xref: /linux/include/uapi/linux/fs.h (revision b7ce6fa90fd9554482847b19756a06232c1dc78c)
1 /* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
2 #ifndef _UAPI_LINUX_FS_H
3 #define _UAPI_LINUX_FS_H
4 
5 /*
6  * This file has definitions for some important file table structures
7  * and constants and structures used by various generic file system
8  * ioctl's.  Please do not make any changes in this file before
9  * sending patches for review to linux-fsdevel@vger.kernel.org and
10  * linux-api@vger.kernel.org.
11  */
12 
13 #include <linux/limits.h>
14 #include <linux/ioctl.h>
15 #include <linux/types.h>
16 #ifndef __KERNEL__
17 #include <linux/fscrypt.h>
18 #endif
19 
20 /* Use of MS_* flags within the kernel is restricted to core mount(2) code. */
21 #if !defined(__KERNEL__)
22 #include <linux/mount.h>
23 #endif
24 
25 /*
26  * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
27  * the file limit at runtime and only root can increase the per-process
28  * nr_file rlimit, so it's safe to set up a ridiculously high absolute
29  * upper limit on files-per-process.
30  *
31  * Some programs (notably those using select()) may have to be
32  * recompiled to take full advantage of the new limits..
33  */
34 
35 /* Fixed constants first: */
36 #undef NR_OPEN
37 #define INR_OPEN_CUR 1024	/* Initial setting for nfile rlimits */
38 #define INR_OPEN_MAX 4096	/* Hard limit for nfile rlimits */
39 
40 #define BLOCK_SIZE_BITS 10
41 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
42 
43 /* flags for integrity meta */
44 #define IO_INTEGRITY_CHK_GUARD		(1U << 0) /* enforce guard check */
45 #define IO_INTEGRITY_CHK_REFTAG		(1U << 1) /* enforce ref check */
46 #define IO_INTEGRITY_CHK_APPTAG		(1U << 2) /* enforce app check */
47 
48 #define IO_INTEGRITY_VALID_FLAGS (IO_INTEGRITY_CHK_GUARD | \
49 				  IO_INTEGRITY_CHK_REFTAG | \
50 				  IO_INTEGRITY_CHK_APPTAG)
51 
52 #define SEEK_SET	0	/* seek relative to beginning of file */
53 #define SEEK_CUR	1	/* seek relative to current file position */
54 #define SEEK_END	2	/* seek relative to end of file */
55 #define SEEK_DATA	3	/* seek to the next data */
56 #define SEEK_HOLE	4	/* seek to the next hole */
57 #define SEEK_MAX	SEEK_HOLE
58 
59 #define RENAME_NOREPLACE	(1 << 0)	/* Don't overwrite target */
60 #define RENAME_EXCHANGE		(1 << 1)	/* Exchange source and dest */
61 #define RENAME_WHITEOUT		(1 << 2)	/* Whiteout source */
62 
63 /*
64  * The root inode of procfs is guaranteed to always have the same inode number.
65  * For programs that make heavy use of procfs, verifying that the root is a
66  * real procfs root and using openat2(RESOLVE_{NO_{XDEV,MAGICLINKS},BENEATH})
67  * will allow you to make sure you are never tricked into operating on the
68  * wrong procfs file.
69  */
70 enum procfs_ino {
71 	PROCFS_ROOT_INO = 1,
72 };
73 
74 struct file_clone_range {
75 	__s64 src_fd;
76 	__u64 src_offset;
77 	__u64 src_length;
78 	__u64 dest_offset;
79 };
80 
81 struct fstrim_range {
82 	__u64 start;
83 	__u64 len;
84 	__u64 minlen;
85 };
86 
87 /*
88  * We include a length field because some filesystems (vfat) have an identifier
89  * that we do want to expose as a UUID, but doesn't have the standard length.
90  *
91  * We use a fixed size buffer beacuse this interface will, by fiat, never
92  * support "UUIDs" longer than 16 bytes; we don't want to force all downstream
93  * users to have to deal with that.
