xfs_dir2_readdir.c 13.3 KB
Newer Older
D
Dave Chinner 已提交
1
// SPDX-License-Identifier: GPL-2.0
2 3 4 5 6 7 8
/*
 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
 * Copyright (c) 2013 Red Hat, Inc.
 * All Rights Reserved.
 */
#include "xfs.h"
#include "xfs_fs.h"
9
#include "xfs_format.h"
10 11
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
12 13
#include "xfs_bit.h"
#include "xfs_mount.h"
14
#include "xfs_da_format.h"
15 16
#include "xfs_da_btree.h"
#include "xfs_inode.h"
17
#include "xfs_dir2.h"
18 19 20 21
#include "xfs_dir2_priv.h"
#include "xfs_error.h"
#include "xfs_trace.h"
#include "xfs_bmap.h"
22
#include "xfs_trans.h"
23

24 25 26 27 28 29 30 31
/*
 * Directory file type support functions
 */
static unsigned char xfs_dir3_filetype_table[] = {
	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK,
	DT_FIFO, DT_SOCK, DT_LNK, DT_WHT,
};

D
Darrick J. Wong 已提交
32
unsigned char
33 34
xfs_dir3_get_dtype(
	struct xfs_mount	*mp,
35
	uint8_t			filetype)
36 37 38 39 40 41 42 43 44 45
{
	if (!xfs_sb_version_hasftype(&mp->m_sb))
		return DT_UNKNOWN;

	if (filetype >= XFS_DIR3_FT_MAX)
		return DT_UNKNOWN;

	return xfs_dir3_filetype_table[filetype];
}

46 47
STATIC int
xfs_dir2_sf_getdents(
48
	struct xfs_da_args	*args,
49 50 51
	struct dir_context	*ctx)
{
	int			i;		/* shortform entry number */
52
	struct xfs_inode	*dp = args->dp;	/* incore directory inode */
53 54 55 56 57 58
	xfs_dir2_dataptr_t	off;		/* current entry's offset */
	xfs_dir2_sf_entry_t	*sfep;		/* shortform directory entry */
	xfs_dir2_sf_hdr_t	*sfp;		/* shortform structure */
	xfs_dir2_dataptr_t	dot_offset;
	xfs_dir2_dataptr_t	dotdot_offset;
	xfs_ino_t		ino;
59
	struct xfs_da_geometry	*geo = args->geo;
60 61 62 63 64 65 66 67 68 69

	ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
	ASSERT(dp->i_df.if_bytes == dp->i_d.di_size);
	ASSERT(dp->i_df.if_u1.if_data != NULL);

	sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;

	/*
	 * If the block number in the offset is out of range, we're done.
	 */
70
	if (xfs_dir2_dataptr_to_db(geo, ctx->pos) > geo->datablk)
71 72 73 74 75 76
		return 0;

	/*
	 * Precalculate offsets for . and .. as we will always need them.
	 *
	 * XXX(hch): the second argument is sometimes 0 and sometimes
77
	 * geo->datablk
78
	 */
79
	dot_offset = xfs_dir2_db_off_to_dataptr(geo, geo->datablk,
80
						dp->d_ops->data_dot_offset);
81
	dotdot_offset = xfs_dir2_db_off_to_dataptr(geo, geo->datablk,
82
						dp->d_ops->data_dotdot_offset);
83 84 85 86 87 88 89 90 91 92 93 94 95 96

	/*
	 * Put . entry unless we're starting past it.
	 */
	if (ctx->pos <= dot_offset) {
		ctx->pos = dot_offset & 0x7fffffff;
		if (!dir_emit(ctx, ".", 1, dp->i_ino, DT_DIR))
			return 0;
	}

	/*
	 * Put .. entry unless we're starting past it.
	 */
	if (ctx->pos <= dotdot_offset) {
97
		ino = dp->d_ops->sf_get_parent_ino(sfp);
98 99 100 101 102 103 104 105 106 107
		ctx->pos = dotdot_offset & 0x7fffffff;
		if (!dir_emit(ctx, "..", 2, ino, DT_DIR))
			return 0;
	}

