refcounttree.c 112.3 KB
Newer Older
T
Tao Ma 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
/* -*- mode: c; c-basic-offset: 8; -*-
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * refcounttree.c
 *
 * Copyright (C) 2009 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 version 2 as published by the Free Software Foundation.
 *
 * 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.
 */

18
#include <linux/sort.h>
T
Tao Ma 已提交
19 20 21 22 23 24 25 26 27 28
#include <cluster/masklog.h>
#include "ocfs2.h"
#include "inode.h"
#include "alloc.h"
#include "suballoc.h"
#include "journal.h"
#include "uptodate.h"
#include "super.h"
#include "buffer_head_io.h"
#include "blockcheck.h"
29
#include "refcounttree.h"
T
Tao Ma 已提交
30
#include "sysfile.h"
31
#include "dlmglue.h"
32
#include "extent_map.h"
T
Tao Ma 已提交
33
#include "aops.h"
T
Tao Ma 已提交
34
#include "xattr.h"
T
Tao Ma 已提交
35
#include "namei.h"
T
Tao Ma 已提交
36
#include "ocfs2_trace.h"
T
Tao Ma 已提交
37 38 39 40 41 42 43

#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
#include <linux/writeback.h>
#include <linux/pagevec.h>
#include <linux/swap.h>
T
Tao Ma 已提交
44 45 46 47 48
#include <linux/security.h>
#include <linux/fsnotify.h>
#include <linux/quotaops.h>
#include <linux/namei.h>
#include <linux/mount.h>
49
#include <linux/posix_acl.h>
T
Tao Ma 已提交
50 51 52 53 54

struct ocfs2_cow_context {
	struct inode *inode;
	u32 cow_start;
	u32 cow_len;
55 56
	struct ocfs2_extent_tree data_et;
	struct ocfs2_refcount_tree *ref_tree;
T
Tao Ma 已提交
57 58 59 60
	struct buffer_head *ref_root_bh;
	struct ocfs2_alloc_context *meta_ac;
	struct ocfs2_alloc_context *data_ac;
	struct ocfs2_cached_dealloc_ctxt dealloc;
T
Tao Ma 已提交
61 62 63
	void *cow_object;
	struct ocfs2_post_refcount *post_refcount;
	int extra_credits;
64 65 66 67 68
	int (*get_clusters)(struct ocfs2_cow_context *context,
			    u32 v_cluster, u32 *p_cluster,
			    u32 *num_clusters,
			    unsigned int *extent_flags);
	int (*cow_duplicate_clusters)(handle_t *handle,
69
				      struct inode *inode,
70 71
				      u32 cpos, u32 old_cluster,
				      u32 new_cluster, u32 new_len);
T
Tao Ma 已提交
72
};
73 74 75 76 77 78

static inline struct ocfs2_refcount_tree *
cache_info_to_refcount(struct ocfs2_caching_info *ci)
{
	return container_of(ci, struct ocfs2_refcount_tree, rf_ci);
}
T
Tao Ma 已提交
79 80 81 82 83 84 85 86

static int ocfs2_validate_refcount_block(struct super_block *sb,
					 struct buffer_head *bh)
{
	int rc;
	struct ocfs2_refcount_block *rb =
		(struct ocfs2_refcount_block *)bh->b_data;

T
Tao Ma 已提交
87
	trace_ocfs2_validate_refcount_block((unsigned long long)bh->b_blocknr);
T
Tao Ma 已提交
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

	BUG_ON(!buffer_uptodate(bh));

	/*
	 * If the ecc fails, we return the error but otherwise
	 * leave the filesystem running.  We know any error is
	 * local to this block.
	 */
	rc = ocfs2_validate_meta_ecc(sb, bh->b_data, &rb->rf_check);
	if (rc) {
		mlog(ML_ERROR, "Checksum failed for refcount block %llu\n",
		     (unsigned long long)bh->b_blocknr);
		return rc;
	}


	if (!OCFS2_IS_VALID_REFCOUNT_BLOCK(rb)) {
		ocfs2_error(sb,
			    "Refcount block #%llu has bad signature %.*s",
			    (unsigned long long)bh->b_blocknr, 7,
			    rb->rf_signature);
		return -EINVAL;
	}

	if (le64_to_cpu(rb->rf_blkno) != bh->b_blocknr) {
		ocfs2_error(sb,
			    "Refcount block #%llu has an invalid rf_blkno "
			    "of %llu",
			    (unsigned long long)bh->b_blocknr,
			    (unsigned long long)le64_to_cpu(rb->rf_blkno));
		return -EINVAL;
	}

	if (le32_to_cpu(rb->rf_fs_generation) != OCFS2_SB(sb)->fs_generation) {
		ocfs2_error(sb,
			    "Refcount block #%llu has an invalid "
			    "rf_fs_generation of #%u",
			    (unsigned long long)bh->b_blocknr,
			    le32_to_cpu(rb->rf_fs_generation));
		return -EINVAL;
	}

	return 0;
}

static int ocfs2_read_refcount_block(struct ocfs2_caching_info *ci,
				     u64 rb_blkno,
				     struct buffer_head **bh)
{
	int rc;
	struct buffer_head *tmp = *bh;

	rc = ocfs2_read_block(ci, rb_blkno, &tmp,
			      ocfs2_validate_refcount_block);

	/* If ocfs2_read_block() got us a new bh, pass it up. */
	if (!rc && !*bh)
		*bh = tmp;

	return rc;
}
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

static u64 ocfs2_refcount_cache_owner(struct ocfs2_caching_info *ci)
{
	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);

	return rf->rf_blkno;
}

static struct super_block *
ocfs2_refcount_cache_get_super(struct ocfs2_caching_info *ci)
{
	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);

	return rf->rf_sb;
}

static void ocfs2_refcount_cache_lock(struct ocfs2_caching_info *ci)
{
	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);

	spin_lock(&rf->rf_lock);
}

static void ocfs2_refcount_cache_unlock(struct ocfs2_caching_info *ci)
{
	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);

	spin_unlock(&rf->rf_lock);
}

static void ocfs2_refcount_cache_io_lock(struct ocfs2_caching_info *ci)
{
	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);

	mutex_lock(&rf->rf_io_mutex);
}

static void ocfs2_refcount_cache_io_unlock(struct ocfs2_caching_info *ci)
{
	struct ocfs2_refcount_tree *rf = cache_info_to_refcount(ci);

	mutex_unlock(&rf->rf_io_mutex);
}

static const struct ocfs2_caching_operations ocfs2_refcount_caching_ops = {
	.co_owner		= ocfs2_refcount_cache_owner,
	.co_get_super		= ocfs2_refcount_cache_get_super,
	.co_cache_lock		= ocfs2_refcount_cache_lock,
	.co_cache_unlock	= ocfs2_refcount_cache_unlock,
	.co_io_lock		= ocfs2_refcount_cache_io_lock,
	.co_io_unlock		= ocfs2_refcount_cache_io_unlock,
};
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

static struct ocfs2_refcount_tree *
ocfs2_find_refcount_tree(struct ocfs2_super *osb, u64 blkno)
{
	struct rb_node *n = osb->osb_rf_lock_tree.rb_node;
	struct ocfs2_refcount_tree *tree = NULL;

	while (n) {
		tree = rb_entry(n, struct ocfs2_refcount_tree, rf_node);

		if (blkno < tree->rf_blkno)
			n = n->rb_left;
		else if (blkno > tree->rf_blkno)
			n = n->rb_right;
		else
			return tree;
	}

	return NULL;
}

/* osb_lock is already locked. */
static void ocfs2_insert_refcount_tree(struct ocfs2_super *osb,
				       struct ocfs2_refcount_tree *new)
{
	u64 rf_blkno = new->rf_blkno;
	struct rb_node *parent = NULL;
	struct rb_node **p = &osb->osb_rf_lock_tree.rb_node;
	struct ocfs2_refcount_tree *tmp;

	while (*p) {
		parent = *p;

		tmp = rb_entry(parent, struct ocfs2_refcount_tree,
			       rf_node);

		if (rf_blkno < tmp->rf_blkno)
			p = &(*p)->rb_left;
		else if (rf_blkno > tmp->rf_blkno)
			p = &(*p)->rb_right;
		else {
			/* This should never happen! */
			mlog(ML_ERROR, "Duplicate refcount block %llu found!\n",
			     (unsigned long long)rf_blkno);
			BUG();
		}
	}

	rb_link_node(&new->rf_node, parent, p);
	rb_insert_color(&new->rf_node, &osb->osb_rf_lock_tree);
}

static void ocfs2_free_refcount_tree(struct ocfs2_refcount_tree *tree)
{
	ocfs2_metadata_cache_exit(&tree->rf_ci);
	ocfs2_simple_drop_lockres(OCFS2_SB(tree->rf_sb), &tree->rf_lockres);
	ocfs2_lock_res_free(&tree->rf_lockres);
	kfree(tree);
}

static inline void
ocfs2_erase_refcount_tree_from_list_no_lock(struct ocfs2_super *osb,
					struct ocfs2_refcount_tree *tree)
{
	rb_erase(&tree->rf_node, &osb->osb_rf_lock_tree);
	if (osb->osb_ref_tree_lru && osb->osb_ref_tree_lru == tree)
		osb->osb_ref_tree_lru = NULL;
}

static void ocfs2_erase_refcount_tree_from_list(struct ocfs2_super *osb,
					struct ocfs2_refcount_tree *tree)
{
	spin_lock(&osb->osb_lock);
	ocfs2_erase_refcount_tree_from_list_no_lock(osb, tree);
	spin_unlock(&osb->osb_lock);
}

T
Tao Ma 已提交
278
static void ocfs2_kref_remove_refcount_tree(struct kref *kref)
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
{
	struct ocfs2_refcount_tree *tree =
		container_of(kref, struct ocfs2_refcount_tree, rf_getcnt);

	ocfs2_free_refcount_tree(tree);
}

static inline void
ocfs2_refcount_tree_get(struct ocfs2_refcount_tree *tree)
{
	kref_get(&tree->rf_getcnt);
}

static inline void
ocfs2_refcount_tree_put(struct ocfs2_refcount_tree *tree)
{
	kref_put(&tree->rf_getcnt, ocfs2_kref_remove_refcount_tree);
}

static inline void ocfs2_init_refcount_tree_ci(struct ocfs2_refcount_tree *new,
					       struct super_block *sb)
{
	ocfs2_metadata_cache_init(&new->rf_ci, &ocfs2_refcount_caching_ops);
	mutex_init(&new->rf_io_mutex);
	new->rf_sb = sb;
	spin_lock_init(&new->rf_lock);
}

static inline void ocfs2_init_refcount_tree_lock(struct ocfs2_super *osb,
					struct ocfs2_refcount_tree *new,
					u64 rf_blkno, u32 generation)
{
	init_rwsem(&new->rf_sem);
	ocfs2_refcount_lock_res_init(&new->rf_lockres, osb,
				     rf_blkno, generation);
}

T
Tao Ma 已提交
316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331
static struct ocfs2_refcount_tree*
ocfs2_allocate_refcount_tree(struct ocfs2_super *osb, u64 rf_blkno)
{
	struct ocfs2_refcount_tree *new;

	new = kzalloc(sizeof(struct ocfs2_refcount_tree), GFP_NOFS);
	if (!new)
		return NULL;

	new->rf_blkno = rf_blkno;
	kref_init(&new->rf_getcnt);
	ocfs2_init_refcount_tree_ci(new, osb->sb);

	return new;
}

332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350
static int ocfs2_get_refcount_tree(struct ocfs2_super *osb, u64 rf_blkno,
				   struct ocfs2_refcount_tree **ret_tree)
{
	int ret = 0;
	struct ocfs2_refcount_tree *tree, *new = NULL;
	struct buffer_head *ref_root_bh = NULL;
	struct ocfs2_refcount_block *ref_rb;

	spin_lock(&osb->osb_lock);
	if (osb->osb_ref_tree_lru &&
	    osb->osb_ref_tree_lru->rf_blkno == rf_blkno)
		tree = osb->osb_ref_tree_lru;
	else
		tree = ocfs2_find_refcount_tree(osb, rf_blkno);
	if (tree)
		goto out;

	spin_unlock(&osb->osb_lock);

T
Tao Ma 已提交
351
	new = ocfs2_allocate_refcount_tree(osb, rf_blkno);
352 353
	if (!new) {
		ret = -ENOMEM;
T
Tao Ma 已提交
354
		mlog_errno(ret);
355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546
		return ret;
	}
	/*
	 * We need the generation to create the refcount tree lock and since
	 * it isn't changed during the tree modification, we are safe here to
	 * read without protection.
	 * We also have to purge the cache after we create the lock since the
	 * refcount block may have the stale data. It can only be trusted when
	 * we hold the refcount lock.
	 */
	ret = ocfs2_read_refcount_block(&new->rf_ci, rf_blkno, &ref_root_bh);
	if (ret) {
		mlog_errno(ret);
		ocfs2_metadata_cache_exit(&new->rf_ci);
		kfree(new);
		return ret;
	}

	ref_rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
	new->rf_generation = le32_to_cpu(ref_rb->rf_generation);
	ocfs2_init_refcount_tree_lock(osb, new, rf_blkno,
				      new->rf_generation);
	ocfs2_metadata_cache_purge(&new->rf_ci);

	spin_lock(&osb->osb_lock);
	tree = ocfs2_find_refcount_tree(osb, rf_blkno);
	if (tree)
		goto out;

	ocfs2_insert_refcount_tree(osb, new);

	tree = new;
	new = NULL;

out:
	*ret_tree = tree;

	osb->osb_ref_tree_lru = tree;

	spin_unlock(&osb->osb_lock);

	if (new)
		ocfs2_free_refcount_tree(new);

	brelse(ref_root_bh);
	return ret;
}

static int ocfs2_get_refcount_block(struct inode *inode, u64 *ref_blkno)
{
	int ret;
	struct buffer_head *di_bh = NULL;
	struct ocfs2_dinode *di;

	ret = ocfs2_read_inode_block(inode, &di_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	BUG_ON(!(OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));

	di = (struct ocfs2_dinode *)di_bh->b_data;
	*ref_blkno = le64_to_cpu(di->i_refcount_loc);
	brelse(di_bh);
out:
	return ret;
}

static int __ocfs2_lock_refcount_tree(struct ocfs2_super *osb,
				      struct ocfs2_refcount_tree *tree, int rw)
{
	int ret;

	ret = ocfs2_refcount_lock(tree, rw);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	if (rw)
		down_write(&tree->rf_sem);
	else
		down_read(&tree->rf_sem);

out:
	return ret;
}

/*
 * Lock the refcount tree pointed by ref_blkno and return the tree.
 * In most case, we lock the tree and read the refcount block.
 * So read it here if the caller really needs it.
 *
 * If the tree has been re-created by other node, it will free the
 * old one and re-create it.
 */
int ocfs2_lock_refcount_tree(struct ocfs2_super *osb,
			     u64 ref_blkno, int rw,
			     struct ocfs2_refcount_tree **ret_tree,
			     struct buffer_head **ref_bh)
{
	int ret, delete_tree = 0;
	struct ocfs2_refcount_tree *tree = NULL;
	struct buffer_head *ref_root_bh = NULL;
	struct ocfs2_refcount_block *rb;

again:
	ret = ocfs2_get_refcount_tree(osb, ref_blkno, &tree);
	if (ret) {
		mlog_errno(ret);
		return ret;
	}

	ocfs2_refcount_tree_get(tree);

	ret = __ocfs2_lock_refcount_tree(osb, tree, rw);
	if (ret) {
		mlog_errno(ret);
		ocfs2_refcount_tree_put(tree);
		goto out;
	}

	ret = ocfs2_read_refcount_block(&tree->rf_ci, tree->rf_blkno,
					&ref_root_bh);
	if (ret) {
		mlog_errno(ret);
		ocfs2_unlock_refcount_tree(osb, tree, rw);
		ocfs2_refcount_tree_put(tree);
		goto out;
	}

	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
	/*
	 * If the refcount block has been freed and re-created, we may need
	 * to recreate the refcount tree also.
	 *
	 * Here we just remove the tree from the rb-tree, and the last
	 * kref holder will unlock and delete this refcount_tree.
	 * Then we goto "again" and ocfs2_get_refcount_tree will create
	 * the new refcount tree for us.
	 */
	if (tree->rf_generation != le32_to_cpu(rb->rf_generation)) {
		if (!tree->rf_removed) {
			ocfs2_erase_refcount_tree_from_list(osb, tree);
			tree->rf_removed = 1;
			delete_tree = 1;
		}

		ocfs2_unlock_refcount_tree(osb, tree, rw);
		/*
		 * We get an extra reference when we create the refcount
		 * tree, so another put will destroy it.
		 */
		if (delete_tree)
			ocfs2_refcount_tree_put(tree);
		brelse(ref_root_bh);
		ref_root_bh = NULL;
		goto again;
	}

	*ret_tree = tree;
	if (ref_bh) {
		*ref_bh = ref_root_bh;
		ref_root_bh = NULL;
	}
out:
	brelse(ref_root_bh);
	return ret;
}

void ocfs2_unlock_refcount_tree(struct ocfs2_super *osb,
				struct ocfs2_refcount_tree *tree, int rw)
{
	if (rw)
		up_write(&tree->rf_sem);
	else
		up_read(&tree->rf_sem);

	ocfs2_refcount_unlock(tree, rw);
	ocfs2_refcount_tree_put(tree);
}

void ocfs2_purge_refcount_trees(struct ocfs2_super *osb)
{
	struct rb_node *node;
	struct ocfs2_refcount_tree *tree;
	struct rb_root *root = &osb->osb_rf_lock_tree;

	while ((node = rb_last(root)) != NULL) {
		tree = rb_entry(node, struct ocfs2_refcount_tree, rf_node);

T
Tao Ma 已提交
547 548
		trace_ocfs2_purge_refcount_trees(
				(unsigned long long) tree->rf_blkno);
549 550 551 552 553

		rb_erase(&tree->rf_node, root);
		ocfs2_free_refcount_tree(tree);
	}
}
T
Tao Ma 已提交
554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572

/*
 * Create a refcount tree for an inode.
 * We take for granted that the inode is already locked.
 */
static int ocfs2_create_refcount_tree(struct inode *inode,
				      struct buffer_head *di_bh)
{
	int ret;
	handle_t *handle = NULL;
	struct ocfs2_alloc_context *meta_ac = NULL;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_inode_info *oi = OCFS2_I(inode);
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	struct buffer_head *new_bh = NULL;
	struct ocfs2_refcount_block *rb;
	struct ocfs2_refcount_tree *new_tree = NULL, *tree = NULL;
	u16 suballoc_bit_start;
	u32 num_got;
573
	u64 suballoc_loc, first_blkno;
T
Tao Ma 已提交
574 575 576

	BUG_ON(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL);

