Version:  2.0.40 2.2.26 2.4.37 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17

Linux/fs/ext4/ext4.h

  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 <linux/ratelimit.h>
 33 #include <crypto/hash.h>
 34 #include <linux/falloc.h>
 35 #ifdef __KERNEL__
 36 #include <linux/compat.h>
 37 #endif
 38 
 39 /*
 40  * The fourth extended filesystem constants/structures
 41  */
 42 
 43 /*
 44  * Define EXT4FS_DEBUG to produce debug messages
 45  */
 46 #undef EXT4FS_DEBUG
 47 
 48 /*
 49  * Debug code
 50  */
 51 #ifdef EXT4FS_DEBUG
 52 #define ext4_debug(f, a...)                                             \
 53         do {                                                            \
 54                 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",        \
 55                         __FILE__, __LINE__, __func__);                  \
 56                 printk(KERN_DEBUG f, ## a);                             \
 57         } while (0)
 58 #else
 59 #define ext4_debug(fmt, ...)    no_printk(fmt, ##__VA_ARGS__)
 60 #endif
 61 
 62 /*
 63  * Turn on EXT_DEBUG to get lots of info about extents operations.
 64  */
 65 #define EXT_DEBUG__
 66 #ifdef EXT_DEBUG
 67 #define ext_debug(fmt, ...)     printk(fmt, ##__VA_ARGS__)
 68 #else
 69 #define ext_debug(fmt, ...)     no_printk(fmt, ##__VA_ARGS__)
 70 #endif
 71 
 72 #define EXT4_ERROR_INODE(inode, fmt, a...) \
 73         ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
 74 
 75 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)                 \
 76         ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
 77 
 78 #define EXT4_ERROR_FILE(file, block, fmt, a...)                         \
 79         ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
 80 
 81 /* data type for block offset of block group */
 82 typedef int ext4_grpblk_t;
 83 
 84 /* data type for filesystem-wide blocks number */
 85 typedef unsigned long long ext4_fsblk_t;
 86 
 87 /* data type for file logical block number */
 88 typedef __u32 ext4_lblk_t;
 89 
 90 /* data type for block group number */
 91 typedef unsigned int ext4_group_t;
 92 
 93 /*
 94  * Flags used in mballoc's allocation_context flags field.
 95  *
 96  * Also used to show what's going on for debugging purposes when the
 97  * flag field is exported via the traceport interface
 98  */
 99 
100 /* prefer goal again. length */
101 #define EXT4_MB_HINT_MERGE              0x0001
102 /* blocks already reserved */
103 #define EXT4_MB_HINT_RESERVED           0x0002
104 /* metadata is being allocated */
105 #define EXT4_MB_HINT_METADATA           0x0004
106 /* first blocks in the file */
107 #define EXT4_MB_HINT_FIRST              0x0008
108 /* search for the best chunk */
109 #define EXT4_MB_HINT_BEST               0x0010
110 /* data is being allocated */
111 #define EXT4_MB_HINT_DATA               0x0020
112 /* don't preallocate (for tails) */
113 #define EXT4_MB_HINT_NOPREALLOC         0x0040
114 /* allocate for locality group */
115 #define EXT4_MB_HINT_GROUP_ALLOC        0x0080
116 /* allocate goal blocks or none */
117 #define EXT4_MB_HINT_GOAL_ONLY          0x0100
118 /* goal is meaningful */
119 #define EXT4_MB_HINT_TRY_GOAL           0x0200
120 /* blocks already pre-reserved by delayed allocation */
121 #define EXT4_MB_DELALLOC_RESERVED       0x0400
122 /* We are doing stream allocation */
123 #define EXT4_MB_STREAM_ALLOC            0x0800
124 /* Use reserved root blocks if needed */
125 #define EXT4_MB_USE_ROOT_BLOCKS         0x1000
126 /* Use blocks from reserved pool */
127 #define EXT4_MB_USE_RESERVED            0x2000
128 
129 struct ext4_allocation_request {
130         /* target inode for block we're allocating */
131         struct inode *inode;
132         /* how many blocks we want to allocate */
133         unsigned int len;
134         /* logical block in target inode */
135         ext4_lblk_t logical;
136         /* the closest logical allocated block to the left */
137         ext4_lblk_t lleft;
138         /* the closest logical allocated block to the right */
139         ext4_lblk_t lright;
140         /* phys. target (a hint) */
141         ext4_fsblk_t goal;
142         /* phys. block for the closest logical allocated block to the left */
143         ext4_fsblk_t pleft;
144         /* phys. block for the closest logical allocated block to the right */
145         ext4_fsblk_t pright;
146         /* flags. see above EXT4_MB_HINT_* */
147         unsigned int flags;
148 };
149 
150 /*
151  * Logical to physical block mapping, used by ext4_map_blocks()
152  *
153  * This structure is used to pass requests into ext4_map_blocks() as
154  * well as to store the information returned by ext4_map_blocks().  It
155  * takes less room on the stack than a struct buffer_head.
156  */
157 #define EXT4_MAP_NEW            (1 << BH_New)
158 #define EXT4_MAP_MAPPED         (1 << BH_Mapped)
159 #define EXT4_MAP_UNWRITTEN      (1 << BH_Unwritten)
160 #define EXT4_MAP_BOUNDARY       (1 << BH_Boundary)
161 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
162  * ext4_map_blocks wants to know whether or not the underlying cluster has
163  * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
164  * the requested mapping was from previously mapped (or delayed allocated)
165  * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
166  * should never appear on buffer_head's state flags.
167  */
168 #define EXT4_MAP_FROM_CLUSTER   (1 << BH_AllocFromCluster)
169 #define EXT4_MAP_FLAGS          (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
170                                  EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
171                                  EXT4_MAP_FROM_CLUSTER)
172 
173 struct ext4_map_blocks {
174         ext4_fsblk_t m_pblk;
175         ext4_lblk_t m_lblk;
176         unsigned int m_len;
177         unsigned int m_flags;
178 };
179 
180 /*
181  * Flags for ext4_io_end->flags
182  */
183 #define EXT4_IO_END_UNWRITTEN   0x0001
184 
185 /*
186  * For converting unwritten extents on a work queue. 'handle' is used for
187  * buffered writeback.
188  */
189 typedef struct ext4_io_end {
190         struct list_head        list;           /* per-file finished IO list */
191         handle_t                *handle;        /* handle reserved for extent
192                                                  * conversion */
193         struct inode            *inode;         /* file being written to */
194         struct bio              *bio;           /* Linked list of completed
195                                                  * bios covering the extent */
196         unsigned int            flag;           /* unwritten or not */
197         loff_t                  offset;         /* offset in the file */
198         ssize_t                 size;           /* size of the extent */
199         atomic_t                count;          /* reference counter */
200 } ext4_io_end_t;
201 
202 struct ext4_io_submit {
203         int                     io_op;
204         struct bio              *io_bio;
205         ext4_io_end_t           *io_end;
206         sector_t                io_next_block;
207 };
208 
209 /*
210  * Special inodes numbers
211  */
212 #define EXT4_BAD_INO             1      /* Bad blocks inode */
213 #define EXT4_ROOT_INO            2      /* Root inode */
214 #define EXT4_USR_QUOTA_INO       3      /* User quota inode */
215 #define EXT4_GRP_QUOTA_INO       4      /* Group quota inode */
216 #define EXT4_BOOT_LOADER_INO     5      /* Boot loader inode */
217 #define EXT4_UNDEL_DIR_INO       6      /* Undelete directory inode */
218 #define EXT4_RESIZE_INO          7      /* Reserved group descriptors inode */
219 #define EXT4_JOURNAL_INO         8      /* Journal inode */
220 
221 /* First non-reserved inode for old ext4 filesystems */
222 #define EXT4_GOOD_OLD_FIRST_INO 11
223 
224 /*
225  * Maximal count of links to a file
226  */
227 #define EXT4_LINK_MAX           65000
228 
229 /*
230  * Macro-instructions used to manage several block sizes
231  */
232 #define EXT4_MIN_BLOCK_SIZE             1024
233 #define EXT4_MAX_BLOCK_SIZE             65536
234 #define EXT4_MIN_BLOCK_LOG_SIZE         10
235 #define EXT4_MAX_BLOCK_LOG_SIZE         16
236 #ifdef __KERNEL__
237 # define EXT4_BLOCK_SIZE(s)             ((s)->s_blocksize)
238 #else
239 # define EXT4_BLOCK_SIZE(s)             (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
240 #endif
241 #define EXT4_ADDR_PER_BLOCK(s)          (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
242 #define EXT4_CLUSTER_SIZE(s)            (EXT4_BLOCK_SIZE(s) << \
243                                          EXT4_SB(s)->s_cluster_bits)
244 #ifdef __KERNEL__
245 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_blocksize_bits)
246 # define EXT4_CLUSTER_BITS(s)           (EXT4_SB(s)->s_cluster_bits)
247 #else
248 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_log_block_size + 10)
249 #endif
250 #ifdef __KERNEL__
251 #define EXT4_ADDR_PER_BLOCK_BITS(s)     (EXT4_SB(s)->s_addr_per_block_bits)
252 #define EXT4_INODE_SIZE(s)              (EXT4_SB(s)->s_inode_size)
253 #define EXT4_FIRST_INO(s)               (EXT4_SB(s)->s_first_ino)
254 #else
255 #define EXT4_INODE_SIZE(s)      (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
256                                  EXT4_GOOD_OLD_INODE_SIZE : \
257                                  (s)->s_inode_size)
258 #define EXT4_FIRST_INO(s)       (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
259                                  EXT4_GOOD_OLD_FIRST_INO : \
260                                  (s)->s_first_ino)
261 #endif
262 #define EXT4_BLOCK_ALIGN(size, blkbits)         ALIGN((size), (1 << (blkbits)))
263 
264 /* Translate a block number to a cluster number */
265 #define EXT4_B2C(sbi, blk)      ((blk) >> (sbi)->s_cluster_bits)
266 /* Translate a cluster number to a block number */
267 #define EXT4_C2B(sbi, cluster)  ((cluster) << (sbi)->s_cluster_bits)
268 /* Translate # of blks to # of clusters */
269 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
270                                  (sbi)->s_cluster_bits)
271 /* Mask out the low bits to get the starting block of the cluster */
272 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &                              \
273                                   ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
274 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &                              \
275                                   ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
276 /* Get the cluster offset */
277 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &                               \
278                                  ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
279 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &                               \
280                                  ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
281 
282 /*
283  * Structure of a blocks group descriptor
284  */
285 struct ext4_group_desc
286 {
287         __le32  bg_block_bitmap_lo;     /* Blocks bitmap block */
288         __le32  bg_inode_bitmap_lo;     /* Inodes bitmap block */
289         __le32  bg_inode_table_lo;      /* Inodes table block */
290         __le16  bg_free_blocks_count_lo;/* Free blocks count */
291         __le16  bg_free_inodes_count_lo;/* Free inodes count */
292         __le16  bg_used_dirs_count_lo;  /* Directories count */
293         __le16  bg_flags;               /* EXT4_BG_flags (INODE_UNINIT, etc) */
294         __le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
295         __le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
296         __le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
297         __le16  bg_itable_unused_lo;    /* Unused inodes count */
298         __le16  bg_checksum;            /* crc16(sb_uuid+group+desc) */
299         __le32  bg_block_bitmap_hi;     /* Blocks bitmap block MSB */
300         __le32  bg_inode_bitmap_hi;     /* Inodes bitmap block MSB */
301         __le32  bg_inode_table_hi;      /* Inodes table block MSB */
302         __le16  bg_free_blocks_count_hi;/* Free blocks count MSB */
303         __le16  bg_free_inodes_count_hi;/* Free inodes count MSB */
304         __le16  bg_used_dirs_count_hi;  /* Directories count MSB */
305         __le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
306         __le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
307         __le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
308         __le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
309         __u32   bg_reserved;
310 };
311 
312 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END        \
313         (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
314          sizeof(__le16))
315 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END        \
316         (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
317          sizeof(__le16))
318 
319 /*
