X-Git-Url: http://pilppa.org/gitweb/gitweb.cgi?a=blobdiff_plain;f=fs%2Fbtrfs%2Fdisk-io.c;h=b9301a5e4608506b2f6701eabca7d847a67bffb1;hb=04160088141a6be06d5cb9e53e34076b51803ffc;hp=f2c1dd40c087871ba435b49ff74d776edfa819b5;hpb=30ae8467483d7ab023b5e728bf7d74a575c78023;p=linux-2.6-omap-h63xx.git

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index f2c1dd40c08..6a70e4916ab 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1,166 +1,325 @@
-#include <linux/module.h>
+/*
+ * Copyright (C) 2007 Oracle.  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/fs.h>
 #include <linux/blkdev.h>
-#include <linux/crypto.h>
+#include <linux/crc32c.h>
 #include <linux/scatterlist.h>
+#include <linux/swap.h>
+#include <linux/radix-tree.h>
+#include <linux/writeback.h>
+#include <linux/buffer_head.h> // for block_sync_page
 #include "ctree.h"
 #include "disk-io.h"
 #include "transaction.h"
+#include "btrfs_inode.h"
+#include "volumes.h"
+#include "print-tree.h"
 
-
-static int check_tree_block(struct btrfs_root *root, struct buffer_head *buf)
+#if 0
+static int check_tree_block(struct btrfs_root *root, struct extent_buffer *buf)
 {
-	struct btrfs_node *node = btrfs_buffer_node(buf);
-	if (buf->b_blocknr != btrfs_header_blocknr(&node->header)) {
-		BUG();
-	}
-	if (root->node && btrfs_header_parentid(&node->header) !=
-	    btrfs_header_parentid(btrfs_buffer_header(root->node))) {
-		BUG();
+	if (extent_buffer_blocknr(buf) != btrfs_header_blocknr(buf)) {
+		printk(KERN_CRIT "buf blocknr(buf) is %llu, header is %llu\n",
+		       (unsigned long long)extent_buffer_blocknr(buf),
+		       (unsigned long long)btrfs_header_blocknr(buf));
+		return 1;
 	}
 	return 0;
 }
+#endif
 
-struct buffer_head *btrfs_find_tree_block(struct btrfs_root *root, u64 blocknr)
+static struct extent_io_ops btree_extent_io_ops;
+
+struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
+					    u64 bytenr, u32 blocksize)
 {
-	struct address_space *mapping = root->fs_info->btree_inode->i_mapping;
-	int blockbits = root->fs_info->sb->s_blocksize_bits;
-	unsigned long index = blocknr >> (PAGE_CACHE_SHIFT - blockbits);
-	struct page *page;
-	struct buffer_head *bh;
-	struct buffer_head *head;
-	struct buffer_head *ret = NULL;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	struct extent_buffer *eb;
+	eb = find_extent_buffer(&BTRFS_I(btree_inode)->io_tree,
+				bytenr, blocksize, GFP_NOFS);
+	return eb;
+}
 
-	page = find_lock_page(mapping, index);
-	if (!page)
-		return NULL;
+struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
+						 u64 bytenr, u32 blocksize)
+{
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	struct extent_buffer *eb;
 
-	if (!page_has_buffers(page))
-		goto out_unlock;
+	eb = alloc_extent_buffer(&BTRFS_I(btree_inode)->io_tree,
+				 bytenr, blocksize, NULL, GFP_NOFS);
+	return eb;
+}
 
-	head = page_buffers(page);
-	bh = head;
-	do {
-		if (buffer_mapped(bh) && bh->b_blocknr == blocknr) {
-			ret = bh;
-			get_bh(bh);
-			goto out_unlock;
-		}
-		bh = bh->b_this_page;
-	} while (bh != head);
-out_unlock:
-	unlock_page(page);
-	page_cache_release(page);
-	return ret;
+struct extent_map *btree_get_extent(struct inode *inode, struct page *page,
+				    size_t page_offset, u64 start, u64 len,
+				    int create)
+{
+	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	struct extent_map *em;
+	int ret;
+
+again:
+	spin_lock(&em_tree->lock);
+	em = lookup_extent_mapping(em_tree, start, len);
+	spin_unlock(&em_tree->lock);
+	if (em) {
+		goto out;
+	}
+	em = alloc_extent_map(GFP_NOFS);
+	if (!em) {
+		em = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+	em->start = 0;
+	em->len = i_size_read(inode);
+	em->block_start = 0;
+	em->bdev = inode->i_sb->s_bdev;
+
+	spin_lock(&em_tree->lock);
+	ret = add_extent_mapping(em_tree, em);
+	spin_unlock(&em_tree->lock);
+
+	if (ret == -EEXIST) {
+		free_extent_map(em);
+		em = NULL;
+		goto again;
+	} else if (ret) {
+		em = ERR_PTR(ret);
+	}
+out:
+	return em;
 }
 
-struct buffer_head *btrfs_find_create_tree_block(struct btrfs_root *root,
-						 u64 blocknr)
+u32 btrfs_csum_data(struct btrfs_root *root, char *data, u32 seed, size_t len)
 {
-	struct address_space *mapping = root->fs_info->btree_inode->i_mapping;
-	int blockbits = root->fs_info->sb->s_blocksize_bits;
-	unsigned long index = blocknr >> (PAGE_CACHE_SHIFT - blockbits);
-	struct page *page;
-	struct buffer_head *bh;
-	struct buffer_head *head;
-	struct buffer_head *ret = NULL;
-	u64 first_block = index << (PAGE_CACHE_SHIFT - blockbits);
-	page = grab_cache_page(mapping, index);
-	if (!page)
-		return NULL;
+	return crc32c(seed, data, len);
+}
 
