linux/fs/xfs/xfs_dir2_sf.c

1180 lines
34 KiB
C
Raw Normal View History

/*
* Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation.
*
* 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.
*
* 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
*/
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_mount.h"
#include "xfs_da_format.h"
#include "xfs_da_btree.h"
#include "xfs_inode.h"
#include "xfs_trans.h"
#include "xfs_inode_item.h"
#include "xfs_error.h"
#include "xfs_dir2.h"
#include "xfs_dir2_priv.h"
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
#include "xfs_trace.h"
#include "xfs_dinode.h"
/*
* Prototypes for internal functions.
*/
static void xfs_dir2_sf_addname_easy(xfs_da_args_t *args,
xfs_dir2_sf_entry_t *sfep,
xfs_dir2_data_aoff_t offset,
int new_isize);
static void xfs_dir2_sf_addname_hard(xfs_da_args_t *args, int objchange,
int new_isize);
static int xfs_dir2_sf_addname_pick(xfs_da_args_t *args, int objchange,
xfs_dir2_sf_entry_t **sfepp,
xfs_dir2_data_aoff_t *offsetp);
#ifdef DEBUG
static void xfs_dir2_sf_check(xfs_da_args_t *args);
#else
#define xfs_dir2_sf_check(args)
#endif /* DEBUG */
#if XFS_BIG_INUMS
static void xfs_dir2_sf_toino4(xfs_da_args_t *args);
static void xfs_dir2_sf_toino8(xfs_da_args_t *args);
#endif /* XFS_BIG_INUMS */
/*
* Given a block directory (dp/block), calculate its size as a shortform (sf)
* directory and a header for the sf directory, if it will fit it the
* space currently present in the inode. If it won't fit, the output
* size is too big (but not accurate).
*/
int /* size for sf form */
xfs_dir2_block_sfsize(
xfs_inode_t *dp, /* incore inode pointer */
xfs_dir2_data_hdr_t *hdr, /* block directory data */
xfs_dir2_sf_hdr_t *sfhp) /* output: header for sf form */
{
xfs_dir2_dataptr_t addr; /* data entry address */
xfs_dir2_leaf_entry_t *blp; /* leaf area of the block */
xfs_dir2_block_tail_t *btp; /* tail area of the block */
int count; /* shortform entry count */
xfs_dir2_data_entry_t *dep; /* data entry in the block */
int i; /* block entry index */
int i8count; /* count of big-inode entries */
int isdot; /* entry is "." */
int isdotdot; /* entry is ".." */
xfs_mount_t *mp; /* mount structure pointer */
int namelen; /* total name bytes */
xfs_ino_t parent = 0; /* parent inode number */
int size=0; /* total computed size */
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
int has_ftype;
mp = dp->i_mount;
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
/*
* if there is a filetype field, add the extra byte to the namelen
* for each entry that we see.
*/
has_ftype = xfs_sb_version_hasftype(&mp->m_sb) ? 1 : 0;
count = i8count = namelen = 0;
btp = xfs_dir2_block_tail_p(mp, hdr);
blp = xfs_dir2_block_leaf_p(btp);
/*
* Iterate over the block's data entries by using the leaf pointers.
*/
for (i = 0; i < be32_to_cpu(btp->count); i++) {
if ((addr = be32_to_cpu(blp[i].address)) == XFS_DIR2_NULL_DATAPTR)
continue;
/*
* Calculate the pointer to the entry at hand.
*/
dep = (xfs_dir2_data_entry_t *)
((char *)hdr + xfs_dir2_dataptr_to_off(mp, addr));
/*
* Detect . and .., so we can special-case them.
* . is not included in sf directories.
* .. is included by just the parent inode number.
*/
isdot = dep->namelen == 1 && dep->name[0] == '.';
isdotdot =
dep->namelen == 2 &&
dep->name[0] == '.' && dep->name[1] == '.';
#if XFS_BIG_INUMS
if (!isdot)
i8count += be64_to_cpu(dep->inumber) > XFS_DIR2_MAX_SHORT_INUM;
#endif
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
/* take into account the file type field */
if (!isdot && !isdotdot) {
count++;
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
namelen += dep->namelen + has_ftype;
} else if (isdotdot)
parent = be64_to_cpu(dep->inumber);
/*
* Calculate the new size, see if we should give up yet.
*/
size = xfs_dir2_sf_hdr_size(i8count) + /* header */
count + /* namelen */
count * (uint)sizeof(xfs_dir2_sf_off_t) + /* offset */
namelen + /* name */
(i8count ? /* inumber */
(uint)sizeof(xfs_dir2_ino8_t) * count :
(uint)sizeof(xfs_dir2_ino4_t) * count);
if (size > XFS_IFORK_DSIZE(dp))
return size; /* size value is a failure */
}
/*
* Create the output header, if it worked.
*/
sfhp->count = count;
sfhp->i8count = i8count;
dp->d_ops->sf_put_parent_ino(sfhp, parent);
return size;
}
/*
* Convert a block format directory to shortform.
* Caller has already checked that it will fit, and built us a header.
