Merge tag 'for-f2fs-3.15' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs

Pull f2fs updates from Jaegeuk Kim:
 "This patch-set includes the following major enhancement patches.
   - introduce large directory support
   - introduce f2fs_issue_flush to merge redundant flush commands
   - merge write IOs as much as possible aligned to the segment
   - add sysfs entries to tune the f2fs configuration
   - use radix_tree for the free_nid_list to reduce in-memory operations
   - remove costly bit operations in f2fs_find_entry
   - enhance the readahead flow for CP/NAT/SIT/SSA blocks

  The other bug fixes are as follows:
   - recover xattr node blocks correctly after sudden-power-cut
   - fix to calculate the maximum number of node ids
   - enhance to handle many error cases

  And, there are a bunch of cleanups"

* tag 'for-f2fs-3.15' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs: (62 commits)
  f2fs: fix wrong statistics of inline data
  f2fs: check the acl's validity before setting
  f2fs: introduce f2fs_issue_flush to avoid redundant flush issue
  f2fs: fix to cover io->bio with io_rwsem
  f2fs: fix error path when fail to read inline data
  f2fs: use list_for_each_entry{_safe} for simplyfying code
  f2fs: avoid free slab cache under spinlock
  f2fs: avoid unneeded lookup when xattr name length is too long
  f2fs: avoid unnecessary bio submit when wait page writeback
  f2fs: return -EIO when node id is not matched
  f2fs: avoid RECLAIM_FS-ON-W warning
  f2fs: skip unnecessary node writes during fsync
  f2fs: introduce fi->i_sem to protect fi's info
  f2fs: change reclaim rate in percentage
  f2fs: add missing documentation for dir_level
  f2fs: remove unnecessary threshold
  f2fs: throttle the memory footprint with a sysfs entry
  f2fs: avoid to drop nat entries due to the negative nr_shrink
  f2fs: call f2fs_wait_on_page_writeback instead of native function
  f2fs: introduce nr_pages_to_write for segment alignment
  ...

Documentation/ABI/testing/sysfs-fs-f2fs

@@ -55,3 +55,15 @@ Date:		January 2014
 Contact:	"Jaegeuk Kim" <jaegeuk.kim@samsung.com>
 Description:
 		 Controls the number of trials to find a victim segment.
+
+What:		/sys/fs/f2fs/<disk>/dir_level
+Date:		March 2014
+Contact:	"Jaegeuk Kim" <jaegeuk.kim@samsung.com>
+Description:
+		 Controls the directory level for large directory.
+
+What:		/sys/fs/f2fs/<disk>/ram_thresh
+Date:		March 2014
+Contact:	"Jaegeuk Kim" <jaegeuk.kim@samsung.com>
+Description:
+		 Controls the memory footprint used by f2fs.

Documentation/filesystems/f2fs.txt

@@ -122,6 +122,10 @@ disable_ext_identify   Disable the extension list configured by mkfs, so f2fs
 inline_xattr           Enable the inline xattrs feature.
 inline_data            Enable the inline data feature: New created small(<~3.4k)
                        files can be written into inode block.
+flush_merge	       Merge concurrent cache_flush commands as much as possible
+                       to eliminate redundant command issues. If the underlying
+		       device handles the cache_flush command relatively slowly,
+		       recommend to enable this option.
 
 ================================================================================
 DEBUGFS ENTRIES
@@ -169,9 +173,11 @@ Files in /sys/fs/f2fs/<devname>
 
  reclaim_segments             This parameter controls the number of prefree
                               segments to be reclaimed. If the number of prefree
-			      segments is larger than this number, f2fs tries to
-			      conduct checkpoint to reclaim the prefree segments
-			      to free segments. By default, 100 segments, 200MB.
+			      segments is larger than the number of segments
+			      in the proportion to the percentage over total
+			      volume size, f2fs tries to conduct checkpoint to
+			      reclaim the prefree segments to free segments.
+			      By default, 5% over total # of segments.
 
  max_small_discards	      This parameter controls the number of discard
 			      commands that consist small blocks less than 2MB.
@@ -195,6 +201,17 @@ Files in /sys/fs/f2fs/<devname>
 			      cleaning operations. The default value is 4096
 			      which covers 8GB block address range.
 
+ dir_level                    This parameter controls the directory level to
+			      support large directory. If a directory has a
+			      number of files, it can reduce the file lookup
+			      latency by increasing this dir_level value.
+			      Otherwise, it needs to decrease this value to
+			      reduce the space overhead. The default value is 0.
+
+ ram_thresh                   This parameter controls the memory footprint used
+			      by free nids and cached nat entries. By default,
+			      10 is set, which indicates 10 MB / 1 GB RAM.
+
 ================================================================================
 USAGE
 ================================================================================
@@ -444,9 +461,11 @@ The number of blocks and buckets are determined by,
   # of blocks in level #n = |
                             `- 4, Otherwise
 
-                             ,- 2^n, if n < MAX_DIR_HASH_DEPTH / 2,
+                             ,- 2^ (n + dir_level),
+			     |            if n < MAX_DIR_HASH_DEPTH / 2,
   # of buckets in level #n = |
-                             `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1), Otherwise
+                             `- 2^((MAX_DIR_HASH_DEPTH / 2 + dir_level) - 1),
+			                  Otherwise
 
 When F2FS finds a file name in a directory, at first a hash value of the file
 name is calculated. Then, F2FS scans the hash table in level #0 to find the

fs/f2fs/acl.c

@@ -174,7 +174,7 @@ struct posix_acl *f2fs_get_acl(struct inode *inode, int type)
 
 	retval = f2fs_getxattr(inode, name_index, "", NULL, 0);
 	if (retval > 0) {
-		value = kmalloc(retval, GFP_KERNEL);
+		value = kmalloc(retval, GFP_F2FS_ZERO);
 		if (!value)
 			return ERR_PTR(-ENOMEM);
 		retval = f2fs_getxattr(inode, name_index, "", value, retval);
@@ -203,6 +203,12 @@ static int __f2fs_set_acl(struct inode *inode, int type,
 	size_t size = 0;
 	int error;
 
+	if (acl) {
+		error = posix_acl_valid(acl);
+		if (error < 0)
+			return error;
+	}
+
 	switch (type) {
 	case ACL_TYPE_ACCESS:
 		name_index = F2FS_XATTR_INDEX_POSIX_ACL_ACCESS;

fs/f2fs/data.c

@@ -45,7 +45,7 @@ static void f2fs_read_end_io(struct bio *bio, int err)
 
 static void f2fs_write_end_io(struct bio *bio, int err)
 {
-	struct f2fs_sb_info *sbi = F2FS_SB(bio->bi_io_vec->bv_page->mapping->host->i_sb);
+	struct f2fs_sb_info *sbi = bio->bi_private;
 	struct bio_vec *bvec;
 	int i;
 
@@ -55,15 +55,16 @@ static void f2fs_write_end_io(struct bio *bio, int err)
 		if (unlikely(err)) {
 			SetPageError(page);
 			set_bit(AS_EIO, &page->mapping->flags);
-			set_ckpt_flags(sbi->ckpt, CP_ERROR_FLAG);
-			sbi->sb->s_flags |= MS_RDONLY;
+			f2fs_stop_checkpoint(sbi);
 		}
 		end_page_writeback(page);
 		dec_page_count(sbi, F2FS_WRITEBACK);
 	}
 
