Merge branch 'kstat'

This branch updates several of the zfs kstats to take advantage
of the improved raw kstat functionality.  In addition, two new
kstats and a script called dbufstat.py are introduced.

Updated+New Kstats
* dbufs        - Stats for all dbufs in the dbuf_hash
* <pool>/txgs  - Stats for the last N txgs synced to disk
* <pool>/reads - Stats for rhe last N reads issues by the ARC
* <pool>/dmu_tx_assign - Histogram of tx assign times

Signed-off-by: Brian Behlendorf <[email protected]>

8 files changed, 83 insertions, 22 deletions

diff --git a/include/sys/arc.h b/include/sys/arc.h
index 8c10d947c..efafb551a 100644
--- a/include/sys/arc.h
+++ b/include/sys/arc.h
@@ -92,6 +92,36 @@ typedef enum arc_space_type {
 	ARC_SPACE_NUMTYPES
 } arc_space_type_t;
 
+typedef enum arc_state_type {
+	ARC_STATE_ANON,
+	ARC_STATE_MRU,
+	ARC_STATE_MRU_GHOST,
+	ARC_STATE_MFU,
+	ARC_STATE_MFU_GHOST,
+	ARC_STATE_L2C_ONLY,
+	ARC_STATE_NUMTYPES
+} arc_state_type_t;
+
+typedef struct arc_buf_info {
+	arc_state_type_t	abi_state_type;
+	arc_buf_contents_t	abi_state_contents;
+	uint64_t		abi_state_index;
+	uint32_t		abi_flags;
+	uint32_t		abi_datacnt;
+	uint64_t		abi_size;
+	uint64_t		abi_spa;
+	uint64_t		abi_access;
+	uint32_t		abi_mru_hits;
+	uint32_t		abi_mru_ghost_hits;
+	uint32_t		abi_mfu_hits;
+	uint32_t		abi_mfu_ghost_hits;
+	uint32_t		abi_l2arc_hits;
+	uint32_t		abi_holds;
+	uint64_t		abi_l2arc_dattr;
+	uint64_t		abi_l2arc_asize;
+	enum zio_compress	abi_l2arc_compress;
+} arc_buf_info_t;
+
 void arc_space_consume(uint64_t space, arc_space_type_t type);
 void arc_space_return(uint64_t space, arc_space_type_t type);
 arc_buf_t *arc_buf_alloc(spa_t *spa, int size, void *tag,
@@ -101,6 +131,7 @@ void arc_return_buf(arc_buf_t *buf, void *tag);
 void arc_loan_inuse_buf(arc_buf_t *buf, void *tag);
 void arc_buf_add_ref(arc_buf_t *buf, void *tag);
 boolean_t arc_buf_remove_ref(arc_buf_t *buf, void *tag);
+void arc_buf_info(arc_buf_t *buf, arc_buf_info_t *abi, int state_index);
 int arc_buf_size(arc_buf_t *buf);
 void arc_release(arc_buf_t *buf, void *tag);
 int arc_released(arc_buf_t *buf);
diff --git a/include/sys/dbuf.h b/include/sys/dbuf.h
index 8cd1fde01..85e967d0f 100644
--- a/include/sys/dbuf.h
+++ b/include/sys/dbuf.h
@@ -282,6 +282,9 @@ void dbuf_free_range(struct dnode *dn, uint64_t start, uint64_t end,
 
 void dbuf_new_size(dmu_buf_impl_t *db, int size, dmu_tx_t *tx);
 
