Btrfs: Create extent_buffer interface for large blocksizes

Signed-off-by: NChris Mason <chris.mason@oracle.com>

fs/btrfs/Makefile

@@ -23,4 +23,6 @@ modules_install:
 clean:
 	$(MAKE) -C $(KERNELDIR) M=`pwd` clean
 
+tester:
+	$(MAKE) -C $(KERNELDIR) M=`pwd` tree-defrag.o transaction.o sysfs.o super.o root-tree.o inode-map.o inode-item.o inode.o file-item.o file.o extent_map.o disk-io.o ctree.o dir-item.o extent-tree.o
 endif

fs/btrfs/dir-item.c

@@ -33,7 +33,7 @@ static struct btrfs_dir_item *insert_with_overflow(struct btrfs_trans_handle
 	int ret;
 	char *ptr;
 	struct btrfs_item *item;
-	struct btrfs_leaf *leaf;
+	struct extent_buffer *leaf;
 
 	ret = btrfs_insert_empty_item(trans, root, path, cpu_key, data_size);
 	if (ret == -EEXIST) {
@@ -49,11 +49,11 @@ static struct btrfs_dir_item *insert_with_overflow(struct btrfs_trans_handle
 	if (ret < 0)
 		return ERR_PTR(ret);
 	WARN_ON(ret > 0);
-	leaf = btrfs_buffer_leaf(path->nodes[0]);
-	item = leaf->items + path->slots[0];
+	leaf = path->nodes[0];
+	item = btrfs_item_nr(leaf, path->slots[0]);
 	ptr = btrfs_item_ptr(leaf, path->slots[0], char);
-	BUG_ON(data_size > btrfs_item_size(item));
-	ptr += btrfs_item_size(item) - data_size;
+	BUG_ON(data_size > btrfs_item_size(leaf, item));
+	ptr += btrfs_item_size(leaf, item) - data_size;
 	return (struct btrfs_dir_item *)ptr;
 }
 
@@ -65,12 +65,13 @@ int btrfs_insert_dir_item(struct btrfs_trans_handle *trans, struct btrfs_root
 	int ret2 = 0;
 	struct btrfs_path *path;
 	struct btrfs_dir_item *dir_item;
-	char *name_ptr;
+	struct extent_buffer *leaf;
+	unsigned long name_ptr;
 	struct btrfs_key key;
+	struct btrfs_disk_key disk_key;
 	u32 data_size;
 
 	key.objectid = dir;
-	key.flags = 0;
 	btrfs_set_key_type(&key, BTRFS_DIR_ITEM_KEY);
 	ret = btrfs_name_hash(name, name_len, &key.offset);
 	BUG_ON(ret);
@@ -85,14 +86,16 @@ int btrfs_insert_dir_item(struct btrfs_trans_handle *trans, struct btrfs_root
 		goto out;
 	}
 
-	btrfs_cpu_key_to_disk(&dir_item->location, location);
-	btrfs_set_dir_type(dir_item, type);
-	btrfs_set_dir_flags(dir_item, 0);
-	btrfs_set_dir_name_len(dir_item, name_len);
-	name_ptr = (char *)(dir_item + 1);
+	leaf = path->nodes[0];
+	btrfs_cpu_key_to_disk(&disk_key, location);
+	btrfs_set_dir_item_key(leaf, dir_item, &disk_key);
+	btrfs_set_dir_type(leaf, dir_item, type);
+	btrfs_set_dir_flags(leaf, dir_item, 0);
+	btrfs_set_dir_name_len(leaf, dir_item, name_len);
+	name_ptr = (unsigned long)(dir_item + 1);
 
-	btrfs_memcpy(root, path->nodes[0]->b_data, name_ptr, name, name_len);
-	btrfs_mark_buffer_dirty(path->nodes[0]);
+	write_extent_buffer(leaf, name, name_ptr, name_len);
+	btrfs_mark_buffer_dirty(leaf);
 
 second_insert:
 	/* FIXME, use some real flag for selecting the extra index */
@@ -110,13 +113,15 @@ int btrfs_insert_dir_item(struct btrfs_trans_handle *trans, struct btrfs_root
 		ret2 = PTR_ERR(dir_item);
 		goto out;
 	}
-	btrfs_cpu_key_to_disk(&dir_item->location, location);
-	btrfs_set_dir_type(dir_item, type);
-	btrfs_set_dir_flags(dir_item, 0);
-	btrfs_set_dir_name_len(dir_item, name_len);
-	name_ptr = (char *)(dir_item + 1);
-	btrfs_memcpy(root, path->nodes[0]->b_data, name_ptr, name, name_len);
-	btrfs_mark_buffer_dirty(path->nodes[0]);
+	leaf = path->nodes[0];
+	btrfs_cpu_key_to_disk(&disk_key, location);
+	btrfs_set_dir_item_key(leaf, dir_item, &disk_key);
+	btrfs_set_dir_type(leaf, dir_item, type);
+	btrfs_set_dir_flags(leaf, dir_item, 0);
+	btrfs_set_dir_name_len(leaf, dir_item, name_len);
+	name_ptr = (unsigned long)(dir_item + 1);
+	write_extent_buffer(leaf, name, name_ptr, name_len);
+	btrfs_mark_buffer_dirty(leaf);
 out:
 	btrfs_free_path(path);
 	if (ret)
@@ -136,14 +141,15 @@ struct btrfs_dir_item *btrfs_lookup_dir_item(struct btrfs_trans_handle *trans,
 	struct btrfs_key key;
 	int ins_len = mod < 0 ? -1 : 0;
 	int cow = mod != 0;
-	struct btrfs_disk_key *found_key;
-	struct btrfs_leaf *leaf;
+	struct btrfs_key found_key;
+	struct extent_buffer *leaf;
 
 	key.objectid = dir;
-	key.flags = 0;
 	btrfs_set_key_type(&key, BTRFS_DIR_ITEM_KEY);
+
 	ret = btrfs_name_hash(name, name_len, &key.offset);
 	BUG_ON(ret);
+
 	ret = btrfs_search_slot(trans, root, &key, path, ins_len, cow);
 	if (ret < 0)
 		return ERR_PTR(ret);
@@ -152,12 +158,13 @@ struct btrfs_dir_item *btrfs_lookup_dir_item(struct btrfs_trans_handle *trans,
 			return NULL;
 		path->slots[0]--;
 	}
-	leaf = btrfs_buffer_leaf(path->nodes[0]);
-	found_key = &leaf->items[path->slots[0]].key;
 
