Index: stable/9/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/arc.c =================================================================== --- stable/9/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/arc.c (revision 301698) +++ stable/9/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/arc.c (revision 301699) @@ -1,5611 +1,5610 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. * Copyright (c) 2014 by Saso Kiselkov. All rights reserved. * Copyright 2013 Nexenta Systems, Inc. All rights reserved. */ /* * DVA-based Adjustable Replacement Cache * * While much of the theory of operation used here is * based on the self-tuning, low overhead replacement cache * presented by Megiddo and Modha at FAST 2003, there are some * significant differences: * * 1. The Megiddo and Modha model assumes any page is evictable. * Pages in its cache cannot be "locked" into memory. This makes * the eviction algorithm simple: evict the last page in the list. * This also make the performance characteristics easy to reason * about. Our cache is not so simple. At any given moment, some * subset of the blocks in the cache are un-evictable because we * have handed out a reference to them. Blocks are only evictable * when there are no external references active. This makes * eviction far more problematic: we choose to evict the evictable * blocks that are the "lowest" in the list. * * There are times when it is not possible to evict the requested * space. In these circumstances we are unable to adjust the cache * size. To prevent the cache growing unbounded at these times we * implement a "cache throttle" that slows the flow of new data * into the cache until we can make space available. * * 2. The Megiddo and Modha model assumes a fixed cache size. * Pages are evicted when the cache is full and there is a cache * miss. Our model has a variable sized cache. It grows with * high use, but also tries to react to memory pressure from the * operating system: decreasing its size when system memory is * tight. * * 3. The Megiddo and Modha model assumes a fixed page size. All * elements of the cache are therefore exactly the same size. So * when adjusting the cache size following a cache miss, its simply * a matter of choosing a single page to evict. In our model, we * have variable sized cache blocks (rangeing from 512 bytes to * 128K bytes). We therefore choose a set of blocks to evict to make * space for a cache miss that approximates as closely as possible * the space used by the new block. * * See also: "ARC: A Self-Tuning, Low Overhead Replacement Cache" * by N. Megiddo & D. Modha, FAST 2003 */ /* * The locking model: * * A new reference to a cache buffer can be obtained in two * ways: 1) via a hash table lookup using the DVA as a key, * or 2) via one of the ARC lists. The arc_read() interface * uses method 1, while the internal arc algorithms for * adjusting the cache use method 2. We therefore provide two * types of locks: 1) the hash table lock array, and 2) the * arc list locks. * * Buffers do not have their own mutexs, rather they rely on the * hash table mutexs for the bulk of their protection (i.e. most * fields in the arc_buf_hdr_t are protected by these mutexs). * * buf_hash_find() returns the appropriate mutex (held) when it * locates the requested buffer in the hash table. It returns * NULL for the mutex if the buffer was not in the table. * * buf_hash_remove() expects the appropriate hash mutex to be * already held before it is invoked. * * Each arc state also has a mutex which is used to protect the * buffer list associated with the state. When attempting to * obtain a hash table lock while holding an arc list lock you * must use: mutex_tryenter() to avoid deadlock. Also note that * the active state mutex must be held before the ghost state mutex. * * Arc buffers may have an associated eviction callback function. * This function will be invoked prior to removing the buffer (e.g. * in arc_do_user_evicts()). Note however that the data associated * with the buffer may be evicted prior to the callback. The callback * must be made with *no locks held* (to prevent deadlock). Additionally, * the users of callbacks must ensure that their private data is * protected from simultaneous callbacks from arc_buf_evict() * and arc_do_user_evicts(). * * Note that the majority of the performance stats are manipulated * with atomic operations. * * The L2ARC uses the l2arc_buflist_mtx global mutex for the following: * * - L2ARC buflist creation * - L2ARC buflist eviction * - L2ARC write completion, which walks L2ARC buflists * - ARC header destruction, as it removes from L2ARC buflists * - ARC header release, as it removes from L2ARC buflists */ #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #endif #include #include #include #include #include -#include #ifdef illumos #ifndef _KERNEL /* set with ZFS_DEBUG=watch, to enable watchpoints on frozen buffers */ boolean_t arc_watch = B_FALSE; int arc_procfd; #endif #endif /* illumos */ static kmutex_t arc_reclaim_thr_lock; static kcondvar_t arc_reclaim_thr_cv; /* used to signal reclaim thr */ static uint8_t arc_thread_exit; #define ARC_REDUCE_DNLC_PERCENT 3 uint_t arc_reduce_dnlc_percent = ARC_REDUCE_DNLC_PERCENT; typedef enum arc_reclaim_strategy { ARC_RECLAIM_AGGR, /* Aggressive reclaim strategy */ ARC_RECLAIM_CONS /* Conservative reclaim strategy */ } arc_reclaim_strategy_t; /* * The number of iterations through arc_evict_*() before we * drop & reacquire the lock. */ int arc_evict_iterations = 100; /* number of seconds before growing cache again */ static int arc_grow_retry = 60; /* shift of arc_c for calculating both min and max arc_p */ static int arc_p_min_shift = 4; /* log2(fraction of arc to reclaim) */ static int arc_shrink_shift = 5; /* * minimum lifespan of a prefetch block in clock ticks * (initialized in arc_init()) */ static int arc_min_prefetch_lifespan; /* * If this percent of memory is free, don't throttle. */ int arc_lotsfree_percent = 10; static int arc_dead; extern int zfs_prefetch_disable; /* * The arc has filled available memory and has now warmed up. */ static boolean_t arc_warm; /* * These tunables are for performance analysis. */ uint64_t zfs_arc_max; uint64_t zfs_arc_min; uint64_t zfs_arc_meta_limit = 0; int zfs_arc_grow_retry = 0; int zfs_arc_shrink_shift = 0; int zfs_arc_p_min_shift = 0; int zfs_disable_dup_eviction = 0; TUNABLE_QUAD("vfs.zfs.arc_max", &zfs_arc_max); TUNABLE_QUAD("vfs.zfs.arc_min", &zfs_arc_min); TUNABLE_QUAD("vfs.zfs.arc_meta_limit", &zfs_arc_meta_limit); SYSCTL_DECL(_vfs_zfs); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, arc_max, CTLFLAG_RDTUN, &zfs_arc_max, 0, "Maximum ARC size"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, arc_min, CTLFLAG_RDTUN, &zfs_arc_min, 0, "Minimum ARC size"); /* * Note that buffers can be in one of 6 states: * ARC_anon - anonymous (discussed below) * ARC_mru - recently used, currently cached * ARC_mru_ghost - recentely used, no longer in cache * ARC_mfu - frequently used, currently cached * ARC_mfu_ghost - frequently used, no longer in cache * ARC_l2c_only - exists in L2ARC but not other states * When there are no active references to the buffer, they are * are linked onto a list in one of these arc states. These are * the only buffers that can be evicted or deleted. Within each * state there are multiple lists, one for meta-data and one for * non-meta-data. Meta-data (indirect blocks, blocks of dnodes, * etc.) is tracked separately so that it can be managed more * explicitly: favored over data, limited explicitly. * * Anonymous buffers are buffers that are not associated with * a DVA. These are buffers that hold dirty block copies * before they are written to stable storage. By definition, * they are "ref'd" and are considered part of arc_mru * that cannot be freed. Generally, they will aquire a DVA * as they are written and migrate onto the arc_mru list. * * The ARC_l2c_only state is for buffers that are in the second * level ARC but no longer in any of the ARC_m* lists. The second * level ARC itself may also contain buffers that are in any of * the ARC_m* states - meaning that a buffer can exist in two * places. The reason for the ARC_l2c_only state is to keep the * buffer header in the hash table, so that reads that hit the * second level ARC benefit from these fast lookups. */ #define ARCS_LOCK_PAD CACHE_LINE_SIZE struct arcs_lock { kmutex_t arcs_lock; #ifdef _KERNEL unsigned char pad[(ARCS_LOCK_PAD - sizeof (kmutex_t))]; #endif }; /* * must be power of two for mask use to work * */ #define ARC_BUFC_NUMDATALISTS 16 #define ARC_BUFC_NUMMETADATALISTS 16 #define ARC_BUFC_NUMLISTS (ARC_BUFC_NUMMETADATALISTS + ARC_BUFC_NUMDATALISTS) typedef struct arc_state { uint64_t arcs_lsize[ARC_BUFC_NUMTYPES]; /* amount of evictable data */ uint64_t arcs_size; /* total amount of data in this state */ list_t arcs_lists[ARC_BUFC_NUMLISTS]; /* list of evictable buffers */ struct arcs_lock arcs_locks[ARC_BUFC_NUMLISTS] __aligned(CACHE_LINE_SIZE); } arc_state_t; #define ARCS_LOCK(s, i) (&((s)->arcs_locks[(i)].arcs_lock)) /* The 6 states: */ static arc_state_t ARC_anon; static arc_state_t ARC_mru; static arc_state_t ARC_mru_ghost; static arc_state_t ARC_mfu; static arc_state_t ARC_mfu_ghost; static arc_state_t ARC_l2c_only; typedef struct arc_stats { kstat_named_t arcstat_hits; kstat_named_t arcstat_misses; kstat_named_t arcstat_demand_data_hits; kstat_named_t arcstat_demand_data_misses; kstat_named_t arcstat_demand_metadata_hits; kstat_named_t arcstat_demand_metadata_misses; kstat_named_t arcstat_prefetch_data_hits; kstat_named_t arcstat_prefetch_data_misses; kstat_named_t arcstat_prefetch_metadata_hits; kstat_named_t arcstat_prefetch_metadata_misses; kstat_named_t arcstat_mru_hits; kstat_named_t arcstat_mru_ghost_hits; kstat_named_t arcstat_mfu_hits; kstat_named_t arcstat_mfu_ghost_hits; kstat_named_t arcstat_allocated; kstat_named_t arcstat_deleted; kstat_named_t arcstat_stolen; kstat_named_t arcstat_recycle_miss; /* * Number of buffers that could not be evicted because the hash lock * was held by another thread. The lock may not necessarily be held * by something using the same buffer, since hash locks are shared * by multiple buffers. */ kstat_named_t arcstat_mutex_miss; /* * Number of buffers skipped because they have I/O in progress, are * indrect prefetch buffers that have not lived long enough, or are * not from the spa we're trying to evict from. */ kstat_named_t arcstat_evict_skip; kstat_named_t arcstat_evict_l2_cached; kstat_named_t arcstat_evict_l2_eligible; kstat_named_t arcstat_evict_l2_ineligible; kstat_named_t arcstat_hash_elements; kstat_named_t arcstat_hash_elements_max; kstat_named_t arcstat_hash_collisions; kstat_named_t arcstat_hash_chains; kstat_named_t arcstat_hash_chain_max; kstat_named_t arcstat_p; kstat_named_t arcstat_c; kstat_named_t arcstat_c_min; kstat_named_t arcstat_c_max; kstat_named_t arcstat_size; kstat_named_t arcstat_hdr_size; kstat_named_t arcstat_data_size; kstat_named_t arcstat_other_size; kstat_named_t arcstat_l2_hits; kstat_named_t arcstat_l2_misses; kstat_named_t arcstat_l2_feeds; kstat_named_t arcstat_l2_rw_clash; kstat_named_t arcstat_l2_read_bytes; kstat_named_t arcstat_l2_write_bytes; kstat_named_t arcstat_l2_writes_sent; kstat_named_t arcstat_l2_writes_done; kstat_named_t arcstat_l2_writes_error; kstat_named_t arcstat_l2_writes_hdr_miss; kstat_named_t arcstat_l2_evict_lock_retry; kstat_named_t arcstat_l2_evict_reading; kstat_named_t arcstat_l2_free_on_write; kstat_named_t arcstat_l2_abort_lowmem; kstat_named_t arcstat_l2_cksum_bad; kstat_named_t arcstat_l2_io_error; kstat_named_t arcstat_l2_size; kstat_named_t arcstat_l2_asize; kstat_named_t arcstat_l2_hdr_size; kstat_named_t arcstat_l2_compress_successes; kstat_named_t arcstat_l2_compress_zeros; kstat_named_t arcstat_l2_compress_failures; kstat_named_t arcstat_l2_write_trylock_fail; kstat_named_t arcstat_l2_write_passed_headroom; kstat_named_t arcstat_l2_write_spa_mismatch; kstat_named_t arcstat_l2_write_in_l2; kstat_named_t arcstat_l2_write_hdr_io_in_progress; kstat_named_t arcstat_l2_write_not_cacheable; kstat_named_t arcstat_l2_write_full; kstat_named_t arcstat_l2_write_buffer_iter; kstat_named_t arcstat_l2_write_pios; kstat_named_t arcstat_l2_write_buffer_bytes_scanned; kstat_named_t arcstat_l2_write_buffer_list_iter; kstat_named_t arcstat_l2_write_buffer_list_null_iter; kstat_named_t arcstat_memory_throttle_count; kstat_named_t arcstat_duplicate_buffers; kstat_named_t arcstat_duplicate_buffers_size; kstat_named_t arcstat_duplicate_reads; } arc_stats_t; static arc_stats_t arc_stats = { { "hits", KSTAT_DATA_UINT64 }, { "misses", KSTAT_DATA_UINT64 }, { "demand_data_hits", KSTAT_DATA_UINT64 }, { "demand_data_misses", KSTAT_DATA_UINT64 }, { "demand_metadata_hits", KSTAT_DATA_UINT64 }, { "demand_metadata_misses", KSTAT_DATA_UINT64 }, { "prefetch_data_hits", KSTAT_DATA_UINT64 }, { "prefetch_data_misses", KSTAT_DATA_UINT64 }, { "prefetch_metadata_hits", KSTAT_DATA_UINT64 }, { "prefetch_metadata_misses", KSTAT_DATA_UINT64 }, { "mru_hits", KSTAT_DATA_UINT64 }, { "mru_ghost_hits", KSTAT_DATA_UINT64 }, { "mfu_hits", KSTAT_DATA_UINT64 }, { "mfu_ghost_hits", KSTAT_DATA_UINT64 }, { "allocated", KSTAT_DATA_UINT64 }, { "deleted", KSTAT_DATA_UINT64 }, { "stolen", KSTAT_DATA_UINT64 }, { "recycle_miss", KSTAT_DATA_UINT64 }, { "mutex_miss", KSTAT_DATA_UINT64 }, { "evict_skip", KSTAT_DATA_UINT64 }, { "evict_l2_cached", KSTAT_DATA_UINT64 }, { "evict_l2_eligible", KSTAT_DATA_UINT64 }, { "evict_l2_ineligible", KSTAT_DATA_UINT64 }, { "hash_elements", KSTAT_DATA_UINT64 }, { "hash_elements_max", KSTAT_DATA_UINT64 }, { "hash_collisions", KSTAT_DATA_UINT64 }, { "hash_chains", KSTAT_DATA_UINT64 }, { "hash_chain_max", KSTAT_DATA_UINT64 }, { "p", KSTAT_DATA_UINT64 }, { "c", KSTAT_DATA_UINT64 }, { "c_min", KSTAT_DATA_UINT64 }, { "c_max", KSTAT_DATA_UINT64 }, { "size", KSTAT_DATA_UINT64 }, { "hdr_size", KSTAT_DATA_UINT64 }, { "data_size", KSTAT_DATA_UINT64 }, { "other_size", KSTAT_DATA_UINT64 }, { "l2_hits", KSTAT_DATA_UINT64 }, { "l2_misses", KSTAT_DATA_UINT64 }, { "l2_feeds", KSTAT_DATA_UINT64 }, { "l2_rw_clash", KSTAT_DATA_UINT64 }, { "l2_read_bytes", KSTAT_DATA_UINT64 }, { "l2_write_bytes", KSTAT_DATA_UINT64 }, { "l2_writes_sent", KSTAT_DATA_UINT64 }, { "l2_writes_done", KSTAT_DATA_UINT64 }, { "l2_writes_error", KSTAT_DATA_UINT64 }, { "l2_writes_hdr_miss", KSTAT_DATA_UINT64 }, { "l2_evict_lock_retry", KSTAT_DATA_UINT64 }, { "l2_evict_reading", KSTAT_DATA_UINT64 }, { "l2_free_on_write", KSTAT_DATA_UINT64 }, { "l2_abort_lowmem", KSTAT_DATA_UINT64 }, { "l2_cksum_bad", KSTAT_DATA_UINT64 }, { "l2_io_error", KSTAT_DATA_UINT64 }, { "l2_size", KSTAT_DATA_UINT64 }, { "l2_asize", KSTAT_DATA_UINT64 }, { "l2_hdr_size", KSTAT_DATA_UINT64 }, { "l2_compress_successes", KSTAT_DATA_UINT64 }, { "l2_compress_zeros", KSTAT_DATA_UINT64 }, { "l2_compress_failures", KSTAT_DATA_UINT64 }, { "l2_write_trylock_fail", KSTAT_DATA_UINT64 }, { "l2_write_passed_headroom", KSTAT_DATA_UINT64 }, { "l2_write_spa_mismatch", KSTAT_DATA_UINT64 }, { "l2_write_in_l2", KSTAT_DATA_UINT64 }, { "l2_write_io_in_progress", KSTAT_DATA_UINT64 }, { "l2_write_not_cacheable", KSTAT_DATA_UINT64 }, { "l2_write_full", KSTAT_DATA_UINT64 }, { "l2_write_buffer_iter", KSTAT_DATA_UINT64 }, { "l2_write_pios", KSTAT_DATA_UINT64 }, { "l2_write_buffer_bytes_scanned", KSTAT_DATA_UINT64 }, { "l2_write_buffer_list_iter", KSTAT_DATA_UINT64 }, { "l2_write_buffer_list_null_iter", KSTAT_DATA_UINT64 }, { "memory_throttle_count", KSTAT_DATA_UINT64 }, { "duplicate_buffers", KSTAT_DATA_UINT64 }, { "duplicate_buffers_size", KSTAT_DATA_UINT64 }, { "duplicate_reads", KSTAT_DATA_UINT64 } }; #define ARCSTAT(stat) (arc_stats.stat.value.ui64) #define ARCSTAT_INCR(stat, val) \ atomic_add_64(&arc_stats.stat.value.ui64, (val)) #define ARCSTAT_BUMP(stat) ARCSTAT_INCR(stat, 1) #define ARCSTAT_BUMPDOWN(stat) ARCSTAT_INCR(stat, -1) #define ARCSTAT_MAX(stat, val) { \ uint64_t m; \ while ((val) > (m = arc_stats.stat.value.ui64) && \ (m != atomic_cas_64(&arc_stats.stat.value.ui64, m, (val)))) \ continue; \ } #define ARCSTAT_MAXSTAT(stat) \ ARCSTAT_MAX(stat##_max, arc_stats.stat.value.ui64) /* * We define a macro to allow ARC hits/misses to be easily broken down by * two separate conditions, giving a total of four different subtypes for * each of hits and misses (so eight statistics total). */ #define ARCSTAT_CONDSTAT(cond1, stat1, notstat1, cond2, stat2, notstat2, stat) \ if (cond1) { \ if (cond2) { \ ARCSTAT_BUMP(arcstat_##stat1##_##stat2##_##stat); \ } else { \ ARCSTAT_BUMP(arcstat_##stat1##_##notstat2##_##stat); \ } \ } else { \ if (cond2) { \ ARCSTAT_BUMP(arcstat_##notstat1##_##stat2##_##stat); \ } else { \ ARCSTAT_BUMP(arcstat_##notstat1##_##notstat2##_##stat);\ } \ } kstat_t *arc_ksp; static arc_state_t *arc_anon; static arc_state_t *arc_mru; static arc_state_t *arc_mru_ghost; static arc_state_t *arc_mfu; static arc_state_t *arc_mfu_ghost; static arc_state_t *arc_l2c_only; /* * There are several ARC variables that are critical to export as kstats -- * but we don't want to have to grovel around in the kstat whenever we wish to * manipulate them. For these variables, we therefore define them to be in * terms of the statistic variable. This assures that we are not introducing * the possibility of inconsistency by having shadow copies of the variables, * while still allowing the code to be readable. */ #define arc_size ARCSTAT(arcstat_size) /* actual total arc size */ #define arc_p ARCSTAT(arcstat_p) /* target size of MRU */ #define arc_c ARCSTAT(arcstat_c) /* target size of cache */ #define arc_c_min ARCSTAT(arcstat_c_min) /* min target cache size */ #define arc_c_max ARCSTAT(arcstat_c_max) /* max target cache size */ #define L2ARC_IS_VALID_COMPRESS(_c_) \ ((_c_) == ZIO_COMPRESS_LZ4 || (_c_) == ZIO_COMPRESS_EMPTY) static int arc_no_grow; /* Don't try to grow cache size */ static uint64_t arc_tempreserve; static uint64_t arc_loaned_bytes; static uint64_t arc_meta_used; static uint64_t arc_meta_limit; static uint64_t arc_meta_max = 0; SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, arc_meta_used, CTLFLAG_RD, &arc_meta_used, 0, "ARC metadata used"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, arc_meta_limit, CTLFLAG_RW, &arc_meta_limit, 0, "ARC metadata limit"); typedef struct l2arc_buf_hdr l2arc_buf_hdr_t; typedef struct arc_callback arc_callback_t; struct arc_callback { void *acb_private; arc_done_func_t *acb_done; arc_buf_t *acb_buf; zio_t *acb_zio_dummy; arc_callback_t *acb_next; }; typedef struct arc_write_callback arc_write_callback_t; struct arc_write_callback { void *awcb_private; arc_done_func_t *awcb_ready; arc_done_func_t *awcb_physdone; arc_done_func_t *awcb_done; arc_buf_t *awcb_buf; }; struct arc_buf_hdr { /* protected by hash lock */ dva_t b_dva; uint64_t b_birth; uint64_t b_cksum0; kmutex_t b_freeze_lock; zio_cksum_t *b_freeze_cksum; void *b_thawed; arc_buf_hdr_t *b_hash_next; arc_buf_t *b_buf; uint32_t b_flags; uint32_t b_datacnt; arc_callback_t *b_acb; kcondvar_t b_cv; /* immutable */ arc_buf_contents_t b_type; uint64_t b_size; uint64_t b_spa; /* protected by arc state mutex */ arc_state_t *b_state; list_node_t b_arc_node; /* updated atomically */ clock_t b_arc_access; /* self protecting */ refcount_t b_refcnt; l2arc_buf_hdr_t *b_l2hdr; list_node_t b_l2node; }; static arc_buf_t *arc_eviction_list; static kmutex_t arc_eviction_mtx; static arc_buf_hdr_t arc_eviction_hdr; static void arc_get_data_buf(arc_buf_t *buf); static void arc_access(arc_buf_hdr_t *buf, kmutex_t *hash_lock); static int arc_evict_needed(arc_buf_contents_t type); static void arc_evict_ghost(arc_state_t *state, uint64_t spa, int64_t bytes); #ifdef illumos static void arc_buf_watch(arc_buf_t *buf); #endif /* illumos */ static boolean_t l2arc_write_eligible(uint64_t spa_guid, arc_buf_hdr_t *ab); #define GHOST_STATE(state) \ ((state) == arc_mru_ghost || (state) == arc_mfu_ghost || \ (state) == arc_l2c_only) /* * Private ARC flags. These flags are private ARC only flags that will show up * in b_flags in the arc_hdr_buf_t. Some flags are publicly declared, and can * be passed in as arc_flags in things like arc_read. However, these flags * should never be passed and should only be set by ARC code. When adding new * public flags, make sure not to smash the private ones. */ #define ARC_IN_HASH_TABLE (1 << 9) /* this buffer is hashed */ #define ARC_IO_IN_PROGRESS (1 << 10) /* I/O in progress for buf */ #define ARC_IO_ERROR (1 << 11) /* I/O failed for buf */ #define ARC_FREED_IN_READ (1 << 12) /* buf freed while in read */ #define ARC_BUF_AVAILABLE (1 << 13) /* block not in active use */ #define ARC_INDIRECT (1 << 14) /* this is an indirect block */ #define ARC_FREE_IN_PROGRESS (1 << 15) /* hdr about to be freed */ #define ARC_L2_WRITING (1 << 16) /* L2ARC write in progress */ #define ARC_L2_EVICTED (1 << 17) /* evicted during I/O */ #define ARC_L2_WRITE_HEAD (1 << 18) /* head of write list */ #define HDR_IN_HASH_TABLE(hdr) ((hdr)->b_flags & ARC_IN_HASH_TABLE) #define HDR_IO_IN_PROGRESS(hdr) ((hdr)->b_flags & ARC_IO_IN_PROGRESS) #define HDR_IO_ERROR(hdr) ((hdr)->b_flags & ARC_IO_ERROR) #define HDR_PREFETCH(hdr) ((hdr)->b_flags & ARC_PREFETCH) #define HDR_FREED_IN_READ(hdr) ((hdr)->b_flags & ARC_FREED_IN_READ) #define HDR_BUF_AVAILABLE(hdr) ((hdr)->b_flags & ARC_BUF_AVAILABLE) #define HDR_FREE_IN_PROGRESS(hdr) ((hdr)->b_flags & ARC_FREE_IN_PROGRESS) #define HDR_L2CACHE(hdr) ((hdr)->b_flags & ARC_L2CACHE) #define HDR_L2_READING(hdr) ((hdr)->b_flags & ARC_IO_IN_PROGRESS && \ (hdr)->b_l2hdr != NULL) #define HDR_L2_WRITING(hdr) ((hdr)->b_flags & ARC_L2_WRITING) #define HDR_L2_EVICTED(hdr) ((hdr)->b_flags & ARC_L2_EVICTED) #define HDR_L2_WRITE_HEAD(hdr) ((hdr)->b_flags & ARC_L2_WRITE_HEAD) /* * Other sizes */ #define HDR_SIZE ((int64_t)sizeof (arc_buf_hdr_t)) #define L2HDR_SIZE ((int64_t)sizeof (l2arc_buf_hdr_t)) /* * Hash table routines */ #define HT_LOCK_PAD CACHE_LINE_SIZE struct ht_lock { kmutex_t ht_lock; #ifdef _KERNEL unsigned char pad[(HT_LOCK_PAD - sizeof (kmutex_t))]; #endif }; #define BUF_LOCKS 256 typedef struct buf_hash_table { uint64_t ht_mask; arc_buf_hdr_t **ht_table; struct ht_lock ht_locks[BUF_LOCKS] __aligned(CACHE_LINE_SIZE); } buf_hash_table_t; static buf_hash_table_t buf_hash_table; #define BUF_HASH_INDEX(spa, dva, birth) \ (buf_hash(spa, dva, birth) & buf_hash_table.ht_mask) #define BUF_HASH_LOCK_NTRY(idx) (buf_hash_table.ht_locks[idx & (BUF_LOCKS-1)]) #define BUF_HASH_LOCK(idx) (&(BUF_HASH_LOCK_NTRY(idx).ht_lock)) #define HDR_LOCK(hdr) \ (BUF_HASH_LOCK(BUF_HASH_INDEX(hdr->b_spa, &hdr->b_dva, hdr->b_birth))) uint64_t zfs_crc64_table[256]; /* * Level 2 ARC */ #define L2ARC_WRITE_SIZE (8 * 1024 * 1024) /* initial write max */ #define L2ARC_HEADROOM 2 /* num of writes */ /* * If we discover during ARC scan any buffers to be compressed, we boost * our headroom for the next scanning cycle by this percentage multiple. */ #define L2ARC_HEADROOM_BOOST 200 #define L2ARC_FEED_SECS 1 /* caching interval secs */ #define L2ARC_FEED_MIN_MS 200 /* min caching interval ms */ #define l2arc_writes_sent ARCSTAT(arcstat_l2_writes_sent) #define l2arc_writes_done ARCSTAT(arcstat_l2_writes_done) /* L2ARC Performance Tunables */ uint64_t l2arc_write_max = L2ARC_WRITE_SIZE; /* default max write size */ uint64_t l2arc_write_boost = L2ARC_WRITE_SIZE; /* extra write during warmup */ uint64_t l2arc_headroom = L2ARC_HEADROOM; /* number of dev writes */ uint64_t l2arc_headroom_boost = L2ARC_HEADROOM_BOOST; uint64_t l2arc_feed_secs = L2ARC_FEED_SECS; /* interval seconds */ uint64_t l2arc_feed_min_ms = L2ARC_FEED_MIN_MS; /* min interval milliseconds */ boolean_t l2arc_noprefetch = B_TRUE; /* don't cache prefetch bufs */ boolean_t l2arc_feed_again = B_TRUE; /* turbo warmup */ boolean_t l2arc_norw = B_TRUE; /* no reads during writes */ SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_max, CTLFLAG_RW, &l2arc_write_max, 0, "max write size"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_boost, CTLFLAG_RW, &l2arc_write_boost, 0, "extra write during warmup"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom, CTLFLAG_RW, &l2arc_headroom, 0, "number of dev writes"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_secs, CTLFLAG_RW, &l2arc_feed_secs, 0, "interval seconds"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_min_ms, CTLFLAG_RW, &l2arc_feed_min_ms, 0, "min interval milliseconds"); SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_noprefetch, CTLFLAG_RW, &l2arc_noprefetch, 0, "don't cache prefetch bufs"); SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_feed_again, CTLFLAG_RW, &l2arc_feed_again, 0, "turbo warmup"); SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_norw, CTLFLAG_RW, &l2arc_norw, 0, "no reads during writes"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_size, CTLFLAG_RD, &ARC_anon.arcs_size, 0, "size of anonymous state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_metadata_lsize, CTLFLAG_RD, &ARC_anon.arcs_lsize[ARC_BUFC_METADATA], 0, "size of anonymous state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_data_lsize, CTLFLAG_RD, &ARC_anon.arcs_lsize[ARC_BUFC_DATA], 0, "size of anonymous state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_size, CTLFLAG_RD, &ARC_mru.arcs_size, 0, "size of mru state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_metadata_lsize, CTLFLAG_RD, &ARC_mru.arcs_lsize[ARC_BUFC_METADATA], 0, "size of metadata in mru state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_data_lsize, CTLFLAG_RD, &ARC_mru.arcs_lsize[ARC_BUFC_DATA], 0, "size of data in mru state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_size, CTLFLAG_RD, &ARC_mru_ghost.arcs_size, 0, "size of mru ghost state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_metadata_lsize, CTLFLAG_RD, &ARC_mru_ghost.arcs_lsize[ARC_BUFC_METADATA], 0, "size of metadata in mru ghost state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_data_lsize, CTLFLAG_RD, &ARC_mru_ghost.arcs_lsize[ARC_BUFC_DATA], 0, "size of data in mru ghost state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_size, CTLFLAG_RD, &ARC_mfu.arcs_size, 0, "size of mfu state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_metadata_lsize, CTLFLAG_RD, &ARC_mfu.arcs_lsize[ARC_BUFC_METADATA], 0, "size of metadata in mfu state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_data_lsize, CTLFLAG_RD, &ARC_mfu.arcs_lsize[ARC_BUFC_DATA], 0, "size of data in mfu state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_size, CTLFLAG_RD, &ARC_mfu_ghost.arcs_size, 0, "size of mfu ghost state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_metadata_lsize, CTLFLAG_RD, &ARC_mfu_ghost.arcs_lsize[ARC_BUFC_METADATA], 0, "size of metadata in mfu ghost state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_data_lsize, CTLFLAG_RD, &ARC_mfu_ghost.arcs_lsize[ARC_BUFC_DATA], 0, "size of data in mfu ghost state"); SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2c_only_size, CTLFLAG_RD, &ARC_l2c_only.arcs_size, 0, "size of mru state"); /* * L2ARC Internals */ typedef struct l2arc_dev { vdev_t *l2ad_vdev; /* vdev */ spa_t *l2ad_spa; /* spa */ uint64_t l2ad_hand; /* next write location */ uint64_t l2ad_start; /* first addr on device */ uint64_t l2ad_end; /* last addr on device */ uint64_t l2ad_evict; /* last addr eviction reached */ boolean_t l2ad_first; /* first sweep through */ boolean_t l2ad_writing; /* currently writing */ list_t *l2ad_buflist; /* buffer list */ list_node_t l2ad_node; /* device list node */ } l2arc_dev_t; static list_t L2ARC_dev_list; /* device list */ static list_t *l2arc_dev_list; /* device list pointer */ static kmutex_t l2arc_dev_mtx; /* device list mutex */ static l2arc_dev_t *l2arc_dev_last; /* last device used */ static kmutex_t l2arc_buflist_mtx; /* mutex for all buflists */ static list_t L2ARC_free_on_write; /* free after write buf list */ static list_t *l2arc_free_on_write; /* free after write list ptr */ static kmutex_t l2arc_free_on_write_mtx; /* mutex for list */ static uint64_t l2arc_ndev; /* number of devices */ typedef struct l2arc_read_callback { arc_buf_t *l2rcb_buf; /* read buffer */ spa_t *l2rcb_spa; /* spa */ blkptr_t l2rcb_bp; /* original blkptr */ zbookmark_t l2rcb_zb; /* original bookmark */ int l2rcb_flags; /* original flags */ enum zio_compress l2rcb_compress; /* applied compress */ } l2arc_read_callback_t; typedef struct l2arc_write_callback { l2arc_dev_t *l2wcb_dev; /* device info */ arc_buf_hdr_t *l2wcb_head; /* head of write buflist */ } l2arc_write_callback_t; struct l2arc_buf_hdr { /* protected by arc_buf_hdr mutex */ l2arc_dev_t *b_dev; /* L2ARC device */ uint64_t b_daddr; /* disk address, offset byte */ /* compression applied to buffer data */ enum zio_compress b_compress; /* real alloc'd buffer size depending on b_compress applied */ int b_asize; /* temporary buffer holder for in-flight compressed data */ void *b_tmp_cdata; }; typedef struct l2arc_data_free { /* protected by l2arc_free_on_write_mtx */ void *l2df_data; size_t l2df_size; void (*l2df_func)(void *, size_t); list_node_t l2df_list_node; } l2arc_data_free_t; static kmutex_t l2arc_feed_thr_lock; static kcondvar_t l2arc_feed_thr_cv; static uint8_t l2arc_thread_exit; static void l2arc_read_done(zio_t *zio); static void l2arc_hdr_stat_add(void); static void l2arc_hdr_stat_remove(void); static boolean_t l2arc_compress_buf(l2arc_buf_hdr_t *l2hdr); static void l2arc_decompress_zio(zio_t *zio, arc_buf_hdr_t *hdr, enum zio_compress c); static void l2arc_release_cdata_buf(arc_buf_hdr_t *ab); static void l2arc_trim(const l2arc_buf_hdr_t *l2hdr) { ASSERT(MUTEX_HELD(&l2arc_buflist_mtx)); if (l2hdr->b_asize != 0) { trim_map_free(l2hdr->b_dev->l2ad_vdev, l2hdr->b_daddr, l2hdr->b_asize, 0); } else { ASSERT3U(l2hdr->b_compress, ==, ZIO_COMPRESS_EMPTY); } } static uint64_t buf_hash(uint64_t spa, const dva_t *dva, uint64_t birth) { uint8_t *vdva = (uint8_t *)dva; uint64_t crc = -1ULL; int i; ASSERT(zfs_crc64_table[128] == ZFS_CRC64_POLY); for (i = 0; i < sizeof (dva_t); i++) crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ vdva[i]) & 0xFF]; crc ^= (spa>>8) ^ birth; return (crc); } #define BUF_EMPTY(buf) \ ((buf)->b_dva.dva_word[0] == 0 && \ (buf)->b_dva.dva_word[1] == 0 && \ (buf)->b_cksum0 == 0) #define BUF_EQUAL(spa, dva, birth, buf) \ ((buf)->b_dva.dva_word[0] == (dva)->dva_word[0]) && \ ((buf)->b_dva.dva_word[1] == (dva)->dva_word[1]) && \ ((buf)->b_birth == birth) && ((buf)->b_spa == spa) static void buf_discard_identity(arc_buf_hdr_t *hdr) { hdr->b_dva.dva_word[0] = 0; hdr->b_dva.dva_word[1] = 0; hdr->b_birth = 0; hdr->b_cksum0 = 0; } static arc_buf_hdr_t * buf_hash_find(uint64_t spa, const dva_t *dva, uint64_t birth, kmutex_t **lockp) { uint64_t idx = BUF_HASH_INDEX(spa, dva, birth); kmutex_t *hash_lock = BUF_HASH_LOCK(idx); arc_buf_hdr_t *buf; mutex_enter(hash_lock); for (buf = buf_hash_table.ht_table[idx]; buf != NULL; buf = buf->b_hash_next) { if (BUF_EQUAL(spa, dva, birth, buf)) { *lockp = hash_lock; return (buf); } } mutex_exit(hash_lock); *lockp = NULL; return (NULL); } /* * Insert an entry into the hash table. If there is already an element * equal to elem in the hash table, then the already existing element * will be returned and the new element will not be inserted. * Otherwise returns NULL. */ static arc_buf_hdr_t * buf_hash_insert(arc_buf_hdr_t *buf, kmutex_t **lockp) { uint64_t idx = BUF_HASH_INDEX(buf->b_spa, &buf->b_dva, buf->b_birth); kmutex_t *hash_lock = BUF_HASH_LOCK(idx); arc_buf_hdr_t *fbuf; uint32_t i; ASSERT(!HDR_IN_HASH_TABLE(buf)); *lockp = hash_lock; mutex_enter(hash_lock); for (fbuf = buf_hash_table.ht_table[idx], i = 0; fbuf != NULL; fbuf = fbuf->b_hash_next, i++) { if (BUF_EQUAL(buf->b_spa, &buf->b_dva, buf->b_birth, fbuf)) return (fbuf); } buf->b_hash_next = buf_hash_table.ht_table[idx]; buf_hash_table.ht_table[idx] = buf; buf->b_flags |= ARC_IN_HASH_TABLE; /* collect some hash table performance data */ if (i > 0) { ARCSTAT_BUMP(arcstat_hash_collisions); if (i == 1) ARCSTAT_BUMP(arcstat_hash_chains); ARCSTAT_MAX(arcstat_hash_chain_max, i); } ARCSTAT_BUMP(arcstat_hash_elements); ARCSTAT_MAXSTAT(arcstat_hash_elements); return (NULL); } static void buf_hash_remove(arc_buf_hdr_t *buf) { arc_buf_hdr_t *fbuf, **bufp; uint64_t idx = BUF_HASH_INDEX(buf->b_spa, &buf->b_dva, buf->b_birth); ASSERT(MUTEX_HELD(BUF_HASH_LOCK(idx))); ASSERT(HDR_IN_HASH_TABLE(buf)); bufp = &buf_hash_table.ht_table[idx]; while ((fbuf = *bufp) != buf) { ASSERT(fbuf != NULL); bufp = &fbuf->b_hash_next; } *bufp = buf->b_hash_next; buf->b_hash_next = NULL; buf->b_flags &= ~ARC_IN_HASH_TABLE; /* collect some hash table performance data */ ARCSTAT_BUMPDOWN(arcstat_hash_elements); if (buf_hash_table.ht_table[idx] && buf_hash_table.ht_table[idx]->b_hash_next == NULL) ARCSTAT_BUMPDOWN(arcstat_hash_chains); } /* * Global data structures and functions for the buf kmem cache. */ static kmem_cache_t *hdr_cache; static kmem_cache_t *buf_cache; static void buf_fini(void) { int i; kmem_free(buf_hash_table.ht_table, (buf_hash_table.ht_mask + 1) * sizeof (void *)); for (i = 0; i < BUF_LOCKS; i++) mutex_destroy(&buf_hash_table.ht_locks[i].ht_lock); kmem_cache_destroy(hdr_cache); kmem_cache_destroy(buf_cache); } /* * Constructor callback - called when the cache is empty * and a new buf is requested. */ /* ARGSUSED */ static int hdr_cons(void *vbuf, void *unused, int kmflag) { arc_buf_hdr_t *buf = vbuf; bzero(buf, sizeof (arc_buf_hdr_t)); refcount_create(&buf->b_refcnt); cv_init(&buf->b_cv, NULL, CV_DEFAULT, NULL); mutex_init(&buf->b_freeze_lock, NULL, MUTEX_DEFAULT, NULL); arc_space_consume(sizeof (arc_buf_hdr_t), ARC_SPACE_HDRS); return (0); } /* ARGSUSED */ static int buf_cons(void *vbuf, void *unused, int kmflag) { arc_buf_t *buf = vbuf; bzero(buf, sizeof (arc_buf_t)); mutex_init(&buf->b_evict_lock, NULL, MUTEX_DEFAULT, NULL); arc_space_consume(sizeof (arc_buf_t), ARC_SPACE_HDRS); return (0); } /* * Destructor callback - called when a cached buf is * no longer required. */ /* ARGSUSED */ static void hdr_dest(void *vbuf, void *unused) { arc_buf_hdr_t *buf = vbuf; ASSERT(BUF_EMPTY(buf)); refcount_destroy(&buf->b_refcnt); cv_destroy(&buf->b_cv); mutex_destroy(&buf->b_freeze_lock); arc_space_return(sizeof (arc_buf_hdr_t), ARC_SPACE_HDRS); } /* ARGSUSED */ static void buf_dest(void *vbuf, void *unused) { arc_buf_t *buf = vbuf; mutex_destroy(&buf->b_evict_lock); arc_space_return(sizeof (arc_buf_t), ARC_SPACE_HDRS); } /* * Reclaim callback -- invoked when memory is low. */ /* ARGSUSED */ static void hdr_recl(void *unused) { dprintf("hdr_recl called\n"); /* * umem calls the reclaim func when we destroy the buf cache, * which is after we do arc_fini(). */ if (!arc_dead) cv_signal(&arc_reclaim_thr_cv); } static void buf_init(void) { uint64_t *ct; uint64_t hsize = 1ULL << 12; int i, j; /* * The hash table is big enough to fill all of physical memory * with an average 64K block size. The table will take up * totalmem*sizeof(void*)/64K (eg. 128KB/GB with 8-byte pointers). */ while (hsize * 65536 < (uint64_t)physmem * PAGESIZE) hsize <<= 1; retry: buf_hash_table.ht_mask = hsize - 1; buf_hash_table.ht_table = kmem_zalloc(hsize * sizeof (void*), KM_NOSLEEP); if (buf_hash_table.ht_table == NULL) { ASSERT(hsize > (1ULL << 8)); hsize >>= 1; goto retry; } hdr_cache = kmem_cache_create("arc_buf_hdr_t", sizeof (arc_buf_hdr_t), 0, hdr_cons, hdr_dest, hdr_recl, NULL, NULL, 0); buf_cache = kmem_cache_create("arc_buf_t", sizeof (arc_buf_t), 0, buf_cons, buf_dest, NULL, NULL, NULL, 0); for (i = 0; i < 256; i++) for (ct = zfs_crc64_table + i, *ct = i, j = 8; j > 0; j--) *ct = (*ct >> 1) ^ (-(*ct & 1) & ZFS_CRC64_POLY); for (i = 0; i < BUF_LOCKS; i++) { mutex_init(&buf_hash_table.ht_locks[i].ht_lock, NULL, MUTEX_DEFAULT, NULL); } } #define ARC_MINTIME (hz>>4) /* 62 ms */ static void arc_cksum_verify(arc_buf_t *buf) { zio_cksum_t zc; if (!