xfs_dir2_node.c 59.6 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2 3
 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
 * All Rights Reserved.
L
Linus Torvalds 已提交
4
 *
5 6
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
L
Linus Torvalds 已提交
7 8
 * published by the Free Software Foundation.
 *
9 10 11 12
 * This program is distributed in the hope that it would 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.
L
Linus Torvalds 已提交
13
 *
14 15 16
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write the Free Software Foundation,
 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
L
Linus Torvalds 已提交
17 18
 */
#include "xfs.h"
19
#include "xfs_fs.h"
L
Linus Torvalds 已提交
20 21
#include "xfs_types.h"
#include "xfs_log.h"
22
#include "xfs_inum.h"
L
Linus Torvalds 已提交
23 24
#include "xfs_trans.h"
#include "xfs_sb.h"
25
#include "xfs_ag.h"
L
Linus Torvalds 已提交
26 27 28
#include "xfs_dir2.h"
#include "xfs_dmapi.h"
#include "xfs_mount.h"
29
#include "xfs_da_btree.h"
L
Linus Torvalds 已提交
30 31
#include "xfs_bmap_btree.h"
#include "xfs_dir2_sf.h"
32
#include "xfs_attr_sf.h"
L
Linus Torvalds 已提交
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_bmap.h"
#include "xfs_dir2_data.h"
#include "xfs_dir2_leaf.h"
#include "xfs_dir2_block.h"
#include "xfs_dir2_node.h"
#include "xfs_dir2_trace.h"
#include "xfs_error.h"

/*
 * Function declarations.
 */
static void xfs_dir2_free_log_header(xfs_trans_t *tp, xfs_dabuf_t *bp);
static int xfs_dir2_leafn_add(xfs_dabuf_t *bp, xfs_da_args_t *args, int index);
#ifdef DEBUG
static void xfs_dir2_leafn_check(xfs_inode_t *dp, xfs_dabuf_t *bp);
#else
#define	xfs_dir2_leafn_check(dp, bp)
#endif
static void xfs_dir2_leafn_moveents(xfs_da_args_t *args, xfs_dabuf_t *bp_s,
				    int start_s, xfs_dabuf_t *bp_d, int start_d,
				    int count);
static void xfs_dir2_leafn_rebalance(xfs_da_state_t *state,
				     xfs_da_state_blk_t *blk1,
				     xfs_da_state_blk_t *blk2);
static int xfs_dir2_leafn_remove(xfs_da_args_t *args, xfs_dabuf_t *bp,
				 int index, xfs_da_state_blk_t *dblk,
				 int *rval);
static int xfs_dir2_node_addname_int(xfs_da_args_t *args,
				     xfs_da_state_blk_t *fblk);

/*
 * Log entries from a freespace block.
 */
void
xfs_dir2_free_log_bests(
	xfs_trans_t		*tp,		/* transaction pointer */
	xfs_dabuf_t		*bp,		/* freespace buffer */
	int			first,		/* first entry to log */
	int			last)		/* last entry to log */
{
	xfs_dir2_free_t		*free;		/* freespace structure */

	free = bp->data;
78
	ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
L
Linus Torvalds 已提交
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
	xfs_da_log_buf(tp, bp,
		(uint)((char *)&free->bests[first] - (char *)free),
		(uint)((char *)&free->bests[last] - (char *)free +
		       sizeof(free->bests[0]) - 1));
}

/*
 * Log header from a freespace block.
 */
static void
xfs_dir2_free_log_header(
	xfs_trans_t		*tp,		/* transaction pointer */
	xfs_dabuf_t		*bp)		/* freespace buffer */
{
	xfs_dir2_free_t		*free;		/* freespace structure */

	free = bp->data;
96
	ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
L
Linus Torvalds 已提交
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115
	xfs_da_log_buf(tp, bp, (uint)((char *)&free->hdr - (char *)free),
		(uint)(sizeof(xfs_dir2_free_hdr_t) - 1));
}

/*
 * Convert a leaf-format directory to a node-format directory.
 * We need to change the magic number of the leaf block, and copy
 * the freespace table out of the leaf block into its own block.
 */
int						/* error */
xfs_dir2_leaf_to_node(
	xfs_da_args_t		*args,		/* operation arguments */
	xfs_dabuf_t		*lbp)		/* leaf buffer */
{
	xfs_inode_t		*dp;		/* incore directory inode */
	int			error;		/* error return value */
	xfs_dabuf_t		*fbp;		/* freespace buffer */
	xfs_dir2_db_t		fdb;		/* freespace block number */
	xfs_dir2_free_t		*free;		/* freespace structure */
116
	__be16			*from;		/* pointer to freespace entry */
L
Linus Torvalds 已提交
117 118 119 120 121 122
	int			i;		/* leaf freespace index */
	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
	xfs_mount_t		*mp;		/* filesystem mount point */
	int			n;		/* count of live freespc ents */
	xfs_dir2_data_off_t	off;		/* freespace entry value */
123
	__be16			*to;		/* pointer to freespace entry */
L
Linus Torvalds 已提交
124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139
	xfs_trans_t		*tp;		/* transaction pointer */

	xfs_dir2_trace_args_b("leaf_to_node", args, lbp);
	dp = args->dp;
	mp = dp->i_mount;
	tp = args->trans;
	/*
	 * Add a freespace block to the directory.
	 */
	if ((error = xfs_dir2_grow_inode(args, XFS_DIR2_FREE_SPACE, &fdb))) {
		return error;
	}
	ASSERT(fdb == XFS_DIR2_FREE_FIRSTDB(mp));
	/*
	 * Get the buffer for the new freespace block.
	 */
140
	if ((error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(mp, fdb), -1, &fbp,
L
Linus Torvalds 已提交
141 142 143 144 145 146
			XFS_DATA_FORK))) {
		return error;
	}
	ASSERT(fbp != NULL);
	free = fbp->data;
	leaf = lbp->data;
147
	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
L
Linus Torvalds 已提交
148 149 150
	/*
	 * Initialize the freespace block header.
	 */
151
	free->hdr.magic = cpu_to_be32(XFS_DIR2_FREE_MAGIC);
L
Linus Torvalds 已提交
152
	free->hdr.firstdb = 0;
153
	ASSERT(be32_to_cpu(ltp->bestcount) <= (uint)dp->i_d.di_size / mp->m_dirblksize);
154
	free->hdr.nvalid = ltp->bestcount;
L
Linus Torvalds 已提交
155 156 157 158
	/*
	 * Copy freespace entries from the leaf block to the new block.
	 * Count active entries.
	 */
159
	for (i = n = 0, from = xfs_dir2_leaf_bests_p(ltp), to = free->bests;
160
	     i < be32_to_cpu(ltp->bestcount); i++, from++, to++) {
161
		if ((off = be16_to_cpu(*from)) != NULLDATAOFF)
L
Linus Torvalds 已提交
162
			n++;
163
		*to = cpu_to_be16(off);
L
Linus Torvalds 已提交
164
	}
165
	free->hdr.nused = cpu_to_be32(n);
166
	leaf->hdr.info.magic = cpu_to_be16(XFS_DIR2_LEAFN_MAGIC);
L
Linus Torvalds 已提交
167 168 169 170 171
	/*
	 * Log everything.
	 */
	xfs_dir2_leaf_log_header(tp, lbp);
	xfs_dir2_free_log_header(tp, fbp);
172
	xfs_dir2_free_log_bests(tp, fbp, 0, be32_to_cpu(free->hdr.nvalid) - 1);
L
Linus Torvalds 已提交
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218
	xfs_da_buf_done(fbp);
	xfs_dir2_leafn_check(dp, lbp);
	return 0;
}

/*
 * Add a leaf entry to a leaf block in a node-form directory.
 * The other work necessary is done from the caller.
 */
static int					/* error */
xfs_dir2_leafn_add(
	xfs_dabuf_t		*bp,		/* leaf buffer */
	xfs_da_args_t		*args,		/* operation arguments */
	int			index)		/* insertion pt for new entry */
{
	int			compact;	/* compacting stale leaves */
	xfs_inode_t		*dp;		/* incore directory inode */
	int			highstale;	/* next stale entry */
	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
	int			lfloghigh;	/* high leaf entry logging */
	int			lfloglow;	/* low leaf entry logging */
	int			lowstale;	/* previous stale entry */
	xfs_mount_t		*mp;		/* filesystem mount point */
	xfs_trans_t		*tp;		/* transaction pointer */

	xfs_dir2_trace_args_sb("leafn_add", args, index, bp);
	dp = args->dp;
	mp = dp->i_mount;
	tp = args->trans;
	leaf = bp->data;

	/*
	 * Quick check just to make sure we are not going to index
	 * into other peoples memory
	 */
	if (index < 0)
		return XFS_ERROR(EFSCORRUPTED);

	/*
	 * If there are already the maximum number of leaf entries in
	 * the block, if there are no stale entries it won't fit.
	 * Caller will do a split.  If there are stale entries we'll do
	 * a compact.
	 */

219
	if (be16_to_cpu(leaf->hdr.count) == xfs_dir2_max_leaf_ents(mp)) {
L
Linus Torvalds 已提交
220 221
		if (!leaf->hdr.stale)
			return XFS_ERROR(ENOSPC);
222
		compact = be16_to_cpu(leaf->hdr.stale) > 1;
L
Linus Torvalds 已提交
223 224
	} else
		compact = 0;
225
	ASSERT(index == 0 || be32_to_cpu(leaf->ents[index - 1].hashval) <= args->hashval);
226
	ASSERT(index == be16_to_cpu(leaf->hdr.count) ||
227
	       be32_to_cpu(leaf->ents[index].hashval) >= args->hashval);
L
Linus Torvalds 已提交
228

229
	if (args->op_flags & XFS_DA_OP_JUSTCHECK)
L
Linus Torvalds 已提交
230 231 232 233 234 235 236 237 238 239 240 241 242 243
		return 0;

