Linux Audio

Check our new training course

Loading...
v3.5.6
 
   1/*
   2 *  ext4.h
   3 *
   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
  16#ifndef _EXT4_H
  17#define _EXT4_H
  18
  19#include <linux/types.h>
  20#include <linux/blkdev.h>
  21#include <linux/magic.h>
  22#include <linux/jbd2.h>
  23#include <linux/quota.h>
  24#include <linux/rwsem.h>
  25#include <linux/rbtree.h>
  26#include <linux/seqlock.h>
  27#include <linux/mutex.h>
  28#include <linux/timer.h>
 
  29#include <linux/wait.h>
 
  30#include <linux/blockgroup_lock.h>
  31#include <linux/percpu_counter.h>
 
  32#include <crypto/hash.h>
 
 
 
  33#ifdef __KERNEL__
  34#include <linux/compat.h>
  35#endif
  36
 
 
 
 
 
  37/*
  38 * The fourth extended filesystem constants/structures
  39 */
  40
  41/*
 
 
 
 
 
 
 
 
 
 
 
 
  42 * Define EXT4FS_DEBUG to produce debug messages
  43 */
  44#undef EXT4FS_DEBUG
  45
  46/*
  47 * Debug code
  48 */
  49#ifdef EXT4FS_DEBUG
  50#define ext4_debug(f, a...)						\
  51	do {								\
  52		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
  53			__FILE__, __LINE__, __func__);			\
  54		printk(KERN_DEBUG f, ## a);				\
  55	} while (0)
  56#else
  57#define ext4_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
  58#endif
  59
  60#define EXT4_ERROR_INODE(inode, fmt, a...) \
  61	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
 
 
  62
  63#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)			\
  64	ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
  65
  66#define EXT4_ERROR_FILE(file, block, fmt, a...)				\
  67	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
 
 
 
 
 
 
 
  68
  69/* data type for block offset of block group */
  70typedef int ext4_grpblk_t;
  71
  72/* data type for filesystem-wide blocks number */
  73typedef unsigned long long ext4_fsblk_t;
  74
  75/* data type for file logical block number */
  76typedef __u32 ext4_lblk_t;
  77
  78/* data type for block group number */
  79typedef unsigned int ext4_group_t;
  80
 
 
 
 
 
  81/*
  82 * Flags used in mballoc's allocation_context flags field.
  83 *
  84 * Also used to show what's going on for debugging purposes when the
  85 * flag field is exported via the traceport interface
  86 */
  87
  88/* prefer goal again. length */
  89#define EXT4_MB_HINT_MERGE		0x0001
  90/* blocks already reserved */
  91#define EXT4_MB_HINT_RESERVED		0x0002
  92/* metadata is being allocated */
  93#define EXT4_MB_HINT_METADATA		0x0004
  94/* first blocks in the file */
  95#define EXT4_MB_HINT_FIRST		0x0008
  96/* search for the best chunk */
  97#define EXT4_MB_HINT_BEST		0x0010
  98/* data is being allocated */
  99#define EXT4_MB_HINT_DATA		0x0020
 100/* don't preallocate (for tails) */
 101#define EXT4_MB_HINT_NOPREALLOC		0x0040
 102/* allocate for locality group */
 103#define EXT4_MB_HINT_GROUP_ALLOC	0x0080
 104/* allocate goal blocks or none */
 105#define EXT4_MB_HINT_GOAL_ONLY		0x0100
 106/* goal is meaningful */
 107#define EXT4_MB_HINT_TRY_GOAL		0x0200
 108/* blocks already pre-reserved by delayed allocation */
 109#define EXT4_MB_DELALLOC_RESERVED	0x0400
 110/* We are doing stream allocation */
 111#define EXT4_MB_STREAM_ALLOC		0x0800
 112/* Use reserved root blocks if needed */
 113#define EXT4_MB_USE_ROOT_BLOCKS		0x1000
 
 
 
 
 114
 115struct ext4_allocation_request {
 116	/* target inode for block we're allocating */
 117	struct inode *inode;
 118	/* how many blocks we want to allocate */
 119	unsigned int len;
 120	/* logical block in target inode */
 121	ext4_lblk_t logical;
 122	/* the closest logical allocated block to the left */
 123	ext4_lblk_t lleft;
 124	/* the closest logical allocated block to the right */
 125	ext4_lblk_t lright;
 126	/* phys. target (a hint) */
 127	ext4_fsblk_t goal;
 128	/* phys. block for the closest logical allocated block to the left */
 129	ext4_fsblk_t pleft;
 130	/* phys. block for the closest logical allocated block to the right */
 131	ext4_fsblk_t pright;
 132	/* flags. see above EXT4_MB_HINT_* */
 133	unsigned int flags;
 134};
 135
 136/*
 137 * Logical to physical block mapping, used by ext4_map_blocks()
 138 *
 139 * This structure is used to pass requests into ext4_map_blocks() as
 140 * well as to store the information returned by ext4_map_blocks().  It
 141 * takes less room on the stack than a struct buffer_head.
 142 */
 143#define EXT4_MAP_NEW		(1 << BH_New)
 144#define EXT4_MAP_MAPPED		(1 << BH_Mapped)
 145#define EXT4_MAP_UNWRITTEN	(1 << BH_Unwritten)
 146#define EXT4_MAP_BOUNDARY	(1 << BH_Boundary)
 147#define EXT4_MAP_UNINIT		(1 << BH_Uninit)
 148/* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
 149 * ext4_map_blocks wants to know whether or not the underlying cluster has
 150 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
 151 * the requested mapping was from previously mapped (or delayed allocated)
 152 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
 153 * should never appear on buffer_head's state flags.
 154 */
 155#define EXT4_MAP_FROM_CLUSTER	(1 << BH_AllocFromCluster)
 156#define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
 157				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
 158				 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
 159
 160struct ext4_map_blocks {
 161	ext4_fsblk_t m_pblk;
 162	ext4_lblk_t m_lblk;
 163	unsigned int m_len;
 164	unsigned int m_flags;
 165};
 166
 167/*
 168 * For delayed allocation tracking
 169 */
 170struct mpage_da_data {
 171	struct inode *inode;
 172	sector_t b_blocknr;		/* start block number of extent */
 173	size_t b_size;			/* size of extent */
 174	unsigned long b_state;		/* state of the extent */
 175	unsigned long first_page, next_page;	/* extent of pages */
 176	struct writeback_control *wbc;
 177	int io_done;
 178	int pages_written;
 179	int retval;
 180};
 181
 182/*
 183 * Flags for ext4_io_end->flags
 184 */
 185#define	EXT4_IO_END_UNWRITTEN	0x0001
 186#define EXT4_IO_END_ERROR	0x0002
 187#define EXT4_IO_END_QUEUED	0x0004
 188#define EXT4_IO_END_DIRECT	0x0008
 189#define EXT4_IO_END_IN_FSYNC	0x0010
 190
 191struct ext4_io_page {
 192	struct page	*p_page;
 193	atomic_t	p_count;
 
 194};
 195
 196#define MAX_IO_PAGES 128
 197
 198/*
 199 * For converting uninitialized extents on a work queue.
 200 *
 201 * 'page' is only used from the writepage() path; 'pages' is only used for
 202 * buffered writes; they are used to keep page references until conversion
 203 * takes place.  For AIO/DIO, neither field is filled in.
 204 */
 205typedef struct ext4_io_end {
 206	struct list_head	list;		/* per-file finished IO list */
 
 
 207	struct inode		*inode;		/* file being written to */
 
 
 208	unsigned int		flag;		/* unwritten or not */
 209	struct page		*page;		/* for writepage() path */
 210	loff_t			offset;		/* offset in the file */
 211	ssize_t			size;		/* size of the extent */
 212	struct work_struct	work;		/* data work queue */
 213	struct kiocb		*iocb;		/* iocb struct for AIO */
 214	int			result;		/* error value for AIO */
 215	int			num_io_pages;   /* for writepages() */
 216	struct ext4_io_page	*pages[MAX_IO_PAGES]; /* for writepages() */
 217} ext4_io_end_t;
 218
 219struct ext4_io_submit {
 220	int			io_op;
 221	struct bio		*io_bio;
 222	ext4_io_end_t		*io_end;
 223	struct ext4_io_page	*io_page;
 224	sector_t		io_next_block;
 225};
 226
 227/*
 228 * Special inodes numbers
 229 */
 230#define	EXT4_BAD_INO		 1	/* Bad blocks inode */
 231#define EXT4_ROOT_INO		 2	/* Root inode */
 232#define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
 233#define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
 234#define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
 235#define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
 236#define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
 237#define EXT4_JOURNAL_INO	 8	/* Journal inode */
 238
 239/* First non-reserved inode for old ext4 filesystems */
 240#define EXT4_GOOD_OLD_FIRST_INO	11
 241
 242/*
 243 * Maximal count of links to a file
 244 */
 245#define EXT4_LINK_MAX		65000
 246
 247/*
 248 * Macro-instructions used to manage several block sizes
 249 */
 250#define EXT4_MIN_BLOCK_SIZE		1024
 251#define	EXT4_MAX_BLOCK_SIZE		65536
 252#define EXT4_MIN_BLOCK_LOG_SIZE		10
 253#define EXT4_MAX_BLOCK_LOG_SIZE		16
 
 254#ifdef __KERNEL__
 255# define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
 256#else
 257# define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
 258#endif
 259#define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
 260#define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
 261					 EXT4_SB(s)->s_cluster_bits)
 262#ifdef __KERNEL__
 263# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
 264# define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
 265#else
 266# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
 267#endif
 268#ifdef __KERNEL__
 269#define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
 270#define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
 271#define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
 272#else
 273#define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 274				 EXT4_GOOD_OLD_INODE_SIZE : \
 275				 (s)->s_inode_size)
 276#define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 277				 EXT4_GOOD_OLD_FIRST_INO : \
 278				 (s)->s_first_ino)
 279#endif
 280#define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
 
 
 
 281
 282/* Translate a block number to a cluster number */
 283#define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
 284/* Translate a cluster number to a block number */
 285#define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
 286/* Translate # of blks to # of clusters */
 287#define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
 288				 (sbi)->s_cluster_bits)
 
 
 
 
 
 
 
 
 
 
 
 
 
 289
 290/*
 291 * Structure of a blocks group descriptor
 292 */
 293struct ext4_group_desc
 294{
 295	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
 296	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
 297	__le32	bg_inode_table_lo;	/* Inodes table block */
 298	__le16	bg_free_blocks_count_lo;/* Free blocks count */
 299	__le16	bg_free_inodes_count_lo;/* Free inodes count */
 300	__le16	bg_used_dirs_count_lo;	/* Directories count */
 301	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
 302	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
 303	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
 304	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
 305	__le16  bg_itable_unused_lo;	/* Unused inodes count */
 306	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
 307	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
 308	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
 309	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
 310	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
 311	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
 312	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
 313	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
 314	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
 315	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
 316	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
 317	__u32   bg_reserved;
 318};
 319
 320#define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
 321	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
 322	 sizeof(__le16))
 323#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
 324	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
 325	 sizeof(__le16))
 326
 327/*
 328 * Structure of a flex block group info
 329 */
 330
 331struct flex_groups {
 332	atomic_t free_inodes;
 333	atomic_t free_clusters;
 334	atomic_t used_dirs;
 335};
 336
 337#define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
 338#define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
 339#define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
 340
 341/*
 342 * Macro-instructions used to manage group descriptors
 343 */
 344#define EXT4_MIN_DESC_SIZE		32
 345#define EXT4_MIN_DESC_SIZE_64BIT	64
 346#define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
 347#define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
 348#ifdef __KERNEL__
 349# define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
 350# define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
 351# define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
 352# define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
 353# define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
 354#else
 355# define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
 356# define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
 357# define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
 358#endif
 359
 360/*
 361 * Constants relative to the data blocks
 362 */
 363#define	EXT4_NDIR_BLOCKS		12
 364#define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
 365#define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
 366#define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
 367#define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
 368
 369/*
 370 * Inode flags
 371 */
 372#define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
 373#define	EXT4_UNRM_FL			0x00000002 /* Undelete */
 374#define	EXT4_COMPR_FL			0x00000004 /* Compress file */
 375#define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
 376#define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
 377#define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
 378#define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
 379#define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
 380/* Reserved for compression usage... */
 381#define EXT4_DIRTY_FL			0x00000100
 382#define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
 383#define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
 384#define EXT4_ECOMPR_FL			0x00000800 /* Compression error */
 
 385/* End compression flags --- maybe not all used */
 386#define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
 387#define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
 388#define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
 389#define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
 390#define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
 391#define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
 392#define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
 393#define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
 
 394#define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
 395#define EXT4_EOFBLOCKS_FL		0x00400000 /* Blocks allocated beyond EOF */
 
 
 
 
 
 
 396#define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
 397
 398#define EXT4_FL_USER_VISIBLE		0x004BDFFF /* User visible flags */
 399#define EXT4_FL_USER_MODIFIABLE		0x004B80FF /* User modifiable flags */
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 400
 401/* Flags that should be inherited by new inodes from their parent. */
 402#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
 403			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
 404			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
 405			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
 
 
 406
 407/* Flags that are appropriate for regular files (all but dir-specific ones). */
 408#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
 
 409
 410/* Flags that are appropriate for non-directories/regular files. */
 411#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
 412
 
 
 
 
 
 
 
 413/* Mask out flags that are inappropriate for the given type of inode. */
 414static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
 415{
 416	if (S_ISDIR(mode))
 417		return flags;
 418	else if (S_ISREG(mode))
 419		return flags & EXT4_REG_FLMASK;
 420	else
 421		return flags & EXT4_OTHER_FLMASK;
 422}
 423
 424/*
 425 * Inode flags used for atomic set/get
 426 */
 427enum {
 428	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
 429	EXT4_INODE_UNRM		= 1,	/* Undelete */
 430	EXT4_INODE_COMPR	= 2,	/* Compress file */
 431	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
 432	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
 433	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
 434	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
 435	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
 436/* Reserved for compression usage... */
 437	EXT4_INODE_DIRTY	= 8,
 438	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
 439	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
 440	EXT4_INODE_ECOMPR	= 11,	/* Compression error */
 441/* End compression flags --- maybe not all used */
 442	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
 443	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
 444	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
 445	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
 446	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
 447	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
 448	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
 449	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
 
 450	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
 451	EXT4_INODE_EOFBLOCKS	= 22,	/* Blocks allocated beyond EOF */
 
 
 
 
 452	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
 453};
 454
 455#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
 456#define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
 457	printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
 458		EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
 459
 460/*
 461 * Since it's pretty easy to mix up bit numbers and hex values, and we
 462 * can't do a compile-time test for ENUM values, we use a run-time
 463 * test to make sure that EXT4_XXX_FL is consistent with respect to
 464 * EXT4_INODE_XXX.  If all is well the printk and BUG_ON will all drop
 465 * out so it won't cost any extra space in the compiled kernel image.
 466 * But it's important that these values are the same, since we are
 467 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
 468 * must be consistent with the values of FS_XXX_FL defined in
 469 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
 470 * ext4 filesystems, and of course the values defined in e2fsprogs.
 471 *
 472 * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
 473 */
 
 
 
 474static inline void ext4_check_flag_values(void)
 475{
 476	CHECK_FLAG_VALUE(SECRM);
 477	CHECK_FLAG_VALUE(UNRM);
 478	CHECK_FLAG_VALUE(COMPR);
 479	CHECK_FLAG_VALUE(SYNC);
 480	CHECK_FLAG_VALUE(IMMUTABLE);
 481	CHECK_FLAG_VALUE(APPEND);
 482	CHECK_FLAG_VALUE(NODUMP);
 483	CHECK_FLAG_VALUE(NOATIME);
 484	CHECK_FLAG_VALUE(DIRTY);
 485	CHECK_FLAG_VALUE(COMPRBLK);
 486	CHECK_FLAG_VALUE(NOCOMPR);
 487	CHECK_FLAG_VALUE(ECOMPR);
 488	CHECK_FLAG_VALUE(INDEX);
 489	CHECK_FLAG_VALUE(IMAGIC);
 490	CHECK_FLAG_VALUE(JOURNAL_DATA);
 491	CHECK_FLAG_VALUE(NOTAIL);
 492	CHECK_FLAG_VALUE(DIRSYNC);
 493	CHECK_FLAG_VALUE(TOPDIR);
 494	CHECK_FLAG_VALUE(HUGE_FILE);
 495	CHECK_FLAG_VALUE(EXTENTS);
 
 496	CHECK_FLAG_VALUE(EA_INODE);
 497	CHECK_FLAG_VALUE(EOFBLOCKS);
 
 
 498	CHECK_FLAG_VALUE(RESERVED);
 499}
 500
 501/* Used to pass group descriptor data when online resize is done */
 502struct ext4_new_group_input {
 503	__u32 group;		/* Group number for this data */
 504	__u64 block_bitmap;	/* Absolute block number of block bitmap */
 505	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
 506	__u64 inode_table;	/* Absolute block number of inode table start */
 507	__u32 blocks_count;	/* Total number of blocks in this group */
 508	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
 509	__u16 unused;
 510};
 511
 512#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 513struct compat_ext4_new_group_input {
 514	u32 group;
 515	compat_u64 block_bitmap;
 516	compat_u64 inode_bitmap;
 517	compat_u64 inode_table;
 518	u32 blocks_count;
 519	u16 reserved_blocks;
 520	u16 unused;
 521};
 522#endif
 523
 524/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
 525struct ext4_new_group_data {
 526	__u32 group;
 527	__u64 block_bitmap;
 528	__u64 inode_bitmap;
 529	__u64 inode_table;
 530	__u32 blocks_count;
 531	__u16 reserved_blocks;
 532	__u16 unused;
 533	__u32 free_blocks_count;
 534};
 535
 536/* Indexes used to index group tables in ext4_new_group_data */
 537enum {
 538	BLOCK_BITMAP = 0,	/* block bitmap */
 539	INODE_BITMAP,		/* inode bitmap */
 540	INODE_TABLE,		/* inode tables */
 541	GROUP_TABLE_COUNT,
 542};
 543
 544/*
 545 * Flags used by ext4_map_blocks()
 546 */
 547	/* Allocate any needed blocks and/or convert an unitialized
 548	   extent to be an initialized ext4 */
 549#define EXT4_GET_BLOCKS_CREATE			0x0001
 550	/* Request the creation of an unitialized extent */
 551#define EXT4_GET_BLOCKS_UNINIT_EXT		0x0002
 552#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT	(EXT4_GET_BLOCKS_UNINIT_EXT|\
 553						 EXT4_GET_BLOCKS_CREATE)
 554	/* Caller is from the delayed allocation writeout path,
 555	   so set the magic i_delalloc_reserve_flag after taking the
 556	   inode allocation semaphore for */
 557#define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
 558	/* caller is from the direct IO path, request to creation of an
 559	unitialized extents if not allocated, split the uninitialized
 560	extent if blocks has been preallocated already*/
 561#define EXT4_GET_BLOCKS_PRE_IO			0x0008
 562#define EXT4_GET_BLOCKS_CONVERT			0x0010
 563#define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
 564					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
 565	/* Convert extent to initialized after IO complete */
 566#define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
 567					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
 568	/* Punch out blocks of an extent */
 569#define EXT4_GET_BLOCKS_PUNCH_OUT_EXT		0x0020
 
 570	/* Don't normalize allocation size (used for fallocate) */
 571#define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
 572	/* Request will not result in inode size update (user for fallocate) */
 573#define EXT4_GET_BLOCKS_KEEP_SIZE		0x0080
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 574
 575/*
 576 * Flags used by ext4_free_blocks
 577 */
 578#define EXT4_FREE_BLOCKS_METADATA	0x0001
 579#define EXT4_FREE_BLOCKS_FORGET		0x0002
 580#define EXT4_FREE_BLOCKS_VALIDATED	0x0004
 581#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE	0x0008
 582#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
 583#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
 584
 585/*
 586 * Flags used by ext4_discard_partial_page_buffers
 587 */
 588#define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED	0x0001
 589
 590/*
 591 * ioctl commands
 592 */
 593#define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
 594#define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
 595#define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
 596#define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
 597#define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
 598#define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
 599#define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
 600#define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
 601#define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
 602#define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
 603#define EXT4_IOC_MIGRATE		_IO('f', 9)
 604 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
 605 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
 606#define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
 607#define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
 608#define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 609
 610#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 611/*
 612 * ioctl commands in 32 bit emulation
 613 */
 614#define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
 615#define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
 616#define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
 617#define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
 618#define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
 619#define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
 620#define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
 621#define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
 622#define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
 623#define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
 624#endif
 625
 
 
 
 
 
 
 626/* Max physical block we can address w/o extents */
 627#define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
 628
 
 
 
 629/*
 630 * Structure of an inode on the disk
 631 */
 632struct ext4_inode {
 633	__le16	i_mode;		/* File mode */
 634	__le16	i_uid;		/* Low 16 bits of Owner Uid */
 635	__le32	i_size_lo;	/* Size in bytes */
 636	__le32	i_atime;	/* Access time */
 637	__le32	i_ctime;	/* Inode Change time */
 638	__le32	i_mtime;	/* Modification time */
 639	__le32	i_dtime;	/* Deletion Time */
 640	__le16	i_gid;		/* Low 16 bits of Group Id */
 641	__le16	i_links_count;	/* Links count */
 642	__le32	i_blocks_lo;	/* Blocks count */
 643	__le32	i_flags;	/* File flags */
 644	union {
 645		struct {
 646			__le32  l_i_version;
 647		} linux1;
 648		struct {
 649			__u32  h_i_translator;
 650		} hurd1;
 651		struct {
 652			__u32  m_i_reserved1;
 653		} masix1;
 654	} osd1;				/* OS dependent 1 */
 655	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
 656	__le32	i_generation;	/* File version (for NFS) */
 657	__le32	i_file_acl_lo;	/* File ACL */
 658	__le32	i_size_high;
 659	__le32	i_obso_faddr;	/* Obsoleted fragment address */
 660	union {
 661		struct {
 662			__le16	l_i_blocks_high; /* were l_i_reserved1 */
 663			__le16	l_i_file_acl_high;
 664			__le16	l_i_uid_high;	/* these 2 fields */
 665			__le16	l_i_gid_high;	/* were reserved2[0] */
 666			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
 667			__le16	l_i_reserved;
 668		} linux2;
 669		struct {
 670			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
 671			__u16	h_i_mode_high;
 672			__u16	h_i_uid_high;
 673			__u16	h_i_gid_high;
 674			__u32	h_i_author;
 675		} hurd2;
 676		struct {
 677			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
 678			__le16	m_i_file_acl_high;
 679			__u32	m_i_reserved2[2];
 680		} masix2;
 681	} osd2;				/* OS dependent 2 */
 682	__le16	i_extra_isize;
 683	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
 684	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
 685	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
 686	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
 687	__le32  i_crtime;       /* File Creation time */
 688	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
 689	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
 
 690};
 691
 692struct move_extent {
 693	__u32 reserved;		/* should be zero */
 694	__u32 donor_fd;		/* donor file descriptor */
 695	__u64 orig_start;	/* logical start offset in block for orig */
 696	__u64 donor_start;	/* logical start offset in block for donor */
 697	__u64 len;		/* block length to be moved */
 698	__u64 moved_len;	/* moved block length */
 699};
 700
 701#define EXT4_EPOCH_BITS 2
 702#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
 703#define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
 704
 705/*
 706 * Extended fields will fit into an inode if the filesystem was formatted
 707 * with large inodes (-I 256 or larger) and there are not currently any EAs
 708 * consuming all of the available space. For new inodes we always reserve
 709 * enough space for the kernel's known extended fields, but for inodes
 710 * created with an old kernel this might not have been the case. None of
 711 * the extended inode fields is critical for correct filesystem operation.
 712 * This macro checks if a certain field fits in the inode. Note that
 713 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
 714 */
 715#define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
 716	((offsetof(typeof(*ext4_inode), field) +	\
 717	  sizeof((ext4_inode)->field))			\
 718	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
 719	    (einode)->i_extra_isize))			\
 720
 721static inline __le32 ext4_encode_extra_time(struct timespec *time)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 722{
 723       return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
 724			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
 725                          ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
 726}
 727
 728static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
 
 729{
 730       if (sizeof(time->tv_sec) > 4)
 731	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
 732			       << 32;
 733       time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
 734}
 735
 736#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
 737do {									       \
 738	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
 739	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
 740		(raw_inode)->xtime ## _extra =				       \
 741				ext4_encode_extra_time(&(inode)->xtime);       \
 