	/*
	 * Loop while there are more entries and put'ing works.
	 */
	sfep = xfs_dir2_sf_firstentry(sfp);
	for (i = 0; i < sfp->count; i++) {
108
		uint8_t filetype;
109

110
		off = xfs_dir2_db_off_to_dataptr(geo, geo->datablk,
111 112 113
				xfs_dir2_sf_get_offset(sfep));

		if (ctx->pos > off) {
114
			sfep = dp->d_ops->sf_nextentry(sfp, sfep);
115 116 117
			continue;
		}

118 119
		ino = dp->d_ops->sf_get_ino(sfp, sfep);
		filetype = dp->d_ops->sf_get_ftype(sfep);
120
		ctx->pos = off & 0x7fffffff;
121
		if (!dir_emit(ctx, (char *)sfep->name, sfep->namelen, ino,
122
			    xfs_dir3_get_dtype(dp->i_mount, filetype)))
123
			return 0;
124
		sfep = dp->d_ops->sf_nextentry(sfp, sfep);
125 126
	}

127
	ctx->pos = xfs_dir2_db_off_to_dataptr(geo, geo->datablk + 1, 0) &
128
								0x7fffffff;
129 130 131 132 133 134 135 136
	return 0;
}

/*
 * Readdir for block directories.
 */
STATIC int
xfs_dir2_block_getdents(
137
	struct xfs_da_args	*args,
138 139
	struct dir_context	*ctx)
{
140
	struct xfs_inode	*dp = args->dp;	/* incore directory inode */
141 142 143 144 145 146 147 148 149
	xfs_dir2_data_hdr_t	*hdr;		/* block header */
	struct xfs_buf		*bp;		/* buffer for block */
	xfs_dir2_data_entry_t	*dep;		/* block data entry */
	xfs_dir2_data_unused_t	*dup;		/* block unused entry */
	char			*endptr;	/* end of the data entries */
	int			error;		/* error return value */
	char			*ptr;		/* current data entry */
	int			wantoff;	/* starting block offset */
	xfs_off_t		cook;
150
	struct xfs_da_geometry	*geo = args->geo;
151
	int			lock_mode;
152 153 154 155

	/*
	 * If the block number in the offset is out of range, we're done.
	 */
156
	if (xfs_dir2_dataptr_to_db(geo, ctx->pos) > geo->datablk)
157 158
		return 0;

159
	lock_mode = xfs_ilock_data_map_shared(dp);
160
	error = xfs_dir3_block_read(args->trans, dp, &bp);
161
	xfs_iunlock(dp, lock_mode);
162 163 164 165 166 167 168
	if (error)
		return error;

	/*
	 * Extract the byte offset we start at from the seek pointer.
	 * We'll skip entries before this.
	 */
169
	wantoff = xfs_dir2_dataptr_to_off(geo, ctx->pos);
170 171 172 173 174
	hdr = bp->b_addr;
	xfs_dir3_data_check(dp, bp);
	/*
	 * Set up values for the loop.
	 */
175
	ptr = (char *)dp->d_ops->data_entry_p(hdr);
176
	endptr = xfs_dir3_data_endp(geo, hdr);
177 178 179 180 181 182

	/*
	 * Loop over the data portion of the block.
	 * Each object is a real entry (dep) or an unused one (dup).
	 */
	while (ptr < endptr) {
183
		uint8_t filetype;
184

185 186 187 188 189 190 191 192 193 194 195 196 197 198
		dup = (xfs_dir2_data_unused_t *)ptr;
		/*
		 * Unused, skip it.
		 */
		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
			ptr += be16_to_cpu(dup->length);
			continue;
		}

		dep = (xfs_dir2_data_entry_t *)ptr;

		/*
		 * Bump pointer for the next iteration.
		 */
199
		ptr += dp->d_ops->data_entsize(dep->namelen);
200 201 202 203 204 205
		/*
		 * The entry is before the desired starting point, skip it.
		 */
		if ((char *)dep - (char *)hdr < wantoff)
			continue;