*/
int /* error */
xfs_dir2_block_to_sf(
xfs_da_args_t *args, /* operation arguments */
struct xfs_buf *bp,
int size, /* shortform directory size */
xfs_dir2_sf_hdr_t *sfhp) /* shortform directory hdr */
{
xfs_dir2_data_hdr_t *hdr; /* block header */
xfs_dir2_block_tail_t *btp; /* block tail pointer */
xfs_dir2_data_entry_t *dep; /* data entry pointer */
xfs_inode_t *dp; /* incore directory inode */
xfs_dir2_data_unused_t *dup; /* unused data pointer */
char *endptr; /* end of data entries */
int error; /* error return value */
int logflags; /* inode logging flags */
xfs_mount_t *mp; /* filesystem mount point */
char *ptr; /* current data pointer */
xfs_dir2_sf_entry_t *sfep; /* shortform entry */
xfs_dir2_sf_hdr_t *sfp; /* shortform directory header */
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dir2_block_to_sf(args);
dp = args->dp;
mp = dp->i_mount;
/*
* Make a copy of the block data, so we can shrink the inode
* and add local data.
*/
hdr = kmem_alloc(mp->m_dirblksize, KM_SLEEP);
memcpy(hdr, bp->b_addr, mp->m_dirblksize);
logflags = XFS_ILOG_CORE;
if ((error = xfs_dir2_shrink_inode(args, mp->m_dirdatablk, bp))) {
ASSERT(error != ENOSPC);
goto out;
}
/*
* The buffer is now unconditionally gone, whether
* xfs_dir2_shrink_inode worked or not.
*
* Convert the inode to local format.
*/
dp->i_df.if_flags &= ~XFS_IFEXTENTS;
dp->i_df.if_flags |= XFS_IFINLINE;
dp->i_d.di_format = XFS_DINODE_FMT_LOCAL;
ASSERT(dp->i_df.if_bytes == 0);
xfs_idata_realloc(dp, size, XFS_DATA_FORK);
logflags |= XFS_ILOG_DDATA;
/*
* Copy the header into the newly allocate local space.
*/
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
memcpy(sfp, sfhp, xfs_dir2_sf_hdr_size(sfhp->i8count));
dp->i_d.di_size = size;
/*
* Set up to loop over the block's entries.
*/
btp = xfs_dir2_block_tail_p(mp, hdr);
ptr = (char *)dp->d_ops->data_entry_p(hdr);
endptr = (char *)xfs_dir2_block_leaf_p(btp);
sfep = xfs_dir2_sf_firstentry(sfp);
/*
* Loop over the active and unused entries.
* Stop when we reach the leaf/tail portion of the block.
*/
while (ptr < endptr) {
/*
* If it's unused, just skip over it.
*/
dup = (xfs_dir2_data_unused_t *)ptr;
if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
ptr += be16_to_cpu(dup->length);
continue;
}
dep = (xfs_dir2_data_entry_t *)ptr;
/*
* Skip .
*/
if (dep->namelen == 1 && dep->name[0] == '.')
ASSERT(be64_to_cpu(dep->inumber) == dp->i_ino);
/*
* Skip .., but make sure the inode number is right.
*/
else if (dep->namelen == 2 &&
dep->name[0] == '.' && dep->name[1] == '.')
ASSERT(be64_to_cpu(dep->inumber) ==
dp->d_ops->sf_get_parent_ino(sfp));
/*
* Normal entry, copy it into shortform.
*/
else {
sfep->namelen = dep->namelen;
xfs_dir2_sf_put_offset(sfep,
(xfs_dir2_data_aoff_t)
((char *)dep - (char *)hdr));
memcpy(sfep->name, dep->name, dep->namelen);
dp->d_ops->sf_put_ino(sfp, sfep,
be64_to_cpu(dep->inumber));
dp->d_ops->sf_put_ftype(sfep,
dp->d_ops->data_get_ftype(dep));
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
sfep = dp->d_ops->sf_nextentry(sfp, sfep);
}
ptr += dp->d_ops->data_entsize(dep->namelen);
}
ASSERT((char *)sfep - (char *)sfp == size);
xfs_dir2_sf_check(args);
out:
xfs_trans_log_inode(args->trans, dp, logflags);
kmem_free(hdr);
return error;
}
/*
* Add a name to a shortform directory.
* There are two algorithms, "easy" and "hard" which we decide on
* before changing anything.
* Convert to block form if necessary, if the new entry won't fit.
*/
int /* error */
xfs_dir2_sf_addname(
xfs_da_args_t *args) /* operation arguments */
{
int add_entsize; /* size of the new entry */
xfs_inode_t *dp; /* incore directory inode */
int error; /* error return value */
int incr_isize; /* total change in size */
int new_isize; /* di_size after adding name */
int objchange; /* changing to 8-byte inodes */
xfs_dir2_data_aoff_t offset = 0; /* offset for new entry */
int old_isize; /* di_size before adding name */
int pick; /* which algorithm to use */
xfs_dir2_sf_hdr_t *sfp; /* shortform structure */
xfs_dir2_sf_entry_t *sfep = NULL; /* shortform entry */
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dir2_sf_addname(args);
ASSERT(xfs_dir2_sf_lookup(args) == ENOENT);
dp = args->dp;
ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
/*
* Make sure the shortform value has some of its header.
