xfs_inode_item.c 20.9 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"
20
#include "xfs_format.h"
21 22
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
L
Linus Torvalds 已提交
23
#include "xfs_sb.h"
24
#include "xfs_ag.h"
L
Linus Torvalds 已提交
25 26
#include "xfs_mount.h"
#include "xfs_inode.h"
27
#include "xfs_trans.h"
28
#include "xfs_inode_item.h"
29
#include "xfs_error.h"
C
Christoph Hellwig 已提交
30
#include "xfs_trace.h"
31
#include "xfs_trans_priv.h"
32
#include "xfs_dinode.h"
C
Christoph Hellwig 已提交
33
#include "xfs_log.h"
L
Linus Torvalds 已提交
34 35 36 37


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

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);
}

43
STATIC void
44 45
xfs_inode_item_data_fork_size(
	struct xfs_inode_log_item *iip,
46 47
	int			*nvecs,
	int			*nbytes)
L
Linus Torvalds 已提交
48
{
49
	struct xfs_inode	*ip = iip->ili_inode;
50

L
Linus Torvalds 已提交
51 52
	switch (ip->i_d.di_format) {
	case XFS_DINODE_FMT_EXTENTS:
53
		if ((iip->ili_fields & XFS_ILOG_DEXT) &&
54
		    ip->i_d.di_nextents > 0 &&
55 56 57 58 59
		    ip->i_df.if_bytes > 0) {
			/* worst case, doesn't subtract delalloc extents */
			*nbytes += XFS_IFORK_DSIZE(ip);
			*nvecs += 1;
		}
L
Linus Torvalds 已提交
60 61
		break;
	case XFS_DINODE_FMT_BTREE:
62
		if ((iip->ili_fields & XFS_ILOG_DBROOT) &&
63 64 65 66
		    ip->i_df.if_broot_bytes > 0) {
			*nbytes += ip->i_df.if_broot_bytes;
			*nvecs += 1;
		}
L
Linus Torvalds 已提交
67 68
		break;
	case XFS_DINODE_FMT_LOCAL:
69
		if ((iip->ili_fields & XFS_ILOG_DDATA) &&
70 71 72 73
		    ip->i_df.if_bytes > 0) {
			*nbytes += roundup(ip->i_df.if_bytes, 4);
			*nvecs += 1;
		}
L
Linus Torvalds 已提交
74 75 76 77 78 79 80 81 82
		break;

	case XFS_DINODE_FMT_DEV:
	case XFS_DINODE_FMT_UUID:
		break;
	default:
		ASSERT(0);
		break;
	}
83
}
L
Linus Torvalds 已提交
84

85 86 87 88 89 90 91
STATIC void
xfs_inode_item_attr_fork_size(
	struct xfs_inode_log_item *iip,
	int			*nvecs,
	int			*nbytes)
{
	struct xfs_inode	*ip = iip->ili_inode;
L
Linus Torvalds 已提交
92 93 94

	switch (ip->i_d.di_aformat) {
	case XFS_DINODE_FMT_EXTENTS:
95
		if ((iip->ili_fields & XFS_ILOG_AEXT) &&
96
		    ip->i_d.di_anextents > 0 &&
97 98 99 100 101
		    ip->i_afp->if_bytes > 0) {
			/* worst case, doesn't subtract unused space */
			*nbytes += XFS_IFORK_ASIZE(ip);
			*nvecs += 1;
		}
L
Linus Torvalds 已提交
102 103
		break;
	case XFS_DINODE_FMT_BTREE:
104
		if ((iip->ili_fields & XFS_ILOG_ABROOT) &&
105 106 107 108
		    ip->i_afp->if_broot_bytes > 0) {
			*nbytes += ip->i_afp->if_broot_bytes;
			*nvecs += 1;
		}
L
Linus Torvalds 已提交
109 110
		break;
	case XFS_DINODE_FMT_LOCAL:
111
		if ((iip->ili_fields & XFS_ILOG_ADATA) &&
112 113 114 115
		    ip->i_afp->if_bytes > 0) {
			*nbytes += roundup(ip->i_afp->if_bytes, 4);
			*nvecs += 1;
		}
L
Linus Torvalds 已提交
116 117 118 119 120 121 122
		break;
	default:
		ASSERT(0);
		break;
	}
}

