Btrfs: Split the extent_map code into two parts

There is now extent_map for mapping offsets in the file to disk and
extent_io for state tracking, IO submission and extent_bufers.

The new extent_map code shifts from [start,end] pairs to [start,len], and
pushes the locking out into the caller.  This allows a few performance
optimizations and is easier to use.

A number of extent_map usage bugs were fixed, mostly with failing
to remove extent_map entries when changing the file.
Signed-off-by: Chris Mason <chris.mason@oracle.com>

fs/btrfs/Makefile

@@ -5,7 +5,8 @@ obj-m  := btrfs.o
 btrfs-y := super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \
 	   hash.o file-item.o inode-item.o inode-map.o disk-io.o \
 	   transaction.o bit-radix.o inode.o file.o tree-defrag.o \
-	   extent_map.o sysfs.o struct-funcs.o xattr.o ordered-data.o
+	   extent_map.o sysfs.o struct-funcs.o xattr.o ordered-data.o \
+	   extent_io.o
 
 ifeq ($(CONFIG_FS_POSIX_ACL),y)
 btrfs-y += acl.o

fs/btrfs/btrfs_inode.h

@@ -20,6 +20,7 @@
 #define __BTRFS_I__
 
 #include "extent_map.h"
+#include "extent_io.h"
 
 /* in memory btrfs inode */
 struct btrfs_inode {
@@ -27,6 +28,7 @@ struct btrfs_inode {
 	struct btrfs_block_group_cache *block_group;
 	struct btrfs_key location;
 	struct extent_map_tree extent_tree;
+	struct extent_io_tree io_tree;
 	struct inode vfs_inode;
 
 	u64 ordered_trans;

fs/btrfs/ctree.h

@@ -27,6 +27,7 @@
 #include <linux/completion.h>
 #include <asm/kmap_types.h>
 #include "bit-radix.h"
+#include "extent_io.h"
 #include "extent_map.h"
 
 struct btrfs_trans_handle;
@@ -314,11 +315,11 @@ struct btrfs_fs_info {
 	struct btrfs_root *tree_root;
 	struct radix_tree_root fs_roots_radix;
 
-	struct extent_map_tree free_space_cache;
-	struct extent_map_tree block_group_cache;
-	struct extent_map_tree pinned_extents;
-	struct extent_map_tree pending_del;
-	struct extent_map_tree extent_ins;
+	struct extent_io_tree free_space_cache;
+	struct extent_io_tree block_group_cache;
+	struct extent_io_tree pinned_extents;
+	struct extent_io_tree pending_del;
+	struct extent_io_tree extent_ins;
 
 	u64 generation;
 	u64 last_trans_committed;
@@ -956,7 +957,7 @@ u32 btrfs_count_snapshots_in_path(struct btrfs_root *root,
 				  u64 first_extent);
 int btrfs_extent_post_op(struct btrfs_trans_handle *trans,
 			 struct btrfs_root *root);
-int btrfs_copy_pinned(struct btrfs_root *root, struct extent_map_tree *copy);
+int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy);
 struct btrfs_block_group_cache *btrfs_lookup_block_group(struct
 							 btrfs_fs_info *info,
 							 u64 bytenr);
@@ -1001,7 +1002,7 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root
 		      u64 owner_objectid, u64 owner_offset, int pin);
 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root,
-			       struct extent_map_tree *unpin);
+			       struct extent_io_tree *unpin);
 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
 				u64 bytenr, u64 num_bytes,