94  */
95 struct fsuuid2 {
96 	__u8	len;
97 	__u8	uuid[16];
98 };
99 
100 struct fs_sysfs_path {
101 	__u8			len;
102 	__u8			name[128];
103 };
104 
105 /* Protection info capability flags */
106 #define	LBMD_PI_CAP_INTEGRITY		(1 << 0)
107 #define	LBMD_PI_CAP_REFTAG		(1 << 1)
108 
109 /* Checksum types for Protection Information */
110 #define LBMD_PI_CSUM_NONE		0
111 #define LBMD_PI_CSUM_IP			1
112 #define LBMD_PI_CSUM_CRC16_T10DIF	2
113 #define LBMD_PI_CSUM_CRC64_NVME		4
114 
115 /* sizeof first published struct */
116 #define LBMD_SIZE_VER0			16
117 
118 /*
119  * Logical block metadata capability descriptor
120  * If the device does not support metadata, all the fields will be zero.
121  * Applications must check lbmd_flags to determine whether metadata is
122  * supported or not.
123  */
124 struct logical_block_metadata_cap {
125 	/* Bitmask of logical block metadata capability flags */
126 	__u32	lbmd_flags;
127 	/*
128 	 * The amount of data described by each unit of logical block
129 	 * metadata
130 	 */
131 	__u16	lbmd_interval;
132 	/*
133 	 * Size in bytes of the logical block metadata associated with each
134 	 * interval
135 	 */
136 	__u8	lbmd_size;
137 	/*
138 	 * Size in bytes of the opaque block tag associated with each
139 	 * interval
140 	 */
141 	__u8	lbmd_opaque_size;
142 	/*
143 	 * Offset in bytes of the opaque block tag within the logical block
144 	 * metadata
145 	 */
146 	__u8	lbmd_opaque_offset;
147 	/* Size in bytes of the T10 PI tuple associated with each interval */
148 	__u8	lbmd_pi_size;
149 	/* Offset in bytes of T10 PI tuple within the logical block metadata */
150 	__u8	lbmd_pi_offset;
151 	/* T10 PI guard tag type */
152 	__u8	lbmd_guard_tag_type;
153 	/* Size in bytes of the T10 PI application tag */
154 	__u8	lbmd_app_tag_size;
155 	/* Size in bytes of the T10 PI reference tag */
156 	__u8	lbmd_ref_tag_size;
157 	/* Size in bytes of the T10 PI storage tag */
158 	__u8	lbmd_storage_tag_size;
159 	__u8	pad;
160 };
161 
162 /* extent-same (dedupe) ioctls; these MUST match the btrfs ioctl definitions */
163 #define FILE_DEDUPE_RANGE_SAME		0
164 #define FILE_DEDUPE_RANGE_DIFFERS	1
165 
166 /* from struct btrfs_ioctl_file_extent_same_info */
167 struct file_dedupe_range_info {
168 	__s64 dest_fd;		/* in - destination file */
169 	__u64 dest_offset;	/* in - start of extent in destination */
170 	__u64 bytes_deduped;	/* out - total # of bytes we were able
171 				 * to dedupe from this file. */
172 	/* status of this dedupe operation:
173 	 * < 0 for error
174 	 * == FILE_DEDUPE_RANGE_SAME if dedupe succeeds
175 	 * == FILE_DEDUPE_RANGE_DIFFERS if data differs
176 	 */
177 	__s32 status;		/* out - see above description */
178 	__u32 reserved;		/* must be zero */
179 };
180 
181 /* from struct btrfs_ioctl_file_extent_same_args */
182 struct file_dedupe_range {
183 	__u64 src_offset;	/* in - start of extent in source */
184 	__u64 src_length;	/* in - length of extent */
185 	__u16 dest_count;	/* in - total elements in info array */
186 	__u16 reserved1;	/* must be zero */
187 	__u32 reserved2;	/* must be zero */
188 	struct file_dedupe_range_info info[];
189 };
190 
191 /* And dynamically-tunable limits and defaults: */
192 struct files_stat_struct {
193 	unsigned long nr_files;		/* read only */
194 	unsigned long nr_free_files;	/* read only */
195 	unsigned long max_files;		/* tunable */
196 };
197 
198 struct inodes_stat_t {
199 	long nr_inodes;
200 	long nr_unused;
201 	long dummy[5];		/* padding for sysctl ABI compatibility */
202 };
203 
204 
205 #define NR_FILE  8192	/* this can well be larger on a larger system */
206 
207 /*
208  * Structure for FS_IOC_FSGETXATTR[A] and FS_IOC_FSSETXATTR.
