Btrfs: RAID5 and RAID6

This builds on David Woodhouse's original Btrfs raid5/6 implementation.
The code has changed quite a bit, blame Chris Mason for any bugs.

Read/modify/write is done after the higher levels of the filesystem have
prepared a given bio.  This means the higher layers are not responsible
for building full stripes, and they don't need to query for the topology
of the extents that may get allocated during delayed allocation runs.
It also means different files can easily share the same stripe.

But, it does expose us to incorrect parity if we crash or lose power
while doing a read/modify/write cycle.  This will be addressed in a
later commit.

Scrub is unable to repair crc errors on raid5/6 chunks.

Discard does not work on raid5/6 (yet)

The stripe size is fixed at 64KiB per disk.  This will be tunable
in a later commit.

Signed-off-by: Chris Mason <chris.mason@fusionio.com>

1 files changed, 1647 insertions, 0 deletions

diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c
new file mode 100644
index 000000000000..d02510f34936
--- /dev/null
+++ b/fs/btrfs/raid56.c
@@ -0,0 +1,1647 @@
+/*
+ * Copyright (C) 2012 Fusion-io  All rights reserved.
+ * Copyright (C) 2012 Intel Corp. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License v2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+#include <linux/sched.h>
+#include <linux/wait.h>
+#include <linux/bio.h>
+#include <linux/slab.h>
+#include <linux/buffer_head.h>
+#include <linux/blkdev.h>
+#include <linux/random.h>
+#include <linux/iocontext.h>
+#include <linux/capability.h>
+#include <linux/ratelimit.h>
+#include <linux/kthread.h>
+#include <linux/raid/pq.h>
+#include <linux/hash.h>
+#include <linux/list_sort.h>
+#include <linux/raid/xor.h>
+#include <asm/div64.h>
+#include "compat.h"
+#include "ctree.h"
+#include "extent_map.h"
+#include "disk-io.h"
+#include "transaction.h"
+#include "print-tree.h"
+#include "volumes.h"
+#include "raid56.h"
+#include "async-thread.h"
+#include "check-integrity.h"
+#include "rcu-string.h"
+
+/* set when additional merges to this rbio are not allowed */
+#define RBIO_RMW_LOCKED_BIT	1
+
+struct btrfs_raid_bio {
+	struct btrfs_fs_info *fs_info;
+	struct btrfs_bio *bbio;
+
+	/*
+	 * logical block numbers for the start of each stripe
+	 * The last one or two are p/q.  These are sorted,
+	 * so raid_map[0] is the start of our full stripe
+	 */
+	u64 *raid_map;
+
+	/* while we're doing rmw on a stripe
+	 * we put it into a hash table so we can
+	 * lock the stripe and merge more rbios
+	 * into it.
+	 */
+	struct list_head hash_list;
+
+	/*
+	 * for scheduling work in the helper threads
+	 */
+	struct btrfs_work work;
+
+	/*
+	 * bio list and bio_list_lock are used
+	 * to add more bios into the stripe
+	 * in hopes of avoiding the full rmw
+	 */
+	struct bio_list bio_list;
+	spinlock_t bio_list_lock;
+
+	/*
+	 * also protected by the bio_list_lock, the
+	 * stripe locking code uses plug_list to hand off
+	 * the stripe lock to the next pending IO
+	 */
+	struct list_head plug_list;
+
+	/*
+	 * flags that tell us if it is safe to
+	 * merge with this bio
+	 */
+	unsigned long flags;
+
+	/* size of each individual stripe on disk */
+	int stripe_len;
+
+	/* number of data stripes (no p/q) */
+	int nr_data;
+
+	/*
+	 * set if we're doing a parity rebuild
+	 * for a read from higher up, which is handled
+	 * differently from a parity rebuild as part of
+	 * rmw
+	 */
+	int read_rebuild;
+
+	/* first bad stripe */
+	int faila;
+
+	/* second bad stripe (for raid6 use) */
+	int failb;
+
+	/*
+	 * number of pages needed to represent the full
+	 * stripe
+	 */
+	int nr_pages;
+
+	/*
+	 * size of all the bios in the bio_list.  This
+	 * helps us decide if the rbio maps to a full
+	 * stripe or not
+	 */
+	int bio_list_bytes;
+
+	atomic_t refs;
+
+	/*
+	 * these are two arrays of pointers.  We allocate the
+	 * rbio big enough to hold them both and setup their
+	 * locations when the rbio is allocated
+	 */
+
+	/* pointers to pages that we allocated for
+	 * reading/writing stripes directly from the disk (including P/Q)
+	 */
+	struct page **stripe_pages;
+
+	/*
+	 * pointers to the pages in the bio_list.  Stored
+	 * here for faster lookup
+	 */
+	struct page **bio_pages;
+};
+
+static int __raid56_parity_recover(struct btrfs_raid_bio *rbio);
+static noinline void finish_rmw(struct btrfs_raid_bio *rbio);
+static void rmw_work(struct btrfs_work *work);
+static void read_rebuild_work(struct btrfs_work *work);
+static void async_rmw_stripe(struct btrfs_raid_bio *rbio);
+static void async_read_rebuild(struct btrfs_raid_bio *rbio);
+static int fail_bio_stripe(struct btrfs_raid_bio *rbio, struct bio *bio);
+static int fail_rbio_index(struct btrfs_raid_bio *rbio, int failed);
+static void __free_raid_bio(struct btrfs_raid_bio *rbio);
+static void index_rbio_pages(struct btrfs_raid_bio *rbio);
+static int alloc_rbio_pages(struct btrfs_raid_bio *rbio);
+
+/*
+ * the stripe hash table is used for locking, and to collect
+ * bios in hopes of making a full stripe
+ */
+int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info)
+{
+	struct btrfs_stripe_hash_table *table;
+	struct btrfs_stripe_hash_table *x;
+	struct btrfs_stripe_hash *cur;
+	struct btrfs_stripe_hash *h;
+	int num_entries = 1 << BTRFS_STRIPE_HASH_TABLE_BITS;
+	int i;
+
+	if (info->stripe_hash_table)
+		return 0;
+
+	table = kzalloc(sizeof(*table) + sizeof(*h) * num_entries, GFP_NOFS);
+	if (!table)
+		return -ENOMEM;
+
+	table->table = (void *)(table + 1);
+	h = table->table;
+
+	for (i = 0; i < num_entries; i++) {
+		cur = h + i;
+		INIT_LIST_HEAD(&cur->hash_list);
+		spin_lock_init(&cur->lock);
+		init_waitqueue_head(&cur->wait);
+	}
+
+	x = cmpxchg(&info->stripe_hash_table, NULL, table);
+	if (x)
+		kfree(x);
+	return 0;
+}
+
+/*
+ * we hash on the first logical address of the stripe
+ */
+static int rbio_bucket(struct btrfs_raid_bio *rbio)
+{
+	u64 num = rbio->raid_map[0];
+
+	/*
+	 * we shift down quite a bit.  We're using byte
+	 * addressing, and most of the lower bits are zeros.
