blob: a0637e5057aee0488c9fe0d5c58d72c5b98f684a [file] [log] [blame]
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001/*
Christoph Hellwig3dcf5452008-04-29 18:13:32 -04002 * ext4.h
Dave Kleikampac27a0e2006-10-11 01:20:50 -07003 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
Christoph Hellwig3dcf5452008-04-29 18:13:32 -040016#ifndef _EXT4_H
17#define _EXT4_H
Dave Kleikampac27a0e2006-10-11 01:20:50 -070018
19#include <linux/types.h>
Mingming Cao3a5b2ec2006-10-11 01:21:05 -070020#include <linux/blkdev.h>
Dave Kleikampac27a0e2006-10-11 01:20:50 -070021#include <linux/magic.h>
Aneesh Kumar K.V2ccb5fb2009-01-05 21:49:55 -050022#include <linux/jbd2.h>
Mingming Cao60e58e02009-01-22 18:13:05 +010023#include <linux/quota.h>
Theodore Ts'od444c3c2009-05-01 13:44:33 -040024#include <linux/rwsem.h>
25#include <linux/rbtree.h>
26#include <linux/seqlock.h>
27#include <linux/mutex.h>
Theodore Ts'oca0faba2009-05-03 16:33:44 -040028#include <linux/timer.h>
29#include <linux/wait.h>
30#include <linux/blockgroup_lock.h>
31#include <linux/percpu_counter.h>
Darrick J. Wong04419842012-04-29 18:27:10 -040032#include <crypto/hash.h>
Ben Hutchings4d92dc02010-05-17 06:00:00 -040033#ifdef __KERNEL__
34#include <linux/compat.h>
35#endif
Alex Tomasc9de5602008-01-29 00:19:52 -050036
Dave Kleikampac27a0e2006-10-11 01:20:50 -070037/*
Shen Feng8a356942008-07-11 19:27:31 -040038 * The fourth extended filesystem constants/structures
Dave Kleikampac27a0e2006-10-11 01:20:50 -070039 */
40
41/*
Mingming Cao617ba132006-10-11 01:20:53 -070042 * Define EXT4FS_DEBUG to produce debug messages
Dave Kleikampac27a0e2006-10-11 01:20:50 -070043 */
Mingming Cao617ba132006-10-11 01:20:53 -070044#undef EXT4FS_DEBUG
Dave Kleikampac27a0e2006-10-11 01:20:50 -070045
46/*
Dave Kleikampac27a0e2006-10-11 01:20:50 -070047 * Debug code
48 */
Mingming Cao617ba132006-10-11 01:20:53 -070049#ifdef EXT4FS_DEBUG
50#define ext4_debug(f, a...) \
Dave Kleikampac27a0e2006-10-11 01:20:50 -070051 do { \
Theodore Ts'oaf5bc922008-09-08 22:25:24 -040052 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
Stoyan Gaydarov4db9c542008-07-13 21:03:29 -040053 __FILE__, __LINE__, __func__); \
Theodore Ts'oaf5bc922008-09-08 22:25:24 -040054 printk(KERN_DEBUG f, ## a); \
Dave Kleikampac27a0e2006-10-11 01:20:50 -070055 } while (0)
56#else
Joe Perchesace36ad2012-03-19 23:11:43 -040057#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
Dave Kleikampac27a0e2006-10-11 01:20:50 -070058#endif
59
Theodore Ts'o4a092d72012-11-28 13:03:30 -050060/*
61 * Turn on EXT_DEBUG to get lots of info about extents operations.
62 */
63#define EXT_DEBUG__
64#ifdef EXT_DEBUG
65#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
66#else
67#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
68#endif
69
Frank Mayhar273df552010-03-02 11:46:09 -050070#define EXT4_ERROR_INODE(inode, fmt, a...) \
Theodore Ts'oc398eda2010-07-27 11:56:40 -040071 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
72
73#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
74 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
Frank Mayhar273df552010-03-02 11:46:09 -050075
Theodore Ts'of7c21172011-01-10 12:10:55 -050076#define EXT4_ERROR_FILE(file, block, fmt, a...) \
77 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
Frank Mayhar273df552010-03-02 11:46:09 -050078
Theodore Ts'od444c3c2009-05-01 13:44:33 -040079/* data type for block offset of block group */
80typedef int ext4_grpblk_t;
81
82/* data type for filesystem-wide blocks number */
83typedef unsigned long long ext4_fsblk_t;
84
85/* data type for file logical block number */
86typedef __u32 ext4_lblk_t;
87
88/* data type for block group number */
89typedef unsigned int ext4_group_t;
90
Theodore Ts'o296c3552009-09-30 00:32:42 -040091/*
Theodore Ts'o60e66792010-05-17 07:00:00 -040092 * Flags used in mballoc's allocation_context flags field.
Theodore Ts'o296c3552009-09-30 00:32:42 -040093 *
94 * Also used to show what's going on for debugging purposes when the
95 * flag field is exported via the traceport interface
96 */
Theodore Ts'od444c3c2009-05-01 13:44:33 -040097
Alex Tomasc9de5602008-01-29 00:19:52 -050098/* prefer goal again. length */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -040099#define EXT4_MB_HINT_MERGE 0x0001
Alex Tomasc9de5602008-01-29 00:19:52 -0500100/* blocks already reserved */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400101#define EXT4_MB_HINT_RESERVED 0x0002
Alex Tomasc9de5602008-01-29 00:19:52 -0500102/* metadata is being allocated */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400103#define EXT4_MB_HINT_METADATA 0x0004
Alex Tomasc9de5602008-01-29 00:19:52 -0500104/* first blocks in the file */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400105#define EXT4_MB_HINT_FIRST 0x0008
Alex Tomasc9de5602008-01-29 00:19:52 -0500106/* search for the best chunk */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400107#define EXT4_MB_HINT_BEST 0x0010
Alex Tomasc9de5602008-01-29 00:19:52 -0500108/* data is being allocated */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400109#define EXT4_MB_HINT_DATA 0x0020
Alex Tomasc9de5602008-01-29 00:19:52 -0500110/* don't preallocate (for tails) */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400111#define EXT4_MB_HINT_NOPREALLOC 0x0040
Alex Tomasc9de5602008-01-29 00:19:52 -0500112/* allocate for locality group */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400113#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
Alex Tomasc9de5602008-01-29 00:19:52 -0500114/* allocate goal blocks or none */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400115#define EXT4_MB_HINT_GOAL_ONLY 0x0100
Alex Tomasc9de5602008-01-29 00:19:52 -0500116/* goal is meaningful */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400117#define EXT4_MB_HINT_TRY_GOAL 0x0200
Mingming Caod2a17632008-07-14 17:52:37 -0400118/* blocks already pre-reserved by delayed allocation */
Theodore Ts'o0ef90db2009-08-09 16:46:13 -0400119#define EXT4_MB_DELALLOC_RESERVED 0x0400
Theodore Ts'o4ba74d02009-08-09 22:01:13 -0400120/* We are doing stream allocation */
121#define EXT4_MB_STREAM_ALLOC 0x0800
Allison Henderson55f020d2011-05-25 07:41:26 -0400122/* Use reserved root blocks if needed */
123#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
Alex Tomasc9de5602008-01-29 00:19:52 -0500124
125struct ext4_allocation_request {
126 /* target inode for block we're allocating */
127 struct inode *inode;
Alex Tomasc9de5602008-01-29 00:19:52 -0500128 /* how many blocks we want to allocate */
Theodore Ts'o498e5f22008-11-05 00:14:04 -0500129 unsigned int len;
Eric Sandeen726447d2009-07-13 10:24:17 -0400130 /* logical block in target inode */
131 ext4_lblk_t logical;
132 /* the closest logical allocated block to the left */
133 ext4_lblk_t lleft;
134 /* the closest logical allocated block to the right */
135 ext4_lblk_t lright;
136 /* phys. target (a hint) */
137 ext4_fsblk_t goal;
138 /* phys. block for the closest logical allocated block to the left */
139 ext4_fsblk_t pleft;
140 /* phys. block for the closest logical allocated block to the right */
141 ext4_fsblk_t pright;
Alex Tomasc9de5602008-01-29 00:19:52 -0500142 /* flags. see above EXT4_MB_HINT_* */
Theodore Ts'o498e5f22008-11-05 00:14:04 -0500143 unsigned int flags;
Alex Tomasc9de5602008-01-29 00:19:52 -0500144};
145
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700146/*
Theodore Ts'oe35fd662010-05-16 19:00:00 -0400147 * Logical to physical block mapping, used by ext4_map_blocks()
148 *
149 * This structure is used to pass requests into ext4_map_blocks() as
150 * well as to store the information returned by ext4_map_blocks(). It
151 * takes less room on the stack than a struct buffer_head.
152 */
153#define EXT4_MAP_NEW (1 << BH_New)
154#define EXT4_MAP_MAPPED (1 << BH_Mapped)
155#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
156#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
157#define EXT4_MAP_UNINIT (1 << BH_Uninit)
Aditya Kali7b415bf2011-09-09 19:04:51 -0400158/* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
159 * ext4_map_blocks wants to know whether or not the underlying cluster has
160 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
161 * the requested mapping was from previously mapped (or delayed allocated)
162 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
163 * should never appear on buffer_head's state flags.
164 */
165#define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
Theodore Ts'oe35fd662010-05-16 19:00:00 -0400166#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
167 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
Aditya Kali7b415bf2011-09-09 19:04:51 -0400168 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
Theodore Ts'oe35fd662010-05-16 19:00:00 -0400169
170struct ext4_map_blocks {
171 ext4_fsblk_t m_pblk;
172 ext4_lblk_t m_lblk;
173 unsigned int m_len;
174 unsigned int m_flags;
175};
176
177/*
Theodore Ts'ob3a3ca82009-08-31 23:13:11 -0400178 * For delayed allocation tracking
179 */
180struct mpage_da_data {
181 struct inode *inode;
182 sector_t b_blocknr; /* start block number of extent */
183 size_t b_size; /* size of extent */
184 unsigned long b_state; /* state of the extent */
185 unsigned long first_page, next_page; /* extent of pages */
186 struct writeback_control *wbc;
187 int io_done;
188 int pages_written;
189 int retval;
190};
Theodore Ts'obd2d0212010-10-27 21:30:10 -0400191
192/*
193 * Flags for ext4_io_end->flags
194 */
195#define EXT4_IO_END_UNWRITTEN 0x0001
196#define EXT4_IO_END_ERROR 0x0002
Jan Kara84c17542013-01-28 09:43:46 -0500197#define EXT4_IO_END_DIRECT 0x0004
Theodore Ts'obd2d0212010-10-27 21:30:10 -0400198
199struct ext4_io_page {
200 struct page *p_page;
Theodore Ts'o83668e72010-11-08 13:45:33 -0500201 atomic_t p_count;
Theodore Ts'obd2d0212010-10-27 21:30:10 -0400202};
203
204#define MAX_IO_PAGES 128
205
Curt Wohlgemuth41881882012-03-05 10:40:22 -0500206/*
207 * For converting uninitialized extents on a work queue.
208 *
209 * 'page' is only used from the writepage() path; 'pages' is only used for
210 * buffered writes; they are used to keep page references until conversion
211 * takes place. For AIO/DIO, neither field is filled in.
212 */
Mingming Cao00314622009-09-28 15:49:08 -0400213typedef struct ext4_io_end {
jiayingz@google.com (Jiaying Zhang)5b3ff232010-07-27 11:56:06 -0400214 struct list_head list; /* per-file finished IO list */
Mingming Cao00314622009-09-28 15:49:08 -0400215 struct inode *inode; /* file being written to */
Mingming Cao8d5d02e2009-09-28 15:48:29 -0400216 unsigned int flag; /* unwritten or not */
Eric Sandeena1de02d2010-02-04 23:58:38 -0500217 loff_t offset; /* offset in the file */
218 ssize_t size; /* size of the extent */
jiayingz@google.com (Jiaying Zhang)5b3ff232010-07-27 11:56:06 -0400219 struct kiocb *iocb; /* iocb struct for AIO */
220 int result; /* error value for AIO */
Curt Wohlgemuth41881882012-03-05 10:40:22 -0500221 int num_io_pages; /* for writepages() */
222 struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */
Mingming Cao00314622009-09-28 15:49:08 -0400223} ext4_io_end_t;
224
Theodore Ts'obd2d0212010-10-27 21:30:10 -0400225struct ext4_io_submit {
226 int io_op;
227 struct bio *io_bio;
228 ext4_io_end_t *io_end;
229 struct ext4_io_page *io_page;
230 sector_t io_next_block;
231};
232
Theodore Ts'ob3a3ca82009-08-31 23:13:11 -0400233/*
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700234 * Special inodes numbers
235 */
Mingming Cao617ba132006-10-11 01:20:53 -0700236#define EXT4_BAD_INO 1 /* Bad blocks inode */
237#define EXT4_ROOT_INO 2 /* Root inode */
Aditya Kaliae812302011-05-24 19:00:39 -0400238#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
239#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
Mingming Cao617ba132006-10-11 01:20:53 -0700240#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
241#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
242#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
243#define EXT4_JOURNAL_INO 8 /* Journal inode */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700244
Mingming Cao617ba132006-10-11 01:20:53 -0700245/* First non-reserved inode for old ext4 filesystems */
246#define EXT4_GOOD_OLD_FIRST_INO 11
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700247
248/*
249 * Maximal count of links to a file
250 */
Andreas Dilgerf8628a12007-07-18 08:38:01 -0400251#define EXT4_LINK_MAX 65000
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700252
253/*
254 * Macro-instructions used to manage several block sizes
255 */
Mingming Cao617ba132006-10-11 01:20:53 -0700256#define EXT4_MIN_BLOCK_SIZE 1024
Takashi Satoafc7cbc2008-01-28 23:58:27 -0500257#define EXT4_MAX_BLOCK_SIZE 65536
258#define EXT4_MIN_BLOCK_LOG_SIZE 10
Curt Wohlgemuthfb1813f2010-10-27 21:29:12 -0400259#define EXT4_MAX_BLOCK_LOG_SIZE 16
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700260#ifdef __KERNEL__
Mingming Cao617ba132006-10-11 01:20:53 -0700261# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700262#else
Mingming Cao617ba132006-10-11 01:20:53 -0700263# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700264#endif
Theodore Ts'oaf5bc922008-09-08 22:25:24 -0400265#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
Theodore Ts'o281b5992011-09-09 18:34:51 -0400266#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
267 EXT4_SB(s)->s_cluster_bits)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700268#ifdef __KERNEL__
Mingming Cao617ba132006-10-11 01:20:53 -0700269# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
Theodore Ts'o281b5992011-09-09 18:34:51 -0400270# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700271#else
Mingming Cao617ba132006-10-11 01:20:53 -0700272# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700273#endif
274#ifdef __KERNEL__
Mingming Cao617ba132006-10-11 01:20:53 -0700275#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
276#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
277#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700278#else
Mingming Cao617ba132006-10-11 01:20:53 -0700279#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
280 EXT4_GOOD_OLD_INODE_SIZE : \
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700281 (s)->s_inode_size)
Mingming Cao617ba132006-10-11 01:20:53 -0700282#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
283 EXT4_GOOD_OLD_FIRST_INO : \
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700284 (s)->s_first_ino)
285#endif
Amit Aroraa2df2a62007-07-17 21:42:41 -0400286#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700287
Theodore Ts'od5b8f312011-09-09 18:44:51 -0400288/* Translate a block number to a cluster number */
289#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
290/* Translate a cluster number to a block number */
291#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
292/* Translate # of blks to # of clusters */
293#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
294 (sbi)->s_cluster_bits)
295
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700296/*
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700297 * Structure of a blocks group descriptor
298 */
Mingming Cao617ba132006-10-11 01:20:53 -0700299struct ext4_group_desc
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700300{
Aneesh Kumar K.V3a145892007-10-16 18:38:25 -0400301 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
Aneesh Kumar K.V5272f832007-10-16 18:38:25 -0400302 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
303 __le32 bg_inode_table_lo; /* Inodes table block */
Aneesh Kumar K.V560671a2009-01-05 22:20:24 -0500304 __le16 bg_free_blocks_count_lo;/* Free blocks count */
305 __le16 bg_free_inodes_count_lo;/* Free inodes count */
306 __le16 bg_used_dirs_count_lo; /* Directories count */
Andreas Dilger717d50e2007-10-16 18:38:25 -0400307 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
Darrick J. Wonge6153912012-04-29 18:23:10 -0400308 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
309 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
310 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
Aneesh Kumar K.V560671a2009-01-05 22:20:24 -0500311 __le16 bg_itable_unused_lo; /* Unused inodes count */
Andreas Dilger717d50e2007-10-16 18:38:25 -0400312 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
Alexandre Ratchov8fadc142006-10-11 01:21:15 -0700313 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
314 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
315 __le32 bg_inode_table_hi; /* Inodes table block MSB */
Coly Li91b51a02008-01-28 23:58:27 -0500316 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
317 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
318 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
Aneesh Kumar K.V560671a2009-01-05 22:20:24 -0500319 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
Darrick J. Wonge6153912012-04-29 18:23:10 -0400320 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
321 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
322 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
323 __u32 bg_reserved;
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700324};
325
Darrick J. Wong41a246d2012-04-29 18:33:10 -0400326#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
327 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
328 sizeof(__le16))
329#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
330 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
331 sizeof(__le16))
332
Jose R. Santos772cb7c2008-07-11 19:27:31 -0400333/*
334 * Structure of a flex block group info
335 */
336
337struct flex_groups {
Theodore Ts'o90ba9832013-03-11 23:39:59 -0400338 atomic64_t free_clusters;
339 atomic_t free_inodes;
340 atomic_t used_dirs;
Jose R. Santos772cb7c2008-07-11 19:27:31 -0400341};
342
Andreas Dilger717d50e2007-10-16 18:38:25 -0400343#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
344#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
345#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
346
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700347/*
348 * Macro-instructions used to manage group descriptors
349 */
Alexandre Ratchov0d1ee422006-10-11 01:21:14 -0700350#define EXT4_MIN_DESC_SIZE 32
Alexandre Ratchov8fadc142006-10-11 01:21:15 -0700351#define EXT4_MIN_DESC_SIZE_64BIT 64
Alexandre Ratchov0d1ee422006-10-11 01:21:14 -0700352#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
353#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700354#ifdef __KERNEL__
Mingming Cao617ba132006-10-11 01:20:53 -0700355# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
Theodore Ts'o281b5992011-09-09 18:34:51 -0400356# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
Mingming Cao617ba132006-10-11 01:20:53 -0700357# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
358# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
359# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700360#else
Mingming Cao617ba132006-10-11 01:20:53 -0700361# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
Alexandre Ratchov0d1ee422006-10-11 01:21:14 -0700362# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
Mingming Cao617ba132006-10-11 01:20:53 -0700363# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700364#endif
365
366/*
367 * Constants relative to the data blocks
368 */
Mingming Cao617ba132006-10-11 01:20:53 -0700369#define EXT4_NDIR_BLOCKS 12
370#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
