1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * linux/fs/stat.c
4 *
5 * Copyright (C) 1991, 1992 Linus Torvalds
6 */
7
8 #include <linux/blkdev.h>
9 #include <linux/export.h>
10 #include <linux/mm.h>
11 #include <linux/errno.h>
12 #include <linux/file.h>
13 #include <linux/highuid.h>
14 #include <linux/fs.h>
15 #include <linux/namei.h>
16 #include <linux/security.h>
17 #include <linux/cred.h>
18 #include <linux/syscalls.h>
19 #include <linux/pagemap.h>
20 #include <linux/compat.h>
21 #include <linux/iversion.h>
22
23 #include <linux/uaccess.h>
24 #include <asm/unistd.h>
25
26 #include <trace/events/timestamp.h>
27
28 #include "internal.h"
29 #include "mount.h"
30
31 /**
32 * fill_mg_cmtime - Fill in the mtime and ctime and flag ctime as QUERIED
33 * @stat: where to store the resulting values
34 * @request_mask: STATX_* values requested
35 * @inode: inode from which to grab the c/mtime
36 *
37 * Given @inode, grab the ctime and mtime out if it and store the result
38 * in @stat. When fetching the value, flag it as QUERIED (if not already)
39 * so the next write will record a distinct timestamp.
40 *
41 * NB: The QUERIED flag is tracked in the ctime, but we set it there even
42 * if only the mtime was requested, as that ensures that the next mtime
43 * change will be distinct.
44 */
fill_mg_cmtime(struct kstat * stat,u32 request_mask,struct inode * inode)45 void fill_mg_cmtime(struct kstat *stat, u32 request_mask, struct inode *inode)
46 {
47 atomic_t *pcn = (atomic_t *)&inode->i_ctime_nsec;
48
49 /* If neither time was requested, then don't report them */
50 if (!(request_mask & (STATX_CTIME|STATX_MTIME))) {
51 stat->result_mask &= ~(STATX_CTIME|STATX_MTIME);
52 return;
53 }
54
55 stat->mtime = inode_get_mtime(inode);
56 stat->ctime.tv_sec = inode->i_ctime_sec;
57 stat->ctime.tv_nsec = (u32)atomic_read(pcn);
58 if (!(stat->ctime.tv_nsec & I_CTIME_QUERIED))
59 stat->ctime.tv_nsec = ((u32)atomic_fetch_or(I_CTIME_QUERIED, pcn));
60 stat->ctime.tv_nsec &= ~I_CTIME_QUERIED;
61 trace_fill_mg_cmtime(inode, &stat->ctime, &stat->mtime);
62 }
63 EXPORT_SYMBOL(fill_mg_cmtime);
64
65 /**
66 * generic_fillattr - Fill in the basic attributes from the inode struct
67 * @idmap: idmap of the mount the inode was found from
68 * @request_mask: statx request_mask
69 * @inode: Inode to use as the source
70 * @stat: Where to fill in the attributes
71 *
72 * Fill in the basic attributes in the kstat structure from data that's to be
73 * found on the VFS inode structure. This is the default if no getattr inode
74 * operation is supplied.
75 *
76 * If the inode has been found through an idmapped mount the idmap of
77 * the vfsmount must be passed through @idmap. This function will then
78 * take care to map the inode according to @idmap before filling in the
79 * uid and gid filds. On non-idmapped mounts or if permission checking is to be
80 * performed on the raw inode simply pass @nop_mnt_idmap.
81 */
generic_fillattr(struct mnt_idmap * idmap,u32 request_mask,struct inode * inode,struct kstat * stat)82 void generic_fillattr(struct mnt_idmap *idmap, u32 request_mask,
83 struct inode *inode, struct kstat *stat)
84 {
85 vfsuid_t vfsuid = i_uid_into_vfsuid(idmap, inode);
86 vfsgid_t vfsgid = i_gid_into_vfsgid(idmap, inode);
87
88 stat->dev = inode->i_sb->s_dev;
89 stat->ino = inode->i_ino;
90 stat->mode = inode->i_mode;
91 stat->nlink = inode->i_nlink;
92 stat->uid = vfsuid_into_kuid(vfsuid);
93 stat->gid = vfsgid_into_kgid(vfsgid);
94 stat->rdev = inode->i_rdev;
95 stat->size = i_size_read(inode);
96 stat->atime = inode_get_atime(inode);
97
98 if (is_mgtime(inode)) {
99 fill_mg_cmtime(stat, request_mask, inode);
100 } else {
101 stat->ctime = inode_get_ctime(inode);
102 stat->mtime = inode_get_mtime(inode);
103 }
104
105 stat->blksize = i_blocksize(inode);
106 stat->blocks = inode->i_blocks;
107
108 if ((request_mask & STATX_CHANGE_COOKIE) && IS_I_VERSION(inode)) {
109 stat->result_mask |= STATX_CHANGE_COOKIE;
110 stat->change_cookie = inode_query_iversion(inode);
111 }
112
113 }
114 EXPORT_SYMBOL(generic_fillattr);
115
116 /**
117 * generic_fill_statx_attr - Fill in the statx attributes from the inode flags
118 * @inode: Inode to use as the source
119 * @stat: Where to fill in the attribute flags
120 *
121 * Fill in the STATX_ATTR_* flags in the kstat structure for properties of the
122 * inode that are published on i_flags and enforced by the VFS.