206
		cook = xfs_dir2_db_off_to_dataptr(geo, geo->datablk,
207 208 209
					    (char *)dep - (char *)hdr);

		ctx->pos = cook & 0x7fffffff;
210
		filetype = dp->d_ops->data_get_ftype(dep);
211 212 213 214
		/*
		 * If it didn't fit, set the final offset to here & return.
		 */
		if (!dir_emit(ctx, (char *)dep->name, dep->namelen,
215
			    be64_to_cpu(dep->inumber),
216
			    xfs_dir3_get_dtype(dp->i_mount, filetype))) {
217
			xfs_trans_brelse(args->trans, bp);
218 219 220 221 222 223 224 225
			return 0;
		}
	}

	/*
	 * Reached the end of the block.
	 * Set the offset to a non-existent block 1 and return.
	 */
226
	ctx->pos = xfs_dir2_db_off_to_dataptr(geo, geo->datablk + 1, 0) &
227
								0x7fffffff;
228
	xfs_trans_brelse(args->trans, bp);
229 230 231
	return 0;
}

232 233 234 235 236
/*
 * Read a directory block and initiate readahead for blocks beyond that.
 * We maintain a sliding readahead window of the remaining space in the
 * buffer rounded up to the nearest block.
 */
237 238
STATIC int
xfs_dir2_leaf_readbuf(
239
	struct xfs_da_args	*args,
240
	size_t			bufsize,
241 242 243
	xfs_dir2_off_t		*cur_off,
	xfs_dablk_t		*ra_blk,
	struct xfs_buf		**bpp)
244
{
245
	struct xfs_inode	*dp = args->dp;
246
	struct xfs_buf		*bp = NULL;
247 248 249
	struct xfs_da_geometry	*geo = args->geo;
	struct xfs_ifork	*ifp = XFS_IFORK_PTR(dp, XFS_DATA_FORK);
	struct xfs_bmbt_irec	map;
250
	struct blk_plug		plug;
251 252 253 254
	xfs_dir2_off_t		new_off;
	xfs_dablk_t		next_ra;
	xfs_dablk_t		map_off;
	xfs_dablk_t		last_da;
255
	struct xfs_iext_cursor	icur;
256
	int			ra_want;
257 258
	int			error = 0;

259 260
	if (!(ifp->if_flags & XFS_IFEXTENTS)) {
		error = xfs_iread_extents(args->trans, dp, XFS_DATA_FORK);
261
		if (error)
262
			goto out;
263 264 265
	}

	/*
266 267 268
	 * Look for mapped directory blocks at or above the current offset.
	 * Truncate down to the nearest directory block to start the scanning
	 * operation.
269
	 */
270 271
	last_da = xfs_dir2_byte_to_da(geo, XFS_DIR2_LEAF_OFFSET);
	map_off = xfs_dir2_db_to_da(geo, xfs_dir2_byte_to_db(geo, *cur_off));
272
	if (!xfs_iext_lookup_extent(dp, ifp, map_off, &icur, &map))
273
		goto out;
274 275 276
	if (map.br_startoff >= last_da)
		goto out;
	xfs_trim_extent(&map, map_off, last_da - map_off);
277

278 279 280 281 282
	/* Read the directory block of that first mapping. */
	new_off = xfs_dir2_da_to_byte(geo, map.br_startoff);
	if (new_off > *cur_off)
		*cur_off = new_off;
	error = xfs_dir3_data_read(args->trans, dp, map.br_startoff, -1, &bp);
283
	if (error)
284
		goto out;
285 286

	/*
287 288 289
	 * Start readahead for the next bufsize's worth of dir data blocks.
	 * We may have already issued readahead for some of that range;
	 * ra_blk tracks the last block we tried to read(ahead).
290
	 */
291 292 293 294 295 296 297 298 299
	ra_want = howmany(bufsize + geo->blksize, (1 << geo->fsblog));
	if (*ra_blk >= last_da)
		goto out;
	else if (*ra_blk == 0)
		*ra_blk = map.br_startoff;
	next_ra = map.br_startoff + geo->fsbcount;
	if (next_ra >= last_da)
		goto out_no_ra;
	if (map.br_blockcount < geo->fsbcount &&
300
	    !xfs_iext_next_extent(ifp, &icur, &map))
301 302 303 304 305 306
		goto out_no_ra;
	if (map.br_startoff >= last_da)
		goto out_no_ra;
	xfs_trim_extent(&map, next_ra, last_da - next_ra);

