2005-04-16 22:20:36 +00:00
|
|
|
/*
|
2005-11-02 03:58:39 +00:00
|
|
|
* Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
|
|
|
|
* All Rights Reserved.
|
2005-04-16 22:20:36 +00:00
|
|
|
*
|
2005-11-02 03:58:39 +00:00
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public License as
|
2005-04-16 22:20:36 +00:00
|
|
|
* published by the Free Software Foundation.
|
|
|
|
*
|
2005-11-02 03:58:39 +00:00
|
|
|
* 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.
|
2005-04-16 22:20:36 +00:00
|
|
|
*
|
2005-11-02 03:58:39 +00:00
|
|
|
* 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
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
|
|
|
#include "xfs.h"
|
2005-11-02 03:38:42 +00:00
|
|
|
#include "xfs_fs.h"
|
2005-04-16 22:20:36 +00:00
|
|
|
#include "xfs_types.h"
|
2005-11-02 03:38:42 +00:00
|
|
|
#include "xfs_bit.h"
|
2005-04-16 22:20:36 +00:00
|
|
|
#include "xfs_log.h"
|
2005-11-02 03:38:42 +00:00
|
|
|
#include "xfs_inum.h"
|
2005-04-16 22:20:36 +00:00
|
|
|
#include "xfs_trans.h"
|
|
|
|
#include "xfs_sb.h"
|
|
|
|
#include "xfs_ag.h"
|
|
|
|
#include "xfs_dir2.h"
|
|
|
|
#include "xfs_dmapi.h"
|
|
|
|
#include "xfs_mount.h"
|
|
|
|
#include "xfs_bmap_btree.h"
|
2005-11-02 03:38:42 +00:00
|
|
|
#include "xfs_alloc_btree.h"
|
2005-04-16 22:20:36 +00:00
|
|
|
#include "xfs_ialloc_btree.h"
|
|
|
|
#include "xfs_dir2_sf.h"
|
2005-11-02 03:38:42 +00:00
|
|
|
#include "xfs_attr_sf.h"
|
2005-04-16 22:20:36 +00:00
|
|
|
#include "xfs_dinode.h"
|
|
|
|
#include "xfs_inode.h"
|
2005-11-02 03:38:42 +00:00
|
|
|
#include "xfs_btree.h"
|
|
|
|
#include "xfs_ialloc.h"
|
2005-04-16 22:20:36 +00:00
|
|
|
#include "xfs_alloc.h"
|
|
|
|
#include "xfs_rtalloc.h"
|
|
|
|
#include "xfs_error.h"
|
|
|
|
#include "xfs_bmap.h"
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Allocation group level functions.
|
|
|
|
*/
|
2008-03-27 07:00:38 +00:00
|
|
|
static inline int
|
|
|
|
xfs_ialloc_cluster_alignment(
|
|
|
|
xfs_alloc_arg_t *args)
|
|
|
|
{
|
|
|
|
if (xfs_sb_version_hasalign(&args->mp->m_sb) &&
|
|
|
|
args->mp->m_sb.sb_inoalignmt >=
|
|
|
|
XFS_B_TO_FSBT(args->mp, XFS_INODE_CLUSTER_SIZE(args->mp)))
|
|
|
|
return args->mp->m_sb.sb_inoalignmt;
|
|
|
|
return 1;
|
|
|
|
}
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2008-10-30 05:56:09 +00:00
|
|
|
/*
|
2009-08-31 23:58:21 +00:00
|
|
|
* Lookup a record by ino in the btree given by cur.
|
2008-10-30 05:56:09 +00:00
|
|
|
*/
|
2009-09-01 23:56:55 +00:00
|
|
|
int /* error */
|
2009-08-31 23:58:21 +00:00
|
|
|
xfs_inobt_lookup(
|
2008-10-30 05:56:09 +00:00
|
|
|
struct xfs_btree_cur *cur, /* btree cursor */
|
|
|
|
xfs_agino_t ino, /* starting inode of chunk */
|
2009-08-31 23:58:21 +00:00
|
|
|
xfs_lookup_t dir, /* <=, >=, == */
|
2008-10-30 05:56:09 +00:00
|
|
|
int *stat) /* success/failure */
|
|
|
|
{
|
|
|
|
cur->bc_rec.i.ir_startino = ino;
|
2009-08-31 23:58:21 +00:00
|
|
|
cur->bc_rec.i.ir_freecount = 0;
|
|
|
|
cur->bc_rec.i.ir_free = 0;
|
|
|
|
return xfs_btree_lookup(cur, dir, stat);
|
2008-10-30 05:56:09 +00:00
|
|
|
}
|
|
|
|
|
2008-10-30 05:56:32 +00:00
|
|
|
/*
|
2009-08-31 23:57:03 +00:00
|
|
|
* Update the record referred to by cur to the value given.
|
2008-10-30 05:56:32 +00:00
|
|
|
* This either works (return 0) or gets an EFSCORRUPTED error.
|
|
|
|
*/
|
|
|
|
STATIC int /* error */
|
|
|
|
xfs_inobt_update(
|
|
|
|
struct xfs_btree_cur *cur, /* btree cursor */
|
2009-08-31 23:57:03 +00:00
|
|
|
xfs_inobt_rec_incore_t *irec) /* btree record */
|
2008-10-30 05:56:32 +00:00
|
|
|
{
|
|
|
|
union xfs_btree_rec rec;
|
|
|
|
|
2009-08-31 23:57:03 +00:00
|
|
|
rec.inobt.ir_startino = cpu_to_be32(irec->ir_startino);
|
|
|
|
rec.inobt.ir_freecount = cpu_to_be32(irec->ir_freecount);
|
|
|
|
rec.inobt.ir_free = cpu_to_be64(irec->ir_free);
|
2008-10-30 05:56:32 +00:00
|
|
|
return xfs_btree_update(cur, &rec);
|
|
|
|
}
|
|
|
|
|
2008-10-30 05:58:11 +00:00
|
|
|
/*
|
|
|
|
* Get the data from the pointed-to record.
|
|
|
|
*/
|
|
|
|
int /* error */
|
|
|
|
xfs_inobt_get_rec(
|
|
|
|
struct xfs_btree_cur *cur, /* btree cursor */
|
2009-08-31 23:56:58 +00:00
|
|
|
xfs_inobt_rec_incore_t *irec, /* btree record */
|
2008-10-30 05:58:11 +00:00
|
|
|
int *stat) /* output: success/failure */
|
|
|
|
{
|
|
|
|
union xfs_btree_rec *rec;
|
|
|
|
int error;
|
|
|
|
|
|
|
|
error = xfs_btree_get_rec(cur, &rec, stat);
|
|
|
|
if (!error && *stat == 1) {
|
2009-08-31 23:56:58 +00:00
|
|
|
irec->ir_startino = be32_to_cpu(rec->inobt.ir_startino);
|
|
|
|
irec->ir_freecount = be32_to_cpu(rec->inobt.ir_freecount);
|
|
|
|
irec->ir_free = be64_to_cpu(rec->inobt.ir_free);
|
2008-10-30 05:58:11 +00:00
|
|
|
}
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
2009-08-31 23:57:09 +00:00
|
|
|
/*
|
|
|
|
* Verify that the number of free inodes in the AGI is correct.
|
|
|
|
*/
|
|
|
|
#ifdef DEBUG
|
|
|
|
STATIC int
|
|
|
|
xfs_check_agi_freecount(
|
|
|
|
struct xfs_btree_cur *cur,
|
|
|
|
struct xfs_agi *agi)
|
|
|
|
{
|
|
|
|
if (cur->bc_nlevels == 1) {
|
|
|
|
xfs_inobt_rec_incore_t rec;
|
|
|
|
int freecount = 0;
|
|
|
|
int error;
|
|
|
|
int i;
|
|
|
|
|
2009-08-31 23:58:21 +00:00
|
|
|
error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &i);
|
2009-08-31 23:57:09 +00:00
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
|
|
|
|
do {
|
|
|
|
error = xfs_inobt_get_rec(cur, &rec, &i);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
|
|
|
|
if (i) {
|
|
|
|
freecount += rec.ir_freecount;
|
|
|
|
error = xfs_btree_increment(cur, 0, &i);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
} while (i == 1);
|
|
|
|
|
|
|
|
if (!XFS_FORCED_SHUTDOWN(cur->bc_mp))
|
|
|
|
ASSERT(freecount == be32_to_cpu(agi->agi_freecount));
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
#define xfs_check_agi_freecount(cur, agi) 0
|
|
|
|
#endif
|
|
|
|
|
2009-08-31 23:56:51 +00:00
|
|
|
/*
|
|
|
|
* Initialise a new set of inodes.
|
|
|
|
*/
|
|
|
|
STATIC void
|
|
|
|
xfs_ialloc_inode_init(
|
|
|
|
struct xfs_mount *mp,
|
|
|
|
struct xfs_trans *tp,
|
|
|
|
xfs_agnumber_t agno,
|
|
|
|
xfs_agblock_t agbno,
|
|
|
|
xfs_agblock_t length,
|
|
|
|
unsigned int gen)
|
|
|
|
{
|
|
|
|
struct xfs_buf *fbuf;
|
|
|
|
struct xfs_dinode *free;
|
|
|
|
int blks_per_cluster, nbufs, ninodes;
|
|
|
|
int version;
|
|
|
|
int i, j;
|
|
|
|
xfs_daddr_t d;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Loop over the new block(s), filling in the inodes.
|
|
|
|
* For small block sizes, manipulate the inodes in buffers
|
|
|
|
* which are multiples of the blocks size.
|
|
|
|
*/
|
|
|
|
if (mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(mp)) {
|
|
|
|
blks_per_cluster = 1;
|
|
|
|
nbufs = length;
|
|
|
|
ninodes = mp->m_sb.sb_inopblock;
|
|
|
|
} else {
|
|
|
|
blks_per_cluster = XFS_INODE_CLUSTER_SIZE(mp) /
|
|
|
|
mp->m_sb.sb_blocksize;
|
|
|
|
nbufs = length / blks_per_cluster;
|
|
|
|
ninodes = blks_per_cluster * mp->m_sb.sb_inopblock;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Figure out what version number to use in the inodes we create.
|
|
|
|
* If the superblock version has caught up to the one that supports
|
|
|
|
* the new inode format, then use the new inode version. Otherwise
|
|
|
|
* use the old version so that old kernels will continue to be
|
|
|
|
* able to use the file system.
|
|
|
|
*/
|
|
|
|
if (xfs_sb_version_hasnlink(&mp->m_sb))
|
|
|
|
version = 2;
|
|
|
|
else
|
|
|
|
version = 1;
|
|
|
|
|
|
|
|
for (j = 0; j < nbufs; j++) {
|
|
|
|
/*
|
|
|
|
* Get the block.
|
|
|
|
*/
|
|
|
|
d = XFS_AGB_TO_DADDR(mp, agno, agbno + (j * blks_per_cluster));
|
|
|
|
fbuf = xfs_trans_get_buf(tp, mp->m_ddev_targp, d,
|
|
|
|
mp->m_bsize * blks_per_cluster,
|
|
|
|
XFS_BUF_LOCK);
|
|
|
|
ASSERT(fbuf);
|
|
|
|
ASSERT(!XFS_BUF_GETERROR(fbuf));
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Initialize all inodes in this buffer and then log them.
|
|
|
|
*
|
|
|
|
* XXX: It would be much better if we had just one transaction
|
|
|
|
* to log a whole cluster of inodes instead of all the
|
|
|
|
* individual transactions causing a lot of log traffic.