-	if (bio->bi_private)
-		complete(bio->bi_private);
+	if (sbi->wait_io) {
+		complete(sbi->wait_io);
+		sbi->wait_io = NULL;
+	}
 
 	if (!get_pages(sbi, F2FS_WRITEBACK) &&
 			!list_empty(&sbi->cp_wait.task_list))
@@ -86,6 +87,7 @@ static struct bio *__bio_alloc(struct f2fs_sb_info *sbi, block_t blk_addr,
 	bio->bi_bdev = sbi->sb->s_bdev;
 	bio->bi_iter.bi_sector = SECTOR_FROM_BLOCK(sbi, blk_addr);
 	bio->bi_end_io = is_read ? f2fs_read_end_io : f2fs_write_end_io;
+	bio->bi_private = sbi;
 
 	return bio;
 }
@@ -113,7 +115,7 @@ static void __submit_merged_bio(struct f2fs_bio_info *io)
 		 */
 		if (fio->type == META_FLUSH) {
 			DECLARE_COMPLETION_ONSTACK(wait);
-			io->bio->bi_private = &wait;
+			io->sbi->wait_io = &wait;
 			submit_bio(rw, io->bio);
 			wait_for_completion(&wait);
 		} else {
@@ -132,7 +134,7 @@ void f2fs_submit_merged_bio(struct f2fs_sb_info *sbi,
 
 	io = is_read_io(rw) ? &sbi->read_io : &sbi->write_io[btype];
 
-	mutex_lock(&io->io_mutex);
+	down_write(&io->io_rwsem);
 
 	/* change META to META_FLUSH in the checkpoint procedure */
 	if (type >= META_FLUSH) {
@@ -140,7 +142,7 @@ void f2fs_submit_merged_bio(struct f2fs_sb_info *sbi,
 		io->fio.rw = WRITE_FLUSH_FUA | REQ_META | REQ_PRIO;
 	}
 	__submit_merged_bio(io);
-	mutex_unlock(&io->io_mutex);
+	up_write(&io->io_rwsem);
 }
 
 /*
@@ -178,7 +180,7 @@ void f2fs_submit_page_mbio(struct f2fs_sb_info *sbi, struct page *page,
 
 	verify_block_addr(sbi, blk_addr);
 
-	mutex_lock(&io->io_mutex);
+	down_write(&io->io_rwsem);
 
 	if (!is_read)
 		inc_page_count(sbi, F2FS_WRITEBACK);
@@ -202,7 +204,7 @@ void f2fs_submit_page_mbio(struct f2fs_sb_info *sbi, struct page *page,
 
 	io->last_block_in_bio = blk_addr;
 
-	mutex_unlock(&io->io_mutex);
+	up_write(&io->io_rwsem);
 	trace_f2fs_submit_page_mbio(page, fio->rw, fio->type, blk_addr);
 }
 
@@ -797,48 +799,36 @@ static int f2fs_write_data_page(struct page *page,
 	 */
 	offset = i_size & (PAGE_CACHE_SIZE - 1);
 	if ((page->index >= end_index + 1) || !offset) {
-		if (S_ISDIR(inode->i_mode)) {
-			dec_page_count(sbi, F2FS_DIRTY_DENTS);
-			inode_dec_dirty_dents(inode);
-		}
+		inode_dec_dirty_dents(inode);
 		goto out;
 	}
 
 	zero_user_segment(page, offset, PAGE_CACHE_SIZE);
 write:
-	if (unlikely(sbi->por_doing)) {
-		err = AOP_WRITEPAGE_ACTIVATE;
+	if (unlikely(sbi->por_doing))
 		goto redirty_out;
-	}
 
 	/* Dentry blocks are controlled by checkpoint */
 	if (S_ISDIR(inode->i_mode)) {
-		dec_page_count(sbi, F2FS_DIRTY_DENTS);
 		inode_dec_dirty_dents(inode);
 		err = do_write_data_page(page, &fio);
-	} else {
-		f2fs_lock_op(sbi);
-
-		if (f2fs_has_inline_data(inode) || f2fs_may_inline(inode)) {
-			err = f2fs_write_inline_data(inode, page, offset);
-			f2fs_unlock_op(sbi);
-			goto out;
-		} else {
-			err = do_write_data_page(page, &fio);
-		}
+		goto done;
+	}
 
-		f2fs_unlock_op(sbi);
+	if (!wbc->for_reclaim)
 		need_balance_fs = true;
-	}
-	if (err == -ENOENT)
-		goto out;
-	else if (err)
+	else if (has_not_enough_free_secs(sbi, 0))
 		goto redirty_out;
 
-	if (wbc->for_reclaim) {
-		f2fs_submit_merged_bio(sbi, DATA, WRITE);
-		need_balance_fs = false;
-	}
+	f2fs_lock_op(sbi);
+	if (f2fs_has_inline_data(inode) || f2fs_may_inline(inode))
+		err = f2fs_write_inline_data(inode, page, offset);
+	else
+		err = do_write_data_page(page, &fio);
+	f2fs_unlock_op(sbi);
+done:
+	if (err && err != -ENOENT)
+		goto redirty_out;
 
 	clear_cold_data(page);
 out:
@@ -849,12 +839,11 @@ static int f2fs_write_data_page(struct page *page,
 
 redirty_out:
 	wbc->pages_skipped++;
+	account_page_redirty(page);
 	set_page_dirty(page);
-	return err;
+	return AOP_WRITEPAGE_ACTIVATE;
 }
 
-#define MAX_DESIRED_PAGES_WP	4096
-
 static int __f2fs_writepage(struct page *page, struct writeback_control *wbc,
 			void *data)
 {
@@ -871,17 +860,17 @@ static int f2fs_write_data_pages(struct address_space *mapping,
 	struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb);
 	bool locked = false;
 	int ret;
-	long excess_nrtw = 0, desired_nrtw;
+	long diff;
 
 	/* deal with chardevs and other special file */
 	if (!mapping->a_ops->writepage)
 		return 0;
 
-	if (wbc->nr_to_write < MAX_DESIRED_PAGES_WP) {
-		desired_nrtw = MAX_DESIRED_PAGES_WP;
-		excess_nrtw = desired_nrtw - wbc->nr_to_write;
-		wbc->nr_to_write = desired_nrtw;
-	}
+	if (S_ISDIR(inode->i_mode) && wbc->sync_mode == WB_SYNC_NONE &&
+			get_dirty_dents(inode) < nr_pages_to_skip(sbi, DATA))
+		goto skip_write;
+
+	diff = nr_pages_to_write(sbi, DATA, wbc);
 
 	if (!S_ISDIR(inode->i_mode)) {
 		mutex_lock(&sbi->writepages);
@@ -895,8 +884,12 @@ static int f2fs_write_data_pages(struct address_space *mapping,
 
 	remove_dirty_dir_inode(inode);
 
-	wbc->nr_to_write -= excess_nrtw;
+	wbc->nr_to_write = max((long)0, wbc->nr_to_write - diff);
 	return ret;
+
+skip_write:
+	wbc->pages_skipped += get_dirty_dents(inode);
+	return 0;
 }
 
 static int f2fs_write_begin(struct file *file, struct address_space *mapping,
@@ -949,13 +942,19 @@ static int f2fs_write_begin(struct file *file, struct address_space *mapping,
 	if (dn.data_blkaddr == NEW_ADDR) {
 		zero_user_segment(page, 0, PAGE_CACHE_SIZE);
 	} else {
-		if (f2fs_has_inline_data(inode))
+		if (f2fs_has_inline_data(inode)) {
 			err = f2fs_read_inline_data(inode, page);
-		else
+			if (err) {
+				page_cache_release(page);
+				return err;
+			}
+		} else {
 			err = f2fs_submit_page_bio(sbi, page, dn.data_blkaddr,
 							READ_SYNC);
-		if (err)
-			return err;
+			if (err)
+				return err;
+		}
+
 		lock_page(page);
 		if (unlikely(!PageUptodate(page))) {
 			f2fs_put_page(page, 1);
@@ -1031,11 +1030,8 @@ static void f2fs_invalidate_data_page(struct page *page, unsigned int offset,
 				      unsigned int length)
 {
 	struct inode *inode = page->mapping->host;
-	struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb);
-	if (S_ISDIR(inode->i_mode) && PageDirty(page)) {
-		dec_page_count(sbi, F2FS_DIRTY_DENTS);
+	if (PageDirty(page))
 		inode_dec_dirty_dents(inode);
-	}
 	ClearPagePrivate(page);
 }
 

fs/f2fs/debug.c

@@ -86,7 +86,6 @@ static void update_sit_info(struct f2fs_sb_info *sbi)
 {
 	struct f2fs_stat_info *si = F2FS_STAT(sbi);
 	unsigned int blks_per_sec, hblks_per_sec, total_vblocks, bimodal, dist;
-	struct sit_info *sit_i = SIT_I(sbi);
 	unsigned int segno, vblocks;
 	int ndirty = 0;
 
