xref: /linux/fs/ceph/mds_client.c (revision 55d0969c451159cff86949b38c39171cab962069)
1 // SPDX-License-Identifier: GPL-2.0
2 #include <linux/ceph/ceph_debug.h>
3 
4 #include <linux/fs.h>
5 #include <linux/wait.h>
6 #include <linux/slab.h>
7 #include <linux/gfp.h>
8 #include <linux/sched.h>
9 #include <linux/debugfs.h>
10 #include <linux/seq_file.h>
11 #include <linux/ratelimit.h>
12 #include <linux/bits.h>
13 #include <linux/ktime.h>
14 #include <linux/bitmap.h>
15 #include <linux/mnt_idmapping.h>
16 
17 #include "super.h"
18 #include "mds_client.h"
19 #include "crypto.h"
20 
21 #include <linux/ceph/ceph_features.h>
22 #include <linux/ceph/messenger.h>
23 #include <linux/ceph/decode.h>
24 #include <linux/ceph/pagelist.h>
25 #include <linux/ceph/auth.h>
26 #include <linux/ceph/debugfs.h>
27 
28 #define RECONNECT_MAX_SIZE (INT_MAX - PAGE_SIZE)
29 
30 /*
31  * A cluster of MDS (metadata server) daemons is responsible for
32  * managing the file system namespace (the directory hierarchy and
33  * inodes) and for coordinating shared access to storage.  Metadata is
34  * partitioning hierarchically across a number of servers, and that
35  * partition varies over time as the cluster adjusts the distribution
36  * in order to balance load.
37  *
38  * The MDS client is primarily responsible to managing synchronous
39  * metadata requests for operations like open, unlink, and so forth.
40  * If there is a MDS failure, we find out about it when we (possibly
41  * request and) receive a new MDS map, and can resubmit affected
42  * requests.
43  *
44  * For the most part, though, we take advantage of a lossless
45  * communications channel to the MDS, and do not need to worry about
46  * timing out or resubmitting requests.
47  *
48  * We maintain a stateful "session" with each MDS we interact with.
49  * Within each session, we sent periodic heartbeat messages to ensure
50  * any capabilities or leases we have been issues remain valid.  If
51  * the session times out and goes stale, our leases and capabilities
52  * are no longer valid.
53  */
54 
55 struct ceph_reconnect_state {
56 	struct ceph_mds_session *session;
57 	int nr_caps, nr_realms;
58 	struct ceph_pagelist *pagelist;
59 	unsigned msg_version;
60 	bool allow_multi;
61 };
62 
63 static void __wake_requests(struct ceph_mds_client *mdsc,
64 			    struct list_head *head);
65 static void ceph_cap_release_work(struct work_struct *work);
66 static void ceph_cap_reclaim_work(struct work_struct *work);
67 
68 static const struct ceph_connection_operations mds_con_ops;
69 
70 
71 /*
72  * mds reply parsing
73  */
74 
75 static int parse_reply_info_quota(void **p, void *end,
76 				  struct ceph_mds_reply_info_in *info)
77 {
78 	u8 struct_v, struct_compat;
79 	u32 struct_len;
80 
81 	ceph_decode_8_safe(p, end, struct_v, bad);
82 	ceph_decode_8_safe(p, end, struct_compat, bad);
83 	/* struct_v is expected to be >= 1. we only
84 	 * understand encoding with struct_compat == 1. */
85 	if (!struct_v || struct_compat != 1)
86 		goto bad;
87 	ceph_decode_32_safe(p, end, struct_len, bad);
88 	ceph_decode_need(p, end, struct_len, bad);
89 	end = *p + struct_len;
90 	ceph_decode_64_safe(p, end, info->max_bytes, bad);
91 	ceph_decode_64_safe(p, end, info->max_files, bad);
92 	*p = end;
93 	return 0;
94 bad:
95 	return -EIO;
96 }
97 
98 /*
99  * parse individual inode info
100  */
101 static int parse_reply_info_in(void **p, void *end,
102 			       struct ceph_mds_reply_info_in *info,
103 			       u64 features)
104 {
105 	int err = 0;
106 	u8 struct_v = 0;
107 
108 	if (features == (u64)-1) {
109 		u32 struct_len;
110 		u8 struct_compat;
111 		ceph_decode_8_safe(p, end, struct_v, bad);
112 		ceph_decode_8_safe(p, end, struct_compat, bad);
113 		/* struct_v is expected to be >= 1. we only understand
114 		 * encoding with struct_compat == 1. */
115 		if (!struct_v || struct_compat != 1)
116 			goto bad;
117 		ceph_decode_32_safe(p, end, struct_len, bad);
118 		ceph_decode_need(p, end, struct_len, bad);
119 		end = *p + struct_len;
120 	}
121 
122 	ceph_decode_need(p, end, sizeof(struct ceph_mds_reply_inode), bad);
123 	info->in = *p;
124 	*p += sizeof(struct ceph_mds_reply_inode) +
125 		sizeof(*info->in->fragtree.splits) *
126 		le32_to_cpu(info->in->fragtree.nsplits);
127 
128 	ceph_decode_32_safe(p, end, info->symlink_len, bad);
129 	ceph_decode_need(p, end, info->symlink_len, bad);
130 	info->symlink = *p;
131 	*p += info->symlink_len;
132 
133 	ceph_decode_copy_safe(p, end, &info->dir_layout,
134 			      sizeof(info->dir_layout), bad);
135 	ceph_decode_32_safe(p, end, info->xattr_len, bad);
136 	ceph_decode_need(p, end, info->xattr_len, bad);
137 	info->xattr_data = *p;
138 	*p += info->xattr_len;
139 
140 	if (features == (u64)-1) {
141 		/* inline data */
142 		ceph_decode_64_safe(p, end, info->inline_version, bad);
143 		ceph_decode_32_safe(p, end, info->inline_len, bad);
144 		ceph_decode_need(p, end, info->inline_len, bad);
145 		info->inline_data = *p;
146 		*p += info->inline_len;
147 		/* quota */
148 		err = parse_reply_info_quota(p, end, info);
149 		if (err < 0)
150 			goto out_bad;
151 		/* pool namespace */
152 		ceph_decode_32_safe(p, end, info->pool_ns_len, bad);
153 		if (info->pool_ns_len > 0) {
154 			ceph_decode_need(p, end, info->pool_ns_len, bad);
155 			info->pool_ns_data = *p;
156 			*p += info->pool_ns_len;
157 		}
158 
159 		/* btime */
160 		ceph_decode_need(p, end, sizeof(info->btime), bad);
161 		ceph_decode_copy(p, &info->btime, sizeof(info->btime));
162 
163 		/* change attribute */
164 		ceph_decode_64_safe(p, end, info->change_attr, bad);
165 
166 		/* dir pin */
167 		if (struct_v >= 2) {
168 			ceph_decode_32_safe(p, end, info->dir_pin, bad);
169 		} else {
170 			info->dir_pin = -ENODATA;
171 		}
172 
173 		/* snapshot birth time, remains zero for v<=2 */
174 		if (struct_v >= 3) {
175 			ceph_decode_need(p, end, sizeof(info->snap_btime), bad);
176 			ceph_decode_copy(p, &info->snap_btime,
177 					 sizeof(info->snap_btime));
178 		} else {
179 			memset(&info->snap_btime, 0, sizeof(info->snap_btime));
180 		}
181 
182 		/* snapshot count, remains zero for v<=3 */
183 		if (struct_v >= 4) {
184 			ceph_decode_64_safe(p, end, info->rsnaps, bad);
185 		} else {
186 			info->rsnaps = 0;
187 		}
188 
189 		if (struct_v >= 5) {
190 			u32 alen;
191 
192 			ceph_decode_32_safe(p, end, alen, bad);
193 
194 			while (alen--) {
195 				u32 len;
196 
197 				/* key */
198 				ceph_decode_32_safe(p, end, len, bad);
199 				ceph_decode_skip_n(p, end, len, bad);
200 				/* value */
201 				ceph_decode_32_safe(p, end, len, bad);
202 				ceph_decode_skip_n(p, end, len, bad);
203 			}
204 		}
205 
206 		/* fscrypt flag -- ignore */
207 		if (struct_v >= 6)
208 			ceph_decode_skip_8(p, end, bad);
209 
210 		info->fscrypt_auth = NULL;
211 		info->fscrypt_auth_len = 0;
212 		info->fscrypt_file = NULL;
213 		info->fscrypt_file_len = 0;
214 		if (struct_v >= 7) {
215 			ceph_decode_32_safe(p, end, info->fscrypt_auth_len, bad);
216 			if (info->fscrypt_auth_len) {
217 				info->fscrypt_auth = kmalloc(info->fscrypt_auth_len,
218 							     GFP_KERNEL);
219 				if (!info->fscrypt_auth)
220 					return -ENOMEM;
221 				ceph_decode_copy_safe(p, end, info->fscrypt_auth,
222 						      info->fscrypt_auth_len, bad);
223 			}
224 			ceph_decode_32_safe(p, end, info->fscrypt_file_len, bad);
225 			if (info->fscrypt_file_len) {
226 				info->fscrypt_file = kmalloc(info->fscrypt_file_len,
227 							     GFP_KERNEL);
228 				if (!info->fscrypt_file)
229 					return -ENOMEM;
230 				ceph_decode_copy_safe(p, end, info->fscrypt_file,
231 						      info->fscrypt_file_len, bad);
232 			}
233 		}
234 		*p = end;
235 	} else {
236 		/* legacy (unversioned) struct */
237 		if (features & CEPH_FEATURE_MDS_INLINE_DATA) {
238 			ceph_decode_64_safe(p, end, info->inline_version, bad);
239 			ceph_decode_32_safe(p, end, info->inline_len, bad);
240 			ceph_decode_need(p, end, info->inline_len, bad);
241 			info->inline_data = *p;
242 			*p += info->inline_len;
243 		} else
244 			info->inline_version = CEPH_INLINE_NONE;
245 
246 		if (features & CEPH_FEATURE_MDS_QUOTA) {
247 			err = parse_reply_info_quota(p, end, info);
248 			if (err < 0)
249 				goto out_bad;
250 		} else {
251 			info->max_bytes = 0;
252 			info->max_files = 0;
253 		}
254 
255 		info->pool_ns_len = 0;
256 		info->pool_ns_data = NULL;
257 		if (features & CEPH_FEATURE_FS_FILE_LAYOUT_V2) {
258 			ceph_decode_32_safe(p, end, info->pool_ns_len, bad);
259 			if (info->pool_ns_len > 0) {
260 				ceph_decode_need(p, end, info->pool_ns_len, bad);
261 				info->pool_ns_data = *p;
262 				*p += info->pool_ns_len;
263 			}
264 		}
265 
266 		if (features & CEPH_FEATURE_FS_BTIME) {
267 			ceph_decode_need(p, end, sizeof(info->btime), bad);
268 			ceph_decode_copy(p, &info->btime, sizeof(info->btime));
269 			ceph_decode_64_safe(p, end, info->change_attr, bad);
270 		}
271 
272 		info->dir_pin = -ENODATA;
273 		/* info->snap_btime and info->rsnaps remain zero */
274 	}
275 	return 0;
276 bad:
277 	err = -EIO;
278 out_bad:
279 	return err;
280 }
281 
282 static int parse_reply_info_dir(void **p, void *end,
283 				struct ceph_mds_reply_dirfrag **dirfrag,
284 				u64 features)
285 {
286 	if (features == (u64)-1) {
287 		u8 struct_v, struct_compat;
288 		u32 struct_len;
289 		ceph_decode_8_safe(p, end, struct_v, bad);
290 		ceph_decode_8_safe(p, end, struct_compat, bad);
291 		/* struct_v is expected to be >= 1. we only understand
292 		 * encoding whose struct_compat == 1. */
293 		if (!struct_v || struct_compat != 1)
294 			goto bad;
295 		ceph_decode_32_safe(p, end, struct_len, bad);
296 		ceph_decode_need(p, end, struct_len, bad);
297 		end = *p + struct_len;
298 	}
299 
300 	ceph_decode_need(p, end, sizeof(**dirfrag), bad);
301 	*dirfrag = *p;
302 	*p += sizeof(**dirfrag) + sizeof(u32) * le32_to_cpu((*dirfrag)->ndist);
303 	if (unlikely(*p > end))
304 		goto bad;
305 	if (features == (u64)-1)
306 		*p = end;
307 	return 0;
308 bad:
309 	return -EIO;
310 }
311 
312 static int parse_reply_info_lease(void **p, void *end,
313 				  struct ceph_mds_reply_lease **lease,
314 				  u64 features, u32 *altname_len, u8 **altname)
315 {
316 	u8 struct_v;
317 	u32 struct_len;
318 	void *lend;
319 
320 	if (features == (u64)-1) {
321 		u8 struct_compat;
322 
323 		ceph_decode_8_safe(p, end, struct_v, bad);
324 		ceph_decode_8_safe(p, end, struct_compat, bad);
325 
326 		/* struct_v is expected to be >= 1. we only understand
327 		 * encoding whose struct_compat == 1. */
328 		if (!struct_v || struct_compat != 1)
329 			goto bad;
330 
331 		ceph_decode_32_safe(p, end, struct_len, bad);
332 	} else {
333 		struct_len = sizeof(**lease);
334 		*altname_len = 0;
335 		*altname = NULL;
336 	}
337 
338 	lend = *p + struct_len;
339 	ceph_decode_need(p, end, struct_len, bad);
340 	*lease = *p;
341 	*p += sizeof(**lease);
342 
343 	if (features == (u64)-1) {
344 		if (struct_v >= 2) {
345 			ceph_decode_32_safe(p, end, *altname_len, bad);
346 			ceph_decode_need(p, end, *altname_len, bad);
347 			*altname = *p;
348 			*p += *altname_len;
349 		} else {
350 			*altname = NULL;
351 			*altname_len = 0;
352 		}
353 	}
354 	*p = lend;
355 	return 0;
356 bad:
357 	return -EIO;
358 }
359 
360 /*
361  * parse a normal reply, which may contain a (dir+)dentry and/or a
362  * target inode.
363  */
364 static int parse_reply_info_trace(void **p, void *end,
365 				  struct ceph_mds_reply_info_parsed *info,
366 				  u64 features)
367 {
368 	int err;
369 
370 	if (info->head->is_dentry) {
371 		err = parse_reply_info_in(p, end, &info->diri, features);
372 		if (err < 0)
373 			goto out_bad;
374 
375 		err = parse_reply_info_dir(p, end, &info->dirfrag, features);
376 		if (err < 0)
377 			goto out_bad;
378 
379 		ceph_decode_32_safe(p, end, info->dname_len, bad);
380 		ceph_decode_need(p, end, info->dname_len, bad);
381 		info->dname = *p;
382 		*p += info->dname_len;
383 
384 		err = parse_reply_info_lease(p, end, &info->dlease, features,
385 					     &info->altname_len, &info->altname);
386 		if (err < 0)
387 			goto out_bad;
388 	}
389 
390 	if (info->head->is_target) {
391 		err = parse_reply_info_in(p, end, &info->targeti, features);
392 		if (err < 0)
393 			goto out_bad;
394 	}
395 
396 	if (unlikely(*p != end))
397 		goto bad;
398 	return 0;
399 
400 bad:
401 	err = -EIO;
402 out_bad:
403 	pr_err("problem parsing mds trace %d\n", err);
404 	return err;
405 }
406 
407 /*
408  * parse readdir results
409  */
410 static int parse_reply_info_readdir(void **p, void *end,
411 				    struct ceph_mds_request *req,
412 				    u64 features)
413 {
414 	struct ceph_mds_reply_info_parsed *info = &req->r_reply_info;
415 	struct ceph_client *cl = req->r_mdsc->fsc->client;
416 	u32 num, i = 0;
417 	int err;
418 
419 	err = parse_reply_info_dir(p, end, &info->dir_dir, features);
420 	if (err < 0)
421 		goto out_bad;
422 
423 	ceph_decode_need(p, end, sizeof(num) + 2, bad);
424 	num = ceph_decode_32(p);
425 	{
426 		u16 flags = ceph_decode_16(p);
427 		info->dir_end = !!(flags & CEPH_READDIR_FRAG_END);
428 		info->dir_complete = !!(flags & CEPH_READDIR_FRAG_COMPLETE);
429 		info->hash_order = !!(flags & CEPH_READDIR_HASH_ORDER);
430 		info->offset_hash = !!(flags & CEPH_READDIR_OFFSET_HASH);
431 	}
432 	if (num == 0)
433 		goto done;
434 
435 	BUG_ON(!info->dir_entries);
436 	if ((unsigned long)(info->dir_entries + num) >
437 	    (unsigned long)info->dir_entries + info->dir_buf_size) {
438 		pr_err_client(cl, "dir contents are larger than expected\n");
439 		WARN_ON(1);
440 		goto bad;
441 	}
442 
443 	info->dir_nr = num;
444 	while (num) {
445 		struct inode *inode = d_inode(req->r_dentry);
446 		struct ceph_inode_info *ci = ceph_inode(inode);
447 		struct ceph_mds_reply_dir_entry *rde = info->dir_entries + i;
448 		struct fscrypt_str tname = FSTR_INIT(NULL, 0);
449 		struct fscrypt_str oname = FSTR_INIT(NULL, 0);
450 		struct ceph_fname fname;
451 		u32 altname_len, _name_len;
452 		u8 *altname, *_name;
453 
454 		/* dentry */
455 		ceph_decode_32_safe(p, end, _name_len, bad);
456 		ceph_decode_need(p, end, _name_len, bad);
457 		_name = *p;
458 		*p += _name_len;
459 		doutc(cl, "parsed dir dname '%.*s'\n", _name_len, _name);
460 
461 		if (info->hash_order)
462 			rde->raw_hash = ceph_str_hash(ci->i_dir_layout.dl_dir_hash,
463 						      _name, _name_len);
464 
465 		/* dentry lease */
466 		err = parse_reply_info_lease(p, end, &rde->lease, features,
467 					     &altname_len, &altname);
468 		if (err)
469 			goto out_bad;
470 
471 		/*
472 		 * Try to dencrypt the dentry names and update them
473 		 * in the ceph_mds_reply_dir_entry struct.
474 		 */
475 		fname.dir = inode;
476 		fname.name = _name;
477 		fname.name_len = _name_len;
478 		fname.ctext = altname;
479 		fname.ctext_len = altname_len;
480 		/*
481 		 * The _name_len maybe larger than altname_len, such as
482 		 * when the human readable name length is in range of
483 		 * (CEPH_NOHASH_NAME_MAX, CEPH_NOHASH_NAME_MAX + SHA256_DIGEST_SIZE),
484 		 * then the copy in ceph_fname_to_usr will corrupt the
485 		 * data if there has no encryption key.
486 		 *
487 		 * Just set the no_copy flag and then if there has no
488 		 * encryption key the oname.name will be assigned to
489 		 * _name always.
490 		 */
491 		fname.no_copy = true;
492 		if (altname_len == 0) {
493 			/*
494 			 * Set tname to _name, and this will be used
495 			 * to do the base64_decode in-place. It's
496 			 * safe because the decoded string should
497 			 * always be shorter, which is 3/4 of origin
498 			 * string.
499 			 */
500 			tname.name = _name;
501 
502 			/*
503 			 * Set oname to _name too, and this will be
504 			 * used to do the dencryption in-place.
505 			 */
506 			oname.name = _name;
507 			oname.len = _name_len;
508 		} else {
509 			/*
510 			 * This will do the decryption only in-place
511 			 * from altname cryptext directly.
512 			 */
513 			oname.name = altname;
514 			oname.len = altname_len;
515 		}
516 		rde->is_nokey = false;
517 		err = ceph_fname_to_usr(&fname, &tname, &oname, &rde->is_nokey);
518 		if (err) {
519 			pr_err_client(cl, "unable to decode %.*s, got %d\n",
520 				      _name_len, _name, err);
521 			goto out_bad;
522 		}
523 		rde->name = oname.name;
524 		rde->name_len = oname.len;
525 
526 		/* inode */
527 		err = parse_reply_info_in(p, end, &rde->inode, features);
528 		if (err < 0)
529 			goto out_bad;
530 		/* ceph_readdir_prepopulate() will update it */
531 		rde->offset = 0;
532 		i++;
533 		num--;
534 	}
535 
536 done:
537 	/* Skip over any unrecognized fields */
538 	*p = end;
539 	return 0;
540 
541 bad:
542 	err = -EIO;
543 out_bad:
544 	pr_err_client(cl, "problem parsing dir contents %d\n", err);
545 	return err;
546 }
547 
548 /*
549  * parse fcntl F_GETLK results
550  */
551 static int parse_reply_info_filelock(void **p, void *end,
552 				     struct ceph_mds_reply_info_parsed *info,
553 				     u64 features)
554 {
555 	if (*p + sizeof(*info->filelock_reply) > end)
556 		goto bad;
557 
558 	info->filelock_reply = *p;
559 
560 	/* Skip over any unrecognized fields */
561 	*p = end;
562 	return 0;
563 bad:
564 	return -EIO;
565 }
566 
567 
568 #if BITS_PER_LONG == 64
569 
570 #define DELEGATED_INO_AVAILABLE		xa_mk_value(1)
571 
572 static int ceph_parse_deleg_inos(void **p, void *end,
573 				 struct ceph_mds_session *s)
574 {
575 	struct ceph_client *cl = s->s_mdsc->fsc->client;
576 	u32 sets;
577 
578 	ceph_decode_32_safe(p, end, sets, bad);
579 	doutc(cl, "got %u sets of delegated inodes\n", sets);
580 	while (sets--) {
581 		u64 start, len;
582 
583 		ceph_decode_64_safe(p, end, start, bad);
584 		ceph_decode_64_safe(p, end, len, bad);
585 
586 		/* Don't accept a delegation of system inodes */
587 		if (start < CEPH_INO_SYSTEM_BASE) {
588 			pr_warn_ratelimited_client(cl,
589 				"ignoring reserved inode range delegation (start=0x%llx len=0x%llx)\n",
590 				start, len);
591 			continue;
592 		}
593 		while (len--) {
594 			int err = xa_insert(&s->s_delegated_inos, start++,
595 					    DELEGATED_INO_AVAILABLE,
596 					    GFP_KERNEL);
597 			if (!err) {
598 				doutc(cl, "added delegated inode 0x%llx\n", start - 1);
599 			} else if (err == -EBUSY) {
600 				pr_warn_client(cl,
601 					"MDS delegated inode 0x%llx more than once.\n",
602 					start - 1);
603 			} else {
604 				return err;
605 			}
606 		}
607 	}
608 	return 0;
609 bad:
610 	return -EIO;
611 }
612 
613 u64 ceph_get_deleg_ino(struct ceph_mds_session *s)
614 {
615 	unsigned long ino;
616 	void *val;
617 
618 	xa_for_each(&s->s_delegated_inos, ino, val) {
619 		val = xa_erase(&s->s_delegated_inos, ino);
620 		if (val == DELEGATED_INO_AVAILABLE)
621 			return ino;
622 	}
623 	return 0;
624 }
625 
626 int ceph_restore_deleg_ino(struct ceph_mds_session *s, u64 ino)
627 {
628 	return xa_insert(&s->s_delegated_inos, ino, DELEGATED_INO_AVAILABLE,
629 			 GFP_KERNEL);
630 }
631 #else /* BITS_PER_LONG == 64 */
632 /*
633  * FIXME: xarrays can't handle 64-bit indexes on a 32-bit arch. For now, just
634  * ignore delegated_inos on 32 bit arch. Maybe eventually add xarrays for top
635  * and bottom words?
636  */
637 static int ceph_parse_deleg_inos(void **p, void *end,
638 				 struct ceph_mds_session *s)
639 {
640 	u32 sets;
641 
642 	ceph_decode_32_safe(p, end, sets, bad);
643 	if (sets)
644 		ceph_decode_skip_n(p, end, sets * 2 * sizeof(__le64), bad);
645 	return 0;
646 bad:
647 	return -EIO;
648 }
649 
650 u64 ceph_get_deleg_ino(struct ceph_mds_session *s)
651 {
652 	return 0;
653 }
654 
655 int ceph_restore_deleg_ino(struct ceph_mds_session *s, u64 ino)
656 {
657 	return 0;
658 }
659 #endif /* BITS_PER_LONG == 64 */
660 
661 /*
662  * parse create results
663  */
664 static int parse_reply_info_create(void **p, void *end,
665 				  struct ceph_mds_reply_info_parsed *info,
666 				  u64 features, struct ceph_mds_session *s)
667 {
668 	int ret;
669 
670 	if (features == (u64)-1 ||
671 	    (features & CEPH_FEATURE_REPLY_CREATE_INODE)) {
672 		if (*p == end) {
673 			/* Malformed reply? */
674 			info->has_create_ino = false;
675 		} else if (test_bit(CEPHFS_FEATURE_DELEG_INO, &s->s_features)) {
676 			info->has_create_ino = true;
677 			/* struct_v, struct_compat, and len */
678 			ceph_decode_skip_n(p, end, 2 + sizeof(u32), bad);
679 			ceph_decode_64_safe(p, end, info->ino, bad);
680 			ret = ceph_parse_deleg_inos(p, end, s);
681 			if (ret)
682 				return ret;
683 		} else {
684 			/* legacy */
685 			ceph_decode_64_safe(p, end, info->ino, bad);
686 			info->has_create_ino = true;
687 		}
688 	} else {
689 		if (*p != end)
690 			goto bad;
691 	}
692 
693 	/* Skip over any unrecognized fields */
694 	*p = end;
695 	return 0;
696 bad:
697 	return -EIO;
698 }
699 
700 static int parse_reply_info_getvxattr(void **p, void *end,
701 				      struct ceph_mds_reply_info_parsed *info,
702 				      u64 features)
703 {
704 	u32 value_len;
705 
706 	ceph_decode_skip_8(p, end, bad); /* skip current version: 1 */
707 	ceph_decode_skip_8(p, end, bad); /* skip first version: 1 */
708 	ceph_decode_skip_32(p, end, bad); /* skip payload length */
709 
710 	ceph_decode_32_safe(p, end, value_len, bad);
711 
712 	if (value_len == end - *p) {
713 	  info->xattr_info.xattr_value = *p;
714 	  info->xattr_info.xattr_value_len = value_len;
715 	  *p = end;
716 	  return value_len;
717 	}
718 bad:
719 	return -EIO;
720 }
721 
722 /*
723  * parse extra results
724  */
725 static int parse_reply_info_extra(void **p, void *end,
726 				  struct ceph_mds_request *req,
727 				  u64 features, struct ceph_mds_session *s)
728 {
729 	struct ceph_mds_reply_info_parsed *info = &req->r_reply_info;
730 	u32 op = le32_to_cpu(info->head->op);
731 
732 	if (op == CEPH_MDS_OP_GETFILELOCK)
733 		return parse_reply_info_filelock(p, end, info, features);
734 	else if (op == CEPH_MDS_OP_READDIR || op == CEPH_MDS_OP_LSSNAP)
735 		return parse_reply_info_readdir(p, end, req, features);
736 	else if (op == CEPH_MDS_OP_CREATE)
737 		return parse_reply_info_create(p, end, info, features, s);
738 	else if (op == CEPH_MDS_OP_GETVXATTR)
739 		return parse_reply_info_getvxattr(p, end, info, features);
740 	else
741 		return -EIO;
742 }
743 
744 /*
745  * parse entire mds reply
746  */
747 static int parse_reply_info(struct ceph_mds_session *s, struct ceph_msg *msg,
748 			    struct ceph_mds_request *req, u64 features)
749 {
750 	struct ceph_mds_reply_info_parsed *info = &req->r_reply_info;
751 	struct ceph_client *cl = s->s_mdsc->fsc->client;
752 	void *p, *end;
753 	u32 len;
754 	int err;
755 
756 	info->head = msg->front.iov_base;
757 	p = msg->front.iov_base + sizeof(struct ceph_mds_reply_head);
758 	end = p + msg->front.iov_len - sizeof(struct ceph_mds_reply_head);
759 
760 	/* trace */
761 	ceph_decode_32_safe(&p, end, len, bad);
762 	if (len > 0) {
763 		ceph_decode_need(&p, end, len, bad);
764 		err = parse_reply_info_trace(&p, p+len, info, features);
765 		if (err < 0)
766 			goto out_bad;
767 	}
768 
769 	/* extra */
770 	ceph_decode_32_safe(&p, end, len, bad);
771 	if (len > 0) {
772 		ceph_decode_need(&p, end, len, bad);
773 		err = parse_reply_info_extra(&p, p+len, req, features, s);
774 		if (err < 0)
775 			goto out_bad;
776 	}
777 
778 	/* snap blob */
779 	ceph_decode_32_safe(&p, end, len, bad);
780 	info->snapblob_len = len;
781 	info->snapblob = p;
782 	p += len;
783 
784 	if (p != end)
785 		goto bad;
786 	return 0;
787 
788 bad:
789 	err = -EIO;
790 out_bad:
791 	pr_err_client(cl, "mds parse_reply err %d\n", err);
792 	ceph_msg_dump(msg);
793 	return err;
794 }
795 
796 static void destroy_reply_info(struct ceph_mds_reply_info_parsed *info)
797 {
798 	int i;
799 
800 	kfree(info->diri.fscrypt_auth);
801 	kfree(info->diri.fscrypt_file);
802 	kfree(info->targeti.fscrypt_auth);
803 	kfree(info->targeti.fscrypt_file);
804 	if (!info->dir_entries)
805 		return;
806 
807 	for (i = 0; i < info->dir_nr; i++) {
808 		struct ceph_mds_reply_dir_entry *rde = info->dir_entries + i;
809 
810 		kfree(rde->inode.fscrypt_auth);
811 		kfree(rde->inode.fscrypt_file);
812 	}
813 	free_pages((unsigned long)info->dir_entries, get_order(info->dir_buf_size));
814 }
815 
816 /*
817  * In async unlink case the kclient won't wait for the first reply
818  * from MDS and just drop all the links and unhash the dentry and then
819  * succeeds immediately.
820  *
821  * For any new create/link/rename,etc requests followed by using the
822  * same file names we must wait for the first reply of the inflight
823  * unlink request, or the MDS possibly will fail these following
824  * requests with -EEXIST if the inflight async unlink request was
825  * delayed for some reasons.
826  *
827  * And the worst case is that for the none async openc request it will
828  * successfully open the file if the CDentry hasn't been unlinked yet,
829  * but later the previous delayed async unlink request will remove the
830  * CDenty. That means the just created file is possiblly deleted later
831  * by accident.
832  *
833  * We need to wait for the inflight async unlink requests to finish
834  * when creating new files/directories by using the same file names.
