Online Linux Driver Verification Service (alpha)

1 2 /** 3 * The test checks that double mutex lock is detected on the models 32_1,32_2 4 */ 5 #include <linux/kernel.h> 6 #include <linux/module.h> 7 #include <linux/mutex.h> 8 #include <linux/major.h> 9 #include <linux/fs.h> 10 11 static DEFINE_MUTEX(my_lock); 12 13 static int misc_open(struct inode * inode, struct file * file); 14 15 static const struct file_operations misc_fops = { 16 .owner = THIS_MODULE, 17 .open = misc_open, 18 }; 19 20 static void alock(void) { 21 mutex_lock(&my_lock); 22 mutex_unlock(&my_lock); 23 } 24 25 static int misc_open(struct inode * inode, struct file * file) 26 { 27 mutex_lock(&my_lock); 28 alock(); 29 mutex_unlock(&my_lock); 30 return 0; 31 } 32 33 static int __init my_init(void) 34 { 35 if (register_chrdev(MISC_MAJOR,"misc",&misc_fops)) 36 goto fail_register; 37 return 0; 38 39 fail_register: 40 return -1; 41 } 42 43 static void __exit my_exit(void) 44 { 45 } 46 47 module_init(my_init); 48 module_exit(my_exit); 49 50 MODULE_LICENSE("Apache 2.0"); 51 MODULE_AUTHOR("LDV Project, Vadim Mutilin <mutilin@ispras.ru>"); 52 53 54 55 56 57 58 /* LDV_COMMENT_BEGIN_MAIN */ 59 #ifdef LDV_MAIN0_sequence_infinite_withcheck_stateful 60 61 /*###########################################################################*/ 62 63 /*############## Driver Environment Generator 0.2 output ####################*/ 64 65 /*###########################################################################*/ 66 67 68 69 /* LDV_COMMENT_FUNCTION_DECLARE_LDV Special function for LDV verifier. Test if all kernel resources are correctly released by driver before driver will be unloaded. */ 70 void ldv_check_final_state(void); 71 72 /* LDV_COMMENT_FUNCTION_DECLARE_LDV Special function for LDV verifier. Test correct return result. */ 73 void ldv_check_return_value(int res); 74 75 /* LDV_COMMENT_FUNCTION_DECLARE_LDV Special function for LDV verifier. Initializes the model. */ 76 void ldv_initialize(void); 77 78 /* LDV_COMMENT_FUNCTION_DECLARE_LDV Special function for LDV verifier. Returns arbitrary interger value. */ 79 int nondet_int(void); 80 81 /* LDV_COMMENT_VAR_DECLARE_LDV Special variable for LDV verifier. */ 82 int LDV_IN_INTERRUPT; 83 84 /* LDV_COMMENT_FUNCTION_MAIN Main function for LDV verifier. */ 85 void ldv_main0_sequence_infinite_withcheck_stateful(void) { 86 87 88 89 /* LDV_COMMENT_BEGIN_VARIABLE_DECLARATION_PART */ 90 /*============================= VARIABLE DECLARATION PART =============================*/ 91 /** STRUCT: struct type: file_operations, struct name: misc_fops **/ 92 /* content: static int misc_open(struct inode * inode, struct file * file)*/ 93 /* LDV_COMMENT_END_PREP */ 94 /* LDV_COMMENT_VAR_DECLARE Variable declaration for function "misc_open" */ 95 struct inode * var_group1; 96 /* LDV_COMMENT_VAR_DECLARE Variable declaration for function "misc_open" */ 97 struct file * var_group2; 98 /* LDV_COMMENT_VAR_DECLARE Variable declaration for test return result from function call "misc_open" */ 99 static int res_misc_open_1; 100 101 102 103 104 /* LDV_COMMENT_END_VARIABLE_DECLARATION_PART */ 105 /* LDV_COMMENT_BEGIN_VARIABLE_INITIALIZING_PART */ 106 /*============================= VARIABLE INITIALIZING PART =============================*/ 107 LDV_IN_INTERRUPT=1; 108 109 110 111 112 /* LDV_COMMENT_END_VARIABLE_INITIALIZING_PART */ 113 /* LDV_COMMENT_BEGIN_FUNCTION_CALL_SECTION */ 114 /*============================= FUNCTION CALL SECTION =============================*/ 115 /* LDV_COMMENT_FUNCTION_CALL Initialize LDV model. */ 116 ldv_initialize(); 117 118 /** INIT: init_type: ST_MODULE_INIT **/ 119 /* content: static int __init my_init(void)*/ 120 /* LDV_COMMENT_END_PREP */ 121 /* LDV_COMMENT_FUNCTION_CALL Kernel calls driver init function after driver loading to kernel. This function declared as "MODULE_INIT(function name)". */ 122 if(my_init()) 123 goto ldv_final; 124 int ldv_s_misc_fops_file_operations = 0; 125 126 127 while( nondet_int() 128 || !(ldv_s_misc_fops_file_operations == 0) 129 ) { 130 131 switch(nondet_int()) { 132 133 case 0: { 134 135 /** STRUCT: struct type: file_operations, struct name: misc_fops **/ 136 if(ldv_s_misc_fops_file_operations==0) { 137 138 /* content: static int misc_open(struct inode * inode, struct file * file)*/ 139 /* LDV_COMMENT_END_PREP */ 140 /* LDV_COMMENT_FUNCTION_CALL Function from field "open" from driver structure with callbacks "misc_fops". Standart function test for correct return result. */ 141 res_misc_open_1 = misc_open( var_group1, var_group2); 142 ldv_check_return_value(res_misc_open_1); 143 if(res_misc_open_1) 144 goto ldv_module_exit; 145 ldv_s_misc_fops_file_operations=0; 146 147 } 148 149 } 150 151 break; 152 default: break; 153 154 } 155 156 } 157 158 ldv_module_exit: 159 160 /** INIT: init_type: ST_MODULE_EXIT **/ 161 /* content: static void __exit my_exit(void)*/ 162 /* LDV_COMMENT_END_PREP */ 163 /* LDV_COMMENT_FUNCTION_CALL Kernel calls driver release function before driver will be uploaded from kernel. This function declared as "MODULE_EXIT(function name)". */ 164 my_exit(); 165 166 /* LDV_COMMENT_FUNCTION_CALL Checks that all resources and locks are correctly released before the driver will be unloaded. */ 167 ldv_final: ldv_check_final_state(); 168 169 /* LDV_COMMENT_END_FUNCTION_CALL_SECTION */ 170 return; 171 172 } 173 #endif 174 175 /* LDV_COMMENT_END_MAIN */ 176 /* LDV_COMMENT_BEGIN_MODEL */ 177 178 /* 179 * CONFIG_DEBUG_LOCK_ALLOC must be turned off to apply the given model. 180 * To be independ on the value of this flag there is the corresponding 181 * aspect model. 182 */ 183 184 /* the function works only without aspectator */ 185 long __builtin_expect(long val, long res) { 186 return val; 187 } 188 189 #include "engine-blast.h" 190 191 #include <linux/kernel.h> 192 #include <linux/mutex.h> 193 194 /* Need this because rerouter is buggy!.. */ 195 extern int ldv_mutex_TEMPLATE; 196 /* Now the actual variable goes... */ 197 int ldv_mutex_TEMPLATE = 1; 198 199 /* LDV_COMMENT_MODEL_FUNCTION_DEFINITION(name='mutex_lock_interruptible') Check that the mutex was unlocked and nondeterministically lock it. Return the corresponding error code on fails*/ 200 int mutex_lock_interruptible_TEMPLATE(struct mutex *lock) 201 { 202 int nondetermined; 203 204 /* LDV_COMMENT_ASSERT Mutex must be unlocked*/ 205 ldv_assert(ldv_mutex_TEMPLATE == 1); 206 207 /* LDV_COMMENT_OTHER Construct the nondetermined result*/ 208 nondetermined = ldv_undef_int(); 209 210 /* LDV_COMMENT_ASSERT Nondeterministically lock the mutex*/ 211 if (nondetermined) 212 { 213 /* LDV_COMMENT_CHANGE_STATE Lock the mutex*/ 214 ldv_mutex_TEMPLATE = 2; 215 /* LDV_COMMENT_RETURN Finish with success*/ 216 return 0; 217 } 218 else 219 { 220 /* LDV_COMMENT_RETURN Finish with the fail. The mutex is keeped unlocked*/ 221 return -EINTR; 222 } 223 } 224 225 /* LDV_COMMENT_MODEL_FUNCTION_DEFINITION(name='atomic_dec_and_mutex_lock') Lock if atomic decrement result is zero */ 226 int atomic_dec_and_mutex_lock_TEMPLATE(atomic_t *cnt, struct mutex *lock) 227 { 228 int atomic_value_after_dec; 229 230 /* LDV_COMMENT_ASSERT Mutex must be unlocked (since we may lock it in this function) */ 231 ldv_assert(ldv_mutex_TEMPLATE == 1); 232 233 /* LDV_COMMENT_OTHER Assign the result of atomic decrement */ 234 atomic_value_after_dec = ldv_undef_int(); 235 236 /* LDV_COMMENT_ASSERT Check if atomic decrement returns zero */ 237 if (atomic_value_after_dec == 0) 238 { 239 /* LDV_COMMENT_CHANGE_STATE Lock the mutex, as atomic has decremented to zero*/ 240 ldv_mutex_TEMPLATE = 2; 241 /* LDV_COMMENT_RETURN Return 1 with a locked mutex */ 242 return 1; 243 } 244 245 /* LDV_COMMENT_RETURN Atomic decrement is still not zero, return 0 without locking the mutex */ 246 return 0; 247 } 248 249 /* LDV_COMMENT_MODEL_FUNCTION_DEFINITION(name='mutex_lock_killable') Check that the mutex wasn unlocked and nondeterministically lock it. Return the corresponding error code on fails*/ 250 int __must_check mutex_lock_killable_TEMPLATE(struct mutex *lock) 251 { 252 int nondetermined; 253 254 /* LDV_COMMENT_ASSERT Mutex must be unlocked*/ 255 ldv_assert(ldv_mutex_TEMPLATE == 1); 256 257 /* LDV_COMMENT_OTHER Construct the nondetermined result*/ 258 nondetermined = ldv_undef_int(); 259 260 /* LDV_COMMENT_ASSERT Nondeterministically lock the mutex*/ 261 if (nondetermined) 262 { 263 /* LDV_COMMENT_CHANGE_STATE Lock the mutex*/ 264 ldv_mutex_TEMPLATE = 2; 265 /* LDV_COMMENT_RETURN Finish with success*/ 266 return 0; 267 } 268 else 269 { 270 /* LDV_COMMENT_RETURN Finish with the fail. The mutex is keeped unlocked*/ 271 return -EINTR; 272 } 273 } 274 275 /* LDV_COMMENT_MODEL_FUNCTION_DEFINITION(name='mutex_is_locked') Checks whether the mutex is locked*/ 276 int mutex_is_locked_TEMPLATE(struct mutex *lock) 277 { 278 int nondetermined; 279 280 if(ldv_mutex_TEMPLATE == 1) { 281 /* LDV_COMMENT_OTHER Construct the nondetermined result*/ 282 nondetermined = ldv_undef_int(); 283 if(nondetermined) { 284 /* LDV_COMMENT_RETURN the mutex is unlocked*/ 285 return 0; 286 } else { 287 /* LDV_COMMENT_RETURN the mutex is locked*/ 288 return 1; 289 } 290 } else { 291 /* LDV_COMMENT_RETURN the mutex is locked*/ 292 return 1; 293 } 294 } 295 296 /* LDV_COMMENT_MODEL_FUNCTION_DEFINITION(name='mutex_lock(?!_interruptible|_killable)') Check that the mutex was not locked and lock it*/ 297 void mutex_lock_TEMPLATE(struct mutex *lock) 298 { 299 /* LDV_COMMENT_ASSERT Mutex must be unlocked*/ 300 ldv_assert(ldv_mutex_TEMPLATE == 1); 301 /* LDV_COMMENT_CHANGE_STATE Lock the mutex*/ 302 ldv_mutex_TEMPLATE = 2; 303 } 304 305 /* LDV_COMMENT_MODEL_FUNCTION_DEFINITION(name='mutex_trylock') Check that the mutex was not locked and nondeterministically lock it. Return 0 on fails*/ 306 int mutex_trylock_TEMPLATE(struct mutex *lock) 307 { 308 int is_mutex_held_by_another_thread; 309 310 /* LDV_COMMENT_ASSERT Mutex must be unlocked*/ 311 ldv_assert(ldv_mutex_TEMPLATE == 1); 312 313 /* LDV_COMMENT_OTHER Construct the nondetermined result*/ 314 is_mutex_held_by_another_thread = ldv_undef_int(); 315 316 /* LDV_COMMENT_ASSERT Nondeterministically lock the mutex*/ 317 if (is_mutex_held_by_another_thread) 318 { 319 /* LDV_COMMENT_RETURN Finish with fail*/ 320 return 0; 321 } 322 else 323 { 324 /* LDV_COMMENT_CHANGE_STATE Lock the mutex*/ 325 ldv_mutex_TEMPLATE = 2; 326 /* LDV_COMMENT_RETURN Finish with success*/ 327 return 1; 328 } 329 } 330 331 /* LDV_COMMENT_MODEL_FUNCTION_DEFINITION(name='mutex_unlock') Check that the mutex was locked and unlock it*/ 332 void mutex_unlock_TEMPLATE(struct mutex *lock) 333 { 334 /* LDV_COMMENT_ASSERT Mutex must be locked*/ 335 ldv_assert(ldv_mutex_TEMPLATE == 2); 336 /* LDV_COMMENT_CHANGE_STATE Unlock the mutex*/ 337 ldv_mutex_TEMPLATE = 1; 338 } 339 340 /* LDV_COMMENT_MODEL_FUNCTION_DEFINITION(name='check_final_state') Check that the mutex is unlocked at the end*/ 341 void ldv_check_final_state_TEMPLATE(void) 342 { 343 /* LDV_COMMENT_ASSERT The mutex must be unlocked at the end*/ 344 ldv_assert(ldv_mutex_TEMPLATE == 1); 345 } 346 347 /* LDV_COMMENT_MODEL_FUNCTION_DEFINITION(name='ldv_initialize') Initialize mutex variables*/ 348 void ldv_initialize_TEMPLATE(void) 349 { 350 /* LDV_COMMENT_ASSERT Initialize mutex with initial model value*/ 351 ldv_mutex_TEMPLATE = 1; 352 } 353 354 /* LDV_COMMENT_END_MODEL */