209  */
210 struct fsxattr {
211 	__u32		fsx_xflags;	/* xflags field value (get/set) */
212 	__u32		fsx_extsize;	/* extsize field value (get/set)*/
213 	__u32		fsx_nextents;	/* nextents field value (get)	*/
214 	__u32		fsx_projid;	/* project identifier (get/set) */
215 	__u32		fsx_cowextsize;	/* CoW extsize field value (get/set)*/
216 	unsigned char	fsx_pad[8];
217 };
218 
219 /*
220  * Variable size structure for file_[sg]et_attr().
221  *
222  * Note. This is alternative to the structure 'struct file_kattr'/'struct fsxattr'.
223  * As this structure is passed to/from userspace with its size, this can
224  * be versioned based on the size.
225  */
226 struct file_attr {
227 	__u64 fa_xflags;	/* xflags field value (get/set) */
228 	__u32 fa_extsize;	/* extsize field value (get/set)*/
229 	__u32 fa_nextents;	/* nextents field value (get)   */
230 	__u32 fa_projid;	/* project identifier (get/set) */
231 	__u32 fa_cowextsize;	/* CoW extsize field value (get/set) */
232 };
233 
234 #define FILE_ATTR_SIZE_VER0 24
235 #define FILE_ATTR_SIZE_LATEST FILE_ATTR_SIZE_VER0
236 
237 /*
238  * Flags for the fsx_xflags field
239  */
240 #define FS_XFLAG_REALTIME	0x00000001	/* data in realtime volume */
241 #define FS_XFLAG_PREALLOC	0x00000002	/* preallocated file extents */
242 #define FS_XFLAG_IMMUTABLE	0x00000008	/* file cannot be modified */
243 #define FS_XFLAG_APPEND		0x00000010	/* all writes append */
244 #define FS_XFLAG_SYNC		0x00000020	/* all writes synchronous */
245 #define FS_XFLAG_NOATIME	0x00000040	/* do not update access time */
246 #define FS_XFLAG_NODUMP		0x00000080	/* do not include in backups */
247 #define FS_XFLAG_RTINHERIT	0x00000100	/* create with rt bit set */
248 #define FS_XFLAG_PROJINHERIT	0x00000200	/* create with parents projid */
249 #define FS_XFLAG_NOSYMLINKS	0x00000400	/* disallow symlink creation */
250 #define FS_XFLAG_EXTSIZE	0x00000800	/* extent size allocator hint */
251 #define FS_XFLAG_EXTSZINHERIT	0x00001000	/* inherit inode extent size */
252 #define FS_XFLAG_NODEFRAG	0x00002000	/* do not defragment */
253 #define FS_XFLAG_FILESTREAM	0x00004000	/* use filestream allocator */
254 #define FS_XFLAG_DAX		0x00008000	/* use DAX for IO */
255 #define FS_XFLAG_COWEXTSIZE	0x00010000	/* CoW extent size allocator hint */
256 #define FS_XFLAG_HASATTR	0x80000000	/* no DIFLAG for this	*/
257 
258 /* the read-only stuff doesn't really belong here, but any other place is
259    probably as bad and I don't want to create yet another include file. */
260 
261 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
262 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
263 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
264 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
265 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
266 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
267 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
268 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
269 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
270 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
271 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
272 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
273 #if 0
274 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
275 
276 /* Some people are morons.  Do not use sizeof! */
277 
278 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
279 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
280 /* This was here just to show that the number is taken -
281    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
282 #endif
283 /* A jump here: 108-111 have been used for various private purposes. */