(zfs_flags & ZFS_DEBUG_MODIFY)) return; mutex_enter(&buf->b_hdr->b_freeze_lock); if (buf->b_hdr->b_freeze_cksum == NULL || (buf->b_hdr->b_flags & ARC_IO_ERROR)) { mutex_exit(&buf->b_hdr->b_freeze_lock); return; } fletcher_2_native(buf->b_data, buf->b_hdr->b_size, &zc); if (!ZIO_CHECKSUM_EQUAL(*buf->b_hdr->b_freeze_cksum, zc)) panic("buffer modified while frozen!"); mutex_exit(&buf->b_hdr->b_freeze_lock); } static int arc_cksum_equal(arc_buf_t *buf) { zio_cksum_t zc; int equal; mutex_enter(&buf->b_hdr->b_freeze_lock); fletcher_2_native(buf->b_data, buf->b_hdr->b_size, &zc); equal = ZIO_CHECKSUM_EQUAL(*buf->b_hdr->b_freeze_cksum, zc); mutex_exit(&buf->b_hdr->b_freeze_lock); return (equal); } static void arc_cksum_compute(arc_buf_t *buf, boolean_t force) { if (!force && !(zfs_flags & ZFS_DEBUG_MODIFY)) return; mutex_enter(&buf->b_hdr->b_freeze_lock); if (buf->b_hdr->b_freeze_cksum != NULL) { mutex_exit(&buf->b_hdr->b_freeze_lock); return; } buf->b_hdr->b_freeze_cksum = kmem_alloc(sizeof (zio_cksum_t), KM_SLEEP); fletcher_2_native(buf->b_data, buf->b_hdr->b_size, buf->b_hdr->b_freeze_cksum); mutex_exit(&buf->b_hdr->b_freeze_lock); #ifdef illumos arc_buf_watch(buf); #endif /* illumos */ } #ifdef illumos #ifndef _KERNEL typedef struct procctl { long cmd; prwatch_t prwatch; } procctl_t; #endif /* ARGSUSED */ static void arc_buf_unwatch(arc_buf_t *buf) { #ifndef _KERNEL if (arc_watch) { int result; procctl_t ctl; ctl.cmd = PCWATCH; ctl.prwatch.pr_vaddr = (uintptr_t)buf->b_data; ctl.prwatch.pr_size = 0; ctl.prwatch.pr_wflags = 0; result = write(arc_procfd, &ctl, sizeof (ctl)); ASSERT3U(result, ==, sizeof (ctl)); } #endif } /* ARGSUSED */ static void arc_buf_watch(arc_buf_t *buf) { #ifndef _KERNEL if (arc_watch) { int result; procctl_t ctl; ctl.cmd = PCWATCH; ctl.prwatch.pr_vaddr = (uintptr_t)buf->b_data; ctl.prwatch.pr_size = buf->b_hdr->b_size; ctl.prwatch.pr_wflags = WA_WRITE; result = write(arc_procfd, &ctl, sizeof (ctl)); ASSERT3U(result, ==, sizeof (ctl)); } #endif } #endif /* illumos */ void arc_buf_thaw(arc_buf_t *buf) { if (zfs_flags & ZFS_DEBUG_MODIFY) { if (buf->b_hdr->b_state != arc_anon) panic("modifying non-anon buffer!"); if (buf->b_hdr->b_flags & ARC_IO_IN_PROGRESS) panic("modifying buffer while i/o in progress!"); arc_cksum_verify(buf); } mutex_enter(&buf->b_hdr->b_freeze_lock); if (buf->b_hdr->b_freeze_cksum != NULL) { kmem_free(buf->b_hdr->b_freeze_cksum, sizeof (zio_cksum_t)); buf->b_hdr->b_freeze_cksum = NULL; } if (zfs_flags & ZFS_DEBUG_MODIFY) { if (buf->b_hdr->b_thawed) kmem_free(buf->b_hdr->b_thawed, 1); buf->b_hdr->b_thawed = kmem_alloc(1, KM_SLEEP); } mutex_exit(&buf->b_hdr->b_freeze_lock); #ifdef illumos arc_buf_unwatch(buf); #endif /* illumos */ } void arc_buf_freeze(arc_buf_t *buf) { kmutex_t *hash_lock; if (!(zfs_flags & ZFS_DEBUG_MODIFY)) return; hash_lock = HDR_LOCK(buf->b_hdr); mutex_enter(hash_lock); ASSERT(buf->b_hdr->b_freeze_cksum != NULL || buf->b_hdr->b_state == arc_anon); arc_cksum_compute(buf, B_FALSE); mutex_exit(hash_lock); } static void get_buf_info(arc_buf_hdr_t *ab, arc_state_t *state, list_t **list, kmutex_t **lock) { uint64_t buf_hashid = buf_hash(ab->b_spa, &ab->b_dva, ab->b_birth); if (ab->b_type == ARC_BUFC_METADATA) buf_hashid &= (ARC_BUFC_NUMMETADATALISTS - 1); else { buf_hashid &= (ARC_BUFC_NUMDATALISTS - 1); buf_hashid += ARC_BUFC_NUMMETADATALISTS; } *list = &state->arcs_lists[buf_hashid]; *lock = ARCS_LOCK(state, buf_hashid); } static void add_reference(arc_buf_hdr_t *ab, kmutex_t *hash_lock, void *tag) { ASSERT(MUTEX_HELD(hash_lock)); if ((refcount_add(&ab->b_refcnt, tag) == 1) && (ab->b_state != arc_anon)) { uint64_t delta = ab->b_size * ab->b_datacnt; uint64_t *size = &ab->b_state->arcs_lsize[ab->b_type]; list_t *list; kmutex_t *lock; get_buf_info(ab, ab->b_state, &list, &lock); ASSERT(!MUTEX_HELD(lock)); mutex_enter(lock); ASSERT(list_link_active(&ab->b_arc_node)); list_remove(list, ab); if (GHOST_STATE(ab->b_state)) { ASSERT0(ab->b_datacnt); ASSERT3P(ab->b_buf, ==, NULL); delta = ab->b_size; } ASSERT(delta > 0); ASSERT3U(*size, >=, delta); atomic_add_64(size, -delta); mutex_exit(lock); /* remove the prefetch flag if we get a reference */ if (ab->b_flags & ARC_PREFETCH) ab->b_flags &= ~ARC_PREFETCH; } } static int remove_reference(arc_buf_hdr_t *ab, kmutex_t *hash_lock, void *tag) { int cnt; arc_state_t *state = ab->b_state; ASSERT(state == arc_anon || MUTEX_HELD(hash_lock)); ASSERT(!GHOST_STATE(state)); if (((cnt = refcount_remove(&ab->b_refcnt, tag)) == 0) && (state != arc_anon)) { uint64_t *size = &state->arcs_lsize[ab->b_type]; list_t *list; kmutex_t *lock; get_buf_info(ab, state, &list, &lock); ASSERT(!MUTEX_HELD(lock)); mutex_enter(lock); ASSERT(!list_link_active(&ab->b_arc_node)); list_insert_head(list, ab); ASSERT(ab->b_datacnt > 0); atomic_add_64(size, ab->b_size * ab->b_datacnt); mutex_exit(lock); } return (cnt); } /* * Move the supplied buffer to the indicated state. The mutex * for the buffer must be held by the caller. */ static void arc_change_state(arc_state_t *new_state, arc_buf_hdr_t *ab, kmutex_t *hash_lock) { arc_state_t *old_state = ab->b_state; int64_t refcnt = refcount_count(&ab->b_refcnt); uint64_t from_delta, to_delta; list_t *list; kmutex_t *lock; ASSERT(MUTEX_HELD(hash_lock)); ASSERT3P(new_state, !=, old_state); ASSERT(refcnt == 0 || ab->b_datacnt > 0); ASSERT(ab->b_datacnt == 0 || !GHOST_STATE(new_state)); ASSERT(ab->b_datacnt <= 1 || old_state != arc_anon); from_delta = to_delta = ab->b_datacnt * ab->b_size; /* * If this buffer is evictable, transfer it from the * old state list to the new state list. */ if (refcnt == 0) { if (old_state != arc_anon) { int use_mutex; uint64_t *size = &old_state->arcs_lsize[ab->b_type]; get_buf_info(ab, old_state, &list, &lock); use_mutex = !MUTEX_HELD(lock); if (use_mutex) mutex_enter(lock); ASSERT(list_link_active(&ab->b_arc_node)); list_remove(list, ab); /* * If prefetching out of the ghost cache, * we will have a non-zero datacnt. */ if (GHOST_STATE(old_state) && ab->b_datacnt == 0) { /* ghost elements have a ghost size */ ASSERT(ab->b_buf == NULL); from_delta = ab->b_size; } ASSERT3U(*size, >=, from_delta); atomic_add_64(size, -from_delta); if (use_mutex) mutex_exit(lock); } if (new_state != arc_anon) { int use_mutex; uint64_t *size = &new_state->arcs_lsize[ab->b_type]; get_buf_info(ab, new_state, &list, &lock); use_mutex = !MUTEX_HELD(lock); if (use_mutex) mutex_enter(lock); list_insert_head(list, ab); /* ghost elements have a ghost size */ if (GHOST_STATE(new_state)) { ASSERT(ab->b_datacnt == 0); ASSERT(ab->b_buf == NULL); to_delta = ab->b_size; } atomic_add_64(size, to_delta); if (use_mutex) mutex_exit(lock); } } ASSERT(!BUF_EMPTY(ab)); if (new_state == arc_anon && HDR_IN_HASH_TABLE(ab)) buf_hash_remove(ab); /* adjust state sizes */ if (to_delta) atomic_add_64(&new_state->arcs_size, to_delta); if (from_delta) { ASSERT3U(old_state->arcs_size, >=, from_delta); atomic_add_64(&old_state->arcs_size, -from_delta); } ab->b_state = new_state; /* adjust l2arc hdr stats */ if (new_state == arc_l2c_only) l2arc_hdr_stat_add(); else if (old_state == arc_l2c_only) l2arc_hdr_stat_remove(); } void arc_space_consume(uint64_t space, arc_space_type_t type) { ASSERT(type >= 0 && type < ARC_SPACE_NUMTYPES); switch (type) { case ARC_SPACE_DATA: ARCSTAT_INCR(arcstat_data_size, space); break; case ARC_SPACE_OTHER: ARCSTAT_INCR(arcstat_other_size, space); break; case ARC_SPACE_HDRS: ARCSTAT_INCR(arcstat_hdr_size, space); break; case ARC_SPACE_L2HDRS: ARCSTAT_INCR(arcstat_l2_hdr_size, space); break; } atomic_add_64(&arc_meta_used, space); atomic_add_64(&arc_size, space); } void arc_space_return(uint64_t space, arc_space_type_t type) { ASSERT(type >= 0 && type < ARC_SPACE_NUMTYPES); switch (type) { case ARC_SPACE_DATA: ARCSTAT_INCR(arcstat_data_size, -space); break; case ARC_SPACE_OTHER: ARCSTAT_INCR(arcstat_other_size, -space); break; case ARC_SPACE_HDRS: ARCSTAT_INCR(arcstat_hdr_size, -space); break; case ARC_SPACE_L2HDRS: ARCSTAT_INCR(arcstat_l2_hdr_size, -space); break; } ASSERT(arc_meta_used >= space); if (arc_meta_max < arc_meta_used) arc_meta_max = arc_meta_used; atomic_add_64(&arc_meta_used, -space); ASSERT(arc_size >= space); atomic_add_64(&arc_size, -space); } void * arc_data_buf_alloc(uint64_t size) { if (arc_evict_needed(ARC_BUFC_DATA)) cv_signal(&arc_reclaim_thr_cv); atomic_add_64(&arc_size, size); return (zio_data_buf_alloc(size)); } void arc_data_buf_free(void *buf, uint64_t size) { zio_data_buf_free(buf, size); ASSERT(arc_size >= size); atomic_add_64(&arc_size, -size); } arc_buf_t * arc_buf_alloc(spa_t *spa, int size, void *tag, arc_buf_contents_t type) { arc_buf_hdr_t *hdr; arc_buf_t *buf; ASSERT3U(size, >, 0); hdr = kmem_cache_alloc(hdr_cache, KM_PUSHPAGE); ASSERT(BUF_EMPTY(hdr)); hdr->b_size = size; hdr->b_type = type; hdr->b_spa = spa_load_guid(spa); hdr->b_state = arc_anon; hdr->b_arc_access = 0; buf = kmem_cache_alloc(buf_cache, KM_PUSHPAGE); buf->b_hdr = hdr; buf->b_data = NULL; buf->b_efunc = NULL; buf->b_private = NULL; buf->b_next = NULL; hdr->b_buf = buf; arc_get_data_buf(buf); hdr->b_datacnt = 1; hdr->b_flags = 0; ASSERT(refcount_is_zero(&hdr->b_refcnt)); (void) refcount_add(&hdr->b_refcnt, tag); return (buf); } static char *arc_onloan_tag = "onloan"; /* * Loan out an anonymous arc buffer. Loaned buffers are not counted as in * flight data by arc_tempreserve_space() until they are "returned". Loaned * buffers must be returned to the arc before they can be used by the DMU or * freed. */ arc_buf_t * arc_loan_buf(spa_t *spa, int size) { arc_buf_t *buf; buf = arc_buf_alloc(spa, size, arc_onloan_tag, ARC_BUFC_DATA); atomic_add_64(&arc_loaned_bytes, size); return (buf); } /* * Return a loaned arc buffer to the arc. */ void arc_return_buf(arc_buf_t *buf, void *tag) { arc_buf_hdr_t *hdr = buf->b_hdr; ASSERT(buf->b_data != NULL); (void) refcount_add(&hdr->b_refcnt, tag); (void) refcount_remove(&hdr->b_refcnt, arc_onloan_tag); atomic_add_64(&arc_loaned_bytes, -hdr->b_size); } /* Detach an arc_buf from a dbuf (tag) */ void arc_loan_inuse_buf(arc_buf_t *buf, void *tag) { arc_buf_hdr_t *hdr; ASSERT(buf->b_data != NULL); hdr = buf->b_hdr; (void) refcount_add(&hdr->b_refcnt, arc_onloan_tag); (void) refcount_remove(&hdr->b_refcnt, tag); buf->b_efunc = NULL; buf->b_private = NULL; atomic_add_64(&arc_loaned_bytes, hdr->b_size); } static arc_buf_t * arc_buf_clone(arc_buf_t *from) { arc_buf_t *buf; arc_buf_hdr_t *hdr = from->b_hdr; uint64_t size = hdr->b_size; ASSERT(hdr->b_state != arc_anon); buf = kmem_cache_alloc(buf_cache, KM_PUSHPAGE); buf->b_hdr = hdr; buf->b_data = NULL; buf->b_efunc = NULL; buf->b_private = NULL; buf->b_next = hdr->b_buf; hdr->b_buf = buf; arc_get_data_buf(buf); bcopy(from->b_data, buf->b_data, size); /* * This buffer already exists in the arc so create a duplicate * copy for the caller. If the buffer is associated with user data * then track the size and number of duplicates. These stats will be * updated as duplicate buffers are created and destroyed. */ if (hdr->b_type == ARC_BUFC_DATA) { ARCSTAT_BUMP(arcstat_duplicate_buffers); ARCSTAT_INCR(arcstat_duplicate_buffers_size, size); } hdr->b_datacnt += 1; return (buf); } void arc_buf_add_ref(arc_buf_t *buf, void* tag) { arc_buf_hdr_t *hdr; kmutex_t *hash_lock; /* * Check to see if this buffer is evicted. Callers * must verify b_data != NULL to know if the add_ref * was successful. */ mutex_enter(&buf->b_evict_lock); if (buf->b_data == NULL) { mutex_exit(&buf->b_evict_lock); return; } hash_lock = HDR_LOCK(buf->b_hdr); mutex_enter(hash_lock); hdr = buf->b_hdr; ASSERT3P(hash_lock, ==, HDR_LOCK(hdr)); mutex_exit(&buf->b_evict_lock); ASSERT(hdr->b_state == arc_mru || hdr->b_state == arc_mfu); add_reference(hdr, hash_lock, tag); DTRACE_PROBE1(arc__hit, arc_buf_hdr_t *, hdr); arc_access(hdr, hash_lock); mutex_exit(hash_lock); ARCSTAT_BUMP(arcstat_hits); ARCSTAT_CONDSTAT(!(hdr->b_flags & ARC_PREFETCH), demand, prefetch, hdr->b_type != ARC_BUFC_METADATA, data, metadata, hits); } /* * Free the arc data buffer. If it is an l2arc write in progress, * the buffer is placed on l2arc_free_on_write to be freed later. */ static void arc_buf_data_free(arc_buf_t *buf, void (*free_func)(void *, size_t)) { arc_buf_hdr_t *hdr = buf->b_hdr; if (HDR_L2_WRITING(hdr)) { l2arc_data_free_t *df; df = kmem_alloc(sizeof (l2arc_data_free_t), KM_SLEEP); df->l2df_data = buf->b_data; df->l2df_size = hdr->b_size; df->l2df_func = free_func; mutex_enter(&l2arc_free_on_write_mtx); list_insert_head(l2arc_free_on_write, df); mutex_exit(&l2arc_free_on_write_mtx); ARCSTAT_BUMP(arcstat_l2_free_on_write); } else { free_func(buf->b_data, hdr->b_size); } } static void arc_buf_destroy(arc_buf_t *buf, boolean_t recycle, boolean_t all) { arc_buf_t **bufp; /* free up data associated with the buf */ if (buf->b_data) { arc_state_t *state = buf->b_hdr->b_state; uint64_t size = buf->b_hdr->b_size; arc_buf_contents_t type = buf->b_hdr->b_type; arc_cksum_verify(buf); #ifdef illumos arc_buf_unwatch(buf); #endif /* illumos */ if (!recycle) { if (type == ARC_BUFC_METADATA) { arc_buf_data_free(buf, zio_buf_free); arc_space_return(size, ARC_SPACE_DATA); } else { ASSERT(type == ARC_BUFC_DATA); arc_buf_data_free(buf, zio_data_buf_free); ARCSTAT_INCR(arcstat_data_size, -size); atomic_add_64(&arc_size, -size); } } if (list_link_active(&buf->b_hdr->b_arc_node)) { uint64_t *cnt = &state->arcs_lsize[type]; ASSERT(refcount_is_zero(&buf->b_hdr->b_refcnt)); ASSERT(state != arc_anon); ASSERT3U(*cnt, >=, size); atomic_add_64(cnt, -size); } ASSERT3U(state->arcs_size, >=, size); atomic_add_64(&state->arcs_size, -size); buf->b_data = NULL; /* * If we're destroying a duplicate buffer make sure * that the appropriate statistics are updated. */ if (buf->b_hdr->b_datacnt > 1 && buf->b_hdr->b_type == ARC_BUFC_DATA) { ARCSTAT_BUMPDOWN(arcstat_duplicate_buffers); ARCSTAT_INCR(arcstat_duplicate_buffers_size, -size); } ASSERT(buf->b_hdr->b_datacnt > 0); buf->b_hdr->b_datacnt -= 1; } /* only remove the buf if requested */ if (!all) return; /* remove the buf from the hdr list */ for (bufp = &buf->b_hdr->b_buf; *bufp != buf; bufp = &(*bufp)->b_next) continue; *bufp = buf->b_next; buf->b_next = NULL; ASSERT(buf->b_efunc == NULL); /* clean up the buf */ buf->b_hdr = NULL; kmem_cache_free(buf_cache, buf); } static void arc_hdr_destroy(arc_buf_hdr_t *hdr) { ASSERT(refcount_is_zero(&hdr->b_refcnt)); ASSERT3P(hdr->b_state, ==, arc_anon); ASSERT(!HDR_IO_IN_PROGRESS(hdr)); l2arc_buf_hdr_t *l2hdr = hdr->b_l2hdr; if (l2hdr != NULL) { boolean_t buflist_held = MUTEX_HELD(&l2arc_buflist_mtx); /* * To prevent arc_free() and l2arc_evict() from * attempting to free the same buffer at the same time, * a FREE_IN_PROGRESS flag is given to arc_free() to * give it priority. l2arc_evict() can't destroy this * header while we are waiting on l2arc_buflist_mtx. * * The hdr may be removed from l2ad_buflist before we * grab l2arc_buflist_mtx, so b_l2hdr is rechecked. */ if (!buflist_held) { mutex_enter(&l2arc_buflist_mtx); l2hdr = hdr->b_l2hdr; } if (l2hdr != NULL) { l2arc_trim(l2hdr); list_remove(l2hdr->b_dev->l2ad_buflist, hdr); ARCSTAT_INCR(arcstat_l2_size, -hdr->b_size); ARCSTAT_INCR(arcstat_l2_asize, -l2hdr->b_asize); kmem_free(l2hdr, sizeof (l2arc_buf_hdr_t)); if (hdr->b_state == arc_l2c_only) l2arc_hdr_stat_remove(); hdr->b_l2hdr = NULL; } if (!buflist_held) mutex_exit(&l2arc_buflist_mtx); } if (!BUF_EMPTY(hdr)) { ASSERT(!HDR_IN_HASH_TABLE(hdr)); buf_discard_identity(hdr); } while (hdr->b_buf) { arc_buf_t *buf = hdr->b_buf; if (buf->b_efunc) { mutex_enter(&arc_eviction_mtx); mutex_enter(&buf->b_evict_lock); ASSERT(buf->b_hdr != NULL); arc_buf_destroy(hdr->b_buf, FALSE, FALSE); hdr->b_buf = buf->b_next; buf->b_hdr = &arc_eviction_hdr; buf->b_next = arc_eviction_list; arc_eviction_list = buf; mutex_exit(&buf->b_evict_lock); mutex_exit(&arc_eviction_mtx); } else { arc_buf_destroy(hdr->b_buf, FALSE, TRUE); } } if (hdr->b_freeze_cksum != NULL) { kmem_free(hdr->b_freeze_cksum, sizeof (zio_cksum_t)); hdr->b_freeze_cksum = NULL; } if (hdr->b_thawed) { kmem_free(hdr->b_thawed, 1); hdr->b_thawed = NULL; } ASSERT(!list_link_active(&hdr->b_arc_node)); ASSERT3P(hdr->b_hash_next, ==, NULL); ASSERT3P(hdr->b_acb, ==, NULL); kmem_cache_free(hdr_cache, hdr); } void arc_buf_free(arc_buf_t *buf, void *tag) { arc_buf_hdr_t *hdr = buf->b_hdr; int hashed = hdr->b_state != arc_anon; ASSERT(buf->b_efunc == NULL); ASSERT(buf->b_data != NULL); if (hashed) { kmutex_t *hash_lock = HDR_LOCK(hdr); mutex_enter(hash_lock); hdr = buf->b_hdr; ASSERT3P(hash_lock, ==, HDR_LOCK(hdr)); (void) remove_reference(hdr, hash_lock, tag); if (hdr->b_datacnt > 1) { arc_buf_destroy(buf, FALSE, TRUE); } else { ASSERT(buf == hdr->b_buf); ASSERT(buf->b_efunc == NULL); hdr->b_flags |= ARC_BUF_AVAILABLE; } mutex_exit(hash_lock); } else if (HDR_IO_IN_PROGRESS(hdr)) { int destroy_hdr; /* * We are in the middle of an async write. Don't destroy * this buffer unless the write completes before we finish * decrementing the reference count. */ mutex_enter(&arc_eviction_mtx); (void) remove_reference(hdr, NULL, tag); ASSERT(refcount_is_zero(&hdr->b_refcnt)); destroy_hdr = !HDR_IO_IN_PROGRESS(hdr); mutex_exit(&arc_eviction_mtx); if (destroy_hdr) arc_hdr_destroy(hdr); } else { if (remove_reference(hdr, NULL, tag) > 0) arc_buf_destroy(buf, FALSE, TRUE); else arc_hdr_destroy(hdr); } } boolean_t arc_buf_remove_ref(arc_buf_t *buf, void* tag) { arc_buf_hdr_t *hdr = buf->b_hdr; kmutex_t *hash_lock = HDR_LOCK(hdr); boolean_t no_callback = (buf->b_efunc == NULL); if (hdr->b_state == arc_anon) { ASSERT(hdr->b_datacnt == 1); arc_buf_free(buf, tag); return (no_callback); } mutex_enter(hash_lock); hdr = buf->b_hdr; ASSERT3P(hash_lock, ==, HDR_LOCK(hdr)); ASSERT(hdr->b_state != arc_anon); ASSERT(buf->b_data != NULL); (void) remove_reference(hdr, hash_lock, tag); if (hdr->b_datacnt > 1) { if (no_callback) arc_buf_destroy(buf, FALSE, TRUE); } else if (no_callback) { ASSERT(hdr->b_buf == buf && buf->b_next == NULL); ASSERT(buf->b_efunc == NULL); hdr->b_flags |= ARC_BUF_AVAILABLE; } ASSERT(no_callback || hdr->b_datacnt > 1 || refcount_is_zero(&hdr->b_refcnt)); mutex_exit(hash_lock); return (no_callback); } int arc_buf_size(arc_buf_t *buf) { return (buf->b_hdr->b_size); } /* * Called from the DMU to determine if the current buffer should be * evicted. In order to ensure proper locking, the eviction must be initiated * from the DMU. Return true if the buffer is associated with user data and * duplicate buffers still exist. */ boolean_t arc_buf_eviction_needed(arc_buf_t *buf) { arc_buf_hdr_t *hdr; boolean_t evict_needed = B_FALSE; if (zfs_disable_dup_eviction) return (B_FALSE); mutex_enter(&buf->b_evict_lock); hdr = buf->b_hdr; if (hdr == NULL) { /* * We are in arc_do_user_evicts(); let that function * perform the eviction. */ ASSERT(buf->b_data == NULL); mutex_exit(&buf->b_evict_lock); return (B_FALSE); } else if (buf->b_data == NULL) { /* * We have already been added to the arc eviction list; * recommend eviction. */ ASSERT3P(hdr, ==, &arc_eviction_hdr); mutex_exit(&buf->b_evict_lock); return (B_TRUE); } if (hdr->b_datacnt > 1 && hdr->b_type == ARC_BUFC_DATA) evict_needed = B_TRUE; mutex_exit(&buf->b_evict_lock); return (evict_needed); } /* * Evict buffers from list until we've removed the specified number of * bytes. Move the removed buffers to the appropriate evict state. * If the recycle flag is set, then attempt to "recycle" a buffer: * - look for a buffer to evict that is `bytes' long. * - return the data block from this buffer rather than freeing it. * This flag is used by callers that are trying to make space for a * new buffer in a full arc cache. * * This function makes a "best effort". It skips over any buffers * it can't get a hash_lock on, and so may not catch all candidates. * It may also return without evicting as much space as requested. */ static void * arc_evict(arc_state_t *state, uint64_t spa, int64_t bytes, boolean_t recycle, arc_buf_contents_t type) { arc_state_t *evicted_state; uint64_t bytes_evicted = 0, skipped = 0, missed = 0; int64_t bytes_remaining; arc_buf_hdr_t *ab, *ab_prev = NULL; list_t *evicted_list, *list, *evicted_list_start, *list_start; kmutex_t *lock, *evicted_lock; kmutex_t *hash_lock; boolean_t have_lock; void *stolen = NULL; arc_buf_hdr_t marker = { 0 }; int count = 0; static int evict_metadata_offset, evict_data_offset; int i, idx, offset, list_count, lists; ASSERT(state == arc_mru || state == arc_mfu); evicted_state = (state == arc_mru) ? arc_mru_ghost : arc_mfu_ghost; if (type == ARC_BUFC_METADATA) { offset = 0; list_count = ARC_BUFC_NUMMETADATALISTS; list_start = &state->arcs_lists[0]; evicted_list_start = &evicted_state->arcs_lists[0]; idx = evict_metadata_offset; } else { offset = ARC_BUFC_NUMMETADATALISTS; list_start = &state->arcs_lists[offset]; evicted_list_start = &evicted_state->arcs_lists[offset]; list_count = ARC_BUFC_NUMDATALISTS; idx = evict_data_offset; } bytes_remaining = evicted_state->arcs_lsize[type]; lists = 0; evict_start: list = &list_start[idx]; evicted_list = &evicted_list_start[idx]; lock = ARCS_LOCK(state, (offset + idx)); evicted_lock = ARCS_LOCK(evicted_state, (offset + idx)); mutex_enter(lock); mutex_enter(evicted_lock); for (ab = list_tail(list); ab; ab = ab_prev) { ab_prev = list_prev(list, ab); bytes_remaining -= (ab->b_size * ab->b_datacnt); /* prefetch buffers have a minimum lifespan */ if (HDR_IO_IN_PROGRESS(ab) || (spa && ab->b_spa != spa) || (ab->b_flags & (ARC_PREFETCH|ARC_INDIRECT) && ddi_get_lbolt() - ab->b_arc_access < arc_min_prefetch_lifespan)) { skipped++; continue; } /* "lookahead" for better eviction candidate */ if (recycle && ab->b_size != bytes && ab_prev && ab_prev->b_size == bytes) continue; /* ignore markers */ if (ab->b_spa == 0) continue; /* * It may take a long time to evict all the bufs requested. * To avoid blocking all arc activity, periodically drop * the arcs_mtx and give other threads a chance to run * before reacquiring the lock. * * If we are looking for a buffer to recycle, we are in * the hot code path, so don't sleep. */ if (!recycle && count++ > arc_evict_iterations) { list_insert_after(list, ab, &marker); mutex_exit(evicted_lock); mutex_exit(lock); kpreempt(KPREEMPT_SYNC); mutex_enter(lock); mutex_enter(evicted_lock); ab_prev = list_prev(list, &marker); list_remove(list, &marker); count = 0; continue; } hash_lock = HDR_LOCK(ab); have_lock = MUTEX_HELD(hash_lock); if (have_lock || mutex_tryenter(hash_lock)) { ASSERT0(refcount_count(&ab->b_refcnt)); ASSERT(ab->b_datacnt > 0); while (ab->b_buf) { arc_buf_t *buf = ab->b_buf; if (!mutex_tryenter(&buf->b_evict_lock)) { missed += 1; break; } if (buf->b_data) { bytes_evicted += ab->b_size; if (recycle && ab->b_type == type && ab->b_size == bytes && !HDR_L2_WRITING(ab)) { stolen = buf->b_data; recycle = FALSE; } } if (buf->b_efunc) { mutex_enter(&arc_eviction_mtx); arc_buf_destroy(buf, buf->b_data == stolen, FALSE); ab->b_buf = buf->b_next; buf->b_hdr = &arc_eviction_hdr; buf->b_next = arc_eviction_list; arc_eviction_list = buf; mutex_exit(&arc_eviction_mtx); mutex_exit(&buf->b_evict_lock); } else { mutex_exit(&buf->b_evict_lock); arc_buf_destroy(buf, buf->b_data == stolen, TRUE); } } if (ab->b_l2hdr) { ARCSTAT_INCR(arcstat_evict_l2_cached, ab->b_size); } else { if (l2arc_write_eligible(ab->b_spa, ab)) { ARCSTAT_INCR(arcstat_evict_l2_eligible, ab->b_size); } else { ARCSTAT_INCR( arcstat_evict_l2_ineligible, ab->b_size); } } if (ab->b_datacnt == 0) { arc_change_state(evicted_state, ab, hash_lock); ASSERT(HDR_IN_HASH_TABLE(ab)); ab->b_flags |= ARC_IN_HASH_TABLE; ab->b_flags &= ~ARC_BUF_AVAILABLE; DTRACE_PROBE1(arc__evict, arc_buf_hdr_t *, ab); } if (!have_lock) mutex_exit(hash_lock); if (bytes >= 0 && bytes_evicted >= bytes) break; if (bytes_remaining > 0) { mutex_exit(evicted_lock); mutex_exit(lock); idx = ((idx + 1) & (list_count - 1)); lists++; goto evict_start; } } else { missed += 1; } } mutex_exit(evicted_lock); mutex_exit(lock); idx = ((idx + 1) & (list_count - 1)); lists++; if (bytes_evicted < bytes) { if (lists < list_count) goto evict_start; else dprintf("only evicted %lld bytes from %x", (longlong_t)bytes_evicted, state); } if (type == ARC_BUFC_METADATA) evict_metadata_offset = idx; else evict_data_offset = idx; if (skipped) ARCSTAT_INCR(arcstat_evict_skip, skipped); if (missed) ARCSTAT_INCR(arcstat_mutex_miss, missed); /* * Note: we have just evicted some data into the ghost state, * potentially putting the ghost size over the desired size. Rather * that evicting from the ghost list in this hot code path, leave * this chore to the arc_reclaim_thread(). */ if (stolen) ARCSTAT_BUMP(arcstat_stolen); return (stolen); } /* * Remove buffers from list until we've removed the specified number of * bytes. Destroy the buffers that are removed. */ static void arc_evict_ghost(arc_state_t *state, uint64_t spa, int64_t bytes) { arc_buf_hdr_t *ab, *ab_prev; arc_buf_hdr_t marker = { 0 }; list_t *list, *list_start; kmutex_t *hash_lock, *lock; uint64_t bytes_deleted = 0; uint64_t bufs_skipped = 0; int count = 0; static int evict_offset; int list_count, idx = evict_offset; int offset, lists = 0; ASSERT(GHOST_STATE(state)); /* * data lists come after metadata lists */ list_start = &state->arcs_lists[ARC_BUFC_NUMMETADATALISTS]; list_count = ARC_BUFC_NUMDATALISTS; offset = ARC_BUFC_NUMMETADATALISTS; evict_start: list = &list_start[idx]; lock = ARCS_LOCK(state, idx + offset); mutex_enter(lock); for (ab = list_tail(list); ab; ab = ab_prev) { ab_prev = list_prev(list, ab); if (ab->b_type > ARC_BUFC_NUMTYPES) panic("invalid ab=%p", (void *)ab); if (spa && ab->b_spa != spa) continue; /* ignore markers */ if (ab->b_spa == 0) continue; hash_lock = HDR_LOCK(ab); /* caller may be trying to modify this buffer, skip it */ if (MUTEX_HELD(hash_lock)) continue; /* * It may take a long time to evict all the bufs requested. * To avoid blocking all arc activity, periodically drop * the arcs_mtx and give other threads a chance to run * before reacquiring the lock. */ if (count++ > arc_evict_iterations) { list_insert_after(list, ab, &marker); mutex_exit(lock); kpreempt(KPREEMPT_SYNC); mutex_enter(lock); ab_prev = list_prev(list, &marker); list_remove(list, &marker); count = 0; continue; } if (mutex_tryenter(hash_lock)) { ASSERT(!HDR_IO_IN_PROGRESS(ab)); ASSERT(ab->b_buf == NULL); ARCSTAT_BUMP(arcstat_deleted); bytes_deleted += ab->b_size; if (ab->b_l2hdr != NULL) { /* * This buffer is cached on the 2nd Level ARC; * don't destroy the header. */ arc_change_state(arc_l2c_only, ab, hash_lock); mutex_exit(hash_lock); } else { arc_change_state(arc_anon, ab, hash_lock); mutex_exit(hash_lock); arc_hdr_destroy(ab); } DTRACE_PROBE1(arc__delete, arc_buf_hdr_t *, ab); if (bytes >= 0 && bytes_deleted >= bytes) break; } else if (bytes < 0) { /* * Insert a list marker and then wait for the * hash lock to become available. Once its * available, restart from where we left off. */ list_insert_after(list, ab, &marker); mutex_exit(lock); mutex_enter(hash_lock); mutex_exit(hash_lock); mutex_enter(lock); ab_prev = list_prev(list, &marker); list_remove(list, &marker); } else { bufs_skipped += 1; } } mutex_exit(lock); idx = ((idx + 1) & (ARC_BUFC_NUMDATALISTS - 1)); lists++; if (lists < list_count) goto evict_start; evict_offset = idx; if ((uintptr_t)list > (uintptr_t)&state->arcs_lists[ARC_BUFC_NUMMETADATALISTS] && (bytes < 0 || bytes_deleted < bytes)) { list_start = &state->arcs_lists[0]; list_count = ARC_BUFC_NUMMETADATALISTS; offset = lists = 0; goto evict_start; } if (bufs_skipped) { ARCSTAT_INCR(arcstat_mutex_miss, bufs_skipped); ASSERT(bytes >= 0); } if (bytes_deleted < bytes) dprintf("only deleted %lld bytes from %p", (longlong_t)bytes_deleted, state); } static void arc_adjust(void) { int64_t adjustment, delta; /* * Adjust MRU size */ adjustment = MIN((int64_t)(arc_size - arc_c), (int64_t)(arc_anon->arcs_size + arc_mru->arcs_size + arc_meta_used - arc_p)); if (adjustment > 0 && arc_mru->arcs_lsize[ARC_BUFC_DATA] > 0) { delta = MIN(arc_mru->arcs_lsize[ARC_BUFC_DATA], adjustment); (void) arc_evict(arc_mru, 0, delta, FALSE, ARC_BUFC_DATA); adjustment -= delta; } if (adjustment > 0 && arc_mru->arcs_lsize[ARC_BUFC_METADATA] > 0) { delta = MIN(arc_mru->arcs_lsize[ARC_BUFC_METADATA], adjustment); (void) arc_evict(arc_mru, 0, delta, FALSE, ARC_BUFC_METADATA); } /* * Adjust MFU size */ adjustment = arc_size - arc_c; if (adjustment > 0 && arc_mfu->arcs_lsize[ARC_BUFC_DATA] > 0) { delta = MIN(adjustment, arc_mfu->arcs_lsize[ARC_BUFC_DATA]); (void) arc_evict(arc_mfu, 0, delta, FALSE, ARC_BUFC_DATA); adjustment -= delta; } if (adjustment > 0 && arc_mfu->arcs_lsize[ARC_BUFC_METADATA] > 0) { int64_t delta = MIN(adjustment, arc_mfu->arcs_lsize[ARC_BUFC_METADATA]); (void) arc_evict(arc_mfu, 0, delta, FALSE, ARC_BUFC_METADATA); } /* * Adjust ghost lists */ adjustment = arc_mru->arcs_size + arc_mru_ghost->arcs_size - arc_c; if (adjustment > 0 && arc_mru_ghost->arcs_size > 0) { delta = MIN(arc_mru_ghost->arcs_size, adjustment); arc_evict_ghost(arc_mru_ghost, 0, delta); } adjustment = arc_mru_ghost->arcs_size + arc_mfu_ghost->arcs_size - arc_c; if (adjustment > 0 && arc_mfu_ghost->arcs_size > 0) { delta = MIN(arc_mfu_ghost->arcs_size, adjustment); arc_evict_ghost(arc_mfu_ghost, 0, delta); } } static void arc_do_user_evicts(void) { static arc_buf_t *tmp_arc_eviction_list; /* * Move list over to avoid LOR */ restart: mutex_enter(&arc_eviction_mtx); tmp_arc_eviction_list = arc_eviction_list; arc_eviction_list = NULL; mutex_exit(&arc_eviction_mtx); while (tmp_arc_eviction_list != NULL) { arc_buf_t *buf = tmp_arc_eviction_list; tmp_arc_eviction_list = buf->b_next; mutex_enter(&buf->b_evict_lock); buf->b_hdr = NULL; mutex_exit(&buf->b_evict_lock); if (buf->b_efunc != NULL) VERIFY(buf->b_efunc(buf) == 0); buf->b_efunc = NULL; buf->b_private = NULL; kmem_cache_free(buf_cache, buf); } if (arc_eviction_list != NULL) goto restart; } /* * Flush all *evictable* data from the cache for the given spa. * NOTE: this will not touch "active" (i.e. referenced) data. */ void arc_flush(spa_t *spa) { uint64_t guid = 0; if (spa) guid = spa_load_guid(spa); while (arc_mru->arcs_lsize[ARC_BUFC_DATA]) { (void) arc_evict(arc_mru, guid, -1, FALSE, ARC_BUFC_DATA); if (spa) break; } while (arc_mru->arcs_lsize[ARC_BUFC_METADATA]) { (void) arc_evict(arc_mru, guid, -1, FALSE, ARC_BUFC_METADATA); if (spa) break; } while (arc_mfu->arcs_lsize[ARC_BUFC_DATA]) { (void) arc_evict(arc_mfu, guid, -1, FALSE, ARC_BUFC_DATA); if (spa) break; } while (arc_mfu->arcs_lsize[ARC_BUFC_METADATA]) { (void) arc_evict(arc_mfu, guid, -1, FALSE, ARC_BUFC_METADATA); if (spa) break; } arc_evict_ghost(arc_mru_ghost, guid, -1); arc_evict_ghost(arc_mfu_ghost, guid, -1); mutex_enter(&arc_reclaim_thr_lock); arc_do_user_evicts(); mutex_exit(&arc_reclaim_thr_lock); ASSERT(spa || arc_eviction_list == NULL); } void arc_shrink(void) { if (arc_c > arc_c_min) { uint64_t to_free; #ifdef _KERNEL to_free = arc_c >> arc_shrink_shift; #else to_free = arc_c >> arc_shrink_shift; #endif if (arc_c > arc_c_min + to_free) atomic_add_64(&arc_c, -to_free); else arc_c = arc_c_min; atomic_add_64(&arc_p, -(arc_p >> arc_shrink_shift)); if (arc_c > arc_size) arc_c = MAX(arc_size, arc_c_min); if (arc_p > arc_c) arc_p = (arc_c >> 1); ASSERT(arc_c >= arc_c_min); ASSERT((int64_t)arc_p >= 0); } if (arc_size > arc_c) arc_adjust(); } static int needfree = 0; static int arc_reclaim_needed(void) { #ifdef _KERNEL if (needfree) return (1); /* * Cooperate with pagedaemon when it's time for it to scan * and reclaim some pages. */ if (vm_paging_needed()) return (1); #ifdef sun /* * take 'desfree' extra pages, so we reclaim sooner, rather than later */ extra = desfree; /* * check that we're out of range of the pageout scanner. It starts to * schedule paging if freemem is less than lotsfree and needfree. * lotsfree is the high-water mark for pageout, and needfree is the * number of needed free pages. We add extra pages here to make sure * the scanner doesn't start up while we're freeing memory. */ if (freemem < lotsfree + needfree + extra) return (1); /* * check to make sure that swapfs has enough space so that anon * reservations can still succeed. anon_resvmem() checks that the * availrmem is greater than swapfs_minfree, and the number of reserved * swap pages. We also add a bit of extra here just to prevent * circumstances from getting really dire. */ if (availrmem < swapfs_minfree + swapfs_reserve + extra) return (1); #if defined(__i386) /* * If we're on an i386 platform, it's possible that we'll exhaust the * kernel heap space before we ever run out of available physical * memory. Most checks of the size of the heap_area compare against * tune.t_minarmem, which is the minimum available real memory that we * can have in the system. However, this is generally fixed at 25 pages * which is so low that it's useless. In this comparison, we seek to * calculate the total heap-size, and reclaim if more than 3/4ths of the * heap is allocated. (Or, in the calculation, if less than 1/4th is * free) */ if (btop(vmem_size(heap_arena, VMEM_FREE)) < (btop(vmem_size(heap_arena, VMEM_FREE | VMEM_ALLOC)) >> 2)) return (1); #endif #else /* !sun */ if (kmem_used() > (kmem_size() * 3) / 4) return (1); #endif /* sun */ #else if (spa_get_random(100) == 0) return (1); #endif return (0); } extern kmem_cache_t *zio_buf_cache[]; extern kmem_cache_t *zio_data_buf_cache[]; static void arc_kmem_reap_now(arc_reclaim_strategy_t strat) { size_t i; kmem_cache_t *prev_cache = NULL; kmem_cache_t *prev_data_cache = NULL; #ifdef _KERNEL if (arc_meta_used >= arc_meta_limit) { /* * We are exceeding our meta-data cache limit. * Purge some DNLC entries to release holds on meta-data. */ dnlc_reduce_cache((void *)(uintptr_t)arc_reduce_dnlc_percent); } #if defined(__i386) /* * Reclaim unused memory from all kmem caches. */ kmem_reap(); #endif #endif /* * An aggressive reclamation will shrink the cache size as well as * reap free buffers from the arc kmem caches. */ if (strat == ARC_RECLAIM_AGGR) arc_shrink(); for (i = 0; i < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; i++) { if (zio_buf_cache[i] != prev_cache) { prev_cache = zio_buf_cache[i]; kmem_cache_reap_now(zio_buf_cache[i]); } if (zio_data_buf_cache[i] != prev_data_cache) { prev_data_cache = zio_data_buf_cache[i]; kmem_cache_reap_now(zio_data_buf_cache[i]); } } kmem_cache_reap_now(buf_cache); kmem_cache_reap_now(hdr_cache); } static void arc_reclaim_thread(void *dummy __unused) { clock_t growtime = 0; arc_reclaim_strategy_t last_reclaim = ARC_RECLAIM_CONS; callb_cpr_t cpr; CALLB_CPR_INIT(&cpr, &arc_reclaim_thr_lock, callb_generic_cpr, FTAG); mutex_enter(&arc_reclaim_thr_lock); while (arc_thread_exit == 0) { if (arc_reclaim_needed()) { if (arc_no_grow) { if (last_reclaim == ARC_RECLAIM_CONS) { last_reclaim = ARC_RECLAIM_AGGR; } else { last_reclaim = ARC_RECLAIM_CONS; } } else { arc_no_grow = TRUE; last_reclaim = ARC_RECLAIM_AGGR; membar_producer(); } /* reset the growth delay for every reclaim */ growtime = ddi_get_lbolt() + (arc_grow_retry * hz); if (needfree && last_reclaim == ARC_RECLAIM_CONS) { /* * If needfree is TRUE our vm_lowmem hook * was called and in that case we must free some * memory, so switch to aggressive mode. */ arc_no_grow = TRUE; last_reclaim = ARC_RECLAIM_AGGR; } arc_kmem_reap_now(last_reclaim); arc_warm = B_TRUE; } else if (arc_no_grow && ddi_get_lbolt() >= growtime) { arc_no_grow = FALSE; } arc_adjust(); if (arc_eviction_list != NULL) arc_do_user_evicts(); #ifdef _KERNEL if (needfree) { needfree = 0; wakeup(&needfree); } #endif /* block until needed, or one second, whichever is shorter */ CALLB_CPR_SAFE_BEGIN(&cpr); (void) cv_timedwait(&arc_reclaim_thr_cv, &arc_reclaim_thr_lock, hz); CALLB_CPR_SAFE_END(&cpr, &arc_reclaim_thr_lock); } arc_thread_exit = 0; cv_broadcast(&arc_reclaim_thr_cv); CALLB_CPR_EXIT(&cpr); /* drops arc_reclaim_thr_lock */ thread_exit(); } /* * Adapt arc info given the number of bytes we are trying to add and * the state that we are comming from. This function is only called * when we are adding new content to the cache. */ static void arc_adapt(int bytes, arc_state_t *state) { int mult; uint64_t arc_p_min = (arc_c >> arc_p_min_shift); if (state == arc_l2c_only) return; ASSERT(bytes > 0); /* * Adapt the target size of the MRU list: * - if we just hit in the MRU ghost list, then increase * the target size of the MRU list. * - if we just hit in the MFU ghost list, then increase * the target size of the MFU list by decreasing the * target size of the MRU list. */ if (state == arc_mru_ghost) { mult = ((arc_mru_ghost->arcs_size >= arc_mfu_ghost->arcs_size) ? 1 : (arc_mfu_ghost->arcs_size/arc_mru_ghost->arcs_size)); mult = MIN(mult, 10); /* avoid wild arc_p adjustment */ arc_p = MIN(arc_c - arc_p_min, arc_p + bytes * mult); } else if (state == arc_mfu_ghost) { uint64_t delta; mult = ((arc_mfu_ghost->arcs_size >= arc_mru_ghost->arcs_size) ? 1 : (arc_mru_ghost->arcs_size/arc_mfu_ghost->arcs_size)); mult = MIN(mult, 10); delta = MIN(bytes * mult, arc_p); arc_p = MAX(arc_p_min, arc_p - delta); } ASSERT((int64_t)arc_p >= 0); if (arc_reclaim_needed()) { cv_signal(&arc_reclaim_thr_cv); return; } if (arc_no_grow) return; if (arc_c >= arc_c_max) return; /* * If we're within (2 * maxblocksize) bytes of the target * cache size, increment the target cache size */ if (arc_size > arc_c - (2ULL << SPA_MAXBLOCKSHIFT)) { atomic_add_64(&arc_c, (int64_t)bytes); if (arc_c > arc_c_max) arc_c = arc_c_max; else if (state == arc_anon) atomic_add_64(&arc_p, (int64_t)bytes); if (arc_p > arc_c) arc_p = arc_c; } ASSERT((int64_t)arc_p >= 0); } /* * Check if the cache has reached its limits and eviction is required * prior to insert. */ static int arc_evict_needed(arc_buf_contents_t type) { if (type == ARC_BUFC_METADATA && arc_meta_used >= arc_meta_limit) return (1); #ifdef sun #ifdef _KERNEL /* * If zio data pages are being allocated out of a separate heap segment, * then enforce that the size of available vmem for this area remains * above about 1/32nd free. */ if (type == ARC_BUFC_DATA && zio_arena != NULL && vmem_size(zio_arena, VMEM_FREE) < (vmem_size(zio_arena, VMEM_ALLOC) >> 5)) return (1); #endif #endif /* sun */ if (arc_reclaim_needed()) return (1); return (arc_size > arc_c); } /* * The buffer, supplied as the first argument, needs a data block. * So, if we are at cache max, determine which cache should be victimized. * We have the following cases: * * 1. Insert for MRU, p > sizeof(arc_anon + arc_mru) -> * In this situation if we're out of space, but the resident size of the MFU is * under the limit, victimize the MFU cache to satisfy this insertion request. * * 2. Insert for MRU, p <= sizeof(arc_anon + arc_mru) -> * Here, we've used up all of the available space for the MRU, so we need to * evict from our own cache instead. Evict from the set of resident MRU * entries. * * 3. Insert for MFU (c - p) > sizeof(arc_mfu) -> * c minus p represents the MFU space in the cache, since p is the size of the * cache that is dedicated to the MRU. In this situation there's still space on * the MFU side, so the MRU side needs to be victimized. * * 4. Insert for MFU (c - p) < sizeof(arc_mfu) -> * MFU's resident set is consuming more space than it has been allotted. In * this situation, we must victimize our own cache, the MFU, for this insertion. */ static void arc_get_data_buf(arc_buf_t *buf) { arc_state_t *state = buf->b_hdr->b_state; uint64_t size = buf->b_hdr->b_size; arc_buf_contents_t type = buf->b_hdr->b_type; arc_adapt(size, state); /* * We have not yet reached cache maximum size, * just allocate a new buffer. */ if (!arc_evict_needed(type)) { if (type == ARC_BUFC_METADATA) { buf->b_data = zio_buf_alloc(size); arc_space_consume(size, ARC_SPACE_DATA); } else { ASSERT(type == ARC_BUFC_DATA); buf->b_data = zio_data_buf_alloc(size); ARCSTAT_INCR(arcstat_data_size, size); atomic_add_64(&arc_size, size); } goto out; } /* * If we are prefetching from the mfu ghost list, this buffer * will end up on the mru list; so steal space from there. */ if (state == arc_mfu_ghost) state = buf->b_hdr->b_flags & ARC_PREFETCH ? arc_mru : arc_mfu; else if (state == arc_mru_ghost) state = arc_mru; if (state == arc_mru || state == arc_anon) { uint64_t mru_used = arc_anon->arcs_size + arc_mru->arcs_size; state = (arc_mfu->arcs_lsize[type] >= size && arc_p > mru_used) ? arc_mfu : arc_mru; } else { /* MFU cases */ uint64_t mfu_space = arc_c - arc_p; state = (arc_mru->arcs_lsize[type] >= size && mfu_space > arc_mfu->arcs_size) ? arc_mru : arc_mfu; } if ((buf->b_data = arc_evict(state, 0, size, TRUE, type)) == NULL) { if (type == ARC_BUFC_METADATA) { buf->b_data = zio_buf_alloc(size); arc_space_consume(size, ARC_SPACE_DATA); } else { ASSERT(type == ARC_BUFC_DATA); buf->b_data = zio_data_buf_alloc(size); ARCSTAT_INCR(arcstat_data_size, size); atomic_add_64(&arc_size, size); } ARCSTAT_BUMP(arcstat_recycle_miss); } ASSERT(buf->b_data != NULL); out: /* * Update the state size. Note that ghost states have a * "ghost size" and so don't need to be updated. */ if (!GHOST_STATE(buf->b_hdr->b_state)) { arc_buf_hdr_t *hdr = buf->b_hdr; atomic_add_64(&hdr->b_state->arcs_size, size); if (list_link_active(&hdr->b_arc_node)) { ASSERT(refcount_is_zero(&hdr->b_refcnt)); atomic_add_64(&hdr->b_state->arcs_lsize[type], size); } /* * If we are growing the cache, and we are adding anonymous * data, and we have outgrown arc_p, update arc_p */ if (arc_size < arc_c && hdr->b_state == arc_anon && arc_anon->arcs_size + arc_mru->arcs_size > arc_p) arc_p = MIN(arc_c, arc_p + size); } ARCSTAT_BUMP(arcstat_allocated); } /* * This routine is called whenever a buffer is accessed. * NOTE: the hash lock is dropped in this function. */ static void arc_access(arc_buf_hdr_t *buf, kmutex_t *hash_lock) { clock_t now; ASSERT(MUTEX_HELD(hash_lock)); if (buf->b_state == arc_anon) { /* * This buffer is not in the cache, and does not * appear in our "ghost" list. Add the new buffer * to the MRU state. */ ASSERT(buf->b_arc_access == 0); buf->b_arc_access = ddi_get_lbolt(); DTRACE_PROBE1(new_state__mru, arc_buf_hdr_t *, buf); arc_change_state(arc_mru, buf, hash_lock); } else if (buf->b_state == arc_mru) { now = ddi_get_lbolt(); /* * If this buffer is here because of a prefetch, then either: * - clear the flag if this is a "referencing" read * (any subsequent access will bump this into the MFU state). * or * - move the buffer to the head of the list if this is * another prefetch (to make it less likely to be evicted). */ if ((buf->b_flags & ARC_PREFETCH) != 0) { if (refcount_count(&buf->b_refcnt) == 0) { ASSERT(list_link_active(&buf->b_arc_node)); } else { buf->b_flags &= ~ARC_PREFETCH; ARCSTAT_BUMP(arcstat_mru_hits); } buf->b_arc_access = now; return; } /* * This buffer has been "accessed" only once so far, * but it is still in the cache. Move it to the MFU * state. */ if (now > buf->b_arc_access + ARC_MINTIME) { /* * More than 125ms have passed since we * instantiated this buffer. Move it to the * most frequently used state. */ buf->b_arc_access = now; DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, buf); arc_change_state(arc_mfu, buf, hash_lock); } ARCSTAT_BUMP(arcstat_mru_hits); } else if (buf->b_state == arc_mru_ghost) { arc_state_t *new_state; /* * This buffer has been "accessed" recently, but * was evicted from the cache. Move it to the * MFU state. */ if (buf->b_flags & ARC_PREFETCH) { new_state = arc_mru; if (refcount_count(&buf->b_refcnt) > 0) buf->b_flags &= ~ARC_PREFETCH; DTRACE_PROBE1(new_state__mru, arc_buf_hdr_t *, buf); } else { new_state = arc_mfu; DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, buf); } buf->b_arc_access = ddi_get_lbolt(); arc_change_state(new_state, buf, hash_lock); ARCSTAT_BUMP(arcstat_mru_ghost_hits); } else if (buf->b_state == arc_mfu) { /* * This buffer has been accessed more than once and is * still in the cache. Keep it in the MFU state. * * NOTE: an add_reference() that occurred when we did * the arc_read() will have kicked this off the list. * If it was a prefetch, we will explicitly move it to * the head of the list now. */ if ((buf->b_flags & ARC_PREFETCH) != 0) { ASSERT(refcount_count(&buf->b_refcnt) == 0); ASSERT(list_link_active(&buf->b_arc_node)); } ARCSTAT_BUMP(arcstat_mfu_hits); buf->b_arc_access = ddi_get_lbolt(); } else if (buf->b_state == arc_mfu_ghost) { arc_state_t *new_state = arc_mfu; /* * This buffer has been accessed more than once but has * been evicted from the cache. Move it back to the * MFU state. */ if (buf->b_flags & ARC_PREFETCH) { /* * This is a prefetch access... * move this block back to the MRU state. */ ASSERT0(refcount_count(&buf->b_refcnt)); new_state = arc_mru; } buf->b_arc_access = ddi_get_lbolt(); DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, buf); arc_change_state(new_state, buf, hash_lock); ARCSTAT_BUMP(arcstat_mfu_ghost_hits); } else if (buf->b_state == arc_l2c_only) { /* * This buffer is on the 2nd Level ARC. */ buf->b_arc_access = ddi_get_lbolt(); DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, buf); arc_change_state(arc_mfu, buf, hash_lock); } else { ASSERT(!"invalid arc state"); } } /* a generic arc_done_func_t which you can use */ /* ARGSUSED */ void arc_bcopy_func(zio_t *zio, arc_buf_t *buf, void *arg) { if (zio == NULL || zio->io_error == 0) bcopy(buf->b_data, arg, buf->b_hdr->b_size); VERIFY(arc_buf_remove_ref(buf, arg)); } /* a generic arc_done_func_t */ void arc_getbuf_func(zio_t *zio, arc_buf_t *buf, void *arg) { arc_buf_t **bufp = arg; if (zio && zio->io_error) { VERIFY(arc_buf_remove_ref(buf, arg)); *bufp = NULL; } else { *bufp = buf; ASSERT(buf->b_data); } } static void arc_read_done(zio_t *zio) { arc_buf_hdr_t *hdr, *found; arc_buf_t *buf; arc_buf_t *abuf; /* buffer we're assigning to callback */ kmutex_t *hash_lock; arc_callback_t *callback_list, *acb; int freeable = FALSE; buf = zio->io_private; hdr = buf->b_hdr; /* * The hdr was inserted into hash-table and removed from lists * prior to starting I/O. We should find this header, since * it's in the hash table, and it should be legit since it's * not possible to evict it during the I/O. The only possible * reason for it not to be found is if we were freed during the * read. */ found = buf_hash_find(hdr->b_spa, &hdr->b_dva, hdr->b_birth, &hash_lock); ASSERT((found == NULL && HDR_FREED_IN_READ(hdr) && hash_lock == NULL) || (found == hdr && DVA_EQUAL(&hdr->b_dva, BP_IDENTITY(zio->io_bp))) || (found == hdr && HDR_L2_READING(hdr))); hdr->b_flags &= ~ARC_L2_EVICTED; if (l2arc_noprefetch && (hdr->b_flags & ARC_PREFETCH)) hdr->b_flags &= ~ARC_L2CACHE; /* byteswap if necessary */ callback_list = hdr->b_acb; ASSERT(callback_list != NULL); if (BP_SHOULD_BYTESWAP(zio->io_bp) && zio->io_error == 0) { dmu_object_byteswap_t bswap = DMU_OT_BYTESWAP(BP_GET_TYPE(zio->io_bp)); arc_byteswap_func_t *func = BP_GET_LEVEL(zio->io_bp) > 0 ? byteswap_uint64_array : dmu_ot_byteswap[bswap].ob_func; func(buf->b_data, hdr->b_size); } arc_cksum_compute(buf, B_FALSE); #ifdef illumos arc_buf_watch(buf); #endif /* illumos */ if (hash_lock && zio->io_error == 0 && hdr->b_state == arc_anon) { /* * Only call arc_access on anonymous buffers. This is because * if we've issued an I/O for an evicted buffer, we've already * called arc_access (to prevent any simultaneous readers from * getting confused). */ arc_access(hdr, hash_lock); } /* create copies of the data buffer for the callers */ abuf = buf; for (acb = callback_list; acb; acb = acb->acb_next) { if (acb->acb_done) { if (abuf == NULL) { ARCSTAT_BUMP(arcstat_duplicate_reads); abuf = arc_buf_clone(buf); } acb->acb_buf = abuf; abuf = NULL; } } hdr->b_acb = NULL; hdr->b_flags &= ~ARC_IO_IN_PROGRESS; ASSERT(!HDR_BUF_AVAILABLE(hdr)); if (abuf == buf) { ASSERT(buf->b_efunc == NULL); ASSERT(hdr->b_datacnt == 1); hdr->b_flags |= ARC_BUF_AVAILABLE; } ASSERT(refcount_is_zero(&hdr->b_refcnt) || callback_list != NULL); if (zio->io_error != 0) { hdr->b_flags |= ARC_IO_ERROR; if (hdr->b_state != arc_anon) arc_change_state(arc_anon, hdr, hash_lock); if (HDR_IN_HASH_TABLE(hdr)) buf_hash_remove(hdr); freeable = refcount_is_zero(&hdr->b_refcnt); } /* * Broadcast before we drop the hash_lock to avoid the possibility * that the hdr (and hence the cv) might be freed before we get to * the cv_broadcast(). */ cv_broadcast(&hdr->b_cv); if (hash_lock) { mutex_exit(hash_lock); } else { /* * This block was freed while we waited for the read to * complete. It has been removed from the hash table and * moved to the anonymous state (so that it won't show up * in the cache). */ ASSERT3P(hdr->b_state, ==, arc_anon); freeable = refcount_is_zero(&hdr->b_refcnt); } /* execute each callback and free its structure */ while ((acb = callback_list) != NULL) { if (acb->acb_done) acb->acb_done(zio, acb->acb_buf, acb->acb_private); if (acb->acb_zio_dummy != NULL) { acb->acb_zio_dummy->io_error = zio->io_error; zio_nowait(acb->acb_zio_dummy); } callback_list = acb->acb_next; kmem_free(acb, sizeof (arc_callback_t)); } if (freeable) arc_hdr_destroy(hdr); } /* * "Read" the block block at the specified DVA (in bp) via the * cache. If the block is found in the cache, invoke the provided * callback immediately and return. Note that the `zio' parameter * in the callback will be NULL in this case, since no IO was * required. If the block is not in the cache pass the read request * on to the spa with a substitute callback function, so that the * requested block will be added to the cache. * * If a read request arrives for a block that has a read in-progress, * either wait for the in-progress read to complete (and return the * results); or, if this is a read with a "done" func, add a record * to the read to invoke the "done" func when the read completes, * and return; or just return. * * arc_read_done() will invoke all the requested "done" functions * for readers of this block. */ int arc_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, arc_done_func_t *done, void *private, zio_priority_t priority, int zio_flags, uint32_t *arc_flags, const zbookmark_t *zb) { arc_buf_hdr_t *hdr; arc_buf_t *buf = NULL; kmutex_t *hash_lock; zio_t *rzio; uint64_t guid = spa_load_guid(spa); top: hdr = buf_hash_find(guid, BP_IDENTITY(bp), BP_PHYSICAL_BIRTH(bp), &hash_lock); if (hdr && hdr->b_datacnt > 0) { *arc_flags |= ARC_CACHED; if (HDR_IO_IN_PROGRESS(hdr)) { if (*arc_flags & ARC_WAIT) { cv_wait(&hdr->b_cv, hash_lock); mutex_exit(hash_lock); goto top; } ASSERT(*arc_flags & ARC_NOWAIT); if (done) { arc_callback_t *acb = NULL; acb = kmem_zalloc(sizeof (arc_callback_t), KM_SLEEP); acb->acb_done = done; acb->acb_private = private; if (pio != NULL) acb->acb_zio_dummy = zio_null(pio, spa, NULL, NULL, NULL, zio_flags); ASSERT(acb->acb_done != NULL); acb->acb_next = hdr->b_acb; hdr->b_acb = acb; add_reference(hdr, hash_lock, private); mutex_exit(hash_lock); return (0); } mutex_exit(hash_lock); return (0); } ASSERT(hdr->b_state == arc_mru || hdr->b_state == arc_mfu); if (done) { add_reference(hdr, hash_lock, private); /* * If this block is already in use, create a new * copy of the data so that we will be guaranteed * that arc_release() will always succeed. */ buf = hdr->b_buf; ASSERT(buf); ASSERT(buf->b_data); if (HDR_BUF_AVAILABLE(hdr)) { ASSERT(buf->b_efunc == NULL); hdr->b_flags &= ~ARC_BUF_AVAILABLE; } else { buf = arc_buf_clone(buf); } } else if (*arc_flags & ARC_PREFETCH && refcount_count(&hdr->b_refcnt) == 0) { hdr->b_flags |= ARC_PREFETCH; } DTRACE_PROBE1(arc__hit, arc_buf_hdr_t *, hdr); arc_access(hdr, hash_lock); if (*arc_flags & ARC_L2CACHE) hdr->b_flags |= ARC_L2CACHE; if (*arc_flags & ARC_L2COMPRESS) hdr->b_flags |= ARC_L2COMPRESS; mutex_exit(hash_lock); ARCSTAT_BUMP(arcstat_hits); ARCSTAT_CONDSTAT(!(hdr->b_flags & ARC_PREFETCH), demand, prefetch, hdr->b_type != ARC_BUFC_METADATA, data, metadata, hits); if (done) done(NULL, buf, private); } else { uint64_t size = BP_GET_LSIZE(bp); arc_callback_t *acb; vdev_t *vd = NULL; uint64_t addr = 0; boolean_t devw = B_FALSE; enum zio_compress b_compress = ZIO_COMPRESS_OFF; uint64_t b_asize = 0; if (hdr == NULL) { /* this block is not in the cache */ arc_buf_hdr_t *exists; arc_buf_contents_t type = BP_GET_BUFC_TYPE(bp); buf = arc_buf_alloc(spa, size, private, type); hdr = buf->b_hdr; hdr->b_dva = *BP_IDENTITY(bp); hdr->b_birth = BP_PHYSICAL_BIRTH(bp); hdr->b_cksum0 = bp->blk_cksum.zc_word[0]; exists = buf_hash_insert(hdr, &hash_lock); if (exists) { /* somebody beat us to the hash insert */ mutex_exit(hash_lock); buf_discard_identity(hdr); (void) arc_buf_remove_ref(buf, private); goto top; /* restart the IO request */ } /* if this is a prefetch, we don't have a reference */ if (*arc_flags & ARC_PREFETCH) { (void) remove_reference(hdr, hash_lock, private); hdr->b_flags |= ARC_PREFETCH; } if (*arc_flags & ARC_L2CACHE) hdr->b_flags |= ARC_L2CACHE; if (*arc_flags & ARC_L2COMPRESS) hdr->b_flags |= ARC_L2COMPRESS; if (BP_GET_LEVEL(bp) > 0) hdr->b_flags |= ARC_INDIRECT; } else { /* this block is in the ghost cache */ ASSERT(GHOST_STATE(hdr->b_state)); ASSERT(!HDR_IO_IN_PROGRESS(hdr)); ASSERT0(refcount_count(&hdr->b_refcnt)); ASSERT(hdr->b_buf == NULL); /* if this is a prefetch, we don't have a reference */ if (*arc_flags & ARC_PREFETCH) hdr->b_flags |= ARC_PREFETCH; else add_reference(hdr, hash_lock, private); if (*arc_flags & ARC_L2CACHE) hdr->b_flags |= ARC_L2CACHE; if (*arc_flags & ARC_L2COMPRESS) hdr->b_flags |= ARC_L2COMPRESS; buf = kmem_cache_alloc(buf_cache, KM_PUSHPAGE); buf->b_hdr = hdr; buf->b_data = NULL; buf->b_efunc = NULL; buf->b_private = NULL; buf->b_next = NULL; hdr->b_buf = buf; ASSERT(hdr->b_datacnt == 0); hdr->b_datacnt = 1; arc_get_data_buf(buf); arc_access(hdr, hash_lock); } ASSERT(!GHOST_STATE(hdr->b_state)); acb = kmem_zalloc(sizeof (arc_callback_t), KM_SLEEP); acb->acb_done = done; acb->acb_private = private; ASSERT(hdr->b_acb == NULL); hdr->b_acb = acb; hdr->b_flags |= ARC_IO_IN_PROGRESS; if (hdr->b_l2hdr != NULL && (vd = hdr->b_l2hdr->b_dev->l2ad_vdev) != NULL) { devw = hdr->b_l2hdr->b_dev->l2ad_writing; addr = hdr->b_l2hdr->b_daddr; b_compress = hdr->b_l2hdr->b_compress; b_asize = hdr->b_l2hdr->b_asize; /* * Lock out device removal. */ if (vdev_is_dead(vd) || !spa_config_tryenter(spa, SCL_L2ARC, vd, RW_READER)) vd = NULL; } mutex_exit(hash_lock); /* * At this point, we have a level 1 cache miss. Try again in * L2ARC if possible. */ ASSERT3U(hdr->b_size, ==, size); DTRACE_PROBE4(arc__miss, arc_buf_hdr_t *, hdr, blkptr_t *, bp, uint64_t, size, zbookmark_t *, zb); ARCSTAT_BUMP(arcstat_misses); ARCSTAT_CONDSTAT(!(hdr->b_flags & ARC_PREFETCH), demand, prefetch, hdr->b_type != ARC_BUFC_METADATA, data, metadata, misses); #ifdef _KERNEL curthread->td_ru.ru_inblock++; #endif if (vd != NULL && l2arc_ndev != 0 && !(l2arc_norw && devw)) { /* * Read from the L2ARC if the following are true: * 1. The L2ARC vdev was previously cached. * 2. This buffer still has L2ARC metadata. * 3. This buffer isn't currently writing to the L2ARC. * 4. The L2ARC entry wasn't evicted, which may * also have invalidated the vdev. * 5. This isn't prefetch and l2arc_noprefetch is set. */ if (hdr->b_l2hdr != NULL && !HDR_L2_WRITING(hdr) && !HDR_L2_EVICTED(hdr) && !(l2arc_noprefetch && HDR_PREFETCH(hdr))) { l2arc_read_callback_t *cb; DTRACE_PROBE1(l2arc__hit, arc_buf_hdr_t *, hdr); ARCSTAT_BUMP(arcstat_l2_hits); cb = kmem_zalloc(sizeof (l2arc_read_callback_t), KM_SLEEP); cb->l2rcb_buf = buf; cb->l2rcb_spa = spa; cb->l2rcb_bp = *bp; cb->l2rcb_zb = *zb; cb->l2rcb_flags = zio_flags; cb->l2rcb_compress = b_compress; ASSERT(addr >= VDEV_LABEL_START_SIZE && addr + size < vd->vdev_psize - VDEV_LABEL_END_SIZE); /* * l2arc read. The SCL_L2ARC lock will be * released by l2arc_read_done(). * Issue a null zio if the underlying buffer * was squashed to zero size by compression. */ if (b_compress == ZIO_COMPRESS_EMPTY) { rzio = zio_null(pio, spa, vd, l2arc_read_done, cb, zio_flags | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY); } else { rzio = zio_read_phys(pio, vd, addr, b_asize, buf->b_data, ZIO_CHECKSUM_OFF, l2arc_read_done, cb, priority, zio_flags | ZIO_FLAG_DONT_CACHE | ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY, B_FALSE); } DTRACE_PROBE2(l2arc__read, vdev_t *, vd, zio_t *, rzio); ARCSTAT_INCR(arcstat_l2_read_bytes, b_asize); if (*arc_flags & ARC_NOWAIT) { zio_nowait(rzio); return (0); } ASSERT(*arc_flags & ARC_WAIT); if (zio_wait(rzio) == 0) return (0); /* l2arc read error; goto zio_read() */ } else { DTRACE_PROBE1(l2arc__miss, arc_buf_hdr_t *, hdr); ARCSTAT_BUMP(arcstat_l2_misses); if (HDR_L2_WRITING(hdr)) ARCSTAT_BUMP(arcstat_l2_rw_clash); spa_config_exit(spa, SCL_L2ARC, vd); } } else { if (vd != NULL) spa_config_exit(spa, SCL_L2ARC, vd); if (l2arc_ndev != 0) { DTRACE_PROBE1(l2arc__miss, arc_buf_hdr_t *, hdr); ARCSTAT_BUMP(arcstat_l2_misses); } } rzio = zio_read(pio, spa, bp, buf->b_data, size, arc_read_done, buf, priority, zio_flags, zb); if (*arc_flags & ARC_WAIT) return (zio_wait(rzio)); ASSERT(*arc_flags & ARC_NOWAIT); zio_nowait(rzio); } return (0); } void arc_set_callback(arc_buf_t *buf, arc_evict_func_t *func, void *private) { ASSERT(buf->b_hdr != NULL); ASSERT(buf->b_hdr->b_state != arc_anon); ASSERT(!refcount_is_zero(&buf->b_hdr->b_refcnt) || func == NULL); ASSERT(buf->b_efunc == NULL); ASSERT(!HDR_BUF_AVAILABLE(buf->b_hdr)); buf->b_efunc = func; buf->b_private = private; } /* * Notify the arc that a block was freed, and thus will never be used again. */ void arc_freed(spa_t *spa, const blkptr_t *bp) { arc_buf_hdr_t *hdr; kmutex_t *hash_lock; uint64_t guid = spa_load_guid(spa); hdr = buf_hash_find(guid, BP_IDENTITY(bp), BP_PHYSICAL_BIRTH(bp), &hash_lock); if (hdr == NULL) return; if (HDR_BUF_AVAILABLE(hdr)) { arc_buf_t *buf = hdr->b_buf; add_reference(hdr, hash_lock, FTAG); hdr->b_flags &= ~ARC_BUF_AVAILABLE; mutex_exit(hash_lock); arc_release(buf, FTAG); (void) arc_buf_remove_ref(buf, FTAG); } else { mutex_exit(hash_lock); } } /* * This is used by the DMU to let the ARC know that a buffer is * being evicted, so the ARC should clean up. If this arc buf * is not yet in the evicted state, it will be put there. */ int arc_buf_evict(arc_buf_t *buf) { arc_buf_hdr_t *hdr; kmutex_t *hash_lock; arc_buf_t **bufp; list_t *list, *evicted_list; kmutex_t *lock, *evicted_lock; mutex_enter(&buf->b_evict_lock); hdr = buf->b_hdr; if (hdr == NULL) { /* * We are in arc_do_user_evicts(). */ ASSERT(buf->b_data == NULL); mutex_exit(&buf->b_evict_lock); return (0); } else if (buf->b_data == NULL) { arc_buf_t copy = *buf; /* structure assignment */ /* * We are on the eviction list; process this buffer now * but let arc_do_user_evicts() do the reaping. */ buf->b_efunc = NULL; mutex_exit(&buf->b_evict_lock); VERIFY(copy.b_efunc(©) == 0); return (1); } hash_lock = HDR_LOCK(hdr); mutex_enter(hash_lock); hdr = buf->b_hdr; ASSERT3P(hash_lock, ==, HDR_LOCK(hdr)); ASSERT3U(refcount_count(&hdr->b_refcnt), <, hdr->b_datacnt); ASSERT(hdr->b_state == arc_mru || hdr->b_state == arc_mfu); /* * Pull this buffer off of the hdr */ bufp = &hdr->b_buf; while (*bufp != buf) bufp = &(*bufp)->b_next; *bufp = buf->b_next; ASSERT(buf->b_data != NULL); arc_buf_destroy(buf, FALSE, FALSE); if (hdr->b_datacnt == 0) { arc_state_t *old_state = hdr->b_state; arc_state_t *evicted_state; ASSERT(hdr->b_buf == NULL); ASSERT(refcount_is_zero(&hdr->b_refcnt)); evicted_state = (old_state == arc_mru) ? arc_mru_ghost : arc_mfu_ghost; get_buf_info(hdr, old_state, &list, &lock); get_buf_info(hdr, evicted_state, &evicted_list, &evicted_lock); mutex_enter(lock); mutex_enter(evicted_lock); arc_change_state(evicted_state, hdr, hash_lock); ASSERT(HDR_IN_HASH_TABLE(hdr)); hdr->b_flags |= ARC_IN_HASH_TABLE; hdr->b_flags &= ~ARC_BUF_AVAILABLE; mutex_exit(evicted_lock); mutex_exit(lock); } mutex_exit(hash_lock); mutex_exit(&buf->b_evict_lock); VERIFY(buf->b_efunc(buf) == 0); buf->b_efunc = NULL; buf->b_private = NULL; buf->b_hdr = NULL; buf->b_next = NULL; kmem_cache_free(buf_cache, buf); return (1); } /* * Release this buffer from the cache, making it an anonymous buffer. This * must be done after a read and prior to modifying the buffer contents. * If the buffer has more than one reference, we must make * a new hdr for the buffer. */ void arc_release(arc_buf_t *buf, void *tag) { arc_buf_hdr_t *hdr; kmutex_t *hash_lock = NULL; l2arc_buf_hdr_t *l2hdr; uint64_t buf_size; /* * It would be nice to assert that if it's DMU metadata (level > * 0 || it's the dnode file), then it must be syncing context. * But we don't know that information at this level. */ mutex_enter(&buf->b_evict_lock); hdr = buf->b_hdr; /* this buffer is not on any list */ ASSERT(refcount_count(&hdr->b_refcnt) > 0); if (hdr->b_state == arc_anon) { /* this buffer is already released */ ASSERT(buf->b_efunc == NULL); } else { hash_lock = HDR_LOCK(hdr); mutex_enter(hash_lock); hdr = buf->b_hdr; ASSERT3P(hash_lock, ==, HDR_LOCK(hdr)); } l2hdr = hdr->b_l2hdr; if (l2hdr) { mutex_enter(&l2arc_buflist_mtx); hdr->b_l2hdr = NULL; list_remove(l2hdr->b_dev->l2ad_buflist, hdr); } buf_size = hdr->b_size; /* * Do we have more than one buf? */ if (hdr->b_datacnt > 1) { arc_buf_hdr_t *nhdr; arc_buf_t **bufp; uint64_t blksz = hdr->b_size; uint64_t spa = hdr->b_spa; arc_buf_contents_t type = hdr->b_type; uint32_t flags = hdr->b_flags; ASSERT(hdr->b_buf != buf || buf->b_next != NULL); /* * Pull the data off of this hdr and attach it to * a new anonymous hdr. */ (void) remove_reference(hdr, hash_lock, tag); bufp = &hdr->b_buf; while (*bufp != buf) bufp = &(*bufp)->b_next; *bufp = buf->b_next; buf->b_next = NULL; ASSERT3U(hdr->b_state->arcs_size, >=, hdr->b_size); atomic_add_64(&hdr->b_state->arcs_size, -hdr->b_size); if (refcount_is_zero(&hdr->b_refcnt)) { uint64_t *size = &hdr->b_state->arcs_lsize[hdr->b_type]; ASSERT3U(*size, >=, hdr->b_size); atomic_add_64(size, -hdr->b_size); } /* * We're releasing a duplicate user data buffer, update * our statistics accordingly. */ if (hdr->b_type == ARC_BUFC_DATA) { ARCSTAT_BUMPDOWN(arcstat_duplicate_buffers); ARCSTAT_INCR(arcstat_duplicate_buffers_size, -hdr->b_size); } hdr->b_datacnt -= 1; arc_cksum_verify(buf); #ifdef illumos arc_buf_unwatch(buf); #endif /* illumos */ mutex_exit(hash_lock); nhdr = kmem_cache_alloc(hdr_cache, KM_PUSHPAGE); nhdr->b_size = blksz; nhdr->b_spa = spa; nhdr->b_type = type; nhdr->b_buf = buf; nhdr->b_state = arc_anon; nhdr->b_arc_access = 0; nhdr->b_flags = flags & ARC_L2_WRITING; nhdr->b_l2hdr = NULL; nhdr->b_datacnt = 1; nhdr->b_freeze_cksum = NULL; (void) refcount_add(&nhdr->b_refcnt, tag); buf->b_hdr = nhdr; mutex_exit(&buf->b_evict_lock); atomic_add_64(&arc_anon->arcs_size, blksz); } else { mutex_exit(&buf->b_evict_lock); ASSERT(refcount_count(&hdr->b_refcnt) == 1); ASSERT(!list_link_active(&hdr->b_arc_node)); ASSERT(!HDR_IO_IN_PROGRESS(hdr)); if (hdr->b_state != arc_anon) arc_change_state(arc_anon, hdr, hash_lock); hdr->b_arc_access = 0; if (hash_lock) mutex_exit(hash_lock); buf_discard_identity(hdr); arc_buf_thaw(buf); } buf->b_efunc = NULL; buf->b_private = NULL; if (l2hdr) { ARCSTAT_INCR(arcstat_l2_asize, -l2hdr->b_asize); l2arc_trim(l2hdr); kmem_free(l2hdr, sizeof (l2arc_buf_hdr_t)); ARCSTAT_INCR(arcstat_l2_size, -buf_size); mutex_exit(&l2arc_buflist_mtx); } } int arc_released(arc_buf_t *buf) { int released; mutex_enter(&buf->b_evict_lock); released = (buf->b_data != NULL && buf->b_hdr->b_state == arc_anon); mutex_exit(&buf->b_evict_lock); return (released); } int arc_has_callback(arc_buf_t *buf) { int callback; mutex_enter(&buf->b_evict_lock); callback = (buf->b_efunc != NULL); mutex_exit(&buf->b_evict_lock); return (callback); } #ifdef ZFS_DEBUG int arc_referenced(arc_buf_t *buf) { int referenced; mutex_enter(&buf->b_evict_lock); referenced = (refcount_count(&buf->b_hdr->b_refcnt)); mutex_exit(&buf->b_evict_lock); return (referenced); } #endif static void arc_write_ready(zio_t *zio) { arc_write_callback_t *callback = zio->io_private; arc_buf_t *buf = callback->awcb_buf; arc_buf_hdr_t *hdr = buf->b_hdr; ASSERT(!refcount_is_zero(&buf->b_hdr->b_refcnt)); callback->awcb_ready(zio, buf, callback->awcb_private); /* * If the IO is already in progress, then this is a re-write * attempt, so we need to thaw and re-compute the cksum. * It is the responsibility of the callback to handle the * accounting for any re-write attempt. */ if (HDR_IO_IN_PROGRESS(hdr)) { mutex_enter(&hdr->b_freeze_lock); if (hdr->b_freeze_cksum != NULL) { kmem_free(hdr->b_freeze_cksum, sizeof (zio_cksum_t)); hdr->b_freeze_cksum = NULL; } mutex_exit(&hdr->b_freeze_lock); } arc_cksum_compute(buf, B_FALSE); hdr->b_flags |= ARC_IO_IN_PROGRESS; } /* * The SPA calls this callback for each physical write that happens on behalf * of a logical write. See the comment in dbuf_write_physdone() for details. */ static void arc_write_physdone(zio_t *zio) { arc_write_callback_t *cb = zio->io_private; if (cb->awcb_physdone != NULL) cb->awcb_physdone(zio, cb->awcb_buf, cb->awcb_private); } static void arc_write_done(zio_t *zio) { arc_write_callback_t *callback = zio->io_private; arc_buf_t *buf = callback->awcb_buf; arc_buf_hdr_t *hdr = buf->b_hdr; ASSERT(hdr->b_acb == NULL); if (zio->io_error == 0) { if (BP_IS_HOLE(zio->io_bp)) { buf_discard_identity(hdr); } else { hdr->b_dva = *BP_IDENTITY(zio->io_bp); hdr->b_birth = BP_PHYSICAL_BIRTH(zio->io_bp); hdr->b_cksum0 = zio->io_bp->blk_cksum.zc_word[0]; } } else { ASSERT(BUF_EMPTY(hdr)); } /* * If the block to be written was all-zero, we may have * compressed it away. In this case no write was performed * so there will be no dva/birth/checksum. The buffer must * therefore remain anonymous (and uncached). */ if (!BUF_EMPTY(hdr)) { arc_buf_hdr_t *exists; kmutex_t *hash_lock; ASSERT(zio->io_error == 0); arc_cksum_verify(buf); exists = buf_hash_insert(hdr, &hash_lock); if (exists) { /* * This can only happen if we overwrite for * sync-to-convergence, because we remove * buffers from the hash table when we arc_free(). */ if (zio->io_flags & ZIO_FLAG_IO_REWRITE) { if (!BP_EQUAL(&zio->io_bp_orig, zio->io_bp)) panic("bad overwrite, hdr=%p exists=%p", (void *)hdr, (void *)exists); ASSERT(refcount_is_zero(&exists->b_refcnt)); arc_change_state(arc_anon, exists, hash_lock); mutex_exit(hash_lock); arc_hdr_destroy(exists); exists = buf_hash_insert(hdr, &hash_lock); ASSERT3P(exists, ==, NULL); } else if (zio->io_flags & ZIO_FLAG_NOPWRITE) { /* nopwrite */ ASSERT(zio->io_prop.zp_nopwrite); if (!BP_EQUAL(&zio->io_bp_orig, zio->io_bp)) panic("bad nopwrite, hdr=%p exists=%p", (void *)hdr, (void *)exists); } else { /* Dedup */ ASSERT(hdr->b_datacnt == 1); ASSERT(hdr->b_state == arc_anon); ASSERT(BP_GET_DEDUP(zio->io_bp)); ASSERT(BP_GET_LEVEL(zio->io_bp) == 0); } } hdr->b_flags &= ~ARC_IO_IN_PROGRESS; /* if it's not anon, we are doing a scrub */ if (!exists && hdr->b_state == arc_anon) arc_access(hdr, hash_lock); mutex_exit(hash_lock); } else { hdr->b_flags &= ~ARC_IO_IN_PROGRESS; } ASSERT(!refcount_is_zero(&hdr->b_refcnt)); callback->awcb_done(zio, buf, callback->awcb_private); kmem_free(callback, sizeof (arc_write_callback_t)); } zio_t * arc_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, arc_buf_t *buf, boolean_t l2arc, boolean_t l2arc_compress, const zio_prop_t *zp, arc_done_func_t *ready, arc_done_func_t *physdone, arc_done_func_t *done, void *private, zio_priority_t priority, int zio_flags, const zbookmark_t *zb) { arc_buf_hdr_t *hdr = buf->b_hdr; arc_write_callback_t *callback; zio_t *zio; ASSERT(ready != NULL); ASSERT(done != NULL); ASSERT(!HDR_IO_ERROR(hdr)); ASSERT((hdr->b_flags & ARC_IO_IN_PROGRESS) == 0); ASSERT(hdr->b_acb == NULL); if (l2arc) hdr->b_flags |= ARC_L2CACHE; if (l2arc_compress) hdr->b_flags |= ARC_L2COMPRESS; callback = kmem_zalloc(sizeof (arc_write_callback_t), KM_SLEEP); callback->awcb_ready = ready; callback->awcb_physdone = physdone; callback->awcb_done = done; callback->awcb_private = private; callback->awcb_buf = buf; zio = zio_write(pio, spa, txg, bp, buf->b_data, hdr->b_size, zp, arc_write_ready, arc_write_physdone, arc_write_done, callback, priority, zio_flags, zb); return (zio); } static int arc_memory_throttle(uint64_t reserve, uint64_t txg) { #ifdef _KERNEL uint64_t available_memory = ptoa((uintmax_t)cnt.v_free_count + cnt.v_cache_count); static uint64_t page_load = 0; static uint64_t last_txg = 0; #ifdef sun #if defined(__i386) available_memory = MIN(available_memory, vmem_size(heap_arena, VMEM_FREE)); #endif #endif /* sun */ if (cnt.v_free_count + cnt.v_cache_count > (uint64_t)physmem * arc_lotsfree_percent / 100) return (0); if (txg > last_txg) { last_txg = txg; page_load = 0; } /* * If we are in pageout, we know that memory is already tight, * the arc is already going to be evicting, so we just want to * continue to let page writes occur as quickly as possible. */ if (curproc == pageproc) { if (page_load > available_memory / 4) return (SET_ERROR(ERESTART)); /* Note: reserve is inflated, so we deflate */ page_load += reserve / 8; return (0); } else if (page_load > 0 && arc_reclaim_needed()) { /* memory is low, delay before restarting */ ARCSTAT_INCR(arcstat_memory_throttle_count, 1); return (SET_ERROR(EAGAIN)); } page_load = 0; #endif return (0); } void arc_tempreserve_clear(uint64_t reserve) { atomic_add_64(&arc_tempreserve, -reserve); ASSERT((int64_t)arc_tempreserve >= 0); } int arc_tempreserve_space(uint64_t reserve, uint64_t txg) { int error; uint64_t anon_size; if (reserve > arc_c/4 && !arc_no_grow) arc_c = MIN(arc_c_max, reserve * 4); if (reserve > arc_c) return (SET_ERROR(ENOMEM)); /* * Don't count loaned bufs as in flight dirty data to prevent long * network delays from blocking transactions that are ready to be * assigned to a txg. */ anon_size = MAX((int64_t)(arc_anon->arcs_size - arc_loaned_bytes), 0); /* * Writes will, almost always, require additional memory allocations * in order to compress/encrypt/etc the data. We therefore need to * make sure that there is sufficient available memory for this. */ error = arc_memory_throttle(reserve, txg); if (error != 0) return (error); /* * Throttle writes when the amount of dirty data in the cache * gets too large. We try to keep the cache less than half full * of dirty blocks so that our sync times don't grow too large. * Note: if two requests come in concurrently, we might let them * both succeed, when one of them should fail. Not a huge deal. */ if (reserve + arc_tempreserve + anon_size > arc_c / 2 && anon_size > arc_c / 4) { dprintf("failing, arc_tempreserve=%lluK anon_meta=%lluK " "anon_data=%lluK tempreserve=%lluK arc_c=%lluK\n", arc_tempreserve>>10, arc_anon->arcs_lsize[ARC_BUFC_METADATA]>>10, arc_anon->arcs_lsize[ARC_BUFC_DATA]>>10, reserve>>10, arc_c>>10); return (SET_ERROR(ERESTART)); } atomic_add_64(&arc_tempreserve, reserve); return (0); } static kmutex_t arc_lowmem_lock; #ifdef _KERNEL static eventhandler_tag arc_event_lowmem = NULL; static void arc_lowmem(void *arg __unused, int howto __unused) { /* Serialize access via arc_lowmem_lock. */ mutex_enter(&arc_lowmem_lock); mutex_enter(&arc_reclaim_thr_lock); needfree = 1; cv_signal(&arc_reclaim_thr_cv); /* * It is unsafe to block here in arbitrary threads, because we can come * here from ARC itself and may hold ARC locks and thus risk a deadlock * with ARC reclaim thread. */ if (curproc == pageproc) { while (needfree) msleep(&needfree, &arc_reclaim_thr_lock, 0, "zfs:lowmem", 0); } mutex_exit(&arc_reclaim_thr_lock); mutex_exit(&arc_lowmem_lock); } #endif void arc_init(void) { int i, prefetch_tunable_set = 0; mutex_init(&arc_reclaim_thr_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&arc_reclaim_thr_cv, NULL, CV_DEFAULT, NULL); mutex_init(&arc_lowmem_lock, NULL, MUTEX_DEFAULT, NULL); /* Convert seconds to clock ticks */ arc_min_prefetch_lifespan = 1 * hz; /* Start out with 1/8 of all memory */ arc_c = kmem_size() / 8; #ifdef sun #ifdef _KERNEL /* * On architectures where the physical memory can be larger * than the addressable space (intel in 32-bit mode), we may * need to limit the cache to 1/8 of VM size. */ arc_c = MIN(arc_c, vmem_size(heap_arena, VMEM_ALLOC | VMEM_FREE) / 8); #endif #endif /* sun */ /* set min cache to 1/32 of all memory, or 16MB, whichever is more */ arc_c_min = MAX(arc_c / 4, 16 << 20); /* set max to 1/2 of all memory, or all but 1GB, whichever is more */ if (arc_c * 8 >= 1 << 30) arc_c_max = (arc_c * 8) - (1 << 30); else arc_c_max = arc_c_min; arc_c_max = MAX(arc_c * 5, arc_c_max); #ifdef _KERNEL /* * Allow the tunables to override our calculations if they are * reasonable (ie. over 16MB) */ if (zfs_arc_max > 16 << 20 && zfs_arc_max < kmem_size()) arc_c_max = zfs_arc_max; if (zfs_arc_min > 16 << 20 && zfs_arc_min <= arc_c_max) arc_c_min = zfs_arc_min; #endif arc_c = arc_c_max; arc_p = (arc_c >> 1); /* limit meta-data to 1/4 of the arc capacity */ arc_meta_limit = arc_c_max / 4; /* Allow the tunable to override if it is reasonable */ if (zfs_arc_meta_limit > 0 && zfs_arc_meta_limit <= arc_c_max) arc_meta_limit = zfs_arc_meta_limit; if (arc_c_min < arc_meta_limit / 2 && zfs_arc_min == 0) arc_c_min = arc_meta_limit / 2; if (zfs_arc_grow_retry > 0) arc_grow_retry = zfs_arc_grow_retry; if (zfs_arc_shrink_shift > 0) arc_shrink_shift = zfs_arc_shrink_shift; if (zfs_arc_p_min_shift > 0) arc_p_min_shift = zfs_arc_p_min_shift; /* if kmem_flags are set, lets try to use less memory */ if (kmem_debugging()) arc_c = arc_c / 2; if (arc_c < arc_c_min) arc_c = arc_c_min; zfs_arc_min = arc_c_min; zfs_arc_max = arc_c_max; arc_anon = &ARC_anon; arc_mru = &ARC_mru; arc_mru_ghost = &ARC_mru_ghost; arc_mfu = &ARC_mfu; arc_mfu_ghost = &ARC_mfu_ghost; arc_l2c_only = &ARC_l2c_only; arc_size = 0; for (i = 0; i < ARC_BUFC_NUMLISTS; i++) { mutex_init(&arc_anon->arcs_locks[i].arcs_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&arc_mru->arcs_locks[i].arcs_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&arc_mru_ghost->arcs_locks[i].arcs_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&arc_mfu->arcs_locks[i].arcs_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&arc_mfu_ghost->arcs_locks[i].arcs_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&arc_l2c_only->arcs_locks[i].arcs_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&arc_mru->arcs_lists[i], sizeof (arc_buf_hdr_t), offsetof(arc_buf_hdr_t, b_arc_node)); list_create(&arc_mru_ghost->arcs_lists[i], sizeof (arc_buf_hdr_t), offsetof(arc_buf_hdr_t, b_arc_node)); list_create(&arc_mfu->arcs_lists[i], sizeof (arc_buf_hdr_t), offsetof(arc_buf_hdr_t, b_arc_node)); list_create(&arc_mfu_ghost->arcs_lists[i], sizeof (arc_buf_hdr_t), offsetof(arc_buf_hdr_t, b_arc_node)); list_create(&arc_mfu_ghost->arcs_lists[i], sizeof (arc_buf_hdr_t), offsetof(arc_buf_hdr_t, b_arc_node)); list_create(&arc_l2c_only->arcs_lists[i], sizeof (arc_buf_hdr_t), offsetof(arc_buf_hdr_t, b_arc_node)); } buf_init(); arc_thread_exit = 0; arc_eviction_list = NULL; mutex_init(&arc_eviction_mtx, NULL, MUTEX_DEFAULT, NULL); bzero(&arc_eviction_hdr, sizeof (arc_buf_hdr_t)); arc_ksp = kstat_create("zfs", 0, "arcstats", "misc", KSTAT_TYPE_NAMED, sizeof (arc_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); if (arc_ksp != NULL) { arc_ksp->ks_data = &arc_stats; kstat_install(arc_ksp); } (void) thread_create(NULL, 0, arc_reclaim_thread, NULL, 0, &p0, TS_RUN, minclsyspri); #ifdef _KERNEL arc_event_lowmem = EVENTHANDLER_REGISTER(vm_lowmem, arc_lowmem, NULL, EVENTHANDLER_PRI_FIRST); #endif arc_dead = FALSE; arc_warm = B_FALSE; /* * Calculate maximum amount of dirty data per pool. * * If it has been set by /etc/system, take that. * Otherwise, use a percentage of physical memory defined by * zfs_dirty_data_max_percent (default 10%) with a cap at * zfs_dirty_data_max_max (default 4GB). */ if (zfs_dirty_data_max == 0) { zfs_dirty_data_max = ptob(physmem) * zfs_dirty_data_max_percent / 100; zfs_dirty_data_max = MIN(zfs_dirty_data_max, zfs_dirty_data_max_max); } #ifdef _KERNEL if (TUNABLE_INT_FETCH("vfs.zfs.prefetch_disable", &zfs_prefetch_disable)) prefetch_tunable_set = 1; #ifdef __i386__ if (prefetch_tunable_set == 0) { printf("ZFS NOTICE: Prefetch is disabled by default on i386 " "-- to enable,\n"); printf(" add \"vfs.zfs.prefetch_disable=0\" " "to /boot/loader.conf.\n"); zfs_prefetch_disable = 1; } #else if ((((uint64_t)physmem * PAGESIZE) < (1ULL << 32)) && prefetch_tunable_set == 0) { printf("ZFS NOTICE: Prefetch is disabled by default if less " "than 4GB of RAM is present;\n" " to enable, add \"vfs.zfs.prefetch_disable=0\" " "to /boot/loader.conf.\n"); zfs_prefetch_disable = 1; } #endif /* Warn about ZFS memory and address space requirements. */ if (((uint64_t)physmem * PAGESIZE) < (256 + 128 + 64) * (1 << 20)) { printf("ZFS WARNING: Recommended minimum RAM size is 512MB; " "expect unstable behavior.\n"); } if (kmem_size() < 512 * (1 << 20)) { printf("ZFS WARNING: Recommended minimum kmem_size is 512MB; " "expect unstable behavior.\n"); printf(" Consider tuning vm.kmem_size and " "vm.kmem_size_max\n"); printf(" in /boot/loader.conf.\n"); } #endif } void arc_fini(void) { int i; mutex_enter(&arc_reclaim_thr_lock); arc_thread_exit = 1; cv_signal(&arc_reclaim_thr_cv); while (arc_thread_exit != 0) cv_wait(&arc_reclaim_thr_cv, &arc_reclaim_thr_lock); mutex_exit(&arc_reclaim_thr_lock); arc_flush(NULL); arc_dead = TRUE; if (arc_ksp != NULL) { kstat_delete(arc_ksp); arc_ksp = NULL; } mutex_destroy(&arc_eviction_mtx); mutex_destroy(&arc_reclaim_thr_lock); cv_destroy(&arc_reclaim_thr_cv); for (i = 0; i < ARC_BUFC_NUMLISTS; i++) { list_destroy(&arc_mru->arcs_lists[i]); list_destroy(&arc_mru_ghost->arcs_lists[i]); list_destroy(&arc_mfu->arcs_lists[i]); list_destroy(&arc_mfu_ghost->arcs_lists[i]); list_destroy(&arc_l2c_only->arcs_lists[i]); mutex_destroy(&arc_anon->arcs_locks[i].arcs_lock); mutex_destroy(&arc_mru->arcs_locks[i].arcs_lock); mutex_destroy(&arc_mru_ghost->arcs_locks[i].arcs_lock); mutex_destroy(&arc_mfu->arcs_locks[i].arcs_lock); mutex_destroy(&arc_mfu_ghost->arcs_locks[i].arcs_lock); mutex_destroy(&arc_l2c_only->arcs_locks[i].arcs_lock); } buf_fini(); ASSERT(arc_loaned_bytes == 0); mutex_destroy(&arc_lowmem_lock); #ifdef _KERNEL if (arc_event_lowmem != NULL) EVENTHANDLER_DEREGISTER(vm_lowmem, arc_event_lowmem); #endif } /* * Level 2 ARC * * The level 2 ARC (L2ARC) is a cache layer in-between main memory and disk. * It uses dedicated storage devices to hold cached data, which are populated * using large infrequent writes. The main role of this cache is to boost * the performance of random read workloads. The intended L2ARC devices * include short-stroked disks, solid state disks, and other media with * substantially faster read latency than disk. * * +-----------------------+ * | ARC | * +-----------------------+ * | ^ ^ * | | | * l2arc_feed_thread() arc_read() * | | | * | l2arc read | * V | | * +---------------+ | * | L2ARC | | * +---------------+ | * | ^ | * l2arc_write() | | * | | | * V | | * +-------+ +-------+ * | vdev | | vdev | * | cache | | cache | * +-------+ +-------+ * +=========+ .-----. * : L2ARC : |-_____-| * : devices : | Disks | * +=========+ `-_____-' * * Read requests are satisfied from the following sources, in order: * * 1) ARC * 2) vdev cache of L2ARC devices * 3) L2ARC devices * 4) vdev cache of disks * 5) disks * * Some L2ARC device types exhibit extremely slow write performance. * To accommodate for this there are some significant differences between * the L2ARC and traditional cache design: * * 1. There is no eviction path from the ARC to the L2ARC. Evictions from * the ARC behave as usual, freeing buffers and placing headers on ghost * lists. The ARC does not send buffers to the L2ARC during eviction as * this would add inflated write latencies for all ARC memory pressure. * * 2. The L2ARC attempts to cache data from the ARC before it is evicted. * It does this by periodically scanning buffers from the eviction-end of * the MFU and MRU ARC lists, copying them to the L2ARC devices if they are * not already there. It scans until a headroom of buffers is satisfied, * which itself is a buffer for ARC eviction. If a compressible buffer is * found during scanning and selected for writing to an L2ARC device, we * temporarily boost scanning headroom during the next scan cycle to make * sure we adapt to compression effects (which might significantly reduce * the data volume we write to L2ARC). The thread that does this is * l2arc_feed_thread(), illustrated below; example sizes are included to * provide a better sense of ratio than this diagram: * * head --> tail * +---------------------+----------+ * ARC_mfu |:::::#:::::::::::::::|o#o###o###|-->. # already on L2ARC * +---------------------+----------+ | o L2ARC eligible * ARC_mru |:#:::::::::::::::::::|#o#ooo####|-->| : ARC buffer * +---------------------+----------+ | * 15.9 Gbytes ^ 32 Mbytes | * headroom | * l2arc_feed_thread() * | * l2arc write hand <--[oooo]--' * | 8 Mbyte * | write max * V * +==============================+ * L2ARC dev |####|#|###|###| |####| ... | * +==============================+ * 32 Gbytes * * 3. If an ARC buffer is copied to the L2ARC but then hit instead of * evicted, then the L2ARC has cached a buffer much sooner than it probably * needed to, potentially wasting L2ARC device bandwidth and storage. It is * safe to say that this is an uncommon case, since buffers at the end of * the ARC lists have moved there due to inactivity. * * 4. If the ARC evicts faster than the L2ARC can maintain a headroom, * then the L2ARC simply misses copying some buffers. This serves as a * pressure valve to prevent heavy read workloads from both stalling the ARC * with waits and clogging the L2ARC with writes. This also helps prevent * the potential for the L2ARC to churn if it attempts to cache content too * quickly, such as during backups of the entire pool. * * 5. After system boot and before the ARC has filled main memory, there are * no evictions from the ARC and so the tails of the ARC_mfu and ARC_mru * lists can remain mostly static. Instead of searching from tail of these * lists as pictured, the l2arc_feed_thread() will search from the list heads * for eligible buffers, greatly increasing its chance of finding them. * * The L2ARC device write speed is also boosted during this time so that * the L2ARC warms up faster. Since there have been no ARC evictions yet, * there are no L2ARC reads, and no fear of degrading read performance * through increased writes. * * 6. Writes to the L2ARC devices are grouped and sent in-sequence, so that * the vdev queue can aggregate them into larger and fewer writes. Each * device is written to in a rotor fashion, sweeping writes through * available space then repeating. * * 7. The L2ARC does not store dirty content. It never needs to flush * write buffers back to disk based storage. * * 8. If an ARC buffer is written (and dirtied) which also exists in the * L2ARC, the now stale L2ARC buffer is immediately dropped. * * The performance of the L2ARC can be tweaked by a number of tunables, which * may be necessary for different workloads: * * l2arc_write_max max write bytes per interval * l2arc_write_boost extra write bytes during device warmup * l2arc_noprefetch skip caching prefetched buffers * l2arc_headroom number of max device writes to precache * l2arc_headroom_boost when we find compressed buffers during ARC * scanning, we multiply headroom by this * percentage factor for the next scan cycle, * since more compressed buffers are likely to * be present * l2arc_feed_secs seconds between L2ARC writing * * Tunables may be removed or added as future performance improvements are * integrated, and also may become zpool properties. * * There are three key functions that control how the L2ARC warms up: * * l2arc_write_eligible() check if a buffer is eligible to cache * l2arc_write_size() calculate how much to write * l2arc_write_interval() calculate sleep delay between writes * * These three functions determine what to write, how much, and how quickly * to send writes. */ static boolean_t l2arc_write_eligible(uint64_t spa_guid, arc_buf_hdr_t *ab) { /* * A buffer is *not* eligible for the L2ARC if it: * 1. belongs to a different spa. * 2. is already cached on the L2ARC. * 3. has an I/O in progress (it may be an incomplete read). * 4. is flagged not eligible (zfs property). */ if (ab->b_spa != spa_guid) { ARCSTAT_BUMP(arcstat_l2_write_spa_mismatch); return (B_FALSE); } if (ab->b_l2hdr != NULL) { ARCSTAT_BUMP(arcstat_l2_write_in_l2); return (B_FALSE); } if (HDR_IO_IN_PROGRESS(ab)) { ARCSTAT_BUMP(arcstat_l2_write_hdr_io_in_progress); return (B_FALSE); } if (!HDR_L2CACHE(ab)) { ARCSTAT_BUMP(arcstat_l2_write_not_cacheable); return (B_FALSE); } return (B_TRUE); } static uint64_t l2arc_write_size(void) { uint64_t size; /* * Make sure our globals have meaningful values in case the user * altered them. */ size = l2arc_write_max; if (size == 0) { cmn_err(CE_NOTE, "Bad value for l2arc_write_max, value must " "be greater than zero, resetting it to the default (%d)", L2ARC_WRITE_SIZE); size = l2arc_write_max = L2ARC_WRITE_SIZE; } if (arc_warm == B_FALSE) size += l2arc_write_boost; return (size); } static clock_t l2arc_write_interval(clock_t began, uint64_t wanted, uint64_t wrote) { clock_t interval, next, now; /* * If the ARC lists are busy, increase our write rate; if the * lists are stale, idle back. This is achieved by checking * how much we previously wrote - if it was more than half of * what we wanted, schedule the next write much sooner. */ if (l2arc_feed_again && wrote > (wanted / 2)) interval = (hz * l2arc_feed_min_ms) / 1000; else interval = hz * l2arc_feed_secs; now = ddi_get_lbolt(); next = MAX(now, MIN(now + interval, began + interval)); return (next); } static void l2arc_hdr_stat_add(void) { ARCSTAT_INCR(arcstat_l2_hdr_size, HDR_SIZE + L2HDR_SIZE); ARCSTAT_INCR(arcstat_hdr_size, -HDR_SIZE); } static void l2arc_hdr_stat_remove(void) { ARCSTAT_INCR(arcstat_l2_hdr_size, -(HDR_SIZE + L2HDR_SIZE)); ARCSTAT_INCR(arcstat_hdr_size, HDR_SIZE); } /* * Cycle through L2ARC devices. This is how L2ARC load balances. * If a device is returned, this also returns holding the spa config lock. */ static l2arc_dev_t * l2arc_dev_get_next(void) { l2arc_dev_t *first, *next = NULL; /* * Lock out the removal of spas (spa_namespace_lock), then removal * of cache devices (l2arc_dev_mtx). Once a device has been selected, * both locks will be dropped and a spa config lock held instead. */ mutex_enter(&spa_namespace_lock); mutex_enter(&l2arc_dev_mtx); /* if there are no vdevs, there is nothing to do */ if (l2arc_ndev == 0) goto out; first = NULL; next = l2arc_dev_last; do { /* loop around the list looking for a non-faulted vdev */ if (next == NULL) { next = list_head(l2arc_dev_list); } else { next = list_next(l2arc_dev_list, next); if (next == NULL) next = list_head(l2arc_dev_list); } /* if we have come back to the start, bail out */ if (first == NULL) first = next; else if (next == first) break; } while (vdev_is_dead(next->l2ad_vdev)); /* if we were unable to find any usable vdevs, return NULL */ if (vdev_is_dead(next->l2ad_vdev)) next = NULL; l2arc_dev_last = next; out: mutex_exit(&l2arc_dev_mtx); /* * Grab the config lock to prevent the 'next' device from being * removed while we are writing to it. */ if (next != NULL) spa_config_enter(next->l2ad_spa, SCL_L2ARC, next, RW_READER); mutex_exit(&spa_namespace_lock); return (next); } /* * Free buffers that were tagged for destruction. */ static void l2arc_do_free_on_write() { list_t *buflist; l2arc_data_free_t *df, *df_prev; mutex_enter(&l2arc_free_on_write_mtx); buflist = l2arc_free_on_write; for (df = list_tail(buflist); df; df = df_prev) { df_prev = list_prev(buflist, df); ASSERT(df->l2df_data != NULL); ASSERT(df->l2df_func != NULL); df->l2df_func(df->l2df_data, df->l2df_size); list_remove(buflist, df); kmem_free(df, sizeof (l2arc_data_free_t)); } mutex_exit(&l2arc_free_on_write_mtx); } /* * A write to a cache device has completed. Update all headers to allow * reads from these buffers to begin. */ static void l2arc_write_done(zio_t *zio) { l2arc_write_callback_t *cb; l2arc_dev_t *dev; list_t *buflist; arc_buf_hdr_t *head, *ab, *ab_prev; l2arc_buf_hdr_t *abl2; kmutex_t *hash_lock; cb = zio->io_private; ASSERT(cb != NULL); dev = cb->l2wcb_dev; ASSERT(dev != NULL); head = cb->l2wcb_head; ASSERT(head != NULL); buflist = dev->l2ad_buflist; ASSERT(buflist != NULL); DTRACE_PROBE2(l2arc__iodone, zio_t *, zio, l2arc_write_callback_t *, cb); if (zio->io_error != 0) ARCSTAT_BUMP(arcstat_l2_writes_error); mutex_enter(&l2arc_buflist_mtx); /* * All writes completed, or an error was hit. */ for (ab = list_prev(buflist, head); ab; ab = ab_prev) { ab_prev = list_prev(buflist, ab); abl2 = ab->b_l2hdr; /* * Release the temporary compressed buffer as soon as possible. */ if (abl2->b_compress != ZIO_COMPRESS_OFF) l2arc_release_cdata_buf(ab); hash_lock = HDR_LOCK(ab); if (!mutex_tryenter(hash_lock)) { /* * This buffer misses out. It may be in a stage * of eviction. Its ARC_L2_WRITING flag will be * left set, denying reads to this buffer. */ ARCSTAT_BUMP(arcstat_l2_writes_hdr_miss); continue; } if (zio->io_error != 0) { /* * Error - drop L2ARC entry. */ list_remove(buflist, ab); ARCSTAT_INCR(arcstat_l2_asize, -abl2->b_asize); ab->b_l2hdr = NULL; l2arc_trim(abl2); kmem_free(abl2, sizeof (l2arc_buf_hdr_t)); ARCSTAT_INCR(arcstat_l2_size, -ab->b_size); } /* * Allow ARC to begin reads to this L2ARC entry. */ ab->b_flags &= ~ARC_L2_WRITING; mutex_exit(hash_lock); } atomic_inc_64(&l2arc_writes_done); list_remove(buflist, head); kmem_cache_free(hdr_cache, head); mutex_exit(&l2arc_buflist_mtx); l2arc_do_free_on_write(); kmem_free(cb, sizeof (l2arc_write_callback_t)); } /* * A read to a cache device completed. Validate buffer contents before * handing over to the regular ARC routines. */ static void l2arc_read_done(zio_t *zio) { l2arc_read_callback_t *cb; arc_buf_hdr_t *hdr; arc_buf_t *buf; kmutex_t *hash_lock; int equal; ASSERT(zio->io_vd != NULL); ASSERT(zio->io_flags & ZIO_FLAG_DONT_PROPAGATE); spa_config_exit(zio->io_spa, SCL_L2ARC, zio->io_vd); cb = zio->io_private; ASSERT(cb != NULL); buf = cb->l2rcb_buf; ASSERT(buf != NULL); hash_lock = HDR_LOCK(buf->b_hdr); mutex_enter(hash_lock); hdr = buf->b_hdr; ASSERT3P(hash_lock, ==, HDR_LOCK(hdr)); /* * If the buffer was compressed, decompress it first. */ if (cb->l2rcb_compress != ZIO_COMPRESS_OFF) l2arc_decompress_zio(zio, hdr, cb->l2rcb_compress); ASSERT(zio->io_data != NULL); /* * Check this survived the L2ARC journey. */ equal = arc_cksum_equal(buf); if (equal && zio->io_error == 0 && !HDR_L2_EVICTED(hdr)) { mutex_exit(hash_lock); zio->io_private = buf; zio->io_bp_copy = cb->l2rcb_bp; /* XXX fix in L2ARC 2.0 */ zio->io_bp = &zio->io_bp_copy; /* XXX fix in L2ARC 2.0 */ arc_read_done(zio); } else { mutex_exit(hash_lock); /* * Buffer didn't survive caching. Increment stats and * reissue to the original storage device. */ if (zio->io_error != 0) { ARCSTAT_BUMP(arcstat_l2_io_error); } else { zio->io_error = SET_ERROR(EIO); } if (!equal) ARCSTAT_BUMP(arcstat_l2_cksum_bad); /* * If there's no waiter, issue an async i/o to the primary * storage now. If there *is* a waiter, the caller must * issue the i/o in a context where it's OK to block. */ if (zio->io_waiter == NULL) { zio_t *pio = zio_unique_parent(zio); ASSERT(!pio || pio->io_child_type == ZIO_CHILD_LOGICAL); zio_nowait(zio_read(pio, cb->l2rcb_spa, &cb->l2rcb_bp, buf->b_data, zio->io_size, arc_read_done, buf, zio->io_priority, cb->l2rcb_flags, &cb->l2rcb_zb)); } } kmem_free(cb, sizeof (l2arc_read_callback_t)); } /* * This is the list priority from which the L2ARC will search for pages to * cache. This is used within loops (0..3) to cycle through lists in the * desired order. This order can have a significant effect on cache * performance. * * Currently the metadata lists are hit first, MFU then MRU, followed by * the data lists. This function returns a locked list, and also returns * the lock pointer. */ static list_t * l2arc_list_locked(int list_num, kmutex_t **lock) { list_t *list = NULL; int idx; ASSERT(list_num >= 0 && list_num < 2 * ARC_BUFC_NUMLISTS); if (list_num < ARC_BUFC_NUMMETADATALISTS) { idx = list_num; list = &arc_mfu->arcs_lists[idx]; *lock = ARCS_LOCK(arc_mfu, idx); } else if (list_num < ARC_BUFC_NUMMETADATALISTS * 2) { idx = list_num - ARC_BUFC_NUMMETADATALISTS; list = &arc_mru->arcs_lists[idx]; *lock = ARCS_LOCK(arc_mru, idx); } else if (list_num < (ARC_BUFC_NUMMETADATALISTS * 2 + ARC_BUFC_NUMDATALISTS)) { idx = list_num - ARC_BUFC_NUMMETADATALISTS; list = &arc_mfu->arcs_lists[idx]; *lock = ARCS_LOCK(arc_mfu, idx); } else { idx = list_num - ARC_BUFC_NUMLISTS; list = &arc_mru->arcs_lists[idx]; *lock = ARCS_LOCK(arc_mru, idx); } ASSERT(!(MUTEX_HELD(*lock))); mutex_enter(*lock); return (list); } /* * Evict buffers from the device write hand to the distance specified in * bytes. This distance may span populated buffers, it may span nothing. * This is clearing a region on the L2ARC device ready for writing. * If the 'all' boolean is set, every buffer is evicted. */ static void l2arc_evict(l2arc_dev_t *dev, uint64_t distance, boolean_t all) { list_t *buflist; l2arc_buf_hdr_t *abl2; arc_buf_hdr_t *ab, *ab_prev; kmutex_t *hash_lock; uint64_t taddr; buflist = dev->l2ad_buflist; if (buflist == NULL) return; if (!all && dev->l2ad_first) { /* * This is the first sweep through the device. There is * nothing to evict. */ return; } if (dev->l2ad_hand >= (dev->l2ad_end - (2 * distance))) { /* * When nearing the end of the device, evict to the end * before the device write hand jumps to the start. */ taddr = dev->l2ad_end; } else { taddr = dev->l2ad_hand + distance; } DTRACE_PROBE4(l2arc__evict, l2arc_dev_t *, dev, list_t *, buflist, uint64_t, taddr, boolean_t, all); top: mutex_enter(&l2arc_buflist_mtx); for (ab = list_tail(buflist); ab; ab = ab_prev) { ab_prev = list_prev(buflist, ab); hash_lock = HDR_LOCK(ab); if (!mutex_tryenter(hash_lock)) { /* * Missed the hash lock. Retry. */ ARCSTAT_BUMP(arcstat_l2_evict_lock_retry); mutex_exit(&l2arc_buflist_mtx); mutex_enter(hash_lock); mutex_exit(hash_lock); goto top; } if (HDR_L2_WRITE_HEAD(ab)) { /* * We hit a write head node. Leave it for * l2arc_write_done(). */ list_remove(buflist, ab); mutex_exit(hash_lock); continue; } if (!all && ab->b_l2hdr != NULL && (ab->b_l2hdr->b_daddr > taddr || ab->b_l2hdr->b_daddr < dev->l2ad_hand)) { /* * We've evicted to the target address, * or the end of the device. */ mutex_exit(hash_lock); break; } if (HDR_FREE_IN_PROGRESS(ab)) { /* * Already on the path to destruction. */ mutex_exit(hash_lock); continue; } if (ab->b_state == arc_l2c_only) { ASSERT(!HDR_L2_READING(ab)); /* * This doesn't exist in the ARC. Destroy. * arc_hdr_destroy() will call list_remove() * and decrement arcstat_l2_size. */ arc_change_state(arc_anon, ab, hash_lock); arc_hdr_destroy(ab); } else { /* * Invalidate issued or about to be issued * reads, since we may be about to write * over this location. */ if (HDR_L2_READING(ab)) { ARCSTAT_BUMP(arcstat_l2_evict_reading); ab->b_flags |= ARC_L2_EVICTED; } /* * Tell ARC this no longer exists in L2ARC. */ if (ab->b_l2hdr != NULL) { abl2 = ab->b_l2hdr; ARCSTAT_INCR(arcstat_l2_asize, -abl2->b_asize); ab->b_l2hdr = NULL; kmem_free(abl2, sizeof (l2arc_buf_hdr_t)); ARCSTAT_INCR(arcstat_l2_size, -ab->b_size); } list_remove(buflist, ab); /* * This may have been leftover after a * failed write. */ ab->b_flags &= ~ARC_L2_WRITING; } mutex_exit(hash_lock); } mutex_exit(&l2arc_buflist_mtx); vdev_space_update(dev->l2ad_vdev, -(taddr - dev->l2ad_evict), 0, 0); dev->l2ad_evict = taddr; } /* * Find and write ARC buffers to the L2ARC device. * * An ARC_L2_WRITING flag is set so that the L2ARC buffers are not valid * for reading until they have completed writing. * The headroom_boost is an in-out parameter used to maintain headroom boost * state between calls to this function. * * Returns the number of bytes actually written (which may be smaller than * the delta by which the device hand has changed due to alignment). */ static uint64_t l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz, boolean_t *headroom_boost) { arc_buf_hdr_t *ab, *ab_prev, *head; list_t *list; uint64_t write_asize, write_sz, headroom, buf_compress_minsz; void *buf_data; kmutex_t *list_lock; boolean_t full; l2arc_write_callback_t *cb; zio_t *pio, *wzio; uint64_t guid = spa_load_guid(spa); const boolean_t do_headroom_boost = *headroom_boost; int try; ASSERT(dev->l2ad_vdev != NULL); /* Lower the flag now, we might want to raise it again later. */ *headroom_boost = B_FALSE; pio = NULL; write_sz = write_asize = 0; full = B_FALSE; head = kmem_cache_alloc(hdr_cache, KM_PUSHPAGE); head->b_flags |= ARC_L2_WRITE_HEAD; ARCSTAT_BUMP(arcstat_l2_write_buffer_iter); /* * We will want to try to compress buffers that are at least 2x the * device sector size. */ buf_compress_minsz = 2 << dev->l2ad_vdev->vdev_ashift; /* * Copy buffers for L2ARC writing. */ mutex_enter(&l2arc_buflist_mtx); for (try = 0; try < 2 * ARC_BUFC_NUMLISTS; try++) { uint64_t passed_sz = 0; list = l2arc_list_locked(try, &list_lock); ARCSTAT_BUMP(arcstat_l2_write_buffer_list_iter); /* * L2ARC fast warmup. * * Until the ARC is warm and starts to evict, read from the * head of the ARC lists rather than the tail. */ if (arc_warm == B_FALSE) ab = list_head(list); else ab = list_tail(list); if (ab == NULL) ARCSTAT_BUMP(arcstat_l2_write_buffer_list_null_iter); headroom = target_sz * l2arc_headroom * 2 / ARC_BUFC_NUMLISTS; if (do_headroom_boost) headroom = (headroom * l2arc_headroom_boost) / 100; for (; ab; ab = ab_prev) { l2arc_buf_hdr_t *l2hdr; kmutex_t *hash_lock; uint64_t buf_sz; uint64_t buf_a_sz; if (arc_warm == B_FALSE) ab_prev = list_next(list, ab); else ab_prev = list_prev(list, ab); ARCSTAT_INCR(arcstat_l2_write_buffer_bytes_scanned, ab->b_size); hash_lock = HDR_LOCK(ab); if (!mutex_tryenter(hash_lock)) { ARCSTAT_BUMP(arcstat_l2_write_trylock_fail); /* * Skip this buffer rather than waiting. */ continue; } passed_sz += ab->b_size; if (passed_sz > headroom) { /* * Searched too far. */ mutex_exit(hash_lock); ARCSTAT_BUMP(arcstat_l2_write_passed_headroom); break; } if (!l2arc_write_eligible(guid, ab)) { mutex_exit(hash_lock); continue; } /* * Assume that the buffer is not going to be compressed * and could take more space on disk because of a larger * disk block size. */ buf_sz = ab->b_size; buf_a_sz = vdev_psize_to_asize(dev->l2ad_vdev, buf_sz); if ((write_asize + buf_a_sz) > target_sz) { full = B_TRUE; mutex_exit(hash_lock); ARCSTAT_BUMP(arcstat_l2_write_full); break; } if (pio == NULL) { /* * Insert a dummy header on the buflist so * l2arc_write_done() can find where the * write buffers begin without searching. */ list_insert_head(dev->l2ad_buflist, head); cb = kmem_alloc( sizeof (l2arc_write_callback_t), KM_SLEEP); cb->l2wcb_dev = dev; cb->l2wcb_head = head; pio = zio_root(spa, l2arc_write_done, cb, ZIO_FLAG_CANFAIL); ARCSTAT_BUMP(arcstat_l2_write_pios); } /* * Create and add a new L2ARC header. */ l2hdr = kmem_zalloc(sizeof (l2arc_buf_hdr_t), KM_SLEEP); l2hdr->b_dev = dev; ab->b_flags |= ARC_L2_WRITING; /* * Temporarily stash the data buffer in b_tmp_cdata. * The subsequent write step will pick it up from * there. This is because can't access ab->b_buf * without holding the hash_lock, which we in turn * can't access without holding the ARC list locks * (which we want to avoid during compression/writing). */ l2hdr->b_compress = ZIO_COMPRESS_OFF; l2hdr->b_asize = ab->b_size; l2hdr->b_tmp_cdata = ab->b_buf->b_data; ab->b_l2hdr = l2hdr; list_insert_head(dev->l2ad_buflist, ab); /* * Compute and store the buffer cksum before * writing. On debug the cksum is verified first. */ arc_cksum_verify(ab->b_buf); arc_cksum_compute(ab->b_buf, B_TRUE); mutex_exit(hash_lock); write_sz += buf_sz; write_asize += buf_a_sz; } mutex_exit(list_lock); if (full == B_TRUE) break; } /* No buffers selected for writing? */ if (pio == NULL) { ASSERT0(write_sz); mutex_exit(&l2arc_buflist_mtx); kmem_cache_free(hdr_cache, head); return (0); } /* * Note that elsewhere in this file arcstat_l2_asize * and the used space on l2ad_vdev are updated using b_asize, * which is not necessarily rounded up to the device block size. * Too keep accounting consistent we do the same here as well: * stats_size accumulates the sum of b_asize of the written buffers, * while write_asize accumulates the sum of b_asize rounded up * to the device block size. * The latter sum is used only to validate the corectness of the code. */ uint64_t stats_size = 0; write_asize = 0; /* * Now start writing the buffers. We're starting at the write head * and work backwards, retracing the course of the buffer selector * loop above. */ for (ab = list_prev(dev->l2ad_buflist, head); ab; ab = list_prev(dev->l2ad_buflist, ab)) { l2arc_buf_hdr_t *l2hdr; uint64_t buf_sz; /* * We shouldn't need to lock the buffer here, since we flagged * it as ARC_L2_WRITING in the previous step, but we must take * care to only access its L2 cache parameters. In particular, * ab->b_buf may be invalid by now due to ARC eviction. */ l2hdr = ab->b_l2hdr; l2hdr->b_daddr = dev->l2ad_hand; if ((ab->b_flags & ARC_L2COMPRESS) && l2hdr->b_asize >= buf_compress_minsz) { if (l2arc_compress_buf(l2hdr)) { /* * If compression succeeded, enable headroom * boost on the next scan cycle. */ *headroom_boost = B_TRUE; } } /* * Pick up the buffer data we had previously stashed away * (and now potentially also compressed). */ buf_data = l2hdr->b_tmp_cdata; buf_sz = l2hdr->b_asize; /* Compression may have squashed the buffer to zero length. */ if (buf_sz != 0) { uint64_t buf_a_sz; wzio = zio_write_phys(pio, dev->l2ad_vdev, dev->l2ad_hand, buf_sz, buf_data, ZIO_CHECKSUM_OFF, NULL, NULL, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_CANFAIL, B_FALSE); DTRACE_PROBE2(l2arc__write, vdev_t *, dev->l2ad_vdev, zio_t *, wzio); (void) zio_nowait(wzio); stats_size += buf_sz; /* * Keep the clock hand suitably device-aligned. */ buf_a_sz = vdev_psize_to_asize(dev->l2ad_vdev, buf_sz); write_asize += buf_a_sz; dev->l2ad_hand += buf_a_sz; } } mutex_exit(&l2arc_buflist_mtx); ASSERT3U(write_asize, <=, target_sz); ARCSTAT_BUMP(arcstat_l2_writes_sent); ARCSTAT_INCR(arcstat_l2_write_bytes, write_asize); ARCSTAT_INCR(arcstat_l2_size, write_sz); ARCSTAT_INCR(arcstat_l2_asize, stats_size); vdev_space_update(dev->l2ad_vdev, stats_size, 0, 0); /* * Bump device hand to the device start if it is approaching the end. * l2arc_evict() will already have evicted ahead for this case. */ if (dev->l2ad_hand >= (dev->l2ad_end - target_sz)) { vdev_space_update(dev->l2ad_vdev, dev->l2ad_end - dev->l2ad_hand, 0, 0); dev->l2ad_hand = dev->l2ad_start; dev->l2ad_evict = dev->l2ad_start; dev->l2ad_first = B_FALSE; } dev->l2ad_writing = B_TRUE; (void) zio_wait(pio); dev->l2ad_writing = B_FALSE; return (write_asize); } /* * Compresses an L2ARC buffer. * The data to be compressed must be prefilled in l2hdr->b_tmp_cdata and its * size in l2hdr->b_asize. This routine tries to compress the data and * depending on the compression result there are three possible outcomes: * *) The buffer was incompressible. The original l2hdr contents were left * untouched and are ready for writing to an L2 device. * *) The buffer was all-zeros, so there is no need to write it to an L2 * device. To indicate this situation b_tmp_cdata is NULL'ed, b_asize is * set to zero and b_compress is set to ZIO_COMPRESS_EMPTY. * *) Compression succeeded and b_tmp_cdata was replaced with a temporary * data buffer which holds the compressed data to be written, and b_asize * tells us how much data there is. b_compress is set to the appropriate * compression algorithm. Once writing is done, invoke * l2arc_release_cdata_buf on this l2hdr to free this temporary buffer. * * Returns B_TRUE if compression succeeded, or B_FALSE if it didn't (the * buffer was incompressible). */ static boolean_t l2arc_compress_buf(l2arc_buf_hdr_t *l2hdr) { void *cdata; size_t csize, len; ASSERT(l2hdr->b_compress == ZIO_COMPRESS_OFF); ASSERT(l2hdr->b_tmp_cdata != NULL); len = l2hdr->b_asize; cdata = zio_data_buf_alloc(len); csize = zio_compress_data(ZIO_COMPRESS_LZ4, l2hdr->b_tmp_cdata, cdata, l2hdr->b_asize, (size_t)(1ULL << l2hdr->b_dev->l2ad_vdev->vdev_ashift)); if (csize == 0) { /* zero block, indicate that there's nothing to write */ zio_data_buf_free(cdata, len); l2hdr->b_compress = ZIO_COMPRESS_EMPTY; l2hdr->b_asize = 0; l2hdr->b_tmp_cdata = NULL; ARCSTAT_BUMP(arcstat_l2_compress_zeros); return (B_TRUE); } else if (csize > 0 && csize < len) { /* * Compression succeeded, we'll keep the cdata around for * writing and release it afterwards. */ l2hdr->b_compress = ZIO_COMPRESS_LZ4; l2hdr->b_asize = csize; l2hdr->b_tmp_cdata = cdata; ARCSTAT_BUMP(arcstat_l2_compress_successes); return (B_TRUE); } else { /* * Compression failed, release the compressed buffer. * l2hdr will be left unmodified. */ zio_data_buf_free(cdata, len); ARCSTAT_BUMP(arcstat_l2_compress_failures); return (B_FALSE); } } /* * Decompresses a zio read back from an l2arc device. On success, the * underlying zio's io_data buffer is overwritten by the uncompressed * version. On decompression error (corrupt compressed stream), the * zio->io_error value is set to signal an I/O error. * * Please note that the compressed data stream is not checksummed, so * if the underlying device is experiencing data corruption, we may feed * corrupt data to the decompressor, so the decompressor needs to be * able to handle this situation (LZ4 does). */ static void l2arc_decompress_zio(zio_t *zio, arc_buf_hdr_t *hdr, enum zio_compress c) { ASSERT(L2ARC_IS_VALID_COMPRESS(c)); if (zio->io_error != 0) { /* * An io error has occured, just restore the original io * size in preparation for a main pool read. */ zio->io_orig_size = zio->io_size = hdr->b_size; return; } if (c == ZIO_COMPRESS_EMPTY) { /* * An empty buffer results in a null zio, which means we * need to fill its io_data after we're done restoring the * buffer's contents. */ ASSERT(hdr->b_buf != NULL); bzero(hdr->b_buf->b_data, hdr->b_size); zio->io_data = zio->io_orig_data = hdr->b_buf->b_data; } else { ASSERT(zio->io_data != NULL); /* * We copy the compressed data from the start of the arc buffer * (the zio_read will have pulled in only what we need, the * rest is garbage which we will overwrite at decompression) * and then decompress back to the ARC data buffer. This way we * can minimize copying by simply decompressing back over the * original compressed data (rather than decompressing to an * aux buffer and then copying back the uncompressed buffer, * which is likely to be much larger). */ uint64_t csize; void *cdata; csize = zio->io_size; cdata = zio_data_buf_alloc(csize); bcopy(zio->io_data, cdata, csize); if (zio_decompress_data(c, cdata, zio->io_data, csize, hdr->b_size) != 0) zio->io_error = EIO; zio_data_buf_free(cdata, csize); } /* Restore the expected uncompressed IO size. */ zio->io_orig_size = zio->io_size = hdr->b_size; } /* * Releases the temporary b_tmp_cdata buffer in an l2arc header structure. * This buffer serves as a temporary holder of compressed data while * the buffer entry is being written to an l2arc device. Once that is * done, we can dispose of it. */ static void l2arc_release_cdata_buf(arc_buf_hdr_t *ab) { l2arc_buf_hdr_t *l2hdr = ab->b_l2hdr; if (l2hdr->b_compress == ZIO_COMPRESS_LZ4) { /* * If the data was compressed, then we've allocated a * temporary buffer for it, so now we need to release it. */ ASSERT(l2hdr->b_tmp_cdata != NULL); zio_data_buf_free(l2hdr->b_tmp_cdata, ab->b_size); } l2hdr->b_tmp_cdata = NULL; } /* * This thread feeds the L2ARC at regular intervals. This is the beating * heart of the L2ARC. */ static void l2arc_feed_thread(void *dummy __unused) { callb_cpr_t cpr; l2arc_dev_t *dev; spa_t *spa; uint64_t size, wrote; clock_t begin, next = ddi_get_lbolt(); boolean_t headroom_boost = B_FALSE; CALLB_CPR_INIT(&cpr, &l2arc_feed_thr_lock, callb_generic_cpr, FTAG); mutex_enter(&l2arc_feed_thr_lock); while (l2arc_thread_exit == 0) { CALLB_CPR_SAFE_BEGIN(&cpr); (void) cv_timedwait(&l2arc_feed_thr_cv, &l2arc_feed_thr_lock, next - ddi_get_lbolt()); CALLB_CPR_SAFE_END(&cpr, &l2arc_feed_thr_lock); next = ddi_get_lbolt() + hz; /* * Quick check for L2ARC devices. */ mutex_enter(&l2arc_dev_mtx); if (l2arc_ndev == 0) { mutex_exit(&l2arc_dev_mtx); continue; } mutex_exit(&l2arc_dev_mtx); begin = ddi_get_lbolt(); /* * This selects the next l2arc device to write to, and in * doing so the next spa to feed from: dev->l2ad_spa. This * will return NULL if there are now no l2arc devices or if * they are all faulted. * * If a device is returned, its spa's config lock is also * held to prevent device removal. l2arc_dev_get_next() * will grab and release l2arc_dev_mtx. */ if ((dev = l2arc_dev_get_next()) == NULL) continue; spa = dev->l2ad_spa; ASSERT(spa != NULL); /* * If the pool is read-only then force the feed thread to * sleep a little longer. */ if (!spa_writeable(spa)) { next = ddi_get_lbolt() + 5 * l2arc_feed_secs * hz; spa_config_exit(spa, SCL_L2ARC, dev); continue; } /* * Avoid contributing to memory pressure. */ if (arc_reclaim_needed()) { ARCSTAT_BUMP(arcstat_l2_abort_lowmem); spa_config_exit(spa, SCL_L2ARC, dev); continue; } ARCSTAT_BUMP(arcstat_l2_feeds); size = l2arc_write_size(); /* * Evict L2ARC buffers that will be overwritten. */ l2arc_evict(dev, size, B_FALSE); /* * Write ARC buffers. */ wrote = l2arc_write_buffers(spa, dev, size, &headroom_boost); /* * Calculate interval between writes. */ next = l2arc_write_interval(begin, size, wrote); spa_config_exit(spa, SCL_L2ARC, dev); } l2arc_thread_exit = 0; cv_broadcast(&l2arc_feed_thr_cv); CALLB_CPR_EXIT(&cpr); /* drops l2arc_feed_thr_lock */ thread_exit(); } boolean_t l2arc_vdev_present(vdev_t *vd) { l2arc_dev_t *dev; mutex_enter(&l2arc_dev_mtx); for (dev = list_head(l2arc_dev_list); dev != NULL; dev = list_next(l2arc_dev_list, dev)) { if (dev->l2ad_vdev == vd) break; } mutex_exit(&l2arc_dev_mtx); return (dev != NULL); } /* * Add a vdev for use by the L2ARC. By this point the spa has already * validated the vdev and opened it. */ void l2arc_add_vdev(spa_t *spa, vdev_t *vd) { l2arc_dev_t *adddev; ASSERT(!l2arc_vdev_present(vd)); vdev_ashift_optimize(vd); /* * Create a new l2arc device entry. */ adddev = kmem_zalloc(sizeof (l2arc_dev_t), KM_SLEEP); adddev->l2ad_spa = spa; adddev->l2ad_vdev = vd; adddev->l2ad_start = VDEV_LABEL_START_SIZE; adddev->l2ad_end = VDEV_LABEL_START_SIZE + vdev_get_min_asize(vd); adddev->l2ad_hand = adddev->l2ad_start; adddev->l2ad_evict = adddev->l2ad_start; adddev->l2ad_first = B_TRUE; adddev->l2ad_writing = B_FALSE; /* * This is a list of all ARC buffers that are still valid on the * device. */ adddev->l2ad_buflist = kmem_zalloc(sizeof (list_t), KM_SLEEP); list_create(adddev->l2ad_buflist, sizeof (arc_buf_hdr_t), offsetof(arc_buf_hdr_t, b_l2node)); vdev_space_update(vd, 0, 0, adddev->l2ad_end - adddev->l2ad_hand); /* * Add device to global list */ mutex_enter(&l2arc_dev_mtx); list_insert_head(l2arc_dev_list, adddev); atomic_inc_64(&l2arc_ndev); mutex_exit(&l2arc_dev_mtx); } /* * Remove a vdev from the L2ARC. */ void l2arc_remove_vdev(vdev_t *vd) { l2arc_dev_t *dev, *nextdev, *remdev = NULL; /* * Find the device by vdev */ mutex_enter(&l2arc_dev_mtx); for (dev = list_head(l2arc_dev_list); dev; dev = nextdev) { nextdev = list_next(l2arc_dev_list, dev); if (vd == dev->l2ad_vdev) { remdev = dev; break; } } ASSERT(remdev != NULL); /* * Remove device from global list */ list_remove(l2arc_dev_list, remdev); l2arc_dev_last = NULL; /* may have been invalidated */ atomic_dec_64(&l2arc_ndev); mutex_exit(&l2arc_dev_mtx); /* * Clear all buflists and ARC references. L2ARC device flush. */ l2arc_evict(remdev, 0, B_TRUE); list_destroy(remdev->l2ad_buflist); kmem_free(remdev->l2ad_buflist, sizeof (list_t)); kmem_free(remdev, sizeof (l2arc_dev_t)); } void l2arc_init(void) { l2arc_thread_exit = 0; l2arc_ndev = 0; l2arc_writes_sent = 0; l2arc_writes_done = 0; mutex_init(&l2arc_feed_thr_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&l2arc_feed_thr_cv, NULL, CV_DEFAULT, NULL); mutex_init(&l2arc_dev_mtx, NULL, MUTEX_DEFAULT, NULL); mutex_init(&l2arc_buflist_mtx, NULL, MUTEX_DEFAULT, NULL); mutex_init(&l2arc_free_on_write_mtx, NULL, MUTEX_DEFAULT, NULL); l2arc_dev_list = &L2ARC_dev_list; l2arc_free_on_write = &L2ARC_free_on_write; list_create(l2arc_dev_list, sizeof (l2arc_dev_t), offsetof(l2arc_dev_t, l2ad_node)); list_create(l2arc_free_on_write, sizeof (l2arc_data_free_t), offsetof(l2arc_data_free_t, l2df_list_node)); } void l2arc_fini(void) { /* * This is called from dmu_fini(), which is called from spa_fini(); * Because of this, we can assume that all l2arc devices have * already been removed when the pools themselves were removed. */ l2arc_do_free_on_write(); mutex_destroy(&l2arc_feed_thr_lock); cv_destroy(&l2arc_feed_thr_cv); mutex_destroy(&l2arc_dev_mtx); mutex_destroy(&l2arc_buflist_mtx); mutex_destroy(&l2arc_free_on_write_mtx); list_destroy(l2arc_dev_list); list_destroy(l2arc_free_on_write); } void l2arc_start(void) { if (!(spa_mode_global & FWRITE)) return; (void) thread_create(NULL, 0, l2arc_feed_thread, NULL, 0, &p0, TS_RUN, minclsyspri); } void l2arc_stop(void) { if (!(spa_mode_global & FWRITE)) return; mutex_enter(&l2arc_feed_thr_lock); cv_signal(&l2arc_feed_thr_cv); /* kick thread out of startup */ l2arc_thread_exit = 1; while (l2arc_thread_exit != 0) cv_wait(&l2arc_feed_thr_cv, &l2arc_feed_thr_lock); mutex_exit(&l2arc_feed_thr_lock); } Index: stable/9/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c =================================================================== --- stable/9/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c (revision 301698) +++ stable/9/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c (revision 301699) @@ -1,7077 +1,7076 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013 by Delphix. All rights reserved. */ /* Portions Copyright 2007 Jeremy Teo */ /* Portions Copyright 2010 Robert Milkowski */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include -#include /* * Programming rules. * * Each vnode op performs some logical unit of work. To do this, the ZPL must * properly lock its in-core state, create a DMU transaction, do the work, * record this work in the intent log (ZIL), commit the DMU transaction, * and wait for the intent log to commit if it is a synchronous operation. * Moreover, the vnode ops must work in both normal and log replay context. * The ordering of events is important to avoid deadlocks and references * to freed memory. The example below illustrates the following Big Rules: * * (1) A check must be made in each zfs thread for a mounted file system. * This is done avoiding races using ZFS_ENTER(zfsvfs). * A ZFS_EXIT(zfsvfs) is needed before all returns. Any znodes * must be checked with ZFS_VERIFY_ZP(zp). Both of these macros * can return EIO from the calling function. * * (2) VN_RELE() should always be the last thing except for zil_commit() * (if necessary) and ZFS_EXIT(). This is for 3 reasons: * First, if it's the last reference, the vnode/znode * can be freed, so the zp may point to freed memory. Second, the last * reference will call zfs_zinactive(), which may induce a lot of work -- * pushing cached pages (which acquires range locks) and syncing out * cached atime changes. Third, zfs_zinactive() may require a new tx, * which could deadlock the system if you were already holding one. * If you must call VN_RELE() within a tx then use VN_RELE_ASYNC(). * * (3) All range locks must be grabbed before calling dmu_tx_assign(), * as they can span dmu_tx_assign() calls. * * (4) If ZPL locks are held, pass TXG_NOWAIT as the second argument to * dmu_tx_assign(). This is critical because we don't want to block * while holding locks. * * If no ZPL locks are held (aside from ZFS_ENTER()), use TXG_WAIT. This * reduces lock contention and CPU usage when we must wait (note that if * throughput is constrained by the storage, nearly every transaction * must wait). * * Note, in particular, that if a lock is sometimes acquired before * the tx assigns, and sometimes after (e.g. z_lock), then failing * to use a non-blocking assign can deadlock the system. The scenario: * * Thread A has grabbed a lock before calling dmu_tx_assign(). * Thread B is in an already-assigned tx, and blocks for this lock. * Thread A calls dmu_tx_assign(TXG_WAIT) and blocks in txg_wait_open() * forever, because the previous txg can't quiesce until B's tx commits. * * If dmu_tx_assign() returns ERESTART and zfsvfs->z_assign is TXG_NOWAIT, * then drop all locks, call dmu_tx_wait(), and try again. On subsequent * calls to dmu_tx_assign(), pass TXG_WAITED rather than TXG_NOWAIT, * to indicate that this operation has already called dmu_tx_wait(). * This will ensure that we don't retry forever, waiting a short bit * each time. * * (5) If the operation succeeded, generate the intent log entry for it * before dropping locks. This ensures that the ordering of events * in the intent log matches the order in which they actually occurred. * During ZIL replay the zfs_log_* functions will update the sequence * number to indicate the zil transaction has replayed. * * (6) At the end of each vnode op, the DMU tx must always commit, * regardless of whether there were any errors. * * (7) After dropping all locks, invoke zil_commit(zilog, foid) * to ensure that synchronous semantics are provided when necessary. * * In general, this is how things should be ordered in each vnode op: * * ZFS_ENTER(zfsvfs); // exit if unmounted * top: * zfs_dirent_lock(&dl, ...) // lock directory entry (may VN_HOLD()) * rw_enter(...); // grab any other locks you need * tx = dmu_tx_create(...); // get DMU tx * dmu_tx_hold_*(); // hold each object you might modify * error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); * if (error) { * rw_exit(...); // drop locks * zfs_dirent_unlock(dl); // unlock directory entry * VN_RELE(...); // release held vnodes * if (error == ERESTART) { * waited = B_TRUE; * dmu_tx_wait(tx); * dmu_tx_abort(tx); * goto top; * } * dmu_tx_abort(tx); // abort DMU tx * ZFS_EXIT(zfsvfs); // finished in zfs * return (error); // really out of space * } * error = do_real_work(); // do whatever this VOP does * if (error == 0) * zfs_log_*(...); // on success, make ZIL entry * dmu_tx_commit(tx); // commit DMU tx -- error or not * rw_exit(...); // drop locks * zfs_dirent_unlock(dl); // unlock directory entry * VN_RELE(...); // release held vnodes * zil_commit(zilog, foid); // synchronous when necessary * ZFS_EXIT(zfsvfs); // finished in zfs * return (error); // done, report error */ /* ARGSUSED */ static int zfs_open(vnode_t **vpp, int flag, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(*vpp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if ((flag & FWRITE) && (zp->z_pflags & ZFS_APPENDONLY) && ((flag & FAPPEND) == 0)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if (!zfs_has_ctldir(zp) && zp->z_zfsvfs->z_vscan && ZTOV(zp)->v_type == VREG && !(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0) { if (fs_vscan(*vpp, cr, 0) != 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EACCES)); } } /* Keep a count of the synchronous opens in the znode */ if (flag & (FSYNC | FDSYNC)) atomic_inc_32(&zp->z_sync_cnt); ZFS_EXIT(zfsvfs); return (0); } /* ARGSUSED */ static int zfs_close(vnode_t *vp, int flag, int count, offset_t offset, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; /* * Clean up any locks held by this process on the vp. */ cleanlocks(vp, ddi_get_pid(), 0); cleanshares(vp, ddi_get_pid()); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); /* Decrement the synchronous opens in the znode */ if ((flag & (FSYNC | FDSYNC)) && (count == 1)) atomic_dec_32(&zp->z_sync_cnt); if (!zfs_has_ctldir(zp) && zp->z_zfsvfs->z_vscan && ZTOV(zp)->v_type == VREG && !(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0) VERIFY(fs_vscan(vp, cr, 1) == 0); ZFS_EXIT(zfsvfs); return (0); } /* * Lseek support for finding holes (cmd == _FIO_SEEK_HOLE) and * data (cmd == _FIO_SEEK_DATA). "off" is an in/out parameter. */ static int zfs_holey(vnode_t *vp, u_long cmd, offset_t *off) { znode_t *zp = VTOZ(vp); uint64_t noff = (uint64_t)*off; /* new offset */ uint64_t file_sz; int error; boolean_t hole; file_sz = zp->z_size; if (noff >= file_sz) { return (SET_ERROR(ENXIO)); } if (cmd == _FIO_SEEK_HOLE) hole = B_TRUE; else hole = B_FALSE; error = dmu_offset_next(zp->z_zfsvfs->z_os, zp->z_id, hole, &noff); /* end of file? */ if ((error == ESRCH) || (noff > file_sz)) { /* * Handle the virtual hole at the end of file. */ if (hole) { *off = file_sz; return (0); } return (SET_ERROR(ENXIO)); } if (noff < *off) return (error); *off = noff; return (error); } /* ARGSUSED */ static int zfs_ioctl(vnode_t *vp, u_long com, intptr_t data, int flag, cred_t *cred, int *rvalp, caller_context_t *ct) { offset_t off; int error; zfsvfs_t *zfsvfs; znode_t *zp; switch (com) { case _FIOFFS: return (0); /* * The following two ioctls are used by bfu. Faking out, * necessary to avoid bfu errors. */ case _FIOGDIO: case _FIOSDIO: return (0); case _FIO_SEEK_DATA: case _FIO_SEEK_HOLE: #ifdef sun if (ddi_copyin((void *)data, &off, sizeof (off), flag)) return (SET_ERROR(EFAULT)); #else off = *(offset_t *)data; #endif zp = VTOZ(vp); zfsvfs = zp->z_zfsvfs; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); /* offset parameter is in/out */ error = zfs_holey(vp, com, &off); ZFS_EXIT(zfsvfs); if (error) return (error); #ifdef sun if (ddi_copyout(&off, (void *)data, sizeof (off), flag)) return (SET_ERROR(EFAULT)); #else *(offset_t *)data = off; #endif return (0); } return (SET_ERROR(ENOTTY)); } static vm_page_t page_busy(vnode_t *vp, int64_t start, int64_t off, int64_t nbytes) { vm_object_t obj; vm_page_t pp; int64_t end; /* * At present vm_page_clear_dirty extends the cleared range to DEV_BSIZE * aligned boundaries, if the range is not aligned. As a result a * DEV_BSIZE subrange with partially dirty data may get marked as clean. * It may happen that all DEV_BSIZE subranges are marked clean and thus * the whole page would be considred clean despite have some dirty data. * For this reason we should shrink the range to DEV_BSIZE aligned * boundaries before calling vm_page_clear_dirty. */ end = rounddown2(off + nbytes, DEV_BSIZE); off = roundup2(off, DEV_BSIZE); nbytes = end - off; obj = vp->v_object; VM_OBJECT_LOCK_ASSERT(obj, MA_OWNED); for (;;) { if ((pp = vm_page_lookup(obj, OFF_TO_IDX(start))) != NULL && pp->valid) { if ((pp->oflags & VPO_BUSY) != 0) { /* * Reference the page before unlocking and * sleeping so that the page daemon is less * likely to reclaim it. */ vm_page_reference(pp); vm_page_sleep(pp, "zfsmwb"); continue; } } else if (pp == NULL) { pp = vm_page_alloc(obj, OFF_TO_IDX(start), VM_ALLOC_SYSTEM | VM_ALLOC_IFCACHED | VM_ALLOC_NOBUSY); } else { ASSERT(pp != NULL && !pp->valid); pp = NULL; } if (pp != NULL) { ASSERT3U(pp->valid, ==, VM_PAGE_BITS_ALL); vm_object_pip_add(obj, 1); vm_page_io_start(pp); pmap_remove_write(pp); if (nbytes != 0) vm_page_clear_dirty(pp, off, nbytes); } break; } return (pp); } static void page_unbusy(vm_page_t pp) { vm_page_io_finish(pp); vm_object_pip_subtract(pp->object, 1); } static vm_page_t page_hold(vnode_t *vp, int64_t start) { vm_object_t obj; vm_page_t pp; obj = vp->v_object; VM_OBJECT_LOCK_ASSERT(obj, MA_OWNED); for (;;) { if ((pp = vm_page_lookup(obj, OFF_TO_IDX(start))) != NULL && pp->valid) { if ((pp->oflags & VPO_BUSY) != 0) { /* * Reference the page before unlocking and * sleeping so that the page daemon is less * likely to reclaim it. */ vm_page_reference(pp); vm_page_sleep(pp, "zfsmwb"); continue; } ASSERT3U(pp->valid, ==, VM_PAGE_BITS_ALL); vm_page_lock(pp); vm_page_hold(pp); vm_page_unlock(pp); } else pp = NULL; break; } return (pp); } static void page_unhold(vm_page_t pp) { vm_page_lock(pp); vm_page_unhold(pp); vm_page_unlock(pp); } /* * When a file is memory mapped, we must keep the IO data synchronized * between the DMU cache and the memory mapped pages. What this means: * * On Write: If we find a memory mapped page, we write to *both* * the page and the dmu buffer. */ static void update_pages(vnode_t *vp, int64_t start, int len, objset_t *os, uint64_t oid, int segflg, dmu_tx_t *tx) { vm_object_t obj; struct sf_buf *sf; caddr_t va; int off; ASSERT(segflg != UIO_NOCOPY); ASSERT(vp->v_mount != NULL); obj = vp->v_object; ASSERT(obj != NULL); off = start & PAGEOFFSET; VM_OBJECT_LOCK(obj); for (start &= PAGEMASK; len > 0; start += PAGESIZE) { vm_page_t pp; int nbytes = imin(PAGESIZE - off, len); if ((pp = page_busy(vp, start, off, nbytes)) != NULL) { VM_OBJECT_UNLOCK(obj); va = zfs_map_page(pp, &sf); (void) dmu_read(os, oid, start+off, nbytes, va+off, DMU_READ_PREFETCH);; zfs_unmap_page(sf); VM_OBJECT_LOCK(obj); page_unbusy(pp); } len -= nbytes; off = 0; } vm_object_pip_wakeupn(obj, 0); VM_OBJECT_UNLOCK(obj); } /* * Read with UIO_NOCOPY flag means that sendfile(2) requests * ZFS to populate a range of page cache pages with data. * * NOTE: this function could be optimized to pre-allocate * all pages in advance, drain VPO_BUSY on all of them, * map them into contiguous KVA region and populate them * in one single dmu_read() call. */ static int mappedread_sf(vnode_t *vp, int nbytes, uio_t *uio) { znode_t *zp = VTOZ(vp); objset_t *os = zp->z_zfsvfs->z_os; struct sf_buf *sf; vm_object_t obj; vm_page_t pp; int64_t start; caddr_t va; int len = nbytes; int off; int error = 0; ASSERT(uio->uio_segflg == UIO_NOCOPY); ASSERT(vp->v_mount != NULL); obj = vp->v_object; ASSERT(obj != NULL); ASSERT((uio->uio_loffset & PAGEOFFSET) == 0); VM_OBJECT_LOCK(obj); for (start = uio->uio_loffset; len > 0; start += PAGESIZE) { int bytes = MIN(PAGESIZE, len); pp = vm_page_grab(obj, OFF_TO_IDX(start), VM_ALLOC_NOBUSY | VM_ALLOC_NORMAL | VM_ALLOC_RETRY | VM_ALLOC_IGN_SBUSY); if (pp->valid == 0) { vm_page_io_start(pp); VM_OBJECT_UNLOCK(obj); va = zfs_map_page(pp, &sf); error = dmu_read(os, zp->z_id, start, bytes, va, DMU_READ_PREFETCH); if (bytes != PAGESIZE && error == 0) bzero(va + bytes, PAGESIZE - bytes); zfs_unmap_page(sf); VM_OBJECT_LOCK(obj); vm_page_io_finish(pp); vm_page_lock(pp); if (error) { if (pp->wire_count == 0 && pp->valid == 0 && pp->busy == 0 && !(pp->oflags & VPO_BUSY)) vm_page_free(pp); } else { pp->valid = VM_PAGE_BITS_ALL; vm_page_activate(pp); } vm_page_unlock(pp); } else { ASSERT3U(pp->valid, ==, VM_PAGE_BITS_ALL); } if (error) break; uio->uio_resid -= bytes; uio->uio_offset += bytes; len -= bytes; } VM_OBJECT_UNLOCK(obj); return (error); } /* * When a file is memory mapped, we must keep the IO data synchronized * between the DMU cache and the memory mapped pages. What this means: * * On Read: We "read" preferentially from memory mapped pages, * else we default from the dmu buffer. * * NOTE: We will always "break up" the IO into PAGESIZE uiomoves when * the file is memory mapped. */ static int mappedread(vnode_t *vp, int nbytes, uio_t *uio) { znode_t *zp = VTOZ(vp); objset_t *os = zp->z_zfsvfs->z_os; vm_object_t obj; int64_t start; caddr_t va; int len = nbytes; int off; int error = 0; ASSERT(vp->v_mount != NULL); obj = vp->v_object; ASSERT(obj != NULL); start = uio->uio_loffset; off = start & PAGEOFFSET; VM_OBJECT_LOCK(obj); for (start &= PAGEMASK; len > 0; start += PAGESIZE) { vm_page_t pp; uint64_t bytes = MIN(PAGESIZE - off, len); if (pp = page_hold(vp, start)) { struct sf_buf *sf; caddr_t va; VM_OBJECT_UNLOCK(obj); va = zfs_map_page(pp, &sf); error = uiomove(va + off, bytes, UIO_READ, uio); zfs_unmap_page(sf); VM_OBJECT_LOCK(obj); page_unhold(pp); } else { VM_OBJECT_UNLOCK(obj); error = dmu_read_uio(os, zp->z_id, uio, bytes); VM_OBJECT_LOCK(obj); } len -= bytes; off = 0; if (error) break; } VM_OBJECT_UNLOCK(obj); return (error); } offset_t zfs_read_chunk_size = 1024 * 1024; /* Tunable */ /* * Read bytes from specified file into supplied buffer. * * IN: vp - vnode of file to be read from. * uio - structure supplying read location, range info, * and return buffer. * ioflag - SYNC flags; used to provide FRSYNC semantics. * cr - credentials of caller. * ct - caller context * * OUT: uio - updated offset and range, buffer filled. * * RETURN: 0 on success, error code on failure. * * Side Effects: * vp - atime updated if byte count > 0 */ /* ARGSUSED */ static int zfs_read(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; objset_t *os; ssize_t n, nbytes; int error = 0; rl_t *rl; xuio_t *xuio = NULL; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); os = zfsvfs->z_os; if (zp->z_pflags & ZFS_AV_QUARANTINED) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EACCES)); } /* * Validate file offset */ if (uio->uio_loffset < (offset_t)0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * Fasttrack empty reads */ if (uio->uio_resid == 0) { ZFS_EXIT(zfsvfs); return (0); } /* * Check for mandatory locks */ if (MANDMODE(zp->z_mode)) { if (error = chklock(vp, FREAD, uio->uio_loffset, uio->uio_resid, uio->uio_fmode, ct)) { ZFS_EXIT(zfsvfs); return (error); } } /* * If we're in FRSYNC mode, sync out this znode before reading it. */ if (zfsvfs->z_log && (ioflag & FRSYNC || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)) zil_commit(zfsvfs->z_log, zp->z_id); /* * Lock the range against changes. */ rl = zfs_range_lock(zp, uio->uio_loffset, uio->uio_resid, RL_READER); /* * If we are reading past end-of-file we can skip * to the end; but we might still need to set atime. */ if (uio->uio_loffset >= zp->z_size) { error = 0; goto out; } ASSERT(uio->uio_loffset < zp->z_size); n = MIN(uio->uio_resid, zp->z_size - uio->uio_loffset); #ifdef sun if ((uio->uio_extflg == UIO_XUIO) && (((xuio_t *)uio)->xu_type == UIOTYPE_ZEROCOPY)) { int nblk; int blksz = zp->z_blksz; uint64_t offset = uio->uio_loffset; xuio = (xuio_t *)uio; if ((ISP2(blksz))) { nblk = (P2ROUNDUP(offset + n, blksz) - P2ALIGN(offset, blksz)) / blksz; } else { ASSERT(offset + n <= blksz); nblk = 1; } (void) dmu_xuio_init(xuio, nblk); if (vn_has_cached_data(vp)) { /* * For simplicity, we always allocate a full buffer * even if we only expect to read a portion of a block. */ while (--nblk >= 0) { (void) dmu_xuio_add(xuio, dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl), blksz), 0, blksz); } } } #endif /* sun */ while (n > 0) { nbytes = MIN(n, zfs_read_chunk_size - P2PHASE(uio->uio_loffset, zfs_read_chunk_size)); #ifdef __FreeBSD__ if (uio->uio_segflg == UIO_NOCOPY) error = mappedread_sf(vp, nbytes, uio); else #endif /* __FreeBSD__ */ if (vn_has_cached_data(vp)) error = mappedread(vp, nbytes, uio); else error = dmu_read_uio(os, zp->z_id, uio, nbytes); if (error) { /* convert checksum errors into IO errors */ if (error == ECKSUM) error = SET_ERROR(EIO); break; } n -= nbytes; } out: zfs_range_unlock(rl); ZFS_ACCESSTIME_STAMP(zfsvfs, zp); ZFS_EXIT(zfsvfs); return (error); } /* * Write the bytes to a file. * * IN: vp - vnode of file to be written to. * uio - structure supplying write location, range info, * and data buffer. * ioflag - FAPPEND, FSYNC, and/or FDSYNC. FAPPEND is * set if in append mode. * cr - credentials of caller. * ct - caller context (NFS/CIFS fem monitor only) * * OUT: uio - updated offset and range. * * RETURN: 0 on success, error code on failure. * * Timestamps: * vp - ctime|mtime updated if byte count > 0 */ /* ARGSUSED */ static int zfs_write(vnode_t *vp, uio_t *uio, int ioflag, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); rlim64_t limit = MAXOFFSET_T; ssize_t start_resid = uio->uio_resid; ssize_t tx_bytes; uint64_t end_size; dmu_tx_t *tx; zfsvfs_t *zfsvfs = zp->z_zfsvfs; zilog_t *zilog; offset_t woff; ssize_t n, nbytes; rl_t *rl; int max_blksz = zfsvfs->z_max_blksz; int error = 0; arc_buf_t *abuf; iovec_t *aiov = NULL; xuio_t *xuio = NULL; int i_iov = 0; int iovcnt = uio->uio_iovcnt; iovec_t *iovp = uio->uio_iov; int write_eof; int count = 0; sa_bulk_attr_t bulk[4]; uint64_t mtime[2], ctime[2]; /* * Fasttrack empty write */ n = start_resid; if (n == 0) return (0); if (limit == RLIM64_INFINITY || limit > MAXOFFSET_T) limit = MAXOFFSET_T; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL, &zp->z_size, 8); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL, &zp->z_pflags, 8); /* * If immutable or not appending then return EPERM */ if ((zp->z_pflags & (ZFS_IMMUTABLE | ZFS_READONLY)) || ((zp->z_pflags & ZFS_APPENDONLY) && !(ioflag & FAPPEND) && (uio->uio_loffset < zp->z_size))) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } zilog = zfsvfs->z_log; /* * Validate file offset */ woff = ioflag & FAPPEND ? zp->z_size : uio->uio_loffset; if (woff < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * Check for mandatory locks before calling zfs_range_lock() * in order to prevent a deadlock with locks set via fcntl(). */ if (MANDMODE((mode_t)zp->z_mode) && (error = chklock(vp, FWRITE, woff, n, uio->uio_fmode, ct)) != 0) { ZFS_EXIT(zfsvfs); return (error); } #ifdef sun /* * Pre-fault the pages to ensure slow (eg NFS) pages * don't hold up txg. * Skip this if uio contains loaned arc_buf. */ if ((uio->uio_extflg == UIO_XUIO) && (((xuio_t *)uio)->xu_type == UIOTYPE_ZEROCOPY)) xuio = (xuio_t *)uio; else uio_prefaultpages(MIN(n, max_blksz), uio); #endif /* sun */ /* * If in append mode, set the io offset pointer to eof. */ if (ioflag & FAPPEND) { /* * Obtain an appending range lock to guarantee file append * semantics. We reset the write offset once we have the lock. */ rl = zfs_range_lock(zp, 0, n, RL_APPEND); woff = rl->r_off; if (rl->r_len == UINT64_MAX) { /* * We overlocked the file because this write will cause * the file block size to increase. * Note that zp_size cannot change with this lock held. */ woff = zp->z_size; } uio->uio_loffset = woff; } else { /* * Note that if the file block size will change as a result of * this write, then this range lock will lock the entire file * so that we can re-write the block safely. */ rl = zfs_range_lock(zp, woff, n, RL_WRITER); } if (vn_rlimit_fsize(vp, uio, uio->uio_td)) { zfs_range_unlock(rl); ZFS_EXIT(zfsvfs); return (EFBIG); } if (woff >= limit) { zfs_range_unlock(rl); ZFS_EXIT(zfsvfs); return (SET_ERROR(EFBIG)); } if ((woff + n) > limit || woff > (limit - n)) n = limit - woff; /* Will this write extend the file length? */ write_eof = (woff + n > zp->z_size); end_size = MAX(zp->z_size, woff + n); /* * Write the file in reasonable size chunks. Each chunk is written * in a separate transaction; this keeps the intent log records small * and allows us to do more fine-grained space accounting. */ while (n > 0) { abuf = NULL; woff = uio->uio_loffset; if (zfs_owner_overquota(zfsvfs, zp, B_FALSE) || zfs_owner_overquota(zfsvfs, zp, B_TRUE)) { if (abuf != NULL) dmu_return_arcbuf(abuf); error = SET_ERROR(EDQUOT); break; } if (xuio && abuf == NULL) { ASSERT(i_iov < iovcnt); aiov = &iovp[i_iov]; abuf = dmu_xuio_arcbuf(xuio, i_iov); dmu_xuio_clear(xuio, i_iov); DTRACE_PROBE3(zfs_cp_write, int, i_iov, iovec_t *, aiov, arc_buf_t *, abuf); ASSERT((aiov->iov_base == abuf->b_data) || ((char *)aiov->iov_base - (char *)abuf->b_data + aiov->iov_len == arc_buf_size(abuf))); i_iov++; } else if (abuf == NULL && n >= max_blksz && woff >= zp->z_size && P2PHASE(woff, max_blksz) == 0 && zp->z_blksz == max_blksz) { /* * This write covers a full block. "Borrow" a buffer * from the dmu so that we can fill it before we enter * a transaction. This avoids the possibility of * holding up the transaction if the data copy hangs * up on a pagefault (e.g., from an NFS server mapping). */ size_t cbytes; abuf = dmu_request_arcbuf(sa_get_db(zp->z_sa_hdl), max_blksz); ASSERT(abuf != NULL); ASSERT(arc_buf_size(abuf) == max_blksz); if (error = uiocopy(abuf->b_data, max_blksz, UIO_WRITE, uio, &cbytes)) { dmu_return_arcbuf(abuf); break; } ASSERT(cbytes == max_blksz); } /* * Start a transaction. */ tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); dmu_tx_hold_write(tx, zp->z_id, woff, MIN(n, max_blksz)); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); if (abuf != NULL) dmu_return_arcbuf(abuf); break; } /* * If zfs_range_lock() over-locked we grow the blocksize * and then reduce the lock range. This will only happen * on the first iteration since zfs_range_reduce() will * shrink down r_len to the appropriate size. */ if (rl->r_len == UINT64_MAX) { uint64_t new_blksz; if (zp->z_blksz > max_blksz) { ASSERT(!ISP2(zp->z_blksz)); new_blksz = MIN(end_size, SPA_MAXBLOCKSIZE); } else { new_blksz = MIN(end_size, max_blksz); } zfs_grow_blocksize(zp, new_blksz, tx); zfs_range_reduce(rl, woff, n); } /* * XXX - should we really limit each write to z_max_blksz? * Perhaps we should use SPA_MAXBLOCKSIZE chunks? */ nbytes = MIN(n, max_blksz - P2PHASE(woff, max_blksz)); if (woff + nbytes > zp->z_size) vnode_pager_setsize(vp, woff + nbytes); if (abuf == NULL) { tx_bytes = uio->uio_resid; error = dmu_write_uio_dbuf(sa_get_db(zp->z_sa_hdl), uio, nbytes, tx); tx_bytes -= uio->uio_resid; } else { tx_bytes = nbytes; ASSERT(xuio == NULL || tx_bytes == aiov->iov_len); /* * If this is not a full block write, but we are * extending the file past EOF and this data starts * block-aligned, use assign_arcbuf(). Otherwise, * write via dmu_write(). */ if (tx_bytes < max_blksz && (!write_eof || aiov->iov_base != abuf->b_data)) { ASSERT(xuio); dmu_write(zfsvfs->z_os, zp->z_id, woff, aiov->iov_len, aiov->iov_base, tx); dmu_return_arcbuf(abuf); xuio_stat_wbuf_copied(); } else { ASSERT(xuio || tx_bytes == max_blksz); dmu_assign_arcbuf(sa_get_db(zp->z_sa_hdl), woff, abuf, tx); } ASSERT(tx_bytes <= uio->uio_resid); uioskip(uio, tx_bytes); } if (tx_bytes && vn_has_cached_data(vp)) { update_pages(vp, woff, tx_bytes, zfsvfs->z_os, zp->z_id, uio->uio_segflg, tx); } /* * If we made no progress, we're done. If we made even * partial progress, update the znode and ZIL accordingly. */ if (tx_bytes == 0) { (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs), (void *)&zp->z_size, sizeof (uint64_t), tx); dmu_tx_commit(tx); ASSERT(error != 0); break; } /* * Clear Set-UID/Set-GID bits on successful write if not * privileged and at least one of the excute bits is set. * * It would be nice to to this after all writes have * been done, but that would still expose the ISUID/ISGID * to another app after the partial write is committed. * * Note: we don't call zfs_fuid_map_id() here because * user 0 is not an ephemeral uid. */ mutex_enter(&zp->z_acl_lock); if ((zp->z_mode & (S_IXUSR | (S_IXUSR >> 3) | (S_IXUSR >> 6))) != 0 && (zp->z_mode & (S_ISUID | S_ISGID)) != 0 && secpolicy_vnode_setid_retain(vp, cr, (zp->z_mode & S_ISUID) != 0 && zp->z_uid == 0) != 0) { uint64_t newmode; zp->z_mode &= ~(S_ISUID | S_ISGID); newmode = zp->z_mode; (void) sa_update(zp->z_sa_hdl, SA_ZPL_MODE(zfsvfs), (void *)&newmode, sizeof (uint64_t), tx); } mutex_exit(&zp->z_acl_lock); zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime, B_TRUE); /* * Update the file size (zp_size) if it has changed; * account for possible concurrent updates. */ while ((end_size = zp->z_size) < uio->uio_loffset) { (void) atomic_cas_64(&zp->z_size, end_size, uio->uio_loffset); ASSERT(error == 0); } /* * If we are replaying and eof is non zero then force * the file size to the specified eof. Note, there's no * concurrency during replay. */ if (zfsvfs->z_replay && zfsvfs->z_replay_eof != 0) zp->z_size = zfsvfs->z_replay_eof; error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); zfs_log_write(zilog, tx, TX_WRITE, zp, woff, tx_bytes, ioflag); dmu_tx_commit(tx); if (error != 0) break; ASSERT(tx_bytes == nbytes); n -= nbytes; #ifdef sun if (!xuio && n > 0) uio_prefaultpages(MIN(n, max_blksz), uio); #endif /* sun */ } zfs_range_unlock(rl); /* * If we're in replay mode, or we made no progress, return error. * Otherwise, it's at least a partial write, so it's successful. */ if (zfsvfs->z_replay || uio->uio_resid == start_resid) { ZFS_EXIT(zfsvfs); return (error); } if (ioflag & (FSYNC | FDSYNC) || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, zp->z_id); ZFS_EXIT(zfsvfs); return (0); } void zfs_get_done(zgd_t *zgd, int error) { znode_t *zp = zgd->zgd_private; objset_t *os = zp->z_zfsvfs->z_os; int vfslocked; if (zgd->zgd_db) dmu_buf_rele(zgd->zgd_db, zgd); zfs_range_unlock(zgd->zgd_rl); vfslocked = VFS_LOCK_GIANT(zp->z_zfsvfs->z_vfs); /* * Release the vnode asynchronously as we currently have the * txg stopped from syncing. */ VN_RELE_ASYNC(ZTOV(zp), dsl_pool_vnrele_taskq(dmu_objset_pool(os))); if (error == 0 && zgd->zgd_bp) zil_add_block(zgd->zgd_zilog, zgd->zgd_bp); kmem_free(zgd, sizeof (zgd_t)); VFS_UNLOCK_GIANT(vfslocked); } #ifdef DEBUG static int zil_fault_io = 0; #endif /* * Get data to generate a TX_WRITE intent log record. */ int zfs_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio) { zfsvfs_t *zfsvfs = arg; objset_t *os = zfsvfs->z_os; znode_t *zp; uint64_t object = lr->lr_foid; uint64_t offset = lr->lr_offset; uint64_t size = lr->lr_length; blkptr_t *bp = &lr->lr_blkptr; dmu_buf_t *db; zgd_t *zgd; int error = 0; ASSERT(zio != NULL); ASSERT(size != 0); /* * Nothing to do if the file has been removed */ if (zfs_zget(zfsvfs, object, &zp) != 0) return (SET_ERROR(ENOENT)); if (zp->z_unlinked) { /* * Release the vnode asynchronously as we currently have the * txg stopped from syncing. */ VN_RELE_ASYNC(ZTOV(zp), dsl_pool_vnrele_taskq(dmu_objset_pool(os))); return (SET_ERROR(ENOENT)); } zgd = (zgd_t *)kmem_zalloc(sizeof (zgd_t), KM_SLEEP); zgd->zgd_zilog = zfsvfs->z_log; zgd->zgd_private = zp; /* * Write records come in two flavors: immediate and indirect. * For small writes it's cheaper to store the data with the * log record (immediate); for large writes it's cheaper to * sync the data and get a pointer to it (indirect) so that * we don't have to write the data twice. */ if (buf != NULL) { /* immediate write */ zgd->zgd_rl = zfs_range_lock(zp, offset, size, RL_READER); /* test for truncation needs to be done while range locked */ if (offset >= zp->z_size) { error = SET_ERROR(ENOENT); } else { error = dmu_read(os, object, offset, size, buf, DMU_READ_NO_PREFETCH); } ASSERT(error == 0 || error == ENOENT); } else { /* indirect write */ /* * Have to lock the whole block to ensure when it's * written out and it's checksum is being calculated * that no one can change the data. We need to re-check * blocksize after we get the lock in case it's changed! */ for (;;) { uint64_t blkoff; size = zp->z_blksz; blkoff = ISP2(size) ? P2PHASE(offset, size) : offset; offset -= blkoff; zgd->zgd_rl = zfs_range_lock(zp, offset, size, RL_READER); if (zp->z_blksz == size) break; offset += blkoff; zfs_range_unlock(zgd->zgd_rl); } /* test for truncation needs to be done while range locked */ if (lr->lr_offset >= zp->z_size) error = SET_ERROR(ENOENT); #ifdef DEBUG if (zil_fault_io) { error = SET_ERROR(EIO); zil_fault_io = 0; } #endif if (error == 0) error = dmu_buf_hold(os, object, offset, zgd, &db, DMU_READ_NO_PREFETCH); if (error == 0) { blkptr_t *obp = dmu_buf_get_blkptr(db); if (obp) { ASSERT(BP_IS_HOLE(bp)); *bp = *obp; } zgd->zgd_db = db; zgd->zgd_bp = bp; ASSERT(db->db_offset == offset); ASSERT(db->db_size == size); error = dmu_sync(zio, lr->lr_common.lrc_txg, zfs_get_done, zgd); ASSERT(error || lr->lr_length <= zp->z_blksz); /* * On success, we need to wait for the write I/O * initiated by dmu_sync() to complete before we can * release this dbuf. We will finish everything up * in the zfs_get_done() callback. */ if (error == 0) return (0); if (error == EALREADY) { lr->lr_common.lrc_txtype = TX_WRITE2; error = 0; } } } zfs_get_done(zgd, error); return (error); } /*ARGSUSED*/ static int zfs_access(vnode_t *vp, int mode, int flag, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; int error; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if (flag & V_ACE_MASK) error = zfs_zaccess(zp, mode, flag, B_FALSE, cr); else error = zfs_zaccess_rwx(zp, mode, flag, cr); ZFS_EXIT(zfsvfs); return (error); } /* * If vnode is for a device return a specfs vnode instead. */ static int specvp_check(vnode_t **vpp, cred_t *cr) { int error = 0; if (IS_DEVVP(*vpp)) { struct vnode *svp; svp = specvp(*vpp, (*vpp)->v_rdev, (*vpp)->v_type, cr); VN_RELE(*vpp); if (svp == NULL) error = SET_ERROR(ENOSYS); *vpp = svp; } return (error); } /* * Lookup an entry in a directory, or an extended attribute directory. * If it exists, return a held vnode reference for it. * * IN: dvp - vnode of directory to search. * nm - name of entry to lookup. * pnp - full pathname to lookup [UNUSED]. * flags - LOOKUP_XATTR set if looking for an attribute. * rdir - root directory vnode [UNUSED]. * cr - credentials of caller. * ct - caller context * direntflags - directory lookup flags * realpnp - returned pathname. * * OUT: vpp - vnode of located entry, NULL if not found. * * RETURN: 0 on success, error code on failure. * * Timestamps: * NA */ /* ARGSUSED */ static int zfs_lookup(vnode_t *dvp, char *nm, vnode_t **vpp, struct componentname *cnp, int nameiop, cred_t *cr, kthread_t *td, int flags) { znode_t *zdp = VTOZ(dvp); zfsvfs_t *zfsvfs = zdp->z_zfsvfs; int error = 0; int *direntflags = NULL; void *realpnp = NULL; /* fast path */ if (!(flags & (LOOKUP_XATTR | FIGNORECASE))) { if (dvp->v_type != VDIR) { return (SET_ERROR(ENOTDIR)); } else if (zdp->z_sa_hdl == NULL) { return (SET_ERROR(EIO)); } if (nm[0] == 0 || (nm[0] == '.' && nm[1] == '\0')) { error = zfs_fastaccesschk_execute(zdp, cr); if (!error) { *vpp = dvp; VN_HOLD(*vpp); return (0); } return (error); } else { vnode_t *tvp = dnlc_lookup(dvp, nm); if (tvp) { error = zfs_fastaccesschk_execute(zdp, cr); if (error) { VN_RELE(tvp); return (error); } if (tvp == DNLC_NO_VNODE) { VN_RELE(tvp); return (SET_ERROR(ENOENT)); } else { *vpp = tvp; return (specvp_check(vpp, cr)); } } } } DTRACE_PROBE2(zfs__fastpath__lookup__miss, vnode_t *, dvp, char *, nm); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zdp); *vpp = NULL; if (flags & LOOKUP_XATTR) { #ifdef TODO /* * If the xattr property is off, refuse the lookup request. */ if (!(zfsvfs->z_vfs->vfs_flag & VFS_XATTR)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } #endif /* * We don't allow recursive attributes.. * Maybe someday we will. */ if (zdp->z_pflags & ZFS_XATTR) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } if (error = zfs_get_xattrdir(VTOZ(dvp), vpp, cr, flags)) { ZFS_EXIT(zfsvfs); return (error); } /* * Do we have permission to get into attribute directory? */ if (error = zfs_zaccess(VTOZ(*vpp), ACE_EXECUTE, 0, B_FALSE, cr)) { VN_RELE(*vpp); *vpp = NULL; } ZFS_EXIT(zfsvfs); return (error); } if (dvp->v_type != VDIR) { ZFS_EXIT(zfsvfs); return (SET_ERROR(ENOTDIR)); } /* * Check accessibility of directory. */ if (error = zfs_zaccess(zdp, ACE_EXECUTE, 0, B_FALSE, cr)) { ZFS_EXIT(zfsvfs); return (error); } if (zfsvfs->z_utf8 && u8_validate(nm, strlen(nm), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } error = zfs_dirlook(zdp, nm, vpp, flags, direntflags, realpnp); if (error == 0) error = specvp_check(vpp, cr); /* Translate errors and add SAVENAME when needed. */ if (cnp->cn_flags & ISLASTCN) { switch (nameiop) { case CREATE: case RENAME: if (error == ENOENT) { error = EJUSTRETURN; cnp->cn_flags |= SAVENAME; break; } /* FALLTHROUGH */ case DELETE: if (error == 0) cnp->cn_flags |= SAVENAME; break; } } if (error == 0 && (nm[0] != '.' || nm[1] != '\0')) { int ltype = 0; if (cnp->cn_flags & ISDOTDOT) { ltype = VOP_ISLOCKED(dvp); VOP_UNLOCK(dvp, 0); } ZFS_EXIT(zfsvfs); error = vn_lock(*vpp, cnp->cn_lkflags); if (cnp->cn_flags & ISDOTDOT) vn_lock(dvp, ltype | LK_RETRY); if (error != 0) { VN_RELE(*vpp); *vpp = NULL; return (error); } } else { ZFS_EXIT(zfsvfs); } #ifdef FREEBSD_NAMECACHE /* * Insert name into cache (as non-existent) if appropriate. */ if (error == ENOENT && (cnp->cn_flags & MAKEENTRY) && nameiop != CREATE) cache_enter(dvp, *vpp, cnp); /* * Insert name into cache if appropriate. */ if (error == 0 && (cnp->cn_flags & MAKEENTRY)) { if (!(cnp->cn_flags & ISLASTCN) || (nameiop != DELETE && nameiop != RENAME)) { cache_enter(dvp, *vpp, cnp); } } #endif return (error); } /* * Attempt to create a new entry in a directory. If the entry * already exists, truncate the file if permissible, else return * an error. Return the vp of the created or trunc'd file. * * IN: dvp - vnode of directory to put new file entry in. * name - name of new file entry. * vap - attributes of new file. * excl - flag indicating exclusive or non-exclusive mode. * mode - mode to open file with. * cr - credentials of caller. * flag - large file flag [UNUSED]. * ct - caller context * vsecp - ACL to be set * * OUT: vpp - vnode of created or trunc'd entry. * * RETURN: 0 on success, error code on failure. * * Timestamps: * dvp - ctime|mtime updated if new entry created * vp - ctime|mtime always, atime if new */ /* ARGSUSED */ static int zfs_create(vnode_t *dvp, char *name, vattr_t *vap, int excl, int mode, vnode_t **vpp, cred_t *cr, kthread_t *td) { znode_t *zp, *dzp = VTOZ(dvp); zfsvfs_t *zfsvfs = dzp->z_zfsvfs; zilog_t *zilog; objset_t *os; zfs_dirlock_t *dl; dmu_tx_t *tx; int error; ksid_t *ksid; uid_t uid; gid_t gid = crgetgid(cr); zfs_acl_ids_t acl_ids; boolean_t fuid_dirtied; boolean_t have_acl = B_FALSE; boolean_t waited = B_FALSE; void *vsecp = NULL; int flag = 0; /* * If we have an ephemeral id, ACL, or XVATTR then * make sure file system is at proper version */ ksid = crgetsid(cr, KSID_OWNER); if (ksid) uid = ksid_getid(ksid); else uid = crgetuid(cr); if (zfsvfs->z_use_fuids == B_FALSE && (vsecp || (vap->va_mask & AT_XVATTR) || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid))) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); os = zfsvfs->z_os; zilog = zfsvfs->z_log; if (zfsvfs->z_utf8 && u8_validate(name, strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (vap->va_mask & AT_XVATTR) { if ((error = secpolicy_xvattr(dvp, (xvattr_t *)vap, crgetuid(cr), cr, vap->va_type)) != 0) { ZFS_EXIT(zfsvfs); return (error); } } getnewvnode_reserve(1); top: *vpp = NULL; if ((vap->va_mode & S_ISVTX) && secpolicy_vnode_stky_modify(cr)) vap->va_mode &= ~S_ISVTX; if (*name == '\0') { /* * Null component name refers to the directory itself. */ VN_HOLD(dvp); zp = dzp; dl = NULL; error = 0; } else { /* possible VN_HOLD(zp) */ int zflg = 0; if (flag & FIGNORECASE) zflg |= ZCILOOK; error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL); if (error) { if (have_acl) zfs_acl_ids_free(&acl_ids); if (strcmp(name, "..") == 0) error = SET_ERROR(EISDIR); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (error); } } if (zp == NULL) { uint64_t txtype; /* * Create a new file object and update the directory * to reference it. */ if (error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr)) { if (have_acl) zfs_acl_ids_free(&acl_ids); goto out; } /* * We only support the creation of regular files in * extended attribute directories. */ if ((dzp->z_pflags & ZFS_XATTR) && (vap->va_type != VREG)) { if (have_acl) zfs_acl_ids_free(&acl_ids); error = SET_ERROR(EINVAL); goto out; } if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap, cr, vsecp, &acl_ids)) != 0) goto out; have_acl = B_TRUE; if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) { zfs_acl_ids_free(&acl_ids); error = SET_ERROR(EDQUOT); goto out; } tx = dmu_tx_create(os); dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes + ZFS_SA_BASE_ATTR_SIZE); fuid_dirtied = zfsvfs->z_fuid_dirty; if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name); dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE); if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) { dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, acl_ids.z_aclp->z_acl_bytes); } error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } zfs_acl_ids_free(&acl_ids); dmu_tx_abort(tx); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (error); } zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids); if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); (void) zfs_link_create(dl, zp, tx, ZNEW); txtype = zfs_log_create_txtype(Z_FILE, vsecp, vap); if (flag & FIGNORECASE) txtype |= TX_CI; zfs_log_create(zilog, tx, txtype, dzp, zp, name, vsecp, acl_ids.z_fuidp, vap); zfs_acl_ids_free(&acl_ids); dmu_tx_commit(tx); } else { int aflags = (flag & FAPPEND) ? V_APPEND : 0; if (have_acl) zfs_acl_ids_free(&acl_ids); have_acl = B_FALSE; /* * A directory entry already exists for this name. */ /* * Can't truncate an existing file if in exclusive mode. */ if (excl == EXCL) { error = SET_ERROR(EEXIST); goto out; } /* * Can't open a directory for writing. */ if ((ZTOV(zp)->v_type == VDIR) && (mode & S_IWRITE)) { error = SET_ERROR(EISDIR); goto out; } /* * Verify requested access to file. */ if (mode && (error = zfs_zaccess_rwx(zp, mode, aflags, cr))) { goto out; } mutex_enter(&dzp->z_lock); dzp->z_seq++; mutex_exit(&dzp->z_lock); /* * Truncate regular files if requested. */ if ((ZTOV(zp)->v_type == VREG) && (vap->va_mask & AT_SIZE) && (vap->va_size == 0)) { /* we can't hold any locks when calling zfs_freesp() */ zfs_dirent_unlock(dl); dl = NULL; error = zfs_freesp(zp, 0, 0, mode, TRUE); if (error == 0) { vnevent_create(ZTOV(zp), ct); } } } out: getnewvnode_drop_reserve(); if (dl) zfs_dirent_unlock(dl); if (error) { if (zp) VN_RELE(ZTOV(zp)); } else { *vpp = ZTOV(zp); error = specvp_check(vpp, cr); } if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* * Remove an entry from a directory. * * IN: dvp - vnode of directory to remove entry from. * name - name of entry to remove. * cr - credentials of caller. * ct - caller context * flags - case flags * * RETURN: 0 on success, error code on failure. * * Timestamps: * dvp - ctime|mtime * vp - ctime (if nlink > 0) */ uint64_t null_xattr = 0; /*ARGSUSED*/ static int zfs_remove(vnode_t *dvp, char *name, cred_t *cr, caller_context_t *ct, int flags) { znode_t *zp, *dzp = VTOZ(dvp); znode_t *xzp; vnode_t *vp; zfsvfs_t *zfsvfs = dzp->z_zfsvfs; zilog_t *zilog; uint64_t acl_obj, xattr_obj; uint64_t xattr_obj_unlinked = 0; uint64_t obj = 0; zfs_dirlock_t *dl; dmu_tx_t *tx; boolean_t may_delete_now, delete_now = FALSE; boolean_t unlinked, toobig = FALSE; uint64_t txtype; pathname_t *realnmp = NULL; pathname_t realnm; int error; int zflg = ZEXISTS; boolean_t waited = B_FALSE; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; if (flags & FIGNORECASE) { zflg |= ZCILOOK; pn_alloc(&realnm); realnmp = &realnm; } top: xattr_obj = 0; xzp = NULL; /* * Attempt to lock directory; fail if entry doesn't exist. */ if (error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, realnmp)) { if (realnmp) pn_free(realnmp); ZFS_EXIT(zfsvfs); return (error); } vp = ZTOV(zp); if (error = zfs_zaccess_delete(dzp, zp, cr)) { goto out; } /* * Need to use rmdir for removing directories. */ if (vp->v_type == VDIR) { error = SET_ERROR(EPERM); goto out; } vnevent_remove(vp, dvp, name, ct); if (realnmp) dnlc_remove(dvp, realnmp->pn_buf); else dnlc_remove(dvp, name); VI_LOCK(vp); may_delete_now = vp->v_count == 1 && !vn_has_cached_data(vp); VI_UNLOCK(vp); /* * We may delete the znode now, or we may put it in the unlinked set; * it depends on whether we're the last link, and on whether there are * other holds on the vnode. So we dmu_tx_hold() the right things to * allow for either case. */ obj = zp->z_id; tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); zfs_sa_upgrade_txholds(tx, dzp); if (may_delete_now) { toobig = zp->z_size > zp->z_blksz * DMU_MAX_DELETEBLKCNT; /* if the file is too big, only hold_free a token amount */ dmu_tx_hold_free(tx, zp->z_id, 0, (toobig ? DMU_MAX_ACCESS : DMU_OBJECT_END)); } /* are there any extended attributes? */ error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &xattr_obj, sizeof (xattr_obj)); if (error == 0 && xattr_obj) { error = zfs_zget(zfsvfs, xattr_obj, &xzp); ASSERT0(error); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE); dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE); } mutex_enter(&zp->z_lock); if ((acl_obj = zfs_external_acl(zp)) != 0 && may_delete_now) dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END); mutex_exit(&zp->z_lock); /* charge as an update -- would be nice not to charge at all */ dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); VN_RELE(vp); if (xzp) VN_RELE(ZTOV(xzp)); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } if (realnmp) pn_free(realnmp); dmu_tx_abort(tx); ZFS_EXIT(zfsvfs); return (error); } /* * Remove the directory entry. */ error = zfs_link_destroy(dl, zp, tx, zflg, &unlinked); if (error) { dmu_tx_commit(tx); goto out; } if (unlinked) { /* * Hold z_lock so that we can make sure that the ACL obj * hasn't changed. Could have been deleted due to * zfs_sa_upgrade(). */ mutex_enter(&zp->z_lock); VI_LOCK(vp); (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &xattr_obj_unlinked, sizeof (xattr_obj_unlinked)); delete_now = may_delete_now && !toobig && vp->v_count == 1 && !vn_has_cached_data(vp) && xattr_obj == xattr_obj_unlinked && zfs_external_acl(zp) == acl_obj; VI_UNLOCK(vp); } if (delete_now) { #ifdef __FreeBSD__ panic("zfs_remove: delete_now branch taken"); #endif if (xattr_obj_unlinked) { ASSERT3U(xzp->z_links, ==, 2); mutex_enter(&xzp->z_lock); xzp->z_unlinked = 1; xzp->z_links = 0; error = sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zfsvfs), &xzp->z_links, sizeof (xzp->z_links), tx); ASSERT3U(error, ==, 0); mutex_exit(&xzp->z_lock); zfs_unlinked_add(xzp, tx); if (zp->z_is_sa) error = sa_remove(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), tx); else error = sa_update(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &null_xattr, sizeof (uint64_t), tx); ASSERT0(error); } VI_LOCK(vp); vp->v_count--; ASSERT0(vp->v_count); VI_UNLOCK(vp); mutex_exit(&zp->z_lock); zfs_znode_delete(zp, tx); } else if (unlinked) { mutex_exit(&zp->z_lock); zfs_unlinked_add(zp, tx); #ifdef __FreeBSD__ vp->v_vflag |= VV_NOSYNC; #endif } txtype = TX_REMOVE; if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_remove(zilog, tx, txtype, dzp, name, obj); dmu_tx_commit(tx); out: if (realnmp) pn_free(realnmp); zfs_dirent_unlock(dl); if (!delete_now) VN_RELE(vp); if (xzp) VN_RELE(ZTOV(xzp)); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* * Create a new directory and insert it into dvp using the name * provided. Return a pointer to the inserted directory. * * IN: dvp - vnode of directory to add subdir to. * dirname - name of new directory. * vap - attributes of new directory. * cr - credentials of caller. * ct - caller context * flags - case flags * vsecp - ACL to be set * * OUT: vpp - vnode of created directory. * * RETURN: 0 on success, error code on failure. * * Timestamps: * dvp - ctime|mtime updated * vp - ctime|mtime|atime updated */ /*ARGSUSED*/ static int zfs_mkdir(vnode_t *dvp, char *dirname, vattr_t *vap, vnode_t **vpp, cred_t *cr, caller_context_t *ct, int flags, vsecattr_t *vsecp) { znode_t *zp, *dzp = VTOZ(dvp); zfsvfs_t *zfsvfs = dzp->z_zfsvfs; zilog_t *zilog; zfs_dirlock_t *dl; uint64_t txtype; dmu_tx_t *tx; int error; int zf = ZNEW; ksid_t *ksid; uid_t uid; gid_t gid = crgetgid(cr); zfs_acl_ids_t acl_ids; boolean_t fuid_dirtied; boolean_t waited = B_FALSE; ASSERT(vap->va_type == VDIR); /* * If we have