blob: f4151ee51b034201d3f8077c07c77857f3f4f3f3 [file] [log] [blame]
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001/* SPDX-License-Identifier: GPL-2.0 */
2#ifndef _BCACHEFS_H
3#define _BCACHEFS_H
4
5/*
6 * SOME HIGH LEVEL CODE DOCUMENTATION:
7 *
8 * Bcache mostly works with cache sets, cache devices, and backing devices.
9 *
10 * Support for multiple cache devices hasn't quite been finished off yet, but
11 * it's about 95% plumbed through. A cache set and its cache devices is sort of
12 * like a md raid array and its component devices. Most of the code doesn't care
13 * about individual cache devices, the main abstraction is the cache set.
14 *
15 * Multiple cache devices is intended to give us the ability to mirror dirty
16 * cached data and metadata, without mirroring clean cached data.
17 *
18 * Backing devices are different, in that they have a lifetime independent of a
19 * cache set. When you register a newly formatted backing device it'll come up
20 * in passthrough mode, and then you can attach and detach a backing device from
21 * a cache set at runtime - while it's mounted and in use. Detaching implicitly
22 * invalidates any cached data for that backing device.
23 *
24 * A cache set can have multiple (many) backing devices attached to it.
25 *
26 * There's also flash only volumes - this is the reason for the distinction
27 * between struct cached_dev and struct bcache_device. A flash only volume
28 * works much like a bcache device that has a backing device, except the
29 * "cached" data is always dirty. The end result is that we get thin
30 * provisioning with very little additional code.
31 *
32 * Flash only volumes work but they're not production ready because the moving
33 * garbage collector needs more work. More on that later.
34 *
35 * BUCKETS/ALLOCATION:
36 *
37 * Bcache is primarily designed for caching, which means that in normal
38 * operation all of our available space will be allocated. Thus, we need an
39 * efficient way of deleting things from the cache so we can write new things to
40 * it.
41 *
42 * To do this, we first divide the cache device up into buckets. A bucket is the
43 * unit of allocation; they're typically around 1 mb - anywhere from 128k to 2M+
44 * works efficiently.
45 *
46 * Each bucket has a 16 bit priority, and an 8 bit generation associated with
47 * it. The gens and priorities for all the buckets are stored contiguously and
48 * packed on disk (in a linked list of buckets - aside from the superblock, all
49 * of bcache's metadata is stored in buckets).
50 *
51 * The priority is used to implement an LRU. We reset a bucket's priority when
52 * we allocate it or on cache it, and every so often we decrement the priority
53 * of each bucket. It could be used to implement something more sophisticated,
54 * if anyone ever gets around to it.
55 *
56 * The generation is used for invalidating buckets. Each pointer also has an 8
57 * bit generation embedded in it; for a pointer to be considered valid, its gen
58 * must match the gen of the bucket it points into. Thus, to reuse a bucket all
59 * we have to do is increment its gen (and write its new gen to disk; we batch
60 * this up).
61 *
62 * Bcache is entirely COW - we never write twice to a bucket, even buckets that
63 * contain metadata (including btree nodes).
64 *
65 * THE BTREE:
66 *
67 * Bcache is in large part design around the btree.
68 *
69 * At a high level, the btree is just an index of key -> ptr tuples.
70 *
71 * Keys represent extents, and thus have a size field. Keys also have a variable
72 * number of pointers attached to them (potentially zero, which is handy for
73 * invalidating the cache).
74 *
75 * The key itself is an inode:offset pair. The inode number corresponds to a
76 * backing device or a flash only volume. The offset is the ending offset of the
77 * extent within the inode - not the starting offset; this makes lookups
78 * slightly more convenient.
79 *
80 * Pointers contain the cache device id, the offset on that device, and an 8 bit
81 * generation number. More on the gen later.
82 *
83 * Index lookups are not fully abstracted - cache lookups in particular are
84 * still somewhat mixed in with the btree code, but things are headed in that
85 * direction.
86 *
87 * Updates are fairly well abstracted, though. There are two different ways of
88 * updating the btree; insert and replace.
89 *
90 * BTREE_INSERT will just take a list of keys and insert them into the btree -
91 * overwriting (possibly only partially) any extents they overlap with. This is
92 * used to update the index after a write.
93 *
94 * BTREE_REPLACE is really cmpxchg(); it inserts a key into the btree iff it is
95 * overwriting a key that matches another given key. This is used for inserting
96 * data into the cache after a cache miss, and for background writeback, and for
97 * the moving garbage collector.
98 *
99 * There is no "delete" operation; deleting things from the index is
100 * accomplished by either by invalidating pointers (by incrementing a bucket's
101 * gen) or by inserting a key with 0 pointers - which will overwrite anything
102 * previously present at that location in the index.
103 *
104 * This means that there are always stale/invalid keys in the btree. They're
105 * filtered out by the code that iterates through a btree node, and removed when
106 * a btree node is rewritten.
107 *
108 * BTREE NODES:
109 *
Kent Overstreet3e3e02e2022-10-19 18:31:33 -0400110 * Our unit of allocation is a bucket, and we can't arbitrarily allocate and
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800111 * free smaller than a bucket - so, that's how big our btree nodes are.
112 *
113 * (If buckets are really big we'll only use part of the bucket for a btree node
114 * - no less than 1/4th - but a bucket still contains no more than a single
115 * btree node. I'd actually like to change this, but for now we rely on the
116 * bucket's gen for deleting btree nodes when we rewrite/split a node.)
117 *
118 * Anyways, btree nodes are big - big enough to be inefficient with a textbook
119 * btree implementation.
120 *
121 * The way this is solved is that btree nodes are internally log structured; we
122 * can append new keys to an existing btree node without rewriting it. This
123 * means each set of keys we write is sorted, but the node is not.
124 *
125 * We maintain this log structure in memory - keeping 1Mb of keys sorted would
126 * be expensive, and we have to distinguish between the keys we have written and
127 * the keys we haven't. So to do a lookup in a btree node, we have to search
128 * each sorted set. But we do merge written sets together lazily, so the cost of
129 * these extra searches is quite low (normally most of the keys in a btree node
130 * will be in one big set, and then there'll be one or two sets that are much
131 * smaller).
132 *
133 * This log structure makes bcache's btree more of a hybrid between a
134 * conventional btree and a compacting data structure, with some of the
135 * advantages of both.
136 *
137 * GARBAGE COLLECTION:
138 *
139 * We can't just invalidate any bucket - it might contain dirty data or
140 * metadata. If it once contained dirty data, other writes might overwrite it
141 * later, leaving no valid pointers into that bucket in the index.
142 *
143 * Thus, the primary purpose of garbage collection is to find buckets to reuse.
144 * It also counts how much valid data it each bucket currently contains, so that
145 * allocation can reuse buckets sooner when they've been mostly overwritten.
146 *
147 * It also does some things that are really internal to the btree
148 * implementation. If a btree node contains pointers that are stale by more than
149 * some threshold, it rewrites the btree node to avoid the bucket's generation
150 * wrapping around. It also merges adjacent btree nodes if they're empty enough.
151 *
152 * THE JOURNAL:
153 *
154 * Bcache's journal is not necessary for consistency; we always strictly
155 * order metadata writes so that the btree and everything else is consistent on
156 * disk in the event of an unclean shutdown, and in fact bcache had writeback
157 * caching (with recovery from unclean shutdown) before journalling was
158 * implemented.
159 *
160 * Rather, the journal is purely a performance optimization; we can't complete a
161 * write until we've updated the index on disk, otherwise the cache would be
162 * inconsistent in the event of an unclean shutdown. This means that without the
163 * journal, on random write workloads we constantly have to update all the leaf
164 * nodes in the btree, and those writes will be mostly empty (appending at most
165 * a few keys each) - highly inefficient in terms of amount of metadata writes,
166 * and it puts more strain on the various btree resorting/compacting code.
167 *
168 * The journal is just a log of keys we've inserted; on startup we just reinsert
169 * all the keys in the open journal entries. That means that when we're updating
170 * a node in the btree, we can wait until a 4k block of keys fills up before
171 * writing them out.
172 *
173 * For simplicity, we only journal updates to leaf nodes; updates to parent
174 * nodes are rare enough (since our leaf nodes are huge) that it wasn't worth
175 * the complexity to deal with journalling them (in particular, journal replay)
176 * - updates to non leaf nodes just happen synchronously (see btree_split()).
177 */
178
179#undef pr_fmt
Kent Overstreet365f64f2022-01-04 00:06:49 -0500180#ifdef __KERNEL__
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800181#define pr_fmt(fmt) "bcachefs: %s() " fmt "\n", __func__
Kent Overstreet365f64f2022-01-04 00:06:49 -0500182#else
183#define pr_fmt(fmt) "%s() " fmt "\n", __func__
184#endif
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800185
186#include <linux/backing-dev-defs.h>
187#include <linux/bug.h>
188#include <linux/bio.h>
189#include <linux/closure.h>
190#include <linux/kobject.h>
191#include <linux/list.h>
Kent Overstreet1dd7f9d2019-04-04 21:53:12 -0400192#include <linux/math64.h>
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800193#include <linux/mutex.h>
194#include <linux/percpu-refcount.h>
195#include <linux/percpu-rwsem.h>
Kent Overstreete06af202023-12-15 22:16:51 -0500196#include <linux/refcount.h>
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800197#include <linux/rhashtable.h>
198#include <linux/rwsem.h>
Kent Overstreetef1b2092021-05-18 23:53:43 -0400199#include <linux/semaphore.h>
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800200#include <linux/seqlock.h>
201#include <linux/shrinker.h>
Kent Overstreet876c7af2020-11-15 16:30:22 -0500202#include <linux/srcu.h>
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800203#include <linux/types.h>
204#include <linux/workqueue.h>
205#include <linux/zstd.h>
206
207#include "bcachefs_format.h"
Kent Overstreet1d16c602023-11-09 14:22:46 -0500208#include "disk_accounting_types.h"
Kent Overstreetfc6c01e2021-11-28 13:42:05 -0500209#include "errcode.h"
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800210#include "fifo.h"
Kent Overstreet350175b2022-12-14 20:52:11 -0500211#include "nocow_locking_types.h"
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800212#include "opts.h"
Kent Overstreetd2554262024-03-23 20:07:46 -0400213#include "recovery_passes_types.h"
Kent Overstreetf5d26fa2023-10-25 15:51:16 -0400214#include "sb-errors_types.h"
Kent Overstreeta5b696e2023-06-19 21:01:13 -0400215#include "seqmutex.h"
Kent Overstreetf1ca1ab2024-03-13 20:16:40 -0400216#include "time_stats.h"
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800217#include "util.h"
218
Kent Overstreetd94189a2023-02-09 12:21:45 -0500219#ifdef CONFIG_BCACHEFS_DEBUG
220#define BCH_WRITE_REF_DEBUG
221#endif
222
Kent Overstreetb9fa3752023-03-11 20:38:46 -0500223#ifndef dynamic_fault
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800224#define dynamic_fault(...) 0
Kent Overstreetb9fa3752023-03-11 20:38:46 -0500225#endif
226
227#define race_fault(...) dynamic_fault("bcachefs:race")
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800228
Kent Overstreet74644032023-11-27 22:37:27 -0500229#define count_event(_c, _name) this_cpu_inc((_c)->counters[BCH_COUNTER_##_name])
230
Kent Overstreet674cfc22022-08-27 12:48:36 -0400231#define trace_and_count(_c, _name, ...) \
232do { \
Kent Overstreet74644032023-11-27 22:37:27 -0500233 count_event(_c, _name); \
Kent Overstreet674cfc22022-08-27 12:48:36 -0400234 trace_##_name(__VA_ARGS__); \
235} while (0)
236
Kent Overstreetcd575dd2018-11-01 15:13:19 -0400237#define bch2_fs_init_fault(name) \
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800238 dynamic_fault("bcachefs:bch_fs_init:" name)
239#define bch2_meta_read_fault(name) \
240 dynamic_fault("bcachefs:meta:read:" name)
241#define bch2_meta_write_fault(name) \
242 dynamic_fault("bcachefs:meta:write:" name)
243
244#ifdef __KERNEL__
Kent Overstreetb2d1d562022-11-13 20:01:42 -0500245#define BCACHEFS_LOG_PREFIX
246#endif
247
248#ifdef BCACHEFS_LOG_PREFIX
Kent Overstreet7fec8262022-11-15 20:25:08 -0500249
250#define bch2_log_msg(_c, fmt) "bcachefs (%s): " fmt, ((_c)->name)
251#define bch2_fmt_dev(_ca, fmt) "bcachefs (%s): " fmt "\n", ((_ca)->name)
252#define bch2_fmt_dev_offset(_ca, _offset, fmt) "bcachefs (%s sector %llu): " fmt "\n", ((_ca)->name), (_offset)
253#define bch2_fmt_inum(_c, _inum, fmt) "bcachefs (%s inum %llu): " fmt "\n", ((_c)->name), (_inum)
254#define bch2_fmt_inum_offset(_c, _inum, _offset, fmt) \
255 "bcachefs (%s inum %llu offset %llu): " fmt "\n", ((_c)->name), (_inum), (_offset)
256
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800257#else
Kent Overstreet7fec8262022-11-15 20:25:08 -0500258
259#define bch2_log_msg(_c, fmt) fmt
260#define bch2_fmt_dev(_ca, fmt) "%s: " fmt "\n", ((_ca)->name)
261#define bch2_fmt_dev_offset(_ca, _offset, fmt) "%s sector %llu: " fmt "\n", ((_ca)->name), (_offset)
262#define bch2_fmt_inum(_c, _inum, fmt) "inum %llu: " fmt "\n", (_inum)
263#define bch2_fmt_inum_offset(_c, _inum, _offset, fmt) \
264 "inum %llu offset %llu: " fmt "\n", (_inum), (_offset)
265
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800266#endif
267
Kent Overstreet7fec8262022-11-15 20:25:08 -0500268#define bch2_fmt(_c, fmt) bch2_log_msg(_c, fmt "\n")
269
Kent Overstreet36008d52024-05-29 22:06:00 -0400270void bch2_print_str(struct bch_fs *, const char *);
271
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500272__printf(2, 3)
Kent Overstreetb63570f2024-02-12 17:15:29 -0500273void bch2_print_opts(struct bch_opts *, const char *, ...);
274
275__printf(2, 3)
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500276void __bch2_print(struct bch_fs *c, const char *fmt, ...);
277
278#define maybe_dev_to_fs(_c) _Generic((_c), \
279 struct bch_dev *: ((struct bch_dev *) (_c))->fs, \
280 struct bch_fs *: (_c))
281
282#define bch2_print(_c, ...) __bch2_print(maybe_dev_to_fs(_c), __VA_ARGS__)
283
284#define bch2_print_ratelimited(_c, ...) \
285do { \
286 static DEFINE_RATELIMIT_STATE(_rs, \
287 DEFAULT_RATELIMIT_INTERVAL, \
288 DEFAULT_RATELIMIT_BURST); \
289 \
290 if (__ratelimit(&_rs)) \
291 bch2_print(_c, __VA_ARGS__); \
292} while (0)
293
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800294#define bch_info(c, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500295 bch2_print(c, KERN_INFO bch2_fmt(c, fmt), ##__VA_ARGS__)
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800296#define bch_notice(c, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500297 bch2_print(c, KERN_NOTICE bch2_fmt(c, fmt), ##__VA_ARGS__)
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800298#define bch_warn(c, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500299 bch2_print(c, KERN_WARNING bch2_fmt(c, fmt), ##__VA_ARGS__)
Kent Overstreetac7f0d72019-04-03 20:38:37 -0400300#define bch_warn_ratelimited(c, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500301 bch2_print_ratelimited(c, KERN_WARNING bch2_fmt(c, fmt), ##__VA_ARGS__)
Kent Overstreet7fec8262022-11-15 20:25:08 -0500302
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800303#define bch_err(c, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500304 bch2_print(c, KERN_ERR bch2_fmt(c, fmt), ##__VA_ARGS__)
Kent Overstreet7fec8262022-11-15 20:25:08 -0500305#define bch_err_dev(ca, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500306 bch2_print(c, KERN_ERR bch2_fmt_dev(ca, fmt), ##__VA_ARGS__)
Kent Overstreet7fec8262022-11-15 20:25:08 -0500307#define bch_err_dev_offset(ca, _offset, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500308 bch2_print(c, KERN_ERR bch2_fmt_dev_offset(ca, _offset, fmt), ##__VA_ARGS__)
Kent Overstreet7fec8262022-11-15 20:25:08 -0500309#define bch_err_inum(c, _inum, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500310 bch2_print(c, KERN_ERR bch2_fmt_inum(c, _inum, fmt), ##__VA_ARGS__)
Kent Overstreet7fec8262022-11-15 20:25:08 -0500311#define bch_err_inum_offset(c, _inum, _offset, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500312 bch2_print(c, KERN_ERR bch2_fmt_inum_offset(c, _inum, _offset, fmt), ##__VA_ARGS__)
Kent Overstreet0fefe8d2020-12-03 13:57:22 -0500313
Kent Overstreetdfe9bfb2018-11-24 17:09:44 -0500314#define bch_err_ratelimited(c, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500315 bch2_print_ratelimited(c, KERN_ERR bch2_fmt(c, fmt), ##__VA_ARGS__)
Kent Overstreet7fec8262022-11-15 20:25:08 -0500316#define bch_err_dev_ratelimited(ca, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500317 bch2_print_ratelimited(ca, KERN_ERR bch2_fmt_dev(ca, fmt), ##__VA_ARGS__)
Kent Overstreet7fec8262022-11-15 20:25:08 -0500318#define bch_err_dev_offset_ratelimited(ca, _offset, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500319 bch2_print_ratelimited(ca, KERN_ERR bch2_fmt_dev_offset(ca, _offset, fmt), ##__VA_ARGS__)
Kent Overstreet0fefe8d2020-12-03 13:57:22 -0500320#define bch_err_inum_ratelimited(c, _inum, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500321 bch2_print_ratelimited(c, KERN_ERR bch2_fmt_inum(c, _inum, fmt), ##__VA_ARGS__)
Kent Overstreet7fec8262022-11-15 20:25:08 -0500322#define bch_err_inum_offset_ratelimited(c, _inum, _offset, fmt, ...) \
Kent Overstreet2b41226d2023-12-04 20:15:23 -0500323 bch2_print_ratelimited(c, KERN_ERR bch2_fmt_inum_offset(c, _inum, _offset, fmt), ##__VA_ARGS__)
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800324
Kent Overstreetcf904c82023-12-16 22:43:41 -0500325static inline bool should_print_err(int err)
326{
327 return err && !bch2_err_matches(err, BCH_ERR_transaction_restart);
328}
329
Kent Overstreet1bb3c2a2023-06-20 13:49:25 -0400330#define bch_err_fn(_c, _ret) \
Kent Overstreetd2a990d2023-09-26 16:02:06 -0400331do { \
Kent Overstreetcf904c82023-12-16 22:43:41 -0500332 if (should_print_err(_ret)) \
Kent Overstreetd2a990d2023-09-26 16:02:06 -0400333 bch_err(_c, "%s(): error %s", __func__, bch2_err_str(_ret));\
334} while (0)
335
Kent Overstreeta7dc10c2023-12-19 18:08:19 -0500336#define bch_err_fn_ratelimited(_c, _ret) \
337do { \
338 if (should_print_err(_ret)) \
339 bch_err_ratelimited(_c, "%s(): error %s", __func__, bch2_err_str(_ret));\
340} while (0)
341
Kent Overstreete691b392023-08-06 10:04:05 -0400342#define bch_err_msg(_c, _ret, _msg, ...) \
Kent Overstreetd2a990d2023-09-26 16:02:06 -0400343do { \
Kent Overstreetcf904c82023-12-16 22:43:41 -0500344 if (should_print_err(_ret)) \
Kent Overstreetd2a990d2023-09-26 16:02:06 -0400345 bch_err(_c, "%s(): error " _msg " %s", __func__, \
346 ##__VA_ARGS__, bch2_err_str(_ret)); \
347} while (0)
Kent Overstreet1bb3c2a2023-06-20 13:49:25 -0400348
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800349#define bch_verbose(c, fmt, ...) \
350do { \
Kent Overstreet0b847a12018-12-19 12:58:56 -0500351 if ((c)->opts.verbose) \
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800352 bch_info(c, fmt, ##__VA_ARGS__); \
353} while (0)
354
355#define pr_verbose_init(opts, fmt, ...) \
356do { \
Kent Overstreet0b847a12018-12-19 12:58:56 -0500357 if (opt_get(opts, verbose)) \
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800358 pr_info(fmt, ##__VA_ARGS__); \
359} while (0)
360
361/* Parameters that are useful for debugging, but should always be compiled in: */
362#define BCH_DEBUG_PARAMS_ALWAYS() \
363 BCH_DEBUG_PARAM(key_merging_disabled, \
364 "Disables merging of extents") \
Kent Overstreet55778812024-04-05 16:21:39 -0400365 BCH_DEBUG_PARAM(btree_node_merging_disabled, \
366 "Disables merging of btree nodes") \
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800367 BCH_DEBUG_PARAM(btree_gc_always_rewrite, \
368 "Causes mark and sweep to compact and rewrite every " \
369 "btree node it traverses") \
370 BCH_DEBUG_PARAM(btree_gc_rewrite_disabled, \
371 "Disables rewriting of btree nodes during mark and sweep")\
372 BCH_DEBUG_PARAM(btree_shrinker_disabled, \
Kent Overstreet6adaac02021-04-20 20:21:12 -0400373 "Disables the shrinker callback for the btree node cache")\
374 BCH_DEBUG_PARAM(verify_btree_ondisk, \
375 "Reread btree nodes at various points to verify the " \
376 "mergesort in the read path against modifications " \
Kent Overstreet1ce0cf52021-05-21 23:57:37 -0400377 "done in memory") \
378 BCH_DEBUG_PARAM(verify_all_btree_replicas, \
379 "When reading btree nodes, read all replicas and " \
Kent Overstreeta8c752b2022-03-17 20:51:27 -0400380 "compare them") \
381 BCH_DEBUG_PARAM(backpointers_no_use_write_buffer, \
382 "Don't use the write buffer for backpointers, enabling "\
383 "extra runtime checks")
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800384
Kent Overstreeta1019572022-10-22 15:59:53 -0400385/* Parameters that should only be compiled in debug mode: */
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800386#define BCH_DEBUG_PARAMS_DEBUG() \
387 BCH_DEBUG_PARAM(expensive_debug_checks, \
388 "Enables various runtime debugging checks that " \
389 "significantly affect performance") \
Kent Overstreetf13f5a82019-03-28 01:51:47 -0400390 BCH_DEBUG_PARAM(debug_check_iterators, \
391 "Enables extra verification for btree iterators") \
Kent Overstreet692d4032020-11-02 18:36:08 -0500392 BCH_DEBUG_PARAM(debug_check_btree_accounting, \
393 "Verify btree accounting for keys within a node") \
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800394 BCH_DEBUG_PARAM(journal_seq_verify, \
395 "Store the journal sequence number in the version " \
396 "number of every btree key, and verify that btree " \
397 "update ordering is preserved during recovery") \
398 BCH_DEBUG_PARAM(inject_invalid_keys, \
399 "Store the journal sequence number in the version " \
400 "number of every btree key, and verify that btree " \
401 "update ordering is preserved during recovery") \
Kent Overstreetb29e1972018-07-22 10:43:01 -0400402 BCH_DEBUG_PARAM(test_alloc_startup, \
403 "Force allocator startup to use the slowpath where it" \
404 "can't find enough free buckets without invalidating" \
Kent Overstreetcd575dd2018-11-01 15:13:19 -0400405 "cached data") \
406 BCH_DEBUG_PARAM(force_reconstruct_read, \
407 "Force reads to use the reconstruct path, when reading" \
Kent Overstreet6122ab62019-03-21 19:03:57 -0400408 "from erasure coded extents") \
409 BCH_DEBUG_PARAM(test_restart_gc, \
Kent Overstreetad7e1372019-08-28 13:20:31 -0400410 "Test restarting mark and sweep gc when bucket gens change")
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800411
412#define BCH_DEBUG_PARAMS_ALL() BCH_DEBUG_PARAMS_ALWAYS() BCH_DEBUG_PARAMS_DEBUG()
413
414#ifdef CONFIG_BCACHEFS_DEBUG
415#define BCH_DEBUG_PARAMS() BCH_DEBUG_PARAMS_ALL()
416#else
417#define BCH_DEBUG_PARAMS() BCH_DEBUG_PARAMS_ALWAYS()
418#endif
419
Kent Overstreet29364f32020-11-02 18:20:44 -0500420#define BCH_DEBUG_PARAM(name, description) extern bool bch2_##name;
421BCH_DEBUG_PARAMS()
422#undef BCH_DEBUG_PARAM
423
424#ifndef CONFIG_BCACHEFS_DEBUG
Kent Overstreet96dea3d2023-09-12 18:41:22 -0400425#define BCH_DEBUG_PARAM(name, description) static const __maybe_unused bool bch2_##name;
Kent Overstreet29364f32020-11-02 18:20:44 -0500426BCH_DEBUG_PARAMS_DEBUG()
427#undef BCH_DEBUG_PARAM
428#endif
429
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800430#define BCH_TIME_STATS() \
431 x(btree_node_mem_alloc) \
Kent Overstreetdc3b63d2019-03-21 16:28:57 -0400432 x(btree_node_split) \
Kent Overstreet991ba0212021-12-10 15:41:38 -0500433 x(btree_node_compact) \
434 x(btree_node_merge) \
Kent Overstreetdc3b63d2019-03-21 16:28:57 -0400435 x(btree_node_sort) \
436 x(btree_node_read) \
Kent Overstreetc72e4d72024-01-03 16:42:33 -0500437 x(btree_node_read_done) \
Kent Overstreet991ba0212021-12-10 15:41:38 -0500438 x(btree_interior_update_foreground) \
439 x(btree_interior_update_total) \
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800440 x(btree_gc) \
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800441 x(data_write) \
442 x(data_read) \
443 x(data_promote) \
Kent Overstreet991ba0212021-12-10 15:41:38 -0500444 x(journal_flush_write) \
445 x(journal_noflush_write) \
Kent Overstreet49a67202019-03-18 13:42:10 -0400446 x(journal_flush_seq) \
Kent Overstreet066a2642023-11-09 22:07:42 -0500447 x(blocked_journal_low_on_space) \
448 x(blocked_journal_low_on_pin) \
449 x(blocked_journal_max_in_flight) \
Kent Overstreet06a86932024-08-10 15:48:18 -0400450 x(blocked_key_cache_flush) \
Kent Overstreet49a67202019-03-18 13:42:10 -0400451 x(blocked_allocate) \
Kent Overstreeta8b3a672022-11-02 17:12:00 -0400452 x(blocked_allocate_open_bucket) \
Kent Overstreet09caeab2023-11-02 18:57:19 -0400453 x(blocked_write_buffer_full) \
Kent Overstreeta8b3a672022-11-02 17:12:00 -0400454 x(nocow_lock_contended)
