linux_old1/fs/xfs/xfs_da_format.h

1016 lines
32 KiB
C
Raw Normal View History

/*
* Copyright (c) 2000-2001,2005 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
*/
#ifndef __XFS_DA_FORMAT_H__
#define __XFS_DA_FORMAT_H__
/*========================================================================
* Directory Structure when greater than XFS_LBSIZE(mp) bytes.
*========================================================================*/
/*
* This structure is common to both leaf nodes and non-leaf nodes in the Btree.
*
* It is used to manage a doubly linked list of all blocks at the same
* level in the Btree, and to identify which type of block this is.
*/
#define XFS_DA_NODE_MAGIC 0xfebe /* magic number: non-leaf blocks */
#define XFS_ATTR_LEAF_MAGIC 0xfbee /* magic number: attribute leaf blks */
#define XFS_DIR2_LEAF1_MAGIC 0xd2f1 /* magic number: v2 dirlf single blks */
#define XFS_DIR2_LEAFN_MAGIC 0xd2ff /* magic number: v2 dirlf multi blks */
typedef struct xfs_da_blkinfo {
__be32 forw; /* previous block in list */
__be32 back; /* following block in list */
__be16 magic; /* validity check on block */
__be16 pad; /* unused */
} xfs_da_blkinfo_t;
/*
* CRC enabled directory structure types
*
* The headers change size for the additional verification information, but
* otherwise the tree layouts and contents are unchanged. Hence the da btree
* code can use the struct xfs_da_blkinfo for manipulating the tree links and
* magic numbers without modification for both v2 and v3 nodes.
*/
#define XFS_DA3_NODE_MAGIC 0x3ebe /* magic number: non-leaf blocks */
#define XFS_ATTR3_LEAF_MAGIC 0x3bee /* magic number: attribute leaf blks */
#define XFS_DIR3_LEAF1_MAGIC 0x3df1 /* magic number: v2 dirlf single blks */
#define XFS_DIR3_LEAFN_MAGIC 0x3dff /* magic number: v2 dirlf multi blks */
struct xfs_da3_blkinfo {
/*
* the node link manipulation code relies on the fact that the first
* element of this structure is the struct xfs_da_blkinfo so it can
* ignore the differences in the rest of the structures.
*/
struct xfs_da_blkinfo hdr;
__be32 crc; /* CRC of block */
__be64 blkno; /* first block of the buffer */
__be64 lsn; /* sequence number of last write */
uuid_t uuid; /* filesystem we belong to */
__be64 owner; /* inode that owns the block */
};
/*
* This is the structure of the root and intermediate nodes in the Btree.
* The leaf nodes are defined above.
*
* Entries are not packed.
*
* Since we have duplicate keys, use a binary search but always follow
* all match in the block, not just the first match found.
*/
#define XFS_DA_NODE_MAXDEPTH 5 /* max depth of Btree */
typedef struct xfs_da_node_hdr {
struct xfs_da_blkinfo info; /* block type, links, etc. */
__be16 __count; /* count of active entries */
__be16 __level; /* level above leaves (leaf == 0) */
} xfs_da_node_hdr_t;
struct xfs_da3_node_hdr {
struct xfs_da3_blkinfo info; /* block type, links, etc. */
__be16 __count; /* count of active entries */
__be16 __level; /* level above leaves (leaf == 0) */
__be32 __pad32;
};
#define XFS_DA3_NODE_CRC_OFF (offsetof(struct xfs_da3_node_hdr, info.crc))
typedef struct xfs_da_node_entry {
__be32 hashval; /* hash value for this descendant */
__be32 before; /* Btree block before this key */
} xfs_da_node_entry_t;
typedef struct xfs_da_intnode {
struct xfs_da_node_hdr hdr;
struct xfs_da_node_entry __btree[];
} xfs_da_intnode_t;
struct xfs_da3_intnode {
struct xfs_da3_node_hdr hdr;
struct xfs_da_node_entry __btree[];
};
/*
* In-core version of the node header to abstract the differences in the v2 and
* v3 disk format of the headers. Callers need to convert to/from disk format as
* appropriate.
*/
struct xfs_da3_icnode_hdr {
__uint32_t forw;
__uint32_t back;
__uint16_t magic;
__uint16_t count;
__uint16_t level;
};
#define XFS_LBSIZE(mp) (mp)->m_sb.sb_blocksize
/*
* Directory version 2.
*
* There are 4 possible formats:
* - shortform - embedded into the inode
* - single block - data with embedded leaf at the end
* - multiple data blocks, single leaf+freeindex block
* - data blocks, node and leaf blocks (btree), freeindex blocks
*
* Note: many node blocks structures and constants are shared with the attr
* code and defined in xfs_da_btree.h.
*/
#define XFS_DIR2_BLOCK_MAGIC 0x58443242 /* XD2B: single block dirs */
#define XFS_DIR2_DATA_MAGIC 0x58443244 /* XD2D: multiblock dirs */
#define XFS_DIR2_FREE_MAGIC 0x58443246 /* XD2F: free index blocks */
/*
* Directory Version 3 With CRCs.
*
* The tree formats are the same as for version 2 directories. The difference
* is in the block header and dirent formats. In many cases the v3 structures
* use v2 definitions as they are no different and this makes code sharing much
* easier.
*
* Also, the xfs_dir3_*() functions handle both v2 and v3 formats - if the
* format is v2 then they switch to the existing v2 code, or the format is v3
* they implement the v3 functionality. This means the existing dir2 is a mix of
* xfs_dir2/xfs_dir3 calls and functions. The xfs_dir3 functions are called
* where there is a difference in the formats, otherwise the code is unchanged.
*
* Where it is possible, the code decides what to do based on the magic numbers
* in the blocks rather than feature bits in the superblock. This means the code
* is as independent of the external XFS code as possible as doesn't require
* passing struct xfs_mount pointers into places where it isn't really
* necessary.
