Move xen patchset to new version's subdir.

[ipfire-2.x.git] / src / patches / suse-2.6.27.31 / patches.suse / dm-raid45_2.6.27_20081027.patch

From: "Heinz Mauelshagen <hjm@redhat.de>
Subject: DMRAID45 module
X-URL: http://people.redhat.com/~heinzm/sw/dm/dm-raid45/

 DM-RAID 45 module.

 This driver is used for "Fake RAID" devices.

Acked-by: Jeff Mahoney <jeffm@suse.com>

---

 drivers/md/Kconfig         |    8 
 drivers/md/Makefile        |    4 
 drivers/md/dm-memcache.c   |  301 ++
 drivers/md/dm-memcache.h   |   68 
 drivers/md/dm-message.c    |  182 +
 drivers/md/dm-message.h    |   91 
 drivers/md/dm-raid45.c     | 4516 +++++++++++++++++++++++++++++++++++++++++++++
 drivers/md/dm-raid45.h     |   28 
 drivers/md/dm-regions.c    |  723 +++++++
 drivers/md/dm.c            |    1 
 include/linux/dm-regions.h |  115 +
 11 files changed, 6036 insertions(+), 1 deletion(-)

--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -1680,6 +1680,7 @@ struct gendisk *dm_disk(struct mapped_de
 {
 	return md->disk;
 }
+EXPORT_SYMBOL_GPL(dm_disk);
 
 int dm_suspended(struct mapped_device *md)
 {
--- /dev/null
+++ b/drivers/md/dm-memcache.c
@@ -0,0 +1,301 @@
+/*
+ * Copyright (C) 2006-2008 Red Hat, Inc. All rights reserved.
+ *
+ * Module Author: Heinz Mauelshagen <heinzm@redhat.com>
+ *
+ * Device-mapper memory object handling:
+ *
+ * o allocate/free total_pages in a per client page pool.
+ *
+ * o allocate/free memory objects with chunks (1..n) of
+ *   pages_per_chunk pages hanging off.
+ *
+ * This file is released under the GPL.
+ */
+
+#define	DM_MEM_CACHE_VERSION	"0.2"
+
+#include "dm.h"
+#include "dm-memcache.h"
+#include <linux/dm-io.h>
+
+struct dm_mem_cache_client {
+	spinlock_t lock;
+	mempool_t *objs_pool;
+	struct page_list *free_list;
+	unsigned objects;
+	unsigned chunks;
+	unsigned pages_per_chunk;
+	unsigned free_pages;
+	unsigned total_pages;
+};
+
+/*
+ * Free pages and page_list elements of client.
+ */
+static void free_cache_pages(struct page_list *list)
+{
+	while (list) {
+		struct page_list *pl = list;
+
+		list = pl->next;
+		BUG_ON(!pl->page);
+		__free_page(pl->page);
+		kfree(pl);
+	}
+}
+
+/*
+ * Alloc number of pages and page_list elements as required by client.
+ */
+static struct page_list *alloc_cache_pages(unsigned pages)
+{
+	struct page_list *pl, *ret = NULL;
+	struct page *page;
+
+	while (pages--) {
+		page = alloc_page(GFP_NOIO);
+		if (!page)
+			goto err;
+
+		pl = kmalloc(sizeof(*pl), GFP_NOIO);
+		if (!pl) {
+			__free_page(page);
+			goto err;
+		}
+
+		pl->page = page;
+		pl->next = ret;
+		ret = pl;
+	}
+
+	return ret;
+
+err:
+	free_cache_pages(ret);
+	return NULL;
+}
+
+/*
+ * Allocate page_list elements from the pool to chunks of the memory object.
+ */
+static void alloc_chunks(struct dm_mem_cache_client *cl,
+			 struct dm_mem_cache_object *obj)
+{
+	unsigned chunks = cl->chunks;
+	unsigned long flags;
+
+	local_irq_save(flags);
+	local_irq_disable();
+	while (chunks--) {
+		unsigned p = cl->pages_per_chunk;
+
+		obj[chunks].pl = NULL;
+
+		while (p--) {
+			struct page_list *pl;
+
+			/* Take next element from free list */
+			spin_lock(&cl->lock);
+			pl = cl->free_list;
+			BUG_ON(!pl);
+			cl->free_list = pl->next;
+			spin_unlock(&cl->lock);
+
+			pl->next = obj[chunks].pl;
+			obj[chunks].pl = pl;
+		}
+	}
+
+	local_irq_restore(flags);
+}
+
+/*
+ * Free page_list elements putting them back onto free list
+ */
+static void free_chunks(struct dm_mem_cache_client *cl,
+			struct dm_mem_cache_object *obj)
+{
+	unsigned chunks = cl->chunks;
+	unsigned long flags;
+	struct page_list *next, *pl;
+
+	local_irq_save(flags);
+	local_irq_disable();
+	while (chunks--) {
+		for (pl = obj[chunks].pl; pl; pl = next) {
+			next = pl->next;
+
+			spin_lock(&cl->lock);
+			pl->next = cl->free_list;
+			cl->free_list = pl;
+			cl->free_pages++;
+			spin_unlock(&cl->lock);
+		}
+	}
+
+	local_irq_restore(flags);
+}
+
+/*
+ * Create/destroy dm memory cache client resources.
+ */
+struct dm_mem_cache_client *
+dm_mem_cache_client_create(unsigned objects, unsigned chunks,
+			   unsigned pages_per_chunk)
+{
+	unsigned total_pages = objects * chunks * pages_per_chunk;
+	struct dm_mem_cache_client *client;
+
+	BUG_ON(!total_pages);
+	client = kzalloc(sizeof(*client), GFP_KERNEL);
+	if (!client)
+		return ERR_PTR(-ENOMEM);
+
+	client->objs_pool = mempool_create_kmalloc_pool(objects,
+				chunks * sizeof(struct dm_mem_cache_object));
+	if (!client->objs_pool)
+		goto err;
+
+	client->free_list = alloc_cache_pages(total_pages);
+	if (!client->free_list)
+		goto err1;
+
+	spin_lock_init(&client->lock);
+	client->objects = objects;
+	client->chunks = chunks;
+	client->pages_per_chunk = pages_per_chunk;
+	client->free_pages = client->total_pages = total_pages;
+	return client;
+
+err1:
+	mempool_destroy(client->objs_pool);
+err:
+	kfree(client);
+	return ERR_PTR(-ENOMEM);
+}
+EXPORT_SYMBOL(dm_mem_cache_client_create);
+
+void dm_mem_cache_client_destroy(struct dm_mem_cache_client *cl)
+{
+	BUG_ON(cl->free_pages != cl->total_pages);
+	free_cache_pages(cl->free_list);
+	mempool_destroy(cl->objs_pool);
+	kfree(cl);
+}
+EXPORT_SYMBOL(dm_mem_cache_client_destroy);
+
+/*
+ * Grow a clients cache by an amount of pages.
+ *
+ * Don't call from interrupt context!
+ */
+int dm_mem_cache_grow(struct dm_mem_cache_client *cl, unsigned objects)
+{
+	unsigned pages = objects * cl->chunks * cl->pages_per_chunk;
+	struct page_list *pl, *last;
+
+	BUG_ON(!pages);
+	pl = alloc_cache_pages(pages);
+	if (!pl)
+		return -ENOMEM;
+
+	last = pl;
+	while (last->next)
+		last = last->next;
+
+	spin_lock_irq(&cl->lock);
+	last->next = cl->free_list;
+	cl->free_list = pl;
+	cl->free_pages += pages;
+	cl->total_pages += pages;
+	cl->objects++;
+	spin_unlock_irq(&cl->lock);
+
+	mempool_resize(cl->objs_pool, cl->objects, GFP_NOIO);
+	return 0;
+}
+EXPORT_SYMBOL(dm_mem_cache_grow);
+
+/* Shrink a clients cache by an amount of pages */
+int dm_mem_cache_shrink(struct dm_mem_cache_client *cl, unsigned objects)
+{
+	int r;
+	unsigned pages = objects * cl->chunks * cl->pages_per_chunk, p = pages;
+	unsigned long flags;
+	struct page_list *last = NULL, *pl, *pos;
+
+	BUG_ON(!pages);
+
+	spin_lock_irqsave(&cl->lock, flags);
+	pl = pos = cl->free_list;
+	while (p-- && pos->next) {
+		last = pos;
+		pos = pos->next;
+	}
+
+	if (++p)
+		r = -ENOMEM;
+	else {
+		r = 0;
+		cl->free_list = pos;
+		cl->free_pages -= pages;
+		cl->total_pages -= pages;
+		cl->objects--;
+		last->next = NULL;
+	}
+	spin_unlock_irqrestore(&cl->lock, flags);
+
+	if (!r) {
+		free_cache_pages(pl);
+		mempool_resize(cl->objs_pool, cl->objects, GFP_NOIO);
+	}
+
+	return r;
+}
+EXPORT_SYMBOL(dm_mem_cache_shrink);
+
+/*
+ * Allocate/free a memory object
+ *
+ * Can be called from interrupt context
+ */
+struct dm_mem_cache_object *dm_mem_cache_alloc(struct dm_mem_cache_client *cl)
+{
+	int r = 0;
+	unsigned pages = cl->chunks * cl->pages_per_chunk;
+	unsigned long flags;
+	struct dm_mem_cache_object *obj;
+
+	obj = mempool_alloc(cl->objs_pool, GFP_NOIO);
+	if (!obj)
+		return ERR_PTR(-ENOMEM);
+
+	spin_lock_irqsave(&cl->lock, flags);
+	if (pages > cl->free_pages)
+		r = -ENOMEM;
+	else
+		cl->free_pages -= pages;
+	spin_unlock_irqrestore(&cl->lock, flags);
+
+	if (r) {
+		mempool_free(obj, cl->objs_pool);
+		return ERR_PTR(r);
+	}
+
+	alloc_chunks(cl, obj);
+	return obj;
+}
+EXPORT_SYMBOL(dm_mem_cache_alloc);
+
+void dm_mem_cache_free(struct dm_mem_cache_client *cl,
+		       struct dm_mem_cache_object *obj)
+{
+	free_chunks(cl, obj);
+	mempool_free(obj, cl->objs_pool);
+}
+EXPORT_SYMBOL(dm_mem_cache_free);
+
+MODULE_DESCRIPTION(DM_NAME " dm memory cache");
+MODULE_AUTHOR("Heinz Mauelshagen <hjm@redhat.com>");
+MODULE_LICENSE("GPL");
--- /dev/null
+++ b/drivers/md/dm-memcache.h
@@ -0,0 +1,68 @@
+/*
+ * Copyright (C) 2006-2008 Red Hat, Inc. All rights reserved.
+ *
+ * Module Author: Heinz Mauelshagen <Mauelshagen@RedHat.com>
+ *
+ * Device-mapper memory object handling:
+ *
+ * o allocate/free total_pages in a per client page pool.
+ *
+ * o allocate/free memory objects with chunks (1..n) of
+ *   pages_per_chunk pages hanging off.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _DM_MEM_CACHE_H
+#define _DM_MEM_CACHE_H
+
+#define	DM_MEM_CACHE_H_VERSION	"0.1"
+
+#include "dm.h"
+#include <linux/dm-io.h>
+
+static inline struct page_list *pl_elem(struct page_list *pl, unsigned p)
+{
+	while (pl && p--)
+		pl = pl->next;
+
+	return pl;
+}
+
+struct dm_mem_cache_object {
+	struct page_list *pl; /* Dynamically allocated array */
+	void *private;	      /* Caller context reference */
+};
+
+struct dm_mem_cache_client;
+
+/*
+ * Create/destroy dm memory cache client resources.
+ *
+ * On creation, a number of @objects with @chunks of
+ * @pages_per_chunk pages will be allocated.
+ */
+struct dm_mem_cache_client *
+dm_mem_cache_client_create(unsigned objects, unsigned chunks,
+			   unsigned pages_per_chunk);
+void dm_mem_cache_client_destroy(struct dm_mem_cache_client *client);
+
+/*
+ * Grow/shrink a dm memory cache client resources
+ * by @objetcs amount of objects.
+ */
+int dm_mem_cache_grow(struct dm_mem_cache_client *client, unsigned objects);
+int dm_mem_cache_shrink(struct dm_mem_cache_client *client, unsigned objects);
+
+/*
+ * Allocate/free a memory object
+ *
+ * On allocation one object with an amount of chunks and
+ * an amount of pages per chunk will be returned on success.
+ */
+struct dm_mem_cache_object *
+dm_mem_cache_alloc(struct dm_mem_cache_client *client);
+void dm_mem_cache_free(struct dm_mem_cache_client *client,
+		       struct dm_mem_cache_object *object);
+
+#endif
--- /dev/null
+++ b/drivers/md/dm-message.c
@@ -0,0 +1,182 @@
+/*
+ * Copyright (C) 2007,2008 Red Hat Inc. All rights reserved.
+ *
+ * Module Author: Heinz Mauelshagen <heinzm@redhat.com>
+ *
+ * General device-mapper message interface argument parser.
+ *
+ * This file is released under the GPL.
+ *
+ * device-mapper message parser.
+ *
+ */
+
+#include "dm.h"
+#include "dm-message.h"
+#include <linux/kernel.h>
+
+#define DM_MSG_PREFIX	"dm_message"
+
+/* Basename of a path. */
+static inline char *
+basename(char *s)
+{
+	char *p = strrchr(s, '/');
+
+	return p ? p + 1 : s;
+}
+
+/* Get an argument depending on type. */
+static void
+message_arguments(struct dm_msg *msg, int argc, char **argv)
+{
+
+	if (argc) {
+		int i;
+		struct dm_message_argument *args = msg->spec->args;
+
+		for (i = 0; i < args->num_args; i++) {
+			int r;
+			unsigned long **ptr = args->ptr;
+			enum dm_message_argument_type type = args->types[i];
+
+			switch (type) {
+			case dm_msg_base_t:
+				((char **) ptr)[i] = basename(argv[i]);
+				break;
+
+			case dm_msg_str_t:
+				((char **) ptr)[i] = argv[i];
+				break;
+
+			case dm_msg_int_t:
+				r = sscanf(argv[i], "%d", ((int **) ptr)[i]);
+				goto check;
+
+			case dm_msg_uint_t:
+				r = sscanf(argv[i], "%u",
+					   ((unsigned **) ptr)[i]);
+				goto check;
+
+			case dm_msg_uint64_t:
+				r = sscanf(argv[i], "%llu",
+					   ((unsigned long long **) ptr)[i]);
+
+check:
+				if (r != 1) {
+					set_bit(dm_msg_ret_undef, &msg->ret);
+					set_bit(dm_msg_ret_arg, &msg->ret);
+				}
+			}
+		}
+	}
+}
+
+/* Parse message options. */
+static void
+message_options_parse(struct dm_msg *msg, int argc, char **argv)
+{
+	int hit = 0;
+	unsigned long *action;
+	size_t l1 = strlen(*argv), l_hit = 0;
+	struct dm_message_option *o = msg->spec->options;
+	char **option, **option_end = o->options + o->num_options;
+
+	for (option = o->options, action = o->actions;
+	     option < option_end; option++, action++) {
+		size_t l2 = strlen(*option);
+
+		if (!strnicmp(*argv, *option, min(l1, l2))) {
+			hit++;
+			l_hit = l2;
+			set_bit(*action, &msg->action);
+		}
+	}
+
+	/* Assume error. */
+	msg->ret = 0;
+	set_bit(dm_msg_ret_option, &msg->ret);
+	if (!hit || l1 > l_hit)
+		set_bit(dm_msg_ret_undef, &msg->ret);	/* Undefined option. */
+	else if (hit > 1)
+		set_bit(dm_msg_ret_ambiguous, &msg->ret); /* Ambiguous option.*/
+	else {
+		clear_bit(dm_msg_ret_option, &msg->ret); /* Option OK. */
+		message_arguments(msg, --argc, ++argv);
+	}
+}
+
+static inline void
+print_ret(const char *caller, unsigned long ret)
+{
+	struct {
+		unsigned long err;
+		const char *err_str;
+	} static err_msg[] = {
+		{ dm_msg_ret_ambiguous, "message ambiguous" },
+		{ dm_msg_ret_inval, "message invalid" },
+		{ dm_msg_ret_undef, "message undefined" },
+		{ dm_msg_ret_arg, "message argument" },
+		{ dm_msg_ret_argcount, "message argument count" },
+		{ dm_msg_ret_option, "option" },
+	}, *e = ARRAY_END(err_msg);
+
+	while (e-- > err_msg) {
+		if (test_bit(e->err, &ret))
+			DMERR("%s %s", caller, e->err_str);
+	}
+}
+
+/* Parse a message action. */
+int
+dm_message_parse(const char *caller, struct dm_msg *msg, void *context,
+		 int argc, char **argv)
+{
+	int hit = 0;
+	size_t l1 = strlen(*argv), l_hit = 0;
+	struct dm_msg_spec *s, *s_hit = NULL,
+			   *s_end = msg->specs + msg->num_specs;
+
+	if (argc < 2)
+		return -EINVAL;
+
+	for (s = msg->specs; s < s_end; s++) {
+		size_t l2 = strlen(s->cmd);
+
+		if (!strnicmp(*argv, s->cmd, min(l1, l2))) {
+			hit++;
+			l_hit = l2;
+			s_hit = s;
+		}
+	}
+
+	msg->ret = 0;
+	if (!hit || l1 > l_hit)	/* No hit or message string too long. */
+		set_bit(dm_msg_ret_undef, &msg->ret);
+	else if (hit > 1)	/* Ambiguous message. */
+		set_bit(dm_msg_ret_ambiguous, &msg->ret);
+	else if (argc - 2 != s_hit->args->num_args) {
+		set_bit(dm_msg_ret_undef, &msg->ret);
+		set_bit(dm_msg_ret_argcount, &msg->ret);
+	}
+
+	if (msg->ret)
+		goto bad;
+
+	msg->action = 0;
+	msg->spec = s_hit;
+	set_bit(s_hit->action, &msg->action);
+	message_options_parse(msg, --argc, ++argv);
+
+	if (!msg->ret)
+		return msg->spec->f(msg, context);
+
+bad:
+	print_ret(caller, msg->ret);
+	return -EINVAL;
+}
+EXPORT_SYMBOL(dm_message_parse);
+
+MODULE_DESCRIPTION(DM_NAME " device-mapper target message parser");
+MODULE_AUTHOR("Heinz Mauelshagen <hjm@redhat.com>");
+MODULE_LICENSE("GPL");
--- /dev/null
+++ b/drivers/md/dm-message.h
@@ -0,0 +1,91 @@
+/*
+ * Copyright (C) 2007,2008 Red Hat, Inc. All rights reserved.
+ *
+ * Module Author: Heinz Mauelshagen <Mauelshagen@RedHat.de>
+ *
+ * General device-mapper message interface argument parser.
+ *
+ * This file is released under the GPL.
+ *
+ */
+
+#ifndef DM_MESSAGE_H
+#define DM_MESSAGE_H
+
+/* Factor out to dm.h. */
+/* Reference to array end. */
+#define ARRAY_END(a)    ((a) + ARRAY_SIZE(a))
+
+/* Message return bits. */
+enum dm_message_return {
+	dm_msg_ret_ambiguous,		/* Action ambiguous. */
+	dm_msg_ret_inval,		/* Action invalid. */
+	dm_msg_ret_undef,		/* Action undefined. */
+
+	dm_msg_ret_option,		/* Option error. */
+	dm_msg_ret_arg,			/* Argument error. */
+	dm_msg_ret_argcount,		/* Argument count error. */
+};
+
+/* Message argument type conversions. */
+enum dm_message_argument_type {
+	dm_msg_base_t,		/* Basename string. */
+	dm_msg_str_t,		/* String. */
+	dm_msg_int_t,		/* Signed int. */
+	dm_msg_uint_t,		/* Unsigned int. */
+	dm_msg_uint64_t,	/* Unsigned int 64. */
+};
+
+/* A message option. */
+struct dm_message_option {
+	unsigned num_options;
+	char **options;
+	unsigned long *actions;
+};
+
+/* Message arguments and types. */
+struct dm_message_argument {
+	unsigned num_args;
+	unsigned long **ptr;
+	enum dm_message_argument_type types[];
+};
+
+/* Client message. */
+struct dm_msg {
+	unsigned long action;		/* Identified action. */
+	unsigned long ret;		/* Return bits. */
+	unsigned num_specs;		/* # of sepcifications listed. */
+	struct dm_msg_spec *specs;	/* Specification list. */
+	struct dm_msg_spec *spec;	/* Specification selected. */
+};
+
+/* Secification of the message. */
+struct dm_msg_spec {
+	const char *cmd;	/* Name of the command (i.e. 'bandwidth'). */
+	unsigned long action;
+	struct dm_message_option *options;
+	struct dm_message_argument *args;
+	unsigned long parm;	/* Parameter to pass through to callback. */
+	/* Function to process for action. */
+	int (*f) (struct dm_msg *msg, void *context);
+};
+
+/* Parameter access macros. */
+#define	DM_MSG_PARM(msg) ((msg)->spec->parm)
+
+#define	DM_MSG_STR_ARGS(msg, idx) ((char *) *(msg)->spec->args->ptr[idx])
+#define	DM_MSG_INT_ARGS(msg, idx) ((int) *(msg)->spec->args->ptr[idx])
+#define	DM_MSG_UINT_ARGS(msg, idx) ((unsigned) DM_MSG_INT_ARG(msg, idx))
+#define	DM_MSG_UINT64_ARGS(msg, idx) ((uint64_t)  *(msg)->spec->args->ptr[idx])
+
+#define	DM_MSG_STR_ARG(msg)	DM_MSG_STR_ARGS(msg, 0)
+#define	DM_MSG_INT_ARG(msg)	DM_MSG_INT_ARGS(msg, 0)
+#define	DM_MSG_UINT_ARG(msg)	DM_MSG_UINT_ARGS(msg, 0)
+#define	DM_MSG_UINT64_ARG(msg)	DM_MSG_UINT64_ARGS(msg, 0)
+
+
+/* Parse a message and its options and optionally call a function back. */
+int dm_message_parse(const char *caller, struct dm_msg *msg, void *context,
+		     int argc, char **argv);
+
+#endif
--- /dev/null
+++ b/drivers/md/dm-raid45.c
@@ -0,0 +1,4516 @@
+/*
+ * Copyright (C) 2005-2008 Red Hat, Inc. All rights reserved.
+ *
+ * Module Author: Heinz Mauelshagen <Mauelshagen@RedHat.com>
+ *
+ * This file is released under the GPL.
+ *
+ *
+ * Linux 2.6 Device Mapper RAID4 and RAID5 target.
+ *
+ * Supports:
+ *	o RAID4 with dedicated and selectable parity device
+ *	o RAID5 with rotating parity (left+right, symmetric+asymmetric)
+ *	o run time optimization of xor algorithm used to calculate parity
+ *
+ *
+ * Thanks to MD for:
+ *    o the raid address calculation algorithm
+ *    o the base of the biovec <-> page list copier.
+ *
+ *
+ * Uses region hash to keep track of how many writes are in flight to
+ * regions in order to use dirty log to keep state of regions to recover:
+ *
+ *    o clean regions (those which are synchronized
+ * 	and don't have write io in flight)
+ *    o dirty regions (those with write io in flight)
+ *
+ *
+ * On startup, any dirty regions are migrated to the 'nosync' state
+ * and are subject to recovery by the daemon.
+ *
+ * See raid_ctr() for table definition.
+ *
+ *
+ * FIXME:
+ * o add virtual interface for locking
+ * o remove instrumentation (REMOVEME:)
+ *
+ */
+
+static const char *version = "v0.2431";
+
+#include "dm.h"
+#include "dm-bio-list.h"
+#include "dm-memcache.h"
+#include "dm-message.h"
+#include "dm-raid45.h"
+
+#include <linux/kernel.h>
+#include <linux/vmalloc.h>
+
+#include <linux/dm-io.h>
+#include <linux/dm-dirty-log.h>
+#include <linux/dm-regions.h>
+
+/* # of parallel recovered regions */
+/* FIXME: cope with multiple recovery stripes in raid_set struct. */
+#define MAX_RECOVER	1 /* needs to be 1! */
+
+/*
+ * Configurable parameters
+ */
+#define	INLINE
+
+/* Default # of stripes if not set in constructor. */
+#define	STRIPES			64
+
+/* Minimum/maximum # of selectable stripes. */
+#define	STRIPES_MIN		8
+#define	STRIPES_MAX		16384
+
+/* Default chunk size in sectors if not set in constructor. */
+#define	CHUNK_SIZE		64
+
+/* Default io size in sectors if not set in constructor. */
+#define	IO_SIZE_MIN		SECTORS_PER_PAGE
+#define	IO_SIZE			IO_SIZE_MIN
+
+/* Maximum setable chunk size in sectors. */
+#define	CHUNK_SIZE_MAX		16384
+
+/* Recover io size default in sectors. */
+#define	RECOVER_IO_SIZE_MIN	64
+#define	RECOVER_IO_SIZE		256
+
+/* Default percentage recover io bandwidth. */
+#define	BANDWIDTH		10
+#define	BANDWIDTH_MIN		1
+#define	BANDWIDTH_MAX		100
+/*
+ * END Configurable parameters
+ */
+
+#define	TARGET	"dm-raid45"
+#define	DAEMON	"kraid45d"
+#define	DM_MSG_PREFIX	TARGET
+
+#define	SECTORS_PER_PAGE	(PAGE_SIZE >> SECTOR_SHIFT)
+
+/* Amount/size for __xor(). */
+#define	SECTORS_PER_XOR	SECTORS_PER_PAGE
+#define	XOR_SIZE	PAGE_SIZE
+
+/* Derive raid_set from stripe_cache pointer. */
+#define	RS(x)	container_of(x, struct raid_set, sc)
+
+/* Check value in range. */
+#define	range_ok(i, min, max)	(i >= min && i <= max)
+
+/* Page reference. */
+#define PAGE(stripe, p)	((stripe)->obj[p].pl->page)
+
+/* Bio list reference. */
+#define	BL(stripe, p, rw)	(stripe->ss[p].bl + rw)
+
+/* Page list reference. */
+#define	PL(stripe, p)		(stripe->obj[p].pl)
+
+/* Check argument is power of 2. */
+#define POWER_OF_2(a) (!(a & (a - 1)))
+
+/* Factor out to dm-bio-list.h */
+static inline void bio_list_push(struct bio_list *bl, struct bio *bio)
+{
+	bio->bi_next = bl->head;
+	bl->head = bio;
+
+	if (!bl->tail)
+		bl->tail = bio;
+}
+
+/* Factor out to dm.h */
+#define TI_ERR_RET(str, ret) \
+	do { ti->error = DM_MSG_PREFIX ": " str; return ret; } while (0);
+#define TI_ERR(str)     TI_ERR_RET(str, -EINVAL)
+
+/*-----------------------------------------------------------------
+ * Stripe cache
+ *
+ * Cache for all reads and writes to raid sets (operational or degraded)
+ *
+ * We need to run all data to and from a RAID set through this cache,
+ * because parity chunks need to get calculated from data chunks
+ * or, in the degraded/resynchronization case, missing chunks need
+ * to be reconstructed using the other chunks of the stripe.
