Btrfs: fix unprotected device list access when cloning fs devices
[deliverable/linux.git] / fs / btrfs / volumes.h
CommitLineData
0b86a832
CM
1/*
2 * Copyright (C) 2007 Oracle. All rights reserved.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
17 */
18
19#ifndef __BTRFS_VOLUMES_
20#define __BTRFS_VOLUMES_
8790d502 21
cea9e445 22#include <linux/bio.h>
b2117a39 23#include <linux/sort.h>
55e301fd 24#include <linux/btrfs.h>
8b712842 25#include "async-thread.h"
cea9e445 26
b2117a39
MX
27#define BTRFS_STRIPE_LEN (64 * 1024)
28
f2984462 29struct buffer_head;
ffbd517d
CM
30struct btrfs_pending_bios {
31 struct bio *head;
32 struct bio *tail;
33};
34
7cc8e58d
MX
35/*
36 * Use sequence counter to get consistent device stat data on
37 * 32-bit processors.
38 */
39#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
40#include <linux/seqlock.h>
41#define __BTRFS_NEED_DEVICE_DATA_ORDERED
42#define btrfs_device_data_ordered_init(device) \
43 seqcount_init(&device->data_seqcount)
44#else
45#define btrfs_device_data_ordered_init(device) do { } while (0)
46#endif
47
0b86a832
CM
48struct btrfs_device {
49 struct list_head dev_list;
b3075717 50 struct list_head dev_alloc_list;
2b82032c 51 struct btrfs_fs_devices *fs_devices;
d5ee37bc 52
0b86a832 53 struct btrfs_root *dev_root;
ffbd517d 54
d5ee37bc
MX
55 struct rcu_string *name;
56
57 u64 generation;
58
59 spinlock_t io_lock ____cacheline_aligned;
60 int running_pending;
ffbd517d
CM
61 /* regular prio bios */
62 struct btrfs_pending_bios pending_bios;
63 /* WRITE_SYNC bios */
64 struct btrfs_pending_bios pending_sync_bios;
65
d5ee37bc
MX
66 struct block_device *bdev;
67
68 /* the mode sent to blkdev_get */
69 fmode_t mode;
70
2b82032c 71 int writeable;
dfe25020 72 int in_fs_metadata;
cd02dca5 73 int missing;
d5e2003c 74 int can_discard;
63a212ab 75 int is_tgtdev_for_dev_replace;
b3075717 76
7cc8e58d
MX
77#ifdef __BTRFS_NEED_DEVICE_DATA_ORDERED
78 seqcount_t data_seqcount;
79#endif
80
0b86a832
CM
81 /* the internal btrfs device id */
82 u64 devid;
83
6ba40b61 84 /* size of the device in memory */
0b86a832
CM
85 u64 total_bytes;
86
6ba40b61 87 /* size of the device on disk */
d6397bae
CB
88 u64 disk_total_bytes;
89
0b86a832
CM
90 /* bytes used */
91 u64 bytes_used;
92
93 /* optimal io alignment for this device */
94 u32 io_align;
95
96 /* optimal io width for this device */
97 u32 io_width;
3c45bfc1
DG
98 /* type and info about this device */
99 u64 type;
0b86a832
CM
100
101 /* minimal io size for this device */
102 u32 sector_size;
103
0b86a832 104 /* physical drive uuid (or lvm uuid) */
e17cade2 105 u8 uuid[BTRFS_UUID_SIZE];
8b712842 106
935e5cc9
MX
107 /*
108 * size of the device on the current transaction
109 *
110 * This variant is update when committing the transaction,
111 * and protected by device_list_mutex
112 */
113 u64 commit_total_bytes;
114
ce7213c7
MX
115 /* bytes used on the current transaction */
116 u64 commit_bytes_used;
935e5cc9
MX
117 /*
118 * used to manage the device which is resized
119 *
120 * It is protected by chunk_lock.
