kext.c   [plain text]

/*
 * Boot-time performance cache.
 *
 * We implement a shim layer for a block device which can be advised ahead
 * of time of the likely read pattern for the block device.
 *
 * Armed with a sorted version of this read pattern (the "playlist"), we
 * cluster the reads into a private cache, taking advantage of the
 * sequential I/O pattern. Incoming read requests are then satisfied from
 * the cache (if possible), or by forwarding to the original strategy routine.
 *
 * We also record the pattern of incoming read requests (the "history
 * list"), in order to provide for adaptation to changes in behaviour by
 * the calling layer.
 *
 * The hit rate of the cache is also measured, and caching will be disabled
 * preemptively if the hit rate is too low.
 *
 * The cache is controlled by a sysctl interface.  Typical usage is as
 * follows:
 *
 * - Load a sorted read pattern.
 * - Wait for the operation(s) of interest to be performed.
 * - Fetch the new (unsorted) read pattern.
 *
 * Loading a read pattern initialises the cache; upon successful loading
 * the cache is active and readhead is being performed.  After fetching the
 * recorded read pattern, the cache remains around until jettisoned due
 * to memory pressure or the hit rate falls below a desirable level.
 *
 * Limitations:
 *
 * - only one copy of the cache may be active at a time
 */

/*
 * Build options. 
 */
/*#define BC_DEBUG*/
/*#define BC_EXTRA_DEBUG*/

/*
 * Emulate readahead, don't actually perform it.
 */
/*#define EMULATE_ONLY*/

/*
 * Don't collect request history.
 */
/*#define NO_HISTORY*/

/*
 * Don't use history cluster chains (seems to be a malloc problem with these
 * enabled)
 */
/*#define STATIC_HISTORY*/

/*
 * Only cache for the volumes on the root disk
 */
#define ROOT_DISK_ONLY

/*
 * Don't cache SSDs
 */
#define DONT_CACHE_SSDS

/*
 * Keep a log of read history to aid in debugging.
 */
/*#define READ_HISTORY_BUFFER	1000*/
/*#define READ_HISTORY_ALL*/

/*
 * Ignore the batches on playlist entries.
 */
/*#define IGNORE_BATCH */

/*
 * Tunable parameters.
 */

/*
 * Minimum required physical memory before BootCache will
 * activate.  Insufficient physical memory leads to paging
 * activity during boot, which is not repeatable and can't
 * be distinguished from the repeatable read patterns.
 *
 * Size in megabytes.
 */
static int BC_minimum_mem_size = 256;

/*
 * Number of seconds to wait in BC_strategy for readahead to catch up
 * with our request.   Tuninig issues:
 *  - Waiting too long may hold up the system behind this I/O/.
 *  - Bypassing an I/O too early will reduce the hit rate and adversely
 *    affect performance on slow media.
 */
#define BC_READAHEAD_WAIT_DEFAULT	10
#define BC_READAHEAD_WAIT_CDROM		45
static int BC_wait_for_readahead = BC_READAHEAD_WAIT_DEFAULT;

/*
 * Cache hit ratio monitoring.
 *
 * The time overhead of a miss in the cache is approx 1/5000th of the
 * time it takes to satisfy an IO from a HDD. If our hit rate falls
 * below this, we're no longer providing a performance win and the
 * cache will be jettisoned.
 *
 * Make sure to keep the cache around for at least an hour
 * if the user hasn't logged in yet, though.
 */
static int BC_chit_max_num_IOs_since_last_hit = 10000;
static int BC_chit_prelogin_timeout_seconds = 3600;

/*
 * History recording timeout.
 *
 * If the history recording is not turned off, the history buffer
 * will grow to its maximum size. This timeout will kill the history
 * recording after the specified interval (in hz).
 *
 * User DVDs don't issue a "BootCacheControl stop", so they depend
 * on the history recording timeout.
 */
static int BC_history_timeout = 240 * 100;

/*
 * Trace macros
 */
#ifndef DBG_BOOTCACHE
#define DBG_BOOTCACHE	7
#endif

#define	DBG_BC_TAG	1
#define	DBG_BC_BATCH	2
static int dbg_tag_count = 0;

#ifdef BC_DEBUG
# define MACH_DEBUG
# define MACH_ASSERT 1
# define message(fmt, args...)	\
do { \
	microtime(&debug_currenttime); \
	timersub(&debug_currenttime, &debug_starttime, &debug_currenttime); \
	lck_mtx_lock(&debug_printlock); \
	printf("BootCache: %5d.%03d %24s[%4d]: " fmt "\n", (u_int)(debug_currenttime.tv_sec), (u_int)(debug_currenttime.tv_usec / 1000), __FUNCTION__, __LINE__, ##args); \
	lck_mtx_unlock(&debug_printlock); \
} while (0)

# define debug(fmt, args...)	message("*** " fmt, ##args)
extern void Debugger(char *);
#else
# define debug(fmt, args...)	do {} while (0)
# define message(fmt, args...)	printf("BootCache: " fmt "\n" , ##args)
#endif

#ifdef BC_EXTRA_DEBUG
# define xdebug(fmt, args...)	message("+++ " fmt, ##args)
#else
# define xdebug(fmt, args...)	do {} while (0)
#endif

#include <sys/buf.h>
#include <sys/conf.h>
#include <sys/fcntl.h>
#include <sys/kdebug.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/param.h>
#include <sys/proc.h>
#include <sys/types.h>
#include <sys/sysctl.h>
#include <sys/systm.h>
#include <sys/vnode.h>
#include <sys/mount.h>
#include <sys/ubc.h>
#include <sys/uio.h>

#include <uuid/uuid.h>

#include <mach/kmod.h>
#include <mach/mach_types.h>
#include <mach/vm_param.h>	/* for max_mem */

#include <vm/vm_kern.h>

#include <libkern/libkern.h>
#include <libkern/OSAtomic.h>

#include <kern/assert.h>
#include <kern/host.h>
#include <kern/kalloc.h>
#include <kern/thread_call.h>
#include <kern/locks.h>

#include <IOKit/storage/IOMediaBSDClient.h>

#include <pexpert/pexpert.h>

#include "BootCache.h"

#ifdef BC_DEBUG
static struct timeval debug_starttime;
static struct timeval debug_currenttime;
static lck_mtx_t debug_printlock;
#endif

#define BC_MALLOC(_size, _type, _flags) ((ssize_t)(_size) < 0) ? NULL : _MALLOC((_size), (_type), (_flags))

/**************************************
 * Structures for the readahead cache *
 **************************************/

/*
 * A cache extent.
 *
 * Each extent represents a contiguous range of disk blocks which are
 * held in the cache.  All values in bytes.
 */
struct BC_cache_extent {
	lck_mtx_t   ce_lock;
	u_int64_t   ce_diskoffset;  /* physical offset on device */
	u_int64_t   ce_length;      /* data length */
	off_t       ce_cacheoffset;	/* offset into cache backing store */
	u_int16_t   ce_batch;       /* batch number */
	u_int16_t   ce_flags;       /* flags */
#define CE_ABORTED  (1 << 0)    /* extent no longer has data */
#define CE_IODONE   (1 << 1)    /* extent has been read */
#define CE_LOWPRI   (1 << 2)    /* extent is low priority */
#define CE_SHARED   (1 << 3)    /* extent is in the shared cache */
	int         ce_mount_idx;   /* the index of the mount for this extent */
	u_int32_t  *ce_blockmap;    /* track valid blocks for this extent */
};

/* A mount 
 *
 * The volume to which an extent refers
 */
struct BC_cache_mount {
	lck_rw_t                 cm_lock;      /* lock guards instance variables, extents have their own locks */
	uuid_t                   cm_uuid;      /* this mount's uuid */
	int                      cm_nextents;  /* the number of extents that reference this mount */
	struct BC_cache_extent **cm_pextents;  /* a list of pointers to extents which reference
											this mount, sorted by block address */
	int                      cm_state;     /* this mount's state */
#define CM_SETUP	(0)                    /* fields above are valid. Not mounted yet */
#define CM_READY	(1)                    /* every field is valid. Mounted and ready to go */
#define CM_ABORTED	(2)                    /* mount failed for some reason */
	
	/* fields below filled in when we detect that the volume has mounted */
	dev_t                    cm_dev;       /* the device for this mount */
	struct buf              *cm_bp;        /* struct buf allocated in the reader thread for this device */
	u_int64_t                cm_throttle_mask;/* the throttle mask to use in throttle APIs for this mount */
	u_int64_t                cm_blocksize; /* keep 64-bit for promotion */
	u_int64_t                cm_devsize;   /* range checking */
	u_int64_t                cm_maxread;   /* largest read the dev will support */
	struct BC_cache_disk*    cm_disk;      /* the mount's physical disk */
};

/* A physical disk
 *
 * The disk may contain several mounts
 * and will have one readahead thread
 */
struct BC_cache_disk {
	lck_mtx_t cd_lock;
	u_int64_t cd_disk_id;    /* The throttle mask, used as an identifier */
	int       cd_disk_num; /* used for stat batching */
	int       cd_flags;      /* flags */
#define CD_HAS_THREAD   (1 << 0)  /* there is a readahead thread in progress for this disk */
#define CD_ISSUE_LOWPRI (1 << 1)  /* the readahead thread is reading in low-priority extents */
#define CD_IS_SSD       (1 << 2)  /* this is a solid state disk */
	int       cd_nmounts;    /* The number of mounts that reference this disk */
	int       cd_batch;      /* What playback batch this disk is on */
};

/************************************
 * Structures for history recording *
 ************************************/

/* see BC_history_entry and BC_history_mount in BootCache.h */

/*
 * Wrapper around BC_history_mount so we
 * can keep track of the device
 */
struct BC_history_mount_device {
	u_int64_t                       hmd_disk_id;
	struct BC_history_mount_device* hmd_next;
	int                             hmd_is_ssd;
	dev_t                           hmd_dev;
	int                             hmd_blocksize;
	struct BC_history_mount         hmd_mount;
};

/*
 * A history entry list cluster.
 */
struct BC_history_entry_cluster {
	struct BC_history_entry_cluster *hec_link;
	int                              hec_nentries;	/* number of entries in list */
	struct BC_history_entry          hec_data[0];
};

#ifdef STATIC_HISTORY
# define BC_HISTORY_ALLOC	128000
# define BC_HISTORY_ENTRIES						\
(((BC_HISTORY_ALLOC - sizeof(struct BC_history_entry_cluster)) /	\
sizeof(struct BC_history_entry)) - 1)
# define BC_HISTORY_MAXCLUSTERS	1
#else
# define BC_HISTORY_ALLOC	16000
# define BC_HISTORY_ENTRIES						\
(((BC_HISTORY_ALLOC - sizeof(struct BC_history_entry_cluster)) /	\
sizeof(struct BC_history_entry)) - 1)
# define BC_HISTORY_MAXCLUSTERS	((BC_MAXENTRIES / BC_HISTORY_ENTRIES) + 1)
#endif


/*****************************************
 * To be able to debug live read history *
 *****************************************/
#ifdef READ_HISTORY_BUFFER
struct BC_read_history_entry {
	struct BC_cache_extent	*rh_extent;	/* extent this read is for */
	u_int64_t		rh_blkno;	/* offset of actual read */
	u_int64_t		rh_bcount;	/* length in bytes */
	int			rh_result;
# define BC_RHISTORY_OK		0
# define BC_RHISTORY_FAIL	1
};
#endif


/*
 * The cache control structure.
 */
struct BC_cache_control {
	lck_grp_t  *c_lckgrp;
	
	/* the device we are covering */
	struct bdevsw	*c_devsw;
	
	/* saved switch table routines */
	strategy_fcn_t   *c_strategy;
	open_close_fcn_t *c_close;
	lck_mtx_t         c_handlers_lock;
	
	u_int64_t	c_cachesize;		/* total number of bytes in cache */
	
	u_int64_t   c_root_disk_id;     /* the throttle mask of the root disk, used as an id for the physical device */
	                                /* This needs to be updated to handle multiple physical volumes backing a mount */
	
	/*
	 * The mounts we're tracking
	 */
	int                     c_nmounts; /* number of mounts in the array below */
	struct BC_cache_mount  *c_mounts;  /* the array of mounts the cache extents refer to */
	
	/*
	 * Extents, in optimal read order.
	 *
	 * Each extent tracks the pages it owns in the buffer.
	 * It is assumed that PAGE_SIZE is a multiple of cm_blocksize for each mount.
	 *
	 * Each extent refers to one of the mounts, above, by index into c_mounts
	 */
	int                      c_nextentlists; /* number of extent lists we have */
	int                     *c_nextents;     /* number of extents in a given extent list */
	struct BC_cache_extent **c_extentlists;      /* an array of extent lists in the cache */
	
	int         c_batch_count;		/* number of extent batches */
	vnode_t     c_vp;
	uint32_t    c_vid;
	/* rw lock protects the above fields through c_nmounts, but not their contents.
	 * Shared for general use of the fields,
	 * Exclusive for cache initialization or cleanup.
	 * Each mount/extent has its own lock for modifying its contents.
	 */
	lck_rw_t    c_cache_lock;
	
	
	/* history list, in reverse order */
	int                              c_history_num_clusters;
	struct BC_history_entry_cluster *c_history;
	struct BC_history_mount_device  *c_history_mounts;
	uint64_t                         c_history_size;
	/* lock protects the above history fields and their contents */
	lck_rw_t   c_history_lock;
	
	/* fields below are accessed atomically */
	
	/* flags */
	UInt32		c_flags;
#define	BC_FLAG_SETUP			(1 << 0)	/* cache setup properly during mount */
#define	BC_FLAG_CACHEACTIVE		(1 << 1)	/* cache is active, owns memory */
#define	BC_FLAG_HISTORYACTIVE	(1 << 2)	/* currently recording history */
#define	BC_FLAG_HTRUNCATED		(1 << 3)	/* history list truncated */
#define	BC_FLAG_SHUTDOWN		(1 << 4)	/* readahead shut down */
	SInt32		c_strategycalls;	/* count of busy strategy calls in the cache */
	
	uint32_t  c_readahead_throttles_cutthroughs;
	
	uint32_t  c_num_ios_since_last_hit; /* The number of cache misses since the last hit */
	
	uint32_t  c_num_reader_threads;     /* The number of reader threads active */
	lck_mtx_t c_reader_threads_lock;    /* protects c_num_reader_threads */
	
	/* statistics */
	int c_take_detailed_stats;
	struct BC_statistics c_stats;
	
	/* time-keeping */
	struct timeval c_loadtime;
	struct timeval c_cache_starttime;
	struct timeval c_history_starttime;
	
#ifdef READ_HISTORY_BUFFER
	int		c_rhistory_idx;
	struct BC_read_history_entry *c_rhistory;
#endif
};

/*
 * The number of blocks per page; assumes that a page
 * will always be >= the size of a disk block.
 */
#define CB_PAGEBLOCKS(cm)			(PAGE_SIZE / (cm)->cm_blocksize)

/*
 * Convert block offset to page offset and vice versa.
 */
#define CB_BLOCK_TO_PAGE(cm, block)		((block) / CB_PAGEBLOCKS(cm))
#define CB_PAGE_TO_BLOCK(cm, page)		((page)  * CB_PAGEBLOCKS(cm))
#define CB_BLOCK_TO_BYTE(cm, block)		((block) * (cm)->cm_blocksize)
#define HB_BLOCK_TO_BYTE(hmd, block)	((block) * (hmd)->hmd_blocksize)
#define CB_BYTE_TO_BLOCK(cm, byte)		((byte)  / (cm)->cm_blocksize)

/*
 * The size of an addressable field in the block map.
 * This reflects the u_int32_t type for the blockmap type.
 */
#define CB_MAPFIELDBITS			32
#define CB_MAPFIELDBYTES		4

/*
 * The index into an array of addressable fields, and the bit within
 * the addressable field corresponding to an index in the map.
 */
#define CB_MAPIDX(x)			((x) / CB_MAPFIELDBITS)
#define CB_MAPBIT(x)			(1 << ((x) % CB_MAPFIELDBITS))

/*
 * Blockmap management. Must be called with extent lock held.
 */
#define CB_BLOCK_PRESENT(ce, block) \
((ce)->ce_blockmap != NULL && ((ce)->ce_blockmap[CB_MAPIDX(block)] &   CB_MAPBIT(block)))
#define CB_MARK_BLOCK_PRESENT(ce, block) \
((ce)->ce_blockmap[CB_MAPIDX(block)] |=  CB_MAPBIT(block))
#define CB_MARK_BLOCK_VACANT(ce, block) \
((ce)->ce_blockmap[CB_MAPIDX(block)] &= ~CB_MAPBIT(block))

/*
 * Determine whether a given page is vacant (all blocks are gone).
 * This takes advantage of the expectation that a page's blocks
 * will never cross the boundary of a field in the blockmap.
 */
#define CB_PAGE_VACANT(cm, ce, page)					\
(!((ce)->ce_blockmap[CB_MAPIDX(CB_PAGE_TO_BLOCK(cm, page))] &	\
(((1 << CB_PAGEBLOCKS(cm)) - 1) << (CB_PAGE_TO_BLOCK(cm, page) %	\
CB_MAPFIELDBITS))))

/* Maximum size of the boot cache */
#define BC_MAX_SIZE (max_mem / 2)

/*
 * Sanity macro, frees and zeroes a pointer if it is not already zeroed.
 */
#define _FREE_ZERO(p, c)						\
do {								\
if (p != NULL) {					\
_FREE(p, c);					\
p = NULL;					\
}							\
} while (0);

/*
 * Synchronization macros
 */
#define LOCK_HISTORY_R()		lck_rw_lock_shared(&BC_cache->c_history_lock)
#define UNLOCK_HISTORY_R()		lck_rw_unlock_shared(&BC_cache->c_history_lock)
#define LOCK_HISTORY_W()		lck_rw_lock_exclusive(&BC_cache->c_history_lock)
#define UNLOCK_HISTORY_W()		lck_rw_unlock_exclusive(&BC_cache->c_history_lock)

#define LOCK_EXTENT(e)			lck_mtx_lock(&(e)->ce_lock)
#define UNLOCK_EXTENT(e)		lck_mtx_unlock(&(e)->ce_lock)

/* mount locks should only be held while also holding the cache lock */
#define LOCK_MOUNT_R(m)			lck_rw_lock_shared(&(m)->cm_lock)
#define UNLOCK_MOUNT_R(m)		lck_rw_unlock_shared(&(m)->cm_lock)
#define LOCK_MOUNT_W(m)			lck_rw_lock_exclusive(&(m)->cm_lock)
#define UNLOCK_MOUNT_W(m)		lck_rw_unlock_exclusive(&(m)->cm_lock)
#define LOCK_MOUNT_W_TO_R(m)	lck_rw_lock_exclusive_to_shared(&(m)->cm_lock)
#define LOCK_MOUNT_R_TO_W(m)	lck_rw_lock_shared_to_exclusive(&(m)->cm_lock)

#define LOCK_DISK(d)			lck_mtx_lock(&(d)->cd_lock)
#define UNLOCK_DISK(d)			lck_mtx_unlock(&(d)->cd_lock)

#define LOCK_READERS()			lck_mtx_lock(&BC_cache->c_reader_threads_lock)
#define UNLOCK_READERS()		lck_mtx_unlock(&BC_cache->c_reader_threads_lock)

#define LOCK_HANDLERS()			lck_mtx_lock(&BC_cache->c_handlers_lock)
#define UNLOCK_HANDLERS()		lck_mtx_unlock(&BC_cache->c_handlers_lock)

#define LOCK_CACHE_R()			lck_rw_lock_shared(&BC_cache->c_cache_lock)
#define UNLOCK_CACHE_R()		lck_rw_unlock_shared(&BC_cache->c_cache_lock)
#define LOCK_CACHE_W()			lck_rw_lock_exclusive(&BC_cache->c_cache_lock)
#define UNLOCK_CACHE_W()		lck_rw_unlock_exclusive(&BC_cache->c_cache_lock)
#define LOCK_CACHE_TRY_W()		lck_rw_try_lock(&BC_cache->c_cache_lock, LCK_RW_TYPE_EXCLUSIVE)
#define LOCK_CACHE_TRY_R()		lck_rw_try_lock(&BC_cache->c_cache_lock, LCK_RW_TYPE_SHARED)
#define LOCK_CACHE_W_TO_R()		lck_rw_lock_exclusive_to_shared(&BC_cache->c_cache_lock)
#define LOCK_CACHE_R_TO_W()		lck_rw_lock_shared_to_exclusive(&BC_cache->c_cache_lock)

#define BC_ADD_STAT(stat, inc)	OSAddAtomic((inc), ((SInt32 *)&BC_cache->c_stats.ss_##stat))

/*
 * Only one instance of the cache is supported.
 */
static struct BC_cache_control BC_cache_softc;
static struct BC_cache_control *BC_cache = &BC_cache_softc;

/*
 * We support preloaded cache data by checking for a Mach-O
 * segment/section which can be populated with a playlist by the
 * linker.  This is particularly useful in the CD mastering process,
 * as reading the playlist from CD is very slow and rebuilding the
 * kext at mastering time is not practical.
 */
extern kmod_info_t kmod_info;
static struct BC_playlist_header *BC_preloaded_playlist;
static int	BC_preloaded_playlist_size;

static int	BC_check_intersection(struct BC_cache_extent *ce, u_int64_t offset, u_int64_t length);
static int	BC_find_cache_mount(dev_t dev);
static struct BC_cache_extent**	BC_find_extent(struct BC_cache_mount* cm, u_int64_t offset, u_int64_t length, int contained, int* pnum_extents);
static int	BC_discard_bytes(struct BC_cache_extent *ce, u_int64_t offset, u_int64_t length);
static int	BC_handle_discards(struct BC_cache_mount *cm, u_int64_t offset, u_int64_t length);
static int	BC_blocks_present(struct BC_cache_extent *ce, int base, int nblk);
static void	BC_reader_thread(void *param0, wait_result_t param1);
static int	BC_strategy(struct buf *bp);
static int	BC_close(dev_t dev, int flags, int devtype, struct proc *p);
static int	BC_terminate_readahead(void);
static int	BC_terminate_cache(void);
static int	BC_terminate_history(void);
static void	BC_check_handlers(void);
static void	BC_next_valid_range(struct BC_cache_mount *cm, struct BC_cache_extent *ce, uint32_t from, 
								uint32_t *nextpage, uint32_t *nextoffset, uint32_t *nextlength);
static int	BC_setup_disk(struct BC_cache_disk *cd, u_int64_t disk_id, int is_ssd);
static void	BC_teardown_disk(struct BC_cache_disk *cd);
static void	BC_mount_available(struct BC_cache_mount *cm);
static int	BC_setup_mount(struct BC_cache_mount *cm, struct BC_playlist_mount* pm);
static int	BC_fill_in_mount(struct BC_cache_mount *cm, mount_t mount, vfs_context_t context);
static void	BC_teardown_mount(struct BC_cache_mount *ce);
static int	BC_setup_extent(struct BC_cache_extent *ce, const struct BC_playlist_entry *pe, int batch_adjustment, int cache_mount_idx);
static void	BC_teardown_extent(struct BC_cache_extent *ce);
static int	BC_copyin_playlist(size_t mounts_size, user_addr_t mounts, size_t entries_size, user_addr_t entries);
static int	BC_init_cache(size_t mounts_size, user_addr_t mounts, size_t entries_size, user_addr_t entries, int record_history);
static void BC_update_mounts(void);
static void	BC_timeout_history(void *);
static struct BC_history_mount_device * BC_get_history_mount_device(dev_t dev, int* pindex);
static int	BC_add_history(daddr64_t blkno, u_int64_t length, int pid, int hit, int write, int tag, int shared, dev_t dev);
static int	BC_size_history_mounts(void);
static int	BC_size_history_entries(void);
static int	BC_copyout_history_mounts(user_addr_t uptr);
static int	BC_copyout_history_entries(user_addr_t uptr);
static void	BC_discard_history(void);
static void	BC_auto_start(void);
static int	BC_setup(void);

static int	fill_in_bc_cache_mounts(mount_t mount, void* arg);
static int	check_for_new_mount_itr(mount_t mount, void* arg);
static int	extents_status(struct BC_cache_extent **pce, int nextents) ;
static void	wait_for_extent(struct BC_cache_extent *ce, struct timeval starttime) ;
	