*
* Version 3 includes:
*
* - a larger block header for CRC and identification purposes and so the
* offsets of all the structures inside the blocks are different.
*
* - new magic numbers to be able to detect the v2/v3 types on the fly.
*/
#define XFS_DIR3_BLOCK_MAGIC 0x58444233 /* XDB3: single block dirs */
#define XFS_DIR3_DATA_MAGIC 0x58444433 /* XDD3: multiblock dirs */
#define XFS_DIR3_FREE_MAGIC 0x58444633 /* XDF3: free index blocks */
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 18:50:09 +08:00
/*
* Dirents in version 3 directories have a file type field. Additions to this
* list are an on-disk format change, requiring feature bits. Valid values
* are as follows:
*/
#define XFS_DIR3_FT_UNKNOWN 0
#define XFS_DIR3_FT_REG_FILE 1
#define XFS_DIR3_FT_DIR 2
#define XFS_DIR3_FT_CHRDEV 3
#define XFS_DIR3_FT_BLKDEV 4
#define XFS_DIR3_FT_FIFO 5
#define XFS_DIR3_FT_SOCK 6
#define XFS_DIR3_FT_SYMLINK 7
#define XFS_DIR3_FT_WHT 8
#define XFS_DIR3_FT_MAX 9
/*
* Byte offset in data block and shortform entry.
*/
typedef __uint16_t xfs_dir2_data_off_t;
#define NULLDATAOFF 0xffffU
typedef uint xfs_dir2_data_aoff_t; /* argument form */
/*
* Normalized offset (in a data block) of the entry, really xfs_dir2_data_off_t.
* Only need 16 bits, this is the byte offset into the single block form.
*/
typedef struct { __uint8_t i[2]; } __arch_pack xfs_dir2_sf_off_t;
/*
* Offset in data space of a data entry.
*/
typedef __uint32_t xfs_dir2_dataptr_t;
#define XFS_DIR2_MAX_DATAPTR ((xfs_dir2_dataptr_t)0xffffffff)
#define XFS_DIR2_NULL_DATAPTR ((xfs_dir2_dataptr_t)0)
/*
* Byte offset in a directory.
*/
typedef xfs_off_t xfs_dir2_off_t;
/*
* Directory block number (logical dirblk in file)
*/
typedef __uint32_t xfs_dir2_db_t;
/*
* Inode number stored as 8 8-bit values.
*/
typedef struct { __uint8_t i[8]; } xfs_dir2_ino8_t;
/*
* Inode number stored as 4 8-bit values.
* Works a lot of the time, when all the inode numbers in a directory
* fit in 32 bits.
*/
typedef struct { __uint8_t i[4]; } xfs_dir2_ino4_t;
typedef union {
xfs_dir2_ino8_t i8;
xfs_dir2_ino4_t i4;
} xfs_dir2_inou_t;
#define XFS_DIR2_MAX_SHORT_INUM ((xfs_ino_t)0xffffffffULL)
/*
* Directory layout when stored internal to an inode.
*
* Small directories are packed as tightly as possible so as to fit into the
* literal area of the inode. These "shortform" directories consist of a
* single xfs_dir2_sf_hdr header followed by zero or more xfs_dir2_sf_entry
* structures. Due the different inode number storage size and the variable
* length name field in the xfs_dir2_sf_entry all these structure are
* variable length, and the accessors in this file should be used to iterate
* over them.
*/
typedef struct xfs_dir2_sf_hdr {
__uint8_t count; /* count of entries */
__uint8_t i8count; /* count of 8-byte inode #s */
xfs_dir2_inou_t parent; /* parent dir inode number */
} __arch_pack xfs_dir2_sf_hdr_t;
typedef struct xfs_dir2_sf_entry {
__u8 namelen; /* actual name length */
xfs_dir2_sf_off_t offset; /* saved offset */
__u8 name[]; /* name, variable 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 18:50:09 +08:00
* A single byte containing the file type field follows the inode
* number for version 3 directory entries.
*
* A xfs_dir2_ino8_t or xfs_dir2_ino4_t follows here, at a
* variable offset after the name.
*/
} __arch_pack xfs_dir2_sf_entry_t;
static inline int xfs_dir2_sf_hdr_size(int i8count)
{
return sizeof(struct xfs_dir2_sf_hdr) -
(i8count == 0) *
(sizeof(xfs_dir2_ino8_t) - sizeof(xfs_dir2_ino4_t));
}
static inline xfs_dir2_data_aoff_t
xfs_dir2_sf_get_offset(xfs_dir2_sf_entry_t *sfep)
{
return get_unaligned_be16(&sfep->offset.i);
}
static inline void
xfs_dir2_sf_put_offset(xfs_dir2_sf_entry_t *sfep, xfs_dir2_data_aoff_t off)
{
put_unaligned_be16(off, &sfep->offset.i);
}
static inline struct xfs_dir2_sf_entry *
xfs_dir2_sf_firstentry(struct xfs_dir2_sf_hdr *hdr)
{
return (struct xfs_dir2_sf_entry *)
((char *)hdr + xfs_dir2_sf_hdr_size(hdr->i8count));
}
/*
* Data block structures.
*
* A pure data block looks like the following drawing on disk:
*
* +-------------------------------------------------+
* | xfs_dir2_data_hdr_t |
* +-------------------------------------------------+
* | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
* | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
* | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
* | ... |
* +-------------------------------------------------+
* | unused space |
* +-------------------------------------------------+
*
* As all the entries are variable size structures the accessors below should
* be used to iterate over them.
*
* In addition to the pure data blocks for the data and node formats,
* most structures are also used for the combined data/freespace "block"
* format below.
