xref: /kernel/linux/linux-5.10/fs/erofs/internal.h (revision 8c2ecf20)
1/* SPDX-License-Identifier: GPL-2.0-only */
2/*
3 * Copyright (C) 2017-2018 HUAWEI, Inc.
4 *             https://www.huawei.com/
5 * Created by Gao Xiang <gaoxiang25@huawei.com>
6 */
7#ifndef __EROFS_INTERNAL_H
8#define __EROFS_INTERNAL_H
9
10#include <linux/fs.h>
11#include <linux/dcache.h>
12#include <linux/mm.h>
13#include <linux/pagemap.h>
14#include <linux/bio.h>
15#include <linux/buffer_head.h>
16#include <linux/magic.h>
17#include <linux/slab.h>
18#include <linux/vmalloc.h>
19#include "erofs_fs.h"
20
21/* redefine pr_fmt "erofs: " */
22#undef pr_fmt
23#define pr_fmt(fmt) "erofs: " fmt
24
25__printf(3, 4) void _erofs_err(struct super_block *sb,
26			       const char *function, const char *fmt, ...);
27#define erofs_err(sb, fmt, ...)	\
28	_erofs_err(sb, __func__, fmt "\n", ##__VA_ARGS__)
29__printf(3, 4) void _erofs_info(struct super_block *sb,
30			       const char *function, const char *fmt, ...);
31#define erofs_info(sb, fmt, ...) \
32	_erofs_info(sb, __func__, fmt "\n", ##__VA_ARGS__)
33#ifdef CONFIG_EROFS_FS_DEBUG
34#define erofs_dbg(x, ...)       pr_debug(x "\n", ##__VA_ARGS__)
35#define DBG_BUGON               BUG_ON
36#else
37#define erofs_dbg(x, ...)       ((void)0)
38#define DBG_BUGON(x)            ((void)(x))
39#endif	/* !CONFIG_EROFS_FS_DEBUG */
40
41/* EROFS_SUPER_MAGIC_V1 to represent the whole file system */
42#define EROFS_SUPER_MAGIC   EROFS_SUPER_MAGIC_V1
43
44typedef u64 erofs_nid_t;
45typedef u64 erofs_off_t;
46/* data type for filesystem-wide blocks number */
47typedef u32 erofs_blk_t;
48
49struct erofs_fs_context {
50#ifdef CONFIG_EROFS_FS_ZIP
51	/* current strategy of how to use managed cache */
52	unsigned char cache_strategy;
53
54	/* threshold for decompression synchronously */
55	unsigned int max_sync_decompress_pages;
56#endif
57	unsigned int mount_opt;
58};
59
60struct erofs_sb_info {
61#ifdef CONFIG_EROFS_FS_ZIP
62	/* list for all registered superblocks, mainly for shrinker */
63	struct list_head list;
64	struct mutex umount_mutex;
65
66	/* managed XArray arranged in physical block number */
67	struct xarray managed_pslots;
68
69	unsigned int shrinker_run_no;
70
71	/* pseudo inode to manage cached pages */
72	struct inode *managed_cache;
73#endif	/* CONFIG_EROFS_FS_ZIP */
74	u32 blocks;
75	u32 meta_blkaddr;
76#ifdef CONFIG_EROFS_FS_XATTR
77	u32 xattr_blkaddr;
78#endif
79
80	/* inode slot unit size in bit shift */
81	unsigned char islotbits;
82
83	u32 build_time_nsec;
84	u64 build_time;
85
86	/* what we really care is nid, rather than ino.. */
87	erofs_nid_t root_nid;
88	/* used for statfs, f_files - f_favail */
89	u64 inos;
90
91	u8 uuid[16];                    /* 128-bit uuid for volume */
92	u8 volume_name[16];             /* volume name */
93	u32 feature_compat;
94	u32 feature_incompat;
95
96	struct erofs_fs_context ctx;	/* options */
97};
98
99#define EROFS_SB(sb) ((struct erofs_sb_info *)(sb)->s_fs_info)
100#define EROFS_I_SB(inode) ((struct erofs_sb_info *)(inode)->i_sb->s_fs_info)
101
102/* Mount flags set via mount options or defaults */
103#define EROFS_MOUNT_XATTR_USER		0x00000010
104#define EROFS_MOUNT_POSIX_ACL		0x00000020
105
106#define clear_opt(ctx, option)	((ctx)->mount_opt &= ~EROFS_MOUNT_##option)
107#define set_opt(ctx, option)	((ctx)->mount_opt |= EROFS_MOUNT_##option)
108#define test_opt(ctx, option)	((ctx)->mount_opt & EROFS_MOUNT_##option)
109
110enum {
111	EROFS_ZIP_CACHE_DISABLED,
112	EROFS_ZIP_CACHE_READAHEAD,
113	EROFS_ZIP_CACHE_READAROUND
114};
115
116#ifdef CONFIG_EROFS_FS_ZIP
117#define EROFS_LOCKED_MAGIC     (INT_MIN | 0xE0F510CCL)
118
119/* basic unit of the workstation of a super_block */
120struct erofs_workgroup {
121	/* the workgroup index in the workstation */
122	pgoff_t index;
123
124	/* overall workgroup reference count */
125	atomic_t refcount;
126};
127
128#if defined(CONFIG_SMP)
129static inline bool erofs_workgroup_try_to_freeze(struct erofs_workgroup *grp,
130						 int val)
131{
132	preempt_disable();
133	if (val != atomic_cmpxchg(&grp->refcount, val, EROFS_LOCKED_MAGIC)) {
134		preempt_enable();
135		return false;
136	}
137	return true;
138}
139
140static inline void erofs_workgroup_unfreeze(struct erofs_workgroup *grp,
141					    int orig_val)
142{
143	/*
144	 * other observers should notice all modifications
145	 * in the freezing period.
146	 */
147	smp_mb();
148	atomic_set(&grp->refcount, orig_val);
149	preempt_enable();
150}
151
152static inline int erofs_wait_on_workgroup_freezed(struct erofs_workgroup *grp)
153{
154	return atomic_cond_read_relaxed(&grp->refcount,
155					VAL != EROFS_LOCKED_MAGIC);
156}
157#else
158static inline bool erofs_workgroup_try_to_freeze(struct erofs_workgroup *grp,
159						 int val)
160{
161	preempt_disable();
162	/* no need to spin on UP platforms, let's just disable preemption. */
163	if (val != atomic_read(&grp->refcount)) {
164		preempt_enable();
165		return false;
166	}
167	return true;
168}
169
170static inline void erofs_workgroup_unfreeze(struct erofs_workgroup *grp,
171					    int orig_val)
172{
173	preempt_enable();
174}
175
176static inline int erofs_wait_on_workgroup_freezed(struct erofs_workgroup *grp)
177{
178	int v = atomic_read(&grp->refcount);
179
180	/* workgroup is never freezed on uniprocessor systems */
181	DBG_BUGON(v == EROFS_LOCKED_MAGIC);
182	return v;
183}
184#endif	/* !CONFIG_SMP */
185
186/* hard limit of pages per compressed cluster */
187#define Z_EROFS_CLUSTER_MAX_PAGES       (CONFIG_EROFS_FS_CLUSTER_PAGE_LIMIT)
188#define EROFS_PCPUBUF_NR_PAGES          Z_EROFS_CLUSTER_MAX_PAGES
189#else
190#define EROFS_PCPUBUF_NR_PAGES          0
191#endif	/* !CONFIG_EROFS_FS_ZIP */
192
193/* we strictly follow PAGE_SIZE and no buffer head yet */
194#define LOG_BLOCK_SIZE		PAGE_SHIFT
195
196#undef LOG_SECTORS_PER_BLOCK
197#define LOG_SECTORS_PER_BLOCK	(PAGE_SHIFT - 9)
198
199#undef SECTORS_PER_BLOCK
200#define SECTORS_PER_BLOCK	(1 << SECTORS_PER_BLOCK)
201
202#define EROFS_BLKSIZ		(1 << LOG_BLOCK_SIZE)
203
204#if (EROFS_BLKSIZ % 4096 || !EROFS_BLKSIZ)
205#error erofs cannot be used in this platform
206#endif
207
208#define ROOT_NID(sb)		((sb)->root_nid)
209
210#define erofs_blknr(addr)       ((addr) / EROFS_BLKSIZ)
211#define erofs_blkoff(addr)      ((addr) % EROFS_BLKSIZ)
212#define blknr_to_addr(nr)       ((erofs_off_t)(nr) * EROFS_BLKSIZ)
213
214static inline erofs_off_t iloc(struct erofs_sb_info *sbi, erofs_nid_t nid)
215{
216	return blknr_to_addr(sbi->meta_blkaddr) + (nid << sbi->islotbits);
217}
218
219/* atomic flag definitions */
220#define EROFS_I_EA_INITED_BIT	0
221#define EROFS_I_Z_INITED_BIT	1
222
223/* bitlock definitions (arranged in reverse order) */
224#define EROFS_I_BL_XATTR_BIT	(BITS_PER_LONG - 1)
225#define EROFS_I_BL_Z_BIT	(BITS_PER_LONG - 2)
226
227struct erofs_inode {
228	erofs_nid_t nid;
229
230	/* atomic flags (including bitlocks) */
231	unsigned long flags;
232
233	unsigned char datalayout;
234	unsigned char inode_isize;
235	unsigned int xattr_isize;
236
237	unsigned int xattr_shared_count;
238	unsigned int *xattr_shared_xattrs;
239
240	union {
241		erofs_blk_t raw_blkaddr;
242#ifdef CONFIG_EROFS_FS_ZIP
243		struct {
244			unsigned short z_advise;
245			unsigned char  z_algorithmtype[2];
246			unsigned char  z_logical_clusterbits;
247			unsigned char  z_physical_clusterbits[2];
248		};
249#endif	/* CONFIG_EROFS_FS_ZIP */
250	};
251	/* the corresponding vfs inode */
252	struct inode vfs_inode;
253};
254
255#define EROFS_I(ptr)	\
256	container_of(ptr, struct erofs_inode, vfs_inode)
257
258static inline unsigned long erofs_inode_datablocks(struct inode *inode)
259{
260	/* since i_size cannot be changed */
261	return DIV_ROUND_UP(inode->i_size, EROFS_BLKSIZ);
262}
263
264static inline unsigned int erofs_bitrange(unsigned int value, unsigned int bit,
265					  unsigned int bits)
266{
267
268	return (value >> bit) & ((1 << bits) - 1);
269}
270
271
272static inline unsigned int erofs_inode_version(unsigned int value)
273{
274	return erofs_bitrange(value, EROFS_I_VERSION_BIT,
275			      EROFS_I_VERSION_BITS);
276}
277
278static inline unsigned int erofs_inode_datalayout(unsigned int value)
279{
280	return erofs_bitrange(value, EROFS_I_DATALAYOUT_BIT,
281			      EROFS_I_DATALAYOUT_BITS);
282}
283
284extern const struct super_operations erofs_sops;
285
286extern const struct address_space_operations erofs_raw_access_aops;
287extern const struct address_space_operations z_erofs_aops;
288
289/*
290 * Logical to physical block mapping, used by erofs_map_blocks()
291 *
292 * Different with other file systems, it is used for 2 access modes:
293 *
294 * 1) RAW access mode:
295 *
296 * Users pass a valid (m_lblk, m_lofs -- usually 0) pair,
297 * and get the valid m_pblk, m_pofs and the longest m_len(in bytes).
298 *
299 * Note that m_lblk in the RAW access mode refers to the number of
300 * the compressed ondisk block rather than the uncompressed
301 * in-memory block for the compressed file.
302 *
303 * m_pofs equals to m_lofs except for the inline data page.
304 *
305 * 2) Normal access mode:
306 *
307 * If the inode is not compressed, it has no difference with
308 * the RAW access mode. However, if the inode is compressed,
309 * users should pass a valid (m_lblk, m_lofs) pair, and get
310 * the needed m_pblk, m_pofs, m_len to get the compressed data
311 * and the updated m_lblk, m_lofs which indicates the start
312 * of the corresponding uncompressed data in the file.
313 */
314enum {
315	BH_Zipped = BH_PrivateStart,
316	BH_FullMapped,
317};
318
319/* Has a disk mapping */
320#define EROFS_MAP_MAPPED	(1 << BH_Mapped)
321/* Located in metadata (could be copied from bd_inode) */
322#define EROFS_MAP_META		(1 << BH_Meta)
323/* The extent has been compressed */
324#define EROFS_MAP_ZIPPED	(1 << BH_Zipped)
325/* The length of extent is full */
326#define EROFS_MAP_FULL_MAPPED	(1 << BH_FullMapped)
327
328struct erofs_map_blocks {
329	erofs_off_t m_pa, m_la;
330	u64 m_plen, m_llen;
331
332	unsigned int m_flags;
333
334	struct page *mpage;
335};
336
337/* Flags used by erofs_map_blocks() */
338#define EROFS_GET_BLOCKS_RAW    0x0001
339
340/* zmap.c */
341#ifdef CONFIG_EROFS_FS_ZIP
342int z_erofs_fill_inode(struct inode *inode);
343int z_erofs_map_blocks_iter(struct inode *inode,
344			    struct erofs_map_blocks *map,
345			    int flags);
346#else
347static inline int z_erofs_fill_inode(struct inode *inode) { return -EOPNOTSUPP; }
348static inline int z_erofs_map_blocks_iter(struct inode *inode,
349					  struct erofs_map_blocks *map,
350					  int flags)
351{
352	return -EOPNOTSUPP;
353}
354#endif	/* !CONFIG_EROFS_FS_ZIP */
355
356/* data.c */
357struct page *erofs_get_meta_page(struct super_block *sb, erofs_blk_t blkaddr);
358
359int erofs_map_blocks(struct inode *, struct erofs_map_blocks *, int);
360
361/* inode.c */
362static inline unsigned long erofs_inode_hash(erofs_nid_t nid)
363{
364#if BITS_PER_LONG == 32
365	return (nid >> 32) ^ (nid & 0xffffffff);
366#else
367	return nid;
368#endif
369}
370
371extern const struct inode_operations erofs_generic_iops;
372extern const struct inode_operations erofs_symlink_iops;
373extern const struct inode_operations erofs_fast_symlink_iops;
374
375struct inode *erofs_iget(struct super_block *sb, erofs_nid_t nid, bool dir);
376int erofs_getattr(const struct path *path, struct kstat *stat,
377		  u32 request_mask, unsigned int query_flags);
378
379/* namei.c */
380extern const struct inode_operations erofs_dir_iops;
381
382int erofs_namei(struct inode *dir, struct qstr *name,
383		erofs_nid_t *nid, unsigned int *d_type);
384
385/* dir.c */
386extern const struct file_operations erofs_dir_fops;
387
388/* utils.c / zdata.c */
389struct page *erofs_allocpage(struct list_head *pool, gfp_t gfp);
390
391#if (EROFS_PCPUBUF_NR_PAGES > 0)
392void *erofs_get_pcpubuf(unsigned int pagenr);
393#define erofs_put_pcpubuf(buf) do { \
394	(void)&(buf);	\
395	preempt_enable();	\
396} while (0)
397#else
398static inline void *erofs_get_pcpubuf(unsigned int pagenr)
399{
400	return ERR_PTR(-EOPNOTSUPP);
401}
402
403#define erofs_put_pcpubuf(buf) do {} while (0)
404#endif
405
406#ifdef CONFIG_EROFS_FS_ZIP
407int erofs_workgroup_put(struct erofs_workgroup *grp);
408struct erofs_workgroup *erofs_find_workgroup(struct super_block *sb,
409					     pgoff_t index);
410struct erofs_workgroup *erofs_insert_workgroup(struct super_block *sb,
411					       struct erofs_workgroup *grp);
412void erofs_workgroup_free_rcu(struct erofs_workgroup *grp);
413void erofs_shrinker_register(struct super_block *sb);
414void erofs_shrinker_unregister(struct super_block *sb);
415int __init erofs_init_shrinker(void);
416void erofs_exit_shrinker(void);
417int __init z_erofs_init_zip_subsystem(void);
418void z_erofs_exit_zip_subsystem(void);
419int erofs_try_to_free_all_cached_pages(struct erofs_sb_info *sbi,
420				       struct erofs_workgroup *egrp);
421int erofs_try_to_free_cached_page(struct address_space *mapping,
422				  struct page *page);
423#else
424static inline void erofs_shrinker_register(struct super_block *sb) {}
425static inline void erofs_shrinker_unregister(struct super_block *sb) {}
426static inline int erofs_init_shrinker(void) { return 0; }
427static inline void erofs_exit_shrinker(void) {}
428static inline int z_erofs_init_zip_subsystem(void) { return 0; }
429static inline void z_erofs_exit_zip_subsystem(void) {}
430#endif	/* !CONFIG_EROFS_FS_ZIP */
431
432#define EFSCORRUPTED    EUCLEAN         /* Filesystem is corrupted */
433
434#endif	/* __EROFS_INTERNAL_H */
435
436