123 */
generic_fill_statx_attr(struct inode * inode,struct kstat * stat)124 void generic_fill_statx_attr(struct inode *inode, struct kstat *stat)
125 {
126 if (inode->i_flags & S_IMMUTABLE)
127 stat->attributes |= STATX_ATTR_IMMUTABLE;
128 if (inode->i_flags & S_APPEND)
129 stat->attributes |= STATX_ATTR_APPEND;
130 stat->attributes_mask |= KSTAT_ATTR_VFS_FLAGS;
131 }
132 EXPORT_SYMBOL(generic_fill_statx_attr);
133
134 /**
135 * generic_fill_statx_atomic_writes - Fill in atomic writes statx attributes
136 * @stat: Where to fill in the attribute flags
137 * @unit_min: Minimum supported atomic write length in bytes
138 * @unit_max: Maximum supported atomic write length in bytes
139 *
140 * Fill in the STATX{_ATTR}_WRITE_ATOMIC flags in the kstat structure from
141 * atomic write unit_min and unit_max values.
142 */
generic_fill_statx_atomic_writes(struct kstat * stat,unsigned int unit_min,unsigned int unit_max)143 void generic_fill_statx_atomic_writes(struct kstat *stat,
144 unsigned int unit_min,
145 unsigned int unit_max)
146 {
147 /* Confirm that the request type is known */
148 stat->result_mask |= STATX_WRITE_ATOMIC;
149
150 /* Confirm that the file attribute type is known */
151 stat->attributes_mask |= STATX_ATTR_WRITE_ATOMIC;
152
153 if (unit_min) {
154 stat->atomic_write_unit_min = unit_min;
155 stat->atomic_write_unit_max = unit_max;
156 /* Initially only allow 1x segment */
157 stat->atomic_write_segments_max = 1;
158
159 /* Confirm atomic writes are actually supported */
160 stat->attributes |= STATX_ATTR_WRITE_ATOMIC;
161 }
162 }
163 EXPORT_SYMBOL_GPL(generic_fill_statx_atomic_writes);
164
165 /**
166 * vfs_getattr_nosec - getattr without security checks
167 * @path: file to get attributes from
168 * @stat: structure to return attributes in
169 * @request_mask: STATX_xxx flags indicating what the caller wants
170 * @query_flags: Query mode (AT_STATX_SYNC_TYPE)
171 *
172 * Get attributes without calling security_inode_getattr.
173 *
174 * Currently the only caller other than vfs_getattr is internal to the
175 * filehandle lookup code, which uses only the inode number and returns no
176 * attributes to any user. Any other code probably wants vfs_getattr.
177 */
vfs_getattr_nosec(const struct path * path,struct kstat * stat,u32 request_mask,unsigned int query_flags)178 int vfs_getattr_nosec(const struct path *path, struct kstat *stat,
179 u32 request_mask, unsigned int query_flags)
180 {
181 struct mnt_idmap *idmap;
182 struct inode *inode = d_backing_inode(path->dentry);
183
184 memset(stat, 0, sizeof(*stat));
185 stat->result_mask |= STATX_BASIC_STATS;
186 query_flags &= AT_STATX_SYNC_TYPE;
187
188 /* allow the fs to override these if it really wants to */
189 /* SB_NOATIME means filesystem supplies dummy atime value */
190 if (inode->i_sb->s_flags & SB_NOATIME)
191 stat->result_mask &= ~STATX_ATIME;
192
193 /*
194 * Note: If you add another clause to set an attribute flag, please
195 * update attributes_mask below.
196 */
197 if (IS_AUTOMOUNT(inode))
198 stat->attributes |= STATX_ATTR_AUTOMOUNT;
199
200 if (IS_DAX(inode))
201 stat->attributes |= STATX_ATTR_DAX;
202
203 stat->attributes_mask |= (STATX_ATTR_AUTOMOUNT |
204 STATX_ATTR_DAX);
205
206 idmap = mnt_idmap(path->mnt);
207 if (inode->i_op->getattr)
208 return inode->i_op->getattr(idmap, path, stat,
209 request_mask,
210 query_flags);
211
212 generic_fillattr(idmap, request_mask, inode, stat);
213 return 0;
214 }
215 EXPORT_SYMBOL(vfs_getattr_nosec);
216
217 /*
218 * vfs_getattr - Get the enhanced basic attributes of a file
219 * @path: The file of interest
220 * @stat: Where to return the statistics
221 * @request_mask: STATX_xxx flags indicating what the caller wants
222 * @query_flags: Query mode (AT_STATX_SYNC_TYPE)
223 *
224 * Ask the filesystem for a file's attributes. The caller must indicate in
225 * request_mask and query_flags to indicate what they want.
226 *
227 * If the file is remote, the filesystem can be forced to update the attributes
228 * from the backing store by passing AT_STATX_FORCE_SYNC in query_flags or can
229 * suppress the update by passing AT_STATX_DONT_SYNC.
230 *
231 * Bits must have been set in request_mask to indicate which attributes the
232 * caller wants retrieving. Any such attribute not requested may be returned
233 * anyway, but the value may be approximate, and, if remote, may not have been
234 * synchronised with the server.
235 *
236 * 0 will be returned on success, and a -ve error code if unsuccessful.