+ *---------------------------------------------------------------*/
+/* Protect kmem cache # counter. */
+static atomic_t _stripe_sc_nr = ATOMIC_INIT(-1); /* kmem cache # counter. */
+
+/* A stripe set (holds bios hanging off). */
+struct stripe_set {
+	struct stripe *stripe;	/* Backpointer to stripe for endio(). */
+	struct bio_list bl[3]; /* Reads, writes, and writes merged. */
+#define	WRITE_MERGED	2
+};
+
+#if READ != 0 || WRITE != 1
+#error dm-raid45: READ/WRITE != 0/1 used as index!!!
+#endif
+
+/*
+ * Stripe linked list indexes. Keep order, because the stripe
+ * and the stripe cache rely on the first 3!
+ */
+enum list_types {
+	LIST_IO = 0,	/* Stripes with io pending. */
+	LIST_ENDIO,	/* Stripes to endio. */
+	LIST_LRU,	/* Least recently used stripes. */
+	LIST_HASH,	/* Hashed stripes. */
+	LIST_RECOVER = LIST_HASH,	/* For recovery type stripes only. */
+	NR_LISTS,	/* To size array in struct stripe. */
+};
+
+enum lock_types {
+	LOCK_ENDIO = 0,	/* Protect endio list. */
+	LOCK_LRU,	/* Protect lru list. */
+	NR_LOCKS,	/* To size array in struct stripe_cache. */
+};
+
+/* A stripe: the io object to handle all reads and writes to a RAID set. */
+struct stripe {
+	struct stripe_cache *sc;	/* Backpointer to stripe cache. */
+
+	sector_t key;		/* Hash key. */
+	sector_t region;	/* Region stripe is mapped to. */
+
+	/* Reference count. */
+	atomic_t cnt;
+
+	struct {
+		unsigned long flags;	/* flags (see below). */
+
+		/*
+		 * Pending ios in flight:
+		 *
+		 * used as a 'lock' to control move of stripe to endio list
+		 */
+		atomic_t pending;	/* Pending ios in flight. */
+
+		/* Sectors to read and write for multi page stripe sets. */
+		unsigned size;
+	} io;
+
+	/* Lock on stripe (for clustering). */
+	void *lock;
+
+	/*
+	 * 4 linked lists:
+	 *   o io list to flush io
+	 *   o endio list
+	 *   o LRU list to put stripes w/o reference count on
+	 *   o stripe cache hash
+	 */
+	struct list_head lists[NR_LISTS];
+
+	struct {
+		unsigned short parity;	/* Parity chunk index. */
+		short recover;		/* Recovery chunk index. */
+	} idx;
+
+	/* This sets memory cache object (dm-mem-cache). */
+	struct dm_mem_cache_object *obj;
+
+	/* Array of stripe sets (dynamically allocated). */
+	struct stripe_set ss[0];
+};
+
+/* States stripes can be in (flags field). */
+enum stripe_states {
+	STRIPE_ACTIVE,		/* Active io on stripe. */
+	STRIPE_ERROR,		/* io error on stripe. */
+	STRIPE_MERGED,		/* Writes got merged. */
+	STRIPE_READ,		/* Read. */
+	STRIPE_RBW,		/* Read-before-write. */
+	STRIPE_RECONSTRUCT,	/* reconstruct of a missing chunk required. */
+	STRIPE_RECOVER,		/* Stripe used for RAID set recovery. */
+};
+
+/* ... and macros to access them. */
+#define	BITOPS(name, what, var, flag) \
+static inline int TestClear ## name ## what(struct var *v) \
+{ return test_and_clear_bit(flag, &v->io.flags); } \
+static inline int TestSet ## name ## what(struct var *v) \
+{ return test_and_set_bit(flag, &v->io.flags); } \
+static inline void Clear ## name ## what(struct var *v) \
+{ clear_bit(flag, &v->io.flags); } \
+static inline void Set ## name ## what(struct var *v) \
+{ set_bit(flag, &v->io.flags); } \
+static inline int name ## what(struct var *v) \
+{ return test_bit(flag, &v->io.flags); }
+
+
+BITOPS(Stripe, Active, stripe, STRIPE_ACTIVE)
+BITOPS(Stripe, Merged, stripe, STRIPE_MERGED)
+BITOPS(Stripe, Error, stripe, STRIPE_ERROR)
+BITOPS(Stripe, Read, stripe, STRIPE_READ)
+BITOPS(Stripe, RBW, stripe, STRIPE_RBW)
+BITOPS(Stripe, Reconstruct, stripe, STRIPE_RECONSTRUCT)
+BITOPS(Stripe, Recover, stripe, STRIPE_RECOVER)
+
+/* A stripe hash. */
+struct stripe_hash {
+	struct list_head *hash;
+	unsigned buckets;
+	unsigned mask;
+	unsigned prime;
+	unsigned shift;
+};
+
+/* A stripe cache. */
+struct stripe_cache {
+	/* Stripe hash. */
+	struct stripe_hash hash;
+
+	/* Stripes with io to flush, stripes to endio and LRU lists. */
+	struct list_head lists[3];
+
+	/* Locks to protect endio and lru lists. */
+	spinlock_t locks[NR_LOCKS];
+
+	/* Slab cache to allocate stripes from. */
+	struct {
+		struct kmem_cache *cache;	/* Cache itself. */
+		char name[32];	/* Unique name. */
+	} kc;
+
+	struct dm_io_client *dm_io_client; /* dm-io client resource context. */
+
+	/* dm-mem-cache client resource context. */
+	struct dm_mem_cache_client *mem_cache_client;
+
+	int stripes_parm;	    /* # stripes parameter from constructor. */
+	atomic_t stripes;	    /* actual # of stripes in cache. */
+	atomic_t stripes_to_shrink; /* # of stripes to shrink cache by. */
+	atomic_t stripes_last;	    /* last # of stripes in cache. */
+	atomic_t active_stripes;    /* actual # of active stripes in cache. */
+
+	/* REMOVEME: */
+	atomic_t max_active_stripes; /* actual # of active stripes in cache. */
+};
+
+/* Flag specs for raid_dev */ ;
+enum raid_dev_flags { DEVICE_FAILED, IO_QUEUED };
+
+/* The raid device in a set. */
+struct raid_dev {
+	struct dm_dev *dev;
+	unsigned long flags;	/* raid_dev_flags. */
+	sector_t start;		/* offset to map to. */
+};
+
+/* Flags spec for raid_set. */
+enum raid_set_flags {
+	RS_CHECK_OVERWRITE,	/* Check for chunk overwrites. */
+	RS_DEAD,		/* RAID set inoperational. */
+	RS_DEVEL_STATS,		/* REMOVEME: display status information. */
+	RS_IO_ERROR,		/* io error on set. */
+	RS_RECOVER,		/* Do recovery. */
+	RS_RECOVERY_BANDWIDTH,	/* Allow recovery bandwidth (delayed bios). */
+	RS_REGION_GET,		/* get a region to recover. */
+	RS_SC_BUSY,		/* stripe cache busy -> send an event. */
+	RS_SUSPENDED,		/* RAID set suspendedn. */
+};
+
+/* REMOVEME: devel stats counters. */
+enum stats_types {
+	S_BIOS_READ,
+	S_BIOS_ADDED_READ,
+	S_BIOS_ENDIO_READ,
+	S_BIOS_WRITE,
+	S_BIOS_ADDED_WRITE,
+	S_BIOS_ENDIO_WRITE,
+	S_CAN_MERGE,
+	S_CANT_MERGE,
+	S_CONGESTED,
+	S_DM_IO_READ,
+	S_DM_IO_WRITE,
+	S_ACTIVE_READS,
+	S_BANDWIDTH,
+	S_BARRIER,
+	S_BIO_COPY_PL_NEXT,
+	S_DEGRADED,
+	S_DELAYED_BIOS,
+	S_EVICT,
+	S_FLUSHS,
+	S_HITS_1ST,
+	S_IOS_POST,
+	S_INSCACHE,
+	S_MAX_LOOKUP,
+	S_MERGE_PAGE_LOCKED,
+	S_NO_BANDWIDTH,
+	S_NOT_CONGESTED,
+	S_NO_RW,
+	S_NOSYNC,
+	S_PROHIBITPAGEIO,
+	S_RECONSTRUCT_EI,
+	S_RECONSTRUCT_DEV,
+	S_REDO,
+	S_REQUEUE,
+	S_STRIPE_ERROR,
+	S_SUM_DELAYED_BIOS,
+	S_XORS,
+	S_NR_STATS,	/* # of stats counters. */
+};
+
+/* Status type -> string mappings. */
+struct stats_map {
+	const enum stats_types type;
+	const char *str;
+};
+
+static struct stats_map stats_map[] = {
+	{ S_BIOS_READ, "r=" },
+	{ S_BIOS_ADDED_READ, "/" },
+	{ S_BIOS_ENDIO_READ, "/" },
+	{ S_BIOS_WRITE, " w=" },
+	{ S_BIOS_ADDED_WRITE, "/" },
+	{ S_BIOS_ENDIO_WRITE, "/" },
+	{ S_DM_IO_READ, " rc=" },
+	{ S_DM_IO_WRITE, " wc=" },
+	{ S_ACTIVE_READS, " active_reads=" },
+	{ S_BANDWIDTH, " bandwidth=" },
+	{ S_NO_BANDWIDTH, " no_bandwidth=" },
+	{ S_BARRIER, " barrier=" },
+	{ S_BIO_COPY_PL_NEXT, " bio_copy_pl_next=" },
+	{ S_CAN_MERGE, " can_merge=" },
+	{ S_MERGE_PAGE_LOCKED, "/page_locked=" },
+	{ S_CANT_MERGE, "/cant_merge=" },
+	{ S_CONGESTED, " congested=" },
+	{ S_NOT_CONGESTED, "/not_congested=" },
+	{ S_DEGRADED, " degraded=" },
+	{ S_DELAYED_BIOS, " delayed_bios=" },
+	{ S_SUM_DELAYED_BIOS, "/sum_delayed_bios=" },
+	{ S_EVICT, " evict=" },
+	{ S_FLUSHS, " flushs=" },
+	{ S_HITS_1ST, " hits_1st=" },
+	{ S_IOS_POST, " ios_post=" },
+	{ S_INSCACHE, " inscache=" },
+	{ S_MAX_LOOKUP, " max_lookup=" },
+	{ S_NO_RW, " no_rw=" },
+	{ S_NOSYNC, " nosync=" },
+	{ S_PROHIBITPAGEIO, " ProhibitPageIO=" },
+	{ S_RECONSTRUCT_EI, " reconstruct_ei=" },
+	{ S_RECONSTRUCT_DEV, " reconstruct_dev=" },
+	{ S_REDO, " redo=" },
+	{ S_REQUEUE, " requeue=" },
+	{ S_STRIPE_ERROR, " stripe_error=" },
+	{ S_XORS, " xors=" },
+};
+
+/*
+ * A RAID set.
+ */
+typedef void (*xor_function_t)(unsigned count, unsigned long **data);
+struct raid_set {
+	struct dm_target *ti;	/* Target pointer. */
+
+	struct {
+		unsigned long flags;	/* State flags. */
+		spinlock_t in_lock;	/* Protects central input list below. */
+		struct bio_list in;	/* Pending ios (central input list). */
+		struct bio_list work;	/* ios work set. */
+		wait_queue_head_t suspendq;	/* suspend synchronization. */
+		atomic_t in_process;	/* counter of queued bios (suspendq). */
+		atomic_t in_process_max;/* counter of queued bios max. */
+
+		/* io work. */
+		struct workqueue_struct *wq;
+		struct delayed_work dws;
+	} io;
+
+	/* External locking. */
+	struct dm_raid45_locking_type *locking;
+
+	struct stripe_cache sc;	/* Stripe cache for this set. */
+
+	/* Xor optimization. */
+	struct {
+		struct xor_func *f;
+		unsigned chunks;
+		unsigned speed;
+	} xor;
+
+	/* Recovery parameters. */
+	struct recover {
+		struct dm_dirty_log *dl;	/* Dirty log. */
+		struct dm_rh_client *rh;	/* Region hash. */
+
+		/* dm-mem-cache client resource context for recovery stripes. */
+		struct dm_mem_cache_client *mem_cache_client;
+
+		struct list_head stripes;	/* List of recovery stripes. */
+
+		region_t nr_regions;
+		region_t nr_regions_to_recover;
+		region_t nr_regions_recovered;
+		unsigned long start_jiffies;
+		unsigned long end_jiffies;
+
+		unsigned bandwidth;	     /* Recovery bandwidth [%]. */
+		unsigned bandwidth_work; /* Recovery bandwidth [factor]. */
+		unsigned bandwidth_parm; /*  " constructor parm. */
+		unsigned io_size;        /* io size <= chunk size. */
+		unsigned io_size_parm;   /* io size ctr parameter. */
+
+		/* recovery io throttling. */
+		atomic_t io_count[2];	/* counter recover/regular io. */
+		unsigned long last_jiffies;
+
+		struct dm_region *reg;	/* Actual region to recover. */
+		sector_t pos;	/* Position within region to recover. */
+		sector_t end;	/* End of region to recover. */
+	} recover;
+
+	/* RAID set parameters. */
+	struct {
+		struct raid_type *raid_type;	/* RAID type (eg, RAID4). */
+		unsigned raid_parms;	/* # variable raid parameters. */
+
+		unsigned chunk_size;	/* Sectors per chunk. */
+		unsigned chunk_size_parm;
+		unsigned chunk_mask;	/* Mask for amount. */
+		unsigned chunk_shift;	/* rsector chunk size shift. */
+
+		unsigned io_size;	/* Sectors per io. */
+		unsigned io_size_parm;
+		unsigned io_mask;	/* Mask for amount. */
+		unsigned io_shift_mask;	/* Mask for raid_address(). */
+		unsigned io_shift;	/* rsector io size shift. */
+		unsigned pages_per_io;	/* Pages per io. */
+
+		sector_t sectors_per_dev;	/* Sectors per device. */
+
+		atomic_t failed_devs;		/* Amount of devices failed. */
+
+		/* Index of device to initialize. */
+		int dev_to_init;
+		int dev_to_init_parm;
+
+		/* Raid devices dynamically allocated. */
+		unsigned raid_devs;	/* # of RAID devices below. */
+		unsigned data_devs;	/* # of RAID data devices. */
+
+		int ei;		/* index of failed RAID device. */
+
+		/* index of dedicated parity device (i.e. RAID4). */
+		int pi;
+		int pi_parm;	/* constructor parm for status output. */
+	} set;
+
+	/* REMOVEME: devel stats counters. */
+	atomic_t stats[S_NR_STATS];
+
+	/* Dynamically allocated temporary pointers for xor(). */
+	unsigned long **data;
+
+	/* Dynamically allocated RAID devices. Alignment? */
+	struct raid_dev dev[0];
+};
+
+
+BITOPS(RS, Bandwidth, raid_set, RS_RECOVERY_BANDWIDTH)
+BITOPS(RS, CheckOverwrite, raid_set, RS_CHECK_OVERWRITE)
+BITOPS(RS, Dead, raid_set, RS_DEAD)
+BITOPS(RS, DevelStats, raid_set, RS_DEVEL_STATS)
+BITOPS(RS, IoError, raid_set, RS_IO_ERROR)
+BITOPS(RS, Recover, raid_set, RS_RECOVER)
+BITOPS(RS, RegionGet, raid_set, RS_REGION_GET)
+BITOPS(RS, ScBusy, raid_set, RS_SC_BUSY)
+BITOPS(RS, Suspended, raid_set, RS_SUSPENDED)
+#undef BITOPS
+
+#define	PageIO(page)		PageChecked(page)
+#define	AllowPageIO(page)	SetPageChecked(page)
+#define	ProhibitPageIO(page)	ClearPageChecked(page)
+
+/*-----------------------------------------------------------------
+ * Raid-4/5 set structures.
+ *---------------------------------------------------------------*/
+/* RAID level definitions. */
+enum raid_level {
+	raid4,
+	raid5,
+};
+
+/* Symmetric/Asymmetric, Left/Right parity rotating algorithms. */
+enum raid_algorithm {
+	none,
+	left_asym,
+	right_asym,
+	left_sym,
+	right_sym,
+};
+
+struct raid_type {
+	const char *name;		/* RAID algorithm. */
+	const char *descr;		/* Descriptor text for logging. */
+	const unsigned parity_devs;	/* # of parity devices. */
+	const unsigned minimal_devs;	/* minimal # of devices in set. */
+	const enum raid_level level;		/* RAID level. */
+	const enum raid_algorithm algorithm;	/* RAID algorithm. */
+};
+
+/* Supported raid types and properties. */
+static struct raid_type raid_types[] = {
+	{"raid4", "RAID4 (dedicated parity disk)", 1, 3, raid4, none},
+	{"raid5_la", "RAID5 (left asymmetric)", 1, 3, raid5, left_asym},
+	{"raid5_ra", "RAID5 (right asymmetric)", 1, 3, raid5, right_asym},
+	{"raid5_ls", "RAID5 (left symmetric)", 1, 3, raid5, left_sym},
+	{"raid5_rs", "RAID5 (right symmetric)", 1, 3, raid5, right_sym},
+};
+
+/* Address as calculated by raid_address(). */
+struct address {
+	sector_t key;		/* Hash key (start address of stripe). */
+	unsigned di, pi;	/* Data and parity disks index. */
+};
+
+/* REMOVEME: reset statistics counters. */
+static void stats_reset(struct raid_set *rs)
+{
+	unsigned s = S_NR_STATS;
+
+	while (s--)
+		atomic_set(rs->stats + s, 0);
+}
+
+/*----------------------------------------------------------------
+ * RAID set management routines.
+ *--------------------------------------------------------------*/
+/*
+ * Begin small helper functions.
+ */
+/* Queue (optionally delayed) io work. */
+static void wake_do_raid_delayed(struct raid_set *rs, unsigned long delay)
+{
+	struct delayed_work *dws = &rs->io.dws;
+
+	cancel_delayed_work(dws);
+	queue_delayed_work(rs->io.wq, dws, delay);
+}
+
+/* Queue io work immediately (called from region hash too). */
+static INLINE void wake_do_raid(void *context)
+{
+	wake_do_raid_delayed(context, 0);
+}
+
+/* Wait until all io has been processed. */
+static INLINE void wait_ios(struct raid_set *rs)
+{
+	wait_event(rs->io.suspendq, !atomic_read(&rs->io.in_process));
+}
+
+/* Declare io queued to device. */
+static INLINE void io_dev_queued(struct raid_dev *dev)
+{
+	set_bit(IO_QUEUED, &dev->flags);
+}
+
+/* Io on device and reset ? */
+static inline int io_dev_clear(struct raid_dev *dev)
+{
+	return test_and_clear_bit(IO_QUEUED, &dev->flags);
+}
+
+/* Get an io reference. */
+static INLINE void io_get(struct raid_set *rs)
+{
+	int p = atomic_inc_return(&rs->io.in_process);
+
+	if (p > atomic_read(&rs->io.in_process_max))
+		atomic_set(&rs->io.in_process_max, p); /* REMOVEME: max. */
+}
+
+/* Put the io reference and conditionally wake io waiters. */
+static INLINE void io_put(struct raid_set *rs)
+{
+	/* Intel: rebuild data corrupter? */
+	if (!atomic_read(&rs->io.in_process)) {
+		DMERR("%s would go negative!!!", __func__);
+		return;
+	}
+
+	if (atomic_dec_and_test(&rs->io.in_process))
+		wake_up(&rs->io.suspendq);
+}
+
+/* Calculate device sector offset. */
+static INLINE sector_t _sector(struct raid_set *rs, struct bio *bio)
+{
+	sector_t sector = bio->bi_sector;
+
+	sector_div(sector, rs->set.data_devs);
+	return sector;
+}
+
+/* Test device operational. */
+static INLINE int dev_operational(struct raid_set *rs, unsigned p)
+{
+	return !test_bit(DEVICE_FAILED, &rs->dev[p].flags);
+}
+
+/* Return # of active stripes in stripe cache. */
+static INLINE int sc_active(struct stripe_cache *sc)
+{
+	return atomic_read(&sc->active_stripes);
+}
+
+/* Test io pending on stripe. */
+static INLINE int stripe_io(struct stripe *stripe)
+{
+	return atomic_read(&stripe->io.pending);
+}
+
+static INLINE void stripe_io_inc(struct stripe *stripe)
+{
+	atomic_inc(&stripe->io.pending);
+}
+
+static INLINE void stripe_io_dec(struct stripe *stripe)
+{
+	atomic_dec(&stripe->io.pending);
+}
+
+/* Wrapper needed by for_each_io_dev(). */
+static void _stripe_io_inc(struct stripe *stripe, unsigned p)
+{
+	stripe_io_inc(stripe);
+}
+
+/* Error a stripe. */
+static INLINE void stripe_error(struct stripe *stripe, struct page *page)
+{
+	SetStripeError(stripe);
+	SetPageError(page);
+	atomic_inc(RS(stripe->sc)->stats + S_STRIPE_ERROR);
+}
+
+/* Page IOed ok. */
+enum dirty_type { CLEAN, DIRTY };
+static INLINE void page_set(struct page *page, enum dirty_type type)
+{
+	switch (type) {
+	case DIRTY:
+		SetPageDirty(page);
+		AllowPageIO(page);
+		break;
+
+	case CLEAN:
+		ClearPageDirty(page);
+		break;
+
+	default:
+		BUG();
+	}
+
+	SetPageUptodate(page);
+	ClearPageError(page);
+}
+
+/* Return region state for a sector. */
+static INLINE int
+region_state(struct raid_set *rs, sector_t sector, unsigned long state)
+{
+	struct dm_rh_client *rh = rs->recover.rh;
+
+	return RSRecover(rs) ?
+	       (dm_rh_get_state(rh, dm_rh_sector_to_region(rh, sector), 1) &
+		state) : 0;
+}
+
+/* Check maximum devices which may fail in a raid set. */
+static inline int raid_set_degraded(struct raid_set *rs)
+{
+	return RSIoError(rs);
+}
+
+/* Check # of devices which may fail in a raid set. */
+static INLINE int raid_set_operational(struct raid_set *rs)
+{
+	/* Too many failed devices -> BAD. */
+	return atomic_read(&rs->set.failed_devs) <=
+	       rs->set.raid_type->parity_devs;
+}
+
+/*
+ * Return true in case a page_list should be read/written
+ *
+ * Conditions to read/write:
+ *	o 1st page in list not uptodate
+ *	o 1st page in list dirty
+ *	o if we optimized io away, we flag it using the pages checked bit.
+ */
+static INLINE unsigned page_io(struct page *page)
+{
+	/* Optimization: page was flagged to need io during first run. */
+	if (PagePrivate(page)) {
+		ClearPagePrivate(page);
+		return 1;
+	}
+
+	/* Avoid io if prohibited or a locked page. */
+	if (!PageIO(page) || PageLocked(page))
+		return 0;
+
+	if (!PageUptodate(page) || PageDirty(page)) {
+		/* Flag page needs io for second run optimization. */
+		SetPagePrivate(page);
+		return 1;
+	}
+
+	return 0;
+}
+
+/* Call a function on each page list needing io. */
+static INLINE unsigned
+for_each_io_dev(struct raid_set *rs, struct stripe *stripe,
+		void (*f_io)(struct stripe *stripe, unsigned p))
+{
+	unsigned p = rs->set.raid_devs, r = 0;
+
+	while (p--) {
+		if (page_io(PAGE(stripe, p))) {
+			f_io(stripe, p);
+			r++;
+		}
+	}
+
+	return r;
+}
+
+/* Reconstruct a particular device ?. */
+static INLINE int dev_to_init(struct raid_set *rs)
+{
+	return rs->set.dev_to_init > -1;
+}
+
+/*
+ * Index of device to calculate parity on.
+ * Either the parity device index *or* the selected device to init
+ * after a spare replacement.
