1 /*
2  * Written by Stephen C. Tweedie <sct@redhat.com>, 1999
3  *
4  * Copyright 1998--1999 Red Hat corp --- All Rights Reserved
5  *
6  * This file is part of the Linux kernel and is made available under
7  * the terms of the GNU General Public License, version 2, or at your
8  * option, any later version, incorporated herein by reference.
9  *
10  * Copyright (C) 1992, 1993, 1994, 1995
11  * Remy Card (card@masi.ibp.fr)
12  * Laboratoire MASI - Institut Blaise Pascal
13  * Universite Pierre et Marie Curie (Paris VI)
14  *
15  *  from
16  *
17  *  linux/include/linux/minix_fs.h
18  *
19  *  Copyright (C) 1991, 1992  Linus Torvalds
20  */
21 
22 #include <linux/fs.h>
23 #include <linux/jbd.h>
24 #include <linux/magic.h>
25 #include <linux/bug.h>
26 #include <linux/blockgroup_lock.h>
27 
28 /*
29  * The second extended filesystem constants/structures
30  */
31 
32 /*
33  * Define EXT3FS_DEBUG to produce debug messages
34  */
35 #undef EXT3FS_DEBUG
36 
37 /*
38  * Define EXT3_RESERVATION to reserve data blocks for expanding files
39  */
40 #define EXT3_DEFAULT_RESERVE_BLOCKS     8
41 /*max window size: 1024(direct blocks) + 3([t,d]indirect blocks) */
42 #define EXT3_MAX_RESERVE_BLOCKS         1027
43 #define EXT3_RESERVE_WINDOW_NOT_ALLOCATED 0
44 
45 /*
46  * Debug code
47  */
48 #ifdef EXT3FS_DEBUG
49 #define ext3_debug(f, a...)						\
50 	do {								\
51 		printk (KERN_DEBUG "EXT3-fs DEBUG (%s, %d): %s:",	\
52 			__FILE__, __LINE__, __func__);		\
53 		printk (KERN_DEBUG f, ## a);				\
54 	} while (0)
55 #else
56 #define ext3_debug(f, a...)	do {} while (0)
57 #endif
58 
59 /*
60  * Special inodes numbers
61  */
62 #define	EXT3_BAD_INO		 1	/* Bad blocks inode */
63 #define EXT3_ROOT_INO		 2	/* Root inode */
64 #define EXT3_BOOT_LOADER_INO	 5	/* Boot loader inode */
65 #define EXT3_UNDEL_DIR_INO	 6	/* Undelete directory inode */
66 #define EXT3_RESIZE_INO		 7	/* Reserved group descriptors inode */
67 #define EXT3_JOURNAL_INO	 8	/* Journal inode */
68 
69 /* First non-reserved inode for old ext3 filesystems */
70 #define EXT3_GOOD_OLD_FIRST_INO	11
71 
72 /*
73  * Maximal count of links to a file
74  */
75 #define EXT3_LINK_MAX		32000
76 
77 /*
78  * Macro-instructions used to manage several block sizes
79  */
80 #define EXT3_MIN_BLOCK_SIZE		1024
81 #define	EXT3_MAX_BLOCK_SIZE		65536
82 #define EXT3_MIN_BLOCK_LOG_SIZE		10
83 #define EXT3_BLOCK_SIZE(s)		((s)->s_blocksize)
84 #define	EXT3_ADDR_PER_BLOCK(s)		(EXT3_BLOCK_SIZE(s) / sizeof (__u32))
85 #define EXT3_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
86 #define	EXT3_ADDR_PER_BLOCK_BITS(s)	(EXT3_SB(s)->s_addr_per_block_bits)
87 #define EXT3_INODE_SIZE(s)		(EXT3_SB(s)->s_inode_size)
88 #define EXT3_FIRST_INO(s)		(EXT3_SB(s)->s_first_ino)
89 
90 /*
91  * Macro-instructions used to manage fragments
92  */
93 #define EXT3_MIN_FRAG_SIZE		1024
94 #define	EXT3_MAX_FRAG_SIZE		4096
95 #define EXT3_MIN_FRAG_LOG_SIZE		  10
96 #define EXT3_FRAG_SIZE(s)		(EXT3_SB(s)->s_frag_size)
97 #define EXT3_FRAGS_PER_BLOCK(s)		(EXT3_SB(s)->s_frags_per_block)
98 
99 /*
100  * Structure of a blocks group descriptor
101  */
102 struct ext3_group_desc
103 {
104 	__le32	bg_block_bitmap;		/* Blocks bitmap block */
105 	__le32	bg_inode_bitmap;		/* Inodes bitmap block */
106 	__le32	bg_inode_table;		/* Inodes table block */
107 	__le16	bg_free_blocks_count;	/* Free blocks count */
108 	__le16	bg_free_inodes_count;	/* Free inodes count */
109 	__le16	bg_used_dirs_count;	/* Directories count */
110 	__u16	bg_pad;
111 	__le32	bg_reserved[3];
112 };
113 
114 /*
115  * Macro-instructions used to manage group descriptors
116  */
117 #define EXT3_BLOCKS_PER_GROUP(s)	(EXT3_SB(s)->s_blocks_per_group)
118 #define EXT3_DESC_PER_BLOCK(s)		(EXT3_SB(s)->s_desc_per_block)
119 #define EXT3_INODES_PER_GROUP(s)	(EXT3_SB(s)->s_inodes_per_group)
120 #define EXT3_DESC_PER_BLOCK_BITS(s)	(EXT3_SB(s)->s_desc_per_block_bits)
121 
122 /*
123  * Constants relative to the data blocks
124  */
125 #define	EXT3_NDIR_BLOCKS		12
126 #define	EXT3_IND_BLOCK			EXT3_NDIR_BLOCKS
127 #define	EXT3_DIND_BLOCK			(EXT3_IND_BLOCK + 1)
128 #define	EXT3_TIND_BLOCK			(EXT3_DIND_BLOCK + 1)
129 #define	EXT3_N_BLOCKS			(EXT3_TIND_BLOCK + 1)
130 
131 /*
132  * Inode flags
133  */
134 #define	EXT3_SECRM_FL			0x00000001 /* Secure deletion */
135 #define	EXT3_UNRM_FL			0x00000002 /* Undelete */
136 #define	EXT3_COMPR_FL			0x00000004 /* Compress file */
137 #define EXT3_SYNC_FL			0x00000008 /* Synchronous updates */
138 #define EXT3_IMMUTABLE_FL		0x00000010 /* Immutable file */
139 #define EXT3_APPEND_FL			0x00000020 /* writes to file may only append */
140 #define EXT3_NODUMP_FL			0x00000040 /* do not dump file */
141 #define EXT3_NOATIME_FL			0x00000080 /* do not update atime */
142 /* Reserved for compression usage... */
143 #define EXT3_DIRTY_FL			0x00000100
144 #define EXT3_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
145 #define EXT3_NOCOMPR_FL			0x00000400 /* Don't compress */
146 #define EXT3_ECOMPR_FL			0x00000800 /* Compression error */
147 /* End compression flags --- maybe not all used */
148 #define EXT3_INDEX_FL			0x00001000 /* hash-indexed directory */
149 #define EXT3_IMAGIC_FL			0x00002000 /* AFS directory */
150 #define EXT3_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
151 #define EXT3_NOTAIL_FL			0x00008000 /* file tail should not be merged */
152 #define EXT3_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
153 #define EXT3_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
154 #define EXT3_RESERVED_FL		0x80000000 /* reserved for ext3 lib */
155 
156 #define EXT3_FL_USER_VISIBLE		0x0003DFFF /* User visible flags */
157 #define EXT3_FL_USER_MODIFIABLE		0x000380FF /* User modifiable flags */
158 
159 /* Flags that should be inherited by new inodes from their parent. */
160 #define EXT3_FL_INHERITED (EXT3_SECRM_FL | EXT3_UNRM_FL | EXT3_COMPR_FL |\
161 			   EXT3_SYNC_FL | EXT3_NODUMP_FL |\
162 			   EXT3_NOATIME_FL | EXT3_COMPRBLK_FL |\
163 			   EXT3_NOCOMPR_FL | EXT3_JOURNAL_DATA_FL |\
164 			   EXT3_NOTAIL_FL | EXT3_DIRSYNC_FL)
165 
166 /* Flags that are appropriate for regular files (all but dir-specific ones). */
167 #define EXT3_REG_FLMASK (~(EXT3_DIRSYNC_FL | EXT3_TOPDIR_FL))
168 
169 /* Flags that are appropriate for non-directories/regular files. */
170 #define EXT3_OTHER_FLMASK (EXT3_NODUMP_FL | EXT3_NOATIME_FL)
171 
172 /* Mask out flags that are inappropriate for the given type of inode. */
ext3_mask_flags(umode_t mode,__u32 flags)173 static inline __u32 ext3_mask_flags(umode_t mode, __u32 flags)
174 {
175 	if (S_ISDIR(mode))
176 		return flags;
177 	else if (S_ISREG(mode))
178 		return flags & EXT3_REG_FLMASK;
179 	else
180 		return flags & EXT3_OTHER_FLMASK;
181 }
182 
183 /* Used to pass group descriptor data when online resize is done */
