OpenCloudOS-Kernel/fs/btrfs/ctree.h

1985 lines
64 KiB
C
Raw Normal View History

/*
* Copyright (C) 2007 Oracle. 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 v2 as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will 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 to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*/
#ifndef __BTRFS_CTREE__
#define __BTRFS_CTREE__
#include <linux/version.h>
#include <linux/mm.h>
#include <linux/highmem.h>
#include <linux/fs.h>
#include <linux/completion.h>
#include <linux/backing-dev.h>
#include <linux/wait.h>
#include <asm/kmap_types.h>
#include "extent_io.h"
#include "extent_map.h"
#include "async-thread.h"
struct btrfs_trans_handle;
struct btrfs_transaction;
extern struct kmem_cache *btrfs_trans_handle_cachep;
extern struct kmem_cache *btrfs_transaction_cachep;
extern struct kmem_cache *btrfs_bit_radix_cachep;
extern struct kmem_cache *btrfs_path_cachep;
struct btrfs_ordered_sum;
#define BTRFS_MAGIC "_BDRfS_M"
#define BTRFS_ACL_NOT_CACHED ((void *)-1)
#ifdef CONFIG_LOCKDEP
# define BTRFS_MAX_LEVEL 7
#else
# define BTRFS_MAX_LEVEL 8
#endif
/* holds pointers to all of the tree roots */
#define BTRFS_ROOT_TREE_OBJECTID 1ULL
/* stores information about which extents are in use, and reference counts */
#define BTRFS_EXTENT_TREE_OBJECTID 2ULL
/*
* chunk tree stores translations from logical -> physical block numbering
* the super block points to the chunk tree
*/
#define BTRFS_CHUNK_TREE_OBJECTID 3ULL
/*
* stores information about which areas of a given device are in use.
* one per device. The tree of tree roots points to the device tree
*/
#define BTRFS_DEV_TREE_OBJECTID 4ULL
/* one per subvolume, storing files and directories */
#define BTRFS_FS_TREE_OBJECTID 5ULL
/* directory objectid inside the root tree */
#define BTRFS_ROOT_TREE_DIR_OBJECTID 6ULL
/* orhpan objectid for tracking unlinked/truncated files */
#define BTRFS_ORPHAN_OBJECTID -5ULL
/* does write ahead logging to speed up fsyncs */
#define BTRFS_TREE_LOG_OBJECTID -6ULL
#define BTRFS_TREE_LOG_FIXUP_OBJECTID -7ULL
/* for space balancing */
#define BTRFS_TREE_RELOC_OBJECTID -8ULL
#define BTRFS_DATA_RELOC_TREE_OBJECTID -9ULL
/* dummy objectid represents multiple objectids */
#define BTRFS_MULTIPLE_OBJECTIDS -255ULL
/*
* All files have objectids in this range.
*/
#define BTRFS_FIRST_FREE_OBJECTID 256ULL
#define BTRFS_LAST_FREE_OBJECTID -256ULL
#define BTRFS_FIRST_CHUNK_TREE_OBJECTID 256ULL
/*
* the device items go into the chunk tree. The key is in the form
* [ 1 BTRFS_DEV_ITEM_KEY device_id ]
*/
#define BTRFS_DEV_ITEMS_OBJECTID 1ULL
/*
* we can actually store much bigger names, but lets not confuse the rest
* of linux
*/
#define BTRFS_NAME_LEN 255
/* 32 bytes in various csum fields */
#define BTRFS_CSUM_SIZE 32
/* four bytes for CRC32 */
#define BTRFS_CRC32_SIZE 4
#define BTRFS_EMPTY_DIR_SIZE 0
#define BTRFS_FT_UNKNOWN 0
#define BTRFS_FT_REG_FILE 1
#define BTRFS_FT_DIR 2
#define BTRFS_FT_CHRDEV 3
#define BTRFS_FT_BLKDEV 4
#define BTRFS_FT_FIFO 5
#define BTRFS_FT_SOCK 6
#define BTRFS_FT_SYMLINK 7
#define BTRFS_FT_XATTR 8
#define BTRFS_FT_MAX 9
/*
* the key defines the order in the tree, and so it also defines (optimal)
* block layout. objectid corresonds to the inode number. The flags
* tells us things about the object, and is a kind of stream selector.
* so for a given inode, keys with flags of 1 might refer to the inode
* data, flags of 2 may point to file data in the btree and flags == 3
* may point to extents.
*
* offset is the starting byte offset for this key in the stream.
*
* btrfs_disk_key is in disk byte order. struct btrfs_key is always
* in cpu native order. Otherwise they are identical and their sizes
* should be the same (ie both packed)
*/
struct btrfs_disk_key {
__le64 objectid;
u8 type;
__le64 offset;
} __attribute__ ((__packed__));
struct btrfs_key {
u64 objectid;
u8 type;
u64 offset;
} __attribute__ ((__packed__));
struct btrfs_mapping_tree {
struct extent_map_tree map_tree;
};
#define BTRFS_UUID_SIZE 16
struct btrfs_dev_item {
/* the internal btrfs device id */
__le64 devid;
/* size of the device */
__le64 total_bytes;
/* bytes used */
__le64 bytes_used;
/* optimal io alignment for this device */
__le32 io_align;
/* optimal io width for this device */
__le32 io_width;
/* minimal io size for this device */
__le32 sector_size;
/* type and info about this device */
__le64 type;
/* grouping information for allocation decisions */
__le32 dev_group;
/* seek speed 0-100 where 100 is fastest */
u8 seek_speed;
/* bandwidth 0-100 where 100 is fastest */
u8 bandwidth;
/* btrfs generated uuid for this device */
u8 uuid[BTRFS_UUID_SIZE];
} __attribute__ ((__packed__));
struct btrfs_stripe {
__le64 devid;
__le64 offset;
u8 dev_uuid[BTRFS_UUID_SIZE];
} __attribute__ ((__packed__));
struct btrfs_chunk {
/* size of this chunk in bytes */
__le64 length;
/* objectid of the root referencing this chunk */
__le64 owner;
__le64 stripe_len;
__le64 type;
/* optimal io alignment for this chunk */
__le32 io_align;
/* optimal io width for this chunk */
__le32 io_width;
/* minimal io size for this chunk */
__le32 sector_size;
/* 2^16 stripes is quite a lot, a second limit is the size of a single
* item in the btree
*/
__le16 num_stripes;
/* sub stripes only matter for raid10 */
__le16 sub_stripes;
struct btrfs_stripe stripe;
/* additional stripes go here */
} __attribute__ ((__packed__));
static inline unsigned long btrfs_chunk_item_size(int num_stripes)
{
BUG_ON(num_stripes == 0);
return sizeof(struct btrfs_chunk) +
sizeof(struct btrfs_stripe) * (num_stripes - 1);
}
#define BTRFS_FSID_SIZE 16
#define BTRFS_HEADER_FLAG_WRITTEN (1 << 0)
/*
* every tree block (leaf or node) starts with this header.
*/
struct btrfs_header {
/* these first four must match the super block */
u8 csum[BTRFS_CSUM_SIZE];
u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
__le64 bytenr; /* which block this node is supposed to live in */
__le64 flags;
/* allowed to be different from the super from here on down */
u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
__le64 generation;
__le64 owner;
__le32 nritems;
u8 level;
} __attribute__ ((__packed__));
#define BTRFS_NODEPTRS_PER_BLOCK(r) (((r)->nodesize - \
sizeof(struct btrfs_header)) / \
sizeof(struct btrfs_key_ptr))
#define __BTRFS_LEAF_DATA_SIZE(bs) ((bs) - sizeof(struct btrfs_header))
#define BTRFS_LEAF_DATA_SIZE(r) (__BTRFS_LEAF_DATA_SIZE(r->leafsize))
#define BTRFS_MAX_INLINE_DATA_SIZE(r) (BTRFS_LEAF_DATA_SIZE(r) - \
sizeof(struct btrfs_item) - \
sizeof(struct btrfs_file_extent_item))
/*
* this is a very generous portion of the super block, giving us
* room to translate 14 chunks with 3 stripes each.
*/
#define BTRFS_SYSTEM_CHUNK_ARRAY_SIZE 2048
#define BTRFS_LABEL_SIZE 256
/*
* the super block basically lists the main trees of the FS
* it currently lacks any block count etc etc
*/
struct btrfs_super_block {
u8 csum[BTRFS_CSUM_SIZE];
/* the first 4 fields must match struct btrfs_header */
u8 fsid[16]; /* FS specific uuid */
__le64 bytenr; /* this block number */
__le64 flags;
/* allowed to be different from the btrfs_header from here own down */
__le64 magic;
__le64 generation;
__le64 root;
__le64 chunk_root;
__le64 log_root;
__le64 total_bytes;
__le64 bytes_used;
__le64 root_dir_objectid;
__le64 num_devices;
__le32 sectorsize;
__le32 nodesize;
__le32 leafsize;
__le32 stripesize;
__le32 sys_chunk_array_size;
__le64 chunk_root_generation;
u8 root_level;
u8 chunk_root_level;
u8 log_root_level;
struct btrfs_dev_item dev_item;
char label[BTRFS_LABEL_SIZE];
u8 sys_chunk_array[BTRFS_SYSTEM_CHUNK_ARRAY_SIZE];
} __attribute__ ((__packed__));
/*
* A leaf is full of items. offset and size tell us where to find
* the item in the leaf (relative to the start of the data area)
*/
struct btrfs_item {
struct btrfs_disk_key key;
__le32 offset;
__le32 size;
} __attribute__ ((__packed__));
/*
* leaves have an item area and a data area:
* [item0, item1....itemN] [free space] [dataN...data1, data0]
*
* The data is separate from the items to get the keys closer together
* during searches.
*/
struct btrfs_leaf {
struct btrfs_header header;
struct btrfs_item items[];
} __attribute__ ((__packed__));
/*
* all non-leaf blocks are nodes, they hold only keys and pointers to
* other blocks
*/
struct btrfs_key_ptr {
struct btrfs_disk_key key;
__le64 blockptr;
__le64 generation;
} __attribute__ ((__packed__));
struct btrfs_node {
struct btrfs_header header;
struct btrfs_key_ptr ptrs[];
} __attribute__ ((__packed__));
/*
* btrfs_paths remember the path taken from the root down to the leaf.
* level 0 is always the leaf, and nodes[1...BTRFS_MAX_LEVEL] will point
* to any other levels that are present.
*
* The slots array records the index of the item or block pointer
* used while walking the tree.