237 */
vfs_getattr(const struct path * path,struct kstat * stat,u32 request_mask,unsigned int query_flags)238 int vfs_getattr(const struct path *path, struct kstat *stat,
239 u32 request_mask, unsigned int query_flags)
240 {
241 int retval;
242
243 retval = security_inode_getattr(path);
244 if (retval)
245 return retval;
246 return vfs_getattr_nosec(path, stat, request_mask, query_flags);
247 }
248 EXPORT_SYMBOL(vfs_getattr);
249
250 /**
251 * vfs_fstat - Get the basic attributes by file descriptor
252 * @fd: The file descriptor referring to the file of interest
253 * @stat: The result structure to fill in.
254 *
255 * This function is a wrapper around vfs_getattr(). The main difference is
256 * that it uses a file descriptor to determine the file location.
257 *
258 * 0 will be returned on success, and a -ve error code if unsuccessful.
259 */
vfs_fstat(int fd,struct kstat * stat)260 int vfs_fstat(int fd, struct kstat *stat)
261 {
262 CLASS(fd_raw, f)(fd);
263 if (fd_empty(f))
264 return -EBADF;
265 return vfs_getattr(&fd_file(f)->f_path, stat, STATX_BASIC_STATS, 0);
266 }
267
statx_lookup_flags(int flags)268 static int statx_lookup_flags(int flags)
269 {
270 int lookup_flags = 0;
271
272 if (!(flags & AT_SYMLINK_NOFOLLOW))
273 lookup_flags |= LOOKUP_FOLLOW;
274 if (!(flags & AT_NO_AUTOMOUNT))
275 lookup_flags |= LOOKUP_AUTOMOUNT;
276
277 return lookup_flags;
278 }
279
vfs_statx_path(struct path * path,int flags,struct kstat * stat,u32 request_mask)280 static int vfs_statx_path(struct path *path, int flags, struct kstat *stat,
281 u32 request_mask)
282 {
283 int error = vfs_getattr(path, stat, request_mask, flags);
284 if (error)
285 return error;
286
287 if (request_mask & STATX_MNT_ID_UNIQUE) {
288 stat->mnt_id = real_mount(path->mnt)->mnt_id_unique;
289 stat->result_mask |= STATX_MNT_ID_UNIQUE;
290 } else {
291 stat->mnt_id = real_mount(path->mnt)->mnt_id;
292 stat->result_mask |= STATX_MNT_ID;
293 }
294
295 if (path_mounted(path))
296 stat->attributes |= STATX_ATTR_MOUNT_ROOT;
297 stat->attributes_mask |= STATX_ATTR_MOUNT_ROOT;
298
299 /*
300 * If this is a block device inode, override the filesystem
301 * attributes with the block device specific parameters that need to be
302 * obtained from the bdev backing inode.
303 */
304 if (S_ISBLK(stat->mode))
305 bdev_statx(path, stat, request_mask);
306
307 return 0;
308 }
309
vfs_statx_fd(int fd,int flags,struct kstat * stat,u32 request_mask)310 static int vfs_statx_fd(int fd, int flags, struct kstat *stat,
311 u32 request_mask)
312 {
313 CLASS(fd_raw, f)(fd);
314 if (fd_empty(f))
315 return -EBADF;
316 return vfs_statx_path(&fd_file(f)->f_path, flags, stat, request_mask);
317 }
318
319 /**
320 * vfs_statx - Get basic and extra attributes by filename
321 * @dfd: A file descriptor representing the base dir for a relative filename
322 * @filename: The name of the file of interest
323 * @flags: Flags to control the query
324 * @stat: The result structure to fill in.
325 * @request_mask: STATX_xxx flags indicating what the caller wants
326 *
327 * This function is a wrapper around vfs_getattr(). The main difference is
328 * that it uses a filename and base directory to determine the file location.
329 * Additionally, the use of AT_SYMLINK_NOFOLLOW in flags will prevent a symlink
330 * at the given name from being referenced.
331 *
332 * 0 will be returned on success, and a -ve error code if unsuccessful.
333 */
vfs_statx(int dfd,struct filename * filename,int flags,struct kstat * stat,u32 request_mask)334 static int vfs_statx(int dfd, struct filename *filename, int flags,
335 struct kstat *stat, u32 request_mask)
336 {
337 struct path path;
338 unsigned int lookup_flags = statx_lookup_flags(flags);
339 int error;
340
341 if (flags & ~(AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT | AT_EMPTY_PATH |
342 AT_STATX_SYNC_TYPE))
343 return -EINVAL;
344
345 retry:
346 error = filename_lookup(dfd, filename, lookup_flags, &path, NULL);
347 if (error)
348 return error;
349 error = vfs_statx_path(&path, flags, stat, request_mask);
350 path_put(&path);
351 if (retry_estale(error, lookup_flags)) {
352 lookup_flags |= LOOKUP_REVAL;
353 goto retry;
354 }
355 return error;
356 }
357
vfs_fstatat(int dfd,const char __user * filename,struct kstat * stat,int flags)358 int vfs_fstatat(int dfd, const char __user *filename,
359 struct kstat *stat, int flags)
360 {
361 int ret;
362 int statx_flags = flags | AT_NO_AUTOMOUNT;
363 struct filename *name = getname_maybe_null(filename, flags);
364
365 if (!name && dfd >= 0)
366 return vfs_fstat(dfd, stat);
367
368 ret = vfs_statx(dfd, name, statx_flags, stat, STATX_BASIC_STATS);
369 putname(name);
370
371 return ret;
372 }
373
374 #ifdef __ARCH_WANT_OLD_STAT
375
376 /*
377 * For backward compatibility? Maybe this should be moved
378 * into arch/i386 instead?