184 struct ext3_new_group_input {
185 	__u32 group;            /* Group number for this data */
186 	__u32 block_bitmap;     /* Absolute block number of block bitmap */
187 	__u32 inode_bitmap;     /* Absolute block number of inode bitmap */
188 	__u32 inode_table;      /* Absolute block number of inode table start */
189 	__u32 blocks_count;     /* Total number of blocks in this group */
190 	__u16 reserved_blocks;  /* Number of reserved blocks in this group */
191 	__u16 unused;
192 };
193 
194 /* The struct ext3_new_group_input in kernel space, with free_blocks_count */
195 struct ext3_new_group_data {
196 	__u32 group;
197 	__u32 block_bitmap;
198 	__u32 inode_bitmap;
199 	__u32 inode_table;
200 	__u32 blocks_count;
201 	__u16 reserved_blocks;
202 	__u16 unused;
203 	__u32 free_blocks_count;
204 };
205 
206 
207 /*
208  * ioctl commands
209  */
210 #define	EXT3_IOC_GETFLAGS		FS_IOC_GETFLAGS
211 #define	EXT3_IOC_SETFLAGS		FS_IOC_SETFLAGS
212 #define	EXT3_IOC_GETVERSION		_IOR('f', 3, long)
213 #define	EXT3_IOC_SETVERSION		_IOW('f', 4, long)
214 #define EXT3_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
215 #define EXT3_IOC_GROUP_ADD		_IOW('f', 8,struct ext3_new_group_input)
216 #define	EXT3_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
217 #define	EXT3_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
218 #ifdef CONFIG_JBD_DEBUG
219 #define EXT3_IOC_WAIT_FOR_READONLY	_IOR('f', 99, long)
220 #endif
221 #define EXT3_IOC_GETRSVSZ		_IOR('f', 5, long)
222 #define EXT3_IOC_SETRSVSZ		_IOW('f', 6, long)
223 
224 /*
225  * ioctl commands in 32 bit emulation
226  */
227 #define EXT3_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
228 #define EXT3_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
229 #define EXT3_IOC32_GETVERSION		_IOR('f', 3, int)
230 #define EXT3_IOC32_SETVERSION		_IOW('f', 4, int)
231 #define EXT3_IOC32_GETRSVSZ		_IOR('f', 5, int)
232 #define EXT3_IOC32_SETRSVSZ		_IOW('f', 6, int)
233 #define EXT3_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
234 #ifdef CONFIG_JBD_DEBUG
235 #define EXT3_IOC32_WAIT_FOR_READONLY	_IOR('f', 99, int)
236 #endif
237 #define EXT3_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
238 #define EXT3_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
239 
240 
241 /*
242  *  Mount options
243  */
244 struct ext3_mount_options {
245 	unsigned long s_mount_opt;
246 	uid_t s_resuid;
247 	gid_t s_resgid;
248 	unsigned long s_commit_interval;
249 #ifdef CONFIG_QUOTA
250 	int s_jquota_fmt;
251 	char *s_qf_names[MAXQUOTAS];
252 #endif
253 };
254 
255 /*
256  * Structure of an inode on the disk
257  */
258 struct ext3_inode {
259 	__le16	i_mode;		/* File mode */
260 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
261 	__le32	i_size;		/* Size in bytes */
262 	__le32	i_atime;	/* Access time */
263 	__le32	i_ctime;	/* Creation time */
264 	__le32	i_mtime;	/* Modification time */
265 	__le32	i_dtime;	/* Deletion Time */
266 	__le16	i_gid;		/* Low 16 bits of Group Id */
267 	__le16	i_links_count;	/* Links count */
268 	__le32	i_blocks;	/* Blocks count */
269 	__le32	i_flags;	/* File flags */
270 	union {
271 		struct {
272 			__u32  l_i_reserved1;
273 		} linux1;
274 		struct {
275 			__u32  h_i_translator;
276 		} hurd1;
277 		struct {
278 			__u32  m_i_reserved1;
279 		} masix1;
280 	} osd1;				/* OS dependent 1 */
281 	__le32	i_block[EXT3_N_BLOCKS];/* Pointers to blocks */
282 	__le32	i_generation;	/* File version (for NFS) */
283 	__le32	i_file_acl;	/* File ACL */
284 	__le32	i_dir_acl;	/* Directory ACL */
285 	__le32	i_faddr;	/* Fragment address */
286 	union {
287 		struct {
288 			__u8	l_i_frag;	/* Fragment number */
289 			__u8	l_i_fsize;	/* Fragment size */
290 			__u16	i_pad1;
291 			__le16	l_i_uid_high;	/* these 2 fields    */
292 			__le16	l_i_gid_high;	/* were reserved2[0] */
293 			__u32	l_i_reserved2;
294 		} linux2;
295 		struct {
296 			__u8	h_i_frag;	/* Fragment number */
297 			__u8	h_i_fsize;	/* Fragment size */
298 			__u16	h_i_mode_high;
299 			__u16	h_i_uid_high;
300 			__u16	h_i_gid_high;
301 			__u32	h_i_author;
302 		} hurd2;
303 		struct {
304 			__u8	m_i_frag;	/* Fragment number */
305 			__u8	m_i_fsize;	/* Fragment size */
306 			__u16	m_pad1;
307 			__u32	m_i_reserved2[2];
308 		} masix2;
309 	} osd2;				/* OS dependent 2 */
310 	__le16	i_extra_isize;
311 	__le16	i_pad1;
312 };
313 
314 #define i_size_high	i_dir_acl
315 
316 #define i_reserved1	osd1.linux1.l_i_reserved1
317 #define i_frag		osd2.linux2.l_i_frag
318 #define i_fsize		osd2.linux2.l_i_fsize
319 #define i_uid_low	i_uid
320 #define i_gid_low	i_gid
321 #define i_uid_high	osd2.linux2.l_i_uid_high
322 #define i_gid_high	osd2.linux2.l_i_gid_high
323 #define i_reserved2	osd2.linux2.l_i_reserved2
324 
325 /*
326  * File system states
327  */
328 #define	EXT3_VALID_FS			0x0001	/* Unmounted cleanly */
329 #define	EXT3_ERROR_FS			0x0002	/* Errors detected */
330 #define	EXT3_ORPHAN_FS			0x0004	/* Orphans being recovered */
331 
332 /*
333  * Misc. filesystem flags
334  */
335 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
336 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
337 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
338 
339 /*
340  * Mount flags
341  */
342 #define EXT3_MOUNT_CHECK		0x00001	/* Do mount-time checks */
343 /* EXT3_MOUNT_OLDALLOC was there */
344 #define EXT3_MOUNT_GRPID		0x00004	/* Create files with directory's group */
345 #define EXT3_MOUNT_DEBUG		0x00008	/* Some debugging messages */
346 #define EXT3_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
347 #define EXT3_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
348 #define EXT3_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
349 #define EXT3_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
350 #define EXT3_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
351 #define EXT3_MOUNT_ABORT		0x00200	/* Fatal error detected */
352 #define EXT3_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
353 #define EXT3_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
354 #define EXT3_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
355 #define EXT3_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
356 #define EXT3_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
357 #define EXT3_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
358 #define EXT3_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
359 #define EXT3_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
360 #define EXT3_MOUNT_RESERVATION		0x10000	/* Preallocation */
361 #define EXT3_MOUNT_BARRIER		0x20000 /* Use block barriers */
362 #define EXT3_MOUNT_QUOTA		0x80000 /* Some quota option set */
363 #define EXT3_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