*/
struct btrfs_path {
struct extent_buffer *nodes[BTRFS_MAX_LEVEL];
int slots[BTRFS_MAX_LEVEL];
/* if there is real range locking, this locks field will change */
int locks[BTRFS_MAX_LEVEL];
int reada;
/* keep some upper locks as we walk down */
int keep_locks;
int skip_locking;
int lowest_level;
};
/*
* items in the extent btree are used to record the objectid of the
* owner of the block and the number of references
*/
struct btrfs_extent_item {
__le32 refs;
} __attribute__ ((__packed__));
struct btrfs_extent_ref {
__le64 root;
__le64 generation;
__le64 objectid;
__le32 num_refs;
} __attribute__ ((__packed__));
/* dev extents record free space on individual devices. The owner
* field points back to the chunk allocation mapping tree that allocated
* the extent. The chunk tree uuid field is a way to double check the owner
*/
struct btrfs_dev_extent {
__le64 chunk_tree;
__le64 chunk_objectid;
__le64 chunk_offset;
__le64 length;
u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
} __attribute__ ((__packed__));
struct btrfs_inode_ref {
__le64 index;
__le16 name_len;
/* name goes here */
} __attribute__ ((__packed__));
struct btrfs_timespec {
__le64 sec;
__le32 nsec;
} __attribute__ ((__packed__));
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
typedef enum {
BTRFS_COMPRESS_NONE = 0,
BTRFS_COMPRESS_ZLIB = 1,
BTRFS_COMPRESS_LAST = 2,
} btrfs_compression_type;
/* we don't understand any encryption methods right now */
typedef enum {
BTRFS_ENCRYPTION_NONE = 0,
BTRFS_ENCRYPTION_LAST = 1,
} btrfs_encryption_type;
struct btrfs_inode_item {
/* nfs style generation number */
__le64 generation;
/* transid that last touched this inode */
__le64 transid;
__le64 size;
__le64 nbytes;
__le64 block_group;
__le32 nlink;
__le32 uid;
__le32 gid;
__le32 mode;
__le64 rdev;
__le16 flags;
__le16 compat_flags;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
struct btrfs_timespec atime;
struct btrfs_timespec ctime;
struct btrfs_timespec mtime;
struct btrfs_timespec otime;
} __attribute__ ((__packed__));
struct btrfs_dir_log_item {
__le64 end;
} __attribute__ ((__packed__));
struct btrfs_dir_item {
struct btrfs_disk_key location;
__le64 transid;
__le16 data_len;
__le16 name_len;
u8 type;
} __attribute__ ((__packed__));
struct btrfs_root_item {
struct btrfs_inode_item inode;
__le64 generation;
__le64 root_dirid;
__le64 bytenr;
__le64 byte_limit;
__le64 bytes_used;
__le64 last_snapshot;
__le32 flags;
__le32 refs;
struct btrfs_disk_key drop_progress;
u8 drop_level;
u8 level;
} __attribute__ ((__packed__));
#define BTRFS_FILE_EXTENT_INLINE 0
#define BTRFS_FILE_EXTENT_REG 1
#define BTRFS_FILE_EXTENT_PREALLOC 2
struct btrfs_file_extent_item {
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
/*
* transaction id that created this extent
*/
__le64 generation;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
/*
* max number of bytes to hold this extent in ram
* when we split a compressed extent we can't know how big
* each of the resulting pieces will be. So, this is
* an upper limit on the size of the extent in ram instead of
* an exact limit.
*/
__le64 ram_bytes;
/*
* 32 bits for the various ways we might encode the data,
* including compression and encryption. If any of these
* are set to something a given disk format doesn't understand
* it is treated like an incompat flag for reading and writing,
* but not for stat.
*/
u8 compression;
u8 encryption;
__le16 other_encoding; /* spare for later use */
/* are we inline data or a real extent? */
u8 type;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
/*
* disk space consumed by the extent, checksum blocks are included
* in these numbers
*/
__le64 disk_bytenr;
__le64 disk_num_bytes;
/*
* the logical offset in file blocks (no csums)
* this extent record is for. This allows a file extent to point
* into the middle of an existing extent on disk, sharing it
* between two snapshots (useful if some bytes in the middle of the
* extent have changed
*/
__le64 offset;
/*
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
* the logical number of file blocks (no csums included). This
* always reflects the size uncompressed and without encoding.
*/
__le64 num_bytes;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
} __attribute__ ((__packed__));
struct btrfs_csum_item {
u8 csum;
} __attribute__ ((__packed__));
/* different types of block groups (and chunks) */
#define BTRFS_BLOCK_GROUP_DATA (1 << 0)
#define BTRFS_BLOCK_GROUP_SYSTEM (1 << 1)
#define BTRFS_BLOCK_GROUP_METADATA (1 << 2)
#define BTRFS_BLOCK_GROUP_RAID0 (1 << 3)
#define BTRFS_BLOCK_GROUP_RAID1 (1 << 4)
#define BTRFS_BLOCK_GROUP_DUP (1 << 5)
#define BTRFS_BLOCK_GROUP_RAID10 (1 << 6)
struct btrfs_block_group_item {
__le64 used;
__le64 chunk_objectid;
__le64 flags;
} __attribute__ ((__packed__));
struct btrfs_space_info {
u64 flags;
u64 total_bytes;
u64 bytes_used;
u64 bytes_pinned;
u64 bytes_reserved;
u64 bytes_readonly;
int full;
int force_alloc;
struct list_head list;
Btrfs: free space accounting redo 1) replace the per fs_info extent_io_tree that tracked free space with two rb-trees per block group to track free space areas via offset and size. The reason to do this is because most allocations come with a hint byte where to start, so we can usually find a chunk of free space at that hint byte to satisfy the allocation and get good space packing. If we cannot find free space at or after the given offset we fall back on looking for a chunk of the given size as close to that given offset as possible. When we fall back on the size search we also try to find a slot as close to the size we want as possible, to avoid breaking small chunks off of huge areas if possible. 2) remove the extent_io_tree that tracked the block group cache from fs_info and replaced it with an rb-tree thats tracks block group cache via offset. also added a per space_info list that tracks the block group cache for the particular space so we can lookup related block groups easily. 3) cleaned up the allocation code to make it a little easier to read and a little less complicated. Basically there are 3 steps, first look from our provided hint. If we couldn't find from that given hint, start back at our original search start and look for space from there. If that fails try to allocate space if we can and start looking again. If not we're screwed and need to start over again. 4) small fixes. there were some issues in volumes.c where we wouldn't allocate the rest of the disk. fixed cow_file_range to actually pass the alloc_hint, which has helped a good bit in making the fs_mark test I run have semi-normal results as we run out of space. Generally with data allocations we don't track where we last allocated from, so everytime we did a data allocation we'd search through every block group that we have looking for free space. Now searching a block group with no free space isn't terribly time consuming, it was causing a slight degradation as we got more data block groups. The alloc_hint has fixed this slight degredation and made things semi-normal. There is still one nagging problem I'm working on where we will get ENOSPC when there is definitely plenty of space. This only happens with metadata allocations, and only when we are almost full. So you generally hit the 85% mark first, but sometimes you'll hit the BUG before you hit the 85% wall. I'm still tracking it down, but until then this seems to be pretty stable and make a significant performance gain. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-09-24 01:14:11 +08:00
/* for block groups in our same type */
struct list_head block_groups;
spinlock_t lock;
Btrfs: fix enospc when there is plenty of space So there is an odd case where we can possibly return -ENOSPC when there is in fact space to be had. It only happens with Metadata writes, and happens _very_ infrequently. What has to happen is we have to allocate have allocated out of the first logical byte on the disk, which would set last_alloc to first_logical_byte(root, 0), so search_start == orig_search_start. We then need to allocate for normal metadata, so BTRFS_BLOCK_GROUP_METADATA | BTRFS_BLOCK_GROUP_DUP. We will do a block lookup for the given search_start, block_group_bits() won't match and we'll go to choose another block group. However because search_start matches orig_search_start we go to see if we can allocate a chunk. If we are in the situation that we cannot allocate a chunk, we fail and ENOSPC. This is kind of a big flaw of the way find_free_extent works, as it along with find_free_space loop through _all_ of the block groups, not just the ones that we want to allocate out of. This patch completely kills find_free_space and rolls it into find_free_extent. I've introduced a sort of state machine into this, which will make it easier to get cache miss information out of the allocator, and will work well with my locking changes. The basic flow is this: We have the variable loop which is 0, meaning we are in the hint phase. We lookup the block group for the hint, and lookup the space_info for what we want to allocate out of. If the block group we were pointed at by the hint either isn't of the correct type, or just doesn't have the space we need, we set head to space_info->block_groups, so we start at the beginning of the block groups for this particular space info, and loop through. This is also where we add the empty_cluster to total_needed. At this point loop is set to 1 and we just loop through all of the block groups for this particular space_info looking for the space we need, just as find_free_space would have done, except we only hit the block groups we want and not _all_ of the block groups. If we come full circle we see if we can allocate a chunk. If we cannot of course we exit with -ENOSPC and we are good. If not we start over at space_info->block_groups and loop through again, with loop == 2. If we come full circle and haven't found what we need then we exit with -ENOSPC. I've been running this for a couple of days now and it seems stable, and I haven't yet hit a -ENOSPC when there was plenty of space left. Also I've made a groups_sem to handle the group list for the space_info. This is part of my locking changes, but is relatively safe and seems better than holding the space_info spinlock over that entire search time. Thanks, Signed-off-by: Josef Bacik <jbacik@redhat.com>
2008-10-30 02:49:05 +08:00
struct rw_semaphore groups_sem;
Btrfs: free space accounting redo 1) replace the per fs_info extent_io_tree that tracked free space with two rb-trees per block group to track free space areas via offset and size. The reason to do this is because most allocations come with a hint byte where to start, so we can usually find a chunk of free space at that hint byte to satisfy the allocation and get good space packing. If we cannot find free space at or after the given offset we fall back on looking for a chunk of the given size as close to that given offset as possible. When we fall back on the size search we also try to find a slot as close to the size we want as possible, to avoid breaking small chunks off of huge areas if possible. 2) remove the extent_io_tree that tracked the block group cache from fs_info and replaced it with an rb-tree thats tracks block group cache via offset. also added a per space_info list that tracks the block group cache for the particular space so we can lookup related block groups easily. 3) cleaned up the allocation code to make it a little easier to read and a little less complicated. Basically there are 3 steps, first look from our provided hint. If we couldn't find from that given hint, start back at our original search start and look for space from there. If that fails try to allocate space if we can and start looking again. If not we're screwed and need to start over again. 4) small fixes. there were some issues in volumes.c where we wouldn't allocate the rest of the disk. fixed cow_file_range to actually pass the alloc_hint, which has helped a good bit in making the fs_mark test I run have semi-normal results as we run out of space. Generally with data allocations we don't track where we last allocated from, so everytime we did a data allocation we'd search through every block group that we have looking for free space. Now searching a block group with no free space isn't terribly time consuming, it was causing a slight degradation as we got more data block groups. The alloc_hint has fixed this slight degredation and made things semi-normal. There is still one nagging problem I'm working on where we will get ENOSPC when there is definitely plenty of space. This only happens with metadata allocations, and only when we are almost full. So you generally hit the 85% mark first, but sometimes you'll hit the BUG before you hit the 85% wall. I'm still tracking it down, but until then this seems to be pretty stable and make a significant performance gain. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-09-24 01:14:11 +08:00