+ */
+static INLINE unsigned dev_for_parity(struct stripe *stripe)
+{
+	struct raid_set *rs = RS(stripe->sc);
+
+	return dev_to_init(rs) ? rs->set.dev_to_init : stripe->idx.parity;
+}
+
+/* Return the index of the device to be recovered. */
+static int idx_get(struct raid_set *rs)
+{
+	/* Avoid to read in the pages to be reconstructed anyway. */
+	if (dev_to_init(rs))
+		return rs->set.dev_to_init;
+	else if (rs->set.raid_type->level == raid4)
+		return rs->set.pi;
+
+	return -1;
+}
+
+/* RAID set congested function. */
+static int raid_set_congested(void *congested_data, int bdi_bits)
+{
+	struct raid_set *rs = congested_data;
+	int r = 0; /* Assume uncongested. */
+	unsigned p = rs->set.raid_devs;
+
+	/* If any of our component devices are overloaded. */
+	while (p--) {
+		struct request_queue *q = bdev_get_queue(rs->dev[p].dev->bdev);
+
+		r |= bdi_congested(&q->backing_dev_info, bdi_bits);
+	}
+
+	/* REMOVEME: statistics. */
+	atomic_inc(rs->stats + (r ? S_CONGESTED : S_NOT_CONGESTED));
+	return r;
+}
+
+/* Display RAID set dead message once. */
+static void raid_set_dead(struct raid_set *rs)
+{
+	if (!TestSetRSDead(rs)) {
+		unsigned p;
+		char buf[BDEVNAME_SIZE];
+
+		DMERR("FATAL: too many devices failed -> RAID set dead");
+
+		for (p = 0; p < rs->set.raid_devs; p++) {
+			if (!dev_operational(rs, p))
+				DMERR("device /dev/%s failed",
+				      bdevname(rs->dev[p].dev->bdev, buf));
+		}
+	}
+}
+
+/* RAID set degrade check. */
+static INLINE int
+raid_set_check_and_degrade(struct raid_set *rs,
+			   struct stripe *stripe, unsigned p)
+{
+	if (test_and_set_bit(DEVICE_FAILED, &rs->dev[p].flags))
+		return -EPERM;
+
+	/* Through an event in case of member device errors. */
+	dm_table_event(rs->ti->table);
+	atomic_inc(&rs->set.failed_devs);
+
+	/* Only log the first member error. */
+	if (!TestSetRSIoError(rs)) {
+		char buf[BDEVNAME_SIZE];
+
+		/* Store index for recovery. */
+		mb();
+		rs->set.ei = p;
+		mb();
+
+		DMERR("CRITICAL: %sio error on device /dev/%s "
+		      "in region=%llu; DEGRADING RAID set",
+		      stripe ? "" : "FAKED ",
+		      bdevname(rs->dev[p].dev->bdev, buf),
+		      (unsigned long long) (stripe ? stripe->key : 0));
+		DMERR("further device error messages suppressed");
+	}
+
+	return 0;
+}
+
+static void
+raid_set_check_degrade(struct raid_set *rs, struct stripe *stripe)
+{
+	unsigned p = rs->set.raid_devs;
+
+	while (p--) {
+		struct page *page = PAGE(stripe, p);
+
+		if (PageError(page)) {
+			ClearPageError(page);
+			raid_set_check_and_degrade(rs, stripe, p);
+		}
+	}
+}
+
+/* RAID set upgrade check. */
+static int raid_set_check_and_upgrade(struct raid_set *rs, unsigned p)
+{
+	if (!test_and_clear_bit(DEVICE_FAILED, &rs->dev[p].flags))
+		return -EPERM;
+
+	if (atomic_dec_and_test(&rs->set.failed_devs)) {
+		ClearRSIoError(rs);
+		rs->set.ei = -1;
+	}
+
+	return 0;
+}
+
+/* Lookup a RAID device by name or by major:minor number. */
+union dev_lookup {
+	const char *dev_name;
+	struct raid_dev *dev;
+};
+enum lookup_type { byname, bymajmin, bynumber };
+static int raid_dev_lookup(struct raid_set *rs, enum lookup_type by,
+			   union dev_lookup *dl)
+{
+	unsigned p;
+
+	/*
+	 * Must be an incremental loop, because the device array
+	 * can have empty slots still on calls from raid_ctr()
+	 */
+	for (p = 0; p < rs->set.raid_devs; p++) {
+		char buf[BDEVNAME_SIZE];
+		struct raid_dev *dev = rs->dev + p;
+
+		if (!dev->dev)
+			break;
+
+		/* Format dev string appropriately if necessary. */
+		if (by == byname)
+			bdevname(dev->dev->bdev, buf);
+		else if (by == bymajmin)
+			format_dev_t(buf, dev->dev->bdev->bd_dev);
+
+		/* Do the actual check. */
+		if (by == bynumber) {
+			if (dl->dev->dev->bdev->bd_dev ==
+			    dev->dev->bdev->bd_dev)
+				return p;
+		} else if (!strcmp(dl->dev_name, buf))
+			return p;
+	}
+
+	return -ENODEV;
+}
+
+/* End io wrapper. */
+static INLINE void
+_bio_endio(struct raid_set *rs, struct bio *bio, int error)
+{
+	/* REMOVEME: statistics. */
+	atomic_inc(rs->stats + (bio_data_dir(bio) == WRITE ?
+		   S_BIOS_ENDIO_WRITE : S_BIOS_ENDIO_READ));
+	bio_endio(bio, error);
+	io_put(rs);		/* Wake any suspend waiters. */
+}
+
+/*
+ * End small helper functions.
+ */
+
+
+/*
+ * Stripe hash functions
+ */
+/* Initialize/destroy stripe hash. */
+static int hash_init(struct stripe_hash *hash, unsigned stripes)
+{
+	unsigned buckets = 2, max_buckets = stripes / 4;
+	unsigned hash_primes[] = {
+		/* Table of primes for hash_fn/table size optimization. */
+		3, 7, 13, 27, 53, 97, 193, 389, 769,
+		1543, 3079, 6151, 12289, 24593,
+	};
+
+	/* Calculate number of buckets (2^^n <= stripes / 4). */
+	while (buckets < max_buckets)
+		buckets <<= 1;
+
+	/* Allocate stripe hash. */
+	hash->hash = vmalloc(buckets * sizeof(*hash->hash));
+	if (!hash->hash)
+		return -ENOMEM;
+
+	hash->buckets = buckets;
+	hash->mask = buckets - 1;
+	hash->shift = ffs(buckets);
+	if (hash->shift > ARRAY_SIZE(hash_primes) + 1)
+		hash->shift = ARRAY_SIZE(hash_primes) + 1;
+
+	BUG_ON(hash->shift - 2 > ARRAY_SIZE(hash_primes) + 1);
+	hash->prime = hash_primes[hash->shift - 2];
+
+	/* Initialize buckets. */
+	while (buckets--)
+		INIT_LIST_HEAD(hash->hash + buckets);
+
+	return 0;
+}
+
+static INLINE void hash_exit(struct stripe_hash *hash)
+{
+	if (hash->hash) {
+		vfree(hash->hash);
+		hash->hash = NULL;
+	}
+}
+
+/* List add (head/tail/locked/unlocked) inlines. */
+enum list_lock_type { LIST_LOCKED, LIST_UNLOCKED };
+#define	LIST_DEL(name, list) \
+static void stripe_ ## name ## _del(struct stripe *stripe, \
+				    enum list_lock_type lock) { \
+	struct list_head *lh = stripe->lists + (list); \
+	spinlock_t *l = NULL; \
+\
+	if (lock == LIST_LOCKED) { \
+		l = stripe->sc->locks + LOCK_LRU; \
+		spin_lock_irq(l); \
+	} \
+\
+\
+	if (!list_empty(lh)) \
+		list_del_init(lh); \
+\
+	if (lock == LIST_LOCKED) \
+		spin_unlock_irq(l); \
+}
+
+LIST_DEL(hash, LIST_HASH)
+LIST_DEL(lru, LIST_LRU)
+#undef LIST_DEL
+
+enum list_pos_type { POS_HEAD, POS_TAIL };
+#define	LIST_ADD(name, list) \
+static void stripe_ ## name ## _add(struct stripe *stripe, \
+				    enum list_pos_type pos, \
+				    enum list_lock_type lock) { \
+	struct list_head *lh = stripe->lists + (list); \
+	struct stripe_cache *sc = stripe->sc; \
+	spinlock_t *l = NULL; \
+\
+	if (lock == LIST_LOCKED) { \
+		l = sc->locks + LOCK_LRU; \
+		spin_lock_irq(l); \
+	} \
+\
+	if (list_empty(lh)) { \
+		if (pos == POS_HEAD) \
+			list_add(lh, sc->lists + (list)); \
+		else \
+			list_add_tail(lh, sc->lists + (list)); \
+	} \
+\
+	if (lock == LIST_LOCKED) \
+		spin_unlock_irq(l); \
+}
+
+LIST_ADD(endio, LIST_ENDIO)
+LIST_ADD(io, LIST_IO)
+LIST_ADD(lru, LIST_LRU)
+#undef LIST_ADD
+
+#define POP(list) \
+	do { \
+		if (list_empty(sc->lists + list)) \
+			stripe = NULL; \
+		else { \
+			stripe = list_first_entry(&sc->lists[list], \
+						  struct stripe, \
+						  lists[list]); \
+			list_del_init(&stripe->lists[list]); \
+		} \
+	} while (0);
+
+/* Pop an available stripe off the lru list. */
+static struct stripe *stripe_lru_pop(struct stripe_cache *sc)
+{
+	struct stripe *stripe;
+	spinlock_t *lock = sc->locks + LOCK_LRU;
+
+	spin_lock_irq(lock);
+	POP(LIST_LRU);
+	spin_unlock_irq(lock);
+
+	if (stripe)
+		/* Remove from hash before reuse. */
+		stripe_hash_del(stripe, LIST_UNLOCKED);
+
+	return stripe;
+}
+
+static inline unsigned hash_fn(struct stripe_hash *hash, sector_t key)
+{
+	return (unsigned) (((key * hash->prime) >> hash->shift) & hash->mask);
+}
+
+static inline struct list_head *
+hash_bucket(struct stripe_hash *hash, sector_t key)
+{
+	return hash->hash + hash_fn(hash, key);
+}
+
+/* Insert an entry into a hash. */
+static inline void hash_insert(struct stripe_hash *hash, struct stripe *stripe)
+{
+	list_add(stripe->lists + LIST_HASH, hash_bucket(hash, stripe->key));
+}
+
+/* Insert an entry into the stripe hash. */
+static inline void
+sc_insert(struct stripe_cache *sc, struct stripe *stripe)
+{
+	hash_insert(&sc->hash, stripe);
+}
+
+/* Lookup an entry in the stripe hash. */
+static inline struct stripe *
+stripe_lookup(struct stripe_cache *sc, sector_t key)
+{
+	unsigned c = 0;
+	struct stripe *stripe;
+	struct list_head *bucket = hash_bucket(&sc->hash, key);
+
+	list_for_each_entry(stripe, bucket, lists[LIST_HASH]) {
+		/* REMOVEME: statisics. */
+		if (++c > atomic_read(RS(sc)->stats + S_MAX_LOOKUP))
+			atomic_set(RS(sc)->stats + S_MAX_LOOKUP, c);
+
+		if (stripe->key == key)
+			return stripe;
+	}
+
+	return NULL;
+}
+
+/* Resize the stripe cache hash on size changes. */
+static int hash_resize(struct stripe_cache *sc)
+{
+	/* Resize threshold reached? */
+	if (atomic_read(&sc->stripes) > 2 * atomic_read(&sc->stripes_last)
+	    || atomic_read(&sc->stripes) < atomic_read(&sc->stripes_last) / 4) {
+		int r;
+		struct stripe_hash hash, hash_tmp;
+		spinlock_t *lock;
+
+		r = hash_init(&hash, atomic_read(&sc->stripes));
+		if (r)
+			return r;
+
+		lock = sc->locks + LOCK_LRU;
+		spin_lock_irq(lock);
+		if (sc->hash.hash) {
+			unsigned b = sc->hash.buckets;
+			struct list_head *pos, *tmp;
+
+			/* Walk old buckets and insert into new. */
+			while (b--) {
+				list_for_each_safe(pos, tmp, sc->hash.hash + b)
+				    hash_insert(&hash,
+						list_entry(pos, struct stripe,
+							   lists[LIST_HASH]));
+			}
+
+		}
+
+		memcpy(&hash_tmp, &sc->hash, sizeof(hash_tmp));
+		memcpy(&sc->hash, &hash, sizeof(sc->hash));
+		atomic_set(&sc->stripes_last, atomic_read(&sc->stripes));
+		spin_unlock_irq(lock);
+
+		hash_exit(&hash_tmp);
+	}
+
+	return 0;
+}
+
+/*
+ * Stripe cache locking functions
+ */
+/* Dummy lock function for local RAID4+5. */
+static void *no_lock(sector_t key, enum dm_lock_type type)
+{
+	return &no_lock;
+}
+
+/* Dummy unlock function for local RAID4+5. */
+static void no_unlock(void *lock_handle)
+{
+}
+
+/* No locking (for local RAID 4+5). */
+static struct dm_raid45_locking_type locking_none = {
+	.lock = no_lock,
+	.unlock = no_unlock,
+};
+
+/* Clustered RAID 4+5. */
+/* FIXME: code this. */
+static struct dm_raid45_locking_type locking_cluster = {
+	.lock = no_lock,
+	.unlock = no_unlock,
+};
+
+/* Lock a stripe (for clustering). */
+static int
+stripe_lock(struct raid_set *rs, struct stripe *stripe, int rw, sector_t key)
+{
+	stripe->lock = rs->locking->lock(key, rw == READ ? DM_RAID45_SHARED :
+							   DM_RAID45_EX);
+	return stripe->lock ? 0 : -EPERM;
+}
+
+/* Unlock a stripe (for clustering). */
+static void stripe_unlock(struct raid_set *rs, struct stripe *stripe)
+{
+	rs->locking->unlock(stripe->lock);
+	stripe->lock = NULL;
+}
+
+/*
+ * Stripe cache functions.
+ */
+/*
+ * Invalidate all page lists pages of a stripe.
+ *
+ * I only keep state for the whole list in the first page.
+ */
+static INLINE void
+stripe_pages_invalidate(struct stripe *stripe)
+{
+	unsigned p = RS(stripe->sc)->set.raid_devs;
+
+	while (p--) {
+		struct page *page = PAGE(stripe, p);
+
+		ProhibitPageIO(page);
+		ClearPageChecked(page);
+		ClearPageDirty(page);
+		ClearPageError(page);
+		clear_page_locked(page);
+		ClearPagePrivate(page);
+		ClearPageUptodate(page);
+	}
+}
+
+/* Prepare stripe for (re)use. */
+static INLINE void stripe_invalidate(struct stripe *stripe)
+{
+	stripe->io.flags = 0;
+	stripe_pages_invalidate(stripe);
+}
+
+/* Allow io on all chunks of a stripe. */
+static INLINE void stripe_allow_io(struct stripe *stripe)
+{
+	unsigned p = RS(stripe->sc)->set.raid_devs;
+
+	while (p--)
+		AllowPageIO(PAGE(stripe, p));
+}
+
+/* Initialize a stripe. */
+static void
+stripe_init(struct stripe_cache *sc, struct stripe *stripe)
+{
+	unsigned p = RS(sc)->set.raid_devs;
+	unsigned i;
+
+	/* Work all io chunks. */
+	while (p--) {
+		struct stripe_set *ss = stripe->ss + p;
+
+		stripe->obj[p].private = ss;
+		ss->stripe = stripe;
+
+		i = ARRAY_SIZE(ss->bl);
+		while (i--)
+			bio_list_init(ss->bl + i);
+	}
+
+	stripe->sc = sc;
+
+	i = ARRAY_SIZE(stripe->lists);
+	while (i--)
+		INIT_LIST_HEAD(stripe->lists + i);
+
+	atomic_set(&stripe->cnt, 0);
+	atomic_set(&stripe->io.pending, 0);
+
+	stripe_invalidate(stripe);
+}
+
+/* Number of pages per chunk. */
+static inline unsigned chunk_pages(unsigned io_size)
+{
+	return dm_div_up(io_size, SECTORS_PER_PAGE);
+}
+
+/* Number of pages per stripe. */
+static inline unsigned stripe_pages(struct raid_set *rs, unsigned io_size)
+{
+	return chunk_pages(io_size) * rs->set.raid_devs;
+}
+
+/* Initialize part of page_list (recovery). */
+static INLINE void stripe_zero_pl_part(struct stripe *stripe, unsigned p,
+				       unsigned start, unsigned count)
+{
+	unsigned pages = chunk_pages(count);
+	/* Get offset into the page_list. */
+	struct page_list *pl = pl_elem(PL(stripe, p), start / SECTORS_PER_PAGE);
+
+	BUG_ON(!pl);
+	while (pl && pages--) {
+		BUG_ON(!pl->page);
+		memset(page_address(pl->page), 0, PAGE_SIZE);
+		pl = pl->next;
+	}
+}
+
+/* Initialize parity chunk of stripe. */
+static INLINE void stripe_zero_chunk(struct stripe *stripe, unsigned p)
+{
+	stripe_zero_pl_part(stripe, p, 0, stripe->io.size);
+}
+
+/* Return dynamic stripe structure size. */
+static INLINE size_t stripe_size(struct raid_set *rs)
+{
+	return sizeof(struct stripe) +
+		      rs->set.raid_devs * sizeof(struct stripe_set);
+}
+
+/* Allocate a stripe and its memory object. */
+/* XXX adjust to cope with stripe cache and recovery stripe caches. */
+enum grow { SC_GROW, SC_KEEP };
+static struct stripe *stripe_alloc(struct stripe_cache *sc,
+				   struct dm_mem_cache_client *mc,
+				   enum grow grow)
+{
+	int r;
+	struct stripe *stripe;
+
+	stripe = kmem_cache_zalloc(sc->kc.cache, GFP_KERNEL);
+	if (stripe) {
+		/* Grow the dm-mem-cache by one object. */
+		if (grow == SC_GROW) {
+			r = dm_mem_cache_grow(mc, 1);
+			if (r)
+				goto err_free;
+		}
+
+		stripe->obj = dm_mem_cache_alloc(mc);
+		if (!stripe->obj)
+			goto err_shrink;
+
+		stripe_init(sc, stripe);
+	}
+
+	return stripe;
+
+err_shrink:
+	if (grow == SC_GROW)
+		dm_mem_cache_shrink(mc, 1);
+err_free:
+	kmem_cache_free(sc->kc.cache, stripe);
+	return NULL;
+}
+
+/*
+ * Free a stripes memory object, shrink the
+ * memory cache and free the stripe itself
+ */
+static void stripe_free(struct stripe *stripe, struct dm_mem_cache_client *mc)
+{
+	dm_mem_cache_free(mc, stripe->obj);
+	dm_mem_cache_shrink(mc, 1);
+	kmem_cache_free(stripe->sc->kc.cache, stripe);
+}
+
+/* Free the recovery stripe. */
+static void stripe_recover_free(struct raid_set *rs)
+{
+	struct recover *rec = &rs->recover;
+	struct list_head *stripes = &rec->stripes;
+
+	while (!list_empty(stripes)) {
+		struct stripe *stripe = list_first_entry(stripes, struct stripe,
+							 lists[LIST_RECOVER]);
+		list_del(stripe->lists + LIST_RECOVER);
+		stripe_free(stripe, rec->mem_cache_client);
+	}
+}
+
+/* Push a stripe safely onto the endio list to be handled by do_endios(). */
+static INLINE void stripe_endio_push(struct stripe *stripe)
+{
+	int wake;
+	unsigned long flags;
+	struct stripe_cache *sc = stripe->sc;
+	spinlock_t *lock = sc->locks + LOCK_ENDIO;
+
+	spin_lock_irqsave(lock, flags);
+	wake = list_empty(sc->lists + LIST_ENDIO);
+	stripe_endio_add(stripe, POS_HEAD, LIST_UNLOCKED);
+	spin_unlock_irqrestore(lock, flags);
+
+	if (wake)
+		wake_do_raid(RS(sc));
+}
+
+/* Protected check for stripe cache endio list empty. */
+static INLINE int stripe_endio_empty(struct stripe_cache *sc)
+{
+	int r;
+	spinlock_t *lock = sc->locks + LOCK_ENDIO;
+
+	spin_lock_irq(lock);
+	r = list_empty(sc->lists + LIST_ENDIO);
+	spin_unlock_irq(lock);
+
+	return r;
+}
+
+/* Pop a stripe off safely off the endio list. */
+static struct stripe *stripe_endio_pop(struct stripe_cache *sc)
+{
+	struct stripe *stripe;
+	spinlock_t *lock = sc->locks + LOCK_ENDIO;
+
+	/* This runs in parallel with endio(). */
+	spin_lock_irq(lock);
+	POP(LIST_ENDIO)
+	spin_unlock_irq(lock);
+	return stripe;
+}
+
+#undef POP
+
+/* Evict stripe from cache. */
+static void stripe_evict(struct stripe *stripe)
+{
+	struct raid_set *rs = RS(stripe->sc);
+	stripe_hash_del(stripe, LIST_UNLOCKED);	/* Take off hash. */
+
+	if (list_empty(stripe->lists + LIST_LRU)) {
+		stripe_lru_add(stripe, POS_TAIL, LIST_LOCKED);
+		atomic_inc(rs->stats + S_EVICT); /* REMOVEME: statistics. */
+	}
+}
+
+/* Grow stripe cache. */
+static int
+sc_grow(struct stripe_cache *sc, unsigned stripes, enum grow grow)
+{
+	int r = 0;
+	struct raid_set *rs = RS(sc);
+
+	/* Try to allocate this many (additional) stripes. */
+	while (stripes--) {
+		struct stripe *stripe =
+			stripe_alloc(sc, sc->mem_cache_client, grow);
+
+		if (likely(stripe)) {
+			stripe->io.size = rs->set.io_size;
+			stripe_lru_add(stripe, POS_TAIL, LIST_LOCKED);
+			atomic_inc(&sc->stripes);
+		} else {
+			r = -ENOMEM;
+			break;
+		}
+	}
+
+	ClearRSScBusy(rs);
+	return r ? r : hash_resize(sc);
+}
+
+/* Shrink stripe cache. */
+static int sc_shrink(struct stripe_cache *sc, unsigned stripes)
+{
+	int r = 0;
+
+	/* Try to get unused stripe from LRU list. */
+	while (stripes--) {
+		struct stripe *stripe;
+
+		stripe = stripe_lru_pop(sc);
+		if (stripe) {
+			/* An lru stripe may never have ios pending! */
+			BUG_ON(stripe_io(stripe));
+			stripe_free(stripe, sc->mem_cache_client);
+			atomic_dec(&sc->stripes);
+		} else {
+			r = -ENOENT;
+			break;
+		}
+	}
+
+	/* Check if stats are still sane. */
+	if (atomic_read(&sc->max_active_stripes) >
+	    atomic_read(&sc->stripes))
+		atomic_set(&sc->max_active_stripes, 0);
+
+	if (r)
+		return r;
+
+	ClearRSScBusy(RS(sc));
+	return hash_resize(sc);
+}
+
+/* Create stripe cache. */
+static int sc_init(struct raid_set *rs, unsigned stripes)
+{
+	unsigned i, nr;
+	struct stripe_cache *sc = &rs->sc;
+	struct stripe *stripe;
+	struct recover *rec = &rs->recover;
+
+	/* Initialize lists and locks. */
+	i = ARRAY_SIZE(sc->lists);
+	while (i--)
+		INIT_LIST_HEAD(sc->lists + i);
+
+	i = NR_LOCKS;
+	while (i--)
+		spin_lock_init(sc->locks + i);
+
+	/* Initialize atomic variables. */
+	atomic_set(&sc->stripes, 0);
+	atomic_set(&sc->stripes_last, 0);
+	atomic_set(&sc->stripes_to_shrink, 0);
+	atomic_set(&sc->active_stripes, 0);
+	atomic_set(&sc->max_active_stripes, 0);	/* REMOVEME: statistics. */
+
+	/*
+	 * We need a runtime unique # to suffix the kmem cache name
+	 * because we'll have one for each active RAID set.
+	 */
+	nr = atomic_inc_return(&_stripe_sc_nr);
+	sprintf(sc->kc.name, "%s_%d", TARGET, nr);
+	sc->kc.cache = kmem_cache_create(sc->kc.name, stripe_size(rs),
+					 0, 0, NULL);
+	if (!sc->kc.cache)
+		return -ENOMEM;
+
+	/* Create memory cache client context for RAID stripe cache. */
+	sc->mem_cache_client =
+		dm_mem_cache_client_create(stripes, rs->set.raid_devs,
+					   chunk_pages(rs->set.io_size));
+	if (IS_ERR(sc->mem_cache_client))
+		return PTR_ERR(sc->mem_cache_client);
+
+	/* Create memory cache client context for RAID recovery stripe(s). */
+	rec->mem_cache_client =
+		dm_mem_cache_client_create(MAX_RECOVER, rs->set.raid_devs,
+					   chunk_pages(rec->io_size));
+	if (IS_ERR(rec->mem_cache_client))
+		return PTR_ERR(rec->mem_cache_client);
+
+	/* Allocate stripe for set recovery. */
+	/* XXX: cope with MAX_RECOVERY. */
+	INIT_LIST_HEAD(&rec->stripes);
+	for (i = 0; i < MAX_RECOVER; i++) {
+		stripe = stripe_alloc(sc, rec->mem_cache_client, SC_KEEP);
+		if (!stripe)
+			return -ENOMEM;
+
+		SetStripeRecover(stripe);
+		stripe->io.size = rec->io_size;
+		list_add(stripe->lists + LIST_RECOVER, &rec->stripes);
+	}
+
+	/*
+	 * Allocate the stripe objetcs from the
+	 * cache and add them to the LRU list.
+	 */
+	return sc_grow(sc, stripes, SC_KEEP);
+}
+
+/* Destroy the stripe cache. */
+static void sc_exit(struct stripe_cache *sc)
+{
+	if (sc->kc.cache) {
+		BUG_ON(sc_shrink(sc, atomic_read(&sc->stripes)));
+		kmem_cache_destroy(sc->kc.cache);
+	}
+
+	if (sc->mem_cache_client)
+		dm_mem_cache_client_destroy(sc->mem_cache_client);
+
+	ClearRSRecover(RS(sc));
+	stripe_recover_free(RS(sc));
+	if (RS(sc)->recover.mem_cache_client)
+		dm_mem_cache_client_destroy(RS(sc)->recover.mem_cache_client);
+
+	hash_exit(&sc->hash);
+}
+
+/*
+ * Calculate RAID address
+ *
+ * Delivers tuple with the index of the data disk holding the chunk
+ * in the set, the parity disks index and the start of the stripe
+ * within the address space of the set (used as the stripe cache hash key).
+ */
+/* thx MD. */
+static struct address *
+raid_address(struct raid_set *rs, sector_t sector, struct address *addr)
+{
+	unsigned data_devs = rs->set.data_devs, di, pi,
+		 raid_devs = rs->set.raid_devs;
+	sector_t stripe, tmp;
+
+	/*
+	 * chunk_number = sector / chunk_size
+	 * stripe = chunk_number / data_devs
+	 * di = stripe % data_devs;
+	 */
+	stripe = sector >> rs->set.chunk_shift;
+	di = sector_div(stripe, data_devs);
+
+	switch (rs->set.raid_type->level) {
+	case raid5:
+		tmp = stripe;
+		pi = sector_div(tmp, raid_devs);
+
+		switch (rs->set.raid_type->algorithm) {
+		case left_asym:		/* Left asymmetric. */
+			pi = data_devs - pi;
+		case right_asym:	/* Right asymmetric. */
+			if (di >= pi)
+				di++;
+			break;
+
+		case left_sym:		/* Left symmetric. */
+			pi = data_devs - pi;
+		case right_sym:		/* Right symmetric. */
+			di = (pi + di + 1) % raid_devs;
+			break;
+
+		default:
+			DMERR("Unknown RAID algorithm %d",
+			      rs->set.raid_type->algorithm);
+			goto out;
+		}
+
+		break;
+
+	case raid4:
+		pi = rs->set.pi;
+		if (di >= pi)
+			di++;
+		break;
+
+	default:
+		DMERR("Unknown RAID level %d", rs->set.raid_type->level);
+		goto out;
+	}
+
+	/*
+	 * Hash key = start offset on any single device of the RAID set;
+	 * adjusted in case io size differs from chunk size.
+	 */
+	addr->key = (stripe << rs->set.chunk_shift) +
+		    (sector & rs->set.io_shift_mask);
+	addr->di = di;
+	addr->pi = pi;
+
+out:
+	return addr;
+}
+
+/*
+ * Copy data across between stripe pages and bio vectors.
+ *
+ * Pay attention to data alignment in stripe and bio pages.
+ */
+static void
+bio_copy_page_list(int rw, struct stripe *stripe,
+		   struct page_list *pl, struct bio *bio)
+{
+	unsigned i, page_offset;
+	void *page_addr;
+	struct raid_set *rs = RS(stripe->sc);
+	struct bio_vec *bv;
+
+	/* Get start page in page list for this sector. */
+	i = (bio->bi_sector & rs->set.io_mask) / SECTORS_PER_PAGE;
+	pl = pl_elem(pl, i);
+
+	page_addr = page_address(pl->page);
+	page_offset = to_bytes(bio->bi_sector & (SECTORS_PER_PAGE - 1));
+
+	/* Walk all segments and copy data across between bio_vecs and pages. */
+	bio_for_each_segment(bv, bio, i) {
+		int len = bv->bv_len, size;
+		unsigned bio_offset = 0;
+		void *bio_addr = __bio_kmap_atomic(bio, i, KM_USER0);
+redo:
+		size = (page_offset + len > PAGE_SIZE) ?
+		       PAGE_SIZE - page_offset : len;
+
+		if (rw == READ)
+			memcpy(bio_addr + bio_offset,
+			       page_addr + page_offset, size);
+		else
+			memcpy(page_addr + page_offset,
+			       bio_addr + bio_offset, size);
+
+		page_offset += size;
+		if (page_offset == PAGE_SIZE) {
+			/*
+			 * We reached the end of the chunk page ->
+			 * need refer to the next one to copy more data.
+			 */
+			len -= size;
+			if (len) {
+				/* Get next page. */
+				pl = pl->next;
+				BUG_ON(!pl);
+				page_addr = page_address(pl->page);
+				page_offset = 0;
+				bio_offset += size;
+				/* REMOVEME: statistics. */
+				atomic_inc(rs->stats + S_BIO_COPY_PL_NEXT);
+				goto redo;
+			}
+		}
+
+		__bio_kunmap_atomic(bio_addr, KM_USER0);
+	}
+}
+
+/*
+ * Xor optimization macros.
+ */
+/* Xor data pointer declaration and initialization macros. */
+#define DECLARE_2	unsigned long *d0 = data[0], *d1 = data[1]
+#define DECLARE_3	DECLARE_2, *d2 = data[2]
+#define DECLARE_4	DECLARE_3, *d3 = data[3]
+#define DECLARE_5	DECLARE_4, *d4 = data[4]
+#define DECLARE_6	DECLARE_5, *d5 = data[5]
+#define DECLARE_7	DECLARE_6, *d6 = data[6]
+#define DECLARE_8	DECLARE_7, *d7 = data[7]
+
+/* Xor unrole macros. */
+#define D2(n)	d0[n] = d0[n] ^ d1[n]
+#define D3(n)	D2(n) ^ d2[n]
+#define D4(n)	D3(n) ^ d3[n]
+#define D5(n)	D4(n) ^ d4[n]
+#define D6(n)	D5(n) ^ d5[n]
+#define D7(n)	D6(n) ^ d6[n]
+#define D8(n)	D7(n) ^ d7[n]
+
+#define	X_2(macro, offset)	macro(offset); macro(offset + 1);
+#define	X_4(macro, offset)	X_2(macro, offset); X_2(macro, offset + 2);
+#define	X_8(macro, offset)	X_4(macro, offset); X_4(macro, offset + 4);
+#define	X_16(macro, offset)	X_8(macro, offset); X_8(macro, offset + 8);
+#define	X_32(macro, offset)	X_16(macro, offset); X_16(macro, offset + 16);
+#define	X_64(macro, offset)	X_32(macro, offset); X_32(macro, offset + 32);
+
+/* Define a _xor_#chunks_#xors_per_run() function. */
+#define	_XOR(chunks, xors_per_run) \
+static void _xor ## chunks ## _ ## xors_per_run(unsigned long **data) \
+{ \
+	unsigned end = XOR_SIZE / sizeof(data[0]), i; \
+	DECLARE_ ## chunks; \
+\
+	for (i = 0; i < end; i += xors_per_run) { \
+		X_ ## xors_per_run(D ## chunks, i); \
+	} \
+}
+
+/* Define xor functions for 2 - 8 chunks. */
+#define	MAKE_XOR_PER_RUN(xors_per_run) \
+	_XOR(2, xors_per_run); _XOR(3, xors_per_run); \
+	_XOR(4, xors_per_run); _XOR(5, xors_per_run); \
+	_XOR(6, xors_per_run); _XOR(7, xors_per_run); \
+	_XOR(8, xors_per_run);
+
+MAKE_XOR_PER_RUN(8)	/* Define _xor_*_8() functions. */
+MAKE_XOR_PER_RUN(16)	/* Define _xor_*_16() functions. */
+MAKE_XOR_PER_RUN(32)	/* Define _xor_*_32() functions. */
+MAKE_XOR_PER_RUN(64)	/* Define _xor_*_64() functions. */
+
+#define MAKE_XOR(xors_per_run) \
+struct { \
+	void (*f)(unsigned long **); \
+} static xor_funcs ## xors_per_run[] = { \
+	{ NULL }, \
+	{ NULL }, \
+	{ _xor2_ ## xors_per_run }, \
+	{ _xor3_ ## xors_per_run }, \
+	{ _xor4_ ## xors_per_run }, \
+	{ _xor5_ ## xors_per_run }, \
+	{ _xor6_ ## xors_per_run }, \
+	{ _xor7_ ## xors_per_run }, \
+	{ _xor8_ ## xors_per_run }, \
+}; \
+\
+static void xor_ ## xors_per_run(unsigned n, unsigned long **data) \
+{ \
+	/* Call respective function for amount of chunks. */ \
+	xor_funcs ## xors_per_run[n].f(data); \
+}
+
+/* Define xor_8() - xor_64 functions. */
+MAKE_XOR(8)
+MAKE_XOR(16)
+MAKE_XOR(32)
+MAKE_XOR(64)
+
+/* Maximum number of chunks, which can be xor'ed in one go. */
+#define	XOR_CHUNKS_MAX	(ARRAY_SIZE(xor_funcs8) - 1)
+
+struct xor_func {
+	xor_function_t f;
+	const char *name;
+} static xor_funcs[] = {
+	{xor_8,   "xor_8"},
+	{xor_16,  "xor_16"},
+	{xor_32,  "xor_32"},
+	{xor_64,  "xor_64"},
+};
+
+/*
+ * Calculate crc.
+ *
+ * This indexes into the page list of the stripe.
+ *
+ * All chunks will be xored into the parity chunk
+ * in maximum groups of xor.chunks.
+ *
+ * FIXME: try mapping the pages on discontiguous memory.
+ */
+static void xor(struct stripe *stripe, unsigned pi, unsigned sector)
+{
+	struct raid_set *rs = RS(stripe->sc);
+	unsigned max_chunks = rs->xor.chunks, n, p;
+	unsigned o = sector / SECTORS_PER_PAGE; /* Offset into the page_list. */
+	unsigned long **d = rs->data;
+	xor_function_t xor_f = rs->xor.f->f;
+
+	/* Address of parity page to xor into. */
+	d[0] = page_address(pl_elem(PL(stripe, pi), o)->page);
+
+	/* Preset pointers to data pages. */
+	for (n = 1, p = rs->set.raid_devs; p--; ) {
+		if (p != pi && PageIO(PAGE(stripe, p)))
+			d[n++] = page_address(pl_elem(PL(stripe, p), o)->page);
+
+		/* If max chunks -> xor .*/
+		if (n == max_chunks) {
+			xor_f(n, d);
+			n = 1;
+		}
+	}
+
+	/* If chunks -> xor. */
+	if (n > 1)
+		xor_f(n, d);
+
+	/* Set parity page uptodate and clean. */
+	page_set(PAGE(stripe, pi), CLEAN);
+}
+
+/* Common xor loop through all stripe page lists. */
+static void common_xor(struct stripe *stripe, sector_t count,
+		       unsigned off, unsigned p)
+{
+	unsigned sector;
+
+	for (sector = off; sector < count; sector += SECTORS_PER_XOR)
+		xor(stripe, p, sector);
+
+	atomic_inc(RS(stripe->sc)->stats + S_XORS); /* REMOVEME: statistics. */
+}
+
+/*
+ * Calculate parity sectors on intact stripes.
+ *
+ * Need to calculate raid address for recover stripe, because its
+ * chunk sizes differs and is typically larger than io chunk size.
+ */
+static void parity_xor(struct stripe *stripe)
+{
+	struct raid_set *rs = RS(stripe->sc);
+	unsigned chunk_size = rs->set.chunk_size,
+		 io_size = stripe->io.size,
+		 xor_size = chunk_size > io_size ? io_size : chunk_size;
+	sector_t off;
+
+	/* This can be the recover stripe with a larger io size. */
+	for (off = 0; off < io_size; off += xor_size) {
+		unsigned pi;
+
+		/*
+		 * Recover stripe likely is bigger than regular io
+		 * ones and has no precalculated parity disk index ->
+		 * need to calculate RAID address.
+		 */
+		if (unlikely(StripeRecover(stripe))) {
+			struct address addr;
+
+			raid_address(rs,
+				     (stripe->key + off) * rs->set.data_devs,
+				     &addr);
+			pi = addr.pi;
+			stripe_zero_pl_part(stripe, pi, off,
+					    rs->set.chunk_size);
+		} else
+			pi = stripe->idx.parity;
+
+		common_xor(stripe, xor_size, off, pi);
+		page_set(PAGE(stripe, pi), DIRTY);
+	}
+}
+
+/* Reconstruct missing chunk. */
+static void reconstruct_xor(struct stripe *stripe)
+{
+	struct raid_set *rs = RS(stripe->sc);
+	int p = stripe->idx.recover;
+
+	BUG_ON(p < 0);
+
+	/* REMOVEME: statistics. */
+	atomic_inc(rs->stats + (raid_set_degraded(rs) ?
+		    S_RECONSTRUCT_EI : S_RECONSTRUCT_DEV));
+
+	/* Zero chunk to be reconstructed. */
+	stripe_zero_chunk(stripe, p);
+	common_xor(stripe, stripe->io.size, 0, p);
+}
+
+/*
+ * Try getting a stripe either from the hash or from the lru list
+ */
+static inline void _stripe_get(struct stripe *stripe)
+{
+	atomic_inc(&stripe->cnt);
+}
+
+static struct stripe *stripe_get(struct raid_set *rs, struct address *addr)
+{
+	struct stripe_cache *sc = &rs->sc;
+	struct stripe *stripe;
+
+	stripe = stripe_lookup(sc, addr->key);
+	if (stripe) {
+		_stripe_get(stripe);
+		/* Remove from the lru list if on. */
+		stripe_lru_del(stripe, LIST_LOCKED);
+		atomic_inc(rs->stats + S_HITS_1ST); /* REMOVEME: statistics. */
+	} else {
+		/* Second try to get an LRU stripe. */
+		stripe = stripe_lru_pop(sc);
+		if (stripe) {
+			_stripe_get(stripe);
+			/* Invalidate before reinserting with changed key. */
+			stripe_invalidate(stripe);
+			stripe->key = addr->key;
+			stripe->region = dm_rh_sector_to_region(rs->recover.rh,
+								addr->key);
+			stripe->idx.parity = addr->pi;
+			sc_insert(sc, stripe);
+			/* REMOVEME: statistics. */
+			atomic_inc(rs->stats + S_INSCACHE);
+		}
+	}
+
+	return stripe;
+}
+
+/*
+ * Decrement reference count on a stripe.
+ *
+ * Move it to list of LRU stripes if zero.
+ */
+static void stripe_put(struct stripe *stripe)
+{
+	if (atomic_dec_and_test(&stripe->cnt)) {
+		if (TestClearStripeActive(stripe))
+			atomic_dec(&stripe->sc->active_stripes);
+
+		/* Put stripe onto the LRU list. */
+		stripe_lru_add(stripe, POS_TAIL, LIST_LOCKED);
+	}
+
+	BUG_ON(atomic_read(&stripe->cnt) < 0);
+}
+
+/*
+ * Process end io
+ *
+ * I need to do it here because I can't in interrupt
+ *
+ * Read and write functions are split in order to avoid
+ * conditionals in the main loop for performamce reasons.
+ */
+
+/* Helper read bios on a page list. */
+static void _bio_copy_page_list(struct stripe *stripe, struct page_list *pl,
+				struct bio *bio)
+{
+	bio_copy_page_list(READ, stripe, pl, bio);
+}
+
+/* Helper write bios on a page list. */
+static void _rh_dec(struct stripe *stripe, struct page_list *pl,
+		    struct bio *bio)
+{
+	dm_rh_dec(RS(stripe->sc)->recover.rh, stripe->region);
+}
+
+/* End io all bios on a page list. */
+static inline int
+page_list_endio(int rw, struct stripe *stripe, unsigned p, unsigned *count)
+{
+	int r = 0;
+	struct bio_list *bl = BL(stripe, p, rw);
+
+	if (!bio_list_empty(bl)) {
+		struct page_list *pl = PL(stripe, p);
+		struct page *page = pl->page;
+
+		if (PageLocked(page))
+			r = -EBUSY;
+		/*
+		 * FIXME: PageUptodate() not cleared
+		 *	  properly for missing chunks ?
+		 */
+		else if (PageUptodate(page)) {
+			struct bio *bio;
+			struct raid_set *rs = RS(stripe->sc);
+			void (*h_f)(struct stripe *, struct page_list *,
+				    struct bio *) =
+				(rw == READ) ? _bio_copy_page_list : _rh_dec;
+
+			while ((bio = bio_list_pop(bl))) {
+				h_f(stripe, pl, bio);
+				_bio_endio(rs, bio, 0);
+				stripe_put(stripe);
+				if (count)
+					(*count)++;
+			}
+		} else
+			r = -EAGAIN;
+	}
+
+	return r;
+}
+
+/*
+ * End io all reads/writes on a stripe copying
+ * read date accross from stripe to bios.
+ */
+static int stripe_endio(int rw, struct stripe *stripe, unsigned *count)
+{
+	int r = 0;
+	unsigned p = RS(stripe->sc)->set.raid_devs;
+
+	while (p--) {
+		int rr = page_list_endio(rw, stripe, p, count);
+
+		if (rr && r != -EIO)
+			r = rr;
+	}
+
+	return r;
+}
+
+/* Fail all ios on a bio list and return # of bios. */
+static unsigned
+bio_list_fail(struct raid_set *rs, struct stripe *stripe, struct bio_list *bl)
+{
+	unsigned r;
+	struct bio *bio;
+
+	raid_set_dead(rs);
+
+	/* Update region counters. */
+	if (stripe) {
+		struct dm_rh_client *rh = rs->recover.rh;
+
+		bio_list_for_each(bio, bl) {
+			if (bio_data_dir(bio) == WRITE)
+				dm_rh_dec(rh, stripe->region);
+		}
+	}
+
+	/* Error end io all bios. */
+	for (r = 0; (bio = bio_list_pop(bl)); r++)
+		_bio_endio(rs, bio, -EIO);
+
+	return r;
+}
+
+/* Fail all ios of a bio list of a stripe and drop io pending count. */
+static void
+stripe_bio_list_fail(struct raid_set *rs, struct stripe *stripe,
+		     struct bio_list *bl)
+{
+	unsigned put = bio_list_fail(rs, stripe, bl);
+
+	while (put--)
+		stripe_put(stripe);
+}
+
+/* Fail all ios hanging off all bio lists of a stripe. */
+static void stripe_fail_io(struct stripe *stripe)
+{
+	struct raid_set *rs = RS(stripe->sc);
+	unsigned p = rs->set.raid_devs;
+
+	stripe_evict(stripe);
+
+	while (p--) {
+		struct stripe_set *ss = stripe->ss + p;
+		int i = ARRAY_SIZE(ss->bl);
+
+		while (i--)
+			stripe_bio_list_fail(rs, stripe, ss->bl + i);
+	}
+}
+
+/*
+ * Handle all stripes by handing them to the daemon, because we can't
+ * map their pages to copy the data in interrupt context.
+ *
+ * We don't want to handle them here either, while interrupts are disabled.
+ */
+
+/* Read/write endio function for dm-io (interrupt context). */
+static void endio(unsigned long error, void *context)
+{
+	struct dm_mem_cache_object *obj = context;
+	struct stripe_set *ss = obj->private;
+	struct stripe *stripe = ss->stripe;
+	struct page *page = obj->pl->page;
+
+	if (unlikely(error))
+		stripe_error(stripe, page);
+	else
+		page_set(page, CLEAN);
+
+	clear_page_locked(page);
+	stripe_io_dec(stripe);
+
+	/* Add stripe to endio list and wake daemon. */
+	stripe_endio_push(stripe);
+}
+
+/*
+ * Recovery io throttling
+ */
+/* Conditionally reset io counters. */
+enum count_type { IO_WORK = 0, IO_RECOVER };
+static int recover_io_reset(struct raid_set *rs)
+{
+	unsigned long j = jiffies;
+
+	/* Pay attention to jiffies overflows. */
+	if (j > rs->recover.last_jiffies + HZ
+	    || j < rs->recover.last_jiffies) {
+		rs->recover.last_jiffies = j;
+		atomic_set(rs->recover.io_count + IO_WORK, 0);
+		atomic_set(rs->recover.io_count + IO_RECOVER, 0);
+		return 1;
+	}
+
+	return 0;
+}
+
+/* Count ios. */
+static INLINE void
+recover_io_count(struct raid_set *rs, struct stripe *stripe)
+{
+	if (RSRecover(rs)) {
+		recover_io_reset(rs);
+		atomic_inc(rs->recover.io_count +
+			   (StripeRecover(stripe) ? IO_RECOVER : IO_WORK));
+	}
+}
+
+/* Read/Write a page_list asynchronously. */
+static void page_list_rw(struct stripe *stripe, unsigned p)
+{
+	struct stripe_cache *sc = stripe->sc;
+	struct raid_set *rs = RS(sc);
+	struct dm_mem_cache_object *obj = stripe->obj + p;
+	struct page_list *pl = obj->pl;
+	struct page *page = pl->page;
+	struct raid_dev *dev = rs->dev + p;
+	struct dm_io_region io = {
+		.bdev = dev->dev->bdev,
+		.sector = stripe->key,
+		.count = stripe->io.size,
+	};
+	struct dm_io_request control = {
+		.bi_rw = PageDirty(page) ? WRITE : READ,
+		.mem.type = DM_IO_PAGE_LIST,
+		.mem.ptr.pl = pl,
+		.mem.offset = 0,
+		.notify.fn = endio,
+		.notify.context = obj,
+		.client = sc->dm_io_client,
+	};
+
+	BUG_ON(PageLocked(page));
+
+	/*
+	 * Don't rw past end of device, which can happen, because
+	 * typically sectors_per_dev isn't divisable by io_size.
+	 */
+	if (unlikely(io.sector + io.count > rs->set.sectors_per_dev))
+		io.count = rs->set.sectors_per_dev - io.sector;
+
+	io.sector += dev->start;	/* Add <offset>. */
+	recover_io_count(rs, stripe);	/* Recovery io accounting. */
+
+	/* REMOVEME: statistics. */
+	atomic_inc(rs->stats +
+		    (PageDirty(page) ? S_DM_IO_WRITE : S_DM_IO_READ));
+
+	ClearPageError(page);
+	set_page_locked(page);
+	io_dev_queued(dev);
+	BUG_ON(dm_io(&control, 1, &io, NULL));
+}
+
+/*
+ * Write dirty / read not uptodate page lists of a stripe.
+ */
+static unsigned stripe_page_lists_rw(struct raid_set *rs, struct stripe *stripe)
+{
+	unsigned r;
+
+	/*
+	 * Increment the pending count on the stripe
+	 * first, so that we don't race in endio().
+	 *
+	 * An inc (IO) is needed for any page:
+	 *
+	 * o not uptodate
+	 * o dirtied by writes merged
+	 * o dirtied by parity calculations
+	 */
+	r = for_each_io_dev(rs, stripe, _stripe_io_inc);
+	if (r) {
+		/* io needed: chunks are not uptodate/dirty. */
+		int max;	/* REMOVEME: */
+		struct stripe_cache *sc = &rs->sc;
+
+		if (!TestSetStripeActive(stripe))
+			atomic_inc(&sc->active_stripes);
+
+		/* Take off the lru list in case it got added there. */
+		stripe_lru_del(stripe, LIST_LOCKED);
+
+		/* Submit actual io. */
+		for_each_io_dev(rs, stripe, page_list_rw);
+
+		/* REMOVEME: statistics */
+		max = sc_active(sc);
+		if (atomic_read(&sc->max_active_stripes) < max)
+			atomic_set(&sc->max_active_stripes, max);
+
+		atomic_inc(rs->stats + S_FLUSHS);
+		/* END REMOVEME: statistics */
+	}
+
+	return r;
+}
+
+/* Work in all pending writes. */
+static INLINE void _writes_merge(struct stripe *stripe, unsigned p)
+{
+	struct bio_list *write = BL(stripe, p, WRITE);
+
+	if (!bio_list_empty(write)) {
+		struct page_list *pl = stripe->obj[p].pl;
+		struct bio *bio;
+		struct bio_list *write_merged = BL(stripe, p, WRITE_MERGED);
+
+		/*
+		 * We can play with the lists without holding a lock,
+		 * because it is just us accessing them anyway.
+		 */
+		bio_list_for_each(bio, write)
+			bio_copy_page_list(WRITE, stripe, pl, bio);
+
+		bio_list_merge(write_merged, write);
+		bio_list_init(write);
+		page_set(pl->page, DIRTY);
+	}
+}
+
+/* Merge in all writes hence dirtying respective pages. */
+static INLINE void writes_merge(struct stripe *stripe)
+{
+	unsigned p = RS(stripe->sc)->set.raid_devs;
+
+	while (p--)
+		_writes_merge(stripe, p);
+}
+
+/* Check, if a chunk gets completely overwritten. */
+static INLINE int stripe_check_overwrite(struct stripe *stripe, unsigned p)
+{
+	unsigned sectors = 0;
+	struct bio *bio;
+	struct bio_list *bl = BL(stripe, p, WRITE);
+
+	bio_list_for_each(bio, bl)
+		sectors += bio_sectors(bio);
+
+	return sectors == RS(stripe->sc)->set.io_size;
+}
+
+/*
+ * Prepare stripe to avoid io on broken/reconstructed
+ * drive in order to reconstruct date on endio.
+ */
+enum prepare_type { IO_ALLOW, IO_PROHIBIT };
+static void stripe_prepare(struct stripe *stripe, unsigned p,
+			   enum prepare_type type)
+{
+	struct page *page = PAGE(stripe, p);
+
+	switch (type) {
+	case IO_PROHIBIT:
+		/*
+		 * In case we prohibit, we gotta make sure, that
+		 * io on all other chunks than the one which failed
+		 * or is being reconstructed is allowed and that it
+		 * doesn't have state uptodate.
+		 */
+		stripe_allow_io(stripe);
+		ClearPageUptodate(page);
+		ProhibitPageIO(page);
+
+		/* REMOVEME: statistics. */
+		atomic_inc(RS(stripe->sc)->stats + S_PROHIBITPAGEIO);
+		stripe->idx.recover = p;
+		SetStripeReconstruct(stripe);
+		break;
+
+	case IO_ALLOW:
+		AllowPageIO(page);
+		stripe->idx.recover = -1;
+		ClearStripeReconstruct(stripe);
+		break;
+
+	default:
+		BUG();
+	}
+}
+
+/*
+ * Degraded/reconstruction mode.
+ *
+ * Check stripe state to figure which chunks don't need IO.
+ */
+static INLINE void stripe_check_reconstruct(struct stripe *stripe,
+					    int prohibited)
+{
+	struct raid_set *rs = RS(stripe->sc);
+
+	/*
+	 * Degraded mode (device(s) failed) ->
+	 * avoid io on the failed device.
+	 */
+	if (unlikely(raid_set_degraded(rs))) {
+		/* REMOVEME: statistics. */
+		atomic_inc(rs->stats + S_DEGRADED);
+		stripe_prepare(stripe, rs->set.ei, IO_PROHIBIT);
+		return;
+	} else {
+		/*
+		 * Reconstruction mode (ie. a particular device or
+		 * some (rotating) parity chunk is being resynchronized) ->
+		 *   o make sure all needed pages are read in
+		 *   o writes are allowed to go through
+		 */
+		int r = region_state(rs, stripe->key, DM_RH_NOSYNC);
+
+		if (r) {
+			/* REMOVEME: statistics. */
+			atomic_inc(rs->stats + S_NOSYNC);
+			stripe_prepare(stripe, dev_for_parity(stripe),
+				       IO_PROHIBIT);
+			return;
+		}
+	}
+
+	/*
+	 * All disks good. Avoid reading parity chunk and reconstruct it
+	 * unless we have prohibited io to chunk(s).