371#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
372#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
373#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700374
375/*
376 * Inode flags
377 */
Mingming Cao617ba132006-10-11 01:20:53 -0700378#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
379#define EXT4_UNRM_FL 0x00000002 /* Undelete */
380#define EXT4_COMPR_FL 0x00000004 /* Compress file */
381#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
382#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
383#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
384#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
385#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700386/* Reserved for compression usage... */
Mingming Cao617ba132006-10-11 01:20:53 -0700387#define EXT4_DIRTY_FL 0x00000100
388#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
389#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
390#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700391/* End compression flags --- maybe not all used */
Mingming Cao617ba132006-10-11 01:20:53 -0700392#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
393#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
394#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
395#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
396#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
397#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
Aneesh Kumar K.V8180a562008-01-28 23:58:27 -0500398#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
Alex Tomasa86c6182006-10-11 01:21:03 -0700399#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
Theodore Ts'of710b4b2010-01-25 03:31:32 -0500400#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
Jiaying Zhangc8d46e42010-02-24 09:52:53 -0500401#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
Tao Ma67cf5b02012-12-10 14:04:46 -0500402#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
Aneesh Kumar K.V8180a562008-01-28 23:58:27 -0500403#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700404
Jiaying Zhangc8d46e42010-02-24 09:52:53 -0500405#define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
Theodore Ts'o996bb9f2013-04-03 22:04:52 -0400406#define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700407
Duane Griffin8fa43a82009-02-15 18:57:26 -0500408/* Flags that should be inherited by new inodes from their parent. */
409#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
Theodore Ts'o1cd9f092011-08-31 11:54:51 -0400410 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
Duane Griffin8fa43a82009-02-15 18:57:26 -0500411 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
412 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
413
Duane Griffin2dc6b0d2009-02-15 18:09:20 -0500414/* Flags that are appropriate for regular files (all but dir-specific ones). */
415#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
416
417/* Flags that are appropriate for non-directories/regular files. */
418#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
419
420/* Mask out flags that are inappropriate for the given type of inode. */
421static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
422{
423 if (S_ISDIR(mode))
424 return flags;
425 else if (S_ISREG(mode))
426 return flags & EXT4_REG_FLMASK;
427 else
428 return flags & EXT4_OTHER_FLMASK;
429}
430
Dmitry Monakhov12e9b892010-05-16 22:00:00 -0400431/*
432 * Inode flags used for atomic set/get
433 */
434enum {
435 EXT4_INODE_SECRM = 0, /* Secure deletion */
436 EXT4_INODE_UNRM = 1, /* Undelete */
437 EXT4_INODE_COMPR = 2, /* Compress file */
438 EXT4_INODE_SYNC = 3, /* Synchronous updates */
439 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
440 EXT4_INODE_APPEND = 5, /* writes to file may only append */
441 EXT4_INODE_NODUMP = 6, /* do not dump file */
442 EXT4_INODE_NOATIME = 7, /* do not update atime */
443/* Reserved for compression usage... */
444 EXT4_INODE_DIRTY = 8,
445 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
446 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
447 EXT4_INODE_ECOMPR = 11, /* Compression error */
448/* End compression flags --- maybe not all used */
449 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
450 EXT4_INODE_IMAGIC = 13, /* AFS directory */
451 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
452 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
453 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
454 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
455 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
456 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
457 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
458 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
Tao Ma67cf5b02012-12-10 14:04:46 -0500459 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
Dmitry Monakhov12e9b892010-05-16 22:00:00 -0400460 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
461};
462
Dmitry Monakhov12e9b892010-05-16 22:00:00 -0400463/*
Carlos Maiolino9a4c8012012-12-10 16:30:45 -0500464 * Since it's pretty easy to mix up bit numbers and hex values, we use a
465 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
466 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
467 * any extra space in the compiled kernel image, otherwise, the build will fail.
468 * It's important that these values are the same, since we are using
469 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
470 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
471 * values found in ext2, ext3 and ext4 filesystems, and of course the values
472 * defined in e2fsprogs.
Dmitry Monakhov12e9b892010-05-16 22:00:00 -0400473 *
474 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
475 */
Carlos Maiolino9a4c8012012-12-10 16:30:45 -0500476#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
477#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
478
Dmitry Monakhov12e9b892010-05-16 22:00:00 -0400479static inline void ext4_check_flag_values(void)
480{
481 CHECK_FLAG_VALUE(SECRM);
482 CHECK_FLAG_VALUE(UNRM);
483 CHECK_FLAG_VALUE(COMPR);
484 CHECK_FLAG_VALUE(SYNC);
485 CHECK_FLAG_VALUE(IMMUTABLE);
486 CHECK_FLAG_VALUE(APPEND);
487 CHECK_FLAG_VALUE(NODUMP);
488 CHECK_FLAG_VALUE(NOATIME);
489 CHECK_FLAG_VALUE(DIRTY);
490 CHECK_FLAG_VALUE(COMPRBLK);
491 CHECK_FLAG_VALUE(NOCOMPR);
492 CHECK_FLAG_VALUE(ECOMPR);
493 CHECK_FLAG_VALUE(INDEX);
494 CHECK_FLAG_VALUE(IMAGIC);
495 CHECK_FLAG_VALUE(JOURNAL_DATA);
496 CHECK_FLAG_VALUE(NOTAIL);
497 CHECK_FLAG_VALUE(DIRSYNC);
498 CHECK_FLAG_VALUE(TOPDIR);
499 CHECK_FLAG_VALUE(HUGE_FILE);
500 CHECK_FLAG_VALUE(EXTENTS);
501 CHECK_FLAG_VALUE(EA_INODE);
502 CHECK_FLAG_VALUE(EOFBLOCKS);
Tao Ma67cf5b02012-12-10 14:04:46 -0500503 CHECK_FLAG_VALUE(INLINE_DATA);
Dmitry Monakhov12e9b892010-05-16 22:00:00 -0400504 CHECK_FLAG_VALUE(RESERVED);
505}
506
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700507/* Used to pass group descriptor data when online resize is done */
Mingming Cao617ba132006-10-11 01:20:53 -0700508struct ext4_new_group_input {
Dave Kleikamp8c55e202007-05-24 13:04:54 -0400509 __u32 group; /* Group number for this data */
510 __u64 block_bitmap; /* Absolute block number of block bitmap */
511 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
512 __u64 inode_table; /* Absolute block number of inode table start */
513 __u32 blocks_count; /* Total number of blocks in this group */
514 __u16 reserved_blocks; /* Number of reserved blocks in this group */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700515 __u16 unused;
516};
517
Ben Hutchings4d92dc02010-05-17 06:00:00 -0400518#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
519struct compat_ext4_new_group_input {
520 u32 group;
521 compat_u64 block_bitmap;
522 compat_u64 inode_bitmap;
523 compat_u64 inode_table;
524 u32 blocks_count;
525 u16 reserved_blocks;
526 u16 unused;
527};
528#endif
529
Mingming Cao617ba132006-10-11 01:20:53 -0700530/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
531struct ext4_new_group_data {
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700532 __u32 group;
Laurent Vivierbd81d8e2006-10-11 01:21:10 -0700533 __u64 block_bitmap;
534 __u64 inode_bitmap;
535 __u64 inode_table;
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700536 __u32 blocks_count;
537 __u16 reserved_blocks;
538 __u16 unused;
539 __u32 free_blocks_count;
540};
541
Yongqiang Yang33afdcc2012-01-03 23:32:52 -0500542/* Indexes used to index group tables in ext4_new_group_data */
543enum {
544 BLOCK_BITMAP = 0, /* block bitmap */
545 INODE_BITMAP, /* inode bitmap */
546 INODE_TABLE, /* inode tables */
547 GROUP_TABLE_COUNT,
548};
549
Amit Aroraa2df2a62007-07-17 21:42:41 -0400550/*
Eric Sandeen79e83032010-07-27 11:56:07 -0400551 * Flags used by ext4_map_blocks()
Amit Aroraa2df2a62007-07-17 21:42:41 -0400552 */
Theodore Ts'oc2177052009-05-14 00:58:52 -0400553 /* Allocate any needed blocks and/or convert an unitialized
554 extent to be an initialized ext4 */
Theodore Ts'o2ac3b6e2009-05-14 13:57:08 -0400555#define EXT4_GET_BLOCKS_CREATE 0x0001
Theodore Ts'oc2177052009-05-14 00:58:52 -0400556 /* Request the creation of an unitialized extent */
Theodore Ts'o2ac3b6e2009-05-14 13:57:08 -0400557#define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
Theodore Ts'oc2177052009-05-14 00:58:52 -0400558#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
559 EXT4_GET_BLOCKS_CREATE)
Theodore Ts'oc2177052009-05-14 00:58:52 -0400560 /* Caller is from the delayed allocation writeout path,
Theodore Ts'o60e66792010-05-17 07:00:00 -0400561 so set the magic i_delalloc_reserve_flag after taking the
Theodore Ts'o2ac3b6e2009-05-14 13:57:08 -0400562 inode allocation semaphore for */
Jan Kara03f5d8b2009-06-09 00:17:05 -0400563#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
Mingming Cao00314622009-09-28 15:49:08 -0400564 /* caller is from the direct IO path, request to creation of an
565 unitialized extents if not allocated, split the uninitialized
566 extent if blocks has been preallocated already*/
Jiaying Zhangc7064ef2010-03-02 13:28:44 -0500567#define EXT4_GET_BLOCKS_PRE_IO 0x0008
Aneesh Kumar K.V1296cc82010-01-15 01:27:59 -0500568#define EXT4_GET_BLOCKS_CONVERT 0x0010
Jiaying Zhangc7064ef2010-03-02 13:28:44 -0500569#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
Mingming Cao00314622009-09-28 15:49:08 -0400570 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
Jiaying Zhangc7064ef2010-03-02 13:28:44 -0500571 /* Convert extent to initialized after IO complete */
572#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
Jiaying Zhang744692d2010-03-04 16:14:02 -0500573 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
Allison Henderson55f020d2011-05-25 07:41:26 -0400574 /* Punch out blocks of an extent */
575#define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
Vivek Haldar556b27a2011-05-25 07:41:54 -0400576 /* Don't normalize allocation size (used for fallocate) */
577#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
Dmitry Monakhova4e5d882011-10-25 08:15:12 -0400578 /* Request will not result in inode size update (user for fallocate) */
579#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
Zheng Liu729f52c2012-07-09 16:29:29 -0400580 /* Do not take i_data_sem locking in ext4_map_blocks */
581#define EXT4_GET_BLOCKS_NO_LOCK 0x0100
Zheng Liud100eef2013-02-18 00:29:59 -0500582 /* Do not put hole in extent cache */
583#define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700584
585/*
Theodore Ts'oe6362602009-11-23 07:17:05 -0500586 * Flags used by ext4_free_blocks
587 */
588#define EXT4_FREE_BLOCKS_METADATA 0x0001
589#define EXT4_FREE_BLOCKS_FORGET 0x0002
Theodore Ts'o1f2acb62010-01-22 17:40:42 -0500590#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
Maxim Patlasov7132de72011-07-10 19:37:48 -0400591#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
Theodore Ts'o84130192011-09-09 18:50:51 -0400592#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
593#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
Theodore Ts'oe6362602009-11-23 07:17:05 -0500594
595/*
Allison Henderson4e96b2d2011-09-03 11:51:09 -0400596 * Flags used by ext4_discard_partial_page_buffers
597 */
598#define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700599
600/*
601 * ioctl commands
602 */
Mingming Cao617ba132006-10-11 01:20:53 -0700603#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
604#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
605#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
606#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
Mingming Cao617ba132006-10-11 01:20:53 -0700607#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
608#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
Mingming Cao617ba132006-10-11 01:20:53 -0700609#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
610#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
Theodore Ts'o8eea80d2008-09-13 19:54:35 -0400611#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
612#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
613#define EXT4_IOC_MIGRATE _IO('f', 9)
Theodore Ts'occd25062009-02-26 01:04:07 -0500614 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
Theodore Ts'o8eea80d2008-09-13 19:54:35 -0400615 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
Theodore Ts'occd25062009-02-26 01:04:07 -0500616#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
Akira Fujita748de672009-06-17 19:24:03 -0400617#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
Yongqiang Yang19c52462012-01-04 17:09:44 -0500618#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700619
Ben Hutchings899ad0c2010-05-17 05:00:00 -0400620#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700621/*
622 * ioctl commands in 32 bit emulation
623 */
Mingming Cao617ba132006-10-11 01:20:53 -0700624#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
625#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
626#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
627#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
628#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
629#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
630#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
Ben Hutchings4d92dc02010-05-17 06:00:00 -0400631#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
Mingming Cao617ba132006-10-11 01:20:53 -0700632#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
633#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
Ben Hutchings899ad0c2010-05-17 05:00:00 -0400634#endif
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700635
Uwe Kleine-Königb5950762010-11-01 15:38:34 -0400636/* Max physical block we can address w/o extents */
Eric Sandeenfb0a3872009-09-16 14:45:10 -0400637#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
638
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700639/*
640 * Structure of an inode on the disk
641 */
Mingming Cao617ba132006-10-11 01:20:53 -0700642struct ext4_inode {
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700643 __le16 i_mode; /* File mode */
644 __le16 i_uid; /* Low 16 bits of Owner Uid */
Aneesh Kumar K.Va48380f2008-01-28 23:58:27 -0500645 __le32 i_size_lo; /* Size in bytes */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700646 __le32 i_atime; /* Access time */
Kalpak Shahef7f3832007-07-18 09:15:20 -0400647 __le32 i_ctime; /* Inode Change time */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700648 __le32 i_mtime; /* Modification time */
649 __le32 i_dtime; /* Deletion Time */
650 __le16 i_gid; /* Low 16 bits of Group Id */
651 __le16 i_links_count; /* Links count */
Aneesh Kumar K.V0fc1b452008-01-28 23:58:26 -0500652 __le32 i_blocks_lo; /* Blocks count */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700653 __le32 i_flags; /* File flags */
654 union {
655 struct {
Jean Noel Cordenner25ec56b2008-01-28 23:58:27 -0500656 __le32 l_i_version;
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700657 } linux1;
658 struct {
659 __u32 h_i_translator;
660 } hurd1;
661 struct {
662 __u32 m_i_reserved1;
663 } masix1;
664 } osd1; /* OS dependent 1 */
Mingming Cao617ba132006-10-11 01:20:53 -0700665 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700666 __le32 i_generation; /* File version (for NFS) */
Aneesh Kumar K.V7973c0c2008-01-28 23:58:27 -0500667 __le32 i_file_acl_lo; /* File ACL */
Aneesh Kumar K.Va48380f2008-01-28 23:58:27 -0500668 __le32 i_size_high;
Coly Lif077d0d2007-10-16 18:38:25 -0400669 __le32 i_obso_faddr; /* Obsoleted fragment address */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700670 union {
671 struct {
Aneesh Kumar K.V0fc1b452008-01-28 23:58:26 -0500672 __le16 l_i_blocks_high; /* were l_i_reserved1 */
Badari Pulavartya1ddeb72006-10-11 01:21:09 -0700673 __le16 l_i_file_acl_high;
Dave Kleikamp8c55e202007-05-24 13:04:54 -0400674 __le16 l_i_uid_high; /* these 2 fields */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700675 __le16 l_i_gid_high; /* were reserved2[0] */
Darrick J. Wonge6153912012-04-29 18:23:10 -0400676 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
677 __le16 l_i_reserved;
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700678 } linux2;
679 struct {
Coly Lif077d0d2007-10-16 18:38:25 -0400680 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700681 __u16 h_i_mode_high;
682 __u16 h_i_uid_high;
683 __u16 h_i_gid_high;
684 __u32 h_i_author;
685 } hurd2;
686 struct {
Coly Lif077d0d2007-10-16 18:38:25 -0400687 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
Badari Pulavartya1ddeb72006-10-11 01:21:09 -0700688 __le16 m_i_file_acl_high;
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700689 __u32 m_i_reserved2[2];
690 } masix2;
691 } osd2; /* OS dependent 2 */
692 __le16 i_extra_isize;
Darrick J. Wonge6153912012-04-29 18:23:10 -0400693 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
Kalpak Shahef7f3832007-07-18 09:15:20 -0400694 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
695 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
696 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
697 __le32 i_crtime; /* File Creation time */
698 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
Jean Noel Cordenner25ec56b2008-01-28 23:58:27 -0500699 __le32 i_version_hi; /* high 32 bits for 64-bit version */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700700};
701
Akira Fujita748de672009-06-17 19:24:03 -0400702struct move_extent {
703 __u32 reserved; /* should be zero */
704 __u32 donor_fd; /* donor file descriptor */
705 __u64 orig_start; /* logical start offset in block for orig */
706 __u64 donor_start; /* logical start offset in block for donor */
707 __u64 len; /* block length to be moved */
708 __u64 moved_len; /* moved block length */
709};
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700710
Kalpak Shahef7f3832007-07-18 09:15:20 -0400711#define EXT4_EPOCH_BITS 2
712#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
713#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
714
715/*
716 * Extended fields will fit into an inode if the filesystem was formatted
717 * with large inodes (-I 256 or larger) and there are not currently any EAs
718 * consuming all of the available space. For new inodes we always reserve
719 * enough space for the kernel's known extended fields, but for inodes
720 * created with an old kernel this might not have been the case. None of
721 * the extended inode fields is critical for correct filesystem operation.