fs/btrfs/extent-tree.c

@@ -63,7 +63,7 @@ static int cache_block_group(struct btrfs_root *root,
 	int ret;
 	struct btrfs_key key;
 	struct extent_buffer *leaf;
-	struct extent_map_tree *free_space_cache;
+	struct extent_io_tree *free_space_cache;
 	int slot;
 	u64 last = 0;
 	u64 hole_size;
@@ -158,7 +158,7 @@ struct btrfs_block_group_cache *btrfs_lookup_block_group(struct
 							 btrfs_fs_info *info,
 							 u64 bytenr)
 {
-	struct extent_map_tree *block_group_cache;
+	struct extent_io_tree *block_group_cache;
 	struct btrfs_block_group_cache *block_group = NULL;
 	u64 ptr;
 	u64 start;
@@ -281,7 +281,7 @@ struct btrfs_block_group_cache *btrfs_find_block_group(struct btrfs_root *root,
 						 int data, int owner)
 {
 	struct btrfs_block_group_cache *cache;
-	struct extent_map_tree *block_group_cache;
+	struct extent_io_tree *block_group_cache;
 	struct btrfs_block_group_cache *found_group = NULL;
 	struct btrfs_fs_info *info = root->fs_info;
 	u64 used;
@@ -951,7 +951,7 @@ static int write_one_cache_group(struct btrfs_trans_handle *trans,
 int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
 				   struct btrfs_root *root)
 {
-	struct extent_map_tree *block_group_cache;
+	struct extent_io_tree *block_group_cache;
 	struct btrfs_block_group_cache *cache;
 	int ret;
 	int err = 0;
@@ -1107,12 +1107,12 @@ static int update_pinned_extents(struct btrfs_root *root,
 	return 0;
 }
 
-int btrfs_copy_pinned(struct btrfs_root *root, struct extent_map_tree *copy)
+int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy)
 {
 	u64 last = 0;
 	u64 start;
 	u64 end;
-	struct extent_map_tree *pinned_extents = &root->fs_info->pinned_extents;
+	struct extent_io_tree *pinned_extents = &root->fs_info->pinned_extents;
 	int ret;
 
 	while(1) {
@@ -1128,12 +1128,12 @@ int btrfs_copy_pinned(struct btrfs_root *root, struct extent_map_tree *copy)
 
 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root,
-			       struct extent_map_tree *unpin)
+			       struct extent_io_tree *unpin)
 {
 	u64 start;
 	u64 end;
 	int ret;
-	struct extent_map_tree *free_space_cache;
+	struct extent_io_tree *free_space_cache;
 	free_space_cache = &root->fs_info->free_space_cache;
 
 	while(1) {
@@ -1329,8 +1329,8 @@ static int del_pending_extents(struct btrfs_trans_handle *trans, struct
 	int err = 0;
 	u64 start;
 	u64 end;
-	struct extent_map_tree *pending_del;
-	struct extent_map_tree *pinned_extents;
+	struct extent_io_tree *pending_del;
+	struct extent_io_tree *pinned_extents;
 
 	pending_del = &extent_root->fs_info->pending_del;
 	pinned_extents = &extent_root->fs_info->pinned_extents;
@@ -1802,7 +1802,7 @@ struct extent_buffer *__btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 
 	set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
 			 buf->start + buf->len - 1, GFP_NOFS);
-	set_extent_bits(&BTRFS_I(root->fs_info->btree_inode)->extent_tree,
+	set_extent_bits(&BTRFS_I(root->fs_info->btree_inode)->io_tree,
 			buf->start, buf->start + buf->len - 1,
 			EXTENT_CSUM, GFP_NOFS);
 	buf->flags |= EXTENT_CSUM;
@@ -2166,7 +2166,7 @@ static int noinline relocate_inode_pages(struct inode *inode, u64 start,
 	unsigned long i;
 	struct page *page;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct file_ra_state *ra;
 
 	ra = kzalloc(sizeof(*ra), GFP_NOFS);
@@ -2195,15 +2195,14 @@ static int noinline relocate_inode_pages(struct inode *inode, u64 start,
 		page_start = (u64)page->index << PAGE_CACHE_SHIFT;
 		page_end = page_start + PAGE_CACHE_SIZE - 1;
 
-		lock_extent(em_tree, page_start, page_end, GFP_NOFS);
+		lock_extent(io_tree, page_start, page_end, GFP_NOFS);
 
 		delalloc_start = page_start;
-		existing_delalloc =
-			count_range_bits(&BTRFS_I(inode)->extent_tree,
-					 &delalloc_start, page_end,
-					 PAGE_CACHE_SIZE, EXTENT_DELALLOC);
+		existing_delalloc = count_range_bits(io_tree,
+					     &delalloc_start, page_end,
+					     PAGE_CACHE_SIZE, EXTENT_DELALLOC);
 