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800455
456enum bch_time_stats {
457#define x(name) BCH_TIME_##name,
458 BCH_TIME_STATS()
459#undef x
460 BCH_TIME_STAT_NR
461};
462
463#include "alloc_types.h"
Kent Overstreet088d0de2024-05-27 18:40:50 -0400464#include "btree_gc_types.h"
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800465#include "btree_types.h"
Kent Overstreet4409b802024-03-11 23:11:46 -0400466#include "btree_node_scan_types.h"
Kent Overstreet920e69b2023-01-04 00:00:50 -0500467#include "btree_write_buffer_types.h"
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800468#include "buckets_types.h"
Kent Overstreet21aec962022-01-04 22:32:09 -0500469#include "buckets_waiting_for_journal_types.h"
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800470#include "clock_types.h"
Kent Overstreet37707bb2023-10-22 10:58:38 -0400471#include "disk_groups_types.h"
Kent Overstreetcd575dd2018-11-01 15:13:19 -0400472#include "ec_types.h"
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800473#include "journal_types.h"
474#include "keylist_types.h"
475#include "quota_types.h"
476#include "rebalance_types.h"
Kent Overstreet7a920562018-10-30 14:14:19 -0400477#include "replicas_types.h"
Kent Overstreetd1552722024-04-11 21:18:35 -0400478#include "sb-members_types.h"
Kent Overstreet20278752021-10-11 12:03:19 -0400479#include "subvolume_types.h"
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800480#include "super_types.h"
Kent Overstreet96f37ea2023-12-31 10:04:54 -0500481#include "thread_with_file_types.h"
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800482
483/* Number of nodes btree coalesce will try to coalesce at once */
484#define GC_MERGE_NODES 4U
485
486/* Maximum number of nodes we might need to allocate atomically: */
487#define BTREE_RESERVE_MAX (BTREE_MAX_DEPTH + (BTREE_MAX_DEPTH - 1))
488
489/* Size of the freelist we allocate btree nodes from: */
Kent Overstreet7dd1ebf2020-06-15 17:38:26 -0400490#define BTREE_NODE_RESERVE (BTREE_RESERVE_MAX * 4)
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800491
Kent Overstreetb030f692019-01-19 13:13:29 -0500492#define BTREE_NODE_OPEN_BUCKET_RESERVE (BTREE_RESERVE_MAX * BCH_REPLICAS_MAX)
493
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800494struct btree;
495
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800496struct io_count {
497 u64 sectors[2][BCH_DATA_NR];
498};
499
Kent Overstreet64ee1432024-06-23 00:53:44 -0400500struct discard_in_flight {
501 bool in_progress:1;
502 u64 bucket:63;
503};
504
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800505struct bch_dev {
506 struct kobject kobj;
Kent Overstreet552aa542024-05-03 18:07:40 -0400507#ifdef CONFIG_BCACHEFS_DEBUG
508 atomic_long_t ref;
509 bool dying;
510 unsigned long last_put;
511#else
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800512 struct percpu_ref ref;
Kent Overstreet552aa542024-05-03 18:07:40 -0400513#endif
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800514 struct completion ref_completion;
515 struct percpu_ref io_ref;
516 struct completion io_ref_completion;
517
518 struct bch_fs *fs;
519
520 u8 dev_idx;
521 /*
522 * Cached version of this device's member info from superblock
523 * Committed by bch2_write_super() -> bch_fs_mi_update()
524 */
525 struct bch_member_cpu mi;
Kent Overstreet94119ee2023-10-25 16:29:37 -0400526 atomic64_t errors[BCH_MEMBER_ERROR_NR];
527
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800528 __uuid_t uuid;
529 char name[BDEVNAME_SIZE];
530
531 struct bch_sb_handle disk_sb;
Kent Overstreet03e183c2019-03-21 23:13:46 -0400532 struct bch_sb *sb_read_scratch;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800533 int sb_write_error;
Kent Overstreeteacb2572022-01-02 21:45:35 -0500534 dev_t dev;
Kent Overstreeta8b3a672022-11-02 17:12:00 -0400535 atomic_t flush_seq;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800536
537 struct bch_devs_mask self;
538
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800539 /*
540 * Buckets:
Kent Overstreet9166b412018-11-26 00:13:33 -0500541 * Per-bucket arrays are protected by c->mark_lock, bucket_lock and
Kent Overstreetb0d3ab52024-06-13 17:07:36 -0400542 * gc_gens_lock, for device resize - holding any is sufficient for
543 * access: Or rcu_read_lock(), but only for dev_ptr_stale():
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800544 */
Kent Overstreet2c6a7bf2024-08-24 11:38:21 -0400545 GENRADIX(struct bucket) buckets_gc;
Kent Overstreet80bf2f32022-02-06 19:20:36 -0500546 struct bucket_gens __rcu *bucket_gens;
Kent Overstreetc45c8662021-12-24 04:51:10 -0500547 u8 *oldest_gen;
Kent Overstreet8eb7f3e2018-11-19 01:16:07 -0500548 unsigned long *buckets_nouse;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800549 struct rw_semaphore bucket_lock;
550
Kent Overstreetf5095b92024-01-01 19:42:37 -0500551 struct bch_dev_usage __percpu *usage;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800552
553 /* Allocator: */
Kent Overstreet099433132021-12-24 04:22:20 -0500554 u64 new_fs_bucket_idx;
Kent Overstreetc6705092024-04-20 16:25:34 -0400555 u64 alloc_cursor[3];
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800556
Kent Overstreetcb66fc52021-04-13 09:49:23 -0400557 unsigned nr_open_buckets;
Kent Overstreetf25d8212022-01-09 20:48:31 -0500558 unsigned nr_btree_reserve;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800559
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800560 size_t inc_gen_needs_gc;
561 size_t inc_gen_really_needs_gc;
Kent Overstreet21aec962022-01-04 22:32:09 -0500562 size_t buckets_waiting_on_journal;
Kent Overstreet430735c2018-11-19 01:31:41 -0500563
Kent Overstreet64ee1432024-06-23 00:53:44 -0400564 struct work_struct invalidate_work;
565 struct work_struct discard_work;
566 struct mutex discard_buckets_in_flight_lock;
567 DARRAY(struct discard_in_flight) discard_buckets_in_flight;
568 struct work_struct discard_fast_work;
569
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800570 atomic64_t rebalance_work;
571
572 struct journal_device journal;
Kent Overstreeta28bd482021-01-29 13:58:10 -0500573 u64 prev_journal_sector;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800574
575 struct work_struct io_error_work;
576
577 /* The rest of this all shows up in sysfs */
578 atomic64_t cur_latency[2];
Darrick J. Wong273960b2024-02-01 12:41:42 -0800579 struct bch2_time_stats_quantiles io_latency[2];
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800580
581#define CONGESTED_MAX 1024
582 atomic_t congested;
583 u64 congested_last;
584
585 struct io_count __percpu *io_done;
586};
587
Kent Overstreet3c471b62023-11-26 17:05:02 -0500588/*
Kent Overstreet3c471b62023-11-26 17:05:02 -0500589 * initial_gc_unfixed
590 * error
591 * topology error
592 */
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800593
Kent Overstreet3c471b62023-11-26 17:05:02 -0500594#define BCH_FS_FLAGS() \
Kent Overstreeta292be32024-03-27 22:50:19 -0400595 x(new_fs) \
Kent Overstreet3c471b62023-11-26 17:05:02 -0500596 x(started) \
Kent Overstreet1c0ee432024-09-26 16:19:58 -0400597 x(clean_recovery) \
Kent Overstreet7773df12024-04-25 20:45:00 -0400598 x(btree_running) \
Kent Overstreet5d9667d2023-11-17 00:23:07 -0500599 x(accounting_replay_done) \
Kent Overstreet3c471b62023-11-26 17:05:02 -0500600 x(may_go_rw) \
601 x(rw) \
602 x(was_rw) \
603 x(stopping) \
604 x(emergency_ro) \
605 x(going_ro) \
606 x(write_disable_complete) \
607 x(clean_shutdown) \
Kent Overstreetd55ddf62023-12-31 19:41:45 -0500608 x(fsck_running) \
Kent Overstreet3c471b62023-11-26 17:05:02 -0500609 x(initial_gc_unfixed) \
Kent Overstreet3c471b62023-11-26 17:05:02 -0500610 x(need_delete_dead_snapshots) \
611 x(error) \
612 x(topology_error) \
613 x(errors_fixed) \
Thomas Bertschinger07f9a272024-05-09 12:37:24 -0600614 x(errors_not_fixed) \
615 x(no_invalid_checks)
Kent Overstreet1cab5a82022-04-21 13:13:57 -0400616
Kent Overstreet3c471b62023-11-26 17:05:02 -0500617enum bch_fs_flags {
618#define x(n) BCH_FS_##n,
619 BCH_FS_FLAGS()
620#undef x
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800621};
622