};
struct btrfs_free_space {
struct rb_node bytes_index;
struct rb_node offset_index;
u64 offset;
u64 bytes;
};
struct btrfs_block_group_cache {
struct btrfs_key key;
struct btrfs_block_group_item item;
spinlock_t lock;
struct mutex alloc_mutex;
u64 pinned;
u64 reserved;
u64 flags;
int cached;
int ro;
Btrfs: free space accounting redo 1) replace the per fs_info extent_io_tree that tracked free space with two rb-trees per block group to track free space areas via offset and size. The reason to do this is because most allocations come with a hint byte where to start, so we can usually find a chunk of free space at that hint byte to satisfy the allocation and get good space packing. If we cannot find free space at or after the given offset we fall back on looking for a chunk of the given size as close to that given offset as possible. When we fall back on the size search we also try to find a slot as close to the size we want as possible, to avoid breaking small chunks off of huge areas if possible. 2) remove the extent_io_tree that tracked the block group cache from fs_info and replaced it with an rb-tree thats tracks block group cache via offset. also added a per space_info list that tracks the block group cache for the particular space so we can lookup related block groups easily. 3) cleaned up the allocation code to make it a little easier to read and a little less complicated. Basically there are 3 steps, first look from our provided hint. If we couldn't find from that given hint, start back at our original search start and look for space from there. If that fails try to allocate space if we can and start looking again. If not we're screwed and need to start over again. 4) small fixes. there were some issues in volumes.c where we wouldn't allocate the rest of the disk. fixed cow_file_range to actually pass the alloc_hint, which has helped a good bit in making the fs_mark test I run have semi-normal results as we run out of space. Generally with data allocations we don't track where we last allocated from, so everytime we did a data allocation we'd search through every block group that we have looking for free space. Now searching a block group with no free space isn't terribly time consuming, it was causing a slight degradation as we got more data block groups. The alloc_hint has fixed this slight degredation and made things semi-normal. There is still one nagging problem I'm working on where we will get ENOSPC when there is definitely plenty of space. This only happens with metadata allocations, and only when we are almost full. So you generally hit the 85% mark first, but sometimes you'll hit the BUG before you hit the 85% wall. I'm still tracking it down, but until then this seems to be pretty stable and make a significant performance gain. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-09-24 01:14:11 +08:00
int dirty;
struct btrfs_space_info *space_info;
/* free space cache stuff */
struct rb_root free_space_bytes;
struct rb_root free_space_offset;
/* block group cache stuff */
struct rb_node cache_node;
/* for block groups in the same raid type */
struct list_head list;
};
struct btrfs_leaf_ref_tree {
struct rb_root root;
struct list_head list;
spinlock_t lock;
};
struct btrfs_device;
struct btrfs_fs_devices;
struct btrfs_fs_info {
u8 fsid[BTRFS_FSID_SIZE];
u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
struct btrfs_root *extent_root;
struct btrfs_root *tree_root;
struct btrfs_root *chunk_root;
struct btrfs_root *dev_root;
/* the log root tree is a directory of all the other log roots */
struct btrfs_root *log_root_tree;
struct radix_tree_root fs_roots_radix;
Btrfs: free space accounting redo 1) replace the per fs_info extent_io_tree that tracked free space with two rb-trees per block group to track free space areas via offset and size. The reason to do this is because most allocations come with a hint byte where to start, so we can usually find a chunk of free space at that hint byte to satisfy the allocation and get good space packing. If we cannot find free space at or after the given offset we fall back on looking for a chunk of the given size as close to that given offset as possible. When we fall back on the size search we also try to find a slot as close to the size we want as possible, to avoid breaking small chunks off of huge areas if possible. 2) remove the extent_io_tree that tracked the block group cache from fs_info and replaced it with an rb-tree thats tracks block group cache via offset. also added a per space_info list that tracks the block group cache for the particular space so we can lookup related block groups easily. 3) cleaned up the allocation code to make it a little easier to read and a little less complicated. Basically there are 3 steps, first look from our provided hint. If we couldn't find from that given hint, start back at our original search start and look for space from there. If that fails try to allocate space if we can and start looking again. If not we're screwed and need to start over again. 4) small fixes. there were some issues in volumes.c where we wouldn't allocate the rest of the disk. fixed cow_file_range to actually pass the alloc_hint, which has helped a good bit in making the fs_mark test I run have semi-normal results as we run out of space. Generally with data allocations we don't track where we last allocated from, so everytime we did a data allocation we'd search through every block group that we have looking for free space. Now searching a block group with no free space isn't terribly time consuming, it was causing a slight degradation as we got more data block groups. The alloc_hint has fixed this slight degredation and made things semi-normal. There is still one nagging problem I'm working on where we will get ENOSPC when there is definitely plenty of space. This only happens with metadata allocations, and only when we are almost full. So you generally hit the 85% mark first, but sometimes you'll hit the BUG before you hit the 85% wall. I'm still tracking it down, but until then this seems to be pretty stable and make a significant performance gain. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-09-24 01:14:11 +08:00
/* block group cache stuff */
spinlock_t block_group_cache_lock;
struct rb_root block_group_cache_tree;
struct extent_io_tree pinned_extents;
struct extent_io_tree pending_del;
struct extent_io_tree extent_ins;
/* logical->physical extent mapping */
struct btrfs_mapping_tree mapping_tree;
u64 generation;
u64 last_trans_committed;
u64 last_trans_new_blockgroup;
u64 open_ioctl_trans;
unsigned long mount_opt;
u64 max_extent;
u64 max_inline;
u64 alloc_start;
struct btrfs_transaction *running_transaction;
wait_queue_head_t transaction_throttle;
wait_queue_head_t transaction_wait;
wait_queue_head_t async_submit_wait;
wait_queue_head_t tree_log_wait;
struct btrfs_super_block super_copy;
struct btrfs_super_block super_for_commit;
struct block_device *__bdev;
struct super_block *sb;
struct inode *btree_inode;
struct backing_dev_info bdi;
spinlock_t hash_lock;
struct mutex trans_mutex;
struct mutex tree_log_mutex;
struct mutex transaction_kthread_mutex;
struct mutex cleaner_mutex;
struct mutex extent_ins_mutex;
struct mutex pinned_mutex;
struct mutex chunk_mutex;
struct mutex drop_mutex;
struct mutex volume_mutex;
2008-09-26 22:09:34 +08:00
struct mutex tree_reloc_mutex;
struct list_head trans_list;
struct list_head hashers;
struct list_head dead_roots;
atomic_t nr_async_submits;
atomic_t async_submit_draining;
atomic_t nr_async_bios;
atomic_t async_delalloc_pages;
atomic_t tree_log_writers;
atomic_t tree_log_commit;
unsigned long tree_log_batch;
u64 tree_log_transid;
/*
* this is used by the balancing code to wait for all the pending
* ordered extents
*/
spinlock_t ordered_extent_lock;
struct list_head ordered_extents;
struct list_head delalloc_inodes;
/*
* there is a pool of worker threads for checksumming during writes
* and a pool for checksumming after reads. This is because readers
* can run with FS locks held, and the writers may be waiting for
* those locks. We don't want ordering in the pending list to cause
* deadlocks, and so the two are serviced separately.
*
* A third pool does submit_bio to avoid deadlocking with the other
* two
*/
struct btrfs_workers workers;
struct btrfs_workers delalloc_workers;
struct btrfs_workers endio_workers;
struct btrfs_workers endio_write_workers;
struct btrfs_workers submit_workers;
/*
* fixup workers take dirty pages that didn't properly go through
* the cow mechanism and make them safe to write. It happens
* for the sys_munmap function call path
*/
struct btrfs_workers fixup_workers;
struct task_struct *transaction_kthread;
struct task_struct *cleaner_kthread;
int thread_pool_size;
2008-09-26 22:09:34 +08:00
/* tree relocation relocated fields */
struct list_head dead_reloc_roots;
struct btrfs_leaf_ref_tree reloc_ref_tree;
struct btrfs_leaf_ref_tree shared_ref_tree;
struct kobject super_kobj;
struct completion kobj_unregister;
int do_barriers;
int closing;
int log_root_recovering;
atomic_t throttles;
atomic_t throttle_gen;
u64 total_pinned;
struct list_head dirty_cowonly_roots;
struct btrfs_fs_devices *fs_devices;
struct list_head space_info;
spinlock_t delalloc_lock;
spinlock_t new_trans_lock;
u64 delalloc_bytes;
u64 last_alloc;
u64 last_data_alloc;
spinlock_t ref_cache_lock;
u64 total_ref_cache_size;
u64 avail_data_alloc_bits;
u64 avail_metadata_alloc_bits;
u64 avail_system_alloc_bits;
u64 data_alloc_profile;
u64 metadata_alloc_profile;
u64 system_alloc_profile;
void *bdev_holder;
};
/*
* in ram representation of the tree. extent_root is used for all allocations
* and for the extent tree extent_root root.
*/
struct btrfs_dirty_root;
struct btrfs_root {
struct extent_buffer *node;
/* the node lock is held while changing the node pointer */
spinlock_t node_lock;
struct extent_buffer *commit_root;
struct btrfs_leaf_ref_tree *ref_tree;
struct btrfs_leaf_ref_tree ref_tree_struct;
struct btrfs_dirty_root *dirty_root;
struct btrfs_root *log_root;
2008-09-26 22:09:34 +08:00
struct btrfs_root *reloc_root;
struct btrfs_root_item root_item;
struct btrfs_key root_key;
struct btrfs_fs_info *fs_info;
struct inode *inode;
struct extent_io_tree dirty_log_pages;
struct kobject root_kobj;
struct completion kobj_unregister;
struct mutex objectid_mutex;
struct mutex log_mutex;
u64 objectid;
u64 last_trans;
/* data allocations are done in sectorsize units */
u32 sectorsize;
/* node allocations are done in nodesize units */
u32 nodesize;
/* leaf allocations are done in leafsize units */
u32 leafsize;
u32 stripesize;
u32 type;
u64 highest_inode;
u64 last_inode_alloc;
int ref_cows;
int track_dirty;
u64 defrag_trans_start;
struct btrfs_key defrag_progress;
struct btrfs_key defrag_max;
int defrag_running;
int defrag_level;
char *name;
int in_sysfs;
/* the dirty list is only used by non-reference counted roots */
struct list_head dirty_list;
spinlock_t list_lock;
struct list_head dead_list;
struct list_head orphan_list;
};
/*
* inode items have the data typically returned from stat and store other
* info about object characteristics. There is one for every file and dir in
* the FS
*/
#define BTRFS_INODE_ITEM_KEY 1
#define BTRFS_INODE_REF_KEY 2
#define BTRFS_XATTR_ITEM_KEY 8
#define BTRFS_ORPHAN_ITEM_KEY 9
/* reserve 2-15 close to the inode for later flexibility */
/*
* dir items are the name -> inode pointers in a directory. There is one
* for every name in a directory.