T
Tao Ma 已提交
577 578
	trace_ocfs2_create_refcount_tree(
		(unsigned long long)OCFS2_I(inode)->ip_blkno);
T
Tao Ma 已提交
579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599

	ret = ocfs2_reserve_new_metadata_blocks(osb, 1, &meta_ac);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	handle = ocfs2_start_trans(osb, OCFS2_REFCOUNT_TREE_CREATE_CREDITS);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

600
	ret = ocfs2_claim_metadata(handle, meta_ac, 1, &suballoc_loc,
T
Tao Ma 已提交
601 602 603 604 605 606 607 608 609 610 611 612 613 614 615
				   &suballoc_bit_start, &num_got,
				   &first_blkno);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	new_tree = ocfs2_allocate_refcount_tree(osb, first_blkno);
	if (!new_tree) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out_commit;
	}

	new_bh = sb_getblk(inode->i_sb, first_blkno);
616 617 618 619 620
	if (!new_bh) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out_commit;
	}
T
Tao Ma 已提交
621 622 623 624 625 626 627 628 629 630 631 632 633
	ocfs2_set_new_buffer_uptodate(&new_tree->rf_ci, new_bh);

	ret = ocfs2_journal_access_rb(handle, &new_tree->rf_ci, new_bh,
				      OCFS2_JOURNAL_ACCESS_CREATE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	/* Initialize ocfs2_refcount_block. */
	rb = (struct ocfs2_refcount_block *)new_bh->b_data;
	memset(rb, 0, inode->i_sb->s_blocksize);
	strcpy((void *)rb, OCFS2_REFCOUNT_BLOCK_SIGNATURE);
634
	rb->rf_suballoc_slot = cpu_to_le16(meta_ac->ac_alloc_slot);
635
	rb->rf_suballoc_loc = cpu_to_le64(suballoc_loc);
T
Tao Ma 已提交
636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653
	rb->rf_suballoc_bit = cpu_to_le16(suballoc_bit_start);
	rb->rf_fs_generation = cpu_to_le32(osb->fs_generation);
	rb->rf_blkno = cpu_to_le64(first_blkno);
	rb->rf_count = cpu_to_le32(1);
	rb->rf_records.rl_count =
			cpu_to_le16(ocfs2_refcount_recs_per_rb(osb->sb));
	spin_lock(&osb->osb_lock);
	rb->rf_generation = osb->s_next_generation++;
	spin_unlock(&osb->osb_lock);

	ocfs2_journal_dirty(handle, new_bh);

	spin_lock(&oi->ip_lock);
	oi->ip_dyn_features |= OCFS2_HAS_REFCOUNT_FL;
	di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
	di->i_refcount_loc = cpu_to_le64(first_blkno);
	spin_unlock(&oi->ip_lock);

T
Tao Ma 已提交
654
	trace_ocfs2_create_refcount_tree_blkno((unsigned long long)first_blkno);
T
Tao Ma 已提交
655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798

	ocfs2_journal_dirty(handle, di_bh);

	/*
	 * We have to init the tree lock here since it will use
	 * the generation number to create it.
	 */
	new_tree->rf_generation = le32_to_cpu(rb->rf_generation);
	ocfs2_init_refcount_tree_lock(osb, new_tree, first_blkno,
				      new_tree->rf_generation);

	spin_lock(&osb->osb_lock);
	tree = ocfs2_find_refcount_tree(osb, first_blkno);

	/*
	 * We've just created a new refcount tree in this block.  If
	 * we found a refcount tree on the ocfs2_super, it must be
	 * one we just deleted.  We free the old tree before
	 * inserting the new tree.
	 */
	BUG_ON(tree && tree->rf_generation == new_tree->rf_generation);
	if (tree)
		ocfs2_erase_refcount_tree_from_list_no_lock(osb, tree);
	ocfs2_insert_refcount_tree(osb, new_tree);
	spin_unlock(&osb->osb_lock);
	new_tree = NULL;
	if (tree)
		ocfs2_refcount_tree_put(tree);

out_commit:
	ocfs2_commit_trans(osb, handle);

out:
	if (new_tree) {
		ocfs2_metadata_cache_exit(&new_tree->rf_ci);
		kfree(new_tree);
	}

	brelse(new_bh);
	if (meta_ac)
		ocfs2_free_alloc_context(meta_ac);

	return ret;
}

static int ocfs2_set_refcount_tree(struct inode *inode,
				   struct buffer_head *di_bh,
				   u64 refcount_loc)
{
	int ret;
	handle_t *handle = NULL;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_inode_info *oi = OCFS2_I(inode);
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	struct buffer_head *ref_root_bh = NULL;
	struct ocfs2_refcount_block *rb;
	struct ocfs2_refcount_tree *ref_tree;

	BUG_ON(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL);

	ret = ocfs2_lock_refcount_tree(osb, refcount_loc, 1,
				       &ref_tree, &ref_root_bh);
	if (ret) {
		mlog_errno(ret);
		return ret;
	}

	handle = ocfs2_start_trans(osb, OCFS2_REFCOUNT_TREE_SET_CREDITS);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	ret = ocfs2_journal_access_rb(handle, &ref_tree->rf_ci, ref_root_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
	le32_add_cpu(&rb->rf_count, 1);

	ocfs2_journal_dirty(handle, ref_root_bh);

	spin_lock(&oi->ip_lock);
	oi->ip_dyn_features |= OCFS2_HAS_REFCOUNT_FL;
	di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
	di->i_refcount_loc = cpu_to_le64(refcount_loc);
	spin_unlock(&oi->ip_lock);
	ocfs2_journal_dirty(handle, di_bh);

out_commit:
	ocfs2_commit_trans(osb, handle);
out:
	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
	brelse(ref_root_bh);

	return ret;
}

int ocfs2_remove_refcount_tree(struct inode *inode, struct buffer_head *di_bh)
{
	int ret, delete_tree = 0;
	handle_t *handle = NULL;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_inode_info *oi = OCFS2_I(inode);
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	struct ocfs2_refcount_block *rb;
	struct inode *alloc_inode = NULL;
	struct buffer_head *alloc_bh = NULL;
	struct buffer_head *blk_bh = NULL;
	struct ocfs2_refcount_tree *ref_tree;
	int credits = OCFS2_REFCOUNT_TREE_REMOVE_CREDITS;
	u64 blk = 0, bg_blkno = 0, ref_blkno = le64_to_cpu(di->i_refcount_loc);
	u16 bit = 0;

	if (!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL))
		return 0;

	BUG_ON(!ref_blkno);
	ret = ocfs2_lock_refcount_tree(osb, ref_blkno, 1, &ref_tree, &blk_bh);
	if (ret) {
		mlog_errno(ret);
		return ret;
	}

	rb = (struct ocfs2_refcount_block *)blk_bh->b_data;

	/*
	 * If we are the last user, we need to free the block.
	 * So lock the allocator ahead.
	 */
	if (le32_to_cpu(rb->rf_count) == 1) {
		blk = le64_to_cpu(rb->rf_blkno);
		bit = le16_to_cpu(rb->rf_suballoc_bit);
799 800 801 802
		if (rb->rf_suballoc_loc)
			bg_blkno = le64_to_cpu(rb->rf_suballoc_loc);
		else
			bg_blkno = ocfs2_which_suballoc_group(blk, bit);
T
Tao Ma 已提交
803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882

		alloc_inode = ocfs2_get_system_file_inode(osb,
					EXTENT_ALLOC_SYSTEM_INODE,
					le16_to_cpu(rb->rf_suballoc_slot));
		if (!alloc_inode) {
			ret = -ENOMEM;
			mlog_errno(ret);
			goto out;
		}
		mutex_lock(&alloc_inode->i_mutex);

		ret = ocfs2_inode_lock(alloc_inode, &alloc_bh, 1);
		if (ret) {
			mlog_errno(ret);
			goto out_mutex;
		}

		credits += OCFS2_SUBALLOC_FREE;
	}

	handle = ocfs2_start_trans(osb, credits);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out_unlock;
	}

	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	ret = ocfs2_journal_access_rb(handle, &ref_tree->rf_ci, blk_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	spin_lock(&oi->ip_lock);
	oi->ip_dyn_features &= ~OCFS2_HAS_REFCOUNT_FL;
	di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
	di->i_refcount_loc = 0;
	spin_unlock(&oi->ip_lock);
	ocfs2_journal_dirty(handle, di_bh);

	le32_add_cpu(&rb->rf_count , -1);
	ocfs2_journal_dirty(handle, blk_bh);

	if (!rb->rf_count) {
		delete_tree = 1;
		ocfs2_erase_refcount_tree_from_list(osb, ref_tree);
		ret = ocfs2_free_suballoc_bits(handle, alloc_inode,
					       alloc_bh, bit, bg_blkno, 1);
		if (ret)
			mlog_errno(ret);
	}

out_commit:
	ocfs2_commit_trans(osb, handle);
out_unlock:
	if (alloc_inode) {
		ocfs2_inode_unlock(alloc_inode, 1);
		brelse(alloc_bh);
	}
out_mutex:
	if (alloc_inode) {
		mutex_unlock(&alloc_inode->i_mutex);
		iput(alloc_inode);
	}
out:
	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
	if (delete_tree)
		ocfs2_refcount_tree_put(ref_tree);
	brelse(blk_bh);

	return ret;
}
883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925

static void ocfs2_find_refcount_rec_in_rl(struct ocfs2_caching_info *ci,
					  struct buffer_head *ref_leaf_bh,
					  u64 cpos, unsigned int len,
					  struct ocfs2_refcount_rec *ret_rec,
					  int *index)
{
	int i = 0;
	struct ocfs2_refcount_block *rb =
		(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
	struct ocfs2_refcount_rec *rec = NULL;

	for (; i < le16_to_cpu(rb->rf_records.rl_used); i++) {
		rec = &rb->rf_records.rl_recs[i];

		if (le64_to_cpu(rec->r_cpos) +
		    le32_to_cpu(rec->r_clusters) <= cpos)
			continue;
		else if (le64_to_cpu(rec->r_cpos) > cpos)
			break;

		/* ok, cpos fail in this rec. Just return. */
		if (ret_rec)
			*ret_rec = *rec;
		goto out;
	}

	if (ret_rec) {
		/* We meet with a hole here, so fake the rec. */
		ret_rec->r_cpos = cpu_to_le64(cpos);
		ret_rec->r_refcount = 0;
		if (i < le16_to_cpu(rb->rf_records.rl_used) &&
		    le64_to_cpu(rec->r_cpos) < cpos + len)
			ret_rec->r_clusters =
				cpu_to_le32(le64_to_cpu(rec->r_cpos) - cpos);
		else
			ret_rec->r_clusters = cpu_to_le32(len);
	}

out:
	*index = i;
}

926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961
/*
 * Try to remove refcount tree. The mechanism is:
 * 1) Check whether i_clusters == 0, if no, exit.
 * 2) check whether we have i_xattr_loc in dinode. if yes, exit.
 * 3) Check whether we have inline xattr stored outside, if yes, exit.
 * 4) Remove the tree.
 */
int ocfs2_try_remove_refcount_tree(struct inode *inode,
				   struct buffer_head *di_bh)
{
	int ret;
	struct ocfs2_inode_info *oi = OCFS2_I(inode);
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;

	down_write(&oi->ip_xattr_sem);
	down_write(&oi->ip_alloc_sem);

	if (oi->ip_clusters)
		goto out;

	if ((oi->ip_dyn_features & OCFS2_HAS_XATTR_FL) && di->i_xattr_loc)
		goto out;

	if (oi->ip_dyn_features & OCFS2_INLINE_XATTR_FL &&
	    ocfs2_has_inline_xattr_value_outside(inode, di))
		goto out;

	ret = ocfs2_remove_refcount_tree(inode, di_bh);
	if (ret)
		mlog_errno(ret);
out:
	up_write(&oi->ip_alloc_sem);
	up_write(&oi->ip_xattr_sem);
	return 0;
}

962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
/*
 * Find the end range for a leaf refcount block indicated by
 * el->l_recs[index].e_blkno.
 */
static int ocfs2_get_refcount_cpos_end(struct ocfs2_caching_info *ci,
				       struct buffer_head *ref_root_bh,
				       struct ocfs2_extent_block *eb,
				       struct ocfs2_extent_list *el,
				       int index,  u32 *cpos_end)
{
	int ret, i, subtree_root;
	u32 cpos;
	u64 blkno;
	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
	struct ocfs2_path *left_path = NULL, *right_path = NULL;
	struct ocfs2_extent_tree et;
	struct ocfs2_extent_list *tmp_el;

	if (index < le16_to_cpu(el->l_next_free_rec) - 1) {
		/*
		 * We have a extent rec after index, so just use the e_cpos
		 * of the next extent rec.
		 */
		*cpos_end = le32_to_cpu(el->l_recs[index+1].e_cpos);
		return 0;
	}

	if (!eb || (eb && !eb->h_next_leaf_blk)) {
		/*
		 * We are the last extent rec, so any high cpos should
		 * be stored in this leaf refcount block.
		 */
		*cpos_end = UINT_MAX;
		return 0;
	}

	/*
	 * If the extent block isn't the last one, we have to find
	 * the subtree root between this extent block and the next
	 * leaf extent block and get the corresponding e_cpos from
	 * the subroot. Otherwise we may corrupt the b-tree.
	 */
	ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);

	left_path = ocfs2_new_path_from_et(&et);
	if (!left_path) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	cpos = le32_to_cpu(eb->h_list.l_recs[index].e_cpos);
	ret = ocfs2_find_path(ci, left_path, cpos);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	right_path = ocfs2_new_path_from_path(left_path);
	if (!right_path) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_find_cpos_for_right_leaf(sb, left_path, &cpos);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_find_path(ci, right_path, cpos);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	subtree_root = ocfs2_find_subtree_root(&et, left_path,
					       right_path);

	tmp_el = left_path->p_node[subtree_root].el;
	blkno = left_path->p_node[subtree_root+1].bh->b_blocknr;
1044
	for (i = 0; i < le16_to_cpu(tmp_el->l_next_free_rec); i++) {
1045 1046 1047 1048 1049 1050
		if (le64_to_cpu(tmp_el->l_recs[i].e_blkno) == blkno) {
			*cpos_end = le32_to_cpu(tmp_el->l_recs[i+1].e_cpos);
			break;
		}
	}

1051
	BUG_ON(i == le16_to_cpu(tmp_el->l_next_free_rec));
1052 1053 1054 1055 1056 1057 1058

out:
	ocfs2_free_path(left_path);
	ocfs2_free_path(right_path);
	return ret;
}

1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073
/*
 * Given a cpos and len, try to find the refcount record which contains cpos.
 * 1. If cpos can be found in one refcount record, return the record.
 * 2. If cpos can't be found, return a fake record which start from cpos
 *    and end at a small value between cpos+len and start of the next record.
 *    This fake record has r_refcount = 0.
 */
static int ocfs2_get_refcount_rec(struct ocfs2_caching_info *ci,
				  struct buffer_head *ref_root_bh,
				  u64 cpos, unsigned int len,
				  struct ocfs2_refcount_rec *ret_rec,
				  int *index,
				  struct buffer_head **ret_bh)
{
	int ret = 0, i, found;
1074
	u32 low_cpos, uninitialized_var(cpos_end);
1075
	struct ocfs2_extent_list *el;
1076 1077
	struct ocfs2_extent_rec *rec = NULL;
	struct ocfs2_extent_block *eb = NULL;
1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
	struct buffer_head *eb_bh = NULL, *ref_leaf_bh = NULL;
	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
	struct ocfs2_refcount_block *rb =
			(struct ocfs2_refcount_block *)ref_root_bh->b_data;

	if (!(le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL)) {
		ocfs2_find_refcount_rec_in_rl(ci, ref_root_bh, cpos, len,
					      ret_rec, index);
		*ret_bh = ref_root_bh;
		get_bh(ref_root_bh);
		return 0;
	}

	el = &rb->rf_list;
	low_cpos = cpos & OCFS2_32BIT_POS_MASK;

	if (el->l_tree_depth) {
		ret = ocfs2_find_leaf(ci, el, low_cpos, &eb_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		eb = (struct ocfs2_extent_block *) eb_bh->b_data;
		el = &eb->h_list;

		if (el->l_tree_depth) {
			ocfs2_error(sb,
			"refcount tree %llu has non zero tree "
			"depth in leaf btree tree block %llu\n",
			(unsigned long long)ocfs2_metadata_cache_owner(ci),
			(unsigned long long)eb_bh->b_blocknr);
			ret = -EROFS;
			goto out;
		}
	}

	found = 0;
	for (i = le16_to_cpu(el->l_next_free_rec) - 1; i >= 0; i--) {
		rec = &el->l_recs[i];

		if (le32_to_cpu(rec->e_cpos) <= low_cpos) {
			found = 1;
			break;
		}
	}

1125 1126 1127 1128 1129 1130 1131
	if (found) {
		ret = ocfs2_get_refcount_cpos_end(ci, ref_root_bh,
						  eb, el, i, &cpos_end);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
1132

1133 1134
		if (cpos_end < low_cpos + len)
			len = cpos_end - low_cpos;
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240
	}

	ret = ocfs2_read_refcount_block(ci, le64_to_cpu(rec->e_blkno),
					&ref_leaf_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ocfs2_find_refcount_rec_in_rl(ci, ref_leaf_bh, cpos, len,
				      ret_rec, index);
	*ret_bh = ref_leaf_bh;
out:
	brelse(eb_bh);
	return ret;
}

enum ocfs2_ref_rec_contig {
	REF_CONTIG_NONE = 0,
	REF_CONTIG_LEFT,
	REF_CONTIG_RIGHT,
	REF_CONTIG_LEFTRIGHT,
};

static enum ocfs2_ref_rec_contig
	ocfs2_refcount_rec_adjacent(struct ocfs2_refcount_block *rb,
				    int index)
{
	if ((rb->rf_records.rl_recs[index].r_refcount ==
	    rb->rf_records.rl_recs[index + 1].r_refcount) &&
	    (le64_to_cpu(rb->rf_records.rl_recs[index].r_cpos) +
	    le32_to_cpu(rb->rf_records.rl_recs[index].r_clusters) ==
	    le64_to_cpu(rb->rf_records.rl_recs[index + 1].r_cpos)))
		return REF_CONTIG_RIGHT;

	return REF_CONTIG_NONE;
}

static enum ocfs2_ref_rec_contig
	ocfs2_refcount_rec_contig(struct ocfs2_refcount_block *rb,
				  int index)
{
	enum ocfs2_ref_rec_contig ret = REF_CONTIG_NONE;

	if (index < le16_to_cpu(rb->rf_records.rl_used) - 1)
		ret = ocfs2_refcount_rec_adjacent(rb, index);

	if (index > 0) {
		enum ocfs2_ref_rec_contig tmp;

		tmp = ocfs2_refcount_rec_adjacent(rb, index - 1);

		if (tmp == REF_CONTIG_RIGHT) {
			if (ret == REF_CONTIG_RIGHT)
				ret = REF_CONTIG_LEFTRIGHT;
			else
				ret = REF_CONTIG_LEFT;
		}
	}

	return ret;
}

static void ocfs2_rotate_refcount_rec_left(struct ocfs2_refcount_block *rb,
					   int index)
{
	BUG_ON(rb->rf_records.rl_recs[index].r_refcount !=
	       rb->rf_records.rl_recs[index+1].r_refcount);

	le32_add_cpu(&rb->rf_records.rl_recs[index].r_clusters,
		     le32_to_cpu(rb->rf_records.rl_recs[index+1].r_clusters));

	if (index < le16_to_cpu(rb->rf_records.rl_used) - 2)
		memmove(&rb->rf_records.rl_recs[index + 1],
			&rb->rf_records.rl_recs[index + 2],
			sizeof(struct ocfs2_refcount_rec) *
			(le16_to_cpu(rb->rf_records.rl_used) - index - 2));

	memset(&rb->rf_records.rl_recs[le16_to_cpu(rb->rf_records.rl_used) - 1],
	       0, sizeof(struct ocfs2_refcount_rec));
	le16_add_cpu(&rb->rf_records.rl_used, -1);
}