	/*
	 * Compact out all but one stale leaf entry.  Leaves behind
	 * the entry closest to index.
	 */
	if (compact) {
		xfs_dir2_leaf_compact_x1(bp, &index, &lowstale, &highstale,
			&lfloglow, &lfloghigh);
	}
	/*
	 * Set impossible logging indices for this case.
	 */
	else if (leaf->hdr.stale) {
244
		lfloglow = be16_to_cpu(leaf->hdr.count);
L
Linus Torvalds 已提交
245 246 247 248 249 250 251
		lfloghigh = -1;
	}
	/*
	 * No stale entries, just insert a space for the new entry.
	 */
	if (!leaf->hdr.stale) {
		lep = &leaf->ents[index];
252
		if (index < be16_to_cpu(leaf->hdr.count))
L
Linus Torvalds 已提交
253
			memmove(lep + 1, lep,
254
				(be16_to_cpu(leaf->hdr.count) - index) * sizeof(*lep));
L
Linus Torvalds 已提交
255
		lfloglow = index;
256
		lfloghigh = be16_to_cpu(leaf->hdr.count);
257
		be16_add_cpu(&leaf->hdr.count, 1);
L
Linus Torvalds 已提交
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272
	}
	/*
	 * There are stale entries.  We'll use one for the new entry.
	 */
	else {
		/*
		 * If we didn't do a compact then we need to figure out
		 * which stale entry will be used.
		 */
		if (compact == 0) {
			/*
			 * Find first stale entry before our insertion point.
			 */
			for (lowstale = index - 1;
			     lowstale >= 0 &&
273
				be32_to_cpu(leaf->ents[lowstale].address) !=
L
Linus Torvalds 已提交
274 275 276 277 278 279 280 281 282
				XFS_DIR2_NULL_DATAPTR;
			     lowstale--)
				continue;
			/*
			 * Find next stale entry after insertion point.
			 * Stop looking if the answer would be worse than
			 * lowstale already found.
			 */
			for (highstale = index;
283
			     highstale < be16_to_cpu(leaf->hdr.count) &&
284
				be32_to_cpu(leaf->ents[highstale].address) !=
L
Linus Torvalds 已提交
285 286 287 288 289 290 291 292 293 294 295
				XFS_DIR2_NULL_DATAPTR &&
				(lowstale < 0 ||
				 index - lowstale - 1 >= highstale - index);
			     highstale++)
				continue;
		}
		/*
		 * Using the low stale entry.
		 * Shift entries up toward the stale slot.
		 */
		if (lowstale >= 0 &&
296
		    (highstale == be16_to_cpu(leaf->hdr.count) ||
L
Linus Torvalds 已提交
297
		     index - lowstale - 1 < highstale - index)) {
298
			ASSERT(be32_to_cpu(leaf->ents[lowstale].address) ==
L
Linus Torvalds 已提交
299 300 301 302 303 304 305 306 307 308 309 310 311 312 313
			       XFS_DIR2_NULL_DATAPTR);
			ASSERT(index - lowstale - 1 >= 0);
			if (index - lowstale - 1 > 0)
				memmove(&leaf->ents[lowstale],
					&leaf->ents[lowstale + 1],
					(index - lowstale - 1) * sizeof(*lep));
			lep = &leaf->ents[index - 1];
			lfloglow = MIN(lowstale, lfloglow);
			lfloghigh = MAX(index - 1, lfloghigh);
		}
		/*
		 * Using the high stale entry.
		 * Shift entries down toward the stale slot.
		 */
		else {
314
			ASSERT(be32_to_cpu(leaf->ents[highstale].address) ==
L
Linus Torvalds 已提交
315 316 317 318 319 320 321 322 323 324
			       XFS_DIR2_NULL_DATAPTR);
			ASSERT(highstale - index >= 0);
			if (highstale - index > 0)
				memmove(&leaf->ents[index + 1],
					&leaf->ents[index],
					(highstale - index) * sizeof(*lep));
			lep = &leaf->ents[index];
			lfloglow = MIN(index, lfloglow);
			lfloghigh = MAX(highstale, lfloghigh);
		}
325
		be16_add_cpu(&leaf->hdr.stale, -1);
L
Linus Torvalds 已提交
326 327 328 329
	}
	/*
	 * Insert the new entry, log everything.
	 */
330
	lep->hashval = cpu_to_be32(args->hashval);
331
	lep->address = cpu_to_be32(xfs_dir2_db_off_to_dataptr(mp,
332
				args->blkno, args->index));
L
Linus Torvalds 已提交
333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
	xfs_dir2_leaf_log_header(tp, bp);
	xfs_dir2_leaf_log_ents(tp, bp, lfloglow, lfloghigh);
	xfs_dir2_leafn_check(dp, bp);
	return 0;
}

#ifdef DEBUG
/*
 * Check internal consistency of a leafn block.
 */
void
xfs_dir2_leafn_check(
	xfs_inode_t	*dp,			/* incore directory inode */
	xfs_dabuf_t	*bp)			/* leaf buffer */
{
	int		i;			/* leaf index */
	xfs_dir2_leaf_t	*leaf;			/* leaf structure */
	xfs_mount_t	*mp;			/* filesystem mount point */
	int		stale;			/* count of stale leaves */

	leaf = bp->data;
	mp = dp->i_mount;
355
	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
356
	ASSERT(be16_to_cpu(leaf->hdr.count) <= xfs_dir2_max_leaf_ents(mp));
357 358
	for (i = stale = 0; i < be16_to_cpu(leaf->hdr.count); i++) {
		if (i + 1 < be16_to_cpu(leaf->hdr.count)) {
359 360
			ASSERT(be32_to_cpu(leaf->ents[i].hashval) <=
			       be32_to_cpu(leaf->ents[i + 1].hashval));
L
Linus Torvalds 已提交
361
		}
362
		if (be32_to_cpu(leaf->ents[i].address) == XFS_DIR2_NULL_DATAPTR)
L
Linus Torvalds 已提交
363 364
			stale++;
	}
365
	ASSERT(be16_to_cpu(leaf->hdr.stale) == stale);
L
Linus Torvalds 已提交
366 367 368 369 370 371 372 373 374 375 376 377 378 379 380
}
#endif	/* DEBUG */

/*
 * Return the last hash value in the leaf.
 * Stale entries are ok.
 */
xfs_dahash_t					/* hash value */
xfs_dir2_leafn_lasthash(
	xfs_dabuf_t	*bp,			/* leaf buffer */
	int		*count)			/* count of entries in leaf */
{
	xfs_dir2_leaf_t	*leaf;			/* leaf structure */

	leaf = bp->data;
381
	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
L
Linus Torvalds 已提交
382
	if (count)
383
		*count = be16_to_cpu(leaf->hdr.count);
L
Linus Torvalds 已提交
384 385
	if (!leaf->hdr.count)
		return 0;
386
	return be32_to_cpu(leaf->ents[be16_to_cpu(leaf->hdr.count) - 1].hashval);
L
Linus Torvalds 已提交
387 388 389
}

/*
390 391
 * Look up a leaf entry for space to add a name in a node-format leaf block.
 * The extrablk in state is a freespace block.
L
Linus Torvalds 已提交
392
 */
393 394
STATIC int
xfs_dir2_leafn_lookup_for_addname(
L
Linus Torvalds 已提交
395 396 397 398 399
	xfs_dabuf_t		*bp,		/* leaf buffer */
	xfs_da_args_t		*args,		/* operation arguments */
	int			*indexp,	/* out: leaf entry index */
	xfs_da_state_t		*state)		/* state to fill in */
{
400 401 402
	xfs_dabuf_t		*curbp = NULL;	/* current data/free buffer */
	xfs_dir2_db_t		curdb = -1;	/* current data block number */
	xfs_dir2_db_t		curfdb = -1;	/* current free block number */
L
Linus Torvalds 已提交
403 404 405
	xfs_inode_t		*dp;		/* incore directory inode */
	int			error;		/* error return value */
	int			fi;		/* free entry index */
406
	xfs_dir2_free_t		*free = NULL;	/* free block structure */
L
Linus Torvalds 已提交
407 408
	int			index;		/* leaf entry index */
	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
409
	int			length;		/* length of new data entry */
L
Linus Torvalds 已提交
410 411 412 413 414 415 416 417 418 419
	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
	xfs_mount_t		*mp;		/* filesystem mount point */
	xfs_dir2_db_t		newdb;		/* new data block number */
	xfs_dir2_db_t		newfdb;		/* new free block number */
	xfs_trans_t		*tp;		/* transaction pointer */

	dp = args->dp;
	tp = args->trans;
	mp = dp->i_mount;
	leaf = bp->data;
420
	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
L
Linus Torvalds 已提交
421
#ifdef __KERNEL__
422
	ASSERT(be16_to_cpu(leaf->hdr.count) > 0);
L
Linus Torvalds 已提交
423 424 425 426 427 428 429 430 431
#endif
	xfs_dir2_leafn_check(dp, bp);
	/*
	 * Look up the hash value in the leaf entries.
	 */
	index = xfs_dir2_leaf_search_hash(args, bp);
	/*
	 * Do we have a buffer coming in?
	 */
432 433
	if (state->extravalid) {
		/* If so, it's a free block buffer, get the block number. */
L
Linus Torvalds 已提交
434
		curbp = state->extrablk.bp;
435 436 437
		curfdb = state->extrablk.blkno;
		free = curbp->data;
		ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
L
Linus Torvalds 已提交
438
	}
439
	length = xfs_dir2_data_entsize(args->namelen);
L
Linus Torvalds 已提交
440 441 442
	/*
	 * Loop over leaf entries with the right hash value.
	 */
443 444 445
	for (lep = &leaf->ents[index]; index < be16_to_cpu(leaf->hdr.count) &&
				be32_to_cpu(lep->hashval) == args->hashval;
				lep++, index++) {
L
Linus Torvalds 已提交
446 447 448
		/*
		 * Skip stale leaf entries.
		 */
449
		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
L
Linus Torvalds 已提交
450 451 452 453
			continue;
		/*
		 * Pull the data block number from the entry.
		 */
454
		newdb = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
L
Linus Torvalds 已提交
455 456 457 458
		/*
		 * For addname, we're looking for a place to put the new entry.
		 * We want to use a data block with an entry of equal
		 * hash value to ours if there is one with room.
459 460 461
		 *
		 * If this block isn't the data block we already have
		 * in hand, take a look at it.
L
Linus Torvalds 已提交
462
		 */
463 464
		if (newdb != curdb) {
			curdb = newdb;
L
Linus Torvalds 已提交
465
			/*
466 467
			 * Convert the data block to the free block
			 * holding its freespace information.
L
Linus Torvalds 已提交
468
			 */
469
			newfdb = xfs_dir2_db_to_fdb(mp, newdb);
L
Linus Torvalds 已提交
470
			/*
471
			 * If it's not the one we have in hand, read it in.
L
Linus Torvalds 已提交
472
			 */
473
			if (newfdb != curfdb) {
L
Linus Torvalds 已提交
474
				/*
475
				 * If we had one before, drop it.
L
Linus Torvalds 已提交
476 477 478 479
				 */
				if (curbp)
					xfs_da_brelse(tp, curbp);
				/*
480
				 * Read the free block.
L
Linus Torvalds 已提交
481
				 */
482 483 484 485
				error = xfs_da_read_buf(tp, dp,
						xfs_dir2_db_to_da(mp, newfdb),
						-1, &curbp, XFS_DATA_FORK);
				if (error)
L
Linus Torvalds 已提交
486
					return error;
487 488 489 490 491 492 493 494
				free = curbp->data;
				ASSERT(be32_to_cpu(free->hdr.magic) ==
					XFS_DIR2_FREE_MAGIC);
				ASSERT((be32_to_cpu(free->hdr.firstdb) %
					XFS_DIR2_MAX_FREE_BESTS(mp)) == 0);
				ASSERT(be32_to_cpu(free->hdr.firstdb) <= curdb);
				ASSERT(curdb < be32_to_cpu(free->hdr.firstdb) +
					be32_to_cpu(free->hdr.nvalid));
L
Linus Torvalds 已提交
495 496
			}
			/*
497
			 * Get the index for our entry.
L
Linus Torvalds 已提交
498
			 */
499
			fi = xfs_dir2_db_to_fdindex(mp, curdb);
L
Linus Torvalds 已提交
500
			/*
501
			 * If it has room, return it.
L
Linus Torvalds 已提交
502
			 */
503 504 505 506 507 508
			if (unlikely(be16_to_cpu(free->bests[fi]) == NULLDATAOFF)) {
				XFS_ERROR_REPORT("xfs_dir2_leafn_lookup_int",
							XFS_ERRLEVEL_LOW, mp);
				if (curfdb != newfdb)
					xfs_da_brelse(tp, curbp);
				return XFS_ERROR(EFSCORRUPTED);
L
Linus Torvalds 已提交
509
			}
510 511 512
			curfdb = newfdb;
			if (be16_to_cpu(free->bests[fi]) >= length)
				goto out;
L
Linus Torvalds 已提交
513 514
		}
	}
515 516 517
	/* Didn't find any space */
	fi = -1;
out:
518
	ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
519 520 521 522 523 524 525 526 527 528
	if (curbp) {
		/* Giving back a free block. */
		state->extravalid = 1;
		state->extrablk.bp = curbp;
		state->extrablk.index = fi;
		state->extrablk.blkno = curfdb;
		state->extrablk.magic = XFS_DIR2_FREE_MAGIC;
	} else {
		state->extravalid = 0;
	}
L
Linus Torvalds 已提交
529
	/*
530
	 * Return the index, that will be the insertion point.
L
Linus Torvalds 已提交
531
	 */
532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557
	*indexp = index;
	return XFS_ERROR(ENOENT);
}