623struct btree_debug {
624 unsigned id;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800625};
626
Kent Overstreet4aba7d42022-08-11 19:36:24 -0400627#define BCH_TRANSACTIONS_NR 128
628
629struct btree_transaction_stats {
Kent Overstreet89056f22023-12-23 22:43:33 -0500630 struct bch2_time_stats duration;
Kent Overstreet4aba7d42022-08-11 19:36:24 -0400631 struct bch2_time_stats lock_hold_times;
Kent Overstreet5c0bb662022-08-11 20:14:54 -0400632 struct mutex lock;
633 unsigned nr_max_paths;
Kent Overstreet24de63d2023-12-10 16:48:22 -0500634 unsigned journal_entries_size;
Kent Overstreet616928c2022-08-22 21:49:55 -0400635 unsigned max_mem;
Kent Overstreet5c0bb662022-08-11 20:14:54 -0400636 char *max_paths_text;
Daniel Hillc807ca92022-07-14 20:33:09 +1200637};
638
Kent Overstreet5663a412018-11-27 08:23:22 -0500639struct bch_fs_pcpu {
640 u64 sectors_available;
641};
642
Kent Overstreet1dd7f9d2019-04-04 21:53:12 -0400643struct journal_seq_blacklist_table {
644 size_t nr;
645 struct journal_seq_blacklist_table_entry {
646 u64 start;
647 u64 end;
648 bool dirty;
Gustavo A. R. Silva274c2f82023-11-06 15:40:22 -0600649 } entries[];
Kent Overstreet1dd7f9d2019-04-04 21:53:12 -0400650};
651
Kent Overstreetf1d786a2020-03-25 16:12:33 -0400652struct journal_keys {
Kent Overstreet894d0622024-02-24 00:15:56 -0500653 /* must match layout in darray_types.h */
654 size_t nr, size;
Kent Overstreetf1d786a2020-03-25 16:12:33 -0400655 struct journal_key {
Kent Overstreet1ffb8762022-09-12 02:22:47 -0400656 u64 journal_seq;
657 u32 journal_offset;
Kent Overstreetf1d786a2020-03-25 16:12:33 -0400658 enum btree_id btree_id:8;
659 unsigned level:8;
Kent Overstreet5b593ee2021-01-26 20:15:46 -0500660 bool allocated;
Kent Overstreetdfd41fb2021-12-31 17:54:13 -0500661 bool overwritten;
Kent Overstreetf1d786a2020-03-25 16:12:33 -0400662 struct bkey_i *k;
Kent Overstreet894d0622024-02-24 00:15:56 -0500663 } *data;
Kent Overstreetd1d77372022-04-04 01:09:26 -0400664 /*
665 * Gap buffer: instead of all the empty space in the array being at the
666 * end of the buffer - from @nr to @size - the empty space is at @gap.
667 * This means that sequential insertions are O(n) instead of O(n^2).
668 */
669 size_t gap;
Kent Overstreet8a443d32023-11-17 23:13:49 -0500670 atomic_t ref;
671 bool initial_ref_held;
Kent Overstreetf1d786a2020-03-25 16:12:33 -0400672};
673
Kent Overstreet6bd68ec2023-09-12 17:16:02 -0400674struct btree_trans_buf {
675 struct btree_trans *trans;
Kent Overstreet1a21bf92020-11-05 20:02:01 -0500676};
677
Kent Overstreet14b393e2021-03-16 00:42:25 -0400678#define BCACHEFS_ROOT_SUBVOL_INUM \
679 ((subvol_inum) { BCACHEFS_ROOT_SUBVOL, BCACHEFS_ROOT_INO })
680
Kent Overstreetd94189a2023-02-09 12:21:45 -0500681#define BCH_WRITE_REFS() \
682 x(trans) \
683 x(write) \
684 x(promote) \
685 x(node_rewrite) \
686 x(stripe_create) \
687 x(stripe_delete) \
688 x(reflink) \
689 x(fallocate) \
Kent Overstreet9e203c42024-04-13 00:26:01 -0400690 x(fsync) \
691 x(dio_write) \
Kent Overstreetd94189a2023-02-09 12:21:45 -0500692 x(discard) \
Kent Overstreeta393f332024-02-16 01:08:25 -0500693 x(discard_fast) \
Kent Overstreetd94189a2023-02-09 12:21:45 -0500694 x(invalidate) \
Kent Overstreetd94189a2023-02-09 12:21:45 -0500695 x(delete_dead_snapshots) \
Kent Overstreet103304022024-04-19 22:44:12 -0400696 x(gc_gens) \
Kent Overstreetd94189a2023-02-09 12:21:45 -0500697 x(snapshot_delete_pagecache) \
Kent Overstreet183bcc82023-11-02 19:37:15 -0400698 x(sysfs) \
699 x(btree_write_buffer)
Kent Overstreetd94189a2023-02-09 12:21:45 -0500700
701enum bch_write_ref {
702#define x(n) BCH_WRITE_REF_##n,
703 BCH_WRITE_REFS()
704#undef x
705 BCH_WRITE_REF_NR,
706};
707
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800708struct bch_fs {
709 struct closure cl;
710
711 struct list_head list;
712 struct kobject kobj;
Daniel Hill104c6972022-03-15 21:36:33 +1300713 struct kobject counters_kobj;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800714 struct kobject internal;
715 struct kobject opts_dir;
716 struct kobject time_stats;
717 unsigned long flags;
718
719 int minor;
720 struct device *chardev;
721 struct super_block *vfs_sb;
Kent Overstreetddc7dd62021-05-27 19:15:44 -0400722 dev_t dev;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800723 char name[40];
Kent Overstreet96f37ea2023-12-31 10:04:54 -0500724 struct stdio_redirect *stdio;
725 struct task_struct *stdio_filter;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800726
Kent Overstreet1ada1602020-06-15 14:58:47 -0400727 /* ro/rw, add/remove/resize devices: */
728 struct rw_semaphore state_lock;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800729
730 /* Counts outstanding writes, for clean transition to read-only */
Kent Overstreetd94189a2023-02-09 12:21:45 -0500731#ifdef BCH_WRITE_REF_DEBUG
732 atomic_long_t writes[BCH_WRITE_REF_NR];
733#else
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800734 struct percpu_ref writes;
Kent Overstreetd94189a2023-02-09 12:21:45 -0500735#endif
Kent Overstreet63508b72023-12-06 16:26:18 -0500736 /*
737 * Analagous to c->writes, for asynchronous ops that don't necessarily
738 * need fs to be read-write
739 */
740 refcount_t ro_ref;
741 wait_queue_head_t ro_ref_wait;
742
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800743 struct work_struct read_only_work;
744
745 struct bch_dev __rcu *devs[BCH_SB_MEMBERS_MAX];
746
Kent Overstreet2574e952024-06-06 13:25:28 -0400747 struct bch_accounting_mem accounting;
Kent Overstreet1d16c602023-11-09 14:22:46 -0500748
Kent Overstreet73e6ab92018-12-01 10:32:48 -0500749 struct bch_replicas_cpu replicas;
750 struct bch_replicas_cpu replicas_gc;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800751 struct mutex replicas_gc_lock;
752
Kent Overstreet4b8f89a2021-02-03 13:10:55 -0500753 struct journal_entry_res btree_root_journal_res;
Kent Overstreet4b8f89a2021-02-03 13:10:55 -0500754 struct journal_entry_res clock_journal_res;
Kent Overstreet180fb49d2021-01-21 21:52:06 -0500755
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800756 struct bch_disk_groups_cpu __rcu *disk_groups;
757
758 struct bch_opts opts;
759
760 /* Updated by bch2_sb_update():*/
761 struct {
762 __uuid_t uuid;
763 __uuid_t user_uuid;
764
Kent Overstreet26609b62018-11-01 15:10:01 -0400765 u16 version;
Kent Overstreet84cc7582021-03-21 16:03:23 -0400766 u16 version_min;
Kent Overstreet24964e12023-06-28 19:59:56 -0400767 u16 version_upgrade_complete;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800768
769 u8 nr_devices;
770 u8 clean;
771
772 u8 encryption_type;
773
774 u64 time_base_lo;
775 u32 time_base_hi;
Kent Overstreet595c1e92021-04-28 22:51:42 -0400776 unsigned time_units_per_sec;
777 unsigned nsec_per_time_unit;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800778 u64 features;
Kent Overstreet1df42b52019-02-06 11:56:51 -0500779 u64 compat;
Kent Overstreet8d65b152024-09-26 15:30:17 -0400780 unsigned long errors_silent[BITS_TO_LONGS(BCH_FSCK_ERR_MAX)];
Kent Overstreet55936af2024-03-15 23:03:42 -0400781 u64 btrees_lost_data;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800782 } sb;
783
Kent Overstreet595c1e92021-04-28 22:51:42 -0400784
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800785 struct bch_sb_handle disk_sb;
786
787 unsigned short block_bits; /* ilog2(block_size) */
788
789 u16 btree_foreground_merge_threshold;
790
791 struct closure sb_write;
792 struct mutex sb_lock;
793
Kent Overstreet14b393e2021-03-16 00:42:25 -0400794 /* snapshot.c: */
Kent Overstreet84799382023-07-12 13:55:03 -0400795 struct snapshot_table __rcu *snapshots;
Kent Overstreet14b393e2021-03-16 00:42:25 -0400796 struct mutex snapshot_table_lock;
Kent Overstreet37fad942023-09-29 01:15:33 -0400797 struct rw_semaphore snapshot_create_lock;
Kent Overstreet84799382023-07-12 13:55:03 -0400798
Kent Overstreet14b393e2021-03-16 00:42:25 -0400799 struct work_struct snapshot_delete_work;
Kent Overstreet20278752021-10-11 12:03:19 -0400800 struct work_struct snapshot_wait_for_pagecache_and_delete_work;
Kent Overstreet91d961b2022-03-29 15:48:45 -0400801 snapshot_id_list snapshots_unlinked;