|
|
|
|
*/
|
|
|
|
xfs_biozero(fbuf, 0, ninodes << mp->m_sb.sb_inodelog);
|
|
|
|
for (i = 0; i < ninodes; i++) {
|
|
|
|
int ioffset = i << mp->m_sb.sb_inodelog;
|
|
|
|
uint isize = sizeof(struct xfs_dinode);
|
|
|
|
|
|
|
|
free = xfs_make_iptr(mp, fbuf, i);
|
|
|
|
free->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
|
|
|
|
free->di_version = version;
|
|
|
|
free->di_gen = cpu_to_be32(gen);
|
|
|
|
free->di_next_unlinked = cpu_to_be32(NULLAGINO);
|
|
|
|
xfs_trans_log_buf(tp, fbuf, ioffset, ioffset + isize - 1);
|
|
|
|
}
|
|
|
|
xfs_trans_inode_alloc_buf(tp, fbuf);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Allocate new inodes in the allocation group specified by agbp.
|
|
|
|
* Return 0 for success, else error code.
|
|
|
|
*/
|
|
|
|
STATIC int /* error code or 0 */
|
|
|
|
xfs_ialloc_ag_alloc(
|
|
|
|
xfs_trans_t *tp, /* transaction pointer */
|
|
|
|
xfs_buf_t *agbp, /* alloc group buffer */
|
|
|
|
int *alloc)
|
|
|
|
{
|
|
|
|
xfs_agi_t *agi; /* allocation group header */
|
|
|
|
xfs_alloc_arg_t args; /* allocation argument structure */
|
|
|
|
xfs_btree_cur_t *cur; /* inode btree cursor */
|
[XFS] Lazy Superblock Counters
When we have a couple of hundred transactions on the fly at once, they all
typically modify the on disk superblock in some way.
create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify
free block counts.
When these counts are modified in a transaction, they must eventually lock
the superblock buffer and apply the mods. The buffer then remains locked
until the transaction is committed into the incore log buffer. The result
of this is that with enough transactions on the fly the incore superblock
buffer becomes a bottleneck.
The result of contention on the incore superblock buffer is that
transaction rates fall - the more pressure that is put on the superblock
buffer, the slower things go.
The key to removing the contention is to not require the superblock fields
in question to be locked. We do that by not marking the superblock dirty
in the transaction. IOWs, we modify the incore superblock but do not
modify the cached superblock buffer. In short, we do not log superblock
modifications to critical fields in the superblock on every transaction.
In fact we only do it just before we write the superblock to disk every
sync period or just before unmount.
This creates an interesting problem - if we don't log or write out the
fields in every transaction, then how do the values get recovered after a
crash? the answer is simple - we keep enough duplicate, logged information
in other structures that we can reconstruct the correct count after log
recovery has been performed.
It is the AGF and AGI structures that contain the duplicate information;
after recovery, we walk every AGI and AGF and sum their individual
counters to get the correct value, and we do a transaction into the log to
correct them. An optimisation of this is that if we have a clean unmount
record, we know the value in the superblock is correct, so we can avoid
the summation walk under normal conditions and so mount/recovery times do
not change under normal operation.
One wrinkle that was discovered during development was that the blocks
used in the freespace btrees are never accounted for in the AGF counters.
This was once a valid optimisation to make; when the filesystem is full,
the free space btrees are empty and consume no space. Hence when it
matters, the "accounting" is correct. But that means the when we do the
AGF summations, we would not have a correct count and xfs_check would
complain. Hence a new counter was added to track the number of blocks used
by the free space btrees. This is an *on-disk format change*.
As a result of this, lazy superblock counters are a mkfs option and at the
moment on linux there is no way to convert an old filesystem. This is
possible - xfs_db can be used to twiddle the right bits and then
xfs_repair will do the format conversion for you. Similarly, you can
convert backwards as well. At some point we'll add functionality to
xfs_admin to do the bit twiddling easily....
SGI-PV: 964999
SGI-Modid: xfs-linux-melb:xfs-kern:28652a
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 05:26:31 +00:00
|
|
|
xfs_agnumber_t agno;
|
2005-04-16 22:20:36 +00:00
|
|
|
int error;
|
2009-08-31 23:56:51 +00:00
|
|
|
int i;
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_agino_t newino; /* new first inode's number */
|
|
|
|
xfs_agino_t newlen; /* new number of inodes */
|
|
|
|
xfs_agino_t thisino; /* current inode number, for loop */
|
2006-03-28 23:52:28 +00:00
|
|
|
int isaligned = 0; /* inode allocation at stripe unit */
|
2005-04-16 22:20:36 +00:00
|
|
|
/* boundary */
|
2010-01-11 11:47:43 +00:00
|
|
|
struct xfs_perag *pag;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
args.tp = tp;
|
|
|
|
args.mp = tp->t_mountp;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Locking will ensure that we don't have two callers in here
|
|
|
|
* at one time.
|
|
|
|
*/
|
|
|
|
newlen = XFS_IALLOC_INODES(args.mp);
|
|
|
|
if (args.mp->m_maxicount &&
|
|
|
|
args.mp->m_sb.sb_icount + newlen > args.mp->m_maxicount)
|
|
|
|
return XFS_ERROR(ENOSPC);
|
|
|
|
args.minlen = args.maxlen = XFS_IALLOC_BLOCKS(args.mp);
|
|
|
|
/*
|
2006-03-28 23:52:28 +00:00
|
|
|
* First try to allocate inodes contiguous with the last-allocated
|
|
|
|
* chunk of inodes. If the filesystem is striped, this will fill
|
|
|
|
* an entire stripe unit with inodes.
|
|
|
|
*/
|
2005-04-16 22:20:36 +00:00
|
|
|
agi = XFS_BUF_TO_AGI(agbp);
|
2006-03-28 23:52:28 +00:00
|
|
|
newino = be32_to_cpu(agi->agi_newino);
|
2009-08-31 23:56:51 +00:00
|
|
|
agno = be32_to_cpu(agi->agi_seqno);
|
2006-04-11 05:45:05 +00:00
|
|
|
args.agbno = XFS_AGINO_TO_AGBNO(args.mp, newino) +
|
|
|
|
XFS_IALLOC_BLOCKS(args.mp);
|
|
|
|
if (likely(newino != NULLAGINO &&
|
|
|
|
(args.agbno < be32_to_cpu(agi->agi_length)))) {
|
2009-08-31 23:56:51 +00:00
|
|
|
args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
|
2006-03-28 23:52:28 +00:00
|
|
|
args.type = XFS_ALLOCTYPE_THIS_BNO;
|
|
|
|
args.mod = args.total = args.wasdel = args.isfl =
|
|
|
|
args.userdata = args.minalignslop = 0;
|
|
|
|
args.prod = 1;
|
2008-03-27 07:00:38 +00:00
|
|
|
|
2006-03-28 23:52:28 +00:00
|
|
|
/*
|
2008-03-27 07:00:38 +00:00
|
|
|
* We need to take into account alignment here to ensure that
|
|
|
|
* we don't modify the free list if we fail to have an exact
|
|
|
|
* block. If we don't have an exact match, and every oher
|
|
|
|
* attempt allocation attempt fails, we'll end up cancelling
|
|
|
|
* a dirty transaction and shutting down.
|
|
|
|
*
|
|
|
|
* For an exact allocation, alignment must be 1,
|
|
|
|
* however we need to take cluster alignment into account when
|
|
|
|
* fixing up the freelist. Use the minalignslop field to
|
|
|
|
* indicate that extra blocks might be required for alignment,
|
|
|
|
* but not to use them in the actual exact allocation.
|
2006-03-28 23:52:28 +00:00
|
|
|
*/
|
2008-03-27 07:00:38 +00:00
|
|
|
args.alignment = 1;
|
|
|
|
args.minalignslop = xfs_ialloc_cluster_alignment(&args) - 1;
|
|
|
|
|
|
|
|
/* Allow space for the inode btree to split. */
|
2009-02-09 07:37:14 +00:00
|
|
|
args.minleft = args.mp->m_in_maxlevels - 1;
|
2006-03-28 23:52:28 +00:00
|
|
|
if ((error = xfs_alloc_vextent(&args)))
|
|
|
|
return error;
|
|
|
|
} else
|
|
|
|
args.fsbno = NULLFSBLOCK;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2006-03-28 23:52:28 +00:00
|
|
|
if (unlikely(args.fsbno == NULLFSBLOCK)) {
|
|
|
|
/*
|
|
|
|
* Set the alignment for the allocation.
|
|
|
|
* If stripe alignment is turned on then align at stripe unit
|
|
|
|
* boundary.
|
2006-04-11 05:45:05 +00:00
|
|
|
* If the cluster size is smaller than a filesystem block
|
|
|
|
* then we're doing I/O for inodes in filesystem block size
|
2006-03-28 23:52:28 +00:00
|
|
|
* pieces, so don't need alignment anyway.
|
|
|
|
*/
|
|
|
|
isaligned = 0;
|
|
|
|
if (args.mp->m_sinoalign) {
|
|
|
|
ASSERT(!(args.mp->m_flags & XFS_MOUNT_NOALIGN));
|
|
|
|
args.alignment = args.mp->m_dalign;
|
|
|
|
isaligned = 1;
|
2008-03-27 07:00:38 +00:00
|
|
|
} else
|
|
|
|
args.alignment = xfs_ialloc_cluster_alignment(&args);
|
2006-03-28 23:52:28 +00:00
|
|
|
/*
|
|
|
|
* Need to figure out where to allocate the inode blocks.
|
|
|
|
* Ideally they should be spaced out through the a.g.
|
|
|
|
* For now, just allocate blocks up front.
|
|
|
|
*/
|
|
|
|
args.agbno = be32_to_cpu(agi->agi_root);
|
2009-08-31 23:56:51 +00:00
|
|
|
args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
|
2006-03-28 23:52:28 +00:00
|
|
|
/*
|
|
|
|
* Allocate a fixed-size extent of inodes.
|
|
|
|
*/
|
|
|
|
args.type = XFS_ALLOCTYPE_NEAR_BNO;
|
|
|
|
args.mod = args.total = args.wasdel = args.isfl =
|
|
|
|
args.userdata = args.minalignslop = 0;
|
|
|
|
args.prod = 1;
|
|
|
|
/*
|
|
|
|
* Allow space for the inode btree to split.
|
|
|
|
*/
|
2009-02-09 07:37:14 +00:00
|
|
|
args.minleft = args.mp->m_in_maxlevels - 1;
|
2006-03-28 23:52:28 +00:00
|
|
|
if ((error = xfs_alloc_vextent(&args)))
|
|
|
|
return error;
|
|
|
|
}
|
2006-04-11 05:45:05 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* If stripe alignment is turned on, then try again with cluster
|
|
|
|
* alignment.
|
|
|
|
*/
|
|
|
|
if (isaligned && args.fsbno == NULLFSBLOCK) {
|
|
|
|
args.type = XFS_ALLOCTYPE_NEAR_BNO;
|
2005-11-02 04:11:25 +00:00
|
|
|
args.agbno = be32_to_cpu(agi->agi_root);
|
2009-08-31 23:56:51 +00:00
|
|
|
args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
|
2008-03-27 07:00:38 +00:00
|
|
|
args.alignment = xfs_ialloc_cluster_alignment(&args);
|
2005-04-16 22:20:36 +00:00
|
|
|
if ((error = xfs_alloc_vextent(&args)))
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (args.fsbno == NULLFSBLOCK) {
|
|
|
|
*alloc = 0;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
ASSERT(args.len == args.minlen);
|
|
|
|
|
2008-04-29 02:53:32 +00:00
|
|
|
/*
|
2009-08-31 23:56:51 +00:00
|
|
|
* Stamp and write the inode buffers.