/*
 * Look up a leaf entry in a node-format leaf block.
 * The extrablk in state a data block.
 */
STATIC int
xfs_dir2_leafn_lookup_for_entry(
	xfs_dabuf_t		*bp,		/* leaf buffer */
	xfs_da_args_t		*args,		/* operation arguments */
	int			*indexp,	/* out: leaf entry index */
	xfs_da_state_t		*state)		/* state to fill in */
{
	xfs_dabuf_t		*curbp = NULL;	/* current data/free buffer */
	xfs_dir2_db_t		curdb = -1;	/* current data block number */
	xfs_dir2_data_entry_t	*dep;		/* data block entry */
	xfs_inode_t		*dp;		/* incore directory inode */
	int			error;		/* error return value */
	int			index;		/* leaf entry index */
	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
	xfs_mount_t		*mp;		/* filesystem mount point */
	xfs_dir2_db_t		newdb;		/* new data block number */
	xfs_trans_t		*tp;		/* transaction pointer */
558
	enum xfs_dacmp		cmp;		/* comparison result */
559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590

	dp = args->dp;
	tp = args->trans;
	mp = dp->i_mount;
	leaf = bp->data;
	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
#ifdef __KERNEL__
	ASSERT(be16_to_cpu(leaf->hdr.count) > 0);
#endif
	xfs_dir2_leafn_check(dp, bp);
	/*
	 * Look up the hash value in the leaf entries.
	 */
	index = xfs_dir2_leaf_search_hash(args, bp);
	/*
	 * Do we have a buffer coming in?
	 */
	if (state->extravalid) {
		curbp = state->extrablk.bp;
		curdb = state->extrablk.blkno;
	}
	/*
	 * Loop over leaf entries with the right hash value.
	 */
	for (lep = &leaf->ents[index]; index < be16_to_cpu(leaf->hdr.count) &&
				be32_to_cpu(lep->hashval) == args->hashval;
				lep++, index++) {
		/*
		 * Skip stale leaf entries.
		 */
		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
			continue;
L
Linus Torvalds 已提交
591
		/*
592 593 594 595 596 597 598 599
		 * Pull the data block number from the entry.
		 */
		newdb = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
		/*
		 * Not adding a new entry, so we really want to find
		 * the name given to us.
		 *
		 * If it's a different data block, go get it.
L
Linus Torvalds 已提交
600
		 */
601 602
		if (newdb != curdb) {
			/*
603 604
			 * If we had a block before that we aren't saving
			 * for a CI name, drop it
605
			 */
606 607
			if (curbp && (args->cmpresult == XFS_CMP_DIFFERENT ||
						curdb != state->extrablk.blkno))
608 609
				xfs_da_brelse(tp, curbp);
			/*
610 611
			 * If needing the block that is saved with a CI match,
			 * use it otherwise read in the new data block.
612
			 */
613 614 615 616 617 618 619 620 621 622 623
			if (args->cmpresult != XFS_CMP_DIFFERENT &&
					newdb == state->extrablk.blkno) {
				ASSERT(state->extravalid);
				curbp = state->extrablk.bp;
			} else {
				error = xfs_da_read_buf(tp, dp,
						xfs_dir2_db_to_da(mp, newdb),
						-1, &curbp, XFS_DATA_FORK);
				if (error)
					return error;
			}
624 625
			xfs_dir2_data_check(dp, curbp);
			curdb = newdb;
L
Linus Torvalds 已提交
626 627
		}
		/*
628
		 * Point to the data entry.
L
Linus Torvalds 已提交
629
		 */
630 631 632
		dep = (xfs_dir2_data_entry_t *)((char *)curbp->data +
			xfs_dir2_dataptr_to_off(mp, be32_to_cpu(lep->address)));
		/*
633 634
		 * Compare the entry and if it's an exact match, return
		 * EEXIST immediately. If it's the first case-insensitive
635
		 * match, store the block & inode number and continue looking.
636
		 */
637 638
		cmp = mp->m_dirnameops->compname(args, dep->name, dep->namelen);
		if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
639 640 641 642
			/* If there is a CI match block, drop it */
			if (args->cmpresult != XFS_CMP_DIFFERENT &&
						curdb != state->extrablk.blkno)
				xfs_da_brelse(tp, state->extrablk.bp);
643
			args->cmpresult = cmp;
644
			args->inumber = be64_to_cpu(dep->inumber);
645 646 647 648 649 650 651
			*indexp = index;
			state->extravalid = 1;
			state->extrablk.bp = curbp;
			state->extrablk.blkno = curdb;
			state->extrablk.index = (int)((char *)dep -
							(char *)curbp->data);
			state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
652
			if (cmp == XFS_CMP_EXACT)
653
				return XFS_ERROR(EEXIST);
L
Linus Torvalds 已提交
654 655
		}
	}
656 657
	ASSERT(index == be16_to_cpu(leaf->hdr.count) ||
					(args->op_flags & XFS_DA_OP_OKNOENT));
658
	if (curbp) {
659 660 661 662 663 664 665 666 667 668 669 670
		if (args->cmpresult == XFS_CMP_DIFFERENT) {
			/* Giving back last used data block. */
			state->extravalid = 1;
			state->extrablk.bp = curbp;
			state->extrablk.index = -1;
			state->extrablk.blkno = curdb;
			state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
		} else {
			/* If the curbp is not the CI match block, drop it */
			if (state->extrablk.bp != curbp)
				xfs_da_brelse(tp, curbp);
		}
671 672 673
	} else {
		state->extravalid = 0;
	}
L
Linus Torvalds 已提交
674
	*indexp = index;
675
	return XFS_ERROR(ENOENT);
676 677 678 679 680 681 682 683 684 685 686 687 688 689
}

/*
 * Look up a leaf entry in a node-format leaf block.
 * If this is an addname then the extrablk in state is a freespace block,
 * otherwise it's a data block.
 */
int
xfs_dir2_leafn_lookup_int(
	xfs_dabuf_t		*bp,		/* leaf buffer */
	xfs_da_args_t		*args,		/* operation arguments */
	int			*indexp,	/* out: leaf entry index */
	xfs_da_state_t		*state)		/* state to fill in */
{
690
	if (args->op_flags & XFS_DA_OP_ADDNAME)
691 692 693
		return xfs_dir2_leafn_lookup_for_addname(bp, args, indexp,
							state);
	return xfs_dir2_leafn_lookup_for_entry(bp, args, indexp, state);
L
Linus Torvalds 已提交
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
}

/*
 * Move count leaf entries from source to destination leaf.
 * Log entries and headers.  Stale entries are preserved.
 */
static void
xfs_dir2_leafn_moveents(
	xfs_da_args_t	*args,			/* operation arguments */
	xfs_dabuf_t	*bp_s,			/* source leaf buffer */
	int		start_s,		/* source leaf index */
	xfs_dabuf_t	*bp_d,			/* destination leaf buffer */
	int		start_d,		/* destination leaf index */
	int		count)			/* count of leaves to copy */
{
	xfs_dir2_leaf_t	*leaf_d;		/* destination leaf structure */
	xfs_dir2_leaf_t	*leaf_s;		/* source leaf structure */
	int		stale;			/* count stale leaves copied */
	xfs_trans_t	*tp;			/* transaction pointer */

	xfs_dir2_trace_args_bibii("leafn_moveents", args, bp_s, start_s, bp_d,
		start_d, count);
	/*
	 * Silently return if nothing to do.
	 */
	if (count == 0) {
		return;
	}
	tp = args->trans;
	leaf_s = bp_s->data;
	leaf_d = bp_d->data;
	/*
	 * If the destination index is not the end of the current
	 * destination leaf entries, open up a hole in the destination
	 * to hold the new entries.
	 */
730
	if (start_d < be16_to_cpu(leaf_d->hdr.count)) {
L
Linus Torvalds 已提交
731
		memmove(&leaf_d->ents[start_d + count], &leaf_d->ents[start_d],
732
			(be16_to_cpu(leaf_d->hdr.count) - start_d) *
L
Linus Torvalds 已提交
733 734
			sizeof(xfs_dir2_leaf_entry_t));
		xfs_dir2_leaf_log_ents(tp, bp_d, start_d + count,
735
			count + be16_to_cpu(leaf_d->hdr.count) - 1);
L
Linus Torvalds 已提交
736 737 738 739 740 741 742 743 744
	}
	/*
	 * If the source has stale leaves, count the ones in the copy range
	 * so we can update the header correctly.
	 */
	if (leaf_s->hdr.stale) {
		int	i;			/* temp leaf index */