722 * This macro checks if a certain field fits in the inode. Note that
723 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
724 */
725#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
726 ((offsetof(typeof(*ext4_inode), field) + \
727 sizeof((ext4_inode)->field)) \
728 <= (EXT4_GOOD_OLD_INODE_SIZE + \
729 (einode)->i_extra_isize)) \
730
731static inline __le32 ext4_encode_extra_time(struct timespec *time)
732{
733 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
Theodore Ts'oc1fccc02009-09-30 01:13:55 -0400734 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
735 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
Kalpak Shahef7f3832007-07-18 09:15:20 -0400736}
737
738static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
739{
740 if (sizeof(time->tv_sec) > 4)
741 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
742 << 32;
Theodore Ts'oc1fccc02009-09-30 01:13:55 -0400743 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
Kalpak Shahef7f3832007-07-18 09:15:20 -0400744}
745
746#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
747do { \
748 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
749 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
750 (raw_inode)->xtime ## _extra = \
751 ext4_encode_extra_time(&(inode)->xtime); \
752} while (0)
753
754#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
755do { \
756 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
757 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
758 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
759 (raw_inode)->xtime ## _extra = \
760 ext4_encode_extra_time(&(einode)->xtime); \
761} while (0)
762
763#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
764do { \
765 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
766 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
767 ext4_decode_extra_time(&(inode)->xtime, \
768 raw_inode->xtime ## _extra); \
Eric Sandeenaf0b44a2010-12-19 22:10:31 -0500769 else \
770 (inode)->xtime.tv_nsec = 0; \
Kalpak Shahef7f3832007-07-18 09:15:20 -0400771} while (0)
772
773#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
774do { \
775 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
776 (einode)->xtime.tv_sec = \
777 (signed)le32_to_cpu((raw_inode)->xtime); \
778 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
779 ext4_decode_extra_time(&(einode)->xtime, \
780 raw_inode->xtime ## _extra); \
Eric Sandeenaf0b44a2010-12-19 22:10:31 -0500781 else \
782 (einode)->xtime.tv_nsec = 0; \
Kalpak Shahef7f3832007-07-18 09:15:20 -0400783} while (0)
784
Jean Noel Cordenner25ec56b2008-01-28 23:58:27 -0500785#define i_disk_version osd1.linux1.l_i_version
786
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700787#if defined(__KERNEL__) || defined(__linux__)
788#define i_reserved1 osd1.linux1.l_i_reserved1
Badari Pulavartya1ddeb72006-10-11 01:21:09 -0700789#define i_file_acl_high osd2.linux2.l_i_file_acl_high
Aneesh Kumar K.V0fc1b452008-01-28 23:58:26 -0500790#define i_blocks_high osd2.linux2.l_i_blocks_high
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700791#define i_uid_low i_uid
792#define i_gid_low i_gid
793#define i_uid_high osd2.linux2.l_i_uid_high
794#define i_gid_high osd2.linux2.l_i_gid_high
Darrick J. Wonge6153912012-04-29 18:23:10 -0400795#define i_checksum_lo osd2.linux2.l_i_checksum_lo
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700796
797#elif defined(__GNU__)
798
799#define i_translator osd1.hurd1.h_i_translator
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700800#define i_uid_high osd2.hurd2.h_i_uid_high
801#define i_gid_high osd2.hurd2.h_i_gid_high
802#define i_author osd2.hurd2.h_i_author
803
804#elif defined(__masix__)
805
806#define i_reserved1 osd1.masix1.m_i_reserved1
Badari Pulavartya1ddeb72006-10-11 01:21:09 -0700807#define i_file_acl_high osd2.masix2.m_i_file_acl_high
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700808#define i_reserved2 osd2.masix2.m_i_reserved2
809
810#endif /* defined(__KERNEL__) || defined(__linux__) */
811
Zheng Liuc0677e62012-11-08 15:18:54 -0500812#include "extents_status.h"
813
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400814/*
815 * fourth extended file system inode data in memory
816 */
817struct ext4_inode_info {
818 __le32 i_data[15]; /* unconverted */
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400819 __u32 i_dtime;
Dmitry Monakhov12e9b892010-05-16 22:00:00 -0400820 ext4_fsblk_t i_file_acl;
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400821
822 /*
823 * i_block_group is the number of the block group which contains
824 * this file's inode. Constant across the lifetime of the inode,
825 * it is ued for making block allocation decisions - we try to
826 * place a file's data blocks near its inode block, and new inodes
827 * near to their parent directory's inode.
828 */
829 ext4_group_t i_block_group;
Theodore Ts'o8a2005d2011-01-10 12:13:42 -0500830 ext4_lblk_t i_dir_start_lookup;
Theodore Ts'o353eb832011-01-10 12:18:25 -0500831#if (BITS_PER_LONG < 64)
Theodore Ts'o19f5fb72010-01-24 14:34:07 -0500832 unsigned long i_state_flags; /* Dynamic state flags */
Theodore Ts'o353eb832011-01-10 12:18:25 -0500833#endif
Dmitry Monakhov12e9b892010-05-16 22:00:00 -0400834 unsigned long i_flags;
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400835
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400836 /*
837 * Extended attributes can be read independently of the main file
838 * data. Taking i_mutex even when reading would cause contention
839 * between readers of EAs and writers of regular file data, so
840 * instead we synchronize on xattr_sem when reading or changing
841 * EAs.
842 */
843 struct rw_semaphore xattr_sem;
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400844
845 struct list_head i_orphan; /* unlinked but open inodes */
846
847 /*
848 * i_disksize keeps track of what the inode size is ON DISK, not
849 * in memory. During truncate, i_size is set to the new size by
850 * the VFS prior to calling ext4_truncate(), but the filesystem won't
851 * set i_disksize to 0 until the truncate is actually under way.
852 *
853 * The intent is that i_disksize always represents the blocks which
854 * are used by this file. This allows recovery to restart truncate
855 * on orphans if we crash during truncate. We actually write i_disksize
856 * into the on-disk inode when writing inodes out, instead of i_size.
857 *
858 * The only time when i_disksize and i_size may be different is when
859 * a truncate is in progress. The only things which change i_disksize
860 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
861 */
862 loff_t i_disksize;
863
864 /*
865 * i_data_sem is for serialising ext4_truncate() against
866 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
867 * data tree are chopped off during truncate. We can't do that in
868 * ext4 because whenever we perform intermediate commits during
869 * truncate, the inode and all the metadata blocks *must* be in a
870 * consistent state which allows truncation of the orphans to restart
871 * during recovery. Hence we must fix the get_block-vs-truncate race
872 * by other means, so we have i_data_sem.
873 */
874 struct rw_semaphore i_data_sem;
875 struct inode vfs_inode;
Theodore Ts'o8aefcd52011-01-10 12:29:43 -0500876 struct jbd2_inode *jinode;
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400877
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400878 /*
879 * File creation time. Its function is same as that of
880 * struct timespec i_{a,c,m}time in the generic inode.
881 */
882 struct timespec i_crtime;
883
884 /* mballoc */
885 struct list_head i_prealloc_list;
886 spinlock_t i_prealloc_lock;
887
Zheng Liuc0677e62012-11-08 15:18:54 -0500888 /* extents status tree */
889 struct ext4_es_tree i_es_tree;
890 rwlock_t i_es_lock;
Zheng Liu74cd15c2013-02-18 00:32:55 -0500891 struct list_head i_es_lru;
892 unsigned int i_es_lru_nr; /* protected by i_es_lock */
Zheng Liuc0677e62012-11-08 15:18:54 -0500893
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400894 /* ialloc */
895 ext4_group_t i_last_alloc_group;
896
897 /* allocation reservation info for delalloc */
Theodore Ts'o57042652011-09-09 18:56:51 -0400898 /* In case of bigalloc, these refer to clusters rather than blocks */
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400899 unsigned int i_reserved_data_blocks;
900 unsigned int i_reserved_meta_blocks;
901 unsigned int i_allocated_meta_blocks;
Theodore Ts'o01f49d02011-01-10 12:13:03 -0500902 ext4_lblk_t i_da_metadata_calc_last_lblock;
Theodore Ts'o9d0be502010-01-01 02:41:30 -0500903 int i_da_metadata_calc_len;
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400904
905 /* on-disk additional length */
906 __u16 i_extra_isize;
907
Tao Ma67cf5b02012-12-10 14:04:46 -0500908 /* Indicate the inline data space. */
909 u16 i_inline_off;
910 u16 i_inline_size;
911
Dmitry Monakhova9e7f442009-12-14 15:21:14 +0300912#ifdef CONFIG_QUOTA
913 /* quota space reservation, managed internally by quota code */
914 qsize_t i_reserved_quota;
915#endif
Mingming Cao8d5d02e2009-09-28 15:48:29 -0400916
Jiaying Zhangc7064ef2010-03-02 13:28:44 -0500917 /* completed IOs that might need unwritten extents handling */
918 struct list_head i_completed_io_list;
Jiaying Zhang744692d2010-03-04 16:14:02 -0500919 spinlock_t i_completed_io_lock;
Theodore Ts'o8a2005d2011-01-10 12:13:42 -0500920 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
Dmitry Monakhove27f41e2012-09-28 23:24:52 -0400921 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
Jan Kara84c17542013-01-28 09:43:46 -0500922 struct work_struct i_unwritten_work; /* deferred extent conversion */
Theodore Ts'o8a2005d2011-01-10 12:13:42 -0500923
924 spinlock_t i_block_reservation_lock;
Jan Karab436b9b2009-12-08 23:51:10 -0500925
926 /*
927 * Transactions that contain inode's metadata needed to complete
928 * fsync and fdatasync, respectively.
929 */
930 tid_t i_sync_tid;
931 tid_t i_datasync_tid;
Darrick J. Wong814525f2012-04-29 18:31:10 -0400932
933 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
934 __u32 i_csum_seed;
Theodore Ts'od444c3c2009-05-01 13:44:33 -0400935};
936
937/*
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700938 * File system states
939 */
Mingming Cao617ba132006-10-11 01:20:53 -0700940#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
941#define EXT4_ERROR_FS 0x0002 /* Errors detected */
942#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700943
944/*
Theodore Tso469108f2008-02-10 01:11:44 -0500945 * Misc. filesystem flags
946 */
947#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
948#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
949#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
950
951/*
Lukas Czerner68911002013-04-03 22:12:52 -0400952 * Mount flags set via mount options or defaults
Dave Kleikampac27a0e2006-10-11 01:20:50 -0700953 */
Mingming Cao617ba132006-10-11 01:20:53 -0700954#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
955#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
956#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
957#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
958#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
Theodore Ts'o39ef17f2012-03-03 17:56:23 -0500959#define EXT4_MOUNT_ERRORS_MASK 0x00070
Mingming Cao617ba132006-10-11 01:20:53 -0700960#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
961#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
Mingming Cao617ba132006-10-11 01:20:53 -0700962#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
963#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
964#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
965#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
966#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
967#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
968#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
969#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
Theodore Ts'oafd46722009-03-16 23:12:23 -0400970#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
Mingming Cao617ba132006-10-11 01:20:53 -0700971#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
Mingming Cao617ba132006-10-11 01:20:53 -0700972#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
973#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
974#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
Jiaying Zhang744692d2010-03-04 16:14:02 -0500975#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
Linus Torvaldsd4da6c92009-11-02 10:15:27 -0800976#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
Girish Shilamkar818d2762008-01-28 23:58:27 -0500977#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
Alex Tomas64769242008-07-11 19:27:31 -0400978#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
Hidehiro Kawai5bf56832008-10-10 22:12:43 -0400979#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
Theodore Ts'o6fd058f2009-05-17 15:38:01 -0400980#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
Eric Sandeen5328e632009-11-19 14:25:42 -0500981#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
Lukas Czernerbfff6872010-10-27 21:30:05 -0400982#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
Hidehiro Kawai5bf56832008-10-10 22:12:43 -0400983
Lukas Czerner68911002013-04-03 22:12:52 -0400984/*
985 * Mount flags set either automatically (could not be set by mount option)
986 * based on per file system feature or property or in special cases such as
987 * distinguishing between explicit mount option definition and default.