-	if (btrfs_disk_key_objectid(found_key) != dir ||
-	    btrfs_disk_key_type(found_key) != BTRFS_DIR_ITEM_KEY ||
-	    btrfs_disk_key_offset(found_key) != key.offset)
+	leaf = path->nodes[0];
+	btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+
+	if (found_key.objectid != dir ||
+	    btrfs_key_type(&found_key) != BTRFS_DIR_ITEM_KEY ||
+	    found_key.offset != key.offset)
 		return NULL;
 
 	return btrfs_match_dir_item_name(root, path, name, name_len);
@@ -176,7 +183,6 @@ btrfs_lookup_dir_index_item(struct btrfs_trans_handle *trans,
 	int cow = mod != 0;
 
 	key.objectid = dir;
-	key.flags = 0;
 	btrfs_set_key_type(&key, BTRFS_DIR_INDEX_KEY);
 	key.offset = objectid;
 
@@ -193,21 +199,22 @@ struct btrfs_dir_item *btrfs_match_dir_item_name(struct btrfs_root *root,
 			      const char *name, int name_len)
 {
 	struct btrfs_dir_item *dir_item;
-	char *name_ptr;
+	unsigned long name_ptr;
 	u32 total_len;
 	u32 cur = 0;
 	u32 this_len;
-	struct btrfs_leaf *leaf;
+	struct extent_buffer *leaf;
 
-	leaf = btrfs_buffer_leaf(path->nodes[0]);
+	leaf = path->nodes[0];
 	dir_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dir_item);
-	total_len = btrfs_item_size(leaf->items + path->slots[0]);
+	total_len = btrfs_item_size_nr(leaf, path->slots[0]);
 	while(cur < total_len) {
-		this_len = sizeof(*dir_item) + btrfs_dir_name_len(dir_item);
-		name_ptr = (char *)(dir_item + 1);
+		this_len = sizeof(*dir_item) +
+			btrfs_dir_name_len(leaf, dir_item);
+		name_ptr = (unsigned long)(dir_item + 1);
 
-		if (btrfs_dir_name_len(dir_item) == name_len &&
-		    memcmp(name_ptr, name, name_len) == 0)
+		if (btrfs_dir_name_len(leaf, dir_item) == name_len &&
+		    memcmp_extent_buffer(leaf, name, name_ptr, name_len) == 0)
 			return dir_item;
 
 		cur += this_len;
@@ -223,20 +230,23 @@ int btrfs_delete_one_dir_name(struct btrfs_trans_handle *trans,
 			      struct btrfs_dir_item *di)
 {
 
-	struct btrfs_leaf *leaf;
+	struct extent_buffer *leaf;
 	u32 sub_item_len;
 	u32 item_len;
 	int ret = 0;
 
-	leaf = btrfs_buffer_leaf(path->nodes[0]);
-	sub_item_len = sizeof(*di) + btrfs_dir_name_len(di);
-	item_len = btrfs_item_size(leaf->items + path->slots[0]);
-	if (sub_item_len == btrfs_item_size(leaf->items + path->slots[0])) {
+	leaf = path->nodes[0];
+	sub_item_len = sizeof(*di) + btrfs_dir_name_len(leaf, di);
+	item_len = btrfs_item_size_nr(leaf, path->slots[0]);
+	if (sub_item_len == item_len) {
 		ret = btrfs_del_item(trans, root, path);
 	} else {
-		char *ptr = (char *)di;
-		char *start = btrfs_item_ptr(leaf, path->slots[0], char);
-		btrfs_memmove(root, leaf, ptr, ptr + sub_item_len,
+		/* MARKER */
+		unsigned long ptr = (unsigned long)di;
+		unsigned long start;
+
+		start = btrfs_item_ptr_offset(leaf, path->slots[0]);
+		memmove_extent_buffer(leaf, ptr, ptr + sub_item_len,
 			item_len - (ptr + sub_item_len - start));
 		ret = btrfs_truncate_item(trans, root, path,
 					  item_len - sub_item_len);

fs/btrfs/disk-io.c

@@ -23,138 +23,132 @@
 #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"
 
-u64 bh_blocknr(struct buffer_head *bh)
+#if 0
+static int check_tree_block(struct btrfs_root *root, struct extent_buffer *buf)
 {
-	return bh->b_blocknr;
-}
-
-static int check_tree_block(struct btrfs_root *root, struct buffer_head *buf)
-{
-	struct btrfs_node *node = btrfs_buffer_node(buf);
-	if (bh_blocknr(buf) != btrfs_header_blocknr(&node->header)) {
-		printk(KERN_CRIT "bh_blocknr(buf) is %llu, header is %llu\n",
-		       (unsigned long long)bh_blocknr(buf),
-		       (unsigned long long)btrfs_header_blocknr(&node->header));
+	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)
+struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
+					    u64 blocknr)
 {
-	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;
-
-
-	page = find_lock_page(mapping, index);
-	if (!page)
-		return NULL;
-
-	if (!page_has_buffers(page))
-		goto out_unlock;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	return find_extent_buffer(&BTRFS_I(btree_inode)->extent_tree,
+				   blocknr * root->sectorsize,
+				   root->sectorsize, GFP_NOFS);
+}
 
-	head = page_buffers(page);
-	bh = head;
-	do {
-		if (buffer_mapped(bh) && bh_blocknr(bh) == 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_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
+						 u64 blocknr)
+{
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	return alloc_extent_buffer(&BTRFS_I(btree_inode)->extent_tree,
+				   blocknr * root->sectorsize,
+				   root->sectorsize, GFP_NOFS);
 }
 
-int btrfs_map_bh_to_logical(struct btrfs_root *root, struct buffer_head *bh,
-			     u64 logical)
+struct extent_map *btree_get_extent(struct inode *inode, struct page *page,
+				    size_t page_offset, u64 start, u64 end,
+				    int create)
 {
-	if (logical == 0) {
-		bh->b_bdev = NULL;
-		bh->b_blocknr = 0;
-		set_buffer_mapped(bh);
-	} else {
-		map_bh(bh, root->fs_info->sb, logical);
+	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	struct extent_map *em;
+	int ret;
+
+again:
+	em = lookup_extent_mapping(em_tree, start, end);
+	if (em) {
+		goto out;
 	}
-	return 0;
+	em = alloc_extent_map(GFP_NOFS);
+	if (!em) {
+		em = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+	em->start = 0;
+	em->end = (i_size_read(inode) & ~((u64)PAGE_CACHE_SIZE -1)) - 1;
+	em->block_start = 0;
+	em->block_end = em->end;
+	em->bdev = inode->i_sb->s_bdev;
+	ret = add_extent_mapping(em_tree, em);
+	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)
+static int btree_writepage(struct page *page, struct writeback_control *wbc)
 {
-	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;
-	int err;
-	u64 first_block = index << (PAGE_CACHE_SHIFT - blockbits);
+	struct extent_map_tree *tree;
+	tree = &BTRFS_I(page->mapping->host)->extent_tree;
+	return extent_write_full_page(tree, page, btree_get_extent, wbc);
+}
+int btree_readpage(struct file *file, struct page *page)
+{
+	struct extent_map_tree *tree;
+	tree = &BTRFS_I(page->mapping->host)->extent_tree;
+	return extent_read_full_page(tree, page, btree_get_extent);
+}
 
