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"
24 #include "xfs_trans.h"
27 #include "xfs_mount.h"
28 #include "xfs_bmap_btree.h"
29 #include "xfs_alloc_btree.h"
30 #include "xfs_ialloc_btree.h"
31 #include "xfs_dinode.h"
32 #include "xfs_inode.h"
33 #include "xfs_ialloc.h"
34 #include "xfs_itable.h"
35 #include "xfs_error.h"
36 #include "xfs_btree.h"
37 #include "xfs_trace.h"
44 return (ino
== mp
->m_sb
.sb_rbmino
|| ino
== mp
->m_sb
.sb_rsumino
||
45 (xfs_sb_version_hasquota(&mp
->m_sb
) &&
46 (ino
== mp
->m_sb
.sb_uquotino
|| ino
== mp
->m_sb
.sb_gquotino
)));
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 */
224 xfs_buf_t
*bp
; /* ptr to on-disk inode cluster buf */
227 * Get the last inode value, see if there's nothing to do.
229 ino
= (xfs_ino_t
)*lastinop
;
231 agno
= XFS_INO_TO_AGNO(mp
, ino
);
232 agino
= XFS_INO_TO_AGINO(mp
, ino
);
233 if (agno
>= mp
->m_sb
.sb_agcount
||
234 ino
!= XFS_AGINO_TO_INO(mp
, agno
, agino
)) {
239 if (!ubcountp
|| *ubcountp
<= 0) {
242 ubcount
= *ubcountp
; /* statstruct's */
243 ubleft
= ubcount
* statstruct_size
; /* bytes */
244 *ubcountp
= ubelem
= 0;
248 nicluster
= mp
->m_sb
.sb_blocksize
>= XFS_INODE_CLUSTER_SIZE(mp
) ?
249 mp
->m_sb
.sb_inopblock
:
250 (XFS_INODE_CLUSTER_SIZE(mp
) >> mp
->m_sb
.sb_inodelog
);
251 nimask
= ~(nicluster
- 1);
252 nbcluster
= nicluster
>> mp
->m_sb
.sb_inopblog
;
253 irbuf
= kmem_zalloc_greedy(&irbsize
, PAGE_SIZE
, PAGE_SIZE
* 4);
257 nirbuf
= irbsize
/ sizeof(*irbuf
);
260 * Loop over the allocation groups, starting from the last
261 * inode returned; 0 means start of the allocation group.
264 while (XFS_BULKSTAT_UBLEFT(ubleft
) && agno
< mp
->m_sb
.sb_agcount
) {
267 error
= xfs_ialloc_read_agi(mp
, NULL
, agno
, &agbp
);
270 * Skip this allocation group and go to the next one.
276 agi
= XFS_BUF_TO_AGI(agbp
);
278 * Allocate and initialize a btree cursor for ialloc btree.
280 cur
= xfs_inobt_init_cursor(mp
, NULL
, agbp
, agno
);
282 irbufend
= irbuf
+ nirbuf
;
285 * If we're returning in the middle of an allocation group,
286 * we need to get the remainder of the chunk we're in.
289 xfs_inobt_rec_incore_t r
;
292 * Lookup the inode chunk that this inode lives in.
294 error
= xfs_inobt_lookup(cur
, agino
, XFS_LOOKUP_LE
,
296 if (!error
&& /* no I/O error */
297 tmp
&& /* lookup succeeded */
298 /* got the record, should always work */
299 !(error
= xfs_inobt_get_rec(cur
, &r
, &i
)) &&
301 /* this is the right chunk */
302 agino
< r
.ir_startino
+ XFS_INODES_PER_CHUNK
&&
303 /* lastino was not last in chunk */
304 (chunkidx
= agino
- r
.ir_startino
+ 1) <
305 XFS_INODES_PER_CHUNK
&&
306 /* there are some left allocated */
307 xfs_inobt_maskn(chunkidx
,
308 XFS_INODES_PER_CHUNK
- chunkidx
) &
311 * Grab the chunk record. Mark all the
312 * uninteresting inodes (because they're
313 * before our start point) free.
315 for (i
= 0; i
< chunkidx
; i
++) {
316 if (XFS_INOBT_MASK(i
) & ~r
.ir_free
)
319 r
.ir_free
|= xfs_inobt_maskn(0, chunkidx
);
320 irbp
->ir_startino
= r
.ir_startino
;
321 irbp
->ir_freecount
= r
.ir_freecount
;
322 irbp
->ir_free
= r
.ir_free
;
324 agino
= r
.ir_startino
+ XFS_INODES_PER_CHUNK
;
325 icount
= XFS_INODES_PER_CHUNK
- r
.ir_freecount
;
328 * If any of those tests failed, bump the
329 * inode number (just in case).
335 * In any case, increment to the next record.
338 error
= xfs_btree_increment(cur
, 0, &tmp
);
341 * Start of ag. Lookup the first inode chunk.
