dcache.c 13.8 KB
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506
/* -*- mode: c; c-basic-offset: 8; -*-
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * dcache.c
 *
 * dentry cache handling code
 *
 * Copyright (C) 2002, 2004 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

#include <linux/fs.h>
#include <linux/types.h>
#include <linux/slab.h>
#include <linux/namei.h>

#define MLOG_MASK_PREFIX ML_DCACHE
#include <cluster/masklog.h>

#include "ocfs2.h"

#include "alloc.h"
#include "dcache.h"
#include "dlmglue.h"
#include "file.h"
#include "inode.h"
#include "super.h"


static int ocfs2_dentry_revalidate(struct dentry *dentry,
				   struct nameidata *nd)
{
	struct inode *inode = dentry->d_inode;
	int ret = 0;    /* if all else fails, just return false */
	struct ocfs2_super *osb = OCFS2_SB(dentry->d_sb);

	mlog_entry("(0x%p, '%.*s')\n", dentry,
		   dentry->d_name.len, dentry->d_name.name);

	/* Never trust a negative dentry - force a new lookup. */
	if (inode == NULL) {
		mlog(0, "negative dentry: %.*s\n", dentry->d_name.len,
		     dentry->d_name.name);
		goto bail;
	}

	BUG_ON(!osb);

	if (inode == osb->root_inode || is_bad_inode(inode))
		goto bail;

	spin_lock(&OCFS2_I(inode)->ip_lock);
	/* did we or someone else delete this inode? */
	if (OCFS2_I(inode)->ip_flags & OCFS2_INODE_DELETED) {
		spin_unlock(&OCFS2_I(inode)->ip_lock);
		mlog(0, "inode (%llu) deleted, returning false\n",
		     (unsigned long long)OCFS2_I(inode)->ip_blkno);
		goto bail;
	}
	spin_unlock(&OCFS2_I(inode)->ip_lock);

	/*
	 * We don't need a cluster lock to test this because once an
	 * inode nlink hits zero, it never goes back.
	 */
	if (inode->i_nlink == 0) {
		mlog(0, "Inode %llu orphaned, returning false "
		     "dir = %d\n",
		     (unsigned long long)OCFS2_I(inode)->ip_blkno,
		     S_ISDIR(inode->i_mode));
		goto bail;
	}

	/*
	 * If the last lookup failed to create dentry lock, let us
	 * redo it.
	 */
	if (!dentry->d_fsdata) {
		mlog(0, "Inode %llu doesn't have dentry lock, "
		     "returning false\n",
		     (unsigned long long)OCFS2_I(inode)->ip_blkno);
		goto bail;
	}

	ret = 1;

bail:
	mlog_exit(ret);

	return ret;
}

static int ocfs2_match_dentry(struct dentry *dentry,
			      u64 parent_blkno,
			      int skip_unhashed)
{
	struct inode *parent;

	/*
	 * ocfs2_lookup() does a d_splice_alias() _before_ attaching
	 * to the lock data, so we skip those here, otherwise
	 * ocfs2_dentry_attach_lock() will get its original dentry
	 * back.
	 */
	if (!dentry->d_fsdata)
		return 0;

	if (!dentry->d_parent)
		return 0;

	if (skip_unhashed && d_unhashed(dentry))
		return 0;

	parent = dentry->d_parent->d_inode;
	/* Negative parent dentry? */
	if (!parent)
		return 0;

	/* Name is in a different directory. */
	if (OCFS2_I(parent)->ip_blkno != parent_blkno)
		return 0;

	return 1;
}

/*
 * Walk the inode alias list, and find a dentry which has a given
 * parent. ocfs2_dentry_attach_lock() wants to find _any_ alias as it
 * is looking for a dentry_lock reference. The downconvert thread is
 * looking to unhash aliases, so we allow it to skip any that already
 * have that property.
 */
struct dentry *ocfs2_find_local_alias(struct inode *inode,
				      u64 parent_blkno,
				      int skip_unhashed)
{
	struct list_head *p;
	struct dentry *dentry = NULL;

	spin_lock(&dcache_lock);

	list_for_each(p, &inode->i_dentry) {
		dentry = list_entry(p, struct dentry, d_alias);

		if (ocfs2_match_dentry(dentry, parent_blkno, skip_unhashed)) {
			mlog(0, "dentry found: %.*s\n",
			     dentry->d_name.len, dentry->d_name.name);

			dget_locked(dentry);
			break;
		}

		dentry = NULL;
	}

	spin_unlock(&dcache_lock);

	return dentry;
}

DEFINE_SPINLOCK(dentry_attach_lock);

