xfs_inode_item.c 23.4 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2 3
 * Copyright (c) 2000-2002,2005 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 22 23
#include "xfs_types.h"
#include "xfs_log.h"
#include "xfs_trans.h"
#include "xfs_sb.h"
24
#include "xfs_ag.h"
L
Linus Torvalds 已提交
25 26 27 28 29
#include "xfs_mount.h"
#include "xfs_trans_priv.h"
#include "xfs_bmap_btree.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
30
#include "xfs_inode_item.h"
31
#include "xfs_error.h"
C
Christoph Hellwig 已提交
32
#include "xfs_trace.h"
L
Linus Torvalds 已提交
33 34 35 36


kmem_zone_t	*xfs_ili_zone;		/* inode log item zone */

37 38 39 40 41 42
static inline struct xfs_inode_log_item *INODE_ITEM(struct xfs_log_item *lip)
{
	return container_of(lip, struct xfs_inode_log_item, ili_item);
}


L
Linus Torvalds 已提交
43 44 45 46 47 48 49 50 51
/*
 * This returns the number of iovecs needed to log the given inode item.
 *
 * We need one iovec for the inode log format structure, one for the
 * inode core, and possibly one for the inode data/extents/b-tree root
 * and one for the inode attribute data/extents/b-tree root.
 */
STATIC uint
xfs_inode_item_size(
52
	struct xfs_log_item	*lip)
L
Linus Torvalds 已提交
53
{
54 55 56
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;
	uint			nvecs = 2;
L
Linus Torvalds 已提交
57 58 59

	switch (ip->i_d.di_format) {
	case XFS_DINODE_FMT_EXTENTS:
60
		if ((iip->ili_fields & XFS_ILOG_DEXT) &&
61 62
		    ip->i_d.di_nextents > 0 &&
		    ip->i_df.if_bytes > 0)
L
Linus Torvalds 已提交
63 64 65 66
			nvecs++;
		break;

	case XFS_DINODE_FMT_BTREE:
67
		if ((iip->ili_fields & XFS_ILOG_DBROOT) &&
68
		    ip->i_df.if_broot_bytes > 0)
L
Linus Torvalds 已提交
69 70 71 72
			nvecs++;
		break;

	case XFS_DINODE_FMT_LOCAL:
73
		if ((iip->ili_fields & XFS_ILOG_DDATA) &&
74
		    ip->i_df.if_bytes > 0)
L
Linus Torvalds 已提交
75 76 77 78 79 80 81 82 83 84 85 86
			nvecs++;
		break;

	case XFS_DINODE_FMT_DEV:
	case XFS_DINODE_FMT_UUID:
		break;

	default:
		ASSERT(0);
		break;
	}

87
	if (!XFS_IFORK_Q(ip))
L
Linus Torvalds 已提交
88
		return nvecs;
89

L
Linus Torvalds 已提交
90 91 92 93 94 95

	/*
	 * Log any necessary attribute data.
	 */
	switch (ip->i_d.di_aformat) {
	case XFS_DINODE_FMT_EXTENTS:
96
		if ((iip->ili_fields & XFS_ILOG_AEXT) &&
97 98
		    ip->i_d.di_anextents > 0 &&
		    ip->i_afp->if_bytes > 0)
L
Linus Torvalds 已提交
99 100 101 102
			nvecs++;
		break;

	case XFS_DINODE_FMT_BTREE:
103
		if ((iip->ili_fields & XFS_ILOG_ABROOT) &&
104
		    ip->i_afp->if_broot_bytes > 0)
L
Linus Torvalds 已提交
105 106 107 108
			nvecs++;
		break;

	case XFS_DINODE_FMT_LOCAL:
109
		if ((iip->ili_fields & XFS_ILOG_ADATA) &&
110
		    ip->i_afp->if_bytes > 0)
L
Linus Torvalds 已提交
111 112 113 114 115 116 117 118 119 120 121
			nvecs++;
		break;

	default:
		ASSERT(0);
		break;
	}

	return nvecs;
}

122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156
/*
 * xfs_inode_item_format_extents - convert in-core extents to on-disk form
 *
 * For either the data or attr fork in extent format, we need to endian convert
 * the in-core extent as we place them into the on-disk inode. In this case, we
 * need to do this conversion before we write the extents into the log. Because
 * we don't have the disk inode to write into here, we allocate a buffer and
 * format the extents into it via xfs_iextents_copy(). We free the buffer in
 * the unlock routine after the copy for the log has been made.
 *
 * In the case of the data fork, the in-core and on-disk fork sizes can be
 * different due to delayed allocation extents. We only log on-disk extents
 * here, so always use the physical fork size to determine the size of the
 * buffer we need to allocate.
 */
STATIC void
xfs_inode_item_format_extents(
	struct xfs_inode	*ip,
	struct xfs_log_iovec	*vecp,
	int			whichfork,
	int			type)
{
	xfs_bmbt_rec_t		*ext_buffer;

	ext_buffer = kmem_alloc(XFS_IFORK_SIZE(ip, whichfork), KM_SLEEP);
	if (whichfork == XFS_DATA_FORK)
		ip->i_itemp->ili_extents_buf = ext_buffer;
	else
		ip->i_itemp->ili_aextents_buf = ext_buffer;

	vecp->i_addr = ext_buffer;
	vecp->i_len = xfs_iextents_copy(ip, ext_buffer, whichfork);
	vecp->i_type = type;
}

