]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - fs/btrfs/volumes.h
Btrfs: use bio_clone_bioset_partial to simplify DIO submit
[mirror_ubuntu-artful-kernel.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
67a2c45e
MX
27extern struct mutex uuid_mutex;
28
ee22184b 29#define BTRFS_STRIPE_LEN SZ_64K
b2117a39 30
f2984462 31struct buffer_head;
ffbd517d
CM
32struct btrfs_pending_bios {
33 struct bio *head;
34 struct bio *tail;
35};
36
7cc8e58d
MX
37/*
38 * Use sequence counter to get consistent device stat data on
39 * 32-bit processors.
40 */
41#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
42#include <linux/seqlock.h>
43#define __BTRFS_NEED_DEVICE_DATA_ORDERED
44#define btrfs_device_data_ordered_init(device) \
45 seqcount_init(&device->data_seqcount)
46#else
47#define btrfs_device_data_ordered_init(device) do { } while (0)
48#endif
49
0b86a832
CM
50struct btrfs_device {
51 struct list_head dev_list;
b3075717 52 struct list_head dev_alloc_list;
2b82032c 53 struct btrfs_fs_devices *fs_devices;
fb456252 54 struct btrfs_fs_info *fs_info;
ffbd517d 55
d5ee37bc
MX
56 struct rcu_string *name;
57
58 u64 generation;
59
60 spinlock_t io_lock ____cacheline_aligned;
61 int running_pending;
ffbd517d
CM
62 /* regular prio bios */
63 struct btrfs_pending_bios pending_bios;
70fd7614 64 /* sync bios */
ffbd517d
CM
65 struct btrfs_pending_bios pending_sync_bios;
66
d5ee37bc
MX
67 struct block_device *bdev;
68
69 /* the mode sent to blkdev_get */
70 fmode_t mode;
71
2b82032c 72 int writeable;
dfe25020 73 int in_fs_metadata;
cd02dca5 74 int missing;
d5e2003c 75 int can_discard;
63a212ab 76 int is_tgtdev_for_dev_replace;
401b41e5 77 int last_flush_error;
b3075717 78
7cc8e58d
MX
79#ifdef __BTRFS_NEED_DEVICE_DATA_ORDERED
80 seqcount_t data_seqcount;
81#endif
82
0b86a832
CM
83 /* the internal btrfs device id */
84 u64 devid;
85
6ba40b61 86 /* size of the device in memory */
0b86a832
CM
87 u64 total_bytes;
88
6ba40b61 89 /* size of the device on disk */
d6397bae
CB
90 u64 disk_total_bytes;
91
0b86a832
CM
92 /* bytes used */
93 u64 bytes_used;
94
95 /* optimal io alignment for this device */
96 u32 io_align;
97
98 /* optimal io width for this device */
99 u32 io_width;
3c45bfc1
DG
100 /* type and info about this device */
101 u64 type;
0b86a832
CM
102
103 /* minimal io size for this device */
104 u32 sector_size;
105
0b86a832 106 /* physical drive uuid (or lvm uuid) */
e17cade2 107 u8 uuid[BTRFS_UUID_SIZE];
8b712842 108
935e5cc9
MX
109 /*
110 * size of the device on the current transaction
111 *
112 * This variant is update when committing the transaction,
113 * and protected by device_list_mutex
114 */
115 u64 commit_total_bytes;
116
ce7213c7
MX
117 /* bytes used on the current transaction */
118 u64 commit_bytes_used;
935e5cc9
MX
119 /*
120 * used to manage the device which is resized
121 *
122 * It is protected by chunk_lock.