/*
 * Merge the refcount rec if we are contiguous with the adjacent recs.
 */
static void ocfs2_refcount_rec_merge(struct ocfs2_refcount_block *rb,
				     int index)
{
	enum ocfs2_ref_rec_contig contig =
				ocfs2_refcount_rec_contig(rb, index);

	if (contig == REF_CONTIG_NONE)
		return;

	if (contig == REF_CONTIG_LEFT || contig == REF_CONTIG_LEFTRIGHT) {
		BUG_ON(index == 0);
		index--;
	}

	ocfs2_rotate_refcount_rec_left(rb, index);

	if (contig == REF_CONTIG_LEFTRIGHT)
		ocfs2_rotate_refcount_rec_left(rb, index);
}

1241 1242 1243 1244
/*
 * Change the refcount indexed by "index" in ref_bh.
 * If refcount reaches 0, remove it.
 */
1245 1246 1247
static int ocfs2_change_refcount_rec(handle_t *handle,
				     struct ocfs2_caching_info *ci,
				     struct buffer_head *ref_leaf_bh,
1248
				     int index, int merge, int change)
1249 1250 1251 1252
{
	int ret;
	struct ocfs2_refcount_block *rb =
			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
1253 1254
	struct ocfs2_refcount_list *rl = &rb->rf_records;
	struct ocfs2_refcount_rec *rec = &rl->rl_recs[index];
1255 1256 1257 1258 1259 1260 1261 1262

	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

T
Tao Ma 已提交
1263 1264 1265
	trace_ocfs2_change_refcount_rec(
		(unsigned long long)ocfs2_metadata_cache_owner(ci),
		index, le32_to_cpu(rec->r_refcount), change);
1266 1267
	le32_add_cpu(&rec->r_refcount, change);

1268 1269 1270 1271 1272 1273 1274 1275 1276 1277
	if (!rec->r_refcount) {
		if (index != le16_to_cpu(rl->rl_used) - 1) {
			memmove(rec, rec + 1,
				(le16_to_cpu(rl->rl_used) - index - 1) *
				sizeof(struct ocfs2_refcount_rec));
			memset(&rl->rl_recs[le16_to_cpu(rl->rl_used) - 1],
			       0, sizeof(struct ocfs2_refcount_rec));
		}

		le16_add_cpu(&rl->rl_used, -1);
1278
	} else if (merge)
1279
		ocfs2_refcount_rec_merge(rb, index);
1280

1281
	ocfs2_journal_dirty(handle, ref_leaf_bh);
1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294
out:
	return ret;
}

static int ocfs2_expand_inline_ref_root(handle_t *handle,
					struct ocfs2_caching_info *ci,
					struct buffer_head *ref_root_bh,
					struct buffer_head **ref_leaf_bh,
					struct ocfs2_alloc_context *meta_ac)
{
	int ret;
	u16 suballoc_bit_start;
	u32 num_got;
1295
	u64 suballoc_loc, blkno;
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308
	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
	struct buffer_head *new_bh = NULL;
	struct ocfs2_refcount_block *new_rb;
	struct ocfs2_refcount_block *root_rb =
			(struct ocfs2_refcount_block *)ref_root_bh->b_data;

	ret = ocfs2_journal_access_rb(handle, ci, ref_root_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

1309
	ret = ocfs2_claim_metadata(handle, meta_ac, 1, &suballoc_loc,
1310 1311 1312 1313 1314 1315 1316 1317 1318
				   &suballoc_bit_start, &num_got,
				   &blkno);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	new_bh = sb_getblk(sb, blkno);
	if (new_bh == NULL) {
1319
		ret = -ENOMEM;
1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339
		mlog_errno(ret);
		goto out;
	}
	ocfs2_set_new_buffer_uptodate(ci, new_bh);

	ret = ocfs2_journal_access_rb(handle, ci, new_bh,
				      OCFS2_JOURNAL_ACCESS_CREATE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	/*
	 * Initialize ocfs2_refcount_block.
	 * It should contain the same information as the old root.
	 * so just memcpy it and change the corresponding field.
	 */
	memcpy(new_bh->b_data, ref_root_bh->b_data, sb->s_blocksize);

	new_rb = (struct ocfs2_refcount_block *)new_bh->b_data;
1340
	new_rb->rf_suballoc_slot = cpu_to_le16(meta_ac->ac_alloc_slot);
1341
	new_rb->rf_suballoc_loc = cpu_to_le64(suballoc_loc);
1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360
	new_rb->rf_suballoc_bit = cpu_to_le16(suballoc_bit_start);
	new_rb->rf_blkno = cpu_to_le64(blkno);
	new_rb->rf_cpos = cpu_to_le32(0);
	new_rb->rf_parent = cpu_to_le64(ref_root_bh->b_blocknr);
	new_rb->rf_flags = cpu_to_le32(OCFS2_REFCOUNT_LEAF_FL);
	ocfs2_journal_dirty(handle, new_bh);

	/* Now change the root. */
	memset(&root_rb->rf_list, 0, sb->s_blocksize -
	       offsetof(struct ocfs2_refcount_block, rf_list));
	root_rb->rf_list.l_count = cpu_to_le16(ocfs2_extent_recs_per_rb(sb));
	root_rb->rf_clusters = cpu_to_le32(1);
	root_rb->rf_list.l_next_free_rec = cpu_to_le16(1);
	root_rb->rf_list.l_recs[0].e_blkno = cpu_to_le64(blkno);
	root_rb->rf_list.l_recs[0].e_leaf_clusters = cpu_to_le16(1);
	root_rb->rf_flags = cpu_to_le32(OCFS2_REFCOUNT_TREE_FL);

	ocfs2_journal_dirty(handle, ref_root_bh);

T
Tao Ma 已提交
1361 1362
	trace_ocfs2_expand_inline_ref_root((unsigned long long)blkno,
		le16_to_cpu(new_rb->rf_records.rl_used));
1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410

	*ref_leaf_bh = new_bh;
	new_bh = NULL;
out:
	brelse(new_bh);
	return ret;
}

static int ocfs2_refcount_rec_no_intersect(struct ocfs2_refcount_rec *prev,
					   struct ocfs2_refcount_rec *next)
{
	if (ocfs2_get_ref_rec_low_cpos(prev) + le32_to_cpu(prev->r_clusters) <=
		ocfs2_get_ref_rec_low_cpos(next))
		return 1;

	return 0;
}

static int cmp_refcount_rec_by_low_cpos(const void *a, const void *b)
{
	const struct ocfs2_refcount_rec *l = a, *r = b;
	u32 l_cpos = ocfs2_get_ref_rec_low_cpos(l);
	u32 r_cpos = ocfs2_get_ref_rec_low_cpos(r);

	if (l_cpos > r_cpos)
		return 1;
	if (l_cpos < r_cpos)
		return -1;
	return 0;
}

static int cmp_refcount_rec_by_cpos(const void *a, const void *b)
{
	const struct ocfs2_refcount_rec *l = a, *r = b;
	u64 l_cpos = le64_to_cpu(l->r_cpos);
	u64 r_cpos = le64_to_cpu(r->r_cpos);

	if (l_cpos > r_cpos)
		return 1;
	if (l_cpos < r_cpos)
		return -1;
	return 0;
}

static void swap_refcount_rec(void *a, void *b, int size)
{
	struct ocfs2_refcount_rec *l = a, *r = b, tmp;

1411 1412 1413
	tmp = *l;
	*l = *r;
	*r = tmp;
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472
}

/*
 * The refcount cpos are ordered by their 64bit cpos,
 * But we will use the low 32 bit to be the e_cpos in the b-tree.
 * So we need to make sure that this pos isn't intersected with others.
 *
 * Note: The refcount block is already sorted by their low 32 bit cpos,
 *       So just try the middle pos first, and we will exit when we find
 *       the good position.
 */
static int ocfs2_find_refcount_split_pos(struct ocfs2_refcount_list *rl,
					 u32 *split_pos, int *split_index)
{
	int num_used = le16_to_cpu(rl->rl_used);
	int delta, middle = num_used / 2;

	for (delta = 0; delta < middle; delta++) {
		/* Let's check delta earlier than middle */
		if (ocfs2_refcount_rec_no_intersect(
					&rl->rl_recs[middle - delta - 1],
					&rl->rl_recs[middle - delta])) {
			*split_index = middle - delta;
			break;
		}

		/* For even counts, don't walk off the end */
		if ((middle + delta + 1) == num_used)
			continue;

		/* Now try delta past middle */
		if (ocfs2_refcount_rec_no_intersect(
					&rl->rl_recs[middle + delta],
					&rl->rl_recs[middle + delta + 1])) {
			*split_index = middle + delta + 1;
			break;
		}
	}

	if (delta >= middle)
		return -ENOSPC;

	*split_pos = ocfs2_get_ref_rec_low_cpos(&rl->rl_recs[*split_index]);
	return 0;
}

static int ocfs2_divide_leaf_refcount_block(struct buffer_head *ref_leaf_bh,
					    struct buffer_head *new_bh,
					    u32 *split_cpos)
{
	int split_index = 0, num_moved, ret;
	u32 cpos = 0;
	struct ocfs2_refcount_block *rb =
			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
	struct ocfs2_refcount_list *rl = &rb->rf_records;
	struct ocfs2_refcount_block *new_rb =
			(struct ocfs2_refcount_block *)new_bh->b_data;
	struct ocfs2_refcount_list *new_rl = &new_rb->rf_records;

T
Tao Ma 已提交
1473 1474
	trace_ocfs2_divide_leaf_refcount_block(
		(unsigned long long)ref_leaf_bh->b_blocknr,
A
Al Viro 已提交
1475
		le16_to_cpu(rl->rl_count), le16_to_cpu(rl->rl_used));
1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511

	/*
	 * XXX: Improvement later.
	 * If we know all the high 32 bit cpos is the same, no need to sort.
	 *
	 * In order to make the whole process safe, we do:
	 * 1. sort the entries by their low 32 bit cpos first so that we can
	 *    find the split cpos easily.
	 * 2. call ocfs2_insert_extent to insert the new refcount block.
	 * 3. move the refcount rec to the new block.
	 * 4. sort the entries by their 64 bit cpos.
	 * 5. dirty the new_rb and rb.
	 */
	sort(&rl->rl_recs, le16_to_cpu(rl->rl_used),
	     sizeof(struct ocfs2_refcount_rec),
	     cmp_refcount_rec_by_low_cpos, swap_refcount_rec);

	ret = ocfs2_find_refcount_split_pos(rl, &cpos, &split_index);
	if (ret) {
		mlog_errno(ret);
		return ret;
	}

	new_rb->rf_cpos = cpu_to_le32(cpos);

	/* move refcount records starting from split_index to the new block. */
	num_moved = le16_to_cpu(rl->rl_used) - split_index;
	memcpy(new_rl->rl_recs, &rl->rl_recs[split_index],
	       num_moved * sizeof(struct ocfs2_refcount_rec));

	/*ok, remove the entries we just moved over to the other block. */
	memset(&rl->rl_recs[split_index], 0,
	       num_moved * sizeof(struct ocfs2_refcount_rec));

	/* change old and new rl_used accordingly. */
	le16_add_cpu(&rl->rl_used, -num_moved);
T
Tao Ma 已提交
1512
	new_rl->rl_used = cpu_to_le16(num_moved);
1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534

	sort(&rl->rl_recs, le16_to_cpu(rl->rl_used),
	     sizeof(struct ocfs2_refcount_rec),
	     cmp_refcount_rec_by_cpos, swap_refcount_rec);

	sort(&new_rl->rl_recs, le16_to_cpu(new_rl->rl_used),
	     sizeof(struct ocfs2_refcount_rec),
	     cmp_refcount_rec_by_cpos, swap_refcount_rec);

	*split_cpos = cpos;
	return 0;
}

static int ocfs2_new_leaf_refcount_block(handle_t *handle,
					 struct ocfs2_caching_info *ci,
					 struct buffer_head *ref_root_bh,
					 struct buffer_head *ref_leaf_bh,
					 struct ocfs2_alloc_context *meta_ac)
{
	int ret;
	u16 suballoc_bit_start;
	u32 num_got, new_cpos;
1535
	u64 suballoc_loc, blkno;
1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558
	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
	struct ocfs2_refcount_block *root_rb =
			(struct ocfs2_refcount_block *)ref_root_bh->b_data;
	struct buffer_head *new_bh = NULL;
	struct ocfs2_refcount_block *new_rb;
	struct ocfs2_extent_tree ref_et;

	BUG_ON(!(le32_to_cpu(root_rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL));

	ret = ocfs2_journal_access_rb(handle, ci, ref_root_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

1559
	ret = ocfs2_claim_metadata(handle, meta_ac, 1, &suballoc_loc,
1560 1561 1562 1563 1564 1565 1566 1567 1568
				   &suballoc_bit_start, &num_got,
				   &blkno);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	new_bh = sb_getblk(sb, blkno);
	if (new_bh == NULL) {
1569
		ret = -ENOMEM;
1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585
		mlog_errno(ret);
		goto out;
	}
	ocfs2_set_new_buffer_uptodate(ci, new_bh);

	ret = ocfs2_journal_access_rb(handle, ci, new_bh,
				      OCFS2_JOURNAL_ACCESS_CREATE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	/* Initialize ocfs2_refcount_block. */
	new_rb = (struct ocfs2_refcount_block *)new_bh->b_data;
	memset(new_rb, 0, sb->s_blocksize);
	strcpy((void *)new_rb, OCFS2_REFCOUNT_BLOCK_SIGNATURE);
1586
	new_rb->rf_suballoc_slot = cpu_to_le16(meta_ac->ac_alloc_slot);
1587
	new_rb->rf_suballoc_loc = cpu_to_le64(suballoc_loc);
1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
	new_rb->rf_suballoc_bit = cpu_to_le16(suballoc_bit_start);
	new_rb->rf_fs_generation = cpu_to_le32(OCFS2_SB(sb)->fs_generation);
	new_rb->rf_blkno = cpu_to_le64(blkno);
	new_rb->rf_parent = cpu_to_le64(ref_root_bh->b_blocknr);
	new_rb->rf_flags = cpu_to_le32(OCFS2_REFCOUNT_LEAF_FL);
	new_rb->rf_records.rl_count =
				cpu_to_le16(ocfs2_refcount_recs_per_rb(sb));
	new_rb->rf_generation = root_rb->rf_generation;

	ret = ocfs2_divide_leaf_refcount_block(ref_leaf_bh, new_bh, &new_cpos);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ocfs2_journal_dirty(handle, ref_leaf_bh);
	ocfs2_journal_dirty(handle, new_bh);

	ocfs2_init_refcount_extent_tree(&ref_et, ci, ref_root_bh);

T
Tao Ma 已提交
1608 1609
	trace_ocfs2_new_leaf_refcount_block(
			(unsigned long long)new_bh->b_blocknr, new_cpos);
1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705

	/* Insert the new leaf block with the specific offset cpos. */
	ret = ocfs2_insert_extent(handle, &ref_et, new_cpos, new_bh->b_blocknr,
				  1, 0, meta_ac);
	if (ret)
		mlog_errno(ret);

out:
	brelse(new_bh);
	return ret;
}

static int ocfs2_expand_refcount_tree(handle_t *handle,
				      struct ocfs2_caching_info *ci,
				      struct buffer_head *ref_root_bh,
				      struct buffer_head *ref_leaf_bh,
				      struct ocfs2_alloc_context *meta_ac)
{
	int ret;
	struct buffer_head *expand_bh = NULL;

	if (ref_root_bh == ref_leaf_bh) {
		/*
		 * the old root bh hasn't been expanded to a b-tree,
		 * so expand it first.
		 */
		ret = ocfs2_expand_inline_ref_root(handle, ci, ref_root_bh,
						   &expand_bh, meta_ac);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	} else {
		expand_bh = ref_leaf_bh;
		get_bh(expand_bh);
	}


	/* Now add a new refcount block into the tree.*/
	ret = ocfs2_new_leaf_refcount_block(handle, ci, ref_root_bh,
					    expand_bh, meta_ac);
	if (ret)
		mlog_errno(ret);
out:
	brelse(expand_bh);
	return ret;
}

/*
 * Adjust the extent rec in b-tree representing ref_leaf_bh.
 *
 * Only called when we have inserted a new refcount rec at index 0
 * which means ocfs2_extent_rec.e_cpos may need some change.
 */
static int ocfs2_adjust_refcount_rec(handle_t *handle,
				     struct ocfs2_caching_info *ci,
				     struct buffer_head *ref_root_bh,
				     struct buffer_head *ref_leaf_bh,
				     struct ocfs2_refcount_rec *rec)
{
	int ret = 0, i;
	u32 new_cpos, old_cpos;
	struct ocfs2_path *path = NULL;
	struct ocfs2_extent_tree et;
	struct ocfs2_refcount_block *rb =
		(struct ocfs2_refcount_block *)ref_root_bh->b_data;
	struct ocfs2_extent_list *el;

	if (!(le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL))
		goto out;

	rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
	old_cpos = le32_to_cpu(rb->rf_cpos);
	new_cpos = le64_to_cpu(rec->r_cpos) & OCFS2_32BIT_POS_MASK;
	if (old_cpos <= new_cpos)
		goto out;

	ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);

	path = ocfs2_new_path_from_et(&et);
	if (!path) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_find_path(ci, path, old_cpos);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	/*
	 * 2 more credits, one for the leaf refcount block, one for
	 * the extent block contains the extent rec.
	 */
1706
	ret = ocfs2_extend_trans(handle, 2);
1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752
	if (ret < 0) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret < 0) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_journal_access_eb(handle, ci, path_leaf_bh(path),
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret < 0) {
		mlog_errno(ret);
		goto out;
	}