988 */
Theodore Ts'o56889782011-09-03 18:22:38 -0400989#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
990 specified delalloc */
Lukas Czerner68911002013-04-03 22:12:52 -0400991#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
992 size of blocksize * 8
993 blocks */
Theodore Ts'o56889782011-09-03 18:22:38 -0400994
Theodore Ts'ofd8c37e2010-12-15 20:26:48 -0500995#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
996 ~EXT4_MOUNT_##opt
997#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
998 EXT4_MOUNT_##opt
Mingming Cao617ba132006-10-11 01:20:53 -0700999#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1000 EXT4_MOUNT_##opt)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001001
Theodore Ts'oa2595b82010-12-15 20:30:48 -05001002#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1003 ~EXT4_MOUNT2_##opt
1004#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1005 EXT4_MOUNT2_##opt
1006#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1007 EXT4_MOUNT2_##opt)
1008
Akinobu Mita597d5082011-12-28 20:32:07 -05001009#define ext4_test_and_set_bit __test_and_set_bit_le
1010#define ext4_set_bit __set_bit_le
Mingming Cao617ba132006-10-11 01:20:53 -07001011#define ext4_set_bit_atomic ext2_set_bit_atomic
Akinobu Mita597d5082011-12-28 20:32:07 -05001012#define ext4_test_and_clear_bit __test_and_clear_bit_le
1013#define ext4_clear_bit __clear_bit_le
Mingming Cao617ba132006-10-11 01:20:53 -07001014#define ext4_clear_bit_atomic ext2_clear_bit_atomic
Akinobu Mita50e01682011-03-23 16:42:07 -07001015#define ext4_test_bit test_bit_le
Akinobu Mita50e01682011-03-23 16:42:07 -07001016#define ext4_find_next_zero_bit find_next_zero_bit_le
1017#define ext4_find_next_bit find_next_bit_le
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001018
Yongqiang Yangc3e94d12011-07-26 22:05:53 -04001019extern void ext4_set_bits(void *bm, int cur, int len);
1020
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001021/*
1022 * Maximal mount counts between two filesystem checks
1023 */
Mingming Cao617ba132006-10-11 01:20:53 -07001024#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1025#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001026
1027/*
1028 * Behaviour when detecting errors
1029 */
Mingming Cao617ba132006-10-11 01:20:53 -07001030#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1031#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1032#define EXT4_ERRORS_PANIC 3 /* Panic */
1033#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001034
Darrick J. Wonge6153912012-04-29 18:23:10 -04001035/* Metadata checksum algorithm codes */
1036#define EXT4_CRC32C_CHKSUM 1
1037
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001038/*
1039 * Structure of the super block
1040 */
Mingming Cao617ba132006-10-11 01:20:53 -07001041struct ext4_super_block {
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001042/*00*/ __le32 s_inodes_count; /* Inodes count */
Aneesh Kumar K.V6bc9fef2007-10-16 18:38:25 -04001043 __le32 s_blocks_count_lo; /* Blocks count */
Aneesh Kumar K.V308ba3e2007-10-16 18:38:25 -04001044 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1045 __le32 s_free_blocks_count_lo; /* Free blocks count */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001046/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1047 __le32 s_first_data_block; /* First Data Block */
1048 __le32 s_log_block_size; /* Block size */
Theodore Ts'o281b5992011-09-09 18:34:51 -04001049 __le32 s_log_cluster_size; /* Allocation cluster size */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001050/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
Theodore Ts'o281b5992011-09-09 18:34:51 -04001051 __le32 s_clusters_per_group; /* # Clusters per group */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001052 __le32 s_inodes_per_group; /* # Inodes per group */
1053 __le32 s_mtime; /* Mount time */
1054/*30*/ __le32 s_wtime; /* Write time */
1055 __le16 s_mnt_count; /* Mount count */
1056 __le16 s_max_mnt_count; /* Maximal mount count */
1057 __le16 s_magic; /* Magic signature */
1058 __le16 s_state; /* File system state */
1059 __le16 s_errors; /* Behaviour when detecting errors */
1060 __le16 s_minor_rev_level; /* minor revision level */
1061/*40*/ __le32 s_lastcheck; /* time of last check */
1062 __le32 s_checkinterval; /* max. time between checks */
1063 __le32 s_creator_os; /* OS */
1064 __le32 s_rev_level; /* Revision level */
1065/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1066 __le16 s_def_resgid; /* Default gid for reserved blocks */
1067 /*
Mingming Cao617ba132006-10-11 01:20:53 -07001068 * These fields are for EXT4_DYNAMIC_REV superblocks only.
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001069 *
1070 * Note: the difference between the compatible feature set and
1071 * the incompatible feature set is that if there is a bit set
1072 * in the incompatible feature set that the kernel doesn't
1073 * know about, it should refuse to mount the filesystem.
1074 *
1075 * e2fsck's requirements are more strict; if it doesn't know
1076 * about a feature in either the compatible or incompatible
1077 * feature set, it must abort and not try to meddle with
1078 * things it doesn't understand...
1079 */
1080 __le32 s_first_ino; /* First non-reserved inode */
Alexandre Ratchov0d1ee422006-10-11 01:21:14 -07001081 __le16 s_inode_size; /* size of inode structure */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001082 __le16 s_block_group_nr; /* block group # of this superblock */
1083 __le32 s_feature_compat; /* compatible feature set */
1084/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1085 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1086/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1087/*78*/ char s_volume_name[16]; /* volume name */
1088/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1089/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1090 /*
1091 * Performance hints. Directory preallocation should only
Mingming Cao617ba132006-10-11 01:20:53 -07001092 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001093 */
1094 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1095 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1096 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1097 /*
Mingming Cao617ba132006-10-11 01:20:53 -07001098 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001099 */
1100/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1101/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1102 __le32 s_journal_dev; /* device number of journal file */
1103 __le32 s_last_orphan; /* start of list of inodes to delete */
1104 __le32 s_hash_seed[4]; /* HTREE hash seed */
1105 __u8 s_def_hash_version; /* Default hash version to use */
Theodore Ts'o89eeddf2010-07-27 11:56:04 -04001106 __u8 s_jnl_backup_type;
Alexandre Ratchov0d1ee422006-10-11 01:21:14 -07001107 __le16 s_desc_size; /* size of group descriptor */
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07001108/*100*/ __le32 s_default_mount_opts;
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001109 __le32 s_first_meta_bg; /* First metablock block group */
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07001110 __le32 s_mkfs_time; /* When the filesystem was created */
1111 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1112 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1113/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1114 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1115 __le32 s_free_blocks_count_hi; /* Free blocks count */
Aneesh Kumar K.Vc1bddad2007-10-16 18:38:25 -04001116 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1117 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1118 __le32 s_flags; /* Miscellaneous flags */
1119 __le16 s_raid_stride; /* RAID stride */
Johann Lombardic5e06d12011-05-24 18:31:25 -04001120 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
Aneesh Kumar K.Vc1bddad2007-10-16 18:38:25 -04001121 __le64 s_mmp_block; /* Block for multi-mount protection */
1122 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
Jose R. Santos772cb7c2008-07-11 19:27:31 -04001123 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
Darrick J. Wonge6153912012-04-29 18:23:10 -04001124 __u8 s_checksum_type; /* metadata checksum algorithm used */
Jose R. Santos772cb7c2008-07-11 19:27:31 -04001125 __le16 s_reserved_pad;
Theodore Ts'oafc32f72009-02-28 19:39:58 -05001126 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
Theodore Ts'o4a9cdec2010-06-29 11:00:23 -04001127 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1128 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1129 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1130 snapshot's future use */
1131 __le32 s_snapshot_list; /* inode number of the head of the
1132 on-disk snapshot list */
Theodore Ts'o1c13d5c2010-07-27 11:56:03 -04001133#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1134 __le32 s_error_count; /* number of fs errors */
1135 __le32 s_first_error_time; /* first time an error happened */
1136 __le32 s_first_error_ino; /* inode involved in first error */
1137 __le64 s_first_error_block; /* block involved of first error */
1138 __u8 s_first_error_func[32]; /* function where the error happened */
1139 __le32 s_first_error_line; /* line number where error happened */
1140 __le32 s_last_error_time; /* most recent time of an error */
1141 __le32 s_last_error_ino; /* inode involved in last error */
1142 __le32 s_last_error_line; /* line number where error happened */
1143 __le64 s_last_error_block; /* block involved of last error */
1144 __u8 s_last_error_func[32]; /* function where the error happened */
Theodore Ts'o8b67f042010-08-01 23:14:20 -04001145#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1146 __u8 s_mount_opts[64];
Theodore Ts'o281b5992011-09-09 18:34:51 -04001147 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1148 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1149 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
Darrick J. Wonge6153912012-04-29 18:23:10 -04001150 __le32 s_reserved[108]; /* Padding to the end of the block */
1151 __le32 s_checksum; /* crc32c(superblock) */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001152};
1153
Theodore Ts'o1c13d5c2010-07-27 11:56:03 -04001154#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1155
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001156#ifdef __KERNEL__
Theodore Ts'obc0b0d62009-06-13 10:09:48 -04001157
1158/*
Theodore Ts'o4ab2f152009-06-13 10:09:36 -04001159 * run-time mount flags
Theodore Ts'obc0b0d62009-06-13 10:09:48 -04001160 */
1161#define EXT4_MF_MNTDIR_SAMPLED 0x0001
Theodore Ts'o4ab2f152009-06-13 10:09:36 -04001162#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
Theodore Ts'obc0b0d62009-06-13 10:09:48 -04001163
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001164/*
1165 * fourth extended-fs super-block data in memory
1166 */
1167struct ext4_sb_info {
1168 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1169 unsigned long s_inodes_per_block;/* Number of inodes per block */
1170 unsigned long s_blocks_per_group;/* Number of blocks in a group */
Theodore Ts'o281b5992011-09-09 18:34:51 -04001171 unsigned long s_clusters_per_group; /* Number of clusters in a group */
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001172 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1173 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1174 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1175 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1176 ext4_group_t s_groups_count; /* Number of groups in the fs */
Eric Sandeenfb0a3872009-09-16 14:45:10 -04001177 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
Theodore Ts'o952fc182012-07-09 16:27:05 -04001178 unsigned long s_overhead; /* # of fs overhead clusters */
Theodore Ts'o281b5992011-09-09 18:34:51 -04001179 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1180 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001181 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1182 struct buffer_head * s_sbh; /* Buffer containing the super block */
1183 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1184 struct buffer_head **s_group_desc;
Theodore Ts'o7f4520c2009-06-13 10:09:41 -04001185 unsigned int s_mount_opt;
Theodore Ts'oa2595b82010-12-15 20:30:48 -05001186 unsigned int s_mount_opt2;
Theodore Ts'obc0b0d62009-06-13 10:09:48 -04001187 unsigned int s_mount_flags;
Theodore Ts'o5a916be2012-03-04 19:27:31 -05001188 unsigned int s_def_mount_opt;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001189 ext4_fsblk_t s_sb_block;
Eric W. Biederman08cefc72012-02-07 15:41:49 -08001190 kuid_t s_resuid;
1191 kgid_t s_resgid;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001192 unsigned short s_mount_state;
1193 unsigned short s_pad;
1194 int s_addr_per_block_bits;
1195 int s_desc_per_block_bits;
1196 int s_inode_size;
1197 int s_first_ino;
1198 unsigned int s_inode_readahead_blks;
Andreas Dilger11013912009-06-13 11:45:35 -04001199 unsigned int s_inode_goal;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001200 spinlock_t s_next_gen_lock;
1201 u32 s_next_generation;
1202 u32 s_hash_seed[4];
1203 int s_def_hash_version;
1204 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
Theodore Ts'o57042652011-09-09 18:56:51 -04001205 struct percpu_counter s_freeclusters_counter;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001206 struct percpu_counter s_freeinodes_counter;
1207 struct percpu_counter s_dirs_counter;
Theodore Ts'o57042652011-09-09 18:56:51 -04001208 struct percpu_counter s_dirtyclusters_counter;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001209 struct blockgroup_lock *s_blockgroup_lock;
1210 struct proc_dir_entry *s_proc;
1211 struct kobject s_kobj;
1212 struct completion s_kobj_unregister;
Theodore Ts'o2c0544b2012-05-30 22:56:46 -04001213 struct super_block *s_sb;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001214
1215 /* Journaling */
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001216 struct journal_s *s_journal;
1217 struct list_head s_orphan;
1218 struct mutex s_orphan_lock;
Yongqiang Yang8f82f842011-07-26 21:35:44 -04001219 unsigned long s_resize_flags; /* Flags indicating if there
1220 is a resizer */
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001221 unsigned long s_commit_interval;
1222 u32 s_max_batch_time;
1223 u32 s_min_batch_time;
1224 struct block_device *journal_bdev;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001225#ifdef CONFIG_QUOTA
1226 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1227 int s_jquota_fmt; /* Format of quota to use */
1228#endif
1229 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
Theodore Ts'o6fd058f2009-05-17 15:38:01 -04001230 struct rb_root system_blks;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001231
1232#ifdef EXTENTS_STATS
1233 /* ext4 extents stats */
1234 unsigned long s_ext_min;
1235 unsigned long s_ext_max;
1236 unsigned long s_depth_max;
1237 spinlock_t s_ext_stats_lock;
1238 unsigned long s_ext_blocks;
1239 unsigned long s_ext_extents;
1240#endif
1241
1242 /* for buddy allocator */
1243 struct ext4_group_info ***s_group_info;
1244 struct inode *s_buddy_cache;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001245 spinlock_t s_md_lock;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001246 unsigned short *s_mb_offsets;
1247 unsigned int *s_mb_maxs;
Theodore Ts'o28623c22012-09-05 01:31:50 -04001248 unsigned int s_group_info_size;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001249
1250 /* tunables */
1251 unsigned long s_stripe;
1252 unsigned int s_mb_stream_request;
1253 unsigned int s_mb_max_to_scan;
1254 unsigned int s_mb_min_to_scan;
1255 unsigned int s_mb_stats;
1256 unsigned int s_mb_order2_reqs;
1257 unsigned int s_mb_group_prealloc;
Theodore Ts'o55138e02009-09-29 13:31:31 -04001258 unsigned int s_max_writeback_mb_bump;
Theodore Ts'odf981d02012-08-17 09:48:17 -04001259 unsigned int s_max_dir_size_kb;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001260 /* where last allocation was done - for stream allocation */
1261 unsigned long s_mb_last_group;
1262 unsigned long s_mb_last_start;
1263
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001264 /* stats for buddy allocator */
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001265 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1266 atomic_t s_bal_success; /* we found long enough chunks */
1267 atomic_t s_bal_allocated; /* in blocks */
1268 atomic_t s_bal_ex_scanned; /* total extents scanned */
1269 atomic_t s_bal_goals; /* goal hits */
1270 atomic_t s_bal_breaks; /* too long searches */
1271 atomic_t s_bal_2orders; /* 2^order hits */
1272 spinlock_t s_bal_lock;
1273 unsigned long s_mb_buddies_generated;
1274 unsigned long long s_mb_generation_time;
1275 atomic_t s_mb_lost_chunks;
1276 atomic_t s_mb_preallocated;
1277 atomic_t s_mb_discarded;
Theodore Ts'o50797482009-09-18 13:34:02 -04001278 atomic_t s_lock_busy;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001279
1280 /* locality groups */
Tejun Heo003cb602010-02-02 14:39:01 +09001281 struct ext4_locality_group __percpu *s_locality_groups;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001282
1283 /* for write statistics */
1284 unsigned long s_sectors_written_start;
1285 u64 s_kbytes_written;
1286
Zheng Liu67a5da52012-08-17 09:54:17 -04001287 /* the size of zero-out chunk */
1288 unsigned int s_extent_max_zeroout_kb;
1289
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001290 unsigned int s_log_groups_per_flex;
1291 struct flex_groups *s_flex_groups;
Theodore Ts'o117fff12012-09-05 01:29:50 -04001292 ext4_group_t s_flex_groups_allocated;
Mingming Cao4c0425f2009-09-28 15:48:41 -04001293
1294 /* workqueue for dio unwritten */
1295 struct workqueue_struct *dio_unwritten_wq;
Theodore Ts'o66e61a92010-07-27 11:56:04 -04001296
1297 /* timer for periodic error stats printing */
1298 struct timer_list s_err_report;
Lukas Czernerbfff6872010-10-27 21:30:05 -04001299
1300 /* Lazy inode table initialization info */
1301 struct ext4_li_request *s_li_request;
1302 /* Wait multiplier for lazy initialization thread */
1303 unsigned int s_li_wait_mult;
Johann Lombardic5e06d12011-05-24 18:31:25 -04001304
1305 /* Kernel thread for multiple mount protection */
1306 struct task_struct *s_mmp_tsk;
Tao Ma3d56b8d2011-07-11 00:03:38 -04001307
1308 /* record the last minlen when FITRIM is called. */