*/
if (dp->i_d.di_size < offsetof(xfs_dir2_sf_hdr_t, parent)) {
ASSERT(XFS_FORCED_SHUTDOWN(dp->i_mount));
return XFS_ERROR(EIO);
}
ASSERT(dp->i_df.if_bytes == dp->i_d.di_size);
ASSERT(dp->i_df.if_u1.if_data != NULL);
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
ASSERT(dp->i_d.di_size >= xfs_dir2_sf_hdr_size(sfp->i8count));
/*
* Compute entry (and change in) size.
*/
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
add_entsize = dp->d_ops->sf_entsize(sfp, args->namelen);
incr_isize = add_entsize;
objchange = 0;
#if XFS_BIG_INUMS
/*
* Do we have to change to 8 byte inodes?
*/
if (args->inumber > XFS_DIR2_MAX_SHORT_INUM && sfp->i8count == 0) {
/*
* Yes, adjust the entry size and the total size.
*/
add_entsize +=
(uint)sizeof(xfs_dir2_ino8_t) -
(uint)sizeof(xfs_dir2_ino4_t);
incr_isize +=
(sfp->count + 2) *
((uint)sizeof(xfs_dir2_ino8_t) -
(uint)sizeof(xfs_dir2_ino4_t));
objchange = 1;
}
#endif
old_isize = (int)dp->i_d.di_size;
new_isize = old_isize + incr_isize;
/*
* Won't fit as shortform any more (due to size),
* or the pick routine says it won't (due to offset values).
*/
if (new_isize > XFS_IFORK_DSIZE(dp) ||
(pick =
xfs_dir2_sf_addname_pick(args, objchange, &sfep, &offset)) == 0) {
/*
* Just checking or no space reservation, it doesn't fit.
*/
if ((args->op_flags & XFS_DA_OP_JUSTCHECK) || args->total == 0)
return XFS_ERROR(ENOSPC);
/*
* Convert to block form then add the name.
*/
error = xfs_dir2_sf_to_block(args);
if (error)
return error;
return xfs_dir2_block_addname(args);
}
/*
* Just checking, it fits.
*/
if (args->op_flags & XFS_DA_OP_JUSTCHECK)
return 0;
/*
* Do it the easy way - just add it at the end.
*/
if (pick == 1)
xfs_dir2_sf_addname_easy(args, sfep, offset, new_isize);
/*
* Do it the hard way - look for a place to insert the new entry.
* Convert to 8 byte inode numbers first if necessary.
*/
else {
ASSERT(pick == 2);
#if XFS_BIG_INUMS
if (objchange)
xfs_dir2_sf_toino8(args);
#endif
xfs_dir2_sf_addname_hard(args, objchange, new_isize);
}
xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
return 0;
}
/*
* Add the new entry the "easy" way.
* This is copying the old directory and adding the new entry at the end.
* Since it's sorted by "offset" we need room after the last offset
* that's already there, and then room to convert to a block directory.
* This is already checked by the pick routine.
*/
static void
xfs_dir2_sf_addname_easy(
xfs_da_args_t *args, /* operation arguments */
xfs_dir2_sf_entry_t *sfep, /* pointer to new entry */
xfs_dir2_data_aoff_t offset, /* offset to use for new ent */
int new_isize) /* new directory size */
{
int byteoff; /* byte offset in sf dir */
xfs_inode_t *dp; /* incore directory inode */
xfs_dir2_sf_hdr_t *sfp; /* shortform structure */
dp = args->dp;
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
byteoff = (int)((char *)sfep - (char *)sfp);
/*
* Grow the in-inode space.
*/
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
xfs_idata_realloc(dp, dp->d_ops->sf_entsize(sfp, args->namelen),
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
XFS_DATA_FORK);
/*
* Need to set up again due to realloc of the inode data.
*/
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
sfep = (xfs_dir2_sf_entry_t *)((char *)sfp + byteoff);
/*
* Fill in the new entry.
*/
sfep->namelen = args->namelen;
xfs_dir2_sf_put_offset(sfep, offset);
memcpy(sfep->name, args->name, sfep->namelen);
dp->d_ops->sf_put_ino(sfp, sfep, args->inumber);
dp->d_ops->sf_put_ftype(sfep, args->filetype);
/*
* Update the header and inode.
*/
sfp->count++;
#if XFS_BIG_INUMS
if (args->inumber > XFS_DIR2_MAX_SHORT_INUM)
sfp->i8count++;
#endif
dp->i_d.di_size = new_isize;
xfs_dir2_sf_check(args);
}
/*
* Add the new entry the "hard" way.
* The caller has already converted to 8 byte inode numbers if necessary,
* in which case we need to leave the i8count at 1.
* Find a hole that the new entry will fit into, and copy
* the first part of the entries, the new entry, and the last part of
* the entries.
*/
/* ARGSUSED */
static void
xfs_dir2_sf_addname_hard(
xfs_da_args_t *args, /* operation arguments */
int objchange, /* changing inode number size */
int new_isize) /* new directory size */
{
int add_datasize; /* data size need for new ent */
char *buf; /* buffer for old */
xfs_inode_t *dp; /* incore directory inode */
int eof; /* reached end of old dir */
int nbytes; /* temp for byte copies */
xfs_dir2_data_aoff_t new_offset; /* next offset value */
xfs_dir2_data_aoff_t offset; /* current offset value */
int old_isize; /* previous di_size */
xfs_dir2_sf_entry_t *oldsfep; /* entry in original dir */
xfs_dir2_sf_hdr_t *oldsfp; /* original shortform dir */
xfs_dir2_sf_entry_t *sfep; /* entry in new dir */
xfs_dir2_sf_hdr_t *sfp; /* new shortform dir */
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
struct xfs_mount *mp;
/*
* Copy the old directory to the stack buffer.