379 */
cp_old_stat(struct kstat * stat,struct __old_kernel_stat __user * statbuf)380 static int cp_old_stat(struct kstat *stat, struct __old_kernel_stat __user * statbuf)
381 {
382 static int warncount = 5;
383 struct __old_kernel_stat tmp;
384
385 if (warncount > 0) {
386 warncount--;
387 printk(KERN_WARNING "VFS: Warning: %s using old stat() call. Recompile your binary.\n",
388 current->comm);
389 } else if (warncount < 0) {
390 /* it's laughable, but... */
391 warncount = 0;
392 }
393
394 memset(&tmp, 0, sizeof(struct __old_kernel_stat));
395 tmp.st_dev = old_encode_dev(stat->dev);
396 tmp.st_ino = stat->ino;
397 if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino)
398 return -EOVERFLOW;
399 tmp.st_mode = stat->mode;
400 tmp.st_nlink = stat->nlink;
401 if (tmp.st_nlink != stat->nlink)
402 return -EOVERFLOW;
403 SET_UID(tmp.st_uid, from_kuid_munged(current_user_ns(), stat->uid));
404 SET_GID(tmp.st_gid, from_kgid_munged(current_user_ns(), stat->gid));
405 tmp.st_rdev = old_encode_dev(stat->rdev);
406 #if BITS_PER_LONG == 32
407 if (stat->size > MAX_NON_LFS)
408 return -EOVERFLOW;
409 #endif
410 tmp.st_size = stat->size;
411 tmp.st_atime = stat->atime.tv_sec;
412 tmp.st_mtime = stat->mtime.tv_sec;
413 tmp.st_ctime = stat->ctime.tv_sec;
414 return copy_to_user(statbuf,&tmp,sizeof(tmp)) ? -EFAULT : 0;
415 }
416
SYSCALL_DEFINE2(stat,const char __user *,filename,struct __old_kernel_stat __user *,statbuf)417 SYSCALL_DEFINE2(stat, const char __user *, filename,
418 struct __old_kernel_stat __user *, statbuf)
419 {
420 struct kstat stat;
421 int error;
422
423 error = vfs_stat(filename, &stat);
424 if (error)
425 return error;
426
427 return cp_old_stat(&stat, statbuf);
428 }
429
SYSCALL_DEFINE2(lstat,const char __user *,filename,struct __old_kernel_stat __user *,statbuf)430 SYSCALL_DEFINE2(lstat, const char __user *, filename,
431 struct __old_kernel_stat __user *, statbuf)
432 {
433 struct kstat stat;
434 int error;
435
436 error = vfs_lstat(filename, &stat);
437 if (error)
438 return error;
439
440 return cp_old_stat(&stat, statbuf);
441 }
442
SYSCALL_DEFINE2(fstat,unsigned int,fd,struct __old_kernel_stat __user *,statbuf)443 SYSCALL_DEFINE2(fstat, unsigned int, fd, struct __old_kernel_stat __user *, statbuf)
444 {
445 struct kstat stat;
446 int error = vfs_fstat(fd, &stat);
447
448 if (!error)
449 error = cp_old_stat(&stat, statbuf);
450
451 return error;
452 }
453
454 #endif /* __ARCH_WANT_OLD_STAT */
455
456 #ifdef __ARCH_WANT_NEW_STAT
457
458 #ifndef INIT_STRUCT_STAT_PADDING
459 # define INIT_STRUCT_STAT_PADDING(st) memset(&st, 0, sizeof(st))
460 #endif
461
cp_new_stat(struct kstat * stat,struct stat __user * statbuf)462 static int cp_new_stat(struct kstat *stat, struct stat __user *statbuf)
463 {
464 struct stat tmp;
465
466 if (sizeof(tmp.st_dev) < 4 && !old_valid_dev(stat->dev))
467 return -EOVERFLOW;
468 if (sizeof(tmp.st_rdev) < 4 && !old_valid_dev(stat->rdev))
469 return -EOVERFLOW;
470 #if BITS_PER_LONG == 32
471 if (stat->size > MAX_NON_LFS)
472 return -EOVERFLOW;
473 #endif
474
475 INIT_STRUCT_STAT_PADDING(tmp);
476 tmp.st_dev = new_encode_dev(stat->dev);
477 tmp.st_ino = stat->ino;
478 if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino)
479 return -EOVERFLOW;
480 tmp.st_mode = stat->mode;
481 tmp.st_nlink = stat->nlink;
482 if (tmp.st_nlink != stat->nlink)
483 return -EOVERFLOW;
484 SET_UID(tmp.st_uid, from_kuid_munged(current_user_ns(), stat->uid));
485 SET_GID(tmp.st_gid, from_kgid_munged(current_user_ns(), stat->gid));
486 tmp.st_rdev = new_encode_dev(stat->rdev);
487 tmp.st_size = stat->size;
488 tmp.st_atime = stat->atime.tv_sec;
489 tmp.st_mtime = stat->mtime.tv_sec;
490 tmp.st_ctime = stat->ctime.tv_sec;
491 #ifdef STAT_HAVE_NSEC
492 tmp.st_atime_nsec = stat->atime.tv_nsec;
493 tmp.st_mtime_nsec = stat->mtime.tv_nsec;
494 tmp.st_ctime_nsec = stat->ctime.tv_nsec;
495 #endif
496 tmp.st_blocks = stat->blocks;
497 tmp.st_blksize = stat->blksize;
498 return copy_to_user(statbuf,&tmp,sizeof(tmp)) ? -EFAULT : 0;
499 }
500
SYSCALL_DEFINE2(newstat,const char __user *,filename,struct stat __user *,statbuf)501 SYSCALL_DEFINE2(newstat, const char __user *, filename,
502 struct stat __user *, statbuf)
503 {
504 struct kstat stat;
