diff options
Diffstat (limited to 'sys/contrib/openzfs/module')
| -rw-r--r-- | sys/contrib/openzfs/module/os/freebsd/zfs/sysctl_os.c | 244 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c | 23 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zcommon/zfs_prop.c | 2 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/arc.c | 32 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/dmu_zfetch.c | 10 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/dnode.c | 65 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/mmp.c | 2 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/range_tree.c | 2 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/spa_config.c | 2 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/spa_misc.c | 23 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/vdev.c | 17 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/vdev_label.c | 2 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/vdev_queue.c | 2 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/vdev_removal.c | 4 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/zio.c | 35 | ||||
| -rw-r--r-- | sys/contrib/openzfs/module/zfs/zio_inject.c | 38 |
16 files changed, 450 insertions, 53 deletions
diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/sysctl_os.c b/sys/contrib/openzfs/module/os/freebsd/zfs/sysctl_os.c index 393bfaa65ff5..ace2360c032d 100644 --- a/sys/contrib/openzfs/module/os/freebsd/zfs/sysctl_os.c +++ b/sys/contrib/openzfs/module/os/freebsd/zfs/sysctl_os.c @@ -188,6 +188,11 @@ param_set_arc_max(SYSCTL_HANDLER_ARGS) return (0); } +SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_max, + CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, + NULL, 0, param_set_arc_max, "LU", + "Maximum ARC size in bytes (LEGACY)"); + int param_set_arc_min(SYSCTL_HANDLER_ARGS) { @@ -212,6 +217,11 @@ param_set_arc_min(SYSCTL_HANDLER_ARGS) return (0); } +SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_min, + CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, + NULL, 0, param_set_arc_min, "LU", + "Minimum ARC size in bytes (LEGACY)"); + extern uint_t zfs_arc_free_target; int @@ -235,6 +245,16 @@ param_set_arc_free_target(SYSCTL_HANDLER_ARGS) return (0); } +/* + * NOTE: This sysctl is CTLFLAG_RW not CTLFLAG_RWTUN due to its dependency on + * pagedaemon initialization. + */ +SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_free_target, + CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_MPSAFE, + NULL, 0, param_set_arc_free_target, "IU", + "Desired number of free pages below which ARC triggers reclaim" + " (LEGACY)"); + int param_set_arc_no_grow_shift(SYSCTL_HANDLER_ARGS) { @@ -253,6 +273,187 @@ param_set_arc_no_grow_shift(SYSCTL_HANDLER_ARGS) return (0); } +SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_no_grow_shift, + CTLTYPE_INT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, + NULL, 0, param_set_arc_no_grow_shift, "I", + "log2(fraction of ARC which must be free to allow growing) (LEGACY)"); + +extern uint64_t l2arc_write_max; + +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_max, + CTLFLAG_RWTUN, &l2arc_write_max, 0, + "Max write bytes per interval (LEGACY)"); + +extern uint64_t l2arc_write_boost; + +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_boost, + CTLFLAG_RWTUN, &l2arc_write_boost, 0, + "Extra write bytes during device warmup (LEGACY)"); + +extern uint64_t l2arc_headroom; + +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom, + CTLFLAG_RWTUN, &l2arc_headroom, 0, + "Number of max device writes to precache (LEGACY)"); + +extern uint64_t l2arc_headroom_boost; + +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom_boost, + CTLFLAG_RWTUN, &l2arc_headroom_boost, 0, + "Compressed l2arc_headroom multiplier (LEGACY)"); + +extern uint64_t l2arc_feed_secs; + +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_secs, + CTLFLAG_RWTUN, &l2arc_feed_secs, 0, + "Seconds between