284 #define BLKBSZGET  _IOR(0x12,112,size_t)
285 #define BLKBSZSET  _IOW(0x12,113,size_t)
286 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
287 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
288 #define BLKTRACESTART _IO(0x12,116)
289 #define BLKTRACESTOP _IO(0x12,117)
290 #define BLKTRACETEARDOWN _IO(0x12,118)
291 #define BLKDISCARD _IO(0x12,119)
292 #define BLKIOMIN _IO(0x12,120)
293 #define BLKIOOPT _IO(0x12,121)
294 #define BLKALIGNOFF _IO(0x12,122)
295 #define BLKPBSZGET _IO(0x12,123)
296 #define BLKDISCARDZEROES _IO(0x12,124)
297 #define BLKSECDISCARD _IO(0x12,125)
298 #define BLKROTATIONAL _IO(0x12,126)
299 #define BLKZEROOUT _IO(0x12,127)
300 #define BLKGETDISKSEQ _IOR(0x12,128,__u64)
301 /* 130-136 are used by zoned block device ioctls (uapi/linux/blkzoned.h) */
302 /* 137-141 are used by blk-crypto ioctls (uapi/linux/blk-crypto.h) */
303 
304 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
305 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
306 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
307 #define FIFREEZE	_IOWR('X', 119, int)	/* Freeze */
308 #define FITHAW		_IOWR('X', 120, int)	/* Thaw */
309 #define FITRIM		_IOWR('X', 121, struct fstrim_range)	/* Trim */
310 #define FICLONE		_IOW(0x94, 9, int)
311 #define FICLONERANGE	_IOW(0x94, 13, struct file_clone_range)
312 #define FIDEDUPERANGE	_IOWR(0x94, 54, struct file_dedupe_range)
313 
314 #define FSLABEL_MAX 256	/* Max chars for the interface; each fs may differ */
315 
316 #define	FS_IOC_GETFLAGS			_IOR('f', 1, long)
317 #define	FS_IOC_SETFLAGS			_IOW('f', 2, long)
318 #define	FS_IOC_GETVERSION		_IOR('v', 1, long)
319 #define	FS_IOC_SETVERSION		_IOW('v', 2, long)
320 #define FS_IOC_FIEMAP			_IOWR('f', 11, struct fiemap)
321 #define FS_IOC32_GETFLAGS		_IOR('f', 1, int)
322 #define FS_IOC32_SETFLAGS		_IOW('f', 2, int)
323 #define FS_IOC32_GETVERSION		_IOR('v', 1, int)
324 #define FS_IOC32_SETVERSION		_IOW('v', 2, int)
325 #define FS_IOC_FSGETXATTR		_IOR('X', 31, struct fsxattr)
326 #define FS_IOC_FSSETXATTR		_IOW('X', 32, struct fsxattr)
327 #define FS_IOC_GETFSLABEL		_IOR(0x94, 49, char[FSLABEL_MAX])
328 #define FS_IOC_SETFSLABEL		_IOW(0x94, 50, char[FSLABEL_MAX])
329 /* Returns the external filesystem UUID, the same one blkid returns */
330 #define FS_IOC_GETFSUUID		_IOR(0x15, 0, struct fsuuid2)
331 /*
332  * Returns the path component under /sys/fs/ that refers to this filesystem;
333  * also /sys/kernel/debug/ for filesystems with debugfs exports
334  */
335 #define FS_IOC_GETFSSYSFSPATH		_IOR(0x15, 1, struct fs_sysfs_path)
336 /* Get logical block metadata capability details */
337 #define FS_IOC_GETLBMD_CAP		_IOWR(0x15, 2, struct logical_block_metadata_cap)
338 
339 /*
340  * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
341  *
342  * Note: for historical reasons, these flags were originally used and
343  * defined for use by ext2/ext3, and then other file systems started
344  * using these flags so they wouldn't need to write their own version
345  * of chattr/lsattr (which was shipped as part of e2fsprogs).  You
346  * should think twice before trying to use these flags in new
347  * contexts, or trying to assign these flags, since they are used both
348  * as the UAPI and the on-disk encoding for ext2/3/4.  Also, we are
349  * almost out of 32-bit flags.  :-)
350  *
351  * We have recently hoisted FS_IOC_FSGETXATTR / FS_IOC_FSSETXATTR from
352  * XFS to the generic FS level interface.  This uses a structure that
353  * has padding and hence has more room to grow, so it may be more
354  * appropriate for many new use cases.