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
/*
 * 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 void
xfs_inode_item_size(
	struct xfs_log_item	*lip,
	int			*nvecs,
	int			*nbytes)
{
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;

	*nvecs += 2;
	*nbytes += sizeof(struct xfs_inode_log_format) +
		   xfs_icdinode_size(ip->i_d.di_version);

	xfs_inode_item_data_fork_size(iip, nvecs, nbytes);
	if (XFS_IFORK_Q(ip))
		xfs_inode_item_attr_fork_size(iip, nvecs, nbytes);
}

C
Christoph Hellwig 已提交
148
STATIC void
149 150
xfs_inode_item_format_data_fork(
	struct xfs_inode_log_item *iip,
151 152 153
	struct xfs_inode_log_format *ilf,
	struct xfs_log_vec	*lv,
	struct xfs_log_iovec	**vecp)
L
Linus Torvalds 已提交
154
{
155
	struct xfs_inode	*ip = iip->ili_inode;
L
Linus Torvalds 已提交
156 157 158 159
	size_t			data_bytes;

	switch (ip->i_d.di_format) {
	case XFS_DINODE_FMT_EXTENTS:
160
		iip->ili_fields &=
161 162 163
			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
			  XFS_ILOG_DEV | XFS_ILOG_UUID);

164
		if ((iip->ili_fields & XFS_ILOG_DEXT) &&
165 166
		    ip->i_d.di_nextents > 0 &&
		    ip->i_df.if_bytes > 0) {
167 168
			struct xfs_bmbt_rec *p;

L
Linus Torvalds 已提交
169
			ASSERT(ip->i_df.if_u1.if_extents != NULL);
170
			ASSERT(ip->i_df.if_bytes / sizeof(xfs_bmbt_rec_t) > 0);
171 172 173 174 175 176 177 178

			p = xlog_prepare_iovec(lv, vecp, XLOG_REG_TYPE_IEXT);
			data_bytes = xfs_iextents_copy(ip, p, XFS_DATA_FORK);
			xlog_finish_iovec(lv, *vecp, data_bytes);

			ASSERT(data_bytes <= ip->i_df.if_bytes);

			ilf->ilf_dsize = data_bytes;
179
			ilf->ilf_size++;
180
		} else {
181
			iip->ili_fields &= ~XFS_ILOG_DEXT;
L
Linus Torvalds 已提交
182 183 184
		}
		break;
	case XFS_DINODE_FMT_BTREE:
185
		iip->ili_fields &=
186 187 188
			~(XFS_ILOG_DDATA | XFS_ILOG_DEXT |
			  XFS_ILOG_DEV | XFS_ILOG_UUID);

