forked from torvalds/linux
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathvote.c
1202 lines (994 loc) · 31.2 KB
/
vote.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* vote.c
*
* description here
*
* Copyright (C) 2003, 2004 Oracle. All rights reserved.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*/
#include <linux/types.h>
#include <linux/slab.h>
#include <linux/highmem.h>
#include <linux/smp_lock.h>
#include <linux/kthread.h>
#include <cluster/heartbeat.h>
#include <cluster/nodemanager.h>
#include <cluster/tcp.h>
#include <dlm/dlmapi.h>
#define MLOG_MASK_PREFIX ML_VOTE
#include <cluster/masklog.h>
#include "ocfs2.h"
#include "alloc.h"
#include "dlmglue.h"
#include "extent_map.h"
#include "heartbeat.h"
#include "inode.h"
#include "journal.h"
#include "slot_map.h"
#include "vote.h"
#include "buffer_head_io.h"
#define OCFS2_MESSAGE_TYPE_VOTE (0x1)
#define OCFS2_MESSAGE_TYPE_RESPONSE (0x2)
struct ocfs2_msg_hdr
{
__be32 h_response_id; /* used to lookup message handle on sending
* node. */
__be32 h_request;
__be64 h_blkno;
__be32 h_generation;
__be32 h_node_num; /* node sending this particular message. */
};
/* OCFS2_MAX_FILENAME_LEN is 255 characters, but we want to align this
* for the network. */
#define OCFS2_VOTE_FILENAME_LEN 256
struct ocfs2_vote_msg
{
struct ocfs2_msg_hdr v_hdr;
union {
__be32 v_generic1;
__be32 v_orphaned_slot; /* Used during delete votes */
__be32 v_nlink; /* Used during unlink votes */
} md1; /* Message type dependant 1 */
__be32 v_unlink_namelen;
__be64 v_unlink_parent;
u8 v_unlink_dirent[OCFS2_VOTE_FILENAME_LEN];
};
/* Responses are given these values to maintain backwards
* compatibility with older ocfs2 versions */
#define OCFS2_RESPONSE_OK (0)
#define OCFS2_RESPONSE_BUSY (-16)
#define OCFS2_RESPONSE_BAD_MSG (-22)
struct ocfs2_response_msg
{
struct ocfs2_msg_hdr r_hdr;
__be32 r_response;
__be32 r_orphaned_slot;
};
struct ocfs2_vote_work {
struct list_head w_list;
struct ocfs2_vote_msg w_msg;
};
enum ocfs2_vote_request {
OCFS2_VOTE_REQ_INVALID = 0,
OCFS2_VOTE_REQ_DELETE,
OCFS2_VOTE_REQ_UNLINK,
OCFS2_VOTE_REQ_RENAME,
OCFS2_VOTE_REQ_MOUNT,
OCFS2_VOTE_REQ_UMOUNT,
OCFS2_VOTE_REQ_LAST
};
static inline int ocfs2_is_valid_vote_request(int request)
{
return OCFS2_VOTE_REQ_INVALID < request &&
request < OCFS2_VOTE_REQ_LAST;
}
typedef void (*ocfs2_net_response_callback)(void *priv,
struct ocfs2_response_msg *resp);
struct ocfs2_net_response_cb {
ocfs2_net_response_callback rc_cb;
void *rc_priv;
};
struct ocfs2_net_wait_ctxt {
struct list_head n_list;
u32 n_response_id;
wait_queue_head_t n_event;
struct ocfs2_node_map n_node_map;
int n_response; /* an agreggate response. 0 if
* all nodes are go, < 0 on any
* negative response from any
* node or network error. */
struct ocfs2_net_response_cb *n_callback;
};
static void ocfs2_process_mount_request(struct ocfs2_super *osb,
unsigned int node_num)
{
mlog(0, "MOUNT vote from node %u\n", node_num);
/* The other node only sends us this message when he has an EX
* on the superblock, so our recovery threads (if having been
* launched) are waiting on it.*/
