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>
40 #include <linux/mtd/ubi.h>
42 #include "ubi-media.h"
46 /* Maximum number of supported UBI devices */
47 #define UBI_MAX_DEVICES 32
49 /* UBI name used for character devices, sysfs, etc */
50 #define UBI_NAME_STR "ubi"
52 /* Normal UBI messages */
53 #define ubi_msg(fmt, ...) printk(KERN_NOTICE "UBI: " fmt "\n", ##__VA_ARGS__)
54 /* UBI warning messages */
55 #define ubi_warn(fmt, ...) printk(KERN_WARNING "UBI warning: %s: " fmt "\n", \
56 __func__, ##__VA_ARGS__)
57 /* UBI error messages */
58 #define ubi_err(fmt, ...) printk(KERN_ERR "UBI error: %s: " fmt "\n", \
59 __func__, ##__VA_ARGS__)
61 /* Lowest number PEBs reserved for bad PEB handling */
62 #define MIN_RESEVED_PEBS 2
64 /* Background thread name pattern */
65 #define UBI_BGT_NAME_PATTERN "ubi_bgt%dd"
67 /* This marker in the EBA table means that the LEB is um-mapped */
68 #define UBI_LEB_UNMAPPED -1
71 * In case of errors, UBI tries to repeat the operation several times before
72 * returning error. The below constant defines how many times UBI re-tries.
74 #define UBI_IO_RETRIES 3
77 * Length of the protection queue. The length is effectively equivalent to the
78 * number of (global) erase cycles PEBs are protected from the wear-leveling
81 #define UBI_PROT_QUEUE_LEN 10
84 * Error codes returned by the I/O sub-system.
86 * UBI_IO_PEB_EMPTY: the physical eraseblock is empty, i.e. it contains only
88 * UBI_IO_PEB_FREE: the physical eraseblock is free, i.e. it contains only a
89 * valid erase counter header, and the rest are %0xFF bytes
90 * UBI_IO_BAD_EC_HDR: the erase counter header is corrupted (bad magic or CRC)
91 * UBI_IO_BAD_VID_HDR: the volume identifier header is corrupted (bad magic or
93 * UBI_IO_BITFLIPS: bit-flips were detected and corrected
104 * struct ubi_wl_entry - wear-leveling entry.
105 * @u.rb: link in the corresponding (free/used) RB-tree
106 * @u.list: link in the protection queue
108 * @pnum: physical eraseblock number
110 * This data structure is used in the WL sub-system. Each physical eraseblock
111 * has a corresponding &struct wl_entry object which may be kept in different
112 * RB-trees. See WL sub-system for details.
114 struct ubi_wl_entry {
117 struct list_head list;
124 * struct ubi_ltree_entry - an entry in the lock tree.
125 * @rb: links RB-tree nodes
126 * @vol_id: volume ID of the locked logical eraseblock
127 * @lnum: locked logical eraseblock number
128 * @users: how many tasks are using this logical eraseblock or wait for it
129 * @mutex: read/write mutex to implement read/write access serialization to
130 * the (@vol_id, @lnum) logical eraseblock
132 * This data structure is used in the EBA sub-system to implement per-LEB
133 * locking. When a logical eraseblock is being locked - corresponding
134 * &struct ubi_ltree_entry object is inserted to the lock tree (@ubi->ltree).
135 * See EBA sub-system for details.
137 struct ubi_ltree_entry {
142 struct rw_semaphore mutex;
146 * struct ubi_rename_entry - volume re-name description data structure.
147 * @new_name_len: new volume name length
148 * @new_name: new volume name
149 * @remove: if not zero, this volume should be removed, not re-named
150 * @desc: descriptor of the volume
151 * @list: links re-name entries into a list
153 * This data structure is utilized in the multiple volume re-name code. Namely,
154 * UBI first creates a list of &struct ubi_rename_entry objects from the
155 * &struct ubi_rnvol_req request object, and then utilizes this list to do all
158 struct ubi_rename_entry {
160 char new_name[UBI_VOL_NAME_MAX + 1];
162 struct ubi_volume_desc *desc;
163 struct list_head list;
166 struct ubi_volume_desc;
169 * struct ubi_volume - UBI volume description data structure.
