Commit | Line | Data |
---|---|---|
2522fe45 | 1 | // SPDX-License-Identifier: GPL-2.0-only |
e7fd4179 DT |
2 | /****************************************************************************** |
3 | ******************************************************************************* | |
4 | ** | |
5 | ** Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved. | |
489d8e55 | 6 | ** Copyright (C) 2004-2021 Red Hat, Inc. All rights reserved. |
e7fd4179 | 7 | ** |
e7fd4179 DT |
8 | ** |
9 | ******************************************************************************* | |
10 | ******************************************************************************/ | |
11 | ||
12 | /* | |
13 | * midcomms.c | |
14 | * | |
489d8e55 AA |
15 | * This is the appallingly named "mid-level" comms layer. It takes care about |
16 | * deliver an on application layer "reliable" communication above the used | |
17 | * lowcomms transport layer. | |
e7fd4179 | 18 | * |
489d8e55 | 19 | * How it works: |
e7fd4179 | 20 | * |
489d8e55 AA |
21 | * Each nodes keeps track of all send DLM messages in send_queue with a sequence |
22 | * number. The receive will send an DLM_ACK message back for every DLM message | |
23 | * received at the other side. If a reconnect happens in lowcomms we will send | |
24 | * all unacknowledged dlm messages again. The receiving side might drop any already | |
25 | * received message by comparing sequence numbers. | |
26 | * | |
27 | * How version detection works: | |
28 | * | |
29 | * Due the fact that dlm has pre-configured node addresses on every side | |
30 | * it is in it's nature that every side connects at starts to transmit | |
31 | * dlm messages which ends in a race. However DLM_RCOM_NAMES, DLM_RCOM_STATUS | |
32 | * and their replies are the first messages which are exchanges. Due backwards | |
33 | * compatibility these messages are not covered by the midcomms re-transmission | |
34 | * layer. These messages have their own re-transmission handling in the dlm | |
35 | * application layer. The version field of every node will be set on these RCOM | |
36 | * messages as soon as they arrived and the node isn't yet part of the nodes | |
37 | * hash. There exists also logic to detect version mismatched if something weird | |
38 | * going on or the first messages isn't an expected one. | |
39 | * | |
40 | * Termination: | |
41 | * | |
42 | * The midcomms layer does a 4 way handshake for termination on DLM protocol | |
43 | * like TCP supports it with half-closed socket support. SCTP doesn't support | |
44 | * half-closed socket, so we do it on DLM layer. Also socket shutdown() can be | |
45 | * interrupted by .e.g. tcp reset itself. Additional there exists the othercon | |
46 | * paradigm in lowcomms which cannot be easily without breaking backwards | |
47 | * compatibility. A node cannot send anything to another node when a DLM_FIN | |
48 | * message was send. There exists additional logic to print a warning if | |
49 | * DLM wants to do it. There exists a state handling like RFC 793 but reduced | |
50 | * to termination only. The event "member removal event" describes the cluster | |
51 | * manager removed the node from internal lists, at this point DLM does not | |
52 | * send any message to the other node. There exists two cases: | |
53 | * | |
54 | * 1. The cluster member was removed and we received a FIN | |
55 | * OR | |
56 | * 2. We received a FIN but the member was not removed yet | |
57 | * | |
58 | * One of these cases will do the CLOSE_WAIT to LAST_ACK change. | |
59 | * | |
60 | * | |
61 | * +---------+ | |
62 | * | CLOSED | | |
63 | * +---------+ | |
64 | * | add member/receive RCOM version | |
65 | * | detection msg | |
66 | * V | |
67 | * +---------+ | |
68 | * | ESTAB | | |
69 | * +---------+ | |
70 | * CLOSE | | rcv FIN | |
71 | * ------- | | ------- | |
72 | * +---------+ snd FIN / \ snd ACK +---------+ | |
73 | * | FIN |<----------------- ------------------>| CLOSE | | |
74 | * | WAIT-1 |------------------ | WAIT | | |
75 | * +---------+ rcv FIN \ +---------+ | |
76 | * | rcv ACK of FIN ------- | CLOSE | member | |
77 | * | -------------- snd ACK | ------- | removal | |
78 | * V x V snd FIN V event | |
79 | * +---------+ +---------+ +---------+ | |
80 | * |FINWAIT-2| | CLOSING | | LAST-ACK| | |
81 | * +---------+ +---------+ +---------+ | |
82 | * | rcv ACK of FIN | rcv ACK of FIN | | |
83 | * | rcv FIN -------------- | -------------- | | |
84 | * | ------- x V x V | |
85 | * \ snd ACK +---------+ +---------+ | |
86 | * ------------------------>| CLOSED | | CLOSED | | |
87 | * +---------+ +---------+ | |
88 | * | |
89 | * NOTE: any state can interrupted by midcomms_close() and state will be | |
90 | * switched to CLOSED in case of fencing. There exists also some timeout | |
91 | * handling when we receive the version detection RCOM messages which is | |
92 | * made by observation. | |
93 | * | |
94 | * Future improvements: | |
95 | * | |
96 | * There exists some known issues/improvements of the dlm handling. Some | |
97 | * of them should be done in a next major dlm version bump which makes | |
98 | * it incompatible with previous versions. | |
99 | * | |
100 | * Unaligned memory access: | |
101 | * | |
102 | * There exists cases when the dlm message buffer length is not aligned | |
103 | * to 8 byte. However seems nobody detected any problem with it. This | |
104 | * can be fixed in the next major version bump of dlm. | |
105 | * | |
106 | * Version detection: | |
107 | * | |
108 | * The version detection and how it's done is related to backwards | |
109 | * compatibility. There exists better ways to make a better handling. | |
110 | * However this should be changed in the next major version bump of dlm. | |
111 | * | |
489d8e55 AA |
112 | * Tail Size checking: |
113 | * | |
114 | * There exists a message tail payload in e.g. DLM_MSG however we don't | |
115 | * check it against the message length yet regarding to the receive buffer | |
116 | * length. That need to be validated. | |
117 | * | |
118 | * Fencing bad nodes: | |
119 | * | |
120 | * At timeout places or weird sequence number behaviours we should send | |
121 | * a fencing request to the cluster manager. | |
