mirror_zfs/include/sys/vdev_impl.h

336 lines
11 KiB
C
Raw Normal View History

2008-11-20 23:01:55 +03:00
/*
* 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) 2012 by Delphix. All rights reserved.
2008-11-20 23:01:55 +03:00
*/
#ifndef _SYS_VDEV_IMPL_H
#define _SYS_VDEV_IMPL_H
#include <sys/avl.h>
#include <sys/dmu.h>
#include <sys/metaslab.h>
#include <sys/nvpair.h>
#include <sys/space_map.h>
#include <sys/vdev.h>
#include <sys/dkio.h>
#include <sys/uberblock_impl.h>
#ifdef __cplusplus
extern "C" {
#endif
/*
* Virtual device descriptors.
*
* All storage pool operations go through the virtual device framework,
* which provides data replication and I/O scheduling.
*/
/*
* Forward declarations that lots of things need.
*/
typedef struct vdev_queue vdev_queue_t;
typedef struct vdev_io vdev_io_t;
2008-11-20 23:01:55 +03:00
typedef struct vdev_cache vdev_cache_t;
typedef struct vdev_cache_entry vdev_cache_entry_t;
/*
* Virtual device operations
*/
typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size,
uint64_t *ashift);
2008-11-20 23:01:55 +03:00
typedef void vdev_close_func_t(vdev_t *vd);
typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize);
typedef int vdev_io_start_func_t(zio_t *zio);
typedef void vdev_io_done_func_t(zio_t *zio);
2008-11-20 23:01:55 +03:00
typedef void vdev_state_change_func_t(vdev_t *vd, int, int);
typedef void vdev_hold_func_t(vdev_t *vd);
typedef void vdev_rele_func_t(vdev_t *vd);
2008-11-20 23:01:55 +03:00
typedef struct vdev_ops {
vdev_open_func_t *vdev_op_open;
vdev_close_func_t *vdev_op_close;
vdev_asize_func_t *vdev_op_asize;
vdev_io_start_func_t *vdev_op_io_start;
vdev_io_done_func_t *vdev_op_io_done;
vdev_state_change_func_t *vdev_op_state_change;
vdev_hold_func_t *vdev_op_hold;
vdev_rele_func_t *vdev_op_rele;
2008-11-20 23:01:55 +03:00
char vdev_op_type[16];
boolean_t vdev_op_leaf;
} vdev_ops_t;
/*
* Virtual device properties
*/
struct vdev_cache_entry {
char *ve_data;
uint64_t ve_offset;
uint64_t ve_lastused;
avl_node_t ve_offset_node;
avl_node_t ve_lastused_node;
uint32_t ve_hits;
uint16_t ve_missed_update;
zio_t *ve_fill_io;
};
struct vdev_cache {
avl_tree_t vc_offset_tree;
avl_tree_t vc_lastused_tree;
kmutex_t vc_lock;
};
struct vdev_queue {
avl_tree_t vq_deadline_tree;
avl_tree_t vq_read_tree;
avl_tree_t vq_write_tree;
avl_tree_t vq_pending_tree;
list_t vq_io_list;
2008-11-20 23:01:55 +03:00
kmutex_t vq_lock;
};
struct vdev_io {
char vi_buffer[SPA_MAXBLOCKSIZE]; /* Must be first */
list_node_t vi_node;
};
2008-11-20 23:01:55 +03:00
/*
* Virtual device descriptor
*/
struct vdev {
/*
* Common to all vdev types.