505 int error = vfs_stat(filename, &stat);
506
507 if (error)
508 return error;
509 return cp_new_stat(&stat, statbuf);
510 }
511
SYSCALL_DEFINE2(newlstat,const char __user *,filename,struct stat __user *,statbuf)512 SYSCALL_DEFINE2(newlstat, const char __user *, filename,
513 struct stat __user *, statbuf)
514 {
515 struct kstat stat;
516 int error;
517
518 error = vfs_lstat(filename, &stat);
519 if (error)
520 return error;
521
522 return cp_new_stat(&stat, statbuf);
523 }
524
525 #if !defined(__ARCH_WANT_STAT64) || defined(__ARCH_WANT_SYS_NEWFSTATAT)
SYSCALL_DEFINE4(newfstatat,int,dfd,const char __user *,filename,struct stat __user *,statbuf,int,flag)526 SYSCALL_DEFINE4(newfstatat, int, dfd, const char __user *, filename,
527 struct stat __user *, statbuf, int, flag)
528 {
529 struct kstat stat;
530 int error;
531
532 error = vfs_fstatat(dfd, filename, &stat, flag);
533 if (error)
534 return error;
535 return cp_new_stat(&stat, statbuf);
536 }
537 #endif
538
SYSCALL_DEFINE2(newfstat,unsigned int,fd,struct stat __user *,statbuf)539 SYSCALL_DEFINE2(newfstat, unsigned int, fd, struct stat __user *, statbuf)
540 {
541 struct kstat stat;
542 int error = vfs_fstat(fd, &stat);
543
544 if (!error)
545 error = cp_new_stat(&stat, statbuf);
546
547 return error;
548 }
549 #endif
550
do_readlinkat(int dfd,const char __user * pathname,char __user * buf,int bufsiz)551 static int do_readlinkat(int dfd, const char __user *pathname,
552 char __user *buf, int bufsiz)
553 {
554 struct path path;
555 struct filename *name;
556 int error;
557 unsigned int lookup_flags = LOOKUP_EMPTY;
558
559 if (bufsiz <= 0)
560 return -EINVAL;
561
562 retry:
563 name = getname_flags(pathname, lookup_flags);
564 error = filename_lookup(dfd, name, lookup_flags, &path, NULL);
565 if (unlikely(error)) {
566 putname(name);
567 return error;
568 }
569
570 /*
571 * AFS mountpoints allow readlink(2) but are not symlinks
572 */
573 if (d_is_symlink(path.dentry) ||
574 d_backing_inode(path.dentry)->i_op->readlink) {
575 error = security_inode_readlink(path.dentry);
576 if (!error) {
577 touch_atime(&path);
578 error = vfs_readlink(path.dentry, buf, bufsiz);
579 }
580 } else {
581 error = (name->name[0] == '\0') ? -ENOENT : -EINVAL;
582 }
583 path_put(&path);
584 putname(name);
585 if (retry_estale(error, lookup_flags)) {
586 lookup_flags |= LOOKUP_REVAL;
587 goto retry;
588 }
589 return error;
590 }
591
SYSCALL_DEFINE4(readlinkat,int,dfd,const char __user *,pathname,char __user *,buf,int,bufsiz)592 SYSCALL_DEFINE4(readlinkat, int, dfd, const char __user *, pathname,
593 char __user *, buf, int, bufsiz)
594 {
595 return do_readlinkat(dfd, pathname, buf, bufsiz);
596 }
597
SYSCALL_DEFINE3(readlink,const char __user *,path,char __user *,buf,int,bufsiz)598 SYSCALL_DEFINE3(readlink, const char __user *, path, char __user *, buf,
599 int, bufsiz)
600 {
601 return do_readlinkat(AT_FDCWD, path, buf, bufsiz);
602 }
603
604
605 /* ---------- LFS-64 ----------- */
606 #if defined(__ARCH_WANT_STAT64) || defined(__ARCH_WANT_COMPAT_STAT64)
607
608 #ifndef INIT_STRUCT_STAT64_PADDING
609 # define INIT_STRUCT_STAT64_PADDING(st) memset(&st, 0, sizeof(st))
610 #endif
611
cp_new_stat64(struct kstat * stat,struct stat64 __user * statbuf)612 static long cp_new_stat64(struct kstat *stat, struct stat64 __user *statbuf)
613 {
614 struct stat64 tmp;
615
616 INIT_STRUCT_STAT64_PADDING(tmp);
617 #ifdef CONFIG_MIPS
618 /* mips has weird padding, so we don't get 64 bits there */
619 tmp.st_dev = new_encode_dev(stat->dev);
620 tmp.st_rdev = new_encode_dev(stat->rdev);
621 #else
622 tmp.st_dev = huge_encode_dev(stat->dev);
623 tmp.st_rdev = huge_encode_dev(stat->rdev);
624 #endif
625 tmp.st_ino = stat->ino;
626 if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino)
627 return -EOVERFLOW;
628 #ifdef STAT64_HAS_BROKEN_ST_INO
629 tmp.__st_ino = stat->ino;
630 #endif
631 tmp.st_mode = stat->mode;
632 tmp.st_nlink = stat->nlink;
633 tmp.st_uid = from_kuid_munged(current_user_ns(), stat->uid);
634 tmp.st_gid = from_kgid_munged(current_user_ns(), stat->gid);
635 tmp.st_atime = stat->atime.tv_sec;
636 tmp.st_atime_nsec = stat->atime.tv_nsec;
637 tmp.st_mtime = stat->mtime.tv_sec;
638 tmp.st_mtime_nsec = stat->mtime.tv_nsec;