L
Linus Torvalds 已提交
157 158 159 160 161 162 163 164 165
/*
 * This is called to fill in the vector of log iovecs for the
 * given inode log item.  It fills the first item with an inode
 * log format structure, the second with the on-disk inode structure,
 * and a possible third and/or fourth with the inode data/extents/b-tree
 * root and inode attributes data/extents/b-tree root.
 */
STATIC void
xfs_inode_item_format(
166 167
	struct xfs_log_item	*lip,
	struct xfs_log_iovec	*vecp)
L
Linus Torvalds 已提交
168
{
169 170
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;
L
Linus Torvalds 已提交
171 172 173 174
	uint			nvecs;
	size_t			data_bytes;
	xfs_mount_t		*mp;

175
	vecp->i_addr = &iip->ili_format;
L
Linus Torvalds 已提交
176
	vecp->i_len  = sizeof(xfs_inode_log_format_t);
C
Christoph Hellwig 已提交
177
	vecp->i_type = XLOG_REG_TYPE_IFORMAT;
L
Linus Torvalds 已提交
178 179 180
	vecp++;
	nvecs	     = 1;

181
	vecp->i_addr = &ip->i_d;
C
Christoph Hellwig 已提交
182
	vecp->i_len  = sizeof(struct xfs_icdinode);
C
Christoph Hellwig 已提交
183
	vecp->i_type = XLOG_REG_TYPE_ICORE;
L
Linus Torvalds 已提交
184 185 186 187 188 189 190 191 192 193 194 195
	vecp++;
	nvecs++;

	/*
	 * If this is really an old format inode, then we need to
	 * log it as such.  This means that we have to copy the link
	 * count from the new field to the old.  We don't have to worry
	 * about the new fields, because nothing trusts them as long as
	 * the old inode version number is there.  If the superblock already
	 * has a new version number, then we don't bother converting back.
	 */
	mp = ip->i_mount;
196 197
	ASSERT(ip->i_d.di_version == 1 || xfs_sb_version_hasnlink(&mp->m_sb));
	if (ip->i_d.di_version == 1) {
198
		if (!xfs_sb_version_hasnlink(&mp->m_sb)) {
L
Linus Torvalds 已提交
199 200 201 202 203 204 205 206 207 208 209
			/*
			 * Convert it back.
			 */
			ASSERT(ip->i_d.di_nlink <= XFS_MAXLINK_1);
			ip->i_d.di_onlink = ip->i_d.di_nlink;
		} else {
			/*
			 * The superblock version has already been bumped,
			 * so just make the conversion to the new inode
			 * format permanent.
			 */
210
			ip->i_d.di_version = 2;
L
Linus Torvalds 已提交
211 212 213 214 215 216 217
			ip->i_d.di_onlink = 0;
			memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
		}
	}

	switch (ip->i_d.di_format) {
	case XFS_DINODE_FMT_EXTENTS:
218
		iip->ili_fields &=
219 220 221
			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
			  XFS_ILOG_DEV | XFS_ILOG_UUID);

222
		if ((iip->ili_fields & XFS_ILOG_DEXT) &&
223 224
		    ip->i_d.di_nextents > 0 &&
		    ip->i_df.if_bytes > 0) {
L
Linus Torvalds 已提交
225
			ASSERT(ip->i_df.if_u1.if_extents != NULL);
226
			ASSERT(ip->i_df.if_bytes / sizeof(xfs_bmbt_rec_t) > 0);
L
Linus Torvalds 已提交
227
			ASSERT(iip->ili_extents_buf == NULL);
228

