2 * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation.
9 * This program is distributed in the hope that it would be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, write the Free Software Foundation,
16 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
20 #include "xfs_types.h"
23 #include "xfs_trans.h"
26 #include "xfs_mount.h"
27 #include "xfs_bmap_btree.h"
28 #include "xfs_alloc_btree.h"
29 #include "xfs_ialloc_btree.h"
30 #include "xfs_dinode.h"
31 #include "xfs_inode.h"
32 #include "xfs_ialloc.h"
33 #include "xfs_itable.h"
34 #include "xfs_error.h"
35 #include "xfs_btree.h"
36 #include "xfs_trace.h"
37 #include "xfs_icache.h"
44 return (ino == mp->m_sb.sb_rbmino || ino == mp->m_sb.sb_rsumino ||
45 (xfs_sb_version_hasquota(&mp->m_sb) &&
46 xfs_is_quota_inode(&mp->m_sb, ino)));
50 * Return stat information for one inode.
51 * Return 0 if ok, else errno.
55 struct xfs_mount *mp, /* mount point for filesystem */
56 xfs_ino_t ino, /* inode to get data for */
57 void __user *buffer, /* buffer to place output in */
58 int ubsize, /* size of buffer */
59 bulkstat_one_fmt_pf formatter, /* formatter, copy to user */
60 int *ubused, /* bytes used by me */
61 int *stat) /* BULKSTAT_RV_... */
63 struct xfs_icdinode *dic; /* dinode core info pointer */
64 struct xfs_inode *ip; /* incore inode pointer */
65 struct xfs_bstat *buf; /* return buffer */
66 int error = 0; /* error value */
68 *stat = BULKSTAT_RV_NOTHING;
70 if (!buffer || xfs_internal_inum(mp, ino))
71 return XFS_ERROR(EINVAL);
73 buf = kmem_alloc(sizeof(*buf), KM_SLEEP | KM_MAYFAIL);
75 return XFS_ERROR(ENOMEM);
77 error = xfs_iget(mp, NULL, ino,
78 (XFS_IGET_DONTCACHE | XFS_IGET_UNTRUSTED),
79 XFS_ILOCK_SHARED, &ip);
81 *stat = BULKSTAT_RV_NOTHING;
86 ASSERT(ip->i_imap.im_blkno != 0);
90 /* xfs_iget returns the following without needing
93 buf->bs_nlink = dic->di_nlink;
94 buf->bs_projid_lo = dic->di_projid_lo;
95 buf->bs_projid_hi = dic->di_projid_hi;
97 buf->bs_mode = dic->di_mode;
98 buf->bs_uid = dic->di_uid;
99 buf->bs_gid = dic->di_gid;
100 buf->bs_size = dic->di_size;
101 buf->bs_atime.tv_sec = dic->di_atime.t_sec;
102 buf->bs_atime.tv_nsec = dic->di_atime.t_nsec;
103 buf->bs_mtime.tv_sec = dic->di_mtime.t_sec;
104 buf->bs_mtime.tv_nsec = dic->di_mtime.t_nsec;
105 buf->bs_ctime.tv_sec = dic->di_ctime.t_sec;
106 buf->bs_ctime.tv_nsec = dic->di_ctime.t_nsec;
107 buf->bs_xflags = xfs_ip2xflags(ip);
108 buf->bs_extsize = dic->di_extsize << mp->m_sb.sb_blocklog;
109 buf->bs_extents = dic->di_nextents;
110 buf->bs_gen = dic->di_gen;
111 memset(buf->bs_pad, 0, sizeof(buf->bs_pad));
112 buf->bs_dmevmask = dic->di_dmevmask;
113 buf->bs_dmstate = dic->di_dmstate;
114 buf->bs_aextents = dic->di_anextents;
115 buf->bs_forkoff = XFS_IFORK_BOFF(ip);
117 switch (dic->di_format) {
118 case XFS_DINODE_FMT_DEV:
119 buf->bs_rdev = ip->i_df.if_u2.if_rdev;
120 buf->bs_blksize = BLKDEV_IOSIZE;
123 case XFS_DINODE_FMT_LOCAL:
124 case XFS_DINODE_FMT_UUID:
126 buf->bs_blksize = mp->m_sb.sb_blocksize;
129 case XFS_DINODE_FMT_EXTENTS:
130 case XFS_DINODE_FMT_BTREE:
