]> git.proxmox.com Git - mirror_ubuntu-hirsute-kernel.git/blame - fs/ceph/mds_client.c
projects / mirror_ubuntu-hirsute-kernel.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Documentation: Fix grammatical error in sysctl/fs.txt & clarify negative dentry
[mirror_ubuntu-hirsute-kernel.git] / fs / ceph / mds_client.c
CommitLineData
b2441318 1// SPDX-License-Identifier: GPL-2.0
3d14c5d2 2#include <linux/ceph/ceph_debug.h>
2f2dc053 3
496e5955 4#include <linux/fs.h>
2f2dc053 5#include <linux/wait.h>
5a0e3ad6 6#include <linux/slab.h>
54008399 7#include <linux/gfp.h>
2f2dc053 8#include <linux/sched.h>
3d14c5d2
YS		 9#include <linux/debugfs.h>
10#include <linux/seq_file.h>
3e0708b9 11#include <linux/ratelimit.h>
2f2dc053 12
2f2dc053 13#include "super.h"
3d14c5d2
YS		 14#include "mds_client.h"
15
1fe60e51 16#include <linux/ceph/ceph_features.h>
3d14c5d2
YS		 17#include <linux/ceph/messenger.h>
18#include <linux/ceph/decode.h>
19#include <linux/ceph/pagelist.h>
20#include <linux/ceph/auth.h>
21#include <linux/ceph/debugfs.h>
2f2dc053
SW		 22
23/*
24 * A cluster of MDS (metadata server) daemons is responsible for
25 * managing the file system namespace (the directory hierarchy and
26 * inodes) and for coordinating shared access to storage. Metadata is
27 * partitioning hierarchically across a number of servers, and that
28 * partition varies over time as the cluster adjusts the distribution
29 * in order to balance load.
30 *
31 * The MDS client is primarily responsible to managing synchronous
32 * metadata requests for operations like open, unlink, and so forth.
33 * If there is a MDS failure, we find out about it when we (possibly
34 * request and) receive a new MDS map, and can resubmit affected
35 * requests.
36 *
37 * For the most part, though, we take advantage of a lossless
38 * communications channel to the MDS, and do not need to worry about
39 * timing out or resubmitting requests.
40 *
41 * We maintain a stateful "session" with each MDS we interact with.
42 * Within each session, we sent periodic heartbeat messages to ensure
43 * any capabilities or leases we have been issues remain valid. If
44 * the session times out and goes stale, our leases and capabilities
45 * are no longer valid.
46 */
47
20cb34ae 48struct ceph_reconnect_state {
44c99757 49 int nr_caps;
20cb34ae 50 struct ceph_pagelist *pagelist;
121f22a1 51 unsigned msg_version;
20cb34ae
SW		 52};
53
2f2dc053
SW		 54static void __wake_requests(struct ceph_mds_client *mdsc,
55 struct list_head *head);
56
9e32789f 57static const struct ceph_connection_operations mds_con_ops;
2f2dc053
SW		 58
59
60/*
61 * mds reply parsing
62 */
63
64/*
65 * parse individual inode info
66 */
67static int parse_reply_info_in(void **p, void *end,
14303d20 68 struct ceph_mds_reply_info_in *info,
12b4629a 69 u64 features)
2f2dc053
SW		 70{
71 int err = -EIO;
72
73 info->in = *p;
74 *p += sizeof(struct ceph_mds_reply_inode) +
75 sizeof(*info->in->fragtree.splits) *
76 le32_to_cpu(info->in->fragtree.nsplits);
77
78 ceph_decode_32_safe(p, end, info->symlink_len, bad);
79 ceph_decode_need(p, end, info->symlink_len, bad);
80 info->symlink = *p;
81 *p += info->symlink_len;
82
23c625ce
ID		 83 ceph_decode_copy_safe(p, end, &info->dir_layout,
84 sizeof(info->dir_layout), bad);
2f2dc053
SW		 85 ceph_decode_32_safe(p, end, info->xattr_len, bad);
86 ceph_decode_need(p, end, info->xattr_len, bad);
87 info->xattr_data = *p;
88 *p += info->xattr_len;
fb01d1f8
YZ		 89
90 if (features & CEPH_FEATURE_MDS_INLINE_DATA) {
91 ceph_decode_64_safe(p, end, info->inline_version, bad);
92 ceph_decode_32_safe(p, end, info->inline_len, bad);
93 ceph_decode_need(p, end, info->inline_len, bad);
94 info->inline_data = *p;
95 *p += info->inline_len;
96 } else
97 info->inline_version = CEPH_INLINE_NONE;
98
fb18a575
LH		 99 if (features & CEPH_FEATURE_MDS_QUOTA) {
100 u8 struct_v, struct_compat;
101 u32 struct_len;
102
103 /*
104 * both struct_v and struct_compat are expected to be >= 1
105 */
106 ceph_decode_8_safe(p, end, struct_v, bad);
107 ceph_decode_8_safe(p, end, struct_compat, bad);
108 if (!struct_v || !struct_compat)
109 goto bad;
110 ceph_decode_32_safe(p, end, struct_len, bad);
111 ceph_decode_need(p, end, struct_len, bad);
112 ceph_decode_64_safe(p, end, info->max_bytes, bad);
113 ceph_decode_64_safe(p, end, info->max_files, bad);
114 } else {
115 info->max_bytes = 0;
116 info->max_files = 0;
117 }
118
779fe0fb
YZ		 119 info->pool_ns_len = 0;
120 info->pool_ns_data = NULL;
5ea5c5e0
YZ		 121 if (features & CEPH_FEATURE_FS_FILE_LAYOUT_V2) {
122 ceph_decode_32_safe(p, end, info->pool_ns_len, bad);
779fe0fb
YZ		 123 if (info->pool_ns_len > 0) {
124 ceph_decode_need(p, end, info->pool_ns_len, bad);
125 info->pool_ns_data = *p;
126 *p += info->pool_ns_len;
127 }
5ea5c5e0
YZ		 128 }
129
2f2dc053
SW		 130 return 0;
131bad:
132 return err;
133}
134
135/*
136 * parse a normal reply, which may contain a (dir+)dentry and/or a
137 * target inode.
138 */
139static int parse_reply_info_trace(void **p, void *end,
14303d20 140 struct ceph_mds_reply_info_parsed *info,
12b4629a 141 u64 features)
2f2dc053
SW		 142{
143 int err;
144
145 if (info->head->is_dentry) {
14303d20 146 err = parse_reply_info_in(p, end, &info->diri, features);
2f2dc053
SW		 147 if (err < 0)
148 goto out_bad;
149
150 if (unlikely(*p + sizeof(*info->dirfrag) > end))
151 goto bad;
152 info->dirfrag = *p;
153 *p += sizeof(*info->dirfrag) +
154 sizeof(u32)*le32_to_cpu(info->dirfrag->ndist);
155 if (unlikely(*p > end))
156 goto bad;
157
158 ceph_decode_32_safe(p, end, info->dname_len, bad);
159 ceph_decode_need(p, end, info->dname_len, bad);
160 info->dname = *p;
161 *p += info->dname_len;
162 info->dlease = *p;
163 *p += sizeof(*info->dlease);
164 }
165
166 if (info->head->is_target) {
14303d20 167 err = parse_reply_info_in(p, end, &info->targeti, features);
2f2dc053
SW		 168 if (err < 0)
169 goto out_bad;
170 }
171
172 if (unlikely(*p != end))
173 goto bad;
174 return 0;
175
176bad:
177 err = -EIO;
178out_bad:
179 pr_err("problem parsing mds trace %d\n", err);
180 return err;
181}
182
183/*
184 * parse readdir results
185 */
186static int parse_reply_info_dir(void **p, void *end,
14303d20 187 struct ceph_mds_reply_info_parsed *info,
12b4629a 188 u64 features)
2f2dc053
SW		 189{
190 u32 num, i = 0;
191 int err;
192
193 info->dir_dir = *p;
194 if (*p + sizeof(*info->dir_dir) > end)
195 goto bad;
196 *p += sizeof(*info->dir_dir) +
197 sizeof(u32)*le32_to_cpu(info->dir_dir->ndist);
198 if (*p > end)
199 goto bad;
200
201 ceph_decode_need(p, end, sizeof(num) + 2, bad);
c89136ea 202 num = ceph_decode_32(p);
956d39d6
YZ		 203 {
204 u16 flags = ceph_decode_16(p);
205 info->dir_end = !!(flags & CEPH_READDIR_FRAG_END);
206 info->dir_complete = !!(flags & CEPH_READDIR_FRAG_COMPLETE);
f3c4ebe6 207 info->hash_order = !!(flags & CEPH_READDIR_HASH_ORDER);
79162547 208 info->offset_hash = !!(flags & CEPH_READDIR_OFFSET_HASH);
956d39d6 209 }
2f2dc053
SW		 210 if (num == 0)
211 goto done;
212
2a5beea3
YZ		 213 BUG_ON(!info->dir_entries);
214 if ((unsigned long)(info->dir_entries + num) >
215 (unsigned long)info->dir_entries + info->dir_buf_size) {
54008399
YZ		 216 pr_err("dir contents are larger than expected\n");
217 WARN_ON(1);
218 goto bad;
219 }
2f2dc053 220
54008399 221 info->dir_nr = num;
2f2dc053 222 while (num) {
2a5beea3 223 struct ceph_mds_reply_dir_entry *rde = info->dir_entries + i;
2f2dc053
SW		 224 /* dentry */
225 ceph_decode_need(p, end, sizeof(u32)*2, bad);
2a5beea3
YZ		 226 rde->name_len = ceph_decode_32(p);
227 ceph_decode_need(p, end, rde->name_len, bad);
228 rde->name = *p;
229 *p += rde->name_len;
230 dout("parsed dir dname '%.*s'\n", rde->name_len, rde->name);
231 rde->lease = *p;
2f2dc053
SW		 232 *p += sizeof(struct ceph_mds_reply_lease);
233
234 /* inode */
2a5beea3 235 err = parse_reply_info_in(p, end, &rde->inode, features);
2f2dc053
SW		 236 if (err < 0)
237 goto out_bad;
8974eebd
YZ		 238 /* ceph_readdir_prepopulate() will update it */
239 rde->offset = 0;
2f2dc053
SW		 240 i++;
241 num--;
242 }
243
244done:
245 if (*p != end)
246 goto bad;
247 return 0;
248
249bad:
250 err = -EIO;
251out_bad:
252 pr_err("problem parsing dir contents %d\n", err);
253 return err;
254}
255
25933abd
HS		 256/*
257 * parse fcntl F_GETLK results
258 */
259static int parse_reply_info_filelock(void **p, void *end,
14303d20 260 struct ceph_mds_reply_info_parsed *info,
12b4629a 261 u64 features)
25933abd
HS		 262{
263 if (*p + sizeof(*info->filelock_reply) > end)
264 goto bad;
265
266 info->filelock_reply = *p;
267 *p += sizeof(*info->filelock_reply);
268
269 if (unlikely(*p != end))
270 goto bad;
271 return 0;
272
273bad:
274 return -EIO;
275}
276
6e8575fa
SL		 277/*
278 * parse create results
279 */
280static int parse_reply_info_create(void **p, void *end,
281 struct ceph_mds_reply_info_parsed *info,
12b4629a 282 u64 features)
6e8575fa
SL		 283{
284 if (features & CEPH_FEATURE_REPLY_CREATE_INODE) {
285 if (*p == end) {
286 info->has_create_ino = false;
287 } else {
288 info->has_create_ino = true;
289 info->ino = ceph_decode_64(p);
290 }
291 }
292
293 if (unlikely(*p != end))
294 goto bad;
295 return 0;
296
297bad:
298 return -EIO;
299}
300
25933abd
HS		 301/*
302 * parse extra results
303 */
304static int parse_reply_info_extra(void **p, void *end,
14303d20 305 struct ceph_mds_reply_info_parsed *info,
12b4629a 306 u64 features)
25933abd 307{
6df8c9d8
JL		 308 u32 op = le32_to_cpu(info->head->op);
309
310 if (op == CEPH_MDS_OP_GETFILELOCK)
14303d20 311 return parse_reply_info_filelock(p, end, info, features);
6df8c9d8 312 else if (op == CEPH_MDS_OP_READDIR || op == CEPH_MDS_OP_LSSNAP)
14303d20 313 return parse_reply_info_dir(p, end, info, features);
6df8c9d8 314 else if (op == CEPH_MDS_OP_CREATE)
6e8575fa
SL		 315 return parse_reply_info_create(p, end, info, features);
316 else
317 return -EIO;
25933abd
HS		 318}
319
2f2dc053
SW		 320/*
321 * parse entire mds reply
322 */
323static int parse_reply_info(struct ceph_msg *msg,
14303d20 324 struct ceph_mds_reply_info_parsed *info,
12b4629a 325 u64 features)
2f2dc053
SW		 326{
327 void *p, *end;
328 u32 len;
329 int err;
330
331 info->head = msg->front.iov_base;
332 p = msg->front.iov_base + sizeof(struct ceph_mds_reply_head);
333 end = p + msg->front.iov_len - sizeof(struct ceph_mds_reply_head);
334
335 /* trace */
336 ceph_decode_32_safe(&p, end, len, bad);
337 if (len > 0) {
32852a81 338 ceph_decode_need(&p, end, len, bad);
14303d20 339 err = parse_reply_info_trace(&p, p+len, info, features);
2f2dc053
SW		 340 if (err < 0)
341 goto out_bad;
342 }
343
25933abd 344 /* extra */
2f2dc053
SW		 345 ceph_decode_32_safe(&p, end, len, bad);
346 if (len > 0) {
32852a81 347 ceph_decode_need(&p, end, len, bad);
14303d20 348 err = parse_reply_info_extra(&p, p+len, info, features);
2f2dc053
SW		 349 if (err < 0)
350 goto out_bad;
351 }
352
353 /* snap blob */
354 ceph_decode_32_safe(&p, end, len, bad);
355 info->snapblob_len = len;
356 info->snapblob = p;
357 p += len;
358
359 if (p != end)
360 goto bad;
361 return 0;
362
363bad:
364 err = -EIO;
365out_bad:
366 pr_err("mds parse_reply err %d\n", err);
367 return err;
368}
369
370static void destroy_reply_info(struct ceph_mds_reply_info_parsed *info)
371{
2a5beea3 372 if (!info->dir_entries)
54008399 373 return;
2a5beea3 374 free_pages((unsigned long)info->dir_entries, get_order(info->dir_buf_size));
2f2dc053
SW		 375}
376
377
378/*
379 * sessions
380 */
a687ecaf 381const char *ceph_session_state_name(int s)
2f2dc053
SW		 382{
383 switch (s) {
384 case CEPH_MDS_SESSION_NEW: return "new";
385 case CEPH_MDS_SESSION_OPENING: return "opening";
386 case CEPH_MDS_SESSION_OPEN: return "open";
387 case CEPH_MDS_SESSION_HUNG: return "hung";
388 case CEPH_MDS_SESSION_CLOSING: return "closing";
44ca18f2 389 case CEPH_MDS_SESSION_RESTARTING: return "restarting";
2f2dc053 390 case CEPH_MDS_SESSION_RECONNECTING: return "reconnecting";
fcff415c 391 case CEPH_MDS_SESSION_REJECTED: return "rejected";
2f2dc053
SW		 392 default: return "???";
393 }
394}
395
396static struct ceph_mds_session *get_session(struct ceph_mds_session *s)
397{
3997c01d 398 if (refcount_inc_not_zero(&s->s_ref)) {
2f2dc053 399 dout("mdsc get_session %p %d -> %d\n", s,
3997c01d 400 refcount_read(&s->s_ref)-1, refcount_read(&s->s_ref));
2f2dc053
SW		 401 return s;
402 } else {
4c069a58 403 dout("mdsc get_session %p 0 -- FAIL\n", s);
2f2dc053
SW		 404 return NULL;
405 }
406}
407
408void ceph_put_mds_session(struct ceph_mds_session *s)
409{
410 dout("mdsc put_session %p %d -> %d\n", s,
3997c01d
ER		 411 refcount_read(&s->s_ref), refcount_read(&s->s_ref)-1);
412 if (refcount_dec_and_test(&s->s_ref)) {
6c4a1915 413 if (s->s_auth.authorizer)
6c1ea260 414 ceph_auth_destroy_authorizer(s->s_auth.authorizer);
2f2dc053 415 kfree(s);
4e7a5dcd 416 }
2f2dc053
SW		 417}
418
419/*
420 * called under mdsc->mutex
421 */
422struct ceph_mds_session *__ceph_lookup_mds_session(struct ceph_mds_client *mdsc,
423 int mds)
424{
425 struct ceph_mds_session *session;
426
d37b1d99 427 if (mds >= mdsc->max_sessions || !mdsc->sessions[mds])
2f2dc053
SW		 428 return NULL;
429 session = mdsc->sessions[mds];
430 dout("lookup_mds_session %p %d\n", session,
3997c01d 431 refcount_read(&session->s_ref));
2f2dc053
SW		 432 get_session(session);
433 return session;
434}
435
436static bool __have_session(struct ceph_mds_client *mdsc, int mds)
437{
98cfda81 438 if (mds >= mdsc->max_sessions || !mdsc->sessions[mds])
2f2dc053 439 return false;
98cfda81
CX		 440 else
441 return true;
2f2dc053
SW		 442}
443
2600d2dd
SW		 444static int __verify_registered_session(struct ceph_mds_client *mdsc,
445 struct ceph_mds_session *s)
446{
447 if (s->s_mds >= mdsc->max_sessions ||
448 mdsc->sessions[s->s_mds] != s)
449 return -ENOENT;
450 return 0;
451}
452
2f2dc053
SW		 453/*
454 * create+register a new session for given mds.
455 * called under mdsc->mutex.
456 */
457static struct ceph_mds_session *register_session(struct ceph_mds_client *mdsc,
458 int mds)
459{
460 struct ceph_mds_session *s;
461
76201b63 462 if (mds >= mdsc->mdsmap->m_num_mds)
c338c07c
NY		 463 return ERR_PTR(-EINVAL);
464
2f2dc053 465 s = kzalloc(sizeof(*s), GFP_NOFS);
4736b009
DC		 466 if (!s)
467 return ERR_PTR(-ENOMEM);
47474d0b
CX		 468
469 if (mds >= mdsc->max_sessions) {
470 int newmax = 1 << get_count_order(mds + 1);
471 struct ceph_mds_session **sa;
472
473 dout("%s: realloc to %d\n", __func__, newmax);
474 sa = kcalloc(newmax, sizeof(void *), GFP_NOFS);
475 if (!sa)
476 goto fail_realloc;
477 if (mdsc->sessions) {
478 memcpy(sa, mdsc->sessions,
479 mdsc->max_sessions * sizeof(void *));
480 kfree(mdsc->sessions);
481 }
482 mdsc->sessions = sa;
483 mdsc->max_sessions = newmax;
484 }
485
486 dout("%s: mds%d\n", __func__, mds);
2f2dc053
SW		 487 s->s_mdsc = mdsc;
488 s->s_mds = mds;
489 s->s_state = CEPH_MDS_SESSION_NEW;
490 s->s_ttl = 0;
491 s->s_seq = 0;
492 mutex_init(&s->s_mutex);
493
b7a9e5dd 494 ceph_con_init(&s->s_con, s, &mds_con_ops, &mdsc->fsc->client->msgr);
2f2dc053 495
d8fb02ab 496 spin_lock_init(&s->s_gen_ttl_lock);
2f2dc053 497 s->s_cap_gen = 0;
1ce208a6 498 s->s_cap_ttl = jiffies - 1;
d8fb02ab
AE		 499
500 spin_lock_init(&s->s_cap_lock);
2f2dc053
SW		 501 s->s_renew_requested = 0;
502 s->s_renew_seq = 0;
503 INIT_LIST_HEAD(&s->s_caps);
504 s->s_nr_caps = 0;
5dacf091 505 s->s_trim_caps = 0;
3997c01d 506 refcount_set(&s->s_ref, 1);
2f2dc053
SW		 507 INIT_LIST_HEAD(&s->s_waiting);
508 INIT_LIST_HEAD(&s->s_unsafe);
509 s->s_num_cap_releases = 0;
99a9c273 510 s->s_cap_reconnect = 0;
7c1332b8 511 s->s_cap_iterator = NULL;
2f2dc053 512 INIT_LIST_HEAD(&s->s_cap_releases);
2f2dc053 513 INIT_LIST_HEAD(&s->s_cap_flushing);
2f2dc053 514
2f2dc053 515 mdsc->sessions[mds] = s;
86d8f67b 516 atomic_inc(&mdsc->num_sessions);
3997c01d 517 refcount_inc(&s->s_ref); /* one ref to sessions[], one to caller */
42ce56e5 518
b7a9e5dd
SW		 519 ceph_con_open(&s->s_con, CEPH_ENTITY_TYPE_MDS, mds,
520 ceph_mdsmap_get_addr(mdsc->mdsmap, mds));
42ce56e5 521
2f2dc053 522 return s;
42ce56e5
SW		 523
524fail_realloc:
525 kfree(s);
526 return ERR_PTR(-ENOMEM);
2f2dc053
SW		 527}
528
529/*
530 * called under mdsc->mutex
531 */
2600d2dd 532static void __unregister_session(struct ceph_mds_client *mdsc,
42ce56e5 533 struct ceph_mds_session *s)
2f2dc053 534{
2600d2dd
SW		 535 dout("__unregister_session mds%d %p\n", s->s_mds, s);
536 BUG_ON(mdsc->sessions[s->s_mds] != s);
42ce56e5
SW		 537 mdsc->sessions[s->s_mds] = NULL;
538 ceph_con_close(&s->s_con);
539 ceph_put_mds_session(s);
86d8f67b 540 atomic_dec(&mdsc->num_sessions);
2f2dc053
SW		 541}
542
543/*
544 * drop session refs in request.
545 *
546 * should be last request ref, or hold mdsc->mutex
547 */
548static void put_request_session(struct ceph_mds_request *req)
549{
550 if (req->r_session) {
551 ceph_put_mds_session(req->r_session);
552 req->r_session = NULL;
553 }
554}
555
153c8e6b 556void ceph_mdsc_release_request(struct kref *kref)
2f2dc053 557{
153c8e6b
SW		 558 struct ceph_mds_request *req = container_of(kref,
559 struct ceph_mds_request,
560 r_kref);
54008399 561 destroy_reply_info(&req->r_reply_info);
153c8e6b
SW		 562 if (req->r_request)
563 ceph_msg_put(req->r_request);
54008399 564 if (req->r_reply)
153c8e6b 565 ceph_msg_put(req->r_reply);
153c8e6b 566 if (req->r_inode) {
41b02e1f 567 ceph_put_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
153c8e6b
SW		 568 iput(req->r_inode);
569 }
3dd69aab
JL		 570 if (req->r_parent)
571 ceph_put_cap_refs(ceph_inode(req->r_parent), CEPH_CAP_PIN);
e96a650a 572 iput(req->r_target_inode);
153c8e6b
SW		 573 if (req->r_dentry)
574 dput(req->r_dentry);
844d87c3
SW		 575 if (req->r_old_dentry)
576 dput(req->r_old_dentry);
577 if (req->r_old_dentry_dir) {
41b02e1f
SW		 578 /*
579 * track (and drop pins for) r_old_dentry_dir
580 * separately, since r_old_dentry's d_parent may have
581 * changed between the dir mutex being dropped and
582 * this request being freed.