229
#ifdef XFS_NATIVE_HOST
D
Dave Chinner 已提交
230 231
                       if (ip->i_d.di_nextents == ip->i_df.if_bytes /
                                               (uint)sizeof(xfs_bmbt_rec_t)) {
L
Linus Torvalds 已提交
232 233 234 235 236
				/*
				 * There are no delayed allocation
				 * extents, so just point to the
				 * real extents array.
				 */
237
				vecp->i_addr = ip->i_df.if_u1.if_extents;
L
Linus Torvalds 已提交
238
				vecp->i_len = ip->i_df.if_bytes;
C
Christoph Hellwig 已提交
239
				vecp->i_type = XLOG_REG_TYPE_IEXT;
L
Linus Torvalds 已提交
240 241 242
			} else
#endif
			{
243 244
				xfs_inode_item_format_extents(ip, vecp,
					XFS_DATA_FORK, XLOG_REG_TYPE_IEXT);
L
Linus Torvalds 已提交
245 246 247 248 249
			}
			ASSERT(vecp->i_len <= ip->i_df.if_bytes);
			iip->ili_format.ilf_dsize = vecp->i_len;
			vecp++;
			nvecs++;
250
		} else {
251
			iip->ili_fields &= ~XFS_ILOG_DEXT;
L
Linus Torvalds 已提交
252 253 254 255
		}
		break;

	case XFS_DINODE_FMT_BTREE:
256
		iip->ili_fields &=
257 258 259
			~(XFS_ILOG_DDATA | XFS_ILOG_DEXT |
			  XFS_ILOG_DEV | XFS_ILOG_UUID);

260
		if ((iip->ili_fields & XFS_ILOG_DBROOT) &&
261
		    ip->i_df.if_broot_bytes > 0) {
L
Linus Torvalds 已提交
262
			ASSERT(ip->i_df.if_broot != NULL);
263
			vecp->i_addr = ip->i_df.if_broot;
L
Linus Torvalds 已提交
264
			vecp->i_len = ip->i_df.if_broot_bytes;
C
Christoph Hellwig 已提交
265
			vecp->i_type = XLOG_REG_TYPE_IBROOT;
L
Linus Torvalds 已提交
266 267 268
			vecp++;
			nvecs++;
			iip->ili_format.ilf_dsize = ip->i_df.if_broot_bytes;
269
		} else {
270
			ASSERT(!(iip->ili_fields &
271
				 XFS_ILOG_DBROOT));
272
			iip->ili_fields &= ~XFS_ILOG_DBROOT;
L
Linus Torvalds 已提交
273 274 275 276
		}
		break;

	case XFS_DINODE_FMT_LOCAL:
277
		iip->ili_fields &=
278 279
			~(XFS_ILOG_DEXT | XFS_ILOG_DBROOT |
			  XFS_ILOG_DEV | XFS_ILOG_UUID);
280
		if ((iip->ili_fields & XFS_ILOG_DDATA) &&
281
		    ip->i_df.if_bytes > 0) {
L
Linus Torvalds 已提交
282 283 284
			ASSERT(ip->i_df.if_u1.if_data != NULL);
			ASSERT(ip->i_d.di_size > 0);

285
			vecp->i_addr = ip->i_df.if_u1.if_data;
L
Linus Torvalds 已提交
286 287 288 289 290 291 292 293 294
			/*
			 * Round i_bytes up to a word boundary.
			 * The underlying memory is guaranteed to
			 * to be there by xfs_idata_realloc().
			 */
			data_bytes = roundup(ip->i_df.if_bytes, 4);
			ASSERT((ip->i_df.if_real_bytes == 0) ||
			       (ip->i_df.if_real_bytes == data_bytes));
			vecp->i_len = (int)data_bytes;
C
Christoph Hellwig 已提交
295
			vecp->i_type = XLOG_REG_TYPE_ILOCAL;
L
Linus Torvalds 已提交
296 297 298
			vecp++;
			nvecs++;
			iip->ili_format.ilf_dsize = (unsigned)data_bytes;
299
		} else {
300
			iip->ili_fields &= ~XFS_ILOG_DDATA;
L
Linus Torvalds 已提交
301 302 303 304
		}
		break;

	case XFS_DINODE_FMT_DEV:
305
		iip->ili_fields &=
306 307
			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
			  XFS_ILOG_DEXT | XFS_ILOG_UUID);
308
		if (iip->ili_fields & XFS_ILOG_DEV) {
L
Linus Torvalds 已提交
309 310 311 312 313 314
			iip->ili_format.ilf_u.ilfu_rdev =
				ip->i_df.if_u2.if_rdev;
		}
		break;

	case XFS_DINODE_FMT_UUID:
315
		iip->ili_fields &=
316 317
			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
			  XFS_ILOG_DEXT | XFS_ILOG_DEV);
318
		if (iip->ili_fields & XFS_ILOG_UUID) {
L
Linus Torvalds 已提交
319 320 321 322 323 324 325 326 327 328 329
			iip->ili_format.ilf_u.ilfu_uuid =
				ip->i_df.if_u2.if_uuid;
		}
		break;

	default:
		ASSERT(0);
		break;
	}

	/*
330
	 * If there are no attributes associated with the file, then we're done.
L
Linus Torvalds 已提交
331 332
	 */
	if (!XFS_IFORK_Q(ip)) {
333
		iip->ili_fields &=
334
			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT);
335
		goto out;
L
Linus Torvalds 已提交
336 337 338 339
	}

	switch (ip->i_d.di_aformat) {
	case XFS_DINODE_FMT_EXTENTS:
340
		iip->ili_fields &=
341 342
			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT);