320  * Structure of a flex block group info
321  */
322 
323 struct flex_groups {
324         atomic64_t      free_clusters;
325         atomic_t        free_inodes;
326         atomic_t        used_dirs;
327 };
328 
329 #define EXT4_BG_INODE_UNINIT    0x0001 /* Inode table/bitmap not in use */
330 #define EXT4_BG_BLOCK_UNINIT    0x0002 /* Block bitmap not in use */
331 #define EXT4_BG_INODE_ZEROED    0x0004 /* On-disk itable initialized to zero */
332 
333 /*
334  * Macro-instructions used to manage group descriptors
335  */
336 #define EXT4_MIN_DESC_SIZE              32
337 #define EXT4_MIN_DESC_SIZE_64BIT        64
338 #define EXT4_MAX_DESC_SIZE              EXT4_MIN_BLOCK_SIZE
339 #define EXT4_DESC_SIZE(s)               (EXT4_SB(s)->s_desc_size)
340 #ifdef __KERNEL__
341 # define EXT4_BLOCKS_PER_GROUP(s)       (EXT4_SB(s)->s_blocks_per_group)
342 # define EXT4_CLUSTERS_PER_GROUP(s)     (EXT4_SB(s)->s_clusters_per_group)
343 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_SB(s)->s_desc_per_block)
344 # define EXT4_INODES_PER_GROUP(s)       (EXT4_SB(s)->s_inodes_per_group)
345 # define EXT4_DESC_PER_BLOCK_BITS(s)    (EXT4_SB(s)->s_desc_per_block_bits)
346 #else
347 # define EXT4_BLOCKS_PER_GROUP(s)       ((s)->s_blocks_per_group)
348 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
349 # define EXT4_INODES_PER_GROUP(s)       ((s)->s_inodes_per_group)
350 #endif
351 
352 /*
353  * Constants relative to the data blocks
354  */
355 #define EXT4_NDIR_BLOCKS                12
356 #define EXT4_IND_BLOCK                  EXT4_NDIR_BLOCKS
357 #define EXT4_DIND_BLOCK                 (EXT4_IND_BLOCK + 1)
358 #define EXT4_TIND_BLOCK                 (EXT4_DIND_BLOCK + 1)
359 #define EXT4_N_BLOCKS                   (EXT4_TIND_BLOCK + 1)
360 
361 /*
362  * Inode flags
363  */
364 #define EXT4_SECRM_FL                   0x00000001 /* Secure deletion */
365 #define EXT4_UNRM_FL                    0x00000002 /* Undelete */
366 #define EXT4_COMPR_FL                   0x00000004 /* Compress file */
367 #define EXT4_SYNC_FL                    0x00000008 /* Synchronous updates */
368 #define EXT4_IMMUTABLE_FL               0x00000010 /* Immutable file */
369 #define EXT4_APPEND_FL                  0x00000020 /* writes to file may only append */
370 #define EXT4_NODUMP_FL                  0x00000040 /* do not dump file */
371 #define EXT4_NOATIME_FL                 0x00000080 /* do not update atime */
372 /* Reserved for compression usage... */
373 #define EXT4_DIRTY_FL                   0x00000100
374 #define EXT4_COMPRBLK_FL                0x00000200 /* One or more compressed clusters */
375 #define EXT4_NOCOMPR_FL                 0x00000400 /* Don't compress */
376 #define EXT4_ECOMPR_FL                  0x00000800 /* Compression error */
377 /* End compression flags --- maybe not all used */
378 #define EXT4_INDEX_FL                   0x00001000 /* hash-indexed directory */
379 #define EXT4_IMAGIC_FL                  0x00002000 /* AFS directory */
380 #define EXT4_JOURNAL_DATA_FL            0x00004000 /* file data should be journaled */
381 #define EXT4_NOTAIL_FL                  0x00008000 /* file tail should not be merged */
382 #define EXT4_DIRSYNC_FL                 0x00010000 /* dirsync behaviour (directories only) */
383 #define EXT4_TOPDIR_FL                  0x00020000 /* Top of directory hierarchies*/
384 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
385 #define EXT4_EXTENTS_FL                 0x00080000 /* Inode uses extents */
386 #define EXT4_EA_INODE_FL                0x00200000 /* Inode used for large EA */
387 #define EXT4_EOFBLOCKS_FL               0x00400000 /* Blocks allocated beyond EOF */
388 #define EXT4_INLINE_DATA_FL             0x10000000 /* Inode has inline data. */
389 #define EXT4_RESERVED_FL                0x80000000 /* reserved for ext4 lib */
390 
391 #define EXT4_FL_USER_VISIBLE            0x004BDFFF /* User visible flags */
392 #define EXT4_FL_USER_MODIFIABLE         0x004380FF /* User modifiable flags */
393 
394 /* Flags that should be inherited by new inodes from their parent. */
395 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
396                            EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
397                            EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
398                            EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
399 
400 /* Flags that are appropriate for regular files (all but dir-specific ones). */
401 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
402 
403 /* Flags that are appropriate for non-directories/regular files. */
404 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
405 
406 /* Mask out flags that are inappropriate for the given type of inode. */
407 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
408 {
409         if (S_ISDIR(mode))
410                 return flags;
411         else if (S_ISREG(mode))
412                 return flags & EXT4_REG_FLMASK;
413         else
414                 return flags & EXT4_OTHER_FLMASK;
415 }
416 
417 /*
418  * Inode flags used for atomic set/get
419  */
420 enum {
421         EXT4_INODE_SECRM        = 0,    /* Secure deletion */
422         EXT4_INODE_UNRM         = 1,    /* Undelete */
423         EXT4_INODE_COMPR        = 2,    /* Compress file */
424         EXT4_INODE_SYNC         = 3,    /* Synchronous updates */
425         EXT4_INODE_IMMUTABLE    = 4,    /* Immutable file */
426         EXT4_INODE_APPEND       = 5,    /* writes to file may only append */
427         EXT4_INODE_NODUMP       = 6,    /* do not dump file */
428         EXT4_INODE_NOATIME      = 7,    /* do not update atime */
429 /* Reserved for compression usage... */
430         EXT4_INODE_DIRTY        = 8,
431         EXT4_INODE_COMPRBLK     = 9,    /* One or more compressed clusters */
432         EXT4_INODE_NOCOMPR      = 10,   /* Don't compress */
433         EXT4_INODE_ECOMPR       = 11,   /* Compression error */
434 /* End compression flags --- maybe not all used */
435         EXT4_INODE_INDEX        = 12,   /* hash-indexed directory */
436         EXT4_INODE_IMAGIC       = 13,   /* AFS directory */
437         EXT4_INODE_JOURNAL_DATA = 14,   /* file data should be journaled */
438         EXT4_INODE_NOTAIL       = 15,   /* file tail should not be merged */
439         EXT4_INODE_DIRSYNC      = 16,   /* dirsync behaviour (directories only) */
440         EXT4_INODE_TOPDIR       = 17,   /* Top of directory hierarchies*/
441         EXT4_INODE_HUGE_FILE    = 18,   /* Set to each huge file */
442         EXT4_INODE_EXTENTS      = 19,   /* Inode uses extents */
443         EXT4_INODE_EA_INODE     = 21,   /* Inode used for large EA */
444         EXT4_INODE_EOFBLOCKS    = 22,   /* Blocks allocated beyond EOF */
445         EXT4_INODE_INLINE_DATA  = 28,   /* Data in inode. */
446         EXT4_INODE_RESERVED     = 31,   /* reserved for ext4 lib */
447 };
448 
449 /*
450  * Since it's pretty easy to mix up bit numbers and hex values, we use a
451  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
452  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
453  * any extra space in the compiled kernel image, otherwise, the build will fail.
454  * It's important that these values are the same, since we are using
455  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
456  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
457  * values found in ext2, ext3 and ext4 filesystems, and of course the values
458  * defined in e2fsprogs.
459  *
460  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
461  */
462 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
463 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
464 
465 static inline void ext4_check_flag_values(void)
466 {
467         CHECK_FLAG_VALUE(SECRM);
468         CHECK_FLAG_VALUE(UNRM);
469         CHECK_FLAG_VALUE(COMPR);
470         CHECK_FLAG_VALUE(SYNC);
471         CHECK_FLAG_VALUE(IMMUTABLE);
472         CHECK_FLAG_VALUE(APPEND);
473         CHECK_FLAG_VALUE(NODUMP);
474         CHECK_FLAG_VALUE(NOATIME);
475         CHECK_FLAG_VALUE(DIRTY);
476         CHECK_FLAG_VALUE(COMPRBLK);
477         CHECK_FLAG_VALUE(NOCOMPR);
478         CHECK_FLAG_VALUE(ECOMPR);
479         CHECK_FLAG_VALUE(INDEX);
480         CHECK_FLAG_VALUE(IMAGIC);
481         CHECK_FLAG_VALUE(JOURNAL_DATA);
482         CHECK_FLAG_VALUE(NOTAIL);
483         CHECK_FLAG_VALUE(DIRSYNC);
484         CHECK_FLAG_VALUE(TOPDIR);
485         CHECK_FLAG_VALUE(HUGE_FILE);
486         CHECK_FLAG_VALUE(EXTENTS);
487         CHECK_FLAG_VALUE(EA_INODE);
488         CHECK_FLAG_VALUE(EOFBLOCKS);
489         CHECK_FLAG_VALUE(INLINE_DATA);
490         CHECK_FLAG_VALUE(RESERVED);
491 }
492 
493 /* Used to pass group descriptor data when online resize is done */
494 struct ext4_new_group_input {
495         __u32 group;            /* Group number for this data */
496         __u64 block_bitmap;     /* Absolute block number of block bitmap */
497         __u64 inode_bitmap;     /* Absolute block number of inode bitmap */
498         __u64 inode_table;      /* Absolute block number of inode table start */
499         __u32 blocks_count;     /* Total number of blocks in this group */
500         __u16 reserved_blocks;  /* Number of reserved blocks in this group */
501         __u16 unused;
502 };
503 
504 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
505 struct compat_ext4_new_group_input {
506         u32 group;
507         compat_u64 block_bitmap;
508         compat_u64 inode_bitmap;
509         compat_u64 inode_table;
510         u32 blocks_count;
511         u16 reserved_blocks;
512         u16 unused;
513 };
514 #endif
515 
516 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
517 struct ext4_new_group_data {
518         __u32 group;
519         __u64 block_bitmap;
520         __u64 inode_bitmap;
521         __u64 inode_table;
522         __u32 blocks_count;
523         __u16 reserved_blocks;
524         __u16 unused;
525         __u32 free_blocks_count;
526 };
527 
528 /* Indexes used to index group tables in ext4_new_group_data */
529 enum {
530         BLOCK_BITMAP = 0,       /* block bitmap */
531         INODE_BITMAP,           /* inode bitmap */
532         INODE_TABLE,            /* inode tables */
533         GROUP_TABLE_COUNT,
534 };
535 
536 /*
537  * Flags used by ext4_map_blocks()
538  */
539         /* Allocate any needed blocks and/or convert an unwritten
540            extent to be an initialized ext4 */
541 #define EXT4_GET_BLOCKS_CREATE                  0x0001
542         /* Request the creation of an unwritten extent */
543 #define EXT4_GET_BLOCKS_UNWRIT_EXT              0x0002
544 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT       (EXT4_GET_BLOCKS_UNWRIT_EXT|\
545                                                  EXT4_GET_BLOCKS_CREATE)
546         /* Caller is from the delayed allocation writeout path
547          * finally doing the actual allocation of delayed blocks */
548 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE        0x0004
549         /* caller is from the direct IO path, request to creation of an
550         unwritten extents if not allocated, split the unwritten
551         extent if blocks has been preallocated already*/
552 #define EXT4_GET_BLOCKS_PRE_IO                  0x0008
553 #define EXT4_GET_BLOCKS_CONVERT                 0x0010
554 #define EXT4_GET_BLOCKS_IO_CREATE_EXT           (EXT4_GET_BLOCKS_PRE_IO|\
555                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
556         /* Convert extent to initialized after IO complete */
557 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT          (EXT4_GET_BLOCKS_CONVERT|\
558                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
559         /* Eventual metadata allocation (due to growing extent tree)
560          * should not fail, so try to use reserved blocks for that.*/
561 #define EXT4_GET_BLOCKS_METADATA_NOFAIL         0x0020
562         /* Don't normalize allocation size (used for fallocate) */
563 #define EXT4_GET_BLOCKS_NO_NORMALIZE            0x0040
564         /* Request will not result in inode size update (user for fallocate) */
565 #define EXT4_GET_BLOCKS_KEEP_SIZE               0x0080
566         /* Do not take i_data_sem locking in ext4_map_blocks */
567 #define EXT4_GET_BLOCKS_NO_LOCK                 0x0100
568         /* Do not put hole in extent cache */
569 #define EXT4_GET_BLOCKS_NO_PUT_HOLE             0x0200
570         /* Convert written extents to unwritten */
571 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN       0x0400
572 
573 /*
574  * The bit position of these flags must not overlap with any of the
575  * EXT4_GET_BLOCKS_*.  They are used by ext4_ext_find_extent(),
576  * read_extent_tree_block(), ext4_split_extent_at(),
577  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
578  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
579  * caching the extents when reading from the extent tree while a
580  * truncate or punch hole operation is in progress.