1 #ifndef _LINUX_FS_H 2 #define _LINUX_FS_H 3 4 /* 5 * This file has definitions for some important file table 6 * structures etc. 7 */ 8 9 #include <linux/limits.h> 10 #include <linux/ioctl.h> 11 12 /* 13 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change 14 * the file limit at runtime and only root can increase the per-process 15 * nr_file rlimit, so it's safe to set up a ridiculously high absolute 16 * upper limit on files-per-process. 17 * 18 * Some programs (notably those using select()) may have to be 19 * recompiled to take full advantage of the new limits.. 20 */ 21 22 /* Fixed constants first: */ 23 #undef NR_OPEN 24 #define INR_OPEN 1024 /* Initial setting for nfile rlimits */ 25 26 #define BLOCK_SIZE_BITS 10 27 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS) 28 29 #define SEEK_SET 0 /* seek relative to beginning of file */ 30 #define SEEK_CUR 1 /* seek relative to current file position */ 31 #define SEEK_END 2 /* seek relative to end of file */ 32 #define SEEK_MAX SEEK_END 33 34 /* And dynamically-tunable limits and defaults: */ 35 struct files_stat_struct { 36 int nr_files; /* read only */ 37 int nr_free_files; /* read only */ 38 int max_files; /* tunable */ 39 }; 40 41 struct inodes_stat_t { 42 int nr_inodes; 43 int nr_unused; 44 int dummy[5]; /* padding for sysctl ABI compatibility */ 45 }; 46 47 48 #define NR_FILE 8192 /* this can well be larger on a larger system */ 49 50 #define MAY_EXEC 1 51 #define MAY_WRITE 2 52 #define MAY_READ 4 53 #define MAY_APPEND 8 54 #define MAY_ACCESS 16 55 #define MAY_OPEN 32 56 57 /* 58 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond 59 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open() 60 */ 61 62 /* file is open for reading */ 63 #define FMODE_READ ((__force fmode_t)1) 64 /* file is open for writing */ 65 #define FMODE_WRITE ((__force fmode_t)2) 66 /* file is seekable */ 67 #define FMODE_LSEEK ((__force fmode_t)4) 68 /* file can be accessed using pread */ 69 #define FMODE_PREAD ((__force fmode_t)8) 70 /* file can be accessed using pwrite */ 71 #define FMODE_PWRITE ((__force fmode_t)16) 72 /* File is opened for execution with sys_execve / sys_uselib */ 73 #define FMODE_EXEC ((__force fmode_t)32) 74 /* File is opened with O_NDELAY (only set for block devices) */ 75 #define FMODE_NDELAY ((__force fmode_t)64) 76 /* File is opened with O_EXCL (only set for block devices) */ 77 #define FMODE_EXCL ((__force fmode_t)128) 78 /* File is opened using open(.., 3, ..) and is writeable only for ioctls 79 (specialy hack for floppy.c) */ 80 #define FMODE_WRITE_IOCTL ((__force fmode_t)256) 81 82 /* 83 * Don't update ctime and mtime. 84 * 85 * Currently a special hack for the XFS open_by_handle ioctl, but we'll 86 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon. 87 */ 88 #define FMODE_NOCMTIME ((__force fmode_t)2048) 89 90 /* Expect random access pattern */ 91 #define FMODE_RANDOM ((__force fmode_t)4096) 92 93 /* 94 * The below are the various read and write types that we support. Some of 95 * them include behavioral modifiers that send information down to the 96 * block layer and IO scheduler. Terminology: 97 * 98 * The block layer uses device plugging to defer IO a little bit, in 99 * the hope that we will see more IO very shortly. This increases 100 * coalescing of adjacent IO and thus reduces the number of IOs we 101 * have to send to the device. It also allows for better queuing, 102 * if the IO isn't mergeable. If the caller is going to be waiting 103 * for the IO, then he must ensure that the device is unplugged so 104 * that the IO is dispatched to the driver. 105 * 106 * All IO is handled async in Linux. This is fine for background 107 * writes, but for reads or writes that someone waits for completion 108 * on, we want to notify the block layer and IO scheduler so that they 109 * know about it. That allows them to make better scheduling 110 * decisions. So when the below references 'sync' and 'async', it 111 * is referencing this priority hint. 112 * 113 * With that in mind, the available types are: 114 * 115 * READ A normal read operation. Device will be plugged. 116 * READ_SYNC A synchronous read. Device is not plugged, caller can 117 * immediately wait on this read without caring about 118 * unplugging. 119 * READA Used for read-ahead operations. Lower priority, and the 120 * block layer could (in theory) choose to ignore this 121 * request if it runs into resource problems. 122 * WRITE A normal async write. Device will be plugged. 123 * SWRITE Like WRITE, but a special case for ll_rw_block() that 124 * tells it to lock the buffer first. Normally a buffer 125 * must be locked before doing IO. 126 * WRITE_SYNC_PLUG Synchronous write. Identical to WRITE, but passes down 127 * the hint that someone will be waiting on this IO 128 * shortly. The device must still be unplugged explicitly, 129 * WRITE_SYNC_PLUG does not do this as we could be 130 * submitting more writes before we actually wait on any 131 * of them. 132 * WRITE_SYNC Like WRITE_SYNC_PLUG, but also unplugs the device 133 * immediately after submission. The write equivalent 134 * of READ_SYNC. 135 * WRITE_ODIRECT Special case write for O_DIRECT only. 136 * SWRITE_SYNC 137 * SWRITE_SYNC_PLUG Like WRITE_SYNC/WRITE_SYNC_PLUG, but locks the buffer. 138 * See SWRITE. 139 * WRITE_BARRIER Like WRITE, but tells the block layer that all 140 * previously submitted writes must be safely on storage 141 * before this one is started. Also guarantees that when 142 * this write is complete, it itself is also safely on 143 * storage. Prevents reordering of writes on both sides 144 * of this IO. 145 * 146 */ 147 #define RW_MASK 1 148 #define RWA_MASK 2 149 #define READ 0 150 #define WRITE 1 151 #define READA 2 /* read-ahead - don't block if no resources */ 152 #define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */ 153 #define READ_SYNC (READ | (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_UNPLUG)) 154 #define READ_META (READ | (1 << BIO_RW_META)) 155 #define WRITE_SYNC_PLUG (WRITE | (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_NOIDLE)) 156 #define WRITE_SYNC (WRITE_SYNC_PLUG | (1 << BIO_RW_UNPLUG)) 157 #define WRITE_ODIRECT (WRITE | (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_UNPLUG)) 158 #define SWRITE_SYNC_PLUG \ 159 (SWRITE | (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_NOIDLE)) 160 #define SWRITE_SYNC (SWRITE_SYNC_PLUG | (1 << BIO_RW_UNPLUG)) 161 #define WRITE_BARRIER (WRITE | (1 << BIO_RW_BARRIER)) 162 163 /* 164 * These aren't really reads or writes, they pass down information about 165 * parts of device that are now unused by the file system. 166 */ 167 #define DISCARD_NOBARRIER (WRITE | (1 << BIO_RW_DISCARD)) 168 #define DISCARD_BARRIER (DISCARD_NOBARRIER | (1 << BIO_RW_BARRIER)) 169 170 #define SEL_IN 1 171 #define SEL_OUT 2 172 #define SEL_EX 4 173 174 /* public flags for file_system_type */ 175 #define FS_REQUIRES_DEV 1 176 #define FS_BINARY_MOUNTDATA 2 177 #define FS_HAS_SUBTYPE 4 178 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */ 179 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() 180 * during rename() internally. 181 */ 182 183 /* 184 * These are the fs-independent mount-flags: up to 32 flags are supported 185 */ 186 #define MS_RDONLY 1 /* Mount read-only */ 187 #define MS_NOSUID 2 /* Ignore suid and sgid bits */ 188 #define MS_NODEV 4 /* Disallow access to device special files */ 189 #define MS_NOEXEC 8 /* Disallow program execution */ 190 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */ 191 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */ 192 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */ 193 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */ 194 #define MS_NOATIME 1024 /* Do not update access times. */ 195 #define MS_NODIRATIME 2048 /* Do not update directory access times */ 196 #define MS_BIND 4096 197 #define MS_MOVE 8192 198 #define MS_REC 16384 199 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence. 200 MS_VERBOSE is deprecated. */ 201 #define MS_SILENT 32768 202 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */ 203 #define MS_UNBINDABLE (1<<17) /* change to unbindable */ 204 #define MS_PRIVATE (1<<18) /* change to private */ 205 #define MS_SLAVE (1<<19) /* change to slave */ 206 #define MS_SHARED (1<<20) /* change to shared */ 207 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */ 208 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */ 209 #define MS_I_VERSION (1<<23) /* Update inode I_version field */ 210 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */ 211 #define MS_ACTIVE (1<<30) 212 #define MS_NOUSER (1<<31) 213 214 /* 215 * Superblock flags that can be altered by MS_REMOUNT 216 */ 217 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION) 218 219 /* 220 * Old magic mount flag and mask 221 */ 222 #define MS_MGC_VAL 0xC0ED0000 223 #define MS_MGC_MSK 0xffff0000 224 225 /* Inode flags - they have nothing to superblock flags now */ 226 227 #define S_SYNC 1 /* Writes are synced at once */ 228 #define S_NOATIME 2 /* Do not update access times */ 229 #define S_APPEND 4 /* Append-only file */ 230 #define S_IMMUTABLE 8 /* Immutable file */ 231 #define S_DEAD 16 /* removed, but still open directory */ 232 #define S_NOQUOTA 32 /* Inode is not counted to quota */ 233 #define S_DIRSYNC 64 /* Directory modifications are synchronous */ 234 #define S_NOCMTIME 128 /* Do not update file c/mtime */ 235 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */ 236 #define S_PRIVATE 512 /* Inode is fs-internal */ 237 238 /* 239 * Note that nosuid etc flags are inode-specific: setting some file-system 240 * flags just means all the inodes inherit those flags by default. It might be 241 * possible to override it selectively if you really wanted to with some 242 * ioctl() that is not currently implemented. 243 * 244 * Exception: MS_RDONLY is always applied to the entire file system. 245 * 246 * Unfortunately, it is possible to change a filesystems flags with it mounted 247 * with files in use. This means that all of the inodes will not have their 248 * i_flags updated. Hence, i_flags no longer inherit the superblock mount 249 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org 250 */ 251 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg)) 252 253 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY) 254 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \ 255 ((inode)->i_flags & S_SYNC)) 256 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \ 257 ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) 258 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK) 259 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME) 260 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION) 261 262 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) 263 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) 264 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) 265 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL) 266 267 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) 268 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) 269 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) 270 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) 271 272 /* the read-only stuff doesn't really belong here, but any other place is 273 probably as bad and I don't want to create yet another include file. */ 274 275 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */ 276 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */ 277 #define BLKRRPART _IO(0x12,95) /* re-read partition table */ 278 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */ 279 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */ 280 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */ 281 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */ 282 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */ 283 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */ 284 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */ 285 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */ 286 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */ 287 #if 0 288 #define BLKPG _IO(0x12,105)/* See blkpg.h */ 289 290 /* Some people are morons. Do not use sizeof! */ 291 292 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */ 293 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */ 294 /* This was here just to show that the number is taken - 295 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */ 296 #endif 297 /* A jump here: 108-111 have been used for various private purposes. */ 298 #define BLKBSZGET _IOR(0x12,112,size_t) 299 #define BLKBSZSET _IOW(0x12,113,size_t) 300 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */ 301 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup) 302 #define BLKTRACESTART _IO(0x12,116) 303 #define BLKTRACESTOP _IO(0x12,117) 304 #define BLKTRACETEARDOWN _IO(0x12,118) 305 #define BLKDISCARD _IO(0x12,119) 306 #define BLKIOMIN _IO(0x12,120) 307 #define BLKIOOPT _IO(0x12,121) 308 #define BLKALIGNOFF _IO(0x12,122) 309 #define BLKPBSZGET _IO(0x12,123) 310 311 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */ 312 #define FIBMAP _IO(0x00,1) /* bmap access */ 313 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */ 314 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */ 315 #define FITHAW _IOWR('X', 120, int) /* Thaw */ 316 317 #define FS_IOC_GETFLAGS _IOR('f', 1, long) 318 #define FS_IOC_SETFLAGS _IOW('f', 2, long) 319 #define FS_IOC_GETVERSION _IOR('v', 1, long) 320 #define FS_IOC_SETVERSION _IOW('v', 2, long) 321 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap) 322 #define FS_IOC32_GETFLAGS _IOR('f', 1, int) 323 #define FS_IOC32_SETFLAGS _IOW('f', 2, int) 324 #define FS_IOC32_GETVERSION _IOR('v', 1, int) 325 #define FS_IOC32_SETVERSION _IOW('v', 2, int) 326 327 /* 328 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS) 329 */ 330 #define FS_SECRM_FL 0x00000001 /* Secure deletion */ 331 #define FS_UNRM_FL 0x00000002 /* Undelete */ 332 #define FS_COMPR_FL 0x00000004 /* Compress file */ 333 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */ 334 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 335 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */ 336 #define FS_NODUMP_FL 0x00000040 /* do not dump file */ 337 #define FS_NOATIME_FL 0x00000080 /* do not update atime */ 338 /* Reserved for compression usage... */ 339 #define FS_DIRTY_FL 0x00000100 340 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 341 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */ 342 #define FS_ECOMPR_FL 0x00000800 /* Compression error */ 343 /* End compression flags --- maybe not all used */ 344 #define FS_BTREE_FL 0x00001000 /* btree format dir */ 345 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 346 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */ 347 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */ 348 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 349 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 350 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 351 #define FS_EXTENT_FL 0x00080000 /* Extents */ 352 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */ 353 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */ 354 355 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */ 356 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */ 357 358 359 #define SYNC_FILE_RANGE_WAIT_BEFORE 1 360 #define SYNC_FILE_RANGE_WRITE 2 361 #define SYNC_FILE_RANGE_WAIT_AFTER 4 362 363 #ifdef __KERNEL__ 364 365 #include <linux/linkage.h> 366 #include <linux/wait.h> 367 #include <linux/types.h> 368 #include <linux/kdev_t.h> 369 #include <linux/dcache.h> 370 #include <linux/path.h> 371 #include <linux/stat.h> 372 #include <linux/cache.h> 373 #include <linux/kobject.h> 374 #include <linux/list.h> 375 #include <linux/radix-tree.h> 376 #include <linux/prio_tree.h> 377 #include <linux/init.h> 378 #include <linux/pid.h> 379 #include <linux/mutex.h> 380 #include <linux/capability.h> 381 #include <linux/semaphore.h> 382 #include <linux/fiemap.h> 383 384 #include <asm/atomic.h> 385 #include <asm/byteorder.h> 386 387 struct export_operations; 388 struct hd_geometry; 389 struct iovec; 390 struct nameidata; 391 struct kiocb; 392 struct pipe_inode_info; 393 struct poll_table_struct; 394 struct kstatfs; 395 struct vm_area_struct; 396 struct vfsmount; 397 struct cred; 398 399 extern void __init inode_init(void); 400 extern void __init inode_init_early(void); 401 extern void __init files_init(unsigned long); 402 403 extern struct files_stat_struct files_stat; 404 extern int get_max_files(void); 405 extern int sysctl_nr_open; 406 extern struct inodes_stat_t inodes_stat; 407 extern int leases_enable, lease_break_time; 408 #ifdef CONFIG_DNOTIFY 409 extern int dir_notify_enable; 410 #endif 411 412 struct buffer_head; 413 typedef int (get_block_t)(struct inode *inode, sector_t iblock, 414 struct buffer_head *bh_result, int create); 415 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset, 416 ssize_t bytes, void *private); 417 418 /* 419 * Attribute flags. These should be or-ed together to figure out what 420 * has been changed! 421 */ 422 #define ATTR_MODE (1 << 0) 423 #define ATTR_UID (1 << 1) 424 #define ATTR_GID (1 << 2) 425 #define ATTR_SIZE (1 << 3) 426 #define ATTR_ATIME (1 << 4) 427 #define ATTR_MTIME (1 << 5) 428 #define ATTR_CTIME (1 << 6) 429 #define ATTR_ATIME_SET (1 << 7) 430 #define ATTR_MTIME_SET (1 << 8) 431 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ 432 #define ATTR_ATTR_FLAG (1 << 10) 433 #define ATTR_KILL_SUID (1 << 11) 434 #define ATTR_KILL_SGID (1 << 12) 435 #define ATTR_FILE (1 << 13) 436 #define ATTR_KILL_PRIV (1 << 14) 437 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ 438 #define ATTR_TIMES_SET (1 << 16) 439 440 /* 441 * This is the Inode Attributes structure, used for notify_change(). It 442 * uses the above definitions as flags, to know which values have changed. 443 * Also, in this manner, a Filesystem can look at only the values it cares 444 * about. Basically, these are the attributes that the VFS layer can 445 * request to change from the FS layer. 446 * 447 * Derek Atkins <warlord@MIT.EDU> 94-10-20 448 */ 449 struct iattr { 450 unsigned int ia_valid; 451 umode_t ia_mode; 452 uid_t ia_uid; 453 gid_t ia_gid; 454 loff_t ia_size; 455 struct timespec ia_atime; 456 struct timespec ia_mtime; 457 struct timespec ia_ctime; 458 459 /* 460 * Not an attribute, but an auxilary info for filesystems wanting to 461 * implement an ftruncate() like method. NOTE: filesystem should 462 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). 463 */ 464 struct file *ia_file; 465 }; 466 467 /* 468 * Includes for diskquotas. 469 */ 470 #include <linux/quota.h> 471 472 /** 473 * enum positive_aop_returns - aop return codes with specific semantics 474 * 475 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has 476 * completed, that the page is still locked, and 477 * should be considered active. The VM uses this hint 478 * to return the page to the active list -- it won't 479 * be a candidate for writeback again in the near 480 * future. Other callers must be careful to unlock 481 * the page if they get this return. Returned by 482 * writepage(); 483 * 484 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has 485 * unlocked it and the page might have been truncated. 486 * The caller should back up to acquiring a new page and 487 * trying again. The aop will be taking reasonable 488 * precautions not to livelock. If the caller held a page 489 * reference, it should drop it before retrying. Returned 490 * by readpage(). 491 * 492 * address_space_operation functions return these large constants to indicate 493 * special semantics to the caller. These are much larger than the bytes in a 494 * page to allow for functions that return the number of bytes operated on in a 495 * given page. 496 */ 497 498 enum positive_aop_returns { 499 AOP_WRITEPAGE_ACTIVATE = 0x80000, 500 AOP_TRUNCATED_PAGE = 0x80001, 501 }; 502 503 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */ 504 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */ 505 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct 506 * helper code (eg buffer layer) 507 * to clear GFP_FS from alloc */ 508 509 /* 510 * oh the beauties of C type declarations. 511 */ 512 struct page; 513 struct address_space; 514 struct writeback_control; 515 516 struct iov_iter { 517 const struct iovec *iov; 518 unsigned long nr_segs; 519 size_t iov_offset; 520 size_t count; 521 }; 522 523 size_t iov_iter_copy_from_user_atomic(struct page *page, 524 struct iov_iter *i, unsigned long offset, size_t bytes); 525 size_t iov_iter_copy_from_user(struct page *page, 526 struct iov_iter *i, unsigned long offset, size_t bytes); 527 void iov_iter_advance(struct iov_iter *i, size_t bytes); 528 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes); 529 size_t iov_iter_single_seg_count(struct iov_iter *i); 530 531 static inline void iov_iter_init(struct iov_iter *i, 532 const struct iovec *iov, unsigned long nr_segs, 533 size_t count, size_t written) 534 { 535 i->iov = iov; 536 i->nr_segs = nr_segs; 537 i->iov_offset = 0; 538 i->count = count + written; 539 540 iov_iter_advance(i, written); 541 } 542 543 static inline size_t iov_iter_count(struct iov_iter *i) 544 { 545 return i->count; 546 } 547 548 /* 549 * "descriptor" for what we're up to with a read. 550 * This allows us to use the same read code yet 551 * have multiple different users of the data that 552 * we read from a file. 553 * 554 * The simplest case just copies the data to user 555 * mode. 556 */ 557 typedef struct { 558 size_t written; 559 size_t count; 560 union { 561 char __user *buf; 562 void *data; 563 } arg; 564 int error; 565 } read_descriptor_t; 566 567 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, 568 unsigned long, unsigned long); 569 570 struct address_space_operations { 571 int (*writepage)(struct page *page, struct writeback_control *wbc); 572 int (*readpage)(struct file *, struct page *); 573 void (*sync_page)(struct page *); 574 575 /* Write back some dirty pages from this mapping. */ 576 int (*writepages)(struct address_space *, struct writeback_control *); 577 578 /* Set a page dirty. Return true if this dirtied it */ 579 int (*set_page_dirty)(struct page *page); 580 581 int (*readpages)(struct file *filp, struct address_space *mapping, 582 struct list_head *pages, unsigned nr_pages); 583 584 int (*write_begin)(struct file *, struct address_space *mapping, 585 loff_t pos, unsigned len, unsigned flags, 586 struct page **pagep, void **fsdata); 587 int (*write_end)(struct file *, struct address_space *mapping, 588 loff_t pos, unsigned len, unsigned copied, 589 struct page *page, void *fsdata); 590 591 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ 592 sector_t (*bmap)(struct address_space *, sector_t); 593 void (*invalidatepage) (struct page *, unsigned long); 594 int (*releasepage) (struct page *, gfp_t); 595 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov, 596 loff_t offset, unsigned long nr_segs); 597 int (*get_xip_mem)(struct address_space *, pgoff_t, int, 598 void **, unsigned long *); 599 /* migrate the contents of a page to the specified target */ 600 int (*migratepage) (struct address_space *, 601 struct page *, struct page *); 602 int (*launder_page) (struct page *); 603 int (*is_partially_uptodate) (struct page *, read_descriptor_t *, 604 unsigned long); 605 int (*error_remove_page)(struct address_space *, struct page *); 606 }; 607 608 /* 609 * pagecache_write_begin/pagecache_write_end must be used by general code 610 * to write into the pagecache. 611 */ 612 int pagecache_write_begin(struct file *, struct address_space *mapping, 613 loff_t pos, unsigned len, unsigned flags, 614 struct page **pagep, void **fsdata); 615 616 int pagecache_write_end(struct file *, struct address_space *mapping, 617 loff_t pos, unsigned len, unsigned copied, 618 struct page *page, void *fsdata); 619 620 struct backing_dev_info; 621 struct address_space { 622 struct inode *host; /* owner: inode, block_device */ 623 struct radix_tree_root page_tree; /* radix tree of all pages */ 624 spinlock_t tree_lock; /* and lock protecting it */ 625 unsigned int i_mmap_writable;/* count VM_SHARED mappings */ 626 struct prio_tree_root i_mmap; /* tree of private and shared mappings */ 627 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */ 628 spinlock_t i_mmap_lock; /* protect tree, count, list */ 629 unsigned int truncate_count; /* Cover race condition with truncate */ 630 unsigned long nrpages; /* number of total pages */ 631 pgoff_t writeback_index;/* writeback starts here */ 632 const struct address_space_operations *a_ops; /* methods */ 633 unsigned long flags; /* error bits/gfp mask */ 634 struct backing_dev_info *backing_dev_info; /* device readahead, etc */ 635 spinlock_t private_lock; /* for use by the address_space */ 636 struct list_head private_list; /* ditto */ 637 struct address_space *assoc_mapping; /* ditto */ 638 } __attribute__((aligned(sizeof(long)))); 639 /* 640 * On most architectures that alignment is already the case; but 641 * must be enforced here for CRIS, to let the least signficant bit 642 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. 643 */ 644 645 struct block_device { 646 dev_t bd_dev; /* not a kdev_t - it's a search key */ 647 struct inode * bd_inode; /* will die */ 648 struct super_block * bd_super; 649 int bd_openers; 650 struct mutex bd_mutex; /* open/close mutex */ 651 struct list_head bd_inodes; 652 void * bd_holder; 653 int bd_holders; 654 #ifdef CONFIG_SYSFS 655 struct list_head bd_holder_list; 656 #endif 657 struct block_device * bd_contains; 658 unsigned bd_block_size; 659 struct hd_struct * bd_part; 660 /* number of times partitions within this device have been opened. */ 661 unsigned bd_part_count; 662 int bd_invalidated; 663 struct gendisk * bd_disk; 664 struct list_head bd_list; 665 /* 666 * Private data. You must have bd_claim'ed the block_device 667 * to use this. NOTE: bd_claim allows an owner to claim 668 * the same device multiple times, the owner must take special 669 * care to not mess up bd_private for that case. 670 */ 671 unsigned long bd_private; 672 673 /* The counter of freeze processes */ 674 int bd_fsfreeze_count; 675 /* Mutex for freeze */ 676 struct mutex bd_fsfreeze_mutex; 677 }; 678 679 /* 680 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache 681 * radix trees 682 */ 683 #define PAGECACHE_TAG_DIRTY 0 684 #define PAGECACHE_TAG_WRITEBACK 1 685 686 int mapping_tagged(struct address_space *mapping, int tag); 687 688 /* 689 * Might pages of this file be mapped into userspace? 690 */ 691 static inline int mapping_mapped(struct address_space *mapping) 692 { 693 return !prio_tree_empty(&mapping->i_mmap) || 694 !list_empty(&mapping->i_mmap_nonlinear); 695 } 696 697 /* 698 * Might pages of this file have been modified in userspace? 699 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff 700 * marks vma as VM_SHARED if it is shared, and the file was opened for 701 * writing i.e. vma may be mprotected writable even if now readonly. 702 */ 703 static inline int mapping_writably_mapped(struct address_space *mapping) 704 { 705 return mapping->i_mmap_writable != 0; 706 } 707 708 /* 709 * Use sequence counter to get consistent i_size on 32-bit processors. 