Add visibility in to arc_read

This change is an attempt to add visibility into the arc_read calls
occurring on a system, in real time. To do this, a list was added to the
in memory SPA data structure for a pool, with each element on the list
corresponding to a call to arc_read. These entries are then exported
through the kstat interface, which can then be interpreted in userspace.

For each arc_read call, the following information is exported:

 * A unique identifier (uint64_t)
 * The time the entry was added to the list (hrtime_t)
   (*not* wall clock time; relative to the other entries on the list)
 * The objset ID (uint64_t)
 * The object number (uint64_t)
 * The indirection level (uint64_t)
 * The block ID (uint64_t)
 * The name of the function originating the arc_read call (char[24])
 * The arc_flags from the arc_read call (uint32_t)
 * The PID of the reading thread (pid_t)
 * The command or name of thread originating read (char[16])

From this exported information one can see, in real time, exactly what
is being read, what function is generating the read, and whether or not
the read was found to be already cached.

There is still some work to be done, but this should serve as a good
starting point.

Specifically, dbuf_read's are not accounted for in the currently
exported information. Thus, a follow up patch should probably be added
to export these calls that never call into arc_read (they only hit the
dbuf hash table). In addition, it might be nice to create a utility
similar to "arcstat.py" to digest the exported information and display
it in a more readable format. Or perhaps, log the information and allow
for it to be "replayed" at a later time.

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

4 files changed, 32 insertions, 2 deletions

diff --git a/include/sys/spa.h b/include/sys/spa.h
index 401ae8343..a4a717ac0 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 +536,25 @@ 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_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);
+
 /* 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)