581  */
582 #define EXT4_EX_NOCACHE                         0x0400
583 #define EXT4_EX_FORCE_CACHE                     0x0800
584 
585 /*
586  * Flags used by ext4_free_blocks
587  */
588 #define EXT4_FREE_BLOCKS_METADATA       0x0001
589 #define EXT4_FREE_BLOCKS_FORGET         0x0002
590 #define EXT4_FREE_BLOCKS_VALIDATED      0x0004
591 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
592 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER   0x0010
593 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER    0x0020
594 
595 /*
596  * ioctl commands
597  */
598 #define EXT4_IOC_GETFLAGS               FS_IOC_GETFLAGS
599 #define EXT4_IOC_SETFLAGS               FS_IOC_SETFLAGS
600 #define EXT4_IOC_GETVERSION             _IOR('f', 3, long)
601 #define EXT4_IOC_SETVERSION             _IOW('f', 4, long)
602 #define EXT4_IOC_GETVERSION_OLD         FS_IOC_GETVERSION
603 #define EXT4_IOC_SETVERSION_OLD         FS_IOC_SETVERSION
604 #define EXT4_IOC_GETRSVSZ               _IOR('f', 5, long)
605 #define EXT4_IOC_SETRSVSZ               _IOW('f', 6, long)
606 #define EXT4_IOC_GROUP_EXTEND           _IOW('f', 7, unsigned long)
607 #define EXT4_IOC_GROUP_ADD              _IOW('f', 8, struct ext4_new_group_input)
608 #define EXT4_IOC_MIGRATE                _IO('f', 9)
609  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
610  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
611 #define EXT4_IOC_ALLOC_DA_BLKS          _IO('f', 12)
612 #define EXT4_IOC_MOVE_EXT               _IOWR('f', 15, struct move_extent)
613 #define EXT4_IOC_RESIZE_FS              _IOW('f', 16, __u64)
614 #define EXT4_IOC_SWAP_BOOT              _IO('f', 17)
615 #define EXT4_IOC_PRECACHE_EXTENTS       _IO('f', 18)
616 
617 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
618 /*
619  * ioctl commands in 32 bit emulation
620  */
621 #define EXT4_IOC32_GETFLAGS             FS_IOC32_GETFLAGS
622 #define EXT4_IOC32_SETFLAGS             FS_IOC32_SETFLAGS
623 #define EXT4_IOC32_GETVERSION           _IOR('f', 3, int)
624 #define EXT4_IOC32_SETVERSION           _IOW('f', 4, int)
625 #define EXT4_IOC32_GETRSVSZ             _IOR('f', 5, int)
626 #define EXT4_IOC32_SETRSVSZ             _IOW('f', 6, int)
627 #define EXT4_IOC32_GROUP_EXTEND         _IOW('f', 7, unsigned int)
628 #define EXT4_IOC32_GROUP_ADD            _IOW('f', 8, struct compat_ext4_new_group_input)
629 #define EXT4_IOC32_GETVERSION_OLD       FS_IOC32_GETVERSION
630 #define EXT4_IOC32_SETVERSION_OLD       FS_IOC32_SETVERSION
631 #endif
632 
633 /* Max physical block we can address w/o extents */
634 #define EXT4_MAX_BLOCK_FILE_PHYS        0xFFFFFFFF
635 
636 /*
637  * Structure of an inode on the disk
638  */
639 struct ext4_inode {
640         __le16  i_mode;         /* File mode */
641         __le16  i_uid;          /* Low 16 bits of Owner Uid */
642         __le32  i_size_lo;      /* Size in bytes */
643         __le32  i_atime;        /* Access time */
644         __le32  i_ctime;        /* Inode Change time */
645         __le32  i_mtime;        /* Modification time */
646         __le32  i_dtime;        /* Deletion Time */
647         __le16  i_gid;          /* Low 16 bits of Group Id */
648         __le16  i_links_count;  /* Links count */
649         __le32  i_blocks_lo;    /* Blocks count */
650         __le32  i_flags;        /* File flags */
651         union {
652                 struct {
653                         __le32  l_i_version;
654                 } linux1;
655                 struct {
656                         __u32  h_i_translator;
657                 } hurd1;
658                 struct {
659                         __u32  m_i_reserved1;
660                 } masix1;
661         } osd1;                         /* OS dependent 1 */
662         __le32  i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
663         __le32  i_generation;   /* File version (for NFS) */
664         __le32  i_file_acl_lo;  /* File ACL */
665         __le32  i_size_high;
666         __le32  i_obso_faddr;   /* Obsoleted fragment address */
667         union {
668                 struct {
669                         __le16  l_i_blocks_high; /* were l_i_reserved1 */
670                         __le16  l_i_file_acl_high;
671                         __le16  l_i_uid_high;   /* these 2 fields */
672                         __le16  l_i_gid_high;   /* were reserved2[0] */
673                         __le16  l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
674                         __le16  l_i_reserved;
675                 } linux2;
676                 struct {
677                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
678                         __u16   h_i_mode_high;
679                         __u16   h_i_uid_high;
680                         __u16   h_i_gid_high;
681                         __u32   h_i_author;
682                 } hurd2;
683                 struct {
684                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
685                         __le16  m_i_file_acl_high;
686                         __u32   m_i_reserved2[2];
687                 } masix2;
688         } osd2;                         /* OS dependent 2 */
689         __le16  i_extra_isize;
690         __le16  i_checksum_hi;  /* crc32c(uuid+inum+inode) BE */
691         __le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
692         __le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
693         __le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
694         __le32  i_crtime;       /* File Creation time */
695         __le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
696         __le32  i_version_hi;   /* high 32 bits for 64-bit version */
697 };
698 
699 struct move_extent {
700         __u32 reserved;         /* should be zero */
701         __u32 donor_fd;         /* donor file descriptor */
702         __u64 orig_start;       /* logical start offset in block for orig */
703         __u64 donor_start;      /* logical start offset in block for donor */
704         __u64 len;              /* block length to be moved */
705         __u64 moved_len;        /* moved block length */
706 };
707 
708 #define EXT4_EPOCH_BITS 2
709 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
710 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
711 
712 /*
713  * Extended fields will fit into an inode if the filesystem was formatted
714  * with large inodes (-I 256 or larger) and there are not currently any EAs
715  * consuming all of the available space. For new inodes we always reserve
716  * enough space for the kernel's known extended fields, but for inodes
717  * created with an old kernel this might not have been the case. None of
718  * the extended inode fields is critical for correct filesystem operation.
719  * This macro checks if a certain field fits in the inode. Note that
720  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
721  */
722 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)   \
723         ((offsetof(typeof(*ext4_inode), field) +        \
724           sizeof((ext4_inode)->field))                  \
725         <= (EXT4_GOOD_OLD_INODE_SIZE +                  \
726             (einode)->i_extra_isize))                   \
727 
728 static inline __le32 ext4_encode_extra_time(struct timespec *time)
729 {
730        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
731                            (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
732                           ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
733 }
734 
735 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
736 {
737        if (sizeof(time->tv_sec) > 4)
738                time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
739                                << 32;
740        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
741 }
742 
743 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)                          \
744 do {                                                                           \
745         (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);               \
746         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
747                 (raw_inode)->xtime ## _extra =                                 \
748                                 ext4_encode_extra_time(&(inode)->xtime);       \
749 } while (0)
750 
751 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)                        \
752 do {                                                                           \
753         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
754                 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
755         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
756                 (raw_inode)->xtime ## _extra =                                 \
757                                 ext4_encode_extra_time(&(einode)->xtime);      \
758 } while (0)
759 
760 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)                          \
761 do {                                                                           \
762         (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
763         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
764                 ext4_decode_extra_time(&(inode)->xtime,                        \
765                                        raw_inode->xtime ## _extra);            \
766         else                                                                   \
767                 (inode)->xtime.tv_nsec = 0;                                    \
768 } while (0)
769 
770 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)                        \
771 do {                                                                           \
772         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
773                 (einode)->xtime.tv_sec =                                       \
774                         (signed)le32_to_cpu((raw_inode)->xtime);               \
775         else                                                                   \
776                 (einode)->xtime.tv_sec = 0;                                    \
777         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
778                 ext4_decode_extra_time(&(einode)->xtime,                       \
779                                        raw_inode->xtime ## _extra);            \
780         else                                                                   \
781                 (einode)->xtime.tv_nsec = 0;                                   \
782 } while (0)
783 
784 #define i_disk_version osd1.linux1.l_i_version
785 
786 #if defined(__KERNEL__) || defined(__linux__)
787 #define i_reserved1     osd1.linux1.l_i_reserved1
788 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
789 #define i_blocks_high   osd2.linux2.l_i_blocks_high
790 #define i_uid_low       i_uid
791 #define i_gid_low       i_gid
792 #define i_uid_high      osd2.linux2.l_i_uid_high
793 #define i_gid_high      osd2.linux2.l_i_gid_high
794 #define i_checksum_lo   osd2.linux2.l_i_checksum_lo
795 
796 #elif defined(__GNU__)
797 
798 #define i_translator    osd1.hurd1.h_i_translator
799 #define i_uid_high      osd2.hurd2.h_i_uid_high
800 #define i_gid_high      osd2.hurd2.h_i_gid_high
801 #define i_author        osd2.hurd2.h_i_author
802 
803 #elif defined(__masix__)
804 
805 #define i_reserved1     osd1.masix1.m_i_reserved1
806 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
807 #define i_reserved2     osd2.masix2.m_i_reserved2
808 
809 #endif /* defined(__KERNEL__) || defined(__linux__) */
810 
811 #include "extents_status.h"
812 
813 /*
814  * fourth extended file system inode data in memory
815  */
816 struct 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         /*
836          * Extended attributes can be read independently of the main file
837          * data. Taking i_mutex even when reading would cause contention
838          * between readers of EAs and writers of regular file data, so
839          * instead we synchronize on xattr_sem when reading or changing
840          * EAs.
841          */
842         struct rw_semaphore xattr_sem;
843 
844         struct list_head i_orphan;      /* unlinked but open inodes */
845 
846         /*
847          * i_disksize keeps track of what the inode size is ON DISK, not
848          * in memory.  During truncate, i_size is set to the new size by
849          * the VFS prior to calling ext4_truncate(), but the filesystem won't
850          * set i_disksize to 0 until the truncate is actually under way.
851          *
852          * The intent is that i_disksize always represents the blocks which
853          * are used by this file.  This allows recovery to restart truncate
854          * on orphans if we crash during truncate.  We actually write i_disksize
855          * into the on-disk inode when writing inodes out, instead of i_size.
856          *
857          * The only time when i_disksize and i_size may be different is when
858          * a truncate is in progress.  The only things which change i_disksize
859          * are ext4_get_block (growth) and ext4_truncate (shrinkth).
860          */
861         loff_t  i_disksize;
862 
863         /*
864          * i_data_sem is for serialising ext4_truncate() against
865          * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
866          * data tree are chopped off during truncate. We can't do that in
867          * ext4 because whenever we perform intermediate commits during
868          * truncate, the inode and all the metadata blocks *must* be in a
869          * consistent state which allows truncation of the orphans to restart
870          * during recovery.  Hence we must fix the get_block-vs-truncate race
871          * by other means, so we have i_data_sem.
872          */
873         struct rw_semaphore i_data_sem;
874         struct inode vfs_inode;
875         struct jbd2_inode *jinode;
876 
877         spinlock_t i_raw_lock;  /* protects updates to the raw inode */
878 
879         /*
880          * File creation time. Its function is same as that of
881          * struct timespec i_{a,c,m}time in the generic inode.
882          */
883         struct timespec i_crtime;
884 
885         /* mballoc */
886         struct list_head i_prealloc_list;
887         spinlock_t i_prealloc_lock;
888 
889         /* extents status tree */
890         struct ext4_es_tree i_es_tree;
891         rwlock_t i_es_lock;
892         struct list_head i_es_lru;
893         unsigned int i_es_lru_nr;       /* protected by i_es_lock */
894         unsigned long i_touch_when;     /* jiffies of last accessing */
895 
896         /* ialloc */
897         ext4_group_t    i_last_alloc_group;
898 
899         /* allocation reservation info for delalloc */
900         /* In case of bigalloc, these refer to clusters rather than blocks */
901         unsigned int i_reserved_data_blocks;
902         unsigned int i_reserved_meta_blocks;
903         unsigned int i_allocated_meta_blocks;
904         ext4_lblk_t i_da_metadata_calc_last_lblock;
905         int i_da_metadata_calc_len;
906 
907         /* on-disk additional length */
908         __u16 i_extra_isize;
909 
910         /* Indicate the inline data space. */
911         u16 i_inline_off;
912         u16 i_inline_size;
913 
914 #ifdef CONFIG_QUOTA
915         /* quota space reservation, managed internally by quota code */
916         qsize_t i_reserved_quota;
917 #endif
918 
919         /* Lock protecting lists below */
920         spinlock_t i_completed_io_lock;
921         /*
922          * Completed IOs that need unwritten extents handling and have
923          * transaction reserved
924          */
925         struct list_head i_rsv_conversion_list;
926         /*
927          * Completed IOs that need unwritten extents handling and don't have
928          * transaction reserved
929          */
930         atomic_t i_ioend_count; /* Number of outstanding io_end structs */
931         atomic_t i_unwritten; /* Nr. of inflight conversions pending */
932         struct work_struct i_rsv_conversion_work;
933 
934         spinlock_t i_block_reservation_lock;
935 
936         /*
937          * Transactions that contain inode's metadata needed to complete
938          * fsync and fdatasync, respectively.
939          */
940         tid_t i_sync_tid;
941         tid_t i_datasync_tid;
942 
943         /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
944         __u32 i_csum_seed;
945 };
946 
947 /*
948  * File system states
949  */
950 #define EXT4_VALID_FS                   0x0001  /* Unmounted cleanly */
951 #define EXT4_ERROR_FS                   0x0002  /* Errors detected */
952 #define EXT4_ORPHAN_FS                  0x0004  /* Orphans being recovered */
953 
954 /*
955  * Misc. filesystem flags
956  */
957 #define EXT2_FLAGS_SIGNED_HASH          0x0001  /* Signed dirhash in use */
958 #define EXT2_FLAGS_UNSIGNED_HASH        0x0002  /* Unsigned dirhash in use */
959 #define EXT2_FLAGS_TEST_FILESYS         0x0004  /* to test development code */
960 
961 /*
962  * Mount flags set via mount options or defaults
963  */
964 #define EXT4_MOUNT_GRPID                0x00004 /* Create files with directory's group */
965 #define EXT4_MOUNT_DEBUG                0x00008 /* Some debugging messages */
966 #define EXT4_MOUNT_ERRORS_CONT          0x00010 /* Continue on errors */
967 #define EXT4_MOUNT_ERRORS_RO            0x00020 /* Remount fs ro on errors */
968 #define EXT4_MOUNT_ERRORS_PANIC         0x00040 /* Panic on errors */
969 #define EXT4_MOUNT_ERRORS_MASK          0x00070
970 #define EXT4_MOUNT_MINIX_DF             0x00080 /* Mimics the Minix statfs */
971 #define EXT4_MOUNT_NOLOAD               0x00100 /* Don't use existing journal*/
972 #define EXT4_MOUNT_DATA_FLAGS           0x00C00 /* Mode for data writes: */
973 #define EXT4_MOUNT_JOURNAL_DATA         0x00400 /* Write data to journal */
974 #define EXT4_MOUNT_ORDERED_DATA         0x00800 /* Flush data before commit */
975 #define EXT4_MOUNT_WRITEBACK_DATA       0x00C00 /* No data ordering */
976 #define EXT4_MOUNT_UPDATE_JOURNAL       0x01000 /* Update the journal format */
977 #define EXT4_MOUNT_NO_UID32             0x02000  /* Disable 32-bit UIDs */
978 #define EXT4_MOUNT_XATTR_USER           0x04000 /* Extended user attributes */
979 #define EXT4_MOUNT_POSIX_ACL            0x08000 /* POSIX Access Control Lists */
980 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC     0x10000 /* No auto delalloc mapping */
981 #define EXT4_MOUNT_BARRIER              0x20000 /* Use block barriers */
982 #define EXT4_MOUNT_QUOTA                0x80000 /* Some quota option set */
983 #define EXT4_MOUNT_USRQUOTA             0x100000 /* "old" user quota */
984 #define EXT4_MOUNT_GRPQUOTA             0x200000 /* "old" group quota */
985 #define EXT4_MOUNT_DIOREAD_NOLOCK       0x400000 /* Enable support for dio read nolocking */
986 #define EXT4_MOUNT_JOURNAL_CHECKSUM     0x800000 /* Journal checksums */
987 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
988 #define EXT4_MOUNT_DELALLOC             0x8000000 /* Delalloc support */
989 #define EXT4_MOUNT_DATA_ERR_ABORT       0x10000000 /* Abort on file data write */
990 #define EXT4_MOUNT_BLOCK_VALIDITY       0x20000000 /* Block validity checking */
991 #define EXT4_MOUNT_DISCARD              0x40000000 /* Issue DISCARD requests */
992 #define EXT4_MOUNT_INIT_INODE_TABLE     0x80000000 /* Initialize uninitialized itables */
993 
994 /*
995  * Mount flags set either automatically (could not be set by mount option)
996  * based on per file system feature or property or in special cases such as
997  * distinguishing between explicit mount option definition and default.