-	page = find_or_create_page(mapping, index, GFP_NOFS);
-	if (!page)
-		return NULL;
+static int btree_releasepage(struct page *page, gfp_t unused_gfp_flags)
+{
+	struct extent_map_tree *tree;
+	int ret;
 
-	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)) {
-			err = btrfs_map_bh_to_logical(root, bh, first_block);
-			BUG_ON(err);
-		}
-		if (bh_blocknr(bh) == blocknr) {
-			ret = bh;
-			get_bh(bh);
-			goto out_unlock;
-		}
-		bh = bh->b_this_page;
-		first_block++;
-	} while (bh != head);
-out_unlock:
-	unlock_page(page);
-	if (ret)
-		touch_buffer(ret);
-	page_cache_release(page);
+	BUG_ON(page->private != 1);
+	tree = &BTRFS_I(page->mapping->host)->extent_tree;
+	ret = try_release_extent_mapping(tree, page);
+	if (ret == 1) {
+		ClearPagePrivate(page);
+		set_page_private(page, 0);
+		page_cache_release(page);
+	}
 	return ret;
 }
 
-static int btree_get_block(struct inode *inode, sector_t iblock,
-			   struct buffer_head *bh, int create)
+static void btree_invalidatepage(struct page *page, unsigned long offset)
 {
-	int err;
-	struct btrfs_root *root = BTRFS_I(bh->b_page->mapping->host)->root;
-	err = btrfs_map_bh_to_logical(root, bh, iblock);
-	return err;
+	struct extent_map_tree *tree;
+	tree = &BTRFS_I(page->mapping->host)->extent_tree;
+	extent_invalidatepage(tree, page, offset);
+	btree_releasepage(page, GFP_NOFS);
 }
 
 int btrfs_csum_data(struct btrfs_root * root, char *data, size_t len,
 		    char *result)
 {
+	return 0;
+#if 0
 	u32 crc;
 	crc = crc32c(0, data, len);
 	memcpy(result, &crc, BTRFS_CRC32_SIZE);
 	return 0;
+#endif
 }
 
-static int csum_tree_block(struct btrfs_root *root, struct buffer_head *bh,
+#if 0
+static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf,
 			   int verify)
 {
+	return 0;
 	char result[BTRFS_CRC32_SIZE];
 	int ret;
 	struct btrfs_node *node;
@@ -176,7 +170,9 @@ static int csum_tree_block(struct btrfs_root *root, struct buffer_head *bh,
 	}
 	return 0;
 }
+#endif
 
+#if 0
 static int btree_writepage(struct page *page, struct writeback_control *wbc)
 {
 	struct buffer_head *bh;
@@ -195,87 +191,65 @@ 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,
+	.releasepage	= btree_releasepage,
+	.invalidatepage = btree_invalidatepage,
 	.sync_page	= block_sync_page,
 };
 
 int readahead_tree_block(struct btrfs_root *root, u64 blocknr)
 {
-	struct buffer_head *bh = NULL;
+	struct extent_buffer *buf = NULL;
+	struct inode *btree_inode = root->fs_info->btree_inode;
 	int ret = 0;
 
-	bh = btrfs_find_create_tree_block(root, blocknr);
-	if (!bh)
+	buf = btrfs_find_create_tree_block(root, blocknr);
+	if (!buf)
 		return 0;
-	if (buffer_uptodate(bh)) {
-		ret = 1;
-		goto done;
-	}
-	if (test_set_buffer_locked(bh)) {
-		ret = 1;
-		goto done;
-	}
-	if (!buffer_uptodate(bh)) {
-		get_bh(bh);
-		bh->b_end_io = end_buffer_read_sync;
-		submit_bh(READ, bh);
-	} else {
-		unlock_buffer(bh);
-		ret = 1;
-	}
-done:
-	brelse(bh);
+	read_extent_buffer_pages(&BTRFS_I(btree_inode)->extent_tree,
+				 buf, 0);
+	free_extent_buffer(buf);
 	return ret;
 }
 
-struct buffer_head *read_tree_block(struct btrfs_root *root, u64 blocknr)
+struct extent_buffer *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;
-	if (buffer_uptodate(bh))
-		goto uptodate;
-	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;
-	} else {
-		unlock_buffer(bh);
-	}
-uptodate:
-	if (!buffer_checked(bh)) {
-		csum_tree_block(root, bh, 1);
-		set_buffer_checked(bh);
-	}
-	if (check_tree_block(root, bh))
-		goto fail;
-	return bh;
-fail:
-	brelse(bh);
-	return NULL;
+	struct extent_buffer *buf = NULL;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+
+	buf = btrfs_find_create_tree_block(root, blocknr);
+	if (!buf)
+		return NULL;
+	read_extent_buffer_pages(&BTRFS_I(btree_inode)->extent_tree,
+				 buf, 1);
+	return buf;
 }
 
 int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
-		     struct buffer_head *buf)
+		     struct extent_buffer *buf)
 {
-	WARN_ON(atomic_read(&buf->b_count) == 0);
-	lock_buffer(buf);
-	clear_buffer_dirty(buf);
-	unlock_buffer(buf);
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->extent_tree, buf);
+	return 0;
+}
+
+int wait_on_tree_block_writeback(struct btrfs_root *root,
+				 struct extent_buffer *buf)
+{
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	wait_on_extent_buffer_writeback(&BTRFS_I(btree_inode)->extent_tree,
+					buf);
+	return 0;
+}
+
+int set_tree_block_dirty(struct btrfs_root *root, struct extent_buffer *buf)
+{
+	struct inode *btree_inode = root->fs_info->btree_inode;
+	set_extent_buffer_dirty(&BTRFS_I(btree_inode)->extent_tree, buf);
 	return 0;
 }
 
@@ -287,7 +261,9 @@ static int __setup_root(int blocksize,
 	root->node = NULL;
 	root->inode = NULL;
 	root->commit_root = NULL;
-	root->blocksize = blocksize;
+	root->sectorsize = blocksize;
+	root->nodesize = blocksize;
+	root->leafsize = blocksize;
 	root->ref_cows = 0;
 	root->fs_info = fs_info;
 	root->objectid = objectid;
@@ -332,7 +308,7 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_fs_info *fs_info,
 	struct btrfs_root *root;
 	struct btrfs_root *tree_root = fs_info->tree_root;
 	struct btrfs_path *path;
-	struct btrfs_leaf *l;
+	struct extent_buffer *l;
 	u64 highest_inode;
 	int ret = 0;
 
