn_path_transition(cp, RDS_CONN_DOWN, RDS_CONN_CONNECTING) || rds_conn_path_transition(cp, RDS_CONN_ERROR, RDS_CONN_CONNECTING)) { return cp->cp_transport_data; } } return NULL; } |
467fa1535 RDS-TCP: Support ... |
94 |
int rds_tcp_accept_one(struct socket *sock) |
70041088e RDS: Add TCP tran... |
95 96 97 98 99 |
{ struct socket *new_sock = NULL; struct rds_connection *conn; int ret; struct inet_sock *inet; |
bd7c5f983 RDS: TCP: Synchro... |
100 101 |
struct rds_tcp_connection *rs_tcp = NULL; int conn_state; |
ea3b1ea53 RDS: TCP: make ->... |
102 |
struct rds_conn_path *cp; |
1e2b44e78 rds: Enable RDS I... |
103 |
struct in6_addr *my_addr, *peer_addr; |
e65d4d963 rds: Remove IPv6 ... |
104 105 106 107 |
#if !IS_ENABLED(CONFIG_IPV6) struct in6_addr saddr, daddr; #endif int dev_if = 0; |
70041088e RDS: Add TCP tran... |
108 |
|
37e14f4fe RDS: TCP: rds_tcp... |
109 110 |
if (!sock) /* module unload or netns delete in progress */ return -ENETUNREACH; |
0933a578c rds: tcp: use soc... |
111 |
ret = sock_create_lite(sock->sk->sk_family, |
d5a8ac28a RDS-TCP: Make RDS... |
112 113 |
sock->sk->sk_type, sock->sk->sk_protocol, &new_sock); |
70041088e RDS: Add TCP tran... |
114 115 |
if (ret) goto out; |
cdfbabfb2 net: Work around ... |
116 |
ret = sock->ops->accept(sock, new_sock, O_NONBLOCK, true); |
70041088e RDS: Add TCP tran... |
117 118 |
if (ret < 0) goto out; |
84eef2b21 rds: Incorrect re... |
119 120 121 122 123 124 125 126 127 |
/* sock_create_lite() does not get a hold on the owner module so we * need to do it here. Note that sock_release() uses sock->ops to * determine if it needs to decrement the reference count. So set * sock->ops after calling accept() in case that fails. And there's * no need to do try_module_get() as the listener should have a hold * already. */ new_sock->ops = sock->ops; __module_get(new_sock->ops->owner); |
480aeb963 tcp: add tcp_sock... |
128 |
rds_tcp_keepalive(new_sock); |
70041088e RDS: Add TCP tran... |
129 130 131 |
rds_tcp_tune(new_sock); inet = inet_sk(new_sock->sk); |
e65d4d963 rds: Remove IPv6 ... |
132 |
#if IS_ENABLED(CONFIG_IPV6) |
1e2b44e78 rds: Enable RDS I... |
133 134 |
my_addr = &new_sock->sk->sk_v6_rcv_saddr; peer_addr = &new_sock->sk->sk_v6_daddr; |
e65d4d963 rds: Remove IPv6 ... |
135 136 137 138 139 140 141 142 143 |
#else ipv6_addr_set_v4mapped(inet->inet_saddr, &saddr); ipv6_addr_set_v4mapped(inet->inet_daddr, &daddr); my_addr = &saddr; peer_addr = &daddr; #endif rdsdebug("accepted family %d tcp %pI6c:%u -> %pI6c:%u ", sock->sk->sk_family, |
1e2b44e78 rds: Enable RDS I... |
144 145 |
my_addr, ntohs(inet->inet_sport), peer_addr, ntohs(inet->inet_dport)); |
70041088e RDS: Add TCP tran... |
146 |
|
e65d4d963 rds: Remove IPv6 ... |
147 |
#if IS_ENABLED(CONFIG_IPV6) |
1e2b44e78 rds: Enable RDS I... |
148 149 150 151 152 153 154 155 156 157 158 159 160 |
/* sk_bound_dev_if is not set if the peer address is not link local * address. In this case, it happens that mcast_oif is set. So * just use it. */ if ((ipv6_addr_type(my_addr) & IPV6_ADDR_LINKLOCAL) && !(ipv6_addr_type(peer_addr) & IPV6_ADDR_LINKLOCAL)) { struct ipv6_pinfo *inet6; inet6 = inet6_sk(new_sock->sk); dev_if = inet6->mcast_oif; } else { dev_if = new_sock->sk->sk_bound_dev_if; } |
e65d4d963 rds: Remove IPv6 ... |
161 |
#endif |
d5a8ac28a RDS-TCP: Make RDS... |
162 |
conn = rds_conn_create(sock_net(sock->sk), |
e65d4d963 rds: Remove IPv6 ... |
163 |
my_addr, peer_addr, |
3eb450367 rds: add type of ... |
164 |
&rds_tcp_transport, 0, GFP_KERNEL, dev_if); |
eee2fa6ab rds: Changing IP ... |
165 |
|
70041088e RDS: Add TCP tran... |
166 167 168 169 |