343
		if ((iip->ili_fields & XFS_ILOG_AEXT) &&
344 345 346 347
		    ip->i_d.di_anextents > 0 &&
		    ip->i_afp->if_bytes > 0) {
			ASSERT(ip->i_afp->if_bytes / sizeof(xfs_bmbt_rec_t) ==
				ip->i_d.di_anextents);
348
			ASSERT(ip->i_afp->if_u1.if_extents != NULL);
349
#ifdef XFS_NATIVE_HOST
L
Linus Torvalds 已提交
350 351 352 353
			/*
			 * There are not delayed allocation extents
			 * for attributes, so just point at the array.
			 */
354
			vecp->i_addr = ip->i_afp->if_u1.if_extents;
L
Linus Torvalds 已提交
355
			vecp->i_len = ip->i_afp->if_bytes;
356
			vecp->i_type = XLOG_REG_TYPE_IATTR_EXT;
L
Linus Torvalds 已提交
357 358
#else
			ASSERT(iip->ili_aextents_buf == NULL);
359 360
			xfs_inode_item_format_extents(ip, vecp,
					XFS_ATTR_FORK, XLOG_REG_TYPE_IATTR_EXT);
L
Linus Torvalds 已提交
361 362 363 364
#endif
			iip->ili_format.ilf_asize = vecp->i_len;
			vecp++;
			nvecs++;
365
		} else {
366
			iip->ili_fields &= ~XFS_ILOG_AEXT;
L
Linus Torvalds 已提交
367 368 369 370
		}
		break;

	case XFS_DINODE_FMT_BTREE:
371
		iip->ili_fields &=
372 373
			~(XFS_ILOG_ADATA | XFS_ILOG_AEXT);

374
		if ((iip->ili_fields & XFS_ILOG_ABROOT) &&
375
		    ip->i_afp->if_broot_bytes > 0) {
L
Linus Torvalds 已提交
376
			ASSERT(ip->i_afp->if_broot != NULL);
377

378
			vecp->i_addr = ip->i_afp->if_broot;
L
Linus Torvalds 已提交
379
			vecp->i_len = ip->i_afp->if_broot_bytes;
C
Christoph Hellwig 已提交
380
			vecp->i_type = XLOG_REG_TYPE_IATTR_BROOT;
L
Linus Torvalds 已提交
381 382 383
			vecp++;
			nvecs++;
			iip->ili_format.ilf_asize = ip->i_afp->if_broot_bytes;
384
		} else {
385
			iip->ili_fields &= ~XFS_ILOG_ABROOT;
L
Linus Torvalds 已提交
386 387 388 389
		}
		break;

	case XFS_DINODE_FMT_LOCAL:
390
		iip->ili_fields &=
391 392
			~(XFS_ILOG_AEXT | XFS_ILOG_ABROOT);

393
		if ((iip->ili_fields & XFS_ILOG_ADATA) &&
394
		    ip->i_afp->if_bytes > 0) {
L
Linus Torvalds 已提交
395 396
			ASSERT(ip->i_afp->if_u1.if_data != NULL);

397
			vecp->i_addr = ip->i_afp->if_u1.if_data;
L
Linus Torvalds 已提交
398 399 400 401 402 403 404 405 406
			/*
			 * Round i_bytes up to a word boundary.
			 * The underlying memory is guaranteed to
			 * to be there by xfs_idata_realloc().
			 */
			data_bytes = roundup(ip->i_afp->if_bytes, 4);
			ASSERT((ip->i_afp->if_real_bytes == 0) ||
			       (ip->i_afp->if_real_bytes == data_bytes));
			vecp->i_len = (int)data_bytes;
C
Christoph Hellwig 已提交
407
			vecp->i_type = XLOG_REG_TYPE_IATTR_LOCAL;
L
Linus Torvalds 已提交
408 409 410
			vecp++;
			nvecs++;
			iip->ili_format.ilf_asize = (unsigned)data_bytes;
411
		} else {
412
			iip->ili_fields &= ~XFS_ILOG_ADATA;
L
Linus Torvalds 已提交
413 414 415 416 417 418 419 420
		}
		break;

	default:
		ASSERT(0);
		break;
	}

421 422 423 424 425 426 427
out:
	/*
	 * Now update the log format that goes out to disk from the in-core
	 * values.  We always write the inode core to make the arithmetic
	 * games in recovery easier, which isn't a big deal as just about any
	 * transaction would dirty it anyway.
	 */
428 429
	iip->ili_format.ilf_fields = XFS_ILOG_CORE |
		(iip->ili_fields & ~XFS_ILOG_TIMESTAMP);
L
Linus Torvalds 已提交
430 431 432 433 434 435
	iip->ili_format.ilf_size = nvecs;
}