#ifndef BOOTCACHE_ENTRIES_SORTED_BY_DISK_OFFSET
static int  compare_cache_extents(const void *vfirst, const void *vsecond);
//FIXME: qsort not available in kexts
extern void qsort(void *a, size_t n, size_t es, int (*cmp)(const void *, const void *));
#endif

/*
 * Sysctl interface.
 */
static int	BC_sysctl SYSCTL_HANDLER_ARGS;

SYSCTL_PROC(_kern, OID_AUTO, BootCache, CTLFLAG_RW | CTLFLAG_LOCKED,
			0, 0, BC_sysctl,
			"S,BC_command",
			"BootCache command interface");

/*
 * Netboot exports this hook so that we can determine
 * whether we were netbooted.
 */
extern int	netboot_root(void);

/*
 * bsd_init exports this hook so that we can register for notification
 * once the root filesystem is mounted.
 */
extern void (* mountroot_post_hook)(void);
extern void (* unmountroot_pre_hook)(void);

/*
 * Mach VM-specific functions.
 */
static int	BC_alloc_pagebuffer();
static void	BC_free_pagebuffer();
static void	BC_free_page(struct BC_cache_extent *ce, int page);

/* Functions not prototyped elsewhere */
int     	ubc_range_op(vnode_t, off_t, off_t, int, int *); /* in sys/ubc_internal.h */

/*
 * Mach-o functions.
 */
#define MACH_KERNEL 1
#include <mach-o/loader.h>
extern void *getsectdatafromheader(struct mach_header *, char *, char *, int *);

/*
 * Set or clear a flag atomically and return the previous value of that bit.
 */
static inline UInt32
BC_set_flag(UInt32 flag)
{
	return flag & OSBitOrAtomic(flag, &BC_cache->c_flags);
}

static inline UInt32
BC_clear_flag(UInt32 flag)
{
	return flag & OSBitAndAtomic(~(flag), &BC_cache->c_flags);
}

/*
 * Return a user-readable string for a given uuid
 *
 * Returns a pointer to a static buffer, which is
 * racy, so this should only be used for debugging purposes
 */
static inline const char* uuid_string(uuid_t uuid)
{
	/* Racy, but used for debug output so who cares */
	static uuid_string_t uuidString;
	uuid_unparse(uuid, uuidString);
	return (char*)uuidString;
}

/*
 * Test whether a given byte range on disk intersects with an extent's range.
 * This function assumes the ranges are within the same mount.
 */
static inline int
BC_check_intersection(struct BC_cache_extent *ce, u_int64_t offset, u_int64_t length)
{
	if ((offset < (ce->ce_diskoffset + ce->ce_length)) &&
		((offset + length) > ce->ce_diskoffset))
		return 1;
	return 0;
}

/*
 * Find one of our cache mounts with the given device, if it has been seen this boot
 *
 * Called with the cache read lock held
 *
 * Returns the index of the mount, or -1 if it is not found
 */
static int
BC_find_cache_mount(dev_t dev)
{
	int i;
	
	if (dev == nulldev())
		return(-1);
	
	if (!(BC_cache->c_flags & BC_FLAG_CACHEACTIVE) || (BC_cache->c_mounts == NULL))
		return(-1);
	
	
	for (i = 0; i < BC_cache->c_nmounts; i++) {
		if (BC_cache->c_mounts[i].cm_state == CM_READY) {
			if (BC_cache->c_mounts[i].cm_dev == dev) {
				return i;
			}
		}
	}
	
	return(-1);
}

/*
 * Find a cache extent containing or intersecting the offset/length.
 *
 * We need to be able to find both containing and intersecting
 * extents.  Rather than handle each seperately, we take advantage
 * of the fact that any containing extent will also be an intersecting
 * extent.
 *
 * Thus, a search for a containing extent should be a search for an
 * intersecting extent followed by a test for containment.
 *
 * Containment may include multiple extents. The returned extent will
 * be the first of these extents (lowest offset), but the next extents
 * in the array may also be required to contain the requested range.
 * Upon success, *pnum_extents contains the number of extents required
 * to contain the range.
 *
 * Returns a pointer to a matching the entry in the mount's extent list
 * or NULL if no match is found.
 *
 * Called with the cache mount read lock held
 */
static struct BC_cache_extent **
BC_find_extent(struct BC_cache_mount *cm, u_int64_t offset, u_int64_t length, int contained, int *pnum_extents)
{
	struct BC_cache_extent **p, **base, **last, **next, **prev; /* pointers into the mount's extent index list */
	size_t lim;
	
	/* invariants */
	assert(length > 0);
	
	base = cm->cm_pextents;
	last = base + cm->cm_nextents - 1;
	
	/*
	 * If the cache is inactive, exit.
	 */
	if (!(BC_cache->c_flags & BC_FLAG_CACHEACTIVE) || (base == NULL)) {
		return(NULL);
	}
	
	/*
	 * Range-check against the cache first.
	 *
	 * Note that we check for possible intersection, rather than
	 * containment.
	 */
	if (((offset + length) <= (*base)->ce_diskoffset) ||
	    (offset >= ((*last)->ce_diskoffset + (*last)->ce_length))) {
		return(NULL);
	}
	
	/*
	 * Perform a binary search for an extent intersecting
	 * the offset and length.
	 *
	 * This code is based on bsearch(3), and the description following
	 * is taken directly from it.
	 *
	 * The code below is a bit sneaky.  After a comparison fails, we
	 * divide the work in half by moving either left or right. If lim
	 * is odd, moving left simply involves halving lim: e.g., when lim
	 * is 5 we look at item 2, so we change lim to 2 so that we will
	 * look at items 0 & 1.  If lim is even, the same applies.  If lim
	 * is odd, moving right again involes halving lim, this time moving
	 * the base up one item past p: e.g., when lim is 5 we change base
	 * to item 3 and make lim 2 so that we will look at items 3 and 4.
	 * If lim is even, however, we have to shrink it by one before
	 * halving: e.g., when lim is 4, we still looked at item 2, so we
	 * have to make lim 3, then halve, obtaining 1, so that we will only
	 * look at item 3.
	 */
	for (lim = cm->cm_nextents; lim != 0; lim >>= 1) {
		p = base + (lim >> 1);
		
		if (BC_check_intersection((*p), offset, length)) {
			
			if (contained == 0)
				return(p);
			
			if (pnum_extents) {
				*pnum_extents = 1;
			}
			
			/*
			 * We are checking for containment, verify that here.
			 * Extents may be adjacent, so include neighboring
			 * extents in the containment check
			 */
			
			/* Check the high bound */
			if ((offset + length) > ((*p)->ce_diskoffset + (*p)->ce_length)) {
				
				/* check higher extents to see if they contain the high end of this range */
				prev = p;
				for (next = prev + 1;
					 next <= last;
					 next++) {
					
					/* extents should never overlap */
					assert(((*prev)->ce_diskoffset + (*prev)->ce_length) <= (*next)->ce_diskoffset);
					
					if (((*prev)->ce_diskoffset + (*prev)->ce_length) < (*next)->ce_diskoffset) {
						/* extents are not adjacent */
						if (BC_cache->c_take_detailed_stats) {
							xdebug("Read 0x%llx:%lld on mount %s intersected, but not contained by %d-extent cache range 0x%llx,%lld (missed last %lld bytes)", offset, length, uuid_string(cm->cm_uuid), *pnum_extents, (*p)->ce_diskoffset, ((*prev)->ce_diskoffset + (*prev)->ce_length), (offset + length) - ((*prev)->ce_diskoffset + (*prev)->ce_length));
						}
						return(NULL);
					}
					
					if (pnum_extents) {
						(*pnum_extents)++;
					}					
					
					if ((offset + length) <= ((*next)->ce_diskoffset + (*next)->ce_length)) {
						/* we contain the high end of the range, break so we check the low end below */
						break;
					}
					
					prev = next;
				}
				
				if (next > last) {
					/* ran off the end of the extent list */
					if (BC_cache->c_take_detailed_stats) {
						xdebug("Read 0x%llx:%lld on mount %s intersected, but not contained by %d-extent cache range 0x%llx,%lld (missed last %lld bytes)", offset, length, uuid_string(cm->cm_uuid), *pnum_extents, (*p)->ce_diskoffset, ((*prev)->ce_diskoffset + (*prev)->ce_length), (offset + length) - ((*prev)->ce_diskoffset + (*prev)->ce_length));
					}
					return (NULL);
				}
			}
			
			/* Check the low bound */
			if (offset < (*p)->ce_diskoffset) {
				
				/* check lower extents to see if they contain the low end of this range */
				next = p;
				for (prev = next - 1;
					 prev >= base;
					 prev--) {
					
					/* extents should never overlap */
					assert(((*prev)->ce_diskoffset + (*prev)->ce_length) <= (*next)->ce_diskoffset);
					
					if (((*prev)->ce_diskoffset + (*prev)->ce_length) < (*next)->ce_diskoffset) {
						/* extents are not adjacent */
						if (BC_cache->c_take_detailed_stats) {
							xdebug("Read 0x%llx:%lld on mount %s intersected, but not contained by %d-extent cache range 0x%llx:%lld (missed first %lld bytes)", offset, length, uuid_string(cm->cm_uuid), *pnum_extents, (*next)->ce_diskoffset, ((*p)->ce_diskoffset + (*p)->ce_length), (*next)->ce_diskoffset - offset);
						}
						return(NULL);
					}
					
					if (pnum_extents) {
						(*pnum_extents)++;
					}										
					
					if (offset >= (*prev)->ce_diskoffset) {
						/* we contain the low end of the range (we checked high end above) */
						return(prev);
					}
					
					next = prev;
				}
				
				assert(prev < base); /* only break condition */
				
				if (prev < base) {
					/* ran off the end of the extent list */
					if (BC_cache->c_take_detailed_stats) {
						xdebug("Read 0x%llx:%lld on mount %s intersected, but not contained by %d-extent cache range 0x%llx:%lld (missed first %lld bytes)", offset, length, uuid_string(cm->cm_uuid), *pnum_extents, (*next)->ce_diskoffset, ((*p)->ce_diskoffset + (*p)->ce_length), (*next)->ce_diskoffset - offset);
					}
					return (NULL);
				}
			}
			
			return(p);
		}
		
		if (offset > (*p)->ce_diskoffset) {	/* move right */
			base = p + 1;
			lim--;
		}				/* else move left */
	}
	/* found nothing */
	return(NULL);
}

/*
 * Discard data from the cache and free pages which are no longer required.
 *
 * This should be locked against anyone testing for !vacant pages.
 *
 * We return the number of bytes we marked unused. If the value is
 * zero, the offset/length did not overap this extent.
 *
 * Called with extent lock held.
 */
static int
BC_discard_bytes(struct BC_cache_extent *ce, u_int64_t offset, u_int64_t length)
{
	struct BC_cache_mount *cm;
	u_int64_t estart, eend, dstart, dend, nblk, base, page;
	int i, discarded;
	
	/* invariants */
	assert(length > 0);
#ifdef BC_DEBUG
	lck_mtx_assert(&ce->ce_lock, LCK_MTX_ASSERT_OWNED);
#endif
	
	/*
	 * Extent has been terminated: blockmap may no longer be valid.
	 */
	if (ce->ce_flags & CE_ABORTED ||
		ce->ce_length == 0)
		return 0;
	
	cm = BC_cache->c_mounts + ce->ce_mount_idx;
	
	/*
	 * Constrain offset and length to fit this extent, return immediately if
	 * there is no overlap.
	 */
	dstart = offset;			/* convert to start/end format */
	dend = offset + length;
	assert(dend > dstart);
	estart = ce->ce_diskoffset;
	eend = ce->ce_diskoffset + ce->ce_length;
	assert(eend > estart);
	
	if (dend <= estart)			/* check for overlap */
		return(0);
	if (eend <= dstart)
		return(0);
	
	if (dstart < estart)			/* clip to extent */
		dstart = estart;
	if (dend > eend)
		dend = eend;
	
	/*
	 * Convert length in bytes to a block count.
	 */
	nblk = CB_BYTE_TO_BLOCK(cm, dend - dstart);
	
	/*
	 * Mark blocks vacant.  For each vacated block, check whether the page
	 * holding it can be freed.
	 *
	 * Note that this could be optimised to reduce the number of
	 * page-freeable checks, but the logic would be more complex.
	 */
	base = CB_BYTE_TO_BLOCK(cm, dstart - ce->ce_diskoffset);
	assert(base >= 0);
	discarded = 0;
	assert((base + nblk - 1) < howmany(ce->ce_length, cm->cm_blocksize));
	for (i = 0; i < nblk; i++) {
		
		/* this is a no-op if the block is already gone */
		if (CB_BLOCK_PRESENT(ce, base + i)) {
			
			/* mark the block as vacant */
			CB_MARK_BLOCK_VACANT(ce, base + i);
			discarded++;
			
			page = CB_BLOCK_TO_PAGE(cm, base + i);
			if (CB_PAGE_VACANT(cm, ce, page))
				BC_free_page(ce, (int) page);
		}
	}
	return(discarded * cm->cm_blocksize);
}

/*
 * Blocks in an extent can be invalidated, e.g. by a write, before the
 * reader thread fills the extent. Search for the next range of viable blocks
 * in the extent, starting from offset 'fromoffset'.
 *
 * Values are returned by out parameters:
 * 	'nextpage' takes the page containing the start of the next run, or
 * 		-1 if no more runs are to be found.
 * 	'nextoffset' takes the offset into that page that the run begins.
 * 	'nextlength' takes the length in bytes of the range, capped at maxread.
 *
 * In other words, for this blockmap, if
 * 	    *fromoffset = 3 * blocksize,
 * 	1 1 0 0 0 1 1 1   1 1 1 1 1 1 0 0
 * 	*nextpage = 0
 * 	          *nextoffset = 5 * blocksize
 * 	                            *nextlength = 6 * blocksize
 *
 * Called with the extent lock held
 */
static void 
BC_next_valid_range(struct BC_cache_mount *cm, struct BC_cache_extent *ce, uint32_t fromoffset, 
					uint32_t *nextpage, uint32_t *nextoffset, uint32_t *nextlength)
{
	int maxblks, i, nextblk = 0;
	int found = 0;
	
	maxblks = howmany(ce->ce_length, cm->cm_blocksize);
	i = CB_BYTE_TO_BLOCK(cm, fromoffset);
	/* scan for the next range of valid blocks in the extent */
	for (; i < maxblks; i++) {
		if (CB_BLOCK_PRESENT(ce, i)) {
			if (found == 0) {
				nextblk = i;
			}
			found++;
		} else {
			if (found != 0)
				break;
		}
	}
	
	if (found) {
		/* found a valid range, so convert to (page, offset, length) */
		*nextpage = CB_BLOCK_TO_PAGE(cm, nextblk);
		*nextoffset = CB_BLOCK_TO_BYTE(cm, nextblk) % PAGE_SIZE;
		*nextlength = MIN(CB_BLOCK_TO_BYTE(cm, found), cm->cm_maxread);
	} else {
		*nextpage = -1;
	}
	
	return;
}

/*
 * Test for the presence of a range of blocks in the cache.
 *
 * Called with the extent lock held.
 */
static int
BC_blocks_present(struct BC_cache_extent *ce, int base, int nblk)
{
	int	blk;
	
	assert(base >= 0);
	assert((base + nblk) <= howmany(ce->ce_length, BC_cache->c_mounts[ce->ce_mount_idx].cm_blocksize));
	
	for (blk = 0; blk < nblk; blk++) {
		
		if (!CB_BLOCK_PRESENT(ce, base + blk)) {
			return(0);
		}
	}
	return(1);
}

/*
 * Readahead thread.
 */
static void
BC_reader_thread(void *param0, wait_result_t param1)
{
	struct BC_cache_disk *cd = NULL;
	struct BC_cache_mount *cm = NULL;
	struct BC_cache_extent *ce = NULL;
	int count, num_mounts, ce_idx, cel_idx;
	upl_t	upl;
	kern_return_t kret;
	struct timeval batchstart, batchend;
	int error = 0;
	int issuing_lowpriority_extents = 0;
	
	/* We can access the BC_cache_disk passed in freely without locks
	 * since disks are only freed up in terminate cache after it has
	 * guanteed that no reader threads are running
	 */
	cd = (struct BC_cache_disk*) param0;
	
	num_mounts = cd->cd_nmounts;
		
	if (BC_cache->c_flags & BC_FLAG_SHUTDOWN)
		goto out;
	
	for (;;) {
				
		if (issuing_lowpriority_extents) {
			debug("disk %d starting low priority batch", cd->cd_disk_num);
		} else {
			debug("disk %d starting batch %d", cd->cd_disk_num, cd->cd_batch);
			KERNEL_DEBUG_CONSTANT(FSDBG_CODE(DBG_BOOTCACHE, DBG_BC_BATCH) | 
								  DBG_FUNC_START,
								  cd->cd_batch, cd->cd_disk_num, 0, 0, 0);
		}
		microtime(&batchstart);
		
		LOCK_CACHE_R();
		for (cel_idx = 0;
			 cel_idx < BC_cache->c_nextentlists;
			 cel_idx++) {
			
			/* iterate over extents to populate */
			for (ce_idx = 0;
				 ce_idx < BC_cache->c_nextents[cel_idx];
				 ce_idx++) {
								
				ce = BC_cache->c_extentlists[cel_idx] + ce_idx;
				
				cm = BC_cache->c_mounts + ce->ce_mount_idx;
				
				if (cm->cm_state != CM_READY) {
					continue;
				}
				
				if (cm->cm_disk != cd){
					continue;
				}
				
				/* Only read extents marked for this batch or earlier.
				 * All low-priority IO is read in one batch, however.
				 */
				if (ce->ce_batch > cd->cd_batch && !(cd->cd_flags & CD_ISSUE_LOWPRI)) {
					continue;
				}
								
				/* Check if already done or aborted */
				if (ce->ce_flags & (CE_ABORTED | CE_IODONE)) {
					continue;
				}
				
				/* Check if this extent is low priority and we're not yet reading in low-priority extents */
				if (!(cd->cd_flags & CD_ISSUE_LOWPRI) &&
					(ce->ce_flags & CE_LOWPRI)) {
					continue;
				}
				
				/* Shouldn't happen */
				if((cd->cd_flags & CD_ISSUE_LOWPRI) &&
				   !(ce->ce_flags & CE_LOWPRI)) {
					debug("Saw a regular extent while issuing throttled IO");
				}
				
				LOCK_EXTENT(ce);
				
				/* Check if again with the lock */
				if (ce->ce_flags & (CE_ABORTED | CE_IODONE)) {
					UNLOCK_EXTENT(ce);
					continue;
				}
				
				/* loop reading to fill this extent */
				uint32_t fromoffset = 0;
				
				LOCK_MOUNT_R(cm);
				
				if (cm->cm_state != CM_READY) {
					/* the mount was aborted. Whoever aborted it handles aborting the extents as well */
					UNLOCK_MOUNT_R(cm);
					UNLOCK_EXTENT(ce);
					continue;
				}
				
				for (;;) {
					uint32_t nextpage;
					uint32_t nextoffset;
					uint32_t nextlength;
					
					/* requested shutdown */
					if (BC_cache->c_flags & BC_FLAG_SHUTDOWN) {
						UNLOCK_MOUNT_R(cm);
						goto out;
					}
					
					
					/*
					 * Find the next set of blocks that haven't been invalidated
					 * for this extent.
					 */
					BC_next_valid_range(cm, ce, fromoffset, &nextpage, &nextoffset, &nextlength);
					/* no more blocks to be read */
					if (nextpage == -1)
						break;
					
					/* set up fromoffset to read the next segment of the extent */
					fromoffset = (nextpage * PAGE_SIZE) + nextoffset + nextlength;
					
					kret = vnode_getwithvid(BC_cache->c_vp, BC_cache->c_vid);
					if (kret != KERN_SUCCESS) {
						UNLOCK_MOUNT_R(cm);
						message("reader thread: vnode_getwithvid failed - %d\n", kret);
						goto out;
					}
					
					kret = ubc_create_upl(BC_cache->c_vp, 
										  ce->ce_cacheoffset + (nextpage * PAGE_SIZE), 
										  (int) roundup(nextoffset + nextlength, PAGE_SIZE), 
										  &upl, 
										  NULL, 
										  UPL_SET_LITE|UPL_FILE_IO);
					if (kret != KERN_SUCCESS) {
						UNLOCK_MOUNT_R(cm);
						message("ubc_create_upl returned %d\n", kret);
						(void) vnode_put(BC_cache->c_vp);
						goto out;
					}
					
					/* cm_bp is only accessed from the reader thread,
					 * and there's only one reader thread for a given mount,
					 * so we don't need the write lock to modify it */
					
					/* set buf to fill the requested subset of the upl */
					buf_setblkno(cm->cm_bp, CB_BYTE_TO_BLOCK(cm, ce->ce_diskoffset + nextoffset) + CB_PAGE_TO_BLOCK(cm, nextpage));
					buf_setcount(cm->cm_bp, (unsigned int) nextlength);
					buf_setupl(cm->cm_bp, upl, (unsigned int) nextoffset);
					buf_setresid(cm->cm_bp, buf_count(cm->cm_bp));	/* ask for residual indication */
					buf_reset(cm->cm_bp, B_READ);
					
					/* If this is regular readahead, make sure any throttled IO are throttled by the readahead thread rdar://8592635
					 * If this is low-priority readahead, make sure this thread will be throttled appropriately later rdar://8734858
					 */
					throttle_info_handle_t throttle_info_handle;
					if ((error = throttle_info_ref_by_mask(cm->cm_throttle_mask, &throttle_info_handle)) == 0) {
						throttle_info_update_by_mask(throttle_info_handle, 0x0);
						throttle_info_rel_by_mask(throttle_info_handle);
					} else {
						debug("Unable to update throttle info for mount %s: %d", cm->cm_uuid, error);
					}					
					
					BC_ADD_STAT(initiated_reads, 1);
					if (cd->cd_disk_num < STAT_DISKMAX) {
						if (ce->ce_flags & CE_LOWPRI) {
							BC_ADD_STAT(disk_initiated_reads_lowpri[cd->cd_disk_num], 1);
						} else {
							BC_ADD_STAT(disk_initiated_reads[cd->cd_disk_num], 1);
						}
					}
					
					/* give the buf to the underlying strategy routine */
					KERNEL_DEBUG_CONSTANT(FSDBG_CODE(DBG_DKRW, DKIO_READ) | DBG_FUNC_NONE,
										  (uintptr_t) cm->cm_bp, buf_device(cm->cm_bp), (long)buf_blkno(cm->cm_bp), buf_count(cm->cm_bp), 0);					
					
					BC_cache->c_strategy(cm->cm_bp);
					
					/* wait for the bio to complete */
					buf_biowait(cm->cm_bp);
					
					kret = ubc_upl_commit(upl);
					(void) vnode_put(BC_cache->c_vp);
					
					/*
					 * If the read or commit returned an error, invalidate the bytes
					 * covered by the read (on a residual, we could avoid invalidating
					 * bytes that are claimed to be read as a minor optimisation, but we do
					 * not expect errors as a matter of course).
					 */
					if (kret != KERN_SUCCESS || buf_error(cm->cm_bp) || (buf_resid(cm->cm_bp) != 0)) {
						debug("read error: extent %lu %lu/%lu "
							  "(error buf %ld/%ld flags %08x resid %d)",
							  (unsigned long) ce_idx,
							  (unsigned long)ce->ce_diskoffset, (unsigned long)ce->ce_length,
							  (long)buf_blkno(cm->cm_bp), (long)buf_count(cm->cm_bp),
							  buf_flags(cm->cm_bp), buf_resid(cm->cm_bp));
						
						count = BC_discard_bytes(ce, ce->ce_diskoffset + nextoffset, nextlength);
						debug("read error: discarded %d bytes", count);
						BC_ADD_STAT(read_errors, 1);
						BC_ADD_STAT(read_errors_bytes, count);
					}
#ifdef READ_HISTORY_BUFFER
					if (BC_cache->c_rhistory && BC_cache->c_rhistory_idx < READ_HISTORY_BUFFER) {
						BC_cache->c_rhistory[BC_cache->c_rhistory_idx].rh_extent = ce;
						BC_cache->c_rhistory[BC_cache->c_rhistory_idx].rh_blkno = buf_blkno(cm->cm_bp);
						BC_cache->c_rhistory[BC_cache->c_rhistory_idx].rh_bcount = buf_count(cm->cm_bp);
						if (buf_error(cm->cm_bp) || (buf_resid(cm->cm_bp) != 0)) {
							BC_cache->c_rhistory[BC_cache->c_rhistory_idx].rh_result = BC_RHISTORY_FAIL;
							BC_cache->c_rhistory_idx++;
						} else {
# ifdef READ_HISTORY_ALL
							BC_cache->c_rhistory[BC_cache->c_rhistory_idx].rh_result = BC_RHISTORY_OK;
							BC_cache->c_rhistory_idx++;
# endif
						}
					}
#endif
										