+	 */
+	if (!prohibited) {
+		if (StripeMerged(stripe))
+			stripe_prepare(stripe, stripe->idx.parity, IO_ALLOW);
+		else {
+			stripe_prepare(stripe, stripe->idx.parity, IO_PROHIBIT);
+
+			/*
+			 * Overrule stripe_prepare to reconstruct the
+			 * parity chunk, because it'll be created new anyway.
+			 */
+			ClearStripeReconstruct(stripe);
+		}
+	}
+}
+
+/* Check, if stripe is ready to merge writes. */
+static INLINE int stripe_check_merge(struct stripe *stripe)
+{
+	struct raid_set *rs = RS(stripe->sc);
+	int prohibited = 0;
+	unsigned chunks = 0, p = rs->set.raid_devs;
+
+	/* Walk all chunks. */
+	while (p--) {
+		struct page *page = PAGE(stripe, p);
+
+		/* Can't merge active chunks. */
+		if (PageLocked(page)) {
+			/* REMOVEME: statistics. */
+			atomic_inc(rs->stats + S_MERGE_PAGE_LOCKED);
+			break;
+		}
+
+		/* Can merge uptodate chunks and have to count parity chunk. */
+		if (PageUptodate(page) || p == stripe->idx.parity) {
+			chunks++;
+			continue;
+		}
+
+		/* Read before write ordering. */
+		if (RSCheckOverwrite(rs) &&
+		    bio_list_empty(BL(stripe, p, READ))) {
+			int r = stripe_check_overwrite(stripe, p);
+
+			if (r) {
+				chunks++;
+				/* REMOVEME: statistics. */
+				atomic_inc(RS(stripe->sc)->stats +
+					   S_PROHIBITPAGEIO);
+				ProhibitPageIO(page);
+				prohibited = 1;
+			}
+		}
+	}
+
+	if (chunks == rs->set.raid_devs) {
+		/* All pages are uptodate or get written over or mixture. */
+		/* REMOVEME: statistics. */
+		atomic_inc(rs->stats + S_CAN_MERGE);
+		return 0;
+	} else
+		/* REMOVEME: statistics.*/
+		atomic_inc(rs->stats + S_CANT_MERGE);
+
+	return prohibited ? 1 : -EPERM;
+}
+
+/* Check, if stripe is ready to merge writes. */
+static INLINE int stripe_check_read(struct stripe *stripe)
+{
+	int r = 0;
+	unsigned p = RS(stripe->sc)->set.raid_devs;
+
+	/* Walk all chunks. */
+	while (p--) {
+		struct page *page = PAGE(stripe, p);
+
+		if (!PageLocked(page) &&
+		    bio_list_empty(BL(stripe, p, READ))) {
+			ProhibitPageIO(page);
+			r = 1;
+		}
+	}
+
+	return r;
+}
+
+/*
+ * Read/write a stripe.
+ *
+ * All stripe read/write activity goes through this function.
+ *
+ * States to cover:
+ *   o stripe to read and/or write
+ *   o stripe with error to reconstruct
+ */
+static int stripe_rw(struct stripe *stripe)
+{
+	struct raid_set *rs = RS(stripe->sc);
+	int prohibited = 0, r;
+
+	/*
+	 * Check the state of the RAID set and if degraded (or
+	 * resynchronizing for reads), read in all other chunks but
+	 * the one on the dead/resynchronizing device in order to be
+	 * able to reconstruct the missing one.
+	 *
+	 * Merge all writes hanging off uptodate pages of the stripe.
+	 */
+
+	/* Initially allow io on all chunks and prohibit below, if necessary. */
+	stripe_allow_io(stripe);
+
+	if (StripeRBW(stripe)) {
+		r = stripe_check_merge(stripe);
+		if (!r) {
+			/*
+			 * If I could rely on valid parity (which would only
+			 * be sure in case of a full synchronization),
+			 * I could xor a fraction of chunks out of
+			 * parity and back in.
+			 *
+			 * For the time being, I got to redo parity...
+			 */
+			/* parity_xor(stripe); */	/* Xor chunks out. */
+			stripe_zero_chunk(stripe, stripe->idx.parity);
+			writes_merge(stripe);		/* Merge writes in. */
+			parity_xor(stripe);		/* Update parity. */
+			ClearStripeRBW(stripe);		/* Disable RBW. */
+			SetStripeMerged(stripe);	/* Writes merged. */
+		}
+
+		if (r > 0)
+			prohibited = 1;
+	} else if (!raid_set_degraded(rs))
+		/* Only allow for read avoidance if not degraded. */
+		prohibited = stripe_check_read(stripe);
+
+	/*
+	 * Check, if io needs to be allowed/prohibeted on certain chunks
+	 * because of a degraded set or reconstruction on a region.
+	 */
+	stripe_check_reconstruct(stripe, prohibited);
+
+	/* Now submit any reads/writes. */
+	r = stripe_page_lists_rw(rs, stripe);
+	if (!r) {
+		/*
+		 * No io submitted because of chunk io prohibited or
+		 * locked pages -> push to end io list for processing.
+		 */
+		atomic_inc(rs->stats + S_NO_RW); /* REMOVEME: statistics. */
+		stripe_endio_push(stripe);
+		wake_do_raid(rs);	/* Wake myself. */
+	}
+
+	return 0;
+}
+
+/* Flush stripe either via flush list or imeediately. */
+enum flush_type { FLUSH_DELAY, FLUSH_NOW };
+static int stripe_flush(struct stripe *stripe, enum flush_type type)
+{
+	int r = 0;
+
+	stripe_lru_del(stripe, LIST_LOCKED);
+
+	/* Immediately flush. */
+	if (type == FLUSH_NOW) {
+		if (likely(raid_set_operational(RS(stripe->sc))))
+			r = stripe_rw(stripe); /* Read/write stripe. */
+		else
+			/* Optimization: Fail early on failed sets. */
+			stripe_fail_io(stripe);
+	/* Delay flush by putting it on io list for later processing. */
+	} else if (type == FLUSH_DELAY)
+		stripe_io_add(stripe, POS_TAIL, LIST_UNLOCKED);
+	else
+		BUG();
+
+	return r;
+}
+
+/*
+ * Queue reads and writes to a stripe by hanging
+ * their bios off the stripsets read/write lists.
+ *
+ * Endio reads on uptodate chunks.
+ */
+static INLINE int stripe_queue_bio(struct raid_set *rs, struct bio *bio,
+				   struct bio_list *reject)
+{
+	int r = 0;
+	struct address addr;
+	struct stripe *stripe =
+		stripe_get(rs, raid_address(rs, bio->bi_sector, &addr));
+
+	if (stripe) {
+		int rr, rw = bio_data_dir(bio);
+
+		rr = stripe_lock(rs, stripe, rw, addr.key); /* Lock stripe */
+		if (rr) {
+			stripe_put(stripe);
+			goto out;
+		}
+
+		/* Distinguish read and write cases. */
+		bio_list_add(BL(stripe, addr.di, rw), bio);
+
+		/* REMOVEME: statistics */
+		atomic_inc(rs->stats + (rw == WRITE ?
+			   S_BIOS_ADDED_WRITE : S_BIOS_ADDED_READ));
+
+		if (rw == READ)
+			SetStripeRead(stripe);
+		else {
+			SetStripeRBW(stripe);
+
+			/* Inrement pending write count on region. */
+			dm_rh_inc(rs->recover.rh, stripe->region);
+			r = 1;	/* Region hash needs a flush. */
+		}
+
+		/*
+		 * Optimize stripe flushing:
+		 *
+		 * o directly start io for read stripes.
+		 *
+		 * o put stripe onto stripe caches io_list for RBW,
+		 *   so that do_flush() can belabour it after we put
+		 *   more bios to the stripe for overwrite optimization.
+		 */
+		stripe_flush(stripe,
+			     StripeRead(stripe) ? FLUSH_NOW : FLUSH_DELAY);
+
+	/* Got no stripe from cache -> reject bio. */
+	} else {
+out:
+		bio_list_add(reject, bio);
+		/* REMOVEME: statistics. */
+		atomic_inc(rs->stats + S_IOS_POST);
+	}
+
+	return r;
+}
+
+/*
+ * Recovery functions
+ */
+/* Read a stripe off a raid set for recovery. */
+static int recover_read(struct raid_set *rs, struct stripe *stripe, int idx)
+{
+	/* Invalidate all pages so that they get read in. */
+	stripe_pages_invalidate(stripe);
+
+	/* Allow io on all recovery chunks. */
+	stripe_allow_io(stripe);
+
+	if (idx > -1)
+		ProhibitPageIO(PAGE(stripe, idx));
+
+	stripe->key = rs->recover.pos;
+	return stripe_page_lists_rw(rs, stripe);
+}
+
+/* Write a stripe to a raid set for recovery. */
+static int recover_write(struct raid_set *rs, struct stripe *stripe, int idx)
+{
+	/*
+	 * If this is a reconstruct of a particular device, then
+	 * reconstruct the respective page(s), else create parity page(s).
+	 */
+	if (idx > -1) {
+		struct page *page = PAGE(stripe, idx);
+
+		AllowPageIO(page);
+		stripe_zero_chunk(stripe, idx);
+		common_xor(stripe, stripe->io.size, 0, idx);
+		page_set(page, DIRTY);
+	} else
+		parity_xor(stripe);
+
+	return stripe_page_lists_rw(rs, stripe);
+}
+
+/* Recover bandwidth available ?. */
+static int recover_bandwidth(struct raid_set *rs)
+{
+	int r, work;
+
+	/* On reset -> allow recovery. */
+	r = recover_io_reset(rs);
+	if (r || RSBandwidth(rs))
+		goto out;
+
+	work = atomic_read(rs->recover.io_count + IO_WORK);
+	if (work) {
+		/* Pay attention to larger recover stripe size. */
+		int recover =
+		    atomic_read(rs->recover.io_count + IO_RECOVER) *
+				rs->recover.io_size /
+				rs->set.io_size;
+
+		/*
+		 * Don't use more than given bandwidth of
+		 * the work io for recovery.
+		 */
+		if (recover > work / rs->recover.bandwidth_work) {
+			/* REMOVEME: statistics. */
+			atomic_inc(rs->stats + S_NO_BANDWIDTH);
+			return 0;
+		}
+	}
+
+out:
+	atomic_inc(rs->stats + S_BANDWIDTH);	/* REMOVEME: statistics. */
+	return 1;
+}
+
+/* Try to get a region to recover. */
+static int recover_get_region(struct raid_set *rs)
+{
+	struct recover *rec = &rs->recover;
+	struct dm_rh_client *rh = rec->rh;
+
+	/* Start quiescing some regions. */
+	if (!RSRegionGet(rs)) {
+		int r = recover_bandwidth(rs); /* Enough bandwidth ?. */
+
+		if (r) {
+			r = dm_rh_recovery_prepare(rh);
+			if (r < 0) {
+				DMINFO("No %sregions to recover",
+				       rec->nr_regions_to_recover ?
+				       "more " : "");
+				return -ENOENT;
+			}
+		} else
+			return -EAGAIN;
+
+		SetRSRegionGet(rs);
+	}
+
+	if (!rec->reg) {
+		rec->reg = dm_rh_recovery_start(rh);
+		if (rec->reg) {
+			/*
+			 * A reference for the the region I'll
+			 * keep till I've completely synced it.
+			 */
+			io_get(rs);
+			rec->pos = dm_rh_region_to_sector(rh,
+				dm_rh_get_region_key(rec->reg));
+			rec->end = rec->pos + dm_rh_get_region_size(rh);
+			return 1;
+		} else
+			return -EAGAIN;
+	}
+
+	return 0;
+}
+
+/* Read/write a recovery stripe. */
+static INLINE int recover_stripe_rw(struct raid_set *rs, struct stripe *stripe)
+{
+	/* Read/write flip-flop. */
+	if (TestClearStripeRBW(stripe)) {
+		SetStripeRead(stripe);
+		return recover_read(rs, stripe, idx_get(rs));
+	} else if (TestClearStripeRead(stripe))
+		return recover_write(rs, stripe, idx_get(rs));
+
+	return 0;
+}
+
+/* Reset recovery variables. */
+static void recovery_region_reset(struct raid_set *rs)
+{
+	rs->recover.reg = NULL;
+	ClearRSRegionGet(rs);
+}
+
+/* Update region hash state. */
+static void recover_rh_update(struct raid_set *rs, int error)
+{
+	struct recover *rec = &rs->recover;
+	struct dm_rh_client *rh = rec->rh;
+	struct dm_region *reg = rec->reg;
+
+	if (reg) {
+		dm_rh_recovery_end(rh, reg, error);
+		if (!error)
+			rec->nr_regions_recovered++;
+
+		recovery_region_reset(rs);
+	}
+
+	dm_rh_update_states(rh, 1);
+	dm_rh_flush(rh);
+	io_put(rs);	/* Release the io reference for the region. */
+}
+
+/* Called by main io daemon to recover regions. */
+/* FIXME: cope with MAX_RECOVER > 1. */
+static INLINE void _do_recovery(struct raid_set *rs, struct stripe *stripe)
+{
+	int r;
+	struct recover *rec = &rs->recover;
+
+	/* If recovery is active -> return. */
+	if (StripeActive(stripe))
+		return;
+
+	/* io error is fatal for recovery -> stop it. */
+	if (unlikely(StripeError(stripe)))
+		goto err;
+
+	/* Get a region to recover. */
+	r = recover_get_region(rs);
+	switch (r) {
+	case 1:	/* Got a new region. */
+		/* Flag read before write. */
+		ClearStripeRead(stripe);
+		SetStripeRBW(stripe);
+		break;
+
+	case 0:
+		/* Got a region in the works. */
+		r = recover_bandwidth(rs);
+		if (r) /* Got enough bandwidth. */
+			break;
+
+	case -EAGAIN:
+		/* No bandwidth/quiesced region yet, try later. */
+		wake_do_raid_delayed(rs, HZ / 10);
+		return;
+
+	case -ENOENT:	/* No more regions. */
+		dm_table_event(rs->ti->table);
+		goto free;
+	}
+
+	/* Read/write a recover stripe. */
+	r = recover_stripe_rw(rs, stripe);
+	if (r) {
+		/* IO initiated, get another reference for the IO. */
+		io_get(rs);
+		return;
+	}
+
+	/* Update recovery position within region. */
+	rec->pos += stripe->io.size;
+
+	/* If we're at end of region, update region hash. */
+	if (rec->pos >= rec->end ||
+	    rec->pos >= rs->set.sectors_per_dev)
+		recover_rh_update(rs, 0);
+	else
+		SetStripeRBW(stripe);
+
+	/* Schedule myself for another round... */
+	wake_do_raid(rs);
+	return;
+
+err:
+	raid_set_check_degrade(rs, stripe);
+
+	{
+		char buf[BDEVNAME_SIZE];
+
+		DMERR("stopping recovery due to "
+		      "ERROR on /dev/%s, stripe at offset %llu",
+		      bdevname(rs->dev[rs->set.ei].dev->bdev, buf),
+		      (unsigned long long) stripe->key);
+
+	}
+
+	/* Make sure, that all quiesced regions get released. */
+	do {
+		if (rec->reg)
+			dm_rh_recovery_end(rec->rh, rec->reg, -EIO);
+
+		rec->reg = dm_rh_recovery_start(rec->rh);
+	} while (rec->reg);
+
+	recover_rh_update(rs, -EIO);
+free:
+	rs->set.dev_to_init = -1;
+
+	/* Check for jiffies overrun. */
+	rs->recover.end_jiffies = jiffies;
+	if (rs->recover.end_jiffies < rs->recover.start_jiffies)
+		rs->recover.end_jiffies = ~0;
+
+	ClearRSRecover(rs);
+}
+
+static INLINE void do_recovery(struct raid_set *rs)
+{
+	struct stripe *stripe;
+
+	list_for_each_entry(stripe, &rs->recover.stripes, lists[LIST_RECOVER])
+		_do_recovery(rs, stripe);
+
+	if (!RSRecover(rs))
+		stripe_recover_free(rs);
+}
+
+/*
+ * END recovery functions
+ */
+
+/* End io process all stripes handed in by endio() callback. */
+static void do_endios(struct raid_set *rs)
+{
+	struct stripe_cache *sc = &rs->sc;
+	struct stripe *stripe;
+
+	while ((stripe = stripe_endio_pop(sc))) {
+		unsigned count;
+
+		/* Recovery stripe special case. */
+		if (unlikely(StripeRecover(stripe))) {
+			if (stripe_io(stripe))
+				continue;
+
+			io_put(rs); /* Release region io reference. */
+			ClearStripeActive(stripe);
+
+			/* REMOVEME: statistics*/
+			atomic_dec(&sc->active_stripes);
+			continue;
+		}
+
+		/* Early end io all reads on any uptodate chunks. */
+		stripe_endio(READ, stripe, (count = 0, &count));
+		if (stripe_io(stripe)) {
+			if (count) /* REMOVEME: statistics. */
+				atomic_inc(rs->stats + S_ACTIVE_READS);
+
+			continue;
+		}
+
+		/* Set stripe inactive after all io got processed. */
+		if (TestClearStripeActive(stripe))
+			atomic_dec(&sc->active_stripes);
+
+		/* Unlock stripe (for clustering). */
+		stripe_unlock(rs, stripe);
+
+		/*
+		 * If an io error on a stripe occured and the RAID set
+		 * is still operational, requeue the stripe for io.
+		 */
+		if (TestClearStripeError(stripe)) {
+			raid_set_check_degrade(rs, stripe);
+			ClearStripeReconstruct(stripe);
+
+			if (!StripeMerged(stripe) &&
+			    raid_set_operational(rs)) {
+				stripe_pages_invalidate(stripe);
+				stripe_flush(stripe, FLUSH_DELAY);
+				/* REMOVEME: statistics. */
+				atomic_inc(rs->stats + S_REQUEUE);
+				continue;
+			}
+		}
+
+		/* Check if the RAID set is inoperational to error ios. */
+		if (!raid_set_operational(rs)) {
+			ClearStripeReconstruct(stripe);
+			stripe_fail_io(stripe);
+			BUG_ON(atomic_read(&stripe->cnt));
+			continue;
+		}
+
+		/* Got to reconstruct a missing chunk. */
+		if (TestClearStripeReconstruct(stripe))
+			reconstruct_xor(stripe);
+
+		/*
+		 * Now that we've got a complete stripe, we can
+		 * process the rest of the end ios on reads.
+		 */
+		BUG_ON(stripe_endio(READ, stripe, NULL));
+		ClearStripeRead(stripe);
+
+		/*
+		 * Read-before-write stripes need to be flushed again in
+		 * order to work the write data into the pages *after*
+		 * they were read in.
+		 */
+		if (TestClearStripeMerged(stripe))
+			/* End io all bios which got merged already. */
+			BUG_ON(stripe_endio(WRITE_MERGED, stripe, NULL));
+
+		/* Got to put on flush list because of new writes. */
+		if (StripeRBW(stripe))
+			stripe_flush(stripe, FLUSH_DELAY);
+	}
+}
+
+/*
+ * Stripe cache shrinking.
+ */
+static INLINE void do_sc_shrink(struct raid_set *rs)
+{
+	unsigned shrink = atomic_read(&rs->sc.stripes_to_shrink);
+
+	if (shrink) {
+		unsigned cur = atomic_read(&rs->sc.stripes);
+
+		sc_shrink(&rs->sc, shrink);
+		shrink -= cur - atomic_read(&rs->sc.stripes);
+		atomic_set(&rs->sc.stripes_to_shrink, shrink);
+
+		/*
+		 * Wake myself up in case we failed to shrink the
+		 * requested amount in order to try again later.
+		 */
+		if (shrink)
+			wake_do_raid(rs);
+	}
+}
+
+
+/*
+ * Process all ios
+ *
+ * We do different things with the io depending on the
+ * state of the region that it's in:
+ *
+ * o reads: hang off stripe cache or postpone if full
+ *
+ * o writes:
+ *
+ *  CLEAN/DIRTY/NOSYNC:	increment pending and hang io off stripe's stripe set.
+ *			In case stripe cache is full or busy, postpone the io.
+ *
+ *  RECOVERING:		delay the io until recovery of the region completes.
+ *
+ */
+static INLINE void do_ios(struct raid_set *rs, struct bio_list *ios)
+{
+	int r;
+	unsigned flush = 0;
+	struct dm_rh_client *rh = rs->recover.rh;
+	struct bio *bio;
+	struct bio_list delay, reject;
+
+	bio_list_init(&delay);
+	bio_list_init(&reject);
+
+	/*
+	 * Classify each io:
+	 *    o delay to recovering regions
+	 *    o queue to all other regions
+	 */
+	while ((bio = bio_list_pop(ios))) {
+		/*
+		 * In case we get a barrier bio, push it back onto
+		 * the input queue unless all work queues are empty
+		 * and the stripe cache is inactive.
+		 */
+		if (unlikely(bio_barrier(bio))) {
+			/* REMOVEME: statistics. */
+			atomic_inc(rs->stats + S_BARRIER);
+			if (!list_empty(rs->sc.lists + LIST_IO) ||
+			    !bio_list_empty(&delay) ||
+			    !bio_list_empty(&reject) ||
+			    sc_active(&rs->sc)) {
+				bio_list_push(ios, bio);
+				break;
+			}
+		}
+
+		r = region_state(rs, _sector(rs, bio), DM_RH_RECOVERING);
+		if (unlikely(r)) {
+			/* Got to wait for recovering regions. */
+			bio_list_add(&delay, bio);
+			SetRSBandwidth(rs);
+		} else {
+			/*
+			 * Process ios to non-recovering regions by queueing
+			 * them to stripes (does rh_inc()) for writes).
+			 */
+			flush += stripe_queue_bio(rs, bio, &reject);
+		}
+	}
+
+	if (flush) {
+		r = dm_rh_flush(rh); /* Writes got queued -> flush dirty log. */
+		if (r)
+			DMERR("dirty log flush");
+	}
+
+	/* Delay ios to regions which are recovering. */
+	while ((bio = bio_list_pop(&delay))) {
+		/* REMOVEME: statistics.*/
+		atomic_inc(rs->stats + S_DELAYED_BIOS);
+		atomic_inc(rs->stats + S_SUM_DELAYED_BIOS);
+		dm_rh_delay_by_region(rh, bio,
+			dm_rh_sector_to_region(rh, _sector(rs, bio)));
+
+	}
+
+	/* Merge any rejected bios back to the head of the input list. */
+	bio_list_merge_head(ios, &reject);
+}
+
+/* Flush any stripes on the io list. */
+static INLINE void do_flush(struct raid_set *rs)
+{
+	struct list_head *list = rs->sc.lists + LIST_IO, *pos, *tmp;
+
+	list_for_each_safe(pos, tmp, list) {
+		int r = stripe_flush(list_entry(pos, struct stripe,
+						lists[LIST_IO]), FLUSH_NOW);
+
+		/* Remove from the list only if the stripe got processed. */
+		if (!r)
+			list_del_init(pos);
+	}
+}
+
+/* Send an event in case we're getting too busy. */
+static INLINE void do_busy_event(struct raid_set *rs)
+{
+	if ((sc_active(&rs->sc) > atomic_read(&rs->sc.stripes) * 4 / 5)) {
+		if (!TestSetRSScBusy(rs))
+			dm_table_event(rs->ti->table);
+	} else
+		ClearRSScBusy(rs);
+}
+
+/* Unplug: let the io role on the sets devices. */
+static INLINE void do_unplug(struct raid_set *rs)
+{
+	struct raid_dev *dev = rs->dev + rs->set.raid_devs;
+
+	while (dev-- > rs->dev) {
+		/* Only call any device unplug function, if io got queued. */
+		if (io_dev_clear(dev))
+			blk_unplug(bdev_get_queue(dev->dev->bdev));
+	}
+}
+
+/*-----------------------------------------------------------------
+ * RAID daemon
+ *---------------------------------------------------------------*/
+/*
+ * o belabour all end ios
+ * o optionally shrink the stripe cache
+ * o update the region hash states
+ * o optionally do recovery
+ * o grab the input queue
+ * o work an all requeued or new ios and perform stripe cache flushs
+ *   unless the RAID set is inoperational (when we error ios)
+ * o check, if the stripe cache gets too busy and throw an event if so
+ * o unplug any component raid devices with queued bios
+ */
+static void do_raid(struct work_struct *ws)
+{
+	struct raid_set *rs = container_of(ws, struct raid_set, io.dws.work);
+	struct bio_list *ios = &rs->io.work, *ios_in = &rs->io.in;
+	spinlock_t *lock = &rs->io.in_lock;
+
+	/*
+	 * We always need to end io, so that ios
+	 * can get errored in case the set failed
+	 * and the region counters get decremented
+	 * before we update the region hash states.
+	 */
+redo:
+	do_endios(rs);
+
+	/*
+	 * Now that we've end io'd, which may have put stripes on
+	 * the LRU list, we shrink the stripe cache if requested.
+	 */
+	do_sc_shrink(rs);
+
+	/* Update region hash states before we go any further. */
+	dm_rh_update_states(rs->recover.rh, 1);
+
+	/* Try to recover regions. */
+	if (RSRecover(rs))
+		do_recovery(rs);
+
+	/* More endios -> process. */
+	if (!stripe_endio_empty(&rs->sc)) {
+		atomic_inc(rs->stats + S_REDO);
+		goto redo;
+	}
+
+	/* Quickly grab all new ios queued and add them to the work list. */
+	spin_lock_irq(lock);
+	bio_list_merge(ios, ios_in);
+	bio_list_init(ios_in);
+	spin_unlock_irq(lock);
+
+	/* Let's assume we're operational most of the time ;-). */
+	if (likely(raid_set_operational(rs))) {
+		/* If we got ios, work them into the cache. */
+		if (!bio_list_empty(ios)) {
+			do_ios(rs, ios);
+			do_unplug(rs);	/* Unplug the sets device queues. */
+		}
+
+		do_flush(rs);		/* Flush any stripes on io list. */
+		do_unplug(rs);		/* Unplug the sets device queues. */
+		do_busy_event(rs);	/* Check if we got too busy. */
+
+		/* More endios -> process. */
+		if (!stripe_endio_empty(&rs->sc)) {
+			atomic_inc(rs->stats + S_REDO);
+			goto redo;
+		}
+	} else
+		/* No way to reconstruct data with too many devices failed. */
+		bio_list_fail(rs, NULL, ios);
+}
+
+/*
+ * Callback for region hash to dispatch
+ * delayed bios queued to recovered regions
+ * (Gets called via rh_update_states()).