+	 * This tends to upset hash_64, and it consistently
+	 * returns just one or two different values.
+	 *
+	 * shifting off the lower bits fixes things.
+	 */
+	return hash_64(num >> 16, BTRFS_STRIPE_HASH_TABLE_BITS);
+}
+
+/*
+ * merging means we take the bio_list from the victim and
+ * splice it into the destination.  The victim should
+ * be discarded afterwards.
+ *
+ * must be called with dest->rbio_list_lock held
+ */
+static void merge_rbio(struct btrfs_raid_bio *dest,
+		       struct btrfs_raid_bio *victim)
+{
+	bio_list_merge(&dest->bio_list, &victim->bio_list);
+	dest->bio_list_bytes += victim->bio_list_bytes;
+	bio_list_init(&victim->bio_list);
+}
+
+/*
+ * free the hash table used by unmount
+ */
+void btrfs_free_stripe_hash_table(struct btrfs_fs_info *info)
+{
+	if (!info->stripe_hash_table)
+		return;
+	kfree(info->stripe_hash_table);
+	info->stripe_hash_table = NULL;
+}
+
+/*
+ * helper function to run the xor_blocks api.  It is only
+ * able to do MAX_XOR_BLOCKS at a time, so we need to
+ * loop through.
+ */
+static void run_xor(void **pages, int src_cnt, ssize_t len)
+{
+	int src_off = 0;
+	int xor_src_cnt = 0;
+	void *dest = pages[src_cnt];
+
+	while(src_cnt > 0) {
+		xor_src_cnt = min(src_cnt, MAX_XOR_BLOCKS);
+		xor_blocks(xor_src_cnt, len, dest, pages + src_off);
+
+		src_cnt -= xor_src_cnt;
+		src_off += xor_src_cnt;
+	}
+}
+
+/*
+ * returns true if the bio list inside this rbio
+ * covers an entire stripe (no rmw required).
+ * Must be called with the bio list lock held, or
+ * at a time when you know it is impossible to add
+ * new bios into the list
+ */
+static int __rbio_is_full(struct btrfs_raid_bio *rbio)
+{
+	unsigned long size = rbio->bio_list_bytes;
+	int ret = 1;
+
+	if (size != rbio->nr_data * rbio->stripe_len)
+		ret = 0;
+
+	BUG_ON(size > rbio->nr_data * rbio->stripe_len);
+	return ret;
+}
+
+static int rbio_is_full(struct btrfs_raid_bio *rbio)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&rbio->bio_list_lock, flags);
+	ret = __rbio_is_full(rbio);
+	spin_unlock_irqrestore(&rbio->bio_list_lock, flags);
+	return ret;
+}
+
+/*
+ * returns 1 if it is safe to merge two rbios together.
+ * The merging is safe if the two rbios correspond to
+ * the same stripe and if they are both going in the same
+ * direction (read vs write), and if neither one is
+ * locked for final IO
+ *
+ * The caller is responsible for locking such that
+ * rmw_locked is safe to test
+ */
+static int rbio_can_merge(struct btrfs_raid_bio *last,
+			  struct btrfs_raid_bio *cur)
+{
+	if (test_bit(RBIO_RMW_LOCKED_BIT, &last->flags) ||
+	    test_bit(RBIO_RMW_LOCKED_BIT, &cur->flags))
+		return 0;
+
+	if (last->raid_map[0] !=
+	    cur->raid_map[0])
+		return 0;
+
+	/* reads can't merge with writes */
+	if (last->read_rebuild !=
+	    cur->read_rebuild) {
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * helper to index into the pstripe
+ */
+static struct page *rbio_pstripe_page(struct btrfs_raid_bio *rbio, int index)
+{
+	index += (rbio->nr_data * rbio->stripe_len) >> PAGE_CACHE_SHIFT;
+	return rbio->stripe_pages[index];
+}
+
+/*
+ * helper to index into the qstripe, returns null
+ * if there is no qstripe
+ */
+static struct page *rbio_qstripe_page(struct btrfs_raid_bio *rbio, int index)
+{
+	if (rbio->nr_data + 1 == rbio->bbio->num_stripes)
+		return NULL;
+
+	index += ((rbio->nr_data + 1) * rbio->stripe_len) >>
+		PAGE_CACHE_SHIFT;
+	return rbio->stripe_pages[index];
+}
+
+/*
+ * The first stripe in the table for a logical address
+ * has the lock.  rbios are added in one of three ways:
+ *
+ * 1) Nobody has the stripe locked yet.  The rbio is given
+ * the lock and 0 is returned.  The caller must start the IO
+ * themselves.
+ *
+ * 2) Someone has the stripe locked, but we're able to merge
+ * with the lock owner.  The rbio is freed and the IO will
+ * start automatically along with the existing rbio.  1 is returned.
+ *
+ * 3) Someone has the stripe locked, but we're not able to merge.
+ * The rbio is added to the lock owner's plug list, or merged into
+ * an rbio already on the plug list.  When the lock owner unlocks,
+ * the next rbio on the list is run and the IO is started automatically.
+ * 1 is returned
+ *
+ * If we return 0, the caller still owns the rbio and must continue with
+ * IO submission.  If we return 1, the caller must assume the rbio has
+ * already been freed.