-	if (!page_has_buffers(page))
-		create_empty_buffers(page, root->fs_info->sb->s_blocksize, 0);
-	head = page_buffers(page);
-	bh = head;
-	do {
-		if (!buffer_mapped(bh)) {
-			bh->b_bdev = root->fs_info->sb->s_bdev;
-			bh->b_blocknr = first_block;
-			set_buffer_mapped(bh);
+void btrfs_csum_final(u32 crc, char *result)
+{
+	*(__le32 *)result = ~cpu_to_le32(crc);
+}
+
+static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf,
+			   int verify)
+{
+	char result[BTRFS_CRC32_SIZE];
+	unsigned long len;
+	unsigned long cur_len;
+	unsigned long offset = BTRFS_CSUM_SIZE;
+	char *map_token = NULL;
+	char *kaddr;
+	unsigned long map_start;
+	unsigned long map_len;
+	int err;
+	u32 crc = ~(u32)0;
+
+	len = buf->len - offset;
+	while(len > 0) {
+		err = map_private_extent_buffer(buf, offset, 32,
+					&map_token, &kaddr,
+					&map_start, &map_len, KM_USER0);
+		if (err) {
+			printk("failed to map extent buffer! %lu\n",
+			       offset);
+			return 1;
 		}
-		if (bh->b_blocknr == blocknr) {
-			ret = bh;
-			get_bh(bh);
-			goto out_unlock;
+		cur_len = min(len, map_len - (offset - map_start));
+		crc = btrfs_csum_data(root, kaddr + offset - map_start,
+				      crc, cur_len);
+		len -= cur_len;
+		offset += cur_len;
+		unmap_extent_buffer(buf, map_token, KM_USER0);
+	}
+	btrfs_csum_final(crc, result);
+
+	if (verify) {
+		int from_this_trans = 0;
+
+		if (root->fs_info->running_transaction &&
+		    btrfs_header_generation(buf) ==
+		    root->fs_info->running_transaction->transid)
+			from_this_trans = 1;
+
+		/* FIXME, this is not good */
+		if (from_this_trans == 0 &&
+		    memcmp_extent_buffer(buf, result, 0, BTRFS_CRC32_SIZE)) {
+			u32 val;
+			u32 found = 0;
+			memcpy(&found, result, BTRFS_CRC32_SIZE);
+
+			read_extent_buffer(buf, &val, 0, BTRFS_CRC32_SIZE);
+			printk("btrfs: %s checksum verify failed on %llu "
+			       "wanted %X found %X from_this_trans %d\n",
+			       root->fs_info->sb->s_id,
+			       buf->start, val, found, from_this_trans);
+			return 1;
 		}
-		bh = bh->b_this_page;
-		first_block++;
-	} while (bh != head);
-out_unlock:
-	unlock_page(page);
-	page_cache_release(page);
-	return ret;
+	} else {
+		write_extent_buffer(buf, result, 0, BTRFS_CRC32_SIZE);
+	}
+	return 0;
 }
 
-static sector_t max_block(struct block_device *bdev)
+
+int csum_dirty_buffer(struct btrfs_root *root, struct page *page)
 {
-	sector_t retval = ~((sector_t)0);
-	loff_t sz = i_size_read(bdev->bd_inode);
+	struct extent_io_tree *tree;
+	u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
+	u64 found_start;
+	int found_level;
+	unsigned long len;
+	struct extent_buffer *eb;
+	tree = &BTRFS_I(page->mapping->host)->io_tree;
 
-	if (sz) {
-		unsigned int size = block_size(bdev);
-		unsigned int sizebits = blksize_bits(size);
-		retval = (sz >> sizebits);
+	if (page->private == EXTENT_PAGE_PRIVATE)
+		goto out;
+	if (!page->private)
+		goto out;
+	len = page->private >> 2;
+	if (len == 0) {
+		WARN_ON(1);
+	}
+	eb = alloc_extent_buffer(tree, start, len, page, GFP_NOFS);
+	read_extent_buffer_pages(tree, eb, start + PAGE_CACHE_SIZE, 1,
+				 btree_get_extent);
+	btrfs_clear_buffer_defrag(eb);
+	found_start = btrfs_header_bytenr(eb);
+	if (found_start != start) {
+		printk("warning: eb start incorrect %Lu buffer %Lu len %lu\n",
+		       start, found_start, len);
+		WARN_ON(1);
+		goto err;
+	}
+	if (eb->first_page != page) {
+		printk("bad first page %lu %lu\n", eb->first_page->index,
+		       page->index);
+		WARN_ON(1);
+		goto err;
 	}
-	return retval;
+	if (!PageUptodate(page)) {
+		printk("csum not up to date page %lu\n", page->index);
+		WARN_ON(1);
+		goto err;
+	}
+	found_level = btrfs_header_level(eb);
+	csum_tree_block(root, eb, 0);
+err:
+	free_extent_buffer(eb);
+out:
+	return 0;
 }
 
-static int btree_get_block(struct inode *inode, sector_t iblock,
-			   struct buffer_head *bh, int create)
+static int btree_writepage_io_hook(struct page *page, u64 start, u64 end)
 {
-	if (iblock >= max_block(inode->i_sb->s_bdev)) {
-		if (create)
-			return -EIO;
+	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
+
+	csum_dirty_buffer(root, page);
+	return 0;
+}
 
-		/*
-		 * for reads, we're just trying to fill a partial page.
-		 * return a hole, they will have to call get_block again
-		 * before they can fill it, and they will get -EIO at that
-		 * time
-		 */
+static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	u64 offset;
+	offset = bio->bi_sector << 9;
+	if (offset == BTRFS_SUPER_INFO_OFFSET) {
+		bio->bi_bdev = root->fs_info->sb->s_bdev;
+		submit_bio(rw, bio);
 		return 0;
 	}
-	bh->b_bdev = inode->i_sb->s_bdev;
-	bh->b_blocknr = iblock;
-	set_buffer_mapped(bh);
-	return 0;
+	return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio);
 }
 