L2ARC writing (LEGACY)"); + +extern uint64_t l2arc_feed_min_ms; + +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_min_ms, + CTLFLAG_RWTUN, &l2arc_feed_min_ms, 0, + "Min feed interval in milliseconds (LEGACY)"); + +extern int l2arc_noprefetch; + +SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_noprefetch, + CTLFLAG_RWTUN, &l2arc_noprefetch, 0, + "Skip caching prefetched buffers (LEGACY)"); + +extern int l2arc_feed_again; + +SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_feed_again, + CTLFLAG_RWTUN, &l2arc_feed_again, 0, + "Turbo L2ARC warmup (LEGACY)"); + +extern int l2arc_norw; + +SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_norw, + CTLFLAG_RWTUN, &l2arc_norw, 0, + "No reads during writes (LEGACY)"); + +static int +param_get_arc_state_size(SYSCTL_HANDLER_ARGS) +{ + arc_state_t *state = (arc_state_t *)arg1; + int64_t val; + + val = zfs_refcount_count(&state->arcs_size[ARC_BUFC_DATA]) + + zfs_refcount_count(&state->arcs_size[ARC_BUFC_METADATA]); + return (sysctl_handle_64(oidp, &val, 0, req)); +} + +extern arc_state_t ARC_anon; + +SYSCTL_PROC(_vfs_zfs, OID_AUTO, anon_size, + CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE, + &ARC_anon, 0, param_get_arc_state_size, "Q", + "size of anonymous state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_metadata_esize, CTLFLAG_RD, + &ARC_anon.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, + "size of evictable metadata in anonymous state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_data_esize, CTLFLAG_RD, + &ARC_anon.arcs_esize[ARC_BUFC_DATA].rc_count, 0, + "size of evictable data in anonymous state"); + +extern arc_state_t ARC_mru; + +SYSCTL_PROC(_vfs_zfs, OID_AUTO, mru_size, + CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE, + &ARC_mru, 0, param_get_arc_state_size, "Q", + "size of mru state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_metadata_esize, CTLFLAG_RD, + &ARC_mru.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, + "size of evictable metadata in mru state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_data_esize, CTLFLAG_RD, + &ARC_mru.arcs_esize[ARC_BUFC_DATA].rc_count, 0, + "size of evictable data in mru state"); + +extern arc_state_t ARC_mru_ghost; + +SYSCTL_PROC(_vfs_zfs, OID_AUTO, mru_ghost_size, + CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE, + &ARC_mru_ghost, 0, param_get_arc_state_size, "Q", + "size of mru ghost state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_metadata_esize, CTLFLAG_RD, + &ARC_mru_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, + "size of evictable metadata in mru ghost state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_data_esize, CTLFLAG_RD, + &ARC_mru_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0, + "size of evictable data in mru ghost state"); + +extern arc_state_t ARC_mfu; + +SYSCTL_PROC(_vfs_zfs, OID_AUTO, mfu_size, + CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE, + &ARC_mfu, 0, param_get_arc_state_size, "Q", + "size of mfu state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_metadata_esize, CTLFLAG_RD, + &ARC_mfu.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, + "size of evictable metadata in mfu state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_data_esize, CTLFLAG_RD, + &ARC_mfu.arcs_esize[ARC_BUFC_DATA].rc_count, 0, + "size of evictable data in mfu state"); + +extern arc_state_t ARC_mfu_ghost; + +SYSCTL_PROC(_vfs_zfs, OID_AUTO, mfu_ghost_size, + CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE, + &ARC_mfu_ghost, 0, param_get_arc_state_size, "Q", + "size of mfu ghost state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_metadata_esize, CTLFLAG_RD, + &ARC_mfu_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, + "size of evictable metadata in mfu ghost state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_data_esize, CTLFLAG_RD, + &ARC_mfu_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0, + "size of evictable data in mfu ghost state"); + +extern arc_state_t ARC_uncached; + +SYSCTL_PROC(_vfs_zfs, OID_AUTO, uncached_size, + CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE, + &ARC_uncached, 0, param_get_arc_state_size, "Q", + "size of uncached state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, uncached_metadata_esize, CTLFLAG_RD, + &ARC_uncached.