835  */
836 int ceph_wait_on_conflict_unlink(struct dentry *dentry)
837 {
838 	struct ceph_fs_client *fsc = ceph_sb_to_fs_client(dentry->d_sb);
839 	struct ceph_client *cl = fsc->client;
840 	struct dentry *pdentry = dentry->d_parent;
841 	struct dentry *udentry, *found = NULL;
842 	struct ceph_dentry_info *di;
843 	struct qstr dname;
844 	u32 hash = dentry->d_name.hash;
845 	int err;
846 
847 	dname.name = dentry->d_name.name;
848 	dname.len = dentry->d_name.len;
849 
850 	rcu_read_lock();
851 	hash_for_each_possible_rcu(fsc->async_unlink_conflict, di,
852 				   hnode, hash) {
853 		udentry = di->dentry;
854 
855 		spin_lock(&udentry->d_lock);
856 		if (udentry->d_name.hash != hash)
857 			goto next;
858 		if (unlikely(udentry->d_parent != pdentry))
859 			goto next;
860 		if (!hash_hashed(&di->hnode))
861 			goto next;
862 
863 		if (!test_bit(CEPH_DENTRY_ASYNC_UNLINK_BIT, &di->flags))
864 			pr_warn_client(cl, "dentry %p:%pd async unlink bit is not set\n",
865 				       dentry, dentry);
866 
867 		if (!d_same_name(udentry, pdentry, &dname))
868 			goto next;
869 
870 		found = dget_dlock(udentry);
871 		spin_unlock(&udentry->d_lock);
872 		break;
873 next:
874 		spin_unlock(&udentry->d_lock);
875 	}
876 	rcu_read_unlock();
877 
878 	if (likely(!found))
879 		return 0;
880 
881 	doutc(cl, "dentry %p:%pd conflict with old %p:%pd\n", dentry, dentry,
882 	      found, found);
883 
884 	err = wait_on_bit(&di->flags, CEPH_DENTRY_ASYNC_UNLINK_BIT,
885 			  TASK_KILLABLE);
886 	dput(found);
887 	return err;
888 }
889 
890 
891 /*
892  * sessions
893  */
894 const char *ceph_session_state_name(int s)
895 {
896 	switch (s) {
897 	case CEPH_MDS_SESSION_NEW: return "new";
898 	case CEPH_MDS_SESSION_OPENING: return "opening";
899 	case CEPH_MDS_SESSION_OPEN: return "open";
900 	case CEPH_MDS_SESSION_HUNG: return "hung";
901 	case CEPH_MDS_SESSION_CLOSING: return "closing";
902 	case CEPH_MDS_SESSION_CLOSED: return "closed";
903 	case CEPH_MDS_SESSION_RESTARTING: return "restarting";
904 	case CEPH_MDS_SESSION_RECONNECTING: return "reconnecting";
905 	case CEPH_MDS_SESSION_REJECTED: return "rejected";
906 	default: return "???";
907 	}
908 }
909 
910 struct ceph_mds_session *ceph_get_mds_session(struct ceph_mds_session *s)
911 {
912 	if (refcount_inc_not_zero(&s->s_ref))
913 		return s;
914 	return NULL;
915 }
916 
917 void ceph_put_mds_session(struct ceph_mds_session *s)
918 {
919 	if (IS_ERR_OR_NULL(s))
920 		return;
921 
922 	if (refcount_dec_and_test(&s->s_ref)) {
923 		if (s->s_auth.authorizer)
924 			ceph_auth_destroy_authorizer(s->s_auth.authorizer);
925 		WARN_ON(mutex_is_locked(&s->s_mutex));
926 		xa_destroy(&s->s_delegated_inos);
927 		kfree(s);
928 	}
929 }
930 
931 /*
932  * called under mdsc->mutex
933  */
934 struct ceph_mds_session *__ceph_lookup_mds_session(struct ceph_mds_client *mdsc,
935 						   int mds)
936 {
937 	if (mds >= mdsc->max_sessions || !mdsc->sessions[mds])
938 		return NULL;
939 	return ceph_get_mds_session(mdsc->sessions[mds]);
940 }
941 
942 static bool __have_session(struct ceph_mds_client *mdsc, int mds)
943 {
944 	if (mds >= mdsc->max_sessions || !mdsc->sessions[mds])
945 		return false;
946 	else
947 		return true;
948 }
949 
950 static int __verify_registered_session(struct ceph_mds_client *mdsc,
951 				       struct ceph_mds_session *s)
952 {
953 	if (s->s_mds >= mdsc->max_sessions ||
954 	    mdsc->sessions[s->s_mds] != s)
955 		return -ENOENT;
956 	return 0;
957 }
958 
959 /*
960  * create+register a new session for given mds.
961  * called under mdsc->mutex.
962  */
963 static struct ceph_mds_session *register_session(struct ceph_mds_client *mdsc,
964 						 int mds)
965 {
966 	struct ceph_client *cl = mdsc->fsc->client;
967 	struct ceph_mds_session *s;
968 
969 	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_FENCE_IO)
970 		return ERR_PTR(-EIO);
971 
972 	if (mds >= mdsc->mdsmap->possible_max_rank)
973 		return ERR_PTR(-EINVAL);
974 
975 	s = kzalloc(sizeof(*s), GFP_NOFS);
976 	if (!s)
977 		return ERR_PTR(-ENOMEM);
978 
979 	if (mds >= mdsc->max_sessions) {
980 		int newmax = 1 << get_count_order(mds + 1);
981 		struct ceph_mds_session **sa;
982 
983 		doutc(cl, "realloc to %d\n", newmax);
984 		sa = kcalloc(newmax, sizeof(void *), GFP_NOFS);
985 		if (!sa)
986 			goto fail_realloc;
987 		if (mdsc->sessions) {
988 			memcpy(sa, mdsc->sessions,
989 			       mdsc->max_sessions * sizeof(void *));
990 			kfree(mdsc->sessions);
991 		}
992 		mdsc->sessions = sa;
993 		mdsc->max_sessions = newmax;
994 	}
995 
996 	doutc(cl, "mds%d\n", mds);
997 	s->s_mdsc = mdsc;
998 	s->s_mds = mds;
999 	s->s_state = CEPH_MDS_SESSION_NEW;
1000 	mutex_init(&s->s_mutex);
1001 
1002 	ceph_con_init(&s->s_con, s, &mds_con_ops, &mdsc->fsc->client->msgr);
1003 
1004 	atomic_set(&s->s_cap_gen, 1);
1005 	s->s_cap_ttl = jiffies - 1;
1006 
1007 	spin_lock_init(&s->s_cap_lock);
1008 	INIT_LIST_HEAD(&s->s_caps);
1009 	refcount_set(&s->s_ref, 1);
1010 	INIT_LIST_HEAD(&s->s_waiting);
1011 	INIT_LIST_HEAD(&s->s_unsafe);
1012 	xa_init(&s->s_delegated_inos);
1013 	INIT_LIST_HEAD(&s->s_cap_releases);
1014 	INIT_WORK(&s->s_cap_release_work, ceph_cap_release_work);
1015 
1016 	INIT_LIST_HEAD(&s->s_cap_dirty);
1017 	INIT_LIST_HEAD(&s->s_cap_flushing);
1018 
1019 	mdsc->sessions[mds] = s;
1020 	atomic_inc(&mdsc->num_sessions);
1021 	refcount_inc(&s->s_ref);  /* one ref to sessions[], one to caller */
1022 
1023 	ceph_con_open(&s->s_con, CEPH_ENTITY_TYPE_MDS, mds,
1024 		      ceph_mdsmap_get_addr(mdsc->mdsmap, mds));
1025 
1026 	return s;
1027 
1028 fail_realloc:
1029 	kfree(s);
1030 	return ERR_PTR(-ENOMEM);
1031 }
1032 
1033 /*
1034  * called under mdsc->mutex
1035  */
1036 static void __unregister_session(struct ceph_mds_client *mdsc,
1037 			       struct ceph_mds_session *s)
1038 {
1039 	doutc(mdsc->fsc->client, "mds%d %p\n", s->s_mds, s);
1040 	BUG_ON(mdsc->sessions[s->s_mds] != s);
1041 	mdsc->sessions[s->s_mds] = NULL;
1042 	ceph_con_close(&s->s_con);
1043 	ceph_put_mds_session(s);
1044 	atomic_dec(&mdsc->num_sessions);
1045 }
1046 
1047 /*
1048  * drop session refs in request.
1049  *
1050  * should be last request ref, or hold mdsc->mutex
1051  */
1052 static void put_request_session(struct ceph_mds_request *req)
1053 {
1054 	if (req->r_session) {
1055 		ceph_put_mds_session(req->r_session);
1056 		req->r_session = NULL;
1057 	}
1058 }
1059 
1060 void ceph_mdsc_iterate_sessions(struct ceph_mds_client *mdsc,
1061 				void (*cb)(struct ceph_mds_session *),
1062 				bool check_state)
1063 {
1064 	int mds;
1065 
1066 	mutex_lock(&mdsc->mutex);
1067 	for (mds = 0; mds < mdsc->max_sessions; ++mds) {
1068 		struct ceph_mds_session *s;
1069 
1070 		s = __ceph_lookup_mds_session(mdsc, mds);
1071 		if (!s)
1072 			continue;
1073 
1074 		if (check_state && !check_session_state(s)) {
1075 			ceph_put_mds_session(s);
1076 			continue;
1077 		}
1078 
1079 		mutex_unlock(&mdsc->mutex);
1080 		cb(s);
1081 		ceph_put_mds_session(s);
1082 		mutex_lock(&mdsc->mutex);
1083 	}
1084 	mutex_unlock(&mdsc->mutex);
1085 }
1086 
1087 void ceph_mdsc_release_request(struct kref *kref)
1088 {
1089 	struct ceph_mds_request *req = container_of(kref,
1090 						    struct ceph_mds_request,
1091 						    r_kref);
1092 	ceph_mdsc_release_dir_caps_async(req);
1093 	destroy_reply_info(&req->r_reply_info);
1094 	if (req->r_request)
1095 		ceph_msg_put(req->r_request);
1096 	if (req->r_reply)
1097 		ceph_msg_put(req->r_reply);
1098 	if (req->r_inode) {
1099 		ceph_put_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
1100 		iput(req->r_inode);
1101 	}
1102 	if (req->r_parent) {
1103 		ceph_put_cap_refs(ceph_inode(req->r_parent), CEPH_CAP_PIN);
1104 		iput(req->r_parent);
1105 	}
1106 	iput(req->r_target_inode);
1107 	iput(req->r_new_inode);
1108 	if (req->r_dentry)
1109 		dput(req->r_dentry);
1110 	if (req->r_old_dentry)
1111 		dput(req->r_old_dentry);
1112 	if (req->r_old_dentry_dir) {
1113 		/*
1114 		 * track (and drop pins for) r_old_dentry_dir
1115 		 * separately, since r_old_dentry's d_parent may have
1116 		 * changed between the dir mutex being dropped and
1117 		 * this request being freed.
1118 		 */
1119 		ceph_put_cap_refs(ceph_inode(req->r_old_dentry_dir),
1120 				  CEPH_CAP_PIN);
1121 		iput(req->r_old_dentry_dir);
1122 	}
1123 	kfree(req->r_path1);
1124 	kfree(req->r_path2);
1125 	put_cred(req->r_cred);
1126 	if (req->r_mnt_idmap)
1127 		mnt_idmap_put(req->r_mnt_idmap);
1128 	if (req->r_pagelist)
1129 		ceph_pagelist_release(req->r_pagelist);
1130 	kfree(req->r_fscrypt_auth);
1131 	kfree(req->r_altname);
1132 	put_request_session(req);
1133 	ceph_unreserve_caps(req->r_mdsc, &req->r_caps_reservation);
1134 	WARN_ON_ONCE(!list_empty(&req->r_wait));
1135 	kmem_cache_free(ceph_mds_request_cachep, req);
1136 }
1137 
1138 DEFINE_RB_FUNCS(request, struct ceph_mds_request, r_tid, r_node)
1139 
1140 /*
1141  * lookup session, bump ref if found.
1142  *
1143  * called under mdsc->mutex.
1144  */
1145 static struct ceph_mds_request *
1146 lookup_get_request(struct ceph_mds_client *mdsc, u64 tid)
1147 {
1148 	struct ceph_mds_request *req;
1149 
1150 	req = lookup_request(&mdsc->request_tree, tid);
1151 	if (req)
1152 		ceph_mdsc_get_request(req);
1153 
1154 	return req;
1155 }
1156 
1157 /*
1158  * Register an in-flight request, and assign a tid.  Link to directory
1159  * are modifying (if any).
1160  *
1161  * Called under mdsc->mutex.
1162  */
1163 static void __register_request(struct ceph_mds_client *mdsc,
1164 			       struct ceph_mds_request *req,
1165 			       struct inode *dir)
1166 {
1167 	struct ceph_client *cl = mdsc->fsc->client;
1168 	int ret = 0;
1169 
1170 	req->r_tid = ++mdsc->last_tid;
1171 	if (req->r_num_caps) {
1172 		ret = ceph_reserve_caps(mdsc, &req->r_caps_reservation,
1173 					req->r_num_caps);
1174 		if (ret < 0) {
1175 			pr_err_client(cl, "%p failed to reserve caps: %d\n",
1176 				      req, ret);
1177 			/* set req->r_err to fail early from __do_request */
1178 			req->r_err = ret;
1179 			return;
1180 		}
1181 	}
1182 	doutc(cl, "%p tid %lld\n", req, req->r_tid);
1183 	ceph_mdsc_get_request(req);
1184 	insert_request(&mdsc->request_tree, req);
1185 
1186 	req->r_cred = get_current_cred();
1187 	if (!req->r_mnt_idmap)
1188 		req->r_mnt_idmap = &nop_mnt_idmap;
1189 
1190 	if (mdsc->oldest_tid == 0 && req->r_op != CEPH_MDS_OP_SETFILELOCK)
1191 		mdsc->oldest_tid = req->r_tid;
1192 
1193 	if (dir) {
1194 		struct ceph_inode_info *ci = ceph_inode(dir);
1195 
1196 		ihold(dir);
1197 		req->r_unsafe_dir = dir;
1198 		spin_lock(&ci->i_unsafe_lock);
1199 		list_add_tail(&req->r_unsafe_dir_item, &ci->i_unsafe_dirops);
1200 		spin_unlock(&ci->i_unsafe_lock);
1201 	}
1202 }
1203 
1204 static void __unregister_request(struct ceph_mds_client *mdsc,
1205 				 struct ceph_mds_request *req)
1206 {
1207 	doutc(mdsc->fsc->client, "%p tid %lld\n", req, req->r_tid);
1208 
1209 	/* Never leave an unregistered request on an unsafe list! */
1210 	list_del_init(&req->r_unsafe_item);
1211 
1212 	if (req->r_tid == mdsc->oldest_tid) {
1213 		struct rb_node *p = rb_next(&req->r_node);
1214 		mdsc->oldest_tid = 0;
1215 		while (p) {
1216 			struct ceph_mds_request *next_req =
1217 				rb_entry(p, struct ceph_mds_request, r_node);
1218 			if (next_req->r_op != CEPH_MDS_OP_SETFILELOCK) {
1219 				mdsc->oldest_tid = next_req->r_tid;
1220 				break;
1221 			}
1222 			p = rb_next(p);
1223 		}
1224 	}
1225 
1226 	erase_request(&mdsc->request_tree, req);
1227 
1228 	if (req->r_unsafe_dir) {
1229 		struct ceph_inode_info *ci = ceph_inode(req->r_unsafe_dir);
1230 		spin_lock(&ci->i_unsafe_lock);
1231 		list_del_init(&req->r_unsafe_dir_item);
1232 		spin_unlock(&ci->i_unsafe_lock);
1233 	}
1234 	if (req->r_target_inode &&
1235 	    test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
1236 		struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
1237 		spin_lock(&ci->i_unsafe_lock);
1238 		list_del_init(&req->r_unsafe_target_item);
1239 		spin_unlock(&ci->i_unsafe_lock);
1240 	}
1241 
1242 	if (req->r_unsafe_dir) {
1243 		iput(req->r_unsafe_dir);
1244 		req->r_unsafe_dir = NULL;
1245 	}
1246 
1247 	complete_all(&req->r_safe_completion);
1248 
1249 	ceph_mdsc_put_request(req);
1250 }
1251 
1252 /*
1253  * Walk back up the dentry tree until we hit a dentry representing a
1254  * non-snapshot inode. We do this using the rcu_read_lock (which must be held
1255  * when calling this) to ensure that the objects won't disappear while we're
1256  * working with them. Once we hit a candidate dentry, we attempt to take a
1257  * reference to it, and return that as the result.
1258  */
1259 static struct inode *get_nonsnap_parent(struct dentry *dentry)
1260 {
1261 	struct inode *inode = NULL;
1262 
1263 	while (dentry && !IS_ROOT(dentry)) {
1264 		inode = d_inode_rcu(dentry);
1265 		if (!inode || ceph_snap(inode) == CEPH_NOSNAP)
1266 			break;
1267 		dentry = dentry->d_parent;
1268 	}
1269 	if (inode)
1270 		inode = igrab(inode);
1271 	return inode;
1272 }
1273 
1274 /*
1275  * Choose mds to send request to next.  If there is a hint set in the
1276  * request (e.g., due to a prior forward hint from the mds), use that.
1277  * Otherwise, consult frag tree and/or caps to identify the
1278  * appropriate mds.  If all else fails, choose randomly.
1279  *
1280  * Called under mdsc->mutex.
1281  */
1282 static int __choose_mds(struct ceph_mds_client *mdsc,
1283 			struct ceph_mds_request *req,
1284 			bool *random)
1285 {
1286 	struct inode *inode;
1287 	struct ceph_inode_info *ci;
1288 	struct ceph_cap *cap;
1289 	int mode = req->r_direct_mode;
1290 	int mds = -1;
1291 	u32 hash = req->r_direct_hash;
1292 	bool is_hash = test_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags);
1293 	struct ceph_client *cl = mdsc->fsc->client;
1294 
1295 	if (random)
1296 		*random = false;
1297 
1298 	/*
1299 	 * is there a specific mds we should try?  ignore hint if we have
1300 	 * no session and the mds is not up (active or recovering).
1301 	 */
1302 	if (req->r_resend_mds >= 0 &&
1303 	    (__have_session(mdsc, req->r_resend_mds) ||
1304 	     ceph_mdsmap_get_state(mdsc->mdsmap, req->r_resend_mds) > 0)) {
1305 		doutc(cl, "using resend_mds mds%d\n", req->r_resend_mds);
1306 		return req->r_resend_mds;
1307 	}
1308 
1309 	if (mode == USE_RANDOM_MDS)
1310 		goto random;
1311 
1312 	inode = NULL;
1313 	if (req->r_inode) {
1314 		if (ceph_snap(req->r_inode) != CEPH_SNAPDIR) {
1315 			inode = req->r_inode;
1316 			ihold(inode);
1317 		} else {
1318 			/* req->r_dentry is non-null for LSSNAP request */
1319 			rcu_read_lock();
1320 			inode = get_nonsnap_parent(req->r_dentry);
1321 			rcu_read_unlock();
1322 			doutc(cl, "using snapdir's parent %p %llx.%llx\n",
1323 			      inode, ceph_vinop(inode));
1324 		}
1325 	} else if (req->r_dentry) {
1326 		/* ignore race with rename; old or new d_parent is okay */
1327 		struct dentry *parent;
1328 		struct inode *dir;
1329 
1330 		rcu_read_lock();
1331 		parent = READ_ONCE(req->r_dentry->d_parent);
1332 		dir = req->r_parent ? : d_inode_rcu(parent);
1333 
1334 		if (!dir || dir->i_sb != mdsc->fsc->sb) {
1335 			/*  not this fs or parent went negative */
1336 			inode = d_inode(req->r_dentry);
1337 			if (inode)
1338 				ihold(inode);
1339 		} else if (ceph_snap(dir) != CEPH_NOSNAP) {
1340 			/* direct snapped/virtual snapdir requests
1341 			 * based on parent dir inode */
1342 			inode = get_nonsnap_parent(parent);
1343 			doutc(cl, "using nonsnap parent %p %llx.%llx\n",
1344 			      inode, ceph_vinop(inode));
1345 		} else {
1346 			/* dentry target */
1347 			inode = d_inode(req->r_dentry);
1348 			if (!inode || mode == USE_AUTH_MDS) {
1349 				/* dir + name */
1350 				inode = igrab(dir);
1351 				hash = ceph_dentry_hash(dir, req->r_dentry);
1352 				is_hash = true;
1353 			} else {
1354 				ihold(inode);
1355 			}
1356 		}
1357 		rcu_read_unlock();
1358 	}
1359 
1360 	if (!inode)
1361 		goto random;
1362 
1363 	doutc(cl, "%p %llx.%llx is_hash=%d (0x%x) mode %d\n", inode,
1364 	      ceph_vinop(inode), (int)is_hash, hash, mode);
1365 	ci = ceph_inode(inode);
1366 
1367 	if (is_hash && S_ISDIR(inode->i_mode)) {
1368 		struct ceph_inode_frag frag;
1369 		int found;
1370 
1371 		ceph_choose_frag(ci, hash, &frag, &found);
1372 		if (found) {
1373 			if (mode == USE_ANY_MDS && frag.ndist > 0) {
1374 				u8 r;
1375 
1376 				/* choose a random replica */
1377 				get_random_bytes(&r, 1);
1378 				r %= frag.ndist;
1379 				mds = frag.dist[r];
1380 				doutc(cl, "%p %llx.%llx frag %u mds%d (%d/%d)\n",
1381 				      inode, ceph_vinop(inode), frag.frag,
1382 				      mds, (int)r, frag.ndist);
1383 				if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
1384 				    CEPH_MDS_STATE_ACTIVE &&
1385 				    !ceph_mdsmap_is_laggy(mdsc->mdsmap, mds))
1386 					goto out;
1387 			}
1388 
1389 			/* since this file/dir wasn't known to be
1390 			 * replicated, then we want to look for the
1391 			 * authoritative mds. */
1392 			if (frag.mds >= 0) {
1393 				/* choose auth mds */
1394 				mds = frag.mds;
1395 				doutc(cl, "%p %llx.%llx frag %u mds%d (auth)\n",
1396 				      inode, ceph_vinop(inode), frag.frag, mds);
1397 				if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
1398 				    CEPH_MDS_STATE_ACTIVE) {
1399 					if (!ceph_mdsmap_is_laggy(mdsc->mdsmap,
1400 								  mds))
1401 						goto out;
1402 				}
1403 			}
1404 			mode = USE_AUTH_MDS;
1405 		}
1406 	}
1407 
1408 	spin_lock(&ci->i_ceph_lock);
1409 	cap = NULL;
1410 	if (mode == USE_AUTH_MDS)
1411 		cap = ci->i_auth_cap;
1412 	if (!cap && !RB_EMPTY_ROOT(&ci->i_caps))
1413 		cap = rb_entry(rb_first(&ci->i_caps), struct ceph_cap, ci_node);
1414 	if (!cap) {
1415 		spin_unlock(&ci->i_ceph_lock);
1416 		iput(inode);
1417 		goto random;
1418 	}
1419 	mds = cap->session->s_mds;
1420 	doutc(cl, "%p %llx.%llx mds%d (%scap %p)\n", inode,
1421 	      ceph_vinop(inode), mds,
1422 	      cap == ci->i_auth_cap ? "auth " : "", cap);
1423 	spin_unlock(&ci->i_ceph_lock);
1424 out:
1425 	iput(inode);
1426 	return mds;
1427 
1428 random:
1429 	if (random)
1430 		*random = true;
1431 
1432 	mds = ceph_mdsmap_get_random_mds(mdsc->mdsmap);
1433 	doutc(cl, "chose random mds%d\n", mds);
1434 	return mds;
1435 }
1436 
1437 
1438 /*
1439  * session messages
1440  */
1441 struct ceph_msg *ceph_create_session_msg(u32 op, u64 seq)
1442 {
1443 	struct ceph_msg *msg;
1444 	struct ceph_mds_session_head *h;
1445 
1446 	msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h), GFP_NOFS,
1447 			   false);
1448 	if (!msg) {
1449 		pr_err("ENOMEM creating session %s msg\n",
1450 		       ceph_session_op_name(op));
1451 		return NULL;
1452 	}
1453 	h = msg->front.iov_base;
1454 	h->op = cpu_to_le32(op);
1455 	h->seq = cpu_to_le64(seq);
1456 
1457 	return msg;
1458 }
1459 
1460 static const unsigned char feature_bits[] = CEPHFS_FEATURES_CLIENT_SUPPORTED;
1461 #define FEATURE_BYTES(c) (DIV_ROUND_UP((size_t)feature_bits[c - 1] + 1, 64) * 8)
1462 static int encode_supported_features(void **p, void *end)
1463 {
1464 	static const size_t count = ARRAY_SIZE(feature_bits);
1465 
1466 	if (count > 0) {
1467 		size_t i;
1468 		size_t size = FEATURE_BYTES(count);
1469 		unsigned long bit;
1470 
1471 		if (WARN_ON_ONCE(*p + 4 + size > end))
1472 			return -ERANGE;
1473 
1474 		ceph_encode_32(p, size);
1475 		memset(*p, 0, size);
1476 		for (i = 0; i < count; i++) {
1477 			bit = feature_bits[i];
1478 			((unsigned char *)(*p))[bit / 8] |= BIT(bit % 8);
1479 		}
1480 		*p += size;
1481 	} else {
1482 		if (WARN_ON_ONCE(*p + 4 > end))
1483 			return -ERANGE;
1484 
1485 		ceph_encode_32(p, 0);
1486 	}
1487 
1488 	return 0;
1489 }
1490 
1491 static const unsigned char metric_bits[] = CEPHFS_METRIC_SPEC_CLIENT_SUPPORTED;
1492 #define METRIC_BYTES(cnt) (DIV_ROUND_UP((size_t)metric_bits[cnt - 1] + 1, 64) * 8)
1493 static int encode_metric_spec(void **p, void *end)
1494 {
1495 	static const size_t count = ARRAY_SIZE(metric_bits);
1496 
1497 	/* header */
1498 	if (WARN_ON_ONCE(*p + 2 > end))
1499 		return -ERANGE;
1500 
1501 	ceph_encode_8(p, 1); /* version */
1502 	ceph_encode_8(p, 1); /* compat */
1503 
1504 	if (count > 0) {
1505 		size_t i;
1506 		size_t size = METRIC_BYTES(count);
1507 
1508 		if (WARN_ON_ONCE(*p + 4 + 4 + size > end))
1509 			return -ERANGE;
1510 
1511 		/* metric spec info length */
1512 		ceph_encode_32(p, 4 + size);
1513 
1514 		/* metric spec */
1515 		ceph_encode_32(p, size);
1516 		memset(*p, 0, size);
1517 		for (i = 0; i < count; i++)
1518 			((unsigned char *)(*p))[i / 8] |= BIT(metric_bits[i] % 8);
1519 		*p += size;
1520 	} else {
1521 		if (WARN_ON_ONCE(*p + 4 + 4 > end))
1522 			return -ERANGE;
1523 
1524 		/* metric spec info length */
1525 		ceph_encode_32(p, 4);
1526 		/* metric spec */
1527 		ceph_encode_32(p, 0);
1528 	}
1529 
1530 	return 0;
1531 }
1532 
1533 /*
1534  * session message, specialization for CEPH_SESSION_REQUEST_OPEN
1535  * to include additional client metadata fields.
1536  */
1537 static struct ceph_msg *
1538 create_session_full_msg(struct ceph_mds_client *mdsc, int op, u64 seq)
1539 {
1540 	struct ceph_msg *msg;
1541 	struct ceph_mds_session_head *h;
1542 	int i;
1543 	int extra_bytes = 0;
1544 	int metadata_key_count = 0;
1545 	struct ceph_options *opt = mdsc->fsc->client->options;
1546 	struct ceph_mount_options *fsopt = mdsc->fsc->mount_options;
1547 	struct ceph_client *cl = mdsc->fsc->client;
1548 	size_t size, count;
1549 	void *p, *end;
1550 	int ret;
1551 
1552 	const char* metadata[][2] = {
1553 		{"hostname", mdsc->nodename},
1554 		{"kernel_version", init_utsname()->release},
1555 		{"entity_id", opt->name ? : ""},
1556 		{"root", fsopt->server_path ? : "/"},
1557 		{NULL, NULL}
1558 	};
1559 
1560 	/* Calculate serialized length of metadata */
1561 	extra_bytes = 4;  /* map length */
1562 	for (i = 0; metadata[i][0]; ++i) {
1563 		extra_bytes += 8 + strlen(metadata[i][0]) +
1564 			strlen(metadata[i][1]);
1565 		metadata_key_count++;
1566 	}
1567 
1568 	/* supported feature */
1569 	size = 0;
1570 	count = ARRAY_SIZE(feature_bits);
1571 	if (count > 0)
1572 		size = FEATURE_BYTES(count);
1573 	extra_bytes += 4 + size;
1574 
1575 	/* metric spec */
1576 	size = 0;
1577 	count = ARRAY_SIZE(metric_bits);
1578 	if (count > 0)
1579 		size = METRIC_BYTES(count);
1580 	extra_bytes += 2 + 4 + 4 + size;
1581 
1582 	/* flags, mds auth caps and oldest_client_tid */
1583 	extra_bytes += 4 + 4 + 8;
1584 
1585 	/* Allocate the message */
1586 	msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h) + extra_bytes,
1587 			   GFP_NOFS, false);
1588 	if (!msg) {
1589 		pr_err_client(cl, "ENOMEM creating session open msg\n");
1590 		return ERR_PTR(-ENOMEM);
1591 	}
1592 	p = msg->front.iov_base;
1593 	end = p + msg->front.iov_len;
1594 
1595 	h = p;
1596 	h->op = cpu_to_le32(op);
1597 	h->seq = cpu_to_le64(seq);
1598 
1599 	/*
1600 	 * Serialize client metadata into waiting buffer space, using
1601 	 * the format that userspace expects for map<string, string>
1602 	 *
1603 	 * ClientSession messages with metadata are v7
1604 	 */
1605 	msg->hdr.version = cpu_to_le16(7);
1606 	msg->hdr.compat_version = cpu_to_le16(1);
1607 
1608 	/* The write pointer, following the session_head structure */
1609 	p += sizeof(*h);
1610 
1611 	/* Number of entries in the map */
1612 	ceph_encode_32(&p, metadata_key_count);
1613 
1614 	/* Two length-prefixed strings for each entry in the map */
1615 	for (i = 0; metadata[i][0]; ++i) {
1616 		size_t const key_len = strlen(metadata[i][0]);
1617 		size_t const val_len = strlen(metadata[i][1]);
1618 
1619 		ceph_encode_32(&p, key_len);
1620 		memcpy(p, metadata[i][0], key_len);
1621 		p += key_len;
1622 		ceph_encode_32(&p, val_len);
1623 		memcpy(p, metadata[i][1], val_len);
1624 		p += val_len;
1625 	}
1626 
1627 	ret = encode_supported_features(&p, end);
1628 	if (ret) {
1629 		pr_err_client(cl, "encode_supported_features failed!\n");
1630 		ceph_msg_put(msg);
1631 		return ERR_PTR(ret);
1632 	}
1633 
1634 	ret = encode_metric_spec(&p, end);
1635 	if (ret) {
1636 		pr_err_client(cl, "encode_metric_spec failed!\n");
1637 		ceph_msg_put(msg);
1638 		return ERR_PTR(ret);
1639 	}
1640 
1641 	/* version == 5, flags */
1642 	ceph_encode_32(&p, 0);
1643 
1644 	/* version == 6, mds auth caps */
1645 	ceph_encode_32(&p, 0);
1646 
1647 	/* version == 7, oldest_client_tid */
1648 	ceph_encode_64(&p, mdsc->oldest_tid);
1649 
1650 	msg->front.iov_len = p - msg->front.iov_base;
1651 	msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
1652 
1653 	return msg;
1654 }
1655 
1656 /*
1657  * send session open request.