e7fd4179 DT |
122 | */ |
123 | ||
489d8e55 AA |
124 | /* Debug switch to enable a 5 seconds sleep waiting of a termination. |
125 | * This can be useful to test fencing while termination is running. | |
126 | * This requires a setup with only gfs2 as dlm user, so that the | |
127 | * last umount will terminate the connection. | |
128 | * | |
129 | * However it became useful to test, while the 5 seconds block in umount | |
130 | * just press the reset button. In a lot of dropping the termination | |
131 | * process can could take several seconds. | |
132 | */ | |
133 | #define DLM_DEBUG_FENCE_TERMINATION 0 | |
134 | ||
135 | #include <net/tcp.h> | |
136 | ||
e7fd4179 DT |
137 | #include "dlm_internal.h" |
138 | #include "lowcomms.h" | |
139 | #include "config.h" | |
e7fd4179 | 140 | #include "lock.h" |
489d8e55 | 141 | #include "util.h" |
e7fd4179 DT |
142 | #include "midcomms.h" |
143 | ||
489d8e55 AA |
144 | /* init value for sequence numbers for testing purpose only e.g. overflows */ |
145 | #define DLM_SEQ_INIT 0 | |
146 | /* 3 minutes wait to sync ending of dlm */ | |
147 | #define DLM_SHUTDOWN_TIMEOUT msecs_to_jiffies(3 * 60 * 1000) | |
148 | #define DLM_VERSION_NOT_SET 0 | |
149 | ||
150 | struct midcomms_node { | |
151 | int nodeid; | |
152 | uint32_t version; | |
153 | uint32_t seq_send; | |
154 | uint32_t seq_next; | |
155 | /* These queues are unbound because we cannot drop any message in dlm. | |
156 | * We could send a fence signal for a specific node to the cluster | |
157 | * manager if queues hits some maximum value, however this handling | |
158 | * not supported yet. | |
159 | */ | |
160 | struct list_head send_queue; | |
161 | spinlock_t send_queue_lock; | |
162 | atomic_t send_queue_cnt; | |
163 | #define DLM_NODE_FLAG_CLOSE 1 | |
164 | #define DLM_NODE_FLAG_STOP_TX 2 | |
165 | #define DLM_NODE_FLAG_STOP_RX 3 | |
b97f8525 | 166 | #define DLM_NODE_ULP_DELIVERED 4 |
489d8e55 AA |
167 | unsigned long flags; |
168 | wait_queue_head_t shutdown_wait; | |
169 | ||
170 | /* dlm tcp termination state */ | |
171 | #define DLM_CLOSED 1 | |
172 | #define DLM_ESTABLISHED 2 | |
173 | #define DLM_FIN_WAIT1 3 | |
174 | #define DLM_FIN_WAIT2 4 | |
175 | #define DLM_CLOSE_WAIT 5 | |
176 | #define DLM_LAST_ACK 6 | |
177 | #define DLM_CLOSING 7 | |
178 | int state; | |
179 | spinlock_t state_lock; | |
180 | ||
181 | /* counts how many lockspaces are using this node | |
182 | * this refcount is necessary to determine if the | |
183 | * node wants to disconnect. | |
184 | */ | |
185 | int users; | |
186 | ||
5b2f981f AA |
187 | /* not protected by srcu, node_hash lifetime */ |
188 | void *debugfs; | |
189 | ||
489d8e55 AA |
190 | struct hlist_node hlist; |
191 | struct rcu_head rcu; | |
192 | }; | |
193 | ||
194 | struct dlm_mhandle { | |
195 | const struct dlm_header *inner_hd; | |
196 | struct midcomms_node *node; | |
197 | struct dlm_opts *opts; | |
198 | struct dlm_msg *msg; | |
199 | bool committed; | |
200 | uint32_t seq; | |
201 | ||
202 | void (*ack_rcv)(struct midcomms_node *node); | |
203 | ||
204 | /* get_mhandle/commit srcu idx exchange */ | |
205 | int idx; | |
206 | ||
207 | struct list_head list; | |
208 | struct rcu_head rcu; | |
209 | }; | |
210 | ||
211 | static struct hlist_head node_hash[CONN_HASH_SIZE]; | |
212 | static DEFINE_SPINLOCK(nodes_lock); | |
213 | DEFINE_STATIC_SRCU(nodes_srcu); | |
214 | ||
215 | /* This mutex prevents that midcomms_close() is running while | |
216 | * stop() or remove(). As I experienced invalid memory access | |
217 | * behaviours when DLM_DEBUG_FENCE_TERMINATION is enabled and | |
218 | * resetting machines. I will end in some double deletion in nodes | |
219 | * datastructure. | |
220 | */ | |
221 | static DEFINE_MUTEX(close_lock); | |
222 | ||
223 | static inline const char *dlm_state_str(int state) | |
a070a91c | 224 | { |
489d8e55 AA |
225 | switch (state) { |
226 | case DLM_CLOSED: | |
227 | return "CLOSED"; | |
228 | case DLM_ESTABLISHED: | |
229 | return "ESTABLISHED"; | |
230 | case DLM_FIN_WAIT1: | |
231 | return "FIN_WAIT1"; | |
232 | case DLM_FIN_WAIT2: | |
233 | return "FIN_WAIT2"; | |
234 | case DLM_CLOSE_WAIT: | |
235 | return "CLOSE_WAIT"; | |
236 | case DLM_LAST_ACK: | |
237 | return "LAST_ACK"; | |
238 | case DLM_CLOSING: | |
239 | return "CLOSING"; | |
240 | default: | |
241 | return "UNKNOWN"; | |
242 | } | |
a070a91c AA |
243 | } |
244 | ||
5b2f981f AA |
245 | const char *dlm_midcomms_state(struct midcomms_node *node) |
246 | { | |
247 | return dlm_state_str(node->state); | |
248 | } | |
249 | ||
250 | unsigned long dlm_midcomms_flags(struct midcomms_node *node) | |
251 | { | |
252 | return node->flags; | |
253 | } | |
254 | ||
255 | int dlm_midcomms_send_queue_cnt(struct midcomms_node *node) | |
256 | { | |
257 | return atomic_read(&node->send_queue_cnt); | |
258 | } | |
259 | ||
260 | uint32_t dlm_midcomms_version(struct midcomms_node *node) | |
261 | { | |
262 | return node->version; | |
263 | } | |
264 | ||
489d8e55 | 265 | static struct midcomms_node *__find_node(int nodeid, int r) |
a070a91c | 266 | { |
489d8e55 AA |
267 | struct midcomms_node *node; |
268 | ||
269 | hlist_for_each_entry_rcu(node, &node_hash[r], hlist) { | |
270 | if (node->nodeid == nodeid) | |
271 | return node; | |
272 | } | |
273 | ||
274 | return NULL; | |
a070a91c AA |
275 | } |
276 | ||
489d8e55 AA |
277 | static void dlm_mhandle_release(struct rcu_head *rcu) |
278 | { | |
279 | struct dlm_mhandle *mh = container_of(rcu, struct dlm_mhandle, rcu); | |
a070a91c | 280 | |
489d8e55 AA |
281 | dlm_lowcomms_put_msg(mh->msg); |
282 | kfree(mh); | |
283 | } | |
a070a91c | 284 | |
f5fe8d51 AA |
285 | static void dlm_mhandle_delete(struct midcomms_node *node, |
286 | struct dlm_mhandle *mh) | |
287 | { | |
288 | list_del_rcu(&mh->list); | |
289 | atomic_dec(&node->send_queue_cnt); | |
290 | call_rcu(&mh->rcu, dlm_mhandle_release); | |
291 | } | |
292 | ||
489d8e55 | 293 | static void dlm_send_queue_flush(struct midcomms_node *node) |
a070a91c | 294 | { |
489d8e55 AA |
295 | struct dlm_mhandle *mh; |
296 | ||
297 | pr_debug("flush midcomms send queue of node %d\n", node->nodeid); | |
298 | ||
299 | rcu_read_lock(); | |