@@ -94,7 +93,6 @@ static void update_sit_info(struct f2fs_sb_info *sbi)
 	total_vblocks = 0;
 	blks_per_sec = sbi->segs_per_sec * (1 << sbi->log_blocks_per_seg);
 	hblks_per_sec = blks_per_sec / 2;
-	mutex_lock(&sit_i->sentry_lock);
 	for (segno = 0; segno < TOTAL_SEGS(sbi); segno += sbi->segs_per_sec) {
 		vblocks = get_valid_blocks(sbi, segno, sbi->segs_per_sec);
 		dist = abs(vblocks - hblks_per_sec);
@@ -105,7 +103,6 @@ static void update_sit_info(struct f2fs_sb_info *sbi)
 			ndirty++;
 		}
 	}
-	mutex_unlock(&sit_i->sentry_lock);
 	dist = TOTAL_SECS(sbi) * hblks_per_sec * hblks_per_sec / 100;
 	si->bimodal = bimodal / dist;
 	if (si->dirty_count)
@@ -236,6 +233,7 @@ static int stat_show(struct seq_file *s, void *v)
 			   si->dirty_count);
 		seq_printf(s, "  - Prefree: %d\n  - Free: %d (%d)\n\n",
 			   si->prefree_count, si->free_segs, si->free_secs);
+		seq_printf(s, "CP calls: %d\n", si->cp_count);
 		seq_printf(s, "GC calls: %d (BG: %d)\n",
 			   si->call_count, si->bg_gc);
 		seq_printf(s, "  - data segments : %d\n", si->data_segs);
@@ -252,10 +250,10 @@ static int stat_show(struct seq_file *s, void *v)
 			   si->ndirty_dent, si->ndirty_dirs);
 		seq_printf(s, "  - meta: %4d in %4d\n",
 			   si->ndirty_meta, si->meta_pages);
-		seq_printf(s, "  - NATs: %5d > %lu\n",
-			   si->nats, NM_WOUT_THRESHOLD);
-		seq_printf(s, "  - SITs: %5d\n  - free_nids: %5d\n",
-			   si->sits, si->fnids);
+		seq_printf(s, "  - NATs: %9d\n  - SITs: %9d\n",
+			   si->nats, si->sits);
+		seq_printf(s, "  - free_nids: %9d\n",
+			   si->fnids);
 		seq_puts(s, "\nDistribution of User Blocks:");
 		seq_puts(s, " [ valid | invalid | free ]\n");
 		seq_puts(s, "  [");

fs/f2fs/dir.c

@@ -21,12 +21,12 @@ static unsigned long dir_blocks(struct inode *inode)
 							>> PAGE_CACHE_SHIFT;
 }
 
-static unsigned int dir_buckets(unsigned int level)
+static unsigned int dir_buckets(unsigned int level, int dir_level)
 {
 	if (level < MAX_DIR_HASH_DEPTH / 2)
-		return 1 << level;
+		return 1 << (level + dir_level);
 	else
-		return 1 << ((MAX_DIR_HASH_DEPTH / 2) - 1);
+		return 1 << ((MAX_DIR_HASH_DEPTH / 2 + dir_level) - 1);
 }
 
 static unsigned int bucket_blocks(unsigned int level)
@@ -65,13 +65,14 @@ static void set_de_type(struct f2fs_dir_entry *de, struct inode *inode)
 	de->file_type = f2fs_type_by_mode[(mode & S_IFMT) >> S_SHIFT];
 }
 
-static unsigned long dir_block_index(unsigned int level, unsigned int idx)
+static unsigned long dir_block_index(unsigned int level,
+				int dir_level, unsigned int idx)
 {
 	unsigned long i;
 	unsigned long bidx = 0;
 
 	for (i = 0; i < level; i++)
-		bidx += dir_buckets(i) * bucket_blocks(i);
+		bidx += dir_buckets(i, dir_level) * bucket_blocks(i);
 	bidx += idx * bucket_blocks(level);
 	return bidx;
 }
@@ -93,16 +94,21 @@ static struct f2fs_dir_entry *find_in_block(struct page *dentry_page,
 			f2fs_hash_t namehash, struct page **res_page)
 {
 	struct f2fs_dir_entry *de;
-	unsigned long bit_pos, end_pos, next_pos;
+	unsigned long bit_pos = 0;
 	struct f2fs_dentry_block *dentry_blk = kmap(dentry_page);
-	int slots;
+	const void *dentry_bits = &dentry_blk->dentry_bitmap;
+	int max_len = 0;
 
-	bit_pos = find_next_bit_le(&dentry_blk->dentry_bitmap,
-					NR_DENTRY_IN_BLOCK, 0);
 	while (bit_pos < NR_DENTRY_IN_BLOCK) {
+		if (!test_bit_le(bit_pos, dentry_bits)) {
+			if (bit_pos == 0)
+				max_len = 1;
+			else if (!test_bit_le(bit_pos - 1, dentry_bits))
+				max_len++;
+			bit_pos++;
+			continue;
+		}
 		de = &dentry_blk->dentry[bit_pos];
-		slots = GET_DENTRY_SLOTS(le16_to_cpu(de->name_len));
-
 		if (early_match_name(name, namelen, namehash, de)) {
 			if (!memcmp(dentry_blk->filename[bit_pos],
 							name, namelen)) {
@@ -110,20 +116,18 @@ static struct f2fs_dir_entry *find_in_block(struct page *dentry_page,
 				goto found;
 			}
 		}
-		next_pos = bit_pos + slots;
-		bit_pos = find_next_bit_le(&dentry_blk->dentry_bitmap,
-				NR_DENTRY_IN_BLOCK, next_pos);
-		if (bit_pos >= NR_DENTRY_IN_BLOCK)
-			end_pos = NR_DENTRY_IN_BLOCK;
-		else
-			end_pos = bit_pos;
-		if (*max_slots < end_pos - next_pos)
-			*max_slots = end_pos - next_pos;
+		if (max_len > *max_slots) {
+			*max_slots = max_len;
+			max_len = 0;
+		}
+		bit_pos += GET_DENTRY_SLOTS(le16_to_cpu(de->name_len));
 	}
 
 	de = NULL;
 	kunmap(dentry_page);
 found:
+	if (max_len > *max_slots)
+		*max_slots = max_len;
 	return de;
 }
 
@@ -141,10 +145,11 @@ static struct f2fs_dir_entry *find_in_level(struct inode *dir,
 
 	f2fs_bug_on(level > MAX_DIR_HASH_DEPTH);
 
-	nbucket = dir_buckets(level);
+	nbucket = dir_buckets(level, F2FS_I(dir)->i_dir_level);
 	nblock = bucket_blocks(level);
 
-	bidx = dir_block_index(level, le32_to_cpu(namehash) % nbucket);
+	bidx = dir_block_index(level, F2FS_I(dir)->i_dir_level,
+					le32_to_cpu(namehash) % nbucket);
 	end_block = bidx + nblock;
 
 	for (; bidx < end_block; bidx++) {
@@ -248,7 +253,7 @@ void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
 		struct page *page, struct inode *inode)
 {
 	lock_page(page);
-	wait_on_page_writeback(page);
+	f2fs_wait_on_page_writeback(page, DATA);
 	de->ino = cpu_to_le32(inode->i_ino);
 	set_de_type(de, inode);
 	kunmap(page);
@@ -347,14 +352,11 @@ static struct page *init_inode_metadata(struct inode *inode,
 		err = f2fs_init_security(inode, dir, name, page);
 		if (err)
 			goto put_error;
-
-		wait_on_page_writeback(page);
 	} else {
 		page = get_node_page(F2FS_SB(dir->i_sb), inode->i_ino);
 		if (IS_ERR(page))
 			return page;
 
