linux/fs/xfs/xfs_da_format.c

656 lines
15 KiB
C
Raw Normal View History

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 19:11:46 +08:00
/*
* Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
* Copyright (c) 2013 Red Hat, 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_inode.h"
#include "xfs_dir2.h"
/*
* Shortform directory ops
*/
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 19:11:46 +08:00
static int
xfs_dir2_sf_entsize(
struct xfs_dir2_sf_hdr *hdr,
int len)
{
int count = sizeof(struct xfs_dir2_sf_entry); /* namelen + offset */
count += len; /* name */
count += hdr->i8count ? sizeof(xfs_dir2_ino8_t) :
sizeof(xfs_dir2_ino4_t); /* ino # */
return count;
}
static int
xfs_dir3_sf_entsize(
struct xfs_dir2_sf_hdr *hdr,
int len)
{
return xfs_dir2_sf_entsize(hdr, len) + sizeof(__uint8_t);
}
static struct xfs_dir2_sf_entry *
xfs_dir2_sf_nextentry(
struct xfs_dir2_sf_hdr *hdr,
struct xfs_dir2_sf_entry *sfep)
{
return (struct xfs_dir2_sf_entry *)
((char *)sfep + xfs_dir2_sf_entsize(hdr, sfep->namelen));
}
static struct xfs_dir2_sf_entry *
xfs_dir3_sf_nextentry(
struct xfs_dir2_sf_hdr *hdr,
struct xfs_dir2_sf_entry *sfep)
{
return (struct xfs_dir2_sf_entry *)
((char *)sfep + xfs_dir3_sf_entsize(hdr, sfep->namelen));
}
/*
* For filetype enabled shortform directories, the file type field is stored at
* the end of the name. Because it's only a single byte, endian conversion is
* not necessary. For non-filetype enable directories, the type is always
* unknown and we never store the value.
*/
static __uint8_t
xfs_dir2_sfe_get_ftype(
struct xfs_dir2_sf_entry *sfep)
{
return XFS_DIR3_FT_UNKNOWN;
}
static void
xfs_dir2_sfe_put_ftype(
struct xfs_dir2_sf_entry *sfep,
__uint8_t ftype)
{
ASSERT(ftype < XFS_DIR3_FT_MAX);
}
static __uint8_t
xfs_dir3_sfe_get_ftype(
struct xfs_dir2_sf_entry *sfep)
{
__uint8_t ftype;
ftype = sfep->name[sfep->namelen];
if (ftype >= XFS_DIR3_FT_MAX)
return XFS_DIR3_FT_UNKNOWN;
return ftype;
}
static void
xfs_dir3_sfe_put_ftype(
struct xfs_dir2_sf_entry *sfep,
__uint8_t ftype)
{
ASSERT(ftype < XFS_DIR3_FT_MAX);
sfep->name[sfep->namelen] = ftype;
}
/*
* Inode numbers in short-form directories can come in two versions,
* either 4 bytes or 8 bytes wide. These helpers deal with the
* two forms transparently by looking at the headers i8count field.
*
* For 64-bit inode number the most significant byte must be zero.
*/
static xfs_ino_t
xfs_dir2_sf_get_ino(
struct xfs_dir2_sf_hdr *hdr,
xfs_dir2_inou_t *from)
{
if (hdr->i8count)
return get_unaligned_be64(&from->i8.i) & 0x00ffffffffffffffULL;
else
return get_unaligned_be32(&from->i4.i);
}
static void
xfs_dir2_sf_put_ino(
struct xfs_dir2_sf_hdr *hdr,
xfs_dir2_inou_t *to,
xfs_ino_t ino)
{
ASSERT((ino & 0xff00000000000000ULL) == 0);
if (hdr->i8count)
put_unaligned_be64(ino, &to->i8.i);
else
put_unaligned_be32(ino, &to->i4.i);
}
static xfs_ino_t
xfs_dir2_sf_get_parent_ino(
struct xfs_dir2_sf_hdr *hdr)
{
return xfs_dir2_sf_get_ino(hdr, &hdr->parent);
}
static void
xfs_dir2_sf_put_parent_ino(
struct xfs_dir2_sf_hdr *hdr,
xfs_ino_t ino)
{
xfs_dir2_sf_put_ino(hdr, &hdr->parent, ino);
}
/*
* In short-form directory entries the inode numbers are stored at variable
* offset behind the entry name. If the entry stores a filetype value, then it
* sits between the name and the inode number. Hence the inode numbers may only
* be accessed through the helpers below.