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, + "size of evictable metadata in uncached state"); +SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, uncached_data_esize, CTLFLAG_RD, + &ARC_uncached.arcs_esize[ARC_BUFC_DATA].rc_count, 0, + "size of evictable data in uncached state"); + +extern arc_state_t ARC_l2c_only; + +SYSCTL_PROC(_vfs_zfs, OID_AUTO, l2c_only_size, + CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE, + &ARC_l2c_only, 0, param_get_arc_state_size, "Q", + "size of l2c_only state"); + +/* dbuf.c */ + +/* dmu.c */ + +/* dmu_zfetch.c */ + +SYSCTL_NODE(_vfs_zfs, OID_AUTO, zfetch, CTLFLAG_RW, 0, "ZFS ZFETCH (LEGACY)"); + +extern uint32_t zfetch_max_distance; + +SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_distance, + CTLFLAG_RWTUN, &zfetch_max_distance, 0, + "Max bytes to prefetch per stream (LEGACY)"); + +extern uint32_t zfetch_max_idistance; + +SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_idistance, + CTLFLAG_RWTUN, &zfetch_max_idistance, 0, + "Max bytes to prefetch indirects for per stream (LEGACY)"); + +/* dsl_pool.c */ + +/* dnode.c */ + +/* dsl_scan.c */ + /* metaslab.c */ int @@ -313,6 +514,19 @@ SYSCTL_UINT(_vfs_zfs, OID_AUTO, condense_pct, "Condense on-disk spacemap when it is more than this many percents" " of in-memory counterpart"); +extern uint_t zfs_remove_max_segment; + +SYSCTL_UINT(_vfs_zfs, OID_AUTO, remove_max_segment, + CTLFLAG_RWTUN, &zfs_remove_max_segment, 0, + "Largest contiguous segment ZFS will attempt to allocate when removing" + " a device"); + +extern int zfs_removal_suspend_progress; + +SYSCTL_INT(_vfs_zfs, OID_AUTO, removal_suspend_progress, + CTLFLAG_RWTUN, &zfs_removal_suspend_progress, 0, + "Ensures certain actions can happen while in the middle of a removal"); + /* * Minimum size which forces the dynamic allocator to change * it's allocation strategy. Once the space map cannot satisfy @@ -535,6 +749,12 @@ param_set_min_auto_ashift(SYSCTL_HANDLER_ARGS) return (0); } +SYSCTL_PROC(_vfs_zfs, OID_AUTO, min_auto_ashift, + CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, + &zfs_vdev_min_auto_ashift, sizeof (zfs_vdev_min_auto_ashift), + param_set_min_auto_ashift, "IU", + "Min ashift used when creating new top-level vdev. (LEGACY)"); + int param_set_max_auto_ashift(SYSCTL_HANDLER_ARGS) { @@ -554,6 +774,13 @@ param_set_max_auto_ashift(SYSCTL_HANDLER_ARGS) return (0); } +SYSCTL_PROC(_vfs_zfs, OID_AUTO, max_auto_ashift, + CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, + &zfs_vdev_max_auto_ashift, sizeof (zfs_vdev_max_auto_ashift), + param_set_max_auto_ashift, "IU", + "Max ashift used when optimizing for logical -> physical sector size on" + " new top-level vdevs. (LEGACY)"); + /* * Since the DTL space map of a vdev is not expected to have a lot of * entries, we default its block size to 4K. @@ -575,6 +802,23 @@ SYSCTL_INT(_vfs_zfs, OID_AUTO, standard_sm_blksz, CTLFLAG_RDTUN, &zfs_vdev_standard_sm_blksz, 0, "Block size for standard space map. Power of 2 greater than 4096."); +extern int vdev_validate_skip; + +SYSCTL_INT(_vfs_zfs, OID_AUTO, validate_skip, + CTLFLAG_RDTUN, &vdev_validate_skip, 0, + "Enable to bypass vdev_validate()."); + +/* vdev_mirror.c */ + +/* vdev_queue.c */ + +extern uint_t zfs_vdev_max_active; + +SYSCTL_UINT(_vfs_zfs, OID_AUTO, top_maxinflight, + CTLFLAG_RWTUN, &zfs_vdev_max_active, 0, + "The maximum number of I/Os of all types active for each device." + " (LEGACY)"); + /* zio.c */ SYSCTL_INT(_vfs_zfs_zio, OID_AUTO, exclude_metadata, diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c