132 buf->bs_blksize = mp->m_sb.sb_blocksize;
133 buf->bs_blocks = dic->di_nblocks + ip->i_delayed_blks;
136 xfs_iunlock(ip, XFS_ILOCK_SHARED);
139 error = formatter(buffer, ubsize, ubused, buf);
142 *stat = BULKSTAT_RV_DIDONE;
149 /* Return 0 on success or positive error */
151 xfs_bulkstat_one_fmt(
152 void __user *ubuffer,
155 const xfs_bstat_t *buffer)
157 if (ubsize < sizeof(*buffer))
158 return XFS_ERROR(ENOMEM);
159 if (copy_to_user(ubuffer, buffer, sizeof(*buffer)))
160 return XFS_ERROR(EFAULT);
162 *ubused = sizeof(*buffer);
168 xfs_mount_t *mp, /* mount point for filesystem */
169 xfs_ino_t ino, /* inode number to get data for */
170 void __user *buffer, /* buffer to place output in */
171 int ubsize, /* size of buffer */
172 int *ubused, /* bytes used by me */
173 int *stat) /* BULKSTAT_RV_... */
175 return xfs_bulkstat_one_int(mp, ino, buffer, ubsize,
176 xfs_bulkstat_one_fmt, ubused, stat);
179 #define XFS_BULKSTAT_UBLEFT(ubleft) ((ubleft) >= statstruct_size)
182 * Return stat information in bulk (by-inode) for the filesystem.
184 int /* error status */
186 xfs_mount_t *mp, /* mount point for filesystem */
187 xfs_ino_t *lastinop, /* last inode returned */
188 int *ubcountp, /* size of buffer/count returned */
189 bulkstat_one_pf formatter, /* func that'd fill a single buf */
190 size_t statstruct_size, /* sizeof struct filling */
191 char __user *ubuffer, /* buffer with inode stats */
192 int *done) /* 1 if there are more stats to get */
194 xfs_agblock_t agbno=0;/* allocation group block number */
195 xfs_buf_t *agbp; /* agi header buffer */
196 xfs_agi_t *agi; /* agi header data */
197 xfs_agino_t agino; /* inode # in allocation group */
198 xfs_agnumber_t agno; /* allocation group number */
199 int chunkidx; /* current index into inode chunk */
200 int clustidx; /* current index into inode cluster */
201 xfs_btree_cur_t *cur; /* btree cursor for ialloc btree */
202 int end_of_ag; /* set if we've seen the ag end */
203 int error; /* error code */
204 int fmterror;/* bulkstat formatter result */
205 int i; /* loop index */
206 int icount; /* count of inodes good in irbuf */
207 size_t irbsize; /* size of irec buffer in bytes */
208 xfs_ino_t ino; /* inode number (filesystem) */
209 xfs_inobt_rec_incore_t *irbp; /* current irec buffer pointer */
210 xfs_inobt_rec_incore_t *irbuf; /* start of irec buffer */
211 xfs_inobt_rec_incore_t *irbufend; /* end of good irec buffer entries */
212 xfs_ino_t lastino; /* last inode number returned */
213 int nbcluster; /* # of blocks in a cluster */
214 int nicluster; /* # of inodes in a cluster */
215 int nimask; /* mask for inode clusters */
216 int nirbuf; /* size of irbuf */
217 int rval; /* return value error code */
218 int tmp; /* result value from btree calls */
219 int ubcount; /* size of user's buffer */
220 int ubleft; /* bytes left in user's buffer */
221 char __user *ubufp; /* pointer into user's buffer */
222 int ubelem; /* spaces used in user's buffer */
223 int ubused; /* bytes used by formatter */
226 * Get the last inode value, see if there's nothing to do.
