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  45 * are no longer valid.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