2 * Copyright (c) International Business Machines Corp., 2006
3 * Copyright (c) Nokia Corporation, 2006, 2007
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
13 * the GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19 * Author: Artem Bityutskiy (Битюцкий Артём)
25 #include <linux/init.h>
26 #include <linux/types.h>
27 #include <linux/list.h>
28 #include <linux/rbtree.h>
29 #include <linux/sched.h>
30 #include <linux/wait.h>
31 #include <linux/mutex.h>
32 #include <linux/rwsem.h>
33 #include <linux/spinlock.h>
35 #include <linux/cdev.h>
36 #include <linux/device.h>
37 #include <linux/string.h>
38 #include <linux/vmalloc.h>
39 #include <linux/mtd/mtd.h>
41 #include <mtd/ubi-header.h>
42 #include <linux/mtd/ubi.h>
47 /* Maximum number of supported UBI devices */
48 #define UBI_MAX_DEVICES 32
50 /* UBI name used for character devices, sysfs, etc */
51 #define UBI_NAME_STR "ubi"
53 /* Normal UBI messages */
54 #define ubi_msg(fmt, ...) printk(KERN_NOTICE "UBI: " fmt "\n", ##__VA_ARGS__)
55 /* UBI warning messages */
56 #define ubi_warn(fmt, ...) printk(KERN_WARNING "UBI warning: %s: " fmt "\n", \
57 __FUNCTION__, ##__VA_ARGS__)
58 /* UBI error messages */
59 #define ubi_err(fmt, ...) printk(KERN_ERR "UBI error: %s: " fmt "\n", \
60 __FUNCTION__, ##__VA_ARGS__)
62 /* Lowest number PEBs reserved for bad PEB handling */
63 #define MIN_RESEVED_PEBS 2
65 /* Background thread name pattern */
66 #define UBI_BGT_NAME_PATTERN "ubi_bgt%dd"
68 /* This marker in the EBA table means that the LEB is um-mapped */
69 #define UBI_LEB_UNMAPPED -1
72 * In case of errors, UBI tries to repeat the operation several times before
73 * returning error. The below constant defines how many times UBI re-tries.
75 #define UBI_IO_RETRIES 3
78 * Error codes returned by the I/O unit.
80 * UBI_IO_PEB_EMPTY: the physical eraseblock is empty, i.e. it contains only
82 * UBI_IO_PEB_FREE: the physical eraseblock is free, i.e. it contains only a
83 * valid erase counter header, and the rest are %0xFF bytes
84 * UBI_IO_BAD_EC_HDR: the erase counter header is corrupted (bad magic or CRC)
85 * UBI_IO_BAD_VID_HDR: the volume identifier header is corrupted (bad magic or
87 * UBI_IO_BITFLIPS: bit-flips were detected and corrected
98 * struct ubi_wl_entry - wear-leveling entry.
99 * @rb: link in the corresponding RB-tree
101 * @pnum: physical eraseblock number
103 * This data structure is used in the WL unit. Each physical eraseblock has a
104 * corresponding &struct wl_entry object which may be kept in different
105 * RB-trees. See WL unit for details.
107 struct ubi_wl_entry {
114 * struct ubi_ltree_entry - an entry in the lock tree.
115 * @rb: links RB-tree nodes
116 * @vol_id: volume ID of the locked logical eraseblock
117 * @lnum: locked logical eraseblock number
118 * @users: how many tasks are using this logical eraseblock or wait for it
119 * @mutex: read/write mutex to implement read/write access serialization to
120 * the (@vol_id, @lnum) logical eraseblock
122 * This data structure is used in the EBA unit to implement per-LEB locking.
123 * When a logical eraseblock is being locked - corresponding
124 * &struct ubi_ltree_entry object is inserted to the lock tree (@ubi->ltree).
125 * See EBA unit for details.
127 struct ubi_ltree_entry {
132 struct rw_semaphore mutex;
135 struct ubi_volume_desc;
138 * struct ubi_volume - UBI volume description data structure.