	/* change the leaf extent block first. */
	el = path_leaf_el(path);

	for (i = 0; i < le16_to_cpu(el->l_next_free_rec); i++)
		if (le32_to_cpu(el->l_recs[i].e_cpos) == old_cpos)
			break;

	BUG_ON(i == le16_to_cpu(el->l_next_free_rec));

	el->l_recs[i].e_cpos = cpu_to_le32(new_cpos);

	/* change the r_cpos in the leaf block. */
	rb->rf_cpos = cpu_to_le32(new_cpos);

	ocfs2_journal_dirty(handle, path_leaf_bh(path));
	ocfs2_journal_dirty(handle, ref_leaf_bh);

out:
	ocfs2_free_path(path);
	return ret;
}

static int ocfs2_insert_refcount_rec(handle_t *handle,
				     struct ocfs2_caching_info *ci,
				     struct buffer_head *ref_root_bh,
				     struct buffer_head *ref_leaf_bh,
				     struct ocfs2_refcount_rec *rec,
1753
				     int index, int merge,
1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800
				     struct ocfs2_alloc_context *meta_ac)
{
	int ret;
	struct ocfs2_refcount_block *rb =
			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
	struct ocfs2_refcount_list *rf_list = &rb->rf_records;
	struct buffer_head *new_bh = NULL;

	BUG_ON(le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL);

	if (rf_list->rl_used == rf_list->rl_count) {
		u64 cpos = le64_to_cpu(rec->r_cpos);
		u32 len = le32_to_cpu(rec->r_clusters);

		ret = ocfs2_expand_refcount_tree(handle, ci, ref_root_bh,
						 ref_leaf_bh, meta_ac);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
					     cpos, len, NULL, &index,
					     &new_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		ref_leaf_bh = new_bh;
		rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
		rf_list = &rb->rf_records;
	}

	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	if (index < le16_to_cpu(rf_list->rl_used))
		memmove(&rf_list->rl_recs[index + 1],
			&rf_list->rl_recs[index],
			(le16_to_cpu(rf_list->rl_used) - index) *
			 sizeof(struct ocfs2_refcount_rec));

T
Tao Ma 已提交
1801 1802 1803 1804
	trace_ocfs2_insert_refcount_rec(
		(unsigned long long)ref_leaf_bh->b_blocknr, index,
		(unsigned long long)le64_to_cpu(rec->r_cpos),
		le32_to_cpu(rec->r_clusters), le32_to_cpu(rec->r_refcount));
1805 1806 1807 1808 1809

	rf_list->rl_recs[index] = *rec;

	le16_add_cpu(&rf_list->rl_used, 1);

1810 1811
	if (merge)
		ocfs2_refcount_rec_merge(rb, index);
1812

1813
	ocfs2_journal_dirty(handle, ref_leaf_bh);
1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840

	if (index == 0) {
		ret = ocfs2_adjust_refcount_rec(handle, ci,
						ref_root_bh,
						ref_leaf_bh, rec);
		if (ret)
			mlog_errno(ret);
	}
out:
	brelse(new_bh);
	return ret;
}

/*
 * Split the refcount_rec indexed by "index" in ref_leaf_bh.
 * This is much simple than our b-tree code.
 * split_rec is the new refcount rec we want to insert.
 * If split_rec->r_refcount > 0, we are changing the refcount(in case we
 * increase refcount or decrease a refcount to non-zero).
 * If split_rec->r_refcount == 0, we are punching a hole in current refcount
 * rec( in case we decrease a refcount to zero).
 */
static int ocfs2_split_refcount_rec(handle_t *handle,
				    struct ocfs2_caching_info *ci,
				    struct buffer_head *ref_root_bh,
				    struct buffer_head *ref_leaf_bh,
				    struct ocfs2_refcount_rec *split_rec,
1841
				    int index, int merge,
1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855
				    struct ocfs2_alloc_context *meta_ac,
				    struct ocfs2_cached_dealloc_ctxt *dealloc)
{
	int ret, recs_need;
	u32 len;
	struct ocfs2_refcount_block *rb =
			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
	struct ocfs2_refcount_list *rf_list = &rb->rf_records;
	struct ocfs2_refcount_rec *orig_rec = &rf_list->rl_recs[index];
	struct ocfs2_refcount_rec *tail_rec = NULL;
	struct buffer_head *new_bh = NULL;

	BUG_ON(le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL);

T
Tao Ma 已提交
1856 1857 1858 1859 1860 1861
	trace_ocfs2_split_refcount_rec(le64_to_cpu(orig_rec->r_cpos),
		le32_to_cpu(orig_rec->r_clusters),
		le32_to_cpu(orig_rec->r_refcount),
		le64_to_cpu(split_rec->r_cpos),
		le32_to_cpu(split_rec->r_clusters),
		le32_to_cpu(split_rec->r_refcount));
1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888

	/*
	 * If we just need to split the header or tail clusters,
	 * no more recs are needed, just split is OK.
	 * Otherwise we at least need one new recs.
	 */
	if (!split_rec->r_refcount &&
	    (split_rec->r_cpos == orig_rec->r_cpos ||
	     le64_to_cpu(split_rec->r_cpos) +
	     le32_to_cpu(split_rec->r_clusters) ==
	     le64_to_cpu(orig_rec->r_cpos) + le32_to_cpu(orig_rec->r_clusters)))
		recs_need = 0;
	else
		recs_need = 1;

	/*
	 * We need one more rec if we split in the middle and the new rec have
	 * some refcount in it.
	 */
	if (split_rec->r_refcount &&
	    (split_rec->r_cpos != orig_rec->r_cpos &&
	     le64_to_cpu(split_rec->r_cpos) +
	     le32_to_cpu(split_rec->r_clusters) !=
	     le64_to_cpu(orig_rec->r_cpos) + le32_to_cpu(orig_rec->r_clusters)))
		recs_need++;

	/* If the leaf block don't have enough record, expand it. */
T
Tao Ma 已提交
1889 1890
	if (le16_to_cpu(rf_list->rl_used) + recs_need >
					 le16_to_cpu(rf_list->rl_count)) {
1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951
		struct ocfs2_refcount_rec tmp_rec;
		u64 cpos = le64_to_cpu(orig_rec->r_cpos);
		len = le32_to_cpu(orig_rec->r_clusters);
		ret = ocfs2_expand_refcount_tree(handle, ci, ref_root_bh,
						 ref_leaf_bh, meta_ac);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		/*
		 * We have to re-get it since now cpos may be moved to
		 * another leaf block.
		 */
		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
					     cpos, len, &tmp_rec, &index,
					     &new_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		ref_leaf_bh = new_bh;
		rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
		rf_list = &rb->rf_records;
		orig_rec = &rf_list->rl_recs[index];
	}

	ret = ocfs2_journal_access_rb(handle, ci, ref_leaf_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	/*
	 * We have calculated out how many new records we need and store
	 * in recs_need, so spare enough space first by moving the records
	 * after "index" to the end.
	 */
	if (index != le16_to_cpu(rf_list->rl_used) - 1)
		memmove(&rf_list->rl_recs[index + 1 + recs_need],
			&rf_list->rl_recs[index + 1],
			(le16_to_cpu(rf_list->rl_used) - index - 1) *
			 sizeof(struct ocfs2_refcount_rec));

	len = (le64_to_cpu(orig_rec->r_cpos) +
	      le32_to_cpu(orig_rec->r_clusters)) -
	      (le64_to_cpu(split_rec->r_cpos) +
	      le32_to_cpu(split_rec->r_clusters));

	/*
	 * If we have "len", the we will split in the tail and move it
	 * to the end of the space we have just spared.
	 */
	if (len) {
		tail_rec = &rf_list->rl_recs[index + recs_need];

		memcpy(tail_rec, orig_rec, sizeof(struct ocfs2_refcount_rec));
		le64_add_cpu(&tail_rec->r_cpos,
			     le32_to_cpu(tail_rec->r_clusters) - len);
T
Tao Ma 已提交
1952
		tail_rec->r_clusters = cpu_to_le32(len);
1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974
	}

	/*
	 * If the split pos isn't the same as the original one, we need to
	 * split in the head.
	 *
	 * Note: We have the chance that split_rec.r_refcount = 0,
	 * recs_need = 0 and len > 0, which means we just cut the head from
	 * the orig_rec and in that case we have done some modification in
	 * orig_rec above, so the check for r_cpos is faked.
	 */
	if (split_rec->r_cpos != orig_rec->r_cpos && tail_rec != orig_rec) {
		len = le64_to_cpu(split_rec->r_cpos) -
		      le64_to_cpu(orig_rec->r_cpos);
		orig_rec->r_clusters = cpu_to_le32(len);
		index++;
	}

	le16_add_cpu(&rf_list->rl_used, recs_need);

	if (split_rec->r_refcount) {
		rf_list->rl_recs[index] = *split_rec;
T
Tao Ma 已提交
1975 1976 1977 1978 1979
		trace_ocfs2_split_refcount_rec_insert(
			(unsigned long long)ref_leaf_bh->b_blocknr, index,
			(unsigned long long)le64_to_cpu(split_rec->r_cpos),
			le32_to_cpu(split_rec->r_clusters),
			le32_to_cpu(split_rec->r_refcount));
1980

1981 1982
		if (merge)
			ocfs2_refcount_rec_merge(rb, index);
1983 1984
	}

1985
	ocfs2_journal_dirty(handle, ref_leaf_bh);
1986 1987 1988 1989 1990 1991

out:
	brelse(new_bh);
	return ret;
}

1992 1993 1994 1995 1996 1997
static int __ocfs2_increase_refcount(handle_t *handle,
				     struct ocfs2_caching_info *ci,
				     struct buffer_head *ref_root_bh,
				     u64 cpos, u32 len, int merge,
				     struct ocfs2_alloc_context *meta_ac,
				     struct ocfs2_cached_dealloc_ctxt *dealloc)
1998 1999 2000 2001 2002 2003
{
	int ret = 0, index;
	struct buffer_head *ref_leaf_bh = NULL;
	struct ocfs2_refcount_rec rec;
	unsigned int set_len = 0;

T
Tao Ma 已提交
2004
	trace_ocfs2_increase_refcount_begin(
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030
	     (unsigned long long)ocfs2_metadata_cache_owner(ci),
	     (unsigned long long)cpos, len);

	while (len) {
		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
					     cpos, len, &rec, &index,
					     &ref_leaf_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		set_len = le32_to_cpu(rec.r_clusters);

		/*
		 * Here we may meet with 3 situations:
		 *
		 * 1. If we find an already existing record, and the length
		 *    is the same, cool, we just need to increase the r_refcount
		 *    and it is OK.
		 * 2. If we find a hole, just insert it with r_refcount = 1.
		 * 3. If we are in the middle of one extent record, split
		 *    it.
		 */
		if (rec.r_refcount && le64_to_cpu(rec.r_cpos) == cpos &&
		    set_len <= len) {
T
Tao Ma 已提交
2031 2032 2033
			trace_ocfs2_increase_refcount_change(
				(unsigned long long)cpos, set_len,
				le32_to_cpu(rec.r_refcount));
2034
			ret = ocfs2_change_refcount_rec(handle, ci,
2035 2036
							ref_leaf_bh, index,
							merge, 1);
2037 2038 2039 2040 2041 2042 2043
			if (ret) {
				mlog_errno(ret);
				goto out;
			}
		} else if (!rec.r_refcount) {
			rec.r_refcount = cpu_to_le32(1);

T
Tao Ma 已提交
2044
			trace_ocfs2_increase_refcount_insert(
2045 2046 2047 2048
			     (unsigned long long)le64_to_cpu(rec.r_cpos),
			     set_len);
			ret = ocfs2_insert_refcount_rec(handle, ci, ref_root_bh,
							ref_leaf_bh,
2049 2050
							&rec, index,
							merge, meta_ac);
2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061
			if (ret) {
				mlog_errno(ret);
				goto out;
			}
		} else  {
			set_len = min((u64)(cpos + len),
				      le64_to_cpu(rec.r_cpos) + set_len) - cpos;
			rec.r_cpos = cpu_to_le64(cpos);
			rec.r_clusters = cpu_to_le32(set_len);
			le32_add_cpu(&rec.r_refcount, 1);

T
Tao Ma 已提交
2062
			trace_ocfs2_increase_refcount_split(
2063 2064 2065 2066
			     (unsigned long long)le64_to_cpu(rec.r_cpos),
			     set_len, le32_to_cpu(rec.r_refcount));
			ret = ocfs2_split_refcount_rec(handle, ci,
						       ref_root_bh, ref_leaf_bh,
2067
						       &rec, index, merge,
2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084
						       meta_ac, dealloc);
			if (ret) {
				mlog_errno(ret);
				goto out;
			}
		}

		cpos += set_len;
		len -= set_len;
		brelse(ref_leaf_bh);
		ref_leaf_bh = NULL;
	}

out:
	brelse(ref_leaf_bh);
	return ret;
}
2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100

static int ocfs2_remove_refcount_extent(handle_t *handle,
				struct ocfs2_caching_info *ci,
				struct buffer_head *ref_root_bh,
				struct buffer_head *ref_leaf_bh,
				struct ocfs2_alloc_context *meta_ac,
				struct ocfs2_cached_dealloc_ctxt *dealloc)
{
	int ret;
	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
	struct ocfs2_refcount_block *rb =
			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
	struct ocfs2_extent_tree et;

	BUG_ON(rb->rf_records.rl_used);

T
Tao Ma 已提交
2101 2102 2103 2104 2105
	trace_ocfs2_remove_refcount_extent(
		(unsigned long long)ocfs2_metadata_cache_owner(ci),
		(unsigned long long)ref_leaf_bh->b_blocknr,
		le32_to_cpu(rb->rf_cpos));

2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
	ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);
	ret = ocfs2_remove_extent(handle, &et, le32_to_cpu(rb->rf_cpos),
				  1, meta_ac, dealloc);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ocfs2_remove_from_cache(ci, ref_leaf_bh);

	/*
	 * add the freed block to the dealloc so that it will be freed
	 * when we run dealloc.
	 */
	ret = ocfs2_cache_block_dealloc(dealloc, EXTENT_ALLOC_SYSTEM_INODE,
					le16_to_cpu(rb->rf_suballoc_slot),
2122
					le64_to_cpu(rb->rf_suballoc_loc),
2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147
					le64_to_cpu(rb->rf_blkno),
					le16_to_cpu(rb->rf_suballoc_bit));
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_journal_access_rb(handle, ci, ref_root_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;

	le32_add_cpu(&rb->rf_clusters, -1);

	/*
	 * check whether we need to restore the root refcount block if
	 * there is no leaf extent block at atll.
	 */
	if (!rb->rf_list.l_next_free_rec) {
		BUG_ON(rb->rf_clusters);

T
Tao Ma 已提交
2148
		trace_ocfs2_restore_refcount_block(
2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165
		     (unsigned long long)ref_root_bh->b_blocknr);

		rb->rf_flags = 0;
		rb->rf_parent = 0;
		rb->rf_cpos = 0;
		memset(&rb->rf_records, 0, sb->s_blocksize -
		       offsetof(struct ocfs2_refcount_block, rf_records));
		rb->rf_records.rl_count =
				cpu_to_le16(ocfs2_refcount_recs_per_rb(sb));
	}

	ocfs2_journal_dirty(handle, ref_root_bh);

out:
	return ret;
}

2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177
int ocfs2_increase_refcount(handle_t *handle,
			    struct ocfs2_caching_info *ci,
			    struct buffer_head *ref_root_bh,
			    u64 cpos, u32 len,
			    struct ocfs2_alloc_context *meta_ac,
			    struct ocfs2_cached_dealloc_ctxt *dealloc)
{
	return __ocfs2_increase_refcount(handle, ci, ref_root_bh,
					 cpos, len, 1,
					 meta_ac, dealloc);
}

2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194
static int ocfs2_decrease_refcount_rec(handle_t *handle,
				struct ocfs2_caching_info *ci,
				struct buffer_head *ref_root_bh,
				struct buffer_head *ref_leaf_bh,
				int index, u64 cpos, unsigned int len,
				struct ocfs2_alloc_context *meta_ac,
				struct ocfs2_cached_dealloc_ctxt *dealloc)
{
	int ret;
	struct ocfs2_refcount_block *rb =
			(struct ocfs2_refcount_block *)ref_leaf_bh->b_data;
	struct ocfs2_refcount_rec *rec = &rb->rf_records.rl_recs[index];

	BUG_ON(cpos < le64_to_cpu(rec->r_cpos));
	BUG_ON(cpos + len >
	       le64_to_cpu(rec->r_cpos) + le32_to_cpu(rec->r_clusters));

T
Tao Ma 已提交
2195 2196 2197 2198
	trace_ocfs2_decrease_refcount_rec(
		(unsigned long long)ocfs2_metadata_cache_owner(ci),
		(unsigned long long)cpos, len);

2199 2200 2201
	if (cpos == le64_to_cpu(rec->r_cpos) &&
	    len == le32_to_cpu(rec->r_clusters))
		ret = ocfs2_change_refcount_rec(handle, ci,
2202
						ref_leaf_bh, index, 1, -1);
2203 2204 2205 2206 2207 2208 2209 2210 2211
	else {
		struct ocfs2_refcount_rec split = *rec;
		split.r_cpos = cpu_to_le64(cpos);
		split.r_clusters = cpu_to_le32(len);

		le32_add_cpu(&split.r_refcount, -1);

		ret = ocfs2_split_refcount_rec(handle, ci,
					       ref_root_bh, ref_leaf_bh,
2212
					       &split, index, 1,
2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238
					       meta_ac, dealloc);
	}

	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	/* Remove the leaf refcount block if it contains no refcount record. */
	if (!rb->rf_records.rl_used && ref_leaf_bh != ref_root_bh) {
		ret = ocfs2_remove_refcount_extent(handle, ci, ref_root_bh,
						   ref_leaf_bh, meta_ac,
						   dealloc);
		if (ret)
			mlog_errno(ret);
	}

out:
	return ret;
}

static int __ocfs2_decrease_refcount(handle_t *handle,
				     struct ocfs2_caching_info *ci,
				     struct buffer_head *ref_root_bh,
				     u64 cpos, u32 len,
				     struct ocfs2_alloc_context *meta_ac,
T
Tao Ma 已提交
2239 2240
				     struct ocfs2_cached_dealloc_ctxt *dealloc,
				     int delete)
2241 2242 2243 2244 2245 2246 2247
{
	int ret = 0, index = 0;
	struct ocfs2_refcount_rec rec;
	unsigned int r_count = 0, r_len;
	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
	struct buffer_head *ref_leaf_bh = NULL;