		for (i = start_s, stale = 0; i < start_s + count; i++) {
745
			if (be32_to_cpu(leaf_s->ents[i].address) == XFS_DIR2_NULL_DATAPTR)
L
Linus Torvalds 已提交
746 747 748 749 750 751 752 753 754 755 756 757 758 759
				stale++;
		}
	} else
		stale = 0;
	/*
	 * Copy the leaf entries from source to destination.
	 */
	memcpy(&leaf_d->ents[start_d], &leaf_s->ents[start_s],
		count * sizeof(xfs_dir2_leaf_entry_t));
	xfs_dir2_leaf_log_ents(tp, bp_d, start_d, start_d + count - 1);
	/*
	 * If there are source entries after the ones we copied,
	 * delete the ones we copied by sliding the next ones down.
	 */
760
	if (start_s + count < be16_to_cpu(leaf_s->hdr.count)) {
L
Linus Torvalds 已提交
761 762 763 764 765 766 767
		memmove(&leaf_s->ents[start_s], &leaf_s->ents[start_s + count],
			count * sizeof(xfs_dir2_leaf_entry_t));
		xfs_dir2_leaf_log_ents(tp, bp_s, start_s, start_s + count - 1);
	}
	/*
	 * Update the headers and log them.
	 */
768 769 770 771
	be16_add_cpu(&leaf_s->hdr.count, -(count));
	be16_add_cpu(&leaf_s->hdr.stale, -(stale));
	be16_add_cpu(&leaf_d->hdr.count, count);
	be16_add_cpu(&leaf_d->hdr.stale, stale);
L
Linus Torvalds 已提交
772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791
	xfs_dir2_leaf_log_header(tp, bp_s);
	xfs_dir2_leaf_log_header(tp, bp_d);
	xfs_dir2_leafn_check(args->dp, bp_s);
	xfs_dir2_leafn_check(args->dp, bp_d);
}

/*
 * Determine the sort order of two leaf blocks.
 * Returns 1 if both are valid and leaf2 should be before leaf1, else 0.
 */
int						/* sort order */
xfs_dir2_leafn_order(
	xfs_dabuf_t	*leaf1_bp,		/* leaf1 buffer */
	xfs_dabuf_t	*leaf2_bp)		/* leaf2 buffer */
{
	xfs_dir2_leaf_t	*leaf1;			/* leaf1 structure */
	xfs_dir2_leaf_t	*leaf2;			/* leaf2 structure */

	leaf1 = leaf1_bp->data;
	leaf2 = leaf2_bp->data;
792 793
	ASSERT(be16_to_cpu(leaf1->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
	ASSERT(be16_to_cpu(leaf2->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
794 795
	if (be16_to_cpu(leaf1->hdr.count) > 0 &&
	    be16_to_cpu(leaf2->hdr.count) > 0 &&
796 797 798
	    (be32_to_cpu(leaf2->ents[0].hashval) < be32_to_cpu(leaf1->ents[0].hashval) ||
	     be32_to_cpu(leaf2->ents[be16_to_cpu(leaf2->hdr.count) - 1].hashval) <
	     be32_to_cpu(leaf1->ents[be16_to_cpu(leaf1->hdr.count) - 1].hashval)))
L
Linus Torvalds 已提交
799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840
		return 1;
	return 0;
}

/*
 * Rebalance leaf entries between two leaf blocks.
 * This is actually only called when the second block is new,
 * though the code deals with the general case.
 * A new entry will be inserted in one of the blocks, and that
 * entry is taken into account when balancing.
 */
static void
xfs_dir2_leafn_rebalance(
	xfs_da_state_t		*state,		/* btree cursor */
	xfs_da_state_blk_t	*blk1,		/* first btree block */
	xfs_da_state_blk_t	*blk2)		/* second btree block */
{
	xfs_da_args_t		*args;		/* operation arguments */
	int			count;		/* count (& direction) leaves */
	int			isleft;		/* new goes in left leaf */
	xfs_dir2_leaf_t		*leaf1;		/* first leaf structure */
	xfs_dir2_leaf_t		*leaf2;		/* second leaf structure */
	int			mid;		/* midpoint leaf index */
#ifdef DEBUG
	int			oldstale;	/* old count of stale leaves */
#endif
	int			oldsum;		/* old total leaf count */
	int			swap;		/* swapped leaf blocks */

	args = state->args;
	/*
	 * If the block order is wrong, swap the arguments.
	 */
	if ((swap = xfs_dir2_leafn_order(blk1->bp, blk2->bp))) {
		xfs_da_state_blk_t	*tmp;	/* temp for block swap */

		tmp = blk1;
		blk1 = blk2;
		blk2 = tmp;
	}
	leaf1 = blk1->bp->data;
	leaf2 = blk2->bp->data;
841
	oldsum = be16_to_cpu(leaf1->hdr.count) + be16_to_cpu(leaf2->hdr.count);
L
Linus Torvalds 已提交
842
#ifdef DEBUG
843
	oldstale = be16_to_cpu(leaf1->hdr.stale) + be16_to_cpu(leaf2->hdr.stale);
L
Linus Torvalds 已提交
844 845 846 847 848 849 850 851 852
#endif
	mid = oldsum >> 1;
	/*
	 * If the old leaf count was odd then the new one will be even,
	 * so we need to divide the new count evenly.
	 */
	if (oldsum & 1) {
		xfs_dahash_t	midhash;	/* middle entry hash value */

853
		if (mid >= be16_to_cpu(leaf1->hdr.count))
854
			midhash = be32_to_cpu(leaf2->ents[mid - be16_to_cpu(leaf1->hdr.count)].hashval);
L
Linus Torvalds 已提交
855
		else
856
			midhash = be32_to_cpu(leaf1->ents[mid].hashval);
L
Linus Torvalds 已提交
857 858 859 860 861 862 863 864 865 866 867 868 869
		isleft = args->hashval <= midhash;
	}
	/*
	 * If the old count is even then the new count is odd, so there's
	 * no preferred side for the new entry.
	 * Pick the left one.
	 */
	else
		isleft = 1;
	/*
	 * Calculate moved entry count.  Positive means left-to-right,
	 * negative means right-to-left.  Then move the entries.
	 */
870
	count = be16_to_cpu(leaf1->hdr.count) - mid + (isleft == 0);
L
Linus Torvalds 已提交
871 872
	if (count > 0)
		xfs_dir2_leafn_moveents(args, blk1->bp,
873
			be16_to_cpu(leaf1->hdr.count) - count, blk2->bp, 0, count);
L
Linus Torvalds 已提交
874 875
	else if (count < 0)
		xfs_dir2_leafn_moveents(args, blk2->bp, 0, blk1->bp,
876 877 878
			be16_to_cpu(leaf1->hdr.count), count);
	ASSERT(be16_to_cpu(leaf1->hdr.count) + be16_to_cpu(leaf2->hdr.count) == oldsum);
	ASSERT(be16_to_cpu(leaf1->hdr.stale) + be16_to_cpu(leaf2->hdr.stale) == oldstale);
L
Linus Torvalds 已提交
879 880 881
	/*
	 * Mark whether we're inserting into the old or new leaf.
	 */
882
	if (be16_to_cpu(leaf1->hdr.count) < be16_to_cpu(leaf2->hdr.count))
L
Linus Torvalds 已提交
883
		state->inleaf = swap;
884
	else if (be16_to_cpu(leaf1->hdr.count) > be16_to_cpu(leaf2->hdr.count))
L
Linus Torvalds 已提交
885 886 887
		state->inleaf = !swap;
	else
		state->inleaf =
888
			swap ^ (blk1->index <= be16_to_cpu(leaf1->hdr.count));
L
Linus Torvalds 已提交
889 890 891 892
	/*
	 * Adjust the expected index for insertion.
	 */
	if (!state->inleaf)
893
		blk2->index = blk1->index - be16_to_cpu(leaf1->hdr.count);
894 895 896 897

	/*
	 * Finally sanity check just to make sure we are not returning a
	 * negative index
L
Linus Torvalds 已提交
898 899 900 901 902
	 */
	if(blk2->index < 0) {
		state->inleaf = 1;
		blk2->index = 0;
		cmn_err(CE_ALERT,
903
			"xfs_dir2_leafn_rebalance: picked the wrong leaf? reverting original leaf: "
L
Linus Torvalds 已提交
904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940
			"blk1->index %d\n",
			blk1->index);
	}
}

/*
 * Remove an entry from a node directory.
 * This removes the leaf entry and the data entry,
 * and updates the free block if necessary.
 */
static int					/* error */
xfs_dir2_leafn_remove(
	xfs_da_args_t		*args,		/* operation arguments */
	xfs_dabuf_t		*bp,		/* leaf buffer */
	int			index,		/* leaf entry index */
	xfs_da_state_blk_t	*dblk,		/* data block */
	int			*rval)		/* resulting block needs join */
{
	xfs_dir2_data_t		*data;		/* data block structure */
	xfs_dir2_db_t		db;		/* data block number */
	xfs_dabuf_t		*dbp;		/* data block buffer */
	xfs_dir2_data_entry_t	*dep;		/* data block entry */
	xfs_inode_t		*dp;		/* incore directory inode */
	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
	int			longest;	/* longest data free entry */
	int			off;		/* data block entry offset */
	xfs_mount_t		*mp;		/* filesystem mount point */
	int			needlog;	/* need to log data header */
	int			needscan;	/* need to rescan data frees */
	xfs_trans_t		*tp;		/* transaction pointer */