1658  *
1659  * called under mdsc->mutex
1660  */
1661 static int __open_session(struct ceph_mds_client *mdsc,
1662 			  struct ceph_mds_session *session)
1663 {
1664 	struct ceph_msg *msg;
1665 	int mstate;
1666 	int mds = session->s_mds;
1667 
1668 	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_FENCE_IO)
1669 		return -EIO;
1670 
1671 	/* wait for mds to go active? */
1672 	mstate = ceph_mdsmap_get_state(mdsc->mdsmap, mds);
1673 	doutc(mdsc->fsc->client, "open_session to mds%d (%s)\n", mds,
1674 	      ceph_mds_state_name(mstate));
1675 	session->s_state = CEPH_MDS_SESSION_OPENING;
1676 	session->s_renew_requested = jiffies;
1677 
1678 	/* send connect message */
1679 	msg = create_session_full_msg(mdsc, CEPH_SESSION_REQUEST_OPEN,
1680 				      session->s_seq);
1681 	if (IS_ERR(msg))
1682 		return PTR_ERR(msg);
1683 	ceph_con_send(&session->s_con, msg);
1684 	return 0;
1685 }
1686 
1687 /*
1688  * open sessions for any export targets for the given mds
1689  *
1690  * called under mdsc->mutex
1691  */
1692 static struct ceph_mds_session *
1693 __open_export_target_session(struct ceph_mds_client *mdsc, int target)
1694 {
1695 	struct ceph_mds_session *session;
1696 	int ret;
1697 
1698 	session = __ceph_lookup_mds_session(mdsc, target);
1699 	if (!session) {
1700 		session = register_session(mdsc, target);
1701 		if (IS_ERR(session))
1702 			return session;
1703 	}
1704 	if (session->s_state == CEPH_MDS_SESSION_NEW ||
1705 	    session->s_state == CEPH_MDS_SESSION_CLOSING) {
1706 		ret = __open_session(mdsc, session);
1707 		if (ret)
1708 			return ERR_PTR(ret);
1709 	}
1710 
1711 	return session;
1712 }
1713 
1714 struct ceph_mds_session *
1715 ceph_mdsc_open_export_target_session(struct ceph_mds_client *mdsc, int target)
1716 {
1717 	struct ceph_mds_session *session;
1718 	struct ceph_client *cl = mdsc->fsc->client;
1719 
1720 	doutc(cl, "to mds%d\n", target);
1721 
1722 	mutex_lock(&mdsc->mutex);
1723 	session = __open_export_target_session(mdsc, target);
1724 	mutex_unlock(&mdsc->mutex);
1725 
1726 	return session;
1727 }
1728 
1729 static void __open_export_target_sessions(struct ceph_mds_client *mdsc,
1730 					  struct ceph_mds_session *session)
1731 {
1732 	struct ceph_mds_info *mi;
1733 	struct ceph_mds_session *ts;
1734 	int i, mds = session->s_mds;
1735 	struct ceph_client *cl = mdsc->fsc->client;
1736 
1737 	if (mds >= mdsc->mdsmap->possible_max_rank)
1738 		return;
1739 
1740 	mi = &mdsc->mdsmap->m_info[mds];
1741 	doutc(cl, "for mds%d (%d targets)\n", session->s_mds,
1742 	      mi->num_export_targets);
1743 
1744 	for (i = 0; i < mi->num_export_targets; i++) {
1745 		ts = __open_export_target_session(mdsc, mi->export_targets[i]);
1746 		ceph_put_mds_session(ts);
1747 	}
1748 }
1749 
1750 void ceph_mdsc_open_export_target_sessions(struct ceph_mds_client *mdsc,
1751 					   struct ceph_mds_session *session)
1752 {
1753 	mutex_lock(&mdsc->mutex);
1754 	__open_export_target_sessions(mdsc, session);
1755 	mutex_unlock(&mdsc->mutex);
1756 }
1757 
1758 /*
1759  * session caps
1760  */
1761 
1762 static void detach_cap_releases(struct ceph_mds_session *session,
1763 				struct list_head *target)
1764 {
1765 	struct ceph_client *cl = session->s_mdsc->fsc->client;
1766 
1767 	lockdep_assert_held(&session->s_cap_lock);
1768 
1769 	list_splice_init(&session->s_cap_releases, target);
1770 	session->s_num_cap_releases = 0;
1771 	doutc(cl, "mds%d\n", session->s_mds);
1772 }
1773 
1774 static void dispose_cap_releases(struct ceph_mds_client *mdsc,
1775 				 struct list_head *dispose)
1776 {
1777 	while (!list_empty(dispose)) {
1778 		struct ceph_cap *cap;
1779 		/* zero out the in-progress message */
1780 		cap = list_first_entry(dispose, struct ceph_cap, session_caps);
1781 		list_del(&cap->session_caps);
1782 		ceph_put_cap(mdsc, cap);
1783 	}
1784 }
1785 
1786 static void cleanup_session_requests(struct ceph_mds_client *mdsc,
1787 				     struct ceph_mds_session *session)
1788 {
1789 	struct ceph_client *cl = mdsc->fsc->client;
1790 	struct ceph_mds_request *req;
1791 	struct rb_node *p;
1792 
1793 	doutc(cl, "mds%d\n", session->s_mds);
1794 	mutex_lock(&mdsc->mutex);
1795 	while (!list_empty(&session->s_unsafe)) {
1796 		req = list_first_entry(&session->s_unsafe,
1797 				       struct ceph_mds_request, r_unsafe_item);
1798 		pr_warn_ratelimited_client(cl, " dropping unsafe request %llu\n",
1799 					   req->r_tid);
1800 		if (req->r_target_inode)
1801 			mapping_set_error(req->r_target_inode->i_mapping, -EIO);
1802 		if (req->r_unsafe_dir)
1803 			mapping_set_error(req->r_unsafe_dir->i_mapping, -EIO);
1804 		__unregister_request(mdsc, req);
1805 	}
1806 	/* zero r_attempts, so kick_requests() will re-send requests */
1807 	p = rb_first(&mdsc->request_tree);
1808 	while (p) {
1809 		req = rb_entry(p, struct ceph_mds_request, r_node);
1810 		p = rb_next(p);
1811 		if (req->r_session &&
1812 		    req->r_session->s_mds == session->s_mds)
1813 			req->r_attempts = 0;
1814 	}
1815 	mutex_unlock(&mdsc->mutex);
1816 }
1817 
1818 /*
1819  * Helper to safely iterate over all caps associated with a session, with
1820  * special care taken to handle a racing __ceph_remove_cap().
1821  *
1822  * Caller must hold session s_mutex.
1823  */
1824 int ceph_iterate_session_caps(struct ceph_mds_session *session,
1825 			      int (*cb)(struct inode *, int mds, void *),
1826 			      void *arg)
1827 {
1828 	struct ceph_client *cl = session->s_mdsc->fsc->client;
1829 	struct list_head *p;
1830 	struct ceph_cap *cap;
1831 	struct inode *inode, *last_inode = NULL;
1832 	struct ceph_cap *old_cap = NULL;
1833 	int ret;
1834 
1835 	doutc(cl, "%p mds%d\n", session, session->s_mds);
1836 	spin_lock(&session->s_cap_lock);
1837 	p = session->s_caps.next;
1838 	while (p != &session->s_caps) {
1839 		int mds;
1840 
1841 		cap = list_entry(p, struct ceph_cap, session_caps);
1842 		inode = igrab(&cap->ci->netfs.inode);
1843 		if (!inode) {
1844 			p = p->next;
1845 			continue;
1846 		}
1847 		session->s_cap_iterator = cap;
1848 		mds = cap->mds;
1849 		spin_unlock(&session->s_cap_lock);
1850 
1851 		if (last_inode) {
1852 			iput(last_inode);
1853 			last_inode = NULL;
1854 		}
1855 		if (old_cap) {
1856 			ceph_put_cap(session->s_mdsc, old_cap);
1857 			old_cap = NULL;
1858 		}
1859 
1860 		ret = cb(inode, mds, arg);
1861 		last_inode = inode;
1862 
1863 		spin_lock(&session->s_cap_lock);
1864 		p = p->next;
1865 		if (!cap->ci) {
1866 			doutc(cl, "finishing cap %p removal\n", cap);
1867 			BUG_ON(cap->session != session);
1868 			cap->session = NULL;
1869 			list_del_init(&cap->session_caps);
1870 			session->s_nr_caps--;
1871 			atomic64_dec(&session->s_mdsc->metric.total_caps);
1872 			if (cap->queue_release)
1873 				__ceph_queue_cap_release(session, cap);
1874 			else
1875 				old_cap = cap;  /* put_cap it w/o locks held */
1876 		}
1877 		if (ret < 0)
1878 			goto out;
1879 	}
1880 	ret = 0;
1881 out:
1882 	session->s_cap_iterator = NULL;
1883 	spin_unlock(&session->s_cap_lock);
1884 
1885 	iput(last_inode);
1886 	if (old_cap)
1887 		ceph_put_cap(session->s_mdsc, old_cap);
1888 
1889 	return ret;
1890 }
1891 
1892 static int remove_session_caps_cb(struct inode *inode, int mds, void *arg)
1893 {
1894 	struct ceph_inode_info *ci = ceph_inode(inode);
1895 	struct ceph_client *cl = ceph_inode_to_client(inode);
1896 	bool invalidate = false;
1897 	struct ceph_cap *cap;
1898 	int iputs = 0;
1899 
1900 	spin_lock(&ci->i_ceph_lock);
1901 	cap = __get_cap_for_mds(ci, mds);
1902 	if (cap) {
1903 		doutc(cl, " removing cap %p, ci is %p, inode is %p\n",
1904 		      cap, ci, &ci->netfs.inode);
1905 
1906 		iputs = ceph_purge_inode_cap(inode, cap, &invalidate);
1907 	}
1908 	spin_unlock(&ci->i_ceph_lock);
1909 
1910 	if (cap)
1911 		wake_up_all(&ci->i_cap_wq);
1912 	if (invalidate)
1913 		ceph_queue_invalidate(inode);
1914 	while (iputs--)
1915 		iput(inode);
1916 	return 0;
1917 }
1918 
1919 /*
1920  * caller must hold session s_mutex
1921  */
1922 static void remove_session_caps(struct ceph_mds_session *session)
1923 {
1924 	struct ceph_fs_client *fsc = session->s_mdsc->fsc;
1925 	struct super_block *sb = fsc->sb;
1926 	LIST_HEAD(dispose);
1927 
1928 	doutc(fsc->client, "on %p\n", session);
1929 	ceph_iterate_session_caps(session, remove_session_caps_cb, fsc);
1930 
1931 	wake_up_all(&fsc->mdsc->cap_flushing_wq);
1932 
1933 	spin_lock(&session->s_cap_lock);
1934 	if (session->s_nr_caps > 0) {
1935 		struct inode *inode;
1936 		struct ceph_cap *cap, *prev = NULL;
1937 		struct ceph_vino vino;
1938 		/*
1939 		 * iterate_session_caps() skips inodes that are being
1940 		 * deleted, we need to wait until deletions are complete.
1941 		 * __wait_on_freeing_inode() is designed for the job,
1942 		 * but it is not exported, so use lookup inode function
1943 		 * to access it.
1944 		 */
1945 		while (!list_empty(&session->s_caps)) {
1946 			cap = list_entry(session->s_caps.next,
1947 					 struct ceph_cap, session_caps);
1948 			if (cap == prev)
1949 				break;
1950 			prev = cap;
1951 			vino = cap->ci->i_vino;
1952 			spin_unlock(&session->s_cap_lock);
1953 
1954 			inode = ceph_find_inode(sb, vino);
1955 			iput(inode);
1956 
1957 			spin_lock(&session->s_cap_lock);
1958 		}
1959 	}
1960 
1961 	// drop cap expires and unlock s_cap_lock
1962 	detach_cap_releases(session, &dispose);
1963 
1964 	BUG_ON(session->s_nr_caps > 0);
1965 	BUG_ON(!list_empty(&session->s_cap_flushing));
1966 	spin_unlock(&session->s_cap_lock);
1967 	dispose_cap_releases(session->s_mdsc, &dispose);
1968 }
1969 
1970 enum {
1971 	RECONNECT,
1972 	RENEWCAPS,
1973 	FORCE_RO,
1974 };
1975 
1976 /*
1977  * wake up any threads waiting on this session's caps.  if the cap is
1978  * old (didn't get renewed on the client reconnect), remove it now.
1979  *
1980  * caller must hold s_mutex.
1981  */
1982 static int wake_up_session_cb(struct inode *inode, int mds, void *arg)
1983 {
1984 	struct ceph_inode_info *ci = ceph_inode(inode);
1985 	unsigned long ev = (unsigned long)arg;
1986 
1987 	if (ev == RECONNECT) {
1988 		spin_lock(&ci->i_ceph_lock);
1989 		ci->i_wanted_max_size = 0;
1990 		ci->i_requested_max_size = 0;
1991 		spin_unlock(&ci->i_ceph_lock);
1992 	} else if (ev == RENEWCAPS) {
1993 		struct ceph_cap *cap;
1994 
1995 		spin_lock(&ci->i_ceph_lock);
1996 		cap = __get_cap_for_mds(ci, mds);
1997 		/* mds did not re-issue stale cap */
1998 		if (cap && cap->cap_gen < atomic_read(&cap->session->s_cap_gen))
1999 			cap->issued = cap->implemented = CEPH_CAP_PIN;
2000 		spin_unlock(&ci->i_ceph_lock);
2001 	} else if (ev == FORCE_RO) {
2002 	}
2003 	wake_up_all(&ci->i_cap_wq);
2004 	return 0;
2005 }
2006 
2007 static void wake_up_session_caps(struct ceph_mds_session *session, int ev)
2008 {
2009 	struct ceph_client *cl = session->s_mdsc->fsc->client;
2010 
2011 	doutc(cl, "session %p mds%d\n", session, session->s_mds);
2012 	ceph_iterate_session_caps(session, wake_up_session_cb,
2013 				  (void *)(unsigned long)ev);
2014 }
2015 
2016 /*
2017  * Send periodic message to MDS renewing all currently held caps.  The
2018  * ack will reset the expiration for all caps from this session.
2019  *
2020  * caller holds s_mutex
2021  */
2022 static int send_renew_caps(struct ceph_mds_client *mdsc,
2023 			   struct ceph_mds_session *session)
2024 {
2025 	struct ceph_client *cl = mdsc->fsc->client;
2026 	struct ceph_msg *msg;
2027 	int state;
2028 
2029 	if (time_after_eq(jiffies, session->s_cap_ttl) &&
2030 	    time_after_eq(session->s_cap_ttl, session->s_renew_requested))
2031 		pr_info_client(cl, "mds%d caps stale\n", session->s_mds);
2032 	session->s_renew_requested = jiffies;
2033 
2034 	/* do not try to renew caps until a recovering mds has reconnected
2035 	 * with its clients. */
2036 	state = ceph_mdsmap_get_state(mdsc->mdsmap, session->s_mds);
2037 	if (state < CEPH_MDS_STATE_RECONNECT) {
2038 		doutc(cl, "ignoring mds%d (%s)\n", session->s_mds,
2039 		      ceph_mds_state_name(state));
2040 		return 0;
2041 	}
2042 
2043 	doutc(cl, "to mds%d (%s)\n", session->s_mds,
2044 	      ceph_mds_state_name(state));
2045 	msg = create_session_full_msg(mdsc, CEPH_SESSION_REQUEST_RENEWCAPS,
2046 				      ++session->s_renew_seq);
2047 	if (IS_ERR(msg))
2048 		return PTR_ERR(msg);
2049 	ceph_con_send(&session->s_con, msg);
2050 	return 0;
2051 }
2052 
2053 static int send_flushmsg_ack(struct ceph_mds_client *mdsc,
2054 			     struct ceph_mds_session *session, u64 seq)
2055 {
2056 	struct ceph_client *cl = mdsc->fsc->client;
2057 	struct ceph_msg *msg;
2058 
2059 	doutc(cl, "to mds%d (%s)s seq %lld\n", session->s_mds,
2060 	      ceph_session_state_name(session->s_state), seq);
2061 	msg = ceph_create_session_msg(CEPH_SESSION_FLUSHMSG_ACK, seq);
2062 	if (!msg)
2063 		return -ENOMEM;
2064 	ceph_con_send(&session->s_con, msg);
2065 	return 0;
2066 }
2067 
2068 
2069 /*
2070  * Note new cap ttl, and any transition from stale -> not stale (fresh?).
2071  *
2072  * Called under session->s_mutex
2073  */
2074 static void renewed_caps(struct ceph_mds_client *mdsc,
2075 			 struct ceph_mds_session *session, int is_renew)
2076 {
2077 	struct ceph_client *cl = mdsc->fsc->client;
2078 	int was_stale;
2079 	int wake = 0;
2080 
2081 	spin_lock(&session->s_cap_lock);
2082 	was_stale = is_renew && time_after_eq(jiffies, session->s_cap_ttl);
2083 
2084 	session->s_cap_ttl = session->s_renew_requested +
2085 		mdsc->mdsmap->m_session_timeout*HZ;
2086 
2087 	if (was_stale) {
2088 		if (time_before(jiffies, session->s_cap_ttl)) {
2089 			pr_info_client(cl, "mds%d caps renewed\n",
2090 				       session->s_mds);
2091 			wake = 1;
2092 		} else {
2093 			pr_info_client(cl, "mds%d caps still stale\n",
2094 				       session->s_mds);
2095 		}
2096 	}
2097 	doutc(cl, "mds%d ttl now %lu, was %s, now %s\n", session->s_mds,
2098 	      session->s_cap_ttl, was_stale ? "stale" : "fresh",
2099 	      time_before(jiffies, session->s_cap_ttl) ? "stale" : "fresh");
2100 	spin_unlock(&session->s_cap_lock);
2101 
2102 	if (wake)
2103 		wake_up_session_caps(session, RENEWCAPS);
2104 }
2105 
2106 /*
2107  * send a session close request
2108  */
2109 static int request_close_session(struct ceph_mds_session *session)
2110 {
2111 	struct ceph_client *cl = session->s_mdsc->fsc->client;
2112 	struct ceph_msg *msg;
2113 
2114 	doutc(cl, "mds%d state %s seq %lld\n", session->s_mds,
2115 	      ceph_session_state_name(session->s_state), session->s_seq);
2116 	msg = ceph_create_session_msg(CEPH_SESSION_REQUEST_CLOSE,
2117 				      session->s_seq);
2118 	if (!msg)
2119 		return -ENOMEM;
2120 	ceph_con_send(&session->s_con, msg);
2121 	return 1;
2122 }
2123 
2124 /*
2125  * Called with s_mutex held.
2126  */
2127 static int __close_session(struct ceph_mds_client *mdsc,
2128 			 struct ceph_mds_session *session)
2129 {
2130 	if (session->s_state >= CEPH_MDS_SESSION_CLOSING)
2131 		return 0;
2132 	session->s_state = CEPH_MDS_SESSION_CLOSING;
2133 	return request_close_session(session);
2134 }
2135 
2136 static bool drop_negative_children(struct dentry *dentry)
2137 {
2138 	struct dentry *child;
2139 	bool all_negative = true;
2140 
2141 	if (!d_is_dir(dentry))
2142 		goto out;
2143 
2144 	spin_lock(&dentry->d_lock);
2145 	hlist_for_each_entry(child, &dentry->d_children, d_sib) {
2146 		if (d_really_is_positive(child)) {
2147 			all_negative = false;
2148 			break;
2149 		}
2150 	}
2151 	spin_unlock(&dentry->d_lock);
2152 
2153 	if (all_negative)
2154 		shrink_dcache_parent(dentry);
2155 out:
2156 	return all_negative;
2157 }
2158 
2159 /*
2160  * Trim old(er) caps.
2161  *
2162  * Because we can't cache an inode without one or more caps, we do
2163  * this indirectly: if a cap is unused, we prune its aliases, at which
2164  * point the inode will hopefully get dropped to.
2165  *
2166  * Yes, this is a bit sloppy.  Our only real goal here is to respond to
2167  * memory pressure from the MDS, though, so it needn't be perfect.
2168  */
2169 static int trim_caps_cb(struct inode *inode, int mds, void *arg)
2170 {
2171 	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
2172 	struct ceph_client *cl = mdsc->fsc->client;
2173 	int *remaining = arg;
2174 	struct ceph_inode_info *ci = ceph_inode(inode);
2175 	int used, wanted, oissued, mine;
2176 	struct ceph_cap *cap;
2177 
2178 	if (*remaining <= 0)
2179 		return -1;
2180 
2181 	spin_lock(&ci->i_ceph_lock);
2182 	cap = __get_cap_for_mds(ci, mds);
2183 	if (!cap) {
2184 		spin_unlock(&ci->i_ceph_lock);
2185 		return 0;
2186 	}
2187 	mine = cap->issued | cap->implemented;
2188 	used = __ceph_caps_used(ci);
2189 	wanted = __ceph_caps_file_wanted(ci);
2190 	oissued = __ceph_caps_issued_other(ci, cap);
2191 
2192 	doutc(cl, "%p %llx.%llx cap %p mine %s oissued %s used %s wanted %s\n",
2193 	      inode, ceph_vinop(inode), cap, ceph_cap_string(mine),
2194 	      ceph_cap_string(oissued), ceph_cap_string(used),
2195 	      ceph_cap_string(wanted));
2196 	if (cap == ci->i_auth_cap) {
2197 		if (ci->i_dirty_caps || ci->i_flushing_caps ||
2198 		    !list_empty(&ci->i_cap_snaps))
2199 			goto out;
2200 		if ((used | wanted) & CEPH_CAP_ANY_WR)
2201 			goto out;
2202 		/* Note: it's possible that i_filelock_ref becomes non-zero
2203 		 * after dropping auth caps. It doesn't hurt because reply
2204 		 * of lock mds request will re-add auth caps. */
2205 		if (atomic_read(&ci->i_filelock_ref) > 0)
2206 			goto out;
2207 	}
2208 	/* The inode has cached pages, but it's no longer used.
2209 	 * we can safely drop it */
2210 	if (S_ISREG(inode->i_mode) &&
2211 	    wanted == 0 && used == CEPH_CAP_FILE_CACHE &&
2212 	    !(oissued & CEPH_CAP_FILE_CACHE)) {
2213 	  used = 0;
2214 	  oissued = 0;
2215 	}
2216 	if ((used | wanted) & ~oissued & mine)
2217 		goto out;   /* we need these caps */
2218 
2219 	if (oissued) {
2220 		/* we aren't the only cap.. just remove us */
2221 		ceph_remove_cap(mdsc, cap, true);
2222 		(*remaining)--;
2223 	} else {
2224 		struct dentry *dentry;
2225 		/* try dropping referring dentries */
2226 		spin_unlock(&ci->i_ceph_lock);
2227 		dentry = d_find_any_alias(inode);
2228 		if (dentry && drop_negative_children(dentry)) {
2229 			int count;
2230 			dput(dentry);
2231 			d_prune_aliases(inode);
2232 			count = atomic_read(&inode->i_count);
2233 			if (count == 1)
2234 				(*remaining)--;
2235 			doutc(cl, "%p %llx.%llx cap %p pruned, count now %d\n",
2236 			      inode, ceph_vinop(inode), cap, count);
2237 		} else {
2238 			dput(dentry);
2239 		}
2240 		return 0;
2241 	}
2242 
2243 out:
2244 	spin_unlock(&ci->i_ceph_lock);
2245 	return 0;
2246 }
2247 
2248 /*
2249  * Trim session cap count down to some max number.
2250  */
2251 int ceph_trim_caps(struct ceph_mds_client *mdsc,
2252 		   struct ceph_mds_session *session,
2253 		   int max_caps)
2254 {
2255 	struct ceph_client *cl = mdsc->fsc->client;
2256 	int trim_caps = session->s_nr_caps - max_caps;
2257 
2258 	doutc(cl, "mds%d start: %d / %d, trim %d\n", session->s_mds,
2259 	      session->s_nr_caps, max_caps, trim_caps);
2260 	if (trim_caps > 0) {
2261 		int remaining = trim_caps;
2262 
2263 		ceph_iterate_session_caps(session, trim_caps_cb, &remaining);
2264 		doutc(cl, "mds%d done: %d / %d, trimmed %d\n",
2265 		      session->s_mds, session->s_nr_caps, max_caps,
2266 		      trim_caps - remaining);
2267 	}
2268 
2269 	ceph_flush_session_cap_releases(mdsc, session);
2270 	return 0;
2271 }
2272 
2273 static int check_caps_flush(struct ceph_mds_client *mdsc,
2274 			    u64 want_flush_tid)
2275 {
2276 	struct ceph_client *cl = mdsc->fsc->client;
2277 	int ret = 1;
2278 
2279 	spin_lock(&mdsc->cap_dirty_lock);
2280 	if (!list_empty(&mdsc->cap_flush_list)) {
2281 		struct ceph_cap_flush *cf =
2282 			list_first_entry(&mdsc->cap_flush_list,
2283 					 struct ceph_cap_flush, g_list);
2284 		if (cf->tid <= want_flush_tid) {
2285 			doutc(cl, "still flushing tid %llu <= %llu\n",
2286 			      cf->tid, want_flush_tid);
2287 			ret = 0;
2288 		}
2289 	}
2290 	spin_unlock(&mdsc->cap_dirty_lock);
2291 	return ret;
2292 }
2293 
2294 /*
2295  * flush all dirty inode data to disk.
2296  *
2297  * returns true if we've flushed through want_flush_tid
2298  */
2299 static void wait_caps_flush(struct ceph_mds_client *mdsc,
2300 			    u64 want_flush_tid)
2301 {
2302 	struct ceph_client *cl = mdsc->fsc->client;
2303 
2304 	doutc(cl, "want %llu\n", want_flush_tid);
2305 
2306 	wait_event(mdsc->cap_flushing_wq,
2307 		   check_caps_flush(mdsc, want_flush_tid));
2308 
2309 	doutc(cl, "ok, flushed thru %llu\n", want_flush_tid);
2310 }
2311 
2312 /*
2313  * called under s_mutex
2314  */
2315 static void ceph_send_cap_releases(struct ceph_mds_client *mdsc,
2316 				   struct ceph_mds_session *session)
2317 {
2318 	struct ceph_client *cl = mdsc->fsc->client;
2319 	struct ceph_msg *msg = NULL;
2320 	struct ceph_mds_cap_release *head;
2321 	struct ceph_mds_cap_item *item;
2322 	struct ceph_osd_client *osdc = &mdsc->fsc->client->osdc;
2323 	struct ceph_cap *cap;
2324 	LIST_HEAD(tmp_list);
2325 	int num_cap_releases;
2326 	__le32	barrier, *cap_barrier;
2327 
2328 	down_read(&osdc->lock);
2329 	barrier = cpu_to_le32(osdc->epoch_barrier);
2330 	up_read(&osdc->lock);
2331 
2332 	spin_lock(&session->s_cap_lock);
2333 again:
2334 	list_splice_init(&session->s_cap_releases, &tmp_list);
2335 	num_cap_releases = session->s_num_cap_releases;
2336 	session->s_num_cap_releases = 0;
2337 	spin_unlock(&session->s_cap_lock);
2338 
2339 	while (!list_empty(&tmp_list)) {
2340 		if (!msg) {
2341 			msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPRELEASE,
2342 					PAGE_SIZE, GFP_NOFS, false);
2343 			if (!msg)
2344 				goto out_err;
2345 			head = msg->front.iov_base;
2346 			head->num = cpu_to_le32(0);
2347 			msg->front.iov_len = sizeof(*head);
2348 
2349 			msg->hdr.version = cpu_to_le16(2);
2350 			msg->hdr.compat_version = cpu_to_le16(1);
2351 		}
2352 
2353 		cap = list_first_entry(&tmp_list, struct ceph_cap,
2354 					session_caps);
2355 		list_del(&cap->session_caps);
2356 		num_cap_releases--;
2357 
2358 		head = msg->front.iov_base;
2359 		put_unaligned_le32(get_unaligned_le32(&head->num) + 1,
2360 				   &head->num);
2361 		item = msg->front.iov_base + msg->front.iov_len;
2362 		item->ino = cpu_to_le64(cap->cap_ino);
2363 		item->cap_id = cpu_to_le64(cap->cap_id);
2364 		item->migrate_seq = cpu_to_le32(cap->mseq);
2365 		item->seq = cpu_to_le32(cap->issue_seq);
2366 		msg->front.iov_len += sizeof(*item);
2367 
2368 		ceph_put_cap(mdsc, cap);
2369 
2370 		if (le32_to_cpu(head->num) == CEPH_CAPS_PER_RELEASE) {
2371 			// Append cap_barrier field
2372 			cap_barrier = msg->front.iov_base + msg->front.iov_len;
2373 			*cap_barrier = barrier;
2374 			msg->front.iov_len += sizeof(*cap_barrier);
2375 
2376 			msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
2377 			doutc(cl, "mds%d %p\n", session->s_mds, msg);
2378 			ceph_con_send(&session->s_con, msg);
2379 			msg = NULL;
2380 		}
2381 	}
2382 
2383 	BUG_ON(num_cap_releases != 0);
2384 
2385 	spin_lock(&session->s_cap_lock);
2386 	if (!list_empty(&session->s_cap_releases))
2387 		goto again;
2388 	spin_unlock(&session->s_cap_lock);
2389 
2390 	if (msg) {
2391 		// Append cap_barrier field
2392 		cap_barrier = msg->front.iov_base + msg->front.iov_len;
2393 		*cap_barrier = barrier;
2394 		msg->front.iov_len += sizeof(*cap_barrier);
2395 
2396 		msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
2397 		doutc(cl, "mds%d %p\n", session->s_mds, msg);
2398 		ceph_con_send(&session->s_con, msg);
2399 	}
2400 	return;
2401 out_err:
2402 	pr_err_client(cl, "mds%d, failed to allocate message\n",
2403 		      session->s_mds);
2404 	spin_lock(&session->s_cap_lock);
2405 	list_splice(&tmp_list, &session->s_cap_releases);
2406 	session->s_num_cap_releases += num_cap_releases;
2407 	spin_unlock(&session->s_cap_lock);
2408 }
2409 
2410 static void ceph_cap_release_work(struct work_struct *work)
2411 {
2412 	struct ceph_mds_session *session =
2413 		container_of(work, struct ceph_mds_session, s_cap_release_work);
2414 
2415 	mutex_lock(&session->s_mutex);
2416 	if (session->s_state == CEPH_MDS_SESSION_OPEN ||
2417 	    session->s_state == CEPH_MDS_SESSION_HUNG)
2418 		ceph_send_cap_releases(session->s_mdsc, session);
2419 	mutex_unlock(&session->s_mutex);
2420 	ceph_put_mds_session(session);
2421 }
2422 
2423 void ceph_flush_session_cap_releases(struct ceph_mds_client *mdsc,
2424 		             struct ceph_mds_session *session)
2425 {
2426 	struct ceph_client *cl = mdsc->fsc->client;
2427 	if (mdsc->stopping)
2428 		return;
2429 
2430 	ceph_get_mds_session(session);
2431 	if (queue_work(mdsc->fsc->cap_wq,
2432 		       &session->s_cap_release_work)) {
2433 		doutc(cl, "cap release work queued\n");
2434 	} else {
2435 		ceph_put_mds_session(session);
2436 		doutc(cl, "failed to queue cap release work\n");
2437 	}
2438 }
2439 
2440 /*
2441  * caller holds session->s_cap_lock
2442  */
2443 void __ceph_queue_cap_release(struct ceph_mds_session *session,
2444 			      struct ceph_cap *cap)
2445 {
2446 	list_add_tail(&cap->session_caps, &session->s_cap_releases);
2447 	session->s_num_cap_releases++;
2448 
2449 	if (!(session->s_num_cap_releases % CEPH_CAPS_PER_RELEASE))
2450 		ceph_flush_session_cap_releases(session->s_mdsc, session);
2451 }
2452 
2453 static void ceph_cap_reclaim_work(struct work_struct *work)
2454 {
2455 	struct ceph_mds_client *mdsc =
2456 		container_of(work, struct ceph_mds_client, cap_reclaim_work);
2457 	int ret = ceph_trim_dentries(mdsc);
2458 	if (ret == -EAGAIN)
2459 		ceph_queue_cap_reclaim_work(mdsc);
2460 }
2461 
2462 void ceph_queue_cap_reclaim_work(struct ceph_mds_client *mdsc)
2463 {
2464 	struct ceph_client *cl = mdsc->fsc->client;
2465 	if (mdsc->stopping)
2466 		return;
2467 
2468         if (queue_work(mdsc->fsc->cap_wq, &mdsc->cap_reclaim_work)) {
2469                 doutc(cl, "caps reclaim work queued\n");
2470         } else {
2471                 doutc(cl, "failed to queue caps release work\n");
2472         }
2473 }
2474 
2475 void ceph_reclaim_caps_nr(struct ceph_mds_client *mdsc, int nr)
2476 {
2477 	int val;
2478 	if (!nr)
2479 		return;
2480 	val = atomic_add_return(nr, &mdsc->cap_reclaim_pending);
2481 	if ((val % CEPH_CAPS_PER_RELEASE) < nr) {
2482 		atomic_set(&mdsc->cap_reclaim_pending, 0);
2483 		ceph_queue_cap_reclaim_work(mdsc);
2484 	}
2485 }
2486 
2487 void ceph_queue_cap_unlink_work(struct ceph_mds_client *mdsc)
2488 {
2489 	struct ceph_client *cl = mdsc->fsc->client;
2490 	if (mdsc->stopping)
2491 		return;
2492 
2493         if (queue_work(mdsc->fsc->cap_wq, &mdsc->cap_unlink_work)) {
2494                 doutc(cl, "caps unlink work queued\n");
2495         } else {
2496                 doutc(cl, "failed to queue caps unlink work\n");
2497         }
2498 }
2499 
2500 static void ceph_cap_unlink_work(struct work_struct *work)
2501 {
2502 	struct ceph_mds_client *mdsc =
2503 		container_of(work, struct ceph_mds_client, cap_unlink_work);
2504 	struct ceph_client *cl = mdsc->fsc->client;
2505 
2506 	doutc(cl, "begin\n");
2507 	spin_lock(&mdsc->cap_delay_lock);
2508 	while (!list_empty(&mdsc->cap_unlink_delay_list)) {
2509 		struct ceph_inode_info *ci;
2510 		struct inode *inode;
2511 
2512 		ci = list_first_entry(&mdsc->cap_unlink_delay_list,
2513 				      struct ceph_inode_info,
2514 				      i_cap_delay_list);
2515 		list_del_init(&ci->i_cap_delay_list);
2516 
2517 		inode = igrab(&ci->netfs.inode);
2518 		if (inode) {
2519 			spin_unlock(&mdsc->cap_delay_lock);
2520 			doutc(cl, "on %p %llx.%llx\n", inode,
2521 			      ceph_vinop(inode));
2522 			ceph_check_caps(ci, CHECK_CAPS_FLUSH);
2523 			iput(inode);
2524 			spin_lock(&mdsc->cap_delay_lock);
2525 		}
2526 	}
2527 	spin_unlock(&mdsc->cap_delay_lock);
2528 	doutc(cl, "done\n");
2529 }
2530 
2531 /*
2532  * requests
2533  */
2534 
2535 int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
2536 				    struct inode *dir)
2537 {
2538 	struct ceph_inode_info *ci = ceph_inode(dir);
2539 	struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
2540 	struct ceph_mount_options *opt = req->r_mdsc->fsc->mount_options;
2541 	size_t size = sizeof(struct ceph_mds_reply_dir_entry);
2542 	unsigned int num_entries;
2543 	int order;
2544 
2545 	spin_lock(&ci->i_ceph_lock);
2546 	num_entries = ci->i_files + ci->i_subdirs;
2547 	spin_unlock(&ci->i_ceph_lock);
2548 	num_entries = max(num_entries, 1U);
2549 	num_entries = min(num_entries, opt->max_readdir);
2550 
2551 	order = get_order(size * num_entries);
2552 	while (order >= 0) {
2553 		rinfo->dir_entries = (void*)__get_free_pages(GFP_KERNEL |
2554 							     __GFP_NOWARN |
2555 							     __GFP_ZERO,
2556 							     order);
2557 		if (rinfo->dir_entries)
2558 			break;
2559 		order--;
2560 	}
2561 	if (!rinfo->dir_entries)
2562 		return -ENOMEM;
2563 
2564 	num_entries = (PAGE_SIZE << order) / size;
2565 	num_entries = min(num_entries, opt->max_readdir);
2566 
2567 	rinfo->dir_buf_size = PAGE_SIZE << order;
2568 	req->r_num_caps = num_entries + 1;
2569 	req->r_args.readdir.max_entries = cpu_to_le32(num_entries);
2570 	req->r_args.readdir.max_bytes = cpu_to_le32(opt->max_readdir_bytes);
2571 	return 0;
2572 }
2573 
2574 /*
2575  * Create an mds request.
