Merge tag 'for-f2fs-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs

Pull f2fs updates from Jaegeuk Kim:
 "The major work includes fixing and enhancing the existing extent_cache
  feature, which has been well settling down so far and now it becomes a
  default mount option accordingly.

  Also, this version newly registers a f2fs memory shrinker to reclaim
  several objects consumed by a couple of data structures in order to
  avoid memory pressures.

  Another new feature is to add ioctl(F2FS_GARBAGE_COLLECT) which
  triggers a cleaning job explicitly by users.

  Most of the other patches are to fix bugs occurred in the corner cases
  across the whole code area"

* tag 'for-f2fs-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs: (85 commits)
  f2fs: upset segment_info repair
  f2fs: avoid accessing NULL pointer in f2fs_drop_largest_extent
  f2fs: update extent tree in batches
  f2fs: fix to release inode correctly
  f2fs: handle f2fs_truncate error correctly
  f2fs: avoid unneeded initializing when converting inline dentry
  f2fs: atomically set inode->i_flags
  f2fs: fix wrong pointer access during try_to_free_nids
  f2fs: use __GFP_NOFAIL to avoid infinite loop
  f2fs: lookup neighbor extent nodes for merging later
  f2fs: split __insert_extent_tree_ret for readability
  f2fs: kill dead code in __insert_extent_tree
  f2fs: adjust showing of extent cache stat
  f2fs: add largest/cached stat in extent cache
  f2fs: fix incorrect mapping for bmap
  f2fs: add annotation for space utilization of regular/inline dentry
  f2fs: fix to update cached_en of extent tree properly
  f2fs: fix typo
  f2fs: check the node block address of newly allocated nid
  f2fs: go out for insert_inode_locked failure
  ...

Documentation/filesystems/f2fs.txt

@@ -143,7 +143,9 @@ fastboot               This option is used when a system wants to reduce mount
 extent_cache           Enable an extent cache based on rb-tree, it can cache
                        as many as extent which map between contiguous logical
                        address and physical address per inode, resulting in
-                       increasing the cache hit ratio.
+                       increasing the cache hit ratio. Set by default.
+noextent_cache         Diable an extent cache based on rb-tree explicitly, see
+                       the above extent_cache mount option.
 noinline_data          Disable the inline data feature, inline data feature is
                        enabled by default.
 

MAINTAINERS

@@ -4416,6 +4416,7 @@ F:	include/linux/fscache*.h
 F2FS FILE SYSTEM
 M:	Jaegeuk Kim <jaegeuk@kernel.org>
 M:	Changman Lee <cm224.lee@samsung.com>
+R:	Chao Yu <chao2.yu@samsung.com>
 L:	linux-f2fs-devel@lists.sourceforge.net
 W:	http://en.wikipedia.org/wiki/F2FS
 T:	git git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git
@@ -4424,6 +4425,7 @@ F:	Documentation/filesystems/f2fs.txt
 F:	Documentation/ABI/testing/sysfs-fs-f2fs
 F:	fs/f2fs/
 F:	include/linux/f2fs_fs.h
+F:	include/trace/events/f2fs.h
 
 FUJITSU FR-V (FRV) PORT
 M:	David Howells <dhowells@redhat.com>

fs/f2fs/Kconfig

@@ -45,7 +45,7 @@ config F2FS_FS_POSIX_ACL
 	default y
 	help
 	  Posix Access Control Lists (ACLs) support permissions for users and
-	  gourps beyond the owner/group/world scheme.
+	  groups beyond the owner/group/world scheme.
 
 	  To learn more about Access Control Lists, visit the POSIX ACLs for
 	  Linux website <http://acl.bestbits.at/>.

fs/f2fs/Makefile

@@ -2,6 +2,7 @@ obj-$(CONFIG_F2FS_FS) += f2fs.o
 
 f2fs-y		:= dir.o file.o inode.o namei.o hash.o super.o inline.o
 f2fs-y		+= checkpoint.o gc.o data.o node.o segment.o recovery.o
+f2fs-y		+= shrinker.o extent_cache.o
 f2fs-$(CONFIG_F2FS_STAT_FS) += debug.o
 f2fs-$(CONFIG_F2FS_FS_XATTR) += xattr.o
 f2fs-$(CONFIG_F2FS_FS_POSIX_ACL) += acl.o

fs/f2fs/checkpoint.c

@@ -69,14 +69,24 @@ struct page *get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index)
 
 	fio.page = page;
 