343 error
= xfs_inobt_lookup(cur
, 0, XFS_LOOKUP_GE
, &tmp
);
347 * Loop through inode btree records in this ag,
348 * until we run out of inodes or space in the buffer.
350 while (irbp
< irbufend
&& icount
< ubcount
) {
351 xfs_inobt_rec_incore_t r
;
354 * Loop as long as we're unable to read the
358 agino
+= XFS_INODES_PER_CHUNK
;
359 if (XFS_AGINO_TO_AGBNO(mp
, agino
) >=
360 be32_to_cpu(agi
->agi_length
))
362 error
= xfs_inobt_lookup(cur
, agino
,
363 XFS_LOOKUP_GE
, &tmp
);
367 * If ran off the end of the ag either with an error,
368 * or the normal way, set end and stop collecting.
375 error
= xfs_inobt_get_rec(cur
, &r
, &i
);
376 if (error
|| i
== 0) {
382 * If this chunk has any allocated inodes, save it.
383 * Also start read-ahead now for this chunk.
385 if (r
.ir_freecount
< XFS_INODES_PER_CHUNK
) {
387 * Loop over all clusters in the next chunk.
388 * Do a readahead if there are any allocated
389 * inodes in that cluster.
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
,
401 irbp
->ir_startino
= r
.ir_startino
;
402 irbp
->ir_freecount
= r
.ir_freecount
;
403 irbp
->ir_free
= r
.ir_free
;
405 icount
+= XFS_INODES_PER_CHUNK
- r
.ir_freecount
;
408 * Set agino to after this chunk and bump the cursor.
410 agino
= r
.ir_startino
+ XFS_INODES_PER_CHUNK
;
411 error
= xfs_btree_increment(cur
, 0, &tmp
);
415 * Drop the btree buffers and the agi buffer.
416 * We can't hold any of the locks these represent
419 xfs_btree_del_cursor(cur
, XFS_BTREE_NOERROR
);
422 * Now format all the good inodes into the user's buffer.
426 irbp
< irbufend
&& XFS_BULKSTAT_UBLEFT(ubleft
); irbp
++) {
428 * Now process this chunk of inodes.
430 for (agino
= irbp
->ir_startino
, chunkidx
= clustidx
= 0;
431 XFS_BULKSTAT_UBLEFT(ubleft
) &&
432 irbp
->ir_freecount
< XFS_INODES_PER_CHUNK
;
433 chunkidx
++, clustidx
++, agino
++) {
434 ASSERT(chunkidx
< XFS_INODES_PER_CHUNK
);
436 * Recompute agbno if this is the
437 * first inode of the cluster.
439 * Careful with clustidx. There can be
440 * multiple clusters per chunk, a single
441 * cluster per chunk or a cluster that has
442 * inodes represented from several different
443 * chunks (if blocksize is large).
445 * Because of this, the starting clustidx is
446 * initialized to zero in this loop but must
447 * later be reset after reading in the cluster
450 if ((chunkidx
& (nicluster
- 1)) == 0) {
451 agbno
= XFS_AGINO_TO_AGBNO(mp
,
453 ((chunkidx
& nimask
) >>
454 mp
->m_sb
.sb_inopblog
);
456 ino
= XFS_AGINO_TO_INO(mp
, agno
, agino
);
458 * Skip if this inode is free.
460 if (XFS_INOBT_MASK(chunkidx
) & irbp
->ir_free
) {
465 * Count used inodes as free so we can tell
466 * when the chunk is used up.
468 irbp
->ir_freecount
++;
471 * Get the inode and fill in a single buffer.
473 ubused
= statstruct_size
;
474 error
= formatter(mp
, ino
, ubufp
, ubleft
,
476 if (fmterror
== BULKSTAT_RV_NOTHING
) {
477 if (error
&& error
!= ENOENT
&&
486 if (fmterror
== BULKSTAT_RV_GIVEUP
) {
506 * Set up for the next loop iteration.
508 if (XFS_BULKSTAT_UBLEFT(ubleft
)) {
513 agino
= XFS_INO_TO_AGINO(mp
, lastino
);
518 * Done, we're either out of filesystem or space to put the data.
520 kmem_free_large(irbuf
);
523 * Found some inodes, return them now and return the error next time.
527 if (agno
>= mp
->m_sb
.sb_agcount
) {
529 * If we ran out of filesystem, mark lastino as off
530 * the end of the filesystem, so the next call
531 * will return immediately.
533 *lastinop
= (xfs_ino_t
)XFS_AGINO_TO_INO(mp
, agno
, 0);
536 *lastinop
= (xfs_ino_t
)lastino
;
542 * Return stat information in bulk (by-inode) for the filesystem.
543 * Special case for non-sequential one inode bulkstat.
