1 files changed, 70 insertions, 45 deletions

diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 0554c48cb1fd..ccce8a7e94ed 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -117,11 +117,33 @@ static ext4_fsblk_t ext4_ext_find_goal(struct inode *inode,
 		struct ext4_extent *ex;
 		depth = path->p_depth;
 
-		/* try to predict block placement */
+		/*
+		 * Try to predict block placement assuming that we are
+		 * filling in a file which will eventually be
+		 * non-sparse --- i.e., in the case of libbfd writing
+		 * an ELF object sections out-of-order but in a way
+		 * the eventually results in a contiguous object or
+		 * executable file, or some database extending a table
+		 * space file.  However, this is actually somewhat
+		 * non-ideal if we are writing a sparse file such as
+		 * qemu or KVM writing a raw image file that is going
+		 * to stay fairly sparse, since it will end up
+		 * fragmenting the file system's free space.  Maybe we
+		 * should have some hueristics or some way to allow
+		 * userspace to pass a hint to file system,
+		 * especiially if the latter case turns out to be
+		 * common.
+		 */
 		ex = path[depth].p_ext;
-		if (ex)
-			return (ext4_ext_pblock(ex) +
-				(block - le32_to_cpu(ex->ee_block)));
+		if (ex) {
+			ext4_fsblk_t ext_pblk = ext4_ext_pblock(ex);
+			ext4_lblk_t ext_block = le32_to_cpu(ex->ee_block);
+
+			if (block > ext_block)
+				return ext_pblk + (block - ext_block);
+			else
+				return ext_pblk - (ext_block - block);
+		}
 
 		/* it looks like index is empty;
 		 * try to find starting block from index itself */
@@ -244,7 +266,7 @@ static inline int ext4_ext_space_root_idx(struct inode *inode, int check)
  * to allocate @blocks
  * Worse case is one block per extent
  */
-int ext4_ext_calc_metadata_amount(struct inode *inode, sector_t lblock)
+int ext4_ext_calc_metadata_amount(struct inode *inode, ext4_lblk_t lblock)
 {
 	struct ext4_inode_info *ei = EXT4_I(inode);
 	int idxs, num = 0;
@@ -1872,12 +1894,10 @@ static int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
 			cbex.ec_block = start;
 			cbex.ec_len = end - start;
 			cbex.ec_start = 0;
-			cbex.ec_type = EXT4_EXT_CACHE_GAP;
 		} else {
 			cbex.ec_block = le32_to_cpu(ex->ee_block);
 			cbex.ec_len = ext4_ext_get_actual_len(ex);
 			cbex.ec_start = ext4_ext_pblock(ex);
-			cbex.ec_type = EXT4_EXT_CACHE_EXTENT;
 		}
 
 		if (unlikely(cbex.ec_len == 0)) {
@@ -1917,13 +1937,12 @@ static int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
 
 static void
 ext4_ext_put_in_cache(struct inode *inode, ext4_lblk_t block,
-			__u32 len, ext4_fsblk_t start, int type)
+			__u32 len, ext4_fsblk_t start)
 {
 	struct ext4_ext_cache *cex;
 	BUG_ON(len == 0);
 	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
 	cex = &EXT4_I(inode)->i_cached_extent;
-	cex->ec_type = type;
 	cex->ec_block = block;
 	cex->ec_len = len;
 	cex->ec_start = start;
@@ -1976,15 +1995,18 @@ ext4_ext_put_gap_in_cache(struct inode *inode, struct ext4_ext_path *path,
 	}
 
 	ext_debug(" -> %u:%lu\n", lblock, len);
-	ext4_ext_put_in_cache(inode, lblock, len, 0, EXT4_EXT_CACHE_GAP);
+	ext4_ext_put_in_cache(inode, lblock, len, 0);
 }
 
+/*
+ * Return 0 if cache is invalid; 1 if the cache is valid
+ */
 static int
 ext4_ext_in_cache(struct inode *inode, ext4_lblk_t block,
 			struct ext4_extent *ex)
 {
 	struct ext4_ext_cache *cex;
-	int ret = EXT4_EXT_CACHE_NO;
+	int ret = 0;
 
 	/*
 	 * We borrow i_block_reservation_lock to protect i_cached_extent
@@ -1993,11 +2015,9 @@ ext4_ext_in_cache(struct inode *inode, ext4_lblk_t block,
 	cex = &EXT4_I(inode)->i_cached_extent;
 
 	/* has cache valid data? */
-	if (cex->ec_type == EXT4_EXT_CACHE_NO)
+	if (cex->ec_len == 0)
 		goto errout;
 