@@ -361,11 +337,10 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_fs_info *fs_info,
 			ret = -ENOENT;
 		goto out;
 	}
-	l = btrfs_buffer_leaf(path->nodes[0]);
-	memcpy(&root->root_item,
-	       btrfs_item_ptr(l, path->slots[0], struct btrfs_root_item),
+	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);
@@ -406,21 +381,21 @@ struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info,
 				(unsigned long)root->root_key.objectid,
 				root);
 	if (ret) {
-		brelse(root->node);
+		free_extent_buffer(root->node);
 		kfree(root);
 		return ERR_PTR(ret);
 	}
 
 	ret = btrfs_set_root_name(root, name, namelen);
 	if (ret) {
-		brelse(root->node);
+		free_extent_buffer(root->node);
 		kfree(root);
 		return ERR_PTR(ret);
 	}
 
 	ret = btrfs_sysfs_add_root(root);
 	if (ret) {
-		brelse(root->node);
+		free_extent_buffer(root->node);
 		kfree(root->name);
 		kfree(root);
 		return ERR_PTR(ret);
@@ -471,6 +446,9 @@ struct btrfs_root *open_ctree(struct super_block *sb)
 	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;
+	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree,
+			     fs_info->btree_inode->i_mapping,
+			     GFP_NOFS);
 	fs_info->do_barriers = 1;
 	fs_info->closing = 0;
 
@@ -493,10 +471,14 @@ struct btrfs_root *open_ctree(struct super_block *sb)
 
 	if (!fs_info->sb_buffer)
 		goto fail_iput;
-	disk_super = (struct btrfs_super_block *)fs_info->sb_buffer->b_data;
-	fs_info->disk_super = disk_super;
-	memcpy(&fs_info->super_copy, disk_super, sizeof(fs_info->super_copy));
 
+	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;
 
@@ -530,9 +512,9 @@ struct btrfs_root *open_ctree(struct super_block *sb)
 	return tree_root;
 
 fail_tree_root:
-	btrfs_block_release(tree_root, tree_root->node);
+	free_extent_buffer(tree_root->node);
 fail_sb_buffer:
-	btrfs_block_release(tree_root, fs_info->sb_buffer);
+	free_extent_buffer(fs_info->sb_buffer);
 fail_iput:
 	iput(fs_info->btree_inode);
 fail:
@@ -546,31 +528,13 @@ int write_ctree_super(struct btrfs_trans_handle *trans, struct btrfs_root
 		      *root)
 {
 	int ret;
-	struct buffer_head *bh = root->fs_info->sb_buffer;
-
-	lock_buffer(bh);
-	WARN_ON(atomic_read(&bh->b_count) < 1);
-	clear_buffer_dirty(bh);
-	csum_tree_block(root, bh, 0);
-	bh->b_end_io = end_buffer_write_sync;
-	get_bh(bh);
-	if (root->fs_info->do_barriers)
-		ret = submit_bh(WRITE_BARRIER, bh);
-	else
-		ret = submit_bh(WRITE, bh);
-	if (ret == -EOPNOTSUPP) {
-		get_bh(bh);
-		lock_buffer(bh);
-		set_buffer_uptodate(bh);
-		root->fs_info->do_barriers = 0;
-		ret = submit_bh(WRITE, bh);
-	}
-	wait_on_buffer(bh);
-	if (!buffer_uptodate(bh)) {
-		WARN_ON(1);
-		return -EIO;
-	}
-	return 0;
+	struct extent_buffer *super = root->fs_info->sb_buffer;
+	struct inode *btree_inode = root->fs_info->btree_inode;
+
+	set_extent_buffer_dirty(&BTRFS_I(btree_inode)->extent_tree, super);
+	ret = sync_page_range_nolock(btree_inode, btree_inode->i_mapping,
+				     super->start, super->len);
+	return ret;
 }
 
 int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
@@ -581,9 +545,9 @@ int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
 	if (root->inode)
 		iput(root->inode);
 	if (root->node)
-		brelse(root->node);
+		free_extent_buffer(root->node);
 	if (root->commit_root)
-		brelse(root->commit_root);
+		free_extent_buffer(root->commit_root);
 	if (root->name)
 		kfree(root->name);
 	kfree(root);
@@ -629,12 +593,10 @@ int close_ctree(struct btrfs_root *root)
 	mutex_unlock(&fs_info->fs_mutex);
 
 	if (fs_info->extent_root->node)
-		btrfs_block_release(fs_info->extent_root,
-				    fs_info->extent_root->node);
+		free_extent_buffer(fs_info->extent_root->node);
 	if (fs_info->tree_root->node)
-		btrfs_block_release(fs_info->tree_root,
-				    fs_info->tree_root->node);
-	btrfs_block_release(root, fs_info->sb_buffer);
+		free_extent_buffer(fs_info->tree_root->node);
+	free_extent_buffer(fs_info->sb_buffer);
 	truncate_inode_pages(fs_info->btree_inode->i_mapping, 0);
 	iput(fs_info->btree_inode);
 
@@ -645,25 +607,32 @@ int close_ctree(struct btrfs_root *root)
 	return 0;
 }
 
-void btrfs_mark_buffer_dirty(struct buffer_head *bh)
+int btrfs_buffer_uptodate(struct extent_buffer *buf)
+{
+	struct inode *btree_inode = buf->pages[0]->mapping->host;
+	return extent_buffer_uptodate(&BTRFS_I(btree_inode)->extent_tree, buf);
+}
+
+int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
 {
-	struct btrfs_root *root = BTRFS_I(bh->b_page->mapping->host)->root;
-	u64 transid = btrfs_header_generation(btrfs_buffer_header(bh));
+	struct inode *btree_inode = buf->pages[0]->mapping->host;
+	return set_extent_buffer_uptodate(&BTRFS_I(btree_inode)->extent_tree,
+					  buf);
+}
 
-	WARN_ON(!atomic_read(&bh->b_count));
+void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
+{
+	struct btrfs_root *root = BTRFS_I(buf->pages[0]->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)bh->b_blocknr,
+			(unsigned long long)extent_buffer_blocknr(buf),
 			transid, root->fs_info->generation);
 		WARN_ON(1);
 	}
-	mark_buffer_dirty(bh);
-}
-
-void btrfs_block_release(struct btrfs_root *root, struct buffer_head *buf)
-{
-	brelse(buf);
+	set_extent_buffer_dirty(&BTRFS_I(btree_inode)->extent_tree, buf);
 }
 
 void btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr)

fs/btrfs/disk-io.h

@@ -19,68 +19,35 @@
 #ifndef __DISKIO__
 #define __DISKIO__
 
-#include <linux/buffer_head.h>
-
 #define BTRFS_SUPER_INFO_OFFSET (16 * 1024)
 
-enum btrfs_bh_state_bits {
-	BH_Checked = BH_PrivateStart,
-	BH_Defrag,
-	BH_DefragDone,
-};
-BUFFER_FNS(Checked, checked);
-BUFFER_FNS(Defrag, defrag);
-BUFFER_FNS(DefragDone, defrag_done);
-
-static inline struct btrfs_node *btrfs_buffer_node(struct buffer_head *bh)
-{
-	return (struct btrfs_node *)bh->b_data;
-}
-
-static inline struct btrfs_leaf *btrfs_buffer_leaf(struct buffer_head *bh)
-{
-	return (struct btrfs_leaf *)bh->b_data;
-}
-
-static inline struct btrfs_header *btrfs_buffer_header(struct buffer_head *bh)
-{
-	return &((struct btrfs_node *)bh->b_data)->header;
-}
-
-struct buffer_head *read_tree_block(struct btrfs_root *root, u64 blocknr);
+struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 blocknr);
 int readahead_tree_block(struct btrfs_root *root, u64 blocknr);
-struct buffer_head *btrfs_find_create_tree_block(struct btrfs_root *root,
-						 u64 blocknr);
-int write_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
-		     struct buffer_head *buf);
-int dirty_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
-		     struct buffer_head *buf);
+struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
+						   u64 blocknr);
 int clean_tree_block(struct btrfs_trans_handle *trans,
-		     struct btrfs_root *root, struct buffer_head *buf);
-int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
-			     struct btrfs_root *root);
+		     struct btrfs_root *root, struct extent_buffer *buf);
 struct btrfs_root *open_ctree(struct super_block *sb);
 int close_ctree(struct btrfs_root *root);
-void btrfs_block_release(struct btrfs_root *root, struct buffer_head *buf);
 int write_ctree_super(struct btrfs_trans_handle *trans,
 		      struct btrfs_root *root);
-struct buffer_head *btrfs_find_tree_block(struct btrfs_root *root, u64 blocknr);
-int btrfs_csum_data(struct btrfs_root * root, char *data, size_t len,
-		    char *result);
+struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
+					    u64 blocknr);
 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 *btrfs_read_fs_root_no_radix(struct btrfs_fs_info *fs_info,
 					       struct btrfs_key *location);
-u64 bh_blocknr(struct buffer_head *bh);
 int btrfs_insert_dev_radix(struct btrfs_root *root,
 			   struct block_device *bdev,
 			   u64 device_id,
 			   u64 block_start,
 			   u64 num_blocks);
-int btrfs_map_bh_to_logical(struct btrfs_root *root, struct buffer_head *bh,
-			     u64 logical);
 void btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr);
 int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root);
-void btrfs_mark_buffer_dirty(struct buffer_head *bh);
+void btrfs_mark_buffer_dirty(struct extent_buffer *buf);
+int btrfs_buffer_uptodate(struct extent_buffer *buf);
+int btrfs_set_buffer_uptodate(struct extent_buffer *buf);
+int wait_on_tree_block_writeback(struct btrfs_root *root,
+				 struct extent_buffer *buf);
 #endif

fs/btrfs/extent_map.h

@@ -3,9 +3,20 @@
 
 #include <linux/rbtree.h>
 
+#define EXTENT_MAP_HOLE (u64)-3
 #define EXTENT_MAP_INLINE (u64)-2
 #define EXTENT_MAP_DELALLOC (u64)-1
 
+/* bits for the extent state */
+#define EXTENT_DIRTY 1
+#define EXTENT_WRITEBACK (1 << 1)
+#define EXTENT_UPTODATE (1 << 2)
+#define EXTENT_LOCKED (1 << 3)
+#define EXTENT_NEW (1 << 4)
+#define EXTENT_DELALLOC (1 << 5)
+#define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK)
+
+
 struct extent_map_ops {
 	int (*fill_delalloc)(struct inode *inode, u64 start, u64 end);
 	int (*writepage_io_hook)(struct page *page, u64 start, u64 end);
@@ -51,10 +62,13 @@ struct extent_state {
 	struct list_head list;
 };
 
+#define EXTENT_BUFFER_SIZE(nr) (sizeof(struct extent_buffer) + \
+			       (nr - 1) * sizeof(struct page *))
 struct extent_buffer {
 	u64 start;
-	u64 end; /* inclusive */
-	char *addr;
+	unsigned long len;
+	atomic_t refs;
+	int flags;
 	struct page *pages[];
 };
 
@@ -87,8 +101,12 @@ int set_extent_new(struct extent_map_tree *tree, u64 start, u64 end,
 		   gfp_t mask);
 int set_extent_dirty(struct extent_map_tree *tree, u64 start, u64 end,
 		     gfp_t mask);
+int clear_extent_dirty(struct extent_map_tree *tree, u64 start, u64 end,
+		       gfp_t mask);
 int set_extent_delalloc(struct extent_map_tree *tree, u64 start, u64 end,
 		     gfp_t mask);
+int find_first_extent_bit(struct extent_map_tree *tree, u64 start,
+			  u64 *start_ret, u64 *end_ret, int bits);
 int extent_invalidatepage(struct extent_map_tree *tree,
 			  struct page *page, unsigned long offset);
 int extent_write_full_page(struct extent_map_tree *tree, struct page *page,
@@ -106,4 +124,57 @@ int set_range_dirty(struct extent_map_tree *tree, u64 start, u64 end);
 int set_state_private(struct extent_map_tree *tree, u64 start, u64 private);
 int get_state_private(struct extent_map_tree *tree, u64 start, u64 *private);
 void set_page_extent_mapped(struct page *page);
+
+struct extent_buffer *alloc_extent_buffer(struct extent_map_tree *tree,
+					  u64 start, unsigned long len,
+					  gfp_t mask);
+struct extent_buffer *find_extent_buffer(struct extent_map_tree *tree,
+					 u64 start, unsigned long len,
+					  gfp_t mask);
+void free_extent_buffer(struct extent_buffer *eb);
+int read_extent_buffer_pages(struct extent_map_tree *tree,
+			     struct extent_buffer *eb, int wait);
+
+static inline void extent_buffer_get(struct extent_buffer *eb)
+{
+	atomic_inc(&eb->refs);
+}
+
+static inline u64 extent_buffer_blocknr(struct extent_buffer *eb)
+{
+	return eb->start / 4096;
+}
+
+int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv,
+			  unsigned long start,
+			  unsigned long len);
+void read_extent_buffer(struct extent_buffer *eb, void *dst,
+			unsigned long start,
+			unsigned long len);
+void write_extent_buffer(struct extent_buffer *eb, const void *src,
+			 unsigned long start, unsigned long len);
+void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
+			unsigned long dst_offset, unsigned long src_offset,
+			unsigned long len);
+void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
+			   unsigned long src_offset, unsigned long len);
+void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
+			   unsigned long src_offset, unsigned long len);
+void memset_extent_buffer(struct extent_buffer *eb, char c,
+			  unsigned long start, unsigned long len);
+int wait_on_extent_buffer_writeback(struct extent_map_tree *tree,
+				    struct extent_buffer *eb);
+int clear_extent_buffer_dirty(struct extent_map_tree *tree,
+			      struct extent_buffer *eb);
+int set_extent_buffer_dirty(struct extent_map_tree *tree,
+			     struct extent_buffer *eb);
+int set_extent_buffer_uptodate(struct extent_map_tree *tree,
+			       struct extent_buffer *eb);
+int extent_buffer_uptodate(struct extent_map_tree *tree,
+			   struct extent_buffer *eb);
+int map_extent_buffer(struct extent_buffer *eb, unsigned long offset,
+		      char **token, char **map,
+		      unsigned long *map_start,
+		      unsigned long *map_len, int km);
+void unmap_extent_buffer(struct extent_buffer *eb, char *token, int km);
 #endif

