xref: /openbmc/linux/fs/dlm/lock.c (revision a325f174)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /******************************************************************************
3 *******************************************************************************
4 **
5 **  Copyright (C) 2005-2010 Red Hat, Inc.  All rights reserved.
6 **
7 **
8 *******************************************************************************
9 ******************************************************************************/
10 
11 /* Central locking logic has four stages:
12 
13    dlm_lock()
14    dlm_unlock()
15 
16    request_lock(ls, lkb)
17    convert_lock(ls, lkb)
18    unlock_lock(ls, lkb)
19    cancel_lock(ls, lkb)
20 
21    _request_lock(r, lkb)
22    _convert_lock(r, lkb)
23    _unlock_lock(r, lkb)
24    _cancel_lock(r, lkb)
25 
26    do_request(r, lkb)
27    do_convert(r, lkb)
28    do_unlock(r, lkb)
29    do_cancel(r, lkb)
30 
31    Stage 1 (lock, unlock) is mainly about checking input args and
32    splitting into one of the four main operations:
33 
34        dlm_lock          = request_lock
35        dlm_lock+CONVERT  = convert_lock
36        dlm_unlock        = unlock_lock
37        dlm_unlock+CANCEL = cancel_lock
38 
39    Stage 2, xxxx_lock(), just finds and locks the relevant rsb which is
40    provided to the next stage.
41 
42    Stage 3, _xxxx_lock(), determines if the operation is local or remote.
43    When remote, it calls send_xxxx(), when local it calls do_xxxx().
44 
45    Stage 4, do_xxxx(), is the guts of the operation.  It manipulates the
46    given rsb and lkb and queues callbacks.
47 
48    For remote operations, send_xxxx() results in the corresponding do_xxxx()
49    function being executed on the remote node.  The connecting send/receive
50    calls on local (L) and remote (R) nodes:
51 
52    L: send_xxxx()              ->  R: receive_xxxx()
53                                    R: do_xxxx()
54    L: receive_xxxx_reply()     <-  R: send_xxxx_reply()
55 */
56 #include <trace/events/dlm.h>
57 
58 #include <linux/types.h>
59 #include <linux/rbtree.h>
60 #include <linux/slab.h>
61 #include "dlm_internal.h"
62 #include <linux/dlm_device.h>
63 #include "memory.h"
64 #include "midcomms.h"
65 #include "requestqueue.h"
66 #include "util.h"
67 #include "dir.h"
68 #include "member.h"
69 #include "lockspace.h"
70 #include "ast.h"
71 #include "lock.h"
72 #include "rcom.h"
73 #include "recover.h"
74 #include "lvb_table.h"
75 #include "user.h"
76 #include "config.h"
77 
78 static int send_request(struct dlm_rsb *r, struct dlm_lkb *lkb);
79 static int send_convert(struct dlm_rsb *r, struct dlm_lkb *lkb);
80 static int send_unlock(struct dlm_rsb *r, struct dlm_lkb *lkb);
81 static int send_cancel(struct dlm_rsb *r, struct dlm_lkb *lkb);
82 static int send_grant(struct dlm_rsb *r, struct dlm_lkb *lkb);
83 static int send_bast(struct dlm_rsb *r, struct dlm_lkb *lkb, int mode);
84 static int send_lookup(struct dlm_rsb *r, struct dlm_lkb *lkb);
85 static int send_remove(struct dlm_rsb *r);
86 static int _request_lock(struct dlm_rsb *r, struct dlm_lkb *lkb);
87 static int _cancel_lock(struct dlm_rsb *r, struct dlm_lkb *lkb);
88 static void __receive_convert_reply(struct dlm_rsb *r, struct dlm_lkb *lkb,
89 				    const struct dlm_message *ms, bool local);
90 static int receive_extralen(const struct dlm_message *ms);
91 static void do_purge(struct dlm_ls *ls, int nodeid, int pid);
92 static void toss_rsb(struct kref *kref);
93 
94 /*
95  * Lock compatibilty matrix - thanks Steve
96  * UN = Unlocked state. Not really a state, used as a flag
97  * PD = Padding. Used to make the matrix a nice power of two in size
98  * Other states are the same as the VMS DLM.
99  * Usage: matrix[grmode+1][rqmode+1]  (although m[rq+1][gr+1] is the same)
100  */
101 
102 static const int __dlm_compat_matrix[8][8] = {
103       /* UN NL CR CW PR PW EX PD */
104         {1, 1, 1, 1, 1, 1, 1, 0},       /* UN */
105         {1, 1, 1, 1, 1, 1, 1, 0},       /* NL */
106         {1, 1, 1, 1, 1, 1, 0, 0},       /* CR */
107         {1, 1, 1, 1, 0, 0, 0, 0},       /* CW */
108         {1, 1, 1, 0, 1, 0, 0, 0},       /* PR */
109         {1, 1, 1, 0, 0, 0, 0, 0},       /* PW */
110         {1, 1, 0, 0, 0, 0, 0, 0},       /* EX */
111         {0, 0, 0, 0, 0, 0, 0, 0}        /* PD */
112 };
113 
114 /*
115  * This defines the direction of transfer of LVB data.
116  * Granted mode is the row; requested mode is the column.
117  * Usage: matrix[grmode+1][rqmode+1]
118  * 1 = LVB is returned to the caller
119  * 0 = LVB is written to the resource
120  * -1 = nothing happens to the LVB
121  */
122 
123 const int dlm_lvb_operations[8][8] = {
124         /* UN   NL  CR  CW  PR  PW  EX  PD*/
125         {  -1,  1,  1,  1,  1,  1,  1, -1 }, /* UN */
126         {  -1,  1,  1,  1,  1,  1,  1,  0 }, /* NL */
127         {  -1, -1,  1,  1,  1,  1,  1,  0 }, /* CR */
128         {  -1, -1, -1,  1,  1,  1,  1,  0 }, /* CW */
129         {  -1, -1, -1, -1,  1,  1,  1,  0 }, /* PR */
130         {  -1,  0,  0,  0,  0,  0,  1,  0 }, /* PW */
131         {  -1,  0,  0,  0,  0,  0,  0,  0 }, /* EX */
132         {  -1,  0,  0,  0,  0,  0,  0,  0 }  /* PD */
133 };
134 
135 #define modes_compat(gr, rq) \
136 	__dlm_compat_matrix[(gr)->lkb_grmode + 1][(rq)->lkb_rqmode + 1]
137 
138 int dlm_modes_compat(int mode1, int mode2)
139 {
140 	return __dlm_compat_matrix[mode1 + 1][mode2 + 1];
141 }
142 
143 /*
144  * Compatibility matrix for conversions with QUECVT set.
145  * Granted mode is the row; requested mode is the column.
146  * Usage: matrix[grmode+1][rqmode+1]
147  */
148 
149 static const int __quecvt_compat_matrix[8][8] = {
150       /* UN NL CR CW PR PW EX PD */
151         {0, 0, 0, 0, 0, 0, 0, 0},       /* UN */
152         {0, 0, 1, 1, 1, 1, 1, 0},       /* NL */
153         {0, 0, 0, 1, 1, 1, 1, 0},       /* CR */
154         {0, 0, 0, 0, 1, 1, 1, 0},       /* CW */
155         {0, 0, 0, 1, 0, 1, 1, 0},       /* PR */
156         {0, 0, 0, 0, 0, 0, 1, 0},       /* PW */
157         {0, 0, 0, 0, 0, 0, 0, 0},       /* EX */
158         {0, 0, 0, 0, 0, 0, 0, 0}        /* PD */
159 };
160 
161 void dlm_print_lkb(struct dlm_lkb *lkb)
162 {
163 	printk(KERN_ERR "lkb: nodeid %d id %x remid %x exflags %x flags %x "
164 	       "sts %d rq %d gr %d wait_type %d wait_nodeid %d seq %llu\n",
165 	       lkb->lkb_nodeid, lkb->lkb_id, lkb->lkb_remid, lkb->lkb_exflags,
166 	       dlm_iflags_val(lkb), lkb->lkb_status, lkb->lkb_rqmode,
167 	       lkb->lkb_grmode, lkb->lkb_wait_type, lkb->lkb_wait_nodeid,
168 	       (unsigned long long)lkb->lkb_recover_seq);
169 }
170 
171 static void dlm_print_rsb(struct dlm_rsb *r)
172 {
173 	printk(KERN_ERR "rsb: nodeid %d master %d dir %d flags %lx first %x "
174 	       "rlc %d name %s\n",
175 	       r->res_nodeid, r->res_master_nodeid, r->res_dir_nodeid,
176 	       r->res_flags, r->res_first_lkid, r->res_recover_locks_count,
177 	       r->res_name);
178 }
179 
180 void dlm_dump_rsb(struct dlm_rsb *r)
181 {
182 	struct dlm_lkb *lkb;
183 
184 	dlm_print_rsb(r);
185 
186 	printk(KERN_ERR "rsb: root_list empty %d recover_list empty %d\n",
187 	       list_empty(&r->res_root_list), list_empty(&r->res_recover_list));
188 	printk(KERN_ERR "rsb lookup list\n");
189 	list_for_each_entry(lkb, &r->res_lookup, lkb_rsb_lookup)
190 		dlm_print_lkb(lkb);
191 	printk(KERN_ERR "rsb grant queue:\n");
192 	list_for_each_entry(lkb, &r->res_grantqueue, lkb_statequeue)
193 		dlm_print_lkb(lkb);
194 	printk(KERN_ERR "rsb convert queue:\n");
195 	list_for_each_entry(lkb, &r->res_convertqueue, lkb_statequeue)
196 		dlm_print_lkb(lkb);
197 	printk(KERN_ERR "rsb wait queue:\n");
198 	list_for_each_entry(lkb, &r->res_waitqueue, lkb_statequeue)
199 		dlm_print_lkb(lkb);
200 }
201 
202 /* Threads cannot use the lockspace while it's being recovered */
203 
204 static inline void dlm_lock_recovery(struct dlm_ls *ls)
205 {
206 	down_read(&ls->ls_in_recovery);
207 }
208 
209 void dlm_unlock_recovery(struct dlm_ls *ls)
210 {
211 	up_read(&ls->ls_in_recovery);
212 }
213 
214 int dlm_lock_recovery_try(struct dlm_ls *ls)
215 {
216 	return down_read_trylock(&ls->ls_in_recovery);
217 }
218 
219 static inline int can_be_queued(struct dlm_lkb *lkb)
220 {
221 	return !(lkb->lkb_exflags & DLM_LKF_NOQUEUE);
222 }
223 
224 static inline int force_blocking_asts(struct dlm_lkb *lkb)
225 {
226 	return (lkb->lkb_exflags & DLM_LKF_NOQUEUEBAST);
227 }
228 
229 static inline int is_demoted(struct dlm_lkb *lkb)
230 {
231 	return test_bit(DLM_SBF_DEMOTED_BIT, &lkb->lkb_sbflags);
232 }
233 
234 static inline int is_altmode(struct dlm_lkb *lkb)
235 {
236 	return test_bit(DLM_SBF_ALTMODE_BIT, &lkb->lkb_sbflags);
237 }
238 
239 static inline int is_granted(struct dlm_lkb *lkb)
240 {
241 	return (lkb->lkb_status == DLM_LKSTS_GRANTED);
242 }
243 
244 static inline int is_remote(struct dlm_rsb *r)
245 {
246 	DLM_ASSERT(r->res_nodeid >= 0, dlm_print_rsb(r););
247 	return !!r->res_nodeid;
248 }
249 
250 static inline int is_process_copy(struct dlm_lkb *lkb)
251 {
252 	return lkb->lkb_nodeid &&
253 	       !test_bit(DLM_IFL_MSTCPY_BIT, &lkb->lkb_iflags);
254 }
255 
256 static inline int is_master_copy(struct dlm_lkb *lkb)
257 {
258 	return test_bit(DLM_IFL_MSTCPY_BIT, &lkb->lkb_iflags);
259 }
260 
261 static inline int middle_conversion(struct dlm_lkb *lkb)
262 {
263 	if ((lkb->lkb_grmode==DLM_LOCK_PR && lkb->lkb_rqmode==DLM_LOCK_CW) ||
264 	    (lkb->lkb_rqmode==DLM_LOCK_PR && lkb->lkb_grmode==DLM_LOCK_CW))
265 		return 1;
266 	return 0;
267 }
268 
269 static inline int down_conversion(struct dlm_lkb *lkb)
270 {
271 	return (!middle_conversion(lkb) && lkb->lkb_rqmode < lkb->lkb_grmode);
272 }
273 
274 static inline int is_overlap_unlock(struct dlm_lkb *lkb)
275 {
276 	return test_bit(DLM_IFL_OVERLAP_UNLOCK_BIT, &lkb->lkb_iflags);
277 }
278 
279 static inline int is_overlap_cancel(struct dlm_lkb *lkb)
280 {
281 	return test_bit(DLM_IFL_OVERLAP_CANCEL_BIT, &lkb->lkb_iflags);
282 }
283 
284 static inline int is_overlap(struct dlm_lkb *lkb)
285 {
286 	return test_bit(DLM_IFL_OVERLAP_UNLOCK_BIT, &lkb->lkb_iflags) ||
287 	       test_bit(DLM_IFL_OVERLAP_CANCEL_BIT, &lkb->lkb_iflags);
288 }
289 
290 static void queue_cast(struct dlm_rsb *r, struct dlm_lkb *lkb, int rv)
291 {
292 	if (is_master_copy(lkb))
293 		return;
294 
295 	DLM_ASSERT(lkb->lkb_lksb, dlm_print_lkb(lkb););
296 
297 	if (rv == -DLM_ECANCEL &&
298 	    test_and_clear_bit(DLM_IFL_DEADLOCK_CANCEL_BIT, &lkb->lkb_iflags))
299 		rv = -EDEADLK;
300 
301 	dlm_add_cb(lkb, DLM_CB_CAST, lkb->lkb_grmode, rv, dlm_sbflags_val(lkb));
302 }
303 
304 static inline void queue_cast_overlap(struct dlm_rsb *r, struct dlm_lkb *lkb)
305 {
306 	queue_cast(r, lkb,
307 		   is_overlap_unlock(lkb) ? -DLM_EUNLOCK : -DLM_ECANCEL);
308 }
309 
310 static void queue_bast(struct dlm_rsb *r, struct dlm_lkb *lkb, int rqmode)
311 {
312 	if (is_master_copy(lkb)) {
313 		send_bast(r, lkb, rqmode);
314 	} else {
315 		dlm_add_cb(lkb, DLM_CB_BAST, rqmode, 0, 0);
316 	}
317 }
318 
319 /*
320  * Basic operations on rsb's and lkb's
321  */
322 
323 /* This is only called to add a reference when the code already holds
324    a valid reference to the rsb, so there's no need for locking. */
325 
326 static inline void hold_rsb(struct dlm_rsb *r)
327 {
328 	kref_get(&r->res_ref);
329 }
330 
331 void dlm_hold_rsb(struct dlm_rsb *r)
332 {
333 	hold_rsb(r);
334 }
335 
336 /* When all references to the rsb are gone it's transferred to
337    the tossed list for later disposal. */
338 
339 static void put_rsb(struct dlm_rsb *r)
340 {
341 	struct dlm_ls *ls = r->res_ls;
342 	uint32_t bucket = r->res_bucket;
343 	int rv;
344 
345 	rv = kref_put_lock(&r->res_ref, toss_rsb,
346 			   &ls->ls_rsbtbl[bucket].lock);
347 	if (rv)
348 		spin_unlock(&ls->ls_rsbtbl[bucket].lock);
349 }
350 
351 void dlm_put_rsb(struct dlm_rsb *r)
352 {
353 	put_rsb(r);
354 }
355 
356 static int pre_rsb_struct(struct dlm_ls *ls)
357 {
358 	struct dlm_rsb *r1, *r2;
359 	int count = 0;
360 
361 	spin_lock(&ls->ls_new_rsb_spin);
362 	if (ls->ls_new_rsb_count > dlm_config.ci_new_rsb_count / 2) {
363 		spin_unlock(&ls->ls_new_rsb_spin);
364 		return 0;
365 	}
366 	spin_unlock(&ls->ls_new_rsb_spin);
367 
368 	r1 = dlm_allocate_rsb(ls);
369 	r2 = dlm_allocate_rsb(ls);
370 
371 	spin_lock(&ls->ls_new_rsb_spin);
372 	if (r1) {
373 		list_add(&r1->res_hashchain, &ls->ls_new_rsb);
374 		ls->ls_new_rsb_count++;
375 	}
376 	if (r2) {
377 		list_add(&r2->res_hashchain, &ls->ls_new_rsb);
378 		ls->ls_new_rsb_count++;
379 	}
380 	count = ls->ls_new_rsb_count;
381 	spin_unlock(&ls->ls_new_rsb_spin);
382 
383 	if (!count)
384 		return -ENOMEM;
385 	return 0;
386 }
387 
388 /* If ls->ls_new_rsb is empty, return -EAGAIN, so the caller can
389    unlock any spinlocks, go back and call pre_rsb_struct again.
390    Otherwise, take an rsb off the list and return it. */
391 
392 static int get_rsb_struct(struct dlm_ls *ls, const void *name, int len,
393 			  struct dlm_rsb **r_ret)
394 {
395 	struct dlm_rsb *r;
396 	int count;
397 
398 	spin_lock(&ls->ls_new_rsb_spin);
399 	if (list_empty(&ls->ls_new_rsb)) {
400 		count = ls->ls_new_rsb_count;
401 		spin_unlock(&ls->ls_new_rsb_spin);
402 		log_debug(ls, "find_rsb retry %d %d %s",
403 			  count, dlm_config.ci_new_rsb_count,
404 			  (const char *)name);
405 		return -EAGAIN;
406 	}
407 
408 	r = list_first_entry(&ls->ls_new_rsb, struct dlm_rsb, res_hashchain);
409 	list_del(&r->res_hashchain);
410 	/* Convert the empty list_head to a NULL rb_node for tree usage: */
411 	memset(&r->res_hashnode, 0, sizeof(struct rb_node));
412 	ls->ls_new_rsb_count--;
413 	spin_unlock(&ls->ls_new_rsb_spin);
414 
415 	r->res_ls = ls;
416 	r->res_length = len;
417 	memcpy(r->res_name, name, len);
418 	mutex_init(&r->res_mutex);
419 
420 	INIT_LIST_HEAD(&r->res_lookup);
421 	INIT_LIST_HEAD(&r->res_grantqueue);
422 	INIT_LIST_HEAD(&r->res_convertqueue);
423 	INIT_LIST_HEAD(&r->res_waitqueue);
424 	INIT_LIST_HEAD(&r->res_root_list);
425 	INIT_LIST_HEAD(&r->res_recover_list);
426 
427 	*r_ret = r;
428 	return 0;
429 }
430 
431 static int rsb_cmp(struct dlm_rsb *r, const char *name, int nlen)
432 {
433 	char maxname[DLM_RESNAME_MAXLEN];
434 
435 	memset(maxname, 0, DLM_RESNAME_MAXLEN);
436 	memcpy(maxname, name, nlen);
437 	return memcmp(r->res_name, maxname, DLM_RESNAME_MAXLEN);
438 }
439 
440 int dlm_search_rsb_tree(struct rb_root *tree, const void *name, int len,
441 			struct dlm_rsb **r_ret)
442 {
443 	struct rb_node *node = tree->rb_node;
444 	struct dlm_rsb *r;
445 	int rc;
446 
447 	while (node) {
448 		r = rb_entry(node, struct dlm_rsb, res_hashnode);
449 		rc = rsb_cmp(r, name, len);
450 		if (rc < 0)
451 			node = node->rb_left;
452 		else if (rc > 0)
453 			node = node->rb_right;
454 		else
455 			goto found;
456 	}
457 	*r_ret = NULL;
458 	return -EBADR;
459 
460  found:
461 	*r_ret = r;
462 	return 0;
463 }
464 
465 static int rsb_insert(struct dlm_rsb *rsb, struct rb_root *tree)
466 {
467 	struct rb_node **newn = &tree->rb_node;
468 	struct rb_node *parent = NULL;
469 	int rc;
470 
471 	while (*newn) {
472 		struct dlm_rsb *cur = rb_entry(*newn, struct dlm_rsb,
473 					       res_hashnode);
474 
475 		parent = *newn;
476 		rc = rsb_cmp(cur, rsb->res_name, rsb->res_length);
477 		if (rc < 0)
478 			newn = &parent->rb_left;
479 		else if (rc > 0)
480 			newn = &parent->rb_right;
481 		else {
482 			log_print("rsb_insert match");
483 			dlm_dump_rsb(rsb);
484 			dlm_dump_rsb(cur);
485 			return -EEXIST;
486 		}
487 	}
488 
489 	rb_link_node(&rsb->res_hashnode, parent, newn);
490 	rb_insert_color(&rsb->res_hashnode, tree);
491 	return 0;
492 }
493 
494 /*
495  * Find rsb in rsbtbl and potentially create/add one
496  *
497  * Delaying the release of rsb's has a similar benefit to applications keeping
498  * NL locks on an rsb, but without the guarantee that the cached master value
499  * will still be valid when the rsb is reused.  Apps aren't always smart enough
500  * to keep NL locks on an rsb that they may lock again shortly; this can lead
501  * to excessive master lookups and removals if we don't delay the release.
502  *
503  * Searching for an rsb means looking through both the normal list and toss
504  * list.  When found on the toss list the rsb is moved to the normal list with
505  * ref count of 1; when found on normal list the ref count is incremented.
506  *
507  * rsb's on the keep list are being used locally and refcounted.
508  * rsb's on the toss list are not being used locally, and are not refcounted.
509  *
510  * The toss list rsb's were either
511  * - previously used locally but not any more (were on keep list, then
512  *   moved to toss list when last refcount dropped)
513  * - created and put on toss list as a directory record for a lookup
514  *   (we are the dir node for the res, but are not using the res right now,
515  *   but some other node is)
516  *
517  * The purpose of find_rsb() is to return a refcounted rsb for local use.
518  * So, if the given rsb is on the toss list, it is moved to the keep list
519  * before being returned.
520  *
521  * toss_rsb() happens when all local usage of the rsb is done, i.e. no
522  * more refcounts exist, so the rsb is moved from the keep list to the
523  * toss list.
524  *
525  * rsb's on both keep and toss lists are used for doing a name to master
526  * lookups.  rsb's that are in use locally (and being refcounted) are on
527  * the keep list, rsb's that are not in use locally (not refcounted) and
528  * only exist for name/master lookups are on the toss list.
529  *
530  * rsb's on the toss list who's dir_nodeid is not local can have stale
531  * name/master mappings.  So, remote requests on such rsb's can potentially
532  * return with an error, which means the mapping is stale and needs to
533  * be updated with a new lookup.  (The idea behind MASTER UNCERTAIN and
534  * first_lkid is to keep only a single outstanding request on an rsb
535  * while that rsb has a potentially stale master.)
536  */
537 
538 static int find_rsb_dir(struct dlm_ls *ls, const void *name, int len,
539 			uint32_t hash, uint32_t b,
540 			int dir_nodeid, int from_nodeid,
541 			unsigned int flags, struct dlm_rsb **r_ret)
542 {
543 	struct dlm_rsb *r = NULL;
544 	int our_nodeid = dlm_our_nodeid();
545 	int from_local = 0;
546 	int from_other = 0;
547 	int from_dir = 0;
548 	int create = 0;
549 	int error;
550 
551 	if (flags & R_RECEIVE_REQUEST) {
552 		if (from_nodeid == dir_nodeid)
553 			from_dir = 1;
554 		else
555 			from_other = 1;
556 	} else if (flags & R_REQUEST) {
557 		from_local = 1;
558 	}
559 
560 	/*
561 	 * flags & R_RECEIVE_RECOVER is from dlm_recover_master_copy, so
562 	 * from_nodeid has sent us a lock in dlm_recover_locks, believing
563 	 * we're the new master.  Our local recovery may not have set
564 	 * res_master_nodeid to our_nodeid yet, so allow either.  Don't
565 	 * create the rsb; dlm_recover_process_copy() will handle EBADR
566 	 * by resending.
567 	 *
568 	 * If someone sends us a request, we are the dir node, and we do
569 	 * not find the rsb anywhere, then recreate it.  This happens if
570 	 * someone sends us a request after we have removed/freed an rsb
571 	 * from our toss list.  (They sent a request instead of lookup
572 	 * because they are using an rsb from their toss list.)
573 	 */
574 
575 	if (from_local || from_dir ||
576 	    (from_other && (dir_nodeid == our_nodeid))) {
577 		create = 1;
578 	}
579 
580  retry:
581 	if (create) {
582 		error = pre_rsb_struct(ls);
583 		if (error < 0)
584 			goto out;
585 	}
586 
587 	spin_lock(&ls->ls_rsbtbl[b].lock);
588 
589 	error = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].keep, name, len, &r);
590 	if (error)
591 		goto do_toss;
592 
593 	/*
594 	 * rsb is active, so we can't check master_nodeid without lock_rsb.
595 	 */
596 
597 	kref_get(&r->res_ref);
598 	goto out_unlock;
599 
600 
601  do_toss:
602 	error = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].toss, name, len, &r);
603 	if (error)
604 		goto do_new;
605 
606 	/*
607 	 * rsb found inactive (master_nodeid may be out of date unless
608 	 * we are the dir_nodeid or were the master)  No other thread
609 	 * is using this rsb because it's on the toss list, so we can
610 	 * look at or update res_master_nodeid without lock_rsb.
611 	 */
612 
613 	if ((r->res_master_nodeid != our_nodeid) && from_other) {
614 		/* our rsb was not master, and another node (not the dir node)
615 		   has sent us a request */
616 		log_debug(ls, "find_rsb toss from_other %d master %d dir %d %s",
617 			  from_nodeid, r->res_master_nodeid, dir_nodeid,
618 			  r->res_name);
619 		error = -ENOTBLK;
620 		goto out_unlock;
621 	}
622 
623 	if ((r->res_master_nodeid != our_nodeid) && from_dir) {
624 		/* don't think this should ever happen */
625 		log_error(ls, "find_rsb toss from_dir %d master %d",
626 			  from_nodeid, r->res_master_nodeid);
627 		dlm_print_rsb(r);
628 		/* fix it and go on */
629 		r->res_master_nodeid = our_nodeid;
630 		r->res_nodeid = 0;
631 		rsb_clear_flag(r, RSB_MASTER_UNCERTAIN);
632 		r->res_first_lkid = 0;
633 	}
634 
635 	if (from_local && (r->res_master_nodeid != our_nodeid)) {
636 		/* Because we have held no locks on this rsb,
637 		   res_master_nodeid could have become stale. */
638 		rsb_set_flag(r, RSB_MASTER_UNCERTAIN);
639 		r->res_first_lkid = 0;
640 	}
641 
642 	rb_erase(&r->res_hashnode, &ls->ls_rsbtbl[b].toss);
643 	error = rsb_insert(r, &ls->ls_rsbtbl[b].keep);
644 	goto out_unlock;
645 
646 
647  do_new:
648 	/*
649 	 * rsb not found
650 	 */
651 
652 	if (error == -EBADR && !create)
653 		goto out_unlock;
654 
655 	error = get_rsb_struct(ls, name, len, &r);
656 	if (error == -EAGAIN) {
657 		spin_unlock(&ls->ls_rsbtbl[b].lock);
658 		goto retry;
659 	}
660 	if (error)
661 		goto out_unlock;
662 
663 	r->res_hash = hash;
664 	r->res_bucket = b;
665 	r->res_dir_nodeid = dir_nodeid;
666 	kref_init(&r->res_ref);
667 
668 	if (from_dir) {
669 		/* want to see how often this happens */
670 		log_debug(ls, "find_rsb new from_dir %d recreate %s",
671 			  from_nodeid, r->res_name);
672 		r->res_master_nodeid = our_nodeid;
673 		r->res_nodeid = 0;
674 		goto out_add;
675 	}
676 
677 	if (from_other && (dir_nodeid != our_nodeid)) {
678 		/* should never happen */
679 		log_error(ls, "find_rsb new from_other %d dir %d our %d %s",
680 			  from_nodeid, dir_nodeid, our_nodeid, r->res_name);
681 		dlm_free_rsb(r);
682 		r = NULL;
683 		error = -ENOTBLK;
684 		goto out_unlock;
685 	}
686 
687 	if (from_other) {
688 		log_debug(ls, "find_rsb new from_other %d dir %d %s",
689 			  from_nodeid, dir_nodeid, r->res_name);
690 	}
691 
692 	if (dir_nodeid == our_nodeid) {
693 		/* When we are the dir nodeid, we can set the master
694 		   node immediately */
695 		r->res_master_nodeid = our_nodeid;
696 		r->res_nodeid = 0;
697 	} else {
698 		/* set_master will send_lookup to dir_nodeid */
699 		r->res_master_nodeid = 0;
700 		r->res_nodeid = -1;
701 	}
702 
703  out_add:
704 	error = rsb_insert(r, &ls->ls_rsbtbl[b].keep);
705  out_unlock:
706 	spin_unlock(&ls->ls_rsbtbl[b].lock);
707  out:
708 	*r_ret = r;
709 	return error;
710 }
711 
712 /* During recovery, other nodes can send us new MSTCPY locks (from
713    dlm_recover_locks) before we've made ourself master (in
714    dlm_recover_masters). */
715 
716 static int find_rsb_nodir(struct dlm_ls *ls, const void *name, int len,
717 			  uint32_t hash, uint32_t b,
718 			  int dir_nodeid, int from_nodeid,
719 			  unsigned int flags, struct dlm_rsb **r_ret)
720 {
721 	struct dlm_rsb *r = NULL;
722 	int our_nodeid = dlm_our_nodeid();
723 	int recover = (flags & R_RECEIVE_RECOVER);
724 	int error;
725 
726  retry:
727 	error = pre_rsb_struct(ls);
728 	if (error < 0)
729 		goto out;
730 
731 	spin_lock(&ls->ls_rsbtbl[b].lock);
732 
733 	error = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].keep, name, len, &r);
734 	if (error)
735 		goto do_toss;
736 
737 	/*
738 	 * rsb is active, so we can't check master_nodeid without lock_rsb.
739 	 */
740 
741 	kref_get(&r->res_ref);
742 	goto out_unlock;
743 
744 
745  do_toss:
746 	error = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].toss, name, len, &r);
747 	if (error)
748 		goto do_new;
749 
750 	/*
751 	 * rsb found inactive. No other thread is using this rsb because
752 	 * it's on the toss list, so we can look at or update
753 	 * res_master_nodeid without lock_rsb.