*/
#define BTRFS_DIR_LOG_ITEM_KEY 14
#define BTRFS_DIR_LOG_INDEX_KEY 15
#define BTRFS_DIR_ITEM_KEY 16
#define BTRFS_DIR_INDEX_KEY 17
/*
* extent data is for file data
*/
#define BTRFS_EXTENT_DATA_KEY 18
/*
* csum items have the checksums for data in the extents
*/
#define BTRFS_CSUM_ITEM_KEY 19
/* reserve 21-31 for other file/dir stuff */
/*
* root items point to tree roots. There are typically in the root
* tree used by the super block to find all the other trees
*/
#define BTRFS_ROOT_ITEM_KEY 32
/*
* extent items are in the extent map tree. These record which blocks
* are used, and how many references there are to each block
*/
#define BTRFS_EXTENT_ITEM_KEY 33
#define BTRFS_EXTENT_REF_KEY 34
/*
* block groups give us hints into the extent allocation trees. Which
* blocks are free etc etc
*/
#define BTRFS_BLOCK_GROUP_ITEM_KEY 50
#define BTRFS_DEV_EXTENT_KEY 75
#define BTRFS_DEV_ITEM_KEY 76
#define BTRFS_CHUNK_ITEM_KEY 77
/*
* string items are for debugging. They just store a short string of
* data in the FS
*/
#define BTRFS_STRING_ITEM_KEY 253
#define BTRFS_MOUNT_NODATASUM (1 << 0)
#define BTRFS_MOUNT_NODATACOW (1 << 1)
#define BTRFS_MOUNT_NOBARRIER (1 << 2)
#define BTRFS_MOUNT_SSD (1 << 3)
#define BTRFS_MOUNT_DEGRADED (1 << 4)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
#define BTRFS_MOUNT_COMPRESS (1 << 5)
#define btrfs_clear_opt(o, opt) ((o) &= ~BTRFS_MOUNT_##opt)
#define btrfs_set_opt(o, opt) ((o) |= BTRFS_MOUNT_##opt)
#define btrfs_test_opt(root, opt) ((root)->fs_info->mount_opt & \
BTRFS_MOUNT_##opt)
/*
* Inode flags
*/
#define BTRFS_INODE_NODATASUM (1 << 0)
#define BTRFS_INODE_NODATACOW (1 << 1)
#define BTRFS_INODE_READONLY (1 << 2)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
#define BTRFS_INODE_NOCOMPRESS (1 << 3)
#define BTRFS_INODE_PREALLOC (1 << 4)
#define btrfs_clear_flag(inode, flag) (BTRFS_I(inode)->flags &= \
~BTRFS_INODE_##flag)
#define btrfs_set_flag(inode, flag) (BTRFS_I(inode)->flags |= \
BTRFS_INODE_##flag)
#define btrfs_test_flag(inode, flag) (BTRFS_I(inode)->flags & \
BTRFS_INODE_##flag)
/* some macros to generate set/get funcs for the struct fields. This
* assumes there is a lefoo_to_cpu for every type, so lets make a simple
* one for u8:
*/
#define le8_to_cpu(v) (v)
#define cpu_to_le8(v) (v)
#define __le8 u8
#define read_eb_member(eb, ptr, type, member, result) ( \
read_extent_buffer(eb, (char *)(result), \
((unsigned long)(ptr)) + \
offsetof(type, member), \
sizeof(((type *)0)->member)))
#define write_eb_member(eb, ptr, type, member, result) ( \
write_extent_buffer(eb, (char *)(result), \
((unsigned long)(ptr)) + \
offsetof(type, member), \
sizeof(((type *)0)->member)))
#ifndef BTRFS_SETGET_FUNCS
#define BTRFS_SETGET_FUNCS(name, type, member, bits) \
u##bits btrfs_##name(struct extent_buffer *eb, type *s); \
void btrfs_set_##name(struct extent_buffer *eb, type *s, u##bits val);
#endif
#define BTRFS_SETGET_HEADER_FUNCS(name, type, member, bits) \
static inline u##bits btrfs_##name(struct extent_buffer *eb) \
{ \
type *p = kmap_atomic(eb->first_page, KM_USER0); \
u##bits res = le##bits##_to_cpu(p->member); \
kunmap_atomic(p, KM_USER0); \
return res; \
} \
static inline void btrfs_set_##name(struct extent_buffer *eb, \
u##bits val) \
{ \
type *p = kmap_atomic(eb->first_page, KM_USER0); \
p->member = cpu_to_le##bits(val); \
kunmap_atomic(p, KM_USER0); \
}
#define BTRFS_SETGET_STACK_FUNCS(name, type, member, bits) \
static inline u##bits btrfs_##name(type *s) \
{ \
return le##bits##_to_cpu(s->member); \
} \
static inline void btrfs_set_##name(type *s, u##bits val) \
{ \
s->member = cpu_to_le##bits(val); \
}
BTRFS_SETGET_FUNCS(device_type, struct btrfs_dev_item, type, 64);
BTRFS_SETGET_FUNCS(device_total_bytes, struct btrfs_dev_item, total_bytes, 64);
BTRFS_SETGET_FUNCS(device_bytes_used, struct btrfs_dev_item, bytes_used, 64);
BTRFS_SETGET_FUNCS(device_io_align, struct btrfs_dev_item, io_align, 32);
BTRFS_SETGET_FUNCS(device_io_width, struct btrfs_dev_item, io_width, 32);
BTRFS_SETGET_FUNCS(device_sector_size, struct btrfs_dev_item, sector_size, 32);
BTRFS_SETGET_FUNCS(device_id, struct btrfs_dev_item, devid, 64);
BTRFS_SETGET_FUNCS(device_group, struct btrfs_dev_item, dev_group, 32);
BTRFS_SETGET_FUNCS(device_seek_speed, struct btrfs_dev_item, seek_speed, 8);
BTRFS_SETGET_FUNCS(device_bandwidth, struct btrfs_dev_item, bandwidth, 8);
BTRFS_SETGET_STACK_FUNCS(stack_device_type, struct btrfs_dev_item, type, 64);
BTRFS_SETGET_STACK_FUNCS(stack_device_total_bytes, struct btrfs_dev_item,
total_bytes, 64);
BTRFS_SETGET_STACK_FUNCS(stack_device_bytes_used, struct btrfs_dev_item,
bytes_used, 64);
BTRFS_SETGET_STACK_FUNCS(stack_device_io_align, struct btrfs_dev_item,
io_align, 32);
BTRFS_SETGET_STACK_FUNCS(stack_device_io_width, struct btrfs_dev_item,
io_width, 32);
BTRFS_SETGET_STACK_FUNCS(stack_device_sector_size, struct btrfs_dev_item,
sector_size, 32);
BTRFS_SETGET_STACK_FUNCS(stack_device_id, struct btrfs_dev_item, devid, 64);
BTRFS_SETGET_STACK_FUNCS(stack_device_group, struct btrfs_dev_item,
dev_group, 32);
BTRFS_SETGET_STACK_FUNCS(stack_device_seek_speed, struct btrfs_dev_item,
seek_speed, 8);
BTRFS_SETGET_STACK_FUNCS(stack_device_bandwidth, struct btrfs_dev_item,
bandwidth, 8);
static inline char *btrfs_device_uuid(struct btrfs_dev_item *d)
{
return (char *)d + offsetof(struct btrfs_dev_item, uuid);
}
BTRFS_SETGET_FUNCS(chunk_length, struct btrfs_chunk, length, 64);
BTRFS_SETGET_FUNCS(chunk_owner, struct btrfs_chunk, owner, 64);
BTRFS_SETGET_FUNCS(chunk_stripe_len, struct btrfs_chunk, stripe_len, 64);
BTRFS_SETGET_FUNCS(chunk_io_align, struct btrfs_chunk, io_align, 32);
BTRFS_SETGET_FUNCS(chunk_io_width, struct btrfs_chunk, io_width, 32);
BTRFS_SETGET_FUNCS(chunk_sector_size, struct btrfs_chunk, sector_size, 32);
BTRFS_SETGET_FUNCS(chunk_type, struct btrfs_chunk, type, 64);
BTRFS_SETGET_FUNCS(chunk_num_stripes, struct btrfs_chunk, num_stripes, 16);
BTRFS_SETGET_FUNCS(chunk_sub_stripes, struct btrfs_chunk, sub_stripes, 16);
BTRFS_SETGET_FUNCS(stripe_devid, struct btrfs_stripe, devid, 64);
BTRFS_SETGET_FUNCS(stripe_offset, struct btrfs_stripe, offset, 64);
static inline char *btrfs_stripe_dev_uuid(struct btrfs_stripe *s)
{
return (char *)s + offsetof(struct btrfs_stripe, dev_uuid);
}
BTRFS_SETGET_STACK_FUNCS(stack_chunk_length, struct btrfs_chunk, length, 64);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_owner, struct btrfs_chunk, owner, 64);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_stripe_len, struct btrfs_chunk,
stripe_len, 64);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_io_align, struct btrfs_chunk,
io_align, 32);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_io_width, struct btrfs_chunk,
io_width, 32);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_sector_size, struct btrfs_chunk,
sector_size, 32);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_type, struct btrfs_chunk, type, 64);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_num_stripes, struct btrfs_chunk,
num_stripes, 16);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_sub_stripes, struct btrfs_chunk,
sub_stripes, 16);
BTRFS_SETGET_STACK_FUNCS(stack_stripe_devid, struct btrfs_stripe, devid, 64);
BTRFS_SETGET_STACK_FUNCS(stack_stripe_offset, struct btrfs_stripe, offset, 64);
static inline struct btrfs_stripe *btrfs_stripe_nr(struct btrfs_chunk *c,
int nr)
{
unsigned long offset = (unsigned long)c;
offset += offsetof(struct btrfs_chunk, stripe);
offset += nr * sizeof(struct btrfs_stripe);
return (struct btrfs_stripe *)offset;
}
static inline char *btrfs_stripe_dev_uuid_nr(struct btrfs_chunk *c, int nr)
{
return btrfs_stripe_dev_uuid(btrfs_stripe_nr(c, nr));
}
static inline u64 btrfs_stripe_offset_nr(struct extent_buffer *eb,
struct btrfs_chunk *c, int nr)
{
return btrfs_stripe_offset(eb, btrfs_stripe_nr(c, nr));
}
static inline void btrfs_set_stripe_offset_nr(struct extent_buffer *eb,
struct btrfs_chunk *c, int nr,
u64 val)
{
btrfs_set_stripe_offset(eb, btrfs_stripe_nr(c, nr), val);
