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)0x1) 64 /* file is open for writing */ 65 #define FMODE_WRITE ((__force fmode_t)0x2) 66 /* file is seekable */ 67 #define FMODE_LSEEK ((__force fmode_t)0x4) 68 /* file can be accessed using pread */ 69 #define FMODE_PREAD ((__force fmode_t)0x8) 70 /* file can be accessed using pwrite */ 71 #define FMODE_PWRITE ((__force fmode_t)0x10) 72 /* File is opened for execution with sys_execve / sys_uselib */ 73 #define FMODE_EXEC ((__force fmode_t)0x20) 74 /* File is opened with O_NDELAY (only set for block devices) */ 75 #define FMODE_NDELAY ((__force fmode_t)0x40) 76 /* File is opened with O_EXCL (only set for block devices) */ 77 #define FMODE_EXCL ((__force fmode_t)0x80) 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)0x100) 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)0x800) 89 90 /* Expect random access pattern */ 91 #define FMODE_RANDOM ((__force fmode_t)0x1000) 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_PLUG 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_PLUG (WRITE | (1 << BIO_RW_SYNCIO)) 158 #define WRITE_META (WRITE | (1 << BIO_RW_META)) 159 #define SWRITE_SYNC_PLUG \ 160 (SWRITE | (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_NOIDLE)) 161 #define SWRITE_SYNC (SWRITE_SYNC_PLUG | (1 << BIO_RW_UNPLUG)) 162 #define WRITE_BARRIER (WRITE | (1 << BIO_RW_BARRIER)) 163 164 /* 165 * These aren't really reads or writes, they pass down information about 166 * parts of device that are now unused by the file system. 167 */ 168 #define DISCARD_NOBARRIER (WRITE | (1 << BIO_RW_DISCARD)) 169 #define DISCARD_BARRIER (DISCARD_NOBARRIER | (1 << BIO_RW_BARRIER)) 170 171 #define SEL_IN 1 172 #define SEL_OUT 2 173 #define SEL_EX 4 174 175 /* public flags for file_system_type */ 176 #define FS_REQUIRES_DEV 1 177 #define FS_BINARY_MOUNTDATA 2 178 #define FS_HAS_SUBTYPE 4 179 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */ 180 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() 181 * during rename() internally. 182 */ 183 184 /* 185 * These are the fs-independent mount-flags: up to 32 flags are supported 186 */ 187 #define MS_RDONLY 1 /* Mount read-only */ 188 #define MS_NOSUID 2 /* Ignore suid and sgid bits */ 189 #define MS_NODEV 4 /* Disallow access to device special files */ 190 #define MS_NOEXEC 8 /* Disallow program execution */ 191 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */ 192 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */ 193 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */ 194 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */ 195 #define MS_NOATIME 1024 /* Do not update access times. */ 196 #define MS_NODIRATIME 2048 /* Do not update directory access times */ 197 #define MS_BIND 4096 198 #define MS_MOVE 8192 199 #define MS_REC 16384 200 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence. 201 MS_VERBOSE is deprecated. */ 202 #define MS_SILENT 32768 203 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */ 204 #define MS_UNBINDABLE (1<<17) /* change to unbindable */ 205 #define MS_PRIVATE (1<<18) /* change to private */ 206 #define MS_SLAVE (1<<19) /* change to slave */ 207 #define MS_SHARED (1<<20) /* change to shared */ 208 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */ 209 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */ 210 #define MS_I_VERSION (1<<23) /* Update inode I_version field */ 211 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */ 212 #define MS_ACTIVE (1<<30) 213 #define MS_NOUSER (1<<31) 214 215 /* 216 * Superblock flags that can be altered by MS_REMOUNT 217 */ 218 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION) 219 220 /* 221 * Old magic mount flag and mask 222 */ 223 #define MS_MGC_VAL 0xC0ED0000 224 #define MS_MGC_MSK 0xffff0000 225 226 /* Inode flags - they have nothing to superblock flags now */ 227 228 #define S_SYNC 1 /* Writes are synced at once */ 229 #define S_NOATIME 2 /* Do not update access times */ 230 #define S_APPEND 4 /* Append-only file */ 231 #define S_IMMUTABLE 8 /* Immutable file */ 232 #define S_DEAD 16 /* removed, but still open directory */ 233 #define S_NOQUOTA 32 /* Inode is not counted to quota */ 234 #define S_DIRSYNC 64 /* Directory modifications are synchronous */ 235 #define S_NOCMTIME 128 /* Do not update file c/mtime */ 236 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */ 237 #define S_PRIVATE 512 /* Inode is fs-internal */ 238 239 /* 240 * Note that nosuid etc flags are inode-specific: setting some file-system 241 * flags just means all the inodes inherit those flags by default. It might be 242 * possible to override it selectively if you really wanted to with some 243 * ioctl() that is not currently implemented. 244 * 245 * Exception: MS_RDONLY is always applied to the entire file system. 246 * 247 * Unfortunately, it is possible to change a filesystems flags with it mounted 248 * with files in use. This means that all of the inodes will not have their 249 * i_flags updated. Hence, i_flags no longer inherit the superblock mount 250 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org 251 */ 252 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg)) 253 254 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY) 255 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \ 256 ((inode)->i_flags & S_SYNC)) 257 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \ 258 ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) 259 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK) 260 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME) 261 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION) 262 263 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) 264 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) 265 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) 266 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL) 267 268 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) 269 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) 270 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) 271 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) 272 273 /* the read-only stuff doesn't really belong here, but any other place is 274 probably as bad and I don't want to create yet another include file. */ 275 276 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */ 277 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */ 278 #define BLKRRPART _IO(0x12,95) /* re-read partition table */ 279 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */ 280 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */ 281 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */ 282 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */ 283 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */ 284 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */ 285 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */ 286 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */ 287 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */ 288 #if 0 289 #define BLKPG _IO(0x12,105)/* See blkpg.h */ 290 291 /* Some people are morons. Do not use sizeof! */ 292 293 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */ 294 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */ 295 /* This was here just to show that the number is taken - 296 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */ 297 #endif 298 /* A jump here: 108-111 have been used for various private purposes. */ 299 #define BLKBSZGET _IOR(0x12,112,size_t) 300 #define BLKBSZSET _IOW(0x12,113,size_t) 301 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */ 302 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup) 303 #define BLKTRACESTART _IO(0x12,116) 304 #define BLKTRACESTOP _IO(0x12,117) 305 #define BLKTRACETEARDOWN _IO(0x12,118) 306 #define BLKDISCARD _IO(0x12,119) 307 #define BLKIOMIN _IO(0x12,120) 308 #define BLKIOOPT _IO(0x12,121) 309 #define BLKALIGNOFF _IO(0x12,122) 310 #define BLKPBSZGET _IO(0x12,123) 311 #define BLKDISCARDZEROES _IO(0x12,124) 312 313 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */ 314 #define FIBMAP _IO(0x00,1) /* bmap access */ 315 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */ 316 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */ 317 #define FITHAW _IOWR('X', 120, int) /* Thaw */ 318 319 #define FS_IOC_GETFLAGS _IOR('f', 1, long) 320 #define FS_IOC_SETFLAGS _IOW('f', 2, long) 321 #define FS_IOC_GETVERSION _IOR('v', 1, long) 322 #define FS_IOC_SETVERSION _IOW('v', 2, long) 323 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap) 324 #define FS_IOC32_GETFLAGS _IOR('f', 1, int) 325 #define FS_IOC32_SETFLAGS _IOW('f', 2, int) 326 #define FS_IOC32_GETVERSION _IOR('v', 1, int) 327 #define FS_IOC32_SETVERSION _IOW('v', 2, int) 328 329 /* 330 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS) 331 */ 332 #define FS_SECRM_FL 0x00000001 /* Secure deletion */ 333 #define FS_UNRM_FL 0x00000002 /* Undelete */ 334 #define FS_COMPR_FL 0x00000004 /* Compress file */ 335 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */ 336 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 337 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */ 338 #define FS_NODUMP_FL 0x00000040 /* do not dump file */ 339 #define FS_NOATIME_FL 0x00000080 /* do not update atime */ 340 /* Reserved for compression usage... */ 341 #define FS_DIRTY_FL 0x00000100 342 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 343 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */ 344 #define FS_ECOMPR_FL 0x00000800 /* Compression error */ 345 /* End compression flags --- maybe not all used */ 346 #define FS_BTREE_FL 0x00001000 /* btree format dir */ 347 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 348 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */ 349 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */ 350 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 351 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 352 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 353 #define FS_EXTENT_FL 0x00080000 /* Extents */ 354 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */ 355 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */ 356 357 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */ 358 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */ 359 360 361 #define SYNC_FILE_RANGE_WAIT_BEFORE 1 362 #define SYNC_FILE_RANGE_WRITE 2 363 #define SYNC_FILE_RANGE_WAIT_AFTER 4 364 365 #ifdef __KERNEL__ 366 367 #include <linux/linkage.h> 368 #include <linux/wait.h> 369 #include <linux/types.h> 370 #include <linux/kdev_t.h> 371 #include <linux/dcache.h> 372 #include <linux/path.h> 373 #include <linux/stat.h> 374 #include <linux/cache.h> 375 #include <linux/kobject.h> 376 #include <linux/list.h> 377 #include <linux/radix-tree.h> 378 #include <linux/prio_tree.h> 379 #include <linux/init.h> 380 #include <linux/pid.h> 381 #include <linux/mutex.h> 382 #include <linux/capability.h> 383 #include <linux/semaphore.h> 384 #include <linux/fiemap.h> 385 386 #include <asm/atomic.h> 387 #include <asm/byteorder.h> 388 389 struct export_operations; 390 struct hd_geometry; 391 struct iovec; 392 struct nameidata; 393 struct kiocb; 394 struct pipe_inode_info; 395 struct poll_table_struct; 396 struct kstatfs; 397 struct vm_area_struct; 398 struct vfsmount; 399 struct cred; 400 401 extern void __init inode_init(void); 402 extern void __init inode_init_early(void); 403 extern void __init files_init(unsigned long); 404 405 extern struct files_stat_struct files_stat; 406 extern int get_max_files(void); 407 extern int sysctl_nr_open; 408 extern struct inodes_stat_t inodes_stat; 409 extern int leases_enable, lease_break_time; 410 #ifdef CONFIG_DNOTIFY 411 extern int dir_notify_enable; 412 #endif 413 414 struct buffer_head; 415 typedef int (get_block_t)(struct inode *inode, sector_t iblock, 416 struct buffer_head *bh_result, int create); 417 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset, 418 ssize_t bytes, void *private); 419 420 /* 421 * Attribute flags. These should be or-ed together to figure out what 422 * has been changed! 423 */ 424 #define ATTR_MODE (1 << 0) 425 #define ATTR_UID (1 << 1) 426 #define ATTR_GID (1 << 2) 427 #define ATTR_SIZE (1 << 3) 428 #define ATTR_ATIME (1 << 4) 429 #define ATTR_MTIME (1 << 5) 430 #define ATTR_CTIME (1 << 6) 431 #define ATTR_ATIME_SET (1 << 7) 432 #define ATTR_MTIME_SET (1 << 8) 433 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ 434 #define ATTR_ATTR_FLAG (1 << 10) 435 #define ATTR_KILL_SUID (1 << 11) 436 #define ATTR_KILL_SGID (1 << 12) 437 #define ATTR_FILE (1 << 13) 438 #define ATTR_KILL_PRIV (1 << 14) 439 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ 440 #define ATTR_TIMES_SET (1 << 16) 441 442 /* 443 * This is the Inode Attributes structure, used for notify_change(). It 444 * uses the above definitions as flags, to know which values have changed. 445 * Also, in this manner, a Filesystem can look at only the values it cares 446 * about. Basically, these are the attributes that the VFS layer can 447 * request to change from the FS layer. 448 * 449 * Derek Atkins <warlord@MIT.EDU> 94-10-20 450 */ 451 struct iattr { 452 unsigned int ia_valid; 453 umode_t ia_mode; 454 uid_t ia_uid; 455 gid_t ia_gid; 456 loff_t ia_size; 457 struct timespec ia_atime; 458 struct timespec ia_mtime; 459 struct timespec ia_ctime; 460 461 /* 462 * Not an attribute, but an auxilary info for filesystems wanting to 463 * implement an ftruncate() like method. NOTE: filesystem should 464 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). 465 */ 466 struct file *ia_file; 467 }; 468 469 /* 470 * Includes for diskquotas. 471 */ 472 #include <linux/quota.h> 473 474 /** 475 * enum positive_aop_returns - aop return codes with specific semantics 476 * 477 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has 478 * completed, that the page is still locked, and 479 * should be considered active. The VM uses this hint 480 * to return the page to the active list -- it won't 481 * be a candidate for writeback again in the near 482 * future. Other callers must be careful to unlock 483 * the page if they get this return. Returned by 484 * writepage(); 485 * 486 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has 487 * unlocked it and the page might have been truncated. 488 * The caller should back up to acquiring a new page and 489 * trying again. The aop will be taking reasonable 490 * precautions not to livelock. If the caller held a page 491 * reference, it should drop it before retrying. Returned 492 * by readpage(). 493 * 494 * address_space_operation functions return these large constants to indicate 495 * special semantics to the caller. These are much larger than the bytes in a 496 * page to allow for functions that return the number of bytes operated on in a 497 * given page. 498 */ 499 500 enum positive_aop_returns { 501 AOP_WRITEPAGE_ACTIVATE = 0x80000, 502 AOP_TRUNCATED_PAGE = 0x80001, 503 }; 504 505 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */ 506 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */ 507 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct 508 * helper code (eg buffer layer) 509 * to clear GFP_FS from alloc */ 510 511 /* 512 * oh the beauties of C type declarations. 513 */ 514 struct page; 515 struct address_space; 516 struct writeback_control; 517 518 struct iov_iter { 519 const struct iovec *iov; 520 unsigned long nr_segs; 521 size_t iov_offset; 522 size_t count; 523 }; 524 525 size_t iov_iter_copy_from_user_atomic(struct page *page, 526 struct iov_iter *i, unsigned long offset, size_t bytes); 527 size_t iov_iter_copy_from_user(struct page *page, 528 struct iov_iter *i, unsigned long offset, size_t bytes); 529 void iov_iter_advance(struct iov_iter *i, size_t bytes); 530 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes); 531 size_t iov_iter_single_seg_count(struct iov_iter *i); 532 533 static inline void iov_iter_init(struct iov_iter *i, 534 const struct iovec *iov, unsigned long nr_segs, 535 size_t count, size_t written) 536 { 537 i->iov = iov; 538 i->nr_segs = nr_segs; 539 i->iov_offset = 0; 540 i->count = count + written; 541 542 iov_iter_advance(i, written); 543 } 544 545 static inline size_t iov_iter_count(struct iov_iter *i) 546 { 547 return i->count; 548 } 549 550 /* 551 * "descriptor" for what we're up to with a read. 552 * This allows us to use the same read code yet 553 * have multiple different users of the data that 554 * we read from a file. 555 * 556 * The simplest case just copies the data to user 557 * mode. 558 */ 559 typedef struct { 560 size_t written; 561 size_t count; 562 union { 563 char __user *buf; 564 void *data; 565 } arg; 566 int error; 567 } read_descriptor_t; 568 569 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, 570 unsigned long, unsigned long); 571 572 struct address_space_operations { 573 int (*writepage)(struct page *page, struct writeback_control *wbc); 574 int (*readpage)(struct file *, struct page *); 575 void (*sync_page)(struct page *); 576 577 /* Write back some dirty pages from this mapping. */ 578 int (*writepages)(struct address_space *, struct writeback_control *); 579 580 /* Set a page dirty. Return true if this dirtied it */ 581 int (*set_page_dirty)(struct page *page); 582 583 int (*readpages)(struct file *filp, struct address_space *mapping, 584 struct list_head *pages, unsigned nr_pages); 585 586 int (*write_begin)(struct file *, struct address_space *mapping, 587 loff_t pos, unsigned len, unsigned flags, 588 struct page **pagep, void **fsdata); 589 int (*write_end)(struct file *, struct address_space *mapping, 590 loff_t pos, unsigned len, unsigned copied, 591 struct page *page, void *fsdata); 592 593 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ 594 sector_t (*bmap)(struct address_space *, sector_t); 595 void (*invalidatepage) (struct page *, unsigned long); 596 int (*releasepage) (struct page *, gfp_t); 597 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov, 598 loff_t offset, unsigned long nr_segs); 599 int (*get_xip_mem)(struct address_space *, pgoff_t, int, 600 void **, unsigned long *); 601 /* migrate the contents of a page to the specified target */ 602 int (*migratepage) (struct address_space *, 603 struct page *, struct page *); 604 int (*launder_page) (struct page *); 605 int (*is_partially_uptodate) (struct page *, read_descriptor_t *, 606 unsigned long); 607 int (*error_remove_page)(struct address_space *, struct page *); 608 }; 609 610 /* 611 * pagecache_write_begin/pagecache_write_end must be used by general code 612 * to write into the pagecache. 