b/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c index bac166fcd89e..967a018640e1 100644 --- a/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c +++ b/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c @@ -484,7 +484,28 @@ zvol_request_impl(zvol_state_t *zv, struct bio *bio, struct request *rq, fstrans_cookie_t cookie = spl_fstrans_mark(); uint64_t offset = io_offset(bio, rq); uint64_t size = io_size(bio, rq); - int rw = io_data_dir(bio, rq); + int rw; + + if (rq != NULL) { + /* + * Flush & trim requests go down the zvol_write codepath. Or + * more specifically: + * + * If request is a write, or if it's op_is_sync() and not a + * read, or if it's a flush, or if it's a discard, then send the + * request down the write path. + */ + if (op_is_write(rq->cmd_flags) || + (op_is_sync(rq->cmd_flags) && req_op(rq) != REQ_OP_READ) || + req_op(rq) == REQ_OP_FLUSH || + op_is_discard(rq->cmd_flags)) { + rw = WRITE; + } else { + rw = READ; + } + } else { + rw = bio_data_dir(bio); + } if (unlikely(zv->zv_flags & ZVOL_REMOVING)) { zvol_end_io(bio, rq, SET_ERROR(ENXIO)); diff --git a/sys/contrib/openzfs/module/zcommon/zfs_prop.c b/sys/contrib/openzfs/module/zcommon/zfs_prop.c index 864e3898b365..9190ae0362ea 100644 --- a/sys/contrib/openzfs/module/zcommon/zfs_prop.c +++ b/sys/contrib/openzfs/module/zcommon/zfs_prop.c @@ -364,8 +364,8 @@ zfs_prop_init(void) static const zprop_index_t xattr_table[] = { { "off", ZFS_XATTR_OFF }, - { "on", ZFS_XATTR_SA }, { "sa", ZFS_XATTR_SA }, + { "on", ZFS_XATTR_SA }, { "dir", ZFS_XATTR_DIR }, { NULL } }; diff --git a/sys/contrib/openzfs/module/zfs/arc.c b/sys/contrib/openzfs/module/zfs/arc.c index 591e2dade59e..b677f90280d7 100644 --- a/sys/contrib/openzfs/module/zfs/arc.c +++ b/sys/contrib/openzfs/module/zfs/arc.c @@ -486,13 +486,13 @@ static taskq_t *arc_flush_taskq; static uint_t zfs_arc_evict_threads = 0; /* The 7 states: */ -static arc_state_t ARC_anon; -/* */ arc_state_t ARC_mru; -static arc_state_t ARC_mru_ghost; -/* */ arc_state_t ARC_mfu; -static arc_state_t ARC_mfu_ghost; -static arc_state_t ARC_l2c_only; -static arc_state_t ARC_uncached; +arc_state_t ARC_anon; +arc_state_t ARC_mru; +arc_state_t ARC_mru_ghost; +arc_state_t ARC_mfu; +arc_state_t ARC_mfu_ghost; +arc_state_t ARC_l2c_only; +arc_state_t ARC_uncached; arc_stats_t arc_stats = { { "hits", KSTAT_DATA_UINT64 }, @@ -832,15 +832,15 @@ typedef struct arc_async_flush { #define L2ARC_FEED_TYPES 4 /* L2ARC Performance Tunables */ -static uint64_t l2arc_write_max = L2ARC_WRITE_SIZE; /* def max write size */ -static uint64_t l2arc_write_boost = L2ARC_WRITE_SIZE; /* extra warmup write */ -static uint64_t l2arc_headroom = L2ARC_HEADROOM; /* # of dev writes */ -static uint64_t l2arc_headroom_boost = L2ARC_HEADROOM_BOOST; -static uint64_t l2arc_feed_secs = L2ARC_FEED_SECS; /* interval seconds */ -static uint64_t l2arc_feed_min_ms = L2ARC_FEED_MIN_MS; /* min interval msecs */ -static int l2arc_noprefetch = B_TRUE; /* don't cache prefetch bufs */ -static int l2arc_feed_again = B_TRUE; /* turbo warmup */ -static int l2arc_norw = B_FALSE; /* no reads during writes */ +uint64_t l2arc_write_max = L2ARC_WRITE_SIZE; /* def max write size */ +uint64_t l2arc_write_boost = L2ARC_WRITE_SIZE; /* extra warmup write */ +uint64_t l2arc_headroom = L2ARC_HEADROOM; /* # 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 msecs */ +int l2arc_noprefetch = B_TRUE; /* don't cache prefetch bufs */ +int l2arc_feed_again = B_TRUE; /* turbo warmup */ +int l2arc_norw = B_FALSE; /* no reads during writes */ static uint_t l2arc_meta_percent = 33; /* limit on headers size */ /* diff --git a/sys/contrib/openzfs/module/zfs/dmu_zfetch.c b/sys/contrib/openzfs/module/zfs/dmu_zfetch.c index 3d3a9c713568..51165d0bf723 100644 --- a/sys/contrib/openzfs/module/zfs/dmu_zfetch.c +++ b/sys/contrib/openzfs/module/zfs/dmu_zfetch.c @@ -57,19 +57,19 @@ static unsigned