4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23 * Copyright (c) 2012, Joyent, Inc. All rights reserved.
24 * Copyright (c) 2011, 2017 by Delphix. All rights reserved.
25 * Copyright (c) 2014 by Saso Kiselkov. All rights reserved.
26 * Copyright 2015 Nexenta Systems, Inc. All rights reserved.
30 * DVA-based Adjustable Replacement Cache
32 * While much of the theory of operation used here is
33 * based on the self-tuning, low overhead replacement cache
34 * presented by Megiddo and Modha at FAST 2003, there are some
35 * significant differences:
37 * 1. The Megiddo and Modha model assumes any page is evictable.
38 * Pages in its cache cannot be "locked" into memory. This makes
39 * the eviction algorithm simple: evict the last page in the list.
40 * This also make the performance characteristics easy to reason
41 * about. Our cache is not so simple. At any given moment, some
42 * subset of the blocks in the cache are un-evictable because we
43 * have handed out a reference to them. Blocks are only evictable
44 * when there are no external references active. This makes
45 * eviction far more problematic: we choose to evict the evictable
46 * blocks that are the "lowest" in the list.
48 * There are times when it is not possible to evict the requested
49 * space. In these circumstances we are unable to adjust the cache
50 * size. To prevent the cache growing unbounded at these times we
51 * implement a "cache throttle" that slows the flow of new data
52 * into the cache until we can make space available.
54 * 2. The Megiddo and Modha model assumes a fixed cache size.
55 * Pages are evicted when the cache is full and there is a cache
56 * miss. Our model has a variable sized cache. It grows with
57 * high use, but also tries to react to memory pressure from the
58 * operating system: decreasing its size when system memory is
61 * 3. The Megiddo and Modha model assumes a fixed page size. All
62 * elements of the cache are therefore exactly the same size. So
63 * when adjusting the cache size following a cache miss, its simply
64 * a matter of choosing a single page to evict. In our model, we
65 * have variable sized cache blocks (rangeing from 512 bytes to
66 * 128K bytes). We therefore choose a set of blocks to evict to make
67 * space for a cache miss that approximates as closely as possible
68 * the space used by the new block.
70 * See also: "ARC: A Self-Tuning, Low Overhead Replacement Cache"
71 * by N. Megiddo & D. Modha, FAST 2003
77 * A new reference to a cache buffer can be obtained in two
78 * ways: 1) via a hash table lookup using the DVA as a key,
79 * or 2) via one of the ARC lists. The arc_read() interface
80 * uses method 1, while the internal ARC algorithms for
81 * adjusting the cache use method 2. We therefore provide two
82 * types of locks: 1) the hash table lock array, and 2) the
85 * Buffers do not have their own mutexes, rather they rely on the
86 * hash table mutexes for the bulk of their protection (i.e. most
87 * fields in the arc_buf_hdr_t are protected by these mutexes).
89 * buf_hash_find() returns the appropriate mutex (held) when it
90 * locates the requested buffer in the hash table. It returns
91 * NULL for the mutex if the buffer was not in the table.
93 * buf_hash_remove() expects the appropriate hash mutex to be
94 * already held before it is invoked.
96 * Each ARC state also has a mutex which is used to protect the
97 * buffer list associated with the state. When attempting to
98 * obtain a hash table lock while holding an ARC list lock you
99 * must use: mutex_tryenter() to avoid deadlock. Also note that
100 * the active state mutex must be held before the ghost state mutex.
102 * It as also possible to register a callback which is run when the
103 * arc_meta_limit is reached and no buffers can be safely evicted. In
104 * this case the arc user should drop a reference on some arc buffers so
105 * they can be reclaimed and the arc_meta_limit honored. For example,
106 * when using the ZPL each dentry holds a references on a znode. These
107 * dentries must be pruned before the arc buffer holding the znode can
110 * Note that the majority of the performance stats are manipulated
111 * with atomic operations.
113 * The L2ARC uses the l2ad_mtx on each vdev for the following:
115 * - L2ARC buflist creation
116 * - L2ARC buflist eviction
117 * - L2ARC write completion, which walks L2ARC buflists
118 * - ARC header destruction, as it removes from L2ARC buflists
119 * - ARC header release, as it removes from L2ARC buflists
125 * Every block that is in the ARC is tracked by an arc_buf_hdr_t structure.
126 * This structure can point either to a block that is still in the cache or to
127 * one that is only accessible in an L2 ARC device, or it can provide
128 * information about a block that was recently evicted. If a block is
129 * only accessible in the L2ARC, then the arc_buf_hdr_t only has enough
130 * information to retrieve it from the L2ARC device. This information is
131 * stored in the l2arc_buf_hdr_t sub-structure of the arc_buf_hdr_t. A block
132 * that is in this state cannot access the data directly.
134 * Blocks that are actively being referenced or have not been evicted
135 * are cached in the L1ARC. The L1ARC (l1arc_buf_hdr_t) is a structure within
136 * the arc_buf_hdr_t that will point to the data block in memory. A block can
137 * only be read by a consumer if it has an l1arc_buf_hdr_t. The L1ARC
138 * caches data in two ways -- in a list of ARC buffers (arc_buf_t) and
139 * also in the arc_buf_hdr_t's private physical data block pointer (b_pabd).
141 * The L1ARC's data pointer may or may not be uncompressed. The ARC has the
142 * ability to store the physical data (b_pabd) associated with the DVA of the
143 * arc_buf_hdr_t. Since the b_pabd is a copy of the on-disk physical block,
144 * it will match its on-disk compression characteristics. This behavior can be
145 * disabled by setting 'zfs_compressed_arc_enabled' to B_FALSE. When the
146 * compressed ARC functionality is disabled, the b_pabd will point to an
147 * uncompressed version of the on-disk data.
149 * Data in the L1ARC is not accessed by consumers of the ARC directly. Each
150 * arc_buf_hdr_t can have multiple ARC buffers (arc_buf_t) which reference it.
151 * Each ARC buffer (arc_buf_t) is being actively accessed by a specific ARC
152 * consumer. The ARC will provide references to this data and will keep it
153 * cached until it is no longer in use. The ARC caches only the L1ARC's physical
154 * data block and will evict any arc_buf_t that is no longer referenced. The
155 * amount of memory consumed by the arc_buf_ts' data buffers can be seen via the
156 * "overhead_size" kstat.
158 * Depending on the consumer, an arc_buf_t can be requested in uncompressed or
159 * compressed form. The typical case is that consumers will want uncompressed
160 * data, and when that happens a new data buffer is allocated where the data is
161 * decompressed for them to use. Currently the only consumer who wants
162 * compressed arc_buf_t's is "zfs send", when it streams data exactly as it
163 * exists on disk. When this happens, the arc_buf_t's data buffer is shared
164 * with the arc_buf_hdr_t.
166 * Here is a diagram showing an arc_buf_hdr_t referenced by two arc_buf_t's. The
167 * first one is owned by a compressed send consumer (and therefore references
168 * the same compressed data buffer as the arc_buf_hdr_t) and the second could be
169 * used by any other consumer (and has its own uncompressed copy of the data
184 * | b_buf +------------>+-----------+ arc_buf_t
185 * | b_pabd +-+ |b_next +---->+-----------+
186 * +-----------+ | |-----------| |b_next +-->NULL
187 * | |b_comp = T | +-----------+
188 * | |b_data +-+ |b_comp = F |
189 * | +-----------+ | |b_data +-+
190 * +->+------+ | +-----------+ |
192 * data | |<--------------+ | uncompressed
193 * +------+ compressed, | data
194 * shared +-->+------+
199 * When a consumer reads a block, the ARC must first look to see if the
200 * arc_buf_hdr_t is cached. If the hdr is cached then the ARC allocates a new
201 * arc_buf_t and either copies uncompressed data into a new data buffer from an
202 * existing uncompressed arc_buf_t, decompresses the hdr's b_pabd buffer into a
203 * new data buffer, or shares the hdr's b_pabd buffer, depending on whether the
204 * hdr is compressed and the desired compression characteristics of the
205 * arc_buf_t consumer. If the arc_buf_t ends up sharing data with the
206 * arc_buf_hdr_t and both of them are uncompressed then the arc_buf_t must be
207 * the last buffer in the hdr's b_buf list, however a shared compressed buf can
208 * be anywhere in the hdr's list.
210 * The diagram below shows an example of an uncompressed ARC hdr that is
211 * sharing its data with an arc_buf_t (note that the shared uncompressed buf is
212 * the last element in the buf list):
224 * | | arc_buf_t (shared)
225 * | b_buf +------------>+---------+ arc_buf_t
226 * | | |b_next +---->+---------+
227 * | b_pabd +-+ |---------| |b_next +-->NULL
228 * +-----------+ | | | +---------+
230 * | +---------+ | |b_data +-+
231 * +->+------+ | +---------+ |
233 * uncompressed | | | |
236 * | uncompressed | | |
239 * +---------------------------------+
241 * Writing to the ARC requires that the ARC first discard the hdr's b_pabd
242 * since the physical block is about to be rewritten. The new data contents
243 * will be contained in the arc_buf_t. As the I/O pipeline performs the write,
244 * it may compress the data before writing it to disk. The ARC will be called
245 * with the transformed data and will bcopy the transformed on-disk block into
246 * a newly allocated b_pabd. Writes are always done into buffers which have
247 * either been loaned (and hence are new and don't have other readers) or
248 * buffers which have been released (and hence have their own hdr, if there
249 * were originally other readers of the buf's original hdr). This ensures that
250 * the ARC only needs to update a single buf and its hdr after a write occurs.
252 * When the L2ARC is in use, it will also take advantage of the b_pabd. The
253 * L2ARC will always write the contents of b_pabd to the L2ARC. This means
254 * that when compressed ARC is enabled that the L2ARC blocks are identical
255 * to the on-disk block in the main data pool. This provides a significant
256 * advantage since the ARC can leverage the bp's checksum when reading from the
257 * L2ARC to determine if the contents are valid. However, if the compressed
258 * ARC is disabled, then the L2ARC's block must be transformed to look
259 * like the physical block in the main data pool before comparing the
260 * checksum and determining its validity.
262 * The L1ARC has a slightly different system for storing encrypted data.
263 * Raw (encrypted + possibly compressed) data has a few subtle differences from
264 * data that is just compressed. The biggest difference is that it is not
265 * possible to decrypt encrypted data (or visa versa) if the keys aren't loaded.
266 * The other difference is that encryption cannot be treated as a suggestion.
267 * If a caller would prefer compressed data, but they actually wind up with
268 * uncompressed data the worst thing that could happen is there might be a
269 * performance hit. If the caller requests encrypted data, however, we must be
270 * sure they actually get it or else secret information could be leaked. Raw
271 * data is stored in hdr->b_crypt_hdr.b_rabd. An encrypted header, therefore,
272 * may have both an encrypted version and a decrypted version of its data at
273 * once. When a caller needs a raw arc_buf_t, it is allocated and the data is
274 * copied out of this header. To avoid complications with b_pabd, raw buffers
280 #include <sys/spa_impl.h>
281 #include <sys/zio_compress.h>
282 #include <sys/zio_checksum.h>
283 #include <sys/zfs_context.h>
285 #include <sys/refcount.h>
286 #include <sys/vdev.h>
287 #include <sys/vdev_impl.h>
288 #include <sys/dsl_pool.h>
289 #include <sys/zio_checksum.h>
290 #include <sys/multilist.h>
293 #include <sys/fm/fs/zfs.h>
295 #include <sys/vmsystm.h>
297 #include <sys/fs/swapnode.h>
299 #include <linux/mm_compat.h>
301 #include <sys/callb.h>
302 #include <sys/kstat.h>
303 #include <sys/dmu_tx.h>
304 #include <zfs_fletcher.h>
305 #include <sys/arc_impl.h>
306 #include <sys/trace_arc.h>
309 /* set with ZFS_DEBUG=watch, to enable watchpoints on frozen buffers */
310 boolean_t arc_watch
= B_FALSE
;
313 static kmutex_t arc_reclaim_lock
;
314 static kcondvar_t arc_reclaim_thread_cv
;
315 static boolean_t arc_reclaim_thread_exit
;
316 static kcondvar_t arc_reclaim_waiters_cv
;
319 * The number of headers to evict in arc_evict_state_impl() before
320 * dropping the sublist lock and evicting from another sublist. A lower
321 * value means we're more likely to evict the "correct" header (i.e. the
322 * oldest header in the arc state), but comes with higher overhead
323 * (i.e. more invocations of arc_evict_state_impl()).
325 int zfs_arc_evict_batch_limit
= 10;
327 /* number of seconds before growing cache again */
328 static int arc_grow_retry
= 5;
330 /* shift of arc_c for calculating overflow limit in arc_get_data_impl */
331 int zfs_arc_overflow_shift
= 8;
333 /* shift of arc_c for calculating both min and max arc_p */
334 static int arc_p_min_shift
= 4;
336 /* log2(fraction of arc to reclaim) */
337 static int arc_shrink_shift
= 7;
339 /* percent of pagecache to reclaim arc to */
341 static uint_t zfs_arc_pc_percent
= 0;
345 * log2(fraction of ARC which must be free to allow growing).
346 * I.e. If there is less than arc_c >> arc_no_grow_shift free memory,
347 * when reading a new block into the ARC, we will evict an equal-sized block
350 * This must be less than arc_shrink_shift, so that when we shrink the ARC,
351 * we will still not allow it to grow.
353 int arc_no_grow_shift
= 5;
357 * minimum lifespan of a prefetch block in clock ticks
358 * (initialized in arc_init())
360 static int arc_min_prefetch_ms
;
361 static int arc_min_prescient_prefetch_ms
;
364 * If this percent of memory is free, don't throttle.
366 int arc_lotsfree_percent
= 10;
371 * The arc has filled available memory and has now warmed up.
373 static boolean_t arc_warm
;
376 * log2 fraction of the zio arena to keep free.
378 int arc_zio_arena_free_shift
= 2;
381 * These tunables are for performance analysis.
383 unsigned long zfs_arc_max
= 0;
384 unsigned long zfs_arc_min
= 0;
385 unsigned long zfs_arc_meta_limit
= 0;
386 unsigned long zfs_arc_meta_min
= 0;
387 unsigned long zfs_arc_dnode_limit
= 0;
388 unsigned long zfs_arc_dnode_reduce_percent
= 10;
389 int zfs_arc_grow_retry
= 0;
390 int zfs_arc_shrink_shift
= 0;
391 int zfs_arc_p_min_shift
= 0;
392 int zfs_arc_average_blocksize
= 8 * 1024; /* 8KB */
394 int zfs_compressed_arc_enabled
= B_TRUE
;
397 * ARC will evict meta buffers that exceed arc_meta_limit. This
398 * tunable make arc_meta_limit adjustable for different workloads.
400 unsigned long zfs_arc_meta_limit_percent
= 75;
403 * Percentage that can be consumed by dnodes of ARC meta buffers.
405 unsigned long zfs_arc_dnode_limit_percent
= 10;
408 * These tunables are Linux specific
410 unsigned long zfs_arc_sys_free
= 0;
411 int zfs_arc_min_prefetch_ms
= 0;
412 int zfs_arc_min_prescient_prefetch_ms
= 0;
413 int zfs_arc_p_aggressive_disable
= 1;
414 int zfs_arc_p_dampener_disable
= 1;
415 int zfs_arc_meta_prune
= 10000;
416 int zfs_arc_meta_strategy
= ARC_STRATEGY_META_BALANCED
;
417 int zfs_arc_meta_adjust_restarts
= 4096;
418 int zfs_arc_lotsfree_percent
= 10;
421 static arc_state_t ARC_anon
;
422 static arc_state_t ARC_mru
;
423 static arc_state_t ARC_mru_ghost
;
424 static arc_state_t ARC_mfu
;
425 static arc_state_t ARC_mfu_ghost
;
426 static arc_state_t ARC_l2c_only
;
428 typedef struct arc_stats
{
429 kstat_named_t arcstat_hits
;
430 kstat_named_t arcstat_misses
;
431 kstat_named_t arcstat_demand_data_hits
;
432 kstat_named_t arcstat_demand_data_misses
;
433 kstat_named_t arcstat_demand_metadata_hits
;
434 kstat_named_t arcstat_demand_metadata_misses
;
435 kstat_named_t arcstat_prefetch_data_hits
;
436 kstat_named_t arcstat_prefetch_data_misses
;
437 kstat_named_t arcstat_prefetch_metadata_hits
;
438 kstat_named_t arcstat_prefetch_metadata_misses
;
439 kstat_named_t arcstat_mru_hits
;
440 kstat_named_t arcstat_mru_ghost_hits
;
441 kstat_named_t arcstat_mfu_hits
;
442 kstat_named_t arcstat_mfu_ghost_hits
;
443 kstat_named_t arcstat_deleted
;
445 * Number of buffers that could not be evicted because the hash lock
446 * was held by another thread. The lock may not necessarily be held
447 * by something using the same buffer, since hash locks are shared
448 * by multiple buffers.
450 kstat_named_t arcstat_mutex_miss
;
452 * Number of buffers skipped because they have I/O in progress, are
453 * indrect prefetch buffers that have not lived long enough, or are
454 * not from the spa we're trying to evict from.
456 kstat_named_t arcstat_evict_skip
;
458 * Number of times arc_evict_state() was unable to evict enough
459 * buffers to reach its target amount.
461 kstat_named_t arcstat_evict_not_enough
;
462 kstat_named_t arcstat_evict_l2_cached
;
463 kstat_named_t arcstat_evict_l2_eligible
;
464 kstat_named_t arcstat_evict_l2_ineligible
;
465 kstat_named_t arcstat_evict_l2_skip
;
466 kstat_named_t arcstat_hash_elements
;
467 kstat_named_t arcstat_hash_elements_max
;
468 kstat_named_t arcstat_hash_collisions
;
469 kstat_named_t arcstat_hash_chains
;
470 kstat_named_t arcstat_hash_chain_max
;
471 kstat_named_t arcstat_p
;
472 kstat_named_t arcstat_c
;
473 kstat_named_t arcstat_c_min
;
474 kstat_named_t arcstat_c_max
;
475 kstat_named_t arcstat_size
;
477 * Number of compressed bytes stored in the arc_buf_hdr_t's b_pabd.
478 * Note that the compressed bytes may match the uncompressed bytes
479 * if the block is either not compressed or compressed arc is disabled.
481 kstat_named_t arcstat_compressed_size
;
483 * Uncompressed size of the data stored in b_pabd. If compressed
484 * arc is disabled then this value will be identical to the stat
487 kstat_named_t arcstat_uncompressed_size
;
489 * Number of bytes stored in all the arc_buf_t's. This is classified
490 * as "overhead" since this data is typically short-lived and will
491 * be evicted from the arc when it becomes unreferenced unless the
492 * zfs_keep_uncompressed_metadata or zfs_keep_uncompressed_level
493 * values have been set (see comment in dbuf.c for more information).
495 kstat_named_t arcstat_overhead_size
;
497 * Number of bytes consumed by internal ARC structures necessary
498 * for tracking purposes; these structures are not actually
499 * backed by ARC buffers. This includes arc_buf_hdr_t structures
500 * (allocated via arc_buf_hdr_t_full and arc_buf_hdr_t_l2only
501 * caches), and arc_buf_t structures (allocated via arc_buf_t
504 kstat_named_t arcstat_hdr_size
;
506 * Number of bytes consumed by ARC buffers of type equal to
507 * ARC_BUFC_DATA. This is generally consumed by buffers backing
508 * on disk user data (e.g. plain file contents).
510 kstat_named_t arcstat_data_size
;
512 * Number of bytes consumed by ARC buffers of type equal to
513 * ARC_BUFC_METADATA. This is generally consumed by buffers
514 * backing on disk data that is used for internal ZFS
515 * structures (e.g. ZAP, dnode, indirect blocks, etc).
517 kstat_named_t arcstat_metadata_size
;
519 * Number of bytes consumed by dmu_buf_impl_t objects.
521 kstat_named_t arcstat_dbuf_size
;
523 * Number of bytes consumed by dnode_t objects.
525 kstat_named_t arcstat_dnode_size
;
527 * Number of bytes consumed by bonus buffers.
529 kstat_named_t arcstat_bonus_size
;
531 * Total number of bytes consumed by ARC buffers residing in the
532 * arc_anon state. This includes *all* buffers in the arc_anon
533 * state; e.g. data, metadata, evictable, and unevictable buffers
534 * are all included in this value.
536 kstat_named_t arcstat_anon_size
;
538 * Number of bytes consumed by ARC buffers that meet the
539 * following criteria: backing buffers of type ARC_BUFC_DATA,
540 * residing in the arc_anon state, and are eligible for eviction
541 * (e.g. have no outstanding holds on the buffer).
543 kstat_named_t arcstat_anon_evictable_data
;
545 * Number of bytes consumed by ARC buffers that meet the
546 * following criteria: backing buffers of type ARC_BUFC_METADATA,
547 * residing in the arc_anon state, and are eligible for eviction
548 * (e.g. have no outstanding holds on the buffer).
550 kstat_named_t arcstat_anon_evictable_metadata
;
552 * Total number of bytes consumed by ARC buffers residing in the
553 * arc_mru state. This includes *all* buffers in the arc_mru
554 * state; e.g. data, metadata, evictable, and unevictable buffers
555 * are all included in this value.
557 kstat_named_t arcstat_mru_size
;
559 * Number of bytes consumed by ARC buffers that meet the
560 * following criteria: backing buffers of type ARC_BUFC_DATA,
561 * residing in the arc_mru state, and are eligible for eviction
562 * (e.g. have no outstanding holds on the buffer).
564 kstat_named_t arcstat_mru_evictable_data
;
566 * Number of bytes consumed by ARC buffers that meet the
567 * following criteria: backing buffers of type ARC_BUFC_METADATA,
568 * residing in the arc_mru state, and are eligible for eviction
569 * (e.g. have no outstanding holds on the buffer).
571 kstat_named_t arcstat_mru_evictable_metadata
;
573 * Total number of bytes that *would have been* consumed by ARC
574 * buffers in the arc_mru_ghost state. The key thing to note
575 * here, is the fact that this size doesn't actually indicate
576 * RAM consumption. The ghost lists only consist of headers and
577 * don't actually have ARC buffers linked off of these headers.
578 * Thus, *if* the headers had associated ARC buffers, these
579 * buffers *would have* consumed this number of bytes.
581 kstat_named_t arcstat_mru_ghost_size
;
583 * Number of bytes that *would have been* consumed by ARC
584 * buffers that are eligible for eviction, of type
585 * ARC_BUFC_DATA, and linked off the arc_mru_ghost state.
587 kstat_named_t arcstat_mru_ghost_evictable_data
;
589 * Number of bytes that *would have been* consumed by ARC
590 * buffers that are eligible for eviction, of type
591 * ARC_BUFC_METADATA, and linked off the arc_mru_ghost state.
593 kstat_named_t arcstat_mru_ghost_evictable_metadata
;
595 * Total number of bytes consumed by ARC buffers residing in the
596 * arc_mfu state. This includes *all* buffers in the arc_mfu
597 * state; e.g. data, metadata, evictable, and unevictable buffers
598 * are all included in this value.
600 kstat_named_t arcstat_mfu_size
;
602 * Number of bytes consumed by ARC buffers that are eligible for
603 * eviction, of type ARC_BUFC_DATA, and reside in the arc_mfu
606 kstat_named_t arcstat_mfu_evictable_data
;
608 * Number of bytes consumed by ARC buffers that are eligible for
609 * eviction, of type ARC_BUFC_METADATA, and reside in the
612 kstat_named_t arcstat_mfu_evictable_metadata
;
614 * Total number of bytes that *would have been* consumed by ARC
615 * buffers in the arc_mfu_ghost state. See the comment above
616 * arcstat_mru_ghost_size for more details.
618 kstat_named_t arcstat_mfu_ghost_size
;
620 * Number of bytes that *would have been* consumed by ARC
621 * buffers that are eligible for eviction, of type
622 * ARC_BUFC_DATA, and linked off the arc_mfu_ghost state.
624 kstat_named_t arcstat_mfu_ghost_evictable_data
;
626 * Number of bytes that *would have been* consumed by ARC
627 * buffers that are eligible for eviction, of type
628 * ARC_BUFC_METADATA, and linked off the arc_mru_ghost state.
630 kstat_named_t arcstat_mfu_ghost_evictable_metadata
;
631 kstat_named_t arcstat_l2_hits
;
632 kstat_named_t arcstat_l2_misses
;
633 kstat_named_t arcstat_l2_feeds
;
634 kstat_named_t arcstat_l2_rw_clash
;
635 kstat_named_t arcstat_l2_read_bytes
;
636 kstat_named_t arcstat_l2_write_bytes
;
637 kstat_named_t arcstat_l2_writes_sent
;
638 kstat_named_t arcstat_l2_writes_done
;
639 kstat_named_t arcstat_l2_writes_error
;
640 kstat_named_t arcstat_l2_writes_lock_retry
;
641 kstat_named_t arcstat_l2_evict_lock_retry
;
642 kstat_named_t arcstat_l2_evict_reading
;
643 kstat_named_t arcstat_l2_evict_l1cached
;
644 kstat_named_t arcstat_l2_free_on_write
;
645 kstat_named_t arcstat_l2_abort_lowmem
;
646 kstat_named_t arcstat_l2_cksum_bad
;
647 kstat_named_t arcstat_l2_io_error
;
648 kstat_named_t arcstat_l2_lsize
;
649 kstat_named_t arcstat_l2_psize
;
650 kstat_named_t arcstat_l2_hdr_size
;
651 kstat_named_t arcstat_memory_throttle_count
;
652 kstat_named_t arcstat_memory_direct_count
;
653 kstat_named_t arcstat_memory_indirect_count
;
654 kstat_named_t arcstat_memory_all_bytes
;
655 kstat_named_t arcstat_memory_free_bytes
;
656 kstat_named_t arcstat_memory_available_bytes
;
657 kstat_named_t arcstat_no_grow
;
658 kstat_named_t arcstat_tempreserve
;
659 kstat_named_t arcstat_loaned_bytes
;
660 kstat_named_t arcstat_prune
;
661 kstat_named_t arcstat_meta_used
;
662 kstat_named_t arcstat_meta_limit
;
663 kstat_named_t arcstat_dnode_limit
;
664 kstat_named_t arcstat_meta_max
;
665 kstat_named_t arcstat_meta_min
;
666 kstat_named_t arcstat_async_upgrade_sync
;
667 kstat_named_t arcstat_demand_hit_predictive_prefetch
;
668 kstat_named_t arcstat_demand_hit_prescient_prefetch
;
669 kstat_named_t arcstat_need_free
;
670 kstat_named_t arcstat_sys_free
;
671 kstat_named_t arcstat_raw_size
;
674 static arc_stats_t arc_stats
= {
675 { "hits", KSTAT_DATA_UINT64
},
676 { "misses", KSTAT_DATA_UINT64
},
677 { "demand_data_hits", KSTAT_DATA_UINT64
},
678 { "demand_data_misses", KSTAT_DATA_UINT64
},
679 { "demand_metadata_hits", KSTAT_DATA_UINT64
},
680 { "demand_metadata_misses", KSTAT_DATA_UINT64
},
681 { "prefetch_data_hits", KSTAT_DATA_UINT64
},
682 { "prefetch_data_misses", KSTAT_DATA_UINT64
},
683 { "prefetch_metadata_hits", KSTAT_DATA_UINT64
},
684 { "prefetch_metadata_misses", KSTAT_DATA_UINT64
},
685 { "mru_hits", KSTAT_DATA_UINT64
},
686 { "mru_ghost_hits", KSTAT_DATA_UINT64
},
687 { "mfu_hits", KSTAT_DATA_UINT64
},
688 { "mfu_ghost_hits", KSTAT_DATA_UINT64
},
689 { "deleted", KSTAT_DATA_UINT64
},
690 { "mutex_miss", KSTAT_DATA_UINT64
},
691 { "evict_skip", KSTAT_DATA_UINT64
},
692 { "evict_not_enough", KSTAT_DATA_UINT64
},
693 { "evict_l2_cached", KSTAT_DATA_UINT64
},
694 { "evict_l2_eligible", KSTAT_DATA_UINT64
},
695 { "evict_l2_ineligible", KSTAT_DATA_UINT64
},
696 { "evict_l2_skip", KSTAT_DATA_UINT64
},
697 { "hash_elements", KSTAT_DATA_UINT64
},
698 { "hash_elements_max", KSTAT_DATA_UINT64
},
699 { "hash_collisions", KSTAT_DATA_UINT64
},
700 { "hash_chains", KSTAT_DATA_UINT64
},
701 { "hash_chain_max", KSTAT_DATA_UINT64
},
702 { "p", KSTAT_DATA_UINT64
},
703 { "c", KSTAT_DATA_UINT64
},
704 { "c_min", KSTAT_DATA_UINT64
},
705 { "c_max", KSTAT_DATA_UINT64
},
706 { "size", KSTAT_DATA_UINT64
},
707 { "compressed_size", KSTAT_DATA_UINT64
},
708 { "uncompressed_size", KSTAT_DATA_UINT64
},
709 { "overhead_size", KSTAT_DATA_UINT64
},
710 { "hdr_size", KSTAT_DATA_UINT64
},
711 { "data_size", KSTAT_DATA_UINT64
},
712 { "metadata_size", KSTAT_DATA_UINT64
},
713 { "dbuf_size", KSTAT_DATA_UINT64
},
714 { "dnode_size", KSTAT_DATA_UINT64
},
715 { "bonus_size", KSTAT_DATA_UINT64
},
716 { "anon_size", KSTAT_DATA_UINT64
},
717 { "anon_evictable_data", KSTAT_DATA_UINT64
},
718 { "anon_evictable_metadata", KSTAT_DATA_UINT64
},
719 { "mru_size", KSTAT_DATA_UINT64
},
720 { "mru_evictable_data", KSTAT_DATA_UINT64
},
721 { "mru_evictable_metadata", KSTAT_DATA_UINT64
},
722 { "mru_ghost_size", KSTAT_DATA_UINT64
},
723 { "mru_ghost_evictable_data", KSTAT_DATA_UINT64
},
724 { "mru_ghost_evictable_metadata", KSTAT_DATA_UINT64
},
725 { "mfu_size", KSTAT_DATA_UINT64
},
726 { "mfu_evictable_data", KSTAT_DATA_UINT64
},
727 { "mfu_evictable_metadata", KSTAT_DATA_UINT64
},
728 { "mfu_ghost_size", KSTAT_DATA_UINT64
},
729 { "mfu_ghost_evictable_data", KSTAT_DATA_UINT64
},
730 { "mfu_ghost_evictable_metadata", KSTAT_DATA_UINT64
},
731 { "l2_hits", KSTAT_DATA_UINT64
},
732 { "l2_misses", KSTAT_DATA_UINT64
},
733 { "l2_feeds", KSTAT_DATA_UINT64
},
734 { "l2_rw_clash", KSTAT_DATA_UINT64
},
735 { "l2_read_bytes", KSTAT_DATA_UINT64
},
736 { "l2_write_bytes", KSTAT_DATA_UINT64
},
737 { "l2_writes_sent", KSTAT_DATA_UINT64
},
738 { "l2_writes_done", KSTAT_DATA_UINT64
},
739 { "l2_writes_error", KSTAT_DATA_UINT64
},
740 { "l2_writes_lock_retry", KSTAT_DATA_UINT64
},
741 { "l2_evict_lock_retry", KSTAT_DATA_UINT64
},
742 { "l2_evict_reading", KSTAT_DATA_UINT64
},
743 { "l2_evict_l1cached", KSTAT_DATA_UINT64
},
744 { "l2_free_on_write", KSTAT_DATA_UINT64
},
745 { "l2_abort_lowmem", KSTAT_DATA_UINT64
},
746 { "l2_cksum_bad", KSTAT_DATA_UINT64
},
747 { "l2_io_error", KSTAT_DATA_UINT64
},
748 { "l2_size", KSTAT_DATA_UINT64
},
749 { "l2_asize", KSTAT_DATA_UINT64
},
750 { "l2_hdr_size", KSTAT_DATA_UINT64
},
751 { "memory_throttle_count", KSTAT_DATA_UINT64
},
752 { "memory_direct_count", KSTAT_DATA_UINT64
},
753 { "memory_indirect_count", KSTAT_DATA_UINT64
},
754 { "memory_all_bytes", KSTAT_DATA_UINT64
},
755 { "memory_free_bytes", KSTAT_DATA_UINT64
},
756 { "memory_available_bytes", KSTAT_DATA_INT64
},
757 { "arc_no_grow", KSTAT_DATA_UINT64
},
758 { "arc_tempreserve", KSTAT_DATA_UINT64
},
759 { "arc_loaned_bytes", KSTAT_DATA_UINT64
},
760 { "arc_prune", KSTAT_DATA_UINT64
},
761 { "arc_meta_used", KSTAT_DATA_UINT64
},
762 { "arc_meta_limit", KSTAT_DATA_UINT64
},
763 { "arc_dnode_limit", KSTAT_DATA_UINT64
},
764 { "arc_meta_max", KSTAT_DATA_UINT64
},
765 { "arc_meta_min", KSTAT_DATA_UINT64
},
766 { "async_upgrade_sync", KSTAT_DATA_UINT64
},
767 { "demand_hit_predictive_prefetch", KSTAT_DATA_UINT64
},
768 { "demand_hit_prescient_prefetch", KSTAT_DATA_UINT64
},
769 { "arc_need_free", KSTAT_DATA_UINT64
},
770 { "arc_sys_free", KSTAT_DATA_UINT64
},
771 { "arc_raw_size", KSTAT_DATA_UINT64
}
774 #define ARCSTAT(stat) (arc_stats.stat.value.ui64)
776 #define ARCSTAT_INCR(stat, val) \
777 atomic_add_64(&arc_stats.stat.value.ui64, (val))
779 #define ARCSTAT_BUMP(stat) ARCSTAT_INCR(stat, 1)
780 #define ARCSTAT_BUMPDOWN(stat) ARCSTAT_INCR(stat, -1)
782 #define ARCSTAT_MAX(stat, val) { \
784 while ((val) > (m = arc_stats.stat.value.ui64) && \
785 (m != atomic_cas_64(&arc_stats.stat.value.ui64, m, (val)))) \
789 #define ARCSTAT_MAXSTAT(stat) \
790 ARCSTAT_MAX(stat##_max, arc_stats.stat.value.ui64)
793 * We define a macro to allow ARC hits/misses to be easily broken down by
794 * two separate conditions, giving a total of four different subtypes for
795 * each of hits and misses (so eight statistics total).
797 #define ARCSTAT_CONDSTAT(cond1, stat1, notstat1, cond2, stat2, notstat2, stat) \
800 ARCSTAT_BUMP(arcstat_##stat1##_##stat2##_##stat); \
802 ARCSTAT_BUMP(arcstat_##stat1##_##notstat2##_##stat); \
806 ARCSTAT_BUMP(arcstat_##notstat1##_##stat2##_##stat); \
808 ARCSTAT_BUMP(arcstat_##notstat1##_##notstat2##_##stat);\
813 static arc_state_t
*arc_anon
;
814 static arc_state_t
*arc_mru
;
815 static arc_state_t
*arc_mru_ghost
;
816 static arc_state_t
*arc_mfu
;
817 static arc_state_t
*arc_mfu_ghost
;
818 static arc_state_t
*arc_l2c_only
;
821 * There are several ARC variables that are critical to export as kstats --
822 * but we don't want to have to grovel around in the kstat whenever we wish to
823 * manipulate them. For these variables, we therefore define them to be in
824 * terms of the statistic variable. This assures that we are not introducing
825 * the possibility of inconsistency by having shadow copies of the variables,
826 * while still allowing the code to be readable.
828 #define arc_size ARCSTAT(arcstat_size) /* actual total arc size */
829 #define arc_p ARCSTAT(arcstat_p) /* target size of MRU */
830 #define arc_c ARCSTAT(arcstat_c) /* target size of cache */
831 #define arc_c_min ARCSTAT(arcstat_c_min) /* min target cache size */
832 #define arc_c_max ARCSTAT(arcstat_c_max) /* max target cache size */
833 #define arc_no_grow ARCSTAT(arcstat_no_grow) /* do not grow cache size */
834 #define arc_tempreserve ARCSTAT(arcstat_tempreserve)
835 #define arc_loaned_bytes ARCSTAT(arcstat_loaned_bytes)
836 #define arc_meta_limit ARCSTAT(arcstat_meta_limit) /* max size for metadata */
837 #define arc_dnode_limit ARCSTAT(arcstat_dnode_limit) /* max size for dnodes */
838 #define arc_meta_min ARCSTAT(arcstat_meta_min) /* min size for metadata */
839 #define arc_meta_used ARCSTAT(arcstat_meta_used) /* size of metadata */
840 #define arc_meta_max ARCSTAT(arcstat_meta_max) /* max size of metadata */
841 #define arc_dbuf_size ARCSTAT(arcstat_dbuf_size) /* dbuf metadata */
842 #define arc_dnode_size ARCSTAT(arcstat_dnode_size) /* dnode metadata */
843 #define arc_bonus_size ARCSTAT(arcstat_bonus_size) /* bonus buffer metadata */
844 #define arc_need_free ARCSTAT(arcstat_need_free) /* bytes to be freed */
845 #define arc_sys_free ARCSTAT(arcstat_sys_free) /* target system free bytes */
847 /* size of all b_rabd's in entire arc */
848 #define arc_raw_size ARCSTAT(arcstat_raw_size)
849 /* compressed size of entire arc */
850 #define arc_compressed_size ARCSTAT(arcstat_compressed_size)
851 /* uncompressed size of entire arc */
852 #define arc_uncompressed_size ARCSTAT(arcstat_uncompressed_size)
853 /* number of bytes in the arc from arc_buf_t's */
854 #define arc_overhead_size ARCSTAT(arcstat_overhead_size)
856 static list_t arc_prune_list
;
857 static kmutex_t arc_prune_mtx
;
858 static taskq_t
*arc_prune_taskq
;
860 #define GHOST_STATE(state) \
861 ((state) == arc_mru_ghost || (state) == arc_mfu_ghost || \
862 (state) == arc_l2c_only)
864 #define HDR_IN_HASH_TABLE(hdr) ((hdr)->b_flags & ARC_FLAG_IN_HASH_TABLE)
865 #define HDR_IO_IN_PROGRESS(hdr) ((hdr)->b_flags & ARC_FLAG_IO_IN_PROGRESS)
866 #define HDR_IO_ERROR(hdr) ((hdr)->b_flags & ARC_FLAG_IO_ERROR)
867 #define HDR_PREFETCH(hdr) ((hdr)->b_flags & ARC_FLAG_PREFETCH)
868 #define HDR_PRESCIENT_PREFETCH(hdr) \
869 ((hdr)->b_flags & ARC_FLAG_PRESCIENT_PREFETCH)
870 #define HDR_COMPRESSION_ENABLED(hdr) \
871 ((hdr)->b_flags & ARC_FLAG_COMPRESSED_ARC)
873 #define HDR_L2CACHE(hdr) ((hdr)->b_flags & ARC_FLAG_L2CACHE)
874 #define HDR_L2_READING(hdr) \
875 (((hdr)->b_flags & ARC_FLAG_IO_IN_PROGRESS) && \
876 ((hdr)->b_flags & ARC_FLAG_HAS_L2HDR))
877 #define HDR_L2_WRITING(hdr) ((hdr)->b_flags & ARC_FLAG_L2_WRITING)
878 #define HDR_L2_EVICTED(hdr) ((hdr)->b_flags & ARC_FLAG_L2_EVICTED)
879 #define HDR_L2_WRITE_HEAD(hdr) ((hdr)->b_flags & ARC_FLAG_L2_WRITE_HEAD)
880 #define HDR_PROTECTED(hdr) ((hdr)->b_flags & ARC_FLAG_PROTECTED)
881 #define HDR_NOAUTH(hdr) ((hdr)->b_flags & ARC_FLAG_NOAUTH)
882 #define HDR_SHARED_DATA(hdr) ((hdr)->b_flags & ARC_FLAG_SHARED_DATA)
884 #define HDR_ISTYPE_METADATA(hdr) \
885 ((hdr)->b_flags & ARC_FLAG_BUFC_METADATA)
886 #define HDR_ISTYPE_DATA(hdr) (!HDR_ISTYPE_METADATA(hdr))
888 #define HDR_HAS_L1HDR(hdr) ((hdr)->b_flags & ARC_FLAG_HAS_L1HDR)
889 #define HDR_HAS_L2HDR(hdr) ((hdr)->b_flags & ARC_FLAG_HAS_L2HDR)
890 #define HDR_HAS_RABD(hdr) \
891 (HDR_HAS_L1HDR(hdr) && HDR_PROTECTED(hdr) && \
892 (hdr)->b_crypt_hdr.b_rabd != NULL)
893 #define HDR_ENCRYPTED(hdr) \
894 (HDR_PROTECTED(hdr) && DMU_OT_IS_ENCRYPTED((hdr)->b_crypt_hdr.b_ot))
895 #define HDR_AUTHENTICATED(hdr) \
896 (HDR_PROTECTED(hdr) && !DMU_OT_IS_ENCRYPTED((hdr)->b_crypt_hdr.b_ot))
898 /* For storing compression mode in b_flags */
899 #define HDR_COMPRESS_OFFSET (highbit64(ARC_FLAG_COMPRESS_0) - 1)
901 #define HDR_GET_COMPRESS(hdr) ((enum zio_compress)BF32_GET((hdr)->b_flags, \
902 HDR_COMPRESS_OFFSET, SPA_COMPRESSBITS))
903 #define HDR_SET_COMPRESS(hdr, cmp) BF32_SET((hdr)->b_flags, \
904 HDR_COMPRESS_OFFSET, SPA_COMPRESSBITS, (cmp));
906 #define ARC_BUF_LAST(buf) ((buf)->b_next == NULL)
907 #define ARC_BUF_SHARED(buf) ((buf)->b_flags & ARC_BUF_FLAG_SHARED)
908 #define ARC_BUF_COMPRESSED(buf) ((buf)->b_flags & ARC_BUF_FLAG_COMPRESSED)
909 #define ARC_BUF_ENCRYPTED(buf) ((buf)->b_flags & ARC_BUF_FLAG_ENCRYPTED)
915 #define HDR_FULL_CRYPT_SIZE ((int64_t)sizeof (arc_buf_hdr_t))
916 #define HDR_FULL_SIZE ((int64_t)offsetof(arc_buf_hdr_t, b_crypt_hdr))
917 #define HDR_L2ONLY_SIZE ((int64_t)offsetof(arc_buf_hdr_t, b_l1hdr))
920 * Hash table routines
923 #define HT_LOCK_ALIGN 64
924 #define HT_LOCK_PAD (P2NPHASE(sizeof (kmutex_t), (HT_LOCK_ALIGN)))
929 unsigned char pad
[HT_LOCK_PAD
];
933 #define BUF_LOCKS 8192
934 typedef struct buf_hash_table
{
936 arc_buf_hdr_t
**ht_table
;
937 struct ht_lock ht_locks
[BUF_LOCKS
];
940 static buf_hash_table_t buf_hash_table
;
942 #define BUF_HASH_INDEX(spa, dva, birth) \
943 (buf_hash(spa, dva, birth) & buf_hash_table.ht_mask)
944 #define BUF_HASH_LOCK_NTRY(idx) (buf_hash_table.ht_locks[idx & (BUF_LOCKS-1)])
945 #define BUF_HASH_LOCK(idx) (&(BUF_HASH_LOCK_NTRY(idx).ht_lock))
946 #define HDR_LOCK(hdr) \
947 (BUF_HASH_LOCK(BUF_HASH_INDEX(hdr->b_spa, &hdr->b_dva, hdr->b_birth)))
949 uint64_t zfs_crc64_table
[256];
955 #define L2ARC_WRITE_SIZE (8 * 1024 * 1024) /* initial write max */
956 #define L2ARC_HEADROOM 2 /* num of writes */
959 * If we discover during ARC scan any buffers to be compressed, we boost
960 * our headroom for the next scanning cycle by this percentage multiple.
962 #define L2ARC_HEADROOM_BOOST 200
963 #define L2ARC_FEED_SECS 1 /* caching interval secs */
964 #define L2ARC_FEED_MIN_MS 200 /* min caching interval ms */
967 * We can feed L2ARC from two states of ARC buffers, mru and mfu,
968 * and each of the state has two types: data and metadata.
970 #define L2ARC_FEED_TYPES 4
972 #define l2arc_writes_sent ARCSTAT(arcstat_l2_writes_sent)
973 #define l2arc_writes_done ARCSTAT(arcstat_l2_writes_done)
975 /* L2ARC Performance Tunables */
976 unsigned long l2arc_write_max
= L2ARC_WRITE_SIZE
; /* def max write size */
977 unsigned long l2arc_write_boost
= L2ARC_WRITE_SIZE
; /* extra warmup write */
978 unsigned long l2arc_headroom
= L2ARC_HEADROOM
; /* # of dev writes */
979 unsigned long l2arc_headroom_boost
= L2ARC_HEADROOM_BOOST
;
980 unsigned long l2arc_feed_secs
= L2ARC_FEED_SECS
; /* interval seconds */
981 unsigned long l2arc_feed_min_ms
= L2ARC_FEED_MIN_MS
; /* min interval msecs */
982 int l2arc_noprefetch
= B_TRUE
; /* don't cache prefetch bufs */
983 int l2arc_feed_again
= B_TRUE
; /* turbo warmup */
984 int l2arc_norw
= B_FALSE
; /* no reads during writes */
989 static list_t L2ARC_dev_list
; /* device list */
990 static list_t
*l2arc_dev_list
; /* device list pointer */
991 static kmutex_t l2arc_dev_mtx
; /* device list mutex */
992 static l2arc_dev_t
*l2arc_dev_last
; /* last device used */
993 static list_t L2ARC_free_on_write
; /* free after write buf list */
994 static list_t
*l2arc_free_on_write
; /* free after write list ptr */
995 static kmutex_t l2arc_free_on_write_mtx
; /* mutex for list */
996 static uint64_t l2arc_ndev
; /* number of devices */
998 typedef struct l2arc_read_callback
{
999 arc_buf_hdr_t
*l2rcb_hdr
; /* read header */
1000 blkptr_t l2rcb_bp
; /* original blkptr */
1001 zbookmark_phys_t l2rcb_zb
; /* original bookmark */
1002 int l2rcb_flags
; /* original flags */
1003 abd_t
*l2rcb_abd
; /* temporary buffer */
1004 } l2arc_read_callback_t
;
1006 typedef struct l2arc_data_free
{
1007 /* protected by l2arc_free_on_write_mtx */
1010 arc_buf_contents_t l2df_type
;
1011 list_node_t l2df_list_node
;
1012 } l2arc_data_free_t
;
1014 typedef enum arc_fill_flags
{
1015 ARC_FILL_LOCKED
= 1 << 0, /* hdr lock is held */
1016 ARC_FILL_COMPRESSED
= 1 << 1, /* fill with compressed data */
1017 ARC_FILL_ENCRYPTED
= 1 << 2, /* fill with encrypted data */
1018 ARC_FILL_NOAUTH
= 1 << 3, /* don't attempt to authenticate */
1019 ARC_FILL_IN_PLACE
= 1 << 4 /* fill in place (special case) */
1022 static kmutex_t l2arc_feed_thr_lock
;
1023 static kcondvar_t l2arc_feed_thr_cv
;
1024 static uint8_t l2arc_thread_exit
;
1026 static abd_t
*arc_get_data_abd(arc_buf_hdr_t
*, uint64_t, void *);
1027 static void *arc_get_data_buf(arc_buf_hdr_t
*, uint64_t, void *);
1028 static void arc_get_data_impl(arc_buf_hdr_t
*, uint64_t, void *);
1029 static void arc_free_data_abd(arc_buf_hdr_t
*, abd_t
*, uint64_t, void *);
1030 static void arc_free_data_buf(arc_buf_hdr_t
*, void *, uint64_t, void *);
1031 static void arc_free_data_impl(arc_buf_hdr_t
*hdr
, uint64_t size
, void *tag
);
1032 static void arc_hdr_free_abd(arc_buf_hdr_t
*, boolean_t
);
1033 static void arc_hdr_alloc_abd(arc_buf_hdr_t
*, boolean_t
);
1034 static void arc_access(arc_buf_hdr_t
*, kmutex_t
*);
1035 static boolean_t
arc_is_overflowing(void);
1036 static void arc_buf_watch(arc_buf_t
*);
1037 static void arc_tuning_update(void);
1038 static void arc_prune_async(int64_t);
1039 static uint64_t arc_all_memory(void);
1041 static arc_buf_contents_t
arc_buf_type(arc_buf_hdr_t
*);
1042 static uint32_t arc_bufc_to_flags(arc_buf_contents_t
);
1043 static inline void arc_hdr_set_flags(arc_buf_hdr_t
*hdr
, arc_flags_t flags
);
1044 static inline void arc_hdr_clear_flags(arc_buf_hdr_t
*hdr
, arc_flags_t flags
);
1046 static boolean_t
l2arc_write_eligible(uint64_t, arc_buf_hdr_t
*);
1047 static void l2arc_read_done(zio_t
*);
1050 buf_hash(uint64_t spa
, const dva_t
*dva
, uint64_t birth
)
1052 uint8_t *vdva
= (uint8_t *)dva
;
1053 uint64_t crc
= -1ULL;
1056 ASSERT(zfs_crc64_table
[128] == ZFS_CRC64_POLY
);
1058 for (i
= 0; i
< sizeof (dva_t
); i
++)
1059 crc
= (crc
>> 8) ^ zfs_crc64_table
[(crc
^ vdva
[i
]) & 0xFF];
1061 crc
^= (spa
>>8) ^ birth
;
1066 #define HDR_EMPTY(hdr) \
1067 ((hdr)->b_dva.dva_word[0] == 0 && \
1068 (hdr)->b_dva.dva_word[1] == 0)
1070 #define HDR_EQUAL(spa, dva, birth, hdr) \
1071 ((hdr)->b_dva.dva_word[0] == (dva)->dva_word[0]) && \
1072 ((hdr)->b_dva.dva_word[1] == (dva)->dva_word[1]) && \
1073 ((hdr)->b_birth == birth) && ((hdr)->b_spa == spa)
1076 buf_discard_identity(arc_buf_hdr_t
*hdr
)
1078 hdr
->b_dva
.dva_word
[0] = 0;
1079 hdr
->b_dva
.dva_word
[1] = 0;
1083 static arc_buf_hdr_t
*
1084 buf_hash_find(uint64_t spa
, const blkptr_t
*bp
, kmutex_t
**lockp
)
1086 const dva_t
*dva
= BP_IDENTITY(bp
);
1087 uint64_t birth
= BP_PHYSICAL_BIRTH(bp
);
1088 uint64_t idx
= BUF_HASH_INDEX(spa
, dva
, birth
);
1089 kmutex_t
*hash_lock
= BUF_HASH_LOCK(idx
);
1092 mutex_enter(hash_lock
);
1093 for (hdr
= buf_hash_table
.ht_table
[idx
]; hdr
!= NULL
;
1094 hdr
= hdr
->b_hash_next
) {
1095 if (HDR_EQUAL(spa
, dva
, birth
, hdr
)) {
1100 mutex_exit(hash_lock
);
1106 * Insert an entry into the hash table. If there is already an element
1107 * equal to elem in the hash table, then the already existing element
1108 * will be returned and the new element will not be inserted.
1109 * Otherwise returns NULL.
1110 * If lockp == NULL, the caller is assumed to already hold the hash lock.
1112 static arc_buf_hdr_t
*
1113 buf_hash_insert(arc_buf_hdr_t
*hdr
, kmutex_t
**lockp
)
1115 uint64_t idx
= BUF_HASH_INDEX(hdr
->b_spa
, &hdr
->b_dva
, hdr
->b_birth
);
1116 kmutex_t
*hash_lock
= BUF_HASH_LOCK(idx
);
1117 arc_buf_hdr_t
*fhdr
;
1120 ASSERT(!DVA_IS_EMPTY(&hdr
->b_dva
));
1121 ASSERT(hdr
->b_birth
!= 0);
1122 ASSERT(!HDR_IN_HASH_TABLE(hdr
));
1124 if (lockp
!= NULL
) {
1126 mutex_enter(hash_lock
);
1128 ASSERT(MUTEX_HELD(hash_lock
));
1131 for (fhdr
= buf_hash_table
.ht_table
[idx
], i
= 0; fhdr
!= NULL
;
1132 fhdr
= fhdr
->b_hash_next
, i
++) {
1133 if (HDR_EQUAL(hdr
->b_spa
, &hdr
->b_dva
, hdr
->b_birth
, fhdr
))
1137 hdr
->b_hash_next
= buf_hash_table
.ht_table
[idx
];
1138 buf_hash_table
.ht_table
[idx
] = hdr
;
1139 arc_hdr_set_flags(hdr
, ARC_FLAG_IN_HASH_TABLE
);
1141 /* collect some hash table performance data */
1143 ARCSTAT_BUMP(arcstat_hash_collisions
);
1145 ARCSTAT_BUMP(arcstat_hash_chains
);
1147 ARCSTAT_MAX(arcstat_hash_chain_max
, i
);
1150 ARCSTAT_BUMP(arcstat_hash_elements
);
1151 ARCSTAT_MAXSTAT(arcstat_hash_elements
);
1157 buf_hash_remove(arc_buf_hdr_t
*hdr
)
1159 arc_buf_hdr_t
*fhdr
, **hdrp
;
1160 uint64_t idx
= BUF_HASH_INDEX(hdr
->b_spa
, &hdr
->b_dva
, hdr
->b_birth
);
1162 ASSERT(MUTEX_HELD(BUF_HASH_LOCK(idx
)));
1163 ASSERT(HDR_IN_HASH_TABLE(hdr
));
1165 hdrp
= &buf_hash_table
.ht_table
[idx
];
1166 while ((fhdr
= *hdrp
) != hdr
) {
1167 ASSERT3P(fhdr
, !=, NULL
);
1168 hdrp
= &fhdr
->b_hash_next
;
1170 *hdrp
= hdr
->b_hash_next
;
1171 hdr
->b_hash_next
= NULL
;
1172 arc_hdr_clear_flags(hdr
, ARC_FLAG_IN_HASH_TABLE
);
1174 /* collect some hash table performance data */
1175 ARCSTAT_BUMPDOWN(arcstat_hash_elements
);
1177 if (buf_hash_table
.ht_table
[idx
] &&
1178 buf_hash_table
.ht_table
[idx
]->b_hash_next
== NULL
)
1179 ARCSTAT_BUMPDOWN(arcstat_hash_chains
);
1183 * Global data structures and functions for the buf kmem cache.
1186 static kmem_cache_t
*hdr_full_cache
;
1187 static kmem_cache_t
*hdr_full_crypt_cache
;
1188 static kmem_cache_t
*hdr_l2only_cache
;
1189 static kmem_cache_t
*buf_cache
;
1196 #if defined(_KERNEL) && defined(HAVE_SPL)
1198 * Large allocations which do not require contiguous pages
1199 * should be using vmem_free() in the linux kernel\
1201 vmem_free(buf_hash_table
.ht_table
,
1202 (buf_hash_table
.ht_mask
+ 1) * sizeof (void *));
1204 kmem_free(buf_hash_table
.ht_table
,
1205 (buf_hash_table
.ht_mask
+ 1) * sizeof (void *));
1207 for (i
= 0; i
< BUF_LOCKS
; i
++)
1208 mutex_destroy(&buf_hash_table
.ht_locks
[i
].ht_lock
);
1209 kmem_cache_destroy(hdr_full_cache
);
1210 kmem_cache_destroy(hdr_full_crypt_cache
);
1211 kmem_cache_destroy(hdr_l2only_cache
);
1212 kmem_cache_destroy(buf_cache
);
1216 * Constructor callback - called when the cache is empty
1217 * and a new buf is requested.
1221 hdr_full_cons(void *vbuf
, void *unused
, int kmflag
)
1223 arc_buf_hdr_t
*hdr
= vbuf
;
1225 bzero(hdr
, HDR_FULL_SIZE
);
1226 cv_init(&hdr
->b_l1hdr
.b_cv
, NULL
, CV_DEFAULT
, NULL
);
1227 refcount_create(&hdr
->b_l1hdr
.b_refcnt
);
1228 mutex_init(&hdr
->b_l1hdr
.b_freeze_lock
, NULL
, MUTEX_DEFAULT
, NULL
);
1229 list_link_init(&hdr
->b_l1hdr
.b_arc_node
);
1230 list_link_init(&hdr
->b_l2hdr
.b_l2node
);
1231 multilist_link_init(&hdr
->b_l1hdr
.b_arc_node
);
1232 arc_space_consume(HDR_FULL_SIZE
, ARC_SPACE_HDRS
);
1239 hdr_full_crypt_cons(void *vbuf
, void *unused
, int kmflag
)
1241 arc_buf_hdr_t
*hdr
= vbuf
;
1243 hdr_full_cons(vbuf
, unused
, kmflag
);
1244 bzero(&hdr
->b_crypt_hdr
, sizeof (hdr
->b_crypt_hdr
));
1245 arc_space_consume(sizeof (hdr
->b_crypt_hdr
), ARC_SPACE_HDRS
);
1252 hdr_l2only_cons(void *vbuf
, void *unused
, int kmflag
)
1254 arc_buf_hdr_t
*hdr
= vbuf
;
1256 bzero(hdr
, HDR_L2ONLY_SIZE
);
1257 arc_space_consume(HDR_L2ONLY_SIZE
, ARC_SPACE_L2HDRS
);
1264 buf_cons(void *vbuf
, void *unused
, int kmflag
)
1266 arc_buf_t
*buf
= vbuf
;
1268 bzero(buf
, sizeof (arc_buf_t
));
1269 mutex_init(&buf
->b_evict_lock
, NULL
, MUTEX_DEFAULT
, NULL
);
1270 arc_space_consume(sizeof (arc_buf_t
), ARC_SPACE_HDRS
);
1276 * Destructor callback - called when a cached buf is
1277 * no longer required.
1281 hdr_full_dest(void *vbuf
, void *unused
)
1283 arc_buf_hdr_t
*hdr
= vbuf
;
1285 ASSERT(HDR_EMPTY(hdr
));
1286 cv_destroy(&hdr
->b_l1hdr
.b_cv
);
1287 refcount_destroy(&hdr
->b_l1hdr
.b_refcnt
);
1288 mutex_destroy(&hdr
->b_l1hdr
.b_freeze_lock
);
1289 ASSERT(!multilist_link_active(&hdr
->b_l1hdr
.b_arc_node
));
1290 arc_space_return(HDR_FULL_SIZE
, ARC_SPACE_HDRS
);
1295 hdr_full_crypt_dest(void *vbuf
, void *unused
)
1297 arc_buf_hdr_t
*hdr
= vbuf
;
1299 hdr_full_dest(vbuf
, unused
);
1300 arc_space_return(sizeof (hdr
->b_crypt_hdr
), ARC_SPACE_HDRS
);
1305 hdr_l2only_dest(void *vbuf
, void *unused
)
1307 ASSERTV(arc_buf_hdr_t
*hdr
= vbuf
);
1309 ASSERT(HDR_EMPTY(hdr
));
1310 arc_space_return(HDR_L2ONLY_SIZE
, ARC_SPACE_L2HDRS
);
1315 buf_dest(void *vbuf
, void *unused
)
1317 arc_buf_t
*buf
= vbuf
;
1319 mutex_destroy(&buf
->b_evict_lock
);
1320 arc_space_return(sizeof (arc_buf_t
), ARC_SPACE_HDRS
);
1324 * Reclaim callback -- invoked when memory is low.
1328 hdr_recl(void *unused
)
1330 dprintf("hdr_recl called\n");
1332 * umem calls the reclaim func when we destroy the buf cache,
1333 * which is after we do arc_fini().
1336 cv_signal(&arc_reclaim_thread_cv
);
1342 uint64_t *ct
= NULL
;
1343 uint64_t hsize
= 1ULL << 12;
1347 * The hash table is big enough to fill all of physical memory
1348 * with an average block size of zfs_arc_average_blocksize (default 8K).
1349 * By default, the table will take up
1350 * totalmem * sizeof(void*) / 8K (1MB per GB with 8-byte pointers).
1352 while (hsize
* zfs_arc_average_blocksize
< arc_all_memory())
1355 buf_hash_table
.ht_mask
= hsize
- 1;
1356 #if defined(_KERNEL) && defined(HAVE_SPL)
1358 * Large allocations which do not require contiguous pages
1359 * should be using vmem_alloc() in the linux kernel
1361 buf_hash_table
.ht_table
=
1362 vmem_zalloc(hsize
* sizeof (void*), KM_SLEEP
);
1364 buf_hash_table
.ht_table
=
1365 kmem_zalloc(hsize
* sizeof (void*), KM_NOSLEEP
);
1367 if (buf_hash_table
.ht_table
== NULL
) {
1368 ASSERT(hsize
> (1ULL << 8));
1373 hdr_full_cache
= kmem_cache_create("arc_buf_hdr_t_full", HDR_FULL_SIZE
,
1374 0, hdr_full_cons
, hdr_full_dest
, hdr_recl
, NULL
, NULL
, 0);
1375 hdr_full_crypt_cache
= kmem_cache_create("arc_buf_hdr_t_full_crypt",
1376 HDR_FULL_CRYPT_SIZE
, 0, hdr_full_crypt_cons
, hdr_full_crypt_dest
,
1377 hdr_recl
, NULL
, NULL
, 0);
1378 hdr_l2only_cache
= kmem_cache_create("arc_buf_hdr_t_l2only",
1379 HDR_L2ONLY_SIZE
, 0, hdr_l2only_cons
, hdr_l2only_dest
, hdr_recl
,
1381 buf_cache
= kmem_cache_create("arc_buf_t", sizeof (arc_buf_t
),
1382 0, buf_cons
, buf_dest
, NULL
, NULL
, NULL
, 0);
1384 for (i
= 0; i
< 256; i
++)
1385 for (ct
= zfs_crc64_table
+ i
, *ct
= i
, j
= 8; j
> 0; j
--)
1386 *ct
= (*ct
>> 1) ^ (-(*ct
& 1) & ZFS_CRC64_POLY
);
1388 for (i
= 0; i
< BUF_LOCKS
; i
++) {
1389 mutex_init(&buf_hash_table
.ht_locks
[i
].ht_lock
,
1390 NULL
, MUTEX_DEFAULT
, NULL
);
1394 #define ARC_MINTIME (hz>>4) /* 62 ms */
1397 * This is the size that the buf occupies in memory. If the buf is compressed,
1398 * it will correspond to the compressed size. You should use this method of
1399 * getting the buf size unless you explicitly need the logical size.
1402 arc_buf_size(arc_buf_t
*buf
)
1404 return (ARC_BUF_COMPRESSED(buf
) ?
1405 HDR_GET_PSIZE(buf
->b_hdr
) : HDR_GET_LSIZE(buf
->b_hdr
));
1409 arc_buf_lsize(arc_buf_t
*buf
)
1411 return (HDR_GET_LSIZE(buf
->b_hdr
));
1415 * This function will return B_TRUE if the buffer is encrypted in memory.
1416 * This buffer can be decrypted by calling arc_untransform().
1419 arc_is_encrypted(arc_buf_t
*buf
)
1421 return (ARC_BUF_ENCRYPTED(buf
) != 0);
1425 * Returns B_TRUE if the buffer represents data that has not had its MAC
1429 arc_is_unauthenticated(arc_buf_t
*buf
)
1431 return (HDR_NOAUTH(buf
->b_hdr
) != 0);
1435 arc_get_raw_params(arc_buf_t
*buf
, boolean_t
*byteorder
, uint8_t *salt
,
1436 uint8_t *iv
, uint8_t *mac
)
1438 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
1440 ASSERT(HDR_PROTECTED(hdr
));
1442 bcopy(hdr
->b_crypt_hdr
.b_salt
, salt
, ZIO_DATA_SALT_LEN
);
1443 bcopy(hdr
->b_crypt_hdr
.b_iv
, iv
, ZIO_DATA_IV_LEN
);
1444 bcopy(hdr
->b_crypt_hdr
.b_mac
, mac
, ZIO_DATA_MAC_LEN
);
1445 *byteorder
= (hdr
->b_l1hdr
.b_byteswap
== DMU_BSWAP_NUMFUNCS
) ?
1446 ZFS_HOST_BYTEORDER
: !ZFS_HOST_BYTEORDER
;
1450 * Indicates how this buffer is compressed in memory. If it is not compressed
1451 * the value will be ZIO_COMPRESS_OFF. It can be made normally readable with
1452 * arc_untransform() as long as it is also unencrypted.
1455 arc_get_compression(arc_buf_t
*buf
)
1457 return (ARC_BUF_COMPRESSED(buf
) ?
1458 HDR_GET_COMPRESS(buf
->b_hdr
) : ZIO_COMPRESS_OFF
);
1462 * Return the compression algorithm used to store this data in the ARC. If ARC
1463 * compression is enabled or this is an encrypted block, this will be the same
1464 * as what's used to store it on-disk. Otherwise, this will be ZIO_COMPRESS_OFF.
1466 static inline enum zio_compress
1467 arc_hdr_get_compress(arc_buf_hdr_t
*hdr
)
1469 return (HDR_COMPRESSION_ENABLED(hdr
) ?
1470 HDR_GET_COMPRESS(hdr
) : ZIO_COMPRESS_OFF
);
1473 static inline boolean_t
1474 arc_buf_is_shared(arc_buf_t
*buf
)
1476 boolean_t shared
= (buf
->b_data
!= NULL
&&
1477 buf
->b_hdr
->b_l1hdr
.b_pabd
!= NULL
&&
1478 abd_is_linear(buf
->b_hdr
->b_l1hdr
.b_pabd
) &&
1479 buf
->b_data
== abd_to_buf(buf
->b_hdr
->b_l1hdr
.b_pabd
));
1480 IMPLY(shared
, HDR_SHARED_DATA(buf
->b_hdr
));
1481 IMPLY(shared
, ARC_BUF_SHARED(buf
));
1482 IMPLY(shared
, ARC_BUF_COMPRESSED(buf
) || ARC_BUF_LAST(buf
));
1485 * It would be nice to assert arc_can_share() too, but the "hdr isn't
1486 * already being shared" requirement prevents us from doing that.
1493 * Free the checksum associated with this header. If there is no checksum, this
1497 arc_cksum_free(arc_buf_hdr_t
*hdr
)
1499 ASSERT(HDR_HAS_L1HDR(hdr
));
1501 mutex_enter(&hdr
->b_l1hdr
.b_freeze_lock
);
1502 if (hdr
->b_l1hdr
.b_freeze_cksum
!= NULL
) {
1503 kmem_free(hdr
->b_l1hdr
.b_freeze_cksum
, sizeof (zio_cksum_t
));
1504 hdr
->b_l1hdr
.b_freeze_cksum
= NULL
;
1506 mutex_exit(&hdr
->b_l1hdr
.b_freeze_lock
);
1510 * Return true iff at least one of the bufs on hdr is not compressed.
1511 * Encrypted buffers count as compressed.
1514 arc_hdr_has_uncompressed_buf(arc_buf_hdr_t
*hdr
)
1516 for (arc_buf_t
*b
= hdr
->b_l1hdr
.b_buf
; b
!= NULL
; b
= b
->b_next
) {
1517 if (!ARC_BUF_COMPRESSED(b
)) {
1526 * If we've turned on the ZFS_DEBUG_MODIFY flag, verify that the buf's data
1527 * matches the checksum that is stored in the hdr. If there is no checksum,
1528 * or if the buf is compressed, this is a no-op.
1531 arc_cksum_verify(arc_buf_t
*buf
)
1533 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
1536 if (!(zfs_flags
& ZFS_DEBUG_MODIFY
))
1539 if (ARC_BUF_COMPRESSED(buf
)) {
1540 ASSERT(hdr
->b_l1hdr
.b_freeze_cksum
== NULL
||
1541 arc_hdr_has_uncompressed_buf(hdr
));
1545 ASSERT(HDR_HAS_L1HDR(hdr
));
1547 mutex_enter(&hdr
->b_l1hdr
.b_freeze_lock
);
1548 if (hdr
->b_l1hdr
.b_freeze_cksum
== NULL
|| HDR_IO_ERROR(hdr
)) {
1549 mutex_exit(&hdr
->b_l1hdr
.b_freeze_lock
);
1553 fletcher_2_native(buf
->b_data
, arc_buf_size(buf
), NULL
, &zc
);
1554 if (!ZIO_CHECKSUM_EQUAL(*hdr
->b_l1hdr
.b_freeze_cksum
, zc
))
1555 panic("buffer modified while frozen!");
1556 mutex_exit(&hdr
->b_l1hdr
.b_freeze_lock
);
1560 * This function makes the assumption that data stored in the L2ARC
1561 * will be transformed exactly as it is in the main pool. Because of
1562 * this we can verify the checksum against the reading process's bp.
1565 arc_cksum_is_equal(arc_buf_hdr_t
*hdr
, zio_t
*zio
)
1567 ASSERT(!BP_IS_EMBEDDED(zio
->io_bp
));
1568 VERIFY3U(BP_GET_PSIZE(zio
->io_bp
), ==, HDR_GET_PSIZE(hdr
));
1571 * Block pointers always store the checksum for the logical data.
1572 * If the block pointer has the gang bit set, then the checksum
1573 * it represents is for the reconstituted data and not for an
1574 * individual gang member. The zio pipeline, however, must be able to
1575 * determine the checksum of each of the gang constituents so it
1576 * treats the checksum comparison differently than what we need
1577 * for l2arc blocks. This prevents us from using the
1578 * zio_checksum_error() interface directly. Instead we must call the
1579 * zio_checksum_error_impl() so that we can ensure the checksum is
1580 * generated using the correct checksum algorithm and accounts for the
1581 * logical I/O size and not just a gang fragment.
1583 return (zio_checksum_error_impl(zio
->io_spa
, zio
->io_bp
,
1584 BP_GET_CHECKSUM(zio
->io_bp
), zio
->io_abd
, zio
->io_size
,
1585 zio
->io_offset
, NULL
) == 0);
1589 * Given a buf full of data, if ZFS_DEBUG_MODIFY is enabled this computes a
1590 * checksum and attaches it to the buf's hdr so that we can ensure that the buf
1591 * isn't modified later on. If buf is compressed or there is already a checksum
1592 * on the hdr, this is a no-op (we only checksum uncompressed bufs).
1595 arc_cksum_compute(arc_buf_t
*buf
)
1597 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
1599 if (!(zfs_flags
& ZFS_DEBUG_MODIFY
))
1602 ASSERT(HDR_HAS_L1HDR(hdr
));
1604 mutex_enter(&buf
->b_hdr
->b_l1hdr
.b_freeze_lock
);
1605 if (hdr
->b_l1hdr
.b_freeze_cksum
!= NULL
) {
1606 ASSERT(arc_hdr_has_uncompressed_buf(hdr
));
1607 mutex_exit(&hdr
->b_l1hdr
.b_freeze_lock
);
1609 } else if (ARC_BUF_COMPRESSED(buf
)) {
1610 mutex_exit(&hdr
->b_l1hdr
.b_freeze_lock
);
1614 ASSERT(!ARC_BUF_ENCRYPTED(buf
));
1615 ASSERT(!ARC_BUF_COMPRESSED(buf
));
1616 hdr
->b_l1hdr
.b_freeze_cksum
= kmem_alloc(sizeof (zio_cksum_t
),
1618 fletcher_2_native(buf
->b_data
, arc_buf_size(buf
), NULL
,
1619 hdr
->b_l1hdr
.b_freeze_cksum
);
1620 mutex_exit(&hdr
->b_l1hdr
.b_freeze_lock
);
1626 arc_buf_sigsegv(int sig
, siginfo_t
*si
, void *unused
)
1628 panic("Got SIGSEGV at address: 0x%lx\n", (long)si
->si_addr
);
1634 arc_buf_unwatch(arc_buf_t
*buf
)
1638 ASSERT0(mprotect(buf
->b_data
, arc_buf_size(buf
),
1639 PROT_READ
| PROT_WRITE
));
1646 arc_buf_watch(arc_buf_t
*buf
)
1650 ASSERT0(mprotect(buf
->b_data
, arc_buf_size(buf
),
1655 static arc_buf_contents_t
1656 arc_buf_type(arc_buf_hdr_t
*hdr
)
1658 arc_buf_contents_t type
;
1659 if (HDR_ISTYPE_METADATA(hdr
)) {
1660 type
= ARC_BUFC_METADATA
;
1662 type
= ARC_BUFC_DATA
;
1664 VERIFY3U(hdr
->b_type
, ==, type
);
1669 arc_is_metadata(arc_buf_t
*buf
)
1671 return (HDR_ISTYPE_METADATA(buf
->b_hdr
) != 0);
1675 arc_bufc_to_flags(arc_buf_contents_t type
)
1679 /* metadata field is 0 if buffer contains normal data */
1681 case ARC_BUFC_METADATA
:
1682 return (ARC_FLAG_BUFC_METADATA
);
1686 panic("undefined ARC buffer type!");
1687 return ((uint32_t)-1);
1691 arc_buf_thaw(arc_buf_t
*buf
)
1693 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
1695 ASSERT3P(hdr
->b_l1hdr
.b_state
, ==, arc_anon
);
1696 ASSERT(!HDR_IO_IN_PROGRESS(hdr
));
1698 arc_cksum_verify(buf
);
1701 * Compressed buffers do not manipulate the b_freeze_cksum or
1702 * allocate b_thawed.
1704 if (ARC_BUF_COMPRESSED(buf
)) {
1705 ASSERT(hdr
->b_l1hdr
.b_freeze_cksum
== NULL
||
1706 arc_hdr_has_uncompressed_buf(hdr
));
1710 ASSERT(HDR_HAS_L1HDR(hdr
));
1711 arc_cksum_free(hdr
);
1712 arc_buf_unwatch(buf
);
1716 arc_buf_freeze(arc_buf_t
*buf
)
1718 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
1719 kmutex_t
*hash_lock
;
1721 if (!(zfs_flags
& ZFS_DEBUG_MODIFY
))
1724 if (ARC_BUF_COMPRESSED(buf
)) {
1725 ASSERT(hdr
->b_l1hdr
.b_freeze_cksum
== NULL
||
1726 arc_hdr_has_uncompressed_buf(hdr
));
1730 hash_lock
= HDR_LOCK(hdr
);
1731 mutex_enter(hash_lock
);
1733 ASSERT(HDR_HAS_L1HDR(hdr
));
1734 ASSERT(hdr
->b_l1hdr
.b_freeze_cksum
!= NULL
||
1735 hdr
->b_l1hdr
.b_state
== arc_anon
);
1736 arc_cksum_compute(buf
);
1737 mutex_exit(hash_lock
);
1741 * The arc_buf_hdr_t's b_flags should never be modified directly. Instead,
1742 * the following functions should be used to ensure that the flags are
1743 * updated in a thread-safe way. When manipulating the flags either
1744 * the hash_lock must be held or the hdr must be undiscoverable. This
1745 * ensures that we're not racing with any other threads when updating
1749 arc_hdr_set_flags(arc_buf_hdr_t
*hdr
, arc_flags_t flags
)
1751 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)) || HDR_EMPTY(hdr
));
1752 hdr
->b_flags
|= flags
;
1756 arc_hdr_clear_flags(arc_buf_hdr_t
*hdr
, arc_flags_t flags
)
1758 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)) || HDR_EMPTY(hdr
));
1759 hdr
->b_flags
&= ~flags
;
1763 * Setting the compression bits in the arc_buf_hdr_t's b_flags is
1764 * done in a special way since we have to clear and set bits
1765 * at the same time. Consumers that wish to set the compression bits
1766 * must use this function to ensure that the flags are updated in
1767 * thread-safe manner.
1770 arc_hdr_set_compress(arc_buf_hdr_t
*hdr
, enum zio_compress cmp
)
1772 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)) || HDR_EMPTY(hdr
));
1775 * Holes and embedded blocks will always have a psize = 0 so
1776 * we ignore the compression of the blkptr and set the
1777 * want to uncompress them. Mark them as uncompressed.
1779 if (!zfs_compressed_arc_enabled
|| HDR_GET_PSIZE(hdr
) == 0) {
1780 arc_hdr_clear_flags(hdr
, ARC_FLAG_COMPRESSED_ARC
);
1781 ASSERT(!HDR_COMPRESSION_ENABLED(hdr
));
1783 arc_hdr_set_flags(hdr
, ARC_FLAG_COMPRESSED_ARC
);
1784 ASSERT(HDR_COMPRESSION_ENABLED(hdr
));
1787 HDR_SET_COMPRESS(hdr
, cmp
);
1788 ASSERT3U(HDR_GET_COMPRESS(hdr
), ==, cmp
);
1792 * Looks for another buf on the same hdr which has the data decompressed, copies
1793 * from it, and returns true. If no such buf exists, returns false.
1796 arc_buf_try_copy_decompressed_data(arc_buf_t
*buf
)
1798 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
1799 boolean_t copied
= B_FALSE
;
1801 ASSERT(HDR_HAS_L1HDR(hdr
));
1802 ASSERT3P(buf
->b_data
, !=, NULL
);
1803 ASSERT(!ARC_BUF_COMPRESSED(buf
));
1805 for (arc_buf_t
*from
= hdr
->b_l1hdr
.b_buf
; from
!= NULL
;
1806 from
= from
->b_next
) {
1807 /* can't use our own data buffer */
1812 if (!ARC_BUF_COMPRESSED(from
)) {
1813 bcopy(from
->b_data
, buf
->b_data
, arc_buf_size(buf
));
1820 * There were no decompressed bufs, so there should not be a
1821 * checksum on the hdr either.
1823 EQUIV(!copied
, hdr
->b_l1hdr
.b_freeze_cksum
== NULL
);
1829 * Return the size of the block, b_pabd, that is stored in the arc_buf_hdr_t.
1832 arc_hdr_size(arc_buf_hdr_t
*hdr
)
1836 if (arc_hdr_get_compress(hdr
) != ZIO_COMPRESS_OFF
&&
1837 HDR_GET_PSIZE(hdr
) > 0) {
1838 size
= HDR_GET_PSIZE(hdr
);
1840 ASSERT3U(HDR_GET_LSIZE(hdr
), !=, 0);
1841 size
= HDR_GET_LSIZE(hdr
);
1847 arc_hdr_authenticate(arc_buf_hdr_t
*hdr
, spa_t
*spa
, uint64_t dsobj
)
1851 uint64_t lsize
= HDR_GET_LSIZE(hdr
);
1852 uint64_t psize
= HDR_GET_PSIZE(hdr
);
1853 void *tmpbuf
= NULL
;
1854 abd_t
*abd
= hdr
->b_l1hdr
.b_pabd
;
1856 ASSERT(HDR_LOCK(hdr
) == NULL
|| MUTEX_HELD(HDR_LOCK(hdr
)));
1857 ASSERT(HDR_AUTHENTICATED(hdr
));
1858 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, !=, NULL
);
1861 * The MAC is calculated on the compressed data that is stored on disk.
1862 * However, if compressed arc is disabled we will only have the
1863 * decompressed data available to us now. Compress it into a temporary
1864 * abd so we can verify the MAC. The performance overhead of this will
1865 * be relatively low, since most objects in an encrypted objset will
1866 * be encrypted (instead of authenticated) anyway.
1868 if (HDR_GET_COMPRESS(hdr
) != ZIO_COMPRESS_OFF
&&
1869 !HDR_COMPRESSION_ENABLED(hdr
)) {
1870 tmpbuf
= zio_buf_alloc(lsize
);
1871 abd
= abd_get_from_buf(tmpbuf
, lsize
);
1872 abd_take_ownership_of_buf(abd
, B_TRUE
);
1874 csize
= zio_compress_data(HDR_GET_COMPRESS(hdr
),
1875 hdr
->b_l1hdr
.b_pabd
, tmpbuf
, lsize
);
1876 ASSERT3U(csize
, <=, psize
);
1877 abd_zero_off(abd
, csize
, psize
- csize
);
1881 * Authentication is best effort. We authenticate whenever the key is
1882 * available. If we succeed we clear ARC_FLAG_NOAUTH.
1884 if (hdr
->b_crypt_hdr
.b_ot
== DMU_OT_OBJSET
) {
1885 ASSERT3U(HDR_GET_COMPRESS(hdr
), ==, ZIO_COMPRESS_OFF
);
1886 ASSERT3U(lsize
, ==, psize
);
1887 ret
= spa_do_crypt_objset_mac_abd(B_FALSE
, spa
, dsobj
, abd
,
1888 psize
, hdr
->b_l1hdr
.b_byteswap
!= DMU_BSWAP_NUMFUNCS
);
1890 ret
= spa_do_crypt_mac_abd(B_FALSE
, spa
, dsobj
, abd
, psize
,
1891 hdr
->b_crypt_hdr
.b_mac
);
1895 arc_hdr_clear_flags(hdr
, ARC_FLAG_NOAUTH
);
1896 else if (ret
!= ENOENT
)
1912 * This function will take a header that only has raw encrypted data in
1913 * b_crypt_hdr.b_rabd and decrypt it into a new buffer which is stored in
1914 * b_l1hdr.b_pabd. If designated in the header flags, this function will
1915 * also decompress the data.
1918 arc_hdr_decrypt(arc_buf_hdr_t
*hdr
, spa_t
*spa
, uint64_t dsobj
)
1921 dsl_crypto_key_t
*dck
= NULL
;
1924 boolean_t no_crypt
= B_FALSE
;
1925 boolean_t bswap
= (hdr
->b_l1hdr
.b_byteswap
!= DMU_BSWAP_NUMFUNCS
);
1927 ASSERT(HDR_LOCK(hdr
) == NULL
|| MUTEX_HELD(HDR_LOCK(hdr
)));
1928 ASSERT(HDR_ENCRYPTED(hdr
));
1930 arc_hdr_alloc_abd(hdr
, B_FALSE
);
1933 * We must be careful to use the passed-in dsobj value here and
1934 * not the value in b_dsobj. b_dsobj is meant to be a best guess for
1935 * the L2ARC, which has the luxury of being able to fail without real
1936 * consequences (the data simply won't make it to the L2ARC). In
1937 * reality, the dsobj stored in the header may belong to a dataset
1938 * that has been unmounted or otherwise disowned, meaning the key
1939 * won't be accessible via that dsobj anymore.
1941 ret
= spa_keystore_lookup_key(spa
, dsobj
, FTAG
, &dck
);
1943 ret
= SET_ERROR(EACCES
);
1947 ret
= zio_do_crypt_abd(B_FALSE
, &dck
->dck_key
,
1948 hdr
->b_crypt_hdr
.b_salt
, hdr
->b_crypt_hdr
.b_ot
,
1949 hdr
->b_crypt_hdr
.b_iv
, hdr
->b_crypt_hdr
.b_mac
,
1950 HDR_GET_PSIZE(hdr
), bswap
, hdr
->b_l1hdr
.b_pabd
,
1951 hdr
->b_crypt_hdr
.b_rabd
, &no_crypt
);
1956 abd_copy(hdr
->b_l1hdr
.b_pabd
, hdr
->b_crypt_hdr
.b_rabd
,
1957 HDR_GET_PSIZE(hdr
));
1961 * If this header has disabled arc compression but the b_pabd is
1962 * compressed after decrypting it, we need to decompress the newly
1965 if (HDR_GET_COMPRESS(hdr
) != ZIO_COMPRESS_OFF
&&
1966 !HDR_COMPRESSION_ENABLED(hdr
)) {
1968 * We want to make sure that we are correctly honoring the
1969 * zfs_abd_scatter_enabled setting, so we allocate an abd here
1970 * and then loan a buffer from it, rather than allocating a
1971 * linear buffer and wrapping it in an abd later.
1973 cabd
= arc_get_data_abd(hdr
, arc_hdr_size(hdr
), hdr
);
1974 tmp
= abd_borrow_buf(cabd
, arc_hdr_size(hdr
));
1976 ret
= zio_decompress_data(HDR_GET_COMPRESS(hdr
),
1977 hdr
->b_l1hdr
.b_pabd
, tmp
, HDR_GET_PSIZE(hdr
),
1978 HDR_GET_LSIZE(hdr
));
1980 abd_return_buf(cabd
, tmp
, arc_hdr_size(hdr
));
1984 abd_return_buf_copy(cabd
, tmp
, arc_hdr_size(hdr
));
1985 arc_free_data_abd(hdr
, hdr
->b_l1hdr
.b_pabd
,
1986 arc_hdr_size(hdr
), hdr
);
1987 hdr
->b_l1hdr
.b_pabd
= cabd
;
1990 spa_keystore_dsl_key_rele(spa
, dck
, FTAG
);
1995 arc_hdr_free_abd(hdr
, B_FALSE
);
1997 spa_keystore_dsl_key_rele(spa
, dck
, FTAG
);
1999 arc_free_data_buf(hdr
, cabd
, arc_hdr_size(hdr
), hdr
);
2005 * This function is called during arc_buf_fill() to prepare the header's
2006 * abd plaintext pointer for use. This involves authenticated protected
2007 * data and decrypting encrypted data into the plaintext abd.
2010 arc_fill_hdr_crypt(arc_buf_hdr_t
*hdr
, kmutex_t
*hash_lock
, spa_t
*spa
,
2011 uint64_t dsobj
, boolean_t noauth
)
2015 ASSERT(HDR_PROTECTED(hdr
));
2017 if (hash_lock
!= NULL
)
2018 mutex_enter(hash_lock
);
2020 if (HDR_NOAUTH(hdr
) && !noauth
) {
2022 * The caller requested authenticated data but our data has
2023 * not been authenticated yet. Verify the MAC now if we can.
2025 ret
= arc_hdr_authenticate(hdr
, spa
, dsobj
);
2028 } else if (HDR_HAS_RABD(hdr
) && hdr
->b_l1hdr
.b_pabd
== NULL
) {
2030 * If we only have the encrypted version of the data, but the
2031 * unencrypted version was requested we take this opportunity
2032 * to store the decrypted version in the header for future use.
2034 ret
= arc_hdr_decrypt(hdr
, spa
, dsobj
);
2039 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, !=, NULL
);
2041 if (hash_lock
!= NULL
)
2042 mutex_exit(hash_lock
);
2047 if (hash_lock
!= NULL
)
2048 mutex_exit(hash_lock
);
2054 * This function is used by the dbuf code to decrypt bonus buffers in place.
2055 * The dbuf code itself doesn't have any locking for decrypting a shared dnode
2056 * block, so we use the hash lock here to protect against concurrent calls to
2060 arc_buf_untransform_in_place(arc_buf_t
*buf
, kmutex_t
*hash_lock
)
2062 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
2064 ASSERT(HDR_ENCRYPTED(hdr
));
2065 ASSERT3U(hdr
->b_crypt_hdr
.b_ot
, ==, DMU_OT_DNODE
);
2066 ASSERT(HDR_LOCK(hdr
) == NULL
|| MUTEX_HELD(HDR_LOCK(hdr
)));
2067 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, !=, NULL
);
2069 zio_crypt_copy_dnode_bonus(hdr
->b_l1hdr
.b_pabd
, buf
->b_data
,
2071 buf
->b_flags
&= ~ARC_BUF_FLAG_ENCRYPTED
;
2072 buf
->b_flags
&= ~ARC_BUF_FLAG_COMPRESSED
;
2073 hdr
->b_crypt_hdr
.b_ebufcnt
-= 1;
2077 * Given a buf that has a data buffer attached to it, this function will
2078 * efficiently fill the buf with data of the specified compression setting from
2079 * the hdr and update the hdr's b_freeze_cksum if necessary. If the buf and hdr
2080 * are already sharing a data buf, no copy is performed.
2082 * If the buf is marked as compressed but uncompressed data was requested, this
2083 * will allocate a new data buffer for the buf, remove that flag, and fill the
2084 * buf with uncompressed data. You can't request a compressed buf on a hdr with
2085 * uncompressed data, and (since we haven't added support for it yet) if you
2086 * want compressed data your buf must already be marked as compressed and have
2087 * the correct-sized data buffer.
2090 arc_buf_fill(arc_buf_t
*buf
, spa_t
*spa
, uint64_t dsobj
, arc_fill_flags_t flags
)
2093 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
2094 boolean_t hdr_compressed
=
2095 (arc_hdr_get_compress(hdr
) != ZIO_COMPRESS_OFF
);
2096 boolean_t compressed
= (flags
& ARC_FILL_COMPRESSED
) != 0;
2097 boolean_t encrypted
= (flags
& ARC_FILL_ENCRYPTED
) != 0;
2098 dmu_object_byteswap_t bswap
= hdr
->b_l1hdr
.b_byteswap
;
2099 kmutex_t
*hash_lock
= (flags
& ARC_FILL_LOCKED
) ? NULL
: HDR_LOCK(hdr
);
2101 ASSERT3P(buf
->b_data
, !=, NULL
);
2102 IMPLY(compressed
, hdr_compressed
|| ARC_BUF_ENCRYPTED(buf
));
2103 IMPLY(compressed
, ARC_BUF_COMPRESSED(buf
));
2104 IMPLY(encrypted
, HDR_ENCRYPTED(hdr
));
2105 IMPLY(encrypted
, ARC_BUF_ENCRYPTED(buf
));
2106 IMPLY(encrypted
, ARC_BUF_COMPRESSED(buf
));
2107 IMPLY(encrypted
, !ARC_BUF_SHARED(buf
));
2110 * If the caller wanted encrypted data we just need to copy it from
2111 * b_rabd and potentially byteswap it. We won't be able to do any
2112 * further transforms on it.
2115 ASSERT(HDR_HAS_RABD(hdr
));
2116 abd_copy_to_buf(buf
->b_data
, hdr
->b_crypt_hdr
.b_rabd
,
2117 HDR_GET_PSIZE(hdr
));
2122 * Adjust encrypted and authenticated headers to accomodate the
2123 * request if needed.
2125 if (HDR_PROTECTED(hdr
)) {
2126 error
= arc_fill_hdr_crypt(hdr
, hash_lock
, spa
,
2127 dsobj
, !!(flags
& ARC_FILL_NOAUTH
));
2133 * There is a special case here for dnode blocks which are
2134 * decrypting their bonus buffers. These blocks may request to
2135 * be decrypted in-place. This is necessary because there may
2136 * be many dnodes pointing into this buffer and there is
2137 * currently no method to synchronize replacing the backing
2138 * b_data buffer and updating all of the pointers. Here we use
2139 * the hash lock to ensure there are no races. If the need
2140 * arises for other types to be decrypted in-place, they must
2141 * add handling here as well.
2143 if ((flags
& ARC_FILL_IN_PLACE
) != 0) {
2144 ASSERT(!hdr_compressed
);
2145 ASSERT(!compressed
);
2148 if (HDR_ENCRYPTED(hdr
) && ARC_BUF_ENCRYPTED(buf
)) {
2149 ASSERT3U(hdr
->b_crypt_hdr
.b_ot
, ==, DMU_OT_DNODE
);
2151 if (hash_lock
!= NULL
)
2152 mutex_enter(hash_lock
);
2153 arc_buf_untransform_in_place(buf
, hash_lock
);
2154 if (hash_lock
!= NULL
)
2155 mutex_exit(hash_lock
);
2157 /* Compute the hdr's checksum if necessary */
2158 arc_cksum_compute(buf
);
2164 if (hdr_compressed
== compressed
) {
2165 if (!arc_buf_is_shared(buf
)) {
2166 abd_copy_to_buf(buf
->b_data
, hdr
->b_l1hdr
.b_pabd
,
2170 ASSERT(hdr_compressed
);
2171 ASSERT(!compressed
);
2172 ASSERT3U(HDR_GET_LSIZE(hdr
), !=, HDR_GET_PSIZE(hdr
));
2175 * If the buf is sharing its data with the hdr, unlink it and
2176 * allocate a new data buffer for the buf.