 
 
 742} while (0)
 743
 744#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
 745do {									       \
 746	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
 747		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
 748	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
 749		(raw_inode)->xtime ## _extra =				       \
 750				ext4_encode_extra_time(&(einode)->xtime);      \
 751} while (0)
 752
 753#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
 754do {									       \
 755	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
 756	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
 757		ext4_decode_extra_time(&(inode)->xtime,			       \
 758				       raw_inode->xtime ## _extra);	       \
 759	else								       \
 760		(inode)->xtime.tv_nsec = 0;				       \
 
 761} while (0)
 762
 
 763#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
 764do {									       \
 765	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
 766		(einode)->xtime.tv_sec = 				       \
 767			(signed)le32_to_cpu((raw_inode)->xtime);	       \
 
 
 768	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
 769		ext4_decode_extra_time(&(einode)->xtime,		       \
 770				       raw_inode->xtime ## _extra);	       \
 771	else								       \
 772		(einode)->xtime.tv_nsec = 0;				       \
 773} while (0)
 774
 775#define i_disk_version osd1.linux1.l_i_version
 776
 777#if defined(__KERNEL__) || defined(__linux__)
 778#define i_reserved1	osd1.linux1.l_i_reserved1
 779#define i_file_acl_high	osd2.linux2.l_i_file_acl_high
 780#define i_blocks_high	osd2.linux2.l_i_blocks_high
 781#define i_uid_low	i_uid
 782#define i_gid_low	i_gid
 783#define i_uid_high	osd2.linux2.l_i_uid_high
 784#define i_gid_high	osd2.linux2.l_i_gid_high
 785#define i_checksum_lo	osd2.linux2.l_i_checksum_lo
 786
 787#elif defined(__GNU__)
 788
 789#define i_translator	osd1.hurd1.h_i_translator
 790#define i_uid_high	osd2.hurd2.h_i_uid_high
 791#define i_gid_high	osd2.hurd2.h_i_gid_high
 792#define i_author	osd2.hurd2.h_i_author
 793
 794#elif defined(__masix__)
 795
 796#define i_reserved1	osd1.masix1.m_i_reserved1
 797#define i_file_acl_high	osd2.masix2.m_i_file_acl_high
 798#define i_reserved2	osd2.masix2.m_i_reserved2
 799
 800#endif /* defined(__KERNEL__) || defined(__linux__) */
 801
 
 
 802/*
 803 * storage for cached extent
 804 * If ec_len == 0, then the cache is invalid.
 805 * If ec_start == 0, then the cache represents a gap (null mapping)
 806 */
 807struct ext4_ext_cache {
 808	ext4_fsblk_t	ec_start;
 809	ext4_lblk_t	ec_block;
 810	__u32		ec_len; /* must be 32bit to return holes */
 
 
 
 
 
 
 
 
 
 
 
 811};
 812
 
 813/*
 814 * fourth extended file system inode data in memory
 815 */
 816struct ext4_inode_info {
 817	__le32	i_data[15];	/* unconverted */
 818	__u32	i_dtime;
 819	ext4_fsblk_t	i_file_acl;
 820
 821	/*
 822	 * i_block_group is the number of the block group which contains
 823	 * this file's inode.  Constant across the lifetime of the inode,
 824	 * it is ued for making block allocation decisions - we try to
 825	 * place a file's data blocks near its inode block, and new inodes
 826	 * near to their parent directory's inode.
 827	 */
 828	ext4_group_t	i_block_group;
 829	ext4_lblk_t	i_dir_start_lookup;
 830#if (BITS_PER_LONG < 64)
 831	unsigned long	i_state_flags;		/* Dynamic state flags */
 832#endif
 833	unsigned long	i_flags;
 834
 835#ifdef CONFIG_EXT4_FS_XATTR
 836	/*
 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;
 844#endif
 845
 846	struct list_head i_orphan;	/* unlinked but open inodes */
 847
 848	/*
 849	 * i_disksize keeps track of what the inode size is ON DISK, not
 850	 * in memory.  During truncate, i_size is set to the new size by
 851	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
 852	 * set i_disksize to 0 until the truncate is actually under way.
 853	 *
 854	 * The intent is that i_disksize always represents the blocks which
 855	 * are used by this file.  This allows recovery to restart truncate
 856	 * on orphans if we crash during truncate.  We actually write i_disksize
 857	 * into the on-disk inode when writing inodes out, instead of i_size.
 858	 *
 859	 * The only time when i_disksize and i_size may be different is when
 860	 * a truncate is in progress.  The only things which change i_disksize
 861	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
 862	 */
 863	loff_t	i_disksize;
 864
 865	/*
 866	 * i_data_sem is for serialising ext4_truncate() against
 867	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
 868	 * data tree are chopped off during truncate. We can't do that in
 869	 * ext4 because whenever we perform intermediate commits during
 870	 * truncate, the inode and all the metadata blocks *must* be in a
 871	 * consistent state which allows truncation of the orphans to restart
 872	 * during recovery.  Hence we must fix the get_block-vs-truncate race
 873	 * by other means, so we have i_data_sem.
 874	 */
 875	struct rw_semaphore i_data_sem;
 
 
 
 
 
 
 
 
 
 876	struct inode vfs_inode;
 877	struct jbd2_inode *jinode;
 878
 879	struct ext4_ext_cache i_cached_extent;
 
 880	/*
 881	 * File creation time. Its function is same as that of
 882	 * struct timespec i_{a,c,m}time in the generic inode.
 883	 */
 884	struct timespec i_crtime;
 885
 886	/* mballoc */
 
 887	struct list_head i_prealloc_list;
 888	spinlock_t i_prealloc_lock;
 889
 
 
 
 
 
 
 
 
 
 
 890	/* ialloc */
 891	ext4_group_t	i_last_alloc_group;
 892
 893	/* allocation reservation info for delalloc */
 894	/* In case of bigalloc, these refer to clusters rather than blocks */
 895	unsigned int i_reserved_data_blocks;
 896	unsigned int i_reserved_meta_blocks;
 897	unsigned int i_allocated_meta_blocks;
 898	ext4_lblk_t i_da_metadata_calc_last_lblock;
 899	int i_da_metadata_calc_len;
 900
 901	/* on-disk additional length */
 902	__u16 i_extra_isize;
 903
 
 
 
 
 904#ifdef CONFIG_QUOTA
 905	/* quota space reservation, managed internally by quota code */
 906	qsize_t i_reserved_quota;
 907#endif
 908
 909	/* completed IOs that might need unwritten extents handling */
 910	struct list_head i_completed_io_list;
 911	spinlock_t i_completed_io_lock;
 912	atomic_t i_ioend_count;	/* Number of outstanding io_end structs */
 913	/* current io_end structure for async DIO write*/
 914	ext4_io_end_t *cur_aio_dio;
 915	atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
 
 
 
 916
 917	spinlock_t i_block_reservation_lock;
 918
 919	/*
 920	 * Transactions that contain inode's metadata needed to complete
 921	 * fsync and fdatasync, respectively.
 922	 */
 923	tid_t i_sync_tid;
 924	tid_t i_datasync_tid;
 925
 
 
 
 
 926	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
 927	__u32 i_csum_seed;
 
 
 928};
 929
 930/*
 931 * File system states
 932 */
 933#define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
 934#define	EXT4_ERROR_FS			0x0002	/* Errors detected */
 935#define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
 936
 937/*
 938 * Misc. filesystem flags
 939 */
 940#define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
 941#define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
 942#define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
 943
 944/*
 945 * Mount flags
 946 */
 
 947#define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
 948#define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
 949#define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
 950#define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
 951#define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
 952#define EXT4_MOUNT_ERRORS_MASK		0x00070
 953#define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
 954#define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
 
 
 
 
 
 955#define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
 956#define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
 957#define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
 958#define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
 959#define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
 960#define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
 961#define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
 962#define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
 963#define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
 964#define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
 965#define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
 966#define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
 967#define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
 
 
 
 
 
 
 968#define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
 969#define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
 970#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
 971#define EXT4_MOUNT_MBLK_IO_SUBMIT	0x4000000 /* multi-block io submits */
 
 972#define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
 973#define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
 974#define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
 975#define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
 976#define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
 977
 
 
 
 
 
 978#define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
 979						      specified delalloc */
 
 
 
 
 
 
 
 
 
 
 980
 981#define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
 982						~EXT4_MOUNT_##opt
 983#define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
 984						EXT4_MOUNT_##opt
 985#define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
 986					 EXT4_MOUNT_##opt)
 987
 988#define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
 989						~EXT4_MOUNT2_##opt
 990#define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
 991						EXT4_MOUNT2_##opt
 992#define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
 993					 EXT4_MOUNT2_##opt)
 994
 995#define ext4_test_and_set_bit		__test_and_set_bit_le
 996#define ext4_set_bit			__set_bit_le
 997#define ext4_set_bit_atomic		ext2_set_bit_atomic
 998#define ext4_test_and_clear_bit		__test_and_clear_bit_le
 999#define ext4_clear_bit			__clear_bit_le
1000#define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1001#define ext4_test_bit			test_bit_le
1002#define ext4_find_next_zero_bit		find_next_zero_bit_le
1003#define ext4_find_next_bit		find_next_bit_le
1004
1005extern void ext4_set_bits(void *bm, int cur, int len);
1006
1007/*
1008 * Maximal mount counts between two filesystem checks
1009 */
1010#define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1011#define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1012
1013/*
1014 * Behaviour when detecting errors
1015 */
1016#define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1017#define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1018#define EXT4_ERRORS_PANIC		3	/* Panic */
1019#define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1020
1021/* Metadata checksum algorithm codes */
1022#define EXT4_CRC32C_CHKSUM		1
1023
1024/*
1025 * Structure of the super block
1026 */
1027struct ext4_super_block {
1028/*00*/	__le32	s_inodes_count;		/* Inodes count */
1029	__le32	s_blocks_count_lo;	/* Blocks count */
1030	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1031	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1032/*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1033	__le32	s_first_data_block;	/* First Data Block */
1034	__le32	s_log_block_size;	/* Block size */
1035	__le32	s_log_cluster_size;	/* Allocation cluster size */
1036/*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1037	__le32	s_clusters_per_group;	/* # Clusters per group */
1038	__le32	s_inodes_per_group;	/* # Inodes per group */
1039	__le32	s_mtime;		/* Mount time */
1040/*30*/	__le32	s_wtime;		/* Write time */
1041	__le16	s_mnt_count;		/* Mount count */
1042	__le16	s_max_mnt_count;	/* Maximal mount count */
1043	__le16	s_magic;		/* Magic signature */
1044	__le16	s_state;		/* File system state */
1045	__le16	s_errors;		/* Behaviour when detecting errors */
1046	__le16	s_minor_rev_level;	/* minor revision level */
1047/*40*/	__le32	s_lastcheck;		/* time of last check */
1048	__le32	s_checkinterval;	/* max. time between checks */
1049	__le32	s_creator_os;		/* OS */
1050	__le32	s_rev_level;		/* Revision level */
1051/*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1052	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1053	/*
1054	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1055	 *
1056	 * Note: the difference between the compatible feature set and
1057	 * the incompatible feature set is that if there is a bit set
1058	 * in the incompatible feature set that the kernel doesn't
1059	 * know about, it should refuse to mount the filesystem.
1060	 *
1061	 * e2fsck's requirements are more strict; if it doesn't know
1062	 * about a feature in either the compatible or incompatible
1063	 * feature set, it must abort and not try to meddle with
1064	 * things it doesn't understand...
1065	 */
1066	__le32	s_first_ino;		/* First non-reserved inode */
1067	__le16  s_inode_size;		/* size of inode structure */
1068	__le16	s_block_group_nr;	/* block group # of this superblock */
1069	__le32	s_feature_compat;	/* compatible feature set */
1070/*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1071	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1072/*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1073/*78*/	char	s_volume_name[16];	/* volume name */
1074/*88*/	char	s_last_mounted[64];	/* directory where last mounted */
1075/*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1076	/*
1077	 * Performance hints.  Directory preallocation should only
1078	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1079	 */
1080	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1081	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1082	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1083	/*
1084	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1085	 */
1086/*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1087/*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1088	__le32	s_journal_dev;		/* device number of journal file */
1089	__le32	s_last_orphan;		/* start of list of inodes to delete */
1090	__le32	s_hash_seed[4];		/* HTREE hash seed */
1091	__u8	s_def_hash_version;	/* Default hash version to use */
1092	__u8	s_jnl_backup_type;
1093	__le16  s_desc_size;		/* size of group descriptor */
1094/*100*/	__le32	s_default_mount_opts;
1095	__le32	s_first_meta_bg;	/* First metablock block group */
1096	__le32	s_mkfs_time;		/* When the filesystem was created */
1097	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1098	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1099/*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1100	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1101	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1102	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1103	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1104	__le32	s_flags;		/* Miscellaneous flags */
1105	__le16  s_raid_stride;		/* RAID stride */
1106	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1107	__le64  s_mmp_block;            /* Block for multi-mount protection */
1108	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1109	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1110	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1111	__le16  s_reserved_pad;
 
1112	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1113	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1114	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1115	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1116					      snapshot's future use */
1117	__le32	s_snapshot_list;	/* inode number of the head of the
1118					   on-disk snapshot list */
1119#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1120	__le32	s_error_count;		/* number of fs errors */
1121	__le32	s_first_error_time;	/* first time an error happened */
1122	__le32	s_first_error_ino;	/* inode involved in first error */
1123	__le64	s_first_error_block;	/* block involved of first error */
1124	__u8	s_first_error_func[32];	/* function where the error happened */
1125	__le32	s_first_error_line;	/* line number where error happened */
1126	__le32	s_last_error_time;	/* most recent time of an error */
1127	__le32	s_last_error_ino;	/* inode involved in last error */
1128	__le32	s_last_error_line;	/* line number where error happened */
1129	__le64	s_last_error_block;	/* block involved of last error */
1130	__u8	s_last_error_func[32];	/* function where the error happened */
1131#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1132	__u8	s_mount_opts[64];
1133	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1134	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1135	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1136	__le32	s_reserved[108];	/* Padding to the end of the block */
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1137	__le32	s_checksum;		/* crc32c(superblock) */
1138};
1139
1140#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1141
1142#ifdef __KERNEL__
1143
1144/*
1145 * run-time mount flags
1146 */
1147#define EXT4_MF_MNTDIR_SAMPLED	0x0001
1148#define EXT4_MF_FS_ABORTED	0x0002	/* Fatal error detected */
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1149
1150/*
1151 * fourth extended-fs super-block data in memory
1152 */
1153struct ext4_sb_info {
1154	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1155	unsigned long s_inodes_per_block;/* Number of inodes per block */
1156	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1157	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1158	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1159	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1160	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1161	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1162	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1163	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1164	unsigned long s_overhead;  /* # of fs overhead clusters */
1165	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1166	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1167	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1168	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1169	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1170	struct buffer_head **s_group_desc;
1171	unsigned int s_mount_opt;
1172	unsigned int s_mount_opt2;
1173	unsigned int s_mount_flags;
1174	unsigned int s_def_mount_opt;
1175	ext4_fsblk_t s_sb_block;
 
1176	kuid_t s_resuid;
1177	kgid_t s_resgid;
1178	unsigned short s_mount_state;
1179	unsigned short s_pad;
1180	int s_addr_per_block_bits;
1181	int s_desc_per_block_bits;
1182	int s_inode_size;
1183	int s_first_ino;
1184	unsigned int s_inode_readahead_blks;
1185	unsigned int s_inode_goal;
1186	spinlock_t s_next_gen_lock;
1187	u32 s_next_generation;
1188	u32 s_hash_seed[4];
1189	int s_def_hash_version;
1190	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1191	struct percpu_counter s_freeclusters_counter;
1192	struct percpu_counter s_freeinodes_counter;
1193	struct percpu_counter s_dirs_counter;
1194	struct percpu_counter s_dirtyclusters_counter;
1195	struct blockgroup_lock *s_blockgroup_lock;
1196	struct proc_dir_entry *s_proc;
1197	struct kobject s_kobj;
1198	struct completion s_kobj_unregister;
1199	struct super_block *s_sb;
 
 
 
 
1200
1201	/* Journaling */
1202	struct journal_s *s_journal;
1203	struct list_head s_orphan;
1204	struct mutex s_orphan_lock;
1205	unsigned long s_resize_flags;		/* Flags indicating if there
1206						   is a resizer */
1207	unsigned long s_commit_interval;
1208	u32 s_max_batch_time;
1209	u32 s_min_batch_time;
1210	struct block_device *journal_bdev;
1211#ifdef CONFIG_QUOTA
1212	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
 
1213	int s_jquota_fmt;			/* Format of quota to use */
1214#endif
1215	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1216	struct rb_root system_blks;
1217
1218#ifdef EXTENTS_STATS
1219	/* ext4 extents stats */
1220	unsigned long s_ext_min;
1221	unsigned long s_ext_max;
1222	unsigned long s_depth_max;
1223	spinlock_t s_ext_stats_lock;
1224	unsigned long s_ext_blocks;
1225	unsigned long s_ext_extents;
1226#endif
1227
1228	/* for buddy allocator */
1229	struct ext4_group_info ***s_group_info;
1230	struct inode *s_buddy_cache;
1231	spinlock_t s_md_lock;
1232	unsigned short *s_mb_offsets;
1233	unsigned int *s_mb_maxs;
 
 
 
 
1234
1235	/* tunables */
1236	unsigned long s_stripe;
1237	unsigned int s_mb_stream_request;
1238	unsigned int s_mb_max_to_scan;
1239	unsigned int s_mb_min_to_scan;
1240	unsigned int s_mb_stats;
1241	unsigned int s_mb_order2_reqs;
1242	unsigned int s_mb_group_prealloc;
1243	unsigned int s_max_writeback_mb_bump;
 
1244	/* where last allocation was done - for stream allocation */
1245	unsigned long s_mb_last_group;
1246	unsigned long s_mb_last_start;
 
 
1247
1248	/* stats for buddy allocator */
1249	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1250	atomic_t s_bal_success;	/* we found long enough chunks */
1251	atomic_t s_bal_allocated;	/* in blocks */
1252	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1253	atomic_t s_bal_goals;	/* goal hits */
1254	atomic_t s_bal_breaks;	/* too long searches */
1255	atomic_t s_bal_2orders;	/* 2^order hits */
1256	spinlock_t s_bal_lock;
1257	unsigned long s_mb_buddies_generated;
1258	unsigned long long s_mb_generation_time;
1259	atomic_t s_mb_lost_chunks;
1260	atomic_t s_mb_preallocated;
1261	atomic_t s_mb_discarded;
1262	atomic_t s_lock_busy;
1263
1264	/* locality groups */
1265	struct ext4_locality_group __percpu *s_locality_groups;
1266
1267	/* for write statistics */
1268	unsigned long s_sectors_written_start;
1269	u64 s_kbytes_written;
1270
 
 
 
1271	unsigned int s_log_groups_per_flex;
1272	struct flex_groups *s_flex_groups;
 
1273
1274	/* workqueue for dio unwritten */
1275	struct workqueue_struct *dio_unwritten_wq;
1276
1277	/* timer for periodic error stats printing */
1278	struct timer_list s_err_report;
1279
1280	/* Lazy inode table initialization info */
1281	struct ext4_li_request *s_li_request;
1282	/* Wait multiplier for lazy initialization thread */
1283	unsigned int s_li_wait_mult;
1284
1285	/* Kernel thread for multiple mount protection */
1286	struct task_struct *s_mmp_tsk;
1287
1288	/* record the last minlen when FITRIM is called. */
1289	atomic_t s_last_trim_minblks;
1290
1291	/* Reference to checksum algorithm driver via cryptoapi */
1292	struct crypto_shash *s_chksum_driver;
1293
1294	/* Precomputed FS UUID checksum for seeding other checksums */
1295	__u32 s_csum_seed;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1296};
1297
1298static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1299{
1300	return sb->s_fs_info;
1301}
1302static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1303{
1304	return container_of(inode, struct ext4_inode_info, vfs_inode);
1305}
1306
1307static inline struct timespec ext4_current_time(struct inode *inode)
1308{
1309	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1310		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1311}
1312
1313static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1314{
1315	return ino == EXT4_ROOT_INO ||
1316		ino == EXT4_JOURNAL_INO ||
1317		ino == EXT4_RESIZE_INO ||
1318		(ino >= EXT4_FIRST_INO(sb) &&
1319		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1320}
1321
1322static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1323					      struct ext4_io_end *io_end)
1324{
1325	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1326		io_end->flag |= EXT4_IO_END_UNWRITTEN;
1327		atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1328	}
 
 
 
 
 
 
 
 
 
 
1329}
1330
1331/*
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1332 * Inode dynamic state flags
1333 */
1334enum {
1335	EXT4_STATE_JDATA,		/* journaled data exists */
1336	EXT4_STATE_NEW,			/* inode is newly created */
1337	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1338	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1339	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1340	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1341	EXT4_STATE_DIO_UNWRITTEN,	/* need convert on dio done*/
1342	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1343	EXT4_STATE_DELALLOC_RESERVED,	/* blks already reserved for delalloc */
 
 
 
1344};
1345
1346#define EXT4_INODE_BIT_FNS(name, field, offset)				\
1347static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1348{									\
1349	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1350}									\
1351static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1352{									\
1353	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1354}									\
1355static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1356{									\
1357	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1358}
1359
 
 
 
 
 
1360EXT4_INODE_BIT_FNS(flag, flags, 0)
 
 
 
 
 
 
1361#if (BITS_PER_LONG < 64)
1362EXT4_INODE_BIT_FNS(state, state_flags, 0)
1363
1364static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1365{
1366	(ei)->i_state_flags = 0;
1367}
1368#else
1369EXT4_INODE_BIT_FNS(state, flags, 32)
1370
1371static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1372{
1373	/* We depend on the fact that callers will set i_flags */
1374}
1375#endif
1376#else
1377/* Assume that user mode programs are passing in an ext4fs superblock, not
1378 * a kernel struct super_block.  This will allow us to call the feature-test
1379 * macros from user land. */
1380#define EXT4_SB(sb)	(sb)
1381#endif
1382
 
 
 
 
 
 
1383#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1384
1385/*
1386 * Codes for operating systems
1387 */
1388#define EXT4_OS_LINUX		0
1389#define EXT4_OS_HURD		1
1390#define EXT4_OS_MASIX		2
1391#define EXT4_OS_FREEBSD		3
1392#define EXT4_OS_LITES		4
1393
1394/*
1395 * Revision levels
1396 */
1397#define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1398#define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1399
1400#define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1401#define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1402
1403#define EXT4_GOOD_OLD_INODE_SIZE 128
1404
 
 
 
 
1405/*
1406 * Feature set definitions
1407 */
1408
1409#define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1410	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1411#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1412	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1413#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1414	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1415#define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1416	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1417#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1418	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1419#define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1420	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1421#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1422	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1423#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1424	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1425#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1426	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1427
1428#define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1429#define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1430#define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1431#define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1432#define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1433#define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
 
 
1434
1435#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1436#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1437#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1438#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1439#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1440#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1441#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1442#define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1443#define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1444/*
1445 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1446 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1447 * all other data structures' checksums.  However, the METADATA_CSUM and
1448 * GDT_CSUM bits are mutually exclusive.
1449 */
1450#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
 
 
 
1451
1452#define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1453#define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1454#define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1455#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1456#define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1457#define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1458#define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1459#define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1460#define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1461#define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1462#define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1463#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM	0x2000 /* use crc32c for bg */
1464#define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1465#define EXT4_FEATURE_INCOMPAT_INLINEDATA	0x8000 /* data in inode */
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1466
1467#define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1468#define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1469					 EXT4_FEATURE_INCOMPAT_META_BG)
1470#define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1471					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1472					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1473
1474#define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1475#define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1476					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1477					 EXT4_FEATURE_INCOMPAT_META_BG)
1478#define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1479					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1480					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1481
1482#define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1483#define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1484					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1485					 EXT4_FEATURE_INCOMPAT_META_BG| \
1486					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1487					 EXT4_FEATURE_INCOMPAT_64BIT| \
1488					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1489					 EXT4_FEATURE_INCOMPAT_MMP)
 
 
 
 
 
 
1490#define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1491					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1492					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1493					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1494					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1495					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1496					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1497					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1498					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1499
1500/*
1501 * Default values for user and/or group using reserved blocks
1502 */
1503#define	EXT4_DEF_RESUID		0
1504#define	EXT4_DEF_RESGID		0
1505
 
 
 
 
 