f5fe8d51 | 300 | spin_lock(&node->send_queue_lock); |
489d8e55 | 301 | list_for_each_entry_rcu(mh, &node->send_queue, list) { |
f5fe8d51 | 302 | dlm_mhandle_delete(node, mh); |
489d8e55 | 303 | } |
f5fe8d51 | 304 | spin_unlock(&node->send_queue_lock); |
489d8e55 | 305 | rcu_read_unlock(); |
a070a91c AA |
306 | } |
307 | ||
489d8e55 | 308 | static void midcomms_node_reset(struct midcomms_node *node) |
a070a91c | 309 | { |
489d8e55 AA |
310 | pr_debug("reset node %d\n", node->nodeid); |
311 | ||
312 | node->seq_next = DLM_SEQ_INIT; | |
313 | node->seq_send = DLM_SEQ_INIT; | |
314 | node->version = DLM_VERSION_NOT_SET; | |
315 | node->flags = 0; | |
316 | ||
317 | dlm_send_queue_flush(node); | |
318 | node->state = DLM_CLOSED; | |
319 | wake_up(&node->shutdown_wait); | |
a070a91c AA |
320 | } |
321 | ||
489d8e55 AA |
322 | static struct midcomms_node *nodeid2node(int nodeid, gfp_t alloc) |
323 | { | |
324 | struct midcomms_node *node, *tmp; | |
325 | int r = nodeid_hash(nodeid); | |
326 | ||
327 | node = __find_node(nodeid, r); | |
328 | if (node || !alloc) | |
329 | return node; | |
330 | ||
331 | node = kmalloc(sizeof(*node), alloc); | |
332 | if (!node) | |
333 | return NULL; | |
334 | ||
335 | node->nodeid = nodeid; | |
336 | spin_lock_init(&node->state_lock); | |
337 | spin_lock_init(&node->send_queue_lock); | |
338 | atomic_set(&node->send_queue_cnt, 0); | |
339 | INIT_LIST_HEAD(&node->send_queue); | |
340 | init_waitqueue_head(&node->shutdown_wait); | |
341 | node->users = 0; | |
342 | midcomms_node_reset(node); | |
343 | ||
344 | spin_lock(&nodes_lock); | |
345 | /* check again if there was somebody else | |
346 | * earlier here to add the node | |
347 | */ | |
348 | tmp = __find_node(nodeid, r); | |
349 | if (tmp) { | |
350 | spin_unlock(&nodes_lock); | |
351 | kfree(node); | |
352 | return tmp; | |
353 | } | |
354 | ||
355 | hlist_add_head_rcu(&node->hlist, &node_hash[r]); | |
356 | spin_unlock(&nodes_lock); | |
5b2f981f AA |
357 | |
358 | node->debugfs = dlm_create_debug_comms_file(nodeid, node); | |
489d8e55 AA |
359 | return node; |
360 | } | |
361 | ||
362 | static int dlm_send_ack(int nodeid, uint32_t seq) | |
363 | { | |
364 | int mb_len = sizeof(struct dlm_header); | |
365 | struct dlm_header *m_header; | |
366 | struct dlm_msg *msg; | |
367 | char *ppc; | |
368 | ||
369 | msg = dlm_lowcomms_new_msg(nodeid, mb_len, GFP_NOFS, &ppc, | |
370 | NULL, NULL); | |
371 | if (!msg) | |
372 | return -ENOMEM; | |
373 | ||
374 | m_header = (struct dlm_header *)ppc; | |
375 | ||
376 | m_header->h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR); | |
377 | m_header->h_nodeid = dlm_our_nodeid(); | |
378 | m_header->h_length = mb_len; | |
379 | m_header->h_cmd = DLM_ACK; | |
380 | m_header->u.h_seq = seq; | |
381 | ||
382 | header_out(m_header); | |
383 | dlm_lowcomms_commit_msg(msg); | |
384 | dlm_lowcomms_put_msg(msg); | |
385 | ||
386 | return 0; | |
387 | } | |
388 | ||
389 | static int dlm_send_fin(struct midcomms_node *node, | |
390 | void (*ack_rcv)(struct midcomms_node *node)) | |
391 | { | |
392 | int mb_len = sizeof(struct dlm_header); | |
393 | struct dlm_header *m_header; | |
394 | struct dlm_mhandle *mh; | |
395 | char *ppc; | |
396 | ||
397 | mh = dlm_midcomms_get_mhandle(node->nodeid, mb_len, GFP_NOFS, &ppc); | |
398 | if (!mh) | |
399 | return -ENOMEM; | |
400 | ||
401 | mh->ack_rcv = ack_rcv; | |
402 | ||
403 | m_header = (struct dlm_header *)ppc; | |
404 | ||
405 | m_header->h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR); | |
406 | m_header->h_nodeid = dlm_our_nodeid(); | |
407 | m_header->h_length = mb_len; | |
408 | m_header->h_cmd = DLM_FIN; | |
409 | ||
410 | header_out(m_header); | |
411 | ||
412 | pr_debug("sending fin msg to node %d\n", node->nodeid); | |
413 | dlm_midcomms_commit_mhandle(mh); | |
414 | set_bit(DLM_NODE_FLAG_STOP_TX, &node->flags); | |
415 | ||
416 | return 0; | |
417 | } | |
418 | ||
419 | static void dlm_receive_ack(struct midcomms_node *node, uint32_t seq) | |
420 | { | |
421 | struct dlm_mhandle *mh; | |
422 | ||
423 | rcu_read_lock(); | |
424 | list_for_each_entry_rcu(mh, &node->send_queue, list) { | |
425 | if (before(mh->seq, seq)) { | |
489d8e55 AA |
426 | if (mh->ack_rcv) |
427 | mh->ack_rcv(node); | |
f5fe8d51 AA |
428 | } else { |
429 | /* send queue should be ordered */ | |
430 | break; | |
431 | } | |
432 | } | |
489d8e55 | 433 | |
f5fe8d51 AA |
434 | spin_lock(&node->send_queue_lock); |
435 | list_for_each_entry_rcu(mh, &node->send_queue, list) { | |
436 | if (before(mh->seq, seq)) { | |
437 | dlm_mhandle_delete(node, mh); | |
489d8e55 AA |
438 | } else { |
439 | /* send queue should be ordered */ | |
440 | break; | |
441 | } | |
442 | } | |
f5fe8d51 | 443 | spin_unlock(&node->send_queue_lock); |
489d8e55 AA |
444 | rcu_read_unlock(); |
445 | } | |
446 | ||
447 | static void dlm_pas_fin_ack_rcv(struct midcomms_node *node) | |
448 | { | |
449 | spin_lock(&node->state_lock); | |
450 | pr_debug("receive passive fin ack from node %d with state %s\n", | |
451 | node->nodeid, dlm_state_str(node->state)); | |
452 | ||
453 | switch (node->state) { | |
454 | case DLM_LAST_ACK: | |
455 | /* DLM_CLOSED */ | |
456 | midcomms_node_reset(node); | |
457 | break; | |
458 | case DLM_CLOSED: | |
459 | /* not valid but somehow we got what we want */ | |
460 | wake_up(&node->shutdown_wait); | |
461 | break; | |
462 | default: | |
463 | spin_unlock(&node->state_lock); | |
464 | log_print("%s: unexpected state: %d\n", | |
465 | __func__, node->state); | |
466 | WARN_ON(1); | |
467 | return; | |
468 | } | |
469 | spin_unlock(&node->state_lock); | |
470 | } | |
471 | ||
472 | static void dlm_midcomms_receive_buffer(union dlm_packet *p, | |
473 | struct midcomms_node *node, | |
474 | uint32_t seq) | |
475 | { | |
476 | if (seq == node->seq_next) { | |
477 | node->seq_next++; | |
489d8e55 AA |
478 | |
479 | switch (p->header.h_cmd) { | |
480 | case DLM_FIN: | |
b97f8525 AA |
481 | /* send ack before fin */ |
482 | dlm_send_ack(node->nodeid, node->seq_next); | |
483 | ||
489d8e55 AA |
484 | spin_lock(&node->state_lock); |
485 | pr_debug("receive fin msg from node %d with state %s\n", | |
486 | node->nodeid, dlm_state_str(node->state)); | |
487 | ||