228 ino = (xfs_ino_t)*lastinop;
230 agno = XFS_INO_TO_AGNO(mp, ino);
231 agino = XFS_INO_TO_AGINO(mp, ino);
232 if (agno >= mp->m_sb.sb_agcount ||
233 ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
238 if (!ubcountp || *ubcountp <= 0) {
241 ubcount = *ubcountp; /* statstruct's */
242 ubleft = ubcount * statstruct_size; /* bytes */
243 *ubcountp = ubelem = 0;
247 nicluster = mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(mp) ?
248 mp->m_sb.sb_inopblock :
249 (XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog);
250 nimask = ~(nicluster - 1);
251 nbcluster = nicluster >> mp->m_sb.sb_inopblog;
252 irbuf = kmem_zalloc_greedy(&irbsize, PAGE_SIZE, PAGE_SIZE * 4);
256 nirbuf = irbsize / sizeof(*irbuf);
259 * Loop over the allocation groups, starting from the last
260 * inode returned; 0 means start of the allocation group.
263 while (XFS_BULKSTAT_UBLEFT(ubleft) && agno < mp->m_sb.sb_agcount) {
265 error = xfs_ialloc_read_agi(mp, NULL, agno, &agbp);
268 * Skip this allocation group and go to the next one.
274 agi = XFS_BUF_TO_AGI(agbp);
276 * Allocate and initialize a btree cursor for ialloc btree.
278 cur = xfs_inobt_init_cursor(mp, NULL, agbp, agno);
280 irbufend = irbuf + nirbuf;
283 * If we're returning in the middle of an allocation group,
284 * we need to get the remainder of the chunk we're in.
287 xfs_inobt_rec_incore_t r;
290 * Lookup the inode chunk that this inode lives in.
292 error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE,
294 if (!error && /* no I/O error */
295 tmp && /* lookup succeeded */
296 /* got the record, should always work */
297 !(error = xfs_inobt_get_rec(cur, &r, &i)) &&
299 /* this is the right chunk */
300 agino < r.ir_startino + XFS_INODES_PER_CHUNK &&
301 /* lastino was not last in chunk */
302 (chunkidx = agino - r.ir_startino + 1) <
303 XFS_INODES_PER_CHUNK &&
304 /* there are some left allocated */
305 xfs_inobt_maskn(chunkidx,
306 XFS_INODES_PER_CHUNK - chunkidx) &
309 * Grab the chunk record. Mark all the
310 * uninteresting inodes (because they're
311 * before our start point) free.
313 for (i = 0; i < chunkidx; i++) {
314 if (XFS_INOBT_MASK(i) & ~r.ir_free)
317 r.ir_free |= xfs_inobt_maskn(0, chunkidx);
318 irbp->ir_startino = r.ir_startino;
319 irbp->ir_freecount = r.ir_freecount;
320 irbp->ir_free = r.ir_free;
322 agino = r.ir_startino + XFS_INODES_PER_CHUNK;
323 icount = XFS_INODES_PER_CHUNK - r.ir_freecount;
326 * If any of those tests failed, bump the
327 * inode number (just in case).
333 * In any case, increment to the next record.
336 error = xfs_btree_increment(cur, 0, &tmp);
339 * Start of ag. Lookup the first inode chunk.
341 error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &tmp);
345 * Loop through inode btree records in this ag,
346 * until we run out of inodes or space in the buffer.
348 while (irbp < irbufend && icount < ubcount) {
349 xfs_inobt_rec_incore_t r;
352 * Loop as long as we're unable to read the
356 agino += XFS_INODES_PER_CHUNK;
357 if (XFS_AGINO_TO_AGBNO(mp, agino) >=
358 be32_to_cpu(agi->agi_length))
360 error = xfs_inobt_lookup(cur, agino,
361 XFS_LOOKUP_GE, &tmp);
365 * If ran off the end of the ag either with an error,
366 * or the normal way, set end and stop collecting.