1506#define EXT4_DEF_INODE_READAHEAD_BLKS	32
1507
1508/*
1509 * Default mount options
1510 */
1511#define EXT4_DEFM_DEBUG		0x0001
1512#define EXT4_DEFM_BSDGROUPS	0x0002
1513#define EXT4_DEFM_XATTR_USER	0x0004
1514#define EXT4_DEFM_ACL		0x0008
1515#define EXT4_DEFM_UID16		0x0010
1516#define EXT4_DEFM_JMODE		0x0060
1517#define EXT4_DEFM_JMODE_DATA	0x0020
1518#define EXT4_DEFM_JMODE_ORDERED	0x0040
1519#define EXT4_DEFM_JMODE_WBACK	0x0060
1520#define EXT4_DEFM_NOBARRIER	0x0100
1521#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1522#define EXT4_DEFM_DISCARD	0x0400
1523#define EXT4_DEFM_NODELALLOC	0x0800
1524
1525/*
1526 * Default journal batch times
1527 */
1528#define EXT4_DEF_MIN_BATCH_TIME	0
1529#define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1530
1531/*
1532 * Minimum number of groups in a flexgroup before we separate out
1533 * directories into the first block group of a flexgroup
1534 */
1535#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1536
1537/*
1538 * Structure of a directory entry
1539 */
1540#define EXT4_NAME_LEN 255
1541
1542struct ext4_dir_entry {
1543	__le32	inode;			/* Inode number */
1544	__le16	rec_len;		/* Directory entry length */
1545	__le16	name_len;		/* Name length */
1546	char	name[EXT4_NAME_LEN];	/* File name */
1547};
1548
1549/*
1550 * The new version of the directory entry.  Since EXT4 structures are
1551 * stored in intel byte order, and the name_len field could never be
1552 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1553 * file_type field.
1554 */
1555struct ext4_dir_entry_2 {
1556	__le32	inode;			/* Inode number */
1557	__le16	rec_len;		/* Directory entry length */
1558	__u8	name_len;		/* Name length */
1559	__u8	file_type;
1560	char	name[EXT4_NAME_LEN];	/* File name */
1561};
1562
1563/*
1564 * This is a bogus directory entry at the end of each leaf block that
1565 * records checksums.
1566 */
1567struct ext4_dir_entry_tail {
1568	__le32	det_reserved_zero1;	/* Pretend to be unused */
1569	__le16	det_rec_len;		/* 12 */
1570	__u8	det_reserved_zero2;	/* Zero name length */
1571	__u8	det_reserved_ft;	/* 0xDE, fake file type */
1572	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
1573};
1574
 
 
 
 
 
1575/*
1576 * Ext4 directory file types.  Only the low 3 bits are used.  The
1577 * other bits are reserved for now.
1578 */
1579#define EXT4_FT_UNKNOWN		0
1580#define EXT4_FT_REG_FILE	1
1581#define EXT4_FT_DIR		2
1582#define EXT4_FT_CHRDEV		3
1583#define EXT4_FT_BLKDEV		4
1584#define EXT4_FT_FIFO		5
1585#define EXT4_FT_SOCK		6
1586#define EXT4_FT_SYMLINK		7
1587
1588#define EXT4_FT_MAX		8
1589
1590#define EXT4_FT_DIR_CSUM	0xDE
1591
1592/*
1593 * EXT4_DIR_PAD defines the directory entries boundaries
1594 *
1595 * NOTE: It must be a multiple of 4
1596 */
1597#define EXT4_DIR_PAD			4
1598#define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1599#define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1600					 ~EXT4_DIR_ROUND)
1601#define EXT4_MAX_REC_LEN		((1<<16)-1)
1602
1603/*
1604 * If we ever get support for fs block sizes > page_size, we'll need
1605 * to remove the #if statements in the next two functions...
1606 */
1607static inline unsigned int
1608ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1609{
1610	unsigned len = le16_to_cpu(dlen);
1611
1612#if (PAGE_CACHE_SIZE >= 65536)
1613	if (len == EXT4_MAX_REC_LEN || len == 0)
1614		return blocksize;
1615	return (len & 65532) | ((len & 3) << 16);
1616#else
1617	return len;
1618#endif
1619}
1620
1621static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1622{
1623	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1624		BUG();
1625#if (PAGE_CACHE_SIZE >= 65536)
1626	if (len < 65536)
1627		return cpu_to_le16(len);
1628	if (len == blocksize) {
1629		if (blocksize == 65536)
1630			return cpu_to_le16(EXT4_MAX_REC_LEN);
1631		else
1632			return cpu_to_le16(0);
1633	}
1634	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1635#else
1636	return cpu_to_le16(len);
1637#endif
1638}
1639
1640/*
1641 * Hash Tree Directory indexing
1642 * (c) Daniel Phillips, 2001
1643 */
1644
1645#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1646				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1647		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1648#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
 
1649#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1650
1651/* Legal values for the dx_root hash_version field: */
1652
1653#define DX_HASH_LEGACY		0
1654#define DX_HASH_HALF_MD4	1
1655#define DX_HASH_TEA		2
1656#define DX_HASH_LEGACY_UNSIGNED	3
1657#define DX_HASH_HALF_MD4_UNSIGNED	4
1658#define DX_HASH_TEA_UNSIGNED		5
1659
1660static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1661			      const void *address, unsigned int length)
1662{
1663	struct {
1664		struct shash_desc shash;
1665		char ctx[crypto_shash_descsize(sbi->s_chksum_driver)];
1666	} desc;
1667	int err;
 
1668
1669	desc.shash.tfm = sbi->s_chksum_driver;
1670	desc.shash.flags = 0;
1671	*(u32 *)desc.ctx = crc;
1672
1673	err = crypto_shash_update(&desc.shash, address, length);
1674	BUG_ON(err);
1675
1676	return *(u32 *)desc.ctx;
1677}
1678
1679#ifdef __KERNEL__
1680
1681/* hash info structure used by the directory hash */
1682struct dx_hash_info
1683{
1684	u32		hash;
1685	u32		minor_hash;
1686	int		hash_version;
1687	u32		*seed;
1688};
1689
1690
1691/* 32 and 64 bit signed EOF for dx directories */
1692#define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
1693#define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
1694
1695
1696/*
1697 * Control parameters used by ext4_htree_next_block
1698 */
1699#define HASH_NB_ALWAYS		1
1700
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1701
1702/*
1703 * Describe an inode's exact location on disk and in memory
1704 */
1705struct ext4_iloc
1706{
1707	struct buffer_head *bh;
1708	unsigned long offset;
1709	ext4_group_t block_group;
1710};
1711
1712static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1713{
1714	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1715}
1716
 
 
 
 
 
 
1717/*
1718 * This structure is stuffed into the struct file's private_data field
1719 * for directories.  It is where we put information so that we can do
1720 * readdir operations in hash tree order.
1721 */
1722struct dir_private_info {
1723	struct rb_root	root;
1724	struct rb_node	*curr_node;
1725	struct fname	*extra_fname;
1726	loff_t		last_pos;
1727	__u32		curr_hash;
1728	__u32		curr_minor_hash;
1729	__u32		next_hash;
1730};
1731
1732/* calculate the first block number of the group */
1733static inline ext4_fsblk_t
1734ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1735{
1736	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1737		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1738}
1739
1740/*
1741 * Special error return code only used by dx_probe() and its callers.
1742 */
1743#define ERR_BAD_DX_DIR	-75000
1744
1745void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1746			ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
 
 
 
 
 
 
 
1747
1748/*
1749 * Timeout and state flag for lazy initialization inode thread.
1750 */
1751#define EXT4_DEF_LI_WAIT_MULT			10
1752#define EXT4_DEF_LI_MAX_START_DELAY		5
1753#define EXT4_LAZYINIT_QUIT			0x0001
1754#define EXT4_LAZYINIT_RUNNING			0x0002
1755
1756/*
1757 * Lazy inode table initialization info
1758 */
1759struct ext4_lazy_init {
1760	unsigned long		li_state;
1761	struct list_head	li_request_list;
1762	struct mutex		li_list_mtx;
1763};
1764
 
 
 
 
 
1765struct ext4_li_request {
1766	struct super_block	*lr_super;
1767	struct ext4_sb_info	*lr_sbi;
 
1768	ext4_group_t		lr_next_group;
1769	struct list_head	lr_request;
1770	unsigned long		lr_next_sched;
1771	unsigned long		lr_timeout;
1772};
1773
1774struct ext4_features {
1775	struct kobject f_kobj;
1776	struct completion f_kobj_unregister;
1777};
1778
1779/*
1780 * This structure will be used for multiple mount protection. It will be
1781 * written into the block number saved in the s_mmp_block field in the
1782 * superblock. Programs that check MMP should assume that if
1783 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1784 * to use the filesystem, regardless of how old the timestamp is.
1785 */
1786#define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1787#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1788#define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1789#define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1790
1791struct mmp_struct {
1792	__le32	mmp_magic;		/* Magic number for MMP */
1793	__le32	mmp_seq;		/* Sequence no. updated periodically */
1794
1795	/*
1796	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1797	 * purposes and do not affect the correctness of the algorithm
1798	 */
1799	__le64	mmp_time;		/* Time last updated */
1800	char	mmp_nodename[64];	/* Node which last updated MMP block */
1801	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
1802
1803	/*
1804	 * mmp_check_interval is used to verify if the MMP block has been
1805	 * updated on the block device. The value is updated based on the
1806	 * maximum time to write the MMP block during an update cycle.
1807	 */
1808	__le16	mmp_check_interval;
1809
1810	__le16	mmp_pad1;
1811	__le32	mmp_pad2[226];
1812	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
1813};
1814
1815/* arguments passed to the mmp thread */
1816struct mmpd_data {
1817	struct buffer_head *bh; /* bh from initial read_mmp_block() */
1818	struct super_block *sb;  /* super block of the fs */
1819};
1820
1821/*
1822 * Check interval multiplier
1823 * The MMP block is written every update interval and initially checked every
1824 * update interval x the multiplier (the value is then adapted based on the
1825 * write latency). The reason is that writes can be delayed under load and we
1826 * don't want readers to incorrectly assume that the filesystem is no longer
1827 * in use.
1828 */
1829#define EXT4_MMP_CHECK_MULT		2UL
1830
1831/*
1832 * Minimum interval for MMP checking in seconds.
1833 */
1834#define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
1835
1836/*
1837 * Maximum interval for MMP checking in seconds.
1838 */
1839#define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
1840
1841/*
1842 * Function prototypes
1843 */
1844
1845/*
1846 * Ok, these declarations are also in <linux/kernel.h> but none of the
1847 * ext4 source programs needs to include it so they are duplicated here.
1848 */
1849# define NORET_TYPE	/**/
1850# define ATTRIB_NORET	__attribute__((noreturn))
1851# define NORET_AND	noreturn,
1852
1853/* bitmap.c */
1854extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1855void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1856				struct ext4_group_desc *gdp,
1857				struct buffer_head *bh, int sz);
1858int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1859				  struct ext4_group_desc *gdp,
1860				  struct buffer_head *bh, int sz);
1861void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1862				struct ext4_group_desc *gdp,
1863				struct buffer_head *bh, int sz);
1864int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1865				  struct ext4_group_desc *gdp,
1866				  struct buffer_head *bh, int sz);
1867
1868/* balloc.c */
1869extern void ext4_validate_block_bitmap(struct super_block *sb,
1870				       struct ext4_group_desc *desc,
1871				       unsigned int block_group,
1872				       struct buffer_head *bh);
 
 
 
1873extern unsigned int ext4_block_group(struct super_block *sb,
1874			ext4_fsblk_t blocknr);
1875extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1876			ext4_fsblk_t blocknr);
1877extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1878extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1879			ext4_group_t group);
1880extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1881					 ext4_fsblk_t goal,
1882					 unsigned int flags,
1883					 unsigned long *count,
1884					 int *errp);
1885extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1886				    s64 nclusters, unsigned int flags);
1887extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1888extern void ext4_check_blocks_bitmap(struct super_block *);
1889extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1890						    ext4_group_t block_group,
1891						    struct buffer_head ** bh);
1892extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1893
1894extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1895						ext4_group_t block_group);
 
1896extern int ext4_wait_block_bitmap(struct super_block *sb,
1897				  ext4_group_t block_group,
1898				  struct buffer_head *bh);
1899extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1900						  ext4_group_t block_group);
1901extern void ext4_init_block_bitmap(struct super_block *sb,
1902				   struct buffer_head *bh,
1903				   ext4_group_t group,
1904				   struct ext4_group_desc *desc);
1905extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1906					      ext4_group_t block_group,
1907					      struct ext4_group_desc *gdp);
1908extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1909					   ext4_group_t block_group,
1910					   struct ext4_group_desc *gdp);
1911ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1912
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1913/* dir.c */
1914extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1915				  struct file *,
1916				  struct ext4_dir_entry_2 *,
1917				  struct buffer_head *, unsigned int);
1918#define ext4_check_dir_entry(dir, filp, de, bh, offset)			\
 
1919	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1920					(de), (bh), (offset)))
1921extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1922				    __u32 minor_hash,
1923				    struct ext4_dir_entry_2 *dirent);
 
1924extern void ext4_htree_free_dir_info(struct dir_private_info *p);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1925
1926/* fsync.c */
1927extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
1928extern int ext4_flush_completed_IO(struct inode *);
1929
1930/* hash.c */
1931extern int ext4fs_dirhash(const char *name, int len, struct
1932			  dx_hash_info *hinfo);
1933
1934/* ialloc.c */
1935extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t,
1936				    const struct qstr *qstr, __u32 goal,
1937				    uid_t *owner);
 
 
 
 
 
 
 
 
 
 
 
 
1938extern void ext4_free_inode(handle_t *, struct inode *);
1939extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1940extern unsigned long ext4_count_free_inodes(struct super_block *);
1941extern unsigned long ext4_count_dirs(struct super_block *);
1942extern void ext4_check_inodes_bitmap(struct super_block *);
1943extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1944extern int ext4_init_inode_table(struct super_block *sb,
1945				 ext4_group_t group, int barrier);
1946extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
1947
1948/* mballoc.c */
 
1949extern long ext4_mb_stats;
1950extern long ext4_mb_max_to_scan;
1951extern int ext4_mb_init(struct super_block *);
1952extern int ext4_mb_release(struct super_block *);
1953extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1954				struct ext4_allocation_request *, int *);
1955extern int ext4_mb_reserve_blocks(struct super_block *, int);
1956extern void ext4_discard_preallocations(struct inode *);
1957extern int __init ext4_init_mballoc(void);
1958extern void ext4_exit_mballoc(void);
 
 
 
 
 
 
1959extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1960			     struct buffer_head *bh, ext4_fsblk_t block,
1961			     unsigned long count, int flags);
 
 
1962extern int ext4_mb_add_groupinfo(struct super_block *sb,
1963		ext4_group_t i, struct ext4_group_desc *desc);
1964extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
1965				ext4_fsblk_t block, unsigned long count);
1966extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
 
1967
1968/* inode.c */
1969struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1970						ext4_lblk_t, int, int *);
1971struct buffer_head *ext4_bread(handle_t *, struct inode *,
1972						ext4_lblk_t, int, int *);
 
 
 
1973int ext4_get_block(struct inode *inode, sector_t iblock,
1974				struct buffer_head *bh_result, int create);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1975
1976extern struct inode *ext4_iget(struct super_block *, unsigned long);
1977extern int  ext4_write_inode(struct inode *, struct writeback_control *);
1978extern int  ext4_setattr(struct dentry *, struct iattr *);
1979extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1980				struct kstat *stat);
1981extern void ext4_evict_inode(struct inode *);
1982extern void ext4_clear_inode(struct inode *);
 
1983extern int  ext4_sync_inode(handle_t *, struct inode *);
1984extern void ext4_dirty_inode(struct inode *, int);
1985extern int ext4_change_inode_journal_flag(struct inode *, int);
1986extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
 
1987extern int ext4_can_truncate(struct inode *inode);
1988extern void ext4_truncate(struct inode *);
1989extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
1990extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1991extern void ext4_set_inode_flags(struct inode *);
1992extern void ext4_get_inode_flags(struct ext4_inode_info *);
1993extern int ext4_alloc_da_blocks(struct inode *inode);
1994extern void ext4_set_aops(struct inode *inode);
1995extern int ext4_writepage_trans_blocks(struct inode *);
1996extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1997extern int ext4_discard_partial_page_buffers(handle_t *handle,
1998		struct address_space *mapping, loff_t from,
1999		loff_t length, int flags);
2000extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2001extern qsize_t *ext4_get_reserved_space(struct inode *inode);
 
 
2002extern void ext4_da_update_reserve_space(struct inode *inode,
2003					int used, int quota_claim);
 
 
2004
2005/* indirect.c */
2006extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2007				struct ext4_map_blocks *map, int flags);
2008extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2009				const struct iovec *iov, loff_t offset,
2010				unsigned long nr_segs);
2011extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2012extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2013extern void ext4_ind_truncate(struct inode *inode);
2014
2015/* ioctl.c */
2016extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2017extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2018
2019/* migrate.c */
2020extern int ext4_ext_migrate(struct inode *);
 
2021
2022/* namei.c */
2023extern int ext4_dirent_csum_verify(struct inode *inode,
2024				   struct ext4_dir_entry *dirent);
2025extern int ext4_orphan_add(handle_t *, struct inode *);
2026extern int ext4_orphan_del(handle_t *, struct inode *);
2027extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2028				__u32 start_minor_hash, __u32 *next_hash);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2029
2030/* resize.c */
 
2031extern int ext4_group_add(struct super_block *sb,
2032				struct ext4_new_group_data *input);
2033extern int ext4_group_extend(struct super_block *sb,
2034				struct ext4_super_block *es,
2035				ext4_fsblk_t n_blocks_count);
2036extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2037
2038/* super.c */
 
 
 
2039extern int ext4_calculate_overhead(struct super_block *sb);
2040extern int ext4_superblock_csum_verify(struct super_block *sb,
2041				       struct ext4_super_block *es);
2042extern void ext4_superblock_csum_set(struct super_block *sb,
2043				     struct ext4_super_block *es);
2044extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2045extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2046extern void ext4_kvfree(void *ptr);
2047extern __printf(4, 5)
2048void __ext4_error(struct super_block *, const char *, unsigned int,
 
 
2049		  const char *, ...);
2050#define ext4_error(sb, message...)	__ext4_error(sb, __func__,	\
2051						     __LINE__, ## message)
2052extern __printf(5, 6)
2053void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2054		      const char *, ...);
2055extern __printf(5, 6)
2056void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2057		     const char *, ...);
2058extern void __ext4_std_error(struct super_block *, const char *,
2059			     unsigned int, int);
2060extern __printf(4, 5)
2061void __ext4_abort(struct super_block *, const char *, unsigned int,
2062		  const char *, ...);
2063#define ext4_abort(sb, message...)	__ext4_abort(sb, __func__, \
2064						       __LINE__, ## message)
2065extern __printf(4, 5)
2066void __ext4_warning(struct super_block *, const char *, unsigned int,
2067		    const char *, ...);
2068#define ext4_warning(sb, message...)	__ext4_warning(sb, __func__, \
2069						       __LINE__, ## message)
 
2070extern __printf(3, 4)
2071void ext4_msg(struct super_block *, const char *, const char *, ...);
2072extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2073			   const char *, unsigned int, const char *);
2074#define dump_mmp_msg(sb, mmp, msg)	__dump_mmp_msg(sb, mmp, __func__, \
2075						       __LINE__, msg)
2076extern __printf(7, 8)
2077void __ext4_grp_locked_error(const char *, unsigned int,
2078			     struct super_block *, ext4_group_t,
2079			     unsigned long, ext4_fsblk_t,
2080			     const char *, ...);
2081#define ext4_grp_locked_error(sb, grp, message...) \
2082	__ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2083extern void ext4_update_dynamic_rev(struct super_block *sb);
2084extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2085					__u32 compat);
2086extern int ext4_update_rocompat_feature(handle_t *handle,
2087					struct super_block *sb,	__u32 rocompat);
2088extern int ext4_update_incompat_feature(handle_t *handle,
2089					struct super_block *sb,	__u32 incompat);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2090extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2091				      struct ext4_group_desc *bg);
2092extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2093				      struct ext4_group_desc *bg);
2094extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2095				     struct ext4_group_desc *bg);
2096extern __u32 ext4_free_group_clusters(struct super_block *sb,
2097				      struct ext4_group_desc *bg);
2098extern __u32 ext4_free_inodes_count(struct super_block *sb,
2099				 struct ext4_group_desc *bg);
2100extern __u32 ext4_used_dirs_count(struct super_block *sb,
2101				struct ext4_group_desc *bg);
2102extern __u32 ext4_itable_unused_count(struct super_block *sb,
2103				   struct ext4_group_desc *bg);
2104extern void ext4_block_bitmap_set(struct super_block *sb,
2105				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2106extern void ext4_inode_bitmap_set(struct super_block *sb,
2107				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2108extern void ext4_inode_table_set(struct super_block *sb,
2109				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2110extern void ext4_free_group_clusters_set(struct super_block *sb,
2111					 struct ext4_group_desc *bg,
2112					 __u32 count);
2113extern void ext4_free_inodes_set(struct super_block *sb,
2114				struct ext4_group_desc *bg, __u32 count);
2115extern void ext4_used_dirs_set(struct super_block *sb,
2116				struct ext4_group_desc *bg, __u32 count);
2117extern void ext4_itable_unused_set(struct super_block *sb,
2118				   struct ext4_group_desc *bg, __u32 count);
2119extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2120				       struct ext4_group_desc *gdp);
2121extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2122				     struct ext4_group_desc *gdp);
 
 
 
 
 
 
 
 
 
 
 
2123
2124static inline int ext4_has_group_desc_csum(struct super_block *sb)
2125{
2126	return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2127					  EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2128					  EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2129}
2130
2131static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2132{
2133	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2134		le32_to_cpu(es->s_blocks_count_lo);
2135}
2136
2137static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2138{
2139	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2140		le32_to_cpu(es->s_r_blocks_count_lo);
2141}
2142
2143static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2144{
2145	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2146		le32_to_cpu(es->s_free_blocks_count_lo);
2147}
2148
2149static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2150					 ext4_fsblk_t blk)
2151{
2152	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2153	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2154}
2155
2156static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2157					      ext4_fsblk_t blk)
2158{
2159	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2160	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2161}
2162
2163static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2164					   ext4_fsblk_t blk)
2165{
2166	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2167	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2168}
2169
2170static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
 
2171{
2172	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
 
2173		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2174			le32_to_cpu(raw_inode->i_size_lo);
2175	else
2176		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2177}
2178
2179static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2180{
2181	raw_inode->i_size_lo = cpu_to_le32(i_size);
2182	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2183}
2184
2185static inline
2186struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2187					    ext4_group_t group)
2188{
2189	 struct ext4_group_info ***grp_info;
2190	 long indexv, indexh;
2191	 grp_info = EXT4_SB(sb)->s_group_info;
2192	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2193	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2194	 return grp_info[indexv][indexh];
 
2195}
2196
2197/*
2198 * Reading s_groups_count requires using smp_rmb() afterwards.  See
2199 * the locking protocol documented in the comments of ext4_group_add()
2200 * in resize.c
2201 */
2202static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2203{
2204	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
2205
2206	smp_rmb();
2207	return ngroups;
2208}
2209
2210static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2211					     ext4_group_t block_group)
2212{
2213	return block_group >> sbi->s_log_groups_per_flex;
2214}
2215
2216static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2217{
2218	return 1 << sbi->s_log_groups_per_flex;
2219}
2220
2221#define ext4_std_error(sb, errno)				\
2222do {								\
2223	if ((errno))						\
2224		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
2225} while (0)
2226
2227#ifdef CONFIG_SMP
2228/* Each CPU can accumulate percpu_counter_batch clusters in their local
2229 * counters. So we need to make sure we have free clusters more
2230 * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2231 */
2232#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2233#else
2234#define EXT4_FREECLUSTERS_WATERMARK 0
2235#endif
2236
 