613 */ 614 int pagecache_write_begin(struct file *, struct address_space *mapping, 615 loff_t pos, unsigned len, unsigned flags, 616 struct page **pagep, void **fsdata); 617 618 int pagecache_write_end(struct file *, struct address_space *mapping, 619 loff_t pos, unsigned len, unsigned copied, 620 struct page *page, void *fsdata); 621 622 struct backing_dev_info; 623 struct address_space { 624 struct inode *host; /* owner: inode, block_device */ 625 struct radix_tree_root page_tree; /* radix tree of all pages */ 626 spinlock_t tree_lock; /* and lock protecting it */ 627 unsigned int i_mmap_writable;/* count VM_SHARED mappings */ 628 struct prio_tree_root i_mmap; /* tree of private and shared mappings */ 629 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */ 630 spinlock_t i_mmap_lock; /* protect tree, count, list */ 631 unsigned int truncate_count; /* Cover race condition with truncate */ 632 unsigned long nrpages; /* number of total pages */ 633 pgoff_t writeback_index;/* writeback starts here */ 634 const struct address_space_operations *a_ops; /* methods */ 635 unsigned long flags; /* error bits/gfp mask */ 636 struct backing_dev_info *backing_dev_info; /* device readahead, etc */ 637 spinlock_t private_lock; /* for use by the address_space */ 638 struct list_head private_list; /* ditto */ 639 struct address_space *assoc_mapping; /* ditto */ 640 } __attribute__((aligned(sizeof(long)))); 641 /* 642 * On most architectures that alignment is already the case; but 643 * must be enforced here for CRIS, to let the least signficant bit 644 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. 645 */ 646 647 struct block_device { 648 dev_t bd_dev; /* not a kdev_t - it's a search key */ 649 struct inode * bd_inode; /* will die */ 650 struct super_block * bd_super; 651 int bd_openers; 652 struct mutex bd_mutex; /* open/close mutex */ 653 struct list_head bd_inodes; 654 void * bd_holder; 655 int bd_holders; 656 #ifdef CONFIG_SYSFS 657 struct list_head bd_holder_list; 658 #endif 659 struct block_device * bd_contains; 660 unsigned bd_block_size; 661 struct hd_struct * bd_part; 662 /* number of times partitions within this device have been opened. */ 663 unsigned bd_part_count; 664 int bd_invalidated; 665 struct gendisk * bd_disk; 666 struct list_head bd_list; 667 /* 668 * Private data. You must have bd_claim'ed the block_device 669 * to use this. NOTE: bd_claim allows an owner to claim 670 * the same device multiple times, the owner must take special 671 * care to not mess up bd_private for that case. 672 */ 673 unsigned long bd_private; 674 675 /* The counter of freeze processes */ 676 int bd_fsfreeze_count; 677 /* Mutex for freeze */ 678 struct mutex bd_fsfreeze_mutex; 679 }; 680 681 /* 682 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache 683 * radix trees 684 */ 685 #define PAGECACHE_TAG_DIRTY 0 686 #define PAGECACHE_TAG_WRITEBACK 1 687 688 int mapping_tagged(struct address_space *mapping, int tag); 689 690 /* 691 * Might pages of this file be mapped into userspace? 692 */ 693 static inline int mapping_mapped(struct address_space *mapping) 694 { 695 return !prio_tree_empty(&mapping->i_mmap) || 696 !list_empty(&mapping->i_mmap_nonlinear); 697 } 698 699 /* 700 * Might pages of this file have been modified in userspace? 701 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff 702 * marks vma as VM_SHARED if it is shared, and the file was opened for 703 * writing i.e. vma may be mprotected writable even if now readonly. 704 */ 705 static inline int mapping_writably_mapped(struct address_space *mapping) 706 { 707 return mapping->i_mmap_writable != 0; 708 } 709 710 /* 711 * Use sequence counter to get consistent i_size on 32-bit processors. 712 */ 713 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 714 #include <linux/seqlock.h> 715 #define __NEED_I_SIZE_ORDERED 716 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) 717 #else 718 #define i_size_ordered_init(inode) do { } while (0) 719 #endif 720 721 struct posix_acl; 722 #define ACL_NOT_CACHED ((void *)(-1)) 723 724 struct inode { 725 struct hlist_node i_hash; 726 struct list_head i_list; /* backing dev IO list */ 727 struct list_head i_sb_list; 728 struct list_head i_dentry; 729 unsigned long i_ino; 730 atomic_t i_count; 731 unsigned int i_nlink; 732 uid_t i_uid; 733 gid_t i_gid; 734 dev_t i_rdev; 735 unsigned int i_blkbits; 736 u64 i_version; 737 loff_t i_size; 738 #ifdef __NEED_I_SIZE_ORDERED 739 seqcount_t i_size_seqcount; 740 #endif 741 struct timespec i_atime; 742 struct timespec i_mtime; 743 struct timespec i_ctime; 744 blkcnt_t i_blocks; 745 unsigned short i_bytes; 746 umode_t i_mode; 747 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ 748 struct mutex i_mutex; 749 struct rw_semaphore i_alloc_sem; 750 const struct inode_operations *i_op; 751 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ 752 struct super_block *i_sb; 753 struct file_lock *i_flock; 754 struct address_space *i_mapping; 755 struct address_space i_data; 756 #ifdef CONFIG_QUOTA 757 struct dquot *i_dquot[MAXQUOTAS]; 758 #endif 759 struct list_head i_devices; 760 union { 761 struct pipe_inode_info *i_pipe; 762 struct block_device *i_bdev; 763 struct cdev *i_cdev; 764 }; 765 766 __u32 i_generation; 767 768 #ifdef CONFIG_FSNOTIFY 769 __u32 i_fsnotify_mask; /* all events this inode cares about */ 770 struct hlist_head i_fsnotify_mark_entries; /* fsnotify mark entries */ 771 #endif 772 773 #ifdef CONFIG_INOTIFY 774 struct list_head inotify_watches; /* watches on this inode */ 775 struct mutex inotify_mutex; /* protects the watches list */ 776 #endif 777 778 unsigned long i_state; 779 unsigned long dirtied_when; /* jiffies of first dirtying */ 780 781 unsigned int i_flags; 782 783 atomic_t i_writecount; 784 #ifdef CONFIG_SECURITY 785 void *i_security; 786 #endif 787 #ifdef CONFIG_FS_POSIX_ACL 788 struct posix_acl *i_acl; 789 struct posix_acl *i_default_acl; 790 #endif 791 void *i_private; /* fs or device private pointer */ 792 }; 793 794 /* 795 * inode->i_mutex nesting subclasses for the lock validator: 796 * 797 * 0: the object of the current VFS operation 798 * 1: parent 799 * 2: child/target 800 * 3: quota file 801 * 802 * The locking order between these classes is 803 * parent -> child -> normal -> xattr -> quota 804 */ 805 enum inode_i_mutex_lock_class 806 { 807 I_MUTEX_NORMAL, 808 I_MUTEX_PARENT, 809 I_MUTEX_CHILD, 810 I_MUTEX_XATTR, 811 I_MUTEX_QUOTA 812 }; 813 814 /* 815 * NOTE: in a 32bit arch with a preemptable kernel and 816 * an UP compile the i_size_read/write must be atomic 817 * with respect to the local cpu (unlike with preempt disabled), 818 * but they don't need to be atomic with respect to other cpus like in 819 * true SMP (so they need either to either locally disable irq around 820 * the read or for example on x86 they can be still implemented as a 821 * cmpxchg8b without the need of the lock prefix). For SMP compiles 822 * and 64bit archs it makes no difference if preempt is enabled or not. 823 */ 824 static inline loff_t i_size_read(const struct inode *inode) 825 { 826 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 827 loff_t i_size; 828 unsigned int seq; 829 830 do { 831 seq = read_seqcount_begin(&inode->i_size_seqcount); 832 i_size = inode->i_size; 833 } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); 834 return i_size; 835 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 836 loff_t i_size; 837 838 preempt_disable(); 839 i_size = inode->i_size; 840 preempt_enable(); 841 return i_size; 842 #else 843 return inode->i_size; 844 #endif 845 } 846 847 /* 848 * NOTE: unlike i_size_read(), i_size_write() does need locking around it 849 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount 850 * can be lost, resulting in subsequent i_size_read() calls spinning forever. 851 */ 852 static inline void i_size_write(struct inode *inode, loff_t i_size) 853 { 854 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 855 write_seqcount_begin(&inode->i_size_seqcount); 856 inode->i_size = i_size; 857 write_seqcount_end(&inode->i_size_seqcount); 858 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 859 preempt_disable(); 860 inode->i_size = i_size; 861 preempt_enable(); 862 #else 863 inode->i_size = i_size; 864 #endif 865 } 866 867 static inline unsigned iminor(const struct inode *inode) 868 { 869 return MINOR(inode->i_rdev); 870 } 871 872 static inline unsigned imajor(const struct inode *inode) 873 { 874 return MAJOR(inode->i_rdev); 875 } 876 877 extern struct block_device *I_BDEV(struct inode *inode); 878 879 struct fown_struct { 880 rwlock_t lock; /* protects pid, uid, euid fields */ 881 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ 882 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ 883 uid_t uid, euid; /* uid/euid of process setting the owner */ 884 int signum; /* posix.1b rt signal to be delivered on IO */ 885 }; 886 887 /* 888 * Track a single file's readahead state 889 */ 890 struct file_ra_state { 891 pgoff_t start; /* where readahead started */ 892 unsigned int size; /* # of readahead pages */ 893 unsigned int async_size; /* do asynchronous readahead when 894 there are only # of pages ahead */ 895 896 unsigned int ra_pages; /* Maximum readahead window */ 897 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */ 898 loff_t prev_pos; /* Cache last read() position */ 899 }; 900 901 /* 902 * Check if @index falls in the readahead windows. 