-		wait_on_page_writeback(page);
 		set_cold_node(inode, page);
 	}
 
@@ -372,6 +374,10 @@ static struct page *init_inode_metadata(struct inode *inode,
 
 put_error:
 	f2fs_put_page(page, 1);
+	/* once the failed inode becomes a bad inode, i_mode is S_IFREG */
+	truncate_inode_pages(&inode->i_data, 0);
+	truncate_blocks(inode, 0);
+	remove_dirty_dir_inode(inode);
 error:
 	remove_inode_page(inode);
 	return ERR_PTR(err);
@@ -395,9 +401,6 @@ static void update_parent_metadata(struct inode *dir, struct inode *inode,
 		set_inode_flag(F2FS_I(dir), FI_UPDATE_DIR);
 	}
 
-	if (is_inode_flag_set(F2FS_I(dir), FI_UPDATE_DIR))
-		update_inode_page(dir);
-
 	if (is_inode_flag_set(F2FS_I(inode), FI_INC_LINK))
 		clear_inode_flag(F2FS_I(inode), FI_INC_LINK);
 }
@@ -464,10 +467,11 @@ int __f2fs_add_link(struct inode *dir, const struct qstr *name,
 	if (level == current_depth)
 		++current_depth;
 
-	nbucket = dir_buckets(level);
+	nbucket = dir_buckets(level, F2FS_I(dir)->i_dir_level);
 	nblock = bucket_blocks(level);
 
-	bidx = dir_block_index(level, (le32_to_cpu(dentry_hash) % nbucket));
+	bidx = dir_block_index(level, F2FS_I(dir)->i_dir_level,
+				(le32_to_cpu(dentry_hash) % nbucket));
 
 	for (block = bidx; block <= (bidx + nblock - 1); block++) {
 		dentry_page = get_new_data_page(dir, NULL, block, true);
@@ -487,8 +491,9 @@ int __f2fs_add_link(struct inode *dir, const struct qstr *name,
 	++level;
 	goto start;
 add_dentry:
-	wait_on_page_writeback(dentry_page);
+	f2fs_wait_on_page_writeback(dentry_page, DATA);
 
+	down_write(&F2FS_I(inode)->i_sem);
 	page = init_inode_metadata(inode, dir, name);
 	if (IS_ERR(page)) {
 		err = PTR_ERR(page);
@@ -511,7 +516,12 @@ int __f2fs_add_link(struct inode *dir, const struct qstr *name,
 
 	update_parent_metadata(dir, inode, current_depth);
 fail:
-	clear_inode_flag(F2FS_I(dir), FI_UPDATE_DIR);
+	up_write(&F2FS_I(inode)->i_sem);
+
+	if (is_inode_flag_set(F2FS_I(dir), FI_UPDATE_DIR)) {
+		update_inode_page(dir);
+		clear_inode_flag(F2FS_I(dir), FI_UPDATE_DIR);
+	}
 	kunmap(dentry_page);
 	f2fs_put_page(dentry_page, 1);
 	return err;
@@ -528,13 +538,12 @@ void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
 	unsigned int bit_pos;
 	struct address_space *mapping = page->mapping;
 	struct inode *dir = mapping->host;
-	struct f2fs_sb_info *sbi = F2FS_SB(dir->i_sb);
 	int slots = GET_DENTRY_SLOTS(le16_to_cpu(dentry->name_len));
 	void *kaddr = page_address(page);
 	int i;
 
 	lock_page(page);
-	wait_on_page_writeback(page);
+	f2fs_wait_on_page_writeback(page, DATA);
 
 	dentry_blk = (struct f2fs_dentry_block *)kaddr;
 	bit_pos = dentry - (struct f2fs_dir_entry *)dentry_blk->dentry;
@@ -551,6 +560,10 @@ void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
 	dir->i_ctime = dir->i_mtime = CURRENT_TIME;
 
 	if (inode) {
+		struct f2fs_sb_info *sbi = F2FS_SB(dir->i_sb);
+
+		down_write(&F2FS_I(inode)->i_sem);
+
 		if (S_ISDIR(inode->i_mode)) {
 			drop_nlink(dir);
 			update_inode_page(dir);
@@ -561,6 +574,7 @@ void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
 			drop_nlink(inode);
 			i_size_write(inode, 0);
 		}
+		up_write(&F2FS_I(inode)->i_sem);
 		update_inode_page(inode);
 
 		if (inode->i_nlink == 0)
@@ -573,7 +587,6 @@ void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
 		truncate_hole(dir, page->index, page->index + 1);
 		clear_page_dirty_for_io(page);
 		ClearPageUptodate(page);
-		dec_page_count(sbi, F2FS_DIRTY_DENTS);
 		inode_dec_dirty_dents(dir);
 	}
 	f2fs_put_page(page, 1);

fs/f2fs/file.c

@@ -76,7 +76,7 @@ static int f2fs_vm_page_mkwrite(struct vm_area_struct *vma,
 	trace_f2fs_vm_page_mkwrite(page, DATA);
 mapped:
 	/* fill the page */
-	wait_on_page_writeback(page);
+	f2fs_wait_on_page_writeback(page, DATA);
 out:
 	sb_end_pagefault(inode->i_sb);
 	return block_page_mkwrite_return(err);
@@ -111,11 +111,12 @@ static int get_parent_ino(struct inode *inode, nid_t *pino)
 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 {
 	struct inode *inode = file->f_mapping->host;
+	struct f2fs_inode_info *fi = F2FS_I(inode);
 	struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb);
 	int ret = 0;
 	bool need_cp = false;
 	struct writeback_control wbc = {
-		.sync_mode = WB_SYNC_NONE,
+		.sync_mode = WB_SYNC_ALL,
 		.nr_to_write = LONG_MAX,
 		.for_reclaim = 0,
 	};
@@ -133,7 +134,7 @@ int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	/* guarantee free sections for fsync */
 	f2fs_balance_fs(sbi);
 
-	mutex_lock(&inode->i_mutex);
+	down_read(&fi->i_sem);
 
 	/*
 	 * Both of fdatasync() and fsync() are able to be recovered from
@@ -150,25 +151,33 @@ int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	else if (F2FS_I(inode)->xattr_ver == cur_cp_version(F2FS_CKPT(sbi)))
 		need_cp = true;
 
+	up_read(&fi->i_sem);
+
 	if (need_cp) {
 		nid_t pino;
 
-		F2FS_I(inode)->xattr_ver = 0;
-
 		/* all the dirty node pages should be flushed for POR */
 		ret = f2fs_sync_fs(inode->i_sb, 1);
+
+		down_write(&fi->i_sem);
+		F2FS_I(inode)->xattr_ver = 0;
 		if (file_wrong_pino(inode) && inode->i_nlink == 1 &&
 					get_parent_ino(inode, &pino)) {
 			F2FS_I(inode)->i_pino = pino;
 			file_got_pino(inode);
+			up_write(&fi->i_sem);
 			mark_inode_dirty_sync(inode);
 			ret = f2fs_write_inode(inode, NULL);
 			if (ret)
 				goto out;
+		} else {
+			up_write(&fi->i_sem);
 		}
 	} else {
 		/* if there is no written node page, write its inode page */
 		while (!sync_node_pages(sbi, inode->i_ino, &wbc)) {
+			if (fsync_mark_done(sbi, inode->i_ino))
+				goto out;
 			mark_inode_dirty_sync(inode);
 			ret = f2fs_write_inode(inode, NULL);
 			if (ret)
@@ -177,10 +186,9 @@ int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 		ret = wait_on_node_pages_writeback(sbi, inode->i_ino);
 		if (ret)
 			goto out;
-		ret = blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL, NULL);
+		ret = f2fs_issue_flush(F2FS_SB(inode->i_sb));
 	}
 out:
-	mutex_unlock(&inode->i_mutex);
 	trace_f2fs_sync_file_exit(inode, need_cp, datasync, ret);
 	return ret;
 }
@@ -245,7 +253,7 @@ static void truncate_partial_data_page(struct inode *inode, u64 from)
 		f2fs_put_page(page, 1);
 		return;
 	}
-	wait_on_page_writeback(page);
+	f2fs_wait_on_page_writeback(page, DATA);
 	zero_user(page, offset, PAGE_CACHE_SIZE - offset);
 	set_page_dirty(page);
 	f2fs_put_page(page, 1);
@@ -422,7 +430,7 @@ static void fill_zero(struct inode *inode, pgoff_t index,
 	f2fs_unlock_op(sbi);
 
 	if (!IS_ERR(page)) {
-		wait_on_page_writeback(page);
+		f2fs_wait_on_page_writeback(page, DATA);
 		zero_user(page, start, len);
 		set_page_dirty(page);
 		f2fs_put_page(page, 1);
@@ -560,6 +568,8 @@ static long f2fs_fallocate(struct file *file, int mode,
 	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
 		return -EOPNOTSUPP;
 
+	mutex_lock(&inode->i_mutex);
+
 	if (mode & FALLOC_FL_PUNCH_HOLE)
 		ret = punch_hole(inode, offset, len);
 	else
@@ -569,6 +579,9 @@ static long f2fs_fallocate(struct file *file, int mode,
 		inode->i_mtime = inode->i_ctime = CURRENT_TIME;
 		mark_inode_dirty(inode);
 	}
+
+	mutex_unlock(&inode->i_mutex);
+
 	trace_f2fs_fallocate(inode, mode, offset, len, ret);
 	return ret;
 }

fs/f2fs/gc.c

@@ -531,15 +531,10 @@ static void move_data_page(struct inode *inode, struct page *page, int gc_type)
 		set_page_dirty(page);
 		set_cold_data(page);
 	} else {
-		struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb);
-
 		f2fs_wait_on_page_writeback(page, DATA);
 
-		if (clear_page_dirty_for_io(page) &&
-			S_ISDIR(inode->i_mode)) {
-			dec_page_count(sbi, F2FS_DIRTY_DENTS);
+		if (clear_page_dirty_for_io(page))
 			inode_dec_dirty_dents(inode);
-		}
 		set_cold_data(page);
 		do_write_data_page(page, &fio);
 		clear_cold_data(page);
@@ -701,6 +696,8 @@ int f2fs_gc(struct f2fs_sb_info *sbi)
 gc_more:
 	if (unlikely(!(sbi->sb->s_flags & MS_ACTIVE)))
 		goto stop;
+	if (unlikely(is_set_ckpt_flags(F2FS_CKPT(sbi), CP_ERROR_FLAG)))
+		goto stop;
 
 	if (gc_type == BG_GC && has_not_enough_free_secs(sbi, nfree)) {
 		gc_type = FG_GC;
@@ -711,6 +708,11 @@ int f2fs_gc(struct f2fs_sb_info *sbi)
 		goto stop;
 	ret = 0;
 
+	/* readahead multi ssa blocks those have contiguous address */
+	if (sbi->segs_per_sec > 1)
+		ra_meta_pages(sbi, GET_SUM_BLOCK(sbi, segno), sbi->segs_per_sec,
+								META_SSA);
+
 	for (i = 0; i < sbi->segs_per_sec; i++)
 		do_garbage_collect(sbi, segno + i, &ilist, gc_type);
 
@@ -740,7 +742,7 @@ void build_gc_manager(struct f2fs_sb_info *sbi)
 int __init create_gc_caches(void)
 {
 	winode_slab = f2fs_kmem_cache_create("f2fs_gc_inodes",
-			sizeof(struct inode_entry), NULL);
+			sizeof(struct inode_entry));
 	if (!winode_slab)
 		return -ENOMEM;
 	return 0;

fs/f2fs/inline.c

@@ -45,8 +45,10 @@ int f2fs_read_inline_data(struct inode *inode, struct page *page)
 	}
 
 	ipage = get_node_page(sbi, inode->i_ino);
-	if (IS_ERR(ipage))
+	if (IS_ERR(ipage)) {
+		unlock_page(page);
 		return PTR_ERR(ipage);
+	}
 
 	zero_user_segment(page, MAX_INLINE_DATA, PAGE_CACHE_SIZE);
 

fs/f2fs/inode.c

@@ -107,6 +107,7 @@ static int do_read_inode(struct inode *inode)
 	fi->flags = 0;
 	fi->i_advise = ri->i_advise;
 	fi->i_pino = le32_to_cpu(ri->i_pino);
+	fi->i_dir_level = ri->i_dir_level;
 
 	get_extent_info(&fi->ext, ri->i_ext);
 	get_inline_info(fi, ri);
@@ -204,6 +205,7 @@ void update_inode(struct inode *inode, struct page *node_page)
 	ri->i_flags = cpu_to_le32(F2FS_I(inode)->i_flags);
 	ri->i_pino = cpu_to_le32(F2FS_I(inode)->i_pino);
 	ri->i_generation = cpu_to_le32(inode->i_generation);
+	ri->i_dir_level = F2FS_I(inode)->i_dir_level;
 
 	__set_inode_rdev(inode, ri);
 	set_cold_node(inode, node_page);
@@ -212,24 +214,29 @@ void update_inode(struct inode *inode, struct page *node_page)
 	clear_inode_flag(F2FS_I(inode), FI_DIRTY_INODE);
 }
 
-int update_inode_page(struct inode *inode)
+void update_inode_page(struct inode *inode)
 {
 	struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb);
 	struct page *node_page;
-
+retry:
 	node_page = get_node_page(sbi, inode->i_ino);
-	if (IS_ERR(node_page))
-		return PTR_ERR(node_page);
-
+	if (IS_ERR(node_page)) {
+		int err = PTR_ERR(node_page);
+		if (err == -ENOMEM) {
+			cond_resched();
+			goto retry;
+		} else if (err != -ENOENT) {
+			f2fs_stop_checkpoint(sbi);
+		}
+		return;
+	}
 	update_inode(inode, node_page);
 	f2fs_put_page(node_page, 1);
-	return 0;
 }
 
 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc)
 {
 	struct f2fs_sb_info *sbi = F2FS_SB(inode->i_sb);
-	int ret;
 
 	if (inode->i_ino == F2FS_NODE_INO(sbi) ||
 			inode->i_ino == F2FS_META_INO(sbi))
@@ -243,13 +250,13 @@ int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc)
 	 * during the urgent cleaning time when runing out of free sections.
 	 */
 	f2fs_lock_op(sbi);
-	ret = update_inode_page(inode);
+	update_inode_page(inode);
 	f2fs_unlock_op(sbi);
 
 	if (wbc)
 		f2fs_balance_fs(sbi);
 
-	return ret;
+	return 0;
 }
 
 /*
@@ -266,7 +273,7 @@ void f2fs_evict_inode(struct inode *inode)
 			inode->i_ino == F2FS_META_INO(sbi))
 		goto no_delete;
 