710 */ 711 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 712 #include <linux/seqlock.h> 713 #define __NEED_I_SIZE_ORDERED 714 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) 715 #else 716 #define i_size_ordered_init(inode) do { } while (0) 717 #endif 718 719 struct posix_acl; 720 #define ACL_NOT_CACHED ((void *)(-1)) 721 722 struct inode { 723 struct hlist_node i_hash; 724 struct list_head i_list; /* backing dev IO list */ 725 struct list_head i_sb_list; 726 struct list_head i_dentry; 727 unsigned long i_ino; 728 atomic_t i_count; 729 unsigned int i_nlink; 730 uid_t i_uid; 731 gid_t i_gid; 732 dev_t i_rdev; 733 u64 i_version; 734 loff_t i_size; 735 #ifdef __NEED_I_SIZE_ORDERED 736 seqcount_t i_size_seqcount; 737 #endif 738 struct timespec i_atime; 739 struct timespec i_mtime; 740 struct timespec i_ctime; 741 blkcnt_t i_blocks; 742 unsigned int i_blkbits; 743 unsigned short i_bytes; 744 umode_t i_mode; 745 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ 746 struct mutex i_mutex; 747 struct rw_semaphore i_alloc_sem; 748 const struct inode_operations *i_op; 749 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ 750 struct super_block *i_sb; 751 struct file_lock *i_flock; 752 struct address_space *i_mapping; 753 struct address_space i_data; 754 #ifdef CONFIG_QUOTA 755 struct dquot *i_dquot[MAXQUOTAS]; 756 #endif 757 struct list_head i_devices; 758 union { 759 struct pipe_inode_info *i_pipe; 760 struct block_device *i_bdev; 761 struct cdev *i_cdev; 762 }; 763 764 __u32 i_generation; 765 766 #ifdef CONFIG_FSNOTIFY 767 __u32 i_fsnotify_mask; /* all events this inode cares about */ 768 struct hlist_head i_fsnotify_mark_entries; /* fsnotify mark entries */ 769 #endif 770 771 #ifdef CONFIG_INOTIFY 772 struct list_head inotify_watches; /* watches on this inode */ 773 struct mutex inotify_mutex; /* protects the watches list */ 774 #endif 775 776 unsigned long i_state; 777 unsigned long dirtied_when; /* jiffies of first dirtying */ 778 779 unsigned int i_flags; 780 781 atomic_t i_writecount; 782 #ifdef CONFIG_SECURITY 783 void *i_security; 784 #endif 785 #ifdef CONFIG_FS_POSIX_ACL 786 struct posix_acl *i_acl; 787 struct posix_acl *i_default_acl; 788 #endif 789 void *i_private; /* fs or device private pointer */ 790 }; 791 792 /* 793 * inode->i_mutex nesting subclasses for the lock validator: 794 * 795 * 0: the object of the current VFS operation 796 * 1: parent 797 * 2: child/target 798 * 3: quota file 799 * 800 * The locking order between these classes is 801 * parent -> child -> normal -> xattr -> quota 802 */ 803 enum inode_i_mutex_lock_class 804 { 805 I_MUTEX_NORMAL, 806 I_MUTEX_PARENT, 807 I_MUTEX_CHILD, 808 I_MUTEX_XATTR, 809 I_MUTEX_QUOTA 810 }; 811 812 /* 813 * NOTE: in a 32bit arch with a preemptable kernel and 814 * an UP compile the i_size_read/write must be atomic 815 * with respect to the local cpu (unlike with preempt disabled), 816 * but they don't need to be atomic with respect to other cpus like in 817 * true SMP (so they need either to either locally disable irq around 818 * the read or for example on x86 they can be still implemented as a 819 * cmpxchg8b without the need of the lock prefix). For SMP compiles 820 * and 64bit archs it makes no difference if preempt is enabled or not. 821 */ 822 static inline loff_t i_size_read(const struct inode *inode) 823 { 824 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 825 loff_t i_size; 826 unsigned int seq; 827 828 do { 829 seq = read_seqcount_begin(&inode->i_size_seqcount); 830 i_size = inode->i_size; 831 } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); 832 return i_size; 833 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 834 loff_t i_size; 835 836 preempt_disable(); 837 i_size = inode->i_size; 838 preempt_enable(); 839 return i_size; 840 #else 841 return inode->i_size; 842 #endif 843 } 844 845 /* 846 * NOTE: unlike i_size_read(), i_size_write() does need locking around it 847 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount 848 * can be lost, resulting in subsequent i_size_read() calls spinning forever. 849 */ 850 static inline void i_size_write(struct inode *inode, loff_t i_size) 851 { 852 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 853 write_seqcount_begin(&inode->i_size_seqcount); 854 inode->i_size = i_size; 855 write_seqcount_end(&inode->i_size_seqcount); 856 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 857 preempt_disable(); 858 inode->i_size = i_size; 859 preempt_enable(); 860 #else 861 inode->i_size = i_size; 862 #endif 863 } 864 865 static inline unsigned iminor(const struct inode *inode) 866 { 867 return MINOR(inode->i_rdev); 868 } 869 870 static inline unsigned imajor(const struct inode *inode) 871 { 872 return MAJOR(inode->i_rdev); 873 } 874 875 extern struct block_device *I_BDEV(struct inode *inode); 876 877 struct fown_struct { 878 rwlock_t lock; /* protects pid, uid, euid fields */ 879 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ 880 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ 881 uid_t uid, euid; /* uid/euid of process setting the owner */ 882 int signum; /* posix.1b rt signal to be delivered on IO */ 883 }; 884 885 /* 886 * Track a single file's readahead state 887 */ 888 struct file_ra_state { 889 pgoff_t start; /* where readahead started */ 890 unsigned int size; /* # of readahead pages */ 891 unsigned int async_size; /* do asynchronous readahead when 892 there are only # of pages ahead */ 893 894 unsigned int ra_pages; /* Maximum readahead window */ 895 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */ 896 loff_t prev_pos; /* Cache last read() position */ 897 }; 898 899 /* 900 * Check if @index falls in the readahead windows. 901 */ 902 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) 903 { 904 return (index >= ra->start && 905 index < ra->start + ra->size); 906 } 907 908 #define FILE_MNT_WRITE_TAKEN 1 909 #define FILE_MNT_WRITE_RELEASED 2 910 911 struct file { 912 /* 913 * fu_list becomes invalid after file_free is called and queued via 914 * fu_rcuhead for RCU freeing 915 */ 916 union { 917 struct list_head fu_list; 918 struct rcu_head fu_rcuhead; 919 } f_u; 920 struct path f_path; 921 #define f_dentry f_path.dentry 922 #define f_vfsmnt f_path.mnt 923 const struct file_operations *f_op; 924 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */ 925 atomic_long_t f_count; 926 unsigned int f_flags; 927 fmode_t f_mode; 928 loff_t f_pos; 929 struct fown_struct f_owner; 930 const struct cred *f_cred; 931 struct file_ra_state f_ra; 932 933 u64 f_version; 934 #ifdef CONFIG_SECURITY 935 void *f_security; 936 #endif 937 /* needed for tty driver, and maybe others */ 938 void *private_data; 939 940 #ifdef CONFIG_EPOLL 941 /* Used by fs/eventpoll.c to link all the hooks to this file */ 942 struct list_head f_ep_links; 943 #endif /* #ifdef CONFIG_EPOLL */ 944 struct address_space *f_mapping; 945 #ifdef CONFIG_DEBUG_WRITECOUNT 946 unsigned long f_mnt_write_state; 947 #endif 948 }; 949 extern spinlock_t files_lock; 950 #define file_list_lock() spin_lock(&files_lock); 951 #define file_list_unlock() spin_unlock(&files_lock); 952 953 #define get_file(x) atomic_long_inc(&(x)->f_count) 954 #define file_count(x) atomic_long_read(&(x)->f_count) 955 956 #ifdef CONFIG_DEBUG_WRITECOUNT 957 static inline void file_take_write(struct file *f) 958 { 959 WARN_ON(f->f_mnt_write_state != 0); 960 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN; 961 } 962 static inline void file_release_write(struct file *f) 963 { 964 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED; 965 } 966 static inline void file_reset_write(struct file *f) 967 { 968 f->f_mnt_write_state = 0; 969 } 970 static inline void file_check_state(struct file *f) 971 { 972 /* 973 * At this point, either both or neither of these bits 974 * should be set. 975 */ 976 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN); 977 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED); 978 } 979 static inline int file_check_writeable(struct file *f) 980 { 981 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN) 982 return 0; 983 printk(KERN_WARNING "writeable file with no " 984 "mnt_want_write()\n"); 985 WARN_ON(1); 986 return -EINVAL; 987 } 988 #else /* !CONFIG_DEBUG_WRITECOUNT */ 989 static inline void file_take_write(struct file *filp) {} 990 static inline void file_release_write(struct file *filp) {} 991 static inline void file_reset_write(struct file *filp) {} 992 static inline void file_check_state(struct file *filp) {} 993 static inline int file_check_writeable(struct file *filp) 994 { 995 return 0; 996 } 997 #endif /* CONFIG_DEBUG_WRITECOUNT */ 998 999 #define MAX_NON_LFS ((1UL<<31) - 1) 1000 1001 /* Page cache limit. The filesystems should put that into their s_maxbytes 1002 limits, otherwise bad things can happen in VM. */ 1003 #if BITS_PER_LONG==32 1004 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 1005 #elif BITS_PER_LONG==64 1006 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL 1007 #endif 1008 1009 #define FL_POSIX 1 1010 #define FL_FLOCK 2 1011 #define FL_ACCESS 8 /* not trying to lock, just looking */ 1012 #define FL_EXISTS 16 /* when unlocking, test for existence */ 1013 #define FL_LEASE 32 /* lease held on this file */ 1014 #define FL_CLOSE 64 /* unlock on close */ 1015 #define FL_SLEEP 128 /* A blocking lock */ 1016 1017 /* 1018 * Special return value from posix_lock_file() and vfs_lock_file() for 1019 * asynchronous locking. 1020 */ 1021 #define FILE_LOCK_DEFERRED 1 1022 1023 /* 1024 * The POSIX file lock owner is determined by 1025 * the "struct files_struct" in the thread group 1026 * (or NULL for no owner - BSD locks). 1027 * 1028 * Lockd stuffs a "host" pointer into this. 1029 */ 1030 typedef struct files_struct *fl_owner_t; 1031 1032 struct file_lock_operations { 1033 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 1034 void (*fl_release_private)(struct file_lock *); 1035 }; 1036 1037 struct lock_manager_operations { 1038 int (*fl_compare_owner)(struct file_lock *, struct file_lock *); 1039 void (*fl_notify)(struct file_lock *); /* unblock callback */ 1040 int (*fl_grant)(struct file_lock *, struct file_lock *, int); 1041 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 1042 void (*fl_release_private)(struct file_lock *); 1043 void (*fl_break)(struct file_lock *); 1044 int (*fl_mylease)(struct file_lock *, struct file_lock *); 1045 int (*fl_change)(struct file_lock **, int); 1046 }; 1047 1048 struct lock_manager { 1049 struct list_head list; 1050 }; 1051 1052 void locks_start_grace(struct lock_manager *); 1053 void locks_end_grace(struct lock_manager *); 1054 int locks_in_grace(void); 1055 1056 /* that will die - we need it for nfs_lock_info */ 1057 #include <linux/nfs_fs_i.h> 1058 1059 struct file_lock { 1060 struct file_lock *fl_next; /* singly linked list for this inode */ 1061 struct list_head fl_link; /* doubly linked list of all locks */ 1062 struct list_head fl_block; /* circular list of blocked processes */ 1063 fl_owner_t fl_owner; 1064 unsigned char fl_flags; 1065 unsigned char fl_type; 1066 unsigned int fl_pid; 1067 struct pid *fl_nspid; 1068 wait_queue_head_t fl_wait; 1069 struct file *fl_file; 1070 loff_t fl_start; 1071 loff_t fl_end; 1072 1073 struct fasync_struct * fl_fasync; /* for lease break notifications */ 1074 unsigned long fl_break_time; /* for nonblocking lease breaks */ 1075 1076 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ 1077 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ 1078 union { 1079 struct nfs_lock_info nfs_fl; 1080 struct nfs4_lock_info nfs4_fl; 1081 struct { 1082 struct list_head link; /* link in AFS vnode's pending_locks list */ 1083 int state; /* state of grant or error if -ve */ 1084 } afs; 1085 } fl_u; 1086 }; 1087 1088 /* The following constant reflects the upper bound of the file/locking space */ 1089 #ifndef OFFSET_MAX 1090 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) 1091 #define OFFSET_MAX INT_LIMIT(loff_t) 1092 #define OFFT_OFFSET_MAX INT_LIMIT(off_t) 1093 #endif 1094 1095 #include <linux/fcntl.h> 1096 1097 extern void send_sigio(struct fown_struct *fown, int fd, int band); 1098 1099 /* fs/sync.c */ 1100 extern int do_sync_mapping_range(struct address_space *mapping, loff_t offset, 1101 loff_t endbyte, unsigned int flags); 1102 1103 #ifdef CONFIG_FILE_LOCKING 1104 extern int fcntl_getlk(struct file *, struct flock __user *); 1105 extern int fcntl_setlk(unsigned int, struct file *, unsigned int, 1106 struct flock __user *); 1107 1108 #if BITS_PER_LONG == 32 1109 extern int fcntl_getlk64(struct file *, struct flock64 __user *); 1110 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, 1111 struct flock64 __user *); 1112 #endif 1113 1114 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); 1115 extern int fcntl_getlease(struct file *filp); 1116 1117 /* fs/locks.c */ 1118 extern void locks_init_lock(struct file_lock *); 1119 extern void locks_copy_lock(struct file_lock *, struct file_lock *); 1120 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *); 