T
Tao Ma 已提交
2248 2249 2250
	trace_ocfs2_decrease_refcount(
		(unsigned long long)ocfs2_metadata_cache_owner(ci),
		(unsigned long long)cpos, len, delete);
2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262

	while (len) {
		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
					     cpos, len, &rec, &index,
					     &ref_leaf_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		r_count = le32_to_cpu(rec.r_refcount);
		BUG_ON(r_count == 0);
T
Tao Ma 已提交
2263 2264
		if (!delete)
			BUG_ON(r_count > 1);
2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277

		r_len = min((u64)(cpos + len), le64_to_cpu(rec.r_cpos) +
			      le32_to_cpu(rec.r_clusters)) - cpos;

		ret = ocfs2_decrease_refcount_rec(handle, ci, ref_root_bh,
						  ref_leaf_bh, index,
						  cpos, r_len,
						  meta_ac, dealloc);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

T
Tao Ma 已提交
2278
		if (le32_to_cpu(rec.r_refcount) == 1 && delete) {
2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302
			ret = ocfs2_cache_cluster_dealloc(dealloc,
					  ocfs2_clusters_to_blocks(sb, cpos),
							  r_len);
			if (ret) {
				mlog_errno(ret);
				goto out;
			}
		}

		cpos += r_len;
		len -= r_len;
		brelse(ref_leaf_bh);
		ref_leaf_bh = NULL;
	}

out:
	brelse(ref_leaf_bh);
	return ret;
}

/* Caller must hold refcount tree lock. */
int ocfs2_decrease_refcount(struct inode *inode,
			    handle_t *handle, u32 cpos, u32 len,
			    struct ocfs2_alloc_context *meta_ac,
T
Tao Ma 已提交
2303 2304
			    struct ocfs2_cached_dealloc_ctxt *dealloc,
			    int delete)
2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333
{
	int ret;
	u64 ref_blkno;
	struct ocfs2_inode_info *oi = OCFS2_I(inode);
	struct buffer_head *ref_root_bh = NULL;
	struct ocfs2_refcount_tree *tree;

	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));

	ret = ocfs2_get_refcount_block(inode, &ref_blkno);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_get_refcount_tree(OCFS2_SB(inode->i_sb), ref_blkno, &tree);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_read_refcount_block(&tree->rf_ci, tree->rf_blkno,
					&ref_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = __ocfs2_decrease_refcount(handle, &tree->rf_ci, ref_root_bh,
T
Tao Ma 已提交
2334
					cpos, len, meta_ac, dealloc, delete);
2335 2336 2337 2338 2339 2340
	if (ret)
		mlog_errno(ret);
out:
	brelse(ref_root_bh);
	return ret;
}
2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359

/*
 * Mark the already-existing extent at cpos as refcounted for len clusters.
 * This adds the refcount extent flag.
 *
 * If the existing extent is larger than the request, initiate a
 * split. An attempt will be made at merging with adjacent extents.
 *
 * The caller is responsible for passing down meta_ac if we'll need it.
 */
static int ocfs2_mark_extent_refcounted(struct inode *inode,
				struct ocfs2_extent_tree *et,
				handle_t *handle, u32 cpos,
				u32 len, u32 phys,
				struct ocfs2_alloc_context *meta_ac,
				struct ocfs2_cached_dealloc_ctxt *dealloc)
{
	int ret;

T
Tao Ma 已提交
2360 2361
	trace_ocfs2_mark_extent_refcounted(OCFS2_I(inode)->ip_blkno,
					   cpos, len, phys);
2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379

	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb))) {
		ocfs2_error(inode->i_sb, "Inode %lu want to use refcount "
			    "tree, but the feature bit is not set in the "
			    "super block.", inode->i_ino);
		ret = -EROFS;
		goto out;
	}

	ret = ocfs2_change_extent_flag(handle, et, cpos,
				       len, phys, meta_ac, dealloc,
				       OCFS2_EXT_REFCOUNTED, 0);
	if (ret)
		mlog_errno(ret);

out:
	return ret;
}
2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417

/*
 * Given some contiguous physical clusters, calculate what we need
 * for modifying their refcount.
 */
static int ocfs2_calc_refcount_meta_credits(struct super_block *sb,
					    struct ocfs2_caching_info *ci,
					    struct buffer_head *ref_root_bh,
					    u64 start_cpos,
					    u32 clusters,
					    int *meta_add,
					    int *credits)
{
	int ret = 0, index, ref_blocks = 0, recs_add = 0;
	u64 cpos = start_cpos;
	struct ocfs2_refcount_block *rb;
	struct ocfs2_refcount_rec rec;
	struct buffer_head *ref_leaf_bh = NULL, *prev_bh = NULL;
	u32 len;

	while (clusters) {
		ret = ocfs2_get_refcount_rec(ci, ref_root_bh,
					     cpos, clusters, &rec,
					     &index, &ref_leaf_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		if (ref_leaf_bh != prev_bh) {
			/*
			 * Now we encounter a new leaf block, so calculate
			 * whether we need to extend the old leaf.
			 */
			if (prev_bh) {
				rb = (struct ocfs2_refcount_block *)
							prev_bh->b_data;

A
Al Viro 已提交
2418
				if (le16_to_cpu(rb->rf_records.rl_used) +
2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430
				    recs_add >
				    le16_to_cpu(rb->rf_records.rl_count))
					ref_blocks++;
			}

			recs_add = 0;
			*credits += 1;
			brelse(prev_bh);
			prev_bh = ref_leaf_bh;
			get_bh(prev_bh);
		}

T
Tao Ma 已提交
2431 2432 2433 2434 2435
		trace_ocfs2_calc_refcount_meta_credits_iterate(
				recs_add, (unsigned long long)cpos, clusters,
				(unsigned long long)le64_to_cpu(rec.r_cpos),
				le32_to_cpu(rec.r_clusters),
				le32_to_cpu(rec.r_refcount), index);
2436 2437 2438 2439 2440 2441 2442

		len = min((u64)cpos + clusters, le64_to_cpu(rec.r_cpos) +
			  le32_to_cpu(rec.r_clusters)) - cpos;
		/*
		 * We record all the records which will be inserted to the
		 * same refcount block, so that we can tell exactly whether
		 * we need a new refcount block or not.
2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456
		 *
		 * If we will insert a new one, this is easy and only happens
		 * during adding refcounted flag to the extent, so we don't
		 * have a chance of spliting. We just need one record.
		 *
		 * If the refcount rec already exists, that would be a little
		 * complicated. we may have to:
		 * 1) split at the beginning if the start pos isn't aligned.
		 *    we need 1 more record in this case.
		 * 2) split int the end if the end pos isn't aligned.
		 *    we need 1 more record in this case.
		 * 3) split in the middle because of file system fragmentation.
		 *    we need 2 more records in this case(we can't detect this
		 *    beforehand, so always think of the worst case).
2457 2458
		 */
		if (rec.r_refcount) {
2459
			recs_add += 2;
2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480
			/* Check whether we need a split at the beginning. */
			if (cpos == start_cpos &&
			    cpos != le64_to_cpu(rec.r_cpos))
				recs_add++;

			/* Check whether we need a split in the end. */
			if (cpos + clusters < le64_to_cpu(rec.r_cpos) +
			    le32_to_cpu(rec.r_clusters))
				recs_add++;
		} else
			recs_add++;

		brelse(ref_leaf_bh);
		ref_leaf_bh = NULL;
		clusters -= len;
		cpos += len;
	}

	if (prev_bh) {
		rb = (struct ocfs2_refcount_block *)prev_bh->b_data;

A
Al Viro 已提交
2481
		if (le16_to_cpu(rb->rf_records.rl_used) + recs_add >
2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507
		    le16_to_cpu(rb->rf_records.rl_count))
			ref_blocks++;

		*credits += 1;
	}

	if (!ref_blocks)
		goto out;

	*meta_add += ref_blocks;
	*credits += ref_blocks;

	/*
	 * So we may need ref_blocks to insert into the tree.
	 * That also means we need to change the b-tree and add that number
	 * of records since we never merge them.
	 * We need one more block for expansion since the new created leaf
	 * block is also full and needs split.
	 */
	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
	if (le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL) {
		struct ocfs2_extent_tree et;

		ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);
		*meta_add += ocfs2_extend_meta_needed(et.et_root_el);
		*credits += ocfs2_calc_extend_credits(sb,
2508
						      et.et_root_el);
2509 2510 2511 2512 2513 2514
	} else {
		*credits += OCFS2_EXPAND_REFCOUNT_TREE_CREDITS;
		*meta_add += 1;
	}

out:
T
Tao Ma 已提交
2515 2516 2517 2518

	trace_ocfs2_calc_refcount_meta_credits(
		(unsigned long long)start_cpos, clusters,
		*meta_add, *credits);
2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529
	brelse(ref_leaf_bh);
	brelse(prev_bh);
	return ret;
}

/*
 * For refcount tree, we will decrease some contiguous clusters
 * refcount count, so just go through it to see how many blocks
 * we gonna touch and whether we need to create new blocks.
 *
 * Normally the refcount blocks store these refcount should be
2530
 * contiguous also, so that we can get the number easily.
2531 2532
 * We will at most add split 2 refcount records and 2 more
 * refcount blocks, so just check it in a rough way.
2533 2534 2535 2536
 *
 * Caller must hold refcount tree lock.
 */
int ocfs2_prepare_refcount_change_for_del(struct inode *inode,
2537
					  u64 refcount_loc,
2538 2539 2540
					  u64 phys_blkno,
					  u32 clusters,
					  int *credits,
2541
					  int *ref_blocks)
2542
{
2543
	int ret;
2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559
	struct ocfs2_inode_info *oi = OCFS2_I(inode);
	struct buffer_head *ref_root_bh = NULL;
	struct ocfs2_refcount_tree *tree;
	u64 start_cpos = ocfs2_blocks_to_clusters(inode->i_sb, phys_blkno);

	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb))) {
		ocfs2_error(inode->i_sb, "Inode %lu want to use refcount "
			    "tree, but the feature bit is not set in the "
			    "super block.", inode->i_ino);
		ret = -EROFS;
		goto out;
	}

	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));

	ret = ocfs2_get_refcount_tree(OCFS2_SB(inode->i_sb),
2560
				      refcount_loc, &tree);
2561 2562 2563 2564 2565
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

2566
	ret = ocfs2_read_refcount_block(&tree->rf_ci, refcount_loc,
2567 2568 2569 2570 2571 2572 2573 2574 2575 2576
					&ref_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_calc_refcount_meta_credits(inode->i_sb,
					       &tree->rf_ci,
					       ref_root_bh,
					       start_cpos, clusters,
2577
					       ref_blocks, credits);
2578 2579 2580 2581 2582
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

T
Tao Ma 已提交
2583
	trace_ocfs2_prepare_refcount_change_for_del(*ref_blocks, *credits);
2584 2585 2586 2587 2588

out:
	brelse(ref_root_bh);
	return ret;
}
T
Tao Ma 已提交
2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640

#define	MAX_CONTIG_BYTES	1048576

static inline unsigned int ocfs2_cow_contig_clusters(struct super_block *sb)
{
	return ocfs2_clusters_for_bytes(sb, MAX_CONTIG_BYTES);
}

static inline unsigned int ocfs2_cow_contig_mask(struct super_block *sb)
{
	return ~(ocfs2_cow_contig_clusters(sb) - 1);
}

/*
 * Given an extent that starts at 'start' and an I/O that starts at 'cpos',
 * find an offset (start + (n * contig_clusters)) that is closest to cpos
 * while still being less than or equal to it.
 *
 * The goal is to break the extent at a multiple of contig_clusters.
 */
static inline unsigned int ocfs2_cow_align_start(struct super_block *sb,
						 unsigned int start,
						 unsigned int cpos)
{
	BUG_ON(start > cpos);

	return start + ((cpos - start) & ocfs2_cow_contig_mask(sb));
}

/*
 * Given a cluster count of len, pad it out so that it is a multiple
 * of contig_clusters.
 */
static inline unsigned int ocfs2_cow_align_length(struct super_block *sb,
						  unsigned int len)
{
	unsigned int padded =
		(len + (ocfs2_cow_contig_clusters(sb) - 1)) &
		ocfs2_cow_contig_mask(sb);

	/* Did we wrap? */
	if (padded < len)
		padded = UINT_MAX;

	return padded;
}

/*
 * Calculate out the start and number of virtual clusters we need to to CoW.
 *
 * cpos is vitual start cluster position we want to do CoW in a
 * file and write_len is the cluster length.
2641
 * max_cpos is the place where we want to stop CoW intentionally.
T
Tao Ma 已提交
2642 2643 2644 2645 2646 2647
 *
 * Normal we will start CoW from the beginning of extent record cotaining cpos.
 * We try to break up extents on boundaries of MAX_CONTIG_BYTES so that we
 * get good I/O from the resulting extent tree.
 */
static int ocfs2_refcount_cal_cow_clusters(struct inode *inode,
2648
					   struct ocfs2_extent_list *el,
T
Tao Ma 已提交
2649 2650
					   u32 cpos,
					   u32 write_len,
2651
					   u32 max_cpos,
T
Tao Ma 已提交
2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663
					   u32 *cow_start,
					   u32 *cow_len)
{
	int ret = 0;
	int tree_height = le16_to_cpu(el->l_tree_depth), i;
	struct buffer_head *eb_bh = NULL;
	struct ocfs2_extent_block *eb = NULL;
	struct ocfs2_extent_rec *rec;
	unsigned int want_clusters, rec_end = 0;
	int contig_clusters = ocfs2_cow_contig_clusters(inode->i_sb);
	int leaf_clusters;

2664 2665
	BUG_ON(cpos + write_len > max_cpos);

T
Tao Ma 已提交
2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709
	if (tree_height > 0) {
		ret = ocfs2_find_leaf(INODE_CACHE(inode), el, cpos, &eb_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		eb = (struct ocfs2_extent_block *) eb_bh->b_data;
		el = &eb->h_list;

		if (el->l_tree_depth) {
			ocfs2_error(inode->i_sb,
				    "Inode %lu has non zero tree depth in "
				    "leaf block %llu\n", inode->i_ino,
				    (unsigned long long)eb_bh->b_blocknr);
			ret = -EROFS;
			goto out;
		}
	}

	*cow_len = 0;
	for (i = 0; i < le16_to_cpu(el->l_next_free_rec); i++) {
		rec = &el->l_recs[i];

		if (ocfs2_is_empty_extent(rec)) {
			mlog_bug_on_msg(i != 0, "Inode %lu has empty record in "
					"index %d\n", inode->i_ino, i);
			continue;
		}

		if (le32_to_cpu(rec->e_cpos) +
		    le16_to_cpu(rec->e_leaf_clusters) <= cpos)
			continue;

		if (*cow_len == 0) {
			/*
			 * We should find a refcounted record in the
			 * first pass.
			 */
			BUG_ON(!(rec->e_flags & OCFS2_EXT_REFCOUNTED));
			*cow_start = le32_to_cpu(rec->e_cpos);
		}

		/*
2710 2711
		 * If we encounter a hole, a non-refcounted record or
		 * pass the max_cpos, stop the search.
T
Tao Ma 已提交
2712 2713
		 */
		if ((!(rec->e_flags & OCFS2_EXT_REFCOUNTED)) ||
2714 2715
		    (*cow_len && rec_end != le32_to_cpu(rec->e_cpos)) ||
		    (max_cpos <= le32_to_cpu(rec->e_cpos)))
T
Tao Ma 已提交
2716 2717 2718 2719
			break;

		leaf_clusters = le16_to_cpu(rec->e_leaf_clusters);
		rec_end = le32_to_cpu(rec->e_cpos) + leaf_clusters;
2720 2721 2722 2723
		if (rec_end > max_cpos) {
			rec_end = max_cpos;
			leaf_clusters = rec_end - le32_to_cpu(rec->e_cpos);
		}
T
Tao Ma 已提交
2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878

		/*
		 * How many clusters do we actually need from
		 * this extent?  First we see how many we actually
		 * need to complete the write.  If that's smaller
		 * than contig_clusters, we try for contig_clusters.
		 */
		if (!*cow_len)
			want_clusters = write_len;
		else
			want_clusters = (cpos + write_len) -
				(*cow_start + *cow_len);
		if (want_clusters < contig_clusters)
			want_clusters = contig_clusters;

		/*
		 * If the write does not cover the whole extent, we
		 * need to calculate how we're going to split the extent.
		 * We try to do it on contig_clusters boundaries.
		 *
		 * Any extent smaller than contig_clusters will be
		 * CoWed in its entirety.
		 */
		if (leaf_clusters <= contig_clusters)
			*cow_len += leaf_clusters;
		else if (*cow_len || (*cow_start == cpos)) {
			/*
			 * This extent needs to be CoW'd from its
			 * beginning, so all we have to do is compute
			 * how many clusters to grab.  We align
			 * want_clusters to the edge of contig_clusters
			 * to get better I/O.
			 */
			want_clusters = ocfs2_cow_align_length(inode->i_sb,
							       want_clusters);

			if (leaf_clusters < want_clusters)
				*cow_len += leaf_clusters;
			else
				*cow_len += want_clusters;
		} else if ((*cow_start + contig_clusters) >=
			   (cpos + write_len)) {
			/*
			 * Breaking off contig_clusters at the front
			 * of the extent will cover our write.  That's
			 * easy.
			 */
			*cow_len = contig_clusters;
		} else if ((rec_end - cpos) <= contig_clusters) {
			/*
			 * Breaking off contig_clusters at the tail of
			 * this extent will cover cpos.
			 */
			*cow_start = rec_end - contig_clusters;
			*cow_len = contig_clusters;
		} else if ((rec_end - cpos) <= want_clusters) {
			/*
			 * While we can't fit the entire write in this
			 * extent, we know that the write goes from cpos
			 * to the end of the extent.  Break that off.
			 * We try to break it at some multiple of
			 * contig_clusters from the front of the extent.
			 * Failing that (ie, cpos is within
			 * contig_clusters of the front), we'll CoW the
			 * entire extent.
			 */
			*cow_start = ocfs2_cow_align_start(inode->i_sb,
							   *cow_start, cpos);
			*cow_len = rec_end - *cow_start;
		} else {
			/*
			 * Ok, the entire write lives in the middle of
			 * this extent.  Let's try to slice the extent up
			 * nicely.  Optimally, our CoW region starts at
			 * m*contig_clusters from the beginning of the
			 * extent and goes for n*contig_clusters,
			 * covering the entire write.
			 */
			*cow_start = ocfs2_cow_align_start(inode->i_sb,
							   *cow_start, cpos);

			want_clusters = (cpos + write_len) - *cow_start;
			want_clusters = ocfs2_cow_align_length(inode->i_sb,
							       want_clusters);
			if (*cow_start + want_clusters <= rec_end)
				*cow_len = want_clusters;
			else
				*cow_len = rec_end - *cow_start;
		}