	xfs_dir2_trace_args_sb("leafn_remove", args, index, bp);
	dp = args->dp;
	tp = args->trans;
	mp = dp->i_mount;
	leaf = bp->data;
941
	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
L
Linus Torvalds 已提交
942 943 944 945 946 947 948
	/*
	 * Point to the entry we're removing.
	 */
	lep = &leaf->ents[index];
	/*
	 * Extract the data block and offset from the entry.
	 */
949
	db = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
L
Linus Torvalds 已提交
950
	ASSERT(dblk->blkno == db);
951
	off = xfs_dir2_dataptr_to_off(mp, be32_to_cpu(lep->address));
L
Linus Torvalds 已提交
952 953 954 955 956
	ASSERT(dblk->index == off);
	/*
	 * Kill the leaf entry by marking it stale.
	 * Log the leaf block changes.
	 */
957
	be16_add_cpu(&leaf->hdr.stale, 1);
L
Linus Torvalds 已提交
958
	xfs_dir2_leaf_log_header(tp, bp);
959
	lep->address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
L
Linus Torvalds 已提交
960 961 962 963 964 965 966 967
	xfs_dir2_leaf_log_ents(tp, bp, index, index);
	/*
	 * Make the data entry free.  Keep track of the longest freespace
	 * in the data block in case it changes.
	 */
	dbp = dblk->bp;
	data = dbp->data;
	dep = (xfs_dir2_data_entry_t *)((char *)data + off);
968
	longest = be16_to_cpu(data->hdr.bestfree[0].length);
L
Linus Torvalds 已提交
969 970
	needlog = needscan = 0;
	xfs_dir2_data_make_free(tp, dbp, off,
971
		xfs_dir2_data_entsize(dep->namelen), &needlog, &needscan);
L
Linus Torvalds 已提交
972 973 974 975 976
	/*
	 * Rescan the data block freespaces for bestfree.
	 * Log the data block header if needed.
	 */
	if (needscan)
977
		xfs_dir2_data_freescan(mp, data, &needlog);
L
Linus Torvalds 已提交
978 979 980 981 982 983 984
	if (needlog)
		xfs_dir2_data_log_header(tp, dbp);
	xfs_dir2_data_check(dp, dbp);
	/*
	 * If the longest data block freespace changes, need to update
	 * the corresponding freeblock entry.
	 */
985
	if (longest < be16_to_cpu(data->hdr.bestfree[0].length)) {
L
Linus Torvalds 已提交
986 987 988 989 990 991 992 993 994 995 996
		int		error;		/* error return value */
		xfs_dabuf_t	*fbp;		/* freeblock buffer */
		xfs_dir2_db_t	fdb;		/* freeblock block number */
		int		findex;		/* index in freeblock entries */
		xfs_dir2_free_t	*free;		/* freeblock structure */
		int		logfree;	/* need to log free entry */

		/*
		 * Convert the data block number to a free block,
		 * read in the free block.
		 */
997 998
		fdb = xfs_dir2_db_to_fdb(mp, db);
		if ((error = xfs_da_read_buf(tp, dp, xfs_dir2_db_to_da(mp, fdb),
L
Linus Torvalds 已提交
999 1000 1001 1002
				-1, &fbp, XFS_DATA_FORK))) {
			return error;
		}
		free = fbp->data;
1003 1004
		ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
		ASSERT(be32_to_cpu(free->hdr.firstdb) ==
L
Linus Torvalds 已提交
1005 1006 1007 1008 1009
		       XFS_DIR2_MAX_FREE_BESTS(mp) *
		       (fdb - XFS_DIR2_FREE_FIRSTDB(mp)));
		/*
		 * Calculate which entry we need to fix.
		 */
1010
		findex = xfs_dir2_db_to_fdindex(mp, db);
1011
		longest = be16_to_cpu(data->hdr.bestfree[0].length);
L
Linus Torvalds 已提交
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
		/*
		 * If the data block is now empty we can get rid of it
		 * (usually).
		 */
		if (longest == mp->m_dirblksize - (uint)sizeof(data->hdr)) {
			/*
			 * Try to punch out the data block.
			 */
			error = xfs_dir2_shrink_inode(args, db, dbp);
			if (error == 0) {
				dblk->bp = NULL;
				data = NULL;
			}
			/*
			 * We can get ENOSPC if there's no space reservation.
			 * In this case just drop the buffer and some one else
			 * will eventually get rid of the empty block.
			 */
			else if (error == ENOSPC && args->total == 0)
				xfs_da_buf_done(dbp);
			else
				return error;
		}
		/*
		 * If we got rid of the data block, we can eliminate that entry
		 * in the free block.
		 */
		if (data == NULL) {
			/*
			 * One less used entry in the free table.
			 */
1043
			be32_add_cpu(&free->hdr.nused, -1);
L
Linus Torvalds 已提交
1044 1045 1046 1047 1048 1049 1050
			xfs_dir2_free_log_header(tp, fbp);
			/*
			 * If this was the last entry in the table, we can
			 * trim the table size back.  There might be other
			 * entries at the end referring to non-existent
			 * data blocks, get those too.
			 */
1051
			if (findex == be32_to_cpu(free->hdr.nvalid) - 1) {
L
Linus Torvalds 已提交
1052 1053 1054
				int	i;		/* free entry index */

				for (i = findex - 1;
1055
				     i >= 0 && be16_to_cpu(free->bests[i]) == NULLDATAOFF;
L
Linus Torvalds 已提交
1056 1057
				     i--)
					continue;
1058
				free->hdr.nvalid = cpu_to_be32(i + 1);
L
Linus Torvalds 已提交
1059 1060 1061 1062 1063 1064
				logfree = 0;
			}
			/*
			 * Not the last entry, just punch it out.
			 */
			else {
1065
				free->bests[findex] = cpu_to_be16(NULLDATAOFF);
L
Linus Torvalds 已提交
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
				logfree = 1;
			}
			/*
			 * If there are no useful entries left in the block,
			 * get rid of the block if we can.
			 */
			if (!free->hdr.nused) {
				error = xfs_dir2_shrink_inode(args, fdb, fbp);
				if (error == 0) {
					fbp = NULL;
					logfree = 0;
				} else if (error != ENOSPC || args->total != 0)
					return error;
				/*
				 * It's possible to get ENOSPC if there is no
				 * space reservation.  In this case some one
				 * else will eventually get rid of this block.
				 */
			}
		}
		/*
		 * Data block is not empty, just set the free entry to
		 * the new value.
		 */
		else {
1091
			free->bests[findex] = cpu_to_be16(longest);
L
Linus Torvalds 已提交
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106
			logfree = 1;
		}
		/*
		 * Log the free entry that changed, unless we got rid of it.
		 */
		if (logfree)
			xfs_dir2_free_log_bests(tp, fbp, findex, findex);
		/*
		 * Drop the buffer if we still have it.
		 */
		if (fbp)
			xfs_da_buf_done(fbp);
	}
	xfs_dir2_leafn_check(dp, bp);
	/*
M
Malcolm Parsons 已提交
1107
	 * Return indication of whether this leaf block is empty enough
L
Linus Torvalds 已提交
1108 1109 1110 1111 1112
	 * to justify trying to join it with a neighbor.
	 */
	*rval =
		((uint)sizeof(leaf->hdr) +
		 (uint)sizeof(leaf->ents[0]) *
1113
		 (be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale))) <
L
Linus Torvalds 已提交
1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145
		mp->m_dir_magicpct;
	return 0;
}

/*
 * Split the leaf entries in the old block into old and new blocks.
 */
int						/* error */
xfs_dir2_leafn_split(
	xfs_da_state_t		*state,		/* btree cursor */
	xfs_da_state_blk_t	*oldblk,	/* original block */
	xfs_da_state_blk_t	*newblk)	/* newly created block */
{
	xfs_da_args_t		*args;		/* operation arguments */
	xfs_dablk_t		blkno;		/* new leaf block number */
	int			error;		/* error return value */
	xfs_mount_t		*mp;		/* filesystem mount point */

	/*
	 * Allocate space for a new leaf node.
	 */
	args = state->args;
	mp = args->dp->i_mount;
	ASSERT(args != NULL);
	ASSERT(oldblk->magic == XFS_DIR2_LEAFN_MAGIC);
	error = xfs_da_grow_inode(args, &blkno);
	if (error) {
		return error;
	}
	/*
	 * Initialize the new leaf block.
	 */
1146
	error = xfs_dir2_leaf_init(args, xfs_dir2_da_to_db(mp, blkno),
L
Linus Torvalds 已提交
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
		&newblk->bp, XFS_DIR2_LEAFN_MAGIC);
	if (error) {
		return error;
	}
	newblk->blkno = blkno;
	newblk->magic = XFS_DIR2_LEAFN_MAGIC;
	/*
	 * Rebalance the entries across the two leaves, link the new
	 * block into the leaves.
	 */
	xfs_dir2_leafn_rebalance(state, oldblk, newblk);
	error = xfs_da_blk_link(state, oldblk, newblk);
	if (error) {
		return error;
	}
	/*
	 * Insert the new entry in the correct block.
	 */
	if (state->inleaf)
		error = xfs_dir2_leafn_add(oldblk->bp, args, oldblk->index);
	else
		error = xfs_dir2_leafn_add(newblk->bp, args, newblk->index);
	/*
	 * Update last hashval in each block since we added the name.
	 */
	oldblk->hashval = xfs_dir2_leafn_lasthash(oldblk->bp, NULL);
	newblk->hashval = xfs_dir2_leafn_lasthash(newblk->bp, NULL);
	xfs_dir2_leafn_check(args->dp, oldblk->bp);
	xfs_dir2_leafn_check(args->dp, newblk->bp);
	return error;
}

/*
 * Check a leaf block and its neighbors to see if the block should be
 * collapsed into one or the other neighbor.  Always keep the block
 * with the smaller block number.
 * If the current block is over 50% full, don't try to join it, return 0.
 * If the block is empty, fill in the state structure and return 2.
 * If it can be collapsed, fill in the state structure and return 1.
 * If nothing can be done, return 0.
 */
int						/* error */
xfs_dir2_leafn_toosmall(
	xfs_da_state_t		*state,		/* btree cursor */
	int			*action)	/* resulting action to take */
{
	xfs_da_state_blk_t	*blk;		/* leaf block */
	xfs_dablk_t		blkno;		/* leaf block number */
	xfs_dabuf_t		*bp;		/* leaf buffer */
	int			bytes;		/* bytes in use */
	int			count;		/* leaf live entry count */
	int			error;		/* error return value */
	int			forward;	/* sibling block direction */
	int			i;		/* sibling counter */
	xfs_da_blkinfo_t	*info;		/* leaf block header */
	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
	int			rval;		/* result from path_shift */