*/
#define XFS_DIR2_DATA_ALIGN_LOG 3 /* i.e., 8 bytes */
#define XFS_DIR2_DATA_ALIGN (1 << XFS_DIR2_DATA_ALIGN_LOG)
#define XFS_DIR2_DATA_FREE_TAG 0xffff
#define XFS_DIR2_DATA_FD_COUNT 3
/*
* Directory address space divided into sections,
* spaces separated by 32GB.
*/
#define XFS_DIR2_SPACE_SIZE (1ULL << (32 + XFS_DIR2_DATA_ALIGN_LOG))
#define XFS_DIR2_DATA_SPACE 0
#define XFS_DIR2_DATA_OFFSET (XFS_DIR2_DATA_SPACE * XFS_DIR2_SPACE_SIZE)
#define XFS_DIR2_DATA_FIRSTDB(mp) \
xfs_dir2_byte_to_db(mp, XFS_DIR2_DATA_OFFSET)
/*
* Describe a free area in the data block.
*
* The freespace will be formatted as a xfs_dir2_data_unused_t.
*/
typedef struct xfs_dir2_data_free {
__be16 offset; /* start of freespace */
__be16 length; /* length of freespace */
} xfs_dir2_data_free_t;
/*
* Header for the data blocks.
*
* The code knows that XFS_DIR2_DATA_FD_COUNT is 3.
*/
typedef struct xfs_dir2_data_hdr {
__be32 magic; /* XFS_DIR2_DATA_MAGIC or */
/* XFS_DIR2_BLOCK_MAGIC */
xfs_dir2_data_free_t bestfree[XFS_DIR2_DATA_FD_COUNT];
} xfs_dir2_data_hdr_t;
/*
* define a structure for all the verification fields we are adding to the
* directory block structures. This will be used in several structures.
* The magic number must be the first entry to align with all the dir2
* structures so we determine how to decode them just by the magic number.
*/
struct xfs_dir3_blk_hdr {
__be32 magic; /* magic number */
__be32 crc; /* CRC of block */
__be64 blkno; /* first block of the buffer */
__be64 lsn; /* sequence number of last write */
uuid_t uuid; /* filesystem we belong to */
__be64 owner; /* inode that owns the block */
};
struct xfs_dir3_data_hdr {
struct xfs_dir3_blk_hdr hdr;
xfs_dir2_data_free_t best_free[XFS_DIR2_DATA_FD_COUNT];
xfs: fix implicit padding in directory and attr CRC formats Michael L. Semon has been testing CRC patches on a 32 bit system and been seeing assert failures in the directory code from xfs/080. Thanks to Michael's heroic efforts with printk debugging, we found that the problem was that the last free space being left in the directory structure was too small to fit a unused tag structure and it was being corrupted and attempting to log a region out of bounds. Hence the assert failure looked something like: ..... #5 calling xfs_dir2_data_log_unused() 36 32 #1 4092 4095 4096 #2 8182 8183 4096 XFS: Assertion failed: first <= last && last < BBTOB(bp->b_length), file: fs/xfs/xfs_trans_buf.c, line: 568 Where #1 showed the first region of the dup being logged (i.e. the last 4 bytes of a directory buffer) and #2 shows the corrupt values being calculated from the length of the dup entry which overflowed the size of the buffer. It turns out that the problem was not in the logging code, nor in the freespace handling code. It is an initial condition bug that only shows up on 32 bit systems. When a new buffer is initialised, where's the freespace that is set up: [ 172.316249] calling xfs_dir2_leaf_addname() from xfs_dir_createname() [ 172.316346] #9 calling xfs_dir2_data_log_unused() [ 172.316351] #1 calling xfs_trans_log_buf() 60 63 4096 [ 172.316353] #2 calling xfs_trans_log_buf() 4094 4095 4096 Note the offset of the first region being logged? It's 60 bytes into the buffer. Once I saw that, I pretty much knew that the bug was going to be caused by this. Essentially, all direct entries are rounded to 8 bytes in length, and all entries start with an 8 byte alignment. This means that we can decode inplace as variables are naturally aligned. With the directory data supposedly starting on a 8 byte boundary, and all entries padded to 8 bytes, the minimum freespace in a directory block is supposed to be 8 bytes, which is large enough to fit a unused data entry structure (6 bytes in size). The fact we only have 4 bytes of free space indicates a directory data block alignment problem. And what do you know - there's an implicit hole in the directory data block header for the CRC format, which means the header is 60 byte on 32 bit intel systems and 64 bytes on 64 bit systems. Needs padding. And while looking at the structures, I found the same problem in the attr leaf header. Fix them both. Note that this only affects 32 bit systems with CRCs enabled. Everything else is just fine. Note that CRC enabled filesystems created before this fix on such systems will not be readable with this fix applied. Reported-by: Michael L. Semon <mlsemon35@gmail.com> Debugged-by: Michael L. Semon <mlsemon35@gmail.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Ben Myers <bpm@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com> (cherry picked from commit 8a1fd2950e1fe267e11fc8c85dcaa6b023b51b60)
2013-06-12 10:19:07 +08:00
__be32 pad; /* 64 bit alignment */
};
#define XFS_DIR3_DATA_CRC_OFF offsetof(struct xfs_dir3_data_hdr, hdr.crc)
/*
* Active entry in a data block.
*
* Aligned to 8 bytes. After the variable length name field there is a
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 18:50:09 +08:00
* 2 byte tag field, which can be accessed using xfs_dir3_data_entry_tag_p.
*
* For dir3 structures, there is file type field between the name and the tag.
* This can only be manipulated by helper functions. It is packed hard against
* the end of the name so any padding for rounding is between the file type and
* the tag.
*/
typedef struct xfs_dir2_data_entry {
__be64 inumber; /* inode number */
__u8 namelen; /* name length */
__u8 name[]; /* name bytes, no null */
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 18:50:09 +08:00
/* __u8 filetype; */ /* type of inode we point to */
/* __be16 tag; */ /* starting offset of us */
} xfs_dir2_data_entry_t;
/*
* Unused entry in a data block.