123 */
124 struct list_head resized_list;
125
3c45bfc1 126 /* for sending down flush barriers */
3c45bfc1
DG
127 struct bio *flush_bio;
128 struct completion flush_wait;
129
a2de733c 130 /* per-device scrub information */
d9d181c1 131 struct scrub_ctx *scrub_device;
a2de733c 132
d458b054 133 struct btrfs_work work;
1f78160c
XG
134 struct rcu_head rcu;
135 struct work_struct rcu_work;
90519d66
AJ
136
137 /* readahead state */
138 spinlock_t reada_lock;
139 atomic_t reada_in_flight;
140 u64 reada_next;
141 struct reada_zone *reada_curr_zone;
142 struct radix_tree_root reada_zones;
143 struct radix_tree_root reada_extents;
387125fc 144
442a4f63
SB
145 /* disk I/O failure stats. For detailed description refer to
146 * enum btrfs_dev_stat_values in ioctl.h */
733f4fbb 147 int dev_stats_valid;
addc3fa7
MX
148
149 /* Counter to record the change of device stats */
150 atomic_t dev_stats_ccnt;
442a4f63 151 atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX];
0b86a832
CM
152};
153
7cc8e58d
MX
154/*
155 * If we read those variants at the context of their own lock, we needn't
156 * use the following helpers, reading them directly is safe.
157 */
158#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
159#define BTRFS_DEVICE_GETSET_FUNCS(name) \
160static inline u64 \
161btrfs_device_get_##name(const struct btrfs_device *dev) \
162{ \
163 u64 size; \
164 unsigned int seq; \
165 \
166 do { \
167 seq = read_seqcount_begin(&dev->data_seqcount); \
168 size = dev->name; \
169 } while (read_seqcount_retry(&dev->data_seqcount, seq)); \
170 return size; \
171} \
172 \
173static inline void \
174btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
175{ \
176 preempt_disable(); \
177 write_seqcount_begin(&dev->data_seqcount); \
178 dev->name = size; \
179 write_seqcount_end(&dev->data_seqcount); \
180 preempt_enable(); \
181}
182#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
183#define BTRFS_DEVICE_GETSET_FUNCS(name) \
184static inline u64 \
185btrfs_device_get_##name(const struct btrfs_device *dev) \
186{ \
187 u64 size; \
188 \
189 preempt_disable(); \
190 size = dev->name; \
191 preempt_enable(); \
192 return size; \
193} \
194 \
195static inline void \
196btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
197{ \
198 preempt_disable(); \
199 dev->name = size; \
200 preempt_enable(); \
201}
202#else
203#define BTRFS_DEVICE_GETSET_FUNCS(name) \
204static inline u64 \
205btrfs_device_get_##name(const struct btrfs_device *dev) \
206{ \
207 return dev->name; \
208} \
209 \
210static inline void \
211btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
212{ \
213 dev->name = size; \
214}
215#endif
216
217BTRFS_DEVICE_GETSET_FUNCS(total_bytes);
218BTRFS_DEVICE_GETSET_FUNCS(disk_total_bytes);
219BTRFS_DEVICE_GETSET_FUNCS(bytes_used);
220
8a4b83cc
CM
221struct btrfs_fs_devices {
222 u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
223
8a4b83cc 224 u64 num_devices;
a0af469b 225 u64 open_devices;
2b82032c 226 u64 rw_devices;
cd02dca5 227 u64 missing_devices;
2b82032c 228 u64 total_rw_bytes;
02db0844 229 u64 total_devices;
8a4b83cc 230 struct block_device *latest_bdev;
e5e9a520
CM
231
232 /* all of the devices in the FS, protected by a mutex
233 * so we can safely walk it to write out the supers without
9b011adf
WS
234 * worrying about add/remove by the multi-device code.
235 * Scrubbing super can kick off supers writing by holding
236 * this mutex lock.
e5e9a520
CM
237 */
238 struct mutex device_list_mutex;
8a4b83cc 239 struct list_head devices;
b3075717 240
935e5cc9 241 struct list_head resized_devices;
b3075717
CM
242 /* devices not currently being allocated */
243 struct list_head alloc_list;
8a4b83cc 244 struct list_head list;
2b82032c
YZ
245
246 struct btrfs_fs_devices *seed;
247 int seeding;
2b82032c
YZ
248
249 int opened;
c289811c
CM
250
251 /* set when we find or add a device that doesn't have the
252 * nonrot flag set
253 */
254 int rotating;
2e7910d6 255
5a13f430 256 struct btrfs_fs_info *fs_info;
2e7910d6 257 /* sysfs kobjects */
c1b7e474 258 struct kobject fsid_kobj;
2e7910d6
AJ
259 struct kobject *device_dir_kobj;
260 struct completion kobj_unregister;
8a4b83cc
CM
261};
262
facc8a22
MX
263#define BTRFS_BIO_INLINE_CSUM_SIZE 64
264
9be3395b
CM
265/*
266 * we need the mirror number and stripe index to be passed around
267 * the call chain while we are processing end_io (especially errors).