488 | switch (node->state) { | |
489 | case DLM_ESTABLISHED: | |
490 | node->state = DLM_CLOSE_WAIT; | |
491 | pr_debug("switch node %d to state %s\n", | |
492 | node->nodeid, dlm_state_str(node->state)); | |
493 | /* passive shutdown DLM_LAST_ACK case 1 | |
494 | * additional we check if the node is used by | |
495 | * cluster manager events at all. | |
496 | */ | |
497 | if (node->users == 0) { | |
498 | node->state = DLM_LAST_ACK; | |
499 | pr_debug("switch node %d to state %s case 1\n", | |
500 | node->nodeid, dlm_state_str(node->state)); | |
501 | spin_unlock(&node->state_lock); | |
502 | goto send_fin; | |
503 | } | |
504 | break; | |
505 | case DLM_FIN_WAIT1: | |
506 | node->state = DLM_CLOSING; | |
507 | pr_debug("switch node %d to state %s\n", | |
508 | node->nodeid, dlm_state_str(node->state)); | |
509 | break; | |
510 | case DLM_FIN_WAIT2: | |
511 | midcomms_node_reset(node); | |
512 | pr_debug("switch node %d to state %s\n", | |
513 | node->nodeid, dlm_state_str(node->state)); | |
514 | wake_up(&node->shutdown_wait); | |
515 | break; | |
516 | case DLM_LAST_ACK: | |
517 | /* probably remove_member caught it, do nothing */ | |
518 | break; | |
519 | default: | |
520 | spin_unlock(&node->state_lock); | |
521 | log_print("%s: unexpected state: %d\n", | |
522 | __func__, node->state); | |
523 | WARN_ON(1); | |
524 | return; | |
525 | } | |
526 | spin_unlock(&node->state_lock); | |
527 | ||
528 | set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags); | |
529 | break; | |
530 | default: | |
531 | WARN_ON(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags)); | |
532 | dlm_receive_buffer(p, node->nodeid); | |
b97f8525 | 533 | set_bit(DLM_NODE_ULP_DELIVERED, &node->flags); |
489d8e55 AA |
534 | break; |
535 | } | |
536 | } else { | |
537 | /* retry to ack message which we already have by sending back | |
538 | * current node->seq_next number as ack. | |
539 | */ | |
540 | if (seq < node->seq_next) | |
541 | dlm_send_ack(node->nodeid, node->seq_next); | |
542 | ||
543 | log_print_ratelimited("ignore dlm msg because seq mismatch, seq: %u, expected: %u, nodeid: %d", | |
544 | seq, node->seq_next, node->nodeid); | |
545 | } | |
546 | ||
547 | return; | |
548 | ||
549 | send_fin: | |
550 | set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags); | |
551 | dlm_send_fin(node, dlm_pas_fin_ack_rcv); | |
552 | } | |
553 | ||
554 | static struct midcomms_node * | |
555 | dlm_midcomms_recv_node_lookup(int nodeid, const union dlm_packet *p, | |
556 | uint16_t msglen, int (*cb)(struct midcomms_node *node)) | |
557 | { | |
558 | struct midcomms_node *node = NULL; | |
559 | gfp_t allocation = 0; | |
560 | int ret; | |
561 | ||
562 | switch (p->header.h_cmd) { | |
563 | case DLM_RCOM: | |
564 | if (msglen < sizeof(struct dlm_rcom)) { | |
565 | log_print("rcom msg too small: %u, will skip this message from node %d", | |
566 | msglen, nodeid); | |
567 | return NULL; | |
568 | } | |
569 | ||
570 | switch (le32_to_cpu(p->rcom.rc_type)) { | |
571 | case DLM_RCOM_NAMES: | |
572 | fallthrough; | |
573 | case DLM_RCOM_NAMES_REPLY: | |
574 | fallthrough; | |
575 | case DLM_RCOM_STATUS: | |
576 | fallthrough; | |
577 | case DLM_RCOM_STATUS_REPLY: | |
578 | node = nodeid2node(nodeid, 0); | |
579 | if (node) { | |
580 | spin_lock(&node->state_lock); | |
581 | if (node->state != DLM_ESTABLISHED) | |
582 | pr_debug("receive begin RCOM msg from node %d with state %s\n", | |
583 | node->nodeid, dlm_state_str(node->state)); | |
584 | ||
585 | switch (node->state) { | |
586 | case DLM_CLOSED: | |
587 | node->state = DLM_ESTABLISHED; | |
588 | pr_debug("switch node %d to state %s\n", | |
589 | node->nodeid, dlm_state_str(node->state)); | |
590 | break; | |
591 | case DLM_ESTABLISHED: | |
592 | break; | |
593 | default: | |
594 | /* some invalid state passive shutdown | |
595 | * was failed, we try to reset and | |
596 | * hope it will go on. | |
597 | */ | |
7d3848c0 | 598 | log_print("reset node %d because shutdown stuck", |
489d8e55 AA |
599 | node->nodeid); |
600 | ||
601 | midcomms_node_reset(node); | |
602 | node->state = DLM_ESTABLISHED; | |
603 | break; | |
604 | } | |
605 | spin_unlock(&node->state_lock); | |
606 | } | |
607 | ||
608 | allocation = GFP_NOFS; | |
609 | break; | |
610 | default: | |
611 | break; | |
612 | } | |
613 | ||
614 | break; | |
615 | default: | |
616 | break; | |
617 | } | |
618 | ||
619 | node = nodeid2node(nodeid, allocation); | |
620 | if (!node) { | |
957adb68 AA |
621 | switch (p->header.h_cmd) { |
622 | case DLM_OPTS: | |
623 | if (msglen < sizeof(struct dlm_opts)) { | |
624 | log_print("opts msg too small: %u, will skip this message from node %d", | |
625 | msglen, nodeid); | |
626 | return NULL; | |
627 | } | |
628 | ||
629 | log_print_ratelimited("received dlm opts message nextcmd %d from node %d in an invalid sequence", | |
630 | p->opts.o_nextcmd, nodeid); | |
631 | break; | |
632 | default: | |
633 | log_print_ratelimited("received dlm message cmd %d from node %d in an invalid sequence", | |
634 | p->header.h_cmd, nodeid); | |
635 | break; | |
636 | } | |
637 | ||
489d8e55 AA |
638 | return NULL; |
639 | } | |
640 | ||
641 | ret = cb(node); | |
642 | if (ret < 0) | |
643 | return NULL; | |
644 | ||
645 | return node; | |
646 | } | |
647 | ||
648 | static int dlm_midcomms_version_check_3_2(struct midcomms_node *node) | |
649 | { | |
650 | switch (node->version) { | |
651 | case DLM_VERSION_NOT_SET: | |
652 | node->version = DLM_VERSION_3_2; | |
653 | log_print("version 0x%08x for node %d detected", DLM_VERSION_3_2, | |
654 | node->nodeid); | |
655 | break; | |
656 | case DLM_VERSION_3_2: | |
657 | break; | |
658 | default: | |
659 | log_print_ratelimited("version mismatch detected, assumed 0x%08x but node %d has 0x%08x", | |
660 | DLM_VERSION_3_2, node->nodeid, node->version); | |
661 | return -1; | |
662 | } | |
663 | ||
664 | return 0; | |
665 | } | |
666 | ||
667 | static int dlm_opts_check_msglen(union dlm_packet *p, uint16_t msglen, int nodeid) | |
668 | { | |
669 | int len = msglen; | |
670 | ||
671 | /* we only trust outer header msglen because | |
672 | * it's checked against receive buffer length. | |
673 | */ | |
674 | if (len < sizeof(struct dlm_opts)) | |