*
* Aligned to 8 bytes. Tag appears as the last 2 bytes and must be accessed
* using xfs_dir2_data_unused_tag_p.
*/
typedef struct xfs_dir2_data_unused {
__be16 freetag; /* XFS_DIR2_DATA_FREE_TAG */
__be16 length; /* total free length */
/* variable offset */
__be16 tag; /* starting offset of us */
} xfs_dir2_data_unused_t;
/*
* Pointer to a freespace's tag word.
*/
static inline __be16 *
xfs_dir2_data_unused_tag_p(struct xfs_dir2_data_unused *dup)
{
return (__be16 *)((char *)dup +
be16_to_cpu(dup->length) - sizeof(__be16));
}
/*
* Leaf block structures.
*
* A pure leaf block looks like the following drawing on disk:
*
* +---------------------------+
* | xfs_dir2_leaf_hdr_t |
* +---------------------------+
* | xfs_dir2_leaf_entry_t |
* | xfs_dir2_leaf_entry_t |
* | xfs_dir2_leaf_entry_t |
* | xfs_dir2_leaf_entry_t |
* | ... |
* +---------------------------+
* | xfs_dir2_data_off_t |
* | xfs_dir2_data_off_t |
* | xfs_dir2_data_off_t |
* | ... |
* +---------------------------+
* | xfs_dir2_leaf_tail_t |
* +---------------------------+
*
* The xfs_dir2_data_off_t members (bests) and tail are at the end of the block
* for single-leaf (magic = XFS_DIR2_LEAF1_MAGIC) blocks only, but not present
* for directories with separate leaf nodes and free space blocks
* (magic = XFS_DIR2_LEAFN_MAGIC).
*
* As all the entries are variable size structures the accessors below should
* be used to iterate over them.
*/
/*
* Offset of the leaf/node space. First block in this space
* is the btree root.
*/
#define XFS_DIR2_LEAF_SPACE 1
#define XFS_DIR2_LEAF_OFFSET (XFS_DIR2_LEAF_SPACE * XFS_DIR2_SPACE_SIZE)
#define XFS_DIR2_LEAF_FIRSTDB(mp) \
xfs_dir2_byte_to_db(mp, XFS_DIR2_LEAF_OFFSET)
/*
* Leaf block header.
*/
typedef struct xfs_dir2_leaf_hdr {
xfs_da_blkinfo_t info; /* header for da routines */
__be16 count; /* count of entries */
__be16 stale; /* count of stale entries */
} xfs_dir2_leaf_hdr_t;
struct xfs_dir3_leaf_hdr {
struct xfs_da3_blkinfo info; /* header for da routines */
__be16 count; /* count of entries */
__be16 stale; /* count of stale entries */
xfs: fix implicit padding in directory and attr CRC formats Michael L. Semon has been testing CRC patches on a 32 bit system and been seeing assert failures in the directory code from xfs/080. Thanks to Michael's heroic efforts with printk debugging, we found that the problem was that the last free space being left in the directory structure was too small to fit a unused tag structure and it was being corrupted and attempting to log a region out of bounds. Hence the assert failure looked something like: ..... #5 calling xfs_dir2_data_log_unused() 36 32 #1 4092 4095 4096 #2 8182 8183 4096 XFS: Assertion failed: first <= last && last < BBTOB(bp->b_length), file: fs/xfs/xfs_trans_buf.c, line: 568 Where #1 showed the first region of the dup being logged (i.e. the last 4 bytes of a directory buffer) and #2 shows the corrupt values being calculated from the length of the dup entry which overflowed the size of the buffer. It turns out that the problem was not in the logging code, nor in the freespace handling code. It is an initial condition bug that only shows up on 32 bit systems. When a new buffer is initialised, where's the freespace that is set up: [ 172.316249] calling xfs_dir2_leaf_addname() from xfs_dir_createname() [ 172.316346] #9 calling xfs_dir2_data_log_unused() [ 172.316351] #1 calling xfs_trans_log_buf() 60 63 4096 [ 172.316353] #2 calling xfs_trans_log_buf() 4094 4095 4096 Note the offset of the first region being logged? It's 60 bytes into the buffer. Once I saw that, I pretty much knew that the bug was going to be caused by this. Essentially, all direct entries are rounded to 8 bytes in length, and all entries start with an 8 byte alignment. This means that we can decode inplace as variables are naturally aligned. With the directory data supposedly starting on a 8 byte boundary, and all entries padded to 8 bytes, the minimum freespace in a directory block is supposed to be 8 bytes, which is large enough to fit a unused data entry structure (6 bytes in size). The fact we only have 4 bytes of free space indicates a directory data block alignment problem. And what do you know - there's an implicit hole in the directory data block header for the CRC format, which means the header is 60 byte on 32 bit intel systems and 64 bytes on 64 bit systems. Needs padding. And while looking at the structures, I found the same problem in the attr leaf header. Fix them both. Note that this only affects 32 bit systems with CRCs enabled. Everything else is just fine. Note that CRC enabled filesystems created before this fix on such systems will not be readable with this fix applied. Reported-by: Michael L. Semon <mlsemon35@gmail.com> Debugged-by: Michael L. Semon <mlsemon35@gmail.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Ben Myers <bpm@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com> (cherry picked from commit 8a1fd2950e1fe267e11fc8c85dcaa6b023b51b60)
2013-06-12 10:19:07 +08:00
__be32 pad; /* 64 bit alignment */
};
struct xfs_dir3_icleaf_hdr {
__uint32_t forw;
__uint32_t back;
__uint16_t magic;
__uint16_t count;
__uint16_t stale;
};
/*
* Leaf block entry.