+ */
+static noinline int lock_stripe_add(struct btrfs_raid_bio *rbio)
+{
+	int bucket = rbio_bucket(rbio);
+	struct btrfs_stripe_hash *h = rbio->fs_info->stripe_hash_table->table + bucket;
+	struct btrfs_raid_bio *cur;
+	struct btrfs_raid_bio *pending;
+	unsigned long flags;
+	DEFINE_WAIT(wait);
+	struct btrfs_raid_bio *freeit = NULL;
+	int ret = 0;
+	int walk = 0;
+
+	spin_lock_irqsave(&h->lock, flags);
+	list_for_each_entry(cur, &h->hash_list, hash_list) {
+		walk++;
+		if (cur->raid_map[0] == rbio->raid_map[0]) {
+			spin_lock(&cur->bio_list_lock);
+
+			/* can we merge into the lock owner? */
+			if (rbio_can_merge(cur, rbio)) {
+				merge_rbio(cur, rbio);
+				spin_unlock(&cur->bio_list_lock);
+				freeit = rbio;
+				ret = 1;
+				goto out;
+			}
+
+			/*
+			 * we couldn't merge with the running
+			 * rbio, see if we can merge with the
+			 * pending ones.  We don't have to
+			 * check for rmw_locked because there
+			 * is no way they are inside finish_rmw
+			 * right now
+			 */
+			list_for_each_entry(pending, &cur->plug_list,
+					    plug_list) {
+				if (rbio_can_merge(pending, rbio)) {
+					merge_rbio(pending, rbio);
+					spin_unlock(&cur->bio_list_lock);
+					freeit = rbio;
+					ret = 1;
+					goto out;
+				}
+			}
+
+			/* no merging, put us on the tail of the plug list,
+			 * our rbio will be started with the currently
+			 * running rbio unlocks
+			 */
+			list_add_tail(&rbio->plug_list, &cur->plug_list);
+			spin_unlock(&cur->bio_list_lock);
+			ret = 1;
+			goto out;
+		}
+	}
+
+	atomic_inc(&rbio->refs);
+	list_add(&rbio->hash_list, &h->hash_list);
+out:
+	spin_unlock_irqrestore(&h->lock, flags);
+	if (freeit)
+		__free_raid_bio(freeit);
+	return ret;
+}
+
+/*
+ * called as rmw or parity rebuild is completed.  If the plug list has more
+ * rbios waiting for this stripe, the next one on the list will be started
+ */
+static noinline void unlock_stripe(struct btrfs_raid_bio *rbio)
+{
+	int bucket;
+	struct btrfs_stripe_hash *h;
+	unsigned long flags;
+
+	bucket = rbio_bucket(rbio);
+	h = rbio->fs_info->stripe_hash_table->table + bucket;
+
+	spin_lock_irqsave(&h->lock, flags);
+	spin_lock(&rbio->bio_list_lock);
+
+	if (!list_empty(&rbio->hash_list)) {
+
+		list_del_init(&rbio->hash_list);
+		atomic_dec(&rbio->refs);
+
+		/*
+		 * we use the plug list to hold all the rbios
+		 * waiting for the chance to lock this stripe.
+		 * hand the lock over to one of them.
+		 */
+		if (!list_empty(&rbio->plug_list)) {
+			struct btrfs_raid_bio *next;
+			struct list_head *head = rbio->plug_list.next;
+
+			next = list_entry(head, struct btrfs_raid_bio,
+					  plug_list);
+
+			list_del_init(&rbio->plug_list);
+
+			list_add(&next->hash_list, &h->hash_list);
+			atomic_inc(&next->refs);
+			spin_unlock(&rbio->bio_list_lock);
+			spin_unlock_irqrestore(&h->lock, flags);
+
+			if (next->read_rebuild)
+				async_read_rebuild(next);
+			else
+				async_rmw_stripe(next);
+
+			goto done_nolock;
+
+		} else  if (waitqueue_active(&h->wait)) {
+			spin_unlock(&rbio->bio_list_lock);
+			spin_unlock_irqrestore(&h->lock, flags);
+			wake_up(&h->wait);
+			goto done_nolock;
+		}
+	}
+	spin_unlock(&rbio->bio_list_lock);
+	spin_unlock_irqrestore(&h->lock, flags);
+
+done_nolock:
+	return;
+}
+
+static void __free_raid_bio(struct btrfs_raid_bio *rbio)
+{
+	int i;
+
+	WARN_ON(atomic_read(&rbio->refs) < 0);
+	if (!atomic_dec_and_test(&rbio->refs))
+		return;
+
+	WARN_ON(!list_empty(&rbio->hash_list));
+	WARN_ON(!bio_list_empty(&rbio->bio_list));
+
+	for (i = 0; i < rbio->nr_pages; i++) {
+		if (rbio->stripe_pages[i]) {
+			__free_page(rbio->stripe_pages[i]);
+			rbio->stripe_pages[i] = NULL;
+		}
+	}
+	kfree(rbio->raid_map);
+	kfree(rbio->bbio);
+	kfree(rbio);
+}
+
+static void free_raid_bio(struct btrfs_raid_bio *rbio)
+{
+	unlock_stripe(rbio);
+	__free_raid_bio(rbio);
+}
+
+/*
+ * this frees the rbio and runs through all the bios in the
+ * bio_list and calls end_io on them
+ */
+static void rbio_orig_end_io(struct btrfs_raid_bio *rbio, int err, int uptodate)
+{
+	struct bio *cur = bio_list_get(&rbio->bio_list);
+	struct bio *next;
+	free_raid_bio(rbio);
+
+	while (cur) {
+		next = cur->bi_next;
+		cur->bi_next = NULL;
+		if (uptodate)
+			set_bit(BIO_UPTODATE, &cur->bi_flags);
+		bio_endio(cur, err);
+		cur = next;
+	}
+}
+
+/*
+ * end io function used by finish_rmw.  When we finally
+ * get here, we've written a full stripe
+ */
+static void raid_write_end_io(struct bio *bio, int err)
+{
+	struct btrfs_raid_bio *rbio = bio->bi_private;
+
+	if (err)
+		fail_bio_stripe(rbio, bio);
+
+	bio_put(bio);
+
+	if (!atomic_dec_and_test(&rbio->bbio->stripes_pending))
+		return;
+
+	err = 0;
+
+	/* OK, we have read all the stripes we need to. */
+	if (atomic_read(&rbio->bbio->error) > rbio->bbio->max_errors)
+		err = -EIO;
+
+	rbio_orig_end_io(rbio, err, 0);
+	return;
+}
+
+/*
+ * the read/modify/write code wants to use the original bio for
+ * any pages it included, and then use the rbio for everything
+ * else.  This function decides if a given index (stripe number)
+ * and page number in that stripe fall inside the original bio
+ * or the rbio.
+ *
+ * if you set bio_list_only, you'll get a NULL back for any ranges
+ * that are outside the bio_list
+ *
+ * This doesn't take any refs on anything, you get a bare page pointer
+ * and the caller must bump refs as required.
+ *
+ * You must call index_rbio_pages once before you can trust
+ * the answers from this function.
+ */
+static struct page *page_in_rbio(struct btrfs_raid_bio *rbio,
+				 int index, int pagenr, int bio_list_only)
+{
+	int chunk_page;
+	struct page *p = NULL;
+
+	chunk_page = index * (rbio->stripe_len >> PAGE_SHIFT) + pagenr;
+
+	spin_lock_irq(&rbio->bio_list_lock);
+	p = rbio->bio_pages[chunk_page];
+	spin_unlock_irq(&rbio->bio_list_lock);
+
+	if (p || bio_list_only)
+		return p;
+
+	return rbio->stripe_pages[chunk_page];
+}
+
+/*
+ * number of pages we need for the entire stripe across all the
+ * drives
+ */
+static unsigned long rbio_nr_pages(unsigned long stripe_len, int nr_stripes)
+{
+	unsigned long nr = stripe_len * nr_stripes;
+	return (nr + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+}
+
+/*
+ * allocation and initial setup for the btrfs_raid_bio.  Not
+ * this does not allocate any pages for rbio->pages.
