f2fs: preallocate blocks for buffered aio writes

This patch preallocates data blocks for buffered aio writes.
With this patch, we can avoid redundant locking and unlocking of node pages
given consecutive aio request.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

fs/f2fs/data.c

@@ -571,16 +571,25 @@ ssize_t f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from)
 	ssize_t ret = 0;
 
 	map.m_lblk = F2FS_BYTES_TO_BLK(iocb->ki_pos);
-	map.m_len = F2FS_BYTES_TO_BLK(iov_iter_count(from));
+	map.m_len = F2FS_BLK_ALIGN(iov_iter_count(from));
 	map.m_next_pgofs = NULL;
 
-	if (iocb->ki_flags & IOCB_DIRECT &&
-		!(f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode))) {
+	if (f2fs_encrypted_inode(inode))
+		return 0;
+
+	if (iocb->ki_flags & IOCB_DIRECT) {
+		ret = f2fs_convert_inline_inode(inode);
+		if (ret)
+			return ret;
+		return f2fs_map_blocks(inode, &map, 1, F2FS_GET_BLOCK_PRE_DIO);
+	}
+	if (iocb->ki_pos + iov_iter_count(from) > MAX_INLINE_DATA) {
 		ret = f2fs_convert_inline_inode(inode);
 		if (ret)
 			return ret;
-		ret = f2fs_map_blocks(inode, &map, 1, F2FS_GET_BLOCK_PRE_DIO);
 	}
+	if (!f2fs_has_inline_data(inode))
+		return f2fs_map_blocks(inode, &map, 1, F2FS_GET_BLOCK_PRE_AIO);
 	return ret;
 }
 
@@ -612,7 +621,7 @@ int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
 	/* it only supports block size == page size */
 	pgofs =	(pgoff_t)map->m_lblk;
 
-	if (f2fs_lookup_extent_cache(inode, pgofs, &ei)) {
+	if (!create && f2fs_lookup_extent_cache(inode, pgofs, &ei)) {
 		map->m_pblk = ei.blk + pgofs - ei.fofs;
 		map->m_len = min((pgoff_t)maxblocks, ei.fofs + ei.len - pgofs);
 		map->m_flags = F2FS_MAP_MAPPED;
@@ -647,7 +656,12 @@ next_block:
 				err = -EIO;
 				goto sync_out;
 			}
-			err = __allocate_data_block(&dn);
+			if (flag == F2FS_GET_BLOCK_PRE_AIO) {
+				if (blkaddr == NULL_ADDR)
+					err = reserve_new_block(&dn);
+			} else {
+				err = __allocate_data_block(&dn);
+			}
 			if (err)
 				goto sync_out;
 			allocated = true;
@@ -679,7 +693,8 @@ next_block:
 	} else if ((map->m_pblk != NEW_ADDR &&
 			blkaddr == (map->m_pblk + ofs)) ||
 			(map->m_pblk == NEW_ADDR && blkaddr == NEW_ADDR) ||
-			flag == F2FS_GET_BLOCK_PRE_DIO) {
+			flag == F2FS_GET_BLOCK_PRE_DIO ||
+			flag == F2FS_GET_BLOCK_PRE_AIO) {
 		ofs++;
 		map->m_len++;
 	} else {
@@ -1418,6 +1433,14 @@ static int prepare_write_begin(struct f2fs_sb_info *sbi,
 	struct extent_info ei;
 	int err = 0;
 
+	/*
+	 * we already allocated all the blocks, so we don't need to get
+	 * the block addresses when there is no need to fill the page.
+	 */
+	if (!f2fs_has_inline_data(inode) && !f2fs_encrypted_inode(inode) &&
+					len == PAGE_CACHE_SIZE)
+		return 0;
+
 	if (f2fs_has_inline_data(inode) ||
 			(pos & PAGE_CACHE_MASK) >= i_size_read(inode)) {
 		f2fs_lock_op(sbi);

fs/f2fs/f2fs.h

@@ -392,6 +392,7 @@ struct f2fs_map_blocks {
 #define F2FS_GET_BLOCK_FIEMAP		2
 #define F2FS_GET_BLOCK_BMAP		3
 #define F2FS_GET_BLOCK_PRE_DIO		4
+#define F2FS_GET_BLOCK_PRE_AIO		5
 
 /*
  * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.

include/linux/f2fs_fs.h

@@ -21,7 +21,7 @@
 #define F2FS_BLKSIZE			4096	/* support only 4KB block */
 #define F2FS_BLKSIZE_BITS		12	/* bits for F2FS_BLKSIZE */
 #define F2FS_MAX_EXTENSION		64	/* # of extension entries */
-#define F2FS_BLK_ALIGN(x)	(((x) + F2FS_BLKSIZE - 1) / F2FS_BLKSIZE)
+#define F2FS_BLK_ALIGN(x)	(((x) + F2FS_BLKSIZE - 1) >> F2FS_BLKSIZE_BITS)
 
 #define NULL_ADDR		((block_t)0)	/* used as block_t addresses */
 #define NEW_ADDR		((block_t)-1)	/* used as block_t addresses */