2237static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2238{
2239	/*
2240	 * XXX: replace with spinlock if seen contended -bzzz
2241	 */
2242	down_write(&EXT4_I(inode)->i_data_sem);
2243	if (newsize > EXT4_I(inode)->i_disksize)
2244		EXT4_I(inode)->i_disksize = newsize;
2245	up_write(&EXT4_I(inode)->i_data_sem);
2246	return ;
2247}
2248
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2249struct ext4_group_info {
2250	unsigned long   bb_state;
2251	struct rb_root  bb_free_root;
2252	ext4_grpblk_t	bb_first_free;	/* first free block */
2253	ext4_grpblk_t	bb_free;	/* total free blocks */
2254	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
2255	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
2256	struct          list_head bb_prealloc_list;
2257#ifdef DOUBLE_CHECK
2258	void            *bb_bitmap;
2259#endif
2260	struct rw_semaphore alloc_sem;
2261	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
2262					 * regions, index is order.
2263					 * bb_counters[3] = 5 means
2264					 * 5 free 8-block regions. */
2265};
2266
2267#define EXT4_GROUP_INFO_NEED_INIT_BIT		0
2268#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
 
 
 
 
 
 
 
2269
2270#define EXT4_MB_GRP_NEED_INIT(grp)	\
2271	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
 
 
 
 
2272
2273#define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
2274	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2275#define EXT4_MB_GRP_SET_TRIMMED(grp)	\
2276	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2277#define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
2278	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
 
 
2279
2280#define EXT4_MAX_CONTENTION		8
2281#define EXT4_CONTENTION_THRESHOLD	2
2282
2283static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2284					      ext4_group_t group)
2285{
2286	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2287}
2288
2289/*
2290 * Returns true if the filesystem is busy enough that attempts to
2291 * access the block group locks has run into contention.
2292 */
2293static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2294{
2295	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2296}
2297
2298static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2299{
2300	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2301	if (spin_trylock(lock))
2302		/*
2303		 * We're able to grab the lock right away, so drop the
2304		 * lock contention counter.
2305		 */
2306		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2307	else {
2308		/*
2309		 * The lock is busy, so bump the contention counter,
2310		 * and then wait on the spin lock.
2311		 */
2312		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2313				  EXT4_MAX_CONTENTION);
2314		spin_lock(lock);
2315	}
2316}
2317
2318static inline void ext4_unlock_group(struct super_block *sb,
2319					ext4_group_t group)
2320{
2321	spin_unlock(ext4_group_lock_ptr(sb, group));
2322}
2323
2324static inline void ext4_mark_super_dirty(struct super_block *sb)
2325{
2326	struct ext4_super_block *es = EXT4_SB(sb)->s_es;
2327
2328	ext4_superblock_csum_set(sb, es);
2329	if (EXT4_SB(sb)->s_journal == NULL)
2330		sb->s_dirt =1;
2331}
2332
2333/*
2334 * Block validity checking
2335 */
2336#define ext4_check_indirect_blockref(inode, bh)				\
2337	ext4_check_blockref(__func__, __LINE__, inode,			\
2338			    (__le32 *)(bh)->b_data,			\
2339			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2340
2341#define ext4_ind_check_inode(inode)					\
2342	ext4_check_blockref(__func__, __LINE__, inode,			\
2343			    EXT4_I(inode)->i_data,			\
2344			    EXT4_NDIR_BLOCKS)
2345
2346/*
2347 * Inodes and files operations
2348 */
2349
2350/* dir.c */
2351extern const struct file_operations ext4_dir_operations;
2352
 
 
 
 
2353/* file.c */
2354extern const struct inode_operations ext4_file_inode_operations;
2355extern const struct file_operations ext4_file_operations;
2356extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2357
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2358/* namei.c */
2359extern const struct inode_operations ext4_dir_inode_operations;
2360extern const struct inode_operations ext4_special_inode_operations;
2361extern struct dentry *ext4_get_parent(struct dentry *child);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2362
2363/* symlink.c */
 
2364extern const struct inode_operations ext4_symlink_inode_operations;
2365extern const struct inode_operations ext4_fast_symlink_inode_operations;
2366
 
 
 
 
 
 
2367/* block_validity */
2368extern void ext4_release_system_zone(struct super_block *sb);
2369extern int ext4_setup_system_zone(struct super_block *sb);
2370extern int __init ext4_init_system_zone(void);
2371extern void ext4_exit_system_zone(void);
2372extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2373				 ext4_fsblk_t start_blk,
2374				 unsigned int count);
2375extern int ext4_check_blockref(const char *, unsigned int,
2376			       struct inode *, __le32 *, unsigned int);
2377
2378/* extents.c */
2379extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2380extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2381extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2382				       int chunk);
 
 
 
 
 
 
 
2383extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2384			       struct ext4_map_blocks *map, int flags);
2385extern void ext4_ext_truncate(struct inode *);
2386extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2387				loff_t length);
2388extern void ext4_ext_init(struct super_block *);
2389extern void ext4_ext_release(struct super_block *);
2390extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2391			  loff_t len);
2392extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2393			  ssize_t len);
 
 
2394extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2395			   struct ext4_map_blocks *map, int flags);
 
 
 
 
 
 
 
 
 
 
 
 
2396extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2397			__u64 start, __u64 len);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2398/* move_extent.c */
 
 
 
 
2399extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2400			     __u64 start_orig, __u64 start_donor,
2401			     __u64 len, __u64 *moved_len);
2402
2403/* page-io.c */
2404extern int __init ext4_init_pageio(void);
2405extern void ext4_exit_pageio(void);
2406extern void ext4_ioend_wait(struct inode *);
2407extern void ext4_free_io_end(ext4_io_end_t *io);
2408extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2409extern int ext4_end_io_nolock(ext4_io_end_t *io);
 
 
 
 
 
2410extern void ext4_io_submit(struct ext4_io_submit *io);
2411extern int ext4_bio_write_page(struct ext4_io_submit *io,
2412			       struct page *page,
2413			       int len,
2414			       struct writeback_control *wbc);
 
 
 
2415
2416/* mmp.c */
2417extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2418extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2419extern int ext4_mmp_csum_verify(struct super_block *sb,
2420				struct mmp_struct *mmp);
2421
2422/* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2423enum ext4_state_bits {
2424	BH_Uninit	/* blocks are allocated but uninitialized on disk */
2425	  = BH_JBDPrivateStart,
2426	BH_AllocFromCluster,	/* allocated blocks were part of already
2427				 * allocated cluster. Note that this flag will
2428				 * never, ever appear in a buffer_head's state
2429				 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2430				 * this is used. */
2431	BH_Da_Mapped,	/* Delayed allocated block that now has a mapping. This
2432			 * flag is set when ext4_map_blocks is called on a
2433			 * delayed allocated block to get its real mapping. */
2434};
2435
2436BUFFER_FNS(Uninit, uninit)
2437TAS_BUFFER_FNS(Uninit, uninit)
2438BUFFER_FNS(Da_Mapped, da_mapped)
2439
2440/*
2441 * Add new method to test wether block and inode bitmaps are properly
2442 * initialized. With uninit_bg reading the block from disk is not enough
2443 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2444 */
2445#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2446
2447static inline int bitmap_uptodate(struct buffer_head *bh)
2448{
2449	return (buffer_uptodate(bh) &&
2450			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2451}
2452static inline void set_bitmap_uptodate(struct buffer_head *bh)
2453{
2454	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2455}
2456
2457#define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
2458
2459/* For ioend & aio unwritten conversion wait queues */
2460#define EXT4_WQ_HASH_SZ		37
2461#define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
2462					    EXT4_WQ_HASH_SZ])
2463#define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
2464					     EXT4_WQ_HASH_SZ])
2465extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2466extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2467
2468#define EXT4_RESIZING	0
2469extern int ext4_resize_begin(struct super_block *sb);
2470extern void ext4_resize_end(struct super_block *sb);
2471
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2472#endif	/* __KERNEL__ */
2473
2474#include "ext4_extents.h"
 
2475
2476#endif	/* _EXT4_H */
v5.9
   1// SPDX-License-Identifier: GPL-2.0
   2/*
   3 *  ext4.h
   4 *
   5 * Copyright (C) 1992, 1993, 1994, 1995
   6 * Remy Card (card@masi.ibp.fr)
   7 * Laboratoire MASI - Institut Blaise Pascal
   8 * Universite Pierre et Marie Curie (Paris VI)
   9 *
  10 *  from
  11 *
  12 *  linux/include/linux/minix_fs.h
  13 *
  14 *  Copyright (C) 1991, 1992  Linus Torvalds
  15 */
  16
  17#ifndef _EXT4_H
  18#define _EXT4_H
  19
  20#include <linux/types.h>
  21#include <linux/blkdev.h>
  22#include <linux/magic.h>
  23#include <linux/jbd2.h>
  24#include <linux/quota.h>
  25#include <linux/rwsem.h>
  26#include <linux/rbtree.h>
  27#include <linux/seqlock.h>
  28#include <linux/mutex.h>
  29#include <linux/timer.h>
  30#include <linux/version.h>
  31#include <linux/wait.h>
  32#include <linux/sched/signal.h>
  33#include <linux/blockgroup_lock.h>
  34#include <linux/percpu_counter.h>
  35#include <linux/ratelimit.h>
  36#include <crypto/hash.h>
  37#include <linux/falloc.h>
  38#include <linux/percpu-rwsem.h>
  39#include <linux/fiemap.h>
  40#ifdef __KERNEL__
  41#include <linux/compat.h>
  42#endif
  43
  44#include <linux/fscrypt.h>
  45#include <linux/fsverity.h>
  46
  47#include <linux/compiler.h>
  48
  49/*
  50 * The fourth extended filesystem constants/structures
  51 */
  52
  53/*
  54 * with AGGRESSIVE_CHECK allocator runs consistency checks over
  55 * structures. these checks slow things down a lot
  56 */
  57#define AGGRESSIVE_CHECK__
  58
  59/*
  60 * with DOUBLE_CHECK defined mballoc creates persistent in-core
  61 * bitmaps, maintains and uses them to check for double allocations
  62 */
  63#define DOUBLE_CHECK__
  64
  65/*
  66 * Define EXT4FS_DEBUG to produce debug messages
  67 */
  68#undef EXT4FS_DEBUG
  69
  70/*
  71 * Debug code
  72 */
  73#ifdef EXT4FS_DEBUG
  74#define ext4_debug(f, a...)						\
  75	do {								\
  76		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
  77			__FILE__, __LINE__, __func__);			\
  78		printk(KERN_DEBUG f, ## a);				\
  79	} while (0)
  80#else
  81#define ext4_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
  82#endif
  83
  84 /*
  85  * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
  86  */
  87#define EXT_DEBUG__
  88
  89/*
  90 * Dynamic printk for controlled extents debugging.
  91 */
  92#ifdef CONFIG_EXT4_DEBUG
  93#define ext_debug(ino, fmt, ...)					\
  94	pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt,	\
  95		 current->comm, task_pid_nr(current),			\
  96		 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__,	\
  97		 __func__, ##__VA_ARGS__)
  98#else
  99#define ext_debug(ino, fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
 100#endif
 101
 102/* data type for block offset of block group */
 103typedef int ext4_grpblk_t;
 104
 105/* data type for filesystem-wide blocks number */
 106typedef unsigned long long ext4_fsblk_t;
 107
 108/* data type for file logical block number */
 109typedef __u32 ext4_lblk_t;
 110
 111/* data type for block group number */
 112typedef unsigned int ext4_group_t;
 113
 114enum SHIFT_DIRECTION {
 115	SHIFT_LEFT = 0,
 116	SHIFT_RIGHT,
 117};
 118
 119/*
 120 * Flags used in mballoc's allocation_context flags field.
 121 *
 122 * Also used to show what's going on for debugging purposes when the
 123 * flag field is exported via the traceport interface
 124 */
 125
 126/* prefer goal again. length */
 127#define EXT4_MB_HINT_MERGE		0x0001
 128/* blocks already reserved */
 129#define EXT4_MB_HINT_RESERVED		0x0002
 130/* metadata is being allocated */
 131#define EXT4_MB_HINT_METADATA		0x0004
 132/* first blocks in the file */
 133#define EXT4_MB_HINT_FIRST		0x0008
 134/* search for the best chunk */
 135#define EXT4_MB_HINT_BEST		0x0010
 136/* data is being allocated */
 137#define EXT4_MB_HINT_DATA		0x0020
 138/* don't preallocate (for tails) */
 139#define EXT4_MB_HINT_NOPREALLOC		0x0040
 140/* allocate for locality group */
 141#define EXT4_MB_HINT_GROUP_ALLOC	0x0080
 142/* allocate goal blocks or none */
 143#define EXT4_MB_HINT_GOAL_ONLY		0x0100
 144/* goal is meaningful */
 145#define EXT4_MB_HINT_TRY_GOAL		0x0200
 146/* blocks already pre-reserved by delayed allocation */
 147#define EXT4_MB_DELALLOC_RESERVED	0x0400
 148/* We are doing stream allocation */
 149#define EXT4_MB_STREAM_ALLOC		0x0800
 150/* Use reserved root blocks if needed */
 151#define EXT4_MB_USE_ROOT_BLOCKS		0x1000
 152/* Use blocks from reserved pool */
 153#define EXT4_MB_USE_RESERVED		0x2000
 154/* Do strict check for free blocks while retrying block allocation */
 155#define EXT4_MB_STRICT_CHECK		0x4000
 156
 157struct ext4_allocation_request {
 158	/* target inode for block we're allocating */
 159	struct inode *inode;
 160	/* how many blocks we want to allocate */
 161	unsigned int len;
 162	/* logical block in target inode */
 163	ext4_lblk_t logical;
 164	/* the closest logical allocated block to the left */
 165	ext4_lblk_t lleft;
 166	/* the closest logical allocated block to the right */
 167	ext4_lblk_t lright;
 168	/* phys. target (a hint) */
 169	ext4_fsblk_t goal;
 170	/* phys. block for the closest logical allocated block to the left */
 171	ext4_fsblk_t pleft;
 172	/* phys. block for the closest logical allocated block to the right */
 173	ext4_fsblk_t pright;
 174	/* flags. see above EXT4_MB_HINT_* */
 175	unsigned int flags;
 176};
 177
 178/*
 179 * Logical to physical block mapping, used by ext4_map_blocks()
 180 *
 181 * This structure is used to pass requests into ext4_map_blocks() as
 182 * well as to store the information returned by ext4_map_blocks().  It
 183 * takes less room on the stack than a struct buffer_head.
 184 */
 185#define EXT4_MAP_NEW		BIT(BH_New)
 186#define EXT4_MAP_MAPPED		BIT(BH_Mapped)
 187#define EXT4_MAP_UNWRITTEN	BIT(BH_Unwritten)
 188#define EXT4_MAP_BOUNDARY	BIT(BH_Boundary)
 
 
 
 
 
 
 
 
 
 189#define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
 190				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
 
 191
 192struct ext4_map_blocks {
 193	ext4_fsblk_t m_pblk;
 194	ext4_lblk_t m_lblk;
 195	unsigned int m_len;
 196	unsigned int m_flags;
 197};
 198
 199/*
 200 * Block validity checking, system zone rbtree.
 201 */
 202struct ext4_system_blocks {
 203	struct rb_root root;
 204	struct rcu_head rcu;
 
 
 
 
 
 
 
 205};
 206
 207/*
 208 * Flags for ext4_io_end->flags
 209 */
 210#define	EXT4_IO_END_UNWRITTEN	0x0001
 
 
 
 
 211
 212struct ext4_io_end_vec {
 213	struct list_head list;		/* list of io_end_vec */
 214	loff_t offset;			/* offset in the file */
 215	ssize_t size;			/* size of the extent */
 216};
 217
 
 
 218/*
 219 * For converting unwritten extents on a work queue. 'handle' is used for
 220 * buffered writeback.
 
 
 
 221 */
 222typedef struct ext4_io_end {
 223	struct list_head	list;		/* per-file finished IO list */
 224	handle_t		*handle;	/* handle reserved for extent
 225						 * conversion */
 226	struct inode		*inode;		/* file being written to */
 227	struct bio		*bio;		/* Linked list of completed
 228						 * bios covering the extent */
 229	unsigned int		flag;		/* unwritten or not */
 230	atomic_t		count;		/* reference counter */
 231	struct list_head	list_vec;	/* list of ext4_io_end_vec */
 
 
 
 
 
 
 232} ext4_io_end_t;
 233
 234struct ext4_io_submit {
 235	struct writeback_control *io_wbc;
 236	struct bio		*io_bio;
 237	ext4_io_end_t		*io_end;
 
 238	sector_t		io_next_block;
 239};
 240
 241/*
 242 * Special inodes numbers
 243 */
 244#define	EXT4_BAD_INO		 1	/* Bad blocks inode */
 245#define EXT4_ROOT_INO		 2	/* Root inode */
 246#define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
 247#define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
 248#define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
 249#define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
 250#define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
 251#define EXT4_JOURNAL_INO	 8	/* Journal inode */
 252
 253/* First non-reserved inode for old ext4 filesystems */
 254#define EXT4_GOOD_OLD_FIRST_INO	11
 255
 256/*
 257 * Maximal count of links to a file
 258 */
 259#define EXT4_LINK_MAX		65000
 260
 261/*
 262 * Macro-instructions used to manage several block sizes
 263 */
 264#define EXT4_MIN_BLOCK_SIZE		1024
 265#define	EXT4_MAX_BLOCK_SIZE		65536
 266#define EXT4_MIN_BLOCK_LOG_SIZE		10
 267#define EXT4_MAX_BLOCK_LOG_SIZE		16
 268#define EXT4_MAX_CLUSTER_LOG_SIZE	30
 269#ifdef __KERNEL__
 270# define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
 271#else
 272# define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
 273#endif
 274#define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
 275#define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
 276					 EXT4_SB(s)->s_cluster_bits)
 277#ifdef __KERNEL__
 278# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
 279# define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
 280#else
 281# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
 282#endif
 283#ifdef __KERNEL__
 284#define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
 285#define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
 286#define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
 287#else
 288#define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 289				 EXT4_GOOD_OLD_INODE_SIZE : \
 290				 (s)->s_inode_size)
 291#define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 292				 EXT4_GOOD_OLD_FIRST_INO : \
 293				 (s)->s_first_ino)
 294#endif
 295#define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
 296#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
 297	((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
 298								  blkbits))
 299
 300/* Translate a block number to a cluster number */
 301#define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
 302/* Translate a cluster number to a block number */
 303#define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
 304/* Translate # of blks to # of clusters */
 305#define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
 306				 (sbi)->s_cluster_bits)
 307/* Mask out the low bits to get the starting block of the cluster */
 308#define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
 309				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
 310#define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
 311				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
 312/* Fill in the low bits to get the last block of the cluster */
 313#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |				\
 314				    ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
 315/* Get the cluster offset */
 316#define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
 317				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
 318#define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
 319				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
 320
 321/*
 322 * Structure of a blocks group descriptor
 323 */
 324struct ext4_group_desc
 325{
 326	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
 327	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
 328	__le32	bg_inode_table_lo;	/* Inodes table block */
 329	__le16	bg_free_blocks_count_lo;/* Free blocks count */
 330	__le16	bg_free_inodes_count_lo;/* Free inodes count */
 331	__le16	bg_used_dirs_count_lo;	/* Directories count */
 332	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
 333	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
 334	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
 335	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
 336	__le16  bg_itable_unused_lo;	/* Unused inodes count */
 337	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
 338	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
 339	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
 340	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
 341	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
 342	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
 343	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
 344	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
 345	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
 346	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
 347	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
 348	__u32   bg_reserved;
 349};
 350
 351#define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
 352	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
 353	 sizeof(__le16))
 354#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
 355	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
 356	 sizeof(__le16))
 357
 358/*
 359 * Structure of a flex block group info
 360 */
 361
 362struct flex_groups {
 363	atomic64_t	free_clusters;
 364	atomic_t	free_inodes;
 365	atomic_t	used_dirs;
 366};
 367
 368#define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
 369#define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
 370#define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
 371
 372/*
 373 * Macro-instructions used to manage group descriptors
 374 */
 375#define EXT4_MIN_DESC_SIZE		32
 376#define EXT4_MIN_DESC_SIZE_64BIT	64
 377#define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
 378#define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
 379#ifdef __KERNEL__
 380# define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
 381# define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
 382# define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
 383# define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
 384# define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
 385#else
 386# define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
 387# define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
 388# define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
 389#endif
 390
 391/*
 392 * Constants relative to the data blocks
 393 */
 394#define	EXT4_NDIR_BLOCKS		12
 395#define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
 396#define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
 397#define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
 398#define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
 399
 400/*
 401 * Inode flags
 402 */
 403#define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
 404#define	EXT4_UNRM_FL			0x00000002 /* Undelete */
 405#define	EXT4_COMPR_FL			0x00000004 /* Compress file */
 406#define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
 407#define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
 408#define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
 409#define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
 410#define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
 411/* Reserved for compression usage... */
 412#define EXT4_DIRTY_FL			0x00000100
 413#define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
 414#define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
 415	/* nb: was previously EXT2_ECOMPR_FL */
 416#define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
 417/* End compression flags --- maybe not all used */
 418#define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
 419#define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
 420#define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
 421#define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
 422#define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
 423#define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
 424#define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
 425#define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
 426#define EXT4_VERITY_FL			0x00100000 /* Verity protected inode */
 427#define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
 428/* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
 429
 430#define EXT4_DAX_FL			0x02000000 /* Inode is DAX */
 431
 432#define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
 433#define EXT4_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
 434#define EXT4_CASEFOLD_FL		0x40000000 /* Casefolded directory */
 435#define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
 436
 437/* User modifiable flags */
 438#define EXT4_FL_USER_MODIFIABLE		(EXT4_SECRM_FL | \
 439					 EXT4_UNRM_FL | \
 440					 EXT4_COMPR_FL | \
 441					 EXT4_SYNC_FL | \
 442					 EXT4_IMMUTABLE_FL | \
 443					 EXT4_APPEND_FL | \
 444					 EXT4_NODUMP_FL | \
 445					 EXT4_NOATIME_FL | \
 446					 EXT4_JOURNAL_DATA_FL | \
 447					 EXT4_NOTAIL_FL | \
 448					 EXT4_DIRSYNC_FL | \
 449					 EXT4_TOPDIR_FL | \
 450					 EXT4_EXTENTS_FL | \
 451					 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
 452					 EXT4_DAX_FL | \
 453					 EXT4_PROJINHERIT_FL | \
 454					 EXT4_CASEFOLD_FL)
 455
 456/* User visible flags */
 457#define EXT4_FL_USER_VISIBLE		(EXT4_FL_USER_MODIFIABLE | \
 458					 EXT4_DIRTY_FL | \
 459					 EXT4_COMPRBLK_FL | \
 460					 EXT4_NOCOMPR_FL | \
 461					 EXT4_ENCRYPT_FL | \
 462					 EXT4_INDEX_FL | \
 463					 EXT4_VERITY_FL | \
 464					 EXT4_INLINE_DATA_FL)
 465
 466/* Flags we can manipulate with through FS_IOC_FSSETXATTR */
 467#define EXT4_FL_XFLAG_VISIBLE		(EXT4_SYNC_FL | \
 468					 EXT4_IMMUTABLE_FL | \
 469					 EXT4_APPEND_FL | \
 470					 EXT4_NODUMP_FL | \
 471					 EXT4_NOATIME_FL | \
 472					 EXT4_PROJINHERIT_FL | \
 473					 EXT4_DAX_FL)
 474
 475/* Flags that should be inherited by new inodes from their parent. */
 476#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
 477			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
 478			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
 479			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
 480			   EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
 481			   EXT4_DAX_FL)
 482
 483/* Flags that are appropriate for regular files (all but dir-specific ones). */
 484#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
 485			   EXT4_PROJINHERIT_FL))
 486
 487/* Flags that are appropriate for non-directories/regular files. */
 488#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
 489
 490/* The only flags that should be swapped */
 491#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
 492
 493/* Flags which are mutually exclusive to DAX */
 494#define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
 495			   EXT4_JOURNAL_DATA_FL)
 496
 497/* Mask out flags that are inappropriate for the given type of inode. */
 498static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
 499{
 500	if (S_ISDIR(mode))
 501		return flags;
 502	else if (S_ISREG(mode))
 503		return flags & EXT4_REG_FLMASK;
 504	else
 505		return flags & EXT4_OTHER_FLMASK;
 506}
 507
 508/*
 509 * Inode flags used for atomic set/get
 510 */
 511enum {
 512	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
 513	EXT4_INODE_UNRM		= 1,	/* Undelete */
 514	EXT4_INODE_COMPR	= 2,	/* Compress file */
 515	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
 516	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
 517	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
 518	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
 519	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
 520/* Reserved for compression usage... */
 521	EXT4_INODE_DIRTY	= 8,
 522	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
 523	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
 524	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
 525/* End compression flags --- maybe not all used */
 526	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
 527	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
 528	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
 529	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
 530	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
 531	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
 532	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
 533	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
 534	EXT4_INODE_VERITY	= 20,	/* Verity protected inode */
 535	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
 536/* 22 was formerly EXT4_INODE_EOFBLOCKS */
 537	EXT4_INODE_DAX		= 25,	/* Inode is DAX */
 538	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
 539	EXT4_INODE_PROJINHERIT	= 29,	/* Create with parents projid */
 540	EXT4_INODE_CASEFOLD	= 30,	/* Casefolded directory */
 541	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
 542};
 543
 
 
 
 
 
 544/*
 545 * Since it's pretty easy to mix up bit numbers and hex values, we use a
 546 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
 547 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
 548 * any extra space in the compiled kernel image, otherwise, the build will fail.
 549 * It's important that these values are the same, since we are using
 550 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
 551 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
 552 * values found in ext2, ext3 and ext4 filesystems, and of course the values
 553 * defined in e2fsprogs.
 