*/
dp = args->dp;
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
mp = dp->i_mount;
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
old_isize = (int)dp->i_d.di_size;
buf = kmem_alloc(old_isize, KM_SLEEP);
oldsfp = (xfs_dir2_sf_hdr_t *)buf;
memcpy(oldsfp, sfp, old_isize);
/*
* Loop over the old directory finding the place we're going
* to insert the new entry.
* If it's going to end up at the end then oldsfep will point there.
*/
for (offset = dp->d_ops->data_first_offset(),
oldsfep = xfs_dir2_sf_firstentry(oldsfp),
add_datasize = dp->d_ops->data_entsize(args->namelen),
eof = (char *)oldsfep == &buf[old_isize];
!eof;
offset = new_offset + dp->d_ops->data_entsize(oldsfep->namelen),
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
oldsfep = dp->d_ops->sf_nextentry(oldsfp, oldsfep),
eof = (char *)oldsfep == &buf[old_isize]) {
new_offset = xfs_dir2_sf_get_offset(oldsfep);
if (offset + add_datasize <= new_offset)
break;
}
/*
* Get rid of the old directory, then allocate space for
* the new one. We do this so xfs_idata_realloc won't copy
* the data.
*/
xfs_idata_realloc(dp, -old_isize, XFS_DATA_FORK);
xfs_idata_realloc(dp, new_isize, XFS_DATA_FORK);
/*
* Reset the pointer since the buffer was reallocated.
*/
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
/*
* Copy the first part of the directory, including the header.
*/
nbytes = (int)((char *)oldsfep - (char *)oldsfp);
memcpy(sfp, oldsfp, nbytes);
sfep = (xfs_dir2_sf_entry_t *)((char *)sfp + nbytes);
/*
* Fill in the new entry, and update the header counts.
*/
sfep->namelen = args->namelen;
xfs_dir2_sf_put_offset(sfep, offset);
memcpy(sfep->name, args->name, sfep->namelen);
dp->d_ops->sf_put_ino(sfp, sfep, args->inumber);
dp->d_ops->sf_put_ftype(sfep, args->filetype);
sfp->count++;
#if XFS_BIG_INUMS
if (args->inumber > XFS_DIR2_MAX_SHORT_INUM && !objchange)
sfp->i8count++;
#endif
/*
* If there's more left to copy, do that.
*/
if (!eof) {
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
sfep = dp->d_ops->sf_nextentry(sfp, sfep);
memcpy(sfep, oldsfep, old_isize - nbytes);
}
kmem_free(buf);
dp->i_d.di_size = new_isize;
xfs_dir2_sf_check(args);
}
/*
* Decide if the new entry will fit at all.
* If it will fit, pick between adding the new entry to the end (easy)
* or somewhere else (hard).
* Return 0 (won't fit), 1 (easy), 2 (hard).
*/
/*ARGSUSED*/
static int /* pick result */
xfs_dir2_sf_addname_pick(
xfs_da_args_t *args, /* operation arguments */
int objchange, /* inode # size changes */
xfs_dir2_sf_entry_t **sfepp, /* out(1): new entry ptr */
xfs_dir2_data_aoff_t *offsetp) /* out(1): new offset */
{
xfs_inode_t *dp; /* incore directory inode */
int holefit; /* found hole it will fit in */
int i; /* entry number */
xfs_mount_t *mp; /* filesystem mount point */
xfs_dir2_data_aoff_t offset; /* data block offset */
xfs_dir2_sf_entry_t *sfep; /* shortform entry */
xfs_dir2_sf_hdr_t *sfp; /* shortform structure */
int size; /* entry's data size */
int used; /* data bytes used */
dp = args->dp;
mp = dp->i_mount;
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
size = dp->d_ops->data_entsize(args->namelen);
offset = dp->d_ops->data_first_offset();
sfep = xfs_dir2_sf_firstentry(sfp);
holefit = 0;
/*
* Loop over sf entries.
* Keep track of data offset and whether we've seen a place
* to insert the new entry.
*/
for (i = 0; i < sfp->count; i++) {
if (!holefit)
holefit = offset + size <= xfs_dir2_sf_get_offset(sfep);
offset = xfs_dir2_sf_get_offset(sfep) +
dp->d_ops->data_entsize(sfep->namelen);
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
sfep = dp->d_ops->sf_nextentry(sfp, sfep);
}
/*
* Calculate data bytes used excluding the new entry, if this
* was a data block (block form directory).
*/
used = offset +
(sfp->count + 3) * (uint)sizeof(xfs_dir2_leaf_entry_t) +
(uint)sizeof(xfs_dir2_block_tail_t);
/*
* If it won't fit in a block form then we can't insert it,
* we'll go back, convert to block, then try the insert and convert
* to leaf.
*/
if (used + (holefit ? 0 : size) > mp->m_dirblksize)
return 0;
/*
* If changing the inode number size, do it the hard way.