ocfs2_recovery_map_clear(osb, node_num);
ocfs2_node_map_set_bit(osb, &osb->mounted_map, node_num);
/* We clear the umount map here because a node may have been
* previously mounted, safely unmounted but never stopped
* heartbeating - in which case we'd have a stale entry. */
ocfs2_node_map_clear_bit(osb, &osb->umount_map, node_num);
}
static void ocfs2_process_umount_request(struct ocfs2_super *osb,
unsigned int node_num)
{
mlog(0, "UMOUNT vote from node %u\n", node_num);
ocfs2_node_map_clear_bit(osb, &osb->mounted_map, node_num);
ocfs2_node_map_set_bit(osb, &osb->umount_map, node_num);
}
void ocfs2_mark_inode_remotely_deleted(struct inode *inode)
{
struct ocfs2_inode_info *oi = OCFS2_I(inode);
assert_spin_locked(&oi->ip_lock);
/* We set the SKIP_DELETE flag on the inode so we don't try to
* delete it in delete_inode ourselves, thus avoiding
* unecessary lock pinging. If the other node failed to wipe
* the inode as a result of a crash, then recovery will pick
* up the slack. */
oi->ip_flags |= OCFS2_INODE_DELETED|OCFS2_INODE_SKIP_DELETE;
}
static int ocfs2_process_delete_request(struct inode *inode,
int *orphaned_slot)
{
int response = OCFS2_RESPONSE_BUSY;
mlog(0, "DELETE vote on inode %lu, read lnk_cnt = %u, slot = %d\n",
inode->i_ino, inode->i_nlink, *orphaned_slot);
spin_lock(&OCFS2_I(inode)->ip_lock);
/* Whatever our vote response is, we want to make sure that
* the orphaned slot is recorded properly on this node *and*
* on the requesting node. Technically, if the requesting node
* did not know which slot the inode is orphaned in but we
* respond with BUSY he doesn't actually need the orphaned
* slot, but it doesn't hurt to do it here anyway. */
if ((*orphaned_slot) != OCFS2_INVALID_SLOT) {
mlog_bug_on_msg(OCFS2_I(inode)->ip_orphaned_slot !=
OCFS2_INVALID_SLOT &&
OCFS2_I(inode)->ip_orphaned_slot !=
(*orphaned_slot),
"Inode %"MLFu64": This node thinks it's "
"orphaned in slot %d, messaged it's in %d\n",
OCFS2_I(inode)->ip_blkno,
OCFS2_I(inode)->ip_orphaned_slot,
*orphaned_slot);
mlog(0, "Setting orphaned slot for inode %"MLFu64" to %d\n",
OCFS2_I(inode)->ip_blkno, *orphaned_slot);
OCFS2_I(inode)->ip_orphaned_slot = *orphaned_slot;
} else {
mlog(0, "Sending back orphaned slot %d for inode %"MLFu64"\n",
OCFS2_I(inode)->ip_orphaned_slot,
OCFS2_I(inode)->ip_blkno);
*orphaned_slot = OCFS2_I(inode)->ip_orphaned_slot;
}
/* vote no if the file is still open. */
if (OCFS2_I(inode)->ip_open_count) {
mlog(0, "open count = %u\n",
OCFS2_I(inode)->ip_open_count);
spin_unlock(&OCFS2_I(inode)->ip_lock);
goto done;
}
spin_unlock(&OCFS2_I(inode)->ip_lock);
/* directories are a bit ugly... What if someone is sitting in
* it? We want to make sure the inode is removed completely as
* a result of the iput in process_vote. */
if (S_ISDIR(inode->i_mode) && (atomic_read(&inode->i_count) != 1)) {
mlog(0, "i_count = %u\n", atomic_read(&inode->i_count));
goto done;
}
if (filemap_fdatawrite(inode->i_mapping)) {
mlog(ML_ERROR, "Could not sync inode %"MLFu64" for delete!\n",
OCFS2_I(inode)->ip_blkno);
goto done;
}
sync_mapping_buffers(inode->i_mapping);
truncate_inode_pages(inode->i_mapping, 0);