139 * @dev: device object to make use of the the Linux device model
140 * @cdev: character device object to create character device
141 * @ubi: reference to the UBI device description object
143 * @ref_count: volume reference count
144 * @readers: number of users holding this volume in read-only mode
145 * @writers: number of users holding this volume in read-write mode
146 * @exclusive: whether somebody holds this volume in exclusive mode
148 * @reserved_pebs: how many physical eraseblocks are reserved for this volume
149 * @vol_type: volume type (%UBI_DYNAMIC_VOLUME or %UBI_STATIC_VOLUME)
150 * @usable_leb_size: logical eraseblock size without padding
151 * @used_ebs: how many logical eraseblocks in this volume contain data
152 * @last_eb_bytes: how many bytes are stored in the last logical eraseblock
153 * @used_bytes: how many bytes of data this volume contains
154 * @alignment: volume alignment
155 * @data_pad: how many bytes are not used at the end of physical eraseblocks to
156 * satisfy the requested alignment
157 * @name_len: volume name length
160 * @upd_ebs: how many eraseblocks are expected to be updated
161 * @ch_lnum: LEB number which is being changing by the atomic LEB change
163 * @ch_dtype: data persistency type which is being changing by the atomic LEB
165 * @upd_bytes: how many bytes are expected to be received for volume update or
167 * @upd_received: how many bytes were already received for volume update or
169 * @upd_buf: update buffer which is used to collect update data or data for
172 * @eba_tbl: EBA table of this volume (LEB->PEB mapping)
173 * @checked: %1 if this static volume was checked
174 * @corrupted: %1 if the volume is corrupted (static volumes only)
175 * @upd_marker: %1 if the update marker is set for this volume
176 * @updating: %1 if the volume is being updated
177 * @changing_leb: %1 if the atomic LEB change ioctl command is in progress
179 * @gluebi_desc: gluebi UBI volume descriptor
180 * @gluebi_refcount: reference count of the gluebi MTD device
181 * @gluebi_mtd: MTD device description object of the gluebi MTD device
183 * The @corrupted field indicates that the volume's contents is corrupted.
184 * Since UBI protects only static volumes, this field is not relevant to
185 * dynamic volumes - it is user's responsibility to assure their data
188 * The @upd_marker flag indicates that this volume is either being updated at
189 * the moment or is damaged because of an unclean reboot.
194 struct ubi_device *ubi;
206 long long used_bytes;
210 char name[UBI_VOL_NAME_MAX+1];
216 long long upd_received;
220 unsigned int checked:1;
221 unsigned int corrupted:1;
222 unsigned int upd_marker:1;
223 unsigned int updating:1;
224 unsigned int changing_leb:1;
226 #ifdef CONFIG_MTD_UBI_GLUEBI
228 * Gluebi-related stuff may be compiled out.
229 * TODO: this should not be built into UBI but should be a separate
230 * ubimtd driver which works on top of UBI and emulates MTD devices.
232 struct ubi_volume_desc *gluebi_desc;
234 struct mtd_info gluebi_mtd;
239 * struct ubi_volume_desc - descriptor of the UBI volume returned when it is
241 * @vol: reference to the corresponding volume description object
242 * @mode: open mode (%UBI_READONLY, %UBI_READWRITE, or %UBI_EXCLUSIVE)
244 struct ubi_volume_desc {
245 struct ubi_volume *vol;
252 * struct ubi_device - UBI device description structure
253 * @dev: UBI device object to use the the Linux device model
254 * @cdev: character device object to create character device
255 * @ubi_num: UBI device number
256 * @ubi_name: UBI device name
257 * @vol_count: number of volumes in this UBI device
258 * @volumes: volumes of this UBI device
259 * @volumes_lock: protects @volumes, @rsvd_pebs, @avail_pebs, beb_rsvd_pebs,
260 * @beb_rsvd_level, @bad_peb_count, @good_peb_count, @vol_count,
261 * @vol->readers, @vol->writers, @vol->exclusive,
262 * @vol->ref_count, @vol->mapping and @vol->eba_tbl.