121 */
122 struct list_head resized_list;
123
3c45bfc1
DG
124 /* for sending down flush barriers */
125 int nobarriers;
126 struct bio *flush_bio;
127 struct completion flush_wait;
128
a2de733c 129 /* per-device scrub information */
d9d181c1 130 struct scrub_ctx *scrub_device;
a2de733c 131
d458b054 132 struct btrfs_work work;
1f78160c
XG
133 struct rcu_head rcu;
134 struct work_struct rcu_work;
90519d66
AJ
135
136 /* readahead state */
137 spinlock_t reada_lock;
138 atomic_t reada_in_flight;
139 u64 reada_next;
140 struct reada_zone *reada_curr_zone;
141 struct radix_tree_root reada_zones;
142 struct radix_tree_root reada_extents;
387125fc 143
442a4f63
SB
144 /* disk I/O failure stats. For detailed description refer to
145 * enum btrfs_dev_stat_values in ioctl.h */
733f4fbb 146 int dev_stats_valid;
addc3fa7
MX
147
148 /* Counter to record the change of device stats */
149 atomic_t dev_stats_ccnt;
442a4f63 150 atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX];
0b86a832
CM
151};
152
7cc8e58d
MX
153/*
154 * If we read those variants at the context of their own lock, we needn't
155 * use the following helpers, reading them directly is safe.
156 */
157#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
158#define BTRFS_DEVICE_GETSET_FUNCS(name) \
159static inline u64 \
160btrfs_device_get_##name(const struct btrfs_device *dev) \
161{ \
162 u64 size; \
163 unsigned int seq; \
164 \
165 do { \
166 seq = read_seqcount_begin(&dev->data_seqcount); \
167 size = dev->name; \
168 } while (read_seqcount_retry(&dev->data_seqcount, seq)); \
169 return size; \
170} \
171 \
172static inline void \
173btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
174{ \
175 preempt_disable(); \
176 write_seqcount_begin(&dev->data_seqcount); \
177 dev->name = size; \
178 write_seqcount_end(&dev->data_seqcount); \
179 preempt_enable(); \
180}
181#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
182#define BTRFS_DEVICE_GETSET_FUNCS(name) \
183static inline u64 \
184btrfs_device_get_##name(const struct btrfs_device *dev) \
185{ \
186 u64 size; \
187 \
188 preempt_disable(); \
189 size = dev->name; \
190 preempt_enable(); \
191 return size; \
192} \
193 \
194static inline void \
195btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
196{ \
197 preempt_disable(); \
198 dev->name = size; \
199 preempt_enable(); \
200}
201#else
202#define BTRFS_DEVICE_GETSET_FUNCS(name) \
203static inline u64 \
204btrfs_device_get_##name(const struct btrfs_device *dev) \
205{ \
206 return dev->name; \
207} \
208 \
209static inline void \
210btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
211{ \
212 dev->name = size; \
213}
214#endif
215
216BTRFS_DEVICE_GETSET_FUNCS(total_bytes);
217BTRFS_DEVICE_GETSET_FUNCS(disk_total_bytes);
218BTRFS_DEVICE_GETSET_FUNCS(bytes_used);
219
8a4b83cc
CM
220struct btrfs_fs_devices {
221 u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
222
8a4b83cc 223 u64 num_devices;
a0af469b 224 u64 open_devices;
2b82032c 225 u64 rw_devices;
cd02dca5 226 u64 missing_devices;
2b82032c 227 u64 total_rw_bytes;
02db0844 228 u64 total_devices;
8a4b83cc 229 struct block_device *latest_bdev;
e5e9a520
CM
230
231 /* all of the devices in the FS, protected by a mutex
232 * so we can safely walk it to write out the supers without
9b011adf
WS
233 * worrying about add/remove by the multi-device code.
234 * Scrubbing super can kick off supers writing by holding
235 * this mutex lock.
e5e9a520
CM
236 */
237 struct mutex device_list_mutex;
8a4b83cc 238 struct list_head devices;
b3075717 239
935e5cc9 240 struct list_head resized_devices;
b3075717
CM
241 /* devices not currently being allocated */
242 struct list_head alloc_list;
8a4b83cc 243 struct list_head list;
2b82032c
YZ
244
245 struct btrfs_fs_devices *seed;
246 int seeding;
2b82032c
YZ
247
248 int opened;
c289811c
CM
249
250 /* set when we find or add a device that doesn't have the
251 * nonrot flag set
252 */
253 int rotating;
8a4b83cc
CM
254};
255
facc8a22
MX
256#define BTRFS_BIO_INLINE_CSUM_SIZE 64
257
9be3395b
CM
258/*
259 * we need the mirror number and stripe index to be passed around
260 * the call chain while we are processing end_io (especially errors).