+void dbuf_stats_init(dbuf_hash_table_t *hash);
+void dbuf_stats_destroy(void);
+
 #define	DB_DNODE(_db)		((_db)->db_dnode_handle->dnh_dnode)
 #define	DB_DNODE_LOCK(_db)	((_db)->db_dnode_handle->dnh_zrlock)
 #define	DB_DNODE_ENTER(_db)	(zrl_add(&DB_DNODE_LOCK(_db)))
diff --git a/include/sys/dmu.h b/include/sys/dmu.h
index b0db7604d..fd1460d5c 100644
--- a/include/sys/dmu.h
+++ b/include/sys/dmu.h
@@ -660,6 +660,7 @@ extern const dmu_object_byteswap_info_t dmu_ot_byteswap[DMU_BSWAP_NUMFUNCS];
  * If doi is NULL, just indicates whether the object exists.
  */
 int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi);
+void __dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi);
 void dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi);
 void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi);
 void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize,
diff --git a/include/sys/dsl_pool.h b/include/sys/dsl_pool.h
index 51b588e6a..25abd9c03 100644
--- a/include/sys/dsl_pool.h
+++ b/include/sys/dsl_pool.h
@@ -71,13 +71,6 @@ typedef struct zfs_all_blkstats {
 	zfs_blkstat_t	zab_type[DN_MAX_LEVELS + 1][DMU_OT_TOTAL + 1];
 } zfs_all_blkstats_t;
 
-typedef struct txg_history {
-	kstat_txg_t	th_kstat;
-	vdev_stat_t	th_vs1;
-	vdev_stat_t	th_vs2;
-	kmutex_t	th_lock;
-	list_node_t	th_link;
-} txg_history_t;
 
 typedef struct dsl_pool {
 	/* Immutable */
@@ -89,8 +82,6 @@ typedef struct dsl_pool {
 	struct dsl_dataset *dp_origin_snap;
 	uint64_t dp_root_dir_obj;
 	struct taskq *dp_iput_taskq;
-	kstat_t *dp_txg_kstat;
-	kstat_t *dp_tx_assign_kstat;
 
 	/* No lock needed - sync context only */
 	blkptr_t dp_meta_rootbp;
@@ -111,11 +102,6 @@ typedef struct dsl_pool {
 	uint64_t dp_mos_used_delta;
 	uint64_t dp_mos_compressed_delta;
 	uint64_t dp_mos_uncompressed_delta;
-	uint64_t dp_txg_history_size;
-	list_t dp_txg_history;
-	uint64_t dp_tx_assign_size;
-	kstat_named_t *dp_tx_assign_buckets;
-
 
 	/* Has its own locking */
 	tx_state_t dp_tx;
@@ -171,12 +157,6 @@ int dsl_pool_open_special_dir(dsl_pool_t *dp, const char *name, dsl_dir_t **);
 int dsl_pool_hold(const char *name, void *tag, dsl_pool_t **dp);
 void dsl_pool_rele(dsl_pool_t *dp, void *tag);
 
-void dsl_pool_tx_assign_add_usecs(dsl_pool_t *dp, uint64_t usecs);
-
-txg_history_t *dsl_pool_txg_history_add(dsl_pool_t *dp, uint64_t txg);
-txg_history_t *dsl_pool_txg_history_get(dsl_pool_t *dp, uint64_t txg);
-void dsl_pool_txg_history_put(txg_history_t *th);
-
 #ifdef	__cplusplus
 }
 #endif
diff --git a/include/sys/spa.h b/include/sys/spa.h
index 401ae8343..cb3ce11bc 100644
--- a/include/sys/spa.h
+++ b/include/sys/spa.h
@@ -51,6 +51,8 @@ typedef struct zilog zilog_t;
 typedef struct spa_aux_vdev spa_aux_vdev_t;
 typedef struct ddt ddt_t;
 typedef struct ddt_entry ddt_entry_t;
+typedef struct zbookmark zbookmark_t;
+
 struct dsl_pool;
 struct dsl_dataset;
 
@@ -534,6 +536,41 @@ extern boolean_t spa_refcount_zero(spa_t *spa);
 #define	SCL_ALL		((1 << SCL_LOCKS) - 1)
 #define	SCL_STATE_ALL	(SCL_STATE | SCL_L2ARC | SCL_ZIO)
 