-static int csum_tree_block(struct btrfs_root * root, struct buffer_head *bh,
-			    int verify)
+static int btree_writepage(struct page *page, struct writeback_control *wbc)
 {
-	struct btrfs_node *node = btrfs_buffer_node(bh);
-	struct scatterlist sg;
-	struct crypto_hash *tfm = root->fs_info->hash_tfm;
-	struct hash_desc desc;
+	struct extent_io_tree *tree;
+	tree = &BTRFS_I(page->mapping->host)->io_tree;
+	return extent_write_full_page(tree, page, btree_get_extent, wbc);
+}
+
+static int btree_writepages(struct address_space *mapping,
+			    struct writeback_control *wbc)
+{
+	struct extent_io_tree *tree;
+	tree = &BTRFS_I(mapping->host)->io_tree;
+	if (wbc->sync_mode == WB_SYNC_NONE) {
+		u64 num_dirty;
+		u64 start = 0;
+		unsigned long thresh = 96 * 1024 * 1024;
+
+		if (wbc->for_kupdate)
+			return 0;
+
+		if (current_is_pdflush()) {
+			thresh = 96 * 1024 * 1024;
+		} else {
+			thresh = 8 * 1024 * 1024;
+		}
+		num_dirty = count_range_bits(tree, &start, (u64)-1,
+					     thresh, EXTENT_DIRTY);
+		if (num_dirty < thresh) {
+			return 0;
+		}
+	}
+	return extent_writepages(tree, mapping, btree_get_extent, wbc);
+}
+
+int btree_readpage(struct file *file, struct page *page)
+{
+	struct extent_io_tree *tree;
+	tree = &BTRFS_I(page->mapping->host)->io_tree;
+	return extent_read_full_page(tree, page, btree_get_extent);
+}
+
+static int btree_releasepage(struct page *page, gfp_t gfp_flags)
+{
+	struct extent_io_tree *tree;
+	struct extent_map_tree *map;
 	int ret;
-	char result[32];
-
-	desc.tfm = tfm;
-	desc.flags = 0;
-	sg_init_one(&sg, bh->b_data + 32, bh->b_size - 32);
-	spin_lock(&root->fs_info->hash_lock);
-	ret = crypto_hash_digest(&desc, &sg, bh->b_size - 32, result);
-	spin_unlock(&root->fs_info->hash_lock);
-	if (ret) {
-		printk("sha256 digest failed\n");
+
+	tree = &BTRFS_I(page->mapping->host)->io_tree;
+	map = &BTRFS_I(page->mapping->host)->extent_tree;
+	ret = try_release_extent_mapping(map, tree, page, gfp_flags);
+	if (ret == 1) {
+		ClearPagePrivate(page);
+		set_page_private(page, 0);
+		page_cache_release(page);
 	}
-	if (verify) {
-		if (memcmp(node->header.csum, result, sizeof(result)))
-			printk("csum verify failed on %Lu\n", bh->b_blocknr);
-		return -EINVAL;
-	} else
-		memcpy(node->header.csum, result, sizeof(node->header.csum));
-	return 0;
+	return ret;
 }
 
+static void btree_invalidatepage(struct page *page, unsigned long offset)
+{
+	struct extent_io_tree *tree;
+	tree = &BTRFS_I(page->mapping->host)->io_tree;
+	extent_invalidatepage(tree, page, offset);
+	btree_releasepage(page, GFP_NOFS);
+}
+
+#if 0
 static int btree_writepage(struct page *page, struct writeback_control *wbc)
 {
 	struct buffer_head *bh;
-	struct btrfs_root *root = btrfs_sb(page->mapping->host->i_sb);
+	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
 	struct buffer_head *head;
-
 	if (!page_has_buffers(page)) {
 		create_empty_buffers(page, root->fs_info->sb->s_blocksize,
 					(1 << BH_Dirty)|(1 << BH_Uptodate));
@@ -174,194 +333,648 @@ static int btree_writepage(struct page *page, struct writeback_control *wbc)
 	} while (bh != head);
 	return block_write_full_page(page, btree_get_block, wbc);
 }
-
-static int btree_readpage(struct file * file, struct page * page)
-{
-	return block_read_full_page(page, btree_get_block);
-}
+#endif
 
 static struct address_space_operations btree_aops = {
 	.readpage	= btree_readpage,
 	.writepage	= btree_writepage,
+	.writepages	= btree_writepages,
+	.releasepage	= btree_releasepage,
+	.invalidatepage = btree_invalidatepage,
 	.sync_page	= block_sync_page,
 };
 
-struct buffer_head *read_tree_block(struct btrfs_root *root, u64 blocknr)
-{
-	struct buffer_head *bh = NULL;
-
-	bh = btrfs_find_create_tree_block(root, blocknr);
-	if (!bh)
-		return bh;
-	lock_buffer(bh);
-	if (!buffer_uptodate(bh)) {
-		get_bh(bh);
-		bh->b_end_io = end_buffer_read_sync;
-		submit_bh(READ, bh);
-		wait_on_buffer(bh);
-		if (!buffer_uptodate(bh))
-			goto fail;
-		csum_tree_block(root, bh, 1);
-	} else {
-		unlock_buffer(bh);
-	}
-	if (check_tree_block(root, bh))
-		BUG();
-	return bh;
-fail:
-	brelse(bh);
-	return NULL;
+int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize)
+{
+	struct extent_buffer *buf = NULL;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	int ret = 0;
 
+	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
+	if (!buf)
+		return 0;
+	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
+				 buf, 0, 0, btree_get_extent);
+	free_extent_buffer(buf);
+	return ret;
 }
 
-int dirty_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
-		     struct buffer_head *buf)
+static int close_all_devices(struct btrfs_fs_info *fs_info)
 {
-	mark_buffer_dirty(buf);
+	struct list_head *list;
+	struct list_head *next;
+	struct btrfs_device *device;
+
+	list = &fs_info->fs_devices->devices;
+	list_for_each(next, list) {
+		device = list_entry(next, struct btrfs_device, dev_list);
+		if (device->bdev && device->bdev != fs_info->sb->s_bdev)
+			close_bdev_excl(device->bdev);
+		device->bdev = NULL;
+	}
 	return 0;
 }
 
+struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
+				      u32 blocksize)
+{
+	struct extent_buffer *buf = NULL;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	struct extent_io_tree *io_tree;
+	u64 end;
+	int ret;
+
+	io_tree = &BTRFS_I(btree_inode)->io_tree;
+
+	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
+	if (!buf)
+		return NULL;
+	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree, buf, 0, 1,
+				 btree_get_extent);
+
+	if (buf->flags & EXTENT_CSUM)
+		return buf;
+
+	end = buf->start + PAGE_CACHE_SIZE - 1;
+	if (test_range_bit(io_tree, buf->start, end, EXTENT_CSUM, 1)) {
+		buf->flags |= EXTENT_CSUM;
+		return buf;
+	}
+
+	lock_extent(io_tree, buf->start, end, GFP_NOFS);
+
+	if (test_range_bit(io_tree, buf->start, end, EXTENT_CSUM, 1)) {
+		buf->flags |= EXTENT_CSUM;
+		goto out_unlock;
+	}
+
+	ret = csum_tree_block(root, buf, 1);
+	set_extent_bits(io_tree, buf->start, end, EXTENT_CSUM, GFP_NOFS);
+	buf->flags |= EXTENT_CSUM;
+
+out_unlock:
+	unlock_extent(io_tree, buf->start, end, GFP_NOFS);
+	return buf;
+}
+
 int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
-		     struct buffer_head *buf)
+		     struct extent_buffer *buf)
+{
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	if (btrfs_header_generation(buf) ==
+	    root->fs_info->running_transaction->transid)
+		clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree,
+					  buf);
+	return 0;
+}
+
+int wait_on_tree_block_writeback(struct btrfs_root *root,
+				 struct extent_buffer *buf)
 {
-	clear_buffer_dirty(buf);
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	wait_on_extent_buffer_writeback(&BTRFS_I(btree_inode)->io_tree,
+					buf);
 	return 0;
 }
 