-	f2fs_bug_on(atomic_read(&F2FS_I(inode)->dirty_dents));
+	f2fs_bug_on(get_dirty_dents(inode));
 	remove_dirty_dir_inode(inode);
 
 	if (inode->i_nlink || is_bad_inode(inode))

fs/f2fs/namei.c

@@ -207,6 +207,8 @@ static struct dentry *f2fs_lookup(struct inode *dir, struct dentry *dentry,
 		inode = f2fs_iget(dir->i_sb, ino);
 		if (IS_ERR(inode))
 			return ERR_CAST(inode);
+
+		stat_inc_inline_inode(inode);
 	}
 
 	return d_splice_alias(inode, dentry);
@@ -424,12 +426,17 @@ static int f2fs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		}
 
 		f2fs_set_link(new_dir, new_entry, new_page, old_inode);
+		down_write(&F2FS_I(old_inode)->i_sem);
 		F2FS_I(old_inode)->i_pino = new_dir->i_ino;
+		up_write(&F2FS_I(old_inode)->i_sem);
 
 		new_inode->i_ctime = CURRENT_TIME;
+		down_write(&F2FS_I(new_inode)->i_sem);
 		if (old_dir_entry)
 			drop_nlink(new_inode);
 		drop_nlink(new_inode);
+		up_write(&F2FS_I(new_inode)->i_sem);
+
 		mark_inode_dirty(new_inode);
 
 		if (!new_inode->i_nlink)
@@ -459,7 +466,9 @@ static int f2fs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		if (old_dir != new_dir) {
 			f2fs_set_link(old_inode, old_dir_entry,
 						old_dir_page, new_dir);
+			down_write(&F2FS_I(old_inode)->i_sem);
 			F2FS_I(old_inode)->i_pino = new_dir->i_ino;
+			up_write(&F2FS_I(old_inode)->i_sem);
 			update_inode_page(old_inode);
 		} else {
 			kunmap(old_dir_page);

fs/f2fs/node.h

@@ -17,14 +17,11 @@
 /* # of pages to perform readahead before building free nids */
 #define FREE_NID_PAGES 4
 
-/* maximum # of free node ids to produce during build_free_nids */
-#define MAX_FREE_NIDS (NAT_ENTRY_PER_BLOCK * FREE_NID_PAGES)
-
 /* maximum readahead size for node during getting data blocks */
 #define MAX_RA_NODE		128
 
-/* maximum cached nat entries to manage memory footprint */
-#define NM_WOUT_THRESHOLD	(64 * NAT_ENTRY_PER_BLOCK)
+/* control the memory footprint threshold (10MB per 1GB ram) */
+#define DEF_RAM_THRESHOLD	10
 
 /* vector size for gang look-up from nat cache that consists of radix tree */
 #define NATVEC_SIZE	64
@@ -45,6 +42,7 @@ struct node_info {
 struct nat_entry {
 	struct list_head list;	/* for clean or dirty nat list */
 	bool checkpointed;	/* whether it is checkpointed or not */
+	bool fsync_done;	/* whether the latest node has fsync mark */
 	struct node_info ni;	/* in-memory node information */
 };
 
@@ -58,9 +56,15 @@ struct nat_entry {
 #define nat_set_version(nat, v)		(nat->ni.version = v)
 
 #define __set_nat_cache_dirty(nm_i, ne)					\
-	list_move_tail(&ne->list, &nm_i->dirty_nat_entries);
+	do {								\
+		ne->checkpointed = false;				\
+		list_move_tail(&ne->list, &nm_i->dirty_nat_entries);	\
+	} while (0);
 #define __clear_nat_cache_dirty(nm_i, ne)				\
-	list_move_tail(&ne->list, &nm_i->nat_entries);
+	do {								\
+		ne->checkpointed = true;				\
+		list_move_tail(&ne->list, &nm_i->nat_entries);		\
+	} while (0);
 #define inc_node_version(version)	(++version)
 
 static inline void node_info_from_raw_nat(struct node_info *ni,
@@ -71,6 +75,11 @@ static inline void node_info_from_raw_nat(struct node_info *ni,
 	ni->version = raw_ne->version;
 }
 
+enum nid_type {
+	FREE_NIDS,	/* indicates the free nid list */
+	NAT_ENTRIES	/* indicates the cached nat entry */
+};
+
 /*
  * For free nid mangement
  */
@@ -236,7 +245,7 @@ static inline bool IS_DNODE(struct page *node_page)
 {
 	unsigned int ofs = ofs_of_node(node_page);
 
-	if (ofs == XATTR_NODE_OFFSET)
+	if (f2fs_has_xattr_block(ofs))
 		return false;
 
 	if (ofs == 3 || ofs == 4 + NIDS_PER_BLOCK ||

fs/f2fs/recovery.c

@@ -27,14 +27,12 @@ bool space_for_roll_forward(struct f2fs_sb_info *sbi)
 static struct fsync_inode_entry *get_fsync_inode(struct list_head *head,
 								nid_t ino)
 {
-	struct list_head *this;
 	struct fsync_inode_entry *entry;
 
-	list_for_each(this, head) {
-		entry = list_entry(this, struct fsync_inode_entry, list);
+	list_for_each_entry(entry, head, list)
 		if (entry->inode->i_ino == ino)
 			return entry;
-	}
+
 	return NULL;
 }
 
@@ -136,7 +134,7 @@ static int find_fsync_dnodes(struct f2fs_sb_info *sbi, struct list_head *head)
 
 	/* get node pages in the current segment */
 	curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
-	blkaddr = START_BLOCK(sbi, curseg->segno) + curseg->next_blkoff;
+	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
 
 	/* read node page */
 	page = alloc_page(GFP_F2FS_ZERO);
@@ -218,13 +216,12 @@ static int check_index_in_prev_nodes(struct f2fs_sb_info *sbi,
 {
 	struct seg_entry *sentry;
 	unsigned int segno = GET_SEGNO(sbi, blkaddr);
-	unsigned short blkoff = GET_SEGOFF_FROM_SEG0(sbi, blkaddr) &
-					(sbi->blocks_per_seg - 1);
+	unsigned short blkoff = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
+	struct f2fs_summary_block *sum_node;
 	struct f2fs_summary sum;
+	struct page *sum_page, *node_page;
 	nid_t ino, nid;
-	void *kaddr;
 	struct inode *inode;
-	struct page *node_page;
 	unsigned int offset;
 	block_t bidx;
 	int i;
@@ -238,18 +235,15 @@ static int check_index_in_prev_nodes(struct f2fs_sb_info *sbi,
 		struct curseg_info *curseg = CURSEG_I(sbi, i);
 		if (curseg->segno == segno) {
 			sum = curseg->sum_blk->entries[blkoff];
-			break;
+			goto got_it;
 		}
 	}
-	if (i > CURSEG_COLD_DATA) {
-		struct page *sum_page = get_sum_page(sbi, segno);
-		struct f2fs_summary_block *sum_node;
-		kaddr = page_address(sum_page);
-		sum_node = (struct f2fs_summary_block *)kaddr;
-		sum = sum_node->entries[blkoff];
-		f2fs_put_page(sum_page, 1);
-	}
 
+	sum_page = get_sum_page(sbi, segno);
+	sum_node = (struct f2fs_summary_block *)page_address(sum_page);
+	sum = sum_node->entries[blkoff];
+	f2fs_put_page(sum_page, 1);
+got_it:
 	/* Use the locked dnode page and inode */
 	nid = le32_to_cpu(sum.nid);
 	if (dn->inode->i_ino == nid) {
@@ -301,6 +295,9 @@ static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
 	if (recover_inline_data(inode, page))
 		goto out;
 
+	if (recover_xattr_data(inode, page, blkaddr))
+		goto out;
+
 	start = start_bidx_of_node(ofs_of_node(page), fi);
 	if (IS_INODE(page))
 		end = start + ADDRS_PER_INODE(fi);
@@ -317,7 +314,7 @@ static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
 		goto out;
 	}
 
-	wait_on_page_writeback(dn.node_page);
+	f2fs_wait_on_page_writeback(dn.node_page, NODE);
 
 	get_node_info(sbi, dn.nid, &ni);
 	f2fs_bug_on(ni.ino != ino_of_node(page));
@@ -437,7 +434,7 @@ int recover_fsync_data(struct f2fs_sb_info *sbi)
 	bool need_writecp = false;
 
 	fsync_entry_slab = f2fs_kmem_cache_create("f2fs_fsync_inode_entry",
-			sizeof(struct fsync_inode_entry), NULL);
+			sizeof(struct fsync_inode_entry));
 	if (!fsync_entry_slab)
 		return -ENOMEM;
 

fs/f2fs/segment.h

@@ -14,7 +14,7 @@
 #define NULL_SEGNO			((unsigned int)(~0))
 #define NULL_SECNO			((unsigned int)(~0))
 