263 * @ref_count: count of references on the UBI device
265 * @rsvd_pebs: count of reserved physical eraseblocks
266 * @avail_pebs: count of available physical eraseblocks
267 * @beb_rsvd_pebs: how many physical eraseblocks are reserved for bad PEB
269 * @beb_rsvd_level: normal level of PEBs reserved for bad PEB handling
271 * @autoresize_vol_id: ID of the volume which has to be auto-resized at the end
272 * of UBI ititializetion
273 * @vtbl_slots: how many slots are available in the volume table
274 * @vtbl_size: size of the volume table in bytes
275 * @vtbl: in-RAM volume table copy
276 * @volumes_mutex: protects on-flash volume table and serializes volume
277 * changes, like creation, deletion, update, resize
279 * @max_ec: current highest erase counter value
280 * @mean_ec: current mean erase counter value
282 * @global_sqnum: global sequence number
283 * @ltree_lock: protects the lock tree and @global_sqnum
284 * @ltree: the lock tree
285 * @alc_mutex: serializes "atomic LEB change" operations
287 * @used: RB-tree of used physical eraseblocks
288 * @free: RB-tree of free physical eraseblocks
289 * @scrub: RB-tree of physical eraseblocks which need scrubbing
290 * @prot: protection trees
291 * @prot.pnum: protection tree indexed by physical eraseblock numbers
292 * @prot.aec: protection tree indexed by absolute erase counter value
293 * @wl_lock: protects the @used, @free, @prot, @lookuptbl, @abs_ec, @move_from,
294 * @move_to, @move_to_put @erase_pending, @wl_scheduled, and @works
296 * @move_mutex: serializes eraseblock moves
297 * @wl_scheduled: non-zero if the wear-leveling was scheduled
298 * @lookuptbl: a table to quickly find a &struct ubi_wl_entry object for any
299 * physical eraseblock
300 * @abs_ec: absolute erase counter
301 * @move_from: physical eraseblock from where the data is being moved
302 * @move_to: physical eraseblock where the data is being moved to
303 * @move_to_put: if the "to" PEB was put
304 * @works: list of pending works
305 * @works_count: count of pending works
306 * @bgt_thread: background thread description object
307 * @thread_enabled: if the background thread is enabled
308 * @bgt_name: background thread name
310 * @flash_size: underlying MTD device size (in bytes)
311 * @peb_count: count of physical eraseblocks on the MTD device
312 * @peb_size: physical eraseblock size
313 * @bad_peb_count: count of bad physical eraseblocks
314 * @good_peb_count: count of good physical eraseblocks
315 * @min_io_size: minimal input/output unit size of the underlying MTD device
316 * @hdrs_min_io_size: minimal I/O unit size used for VID and EC headers
317 * @ro_mode: if the UBI device is in read-only mode
318 * @leb_size: logical eraseblock size
319 * @leb_start: starting offset of logical eraseblocks within physical
321 * @ec_hdr_alsize: size of the EC header aligned to @hdrs_min_io_size
322 * @vid_hdr_alsize: size of the VID header aligned to @hdrs_min_io_size
323 * @vid_hdr_offset: starting offset of the volume identifier header (might be
325 * @vid_hdr_aloffset: starting offset of the VID header aligned to
327 * @vid_hdr_shift: contains @vid_hdr_offset - @vid_hdr_aloffset
328 * @bad_allowed: whether the MTD device admits of bad physical eraseblocks or
330 * @mtd: MTD device descriptor
332 * @peb_buf1: a buffer of PEB size used for different purposes
333 * @peb_buf2: another buffer of PEB size used for different purposes
334 * @buf_mutex: proptects @peb_buf1 and @peb_buf2
335 * @dbg_peb_buf: buffer of PEB size used for debugging
336 * @dbg_buf_mutex: proptects @dbg_peb_buf
342 char ubi_name[sizeof(UBI_NAME_STR)+5];
344 struct ubi_volume *volumes[UBI_MAX_VOLUMES+UBI_INT_VOL_COUNT];
345 spinlock_t volumes_lock;
353 int autoresize_vol_id;
356 struct ubi_vtbl_record *vtbl;
357 struct mutex volumes_mutex;
360 /* TODO: mean_ec is not updated run-time, fix */
363 /* EBA unit's stuff */
364 unsigned long long global_sqnum;
365 spinlock_t ltree_lock;
366 struct rb_root ltree;
367 struct mutex alc_mutex;
369 /* Wear-leveling unit's stuff */
372 struct rb_root scrub;
378 struct mutex move_mutex;
379 struct rw_semaphore work_sem;
381 struct ubi_wl_entry **lookuptbl;
382 unsigned long long abs_ec;
383 struct ubi_wl_entry *move_from;
384 struct ubi_wl_entry *move_to;
386 struct list_head works;
388 struct task_struct *bgt_thread;
390 char bgt_name[sizeof(UBI_BGT_NAME_PATTERN)+2];
392 /* I/O unit's stuff */