+ */
+static void dispatch_delayed_bios(void *context, struct bio_list *bl, int dummy)
+{
+	struct raid_set *rs = context;
+	struct bio *bio;
+
+	/* REMOVEME: decrement pending delayed bios counter. */
+	bio_list_for_each(bio, bl)
+		atomic_dec(rs->stats + S_DELAYED_BIOS);
+
+	/* Merge region hash private list to work list. */
+	bio_list_merge_head(&rs->io.work, bl);
+	bio_list_init(bl);
+	ClearRSBandwidth(rs);
+}
+
+/*************************************************************
+ * Constructor helpers
+ *************************************************************/
+/* Calculate MB/sec. */
+static INLINE unsigned mbpers(struct raid_set *rs, unsigned speed)
+{
+	return to_bytes(speed * rs->set.data_devs *
+			rs->recover.io_size * HZ >> 10) >> 10;
+}
+
+/*
+ * Discover fastest xor algorithm and # of chunks combination.
+ */
+/* Calculate speed for algorithm and # of chunks. */
+static INLINE unsigned xor_speed(struct stripe *stripe)
+{
+	unsigned r = 0;
+	unsigned long j;
+
+	/* Wait for next tick. */
+	for (j = jiffies; j == jiffies;)
+		;
+
+	/* Do xors for a full tick. */
+	for (j = jiffies; j == jiffies;) {
+		mb();
+		common_xor(stripe, stripe->io.size, 0, 0);
+		mb();
+		r++;
+		mb();
+	}
+
+	return r;
+}
+
+/* Optimize xor algorithm for this RAID set. */
+static unsigned xor_optimize(struct raid_set *rs)
+{
+	unsigned chunks_max = 2, speed_max = 0;
+	struct xor_func *f = ARRAY_END(xor_funcs), *f_max = NULL;
+	struct stripe *stripe;
+
+	BUG_ON(list_empty(&rs->recover.stripes));
+	stripe = list_first_entry(&rs->recover.stripes, struct stripe,
+			    lists[LIST_RECOVER]);
+
+	/*
+	 * Got to allow io on all chunks, so that
+	 * xor() will actually work on them.
+	 */
+	stripe_allow_io(stripe);
+
+	/* Try all xor functions. */
+	while (f-- > xor_funcs) {
+		unsigned speed;
+
+		/* Set actual xor function for common_xor(). */
+		rs->xor.f = f;
+		rs->xor.chunks = XOR_CHUNKS_MAX + 1;
+
+		while (rs->xor.chunks-- > 2) {
+			speed = xor_speed(stripe);
+			if (speed > speed_max) {
+				speed_max = speed;
+				chunks_max = rs->xor.chunks;
+				f_max = f;
+			}
+		}
+	}
+
+	/* Memorize optimum parameters. */
+	rs->xor.f = f_max;
+	rs->xor.chunks = chunks_max;
+	return speed_max;
+}
+
+/*
+ * Allocate a RAID context (a RAID set)
+ */
+static int
+context_alloc(struct raid_set **raid_set, struct raid_type *raid_type,
+	      unsigned stripes, unsigned chunk_size, unsigned io_size,
+	      unsigned recover_io_size, unsigned raid_devs,
+	      sector_t sectors_per_dev,
+	      struct dm_target *ti, unsigned dl_parms, char **argv)
+{
+	int r;
+	unsigned p;
+	size_t len;
+	sector_t region_size, ti_len;
+	struct raid_set *rs = NULL;
+	struct dm_dirty_log *dl;
+	struct recover *rec;
+
+	/*
+	 * Create the dirty log
+	 *
+	 * We need to change length for the dirty log constructor,
+	 * because we want an amount of regions for all stripes derived
+	 * from the single device size, so that we can keep region
+	 * size = 2^^n independant of the number of devices
+	 */
+	ti_len = ti->len;
+	ti->len = sectors_per_dev;
+	dl = dm_dirty_log_create(argv[0], ti, dl_parms, argv + 2);
+	ti->len = ti_len;
+	if (!dl)
+		goto bad_dirty_log;
+
+	/* Chunk size *must* be smaller than region size. */
+	region_size = dl->type->get_region_size(dl);
+	if (chunk_size > region_size)
+		goto bad_chunk_size;
+
+	/* Recover io size *must* be smaller than region size as well. */
+	if (recover_io_size > region_size)
+		goto bad_recover_io_size;
+
+	/* Size and allocate the RAID set structure. */
+	len = sizeof(*rs->data) + sizeof(*rs->dev);
+	if (array_too_big(sizeof(*rs), len, raid_devs))
+		goto bad_array;
+
+	len = sizeof(*rs) + raid_devs * len;
+	rs = kzalloc(len, GFP_KERNEL);
+	if (!rs)
+		goto bad_alloc;
+
+	rec = &rs->recover;
+	atomic_set(&rs->io.in_process, 0);
+	atomic_set(&rs->io.in_process_max, 0);
+	rec->io_size = recover_io_size;
+
+	/* Pointer to data array. */
+	rs->data = (unsigned long **)
+		   ((void *) rs->dev + raid_devs * sizeof(*rs->dev));
+	rec->dl = dl;
+	rs->set.raid_devs = p = raid_devs;
+	rs->set.data_devs = raid_devs - raid_type->parity_devs;
+	rs->set.raid_type = raid_type;
+
+	/*
+	 * Set chunk and io size and respective shifts
+	 * (used to avoid divisions)
+	 */
+	rs->set.chunk_size = chunk_size;
+	rs->set.chunk_mask = chunk_size - 1;
+	rs->set.chunk_shift = ffs(chunk_size) - 1;
+
+	rs->set.io_size = io_size;
+	rs->set.io_mask = io_size - 1;
+	rs->set.io_shift = ffs(io_size) - 1;
+	rs->set.io_shift_mask = rs->set.chunk_mask & ~rs->set.io_mask;
+
+	rs->set.pages_per_io = chunk_pages(io_size);
+	rs->set.sectors_per_dev = sectors_per_dev;
+
+	rs->set.ei = -1;	/* Indicate no failed device. */
+	atomic_set(&rs->set.failed_devs, 0);
+
+	rs->ti = ti;
+
+	atomic_set(rec->io_count + IO_WORK, 0);
+	atomic_set(rec->io_count + IO_RECOVER, 0);
+
+	/* Initialize io lock and queues. */
+	spin_lock_init(&rs->io.in_lock);
+	bio_list_init(&rs->io.in);
+	bio_list_init(&rs->io.work);
+
+	init_waitqueue_head(&rs->io.suspendq);	/* Suspend waiters (dm-io). */
+
+	rec->nr_regions = dm_sector_div_up(sectors_per_dev, region_size);
+	rec->rh = dm_rh_client_create(MAX_RECOVER, dispatch_delayed_bios, rs,
+				      wake_do_raid, rs, dl, region_size,
+				      rs->recover.nr_regions);
+	if (IS_ERR(rec->rh))
+		goto bad_rh;
+
+	/* Initialize stripe cache. */
+	r = sc_init(rs, stripes);
+	if (r)
+		goto bad_sc;
+
+	/* Create dm-io client context. */
+	rs->sc.dm_io_client = dm_io_client_create(rs->set.raid_devs *
+						  rs->set.pages_per_io);
+	if (IS_ERR(rs->sc.dm_io_client))
+		goto bad_dm_io_client;
+
+	/* REMOVEME: statistics. */
+	stats_reset(rs);
+	ClearRSDevelStats(rs);	/* Disnable development status. */
+
+	*raid_set = rs;
+	return 0;
+
+bad_dirty_log:
+	TI_ERR_RET("Error creating dirty log", -ENOMEM);
+
+
+bad_chunk_size:
+	dm_dirty_log_destroy(dl);
+	TI_ERR("Chunk size larger than region size");
+
+bad_recover_io_size:
+	dm_dirty_log_destroy(dl);
+	TI_ERR("Recover stripe io size larger than region size");
+
+bad_array:
+	dm_dirty_log_destroy(dl);
+	TI_ERR("Arry too big");
+
+bad_alloc:
+	dm_dirty_log_destroy(dl);
+	TI_ERR_RET("Cannot allocate raid context", -ENOMEM);
+
+bad_rh:
+	dm_dirty_log_destroy(dl);
+	ti->error = DM_MSG_PREFIX "Error creating dirty region hash";
+	goto free_rs;
+
+bad_sc:
+	ti->error = DM_MSG_PREFIX "Error creating stripe cache";
+	goto free;
+
+bad_dm_io_client:
+	ti->error = DM_MSG_PREFIX "Error allocating dm-io resources";
+free:
+	dm_rh_client_destroy(rec->rh);
+	sc_exit(&rs->sc);
+	dm_rh_client_destroy(rec->rh); /* Destroys dirty log as well. */
+free_rs:
+	kfree(rs);
+	return -ENOMEM;
+}
+
+/* Free a RAID context (a RAID set). */
+static void
+context_free(struct raid_set *rs, struct dm_target *ti, unsigned r)
+{
+	while (r--)
+		dm_put_device(ti, rs->dev[r].dev);
+
+	dm_io_client_destroy(rs->sc.dm_io_client);
+	sc_exit(&rs->sc);
+	dm_rh_client_destroy(rs->recover.rh);
+	dm_dirty_log_destroy(rs->recover.dl);
+	kfree(rs);
+}
+
+/* Create work queue and initialize work. */
+static int rs_workqueue_init(struct raid_set *rs)
+{
+	struct dm_target *ti = rs->ti;
+
+	rs->io.wq = create_singlethread_workqueue(DAEMON);
+	if (!rs->io.wq)
+		TI_ERR_RET("failed to create " DAEMON, -ENOMEM);
+
+	INIT_DELAYED_WORK(&rs->io.dws, do_raid);
+	return 0;
+}
+
+/* Return pointer to raid_type structure for raid name. */
+static struct raid_type *get_raid_type(char *name)
+{
+	struct raid_type *r = ARRAY_END(raid_types);
+
+	while (r-- > raid_types) {
+		if (!strnicmp(STR_LEN(r->name, name)))
+			return r;
+	}
+
+	return NULL;
+}
+
+/* FIXME: factor out to dm core. */
+static int multiple(sector_t a, sector_t b, sector_t *n)
+{
+	sector_t r = a;
+
+	sector_div(r, b);
+	*n = r;
+	return a == r * b;
+}
+
+/* Log RAID set information to kernel log. */
+static void raid_set_log(struct raid_set *rs, unsigned speed)
+{
+	unsigned p;
+	char buf[BDEVNAME_SIZE];
+
+	for (p = 0; p < rs->set.raid_devs; p++)
+		DMINFO("/dev/%s is raid disk %u",
+		       bdevname(rs->dev[p].dev->bdev, buf), p);
+
+	DMINFO("%d/%d/%d sectors chunk/io/recovery size, %u stripes",
+	       rs->set.chunk_size, rs->set.io_size, rs->recover.io_size,
+	       atomic_read(&rs->sc.stripes));
+	DMINFO("algorithm \"%s\", %u chunks with %uMB/s", rs->xor.f->name,
+	       rs->xor.chunks, mbpers(rs, speed));
+	DMINFO("%s set with net %u/%u devices", rs->set.raid_type->descr,
+	       rs->set.data_devs, rs->set.raid_devs);
+}
+
+/* Get all devices and offsets. */
+static int
+dev_parms(struct dm_target *ti, struct raid_set *rs,
+	  char **argv, int *p)
+{
+	for (*p = 0; *p < rs->set.raid_devs; (*p)++, argv += 2) {
+		int r;
+		unsigned long long tmp;
+		struct raid_dev *dev = rs->dev + *p;
+		union dev_lookup dl = {.dev = dev };
+
+		/* Get offset and device. */
+		r = sscanf(argv[1], "%llu", &tmp);
+		if (r != 1)
+			TI_ERR("Invalid RAID device offset parameter");
+
+		dev->start = tmp;
+		r = dm_get_device(ti, argv[0], dev->start,
+				  rs->set.sectors_per_dev,
+				  dm_table_get_mode(ti->table), &dev->dev);
+		if (r)
+			TI_ERR_RET("RAID device lookup failure", r);
+
+		r = raid_dev_lookup(rs, bynumber, &dl);
+		if (r != -ENODEV && r < *p) {
+			(*p)++;	/* Ensure dm_put_device() on actual device. */
+			TI_ERR_RET("Duplicate RAID device", -ENXIO);
+		}
+	}
+
+	return 0;
+}
+
+/* Set recovery bandwidth. */
+static INLINE void
+recover_set_bandwidth(struct raid_set *rs, unsigned bandwidth)
+{
+	rs->recover.bandwidth = bandwidth;
+	rs->recover.bandwidth_work = 100 / bandwidth;
+}
+
+/* Handle variable number of RAID parameters. */
+static int
+raid_variable_parms(struct dm_target *ti, char **argv,
+		    unsigned i, int *raid_parms,
+		    int *chunk_size, int *chunk_size_parm,
+		    int *stripes, int *stripes_parm,
+		    int *io_size, int *io_size_parm,
+		    int *recover_io_size, int *recover_io_size_parm,
+		    int *bandwidth, int *bandwidth_parm)
+{
+	/* Fetch # of variable raid parameters. */
+	if (sscanf(argv[i++], "%d", raid_parms) != 1 ||
+	    !range_ok(*raid_parms, 0, 5))
+		TI_ERR("Bad variable raid parameters number");
+
+	if (*raid_parms) {
+		/*
+		 * If we've got variable RAID parameters,
+		 * chunk size is the first one
+		 */
+		if (sscanf(argv[i++], "%d", chunk_size) != 1 ||
+		    (*chunk_size != -1 &&
+		     (!POWER_OF_2(*chunk_size) ||
+		      !range_ok(*chunk_size, IO_SIZE_MIN, CHUNK_SIZE_MAX))))
+			TI_ERR("Invalid chunk size; must be 2^^n and <= 16384");
+
+		*chunk_size_parm = *chunk_size;
+		if (*chunk_size == -1)
+			*chunk_size = CHUNK_SIZE;
+
+		/*
+		 * In case we've got 2 or more variable raid
+		 * parameters, the number of stripes is the second one
+		 */
+		if (*raid_parms > 1) {
+			if (sscanf(argv[i++], "%d", stripes) != 1 ||
+			    (*stripes != -1 &&
+			     !range_ok(*stripes, STRIPES_MIN,
+				       STRIPES_MAX)))
+				TI_ERR("Invalid number of stripes: must "
+				       "be >= 8 and <= 8192");
+		}
+
+		*stripes_parm = *stripes;
+		if (*stripes == -1)
+			*stripes = STRIPES;
+
+		/*
+		 * In case we've got 3 or more variable raid
+		 * parameters, the io size is the third one.
+		 */
+		if (*raid_parms > 2) {
+			if (sscanf(argv[i++], "%d", io_size) != 1 ||
+			    (*io_size != -1 &&
+			     (!POWER_OF_2(*io_size) ||
+			      !range_ok(*io_size, IO_SIZE_MIN,
+					min(BIO_MAX_SECTORS / 2,
+					*chunk_size)))))
+				TI_ERR("Invalid io size; must "
+				       "be 2^^n and less equal "
+				       "min(BIO_MAX_SECTORS/2, chunk size)");
+		} else
+			*io_size = *chunk_size;
+
+		*io_size_parm = *io_size;
+		if (*io_size == -1)
+			*io_size = *chunk_size;
+
+		/*
+		 * In case we've got 4 variable raid parameters,
+		 * the recovery stripe io_size is the fourth one
+		 */
+		if (*raid_parms > 3) {
+			if (sscanf(argv[i++], "%d", recover_io_size) != 1 ||
+			    (*recover_io_size != -1 &&
+			     (!POWER_OF_2(*recover_io_size) ||
+			     !range_ok(*recover_io_size, RECOVER_IO_SIZE_MIN,
+				       BIO_MAX_SECTORS / 2))))
+				TI_ERR("Invalid recovery io size; must be "
+				       "2^^n and less equal BIO_MAX_SECTORS/2");
+		}
+
+		*recover_io_size_parm = *recover_io_size;
+		if (*recover_io_size == -1)
+			*recover_io_size = RECOVER_IO_SIZE;
+
+		/*
+		 * In case we've got 5 variable raid parameters,
+		 * the recovery io bandwidth is the fifth one
+		 */
+		if (*raid_parms > 4) {
+			if (sscanf(argv[i++], "%d", bandwidth) != 1 ||
+			    (*bandwidth != -1 &&
+			     !range_ok(*bandwidth, BANDWIDTH_MIN,
+				       BANDWIDTH_MAX)))
+				TI_ERR("Invalid recovery bandwidth "
+				       "percentage; must be > 0 and <= 100");
+		}
+
+		*bandwidth_parm = *bandwidth;
+		if (*bandwidth == -1)
+			*bandwidth = BANDWIDTH;
+	}
+
+	return 0;
+}
+
+/* Parse optional locking parameters. */
+static int
+raid_locking_parms(struct dm_target *ti, char **argv,
+		   unsigned i, int *locking_parms,
+		   struct dm_raid45_locking_type **locking_type)
+{
+	*locking_parms = 0;
+	*locking_type = &locking_none;
+
+	if (!strnicmp(argv[i], "none", strlen(argv[i])))
+		*locking_parms = 1;
+	else if (!strnicmp(argv[i + 1], "locking", strlen(argv[i + 1]))) {
+		*locking_type = &locking_none;
+		*locking_parms = 2;
+	} else if (!strnicmp(argv[i + 1], "cluster", strlen(argv[i + 1]))) {
+		*locking_type = &locking_cluster;
+		/* FIXME: namespace. */
+		*locking_parms = 3;
+	}
+
+	return *locking_parms == 1 ? -EINVAL : 0;
+}
+
+/* Set backing device information properties of RAID set. */
+static void rs_set_bdi(struct raid_set *rs, unsigned stripes, unsigned chunks)
+{
+	unsigned p, ra_pages;
+	struct mapped_device *md = dm_table_get_md(rs->ti->table);
+	struct backing_dev_info *bdi = &dm_disk(md)->queue->backing_dev_info;
+
+	/* Set read-ahead for the RAID set and the component devices. */
+	bdi->ra_pages = stripes * stripe_pages(rs, rs->set.io_size);
+	ra_pages = chunks * chunk_pages(rs->set.io_size);
+	for (p = rs->set.raid_devs; p--; ) {
+		struct request_queue *q = bdev_get_queue(rs->dev[p].dev->bdev);
+
+		q->backing_dev_info.ra_pages = ra_pages;
+	}
+
+	/* Set congested function and data. */
+	bdi->congested_fn = raid_set_congested;
+	bdi->congested_data = rs;
+
+	dm_put(md);
+}
+
+/* Get backing device information properties of RAID set. */
+static void rs_get_ra(struct raid_set *rs, unsigned *stripes, unsigned *chunks)
+{
+	struct mapped_device *md = dm_table_get_md(rs->ti->table);
+
+	 *stripes = dm_disk(md)->queue->backing_dev_info.ra_pages
+		    / stripe_pages(rs, rs->set.io_size);
+	*chunks = bdev_get_queue(rs->dev->dev->bdev)->backing_dev_info.ra_pages
+		  / chunk_pages(rs->set.io_size);
+
+	dm_put(md);
+}
+
+/*
+ * Construct a RAID4/5 mapping:
+ *
+ * log_type #log_params <log_params> \
+ * raid_type [#parity_dev] #raid_variable_params <raid_params> \
+ * [locking "none"/"cluster"]
+ * #raid_devs #dev_to_initialize [<dev_path> <offset>]{3,}
+ *
+ * log_type = "core"/"disk",
+ * #log_params = 1-3 (1-2 for core dirty log type, 3 for disk dirty log only)
+ * log_params = [dirty_log_path] region_size [[no]sync])
+ *
+ * raid_type = "raid4", "raid5_la", "raid5_ra", "raid5_ls", "raid5_rs"
+ *
+ * #parity_dev = N if raid_type = "raid4"
+ * o N = -1: pick default = last device
+ * o N >= 0 and < #raid_devs: parity device index
+ *
+ * #raid_variable_params = 0-5; raid_params (-1 = default):
+ *   [chunk_size [#stripes [io_size [recover_io_size [%recovery_bandwidth]]]]]
+ *   o chunk_size (unit to calculate drive addresses; must be 2^^n, > 8
+ *     and <= CHUNK_SIZE_MAX)
+ *   o #stripes is number of stripes allocated to stripe cache
+ *     (must be > 1 and < STRIPES_MAX)
+ *   o io_size (io unit size per device in sectors; must be 2^^n and > 8)
+ *   o recover_io_size (io unit size per device for recovery in sectors;
+       must be 2^^n, > SECTORS_PER_PAGE and <= region_size)
+ *   o %recovery_bandwith is the maximum amount spend for recovery during
+ *     application io (1-100%)
+ * If raid_variable_params = 0, defaults will be used.
+ * Any raid_variable_param can be set to -1 to apply a default
+ *
+ * #raid_devs = N (N >= 3)
+ *
+ * #dev_to_initialize = N
+ * -1: initialize parity on all devices
+ * >= 0 and < #raid_devs: initialize raid_path; used to force reconstruction
+ * of a failed devices content after replacement
+ *
+ * <dev_path> = device_path (eg, /dev/sdd1)
+ * <offset>   = begin at offset on <dev_path>
+ *
+ */
+#define	MIN_PARMS	13
+static int raid_ctr(struct dm_target *ti, unsigned argc, char **argv)
+{
+	int bandwidth = BANDWIDTH, bandwidth_parm = -1,
+	    chunk_size = CHUNK_SIZE, chunk_size_parm = -1,
+	    dev_to_init, dl_parms, locking_parms, parity_parm, pi = -1,
+	    i, io_size = IO_SIZE, io_size_parm = -1,
+	    r, raid_devs, raid_parms,
+	    recover_io_size = RECOVER_IO_SIZE, recover_io_size_parm = -1,
+	    stripes = STRIPES, stripes_parm = -1;
+	unsigned speed;
+	sector_t tmp, sectors_per_dev;
+	struct dm_raid45_locking_type *locking;
+	struct raid_set *rs;
+	struct raid_type *raid_type;
+
+	/* Ensure minimum number of parameters. */
+	if (argc < MIN_PARMS)
+		TI_ERR("Not enough parameters");
+
+	/* Fetch # of dirty log parameters. */
+	if (sscanf(argv[1], "%d", &dl_parms) != 1
+	    || !range_ok(dl_parms, 1, 4711))
+		TI_ERR("Bad dirty log parameters number");
+
+	/* Check raid_type. */
+	raid_type = get_raid_type(argv[dl_parms + 2]);
+	if (!raid_type)
+		TI_ERR("Bad raid type");
+
+	/* In case of RAID4, parity drive is selectable. */
+	parity_parm = !!(raid_type->level == raid4);
+
+	/* Handle variable number of RAID parameters. */
+	r = raid_variable_parms(ti, argv, dl_parms + parity_parm + 3,
+				&raid_parms,
+				&chunk_size, &chunk_size_parm,
+				&stripes, &stripes_parm,
+				&io_size, &io_size_parm,
+				&recover_io_size, &recover_io_size_parm,
+				&bandwidth, &bandwidth_parm);
+	if (r)
+		return r;
+
+	r = raid_locking_parms(ti, argv,
+			       dl_parms + parity_parm + raid_parms + 4,
+			       &locking_parms, &locking);
+	if (r)
+		return r;
+
+	/* # of raid devices. */
+	i = dl_parms + parity_parm + raid_parms + locking_parms + 4;
+	if (sscanf(argv[i], "%d", &raid_devs) != 1 ||
+	    raid_devs < raid_type->minimal_devs)
+		TI_ERR("Invalid number of raid devices");
+
+	/* In case of RAID4, check parity drive index is in limits. */
+	if (raid_type->level == raid4) {
+		/* Fetch index of parity device. */
+		if (sscanf(argv[dl_parms + 3], "%d", &pi) != 1 ||
+		    !range_ok(pi, 0, raid_devs - 1))
+			TI_ERR("Invalid RAID4 parity device index");
+	}
+
+	/*
+	 * Index of device to initialize starts at 0
+	 *
+	 * o -1 -> don't initialize a particular device,
+	 * o 0..raid_devs-1 -> initialize respective device
+	 *   (used for reconstruction of a replaced device)
+	 */
+	if (sscanf
+	    (argv[dl_parms + parity_parm + raid_parms + locking_parms + 5],
+	     "%d", &dev_to_init) != 1
+	    || !range_ok(dev_to_init, -1, raid_devs - 1))
+		TI_ERR("Invalid number for raid device to initialize");
+
+	/* Check # of raid device arguments. */
+	if (argc - dl_parms - parity_parm - raid_parms - 6 !=
+	    2 * raid_devs)
+		TI_ERR("Wrong number of raid device/offset arguments");
+
+	/*
+	 * Check that the table length is devisable
+	 * w/o rest by (raid_devs - parity_devs)
+	 */
+	if (!multiple(ti->len, raid_devs - raid_type->parity_devs,
+		      &sectors_per_dev))
+		TI_ERR
+		    ("Target length not divisable by number of data devices");
+
+	/*
+	 * Check that the device size is
+	 * devisable w/o rest by chunk size
+	 */
+	if (!multiple(sectors_per_dev, chunk_size, &tmp))
+		TI_ERR("Device length not divisable by chunk_size");
+
+	/****************************************************************
+	 * Now that we checked the constructor arguments ->
+	 * let's allocate the RAID set
+	 ****************************************************************/
+	r = context_alloc(&rs, raid_type, stripes, chunk_size, io_size,
+			  recover_io_size, raid_devs, sectors_per_dev,
+			  ti, dl_parms, argv);
+	if (r)
+		return r;
+
+	/*
+	 * Set these here in order to avoid passing
+	 * too many arguments to context_alloc()
+	 */
+	rs->set.dev_to_init_parm = dev_to_init;
+	rs->set.dev_to_init = dev_to_init;
+	rs->set.pi_parm = pi;
+	rs->set.pi = (pi == -1) ? rs->set.data_devs : pi;
+	rs->set.raid_parms = raid_parms;
+	rs->set.chunk_size_parm = chunk_size_parm;
+	rs->set.io_size_parm = io_size_parm;
+	rs->sc.stripes_parm = stripes_parm;
+	rs->recover.io_size_parm = recover_io_size_parm;
+	rs->recover.bandwidth_parm = bandwidth_parm;
+	recover_set_bandwidth(rs, bandwidth);
+
+	/* Use locking type to lock stripe access. */
+	rs->locking = locking;
+
+	/* Get the device/offset tupels. */
+	argv += dl_parms + 6 + parity_parm + raid_parms;
+	r = dev_parms(ti, rs, argv, &i);
+	if (r)
+		goto err;
+
+	/* Initialize recovery. */
+	rs->recover.start_jiffies = jiffies;
+	rs->recover.end_jiffies = 0;
+	recovery_region_reset(rs);
+
+	/* Allow for recovery of any nosync regions. */
+	SetRSRecover(rs);
+
+	/* Set backing device information (eg. read ahead). */
+	rs_set_bdi(rs, chunk_size * 2, io_size * 4);
+	SetRSCheckOverwrite(rs); /* Allow chunk overwrite checks. */
+
+	speed = xor_optimize(rs); /* Select best xor algorithm. */
+
+	/* Initialize work queue to handle this RAID set's io. */
+	r = rs_workqueue_init(rs);
+	if (r)
+		goto err;
+
+	raid_set_log(rs, speed); /* Log information about RAID set. */
+
+	/*
+	 * Make sure that dm core only hands maximum io size
+	 * length down and pays attention to io boundaries.