-static int __setup_root(struct btrfs_super_block *super,
-			struct btrfs_root *root,
+static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
+			u32 stripesize, struct btrfs_root *root,
 			struct btrfs_fs_info *fs_info,
 			u64 objectid)
 {
 	root->node = NULL;
+	root->inode = NULL;
 	root->commit_root = NULL;
-	root->blocksize = btrfs_super_blocksize(super);
+	root->sectorsize = sectorsize;
+	root->nodesize = nodesize;
+	root->leafsize = leafsize;
+	root->stripesize = stripesize;
 	root->ref_cows = 0;
+	root->track_dirty = 0;
+
 	root->fs_info = fs_info;
+	root->objectid = objectid;
+	root->last_trans = 0;
+	root->highest_inode = 0;
+	root->last_inode_alloc = 0;
+	root->name = NULL;
+	root->in_sysfs = 0;
+
+	INIT_LIST_HEAD(&root->dirty_list);
 	memset(&root->root_key, 0, sizeof(root->root_key));
 	memset(&root->root_item, 0, sizeof(root->root_item));
+	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
+	memset(&root->root_kobj, 0, sizeof(root->root_kobj));
+	init_completion(&root->kobj_unregister);
+	root->defrag_running = 0;
+	root->defrag_level = 0;
+	root->root_key.objectid = objectid;
 	return 0;
 }
 
-static int find_and_setup_root(struct btrfs_super_block *super,
-			       struct btrfs_root *tree_root,
+static int find_and_setup_root(struct btrfs_root *tree_root,
 			       struct btrfs_fs_info *fs_info,
 			       u64 objectid,
 			       struct btrfs_root *root)
 {
 	int ret;
+	u32 blocksize;
 
-	__setup_root(super, root, fs_info, objectid);
+	__setup_root(tree_root->nodesize, tree_root->leafsize,
+		     tree_root->sectorsize, tree_root->stripesize,
+		     root, fs_info, objectid);
 	ret = btrfs_find_last_root(tree_root, objectid,
 				   &root->root_item, &root->root_key);
 	BUG_ON(ret);
 
-	root->node = read_tree_block(root,
-				     btrfs_root_blocknr(&root->root_item));
+	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
+	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
+				     blocksize);
+	BUG_ON(!root->node);
+	return 0;
+}
+
+struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_fs_info *fs_info,
+					       struct btrfs_key *location)
+{
+	struct btrfs_root *root;
+	struct btrfs_root *tree_root = fs_info->tree_root;
+	struct btrfs_path *path;
+	struct extent_buffer *l;
+	u64 highest_inode;
+	u32 blocksize;
+	int ret = 0;
+
+	root = kzalloc(sizeof(*root), GFP_NOFS);
+	if (!root)
+		return ERR_PTR(-ENOMEM);
+	if (location->offset == (u64)-1) {
+		ret = find_and_setup_root(tree_root, fs_info,
+					  location->objectid, root);
+		if (ret) {
+			kfree(root);
+			return ERR_PTR(ret);
+		}
+		goto insert;
+	}
+
+	__setup_root(tree_root->nodesize, tree_root->leafsize,
+		     tree_root->sectorsize, tree_root->stripesize,
+		     root, fs_info, location->objectid);
+
+	path = btrfs_alloc_path();
+	BUG_ON(!path);
+	ret = btrfs_search_slot(NULL, tree_root, location, path, 0, 0);
+	if (ret != 0) {
+		if (ret > 0)
+			ret = -ENOENT;
+		goto out;
+	}
+	l = path->nodes[0];
+	read_extent_buffer(l, &root->root_item,
+	       btrfs_item_ptr_offset(l, path->slots[0]),
+	       sizeof(root->root_item));
+	memcpy(&root->root_key, location, sizeof(*location));
+	ret = 0;
+out:
+	btrfs_release_path(root, path);
+	btrfs_free_path(path);
+	if (ret) {
+		kfree(root);
+		return ERR_PTR(ret);
+	}
+	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
+	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
+				     blocksize);
 	BUG_ON(!root->node);
+insert:
+	root->ref_cows = 1;
+	ret = btrfs_find_highest_inode(root, &highest_inode);
+	if (ret == 0) {
+		root->highest_inode = highest_inode;
+		root->last_inode_alloc = highest_inode;
+	}
+	return root;
+}
+
+struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
+					u64 root_objectid)
+{
+	struct btrfs_root *root;
+
+	if (root_objectid == BTRFS_ROOT_TREE_OBJECTID)
+		return fs_info->tree_root;
+	if (root_objectid == BTRFS_EXTENT_TREE_OBJECTID)
+		return fs_info->extent_root;
+
+	root = radix_tree_lookup(&fs_info->fs_roots_radix,
+				 (unsigned long)root_objectid);
+	return root;
+}
+
+struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
+					      struct btrfs_key *location)
+{
+	struct btrfs_root *root;
+	int ret;
+
+	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
+		return fs_info->tree_root;
+	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
+		return fs_info->extent_root;
+
+	root = radix_tree_lookup(&fs_info->fs_roots_radix,
+				 (unsigned long)location->objectid);
+	if (root)
+		return root;
+
+	root = btrfs_read_fs_root_no_radix(fs_info, location);
+	if (IS_ERR(root))
+		return root;
+	ret = radix_tree_insert(&fs_info->fs_roots_radix,
+				(unsigned long)root->root_key.objectid,
+				root);
+	if (ret) {
+		free_extent_buffer(root->node);
+		kfree(root);
+		return ERR_PTR(ret);
+	}
+	ret = btrfs_find_dead_roots(fs_info->tree_root,
+				    root->root_key.objectid, root);
+	BUG_ON(ret);
+
+	return root;
+}
+
+struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info,
+				      struct btrfs_key *location,
+				      const char *name, int namelen)
+{
+	struct btrfs_root *root;
+	int ret;
+
+	root = btrfs_read_fs_root_no_name(fs_info, location);
+	if (!root)
+		return NULL;
+
+	if (root->in_sysfs)
+		return root;
+
+	ret = btrfs_set_root_name(root, name, namelen);
+	if (ret) {
+		free_extent_buffer(root->node);
+		kfree(root);
+		return ERR_PTR(ret);
+	}
+
+	ret = btrfs_sysfs_add_root(root);
+	if (ret) {
+		free_extent_buffer(root->node);
+		kfree(root->name);
+		kfree(root);
+		return ERR_PTR(ret);
+	}
+	root->in_sysfs = 1;
+	return root;
+}
+#if 0
+static int add_hasher(struct btrfs_fs_info *info, char *type) {
+	struct btrfs_hasher *hasher;
+
+	hasher = kmalloc(sizeof(*hasher), GFP_NOFS);
+	if (!hasher)
+		return -ENOMEM;
+	hasher->hash_tfm = crypto_alloc_hash(type, 0, CRYPTO_ALG_ASYNC);
+	if (!hasher->hash_tfm) {
+		kfree(hasher);
+		return -EINVAL;
+	}
+	spin_lock(&info->hash_lock);
+	list_add(&hasher->list, &info->hashers);
+	spin_unlock(&info->hash_lock);
+	return 0;
+}
+#endif
+
+static int btrfs_congested_fn(void *congested_data, int bdi_bits)
+{
+	struct btrfs_fs_info *info = (struct btrfs_fs_info *)congested_data;
+	int ret = 0;
+	struct list_head *cur;
+	struct btrfs_device *device;
+	struct backing_dev_info *bdi;
+
+	list_for_each(cur, &info->fs_devices->devices) {
+		device = list_entry(cur, struct btrfs_device, dev_list);
+		bdi = blk_get_backing_dev_info(device->bdev);
+		if (bdi && bdi_congested(bdi, bdi_bits)) {
+			ret = 1;
+			break;
+		}
+	}
+	return ret;
+}
+
+void btrfs_unplug_io_fn(struct backing_dev_info *bdi, struct page *page)
+{
+	struct list_head *cur;
+	struct btrfs_device *device;
+	struct btrfs_fs_info *info;
+
+	info = (struct btrfs_fs_info *)bdi->unplug_io_data;
+	list_for_each(cur, &info->fs_devices->devices) {
+		device = list_entry(cur, struct btrfs_device, dev_list);
+		bdi = blk_get_backing_dev_info(device->bdev);
+		if (bdi->unplug_io_fn) {
+			bdi->unplug_io_fn(bdi, page);
+		}
+	}
+}
+
+static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
+{
+	bdi_init(bdi);
+	bdi->ra_pages	= default_backing_dev_info.ra_pages;
+	bdi->state		= 0;
+	bdi->capabilities	= default_backing_dev_info.capabilities;
+	bdi->unplug_io_fn	= btrfs_unplug_io_fn;
+	bdi->unplug_io_data	= info;
+	bdi->congested_fn	= btrfs_congested_fn;
+	bdi->congested_data	= info;
 	return 0;
 }
 