583 */
584 ceph_put_cap_refs(ceph_inode(req->r_old_dentry_dir),
585 CEPH_CAP_PIN);
41b02e1f 586 iput(req->r_old_dentry_dir);
2f2dc053 587 }
153c8e6b
SW		 588 kfree(req->r_path1);
589 kfree(req->r_path2);
25e6bae3
YZ		 590 if (req->r_pagelist)
591 ceph_pagelist_release(req->r_pagelist);
153c8e6b 592 put_request_session(req);
37151668 593 ceph_unreserve_caps(req->r_mdsc, &req->r_caps_reservation);
153c8e6b 594 kfree(req);
2f2dc053
SW		 595}
596
fcd00b68
ID		 597DEFINE_RB_FUNCS(request, struct ceph_mds_request, r_tid, r_node)
598
2f2dc053
SW		 599/*
600 * lookup session, bump ref if found.
601 *
602 * called under mdsc->mutex.
603 */
fcd00b68
ID		 604static struct ceph_mds_request *
605lookup_get_request(struct ceph_mds_client *mdsc, u64 tid)
2f2dc053
SW		 606{
607 struct ceph_mds_request *req;
44ca18f2 608
fcd00b68
ID		 609 req = lookup_request(&mdsc->request_tree, tid);
610 if (req)
611 ceph_mdsc_get_request(req);
44ca18f2 612
fcd00b68 613 return req;
2f2dc053
SW		 614}
615
616/*
617 * Register an in-flight request, and assign a tid. Link to directory
618 * are modifying (if any).
619 *
620 * Called under mdsc->mutex.
621 */
622static void __register_request(struct ceph_mds_client *mdsc,
623 struct ceph_mds_request *req,
624 struct inode *dir)
625{
e30ee581
ZZ		 626 int ret = 0;
627
2f2dc053 628 req->r_tid = ++mdsc->last_tid;
e30ee581
ZZ		 629 if (req->r_num_caps) {
630 ret = ceph_reserve_caps(mdsc, &req->r_caps_reservation,
631 req->r_num_caps);
632 if (ret < 0) {
633 pr_err("__register_request %p "
634 "failed to reserve caps: %d\n", req, ret);
635 /* set req->r_err to fail early from __do_request */
636 req->r_err = ret;
637 return;
638 }
639 }
2f2dc053
SW		 640 dout("__register_request %p tid %lld\n", req, req->r_tid);
641 ceph_mdsc_get_request(req);
fcd00b68 642 insert_request(&mdsc->request_tree, req);
2f2dc053 643
cb4276cc
SW		 644 req->r_uid = current_fsuid();
645 req->r_gid = current_fsgid();
646
e8a7b8b1
YZ		 647 if (mdsc->oldest_tid == 0 && req->r_op != CEPH_MDS_OP_SETFILELOCK)
648 mdsc->oldest_tid = req->r_tid;
649
2f2dc053 650 if (dir) {
3b663780 651 ihold(dir);
2f2dc053 652 req->r_unsafe_dir = dir;
2f2dc053
SW		 653 }
654}
655
656static void __unregister_request(struct ceph_mds_client *mdsc,
657 struct ceph_mds_request *req)
658{
659 dout("__unregister_request %p tid %lld\n", req, req->r_tid);
e8a7b8b1 660
df963ea8
JL		 661 /* Never leave an unregistered request on an unsafe list! */
662 list_del_init(&req->r_unsafe_item);
663
e8a7b8b1
YZ		 664 if (req->r_tid == mdsc->oldest_tid) {
665 struct rb_node *p = rb_next(&req->r_node);
666 mdsc->oldest_tid = 0;
667 while (p) {
668 struct ceph_mds_request *next_req =
669 rb_entry(p, struct ceph_mds_request, r_node);
670 if (next_req->r_op != CEPH_MDS_OP_SETFILELOCK) {
671 mdsc->oldest_tid = next_req->r_tid;
672 break;
673 }
674 p = rb_next(p);
675 }
676 }
677
fcd00b68 678 erase_request(&mdsc->request_tree, req);
2f2dc053 679
bc2de10d
JL		 680 if (req->r_unsafe_dir &&
681 test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
2f2dc053 682 struct ceph_inode_info *ci = ceph_inode(req->r_unsafe_dir);
2f2dc053
SW		 683 spin_lock(&ci->i_unsafe_lock);
684 list_del_init(&req->r_unsafe_dir_item);
685 spin_unlock(&ci->i_unsafe_lock);
4c06ace8 686 }
bc2de10d
JL		 687 if (req->r_target_inode &&
688 test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
68cd5b4b
YZ		 689 struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
690 spin_lock(&ci->i_unsafe_lock);
691 list_del_init(&req->r_unsafe_target_item);
692 spin_unlock(&ci->i_unsafe_lock);
693 }
3b663780 694
4c06ace8 695 if (req->r_unsafe_dir) {
3b663780
SW		 696 iput(req->r_unsafe_dir);
697 req->r_unsafe_dir = NULL;
2f2dc053 698 }
94aa8ae1 699
fc55d2c9
YZ		 700 complete_all(&req->r_safe_completion);
701
94aa8ae1 702 ceph_mdsc_put_request(req);
2f2dc053
SW		 703}
704
30c71233
JL		 705/*
706 * Walk back up the dentry tree until we hit a dentry representing a
707 * non-snapshot inode. We do this using the rcu_read_lock (which must be held
708 * when calling this) to ensure that the objects won't disappear while we're
709 * working with them. Once we hit a candidate dentry, we attempt to take a
710 * reference to it, and return that as the result.
711 */
f1075480
DC		 712static struct inode *get_nonsnap_parent(struct dentry *dentry)
713{
714 struct inode *inode = NULL;
30c71233
JL		 715
716 while (dentry && !IS_ROOT(dentry)) {
717 inode = d_inode_rcu(dentry);
718 if (!inode || ceph_snap(inode) == CEPH_NOSNAP)
719 break;
720 dentry = dentry->d_parent;
721 }
722 if (inode)
723 inode = igrab(inode);
724 return inode;
725}
726
2f2dc053
SW		 727/*
728 * Choose mds to send request to next. If there is a hint set in the
729 * request (e.g., due to a prior forward hint from the mds), use that.
730 * Otherwise, consult frag tree and/or caps to identify the
731 * appropriate mds. If all else fails, choose randomly.
732 *
733 * Called under mdsc->mutex.
734 */
735static int __choose_mds(struct ceph_mds_client *mdsc,
736 struct ceph_mds_request *req)
737{
738 struct inode *inode;
739 struct ceph_inode_info *ci;
740 struct ceph_cap *cap;
741 int mode = req->r_direct_mode;
742 int mds = -1;
743 u32 hash = req->r_direct_hash;
bc2de10d 744 bool is_hash = test_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags);
2f2dc053
SW		 745
746 /*
747 * is there a specific mds we should try? ignore hint if we have
748 * no session and the mds is not up (active or recovering).
749 */
750 if (req->r_resend_mds >= 0 &&
751 (__have_session(mdsc, req->r_resend_mds) ||
752 ceph_mdsmap_get_state(mdsc->mdsmap, req->r_resend_mds) > 0)) {
753 dout("choose_mds using resend_mds mds%d\n",
754 req->r_resend_mds);
755 return req->r_resend_mds;
756 }
757
758 if (mode == USE_RANDOM_MDS)
759 goto random;
760
761 inode = NULL;
762 if (req->r_inode) {
5d37ca14
YZ		 763 if (ceph_snap(req->r_inode) != CEPH_SNAPDIR) {
764 inode = req->r_inode;
765 ihold(inode);
766 } else {
38f340cc
YZ		 767 /* req->r_dentry is non-null for LSSNAP request */
768 rcu_read_lock();
769 inode = get_nonsnap_parent(req->r_dentry);
770 rcu_read_unlock();
771 dout("__choose_mds using snapdir's parent %p\n", inode);
5d37ca14 772 }
38f340cc 773 } else if (req->r_dentry) {
d79698da 774 /* ignore race with rename; old or new d_parent is okay */
30c71233
JL		 775 struct dentry *parent;
776 struct inode *dir;
777
778 rcu_read_lock();
779 parent = req->r_dentry->d_parent;
3dd69aab 780 dir = req->r_parent ? : d_inode_rcu(parent);
eb6bb1c5 781
30c71233
JL		 782 if (!dir || dir->i_sb != mdsc->fsc->sb) {
783 /* not this fs or parent went negative */
2b0143b5 784 inode = d_inode(req->r_dentry);
30c71233
JL		 785 if (inode)
786 ihold(inode);
eb6bb1c5
SW		 787 } else if (ceph_snap(dir) != CEPH_NOSNAP) {
788 /* direct snapped/virtual snapdir requests
789 * based on parent dir inode */
30c71233 790 inode = get_nonsnap_parent(parent);
eb6bb1c5 791 dout("__choose_mds using nonsnap parent %p\n", inode);
ca18bede 792 } else {
eb6bb1c5 793 /* dentry target */
2b0143b5 794 inode = d_inode(req->r_dentry);
ca18bede
YZ		 795 if (!inode || mode == USE_AUTH_MDS) {
796 /* dir + name */
30c71233 797 inode = igrab(dir);
ca18bede
YZ		 798 hash = ceph_dentry_hash(dir, req->r_dentry);
799 is_hash = true;
30c71233
JL		 800 } else {
801 ihold(inode);
ca18bede 802 }
2f2dc053 803 }
30c71233 804 rcu_read_unlock();
2f2dc053 805 }
eb6bb1c5 806
2f2dc053
SW		 807 dout("__choose_mds %p is_hash=%d (%d) mode %d\n", inode, (int)is_hash,
808 (int)hash, mode);
809 if (!inode)
810 goto random;
811 ci = ceph_inode(inode);
812
813 if (is_hash && S_ISDIR(inode->i_mode)) {
814 struct ceph_inode_frag frag;
815 int found;
816
817 ceph_choose_frag(ci, hash, &frag, &found);
818 if (found) {
819 if (mode == USE_ANY_MDS && frag.ndist > 0) {
820 u8 r;
821
822 /* choose a random replica */
823 get_random_bytes(&r, 1);
824 r %= frag.ndist;
825 mds = frag.dist[r];
826 dout("choose_mds %p %llx.%llx "
827 "frag %u mds%d (%d/%d)\n",
828 inode, ceph_vinop(inode),
d66bbd44 829 frag.frag, mds,
2f2dc053 830 (int)r, frag.ndist);
d66bbd44
SW		 831 if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
832 CEPH_MDS_STATE_ACTIVE)
30c71233 833 goto out;
2f2dc053
SW		 834 }
835
836 /* since this file/dir wasn't known to be
837 * replicated, then we want to look for the
838 * authoritative mds. */
839 mode = USE_AUTH_MDS;
840 if (frag.mds >= 0) {
841 /* choose auth mds */
842 mds = frag.mds;
843 dout("choose_mds %p %llx.%llx "
844 "frag %u mds%d (auth)\n",
845 inode, ceph_vinop(inode), frag.frag, mds);
d66bbd44
SW		 846 if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
847 CEPH_MDS_STATE_ACTIVE)
30c71233 848 goto out;
2f2dc053
SW		 849 }
850 }
851 }
852
be655596 853 spin_lock(&ci->i_ceph_lock);
2f2dc053
SW		 854 cap = NULL;
855 if (mode == USE_AUTH_MDS)
856 cap = ci->i_auth_cap;
857 if (!cap && !RB_EMPTY_ROOT(&ci->i_caps))
858 cap = rb_entry(rb_first(&ci->i_caps), struct ceph_cap, ci_node);
859 if (!cap) {
be655596 860 spin_unlock(&ci->i_ceph_lock);
30c71233 861 iput(inode);
2f2dc053
SW		 862 goto random;
863 }
864 mds = cap->session->s_mds;
865 dout("choose_mds %p %llx.%llx mds%d (%scap %p)\n",
866 inode, ceph_vinop(inode), mds,
867 cap == ci->i_auth_cap ? "auth " : "", cap);
be655596 868 spin_unlock(&ci->i_ceph_lock);
30c71233
JL		 869out:
870 iput(inode);
2f2dc053
SW		 871 return mds;
872
873random:
874 mds = ceph_mdsmap_get_random_mds(mdsc->mdsmap);
875 dout("choose_mds chose random mds%d\n", mds);
876 return mds;
877}
878
879
880/*
881 * session messages
882 */
883static struct ceph_msg *create_session_msg(u32 op, u64 seq)
884{
885 struct ceph_msg *msg;
886 struct ceph_mds_session_head *h;
887
b61c2763
SW		 888 msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h), GFP_NOFS,
889 false);
a79832f2 890 if (!msg) {
2f2dc053 891 pr_err("create_session_msg ENOMEM creating msg\n");
a79832f2 892 return NULL;
2f2dc053
SW		 893 }
894 h = msg->front.iov_base;
895 h->op = cpu_to_le32(op);
896 h->seq = cpu_to_le64(seq);
dbd0c8bf
JS		 897
898 return msg;
899}
900
342ce182
YZ		 901static void encode_supported_features(void **p, void *end)
902{
903 static const unsigned char bits[] = CEPHFS_FEATURES_CLIENT_SUPPORTED;
904 static const size_t count = ARRAY_SIZE(bits);
905
906 if (count > 0) {
907 size_t i;
908 size_t size = ((size_t)bits[count - 1] + 64) / 64 * 8;
909
910 BUG_ON(*p + 4 + size > end);
911 ceph_encode_32(p, size);
912 memset(*p, 0, size);
913 for (i = 0; i < count; i++)
914 ((unsigned char*)(*p))[i / 8] |= 1 << (bits[i] % 8);
915 *p += size;
916 } else {
917 BUG_ON(*p + 4 > end);
918 ceph_encode_32(p, 0);
919 }
920}
921
dbd0c8bf
JS		 922/*
923 * session message, specialization for CEPH_SESSION_REQUEST_OPEN
924 * to include additional client metadata fields.
925 */
926static struct ceph_msg *create_session_open_msg(struct ceph_mds_client *mdsc, u64 seq)
927{
928 struct ceph_msg *msg;
929 struct ceph_mds_session_head *h;
930 int i = -1;
342ce182 931 int extra_bytes = 0;
dbd0c8bf
JS		 932 int metadata_key_count = 0;
933 struct ceph_options *opt = mdsc->fsc->client->options;
3f384954 934 struct ceph_mount_options *fsopt = mdsc->fsc->mount_options;
342ce182 935 void *p, *end;
dbd0c8bf 936
a6a5ce4f 937 const char* metadata[][2] = {
717e6f28
YZ		 938 {"hostname", mdsc->nodename},
939 {"kernel_version", init_utsname()->release},
3f384954
YZ		 940 {"entity_id", opt->name ? : ""},
941 {"root", fsopt->server_path ? : "/"},
dbd0c8bf
JS		 942 {NULL, NULL}
943 };
944
945 /* Calculate serialized length of metadata */
342ce182 946 extra_bytes = 4; /* map length */
d37b1d99 947 for (i = 0; metadata[i][0]; ++i) {
342ce182 948 extra_bytes += 8 + strlen(metadata[i][0]) +
dbd0c8bf
JS		 949 strlen(metadata[i][1]);
950 metadata_key_count++;
951 }
342ce182
YZ		 952 /* supported feature */
953 extra_bytes += 4 + 8;
dbd0c8bf
JS		 954
955 /* Allocate the message */
342ce182 956 msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h) + extra_bytes,
dbd0c8bf
JS		 957 GFP_NOFS, false);
958 if (!msg) {
959 pr_err("create_session_msg ENOMEM creating msg\n");
960 return NULL;
961 }
342ce182
YZ		 962 p = msg->front.iov_base;
963 end = p + msg->front.iov_len;
964
965 h = p;
dbd0c8bf
JS		 966 h->op = cpu_to_le32(CEPH_SESSION_REQUEST_OPEN);
967 h->seq = cpu_to_le64(seq);
968
969 /*
970 * Serialize client metadata into waiting buffer space, using
971 * the format that userspace expects for map<string, string>
7cfa0313
JS		 972 *
973 * ClientSession messages with metadata are v2
dbd0c8bf 974 */
342ce182 975 msg->hdr.version = cpu_to_le16(3);
7cfa0313 976 msg->hdr.compat_version = cpu_to_le16(1);
dbd0c8bf
JS		 977
978 /* The write pointer, following the session_head structure */
342ce182 979 p += sizeof(*h);
dbd0c8bf
JS		 980
981 /* Number of entries in the map */
982 ceph_encode_32(&p, metadata_key_count);
983
984 /* Two length-prefixed strings for each entry in the map */
d37b1d99 985 for (i = 0; metadata[i][0]; ++i) {
dbd0c8bf
JS		 986 size_t const key_len = strlen(metadata[i][0]);
987 size_t const val_len = strlen(metadata[i][1]);
988
989 ceph_encode_32(&p, key_len);
990 memcpy(p, metadata[i][0], key_len);
991 p += key_len;
992 ceph_encode_32(&p, val_len);
993 memcpy(p, metadata[i][1], val_len);
994 p += val_len;
995 }
996
342ce182
YZ		 997 encode_supported_features(&p, end);
998 msg->front.iov_len = p - msg->front.iov_base;
999 msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
1000
2f2dc053
SW		 1001 return msg;
1002}
1003
1004/*
1005 * send session open request.
1006 *
1007 * called under mdsc->mutex
1008 */
1009static int __open_session(struct ceph_mds_client *mdsc,
1010 struct ceph_mds_session *session)
1011{
1012 struct ceph_msg *msg;
1013 int mstate;
1014 int mds = session->s_mds;
2f2dc053
SW		 1015
1016 /* wait for mds to go active? */
1017 mstate = ceph_mdsmap_get_state(mdsc->mdsmap, mds);
1018 dout("open_session to mds%d (%s)\n", mds,
1019 ceph_mds_state_name(mstate));
1020 session->s_state = CEPH_MDS_SESSION_OPENING;
1021 session->s_renew_requested = jiffies;
1022
1023 /* send connect message */
dbd0c8bf 1024 msg = create_session_open_msg(mdsc, session->s_seq);
a79832f2
SW		 1025 if (!msg)
1026 return -ENOMEM;
2f2dc053 1027 ceph_con_send(&session->s_con, msg);
2f2dc053
SW		 1028 return 0;
1029}
1030
ed0552a1
SW		 1031/*
1032 * open sessions for any export targets for the given mds
1033 *
1034 * called under mdsc->mutex
1035 */
5d72d13c
YZ		 1036static struct ceph_mds_session *
1037__open_export_target_session(struct ceph_mds_client *mdsc, int target)
1038{
1039 struct ceph_mds_session *session;
1040
1041 session = __ceph_lookup_mds_session(mdsc, target);
1042 if (!session) {
1043 session = register_session(mdsc, target);
1044 if (IS_ERR(session))
1045 return session;
1046 }
1047 if (session->s_state == CEPH_MDS_SESSION_NEW ||
1048 session->s_state == CEPH_MDS_SESSION_CLOSING)
1049 __open_session(mdsc, session);
1050
1051 return session;
1052}
1053
1054struct ceph_mds_session *
1055ceph_mdsc_open_export_target_session(struct ceph_mds_client *mdsc, int target)
1056{
1057 struct ceph_mds_session *session;
1058
1059 dout("open_export_target_session to mds%d\n", target);
1060
1061 mutex_lock(&mdsc->mutex);
1062 session = __open_export_target_session(mdsc, target);
1063 mutex_unlock(&mdsc->mutex);
1064
1065 return session;
1066}
1067
ed0552a1
SW		 1068static void __open_export_target_sessions(struct ceph_mds_client *mdsc,
1069 struct ceph_mds_session *session)
1070{
1071 struct ceph_mds_info *mi;
1072 struct ceph_mds_session *ts;
1073 int i, mds = session->s_mds;
ed0552a1 1074
76201b63 1075 if (mds >= mdsc->mdsmap->m_num_mds)
ed0552a1 1076 return;
5d72d13c 1077
ed0552a1
SW		 1078 mi = &mdsc->mdsmap->m_info[mds];
1079 dout("open_export_target_sessions for mds%d (%d targets)\n",
1080 session->s_mds, mi->num_export_targets);
1081
1082 for (i = 0; i < mi->num_export_targets; i++) {
5d72d13c
YZ		 1083 ts = __open_export_target_session(mdsc, mi->export_targets[i]);
1084 if (!IS_ERR(ts))
1085 ceph_put_mds_session(ts);
ed0552a1
SW		 1086 }
1087}
1088
154f42c2
SW		 1089void ceph_mdsc_open_export_target_sessions(struct ceph_mds_client *mdsc,
1090 struct ceph_mds_session *session)
1091{
1092 mutex_lock(&mdsc->mutex);
1093 __open_export_target_sessions(mdsc, session);
1094 mutex_unlock(&mdsc->mutex);
1095}
1096
2f2dc053
SW		 1097/*
1098 * session caps
1099 */
1100
c8a96a31
JL		 1101static void detach_cap_releases(struct ceph_mds_session *session,
1102 struct list_head *target)
2f2dc053 1103{
c8a96a31
JL		 1104 lockdep_assert_held(&session->s_cap_lock);
1105
1106 list_splice_init(&session->s_cap_releases, target);
745a8e3b 1107 session->s_num_cap_releases = 0;
c8a96a31
JL		 1108 dout("dispose_cap_releases mds%d\n", session->s_mds);
1109}
2f2dc053 1110
c8a96a31
JL		 1111static void dispose_cap_releases(struct ceph_mds_client *mdsc,
1112 struct list_head *dispose)
1113{
1114 while (!list_empty(dispose)) {
745a8e3b
YZ		 1115 struct ceph_cap *cap;
1116 /* zero out the in-progress message */
c8a96a31 1117 cap = list_first_entry(dispose, struct ceph_cap, session_caps);
745a8e3b
YZ		 1118 list_del(&cap->session_caps);
1119 ceph_put_cap(mdsc, cap);
2f2dc053 1120 }
2f2dc053
SW		 1121}
1122
1c841a96
YZ		 1123static void cleanup_session_requests(struct ceph_mds_client *mdsc,
1124 struct ceph_mds_session *session)
1125{
1126 struct ceph_mds_request *req;
1127 struct rb_node *p;
1128
1129 dout("cleanup_session_requests mds%d\n", session->s_mds);
1130 mutex_lock(&mdsc->mutex);
1131 while (!list_empty(&session->s_unsafe)) {
1132 req = list_first_entry(&session->s_unsafe,
1133 struct ceph_mds_request, r_unsafe_item);
3e0708b9
YZ		 1134 pr_warn_ratelimited(" dropping unsafe request %llu\n",
1135 req->r_tid);
1c841a96
YZ		 1136 __unregister_request(mdsc, req);
1137 }
1138 /* zero r_attempts, so kick_requests() will re-send requests */
1139 p = rb_first(&mdsc->request_tree);
1140 while (p) {
1141 req = rb_entry(p, struct ceph_mds_request, r_node);
1142 p = rb_next(p);
1143 if (req->r_session &&
1144 req->r_session->s_mds == session->s_mds)
1145 req->r_attempts = 0;
1146 }
1147 mutex_unlock(&mdsc->mutex);
1148}
1149
2f2dc053 1150/*
f818a736
SW		 1151 * Helper to safely iterate over all caps associated with a session, with
1152 * special care taken to handle a racing __ceph_remove_cap().
2f2dc053 1153 *
f818a736 1154 * Caller must hold session s_mutex.