754 	 */
755 
756 	if (!recover && (r->res_master_nodeid != our_nodeid) && from_nodeid) {
757 		/* our rsb is not master, and another node has sent us a
758 		   request; this should never happen */
759 		log_error(ls, "find_rsb toss from_nodeid %d master %d dir %d",
760 			  from_nodeid, r->res_master_nodeid, dir_nodeid);
761 		dlm_print_rsb(r);
762 		error = -ENOTBLK;
763 		goto out_unlock;
764 	}
765 
766 	if (!recover && (r->res_master_nodeid != our_nodeid) &&
767 	    (dir_nodeid == our_nodeid)) {
768 		/* our rsb is not master, and we are dir; may as well fix it;
769 		   this should never happen */
770 		log_error(ls, "find_rsb toss our %d master %d dir %d",
771 			  our_nodeid, r->res_master_nodeid, dir_nodeid);
772 		dlm_print_rsb(r);
773 		r->res_master_nodeid = our_nodeid;
774 		r->res_nodeid = 0;
775 	}
776 
777 	rb_erase(&r->res_hashnode, &ls->ls_rsbtbl[b].toss);
778 	error = rsb_insert(r, &ls->ls_rsbtbl[b].keep);
779 	goto out_unlock;
780 
781 
782  do_new:
783 	/*
784 	 * rsb not found
785 	 */
786 
787 	error = get_rsb_struct(ls, name, len, &r);
788 	if (error == -EAGAIN) {
789 		spin_unlock(&ls->ls_rsbtbl[b].lock);
790 		goto retry;
791 	}
792 	if (error)
793 		goto out_unlock;
794 
795 	r->res_hash = hash;
796 	r->res_bucket = b;
797 	r->res_dir_nodeid = dir_nodeid;
798 	r->res_master_nodeid = dir_nodeid;
799 	r->res_nodeid = (dir_nodeid == our_nodeid) ? 0 : dir_nodeid;
800 	kref_init(&r->res_ref);
801 
802 	error = rsb_insert(r, &ls->ls_rsbtbl[b].keep);
803  out_unlock:
804 	spin_unlock(&ls->ls_rsbtbl[b].lock);
805  out:
806 	*r_ret = r;
807 	return error;
808 }
809 
810 static int find_rsb(struct dlm_ls *ls, const void *name, int len,
811 		    int from_nodeid, unsigned int flags,
812 		    struct dlm_rsb **r_ret)
813 {
814 	uint32_t hash, b;
815 	int dir_nodeid;
816 
817 	if (len > DLM_RESNAME_MAXLEN)
818 		return -EINVAL;
819 
820 	hash = jhash(name, len, 0);
821 	b = hash & (ls->ls_rsbtbl_size - 1);
822 
823 	dir_nodeid = dlm_hash2nodeid(ls, hash);
824 
825 	if (dlm_no_directory(ls))
826 		return find_rsb_nodir(ls, name, len, hash, b, dir_nodeid,
827 				      from_nodeid, flags, r_ret);
828 	else
829 		return find_rsb_dir(ls, name, len, hash, b, dir_nodeid,
830 				      from_nodeid, flags, r_ret);
831 }
832 
833 /* we have received a request and found that res_master_nodeid != our_nodeid,
834    so we need to return an error or make ourself the master */
835 
836 static int validate_master_nodeid(struct dlm_ls *ls, struct dlm_rsb *r,
837 				  int from_nodeid)
838 {
839 	if (dlm_no_directory(ls)) {
840 		log_error(ls, "find_rsb keep from_nodeid %d master %d dir %d",
841 			  from_nodeid, r->res_master_nodeid,
842 			  r->res_dir_nodeid);
843 		dlm_print_rsb(r);
844 		return -ENOTBLK;
845 	}
846 
847 	if (from_nodeid != r->res_dir_nodeid) {
848 		/* our rsb is not master, and another node (not the dir node)
849 	   	   has sent us a request.  this is much more common when our
850 	   	   master_nodeid is zero, so limit debug to non-zero.  */
851 
852 		if (r->res_master_nodeid) {
853 			log_debug(ls, "validate master from_other %d master %d "
854 				  "dir %d first %x %s", from_nodeid,
855 				  r->res_master_nodeid, r->res_dir_nodeid,
856 				  r->res_first_lkid, r->res_name);
857 		}
858 		return -ENOTBLK;
859 	} else {
860 		/* our rsb is not master, but the dir nodeid has sent us a
861 	   	   request; this could happen with master 0 / res_nodeid -1 */
862 
863 		if (r->res_master_nodeid) {
864 			log_error(ls, "validate master from_dir %d master %d "
865 				  "first %x %s",
866 				  from_nodeid, r->res_master_nodeid,
867 				  r->res_first_lkid, r->res_name);
868 		}
869 
870 		r->res_master_nodeid = dlm_our_nodeid();
871 		r->res_nodeid = 0;
872 		return 0;
873 	}
874 }
875 
876 static void __dlm_master_lookup(struct dlm_ls *ls, struct dlm_rsb *r, int our_nodeid,
877 				int from_nodeid, bool toss_list, unsigned int flags,
878 				int *r_nodeid, int *result)
879 {
880 	int fix_master = (flags & DLM_LU_RECOVER_MASTER);
881 	int from_master = (flags & DLM_LU_RECOVER_DIR);
882 
883 	if (r->res_dir_nodeid != our_nodeid) {
884 		/* should not happen, but may as well fix it and carry on */
885 		log_error(ls, "%s res_dir %d our %d %s", __func__,
886 			  r->res_dir_nodeid, our_nodeid, r->res_name);
887 		r->res_dir_nodeid = our_nodeid;
888 	}
889 
890 	if (fix_master && dlm_is_removed(ls, r->res_master_nodeid)) {
891 		/* Recovery uses this function to set a new master when
892 		 * the previous master failed.  Setting NEW_MASTER will
893 		 * force dlm_recover_masters to call recover_master on this
894 		 * rsb even though the res_nodeid is no longer removed.
895 		 */
896 
897 		r->res_master_nodeid = from_nodeid;
898 		r->res_nodeid = from_nodeid;
899 		rsb_set_flag(r, RSB_NEW_MASTER);
900 
901 		if (toss_list) {
902 			/* I don't think we should ever find it on toss list. */
903 			log_error(ls, "%s fix_master on toss", __func__);
904 			dlm_dump_rsb(r);
905 		}
906 	}
907 
908 	if (from_master && (r->res_master_nodeid != from_nodeid)) {
909 		/* this will happen if from_nodeid became master during
910 		 * a previous recovery cycle, and we aborted the previous
911 		 * cycle before recovering this master value
912 		 */
913 
914 		log_limit(ls, "%s from_master %d master_nodeid %d res_nodeid %d first %x %s",
915 			  __func__, from_nodeid, r->res_master_nodeid,
916 			  r->res_nodeid, r->res_first_lkid, r->res_name);
917 
918 		if (r->res_master_nodeid == our_nodeid) {
919 			log_error(ls, "from_master %d our_master", from_nodeid);
920 			dlm_dump_rsb(r);
921 			goto ret_assign;
922 		}
923 
924 		r->res_master_nodeid = from_nodeid;
925 		r->res_nodeid = from_nodeid;
926 		rsb_set_flag(r, RSB_NEW_MASTER);
927 	}
928 
929 	if (!r->res_master_nodeid) {
930 		/* this will happen if recovery happens while we're looking
931 		 * up the master for this rsb
932 		 */
933 
934 		log_debug(ls, "%s master 0 to %d first %x %s", __func__,
935 			  from_nodeid, r->res_first_lkid, r->res_name);
936 		r->res_master_nodeid = from_nodeid;
937 		r->res_nodeid = from_nodeid;
938 	}
939 
940 	if (!from_master && !fix_master &&
941 	    (r->res_master_nodeid == from_nodeid)) {
942 		/* this can happen when the master sends remove, the dir node
943 		 * finds the rsb on the keep list and ignores the remove,
944 		 * and the former master sends a lookup
945 		 */
946 
947 		log_limit(ls, "%s from master %d flags %x first %x %s",
948 			  __func__, from_nodeid, flags, r->res_first_lkid,
949 			  r->res_name);
950 	}
951 
952  ret_assign:
953 	*r_nodeid = r->res_master_nodeid;
954 	if (result)
955 		*result = DLM_LU_MATCH;
956 }
957 
958 /*
959  * We're the dir node for this res and another node wants to know the
960  * master nodeid.  During normal operation (non recovery) this is only
961  * called from receive_lookup(); master lookups when the local node is
962  * the dir node are done by find_rsb().
963  *
964  * normal operation, we are the dir node for a resource
965  * . _request_lock
966  * . set_master
967  * . send_lookup
968  * . receive_lookup
969  * . dlm_master_lookup flags 0
970  *
971  * recover directory, we are rebuilding dir for all resources
972  * . dlm_recover_directory
973  * . dlm_rcom_names
974  *   remote node sends back the rsb names it is master of and we are dir of
975  * . dlm_master_lookup RECOVER_DIR (fix_master 0, from_master 1)
976  *   we either create new rsb setting remote node as master, or find existing
977  *   rsb and set master to be the remote node.
978  *
979  * recover masters, we are finding the new master for resources
980  * . dlm_recover_masters
981  * . recover_master
982  * . dlm_send_rcom_lookup
983  * . receive_rcom_lookup
984  * . dlm_master_lookup RECOVER_MASTER (fix_master 1, from_master 0)
985  */
986 
987 int dlm_master_lookup(struct dlm_ls *ls, int from_nodeid, const char *name,
988 		      int len, unsigned int flags, int *r_nodeid, int *result)
989 {
990 	struct dlm_rsb *r = NULL;
991 	uint32_t hash, b;
992 	int our_nodeid = dlm_our_nodeid();
993 	int dir_nodeid, error;
994 
995 	if (len > DLM_RESNAME_MAXLEN)
996 		return -EINVAL;
997 
998 	if (from_nodeid == our_nodeid) {
999 		log_error(ls, "dlm_master_lookup from our_nodeid %d flags %x",
1000 			  our_nodeid, flags);
1001 		return -EINVAL;
1002 	}
1003 
1004 	hash = jhash(name, len, 0);
1005 	b = hash & (ls->ls_rsbtbl_size - 1);
1006 
1007 	dir_nodeid = dlm_hash2nodeid(ls, hash);
1008 	if (dir_nodeid != our_nodeid) {
1009 		log_error(ls, "dlm_master_lookup from %d dir %d our %d h %x %d",
1010 			  from_nodeid, dir_nodeid, our_nodeid, hash,
1011 			  ls->ls_num_nodes);
1012 		*r_nodeid = -1;
1013 		return -EINVAL;
1014 	}
1015 
1016  retry:
1017 	error = pre_rsb_struct(ls);
1018 	if (error < 0)
1019 		return error;
1020 
1021 	spin_lock(&ls->ls_rsbtbl[b].lock);
1022 	error = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].keep, name, len, &r);
1023 	if (!error) {
1024 		/* because the rsb is active, we need to lock_rsb before
1025 		 * checking/changing re_master_nodeid
1026 		 */
1027 
1028 		hold_rsb(r);
1029 		spin_unlock(&ls->ls_rsbtbl[b].lock);
1030 		lock_rsb(r);
1031 
1032 		__dlm_master_lookup(ls, r, our_nodeid, from_nodeid, false,
1033 				    flags, r_nodeid, result);
1034 
1035 		/* the rsb was active */
1036 		unlock_rsb(r);
1037 		put_rsb(r);
1038 
1039 		return 0;
1040 	}
1041 
1042 	error = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].toss, name, len, &r);
1043 	if (error)
1044 		goto not_found;
1045 
1046 	/* because the rsb is inactive (on toss list), it's not refcounted
1047 	 * and lock_rsb is not used, but is protected by the rsbtbl lock
1048 	 */
1049 
1050 	__dlm_master_lookup(ls, r, our_nodeid, from_nodeid, true, flags,
1051 			    r_nodeid, result);
1052 
1053 	r->res_toss_time = jiffies;
1054 	/* the rsb was inactive (on toss list) */
1055 	spin_unlock(&ls->ls_rsbtbl[b].lock);
1056 
1057 	return 0;
1058 
1059  not_found:
1060 	error = get_rsb_struct(ls, name, len, &r);
1061 	if (error == -EAGAIN) {
1062 		spin_unlock(&ls->ls_rsbtbl[b].lock);
1063 		goto retry;
1064 	}
1065 	if (error)
1066 		goto out_unlock;
1067 
1068 	r->res_hash = hash;
1069 	r->res_bucket = b;
1070 	r->res_dir_nodeid = our_nodeid;
1071 	r->res_master_nodeid = from_nodeid;
1072 	r->res_nodeid = from_nodeid;
1073 	kref_init(&r->res_ref);
1074 	r->res_toss_time = jiffies;
1075 
1076 	error = rsb_insert(r, &ls->ls_rsbtbl[b].toss);
1077 	if (error) {
1078 		/* should never happen */
1079 		dlm_free_rsb(r);
1080 		spin_unlock(&ls->ls_rsbtbl[b].lock);
1081 		goto retry;
1082 	}
1083 
1084 	if (result)
1085 		*result = DLM_LU_ADD;
1086 	*r_nodeid = from_nodeid;
1087  out_unlock:
1088 	spin_unlock(&ls->ls_rsbtbl[b].lock);
1089 	return error;
1090 }
1091 
1092 static void dlm_dump_rsb_hash(struct dlm_ls *ls, uint32_t hash)
1093 {
1094 	struct rb_node *n;
1095 	struct dlm_rsb *r;
1096 	int i;
1097 
1098 	for (i = 0; i < ls->ls_rsbtbl_size; i++) {
1099 		spin_lock(&ls->ls_rsbtbl[i].lock);
1100 		for (n = rb_first(&ls->ls_rsbtbl[i].keep); n; n = rb_next(n)) {
1101 			r = rb_entry(n, struct dlm_rsb, res_hashnode);
1102 			if (r->res_hash == hash)
1103 				dlm_dump_rsb(r);
1104 		}
1105 		spin_unlock(&ls->ls_rsbtbl[i].lock);
1106 	}
1107 }
1108 
1109 void dlm_dump_rsb_name(struct dlm_ls *ls, const char *name, int len)
1110 {
1111 	struct dlm_rsb *r = NULL;
1112 	uint32_t hash, b;
1113 	int error;
1114 
1115 	hash = jhash(name, len, 0);
1116 	b = hash & (ls->ls_rsbtbl_size - 1);
1117 
1118 	spin_lock(&ls->ls_rsbtbl[b].lock);
1119 	error = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].keep, name, len, &r);
1120 	if (!error)
1121 		goto out_dump;
1122 
1123 	error = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].toss, name, len, &r);
1124 	if (error)
1125 		goto out;
1126  out_dump:
1127 	dlm_dump_rsb(r);
1128  out:
1129 	spin_unlock(&ls->ls_rsbtbl[b].lock);
1130 }
1131 
1132 static void toss_rsb(struct kref *kref)
1133 {
1134 	struct dlm_rsb *r = container_of(kref, struct dlm_rsb, res_ref);
1135 	struct dlm_ls *ls = r->res_ls;
1136 
1137 	DLM_ASSERT(list_empty(&r->res_root_list), dlm_print_rsb(r););
1138 	kref_init(&r->res_ref);
1139 	rb_erase(&r->res_hashnode, &ls->ls_rsbtbl[r->res_bucket].keep);
1140 	rsb_insert(r, &ls->ls_rsbtbl[r->res_bucket].toss);
1141 	r->res_toss_time = jiffies;
1142 	set_bit(DLM_RTF_SHRINK_BIT, &ls->ls_rsbtbl[r->res_bucket].flags);
1143 	if (r->res_lvbptr) {
1144 		dlm_free_lvb(r->res_lvbptr);
1145 		r->res_lvbptr = NULL;
1146 	}
1147 }
1148 
1149 /* See comment for unhold_lkb */
1150 
1151 static void unhold_rsb(struct dlm_rsb *r)
1152 {
1153 	int rv;
1154 	rv = kref_put(&r->res_ref, toss_rsb);
1155 	DLM_ASSERT(!rv, dlm_dump_rsb(r););
1156 }
1157 
1158 static void kill_rsb(struct kref *kref)
1159 {
1160 	struct dlm_rsb *r = container_of(kref, struct dlm_rsb, res_ref);
1161 
1162 	/* All work is done after the return from kref_put() so we
1163 	   can release the write_lock before the remove and free. */
1164 
1165 	DLM_ASSERT(list_empty(&r->res_lookup), dlm_dump_rsb(r););
1166 	DLM_ASSERT(list_empty(&r->res_grantqueue), dlm_dump_rsb(r););
1167 	DLM_ASSERT(list_empty(&r->res_convertqueue), dlm_dump_rsb(r););
1168 	DLM_ASSERT(list_empty(&r->res_waitqueue), dlm_dump_rsb(r););
1169 	DLM_ASSERT(list_empty(&r->res_root_list), dlm_dump_rsb(r););
1170 	DLM_ASSERT(list_empty(&r->res_recover_list), dlm_dump_rsb(r););
1171 }
1172 
1173 /* Attaching/detaching lkb's from rsb's is for rsb reference counting.
1174    The rsb must exist as long as any lkb's for it do. */
1175 
1176 static void attach_lkb(struct dlm_rsb *r, struct dlm_lkb *lkb)
1177 {
1178 	hold_rsb(r);
1179 	lkb->lkb_resource = r;
1180 }
1181 
1182 static void detach_lkb(struct dlm_lkb *lkb)
1183 {
1184 	if (lkb->lkb_resource) {
1185 		put_rsb(lkb->lkb_resource);
1186 		lkb->lkb_resource = NULL;
1187 	}
1188 }
1189 
1190 static int _create_lkb(struct dlm_ls *ls, struct dlm_lkb **lkb_ret,
1191 		       int start, int end)
1192 {
1193 	struct dlm_lkb *lkb;
1194 	int rv;
1195 
1196 	lkb = dlm_allocate_lkb(ls);
1197 	if (!lkb)
1198 		return -ENOMEM;
1199 
1200 	lkb->lkb_last_bast_mode = -1;
1201 	lkb->lkb_nodeid = -1;
1202 	lkb->lkb_grmode = DLM_LOCK_IV;
1203 	kref_init(&lkb->lkb_ref);
1204 	INIT_LIST_HEAD(&lkb->lkb_ownqueue);
1205 	INIT_LIST_HEAD(&lkb->lkb_rsb_lookup);
1206 	INIT_LIST_HEAD(&lkb->lkb_cb_list);
1207 	INIT_LIST_HEAD(&lkb->lkb_callbacks);
1208 	spin_lock_init(&lkb->lkb_cb_lock);
1209 	INIT_WORK(&lkb->lkb_cb_work, dlm_callback_work);
1210 
1211 	idr_preload(GFP_NOFS);
1212 	spin_lock(&ls->ls_lkbidr_spin);
1213 	rv = idr_alloc(&ls->ls_lkbidr, lkb, start, end, GFP_NOWAIT);
1214 	if (rv >= 0)
1215 		lkb->lkb_id = rv;
1216 	spin_unlock(&ls->ls_lkbidr_spin);
1217 	idr_preload_end();
1218 
1219 	if (rv < 0) {
1220 		log_error(ls, "create_lkb idr error %d", rv);
1221 		dlm_free_lkb(lkb);
1222 		return rv;
1223 	}
1224 
1225 	*lkb_ret = lkb;
1226 	return 0;
1227 }
1228 
1229 static int create_lkb(struct dlm_ls *ls, struct dlm_lkb **lkb_ret)
1230 {
1231 	return _create_lkb(ls, lkb_ret, 1, 0);
1232 }
1233 
1234 static int find_lkb(struct dlm_ls *ls, uint32_t lkid, struct dlm_lkb **lkb_ret)
1235 {
1236 	struct dlm_lkb *lkb;
1237 
1238 	spin_lock(&ls->ls_lkbidr_spin);
1239 	lkb = idr_find(&ls->ls_lkbidr, lkid);
1240 	if (lkb)
1241 		kref_get(&lkb->lkb_ref);
1242 	spin_unlock(&ls->ls_lkbidr_spin);
1243 
1244 	*lkb_ret = lkb;
1245 	return lkb ? 0 : -ENOENT;
1246 }
1247 
1248 static void kill_lkb(struct kref *kref)
1249 {
1250 	struct dlm_lkb *lkb = container_of(kref, struct dlm_lkb, lkb_ref);
1251 
1252 	/* All work is done after the return from kref_put() so we
1253 	   can release the write_lock before the detach_lkb */
1254 
1255 	DLM_ASSERT(!lkb->lkb_status, dlm_print_lkb(lkb););
1256 }
1257 
1258 /* __put_lkb() is used when an lkb may not have an rsb attached to
1259    it so we need to provide the lockspace explicitly */
1260 
1261 static int __put_lkb(struct dlm_ls *ls, struct dlm_lkb *lkb)
1262 {
1263 	uint32_t lkid = lkb->lkb_id;
1264 	int rv;
1265 
1266 	rv = kref_put_lock(&lkb->lkb_ref, kill_lkb,
1267 			   &ls->ls_lkbidr_spin);
1268 	if (rv) {
1269 		idr_remove(&ls->ls_lkbidr, lkid);
1270 		spin_unlock(&ls->ls_lkbidr_spin);
1271 
1272 		detach_lkb(lkb);
1273 
1274 		/* for local/process lkbs, lvbptr points to caller's lksb */
1275 		if (lkb->lkb_lvbptr && is_master_copy(lkb))
1276 			dlm_free_lvb(lkb->lkb_lvbptr);
1277 		dlm_free_lkb(lkb);
1278 	}
1279 
1280 	return rv;
1281 }
1282 
1283 int dlm_put_lkb(struct dlm_lkb *lkb)
1284 {
1285 	struct dlm_ls *ls;
1286 
1287 	DLM_ASSERT(lkb->lkb_resource, dlm_print_lkb(lkb););
1288 	DLM_ASSERT(lkb->lkb_resource->res_ls, dlm_print_lkb(lkb););
1289 
1290 	ls = lkb->lkb_resource->res_ls;
1291 	return __put_lkb(ls, lkb);
1292 }
1293 
1294 /* This is only called to add a reference when the code already holds
1295    a valid reference to the lkb, so there's no need for locking. */
1296 
1297 static inline void hold_lkb(struct dlm_lkb *lkb)
1298 {
1299 	kref_get(&lkb->lkb_ref);
1300 }
1301 
1302 static void unhold_lkb_assert(struct kref *kref)
1303 {
1304 	struct dlm_lkb *lkb = container_of(kref, struct dlm_lkb, lkb_ref);
1305 
1306 	DLM_ASSERT(false, dlm_print_lkb(lkb););
1307 }
1308 
1309 /* This is called when we need to remove a reference and are certain
1310    it's not the last ref.  e.g. del_lkb is always called between a
1311    find_lkb/put_lkb and is always the inverse of a previous add_lkb.
1312    put_lkb would work fine, but would involve unnecessary locking */
1313 
1314 static inline void unhold_lkb(struct dlm_lkb *lkb)
1315 {
1316 	kref_put(&lkb->lkb_ref, unhold_lkb_assert);
1317 }
1318 
1319 static void lkb_add_ordered(struct list_head *new, struct list_head *head,
1320 			    int mode)
1321 {
1322 	struct dlm_lkb *lkb = NULL, *iter;
1323 
1324 	list_for_each_entry(iter, head, lkb_statequeue)
1325 		if (iter->lkb_rqmode < mode) {
1326 			lkb = iter;
1327 			list_add_tail(new, &iter->lkb_statequeue);
1328 			break;
1329 		}
1330 
1331 	if (!lkb)
1332 		list_add_tail(new, head);
1333 }
1334 
1335 /* add/remove lkb to rsb's grant/convert/wait queue */
1336 
1337 static void add_lkb(struct dlm_rsb *r, struct dlm_lkb *lkb, int status)
1338 {
1339 	kref_get(&lkb->lkb_ref);
1340 
1341 	DLM_ASSERT(!lkb->lkb_status, dlm_print_lkb(lkb););
1342 
1343 	lkb->lkb_timestamp = ktime_get();
1344 
1345 	lkb->lkb_status = status;
1346 
1347 	switch (status) {
1348 	case DLM_LKSTS_WAITING:
1349 		if (lkb->lkb_exflags & DLM_LKF_HEADQUE)
1350 			list_add(&lkb->lkb_statequeue, &r->res_waitqueue);
1351 		else
1352 			list_add_tail(&lkb->lkb_statequeue, &r->res_waitqueue);
1353 		break;
1354 	case DLM_LKSTS_GRANTED:
1355 		/* convention says granted locks kept in order of grmode */
1356 		lkb_add_ordered(&lkb->lkb_statequeue, &r->res_grantqueue,
1357 				lkb->lkb_grmode);
1358 		break;
1359 	case DLM_LKSTS_CONVERT:
1360 		if (lkb->lkb_exflags & DLM_LKF_HEADQUE)
1361 			list_add(&lkb->lkb_statequeue, &r->res_convertqueue);
1362 		else
1363 			list_add_tail(&lkb->lkb_statequeue,
1364 				      &r->res_convertqueue);
1365 		break;
1366 	default:
1367 		DLM_ASSERT(0, dlm_print_lkb(lkb); printk("sts=%d\n", status););
1368 	}
1369 }
1370 
1371 static void del_lkb(struct dlm_rsb *r, struct dlm_lkb *lkb)
1372 {
1373 	lkb->lkb_status = 0;
1374 	list_del(&lkb->lkb_statequeue);
1375 	unhold_lkb(lkb);
1376 }
1377 
1378 static void move_lkb(struct dlm_rsb *r, struct dlm_lkb *lkb, int sts)
1379 {
1380 	hold_lkb(lkb);
1381 	del_lkb(r, lkb);
1382 	add_lkb(r, lkb, sts);
1383 	unhold_lkb(lkb);
1384 }
1385 
1386 static int msg_reply_type(int mstype)
1387 {
1388 	switch (mstype) {
1389 	case DLM_MSG_REQUEST:
1390 		return DLM_MSG_REQUEST_REPLY;
1391 	case DLM_MSG_CONVERT:
1392 		return DLM_MSG_CONVERT_REPLY;
1393 	case DLM_MSG_UNLOCK:
1394 		return DLM_MSG_UNLOCK_REPLY;
1395 	case DLM_MSG_CANCEL:
1396 		return DLM_MSG_CANCEL_REPLY;
1397 	case DLM_MSG_LOOKUP:
1398 		return DLM_MSG_LOOKUP_REPLY;
1399 	}
1400 	return -1;
1401 }
1402 
1403 /* add/remove lkb from global waiters list of lkb's waiting for
1404    a reply from a remote node */
1405 
1406 static int add_to_waiters(struct dlm_lkb *lkb, int mstype, int to_nodeid)
1407 {
1408 	struct dlm_ls *ls = lkb->lkb_resource->res_ls;
1409 	int error = 0;
1410 	int wc;
1411 
1412 	mutex_lock(&ls->ls_waiters_mutex);
1413 
1414 	if (is_overlap_unlock(lkb) ||
1415 	    (is_overlap_cancel(lkb) && (mstype == DLM_MSG_CANCEL))) {
1416 		error = -EINVAL;
1417 		goto out;
1418 	}
1419 
1420 	if (lkb->lkb_wait_type || is_overlap_cancel(lkb)) {
1421 		switch (mstype) {
1422 		case DLM_MSG_UNLOCK:
1423 			set_bit(DLM_IFL_OVERLAP_UNLOCK_BIT, &lkb->lkb_iflags);
1424 			break;
1425 		case DLM_MSG_CANCEL:
1426 			set_bit(DLM_IFL_OVERLAP_CANCEL_BIT, &lkb->lkb_iflags);
1427 			break;
1428 		default:
1429 			error = -EBUSY;
1430 			goto out;
1431 		}
1432 		wc = atomic_inc_return(&lkb->lkb_wait_count);
1433 		hold_lkb(lkb);
1434 
1435 		log_debug(ls, "addwait %x cur %d overlap %d count %d f %x",
1436 			  lkb->lkb_id, lkb->lkb_wait_type, mstype, wc,
1437 			  dlm_iflags_val(lkb));
1438 		goto out;
1439 	}
1440 
1441 	wc = atomic_fetch_inc(&lkb->lkb_wait_count);
1442 	DLM_ASSERT(!wc, dlm_print_lkb(lkb); printk("wait_count %d\n", wc););
1443 	lkb->lkb_wait_type = mstype;
1444 	lkb->lkb_wait_nodeid = to_nodeid; /* for debugging */
1445 	hold_lkb(lkb);
1446 	list_add(&lkb->lkb_wait_reply, &ls->ls_waiters);
1447  out:
1448 	if (error)
1449 		log_error(ls, "addwait error %x %d flags %x %d %d %s",
1450 			  lkb->lkb_id, error, dlm_iflags_val(lkb), mstype,
1451 			  lkb->lkb_wait_type, lkb->lkb_resource->res_name);
1452 	mutex_unlock(&ls->ls_waiters_mutex);
1453 	return error;
1454 }
1455 
1456 /* We clear the RESEND flag because we might be taking an lkb off the waiters
1457    list as part of process_requestqueue (e.g. a lookup that has an optimized
1458    request reply on the requestqueue) between dlm_recover_waiters_pre() which
1459    set RESEND and dlm_recover_waiters_post() */
1460 
1461 static int _remove_from_waiters(struct dlm_lkb *lkb, int mstype,
1462 				const struct dlm_message *ms)
1463 {
1464 	struct dlm_ls *ls = lkb->lkb_resource->res_ls;
1465 	int overlap_done = 0;
1466 
1467 	if (mstype == DLM_MSG_UNLOCK_REPLY &&
1468 	    test_and_clear_bit(DLM_IFL_OVERLAP_UNLOCK_BIT, &lkb->lkb_iflags)) {
1469 		log_debug(ls, "remwait %x unlock_reply overlap", lkb->lkb_id);
1470 		overlap_done = 1;
1471 		goto out_del;
1472 	}
1473 
1474 	if (mstype == DLM_MSG_CANCEL_REPLY &&
1475 	    test_and_clear_bit(DLM_IFL_OVERLAP_CANCEL_BIT, &lkb->lkb_iflags)) {
1476 		log_debug(ls, "remwait %x cancel_reply overlap", lkb->lkb_id);
1477 		overlap_done = 1;
1478 		goto out_del;
1479 	}
1480 
1481 	/* Cancel state was preemptively cleared by a successful convert,
1482 	   see next comment, nothing to do. */
1483 
1484 	if ((mstype == DLM_MSG_CANCEL_REPLY) &&
1485 	    (lkb->lkb_wait_type != DLM_MSG_CANCEL)) {
1486 		log_debug(ls, "remwait %x cancel_reply wait_type %d",
1487 			  lkb->lkb_id, lkb->lkb_wait_type);
1488 		return -1;
1489 	}
1490 
1491 	/* Remove for the convert reply, and premptively remove for the
1492 	   cancel reply.  A convert has been granted while there's still
1493 	   an outstanding cancel on it (the cancel is moot and the result
1494 	   in the cancel reply should be 0).  We preempt the cancel reply
1495 	   because the app gets the convert result and then can follow up
1496 	   with another op, like convert.  This subsequent op would see the
1497 	   lingering state of the cancel and fail with -EBUSY. */
1498 
1499 	if ((mstype == DLM_MSG_CONVERT_REPLY) &&
1500 	    (lkb->lkb_wait_type == DLM_MSG_CONVERT) && ms && !ms->m_result &&
1501 	    test_and_clear_bit(DLM_IFL_OVERLAP_CANCEL_BIT, &lkb->lkb_iflags)) {
1502 		log_debug(ls, "remwait %x convert_reply zap overlap_cancel",
1503 			  lkb->lkb_id);
1504 		lkb->lkb_wait_type = 0;
1505 		atomic_dec(&lkb->lkb_wait_count);
1506 		unhold_lkb(lkb);
1507 		goto out_del;
1508 	}
1509 
1510 	/* N.B. type of reply may not always correspond to type of original
1511 	   msg due to lookup->request optimization, verify others? */
1512 
1513 	if (lkb->lkb_wait_type) {
1514 		lkb->lkb_wait_type = 0;
1515 		goto out_del;
1516 	}
1517 
1518 	log_error(ls, "remwait error %x remote %d %x msg %d flags %x no wait",
1519 		  lkb->lkb_id, ms ? le32_to_cpu(ms->m_header.h_nodeid) : 0,
1520 		  lkb->lkb_remid, mstype, dlm_iflags_val(lkb));
1521 	return -1;
1522 
1523  out_del:
1524 	/* the force-unlock/cancel has completed and we haven't recvd a reply
1525 	   to the op that was in progress prior to the unlock/cancel; we
1526 	   give up on any reply to the earlier op.  FIXME: not sure when/how
1527 	   this would happen */
1528 
1529 	if (overlap_done && lkb->lkb_wait_type) {
1530 		log_error(ls, "remwait error %x reply %d wait_type %d overlap",
1531 			  lkb->lkb_id, mstype, lkb->lkb_wait_type);
1532 		atomic_dec(&lkb->lkb_wait_count);
1533 		unhold_lkb(lkb);
1534 		lkb->lkb_wait_type = 0;
1535 	}
1536 
1537 	DLM_ASSERT(atomic_read(&lkb->lkb_wait_count), dlm_print_lkb(lkb););
1538 
1539 	clear_bit(DLM_IFL_RESEND_BIT, &lkb->lkb_iflags);
1540 	if (atomic_dec_and_test(&lkb->lkb_wait_count))
1541 		list_del_init(&lkb->lkb_wait_reply);
1542 	unhold_lkb(lkb);
1543 	return 0;
1544 }
1545 
1546 static int remove_from_waiters(struct dlm_lkb *lkb, int mstype)
1547 {
1548 	struct dlm_ls *ls = lkb->lkb_resource->res_ls;
1549 	int error;
1550 
1551 	mutex_lock(&ls->ls_waiters_mutex);
1552 	error = _remove_from_waiters(lkb, mstype, NULL);
1553 	mutex_unlock(&ls->ls_waiters_mutex);
1554 	return error;
1555 }
1556 
1557 /* Handles situations where we might be processing a "fake" or "local" reply in
1558    which we can't try to take waiters_mutex again. */
1559 
1560 static int remove_from_waiters_ms(struct dlm_lkb *lkb,
1561 				  const struct dlm_message *ms, bool local)
1562 {
1563 	struct dlm_ls *ls = lkb->lkb_resource->res_ls;
1564 	int error;
1565 
1566 	if (!local)
1567 		mutex_lock(&ls->ls_waiters_mutex);
1568 	error = _remove_from_waiters(lkb, le32_to_cpu(ms->m_type), ms);
1569 	if (!local)
1570 		mutex_unlock(&ls->ls_waiters_mutex);
1571 	return error;
1572 }
1573 
1574 static void shrink_bucket(struct dlm_ls *ls, int b)
1575 {
1576 	struct rb_node *n, *next;
1577 	struct dlm_rsb *r;
1578 	char *name;
1579 	int our_nodeid = dlm_our_nodeid();
1580 	int remote_count = 0;
1581 	int need_shrink = 0;
1582 	int i, len, rv;
1583 
1584 	memset(&ls->ls_remove_lens, 0, sizeof(int) * DLM_REMOVE_NAMES_MAX);
1585 
1586 	spin_lock(&ls->ls_rsbtbl[b].lock);
1587 
1588 	if (!test_bit(DLM_RTF_SHRINK_BIT, &ls->ls_rsbtbl[b].flags)) {
1589 		spin_unlock(&ls->ls_rsbtbl[b].lock);
1590 		return;
1591 	}
1592 
1593 	for (n = rb_first(&ls->ls_rsbtbl[b].toss); n; n = next) {
1594 		next = rb_next(n);
1595 		r = rb_entry(n, struct dlm_rsb, res_hashnode);
1596 
1597 		/* If we're the directory record for this rsb, and
1598 		   we're not the master of it, then we need to wait
1599 		   for the master node to send us a dir remove for
1600 		   before removing the dir record. */
1601 
1602 		if (!dlm_no_directory(ls) &&
1603 		    (r->res_master_nodeid != our_nodeid) &&
1604 		    (dlm_dir_nodeid(r) == our_nodeid)) {
1605 			continue;
1606 		}
1607 
1608 		need_shrink = 1;
1609 
1610 		if (!time_after_eq(jiffies, r->res_toss_time +
1611 				   dlm_config.ci_toss_secs * HZ)) {
1612 			continue;
1613 		}
1614 
1615 		if (!dlm_no_directory(ls) &&
1616 		    (r->res_master_nodeid == our_nodeid) &&
1617 		    (dlm_dir_nodeid(r) != our_nodeid)) {
1618 
1619 			/* We're the master of this rsb but we're not
1620 			   the directory record, so we need to tell the
1621 			   dir node to remove the dir record. */
1622 
1623 			ls->ls_remove_lens[remote_count] = r->res_length;
1624 			memcpy(ls->ls_remove_names[remote_count], r->res_name,
1625 			       DLM_RESNAME_MAXLEN);
1626 			remote_count++;
1627 
1628 			if (remote_count >= DLM_REMOVE_NAMES_MAX)
1629 				break;
1630 			continue;
1631 		}
1632 
1633 		if (!kref_put(&r->res_ref, kill_rsb)) {
1634 			log_error(ls, "tossed rsb in use %s", r->res_name);
1635 			continue;
1636 		}
1637 
1638 		rb_erase(&r->res_hashnode, &ls->ls_rsbtbl[b].toss);
1639 		dlm_free_rsb(r);
1640 	}
1641 
1642 	if (need_shrink)
1643 		set_bit(DLM_RTF_SHRINK_BIT, &ls->ls_rsbtbl[b].flags);
1644 	else
1645 		clear_bit(DLM_RTF_SHRINK_BIT, &ls->ls_rsbtbl[b].flags);
1646 	spin_unlock(&ls->ls_rsbtbl[b].lock);
1647 
1648 	/*
1649 	 * While searching for rsb's to free, we found some that require
1650 	 * remote removal.  We leave them in place and find them again here
1651 	 * so there is a very small gap between removing them from the toss
1652 	 * list and sending the removal.  Keeping this gap small is
1653 	 * important to keep us (the master node) from being out of sync
1654 	 * with the remote dir node for very long.