*/
typedef struct xfs_dir2_leaf_entry {
__be32 hashval; /* hash value of name */
__be32 address; /* address of data entry */
} xfs_dir2_leaf_entry_t;
/*
* Leaf block tail.
*/
typedef struct xfs_dir2_leaf_tail {
__be32 bestcount;
} xfs_dir2_leaf_tail_t;
/*
* Leaf block.
*/
typedef struct xfs_dir2_leaf {
xfs_dir2_leaf_hdr_t hdr; /* leaf header */
xfs_dir2_leaf_entry_t __ents[]; /* entries */
} xfs_dir2_leaf_t;
struct xfs_dir3_leaf {
struct xfs_dir3_leaf_hdr hdr; /* leaf header */
struct xfs_dir2_leaf_entry __ents[]; /* entries */
};
#define XFS_DIR3_LEAF_CRC_OFF offsetof(struct xfs_dir3_leaf_hdr, info.crc)
/*
* Get address of the bestcount field in the single-leaf block.
*/
static inline struct xfs_dir2_leaf_tail *
xfs_dir2_leaf_tail_p(struct xfs_mount *mp, struct xfs_dir2_leaf *lp)
{
return (struct xfs_dir2_leaf_tail *)
((char *)lp + mp->m_dirblksize -
sizeof(struct xfs_dir2_leaf_tail));
}
/*
* Get address of the bests array in the single-leaf block.
*/
static inline __be16 *
xfs_dir2_leaf_bests_p(struct xfs_dir2_leaf_tail *ltp)
{
return (__be16 *)ltp - be32_to_cpu(ltp->bestcount);
}
/*
* DB blocks here are logical directory block numbers, not filesystem blocks.
*/
/*
* Convert dataptr to byte in file space
*/
static inline xfs_dir2_off_t
xfs_dir2_dataptr_to_byte(struct xfs_mount *mp, xfs_dir2_dataptr_t dp)
{
return (xfs_dir2_off_t)dp << XFS_DIR2_DATA_ALIGN_LOG;
}
/*
* Convert byte in file space to dataptr. It had better be aligned.
*/
static inline xfs_dir2_dataptr_t
xfs_dir2_byte_to_dataptr(struct xfs_mount *mp, xfs_dir2_off_t by)
{
return (xfs_dir2_dataptr_t)(by >> XFS_DIR2_DATA_ALIGN_LOG);
}
/*
* Convert byte in space to (DB) block
*/
static inline xfs_dir2_db_t
xfs_dir2_byte_to_db(struct xfs_mount *mp, xfs_dir2_off_t by)
{
return (xfs_dir2_db_t)
(by >> (mp->m_sb.sb_blocklog + mp->m_sb.sb_dirblklog));
}
/*
* Convert dataptr to a block number
*/
static inline xfs_dir2_db_t
xfs_dir2_dataptr_to_db(struct xfs_mount *mp, xfs_dir2_dataptr_t dp)
{
return xfs_dir2_byte_to_db(mp, xfs_dir2_dataptr_to_byte(mp, dp));
}
/*
* Convert byte in space to offset in a block
*/
static inline xfs_dir2_data_aoff_t
xfs_dir2_byte_to_off(struct xfs_mount *mp, xfs_dir2_off_t by)
{
return (xfs_dir2_data_aoff_t)(by &
((1 << (mp->m_sb.sb_blocklog + mp->m_sb.sb_dirblklog)) - 1));
}
/*
* Convert dataptr to a byte offset in a block
*/
static inline xfs_dir2_data_aoff_t
xfs_dir2_dataptr_to_off(struct xfs_mount *mp, xfs_dir2_dataptr_t dp)
{
return xfs_dir2_byte_to_off(mp, xfs_dir2_dataptr_to_byte(mp, dp));
}
/*
* Convert block and offset to byte in space
*/
static inline xfs_dir2_off_t
xfs_dir2_db_off_to_byte(struct xfs_mount *mp, xfs_dir2_db_t db,
xfs_dir2_data_aoff_t o)
{
return ((xfs_dir2_off_t)db <<
(mp->m_sb.sb_blocklog + mp->m_sb.sb_dirblklog)) + o;
}
/*
* Convert block (DB) to block (dablk)
*/
static inline xfs_dablk_t
xfs_dir2_db_to_da(struct xfs_mount *mp, xfs_dir2_db_t db)
{
return (xfs_dablk_t)(db << mp->m_sb.sb_dirblklog);
}
/*
* Convert byte in space to (DA) block
*/
static inline xfs_dablk_t
xfs_dir2_byte_to_da(struct xfs_mount *mp, xfs_dir2_off_t by)
{
return xfs_dir2_db_to_da(mp, xfs_dir2_byte_to_db(mp, by));
}
/*
* Convert block and offset to dataptr
*/
static inline xfs_dir2_dataptr_t
xfs_dir2_db_off_to_dataptr(struct xfs_mount *mp, xfs_dir2_db_t db,
xfs_dir2_data_aoff_t o)
{
return xfs_dir2_byte_to_dataptr(mp, xfs_dir2_db_off_to_byte(mp, db, o));
}
/*
* Convert block (dablk) to block (DB)
*/
static inline xfs_dir2_db_t
xfs_dir2_da_to_db(struct xfs_mount *mp, xfs_dablk_t da)
{
return (xfs_dir2_db_t)(da >> mp->m_sb.sb_dirblklog);
}
/*
* Convert block (dablk) to byte offset in space
*/
static inline xfs_dir2_off_t
xfs_dir2_da_to_byte(struct xfs_mount *mp, xfs_dablk_t da)
{
return xfs_dir2_db_off_to_byte(mp, xfs_dir2_da_to_db(mp, da), 0);
}
/*
* Free space block defintions for the node format.
*/
/*
* Offset of the freespace index.