}
static inline u64 btrfs_stripe_devid_nr(struct extent_buffer *eb,
struct btrfs_chunk *c, int nr)
{
return btrfs_stripe_devid(eb, btrfs_stripe_nr(c, nr));
}
static inline void btrfs_set_stripe_devid_nr(struct extent_buffer *eb,
struct btrfs_chunk *c, int nr,
u64 val)
{
btrfs_set_stripe_devid(eb, btrfs_stripe_nr(c, nr), val);
}
/* struct btrfs_block_group_item */
BTRFS_SETGET_STACK_FUNCS(block_group_used, struct btrfs_block_group_item,
used, 64);
BTRFS_SETGET_FUNCS(disk_block_group_used, struct btrfs_block_group_item,
used, 64);
BTRFS_SETGET_STACK_FUNCS(block_group_chunk_objectid,
struct btrfs_block_group_item, chunk_objectid, 64);
BTRFS_SETGET_FUNCS(disk_block_group_chunk_objectid,
struct btrfs_block_group_item, chunk_objectid, 64);
BTRFS_SETGET_FUNCS(disk_block_group_flags,
struct btrfs_block_group_item, flags, 64);
BTRFS_SETGET_STACK_FUNCS(block_group_flags,
struct btrfs_block_group_item, flags, 64);
/* struct btrfs_inode_ref */
BTRFS_SETGET_FUNCS(inode_ref_name_len, struct btrfs_inode_ref, name_len, 16);
BTRFS_SETGET_FUNCS(inode_ref_index, struct btrfs_inode_ref, index, 64);
/* struct btrfs_inode_item */
BTRFS_SETGET_FUNCS(inode_generation, struct btrfs_inode_item, generation, 64);
BTRFS_SETGET_FUNCS(inode_transid, struct btrfs_inode_item, transid, 64);
BTRFS_SETGET_FUNCS(inode_size, struct btrfs_inode_item, size, 64);
BTRFS_SETGET_FUNCS(inode_nbytes, struct btrfs_inode_item, nbytes, 64);
BTRFS_SETGET_FUNCS(inode_block_group, struct btrfs_inode_item, block_group, 64);
BTRFS_SETGET_FUNCS(inode_nlink, struct btrfs_inode_item, nlink, 32);
BTRFS_SETGET_FUNCS(inode_uid, struct btrfs_inode_item, uid, 32);
BTRFS_SETGET_FUNCS(inode_gid, struct btrfs_inode_item, gid, 32);
BTRFS_SETGET_FUNCS(inode_mode, struct btrfs_inode_item, mode, 32);
BTRFS_SETGET_FUNCS(inode_rdev, struct btrfs_inode_item, rdev, 64);
BTRFS_SETGET_FUNCS(inode_flags, struct btrfs_inode_item, flags, 16);
BTRFS_SETGET_FUNCS(inode_compat_flags, struct btrfs_inode_item,
compat_flags, 16);
static inline struct btrfs_timespec *
btrfs_inode_atime(struct btrfs_inode_item *inode_item)
{
unsigned long ptr = (unsigned long)inode_item;
ptr += offsetof(struct btrfs_inode_item, atime);
return (struct btrfs_timespec *)ptr;
}
static inline struct btrfs_timespec *
btrfs_inode_mtime(struct btrfs_inode_item *inode_item)
{
unsigned long ptr = (unsigned long)inode_item;
ptr += offsetof(struct btrfs_inode_item, mtime);
return (struct btrfs_timespec *)ptr;
}
static inline struct btrfs_timespec *
btrfs_inode_ctime(struct btrfs_inode_item *inode_item)
{
unsigned long ptr = (unsigned long)inode_item;
ptr += offsetof(struct btrfs_inode_item, ctime);
return (struct btrfs_timespec *)ptr;
}
static inline struct btrfs_timespec *
btrfs_inode_otime(struct btrfs_inode_item *inode_item)
{
unsigned long ptr = (unsigned long)inode_item;
ptr += offsetof(struct btrfs_inode_item, otime);
return (struct btrfs_timespec *)ptr;
}
BTRFS_SETGET_FUNCS(timespec_sec, struct btrfs_timespec, sec, 64);
BTRFS_SETGET_FUNCS(timespec_nsec, struct btrfs_timespec, nsec, 32);
/* struct btrfs_dev_extent */
BTRFS_SETGET_FUNCS(dev_extent_chunk_tree, struct btrfs_dev_extent,
chunk_tree, 64);
BTRFS_SETGET_FUNCS(dev_extent_chunk_objectid, struct btrfs_dev_extent,
chunk_objectid, 64);
BTRFS_SETGET_FUNCS(dev_extent_chunk_offset, struct btrfs_dev_extent,
chunk_offset, 64);
BTRFS_SETGET_FUNCS(dev_extent_length, struct btrfs_dev_extent, length, 64);
static inline u8 *btrfs_dev_extent_chunk_tree_uuid(struct btrfs_dev_extent *dev)
{
unsigned long ptr = offsetof(struct btrfs_dev_extent, chunk_tree_uuid);
return (u8 *)((unsigned long)dev + ptr);
}
/* struct btrfs_extent_ref */
BTRFS_SETGET_FUNCS(ref_root, struct btrfs_extent_ref, root, 64);
BTRFS_SETGET_FUNCS(ref_generation, struct btrfs_extent_ref, generation, 64);
BTRFS_SETGET_FUNCS(ref_objectid, struct btrfs_extent_ref, objectid, 64);
BTRFS_SETGET_FUNCS(ref_num_refs, struct btrfs_extent_ref, num_refs, 32);
BTRFS_SETGET_STACK_FUNCS(stack_ref_root, struct btrfs_extent_ref, root, 64);
BTRFS_SETGET_STACK_FUNCS(stack_ref_generation, struct btrfs_extent_ref,
generation, 64);
BTRFS_SETGET_STACK_FUNCS(stack_ref_objectid, struct btrfs_extent_ref,
objectid, 64);
BTRFS_SETGET_STACK_FUNCS(stack_ref_num_refs, struct btrfs_extent_ref,
num_refs, 32);
/* struct btrfs_extent_item */
BTRFS_SETGET_FUNCS(extent_refs, struct btrfs_extent_item, refs, 32);
BTRFS_SETGET_STACK_FUNCS(stack_extent_refs, struct btrfs_extent_item,
refs, 32);
/* struct btrfs_node */
BTRFS_SETGET_FUNCS(key_blockptr, struct btrfs_key_ptr, blockptr, 64);
BTRFS_SETGET_FUNCS(key_generation, struct btrfs_key_ptr, generation, 64);
static inline u64 btrfs_node_blockptr(struct extent_buffer *eb, int nr)
{
unsigned long ptr;
ptr = offsetof(struct btrfs_node, ptrs) +
sizeof(struct btrfs_key_ptr) * nr;
return btrfs_key_blockptr(eb, (struct btrfs_key_ptr *)ptr);
}
static inline void btrfs_set_node_blockptr(struct extent_buffer *eb,
int nr, u64 val)
{
unsigned long ptr;
ptr = offsetof(struct btrfs_node, ptrs) +
sizeof(struct btrfs_key_ptr) * nr;
btrfs_set_key_blockptr(eb, (struct btrfs_key_ptr *)ptr, val);
}
static inline u64 btrfs_node_ptr_generation(struct extent_buffer *eb, int nr)
{
unsigned long ptr;
ptr = offsetof(struct btrfs_node, ptrs) +
sizeof(struct btrfs_key_ptr) * nr;
return btrfs_key_generation(eb, (struct btrfs_key_ptr *)ptr);
}
static inline void btrfs_set_node_ptr_generation(struct extent_buffer *eb,
int nr, u64 val)
{
unsigned long ptr;
ptr = offsetof(struct btrfs_node, ptrs) +
sizeof(struct btrfs_key_ptr) * nr;
btrfs_set_key_generation(eb, (struct btrfs_key_ptr *)ptr, val);
}
static inline unsigned long btrfs_node_key_ptr_offset(int nr)
{
return offsetof(struct btrfs_node, ptrs) +
sizeof(struct btrfs_key_ptr) * nr;
}
void btrfs_node_key(struct extent_buffer *eb,
struct btrfs_disk_key *disk_key, int nr);
static inline void btrfs_set_node_key(struct extent_buffer *eb,
struct btrfs_disk_key *disk_key, int nr)
{
unsigned long ptr;
ptr = btrfs_node_key_ptr_offset(nr);
write_eb_member(eb, (struct btrfs_key_ptr *)ptr,
struct btrfs_key_ptr, key, disk_key);
}
/* struct btrfs_item */
BTRFS_SETGET_FUNCS(item_offset, struct btrfs_item, offset, 32);
BTRFS_SETGET_FUNCS(item_size, struct btrfs_item, size, 32);
static inline unsigned long btrfs_item_nr_offset(int nr)
{
return offsetof(struct btrfs_leaf, items) +
sizeof(struct btrfs_item) * nr;
}
static inline struct btrfs_item *btrfs_item_nr(struct extent_buffer *eb,
int nr)
{
return (struct btrfs_item *)btrfs_item_nr_offset(nr);
}
static inline u32 btrfs_item_end(struct extent_buffer *eb,
struct btrfs_item *item)
{
return btrfs_item_offset(eb, item) + btrfs_item_size(eb, item);
}
static inline u32 btrfs_item_end_nr(struct extent_buffer *eb, int nr)
{
return btrfs_item_end(eb, btrfs_item_nr(eb, nr));
}
static inline u32 btrfs_item_offset_nr(struct extent_buffer *eb, int nr)
{
return btrfs_item_offset(eb, btrfs_item_nr(eb, nr));
}
static inline u32 btrfs_item_size_nr(struct extent_buffer *eb, int nr)
{
return btrfs_item_size(eb, btrfs_item_nr(eb, nr));
}
static inline void btrfs_item_key(struct extent_buffer *eb,
struct btrfs_disk_key *disk_key, int nr)
{
struct btrfs_item *item = btrfs_item_nr(eb, nr);
read_eb_member(eb, item, struct btrfs_item, key, disk_key);
}
static inline void btrfs_set_item_key(struct extent_buffer *eb,
struct btrfs_disk_key *disk_key, int nr)
{
struct btrfs_item *item = btrfs_item_nr(eb, nr);
write_eb_member(eb, item, struct btrfs_item, key, disk_key);
}
BTRFS_SETGET_FUNCS(dir_log_end, struct btrfs_dir_log_item, end, 64);
/* struct btrfs_dir_item */
BTRFS_SETGET_FUNCS(dir_data_len, struct btrfs_dir_item, data_len, 16);
BTRFS_SETGET_FUNCS(dir_type, struct btrfs_dir_item, type, 8);
BTRFS_SETGET_FUNCS(dir_name_len, struct btrfs_dir_item, name_len, 16);
BTRFS_SETGET_FUNCS(dir_transid, struct btrfs_dir_item, transid, 64);
static inline void btrfs_dir_item_key(struct extent_buffer *eb,
struct btrfs_dir_item *item,
struct btrfs_disk_key *key)