*/
uint64_t vdev_id; /* child number in vdev parent */
uint64_t vdev_guid; /* unique ID for this vdev */
uint64_t vdev_guid_sum; /* self guid + all child guids */
uint64_t vdev_orig_guid; /* orig. guid prior to remove */
2008-11-20 23:01:55 +03:00
uint64_t vdev_asize; /* allocatable device capacity */
2009-07-03 02:44:48 +04:00
uint64_t vdev_min_asize; /* min acceptable asize */
uint64_t vdev_max_asize; /* max acceptable asize */
2008-11-20 23:01:55 +03:00
uint64_t vdev_ashift; /* block alignment shift */
uint64_t vdev_state; /* see VDEV_STATE_* #defines */
uint64_t vdev_prevstate; /* used when reopening a vdev */
vdev_ops_t *vdev_ops; /* vdev operations */
spa_t *vdev_spa; /* spa for this vdev */
void *vdev_tsd; /* type-specific data */
vnode_t *vdev_name_vp; /* vnode for pathname */
vnode_t *vdev_devid_vp; /* vnode for devid */
2008-11-20 23:01:55 +03:00
vdev_t *vdev_top; /* top-level vdev */
vdev_t *vdev_parent; /* parent vdev */
vdev_t **vdev_child; /* array of children */
uint64_t vdev_children; /* number of children */
2009-01-16 00:59:39 +03:00
space_map_t vdev_dtl[DTL_TYPES]; /* in-core dirty time logs */
2008-11-20 23:01:55 +03:00
vdev_stat_t vdev_stat; /* virtual device statistics */
2009-07-03 02:44:48 +04:00
boolean_t vdev_expanding; /* expand the vdev? */
boolean_t vdev_reopening; /* reopen in progress? */
2009-08-18 22:43:27 +04:00
int vdev_open_error; /* error on last open */
kthread_t *vdev_open_thread; /* thread opening children */
uint64_t vdev_crtxg; /* txg when top-level was added */
2008-11-20 23:01:55 +03:00
/*
* Top-level vdev state.
*/
uint64_t vdev_ms_array; /* metaslab array object */
uint64_t vdev_ms_shift; /* metaslab size shift */
uint64_t vdev_ms_count; /* number of metaslabs */
metaslab_group_t *vdev_mg; /* metaslab group */
metaslab_t **vdev_ms; /* metaslab array */
Add FASTWRITE algorithm for synchronous writes. Currently, ZIL blocks are spread over vdevs using hint block pointers managed by the ZIL commit code and passed to metaslab_alloc(). Spreading log blocks accross vdevs is important for performance: indeed, using mutliple disks in parallel decreases the ZIL commit latency, which is the main performance metric for synchronous writes. However, the current implementation suffers from the following issues: 1) It would be best if the ZIL module was not aware of such low-level details. They should be handled by the ZIO and metaslab modules; 2) Because the hint block pointer is managed per log, simultaneous commits from multiple logs might use the same vdevs at the same time, which is inefficient; 3) Because dmu_write() does not honor the block pointer hint, indirect writes are not spread. The naive solution of rotating the metaslab rotor each time a block is allocated for the ZIL or dmu_sync() doesn't work in practice because the first ZIL block to be written is actually allocated during the previous commit. Consequently, when metaslab_alloc() decides the vdev for this block, it will do so while a bunch of other allocations are happening at the same time (from dmu_sync() and other ZILs). This means the vdev for this block is chosen more or less at random. When the next commit happens, there is a high chance (especially when the number of blocks per commit is slightly less than the number of the disks) that one disk will have to write two blocks (with a potential seek) while other disks are sitting idle, which defeats spreading and increases the commit latency. This commit introduces a new concept in the metaslab allocator: fastwrites. Basically, each top-level vdev maintains a counter indicating the number of synchronous writes (from dmu_sync() and the ZIL) which have been allocated but not yet completed. When the metaslab is called with the FASTWRITE flag, it will choose the vdev with the least amount of pending synchronous writes. If there are multiple vdevs with the same value, the first matching vdev (starting from the rotor) is used. Once metaslab_alloc() has decided which vdev the block is allocated to, it updates the fastwrite counter for this vdev. The rationale goes like this: when an allocation is done with FASTWRITE, it "reserves" the vdev until the data is written. Until then, all future allocations will naturally avoid this vdev, even after a full rotation of the rotor. As a result, pending synchronous writes at a given point in time will be nicely spread over all vdevs. This contrasts with the previous algorithm, which is based on the implicit assumption that blocks are written instantaneously after they're allocated. metaslab_fastwrite_mark() and