ocfs2_extent_map_trunc(inode, 0);
spin_lock(&OCFS2_I(inode)->ip_lock);
/* double check open count - someone might have raced this
* thread into ocfs2_file_open while we were writing out
* data. If we're to allow a wipe of this inode now, we *must*
* hold the spinlock until we've marked it. */
if (OCFS2_I(inode)->ip_open_count) {
mlog(0, "Raced to wipe! open count = %u\n",
OCFS2_I(inode)->ip_open_count);
spin_unlock(&OCFS2_I(inode)->ip_lock);
goto done;
}
/* Mark the inode as being wiped from disk. */
ocfs2_mark_inode_remotely_deleted(inode);
spin_unlock(&OCFS2_I(inode)->ip_lock);
/* Not sure this is necessary anymore. */
d_prune_aliases(inode);
/* If we get here, then we're voting 'yes', so commit the
* delete on our side. */
response = OCFS2_RESPONSE_OK;
done:
return response;
}
static int ocfs2_match_dentry(struct dentry *dentry,
u64 parent_blkno,
unsigned int namelen,
const char *name)
{
struct inode *parent;
if (!dentry->d_parent) {
mlog(0, "Detached from parent.\n");
return 0;
}
parent = dentry->d_parent->d_inode;
/* Negative parent dentry? */
if (!parent)
return 0;
/* Name is in a different directory. */
if (OCFS2_I(parent)->ip_blkno != parent_blkno)
return 0;
if (dentry->d_name.len != namelen)
return 0;
/* comparison above guarantees this is safe. */
if (memcmp(dentry->d_name.name, name, namelen))
return 0;
return 1;
}
static void ocfs2_process_dentry_request(struct inode *inode,
int rename,
unsigned int new_nlink,
u64 parent_blkno,
unsigned int namelen,
const char *name)
{
struct dentry *dentry = NULL;
struct list_head *p;
struct ocfs2_inode_info *oi = OCFS2_I(inode);
mlog(0, "parent %"MLFu64", namelen = %u, name = %.*s\n", parent_blkno,
namelen, namelen, name);
spin_lock(&dcache_lock);
/* Another node is removing this name from the system. It is
* up to us to find the corresponding dentry and if it exists,
* unhash it from the dcache. */
list_for_each(p, &inode->i_dentry) {
dentry = list_entry(p, struct dentry, d_alias);
if (ocfs2_match_dentry(dentry, parent_blkno, namelen, name)) {
mlog(0, "dentry found: %.*s\n",
dentry->d_name.len, dentry->d_name.name);
dget_locked(dentry);
break;
}
dentry = NULL;
}
spin_unlock(&dcache_lock);
if (dentry) {
d_delete(dentry);
dput(dentry);
}
/* rename votes don't send link counts */
if (!rename) {
mlog(0, "new_nlink = %u\n", new_nlink);
/* We don't have the proper locks here to directly
* change i_nlink and besides, the vote is sent
* *before* the operation so it may have failed on the
* other node. This passes a hint to ocfs2_drop_inode
* to force ocfs2_delete_inode, who will take the
* proper cluster locks to sort things out. */
if (new_nlink == 0) {
spin_lock(&oi->ip_lock);
oi->ip_flags |= OCFS2_INODE_MAYBE_ORPHANED;
spin_unlock(&OCFS2_I(inode)->ip_lock);
}
}
}
static void ocfs2_process_vote(struct ocfs2_super *osb,
struct ocfs2_vote_msg *msg)
{
int net_status, vote_response;
int orphaned_slot = 0;
int rename = 0;
unsigned int node_num, generation, new_nlink, namelen;
u64 blkno, parent_blkno;
enum ocfs2_vote_request request;
struct inode *inode = NULL;
struct ocfs2_msg_hdr *hdr = &msg->v_hdr;