675 | return -1; | |
676 | len -= sizeof(struct dlm_opts); | |
677 | ||
678 | if (len < le16_to_cpu(p->opts.o_optlen)) | |
679 | return -1; | |
680 | len -= le16_to_cpu(p->opts.o_optlen); | |
681 | ||
682 | switch (p->opts.o_nextcmd) { | |
683 | case DLM_FIN: | |
684 | if (len < sizeof(struct dlm_header)) { | |
685 | log_print("fin too small: %d, will skip this message from node %d", | |
686 | len, nodeid); | |
687 | return -1; | |
688 | } | |
689 | ||
690 | break; | |
691 | case DLM_MSG: | |
692 | if (len < sizeof(struct dlm_message)) { | |
693 | log_print("msg too small: %d, will skip this message from node %d", | |
694 | msglen, nodeid); | |
695 | return -1; | |
696 | } | |
697 | ||
698 | break; | |
699 | case DLM_RCOM: | |
700 | if (len < sizeof(struct dlm_rcom)) { | |
701 | log_print("rcom msg too small: %d, will skip this message from node %d", | |
702 | len, nodeid); | |
703 | return -1; | |
704 | } | |
705 | ||
706 | break; | |
707 | default: | |
708 | log_print("unsupported o_nextcmd received: %u, will skip this message from node %d", | |
709 | p->opts.o_nextcmd, nodeid); | |
710 | return -1; | |
711 | } | |
712 | ||
713 | return 0; | |
714 | } | |
715 | ||
716 | static void dlm_midcomms_receive_buffer_3_2(union dlm_packet *p, int nodeid) | |
717 | { | |
718 | uint16_t msglen = le16_to_cpu(p->header.h_length); | |
719 | struct midcomms_node *node; | |
720 | uint32_t seq; | |
721 | int ret, idx; | |
722 | ||
723 | idx = srcu_read_lock(&nodes_srcu); | |
724 | node = dlm_midcomms_recv_node_lookup(nodeid, p, msglen, | |
725 | dlm_midcomms_version_check_3_2); | |
726 | if (!node) | |
727 | goto out; | |
728 | ||
729 | switch (p->header.h_cmd) { | |
730 | case DLM_RCOM: | |
731 | /* these rcom message we use to determine version. | |
732 | * they have their own retransmission handling and | |
733 | * are the first messages of dlm. | |
734 | * | |
735 | * length already checked. | |
736 | */ | |
737 | switch (le32_to_cpu(p->rcom.rc_type)) { | |
738 | case DLM_RCOM_NAMES: | |
739 | fallthrough; | |
740 | case DLM_RCOM_NAMES_REPLY: | |
741 | fallthrough; | |
742 | case DLM_RCOM_STATUS: | |
743 | fallthrough; | |
744 | case DLM_RCOM_STATUS_REPLY: | |
745 | break; | |
746 | default: | |
747 | log_print("unsupported rcom type received: %u, will skip this message from node %d", | |
748 | le32_to_cpu(p->rcom.rc_type), nodeid); | |
749 | goto out; | |
750 | } | |
751 | ||
752 | WARN_ON(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags)); | |
753 | dlm_receive_buffer(p, nodeid); | |
754 | break; | |
755 | case DLM_OPTS: | |
756 | seq = le32_to_cpu(p->header.u.h_seq); | |
757 | ||
758 | ret = dlm_opts_check_msglen(p, msglen, nodeid); | |
759 | if (ret < 0) { | |
760 | log_print("opts msg too small: %u, will skip this message from node %d", | |
761 | msglen, nodeid); | |
762 | goto out; | |
763 | } | |
764 | ||
765 | p = (union dlm_packet *)((unsigned char *)p->opts.o_opts + | |
766 | le16_to_cpu(p->opts.o_optlen)); | |
767 | ||
768 | /* recheck inner msglen just if it's not garbage */ | |
769 | msglen = le16_to_cpu(p->header.h_length); | |
770 | switch (p->header.h_cmd) { | |
771 | case DLM_RCOM: | |
772 | if (msglen < sizeof(struct dlm_rcom)) { | |
773 | log_print("inner rcom msg too small: %u, will skip this message from node %d", | |
774 | msglen, nodeid); | |
775 | goto out; | |
776 | } | |
777 | ||
778 | break; | |
779 | case DLM_MSG: | |
780 | if (msglen < sizeof(struct dlm_message)) { | |
781 | log_print("inner msg too small: %u, will skip this message from node %d", | |
782 | msglen, nodeid); | |
783 | goto out; | |
784 | } | |
785 | ||
786 | break; | |
787 | case DLM_FIN: | |
788 | if (msglen < sizeof(struct dlm_header)) { | |
789 | log_print("inner fin too small: %u, will skip this message from node %d", | |
790 | msglen, nodeid); | |
791 | goto out; | |
792 | } | |
793 | ||
794 | break; | |
795 | default: | |
796 | log_print("unsupported inner h_cmd received: %u, will skip this message from node %d", | |
797 | msglen, nodeid); | |
798 | goto out; | |
799 | } | |
800 | ||
801 | dlm_midcomms_receive_buffer(p, node, seq); | |
802 | break; | |
803 | case DLM_ACK: | |
804 | seq = le32_to_cpu(p->header.u.h_seq); | |
805 | dlm_receive_ack(node, seq); | |
806 | break; | |
807 | default: | |
808 | log_print("unsupported h_cmd received: %u, will skip this message from node %d", | |
809 | p->header.h_cmd, nodeid); | |
810 | break; | |
811 | } | |
812 | ||
813 | out: | |
814 | srcu_read_unlock(&nodes_srcu, idx); | |
815 | } | |
816 | ||
817 | static int dlm_midcomms_version_check_3_1(struct midcomms_node *node) | |
a070a91c | 818 | { |
489d8e55 AA |
819 | switch (node->version) { |
820 | case DLM_VERSION_NOT_SET: | |
821 | node->version = DLM_VERSION_3_1; | |
822 | log_print("version 0x%08x for node %d detected", DLM_VERSION_3_1, | |
823 | node->nodeid); | |
824 | break; | |
825 | case DLM_VERSION_3_1: | |
826 | break; | |
827 | default: | |
828 | log_print_ratelimited("version mismatch detected, assumed 0x%08x but node %d has 0x%08x", | |
829 | DLM_VERSION_3_1, node->nodeid, node->version); | |
830 | return -1; | |
831 | } | |
832 | ||
833 | return 0; | |
834 | } | |
835 | ||
836 | static void dlm_midcomms_receive_buffer_3_1(union dlm_packet *p, int nodeid) | |
837 | { | |
838 | uint16_t msglen = le16_to_cpu(p->header.h_length); | |
839 | struct midcomms_node *node; | |
840 | int idx; | |
841 | ||
842 | idx = srcu_read_lock(&nodes_srcu); | |
843 | node = dlm_midcomms_recv_node_lookup(nodeid, p, msglen, | |
844 | dlm_midcomms_version_check_3_1); | |
845 | if (!node) { | |
846 | srcu_read_unlock(&nodes_srcu, idx); | |
847 | return; | |
848 | } | |
849 | srcu_read_unlock(&nodes_srcu, idx); | |
850 | ||
851 | switch (p->header.h_cmd) { | |
852 | case DLM_RCOM: | |
853 | /* length already checked */ | |
854 | break; | |
855 | case DLM_MSG: | |
856 | if (msglen < sizeof(struct dlm_message)) { | |
857 | log_print("msg too small: %u, will skip this message from node %d", | |
858 | msglen, nodeid); | |
859 | return; | |
860 | } | |
861 | ||
862 | break; | |
863 | default: | |
864 | log_print("unsupported h_cmd received: %u, will skip this message from node %d", | |