if (IS_ERR(conn)) { ret = PTR_ERR(conn); goto out; } |
f711a6ae0 net/rds: RDS-TCP:... |
170 |
/* An incoming SYN request came in, and TCP just accepted it. |
f711a6ae0 net/rds: RDS-TCP:... |
171 172 173 174 |
* * If the client reboots, this conn will need to be cleaned up. * rds_tcp_state_change() will do that cleanup */ |
5916e2c15 RDS: TCP: Enable ... |
175 176 177 |
rs_tcp = rds_tcp_accept_one_path(conn); if (!rs_tcp) goto rst_nsk; |
02105b2cc RDS: TCP: Make rd... |
178 |
mutex_lock(&rs_tcp->t_conn_path_lock); |
5916e2c15 RDS: TCP: Enable ... |
179 180 |
cp = rs_tcp->t_cpath; conn_state = rds_conn_path_state(cp); |
1a0e100fb RDS: TCP: Force e... |
181 182 |
WARN_ON(conn_state == RDS_CONN_UP); if (conn_state != RDS_CONN_CONNECTING && conn_state != RDS_CONN_ERROR) |
bd7c5f983 RDS: TCP: Synchro... |
183 |
goto rst_nsk; |
eb1928402 RDS:TCP: Synchron... |
184 |
if (rs_tcp->t_sock) { |
41500c3e2 rds: tcp: remove ... |
185 186 187 188 |
/* Duelling SYN has been handled in rds_tcp_accept_one() */ rds_tcp_reset_callbacks(new_sock, cp); /* rds_connect_path_complete() marks RDS_CONN_UP */ rds_connect_path_complete(cp, RDS_CONN_RESETTING); |
335b48d98 RDS: TCP: Add/use... |
189 |
} else { |
ea3b1ea53 RDS: TCP: make ->... |
190 191 |
rds_tcp_set_callbacks(new_sock, cp); rds_connect_path_complete(cp, RDS_CONN_CONNECTING); |
eb1928402 RDS:TCP: Synchron... |
192 |
} |
70041088e RDS: Add TCP tran... |
193 194 |
new_sock = NULL; ret = 0; |
69b92b5b7 rds: tcp: send ha... |
195 196 |
if (conn->c_npaths == 0) rds_send_ping(cp->cp_conn, cp->cp_index); |
bd7c5f983 RDS: TCP: Synchro... |
197 198 |
goto out; rst_nsk: |
10beea7d7 rds: tcp: Set lin... |
199 200 201 202 203 204 |
/* reset the newly returned accept sock and bail. * It is safe to set linger on new_sock because the RDS connection * has not been brought up on new_sock, so no RDS-level data could * be pending on it. By setting linger, we achieve the side-effect * of avoiding TIME_WAIT state on new_sock. */ |
c433594c0 net: add sock_no_... |
205 |
sock_no_linger(new_sock->sk); |
335b48d98 RDS: TCP: Add/use... |
206 |
kernel_sock_shutdown(new_sock, SHUT_RDWR); |
bd7c5f983 RDS: TCP: Synchro... |
207 |
ret = 0; |
70041088e RDS: Add TCP tran... |
208 |
out: |
bd7c5f983 RDS: TCP: Synchro... |
209 |
if (rs_tcp) |
02105b2cc RDS: TCP: Make rd... |
210 |
mutex_unlock(&rs_tcp->t_conn_path_lock); |
70041088e RDS: Add TCP tran... |
211 212 213 214 |
if (new_sock) sock_release(new_sock); return ret; } |
676d23690 net: Fix use afte... |
215 |
void rds_tcp_listen_data_ready(struct sock *sk) |
70041088e RDS: Add TCP tran... |
216 |
{ |
676d23690 net: Fix use afte... |
217 |
void (*ready)(struct sock *sk); |
70041088e RDS: Add TCP tran... |
218 219 220 |
rdsdebug("listen data ready sk %p ", sk); |
38036629c rds: tcp: block B... |
221 |
read_lock_bh(&sk->sk_callback_lock); |
70041088e RDS: Add TCP tran... |
222 |
ready = sk->sk_user_data; |
8690bfa17 RDS: cleanup: rem... |
223 |
if (!ready) { /* check for teardown race */ |
70041088e RDS: Add TCP tran... |
224 225 226 227 228 229 230 231 232 |
ready = sk->sk_data_ready; goto out; } /* * ->sk_data_ready is also called for a newly established child socket * before it has been accepted and the accepter has set up their * data_ready.. we only want to queue listen work for our listening * socket |
b21dd4506 rds: tcp: Sequenc... |
233 234 235 |
* * (*ready)() may be null if we are racing with netns delete, and * the listen socket is being torn down. |
70041088e RDS: Add TCP tran... |
236 237 |
*/ if (sk->sk_state == TCP_LISTEN) |
467fa1535 RDS-TCP: Support ... |
238 |
rds_tcp_accept_work(sk); |
a93d01f57 RDS: TCP: avoid b... |