*/
#define XFS_DIR2_FREE_SPACE 2
#define XFS_DIR2_FREE_OFFSET (XFS_DIR2_FREE_SPACE * XFS_DIR2_SPACE_SIZE)
#define XFS_DIR2_FREE_FIRSTDB(mp) \
xfs_dir2_byte_to_db(mp, XFS_DIR2_FREE_OFFSET)
typedef struct xfs_dir2_free_hdr {
__be32 magic; /* XFS_DIR2_FREE_MAGIC */
__be32 firstdb; /* db of first entry */
__be32 nvalid; /* count of valid entries */
__be32 nused; /* count of used entries */
} xfs_dir2_free_hdr_t;
typedef struct xfs_dir2_free {
xfs_dir2_free_hdr_t hdr; /* block header */
__be16 bests[]; /* best free counts */
/* unused entries are -1 */
} xfs_dir2_free_t;
struct xfs_dir3_free_hdr {
struct xfs_dir3_blk_hdr hdr;
__be32 firstdb; /* db of first entry */
__be32 nvalid; /* count of valid entries */
__be32 nused; /* count of used entries */
xfs: fix implicit padding in directory and attr CRC formats Michael L. Semon has been testing CRC patches on a 32 bit system and been seeing assert failures in the directory code from xfs/080. Thanks to Michael's heroic efforts with printk debugging, we found that the problem was that the last free space being left in the directory structure was too small to fit a unused tag structure and it was being corrupted and attempting to log a region out of bounds. Hence the assert failure looked something like: ..... #5 calling xfs_dir2_data_log_unused() 36 32 #1 4092 4095 4096 #2 8182 8183 4096 XFS: Assertion failed: first <= last && last < BBTOB(bp->b_length), file: fs/xfs/xfs_trans_buf.c, line: 568 Where #1 showed the first region of the dup being logged (i.e. the last 4 bytes of a directory buffer) and #2 shows the corrupt values being calculated from the length of the dup entry which overflowed the size of the buffer. It turns out that the problem was not in the logging code, nor in the freespace handling code. It is an initial condition bug that only shows up on 32 bit systems. When a new buffer is initialised, where's the freespace that is set up: [ 172.316249] calling xfs_dir2_leaf_addname() from xfs_dir_createname() [ 172.316346] #9 calling xfs_dir2_data_log_unused() [ 172.316351] #1 calling xfs_trans_log_buf() 60 63 4096 [ 172.316353] #2 calling xfs_trans_log_buf() 4094 4095 4096 Note the offset of the first region being logged? It's 60 bytes into the buffer. Once I saw that, I pretty much knew that the bug was going to be caused by this. Essentially, all direct entries are rounded to 8 bytes in length, and all entries start with an 8 byte alignment. This means that we can decode inplace as variables are naturally aligned. With the directory data supposedly starting on a 8 byte boundary, and all entries padded to 8 bytes, the minimum freespace in a directory block is supposed to be 8 bytes, which is large enough to fit a unused data entry structure (6 bytes in size). The fact we only have 4 bytes of free space indicates a directory data block alignment problem. And what do you know - there's an implicit hole in the directory data block header for the CRC format, which means the header is 60 byte on 32 bit intel systems and 64 bytes on 64 bit systems. Needs padding. And while looking at the structures, I found the same problem in the attr leaf header. Fix them both. Note that this only affects 32 bit systems with CRCs enabled. Everything else is just fine. Note that CRC enabled filesystems created before this fix on such systems will not be readable with this fix applied. Reported-by: Michael L. Semon <mlsemon35@gmail.com> Debugged-by: Michael L. Semon <mlsemon35@gmail.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Ben Myers <bpm@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com> (cherry picked from commit 8a1fd2950e1fe267e11fc8c85dcaa6b023b51b60)
2013-06-12 10:19:07 +08:00
__be32 pad; /* 64 bit alignment */
};
struct xfs_dir3_free {
struct xfs_dir3_free_hdr hdr;
__be16 bests[]; /* best free counts */
/* unused entries are -1 */
};
#define XFS_DIR3_FREE_CRC_OFF offsetof(struct xfs_dir3_free, hdr.hdr.crc)
/*
* In core version of the free block header, abstracted away from on-disk format
* differences. Use this in the code, and convert to/from the disk version using
* xfs_dir3_free_hdr_from_disk/xfs_dir3_free_hdr_to_disk.
*/
struct xfs_dir3_icfree_hdr {
__uint32_t magic;
__uint32_t firstdb;
__uint32_t nvalid;
__uint32_t nused;
};
/*
* Single block format.
*
* The single block format looks like the following drawing on disk:
*
* +-------------------------------------------------+
* | xfs_dir2_data_hdr_t |
* +-------------------------------------------------+
* | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
* | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t |
* | xfs_dir2_data_entry_t OR xfs_dir2_data_unused_t :
* | ... |
* +-------------------------------------------------+
* | unused space |
* +-------------------------------------------------+
* | ... |
* | xfs_dir2_leaf_entry_t |
* | xfs_dir2_leaf_entry_t |
* +-------------------------------------------------+
* | xfs_dir2_block_tail_t |
* +-------------------------------------------------+
*
* As all the entries are variable size structures the accessors below should
* be used to iterate over them.