1121 extern void locks_remove_posix(struct file *, fl_owner_t); 1122 extern void locks_remove_flock(struct file *); 1123 extern void locks_release_private(struct file_lock *); 1124 extern void posix_test_lock(struct file *, struct file_lock *); 1125 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *); 1126 extern int posix_lock_file_wait(struct file *, struct file_lock *); 1127 extern int posix_unblock_lock(struct file *, struct file_lock *); 1128 extern int vfs_test_lock(struct file *, struct file_lock *); 1129 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *); 1130 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl); 1131 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl); 1132 extern int __break_lease(struct inode *inode, unsigned int flags); 1133 extern void lease_get_mtime(struct inode *, struct timespec *time); 1134 extern int generic_setlease(struct file *, long, struct file_lock **); 1135 extern int vfs_setlease(struct file *, long, struct file_lock **); 1136 extern int lease_modify(struct file_lock **, int); 1137 extern int lock_may_read(struct inode *, loff_t start, unsigned long count); 1138 extern int lock_may_write(struct inode *, loff_t start, unsigned long count); 1139 #else /* !CONFIG_FILE_LOCKING */ 1140 static inline int fcntl_getlk(struct file *file, struct flock __user *user) 1141 { 1142 return -EINVAL; 1143 } 1144 1145 static inline int fcntl_setlk(unsigned int fd, struct file *file, 1146 unsigned int cmd, struct flock __user *user) 1147 { 1148 return -EACCES; 1149 } 1150 1151 #if BITS_PER_LONG == 32 1152 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user) 1153 { 1154 return -EINVAL; 1155 } 1156 1157 static inline int fcntl_setlk64(unsigned int fd, struct file *file, 1158 unsigned int cmd, struct flock64 __user *user) 1159 { 1160 return -EACCES; 1161 } 1162 #endif 1163 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg) 1164 { 1165 return 0; 1166 } 1167 1168 static inline int fcntl_getlease(struct file *filp) 1169 { 1170 return 0; 1171 } 1172 1173 static inline void locks_init_lock(struct file_lock *fl) 1174 { 1175 return; 1176 } 1177 1178 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl) 1179 { 1180 return; 1181 } 1182 1183 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl) 1184 { 1185 return; 1186 } 1187 1188 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner) 1189 { 1190 return; 1191 } 1192 1193 static inline void locks_remove_flock(struct file *filp) 1194 { 1195 return; 1196 } 1197 1198 static inline void posix_test_lock(struct file *filp, struct file_lock *fl) 1199 { 1200 return; 1201 } 1202 1203 static inline int posix_lock_file(struct file *filp, struct file_lock *fl, 1204 struct file_lock *conflock) 1205 { 1206 return -ENOLCK; 1207 } 1208 1209 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl) 1210 { 1211 return -ENOLCK; 1212 } 1213 1214 static inline int posix_unblock_lock(struct file *filp, 1215 struct file_lock *waiter) 1216 { 1217 return -ENOENT; 1218 } 1219 1220 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl) 1221 { 1222 return 0; 1223 } 1224 1225 static inline int vfs_lock_file(struct file *filp, unsigned int cmd, 1226 struct file_lock *fl, struct file_lock *conf) 1227 { 1228 return -ENOLCK; 1229 } 1230 1231 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl) 1232 { 1233 return 0; 1234 } 1235 1236 static inline int flock_lock_file_wait(struct file *filp, 1237 struct file_lock *request) 1238 { 1239 return -ENOLCK; 1240 } 1241 1242 static inline int __break_lease(struct inode *inode, unsigned int mode) 1243 { 1244 return 0; 1245 } 1246 1247 static inline void lease_get_mtime(struct inode *inode, struct timespec *time) 1248 { 1249 return; 1250 } 1251 1252 static inline int generic_setlease(struct file *filp, long arg, 1253 struct file_lock **flp) 1254 { 1255 return -EINVAL; 1256 } 1257 1258 static inline int vfs_setlease(struct file *filp, long arg, 1259 struct file_lock **lease) 1260 { 1261 return -EINVAL; 1262 } 1263 1264 static inline int lease_modify(struct file_lock **before, int arg) 1265 { 1266 return -EINVAL; 1267 } 1268 1269 static inline int lock_may_read(struct inode *inode, loff_t start, 1270 unsigned long len) 1271 { 1272 return 1; 1273 } 1274 1275 static inline int lock_may_write(struct inode *inode, loff_t start, 1276 unsigned long len) 1277 { 1278 return 1; 1279 } 1280 1281 #endif /* !CONFIG_FILE_LOCKING */ 1282 1283 1284 struct fasync_struct { 1285 int magic; 1286 int fa_fd; 1287 struct fasync_struct *fa_next; /* singly linked list */ 1288 struct file *fa_file; 1289 }; 1290 1291 #define FASYNC_MAGIC 0x4601 1292 1293 /* SMP safe fasync helpers: */ 1294 extern int fasync_helper(int, struct file *, int, struct fasync_struct **); 1295 /* can be called from interrupts */ 1296 extern void kill_fasync(struct fasync_struct **, int, int); 1297 /* only for net: no internal synchronization */ 1298 extern void __kill_fasync(struct fasync_struct *, int, int); 1299 1300 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force); 1301 extern int f_setown(struct file *filp, unsigned long arg, int force); 1302 extern void f_delown(struct file *filp); 1303 extern pid_t f_getown(struct file *filp); 1304 extern int send_sigurg(struct fown_struct *fown); 1305 1306 /* 1307 * Umount options 1308 */ 1309 1310 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ 1311 #define MNT_DETACH 0x00000002 /* Just detach from the tree */ 1312 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ 1313 1314 extern struct list_head super_blocks; 1315 extern spinlock_t sb_lock; 1316 1317 #define sb_entry(list) list_entry((list), struct super_block, s_list) 1318 #define S_BIAS (1<<30) 1319 struct super_block { 1320 struct list_head s_list; /* Keep this first */ 1321 dev_t s_dev; /* search index; _not_ kdev_t */ 1322 unsigned long s_blocksize; 1323 unsigned char s_blocksize_bits; 1324 unsigned char s_dirt; 1325 loff_t s_maxbytes; /* Max file size */ 1326 struct file_system_type *s_type; 1327 const struct super_operations *s_op; 1328 const struct dquot_operations *dq_op; 1329 const struct quotactl_ops *s_qcop; 1330 const struct export_operations *s_export_op; 1331 unsigned long s_flags; 1332 unsigned long s_magic; 1333 struct dentry *s_root; 1334 struct rw_semaphore s_umount; 1335 struct mutex s_lock; 1336 int s_count; 1337 int s_need_sync; 1338 atomic_t s_active; 1339 #ifdef CONFIG_SECURITY 1340 void *s_security; 1341 #endif 1342 struct xattr_handler **s_xattr; 1343 1344 struct list_head s_inodes; /* all inodes */ 1345 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */ 1346 struct list_head s_files; 1347 /* s_dentry_lru and s_nr_dentry_unused are protected by dcache_lock */ 1348 struct list_head s_dentry_lru; /* unused dentry lru */ 1349 int s_nr_dentry_unused; /* # of dentry on lru */ 1350 1351 struct block_device *s_bdev; 1352 struct backing_dev_info *s_bdi; 1353 struct mtd_info *s_mtd; 1354 struct list_head s_instances; 1355 struct quota_info s_dquot; /* Diskquota specific options */ 1356 1357 int s_frozen; 1358 wait_queue_head_t s_wait_unfrozen; 1359 1360 char s_id[32]; /* Informational name */ 1361 1362 void *s_fs_info; /* Filesystem private info */ 1363 fmode_t s_mode; 1364 1365 /* 1366 * The next field is for VFS *only*. No filesystems have any business 1367 * even looking at it. You had been warned. 1368 */ 1369 struct mutex s_vfs_rename_mutex; /* Kludge */ 1370 1371 /* Granularity of c/m/atime in ns. 1372 Cannot be worse than a second */ 1373 u32 s_time_gran; 1374 1375 /* 1376 * Filesystem subtype. If non-empty the filesystem type field 1377 * in /proc/mounts will be "type.subtype" 1378 */ 1379 char *s_subtype; 1380 1381 /* 1382 * Saved mount options for lazy filesystems using 1383 * generic_show_options() 1384 */ 1385 char *s_options; 1386 }; 1387 1388 extern struct timespec current_fs_time(struct super_block *sb); 1389 1390 /* 1391 * Snapshotting support. 1392 */ 1393 enum { 1394 SB_UNFROZEN = 0, 1395 SB_FREEZE_WRITE = 1, 1396 SB_FREEZE_TRANS = 2, 1397 }; 1398 1399 #define vfs_check_frozen(sb, level) \ 1400 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level))) 1401 1402 #define get_fs_excl() atomic_inc(&current->fs_excl) 1403 #define put_fs_excl() atomic_dec(&current->fs_excl) 1404 #define has_fs_excl() atomic_read(&current->fs_excl) 1405 1406 #define is_owner_or_cap(inode) \ 1407 ((current_fsuid() == (inode)->i_uid) || capable(CAP_FOWNER)) 1408 1409 /* not quite ready to be deprecated, but... */ 1410 extern void lock_super(struct super_block *); 1411 extern void unlock_super(struct super_block *); 1412 1413 /* 1414 * VFS helper functions.. 1415 */ 1416 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *); 1417 extern int vfs_mkdir(struct inode *, struct dentry *, int); 1418 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t); 1419 extern int vfs_symlink(struct inode *, struct dentry *, const char *); 1420 extern int vfs_link(struct dentry *, struct inode *, struct dentry *); 1421 extern int vfs_rmdir(struct inode *, struct dentry *); 1422 extern int vfs_unlink(struct inode *, struct dentry *); 1423 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); 1424 1425 /* 1426 * VFS dentry helper functions. 1427 */ 1428 extern void dentry_unhash(struct dentry *dentry); 1429 1430 /* 1431 * VFS file helper functions. 1432 */ 1433 extern int file_permission(struct file *, int); 1434 1435 /* 1436 * VFS FS_IOC_FIEMAP helper definitions. 1437 */ 1438 struct fiemap_extent_info { 1439 unsigned int fi_flags; /* Flags as passed from user */ 1440 unsigned int fi_extents_mapped; /* Number of mapped extents */ 1441 unsigned int fi_extents_max; /* Size of fiemap_extent array */ 1442 struct fiemap_extent *fi_extents_start; /* Start of fiemap_extent 1443 * array */ 1444 }; 1445 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical, 1446 u64 phys, u64 len, u32 flags); 1447 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags); 1448 1449 /* 1450 * File types 1451 * 1452 * NOTE! These match bits 12..15 of stat.st_mode 1453 * (ie "(i_mode >> 12) & 15"). 1454 */ 1455 #define DT_UNKNOWN 0 1456 #define DT_FIFO 1 1457 #define DT_CHR 2 1458 #define DT_DIR 4 1459 #define DT_BLK 6 1460 #define DT_REG 8 1461 #define DT_LNK 10 1462 #define DT_SOCK 12 1463 #define DT_WHT 14 1464 1465 /* 1466 * This is the "filldir" function type, used by readdir() to let 1467 * the kernel specify what kind of dirent layout it wants to have. 1468 * This allows the kernel to read directories into kernel space or 1469 * to have different dirent layouts depending on the binary type. 1470 */ 1471 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned); 1472 struct block_device_operations; 1473 1474 /* These macros are for out of kernel modules to test that 1475 * the kernel supports the unlocked_ioctl and compat_ioctl 1476 * fields in struct file_operations. */ 1477 #define HAVE_COMPAT_IOCTL 1 1478 #define HAVE_UNLOCKED_IOCTL 1 1479 1480 /* 1481 * NOTE: 1482 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl 1483 * can be called without the big kernel lock held in all filesystems. 