/*
 * Attach this dentry to a cluster lock.
 *
 * Dentry locks cover all links in a given directory to a particular
 * inode. We do this so that ocfs2 can build a lock name which all
 * nodes in the cluster can agree on at all times. Shoving full names
 * in the cluster lock won't work due to size restrictions. Covering
 * links inside of a directory is a good compromise because it still
 * allows us to use the parent directory lock to synchronize
 * operations.
 *
 * Call this function with the parent dir semaphore and the parent dir
 * cluster lock held.
 *
 * The dir semaphore will protect us from having to worry about
 * concurrent processes on our node trying to attach a lock at the
 * same time.
 *
 * The dir cluster lock (held at either PR or EX mode) protects us
 * from unlink and rename on other nodes.
 *
 * A dput() can happen asynchronously due to pruning, so we cover
 * attaching and detaching the dentry lock with a
 * dentry_attach_lock.
 *
 * A node which has done lookup on a name retains a protected read
 * lock until final dput. If the user requests and unlink or rename,
 * the protected read is upgraded to an exclusive lock. Other nodes
 * who have seen the dentry will then be informed that they need to
 * downgrade their lock, which will involve d_delete on the
 * dentry. This happens in ocfs2_dentry_convert_worker().
 */
int ocfs2_dentry_attach_lock(struct dentry *dentry,
			     struct inode *inode,
			     u64 parent_blkno)
{
	int ret;
	struct dentry *alias;
	struct ocfs2_dentry_lock *dl = dentry->d_fsdata;

	mlog(0, "Attach \"%.*s\", parent %llu, fsdata: %p\n",
	     dentry->d_name.len, dentry->d_name.name,
	     (unsigned long long)parent_blkno, dl);

	/*
	 * Negative dentry. We ignore these for now.
	 *
	 * XXX: Could we can improve ocfs2_dentry_revalidate() by
	 * tracking these?
	 */
	if (!inode)
		return 0;

	if (dl) {
		mlog_bug_on_msg(dl->dl_parent_blkno != parent_blkno,
				" \"%.*s\": old parent: %llu, new: %llu\n",
				dentry->d_name.len, dentry->d_name.name,
				(unsigned long long)parent_blkno,
				(unsigned long long)dl->dl_parent_blkno);
		return 0;
	}

	alias = ocfs2_find_local_alias(inode, parent_blkno, 0);
	if (alias) {
		/*
		 * Great, an alias exists, which means we must have a
		 * dentry lock already. We can just grab the lock off
		 * the alias and add it to the list.
		 *
		 * We're depending here on the fact that this dentry
		 * was found and exists in the dcache and so must have
		 * a reference to the dentry_lock because we can't
		 * race creates. Final dput() cannot happen on it
		 * since we have it pinned, so our reference is safe.
		 */
		dl = alias->d_fsdata;
		mlog_bug_on_msg(!dl, "parent %llu, ino %llu\n",
				(unsigned long long)parent_blkno,
				(unsigned long long)OCFS2_I(inode)->ip_blkno);

		mlog_bug_on_msg(dl->dl_parent_blkno != parent_blkno,
				" \"%.*s\": old parent: %llu, new: %llu\n",
				dentry->d_name.len, dentry->d_name.name,
				(unsigned long long)parent_blkno,
				(unsigned long long)dl->dl_parent_blkno);

		mlog(0, "Found: %s\n", dl->dl_lockres.l_name);

		goto out_attach;
	}

	/*
	 * There are no other aliases
	 */
	dl = kmalloc(sizeof(*dl), GFP_NOFS);
	if (!dl) {
		ret = -ENOMEM;
		mlog_errno(ret);
		return ret;
	}

	dl->dl_count = 0;
	/*
	 * Does this have to happen below, for all attaches, in case
	 * the struct inode gets blown away by the downconvert thread?
	 */
	dl->dl_inode = igrab(inode);
	dl->dl_parent_blkno = parent_blkno;
	ocfs2_dentry_lock_res_init(dl, parent_blkno, inode);

out_attach:
	spin_lock(&dentry_attach_lock);
	dentry->d_fsdata = dl;
	dl->dl_count++;
	spin_unlock(&dentry_attach_lock);