2178 if (arc_buf_is_shared(buf
)) {
2179 ASSERT(ARC_BUF_COMPRESSED(buf
));
2181 /* We need to give the buf it's own b_data */
2182 buf
->b_flags
&= ~ARC_BUF_FLAG_SHARED
;
2184 arc_get_data_buf(hdr
, HDR_GET_LSIZE(hdr
), buf
);
2185 arc_hdr_clear_flags(hdr
, ARC_FLAG_SHARED_DATA
);
2187 /* Previously overhead was 0; just add new overhead */
2188 ARCSTAT_INCR(arcstat_overhead_size
, HDR_GET_LSIZE(hdr
));
2189 } else if (ARC_BUF_COMPRESSED(buf
)) {
2190 /* We need to reallocate the buf's b_data */
2191 arc_free_data_buf(hdr
, buf
->b_data
, HDR_GET_PSIZE(hdr
),
2194 arc_get_data_buf(hdr
, HDR_GET_LSIZE(hdr
), buf
);
2196 /* We increased the size of b_data; update overhead */
2197 ARCSTAT_INCR(arcstat_overhead_size
,
2198 HDR_GET_LSIZE(hdr
) - HDR_GET_PSIZE(hdr
));
2202 * Regardless of the buf's previous compression settings, it
2203 * should not be compressed at the end of this function.
2205 buf
->b_flags
&= ~ARC_BUF_FLAG_COMPRESSED
;
2208 * Try copying the data from another buf which already has a
2209 * decompressed version. If that's not possible, it's time to
2210 * bite the bullet and decompress the data from the hdr.
2212 if (arc_buf_try_copy_decompressed_data(buf
)) {
2213 /* Skip byteswapping and checksumming (already done) */
2214 ASSERT3P(hdr
->b_l1hdr
.b_freeze_cksum
, !=, NULL
);
2217 error
= zio_decompress_data(HDR_GET_COMPRESS(hdr
),
2218 hdr
->b_l1hdr
.b_pabd
, buf
->b_data
,
2219 HDR_GET_PSIZE(hdr
), HDR_GET_LSIZE(hdr
));
2222 * Absent hardware errors or software bugs, this should
2223 * be impossible, but log it anyway so we can debug it.
2227 "hdr %p, compress %d, psize %d, lsize %d",
2228 hdr
, arc_hdr_get_compress(hdr
),
2229 HDR_GET_PSIZE(hdr
), HDR_GET_LSIZE(hdr
));
2230 return (SET_ERROR(EIO
));
2236 /* Byteswap the buf's data if necessary */
2237 if (bswap
!= DMU_BSWAP_NUMFUNCS
) {
2238 ASSERT(!HDR_SHARED_DATA(hdr
));
2239 ASSERT3U(bswap
, <, DMU_BSWAP_NUMFUNCS
);
2240 dmu_ot_byteswap
[bswap
].ob_func(buf
->b_data
, HDR_GET_LSIZE(hdr
));
2243 /* Compute the hdr's checksum if necessary */
2244 arc_cksum_compute(buf
);
2250 * If this function is being called to decrypt an encrypted buffer or verify an
2251 * authenticated one, the key must be loaded and a mapping must be made
2252 * available in the keystore via spa_keystore_create_mapping() or one of its
2256 arc_untransform(arc_buf_t
*buf
, spa_t
*spa
, uint64_t dsobj
, boolean_t in_place
)
2258 arc_fill_flags_t flags
= 0;
2261 flags
|= ARC_FILL_IN_PLACE
;
2263 return (arc_buf_fill(buf
, spa
, dsobj
, flags
));
2267 * Increment the amount of evictable space in the arc_state_t's refcount.
2268 * We account for the space used by the hdr and the arc buf individually
2269 * so that we can add and remove them from the refcount individually.
2272 arc_evictable_space_increment(arc_buf_hdr_t
*hdr
, arc_state_t
*state
)
2274 arc_buf_contents_t type
= arc_buf_type(hdr
);
2276 ASSERT(HDR_HAS_L1HDR(hdr
));
2278 if (GHOST_STATE(state
)) {
2279 ASSERT0(hdr
->b_l1hdr
.b_bufcnt
);
2280 ASSERT3P(hdr
->b_l1hdr
.b_buf
, ==, NULL
);
2281 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
2282 ASSERT(!HDR_HAS_RABD(hdr
));
2283 (void) refcount_add_many(&state
->arcs_esize
[type
],
2284 HDR_GET_LSIZE(hdr
), hdr
);
2288 ASSERT(!GHOST_STATE(state
));
2289 if (hdr
->b_l1hdr
.b_pabd
!= NULL
) {
2290 (void) refcount_add_many(&state
->arcs_esize
[type
],
2291 arc_hdr_size(hdr
), hdr
);
2293 if (HDR_HAS_RABD(hdr
)) {
2294 (void) refcount_add_many(&state
->arcs_esize
[type
],
2295 HDR_GET_PSIZE(hdr
), hdr
);
2298 for (arc_buf_t
*buf
= hdr
->b_l1hdr
.b_buf
; buf
!= NULL
;
2299 buf
= buf
->b_next
) {
2300 if (arc_buf_is_shared(buf
))
2302 (void) refcount_add_many(&state
->arcs_esize
[type
],
2303 arc_buf_size(buf
), buf
);
2308 * Decrement the amount of evictable space in the arc_state_t's refcount.
2309 * We account for the space used by the hdr and the arc buf individually
2310 * so that we can add and remove them from the refcount individually.
2313 arc_evictable_space_decrement(arc_buf_hdr_t
*hdr
, arc_state_t
*state
)
2315 arc_buf_contents_t type
= arc_buf_type(hdr
);
2317 ASSERT(HDR_HAS_L1HDR(hdr
));
2319 if (GHOST_STATE(state
)) {
2320 ASSERT0(hdr
->b_l1hdr
.b_bufcnt
);
2321 ASSERT3P(hdr
->b_l1hdr
.b_buf
, ==, NULL
);
2322 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
2323 ASSERT(!HDR_HAS_RABD(hdr
));
2324 (void) refcount_remove_many(&state
->arcs_esize
[type
],
2325 HDR_GET_LSIZE(hdr
), hdr
);
2329 ASSERT(!GHOST_STATE(state
));
2330 if (hdr
->b_l1hdr
.b_pabd
!= NULL
) {
2331 (void) refcount_remove_many(&state
->arcs_esize
[type
],
2332 arc_hdr_size(hdr
), hdr
);
2334 if (HDR_HAS_RABD(hdr
)) {
2335 (void) refcount_remove_many(&state
->arcs_esize
[type
],
2336 HDR_GET_PSIZE(hdr
), hdr
);
2339 for (arc_buf_t
*buf
= hdr
->b_l1hdr
.b_buf
; buf
!= NULL
;
2340 buf
= buf
->b_next
) {
2341 if (arc_buf_is_shared(buf
))
2343 (void) refcount_remove_many(&state
->arcs_esize
[type
],
2344 arc_buf_size(buf
), buf
);
2349 * Add a reference to this hdr indicating that someone is actively
2350 * referencing that memory. When the refcount transitions from 0 to 1,
2351 * we remove it from the respective arc_state_t list to indicate that
2352 * it is not evictable.
2355 add_reference(arc_buf_hdr_t
*hdr
, void *tag
)
2359 ASSERT(HDR_HAS_L1HDR(hdr
));
2360 if (!MUTEX_HELD(HDR_LOCK(hdr
))) {
2361 ASSERT(hdr
->b_l1hdr
.b_state
== arc_anon
);
2362 ASSERT(refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
));
2363 ASSERT3P(hdr
->b_l1hdr
.b_buf
, ==, NULL
);
2366 state
= hdr
->b_l1hdr
.b_state
;
2368 if ((refcount_add(&hdr
->b_l1hdr
.b_refcnt
, tag
) == 1) &&
2369 (state
!= arc_anon
)) {
2370 /* We don't use the L2-only state list. */
2371 if (state
!= arc_l2c_only
) {
2372 multilist_remove(state
->arcs_list
[arc_buf_type(hdr
)],
2374 arc_evictable_space_decrement(hdr
, state
);
2376 /* remove the prefetch flag if we get a reference */
2377 arc_hdr_clear_flags(hdr
, ARC_FLAG_PREFETCH
);
2382 * Remove a reference from this hdr. When the reference transitions from
2383 * 1 to 0 and we're not anonymous, then we add this hdr to the arc_state_t's
2384 * list making it eligible for eviction.
2387 remove_reference(arc_buf_hdr_t
*hdr
, kmutex_t
*hash_lock
, void *tag
)
2390 arc_state_t
*state
= hdr
->b_l1hdr
.b_state
;
2392 ASSERT(HDR_HAS_L1HDR(hdr
));
2393 ASSERT(state
== arc_anon
|| MUTEX_HELD(hash_lock
));
2394 ASSERT(!GHOST_STATE(state
));
2397 * arc_l2c_only counts as a ghost state so we don't need to explicitly
2398 * check to prevent usage of the arc_l2c_only list.
2400 if (((cnt
= refcount_remove(&hdr
->b_l1hdr
.b_refcnt
, tag
)) == 0) &&
2401 (state
!= arc_anon
)) {
2402 multilist_insert(state
->arcs_list
[arc_buf_type(hdr
)], hdr
);
2403 ASSERT3U(hdr
->b_l1hdr
.b_bufcnt
, >, 0);
2404 arc_evictable_space_increment(hdr
, state
);
2410 * Returns detailed information about a specific arc buffer. When the
2411 * state_index argument is set the function will calculate the arc header
2412 * list position for its arc state. Since this requires a linear traversal
2413 * callers are strongly encourage not to do this. However, it can be helpful
2414 * for targeted analysis so the functionality is provided.
2417 arc_buf_info(arc_buf_t
*ab
, arc_buf_info_t
*abi
, int state_index
)
2419 arc_buf_hdr_t
*hdr
= ab
->b_hdr
;
2420 l1arc_buf_hdr_t
*l1hdr
= NULL
;
2421 l2arc_buf_hdr_t
*l2hdr
= NULL
;
2422 arc_state_t
*state
= NULL
;
2424 memset(abi
, 0, sizeof (arc_buf_info_t
));
2429 abi
->abi_flags
= hdr
->b_flags
;
2431 if (HDR_HAS_L1HDR(hdr
)) {
2432 l1hdr
= &hdr
->b_l1hdr
;
2433 state
= l1hdr
->b_state
;
2435 if (HDR_HAS_L2HDR(hdr
))
2436 l2hdr
= &hdr
->b_l2hdr
;
2439 abi
->abi_bufcnt
= l1hdr
->b_bufcnt
;
2440 abi
->abi_access
= l1hdr
->b_arc_access
;
2441 abi
->abi_mru_hits
= l1hdr
->b_mru_hits
;
2442 abi
->abi_mru_ghost_hits
= l1hdr
->b_mru_ghost_hits
;
2443 abi
->abi_mfu_hits
= l1hdr
->b_mfu_hits
;
2444 abi
->abi_mfu_ghost_hits
= l1hdr
->b_mfu_ghost_hits
;
2445 abi
->abi_holds
= refcount_count(&l1hdr
->b_refcnt
);
2449 abi
->abi_l2arc_dattr
= l2hdr
->b_daddr
;
2450 abi
->abi_l2arc_hits
= l2hdr
->b_hits
;
2453 abi
->abi_state_type
= state
? state
->arcs_state
: ARC_STATE_ANON
;
2454 abi
->abi_state_contents
= arc_buf_type(hdr
);
2455 abi
->abi_size
= arc_hdr_size(hdr
);
2459 * Move the supplied buffer to the indicated state. The hash lock
2460 * for the buffer must be held by the caller.
2463 arc_change_state(arc_state_t
*new_state
, arc_buf_hdr_t
*hdr
,
2464 kmutex_t
*hash_lock
)
2466 arc_state_t
*old_state
;
2469 boolean_t update_old
, update_new
;
2470 arc_buf_contents_t buftype
= arc_buf_type(hdr
);
2473 * We almost always have an L1 hdr here, since we call arc_hdr_realloc()
2474 * in arc_read() when bringing a buffer out of the L2ARC. However, the
2475 * L1 hdr doesn't always exist when we change state to arc_anon before
2476 * destroying a header, in which case reallocating to add the L1 hdr is
2479 if (HDR_HAS_L1HDR(hdr
)) {
2480 old_state
= hdr
->b_l1hdr
.b_state
;
2481 refcnt
= refcount_count(&hdr
->b_l1hdr
.b_refcnt
);
2482 bufcnt
= hdr
->b_l1hdr
.b_bufcnt
;
2483 update_old
= (bufcnt
> 0 || hdr
->b_l1hdr
.b_pabd
!= NULL
||
2486 old_state
= arc_l2c_only
;
2489 update_old
= B_FALSE
;
2491 update_new
= update_old
;
2493 ASSERT(MUTEX_HELD(hash_lock
));
2494 ASSERT3P(new_state
, !=, old_state
);
2495 ASSERT(!GHOST_STATE(new_state
) || bufcnt
== 0);
2496 ASSERT(old_state
!= arc_anon
|| bufcnt
<= 1);
2499 * If this buffer is evictable, transfer it from the
2500 * old state list to the new state list.
2503 if (old_state
!= arc_anon
&& old_state
!= arc_l2c_only
) {
2504 ASSERT(HDR_HAS_L1HDR(hdr
));
2505 multilist_remove(old_state
->arcs_list
[buftype
], hdr
);
2507 if (GHOST_STATE(old_state
)) {
2509 ASSERT3P(hdr
->b_l1hdr
.b_buf
, ==, NULL
);
2510 update_old
= B_TRUE
;
2512 arc_evictable_space_decrement(hdr
, old_state
);
2514 if (new_state
!= arc_anon
&& new_state
!= arc_l2c_only
) {
2516 * An L1 header always exists here, since if we're
2517 * moving to some L1-cached state (i.e. not l2c_only or
2518 * anonymous), we realloc the header to add an L1hdr
2521 ASSERT(HDR_HAS_L1HDR(hdr
));
2522 multilist_insert(new_state
->arcs_list
[buftype
], hdr
);
2524 if (GHOST_STATE(new_state
)) {
2526 ASSERT3P(hdr
->b_l1hdr
.b_buf
, ==, NULL
);
2527 update_new
= B_TRUE
;
2529 arc_evictable_space_increment(hdr
, new_state
);
2533 ASSERT(!HDR_EMPTY(hdr
));
2534 if (new_state
== arc_anon
&& HDR_IN_HASH_TABLE(hdr
))
2535 buf_hash_remove(hdr
);
2537 /* adjust state sizes (ignore arc_l2c_only) */
2539 if (update_new
&& new_state
!= arc_l2c_only
) {
2540 ASSERT(HDR_HAS_L1HDR(hdr
));
2541 if (GHOST_STATE(new_state
)) {
2545 * When moving a header to a ghost state, we first
2546 * remove all arc buffers. Thus, we'll have a
2547 * bufcnt of zero, and no arc buffer to use for
2548 * the reference. As a result, we use the arc
2549 * header pointer for the reference.
2551 (void) refcount_add_many(&new_state
->arcs_size
,
2552 HDR_GET_LSIZE(hdr
), hdr
);
2553 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
2554 ASSERT(!HDR_HAS_RABD(hdr
));
2556 uint32_t buffers
= 0;
2559 * Each individual buffer holds a unique reference,
2560 * thus we must remove each of these references one
2563 for (arc_buf_t
*buf
= hdr
->b_l1hdr
.b_buf
; buf
!= NULL
;
2564 buf
= buf
->b_next
) {
2565 ASSERT3U(bufcnt
, !=, 0);
2569 * When the arc_buf_t is sharing the data
2570 * block with the hdr, the owner of the
2571 * reference belongs to the hdr. Only
2572 * add to the refcount if the arc_buf_t is
2575 if (arc_buf_is_shared(buf
))
2578 (void) refcount_add_many(&new_state
->arcs_size
,
2579 arc_buf_size(buf
), buf
);
2581 ASSERT3U(bufcnt
, ==, buffers
);
2583 if (hdr
->b_l1hdr
.b_pabd
!= NULL
) {
2584 (void) refcount_add_many(&new_state
->arcs_size
,
2585 arc_hdr_size(hdr
), hdr
);
2588 if (HDR_HAS_RABD(hdr
)) {
2589 (void) refcount_add_many(&new_state
->arcs_size
,
2590 HDR_GET_PSIZE(hdr
), hdr
);
2595 if (update_old
&& old_state
!= arc_l2c_only
) {
2596 ASSERT(HDR_HAS_L1HDR(hdr
));
2597 if (GHOST_STATE(old_state
)) {
2599 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
2600 ASSERT(!HDR_HAS_RABD(hdr
));
2603 * When moving a header off of a ghost state,
2604 * the header will not contain any arc buffers.
2605 * We use the arc header pointer for the reference
2606 * which is exactly what we did when we put the
2607 * header on the ghost state.
2610 (void) refcount_remove_many(&old_state
->arcs_size
,
2611 HDR_GET_LSIZE(hdr
), hdr
);
2613 uint32_t buffers
= 0;
2616 * Each individual buffer holds a unique reference,
2617 * thus we must remove each of these references one
2620 for (arc_buf_t
*buf
= hdr
->b_l1hdr
.b_buf
; buf
!= NULL
;
2621 buf
= buf
->b_next
) {
2622 ASSERT3U(bufcnt
, !=, 0);
2626 * When the arc_buf_t is sharing the data
2627 * block with the hdr, the owner of the
2628 * reference belongs to the hdr. Only
2629 * add to the refcount if the arc_buf_t is
2632 if (arc_buf_is_shared(buf
))
2635 (void) refcount_remove_many(
2636 &old_state
->arcs_size
, arc_buf_size(buf
),
2639 ASSERT3U(bufcnt
, ==, buffers
);
2640 ASSERT(hdr
->b_l1hdr
.b_pabd
!= NULL
||
2643 if (hdr
->b_l1hdr
.b_pabd
!= NULL
) {
2644 (void) refcount_remove_many(
2645 &old_state
->arcs_size
, arc_hdr_size(hdr
),
2649 if (HDR_HAS_RABD(hdr
)) {
2650 (void) refcount_remove_many(
2651 &old_state
->arcs_size
, HDR_GET_PSIZE(hdr
),
2657 if (HDR_HAS_L1HDR(hdr
))
2658 hdr
->b_l1hdr
.b_state
= new_state
;
2661 * L2 headers should never be on the L2 state list since they don't
2662 * have L1 headers allocated.
2664 ASSERT(multilist_is_empty(arc_l2c_only
->arcs_list
[ARC_BUFC_DATA
]) &&
2665 multilist_is_empty(arc_l2c_only
->arcs_list
[ARC_BUFC_METADATA
]));
2669 arc_space_consume(uint64_t space
, arc_space_type_t type
)
2671 ASSERT(type
>= 0 && type
< ARC_SPACE_NUMTYPES
);
2676 case ARC_SPACE_DATA
:
2677 ARCSTAT_INCR(arcstat_data_size
, space
);
2679 case ARC_SPACE_META
:
2680 ARCSTAT_INCR(arcstat_metadata_size
, space
);
2682 case ARC_SPACE_BONUS
:
2683 ARCSTAT_INCR(arcstat_bonus_size
, space
);
2685 case ARC_SPACE_DNODE
:
2686 ARCSTAT_INCR(arcstat_dnode_size
, space
);
2688 case ARC_SPACE_DBUF
:
2689 ARCSTAT_INCR(arcstat_dbuf_size
, space
);
2691 case ARC_SPACE_HDRS
:
2692 ARCSTAT_INCR(arcstat_hdr_size
, space
);
2694 case ARC_SPACE_L2HDRS
:
2695 ARCSTAT_INCR(arcstat_l2_hdr_size
, space
);
2699 if (type
!= ARC_SPACE_DATA
)
2700 ARCSTAT_INCR(arcstat_meta_used
, space
);
2702 atomic_add_64(&arc_size
, space
);
2706 arc_space_return(uint64_t space
, arc_space_type_t type
)
2708 ASSERT(type
>= 0 && type
< ARC_SPACE_NUMTYPES
);
2713 case ARC_SPACE_DATA
:
2714 ARCSTAT_INCR(arcstat_data_size
, -space
);
2716 case ARC_SPACE_META
:
2717 ARCSTAT_INCR(arcstat_metadata_size
, -space
);
2719 case ARC_SPACE_BONUS
:
2720 ARCSTAT_INCR(arcstat_bonus_size
, -space
);
2722 case ARC_SPACE_DNODE
:
2723 ARCSTAT_INCR(arcstat_dnode_size
, -space
);
2725 case ARC_SPACE_DBUF
:
2726 ARCSTAT_INCR(arcstat_dbuf_size
, -space
);
2728 case ARC_SPACE_HDRS
:
2729 ARCSTAT_INCR(arcstat_hdr_size
, -space
);
2731 case ARC_SPACE_L2HDRS
:
2732 ARCSTAT_INCR(arcstat_l2_hdr_size
, -space
);
2736 if (type
!= ARC_SPACE_DATA
) {
2737 ASSERT(arc_meta_used
>= space
);
2738 if (arc_meta_max
< arc_meta_used
)
2739 arc_meta_max
= arc_meta_used
;
2740 ARCSTAT_INCR(arcstat_meta_used
, -space
);
2743 ASSERT(arc_size
>= space
);
2744 atomic_add_64(&arc_size
, -space
);
2748 * Given a hdr and a buf, returns whether that buf can share its b_data buffer
2749 * with the hdr's b_pabd.
2752 arc_can_share(arc_buf_hdr_t
*hdr
, arc_buf_t
*buf
)
2755 * The criteria for sharing a hdr's data are:
2756 * 1. the buffer is not encrypted
2757 * 2. the hdr's compression matches the buf's compression
2758 * 3. the hdr doesn't need to be byteswapped
2759 * 4. the hdr isn't already being shared
2760 * 5. the buf is either compressed or it is the last buf in the hdr list
2762 * Criterion #5 maintains the invariant that shared uncompressed
2763 * bufs must be the final buf in the hdr's b_buf list. Reading this, you
2764 * might ask, "if a compressed buf is allocated first, won't that be the
2765 * last thing in the list?", but in that case it's impossible to create
2766 * a shared uncompressed buf anyway (because the hdr must be compressed
2767 * to have the compressed buf). You might also think that #3 is
2768 * sufficient to make this guarantee, however it's possible
2769 * (specifically in the rare L2ARC write race mentioned in
2770 * arc_buf_alloc_impl()) there will be an existing uncompressed buf that
2771 * is sharable, but wasn't at the time of its allocation. Rather than
2772 * allow a new shared uncompressed buf to be created and then shuffle
2773 * the list around to make it the last element, this simply disallows
2774 * sharing if the new buf isn't the first to be added.
2776 ASSERT3P(buf
->b_hdr
, ==, hdr
);
2777 boolean_t hdr_compressed
=
2778 arc_hdr_get_compress(hdr
) != ZIO_COMPRESS_OFF
;
2779 boolean_t buf_compressed
= ARC_BUF_COMPRESSED(buf
) != 0;
2780 return (!ARC_BUF_ENCRYPTED(buf
) &&
2781 buf_compressed
== hdr_compressed
&&
2782 hdr
->b_l1hdr
.b_byteswap
== DMU_BSWAP_NUMFUNCS
&&
2783 !HDR_SHARED_DATA(hdr
) &&
2784 (ARC_BUF_LAST(buf
) || ARC_BUF_COMPRESSED(buf
)));
2788 * Allocate a buf for this hdr. If you care about the data that's in the hdr,
2789 * or if you want a compressed buffer, pass those flags in. Returns 0 if the
2790 * copy was made successfully, or an error code otherwise.
2793 arc_buf_alloc_impl(arc_buf_hdr_t
*hdr
, spa_t
*spa
, uint64_t dsobj
, void *tag
,
2794 boolean_t encrypted
, boolean_t compressed
, boolean_t noauth
,
2795 boolean_t fill
, arc_buf_t
**ret
)
2798 arc_fill_flags_t flags
= ARC_FILL_LOCKED
;
2800 ASSERT(HDR_HAS_L1HDR(hdr
));
2801 ASSERT3U(HDR_GET_LSIZE(hdr
), >, 0);
2802 VERIFY(hdr
->b_type
== ARC_BUFC_DATA
||
2803 hdr
->b_type
== ARC_BUFC_METADATA
);
2804 ASSERT3P(ret
, !=, NULL
);
2805 ASSERT3P(*ret
, ==, NULL
);
2806 IMPLY(encrypted
, compressed
);
2808 hdr
->b_l1hdr
.b_mru_hits
= 0;
2809 hdr
->b_l1hdr
.b_mru_ghost_hits
= 0;
2810 hdr
->b_l1hdr
.b_mfu_hits
= 0;
2811 hdr
->b_l1hdr
.b_mfu_ghost_hits
= 0;
2812 hdr
->b_l1hdr
.b_l2_hits
= 0;
2814 buf
= *ret
= kmem_cache_alloc(buf_cache
, KM_PUSHPAGE
);
2817 buf
->b_next
= hdr
->b_l1hdr
.b_buf
;
2820 add_reference(hdr
, tag
);
2823 * We're about to change the hdr's b_flags. We must either
2824 * hold the hash_lock or be undiscoverable.
2826 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)) || HDR_EMPTY(hdr
));
2829 * Only honor requests for compressed bufs if the hdr is actually
2830 * compressed. This must be overriden if the buffer is encrypted since
2831 * encrypted buffers cannot be decompressed.
2834 buf
->b_flags
|= ARC_BUF_FLAG_COMPRESSED
;
2835 buf
->b_flags
|= ARC_BUF_FLAG_ENCRYPTED
;
2836 flags
|= ARC_FILL_COMPRESSED
| ARC_FILL_ENCRYPTED
;
2837 } else if (compressed
&&
2838 arc_hdr_get_compress(hdr
) != ZIO_COMPRESS_OFF
) {
2839 buf
->b_flags
|= ARC_BUF_FLAG_COMPRESSED
;
2840 flags
|= ARC_FILL_COMPRESSED
;
2845 flags
|= ARC_FILL_NOAUTH
;
2849 * If the hdr's data can be shared then we share the data buffer and
2850 * set the appropriate bit in the hdr's b_flags to indicate the hdr is
2851 * allocate a new buffer to store the buf's data.
2853 * There are two additional restrictions here because we're sharing
2854 * hdr -> buf instead of the usual buf -> hdr. First, the hdr can't be
2855 * actively involved in an L2ARC write, because if this buf is used by
2856 * an arc_write() then the hdr's data buffer will be released when the
2857 * write completes, even though the L2ARC write might still be using it.
2858 * Second, the hdr's ABD must be linear so that the buf's user doesn't
2859 * need to be ABD-aware.
2861 boolean_t can_share
= arc_can_share(hdr
, buf
) && !HDR_L2_WRITING(hdr
) &&
2862 hdr
->b_l1hdr
.b_pabd
!= NULL
&& abd_is_linear(hdr
->b_l1hdr
.b_pabd
);
2864 /* Set up b_data and sharing */
2866 buf
->b_data
= abd_to_buf(hdr
->b_l1hdr
.b_pabd
);
2867 buf
->b_flags
|= ARC_BUF_FLAG_SHARED
;
2868 arc_hdr_set_flags(hdr
, ARC_FLAG_SHARED_DATA
);
2871 arc_get_data_buf(hdr
, arc_buf_size(buf
), buf
);
2872 ARCSTAT_INCR(arcstat_overhead_size
, arc_buf_size(buf
));
2874 VERIFY3P(buf
->b_data
, !=, NULL
);
2876 hdr
->b_l1hdr
.b_buf
= buf
;
2877 hdr
->b_l1hdr
.b_bufcnt
+= 1;
2879 hdr
->b_crypt_hdr
.b_ebufcnt
+= 1;
2882 * If the user wants the data from the hdr, we need to either copy or
2883 * decompress the data.
2886 return (arc_buf_fill(buf
, spa
, dsobj
, flags
));
2892 static char *arc_onloan_tag
= "onloan";
2895 arc_loaned_bytes_update(int64_t delta
)
2897 atomic_add_64(&arc_loaned_bytes
, delta
);
2899 /* assert that it did not wrap around */
2900 ASSERT3S(atomic_add_64_nv(&arc_loaned_bytes
, 0), >=, 0);
2904 * Loan out an anonymous arc buffer. Loaned buffers are not counted as in
2905 * flight data by arc_tempreserve_space() until they are "returned". Loaned
2906 * buffers must be returned to the arc before they can be used by the DMU or
2910 arc_loan_buf(spa_t
*spa
, boolean_t is_metadata
, int size
)
2912 arc_buf_t
*buf
= arc_alloc_buf(spa
, arc_onloan_tag
,
2913 is_metadata
? ARC_BUFC_METADATA
: ARC_BUFC_DATA
, size
);
2915 arc_loaned_bytes_update(size
);
2921 arc_loan_compressed_buf(spa_t
*spa
, uint64_t psize
, uint64_t lsize
,
2922 enum zio_compress compression_type
)
2924 arc_buf_t
*buf
= arc_alloc_compressed_buf(spa
, arc_onloan_tag
,
2925 psize
, lsize
, compression_type
);
2927 arc_loaned_bytes_update(psize
);
2933 arc_loan_raw_buf(spa_t
*spa
, uint64_t dsobj
, boolean_t byteorder
,
2934 const uint8_t *salt
, const uint8_t *iv
, const uint8_t *mac
,
2935 dmu_object_type_t ot
, uint64_t psize
, uint64_t lsize
,
2936 enum zio_compress compression_type
)
2938 arc_buf_t
*buf
= arc_alloc_raw_buf(spa
, arc_onloan_tag
, dsobj
,
2939 byteorder
, salt
, iv
, mac
, ot
, psize
, lsize
, compression_type
);
2941 atomic_add_64(&arc_loaned_bytes
, psize
);
2947 * Return a loaned arc buffer to the arc.
2950 arc_return_buf(arc_buf_t
*buf
, void *tag
)
2952 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
2954 ASSERT3P(buf
->b_data
, !=, NULL
);
2955 ASSERT(HDR_HAS_L1HDR(hdr
));
2956 (void) refcount_add(&hdr
->b_l1hdr
.b_refcnt
, tag
);
2957 (void) refcount_remove(&hdr
->b_l1hdr
.b_refcnt
, arc_onloan_tag
);
2959 arc_loaned_bytes_update(-arc_buf_size(buf
));
2962 /* Detach an arc_buf from a dbuf (tag) */
2964 arc_loan_inuse_buf(arc_buf_t
*buf
, void *tag
)
2966 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
2968 ASSERT3P(buf
->b_data
, !=, NULL
);
2969 ASSERT(HDR_HAS_L1HDR(hdr
));
2970 (void) refcount_add(&hdr
->b_l1hdr
.b_refcnt
, arc_onloan_tag
);
2971 (void) refcount_remove(&hdr
->b_l1hdr
.b_refcnt
, tag
);
2973 arc_loaned_bytes_update(arc_buf_size(buf
));
2977 l2arc_free_abd_on_write(abd_t
*abd
, size_t size
, arc_buf_contents_t type
)
2979 l2arc_data_free_t
*df
= kmem_alloc(sizeof (*df
), KM_SLEEP
);
2982 df
->l2df_size
= size
;
2983 df
->l2df_type
= type
;
2984 mutex_enter(&l2arc_free_on_write_mtx
);
2985 list_insert_head(l2arc_free_on_write
, df
);
2986 mutex_exit(&l2arc_free_on_write_mtx
);
2990 arc_hdr_free_on_write(arc_buf_hdr_t
*hdr
, boolean_t free_rdata
)
2992 arc_state_t
*state
= hdr
->b_l1hdr
.b_state
;
2993 arc_buf_contents_t type
= arc_buf_type(hdr
);
2994 uint64_t size
= (free_rdata
) ? HDR_GET_PSIZE(hdr
) : arc_hdr_size(hdr
);
2996 /* protected by hash lock, if in the hash table */
2997 if (multilist_link_active(&hdr
->b_l1hdr
.b_arc_node
)) {
2998 ASSERT(refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
));
2999 ASSERT(state
!= arc_anon
&& state
!= arc_l2c_only
);
3001 (void) refcount_remove_many(&state
->arcs_esize
[type
],
3004 (void) refcount_remove_many(&state
->arcs_size
, size
, hdr
);
3005 if (type
== ARC_BUFC_METADATA
) {
3006 arc_space_return(size
, ARC_SPACE_META
);
3008 ASSERT(type
== ARC_BUFC_DATA
);
3009 arc_space_return(size
, ARC_SPACE_DATA
);
3013 l2arc_free_abd_on_write(hdr
->b_crypt_hdr
.b_rabd
, size
, type
);
3015 l2arc_free_abd_on_write(hdr
->b_l1hdr
.b_pabd
, size
, type
);
3020 * Share the arc_buf_t's data with the hdr. Whenever we are sharing the
3021 * data buffer, we transfer the refcount ownership to the hdr and update
3022 * the appropriate kstats.
3025 arc_share_buf(arc_buf_hdr_t
*hdr
, arc_buf_t
*buf
)
3027 ASSERT(arc_can_share(hdr
, buf
));
3028 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
3029 ASSERT(!ARC_BUF_ENCRYPTED(buf
));
3030 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)) || HDR_EMPTY(hdr
));
3033 * Start sharing the data buffer. We transfer the
3034 * refcount ownership to the hdr since it always owns
3035 * the refcount whenever an arc_buf_t is shared.
3037 refcount_transfer_ownership(&hdr
->b_l1hdr
.b_state
->arcs_size
, buf
, hdr
);
3038 hdr
->b_l1hdr
.b_pabd
= abd_get_from_buf(buf
->b_data
, arc_buf_size(buf
));
3039 abd_take_ownership_of_buf(hdr
->b_l1hdr
.b_pabd
,
3040 HDR_ISTYPE_METADATA(hdr
));
3041 arc_hdr_set_flags(hdr
, ARC_FLAG_SHARED_DATA
);
3042 buf
->b_flags
|= ARC_BUF_FLAG_SHARED
;
3045 * Since we've transferred ownership to the hdr we need
3046 * to increment its compressed and uncompressed kstats and
3047 * decrement the overhead size.
3049 ARCSTAT_INCR(arcstat_compressed_size
, arc_hdr_size(hdr
));
3050 ARCSTAT_INCR(arcstat_uncompressed_size
, HDR_GET_LSIZE(hdr
));
3051 ARCSTAT_INCR(arcstat_overhead_size
, -arc_buf_size(buf
));
3055 arc_unshare_buf(arc_buf_hdr_t
*hdr
, arc_buf_t
*buf
)
3057 ASSERT(arc_buf_is_shared(buf
));
3058 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, !=, NULL
);
3059 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)) || HDR_EMPTY(hdr
));
3062 * We are no longer sharing this buffer so we need
3063 * to transfer its ownership to the rightful owner.
3065 refcount_transfer_ownership(&hdr
->b_l1hdr
.b_state
->arcs_size
, hdr
, buf
);
3066 arc_hdr_clear_flags(hdr
, ARC_FLAG_SHARED_DATA
);
3067 abd_release_ownership_of_buf(hdr
->b_l1hdr
.b_pabd
);
3068 abd_put(hdr
->b_l1hdr
.b_pabd
);
3069 hdr
->b_l1hdr
.b_pabd
= NULL
;
3070 buf
->b_flags
&= ~ARC_BUF_FLAG_SHARED
;
3073 * Since the buffer is no longer shared between
3074 * the arc buf and the hdr, count it as overhead.
3076 ARCSTAT_INCR(arcstat_compressed_size
, -arc_hdr_size(hdr
));
3077 ARCSTAT_INCR(arcstat_uncompressed_size
, -HDR_GET_LSIZE(hdr
));
3078 ARCSTAT_INCR(arcstat_overhead_size
, arc_buf_size(buf
));
3082 * Remove an arc_buf_t from the hdr's buf list and return the last
3083 * arc_buf_t on the list. If no buffers remain on the list then return
3087 arc_buf_remove(arc_buf_hdr_t
*hdr
, arc_buf_t
*buf
)
3089 ASSERT(HDR_HAS_L1HDR(hdr
));
3090 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)) || HDR_EMPTY(hdr
));
3092 arc_buf_t
**bufp
= &hdr
->b_l1hdr
.b_buf
;
3093 arc_buf_t
*lastbuf
= NULL
;
3096 * Remove the buf from the hdr list and locate the last
3097 * remaining buffer on the list.
3099 while (*bufp
!= NULL
) {
3101 *bufp
= buf
->b_next
;
3104 * If we've removed a buffer in the middle of
3105 * the list then update the lastbuf and update
3108 if (*bufp
!= NULL
) {
3110 bufp
= &(*bufp
)->b_next
;
3114 ASSERT3P(lastbuf
, !=, buf
);
3115 IMPLY(hdr
->b_l1hdr
.b_bufcnt
> 0, lastbuf
!= NULL
);
3116 IMPLY(hdr
->b_l1hdr
.b_bufcnt
> 0, hdr
->b_l1hdr
.b_buf
!= NULL
);
3117 IMPLY(lastbuf
!= NULL
, ARC_BUF_LAST(lastbuf
));
3123 * Free up buf->b_data and pull the arc_buf_t off of the the arc_buf_hdr_t's
3127 arc_buf_destroy_impl(arc_buf_t
*buf
)
3129 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
3132 * Free up the data associated with the buf but only if we're not
3133 * sharing this with the hdr. If we are sharing it with the hdr, the
3134 * hdr is responsible for doing the free.
3136 if (buf
->b_data
!= NULL
) {
3138 * We're about to change the hdr's b_flags. We must either
3139 * hold the hash_lock or be undiscoverable.
3141 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)) || HDR_EMPTY(hdr
));
3143 arc_cksum_verify(buf
);
3144 arc_buf_unwatch(buf
);
3146 if (arc_buf_is_shared(buf
)) {
3147 arc_hdr_clear_flags(hdr
, ARC_FLAG_SHARED_DATA
);
3149 uint64_t size
= arc_buf_size(buf
);
3150 arc_free_data_buf(hdr
, buf
->b_data
, size
, buf
);
3151 ARCSTAT_INCR(arcstat_overhead_size
, -size
);
3155 ASSERT(hdr
->b_l1hdr
.b_bufcnt
> 0);
3156 hdr
->b_l1hdr
.b_bufcnt
-= 1;
3158 if (ARC_BUF_ENCRYPTED(buf
)) {
3159 hdr
->b_crypt_hdr
.b_ebufcnt
-= 1;
3162 * If we have no more encrypted buffers and we've
3163 * already gotten a copy of the decrypted data we can
3164 * free b_rabd to save some space.
3166 if (hdr
->b_crypt_hdr
.b_ebufcnt
== 0 &&
3167 HDR_HAS_RABD(hdr
) && hdr
->b_l1hdr
.b_pabd
!= NULL
&&
3168 !HDR_IO_IN_PROGRESS(hdr
)) {
3169 arc_hdr_free_abd(hdr
, B_TRUE
);
3174 arc_buf_t
*lastbuf
= arc_buf_remove(hdr
, buf
);
3176 if (ARC_BUF_SHARED(buf
) && !ARC_BUF_COMPRESSED(buf
)) {
3178 * If the current arc_buf_t is sharing its data buffer with the
3179 * hdr, then reassign the hdr's b_pabd to share it with the new
3180 * buffer at the end of the list. The shared buffer is always
3181 * the last one on the hdr's buffer list.
3183 * There is an equivalent case for compressed bufs, but since
3184 * they aren't guaranteed to be the last buf in the list and
3185 * that is an exceedingly rare case, we just allow that space be
3186 * wasted temporarily. We must also be careful not to share
3187 * encrypted buffers, since they cannot be shared.
3189 if (lastbuf
!= NULL
&& !ARC_BUF_ENCRYPTED(lastbuf
)) {
3190 /* Only one buf can be shared at once */
3191 VERIFY(!arc_buf_is_shared(lastbuf
));
3192 /* hdr is uncompressed so can't have compressed buf */
3193 VERIFY(!ARC_BUF_COMPRESSED(lastbuf
));
3195 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, !=, NULL
);
3196 arc_hdr_free_abd(hdr
, B_FALSE
);
3199 * We must setup a new shared block between the
3200 * last buffer and the hdr. The data would have
3201 * been allocated by the arc buf so we need to transfer
3202 * ownership to the hdr since it's now being shared.
3204 arc_share_buf(hdr
, lastbuf
);
3206 } else if (HDR_SHARED_DATA(hdr
)) {
3208 * Uncompressed shared buffers are always at the end
3209 * of the list. Compressed buffers don't have the
3210 * same requirements. This makes it hard to
3211 * simply assert that the lastbuf is shared so
3212 * we rely on the hdr's compression flags to determine
3213 * if we have a compressed, shared buffer.
3215 ASSERT3P(lastbuf
, !=, NULL
);
3216 ASSERT(arc_buf_is_shared(lastbuf
) ||
3217 arc_hdr_get_compress(hdr
) != ZIO_COMPRESS_OFF
);
3221 * Free the checksum if we're removing the last uncompressed buf from
3224 if (!arc_hdr_has_uncompressed_buf(hdr
)) {
3225 arc_cksum_free(hdr
);
3228 /* clean up the buf */
3230 kmem_cache_free(buf_cache
, buf
);
3234 arc_hdr_alloc_abd(arc_buf_hdr_t
*hdr
, boolean_t alloc_rdata
)
3238 ASSERT3U(HDR_GET_LSIZE(hdr
), >, 0);
3239 ASSERT(HDR_HAS_L1HDR(hdr
));
3240 ASSERT(!HDR_SHARED_DATA(hdr
) || alloc_rdata
);
3241 IMPLY(alloc_rdata
, HDR_PROTECTED(hdr
));
3243 if (hdr
->b_l1hdr
.b_pabd
== NULL
&& !HDR_HAS_RABD(hdr
))
3244 hdr
->b_l1hdr
.b_byteswap
= DMU_BSWAP_NUMFUNCS
;
3247 size
= HDR_GET_PSIZE(hdr
);
3248 ASSERT3P(hdr
->b_crypt_hdr
.b_rabd
, ==, NULL
);
3249 hdr
->b_crypt_hdr
.b_rabd
= arc_get_data_abd(hdr
, size
, hdr
);
3250 ASSERT3P(hdr
->b_crypt_hdr
.b_rabd
, !=, NULL
);
3251 ARCSTAT_INCR(arcstat_raw_size
, size
);
3253 size
= arc_hdr_size(hdr
);
3254 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
3255 hdr
->b_l1hdr
.b_pabd
= arc_get_data_abd(hdr
, size
, hdr
);
3256 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, !=, NULL
);
3259 ARCSTAT_INCR(arcstat_compressed_size
, size
);
3260 ARCSTAT_INCR(arcstat_uncompressed_size
, HDR_GET_LSIZE(hdr
));
3264 arc_hdr_free_abd(arc_buf_hdr_t
*hdr
, boolean_t free_rdata
)
3266 uint64_t size
= (free_rdata
) ? HDR_GET_PSIZE(hdr
) : arc_hdr_size(hdr
);
3268 ASSERT(HDR_HAS_L1HDR(hdr
));
3269 ASSERT(hdr
->b_l1hdr
.b_pabd
!= NULL
|| HDR_HAS_RABD(hdr
));
3270 IMPLY(free_rdata
, HDR_HAS_RABD(hdr
));
3273 * If the hdr is currently being written to the l2arc then
3274 * we defer freeing the data by adding it to the l2arc_free_on_write
3275 * list. The l2arc will free the data once it's finished
3276 * writing it to the l2arc device.
3278 if (HDR_L2_WRITING(hdr
)) {
3279 arc_hdr_free_on_write(hdr
, free_rdata
);
3280 ARCSTAT_BUMP(arcstat_l2_free_on_write
);
3281 } else if (free_rdata
) {
3282 arc_free_data_abd(hdr
, hdr
->b_crypt_hdr
.b_rabd
, size
, hdr
);
3284 arc_free_data_abd(hdr
, hdr
->b_l1hdr
.b_pabd
, size
, hdr
);
3288 hdr
->b_crypt_hdr
.b_rabd
= NULL
;
3289 ARCSTAT_INCR(arcstat_raw_size
, -size
);
3291 hdr
->b_l1hdr
.b_pabd
= NULL
;
3294 if (hdr
->b_l1hdr
.b_pabd
== NULL
&& !HDR_HAS_RABD(hdr
))
3295 hdr
->b_l1hdr
.b_byteswap
= DMU_BSWAP_NUMFUNCS
;
3297 ARCSTAT_INCR(arcstat_compressed_size
, -size
);
3298 ARCSTAT_INCR(arcstat_uncompressed_size
, -HDR_GET_LSIZE(hdr
));
3301 static arc_buf_hdr_t
*
3302 arc_hdr_alloc(uint64_t spa
, int32_t psize
, int32_t lsize
,
3303 boolean_t
protected, enum zio_compress compression_type
,
3304 arc_buf_contents_t type
, boolean_t alloc_rdata
)
3308 VERIFY(type
== ARC_BUFC_DATA
|| type
== ARC_BUFC_METADATA
);
3310 hdr
= kmem_cache_alloc(hdr_full_crypt_cache
, KM_PUSHPAGE
);
3312 hdr
= kmem_cache_alloc(hdr_full_cache
, KM_PUSHPAGE
);
3315 ASSERT(HDR_EMPTY(hdr
));
3316 ASSERT3P(hdr
->b_l1hdr
.b_freeze_cksum
, ==, NULL
);
3317 HDR_SET_PSIZE(hdr
, psize
);
3318 HDR_SET_LSIZE(hdr
, lsize
);
3322 arc_hdr_set_flags(hdr
, arc_bufc_to_flags(type
) | ARC_FLAG_HAS_L1HDR
);
3323 arc_hdr_set_compress(hdr
, compression_type
);
3325 arc_hdr_set_flags(hdr
, ARC_FLAG_PROTECTED
);
3327 hdr
->b_l1hdr
.b_state
= arc_anon
;
3328 hdr
->b_l1hdr
.b_arc_access
= 0;
3329 hdr
->b_l1hdr
.b_bufcnt
= 0;
3330 hdr
->b_l1hdr
.b_buf
= NULL
;
3333 * Allocate the hdr's buffer. This will contain either
3334 * the compressed or uncompressed data depending on the block
3335 * it references and compressed arc enablement.
3337 arc_hdr_alloc_abd(hdr
, alloc_rdata
);
3338 ASSERT(refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
));
3344 * Transition between the two allocation states for the arc_buf_hdr struct.
3345 * The arc_buf_hdr struct can be allocated with (hdr_full_cache) or without
3346 * (hdr_l2only_cache) the fields necessary for the L1 cache - the smaller
3347 * version is used when a cache buffer is only in the L2ARC in order to reduce
3350 static arc_buf_hdr_t
*
3351 arc_hdr_realloc(arc_buf_hdr_t
*hdr
, kmem_cache_t
*old
, kmem_cache_t
*new)
3353 ASSERT(HDR_HAS_L2HDR(hdr
));
3355 arc_buf_hdr_t
*nhdr
;
3356 l2arc_dev_t
*dev
= hdr
->b_l2hdr
.b_dev
;
3358 ASSERT((old
== hdr_full_cache
&& new == hdr_l2only_cache
) ||
3359 (old
== hdr_l2only_cache
&& new == hdr_full_cache
));
3362 * if the caller wanted a new full header and the header is to be
3363 * encrypted we will actually allocate the header from the full crypt
3364 * cache instead. The same applies to freeing from the old cache.
3366 if (HDR_PROTECTED(hdr
) && new == hdr_full_cache
)
3367 new = hdr_full_crypt_cache
;
3368 if (HDR_PROTECTED(hdr
) && old
== hdr_full_cache
)
3369 old
= hdr_full_crypt_cache
;
3371 nhdr
= kmem_cache_alloc(new, KM_PUSHPAGE
);
3373 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)));
3374 buf_hash_remove(hdr
);
3376 bcopy(hdr
, nhdr
, HDR_L2ONLY_SIZE
);
3378 if (new == hdr_full_cache
|| new == hdr_full_crypt_cache
) {
3379 arc_hdr_set_flags(nhdr
, ARC_FLAG_HAS_L1HDR
);
3381 * arc_access and arc_change_state need to be aware that a
3382 * header has just come out of L2ARC, so we set its state to
3383 * l2c_only even though it's about to change.
3385 nhdr
->b_l1hdr
.b_state
= arc_l2c_only
;
3387 /* Verify previous threads set to NULL before freeing */
3388 ASSERT3P(nhdr
->b_l1hdr
.b_pabd
, ==, NULL
);
3389 ASSERT(!HDR_HAS_RABD(hdr
));
3391 ASSERT3P(hdr
->b_l1hdr
.b_buf
, ==, NULL
);
3392 ASSERT0(hdr
->b_l1hdr
.b_bufcnt
);
3393 ASSERT3P(hdr
->b_l1hdr
.b_freeze_cksum
, ==, NULL
);
3396 * If we've reached here, We must have been called from
3397 * arc_evict_hdr(), as such we should have already been
3398 * removed from any ghost list we were previously on
3399 * (which protects us from racing with arc_evict_state),
3400 * thus no locking is needed during this check.
3402 ASSERT(!multilist_link_active(&hdr
->b_l1hdr
.b_arc_node
));
3405 * A buffer must not be moved into the arc_l2c_only
3406 * state if it's not finished being written out to the
3407 * l2arc device. Otherwise, the b_l1hdr.b_pabd field
3408 * might try to be accessed, even though it was removed.
3410 VERIFY(!HDR_L2_WRITING(hdr
));
3411 VERIFY3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
3412 ASSERT(!HDR_HAS_RABD(hdr
));
3414 arc_hdr_clear_flags(nhdr
, ARC_FLAG_HAS_L1HDR
);
3417 * The header has been reallocated so we need to re-insert it into any
3420 (void) buf_hash_insert(nhdr
, NULL
);
3422 ASSERT(list_link_active(&hdr
->b_l2hdr
.b_l2node
));
3424 mutex_enter(&dev
->l2ad_mtx
);
3427 * We must place the realloc'ed header back into the list at
3428 * the same spot. Otherwise, if it's placed earlier in the list,
3429 * l2arc_write_buffers() could find it during the function's
3430 * write phase, and try to write it out to the l2arc.
3432 list_insert_after(&dev
->l2ad_buflist
, hdr
, nhdr
);
3433 list_remove(&dev
->l2ad_buflist
, hdr
);
3435 mutex_exit(&dev
->l2ad_mtx
);
3438 * Since we're using the pointer address as the tag when
3439 * incrementing and decrementing the l2ad_alloc refcount, we
3440 * must remove the old pointer (that we're about to destroy) and
3441 * add the new pointer to the refcount. Otherwise we'd remove
3442 * the wrong pointer address when calling arc_hdr_destroy() later.
3445 (void) refcount_remove_many(&dev
->l2ad_alloc
, arc_hdr_size(hdr
), hdr
);
3446 (void) refcount_add_many(&dev
->l2ad_alloc
, arc_hdr_size(nhdr
), nhdr
);
3448 buf_discard_identity(hdr
);
3449 kmem_cache_free(old
, hdr
);
3455 * This function allows an L1 header to be reallocated as a crypt
3456 * header and vice versa. If we are going to a crypt header, the
3457 * new fields will be zeroed out.
3459 static arc_buf_hdr_t
*
3460 arc_hdr_realloc_crypt(arc_buf_hdr_t
*hdr
, boolean_t need_crypt
)
3462 arc_buf_hdr_t
*nhdr
;
3464 kmem_cache_t
*ncache
, *ocache
;
3466 ASSERT(HDR_HAS_L1HDR(hdr
));
3467 ASSERT3U(!!HDR_PROTECTED(hdr
), !=, need_crypt
);
3468 ASSERT3P(hdr
->b_l1hdr
.b_state
, ==, arc_anon
);
3469 ASSERT(!multilist_link_active(&hdr
->b_l1hdr
.b_arc_node
));
3472 ncache
= hdr_full_crypt_cache
;
3473 ocache
= hdr_full_cache
;
3475 ncache
= hdr_full_cache
;
3476 ocache
= hdr_full_crypt_cache
;
3479 nhdr
= kmem_cache_alloc(ncache
, KM_PUSHPAGE
);
3480 bcopy(hdr
, nhdr
, HDR_L2ONLY_SIZE
);
3481 nhdr
->b_l1hdr
.b_freeze_cksum
= hdr
->b_l1hdr
.b_freeze_cksum
;
3482 nhdr
->b_l1hdr
.b_bufcnt
= hdr
->b_l1hdr
.b_bufcnt
;
3483 nhdr
->b_l1hdr
.b_byteswap
= hdr
->b_l1hdr
.b_byteswap
;
3484 nhdr
->b_l1hdr
.b_state
= hdr
->b_l1hdr
.b_state
;
3485 nhdr
->b_l1hdr
.b_arc_access
= hdr
->b_l1hdr
.b_arc_access
;
3486 nhdr
->b_l1hdr
.b_mru_hits
= hdr
->b_l1hdr
.b_mru_hits
;
3487 nhdr
->b_l1hdr
.b_mru_ghost_hits
= hdr
->b_l1hdr
.b_mru_ghost_hits
;
3488 nhdr
->b_l1hdr
.b_mfu_hits
= hdr
->b_l1hdr
.b_mfu_hits
;
3489 nhdr
->b_l1hdr
.b_mfu_ghost_hits
= hdr
->b_l1hdr
.b_mfu_ghost_hits
;
3490 nhdr
->b_l1hdr
.b_l2_hits
= hdr
->b_l1hdr
.b_l2_hits
;
3491 nhdr
->b_l1hdr
.b_acb
= hdr
->b_l1hdr
.b_acb
;
3492 nhdr
->b_l1hdr
.b_pabd
= hdr
->b_l1hdr
.b_pabd
;
3493 nhdr
->b_l1hdr
.b_buf
= hdr
->b_l1hdr
.b_buf
;
3496 * This refcount_add() exists only to ensure that the individual
3497 * arc buffers always point to a header that is referenced, avoiding
3498 * a small race condition that could trigger ASSERTs.
3500 (void) refcount_add(&nhdr
->b_l1hdr
.b_refcnt
, FTAG
);
3502 for (buf
= nhdr
->b_l1hdr
.b_buf
; buf
!= NULL
; buf
= buf
->b_next
) {
3503 mutex_enter(&buf
->b_evict_lock
);
3505 mutex_exit(&buf
->b_evict_lock
);
3508 refcount_transfer(&nhdr
->b_l1hdr
.b_refcnt
, &hdr
->b_l1hdr
.b_refcnt
);
3509 (void) refcount_remove(&nhdr
->b_l1hdr
.b_refcnt
, FTAG
);
3512 arc_hdr_set_flags(nhdr
, ARC_FLAG_PROTECTED
);
3514 arc_hdr_clear_flags(nhdr
, ARC_FLAG_PROTECTED
);
3517 buf_discard_identity(hdr
);
3518 kmem_cache_free(ocache
, hdr
);
3524 * This function is used by the send / receive code to convert a newly
3525 * allocated arc_buf_t to one that is suitable for a raw encrypted write. It
3526 * is also used to allow the root objset block to be uupdated without altering
3527 * its embedded MACs. Both block types will always be uncompressed so we do not
3528 * have to worry about compression type or psize.
3531 arc_convert_to_raw(arc_buf_t
*buf
, uint64_t dsobj
, boolean_t byteorder
,
3532 dmu_object_type_t ot
, const uint8_t *salt
, const uint8_t *iv
,
3535 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
3537 ASSERT(ot
== DMU_OT_DNODE
|| ot
== DMU_OT_OBJSET
);
3538 ASSERT(HDR_HAS_L1HDR(hdr
));
3539 ASSERT3P(hdr
->b_l1hdr
.b_state
, ==, arc_anon
);
3541 buf
->b_flags
|= (ARC_BUF_FLAG_COMPRESSED
| ARC_BUF_FLAG_ENCRYPTED
);
3542 if (!HDR_PROTECTED(hdr
))
3543 hdr
= arc_hdr_realloc_crypt(hdr
, B_TRUE
);
3544 hdr
->b_crypt_hdr
.b_dsobj
= dsobj
;
3545 hdr
->b_crypt_hdr
.b_ot
= ot
;
3546 hdr
->b_l1hdr
.b_byteswap
= (byteorder
== ZFS_HOST_BYTEORDER
) ?
3547 DMU_BSWAP_NUMFUNCS
: DMU_OT_BYTESWAP(ot
);
3548 if (!arc_hdr_has_uncompressed_buf(hdr
))
3549 arc_cksum_free(hdr
);
3552 bcopy(salt
, hdr
->b_crypt_hdr
.b_salt
, ZIO_DATA_SALT_LEN
);
3554 bcopy(iv
, hdr
->b_crypt_hdr
.b_iv
, ZIO_DATA_IV_LEN
);
3556 bcopy(mac
, hdr
->b_crypt_hdr
.b_mac
, ZIO_DATA_MAC_LEN
);
3560 * Allocate a new arc_buf_hdr_t and arc_buf_t and return the buf to the caller.
3561 * The buf is returned thawed since we expect the consumer to modify it.
3564 arc_alloc_buf(spa_t
*spa
, void *tag
, arc_buf_contents_t type
, int32_t size
)
3566 arc_buf_hdr_t
*hdr
= arc_hdr_alloc(spa_load_guid(spa
), size
, size
,
3567 B_FALSE
, ZIO_COMPRESS_OFF
, type
, B_FALSE
);
3568 ASSERT(!MUTEX_HELD(HDR_LOCK(hdr
)));
3570 arc_buf_t
*buf
= NULL
;
3571 VERIFY0(arc_buf_alloc_impl(hdr
, spa
, 0, tag
, B_FALSE
, B_FALSE
,
3572 B_FALSE
, B_FALSE
, &buf
));
3579 * Allocate a compressed buf in the same manner as arc_alloc_buf. Don't use this
3580 * for bufs containing metadata.
3583 arc_alloc_compressed_buf(spa_t
*spa
, void *tag
, uint64_t psize
, uint64_t lsize
,
3584 enum zio_compress compression_type
)
3586 ASSERT3U(lsize
, >, 0);
3587 ASSERT3U(lsize
, >=, psize
);
3588 ASSERT3U(compression_type
, >, ZIO_COMPRESS_OFF
);
3589 ASSERT3U(compression_type
, <, ZIO_COMPRESS_FUNCTIONS
);
3591 arc_buf_hdr_t
*hdr
= arc_hdr_alloc(spa_load_guid(spa
), psize
, lsize
,
3592 B_FALSE
, compression_type
, ARC_BUFC_DATA
, B_FALSE
);
3593 ASSERT(!MUTEX_HELD(HDR_LOCK(hdr
)));
3595 arc_buf_t
*buf
= NULL
;
3596 VERIFY0(arc_buf_alloc_impl(hdr
, spa
, 0, tag
, B_FALSE
,
3597 B_TRUE
, B_FALSE
, B_FALSE
, &buf
));
3599 ASSERT3P(hdr
->b_l1hdr
.b_freeze_cksum
, ==, NULL
);
3601 if (!arc_buf_is_shared(buf
)) {
3603 * To ensure that the hdr has the correct data in it if we call
3604 * arc_untransform() on this buf before it's been written to
3605 * disk, it's easiest if we just set up sharing between the
3608 ASSERT(!abd_is_linear(hdr
->b_l1hdr
.b_pabd
));
3609 arc_hdr_free_abd(hdr
, B_FALSE
);
3610 arc_share_buf(hdr
, buf
);
3617 arc_alloc_raw_buf(spa_t
*spa
, void *tag
, uint64_t dsobj
, boolean_t byteorder
,
3618 const uint8_t *salt
, const uint8_t *iv
, const uint8_t *mac
,
3619 dmu_object_type_t ot
, uint64_t psize
, uint64_t lsize
,
3620 enum zio_compress compression_type
)
3624 arc_buf_contents_t type
= DMU_OT_IS_METADATA(ot
) ?
3625 ARC_BUFC_METADATA
: ARC_BUFC_DATA
;
3627 ASSERT3U(lsize
, >, 0);
3628 ASSERT3U(lsize
, >=, psize
);
3629 ASSERT3U(compression_type
, >=, ZIO_COMPRESS_OFF
);
3630 ASSERT3U(compression_type
, <, ZIO_COMPRESS_FUNCTIONS
);
3632 hdr
= arc_hdr_alloc(spa_load_guid(spa
), psize
, lsize
, B_TRUE
,
3633 compression_type
, type
, B_TRUE
);
3634 ASSERT(!MUTEX_HELD(HDR_LOCK(hdr
)));
3636 hdr
->b_crypt_hdr
.b_dsobj
= dsobj
;
3637 hdr
->b_crypt_hdr
.b_ot
= ot
;
3638 hdr
->b_l1hdr
.b_byteswap
= (byteorder
== ZFS_HOST_BYTEORDER
) ?
3639 DMU_BSWAP_NUMFUNCS
: DMU_OT_BYTESWAP(ot
);
3640 bcopy(salt
, hdr
->b_crypt_hdr
.b_salt
, ZIO_DATA_SALT_LEN
);
3641 bcopy(iv
, hdr
->b_crypt_hdr
.b_iv
, ZIO_DATA_IV_LEN
);
3642 bcopy(mac
, hdr
->b_crypt_hdr
.b_mac
, ZIO_DATA_MAC_LEN
);
3645 * This buffer will be considered encrypted even if the ot is not an
3646 * encrypted type. It will become authenticated instead in
3647 * arc_write_ready().
3650 VERIFY0(arc_buf_alloc_impl(hdr
, spa
, dsobj
, tag
, B_TRUE
, B_TRUE
,
3651 B_FALSE
, B_FALSE
, &buf
));
3653 ASSERT3P(hdr
->b_l1hdr
.b_freeze_cksum
, ==, NULL
);
3659 arc_hdr_l2hdr_destroy(arc_buf_hdr_t
*hdr
)
3661 l2arc_buf_hdr_t
*l2hdr
= &hdr
->b_l2hdr
;
3662 l2arc_dev_t
*dev
= l2hdr
->b_dev
;
3663 uint64_t psize
= arc_hdr_size(hdr
);
3665 ASSERT(MUTEX_HELD(&dev
->l2ad_mtx
));
3666 ASSERT(HDR_HAS_L2HDR(hdr
));
3668 list_remove(&dev
->l2ad_buflist
, hdr
);
3670 ARCSTAT_INCR(arcstat_l2_psize
, -psize
);
3671 ARCSTAT_INCR(arcstat_l2_lsize
, -HDR_GET_LSIZE(hdr
));
3673 vdev_space_update(dev
->l2ad_vdev
, -psize
, 0, 0);
3675 (void) refcount_remove_many(&dev
->l2ad_alloc
, psize
, hdr
);
3676 arc_hdr_clear_flags(hdr
, ARC_FLAG_HAS_L2HDR
);
3680 arc_hdr_destroy(arc_buf_hdr_t
*hdr
)
3682 if (HDR_HAS_L1HDR(hdr
)) {
3683 ASSERT(hdr
->b_l1hdr
.b_buf
== NULL
||
3684 hdr
->b_l1hdr
.b_bufcnt
> 0);
3685 ASSERT(refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
));
3686 ASSERT3P(hdr
->b_l1hdr
.b_state
, ==, arc_anon
);
3688 ASSERT(!HDR_IO_IN_PROGRESS(hdr
));
3689 ASSERT(!HDR_IN_HASH_TABLE(hdr
));
3691 if (!HDR_EMPTY(hdr
))
3692 buf_discard_identity(hdr
);
3694 if (HDR_HAS_L2HDR(hdr
)) {
3695 l2arc_dev_t
*dev
= hdr
->b_l2hdr
.b_dev
;
3696 boolean_t buflist_held
= MUTEX_HELD(&dev
->l2ad_mtx
);
3699 mutex_enter(&dev
->l2ad_mtx
);
3702 * Even though we checked this conditional above, we
3703 * need to check this again now that we have the
3704 * l2ad_mtx. This is because we could be racing with
3705 * another thread calling l2arc_evict() which might have
3706 * destroyed this header's L2 portion as we were waiting
3707 * to acquire the l2ad_mtx. If that happens, we don't
3708 * want to re-destroy the header's L2 portion.
3710 if (HDR_HAS_L2HDR(hdr
))
3711 arc_hdr_l2hdr_destroy(hdr
);
3714 mutex_exit(&dev
->l2ad_mtx
);
3717 if (HDR_HAS_L1HDR(hdr
)) {
3718 arc_cksum_free(hdr
);
3720 while (hdr
->b_l1hdr
.b_buf
!= NULL
)
3721 arc_buf_destroy_impl(hdr
->b_l1hdr
.b_buf
);
3723 if (hdr
->b_l1hdr
.b_pabd
!= NULL
) {
3724 arc_hdr_free_abd(hdr
, B_FALSE
);
3727 if (HDR_HAS_RABD(hdr
))
3728 arc_hdr_free_abd(hdr
, B_TRUE
);
3731 ASSERT3P(hdr
->b_hash_next
, ==, NULL
);
3732 if (HDR_HAS_L1HDR(hdr
)) {
3733 ASSERT(!multilist_link_active(&hdr
->b_l1hdr
.b_arc_node
));
3734 ASSERT3P(hdr
->b_l1hdr
.b_acb
, ==, NULL
);
3736 if (!HDR_PROTECTED(hdr
)) {
3737 kmem_cache_free(hdr_full_cache
, hdr
);
3739 kmem_cache_free(hdr_full_crypt_cache
, hdr
);
3742 kmem_cache_free(hdr_l2only_cache
, hdr
);
3747 arc_buf_destroy(arc_buf_t
*buf
, void* tag
)
3749 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
3750 kmutex_t
*hash_lock
= HDR_LOCK(hdr
);
3752 if (hdr
->b_l1hdr
.b_state
== arc_anon
) {
3753 ASSERT3U(hdr
->b_l1hdr
.b_bufcnt
, ==, 1);
3754 ASSERT(!HDR_IO_IN_PROGRESS(hdr
));
3755 VERIFY0(remove_reference(hdr
, NULL
, tag
));
3756 arc_hdr_destroy(hdr
);
3760 mutex_enter(hash_lock
);
3761 ASSERT3P(hdr
, ==, buf
->b_hdr
);
3762 ASSERT(hdr
->b_l1hdr
.b_bufcnt
> 0);
3763 ASSERT3P(hash_lock
, ==, HDR_LOCK(hdr
));
3764 ASSERT3P(hdr
->b_l1hdr
.b_state
, !=, arc_anon
);
3765 ASSERT3P(buf
->b_data
, !=, NULL
);
3767 (void) remove_reference(hdr
, hash_lock
, tag
);
3768 arc_buf_destroy_impl(buf
);
3769 mutex_exit(hash_lock
);
3773 * Evict the arc_buf_hdr that is provided as a parameter. The resultant
3774 * state of the header is dependent on its state prior to entering this
3775 * function. The following transitions are possible:
3777 * - arc_mru -> arc_mru_ghost
3778 * - arc_mfu -> arc_mfu_ghost
3779 * - arc_mru_ghost -> arc_l2c_only
3780 * - arc_mru_ghost -> deleted
3781 * - arc_mfu_ghost -> arc_l2c_only
3782 * - arc_mfu_ghost -> deleted
3785 arc_evict_hdr(arc_buf_hdr_t
*hdr
, kmutex_t
*hash_lock
)
3787 arc_state_t
*evicted_state
, *state
;
3788 int64_t bytes_evicted
= 0;
3789 int min_lifetime
= HDR_PRESCIENT_PREFETCH(hdr
) ?
3790 arc_min_prescient_prefetch_ms
: arc_min_prefetch_ms
;
3792 ASSERT(MUTEX_HELD(hash_lock
));
3793 ASSERT(HDR_HAS_L1HDR(hdr
));
3795 state
= hdr
->b_l1hdr
.b_state
;
3796 if (GHOST_STATE(state
)) {
3797 ASSERT(!HDR_IO_IN_PROGRESS(hdr
));
3798 ASSERT3P(hdr
->b_l1hdr
.b_buf
, ==, NULL
);
3801 * l2arc_write_buffers() relies on a header's L1 portion
3802 * (i.e. its b_pabd field) during it's write phase.
3803 * Thus, we cannot push a header onto the arc_l2c_only
3804 * state (removing its L1 piece) until the header is
3805 * done being written to the l2arc.
3807 if (HDR_HAS_L2HDR(hdr
) && HDR_L2_WRITING(hdr
)) {
3808 ARCSTAT_BUMP(arcstat_evict_l2_skip
);
3809 return (bytes_evicted
);
3812 ARCSTAT_BUMP(arcstat_deleted
);
3813 bytes_evicted
+= HDR_GET_LSIZE(hdr
);
3815 DTRACE_PROBE1(arc__delete
, arc_buf_hdr_t
*, hdr
);
3817 if (HDR_HAS_L2HDR(hdr
)) {
3818 ASSERT(hdr
->b_l1hdr
.b_pabd
== NULL
);
3819 ASSERT(!HDR_HAS_RABD(hdr
));
3821 * This buffer is cached on the 2nd Level ARC;
3822 * don't destroy the header.
3824 arc_change_state(arc_l2c_only
, hdr
, hash_lock
);
3826 * dropping from L1+L2 cached to L2-only,
3827 * realloc to remove the L1 header.
3829 hdr
= arc_hdr_realloc(hdr
, hdr_full_cache
,
3832 arc_change_state(arc_anon
, hdr
, hash_lock
);
3833 arc_hdr_destroy(hdr
);
3835 return (bytes_evicted
);
3838 ASSERT(state
== arc_mru
|| state
== arc_mfu
);
3839 evicted_state
= (state
== arc_mru
) ? arc_mru_ghost
: arc_mfu_ghost
;
3841 /* prefetch buffers have a minimum lifespan */
3842 if (HDR_IO_IN_PROGRESS(hdr
) ||
3843 ((hdr
->b_flags
& (ARC_FLAG_PREFETCH
| ARC_FLAG_INDIRECT
)) &&
3844 ddi_get_lbolt() - hdr
->b_l1hdr
.b_arc_access
< min_lifetime
* hz
)) {
3845 ARCSTAT_BUMP(arcstat_evict_skip
);
3846 return (bytes_evicted
);
3849 ASSERT0(refcount_count(&hdr
->b_l1hdr
.b_refcnt
));
3850 while (hdr
->b_l1hdr
.b_buf
) {
3851 arc_buf_t
*buf
= hdr
->b_l1hdr
.b_buf
;
3852 if (!mutex_tryenter(&buf
->b_evict_lock
)) {
3853 ARCSTAT_BUMP(arcstat_mutex_miss
);
3856 if (buf
->b_data
!= NULL
)
3857 bytes_evicted
+= HDR_GET_LSIZE(hdr
);
3858 mutex_exit(&buf
->b_evict_lock
);
3859 arc_buf_destroy_impl(buf
);
3862 if (HDR_HAS_L2HDR(hdr
)) {
3863 ARCSTAT_INCR(arcstat_evict_l2_cached
, HDR_GET_LSIZE(hdr
));
3865 if (l2arc_write_eligible(hdr
->b_spa
, hdr
)) {
3866 ARCSTAT_INCR(arcstat_evict_l2_eligible
,
3867 HDR_GET_LSIZE(hdr
));
3869 ARCSTAT_INCR(arcstat_evict_l2_ineligible
,
3870 HDR_GET_LSIZE(hdr
));
3874 if (hdr
->b_l1hdr
.b_bufcnt
== 0) {
3875 arc_cksum_free(hdr
);
3877 bytes_evicted
+= arc_hdr_size(hdr
);
3880 * If this hdr is being evicted and has a compressed
3881 * buffer then we discard it here before we change states.
3882 * This ensures that the accounting is updated correctly
3883 * in arc_free_data_impl().
3885 if (hdr
->b_l1hdr
.b_pabd
!= NULL
)
3886 arc_hdr_free_abd(hdr
, B_FALSE
);
3888 if (HDR_HAS_RABD(hdr
))
3889 arc_hdr_free_abd(hdr
, B_TRUE
);
3891 arc_change_state(evicted_state
, hdr
, hash_lock
);
3892 ASSERT(HDR_IN_HASH_TABLE(hdr
));
3893 arc_hdr_set_flags(hdr
, ARC_FLAG_IN_HASH_TABLE
);
3894 DTRACE_PROBE1(arc__evict
, arc_buf_hdr_t
*, hdr
);
3897 return (bytes_evicted
);
3901 arc_evict_state_impl(multilist_t
*ml
, int idx
, arc_buf_hdr_t
*marker
,
3902 uint64_t spa
, int64_t bytes
)
3904 multilist_sublist_t
*mls
;
3905 uint64_t bytes_evicted
= 0;
3907 kmutex_t
*hash_lock
;
3908 int evict_count
= 0;
3910 ASSERT3P(marker
, !=, NULL
);
3911 IMPLY(bytes
< 0, bytes
== ARC_EVICT_ALL
);
3913 mls
= multilist_sublist_lock(ml
, idx
);
3915 for (hdr
= multilist_sublist_prev(mls
, marker
); hdr
!= NULL
;
3916 hdr
= multilist_sublist_prev(mls
, marker
)) {
3917 if ((bytes
!= ARC_EVICT_ALL
&& bytes_evicted
>= bytes
) ||
3918 (evict_count
>= zfs_arc_evict_batch_limit
))
3922 * To keep our iteration location, move the marker
3923 * forward. Since we're not holding hdr's hash lock, we
3924 * must be very careful and not remove 'hdr' from the
3925 * sublist. Otherwise, other consumers might mistake the
3926 * 'hdr' as not being on a sublist when they call the
3927 * multilist_link_active() function (they all rely on
3928 * the hash lock protecting concurrent insertions and
3929 * removals). multilist_sublist_move_forward() was
3930 * specifically implemented to ensure this is the case
3931 * (only 'marker' will be removed and re-inserted).
3933 multilist_sublist_move_forward(mls
, marker
);
3936 * The only case where the b_spa field should ever be
3937 * zero, is the marker headers inserted by
3938 * arc_evict_state(). It's possible for multiple threads
3939 * to be calling arc_evict_state() concurrently (e.g.
3940 * dsl_pool_close() and zio_inject_fault()), so we must
3941 * skip any markers we see from these other threads.
3943 if (hdr
->b_spa
== 0)
3946 /* we're only interested in evicting buffers of a certain spa */
3947 if (spa
!= 0 && hdr
->b_spa
!= spa
) {
3948 ARCSTAT_BUMP(arcstat_evict_skip
);
3952 hash_lock
= HDR_LOCK(hdr
);
3955 * We aren't calling this function from any code path
3956 * that would already be holding a hash lock, so we're
3957 * asserting on this assumption to be defensive in case
3958 * this ever changes. Without this check, it would be
3959 * possible to incorrectly increment arcstat_mutex_miss
3960 * below (e.g. if the code changed such that we called
3961 * this function with a hash lock held).
3963 ASSERT(!MUTEX_HELD(hash_lock
));
3965 if (mutex_tryenter(hash_lock
)) {
3966 uint64_t evicted
= arc_evict_hdr(hdr
, hash_lock
);
3967 mutex_exit(hash_lock
);
3969 bytes_evicted
+= evicted
;
3972 * If evicted is zero, arc_evict_hdr() must have
3973 * decided to skip this header, don't increment
3974 * evict_count in this case.
3980 * If arc_size isn't overflowing, signal any
3981 * threads that might happen to be waiting.
3983 * For each header evicted, we wake up a single
3984 * thread. If we used cv_broadcast, we could
3985 * wake up "too many" threads causing arc_size
3986 * to significantly overflow arc_c; since
3987 * arc_get_data_impl() doesn't check for overflow
3988 * when it's woken up (it doesn't because it's
3989 * possible for the ARC to be overflowing while
3990 * full of un-evictable buffers, and the
3991 * function should proceed in this case).
3993 * If threads are left sleeping, due to not
3994 * using cv_broadcast, they will be woken up
3995 * just before arc_reclaim_thread() sleeps.
3997 mutex_enter(&arc_reclaim_lock
);
3998 if (!arc_is_overflowing())
3999 cv_signal(&arc_reclaim_waiters_cv
);
4000 mutex_exit(&arc_reclaim_lock
);
4002 ARCSTAT_BUMP(arcstat_mutex_miss
);
4006 multilist_sublist_unlock(mls
);
4008 return (bytes_evicted
);
4012 * Evict buffers from the given arc state, until we've removed the
4013 * specified number of bytes. Move the removed buffers to the
4014 * appropriate evict state.
4016 * This function makes a "best effort". It skips over any buffers
4017 * it can't get a hash_lock on, and so, may not catch all candidates.
4018 * It may also return without evicting as much space as requested.
4020 * If bytes is specified using the special value ARC_EVICT_ALL, this
4021 * will evict all available (i.e. unlocked and evictable) buffers from
4022 * the given arc state; which is used by arc_flush().
4025 arc_evict_state(arc_state_t
*state
, uint64_t spa
, int64_t bytes
,
4026 arc_buf_contents_t type
)
4028 uint64_t total_evicted
= 0;
4029 multilist_t
*ml
= state
->arcs_list
[type
];
4031 arc_buf_hdr_t
**markers
;
4033 IMPLY(bytes
< 0, bytes
== ARC_EVICT_ALL
);
4035 num_sublists
= multilist_get_num_sublists(ml
);
4038 * If we've tried to evict from each sublist, made some
4039 * progress, but still have not hit the target number of bytes
4040 * to evict, we want to keep trying. The markers allow us to
4041 * pick up where we left off for each individual sublist, rather
4042 * than starting from the tail each time.
4044 markers
= kmem_zalloc(sizeof (*markers
) * num_sublists
, KM_SLEEP
);
4045 for (int i
= 0; i
< num_sublists
; i
++) {
4046 multilist_sublist_t
*mls
;
4048 markers
[i
] = kmem_cache_alloc(hdr_full_cache
, KM_SLEEP
);
4051 * A b_spa of 0 is used to indicate that this header is
4052 * a marker. This fact is used in arc_adjust_type() and
4053 * arc_evict_state_impl().
4055 markers
[i
]->b_spa
= 0;
4057 mls
= multilist_sublist_lock(ml
, i
);
4058 multilist_sublist_insert_tail(mls
, markers
[i
]);
4059 multilist_sublist_unlock(mls
);
4063 * While we haven't hit our target number of bytes to evict, or
4064 * we're evicting all available buffers.
4066 while (total_evicted
< bytes
|| bytes
== ARC_EVICT_ALL
) {
4067 int sublist_idx
= multilist_get_random_index(ml
);
4068 uint64_t scan_evicted
= 0;
4071 * Try to reduce pinned dnodes with a floor of arc_dnode_limit.
4072 * Request that 10% of the LRUs be scanned by the superblock
4075 if (type
== ARC_BUFC_DATA
&& arc_dnode_size
> arc_dnode_limit
)
4076 arc_prune_async((arc_dnode_size
- arc_dnode_limit
) /
4077 sizeof (dnode_t
) / zfs_arc_dnode_reduce_percent
);
4080 * Start eviction using a randomly selected sublist,
4081 * this is to try and evenly balance eviction across all
4082 * sublists. Always starting at the same sublist
4083 * (e.g. index 0) would cause evictions to favor certain
4084 * sublists over others.
4086 for (int i
= 0; i
< num_sublists
; i
++) {
4087 uint64_t bytes_remaining
;
4088 uint64_t bytes_evicted
;
4090 if (bytes
== ARC_EVICT_ALL
)
4091 bytes_remaining
= ARC_EVICT_ALL
;
4092 else if (total_evicted
< bytes
)
4093 bytes_remaining
= bytes
- total_evicted
;
4097 bytes_evicted
= arc_evict_state_impl(ml
, sublist_idx
,
4098 markers
[sublist_idx
], spa
, bytes_remaining
);
4100 scan_evicted
+= bytes_evicted
;
4101 total_evicted
+= bytes_evicted
;
4103 /* we've reached the end, wrap to the beginning */
4104 if (++sublist_idx
>= num_sublists
)
4109 * If we didn't evict anything during this scan, we have
4110 * no reason to believe we'll evict more during another
4111 * scan, so break the loop.
4113 if (scan_evicted
== 0) {
4114 /* This isn't possible, let's make that obvious */
4115 ASSERT3S(bytes
, !=, 0);
4118 * When bytes is ARC_EVICT_ALL, the only way to
4119 * break the loop is when scan_evicted is zero.
4120 * In that case, we actually have evicted enough,
4121 * so we don't want to increment the kstat.
4123 if (bytes
!= ARC_EVICT_ALL
) {
4124 ASSERT3S(total_evicted
, <, bytes
);
4125 ARCSTAT_BUMP(arcstat_evict_not_enough
);
4132 for (int i
= 0; i
< num_sublists
; i
++) {
4133 multilist_sublist_t
*mls
= multilist_sublist_lock(ml
, i
);
4134 multilist_sublist_remove(mls
, markers
[i
]);
4135 multilist_sublist_unlock(mls
);
4137 kmem_cache_free(hdr_full_cache
, markers
[i
]);
4139 kmem_free(markers
, sizeof (*markers
) * num_sublists
);
4141 return (total_evicted
);
4145 * Flush all "evictable" data of the given type from the arc state
4146 * specified. This will not evict any "active" buffers (i.e. referenced).
4148 * When 'retry' is set to B_FALSE, the function will make a single pass
4149 * over the state and evict any buffers that it can. Since it doesn't
4150 * continually retry the eviction, it might end up leaving some buffers
4151 * in the ARC due to lock misses.
4153 * When 'retry' is set to B_TRUE, the function will continually retry the
4154 * eviction until *all* evictable buffers have been removed from the
4155 * state. As a result, if concurrent insertions into the state are
4156 * allowed (e.g. if the ARC isn't shutting down), this function might
4157 * wind up in an infinite loop, continually trying to evict buffers.
4160 arc_flush_state(arc_state_t
*state
, uint64_t spa
, arc_buf_contents_t type
,
4163 uint64_t evicted
= 0;
4165 while (refcount_count(&state
->arcs_esize
[type
]) != 0) {
4166 evicted
+= arc_evict_state(state
, spa
, ARC_EVICT_ALL
, type
);
4176 * Helper function for arc_prune_async() it is responsible for safely
4177 * handling the execution of a registered arc_prune_func_t.
4180 arc_prune_task(void *ptr
)
4182 arc_prune_t
*ap
= (arc_prune_t
*)ptr
;
4183 arc_prune_func_t
*func
= ap
->p_pfunc
;
4186 func(ap
->p_adjust
, ap
->p_private
);
4188 refcount_remove(&ap
->p_refcnt
, func
);
4192 * Notify registered consumers they must drop holds on a portion of the ARC
4193 * buffered they reference. This provides a mechanism to ensure the ARC can
4194 * honor the arc_meta_limit and reclaim otherwise pinned ARC buffers. This
4195 * is analogous to dnlc_reduce_cache() but more generic.
4197 * This operation is performed asynchronously so it may be safely called
4198 * in the context of the arc_reclaim_thread(). A reference is taken here
4199 * for each registered arc_prune_t and the arc_prune_task() is responsible
4200 * for releasing it once the registered arc_prune_func_t has completed.
4203 arc_prune_async(int64_t adjust
)
4207 mutex_enter(&arc_prune_mtx
);
4208 for (ap
= list_head(&arc_prune_list
); ap
!= NULL
;
4209 ap
= list_next(&arc_prune_list
, ap
)) {
4211 if (refcount_count(&ap
->p_refcnt
) >= 2)
4214 refcount_add(&ap
->p_refcnt
, ap
->p_pfunc
);
4215 ap
->p_adjust
= adjust
;
4216 if (taskq_dispatch(arc_prune_taskq
, arc_prune_task
,
4217 ap
, TQ_SLEEP
) == TASKQID_INVALID
) {
4218 refcount_remove(&ap
->p_refcnt
, ap
->p_pfunc
);
4221 ARCSTAT_BUMP(arcstat_prune
);
4223 mutex_exit(&arc_prune_mtx
);
4227 * Evict the specified number of bytes from the state specified,
4228 * restricting eviction to the spa and type given. This function
4229 * prevents us from trying to evict more from a state's list than
4230 * is "evictable", and to skip evicting altogether when passed a
4231 * negative value for "bytes". In contrast, arc_evict_state() will
4232 * evict everything it can, when passed a negative value for "bytes".
4235 arc_adjust_impl(arc_state_t
*state
, uint64_t spa
, int64_t bytes
,
4236 arc_buf_contents_t type
)
4240 if (bytes
> 0 && refcount_count(&state
->arcs_esize
[type
]) > 0) {
4241 delta
= MIN(refcount_count(&state
->arcs_esize
[type
]), bytes
);
4242 return (arc_evict_state(state
, spa
, delta
, type
));
4249 * The goal of this function is to evict enough meta data buffers from the
4250 * ARC in order to enforce the arc_meta_limit. Achieving this is slightly
4251 * more complicated than it appears because it is common for data buffers
4252 * to have holds on meta data buffers. In addition, dnode meta data buffers
4253 * will be held by the dnodes in the block preventing them from being freed.
4254 * This means we can't simply traverse the ARC and expect to always find
4255 * enough unheld meta data buffer to release.
4257 * Therefore, this function has been updated to make alternating passes
4258 * over the ARC releasing data buffers and then newly unheld meta data
4259 * buffers. This ensures forward progress is maintained and arc_meta_used
4260 * will decrease. Normally this is sufficient, but if required the ARC
4261 * will call the registered prune callbacks causing dentry and inodes to
4262 * be dropped from the VFS cache. This will make dnode meta data buffers
4263 * available for reclaim.
4266 arc_adjust_meta_balanced(void)
4268 int64_t delta
, prune
= 0, adjustmnt
;
4269 uint64_t total_evicted
= 0;
4270 arc_buf_contents_t type
= ARC_BUFC_DATA
;
4271 int restarts
= MAX(zfs_arc_meta_adjust_restarts
, 0);
4275 * This slightly differs than the way we evict from the mru in
4276 * arc_adjust because we don't have a "target" value (i.e. no
4277 * "meta" arc_p). As a result, I think we can completely
4278 * cannibalize the metadata in the MRU before we evict the
4279 * metadata from the MFU. I think we probably need to implement a
4280 * "metadata arc_p" value to do this properly.
4282 adjustmnt
= arc_meta_used
- arc_meta_limit
;
4284 if (adjustmnt
> 0 && refcount_count(&arc_mru
->arcs_esize
[type
]) > 0) {
4285 delta
= MIN(refcount_count(&arc_mru
->arcs_esize
[type
]),
4287 total_evicted
+= arc_adjust_impl(arc_mru
, 0, delta
, type
);
4292 * We can't afford to recalculate adjustmnt here. If we do,
4293 * new metadata buffers can sneak into the MRU or ANON lists,
4294 * thus penalize the MFU metadata. Although the fudge factor is
4295 * small, it has been empirically shown to be significant for
4296 * certain workloads (e.g. creating many empty directories). As
4297 * such, we use the original calculation for adjustmnt, and
4298 * simply decrement the amount of data evicted from the MRU.
4301 if (adjustmnt
> 0 && refcount_count(&arc_mfu
->arcs_esize
[type
]) > 0) {
4302 delta
= MIN(refcount_count(&arc_mfu
->arcs_esize
[type
]),
4304 total_evicted
+= arc_adjust_impl(arc_mfu
, 0, delta
, type
);
4307 adjustmnt
= arc_meta_used
- arc_meta_limit
;
4309 if (adjustmnt
> 0 &&
4310 refcount_count(&arc_mru_ghost
->arcs_esize
[type
]) > 0) {
4311 delta
= MIN(adjustmnt
,
4312 refcount_count(&arc_mru_ghost
->arcs_esize
[type
]));
4313 total_evicted
+= arc_adjust_impl(arc_mru_ghost
, 0, delta
, type
);
4317 if (adjustmnt
> 0 &&
4318 refcount_count(&arc_mfu_ghost
->arcs_esize
[type
]) > 0) {
4319 delta
= MIN(adjustmnt
,
4320 refcount_count(&arc_mfu_ghost
->arcs_esize
[type
]));
4321 total_evicted
+= arc_adjust_impl(arc_mfu_ghost
, 0, delta
, type
);
4325 * If after attempting to make the requested adjustment to the ARC
4326 * the meta limit is still being exceeded then request that the
4327 * higher layers drop some cached objects which have holds on ARC
4328 * meta buffers. Requests to the upper layers will be made with
4329 * increasingly large scan sizes until the ARC is below the limit.
4331 if (arc_meta_used
> arc_meta_limit
) {
4332 if (type
== ARC_BUFC_DATA
) {
4333 type
= ARC_BUFC_METADATA
;
4335 type
= ARC_BUFC_DATA
;
4337 if (zfs_arc_meta_prune
) {
4338 prune
+= zfs_arc_meta_prune
;
4339 arc_prune_async(prune
);
4348 return (total_evicted
);
4352 * Evict metadata buffers from the cache, such that arc_meta_used is
4353 * capped by the arc_meta_limit tunable.
4356 arc_adjust_meta_only(void)
4358 uint64_t total_evicted
= 0;
4362 * If we're over the meta limit, we want to evict enough
4363 * metadata to get back under the meta limit. We don't want to
4364 * evict so much that we drop the MRU below arc_p, though. If
4365 * we're over the meta limit more than we're over arc_p, we
4366 * evict some from the MRU here, and some from the MFU below.
4368 target
= MIN((int64_t)(arc_meta_used
- arc_meta_limit
),
4369 (int64_t)(refcount_count(&arc_anon
->arcs_size
) +
4370 refcount_count(&arc_mru
->arcs_size
) - arc_p
));
4372 total_evicted
+= arc_adjust_impl(arc_mru
, 0, target
, ARC_BUFC_METADATA
);
4375 * Similar to the above, we want to evict enough bytes to get us
4376 * below the meta limit, but not so much as to drop us below the
4377 * space allotted to the MFU (which is defined as arc_c - arc_p).
4379 target
= MIN((int64_t)(arc_meta_used
- arc_meta_limit
),
4380 (int64_t)(refcount_count(&arc_mfu
->arcs_size
) - (arc_c
- arc_p
)));
4382 total_evicted
+= arc_adjust_impl(arc_mfu
, 0, target
, ARC_BUFC_METADATA
);
4384 return (total_evicted
);
4388 arc_adjust_meta(void)
4390 if (zfs_arc_meta_strategy
== ARC_STRATEGY_META_ONLY
)
4391 return (arc_adjust_meta_only());
4393 return (arc_adjust_meta_balanced());
4397 * Return the type of the oldest buffer in the given arc state
4399 * This function will select a random sublist of type ARC_BUFC_DATA and
4400 * a random sublist of type ARC_BUFC_METADATA. The tail of each sublist
4401 * is compared, and the type which contains the "older" buffer will be
4404 static arc_buf_contents_t
4405 arc_adjust_type(arc_state_t
*state
)
4407 multilist_t
*data_ml
= state
->arcs_list
[ARC_BUFC_DATA
];
4408 multilist_t
*meta_ml
= state
->arcs_list
[ARC_BUFC_METADATA
];
4409 int data_idx
= multilist_get_random_index(data_ml
);
4410 int meta_idx
= multilist_get_random_index(meta_ml
);
4411 multilist_sublist_t
*data_mls
;
4412 multilist_sublist_t
*meta_mls
;
4413 arc_buf_contents_t type
;
4414 arc_buf_hdr_t
*data_hdr
;
4415 arc_buf_hdr_t
*meta_hdr
;
4418 * We keep the sublist lock until we're finished, to prevent
4419 * the headers from being destroyed via arc_evict_state().
4421 data_mls
= multilist_sublist_lock(data_ml
, data_idx
);
4422 meta_mls
= multilist_sublist_lock(meta_ml
, meta_idx
);
4425 * These two loops are to ensure we skip any markers that
4426 * might be at the tail of the lists due to arc_evict_state().
4429 for (data_hdr
= multilist_sublist_tail(data_mls
); data_hdr
!= NULL
;
4430 data_hdr
= multilist_sublist_prev(data_mls
, data_hdr
)) {
4431 if (data_hdr
->b_spa
!= 0)
4435 for (meta_hdr
= multilist_sublist_tail(meta_mls
); meta_hdr
!= NULL
;
4436 meta_hdr
= multilist_sublist_prev(meta_mls
, meta_hdr
)) {
4437 if (meta_hdr
->b_spa
!= 0)
4441 if (data_hdr
== NULL
&& meta_hdr
== NULL
) {
4442 type
= ARC_BUFC_DATA
;
4443 } else if (data_hdr
== NULL
) {
4444 ASSERT3P(meta_hdr
, !=, NULL
);
4445 type
= ARC_BUFC_METADATA
;
4446 } else if (meta_hdr
== NULL
) {
4447 ASSERT3P(data_hdr
, !=, NULL
);
4448 type
= ARC_BUFC_DATA
;
4450 ASSERT3P(data_hdr
, !=, NULL
);
4451 ASSERT3P(meta_hdr
, !=, NULL
);
4453 /* The headers can't be on the sublist without an L1 header */
4454 ASSERT(HDR_HAS_L1HDR(data_hdr
));
4455 ASSERT(HDR_HAS_L1HDR(meta_hdr
));
4457 if (data_hdr
->b_l1hdr
.b_arc_access
<
4458 meta_hdr
->b_l1hdr
.b_arc_access
) {
4459 type
= ARC_BUFC_DATA
;
4461 type
= ARC_BUFC_METADATA
;
4465 multilist_sublist_unlock(meta_mls
);
4466 multilist_sublist_unlock(data_mls
);
4472 * Evict buffers from the cache, such that arc_size is capped by arc_c.
4477 uint64_t total_evicted
= 0;
4482 * If we're over arc_meta_limit, we want to correct that before
4483 * potentially evicting data buffers below.
4485 total_evicted
+= arc_adjust_meta();
4490 * If we're over the target cache size, we want to evict enough
4491 * from the list to get back to our target size. We don't want
4492 * to evict too much from the MRU, such that it drops below
4493 * arc_p. So, if we're over our target cache size more than
4494 * the MRU is over arc_p, we'll evict enough to get back to
4495 * arc_p here, and then evict more from the MFU below.
4497 target
= MIN((int64_t)(arc_size
- arc_c
),
4498 (int64_t)(refcount_count(&arc_anon
->arcs_size
) +
4499 refcount_count(&arc_mru
->arcs_size
) + arc_meta_used
- arc_p
));
4502 * If we're below arc_meta_min, always prefer to evict data.
4503 * Otherwise, try to satisfy the requested number of bytes to
4504 * evict from the type which contains older buffers; in an
4505 * effort to keep newer buffers in the cache regardless of their
4506 * type. If we cannot satisfy the number of bytes from this
4507 * type, spill over into the next type.
4509 if (arc_adjust_type(arc_mru
) == ARC_BUFC_METADATA
&&
4510 arc_meta_used
> arc_meta_min
) {
4511 bytes
= arc_adjust_impl(arc_mru
, 0, target
, ARC_BUFC_METADATA
);
4512 total_evicted
+= bytes
;
4515 * If we couldn't evict our target number of bytes from
4516 * metadata, we try to get the rest from data.
4521 arc_adjust_impl(arc_mru
, 0, target
, ARC_BUFC_DATA
);
4523 bytes
= arc_adjust_impl(arc_mru
, 0, target
, ARC_BUFC_DATA
);
4524 total_evicted
+= bytes
;
4527 * If we couldn't evict our target number of bytes from
4528 * data, we try to get the rest from metadata.
4533 arc_adjust_impl(arc_mru
, 0, target
, ARC_BUFC_METADATA
);
4539 * Now that we've tried to evict enough from the MRU to get its
4540 * size back to arc_p, if we're still above the target cache
4541 * size, we evict the rest from the MFU.
4543 target
= arc_size
- arc_c
;
4545 if (arc_adjust_type(arc_mfu
) == ARC_BUFC_METADATA
&&
4546 arc_meta_used
> arc_meta_min
) {
4547 bytes
= arc_adjust_impl(arc_mfu
, 0, target
, ARC_BUFC_METADATA
);
4548 total_evicted
+= bytes
;
4551 * If we couldn't evict our target number of bytes from
4552 * metadata, we try to get the rest from data.
4557 arc_adjust_impl(arc_mfu
, 0, target
, ARC_BUFC_DATA
);
4559 bytes
= arc_adjust_impl(arc_mfu
, 0, target
, ARC_BUFC_DATA
);
4560 total_evicted
+= bytes
;
4563 * If we couldn't evict our target number of bytes from
4564 * data, we try to get the rest from data.
4569 arc_adjust_impl(arc_mfu
, 0, target
, ARC_BUFC_METADATA
);
4573 * Adjust ghost lists
4575 * In addition to the above, the ARC also defines target values
4576 * for the ghost lists. The sum of the mru list and mru ghost
4577 * list should never exceed the target size of the cache, and
4578 * the sum of the mru list, mfu list, mru ghost list, and mfu
4579 * ghost list should never exceed twice the target size of the
4580 * cache. The following logic enforces these limits on the ghost
4581 * caches, and evicts from them as needed.
4583 target
= refcount_count(&arc_mru
->arcs_size
) +
4584 refcount_count(&arc_mru_ghost
->arcs_size
) - arc_c
;
4586 bytes
= arc_adjust_impl(arc_mru_ghost
, 0, target
, ARC_BUFC_DATA
);
4587 total_evicted
+= bytes
;
4592 arc_adjust_impl(arc_mru_ghost
, 0, target
, ARC_BUFC_METADATA
);
4595 * We assume the sum of the mru list and mfu list is less than
4596 * or equal to arc_c (we enforced this above), which means we
4597 * can use the simpler of the two equations below:
4599 * mru + mfu + mru ghost + mfu ghost <= 2 * arc_c
4600 * mru ghost + mfu ghost <= arc_c
4602 target
= refcount_count(&arc_mru_ghost
->arcs_size
) +
4603 refcount_count(&arc_mfu_ghost
->arcs_size
) - arc_c
;
4605 bytes
= arc_adjust_impl(arc_mfu_ghost
, 0, target
, ARC_BUFC_DATA
);
4606 total_evicted
+= bytes
;
4611 arc_adjust_impl(arc_mfu_ghost
, 0, target
, ARC_BUFC_METADATA
);
4613 return (total_evicted
);
4617 arc_flush(spa_t
*spa
, boolean_t retry
)
4622 * If retry is B_TRUE, a spa must not be specified since we have
4623 * no good way to determine if all of a spa's buffers have been
4624 * evicted from an arc state.
4626 ASSERT(!retry
|| spa
== 0);
4629 guid
= spa_load_guid(spa
);
4631 (void) arc_flush_state(arc_mru
, guid
, ARC_BUFC_DATA
, retry
);
4632 (void) arc_flush_state(arc_mru
, guid
, ARC_BUFC_METADATA
, retry
);
4634 (void) arc_flush_state(arc_mfu
, guid
, ARC_BUFC_DATA
, retry
);
4635 (void) arc_flush_state(arc_mfu
, guid
, ARC_BUFC_METADATA
, retry
);
4637 (void) arc_flush_state(arc_mru_ghost
, guid
, ARC_BUFC_DATA
, retry
);
4638 (void) arc_flush_state(arc_mru_ghost
, guid
, ARC_BUFC_METADATA
, retry
);
4640 (void) arc_flush_state(arc_mfu_ghost
, guid
, ARC_BUFC_DATA
, retry
);
4641 (void) arc_flush_state(arc_mfu_ghost
, guid
, ARC_BUFC_METADATA
, retry
);
4645 arc_shrink(int64_t to_free
)
4649 if (c
> to_free
&& c
- to_free
> arc_c_min
) {
4650 arc_c
= c
- to_free
;
4651 atomic_add_64(&arc_p
, -(arc_p
>> arc_shrink_shift
));
4652 if (arc_c
> arc_size
)
4653 arc_c
= MAX(arc_size
, arc_c_min
);
4655 arc_p
= (arc_c
>> 1);
4656 ASSERT(arc_c
>= arc_c_min
);
4657 ASSERT((int64_t)arc_p
>= 0);
4662 if (arc_size
> arc_c
)
4663 (void) arc_adjust();
4667 * Return maximum amount of memory that we could possibly use. Reduced
4668 * to half of all memory in user space which is primarily used for testing.
4671 arc_all_memory(void)
4674 #ifdef CONFIG_HIGHMEM
4675 return (ptob(totalram_pages
- totalhigh_pages
));
4677 return (ptob(totalram_pages
));
4678 #endif /* CONFIG_HIGHMEM */
4680 return (ptob(physmem
) / 2);
4681 #endif /* _KERNEL */
4685 * Return the amount of memory that is considered free. In user space
4686 * which is primarily used for testing we pretend that free memory ranges
4687 * from 0-20% of all memory.
4690 arc_free_memory(void)
4693 #ifdef CONFIG_HIGHMEM
4696 return (ptob(si
.freeram
- si
.freehigh
));
4698 #ifdef ZFS_GLOBAL_NODE_PAGE_STATE
4699 return (ptob(nr_free_pages() +
4700 global_node_page_state(NR_INACTIVE_FILE
) +
4701 global_node_page_state(NR_INACTIVE_ANON
) +
4702 global_node_page_state(NR_SLAB_RECLAIMABLE
)));
4704 return (ptob(nr_free_pages() +
4705 global_page_state(NR_INACTIVE_FILE
) +
4706 global_page_state(NR_INACTIVE_ANON
) +
4707 global_page_state(NR_SLAB_RECLAIMABLE
)));
4708 #endif /* ZFS_GLOBAL_NODE_PAGE_STATE */
4709 #endif /* CONFIG_HIGHMEM */
4711 return (spa_get_random(arc_all_memory() * 20 / 100));
4712 #endif /* _KERNEL */
4715 typedef enum free_memory_reason_t
{
4720 FMR_PAGES_PP_MAXIMUM
,
4723 } free_memory_reason_t
;
4725 int64_t last_free_memory
;
4726 free_memory_reason_t last_free_reason
;
4730 * Additional reserve of pages for pp_reserve.
4732 int64_t arc_pages_pp_reserve
= 64;
4735 * Additional reserve of pages for swapfs.
4737 int64_t arc_swapfs_reserve
= 64;
4738 #endif /* _KERNEL */
4741 * Return the amount of memory that can be consumed before reclaim will be
4742 * needed. Positive if there is sufficient free memory, negative indicates
4743 * the amount of memory that needs to be freed up.
4746 arc_available_memory(void)
4748 int64_t lowest
= INT64_MAX
;
4749 free_memory_reason_t r
= FMR_UNKNOWN
;
4756 pgcnt_t needfree
= btop(arc_need_free
);
4757 pgcnt_t lotsfree
= btop(arc_sys_free
);
4758 pgcnt_t desfree
= 0;
4759 pgcnt_t freemem
= btop(arc_free_memory());
4763 n
= PAGESIZE
* (-needfree
);
4771 * check that we're out of range of the pageout scanner. It starts to
4772 * schedule paging if freemem is less than lotsfree and needfree.
4773 * lotsfree is the high-water mark for pageout, and needfree is the
4774 * number of needed free pages. We add extra pages here to make sure
4775 * the scanner doesn't start up while we're freeing memory.
4777 n
= PAGESIZE
* (freemem
- lotsfree
- needfree
- desfree
);
4785 * check to make sure that swapfs has enough space so that anon
4786 * reservations can still succeed. anon_resvmem() checks that the
4787 * availrmem is greater than swapfs_minfree, and the number of reserved
4788 * swap pages. We also add a bit of extra here just to prevent
4789 * circumstances from getting really dire.
4791 n
= PAGESIZE
* (availrmem
- swapfs_minfree
- swapfs_reserve
-
4792 desfree
- arc_swapfs_reserve
);
4795 r
= FMR_SWAPFS_MINFREE
;
4799 * Check that we have enough availrmem that memory locking (e.g., via
4800 * mlock(3C) or memcntl(2)) can still succeed. (pages_pp_maximum
4801 * stores the number of pages that cannot be locked; when availrmem
4802 * drops below pages_pp_maximum, page locking mechanisms such as
4803 * page_pp_lock() will fail.)
4805 n
= PAGESIZE
* (availrmem
- pages_pp_maximum
-
4806 arc_pages_pp_reserve
);
4809 r
= FMR_PAGES_PP_MAXIMUM
;
4815 * If we're on a 32-bit platform, it's possible that we'll exhaust the
4816 * kernel heap space before we ever run out of available physical
4817 * memory. Most checks of the size of the heap_area compare against
4818 * tune.t_minarmem, which is the minimum available real memory that we
4819 * can have in the system. However, this is generally fixed at 25 pages
4820 * which is so low that it's useless. In this comparison, we seek to
4821 * calculate the total heap-size, and reclaim if more than 3/4ths of the
4822 * heap is allocated. (Or, in the calculation, if less than 1/4th is
4825 n
= vmem_size(heap_arena
, VMEM_FREE
) -
4826 (vmem_size(heap_arena
, VMEM_FREE
| VMEM_ALLOC
) >> 2);
4834 * If zio data pages are being allocated out of a separate heap segment,
4835 * then enforce that the size of available vmem for this arena remains
4836 * above about 1/4th (1/(2^arc_zio_arena_free_shift)) free.
4838 * Note that reducing the arc_zio_arena_free_shift keeps more virtual
4839 * memory (in the zio_arena) free, which can avoid memory
4840 * fragmentation issues.
4842 if (zio_arena
!= NULL
) {
4843 n
= (int64_t)vmem_size(zio_arena
, VMEM_FREE
) -
4844 (vmem_size(zio_arena
, VMEM_ALLOC
) >>
4845 arc_zio_arena_free_shift
);
4852 /* Every 100 calls, free a small amount */
4853 if (spa_get_random(100) == 0)
4855 #endif /* _KERNEL */
4857 last_free_memory
= lowest
;
4858 last_free_reason
= r
;
4864 * Determine if the system is under memory pressure and is asking
4865 * to reclaim memory. A return value of B_TRUE indicates that the system
4866 * is under memory pressure and that the arc should adjust accordingly.
4869 arc_reclaim_needed(void)
4871 return (arc_available_memory() < 0);
4875 arc_kmem_reap_now(void)
4878 kmem_cache_t
*prev_cache
= NULL
;
4879 kmem_cache_t
*prev_data_cache
= NULL
;
4880 extern kmem_cache_t
*zio_buf_cache
[];
4881 extern kmem_cache_t
*zio_data_buf_cache
[];
4882 extern kmem_cache_t
*range_seg_cache
;
4885 if ((arc_meta_used
>= arc_meta_limit
) && zfs_arc_meta_prune
) {
4887 * We are exceeding our meta-data cache limit.
4888 * Prune some entries to release holds on meta-data.
4890 arc_prune_async(zfs_arc_meta_prune
);
4894 * Reclaim unused memory from all kmem caches.
4900 for (i
= 0; i
< SPA_MAXBLOCKSIZE
>> SPA_MINBLOCKSHIFT
; i
++) {
4902 /* reach upper limit of cache size on 32-bit */
4903 if (zio_buf_cache
[i
] == NULL
)
4906 if (zio_buf_cache
[i
] != prev_cache
) {
4907 prev_cache
= zio_buf_cache
[i
];
4908 kmem_cache_reap_now(zio_buf_cache
[i
]);
4910 if (zio_data_buf_cache
[i
] != prev_data_cache
) {
4911 prev_data_cache
= zio_data_buf_cache
[i
];
4912 kmem_cache_reap_now(zio_data_buf_cache
[i
]);
4915 kmem_cache_reap_now(buf_cache
);
4916 kmem_cache_reap_now(hdr_full_cache
);
4917 kmem_cache_reap_now(hdr_l2only_cache
);
4918 kmem_cache_reap_now(range_seg_cache
);
4920 if (zio_arena
!= NULL
) {
4922 * Ask the vmem arena to reclaim unused memory from its
4925 vmem_qcache_reap(zio_arena
);
4930 * Threads can block in arc_get_data_impl() waiting for this thread to evict
4931 * enough data and signal them to proceed. When this happens, the threads in
4932 * arc_get_data_impl() are sleeping while holding the hash lock for their
4933 * particular arc header. Thus, we must be careful to never sleep on a
4934 * hash lock in this thread. This is to prevent the following deadlock:
4936 * - Thread A sleeps on CV in arc_get_data_impl() holding hash lock "L",
4937 * waiting for the reclaim thread to signal it.
4939 * - arc_reclaim_thread() tries to acquire hash lock "L" using mutex_enter,
4940 * fails, and goes to sleep forever.
4942 * This possible deadlock is avoided by always acquiring a hash lock
4943 * using mutex_tryenter() from arc_reclaim_thread().
4947 arc_reclaim_thread(void *unused
)
4949 fstrans_cookie_t cookie
= spl_fstrans_mark();
4950 hrtime_t growtime
= 0;
4953 CALLB_CPR_INIT(&cpr
, &arc_reclaim_lock
, callb_generic_cpr
, FTAG
);
4955 mutex_enter(&arc_reclaim_lock
);
4956 while (!arc_reclaim_thread_exit
) {
4957 uint64_t evicted
= 0;
4958 uint64_t need_free
= arc_need_free
;
4959 arc_tuning_update();
4962 * This is necessary in order for the mdb ::arc dcmd to
4963 * show up to date information. Since the ::arc command
4964 * does not call the kstat's update function, without
4965 * this call, the command may show stale stats for the
4966 * anon, mru, mru_ghost, mfu, and mfu_ghost lists. Even
4967 * with this change, the data might be up to 1 second
4968 * out of date; but that should suffice. The arc_state_t
4969 * structures can be queried directly if more accurate
4970 * information is needed.
4973 if (arc_ksp
!= NULL
)
4974 arc_ksp
->ks_update(arc_ksp
, KSTAT_READ
);
4976 mutex_exit(&arc_reclaim_lock
);
4979 * We call arc_adjust() before (possibly) calling
4980 * arc_kmem_reap_now(), so that we can wake up
4981 * arc_get_data_buf() sooner.
4983 evicted
= arc_adjust();
4985 int64_t free_memory
= arc_available_memory();
4986 if (free_memory
< 0) {
4988 arc_no_grow
= B_TRUE
;
4992 * Wait at least zfs_grow_retry (default 5) seconds
4993 * before considering growing.
4995 growtime
= gethrtime() + SEC2NSEC(arc_grow_retry
);
4997 arc_kmem_reap_now();
5000 * If we are still low on memory, shrink the ARC
5001 * so that we have arc_shrink_min free space.
5003 free_memory
= arc_available_memory();
5006 (arc_c
>> arc_shrink_shift
) - free_memory
;
5009 to_free
= MAX(to_free
, need_free
);
5011 arc_shrink(to_free
);
5013 } else if (free_memory
< arc_c
>> arc_no_grow_shift
) {
5014 arc_no_grow
= B_TRUE
;
5015 } else if (gethrtime() >= growtime
) {
5016 arc_no_grow
= B_FALSE
;
5019 mutex_enter(&arc_reclaim_lock
);
5022 * If evicted is zero, we couldn't evict anything via
5023 * arc_adjust(). This could be due to hash lock
5024 * collisions, but more likely due to the majority of
5025 * arc buffers being unevictable. Therefore, even if
5026 * arc_size is above arc_c, another pass is unlikely to
5027 * be helpful and could potentially cause us to enter an
5030 if (arc_size
<= arc_c
|| evicted
== 0) {
5032 * We're either no longer overflowing, or we
5033 * can't evict anything more, so we should wake
5034 * up any threads before we go to sleep and remove
5035 * the bytes we were working on from arc_need_free
5036 * since nothing more will be done here.
5038 cv_broadcast(&arc_reclaim_waiters_cv
);
5039 ARCSTAT_INCR(arcstat_need_free
, -need_free
);
5042 * Block until signaled, or after one second (we
5043 * might need to perform arc_kmem_reap_now()
5044 * even if we aren't being signalled)
5046 CALLB_CPR_SAFE_BEGIN(&cpr
);
5047 (void) cv_timedwait_sig_hires(&arc_reclaim_thread_cv
,
5048 &arc_reclaim_lock
, SEC2NSEC(1), MSEC2NSEC(1), 0);
5049 CALLB_CPR_SAFE_END(&cpr
, &arc_reclaim_lock
);
5053 arc_reclaim_thread_exit
= B_FALSE
;
5054 cv_broadcast(&arc_reclaim_thread_cv
);
5055 CALLB_CPR_EXIT(&cpr
); /* drops arc_reclaim_lock */
5056 spl_fstrans_unmark(cookie
);
5062 * Determine the amount of memory eligible for eviction contained in the
5063 * ARC. All clean data reported by the ghost lists can always be safely
5064 * evicted. Due to arc_c_min, the same does not hold for all clean data
5065 * contained by the regular mru and mfu lists.
5067 * In the case of the regular mru and mfu lists, we need to report as
5068 * much clean data as possible, such that evicting that same reported
5069 * data will not bring arc_size below arc_c_min. Thus, in certain
5070 * circumstances, the total amount of clean data in the mru and mfu
5071 * lists might not actually be evictable.
5073 * The following two distinct cases are accounted for:
5075 * 1. The sum of the amount of dirty data contained by both the mru and
5076 * mfu lists, plus the ARC's other accounting (e.g. the anon list),
5077 * is greater than or equal to arc_c_min.
5078 * (i.e. amount of dirty data >= arc_c_min)
5080 * This is the easy case; all clean data contained by the mru and mfu
5081 * lists is evictable. Evicting all clean data can only drop arc_size
5082 * to the amount of dirty data, which is greater than arc_c_min.
5084 * 2. The sum of the amount of dirty data contained by both the mru and
5085 * mfu lists, plus the ARC's other accounting (e.g. the anon list),
5086 * is less than arc_c_min.
5087 * (i.e. arc_c_min > amount of dirty data)
5089 * 2.1. arc_size is greater than or equal arc_c_min.
5090 * (i.e. arc_size >= arc_c_min > amount of dirty data)
5092 * In this case, not all clean data from the regular mru and mfu
5093 * lists is actually evictable; we must leave enough clean data
5094 * to keep arc_size above arc_c_min. Thus, the maximum amount of
5095 * evictable data from the two lists combined, is exactly the
5096 * difference between arc_size and arc_c_min.
5098 * 2.2. arc_size is less than arc_c_min
5099 * (i.e. arc_c_min > arc_size > amount of dirty data)
5101 * In this case, none of the data contained in the mru and mfu
5102 * lists is evictable, even if it's clean. Since arc_size is
5103 * already below arc_c_min, evicting any more would only
5104 * increase this negative difference.
5107 arc_evictable_memory(void)
5109 uint64_t arc_clean
=
5110 refcount_count(&arc_mru
->arcs_esize
[ARC_BUFC_DATA
]) +
5111 refcount_count(&arc_mru
->arcs_esize
[ARC_BUFC_METADATA
]) +
5112 refcount_count(&arc_mfu
->arcs_esize
[ARC_BUFC_DATA
]) +
5113 refcount_count(&arc_mfu
->arcs_esize
[ARC_BUFC_METADATA
]);
5114 uint64_t arc_dirty
= MAX((int64_t)arc_size
- (int64_t)arc_clean
, 0);
5117 * Scale reported evictable memory in proportion to page cache, cap
5118 * at specified min/max.
5120 #ifdef ZFS_GLOBAL_NODE_PAGE_STATE
5121 uint64_t min
= (ptob(global_node_page_state(NR_FILE_PAGES
)) / 100) *
5124 uint64_t min
= (ptob(global_page_state(NR_FILE_PAGES
)) / 100) *
5127 min
= MAX(arc_c_min
, MIN(arc_c_max
, min
));
5129 if (arc_dirty
>= min
)
5132 return (MAX((int64_t)arc_size
- (int64_t)min
, 0));
5136 * If sc->nr_to_scan is zero, the caller is requesting a query of the
5137 * number of objects which can potentially be freed. If it is nonzero,
5138 * the request is to free that many objects.
5140 * Linux kernels >= 3.12 have the count_objects and scan_objects callbacks
5141 * in struct shrinker and also require the shrinker to return the number
5144 * Older kernels require the shrinker to return the number of freeable
5145 * objects following the freeing of nr_to_free.
5147 static spl_shrinker_t
5148 __arc_shrinker_func(struct shrinker
*shrink
, struct shrink_control
*sc
)
5152 /* The arc is considered warm once reclaim has occurred */
5153 if (unlikely(arc_warm
== B_FALSE
))
5156 /* Return the potential number of reclaimable pages */
5157 pages
= btop((int64_t)arc_evictable_memory());
5158 if (sc
->nr_to_scan
== 0)
5161 /* Not allowed to perform filesystem reclaim */
5162 if (!(sc
->gfp_mask
& __GFP_FS
))
5163 return (SHRINK_STOP
);
5165 /* Reclaim in progress */
5166 if (mutex_tryenter(&arc_reclaim_lock
) == 0) {
5167 ARCSTAT_INCR(arcstat_need_free
, ptob(sc
->nr_to_scan
));
5171 mutex_exit(&arc_reclaim_lock
);
5174 * Evict the requested number of pages by shrinking arc_c the
5178 arc_shrink(ptob(sc
->nr_to_scan
));
5179 if (current_is_kswapd())
5180 arc_kmem_reap_now();
5181 #ifdef HAVE_SPLIT_SHRINKER_CALLBACK
5182 pages
= MAX((int64_t)pages
-
5183 (int64_t)btop(arc_evictable_memory()), 0);
5185 pages
= btop(arc_evictable_memory());
5188 * We've shrunk what we can, wake up threads.
5190 cv_broadcast(&arc_reclaim_waiters_cv
);
5192 pages
= SHRINK_STOP
;
5195 * When direct reclaim is observed it usually indicates a rapid
5196 * increase in memory pressure. This occurs because the kswapd
5197 * threads were unable to asynchronously keep enough free memory
5198 * available. In this case set arc_no_grow to briefly pause arc
5199 * growth to avoid compounding the memory pressure.
5201 if (current_is_kswapd()) {
5202 ARCSTAT_BUMP(arcstat_memory_indirect_count
);
5204 arc_no_grow
= B_TRUE
;
5205 arc_kmem_reap_now();
5206 ARCSTAT_BUMP(arcstat_memory_direct_count
);
5211 SPL_SHRINKER_CALLBACK_WRAPPER(arc_shrinker_func
);
5213 SPL_SHRINKER_DECLARE(arc_shrinker
, arc_shrinker_func
, DEFAULT_SEEKS
);
5214 #endif /* _KERNEL */
5217 * Adapt arc info given the number of bytes we are trying to add and
5218 * the state that we are coming from. This function is only called
5219 * when we are adding new content to the cache.
5222 arc_adapt(int bytes
, arc_state_t
*state
)
5225 uint64_t arc_p_min
= (arc_c
>> arc_p_min_shift
);
5226 int64_t mrug_size
= refcount_count(&arc_mru_ghost
->arcs_size
);
5227 int64_t mfug_size
= refcount_count(&arc_mfu_ghost
->arcs_size
);
5229 if (state
== arc_l2c_only
)
5234 * Adapt the target size of the MRU list:
5235 * - if we just hit in the MRU ghost list, then increase
5236 * the target size of the MRU list.
5237 * - if we just hit in the MFU ghost list, then increase
5238 * the target size of the MFU list by decreasing the
5239 * target size of the MRU list.
5241 if (state
== arc_mru_ghost
) {
5242 mult
= (mrug_size
>= mfug_size
) ? 1 : (mfug_size
/ mrug_size
);
5243 if (!zfs_arc_p_dampener_disable
)
5244 mult
= MIN(mult
, 10); /* avoid wild arc_p adjustment */
5246 arc_p
= MIN(arc_c
- arc_p_min
, arc_p
+ bytes
* mult
);
5247 } else if (state
== arc_mfu_ghost
) {
5250 mult
= (mfug_size
>= mrug_size
) ? 1 : (mrug_size
/ mfug_size
);
5251 if (!zfs_arc_p_dampener_disable
)
5252 mult
= MIN(mult
, 10);
5254 delta
= MIN(bytes
* mult
, arc_p
);
5255 arc_p
= MAX(arc_p_min
, arc_p
- delta
);
5257 ASSERT((int64_t)arc_p
>= 0);
5259 if (arc_reclaim_needed()) {
5260 cv_signal(&arc_reclaim_thread_cv
);
5267 if (arc_c
>= arc_c_max
)
5271 * If we're within (2 * maxblocksize) bytes of the target
5272 * cache size, increment the target cache size
5274 ASSERT3U(arc_c
, >=, 2ULL << SPA_MAXBLOCKSHIFT
);
5275 if (arc_size
>= arc_c
- (2ULL << SPA_MAXBLOCKSHIFT
)) {
5276 atomic_add_64(&arc_c
, (int64_t)bytes
);
5277 if (arc_c
> arc_c_max
)
5279 else if (state
== arc_anon
)
5280 atomic_add_64(&arc_p
, (int64_t)bytes
);
5284 ASSERT((int64_t)arc_p
>= 0);
5288 * Check if arc_size has grown past our upper threshold, determined by
5289 * zfs_arc_overflow_shift.
5292 arc_is_overflowing(void)
5294 /* Always allow at least one block of overflow */
5295 uint64_t overflow
= MAX(SPA_MAXBLOCKSIZE
,
5296 arc_c
>> zfs_arc_overflow_shift
);
5298 return (arc_size
>= arc_c
+ overflow
);
5302 arc_get_data_abd(arc_buf_hdr_t
*hdr
, uint64_t size
, void *tag
)
5304 arc_buf_contents_t type
= arc_buf_type(hdr
);
5306 arc_get_data_impl(hdr
, size
, tag
);
5307 if (type
== ARC_BUFC_METADATA
) {
5308 return (abd_alloc(size
, B_TRUE
));
5310 ASSERT(type
== ARC_BUFC_DATA
);
5311 return (abd_alloc(size
, B_FALSE
));
5316 arc_get_data_buf(arc_buf_hdr_t
*hdr
, uint64_t size
, void *tag
)
5318 arc_buf_contents_t type
= arc_buf_type(hdr
);
5320 arc_get_data_impl(hdr
, size
, tag
);
5321 if (type
== ARC_BUFC_METADATA
) {
5322 return (zio_buf_alloc(size
));
5324 ASSERT(type
== ARC_BUFC_DATA
);
5325 return (zio_data_buf_alloc(size
));
5330 * Allocate a block and return it to the caller. If we are hitting the
5331 * hard limit for the cache size, we must sleep, waiting for the eviction
5332 * thread to catch up. If we're past the target size but below the hard
5333 * limit, we'll only signal the reclaim thread and continue on.
5336 arc_get_data_impl(arc_buf_hdr_t
*hdr
, uint64_t size
, void *tag
)
5338 arc_state_t
*state
= hdr
->b_l1hdr
.b_state
;
5339 arc_buf_contents_t type
= arc_buf_type(hdr
);
5341 arc_adapt(size
, state
);
5344 * If arc_size is currently overflowing, and has grown past our
5345 * upper limit, we must be adding data faster than the evict
5346 * thread can evict. Thus, to ensure we don't compound the
5347 * problem by adding more data and forcing arc_size to grow even
5348 * further past it's target size, we halt and wait for the
5349 * eviction thread to catch up.
5351 * It's also possible that the reclaim thread is unable to evict
5352 * enough buffers to get arc_size below the overflow limit (e.g.
5353 * due to buffers being un-evictable, or hash lock collisions).
5354 * In this case, we want to proceed regardless if we're
5355 * overflowing; thus we don't use a while loop here.
5357 if (arc_is_overflowing()) {
5358 mutex_enter(&arc_reclaim_lock
);
5361 * Now that we've acquired the lock, we may no longer be
5362 * over the overflow limit, lets check.
5364 * We're ignoring the case of spurious wake ups. If that
5365 * were to happen, it'd let this thread consume an ARC
5366 * buffer before it should have (i.e. before we're under
5367 * the overflow limit and were signalled by the reclaim
5368 * thread). As long as that is a rare occurrence, it
5369 * shouldn't cause any harm.
5371 if (arc_is_overflowing()) {
5372 cv_signal(&arc_reclaim_thread_cv
);
5373 cv_wait(&arc_reclaim_waiters_cv
, &arc_reclaim_lock
);
5376 mutex_exit(&arc_reclaim_lock
);
5379 VERIFY3U(hdr
->b_type
, ==, type
);
5380 if (type
== ARC_BUFC_METADATA
) {
5381 arc_space_consume(size
, ARC_SPACE_META
);
5383 arc_space_consume(size
, ARC_SPACE_DATA
);
5387 * Update the state size. Note that ghost states have a
5388 * "ghost size" and so don't need to be updated.
5390 if (!GHOST_STATE(state
)) {
5392 (void) refcount_add_many(&state
->arcs_size
, size
, tag
);
5395 * If this is reached via arc_read, the link is
5396 * protected by the hash lock. If reached via
5397 * arc_buf_alloc, the header should not be accessed by
5398 * any other thread. And, if reached via arc_read_done,
5399 * the hash lock will protect it if it's found in the
5400 * hash table; otherwise no other thread should be
5401 * trying to [add|remove]_reference it.
5403 if (multilist_link_active(&hdr
->b_l1hdr
.b_arc_node
)) {
5404 ASSERT(refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
));
5405 (void) refcount_add_many(&state
->arcs_esize
[type
],
5410 * If we are growing the cache, and we are adding anonymous
5411 * data, and we have outgrown arc_p, update arc_p
5413 if (arc_size
< arc_c
&& hdr
->b_l1hdr
.b_state
== arc_anon
&&
5414 (refcount_count(&arc_anon
->arcs_size
) +
5415 refcount_count(&arc_mru
->arcs_size
) > arc_p
))
5416 arc_p
= MIN(arc_c
, arc_p
+ size
);
5421 arc_free_data_abd(arc_buf_hdr_t
*hdr
, abd_t
*abd
, uint64_t size
, void *tag
)
5423 arc_free_data_impl(hdr
, size
, tag
);
5428 arc_free_data_buf(arc_buf_hdr_t
*hdr
, void *buf
, uint64_t size
, void *tag
)
5430 arc_buf_contents_t type
= arc_buf_type(hdr
);
5432 arc_free_data_impl(hdr
, size
, tag
);
5433 if (type
== ARC_BUFC_METADATA
) {
5434 zio_buf_free(buf
, size
);
5436 ASSERT(type
== ARC_BUFC_DATA
);
5437 zio_data_buf_free(buf
, size
);
5442 * Free the arc data buffer.
5445 arc_free_data_impl(arc_buf_hdr_t
*hdr
, uint64_t size
, void *tag
)
5447 arc_state_t
*state
= hdr
->b_l1hdr
.b_state
;
5448 arc_buf_contents_t type
= arc_buf_type(hdr
);
5450 /* protected by hash lock, if in the hash table */
5451 if (multilist_link_active(&hdr
->b_l1hdr
.b_arc_node
)) {
5452 ASSERT(refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
));
5453 ASSERT(state
!= arc_anon
&& state
!= arc_l2c_only
);
5455 (void) refcount_remove_many(&state
->arcs_esize
[type
],
5458 (void) refcount_remove_many(&state
->arcs_size
, size
, tag
);
5460 VERIFY3U(hdr
->b_type
, ==, type
);
5461 if (type
== ARC_BUFC_METADATA
) {
5462 arc_space_return(size
, ARC_SPACE_META
);
5464 ASSERT(type
== ARC_BUFC_DATA
);
5465 arc_space_return(size
, ARC_SPACE_DATA
);
5470 * This routine is called whenever a buffer is accessed.
5471 * NOTE: the hash lock is dropped in this function.
5474 arc_access(arc_buf_hdr_t
*hdr
, kmutex_t
*hash_lock
)
5478 ASSERT(MUTEX_HELD(hash_lock
));
5479 ASSERT(HDR_HAS_L1HDR(hdr
));
5481 if (hdr
->b_l1hdr
.b_state
== arc_anon
) {
5483 * This buffer is not in the cache, and does not
5484 * appear in our "ghost" list. Add the new buffer
5488 ASSERT0(hdr
->b_l1hdr
.b_arc_access
);
5489 hdr
->b_l1hdr
.b_arc_access
= ddi_get_lbolt();
5490 DTRACE_PROBE1(new_state__mru
, arc_buf_hdr_t
*, hdr
);
5491 arc_change_state(arc_mru
, hdr
, hash_lock
);
5493 } else if (hdr
->b_l1hdr
.b_state
== arc_mru
) {
5494 now
= ddi_get_lbolt();
5497 * If this buffer is here because of a prefetch, then either:
5498 * - clear the flag if this is a "referencing" read
5499 * (any subsequent access will bump this into the MFU state).
5501 * - move the buffer to the head of the list if this is
5502 * another prefetch (to make it less likely to be evicted).
5504 if (HDR_PREFETCH(hdr
) || HDR_PRESCIENT_PREFETCH(hdr
)) {
5505 if (refcount_count(&hdr
->b_l1hdr
.b_refcnt
) == 0) {
5506 /* link protected by hash lock */
5507 ASSERT(multilist_link_active(
5508 &hdr
->b_l1hdr
.b_arc_node
));
5510 arc_hdr_clear_flags(hdr
,
5512 ARC_FLAG_PRESCIENT_PREFETCH
);
5513 atomic_inc_32(&hdr
->b_l1hdr
.b_mru_hits
);
5514 ARCSTAT_BUMP(arcstat_mru_hits
);
5516 hdr
->b_l1hdr
.b_arc_access
= now
;
5521 * This buffer has been "accessed" only once so far,
5522 * but it is still in the cache. Move it to the MFU
5525 if (ddi_time_after(now
, hdr
->b_l1hdr
.b_arc_access
+
5528 * More than 125ms have passed since we
5529 * instantiated this buffer. Move it to the
5530 * most frequently used state.
5532 hdr
->b_l1hdr
.b_arc_access
= now
;
5533 DTRACE_PROBE1(new_state__mfu
, arc_buf_hdr_t
*, hdr
);
5534 arc_change_state(arc_mfu
, hdr
, hash_lock
);
5536 atomic_inc_32(&hdr
->b_l1hdr
.b_mru_hits
);
5537 ARCSTAT_BUMP(arcstat_mru_hits
);
5538 } else if (hdr
->b_l1hdr
.b_state
== arc_mru_ghost
) {
5539 arc_state_t
*new_state
;
5541 * This buffer has been "accessed" recently, but
5542 * was evicted from the cache. Move it to the
5546 if (HDR_PREFETCH(hdr
) || HDR_PRESCIENT_PREFETCH(hdr
)) {
5547 new_state
= arc_mru
;
5548 if (refcount_count(&hdr
->b_l1hdr
.b_refcnt
) > 0) {
5549 arc_hdr_clear_flags(hdr
,
5551 ARC_FLAG_PRESCIENT_PREFETCH
);
5553 DTRACE_PROBE1(new_state__mru
, arc_buf_hdr_t
*, hdr
);
5555 new_state
= arc_mfu
;
5556 DTRACE_PROBE1(new_state__mfu
, arc_buf_hdr_t
*, hdr
);
5559 hdr
->b_l1hdr
.b_arc_access
= ddi_get_lbolt();
5560 arc_change_state(new_state
, hdr
, hash_lock
);
5562 atomic_inc_32(&hdr
->b_l1hdr
.b_mru_ghost_hits
);
5563 ARCSTAT_BUMP(arcstat_mru_ghost_hits
);
5564 } else if (hdr
->b_l1hdr
.b_state
== arc_mfu
) {
5566 * This buffer has been accessed more than once and is
5567 * still in the cache. Keep it in the MFU state.
5569 * NOTE: an add_reference() that occurred when we did
5570 * the arc_read() will have kicked this off the list.
5571 * If it was a prefetch, we will explicitly move it to
5572 * the head of the list now.
5575 atomic_inc_32(&hdr
->b_l1hdr
.b_mfu_hits
);
5576 ARCSTAT_BUMP(arcstat_mfu_hits
);
5577 hdr
->b_l1hdr
.b_arc_access
= ddi_get_lbolt();
5578 } else if (hdr
->b_l1hdr
.b_state
== arc_mfu_ghost
) {
5579 arc_state_t
*new_state
= arc_mfu
;
5581 * This buffer has been accessed more than once but has
5582 * been evicted from the cache. Move it back to the
5586 if (HDR_PREFETCH(hdr
) || HDR_PRESCIENT_PREFETCH(hdr
)) {
5588 * This is a prefetch access...
5589 * move this block back to the MRU state.
5591 new_state
= arc_mru
;
5594 hdr
->b_l1hdr
.b_arc_access
= ddi_get_lbolt();
5595 DTRACE_PROBE1(new_state__mfu
, arc_buf_hdr_t
*, hdr
);
5596 arc_change_state(new_state
, hdr
, hash_lock
);
5598 atomic_inc_32(&hdr
->b_l1hdr
.b_mfu_ghost_hits
);
5599 ARCSTAT_BUMP(arcstat_mfu_ghost_hits
);
5600 } else if (hdr
->b_l1hdr
.b_state
== arc_l2c_only
) {
5602 * This buffer is on the 2nd Level ARC.
5605 hdr
->b_l1hdr
.b_arc_access
= ddi_get_lbolt();
5606 DTRACE_PROBE1(new_state__mfu
, arc_buf_hdr_t
*, hdr
);
5607 arc_change_state(arc_mfu
, hdr
, hash_lock
);
5609 cmn_err(CE_PANIC
, "invalid arc state 0x%p",
5610 hdr
->b_l1hdr
.b_state
);
5614 /* a generic arc_read_done_func_t which you can use */
5617 arc_bcopy_func(zio_t
*zio
, const zbookmark_phys_t
*zb
, const blkptr_t
*bp
,
5618 arc_buf_t
*buf
, void *arg
)
5623 bcopy(buf
->b_data
, arg
, arc_buf_size(buf
));
5624 arc_buf_destroy(buf
, arg
);
5627 /* a generic arc_read_done_func_t */
5630 arc_getbuf_func(zio_t
*zio
, const zbookmark_phys_t
*zb
, const blkptr_t
*bp
,
5631 arc_buf_t
*buf
, void *arg
)
5633 arc_buf_t
**bufp
= arg
;
5639 ASSERT(buf
->b_data
);
5644 arc_hdr_verify(arc_buf_hdr_t
*hdr
, blkptr_t
*bp
)
5646 if (BP_IS_HOLE(bp
) || BP_IS_EMBEDDED(bp
)) {
5647 ASSERT3U(HDR_GET_PSIZE(hdr
), ==, 0);
5648 ASSERT3U(arc_hdr_get_compress(hdr
), ==, ZIO_COMPRESS_OFF
);
5650 if (HDR_COMPRESSION_ENABLED(hdr
)) {
5651 ASSERT3U(arc_hdr_get_compress(hdr
), ==,
5652 BP_GET_COMPRESS(bp
));
5654 ASSERT3U(HDR_GET_LSIZE(hdr
), ==, BP_GET_LSIZE(bp
));
5655 ASSERT3U(HDR_GET_PSIZE(hdr
), ==, BP_GET_PSIZE(bp
));
5656 ASSERT3U(!!HDR_PROTECTED(hdr
), ==, BP_IS_PROTECTED(bp
));
5661 arc_read_done(zio_t
*zio
)
5663 blkptr_t
*bp
= zio
->io_bp
;
5664 arc_buf_hdr_t
*hdr
= zio
->io_private
;
5665 kmutex_t
*hash_lock
= NULL
;
5666 arc_callback_t
*callback_list
;
5667 arc_callback_t
*acb
;
5668 boolean_t freeable
= B_FALSE
;
5671 * The hdr was inserted into hash-table and removed from lists
5672 * prior to starting I/O. We should find this header, since
5673 * it's in the hash table, and it should be legit since it's
5674 * not possible to evict it during the I/O. The only possible
5675 * reason for it not to be found is if we were freed during the
5678 if (HDR_IN_HASH_TABLE(hdr
)) {
5679 arc_buf_hdr_t
*found
;
5681 ASSERT3U(hdr
->b_birth
, ==, BP_PHYSICAL_BIRTH(zio
->io_bp
));
5682 ASSERT3U(hdr
->b_dva
.dva_word
[0], ==,
5683 BP_IDENTITY(zio
->io_bp
)->dva_word
[0]);
5684 ASSERT3U(hdr
->b_dva
.dva_word
[1], ==,
5685 BP_IDENTITY(zio
->io_bp
)->dva_word
[1]);
5687 found
= buf_hash_find(hdr
->b_spa
, zio
->io_bp
, &hash_lock
);
5689 ASSERT((found
== hdr
&&
5690 DVA_EQUAL(&hdr
->b_dva
, BP_IDENTITY(zio
->io_bp
))) ||
5691 (found
== hdr
&& HDR_L2_READING(hdr
)));
5692 ASSERT3P(hash_lock
, !=, NULL
);
5695 if (BP_IS_PROTECTED(bp
)) {
5696 hdr
->b_crypt_hdr
.b_ot
= BP_GET_TYPE(bp
);
5697 hdr
->b_crypt_hdr
.b_dsobj
= zio
->io_bookmark
.zb_objset
;
5698 zio_crypt_decode_params_bp(bp
, hdr
->b_crypt_hdr
.b_salt
,
5699 hdr
->b_crypt_hdr
.b_iv
);
5701 if (BP_GET_TYPE(bp
) == DMU_OT_INTENT_LOG
) {
5704 tmpbuf
= abd_borrow_buf_copy(zio
->io_abd
,
5705 sizeof (zil_chain_t
));
5706 zio_crypt_decode_mac_zil(tmpbuf
,
5707 hdr
->b_crypt_hdr
.b_mac
);
5708 abd_return_buf(zio
->io_abd
, tmpbuf
,
5709 sizeof (zil_chain_t
));
5711 zio_crypt_decode_mac_bp(bp
, hdr
->b_crypt_hdr
.b_mac
);
5715 if (zio
->io_error
== 0) {
5716 /* byteswap if necessary */
5717 if (BP_SHOULD_BYTESWAP(zio
->io_bp
)) {
5718 if (BP_GET_LEVEL(zio
->io_bp
) > 0) {
5719 hdr
->b_l1hdr
.b_byteswap
= DMU_BSWAP_UINT64
;
5721 hdr
->b_l1hdr
.b_byteswap
=
5722 DMU_OT_BYTESWAP(BP_GET_TYPE(zio
->io_bp
));
5725 hdr
->b_l1hdr
.b_byteswap
= DMU_BSWAP_NUMFUNCS
;
5729 arc_hdr_clear_flags(hdr
, ARC_FLAG_L2_EVICTED
);
5730 if (l2arc_noprefetch
&& HDR_PREFETCH(hdr
))
5731 arc_hdr_clear_flags(hdr
, ARC_FLAG_L2CACHE
);
5733 callback_list
= hdr
->b_l1hdr
.b_acb
;
5734 ASSERT3P(callback_list
, !=, NULL
);
5736 if (hash_lock
&& zio
->io_error
== 0 &&
5737 hdr
->b_l1hdr
.b_state
== arc_anon
) {
5739 * Only call arc_access on anonymous buffers. This is because
5740 * if we've issued an I/O for an evicted buffer, we've already
5741 * called arc_access (to prevent any simultaneous readers from
5742 * getting confused).