639 tmp.st_ctime = stat->ctime.tv_sec;
640 tmp.st_ctime_nsec = stat->ctime.tv_nsec;
641 tmp.st_size = stat->size;
642 tmp.st_blocks = stat->blocks;
643 tmp.st_blksize = stat->blksize;
644 return copy_to_user(statbuf,&tmp,sizeof(tmp)) ? -EFAULT : 0;
645 }
646
SYSCALL_DEFINE2(stat64,const char __user *,filename,struct stat64 __user *,statbuf)647 SYSCALL_DEFINE2(stat64, const char __user *, filename,
648 struct stat64 __user *, statbuf)
649 {
650 struct kstat stat;
651 int error = vfs_stat(filename, &stat);
652
653 if (!error)
654 error = cp_new_stat64(&stat, statbuf);
655
656 return error;
657 }
658
SYSCALL_DEFINE2(lstat64,const char __user *,filename,struct stat64 __user *,statbuf)659 SYSCALL_DEFINE2(lstat64, const char __user *, filename,
660 struct stat64 __user *, statbuf)
661 {
662 struct kstat stat;
663 int error = vfs_lstat(filename, &stat);
664
665 if (!error)
666 error = cp_new_stat64(&stat, statbuf);
667
668 return error;
669 }
670
SYSCALL_DEFINE2(fstat64,unsigned long,fd,struct stat64 __user *,statbuf)671 SYSCALL_DEFINE2(fstat64, unsigned long, fd, struct stat64 __user *, statbuf)
672 {
673 struct kstat stat;
674 int error = vfs_fstat(fd, &stat);
675
676 if (!error)
677 error = cp_new_stat64(&stat, statbuf);
678
679 return error;
680 }
681
SYSCALL_DEFINE4(fstatat64,int,dfd,const char __user *,filename,struct stat64 __user *,statbuf,int,flag)682 SYSCALL_DEFINE4(fstatat64, int, dfd, const char __user *, filename,
683 struct stat64 __user *, statbuf, int, flag)
684 {
685 struct kstat stat;
686 int error;
687
688 error = vfs_fstatat(dfd, filename, &stat, flag);
689 if (error)
690 return error;
691 return cp_new_stat64(&stat, statbuf);
692 }
693 #endif /* __ARCH_WANT_STAT64 || __ARCH_WANT_COMPAT_STAT64 */
694
695 static noinline_for_stack int
cp_statx(const struct kstat * stat,struct statx __user * buffer)696 cp_statx(const struct kstat *stat, struct statx __user *buffer)
697 {
698 struct statx tmp;
699
700 memset(&tmp, 0, sizeof(tmp));
701
702 /* STATX_CHANGE_COOKIE is kernel-only for now */
703 tmp.stx_mask = stat->result_mask & ~STATX_CHANGE_COOKIE;
704 tmp.stx_blksize = stat->blksize;
705 /* STATX_ATTR_CHANGE_MONOTONIC is kernel-only for now */
706 tmp.stx_attributes = stat->attributes & ~STATX_ATTR_CHANGE_MONOTONIC;
707 tmp.stx_nlink = stat->nlink;
708 tmp.stx_uid = from_kuid_munged(current_user_ns(), stat->uid);
709 tmp.stx_gid = from_kgid_munged(current_user_ns(), stat->gid);
710 tmp.stx_mode = stat->mode;
711 tmp.stx_ino = stat->ino;
712 tmp.stx_size = stat->size;
713 tmp.stx_blocks = stat->blocks;
714 tmp.stx_attributes_mask = stat->attributes_mask;
715 tmp.stx_atime.tv_sec = stat->atime.tv_sec;
716 tmp.stx_atime.tv_nsec = stat->atime.tv_nsec;
717 tmp.stx_btime.tv_sec = stat->btime.tv_sec;
718 tmp.stx_btime.tv_nsec = stat->btime.tv_nsec;
719 tmp.stx_ctime.tv_sec = stat->ctime.tv_sec;
720 tmp.stx_ctime.tv_nsec = stat->ctime.tv_nsec;
721 tmp.stx_mtime.tv_sec = stat->mtime.tv_sec;
722 tmp.stx_mtime.tv_nsec = stat->mtime.tv_nsec;
723 tmp.stx_rdev_major = MAJOR(stat->rdev);
724 tmp.stx_rdev_minor = MINOR(stat->rdev);
725 tmp.stx_dev_major = MAJOR(stat->dev);
726 tmp.stx_dev_minor = MINOR(stat->dev);
727 tmp.stx_mnt_id = stat->mnt_id;
728 tmp.stx_dio_mem_align = stat->dio_mem_align;
729 tmp.stx_dio_offset_align = stat->dio_offset_align;
730 tmp.stx_dio_read_offset_align = stat->dio_read_offset_align;
731 tmp.stx_subvol = stat->subvol;
732 tmp.stx_atomic_write_unit_min = stat->atomic_write_unit_min;
733 tmp.stx_atomic_write_unit_max = stat->atomic_write_unit_max;
734 tmp.stx_atomic_write_segments_max = stat->atomic_write_segments_max;
735
736 return copy_to_user(buffer, &tmp, sizeof(tmp)) ? -EFAULT : 0;
737 }
738
do_statx(int dfd,struct filename * filename,unsigned int flags,unsigned int mask,struct statx __user * buffer)739 int do_statx(int dfd, struct filename *filename, unsigned int flags,
740 unsigned int mask, struct statx __user *buffer)
741 {
742 struct kstat stat;
743 int error;
744
745 if (mask & STATX__RESERVED)
746 return -EINVAL;
747 if ((flags & AT_STATX_SYNC_TYPE) == AT_STATX_SYNC_TYPE)
748 return -EINVAL;
749
750 /*
751 * STATX_CHANGE_COOKIE is kernel-only for now. Ignore requests
752 * from userland.