261 * Really, what we need is a btrfs_bio structure that has this info
262 * and is properly sized with its stripe array, but we're not there
263 * quite yet. We have our own btrfs bioset, and all of the bios
264 * we allocate are actually btrfs_io_bios. We'll cram as much of
265 * struct btrfs_bio as we can into this over time.
266 */
facc8a22 267typedef void (btrfs_io_bio_end_io_t) (struct btrfs_io_bio *bio, int err);
9be3395b
CM
268struct btrfs_io_bio {
269 unsigned long mirror_num;
270 unsigned long stripe_index;
facc8a22
MX
271 u8 *csum;
272 u8 csum_inline[BTRFS_BIO_INLINE_CSUM_SIZE];
273 u8 *csum_allocated;
274 btrfs_io_bio_end_io_t *end_io;
9be3395b
CM
275 struct bio bio;
276};
277
278static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio)
279{
280 return container_of(bio, struct btrfs_io_bio, bio);
281}
282
cea9e445
CM
283struct btrfs_bio_stripe {
284 struct btrfs_device *dev;
285 u64 physical;
fce3bb9a 286 u64 length; /* only used for discard mappings */
cea9e445
CM
287};
288
a1d3c478
JS
289struct btrfs_bio;
290typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
291
c55f1396
MX
292#define BTRFS_BIO_ORIG_BIO_SUBMITTED 0x1
293
a1d3c478 294struct btrfs_bio {
cea9e445 295 atomic_t stripes_pending;
c404e0dc 296 struct btrfs_fs_info *fs_info;
cea9e445 297 bio_end_io_t *end_io;
7d2b4daa 298 struct bio *orig_bio;
c55f1396 299 unsigned long flags;
cea9e445 300 void *private;
a236aed1
CM
301 atomic_t error;
302 int max_errors;
cea9e445 303 int num_stripes;
a1d3c478 304 int mirror_num;
cea9e445
CM
305 struct btrfs_bio_stripe stripes[];
306};
307
b2117a39
MX
308struct btrfs_device_info {
309 struct btrfs_device *dev;
310 u64 dev_offset;
311 u64 max_avail;
73c5de00 312 u64 total_avail;
b2117a39
MX
313};
314
31e50229
LB
315struct btrfs_raid_attr {
316 int sub_stripes; /* sub_stripes info for map */
317 int dev_stripes; /* stripes per dev */
318 int devs_max; /* max devs to use */
319 int devs_min; /* min devs needed */
320 int devs_increment; /* ndevs has to be a multiple of this */
321 int ncopies; /* how many copies to data has */
322};
323
1abe9b8a 324struct map_lookup {
325 u64 type;
326 int io_align;
327 int io_width;
328 int stripe_len;
329 int sector_size;
330 int num_stripes;
331 int sub_stripes;
332 struct btrfs_bio_stripe stripes[];
333};
334
a2de733c
AJ
335#define map_lookup_size(n) (sizeof(struct map_lookup) + \
336 (sizeof(struct btrfs_bio_stripe) * (n)))
337
f43ffb60
ID
338/*
339 * Restriper's general type filter
340 */
341#define BTRFS_BALANCE_DATA (1ULL << 0)
342#define BTRFS_BALANCE_SYSTEM (1ULL << 1)
343#define BTRFS_BALANCE_METADATA (1ULL << 2)
344
345#define BTRFS_BALANCE_TYPE_MASK (BTRFS_BALANCE_DATA | \
346 BTRFS_BALANCE_SYSTEM | \
347 BTRFS_BALANCE_METADATA)
e4d8ec0f
ID
348
349#define BTRFS_BALANCE_FORCE (1ULL << 3)
59641015 350#define BTRFS_BALANCE_RESUME (1ULL << 4)
e4d8ec0f 351
ed25e9b2
ID
352/*
353 * Balance filters
354 */
355#define BTRFS_BALANCE_ARGS_PROFILES (1ULL << 0)
5ce5b3c0 356#define BTRFS_BALANCE_ARGS_USAGE (1ULL << 1)
409d404b 357#define BTRFS_BALANCE_ARGS_DEVID (1ULL << 2)
94e60d5a 358#define BTRFS_BALANCE_ARGS_DRANGE (1ULL << 3)
ea67176a 359#define BTRFS_BALANCE_ARGS_VRANGE (1ULL << 4)
7d824b6f 360#define BTRFS_BALANCE_ARGS_LIMIT (1ULL << 5)
f43ffb60 361
cfa4c961
ID
362/*
363 * Profile changing flags. When SOFT is set we won't relocate chunk if
364 * it already has the target profile (even though it may be
365 * half-filled).