545 int /* error status */
547 xfs_mount_t
*mp
, /* mount point for filesystem */
548 xfs_ino_t
*lastinop
, /* inode to return */
549 char __user
*buffer
, /* buffer with inode stats */
550 int *done
) /* 1 if there are more stats to get */
552 int count
; /* count value for bulkstat call */
553 int error
; /* return value */
554 xfs_ino_t ino
; /* filesystem inode number */
555 int res
; /* result from bs1 */
558 * note that requesting valid inode numbers which are not allocated
559 * to inodes will most likely cause xfs_itobp to generate warning
560 * messages about bad magic numbers. This is ok. The fact that
561 * the inode isn't actually an inode is handled by the
562 * error check below. Done this way to make the usual case faster
563 * at the expense of the error case.
566 ino
= (xfs_ino_t
)*lastinop
;
567 error
= xfs_bulkstat_one(mp
, ino
, buffer
, sizeof(xfs_bstat_t
), 0, &res
);
570 * Special case way failed, do it the "long" way
571 * to see if that works.
575 if (xfs_bulkstat(mp
, lastinop
, &count
, xfs_bulkstat_one
,
576 sizeof(xfs_bstat_t
), buffer
, done
))
578 if (count
== 0 || (xfs_ino_t
)*lastinop
!= ino
)
579 return error
== EFSCORRUPTED
?
580 XFS_ERROR(EINVAL
) : error
;
590 void __user
*ubuffer
, /* buffer to write to */
591 const xfs_inogrp_t
*buffer
, /* buffer to read from */
592 long count
, /* # of elements to read */
593 long *written
) /* # of bytes written */
595 if (copy_to_user(ubuffer
, buffer
, count
* sizeof(*buffer
)))
597 *written
= count
* sizeof(*buffer
);
602 * Return inode number table for the filesystem.
604 int /* error status */
606 xfs_mount_t
*mp
, /* mount point for filesystem */
607 xfs_ino_t
*lastino
, /* last inode returned */
608 int *count
, /* size of buffer/count returned */
609 void __user
*ubuffer
,/* buffer with inode descriptions */
610 inumbers_fmt_pf formatter
)
616 xfs_inogrp_t
*buffer
;
618 xfs_btree_cur_t
*cur
;
620 xfs_inobt_rec_incore_t r
;
626 ino
= (xfs_ino_t
)*lastino
;
627 agno
= XFS_INO_TO_AGNO(mp
, ino
);
628 agino
= XFS_INO_TO_AGINO(mp
, ino
);
631 bcount
= MIN(left
, (int)(PAGE_SIZE
/ sizeof(*buffer
)));
632 buffer
= kmem_alloc(bcount
* sizeof(*buffer
), KM_SLEEP
);
636 while (left
> 0 && agno
< mp
->m_sb
.sb_agcount
) {
638 error
= xfs_ialloc_read_agi(mp
, NULL
, agno
, &agbp
);
641 * If we can't read the AGI of this ag,
642 * then just skip to the next one.
650 cur
= xfs_inobt_init_cursor(mp
, NULL
, agbp
, agno
);
651 error
= xfs_inobt_lookup(cur
, agino
, XFS_LOOKUP_GE
,
654 xfs_btree_del_cursor(cur
, XFS_BTREE_ERROR
);
659 * Move up the last inode in the current
660 * chunk. The lookup_ge will always get
661 * us the first inode in the next chunk.
663 agino
+= XFS_INODES_PER_CHUNK
- 1;
667 error
= xfs_inobt_get_rec(cur
, &r
, &i
);
668 if (error
|| i
== 0) {
671 xfs_btree_del_cursor(cur
, XFS_BTREE_NOERROR
);
677 agino
= r
.ir_startino
+ XFS_INODES_PER_CHUNK
- 1;
678 buffer
[bufidx
].xi_startino
=
679 XFS_AGINO_TO_INO(mp
, agno
, r
.ir_startino
);
680 buffer
[bufidx
].xi_alloccount
=
681 XFS_INODES_PER_CHUNK
- r
.ir_freecount
;
682 buffer
[bufidx
].xi_allocmask
= ~r
.ir_free
;
685 if (bufidx
== bcount
) {
687 if (formatter(ubuffer
, buffer
, bufidx
, &written
)) {
688 error
= XFS_ERROR(EFAULT
);
696 error
= xfs_btree_increment(cur
, 0, &tmp
);
698 xfs_btree_del_cursor(cur
, XFS_BTREE_ERROR
);
703 * The agino value has already been bumped.
704 * Just try to skip up to it.
706 agino
+= XFS_INODES_PER_CHUNK
;
714 if (formatter(ubuffer
, buffer
, bufidx
, &written
))
715 error
= XFS_ERROR(EFAULT
);
719 *lastino
= XFS_AGINO_TO_INO(mp
, agno
, agino
);
723 xfs_btree_del_cursor(cur
, (error
? XFS_BTREE_ERROR
: