Linux Audio

Check our new training course

Loading...
v6.8
  1/* SPDX-License-Identifier: GPL-2.0 */
  2/*
  3 * Copyright (C) 2007 Oracle.  All rights reserved.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  4 */
  5
  6#ifndef BTRFS_VOLUMES_H
  7#define BTRFS_VOLUMES_H
  8
 
  9#include <linux/sort.h>
 10#include <linux/btrfs.h>
 11#include "async-thread.h"
 12#include "messages.h"
 13#include "tree-checker.h"
 14#include "rcu-string.h"
 15
 16#define BTRFS_MAX_DATA_CHUNK_SIZE	(10ULL * SZ_1G)
 17
 18extern struct mutex uuid_mutex;
 19
 20#define BTRFS_STRIPE_LEN		SZ_64K
 21#define BTRFS_STRIPE_LEN_SHIFT		(16)
 22#define BTRFS_STRIPE_LEN_MASK		(BTRFS_STRIPE_LEN - 1)
 23
 24static_assert(const_ilog2(BTRFS_STRIPE_LEN) == BTRFS_STRIPE_LEN_SHIFT);
 25
 26/* Used by sanity check for btrfs_raid_types. */
 27#define const_ffs(n) (__builtin_ctzll(n) + 1)
 28
 29/*
 30 * The conversion from BTRFS_BLOCK_GROUP_* bits to btrfs_raid_type requires
 31 * RAID0 always to be the lowest profile bit.
 32 * Although it's part of on-disk format and should never change, do extra
 33 * compile-time sanity checks.
 34 */
 35static_assert(const_ffs(BTRFS_BLOCK_GROUP_RAID0) <
 36	      const_ffs(BTRFS_BLOCK_GROUP_PROFILE_MASK & ~BTRFS_BLOCK_GROUP_RAID0));
 37static_assert(const_ilog2(BTRFS_BLOCK_GROUP_RAID0) >
 38	      ilog2(BTRFS_BLOCK_GROUP_TYPE_MASK));
 39
 40/* ilog2() can handle both constants and variables */
 41#define BTRFS_BG_FLAG_TO_INDEX(profile)					\
 42	ilog2((profile) >> (ilog2(BTRFS_BLOCK_GROUP_RAID0) - 1))
 43
 44enum btrfs_raid_types {
 45	/* SINGLE is the special one as it doesn't have on-disk bit. */
 46	BTRFS_RAID_SINGLE  = 0,
 47
 48	BTRFS_RAID_RAID0   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID0),
 49	BTRFS_RAID_RAID1   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID1),
 50	BTRFS_RAID_DUP	   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_DUP),
 51	BTRFS_RAID_RAID10  = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID10),
 52	BTRFS_RAID_RAID5   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID5),
 53	BTRFS_RAID_RAID6   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID6),
 54	BTRFS_RAID_RAID1C3 = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID1C3),
 55	BTRFS_RAID_RAID1C4 = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID1C4),
 56
 57	BTRFS_NR_RAID_TYPES
 58};
 59
 60/*
 61 * Use sequence counter to get consistent device stat data on
 62 * 32-bit processors.
 63 */
 64#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 65#include <linux/seqlock.h>
 66#define __BTRFS_NEED_DEVICE_DATA_ORDERED
 67#define btrfs_device_data_ordered_init(device)	\
 68	seqcount_init(&device->data_seqcount)
 69#else
 70#define btrfs_device_data_ordered_init(device) do { } while (0)
 71#endif
 72
 73#define BTRFS_DEV_STATE_WRITEABLE	(0)
 74#define BTRFS_DEV_STATE_IN_FS_METADATA	(1)
 75#define BTRFS_DEV_STATE_MISSING		(2)
 76#define BTRFS_DEV_STATE_REPLACE_TGT	(3)
 77#define BTRFS_DEV_STATE_FLUSH_SENT	(4)
 78#define BTRFS_DEV_STATE_NO_READA	(5)
 79
 80struct btrfs_zoned_device_info;
 81
 82struct btrfs_device {
 83	struct list_head dev_list; /* device_list_mutex */
 84	struct list_head dev_alloc_list; /* chunk mutex */
 85	struct list_head post_commit_list; /* chunk mutex */
 86	struct btrfs_fs_devices *fs_devices;
 87	struct btrfs_fs_info *fs_info;
 88
 89	struct rcu_string __rcu *name;
 
 
 
 90
 
 91	u64 generation;
 92
 93	struct bdev_handle *bdev_handle;
 
 
 
 
 
 
 94	struct block_device *bdev;
 95
 96	struct btrfs_zoned_device_info *zone_info;
 
 97
 98	/*
 99	 * Device's major-minor number. Must be set even if the device is not
100	 * opened (bdev == NULL), unless the device is missing.
101	 */
102	dev_t devt;
103	unsigned long dev_state;
104	blk_status_t last_flush_error;
105
106#ifdef __BTRFS_NEED_DEVICE_DATA_ORDERED
107	seqcount_t data_seqcount;
108#endif
109
110	/* the internal btrfs device id */
111	u64 devid;
112
113	/* size of the device in memory */
114	u64 total_bytes;
115
116	/* size of the device on disk */
117	u64 disk_total_bytes;
118
119	/* bytes used */
120	u64 bytes_used;
121
122	/* optimal io alignment for this device */
123	u32 io_align;
124
125	/* optimal io width for this device */
126	u32 io_width;
127	/* type and info about this device */
128	u64 type;
129
130	/* minimal io size for this device */
131	u32 sector_size;
132
 