	/*
	 * Check for the degenerate case of the block being over 50% full.
	 * If so, it's not worth even looking to see if we might be able
	 * to coalesce with a sibling.
	 */
	blk = &state->path.blk[state->path.active - 1];
	info = blk->bp->data;
1212
	ASSERT(be16_to_cpu(info->magic) == XFS_DIR2_LEAFN_MAGIC);
L
Linus Torvalds 已提交
1213
	leaf = (xfs_dir2_leaf_t *)info;
1214
	count = be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale);
L
Linus Torvalds 已提交
1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
	bytes = (uint)sizeof(leaf->hdr) + count * (uint)sizeof(leaf->ents[0]);
	if (bytes > (state->blocksize >> 1)) {
		/*
		 * Blk over 50%, don't try to join.
		 */
		*action = 0;
		return 0;
	}
	/*
	 * Check for the degenerate case of the block being empty.
	 * If the block is empty, we'll simply delete it, no need to
	 * coalesce it with a sibling block.  We choose (arbitrarily)
	 * to merge with the forward block unless it is NULL.
	 */
	if (count == 0) {
		/*
		 * Make altpath point to the block we want to keep and
		 * path point to the block we want to drop (this one).
		 */
1234
		forward = (info->forw != 0);
L
Linus Torvalds 已提交
1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249
		memcpy(&state->altpath, &state->path, sizeof(state->path));
		error = xfs_da_path_shift(state, &state->altpath, forward, 0,
			&rval);
		if (error)
			return error;
		*action = rval ? 2 : 0;
		return 0;
	}
	/*
	 * Examine each sibling block to see if we can coalesce with
	 * at least 25% free space to spare.  We need to figure out
	 * whether to merge with the forward or the backward block.
	 * We prefer coalescing with the lower numbered sibling so as
	 * to shrink a directory over time.
	 */
1250
	forward = be32_to_cpu(info->forw) < be32_to_cpu(info->back);
L
Linus Torvalds 已提交
1251
	for (i = 0, bp = NULL; i < 2; forward = !forward, i++) {
1252
		blkno = forward ? be32_to_cpu(info->forw) : be32_to_cpu(info->back);
L
Linus Torvalds 已提交
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267
		if (blkno == 0)
			continue;
		/*
		 * Read the sibling leaf block.
		 */
		if ((error =
		    xfs_da_read_buf(state->args->trans, state->args->dp, blkno,
			    -1, &bp, XFS_DATA_FORK))) {
			return error;
		}
		ASSERT(bp != NULL);
		/*
		 * Count bytes in the two blocks combined.
		 */
		leaf = (xfs_dir2_leaf_t *)info;
1268
		count = be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale);
L
Linus Torvalds 已提交
1269 1270
		bytes = state->blocksize - (state->blocksize >> 2);
		leaf = bp->data;
1271
		ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
1272
		count += be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale);
L
Linus Torvalds 已提交
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329
		bytes -= count * (uint)sizeof(leaf->ents[0]);
		/*
		 * Fits with at least 25% to spare.
		 */
		if (bytes >= 0)
			break;
		xfs_da_brelse(state->args->trans, bp);
	}
	/*
	 * Didn't like either block, give up.
	 */
	if (i >= 2) {
		*action = 0;
		return 0;
	}
	/*
	 * Done with the sibling leaf block here, drop the dabuf
	 * so path_shift can get it.
	 */
	xfs_da_buf_done(bp);
	/*
	 * Make altpath point to the block we want to keep (the lower
	 * numbered block) and path point to the block we want to drop.
	 */
	memcpy(&state->altpath, &state->path, sizeof(state->path));
	if (blkno < blk->blkno)
		error = xfs_da_path_shift(state, &state->altpath, forward, 0,
			&rval);
	else
		error = xfs_da_path_shift(state, &state->path, forward, 0,
			&rval);
	if (error) {
		return error;
	}
	*action = rval ? 0 : 1;
	return 0;
}

/*
 * Move all the leaf entries from drop_blk to save_blk.
 * This is done as part of a join operation.
 */
void
xfs_dir2_leafn_unbalance(
	xfs_da_state_t		*state,		/* cursor */
	xfs_da_state_blk_t	*drop_blk,	/* dead block */
	xfs_da_state_blk_t	*save_blk)	/* surviving block */
{
	xfs_da_args_t		*args;		/* operation arguments */
	xfs_dir2_leaf_t		*drop_leaf;	/* dead leaf structure */
	xfs_dir2_leaf_t		*save_leaf;	/* surviving leaf structure */

	args = state->args;
	ASSERT(drop_blk->magic == XFS_DIR2_LEAFN_MAGIC);
	ASSERT(save_blk->magic == XFS_DIR2_LEAFN_MAGIC);
	drop_leaf = drop_blk->bp->data;
	save_leaf = save_blk->bp->data;
1330 1331
	ASSERT(be16_to_cpu(drop_leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
	ASSERT(be16_to_cpu(save_leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
L
Linus Torvalds 已提交
1332 1333 1334 1335
	/*
	 * If there are any stale leaf entries, take this opportunity
	 * to purge them.
	 */
1336
	if (drop_leaf->hdr.stale)
L
Linus Torvalds 已提交
1337
		xfs_dir2_leaf_compact(args, drop_blk->bp);
1338
	if (save_leaf->hdr.stale)
L
Linus Torvalds 已提交
1339 1340 1341 1342
		xfs_dir2_leaf_compact(args, save_blk->bp);
	/*
	 * Move the entries from drop to the appropriate end of save.
	 */
1343
	drop_blk->hashval = be32_to_cpu(drop_leaf->ents[be16_to_cpu(drop_leaf->hdr.count) - 1].hashval);
L
Linus Torvalds 已提交
1344 1345
	if (xfs_dir2_leafn_order(save_blk->bp, drop_blk->bp))
		xfs_dir2_leafn_moveents(args, drop_blk->bp, 0, save_blk->bp, 0,
1346
			be16_to_cpu(drop_leaf->hdr.count));
L
Linus Torvalds 已提交
1347 1348
	else
		xfs_dir2_leafn_moveents(args, drop_blk->bp, 0, save_blk->bp,
1349
			be16_to_cpu(save_leaf->hdr.count), be16_to_cpu(drop_leaf->hdr.count));
1350
	save_blk->hashval = be32_to_cpu(save_leaf->ents[be16_to_cpu(save_leaf->hdr.count) - 1].hashval);
L
Linus Torvalds 已提交
1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 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
	xfs_dir2_leafn_check(args->dp, save_blk->bp);
}

/*
 * Top-level node form directory addname routine.
 */
int						/* error */
xfs_dir2_node_addname(
	xfs_da_args_t		*args)		/* operation arguments */
{
	xfs_da_state_blk_t	*blk;		/* leaf block for insert */
	int			error;		/* error return value */
	int			rval;		/* sub-return value */
	xfs_da_state_t		*state;		/* btree cursor */

	xfs_dir2_trace_args("node_addname", args);
	/*
	 * Allocate and initialize the state (btree cursor).
	 */
	state = xfs_da_state_alloc();
	state->args = args;
	state->mp = args->dp->i_mount;
	state->blocksize = state->mp->m_dirblksize;
	state->node_ents = state->mp->m_dir_node_ents;
	/*
	 * Look up the name.  We're not supposed to find it, but
	 * this gives us the insertion point.
	 */
	error = xfs_da_node_lookup_int(state, &rval);
	if (error)
		rval = error;
	if (rval != ENOENT) {
		goto done;
	}
	/*
	 * Add the data entry to a data block.
	 * Extravalid is set to a freeblock found by lookup.
	 */
	rval = xfs_dir2_node_addname_int(args,
		state->extravalid ? &state->extrablk : NULL);
	if (rval) {
		goto done;
	}
	blk = &state->path.blk[state->path.active - 1];
	ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
	/*
	 * Add the new leaf entry.
	 */
	rval = xfs_dir2_leafn_add(blk->bp, args, blk->index);
	if (rval == 0) {
		/*
		 * It worked, fix the hash values up the btree.
		 */
1404
		if (!(args->op_flags & XFS_DA_OP_JUSTCHECK))
L
Linus Torvalds 已提交
1405 1406 1407 1408 1409 1410 1411 1412 1413 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
			xfs_da_fixhashpath(state, &state->path);
	} else {
		/*
		 * It didn't work, we need to split the leaf block.
		 */
		if (args->total == 0) {
			ASSERT(rval == ENOSPC);
			goto done;
		}
		/*
		 * Split the leaf block and insert the new entry.
		 */
		rval = xfs_da_split(state);
	}
done:
	xfs_da_state_free(state);
	return rval;
}

/*
 * Add the data entry for a node-format directory name addition.
 * The leaf entry is added in xfs_dir2_leafn_add.
 * We may enter with a freespace block that the lookup found.
 */
static int					/* error */
xfs_dir2_node_addname_int(
	xfs_da_args_t		*args,		/* operation arguments */
	xfs_da_state_blk_t	*fblk)		/* optional freespace block */
{
	xfs_dir2_data_t		*data;		/* data block structure */
	xfs_dir2_db_t		dbno;		/* data block number */
	xfs_dabuf_t		*dbp;		/* data block buffer */
	xfs_dir2_data_entry_t	*dep;		/* data entry pointer */
	xfs_inode_t		*dp;		/* incore directory inode */
	xfs_dir2_data_unused_t	*dup;		/* data unused entry pointer */
	int			error;		/* error return value */
	xfs_dir2_db_t		fbno;		/* freespace block number */
	xfs_dabuf_t		*fbp;		/* freespace buffer */
	int			findex;		/* freespace entry index */
	xfs_dir2_free_t		*free=NULL;	/* freespace block structure */
	xfs_dir2_db_t		ifbno;		/* initial freespace block no */
	xfs_dir2_db_t		lastfbno=0;	/* highest freespace block no */
	int			length;		/* length of the new entry */
	int			logfree;	/* need to log free entry */
	xfs_mount_t		*mp;		/* filesystem mount point */
	int			needlog;	/* need to log data header */
	int			needscan;	/* need to rescan data frees */
1452
	__be16			*tagp;		/* data entry tag pointer */
L
Linus Torvalds 已提交
1453 1454 1455 1456 1457
	xfs_trans_t		*tp;		/* transaction pointer */

	dp = args->dp;
	mp = dp->i_mount;
	tp = args->trans;
1458
	length = xfs_dir2_data_entsize(args->namelen);
L
Linus Torvalds 已提交
1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470
	/*
	 * If we came in with a freespace block that means that lookup
	 * found an entry with our hash value.  This is the freespace
	 * block for that data entry.
	 */
	if (fblk) {
		fbp = fblk->bp;
		/*
		 * Remember initial freespace block number.
		 */
		ifbno = fblk->blkno;
		free = fbp->data;
1471
		ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
L
Linus Torvalds 已提交
1472 1473 1474 1475 1476 1477 1478
		findex = fblk->index;
		/*
		 * This means the free entry showed that the data block had
		 * space for our entry, so we remembered it.
		 * Use that data block.
		 */
		if (findex >= 0) {
1479 1480 1481 1482
			ASSERT(findex < be32_to_cpu(free->hdr.nvalid));
			ASSERT(be16_to_cpu(free->bests[findex]) != NULLDATAOFF);
			ASSERT(be16_to_cpu(free->bests[findex]) >= length);
			dbno = be32_to_cpu(free->hdr.firstdb) + findex;
L
Linus Torvalds 已提交
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
		}
		/*
		 * The data block looked at didn't have enough room.
		 * We'll start at the beginning of the freespace entries.
		 */
		else {
			dbno = -1;
			findex = 0;
		}
	}
	/*
	 * Didn't come in with a freespace block, so don't have a data block.
	 */
	else {
		ifbno = dbno = -1;
		fbp = NULL;
		findex = 0;
	}
	/*
	 * If we don't have a data block yet, we're going to scan the
	 * freespace blocks looking for one.  Figure out what the
	 * highest freespace block number is.
	 */
	if (dbno == -1) {
		xfs_fileoff_t	fo;		/* freespace block number */