int zfetch_max_sec_reap = 2; /* min bytes to prefetch per stream (default 2MB) */ static unsigned int zfetch_min_distance = 2 * 1024 * 1024; /* max bytes to prefetch per stream (default 8MB) */ -static unsigned int zfetch_max_distance = 8 * 1024 * 1024; +unsigned int zfetch_max_distance = 8 * 1024 * 1024; #else /* min bytes to prefetch per stream (default 4MB) */ static unsigned int zfetch_min_distance = 4 * 1024 * 1024; /* max bytes to prefetch per stream (default 64MB) */ -static unsigned int zfetch_max_distance = 64 * 1024 * 1024; +unsigned int zfetch_max_distance = 64 * 1024 * 1024; #endif /* max bytes to prefetch indirects for per stream (default 128MB) */ -static unsigned int zfetch_max_idistance = 128 * 1024 * 1024; +unsigned int zfetch_max_idistance = 128 * 1024 * 1024; /* max request reorder distance within a stream (default 16MB) */ -static unsigned int zfetch_max_reorder = 16 * 1024 * 1024; +unsigned int zfetch_max_reorder = 16 * 1024 * 1024; /* Max log2 fraction of holes in a stream */ -static unsigned int zfetch_hole_shift = 2; +unsigned int zfetch_hole_shift = 2; typedef struct zfetch_stats { kstat_named_t zfetchstat_hits; diff --git a/sys/contrib/openzfs/module/zfs/dnode.c b/sys/contrib/openzfs/module/zfs/dnode.c index 6c150d31c669..e88d394b5229 100644 --- a/sys/contrib/openzfs/module/zfs/dnode.c +++ b/sys/contrib/openzfs/module/zfs/dnode.c @@ -2656,6 +2656,32 @@ dnode_next_offset_level(dnode_t *dn, int flags, uint64_t *offset, } /* + * Adjust *offset to the next (or previous) block byte offset at lvl. + * Returns FALSE if *offset would overflow or underflow. + */ +static boolean_t +dnode_next_block(dnode_t *dn, int flags, uint64_t *offset, int lvl) +{ + int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; + int span = lvl * epbs + dn->dn_datablkshift; + uint64_t blkid, maxblkid; + + if (span >= 8 * sizeof (uint64_t)) + return (B_FALSE); + + blkid = *offset >> span; + maxblkid = 1ULL << (8 * sizeof (*offset) - span); + if (!(flags & DNODE_FIND_BACKWARDS) && blkid + 1 < maxblkid) + *offset = (blkid + 1) << span; + else if ((flags & DNODE_FIND_BACKWARDS) && blkid > 0) + *offset = (blkid << span) - 1; + else + return (B_FALSE); + + return (B_TRUE); +} + +/* * Find the next hole, data, or sparse region at or after *offset. * The value 'blkfill' tells us how many items we expect to find * in an L0 data block; this value is 1 for normal objects, @@ -2682,7 +2708,7 @@ int dnode_next_offset(dnode_t *dn, int flags, uint64_t *offset, int minlvl, uint64_t blkfill, uint64_t txg) { - uint64_t initial_offset = *offset; + uint64_t matched = *offset; int lvl, maxlvl; int error = 0; @@ -2706,16 +2732,36 @@ dnode_next_offset(dnode_t *dn, int flags, uint64_t *offset, maxlvl = dn->dn_phys->dn_nlevels; - for (lvl = minlvl; lvl <= maxlvl; lvl++) { + for (lvl = minlvl; lvl <= maxlvl; ) { error = dnode_next_offset_level(dn, flags, offset, lvl, blkfill, txg); - if (error != ESRCH) + if (error == 0 && lvl > minlvl) { + --lvl; + matched = *offset; + } else if (error == ESRCH && lvl < maxlvl && + dnode_next_block(dn, flags, &matched, lvl)) { + /* + * Continue search at next/prev offset in lvl+1 block. + * + * Usually we only search upwards at the start of the + * search as higher level blocks point at a matching + * minlvl block in most cases, but we backtrack if not. + * + * This can happen for txg > 0 searches if the block + * contains only BPs/dnodes freed at that txg. It also + * happens if we are still syncing out the tree, and + * some BP's at higher levels are not updated yet. + * + * We must adjust offset to avoid coming back to the + * same offset and getting stuck looping forever. This + * also deals with the case where offset is already at + * the beginning or end of the object. + */ + ++lvl; + *offset = matched; + } else { break; - } - - while (error == 0 && --lvl >= minlvl) { - error = dnode_next_offset_level(dn, - flags, offset, lvl, blkfill, txg); + } } /* @@ -2727,9 +2773,6 @@ dnode_next_offset(dnode_t *dn, int flags, uint64_t *offset, error = 0; } - if (error == 0 && (flags & DNODE_FIND_BACKWARDS ? - initial_offset < *offset : initial_offset > *offset)) - error = SET_ERROR(ESRCH); out: if (!