998  */
999 #define EXT4_MOUNT2_EXPLICIT_DELALLOC   0x00000001 /* User explicitly
1000                                                       specified delalloc */
1001 #define EXT4_MOUNT2_STD_GROUP_SIZE      0x00000002 /* We have standard group
1002                                                       size of blocksize * 8
1003                                                       blocks */
1004 #define EXT4_MOUNT2_HURD_COMPAT         0x00000004 /* Support HURD-castrated
1005                                                       file systems */
1006 
1007 #define clear_opt(sb, opt)              EXT4_SB(sb)->s_mount_opt &= \
1008                                                 ~EXT4_MOUNT_##opt
1009 #define set_opt(sb, opt)                EXT4_SB(sb)->s_mount_opt |= \
1010                                                 EXT4_MOUNT_##opt
1011 #define test_opt(sb, opt)               (EXT4_SB(sb)->s_mount_opt & \
1012                                          EXT4_MOUNT_##opt)
1013 
1014 #define clear_opt2(sb, opt)             EXT4_SB(sb)->s_mount_opt2 &= \
1015                                                 ~EXT4_MOUNT2_##opt
1016 #define set_opt2(sb, opt)               EXT4_SB(sb)->s_mount_opt2 |= \
1017                                                 EXT4_MOUNT2_##opt
1018 #define test_opt2(sb, opt)              (EXT4_SB(sb)->s_mount_opt2 & \
1019                                          EXT4_MOUNT2_##opt)
1020 
1021 #define ext4_test_and_set_bit           __test_and_set_bit_le
1022 #define ext4_set_bit                    __set_bit_le
1023 #define ext4_set_bit_atomic             ext2_set_bit_atomic
1024 #define ext4_test_and_clear_bit         __test_and_clear_bit_le
1025 #define ext4_clear_bit                  __clear_bit_le
1026 #define ext4_clear_bit_atomic           ext2_clear_bit_atomic
1027 #define ext4_test_bit                   test_bit_le
1028 #define ext4_find_next_zero_bit         find_next_zero_bit_le
1029 #define ext4_find_next_bit              find_next_bit_le
1030 
1031 extern void ext4_set_bits(void *bm, int cur, int len);
1032 
1033 /*
1034  * Maximal mount counts between two filesystem checks
1035  */
1036 #define EXT4_DFL_MAX_MNT_COUNT          20      /* Allow 20 mounts */
1037 #define EXT4_DFL_CHECKINTERVAL          0       /* Don't use interval check */
1038 
1039 /*
1040  * Behaviour when detecting errors
1041  */
1042 #define EXT4_ERRORS_CONTINUE            1       /* Continue execution */
1043 #define EXT4_ERRORS_RO                  2       /* Remount fs read-only */
1044 #define EXT4_ERRORS_PANIC               3       /* Panic */
1045 #define EXT4_ERRORS_DEFAULT             EXT4_ERRORS_CONTINUE
1046 
1047 /* Metadata checksum algorithm codes */
1048 #define EXT4_CRC32C_CHKSUM              1
1049 
1050 /*
1051  * Structure of the super block
1052  */
1053 struct ext4_super_block {
1054 /*00*/  __le32  s_inodes_count;         /* Inodes count */
1055         __le32  s_blocks_count_lo;      /* Blocks count */
1056         __le32  s_r_blocks_count_lo;    /* Reserved blocks count */
1057         __le32  s_free_blocks_count_lo; /* Free blocks count */
1058 /*10*/  __le32  s_free_inodes_count;    /* Free inodes count */
1059         __le32  s_first_data_block;     /* First Data Block */
1060         __le32  s_log_block_size;       /* Block size */
1061         __le32  s_log_cluster_size;     /* Allocation cluster size */
1062 /*20*/  __le32  s_blocks_per_group;     /* # Blocks per group */
1063         __le32  s_clusters_per_group;   /* # Clusters per group */
1064         __le32  s_inodes_per_group;     /* # Inodes per group */
1065         __le32  s_mtime;                /* Mount time */
1066 /*30*/  __le32  s_wtime;                /* Write time */
1067         __le16  s_mnt_count;            /* Mount count */
1068         __le16  s_max_mnt_count;        /* Maximal mount count */
1069         __le16  s_magic;                /* Magic signature */
1070         __le16  s_state;                /* File system state */
1071         __le16  s_errors;               /* Behaviour when detecting errors */
1072         __le16  s_minor_rev_level;      /* minor revision level */
1073 /*40*/  __le32  s_lastcheck;            /* time of last check */
1074         __le32  s_checkinterval;        /* max. time between checks */
1075         __le32  s_creator_os;           /* OS */
1076         __le32  s_rev_level;            /* Revision level */
1077 /*50*/  __le16  s_def_resuid;           /* Default uid for reserved blocks */
1078         __le16  s_def_resgid;           /* Default gid for reserved blocks */
1079         /*
1080          * These fields are for EXT4_DYNAMIC_REV superblocks only.
1081          *
1082          * Note: the difference between the compatible feature set and
1083          * the incompatible feature set is that if there is a bit set
1084          * in the incompatible feature set that the kernel doesn't
1085          * know about, it should refuse to mount the filesystem.
1086          *
1087          * e2fsck's requirements are more strict; if it doesn't know
1088          * about a feature in either the compatible or incompatible
1089          * feature set, it must abort and not try to meddle with
1090          * things it doesn't understand...
1091          */
1092         __le32  s_first_ino;            /* First non-reserved inode */
1093         __le16  s_inode_size;           /* size of inode structure */
1094         __le16  s_block_group_nr;       /* block group # of this superblock */
1095         __le32  s_feature_compat;       /* compatible feature set */
1096 /*60*/  __le32  s_feature_incompat;     /* incompatible feature set */
1097         __le32  s_feature_ro_compat;    /* readonly-compatible feature set */
1098 /*68*/  __u8    s_uuid[16];             /* 128-bit uuid for volume */
1099 /*78*/  char    s_volume_name[16];      /* volume name */
1100 /*88*/  char    s_last_mounted[64];     /* directory where last mounted */
1101 /*C8*/  __le32  s_algorithm_usage_bitmap; /* For compression */
1102         /*
1103          * Performance hints.  Directory preallocation should only
1104          * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1105          */
1106         __u8    s_prealloc_blocks;      /* Nr of blocks to try to preallocate*/
1107         __u8    s_prealloc_dir_blocks;  /* Nr to preallocate for dirs */
1108         __le16  s_reserved_gdt_blocks;  /* Per group desc for online growth */
1109         /*
1110          * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1111          */
1112 /*D0*/  __u8    s_journal_uuid[16];     /* uuid of journal superblock */
1113 /*E0*/  __le32  s_journal_inum;         /* inode number of journal file */
1114         __le32  s_journal_dev;          /* device number of journal file */
1115         __le32  s_last_orphan;          /* start of list of inodes to delete */
1116         __le32  s_hash_seed[4];         /* HTREE hash seed */
1117         __u8    s_def_hash_version;     /* Default hash version to use */
1118         __u8    s_jnl_backup_type;
1119         __le16  s_desc_size;            /* size of group descriptor */
1120 /*100*/ __le32  s_default_mount_opts;
1121         __le32  s_first_meta_bg;        /* First metablock block group */
1122         __le32  s_mkfs_time;            /* When the filesystem was created */
1123         __le32  s_jnl_blocks[17];       /* Backup of the journal inode */
1124         /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1125 /*150*/ __le32  s_blocks_count_hi;      /* Blocks count */
1126         __le32  s_r_blocks_count_hi;    /* Reserved blocks count */
1127         __le32  s_free_blocks_count_hi; /* Free blocks count */
1128         __le16  s_min_extra_isize;      /* All inodes have at least # bytes */
1129         __le16  s_want_extra_isize;     /* New inodes should reserve # bytes */
1130         __le32  s_flags;                /* Miscellaneous flags */
1131         __le16  s_raid_stride;          /* RAID stride */
1132         __le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1133         __le64  s_mmp_block;            /* Block for multi-mount protection */
1134         __le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1135         __u8    s_log_groups_per_flex;  /* FLEX_BG group size */
1136         __u8    s_checksum_type;        /* metadata checksum algorithm used */
1137         __le16  s_reserved_pad;
1138         __le64  s_kbytes_written;       /* nr of lifetime kilobytes written */
1139         __le32  s_snapshot_inum;        /* Inode number of active snapshot */
1140         __le32  s_snapshot_id;          /* sequential ID of active snapshot */
1141         __le64  s_snapshot_r_blocks_count; /* reserved blocks for active
1142                                               snapshot's future use */
1143         __le32  s_snapshot_list;        /* inode number of the head of the
1144                                            on-disk snapshot list */
1145 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1146         __le32  s_error_count;          /* number of fs errors */
1147         __le32  s_first_error_time;     /* first time an error happened */
1148         __le32  s_first_error_ino;      /* inode involved in first error */
1149         __le64  s_first_error_block;    /* block involved of first error */
1150         __u8    s_first_error_func[32]; /* function where the error happened */
1151         __le32  s_first_error_line;     /* line number where error happened */
1152         __le32  s_last_error_time;      /* most recent time of an error */
1153         __le32  s_last_error_ino;       /* inode involved in last error */
1154         __le32  s_last_error_line;      /* line number where error happened */
1155         __le64  s_last_error_block;     /* block involved of last error */
1156         __u8    s_last_error_func[32];  /* function where the error happened */
1157 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1158         __u8    s_mount_opts[64];
1159         __le32  s_usr_quota_inum;       /* inode for tracking user quota */
1160         __le32  s_grp_quota_inum;       /* inode for tracking group quota */
1161         __le32  s_overhead_clusters;    /* overhead blocks/clusters in fs */
1162         __le32  s_backup_bgs[2];        /* groups with sparse_super2 SBs */
1163         __le32  s_reserved[106];        /* Padding to the end of the block */
1164         __le32  s_checksum;             /* crc32c(superblock) */
1165 };
1166 
1167 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1168 
1169 #ifdef __KERNEL__
1170 
1171 /*
1172  * run-time mount flags
1173  */
1174 #define EXT4_MF_MNTDIR_SAMPLED  0x0001
1175 #define EXT4_MF_FS_ABORTED      0x0002  /* Fatal error detected */
1176 
1177 /*
1178  * fourth extended-fs super-block data in memory
1179  */
1180 struct ext4_sb_info {
1181         unsigned long s_desc_size;      /* Size of a group descriptor in bytes */
1182         unsigned long s_inodes_per_block;/* Number of inodes per block */
1183         unsigned long s_blocks_per_group;/* Number of blocks in a group */
1184         unsigned long s_clusters_per_group; /* Number of clusters in a group */
1185         unsigned long s_inodes_per_group;/* Number of inodes in a group */
1186         unsigned long s_itb_per_group;  /* Number of inode table blocks per group */
1187         unsigned long s_gdb_count;      /* Number of group descriptor blocks */
1188         unsigned long s_desc_per_block; /* Number of group descriptors per block */
1189         ext4_group_t s_groups_count;    /* Number of groups in the fs */
1190         ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1191         unsigned long s_overhead;  /* # of fs overhead clusters */
1192         unsigned int s_cluster_ratio;   /* Number of blocks per cluster */
1193         unsigned int s_cluster_bits;    /* log2 of s_cluster_ratio */
1194         loff_t s_bitmap_maxbytes;       /* max bytes for bitmap files */
1195         struct buffer_head * s_sbh;     /* Buffer containing the super block */
1196         struct ext4_super_block *s_es;  /* Pointer to the super block in the buffer */
1197         struct buffer_head **s_group_desc;
1198         unsigned int s_mount_opt;
1199         unsigned int s_mount_opt2;
1200         unsigned int s_mount_flags;
1201         unsigned int s_def_mount_opt;
1202         ext4_fsblk_t s_sb_block;
1203         atomic64_t s_resv_clusters;
1204         kuid_t s_resuid;
1205         kgid_t s_resgid;
1206         unsigned short s_mount_state;
1207         unsigned short s_pad;
1208         int s_addr_per_block_bits;
1209         int s_desc_per_block_bits;
1210         int s_inode_size;
1211         int s_first_ino;
1212         unsigned int s_inode_readahead_blks;
1213         unsigned int s_inode_goal;
1214         spinlock_t s_next_gen_lock;
1215         u32 s_next_generation;
1216         u32 s_hash_seed[4];
1217         int s_def_hash_version;
1218         int s_hash_unsigned;    /* 3 if hash should be signed, 0 if not */
1219         struct percpu_counter s_freeclusters_counter;
1220         struct percpu_counter s_freeinodes_counter;
1221         struct percpu_counter s_dirs_counter;
1222         struct percpu_counter s_dirtyclusters_counter;
1223         struct blockgroup_lock *s_blockgroup_lock;
1224         struct proc_dir_entry *s_proc;
1225         struct kobject s_kobj;
1226         struct completion s_kobj_unregister;
1227         struct super_block *s_sb;
1228 
1229         /* Journaling */
1230         struct journal_s *s_journal;
1231         struct list_head s_orphan;
1232         struct mutex s_orphan_lock;
1233         unsigned long s_resize_flags;           /* Flags indicating if there
1234                                                    is a resizer */
1235         unsigned long s_commit_interval;
1236         u32 s_max_batch_time;
1237         u32 s_min_batch_time;
1238         struct block_device *journal_bdev;
1239 #ifdef CONFIG_QUOTA
1240         char *s_qf_names[MAXQUOTAS];            /* Names of quota files with journalled quota */
1241         int s_jquota_fmt;                       /* Format of quota to use */
1242 #endif
1243         unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1244         struct rb_root system_blks;
1245 
1246 #ifdef EXTENTS_STATS
1247         /* ext4 extents stats */
1248         unsigned long s_ext_min;
1249         unsigned long s_ext_max;
1250         unsigned long s_depth_max;
1251         spinlock_t s_ext_stats_lock;
1252         unsigned long s_ext_blocks;
1253         unsigned long s_ext_extents;
1254 #endif
1255 
1256         /* for buddy allocator */
1257         struct ext4_group_info ***s_group_info;
1258         struct inode *s_buddy_cache;
1259         spinlock_t s_md_lock;
1260         unsigned short *s_mb_offsets;
1261         unsigned int *s_mb_maxs;
1262         unsigned int s_group_info_size;
1263 
1264         /* tunables */
1265         unsigned long s_stripe;
1266         unsigned int s_mb_stream_request;
1267         unsigned int s_mb_max_to_scan;
1268         unsigned int s_mb_min_to_scan;
1269         unsigned int s_mb_stats;
1270         unsigned int s_mb_order2_reqs;
1271         unsigned int s_mb_group_prealloc;
1272         unsigned int s_max_dir_size_kb;
1273         /* where last allocation was done - for stream allocation */
1274         unsigned long s_mb_last_group;
1275         unsigned long s_mb_last_start;
1276 
1277         /* stats for buddy allocator */
1278         atomic_t s_bal_reqs;    /* number of reqs with len > 1 */
1279         atomic_t s_bal_success; /* we found long enough chunks */
1280         atomic_t s_bal_allocated;       /* in blocks */
1281         atomic_t s_bal_ex_scanned;      /* total extents scanned */
1282         atomic_t s_bal_goals;   /* goal hits */
1283         atomic_t s_bal_breaks;  /* too long searches */
1284         atomic_t s_bal_2orders; /* 2^order hits */
1285         spinlock_t s_bal_lock;
1286         unsigned long s_mb_buddies_generated;
1287         unsigned long long s_mb_generation_time;
1288         atomic_t s_mb_lost_chunks;
1289         atomic_t s_mb_preallocated;