364 #define EXT3_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
365 #define EXT3_MOUNT_DATA_ERR_ABORT	0x400000 /* Abort on file data write
366 						  * error in ordered mode */
367 
368 /* Compatibility, for having both ext2_fs.h and ext3_fs.h included at once */
369 #ifndef _LINUX_EXT2_FS_H
370 #define clear_opt(o, opt)		o &= ~EXT3_MOUNT_##opt
371 #define set_opt(o, opt)			o |= EXT3_MOUNT_##opt
372 #define test_opt(sb, opt)		(EXT3_SB(sb)->s_mount_opt & \
373 					 EXT3_MOUNT_##opt)
374 #else
375 #define EXT2_MOUNT_NOLOAD		EXT3_MOUNT_NOLOAD
376 #define EXT2_MOUNT_ABORT		EXT3_MOUNT_ABORT
377 #define EXT2_MOUNT_DATA_FLAGS		EXT3_MOUNT_DATA_FLAGS
378 #endif
379 
380 #define ext3_set_bit			__set_bit_le
381 #define ext3_set_bit_atomic		ext2_set_bit_atomic
382 #define ext3_clear_bit			__clear_bit_le
383 #define ext3_clear_bit_atomic		ext2_clear_bit_atomic
384 #define ext3_test_bit			test_bit_le
385 #define ext3_find_next_zero_bit		find_next_zero_bit_le
386 
387 /*
388  * Maximal mount counts between two filesystem checks
389  */
390 #define EXT3_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
391 #define EXT3_DFL_CHECKINTERVAL		0	/* Don't use interval check */
392 
393 /*
394  * Behaviour when detecting errors
395  */
396 #define EXT3_ERRORS_CONTINUE		1	/* Continue execution */
397 #define EXT3_ERRORS_RO			2	/* Remount fs read-only */
398 #define EXT3_ERRORS_PANIC		3	/* Panic */
399 #define EXT3_ERRORS_DEFAULT		EXT3_ERRORS_CONTINUE
400 
401 /*
402  * Structure of the super block
403  */
404 struct ext3_super_block {
405 /*00*/	__le32	s_inodes_count;		/* Inodes count */
406 	__le32	s_blocks_count;		/* Blocks count */
407 	__le32	s_r_blocks_count;	/* Reserved blocks count */
408 	__le32	s_free_blocks_count;	/* Free blocks count */
409 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
410 	__le32	s_first_data_block;	/* First Data Block */
411 	__le32	s_log_block_size;	/* Block size */
412 	__le32	s_log_frag_size;	/* Fragment size */
413 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
414 	__le32	s_frags_per_group;	/* # Fragments per group */
415 	__le32	s_inodes_per_group;	/* # Inodes per group */
416 	__le32	s_mtime;		/* Mount time */
417 /*30*/	__le32	s_wtime;		/* Write time */
418 	__le16	s_mnt_count;		/* Mount count */
419 	__le16	s_max_mnt_count;	/* Maximal mount count */
420 	__le16	s_magic;		/* Magic signature */
421 	__le16	s_state;		/* File system state */
422 	__le16	s_errors;		/* Behaviour when detecting errors */
423 	__le16	s_minor_rev_level;	/* minor revision level */
424 /*40*/	__le32	s_lastcheck;		/* time of last check */
425 	__le32	s_checkinterval;	/* max. time between checks */
426 	__le32	s_creator_os;		/* OS */
427 	__le32	s_rev_level;		/* Revision level */
428 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
429 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
430 	/*
431 	 * These fields are for EXT3_DYNAMIC_REV superblocks only.
432 	 *
433 	 * Note: the difference between the compatible feature set and
434 	 * the incompatible feature set is that if there is a bit set
435 	 * in the incompatible feature set that the kernel doesn't
436 	 * know about, it should refuse to mount the filesystem.
437 	 *
438 	 * e2fsck's requirements are more strict; if it doesn't know
439 	 * about a feature in either the compatible or incompatible
440 	 * feature set, it must abort and not try to meddle with
441 	 * things it doesn't understand...
442 	 */
443 	__le32	s_first_ino;		/* First non-reserved inode */
444 	__le16   s_inode_size;		/* size of inode structure */
445 	__le16	s_block_group_nr;	/* block group # of this superblock */
446 	__le32	s_feature_compat;	/* compatible feature set */
447 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
448 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
449 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
450 /*78*/	char	s_volume_name[16];	/* volume name */
451 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
452 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
453 	/*
454 	 * Performance hints.  Directory preallocation should only
455 	 * happen if the EXT3_FEATURE_COMPAT_DIR_PREALLOC flag is on.
456 	 */
457 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
458 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
459 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
460 	/*
461 	 * Journaling support valid if EXT3_FEATURE_COMPAT_HAS_JOURNAL set.
462 	 */
463 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
464 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
465 	__le32	s_journal_dev;		/* device number of journal file */
466 	__le32	s_last_orphan;		/* start of list of inodes to delete */
467 	__le32	s_hash_seed[4];		/* HTREE hash seed */
468 	__u8	s_def_hash_version;	/* Default hash version to use */
469 	__u8	s_reserved_char_pad;
470 	__u16	s_reserved_word_pad;
471 	__le32	s_default_mount_opts;
472 	__le32	s_first_meta_bg;	/* First metablock block group */
473 	__le32	s_mkfs_time;		/* When the filesystem was created */
474 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
475 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
476 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
477 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
478 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
479 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
480 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
481 	__le32	s_flags;		/* Miscellaneous flags */
482 	__le16  s_raid_stride;		/* RAID stride */
483 	__le16  s_mmp_interval;         /* # seconds to wait in MMP checking */
484 	__le64  s_mmp_block;            /* Block for multi-mount protection */
485 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
486 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
487 	__u8	s_reserved_char_pad2;
488 	__le16  s_reserved_pad;
489 	__u32   s_reserved[162];        /* Padding to the end of the block */
490 };
491 
492 /* data type for block offset of block group */
493 typedef int ext3_grpblk_t;
494 
495 /* data type for filesystem-wide blocks number */
496 typedef unsigned long ext3_fsblk_t;
497 
498 #define E3FSBLK "%lu"
499 
500 struct ext3_reserve_window {
501 	ext3_fsblk_t	_rsv_start;	/* First byte reserved */
502 	ext3_fsblk_t	_rsv_end;	/* Last byte reserved or 0 */
503 };
504 
505 struct ext3_reserve_window_node {
506 	struct rb_node		rsv_node;
507 	__u32			rsv_goal_size;
508 	__u32			rsv_alloc_hit;
509 	struct ext3_reserve_window	rsv_window;
510 };
511 
512 struct ext3_block_alloc_info {
513 	/* information about reservation window */
514 	struct ext3_reserve_window_node	rsv_window_node;
515 	/*
516 	 * was i_next_alloc_block in ext3_inode_info
517 	 * is the logical (file-relative) number of the
518 	 * most-recently-allocated block in this file.