 
 
133	/* physical drive uuid (or lvm uuid) */
134	u8 uuid[BTRFS_UUID_SIZE];
135
136	/*
137	 * size of the device on the current transaction
138	 *
139	 * This variant is update when committing the transaction,
140	 * and protected by chunk mutex
141	 */
142	u64 commit_total_bytes;
143
144	/* bytes used on the current transaction */
145	u64 commit_bytes_used;
 
 
 
 
 
 
 
 
 
146
147	/* Bio used for flushing device barriers */
148	struct bio flush_bio;
149	struct completion flush_wait;
150
151	/* per-device scrub information */
152	struct scrub_ctx *scrub_ctx;
153
154	/* disk I/O failure stats. For detailed description refer to
155	 * enum btrfs_dev_stat_values in ioctl.h */
156	int dev_stats_valid;
157
158	/* Counter to record the change of device stats */
159	atomic_t dev_stats_ccnt;
160	atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX];
161
162	struct extent_io_tree alloc_state;
163
164	struct completion kobj_unregister;
165	/* For sysfs/FSID/devinfo/devid/ */
166	struct kobject devid_kobj;
167
168	/* Bandwidth limit for scrub, in bytes */
169	u64 scrub_speed_max;
170};
171
172/*
173 * Block group or device which contains an active swapfile. Used for preventing
174 * unsafe operations while a swapfile is active.
175 *
176 * These are sorted on (ptr, inode) (note that a block group or device can
177 * contain more than one swapfile). We compare the pointer values because we
178 * don't actually care what the object is, we just need a quick check whether
179 * the object exists in the rbtree.
180 */
181struct btrfs_swapfile_pin {
182	struct rb_node node;
183	void *ptr;
184	struct inode *inode;
185	/*
186	 * If true, ptr points to a struct btrfs_block_group. Otherwise, ptr
187	 * points to a struct btrfs_device.
188	 */
189	bool is_block_group;
190	/*
191	 * Only used when 'is_block_group' is true and it is the number of
192	 * extents used by a swapfile for this block group ('ptr' field).
193	 */
194	int bg_extent_count;
195};
196
197/*
198 * If we read those variants at the context of their own lock, we needn't
199 * use the following helpers, reading them directly is safe.
200 */
201#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
202#define BTRFS_DEVICE_GETSET_FUNCS(name)					\
203static inline u64							\
204btrfs_device_get_##name(const struct btrfs_device *dev)			\
205{									\
206	u64 size;							\
207	unsigned int seq;						\
208									\
209	do {								\
210		seq = read_seqcount_begin(&dev->data_seqcount);		\
211		size = dev->name;					\
212	} while (read_seqcount_retry(&dev->data_seqcount, seq));	\
213	return size;							\
214}									\
215									\
216static inline void							\
217btrfs_device_set_##name(struct btrfs_device *dev, u64 size)		\
218{									\
219	preempt_disable();						\
220	write_seqcount_begin(&dev->data_seqcount);			\
221	dev->name = size;						\
222	write_seqcount_end(&dev->data_seqcount);			\
223	preempt_enable();						\
224}
225#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
226#define BTRFS_DEVICE_GETSET_FUNCS(name)					\
227static inline u64							\
228btrfs_device_get_##name(const struct btrfs_device *dev)			\
229{									\
230	u64 size;							\
231									\
232	preempt_disable();						\
233	size = dev->name;						\
234	preempt_enable();						\
235	return size;							\
236}									\
237									\
238static inline void							\
239btrfs_device_set_##name(struct btrfs_device *dev, u64 size)		\
240{									\
241	preempt_disable();						\
242	dev->name = size;						\
243	preempt_enable();						\
244}
245#else
246#define BTRFS_DEVICE_GETSET_FUNCS(name)					\
247static inline u64							\
248btrfs_device_get_##name(const struct btrfs_device *dev)			\
249{									\
250	return dev->name;						\
251}									\
252									\
253static inline void							\
254btrfs_device_set_##name(struct btrfs_device *dev, u64 size)		\
255{									\
256	dev->name = size;						\
257}
258#endif
259
260BTRFS_DEVICE_GETSET_FUNCS(total_bytes);
261BTRFS_DEVICE_GETSET_FUNCS(disk_total_bytes);
262BTRFS_DEVICE_GETSET_FUNCS(bytes_used);
263
264enum btrfs_chunk_allocation_policy {
265	BTRFS_CHUNK_ALLOC_REGULAR,
266	BTRFS_CHUNK_ALLOC_ZONED,
267};
268
269/*
270 * Read policies for mirrored block group profiles, read picks the stripe based
271 * on these policies.
272 */
273enum btrfs_read_policy {
274	/* Use process PID to choose the stripe */
275	BTRFS_READ_POLICY_PID,
276	BTRFS_NR_READ_POLICY,
277};
278
279struct btrfs_fs_devices {
280	u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
281
282	/*
283	 * UUID written into the btree blocks:
284	 *
285	 * - If metadata_uuid != fsid then super block must have
286	 *   BTRFS_FEATURE_INCOMPAT_METADATA_UUID flag set.
287	 *
288	 * - Following shall be true at all times:
289	 *   - metadata_uuid == btrfs_header::fsid
290	 *   - metadata_uuid == btrfs_dev_item::fsid
291	 *
292	 * - Relations between fsid and metadata_uuid in sb and fs_devices:
293	 *   - Normal:
294	 *       fs_devices->fsid == fs_devices->metadata_uuid == sb->fsid
295	 *       sb->metadata_uuid == 0
296	 *
297	 *   - When the BTRFS_FEATURE_INCOMPAT_METADATA_UUID flag is set:
298	 *       fs_devices->fsid == sb->fsid
299	 *       fs_devices->metadata_uuid == sb->metadata_uuid
300	 *
301	 *   - When in-memory fs_devices->temp_fsid is true
302	 *	 fs_devices->fsid = random
303	 *	 fs_devices->metadata_uuid == sb->fsid
304	 */
305	u8 metadata_uuid[BTRFS_FSID_SIZE];
306
307	struct list_head fs_list;
308
309	/*
310	 * Number of devices under this fsid including missing and
311	 * replace-target device and excludes seed devices.
312	 */
313	u64 num_devices;
314
315	/*
316	 * The number of devices that successfully opened, including
317	 * replace-target, excludes seed devices.
318	 */
319	u64 open_devices;
320
321	/* The number of devices that are under the chunk allocation list. */
322	u64 rw_devices;
323
324	/* Count of missing devices under this fsid excluding seed device. */
325	u64 missing_devices;
326	u64 total_rw_bytes;
 