393 long long flash_size;
399 int hdrs_min_io_size;
406 int vid_hdr_aloffset;
409 struct mtd_info *mtd;
413 struct mutex buf_mutex;
414 struct mutex ckvol_mutex;
415 #ifdef CONFIG_MTD_UBI_DEBUG
417 struct mutex dbg_buf_mutex;
421 extern struct kmem_cache *ubi_wl_entry_slab;
422 extern struct file_operations ubi_ctrl_cdev_operations;
423 extern struct file_operations ubi_cdev_operations;
424 extern struct file_operations ubi_vol_cdev_operations;
425 extern struct class *ubi_class;
426 extern struct mutex ubi_devices_mutex;
429 int ubi_change_vtbl_record(struct ubi_device *ubi, int idx,
430 struct ubi_vtbl_record *vtbl_rec);
431 int ubi_read_volume_table(struct ubi_device *ubi, struct ubi_scan_info *si);
434 int ubi_create_volume(struct ubi_device *ubi, struct ubi_mkvol_req *req);
435 int ubi_remove_volume(struct ubi_volume_desc *desc);
436 int ubi_resize_volume(struct ubi_volume_desc *desc, int reserved_pebs);
437 int ubi_add_volume(struct ubi_device *ubi, struct ubi_volume *vol);
438 void ubi_free_volume(struct ubi_device *ubi, struct ubi_volume *vol);
441 int ubi_start_update(struct ubi_device *ubi, struct ubi_volume *vol,
443 int ubi_more_update_data(struct ubi_device *ubi, struct ubi_volume *vol,
444 const void __user *buf, int count);
445 int ubi_start_leb_change(struct ubi_device *ubi, struct ubi_volume *vol,
446 const struct ubi_leb_change_req *req);
447 int ubi_more_leb_change_data(struct ubi_device *ubi, struct ubi_volume *vol,
448 const void __user *buf, int count);
451 int ubi_calc_data_len(const struct ubi_device *ubi, const void *buf, int length);
452 int ubi_check_volume(struct ubi_device *ubi, int vol_id);
453 void ubi_calculate_reserved(struct ubi_device *ubi);
456 #ifdef CONFIG_MTD_UBI_GLUEBI
457 int ubi_create_gluebi(struct ubi_device *ubi, struct ubi_volume *vol);
458 int ubi_destroy_gluebi(struct ubi_volume *vol);
459 void ubi_gluebi_updated(struct ubi_volume *vol);
461 #define ubi_create_gluebi(ubi, vol) 0
462 #define ubi_destroy_gluebi(vol) 0
463 #define ubi_gluebi_updated(vol)
467 int ubi_eba_unmap_leb(struct ubi_device *ubi, struct ubi_volume *vol,
469 int ubi_eba_read_leb(struct ubi_device *ubi, struct ubi_volume *vol, int lnum,
470 void *buf, int offset, int len, int check);
471 int ubi_eba_write_leb(struct ubi_device *ubi, struct ubi_volume *vol, int lnum,
472 const void *buf, int offset, int len, int dtype);
473 int ubi_eba_write_leb_st(struct ubi_device *ubi, struct ubi_volume *vol,
474 int lnum, const void *buf, int len, int dtype,
476 int ubi_eba_atomic_leb_change(struct ubi_device *ubi, struct ubi_volume *vol,
477 int lnum, const void *buf, int len, int dtype);
478 int ubi_eba_copy_leb(struct ubi_device *ubi, int from, int to,
479 struct ubi_vid_hdr *vid_hdr);
480 int ubi_eba_init_scan(struct ubi_device *ubi, struct ubi_scan_info *si);
481 void ubi_eba_close(const struct ubi_device *ubi);
484 int ubi_wl_get_peb(struct ubi_device *ubi, int dtype);
485 int ubi_wl_put_peb(struct ubi_device *ubi, int pnum, int torture);
486 int ubi_wl_flush(struct ubi_device *ubi);
487 int ubi_wl_scrub_peb(struct ubi_device *ubi, int pnum);
488 int ubi_wl_init_scan(struct ubi_device *ubi, struct ubi_scan_info *si);
489 void ubi_wl_close(struct ubi_device *ubi);
490 int ubi_thread(void *u);
493 int ubi_io_read(const struct ubi_device *ubi, void *buf, int pnum, int offset,
495 int ubi_io_write(struct ubi_device *ubi, const void *buf, int pnum, int offset,
497 int ubi_io_sync_erase(struct ubi_device *ubi, int pnum, int torture);
498 int ubi_io_is_bad(const struct ubi_device *ubi, int pnum);
499 int ubi_io_mark_bad(const struct ubi_device *ubi, int pnum);
500 int ubi_io_read_ec_hdr(struct ubi_device *ubi, int pnum,
501 struct ubi_ec_hdr *ec_hdr, int verbose);
502 int ubi_io_write_ec_hdr(struct ubi_device *ubi, int pnum,
503 struct ubi_ec_hdr *ec_hdr);
504 int ubi_io_read_vid_hdr(struct ubi_device *ubi, int pnum,
505 struct ubi_vid_hdr *vid_hdr, int verbose);
506 int ubi_io_write_vid_hdr(struct ubi_device *ubi, int pnum,
507 struct ubi_vid_hdr *vid_hdr);
510 int ubi_attach_mtd_dev(struct mtd_info *mtd, int ubi_num, int vid_hdr_offset);
511 int ubi_detach_mtd_dev(int ubi_num, int anyway);
512 struct ubi_device *ubi_get_device(int ubi_num);
513 void ubi_put_device(struct ubi_device *ubi);
514 struct ubi_device *ubi_get_by_major(int major);
515 int ubi_major2num(int major);
518 * ubi_rb_for_each_entry - walk an RB-tree.