*/
#if XFS_BIG_INUMS
if (objchange) {
return 2;
}
#else
ASSERT(objchange == 0);
#endif
/*
* If it won't fit at the end then do it the hard way (use the hole).
*/
if (used + size > mp->m_dirblksize)
return 2;
/*
* Do it the easy way.
*/
*sfepp = sfep;
*offsetp = offset;
return 1;
}
#ifdef DEBUG
/*
* Check consistency of shortform directory, assert if bad.
*/
static void
xfs_dir2_sf_check(
xfs_da_args_t *args) /* operation arguments */
{
xfs_inode_t *dp; /* incore directory inode */
int i; /* entry number */
int i8count; /* number of big inode#s */
xfs_ino_t ino; /* entry inode number */
int offset; /* data offset */
xfs_dir2_sf_entry_t *sfep; /* shortform dir entry */
xfs_dir2_sf_hdr_t *sfp; /* shortform structure */
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
struct xfs_mount *mp;
dp = args->dp;
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
mp = dp->i_mount;
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
offset = dp->d_ops->data_first_offset();
ino = dp->d_ops->sf_get_parent_ino(sfp);
i8count = ino > XFS_DIR2_MAX_SHORT_INUM;
for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp);
i < sfp->count;
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
i++, sfep = dp->d_ops->sf_nextentry(sfp, sfep)) {
ASSERT(xfs_dir2_sf_get_offset(sfep) >= offset);
ino = dp->d_ops->sf_get_ino(sfp, sfep);
i8count += ino > XFS_DIR2_MAX_SHORT_INUM;
offset =
xfs_dir2_sf_get_offset(sfep) +
dp->d_ops->data_entsize(sfep->namelen);
ASSERT(dp->d_ops->sf_get_ftype(sfep) < XFS_DIR3_FT_MAX);
}
ASSERT(i8count == sfp->i8count);
ASSERT(XFS_BIG_INUMS || i8count == 0);
ASSERT((char *)sfep - (char *)sfp == dp->i_d.di_size);
ASSERT(offset +
(sfp->count + 2) * (uint)sizeof(xfs_dir2_leaf_entry_t) +
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
(uint)sizeof(xfs_dir2_block_tail_t) <= mp->m_dirblksize);
}
#endif /* DEBUG */
/*
* Create a new (shortform) directory.
*/
int /* error, always 0 */
xfs_dir2_sf_create(
xfs_da_args_t *args, /* operation arguments */
xfs_ino_t pino) /* parent inode number */
{
xfs_inode_t *dp; /* incore directory inode */
int i8count; /* parent inode is an 8-byte number */
xfs_dir2_sf_hdr_t *sfp; /* shortform structure */
int size; /* directory size */
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dir2_sf_create(args);
dp = args->dp;
ASSERT(dp != NULL);
ASSERT(dp->i_d.di_size == 0);
/*
* If it's currently a zero-length extent file,
* convert it to local format.
*/
if (dp->i_d.di_format == XFS_DINODE_FMT_EXTENTS) {
dp->i_df.if_flags &= ~XFS_IFEXTENTS; /* just in case */
dp->i_d.di_format = XFS_DINODE_FMT_LOCAL;
xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE);
dp->i_df.if_flags |= XFS_IFINLINE;
}
ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
ASSERT(dp->i_df.if_bytes == 0);
i8count = pino > XFS_DIR2_MAX_SHORT_INUM;
size = xfs_dir2_sf_hdr_size(i8count);
/*
* Make a buffer for the data.
*/
xfs_idata_realloc(dp, size, XFS_DATA_FORK);
/*
* Fill in the header,
*/
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
sfp->i8count = i8count;
/*
* Now can put in the inode number, since i8count is set.
*/
dp->d_ops->sf_put_parent_ino(sfp, pino);
sfp->count = 0;
dp->i_d.di_size = size;
xfs_dir2_sf_check(args);
xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
return 0;
}
/*
* Lookup an entry in a shortform directory.
* Returns EEXIST if found, ENOENT if not found.
*/
int /* error */
xfs_dir2_sf_lookup(
xfs_da_args_t *args) /* operation arguments */
{
xfs_inode_t *dp; /* incore directory inode */
int i; /* entry index */
int error;
xfs_dir2_sf_entry_t *sfep; /* shortform directory entry */
xfs_dir2_sf_hdr_t *sfp; /* shortform structure */
enum xfs_dacmp cmp; /* comparison result */
xfs_dir2_sf_entry_t *ci_sfep; /* case-insens. entry */
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dir2_sf_lookup(args);
xfs_dir2_sf_check(args);
dp = args->dp;
ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
/*
* Bail out if the directory is way too short.
*/
if (dp->i_d.di_size < offsetof(xfs_dir2_sf_hdr_t, parent)) {
ASSERT(XFS_FORCED_SHUTDOWN(dp->i_mount));
return XFS_ERROR(EIO);
}
ASSERT(dp->i_df.if_bytes == dp->i_d.di_size);
ASSERT(dp->i_df.if_u1.if_data != NULL);
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
ASSERT(dp->i_d.di_size >= xfs_dir2_sf_hdr_size(sfp->i8count));
/*
* Special case for .
*/
if (args->namelen == 1 && args->name[0] == '.') {
args->inumber = dp->i_ino;
args->cmpresult = XFS_CMP_EXACT;
args->filetype = XFS_DIR3_FT_DIR;
return XFS_ERROR(EEXIST);
}
/*
* Special case for ..