|
|
|
|
*
|
2008-04-29 02:53:32 +00:00
|
|
|
* Seed the new inode cluster with a random generation number. This
|
|
|
|
* prevents short-term reuse of generation numbers if a chunk is
|
|
|
|
* freed and then immediately reallocated. We use random numbers
|
|
|
|
* rather than a linear progression to prevent the next generation
|
|
|
|
* number from being easily guessable.
|
|
|
|
*/
|
2009-08-31 23:56:51 +00:00
|
|
|
xfs_ialloc_inode_init(args.mp, tp, agno, args.agbno, args.len,
|
|
|
|
random32());
|
2008-11-28 03:23:38 +00:00
|
|
|
|
2009-08-31 23:56:51 +00:00
|
|
|
/*
|
|
|
|
* Convert the results.
|
|
|
|
*/
|
|
|
|
newino = XFS_OFFBNO_TO_AGINO(args.mp, args.agbno, 0);
|
2008-02-13 23:03:29 +00:00
|
|
|
be32_add_cpu(&agi->agi_count, newlen);
|
|
|
|
be32_add_cpu(&agi->agi_freecount, newlen);
|
2005-04-16 22:20:36 +00:00
|
|
|
down_read(&args.mp->m_peraglock);
|
2010-01-11 11:47:43 +00:00
|
|
|
pag = xfs_perag_get(args.mp, agno);
|
|
|
|
pag->pagi_freecount += newlen;
|
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
up_read(&args.mp->m_peraglock);
|
2005-11-02 04:11:25 +00:00
|
|
|
agi->agi_newino = cpu_to_be32(newino);
|
2009-08-31 23:56:51 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Insert records describing the new inode chunk into the btree.
|
|
|
|
*/
|
2008-10-30 05:53:59 +00:00
|
|
|
cur = xfs_inobt_init_cursor(args.mp, tp, agbp, agno);
|
2005-04-16 22:20:36 +00:00
|
|
|
for (thisino = newino;
|
|
|
|
thisino < newino + newlen;
|
|
|
|
thisino += XFS_INODES_PER_CHUNK) {
|
2009-08-31 23:58:21 +00:00
|
|
|
cur->bc_rec.i.ir_startino = thisino;
|
|
|
|
cur->bc_rec.i.ir_freecount = XFS_INODES_PER_CHUNK;
|
|
|
|
cur->bc_rec.i.ir_free = XFS_INOBT_ALL_FREE;
|
|
|
|
error = xfs_btree_lookup(cur, XFS_LOOKUP_EQ, &i);
|
|
|
|
if (error) {
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
ASSERT(i == 0);
|
2009-08-31 23:58:21 +00:00
|
|
|
error = xfs_btree_insert(cur, &i);
|
|
|
|
if (error) {
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
ASSERT(i == 1);
|
|
|
|
}
|
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
|
|
|
|
/*
|
|
|
|
* Log allocation group header fields
|
|
|
|
*/
|
|
|
|
xfs_ialloc_log_agi(tp, agbp,
|
|
|
|
XFS_AGI_COUNT | XFS_AGI_FREECOUNT | XFS_AGI_NEWINO);
|
|
|
|
/*
|
|
|
|
* Modify/log superblock values for inode count and inode free count.
|
|
|
|
*/
|
|
|
|
xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, (long)newlen);
|
|
|
|
xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, (long)newlen);
|
|
|
|
*alloc = 1;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2009-11-14 16:17:22 +00:00
|
|
|
STATIC xfs_agnumber_t
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_ialloc_next_ag(
|
|
|
|
xfs_mount_t *mp)
|
|
|
|
{
|
|
|
|
xfs_agnumber_t agno;
|
|
|
|
|
|
|
|
spin_lock(&mp->m_agirotor_lock);
|
|
|
|
agno = mp->m_agirotor;
|
|
|
|
if (++mp->m_agirotor == mp->m_maxagi)
|
|
|
|
mp->m_agirotor = 0;
|
|
|
|
spin_unlock(&mp->m_agirotor_lock);
|
|
|
|
|
|
|
|
return agno;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Select an allocation group to look for a free inode in, based on the parent
|
|
|
|
* inode and then mode. Return the allocation group buffer.
|
|
|
|
*/
|
|
|
|
STATIC xfs_buf_t * /* allocation group buffer */
|
|
|
|
xfs_ialloc_ag_select(
|
|
|
|
xfs_trans_t *tp, /* transaction pointer */
|
|
|
|
xfs_ino_t parent, /* parent directory inode number */
|
|
|
|
mode_t mode, /* bits set to indicate file type */
|
|
|
|
int okalloc) /* ok to allocate more space */
|
|
|
|
{
|
|
|
|
xfs_buf_t *agbp; /* allocation group header buffer */
|
|
|
|
xfs_agnumber_t agcount; /* number of ag's in the filesystem */
|
|
|
|
xfs_agnumber_t agno; /* current ag number */
|
|
|
|
int flags; /* alloc buffer locking flags */
|
|
|
|
xfs_extlen_t ineed; /* blocks needed for inode allocation */
|
|
|
|
xfs_extlen_t longest = 0; /* longest extent available */
|
|
|
|
xfs_mount_t *mp; /* mount point structure */
|
|
|
|
int needspace; /* file mode implies space allocated */
|
|
|
|
xfs_perag_t *pag; /* per allocation group data */
|
|
|
|
xfs_agnumber_t pagno; /* parent (starting) ag number */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Files of these types need at least one block if length > 0
|
|
|
|
* (and they won't fit in the inode, but that's hard to figure out).
|
|
|
|
*/
|
|
|
|
needspace = S_ISDIR(mode) || S_ISREG(mode) || S_ISLNK(mode);
|
|
|
|
mp = tp->t_mountp;
|
|
|
|
agcount = mp->m_maxagi;
|
|
|
|
if (S_ISDIR(mode))
|
|
|
|
pagno = xfs_ialloc_next_ag(mp);
|
|
|
|
else {
|
|
|
|
pagno = XFS_INO_TO_AGNO(mp, parent);
|
|
|
|
if (pagno >= agcount)
|
|
|
|
pagno = 0;
|
|
|
|
}
|
|
|
|
ASSERT(pagno < agcount);
|
|
|
|
/*
|
|
|
|
* Loop through allocation groups, looking for one with a little
|
|
|
|
* free space in it. Note we don't look for free inodes, exactly.
|
|
|
|
* Instead, we include whether there is a need to allocate inodes
|
|
|
|
* to mean that blocks must be allocated for them,
|
|
|
|
* if none are currently free.
|
|
|
|
*/
|
|
|
|
agno = pagno;
|
|
|
|
flags = XFS_ALLOC_FLAG_TRYLOCK;
|
|
|
|
down_read(&mp->m_peraglock);
|
|
|
|
for (;;) {
|
2010-01-11 11:47:43 +00:00
|
|
|
pag = xfs_perag_get(mp, agno);
|
2005-04-16 22:20:36 +00:00
|
|
|
if (!pag->pagi_init) {
|
|
|
|
if (xfs_ialloc_read_agi(mp, tp, agno, &agbp)) {
|
|
|
|
agbp = NULL;
|
|
|
|
goto nextag;
|
|
|
|
}
|
|
|
|
} else
|
|
|
|
agbp = NULL;
|
|
|
|
|
|
|
|
if (!pag->pagi_inodeok) {
|
|
|
|
xfs_ialloc_next_ag(mp);
|
|
|
|
goto unlock_nextag;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Is there enough free space for the file plus a block
|
|
|
|
* of inodes (if we need to allocate some)?
|
|
|
|
*/
|
|
|
|
ineed = pag->pagi_freecount ? 0 : XFS_IALLOC_BLOCKS(mp);
|
|
|
|
if (ineed && !pag->pagf_init) {
|
|
|
|
if (agbp == NULL &&
|
|
|
|
xfs_ialloc_read_agi(mp, tp, agno, &agbp)) {
|
|
|
|
agbp = NULL;
|
|
|
|
goto nextag;
|
|
|
|
}
|
|
|
|
(void)xfs_alloc_pagf_init(mp, tp, agno, flags);
|
|
|
|
}
|
|
|
|
if (!ineed || pag->pagf_init) {
|
|
|
|
if (ineed && !(longest = pag->pagf_longest))
|
|
|
|
longest = pag->pagf_flcount > 0;
|
|
|
|
if (!ineed ||
|
|
|
|
(pag->pagf_freeblks >= needspace + ineed &&
|
|
|
|
longest >= ineed &&
|
|
|
|
okalloc)) {
|
|
|
|
if (agbp == NULL &&
|
|
|
|
xfs_ialloc_read_agi(mp, tp, agno, &agbp)) {
|
|
|
|
agbp = NULL;
|
|
|
|
goto nextag;
|
|
|
|
}
|
2010-01-11 11:47:43 +00:00
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
up_read(&mp->m_peraglock);
|
|
|
|
return agbp;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
unlock_nextag:
|
|
|
|
if (agbp)
|
|
|
|
xfs_trans_brelse(tp, agbp);
|
|
|
|
nextag:
|
2010-01-11 11:47:43 +00:00
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* No point in iterating over the rest, if we're shutting
|
|
|
|
* down.
|
|
|
|
*/
|
|
|
|
if (XFS_FORCED_SHUTDOWN(mp)) {
|
|
|
|
up_read(&mp->m_peraglock);
|
2006-09-28 00:58:40 +00:00
|
|
|
return NULL;
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
|
|
|
agno++;
|
|
|
|
if (agno >= agcount)
|
|
|
|
agno = 0;
|
|
|
|
if (agno == pagno) {
|
|
|
|
if (flags == 0) {
|
|
|
|
up_read(&mp->m_peraglock);
|
2006-09-28 00:58:40 +00:00
|
|
|
return NULL;
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
|
|
|
flags = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-08-31 23:57:14 +00:00
|
|
|
/*
|
|
|
|
* Try to retrieve the next record to the left/right from the current one.
|
|
|
|
*/
|
|
|
|
STATIC int
|
|
|
|
xfs_ialloc_next_rec(
|
|
|
|
struct xfs_btree_cur *cur,
|
|
|
|
xfs_inobt_rec_incore_t *rec,
|
|
|
|
int *done,
|
|
|
|
int left)
|
|
|
|
{
|
|
|
|
int error;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
if (left)
|
|
|
|
error = xfs_btree_decrement(cur, 0, &i);
|
|
|
|
else
|
|
|
|
error = xfs_btree_increment(cur, 0, &i);
|
|
|
|
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
*done = !i;
|
|
|
|
if (i) {
|
|
|
|
error = xfs_inobt_get_rec(cur, rec, &i);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
XFS_WANT_CORRUPTED_RETURN(i == 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2009-08-31 23:58:28 +00:00
|
|
|
STATIC int
|
|
|
|
xfs_ialloc_get_rec(
|
|
|
|
struct xfs_btree_cur *cur,
|
|
|
|
xfs_agino_t agino,
|
|
|
|
xfs_inobt_rec_incore_t *rec,
|
|
|
|
int *done,
|
|
|
|
int left)
|
|
|
|
{
|
|
|
|
int error;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_EQ, &i);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
*done = !i;
|
|
|
|
if (i) {
|
|
|
|
error = xfs_inobt_get_rec(cur, rec, &i);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
XFS_WANT_CORRUPTED_RETURN(i == 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
2009-08-31 23:57:09 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Visible inode allocation functions.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Allocate an inode on disk.