2576  */
2577 struct ceph_mds_request *
2578 ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode)
2579 {
2580 	struct ceph_mds_request *req;
2581 
2582 	req = kmem_cache_zalloc(ceph_mds_request_cachep, GFP_NOFS);
2583 	if (!req)
2584 		return ERR_PTR(-ENOMEM);
2585 
2586 	mutex_init(&req->r_fill_mutex);
2587 	req->r_mdsc = mdsc;
2588 	req->r_started = jiffies;
2589 	req->r_start_latency = ktime_get();
2590 	req->r_resend_mds = -1;
2591 	INIT_LIST_HEAD(&req->r_unsafe_dir_item);
2592 	INIT_LIST_HEAD(&req->r_unsafe_target_item);
2593 	req->r_fmode = -1;
2594 	req->r_feature_needed = -1;
2595 	kref_init(&req->r_kref);
2596 	RB_CLEAR_NODE(&req->r_node);
2597 	INIT_LIST_HEAD(&req->r_wait);
2598 	init_completion(&req->r_completion);
2599 	init_completion(&req->r_safe_completion);
2600 	INIT_LIST_HEAD(&req->r_unsafe_item);
2601 
2602 	ktime_get_coarse_real_ts64(&req->r_stamp);
2603 
2604 	req->r_op = op;
2605 	req->r_direct_mode = mode;
2606 	return req;
2607 }
2608 
2609 /*
2610  * return oldest (lowest) request, tid in request tree, 0 if none.
2611  *
2612  * called under mdsc->mutex.
2613  */
2614 static struct ceph_mds_request *__get_oldest_req(struct ceph_mds_client *mdsc)
2615 {
2616 	if (RB_EMPTY_ROOT(&mdsc->request_tree))
2617 		return NULL;
2618 	return rb_entry(rb_first(&mdsc->request_tree),
2619 			struct ceph_mds_request, r_node);
2620 }
2621 
2622 static inline  u64 __get_oldest_tid(struct ceph_mds_client *mdsc)
2623 {
2624 	return mdsc->oldest_tid;
2625 }
2626 
2627 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
2628 static u8 *get_fscrypt_altname(const struct ceph_mds_request *req, u32 *plen)
2629 {
2630 	struct inode *dir = req->r_parent;
2631 	struct dentry *dentry = req->r_dentry;
2632 	u8 *cryptbuf = NULL;
2633 	u32 len = 0;
2634 	int ret = 0;
2635 
2636 	/* only encode if we have parent and dentry */
2637 	if (!dir || !dentry)
2638 		goto success;
2639 
2640 	/* No-op unless this is encrypted */
2641 	if (!IS_ENCRYPTED(dir))
2642 		goto success;
2643 
2644 	ret = ceph_fscrypt_prepare_readdir(dir);
2645 	if (ret < 0)
2646 		return ERR_PTR(ret);
2647 
2648 	/* No key? Just ignore it. */
2649 	if (!fscrypt_has_encryption_key(dir))
2650 		goto success;
2651 
2652 	if (!fscrypt_fname_encrypted_size(dir, dentry->d_name.len, NAME_MAX,
2653 					  &len)) {
2654 		WARN_ON_ONCE(1);
2655 		return ERR_PTR(-ENAMETOOLONG);
2656 	}
2657 
2658 	/* No need to append altname if name is short enough */
2659 	if (len <= CEPH_NOHASH_NAME_MAX) {
2660 		len = 0;
2661 		goto success;
2662 	}
2663 
2664 	cryptbuf = kmalloc(len, GFP_KERNEL);
2665 	if (!cryptbuf)
2666 		return ERR_PTR(-ENOMEM);
2667 
2668 	ret = fscrypt_fname_encrypt(dir, &dentry->d_name, cryptbuf, len);
2669 	if (ret) {
2670 		kfree(cryptbuf);
2671 		return ERR_PTR(ret);
2672 	}
2673 success:
2674 	*plen = len;
2675 	return cryptbuf;
2676 }
2677 #else
2678 static u8 *get_fscrypt_altname(const struct ceph_mds_request *req, u32 *plen)
2679 {
2680 	*plen = 0;
2681 	return NULL;
2682 }
2683 #endif
2684 
2685 /**
2686  * ceph_mdsc_build_path - build a path string to a given dentry
2687  * @mdsc: mds client
2688  * @dentry: dentry to which path should be built
2689  * @plen: returned length of string
2690  * @pbase: returned base inode number
2691  * @for_wire: is this path going to be sent to the MDS?
2692  *
2693  * Build a string that represents the path to the dentry. This is mostly called
2694  * for two different purposes:
2695  *
2696  * 1) we need to build a path string to send to the MDS (for_wire == true)
2697  * 2) we need a path string for local presentation (e.g. debugfs)
2698  *    (for_wire == false)
2699  *
2700  * The path is built in reverse, starting with the dentry. Walk back up toward
2701  * the root, building the path until the first non-snapped inode is reached
2702  * (for_wire) or the root inode is reached (!for_wire).
2703  *
2704  * Encode hidden .snap dirs as a double /, i.e.
2705  *   foo/.snap/bar -> foo//bar
2706  */
2707 char *ceph_mdsc_build_path(struct ceph_mds_client *mdsc, struct dentry *dentry,
2708 			   int *plen, u64 *pbase, int for_wire)
2709 {
2710 	struct ceph_client *cl = mdsc->fsc->client;
2711 	struct dentry *cur;
2712 	struct inode *inode;
2713 	char *path;
2714 	int pos;
2715 	unsigned seq;
2716 	u64 base;
2717 
2718 	if (!dentry)
2719 		return ERR_PTR(-EINVAL);
2720 
2721 	path = __getname();
2722 	if (!path)
2723 		return ERR_PTR(-ENOMEM);
2724 retry:
2725 	pos = PATH_MAX - 1;
2726 	path[pos] = '\0';
2727 
2728 	seq = read_seqbegin(&rename_lock);
2729 	cur = dget(dentry);
2730 	for (;;) {
2731 		struct dentry *parent;
2732 
2733 		spin_lock(&cur->d_lock);
2734 		inode = d_inode(cur);
2735 		if (inode && ceph_snap(inode) == CEPH_SNAPDIR) {
2736 			doutc(cl, "path+%d: %p SNAPDIR\n", pos, cur);
2737 			spin_unlock(&cur->d_lock);
2738 			parent = dget_parent(cur);
2739 		} else if (for_wire && inode && dentry != cur &&
2740 			   ceph_snap(inode) == CEPH_NOSNAP) {
2741 			spin_unlock(&cur->d_lock);
2742 			pos++; /* get rid of any prepended '/' */
2743 			break;
2744 		} else if (!for_wire || !IS_ENCRYPTED(d_inode(cur->d_parent))) {
2745 			pos -= cur->d_name.len;
2746 			if (pos < 0) {
2747 				spin_unlock(&cur->d_lock);
2748 				break;
2749 			}
2750 			memcpy(path + pos, cur->d_name.name, cur->d_name.len);
2751 			spin_unlock(&cur->d_lock);
2752 			parent = dget_parent(cur);
2753 		} else {
2754 			int len, ret;
2755 			char buf[NAME_MAX];
2756 
2757 			/*
2758 			 * Proactively copy name into buf, in case we need to
2759 			 * present it as-is.
2760 			 */
2761 			memcpy(buf, cur->d_name.name, cur->d_name.len);
2762 			len = cur->d_name.len;
2763 			spin_unlock(&cur->d_lock);
2764 			parent = dget_parent(cur);
2765 
2766 			ret = ceph_fscrypt_prepare_readdir(d_inode(parent));
2767 			if (ret < 0) {
2768 				dput(parent);
2769 				dput(cur);
2770 				return ERR_PTR(ret);
2771 			}
2772 
2773 			if (fscrypt_has_encryption_key(d_inode(parent))) {
2774 				len = ceph_encode_encrypted_fname(d_inode(parent),
2775 								  cur, buf);
2776 				if (len < 0) {
2777 					dput(parent);
2778 					dput(cur);
2779 					return ERR_PTR(len);
2780 				}
2781 			}
2782 			pos -= len;
2783 			if (pos < 0) {
2784 				dput(parent);
2785 				break;
2786 			}
2787 			memcpy(path + pos, buf, len);
2788 		}
2789 		dput(cur);
2790 		cur = parent;
2791 
2792 		/* Are we at the root? */
2793 		if (IS_ROOT(cur))
2794 			break;
2795 
2796 		/* Are we out of buffer? */
2797 		if (--pos < 0)
2798 			break;
2799 
2800 		path[pos] = '/';
2801 	}
2802 	inode = d_inode(cur);
2803 	base = inode ? ceph_ino(inode) : 0;
2804 	dput(cur);
2805 
2806 	if (read_seqretry(&rename_lock, seq))
2807 		goto retry;
2808 
2809 	if (pos < 0) {
2810 		/*
2811 		 * A rename didn't occur, but somehow we didn't end up where
2812 		 * we thought we would. Throw a warning and try again.
2813 		 */
2814 		pr_warn_client(cl, "did not end path lookup where expected (pos = %d)\n",
2815 			       pos);
2816 		goto retry;
2817 	}
2818 
2819 	*pbase = base;
2820 	*plen = PATH_MAX - 1 - pos;
2821 	doutc(cl, "on %p %d built %llx '%.*s'\n", dentry, d_count(dentry),
2822 	      base, *plen, path + pos);
2823 	return path + pos;
2824 }
2825 
2826 static int build_dentry_path(struct ceph_mds_client *mdsc, struct dentry *dentry,
2827 			     struct inode *dir, const char **ppath, int *ppathlen,
2828 			     u64 *pino, bool *pfreepath, bool parent_locked)
2829 {
2830 	char *path;
2831 
2832 	rcu_read_lock();
2833 	if (!dir)
2834 		dir = d_inode_rcu(dentry->d_parent);
2835 	if (dir && parent_locked && ceph_snap(dir) == CEPH_NOSNAP &&
2836 	    !IS_ENCRYPTED(dir)) {
2837 		*pino = ceph_ino(dir);
2838 		rcu_read_unlock();
2839 		*ppath = dentry->d_name.name;
2840 		*ppathlen = dentry->d_name.len;
2841 		return 0;
2842 	}
2843 	rcu_read_unlock();
2844 	path = ceph_mdsc_build_path(mdsc, dentry, ppathlen, pino, 1);
2845 	if (IS_ERR(path))
2846 		return PTR_ERR(path);
2847 	*ppath = path;
2848 	*pfreepath = true;
2849 	return 0;
2850 }
2851 
2852 static int build_inode_path(struct inode *inode,
2853 			    const char **ppath, int *ppathlen, u64 *pino,
2854 			    bool *pfreepath)
2855 {
2856 	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
2857 	struct dentry *dentry;
2858 	char *path;
2859 
2860 	if (ceph_snap(inode) == CEPH_NOSNAP) {
2861 		*pino = ceph_ino(inode);
2862 		*ppathlen = 0;
2863 		return 0;
2864 	}
2865 	dentry = d_find_alias(inode);
2866 	path = ceph_mdsc_build_path(mdsc, dentry, ppathlen, pino, 1);
2867 	dput(dentry);
2868 	if (IS_ERR(path))
2869 		return PTR_ERR(path);
2870 	*ppath = path;
2871 	*pfreepath = true;
2872 	return 0;
2873 }
2874 
2875 /*
2876  * request arguments may be specified via an inode *, a dentry *, or
2877  * an explicit ino+path.
2878  */
2879 static int set_request_path_attr(struct ceph_mds_client *mdsc, struct inode *rinode,
2880 				 struct dentry *rdentry, struct inode *rdiri,
2881 				 const char *rpath, u64 rino, const char **ppath,
2882 				 int *pathlen, u64 *ino, bool *freepath,
2883 				 bool parent_locked)
2884 {
2885 	struct ceph_client *cl = mdsc->fsc->client;
2886 	int r = 0;
2887 
2888 	if (rinode) {
2889 		r = build_inode_path(rinode, ppath, pathlen, ino, freepath);
2890 		doutc(cl, " inode %p %llx.%llx\n", rinode, ceph_ino(rinode),
2891 		      ceph_snap(rinode));
2892 	} else if (rdentry) {
2893 		r = build_dentry_path(mdsc, rdentry, rdiri, ppath, pathlen, ino,
2894 					freepath, parent_locked);
2895 		doutc(cl, " dentry %p %llx/%.*s\n", rdentry, *ino, *pathlen, *ppath);
2896 	} else if (rpath || rino) {
2897 		*ino = rino;
2898 		*ppath = rpath;
2899 		*pathlen = rpath ? strlen(rpath) : 0;
2900 		doutc(cl, " path %.*s\n", *pathlen, rpath);
2901 	}
2902 
2903 	return r;
2904 }
2905 
2906 static void encode_mclientrequest_tail(void **p,
2907 				       const struct ceph_mds_request *req)
2908 {
2909 	struct ceph_timespec ts;
2910 	int i;
2911 
2912 	ceph_encode_timespec64(&ts, &req->r_stamp);
2913 	ceph_encode_copy(p, &ts, sizeof(ts));
2914 
2915 	/* v4: gid_list */
2916 	ceph_encode_32(p, req->r_cred->group_info->ngroups);
2917 	for (i = 0; i < req->r_cred->group_info->ngroups; i++)
2918 		ceph_encode_64(p, from_kgid(&init_user_ns,
2919 					    req->r_cred->group_info->gid[i]));
2920 
2921 	/* v5: altname */
2922 	ceph_encode_32(p, req->r_altname_len);
2923 	ceph_encode_copy(p, req->r_altname, req->r_altname_len);
2924 
2925 	/* v6: fscrypt_auth and fscrypt_file */
2926 	if (req->r_fscrypt_auth) {
2927 		u32 authlen = ceph_fscrypt_auth_len(req->r_fscrypt_auth);
2928 
2929 		ceph_encode_32(p, authlen);
2930 		ceph_encode_copy(p, req->r_fscrypt_auth, authlen);
2931 	} else {
2932 		ceph_encode_32(p, 0);
2933 	}
2934 	if (test_bit(CEPH_MDS_R_FSCRYPT_FILE, &req->r_req_flags)) {
2935 		ceph_encode_32(p, sizeof(__le64));
2936 		ceph_encode_64(p, req->r_fscrypt_file);
2937 	} else {
2938 		ceph_encode_32(p, 0);
2939 	}
2940 }
2941 
2942 static inline u16 mds_supported_head_version(struct ceph_mds_session *session)
2943 {
2944 	if (!test_bit(CEPHFS_FEATURE_32BITS_RETRY_FWD, &session->s_features))
2945 		return 1;
2946 
2947 	if (!test_bit(CEPHFS_FEATURE_HAS_OWNER_UIDGID, &session->s_features))
2948 		return 2;
2949 
2950 	return CEPH_MDS_REQUEST_HEAD_VERSION;
2951 }
2952 
2953 static struct ceph_mds_request_head_legacy *
2954 find_legacy_request_head(void *p, u64 features)
2955 {
2956 	bool legacy = !(features & CEPH_FEATURE_FS_BTIME);
2957 	struct ceph_mds_request_head_old *ohead;
2958 
2959 	if (legacy)
2960 		return (struct ceph_mds_request_head_legacy *)p;
2961 	ohead = (struct ceph_mds_request_head_old *)p;
2962 	return (struct ceph_mds_request_head_legacy *)&ohead->oldest_client_tid;
2963 }
2964 
2965 /*
2966  * called under mdsc->mutex
2967  */
2968 static struct ceph_msg *create_request_message(struct ceph_mds_session *session,
2969 					       struct ceph_mds_request *req,
2970 					       bool drop_cap_releases)
2971 {
2972 	int mds = session->s_mds;
2973 	struct ceph_mds_client *mdsc = session->s_mdsc;
2974 	struct ceph_client *cl = mdsc->fsc->client;
2975 	struct ceph_msg *msg;
2976 	struct ceph_mds_request_head_legacy *lhead;
2977 	const char *path1 = NULL;
2978 	const char *path2 = NULL;
2979 	u64 ino1 = 0, ino2 = 0;
2980 	int pathlen1 = 0, pathlen2 = 0;
2981 	bool freepath1 = false, freepath2 = false;
2982 	struct dentry *old_dentry = NULL;
2983 	int len;
2984 	u16 releases;
2985 	void *p, *end;
2986 	int ret;
2987 	bool legacy = !(session->s_con.peer_features & CEPH_FEATURE_FS_BTIME);
2988 	u16 request_head_version = mds_supported_head_version(session);
2989 	kuid_t caller_fsuid = req->r_cred->fsuid;
2990 	kgid_t caller_fsgid = req->r_cred->fsgid;
2991 
2992 	ret = set_request_path_attr(mdsc, req->r_inode, req->r_dentry,
2993 			      req->r_parent, req->r_path1, req->r_ino1.ino,
2994 			      &path1, &pathlen1, &ino1, &freepath1,
2995 			      test_bit(CEPH_MDS_R_PARENT_LOCKED,
2996 					&req->r_req_flags));
2997 	if (ret < 0) {
2998 		msg = ERR_PTR(ret);
2999 		goto out;
3000 	}
3001 
3002 	/* If r_old_dentry is set, then assume that its parent is locked */
3003 	if (req->r_old_dentry &&
3004 	    !(req->r_old_dentry->d_flags & DCACHE_DISCONNECTED))
3005 		old_dentry = req->r_old_dentry;
3006 	ret = set_request_path_attr(mdsc, NULL, old_dentry,
3007 			      req->r_old_dentry_dir,
3008 			      req->r_path2, req->r_ino2.ino,
3009 			      &path2, &pathlen2, &ino2, &freepath2, true);
3010 	if (ret < 0) {
3011 		msg = ERR_PTR(ret);
3012 		goto out_free1;
3013 	}
3014 
3015 	req->r_altname = get_fscrypt_altname(req, &req->r_altname_len);
3016 	if (IS_ERR(req->r_altname)) {
3017 		msg = ERR_CAST(req->r_altname);
3018 		req->r_altname = NULL;
3019 		goto out_free2;
3020 	}
3021 
3022 	/*
3023 	 * For old cephs without supporting the 32bit retry/fwd feature
3024 	 * it will copy the raw memories directly when decoding the
3025 	 * requests. While new cephs will decode the head depending the
3026 	 * version member, so we need to make sure it will be compatible
3027 	 * with them both.
3028 	 */
3029 	if (legacy)
3030 		len = sizeof(struct ceph_mds_request_head_legacy);
3031 	else if (request_head_version == 1)
3032 		len = sizeof(struct ceph_mds_request_head_old);
3033 	else if (request_head_version == 2)
3034 		len = offsetofend(struct ceph_mds_request_head, ext_num_fwd);
3035 	else
3036 		len = sizeof(struct ceph_mds_request_head);
3037 
3038 	/* filepaths */
3039 	len += 2 * (1 + sizeof(u32) + sizeof(u64));
3040 	len += pathlen1 + pathlen2;
3041 
3042 	/* cap releases */
3043 	len += sizeof(struct ceph_mds_request_release) *
3044 		(!!req->r_inode_drop + !!req->r_dentry_drop +
3045 		 !!req->r_old_inode_drop + !!req->r_old_dentry_drop);
3046 
3047 	if (req->r_dentry_drop)
3048 		len += pathlen1;
3049 	if (req->r_old_dentry_drop)
3050 		len += pathlen2;
3051 
3052 	/* MClientRequest tail */
3053 
3054 	/* req->r_stamp */
3055 	len += sizeof(struct ceph_timespec);
3056 
3057 	/* gid list */
3058 	len += sizeof(u32) + (sizeof(u64) * req->r_cred->group_info->ngroups);
3059 
3060 	/* alternate name */
3061 	len += sizeof(u32) + req->r_altname_len;
3062 
3063 	/* fscrypt_auth */
3064 	len += sizeof(u32); // fscrypt_auth
3065 	if (req->r_fscrypt_auth)
3066 		len += ceph_fscrypt_auth_len(req->r_fscrypt_auth);
3067 
3068 	/* fscrypt_file */
3069 	len += sizeof(u32);
3070 	if (test_bit(CEPH_MDS_R_FSCRYPT_FILE, &req->r_req_flags))
3071 		len += sizeof(__le64);
3072 
3073 	msg = ceph_msg_new2(CEPH_MSG_CLIENT_REQUEST, len, 1, GFP_NOFS, false);
3074 	if (!msg) {
3075 		msg = ERR_PTR(-ENOMEM);
3076 		goto out_free2;
3077 	}
3078 
3079 	msg->hdr.tid = cpu_to_le64(req->r_tid);
3080 
3081 	lhead = find_legacy_request_head(msg->front.iov_base,
3082 					 session->s_con.peer_features);
3083 
3084 	if ((req->r_mnt_idmap != &nop_mnt_idmap) &&
3085 	    !test_bit(CEPHFS_FEATURE_HAS_OWNER_UIDGID, &session->s_features)) {
3086 		WARN_ON_ONCE(!IS_CEPH_MDS_OP_NEWINODE(req->r_op));
3087 
3088 		if (enable_unsafe_idmap) {
3089 			pr_warn_once_client(cl,
3090 				"idmapped mount is used and CEPHFS_FEATURE_HAS_OWNER_UIDGID"
3091 				" is not supported by MDS. UID/GID-based restrictions may"
3092 				" not work properly.\n");
3093 
3094 			caller_fsuid = from_vfsuid(req->r_mnt_idmap, &init_user_ns,
3095 						   VFSUIDT_INIT(req->r_cred->fsuid));
3096 			caller_fsgid = from_vfsgid(req->r_mnt_idmap, &init_user_ns,
3097 						   VFSGIDT_INIT(req->r_cred->fsgid));
3098 		} else {
3099 			pr_err_ratelimited_client(cl,
3100 				"idmapped mount is used and CEPHFS_FEATURE_HAS_OWNER_UIDGID"
3101 				" is not supported by MDS. Fail request with -EIO.\n");
3102 
3103 			ret = -EIO;
3104 			goto out_err;
3105 		}
3106 	}
3107 
3108 	/*
3109 	 * The ceph_mds_request_head_legacy didn't contain a version field, and
3110 	 * one was added when we moved the message version from 3->4.
3111 	 */
3112 	if (legacy) {
3113 		msg->hdr.version = cpu_to_le16(3);
3114 		p = msg->front.iov_base + sizeof(*lhead);
3115 	} else if (request_head_version == 1) {
3116 		struct ceph_mds_request_head_old *ohead = msg->front.iov_base;
3117 
3118 		msg->hdr.version = cpu_to_le16(4);
3119 		ohead->version = cpu_to_le16(1);
3120 		p = msg->front.iov_base + sizeof(*ohead);
3121 	} else if (request_head_version == 2) {
3122 		struct ceph_mds_request_head *nhead = msg->front.iov_base;
3123 
3124 		msg->hdr.version = cpu_to_le16(6);
3125 		nhead->version = cpu_to_le16(2);
3126 
3127 		p = msg->front.iov_base + offsetofend(struct ceph_mds_request_head, ext_num_fwd);
3128 	} else {
3129 		struct ceph_mds_request_head *nhead = msg->front.iov_base;
3130 		kuid_t owner_fsuid;
3131 		kgid_t owner_fsgid;
3132 
3133 		msg->hdr.version = cpu_to_le16(6);
3134 		nhead->version = cpu_to_le16(CEPH_MDS_REQUEST_HEAD_VERSION);
3135 		nhead->struct_len = cpu_to_le32(sizeof(struct ceph_mds_request_head));
3136 
3137 		if (IS_CEPH_MDS_OP_NEWINODE(req->r_op)) {
3138 			owner_fsuid = from_vfsuid(req->r_mnt_idmap, &init_user_ns,
3139 						VFSUIDT_INIT(req->r_cred->fsuid));
3140 			owner_fsgid = from_vfsgid(req->r_mnt_idmap, &init_user_ns,
3141 						VFSGIDT_INIT(req->r_cred->fsgid));
3142 			nhead->owner_uid = cpu_to_le32(from_kuid(&init_user_ns, owner_fsuid));
3143 			nhead->owner_gid = cpu_to_le32(from_kgid(&init_user_ns, owner_fsgid));
3144 		} else {
3145 			nhead->owner_uid = cpu_to_le32(-1);
3146 			nhead->owner_gid = cpu_to_le32(-1);
3147 		}
3148 
3149 		p = msg->front.iov_base + sizeof(*nhead);
3150 	}
3151 
3152 	end = msg->front.iov_base + msg->front.iov_len;
3153 
3154 	lhead->mdsmap_epoch = cpu_to_le32(mdsc->mdsmap->m_epoch);
3155 	lhead->op = cpu_to_le32(req->r_op);
3156 	lhead->caller_uid = cpu_to_le32(from_kuid(&init_user_ns,
3157 						  caller_fsuid));
3158 	lhead->caller_gid = cpu_to_le32(from_kgid(&init_user_ns,
3159 						  caller_fsgid));
3160 	lhead->ino = cpu_to_le64(req->r_deleg_ino);
3161 	lhead->args = req->r_args;
3162 
3163 	ceph_encode_filepath(&p, end, ino1, path1);
3164 	ceph_encode_filepath(&p, end, ino2, path2);
3165 
3166 	/* make note of release offset, in case we need to replay */
3167 	req->r_request_release_offset = p - msg->front.iov_base;
3168 
3169 	/* cap releases */
3170 	releases = 0;
3171 	if (req->r_inode_drop)
3172 		releases += ceph_encode_inode_release(&p,
3173 		      req->r_inode ? req->r_inode : d_inode(req->r_dentry),
3174 		      mds, req->r_inode_drop, req->r_inode_unless,
3175 		      req->r_op == CEPH_MDS_OP_READDIR);
3176 	if (req->r_dentry_drop) {
3177 		ret = ceph_encode_dentry_release(&p, req->r_dentry,
3178 				req->r_parent, mds, req->r_dentry_drop,
3179 				req->r_dentry_unless);
3180 		if (ret < 0)
3181 			goto out_err;
3182 		releases += ret;
3183 	}
3184 	if (req->r_old_dentry_drop) {
3185 		ret = ceph_encode_dentry_release(&p, req->r_old_dentry,
3186 				req->r_old_dentry_dir, mds,
3187 				req->r_old_dentry_drop,
3188 				req->r_old_dentry_unless);
3189 		if (ret < 0)
3190 			goto out_err;
3191 		releases += ret;
3192 	}
3193 	if (req->r_old_inode_drop)
3194 		releases += ceph_encode_inode_release(&p,
3195 		      d_inode(req->r_old_dentry),
3196 		      mds, req->r_old_inode_drop, req->r_old_inode_unless, 0);
3197 
3198 	if (drop_cap_releases) {
3199 		releases = 0;
3200 		p = msg->front.iov_base + req->r_request_release_offset;
3201 	}
3202 
3203 	lhead->num_releases = cpu_to_le16(releases);
3204 
3205 	encode_mclientrequest_tail(&p, req);
3206 
3207 	if (WARN_ON_ONCE(p > end)) {
3208 		ceph_msg_put(msg);
3209 		msg = ERR_PTR(-ERANGE);
3210 		goto out_free2;
3211 	}
3212 
3213 	msg->front.iov_len = p - msg->front.iov_base;
3214 	msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
3215 
3216 	if (req->r_pagelist) {
3217 		struct ceph_pagelist *pagelist = req->r_pagelist;
3218 		ceph_msg_data_add_pagelist(msg, pagelist);
3219 		msg->hdr.data_len = cpu_to_le32(pagelist->length);
3220 	} else {
3221 		msg->hdr.data_len = 0;
3222 	}
3223 
3224 	msg->hdr.data_off = cpu_to_le16(0);
3225 
3226 out_free2:
3227 	if (freepath2)
3228 		ceph_mdsc_free_path((char *)path2, pathlen2);
3229 out_free1:
3230 	if (freepath1)
3231 		ceph_mdsc_free_path((char *)path1, pathlen1);
3232 out:
3233 	return msg;
3234 out_err:
3235 	ceph_msg_put(msg);
3236 	msg = ERR_PTR(ret);
3237 	goto out_free2;
3238 }
3239 
3240 /*
3241  * called under mdsc->mutex if error, under no mutex if
3242  * success.
3243  */
3244 static void complete_request(struct ceph_mds_client *mdsc,
3245 			     struct ceph_mds_request *req)
3246 {
3247 	req->r_end_latency = ktime_get();
3248 
3249 	if (req->r_callback)
3250 		req->r_callback(mdsc, req);
3251 	complete_all(&req->r_completion);
3252 }
3253 
3254 /*
3255  * called under mdsc->mutex
3256  */
3257 static int __prepare_send_request(struct ceph_mds_session *session,
3258 				  struct ceph_mds_request *req,
3259 				  bool drop_cap_releases)
3260 {
3261 	int mds = session->s_mds;
3262 	struct ceph_mds_client *mdsc = session->s_mdsc;
3263 	struct ceph_client *cl = mdsc->fsc->client;
3264 	struct ceph_mds_request_head_legacy *lhead;
3265 	struct ceph_mds_request_head *nhead;
3266 	struct ceph_msg *msg;
3267 	int flags = 0, old_max_retry;
3268 	bool old_version = !test_bit(CEPHFS_FEATURE_32BITS_RETRY_FWD,
3269 				     &session->s_features);
3270 
3271 	/*
3272 	 * Avoid inifinite retrying after overflow. The client will
3273 	 * increase the retry count and if the MDS is old version,
3274 	 * so we limit to retry at most 256 times.
3275 	 */
3276 	if (req->r_attempts) {
3277 	       old_max_retry = sizeof_field(struct ceph_mds_request_head_old,
3278 					    num_retry);
3279 	       old_max_retry = 1 << (old_max_retry * BITS_PER_BYTE);
3280 	       if ((old_version && req->r_attempts >= old_max_retry) ||
3281 		   ((uint32_t)req->r_attempts >= U32_MAX)) {
3282 			pr_warn_ratelimited_client(cl, "request tid %llu seq overflow\n",
3283 						   req->r_tid);
3284 			return -EMULTIHOP;
3285 	       }
3286 	}
3287 
3288 	req->r_attempts++;
3289 	if (req->r_inode) {
3290 		struct ceph_cap *cap =
3291 			ceph_get_cap_for_mds(ceph_inode(req->r_inode), mds);
3292 
3293 		if (cap)
3294 			req->r_sent_on_mseq = cap->mseq;
3295 		else
3296 			req->r_sent_on_mseq = -1;
3297 	}
3298 	doutc(cl, "%p tid %lld %s (attempt %d)\n", req, req->r_tid,
3299 	      ceph_mds_op_name(req->r_op), req->r_attempts);
3300 
3301 	if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
3302 		void *p;
3303 
3304 		/*
3305 		 * Replay.  Do not regenerate message (and rebuild
3306 		 * paths, etc.); just use the original message.
3307 		 * Rebuilding paths will break for renames because
3308 		 * d_move mangles the src name.