 struct btrfs_root *open_ctree(struct super_block *sb,
-			      struct buffer_head *sb_buffer,
-			      struct btrfs_super_block *disk_super)
+			      struct btrfs_fs_devices *fs_devices)
 {
-	struct btrfs_root *root = kmalloc(sizeof(struct btrfs_root),
-					  GFP_NOFS);
+	u32 sectorsize;
+	u32 nodesize;
+	u32 leafsize;
+	u32 blocksize;
+	u32 stripesize;
 	struct btrfs_root *extent_root = kmalloc(sizeof(struct btrfs_root),
 						 GFP_NOFS);
 	struct btrfs_root *tree_root = kmalloc(sizeof(struct btrfs_root),
 					       GFP_NOFS);
-	struct btrfs_root *inode_root = kmalloc(sizeof(struct btrfs_root),
-						GFP_NOFS);
 	struct btrfs_fs_info *fs_info = kmalloc(sizeof(*fs_info),
 						GFP_NOFS);
+	struct btrfs_root *chunk_root = kmalloc(sizeof(struct btrfs_root),
+						GFP_NOFS);
+	struct btrfs_root *dev_root = kmalloc(sizeof(struct btrfs_root),
+					      GFP_NOFS);
 	int ret;
+	int err = -EIO;
+	struct btrfs_super_block *disk_super;
 
-	if (!btrfs_super_root(disk_super)) {
-		return NULL;
+	if (!extent_root || !tree_root || !fs_info) {
+		err = -ENOMEM;
+		goto fail;
 	}
-	init_bit_radix(&fs_info->pinned_radix);
-	init_bit_radix(&fs_info->pending_del_radix);
-	sb_set_blocksize(sb, sb_buffer->b_size);
+	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_NOFS);
+	INIT_LIST_HEAD(&fs_info->trans_list);
+	INIT_LIST_HEAD(&fs_info->dead_roots);
+	INIT_LIST_HEAD(&fs_info->hashers);
+	spin_lock_init(&fs_info->hash_lock);
+	spin_lock_init(&fs_info->delalloc_lock);
+	spin_lock_init(&fs_info->new_trans_lock);
+
+	memset(&fs_info->super_kobj, 0, sizeof(fs_info->super_kobj));
+	init_completion(&fs_info->kobj_unregister);
+	sb_set_blocksize(sb, 4096);
 	fs_info->running_transaction = NULL;
-	fs_info->fs_root = root;
+	fs_info->last_trans_committed = 0;
 	fs_info->tree_root = tree_root;
 	fs_info->extent_root = extent_root;
-	fs_info->inode_root = inode_root;
-	fs_info->last_inode_alloc = 0;
-	fs_info->last_inode_alloc_dirid = 0;
-	fs_info->disk_super = disk_super;
+	fs_info->chunk_root = chunk_root;
+	fs_info->dev_root = dev_root;
+	fs_info->fs_devices = fs_devices;
+	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
+	INIT_LIST_HEAD(&fs_info->space_info);
+	btrfs_mapping_init(&fs_info->mapping_tree);
 	fs_info->sb = sb;
+	fs_info->throttles = 0;
+	fs_info->mount_opt = 0;
+	fs_info->max_extent = (u64)-1;
+	fs_info->max_inline = 8192 * 1024;
+	fs_info->delalloc_bytes = 0;
+	setup_bdi(fs_info, &fs_info->bdi);
 	fs_info->btree_inode = new_inode(sb);
 	fs_info->btree_inode->i_ino = 1;
+	fs_info->btree_inode->i_nlink = 1;
 	fs_info->btree_inode->i_size = sb->s_bdev->bd_inode->i_size;
 	fs_info->btree_inode->i_mapping->a_ops = &btree_aops;
-	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
-	fs_info->hash_tfm = crypto_alloc_hash("sha256", 0, CRYPTO_ALG_ASYNC);
-	spin_lock_init(&fs_info->hash_lock);
+	fs_info->btree_inode->i_mapping->backing_dev_info = &fs_info->bdi;
 
-	if (!fs_info->hash_tfm || IS_ERR(fs_info->hash_tfm)) {
-		printk("failed to allocate sha256 hash\n");
-		return NULL;
-	}
+	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
+			     fs_info->btree_inode->i_mapping,
+			     GFP_NOFS);
+	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree,
+			     GFP_NOFS);
+
+	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
+
+	extent_io_tree_init(&fs_info->free_space_cache,
+			     fs_info->btree_inode->i_mapping, GFP_NOFS);
+	extent_io_tree_init(&fs_info->block_group_cache,
+			     fs_info->btree_inode->i_mapping, GFP_NOFS);
+	extent_io_tree_init(&fs_info->pinned_extents,
+			     fs_info->btree_inode->i_mapping, GFP_NOFS);
+	extent_io_tree_init(&fs_info->pending_del,
+			     fs_info->btree_inode->i_mapping, GFP_NOFS);
+	extent_io_tree_init(&fs_info->extent_ins,
+			     fs_info->btree_inode->i_mapping, GFP_NOFS);
+	fs_info->do_barriers = 1;
+	fs_info->closing = 0;
+	fs_info->total_pinned = 0;
+	fs_info->last_alloc = 0;
+	fs_info->last_data_alloc = 0;
+	fs_info->extra_alloc_bits = 0;
+	fs_info->extra_data_alloc_bits = 0;
+
+#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18)
+	INIT_WORK(&fs_info->trans_work, btrfs_transaction_cleaner, fs_info);
+#else
+	INIT_DELAYED_WORK(&fs_info->trans_work, btrfs_transaction_cleaner);
+#endif
+	BTRFS_I(fs_info->btree_inode)->root = tree_root;
+	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
+	       sizeof(struct btrfs_key));
+	insert_inode_hash(fs_info->btree_inode);
+	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
 
 	mutex_init(&fs_info->trans_mutex);
 	mutex_init(&fs_info->fs_mutex);
-	memset(&fs_info->current_insert, 0, sizeof(fs_info->current_insert));
-	memset(&fs_info->last_insert, 0, sizeof(fs_info->last_insert));
 
-	__setup_root(disk_super, tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
+#if 0
+	ret = add_hasher(fs_info, "crc32c");
+	if (ret) {
+		printk("btrfs: failed hash setup, modprobe cryptomgr?\n");
+		err = -ENOMEM;
+		goto fail_iput;
+	}
+#endif
+	__setup_root(4096, 4096, 4096, 4096, tree_root,
+		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
 
-	fs_info->sb_buffer = read_tree_block(tree_root, sb_buffer->b_blocknr);
+	fs_info->sb_buffer = read_tree_block(tree_root,
+					     BTRFS_SUPER_INFO_OFFSET,
+					     4096);
 
-	if (!fs_info->sb_buffer) {
-printk("failed2\n");
-		return NULL;
+	if (!fs_info->sb_buffer)
+		goto fail_iput;
+
+	read_extent_buffer(fs_info->sb_buffer, &fs_info->super_copy, 0,
+			   sizeof(fs_info->super_copy));
+
+	read_extent_buffer(fs_info->sb_buffer, fs_info->fsid,
+			   (unsigned long)btrfs_super_fsid(fs_info->sb_buffer),
+			   BTRFS_FSID_SIZE);
+
+	disk_super = &fs_info->super_copy;
+	if (!btrfs_super_root(disk_super))
+		goto fail_sb_buffer;
+
+	if (btrfs_super_num_devices(disk_super) != fs_devices->num_devices) {
+		printk("Btrfs: wanted %llu devices, but found %llu\n",
+		       (unsigned long long)btrfs_super_num_devices(disk_super),
+		       (unsigned long long)fs_devices->num_devices);
+		goto fail_sb_buffer;
 	}
-	brelse(sb_buffer);
-	sb_buffer = NULL;
-	disk_super = (struct btrfs_super_block *)fs_info->sb_buffer->b_data;
-	fs_info->disk_super = disk_super;
+	nodesize = btrfs_super_nodesize(disk_super);
+	leafsize = btrfs_super_leafsize(disk_super);
+	sectorsize = btrfs_super_sectorsize(disk_super);
+	stripesize = btrfs_super_stripesize(disk_super);
+	tree_root->nodesize = nodesize;
+	tree_root->leafsize = leafsize;
+	tree_root->sectorsize = sectorsize;
+	tree_root->stripesize = stripesize;
+	sb_set_blocksize(sb, sectorsize);
 
-	tree_root->node = read_tree_block(tree_root,
-					  btrfs_super_root(disk_super));
-	BUG_ON(!tree_root->node);
+	i_size_write(fs_info->btree_inode,
+		     btrfs_super_total_bytes(disk_super));
 
-	ret = find_and_setup_root(disk_super, tree_root, fs_info,
-				  BTRFS_EXTENT_TREE_OBJECTID, extent_root);
-	BUG_ON(ret);
+	if (strncmp((char *)(&disk_super->magic), BTRFS_MAGIC,
+		    sizeof(disk_super->magic))) {
+		printk("btrfs: valid FS not found on %s\n", sb->s_id);
+		goto fail_sb_buffer;
+	}
+
+	mutex_lock(&fs_info->fs_mutex);
 
-	ret = find_and_setup_root(disk_super, tree_root, fs_info,
-				  BTRFS_INODE_MAP_OBJECTID, inode_root);
+	ret = btrfs_read_sys_array(tree_root);
 	BUG_ON(ret);
 