 554 *
 555 * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
 556 */
 557#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
 558#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
 559
 560static inline void ext4_check_flag_values(void)
 561{
 562	CHECK_FLAG_VALUE(SECRM);
 563	CHECK_FLAG_VALUE(UNRM);
 564	CHECK_FLAG_VALUE(COMPR);
 565	CHECK_FLAG_VALUE(SYNC);
 566	CHECK_FLAG_VALUE(IMMUTABLE);
 567	CHECK_FLAG_VALUE(APPEND);
 568	CHECK_FLAG_VALUE(NODUMP);
 569	CHECK_FLAG_VALUE(NOATIME);
 570	CHECK_FLAG_VALUE(DIRTY);
 571	CHECK_FLAG_VALUE(COMPRBLK);
 572	CHECK_FLAG_VALUE(NOCOMPR);
 573	CHECK_FLAG_VALUE(ENCRYPT);
 574	CHECK_FLAG_VALUE(INDEX);
 575	CHECK_FLAG_VALUE(IMAGIC);
 576	CHECK_FLAG_VALUE(JOURNAL_DATA);
 577	CHECK_FLAG_VALUE(NOTAIL);
 578	CHECK_FLAG_VALUE(DIRSYNC);
 579	CHECK_FLAG_VALUE(TOPDIR);
 580	CHECK_FLAG_VALUE(HUGE_FILE);
 581	CHECK_FLAG_VALUE(EXTENTS);
 582	CHECK_FLAG_VALUE(VERITY);
 583	CHECK_FLAG_VALUE(EA_INODE);
 584	CHECK_FLAG_VALUE(INLINE_DATA);
 585	CHECK_FLAG_VALUE(PROJINHERIT);
 586	CHECK_FLAG_VALUE(CASEFOLD);
 587	CHECK_FLAG_VALUE(RESERVED);
 588}
 589
 590/* Used to pass group descriptor data when online resize is done */
 591struct ext4_new_group_input {
 592	__u32 group;		/* Group number for this data */
 593	__u64 block_bitmap;	/* Absolute block number of block bitmap */
 594	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
 595	__u64 inode_table;	/* Absolute block number of inode table start */
 596	__u32 blocks_count;	/* Total number of blocks in this group */
 597	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
 598	__u16 unused;
 599};
 600
 601#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 602struct compat_ext4_new_group_input {
 603	u32 group;
 604	compat_u64 block_bitmap;
 605	compat_u64 inode_bitmap;
 606	compat_u64 inode_table;
 607	u32 blocks_count;
 608	u16 reserved_blocks;
 609	u16 unused;
 610};
 611#endif
 612
 613/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
 614struct ext4_new_group_data {
 615	__u32 group;
 616	__u64 block_bitmap;
 617	__u64 inode_bitmap;
 618	__u64 inode_table;
 619	__u32 blocks_count;
 620	__u16 reserved_blocks;
 621	__u16 mdata_blocks;
 622	__u32 free_clusters_count;
 623};
 624
 625/* Indexes used to index group tables in ext4_new_group_data */
 626enum {
 627	BLOCK_BITMAP = 0,	/* block bitmap */
 628	INODE_BITMAP,		/* inode bitmap */
 629	INODE_TABLE,		/* inode tables */
 630	GROUP_TABLE_COUNT,
 631};
 632
 633/*
 634 * Flags used by ext4_map_blocks()
 635 */
 636	/* Allocate any needed blocks and/or convert an unwritten
 637	   extent to be an initialized ext4 */
 638#define EXT4_GET_BLOCKS_CREATE			0x0001
 639	/* Request the creation of an unwritten extent */
 640#define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
 641#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
 642						 EXT4_GET_BLOCKS_CREATE)
 643	/* Caller is from the delayed allocation writeout path
 644	 * finally doing the actual allocation of delayed blocks */
 
 645#define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
 646	/* caller is from the direct IO path, request to creation of an
 647	unwritten extents if not allocated, split the unwritten
 648	extent if blocks has been preallocated already*/
 649#define EXT4_GET_BLOCKS_PRE_IO			0x0008
 650#define EXT4_GET_BLOCKS_CONVERT			0x0010
 651#define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
 652					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
 653	/* Convert extent to initialized after IO complete */
 654#define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
 655					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
 656	/* Eventual metadata allocation (due to growing extent tree)
 657	 * should not fail, so try to use reserved blocks for that.*/
 658#define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
 659	/* Don't normalize allocation size (used for fallocate) */
 660#define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
 661	/* Convert written extents to unwritten */
 662#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
 663	/* Write zeros to newly created written extents */
 664#define EXT4_GET_BLOCKS_ZERO			0x0200
 665#define EXT4_GET_BLOCKS_CREATE_ZERO		(EXT4_GET_BLOCKS_CREATE |\
 666					EXT4_GET_BLOCKS_ZERO)
 667	/* Caller will submit data before dropping transaction handle. This
 668	 * allows jbd2 to avoid submitting data before commit. */
 669#define EXT4_GET_BLOCKS_IO_SUBMIT		0x0400
 670
 671/*
 672 * The bit position of these flags must not overlap with any of the
 673 * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
 674 * read_extent_tree_block(), ext4_split_extent_at(),
 675 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
 676 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
 677 * caching the extents when reading from the extent tree while a
 678 * truncate or punch hole operation is in progress.
 679 */
 680#define EXT4_EX_NOCACHE				0x40000000
 681#define EXT4_EX_FORCE_CACHE			0x20000000
 682#define EXT4_EX_NOFAIL				0x10000000
 683
 684/*
 685 * Flags used by ext4_free_blocks
 686 */
 687#define EXT4_FREE_BLOCKS_METADATA		0x0001
 688#define EXT4_FREE_BLOCKS_FORGET			0x0002
 689#define EXT4_FREE_BLOCKS_VALIDATED		0x0004
 690#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE		0x0008
 691#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
 692#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
 693#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
 
 
 
 
 694
 695/*
 696 * ioctl commands
 697 */
 
 
 698#define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
 699#define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
 700#define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
 701#define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
 702#define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
 703#define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
 704#define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
 705#define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
 706#define EXT4_IOC_MIGRATE		_IO('f', 9)
 707 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
 708 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
 709#define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
 710#define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
 711#define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
 712#define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
 713#define EXT4_IOC_PRECACHE_EXTENTS	_IO('f', 18)
 714/* ioctl codes 19--39 are reserved for fscrypt */
 715#define EXT4_IOC_CLEAR_ES_CACHE		_IO('f', 40)
 716#define EXT4_IOC_GETSTATE		_IOW('f', 41, __u32)
 717#define EXT4_IOC_GET_ES_CACHE		_IOWR('f', 42, struct fiemap)
 718
 719#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
 720
 721/*
 722 * Flags for going down operation
 723 */
 724#define EXT4_GOING_FLAGS_DEFAULT		0x0	/* going down */
 725#define EXT4_GOING_FLAGS_LOGFLUSH		0x1	/* flush log but not data */
 726#define EXT4_GOING_FLAGS_NOLOGFLUSH		0x2	/* don't flush log nor data */
 727
 728/*
 729 * Flags returned by EXT4_IOC_GETSTATE
 730 *
 731 * We only expose to userspace a subset of the state flags in
 732 * i_state_flags
 733 */
 734#define EXT4_STATE_FLAG_EXT_PRECACHED	0x00000001
 735#define EXT4_STATE_FLAG_NEW		0x00000002
 736#define EXT4_STATE_FLAG_NEWENTRY	0x00000004
 737#define EXT4_STATE_FLAG_DA_ALLOC_CLOSE	0x00000008
 738
 739#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 740/*
 741 * ioctl commands in 32 bit emulation
 742 */
 
 
 743#define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
 744#define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
 745#define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
 746#define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
 747#define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
 748#define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
 749#define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
 750#define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
 751#endif
 752
 753/*
 754 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
 755 * It indicates that the entry in extent status cache is for a hole.
 756 */
 757#define EXT4_FIEMAP_EXTENT_HOLE		0x08000000
 758
 759/* Max physical block we can address w/o extents */
 760#define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
 761
 762/* Max logical block we can support */
 763#define EXT4_MAX_LOGICAL_BLOCK		0xFFFFFFFE
 764
 765/*
 766 * Structure of an inode on the disk
 767 */
 768struct ext4_inode {
 769	__le16	i_mode;		/* File mode */
 770	__le16	i_uid;		/* Low 16 bits of Owner Uid */
 771	__le32	i_size_lo;	/* Size in bytes */
 772	__le32	i_atime;	/* Access time */
 773	__le32	i_ctime;	/* Inode Change time */
 774	__le32	i_mtime;	/* Modification time */
 775	__le32	i_dtime;	/* Deletion Time */
 776	__le16	i_gid;		/* Low 16 bits of Group Id */
 777	__le16	i_links_count;	/* Links count */
 778	__le32	i_blocks_lo;	/* Blocks count */
 779	__le32	i_flags;	/* File flags */
 780	union {
 781		struct {
 782			__le32  l_i_version;
 783		} linux1;
 784		struct {
 785			__u32  h_i_translator;
 786		} hurd1;
 787		struct {
 788			__u32  m_i_reserved1;
 789		} masix1;
 790	} osd1;				/* OS dependent 1 */
 791	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
 792	__le32	i_generation;	/* File version (for NFS) */
 793	__le32	i_file_acl_lo;	/* File ACL */
 794	__le32	i_size_high;
 795	__le32	i_obso_faddr;	/* Obsoleted fragment address */
 796	union {
 797		struct {
 798			__le16	l_i_blocks_high; /* were l_i_reserved1 */
 799			__le16	l_i_file_acl_high;
 800			__le16	l_i_uid_high;	/* these 2 fields */
 801			__le16	l_i_gid_high;	/* were reserved2[0] */
 802			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
 803			__le16	l_i_reserved;
 804		} linux2;
 805		struct {
 806			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
 807			__u16	h_i_mode_high;
 808			__u16	h_i_uid_high;
 809			__u16	h_i_gid_high;
 810			__u32	h_i_author;
 811		} hurd2;
 812		struct {
 813			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
 814			__le16	m_i_file_acl_high;
 815			__u32	m_i_reserved2[2];
 816		} masix2;
 817	} osd2;				/* OS dependent 2 */
 818	__le16	i_extra_isize;
 819	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
 820	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
 821	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
 822	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
 823	__le32  i_crtime;       /* File Creation time */
 824	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
 825	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
 826	__le32	i_projid;	/* Project ID */
 827};
 828
 829struct move_extent {
 830	__u32 reserved;		/* should be zero */
 831	__u32 donor_fd;		/* donor file descriptor */
 832	__u64 orig_start;	/* logical start offset in block for orig */
 833	__u64 donor_start;	/* logical start offset in block for donor */
 834	__u64 len;		/* block length to be moved */
 835	__u64 moved_len;	/* moved block length */
 836};
 837
 838#define EXT4_EPOCH_BITS 2
 839#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
 840#define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
 841
 842/*
 843 * Extended fields will fit into an inode if the filesystem was formatted
 844 * with large inodes (-I 256 or larger) and there are not currently any EAs
 845 * consuming all of the available space. For new inodes we always reserve
 846 * enough space for the kernel's known extended fields, but for inodes
 847 * created with an old kernel this might not have been the case. None of
 848 * the extended inode fields is critical for correct filesystem operation.
 849 * This macro checks if a certain field fits in the inode. Note that
 850 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
 851 */
 852#define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
 853	((offsetof(typeof(*ext4_inode), field) +	\
 854	  sizeof((ext4_inode)->field))			\
 855	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
 856	    (einode)->i_extra_isize))			\
 857
 858/*
 859 * We use an encoding that preserves the times for extra epoch "00":
 860 *
 861 * extra  msb of                         adjust for signed
 862 * epoch  32-bit                         32-bit tv_sec to
 863 * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
 864 * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
 865 * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
 866 * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
 867 * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
 868 * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
 869 * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
 870 * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
 871 * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
 872 *
 873 * Note that previous versions of the kernel on 64-bit systems would
 874 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
 875 * 1970.  e2fsck will correct this, assuming that it is run on the
 876 * affected filesystem before 2242.
 877 */
 878
 879static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
 880{
 881	u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
 882	return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
 
 883}
 884
 885static inline void ext4_decode_extra_time(struct timespec64 *time,
 886					  __le32 extra)
 887{
 888	if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
 889		time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
 890	time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
 
 891}
 892
 893#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)				\
 894do {										\
 895	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     {\
 896		(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	\
 897		(raw_inode)->xtime ## _extra =					\
 898				ext4_encode_extra_time(&(inode)->xtime);	\
 899		}								\
 900	else	\
 901		(raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX));	\
 902} while (0)
 903
 904#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
 905do {									       \
 906	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
 907		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
 908	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
 909		(raw_inode)->xtime ## _extra =				       \
 910				ext4_encode_extra_time(&(einode)->xtime);      \
 911} while (0)
 912
 913#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)				\
 914do {										\
 915	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);	\
 916	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {	\
 917		ext4_decode_extra_time(&(inode)->xtime,				\
 918				       raw_inode->xtime ## _extra);		\
 919		}								\
 920	else									\
 921		(inode)->xtime.tv_nsec = 0;					\
 922} while (0)
 923
 924
 925#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
 926do {									       \
 927	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
 928		(einode)->xtime.tv_sec = 				       \
 929			(signed)le32_to_cpu((raw_inode)->xtime);	       \
 930	else								       \
 931		(einode)->xtime.tv_sec = 0;				       \
 932	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
 933		ext4_decode_extra_time(&(einode)->xtime,		       \
 934				       raw_inode->xtime ## _extra);	       \
 935	else								       \
 936		(einode)->xtime.tv_nsec = 0;				       \
 937} while (0)
 938
 939#define i_disk_version osd1.linux1.l_i_version
 940
 941#if defined(__KERNEL__) || defined(__linux__)
 942#define i_reserved1	osd1.linux1.l_i_reserved1
 943#define i_file_acl_high	osd2.linux2.l_i_file_acl_high
 944#define i_blocks_high	osd2.linux2.l_i_blocks_high
 945#define i_uid_low	i_uid
 946#define i_gid_low	i_gid
 947#define i_uid_high	osd2.linux2.l_i_uid_high
 948#define i_gid_high	osd2.linux2.l_i_gid_high
 949#define i_checksum_lo	osd2.linux2.l_i_checksum_lo
 950
 951#elif defined(__GNU__)
 952
 953#define i_translator	osd1.hurd1.h_i_translator
 954#define i_uid_high	osd2.hurd2.h_i_uid_high
 955#define i_gid_high	osd2.hurd2.h_i_gid_high
 956#define i_author	osd2.hurd2.h_i_author
 957
 958#elif defined(__masix__)
 959
 960#define i_reserved1	osd1.masix1.m_i_reserved1
 961#define i_file_acl_high	osd2.masix2.m_i_file_acl_high
 962#define i_reserved2	osd2.masix2.m_i_reserved2
 963
 964#endif /* defined(__KERNEL__) || defined(__linux__) */
 965
 966#include "extents_status.h"
 967
 968/*
 969 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
 970 *
 971 * These are needed to avoid lockdep false positives when we need to
 972 * allocate blocks to the quota inode during ext4_map_blocks(), while
 973 * holding i_data_sem for a normal (non-quota) inode.  Since we don't
 974 * do quota tracking for the quota inode, this avoids deadlock (as
 975 * well as infinite recursion, since it isn't turtles all the way
 976 * down...)
 977 *
 978 *  I_DATA_SEM_NORMAL - Used for most inodes
 979 *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
 980 *			  where the second inode has larger inode number
 981 *			  than the first
 982 *  I_DATA_SEM_QUOTA  - Used for quota inodes only
 983 */
 984enum {
 985	I_DATA_SEM_NORMAL = 0,
 986	I_DATA_SEM_OTHER,
 987	I_DATA_SEM_QUOTA,
 988};
 989
 990
 991/*
 992 * fourth extended file system inode data in memory
 993 */
 994struct ext4_inode_info {
 995	__le32	i_data[15];	/* unconverted */
 996	__u32	i_dtime;
 997	ext4_fsblk_t	i_file_acl;
 998
 999	/*
1000	 * i_block_group is the number of the block group which contains
1001	 * this file's inode.  Constant across the lifetime of the inode,
1002	 * it is used for making block allocation decisions - we try to
1003	 * place a file's data blocks near its inode block, and new inodes
1004	 * near to their parent directory's inode.
1005	 */
1006	ext4_group_t	i_block_group;
1007	ext4_lblk_t	i_dir_start_lookup;
1008#if (BITS_PER_LONG < 64)
1009	unsigned long	i_state_flags;		/* Dynamic state flags */
1010#endif
1011	unsigned long	i_flags;
1012
 
1013	/*
1014	 * Extended attributes can be read independently of the main file
1015	 * data. Taking i_mutex even when reading would cause contention
1016	 * between readers of EAs and writers of regular file data, so
1017	 * instead we synchronize on xattr_sem when reading or changing
1018	 * EAs.
1019	 */
1020	struct rw_semaphore xattr_sem;
 
1021
1022	struct list_head i_orphan;	/* unlinked but open inodes */
1023
1024	/*
1025	 * i_disksize keeps track of what the inode size is ON DISK, not
1026	 * in memory.  During truncate, i_size is set to the new size by
1027	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1028	 * set i_disksize to 0 until the truncate is actually under way.
1029	 *
1030	 * The intent is that i_disksize always represents the blocks which
1031	 * are used by this file.  This allows recovery to restart truncate
1032	 * on orphans if we crash during truncate.  We actually write i_disksize
1033	 * into the on-disk inode when writing inodes out, instead of i_size.
1034	 *
1035	 * The only time when i_disksize and i_size may be different is when
1036	 * a truncate is in progress.  The only things which change i_disksize
1037	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1038	 */
1039	loff_t	i_disksize;
1040
1041	/*
1042	 * i_data_sem is for serialising ext4_truncate() against
1043	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1044	 * data tree are chopped off during truncate. We can't do that in
1045	 * ext4 because whenever we perform intermediate commits during
1046	 * truncate, the inode and all the metadata blocks *must* be in a
1047	 * consistent state which allows truncation of the orphans to restart
1048	 * during recovery.  Hence we must fix the get_block-vs-truncate race
1049	 * by other means, so we have i_data_sem.
1050	 */
1051	struct rw_semaphore i_data_sem;
1052	/*
1053	 * i_mmap_sem is for serializing page faults with truncate / punch hole
1054	 * operations. We have to make sure that new page cannot be faulted in
1055	 * a section of the inode that is being punched. We cannot easily use
1056	 * i_data_sem for this since we need protection for the whole punch
1057	 * operation and i_data_sem ranks below transaction start so we have
1058	 * to occasionally drop it.
1059	 */
1060	struct rw_semaphore i_mmap_sem;
1061	struct inode vfs_inode;
1062	struct jbd2_inode *jinode;
1063
1064	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
1065
1066	/*
1067	 * File creation time. Its function is same as that of
1068	 * struct timespec64 i_{a,c,m}time in the generic inode.
1069	 */
1070	struct timespec64 i_crtime;
1071
1072	/* mballoc */
1073	atomic_t i_prealloc_active;
1074	struct list_head i_prealloc_list;
1075	spinlock_t i_prealloc_lock;
1076
1077	/* extents status tree */
1078	struct ext4_es_tree i_es_tree;
1079	rwlock_t i_es_lock;
1080	struct list_head i_es_list;
1081	unsigned int i_es_all_nr;	/* protected by i_es_lock */
1082	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
1083	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
1084					   extents to shrink. Protected by
1085					   i_es_lock  */
1086
1087	/* ialloc */
1088	ext4_group_t	i_last_alloc_group;
1089
1090	/* allocation reservation info for delalloc */
1091	/* In case of bigalloc, this refer to clusters rather than blocks */
1092	unsigned int i_reserved_data_blocks;
1093
1094	/* pending cluster reservations for bigalloc file systems */
1095	struct ext4_pending_tree i_pending_tree;
 
1096
1097	/* on-disk additional length */
1098	__u16 i_extra_isize;
1099
1100	/* Indicate the inline data space. */
1101	u16 i_inline_off;
1102	u16 i_inline_size;
1103
1104#ifdef CONFIG_QUOTA
1105	/* quota space reservation, managed internally by quota code */
1106	qsize_t i_reserved_quota;
1107#endif
1108
1109	/* Lock protecting lists below */
 