1655 	 */
1656 
1657 	for (i = 0; i < remote_count; i++) {
1658 		name = ls->ls_remove_names[i];
1659 		len = ls->ls_remove_lens[i];
1660 
1661 		spin_lock(&ls->ls_rsbtbl[b].lock);
1662 		rv = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].toss, name, len, &r);
1663 		if (rv) {
1664 			spin_unlock(&ls->ls_rsbtbl[b].lock);
1665 			log_debug(ls, "remove_name not toss %s", name);
1666 			continue;
1667 		}
1668 
1669 		if (r->res_master_nodeid != our_nodeid) {
1670 			spin_unlock(&ls->ls_rsbtbl[b].lock);
1671 			log_debug(ls, "remove_name master %d dir %d our %d %s",
1672 				  r->res_master_nodeid, r->res_dir_nodeid,
1673 				  our_nodeid, name);
1674 			continue;
1675 		}
1676 
1677 		if (r->res_dir_nodeid == our_nodeid) {
1678 			/* should never happen */
1679 			spin_unlock(&ls->ls_rsbtbl[b].lock);
1680 			log_error(ls, "remove_name dir %d master %d our %d %s",
1681 				  r->res_dir_nodeid, r->res_master_nodeid,
1682 				  our_nodeid, name);
1683 			continue;
1684 		}
1685 
1686 		if (!time_after_eq(jiffies, r->res_toss_time +
1687 				   dlm_config.ci_toss_secs * HZ)) {
1688 			spin_unlock(&ls->ls_rsbtbl[b].lock);
1689 			log_debug(ls, "remove_name toss_time %lu now %lu %s",
1690 				  r->res_toss_time, jiffies, name);
1691 			continue;
1692 		}
1693 
1694 		if (!kref_put(&r->res_ref, kill_rsb)) {
1695 			spin_unlock(&ls->ls_rsbtbl[b].lock);
1696 			log_error(ls, "remove_name in use %s", name);
1697 			continue;
1698 		}
1699 
1700 		rb_erase(&r->res_hashnode, &ls->ls_rsbtbl[b].toss);
1701 		send_remove(r);
1702 		spin_unlock(&ls->ls_rsbtbl[b].lock);
1703 
1704 		dlm_free_rsb(r);
1705 	}
1706 }
1707 
1708 void dlm_scan_rsbs(struct dlm_ls *ls)
1709 {
1710 	int i;
1711 
1712 	for (i = 0; i < ls->ls_rsbtbl_size; i++) {
1713 		shrink_bucket(ls, i);
1714 		if (dlm_locking_stopped(ls))
1715 			break;
1716 		cond_resched();
1717 	}
1718 }
1719 
1720 /* lkb is master or local copy */
1721 
1722 static void set_lvb_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
1723 {
1724 	int b, len = r->res_ls->ls_lvblen;
1725 
1726 	/* b=1 lvb returned to caller
1727 	   b=0 lvb written to rsb or invalidated
1728 	   b=-1 do nothing */
1729 
1730 	b =  dlm_lvb_operations[lkb->lkb_grmode + 1][lkb->lkb_rqmode + 1];
1731 
1732 	if (b == 1) {
1733 		if (!lkb->lkb_lvbptr)
1734 			return;
1735 
1736 		if (!(lkb->lkb_exflags & DLM_LKF_VALBLK))
1737 			return;
1738 
1739 		if (!r->res_lvbptr)
1740 			return;
1741 
1742 		memcpy(lkb->lkb_lvbptr, r->res_lvbptr, len);
1743 		lkb->lkb_lvbseq = r->res_lvbseq;
1744 
1745 	} else if (b == 0) {
1746 		if (lkb->lkb_exflags & DLM_LKF_IVVALBLK) {
1747 			rsb_set_flag(r, RSB_VALNOTVALID);
1748 			return;
1749 		}
1750 
1751 		if (!lkb->lkb_lvbptr)
1752 			return;
1753 
1754 		if (!(lkb->lkb_exflags & DLM_LKF_VALBLK))
1755 			return;
1756 
1757 		if (!r->res_lvbptr)
1758 			r->res_lvbptr = dlm_allocate_lvb(r->res_ls);
1759 
1760 		if (!r->res_lvbptr)
1761 			return;
1762 
1763 		memcpy(r->res_lvbptr, lkb->lkb_lvbptr, len);
1764 		r->res_lvbseq++;
1765 		lkb->lkb_lvbseq = r->res_lvbseq;
1766 		rsb_clear_flag(r, RSB_VALNOTVALID);
1767 	}
1768 
1769 	if (rsb_flag(r, RSB_VALNOTVALID))
1770 		set_bit(DLM_SBF_VALNOTVALID_BIT, &lkb->lkb_sbflags);
1771 }
1772 
1773 static void set_lvb_unlock(struct dlm_rsb *r, struct dlm_lkb *lkb)
1774 {
1775 	if (lkb->lkb_grmode < DLM_LOCK_PW)
1776 		return;
1777 
1778 	if (lkb->lkb_exflags & DLM_LKF_IVVALBLK) {
1779 		rsb_set_flag(r, RSB_VALNOTVALID);
1780 		return;
1781 	}
1782 
1783 	if (!lkb->lkb_lvbptr)
1784 		return;
1785 
1786 	if (!(lkb->lkb_exflags & DLM_LKF_VALBLK))
1787 		return;
1788 
1789 	if (!r->res_lvbptr)
1790 		r->res_lvbptr = dlm_allocate_lvb(r->res_ls);
1791 
1792 	if (!r->res_lvbptr)
1793 		return;
1794 
1795 	memcpy(r->res_lvbptr, lkb->lkb_lvbptr, r->res_ls->ls_lvblen);
1796 	r->res_lvbseq++;
1797 	rsb_clear_flag(r, RSB_VALNOTVALID);
1798 }
1799 
1800 /* lkb is process copy (pc) */
1801 
1802 static void set_lvb_lock_pc(struct dlm_rsb *r, struct dlm_lkb *lkb,
1803 			    const struct dlm_message *ms)
1804 {
1805 	int b;
1806 
1807 	if (!lkb->lkb_lvbptr)
1808 		return;
1809 
1810 	if (!(lkb->lkb_exflags & DLM_LKF_VALBLK))
1811 		return;
1812 
1813 	b = dlm_lvb_operations[lkb->lkb_grmode + 1][lkb->lkb_rqmode + 1];
1814 	if (b == 1) {
1815 		int len = receive_extralen(ms);
1816 		if (len > r->res_ls->ls_lvblen)
1817 			len = r->res_ls->ls_lvblen;
1818 		memcpy(lkb->lkb_lvbptr, ms->m_extra, len);
1819 		lkb->lkb_lvbseq = le32_to_cpu(ms->m_lvbseq);
1820 	}
1821 }
1822 
1823 /* Manipulate lkb's on rsb's convert/granted/waiting queues
1824    remove_lock -- used for unlock, removes lkb from granted
1825    revert_lock -- used for cancel, moves lkb from convert to granted
1826    grant_lock  -- used for request and convert, adds lkb to granted or
1827                   moves lkb from convert or waiting to granted
1828 
1829    Each of these is used for master or local copy lkb's.  There is
1830    also a _pc() variation used to make the corresponding change on
1831    a process copy (pc) lkb. */
1832 
1833 static void _remove_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
1834 {
1835 	del_lkb(r, lkb);
1836 	lkb->lkb_grmode = DLM_LOCK_IV;
1837 	/* this unhold undoes the original ref from create_lkb()
1838 	   so this leads to the lkb being freed */
1839 	unhold_lkb(lkb);
1840 }
1841 
1842 static void remove_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
1843 {
1844 	set_lvb_unlock(r, lkb);
1845 	_remove_lock(r, lkb);
1846 }
1847 
1848 static void remove_lock_pc(struct dlm_rsb *r, struct dlm_lkb *lkb)
1849 {
1850 	_remove_lock(r, lkb);
1851 }
1852 
1853 /* returns: 0 did nothing
1854 	    1 moved lock to granted
1855 	   -1 removed lock */
1856 
1857 static int revert_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
1858 {
1859 	int rv = 0;
1860 
1861 	lkb->lkb_rqmode = DLM_LOCK_IV;
1862 
1863 	switch (lkb->lkb_status) {
1864 	case DLM_LKSTS_GRANTED:
1865 		break;
1866 	case DLM_LKSTS_CONVERT:
1867 		move_lkb(r, lkb, DLM_LKSTS_GRANTED);
1868 		rv = 1;
1869 		break;
1870 	case DLM_LKSTS_WAITING:
1871 		del_lkb(r, lkb);
1872 		lkb->lkb_grmode = DLM_LOCK_IV;
1873 		/* this unhold undoes the original ref from create_lkb()
1874 		   so this leads to the lkb being freed */
1875 		unhold_lkb(lkb);
1876 		rv = -1;
1877 		break;
1878 	default:
1879 		log_print("invalid status for revert %d", lkb->lkb_status);
1880 	}
1881 	return rv;
1882 }
1883 
1884 static int revert_lock_pc(struct dlm_rsb *r, struct dlm_lkb *lkb)
1885 {
1886 	return revert_lock(r, lkb);
1887 }
1888 
1889 static void _grant_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
1890 {
1891 	if (lkb->lkb_grmode != lkb->lkb_rqmode) {
1892 		lkb->lkb_grmode = lkb->lkb_rqmode;
1893 		if (lkb->lkb_status)
1894 			move_lkb(r, lkb, DLM_LKSTS_GRANTED);
1895 		else
1896 			add_lkb(r, lkb, DLM_LKSTS_GRANTED);
1897 	}
1898 
1899 	lkb->lkb_rqmode = DLM_LOCK_IV;
1900 	lkb->lkb_highbast = 0;
1901 }
1902 
1903 static void grant_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
1904 {
1905 	set_lvb_lock(r, lkb);
1906 	_grant_lock(r, lkb);
1907 }
1908 
1909 static void grant_lock_pc(struct dlm_rsb *r, struct dlm_lkb *lkb,
1910 			  const struct dlm_message *ms)
1911 {
1912 	set_lvb_lock_pc(r, lkb, ms);
1913 	_grant_lock(r, lkb);
1914 }
1915 
1916 /* called by grant_pending_locks() which means an async grant message must
1917    be sent to the requesting node in addition to granting the lock if the
1918    lkb belongs to a remote node. */
1919 
1920 static void grant_lock_pending(struct dlm_rsb *r, struct dlm_lkb *lkb)
1921 {
1922 	grant_lock(r, lkb);
1923 	if (is_master_copy(lkb))
1924 		send_grant(r, lkb);
1925 	else
1926 		queue_cast(r, lkb, 0);
1927 }
1928 
1929 /* The special CONVDEADLK, ALTPR and ALTCW flags allow the master to
1930    change the granted/requested modes.  We're munging things accordingly in
1931    the process copy.
1932    CONVDEADLK: our grmode may have been forced down to NL to resolve a
1933    conversion deadlock
1934    ALTPR/ALTCW: our rqmode may have been changed to PR or CW to become
1935    compatible with other granted locks */
1936 
1937 static void munge_demoted(struct dlm_lkb *lkb)
1938 {
1939 	if (lkb->lkb_rqmode == DLM_LOCK_IV || lkb->lkb_grmode == DLM_LOCK_IV) {
1940 		log_print("munge_demoted %x invalid modes gr %d rq %d",
1941 			  lkb->lkb_id, lkb->lkb_grmode, lkb->lkb_rqmode);
1942 		return;
1943 	}
1944 
1945 	lkb->lkb_grmode = DLM_LOCK_NL;
1946 }
1947 
1948 static void munge_altmode(struct dlm_lkb *lkb, const struct dlm_message *ms)
1949 {
1950 	if (ms->m_type != cpu_to_le32(DLM_MSG_REQUEST_REPLY) &&
1951 	    ms->m_type != cpu_to_le32(DLM_MSG_GRANT)) {
1952 		log_print("munge_altmode %x invalid reply type %d",
1953 			  lkb->lkb_id, le32_to_cpu(ms->m_type));
1954 		return;
1955 	}
1956 
1957 	if (lkb->lkb_exflags & DLM_LKF_ALTPR)
1958 		lkb->lkb_rqmode = DLM_LOCK_PR;
1959 	else if (lkb->lkb_exflags & DLM_LKF_ALTCW)
1960 		lkb->lkb_rqmode = DLM_LOCK_CW;
1961 	else {
1962 		log_print("munge_altmode invalid exflags %x", lkb->lkb_exflags);
1963 		dlm_print_lkb(lkb);
1964 	}
1965 }
1966 
1967 static inline int first_in_list(struct dlm_lkb *lkb, struct list_head *head)
1968 {
1969 	struct dlm_lkb *first = list_entry(head->next, struct dlm_lkb,
1970 					   lkb_statequeue);
1971 	if (lkb->lkb_id == first->lkb_id)
1972 		return 1;
1973 
1974 	return 0;
1975 }
1976 
1977 /* Check if the given lkb conflicts with another lkb on the queue. */
1978 
1979 static int queue_conflict(struct list_head *head, struct dlm_lkb *lkb)
1980 {
1981 	struct dlm_lkb *this;
1982 
1983 	list_for_each_entry(this, head, lkb_statequeue) {
1984 		if (this == lkb)
1985 			continue;
1986 		if (!modes_compat(this, lkb))
1987 			return 1;
1988 	}
1989 	return 0;
1990 }
1991 
1992 /*
1993  * "A conversion deadlock arises with a pair of lock requests in the converting
1994  * queue for one resource.  The granted mode of each lock blocks the requested
1995  * mode of the other lock."
1996  *
1997  * Part 2: if the granted mode of lkb is preventing an earlier lkb in the
1998  * convert queue from being granted, then deadlk/demote lkb.
1999  *
2000  * Example:
2001  * Granted Queue: empty
2002  * Convert Queue: NL->EX (first lock)
2003  *                PR->EX (second lock)
2004  *
2005  * The first lock can't be granted because of the granted mode of the second
2006  * lock and the second lock can't be granted because it's not first in the
2007  * list.  We either cancel lkb's conversion (PR->EX) and return EDEADLK, or we
2008  * demote the granted mode of lkb (from PR to NL) if it has the CONVDEADLK
2009  * flag set and return DEMOTED in the lksb flags.
2010  *
2011  * Originally, this function detected conv-deadlk in a more limited scope:
2012  * - if !modes_compat(lkb1, lkb2) && !modes_compat(lkb2, lkb1), or
2013  * - if lkb1 was the first entry in the queue (not just earlier), and was
2014  *   blocked by the granted mode of lkb2, and there was nothing on the
2015  *   granted queue preventing lkb1 from being granted immediately, i.e.
2016  *   lkb2 was the only thing preventing lkb1 from being granted.
2017  *
2018  * That second condition meant we'd only say there was conv-deadlk if
2019  * resolving it (by demotion) would lead to the first lock on the convert
2020  * queue being granted right away.  It allowed conversion deadlocks to exist
2021  * between locks on the convert queue while they couldn't be granted anyway.
2022  *
2023  * Now, we detect and take action on conversion deadlocks immediately when
2024  * they're created, even if they may not be immediately consequential.  If
2025  * lkb1 exists anywhere in the convert queue and lkb2 comes in with a granted
2026  * mode that would prevent lkb1's conversion from being granted, we do a
2027  * deadlk/demote on lkb2 right away and don't let it onto the convert queue.
2028  * I think this means that the lkb_is_ahead condition below should always
2029  * be zero, i.e. there will never be conv-deadlk between two locks that are
2030  * both already on the convert queue.
2031  */
2032 
2033 static int conversion_deadlock_detect(struct dlm_rsb *r, struct dlm_lkb *lkb2)
2034 {
2035 	struct dlm_lkb *lkb1;
2036 	int lkb_is_ahead = 0;
2037 
2038 	list_for_each_entry(lkb1, &r->res_convertqueue, lkb_statequeue) {
2039 		if (lkb1 == lkb2) {
2040 			lkb_is_ahead = 1;
2041 			continue;
2042 		}
2043 
2044 		if (!lkb_is_ahead) {
2045 			if (!modes_compat(lkb2, lkb1))
2046 				return 1;
2047 		} else {
2048 			if (!modes_compat(lkb2, lkb1) &&
2049 			    !modes_compat(lkb1, lkb2))
2050 				return 1;
2051 		}
2052 	}
2053 	return 0;
2054 }
2055 
2056 /*
2057  * Return 1 if the lock can be granted, 0 otherwise.
2058  * Also detect and resolve conversion deadlocks.
2059  *
2060  * lkb is the lock to be granted
2061  *
2062  * now is 1 if the function is being called in the context of the
2063  * immediate request, it is 0 if called later, after the lock has been
2064  * queued.
2065  *
2066  * recover is 1 if dlm_recover_grant() is trying to grant conversions
2067  * after recovery.
2068  *
2069  * References are from chapter 6 of "VAXcluster Principles" by Roy Davis
2070  */
2071 
2072 static int _can_be_granted(struct dlm_rsb *r, struct dlm_lkb *lkb, int now,
2073 			   int recover)
2074 {
2075 	int8_t conv = (lkb->lkb_grmode != DLM_LOCK_IV);
2076 
2077 	/*
2078 	 * 6-10: Version 5.4 introduced an option to address the phenomenon of
2079 	 * a new request for a NL mode lock being blocked.
2080 	 *
2081 	 * 6-11: If the optional EXPEDITE flag is used with the new NL mode
2082 	 * request, then it would be granted.  In essence, the use of this flag
2083 	 * tells the Lock Manager to expedite theis request by not considering
2084 	 * what may be in the CONVERTING or WAITING queues...  As of this
2085 	 * writing, the EXPEDITE flag can be used only with new requests for NL
2086 	 * mode locks.  This flag is not valid for conversion requests.
2087 	 *
2088 	 * A shortcut.  Earlier checks return an error if EXPEDITE is used in a
2089 	 * conversion or used with a non-NL requested mode.  We also know an
2090 	 * EXPEDITE request is always granted immediately, so now must always
2091 	 * be 1.  The full condition to grant an expedite request: (now &&
2092 	 * !conv && lkb->rqmode == DLM_LOCK_NL && (flags & EXPEDITE)) can
2093 	 * therefore be shortened to just checking the flag.
2094 	 */
2095 
2096 	if (lkb->lkb_exflags & DLM_LKF_EXPEDITE)
2097 		return 1;
2098 
2099 	/*
2100 	 * A shortcut. Without this, !queue_conflict(grantqueue, lkb) would be
2101 	 * added to the remaining conditions.
2102 	 */
2103 
2104 	if (queue_conflict(&r->res_grantqueue, lkb))
2105 		return 0;
2106 
2107 	/*
2108 	 * 6-3: By default, a conversion request is immediately granted if the
2109 	 * requested mode is compatible with the modes of all other granted
2110 	 * locks
2111 	 */
2112 
2113 	if (queue_conflict(&r->res_convertqueue, lkb))
2114 		return 0;
2115 
2116 	/*
2117 	 * The RECOVER_GRANT flag means dlm_recover_grant() is granting
2118 	 * locks for a recovered rsb, on which lkb's have been rebuilt.
2119 	 * The lkb's may have been rebuilt on the queues in a different
2120 	 * order than they were in on the previous master.  So, granting
2121 	 * queued conversions in order after recovery doesn't make sense
2122 	 * since the order hasn't been preserved anyway.  The new order
2123 	 * could also have created a new "in place" conversion deadlock.
2124 	 * (e.g. old, failed master held granted EX, with PR->EX, NL->EX.
2125 	 * After recovery, there would be no granted locks, and possibly
2126 	 * NL->EX, PR->EX, an in-place conversion deadlock.)  So, after
2127 	 * recovery, grant conversions without considering order.
2128 	 */
2129 
2130 	if (conv && recover)
2131 		return 1;
2132 
2133 	/*
2134 	 * 6-5: But the default algorithm for deciding whether to grant or
2135 	 * queue conversion requests does not by itself guarantee that such
2136 	 * requests are serviced on a "first come first serve" basis.  This, in
2137 	 * turn, can lead to a phenomenon known as "indefinate postponement".
2138 	 *
2139 	 * 6-7: This issue is dealt with by using the optional QUECVT flag with
2140 	 * the system service employed to request a lock conversion.  This flag
2141 	 * forces certain conversion requests to be queued, even if they are
2142 	 * compatible with the granted modes of other locks on the same
2143 	 * resource.  Thus, the use of this flag results in conversion requests
2144 	 * being ordered on a "first come first servce" basis.
2145 	 *
2146 	 * DCT: This condition is all about new conversions being able to occur
2147 	 * "in place" while the lock remains on the granted queue (assuming
2148 	 * nothing else conflicts.)  IOW if QUECVT isn't set, a conversion
2149 	 * doesn't _have_ to go onto the convert queue where it's processed in
2150 	 * order.  The "now" variable is necessary to distinguish converts
2151 	 * being received and processed for the first time now, because once a
2152 	 * convert is moved to the conversion queue the condition below applies
2153 	 * requiring fifo granting.
2154 	 */
2155 
2156 	if (now && conv && !(lkb->lkb_exflags & DLM_LKF_QUECVT))
2157 		return 1;
2158 
2159 	/*
2160 	 * Even if the convert is compat with all granted locks,
2161 	 * QUECVT forces it behind other locks on the convert queue.
2162 	 */
2163 
2164 	if (now && conv && (lkb->lkb_exflags & DLM_LKF_QUECVT)) {
2165 		if (list_empty(&r->res_convertqueue))
2166 			return 1;
2167 		else
2168 			return 0;
2169 	}
2170 
2171 	/*
2172 	 * The NOORDER flag is set to avoid the standard vms rules on grant
2173 	 * order.
2174 	 */
2175 
2176 	if (lkb->lkb_exflags & DLM_LKF_NOORDER)
2177 		return 1;
2178 
2179 	/*
2180 	 * 6-3: Once in that queue [CONVERTING], a conversion request cannot be
2181 	 * granted until all other conversion requests ahead of it are granted
2182 	 * and/or canceled.
2183 	 */
2184 
2185 	if (!now && conv && first_in_list(lkb, &r->res_convertqueue))
2186 		return 1;
2187 
2188 	/*
2189 	 * 6-4: By default, a new request is immediately granted only if all
2190 	 * three of the following conditions are satisfied when the request is
2191 	 * issued:
2192 	 * - The queue of ungranted conversion requests for the resource is
2193 	 *   empty.
2194 	 * - The queue of ungranted new requests for the resource is empty.
2195 	 * - The mode of the new request is compatible with the most
2196 	 *   restrictive mode of all granted locks on the resource.
2197 	 */
2198 
2199 	if (now && !conv && list_empty(&r->res_convertqueue) &&
2200 	    list_empty(&r->res_waitqueue))
2201 		return 1;
2202 
2203 	/*
2204 	 * 6-4: Once a lock request is in the queue of ungranted new requests,
2205 	 * it cannot be granted until the queue of ungranted conversion
2206 	 * requests is empty, all ungranted new requests ahead of it are
2207 	 * granted and/or canceled, and it is compatible with the granted mode
2208 	 * of the most restrictive lock granted on the resource.
2209 	 */
2210 
2211 	if (!now && !conv && list_empty(&r->res_convertqueue) &&
2212 	    first_in_list(lkb, &r->res_waitqueue))
2213 		return 1;
2214 
2215 	return 0;
2216 }
2217 
2218 static int can_be_granted(struct dlm_rsb *r, struct dlm_lkb *lkb, int now,
2219 			  int recover, int *err)
2220 {
2221 	int rv;
2222 	int8_t alt = 0, rqmode = lkb->lkb_rqmode;
2223 	int8_t is_convert = (lkb->lkb_grmode != DLM_LOCK_IV);
2224 
2225 	if (err)
2226 		*err = 0;
2227 
2228 	rv = _can_be_granted(r, lkb, now, recover);
2229 	if (rv)
2230 		goto out;
2231 
2232 	/*
2233 	 * The CONVDEADLK flag is non-standard and tells the dlm to resolve
2234 	 * conversion deadlocks by demoting grmode to NL, otherwise the dlm
2235 	 * cancels one of the locks.
2236 	 */
2237 
2238 	if (is_convert && can_be_queued(lkb) &&
2239 	    conversion_deadlock_detect(r, lkb)) {
2240 		if (lkb->lkb_exflags & DLM_LKF_CONVDEADLK) {
2241 			lkb->lkb_grmode = DLM_LOCK_NL;
2242 			set_bit(DLM_SBF_DEMOTED_BIT, &lkb->lkb_sbflags);
2243 		} else if (err) {
2244 			*err = -EDEADLK;
2245 		} else {
2246 			log_print("can_be_granted deadlock %x now %d",
2247 				  lkb->lkb_id, now);
2248 			dlm_dump_rsb(r);
2249 		}
2250 		goto out;
2251 	}
2252 
2253 	/*
2254 	 * The ALTPR and ALTCW flags are non-standard and tell the dlm to try
2255 	 * to grant a request in a mode other than the normal rqmode.  It's a
2256 	 * simple way to provide a big optimization to applications that can
2257 	 * use them.