*/
if (args->namelen == 2 &&
args->name[0] == '.' && args->name[1] == '.') {
args->inumber = dp->d_ops->sf_get_parent_ino(sfp);
args->cmpresult = XFS_CMP_EXACT;
args->filetype = XFS_DIR3_FT_DIR;
return XFS_ERROR(EEXIST);
}
/*
* Loop over all the entries trying to match ours.
*/
ci_sfep = NULL;
for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp); i < sfp->count;
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
i++, sfep = dp->d_ops->sf_nextentry(sfp, sfep)) {
/*
* Compare name and if it's an exact match, return the inode
* number. If it's the first case-insensitive match, store the
* inode number and continue looking for an exact match.
*/
cmp = dp->i_mount->m_dirnameops->compname(args, sfep->name,
sfep->namelen);
if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
args->cmpresult = cmp;
args->inumber = dp->d_ops->sf_get_ino(sfp, sfep);
args->filetype = dp->d_ops->sf_get_ftype(sfep);
if (cmp == XFS_CMP_EXACT)
return XFS_ERROR(EEXIST);
ci_sfep = sfep;
}
}
ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
/*
* Here, we can only be doing a lookup (not a rename or replace).
* If a case-insensitive match was not found, return ENOENT.
*/
if (!ci_sfep)
return XFS_ERROR(ENOENT);
/* otherwise process the CI match as required by the caller */
error = xfs_dir_cilookup_result(args, ci_sfep->name, ci_sfep->namelen);
return XFS_ERROR(error);
}
/*
* Remove an entry from a shortform directory.
*/
int /* error */
xfs_dir2_sf_removename(
xfs_da_args_t *args)
{
int byteoff; /* offset of removed entry */
xfs_inode_t *dp; /* incore directory inode */
int entsize; /* this entry's size */
int i; /* shortform entry index */
int newsize; /* new inode size */
int oldsize; /* old inode size */
xfs_dir2_sf_entry_t *sfep; /* shortform directory entry */
xfs_dir2_sf_hdr_t *sfp; /* shortform structure */
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dir2_sf_removename(args);
dp = args->dp;
ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
oldsize = (int)dp->i_d.di_size;
/*
* Bail out if the directory is way too short.
*/
if (oldsize < offsetof(xfs_dir2_sf_hdr_t, parent)) {
ASSERT(XFS_FORCED_SHUTDOWN(dp->i_mount));
return XFS_ERROR(EIO);
}
ASSERT(dp->i_df.if_bytes == oldsize);
ASSERT(dp->i_df.if_u1.if_data != NULL);
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
ASSERT(oldsize >= xfs_dir2_sf_hdr_size(sfp->i8count));
/*
* Loop over the old directory entries.
* Find the one we're deleting.
*/
for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp); i < sfp->count;
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
i++, sfep = dp->d_ops->sf_nextentry(sfp, sfep)) {
if (xfs_da_compname(args, sfep->name, sfep->namelen) ==
XFS_CMP_EXACT) {
ASSERT(dp->d_ops->sf_get_ino(sfp, sfep) ==
args->inumber);
break;
}
}
/*
* Didn't find it.
*/
if (i == sfp->count)
return XFS_ERROR(ENOENT);
/*
* Calculate sizes.
*/
byteoff = (int)((char *)sfep - (char *)sfp);
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
entsize = dp->d_ops->sf_entsize(sfp, args->namelen);
newsize = oldsize - entsize;
/*
* Copy the part if any after the removed entry, sliding it down.
*/
if (byteoff + entsize < oldsize)
memmove((char *)sfp + byteoff, (char *)sfp + byteoff + entsize,
oldsize - (byteoff + entsize));
/*
* Fix up the header and file size.
*/
sfp->count--;
dp->i_d.di_size = newsize;
/*
* Reallocate, making it smaller.
*/
xfs_idata_realloc(dp, newsize - oldsize, XFS_DATA_FORK);
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
#if XFS_BIG_INUMS
/*
* Are we changing inode number size?
*/
if (args->inumber > XFS_DIR2_MAX_SHORT_INUM) {
if (sfp->i8count == 1)
xfs_dir2_sf_toino4(args);
else
sfp->i8count--;
}
#endif
xfs_dir2_sf_check(args);
xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
return 0;
}
/*
* Replace the inode number of an entry in a shortform directory.
*/
int /* error */
xfs_dir2_sf_replace(
xfs_da_args_t *args) /* operation arguments */
{
xfs_inode_t *dp; /* incore directory inode */
int i; /* entry index */
#if XFS_BIG_INUMS || defined(DEBUG)
xfs_ino_t ino=0; /* entry old inode number */
#endif
#if XFS_BIG_INUMS
int i8elevated; /* sf_toino8 set i8count=1 */
#endif
xfs_dir2_sf_entry_t *sfep; /* shortform directory entry */
xfs_dir2_sf_hdr_t *sfp; /* shortform structure */
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dir2_sf_replace(args);
dp = args->dp;
ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
/*
* Bail out if the shortform directory is way too small.