519 	 * We use this for detecting linearly ascending allocation requests.
520 	 */
521 	__u32                   last_alloc_logical_block;
522 	/*
523 	 * Was i_next_alloc_goal in ext3_inode_info
524 	 * is the *physical* companion to i_next_alloc_block.
525 	 * it the physical block number of the block which was most-recentl
526 	 * allocated to this file.  This give us the goal (target) for the next
527 	 * allocation when we detect linearly ascending requests.
528 	 */
529 	ext3_fsblk_t		last_alloc_physical_block;
530 };
531 
532 #define rsv_start rsv_window._rsv_start
533 #define rsv_end rsv_window._rsv_end
534 
535 /*
536  * third extended file system inode data in memory
537  */
538 struct ext3_inode_info {
539 	__le32	i_data[15];	/* unconverted */
540 	__u32	i_flags;
541 #ifdef EXT3_FRAGMENTS
542 	__u32	i_faddr;
543 	__u8	i_frag_no;
544 	__u8	i_frag_size;
545 #endif
546 	ext3_fsblk_t	i_file_acl;
547 	__u32	i_dir_acl;
548 	__u32	i_dtime;
549 
550 	/*
551 	 * i_block_group is the number of the block group which contains
552 	 * this file's inode.  Constant across the lifetime of the inode,
553 	 * it is ued for making block allocation decisions - we try to
554 	 * place a file's data blocks near its inode block, and new inodes
555 	 * near to their parent directory's inode.
556 	 */
557 	__u32	i_block_group;
558 	unsigned long	i_state_flags;	/* Dynamic state flags for ext3 */
559 
560 	/* block reservation info */
561 	struct ext3_block_alloc_info *i_block_alloc_info;
562 
563 	__u32	i_dir_start_lookup;
564 #ifdef CONFIG_EXT3_FS_XATTR
565 	/*
566 	 * Extended attributes can be read independently of the main file
567 	 * data. Taking i_mutex even when reading would cause contention
568 	 * between readers of EAs and writers of regular file data, so
569 	 * instead we synchronize on xattr_sem when reading or changing
570 	 * EAs.
571 	 */
572 	struct rw_semaphore xattr_sem;
573 #endif
574 
575 	struct list_head i_orphan;	/* unlinked but open inodes */
576 
577 	/*
578 	 * i_disksize keeps track of what the inode size is ON DISK, not
579 	 * in memory.  During truncate, i_size is set to the new size by
580 	 * the VFS prior to calling ext3_truncate(), but the filesystem won't
581 	 * set i_disksize to 0 until the truncate is actually under way.
582 	 *
583 	 * The intent is that i_disksize always represents the blocks which
584 	 * are used by this file.  This allows recovery to restart truncate
585 	 * on orphans if we crash during truncate.  We actually write i_disksize
586 	 * into the on-disk inode when writing inodes out, instead of i_size.
587 	 *
588 	 * The only time when i_disksize and i_size may be different is when
589 	 * a truncate is in progress.  The only things which change i_disksize
590 	 * are ext3_get_block (growth) and ext3_truncate (shrinkth).
591 	 */
592 	loff_t	i_disksize;
593 
594 	/* on-disk additional length */
595 	__u16 i_extra_isize;
596 
597 	/*
598 	 * truncate_mutex is for serialising ext3_truncate() against
599 	 * ext3_getblock().  In the 2.4 ext2 design, great chunks of inode's
600 	 * data tree are chopped off during truncate. We can't do that in
601 	 * ext3 because whenever we perform intermediate commits during
602 	 * truncate, the inode and all the metadata blocks *must* be in a
603 	 * consistent state which allows truncation of the orphans to restart
604 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
605 	 * by other means, so we have truncate_mutex.
606 	 */
607 	struct mutex truncate_mutex;
608 
609 	/*
610 	 * Transactions that contain inode's metadata needed to complete
611 	 * fsync and fdatasync, respectively.
612 	 */
613 	atomic_t i_sync_tid;
614 	atomic_t i_datasync_tid;
615 
616 	struct inode vfs_inode;
617 };
618 
619 /*
620  * third extended-fs super-block data in memory
621  */
622 struct ext3_sb_info {
623 	unsigned long s_frag_size;	/* Size of a fragment in bytes */
624 	unsigned long s_frags_per_block;/* Number of fragments per block */
625 	unsigned long s_inodes_per_block;/* Number of inodes per block */
626 	unsigned long s_frags_per_group;/* Number of fragments in a group */
627 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
628 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
629 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
630 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
631 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
632 	unsigned long s_groups_count;	/* Number of groups in the fs */
633 	unsigned long s_overhead_last;  /* Last calculated overhead */
634 	unsigned long s_blocks_last;    /* Last seen block count */
635 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
636 	struct ext3_super_block * s_es;	/* Pointer to the super block in the buffer */
637 	struct buffer_head ** s_group_desc;
638 	unsigned long  s_mount_opt;
639 	ext3_fsblk_t s_sb_block;
640 	uid_t s_resuid;
641 	gid_t s_resgid;
642 	unsigned short s_mount_state;
643 	unsigned short s_pad;
644 	int s_addr_per_block_bits;
645 	int s_desc_per_block_bits;
646 	int s_inode_size;
647 	int s_first_ino;
648 	spinlock_t s_next_gen_lock;
649 	u32 s_next_generation;
650 	u32 s_hash_seed[4];
651 	int s_def_hash_version;
652 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
653 	struct percpu_counter s_freeblocks_counter;
654 	struct percpu_counter s_freeinodes_counter;
655 	struct percpu_counter s_dirs_counter;
656 	struct blockgroup_lock *s_blockgroup_lock;
657 
658 	/* root of the per fs reservation window tree */
659 	spinlock_t s_rsv_window_lock;
660 	struct rb_root s_rsv_window_root;
661 	struct ext3_reserve_window_node s_rsv_window_head;
662 
663 	/* Journaling */
664 	struct inode * s_journal_inode;
665 	struct journal_s * s_journal;
666 	struct list_head s_orphan;
667 	struct mutex s_orphan_lock;
668 	struct mutex s_resize_lock;
669 	unsigned long s_commit_interval;
670 	struct block_device *journal_bdev;
671 #ifdef CONFIG_QUOTA
672 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
673 	int s_jquota_fmt;			/* Format of quota to use */
674 #endif
675 };
676 
677 static inline spinlock_t *
sb_bgl_lock(struct ext3_sb_info * sbi,unsigned int block_group)678 sb_bgl_lock(struct ext3_sb_info *sbi, unsigned int block_group)