2258 	 */
2259 
2260 	if (rqmode != DLM_LOCK_PR && (lkb->lkb_exflags & DLM_LKF_ALTPR))
2261 		alt = DLM_LOCK_PR;
2262 	else if (rqmode != DLM_LOCK_CW && (lkb->lkb_exflags & DLM_LKF_ALTCW))
2263 		alt = DLM_LOCK_CW;
2264 
2265 	if (alt) {
2266 		lkb->lkb_rqmode = alt;
2267 		rv = _can_be_granted(r, lkb, now, 0);
2268 		if (rv)
2269 			set_bit(DLM_SBF_ALTMODE_BIT, &lkb->lkb_sbflags);
2270 		else
2271 			lkb->lkb_rqmode = rqmode;
2272 	}
2273  out:
2274 	return rv;
2275 }
2276 
2277 /* Returns the highest requested mode of all blocked conversions; sets
2278    cw if there's a blocked conversion to DLM_LOCK_CW. */
2279 
2280 static int grant_pending_convert(struct dlm_rsb *r, int high, int *cw,
2281 				 unsigned int *count)
2282 {
2283 	struct dlm_lkb *lkb, *s;
2284 	int recover = rsb_flag(r, RSB_RECOVER_GRANT);
2285 	int hi, demoted, quit, grant_restart, demote_restart;
2286 	int deadlk;
2287 
2288 	quit = 0;
2289  restart:
2290 	grant_restart = 0;
2291 	demote_restart = 0;
2292 	hi = DLM_LOCK_IV;
2293 
2294 	list_for_each_entry_safe(lkb, s, &r->res_convertqueue, lkb_statequeue) {
2295 		demoted = is_demoted(lkb);
2296 		deadlk = 0;
2297 
2298 		if (can_be_granted(r, lkb, 0, recover, &deadlk)) {
2299 			grant_lock_pending(r, lkb);
2300 			grant_restart = 1;
2301 			if (count)
2302 				(*count)++;
2303 			continue;
2304 		}
2305 
2306 		if (!demoted && is_demoted(lkb)) {
2307 			log_print("WARN: pending demoted %x node %d %s",
2308 				  lkb->lkb_id, lkb->lkb_nodeid, r->res_name);
2309 			demote_restart = 1;
2310 			continue;
2311 		}
2312 
2313 		if (deadlk) {
2314 			/*
2315 			 * If DLM_LKB_NODLKWT flag is set and conversion
2316 			 * deadlock is detected, we request blocking AST and
2317 			 * down (or cancel) conversion.
2318 			 */
2319 			if (lkb->lkb_exflags & DLM_LKF_NODLCKWT) {
2320 				if (lkb->lkb_highbast < lkb->lkb_rqmode) {
2321 					queue_bast(r, lkb, lkb->lkb_rqmode);
2322 					lkb->lkb_highbast = lkb->lkb_rqmode;
2323 				}
2324 			} else {
2325 				log_print("WARN: pending deadlock %x node %d %s",
2326 					  lkb->lkb_id, lkb->lkb_nodeid,
2327 					  r->res_name);
2328 				dlm_dump_rsb(r);
2329 			}
2330 			continue;
2331 		}
2332 
2333 		hi = max_t(int, lkb->lkb_rqmode, hi);
2334 
2335 		if (cw && lkb->lkb_rqmode == DLM_LOCK_CW)
2336 			*cw = 1;
2337 	}
2338 
2339 	if (grant_restart)
2340 		goto restart;
2341 	if (demote_restart && !quit) {
2342 		quit = 1;
2343 		goto restart;
2344 	}
2345 
2346 	return max_t(int, high, hi);
2347 }
2348 
2349 static int grant_pending_wait(struct dlm_rsb *r, int high, int *cw,
2350 			      unsigned int *count)
2351 {
2352 	struct dlm_lkb *lkb, *s;
2353 
2354 	list_for_each_entry_safe(lkb, s, &r->res_waitqueue, lkb_statequeue) {
2355 		if (can_be_granted(r, lkb, 0, 0, NULL)) {
2356 			grant_lock_pending(r, lkb);
2357 			if (count)
2358 				(*count)++;
2359 		} else {
2360 			high = max_t(int, lkb->lkb_rqmode, high);
2361 			if (lkb->lkb_rqmode == DLM_LOCK_CW)
2362 				*cw = 1;
2363 		}
2364 	}
2365 
2366 	return high;
2367 }
2368 
2369 /* cw of 1 means there's a lock with a rqmode of DLM_LOCK_CW that's blocked
2370    on either the convert or waiting queue.
2371    high is the largest rqmode of all locks blocked on the convert or
2372    waiting queue. */
2373 
2374 static int lock_requires_bast(struct dlm_lkb *gr, int high, int cw)
2375 {
2376 	if (gr->lkb_grmode == DLM_LOCK_PR && cw) {
2377 		if (gr->lkb_highbast < DLM_LOCK_EX)
2378 			return 1;
2379 		return 0;
2380 	}
2381 
2382 	if (gr->lkb_highbast < high &&
2383 	    !__dlm_compat_matrix[gr->lkb_grmode+1][high+1])
2384 		return 1;
2385 	return 0;
2386 }
2387 
2388 static void grant_pending_locks(struct dlm_rsb *r, unsigned int *count)
2389 {
2390 	struct dlm_lkb *lkb, *s;
2391 	int high = DLM_LOCK_IV;
2392 	int cw = 0;
2393 
2394 	if (!is_master(r)) {
2395 		log_print("grant_pending_locks r nodeid %d", r->res_nodeid);
2396 		dlm_dump_rsb(r);
2397 		return;
2398 	}
2399 
2400 	high = grant_pending_convert(r, high, &cw, count);
2401 	high = grant_pending_wait(r, high, &cw, count);
2402 
2403 	if (high == DLM_LOCK_IV)
2404 		return;
2405 
2406 	/*
2407 	 * If there are locks left on the wait/convert queue then send blocking
2408 	 * ASTs to granted locks based on the largest requested mode (high)
2409 	 * found above.
2410 	 */
2411 
2412 	list_for_each_entry_safe(lkb, s, &r->res_grantqueue, lkb_statequeue) {
2413 		if (lkb->lkb_bastfn && lock_requires_bast(lkb, high, cw)) {
2414 			if (cw && high == DLM_LOCK_PR &&
2415 			    lkb->lkb_grmode == DLM_LOCK_PR)
2416 				queue_bast(r, lkb, DLM_LOCK_CW);
2417 			else
2418 				queue_bast(r, lkb, high);
2419 			lkb->lkb_highbast = high;
2420 		}
2421 	}
2422 }
2423 
2424 static int modes_require_bast(struct dlm_lkb *gr, struct dlm_lkb *rq)
2425 {
2426 	if ((gr->lkb_grmode == DLM_LOCK_PR && rq->lkb_rqmode == DLM_LOCK_CW) ||
2427 	    (gr->lkb_grmode == DLM_LOCK_CW && rq->lkb_rqmode == DLM_LOCK_PR)) {
2428 		if (gr->lkb_highbast < DLM_LOCK_EX)
2429 			return 1;
2430 		return 0;
2431 	}
2432 
2433 	if (gr->lkb_highbast < rq->lkb_rqmode && !modes_compat(gr, rq))
2434 		return 1;
2435 	return 0;
2436 }
2437 
2438 static void send_bast_queue(struct dlm_rsb *r, struct list_head *head,
2439 			    struct dlm_lkb *lkb)
2440 {
2441 	struct dlm_lkb *gr;
2442 
2443 	list_for_each_entry(gr, head, lkb_statequeue) {
2444 		/* skip self when sending basts to convertqueue */
2445 		if (gr == lkb)
2446 			continue;
2447 		if (gr->lkb_bastfn && modes_require_bast(gr, lkb)) {
2448 			queue_bast(r, gr, lkb->lkb_rqmode);
2449 			gr->lkb_highbast = lkb->lkb_rqmode;
2450 		}
2451 	}
2452 }
2453 
2454 static void send_blocking_asts(struct dlm_rsb *r, struct dlm_lkb *lkb)
2455 {
2456 	send_bast_queue(r, &r->res_grantqueue, lkb);
2457 }
2458 
2459 static void send_blocking_asts_all(struct dlm_rsb *r, struct dlm_lkb *lkb)
2460 {
2461 	send_bast_queue(r, &r->res_grantqueue, lkb);
2462 	send_bast_queue(r, &r->res_convertqueue, lkb);
2463 }
2464 
2465 /* set_master(r, lkb) -- set the master nodeid of a resource
2466 
2467    The purpose of this function is to set the nodeid field in the given
2468    lkb using the nodeid field in the given rsb.  If the rsb's nodeid is
2469    known, it can just be copied to the lkb and the function will return
2470    0.  If the rsb's nodeid is _not_ known, it needs to be looked up
2471    before it can be copied to the lkb.
2472 
2473    When the rsb nodeid is being looked up remotely, the initial lkb
2474    causing the lookup is kept on the ls_waiters list waiting for the
2475    lookup reply.  Other lkb's waiting for the same rsb lookup are kept
2476    on the rsb's res_lookup list until the master is verified.
2477 
2478    Return values:
2479    0: nodeid is set in rsb/lkb and the caller should go ahead and use it
2480    1: the rsb master is not available and the lkb has been placed on
2481       a wait queue
2482 */
2483 
2484 static int set_master(struct dlm_rsb *r, struct dlm_lkb *lkb)
2485 {
2486 	int our_nodeid = dlm_our_nodeid();
2487 
2488 	if (rsb_flag(r, RSB_MASTER_UNCERTAIN)) {
2489 		rsb_clear_flag(r, RSB_MASTER_UNCERTAIN);
2490 		r->res_first_lkid = lkb->lkb_id;
2491 		lkb->lkb_nodeid = r->res_nodeid;
2492 		return 0;
2493 	}
2494 
2495 	if (r->res_first_lkid && r->res_first_lkid != lkb->lkb_id) {
2496 		list_add_tail(&lkb->lkb_rsb_lookup, &r->res_lookup);
2497 		return 1;
2498 	}
2499 
2500 	if (r->res_master_nodeid == our_nodeid) {
2501 		lkb->lkb_nodeid = 0;
2502 		return 0;
2503 	}
2504 
2505 	if (r->res_master_nodeid) {
2506 		lkb->lkb_nodeid = r->res_master_nodeid;
2507 		return 0;
2508 	}
2509 
2510 	if (dlm_dir_nodeid(r) == our_nodeid) {
2511 		/* This is a somewhat unusual case; find_rsb will usually
2512 		   have set res_master_nodeid when dir nodeid is local, but
2513 		   there are cases where we become the dir node after we've
2514 		   past find_rsb and go through _request_lock again.
2515 		   confirm_master() or process_lookup_list() needs to be
2516 		   called after this. */
2517 		log_debug(r->res_ls, "set_master %x self master %d dir %d %s",
2518 			  lkb->lkb_id, r->res_master_nodeid, r->res_dir_nodeid,
2519 			  r->res_name);
2520 		r->res_master_nodeid = our_nodeid;
2521 		r->res_nodeid = 0;
2522 		lkb->lkb_nodeid = 0;
2523 		return 0;
2524 	}
2525 
2526 	r->res_first_lkid = lkb->lkb_id;
2527 	send_lookup(r, lkb);
2528 	return 1;
2529 }
2530 
2531 static void process_lookup_list(struct dlm_rsb *r)
2532 {
2533 	struct dlm_lkb *lkb, *safe;
2534 
2535 	list_for_each_entry_safe(lkb, safe, &r->res_lookup, lkb_rsb_lookup) {
2536 		list_del_init(&lkb->lkb_rsb_lookup);
2537 		_request_lock(r, lkb);
2538 		schedule();
2539 	}
2540 }
2541 
2542 /* confirm_master -- confirm (or deny) an rsb's master nodeid */
2543 
2544 static void confirm_master(struct dlm_rsb *r, int error)
2545 {
2546 	struct dlm_lkb *lkb;
2547 
2548 	if (!r->res_first_lkid)
2549 		return;
2550 
2551 	switch (error) {
2552 	case 0:
2553 	case -EINPROGRESS:
2554 		r->res_first_lkid = 0;
2555 		process_lookup_list(r);
2556 		break;
2557 
2558 	case -EAGAIN:
2559 	case -EBADR:
2560 	case -ENOTBLK:
2561 		/* the remote request failed and won't be retried (it was
2562 		   a NOQUEUE, or has been canceled/unlocked); make a waiting
2563 		   lkb the first_lkid */
2564 
2565 		r->res_first_lkid = 0;
2566 
2567 		if (!list_empty(&r->res_lookup)) {
2568 			lkb = list_entry(r->res_lookup.next, struct dlm_lkb,
2569 					 lkb_rsb_lookup);
2570 			list_del_init(&lkb->lkb_rsb_lookup);
2571 			r->res_first_lkid = lkb->lkb_id;
2572 			_request_lock(r, lkb);
2573 		}
2574 		break;
2575 
2576 	default:
2577 		log_error(r->res_ls, "confirm_master unknown error %d", error);
2578 	}
2579 }
2580 
2581 static int set_lock_args(int mode, struct dlm_lksb *lksb, uint32_t flags,
2582 			 int namelen, void (*ast)(void *astparam),
2583 			 void *astparam,
2584 			 void (*bast)(void *astparam, int mode),
2585 			 struct dlm_args *args)
2586 {
2587 	int rv = -EINVAL;
2588 
2589 	/* check for invalid arg usage */
2590 
2591 	if (mode < 0 || mode > DLM_LOCK_EX)
2592 		goto out;
2593 
2594 	if (!(flags & DLM_LKF_CONVERT) && (namelen > DLM_RESNAME_MAXLEN))
2595 		goto out;
2596 
2597 	if (flags & DLM_LKF_CANCEL)
2598 		goto out;
2599 
2600 	if (flags & DLM_LKF_QUECVT && !(flags & DLM_LKF_CONVERT))
2601 		goto out;
2602 
2603 	if (flags & DLM_LKF_CONVDEADLK && !(flags & DLM_LKF_CONVERT))
2604 		goto out;
2605 
2606 	if (flags & DLM_LKF_CONVDEADLK && flags & DLM_LKF_NOQUEUE)
2607 		goto out;
2608 
2609 	if (flags & DLM_LKF_EXPEDITE && flags & DLM_LKF_CONVERT)
2610 		goto out;
2611 
2612 	if (flags & DLM_LKF_EXPEDITE && flags & DLM_LKF_QUECVT)
2613 		goto out;
2614 
2615 	if (flags & DLM_LKF_EXPEDITE && flags & DLM_LKF_NOQUEUE)
2616 		goto out;
2617 
2618 	if (flags & DLM_LKF_EXPEDITE && mode != DLM_LOCK_NL)
2619 		goto out;
2620 
2621 	if (!ast || !lksb)
2622 		goto out;
2623 
2624 	if (flags & DLM_LKF_VALBLK && !lksb->sb_lvbptr)
2625 		goto out;
2626 
2627 	if (flags & DLM_LKF_CONVERT && !lksb->sb_lkid)
2628 		goto out;
2629 
2630 	/* these args will be copied to the lkb in validate_lock_args,
2631 	   it cannot be done now because when converting locks, fields in
2632 	   an active lkb cannot be modified before locking the rsb */
2633 
2634 	args->flags = flags;
2635 	args->astfn = ast;
2636 	args->astparam = astparam;
2637 	args->bastfn = bast;
2638 	args->mode = mode;
2639 	args->lksb = lksb;
2640 	rv = 0;
2641  out:
2642 	return rv;
2643 }
2644 
2645 static int set_unlock_args(uint32_t flags, void *astarg, struct dlm_args *args)
2646 {
2647 	if (flags & ~(DLM_LKF_CANCEL | DLM_LKF_VALBLK | DLM_LKF_IVVALBLK |
2648  		      DLM_LKF_FORCEUNLOCK))
2649 		return -EINVAL;
2650 
2651 	if (flags & DLM_LKF_CANCEL && flags & DLM_LKF_FORCEUNLOCK)
2652 		return -EINVAL;
2653 
2654 	args->flags = flags;
2655 	args->astparam = astarg;
2656 	return 0;
2657 }
2658 
2659 static int validate_lock_args(struct dlm_ls *ls, struct dlm_lkb *lkb,
2660 			      struct dlm_args *args)
2661 {
2662 	int rv = -EBUSY;
2663 
2664 	if (args->flags & DLM_LKF_CONVERT) {
2665 		if (lkb->lkb_status != DLM_LKSTS_GRANTED)
2666 			goto out;
2667 
2668 		/* lock not allowed if there's any op in progress */
2669 		if (lkb->lkb_wait_type || atomic_read(&lkb->lkb_wait_count))
2670 			goto out;
2671 
2672 		if (is_overlap(lkb))
2673 			goto out;
2674 
2675 		rv = -EINVAL;
2676 		if (test_bit(DLM_IFL_MSTCPY_BIT, &lkb->lkb_iflags))
2677 			goto out;
2678 
2679 		if (args->flags & DLM_LKF_QUECVT &&
2680 		    !__quecvt_compat_matrix[lkb->lkb_grmode+1][args->mode+1])
2681 			goto out;
2682 	}
2683 
2684 	lkb->lkb_exflags = args->flags;
2685 	dlm_set_sbflags_val(lkb, 0);
2686 	lkb->lkb_astfn = args->astfn;
2687 	lkb->lkb_astparam = args->astparam;
2688 	lkb->lkb_bastfn = args->bastfn;
2689 	lkb->lkb_rqmode = args->mode;
2690 	lkb->lkb_lksb = args->lksb;
2691 	lkb->lkb_lvbptr = args->lksb->sb_lvbptr;
2692 	lkb->lkb_ownpid = (int) current->pid;
2693 	rv = 0;
2694  out:
2695 	switch (rv) {
2696 	case 0:
2697 		break;
2698 	case -EINVAL:
2699 		/* annoy the user because dlm usage is wrong */
2700 		WARN_ON(1);
2701 		log_error(ls, "%s %d %x %x %x %d %d %s", __func__,
2702 			  rv, lkb->lkb_id, dlm_iflags_val(lkb), args->flags,
2703 			  lkb->lkb_status, lkb->lkb_wait_type,
2704 			  lkb->lkb_resource->res_name);
2705 		break;
2706 	default:
2707 		log_debug(ls, "%s %d %x %x %x %d %d %s", __func__,
2708 			  rv, lkb->lkb_id, dlm_iflags_val(lkb), args->flags,
2709 			  lkb->lkb_status, lkb->lkb_wait_type,
2710 			  lkb->lkb_resource->res_name);
2711 		break;
2712 	}
2713 
2714 	return rv;
2715 }
2716 
2717 /* when dlm_unlock() sees -EBUSY with CANCEL/FORCEUNLOCK it returns 0
2718    for success */
2719 
2720 /* note: it's valid for lkb_nodeid/res_nodeid to be -1 when we get here
2721    because there may be a lookup in progress and it's valid to do
2722    cancel/unlockf on it */
2723 
2724 static int validate_unlock_args(struct dlm_lkb *lkb, struct dlm_args *args)
2725 {
2726 	struct dlm_ls *ls = lkb->lkb_resource->res_ls;
2727 	int rv = -EBUSY;
2728 
2729 	/* normal unlock not allowed if there's any op in progress */
2730 	if (!(args->flags & (DLM_LKF_CANCEL | DLM_LKF_FORCEUNLOCK)) &&
2731 	    (lkb->lkb_wait_type || atomic_read(&lkb->lkb_wait_count)))
2732 		goto out;
2733 
2734 	/* an lkb may be waiting for an rsb lookup to complete where the
2735 	   lookup was initiated by another lock */
2736 
2737 	if (!list_empty(&lkb->lkb_rsb_lookup)) {
2738 		if (args->flags & (DLM_LKF_CANCEL | DLM_LKF_FORCEUNLOCK)) {
2739 			log_debug(ls, "unlock on rsb_lookup %x", lkb->lkb_id);
2740 			list_del_init(&lkb->lkb_rsb_lookup);
2741 			queue_cast(lkb->lkb_resource, lkb,
2742 				   args->flags & DLM_LKF_CANCEL ?
2743 				   -DLM_ECANCEL : -DLM_EUNLOCK);
2744 			unhold_lkb(lkb); /* undoes create_lkb() */
2745 		}
2746 		/* caller changes -EBUSY to 0 for CANCEL and FORCEUNLOCK */
2747 		goto out;
2748 	}
2749 
2750 	rv = -EINVAL;
2751 	if (test_bit(DLM_IFL_MSTCPY_BIT, &lkb->lkb_iflags)) {
2752 		log_error(ls, "unlock on MSTCPY %x", lkb->lkb_id);
2753 		dlm_print_lkb(lkb);
2754 		goto out;
2755 	}
2756 
2757 	/* an lkb may still exist even though the lock is EOL'ed due to a
2758 	 * cancel, unlock or failed noqueue request; an app can't use these
2759 	 * locks; return same error as if the lkid had not been found at all
2760 	 */
2761 
2762 	if (test_bit(DLM_IFL_ENDOFLIFE_BIT, &lkb->lkb_iflags)) {
2763 		log_debug(ls, "unlock on ENDOFLIFE %x", lkb->lkb_id);
2764 		rv = -ENOENT;
2765 		goto out;
2766 	}
2767 
2768 	/* cancel not allowed with another cancel/unlock in progress */
2769 
2770 	if (args->flags & DLM_LKF_CANCEL) {
2771 		if (lkb->lkb_exflags & DLM_LKF_CANCEL)
2772 			goto out;
2773 
2774 		if (is_overlap(lkb))
2775 			goto out;
2776 
2777 		if (test_bit(DLM_IFL_RESEND_BIT, &lkb->lkb_iflags)) {
2778 			set_bit(DLM_IFL_OVERLAP_CANCEL_BIT, &lkb->lkb_iflags);
2779 			rv = -EBUSY;
2780 			goto out;
2781 		}
2782 
2783 		/* there's nothing to cancel */
2784 		if (lkb->lkb_status == DLM_LKSTS_GRANTED &&
2785 		    !lkb->lkb_wait_type) {
2786 			rv = -EBUSY;
2787 			goto out;
2788 		}
2789 
2790 		switch (lkb->lkb_wait_type) {
2791 		case DLM_MSG_LOOKUP:
2792 		case DLM_MSG_REQUEST:
2793 			set_bit(DLM_IFL_OVERLAP_CANCEL_BIT, &lkb->lkb_iflags);
2794 			rv = -EBUSY;
2795 			goto out;
2796 		case DLM_MSG_UNLOCK:
2797 		case DLM_MSG_CANCEL:
2798 			goto out;
2799 		}
2800 		/* add_to_waiters() will set OVERLAP_CANCEL */
2801 		goto out_ok;
2802 	}
2803 
2804 	/* do we need to allow a force-unlock if there's a normal unlock
2805 	   already in progress?  in what conditions could the normal unlock
2806 	   fail such that we'd want to send a force-unlock to be sure? */
2807 
2808 	if (args->flags & DLM_LKF_FORCEUNLOCK) {
2809 		if (lkb->lkb_exflags & DLM_LKF_FORCEUNLOCK)
2810 			goto out;
2811 
2812 		if (is_overlap_unlock(lkb))
2813 			goto out;
2814 
2815 		if (test_bit(DLM_IFL_RESEND_BIT, &lkb->lkb_iflags)) {
2816 			set_bit(DLM_IFL_OVERLAP_UNLOCK_BIT, &lkb->lkb_iflags);
2817 			rv = -EBUSY;
2818 			goto out;
2819 		}
2820 
2821 		switch (lkb->lkb_wait_type) {
2822 		case DLM_MSG_LOOKUP:
2823 		case DLM_MSG_REQUEST:
2824 			set_bit(DLM_IFL_OVERLAP_UNLOCK_BIT, &lkb->lkb_iflags);
2825 			rv = -EBUSY;
2826 			goto out;
2827 		case DLM_MSG_UNLOCK:
2828 			goto out;
2829 		}
2830 		/* add_to_waiters() will set OVERLAP_UNLOCK */
2831 	}
2832 
2833  out_ok:
2834 	/* an overlapping op shouldn't blow away exflags from other op */
2835 	lkb->lkb_exflags |= args->flags;
2836 	dlm_set_sbflags_val(lkb, 0);
2837 	lkb->lkb_astparam = args->astparam;
2838 	rv = 0;
2839  out:
2840 	switch (rv) {
2841 	case 0:
2842 		break;
2843 	case -EINVAL:
2844 		/* annoy the user because dlm usage is wrong */
2845 		WARN_ON(1);
2846 		log_error(ls, "%s %d %x %x %x %x %d %s", __func__, rv,
2847 			  lkb->lkb_id, dlm_iflags_val(lkb), lkb->lkb_exflags,
2848 			  args->flags, lkb->lkb_wait_type,
2849 			  lkb->lkb_resource->res_name);
2850 		break;
2851 	default:
2852 		log_debug(ls, "%s %d %x %x %x %x %d %s", __func__, rv,
2853 			  lkb->lkb_id, dlm_iflags_val(lkb), lkb->lkb_exflags,
2854 			  args->flags, lkb->lkb_wait_type,
2855 			  lkb->lkb_resource->res_name);
2856 		break;
2857 	}
2858 
2859 	return rv;
2860 }
2861 
2862 /*
2863  * Four stage 4 varieties:
2864  * do_request(), do_convert(), do_unlock(), do_cancel()
2865  * These are called on the master node for the given lock and
2866  * from the central locking logic.