 
327
328	/*
329	 * Count of devices from btrfs_super_block::num_devices for this fsid,
330	 * which includes the seed device, excludes the transient replace-target
331	 * device.
332	 */
333	u64 total_devices;
334
335	/* Highest generation number of seen devices */
336	u64 latest_generation;
337
338	/*
339	 * The mount device or a device with highest generation after removal
340	 * or replace.
341	 */
342	struct btrfs_device *latest_dev;
343
344	/*
345	 * All of the devices in the filesystem, protected by a mutex so we can
346	 * safely walk it to write out the super blocks without worrying about
347	 * adding/removing by the multi-device code. Scrubbing super block can
348	 * kick off supers writing by holding this mutex lock.
349	 */
350	struct mutex device_list_mutex;
351
352	/* List of all devices, protected by device_list_mutex */
353	struct list_head devices;
354
355	/* Devices which can satisfy space allocation. Protected by * chunk_mutex. */
356	struct list_head alloc_list;
 
357
358	struct list_head seed_list;
 
359
360	/* Count fs-devices opened. */
361	int opened;
362
363	/* Set when we find or add a device that doesn't have the nonrot flag set. */
364	bool rotating;
365	/* Devices support TRIM/discard commands. */
366	bool discardable;
367	/* The filesystem is a seed filesystem. */
368	bool seeding;
369	/* The mount needs to use a randomly generated fsid. */
370	bool temp_fsid;
371
372	struct btrfs_fs_info *fs_info;
373	/* sysfs kobjects */
374	struct kobject fsid_kobj;
375	struct kobject *devices_kobj;
376	struct kobject *devinfo_kobj;
377	struct completion kobj_unregister;
378
379	enum btrfs_chunk_allocation_policy chunk_alloc_policy;
380
381	/* Policy used to read the mirrored stripes. */
382	enum btrfs_read_policy read_policy;
383};
384
385#define BTRFS_MAX_DEVS(info) ((BTRFS_MAX_ITEM_SIZE(info)	\
386			- sizeof(struct btrfs_chunk))		\
387			/ sizeof(struct btrfs_stripe) + 1)
388
389#define BTRFS_MAX_DEVS_SYS_CHUNK ((BTRFS_SYSTEM_CHUNK_ARRAY_SIZE	\
390				- 2 * sizeof(struct btrfs_disk_key)	\
391				- 2 * sizeof(struct btrfs_chunk))	\
392				/ sizeof(struct btrfs_stripe) + 1)
393
394struct btrfs_io_stripe {
395	struct btrfs_device *dev;
396	/* Block mapping. */
397	u64 physical;
398	u64 length;
399	bool is_scrub;
400	/* For the endio handler. */
401	struct btrfs_io_context *bioc;
402};
403
404struct btrfs_discard_stripe {
405	struct btrfs_device *dev;
406	u64 physical;
407	u64 length;
408};
409
410/*
411 * Context for IO subsmission for device stripe.
412 *
413 * - Track the unfinished mirrors for mirror based profiles
414 *   Mirror based profiles are SINGLE/DUP/RAID1/RAID10.
415 *
416 * - Contain the logical -> physical mapping info
417 *   Used by submit_stripe_bio() for mapping logical bio
418 *   into physical device address.
419 *
420 * - Contain device replace info
421 *   Used by handle_ops_on_dev_replace() to copy logical bios
422 *   into the new device.
423 *
424 * - Contain RAID56 full stripe logical bytenrs
425 */
426struct btrfs_io_context {
427	refcount_t refs;
428	struct btrfs_fs_info *fs_info;
429	/* Taken from struct btrfs_chunk_map::type. */
430	u64 map_type;
431	struct bio *orig_bio;
 