2f2dc053
SW		 1155 */
1156static int iterate_session_caps(struct ceph_mds_session *session,
1157 int (*cb)(struct inode *, struct ceph_cap *,
1158 void *), void *arg)
1159{
7c1332b8
SW		 1160 struct list_head *p;
1161 struct ceph_cap *cap;
1162 struct inode *inode, *last_inode = NULL;
1163 struct ceph_cap *old_cap = NULL;
2f2dc053
SW		 1164 int ret;
1165
1166 dout("iterate_session_caps %p mds%d\n", session, session->s_mds);
1167 spin_lock(&session->s_cap_lock);
7c1332b8
SW		 1168 p = session->s_caps.next;
1169 while (p != &session->s_caps) {
1170 cap = list_entry(p, struct ceph_cap, session_caps);
2f2dc053 1171 inode = igrab(&cap->ci->vfs_inode);
7c1332b8
SW		 1172 if (!inode) {
1173 p = p->next;
2f2dc053 1174 continue;
7c1332b8
SW		 1175 }
1176 session->s_cap_iterator = cap;
2f2dc053 1177 spin_unlock(&session->s_cap_lock);
7c1332b8
SW		 1178
1179 if (last_inode) {
1180 iput(last_inode);
1181 last_inode = NULL;
1182 }
1183 if (old_cap) {
37151668 1184 ceph_put_cap(session->s_mdsc, old_cap);
7c1332b8
SW		 1185 old_cap = NULL;
1186 }
1187
2f2dc053 1188 ret = cb(inode, cap, arg);
7c1332b8
SW		 1189 last_inode = inode;
1190
2f2dc053 1191 spin_lock(&session->s_cap_lock);
7c1332b8 1192 p = p->next;
d37b1d99 1193 if (!cap->ci) {
7c1332b8
SW		 1194 dout("iterate_session_caps finishing cap %p removal\n",
1195 cap);
1196 BUG_ON(cap->session != session);
745a8e3b 1197 cap->session = NULL;
7c1332b8
SW		 1198 list_del_init(&cap->session_caps);
1199 session->s_nr_caps--;
745a8e3b
YZ		 1200 if (cap->queue_release) {
1201 list_add_tail(&cap->session_caps,
1202 &session->s_cap_releases);
1203 session->s_num_cap_releases++;
1204 } else {
1205 old_cap = cap; /* put_cap it w/o locks held */
1206 }
7c1332b8 1207 }
5dacf091
SW		 1208 if (ret < 0)
1209 goto out;
2f2dc053 1210 }
5dacf091
SW		 1211 ret = 0;
1212out:
7c1332b8 1213 session->s_cap_iterator = NULL;
2f2dc053 1214 spin_unlock(&session->s_cap_lock);
7c1332b8 1215
e96a650a 1216 iput(last_inode);
7c1332b8 1217 if (old_cap)
37151668 1218 ceph_put_cap(session->s_mdsc, old_cap);
7c1332b8 1219
5dacf091 1220 return ret;
2f2dc053
SW		 1221}
1222
1223static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
6c99f254 1224 void *arg)
2f2dc053 1225{
6c93df5d 1226 struct ceph_fs_client *fsc = (struct ceph_fs_client *)arg;
2f2dc053 1227 struct ceph_inode_info *ci = ceph_inode(inode);
553adfd9 1228 LIST_HEAD(to_remove);
6c93df5d
YZ		 1229 bool drop = false;
1230 bool invalidate = false;
6c99f254 1231
2f2dc053
SW		 1232 dout("removing cap %p, ci is %p, inode is %p\n",
1233 cap, ci, &ci->vfs_inode);
be655596 1234 spin_lock(&ci->i_ceph_lock);
d2f8bb27
YZ		 1235 if (cap->mds_wanted | cap->issued)
1236 ci->i_ceph_flags |= CEPH_I_CAP_DROPPED;
a096b09a 1237 __ceph_remove_cap(cap, false);
571ade33 1238 if (!ci->i_auth_cap) {
553adfd9 1239 struct ceph_cap_flush *cf;
6c93df5d 1240 struct ceph_mds_client *mdsc = fsc->mdsc;
6c99f254 1241
6c93df5d 1242 if (ci->i_wrbuffer_ref > 0 &&
52953d55 1243 READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
6c93df5d
YZ		 1244 invalidate = true;
1245
e4500b5e
YZ		 1246 while (!list_empty(&ci->i_cap_flush_list)) {
1247 cf = list_first_entry(&ci->i_cap_flush_list,
1248 struct ceph_cap_flush, i_list);
8cdcc07d 1249 list_move(&cf->i_list, &to_remove);
553adfd9
YZ		 1250 }
1251
6c99f254 1252 spin_lock(&mdsc->cap_dirty_lock);
8310b089 1253
e4500b5e
YZ		 1254 list_for_each_entry(cf, &to_remove, i_list)
1255 list_del(&cf->g_list);
8310b089 1256
6c99f254 1257 if (!list_empty(&ci->i_dirty_item)) {
3e0708b9
YZ		 1258 pr_warn_ratelimited(
1259 " dropping dirty %s state for %p %lld\n",
6c99f254
SW		 1260 ceph_cap_string(ci->i_dirty_caps),
1261 inode, ceph_ino(inode));
1262 ci->i_dirty_caps = 0;
1263 list_del_init(&ci->i_dirty_item);
6c93df5d 1264 drop = true;
6c99f254
SW		 1265 }
1266 if (!list_empty(&ci->i_flushing_item)) {
3e0708b9
YZ		 1267 pr_warn_ratelimited(
1268 " dropping dirty+flushing %s state for %p %lld\n",
6c99f254
SW		 1269 ceph_cap_string(ci->i_flushing_caps),
1270 inode, ceph_ino(inode));
1271 ci->i_flushing_caps = 0;
1272 list_del_init(&ci->i_flushing_item);
1273 mdsc->num_cap_flushing--;
6c93df5d 1274 drop = true;
6c99f254 1275 }
6c99f254 1276 spin_unlock(&mdsc->cap_dirty_lock);
553adfd9 1277
b3f8d68f
YZ		 1278 if (atomic_read(&ci->i_filelock_ref) > 0) {
1279 /* make further file lock syscall return -EIO */
1280 ci->i_ceph_flags |= CEPH_I_ERROR_FILELOCK;
1281 pr_warn_ratelimited(" dropping file locks for %p %lld\n",
1282 inode, ceph_ino(inode));
1283 }
1284
f66fd9f0 1285 if (!ci->i_dirty_caps && ci->i_prealloc_cap_flush) {
e4500b5e 1286 list_add(&ci->i_prealloc_cap_flush->i_list, &to_remove);
f66fd9f0
YZ		 1287 ci->i_prealloc_cap_flush = NULL;
1288 }
6c99f254 1289 }
be655596 1290 spin_unlock(&ci->i_ceph_lock);
553adfd9
YZ		 1291 while (!list_empty(&to_remove)) {
1292 struct ceph_cap_flush *cf;
1293 cf = list_first_entry(&to_remove,
e4500b5e
YZ		 1294 struct ceph_cap_flush, i_list);
1295 list_del(&cf->i_list);
f66fd9f0 1296 ceph_free_cap_flush(cf);
553adfd9 1297 }
77310320
YZ		 1298
1299 wake_up_all(&ci->i_cap_wq);
6c93df5d
YZ		 1300 if (invalidate)
1301 ceph_queue_invalidate(inode);
77310320 1302 if (drop)
6c99f254 1303 iput(inode);
2f2dc053
SW		 1304 return 0;
1305}
1306
1307/*
1308 * caller must hold session s_mutex
1309 */
1310static void remove_session_caps(struct ceph_mds_session *session)
1311{
6c93df5d
YZ		 1312 struct ceph_fs_client *fsc = session->s_mdsc->fsc;
1313 struct super_block *sb = fsc->sb;
c8a96a31
JL		 1314 LIST_HEAD(dispose);
1315
2f2dc053 1316 dout("remove_session_caps on %p\n", session);
6c93df5d 1317 iterate_session_caps(session, remove_session_caps_cb, fsc);
6f60f889 1318
c8799fc4
YZ		 1319 wake_up_all(&fsc->mdsc->cap_flushing_wq);
1320
6f60f889
YZ		 1321 spin_lock(&session->s_cap_lock);
1322 if (session->s_nr_caps > 0) {
6f60f889
YZ		 1323 struct inode *inode;
1324 struct ceph_cap *cap, *prev = NULL;
1325 struct ceph_vino vino;
1326 /*
1327 * iterate_session_caps() skips inodes that are being
1328 * deleted, we need to wait until deletions are complete.
1329 * __wait_on_freeing_inode() is designed for the job,
1330 * but it is not exported, so use lookup inode function
1331 * to access it.
1332 */
1333 while (!list_empty(&session->s_caps)) {
1334 cap = list_entry(session->s_caps.next,
1335 struct ceph_cap, session_caps);
1336 if (cap == prev)
1337 break;
1338 prev = cap;
1339 vino = cap->ci->i_vino;
1340 spin_unlock(&session->s_cap_lock);
1341
ed284c49 1342 inode = ceph_find_inode(sb, vino);
6f60f889
YZ		 1343 iput(inode);
1344
1345 spin_lock(&session->s_cap_lock);
1346 }
1347 }
745a8e3b
YZ		 1348
1349 // drop cap expires and unlock s_cap_lock
c8a96a31 1350 detach_cap_releases(session, &dispose);
6f60f889 1351
2f2dc053 1352 BUG_ON(session->s_nr_caps > 0);
6c99f254 1353 BUG_ON(!list_empty(&session->s_cap_flushing));
c8a96a31
JL		 1354 spin_unlock(&session->s_cap_lock);
1355 dispose_cap_releases(session->s_mdsc, &dispose);
2f2dc053
SW		 1356}
1357
d2f8bb27
YZ		 1358enum {
1359 RECONNECT,
1360 RENEWCAPS,
1361 FORCE_RO,
1362};
1363
2f2dc053
SW		 1364/*
1365 * wake up any threads waiting on this session's caps. if the cap is
1366 * old (didn't get renewed on the client reconnect), remove it now.
1367 *
1368 * caller must hold s_mutex.
1369 */
1370static int wake_up_session_cb(struct inode *inode, struct ceph_cap *cap,
1371 void *arg)
1372{
0dc2570f 1373 struct ceph_inode_info *ci = ceph_inode(inode);
d2f8bb27 1374 unsigned long ev = (unsigned long)arg;
0dc2570f 1375
d2f8bb27 1376 if (ev == RECONNECT) {
be655596 1377 spin_lock(&ci->i_ceph_lock);
0dc2570f
SW		 1378 ci->i_wanted_max_size = 0;
1379 ci->i_requested_max_size = 0;
be655596 1380 spin_unlock(&ci->i_ceph_lock);
d2f8bb27
YZ		 1381 } else if (ev == RENEWCAPS) {
1382 if (cap->cap_gen < cap->session->s_cap_gen) {
1383 /* mds did not re-issue stale cap */
1384 spin_lock(&ci->i_ceph_lock);
1385 cap->issued = cap->implemented = CEPH_CAP_PIN;
1386 /* make sure mds knows what we want */
1387 if (__ceph_caps_file_wanted(ci) & ~cap->mds_wanted)
1388 ci->i_ceph_flags |= CEPH_I_CAP_DROPPED;
1389 spin_unlock(&ci->i_ceph_lock);
1390 }
1391 } else if (ev == FORCE_RO) {
0dc2570f 1392 }
e5360309 1393 wake_up_all(&ci->i_cap_wq);
2f2dc053
SW		 1394 return 0;
1395}
1396
d2f8bb27 1397static void wake_up_session_caps(struct ceph_mds_session *session, int ev)
2f2dc053
SW		 1398{
1399 dout("wake_up_session_caps %p mds%d\n", session, session->s_mds);
0dc2570f 1400 iterate_session_caps(session, wake_up_session_cb,
d2f8bb27 1401 (void *)(unsigned long)ev);
2f2dc053
SW		 1402}
1403
1404/*
1405 * Send periodic message to MDS renewing all currently held caps. The
1406 * ack will reset the expiration for all caps from this session.
1407 *
1408 * caller holds s_mutex
1409 */
1410static int send_renew_caps(struct ceph_mds_client *mdsc,
1411 struct ceph_mds_session *session)
1412{
1413 struct ceph_msg *msg;
1414 int state;
1415
1416 if (time_after_eq(jiffies, session->s_cap_ttl) &&
1417 time_after_eq(session->s_cap_ttl, session->s_renew_requested))
1418 pr_info("mds%d caps stale\n", session->s_mds);
e4cb4cb8 1419 session->s_renew_requested = jiffies;
2f2dc053
SW		 1420
1421 /* do not try to renew caps until a recovering mds has reconnected
1422 * with its clients. */
1423 state = ceph_mdsmap_get_state(mdsc->mdsmap, session->s_mds);
1424 if (state < CEPH_MDS_STATE_RECONNECT) {
1425 dout("send_renew_caps ignoring mds%d (%s)\n",
1426 session->s_mds, ceph_mds_state_name(state));
1427 return 0;
1428 }
1429
1430 dout("send_renew_caps to mds%d (%s)\n", session->s_mds,
1431 ceph_mds_state_name(state));
2f2dc053
SW		 1432 msg = create_session_msg(CEPH_SESSION_REQUEST_RENEWCAPS,
1433 ++session->s_renew_seq);
a79832f2
SW		 1434 if (!msg)
1435 return -ENOMEM;
2f2dc053
SW		 1436 ceph_con_send(&session->s_con, msg);
1437 return 0;
1438}
1439
186e4f7a
YZ		 1440static int send_flushmsg_ack(struct ceph_mds_client *mdsc,
1441 struct ceph_mds_session *session, u64 seq)
1442{
1443 struct ceph_msg *msg;
1444
1445 dout("send_flushmsg_ack to mds%d (%s)s seq %lld\n",
a687ecaf 1446 session->s_mds, ceph_session_state_name(session->s_state), seq);
186e4f7a
YZ		 1447 msg = create_session_msg(CEPH_SESSION_FLUSHMSG_ACK, seq);
1448 if (!msg)
1449 return -ENOMEM;
1450 ceph_con_send(&session->s_con, msg);
1451 return 0;
1452}
1453
1454
2f2dc053
SW		 1455/*
1456 * Note new cap ttl, and any transition from stale -> not stale (fresh?).
0dc2570f
SW		 1457 *
1458 * Called under session->s_mutex
2f2dc053
SW		 1459 */
1460static void renewed_caps(struct ceph_mds_client *mdsc,
1461 struct ceph_mds_session *session, int is_renew)
1462{
1463 int was_stale;
1464 int wake = 0;
1465
1466 spin_lock(&session->s_cap_lock);
1ce208a6 1467 was_stale = is_renew && time_after_eq(jiffies, session->s_cap_ttl);
2f2dc053
SW		 1468
1469 session->s_cap_ttl = session->s_renew_requested +
1470 mdsc->mdsmap->m_session_timeout*HZ;
1471
1472 if (was_stale) {
1473 if (time_before(jiffies, session->s_cap_ttl)) {
1474 pr_info("mds%d caps renewed\n", session->s_mds);
1475 wake = 1;
1476 } else {
1477 pr_info("mds%d caps still stale\n", session->s_mds);
1478 }
1479 }
1480 dout("renewed_caps mds%d ttl now %lu, was %s, now %s\n",
1481 session->s_mds, session->s_cap_ttl, was_stale ? "stale" : "fresh",
1482 time_before(jiffies, session->s_cap_ttl) ? "stale" : "fresh");
1483 spin_unlock(&session->s_cap_lock);
1484
1485 if (wake)
d2f8bb27 1486 wake_up_session_caps(session, RENEWCAPS);
2f2dc053
SW		 1487}
1488
1489/*
1490 * send a session close request
1491 */
1492static int request_close_session(struct ceph_mds_client *mdsc,
1493 struct ceph_mds_session *session)
1494{
1495 struct ceph_msg *msg;
2f2dc053
SW		 1496
1497 dout("request_close_session mds%d state %s seq %lld\n",
a687ecaf 1498 session->s_mds, ceph_session_state_name(session->s_state),
2f2dc053
SW		 1499 session->s_seq);
1500 msg = create_session_msg(CEPH_SESSION_REQUEST_CLOSE, session->s_seq);
a79832f2
SW		 1501 if (!msg)
1502 return -ENOMEM;
1503 ceph_con_send(&session->s_con, msg);
fcff415c 1504 return 1;
2f2dc053
SW		 1505}
1506
1507/*
1508 * Called with s_mutex held.
1509 */
1510static int __close_session(struct ceph_mds_client *mdsc,
1511 struct ceph_mds_session *session)
1512{
1513 if (session->s_state >= CEPH_MDS_SESSION_CLOSING)
1514 return 0;
1515 session->s_state = CEPH_MDS_SESSION_CLOSING;
1516 return request_close_session(mdsc, session);
1517}
1518
040d7860
YZ		 1519static bool drop_negative_children(struct dentry *dentry)
1520{
1521 struct dentry *child;
1522 bool all_negative = true;
1523
1524 if (!d_is_dir(dentry))
1525 goto out;
1526
1527 spin_lock(&dentry->d_lock);
1528 list_for_each_entry(child, &dentry->d_subdirs, d_child) {
1529 if (d_really_is_positive(child)) {
1530 all_negative = false;
1531 break;
1532 }
1533 }
1534 spin_unlock(&dentry->d_lock);
1535
1536 if (all_negative)
1537 shrink_dcache_parent(dentry);
1538out:
1539 return all_negative;
1540}
1541
2f2dc053
SW		 1542/*
1543 * Trim old(er) caps.
1544 *
1545 * Because we can't cache an inode without one or more caps, we do
1546 * this indirectly: if a cap is unused, we prune its aliases, at which
1547 * point the inode will hopefully get dropped to.
1548 *
1549 * Yes, this is a bit sloppy. Our only real goal here is to respond to
1550 * memory pressure from the MDS, though, so it needn't be perfect.
1551 */
1552static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)
1553{
1554 struct ceph_mds_session *session = arg;
1555 struct ceph_inode_info *ci = ceph_inode(inode);
979abfdd 1556 int used, wanted, oissued, mine;
2f2dc053
SW		 1557
1558 if (session->s_trim_caps <= 0)
1559 return -1;
1560
be655596 1561 spin_lock(&ci->i_ceph_lock);
2f2dc053
SW		 1562 mine = cap->issued | cap->implemented;
1563 used = __ceph_caps_used(ci);
979abfdd 1564 wanted = __ceph_caps_file_wanted(ci);
2f2dc053
SW		 1565 oissued = __ceph_caps_issued_other(ci, cap);
1566
979abfdd 1567 dout("trim_caps_cb %p cap %p mine %s oissued %s used %s wanted %s\n",
2f2dc053 1568 inode, cap, ceph_cap_string(mine), ceph_cap_string(oissued),
979abfdd
YZ		 1569 ceph_cap_string(used), ceph_cap_string(wanted));
1570 if (cap == ci->i_auth_cap) {
622f3e25
YZ		 1571 if (ci->i_dirty_caps || ci->i_flushing_caps ||
1572 !list_empty(&ci->i_cap_snaps))
979abfdd
YZ		 1573 goto out;
1574 if ((used | wanted) & CEPH_CAP_ANY_WR)
1575 goto out;
89aa5930
YZ		 1576 /* Note: it's possible that i_filelock_ref becomes non-zero
1577 * after dropping auth caps. It doesn't hurt because reply
1578 * of lock mds request will re-add auth caps. */
1579 if (atomic_read(&ci->i_filelock_ref) > 0)
1580 goto out;
979abfdd 1581 }
5e804ac4
YZ		 1582 /* The inode has cached pages, but it's no longer used.
1583 * we can safely drop it */
1584 if (wanted == 0 && used == CEPH_CAP_FILE_CACHE &&
1585 !(oissued & CEPH_CAP_FILE_CACHE)) {
1586 used = 0;
1587 oissued = 0;
1588 }
979abfdd 1589 if ((used | wanted) & ~oissued & mine)
2f2dc053
SW		 1590 goto out; /* we need these caps */
1591
2f2dc053
SW		 1592 if (oissued) {
1593 /* we aren't the only cap.. just remove us */
a096b09a 1594 __ceph_remove_cap(cap, true);
040d7860 1595 session->s_trim_caps--;
2f2dc053 1596 } else {
040d7860 1597 struct dentry *dentry;
5e804ac4 1598 /* try dropping referring dentries */
be655596 1599 spin_unlock(&ci->i_ceph_lock);
040d7860
YZ		 1600 dentry = d_find_any_alias(inode);
1601 if (dentry && drop_negative_children(dentry)) {
1602 int count;
1603 dput(dentry);
1604 d_prune_aliases(inode);
1605 count = atomic_read(&inode->i_count);
1606 if (count == 1)
1607 session->s_trim_caps--;
1608 dout("trim_caps_cb %p cap %p pruned, count now %d\n",
1609 inode, cap, count);
1610 } else {
1611 dput(dentry);
1612 }
2f2dc053
SW		 1613 return 0;
1614 }
1615
1616out:
be655596 1617 spin_unlock(&ci->i_ceph_lock);
2f2dc053
SW		 1618 return 0;
1619}
1620
1621/*
1622 * Trim session cap count down to some max number.
1623 */
e30ee581
ZZ		 1624int ceph_trim_caps(struct ceph_mds_client *mdsc,
1625 struct ceph_mds_session *session,
1626 int max_caps)
2f2dc053
SW		 1627{
1628 int trim_caps = session->s_nr_caps - max_caps;
1629
1630 dout("trim_caps mds%d start: %d / %d, trim %d\n",
1631 session->s_mds, session->s_nr_caps, max_caps, trim_caps);
1632 if (trim_caps > 0) {
1633 session->s_trim_caps = trim_caps;
1634 iterate_session_caps(session, trim_caps_cb, session);
1635 dout("trim_caps mds%d done: %d / %d, trimmed %d\n",
1636 session->s_mds, session->s_nr_caps, max_caps,
1637 trim_caps - session->s_trim_caps);
5dacf091 1638 session->s_trim_caps = 0;
2f2dc053 1639 }
a56371d9 1640
a56371d9 1641 ceph_send_cap_releases(mdsc, session);
2f2dc053
SW		 1642 return 0;
1643}
1644
8310b089
YZ		 1645static int check_caps_flush(struct ceph_mds_client *mdsc,
1646 u64 want_flush_tid)
1647{
8310b089
YZ		 1648 int ret = 1;
1649
1650 spin_lock(&mdsc->cap_dirty_lock);
e4500b5e
YZ		 1651 if (!list_empty(&mdsc->cap_flush_list)) {
1652 struct ceph_cap_flush *cf =
1653 list_first_entry(&mdsc->cap_flush_list,
1654 struct ceph_cap_flush, g_list);
1655 if (cf->tid <= want_flush_tid) {
1656 dout("check_caps_flush still flushing tid "
1657 "%llu <= %llu\n", cf->tid, want_flush_tid);
1658 ret = 0;
1659 }
8310b089
YZ		 1660 }
1661 spin_unlock(&mdsc->cap_dirty_lock);
1662 return ret;
d3383a8e
YZ		 1663}
1664
2f2dc053
SW		 1665/*
1666 * flush all dirty inode data to disk.