/*
 * This is called to pin the inode associated with the inode log
436
 * item in memory so it cannot be written out.
L
Linus Torvalds 已提交
437 438 439
 */
STATIC void
xfs_inode_item_pin(
440
	struct xfs_log_item	*lip)
L
Linus Torvalds 已提交
441
{
442
	struct xfs_inode	*ip = INODE_ITEM(lip)->ili_inode;
443

444 445 446 447
	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));

	trace_xfs_inode_pin(ip, _RET_IP_);
	atomic_inc(&ip->i_pincount);
L
Linus Torvalds 已提交
448 449 450 451 452 453
}


/*
 * This is called to unpin the inode associated with the inode log
 * item which was previously pinned with a call to xfs_inode_item_pin().
454 455
 *
 * Also wake up anyone in xfs_iunpin_wait() if the count goes to 0.
L
Linus Torvalds 已提交
456 457 458
 */
STATIC void
xfs_inode_item_unpin(
459
	struct xfs_log_item	*lip,
460
	int			remove)
L
Linus Torvalds 已提交
461
{
462
	struct xfs_inode	*ip = INODE_ITEM(lip)->ili_inode;
463

464
	trace_xfs_inode_unpin(ip, _RET_IP_);
465 466
	ASSERT(atomic_read(&ip->i_pincount) > 0);
	if (atomic_dec_and_test(&ip->i_pincount))
467
		wake_up_bit(&ip->i_flags, __XFS_IPINNED_BIT);
L
Linus Torvalds 已提交
468 469 470
}

STATIC uint
471 472 473
xfs_inode_item_push(
	struct xfs_log_item	*lip,
	struct list_head	*buffer_list)
L
Linus Torvalds 已提交
474
{
475 476
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;
477 478 479
	struct xfs_buf		*bp = NULL;
	uint			rval = XFS_ITEM_SUCCESS;
	int			error;
L
Linus Torvalds 已提交
480

481
	if (xfs_ipincount(ip) > 0)
L
Linus Torvalds 已提交
482 483
		return XFS_ITEM_PINNED;

484
	if (!xfs_ilock_nowait(ip, XFS_ILOCK_SHARED))
L
Linus Torvalds 已提交
485 486
		return XFS_ITEM_LOCKED;

487 488 489 490 491
	/*
	 * Re-check the pincount now that we stabilized the value by
	 * taking the ilock.
	 */
	if (xfs_ipincount(ip) > 0) {
492 493
		rval = XFS_ITEM_PINNED;
		goto out_unlock;
494 495
	}

496 497 498 499 500 501 502 503
	/*
	 * Stale inode items should force out the iclog.
	 */
	if (ip->i_flags & XFS_ISTALE) {
		rval = XFS_ITEM_PINNED;
		goto out_unlock;
	}

504 505 506 507 508
	/*
	 * Someone else is already flushing the inode.  Nothing we can do
	 * here but wait for the flush to finish and remove the item from
	 * the AIL.
	 */
L
Linus Torvalds 已提交
509
	if (!xfs_iflock_nowait(ip)) {
510 511
		rval = XFS_ITEM_FLUSHING;
		goto out_unlock;
L
Linus Torvalds 已提交
512 513
	}

514 515 516 517 518 519 520 521 522 523
	ASSERT(iip->ili_fields != 0 || XFS_FORCED_SHUTDOWN(ip->i_mount));
	ASSERT(iip->ili_logged == 0 || XFS_FORCED_SHUTDOWN(ip->i_mount));

	spin_unlock(&lip->li_ailp->xa_lock);

	error = xfs_iflush(ip, &bp);
	if (!error) {
		if (!xfs_buf_delwri_queue(bp, buffer_list))
			rval = XFS_ITEM_FLUSHING;
		xfs_buf_relse(bp);
L
Linus Torvalds 已提交
524
	}
525 526 527 528 529

	spin_lock(&lip->li_ailp->xa_lock);
out_unlock:
	xfs_iunlock(ip, XFS_ILOCK_SHARED);
	return rval;
L
Linus Torvalds 已提交
530 531 532 533 534 535 536 537 538 539
}

/*
 * Unlock the inode associated with the inode log item.
 * Clear the fields of the inode and inode log item that
 * are specific to the current transaction.  If the
 * hold flags is set, do not unlock the inode.
 */
STATIC void
xfs_inode_item_unlock(
540
	struct xfs_log_item	*lip)
L
Linus Torvalds 已提交
541
{
542 543
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;
544
	unsigned short		lock_flags;
L
Linus Torvalds 已提交
545

C
Christoph Hellwig 已提交
546 547
	ASSERT(ip->i_itemp != NULL);
	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
L
Linus Torvalds 已提交
548 549 550 551 552 553 554 555