170 * @dev: device object to make use of the the Linux device model
171 * @cdev: character device object to create character device
172 * @ubi: reference to the UBI device description object
174 * @ref_count: volume reference count
175 * @readers: number of users holding this volume in read-only mode
176 * @writers: number of users holding this volume in read-write mode
177 * @exclusive: whether somebody holds this volume in exclusive mode
179 * @reserved_pebs: how many physical eraseblocks are reserved for this volume
180 * @vol_type: volume type (%UBI_DYNAMIC_VOLUME or %UBI_STATIC_VOLUME)
181 * @usable_leb_size: logical eraseblock size without padding
182 * @used_ebs: how many logical eraseblocks in this volume contain data
183 * @last_eb_bytes: how many bytes are stored in the last logical eraseblock
184 * @used_bytes: how many bytes of data this volume contains
185 * @alignment: volume alignment
186 * @data_pad: how many bytes are not used at the end of physical eraseblocks to
187 * satisfy the requested alignment
188 * @name_len: volume name length
191 * @upd_ebs: how many eraseblocks are expected to be updated
192 * @ch_lnum: LEB number which is being changing by the atomic LEB change
194 * @ch_dtype: data persistency type which is being changing by the atomic LEB
196 * @upd_bytes: how many bytes are expected to be received for volume update or
198 * @upd_received: how many bytes were already received for volume update or
200 * @upd_buf: update buffer which is used to collect update data or data for
203 * @eba_tbl: EBA table of this volume (LEB->PEB mapping)
204 * @checked: %1 if this static volume was checked
205 * @corrupted: %1 if the volume is corrupted (static volumes only)
206 * @upd_marker: %1 if the update marker is set for this volume
207 * @updating: %1 if the volume is being updated
208 * @changing_leb: %1 if the atomic LEB change ioctl command is in progress
210 * @gluebi_desc: gluebi UBI volume descriptor
211 * @gluebi_refcount: reference count of the gluebi MTD device
212 * @gluebi_mtd: MTD device description object of the gluebi MTD device
214 * The @corrupted field indicates that the volume's contents is corrupted.
215 * Since UBI protects only static volumes, this field is not relevant to
216 * dynamic volumes - it is user's responsibility to assure their data
219 * The @upd_marker flag indicates that this volume is either being updated at
220 * the moment or is damaged because of an unclean reboot.
225 struct ubi_device *ubi;
237 long long used_bytes;
241 char name[UBI_VOL_NAME_MAX + 1];
247 long long upd_received;
251 unsigned int checked:1;
252 unsigned int corrupted:1;
253 unsigned int upd_marker:1;
254 unsigned int updating:1;
255 unsigned int changing_leb:1;
257 #ifdef CONFIG_MTD_UBI_GLUEBI
259 * Gluebi-related stuff may be compiled out.
260 * Note: this should not be built into UBI but should be a separate
261 * ubimtd driver which works on top of UBI and emulates MTD devices.
263 struct ubi_volume_desc *gluebi_desc;
265 struct mtd_info gluebi_mtd;
270 * struct ubi_volume_desc - UBI volume descriptor returned when it is opened.
271 * @vol: reference to the corresponding volume description object
272 * @mode: open mode (%UBI_READONLY, %UBI_READWRITE, or %UBI_EXCLUSIVE)
274 struct ubi_volume_desc {
275 struct ubi_volume *vol;
282 * struct ubi_device - UBI device description structure
283 * @dev: UBI device object to use the the Linux device model
284 * @cdev: character device object to create character device
285 * @ubi_num: UBI device number
286 * @ubi_name: UBI device name
287 * @vol_count: number of volumes in this UBI device
288 * @volumes: volumes of this UBI device
289 * @volumes_lock: protects @volumes, @rsvd_pebs, @avail_pebs, beb_rsvd_pebs,
290 * @beb_rsvd_level, @bad_peb_count, @good_peb_count, @vol_count,
291 * @vol->readers, @vol->writers, @vol->exclusive,
292 * @vol->ref_count, @vol->mapping and @vol->eba_tbl.