-		set_extent_delalloc(em_tree, page_start,
+		set_extent_delalloc(io_tree, page_start,
 				    page_end, GFP_NOFS);
 
 		spin_lock(&root->fs_info->delalloc_lock);
@@ -2211,7 +2210,7 @@ static int noinline relocate_inode_pages(struct inode *inode, u64 start,
 						 existing_delalloc;
 		spin_unlock(&root->fs_info->delalloc_lock);
 
-		unlock_extent(em_tree, page_start, page_end, GFP_NOFS);
+		unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
 		set_page_dirty(page);
 		unlock_page(page);
 		page_cache_release(page);
@@ -2379,7 +2378,7 @@ int btrfs_shrink_extent_tree(struct btrfs_root *root, u64 new_size)
 	u64 cur_byte;
 	u64 total_found;
 	struct btrfs_fs_info *info = root->fs_info;
-	struct extent_map_tree *block_group_cache;
+	struct extent_io_tree *block_group_cache;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
 	struct extent_buffer *leaf;
@@ -2561,7 +2560,7 @@ int btrfs_grow_extent_tree(struct btrfs_trans_handle *trans,
 	struct btrfs_block_group_cache *cache;
 	struct btrfs_block_group_item *item;
 	struct btrfs_fs_info *info = root->fs_info;
-	struct extent_map_tree *block_group_cache;
+	struct extent_io_tree *block_group_cache;
 	struct btrfs_key key;
 	struct extent_buffer *leaf;
 	int ret;
@@ -2645,7 +2644,7 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 	int bit;
 	struct btrfs_block_group_cache *cache;
 	struct btrfs_fs_info *info = root->fs_info;
-	struct extent_map_tree *block_group_cache;
+	struct extent_io_tree *block_group_cache;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
 	struct extent_buffer *leaf;