struct ocfs2_response_msg response;
/* decode the network mumbo jumbo into local variables. */
request = be32_to_cpu(hdr->h_request);
blkno = be64_to_cpu(hdr->h_blkno);
generation = be32_to_cpu(hdr->h_generation);
node_num = be32_to_cpu(hdr->h_node_num);
if (request == OCFS2_VOTE_REQ_DELETE)
orphaned_slot = be32_to_cpu(msg->md1.v_orphaned_slot);
mlog(0, "processing vote: request = %u, blkno = %"MLFu64", "
"generation = %u, node_num = %u, priv1 = %u\n", request,
blkno, generation, node_num, be32_to_cpu(msg->md1.v_generic1));
if (!ocfs2_is_valid_vote_request(request)) {
mlog(ML_ERROR, "Invalid vote request %d from node %u\n",
request, node_num);
vote_response = OCFS2_RESPONSE_BAD_MSG;
goto respond;
}
vote_response = OCFS2_RESPONSE_OK;
switch (request) {
case OCFS2_VOTE_REQ_UMOUNT:
ocfs2_process_umount_request(osb, node_num);
goto respond;
case OCFS2_VOTE_REQ_MOUNT:
ocfs2_process_mount_request(osb, node_num);
goto respond;
default:
/* avoids a gcc warning */
break;
}
/* We cannot process the remaining message types before we're
* fully mounted. It's perfectly safe however to send a 'yes'
* response as we can't possibly have any of the state they're
* asking us to modify yet. */
if (atomic_read(&osb->vol_state) == VOLUME_INIT)
goto respond;
/* If we get here, then the request is against an inode. */
inode = ocfs2_ilookup_for_vote(osb, blkno,
request == OCFS2_VOTE_REQ_DELETE);
/* Not finding the inode is perfectly valid - it means we're
* not interested in what the other node is about to do to it
* so in those cases we automatically respond with an
* affirmative. Cluster locking ensures that we won't race
* interest in the inode with this vote request. */
if (!inode)
goto respond;
/* Check generation values. It's possible for us to get a
* request against a stale inode. If so then we proceed as if
* we had not found an inode in the first place. */
if (inode->i_generation != generation) {
mlog(0, "generation passed %u != inode generation = %u, "
"ip_flags = %x, ip_blkno = %"MLFu64", msg %"MLFu64", "
"i_count = %u, message type = %u\n",
generation, inode->i_generation, OCFS2_I(inode)->ip_flags,
OCFS2_I(inode)->ip_blkno, blkno,
atomic_read(&inode->i_count), request);
iput(inode);
inode = NULL;
goto respond;
}
switch (request) {
case OCFS2_VOTE_REQ_DELETE:
vote_response = ocfs2_process_delete_request(inode,
&orphaned_slot);
break;
case OCFS2_VOTE_REQ_RENAME:
rename = 1;
/* fall through */
case OCFS2_VOTE_REQ_UNLINK:
parent_blkno = be64_to_cpu(msg->v_unlink_parent);
namelen = be32_to_cpu(msg->v_unlink_namelen);
/* new_nlink will be ignored in case of a rename vote */
new_nlink = be32_to_cpu(msg->md1.v_nlink);
ocfs2_process_dentry_request(inode, rename, new_nlink,
parent_blkno, namelen,
msg->v_unlink_dirent);
break;
default:
mlog(ML_ERROR, "node %u, invalid request: %u\n",
node_num, request);
vote_response = OCFS2_RESPONSE_BAD_MSG;
}
respond:
/* Response struture is small so we just put it on the stack
* and stuff it inline. */
memset(&response, 0, sizeof(struct ocfs2_response_msg));
response.r_hdr.h_response_id = hdr->h_response_id;
response.r_hdr.h_blkno = hdr->h_blkno;