-	BUG_ON(cex->ec_type != EXT4_EXT_CACHE_GAP &&
-			cex->ec_type != EXT4_EXT_CACHE_EXTENT);
 	if (in_range(block, cex->ec_block, cex->ec_len)) {
 		ex->ee_block = cpu_to_le32(cex->ec_block);
 		ext4_ext_store_pblock(ex, cex->ec_start);
@@ -2005,7 +2025,7 @@ ext4_ext_in_cache(struct inode *inode, ext4_lblk_t block,
 		ext_debug("%u cached by %u:%u:%llu\n",
 				block,
 				cex->ec_block, cex->ec_len, cex->ec_start);
-		ret = cex->ec_type;
+		ret = 1;
 	}
 errout:
 	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
@@ -2825,14 +2845,14 @@ fix_extent_len:
  * to an uninitialized extent.
  *
  * Writing to an uninitized extent may result in splitting the uninitialized
- * extent into multiple /intialized unintialized extents (up to three)
+ * extent into multiple /initialized uninitialized extents (up to three)
  * There are three possibilities:
  *   a> There is no split required: Entire extent should be uninitialized
  *   b> Splits in two extents: Write is happening at either end of the extent
  *   c> Splits in three extents: Somone is writing in middle of the extent
  *
  * One of more index blocks maybe needed if the extent tree grow after
- * the unintialized extent split. To prevent ENOSPC occur at the IO
+ * the uninitialized extent split. To prevent ENOSPC occur at the IO
  * complete, we need to split the uninitialized extent before DIO submit
  * the IO. The uninitialized extent called at this time will be split
  * into three uninitialized extent(at most). After IO complete, the part
@@ -3082,7 +3102,7 @@ static void unmap_underlying_metadata_blocks(struct block_device *bdev,
  * Handle EOFBLOCKS_FL flag, clearing it if necessary
  */
 static int check_eofblocks_fl(handle_t *handle, struct inode *inode,
-			      struct ext4_map_blocks *map,
+			      ext4_lblk_t lblk,
 			      struct ext4_ext_path *path,
 			      unsigned int len)
 {
@@ -3112,7 +3132,7 @@ static int check_eofblocks_fl(handle_t *handle, struct inode *inode,
 	 * this turns out to be false, we can bail out from this
 	 * function immediately.
 	 */
-	if (map->m_lblk + len < le32_to_cpu(last_ex->ee_block) +
+	if (lblk + len < le32_to_cpu(last_ex->ee_block) +
 	    ext4_ext_get_actual_len(last_ex))
 		return 0;
 	/*
@@ -3154,9 +3174,10 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode,
 		 * that this IO needs to convertion to written when IO is
 		 * completed
 		 */
-		if (io)
+		if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) {
 			io->flag = EXT4_IO_END_UNWRITTEN;
-		else
+			atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
+		} else
 			ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN);
 		if (ext4_should_dioread_nolock(inode))
 			map->m_flags |= EXT4_MAP_UNINIT;
@@ -3168,8 +3189,8 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode,
 							path);
 		if (ret >= 0) {
 			ext4_update_inode_fsync_trans(handle, inode, 1);
-			err = check_eofblocks_fl(handle, inode, map, path,
-						 map->m_len);
+			err = check_eofblocks_fl(handle, inode, map->m_lblk,
+						 path, map->m_len);
 		} else
 			err = ret;
 		goto out2;
@@ -3199,7 +3220,8 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode,
 	ret = ext4_ext_convert_to_initialized(handle, inode, map, path);
 	if (ret >= 0) {
 		ext4_update_inode_fsync_trans(handle, inode, 1);
-		err = check_eofblocks_fl(handle, inode, map, path, map->m_len);
+		err = check_eofblocks_fl(handle, inode, map->m_lblk, path,
+					 map->m_len);
 		if (err < 0)
 			goto out2;
 	}
@@ -3276,7 +3298,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 	struct ext4_extent_header *eh;
 	struct ext4_extent newex, *ex;
 	ext4_fsblk_t newblock;
-	int err = 0, depth, ret, cache_type;
+	int err = 0, depth, ret;
 	unsigned int allocated = 0;
 	struct ext4_allocation_request ar;
 	ext4_io_end_t *io = EXT4_I(inode)->cur_aio_dio;
@@ -3285,9 +3307,8 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 		  map->m_lblk, map->m_len, inode->i_ino);
 