753 */
754 mask &= ~STATX_CHANGE_COOKIE;
755
756 error = vfs_statx(dfd, filename, flags, &stat, mask);
757 if (error)
758 return error;
759
760 return cp_statx(&stat, buffer);
761 }
762
do_statx_fd(int fd,unsigned int flags,unsigned int mask,struct statx __user * buffer)763 int do_statx_fd(int fd, unsigned int flags, unsigned int mask,
764 struct statx __user *buffer)
765 {
766 struct kstat stat;
767 int error;
768
769 if (mask & STATX__RESERVED)
770 return -EINVAL;
771 if ((flags & AT_STATX_SYNC_TYPE) == AT_STATX_SYNC_TYPE)
772 return -EINVAL;
773
774 /*
775 * STATX_CHANGE_COOKIE is kernel-only for now. Ignore requests
776 * from userland.
777 */
778 mask &= ~STATX_CHANGE_COOKIE;
779
780 error = vfs_statx_fd(fd, flags, &stat, mask);
781 if (error)
782 return error;
783
784 return cp_statx(&stat, buffer);
785 }
786
787 /**
788 * sys_statx - System call to get enhanced stats
789 * @dfd: Base directory to pathwalk from *or* fd to stat.
790 * @filename: File to stat or either NULL or "" with AT_EMPTY_PATH
791 * @flags: AT_* flags to control pathwalk.
792 * @mask: Parts of statx struct actually required.
793 * @buffer: Result buffer.
794 *
795 * Note that fstat() can be emulated by setting dfd to the fd of interest,
796 * supplying "" (or preferably NULL) as the filename and setting AT_EMPTY_PATH
797 * in the flags.
798 */
SYSCALL_DEFINE5(statx,int,dfd,const char __user *,filename,unsigned,flags,unsigned int,mask,struct statx __user *,buffer)799 SYSCALL_DEFINE5(statx,
800 int, dfd, const char __user *, filename, unsigned, flags,
801 unsigned int, mask,
802 struct statx __user *, buffer)
803 {
804 int ret;
805 struct filename *name = getname_maybe_null(filename, flags);
806
807 if (!name && dfd >= 0)
808 return do_statx_fd(dfd, flags & ~AT_NO_AUTOMOUNT, mask, buffer);
809
810 ret = do_statx(dfd, name, flags, mask, buffer);
811 putname(name);
812
813 return ret;
814 }
815
816 #if defined(CONFIG_COMPAT) && defined(__ARCH_WANT_COMPAT_STAT)
cp_compat_stat(struct kstat * stat,struct compat_stat __user * ubuf)817 static int cp_compat_stat(struct kstat *stat, struct compat_stat __user *ubuf)
818 {
819 struct compat_stat tmp;
820
821 if (sizeof(tmp.st_dev) < 4 && !old_valid_dev(stat->dev))
822 return -EOVERFLOW;
823 if (sizeof(tmp.st_rdev) < 4 && !old_valid_dev(stat->rdev))
824 return -EOVERFLOW;
825
826 memset(&tmp, 0, sizeof(tmp));
827 tmp.st_dev = new_encode_dev(stat->dev);
828 tmp.st_ino = stat->ino;
829 if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino)
830 return -EOVERFLOW;
831 tmp.st_mode = stat->mode;
832 tmp.st_nlink = stat->nlink;
833 if (tmp.st_nlink != stat->nlink)
834 return -EOVERFLOW;
835 SET_UID(tmp.st_uid, from_kuid_munged(current_user_ns(), stat->uid));
836 SET_GID(tmp.st_gid, from_kgid_munged(current_user_ns(), stat->gid));
837 tmp.st_rdev = new_encode_dev(stat->rdev);
838 if ((u64) stat->size > MAX_NON_LFS)
839 return -EOVERFLOW;
840 tmp.st_size = stat->size;
841 tmp.st_atime = stat->atime.tv_sec;
842 tmp.st_atime_nsec = stat->atime.tv_nsec;
843 tmp.st_mtime = stat->mtime.tv_sec;
844 tmp.st_mtime_nsec = stat->mtime.tv_nsec;
845 tmp.st_ctime = stat->ctime.tv_sec;
846 tmp.st_ctime_nsec = stat->ctime.tv_nsec;
847 tmp.st_blocks = stat->blocks;
848 tmp.st_blksize = stat->blksize;
849 return copy_to_user(ubuf, &tmp, sizeof(tmp)) ? -EFAULT : 0;
850 }
851
COMPAT_SYSCALL_DEFINE2(newstat,const char __user *,filename,struct compat_stat __user *,statbuf)852 COMPAT_SYSCALL_DEFINE2(newstat, const char __user *, filename,