|
|
|
|
* Mode is used to tell whether the new inode will need space, and whether
|
|
|
|
* it is a directory.
|
|
|
|
*
|
|
|
|
* The arguments IO_agbp and alloc_done are defined to work within
|
|
|
|
* the constraint of one allocation per transaction.
|
|
|
|
* xfs_dialloc() is designed to be called twice if it has to do an
|
|
|
|
* allocation to make more free inodes. On the first call,
|
|
|
|
* IO_agbp should be set to NULL. If an inode is available,
|
|
|
|
* i.e., xfs_dialloc() did not need to do an allocation, an inode
|
|
|
|
* number is returned. In this case, IO_agbp would be set to the
|
|
|
|
* current ag_buf and alloc_done set to false.
|
|
|
|
* If an allocation needed to be done, xfs_dialloc would return
|
|
|
|
* the current ag_buf in IO_agbp and set alloc_done to true.
|
|
|
|
* The caller should then commit the current transaction, allocate a new
|
|
|
|
* transaction, and call xfs_dialloc() again, passing in the previous
|
|
|
|
* value of IO_agbp. IO_agbp should be held across the transactions.
|
|
|
|
* Since the agbp is locked across the two calls, the second call is
|
|
|
|
* guaranteed to have a free inode available.
|
|
|
|
*
|
|
|
|
* Once we successfully pick an inode its number is returned and the
|
|
|
|
* on-disk data structures are updated. The inode itself is not read
|
|
|
|
* in, since doing so would break ordering constraints with xfs_reclaim.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
xfs_dialloc(
|
|
|
|
xfs_trans_t *tp, /* transaction pointer */
|
|
|
|
xfs_ino_t parent, /* parent inode (directory) */
|
|
|
|
mode_t mode, /* mode bits for new inode */
|
|
|
|
int okalloc, /* ok to allocate more space */
|
|
|
|
xfs_buf_t **IO_agbp, /* in/out ag header's buffer */
|
|
|
|
boolean_t *alloc_done, /* true if we needed to replenish
|
|
|
|
inode freelist */
|
|
|
|
xfs_ino_t *inop) /* inode number allocated */
|
|
|
|
{
|
|
|
|
xfs_agnumber_t agcount; /* number of allocation groups */
|
|
|
|
xfs_buf_t *agbp; /* allocation group header's buffer */
|
|
|
|
xfs_agnumber_t agno; /* allocation group number */
|
|
|
|
xfs_agi_t *agi; /* allocation group header structure */
|
|
|
|
xfs_btree_cur_t *cur; /* inode allocation btree cursor */
|
|
|
|
int error; /* error return value */
|
|
|
|
int i; /* result code */
|
|
|
|
int ialloced; /* inode allocation status */
|
|
|
|
int noroom = 0; /* no space for inode blk allocation */
|
|
|
|
xfs_ino_t ino; /* fs-relative inode to be returned */
|
|
|
|
/* REFERENCED */
|
|
|
|
int j; /* result code */
|
|
|
|
xfs_mount_t *mp; /* file system mount structure */
|
|
|
|
int offset; /* index of inode in chunk */
|
2009-08-31 23:57:14 +00:00
|
|
|
xfs_agino_t pagino; /* parent's AG relative inode # */
|
|
|
|
xfs_agnumber_t pagno; /* parent's AG number */
|
2006-09-28 00:57:04 +00:00
|
|
|
xfs_inobt_rec_incore_t rec; /* inode allocation record */
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_agnumber_t tagno; /* testing allocation group number */
|
|
|
|
xfs_btree_cur_t *tcur; /* temp cursor */
|
2006-09-28 00:57:04 +00:00
|
|
|
xfs_inobt_rec_incore_t trec; /* temp inode allocation record */
|
2010-01-11 11:47:43 +00:00
|
|
|
struct xfs_perag *pag;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
|
|
|
|
if (*IO_agbp == NULL) {
|
|
|
|
/*
|
|
|
|
* We do not have an agbp, so select an initial allocation
|
|
|
|
* group for inode allocation.
|
|
|
|
*/
|
|
|
|
agbp = xfs_ialloc_ag_select(tp, parent, mode, okalloc);
|
|
|
|
/*
|
|
|
|
* Couldn't find an allocation group satisfying the
|
|
|
|
* criteria, give up.
|
|
|
|
*/
|
|
|
|
if (!agbp) {
|
|
|
|
*inop = NULLFSINO;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
agi = XFS_BUF_TO_AGI(agbp);
|
2005-11-02 04:11:25 +00:00
|
|
|
ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
|
2005-04-16 22:20:36 +00:00
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* Continue where we left off before. In this case, we
|
|
|
|
* know that the allocation group has free inodes.
|
|
|
|
*/
|
|
|
|
agbp = *IO_agbp;
|
|
|
|
agi = XFS_BUF_TO_AGI(agbp);
|
2005-11-02 04:11:25 +00:00
|
|
|
ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
|
|
|
|
ASSERT(be32_to_cpu(agi->agi_freecount) > 0);
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
|
|
|
mp = tp->t_mountp;
|
|
|
|
agcount = mp->m_sb.sb_agcount;
|
2005-11-02 04:11:25 +00:00
|
|
|
agno = be32_to_cpu(agi->agi_seqno);
|
2005-04-16 22:20:36 +00:00
|
|
|
tagno = agno;
|
|
|
|
pagno = XFS_INO_TO_AGNO(mp, parent);
|
|
|
|
pagino = XFS_INO_TO_AGINO(mp, parent);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If we have already hit the ceiling of inode blocks then clear
|
|
|
|
* okalloc so we scan all available agi structures for a free
|
|
|
|
* inode.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (mp->m_maxicount &&
|
|
|
|
mp->m_sb.sb_icount + XFS_IALLOC_INODES(mp) > mp->m_maxicount) {
|
|
|
|
noroom = 1;
|
|
|
|
okalloc = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Loop until we find an allocation group that either has free inodes
|
|
|
|
* or in which we can allocate some inodes. Iterate through the
|
|
|
|
* allocation groups upward, wrapping at the end.
|
|
|
|
*/
|
|
|
|
*alloc_done = B_FALSE;
|
|
|
|
while (!agi->agi_freecount) {
|
|
|
|
/*
|
|
|
|
* Don't do anything if we're not supposed to allocate
|
|
|
|
* any blocks, just go on to the next ag.
|
|
|
|
*/
|
|
|
|
if (okalloc) {
|
|
|
|
/*
|
|
|
|
* Try to allocate some new inodes in the allocation
|
|
|
|
* group.
|
|
|
|
*/
|
|
|
|
if ((error = xfs_ialloc_ag_alloc(tp, agbp, &ialloced))) {
|
|
|
|
xfs_trans_brelse(tp, agbp);
|
|
|
|
if (error == ENOSPC) {
|
|
|
|
*inop = NULLFSINO;
|
|
|
|
return 0;
|
|
|
|
} else
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
if (ialloced) {
|
|
|
|
/*
|
|
|
|
* We successfully allocated some inodes, return
|
|
|
|
* the current context to the caller so that it
|
|
|
|
* can commit the current transaction and call
|
|
|
|
* us again where we left off.
|
|
|
|
*/
|
2005-11-02 04:11:25 +00:00
|
|
|
ASSERT(be32_to_cpu(agi->agi_freecount) > 0);
|
2005-04-16 22:20:36 +00:00
|
|
|
*alloc_done = B_TRUE;
|
|
|
|
*IO_agbp = agbp;
|
|
|
|
*inop = NULLFSINO;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* If it failed, give up on this ag.
|
|
|
|
*/
|
|
|
|
xfs_trans_brelse(tp, agbp);
|
|
|
|
/*
|
|
|
|
* Go on to the next ag: get its ag header.
|
|
|
|
*/
|
|
|
|
nextag:
|
|
|
|
if (++tagno == agcount)
|
|
|
|
tagno = 0;
|
|
|
|
if (tagno == agno) {
|
|
|
|
*inop = NULLFSINO;
|
|
|
|
return noroom ? ENOSPC : 0;
|
|
|
|
}
|
|
|
|
down_read(&mp->m_peraglock);
|
2010-01-11 11:47:43 +00:00
|
|
|
pag = xfs_perag_get(mp, tagno);
|
|
|
|
if (pag->pagi_inodeok == 0) {
|
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
up_read(&mp->m_peraglock);
|
|
|
|
goto nextag;
|
|
|
|
}
|
|
|
|
error = xfs_ialloc_read_agi(mp, tp, tagno, &agbp);
|
2010-01-11 11:47:43 +00:00
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
up_read(&mp->m_peraglock);
|
|
|
|
if (error)
|
|
|
|
goto nextag;
|
|
|
|
agi = XFS_BUF_TO_AGI(agbp);
|
2005-11-02 04:11:25 +00:00
|
|
|
ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
|
|
|
/*
|
|
|
|
* Here with an allocation group that has a free inode.
|
|
|
|
* Reset agno since we may have chosen a new ag in the
|
|
|
|
* loop above.
|
|
|
|
*/
|
|
|
|
agno = tagno;
|
|
|
|
*IO_agbp = NULL;
|
2010-01-11 11:47:43 +00:00
|
|
|
pag = xfs_perag_get(mp, agno);
|
2009-08-31 23:58:28 +00:00
|
|
|
|
|
|
|
restart_pagno:
|
2008-10-30 05:53:59 +00:00
|
|
|
cur = xfs_inobt_init_cursor(mp, tp, agbp, be32_to_cpu(agi->agi_seqno));
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* If pagino is 0 (this is the root inode allocation) use newino.
|
|
|
|
* This must work because we've just allocated some.
|
|
|
|
*/
|
|
|
|
if (!pagino)
|
2005-11-02 04:11:25 +00:00
|
|
|
pagino = be32_to_cpu(agi->agi_newino);
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2009-08-31 23:57:09 +00:00
|
|
|
error = xfs_check_agi_freecount(cur, agi);
|
|
|
|
if (error)
|
|
|
|
goto error0;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
/*
|
2009-08-31 23:57:14 +00:00
|
|
|
* If in the same AG as the parent, try to get near the parent.
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
|
|
|
if (pagno == agno) {
|
2009-08-31 23:57:14 +00:00
|
|
|
int doneleft; /* done, to the left */
|
|
|
|
int doneright; /* done, to the right */
|
2009-08-31 23:58:28 +00:00
|
|
|
int searchdistance = 10;
|
2009-08-31 23:57:14 +00:00
|
|
|
|
2009-08-31 23:58:21 +00:00
|
|
|
error = xfs_inobt_lookup(cur, pagino, XFS_LOOKUP_LE, &i);
|
2009-08-31 23:57:14 +00:00
|
|
|
if (error)
|
2005-04-16 22:20:36 +00:00
|
|
|
goto error0;
|
2009-08-31 23:57:14 +00:00
|
|
|
XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
|
|
|
|
|
|
|
|
error = xfs_inobt_get_rec(cur, &rec, &j);
|
|
|
|
if (error)
|
|
|
|
goto error0;
|
|
|
|
XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
|
|
|
|
|
|
|
|
if (rec.ir_freecount > 0) {
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Found a free inode in the same chunk
|
2009-08-31 23:57:14 +00:00
|
|
|
* as the parent, done.