		if ((error = xfs_bmap_last_offset(tp, dp, &fo, XFS_DATA_FORK)))
			return error;
1511
		lastfbno = xfs_dir2_da_to_db(mp, (xfs_dablk_t)fo);
L
Linus Torvalds 已提交
1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546
		fbno = ifbno;
	}
	/*
	 * While we haven't identified a data block, search the freeblock
	 * data for a good data block.  If we find a null freeblock entry,
	 * indicating a hole in the data blocks, remember that.
	 */
	while (dbno == -1) {
		/*
		 * If we don't have a freeblock in hand, get the next one.
		 */
		if (fbp == NULL) {
			/*
			 * Happens the first time through unless lookup gave
			 * us a freespace block to start with.
			 */
			if (++fbno == 0)
				fbno = XFS_DIR2_FREE_FIRSTDB(mp);
			/*
			 * If it's ifbno we already looked at it.
			 */
			if (fbno == ifbno)
				fbno++;
			/*
			 * If it's off the end we're done.
			 */
			if (fbno >= lastfbno)
				break;
			/*
			 * Read the block.  There can be holes in the
			 * freespace blocks, so this might not succeed.
			 * This should be really rare, so there's no reason
			 * to avoid it.
			 */
			if ((error = xfs_da_read_buf(tp, dp,
1547
					xfs_dir2_db_to_da(mp, fbno), -2, &fbp,
L
Linus Torvalds 已提交
1548 1549 1550 1551 1552 1553 1554
					XFS_DATA_FORK))) {
				return error;
			}
			if (unlikely(fbp == NULL)) {
				continue;
			}
			free = fbp->data;
1555
			ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
L
Linus Torvalds 已提交
1556 1557 1558 1559 1560
			findex = 0;
		}
		/*
		 * Look at the current free entry.  Is it good enough?
		 */
1561 1562 1563
		if (be16_to_cpu(free->bests[findex]) != NULLDATAOFF &&
		    be16_to_cpu(free->bests[findex]) >= length)
			dbno = be32_to_cpu(free->hdr.firstdb) + findex;
L
Linus Torvalds 已提交
1564 1565 1566 1567
		else {
			/*
			 * Are we done with the freeblock?
			 */
1568
			if (++findex == be32_to_cpu(free->hdr.nvalid)) {
L
Linus Torvalds 已提交
1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586
				/*
				 * Drop the block.
				 */
				xfs_da_brelse(tp, fbp);
				fbp = NULL;
				if (fblk && fblk->bp)
					fblk->bp = NULL;
			}
		}
	}
	/*
	 * If we don't have a data block, we need to allocate one and make
	 * the freespace entries refer to it.
	 */
	if (unlikely(dbno == -1)) {
		/*
		 * Not allowed to allocate, return failure.
		 */
1587 1588
		if ((args->op_flags & XFS_DA_OP_JUSTCHECK) ||
							args->total == 0) {
L
Linus Torvalds 已提交
1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623
			/*
			 * Drop the freespace buffer unless it came from our
			 * caller.
			 */
			if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
				xfs_da_buf_done(fbp);
			return XFS_ERROR(ENOSPC);
		}
		/*
		 * Allocate and initialize the new data block.
		 */
		if (unlikely((error = xfs_dir2_grow_inode(args,
							 XFS_DIR2_DATA_SPACE,
							 &dbno)) ||
		    (error = xfs_dir2_data_init(args, dbno, &dbp)))) {
			/*
			 * Drop the freespace buffer unless it came from our
			 * caller.
			 */
			if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
				xfs_da_buf_done(fbp);
			return error;
		}
		/*
		 * If (somehow) we have a freespace block, get rid of it.
		 */
		if (fbp)
			xfs_da_brelse(tp, fbp);
		if (fblk && fblk->bp)
			fblk->bp = NULL;

		/*
		 * Get the freespace block corresponding to the data block
		 * that was just allocated.
		 */
1624
		fbno = xfs_dir2_db_to_fdb(mp, dbno);
L
Linus Torvalds 已提交
1625
		if (unlikely(error = xfs_da_read_buf(tp, dp,
1626
				xfs_dir2_db_to_da(mp, fbno), -2, &fbp,
L
Linus Torvalds 已提交
1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640
				XFS_DATA_FORK))) {
			xfs_da_buf_done(dbp);
			return error;
  		}
		/*
		 * If there wasn't a freespace block, the read will
		 * return a NULL fbp.  Allocate and initialize a new one.
		 */
		if( fbp == NULL ) {
			if ((error = xfs_dir2_grow_inode(args, XFS_DIR2_FREE_SPACE,
							&fbno))) {
				return error;
			}

1641
			if (unlikely(xfs_dir2_db_to_fdb(mp, dbno) != fbno)) {
L
Linus Torvalds 已提交
1642 1643 1644 1645 1646 1647
				cmn_err(CE_ALERT,
					"xfs_dir2_node_addname_int: dir ino "
					"%llu needed freesp block %lld for\n"
					"  data block %lld, got %lld\n"
					"  ifbno %llu lastfbno %d\n",
					(unsigned long long)dp->i_ino,
1648
					(long long)xfs_dir2_db_to_fdb(mp, dbno),
L
Linus Torvalds 已提交
1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671
					(long long)dbno, (long long)fbno,
					(unsigned long long)ifbno, lastfbno);
				if (fblk) {
					cmn_err(CE_ALERT,
						" fblk 0x%p blkno %llu "
						"index %d magic 0x%x\n",
						fblk,
						(unsigned long long)fblk->blkno,
						fblk->index,
						fblk->magic);
				} else {
					cmn_err(CE_ALERT,
						" ... fblk is NULL\n");
				}
				XFS_ERROR_REPORT("xfs_dir2_node_addname_int",
						 XFS_ERRLEVEL_LOW, mp);
				return XFS_ERROR(EFSCORRUPTED);
			}

			/*
			 * Get a buffer for the new block.
			 */
			if ((error = xfs_da_get_buf(tp, dp,
1672
						   xfs_dir2_db_to_da(mp, fbno),
L
Linus Torvalds 已提交
1673 1674 1675 1676 1677 1678 1679 1680 1681 1682
						   -1, &fbp, XFS_DATA_FORK))) {
				return error;
			}
			ASSERT(fbp != NULL);

			/*
			 * Initialize the new block to be empty, and remember
			 * its first slot as our empty slot.
			 */
			free = fbp->data;
1683 1684
			free->hdr.magic = cpu_to_be32(XFS_DIR2_FREE_MAGIC);
			free->hdr.firstdb = cpu_to_be32(
L
Linus Torvalds 已提交
1685 1686 1687 1688 1689 1690
				(fbno - XFS_DIR2_FREE_FIRSTDB(mp)) *
				XFS_DIR2_MAX_FREE_BESTS(mp));
			free->hdr.nvalid = 0;
			free->hdr.nused = 0;
		} else {
			free = fbp->data;
1691
			ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
L
Linus Torvalds 已提交
1692 1693 1694 1695 1696
		}

		/*
		 * Set the freespace block index from the data block number.
		 */
1697
		findex = xfs_dir2_db_to_fdindex(mp, dbno);
L
Linus Torvalds 已提交
1698 1699 1700 1701
		/*
		 * If it's after the end of the current entries in the
		 * freespace block, extend that table.
		 */
1702
		if (findex >= be32_to_cpu(free->hdr.nvalid)) {
L
Linus Torvalds 已提交
1703
			ASSERT(findex < XFS_DIR2_MAX_FREE_BESTS(mp));
1704
			free->hdr.nvalid = cpu_to_be32(findex + 1);
L
Linus Torvalds 已提交
1705 1706 1707
			/*
			 * Tag new entry so nused will go up.
			 */
1708
			free->bests[findex] = cpu_to_be16(NULLDATAOFF);
L
Linus Torvalds 已提交
1709 1710 1711 1712 1713
		}
		/*
		 * If this entry was for an empty data block
		 * (this should always be true) then update the header.
		 */
1714
		if (be16_to_cpu(free->bests[findex]) == NULLDATAOFF) {
1715
			be32_add_cpu(&free->hdr.nused, 1);
L
Linus Torvalds 已提交
1716 1717 1718 1719 1720 1721 1722 1723
			xfs_dir2_free_log_header(tp, fbp);
		}
		/*
		 * Update the real value in the table.
		 * We haven't allocated the data entry yet so this will
		 * change again.
		 */
		data = dbp->data;
1724
		free->bests[findex] = data->hdr.bestfree[0].length;
L
Linus Torvalds 已提交
1725 1726 1727 1728 1729 1730 1731 1732 1733
		logfree = 1;
	}
	/*
	 * We had a data block so we don't have to make a new one.
	 */
	else {
		/*
		 * If just checking, we succeeded.
		 */
1734
		if (args->op_flags & XFS_DA_OP_JUSTCHECK) {
L
Linus Torvalds 已提交
1735 1736 1737 1738 1739 1740 1741 1742
			if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
				xfs_da_buf_done(fbp);
			return 0;
		}
		/*
		 * Read the data block in.
		 */
		if (unlikely(
1743
		    error = xfs_da_read_buf(tp, dp, xfs_dir2_db_to_da(mp, dbno),
L
Linus Torvalds 已提交
1744 1745 1746 1747 1748 1749 1750 1751
				-1, &dbp, XFS_DATA_FORK))) {
			if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
				xfs_da_buf_done(fbp);
			return error;
		}
		data = dbp->data;
		logfree = 0;
	}
1752
	ASSERT(be16_to_cpu(data->hdr.bestfree[0].length) >= length);
L
Linus Torvalds 已提交
1753 1754 1755 1756
	/*
	 * Point to the existing unused space.
	 */
	dup = (xfs_dir2_data_unused_t *)
1757
	      ((char *)data + be16_to_cpu(data->hdr.bestfree[0].offset));
L
Linus Torvalds 已提交
1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768
	needscan = needlog = 0;
	/*
	 * Mark the first part of the unused space, inuse for us.
	 */
	xfs_dir2_data_use_free(tp, dbp, dup,
		(xfs_dir2_data_aoff_t)((char *)dup - (char *)data), length,
		&needlog, &needscan);
	/*
	 * Fill in the new entry and log it.
	 */
	dep = (xfs_dir2_data_entry_t *)dup;
1769
	dep->inumber = cpu_to_be64(args->inumber);
L
Linus Torvalds 已提交
1770 1771
	dep->namelen = args->namelen;
	memcpy(dep->name, args->name, dep->namelen);
1772
	tagp = xfs_dir2_data_entry_tag_p(dep);
1773
	*tagp = cpu_to_be16((char *)dep - (char *)data);
L
Linus Torvalds 已提交
1774 1775 1776 1777 1778
	xfs_dir2_data_log_entry(tp, dbp, dep);
	/*
	 * Rescan the block for bestfree if needed.
	 */
	if (needscan)
1779
		xfs_dir2_data_freescan(mp, data, &needlog);
L
Linus Torvalds 已提交
1780 1781 1782 1783 1784 1785 1786 1787
	/*
	 * Log the data block header if needed.
	 */
	if (needlog)
		xfs_dir2_data_log_header(tp, dbp);
	/*
	 * If the freespace entry is now wrong, update it.
	 */
1788
	if (be16_to_cpu(free->bests[findex]) != be16_to_cpu(data->hdr.bestfree[0].length)) {
1789
		free->bests[findex] = data->hdr.bestfree[0].length;
L
Linus Torvalds 已提交
1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805
		logfree = 1;
	}
	/*
	 * Log the freespace entry if needed.
	 */
	if (logfree)
		xfs_dir2_free_log_bests(tp, fbp, findex, findex);
	/*
	 * If the caller didn't hand us the freespace block, drop it.
	 */
	if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
		xfs_da_buf_done(fbp);
	/*
	 * Return the data block and offset in args, then drop the data block.
	 */
	args->blkno = (xfs_dablk_t)dbno;
1806
	args->index = be16_to_cpu(*tagp);
L
Linus Torvalds 已提交
1807 1808 1809 1810 1811 1812 1813 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
	xfs_da_buf_done(dbp);
	return 0;
}