*/
if (dp->i_d.di_size < offsetof(xfs_dir2_sf_hdr_t, parent)) {
ASSERT(XFS_FORCED_SHUTDOWN(dp->i_mount));
return XFS_ERROR(EIO);
}
ASSERT(dp->i_df.if_bytes == dp->i_d.di_size);
ASSERT(dp->i_df.if_u1.if_data != NULL);
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
ASSERT(dp->i_d.di_size >= xfs_dir2_sf_hdr_size(sfp->i8count));
#if XFS_BIG_INUMS
/*
* New inode number is large, and need to convert to 8-byte inodes.
*/
if (args->inumber > XFS_DIR2_MAX_SHORT_INUM && sfp->i8count == 0) {
int error; /* error return value */
int newsize; /* new inode size */
newsize =
dp->i_df.if_bytes +
(sfp->count + 1) *
((uint)sizeof(xfs_dir2_ino8_t) -
(uint)sizeof(xfs_dir2_ino4_t));
/*
* Won't fit as shortform, convert to block then do replace.
*/
if (newsize > XFS_IFORK_DSIZE(dp)) {
error = xfs_dir2_sf_to_block(args);
if (error) {
return error;
}
return xfs_dir2_block_replace(args);
}
/*
* Still fits, convert to 8-byte now.
*/
xfs_dir2_sf_toino8(args);
i8elevated = 1;
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
} else
i8elevated = 0;
#endif
ASSERT(args->namelen != 1 || args->name[0] != '.');
/*
* Replace ..'s entry.
*/
if (args->namelen == 2 &&
args->name[0] == '.' && args->name[1] == '.') {
#if XFS_BIG_INUMS || defined(DEBUG)
ino = dp->d_ops->sf_get_parent_ino(sfp);
ASSERT(args->inumber != ino);
#endif
dp->d_ops->sf_put_parent_ino(sfp, args->inumber);
}
/*
* Normal entry, look for the name.
*/
else {
xfs: Add read-only support for dirent filetype field Add support for the file type field in directory entries so that readdir can return the type of the inode the dirent points to to userspace without first having to read the inode off disk. The encoding of the type field is a single byte that is added to the end of the directory entry name length. For all intents and purposes, it appends a "hidden" byte to the name field which contains the type information. As the directory entry is already of dynamic size, helpers are already required to access and decode the direct entry structures. Hence the relevent extraction and iteration helpers are updated to understand the hidden byte. Helpers for reading and writing the filetype field from the directory entries are also added. Only the read helpers are used by this patch. It also adds all the code necessary to read the type information out of the dirents on disk. Further we add the superblock feature bit and helpers to indicate that we understand the on-disk format change. This is not a compatible change - existing kernels cannot read the new format successfully - so an incompatible feature flag is added. We don't yet allow filesystems to mount with this flag yet - that will be added once write support is added. Finally, the code to take the type from the VFS, convert it to an XFS on-disk type and put it into the xfs_name structures passed around is added, but the directory code does not use this field yet. That will be in the next patch. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-08-12 10:50:09 +00:00
for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp); i < sfp->count;
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
i++, sfep = dp->d_ops->sf_nextentry(sfp, sfep)) {
if (xfs_da_compname(args, sfep->name, sfep->namelen) ==
XFS_CMP_EXACT) {
#if XFS_BIG_INUMS || defined(DEBUG)
ino = dp->d_ops->sf_get_ino(sfp, sfep);
ASSERT(args->inumber != ino);
#endif
dp->d_ops->sf_put_ino(sfp, sfep, args->inumber);
dp->d_ops->sf_put_ftype(sfep, args->filetype);
break;
}
}
/*
* Didn't find it.
*/
if (i == sfp->count) {
ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
#if XFS_BIG_INUMS
if (i8elevated)
xfs_dir2_sf_toino4(args);
#endif
return XFS_ERROR(ENOENT);
}
}
#if XFS_BIG_INUMS
/*
* See if the old number was large, the new number is small.
*/
if (ino > XFS_DIR2_MAX_SHORT_INUM &&
args->inumber <= XFS_DIR2_MAX_SHORT_INUM) {
/*
* And the old count was one, so need to convert to small.
*/
if (sfp->i8count == 1)
xfs_dir2_sf_toino4(args);
else
sfp->i8count--;
}
/*
* See if the old number was small, the new number is large.
*/
if (ino <= XFS_DIR2_MAX_SHORT_INUM &&
args->inumber > XFS_DIR2_MAX_SHORT_INUM) {
/*
* add to the i8count unless we just converted to 8-byte
* inodes (which does an implied i8count = 1)
*/
ASSERT(sfp->i8count != 0);
if (!i8elevated)
sfp->i8count++;
}
#endif
xfs_dir2_sf_check(args);
xfs_trans_log_inode(args->trans, dp, XFS_ILOG_DDATA);
return 0;
}
#if XFS_BIG_INUMS
/*
* Convert from 8-byte inode numbers to 4-byte inode numbers.
* The last 8-byte inode number is gone, but the count is still 1.
*/
static void
xfs_dir2_sf_toino4(
xfs_da_args_t *args) /* operation arguments */
{
char *buf; /* old dir's buffer */
xfs_inode_t *dp; /* incore directory inode */
int i; /* entry index */
int newsize; /* new inode size */
xfs_dir2_sf_entry_t *oldsfep; /* old sf entry */
xfs_dir2_sf_hdr_t *oldsfp; /* old sf directory */
int oldsize; /* old inode size */
xfs_dir2_sf_entry_t *sfep; /* new sf entry */
xfs_dir2_sf_hdr_t *sfp; /* new sf directory */
struct xfs_mount *mp;
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dir2_sf_toino4(args);
dp = args->dp;
mp = dp->i_mount;
/*
* Copy the old directory to the buffer.