+ */
+static struct btrfs_raid_bio *alloc_rbio(struct btrfs_root *root,
+			  struct btrfs_bio *bbio, u64 *raid_map,
+			  u64 stripe_len)
+{
+	struct btrfs_raid_bio *rbio;
+	int nr_data = 0;
+	int num_pages = rbio_nr_pages(stripe_len, bbio->num_stripes);
+	void *p;
+
+	rbio = kzalloc(sizeof(*rbio) + num_pages * sizeof(struct page *) * 2,
+			GFP_NOFS);
+	if (!rbio) {
+		kfree(raid_map);
+		kfree(bbio);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	bio_list_init(&rbio->bio_list);
+	INIT_LIST_HEAD(&rbio->plug_list);
+	spin_lock_init(&rbio->bio_list_lock);
+	INIT_LIST_HEAD(&rbio->hash_list);
+	rbio->bbio = bbio;
+	rbio->raid_map = raid_map;
+	rbio->fs_info = root->fs_info;
+	rbio->stripe_len = stripe_len;
+	rbio->nr_pages = num_pages;
+	rbio->faila = -1;
+	rbio->failb = -1;
+	atomic_set(&rbio->refs, 1);
+
+	/*
+	 * the stripe_pages and bio_pages array point to the extra
+	 * memory we allocated past the end of the rbio
+	 */
+	p = rbio + 1;
+	rbio->stripe_pages = p;
+	rbio->bio_pages = p + sizeof(struct page *) * num_pages;
+
+	if (raid_map[bbio->num_stripes - 1] == RAID6_Q_STRIPE)
+		nr_data = bbio->num_stripes - 2;
+	else
+		nr_data = bbio->num_stripes - 1;
+
+	rbio->nr_data = nr_data;
+	return rbio;
+}
+
+/* allocate pages for all the stripes in the bio, including parity */
+static int alloc_rbio_pages(struct btrfs_raid_bio *rbio)
+{
+	int i;
+	struct page *page;
+
+	for (i = 0; i < rbio->nr_pages; i++) {
+		if (rbio->stripe_pages[i])
+			continue;
+		page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
+		if (!page)
+			return -ENOMEM;
+		rbio->stripe_pages[i] = page;
+		ClearPageUptodate(page);
+	}
+	return 0;
+}
+
+/* allocate pages for just the p/q stripes */
+static int alloc_rbio_parity_pages(struct btrfs_raid_bio *rbio)
+{
+	int i;
+	struct page *page;
+
+	i = (rbio->nr_data * rbio->stripe_len) >> PAGE_CACHE_SHIFT;
+
+	for (; i < rbio->nr_pages; i++) {
+		if (rbio->stripe_pages[i])
+			continue;
+		page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
+		if (!page)
+			return -ENOMEM;
+		rbio->stripe_pages[i] = page;
+	}
+	return 0;
+}
+
+/*
+ * add a single page from a specific stripe into our list of bios for IO
+ * this will try to merge into existing bios if possible, and returns
+ * zero if all went well.
+ */
+int rbio_add_io_page(struct btrfs_raid_bio *rbio,
+		     struct bio_list *bio_list,
+		     struct page *page,
+		     int stripe_nr,
+		     unsigned long page_index,
+		     unsigned long bio_max_len)
+{
+	struct bio *last = bio_list->tail;
+	u64 last_end = 0;
+	int ret;
+	struct bio *bio;
+	struct btrfs_bio_stripe *stripe;
+	u64 disk_start;
+
+	stripe = &rbio->bbio->stripes[stripe_nr];
+	disk_start = stripe->physical + (page_index << PAGE_CACHE_SHIFT);
+
+	/* if the device is missing, just fail this stripe */
+	if (!stripe->dev->bdev)
+		return fail_rbio_index(rbio, stripe_nr);
+
+	/* see if we can add this page onto our existing bio */
+	if (last) {
+		last_end = (u64)last->bi_sector << 9;
+		last_end += last->bi_size;
+
+		/*
+		 * we can't merge these if they are from different
+		 * devices or if they are not contiguous
+		 */
+		if (last_end == disk_start && stripe->dev->bdev &&
+		    test_bit(BIO_UPTODATE, &last->bi_flags) &&
+		    last->bi_bdev == stripe->dev->bdev) {
+			ret = bio_add_page(last, page, PAGE_CACHE_SIZE, 0);
+			if (ret == PAGE_CACHE_SIZE)
+				return 0;
+		}
+	}
+
+	/* put a new bio on the list */
+	bio = bio_alloc(GFP_NOFS, bio_max_len >> PAGE_SHIFT?:1);
+	if (!bio)
+		return -ENOMEM;
+
+	bio->bi_size = 0;
+	bio->bi_bdev = stripe->dev->bdev;
+	bio->bi_sector = disk_start >> 9;
+	set_bit(BIO_UPTODATE, &bio->bi_flags);
+
+	bio_add_page(bio, page, PAGE_CACHE_SIZE, 0);
+	bio_list_add(bio_list, bio);
+	return 0;
+}
+
+/*
+ * while we're doing the read/modify/write cycle, we could
+ * have errors in reading pages off the disk.  This checks
+ * for errors and if we're not able to read the page it'll
+ * trigger parity reconstruction.  The rmw will be finished
+ * after we've reconstructed the failed stripes
+ */
+static void validate_rbio_for_rmw(struct btrfs_raid_bio *rbio)
+{
+	if (rbio->faila >= 0 || rbio->failb >= 0) {
+		BUG_ON(rbio->faila == rbio->bbio->num_stripes - 1);
+		__raid56_parity_recover(rbio);
+	} else {
+		finish_rmw(rbio);
+	}
+}
+
+/*
+ * these are just the pages from the rbio array, not from anything
+ * the FS sent down to us
+ */
+static struct page *rbio_stripe_page(struct btrfs_raid_bio *rbio, int stripe, int page)
+{
+	int index;
+	index = stripe * (rbio->stripe_len >> PAGE_CACHE_SHIFT);
+	index += page;
+	return rbio->stripe_pages[index];
+}
+
+/*
+ * helper function to walk our bio list and populate the bio_pages array with
+ * the result.  This seems expensive, but it is faster than constantly
+ * searching through the bio list as we setup the IO in finish_rmw or stripe
+ * reconstruction.