1309 atomic_t s_last_trim_minblks;
Darrick J. Wong04419842012-04-29 18:27:10 -04001310
1311 /* Reference to checksum algorithm driver via cryptoapi */
1312 struct crypto_shash *s_chksum_driver;
Darrick J. Wonga9c473172012-04-29 18:29:10 -04001313
1314 /* Precomputed FS UUID checksum for seeding other checksums */
1315 __u32 s_csum_seed;
Zheng Liu74cd15c2013-02-18 00:32:55 -05001316
1317 /* Reclaim extents from extent status tree */
1318 struct shrinker s_es_shrinker;
1319 struct list_head s_es_lru;
Theodore Ts'o1ac64662013-03-02 10:27:46 -05001320 struct percpu_counter s_extent_cache_cnt;
Zheng Liu74cd15c2013-02-18 00:32:55 -05001321 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
Theodore Ts'oca0faba2009-05-03 16:33:44 -04001322};
1323
Theodore Ts'oaf5bc922008-09-08 22:25:24 -04001324static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001325{
1326 return sb->s_fs_info;
1327}
Mingming Cao617ba132006-10-11 01:20:53 -07001328static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001329{
Mingming Cao617ba132006-10-11 01:20:53 -07001330 return container_of(inode, struct ext4_inode_info, vfs_inode);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001331}
1332
Kalpak Shahef7f3832007-07-18 09:15:20 -04001333static inline struct timespec ext4_current_time(struct inode *inode)
1334{
1335 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1336 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1337}
1338
Mingming Cao617ba132006-10-11 01:20:53 -07001339static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001340{
Mingming Cao617ba132006-10-11 01:20:53 -07001341 return ino == EXT4_ROOT_INO ||
Aditya Kali7c319d32012-07-22 20:21:31 -04001342 ino == EXT4_USR_QUOTA_INO ||
1343 ino == EXT4_GRP_QUOTA_INO ||
Mingming Cao617ba132006-10-11 01:20:53 -07001344 ino == EXT4_JOURNAL_INO ||
1345 ino == EXT4_RESIZE_INO ||
1346 (ino >= EXT4_FIRST_INO(sb) &&
1347 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001348}
Theodore Ts'o19f5fb72010-01-24 14:34:07 -05001349
Tao Ma0edeb712011-10-31 17:30:44 -04001350static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1351 struct ext4_io_end *io_end)
1352{
1353 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1354 io_end->flag |= EXT4_IO_END_UNWRITTEN;
Dmitry Monakhove27f41e2012-09-28 23:24:52 -04001355 atomic_inc(&EXT4_I(inode)->i_unwritten);
Tao Ma0edeb712011-10-31 17:30:44 -04001356 }
1357}
1358
Dmitry Monakhovf45ee3a2012-09-28 23:21:09 -04001359static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1360{
1361 return inode->i_private;
1362}
1363
1364static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1365{
1366 inode->i_private = io;
1367}
1368
Theodore Ts'o19f5fb72010-01-24 14:34:07 -05001369/*
1370 * Inode dynamic state flags
1371 */
1372enum {
1373 EXT4_STATE_JDATA, /* journaled data exists */
1374 EXT4_STATE_NEW, /* inode is newly created */
1375 EXT4_STATE_XATTR, /* has in-inode xattrs */
1376 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1377 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1378 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1379 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
Frank Mayhar14ece102010-05-17 08:00:00 -04001380 EXT4_STATE_NEWENTRY, /* File just added to dir */
Theodore Ts'of2321092011-01-10 12:12:36 -05001381 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
Dmitry Monakhov17335dc2012-09-29 00:41:21 -04001382 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1383 nolocking */
Tao Ma67cf5b02012-12-10 14:04:46 -05001384 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
Theodore Ts'o74d553a2013-04-03 12:39:17 -04001385 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
Theodore Ts'o19f5fb72010-01-24 14:34:07 -05001386};
1387
Theodore Ts'o353eb832011-01-10 12:18:25 -05001388#define EXT4_INODE_BIT_FNS(name, field, offset) \
Dmitry Monakhov12e9b892010-05-16 22:00:00 -04001389static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1390{ \
Theodore Ts'o353eb832011-01-10 12:18:25 -05001391 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
Dmitry Monakhov12e9b892010-05-16 22:00:00 -04001392} \
1393static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1394{ \
Theodore Ts'o353eb832011-01-10 12:18:25 -05001395 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
Dmitry Monakhov12e9b892010-05-16 22:00:00 -04001396} \
1397static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1398{ \
Theodore Ts'o353eb832011-01-10 12:18:25 -05001399 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
Theodore Ts'o19f5fb72010-01-24 14:34:07 -05001400}
1401
Theodore Ts'o353eb832011-01-10 12:18:25 -05001402EXT4_INODE_BIT_FNS(flag, flags, 0)
1403#if (BITS_PER_LONG < 64)
1404EXT4_INODE_BIT_FNS(state, state_flags, 0)
1405
1406static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1407{
1408 (ei)->i_state_flags = 0;
1409}
1410#else
1411EXT4_INODE_BIT_FNS(state, flags, 32)
1412
1413static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1414{
1415 /* We depend on the fact that callers will set i_flags */
1416}
1417#endif
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001418#else
Mingming Cao617ba132006-10-11 01:20:53 -07001419/* Assume that user mode programs are passing in an ext4fs superblock, not
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001420 * a kernel struct super_block. This will allow us to call the feature-test
1421 * macros from user land. */
Mingming Cao617ba132006-10-11 01:20:53 -07001422#define EXT4_SB(sb) (sb)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001423#endif
1424
Mingming Cao617ba132006-10-11 01:20:53 -07001425#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001426
1427/*
1428 * Codes for operating systems
1429 */
Mingming Cao617ba132006-10-11 01:20:53 -07001430#define EXT4_OS_LINUX 0
1431#define EXT4_OS_HURD 1
1432#define EXT4_OS_MASIX 2
1433#define EXT4_OS_FREEBSD 3
1434#define EXT4_OS_LITES 4
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001435
1436/*
1437 * Revision levels
1438 */
Mingming Cao617ba132006-10-11 01:20:53 -07001439#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1440#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001441
Mingming Cao617ba132006-10-11 01:20:53 -07001442#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1443#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001444
Mingming Cao617ba132006-10-11 01:20:53 -07001445#define EXT4_GOOD_OLD_INODE_SIZE 128
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001446
1447/*
1448 * Feature set definitions
1449 */
1450
Mingming Cao617ba132006-10-11 01:20:53 -07001451#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
Aneesh Kumar K.V3a06d772008-11-22 15:04:59 -05001452 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
Mingming Cao617ba132006-10-11 01:20:53 -07001453#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
Aneesh Kumar K.V3a06d772008-11-22 15:04:59 -05001454 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
Mingming Cao617ba132006-10-11 01:20:53 -07001455#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
Aneesh Kumar K.V3a06d772008-11-22 15:04:59 -05001456 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
Mingming Cao617ba132006-10-11 01:20:53 -07001457#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1458 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1459#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1460 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1461#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1462 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1463#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1464 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1465#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1466 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1467#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1468 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001469
Mingming Cao617ba132006-10-11 01:20:53 -07001470#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1471#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1472#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1473#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1474#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1475#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001476
Mingming Cao617ba132006-10-11 01:20:53 -07001477#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1478#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1479#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
Aneesh Kumar K.V0fc1b452008-01-28 23:58:26 -05001480#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
Andreas Dilger717d50e2007-10-16 18:38:25 -04001481#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
Andreas Dilgerf8628a12007-07-18 08:38:01 -04001482#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
Kalpak Shahef7f3832007-07-18 09:15:20 -04001483#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
Aditya Kaliae812302011-05-24 19:00:39 -04001484#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
Theodore Ts'o281b5992011-09-09 18:34:51 -04001485#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
Darrick J. Wonge6153912012-04-29 18:23:10 -04001486/*
1487 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1488 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1489 * all other data structures' checksums. However, the METADATA_CSUM and
1490 * GDT_CSUM bits are mutually exclusive.
1491 */
Theodore Ts'o9b90e5e2012-01-04 22:01:53 -05001492#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001493
Mingming Cao617ba132006-10-11 01:20:53 -07001494#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1495#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1496#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1497#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1498#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
Alex Tomasa86c6182006-10-11 01:21:03 -07001499#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07001500#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
Aneesh Kumar K.V0fc1b452008-01-28 23:58:26 -05001501#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
Jose R. Santosce421582007-10-16 18:38:25 -04001502#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
Theodore Ts'of710b4b2010-01-25 03:31:32 -05001503#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1504#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
Theodore Ts'o856cbcf2012-02-20 17:53:01 -05001505#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
Theodore Ts'o9b90e5e2012-01-04 22:01:53 -05001506#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
Tao Ma67cf5b02012-12-10 14:04:46 -05001507#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001508
Theodore Ts'o2035e772011-04-18 17:29:14 -04001509#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1510#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1511 EXT4_FEATURE_INCOMPAT_META_BG)
1512#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1513 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1514 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1515
1516#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1517#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1518 EXT4_FEATURE_INCOMPAT_RECOVER| \
1519 EXT4_FEATURE_INCOMPAT_META_BG)
1520#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1521 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1522 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1523
Mingming Cao617ba132006-10-11 01:20:53 -07001524#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1525#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1526 EXT4_FEATURE_INCOMPAT_RECOVER| \
Alex Tomasa86c6182006-10-11 01:21:03 -07001527 EXT4_FEATURE_INCOMPAT_META_BG| \
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07001528 EXT4_FEATURE_INCOMPAT_EXTENTS| \
Jose R. Santosce421582007-10-16 18:38:25 -04001529 EXT4_FEATURE_INCOMPAT_64BIT| \
Johann Lombardic5e06d12011-05-24 18:31:25 -04001530 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
Tao Maf08225d2012-12-10 14:06:03 -05001531 EXT4_FEATURE_INCOMPAT_MMP | \
1532 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
Mingming Cao617ba132006-10-11 01:20:53 -07001533#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1534 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
Andreas Dilger717d50e2007-10-16 18:38:25 -04001535 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
Andreas Dilgerf8628a12007-07-18 08:38:01 -04001536 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
Kalpak Shahef7f3832007-07-18 09:15:20 -04001537 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
Aneesh Kumar K.V0fc1b452008-01-28 23:58:26 -05001538 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
Theodore Ts'o6f16b602011-09-09 19:06:51 -04001539 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
Darrick J. Wonge93376c2012-05-27 08:12:42 -04001540 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
Aditya Kali7c319d32012-07-22 20:21:31 -04001541 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1542 EXT4_FEATURE_RO_COMPAT_QUOTA)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001543
1544/*
1545 * Default values for user and/or group using reserved blocks
1546 */
Mingming Cao617ba132006-10-11 01:20:53 -07001547#define EXT4_DEF_RESUID 0
1548#define EXT4_DEF_RESGID 0
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001549
Theodore Ts'o240799c2008-10-09 23:53:47 -04001550#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1551
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001552/*
1553 * Default mount options
1554 */
Mingming Cao617ba132006-10-11 01:20:53 -07001555#define EXT4_DEFM_DEBUG 0x0001
1556#define EXT4_DEFM_BSDGROUPS 0x0002
1557#define EXT4_DEFM_XATTR_USER 0x0004
1558#define EXT4_DEFM_ACL 0x0008
1559#define EXT4_DEFM_UID16 0x0010
1560#define EXT4_DEFM_JMODE 0x0060
1561#define EXT4_DEFM_JMODE_DATA 0x0020
1562#define EXT4_DEFM_JMODE_ORDERED 0x0040
1563#define EXT4_DEFM_JMODE_WBACK 0x0060
Theodore Ts'o8b67f042010-08-01 23:14:20 -04001564#define EXT4_DEFM_NOBARRIER 0x0100
1565#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1566#define EXT4_DEFM_DISCARD 0x0400
1567#define EXT4_DEFM_NODELALLOC 0x0800
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001568
1569/*
Theodore Ts'o30773842009-01-03 20:27:38 -05001570 * Default journal batch times
1571 */
1572#define EXT4_DEF_MIN_BATCH_TIME 0
1573#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1574
1575/*
Theodore Ts'oa4912122009-03-12 12:18:34 -04001576 * Minimum number of groups in a flexgroup before we separate out
1577 * directories into the first block group of a flexgroup
1578 */
1579#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1580
1581/*
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001582 * Structure of a directory entry
1583 */
Mingming Cao617ba132006-10-11 01:20:53 -07001584#define EXT4_NAME_LEN 255
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001585
Mingming Cao617ba132006-10-11 01:20:53 -07001586struct ext4_dir_entry {
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001587 __le32 inode; /* Inode number */
1588 __le16 rec_len; /* Directory entry length */
1589 __le16 name_len; /* Name length */
Mingming Cao617ba132006-10-11 01:20:53 -07001590 char name[EXT4_NAME_LEN]; /* File name */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001591};
1592
1593/*
Mingming Cao617ba132006-10-11 01:20:53 -07001594 * The new version of the directory entry. Since EXT4 structures are
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001595 * stored in intel byte order, and the name_len field could never be
1596 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1597 * file_type field.
1598 */
Mingming Cao617ba132006-10-11 01:20:53 -07001599struct ext4_dir_entry_2 {
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001600 __le32 inode; /* Inode number */
1601 __le16 rec_len; /* Directory entry length */
1602 __u8 name_len; /* Name length */
1603 __u8 file_type;
Mingming Cao617ba132006-10-11 01:20:53 -07001604 char name[EXT4_NAME_LEN]; /* File name */
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001605};
1606
1607/*
Darrick J. Wonge6153912012-04-29 18:23:10 -04001608 * This is a bogus directory entry at the end of each leaf block that
1609 * records checksums.
1610 */
1611struct ext4_dir_entry_tail {
1612 __le32 det_reserved_zero1; /* Pretend to be unused */
1613 __le16 det_rec_len; /* 12 */
1614 __u8 det_reserved_zero2; /* Zero name length */
1615 __u8 det_reserved_ft; /* 0xDE, fake file type */
1616 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1617};
1618
Tao Ma3c47d542012-12-10 14:05:59 -05001619#define EXT4_DIRENT_TAIL(block, blocksize) \
1620 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1621 ((blocksize) - \
1622 sizeof(struct ext4_dir_entry_tail))))
1623
Darrick J. Wonge6153912012-04-29 18:23:10 -04001624/*
Mingming Cao617ba132006-10-11 01:20:53 -07001625 * Ext4 directory file types. Only the low 3 bits are used. The
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001626 * other bits are reserved for now.
1627 */
Mingming Cao617ba132006-10-11 01:20:53 -07001628#define EXT4_FT_UNKNOWN 0
1629#define EXT4_FT_REG_FILE 1
1630#define EXT4_FT_DIR 2
1631#define EXT4_FT_CHRDEV 3
1632#define EXT4_FT_BLKDEV 4
1633#define EXT4_FT_FIFO 5
1634#define EXT4_FT_SOCK 6
1635#define EXT4_FT_SYMLINK 7
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001636
Mingming Cao617ba132006-10-11 01:20:53 -07001637#define EXT4_FT_MAX 8
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001638
Darrick J. Wonge6153912012-04-29 18:23:10 -04001639#define EXT4_FT_DIR_CSUM 0xDE
1640
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001641/*
Mingming Cao617ba132006-10-11 01:20:53 -07001642 * EXT4_DIR_PAD defines the directory entries boundaries
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001643 *
1644 * NOTE: It must be a multiple of 4
1645 */
Mingming Cao617ba132006-10-11 01:20:53 -07001646#define EXT4_DIR_PAD 4
1647#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1648#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1649 ~EXT4_DIR_ROUND)
Jan Karaa72d7f82008-01-28 23:58:27 -05001650#define EXT4_MAX_REC_LEN ((1<<16)-1)
1651
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001652/*
Eric Sandeen0cfc9252010-08-05 01:46:37 -04001653 * If we ever get support for fs block sizes > page_size, we'll need
1654 * to remove the #if statements in the next two functions...