	/*
	 * This actually gets us our PRMODE level lock. From now on,
	 * we'll have a notification if one of these names is
	 * destroyed on another node.
	 */
	ret = ocfs2_dentry_lock(dentry, 0);
	if (!ret)
		ocfs2_dentry_unlock(dentry, 0);
	else
		mlog_errno(ret);

	/*
	 * In case of error, manually free the allocation and do the iput().
	 * We need to do this because error here means no d_instantiate(),
	 * which means iput() will not be called during dput(dentry).
	 */
	if (ret < 0 && !alias) {
		ocfs2_lock_res_free(&dl->dl_lockres);
		BUG_ON(dl->dl_count != 1);
		spin_lock(&dentry_attach_lock);
		dentry->d_fsdata = NULL;
		spin_unlock(&dentry_attach_lock);
		kfree(dl);
		iput(inode);
	}

	dput(alias);

	return ret;
}

DEFINE_SPINLOCK(dentry_list_lock);

/* We limit the number of dentry locks to drop in one go. We have
 * this limit so that we don't starve other users of ocfs2_wq. */
#define DL_INODE_DROP_COUNT 64

/* Drop inode references from dentry locks */
static void __ocfs2_drop_dl_inodes(struct ocfs2_super *osb, int drop_count)
{
	struct ocfs2_dentry_lock *dl;

	spin_lock(&dentry_list_lock);
	while (osb->dentry_lock_list && (drop_count < 0 || drop_count--)) {
		dl = osb->dentry_lock_list;
		osb->dentry_lock_list = dl->dl_next;
		spin_unlock(&dentry_list_lock);
		iput(dl->dl_inode);
		kfree(dl);
		spin_lock(&dentry_list_lock);
	}
	spin_unlock(&dentry_list_lock);
}

void ocfs2_drop_dl_inodes(struct work_struct *work)
{
	struct ocfs2_super *osb = container_of(work, struct ocfs2_super,
					       dentry_lock_work);

	__ocfs2_drop_dl_inodes(osb, DL_INODE_DROP_COUNT);
	/*
	 * Don't queue dropping if umount is in progress. We flush the
	 * list in ocfs2_dismount_volume
	 */
	spin_lock(&dentry_list_lock);
	if (osb->dentry_lock_list &&
	    !ocfs2_test_osb_flag(osb, OCFS2_OSB_DROP_DENTRY_LOCK_IMMED))
		queue_work(ocfs2_wq, &osb->dentry_lock_work);
	spin_unlock(&dentry_list_lock);
}

/* Flush the whole work queue */
void ocfs2_drop_all_dl_inodes(struct ocfs2_super *osb)
{
	__ocfs2_drop_dl_inodes(osb, -1);
}

/*
 * ocfs2_dentry_iput() and friends.
 *
 * At this point, our particular dentry is detached from the inodes
 * alias list, so there's no way that the locking code can find it.
 *
 * The interesting stuff happens when we determine that our lock needs
 * to go away because this is the last subdir alias in the
 * system. This function needs to handle a couple things:
 *
 * 1) Synchronizing lock shutdown with the downconvert threads. This
 *    is already handled for us via the lockres release drop function
 *    called in ocfs2_release_dentry_lock()
 *
 * 2) A race may occur when we're doing our lock shutdown and
 *    another process wants to create a new dentry lock. Right now we
 *    let them race, which means that for a very short while, this
 *    node might have two locks on a lock resource. This should be a
 *    problem though because one of them is in the process of being
 *    thrown out.
 */
static void ocfs2_drop_dentry_lock(struct ocfs2_super *osb,
				   struct ocfs2_dentry_lock *dl)
{
	ocfs2_simple_drop_lockres(osb, &dl->dl_lockres);
	ocfs2_lock_res_free(&dl->dl_lockres);