*/
typedef struct xfs_dir2_block_tail {
__be32 count; /* count of leaf entries */
__be32 stale; /* count of stale lf entries */
} xfs_dir2_block_tail_t;
/*
* Pointer to the leaf header embedded in a data block (1-block format)
*/
static inline struct xfs_dir2_block_tail *
xfs_dir2_block_tail_p(struct xfs_mount *mp, struct xfs_dir2_data_hdr *hdr)
{
return ((struct xfs_dir2_block_tail *)
((char *)hdr + mp->m_dirblksize)) - 1;
}
/*
* Pointer to the leaf entries embedded in a data block (1-block format)
*/
static inline struct xfs_dir2_leaf_entry *
xfs_dir2_block_leaf_p(struct xfs_dir2_block_tail *btp)
{
return ((struct xfs_dir2_leaf_entry *)btp) - be32_to_cpu(btp->count);
}
/*
* Attribute storage layout
*
* Attribute lists are structured around Btrees where all the data
* elements are in the leaf nodes. Attribute names are hashed into an int,
* then that int is used as the index into the Btree. Since the hashval
* of an attribute name may not be unique, we may have duplicate keys. The
* internal links in the Btree are logical block offsets into the file.
*
*========================================================================
* Attribute structure when equal to XFS_LBSIZE(mp) bytes.
*========================================================================
*
* Struct leaf_entry's are packed from the top. Name/values grow from the
* bottom but are not packed. The freemap contains run-length-encoded entries
* for the free bytes after the leaf_entry's, but only the N largest such,
* smaller runs are dropped. When the freemap doesn't show enough space
* for an allocation, we compact the name/value area and try again. If we
* still don't have enough space, then we have to split the block. The
* name/value structs (both local and remote versions) must be 32bit aligned.
*
* Since we have duplicate hash keys, for each key that matches, compare
* the actual name string. The root and intermediate node search always
* takes the first-in-the-block key match found, so we should only have
* to work "forw"ard. If none matches, continue with the "forw"ard leaf
* nodes until the hash key changes or the attribute name is found.
*
* We store the fact that an attribute is a ROOT/USER/SECURE attribute in
* the leaf_entry. The namespaces are independent only because we also look
* at the namespace bit when we are looking for a matching attribute name.
*
* We also store an "incomplete" bit in the leaf_entry. It shows that an
* attribute is in the middle of being created and should not be shown to
* the user if we crash during the time that the bit is set. We clear the
* bit when we have finished setting up the attribute. We do this because
* we cannot create some large attributes inside a single transaction, and we
* need some indication that we weren't finished if we crash in the middle.
*/
#define XFS_ATTR_LEAF_MAPSIZE 3 /* how many freespace slots */
typedef struct xfs_attr_leaf_map { /* RLE map of free bytes */
__be16 base; /* base of free region */
__be16 size; /* length of free region */
} xfs_attr_leaf_map_t;
typedef struct xfs_attr_leaf_hdr { /* constant-structure header block */
xfs_da_blkinfo_t info; /* block type, links, etc. */
__be16 count; /* count of active leaf_entry's */
__be16 usedbytes; /* num bytes of names/values stored */
__be16 firstused; /* first used byte in name area */
__u8 holes; /* != 0 if blk needs compaction */
__u8 pad1;
xfs_attr_leaf_map_t freemap[XFS_ATTR_LEAF_MAPSIZE];
/* N largest free regions */
} xfs_attr_leaf_hdr_t;
typedef struct xfs_attr_leaf_entry { /* sorted on key, not name */
__be32 hashval; /* hash value of name */
__be16 nameidx; /* index into buffer of name/value */
__u8 flags; /* LOCAL/ROOT/SECURE/INCOMPLETE flag */
__u8 pad2; /* unused pad byte */
} xfs_attr_leaf_entry_t;
typedef struct xfs_attr_leaf_name_local {
__be16 valuelen; /* number of bytes in value */
__u8 namelen; /* length of name bytes */
__u8 nameval[1]; /* name/value bytes */
} xfs_attr_leaf_name_local_t;
typedef struct xfs_attr_leaf_name_remote {
__be32 valueblk; /* block number of value bytes */
__be32 valuelen; /* number of bytes in value */
__u8 namelen; /* length of name bytes */
__u8 name[1]; /* name bytes */
} xfs_attr_leaf_name_remote_t;
typedef struct xfs_attr_leafblock {
xfs_attr_leaf_hdr_t hdr; /* constant-structure header block */
xfs_attr_leaf_entry_t entries[1]; /* sorted on key, not name */
xfs_attr_leaf_name_local_t namelist; /* grows from bottom of buf */
xfs_attr_leaf_name_remote_t valuelist; /* grows from bottom of buf */
} xfs_attr_leafblock_t;
/*
* CRC enabled leaf structures. Called "version 3" structures to match the
* version number of the directory and dablk structures for this feature, and
* attr2 is already taken by the variable inode attribute fork size feature.
*/
struct xfs_attr3_leaf_hdr {
struct xfs_da3_blkinfo info;
__be16 count;
__be16 usedbytes;
__be16 firstused;
__u8 holes;
__u8 pad1;
struct xfs_attr_leaf_map freemap[XFS_ATTR_LEAF_MAPSIZE];
__be32 pad2; /* 64 bit alignment */
};
#define XFS_ATTR3_LEAF_CRC_OFF (offsetof(struct xfs_attr3_leaf_hdr, info.crc))
struct xfs_attr3_leafblock {
struct xfs_attr3_leaf_hdr hdr;
struct xfs_attr_leaf_entry entries[1];
/*
* The rest of the block contains the following structures after the
* leaf entries, growing from the bottom up. The variables are never
* referenced, the locations accessed purely from helper functions.
*
* struct xfs_attr_leaf_name_local
* struct xfs_attr_leaf_name_remote
*/
};
/*
* incore, neutral version of the attribute leaf header
*/
struct xfs_attr3_icleaf_hdr {
__uint32_t forw;
__uint32_t back;
__uint16_t magic;
__uint16_t count;
__uint16_t usedbytes;
__uint16_t firstused;
__u8 holes;
struct {
__uint16_t base;
__uint16_t size;
} freemap[XFS_ATTR_LEAF_MAPSIZE];
};
/*
* Flags used in the leaf_entry[i].flags field.
* NOTE: the INCOMPLETE bit must not collide with the flags bits specified
* on the system call, they are "or"ed together for various operations.