2867  */
2868 
2869 static int do_request(struct dlm_rsb *r, struct dlm_lkb *lkb)
2870 {
2871 	int error = 0;
2872 
2873 	if (can_be_granted(r, lkb, 1, 0, NULL)) {
2874 		grant_lock(r, lkb);
2875 		queue_cast(r, lkb, 0);
2876 		goto out;
2877 	}
2878 
2879 	if (can_be_queued(lkb)) {
2880 		error = -EINPROGRESS;
2881 		add_lkb(r, lkb, DLM_LKSTS_WAITING);
2882 		goto out;
2883 	}
2884 
2885 	error = -EAGAIN;
2886 	queue_cast(r, lkb, -EAGAIN);
2887  out:
2888 	return error;
2889 }
2890 
2891 static void do_request_effects(struct dlm_rsb *r, struct dlm_lkb *lkb,
2892 			       int error)
2893 {
2894 	switch (error) {
2895 	case -EAGAIN:
2896 		if (force_blocking_asts(lkb))
2897 			send_blocking_asts_all(r, lkb);
2898 		break;
2899 	case -EINPROGRESS:
2900 		send_blocking_asts(r, lkb);
2901 		break;
2902 	}
2903 }
2904 
2905 static int do_convert(struct dlm_rsb *r, struct dlm_lkb *lkb)
2906 {
2907 	int error = 0;
2908 	int deadlk = 0;
2909 
2910 	/* changing an existing lock may allow others to be granted */
2911 
2912 	if (can_be_granted(r, lkb, 1, 0, &deadlk)) {
2913 		grant_lock(r, lkb);
2914 		queue_cast(r, lkb, 0);
2915 		goto out;
2916 	}
2917 
2918 	/* can_be_granted() detected that this lock would block in a conversion
2919 	   deadlock, so we leave it on the granted queue and return EDEADLK in
2920 	   the ast for the convert. */
2921 
2922 	if (deadlk && !(lkb->lkb_exflags & DLM_LKF_NODLCKWT)) {
2923 		/* it's left on the granted queue */
2924 		revert_lock(r, lkb);
2925 		queue_cast(r, lkb, -EDEADLK);
2926 		error = -EDEADLK;
2927 		goto out;
2928 	}
2929 
2930 	/* is_demoted() means the can_be_granted() above set the grmode
2931 	   to NL, and left us on the granted queue.  This auto-demotion
2932 	   (due to CONVDEADLK) might mean other locks, and/or this lock, are
2933 	   now grantable.  We have to try to grant other converting locks
2934 	   before we try again to grant this one. */
2935 
2936 	if (is_demoted(lkb)) {
2937 		grant_pending_convert(r, DLM_LOCK_IV, NULL, NULL);
2938 		if (_can_be_granted(r, lkb, 1, 0)) {
2939 			grant_lock(r, lkb);
2940 			queue_cast(r, lkb, 0);
2941 			goto out;
2942 		}
2943 		/* else fall through and move to convert queue */
2944 	}
2945 
2946 	if (can_be_queued(lkb)) {
2947 		error = -EINPROGRESS;
2948 		del_lkb(r, lkb);
2949 		add_lkb(r, lkb, DLM_LKSTS_CONVERT);
2950 		goto out;
2951 	}
2952 
2953 	error = -EAGAIN;
2954 	queue_cast(r, lkb, -EAGAIN);
2955  out:
2956 	return error;
2957 }
2958 
2959 static void do_convert_effects(struct dlm_rsb *r, struct dlm_lkb *lkb,
2960 			       int error)
2961 {
2962 	switch (error) {
2963 	case 0:
2964 		grant_pending_locks(r, NULL);
2965 		/* grant_pending_locks also sends basts */
2966 		break;
2967 	case -EAGAIN:
2968 		if (force_blocking_asts(lkb))
2969 			send_blocking_asts_all(r, lkb);
2970 		break;
2971 	case -EINPROGRESS:
2972 		send_blocking_asts(r, lkb);
2973 		break;
2974 	}
2975 }
2976 
2977 static int do_unlock(struct dlm_rsb *r, struct dlm_lkb *lkb)
2978 {
2979 	remove_lock(r, lkb);
2980 	queue_cast(r, lkb, -DLM_EUNLOCK);
2981 	return -DLM_EUNLOCK;
2982 }
2983 
2984 static void do_unlock_effects(struct dlm_rsb *r, struct dlm_lkb *lkb,
2985 			      int error)
2986 {
2987 	grant_pending_locks(r, NULL);
2988 }
2989 
2990 /* returns: 0 did nothing, -DLM_ECANCEL canceled lock */
2991 
2992 static int do_cancel(struct dlm_rsb *r, struct dlm_lkb *lkb)
2993 {
2994 	int error;
2995 
2996 	error = revert_lock(r, lkb);
2997 	if (error) {
2998 		queue_cast(r, lkb, -DLM_ECANCEL);
2999 		return -DLM_ECANCEL;
3000 	}
3001 	return 0;
3002 }
3003 
3004 static void do_cancel_effects(struct dlm_rsb *r, struct dlm_lkb *lkb,
3005 			      int error)
3006 {
3007 	if (error)
3008 		grant_pending_locks(r, NULL);
3009 }
3010 
3011 /*
3012  * Four stage 3 varieties:
3013  * _request_lock(), _convert_lock(), _unlock_lock(), _cancel_lock()
3014  */
3015 
3016 /* add a new lkb to a possibly new rsb, called by requesting process */
3017 
3018 static int _request_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
3019 {
3020 	int error;
3021 
3022 	/* set_master: sets lkb nodeid from r */
3023 
3024 	error = set_master(r, lkb);
3025 	if (error < 0)
3026 		goto out;
3027 	if (error) {
3028 		error = 0;
3029 		goto out;
3030 	}
3031 
3032 	if (is_remote(r)) {
3033 		/* receive_request() calls do_request() on remote node */
3034 		error = send_request(r, lkb);
3035 	} else {
3036 		error = do_request(r, lkb);
3037 		/* for remote locks the request_reply is sent
3038 		   between do_request and do_request_effects */
3039 		do_request_effects(r, lkb, error);
3040 	}
3041  out:
3042 	return error;
3043 }
3044 
3045 /* change some property of an existing lkb, e.g. mode */
3046 
3047 static int _convert_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
3048 {
3049 	int error;
3050 
3051 	if (is_remote(r)) {
3052 		/* receive_convert() calls do_convert() on remote node */
3053 		error = send_convert(r, lkb);
3054 	} else {
3055 		error = do_convert(r, lkb);
3056 		/* for remote locks the convert_reply is sent
3057 		   between do_convert and do_convert_effects */
3058 		do_convert_effects(r, lkb, error);
3059 	}
3060 
3061 	return error;
3062 }
3063 
3064 /* remove an existing lkb from the granted queue */
3065 
3066 static int _unlock_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
3067 {
3068 	int error;
3069 
3070 	if (is_remote(r)) {
3071 		/* receive_unlock() calls do_unlock() on remote node */
3072 		error = send_unlock(r, lkb);
3073 	} else {
3074 		error = do_unlock(r, lkb);
3075 		/* for remote locks the unlock_reply is sent
3076 		   between do_unlock and do_unlock_effects */
3077 		do_unlock_effects(r, lkb, error);
3078 	}
3079 
3080 	return error;
3081 }
3082 
3083 /* remove an existing lkb from the convert or wait queue */
3084 
3085 static int _cancel_lock(struct dlm_rsb *r, struct dlm_lkb *lkb)
3086 {
3087 	int error;
3088 
3089 	if (is_remote(r)) {
3090 		/* receive_cancel() calls do_cancel() on remote node */
3091 		error = send_cancel(r, lkb);
3092 	} else {
3093 		error = do_cancel(r, lkb);
3094 		/* for remote locks the cancel_reply is sent
3095 		   between do_cancel and do_cancel_effects */
3096 		do_cancel_effects(r, lkb, error);
3097 	}
3098 
3099 	return error;
3100 }
3101 
3102 /*
3103  * Four stage 2 varieties:
3104  * request_lock(), convert_lock(), unlock_lock(), cancel_lock()
3105  */
3106 
3107 static int request_lock(struct dlm_ls *ls, struct dlm_lkb *lkb,
3108 			const void *name, int len,
3109 			struct dlm_args *args)
3110 {
3111 	struct dlm_rsb *r;
3112 	int error;
3113 
3114 	error = validate_lock_args(ls, lkb, args);
3115 	if (error)
3116 		return error;
3117 
3118 	error = find_rsb(ls, name, len, 0, R_REQUEST, &r);
3119 	if (error)
3120 		return error;
3121 
3122 	lock_rsb(r);
3123 
3124 	attach_lkb(r, lkb);
3125 	lkb->lkb_lksb->sb_lkid = lkb->lkb_id;
3126 
3127 	error = _request_lock(r, lkb);
3128 
3129 	unlock_rsb(r);
3130 	put_rsb(r);
3131 	return error;
3132 }
3133 
3134 static int convert_lock(struct dlm_ls *ls, struct dlm_lkb *lkb,
3135 			struct dlm_args *args)
3136 {
3137 	struct dlm_rsb *r;
3138 	int error;
3139 
3140 	r = lkb->lkb_resource;
3141 
3142 	hold_rsb(r);
3143 	lock_rsb(r);
3144 
3145 	error = validate_lock_args(ls, lkb, args);
3146 	if (error)
3147 		goto out;
3148 
3149 	error = _convert_lock(r, lkb);
3150  out:
3151 	unlock_rsb(r);
3152 	put_rsb(r);
3153 	return error;
3154 }
3155 
3156 static int unlock_lock(struct dlm_ls *ls, struct dlm_lkb *lkb,
3157 		       struct dlm_args *args)
3158 {
3159 	struct dlm_rsb *r;
3160 	int error;
3161 
3162 	r = lkb->lkb_resource;
3163 
3164 	hold_rsb(r);
3165 	lock_rsb(r);
3166 
3167 	error = validate_unlock_args(lkb, args);
3168 	if (error)
3169 		goto out;
3170 
3171 	error = _unlock_lock(r, lkb);
3172  out:
3173 	unlock_rsb(r);
3174 	put_rsb(r);
3175 	return error;
3176 }
3177 
3178 static int cancel_lock(struct dlm_ls *ls, struct dlm_lkb *lkb,
3179 		       struct dlm_args *args)
3180 {
3181 	struct dlm_rsb *r;
3182 	int error;
3183 
3184 	r = lkb->lkb_resource;
3185 
3186 	hold_rsb(r);
3187 	lock_rsb(r);
3188 
3189 	error = validate_unlock_args(lkb, args);
3190 	if (error)
3191 		goto out;
3192 
3193 	error = _cancel_lock(r, lkb);
3194  out:
3195 	unlock_rsb(r);
3196 	put_rsb(r);
3197 	return error;
3198 }
3199 
3200 /*
3201  * Two stage 1 varieties:  dlm_lock() and dlm_unlock()
3202  */
3203 
3204 int dlm_lock(dlm_lockspace_t *lockspace,
3205 	     int mode,
3206 	     struct dlm_lksb *lksb,
3207 	     uint32_t flags,
3208 	     const void *name,
3209 	     unsigned int namelen,
3210 	     uint32_t parent_lkid,
3211 	     void (*ast) (void *astarg),
3212 	     void *astarg,
3213 	     void (*bast) (void *astarg, int mode))
3214 {
3215 	struct dlm_ls *ls;
3216 	struct dlm_lkb *lkb;
3217 	struct dlm_args args;
3218 	int error, convert = flags & DLM_LKF_CONVERT;
3219 
3220 	ls = dlm_find_lockspace_local(lockspace);
3221 	if (!ls)
3222 		return -EINVAL;
3223 
3224 	dlm_lock_recovery(ls);
3225 
3226 	if (convert)
3227 		error = find_lkb(ls, lksb->sb_lkid, &lkb);
3228 	else
3229 		error = create_lkb(ls, &lkb);
3230 
3231 	if (error)
3232 		goto out;
3233 
3234 	trace_dlm_lock_start(ls, lkb, name, namelen, mode, flags);
3235 
3236 	error = set_lock_args(mode, lksb, flags, namelen, ast, astarg, bast,
3237 			      &args);
3238 	if (error)
3239 		goto out_put;
3240 
3241 	if (convert)
3242 		error = convert_lock(ls, lkb, &args);
3243 	else
3244 		error = request_lock(ls, lkb, name, namelen, &args);
3245 
3246 	if (error == -EINPROGRESS)
3247 		error = 0;
3248  out_put:
3249 	trace_dlm_lock_end(ls, lkb, name, namelen, mode, flags, error, true);
3250 
3251 	if (convert || error)
3252 		__put_lkb(ls, lkb);
3253 	if (error == -EAGAIN || error == -EDEADLK)
3254 		error = 0;
3255  out:
3256 	dlm_unlock_recovery(ls);
3257 	dlm_put_lockspace(ls);
3258 	return error;
3259 }
3260 
3261 int dlm_unlock(dlm_lockspace_t *lockspace,
3262 	       uint32_t lkid,
3263 	       uint32_t flags,
3264 	       struct dlm_lksb *lksb,
3265 	       void *astarg)
3266 {
3267 	struct dlm_ls *ls;
3268 	struct dlm_lkb *lkb;
3269 	struct dlm_args args;
3270 	int error;
3271 
3272 	ls = dlm_find_lockspace_local(lockspace);
3273 	if (!ls)
3274 		return -EINVAL;
3275 
3276 	dlm_lock_recovery(ls);
3277 
3278 	error = find_lkb(ls, lkid, &lkb);
3279 	if (error)
3280 		goto out;
3281 
3282 	trace_dlm_unlock_start(ls, lkb, flags);
3283 
3284 	error = set_unlock_args(flags, astarg, &args);
3285 	if (error)
3286 		goto out_put;
3287 
3288 	if (flags & DLM_LKF_CANCEL)
3289 		error = cancel_lock(ls, lkb, &args);
3290 	else
3291 		error = unlock_lock(ls, lkb, &args);
3292 
3293 	if (error == -DLM_EUNLOCK || error == -DLM_ECANCEL)
3294 		error = 0;
3295 	if (error == -EBUSY && (flags & (DLM_LKF_CANCEL | DLM_LKF_FORCEUNLOCK)))
3296 		error = 0;
3297  out_put:
3298 	trace_dlm_unlock_end(ls, lkb, flags, error);
3299 
3300 	dlm_put_lkb(lkb);
3301  out:
3302 	dlm_unlock_recovery(ls);
3303 	dlm_put_lockspace(ls);
3304 	return error;
3305 }
3306 
3307 /*
3308  * send/receive routines for remote operations and replies
3309  *
3310  * send_args
3311  * send_common
3312  * send_request			receive_request
3313  * send_convert			receive_convert
3314  * send_unlock			receive_unlock
3315  * send_cancel			receive_cancel
3316  * send_grant			receive_grant
3317  * send_bast			receive_bast
3318  * send_lookup			receive_lookup
3319  * send_remove			receive_remove
3320  *
3321  * 				send_common_reply
3322  * receive_request_reply	send_request_reply
3323  * receive_convert_reply	send_convert_reply
3324  * receive_unlock_reply		send_unlock_reply
3325  * receive_cancel_reply		send_cancel_reply
3326  * receive_lookup_reply		send_lookup_reply
3327  */
3328 
3329 static int _create_message(struct dlm_ls *ls, int mb_len,
3330 			   int to_nodeid, int mstype,
3331 			   struct dlm_message **ms_ret,
3332 			   struct dlm_mhandle **mh_ret,
3333 			   gfp_t allocation)
3334 {
3335 	struct dlm_message *ms;
3336 	struct dlm_mhandle *mh;
3337 	char *mb;
3338 
3339 	/* get_buffer gives us a message handle (mh) that we need to
3340 	   pass into midcomms_commit and a message buffer (mb) that we
3341 	   write our data into */
3342 
3343 	mh = dlm_midcomms_get_mhandle(to_nodeid, mb_len, allocation, &mb);
3344 	if (!mh)
3345 		return -ENOBUFS;
3346 
3347 	ms = (struct dlm_message *) mb;
3348 
3349 	ms->m_header.h_version = cpu_to_le32(DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
3350 	ms->m_header.u.h_lockspace = cpu_to_le32(ls->ls_global_id);
3351 	ms->m_header.h_nodeid = cpu_to_le32(dlm_our_nodeid());
3352 	ms->m_header.h_length = cpu_to_le16(mb_len);
3353 	ms->m_header.h_cmd = DLM_MSG;
3354 
3355 	ms->m_type = cpu_to_le32(mstype);
3356 
3357 	*mh_ret = mh;
3358 	*ms_ret = ms;
3359 	return 0;
3360 }
3361 
3362 static int create_message(struct dlm_rsb *r, struct dlm_lkb *lkb,
3363 			  int to_nodeid, int mstype,
3364 			  struct dlm_message **ms_ret,
3365 			  struct dlm_mhandle **mh_ret,
3366 			  gfp_t allocation)
3367 {
3368 	int mb_len = sizeof(struct dlm_message);
3369 
3370 	switch (mstype) {
3371 	case DLM_MSG_REQUEST:
3372 	case DLM_MSG_LOOKUP:
3373 	case DLM_MSG_REMOVE:
3374 		mb_len += r->res_length;
3375 		break;
3376 	case DLM_MSG_CONVERT:
3377 	case DLM_MSG_UNLOCK:
3378 	case DLM_MSG_REQUEST_REPLY:
3379 	case DLM_MSG_CONVERT_REPLY:
3380 	case DLM_MSG_GRANT:
3381 		if (lkb && lkb->lkb_lvbptr && (lkb->lkb_exflags & DLM_LKF_VALBLK))
3382 			mb_len += r->res_ls->ls_lvblen;
3383 		break;
3384 	}
3385 
3386 	return _create_message(r->res_ls, mb_len, to_nodeid, mstype,
3387 			       ms_ret, mh_ret, allocation);
3388 }
3389 
3390 /* further lowcomms enhancements or alternate implementations may make
3391    the return value from this function useful at some point */
3392 
3393 static int send_message(struct dlm_mhandle *mh, struct dlm_message *ms,
3394 			const void *name, int namelen)
3395 {
3396 	dlm_midcomms_commit_mhandle(mh, name, namelen);
3397 	return 0;
3398 }
3399 
3400 static void send_args(struct dlm_rsb *r, struct dlm_lkb *lkb,
3401 		      struct dlm_message *ms)
3402 {
3403 	ms->m_nodeid   = cpu_to_le32(lkb->lkb_nodeid);
3404 	ms->m_pid      = cpu_to_le32(lkb->lkb_ownpid);
3405 	ms->m_lkid     = cpu_to_le32(lkb->lkb_id);
3406 	ms->m_remid    = cpu_to_le32(lkb->lkb_remid);
3407 	ms->m_exflags  = cpu_to_le32(lkb->lkb_exflags);
3408 	ms->m_sbflags  = cpu_to_le32(dlm_sbflags_val(lkb));
3409 	ms->m_flags    = cpu_to_le32(dlm_dflags_val(lkb));
3410 	ms->m_lvbseq   = cpu_to_le32(lkb->lkb_lvbseq);
3411 	ms->m_status   = cpu_to_le32(lkb->lkb_status);
3412 	ms->m_grmode   = cpu_to_le32(lkb->lkb_grmode);
3413 	ms->m_rqmode   = cpu_to_le32(lkb->lkb_rqmode);
3414 	ms->m_hash     = cpu_to_le32(r->res_hash);
3415 
3416 	/* m_result and m_bastmode are set from function args,
3417 	   not from lkb fields */
3418 
3419 	if (lkb->lkb_bastfn)
3420 		ms->m_asts |= cpu_to_le32(DLM_CB_BAST);
3421 	if (lkb->lkb_astfn)
3422 		ms->m_asts |= cpu_to_le32(DLM_CB_CAST);
3423 
3424 	/* compare with switch in create_message; send_remove() doesn't
3425 	   use send_args() */
3426 
3427 	switch (ms->m_type) {
3428 	case cpu_to_le32(DLM_MSG_REQUEST):
3429 	case cpu_to_le32(DLM_MSG_LOOKUP):
3430 		memcpy(ms->m_extra, r->res_name, r->res_length);
3431 		break;
3432 	case cpu_to_le32(DLM_MSG_CONVERT):
3433 	case cpu_to_le32(DLM_MSG_UNLOCK):
3434 	case cpu_to_le32(DLM_MSG_REQUEST_REPLY):
3435 	case cpu_to_le32(DLM_MSG_CONVERT_REPLY):
3436 	case cpu_to_le32(DLM_MSG_GRANT):
3437 		if (!lkb->lkb_lvbptr || !(lkb->lkb_exflags & DLM_LKF_VALBLK))
3438 			break;
3439 		memcpy(ms->m_extra, lkb->lkb_lvbptr, r->res_ls->ls_lvblen);
3440 		break;
3441 	}
3442 }
3443 
3444 static int send_common(struct dlm_rsb *r, struct dlm_lkb *lkb, int mstype)
3445 {
3446 	struct dlm_message *ms;
3447 	struct dlm_mhandle *mh;
3448 	int to_nodeid, error;
3449 
3450 	to_nodeid = r->res_nodeid;
3451 
3452 	error = add_to_waiters(lkb, mstype, to_nodeid);
3453 	if (error)
3454 		return error;
3455 
3456 	error = create_message(r, lkb, to_nodeid, mstype, &ms, &mh, GFP_NOFS);
3457 	if (error)
3458 		goto fail;
3459 
3460 	send_args(r, lkb, ms);
3461 
3462 	error = send_message(mh, ms, r->res_name, r->res_length);
3463 	if (error)
3464 		goto fail;
3465 	return 0;
3466 
3467  fail:
3468 	remove_from_waiters(lkb, msg_reply_type(mstype));
3469 	return error;
3470 }
3471 
3472 static int send_request(struct dlm_rsb *r, struct dlm_lkb *lkb)
3473 {
3474 	return send_common(r, lkb, DLM_MSG_REQUEST);
3475 }
3476 
3477 static int send_convert(struct dlm_rsb *r, struct dlm_lkb *lkb)
3478 {
3479 	int error;
3480 
3481 	error = send_common(r, lkb, DLM_MSG_CONVERT);
3482 
3483 	/* down conversions go without a reply from the master */
3484 	if (!error && down_conversion(lkb)) {
3485 		remove_from_waiters(lkb, DLM_MSG_CONVERT_REPLY);
3486 		r->res_ls->ls_local_ms.m_type = cpu_to_le32(DLM_MSG_CONVERT_REPLY);
3487 		r->res_ls->ls_local_ms.m_result = 0;
3488 		__receive_convert_reply(r, lkb, &r->res_ls->ls_local_ms, true);
3489 	}
3490 
3491 	return error;
3492 }
3493 
3494 /* FIXME: if this lkb is the only lock we hold on the rsb, then set
3495    MASTER_UNCERTAIN to force the next request on the rsb to confirm
3496    that the master is still correct. */
3497 
3498 static int send_unlock(struct dlm_rsb *r, struct dlm_lkb *lkb)
3499 {
3500 	return send_common(r, lkb, DLM_MSG_UNLOCK);
3501 }
3502 
3503 static int send_cancel(struct dlm_rsb *r, struct dlm_lkb *lkb)
3504 {
3505 	return send_common(r, lkb, DLM_MSG_CANCEL);
3506 }
3507 
3508 static int send_grant(struct dlm_rsb *r, struct dlm_lkb *lkb)
3509 {
3510 	struct dlm_message *ms;
3511 	struct dlm_mhandle *mh;
3512 	int to_nodeid, error;
3513 
3514 	to_nodeid = lkb->lkb_nodeid;
3515 
3516 	error = create_message(r, lkb, to_nodeid, DLM_MSG_GRANT, &ms, &mh,
3517 			       GFP_NOFS);
3518 	if (error)
3519 		goto out;
3520 
3521 	send_args(r, lkb, ms);
3522 
3523 	ms->m_result = 0;
3524 
3525 	error = send_message(mh, ms, r->res_name, r->res_length);
3526  out:
3527 	return error;
3528 }
3529 
3530 static int send_bast(struct dlm_rsb *r, struct dlm_lkb *lkb, int mode)
3531 {
3532 	struct dlm_message *ms;
3533 	struct dlm_mhandle *mh;
3534 	int to_nodeid, error;
3535 
3536 	to_nodeid = lkb->lkb_nodeid;
3537 
3538 	error = create_message(r, NULL, to_nodeid, DLM_MSG_BAST, &ms, &mh,
3539 			       GFP_NOFS);
3540 	if (error)
3541 		goto out;
3542 
3543 	send_args(r, lkb, ms);
3544 
3545 	ms->m_bastmode = cpu_to_le32(mode);
3546 
3547 	error = send_message(mh, ms, r->res_name, r->res_length);
3548  out:
3549 	return error;
3550 }
3551 
3552 static int send_lookup(struct dlm_rsb *r, struct dlm_lkb *lkb)
3553 {
3554 	struct dlm_message *ms;
3555 	struct dlm_mhandle *mh;
3556 	int to_nodeid, error;
3557 
3558 	to_nodeid = dlm_dir_nodeid(r);
3559 
3560 	error = add_to_waiters(lkb, DLM_MSG_LOOKUP, to_nodeid);
3561 	if (error)
3562 		return error;
3563 
3564 	error = create_message(r, NULL, to_nodeid, DLM_MSG_LOOKUP, &ms, &mh,
3565 			       GFP_NOFS);
3566 	if (error)
3567 		goto fail;
3568 
3569 	send_args(r, lkb, ms);
3570 
3571 	error = send_message(mh, ms, r->res_name, r->res_length);
3572 	if (error)
3573 		goto fail;
3574 	return 0;
3575 
3576  fail:
3577 	remove_from_waiters(lkb, DLM_MSG_LOOKUP_REPLY);
3578 	return error;
3579 }
3580 
3581 static int send_remove(struct dlm_rsb *r)
3582 {
3583 	struct dlm_message *ms;
3584 	struct dlm_mhandle *mh;
3585 	int to_nodeid, error;
3586 
3587 	to_nodeid = dlm_dir_nodeid(r);
3588 
3589 	error = create_message(r, NULL, to_nodeid, DLM_MSG_REMOVE, &ms, &mh,
3590 			       GFP_ATOMIC);
3591 	if (error)
3592 		goto out;
3593 
3594 	memcpy(ms->m_extra, r->res_name, r->res_length);
3595 	ms->m_hash = cpu_to_le32(r->res_hash);
3596 
3597 	error = send_message(mh, ms, r->res_name, r->res_length);
3598  out:
3599 	return error;
3600 }
3601 
3602 static int send_common_reply(struct dlm_rsb *r, struct dlm_lkb *lkb,
3603 			     int mstype, int rv)
3604 {
3605 	struct dlm_message *ms;
3606 	struct dlm_mhandle *mh;
3607 	int to_nodeid, error;
3608 
3609 	to_nodeid = lkb->lkb_nodeid;
3610 
3611 	error = create_message(r, lkb, to_nodeid, mstype, &ms, &mh, GFP_NOFS);
3612 	if (error)
3613 		goto out;
3614 
3615 	send_args(r, lkb, ms);
3616 
3617 	ms->m_result = cpu_to_le32(to_dlm_errno(rv));
3618 
3619 	error = send_message(mh, ms, r->res_name, r->res_length);
3620  out:
3621 	return error;
3622 }
3623 
3624 static int send_request_reply(struct dlm_rsb *r, struct dlm_lkb *lkb, int rv)
3625 {
3626 	return send_common_reply(r, lkb, DLM_MSG_REQUEST_REPLY, rv);
3627 }
3628 
3629 static int send_convert_reply(struct dlm_rsb *r, struct dlm_lkb *lkb, int rv)
3630 {
3631 	return send_common_reply(r, lkb, DLM_MSG_CONVERT_REPLY, rv);
3632 }
3633 
3634 static int send_unlock_reply(struct dlm_rsb *r, struct dlm_lkb *lkb, int rv)
3635 {
3636 	return send_common_reply(r, lkb, DLM_MSG_UNLOCK_REPLY, rv);
3637 }
3638 
3639 static int send_cancel_reply(struct dlm_rsb *r, struct dlm_lkb *lkb, int rv)
3640 {
3641 	return send_common_reply(r, lkb, DLM_MSG_CANCEL_REPLY, rv);
3642 }
3643 
3644 static int send_lookup_reply(struct dlm_ls *ls,
3645 			     const struct dlm_message *ms_in, int ret_nodeid,
3646 			     int rv)
3647 {
3648 	struct dlm_rsb *r = &ls->ls_local_rsb;
3649 	struct dlm_message *ms;
3650 	struct dlm_mhandle *mh;
3651 	int error, nodeid = le32_to_cpu(ms_in->m_header.h_nodeid);
3652 
3653 	error = create_message(r, NULL, nodeid, DLM_MSG_LOOKUP_REPLY, &ms, &mh,
3654 			       GFP_NOFS);
3655 	if (error)
3656 		goto out;
3657 
3658 	ms->m_lkid = ms_in->m_lkid;
3659 	ms->m_result = cpu_to_le32(to_dlm_errno(rv));
3660 	ms->m_nodeid = cpu_to_le32(ret_nodeid);
3661 
3662 	error = send_message(mh, ms, ms_in->m_extra, receive_extralen(ms_in));
3663  out:
3664 	return error;
3665 }
3666 
3667 /* which args we save from a received message depends heavily on the type
3668    of message, unlike the send side where we can safely send everything about
3669    the lkb for any type of message */
3670 
3671 static void receive_flags(struct dlm_lkb *lkb, const struct dlm_message *ms)
3672 {
3673 	lkb->lkb_exflags = le32_to_cpu(ms->m_exflags);
3674 	dlm_set_sbflags_val(lkb, le32_to_cpu(ms->m_sbflags));
3675 	dlm_set_dflags_val(lkb, le32_to_cpu(ms->m_flags));
3676 }
3677 
3678 static void receive_flags_reply(struct dlm_lkb *lkb,
3679 				const struct dlm_message *ms,
3680 				bool local)
3681 {
3682 	if (local)
3683 		return;
3684 
3685 	dlm_set_sbflags_val(lkb, le32_to_cpu(ms->m_sbflags));
3686 	dlm_set_dflags_val(lkb, le32_to_cpu(ms->m_flags));
3687 }
3688 
3689 static int receive_extralen(const struct dlm_message *ms)
3690 {
3691 	return (le16_to_cpu(ms->m_header.h_length) -
3692 		sizeof(struct dlm_message));
3693 }
3694 
3695 static int receive_lvb(struct dlm_ls *ls, struct dlm_lkb *lkb,
3696 		       const struct dlm_message *ms)
3697 {
3698 	int len;
3699 
3700 	if (lkb->lkb_exflags & DLM_LKF_VALBLK) {
3701 		if (!lkb->lkb_lvbptr)
3702 			lkb->lkb_lvbptr = dlm_allocate_lvb(ls);
3703 		if (!lkb->lkb_lvbptr)
3704 			return -ENOMEM;
3705 		len = receive_extralen(ms);
3706 		if (len > ls->ls_lvblen)
3707 			len = ls->ls_lvblen;
3708 		memcpy(lkb->lkb_lvbptr, ms->m_extra, len);
3709 	}
3710 	return 0;
3711 }
3712 
3713 static void fake_bastfn(void *astparam, int mode)
3714 {
3715 	log_print("fake_bastfn should not be called");
3716 }
3717 
3718 static void fake_astfn(void *astparam)
3719 {
3720 	log_print("fake_astfn should not be called");
3721 }
3722 
3723 static int receive_request_args(struct dlm_ls *ls, struct dlm_lkb *lkb,
3724 				const struct dlm_message *ms)
3725 {
3726 	lkb->lkb_nodeid = le32_to_cpu(ms->m_header.h_nodeid);
3727 	lkb->lkb_ownpid = le32_to_cpu(ms->m_pid);
3728 	lkb->lkb_remid = le32_to_cpu(ms->m_lkid);
3729 	lkb->lkb_grmode = DLM_LOCK_IV;
3730 	lkb->lkb_rqmode = le32_to_cpu(ms->m_rqmode);
3731 
3732 	lkb->lkb_bastfn = (ms->m_asts & cpu_to_le32(DLM_CB_BAST)) ? &fake_bastfn : NULL;
3733 	lkb->lkb_astfn = (ms->m_asts & cpu_to_le32(DLM_CB_CAST)) ? &fake_astfn : NULL;
3734 
3735 	if (lkb->lkb_exflags & DLM_LKF_VALBLK) {
3736 		/* lkb was just created so there won't be an lvb yet */
3737 		lkb->lkb_lvbptr = dlm_allocate_lvb(ls);
3738 		if (!lkb->lkb_lvbptr)
3739 			return -ENOMEM;
3740 	}
3741 
3742 	return 0;
3743 }
3744 
3745 static int receive_convert_args(struct dlm_ls *ls, struct dlm_lkb *lkb,
3746 				const struct dlm_message *ms)
3747 {
3748 	if (lkb->lkb_status != DLM_LKSTS_GRANTED)
3749 		return -EBUSY;
3750 
3751 	if (receive_lvb(ls, lkb, ms))
3752 		return -ENOMEM;
3753 
3754 	lkb->lkb_rqmode = le32_to_cpu(ms->m_rqmode);
3755 	lkb->lkb_lvbseq = le32_to_cpu(ms->m_lvbseq);
3756 
3757 	return 0;
3758 }
3759 
3760 static int receive_unlock_args(struct dlm_ls *ls, struct dlm_lkb *lkb,
3761 			       const struct dlm_message *ms)
3762 {
3763 	if (receive_lvb(ls, lkb, ms))
3764 		return -ENOMEM;
3765 	return 0;
3766 }
3767 
3768 /* We fill in the local-lkb fields with the info that send_xxxx_reply()
3769    uses to send a reply and that the remote end uses to process the reply. */
3770 
3771 static void setup_local_lkb(struct dlm_ls *ls, const struct dlm_message *ms)
3772 {
3773 	struct dlm_lkb *lkb = &ls->ls_local_lkb;
3774 	lkb->lkb_nodeid = le32_to_cpu(ms->m_header.h_nodeid);
3775 	lkb->lkb_remid = le32_to_cpu(ms->m_lkid);
3776 }
3777 
3778 /* This is called after the rsb is locked so that we can safely inspect
3779    fields in the lkb. */
3780 
3781 static int validate_message(struct dlm_lkb *lkb, const struct dlm_message *ms)
3782 {
3783 	int from = le32_to_cpu(ms->m_header.h_nodeid);
3784 	int error = 0;
3785 
3786 	/* currently mixing of user/kernel locks are not supported */
3787 	if (ms->m_flags & cpu_to_le32(BIT(DLM_DFL_USER_BIT)) &&
3788 	    !test_bit(DLM_DFL_USER_BIT, &lkb->lkb_dflags)) {
3789 		log_error(lkb->lkb_resource->res_ls,
3790 			  "got user dlm message for a kernel lock");
3791 		error = -EINVAL;
3792 		goto out;
3793 	}
3794 
3795 	switch (ms->m_type) {
3796 	case cpu_to_le32(DLM_MSG_CONVERT):
3797 	case cpu_to_le32(DLM_MSG_UNLOCK):
3798 	case cpu_to_le32(DLM_MSG_CANCEL):
3799 		if (!is_master_copy(lkb) || lkb->lkb_nodeid != from)
3800 			error = -EINVAL;
3801 		break;
3802 
3803 	case cpu_to_le32(DLM_MSG_CONVERT_REPLY):
