Btrfs: btrfs_drop_extent_cache should never fail

I noticed this when I was doing the fsync stuff, we allocate split extents if we
drop an extent range that is in the middle of an existing extent.  This BUG()'s
if we fail to allocate memory, but the fact is this is just a cache, we will
just regenerate the cache if we need it, the important part is that we free the
range we are given.  This can be done without allocations, so if we fail to
allocate splits just skip the splitting stage and free our em and look for more
extents to drop.  This also makes btrfs_drop_extent_cache a void since nobody
was checking the return value anyway.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>

fs/btrfs/ctree.h

@@ -3316,8 +3316,8 @@ int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans,
 			   struct inode *inode);
 int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info);
 int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
-int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
-			    int skip_pinned);
+void btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
+			     int skip_pinned);
 int btrfs_replace_extent_cache(struct inode *inode, struct extent_map *replace,
 			       u64 start, u64 end, int skip_pinned,
 			       int modified);

fs/btrfs/file.c

@@ -459,8 +459,8 @@ int btrfs_dirty_pages(struct btrfs_root *root, struct inode *inode,
  * this drops all the extents in the cache that intersect the range
  * [start, end].  Existing extents are split as required.
  */
-int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
-			       int skip_pinned)
+void btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
+			     int skip_pinned)
 {
 	struct extent_map *em;
 	struct extent_map *split = NULL;
@@ -479,11 +479,14 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 		testend = 0;
 	}
 	while (1) {
+		int no_splits = 0;
+
 		if (!split)
 			split = alloc_extent_map();
 		if (!split2)
 			split2 = alloc_extent_map();
-		BUG_ON(!split || !split2); /* -ENOMEM */
+		if (!split || !split2)
+			no_splits = 1;
 
 		write_lock(&em_tree->lock);
 		em = lookup_extent_mapping(em_tree, start, len);
@@ -509,6 +512,8 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 		compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
 		clear_bit(EXTENT_FLAG_PINNED, &em->flags);
 		remove_extent_mapping(em_tree, em);
+		if (no_splits)
+			goto next;
 
 		if (em->block_start < EXTENT_MAP_LAST_BYTE &&
 		    em->start < start) {
@@ -559,6 +564,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 			free_extent_map(split);
 			split = NULL;
 		}
+next:
 		write_unlock(&em_tree->lock);
 
 		/* once for us */
@@ -570,7 +576,6 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 		free_extent_map(split);
 	if (split2)
 		free_extent_map(split2);
-	return 0;
 }
 
 /*