3309 		 */
3310 		msg = req->r_request;
3311 		lhead = find_legacy_request_head(msg->front.iov_base,
3312 						 session->s_con.peer_features);
3313 
3314 		flags = le32_to_cpu(lhead->flags);
3315 		flags |= CEPH_MDS_FLAG_REPLAY;
3316 		lhead->flags = cpu_to_le32(flags);
3317 
3318 		if (req->r_target_inode)
3319 			lhead->ino = cpu_to_le64(ceph_ino(req->r_target_inode));
3320 
3321 		lhead->num_retry = req->r_attempts - 1;
3322 		if (!old_version) {
3323 			nhead = (struct ceph_mds_request_head*)msg->front.iov_base;
3324 			nhead->ext_num_retry = cpu_to_le32(req->r_attempts - 1);
3325 		}
3326 
3327 		/* remove cap/dentry releases from message */
3328 		lhead->num_releases = 0;
3329 
3330 		p = msg->front.iov_base + req->r_request_release_offset;
3331 		encode_mclientrequest_tail(&p, req);
3332 
3333 		msg->front.iov_len = p - msg->front.iov_base;
3334 		msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
3335 		return 0;
3336 	}
3337 
3338 	if (req->r_request) {
3339 		ceph_msg_put(req->r_request);
3340 		req->r_request = NULL;
3341 	}
3342 	msg = create_request_message(session, req, drop_cap_releases);
3343 	if (IS_ERR(msg)) {
3344 		req->r_err = PTR_ERR(msg);
3345 		return PTR_ERR(msg);
3346 	}
3347 	req->r_request = msg;
3348 
3349 	lhead = find_legacy_request_head(msg->front.iov_base,
3350 					 session->s_con.peer_features);
3351 	lhead->oldest_client_tid = cpu_to_le64(__get_oldest_tid(mdsc));
3352 	if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
3353 		flags |= CEPH_MDS_FLAG_REPLAY;
3354 	if (test_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags))
3355 		flags |= CEPH_MDS_FLAG_ASYNC;
3356 	if (req->r_parent)
3357 		flags |= CEPH_MDS_FLAG_WANT_DENTRY;
3358 	lhead->flags = cpu_to_le32(flags);
3359 	lhead->num_fwd = req->r_num_fwd;
3360 	lhead->num_retry = req->r_attempts - 1;
3361 	if (!old_version) {
3362 		nhead = (struct ceph_mds_request_head*)msg->front.iov_base;
3363 		nhead->ext_num_fwd = cpu_to_le32(req->r_num_fwd);
3364 		nhead->ext_num_retry = cpu_to_le32(req->r_attempts - 1);
3365 	}
3366 
3367 	doutc(cl, " r_parent = %p\n", req->r_parent);
3368 	return 0;
3369 }
3370 
3371 /*
3372  * called under mdsc->mutex
3373  */
3374 static int __send_request(struct ceph_mds_session *session,
3375 			  struct ceph_mds_request *req,
3376 			  bool drop_cap_releases)
3377 {
3378 	int err;
3379 
3380 	err = __prepare_send_request(session, req, drop_cap_releases);
3381 	if (!err) {
3382 		ceph_msg_get(req->r_request);
3383 		ceph_con_send(&session->s_con, req->r_request);
3384 	}
3385 
3386 	return err;
3387 }
3388 
3389 /*
3390  * send request, or put it on the appropriate wait list.
3391  */
3392 static void __do_request(struct ceph_mds_client *mdsc,
3393 			struct ceph_mds_request *req)
3394 {
3395 	struct ceph_client *cl = mdsc->fsc->client;
3396 	struct ceph_mds_session *session = NULL;
3397 	int mds = -1;
3398 	int err = 0;
3399 	bool random;
3400 
3401 	if (req->r_err || test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
3402 		if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags))
3403 			__unregister_request(mdsc, req);
3404 		return;
3405 	}
3406 
3407 	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_FENCE_IO) {
3408 		doutc(cl, "metadata corrupted\n");
3409 		err = -EIO;
3410 		goto finish;
3411 	}
3412 	if (req->r_timeout &&
3413 	    time_after_eq(jiffies, req->r_started + req->r_timeout)) {
3414 		doutc(cl, "timed out\n");
3415 		err = -ETIMEDOUT;
3416 		goto finish;
3417 	}
3418 	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
3419 		doutc(cl, "forced umount\n");
3420 		err = -EIO;
3421 		goto finish;
3422 	}
3423 	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_MOUNTING) {
3424 		if (mdsc->mdsmap_err) {
3425 			err = mdsc->mdsmap_err;
3426 			doutc(cl, "mdsmap err %d\n", err);
3427 			goto finish;
3428 		}
3429 		if (mdsc->mdsmap->m_epoch == 0) {
3430 			doutc(cl, "no mdsmap, waiting for map\n");
3431 			list_add(&req->r_wait, &mdsc->waiting_for_map);
3432 			return;
3433 		}
3434 		if (!(mdsc->fsc->mount_options->flags &
3435 		      CEPH_MOUNT_OPT_MOUNTWAIT) &&
3436 		    !ceph_mdsmap_is_cluster_available(mdsc->mdsmap)) {
3437 			err = -EHOSTUNREACH;
3438 			goto finish;
3439 		}
3440 	}
3441 
3442 	put_request_session(req);
3443 
3444 	mds = __choose_mds(mdsc, req, &random);
3445 	if (mds < 0 ||
3446 	    ceph_mdsmap_get_state(mdsc->mdsmap, mds) < CEPH_MDS_STATE_ACTIVE) {
3447 		if (test_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags)) {
3448 			err = -EJUKEBOX;
3449 			goto finish;
3450 		}
3451 		doutc(cl, "no mds or not active, waiting for map\n");
3452 		list_add(&req->r_wait, &mdsc->waiting_for_map);
3453 		return;
3454 	}
3455 
3456 	/* get, open session */
3457 	session = __ceph_lookup_mds_session(mdsc, mds);
3458 	if (!session) {
3459 		session = register_session(mdsc, mds);
3460 		if (IS_ERR(session)) {
3461 			err = PTR_ERR(session);
3462 			goto finish;
3463 		}
3464 	}
3465 	req->r_session = ceph_get_mds_session(session);
3466 
3467 	doutc(cl, "mds%d session %p state %s\n", mds, session,
3468 	      ceph_session_state_name(session->s_state));
3469 
3470 	/*
3471 	 * The old ceph will crash the MDSs when see unknown OPs
3472 	 */
3473 	if (req->r_feature_needed > 0 &&
3474 	    !test_bit(req->r_feature_needed, &session->s_features)) {
3475 		err = -EOPNOTSUPP;
3476 		goto out_session;
3477 	}
3478 
3479 	if (session->s_state != CEPH_MDS_SESSION_OPEN &&
3480 	    session->s_state != CEPH_MDS_SESSION_HUNG) {
3481 		/*
3482 		 * We cannot queue async requests since the caps and delegated
3483 		 * inodes are bound to the session. Just return -EJUKEBOX and
3484 		 * let the caller retry a sync request in that case.
3485 		 */
3486 		if (test_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags)) {
3487 			err = -EJUKEBOX;
3488 			goto out_session;
3489 		}
3490 
3491 		/*
3492 		 * If the session has been REJECTED, then return a hard error,
3493 		 * unless it's a CLEANRECOVER mount, in which case we'll queue
3494 		 * it to the mdsc queue.
3495 		 */
3496 		if (session->s_state == CEPH_MDS_SESSION_REJECTED) {
3497 			if (ceph_test_mount_opt(mdsc->fsc, CLEANRECOVER))
3498 				list_add(&req->r_wait, &mdsc->waiting_for_map);
3499 			else
3500 				err = -EACCES;
3501 			goto out_session;
3502 		}
3503 
3504 		if (session->s_state == CEPH_MDS_SESSION_NEW ||
3505 		    session->s_state == CEPH_MDS_SESSION_CLOSING) {
3506 			err = __open_session(mdsc, session);
3507 			if (err)
3508 				goto out_session;
3509 			/* retry the same mds later */
3510 			if (random)
3511 				req->r_resend_mds = mds;
3512 		}
3513 		list_add(&req->r_wait, &session->s_waiting);
3514 		goto out_session;
3515 	}
3516 
3517 	/* send request */
3518 	req->r_resend_mds = -1;   /* forget any previous mds hint */
3519 
3520 	if (req->r_request_started == 0)   /* note request start time */
3521 		req->r_request_started = jiffies;
3522 
3523 	/*
3524 	 * For async create we will choose the auth MDS of frag in parent
3525 	 * directory to send the request and ususally this works fine, but
3526 	 * if the migrated the dirtory to another MDS before it could handle
3527 	 * it the request will be forwarded.
3528 	 *
3529 	 * And then the auth cap will be changed.
3530 	 */
3531 	if (test_bit(CEPH_MDS_R_ASYNC, &req->r_req_flags) && req->r_num_fwd) {
3532 		struct ceph_dentry_info *di = ceph_dentry(req->r_dentry);
3533 		struct ceph_inode_info *ci;
3534 		struct ceph_cap *cap;
3535 
3536 		/*
3537 		 * The request maybe handled very fast and the new inode
3538 		 * hasn't been linked to the dentry yet. We need to wait
3539 		 * for the ceph_finish_async_create(), which shouldn't be
3540 		 * stuck too long or fail in thoery, to finish when forwarding
3541 		 * the request.
3542 		 */
3543 		if (!d_inode(req->r_dentry)) {
3544 			err = wait_on_bit(&di->flags, CEPH_DENTRY_ASYNC_CREATE_BIT,
3545 					  TASK_KILLABLE);
3546 			if (err) {
3547 				mutex_lock(&req->r_fill_mutex);
3548 				set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
3549 				mutex_unlock(&req->r_fill_mutex);
3550 				goto out_session;
3551 			}
3552 		}
3553 
3554 		ci = ceph_inode(d_inode(req->r_dentry));
3555 
3556 		spin_lock(&ci->i_ceph_lock);
3557 		cap = ci->i_auth_cap;
3558 		if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE && mds != cap->mds) {
3559 			doutc(cl, "session changed for auth cap %d -> %d\n",
3560 			      cap->session->s_mds, session->s_mds);
3561 
3562 			/* Remove the auth cap from old session */
3563 			spin_lock(&cap->session->s_cap_lock);
3564 			cap->session->s_nr_caps--;
3565 			list_del_init(&cap->session_caps);
3566 			spin_unlock(&cap->session->s_cap_lock);
3567 
3568 			/* Add the auth cap to the new session */
3569 			cap->mds = mds;
3570 			cap->session = session;
3571 			spin_lock(&session->s_cap_lock);
3572 			session->s_nr_caps++;
3573 			list_add_tail(&cap->session_caps, &session->s_caps);
3574 			spin_unlock(&session->s_cap_lock);
3575 
3576 			change_auth_cap_ses(ci, session);
3577 		}
3578 		spin_unlock(&ci->i_ceph_lock);
3579 	}
3580 
3581 	err = __send_request(session, req, false);
3582 
3583 out_session:
3584 	ceph_put_mds_session(session);
3585 finish:
3586 	if (err) {
3587 		doutc(cl, "early error %d\n", err);
3588 		req->r_err = err;
3589 		complete_request(mdsc, req);
3590 		__unregister_request(mdsc, req);
3591 	}
3592 	return;
3593 }
3594 
3595 /*
3596  * called under mdsc->mutex
3597  */
3598 static void __wake_requests(struct ceph_mds_client *mdsc,
3599 			    struct list_head *head)
3600 {
3601 	struct ceph_client *cl = mdsc->fsc->client;
3602 	struct ceph_mds_request *req;
3603 	LIST_HEAD(tmp_list);
3604 
3605 	list_splice_init(head, &tmp_list);
3606 
3607 	while (!list_empty(&tmp_list)) {
3608 		req = list_entry(tmp_list.next,
3609 				 struct ceph_mds_request, r_wait);
3610 		list_del_init(&req->r_wait);
3611 		doutc(cl, " wake request %p tid %llu\n", req,
3612 		      req->r_tid);
3613 		__do_request(mdsc, req);
3614 	}
3615 }
3616 
3617 /*
3618  * Wake up threads with requests pending for @mds, so that they can
3619  * resubmit their requests to a possibly different mds.
3620  */
3621 static void kick_requests(struct ceph_mds_client *mdsc, int mds)
3622 {
3623 	struct ceph_client *cl = mdsc->fsc->client;
3624 	struct ceph_mds_request *req;
3625 	struct rb_node *p = rb_first(&mdsc->request_tree);
3626 
3627 	doutc(cl, "kick_requests mds%d\n", mds);
3628 	while (p) {
3629 		req = rb_entry(p, struct ceph_mds_request, r_node);
3630 		p = rb_next(p);
3631 		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
3632 			continue;
3633 		if (req->r_attempts > 0)
3634 			continue; /* only new requests */
3635 		if (req->r_session &&
3636 		    req->r_session->s_mds == mds) {
3637 			doutc(cl, " kicking tid %llu\n", req->r_tid);
3638 			list_del_init(&req->r_wait);
3639 			__do_request(mdsc, req);
3640 		}
3641 	}
3642 }
3643 
3644 int ceph_mdsc_submit_request(struct ceph_mds_client *mdsc, struct inode *dir,
3645 			      struct ceph_mds_request *req)
3646 {
3647 	struct ceph_client *cl = mdsc->fsc->client;
3648 	int err = 0;
3649 
3650 	/* take CAP_PIN refs for r_inode, r_parent, r_old_dentry */
3651 	if (req->r_inode)
3652 		ceph_get_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
3653 	if (req->r_parent) {
3654 		struct ceph_inode_info *ci = ceph_inode(req->r_parent);
3655 		int fmode = (req->r_op & CEPH_MDS_OP_WRITE) ?
3656 			    CEPH_FILE_MODE_WR : CEPH_FILE_MODE_RD;
3657 		spin_lock(&ci->i_ceph_lock);
3658 		ceph_take_cap_refs(ci, CEPH_CAP_PIN, false);
3659 		__ceph_touch_fmode(ci, mdsc, fmode);
3660 		spin_unlock(&ci->i_ceph_lock);
3661 	}
3662 	if (req->r_old_dentry_dir)
3663 		ceph_get_cap_refs(ceph_inode(req->r_old_dentry_dir),
3664 				  CEPH_CAP_PIN);
3665 
3666 	if (req->r_inode) {
3667 		err = ceph_wait_on_async_create(req->r_inode);
3668 		if (err) {
3669 			doutc(cl, "wait for async create returned: %d\n", err);
3670 			return err;
3671 		}
3672 	}
3673 
3674 	if (!err && req->r_old_inode) {
3675 		err = ceph_wait_on_async_create(req->r_old_inode);
3676 		if (err) {
3677 			doutc(cl, "wait for async create returned: %d\n", err);
3678 			return err;
3679 		}
3680 	}
3681 
3682 	doutc(cl, "submit_request on %p for inode %p\n", req, dir);
3683 	mutex_lock(&mdsc->mutex);
3684 	__register_request(mdsc, req, dir);
3685 	__do_request(mdsc, req);
3686 	err = req->r_err;
3687 	mutex_unlock(&mdsc->mutex);
3688 	return err;
3689 }
3690 
3691 int ceph_mdsc_wait_request(struct ceph_mds_client *mdsc,
3692 			   struct ceph_mds_request *req,
3693 			   ceph_mds_request_wait_callback_t wait_func)
3694 {
3695 	struct ceph_client *cl = mdsc->fsc->client;
3696 	int err;
3697 
3698 	/* wait */
3699 	doutc(cl, "do_request waiting\n");
3700 	if (wait_func) {
3701 		err = wait_func(mdsc, req);
3702 	} else {
3703 		long timeleft = wait_for_completion_killable_timeout(
3704 					&req->r_completion,
3705 					ceph_timeout_jiffies(req->r_timeout));
3706 		if (timeleft > 0)
3707 			err = 0;
3708 		else if (!timeleft)
3709 			err = -ETIMEDOUT;  /* timed out */
3710 		else
3711 			err = timeleft;  /* killed */
3712 	}
3713 	doutc(cl, "do_request waited, got %d\n", err);
3714 	mutex_lock(&mdsc->mutex);
3715 
3716 	/* only abort if we didn't race with a real reply */
3717 	if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
3718 		err = le32_to_cpu(req->r_reply_info.head->result);
3719 	} else if (err < 0) {
3720 		doutc(cl, "aborted request %lld with %d\n", req->r_tid, err);
3721 
3722 		/*
3723 		 * ensure we aren't running concurrently with
3724 		 * ceph_fill_trace or ceph_readdir_prepopulate, which
3725 		 * rely on locks (dir mutex) held by our caller.
3726 		 */
3727 		mutex_lock(&req->r_fill_mutex);
3728 		req->r_err = err;
3729 		set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
3730 		mutex_unlock(&req->r_fill_mutex);
3731 
3732 		if (req->r_parent &&
3733 		    (req->r_op & CEPH_MDS_OP_WRITE))
3734 			ceph_invalidate_dir_request(req);
3735 	} else {
3736 		err = req->r_err;
3737 	}
3738 
3739 	mutex_unlock(&mdsc->mutex);
3740 	return err;
3741 }
3742 
3743 /*
3744  * Synchrously perform an mds request.  Take care of all of the
3745  * session setup, forwarding, retry details.
3746  */
3747 int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
3748 			 struct inode *dir,
3749 			 struct ceph_mds_request *req)
3750 {
3751 	struct ceph_client *cl = mdsc->fsc->client;
3752 	int err;
3753 
3754 	doutc(cl, "do_request on %p\n", req);
3755 
3756 	/* issue */
3757 	err = ceph_mdsc_submit_request(mdsc, dir, req);
3758 	if (!err)
3759 		err = ceph_mdsc_wait_request(mdsc, req, NULL);
3760 	doutc(cl, "do_request %p done, result %d\n", req, err);
3761 	return err;
3762 }
3763 
3764 /*
3765  * Invalidate dir's completeness, dentry lease state on an aborted MDS
3766  * namespace request.
3767  */
3768 void ceph_invalidate_dir_request(struct ceph_mds_request *req)
3769 {
3770 	struct inode *dir = req->r_parent;
3771 	struct inode *old_dir = req->r_old_dentry_dir;
3772 	struct ceph_client *cl = req->r_mdsc->fsc->client;
3773 
3774 	doutc(cl, "invalidate_dir_request %p %p (complete, lease(s))\n",
3775 	      dir, old_dir);
3776 
3777 	ceph_dir_clear_complete(dir);
3778 	if (old_dir)
3779 		ceph_dir_clear_complete(old_dir);
3780 	if (req->r_dentry)
3781 		ceph_invalidate_dentry_lease(req->r_dentry);
3782 	if (req->r_old_dentry)
3783 		ceph_invalidate_dentry_lease(req->r_old_dentry);
3784 }
3785 
3786 /*
3787  * Handle mds reply.
3788  *
3789  * We take the session mutex and parse and process the reply immediately.
3790  * This preserves the logical ordering of replies, capabilities, etc., sent
3791  * by the MDS as they are applied to our local cache.
3792  */
3793 static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
3794 {
3795 	struct ceph_mds_client *mdsc = session->s_mdsc;
3796 	struct ceph_client *cl = mdsc->fsc->client;
3797 	struct ceph_mds_request *req;
3798 	struct ceph_mds_reply_head *head = msg->front.iov_base;
3799 	struct ceph_mds_reply_info_parsed *rinfo;  /* parsed reply info */
3800 	struct ceph_snap_realm *realm;
3801 	u64 tid;
3802 	int err, result;
3803 	int mds = session->s_mds;
3804 	bool close_sessions = false;
3805 
3806 	if (msg->front.iov_len < sizeof(*head)) {
3807 		pr_err_client(cl, "got corrupt (short) reply\n");
3808 		ceph_msg_dump(msg);
3809 		return;
3810 	}
3811 
3812 	/* get request, session */
3813 	tid = le64_to_cpu(msg->hdr.tid);
3814 	mutex_lock(&mdsc->mutex);
3815 	req = lookup_get_request(mdsc, tid);
3816 	if (!req) {
3817 		doutc(cl, "on unknown tid %llu\n", tid);
3818 		mutex_unlock(&mdsc->mutex);
3819 		return;
3820 	}
3821 	doutc(cl, "handle_reply %p\n", req);
3822 
3823 	/* correct session? */
3824 	if (req->r_session != session) {
3825 		pr_err_client(cl, "got %llu on session mds%d not mds%d\n",
3826 			      tid, session->s_mds,
3827 			      req->r_session ? req->r_session->s_mds : -1);
3828 		mutex_unlock(&mdsc->mutex);
3829 		goto out;
3830 	}
3831 
3832 	/* dup? */
3833 	if ((test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags) && !head->safe) ||
3834 	    (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags) && head->safe)) {
3835 		pr_warn_client(cl, "got a dup %s reply on %llu from mds%d\n",
3836 			       head->safe ? "safe" : "unsafe", tid, mds);
3837 		mutex_unlock(&mdsc->mutex);
3838 		goto out;
3839 	}
3840 	if (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags)) {
3841 		pr_warn_client(cl, "got unsafe after safe on %llu from mds%d\n",
3842 			       tid, mds);
3843 		mutex_unlock(&mdsc->mutex);
3844 		goto out;
3845 	}
3846 
3847 	result = le32_to_cpu(head->result);
3848 
3849 	if (head->safe) {
3850 		set_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags);
3851 		__unregister_request(mdsc, req);
3852 
3853 		/* last request during umount? */
3854 		if (mdsc->stopping && !__get_oldest_req(mdsc))
3855 			complete_all(&mdsc->safe_umount_waiters);
3856 
3857 		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
3858 			/*
3859 			 * We already handled the unsafe response, now do the
3860 			 * cleanup.  No need to examine the response; the MDS
3861 			 * doesn't include any result info in the safe
3862 			 * response.  And even if it did, there is nothing
3863 			 * useful we could do with a revised return value.
3864 			 */
3865 			doutc(cl, "got safe reply %llu, mds%d\n", tid, mds);
3866 
3867 			mutex_unlock(&mdsc->mutex);
3868 			goto out;
3869 		}
3870 	} else {
3871 		set_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags);
3872 		list_add_tail(&req->r_unsafe_item, &req->r_session->s_unsafe);
3873 	}
3874 
3875 	doutc(cl, "tid %lld result %d\n", tid, result);
3876 	if (test_bit(CEPHFS_FEATURE_REPLY_ENCODING, &session->s_features))
3877 		err = parse_reply_info(session, msg, req, (u64)-1);
3878 	else
3879 		err = parse_reply_info(session, msg, req,
3880 				       session->s_con.peer_features);
3881 	mutex_unlock(&mdsc->mutex);
3882 
3883 	/* Must find target inode outside of mutexes to avoid deadlocks */
3884 	rinfo = &req->r_reply_info;
3885 	if ((err >= 0) && rinfo->head->is_target) {
3886 		struct inode *in = xchg(&req->r_new_inode, NULL);
3887 		struct ceph_vino tvino = {
3888 			.ino  = le64_to_cpu(rinfo->targeti.in->ino),
3889 			.snap = le64_to_cpu(rinfo->targeti.in->snapid)
3890 		};
3891 
3892 		/*
3893 		 * If we ended up opening an existing inode, discard
3894 		 * r_new_inode
3895 		 */
3896 		if (req->r_op == CEPH_MDS_OP_CREATE &&
3897 		    !req->r_reply_info.has_create_ino) {
3898 			/* This should never happen on an async create */
3899 			WARN_ON_ONCE(req->r_deleg_ino);
3900 			iput(in);
3901 			in = NULL;
3902 		}
3903 
3904 		in = ceph_get_inode(mdsc->fsc->sb, tvino, in);
3905 		if (IS_ERR(in)) {
3906 			err = PTR_ERR(in);
3907 			mutex_lock(&session->s_mutex);
3908 			goto out_err;
3909 		}
3910 		req->r_target_inode = in;
3911 	}
3912 
3913 	mutex_lock(&session->s_mutex);
3914 	if (err < 0) {
3915 		pr_err_client(cl, "got corrupt reply mds%d(tid:%lld)\n",
3916 			      mds, tid);
3917 		ceph_msg_dump(msg);
3918 		goto out_err;
3919 	}
3920 
3921 	/* snap trace */
3922 	realm = NULL;
3923 	if (rinfo->snapblob_len) {
3924 		down_write(&mdsc->snap_rwsem);
3925 		err = ceph_update_snap_trace(mdsc, rinfo->snapblob,
3926 				rinfo->snapblob + rinfo->snapblob_len,
3927 				le32_to_cpu(head->op) == CEPH_MDS_OP_RMSNAP,
3928 				&realm);
3929 		if (err) {
3930 			up_write(&mdsc->snap_rwsem);
3931 			close_sessions = true;
3932 			if (err == -EIO)
3933 				ceph_msg_dump(msg);
3934 			goto out_err;
3935 		}
3936 		downgrade_write(&mdsc->snap_rwsem);
3937 	} else {
3938 		down_read(&mdsc->snap_rwsem);
3939 	}
3940 
3941 	/* insert trace into our cache */
3942 	mutex_lock(&req->r_fill_mutex);
3943 	current->journal_info = req;
3944 	err = ceph_fill_trace(mdsc->fsc->sb, req);
3945 	if (err == 0) {
3946 		if (result == 0 && (req->r_op == CEPH_MDS_OP_READDIR ||
3947 				    req->r_op == CEPH_MDS_OP_LSSNAP))
3948 			err = ceph_readdir_prepopulate(req, req->r_session);
3949 	}
3950 	current->journal_info = NULL;
3951 	mutex_unlock(&req->r_fill_mutex);
3952 
3953 	up_read(&mdsc->snap_rwsem);
3954 	if (realm)
3955 		ceph_put_snap_realm(mdsc, realm);
3956 
3957 	if (err == 0) {
3958 		if (req->r_target_inode &&
3959 		    test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
3960 			struct ceph_inode_info *ci =
3961 				ceph_inode(req->r_target_inode);
3962 			spin_lock(&ci->i_unsafe_lock);
3963 			list_add_tail(&req->r_unsafe_target_item,
3964 				      &ci->i_unsafe_iops);
3965 			spin_unlock(&ci->i_unsafe_lock);
3966 		}
3967 
3968 		ceph_unreserve_caps(mdsc, &req->r_caps_reservation);
3969 	}
3970 out_err:
3971 	mutex_lock(&mdsc->mutex);
3972 	if (!test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
3973 		if (err) {
3974 			req->r_err = err;
3975 		} else {
3976 			req->r_reply =  ceph_msg_get(msg);
3977 			set_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags);
3978 		}
3979 	} else {
3980 		doutc(cl, "reply arrived after request %lld was aborted\n", tid);
3981 	}
3982 	mutex_unlock(&mdsc->mutex);
3983 
3984 	mutex_unlock(&session->s_mutex);
3985 
3986 	/* kick calling process */
3987 	complete_request(mdsc, req);
3988 
3989 	ceph_update_metadata_metrics(&mdsc->metric, req->r_start_latency,
3990 				     req->r_end_latency, err);
3991 out:
3992 	ceph_mdsc_put_request(req);
3993 
3994 	/* Defer closing the sessions after s_mutex lock being released */
3995 	if (close_sessions)
3996 		ceph_mdsc_close_sessions(mdsc);
3997 	return;
3998 }
3999 
4000 
4001 
4002 /*
4003  * handle mds notification that our request has been forwarded.
4004  */
4005 static void handle_forward(struct ceph_mds_client *mdsc,
4006 			   struct ceph_mds_session *session,
4007 			   struct ceph_msg *msg)
4008 {
4009 	struct ceph_client *cl = mdsc->fsc->client;
4010 	struct ceph_mds_request *req;
4011 	u64 tid = le64_to_cpu(msg->hdr.tid);
4012 	u32 next_mds;
4013 	u32 fwd_seq;
4014 	int err = -EINVAL;
4015 	void *p = msg->front.iov_base;
4016 	void *end = p + msg->front.iov_len;
4017 	bool aborted = false;
4018 
4019 	ceph_decode_need(&p, end, 2*sizeof(u32), bad);
4020 	next_mds = ceph_decode_32(&p);
4021 	fwd_seq = ceph_decode_32(&p);
4022 
4023 	mutex_lock(&mdsc->mutex);
4024 	req = lookup_get_request(mdsc, tid);
4025 	if (!req) {
4026 		mutex_unlock(&mdsc->mutex);
4027 		doutc(cl, "forward tid %llu to mds%d - req dne\n", tid, next_mds);
4028 		return;  /* dup reply? */
4029 	}
4030 
4031 	if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
4032 		doutc(cl, "forward tid %llu aborted, unregistering\n", tid);
4033 		__unregister_request(mdsc, req);
4034 	} else if (fwd_seq <= req->r_num_fwd || (uint32_t)fwd_seq >= U32_MAX) {
4035 		/*
4036 		 * Avoid inifinite retrying after overflow.
4037 		 *
4038 		 * The MDS will increase the fwd count and in client side
4039 		 * if the num_fwd is less than the one saved in request
4040 		 * that means the MDS is an old version and overflowed of
4041 		 * 8 bits.
4042 		 */
4043 		mutex_lock(&req->r_fill_mutex);
4044 		req->r_err = -EMULTIHOP;
4045 		set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
4046 		mutex_unlock(&req->r_fill_mutex);
4047 		aborted = true;
4048 		pr_warn_ratelimited_client(cl, "forward tid %llu seq overflow\n",
4049 					   tid);
4050 	} else {
4051 		/* resend. forward race not possible; mds would drop */
4052 		doutc(cl, "forward tid %llu to mds%d (we resend)\n", tid, next_mds);
4053 		BUG_ON(req->r_err);
4054 		BUG_ON(test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags));
4055 		req->r_attempts = 0;
4056 		req->r_num_fwd = fwd_seq;
4057 		req->r_resend_mds = next_mds;
4058 		put_request_session(req);
4059 		__do_request(mdsc, req);
4060 	}
4061 	mutex_unlock(&mdsc->mutex);
4062 
4063 	/* kick calling process */
4064 	if (aborted)
4065 		complete_request(mdsc, req);
4066 	ceph_mdsc_put_request(req);
4067 	return;
4068 
4069 bad:
4070 	pr_err_client(cl, "decode error err=%d\n", err);
4071 	ceph_msg_dump(msg);
4072 }
4073 
4074 static int __decode_session_metadata(void **p, void *end,
4075 				     bool *blocklisted)
4076 {
4077 	/* map<string,string> */
4078 	u32 n;
4079 	bool err_str;
4080 	ceph_decode_32_safe(p, end, n, bad);
4081 	while (n-- > 0) {
4082 		u32 len;
4083 		ceph_decode_32_safe(p, end, len, bad);
4084 		ceph_decode_need(p, end, len, bad);
4085 		err_str = !strncmp(*p, "error_string", len);
4086 		*p += len;
4087 		ceph_decode_32_safe(p, end, len, bad);
4088 		ceph_decode_need(p, end, len, bad);
4089 		/*
4090 		 * Match "blocklisted (blacklisted)" from newer MDSes,
4091 		 * or "blacklisted" from older MDSes.
4092 		 */
4093 		if (err_str && strnstr(*p, "blacklisted", len))
4094 			*blocklisted = true;
4095 		*p += len;
4096 	}
4097 	return 0;
4098 bad:
4099 	return -1;
4100 }
4101 
4102 /*
4103  * handle a mds session control message
4104  */
4105 static void handle_session(struct ceph_mds_session *session,
4106 			   struct ceph_msg *msg)
4107 {
4108 	struct ceph_mds_client *mdsc = session->s_mdsc;
4109 	struct ceph_client *cl = mdsc->fsc->client;
4110 	int mds = session->s_mds;
4111 	int msg_version = le16_to_cpu(msg->hdr.version);
4112 	void *p = msg->front.iov_base;
4113 	void *end = p + msg->front.iov_len;
4114 	struct ceph_mds_session_head *h;
4115 	struct ceph_mds_cap_auth *cap_auths = NULL;
4116 	u32 op, cap_auths_num = 0;
4117 	u64 seq, features = 0;
4118 	int wake = 0;
4119 	bool blocklisted = false;
4120 	u32 i;
4121 
4122 
4123 	/* decode */
4124 	ceph_decode_need(&p, end, sizeof(*h), bad);
4125 	h = p;
4126 	p += sizeof(*h);
4127 
4128 	op = le32_to_cpu(h->op);
4129 	seq = le64_to_cpu(h->seq);
4130 
4131 	if (msg_version >= 3) {
4132 		u32 len;
4133 		/* version >= 2 and < 5, decode metadata, skip otherwise
4134 		 * as it's handled via flags.