+	 */
+	ti->split_io = rs->set.io_size;
+	ti->private = rs;
+	return 0;
+
+err:
+	context_free(rs, ti, i);
+	return r;
+}
+
+/*
+ * Destruct a raid mapping
+ */
+static void raid_dtr(struct dm_target *ti)
+{
+	struct raid_set *rs = ti->private;
+
+	/* Indicate recovery end so that ios in flight drain. */
+	ClearRSRecover(rs);
+
+	wake_do_raid(rs);	/* Wake daemon. */
+	wait_ios(rs);		/* Wait for any io still being processed. */
+	destroy_workqueue(rs->io.wq);
+	context_free(rs, ti, rs->set.raid_devs);
+}
+
+/* Queues ios to RAID sets. */
+static inline void queue_bio(struct raid_set *rs, struct bio *bio)
+{
+	int wake;
+	struct bio_list *in = &rs->io.in;
+	spinlock_t *in_lock = &rs->io.in_lock;
+
+	spin_lock_irq(in_lock);
+	wake = bio_list_empty(in);
+	bio_list_add(in, bio);
+	spin_unlock_irq(in_lock);
+
+	/* Wake daemon if input list was empty. */
+	if (wake)
+		wake_do_raid(rs);
+}
+
+/* Raid mapping function. */
+static int raid_map(struct dm_target *ti, struct bio *bio,
+		    union map_info *map_context)
+{
+	/* I don't want to waste stripe cache capacity. */
+	if (bio_rw(bio) == READA)
+		return -EIO;
+	else {
+		struct raid_set *rs = ti->private;
+
+		/* REMOVEME: statistics. */
+		atomic_inc(rs->stats +
+			   (bio_data_dir(bio) == WRITE ?
+			    S_BIOS_WRITE : S_BIOS_READ));
+
+		/*
+		 * Get io reference to be waiting for to drop
+		 * to zero on device suspension/destruction.
+		 */
+		io_get(rs);
+		bio->bi_sector -= ti->begin;	/* Remap sector. */
+		queue_bio(rs, bio);		/* Queue to the daemon. */
+		return DM_MAPIO_SUBMITTED;	/* Handle later. */
+	}
+}
+
+/* Device suspend. */
+static void raid_postsuspend(struct dm_target *ti)
+{
+	struct raid_set *rs = ti->private;
+	struct dm_dirty_log *dl = rs->recover.dl;
+
+	SetRSSuspended(rs);
+
+	if (RSRecover(rs))
+		dm_rh_stop_recovery(rs->recover.rh); /* Wakes do_raid(). */
+	else
+		wake_do_raid(rs);
+
+	wait_ios(rs);	/* Wait for completion of all ios being processed. */
+	if (dl->type->postsuspend && dl->type->postsuspend(dl))
+		/* Suspend dirty log. */
+		/* FIXME: need better error handling. */
+		DMWARN("log suspend failed");
+}
+
+/* Device resume. */
+static void raid_resume(struct dm_target *ti)
+{
+	struct raid_set *rs = ti->private;
+	struct recover *rec = &rs->recover;
+	struct dm_dirty_log *dl = rec->dl;
+
+	if (dl->type->resume && dl->type->resume(dl))
+		/* Resume dirty log. */
+		/* FIXME: need better error handling. */
+		DMWARN("log resume failed");
+
+	rec->nr_regions_to_recover =
+	    rec->nr_regions - dl->type->get_sync_count(dl);
+
+	ClearRSSuspended(rs);
+
+	/* Reset any unfinished recovery. */
+	if (RSRecover(rs)) {
+		recovery_region_reset(rs);
+		dm_rh_start_recovery(rec->rh);/* Calls wake_do_raid(). */
+	} else
+		wake_do_raid(rs);
+}
+
+static INLINE unsigned sc_size(struct raid_set *rs)
+{
+	return to_sector(atomic_read(&rs->sc.stripes) *
+			 (sizeof(struct stripe) +
+			  (sizeof(struct stripe_set) +
+			   (sizeof(struct page_list) +
+			    to_bytes(rs->set.io_size) *
+			    rs->set.raid_devs)) +
+			  (rs->recover.
+			   end_jiffies ? 0 : to_bytes(rs->set.raid_devs *
+						      rs->recover.
+						      io_size))));
+}
+
+/* REMOVEME: status output for development. */
+static void
+raid_devel_stats(struct dm_target *ti, char *result,
+		 unsigned *size, unsigned maxlen)
+{
+	unsigned chunks, stripes, sz = *size;
+	unsigned long j;
+	char buf[BDEVNAME_SIZE], *p;
+	struct stats_map *sm, *sm_end = ARRAY_END(stats_map);
+	struct raid_set *rs = ti->private;
+	struct recover *rec = &rs->recover;
+	struct timespec ts;
+
+	DMEMIT("%s ", version);
+	DMEMIT("io_inprocess=%d ", atomic_read(&rs->io.in_process));
+	DMEMIT("io_inprocess_max=%d ", atomic_read(&rs->io.in_process_max));
+
+	for (sm = stats_map; sm < sm_end; sm++)
+		DMEMIT("%s%d", sm->str, atomic_read(rs->stats + sm->type));
+
+	DMEMIT(" overwrite=%s ", RSCheckOverwrite(rs) ? "on" : "off");
+	DMEMIT("sc=%u/%u/%u/%u/%u ", rs->set.chunk_size, rs->set.io_size,
+	       atomic_read(&rs->sc.stripes), rs->sc.hash.buckets,
+	       sc_size(rs));
+
+	j = (rec->end_jiffies ? rec->end_jiffies : jiffies) -
+	    rec->start_jiffies;
+	jiffies_to_timespec(j, &ts);
+	sprintf(buf, "%ld.%ld", ts.tv_sec, ts.tv_nsec);
+	p = strchr(buf, '.');
+	p[3] = 0;
+
+	DMEMIT("rg=%llu%s/%llu/%llu/%u %s ",
+	       (unsigned long long) rec->nr_regions_recovered,
+	       RSRegionGet(rs) ? "+" : "",
+	       (unsigned long long) rec->nr_regions_to_recover,
+	       (unsigned long long) rec->nr_regions, rec->bandwidth, buf);
+
+	rs_get_ra(rs, &stripes, &chunks);
+	DMEMIT("ra=%u/%u ", stripes, chunks);
+
+	*size = sz;
+}
+
+static int
+raid_status(struct dm_target *ti, status_type_t type,
+	    char *result, unsigned maxlen)
+{
+	unsigned i, sz = 0;
+	char buf[BDEVNAME_SIZE];
+	struct raid_set *rs = ti->private;
+
+	switch (type) {
+	case STATUSTYPE_INFO:
+		/* REMOVEME: statistics. */
+		if (RSDevelStats(rs))
+			raid_devel_stats(ti, result, &sz, maxlen);
+
+		DMEMIT("%u ", rs->set.raid_devs);
+
+		for (i = 0; i < rs->set.raid_devs; i++)
+			DMEMIT("%s ",
+			       format_dev_t(buf, rs->dev[i].dev->bdev->bd_dev));
+
+		DMEMIT("1 ");
+		for (i = 0; i < rs->set.raid_devs; i++) {
+			DMEMIT("%c", dev_operational(rs, i) ? 'A' : 'D');
+
+			if (rs->set.raid_type->level == raid4 &&
+			    i == rs->set.pi)
+				DMEMIT("p");
+
+			if (rs->set.dev_to_init == i)
+				DMEMIT("i");
+		}
+
+		break;
+
+	case STATUSTYPE_TABLE:
+		sz = rs->recover.dl->type->status(rs->recover.dl, type,
+						  result, maxlen);
+		DMEMIT("%s %u ", rs->set.raid_type->name,
+		       rs->set.raid_parms);
+
+		if (rs->set.raid_type->level == raid4)
+			DMEMIT("%d ", rs->set.pi_parm);
+
+		if (rs->set.raid_parms)
+			DMEMIT("%d ", rs->set.chunk_size_parm);
+
+		if (rs->set.raid_parms > 1)
+			DMEMIT("%d ", rs->sc.stripes_parm);
+
+		if (rs->set.raid_parms > 2)
+			DMEMIT("%d ", rs->set.io_size_parm);
+
+		if (rs->set.raid_parms > 3)
+			DMEMIT("%d ", rs->recover.io_size_parm);
+
+		if (rs->set.raid_parms > 4)
+			DMEMIT("%d ", rs->recover.bandwidth_parm);
+
+		DMEMIT("%u %d ", rs->set.raid_devs, rs->set.dev_to_init);
+
+		for (i = 0; i < rs->set.raid_devs; i++)
+			DMEMIT("%s %llu ",
+			       format_dev_t(buf,
+					    rs->dev[i].dev->bdev->bd_dev),
+			       (unsigned long long) rs->dev[i].start);
+	}
+
+	return 0;
+}
+
+/*
+ * Message interface
+ */
+enum raid_msg_actions {
+	act_bw,			/* Recovery bandwidth switch. */
+	act_dev,		/* Device failure switch. */
+	act_overwrite,		/* Stripe overwrite check. */
+	act_read_ahead,		/* Set read ahead. */
+	act_stats,		/* Development statistics switch. */
+	act_sc,			/* Stripe cache switch. */
+
+	act_on,			/* Set entity on. */
+	act_off,		/* Set entity off. */
+	act_reset,		/* Reset entity. */
+
+	act_set = act_on,	/* Set # absolute. */
+	act_grow = act_off,	/* Grow # by an amount. */
+	act_shrink = act_reset,	/* Shrink # by an amount. */
+};
+
+/* Turn a delta to absolute. */
+static int _absolute(unsigned long action, int act, int r)
+{
+	/* Make delta absolute. */
+	if (test_bit(act_set, &action))
+		;
+	else if (test_bit(act_grow, &action))
+		r += act;
+	else if (test_bit(act_shrink, &action))
+		r = act - r;
+	else
+		r = -EINVAL;
+
+	return r;
+}
+
+ /* Change recovery io bandwidth. */
+static int bandwidth_change(struct dm_msg *msg, void *context)
+{
+	struct raid_set *rs = context;
+	int act = rs->recover.bandwidth;
+	int bandwidth = DM_MSG_INT_ARG(msg);
+
+	if (range_ok(bandwidth, BANDWIDTH_MIN, BANDWIDTH_MAX)) {
+		/* Make delta bandwidth absolute. */
+		bandwidth = _absolute(msg->action, act, bandwidth);
+
+		/* Check range. */
+		if (range_ok(bandwidth, BANDWIDTH_MIN, BANDWIDTH_MAX)) {
+			recover_set_bandwidth(rs, bandwidth);
+			return 0;
+		}
+	}
+
+	set_bit(dm_msg_ret_arg, &msg->ret);
+	set_bit(dm_msg_ret_inval, &msg->ret);
+	return -EINVAL;
+}
+
+/* Change state of a device (running/offline). */
+/* FIXME: this only works while recovering!. */
+static int device_state(struct dm_msg *msg, void *context)
+{
+	int r;
+	const char *str = "is already ";
+	union dev_lookup dl = { .dev_name = DM_MSG_STR_ARG(msg) };
+	struct raid_set *rs = context;
+
+	r = raid_dev_lookup(rs, strchr(dl.dev_name, ':') ?
+			    bymajmin : byname, &dl);
+	if (r == -ENODEV) {
+		DMERR("device %s is no member of this set", dl.dev_name);
+		return r;
+	}
+
+	if (test_bit(act_off, &msg->action)) {
+		if (dev_operational(rs, r))
+			str = "";
+	} else if (!dev_operational(rs, r))
+		str = "";
+
+	DMINFO("/dev/%s %s%s", dl.dev_name, str,
+	       test_bit(act_off, &msg->action) ? "offline" : "running");
+
+	return test_bit(act_off, &msg->action) ?
+	       raid_set_check_and_degrade(rs, NULL, r) :
+	       raid_set_check_and_upgrade(rs, r);
+}
+
+/* Set/reset development feature flags. */
+static int devel_flags(struct dm_msg *msg, void *context)
+{
+	struct raid_set *rs = context;
+
+	if (test_bit(act_on, &msg->action))
+		return test_and_set_bit(msg->spec->parm,
+					&rs->io.flags) ? -EPERM : 0;
+	else if (test_bit(act_off, &msg->action))
+		return test_and_clear_bit(msg->spec->parm,
+					  &rs->io.flags) ? 0 : -EPERM;
+	else if (test_bit(act_reset, &msg->action)) {
+		if (test_bit(act_stats, &msg->action)) {
+			stats_reset(rs);
+			goto on;
+		} else if (test_bit(act_overwrite, &msg->action)) {
+on:
+			set_bit(msg->spec->parm, &rs->io.flags);
+			return 0;
+		}
+	}
+
+	return -EINVAL;
+}
+
+ /* Set stripe and chunk read ahead pages. */
+static int read_ahead_set(struct dm_msg *msg, void *context)
+{
+	int stripes = DM_MSG_INT_ARGS(msg, 0);
+	int chunks  = DM_MSG_INT_ARGS(msg, 1);
+
+	if (range_ok(stripes, 1, 512) &&
+	    range_ok(chunks, 1, 512)) {
+		rs_set_bdi(context, stripes, chunks);
+		return 0;
+	}
+
+	set_bit(dm_msg_ret_arg, &msg->ret);
+	set_bit(dm_msg_ret_inval, &msg->ret);
+	return -EINVAL;
+}
+
+/* Resize the stripe cache. */
+static int stripecache_resize(struct dm_msg *msg, void *context)
+{
+	int act, stripes;
+	struct raid_set *rs = context;
+
+	/* Deny permission in case the daemon is still shrinking!. */
+	if (atomic_read(&rs->sc.stripes_to_shrink))
+		return -EPERM;
+
+	stripes = DM_MSG_INT_ARG(msg);
+	if (stripes > 0) {
+		act = atomic_read(&rs->sc.stripes);
+
+		/* Make delta stripes absolute. */
+		stripes = _absolute(msg->action, act, stripes);
+
+		/*
+		 * Check range and that the # of stripes changes.
+		 * We can grow from gere but need to leave any
+		 * shrinking to the worker for synchronization.
+		 */
+		if (range_ok(stripes, STRIPES_MIN, STRIPES_MAX)) {
+			if (stripes > act)
+				return sc_grow(&rs->sc, stripes - act, SC_GROW);
+			else if (stripes < act) {
+				atomic_set(&rs->sc.stripes_to_shrink,
+					   act - stripes);
+				wake_do_raid(rs);
+			}
+
+			return 0;
+		}
+	}
+
+	set_bit(dm_msg_ret_arg, &msg->ret);
+	set_bit(dm_msg_ret_inval, &msg->ret);
+	return -EINVAL;
+}
+
+/* Parse the RAID message action. */
+/*
+ * 'ba[ndwidth] {se[t],g[row],sh[rink]} #'	# e.g 'ba se 50'
+ * 'de{vice] o[ffline]/r[unning] DevName/maj:min' # e.g 'device o /dev/sda'
+ * "o[verwrite]  {on,of[f],r[eset]}'		# e.g. 'o of'
+ * "r[ead_ahead] set #stripes #chunks		# e.g. 'r se 3 2'
+ * 'sta[tistics] {on,of[f],r[eset]}'		# e.g. 'stat of'
+ * 'str[ipecache] {se[t],g[row],sh[rink]} #'	# e.g. 'stripe set 1024'
+ *
+ */
+static int
+raid_message(struct dm_target *ti, unsigned argc, char **argv)
+{
+	/* Variables to store the parsed parameters im. */
+	static int i[2];
+	static unsigned long *i_arg[] = {
+		(unsigned long *) i + 0,
+		(unsigned long *) i + 1,
+	};
+	static char *p;
+	static unsigned long *p_arg[] = { (unsigned long *) &p };
+
+	/* Declare all message option strings. */
+	static char *str_sgs[] = { "set", "grow", "shrink" };
+	static char *str_dev[] = { "running", "offline" };
+	static char *str_oor[] = { "on", "off", "reset" };
+
+	/* Declare all actions. */
+	static unsigned long act_sgs[] = { act_set, act_grow, act_shrink };
+	static unsigned long act_oor[] = { act_on, act_off, act_reset };
+
+	/* Bandwidth option. */
+	static struct dm_message_option bw_opt = { 3, str_sgs, act_sgs };
+	static struct dm_message_argument bw_args = {
+		1, i_arg, { dm_msg_int_t }
+	};
+
+	/* Device option. */
+	static struct dm_message_option dev_opt = { 2, str_dev, act_oor };
+	static struct dm_message_argument dev_args = {
+		1, p_arg, { dm_msg_base_t }
+	};
+
+	/* Read ahead option. */
+	static struct dm_message_option ra_opt = { 1, str_sgs, act_sgs };
+	static struct dm_message_argument ra_args = {
+		2, i_arg, { dm_msg_int_t, dm_msg_int_t }
+	};
+
+	static struct dm_message_argument null_args = {
+		0, NULL, { dm_msg_int_t }
+	};
+
+	/* Overwrite and statistics option. */
+	static struct dm_message_option ovr_stats_opt = { 3, str_oor, act_oor };
+
+	/* Sripecache option. */
+	static struct dm_message_option stripe_opt = { 3, str_sgs, act_sgs };
+
+	/* Declare messages. */
+	static struct dm_msg_spec specs[] = {
+		{ "bandwidth", act_bw, &bw_opt, &bw_args,
+		  0, bandwidth_change },
+		{ "device", act_dev, &dev_opt, &dev_args,
+		  0, device_state },
+		{ "overwrite", act_overwrite, &ovr_stats_opt, &null_args,
+		  RS_CHECK_OVERWRITE, devel_flags },
+		{ "read_ahead", act_read_ahead, &ra_opt, &ra_args,
+		  0, read_ahead_set },
+		{ "statistics", act_stats, &ovr_stats_opt, &null_args,
+		  RS_DEVEL_STATS, devel_flags },
+		{ "stripecache", act_sc, &stripe_opt, &bw_args,
+		  0, stripecache_resize },
+	};
+
+	/* The message for the parser. */
+	struct dm_msg msg = {
+		.num_specs = ARRAY_SIZE(specs),
+		.specs = specs,
+	};
+
+	return dm_message_parse(TARGET, &msg, ti->private, argc, argv);
+}
+/*
+ * END message interface
+ */
+
+static struct target_type raid_target = {
+	.name = "raid45",
+	.version = {1, 0, 0},
+	.module = THIS_MODULE,
+	.ctr = raid_ctr,
+	.dtr = raid_dtr,
+	.map = raid_map,
+	.postsuspend = raid_postsuspend,
+	.resume = raid_resume,
+	.status = raid_status,
+	.message = raid_message,
+};
+
+static void init_exit(const char *bad_msg, const char *good_msg, int r)
+{
+	if (r)
+		DMERR("Failed to %sregister target [%d]", bad_msg, r);
+	else
+		DMINFO("%s %s", good_msg, version);
+}
+
+static int __init dm_raid_init(void)
+{
+	int r;
+
+	r = dm_register_target(&raid_target);
+	init_exit("", "initialized", r);
+	return r;
+}
+
+static void __exit dm_raid_exit(void)
+{
+	int r;
+
+	r = dm_unregister_target(&raid_target);
+	init_exit("un", "exit", r);
+}
+
+/* Module hooks. */
+module_init(dm_raid_init);
+module_exit(dm_raid_exit);
+
+MODULE_DESCRIPTION(DM_NAME " raid4/5 target");
+MODULE_AUTHOR("Heinz Mauelshagen <hjm@redhat.com>");
+MODULE_LICENSE("GPL");
--- /dev/null
+++ b/drivers/md/dm-raid45.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright (C) 2006-2008 Red Hat, Inc. All rights reserved.
+ *
+ * Module Author: Heinz Mauelshagen (Mauelshagen@RedHat.com)
+ *
+ * Locking definitions for the device-mapper RAID45 target.
+ *
+ * This file is released under the GPL.
+ *
+ */
+
+#ifndef _DM_RAID45_H
+#define _DM_RAID45_H
+
+/* Factor out to dm.h! */
+#define	STR_LEN(ptr, str)	(ptr), (str), strlen((ptr))
+
+enum dm_lock_type { DM_RAID45_EX, DM_RAID45_SHARED };
+
+struct dm_raid45_locking_type {
+	/* Request a lock on a stripe. */
+	void* (*lock)(sector_t key, enum dm_lock_type type);
+
+	/* Release a lock on a stripe. */
+	void (*unlock)(void *lock_handle);
+};
+
+#endif
--- /dev/null
+++ b/drivers/md/dm-regions.c
@@ -0,0 +1,723 @@
+/*
+ * Copyright (C) 2003 Sistina Software Limited.
+ * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
+ *
+ * This file is released under the GPL.
+ */
+
+#include <linux/dm-dirty-log.h>
+#include <linux/dm-regions.h>
+
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/vmalloc.h>
+
+#include "dm.h"
+#include "dm-bio-list.h"
+
+#define	DM_MSG_PREFIX	"region hash"
+
+/*-----------------------------------------------------------------
+ * Region hash
+ *
+ * A storage set (eg. RAID1, RAID5) splits itself up into discrete regions.
+ * Each region can be in one of three states:
+ *
+ * o clean
+ * o dirty,
+ * o nosync.
+ *
+ * There is no need to put clean regions in the hash.
+ *
+ *
+ * In addition to being present in the hash table a region _may_
+ * be present on one of three lists.
+ *
+ *   clean_regions: Regions on this list have no io pending to
+ *   them, they are in sync, we are no longer interested in them,
+ *   they are dull.  dm_rh_update_states() will remove them from the
+ *   hash table.
+ *
+ *   quiesced_regions: These regions have been spun down, ready
+ *   for recovery.  dm_rh_recovery_start() will remove regions from
+ *   this list and hand them to the caller, which will schedule the
+ *   recovery io.
+ *
+ *   recovered_regions: Regions that the caller has successfully
+ *   recovered.  dm_rh_update_states() will now schedule any delayed
+ *   io, up the recovery_count, and remove the region from the hash.
+ *
+ * There are 2 locks:
+ *   A rw spin lock 'hash_lock' protects just the hash table,
+ *   this is never held in write mode from interrupt context,
+ *   which I believe means that we only have to disable irqs when
+ *   doing a write lock.
+ *
+ *   An ordinary spin lock 'region_lock' that protects the three
+ *   lists in the region_hash, with the 'state', 'list' and
+ *   'delayed_bios' fields of the regions.  This is used from irq
+ *   context, so all other uses will have to suspend local irqs.
+ *---------------------------------------------------------------*/
+struct region_hash {
+	unsigned max_recovery; /* Max # of regions to recover in parallel */
+
+	/* Callback function to dispatch queued writes on recovered regions. */
+	void (*dispatch)(void *context, struct bio_list *bios, int error);
+	void *dispatch_context;
+
+	/* Callback function to wakeup callers worker thread. */
+	void (*wake)(void *context);
+	void *wake_context;
+
+	uint32_t region_size;
+	unsigned region_shift;
+
+	/* holds persistent region state */
+	struct dm_dirty_log *log;
+
+	/* hash table */
+	rwlock_t hash_lock;
+	mempool_t *region_pool;
+	unsigned mask;
+	unsigned nr_buckets;
+	unsigned prime;
+	unsigned shift;
+	struct list_head *buckets;
+
+	spinlock_t region_lock;
+	atomic_t recovery_in_flight;
+	struct semaphore recovery_count;
+	struct list_head clean_regions;
+	struct list_head quiesced_regions;
+	struct list_head recovered_regions;
+	struct list_head failed_recovered_regions;
+};
+
+struct region {
+	region_t key;
+	enum dm_rh_region_states state;
+	void *context;	/* Caller context. */
+
+	struct list_head hash_list;
+	struct list_head list;
+
+	atomic_t pending;
+	struct bio_list delayed_bios;
+};
+
+/*
+ * Conversion fns
+ */
+region_t dm_rh_sector_to_region(struct dm_rh_client *rh, sector_t sector)
+{
+	return sector >> ((struct region_hash *) rh)->region_shift;
+}
+EXPORT_SYMBOL_GPL(dm_rh_sector_to_region);
+
+region_t dm_rh_bio_to_region(struct dm_rh_client *rh, struct bio *bio)
+{
+	return dm_rh_sector_to_region(rh, bio->bi_sector);
+}
+EXPORT_SYMBOL_GPL(dm_rh_bio_to_region);
+
+sector_t dm_rh_region_to_sector(struct dm_rh_client *rh, region_t region)
+{
+	return region << ((struct region_hash *) rh)->region_shift;
+}
+EXPORT_SYMBOL_GPL(dm_rh_region_to_sector);
+
+/*
+ * Retrival fns.