189
		if ((iip->ili_fields & XFS_ILOG_DBROOT) &&
190
		    ip->i_df.if_broot_bytes > 0) {
L
Linus Torvalds 已提交
191
			ASSERT(ip->i_df.if_broot != NULL);
192
			xlog_copy_iovec(lv, vecp, XLOG_REG_TYPE_IBROOT,
C
Christoph Hellwig 已提交
193 194
					ip->i_df.if_broot,
					ip->i_df.if_broot_bytes);
195 196
			ilf->ilf_dsize = ip->i_df.if_broot_bytes;
			ilf->ilf_size++;
197
		} else {
198
			ASSERT(!(iip->ili_fields &
199
				 XFS_ILOG_DBROOT));
200
			iip->ili_fields &= ~XFS_ILOG_DBROOT;
L
Linus Torvalds 已提交
201 202 203
		}
		break;
	case XFS_DINODE_FMT_LOCAL:
204
		iip->ili_fields &=
205 206
			~(XFS_ILOG_DEXT | XFS_ILOG_DBROOT |
			  XFS_ILOG_DEV | XFS_ILOG_UUID);
207
		if ((iip->ili_fields & XFS_ILOG_DDATA) &&
208
		    ip->i_df.if_bytes > 0) {
L
Linus Torvalds 已提交
209 210 211 212 213 214
			/*
			 * 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);
C
Christoph Hellwig 已提交
215 216 217 218
			ASSERT(ip->i_df.if_real_bytes == 0 ||
			       ip->i_df.if_real_bytes == data_bytes);
			ASSERT(ip->i_df.if_u1.if_data != NULL);
			ASSERT(ip->i_d.di_size > 0);
219
			xlog_copy_iovec(lv, vecp, XLOG_REG_TYPE_ILOCAL,
C
Christoph Hellwig 已提交
220
					ip->i_df.if_u1.if_data, data_bytes);
221 222
			ilf->ilf_dsize = (unsigned)data_bytes;
			ilf->ilf_size++;
223
		} else {
224
			iip->ili_fields &= ~XFS_ILOG_DDATA;
L
Linus Torvalds 已提交
225 226 227
		}
		break;
	case XFS_DINODE_FMT_DEV:
228
		iip->ili_fields &=
229 230
			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
			  XFS_ILOG_DEXT | XFS_ILOG_UUID);
231 232
		if (iip->ili_fields & XFS_ILOG_DEV)
			ilf->ilf_u.ilfu_rdev = ip->i_df.if_u2.if_rdev;
L
Linus Torvalds 已提交
233 234
		break;
	case XFS_DINODE_FMT_UUID:
235
		iip->ili_fields &=
236 237
			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
			  XFS_ILOG_DEXT | XFS_ILOG_DEV);
238 239
		if (iip->ili_fields & XFS_ILOG_UUID)
			ilf->ilf_u.ilfu_uuid = ip->i_df.if_u2.if_uuid;
L
Linus Torvalds 已提交
240 241 242 243 244
		break;
	default:
		ASSERT(0);
		break;
	}
245 246
}

C
Christoph Hellwig 已提交
247
STATIC void
248 249
xfs_inode_item_format_attr_fork(
	struct xfs_inode_log_item *iip,
250 251 252
	struct xfs_inode_log_format *ilf,
	struct xfs_log_vec	*lv,
	struct xfs_log_iovec	**vecp)
253 254 255
{
	struct xfs_inode	*ip = iip->ili_inode;
	size_t			data_bytes;
L
Linus Torvalds 已提交
256 257 258

	switch (ip->i_d.di_aformat) {
	case XFS_DINODE_FMT_EXTENTS:
259
		iip->ili_fields &=
260 261
			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT);

262
		if ((iip->ili_fields & XFS_ILOG_AEXT) &&
263 264
		    ip->i_d.di_anextents > 0 &&
		    ip->i_afp->if_bytes > 0) {
265 266
			struct xfs_bmbt_rec *p;

267 268
			ASSERT(ip->i_afp->if_bytes / sizeof(xfs_bmbt_rec_t) ==
				ip->i_d.di_anextents);
269
			ASSERT(ip->i_afp->if_u1.if_extents != NULL);
270 271 272 273 274 275

			p = xlog_prepare_iovec(lv, vecp, XLOG_REG_TYPE_IATTR_EXT);
			data_bytes = xfs_iextents_copy(ip, p, XFS_ATTR_FORK);
			xlog_finish_iovec(lv, *vecp, data_bytes);

			ilf->ilf_asize = data_bytes;
276
			ilf->ilf_size++;
277
		} else {
278
			iip->ili_fields &= ~XFS_ILOG_AEXT;
L
Linus Torvalds 已提交
279 280 281
		}
		break;
	case XFS_DINODE_FMT_BTREE:
282
		iip->ili_fields &=
283 284
			~(XFS_ILOG_ADATA | XFS_ILOG_AEXT);

285
		if ((iip->ili_fields & XFS_ILOG_ABROOT) &&
286
		    ip->i_afp->if_broot_bytes > 0) {
L
Linus Torvalds 已提交
287
			ASSERT(ip->i_afp->if_broot != NULL);
288

289
			xlog_copy_iovec(lv, vecp, XLOG_REG_TYPE_IATTR_BROOT,
C
Christoph Hellwig 已提交
290 291
					ip->i_afp->if_broot,
					ip->i_afp->if_broot_bytes);
292 293
			ilf->ilf_asize = ip->i_afp->if_broot_bytes;
			ilf->ilf_size++;
294
		} else {
295
			iip->ili_fields &= ~XFS_ILOG_ABROOT;
L
Linus Torvalds 已提交
296 297 298
		}
		break;
	case XFS_DINODE_FMT_LOCAL:
299
		iip->ili_fields &=
300 301
			~(XFS_ILOG_AEXT | XFS_ILOG_ABROOT);