+ *
+ * This must be called before you trust the answers from page_in_rbio
+ */
+static void index_rbio_pages(struct btrfs_raid_bio *rbio)
+{
+	struct bio *bio;
+	u64 start;
+	unsigned long stripe_offset;
+	unsigned long page_index;
+	struct page *p;
+	int i;
+
+	spin_lock_irq(&rbio->bio_list_lock);
+	bio_list_for_each(bio, &rbio->bio_list) {
+		start = (u64)bio->bi_sector << 9;
+		stripe_offset = start - rbio->raid_map[0];
+		page_index = stripe_offset >> PAGE_CACHE_SHIFT;
+
+		for (i = 0; i < bio->bi_vcnt; i++) {
+			p = bio->bi_io_vec[i].bv_page;
+			rbio->bio_pages[page_index + i] = p;
+		}
+	}
+	spin_unlock_irq(&rbio->bio_list_lock);
+}
+
+/*
+ * this is called from one of two situations.  We either
+ * have a full stripe from the higher layers, or we've read all
+ * the missing bits off disk.
+ *
+ * This will calculate the parity and then send down any
+ * changed blocks.
+ */
+static noinline void finish_rmw(struct btrfs_raid_bio *rbio)
+{
+	struct btrfs_bio *bbio = rbio->bbio;
+	void *pointers[bbio->num_stripes];
+	int stripe_len = rbio->stripe_len;
+	int nr_data = rbio->nr_data;
+	int stripe;
+	int pagenr;
+	int p_stripe = -1;
+	int q_stripe = -1;
+	struct bio_list bio_list;
+	struct bio *bio;
+	int pages_per_stripe = stripe_len >> PAGE_CACHE_SHIFT;
+	int ret;
+
+	bio_list_init(&bio_list);
+
+	if (bbio->num_stripes - rbio->nr_data == 1) {
+		p_stripe = bbio->num_stripes - 1;
+	} else if (bbio->num_stripes - rbio->nr_data == 2) {
+		p_stripe = bbio->num_stripes - 2;
+		q_stripe = bbio->num_stripes - 1;
+	} else {
+		BUG();
+	}
+
+	/* at this point we either have a full stripe,
+	 * or we've read the full stripe from the drive.
+	 * recalculate the parity and write the new results.
+	 *
+	 * We're not allowed to add any new bios to the
+	 * bio list here, anyone else that wants to
+	 * change this stripe needs to do their own rmw.
+	 */
+	spin_lock_irq(&rbio->bio_list_lock);
+	set_bit(RBIO_RMW_LOCKED_BIT, &rbio->flags);
+	spin_unlock_irq(&rbio->bio_list_lock);
+
+	atomic_set(&rbio->bbio->error, 0);
+
+	/*
+	 * now that we've set rmw_locked, run through the
+	 * bio list one last time and map the page pointers
+	 */
+	index_rbio_pages(rbio);
+
+	for (pagenr = 0; pagenr < pages_per_stripe; pagenr++) {
+		struct page *p;
+		/* first collect one page from each data stripe */
+		for (stripe = 0; stripe < nr_data; stripe++) {
+			p = page_in_rbio(rbio, stripe, pagenr, 0);
+			pointers[stripe] = kmap(p);
+		}
+
+		/* then add the parity stripe */
+		p = rbio_pstripe_page(rbio, pagenr);
+		SetPageUptodate(p);
+		pointers[stripe++] = kmap(p);
+
+		if (q_stripe != -1) {
+
+			/*
+			 * raid6, add the qstripe and call the
+			 * library function to fill in our p/q
+			 */
+			p = rbio_qstripe_page(rbio, pagenr);
+			SetPageUptodate(p);
+			pointers[stripe++] = kmap(p);
+
+			raid6_call.gen_syndrome(bbio->num_stripes, PAGE_SIZE,
+						pointers);
+		} else {
+			/* raid5 */
+			memcpy(pointers[nr_data], pointers[0], PAGE_SIZE);
+			run_xor(pointers + 1, nr_data - 1, PAGE_CACHE_SIZE);
+		}
+
+
+		for (stripe = 0; stripe < bbio->num_stripes; stripe++)
+			kunmap(page_in_rbio(rbio, stripe, pagenr, 0));
+	}
+
+	/*
+	 * time to start writing.  Make bios for everything from the
+	 * higher layers (the bio_list in our rbio) and our p/q.  Ignore
+	 * everything else.
+	 */
+	for (stripe = 0; stripe < bbio->num_stripes; stripe++) {
+		for (pagenr = 0; pagenr < pages_per_stripe; pagenr++) {
+			struct page *page;
+			if (stripe < rbio->nr_data) {
+				page = page_in_rbio(rbio, stripe, pagenr, 1);
+				if (!page)
+					continue;
+			} else {
+			       page = rbio_stripe_page(rbio, stripe, pagenr);
+			}
+
+			ret = rbio_add_io_page(rbio, &bio_list,
+				       page, stripe, pagenr, rbio->stripe_len);
+			if (ret)
+				goto cleanup;
+		}
+	}
+
+	atomic_set(&bbio->stripes_pending, bio_list_size(&bio_list));
+	BUG_ON(atomic_read(&bbio->stripes_pending) == 0);
+
+	while (1) {
+		bio = bio_list_pop(&bio_list);
+		if (!bio)
+			break;
+
+		bio->bi_private = rbio;
+		bio->bi_end_io = raid_write_end_io;
+		BUG_ON(!test_bit(BIO_UPTODATE, &bio->bi_flags));
+		submit_bio(WRITE, bio);
+	}
+	return;
+
+cleanup:
+	rbio_orig_end_io(rbio, -EIO, 0);
+}
+
+/*
+ * helper to find the stripe number for a given bio.  Used to figure out which
+ * stripe has failed.  This expects the bio to correspond to a physical disk,
+ * so it looks up based on physical sector numbers.
+ */
+static int find_bio_stripe(struct btrfs_raid_bio *rbio,
+			   struct bio *bio)
+{
+	u64 physical = bio->bi_sector;
+	u64 stripe_start;
+	int i;
+	struct btrfs_bio_stripe *stripe;
+
+	physical <<= 9;
+
+	for (i = 0; i < rbio->bbio->num_stripes; i++) {
+		stripe = &rbio->bbio->stripes[i];
+		stripe_start = stripe->physical;
+		if (physical >= stripe_start &&
+		    physical < stripe_start + rbio->stripe_len) {
+			return i;
+		}
+	}
+	return -1;
+}
+
+/*
+ * helper to find the stripe number for a given
+ * bio (before mapping).  Used to figure out which stripe has
+ * failed.  This looks up based on logical block numbers.