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
2009-08-31 23:57:14 +00:00
|
|
|
goto alloc_inode;
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
2009-08-31 23:57:14 +00:00
|
|
|
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
2009-08-31 23:57:14 +00:00
|
|
|
* In the same AG as parent, but parent's chunk is full.
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
|
|
|
|
2009-08-31 23:57:14 +00:00
|
|
|
/* duplicate the cursor, search left & right simultaneously */
|
|
|
|
error = xfs_btree_dup_cursor(cur, &tcur);
|
|
|
|
if (error)
|
|
|
|
goto error0;
|
|
|
|
|
2009-08-31 23:58:28 +00:00
|
|
|
/*
|
|
|
|
* Skip to last blocks looked up if same parent inode.
|
|
|
|
*/
|
|
|
|
if (pagino != NULLAGINO &&
|
|
|
|
pag->pagl_pagino == pagino &&
|
|
|
|
pag->pagl_leftrec != NULLAGINO &&
|
|
|
|
pag->pagl_rightrec != NULLAGINO) {
|
|
|
|
error = xfs_ialloc_get_rec(tcur, pag->pagl_leftrec,
|
|
|
|
&trec, &doneleft, 1);
|
|
|
|
if (error)
|
|
|
|
goto error1;
|
2009-08-31 23:57:14 +00:00
|
|
|
|
2009-08-31 23:58:28 +00:00
|
|
|
error = xfs_ialloc_get_rec(cur, pag->pagl_rightrec,
|
|
|
|
&rec, &doneright, 0);
|
|
|
|
if (error)
|
|
|
|
goto error1;
|
|
|
|
} else {
|
|
|
|
/* search left with tcur, back up 1 record */
|
|
|
|
error = xfs_ialloc_next_rec(tcur, &trec, &doneleft, 1);
|
|
|
|
if (error)
|
|
|
|
goto error1;
|
|
|
|
|
|
|
|
/* search right with cur, go forward 1 record. */
|
|
|
|
error = xfs_ialloc_next_rec(cur, &rec, &doneright, 0);
|
|
|
|
if (error)
|
|
|
|
goto error1;
|
|
|
|
}
|
2009-08-31 23:57:14 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Loop until we find an inode chunk with a free inode.
|
|
|
|
*/
|
|
|
|
while (!doneleft || !doneright) {
|
|
|
|
int useleft; /* using left inode chunk this time */
|
|
|
|
|
2009-08-31 23:58:28 +00:00
|
|
|
if (!--searchdistance) {
|
|
|
|
/*
|
|
|
|
* Not in range - save last search
|
|
|
|
* location and allocate a new inode
|
|
|
|
*/
|
2009-10-30 08:27:07 +00:00
|
|
|
xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
|
2009-08-31 23:58:28 +00:00
|
|
|
pag->pagl_leftrec = trec.ir_startino;
|
|
|
|
pag->pagl_rightrec = rec.ir_startino;
|
|
|
|
pag->pagl_pagino = pagino;
|
|
|
|
goto newino;
|
|
|
|
}
|
|
|
|
|
2009-08-31 23:57:14 +00:00
|
|
|
/* figure out the closer block if both are valid. */
|
|
|
|
if (!doneleft && !doneright) {
|
|
|
|
useleft = pagino -
|
|
|
|
(trec.ir_startino + XFS_INODES_PER_CHUNK - 1) <
|
|
|
|
rec.ir_startino - pagino;
|
|
|
|
} else {
|
|
|
|
useleft = !doneleft;
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
2009-08-31 23:57:14 +00:00
|
|
|
|
|
|
|
/* free inodes to the left? */
|
|
|
|
if (useleft && trec.ir_freecount) {
|
|
|
|
rec = trec;
|
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
|
|
|
|
cur = tcur;
|
2009-08-31 23:58:28 +00:00
|
|
|
|
|
|
|
pag->pagl_leftrec = trec.ir_startino;
|
|
|
|
pag->pagl_rightrec = rec.ir_startino;
|
|
|
|
pag->pagl_pagino = pagino;
|
2009-08-31 23:57:14 +00:00
|
|
|
goto alloc_inode;
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
|
|
|
|
2009-08-31 23:57:14 +00:00
|
|
|
/* free inodes to the right? */
|
|
|
|
if (!useleft && rec.ir_freecount) {
|
|
|
|
xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
|
2009-08-31 23:58:28 +00:00
|
|
|
|
|
|
|
pag->pagl_leftrec = trec.ir_startino;
|
|
|
|
pag->pagl_rightrec = rec.ir_startino;
|
|
|
|
pag->pagl_pagino = pagino;
|
2009-08-31 23:57:14 +00:00
|
|
|
goto alloc_inode;
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
2009-08-31 23:57:14 +00:00
|
|
|
|
|
|
|
/* get next record to check */
|
|
|
|
if (useleft) {
|
|
|
|
error = xfs_ialloc_next_rec(tcur, &trec,
|
|
|
|
&doneleft, 1);
|
|
|
|
} else {
|
|
|
|
error = xfs_ialloc_next_rec(cur, &rec,
|
|
|
|
&doneright, 0);
|
|
|
|
}
|
|
|
|
if (error)
|
|
|
|
goto error1;
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
2009-08-31 23:58:28 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We've reached the end of the btree. because
|
|
|
|
* we are only searching a small chunk of the
|
|
|
|
* btree each search, there is obviously free
|
|
|
|
* inodes closer to the parent inode than we
|
|
|
|
* are now. restart the search again.
|
|
|
|
*/
|
|
|
|
pag->pagl_pagino = NULLAGINO;
|
|
|
|
pag->pagl_leftrec = NULLAGINO;
|
|
|
|
pag->pagl_rightrec = NULLAGINO;
|
|
|
|
xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
|
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
|
|
|
|
goto restart_pagno;
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
2009-08-31 23:57:14 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
2009-08-31 23:57:14 +00:00
|
|
|
* In a different AG from the parent.
|
2005-04-16 22:20:36 +00:00
|
|
|
* See if the most recently allocated block has any free.
|
|
|
|
*/
|
2009-08-31 23:58:28 +00:00
|
|
|
newino:
|
|
|
|
if (be32_to_cpu(agi->agi_newino) != NULLAGINO) {
|
2009-08-31 23:58:21 +00:00
|
|
|
error = xfs_inobt_lookup(cur, be32_to_cpu(agi->agi_newino),
|
|
|
|
XFS_LOOKUP_EQ, &i);
|
2009-08-31 23:57:14 +00:00
|
|
|
if (error)
|
2005-04-16 22:20:36 +00:00
|
|
|
goto error0;
|
2009-08-31 23:57:14 +00:00
|
|
|
|
|
|
|
if (i == 1) {
|
|
|
|
error = xfs_inobt_get_rec(cur, &rec, &j);
|
|
|
|
if (error)
|
|
|
|
goto error0;
|
|
|
|
|
|
|
|
if (j == 1 && rec.ir_freecount > 0) {
|
|
|
|
/*
|
|
|
|
* The last chunk allocated in the group
|
|
|
|
* still has a free inode.
|
|
|
|
*/
|
|
|
|
goto alloc_inode;
|
|
|
|
}
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
2009-08-31 23:58:28 +00:00
|
|
|
}
|
2009-08-31 23:57:14 +00:00
|
|
|
|
2009-08-31 23:58:28 +00:00
|
|
|
/*
|
|
|
|
* None left in the last group, search the whole AG
|
|
|
|
*/
|
|
|
|
error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &i);
|
|
|
|
if (error)
|
|
|
|
goto error0;
|
|
|
|
XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
error = xfs_inobt_get_rec(cur, &rec, &i);
|
|
|
|
if (error)
|
|
|
|
goto error0;
|
|
|
|
XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
|
|
|
|
if (rec.ir_freecount > 0)
|
|
|
|
break;
|
|
|
|
error = xfs_btree_increment(cur, 0, &i);
|
2009-08-31 23:57:14 +00:00
|
|
|
if (error)
|
|
|
|
goto error0;
|
|
|
|
XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
2009-08-31 23:57:14 +00:00
|
|
|
|
|
|
|
alloc_inode:
|
2009-01-15 05:22:07 +00:00
|
|
|
offset = xfs_ialloc_find_free(&rec.ir_free);
|
2005-04-16 22:20:36 +00:00
|
|
|
ASSERT(offset >= 0);
|
|
|
|
ASSERT(offset < XFS_INODES_PER_CHUNK);
|
|
|
|
ASSERT((XFS_AGINO_TO_OFFSET(mp, rec.ir_startino) %
|
|
|
|
XFS_INODES_PER_CHUNK) == 0);
|
|
|
|
ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino + offset);
|
2009-02-09 07:37:14 +00:00
|
|
|
rec.ir_free &= ~XFS_INOBT_MASK(offset);
|
2005-04-16 22:20:36 +00:00
|
|
|
rec.ir_freecount--;
|
2009-08-31 23:57:03 +00:00
|
|
|
error = xfs_inobt_update(cur, &rec);
|
|
|
|
if (error)
|
2005-04-16 22:20:36 +00:00
|
|
|
goto error0;
|
2008-02-13 23:03:29 +00:00
|
|
|
be32_add_cpu(&agi->agi_freecount, -1);
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
|
|
|
|
down_read(&mp->m_peraglock);
|
2010-01-11 11:47:43 +00:00
|
|
|
pag->pagi_freecount--;
|
2005-04-16 22:20:36 +00:00
|
|
|
up_read(&mp->m_peraglock);
|
|
|
|
|
2009-08-31 23:57:09 +00:00
|
|
|
error = xfs_check_agi_freecount(cur, agi);
|
|
|
|
if (error)
|
|
|
|
goto error0;
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
|
|
|
|
xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -1);
|
2010-01-11 11:47:43 +00:00
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
*inop = ino;
|
|
|
|
return 0;
|
|
|
|
error1:
|
|
|
|
xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
|
|
|
|
error0:
|
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
|
2010-01-11 11:47:43 +00:00
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Free disk inode. Carefully avoids touching the incore inode, all
|
|
|
|
* manipulations incore are the caller's responsibility.
|
|
|
|
* The on-disk inode is not changed by this operation, only the
|
|
|
|
* btree (free inode mask) is changed.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
xfs_difree(
|
|
|
|
xfs_trans_t *tp, /* transaction pointer */
|
|
|
|
xfs_ino_t inode, /* inode to be freed */
|
|
|
|
xfs_bmap_free_t *flist, /* extents to free */
|
|
|
|
int *delete, /* set if inode cluster was deleted */
|
|
|
|
xfs_ino_t *first_ino) /* first inode in deleted cluster */
|
|
|
|
{
|
|
|
|
/* REFERENCED */
|
|
|
|
xfs_agblock_t agbno; /* block number containing inode */
|
|
|
|
xfs_buf_t *agbp; /* buffer containing allocation group header */
|
|
|
|
xfs_agino_t agino; /* inode number relative to allocation group */
|
|
|
|
xfs_agnumber_t agno; /* allocation group number */
|
|
|
|
xfs_agi_t *agi; /* allocation group header */
|
|
|
|
xfs_btree_cur_t *cur; /* inode btree cursor */
|
|
|
|
int error; /* error return value */
|
|
|
|
int i; /* result code */
|
|
|
|
int ilen; /* inodes in an inode cluster */
|
|
|
|
xfs_mount_t *mp; /* mount structure for filesystem */
|
|
|
|
int off; /* offset of inode in inode chunk */
|
2006-09-28 00:57:04 +00:00
|
|
|
xfs_inobt_rec_incore_t rec; /* btree record */
|
2010-01-11 11:47:43 +00:00
|
|
|
struct xfs_perag *pag;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
mp = tp->t_mountp;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Break up inode number into its components.
