Commit | Line | Data |
---|---|---|
b97bf3fd PL |
1 | /* |
2 | * net/tipc/net.c: TIPC network routing code | |
3 | * | |
4 | * Copyright (c) 2003-2005, Ericsson Research Canada | |
5 | * Copyright (c) 2005, Wind River Systems | |
6 | * Copyright (c) 2005-2006, Ericsson AB | |
7 | * All rights reserved. | |
8 | * | |
9 | * Redistribution and use in source and binary forms, with or without | |
10 | * modification, are permitted provided that the following conditions are met: | |
11 | * | |
12 | * Redistributions of source code must retain the above copyright notice, this | |
13 | * list of conditions and the following disclaimer. | |
14 | * Redistributions in binary form must reproduce the above copyright notice, | |
15 | * this list of conditions and the following disclaimer in the documentation | |
16 | * and/or other materials provided with the distribution. | |
17 | * Neither the names of the copyright holders nor the names of its | |
18 | * contributors may be used to endorse or promote products derived from this | |
19 | * software without specific prior written permission. | |
20 | * | |
21 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" | |
22 | * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE | |
23 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE | |
24 | * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE | |
25 | * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR | |
26 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF | |
27 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS | |
28 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN | |
29 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) | |
30 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE | |
31 | * POSSIBILITY OF SUCH DAMAGE. | |
32 | */ | |
33 | ||
34 | #include "core.h" | |
35 | #include "bearer.h" | |
36 | #include "net.h" | |
37 | #include "zone.h" | |
38 | #include "addr.h" | |
39 | #include "name_table.h" | |
40 | #include "name_distr.h" | |
41 | #include "subscr.h" | |
42 | #include "link.h" | |
43 | #include "msg.h" | |
44 | #include "port.h" | |
45 | #include "bcast.h" | |
46 | #include "discover.h" | |
47 | #include "config.h" | |
48 | ||
49 | /* | |
50 | * The TIPC locking policy is designed to ensure a very fine locking | |
51 | * granularity, permitting complete parallel access to individual | |
52 | * port and node/link instances. The code consists of three major | |
53 | * locking domains, each protected with their own disjunct set of locks. | |
54 | * | |
55 | * 1: The routing hierarchy. | |
56 | * Comprises the structures 'zone', 'cluster', 'node', 'link' | |
57 | * and 'bearer'. The whole hierarchy is protected by a big | |
58 | * read/write lock, net_lock, to enssure that nothing is added | |
59 | * or removed while code is accessing any of these structures. | |
60 | * This layer must not be called from the two others while they | |
61 | * hold any of their own locks. | |
62 | * Neither must it itself do any upcalls to the other two before | |
63 | * it has released net_lock and other protective locks. | |
64 | * | |
65 | * Within the net_lock domain there are two sub-domains;'node' and | |
66 | * 'bearer', where local write operations are permitted, | |
67 | * provided that those are protected by individual spin_locks | |
68 | * per instance. Code holding net_lock(read) and a node spin_lock | |
69 | * is permitted to poke around in both the node itself and its | |
70 | * subordinate links. I.e, it can update link counters and queues, | |
71 | * change link state, send protocol messages, and alter the | |
72 | * "active_links" array in the node; but it can _not_ remove a link | |
73 | * or a node from the overall structure. | |
74 | * Correspondingly, individual bearers may change status within a | |
75 | * net_lock(read), protected by an individual spin_lock ber bearer | |
76 | * instance, but it needs net_lock(write) to remove/add any bearers. | |
77 | * | |
78 | * | |
79 | * 2: The transport level of the protocol. | |
80 | * This consists of the structures port, (and its user level | |
81 | * representations, such as user_port and tipc_sock), reference and | |
82 | * tipc_user (port.c, reg.c, socket.c). | |
83 | * | |
84 | * This layer has four different locks: | |
85 | * - The tipc_port spin_lock. This is protecting each port instance | |
86 | * from parallel data access and removal. Since we can not place | |
87 | * this lock in the port itself, it has been placed in the | |
88 | * corresponding reference table entry, which has the same life | |
89 | * cycle as the module. This entry is difficult to access from | |
90 | * outside the TIPC core, however, so a pointer to the lock has | |
91 | * been added in the port instance, -to be used for unlocking | |
92 | * only. | |
93 | * - A read/write lock to protect the reference table itself (teg.c). | |
94 | * (Nobody is using read-only access to this, so it can just as | |
95 | * well be changed to a spin_lock) | |
96 | * - A spin lock to protect the registry of kernel/driver users (reg.c) | |
97 | * - A global spin_lock (port_lock), which only task is to ensure | |
98 | * consistency where more than one port is involved in an operation, | |
99 | * i.e., whe a port is part of a linked list of ports. | |
100 | * There are two such lists; 'port_list', which is used for management, | |
101 | * and 'wait_list', which is used to queue ports during congestion. | |
102 | * | |
103 | * 3: The name table (name_table.c, name_distr.c, subscription.c) | |
104 | * - There is one big read/write-lock (nametbl_lock) protecting the | |
105 | * overall name table structure. Nothing must be added/removed to | |
106 | * this structure without holding write access to it. | |
107 | * - There is one local spin_lock per sub_sequence, which can be seen | |
108 | * as a sub-domain to the nametbl_lock domain. It is used only | |
109 | * for translation operations, and is needed because a translation | |
110 | * steps the root of the 'publication' linked list between each lookup. | |
111 | * This is always used within the scope of a nametbl_lock(read). | |
112 | * - A local spin_lock protecting the queue of subscriber events. | |
113 | */ | |
114 | ||
115 | rwlock_t net_lock = RW_LOCK_UNLOCKED; | |
116 | struct network net = { 0 }; | |
117 | ||
118 | struct node *net_select_remote_node(u32 addr, u32 ref) | |
119 | { | |
120 | return zone_select_remote_node(net.zones[tipc_zone(addr)], addr, ref); | |
121 | } | |
122 | ||
123 | u32 net_select_router(u32 addr, u32 ref) | |
124 | { | |
125 | return zone_select_router(net.zones[tipc_zone(addr)], addr, ref); | |
126 | } | |
127 | ||
128 | ||
129 | u32 net_next_node(u32 a) | |
130 | { | |
131 | if (net.zones[tipc_zone(a)]) | |
132 | return zone_next_node(a); | |
133 | return 0; | |
134 | } | |
135 | ||
136 | void net_remove_as_router(u32 router) | |
137 | { | |
138 | u32 z_num; | |
139 | ||
140 | for (z_num = 1; z_num <= tipc_max_zones; z_num++) { | |
141 | if (!net.zones[z_num]) | |
142 | continue; | |
143 | zone_remove_as_router(net.zones[z_num], router); | |
144 | } | |
145 | } | |
146 | ||
147 | void net_send_external_routes(u32 dest) | |
148 | { | |
149 | u32 z_num; | |
150 | ||
151 | for (z_num = 1; z_num <= tipc_max_zones; z_num++) { | |
152 | if (net.zones[z_num]) | |
153 | zone_send_external_routes(net.zones[z_num], dest); | |
154 | } | |
155 | } | |
156 | ||
157 | int net_init(void) | |
158 | { | |
159 | u32 sz = sizeof(struct _zone *) * (tipc_max_zones + 1); | |
160 | ||
161 | memset(&net, 0, sizeof(net)); | |
162 | net.zones = (struct _zone **)kmalloc(sz, GFP_ATOMIC); | |
163 | if (!net.zones) { | |
164 | return -ENOMEM; | |
165 | } | |
166 | memset(net.zones, 0, sz); | |
167 | return TIPC_OK; | |
168 | } | |
169 | ||
170 | void net_stop(void) | |
171 | { | |