	/* Start ra for each dir (not fs) block that has a mapping. */
307
	blk_start_plug(&plug);
308 309 310 311 312 313 314 315 316 317 318
	while (ra_want > 0) {
		next_ra = roundup((xfs_dablk_t)map.br_startoff, geo->fsbcount);
		while (ra_want > 0 &&
		       next_ra < map.br_startoff + map.br_blockcount) {
			if (next_ra >= last_da) {
				*ra_blk = last_da;
				break;
			}
			if (next_ra > *ra_blk) {
				xfs_dir3_data_readahead(dp, next_ra, -2);
				*ra_blk = next_ra;
319
			}
320 321 322
			ra_want -= geo->fsbcount;
			next_ra += geo->fsbcount;
		}
323
		if (!xfs_iext_next_extent(ifp, &icur, &map)) {
324 325
			*ra_blk = last_da;
			break;
326 327 328 329 330 331 332
		}
	}
	blk_finish_plug(&plug);

out:
	*bpp = bp;
	return error;
333 334 335
out_no_ra:
	*ra_blk = last_da;
	goto out;
336 337 338 339 340 341 342 343
}

/*
 * Getdents (readdir) for leaf and node directories.
 * This reads the data blocks only, so is the same for both forms.
 */
STATIC int
xfs_dir2_leaf_getdents(
344
	struct xfs_da_args	*args,
345 346 347
	struct dir_context	*ctx,
	size_t			bufsize)
{
348
	struct xfs_inode	*dp = args->dp;
349 350 351 352 353
	struct xfs_buf		*bp = NULL;	/* data block buffer */
	xfs_dir2_data_hdr_t	*hdr;		/* data block header */
	xfs_dir2_data_entry_t	*dep;		/* data entry */
	xfs_dir2_data_unused_t	*dup;		/* unused entry */
	char			*ptr = NULL;	/* pointer to current data */
354
	struct xfs_da_geometry	*geo = args->geo;
355 356 357 358 359 360
	xfs_dablk_t		rablk = 0;	/* current readahead block */
	xfs_dir2_off_t		curoff;		/* current overall offset */
	int			length;		/* temporary length value */
	int			byteoff;	/* offset in current block */
	int			lock_mode;
	int			error = 0;	/* error return value */
361 362 363 364 365 366 367 368 369 370 371 372

	/*
	 * If the offset is at or past the largest allowed value,
	 * give up right away.
	 */
	if (ctx->pos >= XFS_DIR2_MAX_DATAPTR)
		return 0;

	/*
	 * Inside the loop we keep the main offset value as a byte offset
	 * in the directory file.
	 */
373
	curoff = xfs_dir2_dataptr_to_byte(ctx->pos);
374 375 376 377 378 379

	/*
	 * Loop over directory entries until we reach the end offset.
	 * Get more blocks and readahead as necessary.
	 */
	while (curoff < XFS_DIR2_LEAF_OFFSET) {
380
		uint8_t filetype;
381

382 383 384 385
		/*
		 * If we have no buffer, or we're off the end of the
		 * current buffer, need to get another one.
		 */
386
		if (!bp || ptr >= (char *)bp->b_addr + geo->blksize) {
387
			if (bp) {
388
				xfs_trans_brelse(args->trans, bp);
389 390
				bp = NULL;
			}
391

392
			lock_mode = xfs_ilock_data_map_shared(dp);
393 394
			error = xfs_dir2_leaf_readbuf(args, bufsize, &curoff,
					&rablk, &bp);
395
			xfs_iunlock(dp, lock_mode);
396
			if (error || !bp)
397 398 399 400 401 402 403
				break;

			hdr = bp->b_addr;
			xfs_dir3_data_check(dp, bp);
			/*
			 * Find our position in the block.
			 */
404
			ptr = (char *)dp->d_ops->data_entry_p(hdr);
405
			byteoff = xfs_dir2_byte_to_off(geo, curoff);
406 407 408 409
			/*
			 * Skip past the header.
			 */
			if (byteoff == 0)
410
				curoff += dp->d_ops->data_entry_offset;
411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426
			/*
			 * Skip past entries until we reach our offset.
			 */
			else {
				while ((char *)ptr - (char *)hdr < byteoff) {
					dup = (xfs_dir2_data_unused_t *)ptr;