|
|
|
|
*/
|
|
|
|
agno = XFS_INO_TO_AGNO(mp, inode);
|
|
|
|
if (agno >= mp->m_sb.sb_agcount) {
|
|
|
|
cmn_err(CE_WARN,
|
|
|
|
"xfs_difree: agno >= mp->m_sb.sb_agcount (%d >= %d) on %s. Returning EINVAL.",
|
|
|
|
agno, mp->m_sb.sb_agcount, mp->m_fsname);
|
|
|
|
ASSERT(0);
|
|
|
|
return XFS_ERROR(EINVAL);
|
|
|
|
}
|
|
|
|
agino = XFS_INO_TO_AGINO(mp, inode);
|
|
|
|
if (inode != XFS_AGINO_TO_INO(mp, agno, agino)) {
|
|
|
|
cmn_err(CE_WARN,
|
2005-11-01 23:21:35 +00:00
|
|
|
"xfs_difree: inode != XFS_AGINO_TO_INO() "
|
|
|
|
"(%llu != %llu) on %s. Returning EINVAL.",
|
|
|
|
(unsigned long long)inode,
|
|
|
|
(unsigned long long)XFS_AGINO_TO_INO(mp, agno, agino),
|
|
|
|
mp->m_fsname);
|
2005-04-16 22:20:36 +00:00
|
|
|
ASSERT(0);
|
|
|
|
return XFS_ERROR(EINVAL);
|
|
|
|
}
|
|
|
|
agbno = XFS_AGINO_TO_AGBNO(mp, agino);
|
|
|
|
if (agbno >= mp->m_sb.sb_agblocks) {
|
|
|
|
cmn_err(CE_WARN,
|
|
|
|
"xfs_difree: agbno >= mp->m_sb.sb_agblocks (%d >= %d) on %s. Returning EINVAL.",
|
|
|
|
agbno, mp->m_sb.sb_agblocks, mp->m_fsname);
|
|
|
|
ASSERT(0);
|
|
|
|
return XFS_ERROR(EINVAL);
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* Get the allocation group header.
|
|
|
|
*/
|
|
|
|
down_read(&mp->m_peraglock);
|
|
|
|
error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
|
|
|
|
up_read(&mp->m_peraglock);
|
|
|
|
if (error) {
|
|
|
|
cmn_err(CE_WARN,
|
|
|
|
"xfs_difree: xfs_ialloc_read_agi() returned an error %d on %s. Returning error.",
|
|
|
|
error, mp->m_fsname);
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
agi = XFS_BUF_TO_AGI(agbp);
|
2005-11-02 04:11:25 +00:00
|
|
|
ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
|
|
|
|
ASSERT(agbno < be32_to_cpu(agi->agi_length));
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Initialize the cursor.
|
|
|
|
*/
|
2008-10-30 05:53:59 +00:00
|
|
|
cur = xfs_inobt_init_cursor(mp, tp, agbp, agno);
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2009-08-31 23:57:09 +00:00
|
|
|
error = xfs_check_agi_freecount(cur, agi);
|
|
|
|
if (error)
|
|
|
|
goto error0;
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Look for the entry describing this inode.
|
|
|
|
*/
|
2009-08-31 23:58:21 +00:00
|
|
|
if ((error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE, &i))) {
|
2005-04-16 22:20:36 +00:00
|
|
|
cmn_err(CE_WARN,
|
2009-08-31 23:58:21 +00:00
|
|
|
"xfs_difree: xfs_inobt_lookup returned() an error %d on %s. Returning error.",
|
2005-04-16 22:20:36 +00:00
|
|
|
error, mp->m_fsname);
|
|
|
|
goto error0;
|
|
|
|
}
|
|
|
|
XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
|
2009-08-31 23:56:58 +00:00
|
|
|
error = xfs_inobt_get_rec(cur, &rec, &i);
|
|
|
|
if (error) {
|
2005-04-16 22:20:36 +00:00
|
|
|
cmn_err(CE_WARN,
|
|
|
|
"xfs_difree: xfs_inobt_get_rec() returned an error %d on %s. Returning error.",
|
|
|
|
error, mp->m_fsname);
|
|
|
|
goto error0;
|
|
|
|
}
|
|
|
|
XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
|
|
|
|
/*
|
|
|
|
* Get the offset in the inode chunk.
|
|
|
|
*/
|
|
|
|
off = agino - rec.ir_startino;
|
|
|
|
ASSERT(off >= 0 && off < XFS_INODES_PER_CHUNK);
|
2009-02-09 07:37:14 +00:00
|
|
|
ASSERT(!(rec.ir_free & XFS_INOBT_MASK(off)));
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Mark the inode free & increment the count.
|
|
|
|
*/
|
2009-02-09 07:37:14 +00:00
|
|
|
rec.ir_free |= XFS_INOBT_MASK(off);
|
2005-04-16 22:20:36 +00:00
|
|
|
rec.ir_freecount++;
|
|
|
|
|
|
|
|
/*
|
2006-03-28 22:55:14 +00:00
|
|
|
* When an inode cluster is free, it becomes eligible for removal
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
2008-02-29 02:58:40 +00:00
|
|
|
if (!(mp->m_flags & XFS_MOUNT_IKEEP) &&
|
2005-04-16 22:20:36 +00:00
|
|
|
(rec.ir_freecount == XFS_IALLOC_INODES(mp))) {
|
|
|
|
|
|
|
|
*delete = 1;
|
|
|
|
*first_ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Remove the inode cluster from the AGI B+Tree, adjust the
|
|
|
|
* AGI and Superblock inode counts, and mark the disk space
|
|
|
|
* to be freed when the transaction is committed.
|
|
|
|
*/
|
|
|
|
ilen = XFS_IALLOC_INODES(mp);
|
2008-02-13 23:03:29 +00:00
|
|
|
be32_add_cpu(&agi->agi_count, -ilen);
|
|
|
|
be32_add_cpu(&agi->agi_freecount, -(ilen - 1));
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_ialloc_log_agi(tp, agbp, XFS_AGI_COUNT | XFS_AGI_FREECOUNT);
|
|
|
|
down_read(&mp->m_peraglock);
|
2010-01-11 11:47:43 +00:00
|
|
|
pag = xfs_perag_get(mp, agno);
|
|
|
|
pag->pagi_freecount -= ilen - 1;
|
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
up_read(&mp->m_peraglock);
|
|
|
|
xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, -ilen);
|
|
|
|
xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -(ilen - 1));
|
|
|
|
|
2008-10-30 05:58:01 +00:00
|
|
|
if ((error = xfs_btree_delete(cur, &i))) {
|
|
|
|
cmn_err(CE_WARN, "xfs_difree: xfs_btree_delete returned an error %d on %s.\n",
|
2005-04-16 22:20:36 +00:00
|
|
|
error, mp->m_fsname);
|
|
|
|
goto error0;
|
|
|
|
}
|
|
|
|
|
|
|
|
xfs_bmap_add_free(XFS_AGB_TO_FSB(mp,
|
|
|
|
agno, XFS_INO_TO_AGBNO(mp,rec.ir_startino)),
|
|
|
|
XFS_IALLOC_BLOCKS(mp), flist, mp);
|
|
|
|
} else {
|
|
|
|
*delete = 0;
|
|
|
|
|
2009-08-31 23:57:03 +00:00
|
|
|
error = xfs_inobt_update(cur, &rec);
|
|
|
|
if (error) {
|
2005-04-16 22:20:36 +00:00
|
|
|
cmn_err(CE_WARN,
|
2009-08-31 23:57:03 +00:00
|
|
|
"xfs_difree: xfs_inobt_update returned an error %d on %s.",
|
2005-04-16 22:20:36 +00:00
|
|
|
error, mp->m_fsname);
|
|
|
|
goto error0;
|
|
|
|
}
|
2009-08-31 23:57:03 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Change the inode free counts and log the ag/sb changes.
|
|
|
|
*/
|
2008-02-13 23:03:29 +00:00
|
|
|
be32_add_cpu(&agi->agi_freecount, 1);
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
|
|
|
|
down_read(&mp->m_peraglock);
|
2010-01-11 11:47:43 +00:00
|
|
|
pag = xfs_perag_get(mp, agno);
|
|
|
|
pag->pagi_freecount++;
|
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
up_read(&mp->m_peraglock);
|
|
|
|
xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, 1);
|
|
|
|
}
|
|
|
|
|
2009-08-31 23:57:09 +00:00
|
|
|
error = xfs_check_agi_freecount(cur, agi);
|
|
|
|
if (error)
|
|
|
|
goto error0;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
error0:
|
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2008-11-28 03:23:41 +00:00
|
|
|
* Return the location of the inode in imap, for mapping it into a buffer.
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
|
|
|
int
|
2008-11-28 03:23:41 +00:00
|
|
|
xfs_imap(
|
|
|
|
xfs_mount_t *mp, /* file system mount structure */
|
|
|
|
xfs_trans_t *tp, /* transaction pointer */
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_ino_t ino, /* inode to locate */
|
2008-11-28 03:23:41 +00:00
|
|
|
struct xfs_imap *imap, /* location map structure */
|
|
|
|
uint flags) /* flags for inode btree lookup */
|
2005-04-16 22:20:36 +00:00
|
|
|
{
|
|
|
|
xfs_agblock_t agbno; /* block number of inode in the alloc group */
|
|
|
|
xfs_agino_t agino; /* inode number within alloc group */
|
|
|
|
xfs_agnumber_t agno; /* allocation group number */
|
|
|
|
int blks_per_cluster; /* num blocks per inode cluster */
|
|
|
|
xfs_agblock_t chunk_agbno; /* first block in inode chunk */
|
|
|
|
xfs_agblock_t cluster_agbno; /* first block in inode cluster */
|
|
|
|
int error; /* error code */
|
|
|
|
int offset; /* index of inode in its buffer */
|
|
|
|
int offset_agbno; /* blks from chunk start to inode */
|
|
|
|
|
|
|
|
ASSERT(ino != NULLFSINO);
|
2008-11-28 03:23:41 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Split up the inode number into its parts.