172 | u32 z_num; | |
173 | ||
174 | if (!net.zones) | |
175 | return; | |
176 | ||
177 | for (z_num = 1; z_num <= tipc_max_zones; z_num++) { | |
178 | zone_delete(net.zones[z_num]); | |
179 | } | |
180 | kfree(net.zones); | |
181 | net.zones = 0; | |
182 | } | |
183 | ||
184 | static void net_route_named_msg(struct sk_buff *buf) | |
185 | { | |
186 | struct tipc_msg *msg = buf_msg(buf); | |
187 | u32 dnode; | |
188 | u32 dport; | |
189 | ||
190 | if (!msg_named(msg)) { | |
191 | msg_dbg(msg, "net->drop_nam:"); | |
192 | buf_discard(buf); | |
193 | return; | |
194 | } | |
195 | ||
196 | dnode = addr_domain(msg_lookup_scope(msg)); | |
197 | dport = nametbl_translate(msg_nametype(msg), msg_nameinst(msg), &dnode); | |
198 | dbg("net->lookup<%u,%u>-><%u,%x>\n", | |
199 | msg_nametype(msg), msg_nameinst(msg), dport, dnode); | |
200 | if (dport) { | |
201 | msg_set_destnode(msg, dnode); | |
202 | msg_set_destport(msg, dport); | |
203 | net_route_msg(buf); | |
204 | return; | |
205 | } | |
206 | msg_dbg(msg, "net->rej:NO NAME: "); | |
207 | tipc_reject_msg(buf, TIPC_ERR_NO_NAME); | |
208 | } | |
209 | ||
210 | void net_route_msg(struct sk_buff *buf) | |
211 | { | |
212 | struct tipc_msg *msg; | |
213 | u32 dnode; | |
214 | ||
215 | if (!buf) | |
216 | return; | |
217 | msg = buf_msg(buf); | |
218 | ||
219 | msg_incr_reroute_cnt(msg); | |
220 | if (msg_reroute_cnt(msg) > 6) { | |
221 | if (msg_errcode(msg)) { | |
222 | msg_dbg(msg, "NET>DISC>:"); | |
223 | buf_discard(buf); | |
224 | } else { | |
225 | msg_dbg(msg, "NET>REJ>:"); | |
226 | tipc_reject_msg(buf, msg_destport(msg) ? | |
227 | TIPC_ERR_NO_PORT : TIPC_ERR_NO_NAME); | |
228 | } | |
229 | return; | |
230 | } | |
231 | ||
232 | msg_dbg(msg, "net->rout: "); | |
233 | ||
234 | /* Handle message for this node */ | |
235 | dnode = msg_short(msg) ? tipc_own_addr : msg_destnode(msg); | |
236 | if (in_scope(dnode, tipc_own_addr)) { | |
237 | if (msg_isdata(msg)) { | |
238 | if (msg_mcast(msg)) | |
239 | port_recv_mcast(buf, NULL); | |
240 | else if (msg_destport(msg)) | |
241 | port_recv_msg(buf); | |
242 | else | |
243 | net_route_named_msg(buf); | |
244 | return; | |
245 | } | |
246 | switch (msg_user(msg)) { | |
247 | case ROUTE_DISTRIBUTOR: | |
248 | cluster_recv_routing_table(buf); | |
249 | break; | |
250 | case NAME_DISTRIBUTOR: | |
251 | named_recv(buf); | |
252 | break; | |
253 | case CONN_MANAGER: | |
254 | port_recv_proto_msg(buf); | |
255 | break; | |
256 | default: | |
257 | msg_dbg(msg,"DROP/NET/<REC<"); | |
258 | buf_discard(buf); | |
259 | } | |
260 | return; | |
261 | } | |
262 | ||
263 | /* Handle message for another node */ | |
264 | msg_dbg(msg, "NET>SEND>: "); | |
265 | link_send(buf, dnode, msg_link_selector(msg)); | |
266 | } | |
267 | ||
268 | int tipc_start_net(void) | |
269 | { | |
270 | char addr_string[16]; | |
271 | int res; | |
272 | ||
273 | if (tipc_mode != TIPC_NODE_MODE) | |
274 | return -ENOPROTOOPT; | |
275 | ||
276 | tipc_mode = TIPC_NET_MODE; | |
277 | named_reinit(); | |
278 | port_reinit(); | |
279 | ||
280 | if ((res = bearer_init()) || | |
281 | (res = net_init()) || | |
282 | (res = cluster_init()) || | |
283 | (res = bclink_init())) { | |
284 | return res; | |
285 | } | |
286 | subscr_stop(); | |
287 | cfg_stop(); | |
288 | k_signal((Handler)subscr_start, 0); | |
289 | k_signal((Handler)cfg_init, 0); | |
290 | info("Started in network mode\n"); | |
291 | info("Own node address %s, network identity %u\n", | |
292 | addr_string_fill(addr_string, tipc_own_addr), tipc_net_id); | |
293 | return TIPC_OK; | |
294 | } | |
295 | ||
296 | void tipc_stop_net(void) | |
297 | { | |
298 | if (tipc_mode != TIPC_NET_MODE) | |
299 | return; | |
300 | write_lock_bh(&net_lock); | |
301 | bearer_stop(); | |
302 | tipc_mode = TIPC_NODE_MODE; | |
303 | bclink_stop(); | |
304 | net_stop(); | |
305 | write_unlock_bh(&net_lock); | |
306 | info("Left network mode \n"); | |
307 | } | |
308 |