865 | p->header.h_cmd, nodeid); | |
866 | return; | |
867 | } | |
868 | ||
869 | dlm_receive_buffer(p, nodeid); | |
a070a91c AA |
870 | } |
871 | ||
e7fd4179 DT |
872 | /* |
873 | * Called from the low-level comms layer to process a buffer of | |
874 | * commands. | |
e7fd4179 DT |
875 | */ |
876 | ||
4798cbbf | 877 | int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len) |
e7fd4179 | 878 | { |
4798cbbf AA |
879 | const unsigned char *ptr = buf; |
880 | const struct dlm_header *hd; | |
e7fd4179 | 881 | uint16_t msglen; |
4798cbbf | 882 | int ret = 0; |
e7fd4179 | 883 | |
4798cbbf AA |
884 | while (len >= sizeof(struct dlm_header)) { |
885 | hd = (struct dlm_header *)ptr; | |
886 | ||
d10a0b88 | 887 | /* no message should be more than DLM_MAX_SOCKET_BUFSIZE or |
df9e06b8 AA |
888 | * less than dlm_header size. |
889 | * | |
890 | * Some messages does not have a 8 byte length boundary yet | |
891 | * which can occur in a unaligned memory access of some dlm | |
892 | * messages. However this problem need to be fixed at the | |
893 | * sending side, for now it seems nobody run into architecture | |
894 | * related issues yet but it slows down some processing. | |
895 | * Fixing this issue should be scheduled in future by doing | |
896 | * the next major version bump. | |
4798cbbf | 897 | */ |
df9e06b8 | 898 | msglen = le16_to_cpu(hd->h_length); |
d10a0b88 | 899 | if (msglen > DLM_MAX_SOCKET_BUFSIZE || |
710176e8 AA |
900 | msglen < sizeof(struct dlm_header)) { |
901 | log_print("received invalid length header: %u from node %d, will abort message parsing", | |
902 | msglen, nodeid); | |
4798cbbf | 903 | return -EBADMSG; |
e7fd4179 | 904 | } |
e7fd4179 | 905 | |
4798cbbf AA |
906 | /* caller will take care that leftover |
907 | * will be parsed next call with more data | |
908 | */ | |
e7fd4179 DT |
909 | if (msglen > len) |
910 | break; | |
911 | ||
489d8e55 AA |
912 | switch (le32_to_cpu(hd->h_version)) { |
913 | case DLM_VERSION_3_1: | |
914 | dlm_midcomms_receive_buffer_3_1((union dlm_packet *)ptr, nodeid); | |
4798cbbf | 915 | break; |
489d8e55 AA |
916 | case DLM_VERSION_3_2: |
917 | dlm_midcomms_receive_buffer_3_2((union dlm_packet *)ptr, nodeid); | |
4798cbbf AA |
918 | break; |
919 | default: | |
489d8e55 AA |
920 | log_print("received invalid version header: %u from node %d, will skip this message", |
921 | le32_to_cpu(hd->h_version), nodeid); | |
922 | break; | |
4798cbbf | 923 | } |
e7fd4179 DT |
924 | |
925 | ret += msglen; | |
e7fd4179 | 926 | len -= msglen; |
4798cbbf | 927 | ptr += msglen; |
e7fd4179 DT |
928 | } |
929 | ||
4798cbbf | 930 | return ret; |
e7fd4179 | 931 | } |
489d8e55 | 932 | |
b97f8525 AA |
933 | void dlm_midcomms_receive_done(int nodeid) |
934 | { | |
935 | struct midcomms_node *node; | |
936 | int idx; | |
937 | ||
938 | idx = srcu_read_lock(&nodes_srcu); | |
939 | node = nodeid2node(nodeid, 0); | |
940 | if (!node) { | |
941 | srcu_read_unlock(&nodes_srcu, idx); | |
942 | return; | |
943 | } | |
944 | ||
945 | /* old protocol, we do nothing */ | |
946 | switch (node->version) { | |
947 | case DLM_VERSION_3_2: | |
948 | break; | |
949 | default: | |
950 | srcu_read_unlock(&nodes_srcu, idx); | |
951 | return; | |
952 | } | |
953 | ||
954 | /* do nothing if we didn't delivered stateful to ulp */ | |
955 | if (!test_and_clear_bit(DLM_NODE_ULP_DELIVERED, | |
956 | &node->flags)) { | |
957 | srcu_read_unlock(&nodes_srcu, idx); | |
958 | return; | |
959 | } | |
960 | ||
961 | spin_lock(&node->state_lock); | |
962 | /* we only ack if state is ESTABLISHED */ | |
963 | switch (node->state) { | |
964 | case DLM_ESTABLISHED: | |
965 | spin_unlock(&node->state_lock); | |
966 | dlm_send_ack(node->nodeid, node->seq_next); | |
967 | break; | |
968 | default: | |
969 | spin_unlock(&node->state_lock); | |
970 | /* do nothing FIN has it's own ack send */ | |
971 | break; | |
972 | }; | |
973 | srcu_read_unlock(&nodes_srcu, idx); | |
974 | } | |
975 | ||
489d8e55 AA |
976 | void dlm_midcomms_unack_msg_resend(int nodeid) |
977 | { | |
978 | struct midcomms_node *node; | |
979 | struct dlm_mhandle *mh; | |
980 | int idx, ret; | |
981 | ||
982 | idx = srcu_read_lock(&nodes_srcu); | |
983 | node = nodeid2node(nodeid, 0); | |
984 | if (!node) { | |
985 | srcu_read_unlock(&nodes_srcu, idx); | |
986 | return; | |
987 | } | |
988 | ||
989 | /* old protocol, we don't support to retransmit on failure */ | |
990 | switch (node->version) { | |
991 | case DLM_VERSION_3_2: | |
992 | break; | |
993 | default: | |
994 | srcu_read_unlock(&nodes_srcu, idx); | |
995 | return; | |
996 | } | |
997 | ||
998 | rcu_read_lock(); | |
999 | list_for_each_entry_rcu(mh, &node->send_queue, list) { | |
1000 | if (!mh->committed) | |
1001 | continue; | |
1002 | ||
1003 | ret = dlm_lowcomms_resend_msg(mh->msg); | |
1004 | if (!ret) | |
1005 | log_print_ratelimited("retransmit dlm msg, seq %u, nodeid %d", | |
1006 | mh->seq, node->nodeid); | |
1007 | } | |
1008 | rcu_read_unlock(); | |
1009 | srcu_read_unlock(&nodes_srcu, idx); | |
1010 | } | |
1011 | ||
1012 | static void dlm_fill_opts_header(struct dlm_opts *opts, uint16_t inner_len, | |
1013 | uint32_t seq) | |
1014 | { | |
1015 | opts->o_header.h_cmd = DLM_OPTS; | |
1016 | opts->o_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR); | |
1017 | opts->o_header.h_nodeid = dlm_our_nodeid(); | |
1018 | opts->o_header.h_length = DLM_MIDCOMMS_OPT_LEN + inner_len; | |
1019 | opts->o_header.u.h_seq = seq; | |
1020 | header_out(&opts->o_header); | |
1021 | } | |
1022 | ||
1023 | static void midcomms_new_msg_cb(struct dlm_mhandle *mh) | |
1024 | { | |
1025 | atomic_inc(&mh->node->send_queue_cnt); | |
1026 | ||
1027 | spin_lock(&mh->node->send_queue_lock); | |
1028 | list_add_tail_rcu(&mh->list, &mh->node->send_queue); | |
1029 | spin_unlock(&mh->node->send_queue_lock); | |
1030 | ||
1031 | mh->seq = mh->node->seq_send++; | |
1032 | } | |
1033 | ||
1034 | static struct dlm_msg *dlm_midcomms_get_msg_3_2(struct dlm_mhandle *mh, int nodeid, | |
1035 | int len, gfp_t allocation, char **ppc) | |
1036 | { | |
1037 | struct dlm_opts *opts; | |
1038 | struct dlm_msg *msg; | |
1039 | ||
1040 | msg = dlm_lowcomms_new_msg(nodeid, len + DLM_MIDCOMMS_OPT_LEN, | |