355  *
356  * Please do not change these flags or interfaces before checking with
357  * linux-fsdevel@vger.kernel.org and linux-api@vger.kernel.org.
358  */
359 #define	FS_SECRM_FL			0x00000001 /* Secure deletion */
360 #define	FS_UNRM_FL			0x00000002 /* Undelete */
361 #define	FS_COMPR_FL			0x00000004 /* Compress file */
362 #define FS_SYNC_FL			0x00000008 /* Synchronous updates */
363 #define FS_IMMUTABLE_FL			0x00000010 /* Immutable file */
364 #define FS_APPEND_FL			0x00000020 /* writes to file may only append */
365 #define FS_NODUMP_FL			0x00000040 /* do not dump file */
366 #define FS_NOATIME_FL			0x00000080 /* do not update atime */
367 /* Reserved for compression usage... */
368 #define FS_DIRTY_FL			0x00000100
369 #define FS_COMPRBLK_FL			0x00000200 /* One or more compressed clusters */
370 #define FS_NOCOMP_FL			0x00000400 /* Don't compress */
371 /* End compression flags --- maybe not all used */
372 #define FS_ENCRYPT_FL			0x00000800 /* Encrypted file */
373 #define FS_BTREE_FL			0x00001000 /* btree format dir */
374 #define FS_INDEX_FL			0x00001000 /* hash-indexed directory */
375 #define FS_IMAGIC_FL			0x00002000 /* AFS directory */
376 #define FS_JOURNAL_DATA_FL		0x00004000 /* Reserved for ext3 */
377 #define FS_NOTAIL_FL			0x00008000 /* file tail should not be merged */
378 #define FS_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
379 #define FS_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
380 #define FS_HUGE_FILE_FL			0x00040000 /* Reserved for ext4 */
381 #define FS_EXTENT_FL			0x00080000 /* Extents */
382 #define FS_VERITY_FL			0x00100000 /* Verity protected inode */
383 #define FS_EA_INODE_FL			0x00200000 /* Inode used for large EA */
384 #define FS_EOFBLOCKS_FL			0x00400000 /* Reserved for ext4 */
385 #define FS_NOCOW_FL			0x00800000 /* Do not cow file */
386 #define FS_DAX_FL			0x02000000 /* Inode is DAX */
387 #define FS_INLINE_DATA_FL		0x10000000 /* Reserved for ext4 */
388 #define FS_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
389 #define FS_CASEFOLD_FL			0x40000000 /* Folder is case insensitive */
390 #define FS_RESERVED_FL			0x80000000 /* reserved for ext2 lib */
391 
392 #define FS_FL_USER_VISIBLE		0x0003DFFF /* User visible flags */
393 #define FS_FL_USER_MODIFIABLE		0x000380FF /* User modifiable flags */
394 
395 
396 #define SYNC_FILE_RANGE_WAIT_BEFORE	1
397 #define SYNC_FILE_RANGE_WRITE		2
398 #define SYNC_FILE_RANGE_WAIT_AFTER	4
399 #define SYNC_FILE_RANGE_WRITE_AND_WAIT	(SYNC_FILE_RANGE_WRITE | \
400 					 SYNC_FILE_RANGE_WAIT_BEFORE | \
401 					 SYNC_FILE_RANGE_WAIT_AFTER)
402 
403 /*
404  * Flags for preadv2/pwritev2:
405  */
406 
407 typedef int __bitwise __kernel_rwf_t;
408 
409 /* high priority request, poll if possible */
410 #define RWF_HIPRI	((__force __kernel_rwf_t)0x00000001)
411 
412 /* per-IO O_DSYNC */
413 #define RWF_DSYNC	((__force __kernel_rwf_t)0x00000002)
414 
415 /* per-IO O_SYNC */
416 #define RWF_SYNC	((__force __kernel_rwf_t)0x00000004)
417 
418 /* per-IO, return -EAGAIN if operation would block */
419 #define RWF_NOWAIT	((__force __kernel_rwf_t)0x00000008)
420 
421 /* per-IO O_APPEND */
422 #define RWF_APPEND	((__force __kernel_rwf_t)0x00000010)
423 
424 /* per-IO negation of O_APPEND */
425 #define RWF_NOAPPEND	((__force __kernel_rwf_t)0x00000020)
426 
427 /* Atomic Write */
428 #define RWF_ATOMIC	((__force __kernel_rwf_t)0x00000040)
429 
430 /* buffered IO that drops the cache after reading or writing data */
431 #define RWF_DONTCACHE	((__force __kernel_rwf_t)0x00000080)
432 
433 /* prevent pipe and socket writes from raising SIGPIPE */
434 #define RWF_NOSIGNAL	((__force __kernel_rwf_t)0x00000100)
435 
436 /* mask of flags supported by the kernel */
437 #define RWF_SUPPORTED	(RWF_HIPRI | RWF_DSYNC | RWF_SYNC | RWF_NOWAIT |\
438 			 RWF_APPEND | RWF_NOAPPEND | RWF_ATOMIC |\
439 			 RWF_DONTCACHE | RWF_NOSIGNAL)
440 
441 #define PROCFS_IOCTL_MAGIC 'f'
442 
443 /* Pagemap ioctl */
444 #define PAGEMAP_SCAN	_IOWR(PROCFS_IOCTL_MAGIC, 16, struct pm_scan_arg)
445 
446 /* Bitmasks provided in pm_scan_args masks and reported in page_region.categories. */
447 #define PAGE_IS_WPALLOWED	(1 << 0)
448 #define PAGE_IS_WRITTEN		(1 << 1)
449 #define PAGE_IS_FILE		(1 << 2)
450 #define PAGE_IS_PRESENT		(1 << 3)
451 #define PAGE_IS_SWAPPED		(1 << 4)
452 #define PAGE_IS_PFNZERO		(1 << 5)
453 #define PAGE_IS_HUGE		(1 << 6)
454 #define PAGE_IS_SOFT_DIRTY	(1 << 7)
455 #define PAGE_IS_GUARD		(1 << 8)
456 
457 /*
458  * struct page_region - Page region with flags
459  * @start:	Start of the region
460  * @end:	End of the region (exclusive)
461  * @categories:	PAGE_IS_* category bitmask for the region
462  */
463 struct page_region {
464 	__u64 start;
465 	__u64 end;
466 	__u64 categories;
467 };
468 
469 /* Flags for PAGEMAP_SCAN ioctl */
470 #define PM_SCAN_WP_MATCHING	(1 << 0)	/* Write protect the pages matched. */
471 #define PM_SCAN_CHECK_WPASYNC	(1 << 1)	/* Abort the scan when a non-WP-enabled page is found. */
472 
473 /*
474  * struct pm_scan_arg - Pagemap ioctl argument
475  * @size:		Size of the structure
476  * @flags:		Flags for the IOCTL
477  * @start:		Starting address of the region
478  * @end:		Ending address of the region
479  * @walk_end		Address where the scan stopped (written by kernel).
480  *			walk_end == end (address tags cleared) informs that the scan completed on entire range.
481  * @vec:		Address of page_region struct array for output
482  * @vec_len:		Length of the page_region struct array
483  * @max_pages:		Optional limit for number of returned pages (0 = disabled)
484  * @category_inverted:	PAGE_IS_* categories which values match if 0 instead of 1
485  * @category_mask:	Skip pages for which any category doesn't match
486  * @category_anyof_mask: Skip pages for which no category matches
487  * @return_mask:	PAGE_IS_* categories that are to be reported in `page_region`s returned
488  */
489 struct pm_scan_arg {
490 	__u64 size;
491 	__u64 flags;
492 	__u64 start;
493 	__u64 end;
494 	__u64 walk_end;
495 	__u64 vec;
496 	__u64 vec_len;
497 	__u64 max_pages;
498 	__u64 category_inverted;
499 	__u64 category_mask;
500 	__u64 category_anyof_mask;
501 	__u64 return_mask;
502 };
503 
504 /* /proc/<pid>/maps ioctl */
505 #define PROCMAP_QUERY	_IOWR(PROCFS_IOCTL_MAGIC, 17, struct procmap_query)
506 
507 enum procmap_query_flags {
508 	/*
509 	 * VMA permission flags.
510 	 *
511 	 * Can be used as part of procmap_query.query_flags field to look up
512 	 * only VMAs satisfying specified subset of permissions. E.g., specifying
513 	 * PROCMAP_QUERY_VMA_READABLE only will return both readable and read/write VMAs,
514 	 * while having PROCMAP_QUERY_VMA_READABLE | PROCMAP_QUERY_VMA_WRITABLE will only
515 	 * return read/write VMAs, though both executable/non-executable and
516 	 * private/shared will be ignored.
517 	 *
518 	 * PROCMAP_QUERY_VMA_* flags are also returned in procmap_query.vma_flags
519 	 * field to specify actual VMA permissions.
520 	 */
521 	PROCMAP_QUERY_VMA_READABLE		= 0x01,
522 	PROCMAP_QUERY_VMA_WRITABLE		= 0x02,
523 	PROCMAP_QUERY_VMA_EXECUTABLE		= 0x04,
524 	PROCMAP_QUERY_VMA_SHARED		= 0x08,
525 	/*
526 	 * Query modifier flags.
527 	 *
528 	 * By default VMA that covers provided address is returned, or -ENOENT
529 	 * is returned. With PROCMAP_QUERY_COVERING_OR_NEXT_VMA flag set, closest
530 	 * VMA with vma_start > addr will be returned if no covering VMA is
531 	 * found.
532 	 *
533 	 * PROCMAP_QUERY_FILE_BACKED_VMA instructs query to consider only VMAs that
534 	 * have file backing. Can be combined with PROCMAP_QUERY_COVERING_OR_NEXT_VMA
535 	 * to iterate all VMAs with file backing.
536 	 */
537 	PROCMAP_QUERY_COVERING_OR_NEXT_VMA	= 0x10,
538 	PROCMAP_QUERY_FILE_BACKED_VMA		= 0x20,
539 };
540 
541 /*
542  * Input/output argument structured passed into ioctl() call. It can be used
543  * to query a set of VMAs (Virtual Memory Areas) of a process.
544  *
545  * Each field can be one of three kinds, marked in a short comment to the
546  * right of the field:
547  *   - "in", input argument, user has to provide this value, kernel doesn't modify it;
548  *   - "out", output argument, kernel sets this field with VMA data;
549  *   - "in/out", input and output argument; user provides initial value (used
550  *     to specify maximum allowable buffer size), and kernel sets it to actual
551  *     amount of data written (or zero, if there is no data).
552  *
553  * If matching VMA is found (according to criterias specified by
554  * query_addr/query_flags, all the out fields are filled out, and ioctl()
555  * returns 0. If there is no matching VMA, -ENOENT will be returned.
556  * In case of any other error, negative error code other than -ENOENT is
557  * returned.
558  *
559  * Most of the data is similar to the one returned as text in /proc/<pid>/maps
560  * file, but procmap_query provides more querying flexibility. There are no
561  * consistency guarantees between subsequent ioctl() calls, but data returned
562  * for matched VMA is self-consistent.
563  */
564 struct procmap_query {
565 	/* Query struct size, for backwards/forward compatibility */
566 	__u64 size;
567 	/*
568 	 * Query flags, a combination of enum procmap_query_flags values.
569 	 * Defines query filtering and behavior, see enum procmap_query_flags.
570 	 *
571 	 * Input argument, provided by user. Kernel doesn't modify it.
572 	 */
573 	__u64 query_flags;		/* in */
574 	/*
575 	 * Query address. By default, VMA that covers this address will
576 	 * be looked up. PROCMAP_QUERY_* flags above modify this default
577 	 * behavior further.
578 	 *
579 	 * Input argument, provided by user. Kernel doesn't modify it.
580 	 */
581 	__u64 query_addr;		/* in */
582 	/* VMA starting (inclusive) and ending (exclusive) address, if VMA is found. */
583 	__u64 vma_start;		/* out */
584 	__u64 vma_end;			/* out */
585 	/* VMA permissions flags. A combination of PROCMAP_QUERY_VMA_* flags. */
586 	__u64 vma_flags;		/* out */
587 	/* VMA backing page size granularity. */
588 	__u64 vma_page_size;		/* out */
589 	/*
590 	 * VMA file offset. If VMA has file backing, this specifies offset
591 	 * within the file that VMA's start address corresponds to.
592 	 * Is set to zero if VMA has no backing file.
593 	 */
594 	__u64 vma_offset;		/* out */
595 	/* Backing file's inode number, or zero, if VMA has no backing file. */
596 	__u64 inode;			/* out */
597 	/* Backing file's device major/minor number, or zero, if VMA has no backing file. */
598 	__u32 dev_major;		/* out */
599 	__u32 dev_minor;		/* out */
600 	/*
601 	 * If set to non-zero value, signals the request to return VMA name
602 	 * (i.e., VMA's backing file's absolute path, with " (deleted)" suffix
603 	 * appended, if file was unlinked from FS) for matched VMA. VMA name
604 	 * can also be some special name (e.g., "[heap]", "[stack]") or could
605 	 * be even user-supplied with prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME).
606 	 *
607 	 * Kernel will set this field to zero, if VMA has no associated name.
608 	 * Otherwise kernel will return actual amount of bytes filled in
609 	 * user-supplied buffer (see vma_name_addr field below), including the
610 	 * terminating zero.
611 	 *
612 	 * If VMA name is longer that user-supplied maximum buffer size,
613 	 * -E2BIG error is returned.
614 	 *
615 	 * If this field is set to non-zero value, vma_name_addr should point
616 	 * to valid user space memory buffer of at least vma_name_size bytes.
617 	 * If set to zero, vma_name_addr should be set to zero as well
618 	 */
619 	__u32 vma_name_size;		/* in/out */
620 	/*
621 	 * If set to non-zero value, signals the request to extract and return
622 	 * VMA's backing file's build ID, if the backing file is an ELF file
623 	 * and it contains embedded build ID.
624 	 *
625 	 * Kernel will set this field to zero, if VMA has no backing file,
626 	 * backing file is not an ELF file, or ELF file has no build ID
627 	 * embedded.
628 	 *
629 	 * Build ID is a binary value (not a string). Kernel will set
630 	 * build_id_size field to exact number of bytes used for build ID.
631 	 * If build ID is requested and present, but needs more bytes than
632 	 * user-supplied maximum buffer size (see build_id_addr field below),
633 	 * -E2BIG error will be returned.
634 	 *
635 	 * If this field is set to non-zero value, build_id_addr should point
636 	 * to valid user space memory buffer of at least build_id_size bytes.
637 	 * If set to zero, build_id_addr should be set to zero as well
638 	 */
639 	__u32 build_id_size;		/* in/out */
640 	/*
641 	 * User-supplied address of a buffer of at least vma_name_size bytes
642 	 * for kernel to fill with matched VMA's name (see vma_name_size field
643 	 * description above for details).
644 	 *
645 	 * Should be set to zero if VMA name should not be returned.
646 	 */
647 	__u64 vma_name_addr;		/* in */
648 	/*
649 	 * User-supplied address of a buffer of at least build_id_size bytes
650 	 * for kernel to fill with matched VMA's ELF build ID, if available
651 	 * (see build_id_size field description above for details).
652 	 *
653 	 * Should be set to zero if build ID should not be returned.
654 	 */
655 	__u64 build_id_addr;		/* in */
656 };
657 
658 #endif /* _UAPI_LINUX_FS_H */
659