1667 *
8310b089 1668 * returns true if we've flushed through want_flush_tid
2f2dc053 1669 */
affbc19a 1670static void wait_caps_flush(struct ceph_mds_client *mdsc,
0e294387 1671 u64 want_flush_tid)
2f2dc053 1672{
0e294387 1673 dout("check_caps_flush want %llu\n", want_flush_tid);
8310b089
YZ		 1674
1675 wait_event(mdsc->cap_flushing_wq,
1676 check_caps_flush(mdsc, want_flush_tid));
1677
1678 dout("check_caps_flush ok, flushed thru %llu\n", want_flush_tid);
2f2dc053
SW		 1679}
1680
1681/*
1682 * called under s_mutex
1683 */
3d7ded4d
SW		 1684void ceph_send_cap_releases(struct ceph_mds_client *mdsc,
1685 struct ceph_mds_session *session)
2f2dc053 1686{
745a8e3b
YZ		 1687 struct ceph_msg *msg = NULL;
1688 struct ceph_mds_cap_release *head;
1689 struct ceph_mds_cap_item *item;
92475f05 1690 struct ceph_osd_client *osdc = &mdsc->fsc->client->osdc;
745a8e3b
YZ		 1691 struct ceph_cap *cap;
1692 LIST_HEAD(tmp_list);
1693 int num_cap_releases;
92475f05
JL		 1694 __le32 barrier, *cap_barrier;
1695
1696 down_read(&osdc->lock);
1697 barrier = cpu_to_le32(osdc->epoch_barrier);
1698 up_read(&osdc->lock);
2f2dc053 1699
0f8605f2 1700 spin_lock(&session->s_cap_lock);
745a8e3b
YZ		 1701again:
1702 list_splice_init(&session->s_cap_releases, &tmp_list);
1703 num_cap_releases = session->s_num_cap_releases;
1704 session->s_num_cap_releases = 0;
2f2dc053 1705 spin_unlock(&session->s_cap_lock);
e01a5946 1706
745a8e3b
YZ		 1707 while (!list_empty(&tmp_list)) {
1708 if (!msg) {
1709 msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPRELEASE,
09cbfeaf 1710 PAGE_SIZE, GFP_NOFS, false);
745a8e3b
YZ		 1711 if (!msg)
1712 goto out_err;
1713 head = msg->front.iov_base;
1714 head->num = cpu_to_le32(0);
1715 msg->front.iov_len = sizeof(*head);
92475f05
JL		 1716
1717 msg->hdr.version = cpu_to_le16(2);
1718 msg->hdr.compat_version = cpu_to_le16(1);
745a8e3b 1719 }
92475f05 1720
745a8e3b
YZ		 1721 cap = list_first_entry(&tmp_list, struct ceph_cap,
1722 session_caps);
1723 list_del(&cap->session_caps);
1724 num_cap_releases--;
e01a5946 1725
00bd8edb 1726 head = msg->front.iov_base;
745a8e3b
YZ		 1727 le32_add_cpu(&head->num, 1);
1728 item = msg->front.iov_base + msg->front.iov_len;
1729 item->ino = cpu_to_le64(cap->cap_ino);
1730 item->cap_id = cpu_to_le64(cap->cap_id);
1731 item->migrate_seq = cpu_to_le32(cap->mseq);
1732 item->seq = cpu_to_le32(cap->issue_seq);
1733 msg->front.iov_len += sizeof(*item);
1734
1735 ceph_put_cap(mdsc, cap);
1736
1737 if (le32_to_cpu(head->num) == CEPH_CAPS_PER_RELEASE) {
92475f05
JL		 1738 // Append cap_barrier field
1739 cap_barrier = msg->front.iov_base + msg->front.iov_len;
1740 *cap_barrier = barrier;
1741 msg->front.iov_len += sizeof(*cap_barrier);
1742
745a8e3b
YZ		 1743 msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
1744 dout("send_cap_releases mds%d %p\n", session->s_mds, msg);
1745 ceph_con_send(&session->s_con, msg);
1746 msg = NULL;
1747 }
00bd8edb 1748 }
e01a5946 1749
745a8e3b 1750 BUG_ON(num_cap_releases != 0);
e01a5946 1751
745a8e3b
YZ		 1752 spin_lock(&session->s_cap_lock);
1753 if (!list_empty(&session->s_cap_releases))
1754 goto again;
1755 spin_unlock(&session->s_cap_lock);
1756
1757 if (msg) {
92475f05
JL		 1758 // Append cap_barrier field
1759 cap_barrier = msg->front.iov_base + msg->front.iov_len;
1760 *cap_barrier = barrier;
1761 msg->front.iov_len += sizeof(*cap_barrier);
1762
745a8e3b
YZ		 1763 msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
1764 dout("send_cap_releases mds%d %p\n", session->s_mds, msg);
1765 ceph_con_send(&session->s_con, msg);
e01a5946 1766 }
745a8e3b
YZ		 1767 return;
1768out_err:
1769 pr_err("send_cap_releases mds%d, failed to allocate message\n",
1770 session->s_mds);
1771 spin_lock(&session->s_cap_lock);
1772 list_splice(&tmp_list, &session->s_cap_releases);
1773 session->s_num_cap_releases += num_cap_releases;
1774 spin_unlock(&session->s_cap_lock);
e01a5946
SW		 1775}
1776
2f2dc053
SW		 1777/*
1778 * requests
1779 */
1780
54008399
YZ		 1781int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
1782 struct inode *dir)
1783{
1784 struct ceph_inode_info *ci = ceph_inode(dir);
1785 struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
1786 struct ceph_mount_options *opt = req->r_mdsc->fsc->mount_options;
2a5beea3 1787 size_t size = sizeof(struct ceph_mds_reply_dir_entry);
54008399
YZ		 1788 int order, num_entries;
1789
1790 spin_lock(&ci->i_ceph_lock);
1791 num_entries = ci->i_files + ci->i_subdirs;
1792 spin_unlock(&ci->i_ceph_lock);
1793 num_entries = max(num_entries, 1);
1794 num_entries = min(num_entries, opt->max_readdir);
1795
1796 order = get_order(size * num_entries);
1797 while (order >= 0) {
2a5beea3
YZ		 1798 rinfo->dir_entries = (void*)__get_free_pages(GFP_KERNEL |
1799 __GFP_NOWARN,
1800 order);
1801 if (rinfo->dir_entries)
54008399
YZ		 1802 break;
1803 order--;
1804 }
2a5beea3 1805 if (!rinfo->dir_entries)
54008399
YZ		 1806 return -ENOMEM;
1807
1808 num_entries = (PAGE_SIZE << order) / size;
1809 num_entries = min(num_entries, opt->max_readdir);
1810
1811 rinfo->dir_buf_size = PAGE_SIZE << order;
1812 req->r_num_caps = num_entries + 1;
1813 req->r_args.readdir.max_entries = cpu_to_le32(num_entries);
1814 req->r_args.readdir.max_bytes = cpu_to_le32(opt->max_readdir_bytes);
1815 return 0;
1816}
1817
2f2dc053
SW		 1818/*
1819 * Create an mds request.
1820 */
1821struct ceph_mds_request *
1822ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode)
1823{
1824 struct ceph_mds_request *req = kzalloc(sizeof(*req), GFP_NOFS);
63ecae7e 1825 struct timespec64 ts;
2f2dc053
SW		 1826
1827 if (!req)
1828 return ERR_PTR(-ENOMEM);
1829
b4556396 1830 mutex_init(&req->r_fill_mutex);
37151668 1831 req->r_mdsc = mdsc;
2f2dc053
SW		 1832 req->r_started = jiffies;
1833 req->r_resend_mds = -1;
1834 INIT_LIST_HEAD(&req->r_unsafe_dir_item);
68cd5b4b 1835 INIT_LIST_HEAD(&req->r_unsafe_target_item);
2f2dc053 1836 req->r_fmode = -1;
153c8e6b 1837 kref_init(&req->r_kref);
fcd00b68 1838 RB_CLEAR_NODE(&req->r_node);
2f2dc053
SW		 1839 INIT_LIST_HEAD(&req->r_wait);
1840 init_completion(&req->r_completion);
1841 init_completion(&req->r_safe_completion);
1842 INIT_LIST_HEAD(&req->r_unsafe_item);
1843
63ecae7e 1844 ktime_get_coarse_real_ts64(&ts);
0ed1e90a 1845 req->r_stamp = timespec64_trunc(ts, mdsc->fsc->sb->s_time_gran);
b8e69066 1846
2f2dc053
SW		 1847 req->r_op = op;
1848 req->r_direct_mode = mode;
1849 return req;
1850}
1851
1852/*
44ca18f2 1853 * return oldest (lowest) request, tid in request tree, 0 if none.
2f2dc053
SW		 1854 *
1855 * called under mdsc->mutex.
1856 */
44ca18f2
SW		 1857static struct ceph_mds_request *__get_oldest_req(struct ceph_mds_client *mdsc)
1858{
1859 if (RB_EMPTY_ROOT(&mdsc->request_tree))
1860 return NULL;
1861 return rb_entry(rb_first(&mdsc->request_tree),
1862 struct ceph_mds_request, r_node);
1863}
1864
e8a7b8b1 1865static inline u64 __get_oldest_tid(struct ceph_mds_client *mdsc)
2f2dc053 1866{
e8a7b8b1 1867 return mdsc->oldest_tid;
2f2dc053
SW		 1868}
1869
1870/*
1871 * Build a dentry's path. Allocate on heap; caller must kfree. Based
1872 * on build_path_from_dentry in fs/cifs/dir.c.
1873 *
1874 * If @stop_on_nosnap, generate path relative to the first non-snapped
1875 * inode.
1876 *
1877 * Encode hidden .snap dirs as a double /, i.e.
1878 * foo/.snap/bar -> foo//bar
1879 */
1880char *ceph_mdsc_build_path(struct dentry *dentry, int *plen, u64 *base,
1881 int stop_on_nosnap)
1882{
1883 struct dentry *temp;
1884 char *path;
1885 int len, pos;
1b71fe2e 1886 unsigned seq;
2f2dc053 1887
d37b1d99 1888 if (!dentry)
2f2dc053
SW		 1889 return ERR_PTR(-EINVAL);
1890
1891retry:
1892 len = 0;
1b71fe2e
AV		 1893 seq = read_seqbegin(&rename_lock);
1894 rcu_read_lock();
2f2dc053 1895 for (temp = dentry; !IS_ROOT(temp);) {
2b0143b5 1896 struct inode *inode = d_inode(temp);
2f2dc053
SW		 1897 if (inode && ceph_snap(inode) == CEPH_SNAPDIR)
1898 len++; /* slash only */
1899 else if (stop_on_nosnap && inode &&
1900 ceph_snap(inode) == CEPH_NOSNAP)
1901 break;
1902 else
1903 len += 1 + temp->d_name.len;
1904 temp = temp->d_parent;
2f2dc053 1905 }
1b71fe2e 1906 rcu_read_unlock();
2f2dc053
SW		 1907 if (len)
1908 len--; /* no leading '/' */
1909
1910 path = kmalloc(len+1, GFP_NOFS);
d37b1d99 1911 if (!path)
2f2dc053
SW		 1912 return ERR_PTR(-ENOMEM);
1913 pos = len;
1914 path[pos] = 0; /* trailing null */
1b71fe2e 1915 rcu_read_lock();
2f2dc053 1916 for (temp = dentry; !IS_ROOT(temp) && pos != 0; ) {
1b71fe2e 1917 struct inode *inode;
2f2dc053 1918
1b71fe2e 1919 spin_lock(&temp->d_lock);
2b0143b5 1920 inode = d_inode(temp);
2f2dc053 1921 if (inode && ceph_snap(inode) == CEPH_SNAPDIR) {
104648ad 1922 dout("build_path path+%d: %p SNAPDIR\n",
2f2dc053
SW		 1923 pos, temp);
1924 } else if (stop_on_nosnap && inode &&
1925 ceph_snap(inode) == CEPH_NOSNAP) {
9d5a09e6 1926 spin_unlock(&temp->d_lock);
2f2dc053
SW		 1927 break;
1928 } else {
1929 pos -= temp->d_name.len;
1b71fe2e
AV		 1930 if (pos < 0) {
1931 spin_unlock(&temp->d_lock);
2f2dc053 1932 break;
1b71fe2e 1933 }
2f2dc053
SW		 1934 strncpy(path + pos, temp->d_name.name,
1935 temp->d_name.len);
2f2dc053 1936 }
1b71fe2e 1937 spin_unlock(&temp->d_lock);
2f2dc053
SW		 1938 if (pos)
1939 path[--pos] = '/';
1940 temp = temp->d_parent;
2f2dc053 1941 }
1b71fe2e
AV		 1942 rcu_read_unlock();
1943 if (pos != 0 || read_seqretry(&rename_lock, seq)) {
104648ad 1944 pr_err("build_path did not end path lookup where "
2f2dc053
SW		 1945 "expected, namelen is %d, pos is %d\n", len, pos);
1946 /* presumably this is only possible if racing with a
1947 rename of one of the parent directories (we can not
1948 lock the dentries above us to prevent this, but
1949 retrying should be harmless) */
1950 kfree(path);
1951 goto retry;
1952 }
1953
2b0143b5 1954 *base = ceph_ino(d_inode(temp));
2f2dc053 1955 *plen = len;
104648ad 1956 dout("build_path on %p %d built %llx '%.*s'\n",
84d08fa8 1957 dentry, d_count(dentry), *base, len, path);
2f2dc053
SW		 1958 return path;
1959}
1960
fd36a717 1961static int build_dentry_path(struct dentry *dentry, struct inode *dir,
2f2dc053
SW		 1962 const char **ppath, int *ppathlen, u64 *pino,
1963 int *pfreepath)
1964{
1965 char *path;
1966
c6b0b656 1967 rcu_read_lock();
fd36a717
JL		 1968 if (!dir)
1969 dir = d_inode_rcu(dentry->d_parent);
c6b0b656
JL		 1970 if (dir && ceph_snap(dir) == CEPH_NOSNAP) {
1971 *pino = ceph_ino(dir);
1972 rcu_read_unlock();
2f2dc053
SW		 1973 *ppath = dentry->d_name.name;
1974 *ppathlen = dentry->d_name.len;
1975 return 0;
1976 }
c6b0b656 1977 rcu_read_unlock();
2f2dc053
SW		 1978 path = ceph_mdsc_build_path(dentry, ppathlen, pino, 1);
1979 if (IS_ERR(path))
1980 return PTR_ERR(path);
1981 *ppath = path;
1982 *pfreepath = 1;
1983 return 0;
1984}
1985
1986static int build_inode_path(struct inode *inode,
1987 const char **ppath, int *ppathlen, u64 *pino,
1988 int *pfreepath)
1989{
1990 struct dentry *dentry;
1991 char *path;
1992
1993 if (ceph_snap(inode) == CEPH_NOSNAP) {
1994 *pino = ceph_ino(inode);
1995 *ppathlen = 0;
1996 return 0;
1997 }
1998 dentry = d_find_alias(inode);
1999 path = ceph_mdsc_build_path(dentry, ppathlen, pino, 1);
2000 dput(dentry);
2001 if (IS_ERR(path))
2002 return PTR_ERR(path);
2003 *ppath = path;
2004 *pfreepath = 1;
2005 return 0;
2006}
2007
2008/*
2009 * request arguments may be specified via an inode *, a dentry *, or
2010 * an explicit ino+path.
2011 */
2012static int set_request_path_attr(struct inode *rinode, struct dentry *rdentry,
fd36a717
JL		 2013 struct inode *rdiri, const char *rpath,
2014 u64 rino, const char **ppath, int *pathlen,
2f2dc053
SW		 2015 u64 *ino, int *freepath)
2016{
2017 int r = 0;
2018
2019 if (rinode) {
2020 r = build_inode_path(rinode, ppath, pathlen, ino, freepath);
2021 dout(" inode %p %llx.%llx\n", rinode, ceph_ino(rinode),
2022 ceph_snap(rinode));
2023 } else if (rdentry) {
fd36a717
JL		 2024 r = build_dentry_path(rdentry, rdiri, ppath, pathlen, ino,
2025 freepath);
2f2dc053
SW		 2026 dout(" dentry %p %llx/%.*s\n", rdentry, *ino, *pathlen,
2027 *ppath);
795858db 2028 } else if (rpath || rino) {
2f2dc053
SW		 2029 *ino = rino;
2030 *ppath = rpath;
b000056a 2031 *pathlen = rpath ? strlen(rpath) : 0;
2f2dc053
SW		 2032 dout(" path %.*s\n", *pathlen, rpath);
2033 }
2034
2035 return r;
2036}
2037
2038/*
2039 * called under mdsc->mutex
2040 */
2041static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc,
2042 struct ceph_mds_request *req,
6e6f0923 2043 int mds, bool drop_cap_releases)
2f2dc053
SW		 2044{
2045 struct ceph_msg *msg;
2046 struct ceph_mds_request_head *head;
2047 const char *path1 = NULL;
2048 const char *path2 = NULL;
2049 u64 ino1 = 0, ino2 = 0;
2050 int pathlen1 = 0, pathlen2 = 0;
2051 int freepath1 = 0, freepath2 = 0;
2052 int len;
2053 u16 releases;
2054 void *p, *end;
2055 int ret;
2056
2057 ret = set_request_path_attr(req->r_inode, req->r_dentry,
3dd69aab 2058 req->r_parent, req->r_path1, req->r_ino1.ino,
2f2dc053
SW		 2059 &path1, &pathlen1, &ino1, &freepath1);
2060 if (ret < 0) {
2061 msg = ERR_PTR(ret);
2062 goto out;
2063 }
2064
2065 ret = set_request_path_attr(NULL, req->r_old_dentry,
fd36a717 2066 req->r_old_dentry_dir,
2f2dc053
SW		 2067 req->r_path2, req->r_ino2.ino,
2068 &path2, &pathlen2, &ino2, &freepath2);
2069 if (ret < 0) {
2070 msg = ERR_PTR(ret);
2071 goto out_free1;
2072 }
2073
2074 len = sizeof(*head) +
b8e69066 2075 pathlen1 + pathlen2 + 2*(1 + sizeof(u32) + sizeof(u64)) +
777d738a 2076 sizeof(struct ceph_timespec);
2f2dc053
SW		 2077
2078 /* calculate (max) length for cap releases */
2079 len += sizeof(struct ceph_mds_request_release) *
2080 (!!req->r_inode_drop + !!req->r_dentry_drop +
2081 !!req->r_old_inode_drop + !!req->r_old_dentry_drop);
2082 if (req->r_dentry_drop)
2083 len += req->r_dentry->d_name.len;
2084 if (req->r_old_dentry_drop)
2085 len += req->r_old_dentry->d_name.len;
2086
0d9c1ab3 2087 msg = ceph_msg_new2(CEPH_MSG_CLIENT_REQUEST, len, 1, GFP_NOFS, false);
a79832f2
SW		 2088 if (!msg) {
2089 msg = ERR_PTR(-ENOMEM);
2f2dc053 2090 goto out_free2;
a79832f2 2091 }
2f2dc053 2092
7cfa0313 2093 msg->hdr.version = cpu_to_le16(2);
6df058c0
SW		 2094 msg->hdr.tid = cpu_to_le64(req->r_tid);
2095
2f2dc053
SW		 2096 head = msg->front.iov_base;
2097 p = msg->front.iov_base + sizeof(*head);
2098 end = msg->front.iov_base + msg->front.iov_len;
2099
2100 head->mdsmap_epoch = cpu_to_le32(mdsc->mdsmap->m_epoch);
2101 head->op = cpu_to_le32(req->r_op);
ff3d0046
EB		 2102 head->caller_uid = cpu_to_le32(from_kuid(&init_user_ns, req->r_uid));
2103 head->caller_gid = cpu_to_le32(from_kgid(&init_user_ns, req->r_gid));
2f2dc053
SW		 2104 head->args = req->r_args;
2105
2106 ceph_encode_filepath(&p, end, ino1, path1);
2107 ceph_encode_filepath(&p, end, ino2, path2);
2108
e979cf50
SW		 2109 /* make note of release offset, in case we need to replay */
2110 req->r_request_release_offset = p - msg->front.iov_base;
2111
2f2dc053
SW		 2112 /* cap releases */
2113 releases = 0;
2114 if (req->r_inode_drop)
2115 releases += ceph_encode_inode_release(&p,
2b0143b5 2116 req->r_inode ? req->r_inode : d_inode(req->r_dentry),
2f2dc053
SW		 2117 mds, req->r_inode_drop, req->r_inode_unless, 0);
2118 if (req->r_dentry_drop)
2119 releases += ceph_encode_dentry_release(&p, req->r_dentry,
3dd69aab 2120 req->r_parent, mds, req->r_dentry_drop,
ca6c8ae0 2121 req->r_dentry_unless);
2f2dc053
SW		 2122 if (req->r_old_dentry_drop)
2123 releases += ceph_encode_dentry_release(&p, req->r_old_dentry,
ca6c8ae0
JL		 2124 req->r_old_dentry_dir, mds,
2125 req->r_old_dentry_drop,
2126 req->r_old_dentry_unless);
2f2dc053
SW		 2127 if (req->r_old_inode_drop)
2128 releases += ceph_encode_inode_release(&p,
2b0143b5 2129 d_inode(req->r_old_dentry),
2f2dc053 2130 mds, req->r_old_inode_drop, req->r_old_inode_unless, 0);
6e6f0923
YZ		 2131
2132 if (drop_cap_releases) {
2133 releases = 0;
2134 p = msg->front.iov_base + req->r_request_release_offset;
2135 }
2136
2f2dc053
SW		 2137 head->num_releases = cpu_to_le16(releases);
2138
b8e69066 2139 /* time stamp */
1f041a89
YZ		 2140 {
2141 struct ceph_timespec ts;
0ed1e90a 2142 ceph_encode_timespec64(&ts, &req->r_stamp);
1f041a89
YZ		 2143 ceph_encode_copy(&p, &ts, sizeof(ts));
2144 }
b8e69066 2145
2f2dc053
SW		 2146 BUG_ON(p > end);
2147 msg->front.iov_len = p - msg->front.iov_base;
2148 msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
2149
25e6bae3
YZ		 2150 if (req->r_pagelist) {
2151 struct ceph_pagelist *pagelist = req->r_pagelist;
25e6bae3
YZ		 2152 ceph_msg_data_add_pagelist(msg, pagelist);
2153 msg->hdr.data_len = cpu_to_le32(pagelist->length);
2154 } else {
2155 msg->hdr.data_len = 0;
ebf18f47 2156 }
02afca6c 2157
2f2dc053
SW		 2158 msg->hdr.data_off = cpu_to_le16(0);
2159
2160out_free2:
2161 if (freepath2)
2162 kfree((char *)path2);
2163out_free1:
2164 if (freepath1)
2165 kfree((char *)path1);
2166out:
2167 return msg;
2168}
2169
2170/*
2171 * called under mdsc->mutex if error, under no mutex if
2172 * success.
2173 */
2174static void complete_request(struct ceph_mds_client *mdsc,
2175 struct ceph_mds_request *req)
2176{
2177 if (req->r_callback)
2178 req->r_callback(mdsc, req);
2179 else
03066f23 2180 complete_all(&req->r_completion);
2f2dc053
SW		 2181}
2182
2183/*
2184 * called under mdsc->mutex
2185 */
2186static int __prepare_send_request(struct ceph_mds_client *mdsc,
2187 struct ceph_mds_request *req,
6e6f0923 2188 int mds, bool drop_cap_releases)
2f2dc053
SW		 2189{
2190 struct ceph_mds_request_head *rhead;
2191 struct ceph_msg *msg;
2192 int flags = 0;
2193
2f2dc053 2194 req->r_attempts++;
e55b71f8
GF		 2195 if (req->r_inode) {
2196 struct ceph_cap *cap =
2197 ceph_get_cap_for_mds(ceph_inode(req->r_inode), mds);
2198
2199 if (cap)
2200 req->r_sent_on_mseq = cap->mseq;
2201 else
2202 req->r_sent_on_mseq = -1;
2203 }
2f2dc053
SW		 2204 dout("prepare_send_request %p tid %lld %s (attempt %d)\n", req,
2205 req->r_tid, ceph_mds_op_name(req->r_op), req->r_attempts);
2206
bc2de10d 2207 if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
c5c9a0bf 2208 void *p;
01a92f17
SW		 2209 /*
2210 * Replay. Do not regenerate message (and rebuild
2211 * paths, etc.); just use the original message.
2212 * Rebuilding paths will break for renames because
2213 * d_move mangles the src name.
2214 */
2215 msg = req->r_request;
2216 rhead = msg->front.iov_base;
2217
2218 flags = le32_to_cpu(rhead->flags);
2219 flags |= CEPH_MDS_FLAG_REPLAY;
2220 rhead->flags = cpu_to_le32(flags);
2221
2222 if (req->r_target_inode)
2223 rhead->ino = cpu_to_le64(ceph_ino(req->r_target_inode));
2224
2225 rhead->num_retry = req->r_attempts - 1;
e979cf50
SW		 2226
2227 /* remove cap/dentry releases from message */
2228 rhead->num_releases = 0;
c5c9a0bf
YZ		 2229
2230 /* time stamp */
2231 p = msg->front.iov_base + req->r_request_release_offset;
1f041a89
YZ		 2232 {
2233 struct ceph_timespec ts;
0ed1e90a 2234 ceph_encode_timespec64(&ts, &req->r_stamp);
1f041a89
YZ		 2235 ceph_encode_copy(&p, &ts, sizeof(ts));
2236 }
c5c9a0bf
YZ		 2237
2238 msg->front.iov_len = p - msg->front.iov_base;
2239 msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
01a92f17
SW		 2240 return 0;
2241 }
2242
2f2dc053
SW		 2243 if (req->r_request) {
2244 ceph_msg_put(req->r_request);
2245 req->r_request = NULL;
2246 }
6e6f0923 2247 msg = create_request_message(mdsc, req, mds, drop_cap_releases);
2f2dc053 2248 if (IS_ERR(msg)) {
e1518c7c 2249 req->r_err = PTR_ERR(msg);
a79832f2 2250 return PTR_ERR(msg);
2f2dc053
SW		 2251 }
2252 req->r_request = msg;
2253
2254 rhead = msg->front.iov_base;
2f2dc053 2255 rhead->oldest_client_tid = cpu_to_le64(__get_oldest_tid(mdsc));
bc2de10d 2256 if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
2f2dc053 2257 flags |= CEPH_MDS_FLAG_REPLAY;
3dd69aab 2258 if (req->r_parent)
2f2dc053
SW		 2259 flags |= CEPH_MDS_FLAG_WANT_DENTRY;
2260 rhead->flags = cpu_to_le32(flags);
2261 rhead->num_fwd = req->r_num_fwd;
2262 rhead->num_retry = req->r_attempts - 1;
01a92f17 2263 rhead->ino = 0;
2f2dc053 2264
3dd69aab 2265 dout(" r_parent = %p\n", req->r_parent);
2f2dc053
SW		 2266 return 0;
2267}
2268
2269/*
2270 * send request, or put it on the appropriate wait list.
2271 */
d5548492 2272static void __do_request(struct ceph_mds_client *mdsc,
2f2dc053
SW		 2273 struct ceph_mds_request *req)
2274{
2275 struct ceph_mds_session *session = NULL;
2276 int mds = -1;
48fec5d0 2277 int err = 0;
2f2dc053 2278
bc2de10d
JL		 2279 if (req->r_err || test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
2280 if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags))
eb1b8af3 2281 __unregister_request(mdsc, req);
d5548492 2282 return;
eb1b8af3 2283 }
2f2dc053
SW		 2284
2285 if (req->r_timeout &&
2286 time_after_eq(jiffies, req->r_started + req->r_timeout)) {
2287 dout("do_request timed out\n");
2288 err = -EIO;
2289 goto finish;
2290 }
52953d55 2291 if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
48fec5d0
YZ		 2292 dout("do_request forced umount\n");
2293 err = -EIO;
2294 goto finish;
2295 }
52953d55 2296 if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_MOUNTING) {
e9e427f0
YZ		 2297 if (mdsc->mdsmap_err) {
2298 err = mdsc->mdsmap_err;
2299 dout("do_request mdsmap err %d\n", err);
2300 goto finish;
2301 }
cc8e8342
YZ		 2302 if (mdsc->mdsmap->m_epoch == 0) {
2303 dout("do_request no mdsmap, waiting for map\n");
2304 list_add(&req->r_wait, &mdsc->waiting_for_map);
d5548492 2305 return;
cc8e8342 2306 }
e9e427f0
YZ		 2307 if (!(mdsc->fsc->mount_options->flags &
2308 CEPH_MOUNT_OPT_MOUNTWAIT) &&
2309 !ceph_mdsmap_is_cluster_available(mdsc->mdsmap)) {
2310 err = -ENOENT;
2311 pr_info("probably no mds server is up\n");
2312 goto finish;
2313 }
2314 }
2f2dc053 2315
dc69e2e9
SW		 2316 put_request_session(req);
2317
2f2dc053
SW		 2318 mds = __choose_mds(mdsc, req);
2319 if (mds < 0 ||
2320 ceph_mdsmap_get_state(mdsc->mdsmap, mds) < CEPH_MDS_STATE_ACTIVE) {
2321 dout("do_request no mds or not active, waiting for map\n");
2322 list_add(&req->r_wait, &mdsc->waiting_for_map);
d5548492 2323 return;
2f2dc053
SW		 2324 }
2325
2326 /* get, open session */
2327 session = __ceph_lookup_mds_session(mdsc, mds);
9c423956 2328 if (!session) {
2f2dc053 2329 session = register_session(mdsc, mds);
9c423956
SW		 2330 if (IS_ERR(session)) {
2331 err = PTR_ERR(session);
2332 goto finish;
2333 }
2334 }
dc69e2e9
SW		 2335 req->r_session = get_session(session);
2336
2f2dc053 2337 dout("do_request mds%d session %p state %s\n", mds, session,
a687ecaf 2338 ceph_session_state_name(session->s_state));
2f2dc053
SW		 2339 if (session->s_state != CEPH_MDS_SESSION_OPEN &&
2340 session->s_state != CEPH_MDS_SESSION_HUNG) {
fcff415c
YZ		 2341 if (session->s_state == CEPH_MDS_SESSION_REJECTED) {
2342 err = -EACCES;
2343 goto out_session;
2344 }
2f2dc053
SW		 2345 if (session->s_state == CEPH_MDS_SESSION_NEW ||
2346 session->s_state == CEPH_MDS_SESSION_CLOSING)
2347 __open_session(mdsc, session);
2348 list_add(&req->r_wait, &session->s_waiting);
2349 goto out_session;
2350 }
2351
2352 /* send request */
2f2dc053
SW		 2353 req->r_resend_mds = -1; /* forget any previous mds hint */
2354
2355 if (req->r_request_started == 0) /* note request start time */
2356 req->r_request_started = jiffies;
2357
6e6f0923 2358 err = __prepare_send_request(mdsc, req, mds, false);
2f2dc053
SW		 2359 if (!err) {
2360 ceph_msg_get(req->r_request);
2361 ceph_con_send(&session->s_con, req->r_request);
2362 }
2363
2364out_session:
2365 ceph_put_mds_session(session);
48fec5d0
YZ		 2366finish:
2367 if (err) {
2368 dout("__do_request early error %d\n", err);
2369 req->r_err = err;
2370 complete_request(mdsc, req);
2371 __unregister_request(mdsc, req);
2372 }
d5548492 2373 return;
2f2dc053
SW		 2374}
2375
2376/*
2377 * called under mdsc->mutex
2378 */
2379static void __wake_requests(struct ceph_mds_client *mdsc,
2380 struct list_head *head)
2381{
ed75ec2c
YZ		 2382 struct ceph_mds_request *req;
2383 LIST_HEAD(tmp_list);
2384
2385 list_splice_init(head, &tmp_list);
2f2dc053 2386
ed75ec2c
YZ		 2387 while (!list_empty(&tmp_list)) {
2388 req = list_entry(tmp_list.next,
2389 struct ceph_mds_request, r_wait);
2f2dc053 2390 list_del_init(&req->r_wait);
7971bd92 2391 dout(" wake request %p tid %llu\n", req, req->r_tid);
2f2dc053
SW		 2392 __do_request(mdsc, req);
2393 }
2394}
2395
2396/*
2397 * Wake up threads with requests pending for @mds, so that they can
29790f26 2398 * resubmit their requests to a possibly different mds.
2f2dc053 2399 */
29790f26 2400static void kick_requests(struct ceph_mds_client *mdsc, int mds)
2f2dc053 2401{
44ca18f2 2402 struct ceph_mds_request *req;
282c1052 2403 struct rb_node *p = rb_first(&mdsc->request_tree);
2f2dc053
SW		 2404
2405 dout("kick_requests mds%d\n", mds);
282c1052 2406 while (p) {
44ca18f2 2407 req = rb_entry(p, struct ceph_mds_request, r_node);
282c1052 2408 p = rb_next(p);
bc2de10d 2409 if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
44ca18f2 2410 continue;
3de22be6
YZ		 2411 if (req->r_attempts > 0)
2412 continue; /* only new requests */
44ca18f2
SW		 2413 if (req->r_session &&
2414 req->r_session->s_mds == mds) {
2415 dout(" kicking tid %llu\n", req->r_tid);
03974e81 2416 list_del_init(&req->r_wait);
44ca18f2 2417 __do_request(mdsc, req);
2f2dc053
SW		 2418 }
2419 }
2420}
2421
2422void ceph_mdsc_submit_request(struct ceph_mds_client *mdsc,
2423 struct ceph_mds_request *req)
2424{
2425 dout("submit_request on %p\n", req);
2426 mutex_lock(&mdsc->mutex);
2427 __register_request(mdsc, req, NULL);
2428 __do_request(mdsc, req);
2429 mutex_unlock(&mdsc->mutex);
2430}
2431
2432/*
2433 * Synchrously perform an mds request. Take care of all of the
2434 * session setup, forwarding, retry details.
2435 */
2436int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
2437 struct inode *dir,
2438 struct ceph_mds_request *req)
2439{
2440 int err;
2441
2442 dout("do_request on %p\n", req);
2443
3dd69aab 2444 /* take CAP_PIN refs for r_inode, r_parent, r_old_dentry */
2f2dc053
SW		 2445 if (req->r_inode)
2446 ceph_get_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
3dd69aab
JL		 2447 if (req->r_parent)
2448 ceph_get_cap_refs(ceph_inode(req->r_parent), CEPH_CAP_PIN);
844d87c3 2449 if (req->r_old_dentry_dir)
41b02e1f
SW		 2450 ceph_get_cap_refs(ceph_inode(req->r_old_dentry_dir),
2451 CEPH_CAP_PIN);
2f2dc053
SW		 2452
2453 /* issue */
2454 mutex_lock(&mdsc->mutex);
2455 __register_request(mdsc, req, dir);
2456 __do_request(mdsc, req);
2457
e1518c7c
SW		 2458 if (req->r_err) {
2459 err = req->r_err;
e1518c7c 2460 goto out;
2f2dc053
SW		 2461 }
2462
e1518c7c
SW		 2463 /* wait */
2464 mutex_unlock(&mdsc->mutex);
2465 dout("do_request waiting\n");
5be73034 2466 if (!req->r_timeout && req->r_wait_for_completion) {
9280be24 2467 err = req->r_wait_for_completion(mdsc, req);
e1518c7c 2468 } else {
5be73034
ID		 2469 long timeleft = wait_for_completion_killable_timeout(
2470 &req->r_completion,
2471 ceph_timeout_jiffies(req->r_timeout));
2472 if (timeleft > 0)
2473 err = 0;
2474 else if (!timeleft)
2475 err = -EIO; /* timed out */
2476 else
2477 err = timeleft; /* killed */
e1518c7c
SW		 2478 }
2479 dout("do_request waited, got %d\n", err);
2480 mutex_lock(&mdsc->mutex);
5b1daecd 2481
e1518c7c 2482 /* only abort if we didn't race with a real reply */
bc2de10d 2483 if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
e1518c7c
SW		 2484 err = le32_to_cpu(req->r_reply_info.head->result);
2485 } else if (err < 0) {
2486 dout("aborted request %lld with %d\n", req->r_tid, err);
b4556396
SW		 2487
2488 /*
2489 * ensure we aren't running concurrently with
2490 * ceph_fill_trace or ceph_readdir_prepopulate, which
2491 * rely on locks (dir mutex) held by our caller.
2492 */
2493 mutex_lock(&req->r_fill_mutex);
e1518c7c 2494 req->r_err = err;
bc2de10d 2495 set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
b4556396 2496 mutex_unlock(&req->r_fill_mutex);
5b1daecd 2497
3dd69aab 2498 if (req->r_parent &&
167c9e35
SW		 2499 (req->r_op & CEPH_MDS_OP_WRITE))
2500 ceph_invalidate_dir_request(req);
2f2dc053 2501 } else {
e1518c7c 2502 err = req->r_err;
2f2dc053 2503 }
2f2dc053 2504
e1518c7c
SW		 2505out:
2506 mutex_unlock(&mdsc->mutex);
2f2dc053
SW		 2507 dout("do_request %p done, result %d\n", req, err);
2508 return err;
2509}
2510
167c9e35 2511/*
2f276c51 2512 * Invalidate dir's completeness, dentry lease state on an aborted MDS
167c9e35
SW		 2513 * namespace request.
2514 */
2515void ceph_invalidate_dir_request(struct ceph_mds_request *req)
2516{
8d8f371c
YZ		 2517 struct inode *dir = req->r_parent;
2518 struct inode *old_dir = req->r_old_dentry_dir;
167c9e35 2519
8d8f371c 2520 dout("invalidate_dir_request %p %p (complete, lease(s))\n", dir, old_dir);
167c9e35 2521
8d8f371c
YZ		 2522 ceph_dir_clear_complete(dir);
2523 if (old_dir)
2524 ceph_dir_clear_complete(old_dir);
167c9e35
SW		 2525 if (req->r_dentry)
2526 ceph_invalidate_dentry_lease(req->r_dentry);
2527 if (req->r_old_dentry)
2528 ceph_invalidate_dentry_lease(req->r_old_dentry);
2529}
2530
2f2dc053
SW		 2531/*
2532 * Handle mds reply.
2533 *
2534 * We take the session mutex and parse and process the reply immediately.
2535 * This preserves the logical ordering of replies, capabilities, etc., sent
2536 * by the MDS as they are applied to our local cache.
2537 */
2538static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
2539{
2540 struct ceph_mds_client *mdsc = session->s_mdsc;
2541 struct ceph_mds_request *req;
2542 struct ceph_mds_reply_head *head = msg->front.iov_base;
2543 struct ceph_mds_reply_info_parsed *rinfo; /* parsed reply info */
982d6011 2544 struct ceph_snap_realm *realm;
2f2dc053
SW		 2545 u64 tid;
2546 int err, result;
2600d2dd 2547 int mds = session->s_mds;
2f2dc053 2548
2f2dc053
SW		 2549 if (msg->front.iov_len < sizeof(*head)) {
2550 pr_err("mdsc_handle_reply got corrupt (short) reply\n");
9ec7cab1 2551 ceph_msg_dump(msg);
2f2dc053
SW		 2552 return;
2553 }
2554
2555 /* get request, session */
6df058c0 2556 tid = le64_to_cpu(msg->hdr.tid);
2f2dc053 2557 mutex_lock(&mdsc->mutex);
fcd00b68 2558 req = lookup_get_request(mdsc, tid);
2f2dc053
SW		 2559 if (!req) {
2560 dout("handle_reply on unknown tid %llu\n", tid);
2561 mutex_unlock(&mdsc->mutex);
2562 return;
2563 }
2564 dout("handle_reply %p\n", req);
2f2dc053
SW		 2565
2566 /* correct session? */
d96d6049 2567 if (req->r_session != session) {
2f2dc053
SW		 2568 pr_err("mdsc_handle_reply got %llu on session mds%d"
2569 " not mds%d\n", tid, session->s_mds,
2570 req->r_session ? req->r_session->s_mds : -1);
2571 mutex_unlock(&mdsc->mutex);
2572 goto out;
2573 }
2574
2575 /* dup? */
bc2de10d
JL		 2576 if ((test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags) && !head->safe) ||
2577 (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags) && head->safe)) {
f3ae1b97 2578 pr_warn("got a dup %s reply on %llu from mds%d\n",
2f2dc053
SW		 2579 head->safe ? "safe" : "unsafe", tid, mds);
2580 mutex_unlock(&mdsc->mutex);
2581 goto out;
2582 }
bc2de10d 2583 if (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags)) {
f3ae1b97 2584 pr_warn("got unsafe after safe on %llu from mds%d\n",
85792d0d
SW		 2585 tid, mds);
2586 mutex_unlock(&mdsc->mutex);
2587 goto out;
2588 }
2f2dc053
SW		 2589
2590 result = le32_to_cpu(head->result);
2591
2592 /*
e55b71f8
GF		 2593 * Handle an ESTALE
2594 * if we're not talking to the authority, send to them
2595 * if the authority has changed while we weren't looking,
2596 * send to new authority
2597 * Otherwise we just have to return an ESTALE
2f2dc053
SW		 2598 */
2599 if (result == -ESTALE) {
4c069a58 2600 dout("got ESTALE on request %llu\n", req->r_tid);
51da8e8c 2601 req->r_resend_mds = -1;
ca18bede 2602 if (req->r_direct_mode != USE_AUTH_MDS) {
4c069a58 2603 dout("not using auth, setting for that now\n");
e55b71f8 2604 req->r_direct_mode = USE_AUTH_MDS;
2f2dc053
SW		 2605 __do_request(mdsc, req);
2606 mutex_unlock(&mdsc->mutex);
2607 goto out;
e55b71f8 2608 } else {
ca18bede
YZ		 2609 int mds = __choose_mds(mdsc, req);
2610 if (mds >= 0 && mds != req->r_session->s_mds) {
4c069a58 2611 dout("but auth changed, so resending\n");
e55b71f8
GF		 2612 __do_request(mdsc, req);
2613 mutex_unlock(&mdsc->mutex);
2614 goto out;
2615 }
2f2dc053 2616 }
4c069a58 2617 dout("have to return ESTALE on request %llu\n", req->r_tid);
2f2dc053
SW		 2618 }
2619
e55b71f8 2620
2f2dc053 2621 if (head->safe) {
bc2de10d 2622 set_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags);
2f2dc053 2623 __unregister_request(mdsc, req);
2f2dc053 2624
bc2de10d 2625 if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
2f2dc053
SW		 2626 /*
2627 * We already handled the unsafe response, now do the
2628 * cleanup. No need to examine the response; the MDS
2629 * doesn't include any result info in the safe
2630 * response. And even if it did, there is nothing
2631 * useful we could do with a revised return value.
2632 */
2633 dout("got safe reply %llu, mds%d\n", tid, mds);
2f2dc053
SW		 2634
2635 /* last unsafe request during umount? */
44ca18f2 2636 if (mdsc->stopping && !__get_oldest_req(mdsc))
03066f23 2637 complete_all(&mdsc->safe_umount_waiters);
2f2dc053
SW		 2638 mutex_unlock(&mdsc->mutex);
2639 goto out;
2640 }
e1518c7c 2641 } else {
bc2de10d 2642 set_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags);
2f2dc053 2643 list_add_tail(&req->r_unsafe_item, &req->r_session->s_unsafe);
4c06ace8
YZ		 2644 if (req->r_unsafe_dir) {
2645 struct ceph_inode_info *ci =
2646 ceph_inode(req->r_unsafe_dir);
2647 spin_lock(&ci->i_unsafe_lock);
2648 list_add_tail(&req->r_unsafe_dir_item,
2649 &ci->i_unsafe_dirops);
2650 spin_unlock(&ci->i_unsafe_lock);
2651 }
2f2dc053
SW		 2652 }
2653
2654 dout("handle_reply tid %lld result %d\n", tid, result);
2655 rinfo = &req->r_reply_info;
14303d20 2656 err = parse_reply_info(msg, rinfo, session->s_con.peer_features);
2f2dc053
SW		 2657 mutex_unlock(&mdsc->mutex);
2658
2659 mutex_lock(&session->s_mutex);
2660 if (err < 0) {
25933abd 2661 pr_err("mdsc_handle_reply got corrupt reply mds%d(tid:%lld)\n", mds, tid);
9ec7cab1 2662 ceph_msg_dump(msg);
2f2dc053
SW		 2663 goto out_err;
2664 }
2665
2666 /* snap trace */
982d6011 2667 realm = NULL;
2f2dc053
SW		 2668 if (rinfo->snapblob_len) {
2669 down_write(&mdsc->snap_rwsem);
2670 ceph_update_snap_trace(mdsc, rinfo->snapblob,
982d6011
YZ		 2671 rinfo->snapblob + rinfo->snapblob_len,
2672 le32_to_cpu(head->op) == CEPH_MDS_OP_RMSNAP,
2673 &realm);
2f2dc053
SW		 2674 downgrade_write(&mdsc->snap_rwsem);
2675 } else {
2676 down_read(&mdsc->snap_rwsem);
2677 }
2678
2679 /* insert trace into our cache */
b4556396 2680 mutex_lock(&req->r_fill_mutex);
315f2408 2681 current->journal_info = req;
f5a03b08 2682 err = ceph_fill_trace(mdsc->fsc->sb, req);
2f2dc053 2683 if (err == 0) {
6e8575fa 2684 if (result == 0 && (req->r_op == CEPH_MDS_OP_READDIR ||
81c6aea5 2685 req->r_op == CEPH_MDS_OP_LSSNAP))
2f2dc053 2686 ceph_readdir_prepopulate(req, req->r_session);
37151668 2687 ceph_unreserve_caps(mdsc, &req->r_caps_reservation);
2f2dc053 2688 }
315f2408 2689 current->journal_info = NULL;
b4556396 2690 mutex_unlock(&req->r_fill_mutex);
2f2dc053
SW		 2691
2692 up_read(&mdsc->snap_rwsem);
982d6011
YZ		 2693 if (realm)
2694 ceph_put_snap_realm(mdsc, realm);
68cd5b4b 2695
bc2de10d
JL		 2696 if (err == 0 && req->r_target_inode &&
2697 test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
68cd5b4b
YZ		 2698 struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
2699 spin_lock(&ci->i_unsafe_lock);
2700 list_add_tail(&req->r_unsafe_target_item, &ci->i_unsafe_iops);
2701 spin_unlock(&ci->i_unsafe_lock);
2702 }
2f2dc053 2703out_err:
e1518c7c 2704 mutex_lock(&mdsc->mutex);
bc2de10d 2705 if (!test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
e1518c7c
SW		 2706 if (err) {
2707 req->r_err = err;
2708 } else {
5fdb1389 2709 req->r_reply = ceph_msg_get(msg);
bc2de10d 2710 set_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags);
e1518c7c 2711 }
2f2dc053 2712 } else {
e1518c7c 2713 dout("reply arrived after request %lld was aborted\n", tid);
2f2dc053 2714 }
e1518c7c 2715 mutex_unlock(&mdsc->mutex);
2f2dc053 2716
2f2dc053
SW		 2717 mutex_unlock(&session->s_mutex);
2718
2719 /* kick calling process */
2720 complete_request(mdsc, req);
2721out:
2722 ceph_mdsc_put_request(req);
2723 return;
2724}
2725
2726
2727
2728/*
2729 * handle mds notification that our request has been forwarded.