1110	spinlock_t i_completed_io_lock;
1111	/*
1112	 * Completed IOs that need unwritten extents handling and have
1113	 * transaction reserved
1114	 */
1115	struct list_head i_rsv_conversion_list;
1116	struct work_struct i_rsv_conversion_work;
1117	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1118
1119	spinlock_t i_block_reservation_lock;
1120
1121	/*
1122	 * Transactions that contain inode's metadata needed to complete
1123	 * fsync and fdatasync, respectively.
1124	 */
1125	tid_t i_sync_tid;
1126	tid_t i_datasync_tid;
1127
1128#ifdef CONFIG_QUOTA
1129	struct dquot *i_dquot[MAXQUOTAS];
1130#endif
1131
1132	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1133	__u32 i_csum_seed;
1134
1135	kprojid_t i_projid;
1136};
1137
1138/*
1139 * File system states
1140 */
1141#define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
1142#define	EXT4_ERROR_FS			0x0002	/* Errors detected */
1143#define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1144
1145/*
1146 * Misc. filesystem flags
1147 */
1148#define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1149#define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1150#define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1151
1152/*
1153 * Mount flags set via mount options or defaults
1154 */
1155#define EXT4_MOUNT_NO_MBCACHE		0x00001 /* Do not use mbcache */
1156#define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1157#define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1158#define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1159#define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1160#define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1161#define EXT4_MOUNT_ERRORS_MASK		0x00070
1162#define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1163#define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1164#ifdef CONFIG_FS_DAX
1165#define EXT4_MOUNT_DAX_ALWAYS		0x00200	/* Direct Access */
1166#else
1167#define EXT4_MOUNT_DAX_ALWAYS		0
1168#endif
1169#define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1170#define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1171#define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1172#define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1173#define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1174#define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1175#define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1176#define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1177#define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1178#define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1179#define EXT4_MOUNT_QUOTA		0x40000 /* Some quota option set */
1180#define EXT4_MOUNT_USRQUOTA		0x80000 /* "old" user quota,
1181						 * enable enforcement for hidden
1182						 * quota files */
1183#define EXT4_MOUNT_GRPQUOTA		0x100000 /* "old" group quota, enable
1184						  * enforcement for hidden quota
1185						  * files */
1186#define EXT4_MOUNT_PRJQUOTA		0x200000 /* Enable project quota
1187						  * enforcement */
1188#define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1189#define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1190#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1191#define EXT4_MOUNT_WARN_ON_ERROR	0x2000000 /* Trigger WARN_ON on error */
1192#define EXT4_MOUNT_PREFETCH_BLOCK_BITMAPS 0x4000000
1193#define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1194#define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1195#define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1196#define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1197#define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1198
1199/*
1200 * Mount flags set either automatically (could not be set by mount option)
1201 * based on per file system feature or property or in special cases such as
1202 * distinguishing between explicit mount option definition and default.
1203 */
1204#define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1205						      specified delalloc */
1206#define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1207						      size of blocksize * 8
1208						      blocks */
1209#define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1210						      file systems */
1211#define EXT4_MOUNT2_DAX_NEVER		0x00000008 /* Do not allow Direct Access */
1212#define EXT4_MOUNT2_DAX_INODE		0x00000010 /* For printing options only */
1213
1214#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM	0x00000008 /* User explicitly
1215						specified journal checksum */
1216
1217#define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1218						~EXT4_MOUNT_##opt
1219#define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1220						EXT4_MOUNT_##opt
1221#define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1222					 EXT4_MOUNT_##opt)
1223
1224#define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1225						~EXT4_MOUNT2_##opt
1226#define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1227						EXT4_MOUNT2_##opt
1228#define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1229					 EXT4_MOUNT2_##opt)
1230
1231#define ext4_test_and_set_bit		__test_and_set_bit_le
1232#define ext4_set_bit			__set_bit_le
1233#define ext4_set_bit_atomic		ext2_set_bit_atomic
1234#define ext4_test_and_clear_bit		__test_and_clear_bit_le
1235#define ext4_clear_bit			__clear_bit_le
1236#define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1237#define ext4_test_bit			test_bit_le
1238#define ext4_find_next_zero_bit		find_next_zero_bit_le
1239#define ext4_find_next_bit		find_next_bit_le
1240
1241extern void ext4_set_bits(void *bm, int cur, int len);
1242
1243/*
1244 * Maximal mount counts between two filesystem checks
1245 */
1246#define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1247#define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1248
1249/*
1250 * Behaviour when detecting errors
1251 */
1252#define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1253#define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1254#define EXT4_ERRORS_PANIC		3	/* Panic */
1255#define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1256
1257/* Metadata checksum algorithm codes */
1258#define EXT4_CRC32C_CHKSUM		1
1259
1260/*
1261 * Structure of the super block
1262 */
1263struct ext4_super_block {
1264/*00*/	__le32	s_inodes_count;		/* Inodes count */
1265	__le32	s_blocks_count_lo;	/* Blocks count */
1266	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1267	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1268/*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1269	__le32	s_first_data_block;	/* First Data Block */
1270	__le32	s_log_block_size;	/* Block size */
1271	__le32	s_log_cluster_size;	/* Allocation cluster size */
1272/*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1273	__le32	s_clusters_per_group;	/* # Clusters per group */
1274	__le32	s_inodes_per_group;	/* # Inodes per group */
1275	__le32	s_mtime;		/* Mount time */
1276/*30*/	__le32	s_wtime;		/* Write time */
1277	__le16	s_mnt_count;		/* Mount count */
1278	__le16	s_max_mnt_count;	/* Maximal mount count */
1279	__le16	s_magic;		/* Magic signature */
1280	__le16	s_state;		/* File system state */
1281	__le16	s_errors;		/* Behaviour when detecting errors */
1282	__le16	s_minor_rev_level;	/* minor revision level */
1283/*40*/	__le32	s_lastcheck;		/* time of last check */
1284	__le32	s_checkinterval;	/* max. time between checks */
1285	__le32	s_creator_os;		/* OS */
1286	__le32	s_rev_level;		/* Revision level */
1287/*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1288	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1289	/*
1290	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1291	 *
1292	 * Note: the difference between the compatible feature set and
1293	 * the incompatible feature set is that if there is a bit set
1294	 * in the incompatible feature set that the kernel doesn't
1295	 * know about, it should refuse to mount the filesystem.
1296	 *
1297	 * e2fsck's requirements are more strict; if it doesn't know
1298	 * about a feature in either the compatible or incompatible
1299	 * feature set, it must abort and not try to meddle with
1300	 * things it doesn't understand...
1301	 */
1302	__le32	s_first_ino;		/* First non-reserved inode */
1303	__le16  s_inode_size;		/* size of inode structure */
1304	__le16	s_block_group_nr;	/* block group # of this superblock */
1305	__le32	s_feature_compat;	/* compatible feature set */
1306/*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1307	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1308/*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1309/*78*/	char	s_volume_name[16];	/* volume name */
1310/*88*/	char	s_last_mounted[64] __nonstring;	/* directory where last mounted */
1311/*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1312	/*
1313	 * Performance hints.  Directory preallocation should only
1314	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1315	 */
1316	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1317	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1318	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1319	/*
1320	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1321	 */
1322/*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1323/*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1324	__le32	s_journal_dev;		/* device number of journal file */
1325	__le32	s_last_orphan;		/* start of list of inodes to delete */
1326	__le32	s_hash_seed[4];		/* HTREE hash seed */
1327	__u8	s_def_hash_version;	/* Default hash version to use */
1328	__u8	s_jnl_backup_type;
1329	__le16  s_desc_size;		/* size of group descriptor */
1330/*100*/	__le32	s_default_mount_opts;
1331	__le32	s_first_meta_bg;	/* First metablock block group */
1332	__le32	s_mkfs_time;		/* When the filesystem was created */
1333	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1334	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1335/*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1336	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1337	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1338	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1339	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1340	__le32	s_flags;		/* Miscellaneous flags */
1341	__le16  s_raid_stride;		/* RAID stride */
1342	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1343	__le64  s_mmp_block;            /* Block for multi-mount protection */
1344	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1345	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1346	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1347	__u8	s_encryption_level;	/* versioning level for encryption */
1348	__u8	s_reserved_pad;		/* Padding to next 32bits */
1349	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1350	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1351	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1352	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1353					      snapshot's future use */
1354	__le32	s_snapshot_list;	/* inode number of the head of the
1355					   on-disk snapshot list */
1356#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1357	__le32	s_error_count;		/* number of fs errors */
1358	__le32	s_first_error_time;	/* first time an error happened */
1359	__le32	s_first_error_ino;	/* inode involved in first error */
1360	__le64	s_first_error_block;	/* block involved of first error */
1361	__u8	s_first_error_func[32] __nonstring;	/* function where the error happened */
1362	__le32	s_first_error_line;	/* line number where error happened */
1363	__le32	s_last_error_time;	/* most recent time of an error */
1364	__le32	s_last_error_ino;	/* inode involved in last error */
1365	__le32	s_last_error_line;	/* line number where error happened */
1366	__le64	s_last_error_block;	/* block involved of last error */
1367	__u8	s_last_error_func[32] __nonstring;	/* function where the error happened */
1368#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1369	__u8	s_mount_opts[64];
1370	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1371	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1372	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1373	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1374	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1375	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1376	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1377	__le32	s_prj_quota_inum;	/* inode for tracking project quota */
1378	__le32	s_checksum_seed;	/* crc32c(uuid) if csum_seed set */
1379	__u8	s_wtime_hi;
1380	__u8	s_mtime_hi;
1381	__u8	s_mkfs_time_hi;
1382	__u8	s_lastcheck_hi;
1383	__u8	s_first_error_time_hi;
1384	__u8	s_last_error_time_hi;
1385	__u8	s_first_error_errcode;
1386	__u8    s_last_error_errcode;
1387	__le16  s_encoding;		/* Filename charset encoding */
1388	__le16  s_encoding_flags;	/* Filename charset encoding flags */
1389	__le32	s_reserved[95];		/* Padding to the end of the block */
1390	__le32	s_checksum;		/* crc32c(superblock) */
1391};
1392
1393#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1394
1395#ifdef __KERNEL__
1396
1397/*
1398 * run-time mount flags
1399 */
1400#define EXT4_MF_MNTDIR_SAMPLED		0x0001
1401#define EXT4_MF_FS_ABORTED		0x0002	/* Fatal error detected */
1402
1403#ifdef CONFIG_FS_ENCRYPTION
1404#define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_ctx.ctx != NULL)
1405#else
1406#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1407#endif
1408
1409/* Number of quota types we support */
1410#define EXT4_MAXQUOTAS 3
1411
1412#define EXT4_ENC_UTF8_12_1	1
1413
1414/*
1415 * Flags for ext4_sb_info.s_encoding_flags.
1416 */
1417#define EXT4_ENC_STRICT_MODE_FL	(1 << 0)
1418
1419#define ext4_has_strict_mode(sbi) \
1420	(sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1421
1422/*
1423 * fourth extended-fs super-block data in memory
1424 */
1425struct ext4_sb_info {
1426	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1427	unsigned long s_inodes_per_block;/* Number of inodes per block */
1428	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1429	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1430	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1431	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1432	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1433	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1434	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1435	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1436	unsigned long s_overhead;  /* # of fs overhead clusters */
1437	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1438	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1439	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1440	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1441	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1442	struct buffer_head * __rcu *s_group_desc;
1443	unsigned int s_mount_opt;
1444	unsigned int s_mount_opt2;
1445	unsigned int s_mount_flags;
1446	unsigned int s_def_mount_opt;
1447	ext4_fsblk_t s_sb_block;
1448	atomic64_t s_resv_clusters;
1449	kuid_t s_resuid;
1450	kgid_t s_resgid;
1451	unsigned short s_mount_state;
1452	unsigned short s_pad;
1453	int s_addr_per_block_bits;
1454	int s_desc_per_block_bits;
1455	int s_inode_size;
1456	int s_first_ino;
1457	unsigned int s_inode_readahead_blks;
1458	unsigned int s_inode_goal;
 
 
1459	u32 s_hash_seed[4];
1460	int s_def_hash_version;
1461	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1462	struct percpu_counter s_freeclusters_counter;
1463	struct percpu_counter s_freeinodes_counter;
1464	struct percpu_counter s_dirs_counter;
1465	struct percpu_counter s_dirtyclusters_counter;
1466	struct blockgroup_lock *s_blockgroup_lock;
1467	struct proc_dir_entry *s_proc;
1468	struct kobject s_kobj;
1469	struct completion s_kobj_unregister;
1470	struct super_block *s_sb;
1471#ifdef CONFIG_UNICODE
1472	struct unicode_map *s_encoding;
1473	__u16 s_encoding_flags;
1474#endif
1475
1476	/* Journaling */
1477	struct journal_s *s_journal;
1478	struct list_head s_orphan;
1479	struct mutex s_orphan_lock;
1480	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
 
1481	unsigned long s_commit_interval;
1482	u32 s_max_batch_time;
1483	u32 s_min_batch_time;
1484	struct block_device *journal_bdev;
1485#ifdef CONFIG_QUOTA
1486	/* Names of quota files with journalled quota */
1487	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1488	int s_jquota_fmt;			/* Format of quota to use */
1489#endif
1490	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1491	struct ext4_system_blocks __rcu *system_blks;
1492
1493#ifdef EXTENTS_STATS
1494	/* ext4 extents stats */
1495	unsigned long s_ext_min;
1496	unsigned long s_ext_max;
1497	unsigned long s_depth_max;
1498	spinlock_t s_ext_stats_lock;
1499	unsigned long s_ext_blocks;
1500	unsigned long s_ext_extents;
1501#endif
1502
1503	/* for buddy allocator */
1504	struct ext4_group_info ** __rcu *s_group_info;
1505	struct inode *s_buddy_cache;
1506	spinlock_t s_md_lock;
1507	unsigned short *s_mb_offsets;
1508	unsigned int *s_mb_maxs;
1509	unsigned int s_group_info_size;
1510	unsigned int s_mb_free_pending;
1511	struct list_head s_freed_data_list;	/* List of blocks to be freed
1512						   after commit completed */
1513
1514	/* tunables */
1515	unsigned long s_stripe;
1516	unsigned int s_mb_stream_request;
1517	unsigned int s_mb_max_to_scan;
1518	unsigned int s_mb_min_to_scan;
1519	unsigned int s_mb_stats;
1520	unsigned int s_mb_order2_reqs;
1521	unsigned int s_mb_group_prealloc;
1522	unsigned int s_mb_max_inode_prealloc;
1523	unsigned int s_max_dir_size_kb;
1524	/* where last allocation was done - for stream allocation */
1525	unsigned long s_mb_last_group;
1526	unsigned long s_mb_last_start;
1527	unsigned int s_mb_prefetch;
1528	unsigned int s_mb_prefetch_limit;
1529
1530	/* stats for buddy allocator */
1531	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1532	atomic_t s_bal_success;	/* we found long enough chunks */
1533	atomic_t s_bal_allocated;	/* in blocks */
1534	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1535	atomic_t s_bal_goals;	/* goal hits */
1536	atomic_t s_bal_breaks;	/* too long searches */
1537	atomic_t s_bal_2orders;	/* 2^order hits */
1538	spinlock_t s_bal_lock;
1539	unsigned long s_mb_buddies_generated;
1540	unsigned long long s_mb_generation_time;
1541	atomic_t s_mb_lost_chunks;
1542	atomic_t s_mb_preallocated;
1543	atomic_t s_mb_discarded;
1544	atomic_t s_lock_busy;
1545
1546	/* locality groups */
1547	struct ext4_locality_group __percpu *s_locality_groups;
1548
1549	/* for write statistics */
1550	unsigned long s_sectors_written_start;
1551	u64 s_kbytes_written;
1552
1553	/* the size of zero-out chunk */
1554	unsigned int s_extent_max_zeroout_kb;
1555
1556	unsigned int s_log_groups_per_flex;
1557	struct flex_groups * __rcu *s_flex_groups;
1558	ext4_group_t s_flex_groups_allocated;
1559
1560	/* workqueue for reserved extent conversions (buffered io) */
1561	struct workqueue_struct *rsv_conversion_wq;
1562
1563	/* timer for periodic error stats printing */
1564	struct timer_list s_err_report;
1565
1566	/* Lazy inode table initialization info */
1567	struct ext4_li_request *s_li_request;
1568	/* Wait multiplier for lazy initialization thread */
1569	unsigned int s_li_wait_mult;
1570
1571	/* Kernel thread for multiple mount protection */
1572	struct task_struct *s_mmp_tsk;
1573
1574	/* record the last minlen when FITRIM is called. */
1575	atomic_t s_last_trim_minblks;
1576
1577	/* Reference to checksum algorithm driver via cryptoapi */
1578	struct crypto_shash *s_chksum_driver;
1579
1580	/* Precomputed FS UUID checksum for seeding other checksums */
1581	__u32 s_csum_seed;
1582
1583	/* Reclaim extents from extent status tree */
1584	struct shrinker s_es_shrinker;
1585	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1586	long s_es_nr_inode;
1587	struct ext4_es_stats s_es_stats;
1588	struct mb_cache *s_ea_block_cache;
1589	struct mb_cache *s_ea_inode_cache;
1590	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1591
1592	/* Ratelimit ext4 messages. */
1593	struct ratelimit_state s_err_ratelimit_state;
1594	struct ratelimit_state s_warning_ratelimit_state;
1595	struct ratelimit_state s_msg_ratelimit_state;
1596	atomic_t s_warning_count;
1597	atomic_t s_msg_count;
1598
1599	/* Encryption context for '-o test_dummy_encryption' */
1600	struct fscrypt_dummy_context s_dummy_enc_ctx;
1601
1602	/*
1603	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1604	 * or EXTENTS flag.
1605	 */
1606	struct percpu_rw_semaphore s_writepages_rwsem;
1607	struct dax_device *s_daxdev;
1608#ifdef CONFIG_EXT4_DEBUG
1609	unsigned long s_simulate_fail;
1610#endif
1611	/* Record the errseq of the backing block device */
1612	errseq_t s_bdev_wb_err;
1613	spinlock_t s_bdev_wb_lock;
1614};
1615
1616static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1617{
1618	return sb->s_fs_info;
1619}
1620static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1621{
1622	return container_of(inode, struct ext4_inode_info, vfs_inode);
1623}
1624
 
 
 
 
 
 
1625static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1626{
1627	return ino == EXT4_ROOT_INO ||
 
 
1628		(ino >= EXT4_FIRST_INO(sb) &&
1629		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1630}
1631
1632/*
1633 * Returns: sbi->field[index]
1634 * Used to access an array element from the following sbi fields which require
1635 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1636 * - s_group_desc
1637 * - s_group_info
1638 * - s_flex_group
1639 */
1640#define sbi_array_rcu_deref(sbi, field, index)				   \
1641({									   \
1642	typeof(*((sbi)->field)) _v;					   \
1643	rcu_read_lock();						   \
1644	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1645	rcu_read_unlock();						   \
1646	_v;								   \
1647})
1648
1649/*
1650 * Simulate_fail codes
1651 */
1652#define EXT4_SIM_BBITMAP_EIO	1
1653#define EXT4_SIM_BBITMAP_CRC	2
1654#define EXT4_SIM_IBITMAP_EIO	3
1655#define EXT4_SIM_IBITMAP_CRC	4
1656#define EXT4_SIM_INODE_EIO	5
1657#define EXT4_SIM_INODE_CRC	6
1658#define EXT4_SIM_DIRBLOCK_EIO	7
1659#define EXT4_SIM_DIRBLOCK_CRC	8
1660
1661static inline bool ext4_simulate_fail(struct super_block *sb,
1662				     unsigned long code)
1663{
1664#ifdef CONFIG_EXT4_DEBUG
1665	struct ext4_sb_info *sbi = EXT4_SB(sb);
1666
1667	if (unlikely(sbi->s_simulate_fail == code)) {
1668		sbi->s_simulate_fail = 0;
1669		return true;
1670	}
1671#endif
1672	return false;
1673}
1674
1675static inline void ext4_simulate_fail_bh(struct super_block *sb,
1676					 struct buffer_head *bh,
1677					 unsigned long code)
1678{
1679	if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1680		clear_buffer_uptodate(bh);
1681}
1682
1683/*
1684 * Error number codes for s_{first,last}_error_errno
1685 *
1686 * Linux errno numbers are architecture specific, so we need to translate
1687 * them into something which is architecture independent.   We don't define
1688 * codes for all errno's; just the ones which are most likely to be the cause
1689 * of an ext4_error() call.
1690 */
1691#define EXT4_ERR_UNKNOWN	 1
1692#define EXT4_ERR_EIO		 2
1693#define EXT4_ERR_ENOMEM		 3
1694#define EXT4_ERR_EFSBADCRC	 4
1695#define EXT4_ERR_EFSCORRUPTED	 5
1696#define EXT4_ERR_ENOSPC		 6
1697#define EXT4_ERR_ENOKEY		 7
1698#define EXT4_ERR_EROFS		 8
1699#define EXT4_ERR_EFBIG		 9
1700#define EXT4_ERR_EEXIST		10
1701#define EXT4_ERR_ERANGE		11
1702#define EXT4_ERR_EOVERFLOW	12
1703#define EXT4_ERR_EBUSY		13
1704#define EXT4_ERR_ENOTDIR	14
1705#define EXT4_ERR_ENOTEMPTY	15
1706#define EXT4_ERR_ESHUTDOWN	16
1707#define EXT4_ERR_EFAULT		17
1708
1709/*
1710 * Inode dynamic state flags
1711 */
1712enum {
1713	EXT4_STATE_JDATA,		/* journaled data exists */
1714	EXT4_STATE_NEW,			/* inode is newly created */
1715	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1716	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1717	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1718	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
 