		/* Have we covered our entire write yet? */
		if ((*cow_start + *cow_len) >= (cpos + write_len))
			break;

		/*
		 * If we reach the end of the extent block and don't get enough
		 * clusters, continue with the next extent block if possible.
		 */
		if (i + 1 == le16_to_cpu(el->l_next_free_rec) &&
		    eb && eb->h_next_leaf_blk) {
			brelse(eb_bh);
			eb_bh = NULL;

			ret = ocfs2_read_extent_block(INODE_CACHE(inode),
					       le64_to_cpu(eb->h_next_leaf_blk),
					       &eb_bh);
			if (ret) {
				mlog_errno(ret);
				goto out;
			}

			eb = (struct ocfs2_extent_block *) eb_bh->b_data;
			el = &eb->h_list;
			i = -1;
		}
	}

out:
	brelse(eb_bh);
	return ret;
}

/*
 * Prepare meta_ac, data_ac and calculate credits when we want to add some
 * num_clusters in data_tree "et" and change the refcount for the old
 * clusters(starting form p_cluster) in the refcount tree.
 *
 * Note:
 * 1. since we may split the old tree, so we at most will need num_clusters + 2
 *    more new leaf records.
 * 2. In some case, we may not need to reserve new clusters(e.g, reflink), so
 *    just give data_ac = NULL.
 */
static int ocfs2_lock_refcount_allocators(struct super_block *sb,
					u32 p_cluster, u32 num_clusters,
					struct ocfs2_extent_tree *et,
					struct ocfs2_caching_info *ref_ci,
					struct buffer_head *ref_root_bh,
					struct ocfs2_alloc_context **meta_ac,
					struct ocfs2_alloc_context **data_ac,
					int *credits)
{
	int ret = 0, meta_add = 0;
	int num_free_extents = ocfs2_num_free_extents(OCFS2_SB(sb), et);

	if (num_free_extents < 0) {
		ret = num_free_extents;
		mlog_errno(ret);
		goto out;
	}

	if (num_free_extents < num_clusters + 2)
		meta_add =
			ocfs2_extend_meta_needed(et->et_root_el);

2879
	*credits += ocfs2_calc_extend_credits(sb, et->et_root_el);
T
Tao Ma 已提交
2880 2881 2882 2883 2884 2885 2886 2887 2888

	ret = ocfs2_calc_refcount_meta_credits(sb, ref_ci, ref_root_bh,
					       p_cluster, num_clusters,
					       &meta_add, credits);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

T
Tao Ma 已提交
2889
	trace_ocfs2_lock_refcount_allocators(meta_add, *credits);
T
Tao Ma 已提交
2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923
	ret = ocfs2_reserve_new_metadata_blocks(OCFS2_SB(sb), meta_add,
						meta_ac);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	if (data_ac) {
		ret = ocfs2_reserve_clusters(OCFS2_SB(sb), num_clusters,
					     data_ac);
		if (ret)
			mlog_errno(ret);
	}

out:
	if (ret) {
		if (*meta_ac) {
			ocfs2_free_alloc_context(*meta_ac);
			*meta_ac = NULL;
		}
	}

	return ret;
}

static int ocfs2_clear_cow_buffer(handle_t *handle, struct buffer_head *bh)
{
	BUG_ON(buffer_dirty(bh));

	clear_buffer_mapped(bh);

	return 0;
}

2924
int ocfs2_duplicate_clusters_by_page(handle_t *handle,
2925
				     struct inode *inode,
2926 2927
				     u32 cpos, u32 old_cluster,
				     u32 new_cluster, u32 new_len)
T
Tao Ma 已提交
2928 2929
{
	int ret = 0, partial;
2930
	struct super_block *sb = inode->i_sb;
T
Tao Ma 已提交
2931 2932 2933
	u64 new_block = ocfs2_clusters_to_blocks(sb, new_cluster);
	struct page *page;
	pgoff_t page_index;
T
Tao Ma 已提交
2934
	unsigned int from, to, readahead_pages;
T
Tao Ma 已提交
2935
	loff_t offset, end, map_end;
2936
	struct address_space *mapping = inode->i_mapping;
T
Tao Ma 已提交
2937

T
Tao Ma 已提交
2938 2939
	trace_ocfs2_duplicate_clusters_by_page(cpos, old_cluster,
					       new_cluster, new_len);
T
Tao Ma 已提交
2940

T
Tao Ma 已提交
2941 2942 2943
	readahead_pages =
		(ocfs2_cow_contig_clusters(sb) <<
		 OCFS2_SB(sb)->s_clustersize_bits) >> PAGE_CACHE_SHIFT;
T
Tao Ma 已提交
2944 2945
	offset = ((loff_t)cpos) << OCFS2_SB(sb)->s_clustersize_bits;
	end = offset + (new_len << OCFS2_SB(sb)->s_clustersize_bits);
2946 2947 2948 2949
	/*
	 * We only duplicate pages until we reach the page contains i_size - 1.
	 * So trim 'end' to i_size.
	 */
2950 2951
	if (end > i_size_read(inode))
		end = i_size_read(inode);
T
Tao Ma 已提交
2952 2953 2954

	while (offset < end) {
		page_index = offset >> PAGE_CACHE_SHIFT;
2955
		map_end = ((loff_t)page_index + 1) << PAGE_CACHE_SHIFT;
T
Tao Ma 已提交
2956 2957 2958 2959 2960 2961 2962 2963 2964
		if (map_end > end)
			map_end = end;

		/* from, to is the offset within the page. */
		from = offset & (PAGE_CACHE_SIZE - 1);
		to = PAGE_CACHE_SIZE;
		if (map_end & (PAGE_CACHE_SIZE - 1))
			to = map_end & (PAGE_CACHE_SIZE - 1);

2965
		page = find_or_create_page(mapping, page_index, GFP_NOFS);
2966 2967 2968 2969 2970
		if (!page) {
			ret = -ENOMEM;
			mlog_errno(ret);
			break;
		}
T
Tao Ma 已提交
2971

2972 2973 2974 2975 2976 2977
		/*
		 * In case PAGE_CACHE_SIZE <= CLUSTER_SIZE, This page
		 * can't be dirtied before we CoW it out.
		 */
		if (PAGE_CACHE_SIZE <= OCFS2_SB(sb)->s_clustersize)
			BUG_ON(PageDirty(page));
T
Tao Ma 已提交
2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997

		if (!PageUptodate(page)) {
			ret = block_read_full_page(page, ocfs2_get_block);
			if (ret) {
				mlog_errno(ret);
				goto unlock;
			}
			lock_page(page);
		}

		if (page_has_buffers(page)) {
			ret = walk_page_buffers(handle, page_buffers(page),
						from, to, &partial,
						ocfs2_clear_cow_buffer);
			if (ret) {
				mlog_errno(ret);
				goto unlock;
			}
		}

2998 2999
		ocfs2_map_and_dirty_page(inode,
					 handle, from, to,
T
Tao Ma 已提交
3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013
					 page, 0, &new_block);
		mark_page_accessed(page);
unlock:
		unlock_page(page);
		page_cache_release(page);
		page = NULL;
		offset = map_end;
		if (ret)
			break;
	}

	return ret;
}

3014
int ocfs2_duplicate_clusters_by_jbd(handle_t *handle,
3015
				    struct inode *inode,
3016 3017
				    u32 cpos, u32 old_cluster,
				    u32 new_cluster, u32 new_len)
T
Tao Ma 已提交
3018 3019
{
	int ret = 0;
3020 3021
	struct super_block *sb = inode->i_sb;
	struct ocfs2_caching_info *ci = INODE_CACHE(inode);
T
Tao Ma 已提交
3022 3023 3024 3025 3026 3027 3028
	int i, blocks = ocfs2_clusters_to_blocks(sb, new_len);
	u64 old_block = ocfs2_clusters_to_blocks(sb, old_cluster);
	u64 new_block = ocfs2_clusters_to_blocks(sb, new_cluster);
	struct ocfs2_super *osb = OCFS2_SB(sb);
	struct buffer_head *old_bh = NULL;
	struct buffer_head *new_bh = NULL;

T
Tao Ma 已提交
3029 3030
	trace_ocfs2_duplicate_clusters_by_page(cpos, old_cluster,
					       new_cluster, new_len);
T
Tao Ma 已提交
3031 3032 3033 3034

	for (i = 0; i < blocks; i++, old_block++, new_block++) {
		new_bh = sb_getblk(osb->sb, new_block);
		if (new_bh == NULL) {
3035
			ret = -ENOMEM;
T
Tao Ma 已提交
3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055
			mlog_errno(ret);
			break;
		}

		ocfs2_set_new_buffer_uptodate(ci, new_bh);

		ret = ocfs2_read_block(ci, old_block, &old_bh, NULL);
		if (ret) {
			mlog_errno(ret);
			break;
		}

		ret = ocfs2_journal_access(handle, ci, new_bh,
					   OCFS2_JOURNAL_ACCESS_CREATE);
		if (ret) {
			mlog_errno(ret);
			break;
		}

		memcpy(new_bh->b_data, old_bh->b_data, sb->s_blocksize);
3056
		ocfs2_journal_dirty(handle, new_bh);
T
Tao Ma 已提交
3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068

		brelse(new_bh);
		brelse(old_bh);
		new_bh = NULL;
		old_bh = NULL;
	}

	brelse(new_bh);
	brelse(old_bh);
	return ret;
}

T
Tao Ma 已提交
3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082
static int ocfs2_clear_ext_refcount(handle_t *handle,
				    struct ocfs2_extent_tree *et,
				    u32 cpos, u32 p_cluster, u32 len,
				    unsigned int ext_flags,
				    struct ocfs2_alloc_context *meta_ac,
				    struct ocfs2_cached_dealloc_ctxt *dealloc)
{
	int ret, index;
	struct ocfs2_extent_rec replace_rec;
	struct ocfs2_path *path = NULL;
	struct ocfs2_extent_list *el;
	struct super_block *sb = ocfs2_metadata_cache_get_super(et->et_ci);
	u64 ino = ocfs2_metadata_cache_owner(et->et_ci);

T
Tao Ma 已提交
3083 3084
	trace_ocfs2_clear_ext_refcount((unsigned long long)ino,
				       cpos, len, p_cluster, ext_flags);
T
Tao Ma 已提交
3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109

	memset(&replace_rec, 0, sizeof(replace_rec));
	replace_rec.e_cpos = cpu_to_le32(cpos);
	replace_rec.e_leaf_clusters = cpu_to_le16(len);
	replace_rec.e_blkno = cpu_to_le64(ocfs2_clusters_to_blocks(sb,
								   p_cluster));
	replace_rec.e_flags = ext_flags;
	replace_rec.e_flags &= ~OCFS2_EXT_REFCOUNTED;

	path = ocfs2_new_path_from_et(et);
	if (!path) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_find_path(et->et_ci, path, cpos);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	el = path_leaf_el(path);

	index = ocfs2_search_extent_list(el, cpos);
3110
	if (index == -1) {
T
Tao Ma 已提交
3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135
		ocfs2_error(sb,
			    "Inode %llu has an extent at cpos %u which can no "
			    "longer be found.\n",
			    (unsigned long long)ino, cpos);
		ret = -EROFS;
		goto out;
	}

	ret = ocfs2_split_extent(handle, et, path, index,
				 &replace_rec, meta_ac, dealloc);
	if (ret)
		mlog_errno(ret);

out:
	ocfs2_free_path(path);
	return ret;
}

static int ocfs2_replace_clusters(handle_t *handle,
				  struct ocfs2_cow_context *context,
				  u32 cpos, u32 old,
				  u32 new, u32 len,
				  unsigned int ext_flags)
{
	int ret;
3136
	struct ocfs2_caching_info *ci = context->data_et.et_ci;
T
Tao Ma 已提交
3137 3138
	u64 ino = ocfs2_metadata_cache_owner(ci);

T
Tao Ma 已提交
3139 3140
	trace_ocfs2_replace_clusters((unsigned long long)ino,
				     cpos, old, new, len, ext_flags);
T
Tao Ma 已提交
3141 3142 3143

	/*If the old clusters is unwritten, no need to duplicate. */
	if (!(ext_flags & OCFS2_EXT_UNWRITTEN)) {
3144
		ret = context->cow_duplicate_clusters(handle, context->inode,
3145
						      cpos, old, new, len);
T
Tao Ma 已提交
3146 3147 3148 3149 3150 3151
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	}

3152
	ret = ocfs2_clear_ext_refcount(handle, &context->data_et,
T
Tao Ma 已提交
3153 3154 3155 3156 3157 3158 3159 3160
				       cpos, new, len, ext_flags,
				       context->meta_ac, &context->dealloc);
	if (ret)
		mlog_errno(ret);
out:
	return ret;
}

3161 3162 3163
int ocfs2_cow_sync_writeback(struct super_block *sb,
			     struct inode *inode,
			     u32 cpos, u32 num_clusters)
T
Tao Ma 已提交
3164 3165 3166 3167 3168 3169
{
	int ret = 0;
	loff_t offset, end, map_end;
	pgoff_t page_index;
	struct page *page;

3170
	if (ocfs2_should_order_data(inode))
T
Tao Ma 已提交
3171 3172 3173 3174 3175
		return 0;

	offset = ((loff_t)cpos) << OCFS2_SB(sb)->s_clustersize_bits;
	end = offset + (num_clusters << OCFS2_SB(sb)->s_clustersize_bits);

3176
	ret = filemap_fdatawrite_range(inode->i_mapping,
T
Tao Ma 已提交
3177 3178 3179 3180 3181 3182 3183 3184
				       offset, end - 1);
	if (ret < 0) {
		mlog_errno(ret);
		return ret;
	}

	while (offset < end) {
		page_index = offset >> PAGE_CACHE_SHIFT;
3185
		map_end = ((loff_t)page_index + 1) << PAGE_CACHE_SHIFT;
T
Tao Ma 已提交
3186 3187 3188
		if (map_end > end)
			map_end = end;

3189
		page = find_or_create_page(inode->i_mapping,
3190
					   page_index, GFP_NOFS);
T
Tao Ma 已提交
3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210
		BUG_ON(!page);

		wait_on_page_writeback(page);
		if (PageError(page)) {
			ret = -EIO;
			mlog_errno(ret);
		} else
			mark_page_accessed(page);

		unlock_page(page);
		page_cache_release(page);
		page = NULL;
		offset = map_end;
		if (ret)
			break;
	}

	return ret;
}

3211 3212 3213 3214 3215 3216 3217 3218 3219
static int ocfs2_di_get_clusters(struct ocfs2_cow_context *context,
				 u32 v_cluster, u32 *p_cluster,
				 u32 *num_clusters,
				 unsigned int *extent_flags)
{
	return ocfs2_get_clusters(context->inode, v_cluster, p_cluster,
				  num_clusters, extent_flags);
}

T
Tao Ma 已提交
3220 3221 3222 3223 3224
static int ocfs2_make_clusters_writable(struct super_block *sb,
					struct ocfs2_cow_context *context,
					u32 cpos, u32 p_cluster,
					u32 num_clusters, unsigned int e_flags)
{
T
Tao Ma 已提交
3225
	int ret, delete, index, credits =  0;
3226
	u32 new_bit, new_len, orig_num_clusters;
T
Tao Ma 已提交
3227
	unsigned int set_len;
T
Tao Ma 已提交
3228 3229
	struct ocfs2_super *osb = OCFS2_SB(sb);
	handle_t *handle;
T
Tao Ma 已提交
3230
	struct buffer_head *ref_leaf_bh = NULL;
3231
	struct ocfs2_caching_info *ref_ci = &context->ref_tree->rf_ci;
T
Tao Ma 已提交
3232 3233
	struct ocfs2_refcount_rec rec;

T
Tao Ma 已提交
3234 3235
	trace_ocfs2_make_clusters_writable(cpos, p_cluster,
					   num_clusters, e_flags);
T
Tao Ma 已提交
3236 3237

	ret = ocfs2_lock_refcount_allocators(sb, p_cluster, num_clusters,
3238 3239
					     &context->data_et,
					     ref_ci,
T
Tao Ma 已提交
3240 3241 3242 3243 3244 3245 3246 3247
					     context->ref_root_bh,
					     &context->meta_ac,
					     &context->data_ac, &credits);
	if (ret) {
		mlog_errno(ret);
		return ret;
	}

T
Tao Ma 已提交
3248 3249 3250 3251
	if (context->post_refcount)
		credits += context->post_refcount->credits;

	credits += context->extra_credits;
T
Tao Ma 已提交
3252 3253 3254 3255 3256 3257 3258
	handle = ocfs2_start_trans(osb, credits);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out;
	}

3259 3260
	orig_num_clusters = num_clusters;

T
Tao Ma 已提交
3261
	while (num_clusters) {
3262
		ret = ocfs2_get_refcount_rec(ref_ci, context->ref_root_bh,
T
Tao Ma 已提交
3263 3264
					     p_cluster, num_clusters,
					     &rec, &index, &ref_leaf_bh);
T
Tao Ma 已提交
3265 3266 3267 3268 3269
		if (ret) {
			mlog_errno(ret);
			goto out_commit;
		}

T
Tao Ma 已提交
3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283
		BUG_ON(!rec.r_refcount);
		set_len = min((u64)p_cluster + num_clusters,
			      le64_to_cpu(rec.r_cpos) +
			      le32_to_cpu(rec.r_clusters)) - p_cluster;

		/*
		 * There are many different situation here.
		 * 1. If refcount == 1, remove the flag and don't COW.
		 * 2. If refcount > 1, allocate clusters.
		 *    Here we may not allocate r_len once at a time, so continue
		 *    until we reach num_clusters.
		 */
		if (le32_to_cpu(rec.r_refcount) == 1) {
			delete = 0;
3284 3285
			ret = ocfs2_clear_ext_refcount(handle,
						       &context->data_et,
T
Tao Ma 已提交
3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296
						       cpos, p_cluster,
						       set_len, e_flags,
						       context->meta_ac,
						       &context->dealloc);
			if (ret) {
				mlog_errno(ret);
				goto out_commit;
			}
		} else {
			delete = 1;

3297
			ret = __ocfs2_claim_clusters(handle,
T
Tao Ma 已提交
3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315
						     context->data_ac,
						     1, set_len,
						     &new_bit, &new_len);
			if (ret) {
				mlog_errno(ret);
				goto out_commit;
			}

			ret = ocfs2_replace_clusters(handle, context,
						     cpos, p_cluster, new_bit,
						     new_len, e_flags);
			if (ret) {
				mlog_errno(ret);
				goto out_commit;
			}
			set_len = new_len;
		}

3316
		ret = __ocfs2_decrease_refcount(handle, ref_ci,
T
Tao Ma 已提交
3317 3318 3319 3320
						context->ref_root_bh,
						p_cluster, set_len,
						context->meta_ac,
						&context->dealloc, delete);
T
Tao Ma 已提交
3321 3322 3323 3324 3325
		if (ret) {
			mlog_errno(ret);
			goto out_commit;
		}

T
Tao Ma 已提交
3326 3327 3328 3329 3330
		cpos += set_len;
		p_cluster += set_len;
		num_clusters -= set_len;
		brelse(ref_leaf_bh);
		ref_leaf_bh = NULL;
T
Tao Ma 已提交
3331 3332
	}

T
Tao Ma 已提交
3333 3334 3335 3336 3337 3338 3339 3340 3341 3342
	/* handle any post_cow action. */
	if (context->post_refcount && context->post_refcount->func) {
		ret = context->post_refcount->func(context->inode, handle,
						context->post_refcount->para);
		if (ret) {
			mlog_errno(ret);
			goto out_commit;
		}
	}

T
Tao Ma 已提交
3343 3344 3345 3346
	/*
	 * Here we should write the new page out first if we are
	 * in write-back mode.
	 */
T
Tao Ma 已提交
3347
	if (context->get_clusters == ocfs2_di_get_clusters) {
3348
		ret = ocfs2_cow_sync_writeback(sb, context->inode, cpos,
3349
					       orig_num_clusters);
T
Tao Ma 已提交
3350 3351 3352
		if (ret)
			mlog_errno(ret);
	}
T
Tao Ma 已提交
3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365

out_commit:
	ocfs2_commit_trans(osb, handle);

out:
	if (context->data_ac) {
		ocfs2_free_alloc_context(context->data_ac);
		context->data_ac = NULL;
	}
	if (context->meta_ac) {
		ocfs2_free_alloc_context(context->meta_ac);
		context->meta_ac = NULL;
	}
T
Tao Ma 已提交
3366
	brelse(ref_leaf_bh);
T
Tao Ma 已提交
3367 3368 3369 3370

	return ret;
}

3371
static int ocfs2_replace_cow(struct ocfs2_cow_context *context)
T
Tao Ma 已提交
3372 3373
{
	int ret = 0;
3374 3375 3376
	struct inode *inode = context->inode;
	u32 cow_start = context->cow_start, cow_len = context->cow_len;
	u32 p_cluster, num_clusters;
T
Tao Ma 已提交
3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389
	unsigned int ext_flags;
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);

	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb))) {
		ocfs2_error(inode->i_sb, "Inode %lu want to use refcount "
			    "tree, but the feature bit is not set in the "
			    "super block.", inode->i_ino);
		return -EROFS;
	}

	ocfs2_init_dealloc_ctxt(&context->dealloc);

	while (cow_len) {
3390 3391
		ret = context->get_clusters(context, cow_start, &p_cluster,
					    &num_clusters, &ext_flags);
T
Tao Ma 已提交
3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422
		if (ret) {
			mlog_errno(ret);
			break;
		}

		BUG_ON(!(ext_flags & OCFS2_EXT_REFCOUNTED));

		if (cow_len < num_clusters)
			num_clusters = cow_len;

		ret = ocfs2_make_clusters_writable(inode->i_sb, context,
						   cow_start, p_cluster,
						   num_clusters, ext_flags);
		if (ret) {
			mlog_errno(ret);
			break;
		}

		cow_len -= num_clusters;
		cow_start += num_clusters;
	}

	if (ocfs2_dealloc_has_cluster(&context->dealloc)) {
		ocfs2_schedule_truncate_log_flush(osb, 1);
		ocfs2_run_deallocs(osb, &context->dealloc);
	}

	return ret;
}

/*
3423 3424 3425
 * Starting at cpos, try to CoW write_len clusters.  Don't CoW
 * past max_cpos.  This will stop when it runs into a hole or an
 * unrefcounted extent.
T
Tao Ma 已提交
3426 3427 3428
 */
static int ocfs2_refcount_cow_hunk(struct inode *inode,
				   struct buffer_head *di_bh,
3429
				   u32 cpos, u32 write_len, u32 max_cpos)
T
Tao Ma 已提交
3430 3431 3432 3433 3434 3435 3436 3437
{
	int ret;
	u32 cow_start = 0, cow_len = 0;
	struct ocfs2_inode_info *oi = OCFS2_I(inode);
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct buffer_head *ref_root_bh = NULL;
	struct ocfs2_refcount_tree *ref_tree;
3438
	struct ocfs2_cow_context *context = NULL;
T
Tao Ma 已提交
3439 3440 3441

	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));

3442
	ret = ocfs2_refcount_cal_cow_clusters(inode, &di->id2.i_list,
3443
					      cpos, write_len, max_cpos,
T
Tao Ma 已提交
3444 3445 3446 3447 3448
					      &cow_start, &cow_len);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
3449

T
Tao Ma 已提交
3450 3451 3452
	trace_ocfs2_refcount_cow_hunk(OCFS2_I(inode)->ip_blkno,
				      cpos, write_len, max_cpos,
				      cow_start, cow_len);
T
Tao Ma 已提交
3453 3454 3455

	BUG_ON(cow_len == 0);

3456 3457 3458 3459 3460 3461 3462
	context = kzalloc(sizeof(struct ocfs2_cow_context), GFP_NOFS);
	if (!context) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

T
Tao Ma 已提交
3463 3464 3465 3466 3467 3468 3469
	ret = ocfs2_lock_refcount_tree(osb, le64_to_cpu(di->i_refcount_loc),
				       1, &ref_tree, &ref_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481
	context->inode = inode;
	context->cow_start = cow_start;
	context->cow_len = cow_len;
	context->ref_tree = ref_tree;
	context->ref_root_bh = ref_root_bh;
	context->cow_duplicate_clusters = ocfs2_duplicate_clusters_by_page;
	context->get_clusters = ocfs2_di_get_clusters;

	ocfs2_init_dinode_extent_tree(&context->data_et,
				      INODE_CACHE(inode), di_bh);

	ret = ocfs2_replace_cow(context);
T
Tao Ma 已提交
3482 3483 3484
	if (ret)
		mlog_errno(ret);

3485 3486 3487 3488 3489 3490 3491
	/*
	 * truncate the extent map here since no matter whether we meet with
	 * any error during the action, we shouldn't trust cached extent map
	 * any more.
	 */
	ocfs2_extent_map_trunc(inode, cow_start);

T
Tao Ma 已提交
3492 3493 3494
	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
	brelse(ref_root_bh);
out:
3495
	kfree(context);
T
Tao Ma 已提交
3496 3497 3498 3499 3500
	return ret;
}

/*
 * CoW any and all clusters between cpos and cpos+write_len.
3501 3502
 * Don't CoW past max_cpos.  If this returns successfully, all
 * clusters between cpos and cpos+write_len are safe to modify.
T
Tao Ma 已提交
3503 3504 3505
 */
int ocfs2_refcount_cow(struct inode *inode,
		       struct buffer_head *di_bh,
3506
		       u32 cpos, u32 write_len, u32 max_cpos)
T
Tao Ma 已提交
3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523
{
	int ret = 0;
	u32 p_cluster, num_clusters;
	unsigned int ext_flags;

	while (write_len) {
		ret = ocfs2_get_clusters(inode, cpos, &p_cluster,
					 &num_clusters, &ext_flags);
		if (ret) {
			mlog_errno(ret);
			break;
		}

		if (write_len < num_clusters)
			num_clusters = write_len;

		if (ext_flags & OCFS2_EXT_REFCOUNTED) {
3524
			ret = ocfs2_refcount_cow_hunk(inode, di_bh, cpos,
3525
						      num_clusters, max_cpos);
T
Tao Ma 已提交
3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537
			if (ret) {
				mlog_errno(ret);
				break;
			}
		}

		write_len -= num_clusters;
		cpos += num_clusters;
	}

	return ret;
}
3538

T
Tao Ma 已提交
3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600
static int ocfs2_xattr_value_get_clusters(struct ocfs2_cow_context *context,
					  u32 v_cluster, u32 *p_cluster,
					  u32 *num_clusters,
					  unsigned int *extent_flags)
{
	struct inode *inode = context->inode;
	struct ocfs2_xattr_value_root *xv = context->cow_object;

	return ocfs2_xattr_get_clusters(inode, v_cluster, p_cluster,
					num_clusters, &xv->xr_list,
					extent_flags);
}

/*
 * Given a xattr value root, calculate the most meta/credits we need for
 * refcount tree change if we truncate it to 0.
 */
int ocfs2_refcounted_xattr_delete_need(struct inode *inode,
				       struct ocfs2_caching_info *ref_ci,
				       struct buffer_head *ref_root_bh,
				       struct ocfs2_xattr_value_root *xv,
				       int *meta_add, int *credits)
{
	int ret = 0, index, ref_blocks = 0;
	u32 p_cluster, num_clusters;
	u32 cpos = 0, clusters = le32_to_cpu(xv->xr_clusters);
	struct ocfs2_refcount_block *rb;
	struct ocfs2_refcount_rec rec;
	struct buffer_head *ref_leaf_bh = NULL;

	while (cpos < clusters) {
		ret = ocfs2_xattr_get_clusters(inode, cpos, &p_cluster,
					       &num_clusters, &xv->xr_list,
					       NULL);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		cpos += num_clusters;

		while (num_clusters) {
			ret = ocfs2_get_refcount_rec(ref_ci, ref_root_bh,
						     p_cluster, num_clusters,
						     &rec, &index,
						     &ref_leaf_bh);
			if (ret) {
				mlog_errno(ret);
				goto out;
			}

			BUG_ON(!rec.r_refcount);

			rb = (struct ocfs2_refcount_block *)ref_leaf_bh->b_data;

			/*
			 * We really don't know whether the other clusters is in
			 * this refcount block or not, so just take the worst
			 * case that all the clusters are in this block and each
			 * one will split a refcount rec, so totally we need
			 * clusters * 2 new refcount rec.
			 */
A
Al Viro 已提交
3601
			if (le16_to_cpu(rb->rf_records.rl_used) + clusters * 2 >
T
Tao Ma 已提交
3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628
			    le16_to_cpu(rb->rf_records.rl_count))
				ref_blocks++;

			*credits += 1;
			brelse(ref_leaf_bh);
			ref_leaf_bh = NULL;

			if (num_clusters <= le32_to_cpu(rec.r_clusters))
				break;
			else
				num_clusters -= le32_to_cpu(rec.r_clusters);
			p_cluster += num_clusters;
		}
	}

	*meta_add += ref_blocks;
	if (!ref_blocks)
		goto out;

	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
	if (le32_to_cpu(rb->rf_flags) & OCFS2_REFCOUNT_TREE_FL)
		*credits += OCFS2_EXPAND_REFCOUNT_TREE_CREDITS;
	else {
		struct ocfs2_extent_tree et;

		ocfs2_init_refcount_extent_tree(&et, ref_ci, ref_root_bh);
		*credits += ocfs2_calc_extend_credits(inode->i_sb,
3629
						      et.et_root_el);
T
Tao Ma 已提交
3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676
	}

out:
	brelse(ref_leaf_bh);
	return ret;
}

/*
 * Do CoW for xattr.
 */
int ocfs2_refcount_cow_xattr(struct inode *inode,
			     struct ocfs2_dinode *di,
			     struct ocfs2_xattr_value_buf *vb,
			     struct ocfs2_refcount_tree *ref_tree,
			     struct buffer_head *ref_root_bh,
			     u32 cpos, u32 write_len,
			     struct ocfs2_post_refcount *post)
{
	int ret;
	struct ocfs2_xattr_value_root *xv = vb->vb_xv;
	struct ocfs2_inode_info *oi = OCFS2_I(inode);
	struct ocfs2_cow_context *context = NULL;
	u32 cow_start, cow_len;

	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));

	ret = ocfs2_refcount_cal_cow_clusters(inode, &xv->xr_list,
					      cpos, write_len, UINT_MAX,
					      &cow_start, &cow_len);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	BUG_ON(cow_len == 0);

	context = kzalloc(sizeof(struct ocfs2_cow_context), GFP_NOFS);
	if (!context) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	context->inode = inode;
	context->cow_start = cow_start;
	context->cow_len = cow_len;
	context->ref_tree = ref_tree;
3677
	context->ref_root_bh = ref_root_bh;
T
Tao Ma 已提交
3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698
	context->cow_object = xv;

	context->cow_duplicate_clusters = ocfs2_duplicate_clusters_by_jbd;
	/* We need the extra credits for duplicate_clusters by jbd. */
	context->extra_credits =
		ocfs2_clusters_to_blocks(inode->i_sb, 1) * cow_len;
	context->get_clusters = ocfs2_xattr_value_get_clusters;
	context->post_refcount = post;

	ocfs2_init_xattr_value_extent_tree(&context->data_et,
					   INODE_CACHE(inode), vb);

	ret = ocfs2_replace_cow(context);
	if (ret)
		mlog_errno(ret);

out:
	kfree(context);
	return ret;
}

3699 3700 3701 3702 3703 3704 3705 3706 3707
/*
 * Insert a new extent into refcount tree and mark a extent rec
 * as refcounted in the dinode tree.
 */
int ocfs2_add_refcount_flag(struct inode *inode,
			    struct ocfs2_extent_tree *data_et,
			    struct ocfs2_caching_info *ref_ci,
			    struct buffer_head *ref_root_bh,
			    u32 cpos, u32 p_cluster, u32 num_clusters,
3708 3709
			    struct ocfs2_cached_dealloc_ctxt *dealloc,
			    struct ocfs2_post_refcount *post)
3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725
{
	int ret;
	handle_t *handle;
	int credits = 1, ref_blocks = 0;
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	struct ocfs2_alloc_context *meta_ac = NULL;

	ret = ocfs2_calc_refcount_meta_credits(inode->i_sb,
					       ref_ci, ref_root_bh,
					       p_cluster, num_clusters,
					       &ref_blocks, &credits);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

T
Tao Ma 已提交
3726
	trace_ocfs2_add_refcount_flag(ref_blocks, credits);
3727 3728 3729 3730 3731 3732 3733 3734 3735 3736

	if (ref_blocks) {
		ret = ocfs2_reserve_new_metadata_blocks(OCFS2_SB(inode->i_sb),
							ref_blocks, &meta_ac);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	}

3737 3738 3739
	if (post)
		credits += post->credits;

3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754
	handle = ocfs2_start_trans(osb, credits);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_mark_extent_refcounted(inode, data_et, handle,
					   cpos, num_clusters, p_cluster,
					   meta_ac, dealloc);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

3755 3756 3757
	ret = __ocfs2_increase_refcount(handle, ref_ci, ref_root_bh,
					p_cluster, num_clusters, 0,
					meta_ac, dealloc);
3758
	if (ret) {
3759
		mlog_errno(ret);
3760 3761 3762 3763 3764 3765 3766 3767
		goto out_commit;
	}

	if (post && post->func) {
		ret = post->func(inode, handle, post->para);
		if (ret)
			mlog_errno(ret);
	}
3768 3769 3770 3771 3772 3773 3774 3775 3776

out_commit:
	ocfs2_commit_trans(osb, handle);
out:
	if (meta_ac)
		ocfs2_free_alloc_context(meta_ac);
	return ret;
}

3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810
static int ocfs2_change_ctime(struct inode *inode,
			      struct buffer_head *di_bh)
{
	int ret;
	handle_t *handle;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;

	handle = ocfs2_start_trans(OCFS2_SB(inode->i_sb),
				   OCFS2_INODE_UPDATE_CREDITS);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	inode->i_ctime = CURRENT_TIME;
	di->i_ctime = cpu_to_le64(inode->i_ctime.tv_sec);
	di->i_ctime_nsec = cpu_to_le32(inode->i_ctime.tv_nsec);

	ocfs2_journal_dirty(handle, di_bh);

out_commit:
	ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle);
out:
	return ret;
}

3811 3812 3813
static int ocfs2_attach_refcount_tree(struct inode *inode,
				      struct buffer_head *di_bh)
{
3814
	int ret, data_changed = 0;
3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844
	struct buffer_head *ref_root_bh = NULL;
	struct ocfs2_inode_info *oi = OCFS2_I(inode);
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	struct ocfs2_refcount_tree *ref_tree;
	unsigned int ext_flags;
	loff_t size;
	u32 cpos, num_clusters, clusters, p_cluster;
	struct ocfs2_cached_dealloc_ctxt dealloc;
	struct ocfs2_extent_tree di_et;

	ocfs2_init_dealloc_ctxt(&dealloc);

	if (!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL)) {
		ret = ocfs2_create_refcount_tree(inode, di_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	}

	BUG_ON(!di->i_refcount_loc);
	ret = ocfs2_lock_refcount_tree(osb,
				       le64_to_cpu(di->i_refcount_loc), 1,
				       &ref_tree, &ref_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

3845 3846 3847
	if (oi->ip_dyn_features & OCFS2_INLINE_DATA_FL)
		goto attach_xattr;

3848 3849 3850 3851 3852 3853 3854 3855 3856
	ocfs2_init_dinode_extent_tree(&di_et, INODE_CACHE(inode), di_bh);

	size = i_size_read(inode);
	clusters = ocfs2_clusters_for_bytes(inode->i_sb, size);

	cpos = 0;
	while (cpos < clusters) {
		ret = ocfs2_get_clusters(inode, cpos, &p_cluster,
					 &num_clusters, &ext_flags);
3857 3858 3859 3860
		if (ret) {
			mlog_errno(ret);
			goto unlock;
		}
3861 3862 3863 3864 3865
		if (p_cluster && !(ext_flags & OCFS2_EXT_REFCOUNTED)) {
			ret = ocfs2_add_refcount_flag(inode, &di_et,
						      &ref_tree->rf_ci,
						      ref_root_bh, cpos,
						      p_cluster, num_clusters,
3866
						      &dealloc, NULL);
3867 3868
			if (ret) {
				mlog_errno(ret);
3869
				goto unlock;
3870
			}
3871 3872

			data_changed = 1;
3873 3874 3875 3876
		}
		cpos += num_clusters;
	}

3877
attach_xattr:
3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888
	if (oi->ip_dyn_features & OCFS2_HAS_XATTR_FL) {
		ret = ocfs2_xattr_attach_refcount_tree(inode, di_bh,
						       &ref_tree->rf_ci,
						       ref_root_bh,
						       &dealloc);
		if (ret) {
			mlog_errno(ret);
			goto unlock;
		}
	}