268 * Really, what we need is a btrfs_bio structure that has this info
269 * and is properly sized with its stripe array, but we're not there
270 * quite yet. We have our own btrfs bioset, and all of the bios
271 * we allocate are actually btrfs_io_bios. We'll cram as much of
272 * struct btrfs_bio as we can into this over time.
273 */
facc8a22 274typedef void (btrfs_io_bio_end_io_t) (struct btrfs_io_bio *bio, int err);
9be3395b 275struct btrfs_io_bio {
c1dc0896
MX
276 unsigned int mirror_num;
277 unsigned int stripe_index;
278 u64 logical;
facc8a22
MX
279 u8 *csum;
280 u8 csum_inline[BTRFS_BIO_INLINE_CSUM_SIZE];
281 u8 *csum_allocated;
282 btrfs_io_bio_end_io_t *end_io;
9be3395b
CM
283 struct bio bio;
284};
285
286static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio)
287{
288 return container_of(bio, struct btrfs_io_bio, bio);
289}
290
cea9e445
CM
291struct btrfs_bio_stripe {
292 struct btrfs_device *dev;
293 u64 physical;
fce3bb9a 294 u64 length; /* only used for discard mappings */
cea9e445
CM
295};
296
a1d3c478
JS
297struct btrfs_bio;
298typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
299
300struct btrfs_bio {
140475ae 301 refcount_t refs;
cea9e445 302 atomic_t stripes_pending;
c404e0dc 303 struct btrfs_fs_info *fs_info;
10f11900 304 u64 map_type; /* get from map_lookup->type */
cea9e445 305 bio_end_io_t *end_io;
7d2b4daa 306 struct bio *orig_bio;
c55f1396 307 unsigned long flags;
cea9e445 308 void *private;
a236aed1
CM
309 atomic_t error;
310 int max_errors;
cea9e445 311 int num_stripes;
a1d3c478 312 int mirror_num;
2c8cdd6e
MX
313 int num_tgtdevs;
314 int *tgtdev_map;
8e5cfb55
ZL
315 /*
316 * logical block numbers for the start of each stripe
317 * The last one or two are p/q. These are sorted,
318 * so raid_map[0] is the start of our full stripe
319 */
320 u64 *raid_map;
cea9e445
CM
321 struct btrfs_bio_stripe stripes[];
322};
323
b2117a39
MX
324struct btrfs_device_info {
325 struct btrfs_device *dev;
326 u64 dev_offset;
327 u64 max_avail;
73c5de00 328 u64 total_avail;
b2117a39
MX
329};
330
31e50229
LB
331struct btrfs_raid_attr {
332 int sub_stripes; /* sub_stripes info for map */
333 int dev_stripes; /* stripes per dev */
334 int devs_max; /* max devs to use */
335 int devs_min; /* min devs needed */
8789f4fe 336 int tolerated_failures; /* max tolerated fail devs */
31e50229
LB
337 int devs_increment; /* ndevs has to be a multiple of this */
338 int ncopies; /* how many copies to data has */
339};
340
af902047 341extern const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES];
621292ba 342extern const int btrfs_raid_mindev_error[BTRFS_NR_RAID_TYPES];
af902047
ZL
343extern const u64 btrfs_raid_group[BTRFS_NR_RAID_TYPES];
344
1abe9b8a 345struct map_lookup {
346 u64 type;
347 int io_align;
348 int io_width;
3d8da678 349 u64 stripe_len;
1abe9b8a 350 int sector_size;
351 int num_stripes;
352 int sub_stripes;
353 struct btrfs_bio_stripe stripes[];
354};
355
a2de733c
AJ
356#define map_lookup_size(n) (sizeof(struct map_lookup) + \
357 (sizeof(struct btrfs_bio_stripe) * (n)))
358
c9e9f97b 359struct btrfs_balance_args;
19a39dce 360struct btrfs_balance_progress;
c9e9f97b
ID
361struct btrfs_balance_control {
362 struct btrfs_fs_info *fs_info;
363
364 struct btrfs_balance_args data;
365 struct btrfs_balance_args meta;
366 struct btrfs_balance_args sys;
367
368 u64 flags;
19a39dce
ID
369
370 struct btrfs_balance_progress stat;
c9e9f97b
ID
371};
372
cf8cddd3
CH
373enum btrfs_map_op {
374 BTRFS_MAP_READ,
375 BTRFS_MAP_WRITE,
376 BTRFS_MAP_DISCARD,
377 BTRFS_MAP_GET_READ_MIRRORS,
378};
379
380static inline enum btrfs_map_op btrfs_op(struct bio *bio)
381{
382 switch (bio_op(bio)) {
383 case REQ_OP_DISCARD:
384 return BTRFS_MAP_DISCARD;
385 case REQ_OP_WRITE:
386 return BTRFS_MAP_WRITE;
387 default:
388 WARN_ON_ONCE(1);
389 case REQ_OP_READ:
390 return BTRFS_MAP_READ;
391 }
392}
393
6d07bcec
MX
394int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
395 u64 end, u64 *length);
6e9606d2
ZL
396void btrfs_get_bbio(struct btrfs_bio *bbio);
397void btrfs_put_bbio(struct btrfs_bio *bbio);
cf8cddd3 398int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
cea9e445 399 u64 logical, u64 *length,
a1d3c478 400 struct btrfs_bio **bbio_ret, int mirror_num);
cf8cddd3 401int btrfs_map_sblock(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
af8e2d1d 402 u64 logical, u64 *length,
825ad4c9 403 struct btrfs_bio **bbio_ret);
ab8d0fc4 404int btrfs_rmap_block(struct btrfs_fs_info *fs_info,
a512bbf8
YZ
405 u64 chunk_start, u64 physical, u64 devid,
406 u64 **logical, int *naddrs, int *stripe_len);
6bccf3ab 407int btrfs_read_sys_array(struct btrfs_fs_info *fs_info);
5b4aacef 408int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info);
0b86a832 409int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
2ff7e61e 410 struct btrfs_fs_info *fs_info, u64 type);
0b86a832
CM
411void btrfs_mapping_init(struct btrfs_mapping_tree *tree);
412void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree);
2ff7e61e 413int btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
8b712842 414 int mirror_num, int async_submit);
8a4b83cc 415int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
97288f2c
CH
416 fmode_t flags, void *holder);
417int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
8a4b83cc
CM
418 struct btrfs_fs_devices **fs_devices_ret);
419int btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
9eaed21e 420void btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices, int step);
88acff64
AJ
421void btrfs_assign_next_active_device(struct btrfs_fs_info *fs_info,
422 struct btrfs_device *device, struct btrfs_device *this_dev);
2ff7e61e 423int btrfs_find_device_missing_or_by_path(struct btrfs_fs_info *fs_info,
da353f6b 424 const char *device_path,
7ba15b7d 425 struct btrfs_device **device);
2ff7e61e 426int btrfs_find_device_by_devspec(struct btrfs_fs_info *fs_info, u64 devid,
da353f6b 427 const char *devpath,
24e0474b 428 struct btrfs_device **device);
12bd2fc0
ID
429struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
430 const u64 *devid,
431 const u8 *uuid);
2ff7e61e 432int btrfs_rm_device(struct btrfs_fs_info *fs_info,
da353f6b 433 const char *device_path, u64 devid);
143bede5 434void btrfs_cleanup_fs_uuids(void);
5d964051 435int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len);
8f18cf13
CM
436int btrfs_grow_device(struct btrfs_trans_handle *trans,
437 struct btrfs_device *device, u64 new_size);
aa1b8cd4 438struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid,
2b82032c 439 u8 *uuid, u8 *fsid);
8f18cf13 440int btrfs_shrink_device(struct btrfs_device *device, u64 new_size);
da353f6b 441int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *path);
2ff7e61e 442int btrfs_init_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
da353f6b 443 const char *device_path,
1c43366d 444 struct btrfs_device *srcdev,
e93c89c1 445 struct btrfs_device **device_out);
c9e9f97b
ID
446int btrfs_balance(struct btrfs_balance_control *bctl,
447 struct btrfs_ioctl_balance_args *bargs);
2b6ba629 448int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info);
68310a5e 449int btrfs_recover_balance(struct btrfs_fs_info *fs_info);
837d5b6e 450int btrfs_pause_balance(struct btrfs_fs_info *fs_info);
a7e99c69 451int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
f7a81ea4 452int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info);