903 */ 904 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) 905 { 906 return (index >= ra->start && 907 index < ra->start + ra->size); 908 } 909 910 #define FILE_MNT_WRITE_TAKEN 1 911 #define FILE_MNT_WRITE_RELEASED 2 912 913 struct file { 914 /* 915 * fu_list becomes invalid after file_free is called and queued via 916 * fu_rcuhead for RCU freeing 917 */ 918 union { 919 struct list_head fu_list; 920 struct rcu_head fu_rcuhead; 921 } f_u; 922 struct path f_path; 923 #define f_dentry f_path.dentry 924 #define f_vfsmnt f_path.mnt 925 const struct file_operations *f_op; 926 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */ 927 atomic_long_t f_count; 928 unsigned int f_flags; 929 fmode_t f_mode; 930 loff_t f_pos; 931 struct fown_struct f_owner; 932 const struct cred *f_cred; 933 struct file_ra_state f_ra; 934 935 u64 f_version; 936 #ifdef CONFIG_SECURITY 937 void *f_security; 938 #endif 939 /* needed for tty driver, and maybe others */ 940 void *private_data; 941 942 #ifdef CONFIG_EPOLL 943 /* Used by fs/eventpoll.c to link all the hooks to this file */ 944 struct list_head f_ep_links; 945 #endif /* #ifdef CONFIG_EPOLL */ 946 struct address_space *f_mapping; 947 #ifdef CONFIG_DEBUG_WRITECOUNT 948 unsigned long f_mnt_write_state; 949 #endif 950 }; 951 extern spinlock_t files_lock; 952 #define file_list_lock() spin_lock(&files_lock); 953 #define file_list_unlock() spin_unlock(&files_lock); 954 955 #define get_file(x) atomic_long_inc(&(x)->f_count) 956 #define file_count(x) atomic_long_read(&(x)->f_count) 957 958 #ifdef CONFIG_DEBUG_WRITECOUNT 959 static inline void file_take_write(struct file *f) 960 { 961 WARN_ON(f->f_mnt_write_state != 0); 962 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN; 963 } 964 static inline void file_release_write(struct file *f) 965 { 966 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED; 967 } 968 static inline void file_reset_write(struct file *f) 969 { 970 f->f_mnt_write_state = 0; 971 } 972 static inline void file_check_state(struct file *f) 973 { 974 /* 975 * At this point, either both or neither of these bits 976 * should be set. 977 */ 978 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN); 979 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED); 980 } 981 static inline int file_check_writeable(struct file *f) 982 { 983 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN) 984 return 0; 985 printk(KERN_WARNING "writeable file with no " 986 "mnt_want_write()\n"); 987 WARN_ON(1); 988 return -EINVAL; 989 } 990 #else /* !CONFIG_DEBUG_WRITECOUNT */ 991 static inline void file_take_write(struct file *filp) {} 992 static inline void file_release_write(struct file *filp) {} 993 static inline void file_reset_write(struct file *filp) {} 994 static inline void file_check_state(struct file *filp) {} 995 static inline int file_check_writeable(struct file *filp) 996 { 997 return 0; 998 } 999 #endif /* CONFIG_DEBUG_WRITECOUNT */ 1000 1001 #define MAX_NON_LFS ((1UL<<31) - 1) 1002 1003 /* Page cache limit. The filesystems should put that into their s_maxbytes 1004 limits, otherwise bad things can happen in VM. */ 1005 #if BITS_PER_LONG==32 1006 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) 1007 #elif BITS_PER_LONG==64 1008 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL 1009 #endif 1010 1011 #define FL_POSIX 1 1012 #define FL_FLOCK 2 1013 #define FL_ACCESS 8 /* not trying to lock, just looking */ 1014 #define FL_EXISTS 16 /* when unlocking, test for existence */ 1015 #define FL_LEASE 32 /* lease held on this file */ 1016 #define FL_CLOSE 64 /* unlock on close */ 1017 #define FL_SLEEP 128 /* A blocking lock */ 1018 1019 /* 1020 * Special return value from posix_lock_file() and vfs_lock_file() for 1021 * asynchronous locking. 1022 */ 1023 #define FILE_LOCK_DEFERRED 1 1024 1025 /* 1026 * The POSIX file lock owner is determined by 1027 * the "struct files_struct" in the thread group 1028 * (or NULL for no owner - BSD locks). 1029 * 1030 * Lockd stuffs a "host" pointer into this. 1031 */ 1032 typedef struct files_struct *fl_owner_t; 1033 1034 struct file_lock_operations { 1035 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 1036 void (*fl_release_private)(struct file_lock *); 1037 }; 1038 1039 struct lock_manager_operations { 1040 int (*fl_compare_owner)(struct file_lock *, struct file_lock *); 1041 void (*fl_notify)(struct file_lock *); /* unblock callback */ 1042 int (*fl_grant)(struct file_lock *, struct file_lock *, int); 1043 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 1044 void (*fl_release_private)(struct file_lock *); 1045 void (*fl_break)(struct file_lock *); 1046 int (*fl_mylease)(struct file_lock *, struct file_lock *); 1047 int (*fl_change)(struct file_lock **, int); 1048 }; 1049 1050 struct lock_manager { 1051 struct list_head list; 1052 }; 1053 1054 void locks_start_grace(struct lock_manager *); 1055 void locks_end_grace(struct lock_manager *); 1056 int locks_in_grace(void); 1057 1058 /* that will die - we need it for nfs_lock_info */ 1059 #include <linux/nfs_fs_i.h> 1060 1061 struct file_lock { 1062 struct file_lock *fl_next; /* singly linked list for this inode */ 1063 struct list_head fl_link; /* doubly linked list of all locks */ 1064 struct list_head fl_block; /* circular list of blocked processes */ 1065 fl_owner_t fl_owner; 1066 unsigned char fl_flags; 1067 unsigned char fl_type; 1068 unsigned int fl_pid; 1069 struct pid *fl_nspid; 1070 wait_queue_head_t fl_wait; 1071 struct file *fl_file; 1072 loff_t fl_start; 1073 loff_t fl_end; 1074 1075 struct fasync_struct * fl_fasync; /* for lease break notifications */ 1076 unsigned long fl_break_time; /* for nonblocking lease breaks */ 1077 1078 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ 1079 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ 1080 union { 1081 struct nfs_lock_info nfs_fl; 1082 struct nfs4_lock_info nfs4_fl; 1083 struct { 1084 struct list_head link; /* link in AFS vnode's pending_locks list */ 1085 int state; /* state of grant or error if -ve */ 1086 } afs; 1087 } fl_u; 1088 }; 1089 1090 /* The following constant reflects the upper bound of the file/locking space */ 1091 #ifndef OFFSET_MAX 1092 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) 1093 #define OFFSET_MAX INT_LIMIT(loff_t) 1094 #define OFFT_OFFSET_MAX INT_LIMIT(off_t) 1095 #endif 1096 1097 #include <linux/fcntl.h> 1098 1099 extern void send_sigio(struct fown_struct *fown, int fd, int band); 1100 1101 #ifdef CONFIG_FILE_LOCKING 1102 extern int fcntl_getlk(struct file *, struct flock __user *); 1103 extern int fcntl_setlk(unsigned int, struct file *, unsigned int, 1104 struct flock __user *); 1105 1106 #if BITS_PER_LONG == 32 1107 extern int fcntl_getlk64(struct file *, struct flock64 __user *); 1108 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, 1109 struct flock64 __user *); 1110 #endif 1111 1112 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); 1113 extern int fcntl_getlease(struct file *filp); 1114 1115 /* fs/locks.c */ 1116 extern void locks_init_lock(struct file_lock *); 1117 extern void locks_copy_lock(struct file_lock *, struct file_lock *); 1118 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *); 1119 extern void locks_remove_posix(struct file *, fl_owner_t); 1120 extern void locks_remove_flock(struct file *); 1121 extern void locks_release_private(struct file_lock *); 1122 extern void posix_test_lock(struct file *, struct file_lock *); 1123 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *); 1124 extern int posix_lock_file_wait(struct file *, struct file_lock *); 1125 extern int posix_unblock_lock(struct file *, struct file_lock *); 1126 extern int vfs_test_lock(struct file *, struct file_lock *); 1127 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *); 1128 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl); 1129 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl); 1130 extern int __break_lease(struct inode *inode, unsigned int flags); 1131 extern void lease_get_mtime(struct inode *, struct timespec *time); 1132 extern int generic_setlease(struct file *, long, struct file_lock **); 1133 extern int vfs_setlease(struct file *, long, struct file_lock **); 1134 extern int lease_modify(struct file_lock **, int); 1135 extern int lock_may_read(struct inode *, loff_t start, unsigned long count); 1136 extern int lock_may_write(struct inode *, loff_t start, unsigned long count); 1137 #else /* !CONFIG_FILE_LOCKING */ 1138 static inline int fcntl_getlk(struct file *file, struct flock __user *user) 1139 { 1140 return -EINVAL; 1141 } 1142 1143 static inline int fcntl_setlk(unsigned int fd, struct file *file, 1144 unsigned int cmd, struct flock __user *user) 1145 { 1146 return -EACCES; 1147 } 1148 1149 #if BITS_PER_LONG == 32 1150 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user) 1151 { 1152 return -EINVAL; 1153 } 1154 1155 static inline int fcntl_setlk64(unsigned int fd, struct file *file, 1156 unsigned int cmd, struct flock64 __user *user) 1157 { 1158 return -EACCES; 1159 } 1160 #endif 1161 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg) 1162 { 1163 return 0; 1164 } 1165 1166 static inline int fcntl_getlease(struct file *filp) 1167 { 1168 return 0; 1169 } 1170 1171 static inline void locks_init_lock(struct file_lock *fl) 1172 { 1173 return; 1174 } 1175 1176 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl) 1177 { 1178 return; 1179 } 1180 1181 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl) 1182 { 1183 return; 1184 } 1185 1186 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner) 1187 { 1188 return; 1189 } 1190 1191 static inline void locks_remove_flock(struct file *filp) 1192 { 1193 return; 1194 } 1195 1196 static inline void posix_test_lock(struct file *filp, struct file_lock *fl) 1197 { 1198 return; 1199 } 1200 1201 static