1655 */
1656static inline unsigned int
1657ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1658{
1659 unsigned len = le16_to_cpu(dlen);
1660
1661#if (PAGE_CACHE_SIZE >= 65536)
1662 if (len == EXT4_MAX_REC_LEN || len == 0)
1663 return blocksize;
1664 return (len & 65532) | ((len & 3) << 16);
1665#else
1666 return len;
1667#endif
1668}
1669
1670static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1671{
1672 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1673 BUG();
1674#if (PAGE_CACHE_SIZE >= 65536)
1675 if (len < 65536)
1676 return cpu_to_le16(len);
1677 if (len == blocksize) {
1678 if (blocksize == 65536)
1679 return cpu_to_le16(EXT4_MAX_REC_LEN);
1680 else
1681 return cpu_to_le16(0);
1682 }
1683 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1684#else
1685 return cpu_to_le16(len);
1686#endif
1687}
1688
1689/*
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001690 * Hash Tree Directory indexing
1691 * (c) Daniel Phillips, 2001
1692 */
1693
Eric Sandeen4074fe32007-10-16 18:38:25 -04001694#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1695 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
Dmitry Monakhov12e9b892010-05-16 22:00:00 -04001696 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
Mingming Cao617ba132006-10-11 01:20:53 -07001697#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1698#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001699
1700/* Legal values for the dx_root hash_version field: */
1701
1702#define DX_HASH_LEGACY 0
1703#define DX_HASH_HALF_MD4 1
1704#define DX_HASH_TEA 2
Theodore Ts'of99b2582008-10-28 13:21:44 -04001705#define DX_HASH_LEGACY_UNSIGNED 3
1706#define DX_HASH_HALF_MD4_UNSIGNED 4
1707#define DX_HASH_TEA_UNSIGNED 5
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001708
Darrick J. Wong04419842012-04-29 18:27:10 -04001709static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1710 const void *address, unsigned int length)
1711{
1712 struct {
1713 struct shash_desc shash;
Theodore Ts'o3108b542012-07-22 20:25:31 -04001714 char ctx[4];
Darrick J. Wong04419842012-04-29 18:27:10 -04001715 } desc;
1716 int err;
1717
Theodore Ts'o3108b542012-07-22 20:25:31 -04001718 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1719
Darrick J. Wong04419842012-04-29 18:27:10 -04001720 desc.shash.tfm = sbi->s_chksum_driver;
1721 desc.shash.flags = 0;
1722 *(u32 *)desc.ctx = crc;
1723
1724 err = crypto_shash_update(&desc.shash, address, length);
1725 BUG_ON(err);
1726
1727 return *(u32 *)desc.ctx;
1728}
1729
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001730#ifdef __KERNEL__
1731
1732/* hash info structure used by the directory hash */
1733struct dx_hash_info
1734{
1735 u32 hash;
1736 u32 minor_hash;
1737 int hash_version;
1738 u32 *seed;
1739};
1740
Fan Yongd1f52732012-03-18 22:44:40 -04001741
1742/* 32 and 64 bit signed EOF for dx directories */
1743#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1744#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1745
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001746
1747/*
Mingming Cao617ba132006-10-11 01:20:53 -07001748 * Control parameters used by ext4_htree_next_block
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001749 */
1750#define HASH_NB_ALWAYS 1
1751
1752
1753/*
1754 * Describe an inode's exact location on disk and in memory
1755 */
Mingming Cao617ba132006-10-11 01:20:53 -07001756struct ext4_iloc
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001757{
1758 struct buffer_head *bh;
1759 unsigned long offset;
Avantika Mathurfd2d4292008-01-28 23:58:27 -05001760 ext4_group_t block_group;
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001761};
1762
Mingming Cao617ba132006-10-11 01:20:53 -07001763static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001764{
Mingming Cao617ba132006-10-11 01:20:53 -07001765 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001766}
1767
1768/*
1769 * This structure is stuffed into the struct file's private_data field
1770 * for directories. It is where we put information so that we can do
1771 * readdir operations in hash tree order.
1772 */
1773struct dir_private_info {
1774 struct rb_root root;
1775 struct rb_node *curr_node;
1776 struct fname *extra_fname;
1777 loff_t last_pos;
1778 __u32 curr_hash;
1779 __u32 curr_minor_hash;
1780 __u32 next_hash;
1781};
1782
1783/* calculate the first block number of the group */
Mingming Cao617ba132006-10-11 01:20:53 -07001784static inline ext4_fsblk_t
Avantika Mathurfd2d4292008-01-28 23:58:27 -05001785ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001786{
Mingming Cao617ba132006-10-11 01:20:53 -07001787 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1788 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001789}
1790
1791/*
1792 * Special error return code only used by dx_probe() and its callers.
1793 */
1794#define ERR_BAD_DX_DIR -75000
1795
Mingming Cao3a5b2ec2006-10-11 01:21:05 -07001796/*
Lukas Czernerbfff6872010-10-27 21:30:05 -04001797 * Timeout and state flag for lazy initialization inode thread.
1798 */
1799#define EXT4_DEF_LI_WAIT_MULT 10
1800#define EXT4_DEF_LI_MAX_START_DELAY 5
1801#define EXT4_LAZYINIT_QUIT 0x0001
1802#define EXT4_LAZYINIT_RUNNING 0x0002
1803
1804/*
1805 * Lazy inode table initialization info
1806 */
1807struct ext4_lazy_init {
1808 unsigned long li_state;
Lukas Czernerbfff6872010-10-27 21:30:05 -04001809 struct list_head li_request_list;
1810 struct mutex li_list_mtx;
1811};
1812
1813struct ext4_li_request {
1814 struct super_block *lr_super;
1815 struct ext4_sb_info *lr_sbi;
1816 ext4_group_t lr_next_group;
1817 struct list_head lr_request;
1818 unsigned long lr_next_sched;
1819 unsigned long lr_timeout;
1820};
1821
Lukas Czerner857ac882010-10-27 21:30:05 -04001822struct ext4_features {
1823 struct kobject f_kobj;
1824 struct completion f_kobj_unregister;
1825};
1826
Lukas Czernerbfff6872010-10-27 21:30:05 -04001827/*
Johann Lombardic5e06d12011-05-24 18:31:25 -04001828 * This structure will be used for multiple mount protection. It will be
1829 * written into the block number saved in the s_mmp_block field in the
1830 * superblock. Programs that check MMP should assume that if
1831 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1832 * to use the filesystem, regardless of how old the timestamp is.
1833 */
1834#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1835#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1836#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1837#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1838
1839struct mmp_struct {
1840 __le32 mmp_magic; /* Magic number for MMP */
1841 __le32 mmp_seq; /* Sequence no. updated periodically */
1842
1843 /*
1844 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1845 * purposes and do not affect the correctness of the algorithm
1846 */
1847 __le64 mmp_time; /* Time last updated */
1848 char mmp_nodename[64]; /* Node which last updated MMP block */
1849 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1850
1851 /*
1852 * mmp_check_interval is used to verify if the MMP block has been
1853 * updated on the block device. The value is updated based on the
1854 * maximum time to write the MMP block during an update cycle.
1855 */
1856 __le16 mmp_check_interval;
1857
1858 __le16 mmp_pad1;
Darrick J. Wonge6153912012-04-29 18:23:10 -04001859 __le32 mmp_pad2[226];
1860 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
Johann Lombardic5e06d12011-05-24 18:31:25 -04001861};
1862
1863/* arguments passed to the mmp thread */
1864struct mmpd_data {
1865 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1866 struct super_block *sb; /* super block of the fs */
1867};
1868
1869/*
1870 * Check interval multiplier
1871 * The MMP block is written every update interval and initially checked every
1872 * update interval x the multiplier (the value is then adapted based on the
1873 * write latency). The reason is that writes can be delayed under load and we
1874 * don't want readers to incorrectly assume that the filesystem is no longer
1875 * in use.
1876 */
1877#define EXT4_MMP_CHECK_MULT 2UL
1878
1879/*
1880 * Minimum interval for MMP checking in seconds.
1881 */
1882#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1883
1884/*
1885 * Maximum interval for MMP checking in seconds.
1886 */
1887#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1888
1889/*
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001890 * Function prototypes
1891 */
1892
1893/*
1894 * Ok, these declarations are also in <linux/kernel.h> but none of the
Mingming Cao617ba132006-10-11 01:20:53 -07001895 * ext4 source programs needs to include it so they are duplicated here.
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001896 */
Dave Kleikamp8c55e202007-05-24 13:04:54 -04001897# define NORET_TYPE /**/
1898# define ATTRIB_NORET __attribute__((noreturn))
1899# define NORET_AND noreturn,
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001900
Theodore Ts'o498e5f22008-11-05 00:14:04 -05001901/* bitmap.c */
Theodore Ts'of6fb99c2012-06-30 19:14:57 -04001902extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
Darrick J. Wong41a246d2012-04-29 18:33:10 -04001903void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1904 struct ext4_group_desc *gdp,
1905 struct buffer_head *bh, int sz);
1906int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1907 struct ext4_group_desc *gdp,
1908 struct buffer_head *bh, int sz);
Darrick J. Wongfa77dcf2012-04-29 18:35:10 -04001909void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1910 struct ext4_group_desc *gdp,
Tao Ma79f1ba42012-10-22 00:34:32 -04001911 struct buffer_head *bh);
Darrick J. Wongfa77dcf2012-04-29 18:35:10 -04001912int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1913 struct ext4_group_desc *gdp,
Tao Ma79f1ba42012-10-22 00:34:32 -04001914 struct buffer_head *bh);
Theodore Ts'o498e5f22008-11-05 00:14:04 -05001915
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001916/* balloc.c */
Lukas Czernerbd862982013-04-03 23:32:34 -04001917extern void ext4_get_group_no_and_offset(struct super_block *sb,
1918 ext4_fsblk_t blocknr,
1919 ext4_group_t *blockgrpp,
1920 ext4_grpblk_t *offsetp);
1921extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1922 ext4_fsblk_t block);
1923
Darrick J. Wongfa77dcf2012-04-29 18:35:10 -04001924extern void ext4_validate_block_bitmap(struct super_block *sb,
1925 struct ext4_group_desc *desc,
1926 unsigned int block_group,
1927 struct buffer_head *bh);
Mingming Cao3a5b2ec2006-10-11 01:21:05 -07001928extern unsigned int ext4_block_group(struct super_block *sb,
1929 ext4_fsblk_t blocknr);
1930extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1931 ext4_fsblk_t blocknr);
Avantika Mathurfd2d4292008-01-28 23:58:27 -05001932extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1933extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1934 ext4_group_t group);
Aneesh Kumar K.V7061eba2008-07-11 19:27:31 -04001935extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
Allison Henderson55f020d2011-05-25 07:41:26 -04001936 ext4_fsblk_t goal,
1937 unsigned int flags,
1938 unsigned long *count,
1939 int *errp);
Theodore Ts'oe7d5f312011-09-09 19:14:51 -04001940extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1941 s64 nclusters, unsigned int flags);
Theodore Ts'o5dee5432011-09-09 19:10:51 -04001942extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
Theodore Ts'oaf5bc922008-09-08 22:25:24 -04001943extern void ext4_check_blocks_bitmap(struct super_block *);
Mingming Cao617ba132006-10-11 01:20:53 -07001944extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
Avantika Mathurfd2d4292008-01-28 23:58:27 -05001945 ext4_group_t block_group,
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001946 struct buffer_head ** bh);
Mingming Cao617ba132006-10-11 01:20:53 -07001947extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
Theodore Ts'o813e5722012-02-20 17:52:46 -05001948
1949extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1950 ext4_group_t block_group);
1951extern int ext4_wait_block_bitmap(struct super_block *sb,
1952 ext4_group_t block_group,
1953 struct buffer_head *bh);
1954extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1955 ext4_group_t block_group);
Theodore Ts'ofd034a82011-09-09 18:42:51 -04001956extern void ext4_init_block_bitmap(struct super_block *sb,
1957 struct buffer_head *bh,
1958 ext4_group_t group,
1959 struct ext4_group_desc *desc);
Theodore Ts'ocff1dfd72011-09-09 19:12:51 -04001960extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1961 ext4_group_t block_group,
1962 struct ext4_group_desc *gdp);
Theodore Ts'od5b8f312011-09-09 18:44:51 -04001963extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1964 ext4_group_t block_group,
1965 struct ext4_group_desc *gdp);
Eric Sandeenf86186b2011-06-28 10:01:31 -04001966ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001967
1968/* dir.c */
Theodore Ts'o60fd4da2010-07-27 11:54:40 -04001969extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
Theodore Ts'of7c21172011-01-10 12:10:55 -05001970 struct file *,
Theodore Ts'o60fd4da2010-07-27 11:54:40 -04001971 struct ext4_dir_entry_2 *,
Tao Ma226ba972012-12-10 14:05:58 -05001972 struct buffer_head *, char *, int,
1973 unsigned int);
1974#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
Theodore Ts'of7c21172011-01-10 12:10:55 -05001975 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
Tao Ma226ba972012-12-10 14:05:58 -05001976 (de), (bh), (buf), (size), (offset)))
Mingming Cao617ba132006-10-11 01:20:53 -07001977extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
Dave Kleikampac27a0e2006-10-11 01:20:50 -07001978 __u32 minor_hash,
Mingming Cao617ba132006-10-11 01:20:53 -07001979 struct ext4_dir_entry_2 *dirent);
1980extern void ext4_htree_free_dir_info(struct dir_private_info *p);
Tao Ma978fef92012-12-10 14:05:58 -05001981extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1982 struct buffer_head *bh,
1983 void *buf, int buf_size,
1984 const char *name, int namelen,
1985 struct ext4_dir_entry_2 **dest_de);
1986void ext4_insert_dentry(struct inode *inode,
1987 struct ext4_dir_entry_2 *de,
1988 int buf_size,
1989 const char *name, int namelen);
1990static inline void ext4_update_dx_flag(struct inode *inode)
1991{
1992 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
1993 EXT4_FEATURE_COMPAT_DIR_INDEX))
1994 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
1995}
Tao Ma65d165d2012-12-10 14:05:59 -05001996static unsigned char ext4_filetype_table[] = {
1997 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
1998};
1999
2000static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2001{
2002 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2003 (filetype >= EXT4_FT_MAX))
2004 return DT_UNKNOWN;
2005
2006 return ext4_filetype_table[filetype];
2007}
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002008
2009/* fsync.c */
Josef Bacik02c24a82011-07-16 20:44:56 -04002010extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
Dmitry Monakhovc2785312012-10-05 11:31:55 -04002011extern int ext4_flush_unwritten_io(struct inode *);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002012
2013/* hash.c */
Mingming Cao617ba132006-10-11 01:20:53 -07002014extern int ext4fs_dirhash(const char *name, int len, struct
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002015 dx_hash_info *hinfo);
2016
2017/* ialloc.c */
Theodore Ts'o11395752013-02-09 16:27:09 -05002018extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2019 const struct qstr *qstr, __u32 goal,
2020 uid_t *owner, int handle_type,
2021 unsigned int line_no, int nblocks);
2022
2023#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2024 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2025 0, 0, 0)
2026#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2027 type, nblocks) \
2028 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2029 (type), __LINE__, (nblocks))
2030
2031
Theodore Ts'oaf5bc922008-09-08 22:25:24 -04002032extern void ext4_free_inode(handle_t *, struct inode *);
2033extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2034extern unsigned long ext4_count_free_inodes(struct super_block *);
2035extern unsigned long ext4_count_dirs(struct super_block *);
2036extern void ext4_check_inodes_bitmap(struct super_block *);
Theodore Ts'o61d08672010-10-27 21:30:15 -04002037extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
Lukas Czernerbfff6872010-10-27 21:30:05 -04002038extern int ext4_init_inode_table(struct super_block *sb,
2039 ext4_group_t group, int barrier);
Theodore Ts'o813e5722012-02-20 17:52:46 -05002040extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002041
Alex Tomasc9de5602008-01-29 00:19:52 -05002042/* mballoc.c */
2043extern long ext4_mb_stats;
2044extern long ext4_mb_max_to_scan;
Akira Fujita9d990122012-05-28 14:19:25 -04002045extern int ext4_mb_init(struct super_block *);
Alex Tomasc9de5602008-01-29 00:19:52 -05002046extern int ext4_mb_release(struct super_block *);
2047extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2048 struct ext4_allocation_request *, int *);
2049extern int ext4_mb_reserve_blocks(struct super_block *, int);
Theodore Ts'oc2ea3fd2008-10-10 09:40:52 -04002050extern void ext4_discard_preallocations(struct inode *);
Theodore Ts'o5dabfc72010-10-27 21:30:14 -04002051extern int __init ext4_init_mballoc(void);
2052extern void ext4_exit_mballoc(void);
Theodore Ts'o44338712009-11-22 07:44:56 -05002053extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
Theodore Ts'oe6362602009-11-23 07:17:05 -05002054 struct buffer_head *bh, ext4_fsblk_t block,
2055 unsigned long count, int flags);
Theodore Ts'o28623c22012-09-05 01:31:50 -04002056extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2057 ext4_group_t ngroups);
Aneesh Kumar K.V920313a2009-01-05 21:36:19 -05002058extern int ext4_mb_add_groupinfo(struct super_block *sb,
Frederic Bohe5f21b0e2008-07-11 19:27:31 -04002059 ext4_group_t i, struct ext4_group_desc *desc);
Yongqiang Yangcc7365d2011-07-26 21:46:07 -04002060extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
Amir Goldstein2846e822011-05-09 10:46:41 -04002061 ext4_fsblk_t block, unsigned long count);
Lukas Czerner7360d172010-10-27 21:30:12 -04002062extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2063
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002064/* inode.c */
Aneesh Kumar K.V725d26d2008-01-28 23:58:27 -05002065struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2066 ext4_lblk_t, int, int *);