3889 3890 3891 3892 3893 3894 3895
	if (data_changed) {
		ret = ocfs2_change_ctime(inode, di_bh);
		if (ret)
			mlog_errno(ret);
	}

unlock:
3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944
	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
	brelse(ref_root_bh);

	if (!ret && ocfs2_dealloc_has_cluster(&dealloc)) {
		ocfs2_schedule_truncate_log_flush(osb, 1);
		ocfs2_run_deallocs(osb, &dealloc);
	}
out:
	/*
	 * Empty the extent map so that we may get the right extent
	 * record from the disk.
	 */
	ocfs2_extent_map_trunc(inode, 0);

	return ret;
}

static int ocfs2_add_refcounted_extent(struct inode *inode,
				   struct ocfs2_extent_tree *et,
				   struct ocfs2_caching_info *ref_ci,
				   struct buffer_head *ref_root_bh,
				   u32 cpos, u32 p_cluster, u32 num_clusters,
				   unsigned int ext_flags,
				   struct ocfs2_cached_dealloc_ctxt *dealloc)
{
	int ret;
	handle_t *handle;
	int credits = 0;
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	struct ocfs2_alloc_context *meta_ac = NULL;

	ret = ocfs2_lock_refcount_allocators(inode->i_sb,
					     p_cluster, num_clusters,
					     et, ref_ci,
					     ref_root_bh, &meta_ac,
					     NULL, &credits);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	handle = ocfs2_start_trans(osb, credits);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_insert_extent(handle, et, cpos,
T
Tao Ma 已提交
3945
			ocfs2_clusters_to_blocks(inode->i_sb, p_cluster),
3946 3947 3948 3949 3950 3951
			num_clusters, ext_flags, meta_ac);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

T
Tao Ma 已提交
3952 3953 3954
	ret = ocfs2_increase_refcount(handle, ref_ci, ref_root_bh,
				      p_cluster, num_clusters,
				      meta_ac, dealloc);
3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965
	if (ret)
		mlog_errno(ret);

out_commit:
	ocfs2_commit_trans(osb, handle);
out:
	if (meta_ac)
		ocfs2_free_alloc_context(meta_ac);
	return ret;
}

3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008
static int ocfs2_duplicate_inline_data(struct inode *s_inode,
				       struct buffer_head *s_bh,
				       struct inode *t_inode,
				       struct buffer_head *t_bh)
{
	int ret;
	handle_t *handle;
	struct ocfs2_super *osb = OCFS2_SB(s_inode->i_sb);
	struct ocfs2_dinode *s_di = (struct ocfs2_dinode *)s_bh->b_data;
	struct ocfs2_dinode *t_di = (struct ocfs2_dinode *)t_bh->b_data;

	BUG_ON(!(OCFS2_I(s_inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL));

	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_journal_access_di(handle, INODE_CACHE(t_inode), t_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	t_di->id2.i_data.id_count = s_di->id2.i_data.id_count;
	memcpy(t_di->id2.i_data.id_data, s_di->id2.i_data.id_data,
	       le16_to_cpu(s_di->id2.i_data.id_count));
	spin_lock(&OCFS2_I(t_inode)->ip_lock);
	OCFS2_I(t_inode)->ip_dyn_features |= OCFS2_INLINE_DATA_FL;
	t_di->i_dyn_features = cpu_to_le16(OCFS2_I(t_inode)->ip_dyn_features);
	spin_unlock(&OCFS2_I(t_inode)->ip_lock);

	ocfs2_journal_dirty(handle, t_bh);

out_commit:
	ocfs2_commit_trans(osb, handle);
out:
	return ret;
}

4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030
static int ocfs2_duplicate_extent_list(struct inode *s_inode,
				struct inode *t_inode,
				struct buffer_head *t_bh,
				struct ocfs2_caching_info *ref_ci,
				struct buffer_head *ref_root_bh,
				struct ocfs2_cached_dealloc_ctxt *dealloc)
{
	int ret = 0;
	u32 p_cluster, num_clusters, clusters, cpos;
	loff_t size;
	unsigned int ext_flags;
	struct ocfs2_extent_tree et;

	ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(t_inode), t_bh);

	size = i_size_read(s_inode);
	clusters = ocfs2_clusters_for_bytes(s_inode->i_sb, size);

	cpos = 0;
	while (cpos < clusters) {
		ret = ocfs2_get_clusters(s_inode, cpos, &p_cluster,
					 &num_clusters, &ext_flags);
4031 4032 4033 4034
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054
		if (p_cluster) {
			ret = ocfs2_add_refcounted_extent(t_inode, &et,
							  ref_ci, ref_root_bh,
							  cpos, p_cluster,
							  num_clusters,
							  ext_flags,
							  dealloc);
			if (ret) {
				mlog_errno(ret);
				goto out;
			}
		}

		cpos += num_clusters;
	}

out:
	return ret;
}

4055 4056 4057 4058 4059 4060 4061 4062 4063
/*
 * change the new file's attributes to the src.
 *
 * reflink creates a snapshot of a file, that means the attributes
 * must be identical except for three exceptions - nlink, ino, and ctime.
 */
static int ocfs2_complete_reflink(struct inode *s_inode,
				  struct buffer_head *s_bh,
				  struct inode *t_inode,
T
Tao Ma 已提交
4064 4065
				  struct buffer_head *t_bh,
				  bool preserve)
4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093
{
	int ret;
	handle_t *handle;
	struct ocfs2_dinode *s_di = (struct ocfs2_dinode *)s_bh->b_data;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)t_bh->b_data;
	loff_t size = i_size_read(s_inode);

	handle = ocfs2_start_trans(OCFS2_SB(t_inode->i_sb),
				   OCFS2_INODE_UPDATE_CREDITS);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		return ret;
	}

	ret = ocfs2_journal_access_di(handle, INODE_CACHE(t_inode), t_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	spin_lock(&OCFS2_I(t_inode)->ip_lock);
	OCFS2_I(t_inode)->ip_clusters = OCFS2_I(s_inode)->ip_clusters;
	OCFS2_I(t_inode)->ip_attr = OCFS2_I(s_inode)->ip_attr;
	OCFS2_I(t_inode)->ip_dyn_features = OCFS2_I(s_inode)->ip_dyn_features;
	spin_unlock(&OCFS2_I(t_inode)->ip_lock);
	i_size_write(t_inode, size);
4094
	t_inode->i_blocks = s_inode->i_blocks;
4095 4096 4097 4098 4099 4100 4101

	di->i_xattr_inline_size = s_di->i_xattr_inline_size;
	di->i_clusters = s_di->i_clusters;
	di->i_size = s_di->i_size;
	di->i_dyn_features = s_di->i_dyn_features;
	di->i_attr = s_di->i_attr;

T
Tao Ma 已提交
4102
	if (preserve) {
4103 4104 4105
		t_inode->i_uid = s_inode->i_uid;
		t_inode->i_gid = s_inode->i_gid;
		t_inode->i_mode = s_inode->i_mode;
T
Tao Ma 已提交
4106 4107 4108 4109 4110 4111 4112 4113 4114 4115
		di->i_uid = s_di->i_uid;
		di->i_gid = s_di->i_gid;
		di->i_mode = s_di->i_mode;

		/*
		 * update time.
		 * we want mtime to appear identical to the source and
		 * update ctime.
		 */
		t_inode->i_ctime = CURRENT_TIME;
4116

T
Tao Ma 已提交
4117 4118
		di->i_ctime = cpu_to_le64(t_inode->i_ctime.tv_sec);
		di->i_ctime_nsec = cpu_to_le32(t_inode->i_ctime.tv_nsec);
4119

T
Tao Ma 已提交
4120 4121 4122 4123
		t_inode->i_mtime = s_inode->i_mtime;
		di->i_mtime = s_di->i_mtime;
		di->i_mtime_nsec = s_di->i_mtime_nsec;
	}
4124 4125 4126 4127 4128 4129 4130 4131

	ocfs2_journal_dirty(handle, t_bh);

out_commit:
	ocfs2_commit_trans(OCFS2_SB(t_inode->i_sb), handle);
	return ret;
}

4132 4133 4134
static int ocfs2_create_reflink_node(struct inode *s_inode,
				     struct buffer_head *s_bh,
				     struct inode *t_inode,
T
Tao Ma 已提交
4135 4136
				     struct buffer_head *t_bh,
				     bool preserve)
4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154
{
	int ret;
	struct buffer_head *ref_root_bh = NULL;
	struct ocfs2_cached_dealloc_ctxt dealloc;
	struct ocfs2_super *osb = OCFS2_SB(s_inode->i_sb);
	struct ocfs2_refcount_block *rb;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)s_bh->b_data;
	struct ocfs2_refcount_tree *ref_tree;

	ocfs2_init_dealloc_ctxt(&dealloc);

	ret = ocfs2_set_refcount_tree(t_inode, t_bh,
				      le64_to_cpu(di->i_refcount_loc));
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

4155 4156 4157 4158 4159 4160 4161 4162
	if (OCFS2_I(s_inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
		ret = ocfs2_duplicate_inline_data(s_inode, s_bh,
						  t_inode, t_bh);
		if (ret)
			mlog_errno(ret);
		goto out;
	}

4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173
	ret = ocfs2_lock_refcount_tree(osb, le64_to_cpu(di->i_refcount_loc),
				       1, &ref_tree, &ref_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
	rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;

	ret = ocfs2_duplicate_extent_list(s_inode, t_inode, t_bh,
					  &ref_tree->rf_ci, ref_root_bh,
					  &dealloc);
4174 4175 4176 4177 4178 4179
	if (ret) {
		mlog_errno(ret);
		goto out_unlock_refcount;
	}

out_unlock_refcount:
4180 4181 4182 4183 4184 4185 4186 4187 4188 4189
	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
	brelse(ref_root_bh);
out:
	if (ocfs2_dealloc_has_cluster(&dealloc)) {
		ocfs2_schedule_truncate_log_flush(osb, 1);
		ocfs2_run_deallocs(osb, &dealloc);
	}

	return ret;
}
T
Tao Ma 已提交
4190 4191 4192 4193 4194 4195 4196

static int __ocfs2_reflink(struct dentry *old_dentry,
			   struct buffer_head *old_bh,
			   struct inode *new_inode,
			   bool preserve)
{
	int ret;
4197
	struct inode *inode = d_inode(old_dentry);
T
Tao Ma 已提交
4198 4199
	struct buffer_head *new_bh = NULL;

4200 4201 4202 4203 4204 4205
	if (OCFS2_I(inode)->ip_flags & OCFS2_INODE_SYSTEM_FILE) {
		ret = -EINVAL;
		mlog_errno(ret);
		goto out;
	}

T
Tao Ma 已提交
4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217
	ret = filemap_fdatawrite(inode->i_mapping);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_attach_refcount_tree(inode, old_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

T
Tao Ma 已提交
4218 4219 4220
	mutex_lock_nested(&new_inode->i_mutex, I_MUTEX_CHILD);
	ret = ocfs2_inode_lock_nested(new_inode, &new_bh, 1,
				      OI_LS_REFLINK_TARGET);
T
Tao Ma 已提交
4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236
	if (ret) {
		mlog_errno(ret);
		goto out_unlock;
	}

	ret = ocfs2_create_reflink_node(inode, old_bh,
					new_inode, new_bh, preserve);
	if (ret) {
		mlog_errno(ret);
		goto inode_unlock;
	}

	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_XATTR_FL) {
		ret = ocfs2_reflink_xattrs(inode, old_bh,
					   new_inode, new_bh,
					   preserve);
4237
		if (ret) {
T
Tao Ma 已提交
4238
			mlog_errno(ret);
4239 4240
			goto inode_unlock;
		}
T
Tao Ma 已提交
4241
	}
4242 4243 4244 4245 4246 4247

	ret = ocfs2_complete_reflink(inode, old_bh,
				     new_inode, new_bh, preserve);
	if (ret)
		mlog_errno(ret);

T
Tao Ma 已提交
4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265
inode_unlock:
	ocfs2_inode_unlock(new_inode, 1);
	brelse(new_bh);
out_unlock:
	mutex_unlock(&new_inode->i_mutex);
out:
	if (!ret) {
		ret = filemap_fdatawait(inode->i_mapping);
		if (ret)
			mlog_errno(ret);
	}
	return ret;
}

static int ocfs2_reflink(struct dentry *old_dentry, struct inode *dir,
			 struct dentry *new_dentry, bool preserve)
{
	int error;
4266
	struct inode *inode = d_inode(old_dentry);
T
Tao Ma 已提交
4267 4268
	struct buffer_head *old_bh = NULL;
	struct inode *new_orphan_inode = NULL;
4269 4270
	struct posix_acl *default_acl, *acl;
	umode_t mode;
T
Tao Ma 已提交
4271 4272 4273 4274

	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb)))
		return -EOPNOTSUPP;

4275 4276 4277 4278
	mode = inode->i_mode;
	error = posix_acl_create(dir, &mode, &default_acl, &acl);
	if (error) {
		mlog_errno(error);
4279
		return error;
4280 4281 4282
	}

	error = ocfs2_create_inode_in_orphan(dir, mode,
T
Tao Ma 已提交
4283 4284 4285 4286 4287 4288
					     &new_orphan_inode);
	if (error) {
		mlog_errno(error);
		goto out;
	}

4289 4290 4291 4292 4293 4294
	error = ocfs2_rw_lock(inode, 1);
	if (error) {
		mlog_errno(error);
		goto out;
	}

T
Tao Ma 已提交
4295 4296 4297
	error = ocfs2_inode_lock(inode, &old_bh, 1);
	if (error) {
		mlog_errno(error);
4298
		ocfs2_rw_unlock(inode, 1);
T
Tao Ma 已提交
4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309
		goto out;
	}

	down_write(&OCFS2_I(inode)->ip_xattr_sem);
	down_write(&OCFS2_I(inode)->ip_alloc_sem);
	error = __ocfs2_reflink(old_dentry, old_bh,
				new_orphan_inode, preserve);
	up_write(&OCFS2_I(inode)->ip_alloc_sem);
	up_write(&OCFS2_I(inode)->ip_xattr_sem);

	ocfs2_inode_unlock(inode, 1);
4310
	ocfs2_rw_unlock(inode, 1);
T
Tao Ma 已提交
4311 4312 4313 4314 4315 4316 4317 4318 4319
	brelse(old_bh);

	if (error) {
		mlog_errno(error);
		goto out;
	}

	/* If the security isn't preserved, we need to re-initialize them. */
	if (!preserve) {
4320
		error = ocfs2_init_security_and_acl(dir, new_orphan_inode,
4321 4322
						    &new_dentry->d_name,
						    default_acl, acl);
T
Tao Ma 已提交
4323 4324 4325 4326
		if (error)
			mlog_errno(error);
	}
out:
4327 4328 4329 4330
	if (default_acl)
		posix_acl_release(default_acl);
	if (acl)
		posix_acl_release(acl);
T
Tao Ma 已提交
4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349
	if (!error) {
		error = ocfs2_mv_orphaned_inode_to_new(dir, new_orphan_inode,
						       new_dentry);
		if (error)
			mlog_errno(error);
	}

	if (new_orphan_inode) {
		/*
		 * We need to open_unlock the inode no matter whether we
		 * succeed or not, so that other nodes can delete it later.
		 */
		ocfs2_open_unlock(new_orphan_inode);
		if (error)
			iput(new_orphan_inode);
	}

	return error;
}
T
Tao Ma 已提交
4350 4351 4352 4353 4354 4355 4356 4357 4358 4359

/*
 * Below here are the bits used by OCFS2_IOC_REFLINK() to fake
 * sys_reflink().  This will go away when vfs_reflink() exists in
 * fs/namei.c.
 */

/* copied from may_create in VFS. */
static inline int ocfs2_may_create(struct inode *dir, struct dentry *child)
{
4360
	if (d_really_is_positive(child))
T
Tao Ma 已提交
4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374
		return -EEXIST;
	if (IS_DEADDIR(dir))
		return -ENOENT;
	return inode_permission(dir, MAY_WRITE | MAY_EXEC);
}

/**
 * ocfs2_vfs_reflink - Create a reference-counted link
 *
 * @old_dentry:        source dentry + inode
 * @dir:       directory to create the target
 * @new_dentry:        target dentry
 * @preserve:  if true, preserve all file attributes
 */
T
Tao Ma 已提交
4375 4376
static int ocfs2_vfs_reflink(struct dentry *old_dentry, struct inode *dir,
			     struct dentry *new_dentry, bool preserve)
T
Tao Ma 已提交
4377
{
4378
	struct inode *inode = d_inode(old_dentry);
T
Tao Ma 已提交
4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405
	int error;

	if (!inode)
		return -ENOENT;

	error = ocfs2_may_create(dir, new_dentry);
	if (error)
		return error;

	if (dir->i_sb != inode->i_sb)
		return -EXDEV;

	/*
	 * A reflink to an append-only or immutable file cannot be created.
	 */
	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
		return -EPERM;

	/* Only regular files can be reflinked. */
	if (!S_ISREG(inode->i_mode))
		return -EPERM;

	/*
	 * If the caller wants to preserve ownership, they require the
	 * rights to do so.
	 */
	if (preserve) {
4406
		if (!uid_eq(current_fsuid(), inode->i_uid) && !capable(CAP_CHOWN))
T
Tao Ma 已提交
4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423
			return -EPERM;
		if (!in_group_p(inode->i_gid) && !capable(CAP_CHOWN))
			return -EPERM;
	}

	/*
	 * If the caller is modifying any aspect of the attributes, they
	 * are not creating a snapshot.  They need read permission on the
	 * file.
	 */
	if (!preserve) {
		error = inode_permission(inode, MAY_READ);
		if (error)
			return error;
	}

	mutex_lock(&inode->i_mutex);
4424
	dquot_initialize(dir);
T
Tao Ma 已提交
4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439
	error = ocfs2_reflink(old_dentry, dir, new_dentry, preserve);
	mutex_unlock(&inode->i_mutex);
	if (!error)
		fsnotify_create(dir, new_dentry);
	return error;
}
/*
 * Most codes are copied from sys_linkat.
 */
int ocfs2_reflink_ioctl(struct inode *inode,
			const char __user *oldname,
			const char __user *newname,
			bool preserve)
{
	struct dentry *new_dentry;
4440
	struct path old_path, new_path;
T
Tao Ma 已提交
4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451
	int error;

	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb)))
		return -EOPNOTSUPP;

	error = user_path_at(AT_FDCWD, oldname, 0, &old_path);
	if (error) {
		mlog_errno(error);
		return error;
	}

4452 4453 4454
	new_dentry = user_path_create(AT_FDCWD, newname, &new_path, 0);
	error = PTR_ERR(new_dentry);
	if (IS_ERR(new_dentry)) {
T
Tao Ma 已提交
4455 4456 4457 4458 4459
		mlog_errno(error);
		goto out;
	}

	error = -EXDEV;
4460
	if (old_path.mnt != new_path.mnt) {
T
Tao Ma 已提交
4461
		mlog_errno(error);
4462
		goto out_dput;
T
Tao Ma 已提交
4463 4464 4465
	}

	error = ocfs2_vfs_reflink(old_path.dentry,
4466
				  d_inode(new_path.dentry),
T
Tao Ma 已提交
4467 4468
				  new_dentry, preserve);
out_dput:
A
Al Viro 已提交
4469
	done_path_create(&new_path, new_dentry);
T
Tao Ma 已提交
4470 4471 4472 4473 4474
out:
	path_put(&old_path);

	return error;
}