293 * @ref_count: count of references on the UBI device
295 * @rsvd_pebs: count of reserved physical eraseblocks
296 * @avail_pebs: count of available physical eraseblocks
297 * @beb_rsvd_pebs: how many physical eraseblocks are reserved for bad PEB
299 * @beb_rsvd_level: normal level of PEBs reserved for bad PEB handling
301 * @autoresize_vol_id: ID of the volume which has to be auto-resized at the end
302 * of UBI initialization
303 * @vtbl_slots: how many slots are available in the volume table
304 * @vtbl_size: size of the volume table in bytes
305 * @vtbl: in-RAM volume table copy
306 * @volumes_mutex: protects on-flash volume table and serializes volume
307 * changes, like creation, deletion, update, re-size and re-name
309 * @max_ec: current highest erase counter value
310 * @mean_ec: current mean erase counter value
312 * @global_sqnum: global sequence number
313 * @ltree_lock: protects the lock tree and @global_sqnum
314 * @ltree: the lock tree
315 * @alc_mutex: serializes "atomic LEB change" operations
317 * @used: RB-tree of used physical eraseblocks
318 * @free: RB-tree of free physical eraseblocks
319 * @scrub: RB-tree of physical eraseblocks which need scrubbing
320 * @pq: protection queue (contain physical eraseblocks which are temporarily
321 * protected from the wear-leveling worker)
322 * @pq_head: protection queue head
323 * @wl_lock: protects the @used, @free, @pq, @pq_head, @lookuptbl, @move_from,
324 * @move_to, @move_to_put @erase_pending, @wl_scheduled and @works
326 * @move_mutex: serializes eraseblock moves
327 * @work_sem: synchronizes the WL worker with use tasks
328 * @wl_scheduled: non-zero if the wear-leveling was scheduled
329 * @lookuptbl: a table to quickly find a &struct ubi_wl_entry object for any
330 * physical eraseblock
331 * @move_from: physical eraseblock from where the data is being moved
332 * @move_to: physical eraseblock where the data is being moved to
333 * @move_to_put: if the "to" PEB was put
334 * @works: list of pending works
335 * @works_count: count of pending works
336 * @bgt_thread: background thread description object
337 * @thread_enabled: if the background thread is enabled
338 * @bgt_name: background thread name
340 * @flash_size: underlying MTD device size (in bytes)
341 * @peb_count: count of physical eraseblocks on the MTD device
342 * @peb_size: physical eraseblock size
343 * @bad_peb_count: count of bad physical eraseblocks
344 * @good_peb_count: count of good physical eraseblocks
345 * @min_io_size: minimal input/output unit size of the underlying MTD device
346 * @hdrs_min_io_size: minimal I/O unit size used for VID and EC headers
347 * @ro_mode: if the UBI device is in read-only mode
348 * @leb_size: logical eraseblock size
349 * @leb_start: starting offset of logical eraseblocks within physical
351 * @ec_hdr_alsize: size of the EC header aligned to @hdrs_min_io_size
352 * @vid_hdr_alsize: size of the VID header aligned to @hdrs_min_io_size
353 * @vid_hdr_offset: starting offset of the volume identifier header (might be
355 * @vid_hdr_aloffset: starting offset of the VID header aligned to
357 * @vid_hdr_shift: contains @vid_hdr_offset - @vid_hdr_aloffset
358 * @bad_allowed: whether the MTD device admits of bad physical eraseblocks or
360 * @mtd: MTD device descriptor
362 * @peb_buf1: a buffer of PEB size used for different purposes
363 * @peb_buf2: another buffer of PEB size used for different purposes
364 * @buf_mutex: protects @peb_buf1 and @peb_buf2
365 * @ckvol_mutex: serializes static volume checking when opening
366 * @mult_mutex: serializes operations on multiple volumes, like re-naming
367 * @dbg_peb_buf: buffer of PEB size used for debugging
368 * @dbg_buf_mutex: protects @dbg_peb_buf
374 char ubi_name[sizeof(UBI_NAME_STR)+5];
376 struct ubi_volume *volumes[UBI_MAX_VOLUMES+UBI_INT_VOL_COUNT];
377 spinlock_t volumes_lock;
385 int autoresize_vol_id;
388 struct ubi_vtbl_record *vtbl;
389 struct mutex volumes_mutex;
392 /* Note, mean_ec is not updated run-time - should be fixed */
395 /* EBA sub-system's stuff */
396 unsigned long long global_sqnum;
397 spinlock_t ltree_lock;
398 struct rb_root ltree;
399 struct mutex alc_mutex;
401 /* Wear-leveling sub-system's stuff */
404 struct rb_root scrub;
405 struct list_head pq[UBI_PROT_QUEUE_LEN];
408 struct mutex move_mutex;
409 struct rw_semaphore work_sem;
411 struct ubi_wl_entry **lookuptbl;
412 struct ubi_wl_entry *move_from;
413 struct ubi_wl_entry *move_to;
415 struct list_head works;
417 struct task_struct *bgt_thread;
419 char bgt_name[sizeof(UBI_BGT_NAME_PATTERN)+2];