519 * @rb: a pointer to type 'struct rb_node' to to use as a loop counter
520 * @pos: a pointer to RB-tree entry type to use as a loop counter
521 * @root: RB-tree's root
522 * @member: the name of the 'struct rb_node' within the RB-tree entry
524 #define ubi_rb_for_each_entry(rb, pos, root, member) \
525 for (rb = rb_first(root), \
526 pos = (rb ? container_of(rb, typeof(*pos), member) : NULL); \
528 rb = rb_next(rb), pos = container_of(rb, typeof(*pos), member))
531 * ubi_zalloc_vid_hdr - allocate a volume identifier header object.
532 * @ubi: UBI device description object
533 * @gfp_flags: GFP flags to allocate with
535 * This function returns a pointer to the newly allocated and zero-filled
536 * volume identifier header object in case of success and %NULL in case of
539 static inline struct ubi_vid_hdr *
540 ubi_zalloc_vid_hdr(const struct ubi_device *ubi, gfp_t gfp_flags)
544 vid_hdr = kzalloc(ubi->vid_hdr_alsize, gfp_flags);
549 * VID headers may be stored at un-aligned flash offsets, so we shift
552 return vid_hdr + ubi->vid_hdr_shift;
556 * ubi_free_vid_hdr - free a volume identifier header object.
557 * @ubi: UBI device description object
558 * @vid_hdr: the object to free
560 static inline void ubi_free_vid_hdr(const struct ubi_device *ubi,
561 struct ubi_vid_hdr *vid_hdr)
568 kfree(p - ubi->vid_hdr_shift);
572 * This function is equivalent to 'ubi_io_read()', but @offset is relative to
573 * the beginning of the logical eraseblock, not to the beginning of the
574 * physical eraseblock.
576 static inline int ubi_io_read_data(const struct ubi_device *ubi, void *buf,
577 int pnum, int offset, int len)
579 ubi_assert(offset >= 0);
580 return ubi_io_read(ubi, buf, pnum, offset + ubi->leb_start, len);
584 * This function is equivalent to 'ubi_io_write()', but @offset is relative to
585 * the beginning of the logical eraseblock, not to the beginning of the
586 * physical eraseblock.
588 static inline int ubi_io_write_data(struct ubi_device *ubi, const void *buf,
589 int pnum, int offset, int len)
591 ubi_assert(offset >= 0);
592 return ubi_io_write(ubi, buf, pnum, offset + ubi->leb_start, len);
596 * ubi_ro_mode - switch to read-only mode.
597 * @ubi: UBI device description object
599 static inline void ubi_ro_mode(struct ubi_device *ubi)
603 ubi_warn("switch to read-only mode");
608 * vol_id2idx - get table index by volume ID.
609 * @ubi: UBI device description object
612 static inline int vol_id2idx(const struct ubi_device *ubi, int vol_id)
614 if (vol_id >= UBI_INTERNAL_VOL_START)
615 return vol_id - UBI_INTERNAL_VOL_START + ubi->vtbl_slots;
621 * idx2vol_id - get volume ID by table index.
622 * @ubi: UBI device description object
625 static inline int idx2vol_id(const struct ubi_device *ubi, int idx)
627 if (idx >= ubi->vtbl_slots)
628 return idx - ubi->vtbl_slots + UBI_INTERNAL_VOL_START;
633 #endif /* !__UBI_UBI_H__ */