70f80175 453int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info);
2ff7e61e 454int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset);
499f377f
JM
455int find_free_dev_extent_start(struct btrfs_transaction *transaction,
456 struct btrfs_device *device, u64 num_bytes,
457 u64 search_start, u64 *start, u64 *max_avail);
6df9a95e
JB
458int find_free_dev_extent(struct btrfs_trans_handle *trans,
459 struct btrfs_device *device, u64 num_bytes,
ba1bf481 460 u64 *start, u64 *max_avail);
442a4f63 461void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
2ff7e61e 462int btrfs_get_dev_stats(struct btrfs_fs_info *fs_info,
b27f7c0c 463 struct btrfs_ioctl_get_dev_stats *stats);
cb517eab 464void btrfs_init_devices_late(struct btrfs_fs_info *fs_info);
733f4fbb
SB
465int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info);
466int btrfs_run_dev_stats(struct btrfs_trans_handle *trans,
467 struct btrfs_fs_info *fs_info);
084b6e7c
QW
468void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_fs_info *fs_info,
469 struct btrfs_device *srcdev);
470void btrfs_rm_dev_replace_free_srcdev(struct btrfs_fs_info *fs_info,
471 struct btrfs_device *srcdev);
e93c89c1
SB
472void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
473 struct btrfs_device *tgtdev);
474void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info,
475 struct btrfs_device *tgtdev);
da353f6b 476void btrfs_scratch_superblocks(struct block_device *bdev, const char *device_path);
592d92ee 477int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info,
53b381b3 478 u64 logical, u64 len, int mirror_num);
2ff7e61e 479unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info,
53b381b3
DW
480 struct btrfs_mapping_tree *map_tree,
481 u64 logical);
6df9a95e 482int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
6bccf3ab 483 struct btrfs_fs_info *fs_info,
6df9a95e 484 u64 chunk_offset, u64 chunk_size);
47ab2a6c 485int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
5b4aacef 486 struct btrfs_fs_info *fs_info, u64 chunk_offset);
addc3fa7
MX
487
488static inline int btrfs_dev_stats_dirty(struct btrfs_device *dev)
489{
490 return atomic_read(&dev->dev_stats_ccnt);
491}
492
442a4f63
SB
493static inline void btrfs_dev_stat_inc(struct btrfs_device *dev,
494 int index)
495{
496 atomic_inc(dev->dev_stat_values + index);
addc3fa7
MX
497 smp_mb__before_atomic();
498 atomic_inc(&dev->dev_stats_ccnt);
442a4f63
SB
499}
500
501static inline int btrfs_dev_stat_read(struct btrfs_device *dev,
502 int index)
503{
504 return atomic_read(dev->dev_stat_values + index);
505}
506
507static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev,
508 int index)
509{
510 int ret;
511
512 ret = atomic_xchg(dev->dev_stat_values + index, 0);
addc3fa7
MX
513 smp_mb__before_atomic();
514 atomic_inc(&dev->dev_stats_ccnt);
442a4f63
SB
515 return ret;
516}
517
518static inline void btrfs_dev_stat_set(struct btrfs_device *dev,
519 int index, unsigned long val)
520{
521 atomic_set(dev->dev_stat_values + index, val);
addc3fa7
MX
522 smp_mb__before_atomic();
523 atomic_inc(&dev->dev_stats_ccnt);
442a4f63
SB
524}
525
526static inline void btrfs_dev_stat_reset(struct btrfs_device *dev,
527 int index)
528{
529 btrfs_dev_stat_set(dev, index, 0);
530}
935e5cc9
MX
531
532void btrfs_update_commit_device_size(struct btrfs_fs_info *fs_info);
2ff7e61e 533void btrfs_update_commit_device_bytes_used(struct btrfs_fs_info *fs_info,
ce7213c7 534 struct btrfs_transaction *transaction);
04216820 535
c73eccf7 536struct list_head *btrfs_get_fs_uuids(void);
5a13f430
AJ
537void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info);
538void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info);
04216820 539
0b86a832 540#endif