421 /* I/O sub-system's stuff */
422 long long flash_size;
428 int hdrs_min_io_size;
435 int vid_hdr_aloffset;
438 struct mtd_info *mtd;
442 struct mutex buf_mutex;
443 struct mutex ckvol_mutex;
444 struct mutex mult_mutex;
445 #ifdef CONFIG_MTD_UBI_DEBUG
447 struct mutex dbg_buf_mutex;
451 extern struct kmem_cache *ubi_wl_entry_slab;
452 extern struct file_operations ubi_ctrl_cdev_operations;
453 extern struct file_operations ubi_cdev_operations;
454 extern struct file_operations ubi_vol_cdev_operations;
455 extern struct class *ubi_class;
456 extern struct mutex ubi_devices_mutex;
459 int ubi_change_vtbl_record(struct ubi_device *ubi, int idx,
460 struct ubi_vtbl_record *vtbl_rec);
461 int ubi_vtbl_rename_volumes(struct ubi_device *ubi,
462 struct list_head *rename_list);
463 int ubi_read_volume_table(struct ubi_device *ubi, struct ubi_scan_info *si);
466 int ubi_create_volume(struct ubi_device *ubi, struct ubi_mkvol_req *req);
467 int ubi_remove_volume(struct ubi_volume_desc *desc, int no_vtbl);
468 int ubi_resize_volume(struct ubi_volume_desc *desc, int reserved_pebs);
469 int ubi_rename_volumes(struct ubi_device *ubi, struct list_head *rename_list);
470 int ubi_add_volume(struct ubi_device *ubi, struct ubi_volume *vol);
471 void ubi_free_volume(struct ubi_device *ubi, struct ubi_volume *vol);
474 int ubi_start_update(struct ubi_device *ubi, struct ubi_volume *vol,
476 int ubi_more_update_data(struct ubi_device *ubi, struct ubi_volume *vol,
477 const void __user *buf, int count);
478 int ubi_start_leb_change(struct ubi_device *ubi, struct ubi_volume *vol,
479 const struct ubi_leb_change_req *req);
480 int ubi_more_leb_change_data(struct ubi_device *ubi, struct ubi_volume *vol,
481 const void __user *buf, int count);
484 int ubi_calc_data_len(const struct ubi_device *ubi, const void *buf,
486 int ubi_check_volume(struct ubi_device *ubi, int vol_id);
487 void ubi_calculate_reserved(struct ubi_device *ubi);
490 #ifdef CONFIG_MTD_UBI_GLUEBI
491 int ubi_create_gluebi(struct ubi_device *ubi, struct ubi_volume *vol);
492 int ubi_destroy_gluebi(struct ubi_volume *vol);
493 void ubi_gluebi_updated(struct ubi_volume *vol);
495 #define ubi_create_gluebi(ubi, vol) 0
496 #define ubi_destroy_gluebi(vol) 0
497 #define ubi_gluebi_updated(vol)
501 int ubi_eba_unmap_leb(struct ubi_device *ubi, struct ubi_volume *vol,
503 int ubi_eba_read_leb(struct ubi_device *ubi, struct ubi_volume *vol, int lnum,
504 void *buf, int offset, int len, int check);
505 int ubi_eba_write_leb(struct ubi_device *ubi, struct ubi_volume *vol, int lnum,
506 const void *buf, int offset, int len, int dtype);
507 int ubi_eba_write_leb_st(struct ubi_device *ubi, struct ubi_volume *vol,
508 int lnum, const void *buf, int len, int dtype,
510 int ubi_eba_atomic_leb_change(struct ubi_device *ubi, struct ubi_volume *vol,
511 int lnum, const void *buf, int len, int dtype);
512 int ubi_eba_copy_leb(struct ubi_device *ubi, int from, int to,
513 struct ubi_vid_hdr *vid_hdr);
514 int ubi_eba_init_scan(struct ubi_device *ubi, struct ubi_scan_info *si);
517 int ubi_wl_get_peb(struct ubi_device *ubi, int dtype);
518 int ubi_wl_put_peb(struct ubi_device *ubi, int pnum, int torture);
519 int ubi_wl_flush(struct ubi_device *ubi);
520 int ubi_wl_scrub_peb(struct ubi_device *ubi, int pnum);
521 int ubi_wl_init_scan(struct ubi_device *ubi, struct ubi_scan_info *si);
522 void ubi_wl_close(struct ubi_device *ubi);
523 int ubi_thread(void *u);
526 int ubi_io_read(const struct ubi_device *ubi, void *buf, int pnum, int offset,
528 int ubi_io_write(struct ubi_device *ubi, const void *buf, int pnum, int offset,
530 int ubi_io_sync_erase(struct ubi_device *ubi, int pnum, int torture);
531 int ubi_io_is_bad(const struct ubi_device *ubi, int pnum);
532 int ubi_io_mark_bad(const struct ubi_device *ubi, int pnum);
533 int ubi_io_read_ec_hdr(struct ubi_device *ubi, int pnum,
534 struct ubi_ec_hdr *ec_hdr, int verbose);
535 int ubi_io_write_ec_hdr(struct ubi_device *ubi, int pnum,
536 struct ubi_ec_hdr *ec_hdr);
537 int ubi_io_read_vid_hdr(struct ubi_device *ubi, int pnum,
538 struct ubi_vid_hdr *vid_hdr, int verbose);
539 int ubi_io_write_vid_hdr(struct ubi_device *ubi, int pnum,
540 struct ubi_vid_hdr *vid_hdr);
543 int ubi_attach_mtd_dev(struct mtd_info *mtd, int ubi_num, int vid_hdr_offset);
544 int ubi_detach_mtd_dev(int ubi_num, int anyway);
545 struct ubi_device *ubi_get_device(int ubi_num);
546 void ubi_put_device(struct ubi_device *ubi);
547 struct ubi_device *ubi_get_by_major(int major);
548 int ubi_major2num(int major);
551 * ubi_rb_for_each_entry - walk an RB-tree.