-	ret = find_and_setup_root(disk_super, tree_root, fs_info,
-				  BTRFS_FS_TREE_OBJECTID, root);
+	blocksize = btrfs_level_size(tree_root,
+				     btrfs_super_chunk_root_level(disk_super));
+
+	__setup_root(nodesize, leafsize, sectorsize, stripesize,
+		     chunk_root, fs_info, BTRFS_CHUNK_TREE_OBJECTID);
+
+	chunk_root->node = read_tree_block(chunk_root,
+					   btrfs_super_chunk_root(disk_super),
+					   blocksize);
+	BUG_ON(!chunk_root->node);
+
+	ret = btrfs_read_chunk_tree(chunk_root);
 	BUG_ON(ret);
-	root->commit_root = root->node;
-	get_bh(root->node);
-	root->ref_cows = 1;
-	root->fs_info->generation = root->root_key.offset + 1;
-	return root;
+
+	blocksize = btrfs_level_size(tree_root,
+				     btrfs_super_root_level(disk_super));
+
+
+	tree_root->node = read_tree_block(tree_root,
+					  btrfs_super_root(disk_super),
+					  blocksize);
+	if (!tree_root->node)
+		goto fail_sb_buffer;
+
+
+	ret = find_and_setup_root(tree_root, fs_info,
+				  BTRFS_EXTENT_TREE_OBJECTID, extent_root);
+	if (ret)
+		goto fail_tree_root;
+	extent_root->track_dirty = 1;
+
+	ret = find_and_setup_root(tree_root, fs_info,
+				  BTRFS_DEV_TREE_OBJECTID, dev_root);
+	dev_root->track_dirty = 1;
+
+	if (ret)
+		goto fail_extent_root;
+
+	btrfs_read_block_groups(extent_root);
+
+	fs_info->generation = btrfs_super_generation(disk_super) + 1;
+	mutex_unlock(&fs_info->fs_mutex);
+	return tree_root;
+
+fail_extent_root:
+	free_extent_buffer(extent_root->node);
+fail_tree_root:
+	mutex_unlock(&fs_info->fs_mutex);
+	free_extent_buffer(tree_root->node);
+fail_sb_buffer:
+	free_extent_buffer(fs_info->sb_buffer);
+fail_iput:
+	iput(fs_info->btree_inode);
+fail:
+	close_all_devices(fs_info);
+	kfree(extent_root);
+	kfree(tree_root);
+	kfree(fs_info);
+	return ERR_PTR(err);
 }
 
 int write_ctree_super(struct btrfs_trans_handle *trans, struct btrfs_root
 		      *root)
 {
-	struct buffer_head *bh = root->fs_info->sb_buffer;
-	btrfs_set_super_root(root->fs_info->disk_super,
-			     root->fs_info->tree_root->node->b_blocknr);
-	lock_buffer(bh);
-	clear_buffer_dirty(bh);
-	csum_tree_block(root, bh, 0);
-	bh->b_end_io = end_buffer_write_sync;
-	get_bh(bh);
-	submit_bh(WRITE, bh);
-	wait_on_buffer(bh);
-	if (!buffer_uptodate(bh)) {
-		WARN_ON(1);
-		return -EIO;
+	int ret;
+	struct extent_buffer *super = root->fs_info->sb_buffer;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	struct super_block *sb = root->fs_info->sb;
+
+	if (!btrfs_test_opt(root, NOBARRIER))
+		blkdev_issue_flush(sb->s_bdev, NULL);
+	set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree, super);
+	ret = sync_page_range_nolock(btree_inode, btree_inode->i_mapping,
+				     super->start, super->len);
+	if (!btrfs_test_opt(root, NOBARRIER))
+		blkdev_issue_flush(sb->s_bdev, NULL);
+	return ret;
+}
+
+int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
+{
+	radix_tree_delete(&fs_info->fs_roots_radix,
+			  (unsigned long)root->root_key.objectid);
+	if (root->in_sysfs)
+		btrfs_sysfs_del_root(root);
+	if (root->inode)
+		iput(root->inode);
+	if (root->node)
+		free_extent_buffer(root->node);
+	if (root->commit_root)
+		free_extent_buffer(root->commit_root);
+	if (root->name)
+		kfree(root->name);
+	kfree(root);
+	return 0;
+}
+
+static int del_fs_roots(struct btrfs_fs_info *fs_info)
+{
+	int ret;
+	struct btrfs_root *gang[8];
+	int i;
+
+	while(1) {
+		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
+					     (void **)gang, 0,
+					     ARRAY_SIZE(gang));
+		if (!ret)
+			break;
+		for (i = 0; i < ret; i++)
+			btrfs_free_fs_root(fs_info, gang[i]);
 	}
 	return 0;
 }
@@ -370,42 +983,187 @@ int close_ctree(struct btrfs_root *root)
 {
 	int ret;
 	struct btrfs_trans_handle *trans;
+	struct btrfs_fs_info *fs_info = root->fs_info;
 
+	fs_info->closing = 1;
+	btrfs_transaction_flush_work(root);
+	mutex_lock(&fs_info->fs_mutex);
+	btrfs_defrag_dirty_roots(root->fs_info);
 	trans = btrfs_start_transaction(root, 1);
-	btrfs_commit_transaction(trans, root);
+	ret = btrfs_commit_transaction(trans, root);
 	/* run commit again to  drop the original snapshot */
 	trans = btrfs_start_transaction(root, 1);
 	btrfs_commit_transaction(trans, root);
 	ret = btrfs_write_and_wait_transaction(NULL, root);
 	BUG_ON(ret);
 	write_ctree_super(NULL, root);
+	mutex_unlock(&fs_info->fs_mutex);
 