inline int posix_lock_file(struct file *filp, struct file_lock *fl, 1202 struct file_lock *conflock) 1203 { 1204 return -ENOLCK; 1205 } 1206 1207 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl) 1208 { 1209 return -ENOLCK; 1210 } 1211 1212 static inline int posix_unblock_lock(struct file *filp, 1213 struct file_lock *waiter) 1214 { 1215 return -ENOENT; 1216 } 1217 1218 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl) 1219 { 1220 return 0; 1221 } 1222 1223 static inline int vfs_lock_file(struct file *filp, unsigned int cmd, 1224 struct file_lock *fl, struct file_lock *conf) 1225 { 1226 return -ENOLCK; 1227 } 1228 1229 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl) 1230 { 1231 return 0; 1232 } 1233 1234 static inline int flock_lock_file_wait(struct file *filp, 1235 struct file_lock *request) 1236 { 1237 return -ENOLCK; 1238 } 1239 1240 static inline int __break_lease(struct inode *inode, unsigned int mode) 1241 { 1242 return 0; 1243 } 1244 1245 static inline void lease_get_mtime(struct inode *inode, struct timespec *time) 1246 { 1247 return; 1248 } 1249 1250 static inline int generic_setlease(struct file *filp, long arg, 1251 struct file_lock **flp) 1252 { 1253 return -EINVAL; 1254 } 1255 1256 static inline int vfs_setlease(struct file *filp, long arg, 1257 struct file_lock **lease) 1258 { 1259 return -EINVAL; 1260 } 1261 1262 static inline int lease_modify(struct file_lock **before, int arg) 1263 { 1264 return -EINVAL; 1265 } 1266 1267 static inline int lock_may_read(struct inode *inode, loff_t start, 1268 unsigned long len) 1269 { 1270 return 1; 1271 } 1272 1273 static inline int lock_may_write(struct inode *inode, loff_t start, 1274 unsigned long len) 1275 { 1276 return 1; 1277 } 1278 1279 #endif /* !CONFIG_FILE_LOCKING */ 1280 1281 1282 struct fasync_struct { 1283 int magic; 1284 int fa_fd; 1285 struct fasync_struct *fa_next; /* singly linked list */ 1286 struct file *fa_file; 1287 }; 1288 1289 #define FASYNC_MAGIC 0x4601 1290 1291 /* SMP safe fasync helpers: */ 1292 extern int fasync_helper(int, struct file *, int, struct fasync_struct **); 1293 /* can be called from interrupts */ 1294 extern void kill_fasync(struct fasync_struct **, int, int); 1295 /* only for net: no internal synchronization */ 1296 extern void __kill_fasync(struct fasync_struct *, int, int); 1297 1298 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force); 1299 extern int f_setown(struct file *filp, unsigned long arg, int force); 1300 extern void f_delown(struct file *filp); 1301 extern pid_t f_getown(struct file *filp); 1302 extern int send_sigurg(struct fown_struct *fown); 1303 1304 /* 1305 * Umount options 1306 */ 1307 1308 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ 1309 #define MNT_DETACH 0x00000002 /* Just detach from the tree */ 1310 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ 1311 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ 1312 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ 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 char s_dirt; 1323 unsigned char s_blocksize_bits; 1324 unsigned long s_blocksize; 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 /* Granularity of c/m/atime in ns. 1366 Cannot be worse than a second */ 1367 u32 s_time_gran; 1368 1369 /* 1370 * The next field is for VFS *only*. No filesystems have any business 1371 * even looking at it. You had been warned. 1372 */ 1373 struct mutex s_vfs_rename_mutex; /* Kludge */ 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 *, struct writeback_control *wbc); 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_NEW 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 Serves as both a mutex and completion notification. 1605 * New inodes set I_NEW. If two processes both create 1606 * the same inode, one of them will release its inode and 1607 * wait for I_NEW to be released before returning. 1608 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can 1609 * also cause waiting on I_NEW, without I_NEW actually 1610 * being set. find_inode() uses this to prevent returning 1611 * nearly-dead inodes. 1612 * I_WILL_FREE Must be set when calling write_inode_now() if i_count 1613 * is zero. I_FREEING must be set when I_WILL_FREE is 1614 * cleared. 1615 * I_FREEING Set when inode is about to be freed but still has dirty 1616 * pages or buffers attached or the inode itself is still 1617 * dirty. 1618 * I_CLEAR Set by clear_inode(). In this state the inode is clean 1619 * and can be destroyed. 1620 * 1621 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are 1622 * prohibited for many purposes. iget() must wait for 1623 * the inode to be completely released, then create it 1624 * anew. Other functions will just ignore such inodes, 1625 * if appropriate. I_NEW is used for waiting. 1626 * 1627 * I_SYNC Synchonized write of dirty inode data. The bits is 1628 * set during data writeback, and cleared with a wakeup 1629 * on the bit address once it is done. 1630 * 1631 * Q: What is the difference between I_WILL_FREE and I_FREEING? 1632 */ 1633 #define I_DIRTY_SYNC 1 1634 #define I_DIRTY_DATASYNC 2 1635 #define I_DIRTY_PAGES 4 1636 #define __I_NEW 3 1637 #define I_NEW (1 << __I_NEW) 1638 #define I_WILL_FREE 16 1639 #define I_FREEING 32 1640 #define I_CLEAR 64 1641 #define __I_SYNC 7 1642 #define I_SYNC (1 << __I_SYNC) 1643 1644 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) 1645 1646 extern void __mark_inode_dirty(struct inode *, int); 1647 static inline void mark_inode_dirty(struct inode *inode) 1648 { 1649 __mark_inode_dirty(inode, I_DIRTY); 1650 } 1651 1652 static inline void mark_inode_dirty_sync(struct inode *inode) 1653 { 1654 __mark_inode_dirty(inode, I_DIRTY_SYNC); 1655 } 1656 1657 /** 1658 * inc_nlink - directly increment an inode's link count 1659 * @inode: inode 1660 * 1661 * This is a low-level filesystem helper to replace any 1662 * direct filesystem manipulation of i_nlink. Currently, 1663 * it is only here for parity with dec_nlink(). 1664 */ 1665 static inline void inc_nlink(struct inode *inode) 1666 { 1667 inode->i_nlink++; 1668 } 1669 1670 static inline void inode_inc_link_count(struct inode *inode) 1671 { 1672 inc_nlink(inode); 1673 mark_inode_dirty(inode); 1674 } 1675 1676 /** 1677 * drop_nlink - directly drop an inode's link count 1678 * @inode: inode 1679 * 1680 * This is a low-level filesystem helper to replace any 1681 * direct filesystem manipulation of i_nlink. In cases 1682 * where we are attempting to track writes to the 1683 * filesystem, a decrement to zero means an imminent 1684 * write when the file is truncated and actually unlinked 1685 * on the filesystem. 1686 */ 1687 static inline void drop_nlink(struct inode *inode) 1688 { 1689 inode->i_nlink--; 1690 } 1691 1692 /** 1693 * clear_nlink - directly zero an inode's link count 1694 * @inode: inode 1695 * 1696 * This is a low-level filesystem helper to replace any 1697 * direct filesystem manipulation of i_nlink. See 1698 * drop_nlink() for why we care about i_nlink hitting zero. 1699 */ 1700 static inline void clear_nlink(struct inode *inode) 1701 { 1702 inode->i_nlink = 0; 1703 } 1704 1705 static inline void inode_dec_link_count(struct inode *inode) 1706 { 1707 drop_nlink(inode); 1708 mark_inode_dirty(inode); 1709 } 1710 1711 /** 1712 * inode_inc_iversion - increments i_version 1713 * @inode: inode that need to be updated 1714 * 1715 * Every time the inode is modified, the i_version field will be incremented. 1716 * The filesystem has to be mounted with i_version flag 1717 */ 1718 1719 static inline void inode_inc_iversion(struct inode *inode) 1720 { 1721 spin_lock(&inode->i_lock); 1722 inode->i_version++; 1723 spin_unlock(&inode->i_lock); 1724 } 1725 1726 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry); 1727 static inline void file_accessed(struct file *file) 1728 { 1729 if (!(file->f_flags & O_NOATIME)) 1730 touch_atime(file->f_path.mnt, file->f_path.dentry); 1731 } 1732 1733 int sync_inode(struct inode *inode, struct writeback_control *wbc); 1734 1735 struct file_system_type { 1736 const char *name; 1737 int fs_flags; 1738 int (*get_sb) (struct file_system_type *, int, 1739 const char *, void *, struct vfsmount *); 1740 void (*kill_sb) (struct super_block *); 1741 struct module *owner; 1742 struct file_system_type * next; 1743 struct list_head fs_supers; 1744 1745 struct lock_class_key s_lock_key; 1746 struct lock_class_key s_umount_key; 1747 1748 struct lock_class_key i_lock_key; 1749 struct lock_class_key i_mutex_key; 1750 struct lock_class_key i_mutex_dir_key; 1751 struct lock_class_key i_alloc_sem_key; 1752 }; 1753 1754 extern int get_sb_ns(struct file_system_type *fs_type, int flags, void *data, 1755 int (*fill_super)(struct super_block *, void *, int), 1756 struct vfsmount *mnt); 1757 extern int get_sb_bdev(struct file_system_type *fs_type, 1758 int flags, const char *dev_name, void *data, 1759 int (*fill_super)(struct super_block *, void *, int), 1760 struct vfsmount *mnt); 1761 extern int get_sb_single(struct file_system_type *fs_type, 1762 int flags, void *data, 1763 int (*fill_super)(struct super_block *, void *, int), 1764 struct vfsmount *mnt); 1765 extern int get_sb_nodev(struct file_system_type *fs_type, 1766 int flags, void *data, 1767 int (*fill_super)(struct super_block *, void *, int), 1768 struct vfsmount *mnt); 1769 void generic_shutdown_super(struct super_block *sb); 1770 void kill_block_super(struct super_block *sb); 1771 void kill_anon_super(struct