fs/btrfs/file-item.c

@@ -34,12 +34,12 @@ int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
 	struct btrfs_file_extent_item *item;
 	struct btrfs_key file_key;
 	struct btrfs_path *path;
+	struct extent_buffer *leaf;
 
 	path = btrfs_alloc_path();
 	BUG_ON(!path);
 	file_key.objectid = objectid;
 	file_key.offset = pos;
-	file_key.flags = 0;
 	btrfs_set_key_type(&file_key, BTRFS_EXTENT_DATA_KEY);
 
 	ret = btrfs_insert_empty_item(trans, root, path, &file_key,
@@ -47,15 +47,16 @@ int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
 	if (ret < 0)
 		goto out;
 	BUG_ON(ret);
-	item = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), path->slots[0],
+	leaf = path->nodes[0];
+	item = btrfs_item_ptr(leaf, path->slots[0],
 			      struct btrfs_file_extent_item);
-	btrfs_set_file_extent_disk_blocknr(item, offset);
-	btrfs_set_file_extent_disk_num_blocks(item, disk_num_blocks);
-	btrfs_set_file_extent_offset(item, 0);
-	btrfs_set_file_extent_num_blocks(item, num_blocks);
-	btrfs_set_file_extent_generation(item, trans->transid);
-	btrfs_set_file_extent_type(item, BTRFS_FILE_EXTENT_REG);
-	btrfs_mark_buffer_dirty(path->nodes[0]);
+	btrfs_set_file_extent_disk_blocknr(leaf, item, offset);
+	btrfs_set_file_extent_disk_num_blocks(leaf, item, disk_num_blocks);
+	btrfs_set_file_extent_offset(leaf, item, 0);
+	btrfs_set_file_extent_num_blocks(leaf, item, num_blocks);
+	btrfs_set_file_extent_generation(leaf, item, trans->transid);
+	btrfs_set_file_extent_type(leaf, item, BTRFS_FILE_EXTENT_REG);
+	btrfs_mark_buffer_dirty(leaf);
 out:
 	btrfs_free_path(path);
 	return ret;
@@ -71,32 +72,30 @@ struct btrfs_csum_item *btrfs_lookup_csum(struct btrfs_trans_handle *trans,
 	struct btrfs_key file_key;
 	struct btrfs_key found_key;
 	struct btrfs_csum_item *item;
-	struct btrfs_leaf *leaf;
+	struct extent_buffer *leaf;
 	u64 csum_offset = 0;
 	int csums_in_item;
 
 	file_key.objectid = objectid;
 	file_key.offset = offset;
-	file_key.flags = 0;
 	btrfs_set_key_type(&file_key, BTRFS_CSUM_ITEM_KEY);
 	ret = btrfs_search_slot(trans, root, &file_key, path, 0, cow);
 	if (ret < 0)
 		goto fail;
-	leaf = btrfs_buffer_leaf(path->nodes[0]);
+	leaf = path->nodes[0];
 	if (ret > 0) {
 		ret = 1;
 		if (path->slots[0] == 0)
 			goto fail;
 		path->slots[0]--;
-		btrfs_disk_key_to_cpu(&found_key,
-				      &leaf->items[path->slots[0]].key);
+		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
 		if (btrfs_key_type(&found_key) != BTRFS_CSUM_ITEM_KEY ||
 		    found_key.objectid != objectid) {
 			goto fail;
 		}
 		csum_offset = (offset - found_key.offset) >>
 				root->fs_info->sb->s_blocksize_bits;
-		csums_in_item = btrfs_item_size(leaf->items + path->slots[0]);
+		csums_in_item = btrfs_item_size_nr(leaf, path->slots[0]);
 		csums_in_item /= BTRFS_CRC32_SIZE;
 
 		if (csum_offset >= csums_in_item) {
@@ -127,7 +126,6 @@ int btrfs_lookup_file_extent(struct btrfs_trans_handle *trans,
 
 	file_key.objectid = objectid;
 	file_key.offset = offset;
-	file_key.flags = 0;
 	btrfs_set_key_type(&file_key, BTRFS_EXTENT_DATA_KEY);
 	ret = btrfs_search_slot(trans, root, &file_key, path, ins_len, cow);
 	return ret;
@@ -138,12 +136,14 @@ int btrfs_csum_file_block(struct btrfs_trans_handle *trans,
 			  u64 objectid, u64 offset,
 			  char *data, size_t len)
 {
+	return 0;
+#if 0
 	int ret;
 	struct btrfs_key file_key;
 	struct btrfs_key found_key;
 	struct btrfs_path *path;
 	struct btrfs_csum_item *item;
-	struct btrfs_leaf *leaf;
+	struct extent_buffer *leaf;
 	u64 csum_offset;
 