-#define DEF_RECLAIM_PREFREE_SEGMENTS	100	/* 200MB of prefree segments */
+#define DEF_RECLAIM_PREFREE_SEGMENTS	5	/* 5% over total segments */
 
 /* L: Logical segment # in volume, R: Relative segment # in main area */
 #define GET_L2R_SEGNO(free_i, segno)	(segno - free_i->start_segno)
@@ -57,6 +57,9 @@
 	((blk_addr) - SM_I(sbi)->seg0_blkaddr)
 #define GET_SEGNO_FROM_SEG0(sbi, blk_addr)				\
 	(GET_SEGOFF_FROM_SEG0(sbi, blk_addr) >> sbi->log_blocks_per_seg)
+#define GET_BLKOFF_FROM_SEG0(sbi, blk_addr)				\
+	(GET_SEGOFF_FROM_SEG0(sbi, blk_addr) & (sbi->blocks_per_seg - 1))
+
 #define GET_SEGNO(sbi, blk_addr)					\
 	(((blk_addr == NULL_ADDR) || (blk_addr == NEW_ADDR)) ?		\
 	NULL_SEGNO : GET_L2R_SEGNO(FREE_I(sbi),			\
@@ -377,26 +380,12 @@ static inline void get_sit_bitmap(struct f2fs_sb_info *sbi,
 
 static inline block_t written_block_count(struct f2fs_sb_info *sbi)
 {
-	struct sit_info *sit_i = SIT_I(sbi);
-	block_t vblocks;
-
-	mutex_lock(&sit_i->sentry_lock);
-	vblocks = sit_i->written_valid_blocks;
-	mutex_unlock(&sit_i->sentry_lock);
-
-	return vblocks;
+	return SIT_I(sbi)->written_valid_blocks;
 }
 
 static inline unsigned int free_segments(struct f2fs_sb_info *sbi)
 {
-	struct free_segmap_info *free_i = FREE_I(sbi);
-	unsigned int free_segs;
-
-	read_lock(&free_i->segmap_lock);
-	free_segs = free_i->free_segments;
-	read_unlock(&free_i->segmap_lock);
-
-	return free_segs;
+	return FREE_I(sbi)->free_segments;
 }
 
 static inline int reserved_segments(struct f2fs_sb_info *sbi)
@@ -406,14 +395,7 @@ static inline int reserved_segments(struct f2fs_sb_info *sbi)
 
 static inline unsigned int free_sections(struct f2fs_sb_info *sbi)
 {
-	struct free_segmap_info *free_i = FREE_I(sbi);
-	unsigned int free_secs;
-
-	read_lock(&free_i->segmap_lock);
-	free_secs = free_i->free_sections;
-	read_unlock(&free_i->segmap_lock);
-
-	return free_secs;
+	return FREE_I(sbi)->free_sections;
 }
 
 static inline unsigned int prefree_segments(struct f2fs_sb_info *sbi)
@@ -682,3 +664,46 @@ static inline unsigned int max_hw_blocks(struct f2fs_sb_info *sbi)
 	struct request_queue *q = bdev_get_queue(bdev);
 	return SECTOR_TO_BLOCK(sbi, queue_max_sectors(q));
 }
+
+/*
+ * It is very important to gather dirty pages and write at once, so that we can
+ * submit a big bio without interfering other data writes.
+ * By default, 512 pages for directory data,
+ * 512 pages (2MB) * 3 for three types of nodes, and
+ * max_bio_blocks for meta are set.
+ */
+static inline int nr_pages_to_skip(struct f2fs_sb_info *sbi, int type)
+{
+	if (type == DATA)
+		return sbi->blocks_per_seg;
+	else if (type == NODE)
+		return 3 * sbi->blocks_per_seg;
+	else if (type == META)
+		return MAX_BIO_BLOCKS(max_hw_blocks(sbi));
+	else
+		return 0;
+}
+
+/*
+ * When writing pages, it'd better align nr_to_write for segment size.
+ */
+static inline long nr_pages_to_write(struct f2fs_sb_info *sbi, int type,
+					struct writeback_control *wbc)
+{
+	long nr_to_write, desired;
+
+	if (wbc->sync_mode != WB_SYNC_NONE)
+		return 0;
+
+	nr_to_write = wbc->nr_to_write;
+
+	if (type == DATA)
+		desired = 4096;
+	else if (type == NODE)
+		desired = 3 * max_hw_blocks(sbi);
+	else
+		desired = MAX_BIO_BLOCKS(max_hw_blocks(sbi));
+
+	wbc->nr_to_write = desired;
+	return desired - nr_to_write;
+}

fs/f2fs/super.c

@@ -51,6 +51,7 @@ enum {
 	Opt_disable_ext_identify,
 	Opt_inline_xattr,
 	Opt_inline_data,
+	Opt_flush_merge,
 	Opt_err,
 };
 
@@ -67,6 +68,7 @@ static match_table_t f2fs_tokens = {
 	{Opt_disable_ext_identify, "disable_ext_identify"},
 	{Opt_inline_xattr, "inline_xattr"},
 	{Opt_inline_data, "inline_data"},
+	{Opt_flush_merge, "flush_merge"},
 	{Opt_err, NULL},
 };
 
@@ -74,6 +76,7 @@ static match_table_t f2fs_tokens = {
 enum {
 	GC_THREAD,	/* struct f2fs_gc_thread */
 	SM_INFO,	/* struct f2fs_sm_info */
+	NM_INFO,	/* struct f2fs_nm_info */
 	F2FS_SBI,	/* struct f2fs_sb_info */
 };
 
@@ -92,6 +95,8 @@ static unsigned char *__struct_ptr(struct f2fs_sb_info *sbi, int struct_type)
 		return (unsigned char *)sbi->gc_thread;
 	else if (struct_type == SM_INFO)
 		return (unsigned char *)SM_I(sbi);
+	else if (struct_type == NM_INFO)
+		return (unsigned char *)NM_I(sbi);
 	else if (struct_type == F2FS_SBI)
 		return (unsigned char *)sbi;
 	return NULL;
@@ -183,7 +188,9 @@ F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, reclaim_segments, rec_prefree_segments);
 F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, max_small_discards, max_discards);
 F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, ipu_policy, ipu_policy);
 F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_ipu_util, min_ipu_util);
+F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ram_thresh, ram_thresh);
 F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, max_victim_search, max_victim_search);
+F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, dir_level, dir_level);
 
 #define ATTR_LIST(name) (&f2fs_attr_##name.attr)
 static struct attribute *f2fs_attrs[] = {
@@ -196,6 +203,8 @@ static struct attribute *f2fs_attrs[] = {
 	ATTR_LIST(ipu_policy),
 	ATTR_LIST(min_ipu_util),
 	ATTR_LIST(max_victim_search),
+	ATTR_LIST(dir_level),
+	ATTR_LIST(ram_thresh),
 	NULL,
 };
 
@@ -256,9 +265,9 @@ static int parse_options(struct super_block *sb, char *options)
 
 			if (!name)
 				return -ENOMEM;
-			if (!strncmp(name, "on", 2))
+			if (strlen(name) == 2 && !strncmp(name, "on", 2))
 				set_opt(sbi, BG_GC);
-			else if (!strncmp(name, "off", 3))
+			else if (strlen(name) == 3 && !strncmp(name, "off", 3))
 				clear_opt(sbi, BG_GC);
 			else {
 				kfree(name);
@@ -327,6 +336,9 @@ static int parse_options(struct super_block *sb, char *options)
 		case Opt_inline_data:
 			set_opt(sbi, INLINE_DATA);
 			break;
+		case Opt_flush_merge:
+			set_opt(sbi, FLUSH_MERGE);
+			break;
 		default:
 			f2fs_msg(sb, KERN_ERR,
 				"Unrecognized mount option \"%s\" or missing value",
@@ -353,12 +365,16 @@ static struct inode *f2fs_alloc_inode(struct super_block *sb)
 	fi->i_current_depth = 1;
 	fi->i_advise = 0;
 	rwlock_init(&fi->ext.ext_lock);
+	init_rwsem(&fi->i_sem);
 
 	set_inode_flag(fi, FI_NEW_INODE);
 
 	if (test_opt(F2FS_SB(sb), INLINE_XATTR))
 		set_inode_flag(fi, FI_INLINE_XATTR);
 