					/* I'd like to throttle the thread here, but I'm holding the extent and cache locks, so it's throttled below */
				}
				
				/* update stats */
				BC_ADD_STAT(read_blocks, (u_int) CB_BYTE_TO_BLOCK(cm, ce->ce_length));
				BC_ADD_STAT(read_bytes,  (u_int) ce->ce_length);
				if (ce->ce_flags & CE_LOWPRI) {
					BC_ADD_STAT(read_bytes_lowpri,  (u_int) ce->ce_length);
					if (cd->cd_disk_num < STAT_DISKMAX) {
						BC_ADD_STAT(batch_bytes_lowpri[cd->cd_disk_num],  (u_int) ce->ce_length);
					}
				} else {
					if (cd->cd_disk_num < STAT_DISKMAX) {
						BC_ADD_STAT(batch_bytes[cd->cd_disk_num][MIN(cd->cd_batch, STAT_BATCHMAX)],  (u_int) ce->ce_length);
						if (ce->ce_batch < cd->cd_batch) {
							BC_ADD_STAT(batch_late_bytes[cd->cd_disk_num][MIN(cd->cd_batch, STAT_BATCHMAX)], ce->ce_length);
						}
					}
				}
				if (ce->ce_flags & CE_SHARED) {
					BC_ADD_STAT(shared_bytes, (u_int) ce->ce_length);
				}
				UNLOCK_MOUNT_R(cm);
				
				/*
				 * Wake up anyone wanting this extent, as it is now ready for
				 * use.
				 */
				ce->ce_flags |= CE_IODONE;
				UNLOCK_EXTENT(ce);
				wakeup(ce);
				
				/* Let anyone waiting for the write lock to take hold */
				UNLOCK_CACHE_R();

				/* Take this opportunity of locklessness to throttle the reader thread, if necessary */
				if (issuing_lowpriority_extents) {
					throttle_lowpri_io((ce->ce_flags & CE_LOWPRI) ? 1 : 0);
				}
				
				LOCK_CACHE_R();
				if (issuing_lowpriority_extents && !(cd->cd_flags & CD_ISSUE_LOWPRI)) {
					/* New playlist arrived, go back to issuing regular priority extents */
					debug("New extents, interrupting low priority readahead");
					break;
				}

				if (cel_idx >= BC_cache->c_nextentlists ||
					ce_idx  >= BC_cache->c_nextents[cel_idx]) {
					/* cache shrunk while we weren't looking.
					 * Not a hard error, but we're done with this batch at least
					 */
					break;
				}
			}
			ce = NULL;
			if (issuing_lowpriority_extents && !(cd->cd_flags & CD_ISSUE_LOWPRI)) {
				/* New playlist arrived, go back to issuing regular priority extents */
				break;
			}
		}
		UNLOCK_CACHE_R();
		
		microtime(&batchend);
		timersub(&batchend, &batchstart, &batchend);
		if (cd->cd_disk_num < STAT_DISKMAX) {
			if (issuing_lowpriority_extents) {
				BC_cache->c_stats.ss_batch_time_lowpri[cd->cd_disk_num] += 
				(u_int) batchend.tv_sec * 1000 + 
				(u_int) batchend.tv_usec / 1000;
			} else {
				/* Measure times for the first several batches separately */
				BC_cache->c_stats.ss_batch_time[cd->cd_disk_num][MIN(cd->cd_batch, STAT_BATCHMAX)] += 
				(u_int) batchend.tv_sec * 1000 + 
				(u_int) batchend.tv_usec / 1000;
			}
		}
		if (issuing_lowpriority_extents) {
			// debug("disk %d low priority batch done", cd->cd_disk_num);
		} else {
			// debug("disk %d batch %d done", cd->cd_disk_num, cd->cd_batch);
			KERNEL_DEBUG_CONSTANT(FSDBG_CODE(DBG_BOOTCACHE, DBG_BC_BATCH) | 
								  DBG_FUNC_END,
								  cd->cd_batch, cd->cd_disk_num, 0, 0, 0);
		}
		

		LOCK_DISK(cd);
		if (issuing_lowpriority_extents && !(cd->cd_flags & CD_ISSUE_LOWPRI)) {
			/* New playlist arrived, go back to issuing regular extents */
			issuing_lowpriority_extents = 0;
			throttle_set_thread_io_policy(IOPOL_NORMAL);
			cd->cd_batch = 0;
		} else if (cd->cd_batch + 1 < BC_cache->c_batch_count && !issuing_lowpriority_extents) {
			cd->cd_batch++;
		} else {
			if (num_mounts == cd->cd_nmounts) {
				/* no new mounts appeared */
				if ( !(cd->cd_flags & CD_ISSUE_LOWPRI)) {
					/* go through the playlist again, issuing any low-priority extents we have */
					cd->cd_flags |= CD_ISSUE_LOWPRI;
					throttle_set_thread_io_policy(IOPOL_THROTTLE);
					issuing_lowpriority_extents = 1;
				} else {
					/* we're done */
					cd->cd_flags &= (~CD_HAS_THREAD);
					cd->cd_batch = 0;
					UNLOCK_DISK(cd);
					goto out;
				}
			} else {
				/* New mounts appeared. Run through again */
				cd->cd_batch = 0;
				debug("disk %d more mounts detected (%d total)", cd->cd_disk_num, cd->cd_nmounts);
			}
		}
		num_mounts = cd->cd_nmounts;
		UNLOCK_DISK(cd);
	}
	
out:
	/*
	 * If ce != NULL we have bailed out, but we cannot free memory beyond
	 * the bailout point as it may have been read in a previous batch and 
	 * there may be a sleeping strategy routine assuming that it will 
	 * remain valid.  Since we are typically killed immediately before a 
	 * complete cache termination, this does not represent a significant 
	 * problem.
	 *
	 * However, to prevent readers blocking waiting for readahead to
	 * complete for extents that will never be read, we mark all the
	 * extents we have given up on as aborted.
	 */
	if (ce != NULL) {
		UNLOCK_EXTENT(ce);
		
		/* the only errors that get us here are global, so mark the cache as shut down */
		LOCK_DISK(cd);
		BC_set_flag(BC_FLAG_SHUTDOWN);
		cd->cd_flags &= (~CD_HAS_THREAD);
		UNLOCK_DISK(cd);
		
		int tmpcel_idx, tmpce_idx;
		for (tmpcel_idx = 0;
			 tmpcel_idx < BC_cache->c_nextentlists;
			 tmpcel_idx++) {
			for (tmpce_idx = 0;
				 tmpce_idx < BC_cache->c_nextents[tmpcel_idx];
				 tmpce_idx++) {
				ce = BC_cache->c_extentlists[tmpcel_idx] + tmpce_idx;
				LOCK_EXTENT(ce);
				/* abort any extent that hasn't been satisfied */
				if (!(ce->ce_flags & CE_IODONE)) {
					ce->ce_flags |= CE_ABORTED;
					/* wake up anyone asleep on this extent */
					UNLOCK_EXTENT(ce);
					wakeup(ce);
				} else {
					UNLOCK_EXTENT(ce);
				}
			}
		}
		UNLOCK_CACHE_R();
	}
	
	
	debug("disk %d done", cd->cd_disk_num);
	
	/* wake up someone that might be waiting for this reader thread to exit */
	wakeup(&cd->cd_flags);
	
	LOCK_READERS();
	/* wake up someone that might be waiting for all the reader threads to exit */
	BC_cache->c_num_reader_threads--;
	if (BC_cache->c_num_reader_threads == 0) {
		wakeup(&BC_cache->c_num_reader_threads);
	}
	UNLOCK_READERS();
	
}

/*
 * Handle an incoming close request.
 */
static int
BC_close(dev_t dev, int flags, int devtype, struct proc *p)
{
	struct BC_cache_mount *cm;
	int i, cm_idx;

	/* Mark our cache_mount for this device as invalid */

	LOCK_CACHE_R();
	if (BC_cache->c_flags & BC_FLAG_CACHEACTIVE) {
		
		cm_idx = BC_find_cache_mount(dev);
		if (-1 != cm_idx) {
			cm = BC_cache->c_mounts + cm_idx;
			debug("Tearing down closing mount %s with dev 0x%x", uuid_string(cm->cm_uuid), dev);
			LOCK_MOUNT_R(cm);
			if (cm->cm_state != CM_ABORTED) {
				/*
				 * Mark all extents as aborted. This will notify any sleepers in the 
				 * strategy routine that the extent won't have data for them.
				 *
				 * Note that we shouldn't try to take an extent lock while holding
				 * the exclusive mount lock or we risk deadlock with the reader thread
				 */
				for (i = 0; i < cm->cm_nextents; i++) {
					LOCK_EXTENT(cm->cm_pextents[i]);
					BC_teardown_extent(cm->cm_pextents[i]);
					UNLOCK_EXTENT(cm->cm_pextents[i]);
					wakeup(cm->cm_pextents[i]);
				}
				if (! LOCK_MOUNT_R_TO_W(cm)) {
					/* If there is someone waiting for the exclusive lock on this mount,
					 * we want to grab it before they can so they don't waste time.
					 * If we fail to take the exclusive lock it means someone
					 * else is also trying to upgrate the mount lock. We don't keep any
					 * state, so we can simply take the write lock again.
					 */
					LOCK_MOUNT_W(cm);
					if (cm->cm_state == CM_ABORTED) {
						// Mount is aborted, no more work necessary
						UNLOCK_MOUNT_W(cm);
						goto out;	
					}
				}
				BC_teardown_mount(cm);
				UNLOCK_MOUNT_W(cm);
			} else {
				UNLOCK_MOUNT_R(cm);
			}
		}
	}
	
out:
	UNLOCK_CACHE_R();
	
	return BC_cache->c_close(dev, flags, devtype, p);
}

/*
 * Returns CE_ABORTED if any of the extents in the array are aborted
 * Returns CE_LOWPRI  if any of the extents in the array are low priority and not done
 * Returns CE_IODONE  if all the extents in the array are done
 * Returns 0 otherwise
 */
static int extents_status(struct BC_cache_extent **pce, int nextents) {
	int ce_idx;
	int ret = CE_IODONE;
	
	for (ce_idx = 0;
		 ce_idx < nextents;
		 ce_idx++) {
		
		if (pce[ce_idx]->ce_flags & CE_ABORTED) {
			return CE_ABORTED;
		}
		
		if (! (pce[ce_idx]->ce_flags & CE_IODONE)) {
			if (pce[ce_idx]->ce_flags & CE_LOWPRI) {
				ret = CE_LOWPRI;
			} else {
				if (ret == CE_IODONE) {
					ret = 0;
				}
			}
		}
	}
	return ret;
}

/*
 * Waits for the given extent to be filled or aborted
 *
 * Should only be called with the extent lock.
 * NO OTHER LOCKS SHOULD BE HELD INCLUDING THE CACHE LOCK
 *
 * We are guaranteed that the extent won't disappear when we release
 * the cache lock by the cleanup code waiting for BC_cache->c_strategycalls
 * to reach 0, and we're counted in there.
 */
static void wait_for_extent(struct BC_cache_extent *ce, struct timeval starttime) {
	struct timeval time;
	struct timespec timeout;
	microtime(&time);
	timersub(&time,
			 &starttime,
			 &time);
	if (time.tv_sec < BC_wait_for_readahead) {
		timeout.tv_sec = BC_wait_for_readahead - time.tv_sec;
		if (time.tv_usec == 0) {
			timeout.tv_nsec = 0;
		} else {
			timeout.tv_sec--;
			timeout.tv_nsec = NSEC_PER_USEC * (USEC_PER_SEC - time.tv_usec);
		}
		msleep(ce, &ce->ce_lock, PRIBIO, "BC_strategy", &timeout);
	}
}

/*
 * Handle an incoming IO request.
 */
static int
BC_strategy(struct buf *bp)
{
	struct BC_cache_extent *ce = NULL, **pce, **pcontaining_extents = NULL;
	int num_extents;
	uio_t uio = NULL;
	int nbytes, resid, discards = 0;
	struct timeval blocked_start_time, blocked_end_time;
	daddr64_t blkno;
	caddr_t buf_map_p, buf_extent_p;
	off_t disk_offset;
	kern_return_t kret;
	int cm_idx = -1, ce_idx;
	dev_t dev;
	int32_t bufflags;
	int during_cache = 0, during_io = 0, take_detailed_stats = 0, cache_hit = 0;
	int is_root_disk, did_block, status;
	int is_shared = 0, is_swap = 0;
	int should_throttle = 0;
	int is_ssd = 0;
	int unfilled = 0;
	vnode_t vp = NULL;
	int pid = 0;
	int dont_cache = 0;
	
	assert(bp != NULL);
	
	blkno = buf_blkno(bp);
	nbytes = buf_count(bp);
	bufflags = buf_flags(bp);
	dev = buf_device(bp);
	vp = buf_vnode(bp);

	/*
	 * If the buf doesn't have a vnode for some reason, pretend
	 * we never saw the request at all.
	 */
	if (dev == nulldev()) {
		BC_cache->c_strategy(bp);
		BC_ADD_STAT(strategy_unknown, 1);
		BC_ADD_STAT(strategy_unknown_bytes, nbytes);
		return 0;
	}
	
	if (vp && vnode_isswap(vp)) {
		is_swap = 1;
		goto bypass;
	}
	
	if (BC_cache->c_flags & BC_FLAG_HISTORYACTIVE) {
		pid = proc_selfpid();
		
		if (vp) {
			is_shared = vnode_isdyldsharedcache(vp);
 		}
	}
	
	if (vp && (vnode_israge(vp))) {
		dont_cache = 1;
	}
		
	/*
	 * If the cache is not active, bypass the request.  We may
	 * not be able to fill it, but we still want to record it.
	 */
	if (!(BC_cache->c_flags & BC_FLAG_CACHEACTIVE)) {
		goto bypass;
	}
		
	/*
	 * In order to prevent the cache cleanup code from racing with us
	 * when we sleep, we track the number of strategy calls in flight.
	 */
	OSAddAtomic(1, &BC_cache->c_strategycalls);
	LOCK_CACHE_R();
	during_cache = 1; /* we're included in the in_flight count */
	
	/* Get cache mount asap for use in case we bypass */
	cm_idx = BC_find_cache_mount(dev);
	
	if (cm_idx != -1) {
		disk_offset = CB_BLOCK_TO_BYTE(BC_cache->c_mounts + cm_idx, blkno);
	}
	
	if (BC_cache->c_take_detailed_stats) {
		take_detailed_stats = 1;
		BC_ADD_STAT(strategy_calls, 1);
		if (bufflags & B_THROTTLED_IO) {
			BC_ADD_STAT(strategy_throttled, 1);
		}
#ifdef BC_DEBUG
//		if (dont_cache) {
//			char procname[128];
//			proc_selfname(procname, sizeof(procname));
//			const char* filename = vp ? vnode_getname(vp) : NULL;
//			debug("not recording %s%s from app %s for file %s (disk block 0x%llx) which is%s throttled", (vp && vnode_israge(vp)) ? "rapid age " : "", (bufflags & B_READ) ? "read" : "write", procname, filename?:"unknown", blkno, (bufflags & B_THROTTLED_IO) ? "" : " not");
//			if (filename) {
//				vnode_putname(filename);
//			}
//		}
#endif
	}
	
	/*
	 * if we haven't seen this mount before, pass it off
	 *
	 * If this is a mount we want to cache, we'll kick it off when
	 * adding to our history
	 */	
	if (cm_idx == -1) {
		/* don't have a mount for this IO */
		if (take_detailed_stats)
			BC_ADD_STAT(strategy_noncached_mount, 1);
		goto bypass;
	}
	
	/* Get the info from the mount we need */
	LOCK_MOUNT_R(BC_cache->c_mounts + cm_idx);

	if (BC_cache->c_mounts[cm_idx].cm_disk &&
		(BC_cache->c_mounts[cm_idx].cm_disk->cd_flags & CD_HAS_THREAD) &&
		!(BC_cache->c_mounts[cm_idx].cm_disk->cd_flags & CD_ISSUE_LOWPRI)) {
		during_io = 1; /* for statistics */
		if (take_detailed_stats) {
			BC_ADD_STAT(strategy_duringio, 1);
		}
	}
	
	if (BC_cache->c_mounts[cm_idx].cm_state != CM_READY) {
		/* the mount has been aborted, treat it like a missing mount */
		UNLOCK_MOUNT_R(BC_cache->c_mounts + cm_idx);
		cm_idx = -1;

		if (take_detailed_stats)
			BC_ADD_STAT(strategy_noncached_mount, 1);
		goto bypass;
	}
		
	/* if it's not a read, pass it off */
	if ( !(bufflags & B_READ)) {
		UNLOCK_MOUNT_R(BC_cache->c_mounts + cm_idx);
		if (take_detailed_stats)
			BC_ADD_STAT(strategy_nonread, 1);
		goto bypass;
	}	
	
	if ((nbytes % BC_cache->c_mounts[cm_idx].cm_blocksize) != 0) {
		debug("IO with non-blocksize multiple bytes (%d %% %lld = %lld)", nbytes, BC_cache->c_mounts[cm_idx].cm_blocksize, (nbytes % BC_cache->c_mounts[cm_idx].cm_blocksize));
		UNLOCK_MOUNT_R(BC_cache->c_mounts + cm_idx);
		if (take_detailed_stats)
			BC_ADD_STAT(strategy_nonblocksize, 1);
		goto bypass;
	}
			
	if (take_detailed_stats) {
		BC_ADD_STAT(extent_lookups, 1);
		BC_ADD_STAT(requested_bytes, nbytes);
		if (cm_idx < STAT_MOUNTMAX) {
			BC_ADD_STAT(requested_bytes_m[cm_idx], nbytes);
		}
	}
	
	/*
	 * Look for a cache extent containing this request.
	 */
	num_extents = 0;
	pce = BC_find_extent(BC_cache->c_mounts + cm_idx, disk_offset, nbytes, 1, &num_extents);	
	if (pce == NULL) {
#ifdef BC_EXTRA_DEBUG
		if (take_detailed_stats && !dont_cache) {
			char procname[128];
			proc_selfname(procname, sizeof(procname));
			const char* filename = vp ? vnode_getname(vp) : NULL;
			if (num_extents > 0) {
				xdebug("Missed IO from app %s for file %s (disk offset 0x%llx) (intersected, though)", procname, filename?:"unknown", blkno);
			} else {
				xdebug("Missed IO from app %s for file %s (disk offset 0x%llx)", procname, filename?:"unknown", blkno);
			}
			if (filename) {
				vnode_putname(filename);
			}
		}
#endif
		UNLOCK_MOUNT_R(BC_cache->c_mounts + cm_idx);
		goto bypass;
	}
	
	assert(num_extents > 0);
	
	cache_hit = 1;
	
	if (take_detailed_stats) {
		BC_ADD_STAT(extent_hits, 1);
		if (during_io)
			BC_ADD_STAT(hit_duringio, 1);
		if (num_extents > 1)
			BC_ADD_STAT(hit_multiple, 1);
	}
	
	pcontaining_extents = BC_MALLOC(num_extents * sizeof(*pcontaining_extents), M_TEMP, M_WAITOK | M_ZERO);
	if (pcontaining_extents == NULL) {
		UNLOCK_MOUNT_R(BC_cache->c_mounts + cm_idx);
		if (take_detailed_stats)
			BC_ADD_STAT(hit_failure, 1);
		goto bypass;
	}
	memcpy(pcontaining_extents, pce, num_extents * sizeof(*pcontaining_extents));
	
	is_ssd = (BC_cache->c_mounts[cm_idx].cm_disk && 
			 (BC_cache->c_mounts[cm_idx].cm_disk->cd_flags & CD_IS_SSD));
	
	UNLOCK_MOUNT_R(BC_cache->c_mounts + cm_idx);
	
	/* if any extent is aborted, bypass immediately */
	status = extents_status(pcontaining_extents, num_extents);
	if (status & CE_ABORTED) {
		if (take_detailed_stats)
			BC_ADD_STAT(hit_aborted, 1);
		goto bypass;
	}
	
#ifndef EMULATE_ONLY
	
	if (! (status & CE_IODONE)) {
						
		if (is_ssd) {
			/* Don't delay IOs on SSD since cut-throughs aren't harmful */
			unfilled = 1;
			goto bypass;
		}
		
		if (status & CE_LOWPRI) {
			/* Don't wait for low priority extents */
			if (take_detailed_stats)
				BC_ADD_STAT(strategy_unfilled_lowpri, 1);
			cache_hit = 0;
			goto bypass;
		}		
				
		/*
		 * wait for all the extents to finish
		 */
		microtime(&blocked_start_time);
		did_block = 0;
		for (ce_idx = 0;
			 ce_idx < num_extents;
			 ce_idx++) {
			
			ce = pcontaining_extents[ce_idx];
			LOCK_EXTENT(ce);
			if (! (ce->ce_flags & (CE_ABORTED | CE_IODONE))) {
				did_block = 1;
				UNLOCK_CACHE_R();
				/*
				 * We need to release the cache lock since we will msleep in wait_for_extent.
				 * If we don't release the cache read lock and someone wants to add a new playlist,
				 * they go to grab the cache write lock, and everybody stalls for the msleep timeout.
				 *
				 * We are guaranteed that the cache won't disappear during this time since
				 * the cleanup code waits for all the strategy calls (this function) to complete
				 * by checking BC_cache->c_strategycalls, which we incremented above.
				 *
				 * The cache may have grown behind our backs, however. Mount pointers are not valid,
				 * but mount indexes still are. Extent pointers are still valid since the extent list
				 * is not modified (this is a requirement to be able to msleep with the extent's lock).
				 */
				wait_for_extent(ce, blocked_start_time);
				
				/* Cache lock must be grabbed without holding any locks to avoid deadlock rdar://8626772 */
				UNLOCK_EXTENT(ce);
				LOCK_CACHE_R();
				LOCK_EXTENT(ce);
				
				if (! (ce->ce_flags & (CE_ABORTED | CE_IODONE))) {
					/* strategy call timed out waiting for the extent */
					if (take_detailed_stats)
						BC_ADD_STAT(strategy_timedout, 1);
#ifdef BC_DEBUG
					microtime(&blocked_end_time);
					timersub(&blocked_end_time,
							 &blocked_start_time,
							 &blocked_end_time);
					char procname[128];
					proc_selfname(procname, sizeof(procname));
					const char* filename = vp ? vnode_getname(vp) : NULL;
					debug("Strategy routine timed out app %s waiting on file %s (disk offset 0x%llx) after %lu.%03d seconds", procname, filename?:"unknown", disk_offset, blocked_end_time.tv_sec, blocked_end_time.tv_usec / 1000);
					if (filename) {
						vnode_putname(filename);
					}
#endif
					goto bypass;
				}
				UNLOCK_EXTENT(ce);
				ce = NULL;
				