1484 */ 1485 struct file_operations { 1486 struct module *owner; 1487 loff_t (*llseek) (struct file *, loff_t, int); 1488 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); 1489 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); 1490 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t); 1491 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t); 1492 int (*readdir) (struct file *, void *, filldir_t); 1493 unsigned int (*poll) (struct file *, struct poll_table_struct *); 1494 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long); 1495 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); 1496 long (*compat_ioctl) (struct file *, unsigned int, unsigned long); 1497 int (*mmap) (struct file *, struct vm_area_struct *); 1498 int (*open) (struct inode *, struct file *); 1499 int (*flush) (struct file *, fl_owner_t id); 1500 int (*release) (struct inode *, struct file *); 1501 int (*fsync) (struct file *, struct dentry *, int datasync); 1502 int (*aio_fsync) (struct kiocb *, int datasync); 1503 int (*fasync) (int, struct file *, int); 1504 int (*lock) (struct file *, int, struct file_lock *); 1505 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); 1506 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); 1507 int (*check_flags)(int); 1508 int (*flock) (struct file *, int, struct file_lock *); 1509 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); 1510 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); 1511 int (*setlease)(struct file *, long, struct file_lock **); 1512 }; 1513 1514 struct inode_operations { 1515 int (*create) (struct inode *,struct dentry *,int, struct nameidata *); 1516 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *); 1517 int (*link) (struct dentry *,struct inode *,struct dentry *); 1518 int (*unlink) (struct inode *,struct dentry *); 1519 int (*symlink) (struct inode *,struct dentry *,const char *); 1520 int (*mkdir) (struct inode *,struct dentry *,int); 1521 int (*rmdir) (struct inode *,struct dentry *); 1522 int (*mknod) (struct inode *,struct dentry *,int,dev_t); 1523 int (*rename) (struct inode *, struct dentry *, 1524 struct inode *, struct dentry *); 1525 int (*readlink) (struct dentry *, char __user *,int); 1526 void * (*follow_link) (struct dentry *, struct nameidata *); 1527 void (*put_link) (struct dentry *, struct nameidata *, void *); 1528 void (*truncate) (struct inode *); 1529 int (*permission) (struct inode *, int); 1530 int (*check_acl)(struct inode *, int); 1531 int (*setattr) (struct dentry *, struct iattr *); 1532 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *); 1533 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int); 1534 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t); 1535 ssize_t (*listxattr) (struct dentry *, char *, size_t); 1536 int (*removexattr) (struct dentry *, const char *); 1537 void (*truncate_range)(struct inode *, loff_t, loff_t); 1538 long (*fallocate)(struct inode *inode, int mode, loff_t offset, 1539 loff_t len); 1540 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, 1541 u64 len); 1542 }; 1543 1544 struct seq_file; 1545 1546 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector, 1547 unsigned long nr_segs, unsigned long fast_segs, 1548 struct iovec *fast_pointer, 1549 struct iovec **ret_pointer); 1550 1551 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); 1552 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); 1553 extern ssize_t vfs_readv(struct file *, const struct iovec __user *, 1554 unsigned long, loff_t *); 1555 extern ssize_t vfs_writev(struct file *, const struct iovec __user *, 1556 unsigned long, loff_t *); 1557 1558 struct super_operations { 1559 struct inode *(*alloc_inode)(struct super_block *sb); 1560 void (*destroy_inode)(struct inode *); 1561 1562 void (*dirty_inode) (struct inode *); 1563 int (*write_inode) (struct inode *, int); 1564 void (*drop_inode) (struct inode *); 1565 void (*delete_inode) (struct inode *); 1566 void (*put_super) (struct super_block *); 1567 void (*write_super) (struct super_block *); 1568 int (*sync_fs)(struct super_block *sb, int wait); 1569 int (*freeze_fs) (struct super_block *); 1570 int (*unfreeze_fs) (struct super_block *); 1571 int (*statfs) (struct dentry *, struct kstatfs *); 1572 int (*remount_fs) (struct super_block *, int *, char *); 1573 void (*clear_inode) (struct inode *); 1574 void (*umount_begin) (struct super_block *); 1575 1576 int (*show_options)(struct seq_file *, struct vfsmount *); 1577 int (*show_stats)(struct seq_file *, struct vfsmount *); 1578 #ifdef CONFIG_QUOTA 1579 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); 1580 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); 1581 #endif 1582 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t); 1583 }; 1584 1585 /* 1586 * Inode state bits. Protected by inode_lock. 1587 * 1588 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC, 1589 * I_DIRTY_DATASYNC and I_DIRTY_PAGES. 1590 * 1591 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, 1592 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at 1593 * various stages of removing an inode. 1594 * 1595 * Two bits are used for locking and completion notification, I_LOCK and I_SYNC. 1596 * 1597 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on 1598 * fdatasync(). i_atime is the usual cause. 1599 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of 1600 * these changes separately from I_DIRTY_SYNC so that we 1601 * don't have to write inode on fdatasync() when only 1602 * mtime has changed in it. 1603 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. 1604 * I_NEW get_new_inode() sets i_state to I_LOCK|I_NEW. Both 1605 * are cleared by unlock_new_inode(), called from iget(). 1606 * I_WILL_FREE Must be set when calling write_inode_now() if i_count 1607 * is zero. I_FREEING must be set when I_WILL_FREE is 1608 * cleared. 1609 * I_FREEING Set when inode is about to be freed but still has dirty 1610 * pages or buffers attached or the inode itself is still 1611 * dirty. 1612 * I_CLEAR Set by clear_inode(). In this state the inode is clean 1613 * and can be destroyed. 1614 * 1615 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are 1616 * prohibited for many purposes. iget() must wait for 1617 * the inode to be completely released, then create it 1618 * anew. Other functions will just ignore such inodes, 1619 * if appropriate. I_LOCK is used for waiting. 1620 * 1621 * I_LOCK Serves as both a mutex and completion notification. 1622 * New inodes set I_LOCK. If two processes both create 1623 * the same inode, one of them will release its inode and 1624 * wait for I_LOCK to be released before returning. 1625 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can 1626 * also cause waiting on I_LOCK, without I_LOCK actually 1627 * being set. find_inode() uses this to prevent returning 1628 * nearly-dead inodes. 1629 * I_SYNC Similar to I_LOCK, but limited in scope to writeback 1630 * of inode dirty data. Having a separate lock for this 1631 * purpose reduces latency and prevents some filesystem- 1632 * specific deadlocks. 1633 * 1634 * Q: What is the difference between I_WILL_FREE and I_FREEING? 1635 * Q: igrab() only checks on (I_FREEING|I_WILL_FREE). Should it also check on 1636 * I_CLEAR? If not, why? 1637 */ 1638 #define I_DIRTY_SYNC 1 1639 #define I_DIRTY_DATASYNC 2 1640 #define I_DIRTY_PAGES 4 1641 #define I_NEW 8 1642 #define I_WILL_FREE 16 1643 #define I_FREEING 32 1644 #define I_CLEAR 64 1645 #define __I_LOCK 7 1646 #define I_LOCK (1 << __I_LOCK) 1647 #define __I_SYNC 8 1648 #define I_SYNC (1 << __I_SYNC) 1649 1650 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) 1651 1652 extern void __mark_inode_dirty(struct inode *, int); 1653 static inline void mark_inode_dirty(struct inode *inode) 1654 { 1655 __mark_inode_dirty(inode, I_DIRTY); 1656 } 1657 1658 static inline void mark_inode_dirty_sync(struct inode *inode) 1659 { 1660 __mark_inode_dirty(inode, I_DIRTY_SYNC); 1661 } 1662 1663 /** 1664 * inc_nlink - directly increment an inode's link count 1665 * @inode: inode 1666 * 1667 * This is a low-level filesystem helper to replace any 1668 * direct filesystem manipulation of i_nlink. Currently, 1669 * it is only here for parity with dec_nlink(). 1670 */ 1671 static inline void inc_nlink(struct inode *inode) 1672 { 1673 inode->i_nlink++; 1674 } 1675 1676 static inline void inode_inc_link_count(struct inode *inode) 1677 { 1678 inc_nlink(inode); 1679 mark_inode_dirty(inode); 1680 } 1681 1682 /** 1683 * drop_nlink - directly drop an inode's link count 1684 * @inode: inode 1685 * 1686 * This is a low-level filesystem helper to replace any 1687 * direct filesystem manipulation of i_nlink. In cases 1688 * where we are attempting to track writes to the 1689 * filesystem, a decrement to zero means an imminent 1690 * write when the file is truncated and actually unlinked 1691 * on the filesystem. 1692 */ 1693 static inline void drop_nlink(struct inode *inode) 1694 { 1695 inode->i_nlink--; 1696 } 1697 1698 /** 1699 * clear_nlink - directly zero an inode's link count 1700 * @inode: inode 1701 * 1702 * This is a low-level filesystem helper to replace any 1703 * direct filesystem manipulation of i_nlink. See 1704 * drop_nlink() for why we care about i_nlink hitting zero. 1705 */ 1706 static inline void clear_nlink(struct inode *inode) 1707 { 1708 inode->i_nlink = 0; 1709 } 1710 1711 static inline void inode_dec_link_count(struct inode *inode) 1712 { 1713 drop_nlink(inode); 1714 mark_inode_dirty(inode); 1715 } 1716 1717 /** 1718 * inode_inc_iversion - increments i_version 1719 * @inode: inode that need to be updated 1720 * 1721 * Every time the inode is modified, the i_version field will be incremented. 1722 * The filesystem has to be mounted with i_version flag 1723 */ 1724 1725 static inline void inode_inc_iversion(struct inode *inode) 1726 { 1727 spin_lock(&inode->i_lock); 1728 inode->i_version++; 1729 spin_unlock(&inode->i_lock); 1730 } 1731 1732 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry); 1733 static inline void file_accessed(struct file *file) 1734 { 1735 if (!(file->f_flags & O_NOATIME)) 1736 touch_atime(file->f_path.mnt, file->f_path.dentry); 1737 } 1738 1739 int sync_inode(struct inode *inode, struct writeback_control *wbc); 1740 1741 struct file_system_type { 1742 const char *name; 1743 int fs_flags; 1744 int (*get_sb) (struct file_system_type *, int, 1745 const char *, void *, struct vfsmount *); 1746 void (*kill_sb) (struct super_block *); 1747 struct module *owner; 1748 struct file_system_type * next; 1749 struct list_head fs_supers; 1750 1751 struct lock_class_key s_lock_key; 1752 struct lock_class_key s_umount_key; 1753 1754 struct lock_class_key i_lock_key; 1755 struct lock_class_key i_mutex_key; 1756 struct lock_class_key i_mutex_dir_key; 1757 struct lock_class_key i_alloc_sem_key; 1758 }; 1759 1760 extern int get_sb_ns(struct file_system_type *fs_type, int flags, void *data, 1761 int (*fill_super)(struct super_block *, void *, int), 1762 struct vfsmount *mnt); 1763 extern int get_sb_bdev(struct file_system_type *fs_type, 1764 int flags, const char *dev_name, void *data, 1765 int (*fill_super)(struct super_block *, void *, int), 1766 struct vfsmount *mnt); 1767 extern int get_sb_single(struct file_system_type *fs_type, 1768 int flags, void *data, 1769 int (*fill_super)(struct super_block *, void *, int), 1770 struct vfsmount *mnt); 1771 extern int get_sb_nodev(struct file_system_type *fs_type, 1772 int flags, void *data, 1773 int (*fill_super)(struct super_block *, void *, int), 1774 struct vfsmount *mnt); 1775 void generic_shutdown_super(struct super_block *sb); 1776 void kill_block_super(struct super_block *sb); 1777 void kill_anon_super(struct super_block *sb); 1778 void kill_litter_super(struct super_block *sb); 1779 void deactivate_super(struct super_block *sb); 1780 void deactivate_locked_super(struct super_block *sb); 1781 int set_anon_super(struct super_block *s, void *data); 1782 struct super_block *sget(struct file_system_type *type, 1783 int (*test)(struct super_block *,void *), 1784 int (*set)(struct super_block *,void *), 1785 void *data); 1786 extern int get_sb_pseudo(struct file_system_type *, char *, 1787 const struct super_operations *ops, unsigned long, 1788 struct vfsmount *mnt); 1789 extern void simple_set_mnt(struct vfsmount *mnt, struct super_block *sb); 1790 int __put_super_and_need_restart(struct super_block *sb); 1791 void put_super(struct super_block *sb); 1792 1793 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ 1794 #define fops_get(fops) \ 1795 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) 1796 #define fops_put(fops) \ 1797 do { if (fops) module_put((fops)->owner); } while(0) 1798 1799 extern int register_filesystem(struct file_system_type *); 1800 extern int unregister_filesystem(struct file_system_type *); 1801 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data); 1802 #define kern_mount(type) kern_mount_data(type, NULL) 1803 extern int may_umount_tree(struct vfsmount *); 1804 extern int may_umount(struct vfsmount *); 1805 extern long do_mount(char *, char *, char *, unsigned long, void *); 1806 extern struct vfsmount *collect_mounts(struct path *); 1807 extern void drop_collected_mounts(struct vfsmount *); 1808 1809 extern int vfs_statfs(struct dentry *, struct kstatfs *); 1810 1811 extern int current_umask(void); 1812 1813 /* /sys/fs */ 1814 extern struct kobject *fs_kobj; 1815 1816 extern int rw_verify_area(int, struct file *, loff_t *, size_t); 1817 1818 #define FLOCK_VERIFY_READ 1 1819 #define FLOCK_VERIFY_WRITE 2 1820 1821 #ifdef CONFIG_FILE_LOCKING 1822 extern int locks_mandatory_locked(struct inode *); 1823 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t); 1824 1825 /* 1826 * Candidates for mandatory locking have the setgid bit set 1827 * but no group execute bit - an otherwise meaningless combination. 