373 error = xfs_inobt_get_rec(cur, &r, &i);
374 if (error || i == 0) {
380 * If this chunk has any allocated inodes, save it.
381 * Also start read-ahead now for this chunk.
383 if (r.ir_freecount < XFS_INODES_PER_CHUNK) {
384 struct blk_plug plug;
386 * Loop over all clusters in the next chunk.
387 * Do a readahead if there are any allocated
388 * inodes in that cluster.
390 blk_start_plug(&plug);
391 agbno = XFS_AGINO_TO_AGBNO(mp, r.ir_startino);
393 chunkidx < XFS_INODES_PER_CHUNK;
394 chunkidx += nicluster,
395 agbno += nbcluster) {
396 if (xfs_inobt_maskn(chunkidx, nicluster)
398 xfs_btree_reada_bufs(mp, agno,
402 blk_finish_plug(&plug);
403 irbp->ir_startino = r.ir_startino;
404 irbp->ir_freecount = r.ir_freecount;
405 irbp->ir_free = r.ir_free;
407 icount += XFS_INODES_PER_CHUNK - r.ir_freecount;
410 * Set agino to after this chunk and bump the cursor.
412 agino = r.ir_startino + XFS_INODES_PER_CHUNK;
413 error = xfs_btree_increment(cur, 0, &tmp);
417 * Drop the btree buffers and the agi buffer.
418 * We can't hold any of the locks these represent
421 xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
424 * Now format all the good inodes into the user's buffer.
428 irbp < irbufend && XFS_BULKSTAT_UBLEFT(ubleft); irbp++) {
430 * Now process this chunk of inodes.
432 for (agino = irbp->ir_startino, chunkidx = clustidx = 0;
433 XFS_BULKSTAT_UBLEFT(ubleft) &&
434 irbp->ir_freecount < XFS_INODES_PER_CHUNK;
435 chunkidx++, clustidx++, agino++) {
436 ASSERT(chunkidx < XFS_INODES_PER_CHUNK);
438 ino = XFS_AGINO_TO_INO(mp, agno, agino);
440 * Skip if this inode is free.
442 if (XFS_INOBT_MASK(chunkidx) & irbp->ir_free) {
447 * Count used inodes as free so we can tell
448 * when the chunk is used up.
450 irbp->ir_freecount++;
453 * Get the inode and fill in a single buffer.
455 ubused = statstruct_size;
456 error = formatter(mp, ino, ubufp, ubleft,
458 if (fmterror == BULKSTAT_RV_NOTHING) {
459 if (error && error != ENOENT &&
468 if (fmterror == BULKSTAT_RV_GIVEUP) {
484 * Set up for the next loop iteration.
486 if (XFS_BULKSTAT_UBLEFT(ubleft)) {
491 agino = XFS_INO_TO_AGINO(mp, lastino);
496 * Done, we're either out of filesystem or space to put the data.
498 kmem_free_large(irbuf);
501 * Found some inodes, return them now and return the error next time.
505 if (agno >= mp->m_sb.sb_agcount) {
507 * If we ran out of filesystem, mark lastino as off
508 * the end of the filesystem, so the next call
509 * will return immediately.
511 *lastinop = (xfs_ino_t)XFS_AGINO_TO_INO(mp, agno, 0);
514 *lastinop = (xfs_ino_t)lastino;
520 * Return stat information in bulk (by-inode) for the filesystem.
521 * Special case for non-sequential one inode bulkstat.