+ */
+static int find_logical_bio_stripe(struct btrfs_raid_bio *rbio,
+				   struct bio *bio)
+{
+	u64 logical = bio->bi_sector;
+	u64 stripe_start;
+	int i;
+
+	logical <<= 9;
+
+	for (i = 0; i < rbio->nr_data; i++) {
+		stripe_start = rbio->raid_map[i];
+		if (logical >= stripe_start &&
+		    logical < stripe_start + rbio->stripe_len) {
+			return i;
+		}
+	}
+	return -1;
+}
+
+/*
+ * returns -EIO if we had too many failures
+ */
+static int fail_rbio_index(struct btrfs_raid_bio *rbio, int failed)
+{
+	unsigned long flags;
+	int ret = 0;
+
+	spin_lock_irqsave(&rbio->bio_list_lock, flags);
+
+	/* we already know this stripe is bad, move on */
+	if (rbio->faila == failed || rbio->failb == failed)
+		goto out;
+
+	if (rbio->faila == -1) {
+		/* first failure on this rbio */
+		rbio->faila = failed;
+		atomic_inc(&rbio->bbio->error);
+	} else if (rbio->failb == -1) {
+		/* second failure on this rbio */
+		rbio->failb = failed;
+		atomic_inc(&rbio->bbio->error);
+	} else {
+		ret = -EIO;
+	}
+out:
+	spin_unlock_irqrestore(&rbio->bio_list_lock, flags);
+
+	return ret;
+}
+
+/*
+ * helper to fail a stripe based on a physical disk
+ * bio.
+ */
+static int fail_bio_stripe(struct btrfs_raid_bio *rbio,
+			   struct bio *bio)
+{
+	int failed = find_bio_stripe(rbio, bio);
+
+	if (failed < 0)
+		return -EIO;
+
+	return fail_rbio_index(rbio, failed);
+}
+
+/*
+ * this sets each page in the bio uptodate.  It should only be used on private
+ * rbio pages, nothing that comes in from the higher layers
+ */
+static void set_bio_pages_uptodate(struct bio *bio)
+{
+	int i;
+	struct page *p;
+
+	for (i = 0; i < bio->bi_vcnt; i++) {
+		p = bio->bi_io_vec[i].bv_page;
+		SetPageUptodate(p);
+	}
+}
+
+/*
+ * end io for the read phase of the rmw cycle.  All the bios here are physical
+ * stripe bios we've read from the disk so we can recalculate the parity of the
+ * stripe.
+ *
+ * This will usually kick off finish_rmw once all the bios are read in, but it
+ * may trigger parity reconstruction if we had any errors along the way
+ */
+static void raid_rmw_end_io(struct bio *bio, int err)
+{
+	struct btrfs_raid_bio *rbio = bio->bi_private;
+
+	if (err)
+		fail_bio_stripe(rbio, bio);
+	else
+		set_bio_pages_uptodate(bio);
+
+	bio_put(bio);
+
+	if (!atomic_dec_and_test(&rbio->bbio->stripes_pending))
+		return;
+
+	err = 0;
+	if (atomic_read(&rbio->bbio->error) > rbio->bbio->max_errors)
+		goto cleanup;
+
+	/*
+	 * this will normally call finish_rmw to start our write
+	 * but if there are any failed stripes we'll reconstruct
+	 * from parity first
+	 */
+	validate_rbio_for_rmw(rbio);
+	return;
+
+cleanup:
+
+	rbio_orig_end_io(rbio, -EIO, 0);
+}
+
+static void async_rmw_stripe(struct btrfs_raid_bio *rbio)
+{
+	rbio->work.flags = 0;
+	rbio->work.func = rmw_work;
+
+	btrfs_queue_worker(&rbio->fs_info->rmw_workers,
+			   &rbio->work);
+}
+
+static void async_read_rebuild(struct btrfs_raid_bio *rbio)
+{
+	rbio->work.flags = 0;
+	rbio->work.func = read_rebuild_work;
+
+	btrfs_queue_worker(&rbio->fs_info->rmw_workers,
+			   &rbio->work);
+}
+
+/*
+ * the stripe must be locked by the caller.  It will
+ * unlock after all the writes are done
+ */
+static int raid56_rmw_stripe(struct btrfs_raid_bio *rbio)
+{
+	int bios_to_read = 0;
+	struct btrfs_bio *bbio = rbio->bbio;
+	struct bio_list bio_list;
+	int ret;
+	int nr_pages = (rbio->stripe_len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+	int pagenr;
+	int stripe;
+	struct bio *bio;
+
+	bio_list_init(&bio_list);
+
+	ret = alloc_rbio_pages(rbio);
+	if (ret)
+		goto cleanup;
+
+	index_rbio_pages(rbio);
+
+	atomic_set(&rbio->bbio->error, 0);
+	/*
+	 * build a list of bios to read all the missing parts of this
+	 * stripe
+	 */
+	for (stripe = 0; stripe < rbio->nr_data; stripe++) {
+		for (pagenr = 0; pagenr < nr_pages; pagenr++) {
+			struct page *page;
+			/*
+			 * we want to find all the pages missing from
+			 * the rbio and read them from the disk.  If
+			 * page_in_rbio finds a page in the bio list
+			 * we don't need to read it off the stripe.
+			 */
+			page = page_in_rbio(rbio, stripe, pagenr, 1);
+			if (page)
+				continue;
+
+			page = rbio_stripe_page(rbio, stripe, pagenr);
+			ret = rbio_add_io_page(rbio, &bio_list, page,
+				       stripe, pagenr, rbio->stripe_len);
+			if (ret)
+				goto cleanup;
+		}
+	}
+
+	bios_to_read = bio_list_size(&bio_list);
+	if (!bios_to_read) {
+		/*
+		 * this can happen if others have merged with
+		 * us, it means there is nothing left to read.
+		 * But if there are missing devices it may not be
+		 * safe to do the full stripe write yet.
+		 */
+		goto finish;
+	}
+
+	/*
+	 * the bbio may be freed once we submit the last bio.  Make sure
+	 * not to touch it after that
+	 */
+	atomic_set(&bbio->stripes_pending, bios_to_read);
+	while (1) {
+		bio = bio_list_pop(&bio_list);
+		if (!bio)
+			break;
+
+		bio->bi_private = rbio;
+		bio->bi_end_io = raid_rmw_end_io;
+
+		btrfs_bio_wq_end_io(rbio->fs_info, bio,
+				    BTRFS_WQ_ENDIO_RAID56);
+
+		BUG_ON(!test_bit(BIO_UPTODATE, &bio->bi_flags));
+		submit_bio(READ, bio);
+	}
+	/* the actual write will happen once the reads are done */
+	return 0;
+
+cleanup:
+	rbio_orig_end_io(rbio, -EIO, 0);
+	return -EIO;
+
+finish:
+	validate_rbio_for_rmw(rbio);
+	return 0;
+}
+
+/*
+ * if the upper layers pass in a full stripe, we thank them by only allocating
+ * enough pages to hold the parity, and sending it all down quickly.
+ */
+static int full_stripe_write(struct btrfs_raid_bio *rbio)
+{
+	int ret;
+
+	ret = alloc_rbio_parity_pages(rbio);
+	if (ret)
+		return ret;
+
+	ret = lock_stripe_add(rbio);
+	if (ret == 0)
+		finish_rmw(rbio);
+	return 0;
+}
+
+/*
+ * partial stripe writes get handed over to async helpers.