	/*
	 * If the inode needed a separate buffer with which to log
	 * its extents, then free it now.
	 */
	if (iip->ili_extents_buf != NULL) {
		ASSERT(ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS);
		ASSERT(ip->i_d.di_nextents > 0);
556
		ASSERT(iip->ili_fields & XFS_ILOG_DEXT);
L
Linus Torvalds 已提交
557
		ASSERT(ip->i_df.if_bytes > 0);
558
		kmem_free(iip->ili_extents_buf);
L
Linus Torvalds 已提交
559 560 561 562 563
		iip->ili_extents_buf = NULL;
	}
	if (iip->ili_aextents_buf != NULL) {
		ASSERT(ip->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS);
		ASSERT(ip->i_d.di_anextents > 0);
564
		ASSERT(iip->ili_fields & XFS_ILOG_AEXT);
L
Linus Torvalds 已提交
565
		ASSERT(ip->i_afp->if_bytes > 0);
566
		kmem_free(iip->ili_aextents_buf);
L
Linus Torvalds 已提交
567 568 569
		iip->ili_aextents_buf = NULL;
	}

570 571
	lock_flags = iip->ili_lock_flags;
	iip->ili_lock_flags = 0;
572
	if (lock_flags)
C
Christoph Hellwig 已提交
573
		xfs_iunlock(ip, lock_flags);
L
Linus Torvalds 已提交
574 575 576
}

/*
577 578 579 580 581 582 583 584 585 586 587
 * This is called to find out where the oldest active copy of the inode log
 * item in the on disk log resides now that the last log write of it completed
 * at the given lsn.  Since we always re-log all dirty data in an inode, the
 * latest copy in the on disk log is the only one that matters.  Therefore,
 * simply return the given lsn.
 *
 * If the inode has been marked stale because the cluster is being freed, we
 * don't want to (re-)insert this inode into the AIL. There is a race condition
 * where the cluster buffer may be unpinned before the inode is inserted into
 * the AIL during transaction committed processing. If the buffer is unpinned
 * before the inode item has been committed and inserted, then it is possible
588
 * for the buffer to be written and IO completes before the inode is inserted
589 590 591 592 593
 * into the AIL. In that case, we'd be inserting a clean, stale inode into the
 * AIL which will never get removed. It will, however, get reclaimed which
 * triggers an assert in xfs_inode_free() complaining about freein an inode
 * still in the AIL.
 *
594 595 596
 * To avoid this, just unpin the inode directly and return a LSN of -1 so the
 * transaction committed code knows that it does not need to do any further
 * processing on the item.
L
Linus Torvalds 已提交
597 598 599
 */
STATIC xfs_lsn_t
xfs_inode_item_committed(
600
	struct xfs_log_item	*lip,
L
Linus Torvalds 已提交
601 602
	xfs_lsn_t		lsn)
{
603 604 605
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;

606 607 608 609
	if (xfs_iflags_test(ip, XFS_ISTALE)) {
		xfs_inode_item_unpin(lip, 0);
		return -1;
	}
610
	return lsn;
L
Linus Torvalds 已提交
611 612 613 614 615 616 617 618
}

/*
 * XXX rcc - this one really has to do something.  Probably needs
 * to stamp in a new field in the incore inode.
 */
STATIC void
xfs_inode_item_committing(
619
	struct xfs_log_item	*lip,
L
Linus Torvalds 已提交
620 621
	xfs_lsn_t		lsn)
{
622
	INODE_ITEM(lip)->ili_last_lsn = lsn;
L
Linus Torvalds 已提交
623 624 625 626 627
}

/*
 * This is the ops vector shared by all buf log items.
 */
C
Christoph Hellwig 已提交
628
static const struct xfs_item_ops xfs_inode_item_ops = {
629 630 631 632 633 634 635 636
	.iop_size	= xfs_inode_item_size,
	.iop_format	= xfs_inode_item_format,
	.iop_pin	= xfs_inode_item_pin,
	.iop_unpin	= xfs_inode_item_unpin,
	.iop_unlock	= xfs_inode_item_unlock,
	.iop_committed	= xfs_inode_item_committed,
	.iop_push	= xfs_inode_item_push,
	.iop_committing = xfs_inode_item_committing
L
Linus Torvalds 已提交
637 638 639 640 641 642 643 644
};


/*
 * Initialize the inode log item for a newly allocated (in-core) inode.
 */
void
xfs_inode_item_init(
645 646
	struct xfs_inode	*ip,
	struct xfs_mount	*mp)
L
Linus Torvalds 已提交
647
{
648
	struct xfs_inode_log_item *iip;
L
Linus Torvalds 已提交
649 650 651 652 653