5744 arc_access(hdr
, hash_lock
);
5748 * If a read request has a callback (i.e. acb_done is not NULL), then we
5749 * make a buf containing the data according to the parameters which were
5750 * passed in. The implementation of arc_buf_alloc_impl() ensures that we
5751 * aren't needlessly decompressing the data multiple times.
5753 int callback_cnt
= 0;
5754 for (acb
= callback_list
; acb
!= NULL
; acb
= acb
->acb_next
) {
5760 if (zio
->io_error
!= 0)
5763 int error
= arc_buf_alloc_impl(hdr
, zio
->io_spa
,
5764 acb
->acb_dsobj
, acb
->acb_private
, acb
->acb_encrypted
,
5765 acb
->acb_compressed
, acb
->acb_noauth
, B_TRUE
,
5768 arc_buf_destroy(acb
->acb_buf
, acb
->acb_private
);
5769 acb
->acb_buf
= NULL
;
5773 * Assert non-speculative zios didn't fail because an
5774 * encryption key wasn't loaded
5776 ASSERT((zio
->io_flags
& ZIO_FLAG_SPECULATIVE
) || error
== 0);
5779 * If we failed to decrypt, report an error now (as the zio
5780 * layer would have done if it had done the transforms).
5782 if (error
== ECKSUM
) {
5783 ASSERT(BP_IS_PROTECTED(bp
));
5784 error
= SET_ERROR(EIO
);
5785 spa_log_error(zio
->io_spa
, &zio
->io_bookmark
);
5786 if ((zio
->io_flags
& ZIO_FLAG_SPECULATIVE
) == 0) {
5787 zfs_ereport_post(FM_EREPORT_ZFS_AUTHENTICATION
,
5788 zio
->io_spa
, NULL
, &zio
->io_bookmark
, zio
,
5793 if (zio
->io_error
== 0)
5794 zio
->io_error
= error
;
5796 hdr
->b_l1hdr
.b_acb
= NULL
;
5797 arc_hdr_clear_flags(hdr
, ARC_FLAG_IO_IN_PROGRESS
);
5798 if (callback_cnt
== 0)
5799 ASSERT(hdr
->b_l1hdr
.b_pabd
!= NULL
|| HDR_HAS_RABD(hdr
));
5801 ASSERT(refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
) ||
5802 callback_list
!= NULL
);
5804 if (zio
->io_error
== 0) {
5805 arc_hdr_verify(hdr
, zio
->io_bp
);
5807 arc_hdr_set_flags(hdr
, ARC_FLAG_IO_ERROR
);
5808 if (hdr
->b_l1hdr
.b_state
!= arc_anon
)
5809 arc_change_state(arc_anon
, hdr
, hash_lock
);
5810 if (HDR_IN_HASH_TABLE(hdr
))
5811 buf_hash_remove(hdr
);
5812 freeable
= refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
);
5816 * Broadcast before we drop the hash_lock to avoid the possibility
5817 * that the hdr (and hence the cv) might be freed before we get to
5818 * the cv_broadcast().
5820 cv_broadcast(&hdr
->b_l1hdr
.b_cv
);
5822 if (hash_lock
!= NULL
) {
5823 mutex_exit(hash_lock
);
5826 * This block was freed while we waited for the read to
5827 * complete. It has been removed from the hash table and
5828 * moved to the anonymous state (so that it won't show up
5831 ASSERT3P(hdr
->b_l1hdr
.b_state
, ==, arc_anon
);
5832 freeable
= refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
);
5835 /* execute each callback and free its structure */
5836 while ((acb
= callback_list
) != NULL
) {
5837 if (acb
->acb_done
) {
5838 acb
->acb_done(zio
, &zio
->io_bookmark
, zio
->io_bp
,
5839 acb
->acb_buf
, acb
->acb_private
);
5842 if (acb
->acb_zio_dummy
!= NULL
) {
5843 acb
->acb_zio_dummy
->io_error
= zio
->io_error
;
5844 zio_nowait(acb
->acb_zio_dummy
);
5847 callback_list
= acb
->acb_next
;
5848 kmem_free(acb
, sizeof (arc_callback_t
));
5852 arc_hdr_destroy(hdr
);
5856 * "Read" the block at the specified DVA (in bp) via the
5857 * cache. If the block is found in the cache, invoke the provided
5858 * callback immediately and return. Note that the `zio' parameter
5859 * in the callback will be NULL in this case, since no IO was
5860 * required. If the block is not in the cache pass the read request
5861 * on to the spa with a substitute callback function, so that the
5862 * requested block will be added to the cache.
5864 * If a read request arrives for a block that has a read in-progress,
5865 * either wait for the in-progress read to complete (and return the
5866 * results); or, if this is a read with a "done" func, add a record
5867 * to the read to invoke the "done" func when the read completes,
5868 * and return; or just return.
5870 * arc_read_done() will invoke all the requested "done" functions
5871 * for readers of this block.
5874 arc_read(zio_t
*pio
, spa_t
*spa
, const blkptr_t
*bp
,
5875 arc_read_done_func_t
*done
, void *private, zio_priority_t priority
,
5876 int zio_flags
, arc_flags_t
*arc_flags
, const zbookmark_phys_t
*zb
)
5878 arc_buf_hdr_t
*hdr
= NULL
;
5879 kmutex_t
*hash_lock
= NULL
;
5881 uint64_t guid
= spa_load_guid(spa
);
5882 boolean_t compressed_read
= (zio_flags
& ZIO_FLAG_RAW_COMPRESS
) != 0;
5883 boolean_t encrypted_read
= BP_IS_ENCRYPTED(bp
) &&
5884 (zio_flags
& ZIO_FLAG_RAW_ENCRYPT
) != 0;
5885 boolean_t noauth_read
= BP_IS_AUTHENTICATED(bp
) &&
5886 (zio_flags
& ZIO_FLAG_RAW_ENCRYPT
) != 0;
5889 ASSERT(!BP_IS_EMBEDDED(bp
) ||
5890 BPE_GET_ETYPE(bp
) == BP_EMBEDDED_TYPE_DATA
);
5893 if (!BP_IS_EMBEDDED(bp
)) {
5895 * Embedded BP's have no DVA and require no I/O to "read".
5896 * Create an anonymous arc buf to back it.
5898 hdr
= buf_hash_find(guid
, bp
, &hash_lock
);
5902 * Determine if we have an L1 cache hit or a cache miss. For simplicity
5903 * we maintain encrypted data seperately from compressed / uncompressed
5904 * data. If the user is requesting raw encrypted data and we don't have
5905 * that in the header we will read from disk to guarantee that we can
5906 * get it even if the encryption keys aren't loaded.
5908 if (hdr
!= NULL
&& HDR_HAS_L1HDR(hdr
) && (HDR_HAS_RABD(hdr
) ||
5909 (hdr
->b_l1hdr
.b_pabd
!= NULL
&& !encrypted_read
))) {
5910 arc_buf_t
*buf
= NULL
;
5911 *arc_flags
|= ARC_FLAG_CACHED
;
5913 if (HDR_IO_IN_PROGRESS(hdr
)) {
5914 zio_t
*head_zio
= hdr
->b_l1hdr
.b_acb
->acb_zio_head
;
5916 ASSERT3P(head_zio
, !=, NULL
);
5917 if ((hdr
->b_flags
& ARC_FLAG_PRIO_ASYNC_READ
) &&
5918 priority
== ZIO_PRIORITY_SYNC_READ
) {
5920 * This is a sync read that needs to wait for
5921 * an in-flight async read. Request that the
5922 * zio have its priority upgraded.
5924 zio_change_priority(head_zio
, priority
);
5925 DTRACE_PROBE1(arc__async__upgrade__sync
,
5926 arc_buf_hdr_t
*, hdr
);
5927 ARCSTAT_BUMP(arcstat_async_upgrade_sync
);
5929 if (hdr
->b_flags
& ARC_FLAG_PREDICTIVE_PREFETCH
) {
5930 arc_hdr_clear_flags(hdr
,
5931 ARC_FLAG_PREDICTIVE_PREFETCH
);
5934 if (*arc_flags
& ARC_FLAG_WAIT
) {
5935 cv_wait(&hdr
->b_l1hdr
.b_cv
, hash_lock
);
5936 mutex_exit(hash_lock
);
5939 ASSERT(*arc_flags
& ARC_FLAG_NOWAIT
);
5942 arc_callback_t
*acb
= NULL
;
5944 acb
= kmem_zalloc(sizeof (arc_callback_t
),
5946 acb
->acb_done
= done
;
5947 acb
->acb_private
= private;
5948 acb
->acb_compressed
= compressed_read
;
5949 acb
->acb_encrypted
= encrypted_read
;
5950 acb
->acb_noauth
= noauth_read
;
5951 acb
->acb_dsobj
= zb
->zb_objset
;
5953 acb
->acb_zio_dummy
= zio_null(pio
,
5954 spa
, NULL
, NULL
, NULL
, zio_flags
);
5956 ASSERT3P(acb
->acb_done
, !=, NULL
);
5957 acb
->acb_zio_head
= head_zio
;
5958 acb
->acb_next
= hdr
->b_l1hdr
.b_acb
;
5959 hdr
->b_l1hdr
.b_acb
= acb
;
5960 mutex_exit(hash_lock
);
5963 mutex_exit(hash_lock
);
5967 ASSERT(hdr
->b_l1hdr
.b_state
== arc_mru
||
5968 hdr
->b_l1hdr
.b_state
== arc_mfu
);
5971 if (hdr
->b_flags
& ARC_FLAG_PREDICTIVE_PREFETCH
) {
5973 * This is a demand read which does not have to
5974 * wait for i/o because we did a predictive
5975 * prefetch i/o for it, which has completed.
5978 arc__demand__hit__predictive__prefetch
,
5979 arc_buf_hdr_t
*, hdr
);
5981 arcstat_demand_hit_predictive_prefetch
);
5982 arc_hdr_clear_flags(hdr
,
5983 ARC_FLAG_PREDICTIVE_PREFETCH
);
5986 if (hdr
->b_flags
& ARC_FLAG_PRESCIENT_PREFETCH
) {
5988 arcstat_demand_hit_prescient_prefetch
);
5989 arc_hdr_clear_flags(hdr
,
5990 ARC_FLAG_PRESCIENT_PREFETCH
);
5993 ASSERT(!BP_IS_EMBEDDED(bp
) || !BP_IS_HOLE(bp
));
5995 /* Get a buf with the desired data in it. */
5996 rc
= arc_buf_alloc_impl(hdr
, spa
, zb
->zb_objset
,
5997 private, encrypted_read
, compressed_read
,
5998 noauth_read
, B_TRUE
, &buf
);
6000 arc_buf_destroy(buf
, private);
6004 ASSERT((zio_flags
& ZIO_FLAG_SPECULATIVE
) || rc
== 0);
6005 } else if (*arc_flags
& ARC_FLAG_PREFETCH
&&
6006 refcount_count(&hdr
->b_l1hdr
.b_refcnt
) == 0) {
6007 arc_hdr_set_flags(hdr
, ARC_FLAG_PREFETCH
);
6009 DTRACE_PROBE1(arc__hit
, arc_buf_hdr_t
*, hdr
);
6010 arc_access(hdr
, hash_lock
);
6011 if (*arc_flags
& ARC_FLAG_PRESCIENT_PREFETCH
)
6012 arc_hdr_set_flags(hdr
, ARC_FLAG_PRESCIENT_PREFETCH
);
6013 if (*arc_flags
& ARC_FLAG_L2CACHE
)
6014 arc_hdr_set_flags(hdr
, ARC_FLAG_L2CACHE
);
6015 mutex_exit(hash_lock
);
6016 ARCSTAT_BUMP(arcstat_hits
);
6017 ARCSTAT_CONDSTAT(!HDR_PREFETCH(hdr
),
6018 demand
, prefetch
, !HDR_ISTYPE_METADATA(hdr
),
6019 data
, metadata
, hits
);
6022 done(NULL
, zb
, bp
, buf
, private);
6024 uint64_t lsize
= BP_GET_LSIZE(bp
);
6025 uint64_t psize
= BP_GET_PSIZE(bp
);
6026 arc_callback_t
*acb
;
6029 boolean_t devw
= B_FALSE
;
6034 * Gracefully handle a damaged logical block size as a
6037 if (lsize
> spa_maxblocksize(spa
)) {
6038 rc
= SET_ERROR(ECKSUM
);
6043 /* this block is not in the cache */
6044 arc_buf_hdr_t
*exists
= NULL
;
6045 arc_buf_contents_t type
= BP_GET_BUFC_TYPE(bp
);
6046 hdr
= arc_hdr_alloc(spa_load_guid(spa
), psize
, lsize
,
6047 BP_IS_PROTECTED(bp
), BP_GET_COMPRESS(bp
), type
,
6050 if (!BP_IS_EMBEDDED(bp
)) {
6051 hdr
->b_dva
= *BP_IDENTITY(bp
);
6052 hdr
->b_birth
= BP_PHYSICAL_BIRTH(bp
);
6053 exists
= buf_hash_insert(hdr
, &hash_lock
);
6055 if (exists
!= NULL
) {
6056 /* somebody beat us to the hash insert */
6057 mutex_exit(hash_lock
);
6058 buf_discard_identity(hdr
);
6059 arc_hdr_destroy(hdr
);
6060 goto top
; /* restart the IO request */
6064 * This block is in the ghost cache or encrypted data
6065 * was requested and we didn't have it. If it was
6066 * L2-only (and thus didn't have an L1 hdr),
6067 * we realloc the header to add an L1 hdr.
6069 if (!HDR_HAS_L1HDR(hdr
)) {
6070 hdr
= arc_hdr_realloc(hdr
, hdr_l2only_cache
,
6074 if (GHOST_STATE(hdr
->b_l1hdr
.b_state
)) {
6075 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
6076 ASSERT(!HDR_HAS_RABD(hdr
));
6077 ASSERT(!HDR_IO_IN_PROGRESS(hdr
));
6078 ASSERT0(refcount_count(&hdr
->b_l1hdr
.b_refcnt
));
6079 ASSERT3P(hdr
->b_l1hdr
.b_buf
, ==, NULL
);
6080 ASSERT3P(hdr
->b_l1hdr
.b_freeze_cksum
, ==, NULL
);
6081 } else if (HDR_IO_IN_PROGRESS(hdr
)) {
6083 * If this header already had an IO in progress
6084 * and we are performing another IO to fetch
6085 * encrypted data we must wait until the first
6086 * IO completes so as not to confuse
6087 * arc_read_done(). This should be very rare
6088 * and so the performance impact shouldn't
6091 cv_wait(&hdr
->b_l1hdr
.b_cv
, hash_lock
);
6092 mutex_exit(hash_lock
);
6097 * This is a delicate dance that we play here.
6098 * This hdr might be in the ghost list so we access
6099 * it to move it out of the ghost list before we
6100 * initiate the read. If it's a prefetch then
6101 * it won't have a callback so we'll remove the
6102 * reference that arc_buf_alloc_impl() created. We
6103 * do this after we've called arc_access() to
6104 * avoid hitting an assert in remove_reference().
6106 arc_access(hdr
, hash_lock
);
6107 arc_hdr_alloc_abd(hdr
, encrypted_read
);
6110 if (encrypted_read
) {
6111 ASSERT(HDR_HAS_RABD(hdr
));
6112 size
= HDR_GET_PSIZE(hdr
);
6113 hdr_abd
= hdr
->b_crypt_hdr
.b_rabd
;
6114 zio_flags
|= ZIO_FLAG_RAW
;
6116 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, !=, NULL
);
6117 size
= arc_hdr_size(hdr
);
6118 hdr_abd
= hdr
->b_l1hdr
.b_pabd
;
6120 if (arc_hdr_get_compress(hdr
) != ZIO_COMPRESS_OFF
) {
6121 zio_flags
|= ZIO_FLAG_RAW_COMPRESS
;
6125 * For authenticated bp's, we do not ask the ZIO layer
6126 * to authenticate them since this will cause the entire
6127 * IO to fail if the key isn't loaded. Instead, we
6128 * defer authentication until arc_buf_fill(), which will
6129 * verify the data when the key is available.
6131 if (BP_IS_AUTHENTICATED(bp
))
6132 zio_flags
|= ZIO_FLAG_RAW_ENCRYPT
;
6135 if (*arc_flags
& ARC_FLAG_PREFETCH
&&
6136 refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
))
6137 arc_hdr_set_flags(hdr
, ARC_FLAG_PREFETCH
);
6138 if (*arc_flags
& ARC_FLAG_PRESCIENT_PREFETCH
)
6139 arc_hdr_set_flags(hdr
, ARC_FLAG_PRESCIENT_PREFETCH
);
6140 if (*arc_flags
& ARC_FLAG_L2CACHE
)
6141 arc_hdr_set_flags(hdr
, ARC_FLAG_L2CACHE
);
6142 if (BP_IS_AUTHENTICATED(bp
))
6143 arc_hdr_set_flags(hdr
, ARC_FLAG_NOAUTH
);
6144 if (BP_GET_LEVEL(bp
) > 0)
6145 arc_hdr_set_flags(hdr
, ARC_FLAG_INDIRECT
);
6146 if (*arc_flags
& ARC_FLAG_PREDICTIVE_PREFETCH
)
6147 arc_hdr_set_flags(hdr
, ARC_FLAG_PREDICTIVE_PREFETCH
);
6148 ASSERT(!GHOST_STATE(hdr
->b_l1hdr
.b_state
));
6150 acb
= kmem_zalloc(sizeof (arc_callback_t
), KM_SLEEP
);
6151 acb
->acb_done
= done
;
6152 acb
->acb_private
= private;
6153 acb
->acb_compressed
= compressed_read
;
6154 acb
->acb_encrypted
= encrypted_read
;
6155 acb
->acb_noauth
= noauth_read
;
6156 acb
->acb_dsobj
= zb
->zb_objset
;
6158 ASSERT3P(hdr
->b_l1hdr
.b_acb
, ==, NULL
);
6159 hdr
->b_l1hdr
.b_acb
= acb
;
6160 arc_hdr_set_flags(hdr
, ARC_FLAG_IO_IN_PROGRESS
);
6162 if (HDR_HAS_L2HDR(hdr
) &&
6163 (vd
= hdr
->b_l2hdr
.b_dev
->l2ad_vdev
) != NULL
) {
6164 devw
= hdr
->b_l2hdr
.b_dev
->l2ad_writing
;
6165 addr
= hdr
->b_l2hdr
.b_daddr
;
6167 * Lock out device removal.
6169 if (vdev_is_dead(vd
) ||
6170 !spa_config_tryenter(spa
, SCL_L2ARC
, vd
, RW_READER
))
6175 * We count both async reads and scrub IOs as asynchronous so
6176 * that both can be upgraded in the event of a cache hit while
6177 * the read IO is still in-flight.
6179 if (priority
== ZIO_PRIORITY_ASYNC_READ
||
6180 priority
== ZIO_PRIORITY_SCRUB
)
6181 arc_hdr_set_flags(hdr
, ARC_FLAG_PRIO_ASYNC_READ
);
6183 arc_hdr_clear_flags(hdr
, ARC_FLAG_PRIO_ASYNC_READ
);
6186 * At this point, we have a level 1 cache miss. Try again in
6187 * L2ARC if possible.
6189 ASSERT3U(HDR_GET_LSIZE(hdr
), ==, lsize
);
6191 DTRACE_PROBE4(arc__miss
, arc_buf_hdr_t
*, hdr
, blkptr_t
*, bp
,
6192 uint64_t, lsize
, zbookmark_phys_t
*, zb
);
6193 ARCSTAT_BUMP(arcstat_misses
);
6194 ARCSTAT_CONDSTAT(!HDR_PREFETCH(hdr
),
6195 demand
, prefetch
, !HDR_ISTYPE_METADATA(hdr
),
6196 data
, metadata
, misses
);
6198 if (vd
!= NULL
&& l2arc_ndev
!= 0 && !(l2arc_norw
&& devw
)) {
6200 * Read from the L2ARC if the following are true:
6201 * 1. The L2ARC vdev was previously cached.
6202 * 2. This buffer still has L2ARC metadata.
6203 * 3. This buffer isn't currently writing to the L2ARC.
6204 * 4. The L2ARC entry wasn't evicted, which may
6205 * also have invalidated the vdev.
6206 * 5. This isn't prefetch and l2arc_noprefetch is set.
6208 if (HDR_HAS_L2HDR(hdr
) &&
6209 !HDR_L2_WRITING(hdr
) && !HDR_L2_EVICTED(hdr
) &&
6210 !(l2arc_noprefetch
&& HDR_PREFETCH(hdr
))) {
6211 l2arc_read_callback_t
*cb
;
6215 DTRACE_PROBE1(l2arc__hit
, arc_buf_hdr_t
*, hdr
);
6216 ARCSTAT_BUMP(arcstat_l2_hits
);
6217 atomic_inc_32(&hdr
->b_l2hdr
.b_hits
);
6219 cb
= kmem_zalloc(sizeof (l2arc_read_callback_t
),
6221 cb
->l2rcb_hdr
= hdr
;
6224 cb
->l2rcb_flags
= zio_flags
;
6226 asize
= vdev_psize_to_asize(vd
, size
);
6227 if (asize
!= size
) {
6228 abd
= abd_alloc_for_io(asize
,
6229 HDR_ISTYPE_METADATA(hdr
));
6230 cb
->l2rcb_abd
= abd
;
6235 ASSERT(addr
>= VDEV_LABEL_START_SIZE
&&
6236 addr
+ asize
<= vd
->vdev_psize
-
6237 VDEV_LABEL_END_SIZE
);
6240 * l2arc read. The SCL_L2ARC lock will be
6241 * released by l2arc_read_done().
6242 * Issue a null zio if the underlying buffer
6243 * was squashed to zero size by compression.
6245 ASSERT3U(arc_hdr_get_compress(hdr
), !=,
6246 ZIO_COMPRESS_EMPTY
);
6247 rzio
= zio_read_phys(pio
, vd
, addr
,
6250 l2arc_read_done
, cb
, priority
,
6251 zio_flags
| ZIO_FLAG_DONT_CACHE
|
6253 ZIO_FLAG_DONT_PROPAGATE
|
6254 ZIO_FLAG_DONT_RETRY
, B_FALSE
);
6255 acb
->acb_zio_head
= rzio
;
6257 if (hash_lock
!= NULL
)
6258 mutex_exit(hash_lock
);
6260 DTRACE_PROBE2(l2arc__read
, vdev_t
*, vd
,
6262 ARCSTAT_INCR(arcstat_l2_read_bytes
,
6263 HDR_GET_PSIZE(hdr
));
6265 if (*arc_flags
& ARC_FLAG_NOWAIT
) {
6270 ASSERT(*arc_flags
& ARC_FLAG_WAIT
);
6271 if (zio_wait(rzio
) == 0)
6274 /* l2arc read error; goto zio_read() */
6275 if (hash_lock
!= NULL
)
6276 mutex_enter(hash_lock
);
6278 DTRACE_PROBE1(l2arc__miss
,
6279 arc_buf_hdr_t
*, hdr
);
6280 ARCSTAT_BUMP(arcstat_l2_misses
);
6281 if (HDR_L2_WRITING(hdr
))
6282 ARCSTAT_BUMP(arcstat_l2_rw_clash
);
6283 spa_config_exit(spa
, SCL_L2ARC
, vd
);
6287 spa_config_exit(spa
, SCL_L2ARC
, vd
);
6288 if (l2arc_ndev
!= 0) {
6289 DTRACE_PROBE1(l2arc__miss
,
6290 arc_buf_hdr_t
*, hdr
);
6291 ARCSTAT_BUMP(arcstat_l2_misses
);
6295 rzio
= zio_read(pio
, spa
, bp
, hdr_abd
, size
,
6296 arc_read_done
, hdr
, priority
, zio_flags
, zb
);
6297 acb
->acb_zio_head
= rzio
;
6299 if (hash_lock
!= NULL
)
6300 mutex_exit(hash_lock
);
6302 if (*arc_flags
& ARC_FLAG_WAIT
) {
6303 rc
= zio_wait(rzio
);
6307 ASSERT(*arc_flags
& ARC_FLAG_NOWAIT
);
6312 spa_read_history_add(spa
, zb
, *arc_flags
);
6317 arc_add_prune_callback(arc_prune_func_t
*func
, void *private)
6321 p
= kmem_alloc(sizeof (*p
), KM_SLEEP
);
6323 p
->p_private
= private;
6324 list_link_init(&p
->p_node
);
6325 refcount_create(&p
->p_refcnt
);
6327 mutex_enter(&arc_prune_mtx
);
6328 refcount_add(&p
->p_refcnt
, &arc_prune_list
);
6329 list_insert_head(&arc_prune_list
, p
);
6330 mutex_exit(&arc_prune_mtx
);
6336 arc_remove_prune_callback(arc_prune_t
*p
)
6338 boolean_t wait
= B_FALSE
;
6339 mutex_enter(&arc_prune_mtx
);
6340 list_remove(&arc_prune_list
, p
);
6341 if (refcount_remove(&p
->p_refcnt
, &arc_prune_list
) > 0)
6343 mutex_exit(&arc_prune_mtx
);
6345 /* wait for arc_prune_task to finish */
6347 taskq_wait_outstanding(arc_prune_taskq
, 0);
6348 ASSERT0(refcount_count(&p
->p_refcnt
));
6349 refcount_destroy(&p
->p_refcnt
);
6350 kmem_free(p
, sizeof (*p
));
6354 * Notify the arc that a block was freed, and thus will never be used again.
6357 arc_freed(spa_t
*spa
, const blkptr_t
*bp
)
6360 kmutex_t
*hash_lock
;
6361 uint64_t guid
= spa_load_guid(spa
);
6363 ASSERT(!BP_IS_EMBEDDED(bp
));
6365 hdr
= buf_hash_find(guid
, bp
, &hash_lock
);
6370 * We might be trying to free a block that is still doing I/O
6371 * (i.e. prefetch) or has a reference (i.e. a dedup-ed,
6372 * dmu_sync-ed block). If this block is being prefetched, then it
6373 * would still have the ARC_FLAG_IO_IN_PROGRESS flag set on the hdr
6374 * until the I/O completes. A block may also have a reference if it is
6375 * part of a dedup-ed, dmu_synced write. The dmu_sync() function would
6376 * have written the new block to its final resting place on disk but
6377 * without the dedup flag set. This would have left the hdr in the MRU
6378 * state and discoverable. When the txg finally syncs it detects that
6379 * the block was overridden in open context and issues an override I/O.
6380 * Since this is a dedup block, the override I/O will determine if the
6381 * block is already in the DDT. If so, then it will replace the io_bp
6382 * with the bp from the DDT and allow the I/O to finish. When the I/O
6383 * reaches the done callback, dbuf_write_override_done, it will
6384 * check to see if the io_bp and io_bp_override are identical.
6385 * If they are not, then it indicates that the bp was replaced with
6386 * the bp in the DDT and the override bp is freed. This allows
6387 * us to arrive here with a reference on a block that is being
6388 * freed. So if we have an I/O in progress, or a reference to
6389 * this hdr, then we don't destroy the hdr.
6391 if (!HDR_HAS_L1HDR(hdr
) || (!HDR_IO_IN_PROGRESS(hdr
) &&
6392 refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
))) {
6393 arc_change_state(arc_anon
, hdr
, hash_lock
);
6394 arc_hdr_destroy(hdr
);
6395 mutex_exit(hash_lock
);
6397 mutex_exit(hash_lock
);
6403 * Release this buffer from the cache, making it an anonymous buffer. This
6404 * must be done after a read and prior to modifying the buffer contents.
6405 * If the buffer has more than one reference, we must make
6406 * a new hdr for the buffer.
6409 arc_release(arc_buf_t
*buf
, void *tag
)
6411 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
6414 * It would be nice to assert that if its DMU metadata (level >
6415 * 0 || it's the dnode file), then it must be syncing context.
6416 * But we don't know that information at this level.
6419 mutex_enter(&buf
->b_evict_lock
);
6421 ASSERT(HDR_HAS_L1HDR(hdr
));
6424 * We don't grab the hash lock prior to this check, because if
6425 * the buffer's header is in the arc_anon state, it won't be
6426 * linked into the hash table.
6428 if (hdr
->b_l1hdr
.b_state
== arc_anon
) {
6429 mutex_exit(&buf
->b_evict_lock
);
6430 ASSERT(!HDR_IO_IN_PROGRESS(hdr
));
6431 ASSERT(!HDR_IN_HASH_TABLE(hdr
));
6432 ASSERT(!HDR_HAS_L2HDR(hdr
));
6433 ASSERT(HDR_EMPTY(hdr
));
6435 ASSERT3U(hdr
->b_l1hdr
.b_bufcnt
, ==, 1);
6436 ASSERT3S(refcount_count(&hdr
->b_l1hdr
.b_refcnt
), ==, 1);
6437 ASSERT(!list_link_active(&hdr
->b_l1hdr
.b_arc_node
));
6439 hdr
->b_l1hdr
.b_arc_access
= 0;
6442 * If the buf is being overridden then it may already
6443 * have a hdr that is not empty.
6445 buf_discard_identity(hdr
);
6451 kmutex_t
*hash_lock
= HDR_LOCK(hdr
);
6452 mutex_enter(hash_lock
);
6455 * This assignment is only valid as long as the hash_lock is
6456 * held, we must be careful not to reference state or the
6457 * b_state field after dropping the lock.
6459 arc_state_t
*state
= hdr
->b_l1hdr
.b_state
;
6460 ASSERT3P(hash_lock
, ==, HDR_LOCK(hdr
));
6461 ASSERT3P(state
, !=, arc_anon
);
6463 /* this buffer is not on any list */
6464 ASSERT3S(refcount_count(&hdr
->b_l1hdr
.b_refcnt
), >, 0);
6466 if (HDR_HAS_L2HDR(hdr
)) {
6467 mutex_enter(&hdr
->b_l2hdr
.b_dev
->l2ad_mtx
);
6470 * We have to recheck this conditional again now that
6471 * we're holding the l2ad_mtx to prevent a race with
6472 * another thread which might be concurrently calling
6473 * l2arc_evict(). In that case, l2arc_evict() might have
6474 * destroyed the header's L2 portion as we were waiting
6475 * to acquire the l2ad_mtx.
6477 if (HDR_HAS_L2HDR(hdr
))
6478 arc_hdr_l2hdr_destroy(hdr
);
6480 mutex_exit(&hdr
->b_l2hdr
.b_dev
->l2ad_mtx
);
6484 * Do we have more than one buf?
6486 if (hdr
->b_l1hdr
.b_bufcnt
> 1) {
6487 arc_buf_hdr_t
*nhdr
;
6488 uint64_t spa
= hdr
->b_spa
;
6489 uint64_t psize
= HDR_GET_PSIZE(hdr
);
6490 uint64_t lsize
= HDR_GET_LSIZE(hdr
);
6491 boolean_t
protected = HDR_PROTECTED(hdr
);
6492 enum zio_compress compress
= arc_hdr_get_compress(hdr
);
6493 arc_buf_contents_t type
= arc_buf_type(hdr
);
6494 VERIFY3U(hdr
->b_type
, ==, type
);
6496 ASSERT(hdr
->b_l1hdr
.b_buf
!= buf
|| buf
->b_next
!= NULL
);
6497 (void) remove_reference(hdr
, hash_lock
, tag
);
6499 if (arc_buf_is_shared(buf
) && !ARC_BUF_COMPRESSED(buf
)) {
6500 ASSERT3P(hdr
->b_l1hdr
.b_buf
, !=, buf
);
6501 ASSERT(ARC_BUF_LAST(buf
));
6505 * Pull the data off of this hdr and attach it to
6506 * a new anonymous hdr. Also find the last buffer
6507 * in the hdr's buffer list.
6509 arc_buf_t
*lastbuf
= arc_buf_remove(hdr
, buf
);
6510 ASSERT3P(lastbuf
, !=, NULL
);
6513 * If the current arc_buf_t and the hdr are sharing their data
6514 * buffer, then we must stop sharing that block.
6516 if (arc_buf_is_shared(buf
)) {
6517 ASSERT3P(hdr
->b_l1hdr
.b_buf
, !=, buf
);
6518 VERIFY(!arc_buf_is_shared(lastbuf
));
6521 * First, sever the block sharing relationship between
6522 * buf and the arc_buf_hdr_t.
6524 arc_unshare_buf(hdr
, buf
);
6527 * Now we need to recreate the hdr's b_pabd. Since we
6528 * have lastbuf handy, we try to share with it, but if
6529 * we can't then we allocate a new b_pabd and copy the
6530 * data from buf into it.
6532 if (arc_can_share(hdr
, lastbuf
)) {
6533 arc_share_buf(hdr
, lastbuf
);
6535 arc_hdr_alloc_abd(hdr
, B_FALSE
);
6536 abd_copy_from_buf(hdr
->b_l1hdr
.b_pabd
,
6537 buf
->b_data
, psize
);
6539 VERIFY3P(lastbuf
->b_data
, !=, NULL
);
6540 } else if (HDR_SHARED_DATA(hdr
)) {
6542 * Uncompressed shared buffers are always at the end
6543 * of the list. Compressed buffers don't have the
6544 * same requirements. This makes it hard to
6545 * simply assert that the lastbuf is shared so
6546 * we rely on the hdr's compression flags to determine
6547 * if we have a compressed, shared buffer.
6549 ASSERT(arc_buf_is_shared(lastbuf
) ||
6550 arc_hdr_get_compress(hdr
) != ZIO_COMPRESS_OFF
);
6551 ASSERT(!ARC_BUF_SHARED(buf
));
6554 ASSERT(hdr
->b_l1hdr
.b_pabd
!= NULL
|| HDR_HAS_RABD(hdr
));
6555 ASSERT3P(state
, !=, arc_l2c_only
);
6557 (void) refcount_remove_many(&state
->arcs_size
,
6558 arc_buf_size(buf
), buf
);
6560 if (refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
)) {
6561 ASSERT3P(state
, !=, arc_l2c_only
);
6562 (void) refcount_remove_many(&state
->arcs_esize
[type
],
6563 arc_buf_size(buf
), buf
);
6566 hdr
->b_l1hdr
.b_bufcnt
-= 1;
6567 if (ARC_BUF_ENCRYPTED(buf
))
6568 hdr
->b_crypt_hdr
.b_ebufcnt
-= 1;
6570 arc_cksum_verify(buf
);
6571 arc_buf_unwatch(buf
);
6573 /* if this is the last uncompressed buf free the checksum */
6574 if (!arc_hdr_has_uncompressed_buf(hdr
))
6575 arc_cksum_free(hdr
);
6577 mutex_exit(hash_lock
);
6580 * Allocate a new hdr. The new hdr will contain a b_pabd
6581 * buffer which will be freed in arc_write().
6583 nhdr
= arc_hdr_alloc(spa
, psize
, lsize
, protected,
6584 compress
, type
, HDR_HAS_RABD(hdr
));
6585 ASSERT3P(nhdr
->b_l1hdr
.b_buf
, ==, NULL
);
6586 ASSERT0(nhdr
->b_l1hdr
.b_bufcnt
);
6587 ASSERT0(refcount_count(&nhdr
->b_l1hdr
.b_refcnt
));
6588 VERIFY3U(nhdr
->b_type
, ==, type
);
6589 ASSERT(!HDR_SHARED_DATA(nhdr
));
6591 nhdr
->b_l1hdr
.b_buf
= buf
;
6592 nhdr
->b_l1hdr
.b_bufcnt
= 1;
6593 if (ARC_BUF_ENCRYPTED(buf
))
6594 nhdr
->b_crypt_hdr
.b_ebufcnt
= 1;
6595 nhdr
->b_l1hdr
.b_mru_hits
= 0;
6596 nhdr
->b_l1hdr
.b_mru_ghost_hits
= 0;
6597 nhdr
->b_l1hdr
.b_mfu_hits
= 0;
6598 nhdr
->b_l1hdr
.b_mfu_ghost_hits
= 0;
6599 nhdr
->b_l1hdr
.b_l2_hits
= 0;
6600 (void) refcount_add(&nhdr
->b_l1hdr
.b_refcnt
, tag
);
6603 mutex_exit(&buf
->b_evict_lock
);
6604 (void) refcount_add_many(&arc_anon
->arcs_size
,
6605 HDR_GET_LSIZE(nhdr
), buf
);
6607 mutex_exit(&buf
->b_evict_lock
);
6608 ASSERT(refcount_count(&hdr
->b_l1hdr
.b_refcnt
) == 1);
6609 /* protected by hash lock, or hdr is on arc_anon */
6610 ASSERT(!multilist_link_active(&hdr
->b_l1hdr
.b_arc_node
));
6611 ASSERT(!HDR_IO_IN_PROGRESS(hdr
));
6612 hdr
->b_l1hdr
.b_mru_hits
= 0;
6613 hdr
->b_l1hdr
.b_mru_ghost_hits
= 0;
6614 hdr
->b_l1hdr
.b_mfu_hits
= 0;
6615 hdr
->b_l1hdr
.b_mfu_ghost_hits
= 0;
6616 hdr
->b_l1hdr
.b_l2_hits
= 0;
6617 arc_change_state(arc_anon
, hdr
, hash_lock
);
6618 hdr
->b_l1hdr
.b_arc_access
= 0;
6620 mutex_exit(hash_lock
);
6621 buf_discard_identity(hdr
);
6627 arc_released(arc_buf_t
*buf
)
6631 mutex_enter(&buf
->b_evict_lock
);
6632 released
= (buf
->b_data
!= NULL
&&
6633 buf
->b_hdr
->b_l1hdr
.b_state
== arc_anon
);
6634 mutex_exit(&buf
->b_evict_lock
);
6640 arc_referenced(arc_buf_t
*buf
)
6644 mutex_enter(&buf
->b_evict_lock
);
6645 referenced
= (refcount_count(&buf
->b_hdr
->b_l1hdr
.b_refcnt
));
6646 mutex_exit(&buf
->b_evict_lock
);
6647 return (referenced
);
6652 arc_write_ready(zio_t
*zio
)
6654 arc_write_callback_t
*callback
= zio
->io_private
;
6655 arc_buf_t
*buf
= callback
->awcb_buf
;
6656 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
6657 blkptr_t
*bp
= zio
->io_bp
;
6658 uint64_t psize
= BP_IS_HOLE(bp
) ? 0 : BP_GET_PSIZE(bp
);
6659 fstrans_cookie_t cookie
= spl_fstrans_mark();
6661 ASSERT(HDR_HAS_L1HDR(hdr
));
6662 ASSERT(!refcount_is_zero(&buf
->b_hdr
->b_l1hdr
.b_refcnt
));
6663 ASSERT(hdr
->b_l1hdr
.b_bufcnt
> 0);
6666 * If we're reexecuting this zio because the pool suspended, then
6667 * cleanup any state that was previously set the first time the
6668 * callback was invoked.
6670 if (zio
->io_flags
& ZIO_FLAG_REEXECUTED
) {
6671 arc_cksum_free(hdr
);
6672 arc_buf_unwatch(buf
);
6673 if (hdr
->b_l1hdr
.b_pabd
!= NULL
) {
6674 if (arc_buf_is_shared(buf
)) {
6675 arc_unshare_buf(hdr
, buf
);
6677 arc_hdr_free_abd(hdr
, B_FALSE
);
6681 if (HDR_HAS_RABD(hdr
))
6682 arc_hdr_free_abd(hdr
, B_TRUE
);
6684 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
6685 ASSERT(!HDR_HAS_RABD(hdr
));
6686 ASSERT(!HDR_SHARED_DATA(hdr
));
6687 ASSERT(!arc_buf_is_shared(buf
));
6689 callback
->awcb_ready(zio
, buf
, callback
->awcb_private
);
6691 if (HDR_IO_IN_PROGRESS(hdr
))
6692 ASSERT(zio
->io_flags
& ZIO_FLAG_REEXECUTED
);
6694 arc_hdr_set_flags(hdr
, ARC_FLAG_IO_IN_PROGRESS
);
6696 if (BP_IS_PROTECTED(bp
) != !!HDR_PROTECTED(hdr
))
6697 hdr
= arc_hdr_realloc_crypt(hdr
, BP_IS_PROTECTED(bp
));
6699 if (BP_IS_PROTECTED(bp
)) {
6700 /* ZIL blocks are written through zio_rewrite */
6701 ASSERT3U(BP_GET_TYPE(bp
), !=, DMU_OT_INTENT_LOG
);
6702 ASSERT(HDR_PROTECTED(hdr
));
6704 hdr
->b_crypt_hdr
.b_ot
= BP_GET_TYPE(bp
);
6705 hdr
->b_crypt_hdr
.b_dsobj
= zio
->io_bookmark
.zb_objset
;
6706 zio_crypt_decode_params_bp(bp
, hdr
->b_crypt_hdr
.b_salt
,
6707 hdr
->b_crypt_hdr
.b_iv
);
6708 zio_crypt_decode_mac_bp(bp
, hdr
->b_crypt_hdr
.b_mac
);
6712 * If this block was written for raw encryption but the zio layer
6713 * ended up only authenticating it, adjust the buffer flags now.
6715 if (BP_IS_AUTHENTICATED(bp
) && ARC_BUF_ENCRYPTED(buf
)) {
6716 arc_hdr_set_flags(hdr
, ARC_FLAG_NOAUTH
);
6717 buf
->b_flags
&= ~ARC_BUF_FLAG_ENCRYPTED
;
6718 if (BP_GET_COMPRESS(bp
) == ZIO_COMPRESS_OFF
)
6719 buf
->b_flags
&= ~ARC_BUF_FLAG_COMPRESSED
;
6722 /* this must be done after the buffer flags are adjusted */
6723 arc_cksum_compute(buf
);
6725 enum zio_compress compress
;
6726 if (BP_IS_HOLE(bp
) || BP_IS_EMBEDDED(bp
)) {
6727 compress
= ZIO_COMPRESS_OFF
;
6729 ASSERT3U(HDR_GET_LSIZE(hdr
), ==, BP_GET_LSIZE(bp
));
6730 compress
= BP_GET_COMPRESS(bp
);
6732 HDR_SET_PSIZE(hdr
, psize
);
6733 arc_hdr_set_compress(hdr
, compress
);
6735 if (zio
->io_error
!= 0 || psize
== 0)
6739 * Fill the hdr with data. If the buffer is encrypted we have no choice
6740 * but to copy the data into b_radb. If the hdr is compressed, the data
6741 * we want is available from the zio, otherwise we can take it from
6744 * We might be able to share the buf's data with the hdr here. However,
6745 * doing so would cause the ARC to be full of linear ABDs if we write a
6746 * lot of shareable data. As a compromise, we check whether scattered
6747 * ABDs are allowed, and assume that if they are then the user wants
6748 * the ARC to be primarily filled with them regardless of the data being
6749 * written. Therefore, if they're allowed then we allocate one and copy
6750 * the data into it; otherwise, we share the data directly if we can.
6752 if (ARC_BUF_ENCRYPTED(buf
)) {
6753 ASSERT3U(psize
, >, 0);
6754 ASSERT(ARC_BUF_COMPRESSED(buf
));
6755 arc_hdr_alloc_abd(hdr
, B_TRUE
);
6756 abd_copy(hdr
->b_crypt_hdr
.b_rabd
, zio
->io_abd
, psize
);
6757 } else if (zfs_abd_scatter_enabled
|| !arc_can_share(hdr
, buf
)) {
6759 * Ideally, we would always copy the io_abd into b_pabd, but the
6760 * user may have disabled compressed ARC, thus we must check the
6761 * hdr's compression setting rather than the io_bp's.
6763 if (BP_IS_ENCRYPTED(bp
)) {
6764 ASSERT3U(psize
, >, 0);
6765 arc_hdr_alloc_abd(hdr
, B_TRUE
);
6766 abd_copy(hdr
->b_crypt_hdr
.b_rabd
, zio
->io_abd
, psize
);
6767 } else if (arc_hdr_get_compress(hdr
) != ZIO_COMPRESS_OFF
&&
6768 !ARC_BUF_COMPRESSED(buf
)) {
6769 ASSERT3U(psize
, >, 0);
6770 arc_hdr_alloc_abd(hdr
, B_FALSE
);
6771 abd_copy(hdr
->b_l1hdr
.b_pabd
, zio
->io_abd
, psize
);
6773 ASSERT3U(zio
->io_orig_size
, ==, arc_hdr_size(hdr
));
6774 arc_hdr_alloc_abd(hdr
, B_FALSE
);
6775 abd_copy_from_buf(hdr
->b_l1hdr
.b_pabd
, buf
->b_data
,
6779 ASSERT3P(buf
->b_data
, ==, abd_to_buf(zio
->io_orig_abd
));
6780 ASSERT3U(zio
->io_orig_size
, ==, arc_buf_size(buf
));
6781 ASSERT3U(hdr
->b_l1hdr
.b_bufcnt
, ==, 1);
6783 arc_share_buf(hdr
, buf
);
6787 arc_hdr_verify(hdr
, bp
);
6788 spl_fstrans_unmark(cookie
);
6792 arc_write_children_ready(zio_t
*zio
)
6794 arc_write_callback_t
*callback
= zio
->io_private
;
6795 arc_buf_t
*buf
= callback
->awcb_buf
;
6797 callback
->awcb_children_ready(zio
, buf
, callback
->awcb_private
);
6801 * The SPA calls this callback for each physical write that happens on behalf
6802 * of a logical write. See the comment in dbuf_write_physdone() for details.