4135 		 */
4136 		if (msg_version >= 5)
4137 			ceph_decode_skip_map(&p, end, string, string, bad);
4138 		else if (__decode_session_metadata(&p, end, &blocklisted) < 0)
4139 			goto bad;
4140 
4141 		/* version >= 3, feature bits */
4142 		ceph_decode_32_safe(&p, end, len, bad);
4143 		if (len) {
4144 			ceph_decode_64_safe(&p, end, features, bad);
4145 			p += len - sizeof(features);
4146 		}
4147 	}
4148 
4149 	if (msg_version >= 5) {
4150 		u32 flags, len;
4151 
4152 		/* version >= 4 */
4153 		ceph_decode_skip_16(&p, end, bad); /* struct_v, struct_cv */
4154 		ceph_decode_32_safe(&p, end, len, bad); /* len */
4155 		ceph_decode_skip_n(&p, end, len, bad); /* metric_spec */
4156 
4157 		/* version >= 5, flags   */
4158 		ceph_decode_32_safe(&p, end, flags, bad);
4159 		if (flags & CEPH_SESSION_BLOCKLISTED) {
4160 			pr_warn_client(cl, "mds%d session blocklisted\n",
4161 				       session->s_mds);
4162 			blocklisted = true;
4163 		}
4164 	}
4165 
4166 	if (msg_version >= 6) {
4167 		ceph_decode_32_safe(&p, end, cap_auths_num, bad);
4168 		doutc(cl, "cap_auths_num %d\n", cap_auths_num);
4169 
4170 		if (cap_auths_num && op != CEPH_SESSION_OPEN) {
4171 			WARN_ON_ONCE(op != CEPH_SESSION_OPEN);
4172 			goto skip_cap_auths;
4173 		}
4174 
4175 		cap_auths = kcalloc(cap_auths_num,
4176 				    sizeof(struct ceph_mds_cap_auth),
4177 				    GFP_KERNEL);
4178 		if (!cap_auths) {
4179 			pr_err_client(cl, "No memory for cap_auths\n");
4180 			return;
4181 		}
4182 
4183 		for (i = 0; i < cap_auths_num; i++) {
4184 			u32 _len, j;
4185 
4186 			/* struct_v, struct_compat, and struct_len in MDSCapAuth */
4187 			ceph_decode_skip_n(&p, end, 2 + sizeof(u32), bad);
4188 
4189 			/* struct_v, struct_compat, and struct_len in MDSCapMatch */
4190 			ceph_decode_skip_n(&p, end, 2 + sizeof(u32), bad);
4191 			ceph_decode_64_safe(&p, end, cap_auths[i].match.uid, bad);
4192 			ceph_decode_32_safe(&p, end, _len, bad);
4193 			if (_len) {
4194 				cap_auths[i].match.gids = kcalloc(_len, sizeof(u32),
4195 								  GFP_KERNEL);
4196 				if (!cap_auths[i].match.gids) {
4197 					pr_err_client(cl, "No memory for gids\n");
4198 					goto fail;
4199 				}
4200 
4201 				cap_auths[i].match.num_gids = _len;
4202 				for (j = 0; j < _len; j++)
4203 					ceph_decode_32_safe(&p, end,
4204 							    cap_auths[i].match.gids[j],
4205 							    bad);
4206 			}
4207 
4208 			ceph_decode_32_safe(&p, end, _len, bad);
4209 			if (_len) {
4210 				cap_auths[i].match.path = kcalloc(_len + 1, sizeof(char),
4211 								  GFP_KERNEL);
4212 				if (!cap_auths[i].match.path) {
4213 					pr_err_client(cl, "No memory for path\n");
4214 					goto fail;
4215 				}
4216 				ceph_decode_copy(&p, cap_auths[i].match.path, _len);
4217 
4218 				/* Remove the tailing '/' */
4219 				while (_len && cap_auths[i].match.path[_len - 1] == '/') {
4220 					cap_auths[i].match.path[_len - 1] = '\0';
4221 					_len -= 1;
4222 				}
4223 			}
4224 
4225 			ceph_decode_32_safe(&p, end, _len, bad);
4226 			if (_len) {
4227 				cap_auths[i].match.fs_name = kcalloc(_len + 1, sizeof(char),
4228 								     GFP_KERNEL);
4229 				if (!cap_auths[i].match.fs_name) {
4230 					pr_err_client(cl, "No memory for fs_name\n");
4231 					goto fail;
4232 				}
4233 				ceph_decode_copy(&p, cap_auths[i].match.fs_name, _len);
4234 			}
4235 
4236 			ceph_decode_8_safe(&p, end, cap_auths[i].match.root_squash, bad);
4237 			ceph_decode_8_safe(&p, end, cap_auths[i].readable, bad);
4238 			ceph_decode_8_safe(&p, end, cap_auths[i].writeable, bad);
4239 			doutc(cl, "uid %lld, num_gids %u, path %s, fs_name %s, root_squash %d, readable %d, writeable %d\n",
4240 			      cap_auths[i].match.uid, cap_auths[i].match.num_gids,
4241 			      cap_auths[i].match.path, cap_auths[i].match.fs_name,
4242 			      cap_auths[i].match.root_squash,
4243 			      cap_auths[i].readable, cap_auths[i].writeable);
4244 		}
4245 	}
4246 
4247 skip_cap_auths:
4248 	mutex_lock(&mdsc->mutex);
4249 	if (op == CEPH_SESSION_OPEN) {
4250 		if (mdsc->s_cap_auths) {
4251 			for (i = 0; i < mdsc->s_cap_auths_num; i++) {
4252 				kfree(mdsc->s_cap_auths[i].match.gids);
4253 				kfree(mdsc->s_cap_auths[i].match.path);
4254 				kfree(mdsc->s_cap_auths[i].match.fs_name);
4255 			}
4256 			kfree(mdsc->s_cap_auths);
4257 		}
4258 		mdsc->s_cap_auths_num = cap_auths_num;
4259 		mdsc->s_cap_auths = cap_auths;
4260 	}
4261 	if (op == CEPH_SESSION_CLOSE) {
4262 		ceph_get_mds_session(session);
4263 		__unregister_session(mdsc, session);
4264 	}
4265 	/* FIXME: this ttl calculation is generous */
4266 	session->s_ttl = jiffies + HZ*mdsc->mdsmap->m_session_autoclose;
4267 	mutex_unlock(&mdsc->mutex);
4268 
4269 	mutex_lock(&session->s_mutex);
4270 
4271 	doutc(cl, "mds%d %s %p state %s seq %llu\n", mds,
4272 	      ceph_session_op_name(op), session,
4273 	      ceph_session_state_name(session->s_state), seq);
4274 
4275 	if (session->s_state == CEPH_MDS_SESSION_HUNG) {
4276 		session->s_state = CEPH_MDS_SESSION_OPEN;
4277 		pr_info_client(cl, "mds%d came back\n", session->s_mds);
4278 	}
4279 
4280 	switch (op) {
4281 	case CEPH_SESSION_OPEN:
4282 		if (session->s_state == CEPH_MDS_SESSION_RECONNECTING)
4283 			pr_info_client(cl, "mds%d reconnect success\n",
4284 				       session->s_mds);
4285 
4286 		session->s_features = features;
4287 		if (session->s_state == CEPH_MDS_SESSION_OPEN) {
4288 			pr_notice_client(cl, "mds%d is already opened\n",
4289 					 session->s_mds);
4290 		} else {
4291 			session->s_state = CEPH_MDS_SESSION_OPEN;
4292 			renewed_caps(mdsc, session, 0);
4293 			if (test_bit(CEPHFS_FEATURE_METRIC_COLLECT,
4294 				     &session->s_features))
4295 				metric_schedule_delayed(&mdsc->metric);
4296 		}
4297 
4298 		/*
4299 		 * The connection maybe broken and the session in client
4300 		 * side has been reinitialized, need to update the seq
4301 		 * anyway.
4302 		 */
4303 		if (!session->s_seq && seq)
4304 			session->s_seq = seq;
4305 
4306 		wake = 1;
4307 		if (mdsc->stopping)
4308 			__close_session(mdsc, session);
4309 		break;
4310 
4311 	case CEPH_SESSION_RENEWCAPS:
4312 		if (session->s_renew_seq == seq)
4313 			renewed_caps(mdsc, session, 1);
4314 		break;
4315 
4316 	case CEPH_SESSION_CLOSE:
4317 		if (session->s_state == CEPH_MDS_SESSION_RECONNECTING)
4318 			pr_info_client(cl, "mds%d reconnect denied\n",
4319 				       session->s_mds);
4320 		session->s_state = CEPH_MDS_SESSION_CLOSED;
4321 		cleanup_session_requests(mdsc, session);
4322 		remove_session_caps(session);
4323 		wake = 2; /* for good measure */
4324 		wake_up_all(&mdsc->session_close_wq);
4325 		break;
4326 
4327 	case CEPH_SESSION_STALE:
4328 		pr_info_client(cl, "mds%d caps went stale, renewing\n",
4329 			       session->s_mds);
4330 		atomic_inc(&session->s_cap_gen);
4331 		session->s_cap_ttl = jiffies - 1;
4332 		send_renew_caps(mdsc, session);
4333 		break;
4334 
4335 	case CEPH_SESSION_RECALL_STATE:
4336 		ceph_trim_caps(mdsc, session, le32_to_cpu(h->max_caps));
4337 		break;
4338 
4339 	case CEPH_SESSION_FLUSHMSG:
4340 		/* flush cap releases */
4341 		spin_lock(&session->s_cap_lock);
4342 		if (session->s_num_cap_releases)
4343 			ceph_flush_session_cap_releases(mdsc, session);
4344 		spin_unlock(&session->s_cap_lock);
4345 
4346 		send_flushmsg_ack(mdsc, session, seq);
4347 		break;
4348 
4349 	case CEPH_SESSION_FORCE_RO:
4350 		doutc(cl, "force_session_readonly %p\n", session);
4351 		spin_lock(&session->s_cap_lock);
4352 		session->s_readonly = true;
4353 		spin_unlock(&session->s_cap_lock);
4354 		wake_up_session_caps(session, FORCE_RO);
4355 		break;
4356 
4357 	case CEPH_SESSION_REJECT:
4358 		WARN_ON(session->s_state != CEPH_MDS_SESSION_OPENING);
4359 		pr_info_client(cl, "mds%d rejected session\n",
4360 			       session->s_mds);
4361 		session->s_state = CEPH_MDS_SESSION_REJECTED;
4362 		cleanup_session_requests(mdsc, session);
4363 		remove_session_caps(session);
4364 		if (blocklisted)
4365 			mdsc->fsc->blocklisted = true;
4366 		wake = 2; /* for good measure */
4367 		break;
4368 
4369 	default:
4370 		pr_err_client(cl, "bad op %d mds%d\n", op, mds);
4371 		WARN_ON(1);
4372 	}
4373 
4374 	mutex_unlock(&session->s_mutex);
4375 	if (wake) {
4376 		mutex_lock(&mdsc->mutex);
4377 		__wake_requests(mdsc, &session->s_waiting);
4378 		if (wake == 2)
4379 			kick_requests(mdsc, mds);
4380 		mutex_unlock(&mdsc->mutex);
4381 	}
4382 	if (op == CEPH_SESSION_CLOSE)
4383 		ceph_put_mds_session(session);
4384 	return;
4385 
4386 bad:
4387 	pr_err_client(cl, "corrupt message mds%d len %d\n", mds,
4388 		      (int)msg->front.iov_len);
4389 	ceph_msg_dump(msg);
4390 fail:
4391 	for (i = 0; i < cap_auths_num; i++) {
4392 		kfree(cap_auths[i].match.gids);
4393 		kfree(cap_auths[i].match.path);
4394 		kfree(cap_auths[i].match.fs_name);
4395 	}
4396 	kfree(cap_auths);
4397 	return;
4398 }
4399 
4400 void ceph_mdsc_release_dir_caps(struct ceph_mds_request *req)
4401 {
4402 	struct ceph_client *cl = req->r_mdsc->fsc->client;
4403 	int dcaps;
4404 
4405 	dcaps = xchg(&req->r_dir_caps, 0);
4406 	if (dcaps) {
4407 		doutc(cl, "releasing r_dir_caps=%s\n", ceph_cap_string(dcaps));
4408 		ceph_put_cap_refs(ceph_inode(req->r_parent), dcaps);
4409 	}
4410 }
4411 
4412 void ceph_mdsc_release_dir_caps_async(struct ceph_mds_request *req)
4413 {
4414 	struct ceph_client *cl = req->r_mdsc->fsc->client;
4415 	int dcaps;
4416 
4417 	dcaps = xchg(&req->r_dir_caps, 0);
4418 	if (dcaps) {
4419 		doutc(cl, "releasing r_dir_caps=%s\n", ceph_cap_string(dcaps));
4420 		ceph_put_cap_refs_async(ceph_inode(req->r_parent), dcaps);
4421 	}
4422 }
4423 
4424 /*
4425  * called under session->mutex.
4426  */
4427 static void replay_unsafe_requests(struct ceph_mds_client *mdsc,
4428 				   struct ceph_mds_session *session)
4429 {
4430 	struct ceph_mds_request *req, *nreq;
4431 	struct rb_node *p;
4432 
4433 	doutc(mdsc->fsc->client, "mds%d\n", session->s_mds);
4434 
4435 	mutex_lock(&mdsc->mutex);
4436 	list_for_each_entry_safe(req, nreq, &session->s_unsafe, r_unsafe_item)
4437 		__send_request(session, req, true);
4438 
4439 	/*
4440 	 * also re-send old requests when MDS enters reconnect stage. So that MDS
4441 	 * can process completed request in clientreplay stage.
4442 	 */
4443 	p = rb_first(&mdsc->request_tree);
4444 	while (p) {
4445 		req = rb_entry(p, struct ceph_mds_request, r_node);
4446 		p = rb_next(p);
4447 		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
4448 			continue;
4449 		if (req->r_attempts == 0)
4450 			continue; /* only old requests */
4451 		if (!req->r_session)
4452 			continue;
4453 		if (req->r_session->s_mds != session->s_mds)
4454 			continue;
4455 
4456 		ceph_mdsc_release_dir_caps_async(req);
4457 
4458 		__send_request(session, req, true);
4459 	}
4460 	mutex_unlock(&mdsc->mutex);
4461 }
4462 
4463 static int send_reconnect_partial(struct ceph_reconnect_state *recon_state)
4464 {
4465 	struct ceph_msg *reply;
4466 	struct ceph_pagelist *_pagelist;
4467 	struct page *page;
4468 	__le32 *addr;
4469 	int err = -ENOMEM;
4470 
4471 	if (!recon_state->allow_multi)
4472 		return -ENOSPC;
4473 
4474 	/* can't handle message that contains both caps and realm */
4475 	BUG_ON(!recon_state->nr_caps == !recon_state->nr_realms);
4476 
4477 	/* pre-allocate new pagelist */
4478 	_pagelist = ceph_pagelist_alloc(GFP_NOFS);
4479 	if (!_pagelist)
4480 		return -ENOMEM;
4481 
4482 	reply = ceph_msg_new2(CEPH_MSG_CLIENT_RECONNECT, 0, 1, GFP_NOFS, false);
4483 	if (!reply)
4484 		goto fail_msg;
4485 
4486 	/* placeholder for nr_caps */
4487 	err = ceph_pagelist_encode_32(_pagelist, 0);
4488 	if (err < 0)
4489 		goto fail;
4490 
4491 	if (recon_state->nr_caps) {
4492 		/* currently encoding caps */
4493 		err = ceph_pagelist_encode_32(recon_state->pagelist, 0);
4494 		if (err)
4495 			goto fail;
4496 	} else {
4497 		/* placeholder for nr_realms (currently encoding relams) */
4498 		err = ceph_pagelist_encode_32(_pagelist, 0);
4499 		if (err < 0)
4500 			goto fail;
4501 	}
4502 
4503 	err = ceph_pagelist_encode_8(recon_state->pagelist, 1);
4504 	if (err)
4505 		goto fail;
4506 
4507 	page = list_first_entry(&recon_state->pagelist->head, struct page, lru);
4508 	addr = kmap_atomic(page);
4509 	if (recon_state->nr_caps) {
4510 		/* currently encoding caps */
4511 		*addr = cpu_to_le32(recon_state->nr_caps);
4512 	} else {
4513 		/* currently encoding relams */
4514 		*(addr + 1) = cpu_to_le32(recon_state->nr_realms);
4515 	}
4516 	kunmap_atomic(addr);
4517 
4518 	reply->hdr.version = cpu_to_le16(5);
4519 	reply->hdr.compat_version = cpu_to_le16(4);
4520 
4521 	reply->hdr.data_len = cpu_to_le32(recon_state->pagelist->length);
4522 	ceph_msg_data_add_pagelist(reply, recon_state->pagelist);
4523 
4524 	ceph_con_send(&recon_state->session->s_con, reply);
4525 	ceph_pagelist_release(recon_state->pagelist);
4526 
4527 	recon_state->pagelist = _pagelist;
4528 	recon_state->nr_caps = 0;
4529 	recon_state->nr_realms = 0;
4530 	recon_state->msg_version = 5;
4531 	return 0;
4532 fail:
4533 	ceph_msg_put(reply);
4534 fail_msg:
4535 	ceph_pagelist_release(_pagelist);
4536 	return err;
4537 }
4538 
4539 static struct dentry* d_find_primary(struct inode *inode)
4540 {
4541 	struct dentry *alias, *dn = NULL;
4542 
4543 	if (hlist_empty(&inode->i_dentry))
4544 		return NULL;
4545 
4546 	spin_lock(&inode->i_lock);
4547 	if (hlist_empty(&inode->i_dentry))
4548 		goto out_unlock;
4549 
4550 	if (S_ISDIR(inode->i_mode)) {
4551 		alias = hlist_entry(inode->i_dentry.first, struct dentry, d_u.d_alias);
4552 		if (!IS_ROOT(alias))
4553 			dn = dget(alias);
4554 		goto out_unlock;
4555 	}
4556 
4557 	hlist_for_each_entry(alias, &inode->i_dentry, d_u.d_alias) {
4558 		spin_lock(&alias->d_lock);
4559 		if (!d_unhashed(alias) &&
4560 		    (ceph_dentry(alias)->flags & CEPH_DENTRY_PRIMARY_LINK)) {
4561 			dn = dget_dlock(alias);
4562 		}
4563 		spin_unlock(&alias->d_lock);
4564 		if (dn)
4565 			break;
4566 	}
4567 out_unlock:
4568 	spin_unlock(&inode->i_lock);
4569 	return dn;
4570 }
4571 
4572 /*
4573  * Encode information about a cap for a reconnect with the MDS.
4574  */
4575 static int reconnect_caps_cb(struct inode *inode, int mds, void *arg)
4576 {
4577 	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
4578 	struct ceph_client *cl = ceph_inode_to_client(inode);
4579 	union {
4580 		struct ceph_mds_cap_reconnect v2;
4581 		struct ceph_mds_cap_reconnect_v1 v1;
4582 	} rec;
4583 	struct ceph_inode_info *ci = ceph_inode(inode);
4584 	struct ceph_reconnect_state *recon_state = arg;
4585 	struct ceph_pagelist *pagelist = recon_state->pagelist;
4586 	struct dentry *dentry;
4587 	struct ceph_cap *cap;
4588 	char *path;
4589 	int pathlen = 0, err;
4590 	u64 pathbase;
4591 	u64 snap_follows;
4592 
4593 	dentry = d_find_primary(inode);
4594 	if (dentry) {
4595 		/* set pathbase to parent dir when msg_version >= 2 */
4596 		path = ceph_mdsc_build_path(mdsc, dentry, &pathlen, &pathbase,
4597 					    recon_state->msg_version >= 2);
4598 		dput(dentry);
4599 		if (IS_ERR(path)) {
4600 			err = PTR_ERR(path);
4601 			goto out_err;
4602 		}
4603 	} else {
4604 		path = NULL;
4605 		pathbase = 0;
4606 	}
4607 
4608 	spin_lock(&ci->i_ceph_lock);
4609 	cap = __get_cap_for_mds(ci, mds);
4610 	if (!cap) {
4611 		spin_unlock(&ci->i_ceph_lock);
4612 		err = 0;
4613 		goto out_err;
4614 	}
4615 	doutc(cl, " adding %p ino %llx.%llx cap %p %lld %s\n", inode,
4616 	      ceph_vinop(inode), cap, cap->cap_id,
4617 	      ceph_cap_string(cap->issued));
4618 
4619 	cap->seq = 0;        /* reset cap seq */
4620 	cap->issue_seq = 0;  /* and issue_seq */
4621 	cap->mseq = 0;       /* and migrate_seq */
4622 	cap->cap_gen = atomic_read(&cap->session->s_cap_gen);
4623 
4624 	/* These are lost when the session goes away */
4625 	if (S_ISDIR(inode->i_mode)) {
4626 		if (cap->issued & CEPH_CAP_DIR_CREATE) {
4627 			ceph_put_string(rcu_dereference_raw(ci->i_cached_layout.pool_ns));
4628 			memset(&ci->i_cached_layout, 0, sizeof(ci->i_cached_layout));
4629 		}
4630 		cap->issued &= ~CEPH_CAP_ANY_DIR_OPS;
4631 	}
4632 
4633 	if (recon_state->msg_version >= 2) {
4634 		rec.v2.cap_id = cpu_to_le64(cap->cap_id);
4635 		rec.v2.wanted = cpu_to_le32(__ceph_caps_wanted(ci));
4636 		rec.v2.issued = cpu_to_le32(cap->issued);
4637 		rec.v2.snaprealm = cpu_to_le64(ci->i_snap_realm->ino);
4638 		rec.v2.pathbase = cpu_to_le64(pathbase);
4639 		rec.v2.flock_len = (__force __le32)
4640 			((ci->i_ceph_flags & CEPH_I_ERROR_FILELOCK) ? 0 : 1);
4641 	} else {
4642 		struct timespec64 ts;
4643 
4644 		rec.v1.cap_id = cpu_to_le64(cap->cap_id);
4645 		rec.v1.wanted = cpu_to_le32(__ceph_caps_wanted(ci));
4646 		rec.v1.issued = cpu_to_le32(cap->issued);
4647 		rec.v1.size = cpu_to_le64(i_size_read(inode));
4648 		ts = inode_get_mtime(inode);
4649 		ceph_encode_timespec64(&rec.v1.mtime, &ts);
4650 		ts = inode_get_atime(inode);
4651 		ceph_encode_timespec64(&rec.v1.atime, &ts);
4652 		rec.v1.snaprealm = cpu_to_le64(ci->i_snap_realm->ino);
4653 		rec.v1.pathbase = cpu_to_le64(pathbase);
4654 	}
4655 
4656 	if (list_empty(&ci->i_cap_snaps)) {
4657 		snap_follows = ci->i_head_snapc ? ci->i_head_snapc->seq : 0;
4658 	} else {
4659 		struct ceph_cap_snap *capsnap =
4660 			list_first_entry(&ci->i_cap_snaps,
4661 					 struct ceph_cap_snap, ci_item);
4662 		snap_follows = capsnap->follows;
4663 	}
4664 	spin_unlock(&ci->i_ceph_lock);
4665 
4666 	if (recon_state->msg_version >= 2) {
4667 		int num_fcntl_locks, num_flock_locks;
4668 		struct ceph_filelock *flocks = NULL;
4669 		size_t struct_len, total_len = sizeof(u64);
4670 		u8 struct_v = 0;
4671 
4672 encode_again:
4673 		if (rec.v2.flock_len) {
4674 			ceph_count_locks(inode, &num_fcntl_locks, &num_flock_locks);
4675 		} else {
4676 			num_fcntl_locks = 0;
4677 			num_flock_locks = 0;
4678 		}
4679 		if (num_fcntl_locks + num_flock_locks > 0) {
4680 			flocks = kmalloc_array(num_fcntl_locks + num_flock_locks,
4681 					       sizeof(struct ceph_filelock),
4682 					       GFP_NOFS);
4683 			if (!flocks) {
4684 				err = -ENOMEM;
4685 				goto out_err;
4686 			}
4687 			err = ceph_encode_locks_to_buffer(inode, flocks,
4688 							  num_fcntl_locks,
4689 							  num_flock_locks);
4690 			if (err) {
4691 				kfree(flocks);
4692 				flocks = NULL;
4693 				if (err == -ENOSPC)
4694 					goto encode_again;
4695 				goto out_err;
4696 			}
4697 		} else {
4698 			kfree(flocks);
4699 			flocks = NULL;
4700 		}
4701 
4702 		if (recon_state->msg_version >= 3) {
4703 			/* version, compat_version and struct_len */
4704 			total_len += 2 * sizeof(u8) + sizeof(u32);
4705 			struct_v = 2;
4706 		}
4707 		/*
4708 		 * number of encoded locks is stable, so copy to pagelist
4709 		 */
4710 		struct_len = 2 * sizeof(u32) +
4711 			    (num_fcntl_locks + num_flock_locks) *
4712 			    sizeof(struct ceph_filelock);
4713 		rec.v2.flock_len = cpu_to_le32(struct_len);
4714 
4715 		struct_len += sizeof(u32) + pathlen + sizeof(rec.v2);
4716 
4717 		if (struct_v >= 2)
4718 			struct_len += sizeof(u64); /* snap_follows */
4719 
4720 		total_len += struct_len;
4721 
4722 		if (pagelist->length + total_len > RECONNECT_MAX_SIZE) {
4723 			err = send_reconnect_partial(recon_state);
4724 			if (err)
4725 				goto out_freeflocks;
4726 			pagelist = recon_state->pagelist;
4727 		}
4728 
4729 		err = ceph_pagelist_reserve(pagelist, total_len);
4730 		if (err)
4731 			goto out_freeflocks;
4732 
4733 		ceph_pagelist_encode_64(pagelist, ceph_ino(inode));
4734 		if (recon_state->msg_version >= 3) {
4735 			ceph_pagelist_encode_8(pagelist, struct_v);
4736 			ceph_pagelist_encode_8(pagelist, 1);
4737 			ceph_pagelist_encode_32(pagelist, struct_len);
4738 		}
4739 		ceph_pagelist_encode_string(pagelist, path, pathlen);
4740 		ceph_pagelist_append(pagelist, &rec, sizeof(rec.v2));
4741 		ceph_locks_to_pagelist(flocks, pagelist,
4742 				       num_fcntl_locks, num_flock_locks);
4743 		if (struct_v >= 2)
4744 			ceph_pagelist_encode_64(pagelist, snap_follows);
4745 out_freeflocks:
4746 		kfree(flocks);
4747 	} else {
4748 		err = ceph_pagelist_reserve(pagelist,
4749 					    sizeof(u64) + sizeof(u32) +
4750 					    pathlen + sizeof(rec.v1));
4751 		if (err)
4752 			goto out_err;
4753 
4754 		ceph_pagelist_encode_64(pagelist, ceph_ino(inode));
4755 		ceph_pagelist_encode_string(pagelist, path, pathlen);
4756 		ceph_pagelist_append(pagelist, &rec, sizeof(rec.v1));
4757 	}
4758 
4759 out_err:
4760 	ceph_mdsc_free_path(path, pathlen);
4761 	if (!err)
4762 		recon_state->nr_caps++;
4763 	return err;
4764 }
4765 
4766 static int encode_snap_realms(struct ceph_mds_client *mdsc,
4767 			      struct ceph_reconnect_state *recon_state)
4768 {
4769 	struct rb_node *p;
4770 	struct ceph_pagelist *pagelist = recon_state->pagelist;
4771 	struct ceph_client *cl = mdsc->fsc->client;
4772 	int err = 0;
4773 
4774 	if (recon_state->msg_version >= 4) {
4775 		err = ceph_pagelist_encode_32(pagelist, mdsc->num_snap_realms);
4776 		if (err < 0)
4777 			goto fail;
4778 	}
4779 
4780 	/*
4781 	 * snaprealms.  we provide mds with the ino, seq (version), and
4782 	 * parent for all of our realms.  If the mds has any newer info,
4783 	 * it will tell us.
4784 	 */
4785 	for (p = rb_first(&mdsc->snap_realms); p; p = rb_next(p)) {
4786 		struct ceph_snap_realm *realm =
4787 		       rb_entry(p, struct ceph_snap_realm, node);
4788 		struct ceph_mds_snaprealm_reconnect sr_rec;
4789 
4790 		if (recon_state->msg_version >= 4) {
4791 			size_t need = sizeof(u8) * 2 + sizeof(u32) +
4792 				      sizeof(sr_rec);
4793 
4794 			if (pagelist->length + need > RECONNECT_MAX_SIZE) {
4795 				err = send_reconnect_partial(recon_state);
4796 				if (err)
4797 					goto fail;
4798 				pagelist = recon_state->pagelist;
4799 			}
4800 
4801 			err = ceph_pagelist_reserve(pagelist, need);
4802 			if (err)
4803 				goto fail;
4804 
4805 			ceph_pagelist_encode_8(pagelist, 1);
4806 			ceph_pagelist_encode_8(pagelist, 1);
4807 			ceph_pagelist_encode_32(pagelist, sizeof(sr_rec));
4808 		}
4809 
4810 		doutc(cl, " adding snap realm %llx seq %lld parent %llx\n",
4811 		      realm->ino, realm->seq, realm->parent_ino);
4812 		sr_rec.ino = cpu_to_le64(realm->ino);
4813 		sr_rec.seq = cpu_to_le64(realm->seq);
4814 		sr_rec.parent = cpu_to_le64(realm->parent_ino);
4815 
4816 		err = ceph_pagelist_append(pagelist, &sr_rec, sizeof(sr_rec));
4817 		if (err)
4818 			goto fail;
4819 
4820 		recon_state->nr_realms++;
4821 	}
4822 fail:
4823 	return err;
4824 }
4825 
4826 
4827 /*
4828  * If an MDS fails and recovers, clients need to reconnect in order to
4829  * reestablish shared state.  This includes all caps issued through
4830  * this session _and_ the snap_realm hierarchy.  Because it's not
4831  * clear which snap realms the mds cares about, we send everything we
4832  * know about.. that ensures we'll then get any new info the
4833  * recovering MDS might have.
4834  *
4835  * This is a relatively heavyweight operation, but it's rare.
4836  */
4837 static void send_mds_reconnect(struct ceph_mds_client *mdsc,
4838 			       struct ceph_mds_session *session)
4839 {
4840 	struct ceph_client *cl = mdsc->fsc->client;
4841 	struct ceph_msg *reply;
4842 	int mds = session->s_mds;
4843 	int err = -ENOMEM;
4844 	struct ceph_reconnect_state recon_state = {
4845 		.session = session,
4846 	};
4847 	LIST_HEAD(dispose);
4848 
4849 	pr_info_client(cl, "mds%d reconnect start\n", mds);
4850 
4851 	recon_state.pagelist = ceph_pagelist_alloc(GFP_NOFS);
4852 	if (!recon_state.pagelist)
4853 		goto fail_nopagelist;
4854 
4855 	reply = ceph_msg_new2(CEPH_MSG_CLIENT_RECONNECT, 0, 1, GFP_NOFS, false);
4856 	if (!reply)
4857 		goto fail_nomsg;
4858 
4859 	xa_destroy(&session->s_delegated_inos);
4860 
4861 	mutex_lock(&session->s_mutex);
4862 	session->s_state = CEPH_MDS_SESSION_RECONNECTING;
4863 	session->s_seq = 0;
4864 
4865 	doutc(cl, "session %p state %s\n", session,
4866 	      ceph_session_state_name(session->s_state));
4867 
4868 	atomic_inc(&session->s_cap_gen);
4869 
4870 	spin_lock(&session->s_cap_lock);
4871 	/* don't know if session is readonly */
4872 	session->s_readonly = 0;
4873 	/*
4874 	 * notify __ceph_remove_cap() that we are composing cap reconnect.
4875 	 * If a cap get released before being added to the cap reconnect,
4876 	 * __ceph_remove_cap() should skip queuing cap release.