* Then nuke it from the inode, and add the new buffer to the inode.
* Don't want xfs_idata_realloc copying the data here.
*/
oldsize = dp->i_df.if_bytes;
buf = kmem_alloc(oldsize, KM_SLEEP);
oldsfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
ASSERT(oldsfp->i8count == 1);
memcpy(buf, oldsfp, oldsize);
/*
* Compute the new inode size.
*/
newsize =
oldsize -
(oldsfp->count + 1) *
((uint)sizeof(xfs_dir2_ino8_t) - (uint)sizeof(xfs_dir2_ino4_t));
xfs_idata_realloc(dp, -oldsize, XFS_DATA_FORK);
xfs_idata_realloc(dp, newsize, XFS_DATA_FORK);
/*
* Reset our pointers, the data has moved.
*/
oldsfp = (xfs_dir2_sf_hdr_t *)buf;
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
/*
* Fill in the new header.
*/
sfp->count = oldsfp->count;
sfp->i8count = 0;
dp->d_ops->sf_put_parent_ino(sfp, dp->d_ops->sf_get_parent_ino(oldsfp));
/*
* Copy the entries field by field.
*/
for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp),
oldsfep = xfs_dir2_sf_firstentry(oldsfp);
i < sfp->count;
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
i++, sfep = dp->d_ops->sf_nextentry(sfp, sfep),
oldsfep = dp->d_ops->sf_nextentry(oldsfp, oldsfep)) {
sfep->namelen = oldsfep->namelen;
sfep->offset = oldsfep->offset;
memcpy(sfep->name, oldsfep->name, sfep->namelen);
dp->d_ops->sf_put_ino(sfp, sfep,
dp->d_ops->sf_get_ino(oldsfp, oldsfep));
dp->d_ops->sf_put_ftype(sfep, dp->d_ops->sf_get_ftype(oldsfep));
}
/*
* Clean up the inode.
*/
kmem_free(buf);
dp->i_d.di_size = newsize;
xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
}
/*
* Convert from 4-byte inode numbers to 8-byte inode numbers.
* The new 8-byte inode number is not there yet, we leave with the
* count 1 but no corresponding entry.
*/
static void
xfs_dir2_sf_toino8(
xfs_da_args_t *args) /* operation arguments */
{
char *buf; /* old dir's buffer */
xfs_inode_t *dp; /* incore directory inode */
int i; /* entry index */
int newsize; /* new inode size */
xfs_dir2_sf_entry_t *oldsfep; /* old sf entry */
xfs_dir2_sf_hdr_t *oldsfp; /* old sf directory */
int oldsize; /* old inode size */
xfs_dir2_sf_entry_t *sfep; /* new sf entry */
xfs_dir2_sf_hdr_t *sfp; /* new sf directory */
struct xfs_mount *mp;
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dir2_sf_toino8(args);
dp = args->dp;
mp = dp->i_mount;
/*
* Copy the old directory to the buffer.
* Then nuke it from the inode, and add the new buffer to the inode.
* Don't want xfs_idata_realloc copying the data here.
*/
oldsize = dp->i_df.if_bytes;
buf = kmem_alloc(oldsize, KM_SLEEP);
oldsfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
ASSERT(oldsfp->i8count == 0);
memcpy(buf, oldsfp, oldsize);
/*
* Compute the new inode size.
*/
newsize =
oldsize +
(oldsfp->count + 1) *
((uint)sizeof(xfs_dir2_ino8_t) - (uint)sizeof(xfs_dir2_ino4_t));
xfs_idata_realloc(dp, -oldsize, XFS_DATA_FORK);
xfs_idata_realloc(dp, newsize, XFS_DATA_FORK);
/*
* Reset our pointers, the data has moved.
*/
oldsfp = (xfs_dir2_sf_hdr_t *)buf;
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
/*
* Fill in the new header.
*/
sfp->count = oldsfp->count;
sfp->i8count = 1;
dp->d_ops->sf_put_parent_ino(sfp, dp->d_ops->sf_get_parent_ino(oldsfp));
/*
* Copy the entries field by field.
*/
for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp),
oldsfep = xfs_dir2_sf_firstentry(oldsfp);
i < sfp->count;
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 11:11:46 +00:00
i++, sfep = dp->d_ops->sf_nextentry(sfp, sfep),
oldsfep = dp->d_ops->sf_nextentry(oldsfp, oldsfep)) {
sfep->namelen = oldsfep->namelen;
sfep->offset = oldsfep->offset;
memcpy(sfep->name, oldsfep->name, sfep->namelen);
dp->d_ops->sf_put_ino(sfp, sfep,
dp->d_ops->sf_get_ino(oldsfp, oldsfep));
dp->d_ops->sf_put_ftype(sfep, dp->d_ops->sf_get_ftype(oldsfep));
}
/*
* Clean up the inode.
*/
kmem_free(buf);
dp->i_d.di_size = newsize;
xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
}
#endif /* XFS_BIG_INUMS */