3804 	case cpu_to_le32(DLM_MSG_UNLOCK_REPLY):
3805 	case cpu_to_le32(DLM_MSG_CANCEL_REPLY):
3806 	case cpu_to_le32(DLM_MSG_GRANT):
3807 	case cpu_to_le32(DLM_MSG_BAST):
3808 		if (!is_process_copy(lkb) || lkb->lkb_nodeid != from)
3809 			error = -EINVAL;
3810 		break;
3811 
3812 	case cpu_to_le32(DLM_MSG_REQUEST_REPLY):
3813 		if (!is_process_copy(lkb))
3814 			error = -EINVAL;
3815 		else if (lkb->lkb_nodeid != -1 && lkb->lkb_nodeid != from)
3816 			error = -EINVAL;
3817 		break;
3818 
3819 	default:
3820 		error = -EINVAL;
3821 	}
3822 
3823 out:
3824 	if (error)
3825 		log_error(lkb->lkb_resource->res_ls,
3826 			  "ignore invalid message %d from %d %x %x %x %d",
3827 			  le32_to_cpu(ms->m_type), from, lkb->lkb_id,
3828 			  lkb->lkb_remid, dlm_iflags_val(lkb),
3829 			  lkb->lkb_nodeid);
3830 	return error;
3831 }
3832 
3833 static int receive_request(struct dlm_ls *ls, const struct dlm_message *ms)
3834 {
3835 	struct dlm_lkb *lkb;
3836 	struct dlm_rsb *r;
3837 	int from_nodeid;
3838 	int error, namelen = 0;
3839 
3840 	from_nodeid = le32_to_cpu(ms->m_header.h_nodeid);
3841 
3842 	error = create_lkb(ls, &lkb);
3843 	if (error)
3844 		goto fail;
3845 
3846 	receive_flags(lkb, ms);
3847 	set_bit(DLM_IFL_MSTCPY_BIT, &lkb->lkb_iflags);
3848 	error = receive_request_args(ls, lkb, ms);
3849 	if (error) {
3850 		__put_lkb(ls, lkb);
3851 		goto fail;
3852 	}
3853 
3854 	/* The dir node is the authority on whether we are the master
3855 	   for this rsb or not, so if the master sends us a request, we should
3856 	   recreate the rsb if we've destroyed it.   This race happens when we
3857 	   send a remove message to the dir node at the same time that the dir
3858 	   node sends us a request for the rsb. */
3859 
3860 	namelen = receive_extralen(ms);
3861 
3862 	error = find_rsb(ls, ms->m_extra, namelen, from_nodeid,
3863 			 R_RECEIVE_REQUEST, &r);
3864 	if (error) {
3865 		__put_lkb(ls, lkb);
3866 		goto fail;
3867 	}
3868 
3869 	lock_rsb(r);
3870 
3871 	if (r->res_master_nodeid != dlm_our_nodeid()) {
3872 		error = validate_master_nodeid(ls, r, from_nodeid);
3873 		if (error) {
3874 			unlock_rsb(r);
3875 			put_rsb(r);
3876 			__put_lkb(ls, lkb);
3877 			goto fail;
3878 		}
3879 	}
3880 
3881 	attach_lkb(r, lkb);
3882 	error = do_request(r, lkb);
3883 	send_request_reply(r, lkb, error);
3884 	do_request_effects(r, lkb, error);
3885 
3886 	unlock_rsb(r);
3887 	put_rsb(r);
3888 
3889 	if (error == -EINPROGRESS)
3890 		error = 0;
3891 	if (error)
3892 		dlm_put_lkb(lkb);
3893 	return 0;
3894 
3895  fail:
3896 	/* TODO: instead of returning ENOTBLK, add the lkb to res_lookup
3897 	   and do this receive_request again from process_lookup_list once
3898 	   we get the lookup reply.  This would avoid a many repeated
3899 	   ENOTBLK request failures when the lookup reply designating us
3900 	   as master is delayed. */
3901 
3902 	if (error != -ENOTBLK) {
3903 		log_limit(ls, "receive_request %x from %d %d",
3904 			  le32_to_cpu(ms->m_lkid), from_nodeid, error);
3905 	}
3906 
3907 	setup_local_lkb(ls, ms);
3908 	send_request_reply(&ls->ls_local_rsb, &ls->ls_local_lkb, error);
3909 	return error;
3910 }
3911 
3912 static int receive_convert(struct dlm_ls *ls, const struct dlm_message *ms)
3913 {
3914 	struct dlm_lkb *lkb;
3915 	struct dlm_rsb *r;
3916 	int error, reply = 1;
3917 
3918 	error = find_lkb(ls, le32_to_cpu(ms->m_remid), &lkb);
3919 	if (error)
3920 		goto fail;
3921 
3922 	if (lkb->lkb_remid != le32_to_cpu(ms->m_lkid)) {
3923 		log_error(ls, "receive_convert %x remid %x recover_seq %llu "
3924 			  "remote %d %x", lkb->lkb_id, lkb->lkb_remid,
3925 			  (unsigned long long)lkb->lkb_recover_seq,
3926 			  le32_to_cpu(ms->m_header.h_nodeid),
3927 			  le32_to_cpu(ms->m_lkid));
3928 		error = -ENOENT;
3929 		dlm_put_lkb(lkb);
3930 		goto fail;
3931 	}
3932 
3933 	r = lkb->lkb_resource;
3934 
3935 	hold_rsb(r);
3936 	lock_rsb(r);
3937 
3938 	error = validate_message(lkb, ms);
3939 	if (error)
3940 		goto out;
3941 
3942 	receive_flags(lkb, ms);
3943 
3944 	error = receive_convert_args(ls, lkb, ms);
3945 	if (error) {
3946 		send_convert_reply(r, lkb, error);
3947 		goto out;
3948 	}
3949 
3950 	reply = !down_conversion(lkb);
3951 
3952 	error = do_convert(r, lkb);
3953 	if (reply)
3954 		send_convert_reply(r, lkb, error);
3955 	do_convert_effects(r, lkb, error);
3956  out:
3957 	unlock_rsb(r);
3958 	put_rsb(r);
3959 	dlm_put_lkb(lkb);
3960 	return 0;
3961 
3962  fail:
3963 	setup_local_lkb(ls, ms);
3964 	send_convert_reply(&ls->ls_local_rsb, &ls->ls_local_lkb, error);
3965 	return error;
3966 }
3967 
3968 static int receive_unlock(struct dlm_ls *ls, const struct dlm_message *ms)
3969 {
3970 	struct dlm_lkb *lkb;
3971 	struct dlm_rsb *r;
3972 	int error;
3973 
3974 	error = find_lkb(ls, le32_to_cpu(ms->m_remid), &lkb);
3975 	if (error)
3976 		goto fail;
3977 
3978 	if (lkb->lkb_remid != le32_to_cpu(ms->m_lkid)) {
3979 		log_error(ls, "receive_unlock %x remid %x remote %d %x",
3980 			  lkb->lkb_id, lkb->lkb_remid,
3981 			  le32_to_cpu(ms->m_header.h_nodeid),
3982 			  le32_to_cpu(ms->m_lkid));
3983 		error = -ENOENT;
3984 		dlm_put_lkb(lkb);
3985 		goto fail;
3986 	}
3987 
3988 	r = lkb->lkb_resource;
3989 
3990 	hold_rsb(r);
3991 	lock_rsb(r);
3992 
3993 	error = validate_message(lkb, ms);
3994 	if (error)
3995 		goto out;
3996 
3997 	receive_flags(lkb, ms);
3998 
3999 	error = receive_unlock_args(ls, lkb, ms);
4000 	if (error) {
4001 		send_unlock_reply(r, lkb, error);
4002 		goto out;
4003 	}
4004 
4005 	error = do_unlock(r, lkb);
4006 	send_unlock_reply(r, lkb, error);
4007 	do_unlock_effects(r, lkb, error);
4008  out:
4009 	unlock_rsb(r);
4010 	put_rsb(r);
4011 	dlm_put_lkb(lkb);
4012 	return 0;
4013 
4014  fail:
4015 	setup_local_lkb(ls, ms);
4016 	send_unlock_reply(&ls->ls_local_rsb, &ls->ls_local_lkb, error);
4017 	return error;
4018 }
4019 
4020 static int receive_cancel(struct dlm_ls *ls, const struct dlm_message *ms)
4021 {
4022 	struct dlm_lkb *lkb;
4023 	struct dlm_rsb *r;
4024 	int error;
4025 
4026 	error = find_lkb(ls, le32_to_cpu(ms->m_remid), &lkb);
4027 	if (error)
4028 		goto fail;
4029 
4030 	receive_flags(lkb, ms);
4031 
4032 	r = lkb->lkb_resource;
4033 
4034 	hold_rsb(r);
4035 	lock_rsb(r);
4036 
4037 	error = validate_message(lkb, ms);
4038 	if (error)
4039 		goto out;
4040 
4041 	error = do_cancel(r, lkb);
4042 	send_cancel_reply(r, lkb, error);
4043 	do_cancel_effects(r, lkb, error);
4044  out:
4045 	unlock_rsb(r);
4046 	put_rsb(r);
4047 	dlm_put_lkb(lkb);
4048 	return 0;
4049 
4050  fail:
4051 	setup_local_lkb(ls, ms);
4052 	send_cancel_reply(&ls->ls_local_rsb, &ls->ls_local_lkb, error);
4053 	return error;
4054 }
4055 
4056 static int receive_grant(struct dlm_ls *ls, const struct dlm_message *ms)
4057 {
4058 	struct dlm_lkb *lkb;
4059 	struct dlm_rsb *r;
4060 	int error;
4061 
4062 	error = find_lkb(ls, le32_to_cpu(ms->m_remid), &lkb);
4063 	if (error)
4064 		return error;
4065 
4066 	r = lkb->lkb_resource;
4067 
4068 	hold_rsb(r);
4069 	lock_rsb(r);
4070 
4071 	error = validate_message(lkb, ms);
4072 	if (error)
4073 		goto out;
4074 
4075 	receive_flags_reply(lkb, ms, false);
4076 	if (is_altmode(lkb))
4077 		munge_altmode(lkb, ms);
4078 	grant_lock_pc(r, lkb, ms);
4079 	queue_cast(r, lkb, 0);
4080  out:
4081 	unlock_rsb(r);
4082 	put_rsb(r);
4083 	dlm_put_lkb(lkb);
4084 	return 0;
4085 }
4086 
4087 static int receive_bast(struct dlm_ls *ls, const struct dlm_message *ms)
4088 {
4089 	struct dlm_lkb *lkb;
4090 	struct dlm_rsb *r;
4091 	int error;
4092 
4093 	error = find_lkb(ls, le32_to_cpu(ms->m_remid), &lkb);
4094 	if (error)
4095 		return error;
4096 
4097 	r = lkb->lkb_resource;
4098 
4099 	hold_rsb(r);
4100 	lock_rsb(r);
4101 
4102 	error = validate_message(lkb, ms);
4103 	if (error)
4104 		goto out;
4105 
4106 	queue_bast(r, lkb, le32_to_cpu(ms->m_bastmode));
4107 	lkb->lkb_highbast = le32_to_cpu(ms->m_bastmode);
4108  out:
4109 	unlock_rsb(r);
4110 	put_rsb(r);
4111 	dlm_put_lkb(lkb);
4112 	return 0;
4113 }
4114 
4115 static void receive_lookup(struct dlm_ls *ls, const struct dlm_message *ms)
4116 {
4117 	int len, error, ret_nodeid, from_nodeid, our_nodeid;
4118 
4119 	from_nodeid = le32_to_cpu(ms->m_header.h_nodeid);
4120 	our_nodeid = dlm_our_nodeid();
4121 
4122 	len = receive_extralen(ms);
4123 
4124 	error = dlm_master_lookup(ls, from_nodeid, ms->m_extra, len, 0,
4125 				  &ret_nodeid, NULL);
4126 
4127 	/* Optimization: we're master so treat lookup as a request */
4128 	if (!error && ret_nodeid == our_nodeid) {
4129 		receive_request(ls, ms);
4130 		return;
4131 	}
4132 	send_lookup_reply(ls, ms, ret_nodeid, error);
4133 }
4134 
4135 static void receive_remove(struct dlm_ls *ls, const struct dlm_message *ms)
4136 {
4137 	char name[DLM_RESNAME_MAXLEN+1];
4138 	struct dlm_rsb *r;
4139 	uint32_t hash, b;
4140 	int rv, len, dir_nodeid, from_nodeid;
4141 
4142 	from_nodeid = le32_to_cpu(ms->m_header.h_nodeid);
4143 
4144 	len = receive_extralen(ms);
4145 
4146 	if (len > DLM_RESNAME_MAXLEN) {
4147 		log_error(ls, "receive_remove from %d bad len %d",
4148 			  from_nodeid, len);
4149 		return;
4150 	}
4151 
4152 	dir_nodeid = dlm_hash2nodeid(ls, le32_to_cpu(ms->m_hash));
4153 	if (dir_nodeid != dlm_our_nodeid()) {
4154 		log_error(ls, "receive_remove from %d bad nodeid %d",
4155 			  from_nodeid, dir_nodeid);
4156 		return;
4157 	}
4158 
4159 	/* Look for name on rsbtbl.toss, if it's there, kill it.
4160 	   If it's on rsbtbl.keep, it's being used, and we should ignore this
4161 	   message.  This is an expected race between the dir node sending a
4162 	   request to the master node at the same time as the master node sends
4163 	   a remove to the dir node.  The resolution to that race is for the
4164 	   dir node to ignore the remove message, and the master node to
4165 	   recreate the master rsb when it gets a request from the dir node for
4166 	   an rsb it doesn't have. */
4167 
4168 	memset(name, 0, sizeof(name));
4169 	memcpy(name, ms->m_extra, len);
4170 
4171 	hash = jhash(name, len, 0);
4172 	b = hash & (ls->ls_rsbtbl_size - 1);
4173 
4174 	spin_lock(&ls->ls_rsbtbl[b].lock);
4175 
4176 	rv = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].toss, name, len, &r);
4177 	if (rv) {
4178 		/* verify the rsb is on keep list per comment above */
4179 		rv = dlm_search_rsb_tree(&ls->ls_rsbtbl[b].keep, name, len, &r);
4180 		if (rv) {
4181 			/* should not happen */
4182 			log_error(ls, "receive_remove from %d not found %s",
4183 				  from_nodeid, name);
4184 			spin_unlock(&ls->ls_rsbtbl[b].lock);
4185 			return;
4186 		}
4187 		if (r->res_master_nodeid != from_nodeid) {
4188 			/* should not happen */
4189 			log_error(ls, "receive_remove keep from %d master %d",
4190 				  from_nodeid, r->res_master_nodeid);
4191 			dlm_print_rsb(r);
4192 			spin_unlock(&ls->ls_rsbtbl[b].lock);
4193 			return;
4194 		}
4195 
4196 		log_debug(ls, "receive_remove from %d master %d first %x %s",
4197 			  from_nodeid, r->res_master_nodeid, r->res_first_lkid,
4198 			  name);
4199 		spin_unlock(&ls->ls_rsbtbl[b].lock);
4200 		return;
4201 	}
4202 
4203 	if (r->res_master_nodeid != from_nodeid) {
4204 		log_error(ls, "receive_remove toss from %d master %d",
4205 			  from_nodeid, r->res_master_nodeid);
4206 		dlm_print_rsb(r);
4207 		spin_unlock(&ls->ls_rsbtbl[b].lock);
4208 		return;
4209 	}
4210 
4211 	if (kref_put(&r->res_ref, kill_rsb)) {
4212 		rb_erase(&r->res_hashnode, &ls->ls_rsbtbl[b].toss);
4213 		spin_unlock(&ls->ls_rsbtbl[b].lock);
4214 		dlm_free_rsb(r);
4215 	} else {
4216 		log_error(ls, "receive_remove from %d rsb ref error",
4217 			  from_nodeid);
4218 		dlm_print_rsb(r);
4219 		spin_unlock(&ls->ls_rsbtbl[b].lock);
4220 	}
4221 }
4222 
4223 static void receive_purge(struct dlm_ls *ls, const struct dlm_message *ms)
4224 {
4225 	do_purge(ls, le32_to_cpu(ms->m_nodeid), le32_to_cpu(ms->m_pid));
4226 }
4227 
4228 static int receive_request_reply(struct dlm_ls *ls,
4229 				 const struct dlm_message *ms)
4230 {
4231 	struct dlm_lkb *lkb;
4232 	struct dlm_rsb *r;
4233 	int error, mstype, result;
4234 	int from_nodeid = le32_to_cpu(ms->m_header.h_nodeid);
4235 
4236 	error = find_lkb(ls, le32_to_cpu(ms->m_remid), &lkb);
4237 	if (error)
4238 		return error;
4239 
4240 	r = lkb->lkb_resource;
4241 	hold_rsb(r);
4242 	lock_rsb(r);
4243 
4244 	error = validate_message(lkb, ms);
4245 	if (error)
4246 		goto out;
4247 
4248 	mstype = lkb->lkb_wait_type;
4249 	error = remove_from_waiters(lkb, DLM_MSG_REQUEST_REPLY);
4250 	if (error) {
4251 		log_error(ls, "receive_request_reply %x remote %d %x result %d",
4252 			  lkb->lkb_id, from_nodeid, le32_to_cpu(ms->m_lkid),
4253 			  from_dlm_errno(le32_to_cpu(ms->m_result)));
4254 		dlm_dump_rsb(r);
4255 		goto out;
4256 	}
4257 
4258 	/* Optimization: the dir node was also the master, so it took our
4259 	   lookup as a request and sent request reply instead of lookup reply */
4260 	if (mstype == DLM_MSG_LOOKUP) {
4261 		r->res_master_nodeid = from_nodeid;
4262 		r->res_nodeid = from_nodeid;
4263 		lkb->lkb_nodeid = from_nodeid;
4264 	}
4265 
4266 	/* this is the value returned from do_request() on the master */
4267 	result = from_dlm_errno(le32_to_cpu(ms->m_result));
4268 
4269 	switch (result) {
4270 	case -EAGAIN:
4271 		/* request would block (be queued) on remote master */
4272 		queue_cast(r, lkb, -EAGAIN);
4273 		confirm_master(r, -EAGAIN);
4274 		unhold_lkb(lkb); /* undoes create_lkb() */
4275 		break;
4276 
4277 	case -EINPROGRESS:
4278 	case 0:
4279 		/* request was queued or granted on remote master */
4280 		receive_flags_reply(lkb, ms, false);
4281 		lkb->lkb_remid = le32_to_cpu(ms->m_lkid);
4282 		if (is_altmode(lkb))
4283 			munge_altmode(lkb, ms);
4284 		if (result) {
4285 			add_lkb(r, lkb, DLM_LKSTS_WAITING);
4286 		} else {
4287 			grant_lock_pc(r, lkb, ms);
4288 			queue_cast(r, lkb, 0);
4289 		}
4290 		confirm_master(r, result);
4291 		break;
4292 
4293 	case -EBADR:
4294 	case -ENOTBLK:
4295 		/* find_rsb failed to find rsb or rsb wasn't master */
4296 		log_limit(ls, "receive_request_reply %x from %d %d "
4297 			  "master %d dir %d first %x %s", lkb->lkb_id,
4298 			  from_nodeid, result, r->res_master_nodeid,
4299 			  r->res_dir_nodeid, r->res_first_lkid, r->res_name);
4300 
4301 		if (r->res_dir_nodeid != dlm_our_nodeid() &&
4302 		    r->res_master_nodeid != dlm_our_nodeid()) {
4303 			/* cause _request_lock->set_master->send_lookup */
4304 			r->res_master_nodeid = 0;
4305 			r->res_nodeid = -1;
4306 			lkb->lkb_nodeid = -1;
4307 		}
4308 
4309 		if (is_overlap(lkb)) {
4310 			/* we'll ignore error in cancel/unlock reply */
4311 			queue_cast_overlap(r, lkb);
4312 			confirm_master(r, result);
4313 			unhold_lkb(lkb); /* undoes create_lkb() */
4314 		} else {
4315 			_request_lock(r, lkb);
4316 
4317 			if (r->res_master_nodeid == dlm_our_nodeid())
4318 				confirm_master(r, 0);
4319 		}
4320 		break;
4321 
4322 	default:
4323 		log_error(ls, "receive_request_reply %x error %d",
4324 			  lkb->lkb_id, result);
4325 	}
4326 
4327 	if ((result == 0 || result == -EINPROGRESS) &&
4328 	    test_and_clear_bit(DLM_IFL_OVERLAP_UNLOCK_BIT, &lkb->lkb_iflags)) {
4329 		log_debug(ls, "receive_request_reply %x result %d unlock",
4330 			  lkb->lkb_id, result);
4331 		clear_bit(DLM_IFL_OVERLAP_CANCEL_BIT, &lkb->lkb_iflags);
4332 		send_unlock(r, lkb);
4333 	} else if ((result == -EINPROGRESS) &&
4334 		   test_and_clear_bit(DLM_IFL_OVERLAP_CANCEL_BIT,
4335 				      &lkb->lkb_iflags)) {
4336 		log_debug(ls, "receive_request_reply %x cancel", lkb->lkb_id);
4337 		clear_bit(DLM_IFL_OVERLAP_UNLOCK_BIT, &lkb->lkb_iflags);
4338 		send_cancel(r, lkb);
4339 	} else {
4340 		clear_bit(DLM_IFL_OVERLAP_CANCEL_BIT, &lkb->lkb_iflags);
4341 		clear_bit(DLM_IFL_OVERLAP_UNLOCK_BIT, &lkb->lkb_iflags);
4342 	}
4343  out:
4344 	unlock_rsb(r);
4345 	put_rsb(r);
4346 	dlm_put_lkb(lkb);
4347 	return 0;
4348 }
4349 
4350 static void __receive_convert_reply(struct dlm_rsb *r, struct dlm_lkb *lkb,
4351 				    const struct dlm_message *ms, bool local)
4352 {
4353 	/* this is the value returned from do_convert() on the master */
4354 	switch (from_dlm_errno(le32_to_cpu(ms->m_result))) {
4355 	case -EAGAIN:
4356 		/* convert would block (be queued) on remote master */
4357 		queue_cast(r, lkb, -EAGAIN);
4358 		break;
4359 
4360 	case -EDEADLK:
4361 		receive_flags_reply(lkb, ms, local);
4362 		revert_lock_pc(r, lkb);
4363 		queue_cast(r, lkb, -EDEADLK);
4364 		break;
4365 
4366 	case -EINPROGRESS:
4367 		/* convert was queued on remote master */
4368 		receive_flags_reply(lkb, ms, local);
4369 		if (is_demoted(lkb))
4370 			munge_demoted(lkb);
4371 		del_lkb(r, lkb);
4372 		add_lkb(r, lkb, DLM_LKSTS_CONVERT);
4373 		break;
4374 
4375 	case 0:
4376 		/* convert was granted on remote master */
4377 		receive_flags_reply(lkb, ms, local);
4378 		if (is_demoted(lkb))
4379 			munge_demoted(lkb);
4380 		grant_lock_pc(r, lkb, ms);
4381 		queue_cast(r, lkb, 0);
4382 		break;
4383 
4384 	default:
4385 		log_error(r->res_ls, "receive_convert_reply %x remote %d %x %d",
4386 			  lkb->lkb_id, le32_to_cpu(ms->m_header.h_nodeid),
4387 			  le32_to_cpu(ms->m_lkid),
4388 			  from_dlm_errno(le32_to_cpu(ms->m_result)));
4389 		dlm_print_rsb(r);
4390 		dlm_print_lkb(lkb);
4391 	}
4392 }
4393 
4394 static void _receive_convert_reply(struct dlm_lkb *lkb,
4395 				   const struct dlm_message *ms, bool local)
4396 {
4397 	struct dlm_rsb *r = lkb->lkb_resource;
4398 	int error;
4399 
4400 	hold_rsb(r);
4401 	lock_rsb(r);
4402 
4403 	error = validate_message(lkb, ms);
4404 	if (error)
4405 		goto out;
4406 
4407 	/* local reply can happen with waiters_mutex held */
4408 	error = remove_from_waiters_ms(lkb, ms, local);
4409 	if (error)
4410 		goto out;
4411 
4412 	__receive_convert_reply(r, lkb, ms, local);
4413  out:
4414 	unlock_rsb(r);
4415 	put_rsb(r);
4416 }
4417 
4418 static int receive_convert_reply(struct dlm_ls *ls,
4419 				 const struct dlm_message *ms)
4420 {
4421 	struct dlm_lkb *lkb;
4422 	int error;
4423 
4424 	error = find_lkb(ls, le32_to_cpu(ms->m_remid), &lkb);
4425 	if (error)
4426 		return error;
4427 
4428 	_receive_convert_reply(lkb, ms, false);
4429 	dlm_put_lkb(lkb);
4430 	return 0;
4431 }
4432 
4433 static void _receive_unlock_reply(struct dlm_lkb *lkb,
4434 				  const struct dlm_message *ms, bool local)
4435 {
4436 	struct dlm_rsb *r = lkb->lkb_resource;
4437 	int error;
4438 
4439 	hold_rsb(r);
4440 	lock_rsb(r);
4441 
4442 	error = validate_message(lkb, ms);
4443 	if (error)
4444 		goto out;
4445 
4446 	/* local reply can happen with waiters_mutex held */
4447 	error = remove_from_waiters_ms(lkb, ms, local);
4448 	if (error)
4449 		goto out;
4450 
4451 	/* this is the value returned from do_unlock() on the master */
4452 
4453 	switch (from_dlm_errno(le32_to_cpu(ms->m_result))) {
4454 	case -DLM_EUNLOCK:
4455 		receive_flags_reply(lkb, ms, local);
4456 		remove_lock_pc(r, lkb);
4457 		queue_cast(r, lkb, -DLM_EUNLOCK);
4458 		break;
4459 	case -ENOENT:
4460 		break;
4461 	default:
4462 		log_error(r->res_ls, "receive_unlock_reply %x error %d",
4463 			  lkb->lkb_id, from_dlm_errno(le32_to_cpu(ms->m_result)));
4464 	}
4465  out:
4466 	unlock_rsb(r);
4467 	put_rsb(r);
4468 }
4469 
4470 static int receive_unlock_reply(struct dlm_ls *ls,
4471 				const struct dlm_message *ms)
4472 {
4473 	struct dlm_lkb *lkb;
4474 	int error;
4475 
4476 	error = find_lkb(ls, le32_to_cpu(ms->m_remid), &lkb);
4477 	if (error)
4478 		return error;
4479 
4480 	_receive_unlock_reply(lkb, ms, false);
4481 	dlm_put_lkb(lkb);
4482 	return 0;
4483 }
4484 
4485 static void _receive_cancel_reply(struct dlm_lkb *lkb,
4486 				  const struct dlm_message *ms, bool local)
4487 {
4488 	struct dlm_rsb *r = lkb->lkb_resource;
4489 	int error;
4490 
4491 	hold_rsb(r);
4492 	lock_rsb(r);
4493 
4494 	error = validate_message(lkb, ms);
4495 	if (error)
4496 		goto out;
4497 
4498 	/* local reply can happen with waiters_mutex held */
4499 	error = remove_from_waiters_ms(lkb, ms, local);
4500 	if (error)
4501 		goto out;
4502 
4503 	/* this is the value returned from do_cancel() on the master */
4504 
4505 	switch (from_dlm_errno(le32_to_cpu(ms->m_result))) {
4506 	case -DLM_ECANCEL:
4507 		receive_flags_reply(lkb, ms, local);
4508 		revert_lock_pc(r, lkb);
4509 		queue_cast(r, lkb, -DLM_ECANCEL);
4510 		break;
4511 	case 0:
4512 		break;
4513 	default:
4514 		log_error(r->res_ls, "receive_cancel_reply %x error %d",
4515 			  lkb->lkb_id,
4516 			  from_dlm_errno(le32_to_cpu(ms->m_result)));
4517 	}
4518  out:
4519 	unlock_rsb(r);
4520 	put_rsb(r);
4521 }
4522 
4523 static int receive_cancel_reply(struct dlm_ls *ls,
4524 				const struct dlm_message *ms)
4525 {
4526 	struct dlm_lkb *lkb;
4527 	int error;
4528 
4529 	error = find_lkb(ls, le32_to_cpu(ms->m_remid), &lkb);
4530 	if (error)
4531 		return error;
4532 
4533 	_receive_cancel_reply(lkb, ms, false);
4534 	dlm_put_lkb(lkb);
4535 	return 0;
4536 }
4537 
4538 static void receive_lookup_reply(struct dlm_ls *ls,
4539 				 const struct dlm_message *ms)
4540 {
4541 	struct dlm_lkb *lkb;
4542 	struct dlm_rsb *r;
4543 	int error, ret_nodeid;
4544 	int do_lookup_list = 0;
4545 
4546 	error = find_lkb(ls, le32_to_cpu(ms->m_lkid), &lkb);
4547 	if (error) {
4548 		log_error(ls, "%s no lkid %x", __func__,
4549 			  le32_to_cpu(ms->m_lkid));
4550 		return;
4551 	}
4552 
4553 	/* ms->m_result is the value returned by dlm_master_lookup on dir node
4554 	   FIXME: will a non-zero error ever be returned? */
4555 
4556 	r = lkb->lkb_resource;
4557 	hold_rsb(r);
4558 	lock_rsb(r);
4559 
4560 	error = remove_from_waiters(lkb, DLM_MSG_LOOKUP_REPLY);
4561 	if (error)
4562 		goto out;
4563 
4564 	ret_nodeid = le32_to_cpu(ms->m_nodeid);
4565 
4566 	/* We sometimes receive a request from the dir node for this
4567 	   rsb before we've received the dir node's loookup_reply for it.
4568 	   The request from the dir node implies we're the master, so we set
4569 	   ourself as master in receive_request_reply, and verify here that
4570 	   we are indeed the master. */
4571 
4572 	if (r->res_master_nodeid && (r->res_master_nodeid != ret_nodeid)) {
4573 		/* This should never happen */
4574 		log_error(ls, "receive_lookup_reply %x from %d ret %d "
4575 			  "master %d dir %d our %d first %x %s",
4576 			  lkb->lkb_id, le32_to_cpu(ms->m_header.h_nodeid),
4577 			  ret_nodeid, r->res_master_nodeid, r->res_dir_nodeid,
4578 			  dlm_our_nodeid(), r->res_first_lkid, r->res_name);
4579 	}
4580 
4581 	if (ret_nodeid == dlm_our_nodeid()) {
4582 		r->res_master_nodeid = ret_nodeid;
4583 		r->res_nodeid = 0;
4584 		do_lookup_list = 1;
4585 		r->res_first_lkid = 0;
4586 	} else if (ret_nodeid == -1) {
4587 		/* the remote node doesn't believe it's the dir node */
4588 		log_error(ls, "receive_lookup_reply %x from %d bad ret_nodeid",
4589 			  lkb->lkb_id, le32_to_cpu(ms->m_header.h_nodeid));
4590 		r->res_master_nodeid = 0;
4591 		r->res_nodeid = -1;
4592 		lkb->lkb_nodeid = -1;
4593 	} else {
4594 		/* set_master() will set lkb_nodeid from r */
4595 		r->res_master_nodeid = ret_nodeid;
4596 		r->res_nodeid = ret_nodeid;
4597 	}
4598 
4599 	if (is_overlap(lkb)) {
4600 		log_debug(ls, "receive_lookup_reply %x unlock %x",
4601 			  lkb->lkb_id, dlm_iflags_val(lkb));
4602 		queue_cast_overlap(r, lkb);
4603 		unhold_lkb(lkb); /* undoes create_lkb() */
4604 		goto out_list;
4605 	}
4606 
4607 	_request_lock(r, lkb);
4608 
4609  out_list:
4610 	if (do_lookup_list)
4611 		process_lookup_list(r);
4612  out:
4613 	unlock_rsb(r);
4614 	put_rsb(r);
4615 	dlm_put_lkb(lkb);
4616 }
4617 
4618 static void _receive_message(struct dlm_ls *ls, const struct dlm_message *ms,
4619 			     uint32_t saved_seq)
4620 {
4621 	int error = 0, noent = 0;
4622 
4623 	if (WARN_ON_ONCE(!dlm_is_member(ls, le32_to_cpu(ms->m_header.h_nodeid)))) {
4624 		log_limit(ls, "receive %d from non-member %d %x %x %d",
4625 			  le32_to_cpu(ms->m_type),
4626 			  le32_to_cpu(ms->m_header.h_nodeid),
4627 			  le32_to_cpu(ms->m_lkid), le32_to_cpu(ms->m_remid),
4628 			  from_dlm_errno(le32_to_cpu(ms->m_result)));
4629 		return;
4630 	}
4631 
4632 	switch (ms->m_type) {
4633 
4634 	/* messages sent to a master node */
4635 
4636 	case cpu_to_le32(DLM_MSG_REQUEST):
4637 		error = receive_request(ls, ms);
4638 		break;
4639 
4640 	case cpu_to_le32(DLM_MSG_CONVERT):
4641 		error = receive_convert(ls, ms);
4642 		break;
4643 
4644 	case cpu_to_le32(DLM_MSG_UNLOCK):
4645 		error = receive_unlock(ls, ms);
4646 		break;
4647 
4648 	case cpu_to_le32(DLM_MSG_CANCEL):
4649 		noent = 1;
4650 		error = receive_cancel(ls, ms);
4651 		break;
4652 
4653 	/* messages sent from a master node (replies to above) */
4654 
4655 	case cpu_to_le32(DLM_MSG_REQUEST_REPLY):
4656 		error = receive_request_reply(ls, ms);
4657 		break;
4658 
4659 	case cpu_to_le32(DLM_MSG_CONVERT_REPLY):
4660 		error = receive_convert_reply(ls, ms);
4661 		break;
4662 
4663 	case cpu_to_le32(DLM_MSG_UNLOCK_REPLY):
4664 		error = receive_unlock_reply(ls, ms);
4665 		break;
4666 
4667 	case cpu_to_le32(DLM_MSG_CANCEL_REPLY):
4668 		error = receive_cancel_reply(ls, ms);
4669 		break;
4670 
4671 	/* messages sent from a master node (only two types of async msg) */
4672 
4673 	case cpu_to_le32(DLM_MSG_GRANT):
4674 		noent = 1;
4675 		error = receive_grant(ls, ms);
4676 		break;
4677 
4678 	case cpu_to_le32(DLM_MSG_BAST):
4679 		noent = 1;
4680 		error = receive_bast(ls, ms);
4681 		break;
4682 
4683 	/* messages sent to a dir node */
4684 
4685 	case cpu_to_le32(DLM_MSG_LOOKUP):
4686 		receive_lookup(ls, ms);
4687 		break;
4688 
4689 	case cpu_to_le32(DLM_MSG_REMOVE):
4690 		receive_remove(ls, ms);
4691 		break;
4692 
4693 	/* messages sent from a dir node (remove has no reply) */
4694 
4695 	case cpu_to_le32(DLM_MSG_LOOKUP_REPLY):
4696 		receive_lookup_reply(ls, ms);
4697 		break;
4698 
4699 	/* other messages */
4700 
4701 	case cpu_to_le32(DLM_MSG_PURGE):
4702 		receive_purge(ls, ms);
4703 		break;
4704 
4705 	default:
4706 		log_error(ls, "unknown message type %d",
4707 			  le32_to_cpu(ms->m_type));
4708 	}
4709 
4710 	/*
4711 	 * When checking for ENOENT, we're checking the result of
4712 	 * find_lkb(m_remid):
4713 	 *
4714 	 * The lock id referenced in the message wasn't found.  This may
4715 	 * happen in normal usage for the async messages and cancel, so
4716 	 * only use log_debug for them.