4877 	 */
4878 	session->s_cap_reconnect = 1;
4879 	/* drop old cap expires; we're about to reestablish that state */
4880 	detach_cap_releases(session, &dispose);
4881 	spin_unlock(&session->s_cap_lock);
4882 	dispose_cap_releases(mdsc, &dispose);
4883 
4884 	/* trim unused caps to reduce MDS's cache rejoin time */
4885 	if (mdsc->fsc->sb->s_root)
4886 		shrink_dcache_parent(mdsc->fsc->sb->s_root);
4887 
4888 	ceph_con_close(&session->s_con);
4889 	ceph_con_open(&session->s_con,
4890 		      CEPH_ENTITY_TYPE_MDS, mds,
4891 		      ceph_mdsmap_get_addr(mdsc->mdsmap, mds));
4892 
4893 	/* replay unsafe requests */
4894 	replay_unsafe_requests(mdsc, session);
4895 
4896 	ceph_early_kick_flushing_caps(mdsc, session);
4897 
4898 	down_read(&mdsc->snap_rwsem);
4899 
4900 	/* placeholder for nr_caps */
4901 	err = ceph_pagelist_encode_32(recon_state.pagelist, 0);
4902 	if (err)
4903 		goto fail;
4904 
4905 	if (test_bit(CEPHFS_FEATURE_MULTI_RECONNECT, &session->s_features)) {
4906 		recon_state.msg_version = 3;
4907 		recon_state.allow_multi = true;
4908 	} else if (session->s_con.peer_features & CEPH_FEATURE_MDSENC) {
4909 		recon_state.msg_version = 3;
4910 	} else {
4911 		recon_state.msg_version = 2;
4912 	}
4913 	/* traverse this session's caps */
4914 	err = ceph_iterate_session_caps(session, reconnect_caps_cb, &recon_state);
4915 
4916 	spin_lock(&session->s_cap_lock);
4917 	session->s_cap_reconnect = 0;
4918 	spin_unlock(&session->s_cap_lock);
4919 
4920 	if (err < 0)
4921 		goto fail;
4922 
4923 	/* check if all realms can be encoded into current message */
4924 	if (mdsc->num_snap_realms) {
4925 		size_t total_len =
4926 			recon_state.pagelist->length +
4927 			mdsc->num_snap_realms *
4928 			sizeof(struct ceph_mds_snaprealm_reconnect);
4929 		if (recon_state.msg_version >= 4) {
4930 			/* number of realms */
4931 			total_len += sizeof(u32);
4932 			/* version, compat_version and struct_len */
4933 			total_len += mdsc->num_snap_realms *
4934 				     (2 * sizeof(u8) + sizeof(u32));
4935 		}
4936 		if (total_len > RECONNECT_MAX_SIZE) {
4937 			if (!recon_state.allow_multi) {
4938 				err = -ENOSPC;
4939 				goto fail;
4940 			}
4941 			if (recon_state.nr_caps) {
4942 				err = send_reconnect_partial(&recon_state);
4943 				if (err)
4944 					goto fail;
4945 			}
4946 			recon_state.msg_version = 5;
4947 		}
4948 	}
4949 
4950 	err = encode_snap_realms(mdsc, &recon_state);
4951 	if (err < 0)
4952 		goto fail;
4953 
4954 	if (recon_state.msg_version >= 5) {
4955 		err = ceph_pagelist_encode_8(recon_state.pagelist, 0);
4956 		if (err < 0)
4957 			goto fail;
4958 	}
4959 
4960 	if (recon_state.nr_caps || recon_state.nr_realms) {
4961 		struct page *page =
4962 			list_first_entry(&recon_state.pagelist->head,
4963 					struct page, lru);
4964 		__le32 *addr = kmap_atomic(page);
4965 		if (recon_state.nr_caps) {
4966 			WARN_ON(recon_state.nr_realms != mdsc->num_snap_realms);
4967 			*addr = cpu_to_le32(recon_state.nr_caps);
4968 		} else if (recon_state.msg_version >= 4) {
4969 			*(addr + 1) = cpu_to_le32(recon_state.nr_realms);
4970 		}
4971 		kunmap_atomic(addr);
4972 	}
4973 
4974 	reply->hdr.version = cpu_to_le16(recon_state.msg_version);
4975 	if (recon_state.msg_version >= 4)
4976 		reply->hdr.compat_version = cpu_to_le16(4);
4977 
4978 	reply->hdr.data_len = cpu_to_le32(recon_state.pagelist->length);
4979 	ceph_msg_data_add_pagelist(reply, recon_state.pagelist);
4980 
4981 	ceph_con_send(&session->s_con, reply);
4982 
4983 	mutex_unlock(&session->s_mutex);
4984 
4985 	mutex_lock(&mdsc->mutex);
4986 	__wake_requests(mdsc, &session->s_waiting);
4987 	mutex_unlock(&mdsc->mutex);
4988 
4989 	up_read(&mdsc->snap_rwsem);
4990 	ceph_pagelist_release(recon_state.pagelist);
4991 	return;
4992 
4993 fail:
4994 	ceph_msg_put(reply);
4995 	up_read(&mdsc->snap_rwsem);
4996 	mutex_unlock(&session->s_mutex);
4997 fail_nomsg:
4998 	ceph_pagelist_release(recon_state.pagelist);
4999 fail_nopagelist:
5000 	pr_err_client(cl, "error %d preparing reconnect for mds%d\n",
5001 		      err, mds);
5002 	return;
5003 }
5004 
5005 
5006 /*
5007  * compare old and new mdsmaps, kicking requests
5008  * and closing out old connections as necessary
5009  *
5010  * called under mdsc->mutex.
5011  */
5012 static void check_new_map(struct ceph_mds_client *mdsc,
5013 			  struct ceph_mdsmap *newmap,
5014 			  struct ceph_mdsmap *oldmap)
5015 {
5016 	int i, j, err;
5017 	int oldstate, newstate;
5018 	struct ceph_mds_session *s;
5019 	unsigned long targets[DIV_ROUND_UP(CEPH_MAX_MDS, sizeof(unsigned long))] = {0};
5020 	struct ceph_client *cl = mdsc->fsc->client;
5021 
5022 	doutc(cl, "new %u old %u\n", newmap->m_epoch, oldmap->m_epoch);
5023 
5024 	if (newmap->m_info) {
5025 		for (i = 0; i < newmap->possible_max_rank; i++) {
5026 			for (j = 0; j < newmap->m_info[i].num_export_targets; j++)
5027 				set_bit(newmap->m_info[i].export_targets[j], targets);
5028 		}
5029 	}
5030 
5031 	for (i = 0; i < oldmap->possible_max_rank && i < mdsc->max_sessions; i++) {
5032 		if (!mdsc->sessions[i])
5033 			continue;
5034 		s = mdsc->sessions[i];
5035 		oldstate = ceph_mdsmap_get_state(oldmap, i);
5036 		newstate = ceph_mdsmap_get_state(newmap, i);
5037 
5038 		doutc(cl, "mds%d state %s%s -> %s%s (session %s)\n",
5039 		      i, ceph_mds_state_name(oldstate),
5040 		      ceph_mdsmap_is_laggy(oldmap, i) ? " (laggy)" : "",
5041 		      ceph_mds_state_name(newstate),
5042 		      ceph_mdsmap_is_laggy(newmap, i) ? " (laggy)" : "",
5043 		      ceph_session_state_name(s->s_state));
5044 
5045 		if (i >= newmap->possible_max_rank) {
5046 			/* force close session for stopped mds */
5047 			ceph_get_mds_session(s);
5048 			__unregister_session(mdsc, s);
5049 			__wake_requests(mdsc, &s->s_waiting);
5050 			mutex_unlock(&mdsc->mutex);
5051 
5052 			mutex_lock(&s->s_mutex);
5053 			cleanup_session_requests(mdsc, s);
5054 			remove_session_caps(s);
5055 			mutex_unlock(&s->s_mutex);
5056 
5057 			ceph_put_mds_session(s);
5058 
5059 			mutex_lock(&mdsc->mutex);
5060 			kick_requests(mdsc, i);
5061 			continue;
5062 		}
5063 
5064 		if (memcmp(ceph_mdsmap_get_addr(oldmap, i),
5065 			   ceph_mdsmap_get_addr(newmap, i),
5066 			   sizeof(struct ceph_entity_addr))) {
5067 			/* just close it */
5068 			mutex_unlock(&mdsc->mutex);
5069 			mutex_lock(&s->s_mutex);
5070 			mutex_lock(&mdsc->mutex);
5071 			ceph_con_close(&s->s_con);
5072 			mutex_unlock(&s->s_mutex);
5073 			s->s_state = CEPH_MDS_SESSION_RESTARTING;
5074 		} else if (oldstate == newstate) {
5075 			continue;  /* nothing new with this mds */
5076 		}
5077 
5078 		/*
5079 		 * send reconnect?
5080 		 */
5081 		if (s->s_state == CEPH_MDS_SESSION_RESTARTING &&
5082 		    newstate >= CEPH_MDS_STATE_RECONNECT) {
5083 			mutex_unlock(&mdsc->mutex);
5084 			clear_bit(i, targets);
5085 			send_mds_reconnect(mdsc, s);
5086 			mutex_lock(&mdsc->mutex);
5087 		}
5088 
5089 		/*
5090 		 * kick request on any mds that has gone active.
5091 		 */
5092 		if (oldstate < CEPH_MDS_STATE_ACTIVE &&
5093 		    newstate >= CEPH_MDS_STATE_ACTIVE) {
5094 			if (oldstate != CEPH_MDS_STATE_CREATING &&
5095 			    oldstate != CEPH_MDS_STATE_STARTING)
5096 				pr_info_client(cl, "mds%d recovery completed\n",
5097 					       s->s_mds);
5098 			kick_requests(mdsc, i);
5099 			mutex_unlock(&mdsc->mutex);
5100 			mutex_lock(&s->s_mutex);
5101 			mutex_lock(&mdsc->mutex);
5102 			ceph_kick_flushing_caps(mdsc, s);
5103 			mutex_unlock(&s->s_mutex);
5104 			wake_up_session_caps(s, RECONNECT);
5105 		}
5106 	}
5107 
5108 	/*
5109 	 * Only open and reconnect sessions that don't exist yet.
5110 	 */
5111 	for (i = 0; i < newmap->possible_max_rank; i++) {
5112 		/*
5113 		 * In case the import MDS is crashed just after
5114 		 * the EImportStart journal is flushed, so when
5115 		 * a standby MDS takes over it and is replaying
5116 		 * the EImportStart journal the new MDS daemon
5117 		 * will wait the client to reconnect it, but the
5118 		 * client may never register/open the session yet.
5119 		 *
5120 		 * Will try to reconnect that MDS daemon if the
5121 		 * rank number is in the export targets array and
5122 		 * is the up:reconnect state.
5123 		 */
5124 		newstate = ceph_mdsmap_get_state(newmap, i);
5125 		if (!test_bit(i, targets) || newstate != CEPH_MDS_STATE_RECONNECT)
5126 			continue;
5127 
5128 		/*
5129 		 * The session maybe registered and opened by some
5130 		 * requests which were choosing random MDSes during
5131 		 * the mdsc->mutex's unlock/lock gap below in rare
5132 		 * case. But the related MDS daemon will just queue
5133 		 * that requests and be still waiting for the client's
5134 		 * reconnection request in up:reconnect state.
5135 		 */
5136 		s = __ceph_lookup_mds_session(mdsc, i);
5137 		if (likely(!s)) {
5138 			s = __open_export_target_session(mdsc, i);
5139 			if (IS_ERR(s)) {
5140 				err = PTR_ERR(s);
5141 				pr_err_client(cl,
5142 					      "failed to open export target session, err %d\n",
5143 					      err);
5144 				continue;
5145 			}
5146 		}
5147 		doutc(cl, "send reconnect to export target mds.%d\n", i);
5148 		mutex_unlock(&mdsc->mutex);
5149 		send_mds_reconnect(mdsc, s);
5150 		ceph_put_mds_session(s);
5151 		mutex_lock(&mdsc->mutex);
5152 	}
5153 
5154 	for (i = 0; i < newmap->possible_max_rank && i < mdsc->max_sessions; i++) {
5155 		s = mdsc->sessions[i];
5156 		if (!s)
5157 			continue;
5158 		if (!ceph_mdsmap_is_laggy(newmap, i))
5159 			continue;
5160 		if (s->s_state == CEPH_MDS_SESSION_OPEN ||
5161 		    s->s_state == CEPH_MDS_SESSION_HUNG ||
5162 		    s->s_state == CEPH_MDS_SESSION_CLOSING) {
5163 			doutc(cl, " connecting to export targets of laggy mds%d\n", i);
5164 			__open_export_target_sessions(mdsc, s);
5165 		}
5166 	}
5167 }
5168 
5169 
5170 
5171 /*
5172  * leases
5173  */
5174 
5175 /*
5176  * caller must hold session s_mutex, dentry->d_lock
5177  */
5178 void __ceph_mdsc_drop_dentry_lease(struct dentry *dentry)
5179 {
5180 	struct ceph_dentry_info *di = ceph_dentry(dentry);
5181 
5182 	ceph_put_mds_session(di->lease_session);
5183 	di->lease_session = NULL;
5184 }
5185 
5186 static void handle_lease(struct ceph_mds_client *mdsc,
5187 			 struct ceph_mds_session *session,
5188 			 struct ceph_msg *msg)
5189 {
5190 	struct ceph_client *cl = mdsc->fsc->client;
5191 	struct super_block *sb = mdsc->fsc->sb;
5192 	struct inode *inode;
5193 	struct dentry *parent, *dentry;
5194 	struct ceph_dentry_info *di;
5195 	int mds = session->s_mds;
5196 	struct ceph_mds_lease *h = msg->front.iov_base;
5197 	u32 seq;
5198 	struct ceph_vino vino;
5199 	struct qstr dname;
5200 	int release = 0;
5201 
5202 	doutc(cl, "from mds%d\n", mds);
5203 
5204 	if (!ceph_inc_mds_stopping_blocker(mdsc, session))
5205 		return;
5206 
5207 	/* decode */
5208 	if (msg->front.iov_len < sizeof(*h) + sizeof(u32))
5209 		goto bad;
5210 	vino.ino = le64_to_cpu(h->ino);
5211 	vino.snap = CEPH_NOSNAP;
5212 	seq = le32_to_cpu(h->seq);
5213 	dname.len = get_unaligned_le32(h + 1);
5214 	if (msg->front.iov_len < sizeof(*h) + sizeof(u32) + dname.len)
5215 		goto bad;
5216 	dname.name = (void *)(h + 1) + sizeof(u32);
5217 
5218 	/* lookup inode */
5219 	inode = ceph_find_inode(sb, vino);
5220 	doutc(cl, "%s, ino %llx %p %.*s\n", ceph_lease_op_name(h->action),
5221 	      vino.ino, inode, dname.len, dname.name);
5222 
5223 	mutex_lock(&session->s_mutex);
5224 	if (!inode) {
5225 		doutc(cl, "no inode %llx\n", vino.ino);
5226 		goto release;
5227 	}
5228 
5229 	/* dentry */
5230 	parent = d_find_alias(inode);
5231 	if (!parent) {
5232 		doutc(cl, "no parent dentry on inode %p\n", inode);
5233 		WARN_ON(1);
5234 		goto release;  /* hrm... */
5235 	}
5236 	dname.hash = full_name_hash(parent, dname.name, dname.len);
5237 	dentry = d_lookup(parent, &dname);
5238 	dput(parent);
5239 	if (!dentry)
5240 		goto release;
5241 
5242 	spin_lock(&dentry->d_lock);
5243 	di = ceph_dentry(dentry);
5244 	switch (h->action) {
5245 	case CEPH_MDS_LEASE_REVOKE:
5246 		if (di->lease_session == session) {
5247 			if (ceph_seq_cmp(di->lease_seq, seq) > 0)
5248 				h->seq = cpu_to_le32(di->lease_seq);
5249 			__ceph_mdsc_drop_dentry_lease(dentry);
5250 		}
5251 		release = 1;
5252 		break;
5253 
5254 	case CEPH_MDS_LEASE_RENEW:
5255 		if (di->lease_session == session &&
5256 		    di->lease_gen == atomic_read(&session->s_cap_gen) &&
5257 		    di->lease_renew_from &&
5258 		    di->lease_renew_after == 0) {
5259 			unsigned long duration =
5260 				msecs_to_jiffies(le32_to_cpu(h->duration_ms));
5261 
5262 			di->lease_seq = seq;
5263 			di->time = di->lease_renew_from + duration;
5264 			di->lease_renew_after = di->lease_renew_from +
5265 				(duration >> 1);
5266 			di->lease_renew_from = 0;
5267 		}
5268 		break;
5269 	}
5270 	spin_unlock(&dentry->d_lock);
5271 	dput(dentry);
5272 
5273 	if (!release)
5274 		goto out;
5275 
5276 release:
5277 	/* let's just reuse the same message */
5278 	h->action = CEPH_MDS_LEASE_REVOKE_ACK;
5279 	ceph_msg_get(msg);
5280 	ceph_con_send(&session->s_con, msg);
5281 
5282 out:
5283 	mutex_unlock(&session->s_mutex);
5284 	iput(inode);
5285 
5286 	ceph_dec_mds_stopping_blocker(mdsc);
5287 	return;
5288 
5289 bad:
5290 	ceph_dec_mds_stopping_blocker(mdsc);
5291 
5292 	pr_err_client(cl, "corrupt lease message\n");
5293 	ceph_msg_dump(msg);
5294 }
5295 
5296 void ceph_mdsc_lease_send_msg(struct ceph_mds_session *session,
5297 			      struct dentry *dentry, char action,
5298 			      u32 seq)
5299 {
5300 	struct ceph_client *cl = session->s_mdsc->fsc->client;
5301 	struct ceph_msg *msg;
5302 	struct ceph_mds_lease *lease;
5303 	struct inode *dir;
5304 	int len = sizeof(*lease) + sizeof(u32) + NAME_MAX;
5305 
5306 	doutc(cl, "identry %p %s to mds%d\n", dentry, ceph_lease_op_name(action),
5307 	      session->s_mds);
5308 
5309 	msg = ceph_msg_new(CEPH_MSG_CLIENT_LEASE, len, GFP_NOFS, false);
5310 	if (!msg)
5311 		return;
5312 	lease = msg->front.iov_base;
5313 	lease->action = action;
5314 	lease->seq = cpu_to_le32(seq);
5315 
5316 	spin_lock(&dentry->d_lock);
5317 	dir = d_inode(dentry->d_parent);
5318 	lease->ino = cpu_to_le64(ceph_ino(dir));
5319 	lease->first = lease->last = cpu_to_le64(ceph_snap(dir));
5320 
5321 	put_unaligned_le32(dentry->d_name.len, lease + 1);
5322 	memcpy((void *)(lease + 1) + 4,
5323 	       dentry->d_name.name, dentry->d_name.len);
5324 	spin_unlock(&dentry->d_lock);
5325 
5326 	ceph_con_send(&session->s_con, msg);
5327 }
5328 
5329 /*
5330  * lock unlock the session, to wait ongoing session activities
5331  */
5332 static void lock_unlock_session(struct ceph_mds_session *s)
5333 {
5334 	mutex_lock(&s->s_mutex);
5335 	mutex_unlock(&s->s_mutex);
5336 }
5337 
5338 static void maybe_recover_session(struct ceph_mds_client *mdsc)
5339 {
5340 	struct ceph_client *cl = mdsc->fsc->client;
5341 	struct ceph_fs_client *fsc = mdsc->fsc;
5342 
5343 	if (!ceph_test_mount_opt(fsc, CLEANRECOVER))
5344 		return;
5345 
5346 	if (READ_ONCE(fsc->mount_state) != CEPH_MOUNT_MOUNTED)
5347 		return;
5348 
5349 	if (!READ_ONCE(fsc->blocklisted))
5350 		return;
5351 
5352 	pr_info_client(cl, "auto reconnect after blocklisted\n");
5353 	ceph_force_reconnect(fsc->sb);
5354 }
5355 
5356 bool check_session_state(struct ceph_mds_session *s)
5357 {
5358 	struct ceph_client *cl = s->s_mdsc->fsc->client;
5359 
5360 	switch (s->s_state) {
5361 	case CEPH_MDS_SESSION_OPEN:
5362 		if (s->s_ttl && time_after(jiffies, s->s_ttl)) {
5363 			s->s_state = CEPH_MDS_SESSION_HUNG;
5364 			pr_info_client(cl, "mds%d hung\n", s->s_mds);
5365 		}
5366 		break;
5367 	case CEPH_MDS_SESSION_CLOSING:
5368 	case CEPH_MDS_SESSION_NEW:
5369 	case CEPH_MDS_SESSION_RESTARTING:
5370 	case CEPH_MDS_SESSION_CLOSED:
5371 	case CEPH_MDS_SESSION_REJECTED:
5372 		return false;
5373 	}
5374 
5375 	return true;
5376 }
5377 
5378 /*
5379  * If the sequence is incremented while we're waiting on a REQUEST_CLOSE reply,
5380  * then we need to retransmit that request.
5381  */
5382 void inc_session_sequence(struct ceph_mds_session *s)
5383 {
5384 	struct ceph_client *cl = s->s_mdsc->fsc->client;
5385 
5386 	lockdep_assert_held(&s->s_mutex);
5387 
5388 	s->s_seq++;
5389 
5390 	if (s->s_state == CEPH_MDS_SESSION_CLOSING) {
5391 		int ret;
5392 
5393 		doutc(cl, "resending session close request for mds%d\n", s->s_mds);
5394 		ret = request_close_session(s);
5395 		if (ret < 0)
5396 			pr_err_client(cl, "unable to close session to mds%d: %d\n",
5397 				      s->s_mds, ret);
5398 	}
5399 }
5400 
5401 /*
5402  * delayed work -- periodically trim expired leases, renew caps with mds.  If
5403  * the @delay parameter is set to 0 or if it's more than 5 secs, the default
5404  * workqueue delay value of 5 secs will be used.
5405  */
5406 static void schedule_delayed(struct ceph_mds_client *mdsc, unsigned long delay)
5407 {
5408 	unsigned long max_delay = HZ * 5;
5409 
5410 	/* 5 secs default delay */
5411 	if (!delay || (delay > max_delay))
5412 		delay = max_delay;
5413 	schedule_delayed_work(&mdsc->delayed_work,
5414 			      round_jiffies_relative(delay));
5415 }
5416 
5417 static void delayed_work(struct work_struct *work)
5418 {
5419 	struct ceph_mds_client *mdsc =
5420 		container_of(work, struct ceph_mds_client, delayed_work.work);
5421 	unsigned long delay;
5422 	int renew_interval;
5423 	int renew_caps;
5424 	int i;
5425 
5426 	doutc(mdsc->fsc->client, "mdsc delayed_work\n");
5427 
5428 	if (mdsc->stopping >= CEPH_MDSC_STOPPING_FLUSHED)
5429 		return;
5430 
5431 	mutex_lock(&mdsc->mutex);
5432 	renew_interval = mdsc->mdsmap->m_session_timeout >> 2;
5433 	renew_caps = time_after_eq(jiffies, HZ*renew_interval +
5434 				   mdsc->last_renew_caps);
5435 	if (renew_caps)
5436 		mdsc->last_renew_caps = jiffies;
5437 
5438 	for (i = 0; i < mdsc->max_sessions; i++) {
5439 		struct ceph_mds_session *s = __ceph_lookup_mds_session(mdsc, i);
5440 		if (!s)
5441 			continue;
5442 
5443 		if (!check_session_state(s)) {
5444 			ceph_put_mds_session(s);
5445 			continue;
5446 		}
5447 		mutex_unlock(&mdsc->mutex);
5448 
5449 		ceph_flush_session_cap_releases(mdsc, s);
5450 
5451 		mutex_lock(&s->s_mutex);
5452 		if (renew_caps)
5453 			send_renew_caps(mdsc, s);
5454 		else
5455 			ceph_con_keepalive(&s->s_con);
5456 		if (s->s_state == CEPH_MDS_SESSION_OPEN ||
5457 		    s->s_state == CEPH_MDS_SESSION_HUNG)
5458 			ceph_send_cap_releases(mdsc, s);
5459 		mutex_unlock(&s->s_mutex);
5460 		ceph_put_mds_session(s);
5461 
5462 		mutex_lock(&mdsc->mutex);
5463 	}
5464 	mutex_unlock(&mdsc->mutex);
5465 
5466 	delay = ceph_check_delayed_caps(mdsc);
5467 
5468 	ceph_queue_cap_reclaim_work(mdsc);
5469 
5470 	ceph_trim_snapid_map(mdsc);
5471 
5472 	maybe_recover_session(mdsc);
5473 
5474 	schedule_delayed(mdsc, delay);
5475 }
5476 
5477 int ceph_mdsc_init(struct ceph_fs_client *fsc)
5478 
5479 {
5480 	struct ceph_mds_client *mdsc;
5481 	int err;
5482 
5483 	mdsc = kzalloc(sizeof(struct ceph_mds_client), GFP_NOFS);
5484 	if (!mdsc)
5485 		return -ENOMEM;
5486 	mdsc->fsc = fsc;
5487 	mutex_init(&mdsc->mutex);
5488 	mdsc->mdsmap = kzalloc(sizeof(*mdsc->mdsmap), GFP_NOFS);
5489 	if (!mdsc->mdsmap) {
5490 		err = -ENOMEM;
5491 		goto err_mdsc;
5492 	}
5493 
5494 	init_completion(&mdsc->safe_umount_waiters);
5495 	spin_lock_init(&mdsc->stopping_lock);
5496 	atomic_set(&mdsc->stopping_blockers, 0);
5497 	init_completion(&mdsc->stopping_waiter);
5498 	init_waitqueue_head(&mdsc->session_close_wq);
5499 	INIT_LIST_HEAD(&mdsc->waiting_for_map);
5500 	mdsc->quotarealms_inodes = RB_ROOT;
5501 	mutex_init(&mdsc->quotarealms_inodes_mutex);
5502 	init_rwsem(&mdsc->snap_rwsem);
5503 	mdsc->snap_realms = RB_ROOT;
5504 	INIT_LIST_HEAD(&mdsc->snap_empty);
5505 	spin_lock_init(&mdsc->snap_empty_lock);
5506 	mdsc->request_tree = RB_ROOT;
5507 	INIT_DELAYED_WORK(&mdsc->delayed_work, delayed_work);
5508 	mdsc->last_renew_caps = jiffies;
5509 	INIT_LIST_HEAD(&mdsc->cap_delay_list);
5510 #ifdef CONFIG_DEBUG_FS
5511 	INIT_LIST_HEAD(&mdsc->cap_wait_list);
5512 #endif
5513 	spin_lock_init(&mdsc->cap_delay_lock);
5514 	INIT_LIST_HEAD(&mdsc->cap_unlink_delay_list);
5515 	INIT_LIST_HEAD(&mdsc->snap_flush_list);
5516 	spin_lock_init(&mdsc->snap_flush_lock);
5517 	mdsc->last_cap_flush_tid = 1;
5518 	INIT_LIST_HEAD(&mdsc->cap_flush_list);
5519 	INIT_LIST_HEAD(&mdsc->cap_dirty_migrating);
5520 	spin_lock_init(&mdsc->cap_dirty_lock);
5521 	init_waitqueue_head(&mdsc->cap_flushing_wq);
5522 	INIT_WORK(&mdsc->cap_reclaim_work, ceph_cap_reclaim_work);
5523 	INIT_WORK(&mdsc->cap_unlink_work, ceph_cap_unlink_work);
5524 	err = ceph_metric_init(&mdsc->metric);
5525 	if (err)
5526 		goto err_mdsmap;
5527 
5528 	spin_lock_init(&mdsc->dentry_list_lock);
5529 	INIT_LIST_HEAD(&mdsc->dentry_leases);
5530 	INIT_LIST_HEAD(&mdsc->dentry_dir_leases);
5531 
5532 	ceph_caps_init(mdsc);
5533 	ceph_adjust_caps_max_min(mdsc, fsc->mount_options);
5534 
5535 	spin_lock_init(&mdsc->snapid_map_lock);
5536 	mdsc->snapid_map_tree = RB_ROOT;
5537 	INIT_LIST_HEAD(&mdsc->snapid_map_lru);
5538 
5539 	init_rwsem(&mdsc->pool_perm_rwsem);
5540 	mdsc->pool_perm_tree = RB_ROOT;
5541 
5542 	strscpy(mdsc->nodename, utsname()->nodename,
5543 		sizeof(mdsc->nodename));
5544 
5545 	fsc->mdsc = mdsc;
5546 	return 0;
5547 
5548 err_mdsmap:
5549 	kfree(mdsc->mdsmap);
5550 err_mdsc:
5551 	kfree(mdsc);
5552 	return err;
5553 }
5554 
5555 /*
5556  * Wait for safe replies on open mds requests.  If we time out, drop
5557  * all requests from the tree to avoid dangling dentry refs.
5558  */
5559 static void wait_requests(struct ceph_mds_client *mdsc)
5560 {
5561 	struct ceph_client *cl = mdsc->fsc->client;
5562 	struct ceph_options *opts = mdsc->fsc->client->options;
5563 	struct ceph_mds_request *req;
5564 
5565 	mutex_lock(&mdsc->mutex);
5566 	if (__get_oldest_req(mdsc)) {
5567 		mutex_unlock(&mdsc->mutex);
5568 
5569 		doutc(cl, "waiting for requests\n");
5570 		wait_for_completion_timeout(&mdsc->safe_umount_waiters,
5571 				    ceph_timeout_jiffies(opts->mount_timeout));
5572 
5573 		/* tear down remaining requests */
5574 		mutex_lock(&mdsc->mutex);
5575 		while ((req = __get_oldest_req(mdsc))) {
5576 			doutc(cl, "timed out on tid %llu\n", req->r_tid);
5577 			list_del_init(&req->r_wait);
5578 			__unregister_request(mdsc, req);
5579 		}
5580 	}
5581 	mutex_unlock(&mdsc->mutex);
5582 	doutc(cl, "done\n");
5583 }
5584 
5585 void send_flush_mdlog(struct ceph_mds_session *s)
5586 {
5587 	struct ceph_client *cl = s->s_mdsc->fsc->client;
5588 	struct ceph_msg *msg;
5589 
5590 	/*
5591 	 * Pre-luminous MDS crashes when it sees an unknown session request
5592 	 */
5593 	if (!CEPH_HAVE_FEATURE(s->s_con.peer_features, SERVER_LUMINOUS))
5594 		return;
5595 
5596 	mutex_lock(&s->s_mutex);
5597 	doutc(cl, "request mdlog flush to mds%d (%s)s seq %lld\n",
5598 	      s->s_mds, ceph_session_state_name(s->s_state), s->s_seq);
5599 	msg = ceph_create_session_msg(CEPH_SESSION_REQUEST_FLUSH_MDLOG,
5600 				      s->s_seq);
5601 	if (!msg) {
5602 		pr_err_client(cl, "failed to request mdlog flush to mds%d (%s) seq %lld\n",
5603 			      s->s_mds, ceph_session_state_name(s->s_state), s->s_seq);
5604 	} else {
5605 		ceph_con_send(&s->s_con, msg);
5606 	}
5607 	mutex_unlock(&s->s_mutex);
5608 }
5609 
5610 static int ceph_mds_auth_match(struct ceph_mds_client *mdsc,
5611 			       struct ceph_mds_cap_auth *auth,
5612 			       char *tpath)
5613 {
5614 	const struct cred *cred = get_current_cred();
5615 	u32 caller_uid = from_kuid(&init_user_ns, cred->fsuid);
5616 	u32 caller_gid = from_kgid(&init_user_ns, cred->fsgid);
5617 	struct ceph_client *cl = mdsc->fsc->client;
5618 	const char *spath = mdsc->fsc->mount_options->server_path;
5619 	bool gid_matched = false;
5620 	u32 gid, tlen, len;
5621 	int i, j;
5622 
5623 	doutc(cl, "match.uid %lld\n", auth->match.uid);
5624 	if (auth->match.uid != MDS_AUTH_UID_ANY) {
5625 		if (auth->match.uid != caller_uid)
5626 			return 0;
5627 		if (auth->match.num_gids) {
5628 			for (i = 0; i < auth->match.num_gids; i++) {
5629 				if (caller_gid == auth->match.gids[i])
5630 					gid_matched = true;
5631 			}
5632 			if (!gid_matched && cred->group_info->ngroups) {
5633 				for (i = 0; i < cred->group_info->ngroups; i++) {
5634 					gid = from_kgid(&init_user_ns,
5635 							cred->group_info->gid[i]);
5636 					for (j = 0; j < auth->match.num_gids; j++) {
5637 						if (gid == auth->match.gids[j]) {
5638 							gid_matched = true;
5639 							break;
5640 						}
5641 					}
5642 					if (gid_matched)
5643 						break;
5644 				}
5645 			}
5646 			if (!gid_matched)
5647 				return 0;
5648 		}
5649 	}
5650 
5651 	/* path match */
5652 	if (auth->match.path) {
5653 		if (!tpath)
5654 			return 0;
5655 
5656 		tlen = strlen(tpath);
5657 		len = strlen(auth->match.path);
5658 		if (len) {
5659 			char *_tpath = tpath;
5660 			bool free_tpath = false;
5661 			int m, n;
5662 
5663 			doutc(cl, "server path %s, tpath %s, match.path %s\n",
5664 			      spath, tpath, auth->match.path);
5665 			if (spath && (m = strlen(spath)) != 1) {
5666 				/* mount path + '/' + tpath + an extra space */
5667 				n = m + 1 + tlen + 1;
5668 				_tpath = kmalloc(n, GFP_NOFS);
5669 				if (!_tpath)
5670 					return -ENOMEM;
5671 				/* remove the leading '/' */
5672 				snprintf(_tpath, n, "%s/%s", spath + 1, tpath);
5673 				free_tpath = true;
5674 				tlen = strlen(_tpath);
5675 			}
5676 
5677 			/*
5678 			 * Please note the tailing '/' for match.path has already
5679 			 * been removed when parsing.
5680 			 *
5681 			 * Remove the tailing '/' for the target path.