679 {
680 	return bgl_lock_ptr(sbi->s_blockgroup_lock, block_group);
681 }
682 
EXT3_SB(struct super_block * sb)683 static inline struct ext3_sb_info * EXT3_SB(struct super_block *sb)
684 {
685 	return sb->s_fs_info;
686 }
EXT3_I(struct inode * inode)687 static inline struct ext3_inode_info *EXT3_I(struct inode *inode)
688 {
689 	return container_of(inode, struct ext3_inode_info, vfs_inode);
690 }
691 
ext3_valid_inum(struct super_block * sb,unsigned long ino)692 static inline int ext3_valid_inum(struct super_block *sb, unsigned long ino)
693 {
694 	return ino == EXT3_ROOT_INO ||
695 		ino == EXT3_JOURNAL_INO ||
696 		ino == EXT3_RESIZE_INO ||
697 		(ino >= EXT3_FIRST_INO(sb) &&
698 		 ino <= le32_to_cpu(EXT3_SB(sb)->s_es->s_inodes_count));
699 }
700 
701 /*
702  * Inode dynamic state flags
703  */
704 enum {
705 	EXT3_STATE_JDATA,		/* journaled data exists */
706 	EXT3_STATE_NEW,			/* inode is newly created */
707 	EXT3_STATE_XATTR,		/* has in-inode xattrs */
708 	EXT3_STATE_FLUSH_ON_CLOSE,	/* flush dirty pages on close */
709 };
710 
ext3_test_inode_state(struct inode * inode,int bit)711 static inline int ext3_test_inode_state(struct inode *inode, int bit)
712 {
713 	return test_bit(bit, &EXT3_I(inode)->i_state_flags);
714 }
715 
ext3_set_inode_state(struct inode * inode,int bit)716 static inline void ext3_set_inode_state(struct inode *inode, int bit)
717 {
718 	set_bit(bit, &EXT3_I(inode)->i_state_flags);
719 }
720 
ext3_clear_inode_state(struct inode * inode,int bit)721 static inline void ext3_clear_inode_state(struct inode *inode, int bit)
722 {
723 	clear_bit(bit, &EXT3_I(inode)->i_state_flags);
724 }
725 
726 #define NEXT_ORPHAN(inode) EXT3_I(inode)->i_dtime
727 
728 /*
729  * Codes for operating systems
730  */
731 #define EXT3_OS_LINUX		0
732 #define EXT3_OS_HURD		1
733 #define EXT3_OS_MASIX		2
734 #define EXT3_OS_FREEBSD		3
735 #define EXT3_OS_LITES		4
736 
737 /*
738  * Revision levels
739  */
740 #define EXT3_GOOD_OLD_REV	0	/* The good old (original) format */
741 #define EXT3_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
742 
743 #define EXT3_CURRENT_REV	EXT3_GOOD_OLD_REV
744 #define EXT3_MAX_SUPP_REV	EXT3_DYNAMIC_REV
745 
746 #define EXT3_GOOD_OLD_INODE_SIZE 128
747 
748 /*
749  * Feature set definitions
750  */
751 
752 #define EXT3_HAS_COMPAT_FEATURE(sb,mask)			\
753 	( EXT3_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask) )
754 #define EXT3_HAS_RO_COMPAT_FEATURE(sb,mask)			\
755 	( EXT3_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask) )
756 #define EXT3_HAS_INCOMPAT_FEATURE(sb,mask)			\
757 	( EXT3_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask) )
758 #define EXT3_SET_COMPAT_FEATURE(sb,mask)			\
759 	EXT3_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
760 #define EXT3_SET_RO_COMPAT_FEATURE(sb,mask)			\
761 	EXT3_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
762 #define EXT3_SET_INCOMPAT_FEATURE(sb,mask)			\
763 	EXT3_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
764 #define EXT3_CLEAR_COMPAT_FEATURE(sb,mask)			\
765 	EXT3_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
766 #define EXT3_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
767 	EXT3_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
768 #define EXT3_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
769 	EXT3_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
770 
771 #define EXT3_FEATURE_COMPAT_DIR_PREALLOC	0x0001
772 #define EXT3_FEATURE_COMPAT_IMAGIC_INODES	0x0002
773 #define EXT3_FEATURE_COMPAT_HAS_JOURNAL		0x0004
774 #define EXT3_FEATURE_COMPAT_EXT_ATTR		0x0008
775 #define EXT3_FEATURE_COMPAT_RESIZE_INODE	0x0010
776 #define EXT3_FEATURE_COMPAT_DIR_INDEX		0x0020
777 
778 #define EXT3_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
779 #define EXT3_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
780 #define EXT3_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
781 
782 #define EXT3_FEATURE_INCOMPAT_COMPRESSION	0x0001
783 #define EXT3_FEATURE_INCOMPAT_FILETYPE		0x0002
784 #define EXT3_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
785 #define EXT3_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
786 #define EXT3_FEATURE_INCOMPAT_META_BG		0x0010
787 
788 #define EXT3_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
789 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT3_FEATURE_INCOMPAT_FILETYPE| \
790 					 EXT3_FEATURE_INCOMPAT_RECOVER| \
791 					 EXT3_FEATURE_INCOMPAT_META_BG)
792 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT3_FEATURE_RO_COMPAT_SPARSE_SUPER| \
793 					 EXT3_FEATURE_RO_COMPAT_LARGE_FILE| \
794 					 EXT3_FEATURE_RO_COMPAT_BTREE_DIR)
795 
796 /*
797  * Default values for user and/or group using reserved blocks
798  */
799 #define	EXT3_DEF_RESUID		0
800 #define	EXT3_DEF_RESGID		0
801 
802 /*
803  * Default mount options
804  */
805 #define EXT3_DEFM_DEBUG		0x0001
806 #define EXT3_DEFM_BSDGROUPS	0x0002
807 #define EXT3_DEFM_XATTR_USER	0x0004
808 #define EXT3_DEFM_ACL		0x0008
809 #define EXT3_DEFM_UID16		0x0010
810 #define EXT3_DEFM_JMODE		0x0060
811 #define EXT3_DEFM_JMODE_DATA	0x0020
812 #define EXT3_DEFM_JMODE_ORDERED	0x0040
813 #define EXT3_DEFM_JMODE_WBACK	0x0060
814 
815 /*
816  * Structure of a directory entry
817  */
818 #define EXT3_NAME_LEN 255
819 
820 struct ext3_dir_entry {
821 	__le32	inode;			/* Inode number */
822 	__le16	rec_len;		/* Directory entry length */
823 	__le16	name_len;		/* Name length */
824 	char	name[EXT3_NAME_LEN];	/* File name */
825 };
826 
827 /*
828  * The new version of the directory entry.  Since EXT3 structures are
829  * stored in intel byte order, and the name_len field could never be
830  * bigger than 255 chars, it's safe to reclaim the extra byte for the
831  * file_type field.
832  */
833 struct ext3_dir_entry_2 {
834 	__le32	inode;			/* Inode number */
835 	__le16	rec_len;		/* Directory entry length */
836 	__u8	name_len;		/* Name length */
837 	__u8	file_type;
838 	char	name[EXT3_NAME_LEN];	/* File name */
839 };
840 
841 /*
842  * Ext3 directory file types.  Only the low 3 bits are used.  The
843  * other bits are reserved for now.