|
|
|
|
*/
|
|
|
|
agno = XFS_INO_TO_AGNO(mp, ino);
|
|
|
|
agino = XFS_INO_TO_AGINO(mp, ino);
|
|
|
|
agbno = XFS_AGINO_TO_AGBNO(mp, agino);
|
|
|
|
if (agno >= mp->m_sb.sb_agcount || agbno >= mp->m_sb.sb_agblocks ||
|
|
|
|
ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
|
|
|
|
#ifdef DEBUG
|
2006-06-09 07:12:28 +00:00
|
|
|
/* no diagnostics for bulkstat, ino comes from userspace */
|
2008-11-28 03:23:41 +00:00
|
|
|
if (flags & XFS_IGET_BULKSTAT)
|
2006-06-09 07:12:28 +00:00
|
|
|
return XFS_ERROR(EINVAL);
|
2005-04-16 22:20:36 +00:00
|
|
|
if (agno >= mp->m_sb.sb_agcount) {
|
|
|
|
xfs_fs_cmn_err(CE_ALERT, mp,
|
2008-11-28 03:23:41 +00:00
|
|
|
"xfs_imap: agno (%d) >= "
|
2005-04-16 22:20:36 +00:00
|
|
|
"mp->m_sb.sb_agcount (%d)",
|
|
|
|
agno, mp->m_sb.sb_agcount);
|
|
|
|
}
|
|
|
|
if (agbno >= mp->m_sb.sb_agblocks) {
|
|
|
|
xfs_fs_cmn_err(CE_ALERT, mp,
|
2008-11-28 03:23:41 +00:00
|
|
|
"xfs_imap: agbno (0x%llx) >= "
|
2005-04-16 22:20:36 +00:00
|
|
|
"mp->m_sb.sb_agblocks (0x%lx)",
|
|
|
|
(unsigned long long) agbno,
|
|
|
|
(unsigned long) mp->m_sb.sb_agblocks);
|
|
|
|
}
|
|
|
|
if (ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
|
|
|
|
xfs_fs_cmn_err(CE_ALERT, mp,
|
2008-11-28 03:23:41 +00:00
|
|
|
"xfs_imap: ino (0x%llx) != "
|
2005-04-16 22:20:36 +00:00
|
|
|
"XFS_AGINO_TO_INO(mp, agno, agino) "
|
|
|
|
"(0x%llx)",
|
|
|
|
ino, XFS_AGINO_TO_INO(mp, agno, agino));
|
|
|
|
}
|
2006-09-28 01:02:23 +00:00
|
|
|
xfs_stack_trace();
|
2005-04-16 22:20:36 +00:00
|
|
|
#endif /* DEBUG */
|
|
|
|
return XFS_ERROR(EINVAL);
|
|
|
|
}
|
2008-11-28 03:23:41 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If the inode cluster size is the same as the blocksize or
|
|
|
|
* smaller we get to the buffer by simple arithmetics.
|
|
|
|
*/
|
|
|
|
if (XFS_INODE_CLUSTER_SIZE(mp) <= mp->m_sb.sb_blocksize) {
|
2005-04-16 22:20:36 +00:00
|
|
|
offset = XFS_INO_TO_OFFSET(mp, ino);
|
|
|
|
ASSERT(offset < mp->m_sb.sb_inopblock);
|
2008-11-28 03:23:41 +00:00
|
|
|
|
|
|
|
imap->im_blkno = XFS_AGB_TO_DADDR(mp, agno, agbno);
|
|
|
|
imap->im_len = XFS_FSB_TO_BB(mp, 1);
|
|
|
|
imap->im_boffset = (ushort)(offset << mp->m_sb.sb_inodelog);
|
2005-04-16 22:20:36 +00:00
|
|
|
return 0;
|
|
|
|
}
|
2008-11-28 03:23:41 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
blks_per_cluster = XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_blocklog;
|
2008-11-28 03:23:41 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If we get a block number passed from bulkstat we can use it to
|
|
|
|
* find the buffer easily.
|
|
|
|
*/
|
|
|
|
if (imap->im_blkno) {
|
2005-04-16 22:20:36 +00:00
|
|
|
offset = XFS_INO_TO_OFFSET(mp, ino);
|
|
|
|
ASSERT(offset < mp->m_sb.sb_inopblock);
|
2008-11-28 03:23:41 +00:00
|
|
|
|
2009-01-15 05:22:07 +00:00
|
|
|
cluster_agbno = xfs_daddr_to_agbno(mp, imap->im_blkno);
|
2008-11-28 03:23:41 +00:00
|
|
|
offset += (agbno - cluster_agbno) * mp->m_sb.sb_inopblock;
|
|
|
|
|
|
|
|
imap->im_len = XFS_FSB_TO_BB(mp, blks_per_cluster);
|
|
|
|
imap->im_boffset = (ushort)(offset << mp->m_sb.sb_inodelog);
|
2005-04-16 22:20:36 +00:00
|
|
|
return 0;
|
|
|
|
}
|
2008-11-28 03:23:41 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If the inode chunks are aligned then use simple maths to
|
|
|
|
* find the location. Otherwise we have to do a btree
|
|
|
|
* lookup to find the location.
|
|
|
|
*/
|
2005-04-16 22:20:36 +00:00
|
|
|
if (mp->m_inoalign_mask) {
|
|
|
|
offset_agbno = agbno & mp->m_inoalign_mask;
|
|
|
|
chunk_agbno = agbno - offset_agbno;
|
|
|
|
} else {
|
2008-11-28 03:23:41 +00:00
|
|
|
xfs_btree_cur_t *cur; /* inode btree cursor */
|
2009-08-31 23:56:58 +00:00
|
|
|
xfs_inobt_rec_incore_t chunk_rec;
|
2008-11-28 03:23:41 +00:00
|
|
|
xfs_buf_t *agbp; /* agi buffer */
|
|
|
|
int i; /* temp state */
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
down_read(&mp->m_peraglock);
|
|
|
|
error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
|
|
|
|
up_read(&mp->m_peraglock);
|
|
|
|
if (error) {
|
2008-11-28 03:23:41 +00:00
|
|
|
xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
|
2005-04-16 22:20:36 +00:00
|
|
|
"xfs_ialloc_read_agi() returned "
|
|
|
|
"error %d, agno %d",
|
|
|
|
error, agno);
|
|
|
|
return error;
|
|
|
|
}
|
2008-11-28 03:23:41 +00:00
|
|
|
|
2008-10-30 05:53:59 +00:00
|
|
|
cur = xfs_inobt_init_cursor(mp, tp, agbp, agno);
|
2009-08-31 23:58:21 +00:00
|
|
|
error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE, &i);
|
2008-11-28 03:23:41 +00:00
|
|
|
if (error) {
|
|
|
|
xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
|
2009-08-31 23:58:21 +00:00
|
|
|
"xfs_inobt_lookup() failed");
|
2005-04-16 22:20:36 +00:00
|
|
|
goto error0;
|
|
|
|
}
|
2008-11-28 03:23:41 +00:00
|
|
|
|
2009-08-31 23:56:58 +00:00
|
|
|
error = xfs_inobt_get_rec(cur, &chunk_rec, &i);
|
2008-11-28 03:23:41 +00:00
|
|
|
if (error) {
|
|
|
|
xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
|
2005-04-16 22:20:36 +00:00
|
|
|
"xfs_inobt_get_rec() failed");
|
|
|
|
goto error0;
|
|
|
|
}
|
|
|
|
if (i == 0) {
|
|
|
|
#ifdef DEBUG
|
2008-11-28 03:23:41 +00:00
|
|
|
xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
|
2005-04-16 22:20:36 +00:00
|
|
|
"xfs_inobt_get_rec() failed");
|
|
|
|
#endif /* DEBUG */
|
|
|
|
error = XFS_ERROR(EINVAL);
|
|
|
|
}
|
2008-11-28 03:23:41 +00:00
|
|
|
error0:
|
2005-04-16 22:20:36 +00:00
|
|
|
xfs_trans_brelse(tp, agbp);
|
|
|
|
xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
|
|
|
|
if (error)
|
|
|
|
return error;
|
2009-08-31 23:56:58 +00:00
|
|
|
chunk_agbno = XFS_AGINO_TO_AGBNO(mp, chunk_rec.ir_startino);
|
2005-04-16 22:20:36 +00:00
|
|
|
offset_agbno = agbno - chunk_agbno;
|
|
|
|
}
|
2008-11-28 03:23:41 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
ASSERT(agbno >= chunk_agbno);
|
|
|
|
cluster_agbno = chunk_agbno +
|
|
|
|
((offset_agbno / blks_per_cluster) * blks_per_cluster);
|
|
|
|
offset = ((agbno - cluster_agbno) * mp->m_sb.sb_inopblock) +
|
|
|
|
XFS_INO_TO_OFFSET(mp, ino);
|
2008-11-28 03:23:41 +00:00
|
|
|
|
|
|
|
imap->im_blkno = XFS_AGB_TO_DADDR(mp, agno, cluster_agbno);
|
|
|
|
imap->im_len = XFS_FSB_TO_BB(mp, blks_per_cluster);
|
|
|
|
imap->im_boffset = (ushort)(offset << mp->m_sb.sb_inodelog);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the inode number maps to a block outside the bounds
|
|
|
|
* of the file system then return NULL rather than calling
|
|
|
|
* read_buf and panicing when we get an error from the
|
|
|
|
* driver.
|
|
|
|
*/
|
|
|
|
if ((imap->im_blkno + imap->im_len) >
|
|
|
|
XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
|
|
|
|
xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
|
|
|
|
"(imap->im_blkno (0x%llx) + imap->im_len (0x%llx)) > "
|
|
|
|
" XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks) (0x%llx)",
|
|
|
|
(unsigned long long) imap->im_blkno,
|
|
|
|
(unsigned long long) imap->im_len,
|
|
|
|
XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks));
|
|
|
|
return XFS_ERROR(EINVAL);
|
|
|
|
}
|
2005-04-16 22:20:36 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Compute and fill in value of m_in_maxlevels.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
xfs_ialloc_compute_maxlevels(
|
|
|
|
xfs_mount_t *mp) /* file system mount structure */
|
|
|
|
{
|
|
|
|
int level;
|
|
|
|
uint maxblocks;
|
|
|
|
uint maxleafents;
|
|
|
|
int minleafrecs;
|
|
|
|
int minnoderecs;
|
|
|
|
|
|
|
|
maxleafents = (1LL << XFS_INO_AGINO_BITS(mp)) >>
|
|
|
|
XFS_INODES_PER_CHUNK_LOG;
|
|
|
|
minleafrecs = mp->m_alloc_mnr[0];
|
|
|
|
minnoderecs = mp->m_alloc_mnr[1];
|
|
|
|
maxblocks = (maxleafents + minleafrecs - 1) / minleafrecs;
|
|
|
|
for (level = 1; maxblocks > 1; level++)
|
|
|
|
maxblocks = (maxblocks + minnoderecs - 1) / minnoderecs;
|
|
|
|
mp->m_in_maxlevels = level;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Log specified fields for the ag hdr (inode section)
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
xfs_ialloc_log_agi(
|
|
|
|
xfs_trans_t *tp, /* transaction pointer */
|
|
|
|
xfs_buf_t *bp, /* allocation group header buffer */
|
|
|
|
int fields) /* bitmask of fields to log */
|
|
|
|
{
|
|
|
|
int first; /* first byte number */
|
|
|
|
int last; /* last byte number */
|
|
|
|
static const short offsets[] = { /* field starting offsets */
|
|
|
|
/* keep in sync with bit definitions */
|
|
|
|
offsetof(xfs_agi_t, agi_magicnum),
|
|
|
|
offsetof(xfs_agi_t, agi_versionnum),
|
|
|
|
offsetof(xfs_agi_t, agi_seqno),
|
|
|
|
offsetof(xfs_agi_t, agi_length),
|
|
|
|
offsetof(xfs_agi_t, agi_count),
|
|
|
|
offsetof(xfs_agi_t, agi_root),
|
|
|
|
offsetof(xfs_agi_t, agi_level),
|
|
|
|
offsetof(xfs_agi_t, agi_freecount),
|
|
|
|
offsetof(xfs_agi_t, agi_newino),
|
|
|
|
offsetof(xfs_agi_t, agi_dirino),
|
|
|
|
offsetof(xfs_agi_t, agi_unlinked),
|
|
|
|
sizeof(xfs_agi_t)
|
|
|
|
};
|
|
|
|
#ifdef DEBUG
|
|
|
|
xfs_agi_t *agi; /* allocation group header */
|
|
|
|
|
|
|
|
agi = XFS_BUF_TO_AGI(bp);
|
2005-11-02 04:11:25 +00:00
|
|
|
ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
|
2005-04-16 22:20:36 +00:00
|
|
|
#endif
|
|
|
|
/*
|
|
|
|
* Compute byte offsets for the first and last fields.