2730 */
2600d2dd
SW		 2731static void handle_forward(struct ceph_mds_client *mdsc,
2732 struct ceph_mds_session *session,
2733 struct ceph_msg *msg)
2f2dc053
SW		 2734{
2735 struct ceph_mds_request *req;
a1ea787c 2736 u64 tid = le64_to_cpu(msg->hdr.tid);
2f2dc053
SW		 2737 u32 next_mds;
2738 u32 fwd_seq;
2f2dc053
SW		 2739 int err = -EINVAL;
2740 void *p = msg->front.iov_base;
2741 void *end = p + msg->front.iov_len;
2f2dc053 2742
a1ea787c 2743 ceph_decode_need(&p, end, 2*sizeof(u32), bad);
c89136ea
SW		 2744 next_mds = ceph_decode_32(&p);
2745 fwd_seq = ceph_decode_32(&p);
2f2dc053
SW		 2746
2747 mutex_lock(&mdsc->mutex);
fcd00b68 2748 req = lookup_get_request(mdsc, tid);
2f2dc053 2749 if (!req) {
2a8e5e36 2750 dout("forward tid %llu to mds%d - req dne\n", tid, next_mds);
2f2dc053
SW		 2751 goto out; /* dup reply? */
2752 }
2753
bc2de10d 2754 if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
2a8e5e36
SW		 2755 dout("forward tid %llu aborted, unregistering\n", tid);
2756 __unregister_request(mdsc, req);
2757 } else if (fwd_seq <= req->r_num_fwd) {
2758 dout("forward tid %llu to mds%d - old seq %d <= %d\n",
2f2dc053
SW		 2759 tid, next_mds, req->r_num_fwd, fwd_seq);
2760 } else {
2761 /* resend. forward race not possible; mds would drop */
2a8e5e36
SW		 2762 dout("forward tid %llu to mds%d (we resend)\n", tid, next_mds);
2763 BUG_ON(req->r_err);
bc2de10d 2764 BUG_ON(test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags));
3de22be6 2765 req->r_attempts = 0;
2f2dc053
SW		 2766 req->r_num_fwd = fwd_seq;
2767 req->r_resend_mds = next_mds;
2768 put_request_session(req);
2769 __do_request(mdsc, req);
2770 }
2771 ceph_mdsc_put_request(req);
2772out:
2773 mutex_unlock(&mdsc->mutex);
2774 return;
2775
2776bad:
2777 pr_err("mdsc_handle_forward decode error err=%d\n", err);
2778}
2779
2780/*
2781 * handle a mds session control message
2782 */
2783static void handle_session(struct ceph_mds_session *session,
2784 struct ceph_msg *msg)
2785{
2786 struct ceph_mds_client *mdsc = session->s_mdsc;
2787 u32 op;
2788 u64 seq;
2600d2dd 2789 int mds = session->s_mds;
2f2dc053
SW		 2790 struct ceph_mds_session_head *h = msg->front.iov_base;
2791 int wake = 0;
2792
2f2dc053 2793 /* decode */
342ce182 2794 if (msg->front.iov_len < sizeof(*h))
2f2dc053
SW		 2795 goto bad;
2796 op = le32_to_cpu(h->op);
2797 seq = le64_to_cpu(h->seq);
2798
2799 mutex_lock(&mdsc->mutex);
0a07fc8c
YZ		 2800 if (op == CEPH_SESSION_CLOSE) {
2801 get_session(session);
2600d2dd 2802 __unregister_session(mdsc, session);
0a07fc8c 2803 }
2f2dc053
SW		 2804 /* FIXME: this ttl calculation is generous */
2805 session->s_ttl = jiffies + HZ*mdsc->mdsmap->m_session_autoclose;
2806 mutex_unlock(&mdsc->mutex);
2807
2808 mutex_lock(&session->s_mutex);
2809
2810 dout("handle_session mds%d %s %p state %s seq %llu\n",
2811 mds, ceph_session_op_name(op), session,
a687ecaf 2812 ceph_session_state_name(session->s_state), seq);
2f2dc053
SW		 2813
2814 if (session->s_state == CEPH_MDS_SESSION_HUNG) {
2815 session->s_state = CEPH_MDS_SESSION_OPEN;
2816 pr_info("mds%d came back\n", session->s_mds);
2817 }
2818
2819 switch (op) {
2820 case CEPH_SESSION_OPEN:
29790f26
SW		 2821 if (session->s_state == CEPH_MDS_SESSION_RECONNECTING)
2822 pr_info("mds%d reconnect success\n", session->s_mds);
2f2dc053
SW		 2823 session->s_state = CEPH_MDS_SESSION_OPEN;
2824 renewed_caps(mdsc, session, 0);
2825 wake = 1;
2826 if (mdsc->stopping)
2827 __close_session(mdsc, session);
2828 break;
2829
2830 case CEPH_SESSION_RENEWCAPS:
2831 if (session->s_renew_seq == seq)
2832 renewed_caps(mdsc, session, 1);
2833 break;
2834
2835 case CEPH_SESSION_CLOSE:
29790f26
SW		 2836 if (session->s_state == CEPH_MDS_SESSION_RECONNECTING)
2837 pr_info("mds%d reconnect denied\n", session->s_mds);
1c841a96 2838 cleanup_session_requests(mdsc, session);
2f2dc053 2839 remove_session_caps(session);
656e4382 2840 wake = 2; /* for good measure */
f3c60c59 2841 wake_up_all(&mdsc->session_close_wq);
2f2dc053
SW		 2842 break;
2843
2844 case CEPH_SESSION_STALE:
2845 pr_info("mds%d caps went stale, renewing\n",
2846 session->s_mds);
d8fb02ab 2847 spin_lock(&session->s_gen_ttl_lock);
2f2dc053 2848 session->s_cap_gen++;
1ce208a6 2849 session->s_cap_ttl = jiffies - 1;
d8fb02ab 2850 spin_unlock(&session->s_gen_ttl_lock);
2f2dc053
SW		 2851 send_renew_caps(mdsc, session);
2852 break;
2853
2854 case CEPH_SESSION_RECALL_STATE:
e30ee581 2855 ceph_trim_caps(mdsc, session, le32_to_cpu(h->max_caps));
2f2dc053
SW		 2856 break;
2857
186e4f7a
YZ		 2858 case CEPH_SESSION_FLUSHMSG:
2859 send_flushmsg_ack(mdsc, session, seq);
2860 break;
2861
03f4fcb0
YZ		 2862 case CEPH_SESSION_FORCE_RO:
2863 dout("force_session_readonly %p\n", session);
2864 spin_lock(&session->s_cap_lock);
2865 session->s_readonly = true;
2866 spin_unlock(&session->s_cap_lock);
d2f8bb27 2867 wake_up_session_caps(session, FORCE_RO);
03f4fcb0
YZ		 2868 break;
2869
fcff415c
YZ		 2870 case CEPH_SESSION_REJECT:
2871 WARN_ON(session->s_state != CEPH_MDS_SESSION_OPENING);
2872 pr_info("mds%d rejected session\n", session->s_mds);
2873 session->s_state = CEPH_MDS_SESSION_REJECTED;
2874 cleanup_session_requests(mdsc, session);
2875 remove_session_caps(session);
2876 wake = 2; /* for good measure */
2877 break;
2878
2f2dc053
SW		 2879 default:
2880 pr_err("mdsc_handle_session bad op %d mds%d\n", op, mds);
2881 WARN_ON(1);
2882 }
2883
2884 mutex_unlock(&session->s_mutex);
2885 if (wake) {
2886 mutex_lock(&mdsc->mutex);
2887 __wake_requests(mdsc, &session->s_waiting);
656e4382
YZ		 2888 if (wake == 2)
2889 kick_requests(mdsc, mds);
2f2dc053
SW		 2890 mutex_unlock(&mdsc->mutex);
2891 }
0a07fc8c
YZ		 2892 if (op == CEPH_SESSION_CLOSE)
2893 ceph_put_mds_session(session);
2f2dc053
SW		 2894 return;
2895
2896bad:
2897 pr_err("mdsc_handle_session corrupt message mds%d len %d\n", mds,
2898 (int)msg->front.iov_len);
9ec7cab1 2899 ceph_msg_dump(msg);
2f2dc053
SW		 2900 return;
2901}
2902
2903
2904/*
2905 * called under session->mutex.
2906 */
2907static void replay_unsafe_requests(struct ceph_mds_client *mdsc,
2908 struct ceph_mds_session *session)
2909{
2910 struct ceph_mds_request *req, *nreq;
3de22be6 2911 struct rb_node *p;
2f2dc053
SW		 2912 int err;
2913
2914 dout("replay_unsafe_requests mds%d\n", session->s_mds);
2915
2916 mutex_lock(&mdsc->mutex);
2917 list_for_each_entry_safe(req, nreq, &session->s_unsafe, r_unsafe_item) {
6e6f0923 2918 err = __prepare_send_request(mdsc, req, session->s_mds, true);
2f2dc053
SW		 2919 if (!err) {
2920 ceph_msg_get(req->r_request);
2921 ceph_con_send(&session->s_con, req->r_request);
2922 }
2923 }
3de22be6
YZ		 2924
2925 /*
2926 * also re-send old requests when MDS enters reconnect stage. So that MDS
2927 * can process completed request in clientreplay stage.
2928 */
2929 p = rb_first(&mdsc->request_tree);
2930 while (p) {
2931 req = rb_entry(p, struct ceph_mds_request, r_node);
2932 p = rb_next(p);
bc2de10d 2933 if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
3de22be6
YZ		 2934 continue;
2935 if (req->r_attempts == 0)
2936 continue; /* only old requests */
2937 if (req->r_session &&
2938 req->r_session->s_mds == session->s_mds) {
6e6f0923
YZ		 2939 err = __prepare_send_request(mdsc, req,
2940 session->s_mds, true);
3de22be6
YZ		 2941 if (!err) {
2942 ceph_msg_get(req->r_request);
2943 ceph_con_send(&session->s_con, req->r_request);
2944 }
2945 }
2946 }
2f2dc053
SW		 2947 mutex_unlock(&mdsc->mutex);
2948}
2949
2950/*
2951 * Encode information about a cap for a reconnect with the MDS.
2952 */
2f2dc053
SW		 2953static int encode_caps_cb(struct inode *inode, struct ceph_cap *cap,
2954 void *arg)
2955{
20cb34ae
SW		 2956 union {
2957 struct ceph_mds_cap_reconnect v2;
2958 struct ceph_mds_cap_reconnect_v1 v1;
2959 } rec;
b3f8d68f 2960 struct ceph_inode_info *ci = cap->ci;
20cb34ae
SW		 2961 struct ceph_reconnect_state *recon_state = arg;
2962 struct ceph_pagelist *pagelist = recon_state->pagelist;
5ccedf1c 2963 int err;
3469ed0d 2964 u64 snap_follows;
2f2dc053 2965
2f2dc053
SW		 2966 dout(" adding %p ino %llx.%llx cap %p %lld %s\n",
2967 inode, ceph_vinop(inode), cap, cap->cap_id,
2968 ceph_cap_string(cap->issued));
93cea5be
SW		 2969 err = ceph_pagelist_encode_64(pagelist, ceph_ino(inode));
2970 if (err)
2971 return err;
2f2dc053 2972
be655596 2973 spin_lock(&ci->i_ceph_lock);
2f2dc053
SW		 2974 cap->seq = 0; /* reset cap seq */
2975 cap->issue_seq = 0; /* and issue_seq */
667ca05c 2976 cap->mseq = 0; /* and migrate_seq */
99a9c273 2977 cap->cap_gen = cap->session->s_cap_gen;
20cb34ae 2978
121f22a1 2979 if (recon_state->msg_version >= 2) {
20cb34ae
SW		 2980 rec.v2.cap_id = cpu_to_le64(cap->cap_id);
2981 rec.v2.wanted = cpu_to_le32(__ceph_caps_wanted(ci));
2982 rec.v2.issued = cpu_to_le32(cap->issued);
2983 rec.v2.snaprealm = cpu_to_le64(ci->i_snap_realm->ino);
5ccedf1c 2984 rec.v2.pathbase = 0;
ec1dff25
JL		 2985 rec.v2.flock_len = (__force __le32)
2986 ((ci->i_ceph_flags & CEPH_I_ERROR_FILELOCK) ? 0 : 1);
20cb34ae
SW		 2987 } else {
2988 rec.v1.cap_id = cpu_to_le64(cap->cap_id);
2989 rec.v1.wanted = cpu_to_le32(__ceph_caps_wanted(ci));
2990 rec.v1.issued = cpu_to_le32(cap->issued);
2991 rec.v1.size = cpu_to_le64(inode->i_size);
9bbeab41
AB		 2992 ceph_encode_timespec64(&rec.v1.mtime, &inode->i_mtime);
2993 ceph_encode_timespec64(&rec.v1.atime, &inode->i_atime);
20cb34ae 2994 rec.v1.snaprealm = cpu_to_le64(ci->i_snap_realm->ino);
5ccedf1c 2995 rec.v1.pathbase = 0;
20cb34ae 2996 }
3469ed0d
YZ		 2997
2998 if (list_empty(&ci->i_cap_snaps)) {
92776fd2 2999 snap_follows = ci->i_head_snapc ? ci->i_head_snapc->seq : 0;
3469ed0d
YZ		 3000 } else {
3001 struct ceph_cap_snap *capsnap =
3002 list_first_entry(&ci->i_cap_snaps,
3003 struct ceph_cap_snap, ci_item);
3004 snap_follows = capsnap->follows;
20cb34ae 3005 }
be655596 3006 spin_unlock(&ci->i_ceph_lock);
2f2dc053 3007
121f22a1 3008 if (recon_state->msg_version >= 2) {
40819f6f 3009 int num_fcntl_locks, num_flock_locks;
4deb14a2 3010 struct ceph_filelock *flocks = NULL;
121f22a1
YZ		 3011 size_t struct_len, total_len = 0;
3012 u8 struct_v = 0;
39be95e9
JS		 3013
3014encode_again:
b3f8d68f
YZ		 3015 if (rec.v2.flock_len) {
3016 ceph_count_locks(inode, &num_fcntl_locks, &num_flock_locks);
3017 } else {
3018 num_fcntl_locks = 0;
3019 num_flock_locks = 0;
3020 }
4deb14a2 3021 if (num_fcntl_locks + num_flock_locks > 0) {
6da2ec56
KC		 3022 flocks = kmalloc_array(num_fcntl_locks + num_flock_locks,
3023 sizeof(struct ceph_filelock),
3024 GFP_NOFS);
4deb14a2
YZ		 3025 if (!flocks) {
3026 err = -ENOMEM;
5ccedf1c 3027 goto out_err;
4deb14a2
YZ		 3028 }
3029 err = ceph_encode_locks_to_buffer(inode, flocks,
3030 num_fcntl_locks,
3031 num_flock_locks);
3032 if (err) {
3033 kfree(flocks);
3034 flocks = NULL;
3035 if (err == -ENOSPC)
3036 goto encode_again;
5ccedf1c 3037 goto out_err;
4deb14a2
YZ		 3038 }
3039 } else {
39be95e9 3040 kfree(flocks);
4deb14a2 3041 flocks = NULL;
39be95e9 3042 }
121f22a1
YZ		 3043
3044 if (recon_state->msg_version >= 3) {
3045 /* version, compat_version and struct_len */
3046 total_len = 2 * sizeof(u8) + sizeof(u32);
3469ed0d 3047 struct_v = 2;
121f22a1 3048 }
39be95e9
JS		 3049 /*
3050 * number of encoded locks is stable, so copy to pagelist
3051 */
121f22a1
YZ		 3052 struct_len = 2 * sizeof(u32) +
3053 (num_fcntl_locks + num_flock_locks) *
3054 sizeof(struct ceph_filelock);
3055 rec.v2.flock_len = cpu_to_le32(struct_len);
3056
5ccedf1c 3057 struct_len += sizeof(u32) + sizeof(rec.v2);
121f22a1 3058
3469ed0d
YZ		 3059 if (struct_v >= 2)
3060 struct_len += sizeof(u64); /* snap_follows */
3061
121f22a1
YZ		 3062 total_len += struct_len;
3063 err = ceph_pagelist_reserve(pagelist, total_len);
5ccedf1c
YZ		 3064 if (err) {
3065 kfree(flocks);
3066 goto out_err;
3067 }
121f22a1 3068
5ccedf1c
YZ		 3069 if (recon_state->msg_version >= 3) {
3070 ceph_pagelist_encode_8(pagelist, struct_v);
3071 ceph_pagelist_encode_8(pagelist, 1);
3072 ceph_pagelist_encode_32(pagelist, struct_len);
121f22a1 3073 }
5ccedf1c
YZ		 3074 ceph_pagelist_encode_string(pagelist, NULL, 0);
3075 ceph_pagelist_append(pagelist, &rec, sizeof(rec.v2));
3076 ceph_locks_to_pagelist(flocks, pagelist,
3077 num_fcntl_locks, num_flock_locks);
3078 if (struct_v >= 2)
3079 ceph_pagelist_encode_64(pagelist, snap_follows);
3080
39be95e9 3081 kfree(flocks);
3612abbd 3082 } else {
5ccedf1c
YZ		 3083 u64 pathbase = 0;
3084 int pathlen = 0;
3085 char *path = NULL;
3086 struct dentry *dentry;
3087
3088 dentry = d_find_alias(inode);
3089 if (dentry) {
3090 path = ceph_mdsc_build_path(dentry,
3091 &pathlen, &pathbase, 0);
3092 dput(dentry);
3093 if (IS_ERR(path)) {
3094 err = PTR_ERR(path);
3095 goto out_err;
3096 }
3097 rec.v1.pathbase = cpu_to_le64(pathbase);
121f22a1 3098 }
5ccedf1c
YZ		 3099
3100 err = ceph_pagelist_reserve(pagelist,
3101 pathlen + sizeof(u32) + sizeof(rec.v1));
3102 if (err) {
3103 kfree(path);
3104 goto out_err;
3105 }
3106
3107 ceph_pagelist_encode_string(pagelist, path, pathlen);
3108 ceph_pagelist_append(pagelist, &rec, sizeof(rec.v1));
3109
3110 kfree(path);
40819f6f 3111 }
44c99757
YZ		 3112
3113 recon_state->nr_caps++;
5ccedf1c 3114out_err:
93cea5be 3115 return err;
2f2dc053
SW		 3116}
3117
3118
3119/*
3120 * If an MDS fails and recovers, clients need to reconnect in order to
3121 * reestablish shared state. This includes all caps issued through
3122 * this session _and_ the snap_realm hierarchy. Because it's not
3123 * clear which snap realms the mds cares about, we send everything we
3124 * know about.. that ensures we'll then get any new info the
3125 * recovering MDS might have.
3126 *
3127 * This is a relatively heavyweight operation, but it's rare.
3128 *
3129 * called with mdsc->mutex held.
3130 */
34b6c855
SW		 3131static void send_mds_reconnect(struct ceph_mds_client *mdsc,
3132 struct ceph_mds_session *session)
2f2dc053 3133{
2f2dc053 3134 struct ceph_msg *reply;
a105f00c 3135 struct rb_node *p;
34b6c855 3136 int mds = session->s_mds;
9abf82b8 3137 int err = -ENOMEM;
44c99757 3138 int s_nr_caps;
93cea5be 3139 struct ceph_pagelist *pagelist;
20cb34ae 3140 struct ceph_reconnect_state recon_state;
c8a96a31 3141 LIST_HEAD(dispose);
2f2dc053 3142
34b6c855 3143 pr_info("mds%d reconnect start\n", mds);
2f2dc053 3144
33165d47 3145 pagelist = ceph_pagelist_alloc(GFP_NOFS);
93cea5be
SW		 3146 if (!pagelist)
3147 goto fail_nopagelist;
93cea5be 3148
0d9c1ab3 3149 reply = ceph_msg_new2(CEPH_MSG_CLIENT_RECONNECT, 0, 1, GFP_NOFS, false);
a79832f2 3150 if (!reply)
93cea5be 3151 goto fail_nomsg;
93cea5be 3152
34b6c855
SW		 3153 mutex_lock(&session->s_mutex);
3154 session->s_state = CEPH_MDS_SESSION_RECONNECTING;
3155 session->s_seq = 0;
2f2dc053 3156
2f2dc053 3157 dout("session %p state %s\n", session,
a687ecaf 3158 ceph_session_state_name(session->s_state));
2f2dc053 3159
99a9c273
YZ		 3160 spin_lock(&session->s_gen_ttl_lock);
3161 session->s_cap_gen++;
3162 spin_unlock(&session->s_gen_ttl_lock);
3163
3164 spin_lock(&session->s_cap_lock);
03f4fcb0
YZ		 3165 /* don't know if session is readonly */
3166 session->s_readonly = 0;
99a9c273
YZ		 3167 /*
3168 * notify __ceph_remove_cap() that we are composing cap reconnect.
3169 * If a cap get released before being added to the cap reconnect,
3170 * __ceph_remove_cap() should skip queuing cap release.
3171 */
3172 session->s_cap_reconnect = 1;
e01a5946 3173 /* drop old cap expires; we're about to reestablish that state */
c8a96a31
JL		 3174 detach_cap_releases(session, &dispose);
3175 spin_unlock(&session->s_cap_lock);
3176 dispose_cap_releases(mdsc, &dispose);
e01a5946 3177
5d23371f 3178 /* trim unused caps to reduce MDS's cache rejoin time */
c0bd50e2
YZ		 3179 if (mdsc->fsc->sb->s_root)
3180 shrink_dcache_parent(mdsc->fsc->sb->s_root);
5d23371f
YZ		 3181
3182 ceph_con_close(&session->s_con);
3183 ceph_con_open(&session->s_con,
3184 CEPH_ENTITY_TYPE_MDS, mds,
3185 ceph_mdsmap_get_addr(mdsc->mdsmap, mds));
3186
3187 /* replay unsafe requests */
3188 replay_unsafe_requests(mdsc, session);
3189
3190 down_read(&mdsc->snap_rwsem);
3191
2f2dc053 3192 /* traverse this session's caps */
44c99757
YZ		 3193 s_nr_caps = session->s_nr_caps;
3194 err = ceph_pagelist_encode_32(pagelist, s_nr_caps);
93cea5be
SW		 3195 if (err)
3196 goto fail;
20cb34ae 3197
44c99757 3198 recon_state.nr_caps = 0;
20cb34ae 3199 recon_state.pagelist = pagelist;
121f22a1
YZ		 3200 if (session->s_con.peer_features & CEPH_FEATURE_MDSENC)
3201 recon_state.msg_version = 3;
121f22a1 3202 else
23c625ce 3203 recon_state.msg_version = 2;
20cb34ae 3204 err = iterate_session_caps(session, encode_caps_cb, &recon_state);
2f2dc053 3205 if (err < 0)
9abf82b8 3206 goto fail;
2f2dc053 3207
99a9c273
YZ		 3208 spin_lock(&session->s_cap_lock);
3209 session->s_cap_reconnect = 0;
3210 spin_unlock(&session->s_cap_lock);
3211
2f2dc053
SW		 3212 /*
3213 * snaprealms. we provide mds with the ino, seq (version), and
3214 * parent for all of our realms. If the mds has any newer info,
3215 * it will tell us.