fs/btrfs/extent_io.h

+#ifndef __EXTENTIO__
+#define __EXTENTIO__
+
+#include <linux/rbtree.h>
+
+/* 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_DEFRAG (1 << 6)
+#define EXTENT_DEFRAG_DONE (1 << 7)
+#define EXTENT_BUFFER_FILLED (1 << 8)
+#define EXTENT_CSUM (1 << 9)
+#define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK)
+
+/*
+ * page->private values.  Every page that is controlled by the extent
+ * map has page->private set to one.
+ */
+#define EXTENT_PAGE_PRIVATE 1
+#define EXTENT_PAGE_PRIVATE_FIRST_PAGE 3
+
+struct extent_io_ops {
+	int (*fill_delalloc)(struct inode *inode, u64 start, u64 end);
+	int (*writepage_io_hook)(struct page *page, u64 start, u64 end);
+	int (*readpage_io_hook)(struct page *page, u64 start, u64 end);
+	int (*readpage_end_io_hook)(struct page *page, u64 start, u64 end);
+	void (*writepage_end_io_hook)(struct page *page, u64 start, u64 end);
+};
+
+struct extent_io_tree {
+	struct rb_root state;
+	struct address_space *mapping;
+	u64 dirty_bytes;
+	rwlock_t lock;
+	struct extent_io_ops *ops;
+	spinlock_t lru_lock;
+	struct list_head buffer_lru;
+	int lru_size;
+};
+
+struct extent_state {
+	u64 start;
+	u64 end; /* inclusive */
+	int in_tree;
+	struct rb_node rb_node;
+	wait_queue_head_t wq;
+	atomic_t refs;
+	unsigned long state;
+
+	/* for use by the FS */
+	u64 private;
+
+	struct list_head list;
+};
+
+struct extent_buffer {
+	u64 start;
+	unsigned long len;
+	char *map_token;
+	char *kaddr;
+	unsigned long map_start;
+	unsigned long map_len;
+	struct page *first_page;
+	struct list_head lru;
+	atomic_t refs;
+	int flags;
+};
+
+struct extent_map_tree;
+
+typedef struct extent_map *(get_extent_t)(struct inode *inode,
+					  struct page *page,
+					  size_t page_offset,
+					  u64 start, u64 len,
+					  int create);
+
+void extent_io_tree_init(struct extent_io_tree *tree,
+			  struct address_space *mapping, gfp_t mask);
+void extent_io_tree_empty_lru(struct extent_io_tree *tree);
+int try_release_extent_mapping(struct extent_map_tree *map,
+			       struct extent_io_tree *tree, struct page *page);
+int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask);
+int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask);
+int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
+			  get_extent_t *get_extent);
+int __init extent_io_init(void);
+void extent_io_exit(void);
+
+u64 count_range_bits(struct extent_io_tree *tree,
+		     u64 *start, u64 search_end,
+		     u64 max_bytes, unsigned long bits);
+
+int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
+		   int bits, int filled);
+int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
+		      int bits, gfp_t mask);
+int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
+		    int bits, gfp_t mask);
+int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
+			gfp_t mask);
+int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
+		   gfp_t mask);
+int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
+		     gfp_t mask);
+int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
+		       gfp_t mask);
+int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
+		     gfp_t mask);
+int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
+			  u64 *start_ret, u64 *end_ret, int bits);
+int extent_invalidatepage(struct extent_io_tree *tree,
+			  struct page *page, unsigned long offset);
+int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
+			  get_extent_t *get_extent,
+			  struct writeback_control *wbc);
+int extent_writepages(struct extent_io_tree *tree,
+		      struct address_space *mapping,
+		      get_extent_t *get_extent,
+		      struct writeback_control *wbc);
+int extent_readpages(struct extent_io_tree *tree,
+		     struct address_space *mapping,
+		     struct list_head *pages, unsigned nr_pages,
+		     get_extent_t get_extent);
+int extent_prepare_write(struct extent_io_tree *tree,
+			 struct inode *inode, struct page *page,
+			 unsigned from, unsigned to, get_extent_t *get_extent);
+int extent_commit_write(struct extent_io_tree *tree,
+			struct inode *inode, struct page *page,
+			unsigned from, unsigned to);
+sector_t extent_bmap(struct address_space *mapping, sector_t iblock,
+		get_extent_t *get_extent);
+int set_range_dirty(struct extent_io_tree *tree, u64 start, u64 end);
+int set_state_private(struct extent_io_tree *tree, u64 start, u64 private);
+int get_state_private(struct extent_io_tree *tree, u64 start, u64 *private);
+void set_page_extent_mapped(struct page *page);
+
+struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
+					  u64 start, unsigned long len,
+					  struct page *page0,
+					  gfp_t mask);
+struct extent_buffer *find_extent_buffer(struct extent_io_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_io_tree *tree,
+			     struct extent_buffer *eb, u64 start, int wait);
+
+static inline void extent_buffer_get(struct extent_buffer *eb)
+{
+	atomic_inc(&eb->refs);
+}
+
+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_io_tree *tree,
+				    struct extent_buffer *eb);
+int clear_extent_buffer_dirty(struct extent_io_tree *tree,
+			      struct extent_buffer *eb);
+int set_extent_buffer_dirty(struct extent_io_tree *tree,
+			     struct extent_buffer *eb);
+int set_extent_buffer_uptodate(struct extent_io_tree *tree,
+			       struct extent_buffer *eb);
+int extent_buffer_uptodate(struct extent_io_tree *tree,
+			   struct extent_buffer *eb);
+int map_extent_buffer(struct extent_buffer *eb, unsigned long offset,
+		      unsigned long min_len, char **token, char **map,
+		      unsigned long *map_start,
+		      unsigned long *map_len, int km);
+int map_private_extent_buffer(struct extent_buffer *eb, unsigned long offset,
+		      unsigned long min_len, 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/extent_map.h