-	if (f2fs_submit_page_bio(&fio))
+	if (f2fs_submit_page_bio(&fio)) {
+		f2fs_put_page(page, 1);
 		goto repeat;
+	}
 
 	lock_page(page);
 	if (unlikely(page->mapping != mapping)) {
 		f2fs_put_page(page, 1);
 		goto repeat;
 	}
+
+	/*
+	 * if there is any IO error when accessing device, make our filesystem
+	 * readonly and make sure do not write checkpoint with non-uptodate
+	 * meta page.
+	 */
+	if (unlikely(!PageUptodate(page)))
+		f2fs_stop_checkpoint(sbi);
 out:
 	return page;
 }
@@ -326,26 +336,18 @@ const struct address_space_operations f2fs_meta_aops = {
 static void __add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
 {
 	struct inode_management *im = &sbi->im[type];
-	struct ino_entry *e;
+	struct ino_entry *e, *tmp;
+
+	tmp = f2fs_kmem_cache_alloc(ino_entry_slab, GFP_NOFS);
 retry:
-	if (radix_tree_preload(GFP_NOFS)) {
+	radix_tree_preload(GFP_NOFS | __GFP_NOFAIL);
-		cond_resched();
-		goto retry;
-	}
 
 	spin_lock(&im->ino_lock);
-
 	e = radix_tree_lookup(&im->ino_root, ino);
 	if (!e) {
-		e = kmem_cache_alloc(ino_entry_slab, GFP_ATOMIC);
+		e = tmp;
-		if (!e) {
-			spin_unlock(&im->ino_lock);
-			radix_tree_preload_end();
-			goto retry;
-		}
 		if (radix_tree_insert(&im->ino_root, ino, e)) {
 			spin_unlock(&im->ino_lock);
-			kmem_cache_free(ino_entry_slab, e);
 			radix_tree_preload_end();
 			goto retry;
 		}
@@ -358,6 +360,9 @@ static void __add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
 	}
 	spin_unlock(&im->ino_lock);
 	radix_tree_preload_end();
+
+	if (e != tmp)
+		kmem_cache_free(ino_entry_slab, tmp);
 }
 
 static void __remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
@@ -458,24 +463,34 @@ void remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino)
 	__remove_ino_entry(sbi, ino, ORPHAN_INO);
 }
 
-static void recover_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino)
+static int recover_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino)
 {
-	struct inode *inode = f2fs_iget(sbi->sb, ino);
+	struct inode *inode;
-	f2fs_bug_on(sbi, IS_ERR(inode));
+
+	inode = f2fs_iget(sbi->sb, ino);
+	if (IS_ERR(inode)) {
+		/*
+		 * there should be a bug that we can't find the entry
+		 * to orphan inode.
+		 */
+		f2fs_bug_on(sbi, PTR_ERR(inode) == -ENOENT);
+		return PTR_ERR(inode);
+	}
+
 	clear_nlink(inode);
 
 	/* truncate all the data during iput */
 	iput(inode);
+	return 0;
 }
 
-void recover_orphan_inodes(struct f2fs_sb_info *sbi)
+int recover_orphan_inodes(struct f2fs_sb_info *sbi)
 {
 	block_t start_blk, orphan_blocks, i, j;
+	int err;
 
 	if (!is_set_ckpt_flags(F2FS_CKPT(sbi), CP_ORPHAN_PRESENT_FLAG))
-		return;
+		return 0;
-
-	set_sbi_flag(sbi, SBI_POR_DOING);
 
 	start_blk = __start_cp_addr(sbi) + 1 + __cp_payload(sbi);
 	orphan_blocks = __start_sum_addr(sbi) - 1 - __cp_payload(sbi);
@@ -489,14 +504,17 @@ void recover_orphan_inodes(struct f2fs_sb_info *sbi)
 		orphan_blk = (struct f2fs_orphan_block *)page_address(page);
 		for (j = 0; j < le32_to_cpu(orphan_blk->entry_count); j++) {
 			nid_t ino = le32_to_cpu(orphan_blk->ino[j]);
-			recover_orphan_inode(sbi, ino);
+			err = recover_orphan_inode(sbi, ino);
+			if (err) {
+				f2fs_put_page(page, 1);
+				return err;
+			}
 		}
 		f2fs_put_page(page, 1);
 	}
 	/* clear Orphan Flag */
 	clear_ckpt_flags(F2FS_CKPT(sbi), CP_ORPHAN_PRESENT_FLAG);
-	clear_sbi_flag(sbi, SBI_POR_DOING);
+	return 0;
-	return;
 }
 
 static void write_orphan_inodes(struct f2fs_sb_info *sbi, block_t start_blk)
@@ -504,7 +522,7 @@ static void write_orphan_inodes(struct f2fs_sb_info *sbi, block_t start_blk)
 	struct list_head *head;
 	struct f2fs_orphan_block *orphan_blk = NULL;
 	unsigned int nentries = 0;
-	unsigned short index;
+	unsigned short index = 1;
 	unsigned short orphan_blocks;
 	struct page *page = NULL;
 	struct ino_entry *orphan = NULL;
@@ -512,11 +530,6 @@ static void write_orphan_inodes(struct f2fs_sb_info *sbi, block_t start_blk)
 
 	orphan_blocks = GET_ORPHAN_BLOCKS(im->ino_num);
 