4717 	 *
4718 	 * Some errors are expected and normal.
4719 	 */
4720 
4721 	if (error == -ENOENT && noent) {
4722 		log_debug(ls, "receive %d no %x remote %d %x saved_seq %u",
4723 			  le32_to_cpu(ms->m_type), le32_to_cpu(ms->m_remid),
4724 			  le32_to_cpu(ms->m_header.h_nodeid),
4725 			  le32_to_cpu(ms->m_lkid), saved_seq);
4726 	} else if (error == -ENOENT) {
4727 		log_error(ls, "receive %d no %x remote %d %x saved_seq %u",
4728 			  le32_to_cpu(ms->m_type), le32_to_cpu(ms->m_remid),
4729 			  le32_to_cpu(ms->m_header.h_nodeid),
4730 			  le32_to_cpu(ms->m_lkid), saved_seq);
4731 
4732 		if (ms->m_type == cpu_to_le32(DLM_MSG_CONVERT))
4733 			dlm_dump_rsb_hash(ls, le32_to_cpu(ms->m_hash));
4734 	}
4735 
4736 	if (error == -EINVAL) {
4737 		log_error(ls, "receive %d inval from %d lkid %x remid %x "
4738 			  "saved_seq %u",
4739 			  le32_to_cpu(ms->m_type),
4740 			  le32_to_cpu(ms->m_header.h_nodeid),
4741 			  le32_to_cpu(ms->m_lkid), le32_to_cpu(ms->m_remid),
4742 			  saved_seq);
4743 	}
4744 }
4745 
4746 /* If the lockspace is in recovery mode (locking stopped), then normal
4747    messages are saved on the requestqueue for processing after recovery is
4748    done.  When not in recovery mode, we wait for dlm_recoverd to drain saved
4749    messages off the requestqueue before we process new ones. This occurs right
4750    after recovery completes when we transition from saving all messages on
4751    requestqueue, to processing all the saved messages, to processing new
4752    messages as they arrive. */
4753 
4754 static void dlm_receive_message(struct dlm_ls *ls, const struct dlm_message *ms,
4755 				int nodeid)
4756 {
4757 	if (dlm_locking_stopped(ls)) {
4758 		/* If we were a member of this lockspace, left, and rejoined,
4759 		   other nodes may still be sending us messages from the
4760 		   lockspace generation before we left. */
4761 		if (WARN_ON_ONCE(!ls->ls_generation)) {
4762 			log_limit(ls, "receive %d from %d ignore old gen",
4763 				  le32_to_cpu(ms->m_type), nodeid);
4764 			return;
4765 		}
4766 
4767 		dlm_add_requestqueue(ls, nodeid, ms);
4768 	} else {
4769 		dlm_wait_requestqueue(ls);
4770 		_receive_message(ls, ms, 0);
4771 	}
4772 }
4773 
4774 /* This is called by dlm_recoverd to process messages that were saved on
4775    the requestqueue. */
4776 
4777 void dlm_receive_message_saved(struct dlm_ls *ls, const struct dlm_message *ms,
4778 			       uint32_t saved_seq)
4779 {
4780 	_receive_message(ls, ms, saved_seq);
4781 }
4782 
4783 /* This is called by the midcomms layer when something is received for
4784    the lockspace.  It could be either a MSG (normal message sent as part of
4785    standard locking activity) or an RCOM (recovery message sent as part of
4786    lockspace recovery). */
4787 
4788 void dlm_receive_buffer(const union dlm_packet *p, int nodeid)
4789 {
4790 	const struct dlm_header *hd = &p->header;
4791 	struct dlm_ls *ls;
4792 	int type = 0;
4793 
4794 	switch (hd->h_cmd) {
4795 	case DLM_MSG:
4796 		type = le32_to_cpu(p->message.m_type);
4797 		break;
4798 	case DLM_RCOM:
4799 		type = le32_to_cpu(p->rcom.rc_type);
4800 		break;
4801 	default:
4802 		log_print("invalid h_cmd %d from %u", hd->h_cmd, nodeid);
4803 		return;
4804 	}
4805 
4806 	if (le32_to_cpu(hd->h_nodeid) != nodeid) {
4807 		log_print("invalid h_nodeid %d from %d lockspace %x",
4808 			  le32_to_cpu(hd->h_nodeid), nodeid,
4809 			  le32_to_cpu(hd->u.h_lockspace));
4810 		return;
4811 	}
4812 
4813 	ls = dlm_find_lockspace_global(le32_to_cpu(hd->u.h_lockspace));
4814 	if (!ls) {
4815 		if (dlm_config.ci_log_debug) {
4816 			printk_ratelimited(KERN_DEBUG "dlm: invalid lockspace "
4817 				"%u from %d cmd %d type %d\n",
4818 				le32_to_cpu(hd->u.h_lockspace), nodeid,
4819 				hd->h_cmd, type);
4820 		}
4821 
4822 		if (hd->h_cmd == DLM_RCOM && type == DLM_RCOM_STATUS)
4823 			dlm_send_ls_not_ready(nodeid, &p->rcom);
4824 		return;
4825 	}
4826 
4827 	/* this rwsem allows dlm_ls_stop() to wait for all dlm_recv threads to
4828 	   be inactive (in this ls) before transitioning to recovery mode */
4829 
4830 	down_read(&ls->ls_recv_active);
4831 	if (hd->h_cmd == DLM_MSG)
4832 		dlm_receive_message(ls, &p->message, nodeid);
4833 	else if (hd->h_cmd == DLM_RCOM)
4834 		dlm_receive_rcom(ls, &p->rcom, nodeid);
4835 	else
4836 		log_error(ls, "invalid h_cmd %d from %d lockspace %x",
4837 			  hd->h_cmd, nodeid, le32_to_cpu(hd->u.h_lockspace));
4838 	up_read(&ls->ls_recv_active);
4839 
4840 	dlm_put_lockspace(ls);
4841 }
4842 
4843 static void recover_convert_waiter(struct dlm_ls *ls, struct dlm_lkb *lkb,
4844 				   struct dlm_message *ms_local)
4845 {
4846 	if (middle_conversion(lkb)) {
4847 		hold_lkb(lkb);
4848 		memset(ms_local, 0, sizeof(struct dlm_message));
4849 		ms_local->m_type = cpu_to_le32(DLM_MSG_CONVERT_REPLY);
4850 		ms_local->m_result = cpu_to_le32(to_dlm_errno(-EINPROGRESS));
4851 		ms_local->m_header.h_nodeid = cpu_to_le32(lkb->lkb_nodeid);
4852 		_receive_convert_reply(lkb, ms_local, true);
4853 
4854 		/* Same special case as in receive_rcom_lock_args() */
4855 		lkb->lkb_grmode = DLM_LOCK_IV;
4856 		rsb_set_flag(lkb->lkb_resource, RSB_RECOVER_CONVERT);
4857 		unhold_lkb(lkb);
4858 
4859 	} else if (lkb->lkb_rqmode >= lkb->lkb_grmode) {
4860 		set_bit(DLM_IFL_RESEND_BIT, &lkb->lkb_iflags);
4861 	}
4862 
4863 	/* lkb->lkb_rqmode < lkb->lkb_grmode shouldn't happen since down
4864 	   conversions are async; there's no reply from the remote master */
4865 }
4866 
4867 /* A waiting lkb needs recovery if the master node has failed, or
4868    the master node is changing (only when no directory is used) */
4869 
4870 static int waiter_needs_recovery(struct dlm_ls *ls, struct dlm_lkb *lkb,
4871 				 int dir_nodeid)
4872 {
4873 	if (dlm_no_directory(ls))
4874 		return 1;
4875 
4876 	if (dlm_is_removed(ls, lkb->lkb_wait_nodeid))
4877 		return 1;
4878 
4879 	return 0;
4880 }
4881 
4882 /* Recovery for locks that are waiting for replies from nodes that are now
4883    gone.  We can just complete unlocks and cancels by faking a reply from the
4884    dead node.  Requests and up-conversions we flag to be resent after
4885    recovery.  Down-conversions can just be completed with a fake reply like
4886    unlocks.  Conversions between PR and CW need special attention. */
4887 
4888 void dlm_recover_waiters_pre(struct dlm_ls *ls)
4889 {
4890 	struct dlm_lkb *lkb, *safe;
4891 	struct dlm_message *ms_local;
4892 	int wait_type, local_unlock_result, local_cancel_result;
4893 	int dir_nodeid;
4894 
4895 	ms_local = kmalloc(sizeof(*ms_local), GFP_KERNEL);
4896 	if (!ms_local)
4897 		return;
4898 
4899 	mutex_lock(&ls->ls_waiters_mutex);
4900 
4901 	list_for_each_entry_safe(lkb, safe, &ls->ls_waiters, lkb_wait_reply) {
4902 
4903 		dir_nodeid = dlm_dir_nodeid(lkb->lkb_resource);
4904 
4905 		/* exclude debug messages about unlocks because there can be so
4906 		   many and they aren't very interesting */
4907 
4908 		if (lkb->lkb_wait_type != DLM_MSG_UNLOCK) {
4909 			log_debug(ls, "waiter %x remote %x msg %d r_nodeid %d "
4910 				  "lkb_nodeid %d wait_nodeid %d dir_nodeid %d",
4911 				  lkb->lkb_id,
4912 				  lkb->lkb_remid,
4913 				  lkb->lkb_wait_type,
4914 				  lkb->lkb_resource->res_nodeid,
4915 				  lkb->lkb_nodeid,
4916 				  lkb->lkb_wait_nodeid,
4917 				  dir_nodeid);
4918 		}
4919 
4920 		/* all outstanding lookups, regardless of destination  will be
4921 		   resent after recovery is done */
4922 
4923 		if (lkb->lkb_wait_type == DLM_MSG_LOOKUP) {
4924 			set_bit(DLM_IFL_RESEND_BIT, &lkb->lkb_iflags);
4925 			continue;
4926 		}
4927 
4928 		if (!waiter_needs_recovery(ls, lkb, dir_nodeid))
4929 			continue;
4930 
4931 		wait_type = lkb->lkb_wait_type;
4932 		local_unlock_result = -DLM_EUNLOCK;
4933 		local_cancel_result = -DLM_ECANCEL;
4934 
4935 		/* Main reply may have been received leaving a zero wait_type,
4936 		   but a reply for the overlapping op may not have been
4937 		   received.  In that case we need to fake the appropriate
4938 		   reply for the overlap op. */
4939 
4940 		if (!wait_type) {
4941 			if (is_overlap_cancel(lkb)) {
4942 				wait_type = DLM_MSG_CANCEL;
4943 				if (lkb->lkb_grmode == DLM_LOCK_IV)
4944 					local_cancel_result = 0;
4945 			}
4946 			if (is_overlap_unlock(lkb)) {
4947 				wait_type = DLM_MSG_UNLOCK;
4948 				if (lkb->lkb_grmode == DLM_LOCK_IV)
4949 					local_unlock_result = -ENOENT;
4950 			}
4951 
4952 			log_debug(ls, "rwpre overlap %x %x %d %d %d",
4953 				  lkb->lkb_id, dlm_iflags_val(lkb), wait_type,
4954 				  local_cancel_result, local_unlock_result);
4955 		}
4956 
4957 		switch (wait_type) {
4958 
4959 		case DLM_MSG_REQUEST:
4960 			set_bit(DLM_IFL_RESEND_BIT, &lkb->lkb_iflags);
4961 			break;
4962 
4963 		case DLM_MSG_CONVERT:
4964 			recover_convert_waiter(ls, lkb, ms_local);
4965 			break;
4966 
4967 		case DLM_MSG_UNLOCK:
4968 			hold_lkb(lkb);
4969 			memset(ms_local, 0, sizeof(struct dlm_message));
4970 			ms_local->m_type = cpu_to_le32(DLM_MSG_UNLOCK_REPLY);
4971 			ms_local->m_result = cpu_to_le32(to_dlm_errno(local_unlock_result));
4972 			ms_local->m_header.h_nodeid = cpu_to_le32(lkb->lkb_nodeid);
4973 			_receive_unlock_reply(lkb, ms_local, true);
4974 			dlm_put_lkb(lkb);
4975 			break;
4976 
4977 		case DLM_MSG_CANCEL:
4978 			hold_lkb(lkb);
4979 			memset(ms_local, 0, sizeof(struct dlm_message));
4980 			ms_local->m_type = cpu_to_le32(DLM_MSG_CANCEL_REPLY);
4981 			ms_local->m_result = cpu_to_le32(to_dlm_errno(local_cancel_result));
4982 			ms_local->m_header.h_nodeid = cpu_to_le32(lkb->lkb_nodeid);
4983 			_receive_cancel_reply(lkb, ms_local, true);
4984 			dlm_put_lkb(lkb);
4985 			break;
4986 
4987 		default:
4988 			log_error(ls, "invalid lkb wait_type %d %d",
4989 				  lkb->lkb_wait_type, wait_type);
4990 		}
4991 		schedule();
4992 	}
4993 	mutex_unlock(&ls->ls_waiters_mutex);
4994 	kfree(ms_local);
4995 }
4996 
4997 static struct dlm_lkb *find_resend_waiter(struct dlm_ls *ls)
4998 {
4999 	struct dlm_lkb *lkb = NULL, *iter;
5000 
5001 	mutex_lock(&ls->ls_waiters_mutex);
5002 	list_for_each_entry(iter, &ls->ls_waiters, lkb_wait_reply) {
5003 		if (test_bit(DLM_IFL_RESEND_BIT, &iter->lkb_iflags)) {
5004 			hold_lkb(iter);
5005 			lkb = iter;
5006 			break;
5007 		}
5008 	}
5009 	mutex_unlock(&ls->ls_waiters_mutex);
5010 
5011 	return lkb;
5012 }
5013 
5014 /* Deal with lookups and lkb's marked RESEND from _pre.  We may now be the
5015    master or dir-node for r.  Processing the lkb may result in it being placed
5016    back on waiters. */
5017 
5018 /* We do this after normal locking has been enabled and any saved messages
5019    (in requestqueue) have been processed.  We should be confident that at
5020    this point we won't get or process a reply to any of these waiting
5021    operations.  But, new ops may be coming in on the rsbs/locks here from
5022    userspace or remotely. */
5023 
5024 /* there may have been an overlap unlock/cancel prior to recovery or after
5025    recovery.  if before, the lkb may still have a pos wait_count; if after, the
5026    overlap flag would just have been set and nothing new sent.  we can be
5027    confident here than any replies to either the initial op or overlap ops
5028    prior to recovery have been received. */
5029 
5030 int dlm_recover_waiters_post(struct dlm_ls *ls)
5031 {
5032 	struct dlm_lkb *lkb;
5033 	struct dlm_rsb *r;
5034 	int error = 0, mstype, err, oc, ou;
5035 
5036 	while (1) {
5037 		if (dlm_locking_stopped(ls)) {
5038 			log_debug(ls, "recover_waiters_post aborted");
5039 			error = -EINTR;
5040 			break;
5041 		}
5042 
5043 		lkb = find_resend_waiter(ls);
5044 		if (!lkb)
5045 			break;
5046 
5047 		r = lkb->lkb_resource;
5048 		hold_rsb(r);
5049 		lock_rsb(r);
5050 
5051 		mstype = lkb->lkb_wait_type;
5052 		oc = test_and_clear_bit(DLM_IFL_OVERLAP_CANCEL_BIT,
5053 					&lkb->lkb_iflags);
5054 		ou = test_and_clear_bit(DLM_IFL_OVERLAP_UNLOCK_BIT,
5055 					&lkb->lkb_iflags);
5056 		err = 0;
5057 
5058 		log_debug(ls, "waiter %x remote %x msg %d r_nodeid %d "
5059 			  "lkb_nodeid %d wait_nodeid %d dir_nodeid %d "
5060 			  "overlap %d %d", lkb->lkb_id, lkb->lkb_remid, mstype,
5061 			  r->res_nodeid, lkb->lkb_nodeid, lkb->lkb_wait_nodeid,
5062 			  dlm_dir_nodeid(r), oc, ou);
5063 
5064 		/* At this point we assume that we won't get a reply to any
5065 		   previous op or overlap op on this lock.  First, do a big
5066 		   remove_from_waiters() for all previous ops. */
5067 
5068 		clear_bit(DLM_IFL_RESEND_BIT, &lkb->lkb_iflags);
5069 		lkb->lkb_wait_type = 0;
5070 		/* drop all wait_count references we still
5071 		 * hold a reference for this iteration.
5072 		 */
5073 		while (!atomic_dec_and_test(&lkb->lkb_wait_count))
5074 			unhold_lkb(lkb);
5075 
5076 		mutex_lock(&ls->ls_waiters_mutex);
5077 		list_del_init(&lkb->lkb_wait_reply);
5078 		mutex_unlock(&ls->ls_waiters_mutex);
5079 
5080 		if (oc || ou) {
5081 			/* do an unlock or cancel instead of resending */
5082 			switch (mstype) {
5083 			case DLM_MSG_LOOKUP:
5084 			case DLM_MSG_REQUEST:
5085 				queue_cast(r, lkb, ou ? -DLM_EUNLOCK :
5086 							-DLM_ECANCEL);
5087 				unhold_lkb(lkb); /* undoes create_lkb() */
5088 				break;
5089 			case DLM_MSG_CONVERT:
5090 				if (oc) {
5091 					queue_cast(r, lkb, -DLM_ECANCEL);
5092 				} else {
5093 					lkb->lkb_exflags |= DLM_LKF_FORCEUNLOCK;
5094 					_unlock_lock(r, lkb);
5095 				}
5096 				break;
5097 			default:
5098 				err = 1;
5099 			}
5100 		} else {
5101 			switch (mstype) {
5102 			case DLM_MSG_LOOKUP:
5103 			case DLM_MSG_REQUEST:
5104 				_request_lock(r, lkb);
5105 				if (is_master(r))
5106 					confirm_master(r, 0);
5107 				break;
5108 			case DLM_MSG_CONVERT:
5109 				_convert_lock(r, lkb);
5110 				break;
5111 			default:
5112 				err = 1;
5113 			}
5114 		}
5115 
5116 		if (err) {
5117 			log_error(ls, "waiter %x msg %d r_nodeid %d "
5118 				  "dir_nodeid %d overlap %d %d",
5119 				  lkb->lkb_id, mstype, r->res_nodeid,
5120 				  dlm_dir_nodeid(r), oc, ou);
5121 		}
5122 		unlock_rsb(r);
5123 		put_rsb(r);
5124 		dlm_put_lkb(lkb);
5125 	}
5126 
5127 	return error;
5128 }
5129 
5130 static void purge_mstcpy_list(struct dlm_ls *ls, struct dlm_rsb *r,
5131 			      struct list_head *list)
5132 {
5133 	struct dlm_lkb *lkb, *safe;
5134 
5135 	list_for_each_entry_safe(lkb, safe, list, lkb_statequeue) {
5136 		if (!is_master_copy(lkb))
5137 			continue;
5138 
5139 		/* don't purge lkbs we've added in recover_master_copy for
5140 		   the current recovery seq */
5141 
5142 		if (lkb->lkb_recover_seq == ls->ls_recover_seq)
5143 			continue;
5144 
5145 		del_lkb(r, lkb);
5146 
5147 		/* this put should free the lkb */
5148 		if (!dlm_put_lkb(lkb))
5149 			log_error(ls, "purged mstcpy lkb not released");
5150 	}
5151 }
5152 
5153 void dlm_purge_mstcpy_locks(struct dlm_rsb *r)
5154 {
5155 	struct dlm_ls *ls = r->res_ls;
5156 
5157 	purge_mstcpy_list(ls, r, &r->res_grantqueue);
5158 	purge_mstcpy_list(ls, r, &r->res_convertqueue);
5159 	purge_mstcpy_list(ls, r, &r->res_waitqueue);
5160 }
5161 
5162 static void purge_dead_list(struct dlm_ls *ls, struct dlm_rsb *r,
5163 			    struct list_head *list,
5164 			    int nodeid_gone, unsigned int *count)
5165 {
5166 	struct dlm_lkb *lkb, *safe;
5167 
5168 	list_for_each_entry_safe(lkb, safe, list, lkb_statequeue) {
5169 		if (!is_master_copy(lkb))
5170 			continue;
5171 
5172 		if ((lkb->lkb_nodeid == nodeid_gone) ||
5173 		    dlm_is_removed(ls, lkb->lkb_nodeid)) {
5174 
5175 			/* tell recover_lvb to invalidate the lvb
5176 			   because a node holding EX/PW failed */
5177 			if ((lkb->lkb_exflags & DLM_LKF_VALBLK) &&
5178 			    (lkb->lkb_grmode >= DLM_LOCK_PW)) {
5179 				rsb_set_flag(r, RSB_RECOVER_LVB_INVAL);
5180 			}
5181 
5182 			del_lkb(r, lkb);
5183 
5184 			/* this put should free the lkb */
5185 			if (!dlm_put_lkb(lkb))
5186 				log_error(ls, "purged dead lkb not released");
5187 
5188 			rsb_set_flag(r, RSB_RECOVER_GRANT);
5189 
5190 			(*count)++;
5191 		}
5192 	}
5193 }
5194 
5195 /* Get rid of locks held by nodes that are gone. */
5196 
5197 void dlm_recover_purge(struct dlm_ls *ls)
5198 {
5199 	struct dlm_rsb *r;
5200 	struct dlm_member *memb;
5201 	int nodes_count = 0;
5202 	int nodeid_gone = 0;
5203 	unsigned int lkb_count = 0;
5204 
5205 	/* cache one removed nodeid to optimize the common
5206 	   case of a single node removed */
5207 
5208 	list_for_each_entry(memb, &ls->ls_nodes_gone, list) {
5209 		nodes_count++;
5210 		nodeid_gone = memb->nodeid;
5211 	}
5212 
5213 	if (!nodes_count)
5214 		return;
5215 
5216 	down_write(&ls->ls_root_sem);
5217 	list_for_each_entry(r, &ls->ls_root_list, res_root_list) {
5218 		hold_rsb(r);
5219 		lock_rsb(r);
5220 		if (is_master(r)) {
5221 			purge_dead_list(ls, r, &r->res_grantqueue,
5222 					nodeid_gone, &lkb_count);
5223 			purge_dead_list(ls, r, &r->res_convertqueue,
5224 					nodeid_gone, &lkb_count);
5225 			purge_dead_list(ls, r, &r->res_waitqueue,
5226 					nodeid_gone, &lkb_count);
5227 		}
5228 		unlock_rsb(r);
5229 		unhold_rsb(r);
5230 		cond_resched();
5231 	}
5232 	up_write(&ls->ls_root_sem);
5233 
5234 	if (lkb_count)
5235 		log_rinfo(ls, "dlm_recover_purge %u locks for %u nodes",
5236 			  lkb_count, nodes_count);
5237 }
5238 
5239 static struct dlm_rsb *find_grant_rsb(struct dlm_ls *ls, int bucket)
5240 {
5241 	struct rb_node *n;
5242 	struct dlm_rsb *r;
5243 
5244 	spin_lock(&ls->ls_rsbtbl[bucket].lock);
5245 	for (n = rb_first(&ls->ls_rsbtbl[bucket].keep); n; n = rb_next(n)) {
5246 		r = rb_entry(n, struct dlm_rsb, res_hashnode);
5247 
5248 		if (!rsb_flag(r, RSB_RECOVER_GRANT))
5249 			continue;
5250 		if (!is_master(r)) {
5251 			rsb_clear_flag(r, RSB_RECOVER_GRANT);
5252 			continue;
5253 		}
5254 		hold_rsb(r);
5255 		spin_unlock(&ls->ls_rsbtbl[bucket].lock);
5256 		return r;
5257 	}
5258 	spin_unlock(&ls->ls_rsbtbl[bucket].lock);
5259 	return NULL;
5260 }
5261 
5262 /*
5263  * Attempt to grant locks on resources that we are the master of.
5264  * Locks may have become grantable during recovery because locks
5265  * from departed nodes have been purged (or not rebuilt), allowing
5266  * previously blocked locks to now be granted.  The subset of rsb's
5267  * we are interested in are those with lkb's on either the convert or
5268  * waiting queues.
5269  *
5270  * Simplest would be to go through each master rsb and check for non-empty
5271  * convert or waiting queues, and attempt to grant on those rsbs.
5272  * Checking the queues requires lock_rsb, though, for which we'd need
5273  * to release the rsbtbl lock.  This would make iterating through all
5274  * rsb's very inefficient.  So, we rely on earlier recovery routines
5275  * to set RECOVER_GRANT on any rsb's that we should attempt to grant
5276  * locks for.
5277  */
5278 
5279 void dlm_recover_grant(struct dlm_ls *ls)
5280 {
5281 	struct dlm_rsb *r;
5282 	int bucket = 0;
5283 	unsigned int count = 0;
5284 	unsigned int rsb_count = 0;
5285 	unsigned int lkb_count = 0;
5286 
5287 	while (1) {
5288 		r = find_grant_rsb(ls, bucket);
5289 		if (!r) {
5290 			if (bucket == ls->ls_rsbtbl_size - 1)
5291 				break;
5292 			bucket++;
5293 			continue;
5294 		}
5295 		rsb_count++;
5296 		count = 0;
5297 		lock_rsb(r);
5298 		/* the RECOVER_GRANT flag is checked in the grant path */
5299 		grant_pending_locks(r, &count);
5300 		rsb_clear_flag(r, RSB_RECOVER_GRANT);
5301 		lkb_count += count;
5302 		confirm_master(r, 0);
5303 		unlock_rsb(r);
5304 		put_rsb(r);
5305 		cond_resched();
5306 	}
5307 
5308 	if (lkb_count)
5309 		log_rinfo(ls, "dlm_recover_grant %u locks on %u resources",
5310 			  lkb_count, rsb_count);
5311 }
5312 
5313 static struct dlm_lkb *search_remid_list(struct list_head *head, int nodeid,
5314 					 uint32_t remid)
5315 {
5316 	struct dlm_lkb *lkb;
5317 
5318 	list_for_each_entry(lkb, head, lkb_statequeue) {
5319 		if (lkb->lkb_nodeid == nodeid && lkb->lkb_remid == remid)
5320 			return lkb;
5321 	}
5322 	return NULL;
5323 }
5324 
5325 static struct dlm_lkb *search_remid(struct dlm_rsb *r, int nodeid,
5326 				    uint32_t remid)
5327 {
5328 	struct dlm_lkb *lkb;
5329 
5330 	lkb = search_remid_list(&r->res_grantqueue, nodeid, remid);
5331 	if (lkb)
5332 		return lkb;
5333 	lkb = search_remid_list(&r->res_convertqueue, nodeid, remid);
5334 	if (lkb)
5335 		return lkb;
5336 	lkb = search_remid_list(&r->res_waitqueue, nodeid, remid);
5337 	if (lkb)
5338 		return lkb;
5339 	return NULL;
5340 }
5341 
5342 /* needs at least dlm_rcom + rcom_lock */
5343 static int receive_rcom_lock_args(struct dlm_ls *ls, struct dlm_lkb *lkb,
5344 				  struct dlm_rsb *r, const struct dlm_rcom *rc)
5345 {
5346 	struct rcom_lock *rl = (struct rcom_lock *) rc->rc_buf;
5347 
5348 	lkb->lkb_nodeid = le32_to_cpu(rc->rc_header.h_nodeid);
5349 	lkb->lkb_ownpid = le32_to_cpu(rl->rl_ownpid);
5350 	lkb->lkb_remid = le32_to_cpu(rl->rl_lkid);
5351 	lkb->lkb_exflags = le32_to_cpu(rl->rl_exflags);
5352 	dlm_set_dflags_val(lkb, le32_to_cpu(rl->rl_flags));
5353 	set_bit(DLM_IFL_MSTCPY_BIT, &lkb->lkb_iflags);
5354 	lkb->lkb_lvbseq = le32_to_cpu(rl->rl_lvbseq);
5355 	lkb->lkb_rqmode = rl->rl_rqmode;
5356 	lkb->lkb_grmode = rl->rl_grmode;
5357 	/* don't set lkb_status because add_lkb wants to itself */
5358 
5359 	lkb->lkb_bastfn = (rl->rl_asts & DLM_CB_BAST) ? &fake_bastfn : NULL;
5360 	lkb->lkb_astfn = (rl->rl_asts & DLM_CB_CAST) ? &fake_astfn : NULL;
5361 
5362 	if (lkb->lkb_exflags & DLM_LKF_VALBLK) {
5363 		int lvblen = le16_to_cpu(rc->rc_header.h_length) -
5364 			sizeof(struct dlm_rcom) - sizeof(struct rcom_lock);
5365 		if (lvblen > ls->ls_lvblen)
5366 			return -EINVAL;
5367 		lkb->lkb_lvbptr = dlm_allocate_lvb(ls);
5368 		if (!lkb->lkb_lvbptr)
5369 			return -ENOMEM;
5370 		memcpy(lkb->lkb_lvbptr, rl->rl_lvb, lvblen);
5371 	}
5372 
5373 	/* Conversions between PR and CW (middle modes) need special handling.