552 * @rb: a pointer to type 'struct rb_node' to to use as a loop counter
553 * @pos: a pointer to RB-tree entry type to use as a loop counter
554 * @root: RB-tree's root
555 * @member: the name of the 'struct rb_node' within the RB-tree entry
557 #define ubi_rb_for_each_entry(rb, pos, root, member) \
558 for (rb = rb_first(root), \
559 pos = (rb ? container_of(rb, typeof(*pos), member) : NULL); \
561 rb = rb_next(rb), pos = container_of(rb, typeof(*pos), member))
564 * ubi_zalloc_vid_hdr - allocate a volume identifier header object.
565 * @ubi: UBI device description object
566 * @gfp_flags: GFP flags to allocate with
568 * This function returns a pointer to the newly allocated and zero-filled
569 * volume identifier header object in case of success and %NULL in case of
572 static inline struct ubi_vid_hdr *
573 ubi_zalloc_vid_hdr(const struct ubi_device *ubi, gfp_t gfp_flags)
577 vid_hdr = kzalloc(ubi->vid_hdr_alsize, gfp_flags);
582 * VID headers may be stored at un-aligned flash offsets, so we shift
585 return vid_hdr + ubi->vid_hdr_shift;
589 * ubi_free_vid_hdr - free a volume identifier header object.
590 * @ubi: UBI device description object
591 * @vid_hdr: the object to free
593 static inline void ubi_free_vid_hdr(const struct ubi_device *ubi,
594 struct ubi_vid_hdr *vid_hdr)
601 kfree(p - ubi->vid_hdr_shift);
605 * This function is equivalent to 'ubi_io_read()', but @offset is relative to
606 * the beginning of the logical eraseblock, not to the beginning of the
607 * physical eraseblock.
609 static inline int ubi_io_read_data(const struct ubi_device *ubi, void *buf,
610 int pnum, int offset, int len)
612 ubi_assert(offset >= 0);
613 return ubi_io_read(ubi, buf, pnum, offset + ubi->leb_start, len);
617 * This function is equivalent to 'ubi_io_write()', but @offset is relative to
618 * the beginning of the logical eraseblock, not to the beginning of the
619 * physical eraseblock.
621 static inline int ubi_io_write_data(struct ubi_device *ubi, const void *buf,
622 int pnum, int offset, int len)
624 ubi_assert(offset >= 0);
625 return ubi_io_write(ubi, buf, pnum, offset + ubi->leb_start, len);
629 * ubi_ro_mode - switch to read-only mode.
630 * @ubi: UBI device description object
632 static inline void ubi_ro_mode(struct ubi_device *ubi)
636 ubi_warn("switch to read-only mode");
641 * vol_id2idx - get table index by volume ID.
642 * @ubi: UBI device description object
645 static inline int vol_id2idx(const struct ubi_device *ubi, int vol_id)
647 if (vol_id >= UBI_INTERNAL_VOL_START)
648 return vol_id - UBI_INTERNAL_VOL_START + ubi->vtbl_slots;
654 * idx2vol_id - get volume ID by table index.
655 * @ubi: UBI device description object
658 static inline int idx2vol_id(const struct ubi_device *ubi, int idx)
660 if (idx >= ubi->vtbl_slots)
661 return idx - ubi->vtbl_slots + UBI_INTERNAL_VOL_START;
666 #endif /* !__UBI_UBI_H__ */