response.r_hdr.h_generation = hdr->h_generation;
response.r_hdr.h_node_num = cpu_to_be32(osb->node_num);
response.r_response = cpu_to_be32(vote_response);
response.r_orphaned_slot = cpu_to_be32(orphaned_slot);
net_status = o2net_send_message(OCFS2_MESSAGE_TYPE_RESPONSE,
osb->net_key,
&response,
sizeof(struct ocfs2_response_msg),
node_num,
NULL);
/* We still want to error print for ENOPROTOOPT here. The
* sending node shouldn't have unregistered his net handler
* without sending an unmount vote 1st */
if (net_status < 0
&& net_status != -ETIMEDOUT
&& net_status != -ENOTCONN)
mlog(ML_ERROR, "message to node %u fails with error %d!\n",
node_num, net_status);
if (inode)
iput(inode);
}
static void ocfs2_vote_thread_do_work(struct ocfs2_super *osb)
{
unsigned long processed;
struct ocfs2_lock_res *lockres;
struct ocfs2_vote_work *work;
mlog_entry_void();
spin_lock(&osb->vote_task_lock);
/* grab this early so we know to try again if a state change and
* wake happens part-way through our work */
osb->vote_work_sequence = osb->vote_wake_sequence;
processed = osb->blocked_lock_count;
while (processed) {
BUG_ON(list_empty(&osb->blocked_lock_list));
lockres = list_entry(osb->blocked_lock_list.next,
struct ocfs2_lock_res, l_blocked_list);
list_del_init(&lockres->l_blocked_list);
osb->blocked_lock_count--;
spin_unlock(&osb->vote_task_lock);
BUG_ON(!processed);
processed--;
ocfs2_process_blocked_lock(osb, lockres);
spin_lock(&osb->vote_task_lock);
}
while (osb->vote_count) {
BUG_ON(list_empty(&osb->vote_list));
work = list_entry(osb->vote_list.next,
struct ocfs2_vote_work, w_list);
list_del(&work->w_list);
osb->vote_count--;
spin_unlock(&osb->vote_task_lock);
ocfs2_process_vote(osb, &work->w_msg);
kfree(work);
spin_lock(&osb->vote_task_lock);
}
spin_unlock(&osb->vote_task_lock);
mlog_exit_void();
}
static int ocfs2_vote_thread_lists_empty(struct ocfs2_super *osb)
{
int empty = 0;
spin_lock(&osb->vote_task_lock);
if (list_empty(&osb->blocked_lock_list) &&
list_empty(&osb->vote_list))
empty = 1;
spin_unlock(&osb->vote_task_lock);
return empty;
}
static int ocfs2_vote_thread_should_wake(struct ocfs2_super *osb)
{
int should_wake = 0;
spin_lock(&osb->vote_task_lock);
if (osb->vote_work_sequence != osb->vote_wake_sequence)
should_wake = 1;
spin_unlock(&osb->vote_task_lock);
return should_wake;
}
int ocfs2_vote_thread(void *arg)
{
int status = 0;
struct ocfs2_super *osb = arg;
/* only quit once we've been asked to stop and there is no more
* work available */
while (!(kthread_should_stop() &&
ocfs2_vote_thread_lists_empty(osb))) {
wait_event_interruptible(osb->vote_event,
ocfs2_vote_thread_should_wake(osb) ||
kthread_should_stop());
mlog(0, "vote_thread: awoken\n");
ocfs2_vote_thread_do_work(osb);
}
osb->vote_task = NULL;
return status;
}
static struct ocfs2_net_wait_ctxt *ocfs2_new_net_wait_ctxt(unsigned int response_id)
{
struct ocfs2_net_wait_ctxt *w;
w = kcalloc(1, sizeof(*w), GFP_KERNEL);
if (!w) {
mlog_errno(-ENOMEM);
goto bail;
}
INIT_LIST_HEAD(&w->n_list);
init_waitqueue_head(&w->n_event);
ocfs2_node_map_init(&w->n_node_map);
w->n_response_id = response_id;