+ */
+region_t dm_rh_get_region_key(struct dm_region *reg)
+{
+	return ((struct region *) reg)->key;
+}
+EXPORT_SYMBOL_GPL(dm_rh_get_region_key);
+
+sector_t dm_rh_get_region_size(struct dm_rh_client *rh)
+{
+	return ((struct region_hash *) rh)->region_size;
+}
+EXPORT_SYMBOL_GPL(dm_rh_get_region_size);
+
+/* Squirrel a context with a region. */
+void *dm_rh_reg_get_context(struct dm_region *reg)
+{
+	return ((struct region *) reg)->context;
+}
+EXPORT_SYMBOL_GPL(dm_rh_reg_get_context);
+
+void dm_rh_reg_set_context(struct dm_region *reg, void *context)
+{
+	((struct region *) reg)->context = context;
+}
+EXPORT_SYMBOL_GPL(dm_rh_reg_set_context);
+
+/*
+ * Create region hash client.
+ */
+#define MIN_REGIONS 64
+struct dm_rh_client *dm_rh_client_create(
+		 unsigned max_recovery,
+		 void (*dispatch)(void *dispatch_context,
+				  struct bio_list *bios, int error),
+		 void *dispatch_context,
+		 void (*wake)(void *wake_context), void *wake_context,
+		 struct dm_dirty_log *log, uint32_t region_size,
+		 region_t nr_regions)
+{
+	unsigned i;
+	unsigned nr_buckets, max_buckets;
+	unsigned hash_primes[] = {
+		/* Table of primes for rh_hash/table size optimization. */
+		3, 7, 13, 27, 53, 97, 193, 389, 769,
+		1543, 3079, 6151, 12289, 24593,
+	};
+	struct region_hash *rh;
+
+	if (region_size & (region_size - 1)) {
+		DMERR("region size must be 2^^n");
+		return ERR_PTR(-EINVAL);
+	}
+
+	/* Calculate a suitable number of buckets for our hash table. */
+	max_buckets = nr_regions >> 6;
+	for (nr_buckets = 128u; nr_buckets < max_buckets; nr_buckets <<= 1)
+		;
+	nr_buckets >>= 1;
+
+	rh = kmalloc(sizeof(*rh), GFP_KERNEL);
+	if (!rh) {
+		DMERR("unable to allocate region hash memory");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	rh->max_recovery = max_recovery;
+	rh->dispatch = dispatch;
+	rh->dispatch_context = dispatch_context;
+	rh->wake = wake;
+	rh->wake_context = wake_context;
+	rh->log = log;
+	rh->region_size = region_size;
+	rh->region_shift = ffs(region_size) - 1;
+	rwlock_init(&rh->hash_lock);
+	rh->mask = nr_buckets - 1;
+	rh->nr_buckets = nr_buckets;
+	rh->shift = ffs(nr_buckets);
+
+	/* Check prime array limits. */
+	i = rh->shift - 1 > ARRAY_SIZE(hash_primes) ?
+	    ARRAY_SIZE(hash_primes) - 1 : rh->shift - 2;
+	rh->prime = hash_primes[i];
+
+	rh->buckets = vmalloc(nr_buckets * sizeof(*rh->buckets));
+	if (!rh->buckets) {
+		DMERR("unable to allocate region hash bucket memory");
+		kfree(rh);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	for (i = 0; i < nr_buckets; i++)
+		INIT_LIST_HEAD(rh->buckets + i);
+
+	spin_lock_init(&rh->region_lock);
+	sema_init(&rh->recovery_count, 0);
+	atomic_set(&rh->recovery_in_flight, 0);
+	INIT_LIST_HEAD(&rh->clean_regions);
+	INIT_LIST_HEAD(&rh->quiesced_regions);
+	INIT_LIST_HEAD(&rh->recovered_regions);
+	INIT_LIST_HEAD(&rh->failed_recovered_regions);
+
+	rh->region_pool = mempool_create_kmalloc_pool(MIN_REGIONS,
+						      sizeof(struct region));
+	if (!rh->region_pool) {
+		vfree(rh->buckets);
+		kfree(rh);
+		rh = ERR_PTR(-ENOMEM);
+	}
+
+	return (struct dm_rh_client *) rh;
+}
+EXPORT_SYMBOL_GPL(dm_rh_client_create);
+
+void dm_rh_client_destroy(struct dm_rh_client *rh_in)
+{
+	unsigned h;
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg, *tmp;
+
+	BUG_ON(!list_empty(&rh->quiesced_regions));
+
+	for (h = 0; h < rh->nr_buckets; h++) {
+		list_for_each_entry_safe(reg, tmp, rh->buckets + h, hash_list) {
+			BUG_ON(atomic_read(&reg->pending));
+			mempool_free(reg, rh->region_pool);
+		}
+	}
+
+	if (rh->region_pool)
+		mempool_destroy(rh->region_pool);
+
+	vfree(rh->buckets);
+	kfree(rh);
+}
+EXPORT_SYMBOL_GPL(dm_rh_client_destroy);
+
+static inline unsigned rh_hash(struct region_hash *rh, region_t region)
+{
+	return (unsigned) ((region * rh->prime) >> rh->shift) & rh->mask;
+}
+
+static struct region *__rh_lookup(struct region_hash *rh, region_t region)
+{
+	struct region *reg;
+	struct list_head *bucket = rh->buckets + rh_hash(rh, region);
+
+	list_for_each_entry(reg, bucket, hash_list) {
+		if (reg->key == region)
+			return reg;
+	}
+
+	return NULL;
+}
+
+static void __rh_insert(struct region_hash *rh, struct region *reg)
+{
+	list_add(&reg->hash_list, rh->buckets + rh_hash(rh, reg->key));
+}
+
+static struct region *__rh_alloc(struct region_hash *rh, region_t region)
+{
+	struct region *reg, *nreg;
+
+	read_unlock(&rh->hash_lock);
+	nreg = mempool_alloc(rh->region_pool, GFP_ATOMIC);
+	if (unlikely(!nreg))
+		nreg = kmalloc(sizeof(*nreg), GFP_NOIO);
+
+	nreg->state = rh->log->type->in_sync(rh->log, region, 1) ?
+		      DM_RH_CLEAN : DM_RH_NOSYNC;
+	nreg->key = region;
+	INIT_LIST_HEAD(&nreg->list);
+	atomic_set(&nreg->pending, 0);
+	bio_list_init(&nreg->delayed_bios);
+
+	write_lock_irq(&rh->hash_lock);
+	reg = __rh_lookup(rh, region);
+	if (reg)
+		/* We lost the race. */
+		mempool_free(nreg, rh->region_pool);
+	else {
+		__rh_insert(rh, nreg);
+		if (nreg->state == DM_RH_CLEAN) {
+			spin_lock(&rh->region_lock);
+			list_add(&nreg->list, &rh->clean_regions);
+			spin_unlock(&rh->region_lock);
+		}
+
+		reg = nreg;
+	}
+
+	write_unlock_irq(&rh->hash_lock);
+	read_lock(&rh->hash_lock);
+	return reg;
+}
+
+static inline struct region *__rh_find(struct region_hash *rh, region_t region)
+{
+	struct region *reg;
+
+	reg = __rh_lookup(rh, region);
+	return reg ? reg : __rh_alloc(rh, region);
+}
+
+int dm_rh_get_state(struct dm_rh_client *rh_in, region_t region, int may_block)
+{
+	int r;
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg;
+
+	read_lock(&rh->hash_lock);
+	reg = __rh_lookup(rh, region);
+	read_unlock(&rh->hash_lock);
+
+	if (reg)
+		return reg->state;
+
+	/*
+	 * The region wasn't in the hash, so we fall back to the dirty log.
+	 */
+	r = rh->log->type->in_sync(rh->log, region, may_block);
+
+	/*
+	 * Any error from the dirty log (eg. -EWOULDBLOCK)
+	 * gets taken as a DM_RH_NOSYNC
+	 */
+	return r == 1 ? DM_RH_CLEAN : DM_RH_NOSYNC;
+}
+EXPORT_SYMBOL_GPL(dm_rh_get_state);
+
+void dm_rh_set_state(struct dm_rh_client *rh_in, region_t region,
+		     enum dm_rh_region_states state, int may_block)
+{
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg;
+	struct dm_dirty_log *log = rh->log;
+
+	if (state == DM_RH_NOSYNC)
+		log->type->set_region_sync(log, region, 0);
+	else if (state == DM_RH_CLEAN)
+		log->type->clear_region(log, region);
+	else if (state == DM_RH_DIRTY)
+		log->type->mark_region(log, region);
+
+	read_lock(&rh->hash_lock);
+	reg = __rh_find(rh, region);
+	reg->state = state;
+	read_unlock(&rh->hash_lock);
+}
+EXPORT_SYMBOL_GPL(dm_rh_set_state);
+
+void dm_rh_update_states(struct dm_rh_client *rh_in, int errors_handled)
+{
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg, *next;
+	LIST_HEAD(clean);
+	LIST_HEAD(recovered);
+	LIST_HEAD(failed_recovered);
+
+	/*
+	 * Quickly grab the lists and remove any regions from hash.
+	 */
+	write_lock_irq(&rh->hash_lock);
+	spin_lock(&rh->region_lock);
+	if (!list_empty(&rh->clean_regions)) {
+		list_splice_init(&rh->clean_regions, &clean);
+
+		list_for_each_entry(reg, &clean, list)
+			list_del(&reg->hash_list);
+	}
+
+	if (!list_empty(&rh->recovered_regions)) {
+		list_splice_init(&rh->recovered_regions, &recovered);
+
+		list_for_each_entry(reg, &recovered, list)
+			list_del(&reg->hash_list);
+	}
+
+	if (!list_empty(&rh->failed_recovered_regions)) {
+		list_splice_init(&rh->failed_recovered_regions,
+				 &failed_recovered);
+
+		list_for_each_entry(reg, &recovered, list)
+			list_del(&reg->hash_list);
+	}
+
+	spin_unlock(&rh->region_lock);
+	write_unlock_irq(&rh->hash_lock);
+
+	/*
+	 * All the regions on the recovered and clean lists have
+	 * now been pulled out of the system, so no need to do
+	 * any more locking.
+	 */
+	list_for_each_entry_safe(reg, next, &recovered, list) {
+		rh->log->type->clear_region(rh->log, reg->key);
+		rh->log->type->set_region_sync(rh->log, reg->key, 1);
+
+		if (reg->delayed_bios.head)
+			rh->dispatch(rh->dispatch_context,
+				     &reg->delayed_bios, 0);
+
+		up(&rh->recovery_count);
+		mempool_free(reg, rh->region_pool);
+	}
+
+	list_for_each_entry_safe(reg, next, &failed_recovered, list) {
+		rh->log->type->set_region_sync(rh->log, reg->key,
+					       errors_handled ? 0 : 1);
+		if (reg->delayed_bios.head)
+			rh->dispatch(rh->dispatch_context,
+				     &reg->delayed_bios, -EIO);
+
+		up(&rh->recovery_count);
+		mempool_free(reg, rh->region_pool);
+	}
+
+	list_for_each_entry_safe(reg, next, &clean, list) {
+		rh->log->type->clear_region(rh->log, reg->key);
+		mempool_free(reg, rh->region_pool);
+	}
+
+	dm_rh_flush(rh_in);
+}
+EXPORT_SYMBOL_GPL(dm_rh_update_states);
+
+void dm_rh_inc(struct dm_rh_client *rh_in, region_t region)
+{
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg;
+
+	read_lock(&rh->hash_lock);
+	reg = __rh_find(rh, region);
+	if (reg->state == DM_RH_CLEAN) {
+		rh->log->type->mark_region(rh->log, reg->key);
+
+		spin_lock_irq(&rh->region_lock);
+		reg->state = DM_RH_DIRTY;
+		list_del_init(&reg->list);	/* Take off the clean list. */
+		spin_unlock_irq(&rh->region_lock);
+	}
+
+	atomic_inc(&reg->pending);
+	read_unlock(&rh->hash_lock);
+}
+EXPORT_SYMBOL_GPL(dm_rh_inc);
+
+void dm_rh_inc_pending(struct dm_rh_client *rh_in, struct bio_list *bios)
+{
+	struct bio *bio;
+
+	for (bio = bios->head; bio; bio = bio->bi_next)
+		dm_rh_inc(rh_in, dm_rh_bio_to_region(rh_in, bio));
+}
+EXPORT_SYMBOL_GPL(dm_rh_inc_pending);
+
+int dm_rh_dec(struct dm_rh_client *rh_in, region_t region)
+{
+	int r = 0;
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg;
+
+	read_lock(&rh->hash_lock);
+	reg = __rh_lookup(rh, region);
+	read_unlock(&rh->hash_lock);
+
+	BUG_ON(!reg);
+
+	if (atomic_dec_and_test(&reg->pending)) {
+		unsigned long flags;
+
+		/*
+		 * There is no pending I/O for this region.
+		 * We can move the region to corresponding list for next action.
+		 * At this point, the region is not yet connected to any list.
+		 *
+		 * If the state is DM_RH_NOSYNC, the region should be kept off
+		 * from clean list.
+		 * The hash entry for DM_RH_NOSYNC will remain in memory
+		 * until the region is recovered or the map is reloaded.
+		 */
+
+		spin_lock_irqsave(&rh->region_lock, flags);
+		if (reg->state == DM_RH_RECOVERING)
+			list_add_tail(&reg->list, &rh->quiesced_regions);
+		else {
+			reg->state = DM_RH_CLEAN;
+			list_add(&reg->list, &rh->clean_regions);
+		}
+		spin_unlock_irqrestore(&rh->region_lock, flags);
+
+		r = 1;
+	}
+
+	return r;
+}
+EXPORT_SYMBOL_GPL(dm_rh_dec);
+
+/*
+ * Starts quiescing a region in preparation for recovery.
+ */
+static int __rh_recovery_prepare(struct region_hash *rh)
+{
+	int r;
+	region_t region;
+	struct region *reg;
+
+	/*
+	 * Ask the dirty log what's next.
+	 */
+	r = rh->log->type->get_resync_work(rh->log, &region);
+	if (r <= 0)
+		return r;
+
+	/*
+	 * Get this region, and start it quiescing
+	 * by setting the recovering flag.
+	 */
+	read_lock(&rh->hash_lock);
+	reg = __rh_find(rh, region);
+	read_unlock(&rh->hash_lock);
+
+	spin_lock_irq(&rh->region_lock);
+
+	reg->state = DM_RH_RECOVERING;
+
+	/* Already quiesced ? */
+	list_del_init(&reg->list);
+	if (!atomic_read(&reg->pending))
+		list_add(&reg->list, &rh->quiesced_regions);
+
+	spin_unlock_irq(&rh->region_lock);
+	return 1;
+}
+
+int dm_rh_recovery_prepare(struct dm_rh_client *rh_in)
+{
+	int r = 0;
+	struct region_hash *rh = (struct region_hash *) rh_in;
+
+	/* Extra reference to avoid race with rh_stop_recovery */
+	atomic_inc(&rh->recovery_in_flight);
+
+	while (!down_trylock(&rh->recovery_count)) {
+		atomic_inc(&rh->recovery_in_flight);
+
+		if (__rh_recovery_prepare(rh) <= 0) {
+			atomic_dec(&rh->recovery_in_flight);
+			up(&rh->recovery_count);
+			r = -ENOENT;
+			break;
+		}
+	}
+
+	/* Drop the extra reference. */
+	if (atomic_dec_and_test(&rh->recovery_in_flight))
+		r = -ESRCH;
+
+	return r;
+}
+EXPORT_SYMBOL_GPL(dm_rh_recovery_prepare);
+
+/*
+ * Returns any quiesced regions.
+ */
+struct dm_region *dm_rh_recovery_start(struct dm_rh_client *rh_in)
+{
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg = NULL;
+
+	spin_lock_irq(&rh->region_lock);
+	if (!list_empty(&rh->quiesced_regions)) {
+		reg = list_entry(rh->quiesced_regions.next,
+				 struct region, list);
+		list_del_init(&reg->list); /* Remove from the quiesced list. */
+	}
+
+	spin_unlock_irq(&rh->region_lock);
+	return (struct dm_region *) reg;
+}
+EXPORT_SYMBOL_GPL(dm_rh_recovery_start);
+
+/*
+ * Put region on list of recovered ones.
+ */
+void dm_rh_recovery_end(struct dm_rh_client *rh_in, struct dm_region *reg_in,
+			int error)
+{
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg = (struct region *) reg_in;
+
+	spin_lock_irq(&rh->region_lock);
+	if (error) {
+		reg->state = DM_RH_NOSYNC;
+		list_add(&reg->list, &rh->failed_recovered_regions);
+	} else
+		list_add(&reg->list, &rh->recovered_regions);
+
+	atomic_dec(&rh->recovery_in_flight);
+	spin_unlock_irq(&rh->region_lock);
+
+	rh->wake(rh->wake_context);
+	BUG_ON(atomic_read(&rh->recovery_in_flight) < 0);
+}
+EXPORT_SYMBOL_GPL(dm_rh_recovery_end);
+
+/* Return recovery in flight count. */
+int dm_rh_recovery_in_flight(struct dm_rh_client *rh_in)
+{
+	return atomic_read(&((struct region_hash *) rh_in)->recovery_in_flight);
+}
+EXPORT_SYMBOL_GPL(dm_rh_recovery_in_flight);
+
+int dm_rh_flush(struct dm_rh_client *rh_in)
+{
+	struct region_hash *rh = (struct region_hash *) rh_in;
+
+	return rh->log->type->flush(rh->log);
+}
+EXPORT_SYMBOL_GPL(dm_rh_flush);
+
+void dm_rh_delay_by_region(struct dm_rh_client *rh_in,
+			   struct bio *bio, region_t region)
+{
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg;
+
+	/* FIXME: locking. */
+	read_lock(&rh->hash_lock);
+	reg = __rh_find(rh, region);
+	bio_list_add(&reg->delayed_bios, bio);
+	read_unlock(&rh->hash_lock);
+}
+EXPORT_SYMBOL_GPL(dm_rh_delay_by_region);
+
+void dm_rh_delay(struct dm_rh_client *rh_in, struct bio *bio)
+{
+	return dm_rh_delay_by_region(rh_in, bio,
+				     dm_rh_bio_to_region(rh_in, bio));
+}
+EXPORT_SYMBOL_GPL(dm_rh_delay);
+
+void dm_rh_dispatch_bios(struct dm_rh_client *rh_in,
+			 region_t region, int error)
+{
+	struct region_hash *rh = (struct region_hash *) rh_in;
+	struct region *reg;
+	struct bio_list delayed_bios;
+
+	/* FIXME: locking. */
+	read_lock(&rh->hash_lock);
+	reg = __rh_find(rh, region);
+	BUG_ON(!reg);
+	delayed_bios = reg->delayed_bios;
+	bio_list_init(&reg->delayed_bios);
+	read_unlock(&rh->hash_lock);
+
+	if (delayed_bios.head)
+		rh->dispatch(rh->dispatch_context, &delayed_bios, error);
+
+	up(&rh->recovery_count);
+}
+EXPORT_SYMBOL_GPL(dm_rh_dispatch_bios);
+
+void dm_rh_stop_recovery(struct dm_rh_client *rh_in)
+{
+	int i;
+	struct region_hash *rh = (struct region_hash *) rh_in;
+
+	rh->wake(rh->wake_context);
+
+	/* wait for any recovering regions */
+	for (i = 0; i < rh->max_recovery; i++)
+		down(&rh->recovery_count);
+}
+EXPORT_SYMBOL_GPL(dm_rh_stop_recovery);
+
+void dm_rh_start_recovery(struct dm_rh_client *rh_in)
+{
+	int i;
+	struct region_hash *rh = (struct region_hash *) rh_in;
+
+	for (i = 0; i < rh->max_recovery; i++)
+		up(&rh->recovery_count);
+
+	rh->wake(rh->wake_context);
+}
+EXPORT_SYMBOL_GPL(dm_rh_start_recovery);
+
+MODULE_DESCRIPTION(DM_NAME " region hash");
+MODULE_AUTHOR("Joe Thornber/Heinz Mauelshagen <hjm@redhat.com>");
+MODULE_LICENSE("GPL");
--- a/drivers/md/Kconfig
+++ b/drivers/md/Kconfig
@@ -269,6 +269,14 @@ config DM_DELAY
 
 	If unsure, say N.
 
+config DM_RAID45
+	tristate "RAID 4/5 target (EXPERIMENTAL)"
+	depends on BLK_DEV_DM && EXPERIMENTAL
+	---help---
+	A target that supports RAID4 and RAID5 mappings.
+
+	If unsure, say N.
+
 config DM_UEVENT
 	bool "DM uevents (EXPERIMENTAL)"
 	depends on BLK_DEV_DM && EXPERIMENTAL
--- a/drivers/md/Makefile
+++ b/drivers/md/Makefile
@@ -34,7 +34,9 @@ obj-$(CONFIG_DM_CRYPT)		+= dm-crypt.o
 obj-$(CONFIG_DM_DELAY)		+= dm-delay.o
 obj-$(CONFIG_DM_MULTIPATH)	+= dm-multipath.o dm-round-robin.o
 obj-$(CONFIG_DM_SNAPSHOT)	+= dm-snapshot.o
-obj-$(CONFIG_DM_MIRROR)		+= dm-mirror.o dm-log.o
+obj-$(CONFIG_DM_MIRROR)		+= dm-mirror.o dm-regions.o dm-log.o
+obj-$(CONFIG_DM_RAID45)		+= dm-raid45.o dm-log.o dm-memcache.o \
+				   dm-regions.o dm-message.o
 obj-$(CONFIG_DM_ZERO)		+= dm-zero.o
 
 quiet_cmd_unroll = UNROLL  $@
--- /dev/null
+++ b/include/linux/dm-regions.h
@@ -0,0 +1,115 @@
+/*
+ * Copyright (C) 2003 Sistina Software Limited.
+ * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
+ *
+ * Device-Mapper dirty region hash interface.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef DM_REGION_HASH_H
+#define DM_REGION_HASH_H
+
+#include <linux/dm-dirty-log.h>
+
+/*-----------------------------------------------------------------
+ * Region hash
+ *----------------------------------------------------------------*/
+struct dm_rh_client;
+struct dm_region;
+
+/*
+ * States a region can have.
+ */
+enum dm_rh_region_states {
+	DM_RH_CLEAN	 = 0x01,	/* No writes in flight. */
+	DM_RH_DIRTY	 = 0x02,	/* Writes in flight. */
+	DM_RH_NOSYNC	 = 0x04,	/* Out of sync. */
+	DM_RH_RECOVERING = 0x08,	/* Under resynchronization. */
+};
+
+/*
+ * Region hash create/destroy.
+ */
+struct bio_list;
+struct dm_rh_client *dm_rh_client_create(
+		 unsigned max_recovery,
+		 void (*dispatch)(void *dispatch_context,
+				  struct bio_list *bios, int error),
+		 void *dispatch_context,
+		 void (*wake)(void *wake_context), void *wake_context,
+		 struct dm_dirty_log *log, uint32_t region_size,
+		 region_t nr_regions);
+void dm_rh_client_destroy(struct dm_rh_client *rh);
+
+/*
+ * Conversion fns:
+ *
+ *   bio -> region
+ *   sector -> region
+ *   region -> sector
+ */
+region_t dm_rh_bio_to_region(struct dm_rh_client *rh, struct bio *bio);
+region_t dm_rh_sector_to_region(struct dm_rh_client *rh, sector_t sector);
+sector_t dm_rh_region_to_sector(struct dm_rh_client *rh, region_t region);
+
+/*
+ * Functions to set a caller context in a region.
+ */
+void *dm_rh_reg_get_context(struct dm_region *reg);
+void dm_rh_reg_set_context(struct dm_region *reg, void *context);
+
+/*
+ * Get region size and key (ie. number of the region).
+ */
+sector_t dm_rh_get_region_size(struct dm_rh_client *rh);
+sector_t dm_rh_get_region_key(struct dm_region *reg);
+
+/*
+ * Get/set/update region state (and dirty log).
+ *
+ * dm_rh_update_states
+ * 	@errors_handled != 0 influences
+ *	that the state of the region will be kept NOSYNC
+ */
+int dm_rh_get_state(struct dm_rh_client *rh, region_t region, int may_block);
+void dm_rh_set_state(struct dm_rh_client *rh, region_t region,
+		     enum dm_rh_region_states state, int may_block);
+void dm_rh_update_states(struct dm_rh_client *rh, int errors_handled);
+
+/* Flush the region hash and dirty log. */
+int dm_rh_flush(struct dm_rh_client *rh);
+
+/* Inc/dec pending count on regions. */
+void dm_rh_inc(struct dm_rh_client *rh, region_t region);
+void dm_rh_inc_pending(struct dm_rh_client *rh, struct bio_list *bios);
+int dm_rh_dec(struct dm_rh_client *rh, region_t region);
+
+/* Delay bios on regions. */
+void dm_rh_delay(struct dm_rh_client *rh, struct bio *bio);
+void dm_rh_delay_by_region(struct dm_rh_client *rh,
+			   struct bio *bio, region_t region);
+
+/*
+ * Normally, the region hash will automatically call the dispatch function.
+ * dm_rh_dispatch_bios() is for intentional dispatching of bios.
+ */
+void dm_rh_dispatch_bios(struct dm_rh_client *rh, region_t region, int error);
+
+/*
+ * Region recovery control.
+ */
+/* Prepare some regions for recovery by starting to quiesce them. */
+int dm_rh_recovery_prepare(struct dm_rh_client *rh);
+/* Try fetching a quiesced region for recovery. */
+struct dm_region *dm_rh_recovery_start(struct dm_rh_client *rh);
+/* Report recovery end on a region. */
+void dm_rh_recovery_end(struct dm_rh_client *rh, struct dm_region *reg,
+			int error);
+/* Check for amount of recoveries in flight. */
+int dm_rh_recovery_in_flight(struct dm_rh_client *rh);
+/* Start/stop recovery. */
+void dm_rh_stop_recovery(struct dm_rh_client *rh);
+void dm_rh_start_recovery(struct dm_rh_client *rh);
+
+#endif /* #ifdef DM_REGION_HASH_H */