	/* We leave dropping of inode reference to ocfs2_wq as that can
	 * possibly lead to inode deletion which gets tricky */
	spin_lock(&dentry_list_lock);
	if (!osb->dentry_lock_list &&
	    !ocfs2_test_osb_flag(osb, OCFS2_OSB_DROP_DENTRY_LOCK_IMMED))
		queue_work(ocfs2_wq, &osb->dentry_lock_work);
	dl->dl_next = osb->dentry_lock_list;
	osb->dentry_lock_list = dl;
	spin_unlock(&dentry_list_lock);
}

void ocfs2_dentry_lock_put(struct ocfs2_super *osb,
			   struct ocfs2_dentry_lock *dl)
{
	int unlock;

	BUG_ON(dl->dl_count == 0);

	spin_lock(&dentry_attach_lock);
	dl->dl_count--;
	unlock = !dl->dl_count;
	spin_unlock(&dentry_attach_lock);

	if (unlock)
		ocfs2_drop_dentry_lock(osb, dl);
}

static void ocfs2_dentry_iput(struct dentry *dentry, struct inode *inode)
{
	struct ocfs2_dentry_lock *dl = dentry->d_fsdata;

	if (!dl) {
		/*
		 * No dentry lock is ok if we're disconnected or
		 * unhashed.
		 */
		if (!(dentry->d_flags & DCACHE_DISCONNECTED) &&
		    !d_unhashed(dentry)) {
			unsigned long long ino = 0ULL;
			if (inode)
				ino = (unsigned long long)OCFS2_I(inode)->ip_blkno;
			mlog(ML_ERROR, "Dentry is missing cluster lock. "
			     "inode: %llu, d_flags: 0x%x, d_name: %.*s\n",
			     ino, dentry->d_flags, dentry->d_name.len,
			     dentry->d_name.name);
		}

		goto out;
	}

	mlog_bug_on_msg(dl->dl_count == 0, "dentry: %.*s, count: %u\n",
			dentry->d_name.len, dentry->d_name.name,
			dl->dl_count);

	ocfs2_dentry_lock_put(OCFS2_SB(dentry->d_sb), dl);

out:
	iput(inode);
}

/*
 * d_move(), but keep the locks in sync.
 *
 * When we are done, "dentry" will have the parent dir and name of
 * "target", which will be thrown away.
 *
 * We manually update the lock of "dentry" if need be.
 *
 * "target" doesn't have it's dentry lock touched - we allow the later
 * dput() to handle this for us.
 *
 * This is called during ocfs2_rename(), while holding parent
 * directory locks. The dentries have already been deleted on other
 * nodes via ocfs2_remote_dentry_delete().
 *
 * Normally, the VFS handles the d_move() for the file system, after
 * the ->rename() callback. OCFS2 wants to handle this internally, so
 * the new lock can be created atomically with respect to the cluster.
 */
void ocfs2_dentry_move(struct dentry *dentry, struct dentry *target,
		       struct inode *old_dir, struct inode *new_dir)
{
	int ret;
	struct ocfs2_super *osb = OCFS2_SB(old_dir->i_sb);
	struct inode *inode = dentry->d_inode;

	/*
	 * Move within the same directory, so the actual lock info won't
	 * change.
	 *
	 * XXX: Is there any advantage to dropping the lock here?
	 */
	if (old_dir == new_dir)
		goto out_move;

	ocfs2_dentry_lock_put(osb, dentry->d_fsdata);

	dentry->d_fsdata = NULL;
	ret = ocfs2_dentry_attach_lock(dentry, inode, OCFS2_I(new_dir)->ip_blkno);
	if (ret)
		mlog_errno(ret);

out_move:
	d_move(dentry, target);
}

const struct dentry_operations ocfs2_dentry_ops = {
	.d_revalidate		= ocfs2_dentry_revalidate,
	.d_iput			= ocfs2_dentry_iput,
};