w->n_callback = NULL;
bail:
return w;
}
static unsigned int ocfs2_new_response_id(struct ocfs2_super *osb)
{
unsigned int ret;
spin_lock(&osb->net_response_lock);
ret = ++osb->net_response_ids;
spin_unlock(&osb->net_response_lock);
return ret;
}
static void ocfs2_dequeue_net_wait_ctxt(struct ocfs2_super *osb,
struct ocfs2_net_wait_ctxt *w)
{
spin_lock(&osb->net_response_lock);
list_del(&w->n_list);
spin_unlock(&osb->net_response_lock);
}
static void ocfs2_queue_net_wait_ctxt(struct ocfs2_super *osb,
struct ocfs2_net_wait_ctxt *w)
{
spin_lock(&osb->net_response_lock);
list_add_tail(&w->n_list,
&osb->net_response_list);
spin_unlock(&osb->net_response_lock);
}
static void __ocfs2_mark_node_responded(struct ocfs2_super *osb,
struct ocfs2_net_wait_ctxt *w,
int node_num)
{
assert_spin_locked(&osb->net_response_lock);
ocfs2_node_map_clear_bit(osb, &w->n_node_map, node_num);
if (ocfs2_node_map_is_empty(osb, &w->n_node_map))
wake_up(&w->n_event);
}
/* Intended to be called from the node down callback, we fake remove
* the node from all our response contexts */
void ocfs2_remove_node_from_vote_queues(struct ocfs2_super *osb,
int node_num)
{
struct list_head *p;
struct ocfs2_net_wait_ctxt *w = NULL;
spin_lock(&osb->net_response_lock);
list_for_each(p, &osb->net_response_list) {
w = list_entry(p, struct ocfs2_net_wait_ctxt, n_list);
__ocfs2_mark_node_responded(osb, w, node_num);
}
spin_unlock(&osb->net_response_lock);
}
static int ocfs2_broadcast_vote(struct ocfs2_super *osb,
struct ocfs2_vote_msg *request,
unsigned int response_id,
int *response,
struct ocfs2_net_response_cb *callback)
{
int status, i, remote_err;
struct ocfs2_net_wait_ctxt *w = NULL;
int dequeued = 0;
mlog_entry_void();
w = ocfs2_new_net_wait_ctxt(response_id);
if (!w) {
status = -ENOMEM;
mlog_errno(status);
goto bail;
}
w->n_callback = callback;
/* we're pretty much ready to go at this point, and this fills
* in n_response which we need anyway... */
ocfs2_queue_net_wait_ctxt(osb, w);
i = ocfs2_node_map_iterate(osb, &osb->mounted_map, 0);
while (i != O2NM_INVALID_NODE_NUM) {
if (i != osb->node_num) {
mlog(0, "trying to send request to node %i\n", i);
ocfs2_node_map_set_bit(osb, &w->n_node_map, i);
remote_err = 0;
status = o2net_send_message(OCFS2_MESSAGE_TYPE_VOTE,
osb->net_key,
request,
sizeof(*request),
i,
&remote_err);
if (status == -ETIMEDOUT) {
mlog(0, "remote node %d timed out!\n", i);
status = -EAGAIN;
goto bail;
}
if (remote_err < 0) {
status = remote_err;
mlog(0, "remote error %d on node %d!\n",
remote_err, i);
mlog_errno(status);
goto bail;
}
if (status < 0) {
mlog_errno(status);
goto bail;
}
}
i++;
i = ocfs2_node_map_iterate(osb, &osb->mounted_map, i);
mlog(0, "next is %d, i am %d\n", i, osb->node_num);
}
mlog(0, "done sending, now waiting on responses...\n");
wait_event(w->n_event, ocfs2_node_map_is_empty(osb, &w->n_node_map));
ocfs2_dequeue_net_wait_ctxt(osb, w);
dequeued = 1;
*response = w->n_response;
status = 0;
bail:
if (w) {
if (!dequeued)
ocfs2_dequeue_net_wait_ctxt(osb, w);
kfree(w);
}
mlog_exit(status);
return status;
}
static struct ocfs2_vote_msg * ocfs2_new_vote_request(struct ocfs2_super *osb,
u64 blkno,
unsigned int generation,
enum ocfs2_vote_request type,
u32 priv)
{
struct ocfs2_vote_msg *request;
struct ocfs2_msg_hdr *hdr;
BUG_ON(!ocfs2_is_valid_vote_request(type));
request = kcalloc(1, sizeof(*request), GFP_KERNEL);
if (!request) {
mlog_errno(-ENOMEM);
} else {
hdr = &request->v_hdr;
hdr->h_node_num = cpu_to_be32(osb->node_num);
hdr->h_request = cpu_to_be32(type);
hdr->h_blkno = cpu_to_be64(blkno);
hdr->h_generation = cpu_to_be32(generation);
request->md1.v_generic1 = cpu_to_be32(priv);
}
return request;
}
/* Complete the buildup of a new vote request and process the
* broadcast return value. */
static int ocfs2_do_request_vote(struct ocfs2_super *osb,
struct ocfs2_vote_msg *request,
struct ocfs2_net_response_cb *callback)
{
int status, response;
unsigned int response_id;
struct ocfs2_msg_hdr *hdr;
response_id = ocfs2_new_response_id(osb);
hdr = &request->v_hdr;
hdr->h_response_id = cpu_to_be32(response_id);
status = ocfs2_broadcast_vote(osb, request, response_id, &response,
callback);
if (status < 0) {
mlog_errno(status);
goto bail;
}
status = response;
bail:
return status;
}
static int ocfs2_request_vote(struct inode *inode,
struct ocfs2_vote_msg *request,
struct ocfs2_net_response_cb *callback)
{
int status;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
if (ocfs2_inode_is_new(inode))
return 0;
status = -EAGAIN;
while (status == -EAGAIN) {
if (!(osb->s_mount_opt & OCFS2_MOUNT_NOINTR) &&
signal_pending(current))
return -ERESTARTSYS;
status = ocfs2_super_lock(osb, 0);
if (status < 0) {
mlog_errno(status);
break;
}
status = 0;
if (!ocfs2_node_map_is_only(osb, &osb->mounted_map,
osb->node_num))
status = ocfs2_do_request_vote(osb, request, callback);
ocfs2_super_unlock(osb, 0);
}
return status;
}
static void ocfs2_delete_response_cb(void *priv,
struct ocfs2_response_msg *resp)
{
int orphaned_slot, node;
struct inode *inode = priv;
orphaned_slot = be32_to_cpu(resp->r_orphaned_slot);
node = be32_to_cpu(resp->r_hdr.h_node_num);
mlog(0, "node %d tells us that inode %"MLFu64" is orphaned in slot "
"%d\n", node, OCFS2_I(inode)->ip_blkno, orphaned_slot);
/* The other node may not actually know which slot the inode
* is orphaned in. */
if (orphaned_slot == OCFS2_INVALID_SLOT)
return;
/* Ok, the responding node knows which slot this inode is
* orphaned in. We verify that the information is correct and
* then record this in the inode. ocfs2_delete_inode will use
* this information to determine which lock to take. */
spin_lock(&OCFS2_I(inode)->ip_lock);
mlog_bug_on_msg(OCFS2_I(inode)->ip_orphaned_slot != orphaned_slot &&
OCFS2_I(inode)->ip_orphaned_slot
!= OCFS2_INVALID_SLOT, "Inode %"MLFu64": Node %d "
"says it's orphaned in slot %d, we think it's in %d\n",
OCFS2_I(inode)->ip_blkno,
be32_to_cpu(resp->r_hdr.h_node_num),
orphaned_slot, OCFS2_I(inode)->ip_orphaned_slot);
OCFS2_I(inode)->ip_orphaned_slot = orphaned_slot;
spin_unlock(&OCFS2_I(inode)->ip_lock);
}
int ocfs2_request_delete_vote(struct inode *inode)
{
int orphaned_slot, status;
struct ocfs2_net_response_cb delete_cb;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