432	atomic_t error;
433	u16 max_errors;
434
435	u64 logical;
436	u64 size;
437	/* Raid stripe tree ordered entry. */
438	struct list_head rst_ordered_entry;
439
440	/*
441	 * The total number of stripes, including the extra duplicated
442	 * stripe for replace.
443	 */
444	u16 num_stripes;
445
446	/*
447	 * The mirror_num of this bioc.
448	 *
449	 * This is for reads which use 0 as mirror_num, thus we should return a
450	 * valid mirror_num (>0) for the reader.
451	 */
452	u16 mirror_num;
453
454	/*
455	 * The following two members are for dev-replace case only.
456	 *
457	 * @replace_nr_stripes:	Number of duplicated stripes which need to be
458	 *			written to replace target.
459	 *			Should be <= 2 (2 for DUP, otherwise <= 1).
460	 * @replace_stripe_src:	The array indicates where the duplicated stripes
461	 *			are from.
462	 *
463	 * The @replace_stripe_src[] array is mostly for RAID56 cases.
464	 * As non-RAID56 stripes share the same contents of the mapped range,
465	 * thus no need to bother where the duplicated ones are from.
466	 *
467	 * But for RAID56 case, all stripes contain different contents, thus
468	 * we need a way to know the mapping.
469	 *
470	 * There is an example for the two members, using a RAID5 write:
471	 *
472	 *   num_stripes:	4 (3 + 1 duplicated write)
473	 *   stripes[0]:	dev = devid 1, physical = X
474	 *   stripes[1]:	dev = devid 2, physical = Y
475	 *   stripes[2]:	dev = devid 3, physical = Z
476	 *   stripes[3]:	dev = devid 0, physical = Y
477	 *
478	 * replace_nr_stripes = 1
479	 * replace_stripe_src = 1	<- Means stripes[1] is involved in replace.
480	 *				   The duplicated stripe index would be
481	 *				   (@num_stripes - 1).
482	 *
483	 * Note, that we can still have cases replace_nr_stripes = 2 for DUP.
484	 * In that case, all stripes share the same content, thus we don't
485	 * need to bother @replace_stripe_src value at all.
486	 */
487	u16 replace_nr_stripes;
488	s16 replace_stripe_src;
489	/*
490	 * Logical bytenr of the full stripe start, only for RAID56 cases.
491	 *
492	 * When this value is set to other than (u64)-1, the stripes[] should
493	 * follow this pattern:
494	 *
495	 * (real_stripes = num_stripes - replace_nr_stripes)
496	 * (data_stripes = (is_raid6) ? (real_stripes - 2) : (real_stripes - 1))
497	 *
498	 * stripes[0]:			The first data stripe
499	 * stripes[1]:			The second data stripe
500	 * ...
501	 * stripes[data_stripes - 1]:	The last data stripe
502	 * stripes[data_stripes]:	The P stripe
503	 * stripes[data_stripes + 1]:	The Q stripe (only for RAID6).
504	 */
505	u64 full_stripe_logical;
506	struct btrfs_io_stripe stripes[];
507};
508
509struct btrfs_device_info {
510	struct btrfs_device *dev;
511	u64 dev_offset;
512	u64 max_avail;
513	u64 total_avail;
514};
515
516struct btrfs_raid_attr {
517	u8 sub_stripes;		/* sub_stripes info for map */
518	u8 dev_stripes;		/* stripes per dev */
519	u8 devs_max;		/* max devs to use */
520	u8 devs_min;		/* min devs needed */
521	u8 tolerated_failures;	/* max tolerated fail devs */
522	u8 devs_increment;	/* ndevs has to be a multiple of this */
523	u8 ncopies;		/* how many copies to data has */
524	u8 nparity;		/* number of stripes worth of bytes to store
525				 * parity information */
526	u8 mindev_error;	/* error code if min devs requisite is unmet */
527	const char raid_name[8]; /* name of the raid */
528	u64 bg_flag;		/* block group flag of the raid */
529};
530
531extern const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES];
532
533struct btrfs_chunk_map {
534	struct rb_node rb_node;
535	/* For mount time dev extent verification. */
536	int verified_stripes;
537	refcount_t refs;
538	u64 start;
539	u64 chunk_len;
540	u64 stripe_size;
541	u64 type;
542	int io_align;
543	int io_width;
 
 
544	int num_stripes;
545	int sub_stripes;
546	struct btrfs_io_stripe stripes[];
547};
548
549#define btrfs_chunk_map_size(n) (sizeof(struct btrfs_chunk_map) + \
550				 (sizeof(struct btrfs_io_stripe) * (n)))
551
552static inline void btrfs_free_chunk_map(struct btrfs_chunk_map *map)
553{
554	if (map && refcount_dec_and_test(&map->refs)) {
555		ASSERT(RB_EMPTY_NODE(&map->rb_node));
556		kfree(map);
557	}
558}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
559
560struct btrfs_balance_args;
561struct btrfs_balance_progress;
562struct btrfs_balance_control {
 