(flags & DNODE_FIND_HAVELOCK)) rw_exit(&dn->dn_struct_rwlock); diff --git a/sys/contrib/openzfs/module/zfs/mmp.c b/sys/contrib/openzfs/module/zfs/mmp.c index 7db72b9b04b0..fd46127b6068 100644 --- a/sys/contrib/openzfs/module/zfs/mmp.c +++ b/sys/contrib/openzfs/module/zfs/mmp.c @@ -446,7 +446,7 @@ mmp_write_uberblock(spa_t *spa) uint64_t offset; hrtime_t lock_acquire_time = gethrtime(); - spa_config_enter_mmp(spa, SCL_STATE, mmp_tag, RW_READER); + spa_config_enter_priority(spa, SCL_STATE, mmp_tag, RW_READER); lock_acquire_time = gethrtime() - lock_acquire_time; if (lock_acquire_time > (MSEC2NSEC(MMP_MIN_INTERVAL) / 10)) zfs_dbgmsg("MMP SCL_STATE acquisition pool '%s' took %llu ns " diff --git a/sys/contrib/openzfs/module/zfs/range_tree.c b/sys/contrib/openzfs/module/zfs/range_tree.c index ea2d2c7227c8..d73195f1a21f 100644 --- a/sys/contrib/openzfs/module/zfs/range_tree.c +++ b/sys/contrib/openzfs/module/zfs/range_tree.c @@ -585,7 +585,7 @@ zfs_range_tree_remove_impl(zfs_range_tree_t *rt, uint64_t start, uint64_t size, * the size, since we do not support removing partial segments * of range trees with gaps. */ - zfs_zfs_rs_set_fill_raw(rs, rt, zfs_rs_get_end_raw(rs, rt) - + zfs_rs_set_fill_raw(rs, rt, zfs_rs_get_end_raw(rs, rt) - zfs_rs_get_start_raw(rs, rt)); zfs_range_tree_stat_incr(rt, &rs_tmp); diff --git a/sys/contrib/openzfs/module/zfs/spa_config.c b/sys/contrib/openzfs/module/zfs/spa_config.c index cf28955b0c50..f615591e826b 100644 --- a/sys/contrib/openzfs/module/zfs/spa_config.c +++ b/sys/contrib/openzfs/module/zfs/spa_config.c @@ -372,6 +372,8 @@ spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats) fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG, txg); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID, spa_guid(spa)); fnvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA, spa->spa_errata); + fnvlist_add_uint64(config, ZPOOL_CONFIG_MIN_ALLOC, spa->spa_min_alloc); + fnvlist_add_uint64(config, ZPOOL_CONFIG_MAX_ALLOC, spa->spa_max_alloc); if (spa->spa_comment != NULL) fnvlist_add_string(config, ZPOOL_CONFIG_COMMENT, spa->spa_comment); diff --git a/sys/contrib/openzfs/module/zfs/spa_misc.c b/sys/contrib/openzfs/module/zfs/spa_misc.c index 6f7c060f97f8..0bead6d49666 100644 --- a/sys/contrib/openzfs/module/zfs/spa_misc.c +++ b/sys/contrib/openzfs/module/zfs/spa_misc.c @@ -510,7 +510,7 @@ spa_config_tryenter(spa_t *spa, int locks, const void *tag, krw_t rw) static void spa_config_enter_impl(spa_t *spa, int locks, const void *tag, krw_t rw, - int mmp_flag) + int priority_flag) { (void) tag; int wlocks_held = 0; @@ -526,7 +526,7 @@ spa_config_enter_impl(spa_t *spa, int locks, const void *tag, krw_t rw, mutex_enter(&scl->scl_lock); if (rw == RW_READER) { while (scl->scl_writer || - (!mmp_flag && scl->scl_write_wanted)) { + (!priority_flag && scl->scl_write_wanted)) { cv_wait(&scl->scl_cv, &scl->scl_lock); } } else { @@ -551,7 +551,7 @@ spa_config_enter(spa_t *spa, int locks, const void *tag, krw_t rw) } /* - * The spa_config_enter_mmp() allows the mmp thread to cut in front of + * The spa_config_enter_priority() allows the mmp thread to cut in front of * outstanding write lock requests. This is needed since the mmp updates are * time sensitive and failure to service them promptly will result in a * suspended pool. This pool suspension has been seen in practice when there is @@ -560,7 +560,7 @@ spa_config_enter(spa_t *spa, int locks, const void *tag, krw_t rw) */ void -spa_config_enter_mmp(spa_t *spa, int locks, const void *tag, krw_t rw) +spa_config_enter_priority(spa_t *spa, int locks, const void *tag, krw_t rw) { spa_config_enter_impl(spa, locks, tag, rw, 1); } @@ -806,6 +806,7 @@ spa_add(const char *name, nvlist_t *config, const char *altroot) spa->spa_min_ashift = INT_MAX; spa->spa_max_ashift = 0; spa->spa_min_alloc = INT_MAX; + spa->spa_max_alloc = 0; spa->spa_gcd_alloc = INT_MAX; /* Reset cached value */ @@ -1865,6 +1866,19 @@ spa_get_worst_case_asize(spa_t *spa, uint64_t lsize) } /* + * Return the range of minimum allocation sizes for the normal allocation + * class. This can be used by external consumers of the DMU to estimate + * potential wasted capacity when setting the recordsize for an object. + * This is mainly for dRAID pools which always pad to a full stripe width. + */ +void +spa_get_min_alloc_range(spa_t *spa, uint64_t *min_alloc, uint64_t *max_alloc) +{ + *min_alloc = spa->spa_min_alloc; + *max_alloc = spa->spa_max_alloc; +} + +/* * Return the amount of slop space in bytes. It is typically 1/32 of the pool * (3.2%), minus the embedded log space. On very small pools, it may be * slightly larger than this. On very large pools, it will be capped to @@ -3085,6 +3099,7 @@ EXPORT_SYMBOL(spa_version); EXPORT_SYMBOL(spa_state); EXPORT_SYMBOL(spa_load_state); EXPORT_SYMBOL(spa_freeze_txg); +EXPORT_SYMBOL(spa_get_min_alloc_range); /* for Lustre */ EXPORT_SYMBOL(spa_get_dspace); EXPORT_SYMBOL(spa_update_dspace); EXPORT_SYMBOL(spa_deflate); diff --git a/sys/contrib/openzfs/module/zfs/vdev.c b/sys/contrib/openzfs/module/zfs/vdev.c index fc6d445f9785..c8d7280387a2 100644 --- a/sys/contrib/openzfs/module/zfs/vdev.c +++ b/sys/contrib/openzfs/module/zfs/vdev.c @@ -100,7 +100,7 @@ static uint_t zfs_vdev_default_ms_shift = 29; /* upper limit for metaslab size (16G) */ static uint_t zfs_vdev_max_ms_shift = 34; -static int vdev_validate_skip = B_FALSE; +int vdev_validate_skip = B_FALSE; /* * Since the DTL space map of a vdev is not expected to have a lot of @@ -1497,12 +1497,14 @@ vdev_spa_set_alloc(spa_t *spa, uint64_t min_alloc) { if (min_alloc < spa->spa_min_alloc) spa->spa_min_alloc = min_alloc; - if (spa->spa_gcd_alloc == INT_MAX) { + + if (min_alloc > spa->spa_max_alloc) + spa->spa_max_alloc = min_alloc; + + if (spa->spa_gcd_alloc == INT_MAX) spa->spa_gcd_alloc = min_alloc; - } else { - spa->spa_gcd_alloc = vdev_gcd(min_alloc, - spa->spa_gcd_alloc); - } + else + spa->spa_gcd_alloc = vdev_gcd(min_alloc, spa->spa_gcd_alloc); } void @@ -1560,8 +1562,7 @@ vdev_metaslab_group_create(vdev_t *vd) if (vd->vdev_ashift < spa->spa_min_ashift) spa->spa_min_ashift = vd->vdev_ashift; - uint64_t min_alloc = vdev_get_min_alloc(vd); - vdev_spa_set_alloc(spa, min_alloc); + vdev_spa_set_alloc(spa, vdev_get_min_alloc(vd)); } } } diff --git a/sys/contrib/openzfs/module/zfs/vdev_label.c b/sys/contrib/openzfs/module/zfs/vdev_label.c index c44f654b0261..0d4fdaa77ba0 100644 --- a/sys/contrib/openzfs/module/zfs/vdev_label.c +++ b/sys/contrib/openzfs/module/zfs/vdev_label.c @@ -511,6 +511,8 @@ vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats, fnvlist_add_uint64(nv, ZPOOL_CONFIG_ASHIFT, vd->vdev_ashift); fnvlist_add_uint64(nv, ZPOOL_CONFIG_ASIZE, vd->vdev_asize); + fnvlist_add_uint64(nv, ZPOOL_CONFIG_MIN_ALLOC, + vdev_get_min_alloc(vd)); fnvlist_add_uint64(nv, ZPOOL_CONFIG_IS_LOG, vd->vdev_islog); if (vd->vdev_noalloc) { fnvlist_add_uint64(nv, ZPOOL_CONFIG_NONALLOCATING, diff --git a/sys/contrib/openzfs/module/zfs/vdev_queue.c b/sys/contrib/openzfs/module/zfs/vdev_queue.c index e69e5598939e..c12713b107bf 100644 --- a/sys/contrib/openzfs/module/zfs/vdev_queue.c +++ b/sys/contrib/openzfs/module/zfs/vdev_queue.c @@ -122,7 +122,7 @@ * The maximum number of i/os active to each device. Ideally, this will be >= * the sum of each queue's max_active. */ -static uint_t zfs_vdev_max_active = 1000; +uint_t zfs_vdev_max_active = 1000; /* * Per-queue limits on the number of i/os active to each device. If the diff --git a/sys/contrib/openzfs/module/zfs/vdev_removal.c b/sys/contrib/openzfs/module/zfs/vdev_removal.c index 