|
|
|
|
*/
|
|
|
|
xfs_btree_offsets(fields, offsets, XFS_AGI_NUM_BITS, &first, &last);
|
|
|
|
/*
|
|
|
|
* Log the allocation group inode header buffer.
|
|
|
|
*/
|
|
|
|
xfs_trans_log_buf(tp, bp, first, last);
|
|
|
|
}
|
|
|
|
|
2008-11-28 03:23:37 +00:00
|
|
|
#ifdef DEBUG
|
|
|
|
STATIC void
|
|
|
|
xfs_check_agi_unlinked(
|
|
|
|
struct xfs_agi *agi)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < XFS_AGI_UNLINKED_BUCKETS; i++)
|
|
|
|
ASSERT(agi->agi_unlinked[i]);
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
#define xfs_check_agi_unlinked(agi)
|
|
|
|
#endif
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
/*
|
|
|
|
* Read in the allocation group header (inode allocation section)
|
|
|
|
*/
|
|
|
|
int
|
2008-11-28 03:23:37 +00:00
|
|
|
xfs_read_agi(
|
|
|
|
struct xfs_mount *mp, /* file system mount structure */
|
|
|
|
struct xfs_trans *tp, /* transaction pointer */
|
|
|
|
xfs_agnumber_t agno, /* allocation group number */
|
|
|
|
struct xfs_buf **bpp) /* allocation group hdr buf */
|
2005-04-16 22:20:36 +00:00
|
|
|
{
|
2008-11-28 03:23:37 +00:00
|
|
|
struct xfs_agi *agi; /* allocation group header */
|
|
|
|
int agi_ok; /* agi is consistent */
|
|
|
|
int error;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
ASSERT(agno != NULLAGNUMBER);
|
2008-11-28 03:23:37 +00:00
|
|
|
|
|
|
|
error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
|
2005-04-16 22:20:36 +00:00
|
|
|
XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
|
2008-11-28 03:23:37 +00:00
|
|
|
XFS_FSS_TO_BB(mp, 1), 0, bpp);
|
2005-04-16 22:20:36 +00:00
|
|
|
if (error)
|
|
|
|
return error;
|
2008-11-28 03:23:37 +00:00
|
|
|
|
|
|
|
ASSERT(*bpp && !XFS_BUF_GETERROR(*bpp));
|
|
|
|
agi = XFS_BUF_TO_AGI(*bpp);
|
2005-04-16 22:20:36 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Validate the magic number of the agi block.
|
|
|
|
*/
|
2008-11-28 03:23:37 +00:00
|
|
|
agi_ok = be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC &&
|
|
|
|
XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum)) &&
|
|
|
|
be32_to_cpu(agi->agi_seqno) == agno;
|
2005-04-16 22:20:36 +00:00
|
|
|
if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IALLOC_READ_AGI,
|
|
|
|
XFS_RANDOM_IALLOC_READ_AGI))) {
|
2008-11-28 03:23:37 +00:00
|
|
|
XFS_CORRUPTION_ERROR("xfs_read_agi", XFS_ERRLEVEL_LOW,
|
2005-04-16 22:20:36 +00:00
|
|
|
mp, agi);
|
2008-11-28 03:23:37 +00:00
|
|
|
xfs_trans_brelse(tp, *bpp);
|
2005-04-16 22:20:36 +00:00
|
|
|
return XFS_ERROR(EFSCORRUPTED);
|
|
|
|
}
|
2008-11-28 03:23:37 +00:00
|
|
|
|
|
|
|
XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_AGI, XFS_AGI_REF);
|
|
|
|
|
|
|
|
xfs_check_agi_unlinked(agi);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
|
|
|
xfs_ialloc_read_agi(
|
|
|
|
struct xfs_mount *mp, /* file system mount structure */
|
|
|
|
struct xfs_trans *tp, /* transaction pointer */
|
|
|
|
xfs_agnumber_t agno, /* allocation group number */
|
|
|
|
struct xfs_buf **bpp) /* allocation group hdr buf */
|
|
|
|
{
|
|
|
|
struct xfs_agi *agi; /* allocation group header */
|
|
|
|
struct xfs_perag *pag; /* per allocation group data */
|
|
|
|
int error;
|
|
|
|
|
|
|
|
error = xfs_read_agi(mp, tp, agno, bpp);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
|
|
|
|
agi = XFS_BUF_TO_AGI(*bpp);
|
2010-01-11 11:47:43 +00:00
|
|
|
pag = xfs_perag_get(mp, agno);
|
2005-04-16 22:20:36 +00:00
|
|
|
if (!pag->pagi_init) {
|
2005-11-02 04:11:25 +00:00
|
|
|
pag->pagi_freecount = be32_to_cpu(agi->agi_freecount);
|
[XFS] Lazy Superblock Counters
When we have a couple of hundred transactions on the fly at once, they all
typically modify the on disk superblock in some way.
create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify
free block counts.
When these counts are modified in a transaction, they must eventually lock
the superblock buffer and apply the mods. The buffer then remains locked
until the transaction is committed into the incore log buffer. The result
of this is that with enough transactions on the fly the incore superblock
buffer becomes a bottleneck.
The result of contention on the incore superblock buffer is that
transaction rates fall - the more pressure that is put on the superblock
buffer, the slower things go.
The key to removing the contention is to not require the superblock fields
in question to be locked. We do that by not marking the superblock dirty
in the transaction. IOWs, we modify the incore superblock but do not
modify the cached superblock buffer. In short, we do not log superblock
modifications to critical fields in the superblock on every transaction.
In fact we only do it just before we write the superblock to disk every
sync period or just before unmount.
This creates an interesting problem - if we don't log or write out the
fields in every transaction, then how do the values get recovered after a
crash? the answer is simple - we keep enough duplicate, logged information
in other structures that we can reconstruct the correct count after log
recovery has been performed.
It is the AGF and AGI structures that contain the duplicate information;
after recovery, we walk every AGI and AGF and sum their individual
counters to get the correct value, and we do a transaction into the log to
correct them. An optimisation of this is that if we have a clean unmount
record, we know the value in the superblock is correct, so we can avoid
the summation walk under normal conditions and so mount/recovery times do
not change under normal operation.
One wrinkle that was discovered during development was that the blocks
used in the freespace btrees are never accounted for in the AGF counters.
This was once a valid optimisation to make; when the filesystem is full,
the free space btrees are empty and consume no space. Hence when it
matters, the "accounting" is correct. But that means the when we do the
AGF summations, we would not have a correct count and xfs_check would
complain. Hence a new counter was added to track the number of blocks used
by the free space btrees. This is an *on-disk format change*.
As a result of this, lazy superblock counters are a mkfs option and at the
moment on linux there is no way to convert an old filesystem. This is
possible - xfs_db can be used to twiddle the right bits and then
xfs_repair will do the format conversion for you. Similarly, you can
convert backwards as well. At some point we'll add functionality to
xfs_admin to do the bit twiddling easily....
SGI-PV: 964999
SGI-Modid: xfs-linux-melb:xfs-kern:28652a
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 05:26:31 +00:00
|
|
|
pag->pagi_count = be32_to_cpu(agi->agi_count);
|
2005-04-16 22:20:36 +00:00
|
|
|
pag->pagi_init = 1;
|
|
|
|
}
|
|
|
|
|
2008-11-28 03:23:37 +00:00
|
|
|
/*
|
|
|
|
* It's possible for these to be out of sync if
|
|
|
|
* we are in the middle of a forced shutdown.
|
|
|
|
*/
|
|
|
|
ASSERT(pag->pagi_freecount == be32_to_cpu(agi->agi_freecount) ||
|
|
|
|
XFS_FORCED_SHUTDOWN(mp));
|
2010-01-11 11:47:43 +00:00
|
|
|
xfs_perag_put(pag);
|
2005-04-16 22:20:36 +00:00
|
|
|
return 0;
|
|
|
|
}
|
[XFS] Lazy Superblock Counters
When we have a couple of hundred transactions on the fly at once, they all
typically modify the on disk superblock in some way.
create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify
free block counts.
When these counts are modified in a transaction, they must eventually lock
the superblock buffer and apply the mods. The buffer then remains locked
until the transaction is committed into the incore log buffer. The result
of this is that with enough transactions on the fly the incore superblock
buffer becomes a bottleneck.
The result of contention on the incore superblock buffer is that
transaction rates fall - the more pressure that is put on the superblock
buffer, the slower things go.
The key to removing the contention is to not require the superblock fields
in question to be locked. We do that by not marking the superblock dirty
in the transaction. IOWs, we modify the incore superblock but do not
modify the cached superblock buffer. In short, we do not log superblock
modifications to critical fields in the superblock on every transaction.
In fact we only do it just before we write the superblock to disk every
sync period or just before unmount.
This creates an interesting problem - if we don't log or write out the
fields in every transaction, then how do the values get recovered after a
crash? the answer is simple - we keep enough duplicate, logged information
in other structures that we can reconstruct the correct count after log
recovery has been performed.
It is the AGF and AGI structures that contain the duplicate information;
after recovery, we walk every AGI and AGF and sum their individual
counters to get the correct value, and we do a transaction into the log to
correct them. An optimisation of this is that if we have a clean unmount
record, we know the value in the superblock is correct, so we can avoid
the summation walk under normal conditions and so mount/recovery times do
not change under normal operation.
One wrinkle that was discovered during development was that the blocks
used in the freespace btrees are never accounted for in the AGF counters.
This was once a valid optimisation to make; when the filesystem is full,
the free space btrees are empty and consume no space. Hence when it
matters, the "accounting" is correct. But that means the when we do the
AGF summations, we would not have a correct count and xfs_check would
complain. Hence a new counter was added to track the number of blocks used
by the free space btrees. This is an *on-disk format change*.
As a result of this, lazy superblock counters are a mkfs option and at the
moment on linux there is no way to convert an old filesystem. This is
possible - xfs_db can be used to twiddle the right bits and then
xfs_repair will do the format conversion for you. Similarly, you can
convert backwards as well. At some point we'll add functionality to
xfs_admin to do the bit twiddling easily....
SGI-PV: 964999
SGI-Modid: xfs-linux-melb:xfs-kern:28652a
Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 05:26:31 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Read in the agi to initialise the per-ag data in the mount structure
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
xfs_ialloc_pagi_init(
|
|
|
|
xfs_mount_t *mp, /* file system mount structure */
|
|
|
|
xfs_trans_t *tp, /* transaction pointer */
|
|
|
|
xfs_agnumber_t agno) /* allocation group number */
|
|
|
|
{
|
|
|
|
xfs_buf_t *bp = NULL;
|
|
|
|
int error;
|
|
|
|
|
|
|
|
error = xfs_ialloc_read_agi(mp, tp, agno, &bp);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
if (bp)
|
|
|
|
xfs_trans_brelse(tp, bp);
|
|
|
|
return 0;
|
|
|
|
}
|