5374 	   The real granted mode of these converting locks cannot be determined
5375 	   until all locks have been rebuilt on the rsb (recover_conversion) */
5376 
5377 	if (rl->rl_wait_type == cpu_to_le16(DLM_MSG_CONVERT) &&
5378 	    middle_conversion(lkb)) {
5379 		rl->rl_status = DLM_LKSTS_CONVERT;
5380 		lkb->lkb_grmode = DLM_LOCK_IV;
5381 		rsb_set_flag(r, RSB_RECOVER_CONVERT);
5382 	}
5383 
5384 	return 0;
5385 }
5386 
5387 /* This lkb may have been recovered in a previous aborted recovery so we need
5388    to check if the rsb already has an lkb with the given remote nodeid/lkid.
5389    If so we just send back a standard reply.  If not, we create a new lkb with
5390    the given values and send back our lkid.  We send back our lkid by sending
5391    back the rcom_lock struct we got but with the remid field filled in. */
5392 
5393 /* needs at least dlm_rcom + rcom_lock */
5394 int dlm_recover_master_copy(struct dlm_ls *ls, const struct dlm_rcom *rc,
5395 			    __le32 *rl_remid, __le32 *rl_result)
5396 {
5397 	struct rcom_lock *rl = (struct rcom_lock *) rc->rc_buf;
5398 	struct dlm_rsb *r;
5399 	struct dlm_lkb *lkb;
5400 	uint32_t remid = 0;
5401 	int from_nodeid = le32_to_cpu(rc->rc_header.h_nodeid);
5402 	int error;
5403 
5404 	/* init rl_remid with rcom lock rl_remid */
5405 	*rl_remid = rl->rl_remid;
5406 
5407 	if (rl->rl_parent_lkid) {
5408 		error = -EOPNOTSUPP;
5409 		goto out;
5410 	}
5411 
5412 	remid = le32_to_cpu(rl->rl_lkid);
5413 
5414 	/* In general we expect the rsb returned to be R_MASTER, but we don't
5415 	   have to require it.  Recovery of masters on one node can overlap
5416 	   recovery of locks on another node, so one node can send us MSTCPY
5417 	   locks before we've made ourselves master of this rsb.  We can still
5418 	   add new MSTCPY locks that we receive here without any harm; when
5419 	   we make ourselves master, dlm_recover_masters() won't touch the
5420 	   MSTCPY locks we've received early. */
5421 
5422 	error = find_rsb(ls, rl->rl_name, le16_to_cpu(rl->rl_namelen),
5423 			 from_nodeid, R_RECEIVE_RECOVER, &r);
5424 	if (error)
5425 		goto out;
5426 
5427 	lock_rsb(r);
5428 
5429 	if (dlm_no_directory(ls) && (dlm_dir_nodeid(r) != dlm_our_nodeid())) {
5430 		log_error(ls, "dlm_recover_master_copy remote %d %x not dir",
5431 			  from_nodeid, remid);
5432 		error = -EBADR;
5433 		goto out_unlock;
5434 	}
5435 
5436 	lkb = search_remid(r, from_nodeid, remid);
5437 	if (lkb) {
5438 		error = -EEXIST;
5439 		goto out_remid;
5440 	}
5441 
5442 	error = create_lkb(ls, &lkb);
5443 	if (error)
5444 		goto out_unlock;
5445 
5446 	error = receive_rcom_lock_args(ls, lkb, r, rc);
5447 	if (error) {
5448 		__put_lkb(ls, lkb);
5449 		goto out_unlock;
5450 	}
5451 
5452 	attach_lkb(r, lkb);
5453 	add_lkb(r, lkb, rl->rl_status);
5454 	ls->ls_recover_locks_in++;
5455 
5456 	if (!list_empty(&r->res_waitqueue) || !list_empty(&r->res_convertqueue))
5457 		rsb_set_flag(r, RSB_RECOVER_GRANT);
5458 
5459  out_remid:
5460 	/* this is the new value returned to the lock holder for
5461 	   saving in its process-copy lkb */
5462 	*rl_remid = cpu_to_le32(lkb->lkb_id);
5463 
5464 	lkb->lkb_recover_seq = ls->ls_recover_seq;
5465 
5466  out_unlock:
5467 	unlock_rsb(r);
5468 	put_rsb(r);
5469  out:
5470 	if (error && error != -EEXIST)
5471 		log_rinfo(ls, "dlm_recover_master_copy remote %d %x error %d",
5472 			  from_nodeid, remid, error);
5473 	*rl_result = cpu_to_le32(error);
5474 	return error;
5475 }
5476 
5477 /* needs at least dlm_rcom + rcom_lock */
5478 int dlm_recover_process_copy(struct dlm_ls *ls, const struct dlm_rcom *rc,
5479 			     uint64_t seq)
5480 {
5481 	struct rcom_lock *rl = (struct rcom_lock *) rc->rc_buf;
5482 	struct dlm_rsb *r;
5483 	struct dlm_lkb *lkb;
5484 	uint32_t lkid, remid;
5485 	int error, result;
5486 
5487 	lkid = le32_to_cpu(rl->rl_lkid);
5488 	remid = le32_to_cpu(rl->rl_remid);
5489 	result = le32_to_cpu(rl->rl_result);
5490 
5491 	error = find_lkb(ls, lkid, &lkb);
5492 	if (error) {
5493 		log_error(ls, "dlm_recover_process_copy no %x remote %d %x %d",
5494 			  lkid, le32_to_cpu(rc->rc_header.h_nodeid), remid,
5495 			  result);
5496 		return error;
5497 	}
5498 
5499 	r = lkb->lkb_resource;
5500 	hold_rsb(r);
5501 	lock_rsb(r);
5502 
5503 	if (!is_process_copy(lkb)) {
5504 		log_error(ls, "dlm_recover_process_copy bad %x remote %d %x %d",
5505 			  lkid, le32_to_cpu(rc->rc_header.h_nodeid), remid,
5506 			  result);
5507 		dlm_dump_rsb(r);
5508 		unlock_rsb(r);
5509 		put_rsb(r);
5510 		dlm_put_lkb(lkb);
5511 		return -EINVAL;
5512 	}
5513 
5514 	switch (result) {
5515 	case -EBADR:
5516 		/* There's a chance the new master received our lock before
5517 		   dlm_recover_master_reply(), this wouldn't happen if we did
5518 		   a barrier between recover_masters and recover_locks. */
5519 
5520 		log_debug(ls, "dlm_recover_process_copy %x remote %d %x %d",
5521 			  lkid, le32_to_cpu(rc->rc_header.h_nodeid), remid,
5522 			  result);
5523 
5524 		dlm_send_rcom_lock(r, lkb, seq);
5525 		goto out;
5526 	case -EEXIST:
5527 	case 0:
5528 		lkb->lkb_remid = remid;
5529 		break;
5530 	default:
5531 		log_error(ls, "dlm_recover_process_copy %x remote %d %x %d unk",
5532 			  lkid, le32_to_cpu(rc->rc_header.h_nodeid), remid,
5533 			  result);
5534 	}
5535 
5536 	/* an ack for dlm_recover_locks() which waits for replies from
5537 	   all the locks it sends to new masters */
5538 	dlm_recovered_lock(r);
5539  out:
5540 	unlock_rsb(r);
5541 	put_rsb(r);
5542 	dlm_put_lkb(lkb);
5543 
5544 	return 0;
5545 }
5546 
5547 int dlm_user_request(struct dlm_ls *ls, struct dlm_user_args *ua,
5548 		     int mode, uint32_t flags, void *name, unsigned int namelen)
5549 {
5550 	struct dlm_lkb *lkb;
5551 	struct dlm_args args;
5552 	bool do_put = true;
5553 	int error;
5554 
5555 	dlm_lock_recovery(ls);
5556 
5557 	error = create_lkb(ls, &lkb);
5558 	if (error) {
5559 		kfree(ua);
5560 		goto out;
5561 	}
5562 
5563 	trace_dlm_lock_start(ls, lkb, name, namelen, mode, flags);
5564 
5565 	if (flags & DLM_LKF_VALBLK) {
5566 		ua->lksb.sb_lvbptr = kzalloc(DLM_USER_LVB_LEN, GFP_NOFS);
5567 		if (!ua->lksb.sb_lvbptr) {
5568 			kfree(ua);
5569 			error = -ENOMEM;
5570 			goto out_put;
5571 		}
5572 	}
5573 	error = set_lock_args(mode, &ua->lksb, flags, namelen, fake_astfn, ua,
5574 			      fake_bastfn, &args);
5575 	if (error) {
5576 		kfree(ua->lksb.sb_lvbptr);
5577 		ua->lksb.sb_lvbptr = NULL;
5578 		kfree(ua);
5579 		goto out_put;
5580 	}
5581 
5582 	/* After ua is attached to lkb it will be freed by dlm_free_lkb().
5583 	   When DLM_DFL_USER_BIT is set, the dlm knows that this is a userspace
5584 	   lock and that lkb_astparam is the dlm_user_args structure. */
5585 	set_bit(DLM_DFL_USER_BIT, &lkb->lkb_dflags);
5586 	error = request_lock(ls, lkb, name, namelen, &args);
5587 
5588 	switch (error) {
5589 	case 0:
5590 		break;
5591 	case -EINPROGRESS:
5592 		error = 0;
5593 		break;
5594 	case -EAGAIN:
5595 		error = 0;
5596 		fallthrough;
5597 	default:
5598 		goto out_put;
5599 	}
5600 
5601 	/* add this new lkb to the per-process list of locks */
5602 	spin_lock(&ua->proc->locks_spin);
5603 	hold_lkb(lkb);
5604 	list_add_tail(&lkb->lkb_ownqueue, &ua->proc->locks);
5605 	spin_unlock(&ua->proc->locks_spin);
5606 	do_put = false;
5607  out_put:
5608 	trace_dlm_lock_end(ls, lkb, name, namelen, mode, flags, error, false);
5609 	if (do_put)
5610 		__put_lkb(ls, lkb);
5611  out:
5612 	dlm_unlock_recovery(ls);
5613 	return error;
5614 }
5615 
5616 int dlm_user_convert(struct dlm_ls *ls, struct dlm_user_args *ua_tmp,
5617 		     int mode, uint32_t flags, uint32_t lkid, char *lvb_in)
5618 {
5619 	struct dlm_lkb *lkb;
5620 	struct dlm_args args;
5621 	struct dlm_user_args *ua;
5622 	int error;
5623 
5624 	dlm_lock_recovery(ls);
5625 
5626 	error = find_lkb(ls, lkid, &lkb);
5627 	if (error)
5628 		goto out;
5629 
5630 	trace_dlm_lock_start(ls, lkb, NULL, 0, mode, flags);
5631 
5632 	/* user can change the params on its lock when it converts it, or
5633 	   add an lvb that didn't exist before */
5634 
5635 	ua = lkb->lkb_ua;
5636 
5637 	if (flags & DLM_LKF_VALBLK && !ua->lksb.sb_lvbptr) {
5638 		ua->lksb.sb_lvbptr = kzalloc(DLM_USER_LVB_LEN, GFP_NOFS);
5639 		if (!ua->lksb.sb_lvbptr) {
5640 			error = -ENOMEM;
5641 			goto out_put;
5642 		}
5643 	}
5644 	if (lvb_in && ua->lksb.sb_lvbptr)
5645 		memcpy(ua->lksb.sb_lvbptr, lvb_in, DLM_USER_LVB_LEN);
5646 
5647 	ua->xid = ua_tmp->xid;
5648 	ua->castparam = ua_tmp->castparam;
5649 	ua->castaddr = ua_tmp->castaddr;
5650 	ua->bastparam = ua_tmp->bastparam;
5651 	ua->bastaddr = ua_tmp->bastaddr;
5652 	ua->user_lksb = ua_tmp->user_lksb;
5653 
5654 	error = set_lock_args(mode, &ua->lksb, flags, 0, fake_astfn, ua,
5655 			      fake_bastfn, &args);
5656 	if (error)
5657 		goto out_put;
5658 
5659 	error = convert_lock(ls, lkb, &args);
5660 
5661 	if (error == -EINPROGRESS || error == -EAGAIN || error == -EDEADLK)
5662 		error = 0;
5663  out_put:
5664 	trace_dlm_lock_end(ls, lkb, NULL, 0, mode, flags, error, false);
5665 	dlm_put_lkb(lkb);
5666  out:
5667 	dlm_unlock_recovery(ls);
5668 	kfree(ua_tmp);
5669 	return error;
5670 }
5671 
5672 /*
5673  * The caller asks for an orphan lock on a given resource with a given mode.
5674  * If a matching lock exists, it's moved to the owner's list of locks and
5675  * the lkid is returned.
5676  */
5677 
5678 int dlm_user_adopt_orphan(struct dlm_ls *ls, struct dlm_user_args *ua_tmp,
5679 		     int mode, uint32_t flags, void *name, unsigned int namelen,
5680 		     uint32_t *lkid)
5681 {
5682 	struct dlm_lkb *lkb = NULL, *iter;
5683 	struct dlm_user_args *ua;
5684 	int found_other_mode = 0;
5685 	int rv = 0;
5686 
5687 	mutex_lock(&ls->ls_orphans_mutex);
5688 	list_for_each_entry(iter, &ls->ls_orphans, lkb_ownqueue) {
5689 		if (iter->lkb_resource->res_length != namelen)
5690 			continue;
5691 		if (memcmp(iter->lkb_resource->res_name, name, namelen))
5692 			continue;
5693 		if (iter->lkb_grmode != mode) {
5694 			found_other_mode = 1;
5695 			continue;
5696 		}
5697 
5698 		lkb = iter;
5699 		list_del_init(&iter->lkb_ownqueue);
5700 		clear_bit(DLM_DFL_ORPHAN_BIT, &iter->lkb_dflags);
5701 		*lkid = iter->lkb_id;
5702 		break;
5703 	}
5704 	mutex_unlock(&ls->ls_orphans_mutex);
5705 
5706 	if (!lkb && found_other_mode) {
5707 		rv = -EAGAIN;
5708 		goto out;
5709 	}
5710 
5711 	if (!lkb) {
5712 		rv = -ENOENT;
5713 		goto out;
5714 	}
5715 
5716 	lkb->lkb_exflags = flags;
5717 	lkb->lkb_ownpid = (int) current->pid;
5718 
5719 	ua = lkb->lkb_ua;
5720 
5721 	ua->proc = ua_tmp->proc;
5722 	ua->xid = ua_tmp->xid;
5723 	ua->castparam = ua_tmp->castparam;
5724 	ua->castaddr = ua_tmp->castaddr;
5725 	ua->bastparam = ua_tmp->bastparam;
5726 	ua->bastaddr = ua_tmp->bastaddr;
5727 	ua->user_lksb = ua_tmp->user_lksb;
5728 
5729 	/*
5730 	 * The lkb reference from the ls_orphans list was not
5731 	 * removed above, and is now considered the reference
5732 	 * for the proc locks list.
5733 	 */
5734 
5735 	spin_lock(&ua->proc->locks_spin);
5736 	list_add_tail(&lkb->lkb_ownqueue, &ua->proc->locks);
5737 	spin_unlock(&ua->proc->locks_spin);
5738  out:
5739 	kfree(ua_tmp);
5740 	return rv;
5741 }
5742 
5743 int dlm_user_unlock(struct dlm_ls *ls, struct dlm_user_args *ua_tmp,
5744 		    uint32_t flags, uint32_t lkid, char *lvb_in)
5745 {
5746 	struct dlm_lkb *lkb;
5747 	struct dlm_args args;
5748 	struct dlm_user_args *ua;
5749 	int error;
5750 
5751 	dlm_lock_recovery(ls);
5752 
5753 	error = find_lkb(ls, lkid, &lkb);
5754 	if (error)
5755 		goto out;
5756 
5757 	trace_dlm_unlock_start(ls, lkb, flags);
5758 
5759 	ua = lkb->lkb_ua;
5760 
5761 	if (lvb_in && ua->lksb.sb_lvbptr)
5762 		memcpy(ua->lksb.sb_lvbptr, lvb_in, DLM_USER_LVB_LEN);
5763 	if (ua_tmp->castparam)
5764 		ua->castparam = ua_tmp->castparam;
5765 	ua->user_lksb = ua_tmp->user_lksb;
5766 
5767 	error = set_unlock_args(flags, ua, &args);
5768 	if (error)
5769 		goto out_put;
5770 
5771 	error = unlock_lock(ls, lkb, &args);
5772 
5773 	if (error == -DLM_EUNLOCK)
5774 		error = 0;
5775 	/* from validate_unlock_args() */
5776 	if (error == -EBUSY && (flags & DLM_LKF_FORCEUNLOCK))
5777 		error = 0;
5778 	if (error)
5779 		goto out_put;
5780 
5781 	spin_lock(&ua->proc->locks_spin);
5782 	/* dlm_user_add_cb() may have already taken lkb off the proc list */
5783 	if (!list_empty(&lkb->lkb_ownqueue))
5784 		list_move(&lkb->lkb_ownqueue, &ua->proc->unlocking);
5785 	spin_unlock(&ua->proc->locks_spin);
5786  out_put:
5787 	trace_dlm_unlock_end(ls, lkb, flags, error);
5788 	dlm_put_lkb(lkb);
5789  out:
5790 	dlm_unlock_recovery(ls);
5791 	kfree(ua_tmp);
5792 	return error;
5793 }
5794 
5795 int dlm_user_cancel(struct dlm_ls *ls, struct dlm_user_args *ua_tmp,
5796 		    uint32_t flags, uint32_t lkid)
5797 {
5798 	struct dlm_lkb *lkb;
5799 	struct dlm_args args;
5800 	struct dlm_user_args *ua;
5801 	int error;
5802 
5803 	dlm_lock_recovery(ls);
5804 
5805 	error = find_lkb(ls, lkid, &lkb);
5806 	if (error)
5807 		goto out;
5808 
5809 	trace_dlm_unlock_start(ls, lkb, flags);
5810 
5811 	ua = lkb->lkb_ua;
5812 	if (ua_tmp->castparam)
5813 		ua->castparam = ua_tmp->castparam;
5814 	ua->user_lksb = ua_tmp->user_lksb;
5815 
5816 	error = set_unlock_args(flags, ua, &args);
5817 	if (error)
5818 		goto out_put;
5819 
5820 	error = cancel_lock(ls, lkb, &args);
5821 
5822 	if (error == -DLM_ECANCEL)
5823 		error = 0;
5824 	/* from validate_unlock_args() */
5825 	if (error == -EBUSY)
5826 		error = 0;
5827  out_put:
5828 	trace_dlm_unlock_end(ls, lkb, flags, error);
5829 	dlm_put_lkb(lkb);
5830  out:
5831 	dlm_unlock_recovery(ls);
5832 	kfree(ua_tmp);
5833 	return error;
5834 }
5835 
5836 int dlm_user_deadlock(struct dlm_ls *ls, uint32_t flags, uint32_t lkid)
5837 {
5838 	struct dlm_lkb *lkb;
5839 	struct dlm_args args;
5840 	struct dlm_user_args *ua;
5841 	struct dlm_rsb *r;
5842 	int error;
5843 
5844 	dlm_lock_recovery(ls);
5845 
5846 	error = find_lkb(ls, lkid, &lkb);
5847 	if (error)
5848 		goto out;
5849 
5850 	trace_dlm_unlock_start(ls, lkb, flags);
5851 
5852 	ua = lkb->lkb_ua;
5853 
5854 	error = set_unlock_args(flags, ua, &args);
5855 	if (error)
5856 		goto out_put;
5857 
5858 	/* same as cancel_lock(), but set DEADLOCK_CANCEL after lock_rsb */
5859 
5860 	r = lkb->lkb_resource;
5861 	hold_rsb(r);
5862 	lock_rsb(r);
5863 
5864 	error = validate_unlock_args(lkb, &args);
5865 	if (error)
5866 		goto out_r;
5867 	set_bit(DLM_IFL_DEADLOCK_CANCEL_BIT, &lkb->lkb_iflags);
5868 
5869 	error = _cancel_lock(r, lkb);
5870  out_r:
5871 	unlock_rsb(r);
5872 	put_rsb(r);
5873 
5874 	if (error == -DLM_ECANCEL)
5875 		error = 0;
5876 	/* from validate_unlock_args() */
5877 	if (error == -EBUSY)
5878 		error = 0;
5879  out_put:
5880 	trace_dlm_unlock_end(ls, lkb, flags, error);
5881 	dlm_put_lkb(lkb);
5882  out:
5883 	dlm_unlock_recovery(ls);
5884 	return error;
5885 }
5886 
5887 /* lkb's that are removed from the waiters list by revert are just left on the
5888    orphans list with the granted orphan locks, to be freed by purge */
5889 
5890 static int orphan_proc_lock(struct dlm_ls *ls, struct dlm_lkb *lkb)
5891 {
5892 	struct dlm_args args;
5893 	int error;
5894 
5895 	hold_lkb(lkb); /* reference for the ls_orphans list */
5896 	mutex_lock(&ls->ls_orphans_mutex);
5897 	list_add_tail(&lkb->lkb_ownqueue, &ls->ls_orphans);
5898 	mutex_unlock(&ls->ls_orphans_mutex);
5899 
5900 	set_unlock_args(0, lkb->lkb_ua, &args);
5901 
5902 	error = cancel_lock(ls, lkb, &args);
5903 	if (error == -DLM_ECANCEL)
5904 		error = 0;
5905 	return error;
5906 }
5907 
5908 /* The FORCEUNLOCK flag allows the unlock to go ahead even if the lkb isn't
5909    granted.  Regardless of what rsb queue the lock is on, it's removed and
5910    freed.  The IVVALBLK flag causes the lvb on the resource to be invalidated
5911    if our lock is PW/EX (it's ignored if our granted mode is smaller.) */
5912 
5913 static int unlock_proc_lock(struct dlm_ls *ls, struct dlm_lkb *lkb)
5914 {
5915 	struct dlm_args args;
5916 	int error;
5917 
5918 	set_unlock_args(DLM_LKF_FORCEUNLOCK | DLM_LKF_IVVALBLK,
5919 			lkb->lkb_ua, &args);
5920 
5921 	error = unlock_lock(ls, lkb, &args);
5922 	if (error == -DLM_EUNLOCK)
5923 		error = 0;
5924 	return error;
5925 }
5926 
5927 /* We have to release clear_proc_locks mutex before calling unlock_proc_lock()
5928    (which does lock_rsb) due to deadlock with receiving a message that does
5929    lock_rsb followed by dlm_user_add_cb() */
5930 
5931 static struct dlm_lkb *del_proc_lock(struct dlm_ls *ls,
5932 				     struct dlm_user_proc *proc)
5933 {
5934 	struct dlm_lkb *lkb = NULL;
5935 
5936 	spin_lock(&ls->ls_clear_proc_locks);
5937 	if (list_empty(&proc->locks))
5938 		goto out;
5939 
5940 	lkb = list_entry(proc->locks.next, struct dlm_lkb, lkb_ownqueue);
5941 	list_del_init(&lkb->lkb_ownqueue);
5942 
5943 	if (lkb->lkb_exflags & DLM_LKF_PERSISTENT)
5944 		set_bit(DLM_DFL_ORPHAN_BIT, &lkb->lkb_dflags);
5945 	else
5946 		set_bit(DLM_IFL_DEAD_BIT, &lkb->lkb_iflags);
5947  out:
5948 	spin_unlock(&ls->ls_clear_proc_locks);
5949 	return lkb;
5950 }
5951 
5952 /* The ls_clear_proc_locks mutex protects against dlm_user_add_cb() which
5953    1) references lkb->ua which we free here and 2) adds lkbs to proc->asts,
5954    which we clear here. */
5955 
5956 /* proc CLOSING flag is set so no more device_reads should look at proc->asts
5957    list, and no more device_writes should add lkb's to proc->locks list; so we
5958    shouldn't need to take asts_spin or locks_spin here.  this assumes that
5959    device reads/writes/closes are serialized -- FIXME: we may need to serialize
5960    them ourself. */
5961 
5962 void dlm_clear_proc_locks(struct dlm_ls *ls, struct dlm_user_proc *proc)
5963 {
5964 	struct dlm_lkb *lkb, *safe;
5965 
5966 	dlm_lock_recovery(ls);
5967 
5968 	while (1) {
5969 		lkb = del_proc_lock(ls, proc);
5970 		if (!lkb)
5971 			break;
5972 		if (lkb->lkb_exflags & DLM_LKF_PERSISTENT)
5973 			orphan_proc_lock(ls, lkb);
5974 		else
5975 			unlock_proc_lock(ls, lkb);
5976 
5977 		/* this removes the reference for the proc->locks list
5978 		   added by dlm_user_request, it may result in the lkb
5979 		   being freed */
5980 
5981 		dlm_put_lkb(lkb);
5982 	}
5983 
5984 	spin_lock(&ls->ls_clear_proc_locks);
5985 
5986 	/* in-progress unlocks */
5987 	list_for_each_entry_safe(lkb, safe, &proc->unlocking, lkb_ownqueue) {
5988 		list_del_init(&lkb->lkb_ownqueue);
5989 		set_bit(DLM_IFL_DEAD_BIT, &lkb->lkb_iflags);
5990 		dlm_put_lkb(lkb);
5991 	}
5992 
5993 	list_for_each_entry_safe(lkb, safe, &proc->asts, lkb_cb_list) {
5994 		dlm_purge_lkb_callbacks(lkb);
5995 		list_del_init(&lkb->lkb_cb_list);
5996 		dlm_put_lkb(lkb);
5997 	}
5998 
5999 	spin_unlock(&ls->ls_clear_proc_locks);
6000 	dlm_unlock_recovery(ls);
6001 }
6002 
6003 static void purge_proc_locks(struct dlm_ls *ls, struct dlm_user_proc *proc)
6004 {
6005 	struct dlm_lkb *lkb, *safe;
6006 
6007 	while (1) {
6008 		lkb = NULL;
6009 		spin_lock(&proc->locks_spin);
6010 		if (!list_empty(&proc->locks)) {
6011 			lkb = list_entry(proc->locks.next, struct dlm_lkb,
6012 					 lkb_ownqueue);
6013 			list_del_init(&lkb->lkb_ownqueue);
6014 		}
6015 		spin_unlock(&proc->locks_spin);
6016 
6017 		if (!lkb)
6018 			break;
6019 
6020 		set_bit(DLM_IFL_DEAD_BIT, &lkb->lkb_iflags);
6021 		unlock_proc_lock(ls, lkb);
6022 		dlm_put_lkb(lkb); /* ref from proc->locks list */
6023 	}
6024 
6025 	spin_lock(&proc->locks_spin);
6026 	list_for_each_entry_safe(lkb, safe, &proc->unlocking, lkb_ownqueue) {
6027 		list_del_init(&lkb->lkb_ownqueue);
6028 		set_bit(DLM_IFL_DEAD_BIT, &lkb->lkb_iflags);
6029 		dlm_put_lkb(lkb);
6030 	}
6031 	spin_unlock(&proc->locks_spin);
6032 
6033 	spin_lock(&proc->asts_spin);
6034 	list_for_each_entry_safe(lkb, safe, &proc->asts, lkb_cb_list) {
6035 		dlm_purge_lkb_callbacks(lkb);
6036 		list_del_init(&lkb->lkb_cb_list);
6037 		dlm_put_lkb(lkb);
6038 	}
6039 	spin_unlock(&proc->asts_spin);
6040 }
6041 
6042 /* pid of 0 means purge all orphans */
6043 
6044 static void do_purge(struct dlm_ls *ls, int nodeid, int pid)
6045 {
6046 	struct dlm_lkb *lkb, *safe;
6047 
6048 	mutex_lock(&ls->ls_orphans_mutex);
6049 	list_for_each_entry_safe(lkb, safe, &ls->ls_orphans, lkb_ownqueue) {
6050 		if (pid && lkb->lkb_ownpid != pid)
6051 			continue;
6052 		unlock_proc_lock(ls, lkb);
6053 		list_del_init(&lkb->lkb_ownqueue);
6054 		dlm_put_lkb(lkb);
6055 	}
6056 	mutex_unlock(&ls->ls_orphans_mutex);
6057 }
6058 
6059 static int send_purge(struct dlm_ls *ls, int nodeid, int pid)
6060 {
6061 	struct dlm_message *ms;
6062 	struct dlm_mhandle *mh;
6063 	int error;
6064 
6065 	error = _create_message(ls, sizeof(struct dlm_message), nodeid,
6066 				DLM_MSG_PURGE, &ms, &mh, GFP_NOFS);
6067 	if (error)
6068 		return error;
6069 	ms->m_nodeid = cpu_to_le32(nodeid);
6070 	ms->m_pid = cpu_to_le32(pid);
6071 
6072 	return send_message(mh, ms, NULL, 0);
6073 }
6074 
6075 int dlm_user_purge(struct dlm_ls *ls, struct dlm_user_proc *proc,
6076 		   int nodeid, int pid)
6077 {
6078 	int error = 0;
6079 
6080 	if (nodeid && (nodeid != dlm_our_nodeid())) {
6081 		error = send_purge(ls, nodeid, pid);
6082 	} else {
6083 		dlm_lock_recovery(ls);
6084 		if (pid == current->pid)
6085 			purge_proc_locks(ls, proc);
6086 		else
6087 			do_purge(ls, nodeid, pid);
6088 		dlm_unlock_recovery(ls);
6089 	}
6090 	return error;
6091 }
6092 
6093 /* debug functionality */
6094 int dlm_debug_add_lkb(struct dlm_ls *ls, uint32_t lkb_id, char *name, int len,
6095 		      int lkb_nodeid, unsigned int lkb_dflags, int lkb_status)
6096 {
6097 	struct dlm_lksb *lksb;
6098 	struct dlm_lkb *lkb;
6099 	struct dlm_rsb *r;
6100 	int error;
6101 
6102 	/* we currently can't set a valid user lock */
6103 	if (lkb_dflags & BIT(DLM_DFL_USER_BIT))
6104 		return -EOPNOTSUPP;
6105 
6106 	lksb = kzalloc(sizeof(*lksb), GFP_NOFS);
6107 	if (!lksb)
6108 		return -ENOMEM;
6109 
6110 	error = _create_lkb(ls, &lkb, lkb_id, lkb_id + 1);
6111 	if (error) {
6112 		kfree(lksb);
6113 		return error;
6114 	}
6115 
6116 	dlm_set_dflags_val(lkb, lkb_dflags);
6117 	lkb->lkb_nodeid = lkb_nodeid;
6118 	lkb->lkb_lksb = lksb;
6119 	/* user specific pointer, just don't have it NULL for kernel locks */
6120 	if (~lkb_dflags & BIT(DLM_DFL_USER_BIT))
6121 		lkb->lkb_astparam = (void *)0xDEADBEEF;
6122 
6123 	error = find_rsb(ls, name, len, 0, R_REQUEST, &r);
6124 	if (error) {
6125 		kfree(lksb);
6126 		__put_lkb(ls, lkb);
6127 		return error;
6128 	}
6129 
6130 	lock_rsb(r);
6131 	attach_lkb(r, lkb);
6132 	add_lkb(r, lkb, lkb_status);
6133 	unlock_rsb(r);
6134 	put_rsb(r);
6135 
6136 	return 0;
6137 }
6138 
6139 int dlm_debug_add_lkb_to_waiters(struct dlm_ls *ls, uint32_t lkb_id,
6140 				 int mstype, int to_nodeid)
6141 {
6142 	struct dlm_lkb *lkb;
6143 	int error;
6144 
6145 	error = find_lkb(ls, lkb_id, &lkb);
6146 	if (error)
6147 		return error;
6148 
6149 	error = add_to_waiters(lkb, mstype, to_nodeid);
6150 	dlm_put_lkb(lkb);
6151 	return error;
6152 }
6153 
6154