	ASSERT(ip->i_itemp == NULL);
	iip = ip->i_itemp = kmem_zone_zalloc(xfs_ili_zone, KM_SLEEP);

	iip->ili_inode = ip;
654 655
	xfs_log_item_init(mp, &iip->ili_item, XFS_LI_INODE,
						&xfs_inode_item_ops);
L
Linus Torvalds 已提交
656 657
	iip->ili_format.ilf_type = XFS_LI_INODE;
	iip->ili_format.ilf_ino = ip->i_ino;
658 659 660
	iip->ili_format.ilf_blkno = ip->i_imap.im_blkno;
	iip->ili_format.ilf_len = ip->i_imap.im_len;
	iip->ili_format.ilf_boffset = ip->i_imap.im_boffset;
L
Linus Torvalds 已提交
661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679
}

/*
 * Free the inode log item and any memory hanging off of it.
 */
void
xfs_inode_item_destroy(
	xfs_inode_t	*ip)
{
	kmem_zone_free(xfs_ili_zone, ip->i_itemp);
}


/*
 * This is the inode flushing I/O completion routine.  It is called
 * from interrupt level when the buffer containing the inode is
 * flushed to disk.  It is responsible for removing the inode item
 * from the AIL if it has not been re-logged, and unlocking the inode's
 * flush lock.
680 681 682 683 684
 *
 * To reduce AIL lock traffic as much as possible, we scan the buffer log item
 * list for other inodes that will run this function. We remove them from the
 * buffer list so we can process all the inode IO completions in one AIL lock
 * traversal.
L
Linus Torvalds 已提交
685 686 687
 */
void
xfs_iflush_done(
688 689
	struct xfs_buf		*bp,
	struct xfs_log_item	*lip)
L
Linus Torvalds 已提交
690
{
691 692 693 694
	struct xfs_inode_log_item *iip;
	struct xfs_log_item	*blip;
	struct xfs_log_item	*next;
	struct xfs_log_item	*prev;
695
	struct xfs_ail		*ailp = lip->li_ailp;
696 697 698 699 700 701
	int			need_ail = 0;

	/*
	 * Scan the buffer IO completions for other inodes being completed and
	 * attach them to the current inode log item.
	 */
702
	blip = bp->b_fspriv;
703 704 705 706 707 708 709 710 711 712 713
	prev = NULL;
	while (blip != NULL) {
		if (lip->li_cb != xfs_iflush_done) {
			prev = blip;
			blip = blip->li_bio_list;
			continue;
		}

		/* remove from list */
		next = blip->li_bio_list;
		if (!prev) {
714
			bp->b_fspriv = next;
715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737
		} else {
			prev->li_bio_list = next;
		}

		/* add to current list */
		blip->li_bio_list = lip->li_bio_list;
		lip->li_bio_list = blip;

		/*
		 * while we have the item, do the unlocked check for needing
		 * the AIL lock.
		 */
		iip = INODE_ITEM(blip);
		if (iip->ili_logged && blip->li_lsn == iip->ili_flush_lsn)
			need_ail++;

		blip = next;
	}

	/* make sure we capture the state of the initial inode. */
	iip = INODE_ITEM(lip);
	if (iip->ili_logged && lip->li_lsn == iip->ili_flush_lsn)
		need_ail++;
L
Linus Torvalds 已提交
738 739 740 741 742 743 744 745 746 747

	/*
	 * We only want to pull the item from the AIL if it is
	 * actually there and its location in the log has not
	 * changed since we started the flush.  Thus, we only bother
	 * if the ili_logged flag is set and the inode's lsn has not
	 * changed.  First we check the lsn outside
	 * the lock since it's cheaper, and then we recheck while
	 * holding the lock before removing the inode from the AIL.
	 */
748 749 750
	if (need_ail) {
		struct xfs_log_item *log_items[need_ail];
		int i = 0;
751
		spin_lock(&ailp->xa_lock);
752 753 754 755 756 757 758
		for (blip = lip; blip; blip = blip->li_bio_list) {
			iip = INODE_ITEM(blip);
			if (iip->ili_logged &&
			    blip->li_lsn == iip->ili_flush_lsn) {
				log_items[i++] = blip;
			}
			ASSERT(i <= need_ail);
L
Linus Torvalds 已提交
759
		}
760
		/* xfs_trans_ail_delete_bulk() drops the AIL lock. */
761 762
		xfs_trans_ail_delete_bulk(ailp, log_items, i,
					  SHUTDOWN_CORRUPT_INCORE);
L
Linus Torvalds 已提交
763 764 765 766
	}


	/*
767 768 769
	 * clean up and unlock the flush lock now we are done. We can clear the
	 * ili_last_fields bits now that we know that the data corresponding to
	 * them is safely on disk.
L
Linus Torvalds 已提交
770
	 */
771 772 773 774 775 776 777 778 779
	for (blip = lip; blip; blip = next) {
		next = blip->li_bio_list;
		blip->li_bio_list = NULL;

		iip = INODE_ITEM(blip);
		iip->ili_logged = 0;
		iip->ili_last_fields = 0;
		xfs_ifunlock(iip->ili_inode);
	}
L
Linus Torvalds 已提交
780 781 782
}