844  */
845 #define EXT3_FT_UNKNOWN		0
846 #define EXT3_FT_REG_FILE	1
847 #define EXT3_FT_DIR		2
848 #define EXT3_FT_CHRDEV		3
849 #define EXT3_FT_BLKDEV		4
850 #define EXT3_FT_FIFO		5
851 #define EXT3_FT_SOCK		6
852 #define EXT3_FT_SYMLINK		7
853 
854 #define EXT3_FT_MAX		8
855 
856 /*
857  * EXT3_DIR_PAD defines the directory entries boundaries
858  *
859  * NOTE: It must be a multiple of 4
860  */
861 #define EXT3_DIR_PAD			4
862 #define EXT3_DIR_ROUND			(EXT3_DIR_PAD - 1)
863 #define EXT3_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT3_DIR_ROUND) & \
864 					 ~EXT3_DIR_ROUND)
865 #define EXT3_MAX_REC_LEN		((1<<16)-1)
866 
867 /*
868  * Tests against MAX_REC_LEN etc were put in place for 64k block
869  * sizes; if that is not possible on this arch, we can skip
870  * those tests and speed things up.
871  */
ext3_rec_len_from_disk(__le16 dlen)872 static inline unsigned ext3_rec_len_from_disk(__le16 dlen)
873 {
874 	unsigned len = le16_to_cpu(dlen);
875 
876 #if (PAGE_CACHE_SIZE >= 65536)
877 	if (len == EXT3_MAX_REC_LEN)
878 		return 1 << 16;
879 #endif
880 	return len;
881 }
882 
ext3_rec_len_to_disk(unsigned len)883 static inline __le16 ext3_rec_len_to_disk(unsigned len)
884 {
885 #if (PAGE_CACHE_SIZE >= 65536)
886 	if (len == (1 << 16))
887 		return cpu_to_le16(EXT3_MAX_REC_LEN);
888 	else if (len > (1 << 16))
889 		BUG();
890 #endif
891 	return cpu_to_le16(len);
892 }
893 
894 /*
895  * Hash Tree Directory indexing
896  * (c) Daniel Phillips, 2001
897  */
898 
899 #define is_dx(dir) (EXT3_HAS_COMPAT_FEATURE(dir->i_sb, \
900 				      EXT3_FEATURE_COMPAT_DIR_INDEX) && \
901 		      (EXT3_I(dir)->i_flags & EXT3_INDEX_FL))
902 #define EXT3_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT3_LINK_MAX)
903 #define EXT3_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
904 
905 /* Legal values for the dx_root hash_version field: */
906 
907 #define DX_HASH_LEGACY		0
908 #define DX_HASH_HALF_MD4	1
909 #define DX_HASH_TEA		2
910 #define DX_HASH_LEGACY_UNSIGNED	3
911 #define DX_HASH_HALF_MD4_UNSIGNED	4
912 #define DX_HASH_TEA_UNSIGNED		5
913 
914 /* hash info structure used by the directory hash */
915 struct dx_hash_info
916 {
917 	u32		hash;
918 	u32		minor_hash;
919 	int		hash_version;
920 	u32		*seed;
921 };
922 
923 
924 /* 32 and 64 bit signed EOF for dx directories */
925 #define EXT3_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
926 #define EXT3_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
927 
928 
929 /*
930  * Control parameters used by ext3_htree_next_block
931  */
932 #define HASH_NB_ALWAYS		1
933 
934 
935 /*
936  * Describe an inode's exact location on disk and in memory
937  */
938 struct ext3_iloc
939 {
940 	struct buffer_head *bh;
941 	unsigned long offset;
942 	unsigned long block_group;
943 };
944 
ext3_raw_inode(struct ext3_iloc * iloc)945 static inline struct ext3_inode *ext3_raw_inode(struct ext3_iloc *iloc)
946 {
947 	return (struct ext3_inode *) (iloc->bh->b_data + iloc->offset);
948 }
949 
950 /*
951  * This structure is stuffed into the struct file's private_data field
952  * for directories.  It is where we put information so that we can do
953  * readdir operations in hash tree order.
954  */
955 struct dir_private_info {
956 	struct rb_root	root;
957 	struct rb_node	*curr_node;
958 	struct fname	*extra_fname;
959 	loff_t		last_pos;
960 	__u32		curr_hash;
961 	__u32		curr_minor_hash;
962 	__u32		next_hash;
963 };
964 
965 /* calculate the first block number of the group */
966 static inline ext3_fsblk_t
ext3_group_first_block_no(struct super_block * sb,unsigned long group_no)967 ext3_group_first_block_no(struct super_block *sb, unsigned long group_no)
968 {
969 	return group_no * (ext3_fsblk_t)EXT3_BLOCKS_PER_GROUP(sb) +
970 		le32_to_cpu(EXT3_SB(sb)->s_es->s_first_data_block);
971 }
972 
973 /*
974  * Special error return code only used by dx_probe() and its callers.
975  */
976 #define ERR_BAD_DX_DIR	-75000
977 
978 /*
979  * Function prototypes
980  */
981 
982 /*
983  * Ok, these declarations are also in <linux/kernel.h> but none of the
984  * ext3 source programs needs to include it so they are duplicated here.
985  */
986 # define NORET_TYPE    /**/
987 # define ATTRIB_NORET  __attribute__((noreturn))
988 # define NORET_AND     noreturn,
989 
990 /* balloc.c */
991 extern int ext3_bg_has_super(struct super_block *sb, int group);
992 extern unsigned long ext3_bg_num_gdb(struct super_block *sb, int group);
993 extern ext3_fsblk_t ext3_new_block (handle_t *handle, struct inode *inode,
994 			ext3_fsblk_t goal, int *errp);
995 extern ext3_fsblk_t ext3_new_blocks (handle_t *handle, struct inode *inode,
996 			ext3_fsblk_t goal, unsigned long *count, int *errp);
997 extern void ext3_free_blocks (handle_t *handle, struct inode *inode,
998 			ext3_fsblk_t block, unsigned long count);
999 extern void ext3_free_blocks_sb (handle_t *handle, struct super_block *sb,
1000 				 ext3_fsblk_t block, unsigned long count,
1001 				unsigned long *pdquot_freed_blocks);
1002 extern ext3_fsblk_t ext3_count_free_blocks (struct super_block *);
1003 extern void ext3_check_blocks_bitmap (struct super_block *);
1004 extern struct ext3_group_desc * ext3_get_group_desc(struct super_block * sb,
1005 						    unsigned int block_group,
1006 						    struct buffer_head ** bh);
1007 extern int ext3_should_retry_alloc(struct super_block *sb, int *retries);
1008 extern void ext3_init_block_alloc_info(struct inode *);
1009 extern void ext3_rsv_window_add(struct super_block *sb, struct ext3_reserve_window_node *rsv);
1010 extern int ext3_trim_fs(struct super_block *sb, struct fstrim_range *range);
1011 
1012 /* dir.c */
1013 extern int ext3_check_dir_entry(const char *, struct inode *,
1014 				struct ext3_dir_entry_2 *,
1015 				struct buffer_head *, unsigned long);
1016 extern int ext3_htree_store_dirent(struct file *dir_file, __u32 hash,
1017 				    __u32 minor_hash,
1018 				    struct ext3_dir_entry_2 *dirent);
1019 extern void ext3_htree_free_dir_info(struct dir_private_info *p);
1020 
1021 /* fsync.c */
1022 extern int ext3_sync_file(struct file *, loff_t, loff_t, int);
1023 
1024 /* hash.c */
1025 extern int ext3fs_dirhash(const char *name, int len, struct
1026 			  dx_hash_info *hinfo);
1027 
1028 /* ialloc.c */
1029 extern struct inode * ext3_new_inode (handle_t *, struct inode *,
1030 				      const struct qstr *, umode_t);
1031 extern void ext3_free_inode (handle_t *, struct inode *);
1032 extern struct inode * ext3_orphan_get (struct super_block *, unsigned long);
1033 extern unsigned long ext3_count_free_inodes (struct super_block *);
1034 extern unsigned long ext3_count_dirs (struct super_block *);
1035 extern void ext3_check_inodes_bitmap (struct super_block *);
1036 extern unsigned long ext3_count_free (struct buffer_head *, unsigned);
1037 
1038 
1039 /* inode.c */
1040 int ext3_forget(handle_t *handle, int is_metadata, struct inode *inode,
1041 		struct buffer_head *bh, ext3_fsblk_t blocknr);
1042 struct buffer_head * ext3_getblk (handle_t *, struct inode *, long, int, int *);