				/* Check every time we sleep so we can avoid more sleeping if unnecessary */
				status = extents_status(pcontaining_extents, num_extents);
				if (status & CE_ABORTED) {
					if (take_detailed_stats)
						BC_ADD_STAT(hit_aborted, 1);
					goto bypass;
				} else if (status & CE_IODONE) {
					break;
				}
			} else {
				UNLOCK_EXTENT(ce);
				ce = NULL;
			}
		}
		if (take_detailed_stats && did_block) {
			BC_ADD_STAT(strategy_blocked, 1);	
			microtime(&blocked_end_time);
			timersub(&blocked_end_time,
					 &blocked_start_time,
					 &blocked_end_time);
			int ms_blocked = (blocked_end_time.tv_sec * 1000) + (blocked_end_time.tv_usec / (1000));
#ifdef BC_DEBUG
			char procname[128];
			proc_selfname(procname, sizeof(procname));
			const char* filename = vp ? vnode_getname(vp) : NULL;
			debug("Strategy routine blocked app %s waiting on file %s (disk offset 0x%llx) for %lu.%03d seconds", procname, filename?:"unknown", disk_offset, blocked_end_time.tv_sec, blocked_end_time.tv_usec / 1000);
			if (filename) {
				vnode_putname(filename);
			}
#endif
			BC_ADD_STAT(strategy_time_blocked, ms_blocked);
			if (BC_cache->c_stats.ss_strategy_time_longest_blocked < ms_blocked) {
				BC_cache->c_stats.ss_strategy_time_longest_blocked = ms_blocked;
			}
		}
	}
	
	/*
	 * buf_map will do its best to provide access to this
	 * buffer via a kernel accessible address
	 * if it fails, we can still fall through.
	 */
	if (buf_map(bp, &buf_map_p)) {
		/* can't map, let someone else worry about it */
		if (take_detailed_stats)
			BC_ADD_STAT(hit_failure, 1);
		goto bypass;
	}
	buf_extent_p = buf_map_p;
	
	kret = vnode_getwithvid(BC_cache->c_vp, BC_cache->c_vid);
	if (kret != KERN_SUCCESS) {
		debug("vnode_getwithvid failed - %d", kret);
		if (take_detailed_stats)
			BC_ADD_STAT(hit_failure, 1);
		buf_unmap(bp);
		goto bypass;
	}
	
	/*
	 * fill in bp from the extents
	 */		
	for (ce_idx = 0;
		 ce_idx < num_extents;
		 ce_idx++) {
		u_int64_t nbytes_thisextent;
		u_int64_t diskoffset_thisextent;
		off_t     cacheoffset_thisextent;
		
		ce = pcontaining_extents[ce_idx];
		LOCK_EXTENT(ce);
				
		/*
		 * Check that the extent wasn't aborted while we were unlocked.
		 */
		if (ce->ce_flags & CE_ABORTED) {
			if (take_detailed_stats)
				BC_ADD_STAT(hit_aborted, 1);
			vnode_put(BC_cache->c_vp);
			buf_unmap(bp);
			goto bypass;
		}
		
		assert(ce->ce_flags & CE_IODONE);
		
		diskoffset_thisextent  = MAX(ce->ce_diskoffset, disk_offset);
		nbytes_thisextent      = MIN(disk_offset + nbytes, ce->ce_diskoffset + ce->ce_length) - diskoffset_thisextent;
		cacheoffset_thisextent = ce->ce_cacheoffset + (diskoffset_thisextent - ce->ce_diskoffset);

		/* range check against extent */
		assert(diskoffset_thisextent  <  ce->ce_diskoffset + ce->ce_length);
		assert(diskoffset_thisextent  >= ce->ce_diskoffset);
		assert(nbytes_thisextent      <= (ce->ce_diskoffset + ce->ce_length) - diskoffset_thisextent);
		assert(cacheoffset_thisextent <  ce->ce_cacheoffset + ce->ce_length);
		assert(cacheoffset_thisextent >= ce->ce_cacheoffset);
		assert(nbytes_thisextent      <= (ce->ce_cacheoffset + ce->ce_length) - cacheoffset_thisextent);
		
		/* range check against buf_t */
		assert(diskoffset_thisextent    <  disk_offset + nbytes);
		assert(diskoffset_thisextent    >= disk_offset);
		assert(nbytes_thisextent        <= (disk_offset + nbytes) - diskoffset_thisextent);

		/* check our buf_map pointer */
		assert(buf_extent_p - buf_map_p == diskoffset_thisextent - disk_offset); /* didn't skip any bytes */
		assert(buf_map_p + nbytes       >= buf_extent_p + nbytes_thisextent); /* not overflowing the buffer */
		
		/* Make sure we're including the entire buffer requested */
		assert(ce_idx > 0 || disk_offset == diskoffset_thisextent); /* include the first byte */
		assert(ce_idx < (num_extents - 1) || disk_offset + nbytes == diskoffset_thisextent + nbytes_thisextent); /* include the last byte */
		
		/*
		 * Check that the requested blocks are in the buffer.
		 */
		if (!BC_blocks_present(ce,
							   CB_BYTE_TO_BLOCK(BC_cache->c_mounts + cm_idx, diskoffset_thisextent - ce->ce_diskoffset),
							   CB_BYTE_TO_BLOCK(BC_cache->c_mounts + cm_idx, nbytes_thisextent))) {
			if (take_detailed_stats)
				BC_ADD_STAT(hit_blkmissing, 1);
			vnode_put(BC_cache->c_vp);
			buf_unmap(bp);
			goto bypass;
		}
		
		resid = nbytes_thisextent;
		uio = uio_create(1, cacheoffset_thisextent, UIO_SYSSPACE, UIO_READ);
		if (uio == NULL) {
			debug("couldn't allocate uio");
			if (take_detailed_stats)
				BC_ADD_STAT(hit_failure, 1);
			vnode_put(BC_cache->c_vp);
			buf_unmap(bp);
			goto bypass;
		}
		
		kret = uio_addiov(uio, CAST_USER_ADDR_T(buf_extent_p), nbytes_thisextent);
		if (kret != KERN_SUCCESS) {
			debug("couldn't add iov to uio - %d", kret);
			if (take_detailed_stats)
				BC_ADD_STAT(hit_failure, 1);
			vnode_put(BC_cache->c_vp);
			buf_unmap(bp);
			goto bypass;
		}
		
		kret = cluster_copy_ubc_data(BC_cache->c_vp, uio, &resid, 1); 
		if (kret != KERN_SUCCESS) {
			debug("couldn't copy ubc data - %d", kret);
			if (take_detailed_stats)
				BC_ADD_STAT(hit_failure, 1);
			vnode_put(BC_cache->c_vp);
			buf_unmap(bp);
			goto bypass;
		}
				
		if (resid != 0) {
			/* blocks have been stolen for pageout or contiguous memory */
			if (take_detailed_stats)
				BC_ADD_STAT(hit_stolen, 1);
			vnode_put(BC_cache->c_vp);
			buf_unmap(bp);
			goto bypass;
		}
		
		buf_extent_p += nbytes_thisextent;
		
		/* discard blocks we have touched */
		discards += BC_discard_bytes(ce, disk_offset, nbytes_thisextent);

		UNLOCK_EXTENT(ce);
		ce = NULL;
		
		/* copy was successful */
		uio_free(uio);
		uio = NULL;
	}

	vnode_put(BC_cache->c_vp);

	buf_setresid(bp, 0);
	
	/* buf_unmap will take care of all cases */
	buf_unmap(bp);
	
	/* complete the request */
	buf_biodone(bp);
	
	/* can't dereference bp after a buf_biodone has been issued */

#else //ifndef EMULATE_ONLY
	(void) kret;
	(void) p;
	(void) resid;
	
	/* discard blocks we have touched */
	for (ce_idx = 0;
		 ce_idx < num_extents;
		 ce_idx++) {
		ce = pcontaining_extents[ce_idx];
		LOCK_EXTENT(ce);
		discards += BC_discard_bytes(ce, disk_offset, nbytes);
		UNLOCK_EXTENT(ce);
		ce = NULL;
	}
	
	/* send the IO to the disk, emulate hit in statistics */
	BC_cache->c_strategy(bp);
	
#endif //ifndef EMULATE_ONLY else
	
	if (take_detailed_stats) {
		BC_ADD_STAT(hit_bytes, discards);
		if (cm_idx < STAT_MOUNTMAX) {
			BC_ADD_STAT(hit_bytes_m[cm_idx], discards);
		}
		if (dont_cache) {
			BC_ADD_STAT(strategy_hit_nocache, 1);
			BC_ADD_STAT(hit_nocache_bytes, discards);
		}
		if (is_shared) {
			BC_ADD_STAT(hit_shared_bytes, discards);
		}
	} else {
		BC_ADD_STAT(hit_bytes_afterhistory, discards);
	}
			
	BC_cache->c_num_ios_since_last_hit = 0;
	
	/* we are not busy anymore */
	OSAddAtomic(-1, &BC_cache->c_strategycalls);
	UNLOCK_CACHE_R();
		
	_FREE_ZERO(pcontaining_extents, M_TEMP);

	if (! dont_cache) {
		/* record successful fulfilment (may block) */
		BC_add_history(blkno, nbytes, pid, 1, 0, 0, is_shared, dev);
	}
	
	/*
	 * spec_strategy wants to know if the read has been
	 * satisfied by the boot cache in order to avoid
	 * throttling the thread unnecessarily. spec_strategy
	 * will check for the special return value 0xcafefeed
	 * to indicate that the read was satisfied by the
	 * cache.
	 */
	
#define IO_SATISFIED_BY_CACHE ((int)0xcafefeed)
	return IO_SATISFIED_BY_CACHE;
	
bypass:
	if (ce != NULL)
		UNLOCK_EXTENT(ce);
	if (uio != NULL)
		uio_free(uio);
	_FREE_ZERO(pcontaining_extents, M_TEMP);
	
	/* pass the request on */
	BC_cache->c_strategy(bp);
	
	/*
	 * can't dereference bp after c_strategy has been issued
	 * or else we race with buf_biodone
	 */
	bp = NULL;
	
	/* not really "bypassed" if the cache is not active */
	if (during_cache) {
		if (cm_idx != -1) {
			LOCK_MOUNT_R(BC_cache->c_mounts + cm_idx);
			if (BC_cache->c_mounts[cm_idx].cm_state == CM_READY) {
				discards += BC_handle_discards(BC_cache->c_mounts + cm_idx, disk_offset, nbytes);
			}
			
			/* Check if we should throttle this IO */
			if (BC_cache->c_readahead_throttles_cutthroughs && 
				!is_swap &&
				BC_cache->c_mounts[cm_idx].cm_disk && 
				(BC_cache->c_mounts[cm_idx].cm_disk->cd_flags & CD_HAS_THREAD) &&
				!(BC_cache->c_mounts[cm_idx].cm_disk->cd_flags & CD_ISSUE_LOWPRI) &&
				!(BC_cache->c_mounts[cm_idx].cm_disk->cd_flags & CD_IS_SSD)) {
				/* We're currently issuing readahead for this disk.
				 * Throttle this IO so we don't cut-through the readahead so much.
				 */
				should_throttle = 1;
			}
			
			UNLOCK_MOUNT_R(BC_cache->c_mounts + cm_idx);
		}
		if (take_detailed_stats) {
			BC_ADD_STAT(strategy_bypassed, 1);
			if (during_io) {
				BC_ADD_STAT(strategy_bypass_duringio, 1);
			}
			if (dont_cache) {
				BC_ADD_STAT(strategy_bypass_nocache, 1);
				BC_ADD_STAT(bypass_nocache_bytes, nbytes);
				if (during_io) {
					BC_ADD_STAT(strategy_bypass_duringio_nocache, 1);
				}
			}
			if (cache_hit) {
				BC_ADD_STAT(error_discards, discards);
			} else if (dont_cache) {
				BC_ADD_STAT(bypass_nocache_discards, discards);
			} else if (bufflags & B_READ) {
				BC_ADD_STAT(read_discards, discards);
			} else {
				BC_ADD_STAT(write_discards, discards);			
			}
		} else {
			BC_ADD_STAT(lost_bytes_afterhistory, discards);
		}
	}
	
	if (during_cache) {
		OSAddAtomic(-1, &BC_cache->c_strategycalls);
		UNLOCK_CACHE_R();
	}
	
	if (! is_swap) {
		if (! dont_cache) {
			is_root_disk = BC_add_history(blkno, nbytes, pid, 0, ((bufflags & B_READ) ? 0 : 1), 0, is_shared, dev);
			
			if (take_detailed_stats && during_io && is_root_disk) {
				if (cache_hit) {
					if (unfilled) {
						BC_ADD_STAT(strategy_bypass_duringio_unfilled, 1);
					} else {					
						BC_ADD_STAT(strategy_bypass_duringio_rootdisk_failure, 1);
					}
				} else if (bufflags & B_READ) {
					BC_ADD_STAT(strategy_bypass_duringio_rootdisk_read, 1);
				} else {
					BC_ADD_STAT(strategy_bypass_duringio_rootdisk_nonread, 1);
				}
			}
		}
	}
	
	/*
	 * Check to make sure that we're not missing the cache
	 * too often.  If we are, we're no longer providing a
	 * performance win and the best thing would be to give
	 * up.
	 *
	 * Don't count throttled IOs since those aren't time
	 * critical. (We don't want to jettison the cache just
	 * because spotlight is indexing)
	 */
	
	/* if this is a read, and we do have an active cache, and the read isn't throttled */
	if ((bufflags & B_READ) &&
		(during_cache) &&
		!(bufflags & B_THROTTLED_IO)) {
		
		struct timeval current_time;
		if (BC_cache->c_stats.ss_history_num_recordings < 2) {
			microtime(&current_time);
			timersub(&current_time,
					 &BC_cache->c_loadtime,
					 &current_time);
		}
		/* Don't start counting misses until we've started login or hit our prelogin timeout */
		if (BC_cache->c_stats.ss_history_num_recordings >= 2 || current_time.tv_sec > BC_chit_prelogin_timeout_seconds) {
			
			/* increase number of IOs since last hit */
			OSAddAtomic(1, &BC_cache->c_num_ios_since_last_hit);
			
			if (BC_cache->c_num_ios_since_last_hit >= BC_chit_max_num_IOs_since_last_hit) {
				/*
				 * The hit rate is poor, shut the cache down.
				 */
				debug("hit rate below threshold (0 hits in the last %u lookups)",
					  BC_cache->c_num_ios_since_last_hit);
				if (BC_terminate_cache()) {
					message("could not terminate cache on bad hitrate");
				}
			}
		}
	}
	
	if (is_swap && (! (BC_cache->c_flags & BC_FLAG_SHUTDOWN))) {
		/* if we're swapping, stop readahead */
		debug("Detected %s swap file. Terminating readahead", (bufflags & B_READ) ? "read from" : "write to");
		BC_set_flag(BC_FLAG_SHUTDOWN);
	}
	
	if (should_throttle && !(bufflags & B_THROTTLED_IO)) {
		/*
		 * We need to indicate to spec_strategy that we want to
		 * throttle this IO to avoid cutting through readahead
		 * too much. spec_strategy will check for the special
		 * return value 0xcafebeef to indicate that the IO
		 * should be throttled.
		 */
		
		BC_ADD_STAT(strategy_forced_throttled, 1);
		
#define IO_SHOULD_BE_THROTTLED ((int)0xcafebeef)
		return IO_SHOULD_BE_THROTTLED;
	}
	
	return 0;
}

/*
 * Handle a block range that needs to be discarded.
 *
 * Called with the cache mount read lock held
 *
 * Returns the number of bytes discarded
 */
static int
BC_handle_discards(struct BC_cache_mount *cm, u_int64_t offset, u_int64_t length)
{
	struct BC_cache_extent **pce, **p;
	int count, total_discards;
	
	total_discards = 0;
	
	/*
	 * Look for an extent that we overlap.
	 */
	if ((pce = BC_find_extent(cm, offset, length, 0, NULL)) == NULL)
		return 0;		/* doesn't affect us */
	
	/*
	 * Discard bytes in the matched extent.
	 */
	LOCK_EXTENT(*pce);
	count = BC_discard_bytes((*pce), offset, length);
	UNLOCK_EXTENT(*pce);
	total_discards += count;
	
	/*
	 * Scan adjacent extents for possible overlap and discard there as well.
	 */
	p = pce - 1;
	while (p >= cm->cm_pextents && 
		   BC_check_intersection((*p), offset, length)) {
		LOCK_EXTENT(*p);
		count = BC_discard_bytes((*p), offset, length);
		UNLOCK_EXTENT(*p);
		if (count == 0)
			break;
		total_discards += count;
		p--;
	}
	p = pce + 1;
	while (p < (cm->cm_pextents + cm->cm_nextents) && 
		   BC_check_intersection((*p), offset, length)) {
		LOCK_EXTENT(*p);
		count = BC_discard_bytes((*p), offset, length);
		UNLOCK_EXTENT(*p);
		if (count == 0)
			break;
		total_discards += count;
		p++;
	}
	
	return total_discards;
}

/*
 * Shut down readahead operations.
 */
static int
BC_terminate_readahead(void)
{
	int error;
	struct timespec timeout;
	timeout.tv_sec = 10;
	timeout.tv_nsec = 0;
	
	/*
	 * Signal the readahead thread to terminate, and wait for
	 * it to comply.  If this takes > 10 seconds, give up.
	 */
	BC_set_flag(BC_FLAG_SHUTDOWN);
	
	/*
	 * If readahead is still in progress, we have to shut it down
	 * cleanly.  This is an expensive process, but since readahead
	 * will normally complete long before the reads it tries to serve
	 * complete, it will typically only be invoked when the playlist
	 * is out of synch and the cache hitrate falls below the acceptable
	 * threshold.
	 *
	 * Note that if readahead is aborted, the reader thread will mark
	 * the aborted extents and wake up any strategy callers waiting
	 * on them, so we don't have to worry about them here.
	 */
	LOCK_READERS();
	while (BC_cache->c_num_reader_threads > 0) {
		debug("terminating active readahead");
		
		error = msleep(&BC_cache->c_num_reader_threads, &BC_cache->c_reader_threads_lock, PRIBIO, "BC_terminate_readahead", &timeout);
		if (error == EWOULDBLOCK) {
			UNLOCK_READERS();
			
			message("timed out waiting for I/O to stop");
			if (BC_cache->c_num_reader_threads == 0) {
				debug("but I/O has stopped!");
			} 
#ifdef BC_DEBUG
 			Debugger("I/O Kit wedged on us");
#endif
			/*
			 * It might be nice to free all the pages that
			 * aren't actually referenced by the outstanding
			 * region, since otherwise we may be camped on a
			 * very large amount of physical memory.
			 *
			 * Ideally though, this will never happen.
			 */
			return(EBUSY);	/* really EWEDGED */
		}
	}
	UNLOCK_READERS();
	
	return(0);
}

/*
 * Terminate the cache.
 *
 * This prevents any further requests from being satisfied from the cache
 * and releases all the resources owned by it.
 *
 * Must be called with no locks held
 */
static int
BC_terminate_cache(void)
{
	int retry, cm_idx, j, ce_idx, cel_idx;

	BC_terminate_readahead();
	
	/* can't shut down if readahead is still active */
	if (BC_cache->c_num_reader_threads > 0) {
		debug("cannot terminate cache while readahead is in progress");
		return(EBUSY);
	}
	
	LOCK_CACHE_R();
	
	LOCK_HANDLERS();
	if (!BC_clear_flag(BC_FLAG_CACHEACTIVE)) { 
		/* can't shut down if we're not active */
		debug("cannot terminate cache when not active");
		UNLOCK_HANDLERS();
		UNLOCK_CACHE_R();
		return(ENXIO);
	}
	
	debug("terminating cache...");
	
	/* if we're no longer recording history also, disconnect our strategy routine */
	BC_check_handlers();
	UNLOCK_HANDLERS();
	
	/*
	 * Mark all extents as FREED. This will notify any sleepers in the 
	 * strategy routine that the extent won't have data for them.
	 */
	for (cel_idx = 0;
		 cel_idx < BC_cache->c_nextentlists;
		 cel_idx++) {
		for (ce_idx = 0;
			 ce_idx < BC_cache->c_nextents[cel_idx];
			 ce_idx++) {
			struct BC_cache_extent *ce = BC_cache->c_extentlists[cel_idx] + ce_idx;
			LOCK_EXTENT(ce);		
			/* 
			 * Track unused bytes 
			 */
			if (ce->ce_blockmap != NULL && BC_cache->c_mounts[ce->ce_mount_idx].cm_blocksize != 0) {
				for (j = 0; j < howmany(ce->ce_length, BC_cache->c_mounts[ce->ce_mount_idx].cm_blocksize); j++) {
					if (CB_BLOCK_PRESENT(ce, j))
						BC_ADD_STAT(spurious_bytes, BC_cache->c_mounts[ce->ce_mount_idx].cm_blocksize);
				}
			}
			
			BC_teardown_extent(ce);
			UNLOCK_EXTENT(ce);
			wakeup(ce);
		}
	}
	
	for (cm_idx = 0; cm_idx < BC_cache->c_nmounts; cm_idx++) {
		struct BC_cache_mount *cm = BC_cache->c_mounts + cm_idx;
		LOCK_MOUNT_W(cm);
		BC_teardown_mount(cm);
		UNLOCK_MOUNT_W(cm);
	}
	
	
	/*
	 * It is possible that one or more callers are asleep in the
	 * strategy routine (eg. if readahead was terminated early,
	 * or if we are called off the timeout).
	 * Check the count of callers in the strategy code, and sleep
	 * until there are none left (or we time out here).  Note that
	 * by clearing BC_FLAG_CACHEACTIVE above we prevent any new
	 * strategy callers from touching the cache, so the count
	 * must eventually drain to zero.
	 */
	retry = 0;
	while (BC_cache->c_strategycalls > 0) {
		tsleep(BC_cache, PRIBIO, "BC_terminate_cache", 10);
		if (retry++ > 50) {
			message("could not terminate cache, timed out with %d caller%s in BC_strategy",
					(int) BC_cache->c_strategycalls,
					BC_cache->c_strategycalls == 1 ? "" : "s");
			UNLOCK_CACHE_R();
			return(EBUSY);	/* again really EWEDGED */
		}
	}
	
	if (! LOCK_CACHE_R_TO_W()) {
		/* We shouldn't get here. This is the only LOCK_CACHE_R_TO_W call,
		 * so this implies someone is terminating the cache in parallel with us,
		 * but we check for exclusivity by clearing BC_FLAG_CACHEACTIVE.
		 */
		message("Unable to upgrade cache lock to free resources");
		return ENXIO;
	}
	
	for (cm_idx = 0; cm_idx < BC_cache->c_nmounts; cm_idx++) {
		struct BC_cache_mount *cm = BC_cache->c_mounts + cm_idx;
		if (cm->cm_disk != NULL) {
			for (j = cm_idx + 1; j < BC_cache->c_nmounts; j++) {
				if (BC_cache->c_mounts[j].cm_disk == cm->cm_disk) {
					BC_cache->c_mounts[j].cm_disk = NULL;
				}
			}
			BC_teardown_disk(cm->cm_disk);
			lck_mtx_destroy(&cm->cm_disk->cd_lock, BC_cache->c_lckgrp);
			_FREE_ZERO(cm->cm_disk, M_TEMP);
		}
	}	
	
	BC_free_pagebuffer();
	BC_cache->c_cachesize = 0;
	
	/* free memory held by extents and mounts */
	for (cel_idx = 0;
		 cel_idx < BC_cache->c_nextentlists;
		 cel_idx++) {
		for (ce_idx = 0;
			 ce_idx < BC_cache->c_nextents[cel_idx];
			 ce_idx++) {
			lck_mtx_destroy(&BC_cache->c_extentlists[cel_idx][ce_idx].ce_lock, BC_cache->c_lckgrp);
		}
		_FREE_ZERO(BC_cache->c_extentlists[cel_idx], M_TEMP);
	}
	_FREE_ZERO(BC_cache->c_extentlists, M_TEMP);
	_FREE_ZERO(BC_cache->c_nextents, M_TEMP);
	BC_cache->c_nextentlists = 0;

	for (cm_idx = 0; cm_idx < BC_cache->c_nmounts; cm_idx++) {
		lck_rw_destroy(&BC_cache->c_mounts[cm_idx].cm_lock, BC_cache->c_lckgrp);
	}
	_FREE_ZERO(BC_cache->c_mounts, M_TEMP);
	BC_cache->c_nmounts = 0;
	