1290         atomic_t s_mb_discarded;
1291         atomic_t s_lock_busy;
1292 
1293         /* locality groups */
1294         struct ext4_locality_group __percpu *s_locality_groups;
1295 
1296         /* for write statistics */
1297         unsigned long s_sectors_written_start;
1298         u64 s_kbytes_written;
1299 
1300         /* the size of zero-out chunk */
1301         unsigned int s_extent_max_zeroout_kb;
1302 
1303         unsigned int s_log_groups_per_flex;
1304         struct flex_groups *s_flex_groups;
1305         ext4_group_t s_flex_groups_allocated;
1306 
1307         /* workqueue for reserved extent conversions (buffered io) */
1308         struct workqueue_struct *rsv_conversion_wq;
1309 
1310         /* timer for periodic error stats printing */
1311         struct timer_list s_err_report;
1312 
1313         /* Lazy inode table initialization info */
1314         struct ext4_li_request *s_li_request;
1315         /* Wait multiplier for lazy initialization thread */
1316         unsigned int s_li_wait_mult;
1317 
1318         /* Kernel thread for multiple mount protection */
1319         struct task_struct *s_mmp_tsk;
1320 
1321         /* record the last minlen when FITRIM is called. */
1322         atomic_t s_last_trim_minblks;
1323 
1324         /* Reference to checksum algorithm driver via cryptoapi */
1325         struct crypto_shash *s_chksum_driver;
1326 
1327         /* Precomputed FS UUID checksum for seeding other checksums */
1328         __u32 s_csum_seed;
1329 
1330         /* Reclaim extents from extent status tree */
1331         struct shrinker s_es_shrinker;
1332         struct list_head s_es_lru;
1333         unsigned long s_es_last_sorted;
1334         struct percpu_counter s_extent_cache_cnt;
1335         struct mb_cache *s_mb_cache;
1336         spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
1337 
1338         /* Ratelimit ext4 messages. */
1339         struct ratelimit_state s_err_ratelimit_state;
1340         struct ratelimit_state s_warning_ratelimit_state;
1341         struct ratelimit_state s_msg_ratelimit_state;
1342 };
1343 
1344 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1345 {
1346         return sb->s_fs_info;
1347 }
1348 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1349 {
1350         return container_of(inode, struct ext4_inode_info, vfs_inode);
1351 }
1352 
1353 static inline struct timespec ext4_current_time(struct inode *inode)
1354 {
1355         return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1356                 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1357 }
1358 
1359 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1360 {
1361         return ino == EXT4_ROOT_INO ||
1362                 ino == EXT4_USR_QUOTA_INO ||
1363                 ino == EXT4_GRP_QUOTA_INO ||
1364                 ino == EXT4_BOOT_LOADER_INO ||
1365                 ino == EXT4_JOURNAL_INO ||
1366                 ino == EXT4_RESIZE_INO ||
1367                 (ino >= EXT4_FIRST_INO(sb) &&
1368                  ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1369 }
1370 
1371 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1372                                               struct ext4_io_end *io_end)
1373 {
1374         if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1375                 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1376                 atomic_inc(&EXT4_I(inode)->i_unwritten);
1377         }
1378 }
1379 
1380 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1381 {
1382         return inode->i_private;
1383 }
1384 
1385 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1386 {
1387         inode->i_private = io;
1388 }
1389 
1390 /*
1391  * Inode dynamic state flags
1392  */
1393 enum {
1394         EXT4_STATE_JDATA,               /* journaled data exists */
1395         EXT4_STATE_NEW,                 /* inode is newly created */
1396         EXT4_STATE_XATTR,               /* has in-inode xattrs */
1397         EXT4_STATE_NO_EXPAND,           /* No space for expansion */
1398         EXT4_STATE_DA_ALLOC_CLOSE,      /* Alloc DA blks on close */
1399         EXT4_STATE_EXT_MIGRATE,         /* Inode is migrating */
1400         EXT4_STATE_DIO_UNWRITTEN,       /* need convert on dio done*/
1401         EXT4_STATE_NEWENTRY,            /* File just added to dir */
1402         EXT4_STATE_DELALLOC_RESERVED,   /* blks already reserved for delalloc */
1403         EXT4_STATE_DIOREAD_LOCK,        /* Disable support for dio read
1404                                            nolocking */
1405         EXT4_STATE_MAY_INLINE_DATA,     /* may have in-inode data */
1406         EXT4_STATE_ORDERED_MODE,        /* data=ordered mode */
1407         EXT4_STATE_EXT_PRECACHED,       /* extents have been precached */
1408 };
1409 
1410 #define EXT4_INODE_BIT_FNS(name, field, offset)                         \
1411 static inline int ext4_test_inode_##name(struct inode *inode, int bit)  \
1412 {                                                                       \
1413         return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);     \
1414 }                                                                       \
1415 static inline void ext4_set_inode_##name(struct inode *inode, int bit)  \
1416 {                                                                       \
1417         set_bit(bit + (offset), &EXT4_I(inode)->i_##field);             \
1418 }                                                                       \
1419 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1420 {                                                                       \
1421         clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);           \
1422 }
1423 
1424 /* Add these declarations here only so that these functions can be
1425  * found by name.  Otherwise, they are very hard to locate. */
1426 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1427 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1428 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1429 EXT4_INODE_BIT_FNS(flag, flags, 0)
1430 
1431 /* Add these declarations here only so that these functions can be
1432  * found by name.  Otherwise, they are very hard to locate. */
1433 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1434 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1435 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1436 #if (BITS_PER_LONG < 64)
1437 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1438 
1439 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1440 {
1441         (ei)->i_state_flags = 0;
1442 }
1443 #else
1444 EXT4_INODE_BIT_FNS(state, flags, 32)
1445 
1446 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1447 {
1448         /* We depend on the fact that callers will set i_flags */
1449 }
1450 #endif
1451 #else
1452 /* Assume that user mode programs are passing in an ext4fs superblock, not
1453  * a kernel struct super_block.  This will allow us to call the feature-test
1454  * macros from user land. */
1455 #define EXT4_SB(sb)     (sb)
1456 #endif
1457 
1458 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1459 
1460 /*
1461  * Codes for operating systems
1462  */
1463 #define EXT4_OS_LINUX           0
1464 #define EXT4_OS_HURD            1
1465 #define EXT4_OS_MASIX           2
1466 #define EXT4_OS_FREEBSD         3
1467 #define EXT4_OS_LITES           4
1468 
1469 /*
1470  * Revision levels
1471  */
1472 #define EXT4_GOOD_OLD_REV       0       /* The good old (original) format */
1473 #define EXT4_DYNAMIC_REV        1       /* V2 format w/ dynamic inode sizes */
1474 
1475 #define EXT4_CURRENT_REV        EXT4_GOOD_OLD_REV
1476 #define EXT4_MAX_SUPP_REV       EXT4_DYNAMIC_REV
1477 
1478 #define EXT4_GOOD_OLD_INODE_SIZE 128
1479 
1480 /*
1481  * Feature set definitions
1482  */
1483 
1484 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)                        \
1485         ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1486 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)                     \
1487         ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1488 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)                      \
1489         ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1490 #define EXT4_SET_COMPAT_FEATURE(sb,mask)                        \
1491         EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1492 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)                     \
1493         EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1494 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)                      \
1495         EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1496 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)                      \
1497         EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1498 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)                   \
1499         EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1500 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)                    \
1501         EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1502 
1503 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC        0x0001
1504 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES       0x0002
1505 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL         0x0004
1506 #define EXT4_FEATURE_COMPAT_EXT_ATTR            0x0008
1507 #define EXT4_FEATURE_COMPAT_RESIZE_INODE        0x0010
1508 #define EXT4_FEATURE_COMPAT_DIR_INDEX           0x0020
1509 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2       0x0200
1510 
1511 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER     0x0001
1512 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE       0x0002
1513 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR        0x0004
1514 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1515 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM         0x0010
1516 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK        0x0020
1517 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE      0x0040
1518 #define EXT4_FEATURE_RO_COMPAT_QUOTA            0x0100
1519 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC         0x0200
1520 /*
1521  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1522  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1523  * all other data structures' checksums.  However, the METADATA_CSUM and
1524  * GDT_CSUM bits are mutually exclusive.
1525  */
1526 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM    0x0400
1527 
1528 #define EXT4_FEATURE_INCOMPAT_COMPRESSION       0x0001
1529 #define EXT4_FEATURE_INCOMPAT_FILETYPE          0x0002
1530 #define EXT4_FEATURE_INCOMPAT_RECOVER           0x0004 /* Needs recovery */
1531 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV       0x0008 /* Journal device */
1532 #define EXT4_FEATURE_INCOMPAT_META_BG           0x0010
1533 #define EXT4_FEATURE_INCOMPAT_EXTENTS           0x0040 /* extents support */
1534 #define EXT4_FEATURE_INCOMPAT_64BIT             0x0080
1535 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1536 #define EXT4_FEATURE_INCOMPAT_FLEX_BG           0x0200
1537 #define EXT4_FEATURE_INCOMPAT_EA_INODE          0x0400 /* EA in inode */
1538 #define EXT4_FEATURE_INCOMPAT_DIRDATA           0x1000 /* data in dirent */
1539 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM  0x2000 /* use crc32c for bg */
1540 #define EXT4_FEATURE_INCOMPAT_LARGEDIR          0x4000 /* >2GB or 3-lvl htree */
1541 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA       0x8000 /* data in inode */
1542 
1543 #define EXT2_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1544 #define EXT2_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1545                                          EXT4_FEATURE_INCOMPAT_META_BG)
1546 #define EXT2_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1547                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1548                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1549 
1550 #define EXT3_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1551 #define EXT3_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1552                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1553                                          EXT4_FEATURE_INCOMPAT_META_BG)
1554 #define EXT3_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1555                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1556                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1557 
1558 #define EXT4_FEATURE_COMPAT_SUPP        EXT2_FEATURE_COMPAT_EXT_ATTR
1559 #define EXT4_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1560                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1561                                          EXT4_FEATURE_INCOMPAT_META_BG| \
1562                                          EXT4_FEATURE_INCOMPAT_EXTENTS| \
1563                                          EXT4_FEATURE_INCOMPAT_64BIT| \
1564                                          EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1565                                          EXT4_FEATURE_INCOMPAT_MMP |    \
1566                                          EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1567 #define EXT4_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1568                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1569                                          EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1570                                          EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1571                                          EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1572                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1573                                          EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1574                                          EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1575                                          EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1576                                          EXT4_FEATURE_RO_COMPAT_QUOTA)
1577 
1578 /*
1579  * Default values for user and/or group using reserved blocks
1580  */
1581 #define EXT4_DEF_RESUID         0
1582 #define EXT4_DEF_RESGID         0
1583 
1584 #define EXT4_DEF_INODE_READAHEAD_BLKS   32
1585 
1586 /*
1587  * Default mount options
1588  */
1589 #define EXT4_DEFM_DEBUG         0x0001
1590 #define EXT4_DEFM_BSDGROUPS     0x0002
1591 #define EXT4_DEFM_XATTR_USER    0x0004
1592 #define EXT4_DEFM_ACL           0x0008
1593 #define EXT4_DEFM_UID16         0x0010
1594 #define EXT4_DEFM_JMODE         0x0060
1595 #define EXT4_DEFM_JMODE_DATA    0x0020
1596 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1597 #define EXT4_DEFM_JMODE_WBACK   0x0060
1598 #define EXT4_DEFM_NOBARRIER     0x0100
1599 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1600 #define EXT4_DEFM_DISCARD       0x0400
1601 #define EXT4_DEFM_NODELALLOC    0x0800
1602 
1603 /*
1604  * Default journal batch times
1605  */
1606 #define EXT4_DEF_MIN_BATCH_TIME 0
1607 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1608 
1609 /*
1610  * Minimum number of groups in a flexgroup before we separate out
1611  * directories into the first block group of a flexgroup
1612  */
1613 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1614 
1615 /*
1616  * Structure of a directory entry
1617  */
1618 #define EXT4_NAME_LEN 255
1619 
1620 struct ext4_dir_entry {
1621         __le32  inode;                  /* Inode number */
1622         __le16  rec_len;                /* Directory entry length */
1623         __le16  name_len;               /* Name length */
1624         char    name[EXT4_NAME_LEN];    /* File name */
1625 };
1626 
1627 /*
1628  * The new version of the directory entry.  Since EXT4 structures are
1629  * stored in intel byte order, and the name_len field could never be
1630  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1631  * file_type field.