5682 			 */
5683 			while (tlen && _tpath[tlen - 1] == '/') {
5684 				_tpath[tlen - 1] = '\0';
5685 				tlen -= 1;
5686 			}
5687 			doutc(cl, "_tpath %s\n", _tpath);
5688 
5689 			/*
5690 			 * In case first == _tpath && tlen == len:
5691 			 *  match.path=/foo  --> /foo _path=/foo     --> match
5692 			 *  match.path=/foo/ --> /foo _path=/foo     --> match
5693 			 *
5694 			 * In case first == _tmatch.path && tlen > len:
5695 			 *  match.path=/foo/ --> /foo _path=/foo/    --> match
5696 			 *  match.path=/foo  --> /foo _path=/foo/    --> match
5697 			 *  match.path=/foo/ --> /foo _path=/foo/d   --> match
5698 			 *  match.path=/foo  --> /foo _path=/food    --> mismatch
5699 			 *
5700 			 * All the other cases                       --> mismatch
5701 			 */
5702 			char *first = strstr(_tpath, auth->match.path);
5703 			if (first != _tpath) {
5704 				if (free_tpath)
5705 					kfree(_tpath);
5706 				return 0;
5707 			}
5708 
5709 			if (tlen > len && _tpath[len] != '/') {
5710 				if (free_tpath)
5711 					kfree(_tpath);
5712 				return 0;
5713 			}
5714 		}
5715 	}
5716 
5717 	doutc(cl, "matched\n");
5718 	return 1;
5719 }
5720 
5721 int ceph_mds_check_access(struct ceph_mds_client *mdsc, char *tpath, int mask)
5722 {
5723 	const struct cred *cred = get_current_cred();
5724 	u32 caller_uid = from_kuid(&init_user_ns, cred->fsuid);
5725 	u32 caller_gid = from_kgid(&init_user_ns, cred->fsgid);
5726 	struct ceph_mds_cap_auth *rw_perms_s = NULL;
5727 	struct ceph_client *cl = mdsc->fsc->client;
5728 	bool root_squash_perms = true;
5729 	int i, err;
5730 
5731 	doutc(cl, "tpath '%s', mask %d, caller_uid %d, caller_gid %d\n",
5732 	      tpath, mask, caller_uid, caller_gid);
5733 
5734 	for (i = 0; i < mdsc->s_cap_auths_num; i++) {
5735 		struct ceph_mds_cap_auth *s = &mdsc->s_cap_auths[i];
5736 
5737 		err = ceph_mds_auth_match(mdsc, s, tpath);
5738 		if (err < 0) {
5739 			return err;
5740 		} else if (err > 0) {
5741 			/* always follow the last auth caps' permision */
5742 			root_squash_perms = true;
5743 			rw_perms_s = NULL;
5744 			if ((mask & MAY_WRITE) && s->writeable &&
5745 			    s->match.root_squash && (!caller_uid || !caller_gid))
5746 				root_squash_perms = false;
5747 
5748 			if (((mask & MAY_WRITE) && !s->writeable) ||
5749 			    ((mask & MAY_READ) && !s->readable))
5750 				rw_perms_s = s;
5751 		}
5752 	}
5753 
5754 	doutc(cl, "root_squash_perms %d, rw_perms_s %p\n", root_squash_perms,
5755 	      rw_perms_s);
5756 	if (root_squash_perms && rw_perms_s == NULL) {
5757 		doutc(cl, "access allowed\n");
5758 		return 0;
5759 	}
5760 
5761 	if (!root_squash_perms) {
5762 		doutc(cl, "root_squash is enabled and user(%d %d) isn't allowed to write",
5763 		      caller_uid, caller_gid);
5764 	}
5765 	if (rw_perms_s) {
5766 		doutc(cl, "mds auth caps readable/writeable %d/%d while request r/w %d/%d",
5767 		      rw_perms_s->readable, rw_perms_s->writeable,
5768 		      !!(mask & MAY_READ), !!(mask & MAY_WRITE));
5769 	}
5770 	doutc(cl, "access denied\n");
5771 	return -EACCES;
5772 }
5773 
5774 /*
5775  * called before mount is ro, and before dentries are torn down.
5776  * (hmm, does this still race with new lookups?)
5777  */
5778 void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc)
5779 {
5780 	doutc(mdsc->fsc->client, "begin\n");
5781 	mdsc->stopping = CEPH_MDSC_STOPPING_BEGIN;
5782 
5783 	ceph_mdsc_iterate_sessions(mdsc, send_flush_mdlog, true);
5784 	ceph_mdsc_iterate_sessions(mdsc, lock_unlock_session, false);
5785 	ceph_flush_dirty_caps(mdsc);
5786 	wait_requests(mdsc);
5787 
5788 	/*
5789 	 * wait for reply handlers to drop their request refs and
5790 	 * their inode/dcache refs
5791 	 */
5792 	ceph_msgr_flush();
5793 
5794 	ceph_cleanup_quotarealms_inodes(mdsc);
5795 	doutc(mdsc->fsc->client, "done\n");
5796 }
5797 
5798 /*
5799  * flush the mdlog and wait for all write mds requests to flush.
5800  */
5801 static void flush_mdlog_and_wait_mdsc_unsafe_requests(struct ceph_mds_client *mdsc,
5802 						 u64 want_tid)
5803 {
5804 	struct ceph_client *cl = mdsc->fsc->client;
5805 	struct ceph_mds_request *req = NULL, *nextreq;
5806 	struct ceph_mds_session *last_session = NULL;
5807 	struct rb_node *n;
5808 
5809 	mutex_lock(&mdsc->mutex);
5810 	doutc(cl, "want %lld\n", want_tid);
5811 restart:
5812 	req = __get_oldest_req(mdsc);
5813 	while (req && req->r_tid <= want_tid) {
5814 		/* find next request */
5815 		n = rb_next(&req->r_node);
5816 		if (n)
5817 			nextreq = rb_entry(n, struct ceph_mds_request, r_node);
5818 		else
5819 			nextreq = NULL;
5820 		if (req->r_op != CEPH_MDS_OP_SETFILELOCK &&
5821 		    (req->r_op & CEPH_MDS_OP_WRITE)) {
5822 			struct ceph_mds_session *s = req->r_session;
5823 
5824 			if (!s) {
5825 				req = nextreq;
5826 				continue;
5827 			}
5828 
5829 			/* write op */
5830 			ceph_mdsc_get_request(req);
5831 			if (nextreq)
5832 				ceph_mdsc_get_request(nextreq);
5833 			s = ceph_get_mds_session(s);
5834 			mutex_unlock(&mdsc->mutex);
5835 
5836 			/* send flush mdlog request to MDS */
5837 			if (last_session != s) {
5838 				send_flush_mdlog(s);
5839 				ceph_put_mds_session(last_session);
5840 				last_session = s;
5841 			} else {
5842 				ceph_put_mds_session(s);
5843 			}
5844 			doutc(cl, "wait on %llu (want %llu)\n",
5845 			      req->r_tid, want_tid);
5846 			wait_for_completion(&req->r_safe_completion);
5847 
5848 			mutex_lock(&mdsc->mutex);
5849 			ceph_mdsc_put_request(req);
5850 			if (!nextreq)
5851 				break;  /* next dne before, so we're done! */
5852 			if (RB_EMPTY_NODE(&nextreq->r_node)) {
5853 				/* next request was removed from tree */
5854 				ceph_mdsc_put_request(nextreq);
5855 				goto restart;
5856 			}
5857 			ceph_mdsc_put_request(nextreq);  /* won't go away */
5858 		}
5859 		req = nextreq;
5860 	}
5861 	mutex_unlock(&mdsc->mutex);
5862 	ceph_put_mds_session(last_session);
5863 	doutc(cl, "done\n");
5864 }
5865 
5866 void ceph_mdsc_sync(struct ceph_mds_client *mdsc)
5867 {
5868 	struct ceph_client *cl = mdsc->fsc->client;
5869 	u64 want_tid, want_flush;
5870 
5871 	if (READ_ONCE(mdsc->fsc->mount_state) >= CEPH_MOUNT_SHUTDOWN)
5872 		return;
5873 
5874 	doutc(cl, "sync\n");
5875 	mutex_lock(&mdsc->mutex);
5876 	want_tid = mdsc->last_tid;
5877 	mutex_unlock(&mdsc->mutex);
5878 
5879 	ceph_flush_dirty_caps(mdsc);
5880 	ceph_flush_cap_releases(mdsc);
5881 	spin_lock(&mdsc->cap_dirty_lock);
5882 	want_flush = mdsc->last_cap_flush_tid;
5883 	if (!list_empty(&mdsc->cap_flush_list)) {
5884 		struct ceph_cap_flush *cf =
5885 			list_last_entry(&mdsc->cap_flush_list,
5886 					struct ceph_cap_flush, g_list);
5887 		cf->wake = true;
5888 	}
5889 	spin_unlock(&mdsc->cap_dirty_lock);
5890 
5891 	doutc(cl, "sync want tid %lld flush_seq %lld\n", want_tid, want_flush);
5892 
5893 	flush_mdlog_and_wait_mdsc_unsafe_requests(mdsc, want_tid);
5894 	wait_caps_flush(mdsc, want_flush);
5895 }
5896 
5897 /*
5898  * true if all sessions are closed, or we force unmount
5899  */
5900 static bool done_closing_sessions(struct ceph_mds_client *mdsc, int skipped)
5901 {
5902 	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
5903 		return true;
5904 	return atomic_read(&mdsc->num_sessions) <= skipped;
5905 }
5906 
5907 /*
5908  * called after sb is ro or when metadata corrupted.
5909  */
5910 void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc)
5911 {
5912 	struct ceph_options *opts = mdsc->fsc->client->options;
5913 	struct ceph_client *cl = mdsc->fsc->client;
5914 	struct ceph_mds_session *session;
5915 	int i;
5916 	int skipped = 0;
5917 
5918 	doutc(cl, "begin\n");
5919 
5920 	/* close sessions */
5921 	mutex_lock(&mdsc->mutex);
5922 	for (i = 0; i < mdsc->max_sessions; i++) {
5923 		session = __ceph_lookup_mds_session(mdsc, i);
5924 		if (!session)
5925 			continue;
5926 		mutex_unlock(&mdsc->mutex);
5927 		mutex_lock(&session->s_mutex);
5928 		if (__close_session(mdsc, session) <= 0)
5929 			skipped++;
5930 		mutex_unlock(&session->s_mutex);
5931 		ceph_put_mds_session(session);
5932 		mutex_lock(&mdsc->mutex);
5933 	}
5934 	mutex_unlock(&mdsc->mutex);
5935 
5936 	doutc(cl, "waiting for sessions to close\n");
5937 	wait_event_timeout(mdsc->session_close_wq,
5938 			   done_closing_sessions(mdsc, skipped),
5939 			   ceph_timeout_jiffies(opts->mount_timeout));
5940 
5941 	/* tear down remaining sessions */
5942 	mutex_lock(&mdsc->mutex);
5943 	for (i = 0; i < mdsc->max_sessions; i++) {
5944 		if (mdsc->sessions[i]) {
5945 			session = ceph_get_mds_session(mdsc->sessions[i]);
5946 			__unregister_session(mdsc, session);
5947 			mutex_unlock(&mdsc->mutex);
5948 			mutex_lock(&session->s_mutex);
5949 			remove_session_caps(session);
5950 			mutex_unlock(&session->s_mutex);
5951 			ceph_put_mds_session(session);
5952 			mutex_lock(&mdsc->mutex);
5953 		}
5954 	}
5955 	WARN_ON(!list_empty(&mdsc->cap_delay_list));
5956 	mutex_unlock(&mdsc->mutex);
5957 
5958 	ceph_cleanup_snapid_map(mdsc);
5959 	ceph_cleanup_global_and_empty_realms(mdsc);
5960 
5961 	cancel_work_sync(&mdsc->cap_reclaim_work);
5962 	cancel_work_sync(&mdsc->cap_unlink_work);
5963 	cancel_delayed_work_sync(&mdsc->delayed_work); /* cancel timer */
5964 
5965 	doutc(cl, "done\n");
5966 }
5967 
5968 void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc)
5969 {
5970 	struct ceph_mds_session *session;
5971 	int mds;
5972 
5973 	doutc(mdsc->fsc->client, "force umount\n");
5974 
5975 	mutex_lock(&mdsc->mutex);
5976 	for (mds = 0; mds < mdsc->max_sessions; mds++) {
5977 		session = __ceph_lookup_mds_session(mdsc, mds);
5978 		if (!session)
5979 			continue;
5980 
5981 		if (session->s_state == CEPH_MDS_SESSION_REJECTED)
5982 			__unregister_session(mdsc, session);
5983 		__wake_requests(mdsc, &session->s_waiting);
5984 		mutex_unlock(&mdsc->mutex);
5985 
5986 		mutex_lock(&session->s_mutex);
5987 		__close_session(mdsc, session);
5988 		if (session->s_state == CEPH_MDS_SESSION_CLOSING) {
5989 			cleanup_session_requests(mdsc, session);
5990 			remove_session_caps(session);
5991 		}
5992 		mutex_unlock(&session->s_mutex);
5993 		ceph_put_mds_session(session);
5994 
5995 		mutex_lock(&mdsc->mutex);
5996 		kick_requests(mdsc, mds);
5997 	}
5998 	__wake_requests(mdsc, &mdsc->waiting_for_map);
5999 	mutex_unlock(&mdsc->mutex);
6000 }
6001 
6002 static void ceph_mdsc_stop(struct ceph_mds_client *mdsc)
6003 {
6004 	doutc(mdsc->fsc->client, "stop\n");
6005 	/*
6006 	 * Make sure the delayed work stopped before releasing
6007 	 * the resources.
6008 	 *
6009 	 * Because the cancel_delayed_work_sync() will only
6010 	 * guarantee that the work finishes executing. But the
6011 	 * delayed work will re-arm itself again after that.
6012 	 */
6013 	flush_delayed_work(&mdsc->delayed_work);
6014 
6015 	if (mdsc->mdsmap)
6016 		ceph_mdsmap_destroy(mdsc->mdsmap);
6017 	kfree(mdsc->sessions);
6018 	ceph_caps_finalize(mdsc);
6019 
6020 	if (mdsc->s_cap_auths) {
6021 		int i;
6022 
6023 		for (i = 0; i < mdsc->s_cap_auths_num; i++) {
6024 			kfree(mdsc->s_cap_auths[i].match.gids);
6025 			kfree(mdsc->s_cap_auths[i].match.path);
6026 			kfree(mdsc->s_cap_auths[i].match.fs_name);
6027 		}
6028 		kfree(mdsc->s_cap_auths);
6029 	}
6030 
6031 	ceph_pool_perm_destroy(mdsc);
6032 }
6033 
6034 void ceph_mdsc_destroy(struct ceph_fs_client *fsc)
6035 {
6036 	struct ceph_mds_client *mdsc = fsc->mdsc;
6037 	doutc(fsc->client, "%p\n", mdsc);
6038 
6039 	if (!mdsc)
6040 		return;
6041 
6042 	/* flush out any connection work with references to us */
6043 	ceph_msgr_flush();
6044 
6045 	ceph_mdsc_stop(mdsc);
6046 
6047 	ceph_metric_destroy(&mdsc->metric);
6048 
6049 	fsc->mdsc = NULL;
6050 	kfree(mdsc);
6051 	doutc(fsc->client, "%p done\n", mdsc);
6052 }
6053 
6054 void ceph_mdsc_handle_fsmap(struct ceph_mds_client *mdsc, struct ceph_msg *msg)
6055 {
6056 	struct ceph_fs_client *fsc = mdsc->fsc;
6057 	struct ceph_client *cl = fsc->client;
6058 	const char *mds_namespace = fsc->mount_options->mds_namespace;
6059 	void *p = msg->front.iov_base;
6060 	void *end = p + msg->front.iov_len;
6061 	u32 epoch;
6062 	u32 num_fs;
6063 	u32 mount_fscid = (u32)-1;
6064 	int err = -EINVAL;
6065 
6066 	ceph_decode_need(&p, end, sizeof(u32), bad);
6067 	epoch = ceph_decode_32(&p);
6068 
6069 	doutc(cl, "epoch %u\n", epoch);
6070 
6071 	/* struct_v, struct_cv, map_len, epoch, legacy_client_fscid */
6072 	ceph_decode_skip_n(&p, end, 2 + sizeof(u32) * 3, bad);
6073 
6074 	ceph_decode_32_safe(&p, end, num_fs, bad);
6075 	while (num_fs-- > 0) {
6076 		void *info_p, *info_end;
6077 		u32 info_len;
6078 		u32 fscid, namelen;
6079 
6080 		ceph_decode_need(&p, end, 2 + sizeof(u32), bad);
6081 		p += 2;		// info_v, info_cv
6082 		info_len = ceph_decode_32(&p);
6083 		ceph_decode_need(&p, end, info_len, bad);
6084 		info_p = p;
6085 		info_end = p + info_len;
6086 		p = info_end;
6087 
6088 		ceph_decode_need(&info_p, info_end, sizeof(u32) * 2, bad);
6089 		fscid = ceph_decode_32(&info_p);
6090 		namelen = ceph_decode_32(&info_p);
6091 		ceph_decode_need(&info_p, info_end, namelen, bad);
6092 
6093 		if (mds_namespace &&
6094 		    strlen(mds_namespace) == namelen &&
6095 		    !strncmp(mds_namespace, (char *)info_p, namelen)) {
6096 			mount_fscid = fscid;
6097 			break;
6098 		}
6099 	}
6100 
6101 	ceph_monc_got_map(&fsc->client->monc, CEPH_SUB_FSMAP, epoch);
6102 	if (mount_fscid != (u32)-1) {
6103 		fsc->client->monc.fs_cluster_id = mount_fscid;
6104 		ceph_monc_want_map(&fsc->client->monc, CEPH_SUB_MDSMAP,
6105 				   0, true);
6106 		ceph_monc_renew_subs(&fsc->client->monc);
6107 	} else {
6108 		err = -ENOENT;
6109 		goto err_out;
6110 	}
6111 	return;
6112 
6113 bad:
6114 	pr_err_client(cl, "error decoding fsmap %d. Shutting down mount.\n",
6115 		      err);
6116 	ceph_umount_begin(mdsc->fsc->sb);
6117 	ceph_msg_dump(msg);
6118 err_out:
6119 	mutex_lock(&mdsc->mutex);
6120 	mdsc->mdsmap_err = err;
6121 	__wake_requests(mdsc, &mdsc->waiting_for_map);
6122 	mutex_unlock(&mdsc->mutex);
6123 }
6124 
6125 /*
6126  * handle mds map update.
6127  */
6128 void ceph_mdsc_handle_mdsmap(struct ceph_mds_client *mdsc, struct ceph_msg *msg)
6129 {
6130 	struct ceph_client *cl = mdsc->fsc->client;
6131 	u32 epoch;
6132 	u32 maplen;
6133 	void *p = msg->front.iov_base;
6134 	void *end = p + msg->front.iov_len;
6135 	struct ceph_mdsmap *newmap, *oldmap;
6136 	struct ceph_fsid fsid;
6137 	int err = -EINVAL;
6138 
6139 	ceph_decode_need(&p, end, sizeof(fsid)+2*sizeof(u32), bad);
6140 	ceph_decode_copy(&p, &fsid, sizeof(fsid));
6141 	if (ceph_check_fsid(mdsc->fsc->client, &fsid) < 0)
6142 		return;
6143 	epoch = ceph_decode_32(&p);
6144 	maplen = ceph_decode_32(&p);
6145 	doutc(cl, "epoch %u len %d\n", epoch, (int)maplen);
6146 
6147 	/* do we need it? */
6148 	mutex_lock(&mdsc->mutex);
6149 	if (mdsc->mdsmap && epoch <= mdsc->mdsmap->m_epoch) {
6150 		doutc(cl, "epoch %u <= our %u\n", epoch, mdsc->mdsmap->m_epoch);
6151 		mutex_unlock(&mdsc->mutex);
6152 		return;
6153 	}
6154 
6155 	newmap = ceph_mdsmap_decode(mdsc, &p, end, ceph_msgr2(mdsc->fsc->client));
6156 	if (IS_ERR(newmap)) {
6157 		err = PTR_ERR(newmap);
6158 		goto bad_unlock;
6159 	}
6160 
6161 	/* swap into place */
6162 	if (mdsc->mdsmap) {
6163 		oldmap = mdsc->mdsmap;
6164 		mdsc->mdsmap = newmap;
6165 		check_new_map(mdsc, newmap, oldmap);
6166 		ceph_mdsmap_destroy(oldmap);
6167 	} else {
6168 		mdsc->mdsmap = newmap;  /* first mds map */
6169 	}
6170 	mdsc->fsc->max_file_size = min((loff_t)mdsc->mdsmap->m_max_file_size,
6171 					MAX_LFS_FILESIZE);
6172 
6173 	__wake_requests(mdsc, &mdsc->waiting_for_map);
6174 	ceph_monc_got_map(&mdsc->fsc->client->monc, CEPH_SUB_MDSMAP,
6175 			  mdsc->mdsmap->m_epoch);
6176 
6177 	mutex_unlock(&mdsc->mutex);
6178 	schedule_delayed(mdsc, 0);
6179 	return;
6180 
6181 bad_unlock:
6182 	mutex_unlock(&mdsc->mutex);
6183 bad:
6184 	pr_err_client(cl, "error decoding mdsmap %d. Shutting down mount.\n",
6185 		      err);
6186 	ceph_umount_begin(mdsc->fsc->sb);
6187 	ceph_msg_dump(msg);
6188 	return;
6189 }
6190 
6191 static struct ceph_connection *mds_get_con(struct ceph_connection *con)
6192 {
6193 	struct ceph_mds_session *s = con->private;
6194 
6195 	if (ceph_get_mds_session(s))
6196 		return con;
6197 	return NULL;
6198 }
6199 
6200 static void mds_put_con(struct ceph_connection *con)
6201 {
6202 	struct ceph_mds_session *s = con->private;
6203 
6204 	ceph_put_mds_session(s);
6205 }
6206 
6207 /*
6208  * if the client is unresponsive for long enough, the mds will kill
6209  * the session entirely.
6210  */
6211 static void mds_peer_reset(struct ceph_connection *con)
6212 {
6213 	struct ceph_mds_session *s = con->private;
6214 	struct ceph_mds_client *mdsc = s->s_mdsc;
6215 
6216 	pr_warn_client(mdsc->fsc->client, "mds%d closed our session\n",
6217 		       s->s_mds);
6218 	if (READ_ONCE(mdsc->fsc->mount_state) != CEPH_MOUNT_FENCE_IO &&
6219 	    ceph_mdsmap_get_state(mdsc->mdsmap, s->s_mds) >= CEPH_MDS_STATE_RECONNECT)
6220 		send_mds_reconnect(mdsc, s);
6221 }
6222 
6223 static void mds_dispatch(struct ceph_connection *con, struct ceph_msg *msg)
6224 {
6225 	struct ceph_mds_session *s = con->private;
6226 	struct ceph_mds_client *mdsc = s->s_mdsc;
6227 	struct ceph_client *cl = mdsc->fsc->client;
6228 	int type = le16_to_cpu(msg->hdr.type);
6229 
6230 	mutex_lock(&mdsc->mutex);
6231 	if (__verify_registered_session(mdsc, s) < 0) {
6232 		mutex_unlock(&mdsc->mutex);
6233 		goto out;
6234 	}
6235 	mutex_unlock(&mdsc->mutex);
6236 
6237 	switch (type) {
6238 	case CEPH_MSG_MDS_MAP:
6239 		ceph_mdsc_handle_mdsmap(mdsc, msg);
6240 		break;
6241 	case CEPH_MSG_FS_MAP_USER:
6242 		ceph_mdsc_handle_fsmap(mdsc, msg);
6243 		break;
6244 	case CEPH_MSG_CLIENT_SESSION:
6245 		handle_session(s, msg);
6246 		break;
6247 	case CEPH_MSG_CLIENT_REPLY:
6248 		handle_reply(s, msg);
6249 		break;
6250 	case CEPH_MSG_CLIENT_REQUEST_FORWARD:
6251 		handle_forward(mdsc, s, msg);
6252 		break;
6253 	case CEPH_MSG_CLIENT_CAPS:
6254 		ceph_handle_caps(s, msg);
6255 		break;
6256 	case CEPH_MSG_CLIENT_SNAP:
6257 		ceph_handle_snap(mdsc, s, msg);
6258 		break;
6259 	case CEPH_MSG_CLIENT_LEASE:
6260 		handle_lease(mdsc, s, msg);
6261 		break;
6262 	case CEPH_MSG_CLIENT_QUOTA:
6263 		ceph_handle_quota(mdsc, s, msg);
6264 		break;
6265 
6266 	default:
6267 		pr_err_client(cl, "received unknown message type %d %s\n",
6268 			      type, ceph_msg_type_name(type));
6269 	}
6270 out:
6271 	ceph_msg_put(msg);
6272 }
6273 
6274 /*
6275  * authentication
6276  */
6277 
6278 /*
6279  * Note: returned pointer is the address of a structure that's
6280  * managed separately.  Caller must *not* attempt to free it.
6281  */
6282 static struct ceph_auth_handshake *
6283 mds_get_authorizer(struct ceph_connection *con, int *proto, int force_new)
6284 {
6285 	struct ceph_mds_session *s = con->private;
6286 	struct ceph_mds_client *mdsc = s->s_mdsc;
6287 	struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
6288 	struct ceph_auth_handshake *auth = &s->s_auth;
6289 	int ret;
6290 
6291 	ret = __ceph_auth_get_authorizer(ac, auth, CEPH_ENTITY_TYPE_MDS,
6292 					 force_new, proto, NULL, NULL);
6293 	if (ret)
6294 		return ERR_PTR(ret);
6295 
6296 	return auth;
6297 }
6298 
6299 static int mds_add_authorizer_challenge(struct ceph_connection *con,
6300 				    void *challenge_buf, int challenge_buf_len)
6301 {
6302 	struct ceph_mds_session *s = con->private;
6303 	struct ceph_mds_client *mdsc = s->s_mdsc;
6304 	struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
6305 
6306 	return ceph_auth_add_authorizer_challenge(ac, s->s_auth.authorizer,
6307 					    challenge_buf, challenge_buf_len);
6308 }
6309 
6310 static int mds_verify_authorizer_reply(struct ceph_connection *con)
6311 {
6312 	struct ceph_mds_session *s = con->private;
6313 	struct ceph_mds_client *mdsc = s->s_mdsc;
6314 	struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
6315 	struct ceph_auth_handshake *auth = &s->s_auth;
6316 
6317 	return ceph_auth_verify_authorizer_reply(ac, auth->authorizer,
6318 		auth->authorizer_reply_buf, auth->authorizer_reply_buf_len,
6319 		NULL, NULL, NULL, NULL);
6320 }
6321 
6322 static int mds_invalidate_authorizer(struct ceph_connection *con)
6323 {
6324 	struct ceph_mds_session *s = con->private;
6325 	struct ceph_mds_client *mdsc = s->s_mdsc;
6326 	struct ceph_auth_client *ac = mdsc->fsc->client->monc.auth;
6327 
6328 	ceph_auth_invalidate_authorizer(ac, CEPH_ENTITY_TYPE_MDS);
6329 
6330 	return ceph_monc_validate_auth(&mdsc->fsc->client->monc);
6331 }
6332 
6333 static int mds_get_auth_request(struct ceph_connection *con,
6334 				void *buf, int *buf_len,
6335 				void **authorizer, int *authorizer_len)
6336 {
6337 	struct ceph_mds_session *s = con->private;
6338 	struct ceph_auth_client *ac = s->s_mdsc->fsc->client->monc.auth;
6339 	struct ceph_auth_handshake *auth = &s->s_auth;
6340 	int ret;
6341 
6342 	ret = ceph_auth_get_authorizer(ac, auth, CEPH_ENTITY_TYPE_MDS,
6343 				       buf, buf_len);
6344 	if (ret)
6345 		return ret;
6346 
6347 	*authorizer = auth->authorizer_buf;
6348 	*authorizer_len = auth->authorizer_buf_len;
6349 	return 0;
6350 }
6351 
6352 static int mds_handle_auth_reply_more(struct ceph_connection *con,
6353 				      void *reply, int reply_len,
6354 				      void *buf, int *buf_len,
6355 				      void **authorizer, int *authorizer_len)
6356 {
6357 	struct ceph_mds_session *s = con->private;
6358 	struct ceph_auth_client *ac = s->s_mdsc->fsc->client->monc.auth;
6359 	struct ceph_auth_handshake *auth = &s->s_auth;
6360 	int ret;
6361 
6362 	ret = ceph_auth_handle_svc_reply_more(ac, auth, reply, reply_len,
6363 					      buf, buf_len);
6364 	if (ret)
6365 		return ret;
6366 
6367 	*authorizer = auth->authorizer_buf;
6368 	*authorizer_len = auth->authorizer_buf_len;
6369 	return 0;
6370 }
6371 
6372 static int mds_handle_auth_done(struct ceph_connection *con,
6373 				u64 global_id, void *reply, int reply_len,
6374 				u8 *session_key, int *session_key_len,
6375 				u8 *con_secret, int *con_secret_len)
6376 {
6377 	struct ceph_mds_session *s = con->private;
6378 	struct ceph_auth_client *ac = s->s_mdsc->fsc->client->monc.auth;
6379 	struct ceph_auth_handshake *auth = &s->s_auth;
6380 
6381 	return ceph_auth_handle_svc_reply_done(ac, auth, reply, reply_len,
6382 					       session_key, session_key_len,
6383 					       con_secret, con_secret_len);
6384 }
6385 
6386 static int mds_handle_auth_bad_method(struct ceph_connection *con,
6387 				      int used_proto, int result,
6388 				      const int *allowed_protos, int proto_cnt,
6389 				      const int *allowed_modes, int mode_cnt)
6390 {
6391 	struct ceph_mds_session *s = con->private;
6392 	struct ceph_mon_client *monc = &s->s_mdsc->fsc->client->monc;
6393 	int ret;
6394 
6395 	if (ceph_auth_handle_bad_authorizer(monc->auth, CEPH_ENTITY_TYPE_MDS,
6396 					    used_proto, result,
6397 					    allowed_protos, proto_cnt,
6398 					    allowed_modes, mode_cnt)) {
6399 		ret = ceph_monc_validate_auth(monc);
6400 		if (ret)
6401 			return ret;
6402 	}
6403 
6404 	return -EACCES;
6405 }
6406 
6407 static struct ceph_msg *mds_alloc_msg(struct ceph_connection *con,
6408 				struct ceph_msg_header *hdr, int *skip)
6409 {
6410 	struct ceph_msg *msg;
6411 	int type = (int) le16_to_cpu(hdr->type);
6412 	int front_len = (int) le32_to_cpu(hdr->front_len);
6413 
6414 	if (con->in_msg)
6415 		return con->in_msg;
6416 
6417 	*skip = 0;
6418 	msg = ceph_msg_new(type, front_len, GFP_NOFS, false);
6419 	if (!msg) {
6420 		pr_err("unable to allocate msg type %d len %d\n",
6421 		       type, front_len);
6422 		return NULL;
6423 	}
6424 
6425 	return msg;
6426 }
6427 
6428 static int mds_sign_message(struct ceph_msg *msg)
6429 {
6430        struct ceph_mds_session *s = msg->con->private;
6431        struct ceph_auth_handshake *auth = &s->s_auth;
6432 
6433        return ceph_auth_sign_message(auth, msg);
6434 }
6435 
6436 static int mds_check_message_signature(struct ceph_msg *msg)
6437 {
6438        struct ceph_mds_session *s = msg->con->private;
6439        struct ceph_auth_handshake *auth = &s->s_auth;
6440 
6441        return ceph_auth_check_message_signature(auth, msg);
6442 }
6443 
6444 static const struct ceph_connection_operations mds_con_ops = {
6445 	.get = mds_get_con,
6446 	.put = mds_put_con,
6447 	.alloc_msg = mds_alloc_msg,
6448 	.dispatch = mds_dispatch,
6449 	.peer_reset = mds_peer_reset,
6450 	.get_authorizer = mds_get_authorizer,
6451 	.add_authorizer_challenge = mds_add_authorizer_challenge,
6452 	.verify_authorizer_reply = mds_verify_authorizer_reply,
6453 	.invalidate_authorizer = mds_invalidate_authorizer,
6454 	.sign_message = mds_sign_message,
6455 	.check_message_signature = mds_check_message_signature,
6456 	.get_auth_request = mds_get_auth_request,
6457 	.handle_auth_reply_more = mds_handle_auth_reply_more,
6458 	.handle_auth_done = mds_handle_auth_done,
6459 	.handle_auth_bad_method = mds_handle_auth_bad_method,
6460 };
6461 
6462 /* eof */
6463