1828 */ 1829 1830 static inline int __mandatory_lock(struct inode *ino) 1831 { 1832 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; 1833 } 1834 1835 /* 1836 * ... and these candidates should be on MS_MANDLOCK mounted fs, 1837 * otherwise these will be advisory locks 1838 */ 1839 1840 static inline int mandatory_lock(struct inode *ino) 1841 { 1842 return IS_MANDLOCK(ino) && __mandatory_lock(ino); 1843 } 1844 1845 static inline int locks_verify_locked(struct inode *inode) 1846 { 1847 if (mandatory_lock(inode)) 1848 return locks_mandatory_locked(inode); 1849 return 0; 1850 } 1851 1852 static inline int locks_verify_truncate(struct inode *inode, 1853 struct file *filp, 1854 loff_t size) 1855 { 1856 if (inode->i_flock && mandatory_lock(inode)) 1857 return locks_mandatory_area( 1858 FLOCK_VERIFY_WRITE, inode, filp, 1859 size < inode->i_size ? size : inode->i_size, 1860 (size < inode->i_size ? inode->i_size - size 1861 : size - inode->i_size) 1862 ); 1863 return 0; 1864 } 1865 1866 static inline int break_lease(struct inode *inode, unsigned int mode) 1867 { 1868 if (inode->i_flock) 1869 return __break_lease(inode, mode); 1870 return 0; 1871 } 1872 #else /* !CONFIG_FILE_LOCKING */ 1873 static inline int locks_mandatory_locked(struct inode *inode) 1874 { 1875 return 0; 1876 } 1877 1878 static inline int locks_mandatory_area(int rw, struct inode *inode, 1879 struct file *filp, loff_t offset, 1880 size_t count) 1881 { 1882 return 0; 1883 } 1884 1885 static inline int __mandatory_lock(struct inode *inode) 1886 { 1887 return 0; 1888 } 1889 1890 static inline int mandatory_lock(struct inode *inode) 1891 { 1892 return 0; 1893 } 1894 1895 static inline int locks_verify_locked(struct inode *inode) 1896 { 1897 return 0; 1898 } 1899 1900 static inline int locks_verify_truncate(struct inode *inode, struct file *filp, 1901 size_t size) 1902 { 1903 return 0; 1904 } 1905 1906 static inline int break_lease(struct inode *inode, unsigned int mode) 1907 { 1908 return 0; 1909 } 1910 1911 #endif /* CONFIG_FILE_LOCKING */ 1912 1913 /* fs/open.c */ 1914 1915 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs, 1916 struct file *filp); 1917 extern int do_fallocate(struct file *file, int mode, loff_t offset, 1918 loff_t len); 1919 extern long do_sys_open(int dfd, const char __user *filename, int flags, 1920 int mode); 1921 extern struct file *filp_open(const char *, int, int); 1922 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int, 1923 const struct cred *); 1924 extern int filp_close(struct file *, fl_owner_t id); 1925 extern char * getname(const char __user *); 1926 1927 /* fs/ioctl.c */ 1928 1929 extern int ioctl_preallocate(struct file *filp, void __user *argp); 1930 1931 /* fs/dcache.c */ 1932 extern void __init vfs_caches_init_early(void); 1933 extern void __init vfs_caches_init(unsigned long); 1934 1935 extern struct kmem_cache *names_cachep; 1936 1937 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp)) 1938 #define __getname() __getname_gfp(GFP_KERNEL) 1939 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) 1940 #ifndef CONFIG_AUDITSYSCALL 1941 #define putname(name) __putname(name) 1942 #else 1943 extern void putname(const char *name); 1944 #endif 1945 1946 #ifdef CONFIG_BLOCK 1947 extern int register_blkdev(unsigned int, const char *); 1948 extern void unregister_blkdev(unsigned int, const char *); 1949 extern struct block_device *bdget(dev_t); 1950 extern struct block_device *bdgrab(struct block_device *bdev); 1951 extern void bd_set_size(struct block_device *, loff_t size); 1952 extern void bd_forget(struct inode *inode); 1953 extern void bdput(struct block_device *); 1954 extern struct block_device *open_by_devnum(dev_t, fmode_t); 1955 extern void invalidate_bdev(struct block_device *); 1956 extern int sync_blockdev(struct block_device *bdev); 1957 extern struct super_block *freeze_bdev(struct block_device *); 1958 extern void emergency_thaw_all(void); 1959 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb); 1960 extern int fsync_bdev(struct block_device *); 1961 #else 1962 static inline void bd_forget(struct inode *inode) {} 1963 static inline int sync_blockdev(struct block_device *bdev) { return 0; } 1964 static inline void invalidate_bdev(struct block_device *bdev) {} 1965 1966 static inline struct super_block *freeze_bdev(struct block_device *sb) 1967 { 1968 return NULL; 1969 } 1970 1971 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb) 1972 { 1973 return 0; 1974 } 1975 #endif 1976 extern int sync_filesystem(struct super_block *); 1977 extern const struct file_operations def_blk_fops; 1978 extern const struct file_operations def_chr_fops; 1979 extern const struct file_operations bad_sock_fops; 1980 extern const struct file_operations def_fifo_fops; 1981 #ifdef CONFIG_BLOCK 1982 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); 1983 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long); 1984 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); 1985 extern int blkdev_get(struct block_device *, fmode_t); 1986 extern int blkdev_put(struct block_device *, fmode_t); 1987 extern int bd_claim(struct block_device *, void *); 1988 extern void bd_release(struct block_device *); 1989 #ifdef CONFIG_SYSFS 1990 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *); 1991 extern void bd_release_from_disk(struct block_device *, struct gendisk *); 1992 #else 1993 #define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder) 1994 #define bd_release_from_disk(bdev, disk) bd_release(bdev) 1995 #endif 1996 #endif 1997 1998 /* fs/char_dev.c */ 1999 #define CHRDEV_MAJOR_HASH_SIZE 255 2000 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); 2001 extern int register_chrdev_region(dev_t, unsigned, const char *); 2002 extern int __register_chrdev(unsigned int major, unsigned int baseminor, 2003 unsigned int count, const char *name, 2004 const struct file_operations *fops); 2005 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, 2006 unsigned int count, const char *name); 2007 extern void unregister_chrdev_region(dev_t, unsigned); 2008 extern void chrdev_show(struct seq_file *,off_t); 2009 2010 static inline int register_chrdev(unsigned int major, const char *name, 2011 const struct file_operations *fops) 2012 { 2013 return __register_chrdev(major, 0, 256, name, fops); 2014 } 2015 2016 static inline void unregister_chrdev(unsigned int major, const char *name) 2017 { 2018 __unregister_chrdev(major, 0, 256, name); 2019 } 2020 2021 /* fs/block_dev.c */ 2022 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ 2023 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ 2024 2025 #ifdef CONFIG_BLOCK 2026 #define BLKDEV_MAJOR_HASH_SIZE 255 2027 extern const char *__bdevname(dev_t, char *buffer); 2028 extern const char *bdevname(struct block_device *bdev, char *buffer); 2029 extern struct block_device *lookup_bdev(const char *); 2030 extern struct block_device *open_bdev_exclusive(const char *, fmode_t, void *); 2031 extern void close_bdev_exclusive(struct block_device *, fmode_t); 2032 extern void blkdev_show(struct seq_file *,off_t); 2033 2034 #else 2035 #define BLKDEV_MAJOR_HASH_SIZE 0 2036 #endif 2037 2038 extern void init_special_inode(struct inode *, umode_t, dev_t); 2039 2040 /* Invalid inode operations -- fs/bad_inode.c */ 2041 extern void make_bad_inode(struct inode *); 2042 extern int is_bad_inode(struct inode *); 2043 2044 extern const struct file_operations read_pipefifo_fops; 2045 extern const struct file_operations write_pipefifo_fops; 2046 extern const struct file_operations rdwr_pipefifo_fops; 2047 2048 extern int fs_may_remount_ro(struct super_block *); 2049 2050 #ifdef CONFIG_BLOCK 2051 /* 2052 * return READ, READA, or WRITE 2053 */ 2054 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK)) 2055 2056 /* 2057 * return data direction, READ or WRITE 2058 */ 2059 #define bio_data_dir(bio) ((bio)->bi_rw & 1) 2060 2061 extern void check_disk_size_change(struct gendisk *disk, 2062 struct block_device *bdev); 2063 extern int revalidate_disk(struct gendisk *); 2064 extern int check_disk_change(struct block_device *); 2065 extern int __invalidate_device(struct block_device *); 2066 extern int invalidate_partition(struct gendisk *, int); 2067 #endif 2068 extern int invalidate_inodes(struct super_block *); 2069 unsigned long invalidate_mapping_pages(struct address_space *mapping, 2070 pgoff_t start, pgoff_t end); 2071 2072 static inline unsigned long __deprecated 2073 invalidate_inode_pages(struct address_space *mapping) 2074 { 2075 return invalidate_mapping_pages(mapping, 0, ~0UL); 2076 } 2077 2078 static inline void invalidate_remote_inode(struct inode *inode) 2079 { 2080 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || 2081 S_ISLNK(inode->i_mode)) 2082 invalidate_mapping_pages(inode->i_mapping, 0, -1); 2083 } 2084 extern int invalidate_inode_pages2(struct address_space *mapping); 2085 extern int invalidate_inode_pages2_range(struct address_space *mapping, 2086 pgoff_t start, pgoff_t end); 2087 extern int write_inode_now(struct inode *, int); 2088 extern int filemap_fdatawrite(struct address_space *); 2089 extern int filemap_flush(struct address_space *); 2090 extern int filemap_fdatawait(struct address_space *); 2091 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart, 2092 loff_t lend); 2093 extern int filemap_write_and_wait(struct address_space *mapping); 2094 extern int filemap_write_and_wait_range(struct address_space *mapping, 2095 loff_t lstart, loff_t lend); 2096 extern int wait_on_page_writeback_range(struct address_space *mapping, 2097 pgoff_t start, pgoff_t end); 2098 extern int __filemap_fdatawrite_range(struct address_space *mapping, 2099 loff_t start, loff_t end, int sync_mode); 2100 extern int filemap_fdatawrite_range(struct address_space *mapping, 2101 loff_t start, loff_t end); 2102 2103 extern int vfs_fsync_range(struct file *file, struct dentry *dentry, 2104 loff_t start, loff_t end, int datasync); 2105 extern int vfs_fsync(struct file *file, struct dentry *dentry, int datasync); 2106 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count); 2107 extern void sync_supers(void); 2108 extern void emergency_sync(void); 2109 extern void emergency_remount(void); 2110 #ifdef CONFIG_BLOCK 2111 extern sector_t bmap(struct inode *, sector_t); 2112 #endif 2113 extern int notify_change(struct dentry *, struct iattr *); 2114 extern int inode_permission(struct inode *, int); 2115 extern int generic_permission(struct inode *, int, 2116 int (*check_acl)(struct inode *, int)); 2117 2118 static inline bool execute_ok(struct inode *inode) 2119 { 2120 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); 2121 } 2122 2123 extern int get_write_access(struct inode *); 2124 extern int deny_write_access(struct file *); 2125 static inline void put_write_access(struct inode * inode) 2126 { 2127 atomic_dec(&inode->i_writecount); 2128 } 2129 static inline void allow_write_access(struct file *file) 2130 { 2131 if (file) 2132 atomic_inc(&file->f_path.dentry->d_inode->i_writecount); 2133 } 2134 extern int do_pipe_flags(int *, int); 2135 extern struct file *create_read_pipe(struct file *f, int flags); 2136 extern struct file *create_write_pipe(int flags); 2137 extern void free_write_pipe(struct file *); 2138 2139 extern struct file *do_filp_open(int dfd, const char *pathname, 2140 int open_flag, int mode, int acc_mode); 2141 extern int may_open(struct path *, int, int); 2142 2143 extern int kernel_read(struct file *, loff_t, char *, unsigned long); 2144 extern struct file * open_exec(const char *); 2145 2146 /* fs/dcache.c -- generic fs support functions */ 2147 extern int is_subdir(struct dentry *, struct dentry *); 2148 extern ino_t find_inode_number(struct dentry *, struct qstr *); 2149 2150 #include <linux/err.h> 2151 2152 /* needed for stackable file system support */ 2153 extern loff_t default_llseek(struct file *file, loff_t offset, int origin); 2154 2155 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin); 2156 2157 extern int inode_init_always(struct super_block *, struct inode *); 2158 extern void inode_init_once(struct inode *); 2159 extern void inode_add_to_lists(struct super_block *, struct inode *); 2160 extern void iput(struct inode *); 2161 extern struct inode * igrab(struct inode *); 2162 extern ino_t iunique(struct super_block *, ino_t); 2163 extern int inode_needs_sync(struct inode *inode); 2164 extern void generic_delete_inode(struct inode *inode); 2165 extern void generic_drop_inode(struct inode *inode); 2166 extern int generic_detach_inode(struct inode *inode); 2167 2168 extern struct inode *ilookup5_nowait(struct super_block *sb, 2169 unsigned long hashval, int (*test)(struct inode *, void *), 2170 void *data); 2171 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, 2172 int (*test)(struct inode *, void *), void *data); 2173 extern struct inode *ilookup(struct super_block *sb, unsigned long ino); 2174 2175 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); 2176 extern struct inode * iget_locked(struct super_block *, unsigned long); 2177 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); 2178 extern int insert_inode_locked(struct inode *); 2179 extern void unlock_new_inode(struct inode *); 2180 2181 extern void __iget(struct inode * inode); 2182 extern void iget_failed(struct inode *); 2183 extern void clear_inode(struct inode *); 2184 extern void destroy_inode(struct inode *); 2185 extern void __destroy_inode(struct inode *); 2186 extern struct inode *new_inode(struct super_block *); 2187 extern int should_remove_suid(struct dentry *); 2188 extern int file_remove_suid(struct file *); 2189 2190 extern void __insert_inode_hash(struct inode *, unsigned long hashval); 2191 extern void remove_inode_hash(struct inode *); 2192 static inline void insert_inode_hash(struct inode *inode) { 2193 __insert_inode_hash(inode, inode->i_ino); 2194 } 2195 2196 extern