Kent Overstreet20278752021-10-11 12:03:19 -0400802 struct mutex snapshots_unlinked_lock;
Kent Overstreet14b393e2021-03-16 00:42:25 -0400803
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800804 /* BTREE CACHE */
805 struct bio_set btree_bio;
Kent Overstreet161f73c2024-06-05 11:08:20 -0400806 struct workqueue_struct *btree_read_complete_wq;
807 struct workqueue_struct *btree_write_submit_wq;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800808
Kent Overstreetfaa6cb62023-06-28 22:09:13 -0400809 struct btree_root btree_roots_known[BTREE_ID_NR];
810 DARRAY(struct btree_root) btree_roots_extra;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800811 struct mutex btree_root_lock;
812
813 struct btree_cache btree_cache;
814
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800815 /*
816 * Cache of allocated btree nodes - if we allocate a btree node and
817 * don't use it, if we free it that space can't be reused until going
818 * _all_ the way through the allocator (which exposes us to a livelock
819 * when allocating btree reserves fail halfway through) - instead, we
820 * can stick them here:
821 */
822 struct btree_alloc btree_reserve_cache[BTREE_NODE_RESERVE * 2];
823 unsigned btree_reserve_cache_nr;
824 struct mutex btree_reserve_cache_lock;
825
826 mempool_t btree_interior_update_pool;
827 struct list_head btree_interior_update_list;
Kent Overstreetac7c51b2020-02-08 16:39:37 -0500828 struct list_head btree_interior_updates_unwritten;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800829 struct mutex btree_interior_update_lock;
830 struct closure_waitlist btree_interior_update_wait;
831
Kent Overstreet00b8ccf2020-05-25 14:57:06 -0400832 struct workqueue_struct *btree_interior_update_worker;
833 struct work_struct btree_interior_update_work;
834
Kent Overstreeta0a466e2024-03-17 20:25:39 -0400835 struct workqueue_struct *btree_node_rewrite_worker;
836
Kent Overstreeta1f26d72023-02-11 12:57:04 -0500837 struct list_head pending_node_rewrites;
838 struct mutex pending_node_rewrites_lock;
839
Kent Overstreet46fee692022-10-28 17:08:41 -0400840 /* btree_io.c: */
841 spinlock_t btree_write_error_lock;
842 struct btree_write_stats {
843 atomic64_t nr;
844 atomic64_t bytes;
845 } btree_write_stats[BTREE_WRITE_TYPE_NR];
846
Kent Overstreet495aabe2020-06-02 16:36:11 -0400847 /* btree_iter.c: */
Kent Overstreeta5b696e2023-06-19 21:01:13 -0400848 struct seqmutex btree_trans_lock;
Kent Overstreet495aabe2020-06-02 16:36:11 -0400849 struct list_head btree_trans_list;
Kent Overstreet6bd68ec2023-09-12 17:16:02 -0400850 mempool_t btree_trans_pool;
Kent Overstreete131b6a2021-04-24 00:09:06 -0400851 mempool_t btree_trans_mem_pool;
Kent Overstreet6bd68ec2023-09-12 17:16:02 -0400852 struct btree_trans_buf __percpu *btree_trans_bufs;
Kent Overstreet581edb62018-08-08 21:22:46 -0400853
Kent Overstreet876c7af2020-11-15 16:30:22 -0500854 struct srcu_struct btree_trans_barrier;
Kent Overstreet99fafb02021-12-20 18:18:35 -0500855 bool btree_trans_barrier_initialized;
Kent Overstreet876c7af2020-11-15 16:30:22 -0500856
Kent Overstreet2ca88e52019-03-07 19:46:10 -0500857 struct btree_key_cache btree_key_cache;
Kent Overstreet7c8f6f92022-01-12 02:13:21 -0500858 unsigned btree_key_cache_btrees;
Kent Overstreet2ca88e52019-03-07 19:46:10 -0500859
Kent Overstreet920e69b2023-01-04 00:00:50 -0500860 struct btree_write_buffer btree_write_buffer;
861
Kent Overstreet731bdd22021-05-22 17:37:25 -0400862 struct workqueue_struct *btree_update_wq;
Kent Overstreet9f1833c2021-07-10 13:44:42 -0400863 struct workqueue_struct *btree_io_complete_wq;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800864 /* copygc needs its own workqueue for index updates.. */
865 struct workqueue_struct *copygc_wq;
Brian Foster8bff9872023-03-23 14:09:05 -0400866 /*
867 * Use a dedicated wq for write ref holder tasks. Required to avoid
868 * dependency problems with other wq tasks that can block on ref
869 * draining, such as read-only transition.
870 */
871 struct workqueue_struct *write_ref_wq;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800872
873 /* ALLOCATION */
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800874 struct bch_devs_mask rw_devs[BCH_DATA_NR];
Kent Overstreet83ccd9b2024-09-06 19:12:53 -0400875 unsigned long rw_devs_change_count;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800876
877 u64 capacity; /* sectors */
Kent Overstreet26a170a2024-07-11 16:00:46 -0400878 u64 reserved; /* sectors */
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800879
880 /*
881 * When capacity _decreases_ (due to a disk being removed), we
882 * increment capacity_gen - this invalidates outstanding reservations
883 * and forces them to be revalidated
884 */
885 u32 capacity_gen;
Kent Overstreetb092dad2018-11-04 21:55:35 -0500886 unsigned bucket_size_max;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800887
888 atomic64_t sectors_available;
Kent Overstreetfca12232020-12-03 14:17:33 -0500889 struct mutex sectors_available_lock;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800890
Kent Overstreet5663a412018-11-27 08:23:22 -0500891 struct bch_fs_pcpu __percpu *pcpu;
Kent Overstreet9ca53b52018-07-23 05:32:01 -0400892
Kent Overstreet5663a412018-11-27 08:23:22 -0500893 struct percpu_rw_semaphore mark_lock;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800894
Kent Overstreet5e82a9a2019-02-10 19:34:47 -0500895 seqcount_t usage_lock;
Kent Overstreet8bb8d682023-12-27 22:09:25 -0500896 struct bch_fs_usage_base __percpu *usage;
Kent Overstreet5e82a9a2019-02-10 19:34:47 -0500897 u64 __percpu *online_reserved;
Kent Overstreet4d8100d2019-03-15 18:20:46 -0400898
Kent Overstreetcecf7272024-08-07 13:58:57 -0400899 unsigned long allocator_last_stuck;
900
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800901 struct io_clock io_clock[2];
902
Kent Overstreet1dd7f9d2019-04-04 21:53:12 -0400903 /* JOURNAL SEQ BLACKLIST */
904 struct journal_seq_blacklist_table *
905 journal_seq_blacklist_table;
Kent Overstreet1dd7f9d2019-04-04 21:53:12 -0400906
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800907 /* ALLOCATOR */
908 spinlock_t freelist_lock;
Kent Overstreet90541a72018-07-21 23:36:11 -0400909 struct closure_waitlist freelist_wait;
Kent Overstreet9ddffaf2021-12-25 21:43:29 -0500910
Kent Overstreet374153c2020-06-09 15:44:03 -0400911 open_bucket_idx_t open_buckets_freelist;
912 open_bucket_idx_t open_buckets_nr_free;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800913 struct closure_waitlist open_buckets_wait;
914 struct open_bucket open_buckets[OPEN_BUCKETS_COUNT];
Kent Overstreet9ddffaf2021-12-25 21:43:29 -0500915 open_bucket_idx_t open_buckets_hash[OPEN_BUCKETS_COUNT];
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800916
Kent Overstreet39a1ea12023-02-25 00:32:34 -0500917 open_bucket_idx_t open_buckets_partial[OPEN_BUCKETS_COUNT];
918 open_bucket_idx_t open_buckets_partial_nr;
919
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800920 struct write_point btree_write_point;
921 struct write_point rebalance_write_point;
922
Kent Overstreetb092dad2018-11-04 21:55:35 -0500923 struct write_point write_points[WRITE_POINT_MAX];
924 struct hlist_head write_points_hash[WRITE_POINT_HASH_NR];
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800925 struct mutex write_points_hash_lock;
Kent Overstreetb092dad2018-11-04 21:55:35 -0500926 unsigned write_points_nr;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800927
Kent Overstreet21aec962022-01-04 22:32:09 -0500928 struct buckets_waiting_for_journal buckets_waiting_for_journal;
929
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800930 /* GARBAGE COLLECTION */
Kent Overstreet103304022024-04-19 22:44:12 -0400931 struct work_struct gc_gens_work;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800932 unsigned long gc_count;
933
Kent Overstreetac516d02021-04-13 15:00:40 -0400934 enum btree_id gc_gens_btree;
935 struct bpos gc_gens_pos;
936
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800937 /*
938 * Tracks GC's progress - everything in the range [ZERO_KEY..gc_cur_pos]
939 * has been marked by GC.