1041 | allocation, ppc, midcomms_new_msg_cb, mh); | |
1042 | if (!msg) | |
1043 | return NULL; | |
1044 | ||
1045 | opts = (struct dlm_opts *)*ppc; | |
1046 | mh->opts = opts; | |
1047 | ||
1048 | /* add possible options here */ | |
1049 | dlm_fill_opts_header(opts, len, mh->seq); | |
1050 | ||
1051 | *ppc += sizeof(*opts); | |
1052 | mh->inner_hd = (const struct dlm_header *)*ppc; | |
1053 | return msg; | |
1054 | } | |
1055 | ||
1056 | struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len, | |
1057 | gfp_t allocation, char **ppc) | |
1058 | { | |
1059 | struct midcomms_node *node; | |
1060 | struct dlm_mhandle *mh; | |
1061 | struct dlm_msg *msg; | |
1062 | int idx; | |
1063 | ||
1064 | idx = srcu_read_lock(&nodes_srcu); | |
1065 | node = nodeid2node(nodeid, 0); | |
1066 | if (!node) { | |
1067 | WARN_ON_ONCE(1); | |
1068 | goto err; | |
1069 | } | |
1070 | ||
1071 | /* this is a bug, however we going on and hope it will be resolved */ | |
1072 | WARN_ON(test_bit(DLM_NODE_FLAG_STOP_TX, &node->flags)); | |
1073 | ||
1074 | mh = kzalloc(sizeof(*mh), GFP_NOFS); | |
1075 | if (!mh) | |
1076 | goto err; | |
1077 | ||
1078 | mh->idx = idx; | |
1079 | mh->node = node; | |
1080 | ||
1081 | switch (node->version) { | |
1082 | case DLM_VERSION_3_1: | |
1083 | msg = dlm_lowcomms_new_msg(nodeid, len, allocation, ppc, | |
1084 | NULL, NULL); | |
1085 | if (!msg) { | |
1086 | kfree(mh); | |
1087 | goto err; | |
1088 | } | |
1089 | ||
1090 | break; | |
1091 | case DLM_VERSION_3_2: | |
1092 | msg = dlm_midcomms_get_msg_3_2(mh, nodeid, len, allocation, | |
1093 | ppc); | |
1094 | if (!msg) { | |
1095 | kfree(mh); | |
1096 | goto err; | |
1097 | } | |
1098 | ||
1099 | break; | |
1100 | default: | |
1101 | kfree(mh); | |
1102 | WARN_ON(1); | |
1103 | goto err; | |
1104 | } | |
1105 | ||
1106 | mh->msg = msg; | |
1107 | ||
1108 | /* keep in mind that is a must to call | |
1109 | * dlm_midcomms_commit_msg() which releases | |
1110 | * nodes_srcu using mh->idx which is assumed | |
1111 | * here that the application will call it. | |
1112 | */ | |
1113 | return mh; | |
1114 | ||
1115 | err: | |
1116 | srcu_read_unlock(&nodes_srcu, idx); | |
1117 | return NULL; | |
1118 | } | |
1119 | ||
1120 | static void dlm_midcomms_commit_msg_3_2(struct dlm_mhandle *mh) | |
1121 | { | |
1122 | /* nexthdr chain for fast lookup */ | |
1123 | mh->opts->o_nextcmd = mh->inner_hd->h_cmd; | |
1124 | mh->committed = true; | |
1125 | dlm_lowcomms_commit_msg(mh->msg); | |
1126 | } | |
1127 | ||
1128 | void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh) | |
1129 | { | |
1130 | switch (mh->node->version) { | |
1131 | case DLM_VERSION_3_1: | |
1132 | srcu_read_unlock(&nodes_srcu, mh->idx); | |
1133 | ||
1134 | dlm_lowcomms_commit_msg(mh->msg); | |
1135 | dlm_lowcomms_put_msg(mh->msg); | |
1136 | /* mh is not part of rcu list in this case */ | |
1137 | kfree(mh); | |
1138 | break; | |
1139 | case DLM_VERSION_3_2: | |
1140 | dlm_midcomms_commit_msg_3_2(mh); | |
1141 | srcu_read_unlock(&nodes_srcu, mh->idx); | |
1142 | break; | |
1143 | default: | |
1144 | srcu_read_unlock(&nodes_srcu, mh->idx); | |
1145 | WARN_ON(1); | |
1146 | break; | |
1147 | } | |
1148 | } | |
1149 | ||
1150 | int dlm_midcomms_start(void) | |
1151 | { | |
1152 | int i; | |
1153 | ||
1154 | for (i = 0; i < CONN_HASH_SIZE; i++) | |
1155 | INIT_HLIST_HEAD(&node_hash[i]); | |
1156 | ||
1157 | return dlm_lowcomms_start(); | |
1158 | } | |
1159 | ||
1160 | static void dlm_act_fin_ack_rcv(struct midcomms_node *node) | |
1161 | { | |
1162 | spin_lock(&node->state_lock); | |
1163 | pr_debug("receive active fin ack from node %d with state %s\n", | |
1164 | node->nodeid, dlm_state_str(node->state)); | |
1165 | ||
1166 | switch (node->state) { | |
1167 | case DLM_FIN_WAIT1: | |
1168 | node->state = DLM_FIN_WAIT2; | |
1169 | pr_debug("switch node %d to state %s\n", | |
1170 | node->nodeid, dlm_state_str(node->state)); | |
1171 | break; | |
1172 | case DLM_CLOSING: | |
1173 | midcomms_node_reset(node); | |
1174 | pr_debug("switch node %d to state %s\n", | |
1175 | node->nodeid, dlm_state_str(node->state)); | |
1176 | wake_up(&node->shutdown_wait); | |
1177 | break; | |
1178 | case DLM_CLOSED: | |
1179 | /* not valid but somehow we got what we want */ | |
1180 | wake_up(&node->shutdown_wait); | |
1181 | break; | |
1182 | default: | |
1183 | spin_unlock(&node->state_lock); | |
1184 | log_print("%s: unexpected state: %d\n", | |
1185 | __func__, node->state); | |
1186 | WARN_ON(1); | |
1187 | return; | |
1188 | } | |
1189 | spin_unlock(&node->state_lock); | |
1190 | } | |
1191 | ||
1192 | void dlm_midcomms_add_member(int nodeid) | |
1193 | { | |
1194 | struct midcomms_node *node; | |
1195 | int idx; | |
1196 | ||
1197 | if (nodeid == dlm_our_nodeid()) | |
1198 | return; | |
1199 | ||
1200 | idx = srcu_read_lock(&nodes_srcu); | |
1201 | node = nodeid2node(nodeid, GFP_NOFS); | |
1202 | if (!node) { | |
1203 | srcu_read_unlock(&nodes_srcu, idx); | |
1204 | return; | |
1205 | } | |
1206 | ||
1207 | spin_lock(&node->state_lock); | |
1208 | if (!node->users) { | |
1209 | pr_debug("receive add member from node %d with state %s\n", | |
1210 | node->nodeid, dlm_state_str(node->state)); | |
1211 | switch (node->state) { | |
1212 | case DLM_ESTABLISHED: | |
1213 | break; | |
1214 | case DLM_CLOSED: | |
1215 | node->state = DLM_ESTABLISHED; | |
1216 | pr_debug("switch node %d to state %s\n", | |
1217 | node->nodeid, dlm_state_str(node->state)); | |
1218 | break; | |
1219 | default: | |
1220 | /* some invalid state passive shutdown | |
1221 | * was failed, we try to reset and | |
1222 | * hope it will go on. | |
1223 | */ | |
7d3848c0 | 1224 | log_print("reset node %d because shutdown stuck", |
489d8e55 AA |
1225 | node->nodeid); |
1226 | ||
1227 | midcomms_node_reset(node); | |
1228 | node->state = DLM_ESTABLISHED; | |
1229 | break; | |
1230 | } | |
1231 | } | |
1232 | ||
1233 | node->users++; | |
1234 | pr_debug("users inc count %d\n", node->users); | |
1235 | spin_unlock(&node->state_lock); | |
1236 | ||
1237 | srcu_read_unlock(&nodes_srcu, idx); | |
1238 | } | |
1239 | ||
1240 | void dlm_midcomms_remove_member(int nodeid) | |
1241 | { | |
1242 | struct midcomms_node *node; | |