struct file * get_empty_filp(void); 2197 extern void file_move(struct file *f, struct list_head *list); 2198 extern void file_kill(struct file *f); 2199 #ifdef CONFIG_BLOCK 2200 struct bio; 2201 extern void submit_bio(int, struct bio *); 2202 extern int bdev_read_only(struct block_device *); 2203 #endif 2204 extern int set_blocksize(struct block_device *, int); 2205 extern int sb_set_blocksize(struct super_block *, int); 2206 extern int sb_min_blocksize(struct super_block *, int); 2207 2208 extern int generic_file_mmap(struct file *, struct vm_area_struct *); 2209 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); 2210 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size); 2211 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk); 2212 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t); 2213 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, 2214 loff_t *); 2215 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t); 2216 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *, 2217 unsigned long *, loff_t, loff_t *, size_t, size_t); 2218 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *, 2219 unsigned long, loff_t, loff_t *, size_t, ssize_t); 2220 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos); 2221 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos); 2222 extern int generic_segment_checks(const struct iovec *iov, 2223 unsigned long *nr_segs, size_t *count, int access_flags); 2224 2225 /* fs/block_dev.c */ 2226 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov, 2227 unsigned long nr_segs, loff_t pos); 2228 extern int block_fsync(struct file *filp, struct dentry *dentry, int datasync); 2229 2230 /* fs/splice.c */ 2231 extern ssize_t generic_file_splice_read(struct file *, loff_t *, 2232 struct pipe_inode_info *, size_t, unsigned int); 2233 extern ssize_t default_file_splice_read(struct file *, loff_t *, 2234 struct pipe_inode_info *, size_t, unsigned int); 2235 extern ssize_t generic_file_splice_write(struct pipe_inode_info *, 2236 struct file *, loff_t *, size_t, unsigned int); 2237 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, 2238 struct file *out, loff_t *, size_t len, unsigned int flags); 2239 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, 2240 size_t len, unsigned int flags); 2241 2242 extern void 2243 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); 2244 extern loff_t no_llseek(struct file *file, loff_t offset, int origin); 2245 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin); 2246 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset, 2247 int origin); 2248 extern int generic_file_open(struct inode * inode, struct file * filp); 2249 extern int nonseekable_open(struct inode * inode, struct file * filp); 2250 2251 #ifdef CONFIG_FS_XIP 2252 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len, 2253 loff_t *ppos); 2254 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma); 2255 extern ssize_t xip_file_write(struct file *filp, const char __user *buf, 2256 size_t len, loff_t *ppos); 2257 extern int xip_truncate_page(struct address_space *mapping, loff_t from); 2258 #else 2259 static inline int xip_truncate_page(struct address_space *mapping, loff_t from) 2260 { 2261 return 0; 2262 } 2263 #endif 2264 2265 #ifdef CONFIG_BLOCK 2266 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode, 2267 struct block_device *bdev, const struct iovec *iov, loff_t offset, 2268 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io, 2269 int lock_type); 2270 2271 enum { 2272 DIO_LOCKING = 1, /* need locking between buffered and direct access */ 2273 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */ 2274 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */ 2275 }; 2276 2277 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb, 2278 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 2279 loff_t offset, unsigned long nr_segs, get_block_t get_block, 2280 dio_iodone_t end_io) 2281 { 2282 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 2283 nr_segs, get_block, end_io, DIO_LOCKING); 2284 } 2285 2286 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb, 2287 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 2288 loff_t offset, unsigned long nr_segs, get_block_t get_block, 2289 dio_iodone_t end_io) 2290 { 2291 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 2292 nr_segs, get_block, end_io, DIO_NO_LOCKING); 2293 } 2294 2295 static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb, 2296 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 2297 loff_t offset, unsigned long nr_segs, get_block_t get_block, 2298 dio_iodone_t end_io) 2299 { 2300 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 2301 nr_segs, get_block, end_io, DIO_OWN_LOCKING); 2302 } 2303 #endif 2304 2305 extern const struct file_operations generic_ro_fops; 2306 2307 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) 2308 2309 extern int vfs_readlink(struct dentry *, char __user *, int, const char *); 2310 extern int vfs_follow_link(struct nameidata *, const char *); 2311 extern int page_readlink(struct dentry *, char __user *, int); 2312 extern void *page_follow_link_light(struct dentry *, struct nameidata *); 2313 extern void page_put_link(struct dentry *, struct nameidata *, void *); 2314 extern int __page_symlink(struct inode *inode, const char *symname, int len, 2315 int nofs); 2316 extern int page_symlink(struct inode *inode, const char *symname, int len); 2317 extern const struct inode_operations page_symlink_inode_operations; 2318 extern int generic_readlink(struct dentry *, char __user *, int); 2319 extern void generic_fillattr(struct inode *, struct kstat *); 2320 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *); 2321 void __inode_add_bytes(struct inode *inode, loff_t bytes); 2322 void inode_add_bytes(struct inode *inode, loff_t bytes); 2323 void inode_sub_bytes(struct inode *inode, loff_t bytes); 2324 loff_t inode_get_bytes(struct inode *inode); 2325 void inode_set_bytes(struct inode *inode, loff_t bytes); 2326 2327 extern int vfs_readdir(struct file *, filldir_t, void *); 2328 2329 extern int vfs_stat(char __user *, struct kstat *); 2330 extern int vfs_lstat(char __user *, struct kstat *); 2331 extern int vfs_fstat(unsigned int, struct kstat *); 2332 extern int vfs_fstatat(int , char __user *, struct kstat *, int); 2333 2334 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd, 2335 unsigned long arg); 2336 extern int __generic_block_fiemap(struct inode *inode, 2337 struct fiemap_extent_info *fieinfo, u64 start, 2338 u64 len, get_block_t *get_block); 2339 extern int generic_block_fiemap(struct inode *inode, 2340 struct fiemap_extent_info *fieinfo, u64 start, 2341 u64 len, get_block_t *get_block); 2342 2343 extern void get_filesystem(struct file_system_type *fs); 2344 extern void put_filesystem(struct file_system_type *fs); 2345 extern struct file_system_type *get_fs_type(const char *name); 2346 extern struct super_block *get_super(struct block_device *); 2347 extern struct super_block *get_active_super(struct block_device *bdev); 2348 extern struct super_block *user_get_super(dev_t); 2349 extern void drop_super(struct super_block *sb); 2350 2351 extern int dcache_dir_open(struct inode *, struct file *); 2352 extern int dcache_dir_close(struct inode *, struct file *); 2353 extern loff_t dcache_dir_lseek(struct file *, loff_t, int); 2354 extern int dcache_readdir(struct file *, void *, filldir_t); 2355 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *); 2356 extern int simple_statfs(struct dentry *, struct kstatfs *); 2357 extern int simple_link(struct dentry *, struct inode *, struct dentry *); 2358 extern int simple_unlink(struct inode *, struct dentry *); 2359 extern int simple_rmdir(struct inode *, struct dentry *); 2360 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); 2361 extern int simple_sync_file(struct file *, struct dentry *, int); 2362 extern int simple_empty(struct dentry *); 2363 extern int simple_readpage(struct file *file, struct page *page); 2364 extern int simple_prepare_write(struct file *file, struct page *page, 2365 unsigned offset, unsigned to); 2366 extern int simple_write_begin(struct file *file, struct address_space *mapping, 2367 loff_t pos, unsigned len, unsigned flags, 2368 struct page **pagep, void **fsdata); 2369 extern int simple_write_end(struct file *file, struct address_space *mapping, 2370 loff_t pos, unsigned len, unsigned copied, 2371 struct page *page, void *fsdata); 2372 2373 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *); 2374 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); 2375 extern const struct file_operations simple_dir_operations; 2376 extern const struct inode_operations simple_dir_inode_operations; 2377 struct tree_descr { char *name; const struct file_operations *ops; int mode; }; 2378 struct dentry *d_alloc_name(struct dentry *, const char *); 2379 extern int simple_fill_super(struct super_block *, int, struct tree_descr *); 2380 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); 2381 extern void simple_release_fs(struct vfsmount **mount, int *count); 2382 2383 extern ssize_t simple_read_from_buffer(void __user *to, size_t count, 2384 loff_t *ppos, const void *from, size_t available); 2385 2386 extern int simple_fsync(struct file *, struct dentry *, int); 2387 2388 #ifdef CONFIG_MIGRATION 2389 extern int buffer_migrate_page(struct address_space *, 2390 struct page *, struct page *); 2391 #else 2392 #define buffer_migrate_page NULL 2393 #endif 2394 2395 extern int inode_change_ok(const struct inode *, struct iattr *); 2396 extern int inode_newsize_ok(const struct inode *, loff_t offset); 2397 extern int __must_check inode_setattr(struct inode *, struct iattr *); 2398 2399 extern void file_update_time(struct file *file); 2400 2401 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt); 2402 extern void save_mount_options(struct super_block *sb, char *options); 2403 extern void replace_mount_options(struct super_block *sb, char *options); 2404 2405 static inline ino_t parent_ino(struct dentry *dentry) 2406 { 2407 ino_t res; 2408 2409 spin_lock(&dentry->d_lock); 2410 res = dentry->d_parent->d_inode->i_ino; 2411 spin_unlock(&dentry->d_lock); 2412 return res; 2413 } 2414 2415 /* Transaction based IO helpers */ 2416 2417 /* 2418 * An argresp is stored in an allocated page and holds the 2419 * size of the argument or response, along with its content 2420 */ 2421 struct simple_transaction_argresp { 2422 ssize_t size; 2423 char data[0]; 2424 }; 2425 2426 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) 2427 2428 char *simple_transaction_get(struct file *file, const char __user *buf, 2429 size_t size); 2430 ssize_t simple_transaction_read(struct file *file, char __user *buf, 2431 size_t size, loff_t *pos); 2432 int simple_transaction_release(struct inode *inode, struct file *file); 2433 2434 void simple_transaction_set(struct file *file, size_t n); 2435 2436 /* 2437 * simple attribute files 2438 * 2439 * These attributes behave similar to those in sysfs: 2440 * 2441 * Writing to an attribute immediately sets a value, an open file can be 2442 * written to multiple times. 2443 * 2444 * Reading from an attribute creates a buffer from the value that might get 2445 * read with multiple read calls. When the attribute has been read 2446 * completely, no further read calls are possible until the file is opened 2447 * again. 2448 * 2449 * All attributes contain a text representation of a numeric value 2450 * that are accessed with the get() and set() functions. 2451 */ 2452 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ 2453 static int __fops ## _open(struct inode *inode, struct file *file) \ 2454 { \ 2455 __simple_attr_check_format(__fmt, 0ull); \ 2456 return simple_attr_open(inode, file, __get, __set, __fmt); \ 2457 } \ 2458 static const struct file_operations __fops = { \ 2459 .owner = THIS_MODULE, \ 2460 .open = __fops ## _open, \ 2461 .release = simple_attr_release, \ 2462 .read = simple_attr_read, \ 2463 .write = simple_attr_write, \ 2464 }; 2465 2466 static inline void __attribute__((format(printf, 1, 2))) 2467 __simple_attr_check_format(const char *fmt, ...) 2468 { 2469 /* don't do anything, just let the compiler check the arguments; */ 2470 } 2471 2472 int simple_attr_open(struct inode *inode, struct file *file, 2473 int (*get)(void *, u64 *), int (*set)(void *, u64), 2474 const char *fmt); 2475 int simple_attr_release(struct inode *inode, struct file *file); 2476 ssize_t simple_attr_read(struct file *file, char __user *buf, 2477 size_t len, loff_t *ppos); 2478 ssize_t simple_attr_write(struct file *file, const char __user *buf, 2479 size_t len, loff_t *ppos); 2480 2481 struct ctl_table; 2482 int proc_nr_files(struct ctl_table *table, int write, 2483 void __user *buffer, size_t *lenp, loff_t *ppos); 2484 2485 int __init get_filesystem_list(char *buf); 2486 2487 #endif /* __KERNEL__ */ 2488 #endif /* _LINUX_FS_H */