*/
static xfs_ino_t
xfs_dir2_sfe_get_ino(
struct xfs_dir2_sf_hdr *hdr,
struct xfs_dir2_sf_entry *sfep)
{
return xfs_dir2_sf_get_ino(hdr,
(xfs_dir2_inou_t *)&sfep->name[sfep->namelen]);
}
static void
xfs_dir2_sfe_put_ino(
struct xfs_dir2_sf_hdr *hdr,
struct xfs_dir2_sf_entry *sfep,
xfs_ino_t ino)
{
xfs_dir2_sf_put_ino(hdr,
(xfs_dir2_inou_t *)&sfep->name[sfep->namelen], ino);
}
static xfs_ino_t
xfs_dir3_sfe_get_ino(
struct xfs_dir2_sf_hdr *hdr,
struct xfs_dir2_sf_entry *sfep)
{
return xfs_dir2_sf_get_ino(hdr,
(xfs_dir2_inou_t *)&sfep->name[sfep->namelen + 1]);
}
static void
xfs_dir3_sfe_put_ino(
struct xfs_dir2_sf_hdr *hdr,
struct xfs_dir2_sf_entry *sfep,
xfs_ino_t ino)
{
xfs_dir2_sf_put_ino(hdr,
(xfs_dir2_inou_t *)&sfep->name[sfep->namelen + 1], ino);
}
/*
* Directory data block operations
*/
static int
__xfs_dir3_data_entsize(
bool ftype,
int n)
{
int size = offsetof(struct xfs_dir2_data_entry, name[0]);
size += n;
size += sizeof(xfs_dir2_data_off_t);
if (ftype)
size += sizeof(__uint8_t);
return roundup(size, XFS_DIR2_DATA_ALIGN);
}
static int
xfs_dir2_data_entsize(
int n)
{
return __xfs_dir3_data_entsize(false, n);
}
static int
xfs_dir3_data_entsize(
int n)
{
return __xfs_dir3_data_entsize(true, n);
}
static __uint8_t
xfs_dir2_data_get_ftype(
struct xfs_dir2_data_entry *dep)
{
return XFS_DIR3_FT_UNKNOWN;
}
static void
xfs_dir2_data_put_ftype(
struct xfs_dir2_data_entry *dep,
__uint8_t ftype)
{
ASSERT(ftype < XFS_DIR3_FT_MAX);
}
static __uint8_t
xfs_dir3_data_get_ftype(
struct xfs_dir2_data_entry *dep)
{
__uint8_t ftype = dep->name[dep->namelen];
ASSERT(ftype < XFS_DIR3_FT_MAX);
if (ftype >= XFS_DIR3_FT_MAX)
return XFS_DIR3_FT_UNKNOWN;
return ftype;
}
static void
xfs_dir3_data_put_ftype(
struct xfs_dir2_data_entry *dep,
__uint8_t type)
{
ASSERT(type < XFS_DIR3_FT_MAX);
ASSERT(dep->namelen != 0);
dep->name[dep->namelen] = type;
}
/*
* Pointer to an entry's tag word.