1719	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1720	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1721	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1722	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
1723	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
1724};
1725
1726#define EXT4_INODE_BIT_FNS(name, field, offset)				\
1727static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1728{									\
1729	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1730}									\
1731static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1732{									\
1733	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1734}									\
1735static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1736{									\
1737	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1738}
1739
1740/* Add these declarations here only so that these functions can be
1741 * found by name.  Otherwise, they are very hard to locate. */
1742static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1743static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1744static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1745EXT4_INODE_BIT_FNS(flag, flags, 0)
1746
1747/* Add these declarations here only so that these functions can be
1748 * found by name.  Otherwise, they are very hard to locate. */
1749static inline int ext4_test_inode_state(struct inode *inode, int bit);
1750static inline void ext4_set_inode_state(struct inode *inode, int bit);
1751static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1752#if (BITS_PER_LONG < 64)
1753EXT4_INODE_BIT_FNS(state, state_flags, 0)
1754
1755static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1756{
1757	(ei)->i_state_flags = 0;
1758}
1759#else
1760EXT4_INODE_BIT_FNS(state, flags, 32)
1761
1762static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1763{
1764	/* We depend on the fact that callers will set i_flags */
1765}
1766#endif
1767#else
1768/* Assume that user mode programs are passing in an ext4fs superblock, not
1769 * a kernel struct super_block.  This will allow us to call the feature-test
1770 * macros from user land. */
1771#define EXT4_SB(sb)	(sb)
1772#endif
1773
1774static inline bool ext4_verity_in_progress(struct inode *inode)
1775{
1776	return IS_ENABLED(CONFIG_FS_VERITY) &&
1777	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1778}
1779
1780#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1781
1782/*
1783 * Codes for operating systems
1784 */
1785#define EXT4_OS_LINUX		0
1786#define EXT4_OS_HURD		1
1787#define EXT4_OS_MASIX		2
1788#define EXT4_OS_FREEBSD		3
1789#define EXT4_OS_LITES		4
1790
1791/*
1792 * Revision levels
1793 */
1794#define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1795#define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1796
1797#define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1798#define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1799
1800#define EXT4_GOOD_OLD_INODE_SIZE 128
1801
1802#define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
1803#define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
1804#define EXT4_TIMESTAMP_MIN		S32_MIN
1805
1806/*
1807 * Feature set definitions
1808 */
1809
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1810#define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1811#define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1812#define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1813#define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1814#define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1815#define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1816#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1817#define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
1818
1819#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1820#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1821#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1822#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1823#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1824#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1825#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1826#define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1827#define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1828/*
1829 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1830 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1831 * all other data structures' checksums.  However, the METADATA_CSUM and
1832 * GDT_CSUM bits are mutually exclusive.
1833 */
1834#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1835#define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
1836#define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
1837#define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
1838
1839#define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1840#define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1841#define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1842#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1843#define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1844#define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1845#define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1846#define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1847#define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1848#define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1849#define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1850#define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
1851#define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1852#define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
1853#define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
1854#define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
1855
1856extern void ext4_update_dynamic_rev(struct super_block *sb);
1857
1858#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1859static inline bool ext4_has_feature_##name(struct super_block *sb) \
1860{ \
1861	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1862		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1863} \
1864static inline void ext4_set_feature_##name(struct super_block *sb) \
1865{ \
1866	ext4_update_dynamic_rev(sb); \
1867	EXT4_SB(sb)->s_es->s_feature_compat |= \
1868		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1869} \
1870static inline void ext4_clear_feature_##name(struct super_block *sb) \
1871{ \
1872	EXT4_SB(sb)->s_es->s_feature_compat &= \
1873		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1874}
1875
1876#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1877static inline bool ext4_has_feature_##name(struct super_block *sb) \
1878{ \
1879	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1880		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1881} \
1882static inline void ext4_set_feature_##name(struct super_block *sb) \
1883{ \
1884	ext4_update_dynamic_rev(sb); \
1885	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1886		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1887} \
1888static inline void ext4_clear_feature_##name(struct super_block *sb) \
1889{ \
1890	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1891		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1892}
1893
1894#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1895static inline bool ext4_has_feature_##name(struct super_block *sb) \
1896{ \
1897	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1898		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1899} \
1900static inline void ext4_set_feature_##name(struct super_block *sb) \
1901{ \
1902	ext4_update_dynamic_rev(sb); \
1903	EXT4_SB(sb)->s_es->s_feature_incompat |= \
1904		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1905} \
1906static inline void ext4_clear_feature_##name(struct super_block *sb) \
1907{ \
1908	EXT4_SB(sb)->s_es->s_feature_incompat &= \
1909		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1910}
1911
1912EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
1913EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
1914EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
1915EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
1916EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
1917EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
1918EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
1919EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
1920
1921EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
1922EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
1923EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
1924EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
1925EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
1926EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
1927EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
1928EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
1929EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
1930EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
1931EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
1932EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
1933EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
1934
1935EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
1936EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
1937EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
1938EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
1939EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
1940EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
1941EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
1942EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
1943EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
1944EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
1945EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
1946EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
1947EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
1948EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
1949EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
1950EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
1951
1952#define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1953#define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1954					 EXT4_FEATURE_INCOMPAT_META_BG)
1955#define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1956					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1957					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1958
1959#define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1960#define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1961					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1962					 EXT4_FEATURE_INCOMPAT_META_BG)
1963#define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1964					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1965					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1966
1967#define EXT4_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1968#define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1969					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1970					 EXT4_FEATURE_INCOMPAT_META_BG| \
1971					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1972					 EXT4_FEATURE_INCOMPAT_64BIT| \
1973					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1974					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1975					 EXT4_FEATURE_INCOMPAT_MMP | \
1976					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1977					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1978					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1979					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1980					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1981#define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1982					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1983					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1984					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1985					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1986					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1987					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1988					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1989					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1990					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1991					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
1992					 EXT4_FEATURE_RO_COMPAT_VERITY)
1993
1994#define EXTN_FEATURE_FUNCS(ver) \
1995static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1996{ \
1997	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1998		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1999} \
2000static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2001{ \
2002	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2003		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2004} \
2005static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2006{ \
2007	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2008		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2009}
2010
2011EXTN_FEATURE_FUNCS(2)
2012EXTN_FEATURE_FUNCS(3)
2013EXTN_FEATURE_FUNCS(4)
2014
2015static inline bool ext4_has_compat_features(struct super_block *sb)
2016{
2017	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2018}
2019static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2020{
2021	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2022}
2023static inline bool ext4_has_incompat_features(struct super_block *sb)
2024{
2025	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2026}
2027
2028/*
2029 * Superblock flags
2030 */
2031#define EXT4_FLAGS_RESIZING	0
2032#define EXT4_FLAGS_SHUTDOWN	1
2033#define EXT4_FLAGS_BDEV_IS_DAX	2
2034
2035static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2036{
2037	return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2038}
2039
2040
2041/*
2042 * Default values for user and/or group using reserved blocks
2043 */
2044#define	EXT4_DEF_RESUID		0
2045#define	EXT4_DEF_RESGID		0
2046
2047/*
2048 * Default project ID
2049 */
2050#define	EXT4_DEF_PROJID		0
2051
2052#define EXT4_DEF_INODE_READAHEAD_BLKS	32
2053
2054/*
2055 * Default mount options
2056 */
2057#define EXT4_DEFM_DEBUG		0x0001
2058#define EXT4_DEFM_BSDGROUPS	0x0002
2059#define EXT4_DEFM_XATTR_USER	0x0004
2060#define EXT4_DEFM_ACL		0x0008
2061#define EXT4_DEFM_UID16		0x0010
2062#define EXT4_DEFM_JMODE		0x0060
2063#define EXT4_DEFM_JMODE_DATA	0x0020
2064#define EXT4_DEFM_JMODE_ORDERED	0x0040
2065#define EXT4_DEFM_JMODE_WBACK	0x0060
2066#define EXT4_DEFM_NOBARRIER	0x0100
2067#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2068#define EXT4_DEFM_DISCARD	0x0400
2069#define EXT4_DEFM_NODELALLOC	0x0800
2070
2071/*
2072 * Default journal batch times
2073 */
2074#define EXT4_DEF_MIN_BATCH_TIME	0
2075#define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
2076
2077/*
2078 * Minimum number of groups in a flexgroup before we separate out
2079 * directories into the first block group of a flexgroup
2080 */
2081#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
2082
2083/*
2084 * Structure of a directory entry
2085 */
2086#define EXT4_NAME_LEN 255
2087
2088struct ext4_dir_entry {
2089	__le32	inode;			/* Inode number */
2090	__le16	rec_len;		/* Directory entry length */
2091	__le16	name_len;		/* Name length */
2092	char	name[EXT4_NAME_LEN];	/* File name */
2093};
2094
2095/*
2096 * The new version of the directory entry.  Since EXT4 structures are
2097 * stored in intel byte order, and the name_len field could never be
2098 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2099 * file_type field.
2100 */
2101struct ext4_dir_entry_2 {
2102	__le32	inode;			/* Inode number */
2103	__le16	rec_len;		/* Directory entry length */
2104	__u8	name_len;		/* Name length */
2105	__u8	file_type;		/* See file type macros EXT4_FT_* below */
2106	char	name[EXT4_NAME_LEN];	/* File name */
2107};
2108
2109/*
2110 * This is a bogus directory entry at the end of each leaf block that
2111 * records checksums.
2112 */
2113struct ext4_dir_entry_tail {
2114	__le32	det_reserved_zero1;	/* Pretend to be unused */
2115	__le16	det_rec_len;		/* 12 */
2116	__u8	det_reserved_zero2;	/* Zero name length */
2117	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2118	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2119};
2120
2121#define EXT4_DIRENT_TAIL(block, blocksize) \
2122	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2123					((blocksize) - \
2124					 sizeof(struct ext4_dir_entry_tail))))
2125
2126/*
2127 * Ext4 directory file types.  Only the low 3 bits are used.  The
2128 * other bits are reserved for now.
2129 */
2130#define EXT4_FT_UNKNOWN		0
2131#define EXT4_FT_REG_FILE	1
2132#define EXT4_FT_DIR		2
2133#define EXT4_FT_CHRDEV		3
2134#define EXT4_FT_BLKDEV		4
2135#define EXT4_FT_FIFO		5
2136#define EXT4_FT_SOCK		6
2137#define EXT4_FT_SYMLINK		7
2138
2139#define EXT4_FT_MAX		8
2140
2141#define EXT4_FT_DIR_CSUM	0xDE
2142
2143/*
2144 * EXT4_DIR_PAD defines the directory entries boundaries
2145 *
2146 * NOTE: It must be a multiple of 4
2147 */
2148#define EXT4_DIR_PAD			4
2149#define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2150#define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
2151					 ~EXT4_DIR_ROUND)
2152#define EXT4_MAX_REC_LEN		((1<<16)-1)
2153
2154/*
2155 * If we ever get support for fs block sizes > page_size, we'll need
2156 * to remove the #if statements in the next two functions...
2157 */
2158static inline unsigned int
2159ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2160{
2161	unsigned len = le16_to_cpu(dlen);
2162
2163#if (PAGE_SIZE >= 65536)
2164	if (len == EXT4_MAX_REC_LEN || len == 0)
2165		return blocksize;
2166	return (len & 65532) | ((len & 3) << 16);
2167#else
2168	return len;
2169#endif
2170}
2171
2172static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2173{
2174	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2175		BUG();
2176#if (PAGE_SIZE >= 65536)
2177	if (len < 65536)
2178		return cpu_to_le16(len);
2179	if (len == blocksize) {
2180		if (blocksize == 65536)
2181			return cpu_to_le16(EXT4_MAX_REC_LEN);
2182		else
2183			return cpu_to_le16(0);
2184	}
2185	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2186#else
2187	return cpu_to_le16(len);
2188#endif
2189}
2190
2191/*
2192 * Hash Tree Directory indexing
2193 * (c) Daniel Phillips, 2001
2194 */
2195
2196#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
 
2197		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2198#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2199		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2200#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2201
2202/* Legal values for the dx_root hash_version field: */
2203
2204#define DX_HASH_LEGACY			0
2205#define DX_HASH_HALF_MD4		1
2206#define DX_HASH_TEA			2
2207#define DX_HASH_LEGACY_UNSIGNED		3
2208#define DX_HASH_HALF_MD4_UNSIGNED	4
2209#define DX_HASH_TEA_UNSIGNED		5
2210
2211static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2212			      const void *address, unsigned int length)
2213{
2214	struct {
2215		struct shash_desc shash;
2216		char ctx[4];
2217	} desc;
2218
2219	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2220
2221	desc.shash.tfm = sbi->s_chksum_driver;
 
2222	*(u32 *)desc.ctx = crc;
2223
2224	BUG_ON(crypto_shash_update(&desc.shash, address, length));
 
2225
2226	return *(u32 *)desc.ctx;
2227}
2228
2229#ifdef __KERNEL__
2230
2231/* hash info structure used by the directory hash */
2232struct dx_hash_info
2233{
2234	u32		hash;
2235	u32		minor_hash;
2236	int		hash_version;
2237	u32		*seed;
2238};
2239
2240
2241/* 32 and 64 bit signed EOF for dx directories */
2242#define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2243#define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2244
2245
2246/*
2247 * Control parameters used by ext4_htree_next_block
2248 */
2249#define HASH_NB_ALWAYS		1
2250
2251struct ext4_filename {
2252	const struct qstr *usr_fname;
2253	struct fscrypt_str disk_name;
2254	struct dx_hash_info hinfo;
2255#ifdef CONFIG_FS_ENCRYPTION
2256	struct fscrypt_str crypto_buf;
2257#endif
2258#ifdef CONFIG_UNICODE
2259	struct fscrypt_str cf_name;
2260#endif
2261};
2262
2263#define fname_name(p) ((p)->disk_name.name)
2264#define fname_len(p)  ((p)->disk_name.len)
2265
2266/*
2267 * Describe an inode's exact location on disk and in memory
2268 */
2269struct ext4_iloc
2270{
2271	struct buffer_head *bh;
2272	unsigned long offset;
2273	ext4_group_t block_group;
2274};
2275
2276static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2277{
2278	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2279}
2280
2281static inline bool ext4_is_quota_file(struct inode *inode)
2282{
2283	return IS_NOQUOTA(inode) &&
2284	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2285}
2286
2287/*
2288 * This structure is stuffed into the struct file's private_data field
2289 * for directories.  It is where we put information so that we can do
2290 * readdir operations in hash tree order.
2291 */
2292struct dir_private_info {
2293	struct rb_root	root;
2294	struct rb_node	*curr_node;
2295	struct fname	*extra_fname;
2296	loff_t		last_pos;
2297	__u32		curr_hash;
2298	__u32		curr_minor_hash;
2299	__u32		next_hash;
2300};
2301
2302/* calculate the first block number of the group */
2303static inline ext4_fsblk_t
2304ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2305{
2306	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2307		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2308}
2309
2310/*
2311 * Special error return code only used by dx_probe() and its callers.
2312 */
2313#define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2314
2315/* htree levels for ext4 */
2316#define	EXT4_HTREE_LEVEL_COMPAT	2
2317#define	EXT4_HTREE_LEVEL	3
2318
2319static inline int ext4_dir_htree_level(struct super_block *sb)
2320{
2321	return ext4_has_feature_largedir(sb) ?
2322		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2323}
2324
2325/*
2326 * Timeout and state flag for lazy initialization inode thread.
2327 */
2328#define EXT4_DEF_LI_WAIT_MULT			10
2329#define EXT4_DEF_LI_MAX_START_DELAY		5
2330#define EXT4_LAZYINIT_QUIT			0x0001
2331#define EXT4_LAZYINIT_RUNNING			0x0002
2332
2333/*
2334 * Lazy inode table initialization info
2335 */
2336struct ext4_lazy_init {
2337	unsigned long		li_state;
2338	struct list_head	li_request_list;
2339	struct mutex		li_list_mtx;
2340};
2341
2342enum ext4_li_mode {
2343	EXT4_LI_MODE_PREFETCH_BBITMAP,
2344	EXT4_LI_MODE_ITABLE,
2345};
2346
2347struct ext4_li_request {
2348	struct super_block	*lr_super;
2349	enum ext4_li_mode	lr_mode;
2350	ext4_group_t		lr_first_not_zeroed;
2351	ext4_group_t		lr_next_group;
2352	struct list_head	lr_request;
2353	unsigned long		lr_next_sched;
2354	unsigned long		lr_timeout;
2355};
2356
2357struct ext4_features {
2358	struct kobject f_kobj;
2359	struct completion f_kobj_unregister;
2360};
2361
2362/*
2363 * This structure will be used for multiple mount protection. It will be
2364 * written into the block number saved in the s_mmp_block field in the
2365 * superblock. Programs that check MMP should assume that if
2366 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2367 * to use the filesystem, regardless of how old the timestamp is.
2368 */
2369#define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2370#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2371#define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2372#define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2373
2374struct mmp_struct {
2375	__le32	mmp_magic;		/* Magic number for MMP */
2376	__le32	mmp_seq;		/* Sequence no. updated periodically */
2377
2378	/*
2379	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2380	 * purposes and do not affect the correctness of the algorithm
2381	 */
2382	__le64	mmp_time;		/* Time last updated */
2383	char	mmp_nodename[64];	/* Node which last updated MMP block */
2384	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2385
2386	/*
2387	 * mmp_check_interval is used to verify if the MMP block has been
2388	 * updated on the block device. The value is updated based on the
2389	 * maximum time to write the MMP block during an update cycle.
2390	 */
2391	__le16	mmp_check_interval;
2392
2393	__le16	mmp_pad1;
2394	__le32	mmp_pad2[226];
2395	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2396};
2397
2398/* arguments passed to the mmp thread */
2399struct mmpd_data {
2400	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2401	struct super_block *sb;  /* super block of the fs */
2402};
2403
2404/*
2405 * Check interval multiplier
2406 * The MMP block is written every update interval and initially checked every
2407 * update interval x the multiplier (the value is then adapted based on the
2408 * write latency). The reason is that writes can be delayed under load and we
2409 * don't want readers to incorrectly assume that the filesystem is no longer
2410 * in use.
2411 */
2412#define EXT4_MMP_CHECK_MULT		2UL
2413
2414/*
2415 * Minimum interval for MMP checking in seconds.
2416 */
2417#define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2418
2419/*
2420 * Maximum interval for MMP checking in seconds.
2421 */
2422#define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2423
2424/*
2425 * Function prototypes
2426 */
2427
2428/*
2429 * Ok, these declarations are also in <linux/kernel.h> but none of the
2430 * ext4 source programs needs to include it so they are duplicated here.
2431 */
2432# define NORET_TYPE	/**/
2433# define ATTRIB_NORET	__attribute__((noreturn))
2434# define NORET_AND	noreturn,
2435
2436/* bitmap.c */
2437extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2438void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2439				struct ext4_group_desc *gdp,
2440				struct buffer_head *bh, int sz);
2441int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2442				  struct ext4_group_desc *gdp,
2443				  struct buffer_head *bh, int sz);
2444void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2445				struct ext4_group_desc *gdp,
2446				struct buffer_head *bh);
2447int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2448				  struct ext4_group_desc *gdp,
2449				  struct buffer_head *bh);
2450
2451/* balloc.c */
2452extern void ext4_get_group_no_and_offset(struct super_block *sb,
2453					 ext4_fsblk_t blocknr,
2454					 ext4_group_t *blockgrpp,
2455					 ext4_grpblk_t *offsetp);
2456extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2457					  ext4_fsblk_t block);
2458
2459extern unsigned int ext4_block_group(struct super_block *sb,
2460			ext4_fsblk_t blocknr);
2461extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2462			ext4_fsblk_t blocknr);
2463extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2464extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2465			ext4_group_t group);
2466extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2467					 ext4_fsblk_t goal,
2468					 unsigned int flags,
2469					 unsigned long *count,
2470					 int *errp);
2471extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2472				    s64 nclusters, unsigned int flags);
2473extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2474extern void ext4_check_blocks_bitmap(struct super_block *);
2475extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2476						    ext4_group_t block_group,
2477						    struct buffer_head ** bh);
2478extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2479
2480extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2481						ext4_group_t block_group,
2482						bool ignore_locked);
2483extern int ext4_wait_block_bitmap(struct super_block *sb,
2484				  ext4_group_t block_group,
2485				  struct buffer_head *bh);
2486extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2487						  ext4_group_t block_group);
 
 
 
 
2488extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2489					      ext4_group_t block_group,
2490					      struct ext4_group_desc *gdp);
 
 
 
2491ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2492
2493#ifdef CONFIG_UNICODE
2494extern void ext4_fname_setup_ci_filename(struct inode *dir,
2495					 const struct qstr *iname,
2496					 struct fscrypt_str *fname);
2497#endif
2498
2499#ifdef CONFIG_FS_ENCRYPTION
2500static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2501						const struct fscrypt_name *src)
2502{
2503	memset(dst, 0, sizeof(*dst));
2504
2505	dst->usr_fname = src->usr_fname;
2506	dst->disk_name = src->disk_name;
2507	dst->hinfo.hash = src->hash;
2508	dst->hinfo.minor_hash = src->minor_hash;
2509	dst->crypto_buf = src->crypto_buf;
2510}
2511
2512static inline int ext4_fname_setup_filename(struct inode *dir,
2513					    const struct qstr *iname,
2514					    int lookup,
2515					    struct ext4_filename *fname)
2516{
2517	struct fscrypt_name name;
2518	int err;
2519
2520	err = fscrypt_setup_filename(dir, iname, lookup, &name);
2521	if (err)
2522		return err;
2523
2524	ext4_fname_from_fscrypt_name(fname, &name);
2525
2526#ifdef CONFIG_UNICODE
2527	ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2528#endif
2529	return 0;
2530}
2531
2532static inline int ext4_fname_prepare_lookup(struct inode *dir,
2533					    struct dentry *dentry,
2534					    struct ext4_filename *fname)
2535{
2536	struct fscrypt_name name;
2537	int err;
2538
2539	err = fscrypt_prepare_lookup(dir, dentry, &name);
2540	if (err)
2541		return err;
2542
2543	ext4_fname_from_fscrypt_name(fname, &name);
2544
2545#ifdef CONFIG_UNICODE
2546	ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2547#endif
2548	return 0;
2549}
2550
2551static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2552{
2553	struct fscrypt_name name;
2554
2555	name.crypto_buf = fname->crypto_buf;
2556	fscrypt_free_filename(&name);
2557
2558	fname->crypto_buf.name = NULL;
2559	fname->usr_fname = NULL;
2560	fname->disk_name.name = NULL;
2561
2562#ifdef CONFIG_UNICODE
2563	kfree(fname->cf_name.name);
2564	fname->cf_name.name = NULL;
2565#endif
2566}
2567#else /* !CONFIG_FS_ENCRYPTION */
2568static inline int ext4_fname_setup_filename(struct inode *dir,
2569					    const struct qstr *iname,
2570					    int lookup,
2571					    struct ext4_filename *fname)
2572{
2573	fname->usr_fname = iname;
2574	fname->disk_name.name = (unsigned char *) iname->name;
2575	fname->disk_name.len = iname->len;
2576
2577#ifdef CONFIG_UNICODE
2578	ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2579#endif
2580
2581	return 0;
2582}
2583
2584static inline int ext4_fname_prepare_lookup(struct inode *dir,
2585					    struct dentry *dentry,
2586					    struct ext4_filename *fname)
2587{
2588	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2589}
2590
2591static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2592{
2593#ifdef CONFIG_UNICODE
2594	kfree(fname->cf_name.name);
2595	fname->cf_name.name = NULL;
2596#endif
2597}
2598#endif /* !CONFIG_FS_ENCRYPTION */
2599
2600/* dir.c */
2601extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2602				  struct file *,
2603				  struct ext4_dir_entry_2 *,
2604				  struct buffer_head *, char *, int,
2605				  unsigned int);
2606#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)	\
2607	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2608					(de), (bh), (buf), (size), (offset)))
2609extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2610				__u32 minor_hash,
2611				struct ext4_dir_entry_2 *dirent,
2612				struct fscrypt_str *ent_name);
2613extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2614extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2615			     struct buffer_head *bh,
2616			     void *buf, int buf_size,
2617			     struct ext4_filename *fname,
2618			     struct ext4_dir_entry_2 **dest_de);
2619void ext4_insert_dentry(struct inode *inode,
2620			struct ext4_dir_entry_2 *de,
2621			int buf_size,
2622			struct ext4_filename *fname);
2623static inline void ext4_update_dx_flag(struct inode *inode)
2624{
2625	if (!ext4_has_feature_dir_index(inode->i_sb)) {
2626		/* ext4_iget() should have caught this... */
2627		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2628		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2629	}
2630}
2631static const unsigned char ext4_filetype_table[] = {
2632	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2633};
2634
2635static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2636{
2637	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2638		return DT_UNKNOWN;
2639
2640	return ext4_filetype_table[filetype];
2641}
2642extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2643			     void *buf, int buf_size);
2644
2645/* fsync.c */
2646extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
 
2647
2648/* hash.c */
2649extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2650			  struct dx_hash_info *hinfo);
2651
2652/* ialloc.c */
2653extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2654				      const struct qstr *qstr, __u32 goal,
2655				      uid_t *owner, __u32 i_flags,
2656				      int handle_type, unsigned int line_no,
2657				      int nblocks);
2658
2659#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2660	__ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2661			 i_flags, 0, 0, 0)
2662#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2663				    type, nblocks)		    \
2664	__ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2665			 0, (type), __LINE__, (nblocks))
2666
2667
2668extern void ext4_free_inode(handle_t *, struct inode *);
2669extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2670extern unsigned long ext4_count_free_inodes(struct super_block *);
2671extern unsigned long ext4_count_dirs(struct super_block *);
2672extern void ext4_check_inodes_bitmap(struct super_block *);
2673extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2674extern int ext4_init_inode_table(struct super_block *sb,
2675				 ext4_group_t group, int barrier);
2676extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2677
2678/* mballoc.c */
2679extern const struct seq_operations ext4_mb_seq_groups_ops;
2680extern long ext4_mb_stats;
2681extern long ext4_mb_max_to_scan;
2682extern int ext4_mb_init(struct super_block *);
2683extern int ext4_mb_release(struct super_block *);
2684extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2685				struct ext4_allocation_request *, int *);
2686extern int ext4_mb_reserve_blocks(struct super_block *, int);
2687extern void ext4_discard_preallocations(struct inode *, unsigned int);
2688extern int __init ext4_init_mballoc(void);
2689extern void ext4_exit_mballoc(void);
2690extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2691				     ext4_group_t group,
2692				     unsigned int nr, int *cnt);
2693extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2694				  unsigned int nr);
2695
2696extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2697			     struct buffer_head *bh, ext4_fsblk_t block,
2698			     unsigned long count, int flags);
2699extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2700				   ext4_group_t ngroups);
2701extern int ext4_mb_add_groupinfo(struct super_block *sb,
2702		ext4_group_t i, struct ext4_group_desc *desc);
2703extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2704				ext4_fsblk_t block, unsigned long count);
2705extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2706extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2707
2708/* inode.c */
2709int ext4_inode_is_fast_symlink(struct inode *inode);
2710struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2711struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2712int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2713		     bool wait, struct buffer_head **bhs);
2714int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2715			     struct buffer_head *bh_result, int create);
2716int ext4_get_block(struct inode *inode, sector_t iblock,
2717		   struct buffer_head *bh_result, int create);
2718int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2719			   struct buffer_head *bh, int create);
2720int ext4_walk_page_buffers(handle_t *handle,
2721			   struct buffer_head *head,
2722			   unsigned from,
2723			   unsigned to,
2724			   int *partial,
2725			   int (*fn)(handle_t *handle,
2726				     struct buffer_head *bh));
2727int do_journal_get_write_access(handle_t *handle,
2728				struct buffer_head *bh);
2729#define FALL_BACK_TO_NONDELALLOC 1
2730#define CONVERT_INLINE_DATA	 2
2731
2732typedef enum {
2733	EXT4_IGET_NORMAL =	0,
2734	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
2735	EXT4_IGET_HANDLE = 	0x0002	/* Inode # is from a handle */
2736} ext4_iget_flags;
2737
2738extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2739				 ext4_iget_flags flags, const char *function,
2740				 unsigned int line);
2741
2742#define ext4_iget(sb, ino, flags) \
2743	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2744
 
2745extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2746extern int  ext4_setattr(struct dentry *, struct iattr *);
2747extern int  ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
 
2748extern void ext4_evict_inode(struct inode *);
2749extern void ext4_clear_inode(struct inode *);
2750extern int  ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2751extern int  ext4_sync_inode(handle_t *, struct inode *);
2752extern void ext4_dirty_inode(struct inode *, int);
2753extern int ext4_change_inode_journal_flag(struct inode *, int);
2754extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2755extern int ext4_inode_attach_jinode(struct inode *inode);
2756extern int ext4_can_truncate(struct inode *inode);
2757extern int ext4_truncate(struct inode *);
2758extern int ext4_break_layouts(struct inode *);
2759extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2760extern void ext4_set_inode_flags(struct inode *, bool init);
 
2761extern int ext4_alloc_da_blocks(struct inode *inode);
2762extern void ext4_set_aops(struct inode *inode);
2763extern int ext4_writepage_trans_blocks(struct inode *);
2764extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2765extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2766			     loff_t lstart, loff_t lend);
2767extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2768extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2769extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2770extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2771extern void ext4_da_release_space(struct inode *inode, int to_free);
2772extern void ext4_da_update_reserve_space(struct inode *inode,
2773					int used, int quota_claim);
2774extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2775			      ext4_fsblk_t pblk, ext4_lblk_t len);
2776
2777/* indirect.c */
2778extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2779				struct ext4_map_blocks *map, int flags);
2780extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2781extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2782extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2783				 ext4_lblk_t start, ext4_lblk_t end);
 