302
		if ((iip->ili_fields & XFS_ILOG_ADATA) &&
303
		    ip->i_afp->if_bytes > 0) {
L
Linus Torvalds 已提交
304 305 306 307 308 309
			/*
			 * 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);
C
Christoph Hellwig 已提交
310 311 312
			ASSERT(ip->i_afp->if_real_bytes == 0 ||
			       ip->i_afp->if_real_bytes == data_bytes);
			ASSERT(ip->i_afp->if_u1.if_data != NULL);
313
			xlog_copy_iovec(lv, vecp, XLOG_REG_TYPE_IATTR_LOCAL,
C
Christoph Hellwig 已提交
314 315
					ip->i_afp->if_u1.if_data,
					data_bytes);
316 317
			ilf->ilf_asize = (unsigned)data_bytes;
			ilf->ilf_size++;
318
		} else {
319
			iip->ili_fields &= ~XFS_ILOG_ADATA;
L
Linus Torvalds 已提交
320 321 322 323 324 325
		}
		break;
	default:
		ASSERT(0);
		break;
	}
326 327 328 329 330 331 332 333 334 335 336 337
}

/*
 * 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(
	struct xfs_log_item	*lip,
338
	struct xfs_log_vec	*lv)
339 340 341
{
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;
342 343
	struct xfs_inode_log_format *ilf;
	struct xfs_log_iovec	*vecp = NULL;
344

345 346
	ASSERT(ip->i_d.di_version > 1);

347 348 349 350 351 352 353 354 355
	ilf = xlog_prepare_iovec(lv, &vecp, XLOG_REG_TYPE_IFORMAT);
	ilf->ilf_type = XFS_LI_INODE;
	ilf->ilf_ino = ip->i_ino;
	ilf->ilf_blkno = ip->i_imap.im_blkno;
	ilf->ilf_len = ip->i_imap.im_len;
	ilf->ilf_boffset = ip->i_imap.im_boffset;
	ilf->ilf_fields = XFS_ILOG_CORE;
	ilf->ilf_size = 2; /* format + core */
	xlog_finish_iovec(lv, vecp, sizeof(struct xfs_inode_log_format));
356

357 358 359
	xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_ICORE,
			&ip->i_d,
			xfs_icdinode_size(ip->i_d.di_version));
360

361
	xfs_inode_item_format_data_fork(iip, ilf, lv, &vecp);
362
	if (XFS_IFORK_Q(ip)) {
363
		xfs_inode_item_format_attr_fork(iip, ilf, lv, &vecp);
364 365 366 367 368
	} else {
		iip->ili_fields &=
			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT);
	}

369 370
	/* update the format with the exact fields we actually logged */
	ilf->ilf_fields |= (iip->ili_fields & ~XFS_ILOG_TIMESTAMP);
L
Linus Torvalds 已提交
371 372 373 374
}

/*
 * This is called to pin the inode associated with the inode log
375
 * item in memory so it cannot be written out.
L
Linus Torvalds 已提交
376 377 378
 */
STATIC void
xfs_inode_item_pin(
379
	struct xfs_log_item	*lip)
L
Linus Torvalds 已提交
380
{
381
	struct xfs_inode	*ip = INODE_ITEM(lip)->ili_inode;
382

383 384 385 386
	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
387 388 389 390 391 392
}


/*
 * 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().
393 394
 *
 * Also wake up anyone in xfs_iunpin_wait() if the count goes to 0.
L
Linus Torvalds 已提交
395 396 397
 */
STATIC void
xfs_inode_item_unpin(
398
	struct xfs_log_item	*lip,
399
	int			remove)
L
Linus Torvalds 已提交
400
{
401
	struct xfs_inode	*ip = INODE_ITEM(lip)->ili_inode;
402

403
	trace_xfs_inode_unpin(ip, _RET_IP_);
404 405
	ASSERT(atomic_read(&ip->i_pincount) > 0);
	if (atomic_dec_and_test(&ip->i_pincount))
406
		wake_up_bit(&ip->i_flags, __XFS_IPINNED_BIT);
L
Linus Torvalds 已提交
407 408 409
}