/*
 * Lookup an entry in a node-format directory.
 * All the real work happens in xfs_da_node_lookup_int.
 * The only real output is the inode number of the entry.
 */
int						/* error */
xfs_dir2_node_lookup(
	xfs_da_args_t	*args)			/* operation arguments */
{
	int		error;			/* error return value */
	int		i;			/* btree level */
	int		rval;			/* operation return value */
	xfs_da_state_t	*state;			/* btree cursor */

	xfs_dir2_trace_args("node_lookup", args);
	/*
	 * Allocate and initialize the btree cursor.
	 */
	state = xfs_da_state_alloc();
	state->args = args;
	state->mp = args->dp->i_mount;
	state->blocksize = state->mp->m_dirblksize;
	state->node_ents = state->mp->m_dir_node_ents;
	/*
	 * Fill in the path to the entry in the cursor.
	 */
	error = xfs_da_node_lookup_int(state, &rval);
	if (error)
		rval = error;
1840 1841 1842 1843 1844 1845 1846 1847
	else if (rval == ENOENT && args->cmpresult == XFS_CMP_CASE) {
		/* If a CI match, dup the actual name and return EEXIST */
		xfs_dir2_data_entry_t	*dep;

		dep = (xfs_dir2_data_entry_t *)((char *)state->extrablk.bp->
						data + state->extrablk.index);
		rval = xfs_dir_cilookup_result(args, dep->name, dep->namelen);
	}
L
Linus Torvalds 已提交
1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 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 1889 1890
	/*
	 * Release the btree blocks and leaf block.
	 */
	for (i = 0; i < state->path.active; i++) {
		xfs_da_brelse(args->trans, state->path.blk[i].bp);
		state->path.blk[i].bp = NULL;
	}
	/*
	 * Release the data block if we have it.
	 */
	if (state->extravalid && state->extrablk.bp) {
		xfs_da_brelse(args->trans, state->extrablk.bp);
		state->extrablk.bp = NULL;
	}
	xfs_da_state_free(state);
	return rval;
}

/*
 * Remove an entry from a node-format directory.
 */
int						/* error */
xfs_dir2_node_removename(
	xfs_da_args_t		*args)		/* operation arguments */
{
	xfs_da_state_blk_t	*blk;		/* leaf block */
	int			error;		/* error return value */
	int			rval;		/* operation return value */
	xfs_da_state_t		*state;		/* btree cursor */

	xfs_dir2_trace_args("node_removename", args);
	/*
	 * Allocate and initialize the btree cursor.
	 */
	state = xfs_da_state_alloc();
	state->args = args;
	state->mp = args->dp->i_mount;
	state->blocksize = state->mp->m_dirblksize;
	state->node_ents = state->mp->m_dir_node_ents;
	/*
	 * Look up the entry we're deleting, set up the cursor.
	 */
	error = xfs_da_node_lookup_int(state, &rval);
1891
	if (error)
L
Linus Torvalds 已提交
1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908
		rval = error;
	/*
	 * Didn't find it, upper layer screwed up.
	 */
	if (rval != EEXIST) {
		xfs_da_state_free(state);
		return rval;
	}
	blk = &state->path.blk[state->path.active - 1];
	ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
	ASSERT(state->extravalid);
	/*
	 * Remove the leaf and data entries.
	 * Extrablk refers to the data block.
	 */
	error = xfs_dir2_leafn_remove(args, blk->bp, blk->index,
		&state->extrablk, &rval);
1909
	if (error)
L
Linus Torvalds 已提交
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 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980
		return error;
	/*
	 * Fix the hash values up the btree.
	 */
	xfs_da_fixhashpath(state, &state->path);
	/*
	 * If we need to join leaf blocks, do it.
	 */
	if (rval && state->path.active > 1)
		error = xfs_da_join(state);
	/*
	 * If no errors so far, try conversion to leaf format.
	 */
	if (!error)
		error = xfs_dir2_node_to_leaf(state);
	xfs_da_state_free(state);
	return error;
}

/*
 * Replace an entry's inode number in a node-format directory.
 */
int						/* error */
xfs_dir2_node_replace(
	xfs_da_args_t		*args)		/* operation arguments */
{
	xfs_da_state_blk_t	*blk;		/* leaf block */
	xfs_dir2_data_t		*data;		/* data block structure */
	xfs_dir2_data_entry_t	*dep;		/* data entry changed */
	int			error;		/* error return value */
	int			i;		/* btree level */
	xfs_ino_t		inum;		/* new inode number */
	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry being changed */
	int			rval;		/* internal return value */
	xfs_da_state_t		*state;		/* btree cursor */

	xfs_dir2_trace_args("node_replace", args);
	/*
	 * Allocate and initialize the btree cursor.
	 */
	state = xfs_da_state_alloc();
	state->args = args;
	state->mp = args->dp->i_mount;
	state->blocksize = state->mp->m_dirblksize;
	state->node_ents = state->mp->m_dir_node_ents;
	inum = args->inumber;
	/*
	 * Lookup the entry to change in the btree.
	 */
	error = xfs_da_node_lookup_int(state, &rval);
	if (error) {
		rval = error;
	}
	/*
	 * It should be found, since the vnodeops layer has looked it up
	 * and locked it.  But paranoia is good.
	 */
	if (rval == EEXIST) {
		/*
		 * Find the leaf entry.
		 */
		blk = &state->path.blk[state->path.active - 1];
		ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
		leaf = blk->bp->data;
		lep = &leaf->ents[blk->index];
		ASSERT(state->extravalid);
		/*
		 * Point to the data entry.
		 */
		data = state->extrablk.bp->data;
1981
		ASSERT(be32_to_cpu(data->hdr.magic) == XFS_DIR2_DATA_MAGIC);
L
Linus Torvalds 已提交
1982 1983
		dep = (xfs_dir2_data_entry_t *)
		      ((char *)data +
1984
		       xfs_dir2_dataptr_to_off(state->mp, be32_to_cpu(lep->address)));
1985
		ASSERT(inum != be64_to_cpu(dep->inumber));
L
Linus Torvalds 已提交
1986 1987 1988
		/*
		 * Fill in the new inode number and log the entry.
		 */
1989
		dep->inumber = cpu_to_be64(inum);
L
Linus Torvalds 已提交
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 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 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046
		xfs_dir2_data_log_entry(args->trans, state->extrablk.bp, dep);
		rval = 0;
	}
	/*
	 * Didn't find it, and we're holding a data block.  Drop it.
	 */
	else if (state->extravalid) {
		xfs_da_brelse(args->trans, state->extrablk.bp);
		state->extrablk.bp = NULL;
	}
	/*
	 * Release all the buffers in the cursor.
	 */
	for (i = 0; i < state->path.active; i++) {
		xfs_da_brelse(args->trans, state->path.blk[i].bp);
		state->path.blk[i].bp = NULL;
	}
	xfs_da_state_free(state);
	return rval;
}

/*
 * Trim off a trailing empty freespace block.
 * Return (in rvalp) 1 if we did it, 0 if not.
 */
int						/* error */
xfs_dir2_node_trim_free(
	xfs_da_args_t		*args,		/* operation arguments */
	xfs_fileoff_t		fo,		/* free block number */
	int			*rvalp)		/* out: did something */
{
	xfs_dabuf_t		*bp;		/* freespace buffer */
	xfs_inode_t		*dp;		/* incore directory inode */
	int			error;		/* error return code */
	xfs_dir2_free_t		*free;		/* freespace structure */
	xfs_mount_t		*mp;		/* filesystem mount point */
	xfs_trans_t		*tp;		/* transaction pointer */

	dp = args->dp;
	mp = dp->i_mount;
	tp = args->trans;
	/*
	 * Read the freespace block.
	 */
	if (unlikely(error = xfs_da_read_buf(tp, dp, (xfs_dablk_t)fo, -2, &bp,
			XFS_DATA_FORK))) {
		return error;
	}

	/*
	 * There can be holes in freespace.  If fo is a hole, there's
	 * nothing to do.
	 */
	if (bp == NULL) {
		return 0;
	}
	free = bp->data;
2047
	ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
L
Linus Torvalds 已提交
2048 2049 2050
	/*
	 * If there are used entries, there's nothing to do.
	 */
2051
	if (be32_to_cpu(free->hdr.nused) > 0) {
L
Linus Torvalds 已提交
2052 2053 2054 2055 2056 2057 2058 2059
		xfs_da_brelse(tp, bp);
		*rvalp = 0;
		return 0;
	}
	/*
	 * Blow the block away.
	 */
	if ((error =
2060
	    xfs_dir2_shrink_inode(args, xfs_dir2_da_to_db(mp, (xfs_dablk_t)fo),
L
Linus Torvalds 已提交
2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076
		    bp))) {
		/*
		 * Can't fail with ENOSPC since that only happens with no
		 * space reservation, when breaking up an extent into two
		 * pieces.  This is the last block of an extent.
		 */
		ASSERT(error != ENOSPC);
		xfs_da_brelse(tp, bp);
		return error;
	}
	/*
	 * Return that we succeeded.
	 */
	*rvalp = 1;
	return 0;
}