@@ -3,215 +3,53 @@
 
 #include <linux/rbtree.h>
 
+#define EXTENT_MAP_LAST_BYTE (u64)-4
 #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_DEFRAG (1 << 6)
-#define EXTENT_DEFRAG_DONE (1 << 7)
-#define EXTENT_BUFFER_FILLED (1 << 8)
-#define EXTENT_CSUM (1 << 9)
-#define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK)
-
-/*
- * page->private values.  Every page that is controlled by the extent
- * map has page->private set to one.
- */
-#define EXTENT_PAGE_PRIVATE 1
-#define EXTENT_PAGE_PRIVATE_FIRST_PAGE 3
-
-
-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);
-	int (*readpage_io_hook)(struct page *page, u64 start, u64 end);
-	int (*readpage_end_io_hook)(struct page *page, u64 start, u64 end);
-	void (*writepage_end_io_hook)(struct page *page, u64 start, u64 end);
-};
-
-struct extent_map_tree {
-	struct rb_root map;
-	struct rb_root state;
-	struct address_space *mapping;
-	u64 dirty_bytes;
-	rwlock_t lock;
-	struct extent_map_ops *ops;
-	spinlock_t lru_lock;
-	struct list_head buffer_lru;
-	int lru_size;
-};
-
-/* note, this must start with the same fields as fs/extent_map.c:tree_entry */
 struct extent_map {
-	u64 start;
-	u64 end; /* inclusive */
-	int in_tree;
 	struct rb_node rb_node;
-	/* block_start and block_end are in bytes */
+
+	/* all of these are in bytes */
+	u64 start;
+	u64 len;
 	u64 block_start;
-	u64 block_end; /* inclusive */
+	unsigned long flags;
 	struct block_device *bdev;
 	atomic_t refs;
-};
-
-/* note, this must start with the same fields as fs/extent_map.c:tree_entry */
-struct extent_state {
-	u64 start;
-	u64 end; /* inclusive */
 	int in_tree;
-	struct rb_node rb_node;
-	wait_queue_head_t wq;
-	atomic_t refs;
-	unsigned long state;
-
-	/* for use by the FS */
-	u64 private;
-
-	struct list_head list;
 };
 
-struct extent_buffer {
-	u64 start;
-	unsigned long len;
-	char *map_token;
-	char *kaddr;
-	unsigned long map_start;
-	unsigned long map_len;
-	struct page *first_page;
-	struct list_head lru;
-	atomic_t refs;
-	int flags;
+struct extent_map_tree {
+	struct rb_root map;
+	struct extent_map *last;
+	spinlock_t lock;
 };
 
-typedef struct extent_map *(get_extent_t)(struct inode *inode,
-					  struct page *page,
-					  size_t page_offset,
-					  u64 start, u64 end,
-					  int create);
+static inline u64 extent_map_end(struct extent_map *em)
+{
+	if (em->start + em->len < em->start)
+		return (u64)-1;
+	return em->start + em->len;
+}
+
+static inline u64 extent_map_block_end(struct extent_map *em)
+{
+	if (em->block_start + em->len < em->block_start)
+		return (u64)-1;
+	return em->block_start + em->len;
+}
 
-void extent_map_tree_init(struct extent_map_tree *tree,
-			  struct address_space *mapping, gfp_t mask);
-void extent_map_tree_empty_lru(struct extent_map_tree *tree);
+void extent_map_tree_init(struct extent_map_tree *tree, gfp_t mask);
 struct extent_map *lookup_extent_mapping(struct extent_map_tree *tree,
-					 u64 start, u64 end);
+					 u64 start, u64 len);
 int add_extent_mapping(struct extent_map_tree *tree,
 		       struct extent_map *em);
 int remove_extent_mapping(struct extent_map_tree *tree, struct extent_map *em);
-int try_release_extent_mapping(struct extent_map_tree *tree, struct page *page);
-int lock_extent(struct extent_map_tree *tree, u64 start, u64 end, gfp_t mask);
-int unlock_extent(struct extent_map_tree *tree, u64 start, u64 end, gfp_t mask);
+
 struct extent_map *alloc_extent_map(gfp_t mask);
 void free_extent_map(struct extent_map *em);
-int extent_read_full_page(struct extent_map_tree *tree, struct page *page,
-			  get_extent_t *get_extent);
 int __init extent_map_init(void);
 void extent_map_exit(void);
-
-u64 count_range_bits(struct extent_map_tree *tree,
-		     u64 *start, u64 search_end,
-		     u64 max_bytes, unsigned long bits);
-
-int test_range_bit(struct extent_map_tree *tree, u64 start, u64 end,
-		   int bits, int filled);
-int clear_extent_bits(struct extent_map_tree *tree, u64 start, u64 end,
-		      int bits, gfp_t mask);
-int set_extent_bits(struct extent_map_tree *tree, u64 start, u64 end,
-		    int bits, gfp_t mask);
-int set_extent_uptodate(struct extent_map_tree *tree, u64 start, u64 end,
-			gfp_t mask);
-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,
-			  get_extent_t *get_extent,
-			  struct writeback_control *wbc);
-int extent_writepages(struct extent_map_tree *tree,
-		      struct address_space *mapping,
-		      get_extent_t *get_extent,
-		      struct writeback_control *wbc);
-int extent_readpages(struct extent_map_tree *tree,
-		     struct address_space *mapping,
-		     struct list_head *pages, unsigned nr_pages,
-		     get_extent_t get_extent);
-int extent_prepare_write(struct extent_map_tree *tree,
-			 struct inode *inode, struct page *page,
-			 unsigned from, unsigned to, get_extent_t *get_extent);
-int extent_commit_write(struct extent_map_tree *tree,
-			struct inode *inode, struct page *page,
-			unsigned from, unsigned to);
-sector_t extent_bmap(struct address_space *mapping, sector_t iblock,
-		get_extent_t *get_extent);
-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,
-					  struct page *page0,
-					  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, u64 start, int wait);
-
-static inline void extent_buffer_get(struct extent_buffer *eb)
-{
-	atomic_inc(&eb->refs);
-}
-
-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,
-		      unsigned long min_len, char **token, char **map,
-		      unsigned long *map_start,
-		      unsigned long *map_len, int km);
-int map_private_extent_buffer(struct extent_buffer *eb, unsigned long offset,
-		      unsigned long min_len, 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.c