 	path = btrfs_alloc_path();
@@ -161,8 +161,8 @@ int btrfs_csum_file_block(struct btrfs_trans_handle *trans,
 	if (ret == -EFBIG) {
 		u32 item_size;
 		/* we found one, but it isn't big enough yet */
-		leaf = btrfs_buffer_leaf(path->nodes[0]);
-		item_size = btrfs_item_size(leaf->items + path->slots[0]);
+		leaf = path->nodes[0];
+		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
 		if ((item_size / BTRFS_CRC32_SIZE) >= MAX_CSUM_ITEMS(root)) {
 			/* already at max size, make a new one */
 			goto insert;
@@ -188,8 +188,8 @@ int btrfs_csum_file_block(struct btrfs_trans_handle *trans,
 		goto insert;
 	}
 	path->slots[0]--;
-	leaf = btrfs_buffer_leaf(path->nodes[0]);
-	btrfs_disk_key_to_cpu(&found_key, &leaf->items[path->slots[0]].key);
+	leaf = path->nodes[0];
+	btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
 	csum_offset = (offset - found_key.offset) >>
 			root->fs_info->sb->s_blocksize_bits;
 	if (btrfs_key_type(&found_key) != BTRFS_CSUM_ITEM_KEY ||
@@ -197,10 +197,10 @@ int btrfs_csum_file_block(struct btrfs_trans_handle *trans,
 	    csum_offset >= MAX_CSUM_ITEMS(root)) {
 		goto insert;
 	}
-	if (csum_offset >= btrfs_item_size(leaf->items + path->slots[0]) /
+	if (csum_offset >= btrfs_item_size_nr(leaf, path->slots[0]) /
 	    BTRFS_CRC32_SIZE) {
 		u32 diff = (csum_offset + 1) * BTRFS_CRC32_SIZE;
-		diff = diff - btrfs_item_size(leaf->items + path->slots[0]);
+		diff = diff - btrfs_item_size_nr(leaf, path->slots[0]);
 		if (diff != BTRFS_CRC32_SIZE)
 			goto insert;
 		ret = btrfs_extend_item(trans, root, path, diff);
@@ -220,21 +220,20 @@ int btrfs_csum_file_block(struct btrfs_trans_handle *trans,
 		goto fail;
 	}
 csum:
-	item = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), path->slots[0],
-			      struct btrfs_csum_item);
+	leaf = path->nodes[0];
+	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_csum_item);
 	ret = 0;
 	item = (struct btrfs_csum_item *)((unsigned char *)item +
 					  csum_offset * BTRFS_CRC32_SIZE);
 found:
-	btrfs_check_bounds(&item->csum, BTRFS_CRC32_SIZE,
-			   path->nodes[0]->b_data,
-			   root->fs_info->sb->s_blocksize);
+	/* FIXME!!!!!!!!!!!! */
 	ret = btrfs_csum_data(root, data, len, &item->csum);
 	btrfs_mark_buffer_dirty(path->nodes[0]);
 fail:
 	btrfs_release_path(root, path);
 	btrfs_free_path(path);
 	return ret;
+#endif
 }
 
 int btrfs_csum_truncate(struct btrfs_trans_handle *trans,
@@ -242,21 +241,21 @@ int btrfs_csum_truncate(struct btrfs_trans_handle *trans,
 			u64 isize)
 {
 	struct btrfs_key key;
-	struct btrfs_leaf *leaf = btrfs_buffer_leaf(path->nodes[0]);
+	struct extent_buffer *leaf = path->nodes[0];
 	int slot = path->slots[0];
 	int ret;
 	u32 new_item_size;
 	u64 new_item_span;
 	u64 blocks;
 
-	btrfs_disk_key_to_cpu(&key, &leaf->items[slot].key);
+	btrfs_item_key_to_cpu(leaf, &key, slot);
 	if (isize <= key.offset)
 		return 0;
 	new_item_span = isize - key.offset;
-	blocks = (new_item_span + root->blocksize - 1) >>
+	blocks = (new_item_span + root->sectorsize - 1) >>
 		root->fs_info->sb->s_blocksize_bits;
 	new_item_size = blocks * BTRFS_CRC32_SIZE;
-	if (new_item_size >= btrfs_item_size(leaf->items + slot))
+	if (new_item_size >= btrfs_item_size_nr(leaf, slot))
 		return 0;
 	ret = btrfs_truncate_item(trans, root, path, new_item_size);
 	BUG_ON(ret);

fs/btrfs/inode-item.c

@@ -20,24 +20,18 @@
 #include "disk-io.h"
 #include "transaction.h"
 
-int btrfs_insert_inode(struct btrfs_trans_handle *trans, struct btrfs_root
-		       *root, u64 objectid, struct btrfs_inode_item
-		       *inode_item)
+int btrfs_insert_empty_inode(struct btrfs_trans_handle *trans,
+			     struct btrfs_root *root,
+			     struct btrfs_path *path, u64 objectid)
 {
-	struct btrfs_path *path;
 	struct btrfs_key key;
 	int ret;
 	key.objectid = objectid;
-	key.flags = 0;
 	btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY);
 	key.offset = 0;
 
-	path = btrfs_alloc_path();
-	BUG_ON(!path);
-	ret = btrfs_insert_item(trans, root, &key, inode_item,
-				sizeof(*inode_item));
-	btrfs_release_path(root, path);
-	btrfs_free_path(path);
+	ret = btrfs_insert_empty_item(trans, root, path, &key,
+				      sizeof(struct btrfs_inode_item));
 	if (ret == 0 && objectid > root->highest_inode)
 		root->highest_inode = objectid;
 	return ret;
@@ -51,15 +45,15 @@ int btrfs_lookup_inode(struct btrfs_trans_handle *trans, struct btrfs_root
 	int cow = mod != 0;
 	int ret;
 	int slot;
-	struct btrfs_leaf *leaf;
+	struct extent_buffer *leaf;
 	struct btrfs_key found_key;
 
 	ret = btrfs_search_slot(trans, root, location, path, ins_len, cow);
 	if (ret > 0 && btrfs_key_type(location) == BTRFS_ROOT_ITEM_KEY &&
 	    location->offset == (u64)-1 && path->slots[0] != 0) {
 		slot = path->slots[0] - 1;
-		leaf = btrfs_buffer_leaf(path->nodes[0]);
-		btrfs_disk_key_to_cpu(&found_key, &leaf->items[slot].key);
+		leaf = path->nodes[0];
+		btrfs_item_key_to_cpu(leaf, &found_key, slot);
 		if (found_key.objectid == location->objectid &&
 		    btrfs_key_type(&found_key) == btrfs_key_type(location)) {
 			path->slots[0]--;

fs/btrfs/inode-map.c

@@ -24,8 +24,9 @@ int btrfs_find_highest_inode(struct btrfs_root *root, u64 *objectid)
 {
 	struct btrfs_path *path;
 	int ret;
-	struct btrfs_leaf *l;
+	struct extent_buffer *l;
 	struct btrfs_key search_key;
+	struct btrfs_key found_key;
 	int slot;
 