1043 struct buffer_head * ext3_bread (handle_t *, struct inode *, int, int, int *);
1044 int ext3_get_blocks_handle(handle_t *handle, struct inode *inode,
1045 	sector_t iblock, unsigned long maxblocks, struct buffer_head *bh_result,
1046 	int create);
1047 
1048 extern struct inode *ext3_iget(struct super_block *, unsigned long);
1049 extern int  ext3_write_inode (struct inode *, struct writeback_control *);
1050 extern int  ext3_setattr (struct dentry *, struct iattr *);
1051 extern void ext3_evict_inode (struct inode *);
1052 extern int  ext3_sync_inode (handle_t *, struct inode *);
1053 extern void ext3_discard_reservation (struct inode *);
1054 extern void ext3_dirty_inode(struct inode *, int);
1055 extern int ext3_change_inode_journal_flag(struct inode *, int);
1056 extern int ext3_get_inode_loc(struct inode *, struct ext3_iloc *);
1057 extern int ext3_can_truncate(struct inode *inode);
1058 extern void ext3_truncate(struct inode *inode);
1059 extern void ext3_set_inode_flags(struct inode *);
1060 extern void ext3_get_inode_flags(struct ext3_inode_info *);
1061 extern void ext3_set_aops(struct inode *inode);
1062 extern int ext3_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1063 		       u64 start, u64 len);
1064 
1065 /* ioctl.c */
1066 extern long ext3_ioctl(struct file *, unsigned int, unsigned long);
1067 extern long ext3_compat_ioctl(struct file *, unsigned int, unsigned long);
1068 
1069 /* namei.c */
1070 extern int ext3_orphan_add(handle_t *, struct inode *);
1071 extern int ext3_orphan_del(handle_t *, struct inode *);
1072 extern int ext3_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1073 				__u32 start_minor_hash, __u32 *next_hash);
1074 
1075 /* resize.c */
1076 extern int ext3_group_add(struct super_block *sb,
1077 				struct ext3_new_group_data *input);
1078 extern int ext3_group_extend(struct super_block *sb,
1079 				struct ext3_super_block *es,
1080 				ext3_fsblk_t n_blocks_count);
1081 
1082 /* super.c */
1083 extern __printf(3, 4)
1084 void ext3_error(struct super_block *, const char *, const char *, ...);
1085 extern void __ext3_std_error (struct super_block *, const char *, int);
1086 extern __printf(3, 4)
1087 void ext3_abort(struct super_block *, const char *, const char *, ...);
1088 extern __printf(3, 4)
1089 void ext3_warning(struct super_block *, const char *, const char *, ...);
1090 extern __printf(3, 4)
1091 void ext3_msg(struct super_block *, const char *, const char *, ...);
1092 extern void ext3_update_dynamic_rev (struct super_block *sb);
1093 
1094 #define ext3_std_error(sb, errno)				\
1095 do {								\
1096 	if ((errno))						\
1097 		__ext3_std_error((sb), __func__, (errno));	\
1098 } while (0)
1099 
1100 /*
1101  * Inodes and files operations
1102  */
1103 
1104 /* dir.c */
1105 extern const struct file_operations ext3_dir_operations;
1106 
1107 /* file.c */
1108 extern const struct inode_operations ext3_file_inode_operations;
1109 extern const struct file_operations ext3_file_operations;
1110 
1111 /* namei.c */
1112 extern const struct inode_operations ext3_dir_inode_operations;
1113 extern const struct inode_operations ext3_special_inode_operations;
1114 
1115 /* symlink.c */
1116 extern const struct inode_operations ext3_symlink_inode_operations;
1117 extern const struct inode_operations ext3_fast_symlink_inode_operations;
1118 
1119 #define EXT3_JOURNAL(inode)	(EXT3_SB((inode)->i_sb)->s_journal)
1120 
1121 /* Define the number of blocks we need to account to a transaction to
1122  * modify one block of data.
1123  *
1124  * We may have to touch one inode, one bitmap buffer, up to three
1125  * indirection blocks, the group and superblock summaries, and the data
1126  * block to complete the transaction.  */
1127 
1128 #define EXT3_SINGLEDATA_TRANS_BLOCKS	8U
1129 
1130 /* Extended attribute operations touch at most two data buffers,
1131  * two bitmap buffers, and two group summaries, in addition to the inode
1132  * and the superblock, which are already accounted for. */
1133 
1134 #define EXT3_XATTR_TRANS_BLOCKS		6U
1135 
1136 /* Define the minimum size for a transaction which modifies data.  This
1137  * needs to take into account the fact that we may end up modifying two
1138  * quota files too (one for the group, one for the user quota).  The
1139  * superblock only gets updated once, of course, so don't bother
1140  * counting that again for the quota updates. */
1141 
1142 #define EXT3_DATA_TRANS_BLOCKS(sb)	(EXT3_SINGLEDATA_TRANS_BLOCKS + \
1143 					 EXT3_XATTR_TRANS_BLOCKS - 2 + \
1144 					 EXT3_MAXQUOTAS_TRANS_BLOCKS(sb))
1145 
1146 /* Delete operations potentially hit one directory's namespace plus an
1147  * entire inode, plus arbitrary amounts of bitmap/indirection data.  Be
1148  * generous.  We can grow the delete transaction later if necessary. */
1149 
1150 #define EXT3_DELETE_TRANS_BLOCKS(sb)   (EXT3_MAXQUOTAS_TRANS_BLOCKS(sb) + 64)
1151 
1152 /* Define an arbitrary limit for the amount of data we will anticipate
1153  * writing to any given transaction.  For unbounded transactions such as
1154  * write(2) and truncate(2) we can write more than this, but we always
1155  * start off at the maximum transaction size and grow the transaction
1156  * optimistically as we go. */
1157 
1158 #define EXT3_MAX_TRANS_DATA		64U
1159 
1160 /* We break up a large truncate or write transaction once the handle's
1161  * buffer credits gets this low, we need either to extend the
1162  * transaction or to start a new one.  Reserve enough space here for
1163  * inode, bitmap, superblock, group and indirection updates for at least
1164  * one block, plus two quota updates.  Quota allocations are not
1165  * needed. */
1166 
1167 #define EXT3_RESERVE_TRANS_BLOCKS	12U
1168 
1169 #define EXT3_INDEX_EXTRA_TRANS_BLOCKS	8
1170 
1171 #ifdef CONFIG_QUOTA
1172 /* Amount of blocks needed for quota update - we know that the structure was
1173  * allocated so we need to update only inode+data */
1174 #define EXT3_QUOTA_TRANS_BLOCKS(sb) (test_opt(sb, QUOTA) ? 2 : 0)
1175 /* Amount of blocks needed for quota insert/delete - we do some block writes
1176  * but inode, sb and group updates are done only once */
1177 #define EXT3_QUOTA_INIT_BLOCKS(sb) (test_opt(sb, QUOTA) ? (DQUOT_INIT_ALLOC*\
1178 		(EXT3_SINGLEDATA_TRANS_BLOCKS-3)+3+DQUOT_INIT_REWRITE) : 0)
1179 #define EXT3_QUOTA_DEL_BLOCKS(sb) (test_opt(sb, QUOTA) ? (DQUOT_DEL_ALLOC*\
1180 		(EXT3_SINGLEDATA_TRANS_BLOCKS-3)+3+DQUOT_DEL_REWRITE) : 0)
1181 #else
1182 #define EXT3_QUOTA_TRANS_BLOCKS(sb) 0
1183 #define EXT3_QUOTA_INIT_BLOCKS(sb) 0
1184 #define EXT3_QUOTA_DEL_BLOCKS(sb) 0
1185 #endif
1186 #define EXT3_MAXQUOTAS_TRANS_BLOCKS(sb) (MAXQUOTAS*EXT3_QUOTA_TRANS_BLOCKS(sb))
1187 #define EXT3_MAXQUOTAS_INIT_BLOCKS(sb) (MAXQUOTAS*EXT3_QUOTA_INIT_BLOCKS(sb))
1188 #define EXT3_MAXQUOTAS_DEL_BLOCKS(sb) (MAXQUOTAS*EXT3_QUOTA_DEL_BLOCKS(sb))
1189 
1190 int
1191 ext3_mark_iloc_dirty(handle_t *handle,
1192 		     struct inode *inode,
1193 		     struct ext3_iloc *iloc);
1194 
1195 /*
1196  * On success, We end up with an outstanding reference count against
1197  * iloc->bh.  This _must_ be cleaned up later.