@@ -233,8 +233,7 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans,
 	int err = 0;
 	int i;
 	struct inode *inode = fdentry(file)->d_inode;
-	struct extent_map *em;
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	u64 hint_byte;
 	u64 num_bytes;
 	u64 start_pos;
@@ -242,11 +241,6 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans,
 	u64 end_pos = pos + write_bytes;
 	u64 inline_size;
 	loff_t isize = i_size_read(inode);
-	em = alloc_extent_map(GFP_NOFS);
-	if (!em)
-		return -ENOMEM;
-
-	em->bdev = inode->i_sb->s_bdev;
 
 	start_pos = pos & ~((u64)root->sectorsize - 1);
 	num_bytes = (write_bytes + pos - start_pos +
@@ -254,7 +248,7 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans,
 
 	end_of_last_block = start_pos + num_bytes - 1;
 
-	lock_extent(em_tree, start_pos, end_of_last_block, GFP_NOFS);
+	lock_extent(io_tree, start_pos, end_of_last_block, GFP_NOFS);
 	mutex_lock(&root->fs_info->fs_mutex);
 	trans = btrfs_start_transaction(root, 1);
 	if (!trans) {
@@ -268,7 +262,7 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans,
 	if ((end_of_last_block & 4095) == 0) {
 		printk("strange end of last %Lu %zu %Lu\n", start_pos, write_bytes, end_of_last_block);
 	}
-	set_extent_uptodate(em_tree, start_pos, end_of_last_block, GFP_NOFS);
+	set_extent_uptodate(io_tree, start_pos, end_of_last_block, GFP_NOFS);
 
 	/* FIXME...EIEIO, ENOSPC and more */
 
@@ -293,6 +287,8 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans,
 						       inode->i_ino,
 						       last_pos_in_file,
 						       0, 0, hole_size);
+			btrfs_drop_extent_cache(inode, last_pos_in_file,
+					last_pos_in_file + hole_size -1);
 			btrfs_check_file(root, inode);
 		}
 		if (err)
@@ -320,12 +316,12 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans,
 		last_end += PAGE_CACHE_SIZE - 1;
 		if (start_pos < isize) {
 			u64 delalloc_start = start_pos;
-			existing_delalloc = count_range_bits(em_tree,
+			existing_delalloc = count_range_bits(io_tree,
 					     &delalloc_start,
 					     end_of_last_block, (u64)-1,
 					     EXTENT_DELALLOC);
 		}
-		set_extent_delalloc(em_tree, start_pos, end_of_last_block,
+		set_extent_delalloc(io_tree, start_pos, end_of_last_block,
 				 GFP_NOFS);
 		spin_lock(&root->fs_info->delalloc_lock);
 		root->fs_info->delalloc_bytes += (end_of_last_block + 1 -
@@ -346,6 +342,7 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans,
 		inline_size -= start_pos;
 		err = insert_inline_extent(trans, root, inode, start_pos,
 					   inline_size, pages, 0, num_pages);
+		btrfs_drop_extent_cache(inode, start_pos, aligned_end - 1);
 		BUG_ON(err);
 	}
 	if (end_pos > isize) {
@@ -356,8 +353,7 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans,
 	err = btrfs_end_transaction(trans, root);
 out_unlock:
 	mutex_unlock(&root->fs_info->fs_mutex);
-	unlock_extent(em_tree, start_pos, end_of_last_block, GFP_NOFS);
-	free_extent_map(em);
+	unlock_extent(io_tree, start_pos, end_of_last_block, GFP_NOFS);
 	return err;
 }
 