metaslab_fastwrite_unmark() are used to manually increase or decrease fastwrite counters, respectively. They should be used with caution, as there is no per-BP tracking of fastwrite information, so leaks and "double-unmarks" are possible. There is, however, an assert in the vdev teardown code which will fire if the fastwrite counters are not zero when the pool is exported or the vdev removed. Note that as stated above, marking is also done implictly by metaslab_alloc(). ZIO also got a new FASTWRITE flag; when it is used, ZIO will pass it to the metaslab when allocating (assuming ZIO does the allocation, which is only true in the case of dmu_sync). This flag will also trigger an unmark when zio_done() fires. A side-effect of the new algorithm is that when a ZIL stops being used, its last block can stay in the pending state (allocated but not yet written) for a long time, polluting the fastwrite counters. To avoid that, I've implemented a somewhat crude but working solution which unmarks these pending blocks in zil_sync(), thus guaranteeing that linguering fastwrites will get pruned at each sync event. The best performance improvements are observed with pools using a large number of top-level vdevs and heavy synchronous write workflows (especially indirect writes and concurrent writes from multiple ZILs). Real-life testing shows a 200% to 300% performance increase with indirect writes and various commit sizes. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Issue #1013
2012-06-27 17:20:20 +04:00
uint64_t vdev_pending_fastwrite; /* allocated fastwrites */
2008-11-20 23:01:55 +03:00
txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */
txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
boolean_t vdev_remove_wanted; /* async remove wanted? */
boolean_t vdev_probe_wanted; /* async probe wanted? */
uint64_t vdev_removing; /* device is being removed? */
list_node_t vdev_config_dirty_node; /* config dirty list */
list_node_t vdev_state_dirty_node; /* state dirty list */
2008-11-20 23:01:55 +03:00
uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
uint64_t vdev_islog; /* is an intent log device */
uint64_t vdev_ishole; /* is a hole in the namespace */
2008-11-20 23:01:55 +03:00
/*
* Leaf vdev state.
*/
uint64_t vdev_psize; /* physical device capacity */
2009-01-16 00:59:39 +03:00
space_map_obj_t vdev_dtl_smo; /* dirty time log space map obj */
2008-11-20 23:01:55 +03:00
txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */
uint64_t vdev_wholedisk; /* true if this is a whole disk */
uint64_t vdev_offline; /* persistent offline state */
uint64_t vdev_faulted; /* persistent faulted state */
uint64_t vdev_degraded; /* persistent degraded state */
uint64_t vdev_removed; /* persistent removed state */
uint64_t vdev_resilvering; /* persistent resilvering state */
2008-11-20 23:01:55 +03:00
uint64_t vdev_nparity; /* number of parity devices for raidz */
char *vdev_path; /* vdev path (if any) */
char *vdev_devid; /* vdev devid (if any) */
char *vdev_physpath; /* vdev device path (if any) */
2009-07-03 02:44:48 +04:00
char *vdev_fru; /* physical FRU location */
uint64_t vdev_not_present; /* not present during import */
uint64_t vdev_unspare; /* unspare when resilvering done */
hrtime_t vdev_last_try; /* last reopen time */
boolean_t vdev_nowritecache; /* true if flushwritecache failed */
boolean_t vdev_checkremove; /* temporary online test */
boolean_t vdev_forcefault; /* force online fault */
boolean_t vdev_splitting; /* split or repair in progress */
boolean_t vdev_delayed_close; /* delayed device close? */
2008-11-20 23:01:55 +03:00
uint8_t vdev_tmpoffline; /* device taken offline temporarily? */
uint8_t vdev_detached; /* device detached? */
uint8_t vdev_cant_read; /* vdev is failing all reads */
uint8_t vdev_cant_write; /* vdev is failing all writes */
2008-11-20 23:01:55 +03:00
uint64_t vdev_isspare; /* was a hot spare */
uint64_t vdev_isl2cache; /* was a l2cache device */
vdev_queue_t vdev_queue; /* I/O deadline schedule queue */
vdev_cache_t vdev_cache; /* physical block cache */
spa_aux_vdev_t *vdev_aux; /* for l2cache vdevs */
zio_t *vdev_probe_zio; /* root of current probe */
vdev_aux_t vdev_label_aux; /* on-disk aux state */
2008-11-20 23:01:55 +03:00
/*
* For DTrace to work in userland (libzpool) context, these fields must
* remain at the end of the structure. DTrace will use the kernel's
* CTF definition for 'struct vdev', and since the size of a kmutex_t is
* larger in userland, the offsets for the rest of the fields would be
2008-11-20 23:01:55 +03:00
* incorrect.