523 int /* error status */
525 xfs_mount_t *mp, /* mount point for filesystem */
526 xfs_ino_t *lastinop, /* inode to return */
527 char __user *buffer, /* buffer with inode stats */
528 int *done) /* 1 if there are more stats to get */
530 int count; /* count value for bulkstat call */
531 int error; /* return value */
532 xfs_ino_t ino; /* filesystem inode number */
533 int res; /* result from bs1 */
536 * note that requesting valid inode numbers which are not allocated
537 * to inodes will most likely cause xfs_imap_to_bp to generate warning
538 * messages about bad magic numbers. This is ok. The fact that
539 * the inode isn't actually an inode is handled by the
540 * error check below. Done this way to make the usual case faster
541 * at the expense of the error case.
544 ino = (xfs_ino_t)*lastinop;
545 error = xfs_bulkstat_one(mp, ino, buffer, sizeof(xfs_bstat_t), 0, &res);
548 * Special case way failed, do it the "long" way
549 * to see if that works.
553 if (xfs_bulkstat(mp, lastinop, &count, xfs_bulkstat_one,
554 sizeof(xfs_bstat_t), buffer, done))
556 if (count == 0 || (xfs_ino_t)*lastinop != ino)
557 return error == EFSCORRUPTED ?
558 XFS_ERROR(EINVAL) : error;
568 void __user *ubuffer, /* buffer to write to */
569 const xfs_inogrp_t *buffer, /* buffer to read from */
570 long count, /* # of elements to read */
571 long *written) /* # of bytes written */
573 if (copy_to_user(ubuffer, buffer, count * sizeof(*buffer)))
575 *written = count * sizeof(*buffer);
580 * Return inode number table for the filesystem.
582 int /* error status */
584 xfs_mount_t *mp, /* mount point for filesystem */
585 xfs_ino_t *lastino, /* last inode returned */
586 int *count, /* size of buffer/count returned */
587 void __user *ubuffer,/* buffer with inode descriptions */
588 inumbers_fmt_pf formatter)
594 xfs_inogrp_t *buffer;
596 xfs_btree_cur_t *cur;
598 xfs_inobt_rec_incore_t r;
604 ino = (xfs_ino_t)*lastino;
605 agno = XFS_INO_TO_AGNO(mp, ino);
606 agino = XFS_INO_TO_AGINO(mp, ino);
609 bcount = MIN(left, (int)(PAGE_SIZE / sizeof(*buffer)));
610 buffer = kmem_alloc(bcount * sizeof(*buffer), KM_SLEEP);
614 while (left > 0 && agno < mp->m_sb.sb_agcount) {
616 error = xfs_ialloc_read_agi(mp, NULL, agno, &agbp);
619 * If we can't read the AGI of this ag,
620 * then just skip to the next one.
628 cur = xfs_inobt_init_cursor(mp, NULL, agbp, agno);
629 error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_GE,
632 xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
637 * Move up the last inode in the current
638 * chunk. The lookup_ge will always get
639 * us the first inode in the next chunk.
641 agino += XFS_INODES_PER_CHUNK - 1;
645 error = xfs_inobt_get_rec(cur, &r, &i);
646 if (error || i == 0) {
649 xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
655 agino = r.ir_startino + XFS_INODES_PER_CHUNK - 1;
656 buffer[bufidx].xi_startino =
657 XFS_AGINO_TO_INO(mp, agno, r.ir_startino);
658 buffer[bufidx].xi_alloccount =
659 XFS_INODES_PER_CHUNK - r.ir_freecount;
660 buffer[bufidx].xi_allocmask = ~r.ir_free;
663 if (bufidx == bcount) {
665 if (formatter(ubuffer, buffer, bufidx, &written)) {
666 error = XFS_ERROR(EFAULT);
674 error = xfs_btree_increment(cur, 0, &tmp);
676 xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
681 * The agino value has already been bumped.
682 * Just try to skip up to it.
684 agino += XFS_INODES_PER_CHUNK;
692 if (formatter(ubuffer, buffer, bufidx, &written))
693 error = XFS_ERROR(EFAULT);
697 *lastino = XFS_AGINO_TO_INO(mp, agno, agino);
701 xfs_btree_del_cursor(cur, (error ? XFS_BTREE_ERROR :