STATIC uint
410 411 412
xfs_inode_item_push(
	struct xfs_log_item	*lip,
	struct list_head	*buffer_list)
L
Linus Torvalds 已提交
413
{
414 415
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;
416 417 418
	struct xfs_buf		*bp = NULL;
	uint			rval = XFS_ITEM_SUCCESS;
	int			error;
L
Linus Torvalds 已提交
419

420
	if (xfs_ipincount(ip) > 0)
L
Linus Torvalds 已提交
421 422
		return XFS_ITEM_PINNED;

423
	if (!xfs_ilock_nowait(ip, XFS_ILOCK_SHARED))
L
Linus Torvalds 已提交
424 425
		return XFS_ITEM_LOCKED;

426 427 428 429 430
	/*
	 * Re-check the pincount now that we stabilized the value by
	 * taking the ilock.
	 */
	if (xfs_ipincount(ip) > 0) {
431 432
		rval = XFS_ITEM_PINNED;
		goto out_unlock;
433 434
	}

435 436 437 438 439 440 441 442
	/*
	 * Stale inode items should force out the iclog.
	 */
	if (ip->i_flags & XFS_ISTALE) {
		rval = XFS_ITEM_PINNED;
		goto out_unlock;
	}

443 444 445 446 447
	/*
	 * 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
448
	if (!xfs_iflock_nowait(ip)) {
449 450
		rval = XFS_ITEM_FLUSHING;
		goto out_unlock;
L
Linus Torvalds 已提交
451 452
	}

453 454 455 456 457 458 459 460 461 462
	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
463
	}
464 465 466 467 468

	spin_lock(&lip->li_ailp->xa_lock);
out_unlock:
	xfs_iunlock(ip, XFS_ILOCK_SHARED);
	return rval;
L
Linus Torvalds 已提交
469 470 471 472 473 474 475 476 477 478
}

/*
 * 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(
479
	struct xfs_log_item	*lip)
L
Linus Torvalds 已提交
480
{
481 482
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;
483
	unsigned short		lock_flags;
L
Linus Torvalds 已提交
484

C
Christoph Hellwig 已提交
485 486
	ASSERT(ip->i_itemp != NULL);
	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
L
Linus Torvalds 已提交
487

488 489
	lock_flags = iip->ili_lock_flags;
	iip->ili_lock_flags = 0;
490
	if (lock_flags)
C
Christoph Hellwig 已提交
491
		xfs_iunlock(ip, lock_flags);
L
Linus Torvalds 已提交
492 493 494
}

/*
495 496 497 498 499 500 501 502 503 504 505
 * 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
506
 * for the buffer to be written and IO completes before the inode is inserted
507 508 509 510 511
 * 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.
 *
512 513 514
 * 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
515 516 517
 */
STATIC xfs_lsn_t
xfs_inode_item_committed(
518
	struct xfs_log_item	*lip,
L
Linus Torvalds 已提交
519 520
	xfs_lsn_t		lsn)
{
521 522 523
	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
	struct xfs_inode	*ip = iip->ili_inode;

524 525 526 527
	if (xfs_iflags_test(ip, XFS_ISTALE)) {
		xfs_inode_item_unpin(lip, 0);
		return -1;
	}
528
	return lsn;
L
Linus Torvalds 已提交
529 530 531 532 533 534 535 536
}

/*
 * 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(
537
	struct xfs_log_item	*lip,
L
Linus Torvalds 已提交
538 539
	xfs_lsn_t		lsn)
{
540
	INODE_ITEM(lip)->ili_last_lsn = lsn;
L
Linus Torvalds 已提交
541 542 543 544 545
}

/*
 * This is the ops vector shared by all buf log items.
 */
C
Christoph Hellwig 已提交
546
static const struct xfs_item_ops xfs_inode_item_ops = {
547 548 549 550 551 552 553 554
	.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
555 556 557 558 559 560 561 562
};


/*
 * Initialize the inode log item for a newly allocated (in-core) inode.
 */
void
xfs_inode_item_init(
563 564
	struct xfs_inode	*ip,
	struct xfs_mount	*mp)
L
Linus Torvalds 已提交
565
{
566
	struct xfs_inode_log_item *iip;
L
Linus Torvalds 已提交
567 568 569 570 571