	UNLOCK_CACHE_W();
	
	/* record stop time */
	struct timeval endtime;
	microtime(&endtime);
	timersub(&endtime,
			 &BC_cache->c_cache_starttime,
			 &endtime);
	BC_cache->c_stats.ss_cache_time += (u_int) endtime.tv_sec * 1000 + (u_int) endtime.tv_usec / 1000;
	return(0);
}

/*
 * Terminate history recording.
 *
 * This stops us recording any further history events.
 */
static int
BC_terminate_history(void)
{
	struct BC_history_mount_device  *hmd;
	LOCK_HANDLERS();
	if (!BC_clear_flag(BC_FLAG_HISTORYACTIVE)) { 
		/* can't shut down if we're not active */
		debug("cannot terminate history when not active");
		UNLOCK_HANDLERS();
		return(ENXIO);
	}
	
	debug("terminating history collection...");
	
	/* if the cache is no longer active also, disconnect our strategy routine */
	BC_check_handlers();
	UNLOCK_HANDLERS();
	
	/*
	 * Kill the timeout handler; we don't want it going off
	 * now.
	 */
	untimeout(BC_timeout_history, NULL);	
	
	if (BC_cache->c_take_detailed_stats) {
		BC_cache->c_stats.ss_history_mount_no_uuid = 0;
		for (hmd = BC_cache->c_history_mounts; hmd != NULL; hmd = hmd->hmd_next)
			if (uuid_is_null(hmd->hmd_mount.hm_uuid))
				BC_ADD_STAT(history_mount_no_uuid, 1);
	}
	
	/* record stop time */
	if (BC_cache->c_take_detailed_stats) {
		struct timeval endtime;
		/* record stop time */
		microtime(&endtime);
		timersub(&endtime,
				 &BC_cache->c_history_starttime,
				 &endtime);
		BC_ADD_STAT(history_time, (u_int) endtime.tv_sec * 1000 + (u_int) endtime.tv_usec / 1000);
	}
	
	BC_cache->c_take_detailed_stats = 0;

	return(0);
}

/*
 * Check our strategy handlers and make sure they are set/cleared depending on our current state.
 *
 * Should be called after changing BC_FLAG_CACHEACTIVE or BC_FLAG_HISTORYACTIVE.
 * Called with the handlers lock held.
 */
static void
BC_check_handlers(void)
{	
	if (BC_cache->c_devsw == NULL ||
		BC_cache->c_strategy == NULL ||
		BC_cache->c_close == NULL) {
		debug("Cannot check device handlers: cache not setup properly");
		return;
	}
	
	/* if the cache or history recording is active, make sure we've captured the appropriate device handler routines */
	if ((BC_cache->c_flags & BC_FLAG_CACHEACTIVE) ||
		(BC_cache->c_flags & BC_FLAG_HISTORYACTIVE)) {
		
		if (BC_cache->c_devsw->d_strategy != (strategy_fcn_t*) BC_strategy ||
			BC_cache->c_devsw->d_close    != BC_close) {
			
			debug("connecting handlers...");
			
			/* connect the strategy and close routines */
			BC_cache->c_devsw->d_strategy = (strategy_fcn_t*) BC_strategy;
			BC_cache->c_devsw->d_close    = BC_close;
		}
	} else {
		if (BC_cache->c_devsw->d_strategy != BC_cache->c_strategy ||
			BC_cache->c_devsw->d_close    != BC_cache->c_close) {
			
			debug("disconnecting handlers...");
			
			/* disconnect the strategy and close routines */
			BC_cache->c_devsw->d_strategy = BC_cache->c_strategy;
			BC_cache->c_devsw->d_close    = BC_cache->c_close;
		}
	}
}

/*
 * Setup a cache extent with the parameters given by the playlist entry
 *
 * Returns 0 on success
 */
static int 
BC_setup_extent(struct BC_cache_extent *ce, const struct BC_playlist_entry *pe, int batch_adjustment, int cache_mount_idx)
{	
	lck_mtx_init(&ce->ce_lock, BC_cache->c_lckgrp,
				 LCK_ATTR_NULL);
	ce->ce_diskoffset = pe->pe_offset;
	ce->ce_mount_idx = cache_mount_idx;
	ce->ce_length = pe->pe_length;
#ifdef IGNORE_BATCH
	ce->ce_batch = 0;
#else
	ce->ce_batch = pe->pe_batch;
#endif
	ce->ce_batch += batch_adjustment;
	if (ce->ce_batch > BC_MAXBATCHES) {
		ce->ce_batch = BC_MAXBATCHES; /* cap batch number */
	}
	ce->ce_cacheoffset = 0;
	ce->ce_blockmap = NULL;
	ce->ce_flags = 0;
	if (pe->pe_flags & BC_PE_LOWPRIORITY) {
		ce->ce_flags |= CE_LOWPRI;
	}
	if (pe->pe_flags & BC_PE_SHARED) {
		ce->ce_flags |= CE_SHARED;
	}

	/* track highest batch number for this playlist */
	if (ce->ce_batch >= BC_cache->c_batch_count) {
		BC_cache->c_batch_count = ce->ce_batch + 1;
		// debug("Largest batch is now %d from extent with disk offset %llu", BC_cache->c_batch_count, ce->ce_diskoffset);
	}
	
	return 0;
}

/*
 * The blocksize is initialised from the first playlist read, the statistics
 * structure, or it can be pre-set by the caller.  Once set, only playlists with
 * matching sizes can be read/written.
 */
#define	BLOCK_ROUNDDOWN(cm, x)	(((x) / (cm)->cm_blocksize) * (cm)->cm_blocksize)
#define BLOCK_ROUNDUP(cm, x)	((((x) + ((cm)->cm_blocksize - 1)) / (cm)->cm_blocksize) * (cm)->cm_blocksize)
/*
 * optional for power-of-two blocksize roundup:
 * (((x) + ((cm)->cm_blocksize - 1)) & (~((cm)->cm_blocksize - 1)))
 */

/*
 * Fill in a cache extent now that its mount has been filled in
 * 
 * Called with the extent lock held
 *
 * Returns 0 on success
 */
static int 
BC_fill_in_extent(struct BC_cache_extent *ce)
{
	int numblocks, roundsize, i;
	u_int64_t end;
	
	if (ce->ce_flags & CE_ABORTED ||
		ce->ce_length == 0) {
		return 1;
	}
	
	struct BC_cache_mount *cm = BC_cache->c_mounts + ce->ce_mount_idx;
	
	int blocksize = cm->cm_blocksize;
	
	if (0 == blocksize) {
		return 1;
	}
	
	roundsize = roundup(ce->ce_length, PAGE_SIZE);
	
	/* make sure we're on block boundaries */
	end = ce->ce_diskoffset + roundsize;
	ce->ce_diskoffset = BLOCK_ROUNDDOWN(cm, ce->ce_diskoffset);
	ce->ce_length = BLOCK_ROUNDUP(cm, end) - ce->ce_diskoffset;
	
	/* make sure we didn't grow our pagebuffer size since is's already been allocated */
	if (roundup(ce->ce_length, PAGE_SIZE) > roundsize) {
        debug("Clipped extent %llu by a page", ce->ce_diskoffset);
		BC_ADD_STAT(extents_clipped, 1);
		ce->ce_length = roundsize;
	}
	
	numblocks = howmany(ce->ce_length, blocksize);
	
	ce->ce_blockmap = BC_MALLOC(howmany(numblocks, (CB_MAPFIELDBITS / CB_MAPFIELDBYTES)),
							  M_TEMP, M_WAITOK | M_ZERO);
	if (!ce->ce_blockmap) {
		message("can't allocate extent blockmap for %d blocks of %d bytes", numblocks, howmany(numblocks, (CB_MAPFIELDBITS / CB_MAPFIELDBYTES)));
		return 1;
	}
	
	for (i = 0; i < howmany(ce->ce_length, blocksize); i++) {
		CB_MARK_BLOCK_PRESENT((ce), i);
	}
	
	return 0;
}

static void
BC_teardown_extent(struct BC_cache_extent *ce)
{
	ce->ce_flags |= CE_ABORTED;
	_FREE_ZERO(ce->ce_blockmap, M_TEMP);
}

/*
 * Setup a cache disk.
 * Returns 0 on success
 */
static int
BC_setup_disk(struct BC_cache_disk *cd, u_int64_t disk_id, int is_ssd)
{
	static int next_disk_num;
	lck_mtx_init(&cd->cd_lock, BC_cache->c_lckgrp, LCK_ATTR_NULL);
	cd->cd_disk_id = disk_id;
	cd->cd_disk_num = next_disk_num++;
	cd->cd_flags = 0;
	cd->cd_nmounts = 0;
	cd->cd_batch = 0;

	if (is_ssd) {
		cd->cd_flags |= CD_IS_SSD;
	}
	
	debug("Setup disk 0x%llx as disk num %d%s", disk_id, cd->cd_disk_num, (cd->cd_flags & CD_IS_SSD) ? " (ssd)" : "");
	
	return (0);
}

/*
 * Teardown a cache disk.
 */
static void
BC_teardown_disk(struct BC_cache_disk *cd)
{
	/* Nothing to do */
}


/*
 * Check if a mount has become available for readahead
 *
 * If so, make sure the reader thread picks up this mount's IOs.
 * If there is no reader thread for the mount's disk, kick one off.
 *
 * Called with the cache mount read lock held
 */
static void
BC_mount_available(struct BC_cache_mount *cm)
{
#ifdef EMULATE_ONLY
	int i;
	for (i = 0; i < cm->cm_nextents; i++)
			cm->cm_pextents[i]->ce_flags |= CE_IODONE;
#else
	int i, error;
	thread_t rthread;
	if (cm->cm_state != CM_READY) {
		/* not ready */
		return;
	}
	
	struct BC_cache_disk *cd = cm->cm_disk;
	LOCK_DISK(cd);
	cd->cd_nmounts++;
	LOCK_READERS();
	if (!(BC_cache->c_flags & BC_FLAG_SHUTDOWN)) {
		if (cd->cd_flags & CD_HAS_THREAD) {
			/* Make sure the thread is not issuing lowpriority IO */
			if (cd->cd_flags & CD_ISSUE_LOWPRI) {
				debug("Interrupting low-priority thread for new mount %s", uuid_string(cm->cm_uuid));
				cd->cd_flags &= (~CD_ISSUE_LOWPRI);
				/* TODO: Unthrottle the readahead thread here rather than waiting out its current throttle delay */
			}
			UNLOCK_READERS();
			UNLOCK_DISK(cd);
			return;
		}
		
		/* Make sure we issue regular IOs for the new playlist in case we were issuing low-priority previously */
		cd->cd_flags &= (~CD_ISSUE_LOWPRI);

		debug("Kicking off reader thread for disk %d for mount %s", cd->cd_disk_num, uuid_string(cm->cm_uuid));
		if ((error = kernel_thread_start(BC_reader_thread, cd, &rthread)) == KERN_SUCCESS) {
			thread_deallocate(rthread);
			BC_cache->c_num_reader_threads++;
			BC_ADD_STAT(readahead_threads, 1);
			cd->cd_flags |= CD_HAS_THREAD;
			UNLOCK_READERS();
			UNLOCK_DISK(cd);
			return;
		}
		
		message("Unable to start reader thread for disk %d: %d", cd->cd_disk_num, error);
	}
	UNLOCK_READERS();
	UNLOCK_DISK(cd);
	
	/*
	 * Getting here indicates some failure.
	 *
	 * Mark all extents as aborted. This will notify any sleepers in the 
	 * strategy routine that the extent won't have data for them.
	 */
	for (i = 0; i < cm->cm_nextents; i++) {
		LOCK_EXTENT(cm->cm_pextents[i]);
		BC_teardown_extent(cm->cm_pextents[i]);
		UNLOCK_EXTENT(cm->cm_pextents[i]);
		wakeup(cm->cm_pextents[i]);
	}
#endif
}

/*
 * Setup a cache mount from the playlist mount.
 *
 * Allocates cm_pextents large enough to hold pm->pm_nentries extents,
 * but leaves cm_nextents 0 since the array hasn't been initialized.
 *
 * Returns 0 on success
 */
static int
BC_setup_mount(struct BC_cache_mount *cm, struct BC_playlist_mount* pm)
{
	int error = 0;
	
	lck_rw_init(&cm->cm_lock, BC_cache->c_lckgrp, LCK_ATTR_NULL);
	uuid_copy(cm->cm_uuid, pm->pm_uuid);
	
	/* These will be set once we've detected that this volume has been mounted */
	cm->cm_dev = nulldev();
	cm->cm_bp = NULL;
	cm->cm_devsize = 0;
	cm->cm_maxread = 0;
	cm->cm_disk = NULL;
	cm->cm_blocksize = 0;
	cm->cm_nextents = 0;
	
	/* Allocate the sorted extent array.
	 * We'll fill it in while we setup the extents.
	 */
	if (pm->pm_nentries == 0) {
		message("Playlist incuded mount with 0 entries");
		error = EINVAL;
		goto out;
	}
	
	cm->cm_pextents = BC_MALLOC(pm->pm_nentries * sizeof(*cm->cm_pextents), M_TEMP, M_WAITOK | M_ZERO);
	if (cm->cm_pextents == NULL) {
		message("can't allocate mount's extent array (%d entries)", pm->pm_nentries);
		error = ENOMEM;
		goto out;
	}
	
	cm->cm_state = CM_SETUP;
	
out:
	if (error) {
		cm->cm_state = CM_ABORTED;
		_FREE_ZERO(cm->cm_pextents, M_TEMP);
		lck_rw_destroy(&cm->cm_lock, BC_cache->c_lckgrp);
	}
	return error;
}

/*
 * Fill in the rest of the cache mount given the matching mount structure.
 *
 * Called with the cache mount write lock held
 *
 * Returns 0 if the mount was successfully filled in and cm_state will be CM_READY
 * Returns non-0 on failure or it the mount was already filled in
 */
static int
BC_fill_in_mount(struct BC_cache_mount *cm, mount_t mount, vfs_context_t context)
{
	u_int64_t blkcount;
	u_int32_t blksize;
	int error, mount_idx, i, is_ssd, max_byte_count;
	u_int64_t disk_id;
	struct BC_cache_disk *cd;
	vnode_t devvp = NULLVP;
	
	error = 0;
	
	if (CM_SETUP != cm->cm_state) {
		return EINVAL;
	}
	
	cm->cm_throttle_mask = vfs_throttle_mask(mount);
	disk_id = cm->cm_throttle_mask; /* using the throttle mask as a way to identify the physical disk */
	/* debug("Got throttle mask %llx for mount %s", cm->cm_throttle_mask, uuid_string(cm->cm_uuid)); */
	
	devvp = vfs_devvp(mount);
	if (devvp == NULLVP) {
		message("mount %s does not have a vnode", uuid_string(cm->cm_uuid));
		error = EINVAL;
		goto out;
	}
	
#ifdef ROOT_DISK_ONLY
	if (devvp == rootvp) {
		BC_cache->c_root_disk_id = disk_id;
		assert(BC_cache->c_root_disk_id != 0);
	} else if (0 == BC_cache->c_root_disk_id) {
		error = EAGAIN; /* we haven't seen the root mount yet, try again later */
		goto out;
	} else if (BC_cache->c_root_disk_id != disk_id) {
		debug("mount %s (disk 0x%llx) is not on the root disk (disk 0x%llx)", uuid_string(cm->cm_uuid), disk_id, BC_cache->c_root_disk_id);
		error = ENODEV;
		goto out;
	}
#endif
	
	/* See if we already have a cache_disk for this disk */
	for (mount_idx = 0; mount_idx < BC_cache->c_nmounts; mount_idx++) {
		if (BC_cache->c_mounts + mount_idx == cm) continue;
		
		cd = BC_cache->c_mounts[mount_idx].cm_disk;
		if (cd && cd->cd_disk_id == disk_id) {
			cm->cm_disk = cd;
			break;
		}
	}
	
	/* If we don't already have a cache_disk for this disk, allocate one */
	if (cm->cm_disk == NULL) {
		cd = BC_MALLOC(sizeof(*cd), M_TEMP, M_WAITOK | M_ZERO);
		if (cd == NULL) {
			message("can't allocate memory for cache disk");
			error = ENOMEM;
			goto out;
		}
		
		if (VNOP_IOCTL(devvp,       /* vnode */
					   DKIOCISSOLIDSTATE,    /* cmd */
					   (caddr_t)&is_ssd, /* data */
					   0,
					   context))           /* context */
		{
			message("can't determine if disk is a solid state disk for mount %s", uuid_string(cm->cm_uuid));
			is_ssd = 0;
		}
		
		if ((error = BC_setup_disk(cd, disk_id, is_ssd)) != 0) {
			_FREE_ZERO(cd, M_TEMP);
			message("cache disk setup failed: %d", error);
			goto out;
		}
		cm->cm_disk = cd;
	}
	
#ifdef DONT_CACHE_SSDS
	if (cm->cm_disk->cd_flags & CD_IS_SSD) {
		debug("mount %s is on an SSD (disk 0x%llx)", uuid_string(cm->cm_uuid), cm->cm_disk->cd_disk_id);
		error = ENODEV;
		goto out;
	}
#endif
	
	if (VNOP_IOCTL(devvp,
				   DKIOCGETBLOCKCOUNT,
				   (caddr_t)&blkcount,
				   0,
				   context)
		|| blkcount == 0)
	{
		message("can't determine device size, not checking");
		blkcount = 0;
	}
	
	if (VNOP_IOCTL(devvp,
				   DKIOCGETBLOCKSIZE,
				   (caddr_t)&blksize,
				   0,
				   context)
		|| blksize == 0)
	{
		message("can't determine device block size for mount %s, defaulting to 512 bytes", uuid_string(cm->cm_uuid));
		blksize = 512;
	}
	
	if (PAGE_SIZE % blksize != 0) {
		message("PAGE_SIZE (%d) is not a multiple of block size (%d) for mount %s", PAGE_SIZE, blksize, uuid_string(cm->cm_uuid));
		error = EINVAL;
		goto out;
	}
	
	cm->cm_blocksize = blksize;
	cm->cm_devsize = blksize * blkcount;
	
    VNOP_IOCTL(devvp,
			   DKIOCGETMAXSEGMENTBYTECOUNTREAD,
			   (caddr_t)&cm->cm_maxread,
			   0,
			   context);
	if (0 == VNOP_IOCTL(devvp,
						DKIOCGETMAXBYTECOUNTREAD,
						(caddr_t)&max_byte_count,
						0,
						context))
	{
		if (cm->cm_maxread > max_byte_count && max_byte_count > 0) {
			cm->cm_maxread = max_byte_count;
		}
	}
	if (cm->cm_maxread > ubc_upl_maxbufsize()) {
		/* max read size isn't larger than the max UPL size */
		cm->cm_maxread = ubc_upl_maxbufsize();
	}
	if (cm->cm_maxread < MAXPHYS)
	{
		debug("can't determine device read size for mount %s; using default", uuid_string(cm->cm_uuid));
		cm->cm_maxread = MAXPHYS;
	}
    if (cm->cm_maxread % cm->cm_blocksize != 0) {
        debug("Mount max IO size (%llu) not a multiple of block size (%llu)", cm->cm_maxread, cm->cm_blocksize);
        cm->cm_maxread -= cm->cm_maxread % cm->cm_blocksize;
    }
    
	cm->cm_bp = buf_alloc(devvp);
	if (cm->cm_bp == NULL) {
		message("can't allocate buf");
		error = ENOMEM;
		goto out;
	}
	
	cm->cm_dev = vnode_specrdev(devvp);
	if (cm->cm_dev == nulldev()) {
		message("mount %s does not have a device", uuid_string(cm->cm_uuid));
		error = EINVAL;
		goto out;
	}
	
	for (i = 0; i < cm->cm_nextents; i++) {
		LOCK_EXTENT(cm->cm_pextents[i]);
		if (0 != BC_fill_in_extent(cm->cm_pextents[i])) {
			BC_teardown_extent(cm->cm_pextents[i]);
			UNLOCK_EXTENT(cm->cm_pextents[i]);
			wakeup(cm->cm_pextents[i]);
		} else {
			UNLOCK_EXTENT(cm->cm_pextents[i]);
		}
	}
		
	cm->cm_state = CM_READY;
	
	debug("Found new cache mount %s disk %d, %d extents", uuid_string(cm->cm_uuid), cm->cm_disk->cd_disk_num, cm->cm_nextents);
	
out:
	if (error && error != EAGAIN) {
		/*
		 * Mark all extents as aborted. This will notify any sleepers in the 
		 * strategy routine that the extent won't have data for them.
		 *
		 * Note that its fine to take the extent lock here since no one will
		 * have extent lock and try to take the mount lock if the mount isn't CM_READY
		 */
		for (i = 0; i < cm->cm_nextents; i++) {
			LOCK_EXTENT(cm->cm_pextents[i]);
			BC_teardown_extent(cm->cm_pextents[i]);
			UNLOCK_EXTENT(cm->cm_pextents[i]);
			wakeup(cm->cm_pextents[i]);
		}						
		
		BC_teardown_mount(cm);
	}
	if (devvp != NULLVP) {
		vnode_put(devvp);
	}
	return error;
}

/*
 * Called with the cache mount write lock held
 */
static void
BC_teardown_mount(struct BC_cache_mount *cm)
{
	if (cm->cm_bp) {
		buf_free(cm->cm_bp);
		cm->cm_bp = NULL;
	}
	cm->cm_nextents = 0;
	_FREE_ZERO(cm->cm_pextents, M_TEMP);
	cm->cm_state = CM_ABORTED;
}

/*
 * Check to see which mounts have been mounted and finish setting them up.
 *
 * No extent/mount locks required since we have a write lock on the cache
 *
 * Called with the cache write lock held
 */
static int fill_in_bc_cache_mounts(mount_t mount, void* arg)
{
	int mount_idx, error, i;
	struct vfs_attr attr;
	vfs_context_t context;
	int* go_again = (int*) arg;
	
	VFSATTR_INIT(&attr);
	VFSATTR_WANTED(&attr, f_uuid);
	context = vfs_context_create(NULL);
	error = vfs_getattr(mount, &attr, context);
	if ((0 != error) || (! VFSATTR_IS_SUPPORTED(&attr, f_uuid))) {
#ifdef BC_DEBUG
		char name[MFSNAMELEN];
		vfs_name(mount, name);
		if (strncmp("devfs", name, sizeof("devfs")) != 0) {
			debug("Found mount %s for IO without a UUID: error %d", name, error);
		}
#endif
		vfs_context_rele(context);
		return VFS_RETURNED;
	} else {
		// debug("Found mount for IO with UUID %s", uuid_string(attr.f_uuid));
		
		for (mount_idx = 0; mount_idx < BC_cache->c_nmounts; mount_idx++) {
			struct BC_cache_mount *cm = BC_cache->c_mounts + mount_idx;
			int match = 0;
			if (CM_SETUP == cm->cm_state) {
				
				if (0 == uuid_compare(cm->cm_uuid, attr.f_uuid)) {
					match = 1;
				}
				
				/* a null uuid indicates we want to match the root volume, no matter what the uuid (8350414) */
				if ((!match) && uuid_is_null(cm->cm_uuid)) {
					vnode_t devvp = vfs_devvp(mount);
					if (vnode_specrdev(devvp) == rootdev) {
						uuid_copy(cm->cm_uuid, attr.f_uuid);
						match = 1;
						debug("Found root mount %s", uuid_string(cm->cm_uuid));
					}
					vnode_put(devvp);
				}
				
				if (match) {
					/* Found a matching mount */
					
					/* Locking here isn't necessary since we're only called while holding the cache write lock
					 * and no one holds the mount locks without also holding the cache lock
					 */
					if (BC_fill_in_mount(cm, mount, context) == EAGAIN) {
						*go_again = 1;
					}
					vfs_context_rele(context);
					
					/* Check to see if we have any more mounts to fill in */
					for (i = 0; i < BC_cache->c_nmounts; i++) {
						if (CM_SETUP == BC_cache->c_mounts[i].cm_state) {
							/* have more to fill in, keep iterating */
							return VFS_RETURNED;
						}
					}
					
					return VFS_RETURNED_DONE;
				}
			}
		}
	} 
	
	vfs_context_rele(context);