super_block *sb); 1772 void kill_litter_super(struct super_block *sb); 1773 void deactivate_super(struct super_block *sb); 1774 void deactivate_locked_super(struct super_block *sb); 1775 int set_anon_super(struct super_block *s, void *data); 1776 struct super_block *sget(struct file_system_type *type, 1777 int (*test)(struct super_block *,void *), 1778 int (*set)(struct super_block *,void *), 1779 void *data); 1780 extern int get_sb_pseudo(struct file_system_type *, char *, 1781 const struct super_operations *ops, unsigned long, 1782 struct vfsmount *mnt); 1783 extern void simple_set_mnt(struct vfsmount *mnt, struct super_block *sb); 1784 int __put_super_and_need_restart(struct super_block *sb); 1785 void put_super(struct super_block *sb); 1786 1787 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ 1788 #define fops_get(fops) \ 1789 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) 1790 #define fops_put(fops) \ 1791 do { if (fops) module_put((fops)->owner); } while(0) 1792 1793 extern int register_filesystem(struct file_system_type *); 1794 extern int unregister_filesystem(struct file_system_type *); 1795 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data); 1796 #define kern_mount(type) kern_mount_data(type, NULL) 1797 extern int may_umount_tree(struct vfsmount *); 1798 extern int may_umount(struct vfsmount *); 1799 extern long do_mount(char *, char *, char *, unsigned long, void *); 1800 extern struct vfsmount *collect_mounts(struct path *); 1801 extern void drop_collected_mounts(struct vfsmount *); 1802 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *, 1803 struct vfsmount *); 1804 extern int vfs_statfs(struct dentry *, struct kstatfs *); 1805 1806 extern int current_umask(void); 1807 1808 /* /sys/fs */ 1809 extern struct kobject *fs_kobj; 1810 1811 extern int rw_verify_area(int, struct file *, loff_t *, size_t); 1812 1813 #define FLOCK_VERIFY_READ 1 1814 #define FLOCK_VERIFY_WRITE 2 1815 1816 #ifdef CONFIG_FILE_LOCKING 1817 extern int locks_mandatory_locked(struct inode *); 1818 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t); 1819 1820 /* 1821 * Candidates for mandatory locking have the setgid bit set 1822 * but no group execute bit - an otherwise meaningless combination. 1823 */ 1824 1825 static inline int __mandatory_lock(struct inode *ino) 1826 { 1827 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; 1828 } 1829 1830 /* 1831 * ... and these candidates should be on MS_MANDLOCK mounted fs, 1832 * otherwise these will be advisory locks 1833 */ 1834 1835 static inline int mandatory_lock(struct inode *ino) 1836 { 1837 return IS_MANDLOCK(ino) && __mandatory_lock(ino); 1838 } 1839 1840 static inline int locks_verify_locked(struct inode *inode) 1841 { 1842 if (mandatory_lock(inode)) 1843 return locks_mandatory_locked(inode); 1844 return 0; 1845 } 1846 1847 static inline int locks_verify_truncate(struct inode *inode, 1848 struct file *filp, 1849 loff_t size) 1850 { 1851 if (inode->i_flock && mandatory_lock(inode)) 1852 return locks_mandatory_area( 1853 FLOCK_VERIFY_WRITE, inode, filp, 1854 size < inode->i_size ? size : inode->i_size, 1855 (size < inode->i_size ? inode->i_size - size 1856 : size - inode->i_size) 1857 ); 1858 return 0; 1859 } 1860 1861 static inline int break_lease(struct inode *inode, unsigned int mode) 1862 { 1863 if (inode->i_flock) 1864 return __break_lease(inode, mode); 1865 return 0; 1866 } 1867 #else /* !CONFIG_FILE_LOCKING */ 1868 static inline int locks_mandatory_locked(struct inode *inode) 1869 { 1870 return 0; 1871 } 1872 1873 static inline int locks_mandatory_area(int rw, struct inode *inode, 1874 struct file *filp, loff_t offset, 1875 size_t count) 1876 { 1877 return 0; 1878 } 1879 1880 static inline int __mandatory_lock(struct inode *inode) 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 locks_verify_locked(struct inode *inode) 1891 { 1892 return 0; 1893 } 1894 1895 static inline int locks_verify_truncate(struct inode *inode, struct file *filp, 1896 size_t size) 1897 { 1898 return 0; 1899 } 1900 1901 static inline int break_lease(struct inode *inode, unsigned int mode) 1902 { 1903 return 0; 1904 } 1905 1906 #endif /* CONFIG_FILE_LOCKING */ 1907 1908 /* fs/open.c */ 1909 1910 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs, 1911 struct file *filp); 1912 extern int do_fallocate(struct file *file, int mode, loff_t offset, 1913 loff_t len); 1914 extern long do_sys_open(int dfd, const char __user *filename, int flags, 1915 int mode); 1916 extern struct file *filp_open(const char *, int, int); 1917 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int, 1918 const struct cred *); 1919 extern int filp_close(struct file *, fl_owner_t id); 1920 extern char * getname(const char __user *); 1921 1922 /* fs/ioctl.c */ 1923 1924 extern int ioctl_preallocate(struct file *filp, void __user *argp); 1925 1926 /* fs/dcache.c */ 1927 extern void __init vfs_caches_init_early(void); 1928 extern void __init vfs_caches_init(unsigned long); 1929 1930 extern struct kmem_cache *names_cachep; 1931 1932 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp)) 1933 #define __getname() __getname_gfp(GFP_KERNEL) 1934 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) 1935 #ifndef CONFIG_AUDITSYSCALL 1936 #define putname(name) __putname(name) 1937 #else 1938 extern void putname(const char *name); 1939 #endif 1940 1941 #ifdef CONFIG_BLOCK 1942 extern int register_blkdev(unsigned int, const char *); 1943 extern void unregister_blkdev(unsigned int, const char *); 1944 extern struct block_device *bdget(dev_t); 1945 extern struct block_device *bdgrab(struct block_device *bdev); 1946 extern void bd_set_size(struct block_device *, loff_t size); 1947 extern void bd_forget(struct inode *inode); 1948 extern void bdput(struct block_device *); 1949 extern struct block_device *open_by_devnum(dev_t, fmode_t); 1950 extern void invalidate_bdev(struct block_device *); 1951 extern int sync_blockdev(struct block_device *bdev); 1952 extern struct super_block *freeze_bdev(struct block_device *); 1953 extern void emergency_thaw_all(void); 1954 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb); 1955 extern int fsync_bdev(struct block_device *); 1956 #else 1957 static inline void bd_forget(struct inode *inode) {} 1958 static inline int sync_blockdev(struct block_device *bdev) { return 0; } 1959 static inline void invalidate_bdev(struct block_device *bdev) {} 1960 1961 static inline struct super_block *freeze_bdev(struct block_device *sb) 1962 { 1963 return NULL; 1964 } 1965 1966 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb) 1967 { 1968 return 0; 1969 } 1970 #endif 1971 extern int sync_filesystem(struct super_block *); 1972 extern const struct file_operations def_blk_fops; 1973 extern const struct file_operations def_chr_fops; 1974 extern const struct file_operations bad_sock_fops; 1975 extern const struct file_operations def_fifo_fops; 1976 #ifdef CONFIG_BLOCK 1977 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); 1978 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long); 1979 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); 1980 extern int blkdev_get(struct block_device *, fmode_t); 1981 extern int blkdev_put(struct block_device *, fmode_t); 1982 extern int bd_claim(struct block_device *, void *); 1983 extern void bd_release(struct block_device *); 1984 #ifdef CONFIG_SYSFS 1985 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *); 1986 extern void bd_release_from_disk(struct block_device *, struct gendisk *); 1987 #else 1988 #define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder) 1989 #define bd_release_from_disk(bdev, disk) bd_release(bdev) 1990 #endif 1991 #endif 1992 1993 /* fs/char_dev.c */ 1994 #define CHRDEV_MAJOR_HASH_SIZE 255 1995 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); 1996 extern int register_chrdev_region(dev_t, unsigned, const char *); 1997 extern int __register_chrdev(unsigned int major, unsigned int baseminor, 1998 unsigned int count, const char *name, 1999 const struct file_operations *fops); 2000 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, 2001 unsigned int count, const char *name); 2002 extern void unregister_chrdev_region(dev_t, unsigned); 2003 extern void chrdev_show(struct seq_file *,off_t); 2004 2005 static inline int register_chrdev(unsigned int major, const char *name, 2006 const struct file_operations *fops) 2007 { 2008 return __register_chrdev(major, 0, 256, name, fops); 2009 } 2010 2011 static inline void unregister_chrdev(unsigned int major, const char *name) 2012 { 2013 __unregister_chrdev(major, 0, 256, name); 2014 } 2015 2016 /* fs/block_dev.c */ 2017 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ 2018 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ 2019 2020 #ifdef CONFIG_BLOCK 2021 #define BLKDEV_MAJOR_HASH_SIZE 255 2022 extern const char *__bdevname(dev_t, char *buffer); 2023 extern const char *bdevname(struct block_device *bdev, char *buffer); 2024 extern struct block_device *lookup_bdev(const char *); 2025 extern struct block_device *open_bdev_exclusive(const char *, fmode_t, void *); 2026 extern void close_bdev_exclusive(struct block_device *, fmode_t); 2027 extern void blkdev_show(struct seq_file *,off_t); 2028 2029 #else 2030 #define BLKDEV_MAJOR_HASH_SIZE 0 2031 #endif 2032 2033 extern void init_special_inode(struct inode *, umode_t, dev_t); 2034 2035 /* Invalid inode operations -- fs/bad_inode.c */ 2036 extern void