 
563	struct btrfs_balance_args data;
564	struct btrfs_balance_args meta;
565	struct btrfs_balance_args sys;
566
567	u64 flags;
568
569	struct btrfs_balance_progress stat;
570};
571
572/*
573 * Search for a given device by the set parameters
574 */
575struct btrfs_dev_lookup_args {
576	u64 devid;
577	u8 *uuid;
578	u8 *fsid;
579	bool missing;
580};
581
582/* We have to initialize to -1 because BTRFS_DEV_REPLACE_DEVID is 0 */
583#define BTRFS_DEV_LOOKUP_ARGS_INIT { .devid = (u64)-1 }
584
585#define BTRFS_DEV_LOOKUP_ARGS(name) \
586	struct btrfs_dev_lookup_args name = BTRFS_DEV_LOOKUP_ARGS_INIT
587
588enum btrfs_map_op {
589	BTRFS_MAP_READ,
590	BTRFS_MAP_WRITE,
591	BTRFS_MAP_GET_READ_MIRRORS,
592};
593
594static inline enum btrfs_map_op btrfs_op(struct bio *bio)
595{
596	switch (bio_op(bio)) {
597	case REQ_OP_WRITE:
598	case REQ_OP_ZONE_APPEND:
599		return BTRFS_MAP_WRITE;
600	default:
601		WARN_ON_ONCE(1);
602		fallthrough;
603	case REQ_OP_READ:
604		return BTRFS_MAP_READ;
605	}
606}
607
608static inline unsigned long btrfs_chunk_item_size(int num_stripes)
609{
610	ASSERT(num_stripes);
611	return sizeof(struct btrfs_chunk) +
612		sizeof(struct btrfs_stripe) * (num_stripes - 1);
613}
614
615/*
616 * Do the type safe conversion from stripe_nr to offset inside the chunk.
617 *
618 * @stripe_nr is u32, with left shift it can overflow u32 for chunks larger
619 * than 4G.  This does the proper type cast to avoid overflow.
620 */
621static inline u64 btrfs_stripe_nr_to_offset(u32 stripe_nr)
622{
623	return (u64)stripe_nr << BTRFS_STRIPE_LEN_SHIFT;
624}
625
626void btrfs_get_bioc(struct btrfs_io_context *bioc);
627void btrfs_put_bioc(struct btrfs_io_context *bioc);
628int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 
 
629		    u64 logical, u64 *length,
630		    struct btrfs_io_context **bioc_ret,
631		    struct btrfs_io_stripe *smap, int *mirror_num_ret);
632int btrfs_map_repair_block(struct btrfs_fs_info *fs_info,
633			   struct btrfs_io_stripe *smap, u64 logical,
634			   u32 length, int mirror_num);
635struct btrfs_discard_stripe *btrfs_map_discard(struct btrfs_fs_info *fs_info,
636					       u64 logical, u64 *length_ret,
637					       u32 *num_stripes);
638int btrfs_read_sys_array(struct btrfs_fs_info *fs_info);
639int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info);
640struct btrfs_block_group *btrfs_create_chunk(struct btrfs_trans_handle *trans,
641					    u64 type);
642void btrfs_mapping_tree_free(struct btrfs_fs_info *fs_info);
643int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
644		       blk_mode_t flags, void *holder);
645struct btrfs_device *btrfs_scan_one_device(const char *path, blk_mode_t flags,
646					   bool mount_arg_dev);
647int btrfs_forget_devices(dev_t devt);
648void btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
649void btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices);
650void btrfs_assign_next_active_device(struct btrfs_device *device,
651				     struct btrfs_device *this_dev);
652struct btrfs_device *btrfs_find_device_by_devspec(struct btrfs_fs_info *fs_info,
653						  u64 devid,
654						  const char *devpath);
655int btrfs_get_dev_args_from_path(struct btrfs_fs_info *fs_info,
656				 struct btrfs_dev_lookup_args *args,
657				 const char *path);
658struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
659					const u64 *devid, const u8 *uuid,
660					const char *path);
661void btrfs_put_dev_args_from_path(struct btrfs_dev_lookup_args *args);
662int btrfs_rm_device(struct btrfs_fs_info *fs_info,
663		    struct btrfs_dev_lookup_args *args,
664		    struct bdev_handle **bdev_handle);
665void __exit btrfs_cleanup_fs_uuids(void);
666int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len);
667int btrfs_grow_device(struct btrfs_trans_handle *trans,
668		      struct btrfs_device *device, u64 new_size);
669struct btrfs_device *btrfs_find_device(const struct btrfs_fs_devices *fs_devices,
670				       const struct btrfs_dev_lookup_args *args);
671int btrfs_shrink_device(struct btrfs_device *device, u64 new_size);
672int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *path);
673int btrfs_balance(struct btrfs_fs_info *fs_info,
674		  struct btrfs_balance_control *bctl,
675		  struct btrfs_ioctl_balance_args *bargs);
676void btrfs_describe_block_groups(u64 flags, char *buf, u32 size_buf);
677int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info);
678int btrfs_recover_balance(struct btrfs_fs_info *fs_info);
679int btrfs_pause_balance(struct btrfs_fs_info *fs_info);
680int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset);
681int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
682int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info);
683int btrfs_uuid_scan_kthread(void *data);
684bool btrfs_chunk_writeable(struct btrfs_fs_info *fs_info, u64 chunk_offset);
 