 	/* check in cache */
-	cache_type = ext4_ext_in_cache(inode, map->m_lblk, &newex);
-	if (cache_type) {
-		if (cache_type == EXT4_EXT_CACHE_GAP) {
+	if (ext4_ext_in_cache(inode, map->m_lblk, &newex)) {
+		if (!newex.ee_start_lo && !newex.ee_start_hi) {
 			if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
 				/*
 				 * block isn't allocated yet and
@@ -3296,7 +3317,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 				goto out2;
 			}
 			/* we should allocate requested block */
-		} else if (cache_type == EXT4_EXT_CACHE_EXTENT) {
+		} else {
 			/* block is already allocated */
 			newblock = map->m_lblk
 				   - le32_to_cpu(newex.ee_block)
@@ -3305,8 +3326,6 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 			allocated = ext4_ext_get_actual_len(&newex) -
 				(map->m_lblk - le32_to_cpu(newex.ee_block));
 			goto out;
-		} else {
-			BUG();
 		}
 	}
 
@@ -3357,8 +3376,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 			/* Do not put uninitialized extent in the cache */
 			if (!ext4_ext_is_uninitialized(ex)) {
 				ext4_ext_put_in_cache(inode, ee_block,
-							ee_len, ee_start,
-							EXT4_EXT_CACHE_EXTENT);
+							ee_len, ee_start);
 				goto out;
 			}
 			ret = ext4_ext_handle_uninitialized_extents(handle,
@@ -3446,9 +3464,10 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 		 * that we need to perform convertion when IO is done.
 		 */
 		if ((flags & EXT4_GET_BLOCKS_PRE_IO)) {
-			if (io)
+			if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) {
 				io->flag = EXT4_IO_END_UNWRITTEN;
-			else
+				atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
+			} else
 				ext4_set_inode_state(inode,
 						     EXT4_STATE_DIO_UNWRITTEN);
 		}
@@ -3456,7 +3475,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 			map->m_flags |= EXT4_MAP_UNINIT;
 	}
 
-	err = check_eofblocks_fl(handle, inode, map, path, ar.len);
+	err = check_eofblocks_fl(handle, inode, map->m_lblk, path, ar.len);
 	if (err)
 		goto out2;
 
@@ -3490,8 +3509,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 	 * when it is _not_ an uninitialized extent.
 	 */
 	if ((flags & EXT4_GET_BLOCKS_UNINIT_EXT) == 0) {
-		ext4_ext_put_in_cache(inode, map->m_lblk, allocated, newblock,
-						EXT4_EXT_CACHE_EXTENT);
+		ext4_ext_put_in_cache(inode, map->m_lblk, allocated, newblock);
 		ext4_update_inode_fsync_trans(handle, inode, 1);
 	} else
 		ext4_update_inode_fsync_trans(handle, inode, 0);
@@ -3519,6 +3537,12 @@ void ext4_ext_truncate(struct inode *inode)
 	int err = 0;
 
 	/*
+	 * finish any pending end_io work so we won't run the risk of
+	 * converting any truncated blocks to initialized later
+	 */
+	ext4_flush_completed_IO(inode);
+
+	/*
 	 * probably first extent we're gonna free will be last in block
 	 */
 	err = ext4_writepage_trans_blocks(inode);
@@ -3605,14 +3629,15 @@ static void ext4_falloc_update_inode(struct inode *inode,
 }
 
 /*
- * preallocate space for a file. This implements ext4's fallocate inode
+ * preallocate space for a file. This implements ext4's fallocate file
  * operation, which gets called from sys_fallocate system call.
  * For block-mapped files, posix_fallocate should fall back to the method
  * of writing zeroes to the required new blocks (the same behavior which is
  * expected for file systems which do not support fallocate() system call).
  */
-long ext4_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len)
+long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
 {
+	struct inode *inode = file->f_path.dentry->d_inode;
 	handle_t *handle;
 	loff_t new_size;
 	unsigned int max_blocks;
@@ -3622,6 +3647,10 @@ long ext4_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len)
 	struct ext4_map_blocks map;
 	unsigned int credits, blkbits = inode->i_blkbits;
 
+	/* We only support the FALLOC_FL_KEEP_SIZE mode */
+	if (mode & ~FALLOC_FL_KEEP_SIZE)
+		return -EOPNOTSUPP;
+
 	/*
 	 * currently supporting (pre)allocate mode for extent-based
 	 * files _only_
@@ -3629,10 +3658,6 @@ long ext4_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len)
 	if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
 		return -EOPNOTSUPP;
 
-	/* preallocation to directories is currently not supported */
-	if (S_ISDIR(inode->i_mode))
-		return -ENODEV;
-
 	map.m_lblk = offset >> blkbits;
 	/*
 	 * We can't just convert len to max_blocks because
@@ -3767,7 +3792,7 @@ static int ext4_ext_fiemap_cb(struct inode *inode, struct ext4_ext_path *path,
 
 	logical =  (__u64)newex->ec_block << blksize_bits;
 
-	if (newex->ec_type == EXT4_EXT_CACHE_GAP) {
+	if (newex->ec_start == 0) {
 		pgoff_t offset;
 		struct page *page;
 		struct buffer_head *bh = NULL;