make_bad_inode(struct inode *); 2037 extern int is_bad_inode(struct inode *); 2038 2039 extern const struct file_operations read_pipefifo_fops; 2040 extern const struct file_operations write_pipefifo_fops; 2041 extern const struct file_operations rdwr_pipefifo_fops; 2042 2043 extern int fs_may_remount_ro(struct super_block *); 2044 2045 #ifdef CONFIG_BLOCK 2046 /* 2047 * return READ, READA, or WRITE 2048 */ 2049 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK)) 2050 2051 /* 2052 * return data direction, READ or WRITE 2053 */ 2054 #define bio_data_dir(bio) ((bio)->bi_rw & 1) 2055 2056 extern void check_disk_size_change(struct gendisk *disk, 2057 struct block_device *bdev); 2058 extern int revalidate_disk(struct gendisk *); 2059 extern int check_disk_change(struct block_device *); 2060 extern int __invalidate_device(struct block_device *); 2061 extern int invalidate_partition(struct gendisk *, int); 2062 #endif 2063 extern int invalidate_inodes(struct super_block *); 2064 unsigned long invalidate_mapping_pages(struct address_space *mapping, 2065 pgoff_t start, pgoff_t end); 2066 2067 static inline void invalidate_remote_inode(struct inode *inode) 2068 { 2069 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || 2070 S_ISLNK(inode->i_mode)) 2071 invalidate_mapping_pages(inode->i_mapping, 0, -1); 2072 } 2073 extern int invalidate_inode_pages2(struct address_space *mapping); 2074 extern int invalidate_inode_pages2_range(struct address_space *mapping, 2075 pgoff_t start, pgoff_t end); 2076 extern int write_inode_now(struct inode *, int); 2077 extern int filemap_fdatawrite(struct address_space *); 2078 extern int filemap_flush(struct address_space *); 2079 extern int filemap_fdatawait(struct address_space *); 2080 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart, 2081 loff_t lend); 2082 extern int filemap_write_and_wait(struct address_space *mapping); 2083 extern int filemap_write_and_wait_range(struct address_space *mapping, 2084 loff_t lstart, loff_t lend); 2085 extern int __filemap_fdatawrite_range(struct address_space *mapping, 2086 loff_t start, loff_t end, int sync_mode); 2087 extern int filemap_fdatawrite_range(struct address_space *mapping, 2088 loff_t start, loff_t end); 2089 2090 extern int vfs_fsync_range(struct file *file, struct dentry *dentry, 2091 loff_t start, loff_t end, int datasync); 2092 extern int vfs_fsync(struct file *file, struct dentry *dentry, int datasync); 2093 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count); 2094 extern void sync_supers(void); 2095 extern void emergency_sync(void); 2096 extern void emergency_remount(void); 2097 #ifdef CONFIG_BLOCK 2098 extern sector_t bmap(struct inode *, sector_t); 2099 #endif 2100 extern int notify_change(struct dentry *, struct iattr *); 2101 extern int inode_permission(struct inode *, int); 2102 extern int generic_permission(struct inode *, int, 2103 int (*check_acl)(struct inode *, int)); 2104 2105 static inline bool execute_ok(struct inode *inode) 2106 { 2107 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); 2108 } 2109 2110 extern int get_write_access(struct inode *); 2111 extern int deny_write_access(struct file *); 2112 static inline void put_write_access(struct inode * inode) 2113 { 2114 atomic_dec(&inode->i_writecount); 2115 } 2116 static inline void allow_write_access(struct file *file) 2117 { 2118 if (file) 2119 atomic_inc(&file->f_path.dentry->d_inode->i_writecount); 2120 } 2121 extern int do_pipe_flags(int *, int); 2122 extern struct file *create_read_pipe(struct file *f, int flags); 2123 extern struct file *create_write_pipe(int flags); 2124 extern void free_write_pipe(struct file *); 2125 2126 extern struct file *do_filp_open(int dfd, const char *pathname, 2127 int open_flag, int mode, int acc_mode); 2128 extern int may_open(struct path *, int, int); 2129 2130 extern int kernel_read(struct file *, loff_t, char *, unsigned long); 2131 extern struct file * open_exec(const char *); 2132 2133 /* fs/dcache.c -- generic fs support functions */ 2134 extern int is_subdir(struct dentry *, struct dentry *); 2135 extern int path_is_under(struct path *, struct path *); 2136 extern ino_t find_inode_number(struct dentry *, struct qstr *); 2137 2138 #include <linux/err.h> 2139 2140 /* needed for stackable file system support */ 2141 extern loff_t default_llseek(struct file *file, loff_t offset, int origin); 2142 2143 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin); 2144 2145 extern int inode_init_always(struct super_block *, struct inode *); 2146 extern void inode_init_once(struct inode *); 2147 extern void inode_add_to_lists(struct super_block *, struct inode *); 2148 extern void iput(struct inode *); 2149 extern struct inode * igrab(struct inode *); 2150 extern ino_t iunique(struct super_block *, ino_t); 2151 extern int inode_needs_sync(struct inode *inode); 2152 extern void generic_delete_inode(struct inode *inode); 2153 extern void generic_drop_inode(struct inode *inode); 2154 extern int generic_detach_inode(struct inode *inode); 2155 2156 extern struct inode *ilookup5_nowait(struct super_block *sb, 2157 unsigned long hashval, int (*test)(struct inode *, void *), 2158 void *data); 2159 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, 2160 int (*test)(struct inode *, void *), void *data); 2161 extern struct inode *ilookup(struct super_block *sb, unsigned long ino); 2162 2163 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); 2164 extern struct inode * iget_locked(struct super_block *, unsigned long); 2165 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); 2166 extern int insert_inode_locked(struct inode *); 2167 extern void unlock_new_inode(struct inode *); 2168 2169 extern void __iget(struct inode * inode); 2170 extern void iget_failed(struct inode *); 2171 extern void clear_inode(struct inode *); 2172 extern void destroy_inode(struct inode *); 2173 extern void __destroy_inode(struct inode *); 2174 extern struct inode *new_inode(struct super_block *); 2175 extern int should_remove_suid(struct dentry *); 2176 extern int file_remove_suid(struct file *); 2177 2178 extern void __insert_inode_hash(struct inode *, unsigned long hashval); 2179 extern void remove_inode_hash(struct inode *); 2180 static inline void insert_inode_hash(struct inode *inode) { 2181 __insert_inode_hash(inode, inode->i_ino); 2182 } 2183 2184 extern void file_move(struct file *f, struct list_head *list); 2185 extern void file_kill(struct file *f); 2186 #ifdef CONFIG_BLOCK 2187 struct bio; 2188 extern void submit_bio(int, struct bio *); 2189 extern int bdev_read_only(struct block_device *); 2190 #endif 2191 extern int set_blocksize(struct block_device *, int); 2192 extern int sb_set_blocksize(struct super_block *, int); 2193 extern int sb_min_blocksize(struct super_block *, int); 2194 2195 extern int generic_file_mmap(struct file *, struct vm_area_struct *); 2196 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); 2197 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size); 2198 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk); 2199 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t); 2200 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, 2201 loff_t *); 2202 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t); 2203 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *, 2204 unsigned long *, loff_t, loff_t *, size_t, size_t); 2205 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *, 2206 unsigned long, loff_t, loff_t *, size_t, ssize_t); 2207 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos); 2208 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos); 2209 extern int generic_segment_checks(const struct iovec *iov, 2210 unsigned long *nr_segs, size_t *count, int access_flags); 2211 2212 /* fs/block_dev.c */ 2213 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov, 2214 unsigned long nr_segs, loff_t pos); 2215 extern int blkdev_fsync(struct file *filp, struct dentry *dentry, int datasync); 2216 2217 /* fs/splice.c */ 2218 extern ssize_t generic_file_splice_read(struct file *, loff_t *, 2219 struct pipe_inode_info *, size_t, unsigned int); 2220 extern ssize_t default_file_splice_read(struct file *, loff_t *, 2221 struct pipe_inode_info *, size_t, unsigned int); 2222 extern ssize_t generic_file_splice_write(struct pipe_inode_info *, 2223 struct file *, loff_t *, size_t, unsigned int); 2224 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, 2225 struct file *out, loff_t *, size_t len, unsigned int flags); 2226 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, 2227 size_t len, unsigned int flags); 2228 2229 extern void 2230 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); 2231 extern loff_t no_llseek(struct file *file, loff_t offset, int origin); 2232 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin); 2233 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset, 2234 int origin); 2235 extern int generic_file_open(struct inode * inode, struct file * filp); 2236 extern int nonseekable_open(struct inode * inode, struct file * filp); 2237 2238 #ifdef CONFIG_FS_XIP 2239 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len, 2240 loff_t *ppos); 2241 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma); 2242 extern ssize_t xip_file_write(struct file *filp, const char __user *buf, 2243 size_t len, loff_t *ppos); 2244 extern int xip_truncate_page(struct address_space *mapping, loff_t from); 2245 #else 2246 static inline int xip_truncate_page(struct