3216 */
a105f00c
SW		 3217 for (p = rb_first(&mdsc->snap_realms); p; p = rb_next(p)) {
3218 struct ceph_snap_realm *realm =
3219 rb_entry(p, struct ceph_snap_realm, node);
93cea5be 3220 struct ceph_mds_snaprealm_reconnect sr_rec;
2f2dc053
SW		 3221
3222 dout(" adding snap realm %llx seq %lld parent %llx\n",
3223 realm->ino, realm->seq, realm->parent_ino);
93cea5be
SW		 3224 sr_rec.ino = cpu_to_le64(realm->ino);
3225 sr_rec.seq = cpu_to_le64(realm->seq);
3226 sr_rec.parent = cpu_to_le64(realm->parent_ino);
3227 err = ceph_pagelist_append(pagelist, &sr_rec, sizeof(sr_rec));
3228 if (err)
3229 goto fail;
2f2dc053 3230 }
2f2dc053 3231
121f22a1 3232 reply->hdr.version = cpu_to_le16(recon_state.msg_version);
44c99757
YZ		 3233
3234 /* raced with cap release? */
3235 if (s_nr_caps != recon_state.nr_caps) {
3236 struct page *page = list_first_entry(&pagelist->head,
3237 struct page, lru);
3238 __le32 *addr = kmap_atomic(page);
3239 *addr = cpu_to_le32(recon_state.nr_caps);
3240 kunmap_atomic(addr);
ebf18f47 3241 }
44c99757
YZ		 3242
3243 reply->hdr.data_len = cpu_to_le32(pagelist->length);
3244 ceph_msg_data_add_pagelist(reply, pagelist);
e548e9b9
YZ		 3245
3246 ceph_early_kick_flushing_caps(mdsc, session);
3247
2f2dc053
SW		 3248 ceph_con_send(&session->s_con, reply);
3249
9abf82b8
SW		 3250 mutex_unlock(&session->s_mutex);
3251
3252 mutex_lock(&mdsc->mutex);
3253 __wake_requests(mdsc, &session->s_waiting);
3254 mutex_unlock(&mdsc->mutex);
3255
2f2dc053 3256 up_read(&mdsc->snap_rwsem);
89486833 3257 ceph_pagelist_release(pagelist);
2f2dc053
SW		 3258 return;
3259
93cea5be 3260fail:
2f2dc053 3261 ceph_msg_put(reply);
9abf82b8
SW		 3262 up_read(&mdsc->snap_rwsem);
3263 mutex_unlock(&session->s_mutex);
93cea5be
SW		 3264fail_nomsg:
3265 ceph_pagelist_release(pagelist);
93cea5be 3266fail_nopagelist:
9abf82b8 3267 pr_err("error %d preparing reconnect for mds%d\n", err, mds);
9abf82b8 3268 return;
2f2dc053
SW		 3269}
3270
3271
3272/*
3273 * compare old and new mdsmaps, kicking requests
3274 * and closing out old connections as necessary
3275 *
3276 * called under mdsc->mutex.
3277 */
3278static void check_new_map(struct ceph_mds_client *mdsc,
3279 struct ceph_mdsmap *newmap,
3280 struct ceph_mdsmap *oldmap)
3281{
3282 int i;
3283 int oldstate, newstate;
3284 struct ceph_mds_session *s;
3285
3286 dout("check_new_map new %u old %u\n",
3287 newmap->m_epoch, oldmap->m_epoch);
3288
76201b63 3289 for (i = 0; i < oldmap->m_num_mds && i < mdsc->max_sessions; i++) {
d37b1d99 3290 if (!mdsc->sessions[i])
2f2dc053
SW		 3291 continue;
3292 s = mdsc->sessions[i];
3293 oldstate = ceph_mdsmap_get_state(oldmap, i);
3294 newstate = ceph_mdsmap_get_state(newmap, i);
3295
0deb01c9 3296 dout("check_new_map mds%d state %s%s -> %s%s (session %s)\n",
2f2dc053 3297 i, ceph_mds_state_name(oldstate),
0deb01c9 3298 ceph_mdsmap_is_laggy(oldmap, i) ? " (laggy)" : "",
2f2dc053 3299 ceph_mds_state_name(newstate),
0deb01c9 3300 ceph_mdsmap_is_laggy(newmap, i) ? " (laggy)" : "",
a687ecaf 3301 ceph_session_state_name(s->s_state));
2f2dc053 3302
76201b63 3303 if (i >= newmap->m_num_mds ||
3e8f43a0 3304 memcmp(ceph_mdsmap_get_addr(oldmap, i),
2f2dc053
SW		 3305 ceph_mdsmap_get_addr(newmap, i),
3306 sizeof(struct ceph_entity_addr))) {
3307 if (s->s_state == CEPH_MDS_SESSION_OPENING) {
3308 /* the session never opened, just close it
3309 * out now */
0a07fc8c 3310 get_session(s);
2600d2dd 3311 __unregister_session(mdsc, s);
2f2dc053 3312 __wake_requests(mdsc, &s->s_waiting);
0a07fc8c 3313 ceph_put_mds_session(s);
2827528d
YZ		 3314 } else if (i >= newmap->m_num_mds) {
3315 /* force close session for stopped mds */
3316 get_session(s);
2600d2dd 3317 __unregister_session(mdsc, s);
2827528d
YZ		 3318 __wake_requests(mdsc, &s->s_waiting);
3319 kick_requests(mdsc, i);
3320 mutex_unlock(&mdsc->mutex);
3321
3322 mutex_lock(&s->s_mutex);
3323 cleanup_session_requests(mdsc, s);
3324 remove_session_caps(s);
3325 mutex_unlock(&s->s_mutex);
3326
3327 ceph_put_mds_session(s);
3328
3329 mutex_lock(&mdsc->mutex);
2f2dc053
SW		 3330 } else {
3331 /* just close it */
3332 mutex_unlock(&mdsc->mutex);
3333 mutex_lock(&s->s_mutex);
3334 mutex_lock(&mdsc->mutex);
3335 ceph_con_close(&s->s_con);
3336 mutex_unlock(&s->s_mutex);
3337 s->s_state = CEPH_MDS_SESSION_RESTARTING;
3338 }
2f2dc053
SW		 3339 } else if (oldstate == newstate) {
3340 continue; /* nothing new with this mds */
3341 }
3342
3343 /*
3344 * send reconnect?
3345 */
3346 if (s->s_state == CEPH_MDS_SESSION_RESTARTING &&
34b6c855
SW		 3347 newstate >= CEPH_MDS_STATE_RECONNECT) {
3348 mutex_unlock(&mdsc->mutex);
3349 send_mds_reconnect(mdsc, s);
3350 mutex_lock(&mdsc->mutex);
3351 }
2f2dc053
SW		 3352
3353 /*
29790f26 3354 * kick request on any mds that has gone active.
2f2dc053
SW		 3355 */
3356 if (oldstate < CEPH_MDS_STATE_ACTIVE &&
3357 newstate >= CEPH_MDS_STATE_ACTIVE) {
29790f26
SW		 3358 if (oldstate != CEPH_MDS_STATE_CREATING &&
3359 oldstate != CEPH_MDS_STATE_STARTING)
3360 pr_info("mds%d recovery completed\n", s->s_mds);
3361 kick_requests(mdsc, i);
2f2dc053 3362 ceph_kick_flushing_caps(mdsc, s);
d2f8bb27 3363 wake_up_session_caps(s, RECONNECT);
2f2dc053
SW		 3364 }
3365 }
cb170a22 3366
76201b63 3367 for (i = 0; i < newmap->m_num_mds && i < mdsc->max_sessions; i++) {
cb170a22
SW		 3368 s = mdsc->sessions[i];
3369 if (!s)
3370 continue;
3371 if (!ceph_mdsmap_is_laggy(newmap, i))
3372 continue;
3373 if (s->s_state == CEPH_MDS_SESSION_OPEN ||
3374 s->s_state == CEPH_MDS_SESSION_HUNG ||
3375 s->s_state == CEPH_MDS_SESSION_CLOSING) {
3376 dout(" connecting to export targets of laggy mds%d\n",
3377 i);
3378 __open_export_target_sessions(mdsc, s);
3379 }
3380 }
2f2dc053
SW		 3381}
3382
3383
3384
3385/*
3386 * leases
3387 */
3388
3389/*
3390 * caller must hold session s_mutex, dentry->d_lock
3391 */
3392void __ceph_mdsc_drop_dentry_lease(struct dentry *dentry)
3393{
3394 struct ceph_dentry_info *di = ceph_dentry(dentry);
3395
3396 ceph_put_mds_session(di->lease_session);
3397 di->lease_session = NULL;
3398}
3399
2600d2dd
SW		 3400static void handle_lease(struct ceph_mds_client *mdsc,
3401 struct ceph_mds_session *session,
3402 struct ceph_msg *msg)
2f2dc053 3403{
3d14c5d2 3404 struct super_block *sb = mdsc->fsc->sb;
2f2dc053 3405 struct inode *inode;
2f2dc053
SW		 3406 struct dentry *parent, *dentry;
3407 struct ceph_dentry_info *di;
2600d2dd 3408 int mds = session->s_mds;
2f2dc053 3409 struct ceph_mds_lease *h = msg->front.iov_base;
1e5ea23d 3410 u32 seq;
2f2dc053 3411 struct ceph_vino vino;
2f2dc053
SW		 3412 struct qstr dname;
3413 int release = 0;
3414
2f2dc053
SW		 3415 dout("handle_lease from mds%d\n", mds);
3416
3417 /* decode */
3418 if (msg->front.iov_len < sizeof(*h) + sizeof(u32))
3419 goto bad;
3420 vino.ino = le64_to_cpu(h->ino);
3421 vino.snap = CEPH_NOSNAP;
1e5ea23d 3422 seq = le32_to_cpu(h->seq);
0fcf6c02
YZ		 3423 dname.len = get_unaligned_le32(h + 1);
3424 if (msg->front.iov_len < sizeof(*h) + sizeof(u32) + dname.len)
2f2dc053 3425 goto bad;
0fcf6c02 3426 dname.name = (void *)(h + 1) + sizeof(u32);
2f2dc053 3427
2f2dc053
SW		 3428 /* lookup inode */
3429 inode = ceph_find_inode(sb, vino);
2f90b852
SW		 3430 dout("handle_lease %s, ino %llx %p %.*s\n",
3431 ceph_lease_op_name(h->action), vino.ino, inode,
1e5ea23d 3432 dname.len, dname.name);
6cd3bcad
YZ		 3433
3434 mutex_lock(&session->s_mutex);
3435 session->s_seq++;
3436
d37b1d99 3437 if (!inode) {
2f2dc053
SW		 3438 dout("handle_lease no inode %llx\n", vino.ino);
3439 goto release;
3440 }
2f2dc053
SW		 3441
3442 /* dentry */
3443 parent = d_find_alias(inode);
3444 if (!parent) {
3445 dout("no parent dentry on inode %p\n", inode);
3446 WARN_ON(1);
3447 goto release; /* hrm... */
3448 }
8387ff25 3449 dname.hash = full_name_hash(parent, dname.name, dname.len);
2f2dc053
SW		 3450 dentry = d_lookup(parent, &dname);
3451 dput(parent);
3452 if (!dentry)
3453 goto release;
3454
3455 spin_lock(&dentry->d_lock);
3456 di = ceph_dentry(dentry);
3457 switch (h->action) {
3458 case CEPH_MDS_LEASE_REVOKE:
3d8eb7a9 3459 if (di->lease_session == session) {
1e5ea23d
SW		 3460 if (ceph_seq_cmp(di->lease_seq, seq) > 0)
3461 h->seq = cpu_to_le32(di->lease_seq);
2f2dc053
SW		 3462 __ceph_mdsc_drop_dentry_lease(dentry);
3463 }
3464 release = 1;
3465 break;
3466
3467 case CEPH_MDS_LEASE_RENEW:
3d8eb7a9 3468 if (di->lease_session == session &&
2f2dc053
SW		 3469 di->lease_gen == session->s_cap_gen &&
3470 di->lease_renew_from &&
3471 di->lease_renew_after == 0) {
3472 unsigned long duration =
3563dbdd 3473 msecs_to_jiffies(le32_to_cpu(h->duration_ms));
2f2dc053 3474
1e5ea23d 3475 di->lease_seq = seq;
9b16f03c 3476 di->time = di->lease_renew_from + duration;
2f2dc053
SW		 3477 di->lease_renew_after = di->lease_renew_from +
3478 (duration >> 1);
3479 di->lease_renew_from = 0;
3480 }
3481 break;
3482 }
3483 spin_unlock(&dentry->d_lock);
3484 dput(dentry);
3485
3486 if (!release)
3487 goto out;
3488
3489release:
3490 /* let's just reuse the same message */
3491 h->action = CEPH_MDS_LEASE_REVOKE_ACK;
3492 ceph_msg_get(msg);
3493 ceph_con_send(&session->s_con, msg);
3494
3495out:
3496 iput(inode);
3497 mutex_unlock(&session->s_mutex);
2f2dc053
SW		 3498 return;
3499
3500bad:
3501 pr_err("corrupt lease message\n");
9ec7cab1 3502 ceph_msg_dump(msg);
2f2dc053
SW		 3503}
3504
3505void ceph_mdsc_lease_send_msg(struct ceph_mds_session *session,
3506 struct inode *inode,
3507 struct dentry *dentry, char action,
3508 u32 seq)
3509{
3510 struct ceph_msg *msg;
3511 struct ceph_mds_lease *lease;
3512 int len = sizeof(*lease) + sizeof(u32);
3513 int dnamelen = 0;
3514
3515 dout("lease_send_msg inode %p dentry %p %s to mds%d\n",
3516 inode, dentry, ceph_lease_op_name(action), session->s_mds);
3517 dnamelen = dentry->d_name.len;
3518 len += dnamelen;
3519
b61c2763 3520 msg = ceph_msg_new(CEPH_MSG_CLIENT_LEASE, len, GFP_NOFS, false);
a79832f2 3521 if (!msg)
2f2dc053
SW		 3522 return;
3523 lease = msg->front.iov_base;
3524 lease->action = action;
2f2dc053
SW		 3525 lease->ino = cpu_to_le64(ceph_vino(inode).ino);
3526 lease->first = lease->last = cpu_to_le64(ceph_vino(inode).snap);
3527 lease->seq = cpu_to_le32(seq);
3528 put_unaligned_le32(dnamelen, lease + 1);
3529 memcpy((void *)(lease + 1) + 4, dentry->d_name.name, dnamelen);
3530
3531 /*
3532 * if this is a preemptive lease RELEASE, no need to
3533 * flush request stream, since the actual request will
3534 * soon follow.
3535 */
3536 msg->more_to_follow = (action == CEPH_MDS_LEASE_RELEASE);
3537
3538 ceph_con_send(&session->s_con, msg);
3539}
3540
2f2dc053 3541/*
7aac453a 3542 * lock unlock sessions, to wait ongoing session activities
2f2dc053 3543 */
7aac453a 3544static void lock_unlock_sessions(struct ceph_mds_client *mdsc)
2f2dc053
SW		 3545{
3546 int i;
3547
2f2dc053
SW		 3548 mutex_lock(&mdsc->mutex);
3549 for (i = 0; i < mdsc->max_sessions; i++) {
3550 struct ceph_mds_session *s = __ceph_lookup_mds_session(mdsc, i);
3551 if (!s)
3552 continue;
3553 mutex_unlock(&mdsc->mutex);
3554 mutex_lock(&s->s_mutex);
3555 mutex_unlock(&s->s_mutex);
3556 ceph_put_mds_session(s);
3557 mutex_lock(&mdsc->mutex);
3558 }
3559 mutex_unlock(&mdsc->mutex);
3560}
3561
3562
3563
3564/*
3565 * delayed work -- periodically trim expired leases, renew caps with mds
3566 */
3567static void schedule_delayed(struct ceph_mds_client *mdsc)
3568{
3569 int delay = 5;
3570 unsigned hz = round_jiffies_relative(HZ * delay);
3571 schedule_delayed_work(&mdsc->delayed_work, hz);
3572}
3573
3574static void delayed_work(struct work_struct *work)
3575{
3576 int i;
3577 struct ceph_mds_client *mdsc =
3578 container_of(work, struct ceph_mds_client, delayed_work.work);
3579 int renew_interval;
3580 int renew_caps;
3581
3582 dout("mdsc delayed_work\n");
afcdaea3 3583 ceph_check_delayed_caps(mdsc);
2f2dc053
SW		 3584
3585 mutex_lock(&mdsc->mutex);
3586 renew_interval = mdsc->mdsmap->m_session_timeout >> 2;
3587 renew_caps = time_after_eq(jiffies, HZ*renew_interval +
3588 mdsc->last_renew_caps);
3589 if (renew_caps)
3590 mdsc->last_renew_caps = jiffies;
3591
3592 for (i = 0; i < mdsc->max_sessions; i++) {
3593 struct ceph_mds_session *s = __ceph_lookup_mds_session(mdsc, i);
d37b1d99 3594 if (!s)
2f2dc053
SW		 3595 continue;
3596 if (s->s_state == CEPH_MDS_SESSION_CLOSING) {
3597 dout("resending session close request for mds%d\n",
3598 s->s_mds);
3599 request_close_session(mdsc, s);
3600 ceph_put_mds_session(s);
3601 continue;
3602 }
3603 if (s->s_ttl && time_after(jiffies, s->s_ttl)) {
3604 if (s->s_state == CEPH_MDS_SESSION_OPEN) {
3605 s->s_state = CEPH_MDS_SESSION_HUNG;
3606 pr_info("mds%d hung\n", s->s_mds);
3607 }
3608 }
3609 if (s->s_state < CEPH_MDS_SESSION_OPEN) {
3610 /* this mds is failed or recovering, just wait */
3611 ceph_put_mds_session(s);
3612 continue;
3613 }
3614 mutex_unlock(&mdsc->mutex);
3615
3616 mutex_lock(&s->s_mutex);
3617 if (renew_caps)
3618 send_renew_caps(mdsc, s);
3619 else
3620 ceph_con_keepalive(&s->s_con);
aab53dd9
SW		 3621 if (s->s_state == CEPH_MDS_SESSION_OPEN ||
3622 s->s_state == CEPH_MDS_SESSION_HUNG)
3d7ded4d 3623 ceph_send_cap_releases(mdsc, s);
2f2dc053
SW		 3624 mutex_unlock(&s->s_mutex);
3625 ceph_put_mds_session(s);
3626
3627 mutex_lock(&mdsc->mutex);
3628 }
3629 mutex_unlock(&mdsc->mutex);
3630
3631 schedule_delayed(mdsc);
3632}
3633
3d14c5d2 3634int ceph_mdsc_init(struct ceph_fs_client *fsc)
2f2dc053 3635
2f2dc053 3636{
3d14c5d2
YS		 3637 struct ceph_mds_client *mdsc;
3638
3639 mdsc = kzalloc(sizeof(struct ceph_mds_client), GFP_NOFS);
3640 if (!mdsc)
3641 return -ENOMEM;
3642 mdsc->fsc = fsc;
2f2dc053
SW		 3643 mutex_init(&mdsc->mutex);
3644 mdsc->mdsmap = kzalloc(sizeof(*mdsc->mdsmap), GFP_NOFS);
d37b1d99 3645 if (!mdsc->mdsmap) {
fb3101b6 3646 kfree(mdsc);
2d06eeb8 3647 return -ENOMEM;
fb3101b6 3648 }
2d06eeb8 3649
50c55aec 3650 fsc->mdsc = mdsc;
2f2dc053 3651 init_completion(&mdsc->safe_umount_waiters);
f3c60c59 3652 init_waitqueue_head(&mdsc->session_close_wq);
2f2dc053
SW		 3653 INIT_LIST_HEAD(&mdsc->waiting_for_map);
3654 mdsc->sessions = NULL;
86d8f67b 3655 atomic_set(&mdsc->num_sessions, 0);
2f2dc053
SW		 3656 mdsc->max_sessions = 0;
3657 mdsc->stopping = 0;
d557c48d 3658 atomic64_set(&mdsc->quotarealms_count, 0);
affbc19a 3659 mdsc->last_snap_seq = 0;
2f2dc053 3660 init_rwsem(&mdsc->snap_rwsem);
a105f00c 3661 mdsc->snap_realms = RB_ROOT;
2f2dc053
SW		 3662 INIT_LIST_HEAD(&mdsc->snap_empty);
3663 spin_lock_init(&mdsc->snap_empty_lock);
3664 mdsc->last_tid = 0;
e8a7b8b1 3665 mdsc->oldest_tid = 0;
44ca18f2 3666 mdsc->request_tree = RB_ROOT;
2f2dc053
SW		 3667 INIT_DELAYED_WORK(&mdsc->delayed_work, delayed_work);
3668 mdsc->last_renew_caps = jiffies;
3669 INIT_LIST_HEAD(&mdsc->cap_delay_list);
3670 spin_lock_init(&mdsc->cap_delay_lock);
3671 INIT_LIST_HEAD(&mdsc->snap_flush_list);
3672 spin_lock_init(&mdsc->snap_flush_lock);
553adfd9 3673 mdsc->last_cap_flush_tid = 1;
e4500b5e 3674 INIT_LIST_HEAD(&mdsc->cap_flush_list);
2f2dc053 3675 INIT_LIST_HEAD(&mdsc->cap_dirty);
db354052 3676 INIT_LIST_HEAD(&mdsc->cap_dirty_migrating);
2f2dc053
SW		 3677 mdsc->num_cap_flushing = 0;
3678 spin_lock_init(&mdsc->cap_dirty_lock);
3679 init_waitqueue_head(&mdsc->cap_flushing_wq);
3680 spin_lock_init(&mdsc->dentry_lru_lock);
3681 INIT_LIST_HEAD(&mdsc->dentry_lru);
2d06eeb8 3682
37151668 3683 ceph_caps_init(mdsc);
3d14c5d2 3684 ceph_adjust_min_caps(mdsc, fsc->min_caps);
37151668 3685
10183a69
YZ		 3686 init_rwsem(&mdsc->pool_perm_rwsem);
3687 mdsc->pool_perm_tree = RB_ROOT;
3688
dfeb84d4
YZ		 3689 strscpy(mdsc->nodename, utsname()->nodename,
3690 sizeof(mdsc->nodename));
5f44f142 3691 return 0;
2f2dc053
SW		 3692}
3693
3694/*
3695 * Wait for safe replies on open mds requests. If we time out, drop
3696 * all requests from the tree to avoid dangling dentry refs.
3697 */
3698static void wait_requests(struct ceph_mds_client *mdsc)
3699{
a319bf56 3700 struct ceph_options *opts = mdsc->fsc->client->options;
2f2dc053 3701 struct ceph_mds_request *req;
2f2dc053
SW		 3702
3703 mutex_lock(&mdsc->mutex);
44ca18f2 3704 if (__get_oldest_req(mdsc)) {
2f2dc053 3705 mutex_unlock(&mdsc->mutex);
44ca18f2 3706
2f2dc053
SW		 3707 dout("wait_requests waiting for requests\n");
3708 wait_for_completion_timeout(&mdsc->safe_umount_waiters,
a319bf56 3709 ceph_timeout_jiffies(opts->mount_timeout));
2f2dc053
SW		 3710
3711 /* tear down remaining requests */
44ca18f2
SW		 3712 mutex_lock(&mdsc->mutex);
3713 while ((req = __get_oldest_req(mdsc))) {
2f2dc053
SW		 3714 dout("wait_requests timed out on tid %llu\n",
3715 req->r_tid);
44ca18f2 3716 __unregister_request(mdsc, req);
2f2dc053
SW		 3717 }
3718 }
3719 mutex_unlock(&mdsc->mutex);
3720 dout("wait_requests done\n");
3721}
3722
3723/*
3724 * called before mount is ro, and before dentries are torn down.
3725 * (hmm, does this still race with new lookups?)
3726 */
3727void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc)
3728{
3729 dout("pre_umount\n");
3730 mdsc->stopping = 1;
3731
7aac453a 3732 lock_unlock_sessions(mdsc);
afcdaea3 3733 ceph_flush_dirty_caps(mdsc);
2f2dc053 3734 wait_requests(mdsc);
17c688c3
SW		 3735
3736 /*
3737 * wait for reply handlers to drop their request refs and
3738 * their inode/dcache refs
3739 */
3740 ceph_msgr_flush();
2f2dc053
SW		 3741}
3742
3743/*
3744 * wait for all write mds requests to flush.