685void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
686int btrfs_get_dev_stats(struct btrfs_fs_info *fs_info,
687			struct btrfs_ioctl_get_dev_stats *stats);
688int btrfs_init_devices_late(struct btrfs_fs_info *fs_info);
689int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info);
690int btrfs_run_dev_stats(struct btrfs_trans_handle *trans);
691void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_device *srcdev);
692void btrfs_rm_dev_replace_free_srcdev(struct btrfs_device *srcdev);
693void btrfs_destroy_dev_replace_tgtdev(struct btrfs_device *tgtdev);
694int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info,
695			   u64 logical, u64 len);
696unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info,
697				    u64 logical);
698u64 btrfs_calc_stripe_length(const struct btrfs_chunk_map *map);
699int btrfs_nr_parity_stripes(u64 type);
700int btrfs_chunk_alloc_add_chunk_item(struct btrfs_trans_handle *trans,
701				     struct btrfs_block_group *bg);
702int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset);
703
704#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
705struct btrfs_chunk_map *btrfs_alloc_chunk_map(int num_stripes, gfp_t gfp);
706int btrfs_add_chunk_map(struct btrfs_fs_info *fs_info, struct btrfs_chunk_map *map);
707#endif
708
709struct btrfs_chunk_map *btrfs_clone_chunk_map(struct btrfs_chunk_map *map, gfp_t gfp);
710struct btrfs_chunk_map *btrfs_find_chunk_map(struct btrfs_fs_info *fs_info,
711					     u64 logical, u64 length);
712struct btrfs_chunk_map *btrfs_find_chunk_map_nolock(struct btrfs_fs_info *fs_info,
713						    u64 logical, u64 length);
714struct btrfs_chunk_map *btrfs_get_chunk_map(struct btrfs_fs_info *fs_info,
715					    u64 logical, u64 length);
716void btrfs_remove_chunk_map(struct btrfs_fs_info *fs_info, struct btrfs_chunk_map *map);
717void btrfs_release_disk_super(struct btrfs_super_block *super);
718
719static inline void btrfs_dev_stat_inc(struct btrfs_device *dev,
720				      int index)
721{
722	atomic_inc(dev->dev_stat_values + index);
723	/*
724	 * This memory barrier orders stores updating statistics before stores
725	 * updating dev_stats_ccnt.
726	 *
727	 * It pairs with smp_rmb() in btrfs_run_dev_stats().
728	 */
729	smp_mb__before_atomic();
730	atomic_inc(&dev->dev_stats_ccnt);
731}
732
733static inline int btrfs_dev_stat_read(struct btrfs_device *dev,
734				      int index)
735{
736	return atomic_read(dev->dev_stat_values + index);
737}
738
739static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev,
740						int index)
741{
742	int ret;
743
744	ret = atomic_xchg(dev->dev_stat_values + index, 0);
745	/*
746	 * atomic_xchg implies a full memory barriers as per atomic_t.txt:
747	 * - RMW operations that have a return value are fully ordered;
748	 *
749	 * This implicit memory barriers is paired with the smp_rmb in
750	 * btrfs_run_dev_stats
751	 */
752	atomic_inc(&dev->dev_stats_ccnt);
753	return ret;
754}
755
756static inline void btrfs_dev_stat_set(struct btrfs_device *dev,
757				      int index, unsigned long val)
758{
759	atomic_set(dev->dev_stat_values + index, val);
760	/*
761	 * This memory barrier orders stores updating statistics before stores
762	 * updating dev_stats_ccnt.
763	 *
764	 * It pairs with smp_rmb() in btrfs_run_dev_stats().
765	 */
766	smp_mb__before_atomic();
767	atomic_inc(&dev->dev_stats_ccnt);
768}
769
770static inline const char *btrfs_dev_name(const struct btrfs_device *device)
 
771{
772	if (!device || test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state))
773		return "<missing disk>";
774	else
775		return rcu_str_deref(device->name);
776}
777
778void btrfs_commit_device_sizes(struct btrfs_transaction *trans);
779
780struct list_head * __attribute_const__ btrfs_get_fs_uuids(void);
781bool btrfs_check_rw_degradable(struct btrfs_fs_info *fs_info,
782					struct btrfs_device *failing_dev);
783void btrfs_scratch_superblocks(struct btrfs_fs_info *fs_info,
784			       struct block_device *bdev,
785			       const char *device_path);
786
787enum btrfs_raid_types __attribute_const__ btrfs_bg_flags_to_raid_index(u64 flags);
788int btrfs_bg_type_to_factor(u64 flags);
789const char *btrfs_bg_type_to_raid_name(u64 flags);
790int btrfs_verify_dev_extents(struct btrfs_fs_info *fs_info);
791bool btrfs_repair_one_zone(struct btrfs_fs_info *fs_info, u64 logical);
792
793bool btrfs_pinned_by_swapfile(struct btrfs_fs_info *fs_info, void *ptr);
794u8 *btrfs_sb_fsid_ptr(struct btrfs_super_block *sb);
795
796#endif
v3.5.6
 
  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_
 21
 22#include <linux/bio.h>
 23#include <linux/sort.h>
 
 24#include "async-thread.h"
 25#include "ioctl.h"
 
 
 
 
 
 
 26
 27#define BTRFS_STRIPE_LEN	(64 * 1024)
 
 
 28
 29struct buffer_head;
 30struct btrfs_pending_bios {
 31	struct bio *head;
 32	struct bio *tail;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 33};
 34
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 35struct btrfs_device {
 36	struct list_head dev_list;
 37	struct list_head dev_alloc_list;
 