853 struct compat_stat __user *, statbuf)
854 {
855 struct kstat stat;
856 int error;
857
858 error = vfs_stat(filename, &stat);
859 if (error)
860 return error;
861 return cp_compat_stat(&stat, statbuf);
862 }
863
COMPAT_SYSCALL_DEFINE2(newlstat,const char __user *,filename,struct compat_stat __user *,statbuf)864 COMPAT_SYSCALL_DEFINE2(newlstat, const char __user *, filename,
865 struct compat_stat __user *, statbuf)
866 {
867 struct kstat stat;
868 int error;
869
870 error = vfs_lstat(filename, &stat);
871 if (error)
872 return error;
873 return cp_compat_stat(&stat, statbuf);
874 }
875
876 #ifndef __ARCH_WANT_STAT64
COMPAT_SYSCALL_DEFINE4(newfstatat,unsigned int,dfd,const char __user *,filename,struct compat_stat __user *,statbuf,int,flag)877 COMPAT_SYSCALL_DEFINE4(newfstatat, unsigned int, dfd,
878 const char __user *, filename,
879 struct compat_stat __user *, statbuf, int, flag)
880 {
881 struct kstat stat;
882 int error;
883
884 error = vfs_fstatat(dfd, filename, &stat, flag);
885 if (error)
886 return error;
887 return cp_compat_stat(&stat, statbuf);
888 }
889 #endif
890
COMPAT_SYSCALL_DEFINE2(newfstat,unsigned int,fd,struct compat_stat __user *,statbuf)891 COMPAT_SYSCALL_DEFINE2(newfstat, unsigned int, fd,
892 struct compat_stat __user *, statbuf)
893 {
894 struct kstat stat;
895 int error = vfs_fstat(fd, &stat);
896
897 if (!error)
898 error = cp_compat_stat(&stat, statbuf);
899 return error;
900 }
901 #endif
902
903 /* Caller is here responsible for sufficient locking (ie. inode->i_lock) */
__inode_add_bytes(struct inode * inode,loff_t bytes)904 void __inode_add_bytes(struct inode *inode, loff_t bytes)
905 {
906 inode->i_blocks += bytes >> 9;
907 bytes &= 511;
908 inode->i_bytes += bytes;
909 if (inode->i_bytes >= 512) {
910 inode->i_blocks++;
911 inode->i_bytes -= 512;
912 }
913 }
914 EXPORT_SYMBOL(__inode_add_bytes);
915
inode_add_bytes(struct inode * inode,loff_t bytes)916 void inode_add_bytes(struct inode *inode, loff_t bytes)
917 {
918 spin_lock(&inode->i_lock);
919 __inode_add_bytes(inode, bytes);
920 spin_unlock(&inode->i_lock);
921 }
922
923 EXPORT_SYMBOL(inode_add_bytes);
924
__inode_sub_bytes(struct inode * inode,loff_t bytes)925 void __inode_sub_bytes(struct inode *inode, loff_t bytes)
926 {
927 inode->i_blocks -= bytes >> 9;
928 bytes &= 511;
929 if (inode->i_bytes < bytes) {
930 inode->i_blocks--;
931 inode->i_bytes += 512;
932 }
933 inode->i_bytes -= bytes;
934 }
935
936 EXPORT_SYMBOL(__inode_sub_bytes);
937
inode_sub_bytes(struct inode * inode,loff_t bytes)938 void inode_sub_bytes(struct inode *inode, loff_t bytes)
939 {
940 spin_lock(&inode->i_lock);
941 __inode_sub_bytes(inode, bytes);
942 spin_unlock(&inode->i_lock);
943 }
944
945 EXPORT_SYMBOL(inode_sub_bytes);
946
inode_get_bytes(struct inode * inode)947 loff_t inode_get_bytes(struct inode *inode)
948 {
949 loff_t ret;
950
951 spin_lock(&inode->i_lock);
952 ret = __inode_get_bytes(inode);
953 spin_unlock(&inode->i_lock);
954 return ret;
955 }
956
957 EXPORT_SYMBOL(inode_get_bytes);
958
inode_set_bytes(struct inode * inode,loff_t bytes)959 void inode_set_bytes(struct inode *inode, loff_t bytes)
960 {
961 /* Caller is here responsible for sufficient locking
962 * (ie. inode->i_lock) */
963 inode->i_blocks = bytes >> 9;
964 inode->i_bytes = bytes & 511;
965 }
966
967 EXPORT_SYMBOL(inode_set_bytes);
968