	/* Check to see if we have any more mounts to fill in */
	for (i = 0; i < BC_cache->c_nmounts; i++) {
		if (CM_SETUP == BC_cache->c_mounts[i].cm_state) {
			/* have more to fill in, keep iterating */
			return VFS_RETURNED;
		}
	}
	
	debug("No new mounts filled in fill_in_bc_cache_mounts");
	return VFS_RETURNED_DONE;
	
}

#ifndef BOOTCACHE_ENTRIES_SORTED_BY_DISK_OFFSET
/*
 * Sort the extent list.
 */
static int
compare_cache_extents(const void *vfirst, const void *vsecond)
{
	const struct BC_cache_extent *first, *second;
	
	first  = (const struct BC_cache_extent *)vfirst;
	second = (const struct BC_cache_extent *)vsecond;
	
	// Sort by volume first, then by logical block address
	int mount_comparison = first->ce_mount_idx - second->ce_mount_idx;
	if (mount_comparison != 0)
		return((mount_comparison < 0) ? -1 : 1);
	
	if (first->ce_diskoffset == second->ce_diskoffset)
		return(0);
	return((first->ce_diskoffset < second->ce_diskoffset) ? -1 : 1);
}
#endif

/*
 * Fetch the playlist from userspace or the Mach-O segment where it
 * was preloaded. Add it to the cache and readahead in batches after
 * the current cache, if any.
 *
 * Returns 0 on success. In failure cases, the cache is not modified.
 *
 * Called with the cache write lock held.
 */
static int
BC_copyin_playlist(size_t mounts_size, user_addr_t mounts_buf, size_t entries_size, user_addr_t entries_buf)
{
	int nplaylist_mounts, nplaylist_entries;
	struct BC_playlist_mount *playlist_mounts = NULL, *pm;
	struct BC_playlist_entry *playlist_entries = NULL, *pe;
	struct BC_cache_mount  *cm, *old_cm;
	struct BC_cache_extent *ce;
	
	int ncache_mounts = 0, ncache_extents = 0, nnew_mounts = 0;
	struct BC_cache_mount  *cache_mounts = NULL;
	struct BC_cache_extent *cache_extents = NULL;
	int *pmidx_to_cmidx = NULL;
	int *next_old_extent_idx = NULL;
	
	int *cache_nextents = NULL;
	struct BC_cache_extent **cache_extents_list = NULL;
	
	int old_batch_count;
	u_int64_t old_cache_size;

	int error, pm_idx, cm_idx, pe_idx, ce_idx, cel_idx, pm_idx2, max_entries;
	off_t p;
	u_int64_t size;
	int actual, remaining_entries;
	
	int clipped_extents = 0, enveloped_extents = 0; 
	
	playlist_mounts = NULL;
	playlist_entries = NULL;
	
#ifdef BC_DEBUG
	struct timeval start_time;
	microtime(&start_time);
#endif
	
	old_batch_count = BC_cache->c_batch_count;
	old_cache_size = BC_cache->c_cachesize;
	
	nplaylist_mounts = (int) mounts_size / sizeof(*playlist_mounts);
	nplaylist_entries = (int) entries_size / sizeof(*playlist_entries);

	if (BC_cache->c_stats.ss_total_extents > 0) {
		debug("adding %u extents to existing cache of %u extents", nplaylist_entries, BC_cache->c_stats.ss_total_extents);
	} else {
		debug("setting up first cache of %d extents", nplaylist_entries);
	}
	
	for (cm_idx = 0; cm_idx < BC_cache->c_nmounts; cm_idx++) {
		debug("Mount %s has %d extents", uuid_string(BC_cache->c_mounts[cm_idx].cm_uuid), BC_cache->c_mounts[cm_idx].cm_nextents);
	}

	/*
	 * Mount array
	 */
	if (BC_preloaded_playlist) {
		playlist_mounts = CAST_DOWN(struct BC_playlist_mount *, mounts_buf);
	} else {
		playlist_mounts = BC_MALLOC(mounts_size, M_TEMP, M_WAITOK);
		if (playlist_mounts == NULL) {
			message("can't allocate unpack mount buffer of %ld bytes", mounts_size);
			error = ENOMEM;
			goto out;
		}
		if ((error = copyin(mounts_buf, playlist_mounts, mounts_size)) != 0) {
			message("copyin %ld bytes from 0x%llx to %p failed", mounts_size, mounts_buf, playlist_mounts);
			goto out;
		}
		
		/* if BC_preloaded_playlist is non-NULL, playlist_mounts must be freed */
	}
	
	/* map from the playlist mount index to the corresponding cache mount index */
	pmidx_to_cmidx = BC_MALLOC(nplaylist_mounts * sizeof(*pmidx_to_cmidx), M_TEMP, M_WAITOK);
	if (pmidx_to_cmidx == NULL) {
		message("can't allocate playlist index to cache index reference array for %d mounts", nplaylist_mounts);
		error = ENOMEM;
		goto out;
	}
	
	/* In order to merge the mount's current (sorted) extent list with the new (sorted)
	 * extent list from the new playlist we first grow the extent list allocation to fit
	 * both arrays, then merge the two arrays into the new allocation.
	 */
	
	if (BC_cache->c_nmounts > 0) {
		/* next_old_extent_idx saves the index of the next unmerged extent in the original mount's extent list */
		next_old_extent_idx = BC_MALLOC(BC_cache->c_nmounts * sizeof(*next_old_extent_idx), M_TEMP, M_WAITOK | M_ZERO);
		if (next_old_extent_idx == NULL) {
			message("can't allocate index array for %d mounts", BC_cache->c_nmounts);
			error = ENOMEM;
			goto out;
		}
		
		/* 0-filled since all the index start at 0 */
	}
	
	nnew_mounts = nplaylist_mounts;
	
	/* determine how many mounts are new and how many we already have */
	for (pm_idx = 0; pm_idx < nplaylist_mounts; pm_idx++) {
		pmidx_to_cmidx[pm_idx] = -1; /* invalid by default */
		
		for (cm_idx = 0; cm_idx < BC_cache->c_nmounts; cm_idx++) {
			if (0 == uuid_compare(BC_cache->c_mounts[cm_idx].cm_uuid, playlist_mounts[pm_idx].pm_uuid)) {
				/* already have a record of this mount, won't need a new spot for it */
				pmidx_to_cmidx[pm_idx] = cm_idx;
				nnew_mounts--;
				break;
			}
		}
		
		/* check to make sure there aren't any duplicate mounts within the playlist */
		for (pm_idx2 = pm_idx + 1; pm_idx2 < nplaylist_mounts; pm_idx2++) {
			if (0 == uuid_compare(playlist_mounts[pm_idx2].pm_uuid, playlist_mounts[pm_idx].pm_uuid)) {
				message("Playlist had duplicate mount %s", uuid_string(playlist_mounts[pm_idx2].pm_uuid));
				error = EINVAL;
				goto out;
			}
		}
	}
	
	/* cache_mounts is the array of mounts that will replace the one currently in the cache */
	cache_mounts = BC_MALLOC((BC_cache->c_nmounts + nnew_mounts) * sizeof(*cache_mounts), M_TEMP, M_WAITOK | M_ZERO);
	if (cache_mounts == NULL) {
		message("can't allocate memory for %d mounts", (BC_cache->c_nmounts + nnew_mounts));
		error = ENOMEM;
		goto out;
	}
	memcpy(cache_mounts, BC_cache->c_mounts, (BC_cache->c_nmounts * sizeof(*BC_cache->c_mounts)));
	
	/* ncache_mounts is the current number of valid mounts in the mount array */
	ncache_mounts = BC_cache->c_nmounts;
	
	for (pm_idx = 0; pm_idx < nplaylist_mounts; pm_idx++) {
		if (pmidx_to_cmidx[pm_idx] != -1) {
			/* We already have a record for this mount */
			
			if (playlist_mounts[pm_idx].pm_nentries == 0) {
				message("Playlist incuded mount with 0 entries");
				error = EINVAL;
				goto out;
			}

			cm_idx = pmidx_to_cmidx[pm_idx];
			
			assert(cm_idx < BC_cache->c_nmounts);

			/* grow the mount's extent array to fit the new extents (we'll fill in the new extents below) */
			cache_mounts[cm_idx].cm_pextents = BC_MALLOC((BC_cache->c_mounts[cm_idx].cm_nextents + playlist_mounts[pm_idx].pm_nentries) * sizeof(*cache_mounts[cm_idx].cm_pextents), M_TEMP, M_WAITOK | M_ZERO);
			if (cache_mounts[cm_idx].cm_pextents == NULL) {
				message("can't allocate mount's extent array (%d entries)", (BC_cache->c_mounts[cm_idx].cm_nextents + playlist_mounts[pm_idx].pm_nentries));
				error = ENOMEM;
				goto out;
			}

			/* don't free the old extent list yet, the real BC_cache's mount still points to it and we may yet fail */

			/* cm_nextents is the number of valid extents in this mount's extent list. It starts out as 0 and we fill it in below */
			cache_mounts[cm_idx].cm_nextents = 0;
		} else {
			/* New mount for the cache */
			
			if ((error = BC_setup_mount(cache_mounts + ncache_mounts, playlist_mounts + pm_idx)) != 0) {
				goto out;
			}
			
			pmidx_to_cmidx[pm_idx] = ncache_mounts;
			ncache_mounts++;
		}
	}	
	
	/* 
	 * Extent array
	 *
	 * The existing extent array cannot move since a strategy routine may be
	 * in the middle of an msleep with one of the extent locks. So, we allocate
	 * a new array for the new extents.
	 *
	 * Add one extent list to our arrays of extent lists
	 */
		
	/* cache_nextents is an array of ints, each indicating the number of extents in the list in cache_extents_list at the same index */
	cache_nextents = BC_MALLOC((BC_cache->c_nextentlists + 1) * sizeof(*BC_cache->c_nextents), M_TEMP, M_WAITOK | M_ZERO);
	if (cache_nextents == NULL) {
		message("can't allocate memory for %d extent list sizes", nplaylist_entries);
		error = ENOMEM;
		goto out;
	}
	memcpy(cache_nextents, BC_cache->c_nextents, BC_cache->c_nextentlists * sizeof(*BC_cache->c_nextents));
	
	/* cache_extents_list is the array of extent lists. The extent list at the last index is for the new playlist's cache */
	cache_extents_list  = BC_MALLOC((BC_cache->c_nextentlists + 1) * sizeof(*BC_cache->c_extentlists),  M_TEMP, M_WAITOK | M_ZERO);
	if (cache_extents_list == NULL) {
		message("can't allocate memory for %d extent lists", nplaylist_entries);
		error = ENOMEM;
		goto out;
	}
	memcpy(cache_extents_list,  BC_cache->c_extentlists,  BC_cache->c_nextentlists * sizeof(*BC_cache->c_extentlists));
	
	/* The extent list for this new playlist's cache */
	cache_extents = BC_MALLOC(nplaylist_entries * sizeof(*cache_extents), M_TEMP, M_WAITOK | M_ZERO);
	if (cache_extents == NULL) {
		message("can't allocate memory for %d extents", nplaylist_entries);
		error = ENOMEM;
		goto out;
	}
	
	/* TODO: iterate over our history and remove any blocks we've already seen from the new cache */
		
	/* Fill in the new extents.
	 *
	 * We just tack the new playlist onto the end of any cache we currently have as a new batch.
	 *
	 * At this point, we assume that any additional caches should be in additional batches just as
	 * if there was a single recording with a tag separating the new extents. If we did decide to
	 * merge, it would still be hard since that would require reordering the extent list and
	 * BC_strategy assumes that the cache extents never move (see the msleep in wait_for_extent).
	 *
	 * We also don't coalesce the new extents into the old (and lower batch) extents.
	 * This would be a bit of work, we assume we want a new batch anyway, and it's rare that
	 * a new cache comes in while an old cache is still in readahead. So, we don't bother.
	 *
	 * We do clip any overlap with the new cache from the old cache, however.
	 *
	 * We don't clip any overlap with the history from the new extents.
	 * The history list isn't ordered, so we don't have a fast way to compare the ranges. It's also
	 * expected that the overlap is minimal. If we stopped recording history at some point, we don't
	 * have this info anyway. So, we don't bother.
	 */
	
	/* size is the size in bytes of the cache, used to allocate our page buffer */
	size = 0;
		
	if (BC_preloaded_playlist) {
		/*
		 * Unpack the static control entry array into the extent array.
		 */
		playlist_entries = CAST_DOWN(struct BC_playlist_entry *, entries_buf);
	} else {
		/*
		 * Since the playlist control entry structure is not the same as the
		 * extent structure we use, we need to copy the control entries in
		 * and unpack them.
		 */
		playlist_entries = BC_MALLOC(BC_PLC_CHUNK * sizeof(*playlist_entries),
								   M_TEMP, M_WAITOK);
		if (playlist_entries == NULL) {
			message("can't allocate unpack buffer for %d entries", BC_PLC_CHUNK);
			error = ENOMEM;
			goto out;
		}
	}
	
	remaining_entries = nplaylist_entries;
	while (remaining_entries > 0) {
		
		if (BC_preloaded_playlist) {
			actual = remaining_entries;
		} else {
			actual = min(remaining_entries, BC_PLC_CHUNK);
			if ((error = copyin(entries_buf, playlist_entries,
								actual * sizeof(struct BC_playlist_entry))) != 0) {
				message("copyin from 0x%llx to %p failed", entries_buf, playlist_entries);
				goto out;
			}
		}
		
		/* unpack into our array */
		for (pe_idx = 0; pe_idx < actual; pe_idx++) {
			pe = playlist_entries + pe_idx;
			if (pe->pe_length == 0) {
				debug("Bad Playlist: entry %d has 0 length", (nplaylist_entries - remaining_entries) + pe_idx);
				error = EINVAL;
				goto out;
			}
			pm_idx = pe->pe_mount_idx;
			
			if (pm_idx >= nplaylist_mounts || pm_idx < 0) {
				message("Bad playlist: entry %d referenced non-existent mount index %d", (nplaylist_entries - remaining_entries) + pe_idx, pm_idx);
				error = EINVAL;
				goto out;
			}
			
			pm = playlist_mounts + pm_idx;
			cm_idx = pmidx_to_cmidx[pm_idx];
			cm = cache_mounts + cm_idx;
			ce = cache_extents + ncache_extents;
			
			/* The size of the extent list is the number of playlist entries + the number of old cache extents */
			max_entries = pm->pm_nentries + ((cm_idx < BC_cache->c_nmounts) ? BC_cache->c_mounts[cm_idx].cm_nextents : 0);
			
			if (cm->cm_nextents >= max_entries) {
				message("Bad playlist: more entries existed than the mount %s claimed (%d)", uuid_string(pm->pm_uuid), pm->pm_nentries);
				error = EINVAL;
				goto out;
			}			
			
			if ((error = BC_setup_extent(ce, pe, old_batch_count, cm_idx)) != 0) {
				goto out;
			}
			
			/* Merge the new extent with the old extents for this mount. The new extent may get clipped. */
#ifdef BOOTCACHE_ENTRIES_SORTED_BY_DISK_OFFSET
			if (cm_idx < BC_cache->c_nmounts) {
				old_cm = BC_cache->c_mounts + cm_idx;
				/* Copy any lower or equal extents from the existing playlist down to the low end of the array */
				while (next_old_extent_idx[cm_idx] < old_cm->cm_nextents &&
					   old_cm->cm_pextents[next_old_extent_idx[cm_idx]]->ce_diskoffset <= ce->ce_diskoffset) {
					cm->cm_pextents[cm->cm_nextents] = old_cm->cm_pextents[next_old_extent_idx[cm_idx]];
					cm->cm_nextents++;
					next_old_extent_idx[cm_idx]++;
				}
				
				/* check for overlap with the next extent in the old list and clip the new extent */
				if (next_old_extent_idx[cm_idx] < old_cm->cm_nextents) {
					
					/* FIXME: rdar://9153031 If the new extent extends past the end of the old extent, we're losing caching! */
					if (ce->ce_diskoffset + ce->ce_length > (old_cm->cm_pextents[next_old_extent_idx[cm_idx]]->ce_diskoffset + old_cm->cm_pextents[next_old_extent_idx[cm_idx]]->ce_length)) {
						debug("!!! Entry %d (0x%llx:%lld) is getting clipped too much because of previous entry for mount %s (0x%llx:%lld)", pe_idx, ce->ce_diskoffset, ce->ce_length, uuid_string(cm->cm_uuid), old_cm->cm_pextents[next_old_extent_idx[cm_idx]]->ce_diskoffset, old_cm->cm_pextents[next_old_extent_idx[cm_idx]]->ce_length);
					}
					
					u_int64_t max_offset = old_cm->cm_pextents[next_old_extent_idx[cm_idx]]->ce_diskoffset;
					if (max_offset < ce->ce_diskoffset + ce->ce_length) {
						if (max_offset <= ce->ce_diskoffset) {
							ce->ce_length = 0;
						} else {
							ce->ce_length = (max_offset - ce->ce_diskoffset);
						}
					}
				}
			}
			
			/* check for intersection with the next lower extent in the list and clip the new extent */
			if (cm->cm_nextents > 0) {
				u_int64_t min_offset = cm->cm_pextents[cm->cm_nextents - 1]->ce_diskoffset + cm->cm_pextents[cm->cm_nextents - 1]->ce_length;
				if (min_offset > ce->ce_diskoffset) {
					
					/* Check if this extent is overlapping with an extent from the same playlist */
					if (cm->cm_pextents[cm->cm_nextents - 1] >= cache_extents &&
						cm->cm_pextents[cm->cm_nextents - 1] < cache_extents + ncache_extents) {
						message("Bad playlist: entry %d (0x%llx:%lld) overlapped with previous entry for mount %s (0x%llx:%lld)", pe_idx, ce->ce_diskoffset, ce->ce_length, uuid_string(cm->cm_uuid), cm->cm_pextents[cm->cm_nextents - 1]->ce_diskoffset, cm->cm_pextents[cm->cm_nextents - 1]->ce_length);
						error = EINVAL;
						goto out;
					}
					
					if (min_offset >= ce->ce_diskoffset + ce->ce_length) {
						ce->ce_length = 0;
					} else {
						ce->ce_length -= (min_offset - ce->ce_diskoffset);
						ce->ce_diskoffset = min_offset;
					}
				}
			}

			if (ce->ce_length != pe->pe_length) {
				clipped_extents++;
			}
			if (ce->ce_length == 0) {
				/* new extent is entirely captured in extents from the old cache, throw it out */
				enveloped_extents++;
				BC_teardown_extent(ce);
				lck_mtx_destroy(&ce->ce_lock, BC_cache->c_lckgrp);
				
				/* continue without incrementing ncache_extents so we reuse this extent index */
				continue;
			}
			size += roundup(ce->ce_length, PAGE_SIZE);
#endif			
			
			cm->cm_pextents[cm->cm_nextents] = ce;
			cm->cm_nextents++;
			ncache_extents++;
		}
		remaining_entries -= actual;
		entries_buf += actual * sizeof(struct BC_playlist_entry);
	}
	
	/* Fill in any remaining extents from the original cache that are still sitting at the high end of the extent list */
	for (pm_idx = 0; pm_idx < nplaylist_mounts; pm_idx++) {
		cm_idx = pmidx_to_cmidx[pm_idx];
		cm = cache_mounts + cm_idx;
		pm = playlist_mounts + pm_idx;
		
		if (cm->cm_nextents == 0) {
			message("Bad playlist: No entries existed for mount %s (claimed %d)", uuid_string(pm->pm_uuid), pm->pm_nentries);
			error = EINVAL;
			goto out;
		}
		
		if (cm_idx < BC_cache->c_nmounts) {
			if (next_old_extent_idx[cm_idx] < BC_cache->c_mounts[cm_idx].cm_nextents) {
				/* There are more extents in the old extent array, copy them over */
				memcpy(cm->cm_pextents + cm->cm_nextents, BC_cache->c_mounts[cm_idx].cm_pextents + next_old_extent_idx[cm_idx], sizeof(*cm->cm_pextents) * (BC_cache->c_mounts[cm_idx].cm_nextents - next_old_extent_idx[cm_idx]));
				cm->cm_nextents += BC_cache->c_mounts[cm_idx].cm_nextents - next_old_extent_idx[cm_idx];
			}
		}
		
	}
	
#ifndef BOOTCACHE_ENTRIES_SORTED_BY_DISK_OFFSET
	/* We need an ordered list of extents for each mount, but we were given
	 * the extents in a different ordering, so we need to sort the mount's
	 * extent list here
	 */
	for (cm_idx = 0; cm_idx < ncache_mounts; cm_idx++) {
		cm = cache_mounts + cm_idx;
		
		/* If this is a new mount or a mount with a new extent list, it needs sorting */
		if (cm_idx >= BC_cache->c_nmounts ||
			cm->cm_pextents != BC_cache->c_mounts[cm_idx].cm_pextents) {
			
			qsort(cm->cm_pextents,
				  cm->cm_nextents,
				  sizeof(*cm->cm_pextents),
				  compare_cache_extents);
			
			/* Clip the new extents of overlap with the old extents */
			for (ce_idx = 0; ce_idx < cm->cm_nextents; ce_idx++) {
				
				/* Only look at new extents, old extents were already clipped when they were added and can't be clipped further */
				if (cm->cm_pextents[ce_idx] >= cache_extents &&
					cm->cm_pextents[ce_idx] < cache_extents + ncache_extents) {

					u_int64_t old_length = ce->ce_length;

					/* Compare against the previous (lower) extent */
					if (ce_idx > 0) {
						u_int64_t min_offset = cm->cm_pextents[ce_idx - 1]->ce_diskoffset + cm->cm_pextents[ce_idx - 1]->ce_length;
						if (min_offset > ce->ce_diskoffset) {
							if (min_offset >= ce->ce_diskoffset + ce->ce_length) {
								ce->ce_length = 0;
							} else {
								ce->ce_length -= (min_offset - ce->ce_diskoffset);
								ce->ce_diskoffset = min_offset;
							}
						}
					}
					
					/* Compare against the next (higher) extent */
					if (ce_idx < cm_nextents - 1) {
						u_int64_t max_offset = cm->cm_pextents[ce_idx + 1]->ce_diskoffset;
						if (max_offset < ce->ce_diskoffset + ce->ce_length) {
							
							/* Check if this extent is overlapping with an extent from the same playlist */
							if (cm->cm_pextents[ce_idx + 1] >= cache_extents &&
								cm->cm_pextents[ce_idx + 1] < cache_extents + ncache_extents) {
								message("Bad playlist: entry %d (0x%llx:%lld) overlapped with previous entry for mount %s (0x%llx:%lld)", pe_idx, ce->ce_diskoffset, ce->ce_length, uuid_string(cm->cm_uuid), cm->cm_pextents[ce_idx + 1]->ce_diskoffset, cm->cm_pextents[ce_idx + 1]->ce_length);
								error = EINVAL;
								goto out;
							}
							
							if (max_offset <= ce->ce_diskoffset) {
								ce->ce_length = 0;
							} else {
								ce->ce_length = (max_offset - ce->ce_diskoffset);
							}
						}
					}
					
					if (ce->ce_length != old_length) {
						clipped_extents++;
						if (ce->ce_length == 0) {
							/* extent was enveloped, remove this extent from the mount's extent list */
							if ((ce_idx + 1) < cm->cm_nextents) {
								memmove(cm->cm_pextents + ce_idx, cm->cm_pextents + (ce_idx + 1), cm->cm_nextents - (ce_idx + 1));
							}
							cm->cm_nextents--;
							ce_idx--;
							enveloped_extents++;
						}
					}
				}
			}
		}
	}

	for (ce_idx = 0; ce_idx < ncache_extents; ce_idx++) {
		size += roundup(cache_extents->ce_length, PAGE_SIZE);
	}
	