940 *
Kent Overstreet41f8b092021-02-20 19:27:37 -0500941 * gc_cur_phase is a superset of btree_ids (BTREE_ID_extents etc.)
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800942 *
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800943 * Protected by gc_pos_lock. Only written to by GC thread, so GC thread
944 * can read without a lock.
945 */
946 seqcount_t gc_pos_lock;
947 struct gc_pos gc_pos;
948
949 /*
950 * The allocation code needs gc_mark in struct bucket to be correct, but
951 * it's not while a gc is in progress.
952 */
953 struct rw_semaphore gc_lock;
Kent Overstreetc45c8662021-12-24 04:51:10 -0500954 struct mutex gc_gens_lock;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800955
956 /* IO PATH */
Kent Overstreetef1b2092021-05-18 23:53:43 -0400957 struct semaphore io_in_flight;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800958 struct bio_set bio_read;
959 struct bio_set bio_read_split;
960 struct bio_set bio_write;
Kent Overstreetdbd04082024-04-30 20:32:44 -0400961 struct bio_set replica_set;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800962 struct mutex bio_bounce_pages_lock;
Kent Overstreeta8b3a672022-11-02 17:12:00 -0400963 mempool_t bio_bounce_pages;
964 struct bucket_nocow_lock_table
965 nocow_locks;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800966 struct rhashtable promote_table;
967
968 mempool_t compression_bounce[2];
Kent Overstreet1c3ff722019-12-28 20:17:06 -0500969 mempool_t compress_workspace[BCH_COMPRESSION_TYPE_NR];
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800970 mempool_t decompress_workspace;
Kent Overstreetbbc3a462023-11-24 23:12:45 -0500971 size_t zstd_workspace_size;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800972
973 struct crypto_shash *sha256;
974 struct crypto_sync_skcipher *chacha20;
975 struct crypto_shash *poly1305;
976
977 atomic64_t key_version;
978
Kent Overstreet35189e02019-11-09 16:01:15 -0500979 mempool_t large_bkey_pool;
980
Kent Overstreetb9fa3752023-03-11 20:38:46 -0500981 /* MOVE.C */
982 struct list_head moving_context_list;
983 struct mutex moving_context_lock;
984
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -0800985 /* REBALANCE */
986 struct bch_fs_rebalance rebalance;
987
Kent Overstreete6d11612020-07-11 16:28:54 -0400988 /* COPYGC */
989 struct task_struct *copygc_thread;
Kent Overstreete6d11612020-07-11 16:28:54 -0400990 struct write_point copygc_write_point;
Kent Overstreet0fb11e082023-03-17 09:59:17 -0400991 s64 copygc_wait_at;
Kent Overstreet5bbe4bf2021-04-13 14:45:55 -0400992 s64 copygc_wait;
Daniel Hillc91996c2022-06-16 02:06:43 +1200993 bool copygc_running;
994 wait_queue_head_t copygc_running_wq;
Kent Overstreete6d11612020-07-11 16:28:54 -0400995
Kent Overstreetdfe9bfb2018-11-24 17:09:44 -0500996 /* STRIPES: */
Kent Overstreet990d42d2021-12-04 23:07:33 -0500997 GENRADIX(struct stripe) stripes;
998 GENRADIX(struct gc_stripe) gc_stripes;
Kent Overstreetcd575dd2018-11-01 15:13:19 -0400999
Kent Overstreet4b1e6692023-02-18 21:07:25 -05001000 struct hlist_head ec_stripes_new[32];
1001 spinlock_t ec_stripes_new_lock;
1002
Kent Overstreetcd575dd2018-11-01 15:13:19 -04001003 ec_stripes_heap ec_stripes_heap;
Kent Overstreet627a2312023-02-18 20:49:37 -05001004 struct mutex ec_stripes_heap_lock;
Kent Overstreetcd575dd2018-11-01 15:13:19 -04001005
Kent Overstreetdfe9bfb2018-11-24 17:09:44 -05001006 /* ERASURE CODING */
Kent Overstreet703e2a42020-07-06 20:59:46 -04001007 struct list_head ec_stripe_head_list;
1008 struct mutex ec_stripe_head_lock;
1009
1010 struct list_head ec_stripe_new_list;
1011 struct mutex ec_stripe_new_lock;
Kent Overstreetb40901b2023-03-13 22:01:47 -04001012 wait_queue_head_t ec_stripe_new_wait;
Kent Overstreet703e2a42020-07-06 20:59:46 -04001013
1014 struct work_struct ec_stripe_create_work;
Kent Overstreet4e1510c2019-08-22 17:09:16 -04001015 u64 ec_stripe_hint;
Kent Overstreetdfe9bfb2018-11-24 17:09:44 -05001016
Kent Overstreetcd575dd2018-11-01 15:13:19 -04001017 struct work_struct ec_stripe_delete_work;
Kent Overstreetb40901b2023-03-13 22:01:47 -04001018
1019 struct bio_set ec_bioset;
Kent Overstreetcd575dd2018-11-01 15:13:19 -04001020
Kent Overstreet76426092019-08-16 09:59:56 -04001021 /* REFLINK */
Kent Overstreet890b74f2021-05-23 02:31:33 -04001022 reflink_gc_table reflink_gc_table;
1023 size_t reflink_gc_nr;
Kent Overstreet76426092019-08-16 09:59:56 -04001024
Kent Overstreet9edbcc72023-03-15 11:53:51 -04001025 /* fs.c */
1026 struct list_head vfs_inodes_list;
1027 struct mutex vfs_inodes_lock;
Kent Overstreet112d21f2024-06-08 21:41:01 -04001028 struct rhashtable vfs_inodes_table;
Kent Overstreet9edbcc72023-03-15 11:53:51 -04001029
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001030 /* VFS IO PATH - fs-io.c */
1031 struct bio_set writepage_bioset;
1032 struct bio_set dio_write_bioset;
1033 struct bio_set dio_read_bioset;
Kent Overstreeta8b3a672022-11-02 17:12:00 -04001034 struct bio_set nocow_flush_bioset;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001035
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001036 /* QUOTAS */
1037 struct bch_memquota_type quotas[QTYP_NR];
1038
Kent Overstreet78328fe2023-07-08 22:33:29 -04001039 /* RECOVERY */
1040 u64 journal_replay_seq_start;
1041 u64 journal_replay_seq_end;
Kent Overstreet7f391b22023-12-06 14:36:18 -05001042 /*
1043 * Two different uses:
1044 * "Has this fsck pass?" - i.e. should this type of error be an
1045 * emergency read-only
1046 * And, in certain situations fsck will rewind to an earlier pass: used
1047 * for signaling to the toplevel code which pass we want to run now.