{
read_eb_member(eb, item, struct btrfs_dir_item, location, key);
}
static inline void btrfs_set_dir_item_key(struct extent_buffer *eb,
struct btrfs_dir_item *item,
struct btrfs_disk_key *key)
{
write_eb_member(eb, item, struct btrfs_dir_item, location, key);
}
/* struct btrfs_disk_key */
BTRFS_SETGET_STACK_FUNCS(disk_key_objectid, struct btrfs_disk_key,
objectid, 64);
BTRFS_SETGET_STACK_FUNCS(disk_key_offset, struct btrfs_disk_key, offset, 64);
BTRFS_SETGET_STACK_FUNCS(disk_key_type, struct btrfs_disk_key, type, 8);
static inline void btrfs_disk_key_to_cpu(struct btrfs_key *cpu,
struct btrfs_disk_key *disk)
{
cpu->offset = le64_to_cpu(disk->offset);
cpu->type = disk->type;
cpu->objectid = le64_to_cpu(disk->objectid);
}
static inline void btrfs_cpu_key_to_disk(struct btrfs_disk_key *disk,
struct btrfs_key *cpu)
{
disk->offset = cpu_to_le64(cpu->offset);
disk->type = cpu->type;
disk->objectid = cpu_to_le64(cpu->objectid);
}
static inline void btrfs_node_key_to_cpu(struct extent_buffer *eb,
struct btrfs_key *key, int nr)
{
struct btrfs_disk_key disk_key;
btrfs_node_key(eb, &disk_key, nr);
btrfs_disk_key_to_cpu(key, &disk_key);
}
static inline void btrfs_item_key_to_cpu(struct extent_buffer *eb,
struct btrfs_key *key, int nr)
{
struct btrfs_disk_key disk_key;
btrfs_item_key(eb, &disk_key, nr);
btrfs_disk_key_to_cpu(key, &disk_key);
}
static inline void btrfs_dir_item_key_to_cpu(struct extent_buffer *eb,
struct btrfs_dir_item *item,
struct btrfs_key *key)
{
struct btrfs_disk_key disk_key;
btrfs_dir_item_key(eb, item, &disk_key);
btrfs_disk_key_to_cpu(key, &disk_key);
}
static inline u8 btrfs_key_type(struct btrfs_key *key)
{
return key->type;
}
static inline void btrfs_set_key_type(struct btrfs_key *key, u8 val)
{
key->type = val;
}
/* struct btrfs_header */
BTRFS_SETGET_HEADER_FUNCS(header_bytenr, struct btrfs_header, bytenr, 64);
BTRFS_SETGET_HEADER_FUNCS(header_generation, struct btrfs_header,
generation, 64);
BTRFS_SETGET_HEADER_FUNCS(header_owner, struct btrfs_header, owner, 64);
BTRFS_SETGET_HEADER_FUNCS(header_nritems, struct btrfs_header, nritems, 32);
BTRFS_SETGET_HEADER_FUNCS(header_flags, struct btrfs_header, flags, 64);
BTRFS_SETGET_HEADER_FUNCS(header_level, struct btrfs_header, level, 8);
static inline int btrfs_header_flag(struct extent_buffer *eb, u64 flag)
{
return (btrfs_header_flags(eb) & flag) == flag;
}
static inline int btrfs_set_header_flag(struct extent_buffer *eb, u64 flag)
{
u64 flags = btrfs_header_flags(eb);
btrfs_set_header_flags(eb, flags | flag);
return (flags & flag) == flag;
}
static inline int btrfs_clear_header_flag(struct extent_buffer *eb, u64 flag)
{
u64 flags = btrfs_header_flags(eb);
btrfs_set_header_flags(eb, flags & ~flag);
return (flags & flag) == flag;
}
static inline u8 *btrfs_header_fsid(struct extent_buffer *eb)
{
unsigned long ptr = offsetof(struct btrfs_header, fsid);
return (u8 *)ptr;
}
static inline u8 *btrfs_header_chunk_tree_uuid(struct extent_buffer *eb)
{
unsigned long ptr = offsetof(struct btrfs_header, chunk_tree_uuid);
return (u8 *)ptr;
}
static inline u8 *btrfs_super_fsid(struct extent_buffer *eb)
{
unsigned long ptr = offsetof(struct btrfs_super_block, fsid);
return (u8 *)ptr;
}
static inline u8 *btrfs_header_csum(struct extent_buffer *eb)
{
unsigned long ptr = offsetof(struct btrfs_header, csum);
return (u8 *)ptr;
}
static inline struct btrfs_node *btrfs_buffer_node(struct extent_buffer *eb)
{
return NULL;
}
static inline struct btrfs_leaf *btrfs_buffer_leaf(struct extent_buffer *eb)
{
return NULL;
}
static inline struct btrfs_header *btrfs_buffer_header(struct extent_buffer *eb)
{
return NULL;
}
static inline int btrfs_is_leaf(struct extent_buffer *eb)
{
return (btrfs_header_level(eb) == 0);
}
/* struct btrfs_root_item */
BTRFS_SETGET_FUNCS(disk_root_generation, struct btrfs_root_item,
generation, 64);
BTRFS_SETGET_FUNCS(disk_root_refs, struct btrfs_root_item, refs, 32);
BTRFS_SETGET_FUNCS(disk_root_bytenr, struct btrfs_root_item, bytenr, 64);
BTRFS_SETGET_FUNCS(disk_root_level, struct btrfs_root_item, level, 8);
BTRFS_SETGET_STACK_FUNCS(root_generation, struct btrfs_root_item,
generation, 64);
BTRFS_SETGET_STACK_FUNCS(root_bytenr, struct btrfs_root_item, bytenr, 64);
BTRFS_SETGET_STACK_FUNCS(root_level, struct btrfs_root_item, level, 8);
BTRFS_SETGET_STACK_FUNCS(root_dirid, struct btrfs_root_item, root_dirid, 64);
BTRFS_SETGET_STACK_FUNCS(root_refs, struct btrfs_root_item, refs, 32);
BTRFS_SETGET_STACK_FUNCS(root_flags, struct btrfs_root_item, flags, 32);
BTRFS_SETGET_STACK_FUNCS(root_used, struct btrfs_root_item, bytes_used, 64);
BTRFS_SETGET_STACK_FUNCS(root_limit, struct btrfs_root_item, byte_limit, 64);
BTRFS_SETGET_STACK_FUNCS(root_last_snapshot, struct btrfs_root_item,
last_snapshot, 64);
/* struct btrfs_super_block */
BTRFS_SETGET_STACK_FUNCS(super_bytenr, struct btrfs_super_block, bytenr, 64);
BTRFS_SETGET_STACK_FUNCS(super_flags, struct btrfs_super_block, flags, 64);
BTRFS_SETGET_STACK_FUNCS(super_generation, struct btrfs_super_block,
generation, 64);
BTRFS_SETGET_STACK_FUNCS(super_root, struct btrfs_super_block, root, 64);
BTRFS_SETGET_STACK_FUNCS(super_sys_array_size,
struct btrfs_super_block, sys_chunk_array_size, 32);
BTRFS_SETGET_STACK_FUNCS(super_chunk_root_generation,
struct btrfs_super_block, chunk_root_generation, 64);
BTRFS_SETGET_STACK_FUNCS(super_root_level, struct btrfs_super_block,
root_level, 8);
BTRFS_SETGET_STACK_FUNCS(super_chunk_root, struct btrfs_super_block,
chunk_root, 64);
BTRFS_SETGET_STACK_FUNCS(super_chunk_root_level, struct btrfs_super_block,
chunk_root_level, 8);
BTRFS_SETGET_STACK_FUNCS(super_log_root, struct btrfs_super_block,
log_root, 64);
BTRFS_SETGET_STACK_FUNCS(super_log_root_level, struct btrfs_super_block,
log_root_level, 8);
BTRFS_SETGET_STACK_FUNCS(super_total_bytes, struct btrfs_super_block,
total_bytes, 64);
BTRFS_SETGET_STACK_FUNCS(super_bytes_used, struct btrfs_super_block,
bytes_used, 64);
BTRFS_SETGET_STACK_FUNCS(super_sectorsize, struct btrfs_super_block,
sectorsize, 32);
BTRFS_SETGET_STACK_FUNCS(super_nodesize, struct btrfs_super_block,
nodesize, 32);
BTRFS_SETGET_STACK_FUNCS(super_leafsize, struct btrfs_super_block,
leafsize, 32);
BTRFS_SETGET_STACK_FUNCS(super_stripesize, struct btrfs_super_block,
stripesize, 32);
BTRFS_SETGET_STACK_FUNCS(super_root_dir, struct btrfs_super_block,
root_dir_objectid, 64);
BTRFS_SETGET_STACK_FUNCS(super_num_devices, struct btrfs_super_block,
num_devices, 64);
static inline unsigned long btrfs_leaf_data(struct extent_buffer *l)
{
return offsetof(struct btrfs_leaf, items);
}
/* struct btrfs_file_extent_item */
BTRFS_SETGET_FUNCS(file_extent_type, struct btrfs_file_extent_item, type, 8);
static inline unsigned long btrfs_file_extent_inline_start(struct
btrfs_file_extent_item *e)
{
unsigned long offset = (unsigned long)e;
offset += offsetof(struct btrfs_file_extent_item, disk_bytenr);
return offset;
}
static inline u32 btrfs_file_extent_calc_inline_size(u32 datasize)
{
return offsetof(struct btrfs_file_extent_item, disk_bytenr) + datasize;
}
BTRFS_SETGET_FUNCS(file_extent_disk_bytenr, struct btrfs_file_extent_item,
disk_bytenr, 64);
BTRFS_SETGET_FUNCS(file_extent_generation, struct btrfs_file_extent_item,
generation, 64);
BTRFS_SETGET_FUNCS(file_extent_disk_num_bytes, struct btrfs_file_extent_item,
disk_num_bytes, 64);
BTRFS_SETGET_FUNCS(file_extent_offset, struct btrfs_file_extent_item,
offset, 64);
BTRFS_SETGET_FUNCS(file_extent_num_bytes, struct btrfs_file_extent_item,
num_bytes, 64);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
BTRFS_SETGET_FUNCS(file_extent_ram_bytes, struct btrfs_file_extent_item,
ram_bytes, 64);
BTRFS_SETGET_FUNCS(file_extent_compression, struct btrfs_file_extent_item,
compression, 8);
BTRFS_SETGET_FUNCS(file_extent_encryption, struct btrfs_file_extent_item,
encryption, 8);
BTRFS_SETGET_FUNCS(file_extent_other_encoding, struct btrfs_file_extent_item,
other_encoding, 16);
/* this returns the number of file bytes represented by the inline item.