1243 | int idx; | |
1244 | ||
1245 | if (nodeid == dlm_our_nodeid()) | |
1246 | return; | |
1247 | ||
1248 | idx = srcu_read_lock(&nodes_srcu); | |
1249 | node = nodeid2node(nodeid, 0); | |
1250 | if (!node) { | |
1251 | srcu_read_unlock(&nodes_srcu, idx); | |
1252 | return; | |
1253 | } | |
1254 | ||
1255 | spin_lock(&node->state_lock); | |
1256 | node->users--; | |
1257 | pr_debug("users dec count %d\n", node->users); | |
1258 | ||
1259 | /* hitting users count to zero means the | |
1260 | * other side is running dlm_midcomms_stop() | |
1261 | * we meet us to have a clean disconnect. | |
1262 | */ | |
1263 | if (node->users == 0) { | |
1264 | pr_debug("receive remove member from node %d with state %s\n", | |
1265 | node->nodeid, dlm_state_str(node->state)); | |
1266 | switch (node->state) { | |
1267 | case DLM_ESTABLISHED: | |
1268 | break; | |
1269 | case DLM_CLOSE_WAIT: | |
1270 | /* passive shutdown DLM_LAST_ACK case 2 */ | |
1271 | node->state = DLM_LAST_ACK; | |
1272 | spin_unlock(&node->state_lock); | |
1273 | ||
1274 | pr_debug("switch node %d to state %s case 2\n", | |
1275 | node->nodeid, dlm_state_str(node->state)); | |
1276 | goto send_fin; | |
1277 | case DLM_LAST_ACK: | |
1278 | /* probably receive fin caught it, do nothing */ | |
1279 | break; | |
1280 | case DLM_CLOSED: | |
1281 | /* already gone, do nothing */ | |
1282 | break; | |
1283 | default: | |
1284 | log_print("%s: unexpected state: %d\n", | |
1285 | __func__, node->state); | |
1286 | break; | |
1287 | } | |
1288 | } | |
1289 | spin_unlock(&node->state_lock); | |
1290 | ||
1291 | srcu_read_unlock(&nodes_srcu, idx); | |
1292 | return; | |
1293 | ||
1294 | send_fin: | |
1295 | set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags); | |
1296 | dlm_send_fin(node, dlm_pas_fin_ack_rcv); | |
1297 | srcu_read_unlock(&nodes_srcu, idx); | |
1298 | } | |
1299 | ||
1300 | static void midcomms_node_release(struct rcu_head *rcu) | |
1301 | { | |
1302 | struct midcomms_node *node = container_of(rcu, struct midcomms_node, rcu); | |
1303 | ||
1304 | WARN_ON(atomic_read(&node->send_queue_cnt)); | |
1305 | kfree(node); | |
1306 | } | |
1307 | ||
1308 | static void midcomms_shutdown(struct midcomms_node *node) | |
1309 | { | |
1310 | int ret; | |
1311 | ||
1312 | /* old protocol, we don't wait for pending operations */ | |
1313 | switch (node->version) { | |
1314 | case DLM_VERSION_3_2: | |
1315 | break; | |
1316 | default: | |
1317 | return; | |
1318 | } | |
1319 | ||
1320 | spin_lock(&node->state_lock); | |
1321 | pr_debug("receive active shutdown for node %d with state %s\n", | |
1322 | node->nodeid, dlm_state_str(node->state)); | |
1323 | switch (node->state) { | |
1324 | case DLM_ESTABLISHED: | |
1325 | node->state = DLM_FIN_WAIT1; | |
1326 | pr_debug("switch node %d to state %s case 2\n", | |
1327 | node->nodeid, dlm_state_str(node->state)); | |
1328 | break; | |
1329 | case DLM_CLOSED: | |
1330 | /* we have what we want */ | |
1331 | spin_unlock(&node->state_lock); | |
1332 | return; | |
1333 | default: | |
1334 | /* busy to enter DLM_FIN_WAIT1, wait until passive | |
1335 | * done in shutdown_wait to enter DLM_CLOSED. | |
1336 | */ | |
1337 | break; | |
1338 | } | |
1339 | spin_unlock(&node->state_lock); | |
1340 | ||
1341 | if (node->state == DLM_FIN_WAIT1) { | |
1342 | dlm_send_fin(node, dlm_act_fin_ack_rcv); | |
1343 | ||
1344 | if (DLM_DEBUG_FENCE_TERMINATION) | |
1345 | msleep(5000); | |
1346 | } | |
1347 | ||
1348 | /* wait for other side dlm + fin */ | |
1349 | ret = wait_event_timeout(node->shutdown_wait, | |
1350 | node->state == DLM_CLOSED || | |
1351 | test_bit(DLM_NODE_FLAG_CLOSE, &node->flags), | |
1352 | DLM_SHUTDOWN_TIMEOUT); | |
1353 | if (!ret || test_bit(DLM_NODE_FLAG_CLOSE, &node->flags)) { | |
1354 | pr_debug("active shutdown timed out for node %d with state %s\n", | |
1355 | node->nodeid, dlm_state_str(node->state)); | |
1356 | midcomms_node_reset(node); | |
1357 | return; | |
1358 | } | |
1359 | ||
1360 | pr_debug("active shutdown done for node %d with state %s\n", | |
1361 | node->nodeid, dlm_state_str(node->state)); | |
1362 | } | |
1363 | ||
1364 | void dlm_midcomms_shutdown(void) | |
1365 | { | |
1366 | struct midcomms_node *node; | |
1367 | int i, idx; | |
1368 | ||
1369 | mutex_lock(&close_lock); | |
1370 | idx = srcu_read_lock(&nodes_srcu); | |
1371 | for (i = 0; i < CONN_HASH_SIZE; i++) { | |
1372 | hlist_for_each_entry_rcu(node, &node_hash[i], hlist) { | |
1373 | midcomms_shutdown(node); | |
1374 | ||
5b2f981f AA |
1375 | dlm_delete_debug_comms_file(node->debugfs); |
1376 | ||
489d8e55 AA |
1377 | spin_lock(&nodes_lock); |
1378 | hlist_del_rcu(&node->hlist); | |
1379 | spin_unlock(&nodes_lock); | |
1380 | ||
1381 | call_srcu(&nodes_srcu, &node->rcu, midcomms_node_release); | |
1382 | } | |
1383 | } | |
1384 | srcu_read_unlock(&nodes_srcu, idx); | |
1385 | mutex_unlock(&close_lock); | |
1386 | ||
1387 | dlm_lowcomms_shutdown(); | |
1388 | } | |
1389 | ||
1390 | int dlm_midcomms_close(int nodeid) | |
1391 | { | |
1392 | struct midcomms_node *node; | |
1393 | int idx, ret; | |
1394 | ||
1395 | if (nodeid == dlm_our_nodeid()) | |
1396 | return 0; | |
1397 | ||
1398 | idx = srcu_read_lock(&nodes_srcu); | |
1399 | /* Abort pending close/remove operation */ | |
1400 | node = nodeid2node(nodeid, 0); | |
1401 | if (node) { | |
1402 | /* let shutdown waiters leave */ | |
1403 | set_bit(DLM_NODE_FLAG_CLOSE, &node->flags); | |
1404 | wake_up(&node->shutdown_wait); | |
1405 | } | |
1406 | srcu_read_unlock(&nodes_srcu, idx); | |
1407 | ||
1408 | synchronize_srcu(&nodes_srcu); | |
1409 | ||
1410 | idx = srcu_read_lock(&nodes_srcu); | |
1411 | mutex_lock(&close_lock); | |
1412 | node = nodeid2node(nodeid, 0); | |
1413 | if (!node) { | |
1414 | mutex_unlock(&close_lock); | |
1415 | srcu_read_unlock(&nodes_srcu, idx); | |
1416 | return dlm_lowcomms_close(nodeid); | |
1417 | } | |
1418 | ||
1419 | ret = dlm_lowcomms_close(nodeid); | |
1420 | spin_lock(&node->state_lock); | |
1421 | midcomms_node_reset(node); | |
1422 | spin_unlock(&node->state_lock); | |
1423 | srcu_read_unlock(&nodes_srcu, idx); | |
1424 | mutex_unlock(&close_lock); | |
1425 | ||
1426 | return ret; | |
1427 | } |