1198  */
1199 
1200 int ext3_reserve_inode_write(handle_t *handle, struct inode *inode,
1201 			struct ext3_iloc *iloc);
1202 
1203 int ext3_mark_inode_dirty(handle_t *handle, struct inode *inode);
1204 
1205 /*
1206  * Wrapper functions with which ext3 calls into JBD.  The intent here is
1207  * to allow these to be turned into appropriate stubs so ext3 can control
1208  * ext2 filesystems, so ext2+ext3 systems only nee one fs.  This work hasn't
1209  * been done yet.
1210  */
1211 
ext3_journal_release_buffer(handle_t * handle,struct buffer_head * bh)1212 static inline void ext3_journal_release_buffer(handle_t *handle,
1213 						struct buffer_head *bh)
1214 {
1215 	journal_release_buffer(handle, bh);
1216 }
1217 
1218 void ext3_journal_abort_handle(const char *caller, const char *err_fn,
1219 		struct buffer_head *bh, handle_t *handle, int err);
1220 
1221 int __ext3_journal_get_undo_access(const char *where, handle_t *handle,
1222 				struct buffer_head *bh);
1223 
1224 int __ext3_journal_get_write_access(const char *where, handle_t *handle,
1225 				struct buffer_head *bh);
1226 
1227 int __ext3_journal_forget(const char *where, handle_t *handle,
1228 				struct buffer_head *bh);
1229 
1230 int __ext3_journal_revoke(const char *where, handle_t *handle,
1231 				unsigned long blocknr, struct buffer_head *bh);
1232 
1233 int __ext3_journal_get_create_access(const char *where,
1234 				handle_t *handle, struct buffer_head *bh);
1235 
1236 int __ext3_journal_dirty_metadata(const char *where,
1237 				handle_t *handle, struct buffer_head *bh);
1238 
1239 #define ext3_journal_get_undo_access(handle, bh) \
1240 	__ext3_journal_get_undo_access(__func__, (handle), (bh))
1241 #define ext3_journal_get_write_access(handle, bh) \
1242 	__ext3_journal_get_write_access(__func__, (handle), (bh))
1243 #define ext3_journal_revoke(handle, blocknr, bh) \
1244 	__ext3_journal_revoke(__func__, (handle), (blocknr), (bh))
1245 #define ext3_journal_get_create_access(handle, bh) \
1246 	__ext3_journal_get_create_access(__func__, (handle), (bh))
1247 #define ext3_journal_dirty_metadata(handle, bh) \
1248 	__ext3_journal_dirty_metadata(__func__, (handle), (bh))
1249 #define ext3_journal_forget(handle, bh) \
1250 	__ext3_journal_forget(__func__, (handle), (bh))
1251 
1252 int ext3_journal_dirty_data(handle_t *handle, struct buffer_head *bh);
1253 
1254 handle_t *ext3_journal_start_sb(struct super_block *sb, int nblocks);
1255 int __ext3_journal_stop(const char *where, handle_t *handle);
1256 
ext3_journal_start(struct inode * inode,int nblocks)1257 static inline handle_t *ext3_journal_start(struct inode *inode, int nblocks)
1258 {
1259 	return ext3_journal_start_sb(inode->i_sb, nblocks);
1260 }
1261 
1262 #define ext3_journal_stop(handle) \
1263 	__ext3_journal_stop(__func__, (handle))
1264 
ext3_journal_current_handle(void)1265 static inline handle_t *ext3_journal_current_handle(void)
1266 {
1267 	return journal_current_handle();
1268 }
1269 
ext3_journal_extend(handle_t * handle,int nblocks)1270 static inline int ext3_journal_extend(handle_t *handle, int nblocks)
1271 {
1272 	return journal_extend(handle, nblocks);
1273 }
1274 
ext3_journal_restart(handle_t * handle,int nblocks)1275 static inline int ext3_journal_restart(handle_t *handle, int nblocks)
1276 {
1277 	return journal_restart(handle, nblocks);
1278 }
1279 
ext3_journal_blocks_per_page(struct inode * inode)1280 static inline int ext3_journal_blocks_per_page(struct inode *inode)
1281 {
1282 	return journal_blocks_per_page(inode);
1283 }
1284 
ext3_journal_force_commit(journal_t * journal)1285 static inline int ext3_journal_force_commit(journal_t *journal)
1286 {
1287 	return journal_force_commit(journal);
1288 }
1289 
1290 /* super.c */
1291 int ext3_force_commit(struct super_block *sb);
1292 
ext3_should_journal_data(struct inode * inode)1293 static inline int ext3_should_journal_data(struct inode *inode)
1294 {
1295 	if (!S_ISREG(inode->i_mode))
1296 		return 1;
1297 	if (test_opt(inode->i_sb, DATA_FLAGS) == EXT3_MOUNT_JOURNAL_DATA)
1298 		return 1;
1299 	if (EXT3_I(inode)->i_flags & EXT3_JOURNAL_DATA_FL)
1300 		return 1;
1301 	return 0;
1302 }
1303 
ext3_should_order_data(struct inode * inode)1304 static inline int ext3_should_order_data(struct inode *inode)
1305 {
1306 	if (!S_ISREG(inode->i_mode))
1307 		return 0;
1308 	if (EXT3_I(inode)->i_flags & EXT3_JOURNAL_DATA_FL)
1309 		return 0;
1310 	if (test_opt(inode->i_sb, DATA_FLAGS) == EXT3_MOUNT_ORDERED_DATA)
1311 		return 1;
1312 	return 0;
1313 }
1314 
ext3_should_writeback_data(struct inode * inode)1315 static inline int ext3_should_writeback_data(struct inode *inode)
1316 {
1317 	if (!S_ISREG(inode->i_mode))
1318 		return 0;
1319 	if (EXT3_I(inode)->i_flags & EXT3_JOURNAL_DATA_FL)
1320 		return 0;
1321 	if (test_opt(inode->i_sb, DATA_FLAGS) == EXT3_MOUNT_WRITEBACK_DATA)
1322 		return 1;
1323 	return 0;
1324 }
1325 
1326 #include <trace/events/ext3.h>
1327