* If an item is compressed, this is the uncompressed size
*/
static inline u32 btrfs_file_extent_inline_len(struct extent_buffer *eb,
struct btrfs_file_extent_item *e)
{
return btrfs_file_extent_ram_bytes(eb, e);
}
/*
* this returns the number of bytes used by the item on disk, minus the
* size of any extent headers. If a file is compressed on disk, this is
* the compressed size
*/
static inline u32 btrfs_file_extent_inline_item_len(struct extent_buffer *eb,
struct btrfs_item *e)
{
unsigned long offset;
offset = offsetof(struct btrfs_file_extent_item, disk_bytenr);
return btrfs_item_size(eb, e) - offset;
}
static inline struct btrfs_root *btrfs_sb(struct super_block *sb)
{
return sb->s_fs_info;
}
static inline int btrfs_set_root_name(struct btrfs_root *root,
const char *name, int len)
{
/* if we already have a name just free it */
if (root->name)
kfree(root->name);
root->name = kmalloc(len+1, GFP_KERNEL);
if (!root->name)
return -ENOMEM;
memcpy(root->name, name, len);
root->name[len] ='\0';
return 0;
}
static inline u32 btrfs_level_size(struct btrfs_root *root, int level) {
if (level == 0)
return root->leafsize;
return root->nodesize;
}
/* helper function to cast into the data area of the leaf. */
#define btrfs_item_ptr(leaf, slot, type) \
((type *)(btrfs_leaf_data(leaf) + \
btrfs_item_offset_nr(leaf, slot)))
#define btrfs_item_ptr_offset(leaf, slot) \
((unsigned long)(btrfs_leaf_data(leaf) + \
btrfs_item_offset_nr(leaf, slot)))
static inline struct dentry *fdentry(struct file *file)
{
return file->f_path.dentry;
}
/* extent-tree.c */
int btrfs_lookup_extent(struct btrfs_root *root, u64 start, u64 len);
int btrfs_lookup_extent_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 bytenr,
u64 num_bytes, u32 *refs);
int btrfs_update_pinned_extents(struct btrfs_root *root,
u64 bytenr, u64 num, int pin);
int btrfs_drop_leaf_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct extent_buffer *leaf);
int btrfs_cross_ref_exist(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 bytenr);
int btrfs_extent_post_op(struct btrfs_trans_handle *trans,
struct btrfs_root *root);
int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy);
struct btrfs_block_group_cache *btrfs_lookup_block_group(struct
btrfs_fs_info *info,
u64 bytenr);
struct btrfs_block_group_cache *btrfs_find_block_group(struct btrfs_root *root,
struct btrfs_block_group_cache
*hint, u64 search_start,
int data, int owner);
struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u32 blocksize, u64 parent,
u64 root_objectid,
u64 ref_generation,
int level,
u64 hint,
u64 empty_size);
struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 bytenr, u32 blocksize);
int btrfs_alloc_extent(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 num_bytes, u64 parent, u64 min_bytes,
u64 root_objectid, u64 ref_generation,
u64 owner, u64 empty_size, u64 hint_byte,
u64 search_end, struct btrfs_key *ins, u64 data);
int btrfs_alloc_reserved_extent(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 parent,
u64 root_objectid, u64 ref_generation,
u64 owner, struct btrfs_key *ins);
int btrfs_alloc_logged_extent(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 parent,
u64 root_objectid, u64 ref_generation,
u64 owner, struct btrfs_key *ins);
int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 num_bytes, u64 min_alloc_size,
u64 empty_size, u64 hint_byte,
u64 search_end, struct btrfs_key *ins,
u64 data);
int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
struct extent_buffer *orig_buf, struct extent_buffer *buf,
u32 *nr_extents);
int btrfs_cache_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
struct extent_buffer *buf, u32 nr_extents);
int btrfs_update_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct extent_buffer *orig_buf,
struct extent_buffer *buf, int start_slot, int nr);
int btrfs_free_extent(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 bytenr, u64 num_bytes, u64 parent,
u64 root_objectid, u64 ref_generation,
u64 owner_objectid, int pin);
int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len);
int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct extent_io_tree *unpin);
int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 bytenr, u64 num_bytes, u64 parent,
u64 root_objectid, u64 ref_generation,
u64 owner_objectid);
int btrfs_update_extent_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 bytenr,
u64 orig_parent, u64 parent,
u64 root_objectid, u64 ref_generation,
u64 owner_objectid);
int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
struct btrfs_root *root);
int btrfs_free_block_groups(struct btrfs_fs_info *info);
int btrfs_read_block_groups(struct btrfs_root *root);
int btrfs_make_block_group(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 bytes_used,
u64 type, u64 chunk_objectid, u64 chunk_offset,
u64 size);
2008-09-26 22:09:34 +08:00
int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 group_start);
int btrfs_relocate_block_group(struct btrfs_root *root, u64 group_start);
int btrfs_free_reloc_root(struct btrfs_trans_handle *trans,
struct btrfs_root *root);
2008-09-26 22:09:34 +08:00
int btrfs_drop_dead_reloc_roots(struct btrfs_root *root);
int btrfs_reloc_tree_cache_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct extent_buffer *buf, u64 orig_start);
int btrfs_add_dead_reloc_root(struct btrfs_root *root);
int btrfs_cleanup_reloc_trees(struct btrfs_root *root);
/* ctree.c */
int btrfs_previous_item(struct btrfs_root *root,
struct btrfs_path *path, u64 min_objectid,
int type);
2008-09-26 22:09:34 +08:00
int btrfs_merge_path(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_key *node_keys,
u64 *nodes, int lowest_level);
int btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct btrfs_path *path,
struct btrfs_key *new_key);
struct extent_buffer *btrfs_root_node(struct btrfs_root *root);
struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root);
int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path,
struct btrfs_key *key, int lowest_level,
int cache_only, u64 min_trans);
int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key,
struct btrfs_key *max_key,
struct btrfs_path *path, int cache_only,
u64 min_trans);
int btrfs_cow_block(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct extent_buffer *buf,
struct extent_buffer *parent, int parent_slot,
struct extent_buffer **cow_ret, u64 prealloc_dest);
int btrfs_copy_root(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct extent_buffer *buf,
struct extent_buffer **cow_ret, u64 new_root_objectid);
int btrfs_extend_item(struct btrfs_trans_handle *trans, struct btrfs_root
*root, struct btrfs_path *path, u32 data_size);
int btrfs_truncate_item(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
u32 new_size, int from_end);
int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root
*root, struct btrfs_key *key, struct btrfs_path *p, int
ins_len, int cow);
int btrfs_realloc_node(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct extent_buffer *parent,
int start_slot, int cache_only, u64 *last_ret,
struct btrfs_key *progress);
void btrfs_release_path(struct btrfs_root *root, struct btrfs_path *p);
struct btrfs_path *btrfs_alloc_path(void);
void btrfs_free_path(struct btrfs_path *p);
void btrfs_init_path(struct btrfs_path *p);
int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
struct btrfs_path *path, int slot, int nr);
int btrfs_del_leaf(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path, u64 bytenr);
static inline int btrfs_del_item(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path)
{
return btrfs_del_items(trans, root, path, path->slots[0], 1);
}
int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root
*root, struct btrfs_key *key, void *data, u32 data_size);
Btrfs: batch extent inserts/updates/deletions on the extent root While profiling the allocator I noticed a good amount of time was being spent in finish_current_insert and del_pending_extents, and as the filesystem filled up more and more time was being spent in those functions. This patch aims to try and reduce that problem. This happens two ways 1) track if we tried to delete an extent that we are going to update or insert. Once we get into finish_current_insert we discard any of the extents that were marked for deletion. This saves us from doing unnecessary work almost every time finish_current_insert runs. 2) Batch insertion/updates/deletions. Instead of doing a btrfs_search_slot for each individual extent and doing the needed operation, we instead keep the leaf around and see if there is anything else we can do on that leaf. On the insert case I introduced a btrfs_insert_some_items, which will take an array of keys with an array of data_sizes and try and squeeze in as many of those keys as possible, and then return how many keys it was able to insert. In the update case we search for an extent ref, update the ref and then loop through the leaf to see if any of the other refs we are looking to update are on that leaf, and then once we are done we release the path and search for the next ref we need to update. And finally for the deletion we try and delete the extent+ref in pairs, so we will try to find extent+ref pairs next to the extent we are trying to free and free them in bulk if possible. This along with the other cluster fix that Chris pushed out a bit ago helps make the allocator preform more uniformly as it fills up the disk. There is still a slight drop as we fill up the disk since we start having to stick new blocks in odd places which results in more COW's than on a empty fs, but the drop is not nearly as severe as it was before. Signed-off-by: Josef Bacik <jbacik@redhat.com>
2008-11-13 03:19:50 +08:00
int btrfs_insert_some_items(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
struct btrfs_key *cpu_key, u32 *data_size,
int nr);
int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
struct btrfs_key *cpu_key, u32 *data_size, int nr);
static inline int btrfs_insert_empty_item(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
struct btrfs_key *key,
u32 data_size)
{
return btrfs_insert_empty_items(trans, root, path, key, &data_size, 1);
}
int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path);
int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path);
int btrfs_leaf_free_space(struct btrfs_root *root, struct extent_buffer *leaf);
int btrfs_drop_snapshot(struct btrfs_trans_handle *trans, struct btrfs_root
*root);
int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct extent_buffer *node,
struct extent_buffer *parent);
/* root-item.c */
int btrfs_del_root(struct btrfs_trans_handle *trans, struct btrfs_root *root,
struct btrfs_key *key);
int btrfs_insert_root(struct btrfs_trans_handle *trans, struct btrfs_root
*root, struct btrfs_key *key, struct btrfs_root_item
*item);
int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
*root, struct btrfs_key *key, struct btrfs_root_item
*item);
int btrfs_find_last_root(struct btrfs_root *root, u64 objectid, struct
btrfs_root_item *item, struct btrfs_key *key);
int btrfs_search_root(struct btrfs_root *root, u64 search_start,
u64 *found_objectid);
int btrfs_find_dead_roots(struct btrfs_root *root, u64 objectid,
struct btrfs_root *latest_root);
/* dir-item.c */
int btrfs_insert_dir_item(struct btrfs_trans_handle *trans, struct btrfs_root
*root, const char *name, int name_len, u64 dir,
struct btrfs_key *location, u8 type, u64 index);
struct btrfs_dir_item *btrfs_lookup_dir_item(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path, u64 dir,
const char *name, int name_len,
int mod);
struct btrfs_dir_item *
btrfs_lookup_dir_index_item(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path, u64 dir,
u64 objectid, const char *name, int name_len,
int mod);
struct btrfs_dir_item *btrfs_match_dir_item_name(struct btrfs_root *root,
struct btrfs_path *path,
const char *name, int name_len);
int btrfs_delete_one_dir_name(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
struct btrfs_dir_item *di);
int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
struct btrfs_root *root, const char *name,
u16 name_len, const void *data, u16 data_len,
u64 dir);
struct btrfs_dir_item *btrfs_lookup_xattr(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path, u64 dir,
const char *name, u16 name_len,
int mod);
/* orphan.c */
int btrfs_insert_orphan_item(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 offset);
int btrfs_del_orphan_item(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 offset);
/* inode-map.c */
int btrfs_find_free_objectid(struct btrfs_trans_handle *trans,
struct btrfs_root *fs_root,
u64 dirid, u64 *objectid);
int btrfs_find_highest_inode(struct btrfs_root *fs_root, u64 *objectid);
/* inode-item.c */
int btrfs_insert_inode_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
const char *name, int name_len,
u64 inode_objectid, u64 ref_objectid, u64 index);
int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
const char *name, int name_len,
u64 inode_objectid, u64 ref_objectid, u64 *index);
int btrfs_insert_empty_inode(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path, u64 objectid);
int btrfs_lookup_inode(struct btrfs_trans_handle *trans, struct btrfs_root
*root, struct btrfs_path *path,
struct btrfs_key *location, int mod);
/* file-item.c */
int btrfs_lookup_bio_sums(struct btrfs_root *root, struct inode *inode,
struct bio *bio);
int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
struct btrfs_root *root,
u64 objectid, u64 pos,
u64 disk_offset, u64 disk_num_bytes,
u64 num_bytes, u64 offset, u64 ram_bytes,
u8 compression, u8 encryption, u16 other_encoding);