 
2784
2785/* ioctl.c */
2786extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2787extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2788
2789/* migrate.c */
2790extern int ext4_ext_migrate(struct inode *);
2791extern int ext4_ind_migrate(struct inode *inode);
2792
2793/* namei.c */
2794extern int ext4_dirblock_csum_verify(struct inode *inode,
2795				     struct buffer_head *bh);
2796extern int ext4_orphan_add(handle_t *, struct inode *);
2797extern int ext4_orphan_del(handle_t *, struct inode *);
2798extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2799				__u32 start_minor_hash, __u32 *next_hash);
2800extern int ext4_search_dir(struct buffer_head *bh,
2801			   char *search_buf,
2802			   int buf_size,
2803			   struct inode *dir,
2804			   struct ext4_filename *fname,
2805			   unsigned int offset,
2806			   struct ext4_dir_entry_2 **res_dir);
2807extern int ext4_generic_delete_entry(struct inode *dir,
2808				     struct ext4_dir_entry_2 *de_del,
2809				     struct buffer_head *bh,
2810				     void *entry_buf,
2811				     int buf_size,
2812				     int csum_size);
2813extern bool ext4_empty_dir(struct inode *inode);
2814
2815/* resize.c */
2816extern void ext4_kvfree_array_rcu(void *to_free);
2817extern int ext4_group_add(struct super_block *sb,
2818				struct ext4_new_group_data *input);
2819extern int ext4_group_extend(struct super_block *sb,
2820				struct ext4_super_block *es,
2821				ext4_fsblk_t n_blocks_count);
2822extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2823
2824/* super.c */
2825extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2826					 sector_t block, int op_flags);
2827extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2828extern int ext4_calculate_overhead(struct super_block *sb);
2829extern void ext4_superblock_csum_set(struct super_block *sb);
2830extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2831				    ext4_group_t ngroup);
2832extern const char *ext4_decode_error(struct super_block *sb, int errno,
2833				     char nbuf[16]);
2834extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2835					     ext4_group_t block_group,
2836					     unsigned int flags);
2837
2838extern __printf(6, 7)
2839void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64,
2840		  const char *, ...);
2841extern __printf(6, 7)
2842void __ext4_error_inode(struct inode *, const char *, unsigned int,
2843			ext4_fsblk_t, int, const char *, ...);
 
 
2844extern __printf(5, 6)
2845void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2846		     const char *, ...);
2847extern void __ext4_std_error(struct super_block *, const char *,
2848			     unsigned int, int);
2849extern __printf(5, 6)
2850void __ext4_abort(struct super_block *, const char *, unsigned int, int,
2851		  const char *, ...);
 
 
2852extern __printf(4, 5)
2853void __ext4_warning(struct super_block *, const char *, unsigned int,
2854		    const char *, ...);
2855extern __printf(4, 5)
2856void __ext4_warning_inode(const struct inode *inode, const char *function,
2857			  unsigned int line, const char *fmt, ...);
2858extern __printf(3, 4)
2859void __ext4_msg(struct super_block *, const char *, const char *, ...);
2860extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2861			   const char *, unsigned int, const char *);
 
 
2862extern __printf(7, 8)
2863void __ext4_grp_locked_error(const char *, unsigned int,
2864			     struct super_block *, ext4_group_t,
2865			     unsigned long, ext4_fsblk_t,
2866			     const char *, ...);
2867
2868#define EXT4_ERROR_INODE(inode, fmt, a...) \
2869	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2870
2871#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)			\
2872	__ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
2873
2874#define ext4_error_inode_block(inode, block, err, fmt, a...)		\
2875	__ext4_error_inode((inode), __func__, __LINE__, (block), (err),	\
2876			   (fmt), ## a)
2877
2878#define EXT4_ERROR_FILE(file, block, fmt, a...)				\
2879	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2880
2881#ifdef CONFIG_PRINTK
2882
2883#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2884	__ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
2885#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
2886	__ext4_error_inode((inode), (func), (line), (block), 		\
2887			   (err), (fmt), ##__VA_ARGS__)
2888#define ext4_error_file(file, func, line, block, fmt, ...)		\
2889	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2890#define ext4_error(sb, fmt, ...)					\
2891	__ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__)
2892#define ext4_error_err(sb, err, fmt, ...)				\
2893	__ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__)
2894#define ext4_abort(sb, err, fmt, ...)					\
2895	__ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__)
2896#define ext4_warning(sb, fmt, ...)					\
2897	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2898#define ext4_warning_inode(inode, fmt, ...)				\
2899	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2900#define ext4_msg(sb, level, fmt, ...)				\
2901	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2902#define dump_mmp_msg(sb, mmp, msg)					\
2903	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2904#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2905	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2906				fmt, ##__VA_ARGS__)
2907
2908#else
2909
2910#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2911do {									\
2912	no_printk(fmt, ##__VA_ARGS__);					\
2913	__ext4_error_inode(inode, "", 0, block, 0, " ");		\
2914} while (0)
2915#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
2916do {									\
2917	no_printk(fmt, ##__VA_ARGS__);					\
2918	__ext4_error_inode(inode, "", 0, block, err, " ");		\
2919} while (0)
2920#define ext4_error_file(file, func, line, block, fmt, ...)		\
2921do {									\
2922	no_printk(fmt, ##__VA_ARGS__);					\
2923	__ext4_error_file(file, "", 0, block, " ");			\
2924} while (0)
2925#define ext4_error(sb, fmt, ...)					\
2926do {									\
2927	no_printk(fmt, ##__VA_ARGS__);					\
2928	__ext4_error(sb, "", 0, 0, 0, " ");				\
2929} while (0)
2930#define ext4_error_err(sb, err, fmt, ...)				\
2931do {									\
2932	no_printk(fmt, ##__VA_ARGS__);					\
2933	__ext4_error(sb, "", 0, err, 0, " ");				\
2934} while (0)
2935#define ext4_abort(sb, err, fmt, ...)					\
2936do {									\
2937	no_printk(fmt, ##__VA_ARGS__);					\
2938	__ext4_abort(sb, "", 0, err, " ");				\
2939} while (0)
2940#define ext4_warning(sb, fmt, ...)					\
2941do {									\
2942	no_printk(fmt, ##__VA_ARGS__);					\
2943	__ext4_warning(sb, "", 0, " ");					\
2944} while (0)
2945#define ext4_warning_inode(inode, fmt, ...)				\
2946do {									\
2947	no_printk(fmt, ##__VA_ARGS__);					\
2948	__ext4_warning_inode(inode, "", 0, " ");			\
2949} while (0)
2950#define ext4_msg(sb, level, fmt, ...)					\
2951do {									\
2952	no_printk(fmt, ##__VA_ARGS__);					\
2953	__ext4_msg(sb, "", " ");					\
2954} while (0)
2955#define dump_mmp_msg(sb, mmp, msg)					\
2956	__dump_mmp_msg(sb, mmp, "", 0, "")
2957#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2958do {									\
2959	no_printk(fmt, ##__VA_ARGS__);				\
2960	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
2961} while (0)
2962
2963#endif
2964
2965extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2966				      struct ext4_group_desc *bg);
2967extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2968				      struct ext4_group_desc *bg);
2969extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2970				     struct ext4_group_desc *bg);
2971extern __u32 ext4_free_group_clusters(struct super_block *sb,
2972				      struct ext4_group_desc *bg);
2973extern __u32 ext4_free_inodes_count(struct super_block *sb,
2974				 struct ext4_group_desc *bg);
2975extern __u32 ext4_used_dirs_count(struct super_block *sb,
2976				struct ext4_group_desc *bg);
2977extern __u32 ext4_itable_unused_count(struct super_block *sb,
2978				   struct ext4_group_desc *bg);
2979extern void ext4_block_bitmap_set(struct super_block *sb,
2980				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2981extern void ext4_inode_bitmap_set(struct super_block *sb,
2982				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2983extern void ext4_inode_table_set(struct super_block *sb,
2984				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2985extern void ext4_free_group_clusters_set(struct super_block *sb,
2986					 struct ext4_group_desc *bg,
2987					 __u32 count);
2988extern void ext4_free_inodes_set(struct super_block *sb,
2989				struct ext4_group_desc *bg, __u32 count);
2990extern void ext4_used_dirs_set(struct super_block *sb,
2991				struct ext4_group_desc *bg, __u32 count);
2992extern void ext4_itable_unused_set(struct super_block *sb,
2993				   struct ext4_group_desc *bg, __u32 count);
2994extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2995				       struct ext4_group_desc *gdp);
2996extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2997				     struct ext4_group_desc *gdp);
2998extern int ext4_register_li_request(struct super_block *sb,
2999				    ext4_group_t first_not_zeroed);
3000
3001static inline int ext4_has_metadata_csum(struct super_block *sb)
3002{
3003	WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3004		     !EXT4_SB(sb)->s_chksum_driver);
3005
3006	return ext4_has_feature_metadata_csum(sb) &&
3007	       (EXT4_SB(sb)->s_chksum_driver != NULL);
3008}
3009
3010static inline int ext4_has_group_desc_csum(struct super_block *sb)
3011{
3012	return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
 
 
3013}
3014
3015static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3016{
3017	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
3018		le32_to_cpu(es->s_blocks_count_lo);
3019}
3020
3021static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3022{
3023	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
3024		le32_to_cpu(es->s_r_blocks_count_lo);
3025}
3026
3027static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3028{
3029	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
3030		le32_to_cpu(es->s_free_blocks_count_lo);
3031}
3032
3033static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3034					 ext4_fsblk_t blk)
3035{
3036	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3037	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3038}
3039
3040static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3041					      ext4_fsblk_t blk)
3042{
3043	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3044	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3045}
3046
3047static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3048					   ext4_fsblk_t blk)
3049{
3050	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3051	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3052}
3053
3054static inline loff_t ext4_isize(struct super_block *sb,
3055				struct ext4_inode *raw_inode)
3056{
3057	if (ext4_has_feature_largedir(sb) ||
3058	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3059		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3060			le32_to_cpu(raw_inode->i_size_lo);
3061
3062	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3063}
3064
3065static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3066{
3067	raw_inode->i_size_lo = cpu_to_le32(i_size);
3068	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3069}
3070
3071static inline
3072struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3073					    ext4_group_t group)
3074{
3075	 struct ext4_group_info **grp_info;
3076	 long indexv, indexh;
3077	 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3078	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3079	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3080	 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3081	 return grp_info[indexh];
3082}
3083
3084/*
3085 * Reading s_groups_count requires using smp_rmb() afterwards.  See
3086 * the locking protocol documented in the comments of ext4_group_add()
3087 * in resize.c
3088 */
3089static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3090{
3091	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
3092
3093	smp_rmb();
3094	return ngroups;
3095}
3096
3097static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3098					     ext4_group_t block_group)
3099{
3100	return block_group >> sbi->s_log_groups_per_flex;
3101}
3102
3103static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3104{
3105	return 1 << sbi->s_log_groups_per_flex;
3106}
3107
3108#define ext4_std_error(sb, errno)				\
3109do {								\
3110	if ((errno))						\
3111		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3112} while (0)
3113
3114#ifdef CONFIG_SMP
3115/* Each CPU can accumulate percpu_counter_batch clusters in their local
3116 * counters. So we need to make sure we have free clusters more
3117 * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3118 */
3119#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3120#else
3121#define EXT4_FREECLUSTERS_WATERMARK 0
3122#endif
3123
3124/* Update i_disksize. Requires i_mutex to avoid races with truncate */
3125static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3126{
3127	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3128		     !inode_is_locked(inode));
 
3129	down_write(&EXT4_I(inode)->i_data_sem);
3130	if (newsize > EXT4_I(inode)->i_disksize)
3131		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3132	up_write(&EXT4_I(inode)->i_data_sem);
 
3133}
3134
3135/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3136static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3137{
3138	int changed = 0;
3139
3140	if (newsize > inode->i_size) {
3141		i_size_write(inode, newsize);
3142		changed = 1;
3143	}
3144	if (newsize > EXT4_I(inode)->i_disksize) {
3145		ext4_update_i_disksize(inode, newsize);
3146		changed |= 2;
3147	}
3148	return changed;
3149}
3150
3151int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3152				      loff_t len);
3153
3154struct ext4_group_info {
3155	unsigned long   bb_state;
3156	struct rb_root  bb_free_root;
3157	ext4_grpblk_t	bb_first_free;	/* first free block */
3158	ext4_grpblk_t	bb_free;	/* total free blocks */
3159	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3160	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3161	struct          list_head bb_prealloc_list;
3162#ifdef DOUBLE_CHECK
3163	void            *bb_bitmap;
3164#endif
3165	struct rw_semaphore alloc_sem;
3166	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3167					 * regions, index is order.
3168					 * bb_counters[3] = 5 means
3169					 * 5 free 8-block regions. */
3170};
3171
3172#define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3173#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3174#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3175#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3176#define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3177	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3178#define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3179	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3180#define EXT4_GROUP_INFO_BBITMAP_READ_BIT	4
3181
3182#define EXT4_MB_GRP_NEED_INIT(grp)	\
3183	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3184#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3185	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3186#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3187	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3188
3189#define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3190	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3191#define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3192	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3193#define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3194	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3195#define EXT4_MB_GRP_TEST_AND_SET_READ(grp)	\
3196	(test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3197
3198#define EXT4_MAX_CONTENTION		8
3199#define EXT4_CONTENTION_THRESHOLD	2
3200
3201static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3202					      ext4_group_t group)
3203{
3204	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3205}
3206
3207/*
3208 * Returns true if the filesystem is busy enough that attempts to
3209 * access the block group locks has run into contention.
3210 */
3211static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3212{
3213	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3214}
3215
3216static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3217{
3218	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3219	if (spin_trylock(lock))
3220		/*
3221		 * We're able to grab the lock right away, so drop the
3222		 * lock contention counter.
3223		 */
3224		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3225	else {
3226		/*
3227		 * The lock is busy, so bump the contention counter,
3228		 * and then wait on the spin lock.
3229		 */
3230		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3231				  EXT4_MAX_CONTENTION);
3232		spin_lock(lock);
3233	}
3234}
3235
3236static inline void ext4_unlock_group(struct super_block *sb,
3237					ext4_group_t group)
3238{
3239	spin_unlock(ext4_group_lock_ptr(sb, group));
3240}
3241
 
 
 
 
 
 
 
 
 
3242/*
3243 * Block validity checking
3244 */
3245#define ext4_check_indirect_blockref(inode, bh)				\
3246	ext4_check_blockref(__func__, __LINE__, inode,			\
3247			    (__le32 *)(bh)->b_data,			\
3248			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3249
3250#define ext4_ind_check_inode(inode)					\
3251	ext4_check_blockref(__func__, __LINE__, inode,			\
3252			    EXT4_I(inode)->i_data,			\
3253			    EXT4_NDIR_BLOCKS)
3254
3255/*
3256 * Inodes and files operations
3257 */
3258
3259/* dir.c */
3260extern const struct file_operations ext4_dir_operations;
3261
3262#ifdef CONFIG_UNICODE
3263extern const struct dentry_operations ext4_dentry_ops;
3264#endif
3265
3266/* file.c */
3267extern const struct inode_operations ext4_file_inode_operations;
3268extern const struct file_operations ext4_file_operations;
3269extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3270
3271/* inline.c */
3272extern int ext4_get_max_inline_size(struct inode *inode);
3273extern int ext4_find_inline_data_nolock(struct inode *inode);
3274extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3275				 unsigned int len);
3276extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3277
3278extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3279extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3280					 struct inode *inode,
3281					 loff_t pos, unsigned len,
3282					 unsigned flags,
3283					 struct page **pagep);
3284extern int ext4_write_inline_data_end(struct inode *inode,
3285				      loff_t pos, unsigned len,
3286				      unsigned copied,
3287				      struct page *page);
3288extern struct buffer_head *
3289ext4_journalled_write_inline_data(struct inode *inode,
3290				  unsigned len,
3291				  struct page *page);
3292extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3293					   struct inode *inode,
3294					   loff_t pos, unsigned len,
3295					   unsigned flags,
3296					   struct page **pagep,
3297					   void **fsdata);
3298extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3299					 unsigned len, unsigned copied,
3300					 struct page *page);
3301extern int ext4_try_add_inline_entry(handle_t *handle,
3302				     struct ext4_filename *fname,
3303				     struct inode *dir, struct inode *inode);
3304extern int ext4_try_create_inline_dir(handle_t *handle,
3305				      struct inode *parent,
3306				      struct inode *inode);
3307extern int ext4_read_inline_dir(struct file *filp,
3308				struct dir_context *ctx,
3309				int *has_inline_data);
3310extern int ext4_inlinedir_to_tree(struct file *dir_file,
3311				  struct inode *dir, ext4_lblk_t block,
3312				  struct dx_hash_info *hinfo,
3313				  __u32 start_hash, __u32 start_minor_hash,
3314				  int *has_inline_data);
3315extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3316					struct ext4_filename *fname,
3317					struct ext4_dir_entry_2 **res_dir,
3318					int *has_inline_data);
3319extern int ext4_delete_inline_entry(handle_t *handle,
3320				    struct inode *dir,
3321				    struct ext4_dir_entry_2 *de_del,
3322				    struct buffer_head *bh,
3323				    int *has_inline_data);
3324extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3325extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3326					struct ext4_dir_entry_2 **parent_de,
3327					int *retval);
3328extern int ext4_inline_data_fiemap(struct inode *inode,
3329				   struct fiemap_extent_info *fieinfo,
3330				   int *has_inline, __u64 start, __u64 len);
3331
3332struct iomap;
3333extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3334
3335extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3336
3337extern int ext4_convert_inline_data(struct inode *inode);
3338
3339static inline int ext4_has_inline_data(struct inode *inode)
3340{
3341	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3342	       EXT4_I(inode)->i_inline_off;
3343}
3344
3345/* namei.c */
3346extern const struct inode_operations ext4_dir_inode_operations;
3347extern const struct inode_operations ext4_special_inode_operations;
3348extern struct dentry *ext4_get_parent(struct dentry *child);
3349extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3350				 struct ext4_dir_entry_2 *de,
3351				 int blocksize, int csum_size,
3352				 unsigned int parent_ino, int dotdot_real_len);
3353extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3354					unsigned int blocksize);
3355extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3356				      struct buffer_head *bh);
3357extern int ext4_ci_compare(const struct inode *parent,
3358			   const struct qstr *fname,
3359			   const struct qstr *entry, bool quick);
3360
3361#define S_SHIFT 12
3362static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3363	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3364	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3365	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3366	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3367	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3368	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3369	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3370};
3371
3372static inline void ext4_set_de_type(struct super_block *sb,
3373				struct ext4_dir_entry_2 *de,
3374				umode_t mode) {
3375	if (ext4_has_feature_filetype(sb))
3376		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3377}
3378
3379/* readpages.c */
3380extern int ext4_mpage_readpages(struct inode *inode,
3381		struct readahead_control *rac, struct page *page);
3382extern int __init ext4_init_post_read_processing(void);
3383extern void ext4_exit_post_read_processing(void);
3384
3385/* symlink.c */
3386extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3387extern const struct inode_operations ext4_symlink_inode_operations;
3388extern const struct inode_operations ext4_fast_symlink_inode_operations;
3389
3390/* sysfs.c */
3391extern int ext4_register_sysfs(struct super_block *sb);
3392extern void ext4_unregister_sysfs(struct super_block *sb);
3393extern int __init ext4_init_sysfs(void);
3394extern void ext4_exit_sysfs(void);
3395
3396/* block_validity */
3397extern void ext4_release_system_zone(struct super_block *sb);
3398extern int ext4_setup_system_zone(struct super_block *sb);
3399extern int __init ext4_init_system_zone(void);
3400extern void ext4_exit_system_zone(void);
3401extern int ext4_inode_block_valid(struct inode *inode,
3402				  ext4_fsblk_t start_blk,
3403				  unsigned int count);
3404extern int ext4_check_blockref(const char *, unsigned int,
3405			       struct inode *, __le32 *, unsigned int);
3406
3407/* extents.c */
3408struct ext4_ext_path;
3409struct ext4_extent;
3410
3411/*
3412 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3413 * __le32.
3414 */
3415#define EXT_MAX_BLOCKS	0xffffffff
3416
3417extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3418extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3419extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3420			       struct ext4_map_blocks *map, int flags);
3421extern int ext4_ext_truncate(handle_t *, struct inode *);
3422extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3423				 ext4_lblk_t end);
3424extern void ext4_ext_init(struct super_block *);
3425extern void ext4_ext_release(struct super_block *);
3426extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3427			  loff_t len);
3428extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3429					  loff_t offset, ssize_t len);
3430extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3431					     ext4_io_end_t *io_end);
3432extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3433			   struct ext4_map_blocks *map, int flags);
3434extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3435						   int num,
3436						   struct ext4_ext_path *path);
3437extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3438				  struct ext4_ext_path **,
3439				  struct ext4_extent *, int);
3440extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3441					      struct ext4_ext_path **,
3442					      int flags);
3443extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3444extern int ext4_ext_check_inode(struct inode *inode);
3445extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3446extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3447			__u64 start, __u64 len);
3448extern int ext4_get_es_cache(struct inode *inode,
3449			     struct fiemap_extent_info *fieinfo,
3450			     __u64 start, __u64 len);
3451extern int ext4_ext_precache(struct inode *inode);
3452extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3453				struct inode *inode2, ext4_lblk_t lblk1,
3454			     ext4_lblk_t lblk2,  ext4_lblk_t count,
3455			     int mark_unwritten,int *err);
3456extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3457extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3458				       int check_cred, int restart_cred,
3459				       int revoke_cred);
3460
3461
3462/* move_extent.c */
3463extern void ext4_double_down_write_data_sem(struct inode *first,
3464					    struct inode *second);
3465extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3466					  struct inode *donor_inode);
3467extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3468			     __u64 start_orig, __u64 start_donor,
3469			     __u64 len, __u64 *moved_len);
3470
3471/* page-io.c */
3472extern int __init ext4_init_pageio(void);
3473extern void ext4_exit_pageio(void);
 
 
3474extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3475extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3476extern int ext4_put_io_end(ext4_io_end_t *io_end);
3477extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3478extern void ext4_io_submit_init(struct ext4_io_submit *io,
3479				struct writeback_control *wbc);
3480extern void ext4_end_io_rsv_work(struct work_struct *work);
3481extern void ext4_io_submit(struct ext4_io_submit *io);
3482extern int ext4_bio_write_page(struct ext4_io_submit *io,
3483			       struct page *page,
3484			       int len,
3485			       struct writeback_control *wbc,
3486			       bool keep_towrite);
3487extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3488extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3489
3490/* mmp.c */
3491extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3492
3493/* verity.c */
3494extern const struct fsverity_operations ext4_verityops;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3495
3496/*
3497 * Add new method to test whether block and inode bitmaps are properly
3498 * initialized. With uninit_bg reading the block from disk is not enough
3499 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3500 */
3501#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3502
3503static inline int bitmap_uptodate(struct buffer_head *bh)
3504{
3505	return (buffer_uptodate(bh) &&
3506			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3507}
3508static inline void set_bitmap_uptodate(struct buffer_head *bh)
3509{
3510	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3511}
3512
3513#define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3514
3515/* For ioend & aio unwritten conversion wait queues */
3516#define EXT4_WQ_HASH_SZ		37
3517#define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3518					    EXT4_WQ_HASH_SZ])
 
 
3519extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
 
3520
 
3521extern int ext4_resize_begin(struct super_block *sb);
3522extern void ext4_resize_end(struct super_block *sb);
3523
3524static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3525					      struct ext4_io_end *io_end)
3526{
3527	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3528		io_end->flag |= EXT4_IO_END_UNWRITTEN;
3529		atomic_inc(&EXT4_I(inode)->i_unwritten);
3530	}
3531}
3532
3533static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3534{
3535	struct inode *inode = io_end->inode;
3536
3537	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3538		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3539		/* Wake up anyone waiting on unwritten extent conversion */
3540		if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3541			wake_up_all(ext4_ioend_wq(inode));
3542	}
3543}
3544
3545extern const struct iomap_ops ext4_iomap_ops;
3546extern const struct iomap_ops ext4_iomap_overwrite_ops;
3547extern const struct iomap_ops ext4_iomap_report_ops;
3548
3549static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3550{
3551	/*
3552	 * If the buffer has the write error flag, we have failed
3553	 * to write out data in the block.  In this  case, we don't
3554	 * have to read the block because we may read the old data
3555	 * successfully.
3556	 */
3557	if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3558		set_buffer_uptodate(bh);
3559	return buffer_uptodate(bh);
3560}
3561
3562#endif	/* __KERNEL__ */
3563
3564#define EFSBADCRC	EBADMSG		/* Bad CRC detected */
3565#define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
3566
3567#endif	/* _EXT4_H */