#endif
	
	if (clipped_extents > 0) {
		debug("%d extents were clipped, %d of which were completely enveloped", clipped_extents, enveloped_extents);
	}
	
	if (ncache_extents == 0) {
		debug("No extents added, not updating cache");
		error = EALREADY;
		goto out;
	}
		
	if (! BC_preloaded_playlist) {
		_FREE_ZERO(playlist_entries, M_TEMP);
		_FREE_ZERO(playlist_mounts, M_TEMP);
	}
		
	if (ncache_extents < nplaylist_entries) {
		debug("%d extents added out of %d", ncache_extents, nplaylist_entries);
	}
	for (cm_idx = 0; cm_idx < ncache_mounts; cm_idx++) {
		debug("Mount %s now has %d extents", uuid_string(cache_mounts[cm_idx].cm_uuid), cache_mounts[cm_idx].cm_nextents);
	}
	
	
	/*
	 * In order to simplify allocation of the block and page maps, we round
	 * the size up to a multiple of CB_MAPFIELDBITS pages.  This means we
	 * may waste as many as 31 pages in the worst case.
	 */
	size = roundup(size, PAGE_SIZE * CB_MAPFIELDBITS);
	
	/*
	 * Verify that the cache is not larger than 50% of physical memory
	 * to avoid forcing pageout activity.
	 *
	 * max_mem and size are in bytes.  All calculations are done in page
	 * units to avoid overflow/sign issues.
	 *
	 * If the cache is too large, we discard it and go to record-only mode.
	 * This lets us survive the case where the playlist is corrupted or
	 * manually made too large, but the "real" bootstrap footprint
	 * would be OK.
	 */
	BC_cache->c_cachesize += size;
	if (BC_cache->c_cachesize > BC_MAX_SIZE) {
		message("cache size (%lu bytes) too large for physical memory (%llu bytes), capping at %llu bytes",
				(unsigned long)BC_cache->c_cachesize, max_mem, BC_MAX_SIZE);
		BC_cache->c_cachesize = BC_MAX_SIZE;
	}
		
	/*
	 * Allocate/grow the pagebuffer.
	 */
	if (BC_alloc_pagebuffer() != 0) {
		message("can't allocate %lld bytes for cache buffer", BC_cache->c_cachesize);
		error = ENOMEM;
		goto out;
	}

	/*
	 * Fix up the extent cache offsets.
	 */
	p = 0;
	if (BC_cache->c_nextentlists > 0) {
		/* start at the offset of the end of the previous cache list */
		for (cel_idx = BC_cache->c_nextentlists - 1; cel_idx >= 0; cel_idx--) {
			if (BC_cache->c_nextents[cel_idx] > 0) {
				struct BC_cache_extent *last_ce = BC_cache->c_extentlists[cel_idx] + BC_cache->c_nextents[cel_idx] - 1;
				p = last_ce->ce_cacheoffset + roundup(last_ce->ce_length, PAGE_SIZE);
				break;
			}
		}
	}
	for (ce_idx = 0; ce_idx < ncache_extents; ce_idx++) {
		cache_extents[ce_idx].ce_cacheoffset = p;
		p += roundup(cache_extents[ce_idx].ce_length, PAGE_SIZE);
		
		/* Abort any extents that extend past our cache size */
		if (p >= BC_cache->c_cachesize || cache_extents[ce_idx].ce_length == 0) {
			BC_teardown_extent(cache_extents + ce_idx);
		}
	}
	
	/* We've successfully integrated the new playlist with our cache, copy it into place */
	
	/* free up old cache's allocations that we're replacing */
	if (BC_cache->c_nmounts > 0) {
		for (cm_idx = 0; cm_idx < BC_cache->c_nmounts; cm_idx++) {
			/* If we have a new extent array for this mount, free the old one */
			if (BC_cache->c_mounts[cm_idx].cm_pextents != cache_mounts[cm_idx].cm_pextents) {
				_FREE_ZERO(BC_cache->c_mounts[cm_idx].cm_pextents, M_TEMP);
			}
		}
		_FREE_ZERO(BC_cache->c_mounts, M_TEMP)
	}
	BC_cache->c_mounts = cache_mounts;
	BC_cache->c_nmounts = ncache_mounts;
	
	if (BC_cache->c_nextentlists > 0) {
		_FREE_ZERO(BC_cache->c_extentlists, M_TEMP)
		_FREE_ZERO(BC_cache->c_nextents, M_TEMP)		
	}
	
	cache_nextents[BC_cache->c_nextentlists] = ncache_extents;
	cache_extents_list[BC_cache->c_nextentlists] = cache_extents;
	
	BC_cache->c_nextentlists++;
	
	BC_cache->c_nextents = cache_nextents;
	BC_cache->c_extentlists = cache_extents_list;
		
	BC_cache->c_stats.ss_total_mounts = ncache_mounts;
	BC_cache->c_stats.ss_total_extents += ncache_extents;
		
	/* fill (or tear down) the extents for all the mounts we've already seen */
	for (ce_idx = 0; ce_idx < ncache_extents; ce_idx++) {
		ce = cache_extents + ce_idx;
		if (CM_SETUP != BC_cache->c_mounts[ce->ce_mount_idx].cm_state) {
			LOCK_EXTENT(ce);
			if (CM_READY == BC_cache->c_mounts[ce->ce_mount_idx].cm_state) {
				if (0 == BC_fill_in_extent(ce)) {
					UNLOCK_EXTENT(ce);
					continue;
				}
			}
			/* either the mount is aborted or fill_in_extent failed */
			BC_teardown_extent(ce);
			UNLOCK_EXTENT(ce);
		}
	}
		
	/* Check to see if we have any more mounts to fill in */
	for (cm_idx = 0; cm_idx < BC_cache->c_nmounts; cm_idx++) {
		if (CM_SETUP == BC_cache->c_mounts[cm_idx].cm_state) {
			/* have at least one mount to fill in, iterate all of them */
			int go_again = 0;
			vfs_iterate(0, fill_in_bc_cache_mounts, &go_again);
			if (go_again) {
				debug("Missing root mount during last iteration, going again");
				vfs_iterate(0, fill_in_bc_cache_mounts, &go_again);
			}
			break;
		}
	}
		
	error = 0;

	/* all done */
out:
	if (! BC_preloaded_playlist) {
		_FREE_ZERO(playlist_entries, M_TEMP);
		_FREE_ZERO(playlist_mounts, M_TEMP);
	}
	_FREE_ZERO(pmidx_to_cmidx, M_TEMP);
	_FREE_ZERO(next_old_extent_idx, M_TEMP);
	if (error != 0) {
		if (error != EALREADY) {
			debug("cache setup failed, aborting");
		}

		/* If it was possible to fail after growing the page buffer, we'd need a way to shrink it back here */

		if (BC_cache->c_extentlists == NULL) {
			/* If this is our first cache, clear stats and our page buffer */
			BC_cache->c_stats.ss_total_extents = 0;
			BC_cache->c_stats.ss_total_mounts = 0;
			if (BC_cache->c_vp != NULL) {
				BC_free_pagebuffer();
			}
		}
		
		BC_cache->c_cachesize = old_cache_size;
		BC_cache->c_batch_count = old_batch_count;

		/* free up new cache's allocations that aren't part of the old cache */
		if (cache_extents) {
			for (ce_idx = 0; ce_idx < ncache_extents; ce_idx++) {
				BC_teardown_extent(cache_extents + ce_idx);
				lck_mtx_destroy(&cache_extents[ce_idx].ce_lock, BC_cache->c_lckgrp);
			}
			_FREE_ZERO(cache_extents, M_TEMP);
		}
		_FREE_ZERO(cache_nextents, M_TEMP);
		_FREE_ZERO(cache_extents_list, M_TEMP);

		if (cache_mounts) {
			for (cm_idx = 0; cm_idx < ncache_mounts; cm_idx++) {
				if (cm_idx >= BC_cache->c_nmounts) {
					/* newly allocated mounts, teardown completely */
					BC_teardown_mount(cache_mounts + cm_idx);
					lck_rw_destroy(&cache_mounts[cm_idx].cm_lock, BC_cache->c_lckgrp);
				} else if (BC_cache->c_mounts[cm_idx].cm_pextents != cache_mounts[cm_idx].cm_pextents) {
					/* old mounts with a new extent list */
					_FREE_ZERO(cache_mounts[cm_idx].cm_pextents, M_TEMP);
				}
			}
			_FREE_ZERO(cache_mounts, M_TEMP);
		}
	}
	
#ifdef BC_DEBUG
	struct timeval end_time;
	microtime(&end_time);
	timersub(&end_time, &start_time, &end_time);
	debug("BC_copyin_playlist took %ldus", end_time.tv_sec * 1000000 + end_time.tv_usec);
#endif	
	
	return(error);
}

/*
 * Initialise the cache.  Fetch the playlist from userspace and
 * find and attach to our block device.
 */
static int
BC_init_cache(size_t mounts_size, user_addr_t mounts, size_t entries_size, user_addr_t entries, int record_history)
{
	int error = 0, copyin_error, mount_idx;
	struct timeval current_time;
	
	/* Everything should be set, or nothing should be set */
	if ((mounts_size == 0 || mounts == 0 || entries_size == 0 || entries == 0) &&
		(mounts_size != 0 || mounts != 0 || entries_size != 0 || entries != 0)) {
		debug("Invalid cache parameters: %ld, %lld, %ld, %lld", mounts_size, mounts, entries_size, entries);
		return(EINVAL);
	}
	
	
	if (! (BC_cache->c_flags & BC_FLAG_SETUP)) {
		debug("Cache not yet setup");
		return(ENXIO);
	}
		
	microtime(&current_time);

	/*
	 * Start recording history.
	 *
	 * Only one history recording may be active at a time
	 */
	if (record_history) {
		LOCK_HANDLERS();
		if (BC_set_flag(BC_FLAG_HISTORYACTIVE)) {
			debug("history recording already started, only one playlist will be returned");
			BC_ADD_STAT(history_num_recordings, 1);
			UNLOCK_HANDLERS();
		} else {
			debug("starting history recording");
			
			/* start the timeout handler */
			timeout(BC_timeout_history, NULL, BC_history_timeout);

			/* always take stats for the first two recordings (boot + login), even if we don't have a playlist */
			if (BC_cache->c_stats.ss_history_num_recordings < 2)
				BC_cache->c_take_detailed_stats = 1;

			BC_cache->c_history_starttime = current_time;
			BC_cache->c_history_size = 0;
			BC_cache->c_history_num_clusters = 0;
			BC_ADD_STAT(history_num_recordings, 1);
			
			
			/*
			 * Take over the strategy routine for our device; we are now
			 * recording read operations.
			 */
			BC_check_handlers();
			UNLOCK_HANDLERS();
			
			BC_update_mounts();
			
#ifndef NO_HISTORY
# ifdef STATIC_HISTORY
			/* initialise the history buffer */
			if ((BC_cache->c_history = BC_MALLOC(BC_HISTORY_ALLOC, M_TEMP, M_WAITOK)) == NULL) {
				message("can't allocate %d bytes for static history buffer", BC_HISTORY_ALLOC);
				BC_clear_flag(BC_FLAG_HISTORYACTIVE);
				error = ENOMEM;
				goto out;
			}
			
			bzero(BC_cache->c_history, BC_HISTORY_ALLOC);
# endif
#endif
			
		}
	}
	
	/*
	 * If we have playlist data, fetch it.
	 */
	if (entries_size > 0) {
		LOCK_CACHE_W();

		if (BC_cache->c_flags & BC_FLAG_SHUTDOWN) {
			debug("cache already shutdown");
			error = ENXIO;
			UNLOCK_CACHE_W();
			goto out;
		}
		
		/* we're playing back a cache (or at least trying to),
		 * record detailed statistics until we stop recording history
		 */
		BC_cache->c_take_detailed_stats = 1;		
		
		copyin_error = BC_copyin_playlist(mounts_size, mounts, entries_size, entries);
		if (copyin_error != 0 && copyin_error != EALREADY) {
			message("can't load playlist: %d", copyin_error);
			UNLOCK_CACHE_W();
			error = copyin_error;
			goto out;
		}
				
		/* Always throttle cut-through IOs.
		 * I tried throttling only after login had started, but
		 * then users who log in quickly see a very long delay
		 * after entering their password before apps come up.
		 * I'd prefer delay the time until the login screen appears
		 * and provide a fast-as-possible login.
		 * rdar://8592401
		 */
		BC_cache->c_readahead_throttles_cutthroughs = 1;		
		
		LOCK_HANDLERS();
		if (BC_set_flag(BC_FLAG_CACHEACTIVE)) {
			/* cache is already running, we're adding to it */
			UNLOCK_HANDLERS();
		} else {
			/* first cache we've created this boot */
			
#ifdef READ_HISTORY_BUFFER
			/* initialise the read history buffer */
			if (NULL == BC_cache->c_rhistory) {
				if ((BC_cache->c_rhistory = BC_MALLOC(READ_HISTORY_BUFFER * sizeof(struct BC_read_history_entry),
													M_TEMP, M_WAITOK)) == NULL) {
					message("can't allocate %d bytes for read history buffer",
							READ_HISTORY_BUFFER * sizeof(struct BC_read_history_entry));
				}
			}
			if (BC_cache->c_rhistory) {
				bzero(BC_cache->c_rhistory, READ_HISTORY_BUFFER * sizeof(struct BC_read_history_entry));
			}
#endif
			
			BC_cache->c_cache_starttime = current_time;

			/*
			 * Take over the strategy routine for our device; we are now
			 * filling read operations from the cache.
			 *
			 * This must be done before we kick off reader threads to ensure
			 * that we see catch any writes that occur on blocks in our cache.
			 * This ensures that a write won't come in unnoticed on a block we've
			 * already read in so when it's subsequently requested, the cache gives
			 * stale data.
			 */
			BC_check_handlers();
			UNLOCK_HANDLERS();			
		}
		
		LOCK_CACHE_W_TO_R();
		if (copyin_error != EALREADY) {
			/*
			 * Start the reader threads.
			 */
			for (mount_idx = 0; mount_idx < BC_cache->c_nmounts; mount_idx++) {
				if (BC_cache->c_mounts[mount_idx].cm_state == CM_READY) {
					LOCK_MOUNT_R(BC_cache->c_mounts + mount_idx);
					BC_mount_available(BC_cache->c_mounts + mount_idx);
					UNLOCK_MOUNT_R(BC_cache->c_mounts + mount_idx);
				}
			}
		}
		
		UNLOCK_CACHE_R();
		
		/* cache is now active */
		
	}
	
	if (BC_cache->c_stats.ss_preload_time == 0) {
		timersub(&current_time,
				 &BC_cache->c_loadtime,
				 &current_time);
		BC_cache->c_stats.ss_preload_time = (u_int) current_time.tv_sec * 1000 + (u_int) current_time.tv_usec / 1000;
	}
	
out:
	return(error);
}

/*
 * Timeout on collection.
 *
 * We haven't heard from the userland component for too long;
 * stop recording history, but keep the current history
 * around in case they show up.
 */
static void
BC_timeout_history(void *junk)
{
	/* stop recording history */
	debug("history recording timed out");
	BC_terminate_history();
}

/*
 * Check for new mounts. Called with no locks held
 */
static int check_for_new_mount_itr(mount_t mount, void* arg) {
	struct vfs_attr attr;
	vfs_context_t context;
	int error, mount_idx;
	int* go_again = (int*) arg;

	struct BC_history_mount_device *hmd = NULL;
	vnode_t devvp = NULLVP;
	
	LOCK_HISTORY_R();
	if (! (BC_cache->c_flags & BC_FLAG_HISTORYACTIVE)) {
		/* don't do anything if we've stopped recording history */
		UNLOCK_HISTORY_R();
		return VFS_RETURNED_DONE;
	}	
	
#ifdef BC_DEBUG
	char name[MFSNAMELEN];
	vfs_name(mount, name);
#endif
	
	devvp = vfs_devvp(mount);
	if (devvp == NULLVP) {
		//debug("Mount %s (%p)'s vnode is NULL", name, mount);
		UNLOCK_HISTORY_R();
		goto out;
	}
	
	dev_t mount_dev = vnode_specrdev(devvp);
	//debug("Mount %s (%p)'s vnode %p's device is 0x%x", name, mount, devvp, mount_dev);
		
	/* 
	 * A new mount appeared, but unfortunately we don't know which one.
	 * So, for every mount we have to check through our list of history_mounts
	 * to see if we already have info for it.
	 */	
	hmd = BC_get_history_mount_device(mount_dev, NULL);
	if (hmd == NULL || (! uuid_is_null(hmd->hmd_mount.hm_uuid))) {
		/* Unable to allocate new space for a mount, or
		 * we already have info about this mount
		 */
		UNLOCK_HISTORY_R();
		goto out;
	}
	
	/*
	 * We don't yet have info about this mount. It's new!
	 */
	
	hmd->hmd_blocksize = vfs_devblocksize(mount);
	
	VFSATTR_INIT(&attr);
	VFSATTR_WANTED(&attr, f_uuid);
	context = vfs_context_create(NULL);
	error = vfs_getattr(mount, &attr, context);
	
	if ((0 != error) || (! VFSATTR_IS_SUPPORTED(&attr, f_uuid))) {
		vfs_context_rele(context);
#ifdef BC_DEBUG
		if (strncmp("devfs", name, sizeof("devfs")) != 0) {
			debug("Found mount %s without a UUID: error %d", name, error);
		}
#endif
		UNLOCK_HISTORY_R();
		goto out;
	}

	hmd->hmd_disk_id = vfs_throttle_mask(mount);
	
	if (VNOP_IOCTL(devvp,              /* vnode */
				   DKIOCISSOLIDSTATE,    /* cmd */
				   (caddr_t)&hmd->hmd_is_ssd, /* data */
				   0,
				   context))           /* context */
	{
		debug("can't determine if physical disk for history mount %s is an ssd", uuid_string(hmd->hmd_mount.hm_uuid));
		hmd->hmd_is_ssd = 0;
	}
	
	/* Found info for our mount */
	debug("Found new historic mount after %d IOs: %s, dev 0x%x, disk 0x%llx, blk %d%s",
		  hmd->hmd_mount.hm_nentries,
		  uuid_string(attr.f_uuid),
		  hmd->hmd_dev,
		  hmd->hmd_disk_id,
		  hmd->hmd_blocksize,
		  hmd->hmd_is_ssd ? ", ssd" : "");
	
	if (hmd->hmd_blocksize == 0) {
		hmd->hmd_blocksize = 512;
	}
	hmd->hmd_mount.hm_nentries = 0;
	
	uuid_copy(hmd->hmd_mount.hm_uuid, attr.f_uuid);
	
	UNLOCK_HISTORY_R();

	if (BC_cache->c_flags & BC_FLAG_CACHEACTIVE) {
		LOCK_CACHE_R();
		/* Check if we have a playlist mount that matches this mount and set it up */
		for (mount_idx = 0; mount_idx < BC_cache->c_nmounts; mount_idx++) {
			struct BC_cache_mount *cm = BC_cache->c_mounts + mount_idx;
			if (CM_SETUP == cm->cm_state && 0 == uuid_compare(cm->cm_uuid, attr.f_uuid)) {
				/* Found a matching unfilled mount */
				
				LOCK_MOUNT_W(cm);
				if ((error = BC_fill_in_mount(cm, mount, context)) == 0) {
					LOCK_MOUNT_W_TO_R(cm);
					/* Kick off another reader thread if this is a new physical disk */
					BC_mount_available(cm);
					UNLOCK_MOUNT_R(cm);
				} else {
					if (error == EAGAIN) {
						*go_again = 1;
					}
					UNLOCK_MOUNT_W(cm);
				}
				
				break;
			}
		}
		UNLOCK_CACHE_R();
	}
	
	vfs_context_rele(context);
		
out:
	if (devvp != NULLVP) {
		vnode_put(devvp);
	}
	return VFS_RETURNED;
}

/*
 * Check for new or disappearing mounts 
 */
static void BC_update_mounts(void) {
	
	/* don't do anything if we've stopped recording history */
	if (! (BC_cache->c_flags & BC_FLAG_HISTORYACTIVE)) {
		return;
	}
	
	int go_again = 0;
	vfs_iterate(0, check_for_new_mount_itr, &go_again);
	if (go_again) {
		debug("Missing root mount during last iteration, going again");
		vfs_iterate(0, check_for_new_mount_itr, &go_again);
	}
}

/*
 * Find the mount that corresponds to this IO,
 * or the first empty slot.
 *
 * Called with the history read lock
 */
static struct BC_history_mount_device * BC_get_history_mount_device(dev_t dev, int* pindex) {
	struct BC_history_mount_device *tmphmd = NULL, **phmd, *ret = NULL;
	int mount_idx = 0;
	
	for (phmd = &BC_cache->c_history_mounts;
		 (*phmd) == NULL || (*phmd)->hmd_dev != dev;
		 phmd = &(*phmd)->hmd_next) {
		
		if (*phmd == NULL) {
			if (tmphmd == NULL) { /* we may have an allocation from previous iterations */
				tmphmd = kalloc(sizeof(struct BC_history_mount_device));
				if (tmphmd == NULL) {
					message("could not allocate %lu bytes for history mount device",
							sizeof(struct BC_history_mount_device));
					BC_set_flag(BC_FLAG_HTRUNCATED);
					goto out;
				}
				
				/* claim the new entry */
				tmphmd->hmd_next = NULL;
				tmphmd->hmd_mount.hm_nentries = 0;
				uuid_clear(tmphmd->hmd_mount.hm_uuid);
				tmphmd->hmd_disk_id = 0;
				tmphmd->hmd_blocksize = 0;
				tmphmd->hmd_dev = dev;
			}
			
			if (OSCompareAndSwapPtr(NULL, tmphmd, phmd)) {
				tmphmd = NULL;
				if (BC_cache->c_take_detailed_stats) {
					BC_ADD_STAT(history_mounts, 1);
				}
				break;
			}
			
			/* someone else added a mount before we could, check if its the one we want */
			if ((*phmd)->hmd_dev == dev) break;
			
		}
		mount_idx++;
	}
	
	if (pindex)
		*pindex = mount_idx;
	ret = *phmd;
out:
	if (tmphmd)
		kfree(tmphmd, sizeof(struct BC_history_mount_device));
	
	return ret;
}

/*
 * Record an incoming IO in the history list.
 *
 * Returns non-0 if this IO was on the root disk.
 *
 * Note that this function is not reentrant.
 */
static int
BC_add_history(daddr64_t blkno, u_int64_t length, int pid, int hit, int write, int tag, int shared, dev_t dev)
{	
	u_int64_t offset;
	struct BC_history_entry_cluster *hec, *tmphec = NULL;
	struct BC_history_mount_device *hmd = NULL;
	struct BC_history_entry *he;
	int mount_idx, entry_idx;
	int is_root_disk = 0;
	
	/* don't do anything if we've stopped recording history */
	if (! (BC_cache->c_flags & BC_FLAG_HISTORYACTIVE))
		return 0;
	
	LOCK_HISTORY_R();
	
	/* check again, with lock */
	if (! (BC_cache->c_flags & BC_FLAG_HISTORYACTIVE))
		goto out;
	
	/* count IOs during boot (first recording) */
	if (BC_cache->c_take_detailed_stats) {
		if (! tag) {
			if (! write) {
				BC_ADD_STAT(history_reads, 1);
			} else {
				BC_ADD_STAT(history_writes, 1);
			}
		}
	}
	
	/* don't do anything if the history list has been truncated */
	if (!tag && BC_cache->c_flags & BC_FLAG_HTRUNCATED)
		goto out;
	
	/*
	 * In order to avoid recording a playlist larger than we will
	 * allow to be played back, if the total byte count for recorded
	 * reads reaches 50% of the physical memory in the system, we
	 * start ignoring reads.
	 * This is preferable to dropping entries from the end of the
	 * playlist as we copy it in, since it will result in a clean
	 * cessation of caching at a single point, rather than
	 * introducing sporadic cache misses throughout the cache's run.
	 */
	if (!tag && (BC_cache->c_history_size + length) > BC_MAX_SIZE) {
		debug("History recording too large, capping at %lluMB", BC_MAX_SIZE / (1024 * 1024));
		BC_set_flag(BC_FLAG_HTRUNCATED);
		goto out;
	}
	