address_space *mapping, loff_t from) 2247 { 2248 return 0; 2249 } 2250 #endif 2251 2252 #ifdef CONFIG_BLOCK 2253 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode, 2254 struct block_device *bdev, const struct iovec *iov, loff_t offset, 2255 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io, 2256 int lock_type); 2257 2258 enum { 2259 /* need locking between buffered and direct access */ 2260 DIO_LOCKING = 0x01, 2261 2262 /* filesystem does not support filling holes */ 2263 DIO_SKIP_HOLES = 0x02, 2264 }; 2265 2266 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb, 2267 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 2268 loff_t offset, unsigned long nr_segs, get_block_t get_block, 2269 dio_iodone_t end_io) 2270 { 2271 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 2272 nr_segs, get_block, end_io, 2273 DIO_LOCKING | DIO_SKIP_HOLES); 2274 } 2275 2276 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb, 2277 struct inode *inode, struct block_device *bdev, const struct iovec *iov, 2278 loff_t offset, unsigned long nr_segs, get_block_t get_block, 2279 dio_iodone_t end_io) 2280 { 2281 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset, 2282 nr_segs, get_block, end_io, 0); 2283 } 2284 #endif 2285 2286 extern const struct file_operations generic_ro_fops; 2287 2288 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) 2289 2290 extern int vfs_readlink(struct dentry *, char __user *, int, const char *); 2291 extern int vfs_follow_link(struct nameidata *, const char *); 2292 extern int page_readlink(struct dentry *, char __user *, int); 2293 extern void *page_follow_link_light(struct dentry *, struct nameidata *); 2294 extern void page_put_link(struct dentry *, struct nameidata *, void *); 2295 extern int __page_symlink(struct inode *inode, const char *symname, int len, 2296 int nofs); 2297 extern int page_symlink(struct inode *inode, const char *symname, int len); 2298 extern const struct inode_operations page_symlink_inode_operations; 2299 extern int generic_readlink(struct dentry *, char __user *, int); 2300 extern void generic_fillattr(struct inode *, struct kstat *); 2301 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *); 2302 void __inode_add_bytes(struct inode *inode, loff_t bytes); 2303 void inode_add_bytes(struct inode *inode, loff_t bytes); 2304 void inode_sub_bytes(struct inode *inode, loff_t bytes); 2305 loff_t inode_get_bytes(struct inode *inode); 2306 void inode_set_bytes(struct inode *inode, loff_t bytes); 2307 2308 extern int vfs_readdir(struct file *, filldir_t, void *); 2309 2310 extern int vfs_stat(char __user *, struct kstat *); 2311 extern int vfs_lstat(char __user *, struct kstat *); 2312 extern int vfs_fstat(unsigned int, struct kstat *); 2313 extern int vfs_fstatat(int , char __user *, struct kstat *, int); 2314 2315 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd, 2316 unsigned long arg); 2317 extern int __generic_block_fiemap(struct inode *inode, 2318 struct fiemap_extent_info *fieinfo, 2319 loff_t start, loff_t len, 2320 get_block_t *get_block); 2321 extern int generic_block_fiemap(struct inode *inode, 2322 struct fiemap_extent_info *fieinfo, u64 start, 2323 u64 len, get_block_t *get_block); 2324 2325 extern void get_filesystem(struct file_system_type *fs); 2326 extern void put_filesystem(struct file_system_type *fs); 2327 extern struct file_system_type *get_fs_type(const char *name); 2328 extern struct super_block *get_super(struct block_device *); 2329 extern struct super_block *get_active_super(struct block_device *bdev); 2330 extern struct super_block *user_get_super(dev_t); 2331 extern void drop_super(struct super_block *sb); 2332 2333 extern int dcache_dir_open(struct inode *, struct file *); 2334 extern int dcache_dir_close(struct inode *, struct file *); 2335 extern loff_t dcache_dir_lseek(struct file *, loff_t, int); 2336 extern int dcache_readdir(struct file *, void *, filldir_t); 2337 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *); 2338 extern int simple_statfs(struct dentry *, struct kstatfs *); 2339 extern int simple_link(struct dentry *, struct inode *, struct dentry *); 2340 extern int simple_unlink(struct inode *, struct dentry *); 2341 extern int simple_rmdir(struct inode *, struct dentry *); 2342 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); 2343 extern int simple_sync_file(struct file *, struct dentry *, int); 2344 extern int simple_empty(struct dentry *); 2345 extern int simple_readpage(struct file *file, struct page *page); 2346 extern int simple_write_begin(struct file *file, struct address_space *mapping, 2347 loff_t pos, unsigned len, unsigned flags, 2348 struct page **pagep, void **fsdata); 2349 extern int simple_write_end(struct file *file, struct address_space *mapping, 2350 loff_t pos, unsigned len, unsigned copied, 2351 struct page *page, void *fsdata); 2352 2353 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *); 2354 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); 2355 extern const struct file_operations simple_dir_operations; 2356 extern const struct inode_operations simple_dir_inode_operations; 2357 struct tree_descr { char *name; const struct file_operations *ops; int mode; }; 2358 struct dentry *d_alloc_name(struct dentry *, const char *); 2359 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *); 2360 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); 2361 extern void simple_release_fs(struct vfsmount **mount, int *count); 2362 2363 extern ssize_t simple_read_from_buffer(void __user *to, size_t count, 2364 loff_t *ppos, const void *from, size_t available); 2365 2366 extern int simple_fsync(struct file *, struct dentry *, int); 2367 2368 #ifdef CONFIG_MIGRATION 2369 extern int buffer_migrate_page(struct address_space *, 2370 struct page *, struct page *); 2371 #else 2372 #define buffer_migrate_page NULL 2373 #endif 2374 2375 extern int inode_change_ok(const struct inode *, struct iattr *); 2376 extern int inode_newsize_ok(const struct inode *, loff_t offset); 2377 extern int __must_check inode_setattr(struct inode *, struct iattr *); 2378 2379 extern void file_update_time(struct file *file); 2380 2381 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt); 2382 extern void save_mount_options(struct super_block *sb, char *options); 2383 extern void replace_mount_options(struct super_block *sb, char *options); 2384 2385 static inline ino_t parent_ino(struct dentry *dentry) 2386 { 2387 ino_t res; 2388 2389 spin_lock(&dentry->d_lock); 2390 res = dentry->d_parent->d_inode->i_ino; 2391 spin_unlock(&dentry->d_lock); 2392 return res; 2393 } 2394 2395 /* Transaction based IO helpers */ 2396 2397 /* 2398 * An argresp is stored in an allocated page and holds the 2399 * size of the argument or response, along with its content 2400 */ 2401 struct simple_transaction_argresp { 2402 ssize_t size; 2403 char data[0]; 2404 }; 2405 2406 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) 2407 2408 char *simple_transaction_get(struct file *file, const char __user *buf, 2409 size_t size); 2410 ssize_t simple_transaction_read(struct file *file, char __user *buf, 2411 size_t size, loff_t *pos); 2412 int simple_transaction_release(struct inode *inode, struct file *file); 2413 2414 void simple_transaction_set(struct file *file, size_t n); 2415 2416 /* 2417 * simple attribute files 2418 * 2419 * These attributes behave similar to those in sysfs: 2420 * 2421 * Writing to an attribute immediately sets a value, an open file can be 2422 * written to multiple times. 2423 * 2424 * Reading from an attribute creates a buffer from the value that might get 2425 * read with multiple read calls. When the attribute has been read 2426 * completely, no further read calls are possible until the file is opened 2427 * again. 2428 * 2429 * All attributes contain a text representation of a numeric value 2430 * that are accessed with the get() and set() functions. 2431 */ 2432 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ 2433 static int __fops ## _open(struct inode *inode, struct file *file) \ 2434 { \ 2435 __simple_attr_check_format(__fmt, 0ull); \ 2436 return simple_attr_open(inode, file, __get, __set, __fmt); \ 2437 } \ 2438 static const struct file_operations __fops = { \ 2439 .owner = THIS_MODULE, \ 2440 .open = __fops ## _open, \ 2441 .release = simple_attr_release, \ 2442 .read = simple_attr_read, \ 2443 .write = simple_attr_write, \ 2444 }; 2445 2446 static inline void __attribute__((format(printf, 1, 2))) 2447 __simple_attr_check_format(const char *fmt, ...) 2448 { 2449 /* don't do anything, just let the compiler check the arguments; */ 2450 } 2451 2452 int simple_attr_open(struct inode *inode, struct file *file, 2453 int (*get)(void *, u64 *), int (*set)(void *, u64), 2454 const char *fmt); 2455 int simple_attr_release(struct inode *inode, struct file *file); 2456 ssize_t simple_attr_read(struct file *file, char __user *buf, 2457 size_t len, loff_t *ppos); 2458 ssize_t simple_attr_write(struct file *file, const char __user *buf, 2459 size_t len, loff_t *ppos); 2460 2461 struct ctl_table; 2462 int proc_nr_files(struct ctl_table *table, int write, 2463 void __user *buffer, size_t *lenp, loff_t *ppos); 2464 2465 int __init get_filesystem_list(char *buf); 2466 2467 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) 2468 #define OPEN_FMODE(flag) ((__force fmode_t)((flag + 1) & O_ACCMODE)) 2469 2470 #endif /* __KERNEL__ */ 2471 #endif /* _LINUX_FS_H */