*/
static __be16 *
xfs_dir2_data_entry_tag_p(
struct xfs_dir2_data_entry *dep)
{
return (__be16 *)((char *)dep +
xfs_dir2_data_entsize(dep->namelen) - sizeof(__be16));
}
static __be16 *
xfs_dir3_data_entry_tag_p(
struct xfs_dir2_data_entry *dep)
{
return (__be16 *)((char *)dep +
xfs_dir3_data_entsize(dep->namelen) - sizeof(__be16));
}
/*
* Offsets of . and .. in data space (always block 0)
*/
static xfs_dir2_data_aoff_t
xfs_dir2_data_dot_offset(void)
{
return sizeof(struct xfs_dir2_data_hdr);
}
static xfs_dir2_data_aoff_t
xfs_dir2_data_dotdot_offset(void)
{
return xfs_dir2_data_dot_offset() + xfs_dir2_data_entsize(1);
}
static xfs_dir2_data_aoff_t
xfs_dir2_data_first_offset(void)
{
return xfs_dir2_data_dotdot_offset() + xfs_dir2_data_entsize(2);
}
static xfs_dir2_data_aoff_t
xfs_dir3_data_dot_offset(void)
{
return sizeof(struct xfs_dir3_data_hdr);
}
static xfs_dir2_data_aoff_t
xfs_dir3_data_dotdot_offset(void)
{
return xfs_dir3_data_dot_offset() + xfs_dir3_data_entsize(1);
}
static xfs_dir2_data_aoff_t
xfs_dir3_data_first_offset(void)
{
return xfs_dir3_data_dotdot_offset() + xfs_dir3_data_entsize(2);
}
/*
* location of . and .. in data space (always block 0)
*/
static struct xfs_dir2_data_entry *
xfs_dir2_data_dot_entry_p(
struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_entry *)
((char *)hdr + xfs_dir2_data_dot_offset());
}
static struct xfs_dir2_data_entry *
xfs_dir2_data_dotdot_entry_p(
struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_entry *)
((char *)hdr + xfs_dir2_data_dotdot_offset());
}
static struct xfs_dir2_data_entry *
xfs_dir2_data_first_entry_p(
struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_entry *)
((char *)hdr + xfs_dir2_data_first_offset());
}
static struct xfs_dir2_data_entry *
xfs_dir3_data_dot_entry_p(
struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_entry *)
((char *)hdr + xfs_dir3_data_dot_offset());
}
static struct xfs_dir2_data_entry *
xfs_dir3_data_dotdot_entry_p(
struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_entry *)
((char *)hdr + xfs_dir3_data_dotdot_offset());
}
static struct xfs_dir2_data_entry *
xfs_dir3_data_first_entry_p(
struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_entry *)
((char *)hdr + xfs_dir3_data_first_offset());
}
static struct xfs_dir2_data_free *
xfs_dir2_data_bestfree_p(struct xfs_dir2_data_hdr *hdr)
{
return hdr->bestfree;
}
static struct xfs_dir2_data_free *
xfs_dir3_data_bestfree_p(struct xfs_dir2_data_hdr *hdr)
{
return ((struct xfs_dir3_data_hdr *)hdr)->best_free;
}
static size_t
xfs_dir2_data_entry_offset(void)
{
return sizeof(struct xfs_dir2_data_hdr);
}
static struct xfs_dir2_data_entry *
xfs_dir2_data_entry_p(struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_entry *)
((char *)hdr + xfs_dir2_data_entry_offset());
}
static struct xfs_dir2_data_unused *
xfs_dir2_data_unused_p(struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_unused *)
((char *)hdr + xfs_dir2_data_entry_offset());
}
static size_t
xfs_dir3_data_entry_offset(void)
{
return sizeof(struct xfs_dir3_data_hdr);
}
static struct xfs_dir2_data_entry *
xfs_dir3_data_entry_p(struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_entry *)