2ce0121324ad..2f7a739da241 100644 --- a/sys/contrib/openzfs/module/zfs/vdev_removal.c +++ b/sys/contrib/openzfs/module/zfs/vdev_removal.c @@ -105,7 +105,7 @@ static const uint_t zfs_remove_max_copy_bytes = 64 * 1024 * 1024; * * See also the accessor function spa_remove_max_segment(). */ -static uint_t zfs_remove_max_segment = SPA_MAXBLOCKSIZE; +uint_t zfs_remove_max_segment = SPA_MAXBLOCKSIZE; /* * Ignore hard IO errors during device removal. When set if a device @@ -137,7 +137,7 @@ uint_t vdev_removal_max_span = 32 * 1024; * This is used by the test suite so that it can ensure that certain * actions happen while in the middle of a removal. */ -static int zfs_removal_suspend_progress = 0; +int zfs_removal_suspend_progress = 0; #define VDEV_REMOVAL_ZAP_OBJS "lzap" diff --git a/sys/contrib/openzfs/module/zfs/zio.c b/sys/contrib/openzfs/module/zfs/zio.c index 4cf8912d4269..aeea58bedfe4 100644 --- a/sys/contrib/openzfs/module/zfs/zio.c +++ b/sys/contrib/openzfs/module/zfs/zio.c @@ -4574,8 +4574,29 @@ zio_vdev_io_start(zio_t *zio) ASSERT0(zio->io_child_error[ZIO_CHILD_VDEV]); if (vd == NULL) { - if (!(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) - spa_config_enter(spa, SCL_ZIO, zio, RW_READER); + if (!(zio->io_flags & ZIO_FLAG_CONFIG_WRITER)) { + /* + * A deadlock workaround. The ddt_prune_unique_entries() + * -> prune_candidates_sync() code path takes the + * SCL_ZIO reader lock and may request it again here. + * If there is another thread who wants the SCL_ZIO + * writer lock, then scl_write_wanted will be set. + * Thus, the spa_config_enter_priority() is used to + * ignore pending writer requests. + * + * The locking should be revised to remove the need + * for this workaround. If that's not workable then + * it should only be applied to the zios involved in + * the pruning process. This impacts the read/write + * I/O balance while pruning. + */ + if (spa->spa_active_ddt_prune) + spa_config_enter_priority(spa, SCL_ZIO, zio, + RW_READER); + else + spa_config_enter(spa, SCL_ZIO, zio, + RW_READER); + } /* * The mirror_ops handle multiple DVAs in a single BP. @@ -5305,6 +5326,16 @@ zio_ready(zio_t *zio) return (NULL); } + if (zio_injection_enabled) { + hrtime_t target = zio_handle_ready_delay(zio); + if (target != 0 && zio->io_target_timestamp == 0) { + zio->io_stage >>= 1; + zio->io_target_timestamp = target; + zio_delay_interrupt(zio); + return (NULL); + } + } + if (zio->io_ready) { ASSERT(IO_IS_ALLOCATING(zio)); ASSERT(BP_GET_BIRTH(bp) == zio->io_txg || diff --git a/sys/contrib/openzfs/module/zfs/zio_inject.c b/sys/contrib/openzfs/module/zfs/zio_inject.c index 981a1be4847c..287577018ed1 100644 --- a/sys/contrib/openzfs/module/zfs/zio_inject.c +++ b/sys/contrib/openzfs/module/zfs/zio_inject.c @@ -827,6 +827,44 @@ zio_handle_export_delay(spa_t *spa, hrtime_t elapsed) zio_handle_pool_delay(spa, elapsed, ZINJECT_DELAY_EXPORT); } +/* + * For testing, inject a delay before ready state. + */ +hrtime_t +zio_handle_ready_delay(zio_t *zio) +{ + inject_handler_t *handler; + hrtime_t now = gethrtime(); + hrtime_t target = 0; + + /* + * Ignore I/O not associated with any logical data. + */ + if (zio->io_logical == NULL) + return (0); + + rw_enter(&inject_lock, RW_READER); + + for (handler = list_head(&inject_handlers); handler != NULL; + handler = list_next(&inject_handlers, handler)) { + if (zio->io_spa != handler->zi_spa || + handler->zi_record.zi_cmd != ZINJECT_DELAY_READY) + continue; + + /* If this handler matches, inject the delay */ + if (zio_match_iotype(zio, handler->zi_record.zi_iotype) && + zio_match_handler(&zio->io_logical->io_bookmark, + zio->io_bp ? BP_GET_TYPE(zio->io_bp) : DMU_OT_NONE, + zio_match_dva(zio), &handler->zi_record, zio->io_error)) { + target = now + (hrtime_t)handler->zi_record.zi_timer; + break; + } + } + + rw_exit(&inject_lock); + return (target); +} + static int zio_calculate_range(const char *pool, zinject_record_t *record) { |