1632  */
1633 struct ext4_dir_entry_2 {
1634         __le32  inode;                  /* Inode number */
1635         __le16  rec_len;                /* Directory entry length */
1636         __u8    name_len;               /* Name length */
1637         __u8    file_type;
1638         char    name[EXT4_NAME_LEN];    /* File name */
1639 };
1640 
1641 /*
1642  * This is a bogus directory entry at the end of each leaf block that
1643  * records checksums.
1644  */
1645 struct ext4_dir_entry_tail {
1646         __le32  det_reserved_zero1;     /* Pretend to be unused */
1647         __le16  det_rec_len;            /* 12 */
1648         __u8    det_reserved_zero2;     /* Zero name length */
1649         __u8    det_reserved_ft;        /* 0xDE, fake file type */
1650         __le32  det_checksum;           /* crc32c(uuid+inum+dirblock) */
1651 };
1652 
1653 #define EXT4_DIRENT_TAIL(block, blocksize) \
1654         ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1655                                         ((blocksize) - \
1656                                          sizeof(struct ext4_dir_entry_tail))))
1657 
1658 /*
1659  * Ext4 directory file types.  Only the low 3 bits are used.  The
1660  * other bits are reserved for now.
1661  */
1662 #define EXT4_FT_UNKNOWN         0
1663 #define EXT4_FT_REG_FILE        1
1664 #define EXT4_FT_DIR             2
1665 #define EXT4_FT_CHRDEV          3
1666 #define EXT4_FT_BLKDEV          4
1667 #define EXT4_FT_FIFO            5
1668 #define EXT4_FT_SOCK            6
1669 #define EXT4_FT_SYMLINK         7
1670 
1671 #define EXT4_FT_MAX             8
1672 
1673 #define EXT4_FT_DIR_CSUM        0xDE
1674 
1675 /*
1676  * EXT4_DIR_PAD defines the directory entries boundaries
1677  *
1678  * NOTE: It must be a multiple of 4
1679  */
1680 #define EXT4_DIR_PAD                    4
1681 #define EXT4_DIR_ROUND                  (EXT4_DIR_PAD - 1)
1682 #define EXT4_DIR_REC_LEN(name_len)      (((name_len) + 8 + EXT4_DIR_ROUND) & \
1683                                          ~EXT4_DIR_ROUND)
1684 #define EXT4_MAX_REC_LEN                ((1<<16)-1)
1685 
1686 /*
1687  * If we ever get support for fs block sizes > page_size, we'll need
1688  * to remove the #if statements in the next two functions...
1689  */
1690 static inline unsigned int
1691 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1692 {
1693         unsigned len = le16_to_cpu(dlen);
1694 
1695 #if (PAGE_CACHE_SIZE >= 65536)
1696         if (len == EXT4_MAX_REC_LEN || len == 0)
1697                 return blocksize;
1698         return (len & 65532) | ((len & 3) << 16);
1699 #else
1700         return len;
1701 #endif
1702 }
1703 
1704 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1705 {
1706         if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1707                 BUG();
1708 #if (PAGE_CACHE_SIZE >= 65536)
1709         if (len < 65536)
1710                 return cpu_to_le16(len);
1711         if (len == blocksize) {
1712                 if (blocksize == 65536)
1713                         return cpu_to_le16(EXT4_MAX_REC_LEN);
1714                 else
1715                         return cpu_to_le16(0);
1716         }
1717         return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1718 #else
1719         return cpu_to_le16(len);
1720 #endif
1721 }
1722 
1723 /*
1724  * Hash Tree Directory indexing
1725  * (c) Daniel Phillips, 2001
1726  */
1727 
1728 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1729                                       EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1730                     ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1731 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1732 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1733 
1734 /* Legal values for the dx_root hash_version field: */
1735 
1736 #define DX_HASH_LEGACY          0
1737 #define DX_HASH_HALF_MD4        1
1738 #define DX_HASH_TEA             2
1739 #define DX_HASH_LEGACY_UNSIGNED 3
1740 #define DX_HASH_HALF_MD4_UNSIGNED       4
1741 #define DX_HASH_TEA_UNSIGNED            5
1742 
1743 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1744                               const void *address, unsigned int length)
1745 {
1746         struct {
1747                 struct shash_desc shash;
1748                 char ctx[4];
1749         } desc;
1750         int err;
1751 
1752         BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1753 
1754         desc.shash.tfm = sbi->s_chksum_driver;
1755         desc.shash.flags = 0;
1756         *(u32 *)desc.ctx = crc;
1757 
1758         err = crypto_shash_update(&desc.shash, address, length);
1759         BUG_ON(err);
1760 
1761         return *(u32 *)desc.ctx;
1762 }
1763 
1764 #ifdef __KERNEL__
1765 
1766 /* hash info structure used by the directory hash */
1767 struct dx_hash_info
1768 {
1769         u32             hash;
1770         u32             minor_hash;
1771         int             hash_version;
1772         u32             *seed;
1773 };
1774 
1775 
1776 /* 32 and 64 bit signed EOF for dx directories */
1777 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
1778 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
1779 
1780 
1781 /*
1782  * Control parameters used by ext4_htree_next_block
1783  */
1784 #define HASH_NB_ALWAYS          1
1785 
1786 
1787 /*
1788  * Describe an inode's exact location on disk and in memory
1789  */
1790 struct ext4_iloc
1791 {
1792         struct buffer_head *bh;
1793         unsigned long offset;
1794         ext4_group_t block_group;
1795 };
1796 
1797 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1798 {
1799         return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1800 }
1801 
1802 /*
1803  * This structure is stuffed into the struct file's private_data field
1804  * for directories.  It is where we put information so that we can do
1805  * readdir operations in hash tree order.
1806  */
1807 struct dir_private_info {
1808         struct rb_root  root;
1809         struct rb_node  *curr_node;
1810         struct fname    *extra_fname;
1811         loff_t          last_pos;
1812         __u32           curr_hash;
1813         __u32           curr_minor_hash;
1814         __u32           next_hash;
1815 };
1816 
1817 /* calculate the first block number of the group */
1818 static inline ext4_fsblk_t
1819 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1820 {
1821         return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1822                 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1823 }
1824 
1825 /*
1826  * Special error return code only used by dx_probe() and its callers.
1827  */
1828 #define ERR_BAD_DX_DIR  (-(MAX_ERRNO - 1))
1829 
1830 /*
1831  * Timeout and state flag for lazy initialization inode thread.
1832  */
1833 #define EXT4_DEF_LI_WAIT_MULT                   10
1834 #define EXT4_DEF_LI_MAX_START_DELAY             5
1835 #define EXT4_LAZYINIT_QUIT                      0x0001
1836 #define EXT4_LAZYINIT_RUNNING                   0x0002
1837 
1838 /*
1839  * Lazy inode table initialization info
1840  */
1841 struct ext4_lazy_init {
1842         unsigned long           li_state;
1843         struct list_head        li_request_list;
1844         struct mutex            li_list_mtx;
1845 };
1846 
1847 struct ext4_li_request {
1848         struct super_block      *lr_super;
1849         struct ext4_sb_info     *lr_sbi;
1850         ext4_group_t            lr_next_group;
1851         struct list_head        lr_request;
1852         unsigned long           lr_next_sched;
1853         unsigned long           lr_timeout;
1854 };
1855 
1856 struct ext4_features {
1857         struct kobject f_kobj;
1858         struct completion f_kobj_unregister;
1859 };
1860 
1861 /*
1862  * This structure will be used for multiple mount protection. It will be
1863  * written into the block number saved in the s_mmp_block field in the
1864  * superblock. Programs that check MMP should assume that if
1865  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1866  * to use the filesystem, regardless of how old the timestamp is.
1867  */
1868 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1869 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1870 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1871 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1872 
1873 struct mmp_struct {
1874         __le32  mmp_magic;              /* Magic number for MMP */
1875         __le32  mmp_seq;                /* Sequence no. updated periodically */
1876 
1877         /*
1878          * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1879          * purposes and do not affect the correctness of the algorithm
1880          */
1881         __le64  mmp_time;               /* Time last updated */
1882         char    mmp_nodename[64];       /* Node which last updated MMP block */
1883         char    mmp_bdevname[32];       /* Bdev which last updated MMP block */
1884 
1885         /*
1886          * mmp_check_interval is used to verify if the MMP block has been
1887          * updated on the block device. The value is updated based on the
1888          * maximum time to write the MMP block during an update cycle.
1889          */
1890         __le16  mmp_check_interval;
1891 
1892         __le16  mmp_pad1;
1893         __le32  mmp_pad2[226];
1894         __le32  mmp_checksum;           /* crc32c(uuid+mmp_block) */
1895 };
1896 
1897 /* arguments passed to the mmp thread */
1898 struct mmpd_data {
1899         struct buffer_head *bh; /* bh from initial read_mmp_block() */
1900         struct super_block *sb;  /* super block of the fs */
1901 };
1902 
1903 /*
1904  * Check interval multiplier
1905  * The MMP block is written every update interval and initially checked every
1906  * update interval x the multiplier (the value is then adapted based on the
1907  * write latency). The reason is that writes can be delayed under load and we
1908  * don't want readers to incorrectly assume that the filesystem is no longer
1909  * in use.
1910  */
1911 #define EXT4_MMP_CHECK_MULT             2UL
1912 
1913 /*
1914  * Minimum interval for MMP checking in seconds.
1915  */
1916 #define EXT4_MMP_MIN_CHECK_INTERVAL     5UL
1917 
1918 /*
1919  * Maximum interval for MMP checking in seconds.
1920  */
1921 #define EXT4_MMP_MAX_CHECK_INTERVAL     300UL
1922 
1923 /*
1924  * Function prototypes
1925  */
1926 
1927 /*
1928  * Ok, these declarations are also in <linux/kernel.h> but none of the
1929  * ext4 source programs needs to include it so they are duplicated here.