 	path = btrfs_alloc_path();
@@ -39,8 +40,9 @@ int btrfs_find_highest_inode(struct btrfs_root *root, u64 *objectid)
 	BUG_ON(ret == 0);
 	if (path->slots[0] > 0) {
 		slot = path->slots[0] - 1;
-		l = btrfs_buffer_leaf(path->nodes[0]);
-		*objectid = btrfs_disk_key_objectid(&l->items[slot].key);
+		l = path->nodes[0];
+		btrfs_item_key_to_cpu(l, &found_key, slot);
+		*objectid = found_key.objectid;
 	} else {
 		*objectid = BTRFS_FIRST_FREE_OBJECTID;
 	}
@@ -64,13 +66,12 @@ int btrfs_find_free_objectid(struct btrfs_trans_handle *trans,
 	int slot = 0;
 	u64 last_ino = 0;
 	int start_found;
-	struct btrfs_leaf *l;
+	struct extent_buffer *l;
 	struct btrfs_key search_key;
 	u64 search_start = dirid;
 
 	path = btrfs_alloc_path();
 	BUG_ON(!path);
-	search_key.flags = 0;
 	search_start = root->last_inode_alloc;
 	search_start = max(search_start, BTRFS_FIRST_FREE_OBJECTID);
 	search_key.objectid = search_start;
@@ -86,9 +87,9 @@ int btrfs_find_free_objectid(struct btrfs_trans_handle *trans,
 		path->slots[0]--;
 
 	while (1) {
-		l = btrfs_buffer_leaf(path->nodes[0]);
+		l = path->nodes[0];
 		slot = path->slots[0];
-		if (slot >= btrfs_header_nritems(&l->header)) {
+		if (slot >= btrfs_header_nritems(l)) {
 			ret = btrfs_next_leaf(root, path);
 			if (ret == 0)
 				continue;
@@ -103,7 +104,7 @@ int btrfs_find_free_objectid(struct btrfs_trans_handle *trans,
 				last_ino : search_start;
 			goto found;
 		}
-		btrfs_disk_key_to_cpu(&key, &l->items[slot].key);
+		btrfs_item_key_to_cpu(l, &key, slot);
 		if (key.objectid >= search_start) {
 			if (start_found) {
 				if (last_ino < search_start)

fs/btrfs/print-tree.h

@@ -18,6 +18,6 @@
 
 #ifndef __PRINT_TREE_
 #define __PRINT_TREE_
-void btrfs_print_leaf(struct btrfs_root *root, struct btrfs_leaf *l);
-void btrfs_print_tree(struct btrfs_root *root, struct buffer_head *t);
+void btrfs_print_leaf(struct btrfs_root *root, struct extent_buffer *l);
+void btrfs_print_tree(struct btrfs_root *root, struct extent_buffer *t);
 #endif

fs/btrfs/super.c

@@ -41,7 +41,7 @@
 #include "ioctl.h"
 #include "print-tree.h"
 
-#define BTRFS_SUPER_MAGIC 0x9123682E
+#define BTRFS_SUPER_MAGIC 0x9123683E
 
 static struct super_operations btrfs_super_ops;
 
@@ -115,13 +115,12 @@ static int btrfs_fill_super(struct super_block * sb, void * data, int silent)
 		return -EIO;
 	}
 	sb->s_fs_info = tree_root;
-	disk_super = tree_root->fs_info->disk_super;
+	disk_super = &tree_root->fs_info->super_copy;
 	inode = btrfs_iget_locked(sb, btrfs_super_root_dir(disk_super),
 				  tree_root);
 	bi = BTRFS_I(inode);
 	bi->location.objectid = inode->i_ino;
 	bi->location.offset = 0;
-	bi->location.flags = 0;
 	bi->root = tree_root;
 
 	btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
@@ -281,6 +280,7 @@ int btrfs_get_sb_bdev(struct file_system_type *fs_type,
 error_bdev:
 	close_bdev_excl(bdev);
 error:
+printk("get_sb failed\n");
 	return error;
 }
 /* end copy & paste */
@@ -295,6 +295,7 @@ static int btrfs_get_sb(struct file_system_type *fs_type,
 	ret = btrfs_get_sb_bdev(fs_type, flags, dev_name, data,
 			btrfs_fill_super, mnt,
 			subvol_name ? subvol_name : "default");
+printk("btrfs_get_sb returns %d\n", ret);
 	return ret;
 }
 

fs/btrfs/sysfs.c

@@ -31,31 +31,31 @@
 static ssize_t root_blocks_used_show(struct btrfs_root *root, char *buf)
 {
 	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_root_blocks_used(&root->root_item));
+		(unsigned long long)btrfs_root_used(&root->root_item));
 }
 
 static ssize_t root_block_limit_show(struct btrfs_root *root, char *buf)
 {
 	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_root_block_limit(&root->root_item));
+		(unsigned long long)btrfs_root_limit(&root->root_item));
 }
 
 static ssize_t super_blocks_used_show(struct btrfs_fs_info *fs, char *buf)
 {
 	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_super_blocks_used(fs->disk_super));
+		(unsigned long long)btrfs_super_blocks_used(&fs->super_copy));
 }
 
 static ssize_t super_total_blocks_show(struct btrfs_fs_info *fs, char *buf)
 {
 	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_super_total_blocks(fs->disk_super));
+		(unsigned long long)btrfs_super_total_blocks(&fs->super_copy));
 }
 
 static ssize_t super_blocksize_show(struct btrfs_fs_info *fs, char *buf)
 {
 	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_super_blocksize(fs->disk_super));
+		(unsigned long long)btrfs_super_sectorsize(&fs->super_copy));
 }
 
 /* this is for root attrs (subvols/snapshots) */

fs/btrfs/transaction.h

@@ -28,7 +28,7 @@ struct btrfs_transaction {
 	int use_count;
 	int commit_done;
 	struct list_head list;
-	struct radix_tree_root dirty_pages;
+	struct extent_map_tree dirty_pages;
 	unsigned long start_time;
 	wait_queue_head_t writer_wait;
 	wait_queue_head_t commit_wait;
@@ -83,5 +83,6 @@ int btrfs_add_dead_root(struct btrfs_root *root, struct btrfs_root *latest,
 int btrfs_defrag_dirty_roots(struct btrfs_fs_info *info);
 int btrfs_defrag_root(struct btrfs_root *root, int cacheonly);
 int btrfs_clean_old_snapshots(struct btrfs_root *root);
-
+int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
+			     struct btrfs_root *root);
 #endif