an ephemeral id, ACL, or XVATTR then * make sure file system is at proper version */ ksid = crgetsid(cr, KSID_OWNER); if (ksid) uid = ksid_getid(ksid); else uid = crgetuid(cr); if (zfsvfs->z_use_fuids == B_FALSE && (vsecp || (vap->va_mask & AT_XVATTR) || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid))) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; if (dzp->z_pflags & ZFS_XATTR) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } if (zfsvfs->z_utf8 && u8_validate(dirname, strlen(dirname), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (flags & FIGNORECASE) zf |= ZCILOOK; if (vap->va_mask & AT_XVATTR) { if ((error = secpolicy_xvattr(dvp, (xvattr_t *)vap, crgetuid(cr), cr, vap->va_type)) != 0) { ZFS_EXIT(zfsvfs); return (error); } } if ((error = zfs_acl_ids_create(dzp, 0, vap, cr, vsecp, &acl_ids)) != 0) { ZFS_EXIT(zfsvfs); return (error); } getnewvnode_reserve(1); /* * First make sure the new directory doesn't exist. * * Existence is checked first to make sure we don't return * EACCES instead of EEXIST which can cause some applications * to fail. */ top: *vpp = NULL; if (error = zfs_dirent_lock(&dl, dzp, dirname, &zp, zf, NULL, NULL)) { zfs_acl_ids_free(&acl_ids); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (error); } if (error = zfs_zaccess(dzp, ACE_ADD_SUBDIRECTORY, 0, B_FALSE, cr)) { zfs_acl_ids_free(&acl_ids); zfs_dirent_unlock(dl); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (error); } if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) { zfs_acl_ids_free(&acl_ids); zfs_dirent_unlock(dl); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (SET_ERROR(EDQUOT)); } /* * Add a new entry to the directory. */ tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_zap(tx, dzp->z_id, TRUE, dirname); dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL); fuid_dirtied = zfsvfs->z_fuid_dirty; if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) { dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, acl_ids.z_aclp->z_acl_bytes); } dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes + ZFS_SA_BASE_ATTR_SIZE); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } zfs_acl_ids_free(&acl_ids); dmu_tx_abort(tx); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (error); } /* * Create new node. */ zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids); if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); /* * Now put new name in parent dir. */ (void) zfs_link_create(dl, zp, tx, ZNEW); *vpp = ZTOV(zp); txtype = zfs_log_create_txtype(Z_DIR, vsecp, vap); if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_create(zilog, tx, txtype, dzp, zp, dirname, vsecp, acl_ids.z_fuidp, vap); zfs_acl_ids_free(&acl_ids); dmu_tx_commit(tx); getnewvnode_drop_reserve(); zfs_dirent_unlock(dl); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (0); } /* * Remove a directory subdir entry. If the current working * directory is the same as the subdir to be removed, the * remove will fail. * * IN: dvp - vnode of directory to remove from. * name - name of directory to be removed. * cwd - vnode of current working directory. * cr - credentials of caller. * ct - caller context * flags - case flags * * RETURN: 0 on success, error code on failure. * * Timestamps: * dvp - ctime|mtime updated */ /*ARGSUSED*/ static int zfs_rmdir(vnode_t *dvp, char *name, vnode_t *cwd, cred_t *cr, caller_context_t *ct, int flags) { znode_t *dzp = VTOZ(dvp); znode_t *zp; vnode_t *vp; zfsvfs_t *zfsvfs = dzp->z_zfsvfs; zilog_t *zilog; zfs_dirlock_t *dl; dmu_tx_t *tx; int error; int zflg = ZEXISTS; boolean_t waited = B_FALSE; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; if (flags & FIGNORECASE) zflg |= ZCILOOK; top: zp = NULL; /* * Attempt to lock directory; fail if entry doesn't exist. */ if (error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL)) { ZFS_EXIT(zfsvfs); return (error); } vp = ZTOV(zp); if (error = zfs_zaccess_delete(dzp, zp, cr)) { goto out; } if (vp->v_type != VDIR) { error = SET_ERROR(ENOTDIR); goto out; } if (vp == cwd) { error = SET_ERROR(EINVAL); goto out; } vnevent_rmdir(vp, dvp, name, ct); /* * Grab a lock on the directory to make sure that noone is * trying to add (or lookup) entries while we are removing it. */ rw_enter(&zp->z_name_lock, RW_WRITER); /* * Grab a lock on the parent pointer to make sure we play well * with the treewalk and directory rename code. */ rw_enter(&zp->z_parent_lock, RW_WRITER); tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL); zfs_sa_upgrade_txholds(tx, zp); zfs_sa_upgrade_txholds(tx, dzp); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { rw_exit(&zp->z_parent_lock); rw_exit(&zp->z_name_lock); zfs_dirent_unlock(dl); VN_RELE(vp); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } dmu_tx_abort(tx); ZFS_EXIT(zfsvfs); return (error); } #ifdef FREEBSD_NAMECACHE cache_purge(dvp); #endif error = zfs_link_destroy(dl, zp, tx, zflg, NULL); if (error == 0) { uint64_t txtype = TX_RMDIR; if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_remove(zilog, tx, txtype, dzp, name, ZFS_NO_OBJECT); } dmu_tx_commit(tx); rw_exit(&zp->z_parent_lock); rw_exit(&zp->z_name_lock); #ifdef FREEBSD_NAMECACHE cache_purge(vp); #endif out: zfs_dirent_unlock(dl); VN_RELE(vp); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* * Read as many directory entries as will fit into the provided * buffer from the given directory cursor position (specified in * the uio structure). * * IN: vp - vnode of directory to read. * uio - structure supplying read location, range info, * and return buffer. * cr - credentials of caller. * ct - caller context * flags - case flags * * OUT: uio - updated offset and range, buffer filled. * eofp - set to true if end-of-file detected. * * RETURN: 0 on success, error code on failure. * * Timestamps: * vp - atime updated * * Note that the low 4 bits of the cookie returned by zap is always zero. * This allows us to use the low range for "special" directory entries: * We use 0 for '.', and 1 for '..'. If this is the root of the filesystem, * we use the offset 2 for the '.zfs' directory. */ /* ARGSUSED */ static int zfs_readdir(vnode_t *vp, uio_t *uio, cred_t *cr, int *eofp, int *ncookies, u_long **cookies) { znode_t *zp = VTOZ(vp); iovec_t *iovp; edirent_t *eodp; dirent64_t *odp; zfsvfs_t *zfsvfs = zp->z_zfsvfs; objset_t *os; caddr_t outbuf; size_t bufsize; zap_cursor_t zc; zap_attribute_t zap; uint_t bytes_wanted; uint64_t offset; /* must be unsigned; checks for < 1 */ uint64_t parent; int local_eof; int outcount; int error; uint8_t prefetch; boolean_t check_sysattrs; uint8_t type; int ncooks; u_long *cooks = NULL; int flags = 0; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs), &parent, sizeof (parent))) != 0) { ZFS_EXIT(zfsvfs); return (error); } /* * If we are not given an eof variable, * use a local one. */ if (eofp == NULL) eofp = &local_eof; /* * Check for valid iov_len. */ if (uio->uio_iov->iov_len <= 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * Quit if directory has been removed (posix) */ if ((*eofp = zp->z_unlinked) != 0) { ZFS_EXIT(zfsvfs); return (0); } error = 0; os = zfsvfs->z_os; offset = uio->uio_loffset; prefetch = zp->z_zn_prefetch; /* * Initialize the iterator cursor. */ if (offset <= 3) { /* * Start iteration from the beginning of the directory. */ zap_cursor_init(&zc, os, zp->z_id); } else { /* * The offset is a serialized cursor. */ zap_cursor_init_serialized(&zc, os, zp->z_id, offset); } /* * Get space to change directory entries into fs independent format. */ iovp = uio->uio_iov; bytes_wanted = iovp->iov_len; if (uio->uio_segflg != UIO_SYSSPACE || uio->uio_iovcnt != 1) { bufsize = bytes_wanted; outbuf = kmem_alloc(bufsize, KM_SLEEP); odp = (struct dirent64 *)outbuf; } else { bufsize = bytes_wanted; outbuf = NULL; odp = (struct dirent64 *)iovp->iov_base; } eodp = (struct edirent *)odp; if (ncookies != NULL) { /* * Minimum entry size is dirent size and 1 byte for a file name. */ ncooks = uio->uio_resid / (sizeof(struct dirent) - sizeof(((struct dirent *)NULL)->d_name) + 1); cooks = malloc(ncooks * sizeof(u_long), M_TEMP, M_WAITOK); *cookies = cooks; *ncookies = ncooks; } /* * If this VFS supports the system attribute view interface; and * we're looking at an extended attribute directory; and we care * about normalization conflicts on this vfs; then we must check * for normalization conflicts with the sysattr name space. */ #ifdef TODO check_sysattrs = vfs_has_feature(vp->v_vfsp, VFSFT_SYSATTR_VIEWS) && (vp->v_flag & V_XATTRDIR) && zfsvfs->z_norm && (flags & V_RDDIR_ENTFLAGS); #else check_sysattrs = 0; #endif /* * Transform to file-system independent format */ outcount = 0; while (outcount < bytes_wanted) { ino64_t objnum; ushort_t reclen; off64_t *next = NULL; /* * Special case `.', `..', and `.zfs'. */ if (offset == 0) { (void) strcpy(zap.za_name, "."); zap.za_normalization_conflict = 0; objnum = zp->z_id; type = DT_DIR; } else if (offset == 1) { (void) strcpy(zap.za_name, ".."); zap.za_normalization_conflict = 0; objnum = parent; type = DT_DIR; } else if (offset == 2 && zfs_show_ctldir(zp)) { (void) strcpy(zap.za_name, ZFS_CTLDIR_NAME); zap.za_normalization_conflict = 0; objnum = ZFSCTL_INO_ROOT; type = DT_DIR; } else { /* * Grab next entry. */ if (error = zap_cursor_retrieve(&zc, &zap)) { if ((*eofp = (error == ENOENT)) != 0) break; else goto update; } if (zap.za_integer_length != 8 || zap.za_num_integers != 1) { cmn_err(CE_WARN, "zap_readdir: bad directory " "entry, obj = %lld, offset = %lld\n", (u_longlong_t)zp->z_id, (u_longlong_t)offset); error = SET_ERROR(ENXIO); goto update; } objnum = ZFS_DIRENT_OBJ(zap.za_first_integer); /* * MacOS X can extract the object type here such as: * uint8_t type = ZFS_DIRENT_TYPE(zap.za_first_integer); */ type = ZFS_DIRENT_TYPE(zap.za_first_integer); if (check_sysattrs && !zap.za_normalization_conflict) { #ifdef TODO zap.za_normalization_conflict = xattr_sysattr_casechk(zap.za_name); #else panic("%s:%u: TODO", __func__, __LINE__); #endif } } if (flags & V_RDDIR_ACCFILTER) { /* * If we have no access at all, don't include * this entry in the returned information */ znode_t *ezp; if (zfs_zget(zp->z_zfsvfs, objnum, &ezp) != 0) goto skip_entry; if (!zfs_has_access(ezp, cr)) { VN_RELE(ZTOV(ezp)); goto skip_entry; } VN_RELE(ZTOV(ezp)); } if (flags & V_RDDIR_ENTFLAGS) reclen = EDIRENT_RECLEN(strlen(zap.za_name)); else reclen = DIRENT64_RECLEN(strlen(zap.za_name)); /* * Will this entry fit in the buffer? */ if (outcount + reclen > bufsize) { /* * Did we manage to fit anything in the buffer? */ if (!outcount) { error = SET_ERROR(EINVAL); goto update; } break; } if (flags & V_RDDIR_ENTFLAGS) { /* * Add extended flag entry: */ eodp->ed_ino = objnum; eodp->ed_reclen = reclen; /* NOTE: ed_off is the offset for the *next* entry */ next = &(eodp->ed_off); eodp->ed_eflags = zap.za_normalization_conflict ? ED_CASE_CONFLICT : 0; (void) strncpy(eodp->ed_name, zap.za_name, EDIRENT_NAMELEN(reclen)); eodp = (edirent_t *)((intptr_t)eodp + reclen); } else { /* * Add normal entry: */ odp->d_ino = objnum; odp->d_reclen = reclen; odp->d_namlen = strlen(zap.za_name); (void) strlcpy(odp->d_name, zap.za_name, odp->d_namlen + 1); odp->d_type = type; odp = (dirent64_t *)((intptr_t)odp + reclen); } outcount += reclen; ASSERT(outcount <= bufsize); /* Prefetch znode */ if (prefetch) dmu_prefetch(os, objnum, 0, 0); skip_entry: /* * Move to the next entry, fill in the previous offset. */ if (offset > 2 || (offset == 2 && !zfs_show_ctldir(zp))) { zap_cursor_advance(&zc); offset = zap_cursor_serialize(&zc); } else { offset += 1; } if (cooks != NULL) { *cooks++ = offset; ncooks--; KASSERT(ncooks >= 0, ("ncookies=%d", ncooks)); } } zp->z_zn_prefetch = B_FALSE; /* a lookup will re-enable pre-fetching */ /* Subtract unused cookies */ if (ncookies != NULL) *ncookies -= ncooks; if (uio->uio_segflg == UIO_SYSSPACE && uio->uio_iovcnt == 1) { iovp->iov_base += outcount; iovp->iov_len -= outcount; uio->uio_resid -= outcount; } else if (error = uiomove(outbuf, (long)outcount, UIO_READ, uio)) { /* * Reset the pointer. */ offset = uio->uio_loffset; } update: zap_cursor_fini(&zc); if (uio->uio_segflg != UIO_SYSSPACE || uio->uio_iovcnt != 1) kmem_free(outbuf, bufsize); if (error == ENOENT) error = 0; ZFS_ACCESSTIME_STAMP(zfsvfs, zp); uio->uio_loffset = offset; ZFS_EXIT(zfsvfs); if (error != 0 && cookies != NULL) { free(*cookies, M_TEMP); *cookies = NULL; *ncookies = 0; } return (error); } ulong_t zfs_fsync_sync_cnt = 4; static int zfs_fsync(vnode_t *vp, int syncflag, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; (void) tsd_set(zfs_fsyncer_key, (void *)zfs_fsync_sync_cnt); if (zfsvfs->z_os->os_sync != ZFS_SYNC_DISABLED) { ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); zil_commit(zfsvfs->z_log, zp->z_id); ZFS_EXIT(zfsvfs); } return (0); } /* * Get the requested file attributes and place them in the provided * vattr structure. * * IN: vp - vnode of file. * vap - va_mask identifies requested attributes. * If AT_XVATTR set, then optional attrs are requested * flags - ATTR_NOACLCHECK (CIFS server context) * cr - credentials of caller. * ct - caller context * * OUT: vap - attribute values. * * RETURN: 0 (always succeeds). */ /* ARGSUSED */ static int zfs_getattr(vnode_t *vp, vattr_t *vap, int flags, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; int error = 0; uint32_t blksize; u_longlong_t nblocks; uint64_t links; uint64_t mtime[2], ctime[2], crtime[2], rdev; xvattr_t *xvap = (xvattr_t *)vap; /* vap may be an xvattr_t * */ xoptattr_t *xoap = NULL; boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE; sa_bulk_attr_t bulk[4]; int count = 0; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); zfs_fuid_map_ids(zp, cr, &vap->va_uid, &vap->va_gid); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CRTIME(zfsvfs), NULL, &crtime, 16); if (vp->v_type == VBLK || vp->v_type == VCHR) SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_RDEV(zfsvfs), NULL, &rdev, 8); if ((error = sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) != 0) { ZFS_EXIT(zfsvfs); return (error); } /* * If ACL is trivial don't bother looking for ACE_READ_ATTRIBUTES. * Also, if we are the owner don't bother, since owner should * always be allowed to read basic attributes of file. */ if (!(zp->z_pflags & ZFS_ACL_TRIVIAL) && (vap->va_uid != crgetuid(cr))) { if (error = zfs_zaccess(zp, ACE_READ_ATTRIBUTES, 0, skipaclchk, cr)) { ZFS_EXIT(zfsvfs); return (error); } } /* * Return all attributes. It's cheaper to provide the answer * than to determine whether we were asked the question. */ mutex_enter(&zp->z_lock); vap->va_type = IFTOVT(zp->z_mode); vap->va_mode = zp->z_mode & ~S_IFMT; #ifdef sun vap->va_fsid = zp->z_zfsvfs->z_vfs->vfs_dev; #else vap->va_fsid = vp->v_mount->mnt_stat.f_fsid.val[0]; #endif vap->va_nodeid = zp->z_id; if ((vp->v_flag & VROOT) && zfs_show_ctldir(zp)) links = zp->z_links + 1; else links = zp->z_links; vap->va_nlink = MIN(links, LINK_MAX); /* nlink_t limit! */ vap->va_size = zp->z_size; #ifdef sun vap->va_rdev = vp->v_rdev; #else if (vp->v_type == VBLK || vp->v_type == VCHR) vap->va_rdev = zfs_cmpldev(rdev); #endif vap->va_seq = zp->z_seq; vap->va_flags = 0; /* FreeBSD: Reset chflags(2) flags. */ vap->va_filerev = zp->z_seq; /* * Add in any requested optional attributes and the create time. * Also set the corresponding bits in the returned attribute bitmap. */ if ((xoap = xva_getxoptattr(xvap)) != NULL && zfsvfs->z_use_fuids) { if (XVA_ISSET_REQ(xvap, XAT_ARCHIVE)) { xoap->xoa_archive = ((zp->z_pflags & ZFS_ARCHIVE) != 0); XVA_SET_RTN(xvap, XAT_ARCHIVE); } if (XVA_ISSET_REQ(xvap, XAT_READONLY)) { xoap->xoa_readonly = ((zp->z_pflags & ZFS_READONLY) != 0); XVA_SET_RTN(xvap, XAT_READONLY); } if (XVA_ISSET_REQ(xvap, XAT_SYSTEM)) { xoap->xoa_system = ((zp->z_pflags & ZFS_SYSTEM) != 0); XVA_SET_RTN(xvap, XAT_SYSTEM); } if (XVA_ISSET_REQ(xvap, XAT_HIDDEN)) { xoap->xoa_hidden = ((zp->z_pflags & ZFS_HIDDEN) != 0); XVA_SET_RTN(xvap, XAT_HIDDEN); } if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) { xoap->xoa_nounlink = ((zp->z_pflags & ZFS_NOUNLINK) != 0); XVA_SET_RTN(xvap, XAT_NOUNLINK); } if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) { xoap->xoa_immutable = ((zp->z_pflags & ZFS_IMMUTABLE) != 0); XVA_SET_RTN(xvap, XAT_IMMUTABLE); } if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) { xoap->xoa_appendonly = ((zp->z_pflags & ZFS_APPENDONLY) != 0); XVA_SET_RTN(xvap, XAT_APPENDONLY); } if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) { xoap->xoa_nodump = ((zp->z_pflags & ZFS_NODUMP) != 0); XVA_SET_RTN(xvap, XAT_NODUMP); } if (XVA_ISSET_REQ(xvap, XAT_OPAQUE)) { xoap->xoa_opaque = ((zp->z_pflags & ZFS_OPAQUE) != 0); XVA_SET_RTN(xvap, XAT_OPAQUE); } if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) { xoap->xoa_av_quarantined = ((zp->z_pflags & ZFS_AV_QUARANTINED) != 0); XVA_SET_RTN(xvap, XAT_AV_QUARANTINED); } if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) { xoap->xoa_av_modified = ((zp->z_pflags & ZFS_AV_MODIFIED) != 0); XVA_SET_RTN(xvap, XAT_AV_MODIFIED); } if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) && vp->v_type == VREG) { zfs_sa_get_scanstamp(zp, xvap); } if (XVA_ISSET_REQ(xvap, XAT_CREATETIME)) { uint64_t times[2]; (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_CRTIME(zfsvfs), times, sizeof (times)); ZFS_TIME_DECODE(&xoap->xoa_createtime, times); XVA_SET_RTN(xvap, XAT_CREATETIME); } if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) { xoap->xoa_reparse = ((zp->z_pflags & ZFS_REPARSE) != 0); XVA_SET_RTN(xvap, XAT_REPARSE); } if (XVA_ISSET_REQ(xvap, XAT_GEN)) { xoap->xoa_generation = zp->z_gen; XVA_SET_RTN(xvap, XAT_GEN); } if (XVA_ISSET_REQ(xvap, XAT_OFFLINE)) { xoap->xoa_offline = ((zp->z_pflags & ZFS_OFFLINE) != 0); XVA_SET_RTN(xvap, XAT_OFFLINE); } if (XVA_ISSET_REQ(xvap, XAT_SPARSE)) { xoap->xoa_sparse = ((zp->z_pflags & ZFS_SPARSE) != 0); XVA_SET_RTN(xvap, XAT_SPARSE); } } ZFS_TIME_DECODE(&vap->va_atime, zp->z_atime); ZFS_TIME_DECODE(&vap->va_mtime, mtime); ZFS_TIME_DECODE(&vap->va_ctime, ctime); ZFS_TIME_DECODE(&vap->va_birthtime, crtime); mutex_exit(&zp->z_lock); sa_object_size(zp->z_sa_hdl, &blksize, &nblocks); vap->va_blksize = blksize; vap->va_bytes = nblocks << 9; /* nblocks * 512 */ if (zp->z_blksz == 0) { /* * Block size hasn't been set; suggest maximal I/O transfers. */ vap->va_blksize = zfsvfs->z_max_blksz; } ZFS_EXIT(zfsvfs); return (0); } /* * Set the file attributes to the values contained in the * vattr structure. * * IN: vp - vnode of file to be modified. * vap - new attribute values. * If AT_XVATTR set, then optional attrs are being set * flags - ATTR_UTIME set if non-default time values provided. * - ATTR_NOACLCHECK (CIFS context only). * cr - credentials of caller. * ct - caller context * * RETURN: 0 on success, error code on failure. * * Timestamps: * vp - ctime updated, mtime updated if size changed. */ /* ARGSUSED */ static int zfs_setattr(vnode_t *vp, vattr_t *vap, int flags, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; zilog_t *zilog; dmu_tx_t *tx; vattr_t oldva; xvattr_t tmpxvattr; uint_t mask = vap->va_mask; uint_t saved_mask = 0; uint64_t saved_mode; int trim_mask = 0; uint64_t new_mode; uint64_t new_uid, new_gid; uint64_t xattr_obj; uint64_t mtime[2], ctime[2]; znode_t *attrzp; int need_policy = FALSE; int err, err2; zfs_fuid_info_t *fuidp = NULL; xvattr_t *xvap = (xvattr_t *)vap; /* vap may be an xvattr_t * */ xoptattr_t *xoap; zfs_acl_t *aclp; boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE; boolean_t fuid_dirtied = B_FALSE; sa_bulk_attr_t bulk[7], xattr_bulk[7]; int count = 0, xattr_count = 0; if (mask == 0) return (0); if (mask & AT_NOSET) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); zilog = zfsvfs->z_log; /* * Make sure that if we have ephemeral uid/gid or xvattr specified * that file system is at proper version level */ if (zfsvfs->z_use_fuids == B_FALSE && (((mask & AT_UID) && IS_EPHEMERAL(vap->va_uid)) || ((mask & AT_GID) && IS_EPHEMERAL(vap->va_gid)) || (mask & AT_XVATTR))) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } if (mask & AT_SIZE && vp->v_type == VDIR) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EISDIR)); } if (mask & AT_SIZE && vp->v_type != VREG && vp->v_type != VFIFO) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * If this is an xvattr_t, then get a pointer to the structure of * optional attributes. If this is NULL, then we have a vattr_t. */ xoap = xva_getxoptattr(xvap); xva_init(&tmpxvattr); /* * Immutable files can only alter immutable bit and atime */ if ((zp->z_pflags & ZFS_IMMUTABLE) && ((mask & (AT_SIZE|AT_UID|AT_GID|AT_MTIME|AT_MODE)) || ((mask & AT_XVATTR) && XVA_ISSET_REQ(xvap, XAT_CREATETIME)))) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if ((mask & AT_SIZE) && (zp->z_pflags & ZFS_READONLY)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } /* * Verify timestamps doesn't overflow 32 bits. * ZFS can handle large timestamps, but 32bit syscalls can't * handle times greater than 2039. This check should be removed * once large timestamps are fully supported. */ if (mask & (AT_ATIME | AT_MTIME)) { if (((mask & AT_ATIME) && TIMESPEC_OVERFLOW(&vap->va_atime)) || ((mask & AT_MTIME) && TIMESPEC_OVERFLOW(&vap->va_mtime))) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EOVERFLOW)); } } top: attrzp = NULL; aclp = NULL; /* Can this be moved to before the top label? */ if (zfsvfs->z_vfs->vfs_flag & VFS_RDONLY) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EROFS)); } /* * First validate permissions */ if (mask & AT_SIZE) { /* * XXX - Note, we are not providing any open * mode flags here (like FNDELAY), so we may * block if there are locks present... this * should be addressed in openat(). */ /* XXX - would it be OK to generate a log record here? */ err = zfs_freesp(zp, vap->va_size, 0, 0, FALSE); if (err) { ZFS_EXIT(zfsvfs); return (err); } } if (mask & (AT_ATIME|AT_MTIME) || ((mask & AT_XVATTR) && (XVA_ISSET_REQ(xvap, XAT_HIDDEN) || XVA_ISSET_REQ(xvap, XAT_READONLY) || XVA_ISSET_REQ(xvap, XAT_ARCHIVE) || XVA_ISSET_REQ(xvap, XAT_OFFLINE) || XVA_ISSET_REQ(xvap, XAT_SPARSE) || XVA_ISSET_REQ(xvap, XAT_CREATETIME) || XVA_ISSET_REQ(xvap, XAT_SYSTEM)))) { need_policy = zfs_zaccess(zp, ACE_WRITE_ATTRIBUTES, 0, skipaclchk, cr); } if (mask & (AT_UID|AT_GID)) { int idmask = (mask & (AT_UID|AT_GID)); int take_owner; int take_group; /* * NOTE: even if a new mode is being set, * we may clear S_ISUID/S_ISGID bits. */ if (!(mask & AT_MODE)) vap->va_mode = zp->z_mode; /* * Take ownership or chgrp to group we are a member of */ take_owner = (mask & AT_UID) && (vap->va_uid == crgetuid(cr)); take_group = (mask & AT_GID) && zfs_groupmember(zfsvfs, vap->va_gid, cr); /* * If both AT_UID and AT_GID are set then take_owner and * take_group must both be set in order to allow taking * ownership. * * Otherwise, send the check through secpolicy_vnode_setattr() * */ if (((idmask == (AT_UID|AT_GID)) && take_owner && take_group) || ((idmask == AT_UID) && take_owner) || ((idmask == AT_GID) && take_group)) { if (zfs_zaccess(zp, ACE_WRITE_OWNER, 0, skipaclchk, cr) == 0) { /* * Remove setuid/setgid for non-privileged users */ secpolicy_setid_clear(vap, vp, cr); trim_mask = (mask & (AT_UID|AT_GID)); } else { need_policy = TRUE; } } else { need_policy = TRUE; } } mutex_enter(&zp->z_lock); oldva.va_mode = zp->z_mode; zfs_fuid_map_ids(zp, cr, &oldva.va_uid, &oldva.va_gid); if (mask & AT_XVATTR) { /* * Update xvattr mask to include only those attributes * that are actually changing. * * the bits will be restored prior to actually setting * the attributes so the caller thinks they were set. */ if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) { if (xoap->xoa_appendonly != ((zp->z_pflags & ZFS_APPENDONLY) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_APPENDONLY); XVA_SET_REQ(&tmpxvattr, XAT_APPENDONLY); } } if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) { if (xoap->xoa_nounlink != ((zp->z_pflags & ZFS_NOUNLINK) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_NOUNLINK); XVA_SET_REQ(&tmpxvattr, XAT_NOUNLINK); } } if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) { if (xoap->xoa_immutable != ((zp->z_pflags & ZFS_IMMUTABLE) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_IMMUTABLE); XVA_SET_REQ(&tmpxvattr, XAT_IMMUTABLE); } } if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) { if (xoap->xoa_nodump != ((zp->z_pflags & ZFS_NODUMP) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_NODUMP); XVA_SET_REQ(&tmpxvattr, XAT_NODUMP); } } if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) { if (xoap->xoa_av_modified != ((zp->z_pflags & ZFS_AV_MODIFIED) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_AV_MODIFIED); XVA_SET_REQ(&tmpxvattr, XAT_AV_MODIFIED); } } if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) { if ((vp->v_type != VREG && xoap->xoa_av_quarantined) || xoap->xoa_av_quarantined != ((zp->z_pflags & ZFS_AV_QUARANTINED) != 0)) { need_policy = TRUE; } else { XVA_CLR_REQ(xvap, XAT_AV_QUARANTINED); XVA_SET_REQ(&tmpxvattr, XAT_AV_QUARANTINED); } } if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) { mutex_exit(&zp->z_lock); ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if (need_policy == FALSE && (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) || XVA_ISSET_REQ(xvap, XAT_OPAQUE))) { need_policy = TRUE; } } mutex_exit(&zp->z_lock); if (mask & AT_MODE) { if (zfs_zaccess(zp, ACE_WRITE_ACL, 0, skipaclchk, cr) == 0) { err = secpolicy_setid_setsticky_clear(vp, vap, &oldva, cr); if (err) { ZFS_EXIT(zfsvfs); return (err); } trim_mask |= AT_MODE; } else { need_policy = TRUE; } } if (need_policy) { /* * If trim_mask is set then take ownership * has been granted or write_acl is present and user * has the ability to modify mode. In that case remove * UID|GID and or MODE from mask so that * secpolicy_vnode_setattr() doesn't revoke it. */ if (trim_mask) { saved_mask = vap->va_mask; vap->va_mask &= ~trim_mask; if (trim_mask & AT_MODE) { /* * Save the mode, as secpolicy_vnode_setattr() * will overwrite it with ova.va_mode. */ saved_mode = vap->va_mode; } } err = secpolicy_vnode_setattr(cr, vp, vap, &oldva, flags, (int (*)(void *, int, cred_t *))zfs_zaccess_unix, zp); if (err) { ZFS_EXIT(zfsvfs); return (err); } if (trim_mask) { vap->va_mask |= saved_mask; if (trim_mask & AT_MODE) { /* * Recover the mode after * secpolicy_vnode_setattr(). */ vap->va_mode = saved_mode; } } } /* * secpolicy_vnode_setattr, or take ownership may have * changed va_mask */ mask = vap->va_mask; if ((mask & (AT_UID | AT_GID))) { err = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &xattr_obj, sizeof (xattr_obj)); if (err == 0 && xattr_obj) { err = zfs_zget(zp->z_zfsvfs, xattr_obj, &attrzp); if (err) goto out2; } if (mask & AT_UID) { new_uid = zfs_fuid_create(zfsvfs, (uint64_t)vap->va_uid, cr, ZFS_OWNER, &fuidp); if (new_uid != zp->z_uid && zfs_fuid_overquota(zfsvfs, B_FALSE, new_uid)) { if (attrzp) VN_RELE(ZTOV(attrzp)); err = SET_ERROR(EDQUOT); goto out2; } } if (mask & AT_GID) { new_gid = zfs_fuid_create(zfsvfs, (uint64_t)vap->va_gid, cr, ZFS_GROUP, &fuidp); if (new_gid != zp->z_gid && zfs_fuid_overquota(zfsvfs, B_TRUE, new_gid)) { if (attrzp) VN_RELE(ZTOV(attrzp)); err = SET_ERROR(EDQUOT); goto out2; } } } tx = dmu_tx_create(zfsvfs->z_os); if (mask & AT_MODE) { uint64_t pmode = zp->z_mode; uint64_t acl_obj; new_mode = (pmode & S_IFMT) | (vap->va_mode & ~S_IFMT); if (zp->z_zfsvfs->z_acl_mode == ZFS_ACL_RESTRICTED && !(zp->z_pflags & ZFS_ACL_TRIVIAL)) { err = SET_ERROR(EPERM); goto out; } if (err = zfs_acl_chmod_setattr(zp, &aclp, new_mode)) goto out; mutex_enter(&zp->z_lock); if (!zp->z_is_sa && ((acl_obj = zfs_external_acl(zp)) != 0)) { /* * Are we upgrading ACL from old V0 format * to V1 format? */ if (zfsvfs->z_version >= ZPL_VERSION_FUID && zfs_znode_acl_version(zp) == ZFS_ACL_VERSION_INITIAL) { dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END); dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, aclp->z_acl_bytes); } else { dmu_tx_hold_write(tx, acl_obj, 0, aclp->z_acl_bytes); } } else if (!zp->z_is_sa && aclp->z_acl_bytes > ZFS_ACE_SPACE) { dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, aclp->z_acl_bytes); } mutex_exit(&zp->z_lock); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE); } else { if ((mask & AT_XVATTR) && XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE); else dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); } if (attrzp) { dmu_tx_hold_sa(tx, attrzp->z_sa_hdl, B_FALSE); } fuid_dirtied = zfsvfs->z_fuid_dirty; if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); zfs_sa_upgrade_txholds(tx, zp); err = dmu_tx_assign(tx, TXG_WAIT); if (err) goto out; count = 0; /* * Set each attribute requested. * We group settings according to the locks they need to acquire. * * Note: you cannot set ctime directly, although it will be * updated as a side-effect of calling this function. */ if (mask & (AT_UID|AT_GID|AT_MODE)) mutex_enter(&zp->z_acl_lock); mutex_enter(&zp->z_lock); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL, &zp->z_pflags, sizeof (zp->z_pflags)); if (attrzp) { if (mask & (AT_UID|AT_GID|AT_MODE)) mutex_enter(&attrzp->z_acl_lock); mutex_enter(&attrzp->z_lock); SA_ADD_BULK_ATTR(xattr_bulk, xattr_count, SA_ZPL_FLAGS(zfsvfs), NULL, &attrzp->z_pflags, sizeof (attrzp->z_pflags)); } if (mask & (AT_UID|AT_GID)) { if (mask & AT_UID) { SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL, &new_uid, sizeof (new_uid)); zp->z_uid = new_uid; if (attrzp) { SA_ADD_BULK_ATTR(xattr_bulk, xattr_count, SA_ZPL_UID(zfsvfs), NULL, &new_uid, sizeof (new_uid)); attrzp->z_uid = new_uid; } } if (mask & AT_GID) { SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL, &new_gid, sizeof (new_gid)); zp->z_gid = new_gid; if (attrzp) { SA_ADD_BULK_ATTR(xattr_bulk, xattr_count, SA_ZPL_GID(zfsvfs), NULL, &new_gid, sizeof (new_gid)); attrzp->z_gid = new_gid; } } if (!(mask & AT_MODE)) { SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL, &new_mode, sizeof (new_mode)); new_mode = zp->z_mode; } err = zfs_acl_chown_setattr(zp); ASSERT(err == 0); if (attrzp) { err = zfs_acl_chown_setattr(attrzp); ASSERT(err == 0); } } if (mask & AT_MODE) { SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL, &new_mode, sizeof (new_mode)); zp->z_mode = new_mode; ASSERT3U((uintptr_t)aclp, !=, 0); err = zfs_aclset_common(zp, aclp, cr, tx); ASSERT0(err); if (zp->z_acl_cached) zfs_acl_free(zp->z_acl_cached); zp->z_acl_cached = aclp; aclp = NULL; } if (mask & AT_ATIME) { ZFS_TIME_ENCODE(&vap->va_atime, zp->z_atime); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL, &zp->z_atime, sizeof (zp->z_atime)); } if (mask & AT_MTIME) { ZFS_TIME_ENCODE(&vap->va_mtime, mtime); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, mtime, sizeof (mtime)); } /* XXX - shouldn't this be done *before* the ATIME/MTIME checks? */ if (mask & AT_SIZE && !