((char *)hdr + xfs_dir3_data_entry_offset());
}
static struct xfs_dir2_data_unused *
xfs_dir3_data_unused_p(struct xfs_dir2_data_hdr *hdr)
{
return (struct xfs_dir2_data_unused *)
((char *)hdr + xfs_dir3_data_entry_offset());
}
/*
* Directory Leaf block operations
*/
static int
xfs_dir2_leaf_hdr_size(void)
{
return sizeof(struct xfs_dir2_leaf_hdr);
}
static int
xfs_dir2_max_leaf_ents(struct xfs_mount *mp)
{
return (mp->m_dirblksize - xfs_dir2_leaf_hdr_size()) /
(uint)sizeof(struct xfs_dir2_leaf_entry);
}
static struct xfs_dir2_leaf_entry *
xfs_dir2_leaf_ents_p(struct xfs_dir2_leaf *lp)
{
return lp->__ents;
}
static int
xfs_dir3_leaf_hdr_size(void)
{
return sizeof(struct xfs_dir3_leaf_hdr);
}
static inline int
xfs_dir3_max_leaf_ents(struct xfs_mount *mp)
{
return (mp->m_dirblksize - xfs_dir3_leaf_hdr_size()) /
(uint)sizeof(struct xfs_dir2_leaf_entry);
}
static inline struct xfs_dir2_leaf_entry *
xfs_dir3_leaf_ents_p(struct xfs_dir2_leaf *lp)
{
return ((struct xfs_dir3_leaf *)lp)->__ents;
}
/*
* Directory/Attribute Node block operations
*/
static inline int
xfs_da2_node_hdr_size(void)
{
return sizeof(struct xfs_da_node_hdr);
}
static struct xfs_da_node_entry *
xfs_da2_node_tree_p(struct xfs_da_intnode *dap)
{
return dap->__btree;
}
static inline int
xfs_da3_node_hdr_size(void)
{
return sizeof(struct xfs_da3_node_hdr);
}
static inline struct xfs_da_node_entry *
xfs_da3_node_tree_p(struct xfs_da_intnode *dap)
{
return ((struct xfs_da3_intnode *)dap)->__btree;
}
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 19:11:46 +08:00
const struct xfs_dir_ops xfs_dir2_ops = {
.sf_entsize = xfs_dir2_sf_entsize,
.sf_nextentry = xfs_dir2_sf_nextentry,
.sf_get_ftype = xfs_dir2_sfe_get_ftype,
.sf_put_ftype = xfs_dir2_sfe_put_ftype,
.sf_get_ino = xfs_dir2_sfe_get_ino,
.sf_put_ino = xfs_dir2_sfe_put_ino,
.sf_get_parent_ino = xfs_dir2_sf_get_parent_ino,
.sf_put_parent_ino = xfs_dir2_sf_put_parent_ino,
.data_entsize = xfs_dir2_data_entsize,
.data_get_ftype = xfs_dir2_data_get_ftype,
.data_put_ftype = xfs_dir2_data_put_ftype,
.data_entry_tag_p = xfs_dir2_data_entry_tag_p,
.data_bestfree_p = xfs_dir2_data_bestfree_p,
.data_dot_offset = xfs_dir2_data_dot_offset,
.data_dotdot_offset = xfs_dir2_data_dotdot_offset,
.data_first_offset = xfs_dir2_data_first_offset,
.data_entry_offset = xfs_dir2_data_entry_offset,
.data_dot_entry_p = xfs_dir2_data_dot_entry_p,
.data_dotdot_entry_p = xfs_dir2_data_dotdot_entry_p,
.data_first_entry_p = xfs_dir2_data_first_entry_p,
.data_entry_p = xfs_dir2_data_entry_p,
.data_unused_p = xfs_dir2_data_unused_p,
.leaf_hdr_size = xfs_dir2_leaf_hdr_size,
.leaf_max_ents = xfs_dir2_max_leaf_ents,
.leaf_ents_p = xfs_dir2_leaf_ents_p,
.node_hdr_size = xfs_da2_node_hdr_size,
.node_tree_p = xfs_da2_node_tree_p,
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 19:11:46 +08:00
};
const struct xfs_dir_ops xfs_dir2_ftype_ops = {
.sf_entsize = xfs_dir3_sf_entsize,
.sf_nextentry = xfs_dir3_sf_nextentry,
.sf_get_ftype = xfs_dir3_sfe_get_ftype,
.sf_put_ftype = xfs_dir3_sfe_put_ftype,
.sf_get_ino = xfs_dir3_sfe_get_ino,
.sf_put_ino = xfs_dir3_sfe_put_ino,
.sf_get_parent_ino = xfs_dir2_sf_get_parent_ino,
.sf_put_parent_ino = xfs_dir2_sf_put_parent_ino,