1930  */
1931 # define NORET_TYPE     /**/
1932 # define ATTRIB_NORET   __attribute__((noreturn))
1933 # define NORET_AND      noreturn,
1934 
1935 /* bitmap.c */
1936 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1937 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1938                                 struct ext4_group_desc *gdp,
1939                                 struct buffer_head *bh, int sz);
1940 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1941                                   struct ext4_group_desc *gdp,
1942                                   struct buffer_head *bh, int sz);
1943 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1944                                 struct ext4_group_desc *gdp,
1945                                 struct buffer_head *bh);
1946 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1947                                   struct ext4_group_desc *gdp,
1948                                   struct buffer_head *bh);
1949 
1950 /* balloc.c */
1951 extern void ext4_get_group_no_and_offset(struct super_block *sb,
1952                                          ext4_fsblk_t blocknr,
1953                                          ext4_group_t *blockgrpp,
1954                                          ext4_grpblk_t *offsetp);
1955 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1956                                           ext4_fsblk_t block);
1957 
1958 extern unsigned int ext4_block_group(struct super_block *sb,
1959                         ext4_fsblk_t blocknr);
1960 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1961                         ext4_fsblk_t blocknr);
1962 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1963 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1964                         ext4_group_t group);
1965 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1966                                          ext4_fsblk_t goal,
1967                                          unsigned int flags,
1968                                          unsigned long *count,
1969                                          int *errp);
1970 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1971                                     s64 nclusters, unsigned int flags);
1972 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1973 extern void ext4_check_blocks_bitmap(struct super_block *);
1974 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1975                                                     ext4_group_t block_group,
1976                                                     struct buffer_head ** bh);
1977 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1978 
1979 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1980                                                 ext4_group_t block_group);
1981 extern int ext4_wait_block_bitmap(struct super_block *sb,
1982                                   ext4_group_t block_group,
1983                                   struct buffer_head *bh);
1984 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1985                                                   ext4_group_t block_group);
1986 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1987                                               ext4_group_t block_group,
1988                                               struct ext4_group_desc *gdp);
1989 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1990 
1991 /* dir.c */
1992 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1993                                   struct file *,
1994                                   struct ext4_dir_entry_2 *,
1995                                   struct buffer_head *, char *, int,
1996                                   unsigned int);
1997 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)      \
1998         unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1999                                         (de), (bh), (buf), (size), (offset)))
2000 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2001                                     __u32 minor_hash,
2002                                     struct ext4_dir_entry_2 *dirent);
2003 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2004 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2005                              struct buffer_head *bh,
2006                              void *buf, int buf_size,
2007                              const char *name, int namelen,
2008                              struct ext4_dir_entry_2 **dest_de);
2009 void ext4_insert_dentry(struct inode *inode,
2010                         struct ext4_dir_entry_2 *de,
2011                         int buf_size,
2012                         const char *name, int namelen);
2013 static inline void ext4_update_dx_flag(struct inode *inode)
2014 {
2015         if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2016                                      EXT4_FEATURE_COMPAT_DIR_INDEX))
2017                 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2018 }
2019 static unsigned char ext4_filetype_table[] = {
2020         DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2021 };
2022 
2023 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2024 {
2025         if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2026             (filetype >= EXT4_FT_MAX))
2027                 return DT_UNKNOWN;
2028 
2029         return ext4_filetype_table[filetype];
2030 }
2031 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2032                              void *buf, int buf_size);
2033 
2034 /* fsync.c */
2035 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2036 
2037 /* hash.c */
2038 extern int ext4fs_dirhash(const char *name, int len, struct
2039                           dx_hash_info *hinfo);
2040 
2041 /* ialloc.c */
2042 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2043                                       const struct qstr *qstr, __u32 goal,
2044                                       uid_t *owner, int handle_type,
2045                                       unsigned int line_no, int nblocks);
2046 
2047 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2048         __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2049                          0, 0, 0)
2050 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2051                                     type, nblocks)                  \
2052         __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2053                          (type), __LINE__, (nblocks))
2054 
2055 
2056 extern void ext4_free_inode(handle_t *, struct inode *);
2057 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2058 extern unsigned long ext4_count_free_inodes(struct super_block *);
2059 extern unsigned long ext4_count_dirs(struct super_block *);
2060 extern void ext4_check_inodes_bitmap(struct super_block *);
2061 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2062 extern int ext4_init_inode_table(struct super_block *sb,
2063                                  ext4_group_t group, int barrier);
2064 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2065 
2066 /* mballoc.c */
2067 extern long ext4_mb_stats;
2068 extern long ext4_mb_max_to_scan;
2069 extern int ext4_mb_init(struct super_block *);
2070 extern int ext4_mb_release(struct super_block *);
2071 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2072                                 struct ext4_allocation_request *, int *);
2073 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2074 extern void ext4_discard_preallocations(struct inode *);
2075 extern int __init ext4_init_mballoc(void);
2076 extern void ext4_exit_mballoc(void);
2077 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2078                              struct buffer_head *bh, ext4_fsblk_t block,
2079                              unsigned long count, int flags);
2080 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2081                                    ext4_group_t ngroups);
2082 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2083                 ext4_group_t i, struct ext4_group_desc *desc);
2084 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2085                                 ext4_fsblk_t block, unsigned long count);
2086 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2087 
2088 /* inode.c */
2089 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2090                                                 ext4_lblk_t, int, int *);
2091 struct buffer_head *ext4_bread(handle_t *, struct inode *,
2092                                                 ext4_lblk_t, int, int *);
2093 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2094                          struct buffer_head *bh_result, int create);
2095 int ext4_get_block(struct inode *inode, sector_t iblock,
2096                                 struct buffer_head *bh_result, int create);
2097 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2098                            struct buffer_head *bh, int create);
2099 int ext4_walk_page_buffers(handle_t *handle,
2100                            struct buffer_head *head,
2101                            unsigned from,
2102                            unsigned to,
2103                            int *partial,
2104                            int (*fn)(handle_t *handle,
2105                                      struct buffer_head *bh));
2106 int do_journal_get_write_access(handle_t *handle,
2107                                 struct buffer_head *bh);
2108 #define FALL_BACK_TO_NONDELALLOC 1
2109 #define CONVERT_INLINE_DATA      2
2110 
2111 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2112 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2113 extern int  ext4_setattr(struct dentry *, struct iattr *);
2114 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2115                                 struct kstat *stat);
2116 extern void ext4_evict_inode(struct inode *);
2117 extern void ext4_clear_inode(struct inode *);
2118 extern int  ext4_sync_inode(handle_t *, struct inode *);
2119 extern void ext4_dirty_inode(struct inode *, int);
2120 extern int ext4_change_inode_journal_flag(struct inode *, int);
2121 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2122 extern int ext4_inode_attach_jinode(struct inode *inode);
2123 extern int ext4_can_truncate(struct inode *inode);
2124 extern void ext4_truncate(struct inode *);
2125 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2126 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2127 extern void ext4_set_inode_flags(struct inode *);
2128 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2129 extern int ext4_alloc_da_blocks(struct inode *inode);
2130 extern void ext4_set_aops(struct inode *inode);
2131 extern int ext4_writepage_trans_blocks(struct inode *);
2132 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2133 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2134                              loff_t lstart, loff_t lend);
2135 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2136 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2137 extern void ext4_da_update_reserve_space(struct inode *inode,
2138                                         int used, int quota_claim);
2139 
2140 /* indirect.c */
2141 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2142                                 struct ext4_map_blocks *map, int flags);
2143 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2144                                 struct iov_iter *iter, loff_t offset);
2145 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2146 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2147 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2148 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2149                                  ext4_lblk_t start, ext4_lblk_t end);
2150 
2151 /* ioctl.c */
2152 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2153 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2154 
2155 /* migrate.c */
2156 extern int ext4_ext_migrate(struct inode *);
2157 extern int ext4_ind_migrate(struct inode *inode);
2158 
2159 /* namei.c */
2160 extern int ext4_dirent_csum_verify(struct inode *inode,
2161                                    struct ext4_dir_entry *dirent);
2162 extern int ext4_orphan_add(handle_t *, struct inode *);
2163 extern int ext4_orphan_del(handle_t *, struct inode *);
2164 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2165                                 __u32 start_minor_hash, __u32 *next_hash);
2166 extern int search_dir(struct buffer_head *bh,
2167                       char *search_buf,
2168                       int buf_size,
2169                       struct inode *dir,
2170                       const struct qstr *d_name,
2171                       unsigned int offset,
2172                       struct ext4_dir_entry_2 **res_dir);
2173 extern int ext4_generic_delete_entry(handle_t *handle,
2174                                      struct inode *dir,
2175                                      struct ext4_dir_entry_2 *de_del,
2176                                      struct buffer_head *bh,
2177                                      void *entry_buf,
2178                                      int buf_size,
2179                                      int csum_size);
2180 
2181 /* resize.c */
2182 extern int ext4_group_add(struct super_block *sb,
2183                                 struct ext4_new_group_data *input);
2184 extern int ext4_group_extend(struct super_block *sb,
2185                                 struct ext4_super_block *es,
2186                                 ext4_fsblk_t n_blocks_count);
2187 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2188 
2189 /* super.c */
2190 extern int ext4_calculate_overhead(struct super_block *sb);
2191 extern void ext4_superblock_csum_set(struct super_block *sb);
2192 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2193 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2194 extern void ext4_kvfree(void *ptr);
2195 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2196                                     ext4_group_t ngroup);
2197 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2198                                      char nbuf[16]);
2199 
2200 extern __printf(4, 5)
2201 void __ext4_error(struct super_block *, const char *, unsigned int,
2202                   const char *, ...);
2203 extern __printf(5, 6)
2204 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2205                       const char *, ...);
2206 extern __printf(5, 6)
2207 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2208                      const char *, ...);
2209 extern void __ext4_std_error(struct super_block *, const char *,
2210                              unsigned int, int);
2211 extern __printf(4, 5)
2212 void __ext4_abort(struct super_block *, const char *, unsigned int,
2213                   const char *, ...);
2214 extern __printf(4, 5)
2215 void __ext4_warning(struct super_block *, const char *, unsigned int,
2216                     const char *, ...);
2217 extern __printf(3, 4)
2218 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2219 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2220                            const char *, unsigned int, const char *);
2221 extern __printf(7, 8)
2222 void __ext4_grp_locked_error(const char *, unsigned int,
2223                              struct super_block *, ext4_group_t,
2224                              unsigned long, ext4_fsblk_t,
2225                              const char *, ...);
2226 
2227 #ifdef CONFIG_PRINTK
2228 
2229 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2230         __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2231 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2232         __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2233 #define ext4_error(sb, fmt, ...)                                        \
2234         __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2235 #define ext4_abort(sb, fmt, ...)                                        \
2236         __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2237 #define ext4_warning(sb, fmt, ...)                                      \
2238         __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2239 #define ext4_msg(sb, level, fmt, ...)                           \
2240         __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2241 #define dump_mmp_msg(sb, mmp, msg)                                      \
2242         __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2243 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2244         __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2245                                 fmt, ##__VA_ARGS__)
2246 
2247 #else
2248 
2249 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2250 do {                                                                    \
2251         no_printk(fmt, ##__VA_ARGS__);                                  \
2252         __ext4_error_inode(inode, "", 0, block, " ");                   \
2253 } while (0)
2254 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2255 do {                                                                    \
2256         no_printk(fmt, ##__VA_ARGS__);                                  \
2257         __ext4_error_file(file, "", 0, block, " ");                     \
2258 } while (0)
2259 #define ext4_error(sb, fmt, ...)                                        \
2260 do {                                                                    \
2261         no_printk(fmt, ##__VA_ARGS__);                                  \
2262         __ext4_error(sb, "", 0, " ");                                   \
2263 } while (0)
2264 #define ext4_abort(sb, fmt, ...)                                        \
2265 do {                                                                    \
2266         no_printk(fmt, ##__VA_ARGS__);                                  \
2267         __ext4_abort(sb, "", 0, " ");                                   \
2268 } while (0)
2269 #define ext4_warning(sb, fmt, ...)                                      \
2270 do {                                                                    \
2271         no_printk(fmt, ##__VA_ARGS__);                                  \
2272         __ext4_warning(sb, "", 0, " ");                                 \
2273 } while (0)
2274 #define ext4_msg(sb, level, fmt, ...)                                   \
2275 do {                                                                    \
2276         no_printk(fmt, ##__VA_ARGS__);                                  \
2277         __ext4_msg(sb, "", " ");                                        \
2278 } while (0)
2279 #define dump_mmp_msg(sb, mmp, msg)                                      \
2280         __dump_mmp_msg(sb, mmp, "", 0, "")
2281 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2282 do {                                                                    \
2283         no_printk(fmt, ##__VA_ARGS__);                          \
2284         __ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");       \
2285 } while (0)
2286 
2287 #endif
2288 
2289 extern void ext4_update_dynamic_rev(struct super_block *sb);
2290 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2291                                         __u32 compat);
2292 extern int ext4_update_rocompat_feature(handle_t *handle,
2293                                         struct super_block *sb, __u32 rocompat);
2294 extern int ext4_update_incompat_feature(handle_t *handle,
2295                                         struct super_block *sb, __u32 incompat);
2296 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2297                                       struct ext4_group_desc *bg);
2298 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2299                                       struct ext4_group_desc *bg);
2300 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2301                                      struct ext4_group_desc *bg);
2302 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2303                                       struct ext4_group_desc *bg);
2304 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2305                                  struct ext4_group_desc *bg);
2306 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2307                                 struct ext4_group_desc *bg);
2308 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2309                                    struct ext4_group_desc *bg);
2310 extern void ext4_block_bitmap_set(struct super_block *sb,
2311                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2312 extern void ext4_inode_bitmap_set(struct super_block *sb,
2313                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2314 extern void ext4_inode_table_set(struct super_block *sb,
2315                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2316 extern void ext4_free_group_clusters_set(struct super_block *sb,
2317                                          struct ext4_group_desc *bg,
2318                                          __u32 count);
2319 extern void ext4_free_inodes_set(struct super_block *sb,
2320                                 struct ext4_group_desc *bg, __u32 count);
2321 extern void ext4_used_dirs_set(struct super_block *sb,
2322                                 struct ext4_group_desc *bg, __u32 count);
2323 extern void ext4_itable_unused_set(struct super_block *sb,
2324                                    struct ext4_group_desc *bg, __u32 count);
2325 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2326                                        struct ext4_group_desc *gdp);
2327 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2328                                      struct ext4_group_desc *gdp);
2329 extern int ext4_register_li_request(struct super_block *sb,
2330                                     ext4_group_t first_not_zeroed);
2331 
2332 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2333 {
2334         return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2335                                           EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2336                                           EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2337 }
2338 
2339 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2340 {
2341         return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2342                 le32_to_cpu(es->s_blocks_count_lo);
2343 }
2344 
2345 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2346 {
2347         return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2348                 le32_to_cpu(es->s_r_blocks_count_lo);
2349 }
2350 
2351 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2352 {
2353         return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2354                 le32_to_cpu(es->s_free_blocks_count_lo);
2355 }
2356 
2357 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2358                                          ext4_fsblk_t blk)
2359 {
2360         es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2361         es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2362 }
2363 
2364 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2365                                               ext4_fsblk_t blk)
2366 {
2367         es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2368         es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2369 }
2370 
2371 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2372                                            ext4_fsblk_t blk)
2373 {
2374         es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2375         es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2376 }
2377 
2378 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2379 {
2380         if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2381                 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2382                         le32_to_cpu(raw_inode->i_size_lo);
2383         else
2384                 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2385 }
2386 
2387 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2388 {
2389         raw_inode->i_size_lo = cpu_to_le32(i_size);
2390         raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2391 }
2392 
2393 static inline
2394 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2395                                             ext4_group_t group)
2396 {
2397          struct ext4_group_info ***grp_info;
2398          long indexv, indexh;
2399          BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2400          grp_info = EXT4_SB(sb)->s_group_info;
2401          indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2402          indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2403          return grp_info[indexv][indexh];
2404 }
2405 
2406 /*
2407  * Reading s_groups_count requires using smp_rmb() afterwards.  See
2408  * the locking protocol documented in the comments of ext4_group_add()
2409  * in resize.c
2410  */
2411 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2412 {
2413         ext4_group_t    ngroups = EXT4_SB(sb)->s_groups_count;
2414 
2415         smp_rmb();
2416         return ngroups;
2417 }
2418 
2419 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2420                                              ext4_group_t block_group)
2421 {
2422         return block_group >> sbi->s_log_groups_per_flex;
2423 }
2424 
2425 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2426 {
2427         return 1 << sbi->s_log_groups_per_flex;
2428 }
2429 
2430 #define ext4_std_error(sb, errno)                               \
2431 do {                                                            \
2432         if ((errno))                                            \
2433                 __ext4_std_error((sb), __func__, __LINE__, (errno));    \
2434 } while (0)
2435 
2436 #ifdef CONFIG_SMP
2437 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2438  * counters. So we need to make sure we have free clusters more
2439  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2440  */
2441 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2442 #else
2443 #define EXT4_FREECLUSTERS_WATERMARK 0
2444 #endif
2445 
2446 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2447 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2448 {
2449         WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2450                      !mutex_is_locked(&inode->i_mutex));
2451         down_write(&EXT4_I(inode)->i_data_sem);
2452         if (newsize > EXT4_I(inode)->i_disksize)
2453                 EXT4_I(inode)->i_disksize = newsize;
2454         up_write(&EXT4_I(inode)->i_data_sem);
2455 }
2456 
2457 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2458 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2459 {
2460         int changed = 0;
2461 
2462         if (newsize > inode->i_size) {
2463                 i_size_write(inode, newsize);
2464                 changed = 1;
2465         }
2466         if (newsize > EXT4_I(inode)->i_disksize) {
2467                 ext4_update_i_disksize(inode, newsize);
2468                 changed |= 2;
2469         }
2470         return changed;
2471 }
2472 
2473 struct ext4_group_info {
2474         unsigned long   bb_state;
2475         struct rb_root  bb_free_root;
2476         ext4_grpblk_t   bb_first_free;  /* first free block */
2477         ext4_grpblk_t   bb_free;        /* total free blocks */
2478         ext4_grpblk_t   bb_fragments;   /* nr of freespace fragments */
2479         ext4_grpblk_t   bb_largest_free_order;/* order of largest frag in BG */
2480         struct          list_head bb_prealloc_list;
2481 #ifdef DOUBLE_CHECK
2482         void            *bb_bitmap;
2483 #endif
2484         struct rw_semaphore alloc_sem;
2485         ext4_grpblk_t   bb_counters[];  /* Nr of free power-of-two-block
2486                                          * regions, index is order.