					if (be16_to_cpu(dup->freetag)
						  == XFS_DIR2_DATA_FREE_TAG) {

						length = be16_to_cpu(dup->length);
						ptr += length;
						continue;
					}
					dep = (xfs_dir2_data_entry_t *)ptr;
					length =
427
					   dp->d_ops->data_entsize(dep->namelen);
428 429 430 431 432 433
					ptr += length;
				}
				/*
				 * Now set our real offset.
				 */
				curoff =
434 435
					xfs_dir2_db_off_to_byte(geo,
					    xfs_dir2_byte_to_db(geo, curoff),
436
					    (char *)ptr - (char *)hdr);
437
				if (ptr >= (char *)hdr + geo->blksize) {
438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457
					continue;
				}
			}
		}
		/*
		 * We have a pointer to an entry.
		 * Is it a live one?
		 */
		dup = (xfs_dir2_data_unused_t *)ptr;
		/*
		 * No, it's unused, skip over it.
		 */
		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
			length = be16_to_cpu(dup->length);
			ptr += length;
			curoff += length;
			continue;
		}

		dep = (xfs_dir2_data_entry_t *)ptr;
458 459
		length = dp->d_ops->data_entsize(dep->namelen);
		filetype = dp->d_ops->data_get_ftype(dep);
460

461
		ctx->pos = xfs_dir2_byte_to_dataptr(curoff) & 0x7fffffff;
462
		if (!dir_emit(ctx, (char *)dep->name, dep->namelen,
463
			    be64_to_cpu(dep->inumber),
464
			    xfs_dir3_get_dtype(dp->i_mount, filetype)))
465 466 467 468 469 470 471 472 473 474 475 476 477 478
			break;

		/*
		 * Advance to next entry in the block.
		 */
		ptr += length;
		curoff += length;
		/* bufsize may have just been a guess; don't go negative */
		bufsize = bufsize > length ? bufsize - length : 0;
	}

	/*
	 * All done.  Set output offset value to current offset.
	 */
479
	if (curoff > xfs_dir2_dataptr_to_byte(XFS_DIR2_MAX_DATAPTR))
480 481
		ctx->pos = XFS_DIR2_MAX_DATAPTR & 0x7fffffff;
	else
482
		ctx->pos = xfs_dir2_byte_to_dataptr(curoff) & 0x7fffffff;
483
	if (bp)
484
		xfs_trans_brelse(args->trans, bp);
485 486 487 488 489
	return error;
}

/*
 * Read a directory.
490 491 492 493 494
 *
 * If supplied, the transaction collects locked dir buffers to avoid
 * nested buffer deadlocks.  This function does not dirty the
 * transaction.  The caller should ensure that the inode is locked
 * before calling this function.
495 496 497
 */
int
xfs_readdir(
498
	struct xfs_trans	*tp,
499 500 501
	struct xfs_inode	*dp,
	struct dir_context	*ctx,
	size_t			bufsize)
502
{
503
	struct xfs_da_args	args = { NULL };
504 505
	int			rval;
	int			v;
506 507 508 509

	trace_xfs_readdir(dp);

	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
D
Dave Chinner 已提交
510
		return -EIO;
511

D
Dave Chinner 已提交
512
	ASSERT(S_ISDIR(VFS_I(dp)->i_mode));
513
	XFS_STATS_INC(dp->i_mount, xs_dir_getdents);
514

515 516
	args.dp = dp;
	args.geo = dp->i_mount->m_dir_geo;
517
	args.trans = tp;
518

519
	if (dp->i_d.di_format == XFS_DINODE_FMT_LOCAL)
520 521
		rval = xfs_dir2_sf_getdents(&args, ctx);
	else if ((rval = xfs_dir2_isblock(&args, &v)))
522 523
		;
	else if (v)
524
		rval = xfs_dir2_block_getdents(&args, ctx);
525
	else
526
		rval = xfs_dir2_leaf_getdents(&args, ctx, bufsize);
527

528 529
	return rval;
}