366 */
e4d8ec0f 367#define BTRFS_BALANCE_ARGS_CONVERT (1ULL << 8)
cfa4c961 368#define BTRFS_BALANCE_ARGS_SOFT (1ULL << 9)
e4d8ec0f 369
c9e9f97b 370struct btrfs_balance_args;
19a39dce 371struct btrfs_balance_progress;
c9e9f97b
ID
372struct btrfs_balance_control {
373 struct btrfs_fs_info *fs_info;
374
375 struct btrfs_balance_args data;
376 struct btrfs_balance_args meta;
377 struct btrfs_balance_args sys;
378
379 u64 flags;
19a39dce
ID
380
381 struct btrfs_balance_progress stat;
c9e9f97b
ID
382};
383
6d07bcec
MX
384int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
385 u64 end, u64 *length);
386
a1d3c478 387#define btrfs_bio_size(n) (sizeof(struct btrfs_bio) + \
cea9e445
CM
388 (sizeof(struct btrfs_bio_stripe) * (n)))
389
3ec706c8 390int btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
cea9e445 391 u64 logical, u64 *length,
a1d3c478 392 struct btrfs_bio **bbio_ret, int mirror_num);
a512bbf8
YZ
393int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
394 u64 chunk_start, u64 physical, u64 devid,
395 u64 **logical, int *naddrs, int *stripe_len);
e4404d6e 396int btrfs_read_sys_array(struct btrfs_root *root);
0b86a832
CM
397int btrfs_read_chunk_tree(struct btrfs_root *root);
398int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
2b82032c 399 struct btrfs_root *extent_root, u64 type);
0b86a832
CM
400void btrfs_mapping_init(struct btrfs_mapping_tree *tree);
401void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree);
f188591e 402int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
8b712842 403 int mirror_num, int async_submit);
8a4b83cc 404int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
97288f2c
CH
405 fmode_t flags, void *holder);
406int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
8a4b83cc
CM
407 struct btrfs_fs_devices **fs_devices_ret);
408int btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
8dabb742
SB
409void btrfs_close_extra_devices(struct btrfs_fs_info *fs_info,
410 struct btrfs_fs_devices *fs_devices, int step);
7ba15b7d
SB
411int btrfs_find_device_missing_or_by_path(struct btrfs_root *root,
412 char *device_path,
413 struct btrfs_device **device);
12bd2fc0
ID
414struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
415 const u64 *devid,
416 const u8 *uuid);
a061fc8d 417int btrfs_rm_device(struct btrfs_root *root, char *device_path);
143bede5 418void btrfs_cleanup_fs_uuids(void);
5d964051 419int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len);
8f18cf13
CM
420int btrfs_grow_device(struct btrfs_trans_handle *trans,
421 struct btrfs_device *device, u64 new_size);
aa1b8cd4 422struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid,
2b82032c 423 u8 *uuid, u8 *fsid);
8f18cf13 424int btrfs_shrink_device(struct btrfs_device *device, u64 new_size);
788f20eb 425int btrfs_init_new_device(struct btrfs_root *root, char *path);
e93c89c1 426int btrfs_init_dev_replace_tgtdev(struct btrfs_root *root, char *device_path,
1c43366d 427 struct btrfs_device *srcdev,
e93c89c1 428 struct btrfs_device **device_out);
c9e9f97b
ID
429int btrfs_balance(struct btrfs_balance_control *bctl,
430 struct btrfs_ioctl_balance_args *bargs);
2b6ba629 431int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info);
68310a5e 432int btrfs_recover_balance(struct btrfs_fs_info *fs_info);