#ifdef NO_HISTORY
	BC_set_flag(BC_FLAG_HTRUNCATED);
	debug("history disabled, truncating");
	goto out;
#endif
	
	mount_idx = 0;
	offset = 0;
	if (! tag) {
		
		hmd = BC_get_history_mount_device(dev, &mount_idx);
		if (hmd == NULL) goto out;
		
		if (uuid_is_null(hmd->hmd_mount.hm_uuid)) {
			/* Couldn't identify the mount
			 *
			 * Before the mount has appeared we don't want to
			 * include any of this device's IOs in the history
			 * since these IOs will be issued next boot before
			 * we're able to start this mount's part of the
			 * playlist.
			 */
			
			/* Keep track of the number of IOs we've seen until the mount appears */
			OSIncrementAtomic(&hmd->hmd_mount.hm_nentries);
			
			if (BC_cache->c_take_detailed_stats) {
				BC_ADD_STAT(history_unknown, 1);
				BC_ADD_STAT(history_unknown_bytes, length);
			}
			goto out;
		}
		
		offset = HB_BLOCK_TO_BYTE(hmd, blkno);
		
		/* Only track IOs on the root disk.
		 *
		 * We don't expect IOs on other physical disks
		 * to be part of the critical path to boot/login
		 * and we certainly don't expect the other disks
		 * to be contested enough to make a cache worthwhile
		 */
		
		if (hmd->hmd_disk_id != BC_cache->c_root_disk_id) {
#ifdef ROOT_DISK_ONLY
			goto out;
#endif
		} else {
			is_root_disk = 1;
		}
		
#ifdef DONT_CACHE_SSDS
		if (hmd->hmd_is_ssd) {
			goto out;
		}
#endif
		
	}
	
	/*
	 * Get the current entry cluster.
	 */
	while ((hec = BC_cache->c_history) == NULL ||
		   (entry_idx = OSAddAtomic(1, &hec->hec_nentries)) >= BC_HISTORY_ENTRIES) {
		if (hec)
			OSAddAtomic(-1, &hec->hec_nentries);
		
		/* need a new cluster */
		
		/* limit the number of clusters we will allocate */
		if (BC_cache->c_history_num_clusters + 1 >= BC_HISTORY_MAXCLUSTERS) {
			message("too many history clusters (%d, limit %ld)",
					BC_cache->c_history_num_clusters,
					(long)BC_HISTORY_MAXCLUSTERS);
			BC_set_flag(BC_FLAG_HTRUNCATED);
			goto out;
		}
		
		if (tmphec == NULL) { /* we may have an allocation from previous iterations */
			tmphec = kalloc(BC_HISTORY_ALLOC);
			if (tmphec == NULL) {
				message("could not allocate %d bytes for history cluster",
						BC_HISTORY_ALLOC);
				BC_set_flag(BC_FLAG_HTRUNCATED);
				goto out;
			}
			
			/* claim the first entry of the new cluster*/
			tmphec->hec_nentries = 1;
			tmphec->hec_link = hec;
		}
		
		if (OSCompareAndSwapPtr(hec, tmphec, &BC_cache->c_history)) {
			hec = tmphec;
			entry_idx = 0;
			tmphec = NULL;
			OSAddAtomic(1, &BC_cache->c_history_num_clusters);
			break;
		}
	}
	if (tmphec != NULL)
		kfree(tmphec, BC_HISTORY_ALLOC);
		
	/*
	 * We've claimed entry entry_idx of cluster hec, fill it in.
	 */
	he = &(hec->hec_data[entry_idx]);
	assert(he >= &hec->hec_data[0]);
	assert(he < &hec->hec_data[BC_HISTORY_ENTRIES]);
	he->he_offset    = offset;
	he->he_length    = length;
	he->he_pid       = pid;
	he->he_mount_idx = mount_idx;
	he->he_flags     = 0x0;
	if (hit)    he->he_flags |= BC_HE_HIT;
	if (write)  he->he_flags |= BC_HE_WRITE;
	if (tag)    he->he_flags |= BC_HE_TAG;
	if (shared) he->he_flags |= BC_HE_SHARED;
	if (hmd)
		OSIncrementAtomic(&hmd->hmd_mount.hm_nentries);
	
	if (!write) {
		OSAddAtomic64(length, &BC_cache->c_history_size);
		if (BC_cache->c_take_detailed_stats) {
			BC_ADD_STAT(history_bytes, length);
			BC_ADD_STAT(history_entries, 1);
		}
	}
out:
	UNLOCK_HISTORY_R();
	return is_root_disk;
}

/*
 * Return the size of the history buffer.
 */
static int 
BC_size_history_mounts(void)
{
	struct BC_history_mount_device *hmd;
	int nmounts;
	
	/* walk the list of mount clusters and count the number of mounts */
	nmounts = 0;
	for (hmd = BC_cache->c_history_mounts; hmd != NULL; hmd = hmd->hmd_next)
		nmounts ++;
	
	return(nmounts * sizeof(struct BC_history_mount));
}

/*
 * Return the size of the history buffer.
 */
static int 
BC_size_history_entries(void)
{
	struct BC_history_entry_cluster *hec;
	int nentries, cluster_entries;
	
	/* walk the list of clusters and count the number of entries */
	nentries = 0;
	for (hec = BC_cache->c_history; hec != NULL; hec = hec->hec_link) {
		cluster_entries = hec->hec_nentries;
		if (cluster_entries > BC_HISTORY_ENTRIES) { 
			cluster_entries = BC_HISTORY_ENTRIES;
		}
		
		nentries += cluster_entries;
	}
	
	return(nentries * sizeof(struct BC_history_entry));
}

/*
 * Copy out either the history buffer size, or the entire history buffer.
 *
 * Note that the cache must be shut down before calling this function in
 * order to avoid a race around the entry count/buffer size.
 */
static int
BC_copyout_history_mounts(user_addr_t uptr)
{
	struct BC_history_mount_device *hmd;
	int error;
	
	assert(uptr != 0);
	
	/*
	 * Copy out the mounts
	 */
	for (hmd = BC_cache->c_history_mounts;
		 hmd != NULL;
		 hmd = hmd->hmd_next) {
		
		/* copy the cluster out */
		if ((error = copyout(&(hmd->hmd_mount), uptr,
							 sizeof(struct BC_history_mount))) != 0)
			return(error);
		uptr += sizeof(struct BC_history_mount);
		
	}
	return(0);
}

/*
 * Copy out either the history buffer size, or the entire history buffer.
 *
 * Note that the cache must be shut down before calling this function in
 * order to avoid a race around the entry count/buffer size.
 */
static int
BC_copyout_history_entries(user_addr_t uptr)
{
	struct BC_history_entry_cluster *hec, *ohec;
	int error, cluster_entries;
	
	assert(uptr != 0);
	
	/*
	 * Copying the history entires out is a little messy due to the fact that the
	 * clusters are on the list backwards, but we want the entries in order.
	 */
	ohec = NULL;
	for (;;) {
		/* scan the list until we find the last one we haven't copied */
		hec = BC_cache->c_history;
		while (hec->hec_link != ohec)
			hec = hec->hec_link;
		ohec = hec;
		
		cluster_entries = hec->hec_nentries;
		if (cluster_entries > BC_HISTORY_ENTRIES) { 
			cluster_entries = BC_HISTORY_ENTRIES;
		}
		
		/* copy the cluster out */
		if ((error = copyout(hec->hec_data, uptr,
							 cluster_entries * sizeof(struct BC_history_entry))) != 0)
			return(error);
		uptr += cluster_entries * sizeof(struct BC_history_entry);
		
		/* if this was the last cluster, all done */
		if (hec == BC_cache->c_history)
			break;
	}
	return(0);
}

/*
 * Discard the history list.
 */
static void
BC_discard_history(void)
{
	struct BC_history_mount_device  *hmd;
	struct BC_history_entry_cluster *hec;
	
	while (BC_cache->c_history_mounts != NULL) {
		hmd = BC_cache->c_history_mounts;
		BC_cache->c_history_mounts = hmd->hmd_next;
		kfree(hmd, sizeof(struct BC_history_mount_device));
	}	
	
	while (BC_cache->c_history != NULL) {
		hec = BC_cache->c_history;
		BC_cache->c_history = hec->hec_link;
		kfree(hec, BC_HISTORY_ALLOC);
	}
}

/*
 * Setup for the root mounted device
 */
static int
BC_setup(void)
{
	unsigned int boot_arg;
	
	/*
	 * If we have less than the minimum supported physical memory,
	 * bail immediately.  With too little memory, the cache will
	 * become polluted with nondeterministic pagein activity, leaving
	 * large amounts of wired memory around which further exacerbates
	 * the problem.
	 */
	if ((max_mem / (1024 * 1024)) < BC_minimum_mem_size) {
		debug("insufficient physical memory (%dMB < %dMB required)",
			  (int) (max_mem / (1024 * 1024)), BC_minimum_mem_size);
		return(ENOMEM);
	}
	
	/*
	 * Find our block device.
	 *
	 * Note that in the netbooted case, we are loaded but should not
	 * actually run, unless we're explicitly overridden.
	 *
	 * Note also that if the override is set, we'd better have a valid
	 * rootdev...
	 */
	if (netboot_root() && 
		(!PE_parse_boot_argn("BootCacheOverride", &boot_arg, sizeof(boot_arg)) ||
		 (boot_arg != 1))) {                          /* not interesting */
			return(ENXIO);
		}
	
	debug("Major of rootdev is %d", major(rootdev));
	
	BC_cache->c_devsw = &bdevsw[major(rootdev)];
	assert(BC_cache->c_devsw != NULL);
	
	/* Make sure we don't clobber the device strategy routine */
	if ((BC_cache->c_devsw->d_strategy == (strategy_fcn_t*)BC_strategy) ||
		(BC_cache->c_devsw->d_close    == BC_close)) {
		debug("cache was not terminated properly previously");		
		return(ENXIO);
	}
	
	BC_cache->c_strategy = BC_cache->c_devsw->d_strategy;
	BC_cache->c_close    = BC_cache->c_devsw->d_close;
	if (BC_cache->c_strategy == NULL ||
		BC_cache->c_close    == NULL) {	/* not configured */
		return(ENXIO);
	}
	
#ifdef ROOT_DISK_ONLY
#ifdef DONT_CACHE_SSDS
	vfs_context_t context;
	context = vfs_context_create(NULL);
	int is_ssd;
	if (!rootvp ||
		VNOP_IOCTL(rootvp,       /* vnode */
				   DKIOCISSOLIDSTATE,    /* cmd */
				   (caddr_t)&is_ssd, /* data */
				   0,
				   context))           /* context */
	{
		message("can't determine if disk is a solid state disk for root mount");
		is_ssd = 0;
	}
	vfs_context_rele(context);
	if (is_ssd) {
		return(ENODEV);
	}
#endif
#endif
	
	BC_set_flag(BC_FLAG_SETUP);
	
	return(0);
}


/*
 * Called from the root-mount hook.
 */
static void
BC_auto_start(void)
{
	int error;
	user_addr_t pm, pe;
	
	if ((error = BC_setup()) != 0) {
		debug("BootCache setup failed: %d", error);
		return;
	}
	
	pm = CAST_USER_ADDR_T((uintptr_t)BC_preloaded_playlist + sizeof(struct BC_playlist_header));
	pe = pm + (BC_preloaded_playlist->ph_nmounts * sizeof(struct BC_playlist_mount));
	
	/* Initialize the cache.
	 *
	 * auto-start is for DVDs or other read-only media, so don't bother recording history
	 */
	error = BC_init_cache(BC_preloaded_playlist->ph_nmounts  * sizeof(struct BC_playlist_mount), pm,
						  BC_preloaded_playlist->ph_nentries * sizeof(struct BC_playlist_entry), pe,
						  0 /* don't record history */);
	if (error != 0)
		debug("BootCache autostart failed: %d", error);
}

/*
 * Called from the root-unmount hook.
 */
static void
BC_unmount_hook(void)
{
	debug("device unmounted");
	
	/*
	 * If the cache is running, stop it.  If it's already stopped
	 * (it may have stopped itself), that's OK.
	 */
	BC_terminate_cache();
	BC_terminate_history();
	
}


/*
 * Handle the command sysctl.
 */
static int
BC_sysctl SYSCTL_HANDLER_ARGS
{
	struct BC_command bc;
	int error;
	
	/* get the commande structure and validate */
	if ((error = SYSCTL_IN(req, &bc, sizeof(bc))) != 0) {
		debug("couldn't get command");
		return(error);
	}
	if (bc.bc_magic != BC_MAGIC) {
		debug("bad command magic");
		return(EINVAL);
	}
	
	switch (bc.bc_opcode) {
		case BC_OP_START:
			debug("BC_OP_START(%d mounts, %d extents)", (int)(bc.bc_data1_size / sizeof(struct BC_playlist_mount)), (int)(bc.bc_data2_size / sizeof(struct BC_playlist_entry)));
			if (BC_preloaded_playlist) {
				error = EINVAL;
				break;
			}
			error = BC_init_cache(bc.bc_data1_size, (user_addr_t)bc.bc_data1, bc.bc_data2_size, (user_addr_t)bc.bc_data2, 1 /* record history */);
			if (error != 0) {
				message("cache init failed");
			}
			break;
			
		case BC_OP_STOP:
			debug("BC_OP_STOP");
			
			/*
			 * If we're recording history, stop it.  If it's already stopped
			 * (it may have stopped itself), that's OK.
			 */
			BC_terminate_history();
			
			/* return the size of the history buffer */
			LOCK_HISTORY_W();
			bc.bc_data1_size = BC_size_history_mounts();
			bc.bc_data2_size = BC_size_history_entries();
			if ((error = SYSCTL_OUT(req, &bc, sizeof(bc))) != 0)
				debug("could not return history size");
			UNLOCK_HISTORY_W();
			
			break;
			
		case BC_OP_MOUNT:
			/* debug("BC_OP_MOUNT"); */
			
			/*
			 * A notification that mounts have changed.
			 */
			BC_update_mounts();
			break;
			
		case BC_OP_HISTORY:
			debug("BC_OP_HISTORY");
			/* if there's a user buffer, verify the size and copy out */
			LOCK_HISTORY_W();
			if (bc.bc_data1 != 0 && bc.bc_data2 != 0) {
				if (bc.bc_data1_size < BC_size_history_mounts() ||
					bc.bc_data2_size < BC_size_history_entries()) {
					debug("supplied history buffer too small");
					error = EINVAL;
					UNLOCK_HISTORY_W();
					break;
				}
				if ((error = BC_copyout_history_mounts(bc.bc_data1)) != 0) {
					debug("could not copy out history mounts: %d", error);
				}
				if ((error = BC_copyout_history_entries(bc.bc_data2)) != 0) {
					debug("could not copy out history entries: %d", error);
				}
			}
			/*
			 * Release the last of the memory we own.
			 */
			BC_discard_history();
			UNLOCK_HISTORY_W();
			break;
			
		case BC_OP_JETTISON:
			debug("BC_OP_JETTISON");
			
			/*
			 * Jettison the cache. If it's already jettisoned
			 * (it may have jettisoned itself), that's OK.
			 */
			BC_terminate_cache();
			break;
			
		case BC_OP_STATS:
			debug("BC_OP_STATS");
			/* check buffer size and copy out */
			if (bc.bc_data1_size != sizeof(BC_cache->c_stats)) {
				debug("stats structure wrong size");
				error = ENOMEM;
			} else {
				BC_cache->c_stats.ss_cache_flags = BC_cache->c_flags;
				if ((error = copyout(&BC_cache->c_stats, bc.bc_data1, bc.bc_data1_size)) != 0)
					debug("could not copy out statistics");
			}
			break;
			
		case BC_OP_TAG:
			debug("BC_OP_TAG");
			KERNEL_DEBUG_CONSTANT(FSDBG_CODE(DBG_BOOTCACHE, DBG_BC_TAG) | 
								  DBG_FUNC_NONE, proc_selfppid(), dbg_tag_count, 
								  0, 0, 0);
			BC_add_history(0, 0, 0, 0, 0, 1, 0, 0);
			dbg_tag_count++;
			break;
			
		case BC_OP_TEST:
			debug("BC_OP_TEST");
			
			if (! (BC_cache->c_flags & BC_FLAG_SETUP)) {
				error = ENODEV;
			}
			break;
			
		default:
			error = EINVAL;
	}
	return(error);
}


/*
 * Initialise the block of pages we use to back our data.
 *
 */
static int
BC_alloc_pagebuffer()
{
	kern_return_t kret;
	int ret;
	
	if (BC_cache->c_vp == NULL) {
		
		kret = vnode_open(BC_BOOT_BACKINGFILE, (O_CREAT | O_NOFOLLOW | FREAD), 
						  0400, 0, &BC_cache->c_vp, vfs_context_current());
		if (kret != KERN_SUCCESS) {
			debug("vnode_open failed - %d", kret);
			return -1;
		}
		
		BC_cache->c_vid = vnode_vid(BC_cache->c_vp);

	} else {
		ret = vnode_getwithvid(BC_cache->c_vp, BC_cache->c_vid);
		if (ret != 0) {
			debug("vnode_getwithvid failed - %d", ret);
			return -1;
		}
	}
		
	assert(BC_cache->c_cachesize != 0);
	kret = ubc_setsize(BC_cache->c_vp, BC_cache->c_cachesize);
	if (kret != 1) {
		debug("ubc_setsize failed - %d", kret);
		(void) vnode_put(BC_cache->c_vp);
		return -1;
	}
	
	(void) vnode_put(BC_cache->c_vp);
	
	return(0);
}

/*
 * Release all resources associated with our pagebuffer.
 */
static void
BC_free_pagebuffer(void)
{
	kern_return_t kret;
	if (BC_cache->c_vp != NULL) {
		/* Can handle a vnode without ubc info */
		kret = vnode_getwithref(BC_cache->c_vp);
		if (kret != KERN_SUCCESS) {
			debug("vnode_getwithref failed - %d", kret);
			return;
		}
		
		kret = ubc_range_op(BC_cache->c_vp, 0, BC_cache->c_cachesize, 
							UPL_ROP_DUMP, NULL);
		if (kret != KERN_SUCCESS) {
			debug("ubc_range_op failed - %d", kret);
		}
		
#if 0
		/* need this interface to be exported... will eliminate need
		 * for the ubc_range_op above */
		kret = vnode_delete(BC_BOOT_BACKINGFILE, vfs_context_current());
		if (kret) {
			debug("vnode_delete failed - %d", error);
		}
#endif
		kret = vnode_close(BC_cache->c_vp, 0, vfs_context_current());
		if (kret != KERN_SUCCESS) {
			debug("vnode_close failed - %d", kret);
		}
		
		BC_cache->c_vp = NULL;
	}
}

/*
 * Release one page from the pagebuffer.
 */
static void
BC_free_page(struct BC_cache_extent *ce, int page)
{
	off_t vpoffset;
	kern_return_t kret;
	
	vpoffset = (page * PAGE_SIZE) + ce->ce_cacheoffset;
	kret = vnode_getwithvid(BC_cache->c_vp, BC_cache->c_vid);
	if (kret != KERN_SUCCESS) {
		debug("vnode_getwithvid failed - %d", kret);
		return;
	}
	
	kret = ubc_range_op(BC_cache->c_vp, vpoffset, vpoffset + PAGE_SIZE,
						UPL_ROP_DUMP, NULL);
	if (kret != KERN_SUCCESS) {
		debug("ubc_range_op failed - %d", kret);
	}
	
	(void) vnode_put(BC_cache->c_vp);
}

void
BC_hook(void)
{
	int error;
	
	if ((error = BC_setup()) != 0) {
		debug("BootCache setup failed: %d", error);
	}
}

/*
 * Cache start/stop functions.
 */
void
BC_load(void)
{
	struct mach_header	*mh;
	long			xsize;
	int			has64bitness, error;
	size_t			sysctlsize = sizeof(int);
	char			*plsection = "playlist";
	
#ifdef BC_DEBUG
	microtime(&debug_starttime);
#endif
	
	/* register our control interface */
	sysctl_register_oid(&sysctl__kern_BootCache);
	
	/* clear the control structure */
	bzero(BC_cache, sizeof(*BC_cache));
	
	BC_cache->c_lckgrp = lck_grp_alloc_init("BootCache", LCK_GRP_ATTR_NULL);
	lck_rw_init(&BC_cache->c_history_lock, BC_cache->c_lckgrp, LCK_ATTR_NULL);
	lck_rw_init(&BC_cache->c_cache_lock, BC_cache->c_lckgrp, LCK_ATTR_NULL);
	lck_mtx_init(&BC_cache->c_handlers_lock, BC_cache->c_lckgrp, LCK_ATTR_NULL);
	lck_mtx_init(&BC_cache->c_reader_threads_lock, BC_cache->c_lckgrp, LCK_ATTR_NULL);
#ifdef BC_DEBUG
	lck_mtx_init(&debug_printlock, BC_cache->c_lckgrp, LCK_ATTR_NULL);
#endif
	/*
	 * Find the preload section and its size.  If it's not empty
	 * we have a preloaded playlist, so prepare for an early
	 * start.
	 */
	if (kmod_info.info_version != KMOD_INFO_VERSION) {
		message("incompatible kmod_info versions");
		return;
	}
	mh = (struct mach_header *)kmod_info.address;
	
	/*
	 * If booting on a 32-bit only machine, use the "playlist32" section.
	 * Otherwise, fall through to the default "playlist" section.
	 */
	error = sysctlbyname("hw.cpu64bit_capable", &has64bitness, &sysctlsize, 
						 NULL, 0);
	if (error == 0 && has64bitness == 0) {
		plsection = "playlist32";
	}
	BC_preloaded_playlist = getsectdatafromheader(mh, "BootCache", 
												  plsection, &BC_preloaded_playlist_size);
	debug("preload section %s: %d @ %p", 
	      plsection, BC_preloaded_playlist_size, BC_preloaded_playlist);
	
	if (BC_preloaded_playlist != NULL) {
		if (BC_preloaded_playlist_size < sizeof(struct BC_playlist_header)) {
			message("preloaded playlist too small");
			return;
		}
		if (BC_preloaded_playlist->ph_magic != PH_MAGIC) {
			message("preloaded playlist has invalid magic (%x, expected %x)",
					BC_preloaded_playlist->ph_magic, PH_MAGIC);
			return;
		}
		xsize = sizeof(struct BC_playlist_header) +
		(BC_preloaded_playlist->ph_nmounts  * sizeof(struct BC_playlist_mount)) +
		(BC_preloaded_playlist->ph_nentries * sizeof(struct BC_playlist_entry));
		if (xsize > BC_preloaded_playlist_size) {
			message("preloaded playlist too small (%d, expected at least %ld)",
					BC_preloaded_playlist_size, xsize);
			return;
		}
		BC_wait_for_readahead = BC_READAHEAD_WAIT_CDROM;
		mountroot_post_hook = BC_auto_start;
		unmountroot_pre_hook = BC_unmount_hook;
		
		debug("waiting for root mount...");
	} else {
		/* no preload, normal operation */
		BC_preloaded_playlist = NULL;
		mountroot_post_hook = BC_hook;
		debug("waiting for playlist...");
	}
	
	microtime(&BC_cache->c_loadtime);	
}

int
BC_unload(void)
{
	int error;
	
	debug("preparing to unload...");
	unmountroot_pre_hook = NULL;
	
	/*
	 * If the cache is running, stop it.  If it's already stopped
	 * (it may have stopped itself), that's OK.
	 */
	BC_terminate_history();
	if ((error = BC_terminate_cache()) != 0) {
		if (error != ENXIO)
			goto out;
	}
	LOCK_HISTORY_W();
	BC_discard_history();
	UNLOCK_HISTORY_W();
	
	lck_rw_destroy(&BC_cache->c_history_lock, BC_cache->c_lckgrp);
	lck_rw_destroy(&BC_cache->c_cache_lock, BC_cache->c_lckgrp);
	lck_mtx_destroy(&BC_cache->c_handlers_lock, BC_cache->c_lckgrp);
	lck_mtx_destroy(&BC_cache->c_reader_threads_lock, BC_cache->c_lckgrp);
	lck_grp_free(BC_cache->c_lckgrp);
	
	sysctl_unregister_oid(&sysctl__kern_BootCache);
	error = 0;
	
out:
	return(error);
}