@@ -367,10 +363,15 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end)
 	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 
 	while(1) {
+		spin_lock(&em_tree->lock);
 		em = lookup_extent_mapping(em_tree, start, end);
-		if (!em)
+		if (!em) {
+			spin_unlock(&em_tree->lock);
 			break;
+		}
 		remove_extent_mapping(em_tree, em);
+		spin_unlock(&em_tree->lock);
+
 		/* once for us */
 		free_extent_map(em);
 		/* once for the tree*/

fs/btrfs/super.c

@@ -468,10 +468,15 @@ static int __init init_btrfs_fs(void)
 	err = btrfs_init_cachep();
 	if (err)
 		goto free_transaction_sys;
-	err = extent_map_init();
+
+	err = extent_io_init();
 	if (err)
 		goto free_cachep;
 
+	err = extent_map_init();
+	if (err)
+		goto free_extent_io;
+
 	err = register_filesystem(&btrfs_fs_type);
 	if (err)
 		goto free_extent_map;
@@ -479,6 +484,8 @@ static int __init init_btrfs_fs(void)
 
 free_extent_map:
 	extent_map_exit();
+free_extent_io:
+	extent_io_exit();
 free_cachep:
 	btrfs_destroy_cachep();
 free_transaction_sys:
@@ -492,6 +499,7 @@ static void __exit exit_btrfs_fs(void)
 	btrfs_exit_transaction_sys();
 	btrfs_destroy_cachep();
 	extent_map_exit();
+	extent_io_exit();
 	unregister_filesystem(&btrfs_fs_type);
 	btrfs_exit_sysfs();
 }

fs/btrfs/transaction.c

@@ -70,7 +70,7 @@ static int join_transaction(struct btrfs_root *root)
 		INIT_LIST_HEAD(&cur_trans->pending_snapshots);
 		list_add_tail(&cur_trans->list, &root->fs_info->trans_list);
 		btrfs_ordered_inode_tree_init(&cur_trans->ordered_inode_tree);
-		extent_map_tree_init(&cur_trans->dirty_pages,
+		extent_io_tree_init(&cur_trans->dirty_pages,
 				     root->fs_info->btree_inode->i_mapping,
 				     GFP_NOFS);
 	} else {
@@ -153,7 +153,7 @@ int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
 	int ret;
 	int err;
 	int werr = 0;
-	struct extent_map_tree *dirty_pages;
+	struct extent_io_tree *dirty_pages;
 	struct page *page;
 	struct inode *btree_inode = root->fs_info->btree_inode;
 	u64 start;
@@ -610,7 +610,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	struct btrfs_transaction *cur_trans;
 	struct btrfs_transaction *prev_trans = NULL;
 	struct list_head dirty_fs_roots;
-	struct extent_map_tree *pinned_copy;
+	struct extent_io_tree *pinned_copy;
 	DEFINE_WAIT(wait);
 	int ret;
 
@@ -639,7 +639,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	if (!pinned_copy)
 		return -ENOMEM;
 
-	extent_map_tree_init(pinned_copy,
+	extent_io_tree_init(pinned_copy,
 			     root->fs_info->btree_inode->i_mapping, GFP_NOFS);
 
 	trans->transaction->in_commit = 1;

fs/btrfs/transaction.h

@@ -29,7 +29,7 @@ struct btrfs_transaction {
 	int use_count;
 	int commit_done;
 	struct list_head list;
-	struct extent_map_tree dirty_pages;
+	struct extent_io_tree dirty_pages;
 	unsigned long start_time;
 	struct btrfs_ordered_inode_tree ordered_inode_tree;
 	wait_queue_head_t writer_wait;