239 240 |
else ready = rds_tcp_listen_sock_def_readable(sock_net(sk)); |
70041088e RDS: Add TCP tran... |
241 242 |
out: |
38036629c rds: tcp: block B... |
243 |
read_unlock_bh(&sk->sk_callback_lock); |
b21dd4506 rds: tcp: Sequenc... |
244 245 |
if (ready) ready(sk); |
70041088e RDS: Add TCP tran... |
246 |
} |
1e2b44e78 rds: Enable RDS I... |
247 |
struct socket *rds_tcp_listen_init(struct net *net, bool isv6) |
70041088e RDS: Add TCP tran... |
248 |
{ |
70041088e RDS: Add TCP tran... |
249 |
struct socket *sock = NULL; |
1e2b44e78 rds: Enable RDS I... |
250 251 252 253 |
struct sockaddr_storage ss; struct sockaddr_in6 *sin6; struct sockaddr_in *sin; int addr_len; |
70041088e RDS: Add TCP tran... |
254 |
int ret; |
1e2b44e78 rds: Enable RDS I... |
255 256 257 258 259 260 |
ret = sock_create_kern(net, isv6 ? PF_INET6 : PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock); if (ret < 0) { rdsdebug("could not create %s listener socket: %d ", isv6 ? "IPv6" : "IPv4", ret); |
70041088e RDS: Add TCP tran... |
261 |
goto out; |
1e2b44e78 rds: Enable RDS I... |
262 |
} |
70041088e RDS: Add TCP tran... |
263 |
|
4a17fd522 sock: Introduce n... |
264 |
sock->sk->sk_reuse = SK_CAN_REUSE; |
12abc5ee7 tcp: add tcp_sock... |
265 |
tcp_sock_set_nodelay(sock->sk); |
70041088e RDS: Add TCP tran... |
266 267 268 269 270 |
write_lock_bh(&sock->sk->sk_callback_lock); sock->sk->sk_user_data = sock->sk->sk_data_ready; sock->sk->sk_data_ready = rds_tcp_listen_data_ready; write_unlock_bh(&sock->sk->sk_callback_lock); |
1e2b44e78 rds: Enable RDS I... |
271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 |
if (isv6) { sin6 = (struct sockaddr_in6 *)&ss; sin6->sin6_family = PF_INET6; sin6->sin6_addr = in6addr_any; sin6->sin6_port = (__force u16)htons(RDS_TCP_PORT); sin6->sin6_scope_id = 0; sin6->sin6_flowinfo = 0; addr_len = sizeof(*sin6); } else { sin = (struct sockaddr_in *)&ss; sin->sin_family = PF_INET; sin->sin_addr.s_addr = INADDR_ANY; sin->sin_port = (__force u16)htons(RDS_TCP_PORT); addr_len = sizeof(*sin); } |
70041088e RDS: Add TCP tran... |
286 |
|
1e2b44e78 rds: Enable RDS I... |
287 288 289 290 291 |
ret = sock->ops->bind(sock, (struct sockaddr *)&ss, addr_len); if (ret < 0) { rdsdebug("could not bind %s listener socket: %d ", isv6 ? "IPv6" : "IPv4", ret); |
70041088e RDS: Add TCP tran... |
292 |
goto out; |
1e2b44e78 rds: Enable RDS I... |
293 |
} |
70041088e RDS: Add TCP tran... |
294 295 296 297 |
ret = sock->ops->listen(sock, 64); if (ret < 0) goto out; |
467fa1535 RDS-TCP: Support ... |
298 |
return sock; |
70041088e RDS: Add TCP tran... |
299 300 301 |
out: if (sock) sock_release(sock); |
467fa1535 RDS-TCP: Support ... |
302 |
return NULL; |
70041088e RDS: Add TCP tran... |
303 |
} |
b21dd4506 rds: tcp: Sequenc... |
304 |
void rds_tcp_listen_stop(struct socket *sock, struct work_struct *acceptor) |
70041088e RDS: Add TCP tran... |
305 |
{ |
70041088e RDS: Add TCP tran... |
306 |
struct sock *sk; |
8690bfa17 RDS: cleanup: rem... |
307 |
if (!sock) |
70041088e RDS: Add TCP tran... |
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 |
return; sk = sock->sk; /* serialize with and prevent further callbacks */ lock_sock(sk); write_lock_bh(&sk->sk_callback_lock); if (sk->sk_user_data) { sk->sk_data_ready = sk->sk_user_data; sk->sk_user_data = NULL; } write_unlock_bh(&sk->sk_callback_lock); release_sock(sk); /* wait for accepts to stop and close the socket */ flush_workqueue(rds_wq); |
b21dd4506 rds: tcp: Sequenc... |
324 |
flush_work(acceptor); |
70041088e RDS: Add TCP tran... |
325 |
sock_release(sock); |
70041088e RDS: Add TCP tran... |
326 |
} |