/*
783 784 785 786
 * This is the inode flushing abort routine.  It is called from xfs_iflush when
 * the filesystem is shutting down to clean up the inode state.  It is
 * responsible for removing the inode item from the AIL if it has not been
 * re-logged, and unlocking the inode's flush lock.
L
Linus Torvalds 已提交
787 788 789
 */
void
xfs_iflush_abort(
790 791
	xfs_inode_t		*ip,
	bool			stale)
L
Linus Torvalds 已提交
792
{
793
	xfs_inode_log_item_t	*iip = ip->i_itemp;
L
Linus Torvalds 已提交
794 795

	if (iip) {
796
		struct xfs_ail	*ailp = iip->ili_item.li_ailp;
L
Linus Torvalds 已提交
797
		if (iip->ili_item.li_flags & XFS_LI_IN_AIL) {
798
			spin_lock(&ailp->xa_lock);
L
Linus Torvalds 已提交
799
			if (iip->ili_item.li_flags & XFS_LI_IN_AIL) {
800
				/* xfs_trans_ail_delete() drops the AIL lock. */
801 802 803 804
				xfs_trans_ail_delete(ailp, &iip->ili_item,
						stale ?
						     SHUTDOWN_LOG_IO_ERROR :
						     SHUTDOWN_CORRUPT_INCORE);
L
Linus Torvalds 已提交
805
			} else
806
				spin_unlock(&ailp->xa_lock);
L
Linus Torvalds 已提交
807 808 809 810 811 812 813 814 815 816 817
		}
		iip->ili_logged = 0;
		/*
		 * Clear the ili_last_fields bits now that we know that the
		 * data corresponding to them is safely on disk.
		 */
		iip->ili_last_fields = 0;
		/*
		 * Clear the inode logging fields so no more flushes are
		 * attempted.
		 */
818
		iip->ili_fields = 0;
L
Linus Torvalds 已提交
819 820 821 822 823 824 825 826 827
	}
	/*
	 * Release the inode's flush lock since we're done with it.
	 */
	xfs_ifunlock(ip);
}

void
xfs_istale_done(
828 829
	struct xfs_buf		*bp,
	struct xfs_log_item	*lip)
L
Linus Torvalds 已提交
830
{
831
	xfs_iflush_abort(INODE_ITEM(lip)->ili_inode, true);
L
Linus Torvalds 已提交
832
}
833 834 835 836 837 838 839 840 841 842 843

/*
 * convert an xfs_inode_log_format struct from either 32 or 64 bit versions
 * (which can have different field alignments) to the native version
 */
int
xfs_inode_item_format_convert(
	xfs_log_iovec_t		*buf,
	xfs_inode_log_format_t	*in_f)
{
	if (buf->i_len == sizeof(xfs_inode_log_format_32_t)) {
844
		xfs_inode_log_format_32_t *in_f32 = buf->i_addr;
845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860

		in_f->ilf_type = in_f32->ilf_type;
		in_f->ilf_size = in_f32->ilf_size;
		in_f->ilf_fields = in_f32->ilf_fields;
		in_f->ilf_asize = in_f32->ilf_asize;
		in_f->ilf_dsize = in_f32->ilf_dsize;
		in_f->ilf_ino = in_f32->ilf_ino;
		/* copy biggest field of ilf_u */
		memcpy(in_f->ilf_u.ilfu_uuid.__u_bits,
		       in_f32->ilf_u.ilfu_uuid.__u_bits,
		       sizeof(uuid_t));
		in_f->ilf_blkno = in_f32->ilf_blkno;
		in_f->ilf_len = in_f32->ilf_len;
		in_f->ilf_boffset = in_f32->ilf_boffset;
		return 0;
	} else if (buf->i_len == sizeof(xfs_inode_log_format_64_t)){
861
		xfs_inode_log_format_64_t *in_f64 = buf->i_addr;
862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879

		in_f->ilf_type = in_f64->ilf_type;
		in_f->ilf_size = in_f64->ilf_size;
		in_f->ilf_fields = in_f64->ilf_fields;
		in_f->ilf_asize = in_f64->ilf_asize;
		in_f->ilf_dsize = in_f64->ilf_dsize;
		in_f->ilf_ino = in_f64->ilf_ino;
		/* copy biggest field of ilf_u */
		memcpy(in_f->ilf_u.ilfu_uuid.__u_bits,
		       in_f64->ilf_u.ilfu_uuid.__u_bits,
		       sizeof(uuid_t));
		in_f->ilf_blkno = in_f64->ilf_blkno;
		in_f->ilf_len = in_f64->ilf_len;
		in_f->ilf_boffset = in_f64->ilf_boffset;
		return 0;
	}
	return EFSCORRUPTED;
}