 38	struct btrfs_fs_devices *fs_devices;
 39	struct btrfs_root *dev_root;
 40
 41	/* regular prio bios */
 42	struct btrfs_pending_bios pending_bios;
 43	/* WRITE_SYNC bios */
 44	struct btrfs_pending_bios pending_sync_bios;
 45
 46	int running_pending;
 47	u64 generation;
 48
 49	int writeable;
 50	int in_fs_metadata;
 51	int missing;
 52	int can_discard;
 53
 54	spinlock_t io_lock;
 55
 56	struct block_device *bdev;
 57
 58	/* the mode sent to blkdev_get */
 59	fmode_t mode;
 60
 61	struct rcu_string *name;
 
 
 
 
 
 
 
 
 
 
 62
 63	/* the internal btrfs device id */
 64	u64 devid;
 65
 66	/* size of the device */
 67	u64 total_bytes;
 68
 69	/* size of the disk */
 70	u64 disk_total_bytes;
 71
 72	/* bytes used */
 73	u64 bytes_used;
 74
 75	/* optimal io alignment for this device */
 76	u32 io_align;
 77
 78	/* optimal io width for this device */
 79	u32 io_width;
 
 
 80
 81	/* minimal io size for this device */
 82	u32 sector_size;
 83
 84	/* type and info about this device */
 85	u64 type;
 86
 87	/* physical drive uuid (or lvm uuid) */
 88	u8 uuid[BTRFS_UUID_SIZE];
 89
 90	/* per-device scrub information */
 91	struct scrub_dev *scrub_device;
 
 
 
 
 
 92
 93	struct btrfs_work work;
 94	struct rcu_head rcu;
 95	struct work_struct rcu_work;
 96
 97	/* readahead state */
 98	spinlock_t reada_lock;
 99	atomic_t reada_in_flight;
100	u64 reada_next;
101	struct reada_zone *reada_curr_zone;
102	struct radix_tree_root reada_zones;
103	struct radix_tree_root reada_extents;
104
105	/* for sending down flush barriers */
106	struct bio *flush_bio;
107	struct completion flush_wait;
108	int nobarriers;
 
 
109
110	/* disk I/O failure stats. For detailed description refer to
111	 * enum btrfs_dev_stat_values in ioctl.h */
112	int dev_stats_valid;
113	int dev_stats_dirty; /* counters need to be written to disk */
 
 
114	atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX];
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
115};
116
117struct btrfs_fs_devices {
118	u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
119
120	/* the device with this id has the most recent copy of the super */
121	u64 latest_devid;
122	u64 latest_trans;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
123	u64 num_devices;
 
 
 
 
 
124	u64 open_devices;
 
 
125	u64 rw_devices;
 
 
126	u64 missing_devices;
127	u64 total_rw_bytes;
128	u64 num_can_discard;
129	struct block_device *latest_bdev;
130
131	/* all of the devices in the FS, protected by a mutex
132	 * so we can safely walk it to write out the supers without
133	 * worrying about add/remove by the multi-device code
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
134	 */
135	struct mutex device_list_mutex;
 
 
136	struct list_head devices;
137
138	/* devices not currently being allocated */
139	struct list_head alloc_list;
140	struct list_head list;
141
142	struct btrfs_fs_devices *seed;
143	int seeding;
144
 
145	int opened;
146
147	/* set when we find or add a device that doesn't have the
148	 * nonrot flag set
149	 */
150	int rotating;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
151};
152
153struct btrfs_bio_stripe {
 
 
 
 
 
 
 
 
 
154	struct btrfs_device *dev;
 
155	u64 physical;
156	u64 length; /* only used for discard mappings */
 
 
 
157};
158
159struct btrfs_bio;
160typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
 
 
 
161
162struct btrfs_bio {
163	atomic_t stripes_pending;
164	bio_end_io_t *end_io;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
165	struct bio *orig_bio;
166	void *private;
167	atomic_t error;
168	int max_errors;
169	int num_stripes;
170	int mirror_num;
171	struct btrfs_bio_stripe stripes[];
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
172};
173
174struct btrfs_device_info {
175	struct btrfs_device *dev;
176	u64 dev_offset;
177	u64 max_avail;
178	u64 total_avail;
179};
180
181struct map_lookup {
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
182	u64 type;
183	int io_align;
184	int io_width;
185	int stripe_len;
186	int sector_size;
187	int num_stripes;
188	int sub_stripes;
189	struct btrfs_bio_stripe stripes[];
190};
191
192#define map_lookup_size(n) (sizeof(struct map_lookup) + \
193			    (sizeof(struct btrfs_bio_stripe) * (n)))
194
195/*
196 * Restriper's general type filter
197 */
198#define BTRFS_BALANCE_DATA		(1ULL << 0)
199#define BTRFS_BALANCE_SYSTEM		(1ULL << 1)
200#define BTRFS_BALANCE_METADATA		(1ULL << 2)
201
202#define BTRFS_BALANCE_TYPE_MASK		(BTRFS_BALANCE_DATA |	    \
203					 BTRFS_BALANCE_SYSTEM |	    \
204					 BTRFS_BALANCE_METADATA)
205
206#define BTRFS_BALANCE_FORCE		(1ULL << 3)
207#define BTRFS_BALANCE_RESUME		(1ULL << 4)
208
209/*
210 * Balance filters
211 */
212#define BTRFS_BALANCE_ARGS_PROFILES	(1ULL << 0)
213#define BTRFS_BALANCE_ARGS_USAGE	(1ULL << 1)
214#define BTRFS_BALANCE_ARGS_DEVID	(1ULL << 2)
215#define BTRFS_BALANCE_ARGS_DRANGE	(1ULL << 3)
216#define BTRFS_BALANCE_ARGS_VRANGE	(1ULL << 4)
217
218/*
219 * Profile changing flags.  When SOFT is set we won't relocate chunk if
220 * it already has the target profile (even though it may be
221 * half-filled).
222 */
223#define BTRFS_BALANCE_ARGS_CONVERT	(1ULL << 8)
224#define BTRFS_BALANCE_ARGS_SOFT		(1ULL << 9)
225
226struct btrfs_balance_args;
227struct btrfs_balance_progress;
228struct btrfs_balance_control {
229	struct btrfs_fs_info *fs_info;
230
231	struct btrfs_balance_args data;
232	struct btrfs_balance_args meta;
233	struct btrfs_balance_args sys;
234
235	u64 flags;
236
237	struct btrfs_balance_progress stat;
238};
239
240int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
241				   u64 end, u64 *length);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
242
243#define btrfs_bio_size(n) (sizeof(struct btrfs_bio) + \
244			    (sizeof(struct btrfs_bio_stripe) * (n)))
 