1048 */
Kent Overstreet067d2282023-07-07 02:42:28 -04001049 enum bch_recovery_pass curr_recovery_pass;
Kent Overstreet249bf592023-12-10 12:42:49 -05001050 /* bitmask of recovery passes that we actually ran */
Kent Overstreet0ed4ca12023-08-03 20:57:06 -04001051 u64 recovery_passes_complete;
Kent Overstreet249bf592023-12-10 12:42:49 -05001052 /* never rewinds version of curr_recovery_pass */
1053 enum bch_recovery_pass recovery_pass_done;
Kent Overstreet267b8012023-12-04 13:45:33 -05001054 struct semaphore online_fsck_mutex;
Kent Overstreet067d2282023-07-07 02:42:28 -04001055
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001056 /* DEBUG JUNK */
Kent Overstreet75ef2c52022-02-26 11:48:34 -05001057 struct dentry *fs_debug_dir;
1058 struct dentry *btree_debug_dir;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001059 struct btree_debug btree_debug[BTREE_ID_NR];
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001060 struct btree *verify_data;
1061 struct btree_node *verify_ondisk;
1062 struct mutex verify_lock;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001063
Kent Overstreetb5e8a692020-11-02 23:51:33 -05001064 u64 *unused_inode_hints;
1065 unsigned inode_shard_bits;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001066
1067 /*
1068 * A btree node on disk could have too many bsets for an iterator to fit
1069 * on the stack - have to dynamically allocate them
1070 */
1071 mempool_t fill_iter;
1072
1073 mempool_t btree_bounce_pool;
1074
1075 struct journal journal;
Kent Overstreetce6201c2022-03-21 00:15:53 -04001076 GENRADIX(struct journal_replay *) journal_entries;
1077 u64 journal_entries_base_seq;
Kent Overstreetf1d786a2020-03-25 16:12:33 -04001078 struct journal_keys journal_keys;
Kent Overstreet5b593ee2021-01-26 20:15:46 -05001079 struct list_head journal_iters;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001080
Kent Overstreet4409b802024-03-11 23:11:46 -04001081 struct find_btree_nodes found_btree_nodes;
1082
Kent Overstreetc6923992018-07-21 22:57:20 -04001083 u64 last_bucket_seq_cleanup;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001084
Daniel Hill104c6972022-03-15 21:36:33 +13001085 u64 counters_on_mount[BCH_COUNTER_NR];
1086 u64 __percpu *counters;
1087
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001088 unsigned copy_gc_enabled:1;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001089
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001090 struct bch2_time_stats times[BCH_TIME_STAT_NR];
Daniel Hillc807ca92022-07-14 20:33:09 +12001091
Kent Overstreet4aba7d42022-08-11 19:36:24 -04001092 struct btree_transaction_stats btree_transaction_stats[BCH_TRANSACTIONS_NR];
Kent Overstreetf5d26fa2023-10-25 15:51:16 -04001093
1094 /* ERRORS */
1095 struct list_head fsck_error_msgs;
1096 struct mutex fsck_error_msgs_lock;
1097 bool fsck_alloc_msgs_err;
1098
1099 bch_sb_errors_cpu fsck_error_counts;
1100 struct mutex fsck_error_counts_lock;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001101};
1102
Kent Overstreetd94189a2023-02-09 12:21:45 -05001103extern struct wait_queue_head bch2_read_only_wait;
1104
1105static inline void bch2_write_ref_get(struct bch_fs *c, enum bch_write_ref ref)
1106{
1107#ifdef BCH_WRITE_REF_DEBUG
1108 atomic_long_inc(&c->writes[ref]);
1109#else
1110 percpu_ref_get(&c->writes);
1111#endif
1112}
1113
Kent Overstreet09caeab2023-11-02 18:57:19 -04001114static inline bool __bch2_write_ref_tryget(struct bch_fs *c, enum bch_write_ref ref)
1115{
1116#ifdef BCH_WRITE_REF_DEBUG
1117 return !test_bit(BCH_FS_going_ro, &c->flags) &&
1118 atomic_long_inc_not_zero(&c->writes[ref]);
1119#else
1120 return percpu_ref_tryget(&c->writes);
1121#endif
1122}
1123
Kent Overstreetd94189a2023-02-09 12:21:45 -05001124static inline bool bch2_write_ref_tryget(struct bch_fs *c, enum bch_write_ref ref)
1125{
1126#ifdef BCH_WRITE_REF_DEBUG
Kent Overstreet3c471b62023-11-26 17:05:02 -05001127 return !test_bit(BCH_FS_going_ro, &c->flags) &&
Kent Overstreetd94189a2023-02-09 12:21:45 -05001128 atomic_long_inc_not_zero(&c->writes[ref]);
1129#else
1130 return percpu_ref_tryget_live(&c->writes);
1131#endif
1132}
1133
1134static inline void bch2_write_ref_put(struct bch_fs *c, enum bch_write_ref ref)
1135{
1136#ifdef BCH_WRITE_REF_DEBUG
1137 long v = atomic_long_dec_return(&c->writes[ref]);
1138
1139 BUG_ON(v < 0);
1140 if (v)
1141 return;
1142 for (unsigned i = 0; i < BCH_WRITE_REF_NR; i++)
1143 if (atomic_long_read(&c->writes[i]))
1144 return;
1145
Kent Overstreet3c471b62023-11-26 17:05:02 -05001146 set_bit(BCH_FS_write_disable_complete, &c->flags);
Kent Overstreetd94189a2023-02-09 12:21:45 -05001147 wake_up(&bch2_read_only_wait);
1148#else
1149 percpu_ref_put(&c->writes);
1150#endif
1151}
1152
Kent Overstreet63508b72023-12-06 16:26:18 -05001153static inline bool bch2_ro_ref_tryget(struct bch_fs *c)
1154{
1155 if (test_bit(BCH_FS_stopping, &c->flags))
1156 return false;
1157
1158 return refcount_inc_not_zero(&c->ro_ref);
1159}
1160
1161static inline void bch2_ro_ref_put(struct bch_fs *c)
1162{
1163 if (refcount_dec_and_test(&c->ro_ref))
1164 wake_up(&c->ro_ref_wait);
1165}
1166
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001167static inline void bch2_set_ra_pages(struct bch_fs *c, unsigned ra_pages)
1168{
1169#ifndef NO_BCACHEFS_FS
1170 if (c->vfs_sb)
1171 c->vfs_sb->s_bdi->ra_pages = ra_pages;
1172#endif
1173}
1174
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001175static inline unsigned bucket_bytes(const struct bch_dev *ca)
1176{
1177 return ca->mi.bucket_size << 9;
1178}
1179
1180static inline unsigned block_bytes(const struct bch_fs *c)
1181{
Kent Overstreet8244f322021-12-14 14:24:41 -05001182 return c->opts.block_size;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001183}
1184
Kent Overstreet8244f322021-12-14 14:24:41 -05001185static inline unsigned block_sectors(const struct bch_fs *c)
1186{
1187 return c->opts.block_size >> 9;
1188}
1189
Kent Overstreet7c8f6f92022-01-12 02:13:21 -05001190static inline bool btree_id_cached(const struct bch_fs *c, enum btree_id btree)
1191{
1192 return c->btree_key_cache_btrees & (1U << btree);
1193}
1194
Kent Overstreet8244f322021-12-14 14:24:41 -05001195static inline struct timespec64 bch2_time_to_timespec(const struct bch_fs *c, s64 time)
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001196{
Kent Overstreet595c1e92021-04-28 22:51:42 -04001197 struct timespec64 t;
Alyssa Rossa3ed1cc2024-09-07 18:00:26 +02001198 s64 sec;
Kent Overstreet595c1e92021-04-28 22:51:42 -04001199 s32 rem;
1200
1201 time += c->sb.time_base_lo;
1202
Alyssa Rossa3ed1cc2024-09-07 18:00:26 +02001203 sec = div_s64_rem(time, c->sb.time_units_per_sec, &rem);
1204
1205 set_normalized_timespec64(&t, sec, rem * (s64)c->sb.nsec_per_time_unit);
1206
Kent Overstreet595c1e92021-04-28 22:51:42 -04001207 return t;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001208}
1209
Kent Overstreet8244f322021-12-14 14:24:41 -05001210static inline s64 timespec_to_bch2_time(const struct bch_fs *c, struct timespec64 ts)
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001211{
Kent Overstreet595c1e92021-04-28 22:51:42 -04001212 return (ts.tv_sec * c->sb.time_units_per_sec +
1213 (int) ts.tv_nsec / c->sb.nsec_per_time_unit) - c->sb.time_base_lo;
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001214}
1215
Kent Overstreet8244f322021-12-14 14:24:41 -05001216static inline s64 bch2_current_time(const struct bch_fs *c)
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001217{
1218 struct timespec64 now;
1219
Kent Overstreetea4160232019-04-16 16:03:31 -04001220 ktime_get_coarse_real_ts64(&now);
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001221 return timespec_to_bch2_time(c, now);
1222}
1223
Kent Overstreetcff07e22024-06-17 10:06:03 -04001224static inline u64 bch2_current_io_time(const struct bch_fs *c, int rw)
1225{
1226 return max(1ULL, (u64) atomic64_read(&c->io_clock[rw].now) & LRU_TIME_MAX);
1227}
1228
Kent Overstreet96f37ea2023-12-31 10:04:54 -05001229static inline struct stdio_redirect *bch2_fs_stdio_redirect(struct bch_fs *c)
1230{
1231 struct stdio_redirect *stdio = c->stdio;
1232
1233 if (c->stdio_filter && c->stdio_filter != current)
1234 stdio = NULL;
1235 return stdio;
1236}
1237
Kent Overstreet4e074472024-02-10 21:01:40 -05001238static inline unsigned metadata_replicas_required(struct bch_fs *c)
1239{
1240 return min(c->opts.metadata_replicas,
1241 c->opts.metadata_replicas_required);
1242}
1243
1244static inline unsigned data_replicas_required(struct bch_fs *c)
1245{
1246 return min(c->opts.data_replicas,
1247 c->opts.data_replicas_required);
1248}
1249
Kent Overstreet45dd05b2023-03-04 22:36:02 -05001250#define BKEY_PADDED_ONSTACK(key, pad) \
1251 struct { struct bkey_i key; __u64 key ## _pad[pad]; }
1252
Kent Overstreet1c6fdbd2017-03-16 22:18:50 -08001253#endif /* _BCACHEFS_H */