*/
#define XFS_ATTR_LOCAL_BIT 0 /* attr is stored locally */
#define XFS_ATTR_ROOT_BIT 1 /* limit access to trusted attrs */
#define XFS_ATTR_SECURE_BIT 2 /* limit access to secure attrs */
#define XFS_ATTR_INCOMPLETE_BIT 7 /* attr in middle of create/delete */
#define XFS_ATTR_LOCAL (1 << XFS_ATTR_LOCAL_BIT)
#define XFS_ATTR_ROOT (1 << XFS_ATTR_ROOT_BIT)
#define XFS_ATTR_SECURE (1 << XFS_ATTR_SECURE_BIT)
#define XFS_ATTR_INCOMPLETE (1 << XFS_ATTR_INCOMPLETE_BIT)
/*
* Conversion macros for converting namespace bits from argument flags
* to ondisk flags.
*/
#define XFS_ATTR_NSP_ARGS_MASK (ATTR_ROOT | ATTR_SECURE)
#define XFS_ATTR_NSP_ONDISK_MASK (XFS_ATTR_ROOT | XFS_ATTR_SECURE)
#define XFS_ATTR_NSP_ONDISK(flags) ((flags) & XFS_ATTR_NSP_ONDISK_MASK)
#define XFS_ATTR_NSP_ARGS(flags) ((flags) & XFS_ATTR_NSP_ARGS_MASK)
#define XFS_ATTR_NSP_ARGS_TO_ONDISK(x) (((x) & ATTR_ROOT ? XFS_ATTR_ROOT : 0) |\
((x) & ATTR_SECURE ? XFS_ATTR_SECURE : 0))
#define XFS_ATTR_NSP_ONDISK_TO_ARGS(x) (((x) & XFS_ATTR_ROOT ? ATTR_ROOT : 0) |\
((x) & XFS_ATTR_SECURE ? ATTR_SECURE : 0))
/*
* Alignment for namelist and valuelist entries (since they are mixed
* there can be only one alignment value)
*/
#define XFS_ATTR_LEAF_NAME_ALIGN ((uint)sizeof(xfs_dablk_t))
static inline int
xfs_attr3_leaf_hdr_size(struct xfs_attr_leafblock *leafp)
{
if (leafp->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC))
return sizeof(struct xfs_attr3_leaf_hdr);
return sizeof(struct xfs_attr_leaf_hdr);
}
static inline struct xfs_attr_leaf_entry *
xfs_attr3_leaf_entryp(xfs_attr_leafblock_t *leafp)
{
if (leafp->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC))
return &((struct xfs_attr3_leafblock *)leafp)->entries[0];
return &leafp->entries[0];
}
/*
* Cast typed pointers for "local" and "remote" name/value structs.
*/
static inline char *
xfs_attr3_leaf_name(xfs_attr_leafblock_t *leafp, int idx)
{
struct xfs_attr_leaf_entry *entries = xfs_attr3_leaf_entryp(leafp);
return &((char *)leafp)[be16_to_cpu(entries[idx].nameidx)];
}
static inline xfs_attr_leaf_name_remote_t *
xfs_attr3_leaf_name_remote(xfs_attr_leafblock_t *leafp, int idx)
{
return (xfs_attr_leaf_name_remote_t *)xfs_attr3_leaf_name(leafp, idx);
}
static inline xfs_attr_leaf_name_local_t *
xfs_attr3_leaf_name_local(xfs_attr_leafblock_t *leafp, int idx)
{
return (xfs_attr_leaf_name_local_t *)xfs_attr3_leaf_name(leafp, idx);
}
/*
* Calculate total bytes used (including trailing pad for alignment) for
* a "local" name/value structure, a "remote" name/value structure, and
* a pointer which might be either.
*/
static inline int xfs_attr_leaf_entsize_remote(int nlen)
{
return ((uint)sizeof(xfs_attr_leaf_name_remote_t) - 1 + (nlen) + \
XFS_ATTR_LEAF_NAME_ALIGN - 1) & ~(XFS_ATTR_LEAF_NAME_ALIGN - 1);
}
static inline int xfs_attr_leaf_entsize_local(int nlen, int vlen)
{
return ((uint)sizeof(xfs_attr_leaf_name_local_t) - 1 + (nlen) + (vlen) +
XFS_ATTR_LEAF_NAME_ALIGN - 1) & ~(XFS_ATTR_LEAF_NAME_ALIGN - 1);
}
static inline int xfs_attr_leaf_entsize_local_max(int bsize)
{
return (((bsize) >> 1) + ((bsize) >> 2));
}
/*
* Remote attribute block format definition
*
* There is one of these headers per filesystem block in a remote attribute.
* This is done to ensure there is a 1:1 mapping between the attribute value
* length and the number of blocks needed to store the attribute. This makes the
* verification of a buffer a little more complex, but greatly simplifies the
* allocation, reading and writing of these attributes as we don't have to guess
* the number of blocks needed to store the attribute data.
*/
#define XFS_ATTR3_RMT_MAGIC 0x5841524d /* XARM */
struct xfs_attr3_rmt_hdr {
__be32 rm_magic;
__be32 rm_offset;
__be32 rm_bytes;
__be32 rm_crc;
uuid_t rm_uuid;
__be64 rm_owner;
__be64 rm_blkno;
__be64 rm_lsn;
};
#define XFS_ATTR3_RMT_CRC_OFF offsetof(struct xfs_attr3_rmt_hdr, rm_crc)
#define XFS_ATTR3_RMT_BUF_SPACE(mp, bufsize) \
((bufsize) - (xfs_sb_version_hascrc(&(mp)->m_sb) ? \
sizeof(struct xfs_attr3_rmt_hdr) : 0))
#endif /* __XFS_DA_FORMAT_H__ */