6805 arc_write_physdone(zio_t
*zio
)
6807 arc_write_callback_t
*cb
= zio
->io_private
;
6808 if (cb
->awcb_physdone
!= NULL
)
6809 cb
->awcb_physdone(zio
, cb
->awcb_buf
, cb
->awcb_private
);
6813 arc_write_done(zio_t
*zio
)
6815 arc_write_callback_t
*callback
= zio
->io_private
;
6816 arc_buf_t
*buf
= callback
->awcb_buf
;
6817 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
6819 ASSERT3P(hdr
->b_l1hdr
.b_acb
, ==, NULL
);
6821 if (zio
->io_error
== 0) {
6822 arc_hdr_verify(hdr
, zio
->io_bp
);
6824 if (BP_IS_HOLE(zio
->io_bp
) || BP_IS_EMBEDDED(zio
->io_bp
)) {
6825 buf_discard_identity(hdr
);
6827 hdr
->b_dva
= *BP_IDENTITY(zio
->io_bp
);
6828 hdr
->b_birth
= BP_PHYSICAL_BIRTH(zio
->io_bp
);
6831 ASSERT(HDR_EMPTY(hdr
));
6835 * If the block to be written was all-zero or compressed enough to be
6836 * embedded in the BP, no write was performed so there will be no
6837 * dva/birth/checksum. The buffer must therefore remain anonymous
6840 if (!HDR_EMPTY(hdr
)) {
6841 arc_buf_hdr_t
*exists
;
6842 kmutex_t
*hash_lock
;
6844 ASSERT3U(zio
->io_error
, ==, 0);
6846 arc_cksum_verify(buf
);
6848 exists
= buf_hash_insert(hdr
, &hash_lock
);
6849 if (exists
!= NULL
) {
6851 * This can only happen if we overwrite for
6852 * sync-to-convergence, because we remove
6853 * buffers from the hash table when we arc_free().
6855 if (zio
->io_flags
& ZIO_FLAG_IO_REWRITE
) {
6856 if (!BP_EQUAL(&zio
->io_bp_orig
, zio
->io_bp
))
6857 panic("bad overwrite, hdr=%p exists=%p",
6858 (void *)hdr
, (void *)exists
);
6859 ASSERT(refcount_is_zero(
6860 &exists
->b_l1hdr
.b_refcnt
));
6861 arc_change_state(arc_anon
, exists
, hash_lock
);
6862 mutex_exit(hash_lock
);
6863 arc_hdr_destroy(exists
);
6864 exists
= buf_hash_insert(hdr
, &hash_lock
);
6865 ASSERT3P(exists
, ==, NULL
);
6866 } else if (zio
->io_flags
& ZIO_FLAG_NOPWRITE
) {
6868 ASSERT(zio
->io_prop
.zp_nopwrite
);
6869 if (!BP_EQUAL(&zio
->io_bp_orig
, zio
->io_bp
))
6870 panic("bad nopwrite, hdr=%p exists=%p",
6871 (void *)hdr
, (void *)exists
);
6874 ASSERT(hdr
->b_l1hdr
.b_bufcnt
== 1);
6875 ASSERT(hdr
->b_l1hdr
.b_state
== arc_anon
);
6876 ASSERT(BP_GET_DEDUP(zio
->io_bp
));
6877 ASSERT(BP_GET_LEVEL(zio
->io_bp
) == 0);
6880 arc_hdr_clear_flags(hdr
, ARC_FLAG_IO_IN_PROGRESS
);
6881 /* if it's not anon, we are doing a scrub */
6882 if (exists
== NULL
&& hdr
->b_l1hdr
.b_state
== arc_anon
)
6883 arc_access(hdr
, hash_lock
);
6884 mutex_exit(hash_lock
);
6886 arc_hdr_clear_flags(hdr
, ARC_FLAG_IO_IN_PROGRESS
);
6889 ASSERT(!refcount_is_zero(&hdr
->b_l1hdr
.b_refcnt
));
6890 callback
->awcb_done(zio
, buf
, callback
->awcb_private
);
6892 abd_put(zio
->io_abd
);
6893 kmem_free(callback
, sizeof (arc_write_callback_t
));
6897 arc_write(zio_t
*pio
, spa_t
*spa
, uint64_t txg
,
6898 blkptr_t
*bp
, arc_buf_t
*buf
, boolean_t l2arc
,
6899 const zio_prop_t
*zp
, arc_write_done_func_t
*ready
,
6900 arc_write_done_func_t
*children_ready
, arc_write_done_func_t
*physdone
,
6901 arc_write_done_func_t
*done
, void *private, zio_priority_t priority
,
6902 int zio_flags
, const zbookmark_phys_t
*zb
)
6904 arc_buf_hdr_t
*hdr
= buf
->b_hdr
;
6905 arc_write_callback_t
*callback
;
6907 zio_prop_t localprop
= *zp
;
6909 ASSERT3P(ready
, !=, NULL
);
6910 ASSERT3P(done
, !=, NULL
);
6911 ASSERT(!HDR_IO_ERROR(hdr
));
6912 ASSERT(!HDR_IO_IN_PROGRESS(hdr
));
6913 ASSERT3P(hdr
->b_l1hdr
.b_acb
, ==, NULL
);
6914 ASSERT3U(hdr
->b_l1hdr
.b_bufcnt
, >, 0);
6916 arc_hdr_set_flags(hdr
, ARC_FLAG_L2CACHE
);
6918 if (ARC_BUF_ENCRYPTED(buf
)) {
6919 ASSERT(ARC_BUF_COMPRESSED(buf
));
6920 localprop
.zp_encrypt
= B_TRUE
;
6921 localprop
.zp_compress
= HDR_GET_COMPRESS(hdr
);
6922 localprop
.zp_byteorder
=
6923 (hdr
->b_l1hdr
.b_byteswap
== DMU_BSWAP_NUMFUNCS
) ?
6924 ZFS_HOST_BYTEORDER
: !ZFS_HOST_BYTEORDER
;
6925 bcopy(hdr
->b_crypt_hdr
.b_salt
, localprop
.zp_salt
,
6927 bcopy(hdr
->b_crypt_hdr
.b_iv
, localprop
.zp_iv
,
6929 bcopy(hdr
->b_crypt_hdr
.b_mac
, localprop
.zp_mac
,
6931 if (DMU_OT_IS_ENCRYPTED(localprop
.zp_type
)) {
6932 localprop
.zp_nopwrite
= B_FALSE
;
6933 localprop
.zp_copies
=
6934 MIN(localprop
.zp_copies
, SPA_DVAS_PER_BP
- 1);
6936 zio_flags
|= ZIO_FLAG_RAW
;
6937 } else if (ARC_BUF_COMPRESSED(buf
)) {
6938 ASSERT3U(HDR_GET_LSIZE(hdr
), !=, arc_buf_size(buf
));
6939 localprop
.zp_compress
= HDR_GET_COMPRESS(hdr
);
6940 zio_flags
|= ZIO_FLAG_RAW_COMPRESS
;
6942 callback
= kmem_zalloc(sizeof (arc_write_callback_t
), KM_SLEEP
);
6943 callback
->awcb_ready
= ready
;
6944 callback
->awcb_children_ready
= children_ready
;
6945 callback
->awcb_physdone
= physdone
;
6946 callback
->awcb_done
= done
;
6947 callback
->awcb_private
= private;
6948 callback
->awcb_buf
= buf
;
6951 * The hdr's b_pabd is now stale, free it now. A new data block
6952 * will be allocated when the zio pipeline calls arc_write_ready().
6954 if (hdr
->b_l1hdr
.b_pabd
!= NULL
) {
6956 * If the buf is currently sharing the data block with
6957 * the hdr then we need to break that relationship here.
6958 * The hdr will remain with a NULL data pointer and the
6959 * buf will take sole ownership of the block.
6961 if (arc_buf_is_shared(buf
)) {
6962 arc_unshare_buf(hdr
, buf
);
6964 arc_hdr_free_abd(hdr
, B_FALSE
);
6966 VERIFY3P(buf
->b_data
, !=, NULL
);
6969 if (HDR_HAS_RABD(hdr
))
6970 arc_hdr_free_abd(hdr
, B_TRUE
);
6972 if (!(zio_flags
& ZIO_FLAG_RAW
))
6973 arc_hdr_set_compress(hdr
, ZIO_COMPRESS_OFF
);
6975 ASSERT(!arc_buf_is_shared(buf
));
6976 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, ==, NULL
);
6978 zio
= zio_write(pio
, spa
, txg
, bp
,
6979 abd_get_from_buf(buf
->b_data
, HDR_GET_LSIZE(hdr
)),
6980 HDR_GET_LSIZE(hdr
), arc_buf_size(buf
), &localprop
, arc_write_ready
,
6981 (children_ready
!= NULL
) ? arc_write_children_ready
: NULL
,
6982 arc_write_physdone
, arc_write_done
, callback
,
6983 priority
, zio_flags
, zb
);
6989 arc_memory_throttle(uint64_t reserve
, uint64_t txg
)
6992 uint64_t available_memory
= arc_free_memory();
6993 static uint64_t page_load
= 0;
6994 static uint64_t last_txg
= 0;
6998 MIN(available_memory
, vmem_size(heap_arena
, VMEM_FREE
));
7001 if (available_memory
> arc_all_memory() * arc_lotsfree_percent
/ 100)
7004 if (txg
> last_txg
) {
7009 * If we are in pageout, we know that memory is already tight,
7010 * the arc is already going to be evicting, so we just want to
7011 * continue to let page writes occur as quickly as possible.
7013 if (current_is_kswapd()) {
7014 if (page_load
> MAX(arc_sys_free
/ 4, available_memory
) / 4) {
7015 DMU_TX_STAT_BUMP(dmu_tx_memory_reclaim
);
7016 return (SET_ERROR(ERESTART
));
7018 /* Note: reserve is inflated, so we deflate */
7019 page_load
+= reserve
/ 8;
7021 } else if (page_load
> 0 && arc_reclaim_needed()) {
7022 /* memory is low, delay before restarting */
7023 ARCSTAT_INCR(arcstat_memory_throttle_count
, 1);
7024 DMU_TX_STAT_BUMP(dmu_tx_memory_reclaim
);
7025 return (SET_ERROR(EAGAIN
));
7033 arc_tempreserve_clear(uint64_t reserve
)
7035 atomic_add_64(&arc_tempreserve
, -reserve
);
7036 ASSERT((int64_t)arc_tempreserve
>= 0);
7040 arc_tempreserve_space(uint64_t reserve
, uint64_t txg
)
7046 reserve
> arc_c
/4 &&
7047 reserve
* 4 > (2ULL << SPA_MAXBLOCKSHIFT
))
7048 arc_c
= MIN(arc_c_max
, reserve
* 4);
7051 * Throttle when the calculated memory footprint for the TXG
7052 * exceeds the target ARC size.
7054 if (reserve
> arc_c
) {
7055 DMU_TX_STAT_BUMP(dmu_tx_memory_reserve
);
7056 return (SET_ERROR(ERESTART
));
7060 * Don't count loaned bufs as in flight dirty data to prevent long
7061 * network delays from blocking transactions that are ready to be
7062 * assigned to a txg.
7065 /* assert that it has not wrapped around */
7066 ASSERT3S(atomic_add_64_nv(&arc_loaned_bytes
, 0), >=, 0);
7068 anon_size
= MAX((int64_t)(refcount_count(&arc_anon
->arcs_size
) -
7069 arc_loaned_bytes
), 0);
7072 * Writes will, almost always, require additional memory allocations
7073 * in order to compress/encrypt/etc the data. We therefore need to
7074 * make sure that there is sufficient available memory for this.
7076 error
= arc_memory_throttle(reserve
, txg
);
7081 * Throttle writes when the amount of dirty data in the cache
7082 * gets too large. We try to keep the cache less than half full
7083 * of dirty blocks so that our sync times don't grow too large.
7084 * Note: if two requests come in concurrently, we might let them
7085 * both succeed, when one of them should fail. Not a huge deal.
7088 if (reserve
+ arc_tempreserve
+ anon_size
> arc_c
/ 2 &&
7089 anon_size
> arc_c
/ 4) {
7090 uint64_t meta_esize
=
7091 refcount_count(&arc_anon
->arcs_esize
[ARC_BUFC_METADATA
]);
7092 uint64_t data_esize
=
7093 refcount_count(&arc_anon
->arcs_esize
[ARC_BUFC_DATA
]);
7094 dprintf("failing, arc_tempreserve=%lluK anon_meta=%lluK "
7095 "anon_data=%lluK tempreserve=%lluK arc_c=%lluK\n",
7096 arc_tempreserve
>> 10, meta_esize
>> 10,
7097 data_esize
>> 10, reserve
>> 10, arc_c
>> 10);
7098 DMU_TX_STAT_BUMP(dmu_tx_dirty_throttle
);
7099 return (SET_ERROR(ERESTART
));
7101 atomic_add_64(&arc_tempreserve
, reserve
);
7106 arc_kstat_update_state(arc_state_t
*state
, kstat_named_t
*size
,
7107 kstat_named_t
*evict_data
, kstat_named_t
*evict_metadata
)
7109 size
->value
.ui64
= refcount_count(&state
->arcs_size
);
7110 evict_data
->value
.ui64
=
7111 refcount_count(&state
->arcs_esize
[ARC_BUFC_DATA
]);
7112 evict_metadata
->value
.ui64
=
7113 refcount_count(&state
->arcs_esize
[ARC_BUFC_METADATA
]);
7117 arc_kstat_update(kstat_t
*ksp
, int rw
)
7119 arc_stats_t
*as
= ksp
->ks_data
;
7121 if (rw
== KSTAT_WRITE
) {
7122 return (SET_ERROR(EACCES
));
7124 arc_kstat_update_state(arc_anon
,
7125 &as
->arcstat_anon_size
,
7126 &as
->arcstat_anon_evictable_data
,
7127 &as
->arcstat_anon_evictable_metadata
);
7128 arc_kstat_update_state(arc_mru
,
7129 &as
->arcstat_mru_size
,
7130 &as
->arcstat_mru_evictable_data
,
7131 &as
->arcstat_mru_evictable_metadata
);
7132 arc_kstat_update_state(arc_mru_ghost
,
7133 &as
->arcstat_mru_ghost_size
,
7134 &as
->arcstat_mru_ghost_evictable_data
,
7135 &as
->arcstat_mru_ghost_evictable_metadata
);
7136 arc_kstat_update_state(arc_mfu
,
7137 &as
->arcstat_mfu_size
,
7138 &as
->arcstat_mfu_evictable_data
,
7139 &as
->arcstat_mfu_evictable_metadata
);
7140 arc_kstat_update_state(arc_mfu_ghost
,
7141 &as
->arcstat_mfu_ghost_size
,
7142 &as
->arcstat_mfu_ghost_evictable_data
,
7143 &as
->arcstat_mfu_ghost_evictable_metadata
);
7145 as
->arcstat_memory_all_bytes
.value
.ui64
=
7147 as
->arcstat_memory_free_bytes
.value
.ui64
=
7149 as
->arcstat_memory_available_bytes
.value
.i64
=
7150 arc_available_memory();
7157 * This function *must* return indices evenly distributed between all
7158 * sublists of the multilist. This is needed due to how the ARC eviction
7159 * code is laid out; arc_evict_state() assumes ARC buffers are evenly
7160 * distributed between all sublists and uses this assumption when
7161 * deciding which sublist to evict from and how much to evict from it.
7164 arc_state_multilist_index_func(multilist_t
*ml
, void *obj
)
7166 arc_buf_hdr_t
*hdr
= obj
;
7169 * We rely on b_dva to generate evenly distributed index
7170 * numbers using buf_hash below. So, as an added precaution,
7171 * let's make sure we never add empty buffers to the arc lists.
7173 ASSERT(!HDR_EMPTY(hdr
));
7176 * The assumption here, is the hash value for a given
7177 * arc_buf_hdr_t will remain constant throughout its lifetime
7178 * (i.e. its b_spa, b_dva, and b_birth fields don't change).
7179 * Thus, we don't need to store the header's sublist index
7180 * on insertion, as this index can be recalculated on removal.
7182 * Also, the low order bits of the hash value are thought to be
7183 * distributed evenly. Otherwise, in the case that the multilist
7184 * has a power of two number of sublists, each sublists' usage
7185 * would not be evenly distributed.
7187 return (buf_hash(hdr
->b_spa
, &hdr
->b_dva
, hdr
->b_birth
) %
7188 multilist_get_num_sublists(ml
));
7192 * Called during module initialization and periodically thereafter to
7193 * apply reasonable changes to the exposed performance tunings. Non-zero
7194 * zfs_* values which differ from the currently set values will be applied.
7197 arc_tuning_update(void)
7199 uint64_t allmem
= arc_all_memory();
7200 unsigned long limit
;
7202 /* Valid range: 64M - <all physical memory> */
7203 if ((zfs_arc_max
) && (zfs_arc_max
!= arc_c_max
) &&
7204 (zfs_arc_max
> 64 << 20) && (zfs_arc_max
< allmem
) &&
7205 (zfs_arc_max
> arc_c_min
)) {
7206 arc_c_max
= zfs_arc_max
;
7208 arc_p
= (arc_c
>> 1);
7209 if (arc_meta_limit
> arc_c_max
)
7210 arc_meta_limit
= arc_c_max
;
7211 if (arc_dnode_limit
> arc_meta_limit
)
7212 arc_dnode_limit
= arc_meta_limit
;
7215 /* Valid range: 32M - <arc_c_max> */
7216 if ((zfs_arc_min
) && (zfs_arc_min
!= arc_c_min
) &&
7217 (zfs_arc_min
>= 2ULL << SPA_MAXBLOCKSHIFT
) &&
7218 (zfs_arc_min
<= arc_c_max
)) {
7219 arc_c_min
= zfs_arc_min
;
7220 arc_c
= MAX(arc_c
, arc_c_min
);
7223 /* Valid range: 16M - <arc_c_max> */
7224 if ((zfs_arc_meta_min
) && (zfs_arc_meta_min
!= arc_meta_min
) &&
7225 (zfs_arc_meta_min
>= 1ULL << SPA_MAXBLOCKSHIFT
) &&
7226 (zfs_arc_meta_min
<= arc_c_max
)) {
7227 arc_meta_min
= zfs_arc_meta_min
;
7228 if (arc_meta_limit
< arc_meta_min
)
7229 arc_meta_limit
= arc_meta_min
;
7230 if (arc_dnode_limit
< arc_meta_min
)
7231 arc_dnode_limit
= arc_meta_min
;
7234 /* Valid range: <arc_meta_min> - <arc_c_max> */
7235 limit
= zfs_arc_meta_limit
? zfs_arc_meta_limit
:
7236 MIN(zfs_arc_meta_limit_percent
, 100) * arc_c_max
/ 100;
7237 if ((limit
!= arc_meta_limit
) &&
7238 (limit
>= arc_meta_min
) &&
7239 (limit
<= arc_c_max
))
7240 arc_meta_limit
= limit
;
7242 /* Valid range: <arc_meta_min> - <arc_meta_limit> */
7243 limit
= zfs_arc_dnode_limit
? zfs_arc_dnode_limit
:
7244 MIN(zfs_arc_dnode_limit_percent
, 100) * arc_meta_limit
/ 100;
7245 if ((limit
!= arc_dnode_limit
) &&
7246 (limit
>= arc_meta_min
) &&
7247 (limit
<= arc_meta_limit
))
7248 arc_dnode_limit
= limit
;
7250 /* Valid range: 1 - N */
7251 if (zfs_arc_grow_retry
)
7252 arc_grow_retry
= zfs_arc_grow_retry
;
7254 /* Valid range: 1 - N */
7255 if (zfs_arc_shrink_shift
) {
7256 arc_shrink_shift
= zfs_arc_shrink_shift
;
7257 arc_no_grow_shift
= MIN(arc_no_grow_shift
, arc_shrink_shift
-1);
7260 /* Valid range: 1 - N */
7261 if (zfs_arc_p_min_shift
)
7262 arc_p_min_shift
= zfs_arc_p_min_shift
;
7264 /* Valid range: 1 - N ms */
7265 if (zfs_arc_min_prefetch_ms
)
7266 arc_min_prefetch_ms
= zfs_arc_min_prefetch_ms
;
7268 /* Valid range: 1 - N ms */
7269 if (zfs_arc_min_prescient_prefetch_ms
) {
7270 arc_min_prescient_prefetch_ms
=
7271 zfs_arc_min_prescient_prefetch_ms
;
7274 /* Valid range: 0 - 100 */
7275 if ((zfs_arc_lotsfree_percent
>= 0) &&
7276 (zfs_arc_lotsfree_percent
<= 100))
7277 arc_lotsfree_percent
= zfs_arc_lotsfree_percent
;
7279 /* Valid range: 0 - <all physical memory> */
7280 if ((zfs_arc_sys_free
) && (zfs_arc_sys_free
!= arc_sys_free
))
7281 arc_sys_free
= MIN(MAX(zfs_arc_sys_free
, 0), allmem
);
7286 arc_state_init(void)
7288 arc_anon
= &ARC_anon
;
7290 arc_mru_ghost
= &ARC_mru_ghost
;
7292 arc_mfu_ghost
= &ARC_mfu_ghost
;
7293 arc_l2c_only
= &ARC_l2c_only
;
7295 arc_mru
->arcs_list
[ARC_BUFC_METADATA
] =
7296 multilist_create(sizeof (arc_buf_hdr_t
),
7297 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7298 arc_state_multilist_index_func
);
7299 arc_mru
->arcs_list
[ARC_BUFC_DATA
] =
7300 multilist_create(sizeof (arc_buf_hdr_t
),
7301 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7302 arc_state_multilist_index_func
);
7303 arc_mru_ghost
->arcs_list
[ARC_BUFC_METADATA
] =
7304 multilist_create(sizeof (arc_buf_hdr_t
),
7305 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7306 arc_state_multilist_index_func
);
7307 arc_mru_ghost
->arcs_list
[ARC_BUFC_DATA
] =
7308 multilist_create(sizeof (arc_buf_hdr_t
),
7309 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7310 arc_state_multilist_index_func
);
7311 arc_mfu
->arcs_list
[ARC_BUFC_METADATA
] =
7312 multilist_create(sizeof (arc_buf_hdr_t
),
7313 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7314 arc_state_multilist_index_func
);
7315 arc_mfu
->arcs_list
[ARC_BUFC_DATA
] =
7316 multilist_create(sizeof (arc_buf_hdr_t
),
7317 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7318 arc_state_multilist_index_func
);
7319 arc_mfu_ghost
->arcs_list
[ARC_BUFC_METADATA
] =
7320 multilist_create(sizeof (arc_buf_hdr_t
),
7321 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7322 arc_state_multilist_index_func
);
7323 arc_mfu_ghost
->arcs_list
[ARC_BUFC_DATA
] =
7324 multilist_create(sizeof (arc_buf_hdr_t
),
7325 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7326 arc_state_multilist_index_func
);
7327 arc_l2c_only
->arcs_list
[ARC_BUFC_METADATA
] =
7328 multilist_create(sizeof (arc_buf_hdr_t
),
7329 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7330 arc_state_multilist_index_func
);
7331 arc_l2c_only
->arcs_list
[ARC_BUFC_DATA
] =
7332 multilist_create(sizeof (arc_buf_hdr_t
),
7333 offsetof(arc_buf_hdr_t
, b_l1hdr
.b_arc_node
),
7334 arc_state_multilist_index_func
);
7336 refcount_create(&arc_anon
->arcs_esize
[ARC_BUFC_METADATA
]);
7337 refcount_create(&arc_anon
->arcs_esize
[ARC_BUFC_DATA
]);
7338 refcount_create(&arc_mru
->arcs_esize
[ARC_BUFC_METADATA
]);
7339 refcount_create(&arc_mru
->arcs_esize
[ARC_BUFC_DATA
]);
7340 refcount_create(&arc_mru_ghost
->arcs_esize
[ARC_BUFC_METADATA
]);
7341 refcount_create(&arc_mru_ghost
->arcs_esize
[ARC_BUFC_DATA
]);
7342 refcount_create(&arc_mfu
->arcs_esize
[ARC_BUFC_METADATA
]);
7343 refcount_create(&arc_mfu
->arcs_esize
[ARC_BUFC_DATA
]);
7344 refcount_create(&arc_mfu_ghost
->arcs_esize
[ARC_BUFC_METADATA
]);
7345 refcount_create(&arc_mfu_ghost
->arcs_esize
[ARC_BUFC_DATA
]);
7346 refcount_create(&arc_l2c_only
->arcs_esize
[ARC_BUFC_METADATA
]);
7347 refcount_create(&arc_l2c_only
->arcs_esize
[ARC_BUFC_DATA
]);
7349 refcount_create(&arc_anon
->arcs_size
);
7350 refcount_create(&arc_mru
->arcs_size
);
7351 refcount_create(&arc_mru_ghost
->arcs_size
);
7352 refcount_create(&arc_mfu
->arcs_size
);
7353 refcount_create(&arc_mfu_ghost
->arcs_size
);
7354 refcount_create(&arc_l2c_only
->arcs_size
);
7356 arc_anon
->arcs_state
= ARC_STATE_ANON
;
7357 arc_mru
->arcs_state
= ARC_STATE_MRU
;
7358 arc_mru_ghost
->arcs_state
= ARC_STATE_MRU_GHOST
;
7359 arc_mfu
->arcs_state
= ARC_STATE_MFU
;
7360 arc_mfu_ghost
->arcs_state
= ARC_STATE_MFU_GHOST
;
7361 arc_l2c_only
->arcs_state
= ARC_STATE_L2C_ONLY
;
7365 arc_state_fini(void)
7367 refcount_destroy(&arc_anon
->arcs_esize
[ARC_BUFC_METADATA
]);
7368 refcount_destroy(&arc_anon
->arcs_esize
[ARC_BUFC_DATA
]);
7369 refcount_destroy(&arc_mru
->arcs_esize
[ARC_BUFC_METADATA
]);
7370 refcount_destroy(&arc_mru
->arcs_esize
[ARC_BUFC_DATA
]);
7371 refcount_destroy(&arc_mru_ghost
->arcs_esize
[ARC_BUFC_METADATA
]);
7372 refcount_destroy(&arc_mru_ghost
->arcs_esize
[ARC_BUFC_DATA
]);
7373 refcount_destroy(&arc_mfu
->arcs_esize
[ARC_BUFC_METADATA
]);
7374 refcount_destroy(&arc_mfu
->arcs_esize
[ARC_BUFC_DATA
]);
7375 refcount_destroy(&arc_mfu_ghost
->arcs_esize
[ARC_BUFC_METADATA
]);
7376 refcount_destroy(&arc_mfu_ghost
->arcs_esize
[ARC_BUFC_DATA
]);
7377 refcount_destroy(&arc_l2c_only
->arcs_esize
[ARC_BUFC_METADATA
]);
7378 refcount_destroy(&arc_l2c_only
->arcs_esize
[ARC_BUFC_DATA
]);
7380 refcount_destroy(&arc_anon
->arcs_size
);
7381 refcount_destroy(&arc_mru
->arcs_size
);
7382 refcount_destroy(&arc_mru_ghost
->arcs_size
);
7383 refcount_destroy(&arc_mfu
->arcs_size
);
7384 refcount_destroy(&arc_mfu_ghost
->arcs_size
);
7385 refcount_destroy(&arc_l2c_only
->arcs_size
);
7387 multilist_destroy(arc_mru
->arcs_list
[ARC_BUFC_METADATA
]);
7388 multilist_destroy(arc_mru_ghost
->arcs_list
[ARC_BUFC_METADATA
]);
7389 multilist_destroy(arc_mfu
->arcs_list
[ARC_BUFC_METADATA
]);
7390 multilist_destroy(arc_mfu_ghost
->arcs_list
[ARC_BUFC_METADATA
]);
7391 multilist_destroy(arc_mru
->arcs_list
[ARC_BUFC_DATA
]);
7392 multilist_destroy(arc_mru_ghost
->arcs_list
[ARC_BUFC_DATA
]);
7393 multilist_destroy(arc_mfu
->arcs_list
[ARC_BUFC_DATA
]);
7394 multilist_destroy(arc_mfu_ghost
->arcs_list
[ARC_BUFC_DATA
]);
7395 multilist_destroy(arc_l2c_only
->arcs_list
[ARC_BUFC_METADATA
]);
7396 multilist_destroy(arc_l2c_only
->arcs_list
[ARC_BUFC_DATA
]);
7400 arc_target_bytes(void)
7408 uint64_t percent
, allmem
= arc_all_memory();
7410 mutex_init(&arc_reclaim_lock
, NULL
, MUTEX_DEFAULT
, NULL
);
7411 cv_init(&arc_reclaim_thread_cv
, NULL
, CV_DEFAULT
, NULL
);
7412 cv_init(&arc_reclaim_waiters_cv
, NULL
, CV_DEFAULT
, NULL
);
7414 /* Convert seconds to clock ticks */
7415 arc_min_prefetch_ms
= 1;
7416 arc_min_prescient_prefetch_ms
= 6;
7420 * Register a shrinker to support synchronous (direct) memory
7421 * reclaim from the arc. This is done to prevent kswapd from
7422 * swapping out pages when it is preferable to shrink the arc.
7424 spl_register_shrinker(&arc_shrinker
);
7426 /* Set to 1/64 of all memory or a minimum of 512K */
7427 arc_sys_free
= MAX(allmem
/ 64, (512 * 1024));
7431 /* Set max to 1/2 of all memory */
7432 arc_c_max
= allmem
/ 2;
7435 /* Set min cache to 1/32 of all memory, or 32MB, whichever is more */
7436 arc_c_min
= MAX(allmem
/ 32, 2ULL << SPA_MAXBLOCKSHIFT
);
7439 * In userland, there's only the memory pressure that we artificially
7440 * create (see arc_available_memory()). Don't let arc_c get too
7441 * small, because it can cause transactions to be larger than
7442 * arc_c, causing arc_tempreserve_space() to fail.
7444 arc_c_min
= MAX(arc_c_max
/ 2, 2ULL << SPA_MAXBLOCKSHIFT
);
7448 arc_p
= (arc_c
>> 1);
7451 /* Set min to 1/2 of arc_c_min */
7452 arc_meta_min
= 1ULL << SPA_MAXBLOCKSHIFT
;
7453 /* Initialize maximum observed usage to zero */
7456 * Set arc_meta_limit to a percent of arc_c_max with a floor of
7457 * arc_meta_min, and a ceiling of arc_c_max.
7459 percent
= MIN(zfs_arc_meta_limit_percent
, 100);
7460 arc_meta_limit
= MAX(arc_meta_min
, (percent
* arc_c_max
) / 100);
7461 percent
= MIN(zfs_arc_dnode_limit_percent
, 100);
7462 arc_dnode_limit
= (percent
* arc_meta_limit
) / 100;
7464 /* Apply user specified tunings */
7465 arc_tuning_update();
7467 /* if kmem_flags are set, lets try to use less memory */
7468 if (kmem_debugging())
7470 if (arc_c
< arc_c_min
)
7476 list_create(&arc_prune_list
, sizeof (arc_prune_t
),
7477 offsetof(arc_prune_t
, p_node
));
7478 mutex_init(&arc_prune_mtx
, NULL
, MUTEX_DEFAULT
, NULL
);
7480 arc_prune_taskq
= taskq_create("arc_prune", max_ncpus
, defclsyspri
,
7481 max_ncpus
, INT_MAX
, TASKQ_PREPOPULATE
| TASKQ_DYNAMIC
);
7483 arc_reclaim_thread_exit
= B_FALSE
;
7485 arc_ksp
= kstat_create("zfs", 0, "arcstats", "misc", KSTAT_TYPE_NAMED
,
7486 sizeof (arc_stats
) / sizeof (kstat_named_t
), KSTAT_FLAG_VIRTUAL
);
7488 if (arc_ksp
!= NULL
) {
7489 arc_ksp
->ks_data
= &arc_stats
;
7490 arc_ksp
->ks_update
= arc_kstat_update
;
7491 kstat_install(arc_ksp
);
7494 (void) thread_create(NULL
, 0, arc_reclaim_thread
, NULL
, 0, &p0
,
7495 TS_RUN
, defclsyspri
);
7501 * Calculate maximum amount of dirty data per pool.
7503 * If it has been set by a module parameter, take that.
7504 * Otherwise, use a percentage of physical memory defined by
7505 * zfs_dirty_data_max_percent (default 10%) with a cap at
7506 * zfs_dirty_data_max_max (default 4G or 25% of physical memory).
7508 if (zfs_dirty_data_max_max
== 0)
7509 zfs_dirty_data_max_max
= MIN(4ULL * 1024 * 1024 * 1024,
7510 allmem
* zfs_dirty_data_max_max_percent
/ 100);
7512 if (zfs_dirty_data_max
== 0) {
7513 zfs_dirty_data_max
= allmem
*
7514 zfs_dirty_data_max_percent
/ 100;
7515 zfs_dirty_data_max
= MIN(zfs_dirty_data_max
,
7516 zfs_dirty_data_max_max
);
7526 spl_unregister_shrinker(&arc_shrinker
);
7527 #endif /* _KERNEL */
7529 mutex_enter(&arc_reclaim_lock
);
7530 arc_reclaim_thread_exit
= B_TRUE
;
7532 * The reclaim thread will set arc_reclaim_thread_exit back to
7533 * B_FALSE when it is finished exiting; we're waiting for that.
7535 while (arc_reclaim_thread_exit
) {
7536 cv_signal(&arc_reclaim_thread_cv
);
7537 cv_wait(&arc_reclaim_thread_cv
, &arc_reclaim_lock
);
7539 mutex_exit(&arc_reclaim_lock
);
7541 /* Use B_TRUE to ensure *all* buffers are evicted */
7542 arc_flush(NULL
, B_TRUE
);
7546 if (arc_ksp
!= NULL
) {
7547 kstat_delete(arc_ksp
);
7551 taskq_wait(arc_prune_taskq
);
7552 taskq_destroy(arc_prune_taskq
);
7554 mutex_enter(&arc_prune_mtx
);
7555 while ((p
= list_head(&arc_prune_list
)) != NULL
) {
7556 list_remove(&arc_prune_list
, p
);
7557 refcount_remove(&p
->p_refcnt
, &arc_prune_list
);
7558 refcount_destroy(&p
->p_refcnt
);
7559 kmem_free(p
, sizeof (*p
));
7561 mutex_exit(&arc_prune_mtx
);
7563 list_destroy(&arc_prune_list
);
7564 mutex_destroy(&arc_prune_mtx
);
7565 mutex_destroy(&arc_reclaim_lock
);
7566 cv_destroy(&arc_reclaim_thread_cv
);
7567 cv_destroy(&arc_reclaim_waiters_cv
);
7572 ASSERT0(arc_loaned_bytes
);
7578 * The level 2 ARC (L2ARC) is a cache layer in-between main memory and disk.
7579 * It uses dedicated storage devices to hold cached data, which are populated
7580 * using large infrequent writes. The main role of this cache is to boost
7581 * the performance of random read workloads. The intended L2ARC devices
7582 * include short-stroked disks, solid state disks, and other media with
7583 * substantially faster read latency than disk.
7585 * +-----------------------+
7587 * +-----------------------+
7590 * l2arc_feed_thread() arc_read()
7594 * +---------------+ |
7596 * +---------------+ |
7601 * +-------+ +-------+
7603 * | cache | | cache |
7604 * +-------+ +-------+
7605 * +=========+ .-----.
7606 * : L2ARC : |-_____-|
7607 * : devices : | Disks |
7608 * +=========+ `-_____-'
7610 * Read requests are satisfied from the following sources, in order:
7613 * 2) vdev cache of L2ARC devices
7615 * 4) vdev cache of disks
7618 * Some L2ARC device types exhibit extremely slow write performance.
7619 * To accommodate for this there are some significant differences between
7620 * the L2ARC and traditional cache design:
7622 * 1. There is no eviction path from the ARC to the L2ARC. Evictions from
7623 * the ARC behave as usual, freeing buffers and placing headers on ghost
7624 * lists. The ARC does not send buffers to the L2ARC during eviction as
7625 * this would add inflated write latencies for all ARC memory pressure.
7627 * 2. The L2ARC attempts to cache data from the ARC before it is evicted.
7628 * It does this by periodically scanning buffers from the eviction-end of
7629 * the MFU and MRU ARC lists, copying them to the L2ARC devices if they are
7630 * not already there. It scans until a headroom of buffers is satisfied,
7631 * which itself is a buffer for ARC eviction. If a compressible buffer is
7632 * found during scanning and selected for writing to an L2ARC device, we
7633 * temporarily boost scanning headroom during the next scan cycle to make
7634 * sure we adapt to compression effects (which might significantly reduce
7635 * the data volume we write to L2ARC). The thread that does this is
7636 * l2arc_feed_thread(), illustrated below; example sizes are included to
7637 * provide a better sense of ratio than this diagram:
7640 * +---------------------+----------+
7641 * ARC_mfu |:::::#:::::::::::::::|o#o###o###|-->. # already on L2ARC
7642 * +---------------------+----------+ | o L2ARC eligible
7643 * ARC_mru |:#:::::::::::::::::::|#o#ooo####|-->| : ARC buffer
7644 * +---------------------+----------+ |
7645 * 15.9 Gbytes ^ 32 Mbytes |
7647 * l2arc_feed_thread()
7649 * l2arc write hand <--[oooo]--'
7653 * +==============================+
7654 * L2ARC dev |####|#|###|###| |####| ... |
7655 * +==============================+
7658 * 3. If an ARC buffer is copied to the L2ARC but then hit instead of
7659 * evicted, then the L2ARC has cached a buffer much sooner than it probably
7660 * needed to, potentially wasting L2ARC device bandwidth and storage. It is
7661 * safe to say that this is an uncommon case, since buffers at the end of
7662 * the ARC lists have moved there due to inactivity.
7664 * 4. If the ARC evicts faster than the L2ARC can maintain a headroom,
7665 * then the L2ARC simply misses copying some buffers. This serves as a
7666 * pressure valve to prevent heavy read workloads from both stalling the ARC
7667 * with waits and clogging the L2ARC with writes. This also helps prevent
7668 * the potential for the L2ARC to churn if it attempts to cache content too
7669 * quickly, such as during backups of the entire pool.
7671 * 5. After system boot and before the ARC has filled main memory, there are
7672 * no evictions from the ARC and so the tails of the ARC_mfu and ARC_mru
7673 * lists can remain mostly static. Instead of searching from tail of these
7674 * lists as pictured, the l2arc_feed_thread() will search from the list heads
7675 * for eligible buffers, greatly increasing its chance of finding them.
7677 * The L2ARC device write speed is also boosted during this time so that
7678 * the L2ARC warms up faster. Since there have been no ARC evictions yet,
7679 * there are no L2ARC reads, and no fear of degrading read performance
7680 * through increased writes.
7682 * 6. Writes to the L2ARC devices are grouped and sent in-sequence, so that
7683 * the vdev queue can aggregate them into larger and fewer writes. Each
7684 * device is written to in a rotor fashion, sweeping writes through
7685 * available space then repeating.
7687 * 7. The L2ARC does not store dirty content. It never needs to flush
7688 * write buffers back to disk based storage.
7690 * 8. If an ARC buffer is written (and dirtied) which also exists in the
7691 * L2ARC, the now stale L2ARC buffer is immediately dropped.
7693 * The performance of the L2ARC can be tweaked by a number of tunables, which
7694 * may be necessary for different workloads:
7696 * l2arc_write_max max write bytes per interval
7697 * l2arc_write_boost extra write bytes during device warmup
7698 * l2arc_noprefetch skip caching prefetched buffers
7699 * l2arc_headroom number of max device writes to precache
7700 * l2arc_headroom_boost when we find compressed buffers during ARC
7701 * scanning, we multiply headroom by this
7702 * percentage factor for the next scan cycle,
7703 * since more compressed buffers are likely to
7705 * l2arc_feed_secs seconds between L2ARC writing
7707 * Tunables may be removed or added as future performance improvements are
7708 * integrated, and also may become zpool properties.
7710 * There are three key functions that control how the L2ARC warms up:
7712 * l2arc_write_eligible() check if a buffer is eligible to cache
7713 * l2arc_write_size() calculate how much to write
7714 * l2arc_write_interval() calculate sleep delay between writes
7716 * These three functions determine what to write, how much, and how quickly
7721 l2arc_write_eligible(uint64_t spa_guid
, arc_buf_hdr_t
*hdr
)
7724 * A buffer is *not* eligible for the L2ARC if it:
7725 * 1. belongs to a different spa.
7726 * 2. is already cached on the L2ARC.
7727 * 3. has an I/O in progress (it may be an incomplete read).
7728 * 4. is flagged not eligible (zfs property).
7730 if (hdr
->b_spa
!= spa_guid
|| HDR_HAS_L2HDR(hdr
) ||
7731 HDR_IO_IN_PROGRESS(hdr
) || !HDR_L2CACHE(hdr
))
7738 l2arc_write_size(void)
7743 * Make sure our globals have meaningful values in case the user
7746 size
= l2arc_write_max
;
7748 cmn_err(CE_NOTE
, "Bad value for l2arc_write_max, value must "
7749 "be greater than zero, resetting it to the default (%d)",
7751 size
= l2arc_write_max
= L2ARC_WRITE_SIZE
;
7754 if (arc_warm
== B_FALSE
)
7755 size
+= l2arc_write_boost
;
7762 l2arc_write_interval(clock_t began
, uint64_t wanted
, uint64_t wrote
)
7764 clock_t interval
, next
, now
;
7767 * If the ARC lists are busy, increase our write rate; if the
7768 * lists are stale, idle back. This is achieved by checking
7769 * how much we previously wrote - if it was more than half of
7770 * what we wanted, schedule the next write much sooner.
7772 if (l2arc_feed_again
&& wrote
> (wanted
/ 2))
7773 interval
= (hz
* l2arc_feed_min_ms
) / 1000;
7775 interval
= hz
* l2arc_feed_secs
;
7777 now
= ddi_get_lbolt();
7778 next
= MAX(now
, MIN(now
+ interval
, began
+ interval
));
7784 * Cycle through L2ARC devices. This is how L2ARC load balances.
7785 * If a device is returned, this also returns holding the spa config lock.
7787 static l2arc_dev_t
*
7788 l2arc_dev_get_next(void)
7790 l2arc_dev_t
*first
, *next
= NULL
;
7793 * Lock out the removal of spas (spa_namespace_lock), then removal
7794 * of cache devices (l2arc_dev_mtx). Once a device has been selected,
7795 * both locks will be dropped and a spa config lock held instead.
7797 mutex_enter(&spa_namespace_lock
);
7798 mutex_enter(&l2arc_dev_mtx
);
7800 /* if there are no vdevs, there is nothing to do */
7801 if (l2arc_ndev
== 0)
7805 next
= l2arc_dev_last
;
7807 /* loop around the list looking for a non-faulted vdev */
7809 next
= list_head(l2arc_dev_list
);
7811 next
= list_next(l2arc_dev_list
, next
);
7813 next
= list_head(l2arc_dev_list
);
7816 /* if we have come back to the start, bail out */
7819 else if (next
== first
)
7822 } while (vdev_is_dead(next
->l2ad_vdev
));
7824 /* if we were unable to find any usable vdevs, return NULL */
7825 if (vdev_is_dead(next
->l2ad_vdev
))
7828 l2arc_dev_last
= next
;
7831 mutex_exit(&l2arc_dev_mtx
);
7834 * Grab the config lock to prevent the 'next' device from being
7835 * removed while we are writing to it.
7838 spa_config_enter(next
->l2ad_spa
, SCL_L2ARC
, next
, RW_READER
);
7839 mutex_exit(&spa_namespace_lock
);
7845 * Free buffers that were tagged for destruction.
7848 l2arc_do_free_on_write(void)
7851 l2arc_data_free_t
*df
, *df_prev
;
7853 mutex_enter(&l2arc_free_on_write_mtx
);
7854 buflist
= l2arc_free_on_write
;
7856 for (df
= list_tail(buflist
); df
; df
= df_prev
) {
7857 df_prev
= list_prev(buflist
, df
);
7858 ASSERT3P(df
->l2df_abd
, !=, NULL
);
7859 abd_free(df
->l2df_abd
);
7860 list_remove(buflist
, df
);
7861 kmem_free(df
, sizeof (l2arc_data_free_t
));
7864 mutex_exit(&l2arc_free_on_write_mtx
);
7868 * A write to a cache device has completed. Update all headers to allow
7869 * reads from these buffers to begin.
7872 l2arc_write_done(zio_t
*zio
)
7874 l2arc_write_callback_t
*cb
;
7877 arc_buf_hdr_t
*head
, *hdr
, *hdr_prev
;
7878 kmutex_t
*hash_lock
;
7879 int64_t bytes_dropped
= 0;
7881 cb
= zio
->io_private
;
7882 ASSERT3P(cb
, !=, NULL
);
7883 dev
= cb
->l2wcb_dev
;
7884 ASSERT3P(dev
, !=, NULL
);
7885 head
= cb
->l2wcb_head
;
7886 ASSERT3P(head
, !=, NULL
);
7887 buflist
= &dev
->l2ad_buflist
;
7888 ASSERT3P(buflist
, !=, NULL
);
7889 DTRACE_PROBE2(l2arc__iodone
, zio_t
*, zio
,
7890 l2arc_write_callback_t
*, cb
);
7892 if (zio
->io_error
!= 0)
7893 ARCSTAT_BUMP(arcstat_l2_writes_error
);
7896 * All writes completed, or an error was hit.
7899 mutex_enter(&dev
->l2ad_mtx
);
7900 for (hdr
= list_prev(buflist
, head
); hdr
; hdr
= hdr_prev
) {
7901 hdr_prev
= list_prev(buflist
, hdr
);
7903 hash_lock
= HDR_LOCK(hdr
);
7906 * We cannot use mutex_enter or else we can deadlock
7907 * with l2arc_write_buffers (due to swapping the order
7908 * the hash lock and l2ad_mtx are taken).