-	if (root->node)
-		btrfs_block_release(root, root->node);
-	if (root->fs_info->extent_root->node)
-		btrfs_block_release(root->fs_info->extent_root,
-				    root->fs_info->extent_root->node);
-	if (root->fs_info->inode_root->node)
-		btrfs_block_release(root->fs_info->inode_root,
-				    root->fs_info->inode_root->node);
-	if (root->fs_info->tree_root->node)
-		btrfs_block_release(root->fs_info->tree_root,
-				    root->fs_info->tree_root->node);
-	btrfs_block_release(root, root->commit_root);
-	btrfs_block_release(root, root->fs_info->sb_buffer);
-	crypto_free_hash(root->fs_info->hash_tfm);
-	truncate_inode_pages(root->fs_info->btree_inode->i_mapping, 0);
-	iput(root->fs_info->btree_inode);
-	kfree(root->fs_info->extent_root);
-	kfree(root->fs_info->inode_root);
-	kfree(root->fs_info->tree_root);
-	kfree(root->fs_info);
-	kfree(root);
+	if (fs_info->delalloc_bytes) {
+		printk("btrfs: at unmount delalloc count %Lu\n",
+		       fs_info->delalloc_bytes);
+	}
+	if (fs_info->extent_root->node)
+		free_extent_buffer(fs_info->extent_root->node);
+
+	if (fs_info->tree_root->node)
+		free_extent_buffer(fs_info->tree_root->node);
+
+	if (root->fs_info->chunk_root->node);
+		free_extent_buffer(root->fs_info->chunk_root->node);
+
+	if (root->fs_info->dev_root->node);
+		free_extent_buffer(root->fs_info->dev_root->node);
+
+	free_extent_buffer(fs_info->sb_buffer);
+
+	btrfs_free_block_groups(root->fs_info);
+	del_fs_roots(fs_info);
+
+	filemap_write_and_wait(fs_info->btree_inode->i_mapping);
+
+	extent_io_tree_empty_lru(&fs_info->free_space_cache);
+	extent_io_tree_empty_lru(&fs_info->block_group_cache);
+	extent_io_tree_empty_lru(&fs_info->pinned_extents);
+	extent_io_tree_empty_lru(&fs_info->pending_del);
+	extent_io_tree_empty_lru(&fs_info->extent_ins);
+	extent_io_tree_empty_lru(&BTRFS_I(fs_info->btree_inode)->io_tree);
+
+	truncate_inode_pages(fs_info->btree_inode->i_mapping, 0);
+
+	iput(fs_info->btree_inode);
+#if 0
+	while(!list_empty(&fs_info->hashers)) {
+		struct btrfs_hasher *hasher;
+		hasher = list_entry(fs_info->hashers.next, struct btrfs_hasher,
+				    hashers);
+		list_del(&hasher->hashers);
+		crypto_free_hash(&fs_info->hash_tfm);
+		kfree(hasher);
+	}
+#endif
+	close_all_devices(fs_info);
+	btrfs_mapping_tree_free(&fs_info->mapping_tree);
+	bdi_destroy(&fs_info->bdi);
+
+	kfree(fs_info->extent_root);
+	kfree(fs_info->tree_root);
+	kfree(fs_info->chunk_root);
+	kfree(fs_info->dev_root);
 	return 0;
 }
 
-void btrfs_block_release(struct btrfs_root *root, struct buffer_head *buf)
+int btrfs_buffer_uptodate(struct extent_buffer *buf)
+{
+	struct inode *btree_inode = buf->first_page->mapping->host;
+	return extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf);
+}
+
+int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
+{
+	struct inode *btree_inode = buf->first_page->mapping->host;
+	return set_extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree,
+					  buf);
+}
+
+void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
+{
+	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
+	u64 transid = btrfs_header_generation(buf);
+	struct inode *btree_inode = root->fs_info->btree_inode;
+
+	if (transid != root->fs_info->generation) {
+		printk(KERN_CRIT "transid mismatch buffer %llu, found %Lu running %Lu\n",
+			(unsigned long long)buf->start,
+			transid, root->fs_info->generation);
+		WARN_ON(1);
+	}
+	set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree, buf);
+}
+
+void btrfs_throttle(struct btrfs_root *root)
+{
+	struct backing_dev_info *bdi;
+
+	bdi = root->fs_info->sb->s_bdev->bd_inode->i_mapping->backing_dev_info;
+	if (root->fs_info->throttles && bdi_write_congested(bdi)) {
+#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,18)
+		congestion_wait(WRITE, HZ/20);
+#else
+		blk_congestion_wait(WRITE, HZ/20);
+#endif
+	}
+}
+
+void btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr)
+{
+	balance_dirty_pages_ratelimited_nr(
+				   root->fs_info->btree_inode->i_mapping, 1);
+}
+
+void btrfs_set_buffer_defrag(struct extent_buffer *buf)
 {
-	brelse(buf);
+	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	set_extent_bits(&BTRFS_I(btree_inode)->io_tree, buf->start,
+			buf->start + buf->len - 1, EXTENT_DEFRAG, GFP_NOFS);
 }
 
+void btrfs_set_buffer_defrag_done(struct extent_buffer *buf)
+{
+	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	set_extent_bits(&BTRFS_I(btree_inode)->io_tree, buf->start,
+			buf->start + buf->len - 1, EXTENT_DEFRAG_DONE,
+			GFP_NOFS);
+}
+
+int btrfs_buffer_defrag(struct extent_buffer *buf)
+{
+	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	return test_range_bit(&BTRFS_I(btree_inode)->io_tree,
+		     buf->start, buf->start + buf->len - 1, EXTENT_DEFRAG, 0);
+}
+
+int btrfs_buffer_defrag_done(struct extent_buffer *buf)
+{
+	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	return test_range_bit(&BTRFS_I(btree_inode)->io_tree,
+		     buf->start, buf->start + buf->len - 1,
+		     EXTENT_DEFRAG_DONE, 0);
+}
+
+int btrfs_clear_buffer_defrag_done(struct extent_buffer *buf)
+{
+	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	return clear_extent_bits(&BTRFS_I(btree_inode)->io_tree,
+		     buf->start, buf->start + buf->len - 1,
+		     EXTENT_DEFRAG_DONE, GFP_NOFS);
+}
+
+int btrfs_clear_buffer_defrag(struct extent_buffer *buf)
+{
+	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	return clear_extent_bits(&BTRFS_I(btree_inode)->io_tree,
+		     buf->start, buf->start + buf->len - 1,
+		     EXTENT_DEFRAG, GFP_NOFS);
+}
+
+int btrfs_read_buffer(struct extent_buffer *buf)
+{
+	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	return read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
+					buf, 0, 1, btree_get_extent);
+}
+
+static struct extent_io_ops btree_extent_io_ops = {
+	.writepage_io_hook = btree_writepage_io_hook,
+	.submit_bio_hook = btree_submit_bio_hook,
+	/* note we're sharing with inode.c for the merge bio hook */
+	.merge_bio_hook = btrfs_merge_bio_hook,
+};