+	/* Will be used by directory only */
+	fi->i_dir_level = F2FS_SB(sb)->dir_level;
+
 	return &fi->vfs_inode;
 }
 
@@ -526,6 +542,8 @@ static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
 		seq_puts(seq, ",disable_ext_identify");
 	if (test_opt(sbi, INLINE_DATA))
 		seq_puts(seq, ",inline_data");
+	if (test_opt(sbi, FLUSH_MERGE))
+		seq_puts(seq, ",flush_merge");
 	seq_printf(seq, ",active_logs=%u", sbi->active_logs);
 
 	return 0;
@@ -539,13 +557,22 @@ static int segment_info_seq_show(struct seq_file *seq, void *offset)
 			le32_to_cpu(sbi->raw_super->segment_count_main);
 	int i;
 
+	seq_puts(seq, "format: segment_type|valid_blocks\n"
+		"segment_type(0:HD, 1:WD, 2:CD, 3:HN, 4:WN, 5:CN)\n");
+
 	for (i = 0; i < total_segs; i++) {
-		seq_printf(seq, "%u", get_valid_blocks(sbi, i, 1));
-		if (i != 0 && (i % 10) == 0)
-			seq_puts(seq, "\n");
+		struct seg_entry *se = get_seg_entry(sbi, i);
+
+		if ((i % 10) == 0)
+			seq_printf(seq, "%-5d", i);
+		seq_printf(seq, "%d|%-3u", se->type,
+					get_valid_blocks(sbi, i, 1));
+		if ((i % 10) == 9 || i == (total_segs - 1))
+			seq_putc(seq, '\n');
 		else
-			seq_puts(seq, " ");
+			seq_putc(seq, ' ');
 	}
+
 	return 0;
 }
 
@@ -640,6 +667,8 @@ static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
 
 	if (unlikely(ino < F2FS_ROOT_INO(sbi)))
 		return ERR_PTR(-ESTALE);
+	if (unlikely(ino >= NM_I(sbi)->max_nid))
+		return ERR_PTR(-ESTALE);
 
 	/*
 	 * f2fs_iget isn't quite right if the inode is currently unallocated!
@@ -787,6 +816,8 @@ static void init_sb_info(struct f2fs_sb_info *sbi)
 
 	for (i = 0; i < NR_COUNT_TYPE; i++)
 		atomic_set(&sbi->nr_pages[i], 0);
+
+	sbi->dir_level = DEF_DIR_LEVEL;
 }
 
 /*
@@ -898,11 +929,11 @@ static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
 	sbi->por_doing = false;
 	spin_lock_init(&sbi->stat_lock);
 
-	mutex_init(&sbi->read_io.io_mutex);
+	init_rwsem(&sbi->read_io.io_rwsem);
 	sbi->read_io.sbi = sbi;
 	sbi->read_io.bio = NULL;
 	for (i = 0; i < NR_PAGE_TYPE; i++) {
-		mutex_init(&sbi->write_io[i].io_mutex);
+		init_rwsem(&sbi->write_io[i].io_rwsem);
 		sbi->write_io[i].sbi = sbi;
 		sbi->write_io[i].bio = NULL;
 	}
@@ -991,28 +1022,9 @@ static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
 		goto free_root_inode;
 	}
 
-	/* recover fsynced data */
-	if (!test_opt(sbi, DISABLE_ROLL_FORWARD)) {
-		err = recover_fsync_data(sbi);
-		if (err)
-			f2fs_msg(sb, KERN_ERR,
-				"Cannot recover all fsync data errno=%ld", err);
-	}
-
-	/*
-	 * If filesystem is not mounted as read-only then
-	 * do start the gc_thread.
-	 */
-	if (!(sb->s_flags & MS_RDONLY)) {
-		/* After POR, we can run background GC thread.*/
-		err = start_gc_thread(sbi);
-		if (err)
-			goto free_gc;
-	}
-
 	err = f2fs_build_stats(sbi);
 	if (err)
-		goto free_gc;
+		goto free_root_inode;
 
 	if (f2fs_proc_root)
 		sbi->s_proc = proc_mkdir(sb->s_id, f2fs_proc_root);
@@ -1034,17 +1046,36 @@ static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
 	err = kobject_init_and_add(&sbi->s_kobj, &f2fs_ktype, NULL,
 							"%s", sb->s_id);
 	if (err)
-		goto fail;
+		goto free_proc;
+
+	/* recover fsynced data */
+	if (!test_opt(sbi, DISABLE_ROLL_FORWARD)) {
+		err = recover_fsync_data(sbi);
+		if (err)
+			f2fs_msg(sb, KERN_ERR,
+				"Cannot recover all fsync data errno=%ld", err);
+	}
 
+	/*
+	 * If filesystem is not mounted as read-only then
+	 * do start the gc_thread.
+	 */
+	if (!(sb->s_flags & MS_RDONLY)) {
+		/* After POR, we can run background GC thread.*/
+		err = start_gc_thread(sbi);
+		if (err)
+			goto free_kobj;
+	}
 	return 0;
-fail:
+
+free_kobj:
+	kobject_del(&sbi->s_kobj);
+free_proc:
 	if (sbi->s_proc) {
 		remove_proc_entry("segment_info", sbi->s_proc);
 		remove_proc_entry(sb->s_id, f2fs_proc_root);
 	}
 	f2fs_destroy_stats(sbi);
-free_gc:
-	stop_gc_thread(sbi);
 free_root_inode:
 	dput(sb->s_root);
 	sb->s_root = NULL;
@@ -1084,7 +1115,7 @@ MODULE_ALIAS_FS("f2fs");
 static int __init init_inodecache(void)
 {
 	f2fs_inode_cachep = f2fs_kmem_cache_create("f2fs_inode_cache",
-			sizeof(struct f2fs_inode_info), NULL);
+			sizeof(struct f2fs_inode_info));
 	if (!f2fs_inode_cachep)
 		return -ENOMEM;
 	return 0;

fs/f2fs/xattr.c

@@ -275,7 +275,7 @@ static void *read_all_xattrs(struct inode *inode, struct page *ipage)
 
 	inline_size = inline_xattr_size(inode);
 
-	txattr_addr = kzalloc(inline_size + size, GFP_KERNEL);
+	txattr_addr = kzalloc(inline_size + size, GFP_F2FS_ZERO);
 	if (!txattr_addr)
 		return NULL;
 
@@ -407,6 +407,8 @@ int f2fs_getxattr(struct inode *inode, int name_index, const char *name,
 	if (name == NULL)
 		return -EINVAL;
 	name_len = strlen(name);
+	if (name_len > F2FS_NAME_LEN)
+		return -ERANGE;
 
 	base_addr = read_all_xattrs(inode, NULL);
 	if (!base_addr)
@@ -590,7 +592,10 @@ int f2fs_setxattr(struct inode *inode, int name_index, const char *name,
 	f2fs_balance_fs(sbi);
 
 	f2fs_lock_op(sbi);
+	/* protect xattr_ver */
+	down_write(&F2FS_I(inode)->i_sem);
 	err = __f2fs_setxattr(inode, name_index, name, value, value_len, ipage);
+	up_write(&F2FS_I(inode)->i_sem);
 	f2fs_unlock_op(sbi);
 
 	return err;

include/linux/f2fs_fs.h

@@ -183,7 +183,7 @@ struct f2fs_inode {
 	__le32 i_pino;			/* parent inode number */
 	__le32 i_namelen;		/* file name length */
 	__u8 i_name[F2FS_NAME_LEN];	/* file name for SPOR */
-	__u8 i_reserved2;		/* for backward compatibility */
+	__u8 i_dir_level;		/* dentry_level for large dir */
 
 	struct f2fs_extent i_ext;	/* caching a largest extent */