f2fs: use meta_inode cache to improve roll-forward speed

Previously, all the dnode pages should be read during the roll-forward recovery.
Even worsely, whole the chain was traversed twice.
This patch removes that redundant and costly read operations by using page cache
of meta_inode and readahead function as well.

Reviewed-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

fs/f2fs/checkpoint.c

@@ -72,7 +72,23 @@ out:
 	return page;
 }
 
-static inline int get_max_meta_blks(struct f2fs_sb_info *sbi, int type)
+struct page *get_meta_page_ra(struct f2fs_sb_info *sbi, pgoff_t index)
+{
+	bool readahead = false;
+	struct page *page;
+
+	page = find_get_page(META_MAPPING(sbi), index);
+	if (!page || (page && !PageUptodate(page)))
+		readahead = true;
+	f2fs_put_page(page, 0);
+
+	if (readahead)
+		ra_meta_pages(sbi, index,
+				MAX_BIO_BLOCKS(max_hw_blocks(sbi)), META_POR);
+	return get_meta_page(sbi, index);
+}
+
+static inline block_t get_max_meta_blks(struct f2fs_sb_info *sbi, int type)
 {
 	switch (type) {
 	case META_NAT:
@@ -82,6 +98,8 @@ static inline int get_max_meta_blks(struct f2fs_sb_info *sbi, int type)
 	case META_SSA:
 	case META_CP:
 		return 0;
+	case META_POR:
+		return SM_I(sbi)->seg0_blkaddr + TOTAL_BLKS(sbi);
 	default:
 		BUG();
 	}
@@ -90,12 +108,13 @@ static inline int get_max_meta_blks(struct f2fs_sb_info *sbi, int type)
 /*
  * Readahead CP/NAT/SIT/SSA pages
  */
-int ra_meta_pages(struct f2fs_sb_info *sbi, int start, int nrpages, int type)
+int ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, int type)
 {
 	block_t prev_blk_addr = 0;
 	struct page *page;
-	int blkno = start;
-	int max_blks = get_max_meta_blks(sbi, type);
+	block_t blkno = start;
+	block_t max_blks = get_max_meta_blks(sbi, type);
+	block_t min_blks = SM_I(sbi)->seg0_blkaddr;
 
 	struct f2fs_io_info fio = {
 		.type = META,
@@ -125,7 +144,11 @@ int ra_meta_pages(struct f2fs_sb_info *sbi, int start, int nrpages, int type)
 			break;
 		case META_SSA:
 		case META_CP:
-			/* get ssa/cp block addr */
+		case META_POR:
+			if (unlikely(blkno >= max_blks))
+				goto out;
+			if (unlikely(blkno < min_blks))
+				goto out;
 			blk_addr = blkno;
 			break;
 		default:

fs/f2fs/f2fs.h

@@ -103,7 +103,8 @@ enum {
 	META_CP,
 	META_NAT,
 	META_SIT,
-	META_SSA
+	META_SSA,
+	META_POR,
 };
 
 /* for the list of ino */
@@ -1294,7 +1295,8 @@ void destroy_segment_manager_caches(void);
  */
 struct page *grab_meta_page(struct f2fs_sb_info *, pgoff_t);
 struct page *get_meta_page(struct f2fs_sb_info *, pgoff_t);
-int ra_meta_pages(struct f2fs_sb_info *, int, int, int);
+struct page *get_meta_page_ra(struct f2fs_sb_info *, pgoff_t);
+int ra_meta_pages(struct f2fs_sb_info *, block_t, int, int);
 long sync_meta_pages(struct f2fs_sb_info *, enum page_type, long);
 void add_dirty_inode(struct f2fs_sb_info *, nid_t, int type);
 void remove_dirty_inode(struct f2fs_sb_info *, nid_t, int type);

fs/f2fs/recovery.c

@@ -138,7 +138,7 @@ static int find_fsync_dnodes(struct f2fs_sb_info *sbi, struct list_head *head)
 {
 	unsigned long long cp_ver = cur_cp_version(F2FS_CKPT(sbi));
 	struct curseg_info *curseg;
-	struct page *page;
+	struct page *page = NULL;
 	block_t blkaddr;
 	int err = 0;
 
@@ -146,20 +146,14 @@ static int find_fsync_dnodes(struct f2fs_sb_info *sbi, struct list_head *head)
 	curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
 	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
 
-	/* read node page */
-	page = alloc_page(GFP_F2FS_ZERO);
-	if (!page)
-		return -ENOMEM;
-	lock_page(page);
-
 	while (1) {
 		struct fsync_inode_entry *entry;
 
-		err = f2fs_submit_page_bio(sbi, page, blkaddr, READ_SYNC);
-		if (err)
-			return err;
+		if (blkaddr < SM_I(sbi)->main_blkaddr ||
+			blkaddr >= (SM_I(sbi)->seg0_blkaddr + TOTAL_BLKS(sbi)))
+			return 0;
 
-		lock_page(page);
+		page = get_meta_page_ra(sbi, blkaddr);
 
 		if (cp_ver != cpver_of_node(page))
 			break;
@@ -202,11 +196,9 @@ static int find_fsync_dnodes(struct f2fs_sb_info *sbi, struct list_head *head)
 next:
 		/* check next segment */
 		blkaddr = next_blkaddr_of_node(page);
+		f2fs_put_page(page, 1);
 	}
-
-	unlock_page(page);
-	__free_pages(page, 0);
-
+	f2fs_put_page(page, 1);
 	return err;
 }
 
@@ -400,7 +392,7 @@ static int recover_data(struct f2fs_sb_info *sbi,
 {
 	unsigned long long cp_ver = cur_cp_version(F2FS_CKPT(sbi));
 	struct curseg_info *curseg;
-	struct page *page;
+	struct page *page = NULL;
 	int err = 0;
 	block_t blkaddr;
 
@@ -408,32 +400,29 @@ static int recover_data(struct f2fs_sb_info *sbi,
 	curseg = CURSEG_I(sbi, type);
 	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
 
-	/* read node page */
-	page = alloc_page(GFP_F2FS_ZERO);
-	if (!page)
-		return -ENOMEM;
-
-	lock_page(page);
-
 	while (1) {
 		struct fsync_inode_entry *entry;
 
-		err = f2fs_submit_page_bio(sbi, page, blkaddr, READ_SYNC);
-		if (err)
-			return err;
+		if (blkaddr < SM_I(sbi)->main_blkaddr ||
+			blkaddr >= (SM_I(sbi)->seg0_blkaddr + TOTAL_BLKS(sbi)))
+			break;
 
-		lock_page(page);
+		page = get_meta_page_ra(sbi, blkaddr);
 
-		if (cp_ver != cpver_of_node(page))
+		if (cp_ver != cpver_of_node(page)) {
+			f2fs_put_page(page, 1);
 			break;
+		}
 
 		entry = get_fsync_inode(head, ino_of_node(page));
 		if (!entry)
 			goto next;
 
 		err = do_recover_data(sbi, entry->inode, page, blkaddr);
-		if (err)
+		if (err) {
+			f2fs_put_page(page, 1);
 			break;
+		}
 
 		if (entry->blkaddr == blkaddr) {
 			iput(entry->inode);
@@ -443,11 +432,8 @@ static int recover_data(struct f2fs_sb_info *sbi,
 next:
 		/* check next segment */
 		blkaddr = next_blkaddr_of_node(page);
+		f2fs_put_page(page, 1);
 	}
-
-	unlock_page(page);
-	__free_pages(page, 0);
-
 	if (!err)
 		allocate_new_segments(sbi);
 	return err;
@@ -493,6 +479,10 @@ out:
 	destroy_fsync_dnodes(&inode_list);
 	kmem_cache_destroy(fsync_entry_slab);
 
+	/* truncate meta pages to be used by the recovery */
+	truncate_inode_pages_range(META_MAPPING(sbi),
+		SM_I(sbi)->main_blkaddr << PAGE_CACHE_SHIFT, -1);
+
 	if (err) {
 		truncate_inode_pages_final(NODE_MAPPING(sbi));
 		truncate_inode_pages_final(META_MAPPING(sbi));

fs/f2fs/segment.h

@@ -87,6 +87,7 @@
 	(BITS_TO_LONGS(nr) * sizeof(unsigned long))
 #define TOTAL_SEGS(sbi)	(SM_I(sbi)->main_segments)
 #define TOTAL_SECS(sbi)	(sbi->total_sections)
+#define TOTAL_BLKS(sbi)	(SM_I(sbi)->segment_count << sbi->log_blocks_per_seg)
 
 #define SECTOR_FROM_BLOCK(sbi, blk_addr)				\
 	(((sector_t)blk_addr) << (sbi)->log_sectors_per_block)
@@ -553,7 +554,7 @@ static inline void check_seg_range(struct f2fs_sb_info *sbi, unsigned int segno)
 static inline void verify_block_addr(struct f2fs_sb_info *sbi, block_t blk_addr)
 {
 	struct f2fs_sm_info *sm_info = SM_I(sbi);
-	block_t total_blks = sm_info->segment_count << sbi->log_blocks_per_seg;
+	block_t total_blks = TOTAL_BLKS(sbi);
 	block_t start_addr = sm_info->seg0_blkaddr;
 	block_t end_addr = start_addr + total_blks - 1;
 	BUG_ON(blk_addr < start_addr);
@@ -606,7 +607,7 @@ static inline void check_seg_range(struct f2fs_sb_info *sbi, unsigned int segno)
 static inline void verify_block_addr(struct f2fs_sb_info *sbi, block_t blk_addr)
 {
 	struct f2fs_sm_info *sm_info = SM_I(sbi);
-	block_t total_blks = sm_info->segment_count << sbi->log_blocks_per_seg;
+	block_t total_blks = TOTAL_BLKS(sbi);
 	block_t start_addr = sm_info->seg0_blkaddr;
 	block_t end_addr = start_addr + total_blks - 1;