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

	iip->ili_inode = ip;
572 573
	xfs_log_item_init(mp, &iip->ili_item, XFS_LI_INODE,
						&xfs_inode_item_ops);
L
Linus Torvalds 已提交
574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592
}

/*
 * 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.
593 594 595 596 597
 *
 * 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
598 599 600
 */
void
xfs_iflush_done(
601 602
	struct xfs_buf		*bp,
	struct xfs_log_item	*lip)
L
Linus Torvalds 已提交
603
{
604 605 606 607
	struct xfs_inode_log_item *iip;
	struct xfs_log_item	*blip;
	struct xfs_log_item	*next;
	struct xfs_log_item	*prev;
608
	struct xfs_ail		*ailp = lip->li_ailp;
609 610 611 612 613 614
	int			need_ail = 0;

	/*
	 * Scan the buffer IO completions for other inodes being completed and
	 * attach them to the current inode log item.
	 */
615
	blip = bp->b_fspriv;
616 617 618 619 620 621 622 623 624 625 626
	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) {
627
			bp->b_fspriv = next;
628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650
		} 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
651 652 653 654 655 656 657 658 659 660

	/*
	 * 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.
	 */
661 662 663
	if (need_ail) {
		struct xfs_log_item *log_items[need_ail];
		int i = 0;
664
		spin_lock(&ailp->xa_lock);
665 666 667 668 669 670 671
		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
672
		}
673
		/* xfs_trans_ail_delete_bulk() drops the AIL lock. */
674 675
		xfs_trans_ail_delete_bulk(ailp, log_items, i,
					  SHUTDOWN_CORRUPT_INCORE);
L
Linus Torvalds 已提交
676 677 678 679
	}


	/*
680 681 682
	 * 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
683
	 */
684 685 686 687 688 689 690 691 692
	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
693 694 695
}

/*
696 697 698 699
 * 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
700 701 702
 */
void
xfs_iflush_abort(
703 704
	xfs_inode_t		*ip,
	bool			stale)
L
Linus Torvalds 已提交
705
{
706
	xfs_inode_log_item_t	*iip = ip->i_itemp;
L
Linus Torvalds 已提交
707 708

	if (iip) {
709
		struct xfs_ail	*ailp = iip->ili_item.li_ailp;
L
Linus Torvalds 已提交
710
		if (iip->ili_item.li_flags & XFS_LI_IN_AIL) {
711
			spin_lock(&ailp->xa_lock);
L
Linus Torvalds 已提交
712
			if (iip->ili_item.li_flags & XFS_LI_IN_AIL) {
713
				/* xfs_trans_ail_delete() drops the AIL lock. */
714 715 716 717
				xfs_trans_ail_delete(ailp, &iip->ili_item,
						stale ?
						     SHUTDOWN_LOG_IO_ERROR :
						     SHUTDOWN_CORRUPT_INCORE);
L
Linus Torvalds 已提交
718
			} else
719
				spin_unlock(&ailp->xa_lock);
L
Linus Torvalds 已提交
720 721 722 723 724 725 726 727 728 729 730
		}
		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.
		 */
731
		iip->ili_fields = 0;
L
Linus Torvalds 已提交
732 733 734 735 736 737 738 739 740
	}
	/*
	 * Release the inode's flush lock since we're done with it.
	 */
	xfs_ifunlock(ip);
}

void
xfs_istale_done(
741 742
	struct xfs_buf		*bp,
	struct xfs_log_item	*lip)
L
Linus Torvalds 已提交
743
{
744
	xfs_iflush_abort(INODE_ITEM(lip)->ili_inode, true);
L
Linus Torvalds 已提交
745
}
746 747 748 749 750 751 752 753 754 755 756

/*
 * 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)) {
757
		xfs_inode_log_format_32_t *in_f32 = buf->i_addr;
758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773

		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)){
774
		xfs_inode_log_format_64_t *in_f64 = buf->i_addr;
775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790

		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;
	}
D
Dave Chinner 已提交
791
	return -EFSCORRUPTED;
792
}