+ * We're really hoping to merge a few more writes into this
+ * rbio before calculating new parity
+ */
+static int partial_stripe_write(struct btrfs_raid_bio *rbio)
+{
+	int ret;
+
+	ret = lock_stripe_add(rbio);
+	if (ret == 0)
+		async_rmw_stripe(rbio);
+	return 0;
+}
+
+/*
+ * sometimes while we were reading from the drive to
+ * recalculate parity, enough new bios come into create
+ * a full stripe.  So we do a check here to see if we can
+ * go directly to finish_rmw
+ */
+static int __raid56_parity_write(struct btrfs_raid_bio *rbio)
+{
+	/* head off into rmw land if we don't have a full stripe */
+	if (!rbio_is_full(rbio))
+		return partial_stripe_write(rbio);
+	return full_stripe_write(rbio);
+}
+
+/*
+ * our main entry point for writes from the rest of the FS.
+ */
+int raid56_parity_write(struct btrfs_root *root, struct bio *bio,
+			struct btrfs_bio *bbio, u64 *raid_map,
+			u64 stripe_len)
+{
+	struct btrfs_raid_bio *rbio;
+
+	rbio = alloc_rbio(root, bbio, raid_map, stripe_len);
+	if (IS_ERR(rbio)) {
+		kfree(raid_map);
+		kfree(bbio);
+		return PTR_ERR(rbio);
+	}
+	bio_list_add(&rbio->bio_list, bio);
+	rbio->bio_list_bytes = bio->bi_size;
+	return __raid56_parity_write(rbio);
+}
+
+/*
+ * all parity reconstruction happens here.  We've read in everything
+ * we can find from the drives and this does the heavy lifting of
+ * sorting the good from the bad.
+ */
+static void __raid_recover_end_io(struct btrfs_raid_bio *rbio)
+{
+	int pagenr, stripe;
+	void **pointers;
+	int faila = -1, failb = -1;
+	int nr_pages = (rbio->stripe_len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+	struct page *page;
+	int err;
+	int i;
+
+	pointers = kzalloc(rbio->bbio->num_stripes * sizeof(void *),
+			   GFP_NOFS);
+	if (!pointers) {
+		err = -ENOMEM;
+		goto cleanup_io;
+	}
+
+	faila = rbio->faila;
+	failb = rbio->failb;
+
+	if (rbio->read_rebuild) {
+		spin_lock_irq(&rbio->bio_list_lock);
+		set_bit(RBIO_RMW_LOCKED_BIT, &rbio->flags);
+		spin_unlock_irq(&rbio->bio_list_lock);
+	}
+
+	index_rbio_pages(rbio);
+
+	for (pagenr = 0; pagenr < nr_pages; pagenr++) {
+		/* setup our array of pointers with pages
+		 * from each stripe
+		 */
+		for (stripe = 0; stripe < rbio->bbio->num_stripes; stripe++) {
+			/*
+			 * if we're rebuilding a read, we have to use
+			 * pages from the bio list
+			 */
+			if (rbio->read_rebuild &&
+			    (stripe == faila || stripe == failb)) {
+				page = page_in_rbio(rbio, stripe, pagenr, 0);
+			} else {
+				page = rbio_stripe_page(rbio, stripe, pagenr);
+			}
+			pointers[stripe] = kmap(page);
+		}
+
+		/* all raid6 handling here */
+		if (rbio->raid_map[rbio->bbio->num_stripes - 1] ==
+		    RAID6_Q_STRIPE) {
+
+			/*
+			 * single failure, rebuild from parity raid5
+			 * style
+			 */
+			if (failb < 0) {
+				if (faila == rbio->nr_data) {
+					/*
+					 * Just the P stripe has failed, without
+					 * a bad data or Q stripe.
+					 * TODO, we should redo the xor here.
+					 */
+					err = -EIO;
+					goto cleanup;
+				}
+				/*
+				 * a single failure in raid6 is rebuilt
+				 * in the pstripe code below
+				 */
+				goto pstripe;
+			}
+
+			/* make sure our ps and qs are in order */
+			if (faila > failb) {
+				int tmp = failb;
+				failb = faila;
+				faila = tmp;
+			}
+
+			/* if the q stripe is failed, do a pstripe reconstruction
+			 * from the xors.
+			 * If both the q stripe and the P stripe are failed, we're
+			 * here due to a crc mismatch and we can't give them the
+			 * data they want
+			 */
+			if (rbio->raid_map[failb] == RAID6_Q_STRIPE) {
+				if (rbio->raid_map[faila] == RAID5_P_STRIPE) {
+					err = -EIO;
+					goto cleanup;
+				}
+				/*
+				 * otherwise we have one bad data stripe and
+				 * a good P stripe.  raid5!
+				 */
+				goto pstripe;
+			}
+
+			if (rbio->raid_map[failb] == RAID5_P_STRIPE) {
+				raid6_datap_recov(rbio->bbio->num_stripes,
+						  PAGE_SIZE, faila, pointers);
+			} else {
+				raid6_2data_recov(rbio->bbio->num_stripes,
+						  PAGE_SIZE, faila, failb,
+						  pointers);
+			}
+		} else {
+			void *p;
+
+			/* rebuild from P stripe here (raid5 or raid6) */
+			BUG_ON(failb != -1);
+pstripe:
+			/* Copy parity block into failed block to start with */
+			memcpy(pointers[faila],
+			       pointers[rbio->nr_data],
+			       PAGE_CACHE_SIZE);
+
+			/* rearrange the pointer array */
+			p = pointers[faila];
+			for (stripe = faila; stripe < rbio->nr_data - 1; stripe++)
+				pointers[stripe] = pointers[stripe + 1];
+			pointers[rbio->nr_data - 1] = p;
+
+			/* xor in the rest */
+			run_xor(pointers, rbio->nr_data - 1, PAGE_CACHE_SIZE);
+		}
+		/* if we're doing this rebuild as part of an rmw, go through
+		 * and set all of our private rbio pages in the
+		 * failed stripes as uptodate.  This way finish_rmw will
+		 * know they can be trusted.  If this was a read reconstruction,
+		 * other endio functions will fiddle the uptodate bits
+		 */
+		if (!rbio->read_rebuild) {
+			for (i = 0;  i < nr_pages; i++) {
+				if (faila != -1) {
+					page = rbio_stripe_page(rbio, faila, i);
+					SetPageUptodate(page);
+				}
+				if (failb != -1) {
+					page = rbio_stripe_page(rbio, failb, i);
+					SetPageUptodate(page);
+				}
+			}
+		}
+		for (stripe = 0; stripe < rbio->bbio->num_stripes; stripe++) {
+			/*
+			 * if we're rebuilding a read, we have to use
+			 * pages from the bio list
+			 */
+			if (rbio->read_rebuild &&
+			    (stripe == faila || stripe == failb)) {
+				page = page_in_rbio(rbio, stripe, pagenr, 0);
+			} else {
+				page = rbio_stripe_page(rbio, stripe, pagenr);
+			}
+			kunmap(page);
+		}
+	}
+
+	err = 0;
+cleanup:
+	kfree(pointers);
+
+cleanup_io:
+
+	if (rbio->read_rebuild) {
+		rbio_orig_end_io(rbio, err, err == 0);
+	} else if (err == 0) {
+		rbio->faila = -1;
+		rbio->failb = -1;
+		finish_rmw(rbio);
+	} else {
+		rbio_orig_end_io(rbio, err, 0);
+	}
+}
+
+/*
+ * This is called only for stripes we've read from disk to
+ * reconstruct the parity.