2067struct buffer_head *ext4_bread(handle_t *, struct inode *,
2068 ext4_lblk_t, int, int *);
Tao Maf19d5872012-12-10 14:05:51 -05002069int ext4_get_block_write(struct inode *inode, sector_t iblock,
2070 struct buffer_head *bh_result, int create);
Eric Sandeen6873fa02008-10-07 00:46:36 -04002071int ext4_get_block(struct inode *inode, sector_t iblock,
2072 struct buffer_head *bh_result, int create);
Tao Ma9c3569b2012-12-10 14:05:57 -05002073int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2074 struct buffer_head *bh, int create);
Tao Maf19d5872012-12-10 14:05:51 -05002075int ext4_walk_page_buffers(handle_t *handle,
2076 struct buffer_head *head,
2077 unsigned from,
2078 unsigned to,
2079 int *partial,
2080 int (*fn)(handle_t *handle,
2081 struct buffer_head *bh));
2082int do_journal_get_write_access(handle_t *handle,
2083 struct buffer_head *bh);
Tao Ma9c3569b2012-12-10 14:05:57 -05002084#define FALL_BACK_TO_NONDELALLOC 1
2085#define CONVERT_INLINE_DATA 2
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002086
David Howells1d1fe1e2008-02-07 00:15:37 -08002087extern struct inode *ext4_iget(struct super_block *, unsigned long);
Christoph Hellwiga9185b42010-03-05 09:21:37 +01002088extern int ext4_write_inode(struct inode *, struct writeback_control *);
Theodore Ts'oaf5bc922008-09-08 22:25:24 -04002089extern int ext4_setattr(struct dentry *, struct iattr *);
Mingming Cao3e3398a2008-07-11 19:27:31 -04002090extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2091 struct kstat *stat);
Al Viro0930fcc2010-06-07 13:16:22 -04002092extern void ext4_evict_inode(struct inode *);
2093extern void ext4_clear_inode(struct inode *);
Theodore Ts'oaf5bc922008-09-08 22:25:24 -04002094extern int ext4_sync_inode(handle_t *, struct inode *);
Christoph Hellwigaa385722011-05-27 06:53:02 -04002095extern void ext4_dirty_inode(struct inode *, int);
Mingming Cao617ba132006-10-11 01:20:53 -07002096extern int ext4_change_inode_journal_flag(struct inode *, int);
2097extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
Duane Griffin91ef4ca2008-07-11 19:27:31 -04002098extern int ext4_can_truncate(struct inode *inode);
Theodore Ts'oaf5bc922008-09-08 22:25:24 -04002099extern void ext4_truncate(struct inode *);
Allison Hendersona4bb6b62011-05-25 07:41:50 -04002100extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
Jan Kara487caee2009-08-17 22:17:20 -04002101extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
Mingming Cao617ba132006-10-11 01:20:53 -07002102extern void ext4_set_inode_flags(struct inode *);
Jan Karaff9ddf72007-07-18 09:24:20 -04002103extern void ext4_get_inode_flags(struct ext4_inode_info *);
Theodore Ts'occd25062009-02-26 01:04:07 -05002104extern int ext4_alloc_da_blocks(struct inode *inode);
Mingming Cao617ba132006-10-11 01:20:53 -07002105extern void ext4_set_aops(struct inode *inode);
Alex Tomasa86c6182006-10-11 01:21:03 -07002106extern int ext4_writepage_trans_blocks(struct inode *);
Mingming Caof3bd1f32008-08-19 22:16:03 -04002107extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
Allison Henderson4e96b2d2011-09-03 11:51:09 -04002108extern int ext4_discard_partial_page_buffers(handle_t *handle,
2109 struct address_space *mapping, loff_t from,
2110 loff_t length, int flags);
Nick Pigginc2ec1752009-03-31 15:23:21 -07002111extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
Dmitry Monakhova9e7f442009-12-14 15:21:14 +03002112extern qsize_t *ext4_get_reserved_space(struct inode *inode);
Aneesh Kumar K.V5f634d02010-01-25 04:00:31 -05002113extern void ext4_da_update_reserve_space(struct inode *inode,
2114 int used, int quota_claim);
Amir Goldsteindae1e522011-06-27 19:40:50 -04002115
2116/* indirect.c */
2117extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2118 struct ext4_map_blocks *map, int flags);
2119extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2120 const struct iovec *iov, loff_t offset,
2121 unsigned long nr_segs);
2122extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2123extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
Theodore Ts'o819c4922013-04-03 12:47:17 -04002124extern void ext4_ind_truncate(handle_t *, struct inode *inode);
Theodore Ts'o26a4c0c2013-04-03 12:45:17 -04002125extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
2126 ext4_lblk_t first, ext4_lblk_t stop);
Amir Goldsteinff9893d2011-06-27 16:36:31 -04002127
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002128/* ioctl.c */
Andi Kleen5cdd7b22008-04-29 22:03:54 -04002129extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
Theodore Ts'oaf5bc922008-09-08 22:25:24 -04002130extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002131
Aneesh Kumar K.Vc14c6fd2008-01-28 23:58:26 -05002132/* migrate.c */
Aneesh Kumar K.V2a43a872008-09-13 12:52:26 -04002133extern int ext4_ext_migrate(struct inode *);
Wei Yongjun3d0518f2009-02-14 23:01:36 -05002134
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002135/* namei.c */
Darrick J. Wongb0336e82012-04-29 18:41:10 -04002136extern int ext4_dirent_csum_verify(struct inode *inode,
2137 struct ext4_dir_entry *dirent);
Mingming Cao617ba132006-10-11 01:20:53 -07002138extern int ext4_orphan_add(handle_t *, struct inode *);
2139extern int ext4_orphan_del(handle_t *, struct inode *);
2140extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002141 __u32 start_minor_hash, __u32 *next_hash);
Tao Ma7335cd32012-12-10 14:05:59 -05002142extern int search_dir(struct buffer_head *bh,
2143 char *search_buf,
2144 int buf_size,
2145 struct inode *dir,
2146 const struct qstr *d_name,
2147 unsigned int offset,
2148 struct ext4_dir_entry_2 **res_dir);
Tao Ma05019a92012-12-10 14:06:00 -05002149extern int ext4_generic_delete_entry(handle_t *handle,
2150 struct inode *dir,
2151 struct ext4_dir_entry_2 *de_del,
2152 struct buffer_head *bh,
2153 void *entry_buf,
2154 int buf_size,
2155 int csum_size);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002156
2157/* resize.c */
Mingming Cao617ba132006-10-11 01:20:53 -07002158extern int ext4_group_add(struct super_block *sb,
2159 struct ext4_new_group_data *input);
2160extern int ext4_group_extend(struct super_block *sb,
2161 struct ext4_super_block *es,
2162 ext4_fsblk_t n_blocks_count);
Yongqiang Yang19c52462012-01-04 17:09:44 -05002163extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002164
2165/* super.c */
Theodore Ts'o952fc182012-07-09 16:27:05 -04002166extern int ext4_calculate_overhead(struct super_block *sb);
Darrick J. Wonga9c473172012-04-29 18:29:10 -04002167extern int ext4_superblock_csum_verify(struct super_block *sb,
2168 struct ext4_super_block *es);
Theodore Ts'o06db49e2012-10-10 01:06:58 -04002169extern void ext4_superblock_csum_set(struct super_block *sb);
Theodore Ts'o9933fc02011-08-01 08:45:02 -04002170extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2171extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2172extern void ext4_kvfree(void *ptr);
Theodore Ts'o117fff12012-09-05 01:29:50 -04002173extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2174 ext4_group_t ngroup);
Theodore Ts'o722887d2013-02-08 13:00:31 -05002175extern const char *ext4_decode_error(struct super_block *sb, int errno,
2176 char nbuf[16]);
Joe Perchesb9075fa2011-10-31 17:11:33 -07002177extern __printf(4, 5)
2178void __ext4_error(struct super_block *, const char *, unsigned int,
2179 const char *, ...);
Theodore Ts'oc398eda2010-07-27 11:56:40 -04002180#define ext4_error(sb, message...) __ext4_error(sb, __func__, \
2181 __LINE__, ## message)
Joe Perchesb9075fa2011-10-31 17:11:33 -07002182extern __printf(5, 6)
2183void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2184 const char *, ...);
2185extern __printf(5, 6)
2186void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2187 const char *, ...);
Theodore Ts'oc398eda2010-07-27 11:56:40 -04002188extern void __ext4_std_error(struct super_block *, const char *,
2189 unsigned int, int);
Joe Perchesb9075fa2011-10-31 17:11:33 -07002190extern __printf(4, 5)
2191void __ext4_abort(struct super_block *, const char *, unsigned int,
2192 const char *, ...);
Theodore Ts'oc67d8592010-06-29 11:07:07 -04002193#define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
Theodore Ts'oc398eda2010-07-27 11:56:40 -04002194 __LINE__, ## message)
Joe Perchesb9075fa2011-10-31 17:11:33 -07002195extern __printf(4, 5)
2196void __ext4_warning(struct super_block *, const char *, unsigned int,
2197 const char *, ...);
Theodore Ts'oc398eda2010-07-27 11:56:40 -04002198#define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2199 __LINE__, ## message)
Joe Perchesb9075fa2011-10-31 17:11:33 -07002200extern __printf(3, 4)
2201void ext4_msg(struct super_block *, const char *, const char *, ...);
Johann Lombardic5e06d12011-05-24 18:31:25 -04002202extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2203 const char *, unsigned int, const char *);
2204#define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2205 __LINE__, msg)
Joe Perchesb9075fa2011-10-31 17:11:33 -07002206extern __printf(7, 8)
2207void __ext4_grp_locked_error(const char *, unsigned int,
2208 struct super_block *, ext4_group_t,
2209 unsigned long, ext4_fsblk_t,
2210 const char *, ...);
Theodore Ts'oe29136f2010-06-29 12:54:28 -04002211#define ext4_grp_locked_error(sb, grp, message...) \
2212 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
Theodore Ts'oaf5bc922008-09-08 22:25:24 -04002213extern void ext4_update_dynamic_rev(struct super_block *sb);
Aneesh Kumar K.V99e6f822008-01-28 23:58:27 -05002214extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2215 __u32 compat);
2216extern int ext4_update_rocompat_feature(handle_t *handle,
2217 struct super_block *sb, __u32 rocompat);
2218extern int ext4_update_incompat_feature(handle_t *handle,
2219 struct super_block *sb, __u32 incompat);
Alexandre Ratchov8fadc142006-10-11 01:21:15 -07002220extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2221 struct ext4_group_desc *bg);
2222extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2223 struct ext4_group_desc *bg);
2224extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2225 struct ext4_group_desc *bg);
Theodore Ts'o021b65b2011-09-09 19:08:51 -04002226extern __u32 ext4_free_group_clusters(struct super_block *sb,
2227 struct ext4_group_desc *bg);
Aneesh Kumar K.V560671a2009-01-05 22:20:24 -05002228extern __u32 ext4_free_inodes_count(struct super_block *sb,
2229 struct ext4_group_desc *bg);
2230extern __u32 ext4_used_dirs_count(struct super_block *sb,
2231 struct ext4_group_desc *bg);
2232extern __u32 ext4_itable_unused_count(struct super_block *sb,
2233 struct ext4_group_desc *bg);
Alexandre Ratchov8fadc142006-10-11 01:21:15 -07002234extern void ext4_block_bitmap_set(struct super_block *sb,
2235 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2236extern void ext4_inode_bitmap_set(struct super_block *sb,
2237 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2238extern void ext4_inode_table_set(struct super_block *sb,
2239 struct ext4_group_desc *bg, ext4_fsblk_t blk);
Theodore Ts'o021b65b2011-09-09 19:08:51 -04002240extern void ext4_free_group_clusters_set(struct super_block *sb,
2241 struct ext4_group_desc *bg,
2242 __u32 count);
Aneesh Kumar K.V560671a2009-01-05 22:20:24 -05002243extern void ext4_free_inodes_set(struct super_block *sb,
2244 struct ext4_group_desc *bg, __u32 count);
2245extern void ext4_used_dirs_set(struct super_block *sb,
2246 struct ext4_group_desc *bg, __u32 count);
2247extern void ext4_itable_unused_set(struct super_block *sb,
2248 struct ext4_group_desc *bg, __u32 count);
Darrick J. Wongfeb0ab32012-04-29 18:45:10 -04002249extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
Theodore Ts'obb23c202009-05-01 19:44:44 -04002250 struct ext4_group_desc *gdp);
Darrick J. Wongfeb0ab32012-04-29 18:45:10 -04002251extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2252 struct ext4_group_desc *gdp);
Theodore Ts'o7f511862013-01-13 08:41:45 -05002253extern int ext4_register_li_request(struct super_block *sb,
2254 ext4_group_t first_not_zeroed);
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07002255
Darrick J. Wong41a246d2012-04-29 18:33:10 -04002256static inline int ext4_has_group_desc_csum(struct super_block *sb)
2257{
2258 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2259 EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2260 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2261}
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07002262
2263static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2264{
2265 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
Aneesh Kumar K.V6bc9fef2007-10-16 18:38:25 -04002266 le32_to_cpu(es->s_blocks_count_lo);
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07002267}
2268
2269static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2270{
2271 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
Aneesh Kumar K.V308ba3e2007-10-16 18:38:25 -04002272 le32_to_cpu(es->s_r_blocks_count_lo);
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07002273}
2274
2275static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2276{
2277 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
Aneesh Kumar K.V308ba3e2007-10-16 18:38:25 -04002278 le32_to_cpu(es->s_free_blocks_count_lo);
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07002279}
2280
2281static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2282 ext4_fsblk_t blk)
2283{
Aneesh Kumar K.V6bc9fef2007-10-16 18:38:25 -04002284 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07002285 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2286}
2287
2288static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2289 ext4_fsblk_t blk)
2290{
Aneesh Kumar K.V308ba3e2007-10-16 18:38:25 -04002291 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07002292 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2293}
2294
2295static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2296 ext4_fsblk_t blk)
2297{
Aneesh Kumar K.V308ba3e2007-10-16 18:38:25 -04002298 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07002299 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2300}
2301
Aneesh Kumar K.Va48380f2008-01-28 23:58:27 -05002302static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2303{
Theodore Ts'o06a279d2009-01-17 18:41:37 -05002304 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2305 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2306 le32_to_cpu(raw_inode->i_size_lo);
2307 else
2308 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
Aneesh Kumar K.Va48380f2008-01-28 23:58:27 -05002309}
Laurent Vivierbd81d8e2006-10-11 01:21:10 -07002310
Aneesh Kumar K.Va48380f2008-01-28 23:58:27 -05002311static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2312{
2313 raw_inode->i_size_lo = cpu_to_le32(i_size);
2314 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2315}
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002316
Alex Tomasc9de5602008-01-29 00:19:52 -05002317static inline
2318struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
Theodore Ts'oaf5bc922008-09-08 22:25:24 -04002319 ext4_group_t group)
Alex Tomasc9de5602008-01-29 00:19:52 -05002320{
2321 struct ext4_group_info ***grp_info;
2322 long indexv, indexh;
2323 grp_info = EXT4_SB(sb)->s_group_info;
2324 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2325 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2326 return grp_info[indexv][indexh];
2327}
2328
Theodore Ts'o8df96752009-05-01 08:50:38 -04002329/*
2330 * Reading s_groups_count requires using smp_rmb() afterwards. See
2331 * the locking protocol documented in the comments of ext4_group_add()
2332 * in resize.c
2333 */
2334static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2335{
2336 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2337
2338 smp_rmb();
2339 return ngroups;
2340}
Alex Tomasc9de5602008-01-29 00:19:52 -05002341
Jose R. Santos772cb7c2008-07-11 19:27:31 -04002342static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2343 ext4_group_t block_group)
2344{
2345 return block_group >> sbi->s_log_groups_per_flex;
2346}
2347
2348static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2349{
2350 return 1 << sbi->s_log_groups_per_flex;
2351}
2352
Mingming Cao617ba132006-10-11 01:20:53 -07002353#define ext4_std_error(sb, errno) \
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002354do { \
2355 if ((errno)) \
Theodore Ts'oc398eda2010-07-27 11:56:40 -04002356 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002357} while (0)
2358
Aneesh Kumar K.Va30d542a2008-10-09 10:56:23 -04002359#ifdef CONFIG_SMP
Theodore Ts'odf55c992011-09-09 19:16:51 -04002360/* Each CPU can accumulate percpu_counter_batch clusters in their local
2361 * counters. So we need to make sure we have free clusters more
Eric Dumazet179f7eb2009-01-06 14:41:04 -08002362 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
Aneesh Kumar K.Va30d542a2008-10-09 10:56:23 -04002363 */
Theodore Ts'odf55c992011-09-09 19:16:51 -04002364#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
Aneesh Kumar K.Va30d542a2008-10-09 10:56:23 -04002365#else
Theodore Ts'odf55c992011-09-09 19:16:51 -04002366#define EXT4_FREECLUSTERS_WATERMARK 0
Aneesh Kumar K.Va30d542a2008-10-09 10:56:23 -04002367#endif
2368
Aneesh Kumar K.Vcf17fea2008-09-13 13:06:18 -04002369static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2370{
2371 /*
2372 * XXX: replace with spinlock if seen contended -bzzz
2373 */
2374 down_write(&EXT4_I(inode)->i_data_sem);
2375 if (newsize > EXT4_I(inode)->i_disksize)
2376 EXT4_I(inode)->i_disksize = newsize;
2377 up_write(&EXT4_I(inode)->i_data_sem);
2378 return ;
2379}
Aneesh Kumar K.Va30d542a2008-10-09 10:56:23 -04002380
Aneesh Kumar K.V5d1b1b32009-01-05 22:19:52 -05002381struct ext4_group_info {
2382 unsigned long bb_state;
2383 struct rb_root bb_free_root;
Eric Sandeena36b4492009-08-25 22:36:45 -04002384 ext4_grpblk_t bb_first_free; /* first free block */
2385 ext4_grpblk_t bb_free; /* total free blocks */
2386 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
Curt Wohlgemuth8a57d9d2010-05-16 15:00:00 -04002387 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
Aneesh Kumar K.V5d1b1b32009-01-05 22:19:52 -05002388 struct list_head bb_prealloc_list;
2389#ifdef DOUBLE_CHECK
2390 void *bb_bitmap;
2391#endif
2392 struct rw_semaphore alloc_sem;
Eric Sandeena36b4492009-08-25 22:36:45 -04002393 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2394 * regions, index is order.