(mask & AT_MTIME)) { SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, mtime, sizeof (mtime)); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, sizeof (ctime)); zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime, B_TRUE); } else if (mask != 0) { SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, sizeof (ctime)); zfs_tstamp_update_setup(zp, STATE_CHANGED, mtime, ctime, B_TRUE); if (attrzp) { SA_ADD_BULK_ATTR(xattr_bulk, xattr_count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, sizeof (ctime)); zfs_tstamp_update_setup(attrzp, STATE_CHANGED, mtime, ctime, B_TRUE); } } /* * Do this after setting timestamps to prevent timestamp * update from toggling bit */ if (xoap && (mask & AT_XVATTR)) { /* * restore trimmed off masks * so that return masks can be set for caller. */ if (XVA_ISSET_REQ(&tmpxvattr, XAT_APPENDONLY)) { XVA_SET_REQ(xvap, XAT_APPENDONLY); } if (XVA_ISSET_REQ(&tmpxvattr, XAT_NOUNLINK)) { XVA_SET_REQ(xvap, XAT_NOUNLINK); } if (XVA_ISSET_REQ(&tmpxvattr, XAT_IMMUTABLE)) { XVA_SET_REQ(xvap, XAT_IMMUTABLE); } if (XVA_ISSET_REQ(&tmpxvattr, XAT_NODUMP)) { XVA_SET_REQ(xvap, XAT_NODUMP); } if (XVA_ISSET_REQ(&tmpxvattr, XAT_AV_MODIFIED)) { XVA_SET_REQ(xvap, XAT_AV_MODIFIED); } if (XVA_ISSET_REQ(&tmpxvattr, XAT_AV_QUARANTINED)) { XVA_SET_REQ(xvap, XAT_AV_QUARANTINED); } if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) ASSERT(vp->v_type == VREG); zfs_xvattr_set(zp, xvap, tx); } if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); if (mask != 0) zfs_log_setattr(zilog, tx, TX_SETATTR, zp, vap, mask, fuidp); mutex_exit(&zp->z_lock); if (mask & (AT_UID|AT_GID|AT_MODE)) mutex_exit(&zp->z_acl_lock); if (attrzp) { if (mask & (AT_UID|AT_GID|AT_MODE)) mutex_exit(&attrzp->z_acl_lock); mutex_exit(&attrzp->z_lock); } out: if (err == 0 && attrzp) { err2 = sa_bulk_update(attrzp->z_sa_hdl, xattr_bulk, xattr_count, tx); ASSERT(err2 == 0); } if (attrzp) VN_RELE(ZTOV(attrzp)); if (aclp) zfs_acl_free(aclp); if (fuidp) { zfs_fuid_info_free(fuidp); fuidp = NULL; } if (err) { dmu_tx_abort(tx); if (err == ERESTART) goto top; } else { err2 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); dmu_tx_commit(tx); } out2: if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (err); } typedef struct zfs_zlock { krwlock_t *zl_rwlock; /* lock we acquired */ znode_t *zl_znode; /* znode we held */ struct zfs_zlock *zl_next; /* next in list */ } zfs_zlock_t; /* * Drop locks and release vnodes that were held by zfs_rename_lock(). */ static void zfs_rename_unlock(zfs_zlock_t **zlpp) { zfs_zlock_t *zl; while ((zl = *zlpp) != NULL) { if (zl->zl_znode != NULL) VN_RELE(ZTOV(zl->zl_znode)); rw_exit(zl->zl_rwlock); *zlpp = zl->zl_next; kmem_free(zl, sizeof (*zl)); } } /* * Search back through the directory tree, using the ".." entries. * Lock each directory in the chain to prevent concurrent renames. * Fail any attempt to move a directory into one of its own descendants. * XXX - z_parent_lock can overlap with map or grow locks */ static int zfs_rename_lock(znode_t *szp, znode_t *tdzp, znode_t *sdzp, zfs_zlock_t **zlpp) { zfs_zlock_t *zl; znode_t *zp = tdzp; uint64_t rootid = zp->z_zfsvfs->z_root; uint64_t oidp = zp->z_id; krwlock_t *rwlp = &szp->z_parent_lock; krw_t rw = RW_WRITER; /* * First pass write-locks szp and compares to zp->z_id. * Later passes read-lock zp and compare to zp->z_parent. */ do { if (!rw_tryenter(rwlp, rw)) { /* * Another thread is renaming in this path. * Note that if we are a WRITER, we don't have any * parent_locks held yet. */ if (rw == RW_READER && zp->z_id > szp->z_id) { /* * Drop our locks and restart */ zfs_rename_unlock(&zl); *zlpp = NULL; zp = tdzp; oidp = zp->z_id; rwlp = &szp->z_parent_lock; rw = RW_WRITER; continue; } else { /* * Wait for other thread to drop its locks */ rw_enter(rwlp, rw); } } zl = kmem_alloc(sizeof (*zl), KM_SLEEP); zl->zl_rwlock = rwlp; zl->zl_znode = NULL; zl->zl_next = *zlpp; *zlpp = zl; if (oidp == szp->z_id) /* We're a descendant of szp */ return (SET_ERROR(EINVAL)); if (oidp == rootid) /* We've hit the top */ return (0); if (rw == RW_READER) { /* i.e. not the first pass */ int error = zfs_zget(zp->z_zfsvfs, oidp, &zp); if (error) return (error); zl->zl_znode = zp; } (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zp->z_zfsvfs), &oidp, sizeof (oidp)); rwlp = &zp->z_parent_lock; rw = RW_READER; } while (zp->z_id != sdzp->z_id); return (0); } /* * Move an entry from the provided source directory to the target * directory. Change the entry name as indicated. * * IN: sdvp - Source directory containing the "old entry". * snm - Old entry name. * tdvp - Target directory to contain the "new entry". * tnm - New entry name. * cr - credentials of caller. * ct - caller context * flags - case flags * * RETURN: 0 on success, error code on failure. * * Timestamps: * sdvp,tdvp - ctime|mtime updated */ /*ARGSUSED*/ static int zfs_rename(vnode_t *sdvp, char *snm, vnode_t *tdvp, char *tnm, cred_t *cr, caller_context_t *ct, int flags) { znode_t *tdzp, *szp, *tzp; znode_t *sdzp = VTOZ(sdvp); zfsvfs_t *zfsvfs = sdzp->z_zfsvfs; zilog_t *zilog; vnode_t *realvp; zfs_dirlock_t *sdl, *tdl; dmu_tx_t *tx; zfs_zlock_t *zl; int cmp, serr, terr; int error = 0; int zflg = 0; boolean_t waited = B_FALSE; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(sdzp); zilog = zfsvfs->z_log; /* * Make sure we have the real vp for the target directory. */ if (VOP_REALVP(tdvp, &realvp, ct) == 0) tdvp = realvp; if (tdvp->v_vfsp != sdvp->v_vfsp || zfsctl_is_node(tdvp)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EXDEV)); } tdzp = VTOZ(tdvp); ZFS_VERIFY_ZP(tdzp); if (zfsvfs->z_utf8 && u8_validate(tnm, strlen(tnm), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (flags & FIGNORECASE) zflg |= ZCILOOK; top: szp = NULL; tzp = NULL; zl = NULL; /* * This is to prevent the creation of links into attribute space * by renaming a linked file into/outof an attribute directory. * See the comment in zfs_link() for why this is considered bad. */ if ((tdzp->z_pflags & ZFS_XATTR) != (sdzp->z_pflags & ZFS_XATTR)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* * Lock source and target directory entries. To prevent deadlock, * a lock ordering must be defined. We lock the directory with * the smallest object id first, or if it's a tie, the one with * the lexically first name. */ if (sdzp->z_id < tdzp->z_id) { cmp = -1; } else if (sdzp->z_id > tdzp->z_id) { cmp = 1; } else { /* * First compare the two name arguments without * considering any case folding. */ int nofold = (zfsvfs->z_norm & ~U8_TEXTPREP_TOUPPER); cmp = u8_strcmp(snm, tnm, 0, nofold, U8_UNICODE_LATEST, &error); ASSERT(error == 0 || !zfsvfs->z_utf8); if (cmp == 0) { /* * POSIX: "If the old argument and the new argument * both refer to links to the same existing file, * the rename() function shall return successfully * and perform no other action." */ ZFS_EXIT(zfsvfs); return (0); } /* * If the file system is case-folding, then we may * have some more checking to do. A case-folding file * system is either supporting mixed case sensitivity * access or is completely case-insensitive. Note * that the file system is always case preserving. * * In mixed sensitivity mode case sensitive behavior * is the default. FIGNORECASE must be used to * explicitly request case insensitive behavior. * * If the source and target names provided differ only * by case (e.g., a request to rename 'tim' to 'Tim'), * we will treat this as a special case in the * case-insensitive mode: as long as the source name * is an exact match, we will allow this to proceed as * a name-change request. */ if ((zfsvfs->z_case == ZFS_CASE_INSENSITIVE || (zfsvfs->z_case == ZFS_CASE_MIXED && flags & FIGNORECASE)) && u8_strcmp(snm, tnm, 0, zfsvfs->z_norm, U8_UNICODE_LATEST, &error) == 0) { /* * case preserving rename request, require exact * name matches */ zflg |= ZCIEXACT; zflg &= ~ZCILOOK; } } /* * If the source and destination directories are the same, we should * grab the z_name_lock of that directory only once. */ if (sdzp == tdzp) { zflg |= ZHAVELOCK; rw_enter(&sdzp->z_name_lock, RW_READER); } if (cmp < 0) { serr = zfs_dirent_lock(&sdl, sdzp, snm, &szp, ZEXISTS | zflg, NULL, NULL); terr = zfs_dirent_lock(&tdl, tdzp, tnm, &tzp, ZRENAMING | zflg, NULL, NULL); } else { terr = zfs_dirent_lock(&tdl, tdzp, tnm, &tzp, zflg, NULL, NULL); serr = zfs_dirent_lock(&sdl, sdzp, snm, &szp, ZEXISTS | ZRENAMING | zflg, NULL, NULL); } if (serr) { /* * Source entry invalid or not there. */ if (!terr) { zfs_dirent_unlock(tdl); if (tzp) VN_RELE(ZTOV(tzp)); } if (sdzp == tdzp) rw_exit(&sdzp->z_name_lock); /* * FreeBSD: In OpenSolaris they only check if rename source is * ".." here, because "." is handled in their lookup. This is * not the case for FreeBSD, so we check for "." explicitly. */ if (strcmp(snm, ".") == 0 || strcmp(snm, "..") == 0) serr = SET_ERROR(EINVAL); ZFS_EXIT(zfsvfs); return (serr); } if (terr) { zfs_dirent_unlock(sdl); VN_RELE(ZTOV(szp)); if (sdzp == tdzp) rw_exit(&sdzp->z_name_lock); if (strcmp(tnm, "..") == 0) terr = SET_ERROR(EINVAL); ZFS_EXIT(zfsvfs); return (terr); } /* * Must have write access at the source to remove the old entry * and write access at the target to create the new entry. * Note that if target and source are the same, this can be * done in a single check. */ if (error = zfs_zaccess_rename(sdzp, szp, tdzp, tzp, cr)) goto out; if (ZTOV(szp)->v_type == VDIR) { /* * Check to make sure rename is valid. * Can't do a move like this: /usr/a/b to /usr/a/b/c/d */ if (error = zfs_rename_lock(szp, tdzp, sdzp, &zl)) goto out; } /* * Does target exist? */ if (tzp) { /* * Source and target must be the same type. */ if (ZTOV(szp)->v_type == VDIR) { if (ZTOV(tzp)->v_type != VDIR) { error = SET_ERROR(ENOTDIR); goto out; } } else { if (ZTOV(tzp)->v_type == VDIR) { error = SET_ERROR(EISDIR); goto out; } } /* * POSIX dictates that when the source and target * entries refer to the same file object, rename * must do nothing and exit without error. */ if (szp->z_id == tzp->z_id) { error = 0; goto out; } } vnevent_rename_src(ZTOV(szp), sdvp, snm, ct); if (tzp) vnevent_rename_dest(ZTOV(tzp), tdvp, tnm, ct); /* * notify the target directory if it is not the same * as source directory. */ if (tdvp != sdvp) { vnevent_rename_dest_dir(tdvp, ct); } tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE); dmu_tx_hold_sa(tx, sdzp->z_sa_hdl, B_FALSE); dmu_tx_hold_zap(tx, sdzp->z_id, FALSE, snm); dmu_tx_hold_zap(tx, tdzp->z_id, TRUE, tnm); if (sdzp != tdzp) { dmu_tx_hold_sa(tx, tdzp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, tdzp); } if (tzp) { dmu_tx_hold_sa(tx, tzp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, tzp); } zfs_sa_upgrade_txholds(tx, szp); dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { if (zl != NULL) zfs_rename_unlock(&zl); zfs_dirent_unlock(sdl); zfs_dirent_unlock(tdl); if (sdzp == tdzp) rw_exit(&sdzp->z_name_lock); VN_RELE(ZTOV(szp)); if (tzp) VN_RELE(ZTOV(tzp)); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } dmu_tx_abort(tx); ZFS_EXIT(zfsvfs); return (error); } if (tzp) /* Attempt to remove the existing target */ error = zfs_link_destroy(tdl, tzp, tx, zflg, NULL); if (error == 0) { error = zfs_link_create(tdl, szp, tx, ZRENAMING); if (error == 0) { szp->z_pflags |= ZFS_AV_MODIFIED; error = sa_update(szp->z_sa_hdl, SA_ZPL_FLAGS(zfsvfs), (void *)&szp->z_pflags, sizeof (uint64_t), tx); ASSERT0(error); error = zfs_link_destroy(sdl, szp, tx, ZRENAMING, NULL); if (error == 0) { zfs_log_rename(zilog, tx, TX_RENAME | (flags & FIGNORECASE ? TX_CI : 0), sdzp, sdl->dl_name, tdzp, tdl->dl_name, szp); /* * Update path information for the target vnode */ vn_renamepath(tdvp, ZTOV(szp), tnm, strlen(tnm)); } else { /* * At this point, we have successfully created * the target name, but have failed to remove * the source name. Since the create was done * with the ZRENAMING flag, there are * complications; for one, the link count is * wrong. The easiest way to deal with this * is to remove the newly created target, and * return the original error. This must * succeed; fortunately, it is very unlikely to * fail, since we just created it. */ VERIFY3U(zfs_link_destroy(tdl, szp, tx, ZRENAMING, NULL), ==, 0); } } #ifdef FREEBSD_NAMECACHE if (error == 0) { cache_purge(sdvp); cache_purge(tdvp); cache_purge(ZTOV(szp)); if (tzp) cache_purge(ZTOV(tzp)); } #endif } dmu_tx_commit(tx); out: if (zl != NULL) zfs_rename_unlock(&zl); zfs_dirent_unlock(sdl); zfs_dirent_unlock(tdl); if (sdzp == tdzp) rw_exit(&sdzp->z_name_lock); VN_RELE(ZTOV(szp)); if (tzp) VN_RELE(ZTOV(tzp)); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* * Insert the indicated symbolic reference entry into the directory. * * IN: dvp - Directory to contain new symbolic link. * link - Name for new symlink entry. * vap - Attributes of new entry. * cr - credentials of caller. * ct - caller context * flags - case flags * * RETURN: 0 on success, error code on failure. * * Timestamps: * dvp - ctime|mtime updated */ /*ARGSUSED*/ static int zfs_symlink(vnode_t *dvp, vnode_t **vpp, char *name, vattr_t *vap, char *link, cred_t *cr, kthread_t *td) { znode_t *zp, *dzp = VTOZ(dvp); zfs_dirlock_t *dl; dmu_tx_t *tx; zfsvfs_t *zfsvfs = dzp->z_zfsvfs; zilog_t *zilog; uint64_t len = strlen(link); int error; int zflg = ZNEW; zfs_acl_ids_t acl_ids; boolean_t fuid_dirtied; uint64_t txtype = TX_SYMLINK; boolean_t waited = B_FALSE; int flags = 0; ASSERT(vap->va_type == VLNK); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; if (zfsvfs->z_utf8 && u8_validate(name, strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (flags & FIGNORECASE) zflg |= ZCILOOK; if (len > MAXPATHLEN) { ZFS_EXIT(zfsvfs); return (SET_ERROR(ENAMETOOLONG)); } if ((error = zfs_acl_ids_create(dzp, 0, vap, cr, NULL, &acl_ids)) != 0) { ZFS_EXIT(zfsvfs); return (error); } getnewvnode_reserve(1); top: /* * Attempt to lock directory; fail if entry already exists. */ error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL); if (error) { zfs_acl_ids_free(&acl_ids); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (error); } if (error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr)) { zfs_acl_ids_free(&acl_ids); zfs_dirent_unlock(dl); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (error); } if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) { zfs_acl_ids_free(&acl_ids); zfs_dirent_unlock(dl); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (SET_ERROR(EDQUOT)); } tx = dmu_tx_create(zfsvfs->z_os); fuid_dirtied = zfsvfs->z_fuid_dirty; dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, MAX(1, len)); dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name); dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes + ZFS_SA_BASE_ATTR_SIZE + len); dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE); if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) { dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, acl_ids.z_aclp->z_acl_bytes); } if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } zfs_acl_ids_free(&acl_ids); dmu_tx_abort(tx); getnewvnode_drop_reserve(); ZFS_EXIT(zfsvfs); return (error); } /* * Create a new object for the symlink. * for version 4 ZPL datsets the symlink will be an SA attribute */ zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids); if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); mutex_enter(&zp->z_lock); if (zp->z_is_sa) error = sa_update(zp->z_sa_hdl, SA_ZPL_SYMLINK(zfsvfs), link, len, tx); else zfs_sa_symlink(zp, link, len, tx); mutex_exit(&zp->z_lock); zp->z_size = len; (void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs), &zp->z_size, sizeof (zp->z_size), tx); /* * Insert the new object into the directory. */ (void) zfs_link_create(dl, zp, tx, ZNEW); if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_symlink(zilog, tx, txtype, dzp, zp, name, link); *vpp = ZTOV(zp); zfs_acl_ids_free(&acl_ids); dmu_tx_commit(tx); getnewvnode_drop_reserve(); zfs_dirent_unlock(dl); if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } /* * Return, in the buffer contained in the provided uio structure, * the symbolic path referred to by vp. * * IN: vp - vnode of symbolic link. * uio - structure to contain the link path. * cr - credentials of caller. * ct - caller context * * OUT: uio - structure containing the link path. * * RETURN: 0 on success, error code on failure. * * Timestamps: * vp - atime updated */ /* ARGSUSED */ static int zfs_readlink(vnode_t *vp, uio_t *uio, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; int error; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); mutex_enter(&zp->z_lock); if (zp->z_is_sa) error = sa_lookup_uio(zp->z_sa_hdl, SA_ZPL_SYMLINK(zfsvfs), uio); else error = zfs_sa_readlink(zp, uio); mutex_exit(&zp->z_lock); ZFS_ACCESSTIME_STAMP(zfsvfs, zp); ZFS_EXIT(zfsvfs); return (error); } /* * Insert a new entry into directory tdvp referencing svp. * * IN: tdvp - Directory to contain new entry. * svp - vnode of new entry. * name - name of new entry. * cr - credentials of caller. * ct - caller context * * RETURN: 0 on success, error code on failure. * * Timestamps: * tdvp - ctime|mtime updated * svp - ctime updated */ /* ARGSUSED */ static int zfs_link(vnode_t *tdvp, vnode_t *svp, char *name, cred_t *cr, caller_context_t *ct, int flags) { znode_t *dzp = VTOZ(tdvp); znode_t *tzp, *szp; zfsvfs_t *zfsvfs = dzp->z_zfsvfs; zilog_t *zilog; zfs_dirlock_t *dl; dmu_tx_t *tx; vnode_t *realvp; int error; int zf = ZNEW; uint64_t parent; uid_t owner; boolean_t waited = B_FALSE; ASSERT(tdvp->v_type == VDIR); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(dzp); zilog = zfsvfs->z_log; if (VOP_REALVP(svp, &realvp, ct) == 0) svp = realvp; /* * POSIX dictates that we return EPERM here. * Better choices include ENOTSUP or EISDIR. */ if (svp->v_type == VDIR) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if (svp->v_vfsp != tdvp->v_vfsp || zfsctl_is_node(svp)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EXDEV)); } szp = VTOZ(svp); ZFS_VERIFY_ZP(szp); /* Prevent links to .zfs/shares files */ if ((error = sa_lookup(szp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs), &parent, sizeof (uint64_t))) != 0) { ZFS_EXIT(zfsvfs); return (error); } if (parent == zfsvfs->z_shares_dir) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if (zfsvfs->z_utf8 && u8_validate(name, strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EILSEQ)); } if (flags & FIGNORECASE) zf |= ZCILOOK; /* * We do not support links between attributes and non-attributes * because of the potential security risk of creating links * into "normal" file space in order to circumvent restrictions * imposed in attribute space. */ if ((szp->z_pflags & ZFS_XATTR) != (dzp->z_pflags & ZFS_XATTR)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } owner = zfs_fuid_map_id(zfsvfs, szp->z_uid, cr, ZFS_OWNER); if (owner != crgetuid(cr) && secpolicy_basic_link(svp, cr) != 0) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if (error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr)) { ZFS_EXIT(zfsvfs); return (error); } top: /* * Attempt to lock directory; fail if entry already exists. */ error = zfs_dirent_lock(&dl, dzp, name, &tzp, zf, NULL, NULL); if (error) { ZFS_EXIT(zfsvfs); return (error); } tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE); dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name); zfs_sa_upgrade_txholds(tx, szp); zfs_sa_upgrade_txholds(tx, dzp); error = dmu_tx_assign(tx, waited ? TXG_WAITED : TXG_NOWAIT); if (error) { zfs_dirent_unlock(dl); if (error == ERESTART) { waited = B_TRUE; dmu_tx_wait(tx); dmu_tx_abort(tx); goto top; } dmu_tx_abort(tx); ZFS_EXIT(zfsvfs); return (error); } error = zfs_link_create(dl, szp, tx, 0); if (error == 0) { uint64_t txtype = TX_LINK; if (flags & FIGNORECASE) txtype |= TX_CI; zfs_log_link(zilog, tx, txtype, dzp, szp, name); } dmu_tx_commit(tx); zfs_dirent_unlock(dl); if (error == 0) { vnevent_link(svp, ct); } if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zilog, 0); ZFS_EXIT(zfsvfs); return (error); } #ifdef sun /* * zfs_null_putapage() is used when the file system has been force * unmounted. It just drops the pages. */ /* ARGSUSED */ static int zfs_null_putapage(vnode_t *vp, page_t *pp, u_offset_t *offp, size_t *lenp, int flags, cred_t *cr) { pvn_write_done(pp, B_INVAL|B_FORCE|B_ERROR); return (0); } /* * Push a page out to disk, klustering if possible. * * IN: vp - file to push page to. * pp - page to push. * flags - additional flags. * cr - credentials of caller. * * OUT: offp - start of range pushed. * lenp - len of range pushed. * * RETURN: 0 on success, error code on failure. * * NOTE: callers must have locked the page to be pushed. On * exit, the page (and all other pages in the kluster) must be * unlocked. */ /* ARGSUSED */ static int zfs_putapage(vnode_t *vp, page_t *pp, u_offset_t *offp, size_t *lenp, int flags, cred_t *cr) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; dmu_tx_t *tx; u_offset_t off, koff; size_t len, klen; int err; off = pp->p_offset; len = PAGESIZE; /* * If our blocksize is bigger than the page size, try to kluster * multiple pages so that we write a full block (thus avoiding * a read-modify-write). */ if (off < zp->z_size && zp->z_blksz > PAGESIZE) { klen = P2ROUNDUP((ulong_t)zp->z_blksz, PAGESIZE); koff = ISP2(klen) ? P2ALIGN(off, (u_offset_t)klen) : 0; ASSERT(koff <= zp->z_size); if (koff + klen > zp->z_size) klen = P2ROUNDUP(zp->z_size - koff, (uint64_t)PAGESIZE); pp = pvn_write_kluster(vp, pp, &off, &len, koff, klen, flags); } ASSERT3U(btop(len), ==, btopr(len)); /* * Can't push pages past end-of-file. */ if (off >= zp->z_size) { /* ignore all pages */ err = 0; goto out; } else if (off + len > zp->z_size) { int npages = btopr(zp->z_size - off); page_t *trunc; page_list_break(&pp, &trunc, npages); /* ignore pages past end of file */ if (trunc) pvn_write_done(trunc, flags); len = zp->z_size - off; } if (zfs_owner_overquota(zfsvfs, zp, B_FALSE) || zfs_owner_overquota(zfsvfs, zp, B_TRUE)) { err = SET_ERROR(EDQUOT); goto out; } tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_write(tx, zp->z_id, off, len); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); err = dmu_tx_assign(tx, TXG_WAIT); if (err != 0) { dmu_tx_abort(tx); goto out; } if (zp->z_blksz <= PAGESIZE) { caddr_t va = zfs_map_page(pp, S_READ); ASSERT3U(len, <=, PAGESIZE); dmu_write(zfsvfs->z_os, zp->z_id, off, len, va, tx); zfs_unmap_page(pp, va); } else { err = dmu_write_pages(zfsvfs->z_os, zp->z_id, off, len, pp, tx); } if (err == 0) { uint64_t mtime[2], ctime[2]; sa_bulk_attr_t bulk[3]; int count = 0; SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16); SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL, &zp->z_pflags, 8); zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime, B_TRUE); zfs_log_write(zfsvfs->z_log, tx, TX_WRITE, zp, off, len, 0); } dmu_tx_commit(tx); out: pvn_write_done(pp, (err ? B_ERROR : 0) | flags); if (offp) *offp = off; if (lenp) *lenp = len; return (err); } /* * Copy the portion of the file indicated from pages into the file. * The pages are stored in a page list attached to the files vnode. * * IN: vp - vnode of file to push page data to. * off - position in file to put data. * len - amount of data to write. * flags - flags to control the operation. * cr - credentials of caller. * ct - caller context. * * RETURN: 0 on success, error code on failure. * * Timestamps: * vp - ctime|mtime updated */ /*ARGSUSED*/ static int zfs_putpage(vnode_t *vp, offset_t off, size_t len, int flags, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; page_t *pp; size_t io_len; u_offset_t io_off; uint_t blksz; rl_t *rl; int error = 0; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); /* * Align this request to the file block size in case we kluster. * XXX - this can result in pretty aggresive locking, which can * impact simultanious read/write access. One option might be * to break up long requests (len == 0) into block-by-block * operations to get narrower locking. */ blksz = zp->z_blksz; if (ISP2(blksz)) io_off = P2ALIGN_TYPED(off, blksz, u_offset_t); else io_off = 0; if (len > 0 && ISP2(blksz)) io_len = P2ROUNDUP_TYPED(len + (off - io_off), blksz, size_t); else io_len = 0; if (io_len == 0) { /* * Search the entire vp list for pages >= io_off. */ rl = zfs_range_lock(zp, io_off, UINT64_MAX, RL_WRITER); error = pvn_vplist_dirty(vp, io_off, zfs_putapage, flags, cr); goto out; } rl = zfs_range_lock(zp, io_off, io_len, RL_WRITER); if (off > zp->z_size) { /* past end of file */ zfs_range_unlock(rl); ZFS_EXIT(zfsvfs); return (0); } len = MIN(io_len, P2ROUNDUP(zp->z_size, PAGESIZE) - io_off); for (off = io_off; io_off < off + len; io_off += io_len) { if ((flags & B_INVAL) || ((flags & B_ASYNC) == 0)) { pp = page_lookup(vp, io_off, (flags & (B_INVAL | B_FREE)) ? SE_EXCL : SE_SHARED); } else { pp = page_lookup_nowait(vp, io_off, (flags & B_FREE) ? SE_EXCL : SE_SHARED); } if (pp != NULL && pvn_getdirty(pp, flags)) { int err; /* * Found a dirty page to push */ err = zfs_putapage(vp, pp, &io_off, &io_len, flags, cr); if (err) error = err; } else { io_len = PAGESIZE; } } out: zfs_range_unlock(rl); if ((flags & B_ASYNC) == 0 || zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS) zil_commit(zfsvfs->z_log, zp->z_id); ZFS_EXIT(zfsvfs); return (error); } #endif /* sun */ /*ARGSUSED*/ void zfs_inactive(vnode_t *vp, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; int error; rw_enter(&zfsvfs->z_teardown_inactive_lock, RW_READER); if (zp->z_sa_hdl == NULL) { /* * The fs has been unmounted, or we did a * suspend/resume and this file no longer exists. */ rw_exit(&zfsvfs->z_teardown_inactive_lock); vrecycle(vp, curthread); return; } mutex_enter(&zp->z_lock); if (zp->z_unlinked) { /* * Fast path to recycle a vnode of a removed file. */ mutex_exit(&zp->z_lock); rw_exit(&zfsvfs->z_teardown_inactive_lock); vrecycle(vp, curthread); return; } mutex_exit(&zp->z_lock); if (zp->z_atime_dirty && zp->z_unlinked == 0) { dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); } else { mutex_enter(&zp->z_lock); (void) sa_update(zp->z_sa_hdl, SA_ZPL_ATIME(zfsvfs), (void *)&zp->z_atime, sizeof (zp->z_atime), tx); zp->z_atime_dirty = 0; mutex_exit(&zp->z_lock); dmu_tx_commit(tx); } } rw_exit(&zfsvfs->z_teardown_inactive_lock); } #ifdef sun /* * Bounds-check the seek operation. * * IN: vp - vnode seeking within * ooff - old file offset * noffp - pointer to new file offset * ct - caller context * * RETURN: 0 on success, EINVAL if new offset invalid. */ /* ARGSUSED */ static int zfs_seek(vnode_t *vp, offset_t ooff, offset_t *noffp, caller_context_t *ct) { if (vp->v_type == VDIR) return (0); return ((*noffp < 0 || *noffp > MAXOFFSET_T) ? EINVAL : 0); } /* * Pre-filter the generic locking function to trap attempts to place * a mandatory lock on a memory mapped file. */ static int zfs_frlock(vnode_t *vp, int cmd, flock64_t *bfp, int flag, offset_t offset, flk_callback_t *flk_cbp, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); /* * We are following the UFS semantics with respect to mapcnt * here: If we see that the file is mapped already, then we will * return an error, but we don't worry about races between this * function and zfs_map(). */ if (zp->z_mapcnt > 0 && MANDMODE(zp->z_mode)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EAGAIN)); } ZFS_EXIT(zfsvfs); return (fs_frlock(vp, cmd, bfp, flag, offset, flk_cbp, cr, ct)); } /* * If we can't find a page in the cache, we will create a new page * and fill it with file data. For efficiency, we may try to fill * multiple pages at once (klustering) to fill up the supplied page * list. Note that the pages to be filled are held with an exclusive * lock to prevent access by other threads while they are being filled. */ static int zfs_fillpage(vnode_t *vp, u_offset_t off, struct seg *seg, caddr_t addr, page_t *pl[], size_t plsz, enum seg_rw rw) { znode_t *zp = VTOZ(vp); page_t *pp, *cur_pp; objset_t *os = zp->z_zfsvfs->z_os; u_offset_t io_off, total; size_t io_len; int err; if (plsz == PAGESIZE || zp->z_blksz <= PAGESIZE) { /* * We only have a single page, don't bother klustering */ io_off = off; io_len = PAGESIZE; pp = page_create_va(vp, io_off, io_len, PG_EXCL | PG_WAIT, seg, addr); } else { /* * Try to find enough pages to fill the page list */ pp = pvn_read_kluster(vp, off, seg, addr, &io_off, &io_len, off, plsz, 0); } if (pp == NULL) { /* * The page already exists, nothing to do here. */ *pl = NULL; return (0); } /* * Fill the pages in the kluster. */ cur_pp = pp; for (total = io_off + io_len; io_off < total; io_off += PAGESIZE) { caddr_t va; ASSERT3U(io_off, ==, cur_pp->p_offset); va = zfs_map_page(cur_pp, S_WRITE); err = dmu_read(os, zp->z_id, io_off, PAGESIZE, va, DMU_READ_PREFETCH); zfs_unmap_page(cur_pp, va); if (err) { /* On error, toss the entire kluster */ pvn_read_done(pp, B_ERROR); /* convert checksum errors into IO errors */ if (err == ECKSUM) err = SET_ERROR(EIO); return (err); } cur_pp = cur_pp->p_next; } /* * Fill in the page list array from the kluster starting * from the desired offset `off'. * NOTE: the page list will always be null terminated. */ pvn_plist_init(pp, pl, plsz, off, io_len, rw); ASSERT(pl == NULL || (*pl)->p_offset == off); return (0); } /* * Return pointers to the pages for the file region [off, off + len] * in the pl array. If plsz is greater than len, this function may * also return page pointers from after the specified region * (i.e. the region [off, off + plsz]). These additional pages are * only returned if they are already in the cache, or were created as * part of a klustered read. * * IN: vp - vnode of file to get data from. * off - position in file to get data from. * len - amount of data to retrieve. * plsz - length of provided page list. * seg - segment to obtain pages for. * addr - virtual address of fault. * rw - mode of created pages. * cr - credentials of caller. * ct - caller context. * * OUT: protp - protection mode of created pages. * pl - list of pages created. * * RETURN: 0 on success, error code on failure. * * Timestamps: * vp - atime updated */ /* ARGSUSED */ static int zfs_getpage(vnode_t *vp, offset_t off, size_t len, uint_t *protp, page_t *pl[], size_t plsz, struct seg *seg, caddr_t addr, enum seg_rw rw, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; page_t **pl0 = pl; int err = 0; /* we do our own caching, faultahead is unnecessary */ if (pl == NULL) return (0); else if (len > plsz) len = plsz; else len = P2ROUNDUP(len, PAGESIZE); ASSERT(plsz >= len); ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if (protp) *protp = PROT_ALL; /* * Loop through the requested range [off, off + len) looking * for pages. If we don't find a page, we will need to create * a new page and fill it with data from the file. */ while (len > 0) { if (*pl = page_lookup(vp, off, SE_SHARED)) *(pl+1) = NULL; else if (err = zfs_fillpage(vp, off, seg, addr, pl, plsz, rw)) goto out; while (*pl) { ASSERT3U((*pl)->p_offset, ==, off); off += PAGESIZE; addr += PAGESIZE; if (len > 0) { ASSERT3U(len, >=, PAGESIZE); len -= PAGESIZE; } ASSERT3U(plsz, >=, PAGESIZE); plsz -= PAGESIZE; pl++; } } /* * Fill out the page array with any pages already in the cache. */ while (plsz > 0 && (*pl++ = page_lookup_nowait(vp, off, SE_SHARED))) { off += PAGESIZE; plsz -= PAGESIZE; } out: if (err) { /* * Release any pages we have previously locked. */ while (pl > pl0) page_unlock(*--pl); } else { ZFS_ACCESSTIME_STAMP(zfsvfs, zp); } *pl = NULL; ZFS_EXIT(zfsvfs); return (err); } /* * Request a memory map for a section of a file. This code interacts * with common code and the VM system as follows: * * - common code calls mmap(), which ends up in smmap_common() * - this calls VOP_MAP(), which takes you into (say) zfs * - zfs_map() calls as_map(), passing segvn_create() as the callback * - segvn_create() creates the new segment and calls VOP_ADDMAP() * - zfs_addmap() updates z_mapcnt */ /*ARGSUSED*/ static int zfs_map(vnode_t *vp, offset_t off, struct as *as, caddr_t *addrp, size_t len, uchar_t prot, uchar_t maxprot, uint_t flags, cred_t *cr, caller_context_t *ct) { znode_t *zp = VTOZ(vp); zfsvfs_t *zfsvfs = zp->z_zfsvfs; segvn_crargs_t vn_a; int error; ZFS_ENTER(zfsvfs); ZFS_VERIFY_ZP(zp); if ((prot & PROT_WRITE) && (zp->z_pflags & (ZFS_IMMUTABLE | ZFS_READONLY | ZFS_APPENDONLY))) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EPERM)); } if ((prot & (PROT_READ | PROT_EXEC)) && (zp->z_pflags & ZFS_AV_QUARANTINED)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EACCES)); } if (vp->v_flag & VNOMAP) { ZFS_EXIT(zfsvfs); return (SET_ERROR(ENOSYS)); } if (off < 0 || len > MAXOFFSET_T - off) { ZFS_EXIT(zfsvfs); return (SET_ERROR(ENXIO)); } if (vp->v_type != VREG) { ZFS_EXIT(zfsvfs); return (SET_ERROR(ENODEV)); } /* * If file is locked, disallow mapping. */ if (MANDMODE(zp->z_mode) && vn_has_flocks(vp)) { ZFS_EXIT(zfsvfs); return (SET_ERROR(EAGAIN)); } as_rangelock(as); error = choose_addr(as, addrp, len, off, ADDR_VACALIGN, flags); if (error != 0) { as_rangeunlock(as); ZFS_EXIT(zfsvfs); return (error); } vn_a.vp = vp; vn_a.offset = (u_offset_t)off; vn_a.type = flags & MAP_TYPE; vn_a.prot = prot; vn_a.maxprot = maxprot; vn_a.cred = cr; vn_a.amp = NULL; vn_a.flags = flags & ~MAP_TYPE; vn_a.szc = 0; vn_a.lgrp_mem_policy_flags = 0; error = as_map(as, *addrp, len, segvn_create, &vn_a); as_rangeunlock(as); ZFS_EXIT(zfsvfs); return (error); } /* ARGSUSED */ static int zfs_addmap(vnode_t *vp, offset_t off, struct as *as, caddr_t addr, size_t len, uchar_t prot, uchar_t maxprot, uint_t flags, cred_t *cr, caller_context_t *ct) { uint64_t pages = btopr(len); atomic_add_64(&VTOZ(vp)->z_mapcnt, pages); return (0); } /* * The reason we push dirty pages as part of zfs_delmap() is so that we get a * more accurate mtime for the associated file. Since we don't have a way of * detecting when the data was actually modified, we have to resort to * heuristics. If an explicit msync() is done, then we mark the mtime when the * last page is pushed. The problem occurs when the msync() call is omitted, * which by far the most common case: * * open() * mmap() * * munmap() * close() *