 
 
 
 
 
 
 
245
246int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
247			   struct btrfs_device *device,
248			   u64 chunk_tree, u64 chunk_objectid,
249			   u64 chunk_offset, u64 start, u64 num_bytes);
250int btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
251		    u64 logical, u64 *length,
252		    struct btrfs_bio **bbio_ret, int mirror_num);
253int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
254		     u64 chunk_start, u64 physical, u64 devid,
255		     u64 **logical, int *naddrs, int *stripe_len);
256int btrfs_read_sys_array(struct btrfs_root *root);
257int btrfs_read_chunk_tree(struct btrfs_root *root);
258int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
259		      struct btrfs_root *extent_root, u64 type);
260void btrfs_mapping_init(struct btrfs_mapping_tree *tree);
261void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree);
262int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
263		  int mirror_num, int async_submit);
 
264int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
265		       fmode_t flags, void *holder);
266int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
267			  struct btrfs_fs_devices **fs_devices_ret);
268int btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
269void btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices);
270int btrfs_add_device(struct btrfs_trans_handle *trans,
271		     struct btrfs_root *root,
272		     struct btrfs_device *device);
273int btrfs_rm_device(struct btrfs_root *root, char *device_path);
274void btrfs_cleanup_fs_uuids(void);
275int btrfs_num_copies(struct btrfs_mapping_tree *map_tree, u64 logical, u64 len);
 
 
 
 
 
 
 
 
 
 
 
 
276int btrfs_grow_device(struct btrfs_trans_handle *trans,
277		      struct btrfs_device *device, u64 new_size);
278struct btrfs_device *btrfs_find_device(struct btrfs_root *root, u64 devid,
279				       u8 *uuid, u8 *fsid);
280int btrfs_shrink_device(struct btrfs_device *device, u64 new_size);
281int btrfs_init_new_device(struct btrfs_root *root, char *path);
282int btrfs_balance(struct btrfs_balance_control *bctl,
 
283		  struct btrfs_ioctl_balance_args *bargs);
 
284int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info);
285int btrfs_recover_balance(struct btrfs_fs_info *fs_info);
286int btrfs_pause_balance(struct btrfs_fs_info *fs_info);
 
287int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
288int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset);
289int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes,
290			 u64 *start, u64 *max_avail);
291void btrfs_dev_stat_print_on_error(struct btrfs_device *device);
292void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
293int btrfs_get_dev_stats(struct btrfs_root *root,
294			struct btrfs_ioctl_get_dev_stats *stats,
295			int reset_after_read);
296int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info);
297int btrfs_run_dev_stats(struct btrfs_trans_handle *trans,
298			struct btrfs_fs_info *fs_info);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
299
300static inline void btrfs_dev_stat_inc(struct btrfs_device *dev,
301				      int index)
302{
303	atomic_inc(dev->dev_stat_values + index);
304	dev->dev_stats_dirty = 1;
 
 
 
 
 
 
 
305}
306
307static inline int btrfs_dev_stat_read(struct btrfs_device *dev,
308				      int index)
309{
310	return atomic_read(dev->dev_stat_values + index);
311}
312
313static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev,
314						int index)
315{
316	int ret;
317
318	ret = atomic_xchg(dev->dev_stat_values + index, 0);
319	dev->dev_stats_dirty = 1;
 
 
 
 
 
 
 
320	return ret;
321}
322
323static inline void btrfs_dev_stat_set(struct btrfs_device *dev,
324				      int index, unsigned long val)
325{
326	atomic_set(dev->dev_stat_values + index, val);
327	dev->dev_stats_dirty = 1;
 
 
 
 
 
 
 
328}
329
330static inline void btrfs_dev_stat_reset(struct btrfs_device *dev,
331					int index)
332{
333	btrfs_dev_stat_set(dev, index, 0);
 
 
 
334}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
335#endif