2395 * bb_counters[3] = 5 means
2396 * 5 free 8-block regions. */
Aneesh Kumar K.V5d1b1b32009-01-05 22:19:52 -05002397};
2398
Tao Ma3d56b8d2011-07-11 00:03:38 -04002399#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2400#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
Aneesh Kumar K.V5d1b1b32009-01-05 22:19:52 -05002401
2402#define EXT4_MB_GRP_NEED_INIT(grp) \
2403 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2404
Tao Ma3d56b8d2011-07-11 00:03:38 -04002405#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2406 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2407#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2408 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2409#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2410 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2411
Theodore Ts'o50797482009-09-18 13:34:02 -04002412#define EXT4_MAX_CONTENTION 8
2413#define EXT4_CONTENTION_THRESHOLD 2
2414
Aneesh Kumar K.V955ce5f2009-05-02 20:35:09 -04002415static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2416 ext4_group_t group)
2417{
2418 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2419}
2420
Theodore Ts'o50797482009-09-18 13:34:02 -04002421/*
2422 * Returns true if the filesystem is busy enough that attempts to
2423 * access the block group locks has run into contention.
2424 */
2425static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2426{
2427 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2428}
2429
Aneesh Kumar K.V5d1b1b32009-01-05 22:19:52 -05002430static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2431{
Theodore Ts'o50797482009-09-18 13:34:02 -04002432 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2433 if (spin_trylock(lock))
2434 /*
2435 * We're able to grab the lock right away, so drop the
2436 * lock contention counter.
2437 */
2438 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2439 else {
2440 /*
2441 * The lock is busy, so bump the contention counter,
2442 * and then wait on the spin lock.
2443 */
2444 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2445 EXT4_MAX_CONTENTION);
2446 spin_lock(lock);
2447 }
Aneesh Kumar K.V5d1b1b32009-01-05 22:19:52 -05002448}
2449
2450static inline void ext4_unlock_group(struct super_block *sb,
2451 ext4_group_t group)
2452{
Aneesh Kumar K.V955ce5f2009-05-02 20:35:09 -04002453 spin_unlock(ext4_group_lock_ptr(sb, group));
Aneesh Kumar K.V5d1b1b32009-01-05 22:19:52 -05002454}
2455
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002456/*
Theodore Ts'o1f7d1e72011-06-27 19:16:02 -04002457 * Block validity checking
2458 */
2459#define ext4_check_indirect_blockref(inode, bh) \
2460 ext4_check_blockref(__func__, __LINE__, inode, \
2461 (__le32 *)(bh)->b_data, \
2462 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2463
2464#define ext4_ind_check_inode(inode) \
2465 ext4_check_blockref(__func__, __LINE__, inode, \
2466 EXT4_I(inode)->i_data, \
2467 EXT4_NDIR_BLOCKS)
2468
2469/*
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002470 * Inodes and files operations
2471 */
2472
2473/* dir.c */
Mingming Cao617ba132006-10-11 01:20:53 -07002474extern const struct file_operations ext4_dir_operations;
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002475
2476/* file.c */
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002477extern const struct inode_operations ext4_file_inode_operations;
Mingming Cao617ba132006-10-11 01:20:53 -07002478extern const struct file_operations ext4_file_operations;
Toshiyuki Okajimae0d10bf2010-10-27 21:30:06 -04002479extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
Dmitry Monakhovc2785312012-10-05 11:31:55 -04002480extern void ext4_unwritten_wait(struct inode *inode);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002481
Theodore Ts'o95eaefb2013-02-09 15:23:03 -05002482/* inline.c */
2483extern int ext4_has_inline_data(struct inode *inode);
2484extern int ext4_get_inline_size(struct inode *inode);
2485extern int ext4_get_max_inline_size(struct inode *inode);
2486extern int ext4_find_inline_data_nolock(struct inode *inode);
2487extern void ext4_write_inline_data(struct inode *inode,
2488 struct ext4_iloc *iloc,
2489 void *buffer, loff_t pos,
2490 unsigned int len);
2491extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode,
2492 unsigned int len);
2493extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2494 unsigned int len);
2495extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2496
2497extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2498extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2499 struct inode *inode,
2500 loff_t pos, unsigned len,
2501 unsigned flags,
2502 struct page **pagep);
2503extern int ext4_write_inline_data_end(struct inode *inode,
2504 loff_t pos, unsigned len,
2505 unsigned copied,
2506 struct page *page);
2507extern struct buffer_head *
2508ext4_journalled_write_inline_data(struct inode *inode,
2509 unsigned len,
2510 struct page *page);
2511extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2512 struct inode *inode,
2513 loff_t pos, unsigned len,
2514 unsigned flags,
2515 struct page **pagep,
2516 void **fsdata);
2517extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2518 unsigned len, unsigned copied,
2519 struct page *page);
2520extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2521 struct inode *inode);
2522extern int ext4_try_create_inline_dir(handle_t *handle,
2523 struct inode *parent,
2524 struct inode *inode);
2525extern int ext4_read_inline_dir(struct file *filp,
2526 void *dirent, filldir_t filldir,
2527 int *has_inline_data);
2528extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2529 const struct qstr *d_name,
2530 struct ext4_dir_entry_2 **res_dir,
2531 int *has_inline_data);
2532extern int ext4_delete_inline_entry(handle_t *handle,
2533 struct inode *dir,
2534 struct ext4_dir_entry_2 *de_del,
2535 struct buffer_head *bh,
2536 int *has_inline_data);
2537extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2538extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2539 struct ext4_dir_entry_2 **parent_de,
2540 int *retval);
2541extern int ext4_inline_data_fiemap(struct inode *inode,
2542 struct fiemap_extent_info *fieinfo,
2543 int *has_inline);
2544extern int ext4_try_to_evict_inline_data(handle_t *handle,
2545 struct inode *inode,
2546 int needed);
2547extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2548
2549extern int ext4_convert_inline_data(struct inode *inode);
2550
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002551/* namei.c */
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002552extern const struct inode_operations ext4_dir_inode_operations;
2553extern const struct inode_operations ext4_special_inode_operations;
Theodore Ts'o596397b2009-05-01 13:49:15 -04002554extern struct dentry *ext4_get_parent(struct dentry *child);
Tao Maa774f9c2012-12-10 14:05:57 -05002555extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2556 struct ext4_dir_entry_2 *de,
2557 int blocksize, int csum_size,
2558 unsigned int parent_ino, int dotdot_real_len);
Tao Ma3c47d542012-12-10 14:05:59 -05002559extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2560 unsigned int blocksize);
2561extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2562 struct inode *inode,
2563 struct buffer_head *bh);
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002564
2565/* symlink.c */
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002566extern const struct inode_operations ext4_symlink_inode_operations;
2567extern const struct inode_operations ext4_fast_symlink_inode_operations;
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002568
Theodore Ts'o6fd058f2009-05-17 15:38:01 -04002569/* block_validity */
2570extern void ext4_release_system_zone(struct super_block *sb);
2571extern int ext4_setup_system_zone(struct super_block *sb);
Theodore Ts'o5dabfc72010-10-27 21:30:14 -04002572extern int __init ext4_init_system_zone(void);
2573extern void ext4_exit_system_zone(void);
Theodore Ts'o6fd058f2009-05-17 15:38:01 -04002574extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2575 ext4_fsblk_t start_blk,
2576 unsigned int count);
Theodore Ts'o1f7d1e72011-06-27 19:16:02 -04002577extern int ext4_check_blockref(const char *, unsigned int,
2578 struct inode *, __le32 *, unsigned int);
Theodore Ts'o6fd058f2009-05-17 15:38:01 -04002579
Alex Tomasa86c6182006-10-11 01:21:03 -07002580/* extents.c */
Theodore Ts'o4a092d72012-11-28 13:03:30 -05002581struct ext4_ext_path;
2582struct ext4_extent;
2583
Alex Tomasa86c6182006-10-11 01:21:03 -07002584extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2585extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
Mingming Caoa02908f2008-08-19 22:16:07 -04002586extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2587 int chunk);
Theodore Ts'oe35fd662010-05-16 19:00:00 -04002588extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2589 struct ext4_map_blocks *map, int flags);
Theodore Ts'o819c4922013-04-03 12:47:17 -04002590extern void ext4_ext_truncate(handle_t *, struct inode *);
Theodore Ts'o26a4c0c2013-04-03 12:45:17 -04002591extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2592 ext4_lblk_t end);
Alex Tomasa86c6182006-10-11 01:21:03 -07002593extern void ext4_ext_init(struct super_block *);
2594extern void ext4_ext_release(struct super_block *);
Christoph Hellwig2fe17c12011-01-14 13:07:43 +01002595extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
Amit Aroraa2df2a62007-07-17 21:42:41 -04002596 loff_t len);
Mingming Cao00314622009-09-28 15:49:08 -04002597extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
Eric Sandeena1de02d2010-02-04 23:58:38 -05002598 ssize_t len);
Theodore Ts'oe35fd662010-05-16 19:00:00 -04002599extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2600 struct ext4_map_blocks *map, int flags);
Theodore Ts'o4a092d72012-11-28 13:03:30 -05002601extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2602 ext4_lblk_t lblocks);
2603extern int ext4_extent_tree_init(handle_t *, struct inode *);
2604extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2605 int num,
2606 struct ext4_ext_path *path);
2607extern int ext4_can_extents_be_merged(struct inode *inode,
2608 struct ext4_extent *ex1,
2609 struct ext4_extent *ex2);
2610extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2611 struct ext4_ext_path *,
2612 struct ext4_extent *, int);
2613extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2614 struct ext4_ext_path *);
2615extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2616extern int ext4_ext_check_inode(struct inode *inode);
Zheng Liuf7fec032013-02-18 00:28:47 -05002617extern int ext4_find_delalloc_range(struct inode *inode,
2618 ext4_lblk_t lblk_start,
2619 ext4_lblk_t lblk_end);
Theodore Ts'o4a092d72012-11-28 13:03:30 -05002620extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
Aneesh Kumar K.V3a06d772008-11-22 15:04:59 -05002621extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2622 __u64 start, __u64 len);
Theodore Ts'o996bb9f2013-04-03 22:04:52 -04002623extern int ext4_ind_migrate(struct inode *inode);
Theodore Ts'o4a092d72012-11-28 13:03:30 -05002624
2625
Akira Fujita748de672009-06-17 19:24:03 -04002626/* move_extent.c */
2627extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2628 __u64 start_orig, __u64 start_donor,
2629 __u64 len, __u64 *moved_len);
2630
Theodore Ts'obd2d0212010-10-27 21:30:10 -04002631/* page-io.c */
Theodore Ts'o5dabfc72010-10-27 21:30:14 -04002632extern int __init ext4_init_pageio(void);
Dmitry Monakhov28a535f2012-09-29 00:14:55 -04002633extern void ext4_add_complete_io(ext4_io_end_t *io_end);
Theodore Ts'o5dabfc72010-10-27 21:30:14 -04002634extern void ext4_exit_pageio(void);
Theodore Ts'o1ada47d2013-03-20 09:39:42 -04002635extern void ext4_ioend_shutdown(struct inode *);
Theodore Ts'obd2d0212010-10-27 21:30:10 -04002636extern void ext4_free_io_end(ext4_io_end_t *io);
2637extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
Jan Kara84c17542013-01-28 09:43:46 -05002638extern void ext4_end_io_work(struct work_struct *work);
Theodore Ts'obd2d0212010-10-27 21:30:10 -04002639extern void ext4_io_submit(struct ext4_io_submit *io);
2640extern int ext4_bio_write_page(struct ext4_io_submit *io,
2641 struct page *page,
2642 int len,
2643 struct writeback_control *wbc);
Aneesh Kumar K.V3a06d772008-11-22 15:04:59 -05002644
Johann Lombardic5e06d12011-05-24 18:31:25 -04002645/* mmp.c */
2646extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
Darrick J. Wong5c359a42012-04-29 18:47:10 -04002647extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2648extern int ext4_mmp_csum_verify(struct super_block *sb,
2649 struct mmp_struct *mmp);
Johann Lombardic5e06d12011-05-24 18:31:25 -04002650
Jiaying Zhang744692d2010-03-04 16:14:02 -05002651/* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2652enum ext4_state_bits {
2653 BH_Uninit /* blocks are allocated but uninitialized on disk */
2654 = BH_JBDPrivateStart,
Aditya Kali7b415bf2011-09-09 19:04:51 -04002655 BH_AllocFromCluster, /* allocated blocks were part of already
2656 * allocated cluster. Note that this flag will
2657 * never, ever appear in a buffer_head's state
2658 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2659 * this is used. */
Jiaying Zhang744692d2010-03-04 16:14:02 -05002660};
2661
2662BUFFER_FNS(Uninit, uninit)
2663TAS_BUFFER_FNS(Uninit, uninit)
2664
Aneesh Kumar K.V2ccb5fb2009-01-05 21:49:55 -05002665/*
Adam Buchbinder48fc7f72012-09-19 21:48:00 -04002666 * Add new method to test whether block and inode bitmaps are properly
Aneesh Kumar K.V2ccb5fb2009-01-05 21:49:55 -05002667 * initialized. With uninit_bg reading the block from disk is not enough
2668 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2669 */
2670#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2671
2672static inline int bitmap_uptodate(struct buffer_head *bh)
2673{
2674 return (buffer_uptodate(bh) &&
2675 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2676}
2677static inline void set_bitmap_uptodate(struct buffer_head *bh)
2678{
2679 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2680}
2681
Dmitry Monakhov17335dc2012-09-29 00:41:21 -04002682/*
2683 * Disable DIO read nolock optimization, so new dioreaders will be forced
2684 * to grab i_mutex
2685 */
2686static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2687{
2688 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2689 smp_mb();
2690}
2691static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2692{
2693 smp_mb();
2694 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2695}
2696
Akinobu Mita731eb1a2010-03-03 23:55:01 -05002697#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2698
Eric Sandeene9e3bce2011-02-12 08:17:34 -05002699/* For ioend & aio unwritten conversion wait queues */
2700#define EXT4_WQ_HASH_SZ 37
2701#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2702 EXT4_WQ_HASH_SZ])
2703#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2704 EXT4_WQ_HASH_SZ])
2705extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2706extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2707
Yongqiang Yang8f82f842011-07-26 21:35:44 -04002708#define EXT4_RESIZING 0
2709extern int ext4_resize_begin(struct super_block *sb);
2710extern void ext4_resize_end(struct super_block *sb);
2711
Dave Kleikampac27a0e2006-10-11 01:20:50 -07002712#endif /* __KERNEL__ */
2713
Christoph Hellwig3dcf5452008-04-29 18:13:32 -04002714#endif /* _EXT4_H */