int btrfs_lookup_file_extent(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path, u64 objectid,
u64 bytenr, int mod);
int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct inode *inode,
struct btrfs_ordered_sum *sums);
int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
struct bio *bio);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
int btrfs_csum_file_bytes(struct btrfs_root *root, struct inode *inode,
u64 start, unsigned long len);
struct btrfs_csum_item *btrfs_lookup_csum(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
u64 objectid, u64 offset,
int cow);
int btrfs_csum_truncate(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct btrfs_path *path,
u64 isize);
/* inode.c */
/* RHEL and EL kernels have a patch that renames PG_checked to FsMisc */
#if defined(ClearPageFsMisc) && !defined(ClearPageChecked)
#define ClearPageChecked ClearPageFsMisc
#define SetPageChecked SetPageFsMisc
#define PageChecked PageFsMisc
#endif
int btrfs_unlink_inode(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct inode *dir, struct inode *inode,
const char *name, int name_len);
int btrfs_add_link(struct btrfs_trans_handle *trans,
struct inode *parent_inode, struct inode *inode,
const char *name, int name_len, int add_backref, u64 index);
int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct inode *inode, u64 new_size,
u32 min_type);
int btrfs_start_delalloc_inodes(struct btrfs_root *root);
int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end);
int btrfs_writepages(struct address_space *mapping,
struct writeback_control *wbc);
int btrfs_create_subvol_root(struct btrfs_root *new_root, struct dentry *dentry,
struct btrfs_trans_handle *trans, u64 new_dirid,
struct btrfs_block_group_cache *block_group);
void btrfs_invalidate_dcache_root(struct btrfs_root *root, char *name,
int namelen);
int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-30 02:49:59 +08:00
size_t size, struct bio *bio, unsigned long bio_flags);
unsigned long btrfs_force_ra(struct address_space *mapping,
struct file_ra_state *ra, struct file *file,
pgoff_t offset, pgoff_t last_index);
int btrfs_check_free_space(struct btrfs_root *root, u64 num_required,
int for_del);
int btrfs_page_mkwrite(struct vm_area_struct *vma, struct page *page);
int btrfs_readpage(struct file *file, struct page *page);
void btrfs_delete_inode(struct inode *inode);
void btrfs_put_inode(struct inode *inode);
void btrfs_read_locked_inode(struct inode *inode);
int btrfs_write_inode(struct inode *inode, int wait);
void btrfs_dirty_inode(struct inode *inode);
struct inode *btrfs_alloc_inode(struct super_block *sb);
void btrfs_destroy_inode(struct inode *inode);
int btrfs_init_cachep(void);
void btrfs_destroy_cachep(void);
long btrfs_ioctl_trans_end(struct file *file);
struct inode *btrfs_ilookup(struct super_block *s, u64 objectid,
struct btrfs_root *root, int wait);
struct inode *btrfs_iget_locked(struct super_block *s, u64 objectid,
struct btrfs_root *root);
struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location,
struct btrfs_root *root, int *is_new);
int btrfs_commit_write(struct file *file, struct page *page,
unsigned from, unsigned to);
struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page,
size_t page_offset, u64 start, u64 end,
int create);
int btrfs_update_inode(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct inode *inode);
int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode);
int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode);
void btrfs_orphan_cleanup(struct btrfs_root *root);
int btrfs_cont_expand(struct inode *inode, loff_t size);
/* ioctl.c */
long btrfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
/* file.c */
int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync);
int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
int skip_pinned);
int btrfs_check_file(struct btrfs_root *root, struct inode *inode);
extern struct file_operations btrfs_file_operations;
int btrfs_drop_extents(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct inode *inode,
u64 start, u64 end, u64 inline_limit, u64 *hint_block);
int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct inode *inode, u64 start, u64 end);
int btrfs_release_file(struct inode *inode, struct file *file);
/* tree-defrag.c */
int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
struct btrfs_root *root, int cache_only);
/* sysfs.c */
int btrfs_init_sysfs(void);
void btrfs_exit_sysfs(void);
int btrfs_sysfs_add_super(struct btrfs_fs_info *fs);
int btrfs_sysfs_add_root(struct btrfs_root *root);
void btrfs_sysfs_del_root(struct btrfs_root *root);
void btrfs_sysfs_del_super(struct btrfs_fs_info *root);
/* xattr.c */
ssize_t btrfs_listxattr(struct dentry *dentry, char *buffer, size_t size);
/* super.c */
u64 btrfs_parse_size(char *str);
int btrfs_parse_options(struct btrfs_root *root, char *options);
int btrfs_sync_fs(struct super_block *sb, int wait);
/* acl.c */
int btrfs_check_acl(struct inode *inode, int mask);
int btrfs_init_acl(struct inode *inode, struct inode *dir);
int btrfs_acl_chmod(struct inode *inode);
Btrfs: free space accounting redo 1) replace the per fs_info extent_io_tree that tracked free space with two rb-trees per block group to track free space areas via offset and size. The reason to do this is because most allocations come with a hint byte where to start, so we can usually find a chunk of free space at that hint byte to satisfy the allocation and get good space packing. If we cannot find free space at or after the given offset we fall back on looking for a chunk of the given size as close to that given offset as possible. When we fall back on the size search we also try to find a slot as close to the size we want as possible, to avoid breaking small chunks off of huge areas if possible. 2) remove the extent_io_tree that tracked the block group cache from fs_info and replaced it with an rb-tree thats tracks block group cache via offset. also added a per space_info list that tracks the block group cache for the particular space so we can lookup related block groups easily. 3) cleaned up the allocation code to make it a little easier to read and a little less complicated. Basically there are 3 steps, first look from our provided hint. If we couldn't find from that given hint, start back at our original search start and look for space from there. If that fails try to allocate space if we can and start looking again. If not we're screwed and need to start over again. 4) small fixes. there were some issues in volumes.c where we wouldn't allocate the rest of the disk. fixed cow_file_range to actually pass the alloc_hint, which has helped a good bit in making the fs_mark test I run have semi-normal results as we run out of space. Generally with data allocations we don't track where we last allocated from, so everytime we did a data allocation we'd search through every block group that we have looking for free space. Now searching a block group with no free space isn't terribly time consuming, it was causing a slight degradation as we got more data block groups. The alloc_hint has fixed this slight degredation and made things semi-normal. There is still one nagging problem I'm working on where we will get ENOSPC when there is definitely plenty of space. This only happens with metadata allocations, and only when we are almost full. So you generally hit the 85% mark first, but sometimes you'll hit the BUG before you hit the 85% wall. I'm still tracking it down, but until then this seems to be pretty stable and make a significant performance gain. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-09-24 01:14:11 +08:00
/* free-space-cache.c */
int btrfs_add_free_space(struct btrfs_block_group_cache *block_group,
u64 bytenr, u64 size);
int btrfs_add_free_space_lock(struct btrfs_block_group_cache *block_group,
u64 offset, u64 bytes);
Btrfs: free space accounting redo 1) replace the per fs_info extent_io_tree that tracked free space with two rb-trees per block group to track free space areas via offset and size. The reason to do this is because most allocations come with a hint byte where to start, so we can usually find a chunk of free space at that hint byte to satisfy the allocation and get good space packing. If we cannot find free space at or after the given offset we fall back on looking for a chunk of the given size as close to that given offset as possible. When we fall back on the size search we also try to find a slot as close to the size we want as possible, to avoid breaking small chunks off of huge areas if possible. 2) remove the extent_io_tree that tracked the block group cache from fs_info and replaced it with an rb-tree thats tracks block group cache via offset. also added a per space_info list that tracks the block group cache for the particular space so we can lookup related block groups easily. 3) cleaned up the allocation code to make it a little easier to read and a little less complicated. Basically there are 3 steps, first look from our provided hint. If we couldn't find from that given hint, start back at our original search start and look for space from there. If that fails try to allocate space if we can and start looking again. If not we're screwed and need to start over again. 4) small fixes. there were some issues in volumes.c where we wouldn't allocate the rest of the disk. fixed cow_file_range to actually pass the alloc_hint, which has helped a good bit in making the fs_mark test I run have semi-normal results as we run out of space. Generally with data allocations we don't track where we last allocated from, so everytime we did a data allocation we'd search through every block group that we have looking for free space. Now searching a block group with no free space isn't terribly time consuming, it was causing a slight degradation as we got more data block groups. The alloc_hint has fixed this slight degredation and made things semi-normal. There is still one nagging problem I'm working on where we will get ENOSPC when there is definitely plenty of space. This only happens with metadata allocations, and only when we are almost full. So you generally hit the 85% mark first, but sometimes you'll hit the BUG before you hit the 85% wall. I'm still tracking it down, but until then this seems to be pretty stable and make a significant performance gain. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-09-24 01:14:11 +08:00
int btrfs_remove_free_space(struct btrfs_block_group_cache *block_group,
u64 bytenr, u64 size);
int btrfs_remove_free_space_lock(struct btrfs_block_group_cache *block_group,
u64 offset, u64 bytes);
Btrfs: free space accounting redo 1) replace the per fs_info extent_io_tree that tracked free space with two rb-trees per block group to track free space areas via offset and size. The reason to do this is because most allocations come with a hint byte where to start, so we can usually find a chunk of free space at that hint byte to satisfy the allocation and get good space packing. If we cannot find free space at or after the given offset we fall back on looking for a chunk of the given size as close to that given offset as possible. When we fall back on the size search we also try to find a slot as close to the size we want as possible, to avoid breaking small chunks off of huge areas if possible. 2) remove the extent_io_tree that tracked the block group cache from fs_info and replaced it with an rb-tree thats tracks block group cache via offset. also added a per space_info list that tracks the block group cache for the particular space so we can lookup related block groups easily. 3) cleaned up the allocation code to make it a little easier to read and a little less complicated. Basically there are 3 steps, first look from our provided hint. If we couldn't find from that given hint, start back at our original search start and look for space from there. If that fails try to allocate space if we can and start looking again. If not we're screwed and need to start over again. 4) small fixes. there were some issues in volumes.c where we wouldn't allocate the rest of the disk. fixed cow_file_range to actually pass the alloc_hint, which has helped a good bit in making the fs_mark test I run have semi-normal results as we run out of space. Generally with data allocations we don't track where we last allocated from, so everytime we did a data allocation we'd search through every block group that we have looking for free space. Now searching a block group with no free space isn't terribly time consuming, it was causing a slight degradation as we got more data block groups. The alloc_hint has fixed this slight degredation and made things semi-normal. There is still one nagging problem I'm working on where we will get ENOSPC when there is definitely plenty of space. This only happens with metadata allocations, and only when we are almost full. So you generally hit the 85% mark first, but sometimes you'll hit the BUG before you hit the 85% wall. I'm still tracking it down, but until then this seems to be pretty stable and make a significant performance gain. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-09-24 01:14:11 +08:00
void btrfs_remove_free_space_cache(struct btrfs_block_group_cache
*block_group);
struct btrfs_free_space *btrfs_find_free_space(struct btrfs_block_group_cache
*block_group, u64 offset,
u64 bytes);
void btrfs_dump_free_space(struct btrfs_block_group_cache *block_group,
u64 bytes);
u64 btrfs_block_group_free_space(struct btrfs_block_group_cache *block_group);
#endif