*/
kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
kmutex_t vdev_stat_lock; /* vdev_stat */
kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */
2008-11-20 23:01:55 +03:00
};
#define VDEV_RAIDZ_MAXPARITY 3
2009-07-03 02:44:48 +04:00
#define VDEV_PAD_SIZE (8 << 10)
/* 2 padding areas (vl_pad1 and vl_pad2) to skip */
#define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2
2008-11-20 23:01:55 +03:00
#define VDEV_PHYS_SIZE (112 << 10)
#define VDEV_UBERBLOCK_RING (128 << 10)
#define VDEV_UBERBLOCK_SHIFT(vd) \
MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT)
#define VDEV_UBERBLOCK_COUNT(vd) \
(VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
#define VDEV_UBERBLOCK_OFFSET(vd, n) \
offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
#define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd))
typedef struct vdev_phys {
char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)];
zio_eck_t vp_zbt;
2008-11-20 23:01:55 +03:00
} vdev_phys_t;
typedef struct vdev_label {
2009-07-03 02:44:48 +04:00
char vl_pad1[VDEV_PAD_SIZE]; /* 8K */
char vl_pad2[VDEV_PAD_SIZE]; /* 8K */
2008-11-20 23:01:55 +03:00
vdev_phys_t vl_vdev_phys; /* 112K */
char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
} vdev_label_t; /* 256K total */
/*
* vdev_dirty() flags
*/
#define VDD_METASLAB 0x01
#define VDD_DTL 0x02
/*
* Size and offset of embedded boot loader region on each label.
* The total size of the first two labels plus the boot area is 4MB.
*/
#define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
#define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
/*
* Size of label regions at the start and end of each leaf device.
*/
#define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
#define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
#define VDEV_LABELS 4
#define VDEV_BEST_LABEL VDEV_LABELS
2008-11-20 23:01:55 +03:00
#define VDEV_ALLOC_LOAD 0
#define VDEV_ALLOC_ADD 1
#define VDEV_ALLOC_SPARE 2
#define VDEV_ALLOC_L2CACHE 3
2009-07-03 02:44:48 +04:00
#define VDEV_ALLOC_ROOTPOOL 4
#define VDEV_ALLOC_SPLIT 5
#define VDEV_ALLOC_ATTACH 6
2008-11-20 23:01:55 +03:00
/*
* Allocate or free a vdev
*/
extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid,
vdev_ops_t *ops);
2008-11-20 23:01:55 +03:00
extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
vdev_t *parent, uint_t id, int alloctype);
extern void vdev_free(vdev_t *vd);
/*
* Add or remove children and parents
*/
extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
extern void vdev_compact_children(vdev_t *pvd);
extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
extern void vdev_remove_parent(vdev_t *cvd);
/*
* vdev sync load and sync
*/
extern void vdev_load_log_state(vdev_t *nvd, vdev_t *ovd);
extern boolean_t vdev_log_state_valid(vdev_t *vd);
2008-11-20 23:01:55 +03:00
extern void vdev_load(vdev_t *vd);
extern void vdev_sync(vdev_t *vd, uint64_t txg);
extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
/*
* Available vdev types.
*/
extern vdev_ops_t vdev_root_ops;
extern vdev_ops_t vdev_mirror_ops;
extern vdev_ops_t vdev_replacing_ops;
extern vdev_ops_t vdev_raidz_ops;
extern vdev_ops_t vdev_disk_ops;
extern vdev_ops_t vdev_file_ops;
extern vdev_ops_t vdev_missing_ops;
extern vdev_ops_t vdev_hole_ops;
2008-11-20 23:01:55 +03:00
extern vdev_ops_t vdev_spare_ops;
/*
* Common size functions
*/
extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
2009-07-03 02:44:48 +04:00
extern uint64_t vdev_get_min_asize(vdev_t *vd);
extern void vdev_set_min_asize(vdev_t *vd);
2008-11-20 23:01:55 +03:00
/*
* zdb uses this tunable, so it must be declared here to make lint happy.
*/
extern int zfs_vdev_cache_size;
#ifdef __cplusplus
}
#endif
#endif /* _SYS_VDEV_IMPL_H */