+/* Historical pool statistics */
+typedef struct spa_stats_history {
+	kmutex_t		lock;
+	uint64_t		count;
+	uint64_t		size;
+	kstat_t			*kstat;
+	void			*private;
+	list_t			list;
+} spa_stats_history_t;
+
+typedef struct spa_stats {
+	spa_stats_history_t	read_history;
+	spa_stats_history_t	txg_history;
+	spa_stats_history_t	tx_assign_histogram;
+} spa_stats_t;
+
+typedef enum txg_state {
+	TXG_STATE_BIRTH		= 0,
+	TXG_STATE_OPEN		= 1,
+	TXG_STATE_QUIESCED	= 2,
+	TXG_STATE_SYNCED	= 3,
+	TXG_STATE_COMMITTED	= 4,
+} txg_state_t;
+
+extern void spa_stats_init(spa_t *spa);
+extern void spa_stats_destroy(spa_t *spa);
+extern void spa_read_history_add(spa_t *spa, const zbookmark_t *zb,
+    uint32_t aflags);
+extern void spa_txg_history_add(spa_t *spa, uint64_t txg);
+extern int spa_txg_history_set(spa_t *spa,  uint64_t txg,
+    txg_state_t completed_state, hrtime_t completed_time);
+extern int spa_txg_history_set_io(spa_t *spa,  uint64_t txg, uint64_t nread,
+    uint64_t nwritten, uint64_t reads, uint64_t writes, uint64_t nreserved);
+extern void spa_tx_assign_add_nsecs(spa_t *spa, uint64_t nsecs);
+
 /* Pool configuration locks */
 extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw);
 extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw);
diff --git a/include/sys/spa_impl.h b/include/sys/spa_impl.h
index 47dfe432e..107d14a76 100644
--- a/include/sys/spa_impl.h
+++ b/include/sys/spa_impl.h
@@ -236,6 +236,8 @@ struct spa {
 	uint64_t	spa_deadman_calls;	/* number of deadman calls */
 	uint64_t	spa_sync_starttime;	/* starting time fo spa_sync */
 	uint64_t	spa_deadman_synctime;	/* deadman expiration timer */
+	spa_stats_t	spa_stats;		/* assorted spa statistics */
+
 	/*
 	 * spa_refcnt & spa_config_lock must be the last elements
 	 * because refcount_t changes size based on compilation options.
diff --git a/include/sys/zfs_context.h b/include/sys/zfs_context.h
index a126c058e..dfd11d9f1 100644
--- a/include/sys/zfs_context.h
+++ b/include/sys/zfs_context.h
@@ -210,6 +210,7 @@ typedef struct kthread {
 } kthread_t;
 
 #define	curthread			zk_thread_current()
+#define	getcomm()			"unknown"
 #define	thread_exit			zk_thread_exit
 #define	thread_create(stk, stksize, func, arg, len, pp, state, pri)	\
 	zk_thread_create(stk, stksize, (thread_func_t)func, arg,	\
@@ -347,6 +348,10 @@ extern kstat_t *kstat_create(char *, int,
     char *, char *, uchar_t, ulong_t, uchar_t);
 extern void kstat_install(kstat_t *);
 extern void kstat_delete(kstat_t *);
+extern void kstat_set_raw_ops(kstat_t *ksp,
+    int (*headers)(char *buf, size_t size),
+    int (*data)(char *buf, size_t size, void *data),
+    void *(*addr)(kstat_t *ksp, loff_t index));
 
 /*
  * Kernel memory
diff --git a/include/sys/zio.h b/include/sys/zio.h
index 189966bef..f5a128e0b 100644
--- a/include/sys/zio.h
+++ b/include/sys/zio.h
@@ -256,12 +256,13 @@ extern char *zio_type_name[ZIO_TYPES];
  * Therefore it must not change size or alignment between 32/64 bit
  * compilation options.
  */
-typedef struct zbookmark {
+struct zbookmark {
 	uint64_t	zb_objset;
 	uint64_t	zb_object;
 	int64_t		zb_level;
 	uint64_t	zb_blkid;
-} zbookmark_t;
+	char *		zb_func;
+};
 
 #define	SET_BOOKMARK(zb, objset, object, level, blkid)  \
 {                                                       \
@@ -269,6 +270,7 @@ typedef struct zbookmark {
 	(zb)->zb_object = object;                       \
 	(zb)->zb_level = level;                         \
 	(zb)->zb_blkid = blkid;                         \
+	(zb)->zb_func = FTAG;                           \
 }
 
 #define	ZB_DESTROYED_OBJSET	(-1ULL)