7910 if (!mutex_tryenter(hash_lock
)) {
7912 * Missed the hash lock. We must retry so we
7913 * don't leave the ARC_FLAG_L2_WRITING bit set.
7915 ARCSTAT_BUMP(arcstat_l2_writes_lock_retry
);
7918 * We don't want to rescan the headers we've
7919 * already marked as having been written out, so
7920 * we reinsert the head node so we can pick up
7921 * where we left off.
7923 list_remove(buflist
, head
);
7924 list_insert_after(buflist
, hdr
, head
);
7926 mutex_exit(&dev
->l2ad_mtx
);
7929 * We wait for the hash lock to become available
7930 * to try and prevent busy waiting, and increase
7931 * the chance we'll be able to acquire the lock
7932 * the next time around.
7934 mutex_enter(hash_lock
);
7935 mutex_exit(hash_lock
);
7940 * We could not have been moved into the arc_l2c_only
7941 * state while in-flight due to our ARC_FLAG_L2_WRITING
7942 * bit being set. Let's just ensure that's being enforced.
7944 ASSERT(HDR_HAS_L1HDR(hdr
));
7947 * Skipped - drop L2ARC entry and mark the header as no
7948 * longer L2 eligibile.
7950 if (zio
->io_error
!= 0) {
7952 * Error - drop L2ARC entry.
7954 list_remove(buflist
, hdr
);
7955 arc_hdr_clear_flags(hdr
, ARC_FLAG_HAS_L2HDR
);
7957 ARCSTAT_INCR(arcstat_l2_psize
, -arc_hdr_size(hdr
));
7958 ARCSTAT_INCR(arcstat_l2_lsize
, -HDR_GET_LSIZE(hdr
));
7960 bytes_dropped
+= arc_hdr_size(hdr
);
7961 (void) refcount_remove_many(&dev
->l2ad_alloc
,
7962 arc_hdr_size(hdr
), hdr
);
7966 * Allow ARC to begin reads and ghost list evictions to
7969 arc_hdr_clear_flags(hdr
, ARC_FLAG_L2_WRITING
);
7971 mutex_exit(hash_lock
);
7974 atomic_inc_64(&l2arc_writes_done
);
7975 list_remove(buflist
, head
);
7976 ASSERT(!HDR_HAS_L1HDR(head
));
7977 kmem_cache_free(hdr_l2only_cache
, head
);
7978 mutex_exit(&dev
->l2ad_mtx
);
7980 vdev_space_update(dev
->l2ad_vdev
, -bytes_dropped
, 0, 0);
7982 l2arc_do_free_on_write();
7984 kmem_free(cb
, sizeof (l2arc_write_callback_t
));
7988 l2arc_untransform(zio_t
*zio
, l2arc_read_callback_t
*cb
)
7991 spa_t
*spa
= zio
->io_spa
;
7992 arc_buf_hdr_t
*hdr
= cb
->l2rcb_hdr
;
7993 blkptr_t
*bp
= zio
->io_bp
;
7994 dsl_crypto_key_t
*dck
= NULL
;
7995 uint8_t salt
[ZIO_DATA_SALT_LEN
];
7996 uint8_t iv
[ZIO_DATA_IV_LEN
];
7997 uint8_t mac
[ZIO_DATA_MAC_LEN
];
7998 boolean_t no_crypt
= B_FALSE
;
8001 * ZIL data is never be written to the L2ARC, so we don't need
8002 * special handling for its unique MAC storage.
8004 ASSERT3U(BP_GET_TYPE(bp
), !=, DMU_OT_INTENT_LOG
);
8005 ASSERT(MUTEX_HELD(HDR_LOCK(hdr
)));
8006 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, !=, NULL
);
8009 * If the data was encrypted, decrypt it now. Note that
8010 * we must check the bp here and not the hdr, since the
8011 * hdr does not have its encryption parameters updated
8012 * until arc_read_done().
8014 if (BP_IS_ENCRYPTED(bp
)) {
8015 abd_t
*eabd
= arc_get_data_abd(hdr
,
8016 arc_hdr_size(hdr
), hdr
);
8018 zio_crypt_decode_params_bp(bp
, salt
, iv
);
8019 zio_crypt_decode_mac_bp(bp
, mac
);
8021 ret
= spa_keystore_lookup_key(spa
,
8022 cb
->l2rcb_zb
.zb_objset
, FTAG
, &dck
);
8024 arc_free_data_abd(hdr
, eabd
, arc_hdr_size(hdr
), hdr
);
8028 ret
= zio_do_crypt_abd(B_FALSE
, &dck
->dck_key
,
8029 salt
, BP_GET_TYPE(bp
), iv
, mac
, HDR_GET_PSIZE(hdr
),
8030 BP_SHOULD_BYTESWAP(bp
), eabd
, hdr
->b_l1hdr
.b_pabd
,
8033 arc_free_data_abd(hdr
, eabd
, arc_hdr_size(hdr
), hdr
);
8034 spa_keystore_dsl_key_rele(spa
, dck
, FTAG
);
8038 spa_keystore_dsl_key_rele(spa
, dck
, FTAG
);
8041 * If we actually performed decryption, replace b_pabd
8042 * with the decrypted data. Otherwise we can just throw
8043 * our decryption buffer away.
8046 arc_free_data_abd(hdr
, hdr
->b_l1hdr
.b_pabd
,
8047 arc_hdr_size(hdr
), hdr
);
8048 hdr
->b_l1hdr
.b_pabd
= eabd
;
8051 arc_free_data_abd(hdr
, eabd
, arc_hdr_size(hdr
), hdr
);
8056 * If the L2ARC block was compressed, but ARC compression
8057 * is disabled we decompress the data into a new buffer and
8058 * replace the existing data.
8060 if (HDR_GET_COMPRESS(hdr
) != ZIO_COMPRESS_OFF
&&
8061 !HDR_COMPRESSION_ENABLED(hdr
)) {
8062 abd_t
*cabd
= arc_get_data_abd(hdr
, arc_hdr_size(hdr
), hdr
);
8063 void *tmp
= abd_borrow_buf(cabd
, arc_hdr_size(hdr
));
8065 ret
= zio_decompress_data(HDR_GET_COMPRESS(hdr
),
8066 hdr
->b_l1hdr
.b_pabd
, tmp
, HDR_GET_PSIZE(hdr
),
8067 HDR_GET_LSIZE(hdr
));
8069 abd_return_buf_copy(cabd
, tmp
, arc_hdr_size(hdr
));
8070 arc_free_data_abd(hdr
, cabd
, arc_hdr_size(hdr
), hdr
);
8074 abd_return_buf_copy(cabd
, tmp
, arc_hdr_size(hdr
));
8075 arc_free_data_abd(hdr
, hdr
->b_l1hdr
.b_pabd
,
8076 arc_hdr_size(hdr
), hdr
);
8077 hdr
->b_l1hdr
.b_pabd
= cabd
;
8079 zio
->io_size
= HDR_GET_LSIZE(hdr
);
8090 * A read to a cache device completed. Validate buffer contents before
8091 * handing over to the regular ARC routines.
8094 l2arc_read_done(zio_t
*zio
)
8097 l2arc_read_callback_t
*cb
;
8099 kmutex_t
*hash_lock
;
8100 boolean_t valid_cksum
, using_rdata
;
8102 ASSERT3P(zio
->io_vd
, !=, NULL
);
8103 ASSERT(zio
->io_flags
& ZIO_FLAG_DONT_PROPAGATE
);
8105 spa_config_exit(zio
->io_spa
, SCL_L2ARC
, zio
->io_vd
);
8107 cb
= zio
->io_private
;
8108 ASSERT3P(cb
, !=, NULL
);
8109 hdr
= cb
->l2rcb_hdr
;
8110 ASSERT3P(hdr
, !=, NULL
);
8112 hash_lock
= HDR_LOCK(hdr
);
8113 mutex_enter(hash_lock
);
8114 ASSERT3P(hash_lock
, ==, HDR_LOCK(hdr
));
8117 * If the data was read into a temporary buffer,
8118 * move it and free the buffer.
8120 if (cb
->l2rcb_abd
!= NULL
) {
8121 ASSERT3U(arc_hdr_size(hdr
), <, zio
->io_size
);
8122 if (zio
->io_error
== 0) {
8123 abd_copy(hdr
->b_l1hdr
.b_pabd
, cb
->l2rcb_abd
,
8128 * The following must be done regardless of whether
8129 * there was an error:
8130 * - free the temporary buffer
8131 * - point zio to the real ARC buffer
8132 * - set zio size accordingly
8133 * These are required because zio is either re-used for
8134 * an I/O of the block in the case of the error
8135 * or the zio is passed to arc_read_done() and it
8138 abd_free(cb
->l2rcb_abd
);
8139 zio
->io_size
= zio
->io_orig_size
= arc_hdr_size(hdr
);
8141 if (BP_IS_ENCRYPTED(&cb
->l2rcb_bp
) &&
8142 (cb
->l2rcb_flags
& ZIO_FLAG_RAW_ENCRYPT
)) {
8143 ASSERT(HDR_HAS_RABD(hdr
));
8144 zio
->io_abd
= zio
->io_orig_abd
=
8145 hdr
->b_crypt_hdr
.b_rabd
;
8147 ASSERT3P(hdr
->b_l1hdr
.b_pabd
, !=, NULL
);
8148 zio
->io_abd
= zio
->io_orig_abd
= hdr
->b_l1hdr
.b_pabd
;
8152 ASSERT3P(zio
->io_abd
, !=, NULL
);
8155 * Check this survived the L2ARC journey.
8157 ASSERT(zio
->io_abd
== hdr
->b_l1hdr
.b_pabd
||
8158 (HDR_HAS_RABD(hdr
) && zio
->io_abd
== hdr
->b_crypt_hdr
.b_rabd
));
8159 zio
->io_bp_copy
= cb
->l2rcb_bp
; /* XXX fix in L2ARC 2.0 */
8160 zio
->io_bp
= &zio
->io_bp_copy
; /* XXX fix in L2ARC 2.0 */
8162 valid_cksum
= arc_cksum_is_equal(hdr
, zio
);
8163 using_rdata
= (HDR_HAS_RABD(hdr
) &&
8164 zio
->io_abd
== hdr
->b_crypt_hdr
.b_rabd
);
8167 * b_rabd will always match the data as it exists on disk if it is
8168 * being used. Therefore if we are reading into b_rabd we do not
8169 * attempt to untransform the data.
8171 if (valid_cksum
&& !using_rdata
)
8172 tfm_error
= l2arc_untransform(zio
, cb
);
8174 if (valid_cksum
&& tfm_error
== 0 && zio
->io_error
== 0 &&
8175 !HDR_L2_EVICTED(hdr
)) {
8176 mutex_exit(hash_lock
);
8177 zio
->io_private
= hdr
;
8180 mutex_exit(hash_lock
);
8182 * Buffer didn't survive caching. Increment stats and
8183 * reissue to the original storage device.
8185 if (zio
->io_error
!= 0) {
8186 ARCSTAT_BUMP(arcstat_l2_io_error
);
8188 zio
->io_error
= SET_ERROR(EIO
);
8190 if (!valid_cksum
|| tfm_error
!= 0)
8191 ARCSTAT_BUMP(arcstat_l2_cksum_bad
);
8194 * If there's no waiter, issue an async i/o to the primary
8195 * storage now. If there *is* a waiter, the caller must
8196 * issue the i/o in a context where it's OK to block.
8198 if (zio
->io_waiter
== NULL
) {
8199 zio_t
*pio
= zio_unique_parent(zio
);
8200 void *abd
= (using_rdata
) ?
8201 hdr
->b_crypt_hdr
.b_rabd
: hdr
->b_l1hdr
.b_pabd
;
8203 ASSERT(!pio
|| pio
->io_child_type
== ZIO_CHILD_LOGICAL
);
8205 zio_nowait(zio_read(pio
, zio
->io_spa
, zio
->io_bp
,
8206 abd
, zio
->io_size
, arc_read_done
,
8207 hdr
, zio
->io_priority
, cb
->l2rcb_flags
,
8212 kmem_free(cb
, sizeof (l2arc_read_callback_t
));
8216 * This is the list priority from which the L2ARC will search for pages to
8217 * cache. This is used within loops (0..3) to cycle through lists in the
8218 * desired order. This order can have a significant effect on cache
8221 * Currently the metadata lists are hit first, MFU then MRU, followed by
8222 * the data lists. This function returns a locked list, and also returns
8225 static multilist_sublist_t
*
8226 l2arc_sublist_lock(int list_num
)
8228 multilist_t
*ml
= NULL
;
8231 ASSERT(list_num
>= 0 && list_num
< L2ARC_FEED_TYPES
);
8235 ml
= arc_mfu
->arcs_list
[ARC_BUFC_METADATA
];
8238 ml
= arc_mru
->arcs_list
[ARC_BUFC_METADATA
];
8241 ml
= arc_mfu
->arcs_list
[ARC_BUFC_DATA
];
8244 ml
= arc_mru
->arcs_list
[ARC_BUFC_DATA
];
8251 * Return a randomly-selected sublist. This is acceptable
8252 * because the caller feeds only a little bit of data for each
8253 * call (8MB). Subsequent calls will result in different
8254 * sublists being selected.
8256 idx
= multilist_get_random_index(ml
);
8257 return (multilist_sublist_lock(ml
, idx
));
8261 * Evict buffers from the device write hand to the distance specified in
8262 * bytes. This distance may span populated buffers, it may span nothing.
8263 * This is clearing a region on the L2ARC device ready for writing.
8264 * If the 'all' boolean is set, every buffer is evicted.
8267 l2arc_evict(l2arc_dev_t
*dev
, uint64_t distance
, boolean_t all
)
8270 arc_buf_hdr_t
*hdr
, *hdr_prev
;
8271 kmutex_t
*hash_lock
;
8274 buflist
= &dev
->l2ad_buflist
;
8276 if (!all
&& dev
->l2ad_first
) {
8278 * This is the first sweep through the device. There is
8284 if (dev
->l2ad_hand
>= (dev
->l2ad_end
- (2 * distance
))) {
8286 * When nearing the end of the device, evict to the end
8287 * before the device write hand jumps to the start.
8289 taddr
= dev
->l2ad_end
;
8291 taddr
= dev
->l2ad_hand
+ distance
;
8293 DTRACE_PROBE4(l2arc__evict
, l2arc_dev_t
*, dev
, list_t
*, buflist
,
8294 uint64_t, taddr
, boolean_t
, all
);
8297 mutex_enter(&dev
->l2ad_mtx
);
8298 for (hdr
= list_tail(buflist
); hdr
; hdr
= hdr_prev
) {
8299 hdr_prev
= list_prev(buflist
, hdr
);
8301 hash_lock
= HDR_LOCK(hdr
);
8304 * We cannot use mutex_enter or else we can deadlock
8305 * with l2arc_write_buffers (due to swapping the order
8306 * the hash lock and l2ad_mtx are taken).
8308 if (!mutex_tryenter(hash_lock
)) {
8310 * Missed the hash lock. Retry.
8312 ARCSTAT_BUMP(arcstat_l2_evict_lock_retry
);
8313 mutex_exit(&dev
->l2ad_mtx
);
8314 mutex_enter(hash_lock
);
8315 mutex_exit(hash_lock
);
8320 * A header can't be on this list if it doesn't have L2 header.
8322 ASSERT(HDR_HAS_L2HDR(hdr
));
8324 /* Ensure this header has finished being written. */
8325 ASSERT(!HDR_L2_WRITING(hdr
));
8326 ASSERT(!HDR_L2_WRITE_HEAD(hdr
));
8328 if (!all
&& (hdr
->b_l2hdr
.b_daddr
>= taddr
||
8329 hdr
->b_l2hdr
.b_daddr
< dev
->l2ad_hand
)) {
8331 * We've evicted to the target address,
8332 * or the end of the device.
8334 mutex_exit(hash_lock
);
8338 if (!HDR_HAS_L1HDR(hdr
)) {
8339 ASSERT(!HDR_L2_READING(hdr
));
8341 * This doesn't exist in the ARC. Destroy.
8342 * arc_hdr_destroy() will call list_remove()
8343 * and decrement arcstat_l2_lsize.
8345 arc_change_state(arc_anon
, hdr
, hash_lock
);
8346 arc_hdr_destroy(hdr
);
8348 ASSERT(hdr
->b_l1hdr
.b_state
!= arc_l2c_only
);
8349 ARCSTAT_BUMP(arcstat_l2_evict_l1cached
);
8351 * Invalidate issued or about to be issued
8352 * reads, since we may be about to write
8353 * over this location.
8355 if (HDR_L2_READING(hdr
)) {
8356 ARCSTAT_BUMP(arcstat_l2_evict_reading
);
8357 arc_hdr_set_flags(hdr
, ARC_FLAG_L2_EVICTED
);
8360 arc_hdr_l2hdr_destroy(hdr
);
8362 mutex_exit(hash_lock
);
8364 mutex_exit(&dev
->l2ad_mtx
);
8368 * Handle any abd transforms that might be required for writing to the L2ARC.
8369 * If successful, this function will always return an abd with the data
8370 * transformed as it is on disk in a new abd of asize bytes.
8373 l2arc_apply_transforms(spa_t
*spa
, arc_buf_hdr_t
*hdr
, uint64_t asize
,
8378 abd_t
*cabd
= NULL
, *eabd
= NULL
, *to_write
= hdr
->b_l1hdr
.b_pabd
;
8379 enum zio_compress compress
= HDR_GET_COMPRESS(hdr
);
8380 uint64_t psize
= HDR_GET_PSIZE(hdr
);
8381 uint64_t size
= arc_hdr_size(hdr
);
8382 boolean_t ismd
= HDR_ISTYPE_METADATA(hdr
);
8383 boolean_t bswap
= (hdr
->b_l1hdr
.b_byteswap
!= DMU_BSWAP_NUMFUNCS
);
8384 dsl_crypto_key_t
*dck
= NULL
;
8385 uint8_t mac
[ZIO_DATA_MAC_LEN
] = { 0 };
8386 boolean_t no_crypt
= B_FALSE
;
8388 ASSERT((HDR_GET_COMPRESS(hdr
) != ZIO_COMPRESS_OFF
&&
8389 !HDR_COMPRESSION_ENABLED(hdr
)) ||
8390 HDR_ENCRYPTED(hdr
) || HDR_SHARED_DATA(hdr
) || psize
!= asize
);
8391 ASSERT3U(psize
, <=, asize
);
8394 * If this data simply needs its own buffer, we simply allocate it
8395 * and copy the data. This may be done to elimiate a depedency on a
8396 * shared buffer or to reallocate the buffer to match asize.
8398 if (HDR_HAS_RABD(hdr
) && asize
!= psize
) {
8399 ASSERT3U(size
, ==, psize
);
8400 to_write
= abd_alloc_for_io(asize
, ismd
);
8401 abd_copy(to_write
, hdr
->b_crypt_hdr
.b_rabd
, size
);
8403 abd_zero_off(to_write
, size
, asize
- size
);
8407 if ((compress
== ZIO_COMPRESS_OFF
|| HDR_COMPRESSION_ENABLED(hdr
)) &&
8408 !HDR_ENCRYPTED(hdr
)) {
8409 ASSERT3U(size
, ==, psize
);
8410 to_write
= abd_alloc_for_io(asize
, ismd
);
8411 abd_copy(to_write
, hdr
->b_l1hdr
.b_pabd
, size
);
8413 abd_zero_off(to_write
, size
, asize
- size
);
8417 if (compress
!= ZIO_COMPRESS_OFF
&& !HDR_COMPRESSION_ENABLED(hdr
)) {
8418 cabd
= abd_alloc_for_io(asize
, ismd
);
8419 tmp
= abd_borrow_buf(cabd
, asize
);
8421 psize
= zio_compress_data(compress
, to_write
, tmp
, size
);
8422 ASSERT3U(psize
, <=, HDR_GET_PSIZE(hdr
));
8424 bzero((char *)tmp
+ psize
, asize
- psize
);
8425 psize
= HDR_GET_PSIZE(hdr
);
8426 abd_return_buf_copy(cabd
, tmp
, asize
);
8430 if (HDR_ENCRYPTED(hdr
)) {
8431 eabd
= abd_alloc_for_io(asize
, ismd
);
8434 * If the dataset was disowned before the buffer
8435 * made it to this point, the key to re-encrypt
8436 * it won't be available. In this case we simply
8437 * won't write the buffer to the L2ARC.
8439 ret
= spa_keystore_lookup_key(spa
, hdr
->b_crypt_hdr
.b_dsobj
,
8444 ret
= zio_do_crypt_abd(B_TRUE
, &dck
->dck_key
,
8445 hdr
->b_crypt_hdr
.b_salt
, hdr
->b_crypt_hdr
.b_ot
,
8446 hdr
->b_crypt_hdr
.b_iv
, mac
, psize
, bswap
, to_write
,
8452 abd_copy(eabd
, to_write
, psize
);
8455 abd_zero_off(eabd
, psize
, asize
- psize
);
8457 /* assert that the MAC we got here matches the one we saved */
8458 ASSERT0(bcmp(mac
, hdr
->b_crypt_hdr
.b_mac
, ZIO_DATA_MAC_LEN
));
8459 spa_keystore_dsl_key_rele(spa
, dck
, FTAG
);
8461 if (to_write
== cabd
)
8468 ASSERT3P(to_write
, !=, hdr
->b_l1hdr
.b_pabd
);
8469 *abd_out
= to_write
;
8474 spa_keystore_dsl_key_rele(spa
, dck
, FTAG
);
8485 * Find and write ARC buffers to the L2ARC device.
8487 * An ARC_FLAG_L2_WRITING flag is set so that the L2ARC buffers are not valid
8488 * for reading until they have completed writing.
8489 * The headroom_boost is an in-out parameter used to maintain headroom boost
8490 * state between calls to this function.
8492 * Returns the number of bytes actually written (which may be smaller than
8493 * the delta by which the device hand has changed due to alignment).
8496 l2arc_write_buffers(spa_t
*spa
, l2arc_dev_t
*dev
, uint64_t target_sz
)
8498 arc_buf_hdr_t
*hdr
, *hdr_prev
, *head
;
8499 uint64_t write_asize
, write_psize
, write_lsize
, headroom
;
8501 l2arc_write_callback_t
*cb
;
8503 uint64_t guid
= spa_load_guid(spa
);
8505 ASSERT3P(dev
->l2ad_vdev
, !=, NULL
);
8508 write_lsize
= write_asize
= write_psize
= 0;
8510 head
= kmem_cache_alloc(hdr_l2only_cache
, KM_PUSHPAGE
);
8511 arc_hdr_set_flags(head
, ARC_FLAG_L2_WRITE_HEAD
| ARC_FLAG_HAS_L2HDR
);
8514 * Copy buffers for L2ARC writing.
8516 for (int try = 0; try < L2ARC_FEED_TYPES
; try++) {
8517 multilist_sublist_t
*mls
= l2arc_sublist_lock(try);
8518 uint64_t passed_sz
= 0;
8520 VERIFY3P(mls
, !=, NULL
);
8523 * L2ARC fast warmup.
8525 * Until the ARC is warm and starts to evict, read from the
8526 * head of the ARC lists rather than the tail.
8528 if (arc_warm
== B_FALSE
)
8529 hdr
= multilist_sublist_head(mls
);
8531 hdr
= multilist_sublist_tail(mls
);
8533 headroom
= target_sz
* l2arc_headroom
;
8534 if (zfs_compressed_arc_enabled
)
8535 headroom
= (headroom
* l2arc_headroom_boost
) / 100;
8537 for (; hdr
; hdr
= hdr_prev
) {
8538 kmutex_t
*hash_lock
;
8539 abd_t
*to_write
= NULL
;
8541 if (arc_warm
== B_FALSE
)
8542 hdr_prev
= multilist_sublist_next(mls
, hdr
);
8544 hdr_prev
= multilist_sublist_prev(mls
, hdr
);
8546 hash_lock
= HDR_LOCK(hdr
);
8547 if (!mutex_tryenter(hash_lock
)) {
8549 * Skip this buffer rather than waiting.
8554 passed_sz
+= HDR_GET_LSIZE(hdr
);
8555 if (passed_sz
> headroom
) {
8559 mutex_exit(hash_lock
);
8563 if (!l2arc_write_eligible(guid
, hdr
)) {
8564 mutex_exit(hash_lock
);
8569 * We rely on the L1 portion of the header below, so
8570 * it's invalid for this header to have been evicted out
8571 * of the ghost cache, prior to being written out. The
8572 * ARC_FLAG_L2_WRITING bit ensures this won't happen.
8574 ASSERT(HDR_HAS_L1HDR(hdr
));
8576 ASSERT3U(HDR_GET_PSIZE(hdr
), >, 0);
8577 ASSERT3U(arc_hdr_size(hdr
), >, 0);
8578 ASSERT(hdr
->b_l1hdr
.b_pabd
!= NULL
||
8580 uint64_t psize
= HDR_GET_PSIZE(hdr
);
8581 uint64_t asize
= vdev_psize_to_asize(dev
->l2ad_vdev
,
8584 if ((write_asize
+ asize
) > target_sz
) {
8586 mutex_exit(hash_lock
);
8591 * We rely on the L1 portion of the header below, so
8592 * it's invalid for this header to have been evicted out
8593 * of the ghost cache, prior to being written out. The
8594 * ARC_FLAG_L2_WRITING bit ensures this won't happen.
8596 arc_hdr_set_flags(hdr
, ARC_FLAG_L2_WRITING
);
8597 ASSERT(HDR_HAS_L1HDR(hdr
));
8599 ASSERT3U(HDR_GET_PSIZE(hdr
), >, 0);
8600 ASSERT(hdr
->b_l1hdr
.b_pabd
!= NULL
||
8602 ASSERT3U(arc_hdr_size(hdr
), >, 0);
8605 * If this header has b_rabd, we can use this since it
8606 * must always match the data exactly as it exists on
8607 * disk. Otherwise, the L2ARC can normally use the
8608 * hdr's data, but if we're sharing data between the
8609 * hdr and one of its bufs, L2ARC needs its own copy of
8610 * the data so that the ZIO below can't race with the
8611 * buf consumer. To ensure that this copy will be
8612 * available for the lifetime of the ZIO and be cleaned
8613 * up afterwards, we add it to the l2arc_free_on_write
8614 * queue. If we need to apply any transforms to the
8615 * data (compression, encryption) we will also need the
8618 if (HDR_HAS_RABD(hdr
) && psize
== asize
) {
8619 to_write
= hdr
->b_crypt_hdr
.b_rabd
;
8620 } else if ((HDR_COMPRESSION_ENABLED(hdr
) ||
8621 HDR_GET_COMPRESS(hdr
) == ZIO_COMPRESS_OFF
) &&
8622 !HDR_ENCRYPTED(hdr
) && !HDR_SHARED_DATA(hdr
) &&
8624 to_write
= hdr
->b_l1hdr
.b_pabd
;
8627 arc_buf_contents_t type
= arc_buf_type(hdr
);
8629 ret
= l2arc_apply_transforms(spa
, hdr
, asize
,
8632 arc_hdr_clear_flags(hdr
,
8633 ARC_FLAG_L2_WRITING
);
8634 mutex_exit(hash_lock
);
8638 l2arc_free_abd_on_write(to_write
, asize
, type
);
8643 * Insert a dummy header on the buflist so
8644 * l2arc_write_done() can find where the
8645 * write buffers begin without searching.
8647 mutex_enter(&dev
->l2ad_mtx
);
8648 list_insert_head(&dev
->l2ad_buflist
, head
);
8649 mutex_exit(&dev
->l2ad_mtx
);
8652 sizeof (l2arc_write_callback_t
), KM_SLEEP
);
8653 cb
->l2wcb_dev
= dev
;
8654 cb
->l2wcb_head
= head
;
8655 pio
= zio_root(spa
, l2arc_write_done
, cb
,
8659 hdr
->b_l2hdr
.b_dev
= dev
;
8660 hdr
->b_l2hdr
.b_hits
= 0;
8662 hdr
->b_l2hdr
.b_daddr
= dev
->l2ad_hand
;
8663 arc_hdr_set_flags(hdr
, ARC_FLAG_HAS_L2HDR
);
8665 mutex_enter(&dev
->l2ad_mtx
);
8666 list_insert_head(&dev
->l2ad_buflist
, hdr
);
8667 mutex_exit(&dev
->l2ad_mtx
);
8669 (void) refcount_add_many(&dev
->l2ad_alloc
,
8670 arc_hdr_size(hdr
), hdr
);
8672 wzio
= zio_write_phys(pio
, dev
->l2ad_vdev
,
8673 hdr
->b_l2hdr
.b_daddr
, asize
, to_write
,
8674 ZIO_CHECKSUM_OFF
, NULL
, hdr
,
8675 ZIO_PRIORITY_ASYNC_WRITE
,
8676 ZIO_FLAG_CANFAIL
, B_FALSE
);
8678 write_lsize
+= HDR_GET_LSIZE(hdr
);
8679 DTRACE_PROBE2(l2arc__write
, vdev_t
*, dev
->l2ad_vdev
,
8682 write_psize
+= psize
;
8683 write_asize
+= asize
;
8684 dev
->l2ad_hand
+= asize
;
8686 mutex_exit(hash_lock
);
8688 (void) zio_nowait(wzio
);
8691 multilist_sublist_unlock(mls
);
8697 /* No buffers selected for writing? */
8699 ASSERT0(write_lsize
);
8700 ASSERT(!HDR_HAS_L1HDR(head
));
8701 kmem_cache_free(hdr_l2only_cache
, head
);
8705 ASSERT3U(write_asize
, <=, target_sz
);
8706 ARCSTAT_BUMP(arcstat_l2_writes_sent
);
8707 ARCSTAT_INCR(arcstat_l2_write_bytes
, write_psize
);
8708 ARCSTAT_INCR(arcstat_l2_lsize
, write_lsize
);
8709 ARCSTAT_INCR(arcstat_l2_psize
, write_psize
);
8710 vdev_space_update(dev
->l2ad_vdev
, write_psize
, 0, 0);
8713 * Bump device hand to the device start if it is approaching the end.
8714 * l2arc_evict() will already have evicted ahead for this case.
8716 if (dev
->l2ad_hand
>= (dev
->l2ad_end
- target_sz
)) {
8717 dev
->l2ad_hand
= dev
->l2ad_start
;
8718 dev
->l2ad_first
= B_FALSE
;
8721 dev
->l2ad_writing
= B_TRUE
;
8722 (void) zio_wait(pio
);
8723 dev
->l2ad_writing
= B_FALSE
;
8725 return (write_asize
);
8729 * This thread feeds the L2ARC at regular intervals. This is the beating
8730 * heart of the L2ARC.
8734 l2arc_feed_thread(void *unused
)
8739 uint64_t size
, wrote
;
8740 clock_t begin
, next
= ddi_get_lbolt();
8741 fstrans_cookie_t cookie
;
8743 CALLB_CPR_INIT(&cpr
, &l2arc_feed_thr_lock
, callb_generic_cpr
, FTAG
);
8745 mutex_enter(&l2arc_feed_thr_lock
);
8747 cookie
= spl_fstrans_mark();
8748 while (l2arc_thread_exit
== 0) {
8749 CALLB_CPR_SAFE_BEGIN(&cpr
);
8750 (void) cv_timedwait_sig(&l2arc_feed_thr_cv
,
8751 &l2arc_feed_thr_lock
, next
);
8752 CALLB_CPR_SAFE_END(&cpr
, &l2arc_feed_thr_lock
);
8753 next
= ddi_get_lbolt() + hz
;
8756 * Quick check for L2ARC devices.
8758 mutex_enter(&l2arc_dev_mtx
);
8759 if (l2arc_ndev
== 0) {
8760 mutex_exit(&l2arc_dev_mtx
);
8763 mutex_exit(&l2arc_dev_mtx
);
8764 begin
= ddi_get_lbolt();
8767 * This selects the next l2arc device to write to, and in
8768 * doing so the next spa to feed from: dev->l2ad_spa. This
8769 * will return NULL if there are now no l2arc devices or if
8770 * they are all faulted.
8772 * If a device is returned, its spa's config lock is also
8773 * held to prevent device removal. l2arc_dev_get_next()
8774 * will grab and release l2arc_dev_mtx.
8776 if ((dev
= l2arc_dev_get_next()) == NULL
)
8779 spa
= dev
->l2ad_spa
;
8780 ASSERT3P(spa
, !=, NULL
);
8783 * If the pool is read-only then force the feed thread to
8784 * sleep a little longer.
8786 if (!spa_writeable(spa
)) {
8787 next
= ddi_get_lbolt() + 5 * l2arc_feed_secs
* hz
;
8788 spa_config_exit(spa
, SCL_L2ARC
, dev
);
8793 * Avoid contributing to memory pressure.
8795 if (arc_reclaim_needed()) {
8796 ARCSTAT_BUMP(arcstat_l2_abort_lowmem
);
8797 spa_config_exit(spa
, SCL_L2ARC
, dev
);
8801 ARCSTAT_BUMP(arcstat_l2_feeds
);
8803 size
= l2arc_write_size();
8806 * Evict L2ARC buffers that will be overwritten.
8808 l2arc_evict(dev
, size
, B_FALSE
);
8811 * Write ARC buffers.
8813 wrote
= l2arc_write_buffers(spa
, dev
, size
);
8816 * Calculate interval between writes.
8818 next
= l2arc_write_interval(begin
, size
, wrote
);
8819 spa_config_exit(spa
, SCL_L2ARC
, dev
);
8821 spl_fstrans_unmark(cookie
);
8823 l2arc_thread_exit
= 0;
8824 cv_broadcast(&l2arc_feed_thr_cv
);
8825 CALLB_CPR_EXIT(&cpr
); /* drops l2arc_feed_thr_lock */
8830 l2arc_vdev_present(vdev_t
*vd
)
8834 mutex_enter(&l2arc_dev_mtx
);
8835 for (dev
= list_head(l2arc_dev_list
); dev
!= NULL
;
8836 dev
= list_next(l2arc_dev_list
, dev
)) {
8837 if (dev
->l2ad_vdev
== vd
)
8840 mutex_exit(&l2arc_dev_mtx
);
8842 return (dev
!= NULL
);
8846 * Add a vdev for use by the L2ARC. By this point the spa has already
8847 * validated the vdev and opened it.
8850 l2arc_add_vdev(spa_t
*spa
, vdev_t
*vd
)
8852 l2arc_dev_t
*adddev
;
8854 ASSERT(!l2arc_vdev_present(vd
));
8857 * Create a new l2arc device entry.
8859 adddev
= kmem_zalloc(sizeof (l2arc_dev_t
), KM_SLEEP
);
8860 adddev
->l2ad_spa
= spa
;
8861 adddev
->l2ad_vdev
= vd
;
8862 adddev
->l2ad_start
= VDEV_LABEL_START_SIZE
;
8863 adddev
->l2ad_end
= VDEV_LABEL_START_SIZE
+ vdev_get_min_asize(vd
);
8864 adddev
->l2ad_hand
= adddev
->l2ad_start
;
8865 adddev
->l2ad_first
= B_TRUE
;
8866 adddev
->l2ad_writing
= B_FALSE
;
8867 list_link_init(&adddev
->l2ad_node
);
8869 mutex_init(&adddev
->l2ad_mtx
, NULL
, MUTEX_DEFAULT
, NULL
);
8871 * This is a list of all ARC buffers that are still valid on the
8874 list_create(&adddev
->l2ad_buflist
, sizeof (arc_buf_hdr_t
),
8875 offsetof(arc_buf_hdr_t
, b_l2hdr
.b_l2node
));
8877 vdev_space_update(vd
, 0, 0, adddev
->l2ad_end
- adddev
->l2ad_hand
);
8878 refcount_create(&adddev
->l2ad_alloc
);
8881 * Add device to global list
8883 mutex_enter(&l2arc_dev_mtx
);
8884 list_insert_head(l2arc_dev_list
, adddev
);
8885 atomic_inc_64(&l2arc_ndev
);
8886 mutex_exit(&l2arc_dev_mtx
);
8890 * Remove a vdev from the L2ARC.
8893 l2arc_remove_vdev(vdev_t
*vd
)
8895 l2arc_dev_t
*dev
, *nextdev
, *remdev
= NULL
;
8898 * Find the device by vdev
8900 mutex_enter(&l2arc_dev_mtx
);
8901 for (dev
= list_head(l2arc_dev_list
); dev
; dev
= nextdev
) {
8902 nextdev
= list_next(l2arc_dev_list
, dev
);
8903 if (vd
== dev
->l2ad_vdev
) {
8908 ASSERT3P(remdev
, !=, NULL
);
8911 * Remove device from global list
8913 list_remove(l2arc_dev_list
, remdev
);
8914 l2arc_dev_last
= NULL
; /* may have been invalidated */
8915 atomic_dec_64(&l2arc_ndev
);
8916 mutex_exit(&l2arc_dev_mtx
);
8919 * Clear all buflists and ARC references. L2ARC device flush.
8921 l2arc_evict(remdev
, 0, B_TRUE
);
8922 list_destroy(&remdev
->l2ad_buflist
);
8923 mutex_destroy(&remdev
->l2ad_mtx
);
8924 refcount_destroy(&remdev
->l2ad_alloc
);
8925 kmem_free(remdev
, sizeof (l2arc_dev_t
));
8931 l2arc_thread_exit
= 0;
8933 l2arc_writes_sent
= 0;
8934 l2arc_writes_done
= 0;
8936 mutex_init(&l2arc_feed_thr_lock
, NULL
, MUTEX_DEFAULT
, NULL
);
8937 cv_init(&l2arc_feed_thr_cv
, NULL
, CV_DEFAULT
, NULL
);
8938 mutex_init(&l2arc_dev_mtx
, NULL
, MUTEX_DEFAULT
, NULL
);
8939 mutex_init(&l2arc_free_on_write_mtx
, NULL
, MUTEX_DEFAULT
, NULL
);
8941 l2arc_dev_list
= &L2ARC_dev_list
;
8942 l2arc_free_on_write
= &L2ARC_free_on_write
;
8943 list_create(l2arc_dev_list
, sizeof (l2arc_dev_t
),
8944 offsetof(l2arc_dev_t
, l2ad_node
));
8945 list_create(l2arc_free_on_write
, sizeof (l2arc_data_free_t
),
8946 offsetof(l2arc_data_free_t
, l2df_list_node
));
8953 * This is called from dmu_fini(), which is called from spa_fini();
8954 * Because of this, we can assume that all l2arc devices have
8955 * already been removed when the pools themselves were removed.
8958 l2arc_do_free_on_write();
8960 mutex_destroy(&l2arc_feed_thr_lock
);
8961 cv_destroy(&l2arc_feed_thr_cv
);
8962 mutex_destroy(&l2arc_dev_mtx
);
8963 mutex_destroy(&l2arc_free_on_write_mtx
);
8965 list_destroy(l2arc_dev_list
);
8966 list_destroy(l2arc_free_on_write
);
8972 if (!(spa_mode_global
& FWRITE
))
8975 (void) thread_create(NULL
, 0, l2arc_feed_thread
, NULL
, 0, &p0
,
8976 TS_RUN
, defclsyspri
);
8982 if (!(spa_mode_global
& FWRITE
))
8985 mutex_enter(&l2arc_feed_thr_lock
);
8986 cv_signal(&l2arc_feed_thr_cv
); /* kick thread out of startup */
8987 l2arc_thread_exit
= 1;
8988 while (l2arc_thread_exit
!= 0)
8989 cv_wait(&l2arc_feed_thr_cv
, &l2arc_feed_thr_lock
);
8990 mutex_exit(&l2arc_feed_thr_lock
);
8993 #if defined(_KERNEL) && defined(HAVE_SPL)
8994 EXPORT_SYMBOL(arc_buf_size
);
8995 EXPORT_SYMBOL(arc_write
);
8996 EXPORT_SYMBOL(arc_read
);
8997 EXPORT_SYMBOL(arc_buf_info
);
8998 EXPORT_SYMBOL(arc_getbuf_func
);
8999 EXPORT_SYMBOL(arc_add_prune_callback
);
9000 EXPORT_SYMBOL(arc_remove_prune_callback
);
9003 module_param(zfs_arc_min
, ulong
, 0644);
9004 MODULE_PARM_DESC(zfs_arc_min
, "Min arc size");
9006 module_param(zfs_arc_max
, ulong
, 0644);
9007 MODULE_PARM_DESC(zfs_arc_max
, "Max arc size");
9009 module_param(zfs_arc_meta_limit
, ulong
, 0644);
9010 MODULE_PARM_DESC(zfs_arc_meta_limit
, "Meta limit for arc size");
9012 module_param(zfs_arc_meta_limit_percent
, ulong
, 0644);
9013 MODULE_PARM_DESC(zfs_arc_meta_limit_percent
,
9014 "Percent of arc size for arc meta limit");
9016 module_param(zfs_arc_meta_min
, ulong
, 0644);
9017 MODULE_PARM_DESC(zfs_arc_meta_min
, "Min arc metadata");
9019 module_param(zfs_arc_meta_prune
, int, 0644);
9020 MODULE_PARM_DESC(zfs_arc_meta_prune
, "Meta objects to scan for prune");
9022 module_param(zfs_arc_meta_adjust_restarts
, int, 0644);
9023 MODULE_PARM_DESC(zfs_arc_meta_adjust_restarts
,
9024 "Limit number of restarts in arc_adjust_meta");
9026 module_param(zfs_arc_meta_strategy
, int, 0644);
9027 MODULE_PARM_DESC(zfs_arc_meta_strategy
, "Meta reclaim strategy");
9029 module_param(zfs_arc_grow_retry
, int, 0644);
9030 MODULE_PARM_DESC(zfs_arc_grow_retry
, "Seconds before growing arc size");
9032 module_param(zfs_arc_p_aggressive_disable
, int, 0644);
9033 MODULE_PARM_DESC(zfs_arc_p_aggressive_disable
, "disable aggressive arc_p grow");
9035 module_param(zfs_arc_p_dampener_disable
, int, 0644);
9036 MODULE_PARM_DESC(zfs_arc_p_dampener_disable
, "disable arc_p adapt dampener");
9038 module_param(zfs_arc_shrink_shift
, int, 0644);
9039 MODULE_PARM_DESC(zfs_arc_shrink_shift
, "log2(fraction of arc to reclaim)");
9041 module_param(zfs_arc_pc_percent
, uint
, 0644);
9042 MODULE_PARM_DESC(zfs_arc_pc_percent
,
9043 "Percent of pagecache to reclaim arc to");
9045 module_param(zfs_arc_p_min_shift
, int, 0644);
9046 MODULE_PARM_DESC(zfs_arc_p_min_shift
, "arc_c shift to calc min/max arc_p");
9048 module_param(zfs_arc_average_blocksize
, int, 0444);
9049 MODULE_PARM_DESC(zfs_arc_average_blocksize
, "Target average block size");
9051 module_param(zfs_compressed_arc_enabled
, int, 0644);
9052 MODULE_PARM_DESC(zfs_compressed_arc_enabled
, "Disable compressed arc buffers");
9054 module_param(zfs_arc_min_prefetch_ms
, int, 0644);
9055 MODULE_PARM_DESC(zfs_arc_min_prefetch_ms
, "Min life of prefetch block in ms");
9057 module_param(zfs_arc_min_prescient_prefetch_ms
, int, 0644);
9058 MODULE_PARM_DESC(zfs_arc_min_prescient_prefetch_ms
,
9059 "Min life of prescient prefetched block in ms");
9061 module_param(l2arc_write_max
, ulong
, 0644);
9062 MODULE_PARM_DESC(l2arc_write_max
, "Max write bytes per interval");
9064 module_param(l2arc_write_boost
, ulong
, 0644);
9065 MODULE_PARM_DESC(l2arc_write_boost
, "Extra write bytes during device warmup");
9067 module_param(l2arc_headroom
, ulong
, 0644);
9068 MODULE_PARM_DESC(l2arc_headroom
, "Number of max device writes to precache");
9070 module_param(l2arc_headroom_boost
, ulong
, 0644);
9071 MODULE_PARM_DESC(l2arc_headroom_boost
, "Compressed l2arc_headroom multiplier");
9073 module_param(l2arc_feed_secs
, ulong
, 0644);
9074 MODULE_PARM_DESC(l2arc_feed_secs
, "Seconds between L2ARC writing");
9076 module_param(l2arc_feed_min_ms
, ulong
, 0644);
9077 MODULE_PARM_DESC(l2arc_feed_min_ms
, "Min feed interval in milliseconds");
9079 module_param(l2arc_noprefetch
, int, 0644);
9080 MODULE_PARM_DESC(l2arc_noprefetch
, "Skip caching prefetched buffers");
9082 module_param(l2arc_feed_again
, int, 0644);
9083 MODULE_PARM_DESC(l2arc_feed_again
, "Turbo L2ARC warmup");
9085 module_param(l2arc_norw
, int, 0644);
9086 MODULE_PARM_DESC(l2arc_norw
, "No reads during writes");
9088 module_param(zfs_arc_lotsfree_percent
, int, 0644);
9089 MODULE_PARM_DESC(zfs_arc_lotsfree_percent
,
9090 "System free memory I/O throttle in bytes");
9092 module_param(zfs_arc_sys_free
, ulong
, 0644);
9093 MODULE_PARM_DESC(zfs_arc_sys_free
, "System free memory target size in bytes");
9095 module_param(zfs_arc_dnode_limit
, ulong
, 0644);
9096 MODULE_PARM_DESC(zfs_arc_dnode_limit
, "Minimum bytes of dnodes in arc");
9098 module_param(zfs_arc_dnode_limit_percent
, ulong
, 0644);
9099 MODULE_PARM_DESC(zfs_arc_dnode_limit_percent
,
9100 "Percent of ARC meta buffers for dnodes");
9102 module_param(zfs_arc_dnode_reduce_percent
, ulong
, 0644);
9103 MODULE_PARM_DESC(zfs_arc_dnode_reduce_percent
,
9104 "Percentage of excess dnodes to try to unpin");