3745 */
3746static void wait_unsafe_requests(struct ceph_mds_client *mdsc, u64 want_tid)
3747{
80fc7314 3748 struct ceph_mds_request *req = NULL, *nextreq;
44ca18f2 3749 struct rb_node *n;
2f2dc053
SW		 3750
3751 mutex_lock(&mdsc->mutex);
3752 dout("wait_unsafe_requests want %lld\n", want_tid);
80fc7314 3753restart:
44ca18f2
SW		 3754 req = __get_oldest_req(mdsc);
3755 while (req && req->r_tid <= want_tid) {
80fc7314
SW		 3756 /* find next request */
3757 n = rb_next(&req->r_node);
3758 if (n)
3759 nextreq = rb_entry(n, struct ceph_mds_request, r_node);
3760 else
3761 nextreq = NULL;
e8a7b8b1
YZ		 3762 if (req->r_op != CEPH_MDS_OP_SETFILELOCK &&
3763 (req->r_op & CEPH_MDS_OP_WRITE)) {
44ca18f2
SW		 3764 /* write op */
3765 ceph_mdsc_get_request(req);
80fc7314
SW		 3766 if (nextreq)
3767 ceph_mdsc_get_request(nextreq);
44ca18f2
SW		 3768 mutex_unlock(&mdsc->mutex);
3769 dout("wait_unsafe_requests wait on %llu (want %llu)\n",
3770 req->r_tid, want_tid);
3771 wait_for_completion(&req->r_safe_completion);
3772 mutex_lock(&mdsc->mutex);
44ca18f2 3773 ceph_mdsc_put_request(req);
80fc7314
SW		 3774 if (!nextreq)
3775 break; /* next dne before, so we're done! */
3776 if (RB_EMPTY_NODE(&nextreq->r_node)) {
3777 /* next request was removed from tree */
3778 ceph_mdsc_put_request(nextreq);
3779 goto restart;
3780 }
3781 ceph_mdsc_put_request(nextreq); /* won't go away */
44ca18f2 3782 }
80fc7314 3783 req = nextreq;
2f2dc053
SW		 3784 }
3785 mutex_unlock(&mdsc->mutex);
3786 dout("wait_unsafe_requests done\n");
3787}
3788
3789void ceph_mdsc_sync(struct ceph_mds_client *mdsc)
3790{
0e294387 3791 u64 want_tid, want_flush;
2f2dc053 3792
52953d55 3793 if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
56b7cf95
SW		 3794 return;
3795
2f2dc053
SW		 3796 dout("sync\n");
3797 mutex_lock(&mdsc->mutex);
3798 want_tid = mdsc->last_tid;
2f2dc053 3799 mutex_unlock(&mdsc->mutex);
2f2dc053 3800
afcdaea3 3801 ceph_flush_dirty_caps(mdsc);
d3383a8e 3802 spin_lock(&mdsc->cap_dirty_lock);
8310b089 3803 want_flush = mdsc->last_cap_flush_tid;
c8799fc4
YZ		 3804 if (!list_empty(&mdsc->cap_flush_list)) {
3805 struct ceph_cap_flush *cf =
3806 list_last_entry(&mdsc->cap_flush_list,
3807 struct ceph_cap_flush, g_list);
3808 cf->wake = true;
3809 }
d3383a8e
YZ		 3810 spin_unlock(&mdsc->cap_dirty_lock);
3811
0e294387
YZ		 3812 dout("sync want tid %lld flush_seq %lld\n",
3813 want_tid, want_flush);
2f2dc053
SW		 3814
3815 wait_unsafe_requests(mdsc, want_tid);
0e294387 3816 wait_caps_flush(mdsc, want_flush);
2f2dc053
SW		 3817}
3818
f3c60c59
SW		 3819/*
3820 * true if all sessions are closed, or we force unmount
3821 */
fcff415c 3822static bool done_closing_sessions(struct ceph_mds_client *mdsc, int skipped)
f3c60c59 3823{
52953d55 3824 if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
f3c60c59 3825 return true;
fcff415c 3826 return atomic_read(&mdsc->num_sessions) <= skipped;
f3c60c59 3827}
2f2dc053
SW		 3828
3829/*
3830 * called after sb is ro.
3831 */
3832void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc)
3833{
a319bf56 3834 struct ceph_options *opts = mdsc->fsc->client->options;
2f2dc053
SW		 3835 struct ceph_mds_session *session;
3836 int i;
fcff415c 3837 int skipped = 0;
2f2dc053
SW		 3838
3839 dout("close_sessions\n");
3840
2f2dc053 3841 /* close sessions */
f3c60c59
SW		 3842 mutex_lock(&mdsc->mutex);
3843 for (i = 0; i < mdsc->max_sessions; i++) {
3844 session = __ceph_lookup_mds_session(mdsc, i);
3845 if (!session)
3846 continue;
2f2dc053 3847 mutex_unlock(&mdsc->mutex);
f3c60c59 3848 mutex_lock(&session->s_mutex);
fcff415c
YZ		 3849 if (__close_session(mdsc, session) <= 0)
3850 skipped++;
f3c60c59
SW		 3851 mutex_unlock(&session->s_mutex);
3852 ceph_put_mds_session(session);
2f2dc053
SW		 3853 mutex_lock(&mdsc->mutex);
3854 }
f3c60c59
SW		 3855 mutex_unlock(&mdsc->mutex);
3856
3857 dout("waiting for sessions to close\n");
fcff415c
YZ		 3858 wait_event_timeout(mdsc->session_close_wq,
3859 done_closing_sessions(mdsc, skipped),
a319bf56 3860 ceph_timeout_jiffies(opts->mount_timeout));
2f2dc053
SW		 3861
3862 /* tear down remaining sessions */
f3c60c59 3863 mutex_lock(&mdsc->mutex);
2f2dc053
SW		 3864 for (i = 0; i < mdsc->max_sessions; i++) {
3865 if (mdsc->sessions[i]) {
3866 session = get_session(mdsc->sessions[i]);
2600d2dd 3867 __unregister_session(mdsc, session);
2f2dc053
SW		 3868 mutex_unlock(&mdsc->mutex);
3869 mutex_lock(&session->s_mutex);
3870 remove_session_caps(session);
3871 mutex_unlock(&session->s_mutex);
3872 ceph_put_mds_session(session);
3873 mutex_lock(&mdsc->mutex);
3874 }
3875 }
2f2dc053 3876 WARN_ON(!list_empty(&mdsc->cap_delay_list));
2f2dc053
SW		 3877 mutex_unlock(&mdsc->mutex);
3878
3879 ceph_cleanup_empty_realms(mdsc);
3880
3881 cancel_delayed_work_sync(&mdsc->delayed_work); /* cancel timer */
3882
3883 dout("stopped\n");
3884}
3885
48fec5d0
YZ		 3886void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc)
3887{
3888 struct ceph_mds_session *session;
3889 int mds;
3890
3891 dout("force umount\n");
3892
3893 mutex_lock(&mdsc->mutex);
3894 for (mds = 0; mds < mdsc->max_sessions; mds++) {
3895 session = __ceph_lookup_mds_session(mdsc, mds);
3896 if (!session)
3897 continue;
3898 mutex_unlock(&mdsc->mutex);
3899 mutex_lock(&session->s_mutex);
3900 __close_session(mdsc, session);
3901 if (session->s_state == CEPH_MDS_SESSION_CLOSING) {
3902 cleanup_session_requests(mdsc, session);
3903 remove_session_caps(session);
3904 }
3905 mutex_unlock(&session->s_mutex);
3906 ceph_put_mds_session(session);
3907 mutex_lock(&mdsc->mutex);
3908 kick_requests(mdsc, mds);
3909 }
3910 __wake_requests(mdsc, &mdsc->waiting_for_map);
3911 mutex_unlock(&mdsc->mutex);
3912}
3913
3d14c5d2 3914static void ceph_mdsc_stop(struct ceph_mds_client *mdsc)
2f2dc053
SW		 3915{
3916 dout("stop\n");
3917 cancel_delayed_work_sync(&mdsc->delayed_work); /* cancel timer */
3918 if (mdsc->mdsmap)
3919 ceph_mdsmap_destroy(mdsc->mdsmap);
3920 kfree(mdsc->sessions);
37151668 3921 ceph_caps_finalize(mdsc);
10183a69 3922 ceph_pool_perm_destroy(mdsc);
2f2dc053
SW		 3923}
3924
3d14c5d2
YS		 3925void ceph_mdsc_destroy(struct ceph_fs_client *fsc)
3926{
3927 struct ceph_mds_client *mdsc = fsc->mdsc;
ef550f6f 3928 dout("mdsc_destroy %p\n", mdsc);
ef550f6f 3929
50c55aec
CX		 3930 if (!mdsc)
3931 return;
3932
ef550f6f
SW		 3933 /* flush out any connection work with references to us */
3934 ceph_msgr_flush();
3935
62a65f36
YZ		 3936 ceph_mdsc_stop(mdsc);
3937
3d14c5d2
YS		 3938 fsc->mdsc = NULL;
3939 kfree(mdsc);
ef550f6f 3940 dout("mdsc_destroy %p done\n", mdsc);
3d14c5d2
YS		 3941}
3942
430afbad
YZ		 3943void ceph_mdsc_handle_fsmap(struct ceph_mds_client *mdsc, struct ceph_msg *msg)
3944{
3945 struct ceph_fs_client *fsc = mdsc->fsc;
3946 const char *mds_namespace = fsc->mount_options->mds_namespace;
3947 void *p = msg->front.iov_base;
3948 void *end = p + msg->front.iov_len;
3949 u32 epoch;
3950 u32 map_len;
3951 u32 num_fs;
3952 u32 mount_fscid = (u32)-1;
3953 u8 struct_v, struct_cv;
3954 int err = -EINVAL;
3955
3956 ceph_decode_need(&p, end, sizeof(u32), bad);
3957 epoch = ceph_decode_32(&p);
3958
3959 dout("handle_fsmap epoch %u\n", epoch);
3960
3961 ceph_decode_need(&p, end, 2 + sizeof(u32), bad);
3962 struct_v = ceph_decode_8(&p);
3963 struct_cv = ceph_decode_8(&p);
3964 map_len = ceph_decode_32(&p);
3965
3966 ceph_decode_need(&p, end, sizeof(u32) * 3, bad);
3967 p += sizeof(u32) * 2; /* skip epoch and legacy_client_fscid */
3968
3969 num_fs = ceph_decode_32(&p);
3970 while (num_fs-- > 0) {
3971 void *info_p, *info_end;
3972 u32 info_len;
3973 u8 info_v, info_cv;
3974 u32 fscid, namelen;
3975
3976 ceph_decode_need(&p, end, 2 + sizeof(u32), bad);
3977 info_v = ceph_decode_8(&p);
3978 info_cv = ceph_decode_8(&p);
3979 info_len = ceph_decode_32(&p);
3980 ceph_decode_need(&p, end, info_len, bad);
3981 info_p = p;
3982 info_end = p + info_len;
3983 p = info_end;
3984
3985 ceph_decode_need(&info_p, info_end, sizeof(u32) * 2, bad);
3986 fscid = ceph_decode_32(&info_p);
3987 namelen = ceph_decode_32(&info_p);
3988 ceph_decode_need(&info_p, info_end, namelen, bad);
3989
3990 if (mds_namespace &&
3991 strlen(mds_namespace) == namelen &&
3992 !strncmp(mds_namespace, (char *)info_p, namelen)) {
3993 mount_fscid = fscid;
3994 break;
3995 }
3996 }
3997
3998 ceph_monc_got_map(&fsc->client->monc, CEPH_SUB_FSMAP, epoch);
3999 if (mount_fscid != (u32)-1) {
4000 fsc->client->monc.fs_cluster_id = mount_fscid;
4001 ceph_monc_want_map(&fsc->client->monc, CEPH_SUB_MDSMAP,
4002 0, true);
4003 ceph_monc_renew_subs(&fsc->client->monc);
4004 } else {
4005 err = -ENOENT;
4006 goto err_out;
4007 }
4008 return;
76bd6ec4 4009
430afbad
YZ		 4010bad:
4011 pr_err("error decoding fsmap\n");
4012err_out:
4013 mutex_lock(&mdsc->mutex);
76bd6ec4 4014 mdsc->mdsmap_err = err;
430afbad
YZ		 4015 __wake_requests(mdsc, &mdsc->waiting_for_map);
4016 mutex_unlock(&mdsc->mutex);
430afbad 4017}
2f2dc053
SW		 4018
4019/*
4020 * handle mds map update.
4021 */
430afbad 4022void ceph_mdsc_handle_mdsmap(struct ceph_mds_client *mdsc, struct ceph_msg *msg)
2f2dc053
SW		 4023{
4024 u32 epoch;
4025 u32 maplen;
4026 void *p = msg->front.iov_base;
4027 void *end = p + msg->front.iov_len;
4028 struct ceph_mdsmap *newmap, *oldmap;
4029 struct ceph_fsid fsid;
4030 int err = -EINVAL;
4031
4032 ceph_decode_need(&p, end, sizeof(fsid)+2*sizeof(u32), bad);
4033 ceph_decode_copy(&p, &fsid, sizeof(fsid));
3d14c5d2 4034 if (ceph_check_fsid(mdsc->fsc->client, &fsid) < 0)
0743304d 4035 return;
c89136ea
SW		 4036 epoch = ceph_decode_32(&p);
4037 maplen = ceph_decode_32(&p);
2f2dc053
SW		 4038 dout("handle_map epoch %u len %d\n", epoch, (int)maplen);
4039
4040 /* do we need it? */
2f2dc053
SW		 4041 mutex_lock(&mdsc->mutex);
4042 if (mdsc->mdsmap && epoch <= mdsc->mdsmap->m_epoch) {
4043 dout("handle_map epoch %u <= our %u\n",
4044 epoch, mdsc->mdsmap->m_epoch);
4045 mutex_unlock(&mdsc->mutex);
4046 return;
4047 }
4048
4049 newmap = ceph_mdsmap_decode(&p, end);
4050 if (IS_ERR(newmap)) {
4051 err = PTR_ERR(newmap);
4052 goto bad_unlock;
4053 }
4054
4055 /* swap into place */
4056 if (mdsc->mdsmap) {
4057 oldmap = mdsc->mdsmap;
4058 mdsc->mdsmap = newmap;
4059 check_new_map(mdsc, newmap, oldmap);
4060 ceph_mdsmap_destroy(oldmap);
4061 } else {
4062 mdsc->mdsmap = newmap; /* first mds map */
4063 }
719784ba
CX		 4064 mdsc->fsc->max_file_size = min((loff_t)mdsc->mdsmap->m_max_file_size,
4065 MAX_LFS_FILESIZE);
2f2dc053
SW		 4066
4067 __wake_requests(mdsc, &mdsc->waiting_for_map);
82dcabad
ID		 4068 ceph_monc_got_map(&mdsc->fsc->client->monc, CEPH_SUB_MDSMAP,
4069 mdsc->mdsmap->m_epoch);
2f2dc053
SW		 4070
4071 mutex_unlock(&mdsc->mutex);
4072 schedule_delayed(mdsc);
4073 return;
4074
4075bad_unlock:
4076 mutex_unlock(&mdsc->mutex);
4077bad:
4078 pr_err("error decoding mdsmap %d\n", err);
4079 return;
4080}
4081
4082static struct ceph_connection *con_get(struct ceph_connection *con)
4083{
4084 struct ceph_mds_session *s = con->private;
4085
4086 if (get_session(s)) {
3997c01d 4087 dout("mdsc con_get %p ok (%d)\n", s, refcount_read(&s->s_ref));
2f2dc053
SW		 4088 return con;
4089 }
4090 dout("mdsc con_get %p FAIL\n", s);
4091 return NULL;
4092}
4093
4094static void con_put(struct ceph_connection *con)
4095{
4096 struct ceph_mds_session *s = con->private;
4097
3997c01d 4098 dout("mdsc con_put %p (%d)\n", s, refcount_read(&s->s_ref) - 1);
2f2dc053
SW		 4099 ceph_put_mds_session(s);
4100}
4101
4102/*
4103 * if the client is unresponsive for long enough, the mds will kill
4104 * the session entirely.
4105 */
4106static void peer_reset(struct ceph_connection *con)
4107{
4108 struct ceph_mds_session *s = con->private;
7e70f0ed 4109 struct ceph_mds_client *mdsc = s->s_mdsc;
2f2dc053 4110
f3ae1b97 4111 pr_warn("mds%d closed our session\n", s->s_mds);
7e70f0ed 4112 send_mds_reconnect(mdsc, s);
2f2dc053
SW		 4113}
4114
4115static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
4116{
4117 struct ceph_mds_session *s = con->private;
4118 struct ceph_mds_client *mdsc = s->s_mdsc;
4119 int type = le16_to_cpu(msg->hdr.type);
4120
2600d2dd
SW		 4121 mutex_lock(&mdsc->mutex);
4122 if (__verify_registered_session(mdsc, s) < 0) {
4123 mutex_unlock(&mdsc->mutex);
4124 goto out;
4125 }
4126 mutex_unlock(&mdsc->mutex);
4127
2f2dc053
SW		 4128 switch (type) {
4129 case CEPH_MSG_MDS_MAP:
430afbad
YZ		 4130 ceph_mdsc_handle_mdsmap(mdsc, msg);
4131 break;
4132 case CEPH_MSG_FS_MAP_USER:
4133 ceph_mdsc_handle_fsmap(mdsc, msg);
2f2dc053
SW		 4134 break;
4135 case CEPH_MSG_CLIENT_SESSION:
4136 handle_session(s, msg);
4137 break;
4138 case CEPH_MSG_CLIENT_REPLY:
4139 handle_reply(s, msg);
4140 break;
4141 case CEPH_MSG_CLIENT_REQUEST_FORWARD:
2600d2dd 4142 handle_forward(mdsc, s, msg);
2f2dc053
SW		 4143 break;
4144 case CEPH_MSG_CLIENT_CAPS:
4145 ceph_handle_caps(s, msg);
4146 break;
4147 case CEPH_MSG_CLIENT_SNAP:
2600d2dd 4148 ceph_handle_snap(mdsc, s, msg);
2f2dc053
SW		 4149 break;
4150 case CEPH_MSG_CLIENT_LEASE:
2600d2dd 4151 handle_lease(mdsc, s, msg);
2f2dc053 4152 break;
fb18a575
LH		 4153 case CEPH_MSG_CLIENT_QUOTA:
4154 ceph_handle_quota(mdsc, s, msg);
4155 break;
2f2dc053
SW		 4156
4157 default:
4158 pr_err("received unknown message type %d %s\n", type,
4159 ceph_msg_type_name(type));
4160 }
2600d2dd 4161out:
2f2dc053
SW		 4162 ceph_msg_put(msg);
4163}
4164
4e7a5dcd
SW		 4165/*
4166 * authentication
4167 */
a3530df3
AE		 4168
4169/*
4170 * Note: returned pointer is the address of a structure that's
4171 * managed separately. Caller must *not* attempt to free it.
4172 */
4173static struct ceph_auth_handshake *get_authorizer(struct ceph_connection *con,
8f43fb53 4174 int *proto, int force_new)
4e7a5dcd
SW		 4175{
4176 struct ceph_mds_session *s = con->private;
4177 struct ceph_mds_client *mdsc = s->s_mdsc;
3d14c5d2 4178 struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
74f1869f 4179 struct ceph_auth_handshake *auth = &s->s_auth;
4e7a5dcd 4180
74f1869f 4181 if (force_new && auth->authorizer) {
6c1ea260 4182 ceph_auth_destroy_authorizer(auth->authorizer);
74f1869f 4183 auth->authorizer = NULL;
4e7a5dcd 4184 }
27859f97
SW		 4185 if (!auth->authorizer) {
4186 int ret = ceph_auth_create_authorizer(ac, CEPH_ENTITY_TYPE_MDS,
4187 auth);
0bed9b5c
SW		 4188 if (ret)
4189 return ERR_PTR(ret);
27859f97
SW		 4190 } else {
4191 int ret = ceph_auth_update_authorizer(ac, CEPH_ENTITY_TYPE_MDS,
4192 auth);
a255651d 4193 if (ret)
a3530df3 4194 return ERR_PTR(ret);
4e7a5dcd 4195 }
4e7a5dcd 4196 *proto = ac->protocol;
74f1869f 4197
a3530df3 4198 return auth;
4e7a5dcd
SW		 4199}
4200
6daca13d
ID		 4201static int add_authorizer_challenge(struct ceph_connection *con,
4202 void *challenge_buf, int challenge_buf_len)
4203{
4204 struct ceph_mds_session *s = con->private;
4205 struct ceph_mds_client *mdsc = s->s_mdsc;
4206 struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
4207
4208 return ceph_auth_add_authorizer_challenge(ac, s->s_auth.authorizer,
4209 challenge_buf, challenge_buf_len);
4210}
4e7a5dcd 4211
0dde5848 4212static int verify_authorizer_reply(struct ceph_connection *con)
4e7a5dcd
SW		 4213{
4214 struct ceph_mds_session *s = con->private;
4215 struct ceph_mds_client *mdsc = s->s_mdsc;
3d14c5d2 4216 struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
4e7a5dcd 4217
0dde5848 4218 return ceph_auth_verify_authorizer_reply(ac, s->s_auth.authorizer);
4e7a5dcd
SW		 4219}
4220
9bd2e6f8
SW		 4221static int invalidate_authorizer(struct ceph_connection *con)
4222{
4223 struct ceph_mds_session *s = con->private;
4224 struct ceph_mds_client *mdsc = s->s_mdsc;
3d14c5d2 4225 struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
9bd2e6f8 4226
27859f97 4227 ceph_auth_invalidate_authorizer(ac, CEPH_ENTITY_TYPE_MDS);
9bd2e6f8 4228
3d14c5d2 4229 return ceph_monc_validate_auth(&mdsc->fsc->client->monc);
9bd2e6f8
SW		 4230}
4231
53ded495
AE		 4232static struct ceph_msg *mds_alloc_msg(struct ceph_connection *con,
4233 struct ceph_msg_header *hdr, int *skip)
4234{
4235 struct ceph_msg *msg;
4236 int type = (int) le16_to_cpu(hdr->type);
4237 int front_len = (int) le32_to_cpu(hdr->front_len);
4238
4239 if (con->in_msg)
4240 return con->in_msg;
4241
4242 *skip = 0;
4243 msg = ceph_msg_new(type, front_len, GFP_NOFS, false);
4244 if (!msg) {
4245 pr_err("unable to allocate msg type %d len %d\n",
4246 type, front_len);
4247 return NULL;
4248 }
53ded495
AE		 4249
4250 return msg;
4251}
4252
79dbd1ba 4253static int mds_sign_message(struct ceph_msg *msg)
33d07337 4254{
79dbd1ba 4255 struct ceph_mds_session *s = msg->con->private;
33d07337 4256 struct ceph_auth_handshake *auth = &s->s_auth;
79dbd1ba 4257
33d07337
YZ		 4258 return ceph_auth_sign_message(auth, msg);
4259}
4260
79dbd1ba 4261static int mds_check_message_signature(struct ceph_msg *msg)
33d07337 4262{
79dbd1ba 4263 struct ceph_mds_session *s = msg->con->private;
33d07337 4264 struct ceph_auth_handshake *auth = &s->s_auth;
79dbd1ba 4265
33d07337
YZ		 4266 return ceph_auth_check_message_signature(auth, msg);
4267}
4268
9e32789f 4269static const struct ceph_connection_operations mds_con_ops = {
2f2dc053
SW		 4270 .get = con_get,
4271 .put = con_put,
4272 .dispatch = dispatch,
4e7a5dcd 4273 .get_authorizer = get_authorizer,
6daca13d 4274 .add_authorizer_challenge = add_authorizer_challenge,
4e7a5dcd 4275 .verify_authorizer_reply = verify_authorizer_reply,
9bd2e6f8 4276 .invalidate_authorizer = invalidate_authorizer,
2f2dc053 4277 .peer_reset = peer_reset,
53ded495 4278 .alloc_msg = mds_alloc_msg,
79dbd1ba
ID		 4279 .sign_message = mds_sign_message,
4280 .check_message_signature = mds_check_message_signature,
2f2dc053
SW		 4281};
4282
2f2dc053 4283/* eof */
Ubuntu Hirsute Linux kernel mirror