837d5b6e 433int btrfs_pause_balance(struct btrfs_fs_info *fs_info);
a7e99c69 434int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
f7a81ea4 435int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info);
70f80175 436int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info);
2b82032c 437int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset);
6df9a95e
JB
438int find_free_dev_extent(struct btrfs_trans_handle *trans,
439 struct btrfs_device *device, u64 num_bytes,
ba1bf481 440 u64 *start, u64 *max_avail);
442a4f63 441void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
c11d2c23 442int btrfs_get_dev_stats(struct btrfs_root *root,
b27f7c0c 443 struct btrfs_ioctl_get_dev_stats *stats);
cb517eab 444void btrfs_init_devices_late(struct btrfs_fs_info *fs_info);
733f4fbb
SB
445int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info);
446int btrfs_run_dev_stats(struct btrfs_trans_handle *trans,
447 struct btrfs_fs_info *fs_info);
e93c89c1
SB
448void btrfs_rm_dev_replace_srcdev(struct btrfs_fs_info *fs_info,
449 struct btrfs_device *srcdev);
450void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
451 struct btrfs_device *tgtdev);
452void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info,
453 struct btrfs_device *tgtdev);
a8a6dab7 454int btrfs_scratch_superblock(struct btrfs_device *device);
53b381b3
DW
455int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree,
456 u64 logical, u64 len, int mirror_num);
457unsigned long btrfs_full_stripe_len(struct btrfs_root *root,
458 struct btrfs_mapping_tree *map_tree,
459 u64 logical);
6df9a95e
JB
460int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
461 struct btrfs_root *extent_root,
462 u64 chunk_offset, u64 chunk_size);
addc3fa7
MX
463
464static inline int btrfs_dev_stats_dirty(struct btrfs_device *dev)
465{
466 return atomic_read(&dev->dev_stats_ccnt);
467}
468
442a4f63
SB
469static inline void btrfs_dev_stat_inc(struct btrfs_device *dev,
470 int index)
471{
472 atomic_inc(dev->dev_stat_values + index);
addc3fa7
MX
473 smp_mb__before_atomic();
474 atomic_inc(&dev->dev_stats_ccnt);
442a4f63
SB
475}
476
477static inline int btrfs_dev_stat_read(struct btrfs_device *dev,
478 int index)
479{
480 return atomic_read(dev->dev_stat_values + index);
481}
482
483static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev,
484 int index)
485{
486 int ret;
487
488 ret = atomic_xchg(dev->dev_stat_values + index, 0);
addc3fa7
MX
489 smp_mb__before_atomic();
490 atomic_inc(&dev->dev_stats_ccnt);
442a4f63
SB
491 return ret;
492}
493
494static inline void btrfs_dev_stat_set(struct btrfs_device *dev,
495 int index, unsigned long val)
496{
497 atomic_set(dev->dev_stat_values + index, val);
addc3fa7
MX
498 smp_mb__before_atomic();
499 atomic_inc(&dev->dev_stats_ccnt);
442a4f63
SB
500}
501
502static inline void btrfs_dev_stat_reset(struct btrfs_device *dev,
503 int index)
504{
505 btrfs_dev_stat_set(dev, index, 0);
506}
935e5cc9
MX
507
508void btrfs_update_commit_device_size(struct btrfs_fs_info *fs_info);
ce7213c7
MX
509void btrfs_update_commit_device_bytes_used(struct btrfs_root *root,
510 struct btrfs_transaction *transaction);
0b86a832 511#endif
This page took 0.337008 seconds and 5 git commands to generate.