.data_entsize = xfs_dir3_data_entsize,
.data_get_ftype = xfs_dir3_data_get_ftype,
.data_put_ftype = xfs_dir3_data_put_ftype,
.data_entry_tag_p = xfs_dir3_data_entry_tag_p,
.data_bestfree_p = xfs_dir2_data_bestfree_p,
.data_dot_offset = xfs_dir2_data_dot_offset,
.data_dotdot_offset = xfs_dir2_data_dotdot_offset,
.data_first_offset = xfs_dir2_data_first_offset,
.data_entry_offset = xfs_dir2_data_entry_offset,
.data_dot_entry_p = xfs_dir2_data_dot_entry_p,
.data_dotdot_entry_p = xfs_dir2_data_dotdot_entry_p,
.data_first_entry_p = xfs_dir2_data_first_entry_p,
.data_entry_p = xfs_dir2_data_entry_p,
.data_unused_p = xfs_dir2_data_unused_p,
.leaf_hdr_size = xfs_dir2_leaf_hdr_size,
.leaf_max_ents = xfs_dir2_max_leaf_ents,
.leaf_ents_p = xfs_dir2_leaf_ents_p,
.node_hdr_size = xfs_da2_node_hdr_size,
.node_tree_p = xfs_da2_node_tree_p,
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 19:11:46 +08:00
};
const struct xfs_dir_ops xfs_dir3_ops = {
.sf_entsize = xfs_dir3_sf_entsize,
.sf_nextentry = xfs_dir3_sf_nextentry,
.sf_get_ftype = xfs_dir3_sfe_get_ftype,
.sf_put_ftype = xfs_dir3_sfe_put_ftype,
.sf_get_ino = xfs_dir3_sfe_get_ino,
.sf_put_ino = xfs_dir3_sfe_put_ino,
.sf_get_parent_ino = xfs_dir2_sf_get_parent_ino,
.sf_put_parent_ino = xfs_dir2_sf_put_parent_ino,
.data_entsize = xfs_dir3_data_entsize,
.data_get_ftype = xfs_dir3_data_get_ftype,
.data_put_ftype = xfs_dir3_data_put_ftype,
.data_entry_tag_p = xfs_dir3_data_entry_tag_p,
.data_bestfree_p = xfs_dir3_data_bestfree_p,
.data_dot_offset = xfs_dir3_data_dot_offset,
.data_dotdot_offset = xfs_dir3_data_dotdot_offset,
.data_first_offset = xfs_dir3_data_first_offset,
.data_entry_offset = xfs_dir3_data_entry_offset,
.data_dot_entry_p = xfs_dir3_data_dot_entry_p,
.data_dotdot_entry_p = xfs_dir3_data_dotdot_entry_p,
.data_first_entry_p = xfs_dir3_data_first_entry_p,
.data_entry_p = xfs_dir3_data_entry_p,
.data_unused_p = xfs_dir3_data_unused_p,
.leaf_hdr_size = xfs_dir3_leaf_hdr_size,
.leaf_max_ents = xfs_dir3_max_leaf_ents,
.leaf_ents_p = xfs_dir3_leaf_ents_p,
.node_hdr_size = xfs_da3_node_hdr_size,
.node_tree_p = xfs_da3_node_tree_p,
};
const struct xfs_dir_ops xfs_dir2_nondir_ops = {
.node_hdr_size = xfs_da2_node_hdr_size,
.node_tree_p = xfs_da2_node_tree_p,
};
const struct xfs_dir_ops xfs_dir3_nondir_ops = {
.node_hdr_size = xfs_da3_node_hdr_size,
.node_tree_p = xfs_da3_node_tree_p,
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 19:11:46 +08:00
};
/*
* Return the ops structure according to the current config. If we are passed
* an inode, then that overrides the default config we use which is based on
* feature bits.
*/
const struct xfs_dir_ops *
xfs_dir_get_ops(
struct xfs_mount *mp,
struct xfs_inode *dp)
{
if (dp)
return dp->d_ops;
if (mp->m_dir_inode_ops)
return mp->m_dir_inode_ops;
if (xfs_sb_version_hascrc(&mp->m_sb))
return &xfs_dir3_ops;
if (xfs_sb_version_hasftype(&mp->m_sb))
return &xfs_dir2_ftype_ops;
return &xfs_dir2_ops;
}
const struct xfs_dir_ops *
xfs_nondir_get_ops(
struct xfs_mount *mp,
struct xfs_inode *dp)
{
if (dp)
return dp->d_ops;
if (mp->m_nondir_inode_ops)
return mp->m_nondir_inode_ops;
if (xfs_sb_version_hascrc(&mp->m_sb))
return &xfs_dir3_nondir_ops;
return &xfs_dir2_nondir_ops;
}