2487                                          * bb_counters[3] = 5 means
2488                                          * 5 free 8-block regions. */
2489 };
2490 
2491 #define EXT4_GROUP_INFO_NEED_INIT_BIT           0
2492 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT         1
2493 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT     2
2494 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT     3
2495 
2496 #define EXT4_MB_GRP_NEED_INIT(grp)      \
2497         (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2498 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)        \
2499         (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2500 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)        \
2501         (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2502 
2503 #define EXT4_MB_GRP_WAS_TRIMMED(grp)    \
2504         (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2505 #define EXT4_MB_GRP_SET_TRIMMED(grp)    \
2506         (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2507 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)  \
2508         (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2509 
2510 #define EXT4_MAX_CONTENTION             8
2511 #define EXT4_CONTENTION_THRESHOLD       2
2512 
2513 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2514                                               ext4_group_t group)
2515 {
2516         return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2517 }
2518 
2519 /*
2520  * Returns true if the filesystem is busy enough that attempts to
2521  * access the block group locks has run into contention.
2522  */
2523 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2524 {
2525         return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2526 }
2527 
2528 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2529 {
2530         spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2531         if (spin_trylock(lock))
2532                 /*
2533                  * We're able to grab the lock right away, so drop the
2534                  * lock contention counter.
2535                  */
2536                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2537         else {
2538                 /*
2539                  * The lock is busy, so bump the contention counter,
2540                  * and then wait on the spin lock.
2541                  */
2542                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2543                                   EXT4_MAX_CONTENTION);
2544                 spin_lock(lock);
2545         }
2546 }
2547 
2548 static inline void ext4_unlock_group(struct super_block *sb,
2549                                         ext4_group_t group)
2550 {
2551         spin_unlock(ext4_group_lock_ptr(sb, group));
2552 }
2553 
2554 /*
2555  * Block validity checking
2556  */
2557 #define ext4_check_indirect_blockref(inode, bh)                         \
2558         ext4_check_blockref(__func__, __LINE__, inode,                  \
2559                             (__le32 *)(bh)->b_data,                     \
2560                             EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2561 
2562 #define ext4_ind_check_inode(inode)                                     \
2563         ext4_check_blockref(__func__, __LINE__, inode,                  \
2564                             EXT4_I(inode)->i_data,                      \
2565                             EXT4_NDIR_BLOCKS)
2566 
2567 /*
2568  * Inodes and files operations
2569  */
2570 
2571 /* dir.c */
2572 extern const struct file_operations ext4_dir_operations;
2573 
2574 /* file.c */
2575 extern const struct inode_operations ext4_file_inode_operations;
2576 extern const struct file_operations ext4_file_operations;
2577 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2578 
2579 /* inline.c */
2580 extern int ext4_get_max_inline_size(struct inode *inode);
2581 extern int ext4_find_inline_data_nolock(struct inode *inode);
2582 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2583                                  unsigned int len);
2584 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2585 
2586 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2587 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2588                                          struct inode *inode,
2589                                          loff_t pos, unsigned len,
2590                                          unsigned flags,
2591                                          struct page **pagep);
2592 extern int ext4_write_inline_data_end(struct inode *inode,
2593                                       loff_t pos, unsigned len,
2594                                       unsigned copied,
2595                                       struct page *page);
2596 extern struct buffer_head *
2597 ext4_journalled_write_inline_data(struct inode *inode,
2598                                   unsigned len,
2599                                   struct page *page);
2600 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2601                                            struct inode *inode,
2602                                            loff_t pos, unsigned len,
2603                                            unsigned flags,
2604                                            struct page **pagep,
2605                                            void **fsdata);
2606 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2607                                          unsigned len, unsigned copied,
2608                                          struct page *page);
2609 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2610                                      struct inode *inode);
2611 extern int ext4_try_create_inline_dir(handle_t *handle,
2612                                       struct inode *parent,
2613                                       struct inode *inode);
2614 extern int ext4_read_inline_dir(struct file *filp,
2615                                 struct dir_context *ctx,
2616                                 int *has_inline_data);
2617 extern int htree_inlinedir_to_tree(struct file *dir_file,
2618                                    struct inode *dir, ext4_lblk_t block,
2619                                    struct dx_hash_info *hinfo,
2620                                    __u32 start_hash, __u32 start_minor_hash,
2621                                    int *has_inline_data);
2622 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2623                                         const struct qstr *d_name,
2624                                         struct ext4_dir_entry_2 **res_dir,
2625                                         int *has_inline_data);
2626 extern int ext4_delete_inline_entry(handle_t *handle,
2627                                     struct inode *dir,
2628                                     struct ext4_dir_entry_2 *de_del,
2629                                     struct buffer_head *bh,
2630                                     int *has_inline_data);
2631 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2632 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2633                                         struct ext4_dir_entry_2 **parent_de,
2634                                         int *retval);
2635 extern int ext4_inline_data_fiemap(struct inode *inode,
2636                                    struct fiemap_extent_info *fieinfo,
2637                                    int *has_inline);
2638 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2639                                          struct inode *inode,
2640                                          int needed);
2641 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2642 
2643 extern int ext4_convert_inline_data(struct inode *inode);
2644 
2645 static inline int ext4_has_inline_data(struct inode *inode)
2646 {
2647         return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
2648                EXT4_I(inode)->i_inline_off;
2649 }
2650 
2651 /* namei.c */
2652 extern const struct inode_operations ext4_dir_inode_operations;
2653 extern const struct inode_operations ext4_special_inode_operations;
2654 extern struct dentry *ext4_get_parent(struct dentry *child);
2655 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2656                                  struct ext4_dir_entry_2 *de,
2657                                  int blocksize, int csum_size,
2658                                  unsigned int parent_ino, int dotdot_real_len);
2659 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2660                                    unsigned int blocksize);
2661 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2662                                          struct inode *inode,
2663                                          struct buffer_head *bh);
2664 #define S_SHIFT 12
2665 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2666         [S_IFREG >> S_SHIFT]    = EXT4_FT_REG_FILE,
2667         [S_IFDIR >> S_SHIFT]    = EXT4_FT_DIR,
2668         [S_IFCHR >> S_SHIFT]    = EXT4_FT_CHRDEV,
2669         [S_IFBLK >> S_SHIFT]    = EXT4_FT_BLKDEV,
2670         [S_IFIFO >> S_SHIFT]    = EXT4_FT_FIFO,
2671         [S_IFSOCK >> S_SHIFT]   = EXT4_FT_SOCK,
2672         [S_IFLNK >> S_SHIFT]    = EXT4_FT_SYMLINK,
2673 };
2674 
2675 static inline void ext4_set_de_type(struct super_block *sb,
2676                                 struct ext4_dir_entry_2 *de,
2677                                 umode_t mode) {
2678         if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2679                 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2680 }
2681 
2682 
2683 /* symlink.c */
2684 extern const struct inode_operations ext4_symlink_inode_operations;
2685 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2686 
2687 /* block_validity */
2688 extern void ext4_release_system_zone(struct super_block *sb);
2689 extern int ext4_setup_system_zone(struct super_block *sb);
2690 extern int __init ext4_init_system_zone(void);
2691 extern void ext4_exit_system_zone(void);
2692 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2693                                  ext4_fsblk_t start_blk,
2694                                  unsigned int count);
2695 extern int ext4_check_blockref(const char *, unsigned int,
2696                                struct inode *, __le32 *, unsigned int);
2697 
2698 /* extents.c */
2699 struct ext4_ext_path;
2700 struct ext4_extent;
2701 
2702 /*
2703  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
2704  * __le32.
2705  */
2706 #define EXT_MAX_BLOCKS  0xffffffff
2707 
2708 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2709 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2710 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
2711 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2712                                struct ext4_map_blocks *map, int flags);
2713 extern void ext4_ext_truncate(handle_t *, struct inode *);
2714 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2715                                  ext4_lblk_t end);
2716 extern void ext4_ext_init(struct super_block *);
2717 extern void ext4_ext_release(struct super_block *);
2718 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2719                           loff_t len);
2720 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
2721                                           loff_t offset, ssize_t len);
2722 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2723                            struct ext4_map_blocks *map, int flags);
2724 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2725                                          ext4_lblk_t lblocks);
2726 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2727 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2728                                                    int num,
2729                                                    struct ext4_ext_path *path);
2730 extern int ext4_can_extents_be_merged(struct inode *inode,
2731                                       struct ext4_extent *ex1,
2732                                       struct ext4_extent *ex2);
2733 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2734                                   struct ext4_ext_path *,
2735                                   struct ext4_extent *, int);
2736 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2737                                                   struct ext4_ext_path *,
2738                                                   int flags);
2739 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2740 extern int ext4_ext_check_inode(struct inode *inode);
2741 extern int ext4_find_delalloc_range(struct inode *inode,
2742                                     ext4_lblk_t lblk_start,
2743                                     ext4_lblk_t lblk_end);
2744 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2745 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2746                         __u64 start, __u64 len);
2747 extern int ext4_ext_precache(struct inode *inode);
2748 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
2749 
2750 /* move_extent.c */
2751 extern void ext4_double_down_write_data_sem(struct inode *first,
2752                                             struct inode *second);
2753 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2754                                           struct inode *donor_inode);
2755 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2756                              __u64 start_orig, __u64 start_donor,
2757                              __u64 len, __u64 *moved_len);
2758 extern int mext_next_extent(struct inode *inode, struct ext4_ext_path *path,
2759                             struct ext4_extent **extent);
2760 
2761 /* page-io.c */
2762 extern int __init ext4_init_pageio(void);
2763 extern void ext4_exit_pageio(void);
2764 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2765 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2766 extern int ext4_put_io_end(ext4_io_end_t *io_end);
2767 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2768 extern void ext4_io_submit_init(struct ext4_io_submit *io,
2769                                 struct writeback_control *wbc);
2770 extern void ext4_end_io_rsv_work(struct work_struct *work);
2771 extern void ext4_io_submit(struct ext4_io_submit *io);
2772 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2773                                struct page *page,
2774                                int len,
2775                                struct writeback_control *wbc,
2776                                bool keep_towrite);
2777 
2778 /* mmp.c */
2779 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2780 
2781 /*
2782  * Note that these flags will never ever appear in a buffer_head's state flag.
2783  * See EXT4_MAP_... to see where this is used.
2784  */
2785 enum ext4_state_bits {
2786         BH_AllocFromCluster     /* allocated blocks were part of already
2787                                  * allocated cluster. */
2788         = BH_JBDPrivateStart
2789 };
2790 
2791 /*
2792  * Add new method to test whether block and inode bitmaps are properly
2793  * initialized. With uninit_bg reading the block from disk is not enough
2794  * to mark the bitmap uptodate. We need to also zero-out the bitmap
2795  */
2796 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2797 
2798 static inline int bitmap_uptodate(struct buffer_head *bh)
2799 {
2800         return (buffer_uptodate(bh) &&
2801                         test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2802 }
2803 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2804 {
2805         set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2806 }
2807 
2808 /*
2809  * Disable DIO read nolock optimization, so new dioreaders will be forced
2810  * to grab i_mutex
2811  */
2812 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2813 {
2814         ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2815         smp_mb();
2816 }
2817 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2818 {
2819         smp_mb();
2820         ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2821 }
2822 
2823 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2824 
2825 /* For ioend & aio unwritten conversion wait queues */
2826 #define EXT4_WQ_HASH_SZ         37
2827 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
2828                                             EXT4_WQ_HASH_SZ])
2829 #define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
2830                                              EXT4_WQ_HASH_SZ])
2831 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2832 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2833 
2834 #define EXT4_RESIZING   0
2835 extern int ext4_resize_begin(struct super_block *sb);
2836 extern void ext4_resize_end(struct super_block *sb);
2837 
2838 #endif  /* __KERNEL__ */
2839 
2840 #endif  /* _EXT4_H */
2841 

This page was automatically generated by LXR 0.3.1 (source).  •  Linux is a registered trademark of Linus Torvalds  •  Contact us