struct ocfs2_vote_msg *request;
spin_lock(&OCFS2_I(inode)->ip_lock);
orphaned_slot = OCFS2_I(inode)->ip_orphaned_slot;
spin_unlock(&OCFS2_I(inode)->ip_lock);
delete_cb.rc_cb = ocfs2_delete_response_cb;
delete_cb.rc_priv = inode;
mlog(0, "Inode %"MLFu64", we start thinking orphaned slot is %d\n",
OCFS2_I(inode)->ip_blkno, orphaned_slot);
status = -ENOMEM;
request = ocfs2_new_vote_request(osb, OCFS2_I(inode)->ip_blkno,
inode->i_generation,
OCFS2_VOTE_REQ_DELETE, orphaned_slot);
if (request) {
status = ocfs2_request_vote(inode, request, &delete_cb);
kfree(request);
}
return status;
}
static void ocfs2_setup_unlink_vote(struct ocfs2_vote_msg *request,
struct dentry *dentry)
{
struct inode *parent = dentry->d_parent->d_inode;
/* We need some values which will uniquely identify a dentry
* on the other nodes so that they can find it and run
* d_delete against it. Parent directory block and full name
* should suffice. */
mlog(0, "unlink/rename request: parent: %"MLFu64" name: %.*s\n",
OCFS2_I(parent)->ip_blkno, dentry->d_name.len,
dentry->d_name.name);
request->v_unlink_parent = cpu_to_be64(OCFS2_I(parent)->ip_blkno);
request->v_unlink_namelen = cpu_to_be32(dentry->d_name.len);
memcpy(request->v_unlink_dirent, dentry->d_name.name,
dentry->d_name.len);
}
int ocfs2_request_unlink_vote(struct inode *inode,
struct dentry *dentry,
unsigned int nlink)
{
int status;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
struct ocfs2_vote_msg *request;
if (dentry->d_name.len > OCFS2_VOTE_FILENAME_LEN)
return -ENAMETOOLONG;
status = -ENOMEM;
request = ocfs2_new_vote_request(osb, OCFS2_I(inode)->ip_blkno,
inode->i_generation,
OCFS2_VOTE_REQ_UNLINK, nlink);
if (request) {
ocfs2_setup_unlink_vote(request, dentry);
status = ocfs2_request_vote(inode, request, NULL);
kfree(request);
}
return status;
}
int ocfs2_request_rename_vote(struct inode *inode,
struct dentry *dentry)
{
int status;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
struct ocfs2_vote_msg *request;
if (dentry->d_name.len > OCFS2_VOTE_FILENAME_LEN)
return -ENAMETOOLONG;
status = -ENOMEM;
request = ocfs2_new_vote_request(osb, OCFS2_I(inode)->ip_blkno,
inode->i_generation,
OCFS2_VOTE_REQ_RENAME, 0);
if (request) {
ocfs2_setup_unlink_vote(request, dentry);
status = ocfs2_request_vote(inode, request, NULL);
kfree(request);
}
return status;
}
int ocfs2_request_mount_vote(struct ocfs2_super *osb)
{
int status;
struct ocfs2_vote_msg *request = NULL;
request = ocfs2_new_vote_request(osb, 0ULL, 0,
OCFS2_VOTE_REQ_MOUNT, 0);
if (!request) {
status = -ENOMEM;
goto bail;
}
status = -EAGAIN;
while (status == -EAGAIN) {
if (!(osb->s_mount_opt & OCFS2_MOUNT_NOINTR) &&
signal_pending(current)) {
status = -ERESTARTSYS;
goto bail;
}
if (ocfs2_node_map_is_only(osb, &osb->mounted_map,
osb->node_num)) {
status = 0;
goto bail;
}
status = ocfs2_do_request_vote(osb, request, NULL);
}
bail:
if (request)
kfree(request);
return status;
}
int ocfs2_request_umount_vote(struct ocfs2_super *osb)
{
int status;
struct ocfs2_vote_msg *request = NULL;
request = ocfs2_new_vote_request(osb, 0ULL, 0,
OCFS2_VOTE_REQ_UMOUNT, 0);
if (!request) {