+ */
+static void raid_recover_end_io(struct bio *bio, int err)
+{
+	struct btrfs_raid_bio *rbio = bio->bi_private;
+
+	/*
+	 * we only read stripe pages off the disk, set them
+	 * up to date if there were no errors
+	 */
+	if (err)
+		fail_bio_stripe(rbio, bio);
+	else
+		set_bio_pages_uptodate(bio);
+	bio_put(bio);
+
+	if (!atomic_dec_and_test(&rbio->bbio->stripes_pending))
+		return;
+
+	if (atomic_read(&rbio->bbio->error) > rbio->bbio->max_errors)
+		rbio_orig_end_io(rbio, -EIO, 0);
+	else
+		__raid_recover_end_io(rbio);
+}
+
+/*
+ * reads everything we need off the disk to reconstruct
+ * the parity. endio handlers trigger final reconstruction
+ * when the IO is done.
+ *
+ * This is used both for reads from the higher layers and for
+ * parity construction required to finish a rmw cycle.
+ */
+static int __raid56_parity_recover(struct btrfs_raid_bio *rbio)
+{
+	int bios_to_read = 0;
+	struct btrfs_bio *bbio = rbio->bbio;
+	struct bio_list bio_list;
+	int ret;
+	int nr_pages = (rbio->stripe_len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+	int pagenr;
+	int stripe;
+	struct bio *bio;
+
+	bio_list_init(&bio_list);
+
+	ret = alloc_rbio_pages(rbio);
+	if (ret)
+		goto cleanup;
+
+	atomic_set(&rbio->bbio->error, 0);
+
+	/*
+	 * read everything that hasn't failed.
+	 */
+	for (stripe = 0; stripe < bbio->num_stripes; stripe++) {
+		if (rbio->faila == stripe ||
+		    rbio->failb == stripe)
+			continue;
+
+		for (pagenr = 0; pagenr < nr_pages; pagenr++) {
+			struct page *p;
+
+			/*
+			 * the rmw code may have already read this
+			 * page in
+			 */
+			p = rbio_stripe_page(rbio, stripe, pagenr);
+			if (PageUptodate(p))
+				continue;
+
+			ret = rbio_add_io_page(rbio, &bio_list,
+				       rbio_stripe_page(rbio, stripe, pagenr),
+				       stripe, pagenr, rbio->stripe_len);
+			if (ret < 0)
+				goto cleanup;
+		}
+	}
+
+	bios_to_read = bio_list_size(&bio_list);
+	if (!bios_to_read) {
+		/*
+		 * we might have no bios to read just because the pages
+		 * were up to date, or we might have no bios to read because
+		 * the devices were gone.
+		 */
+		if (atomic_read(&rbio->bbio->error) <= rbio->bbio->max_errors) {
+			__raid_recover_end_io(rbio);
+			goto out;
+		} else {
+			goto cleanup;
+		}
+	}
+
+	/*
+	 * the bbio may be freed once we submit the last bio.  Make sure
+	 * not to touch it after that
+	 */
+	atomic_set(&bbio->stripes_pending, bios_to_read);
+	while (1) {
+		bio = bio_list_pop(&bio_list);
+		if (!bio)
+			break;
+
+		bio->bi_private = rbio;
+		bio->bi_end_io = raid_recover_end_io;
+
+		btrfs_bio_wq_end_io(rbio->fs_info, bio,
+				    BTRFS_WQ_ENDIO_RAID56);
+
+		BUG_ON(!test_bit(BIO_UPTODATE, &bio->bi_flags));
+		submit_bio(READ, bio);
+	}
+out:
+	return 0;
+
+cleanup:
+	if (rbio->read_rebuild)
+		rbio_orig_end_io(rbio, -EIO, 0);
+	return -EIO;
+}
+
+/*
+ * the main entry point for reads from the higher layers.  This
+ * is really only called when the normal read path had a failure,
+ * so we assume the bio they send down corresponds to a failed part
+ * of the drive.
+ */
+int raid56_parity_recover(struct btrfs_root *root, struct bio *bio,
+			  struct btrfs_bio *bbio, u64 *raid_map,
+			  u64 stripe_len, int mirror_num)
+{
+	struct btrfs_raid_bio *rbio;
+	int ret;
+
+	rbio = alloc_rbio(root, bbio, raid_map, stripe_len);
+	if (IS_ERR(rbio)) {
+		return PTR_ERR(rbio);
+	}
+
+	rbio->read_rebuild = 1;
+	bio_list_add(&rbio->bio_list, bio);
+	rbio->bio_list_bytes = bio->bi_size;
+
+	rbio->faila = find_logical_bio_stripe(rbio, bio);
+	if (rbio->faila == -1) {
+		BUG();
+		kfree(rbio);
+		return -EIO;
+	}
+
+	/*
+	 * reconstruct from the q stripe if they are
+	 * asking for mirror 3
+	 */
+	if (mirror_num == 3)
+		rbio->failb = bbio->num_stripes - 2;
+
+	ret = lock_stripe_add(rbio);
+
+	/*
+	 * __raid56_parity_recover will end the bio with
+	 * any errors it hits.  We don't want to return
+	 * its error value up the stack because our caller
+	 * will end up calling bio_endio with any nonzero
+	 * return
+	 */
+	if (ret == 0)
+		__raid56_parity_recover(rbio);
+	/*
+	 * our rbio has been added to the list of
+	 * rbios that will be handled after the
+	 * currently lock owner is done
+	 */
+	return 0;
+
+}
+
+static void rmw_work(struct btrfs_work *work)
+{
+	struct btrfs_raid_bio *rbio;
+
+	rbio = container_of(work, struct btrfs_raid_bio, work);
+	raid56_rmw_stripe(rbio);
+}
+
+static void read_rebuild_work(struct btrfs_work *work)
+{
+	struct btrfs_raid_bio *rbio;
+
+	rbio = container_of(work, struct btrfs_raid_bio, work);
+	__raid56_parity_recover(rbio);
+}