-	for (index = 0; index < orphan_blocks; index++)
-		grab_meta_page(sbi, start_blk + index);
-
-	index = 1;
-
 	/*
 	 * we don't need to do spin_lock(&im->ino_lock) here, since all the
 	 * orphan inode operations are covered under f2fs_lock_op().
@@ -527,12 +540,10 @@ static void write_orphan_inodes(struct f2fs_sb_info *sbi, block_t start_blk)
 	/* loop for each orphan inode entry and write them in Jornal block */
 	list_for_each_entry(orphan, head, list) {
 		if (!page) {
-			page = find_get_page(META_MAPPING(sbi), start_blk++);
+			page = grab_meta_page(sbi, start_blk++);
-			f2fs_bug_on(sbi, !page);
 			orphan_blk =
 				(struct f2fs_orphan_block *)page_address(page);
 			memset(orphan_blk, 0, sizeof(*orphan_blk));
-			f2fs_put_page(page, 0);
 		}
 
 		orphan_blk->ino[nentries++] = cpu_to_le32(orphan->ino);
@@ -704,7 +715,8 @@ void update_dirty_page(struct inode *inode, struct page *page)
 	struct inode_entry *new;
 	int ret = 0;
 
-	if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode))
+	if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
+			!S_ISLNK(inode->i_mode))
 		return;
 
 	if (!S_ISDIR(inode->i_mode)) {
@@ -892,12 +904,15 @@ static void do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
 	__u32 crc32 = 0;
 	int i;
 	int cp_payload_blks = __cp_payload(sbi);
+	block_t discard_blk = NEXT_FREE_BLKADDR(sbi, curseg);
+	bool invalidate = false;
 
 	/*
 	 * This avoids to conduct wrong roll-forward operations and uses
 	 * metapages, so should be called prior to sync_meta_pages below.
 	 */
-	discard_next_dnode(sbi, NEXT_FREE_BLKADDR(sbi, curseg));
+	if (discard_next_dnode(sbi, discard_blk))
+		invalidate = true;
 
 	/* Flush all the NAT/SIT pages */
 	while (get_pages(sbi, F2FS_DIRTY_META)) {
@@ -1026,6 +1041,14 @@ static void do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
 	/* wait for previous submitted meta pages writeback */
 	wait_on_all_pages_writeback(sbi);
 
+	/*
+	 * invalidate meta page which is used temporarily for zeroing out
+	 * block at the end of warm node chain.
+	 */
+	if (invalidate)
+		invalidate_mapping_pages(META_MAPPING(sbi), discard_blk,
+								discard_blk);
+
 	release_dirty_inode(sbi);
 
 	if (unlikely(f2fs_cp_error(sbi)))

fs/f2fs/crypto_key.c

@@ -92,8 +92,7 @@ static void f2fs_free_crypt_info(struct f2fs_crypt_info *ci)
 	if (!ci)
 		return;
 
-	if (ci->ci_keyring_key)
+	key_put(ci->ci_keyring_key);
-		key_put(ci->ci_keyring_key);
 	crypto_free_ablkcipher(ci->ci_ctfm);
 	kmem_cache_free(f2fs_crypt_info_cachep, ci);
 }

fs/f2fs/debug.c

@@ -33,8 +33,11 @@ static void update_general_status(struct f2fs_sb_info *sbi)
 	int i;
 
 	/* validation check of the segment numbers */
-	si->hit_ext = sbi->read_hit_ext;
+	si->hit_largest = atomic_read(&sbi->read_hit_largest);
-	si->total_ext = sbi->total_hit_ext;
+	si->hit_cached = atomic_read(&sbi->read_hit_cached);
+	si->hit_rbtree = atomic_read(&sbi->read_hit_rbtree);
+	si->hit_total = si->hit_largest + si->hit_cached + si->hit_rbtree;
+	si->total_ext = atomic_read(&sbi->total_hit_ext);
 	si->ext_tree = sbi->total_ext_tree;
 	si->ext_node = atomic_read(&sbi->total_ext_node);
 	si->ndirty_node = get_pages(sbi, F2FS_DIRTY_NODES);
@@ -49,6 +52,7 @@ static void update_general_status(struct f2fs_sb_info *sbi)
 	si->valid_count = valid_user_blocks(sbi);
 	si->valid_node_count = valid_node_count(sbi);
 	si->valid_inode_count = valid_inode_count(sbi);
+	si->inline_xattr = atomic_read(&sbi->inline_xattr);
 	si->inline_inode = atomic_read(&sbi->inline_inode);
 	si->inline_dir = atomic_read(&sbi->inline_dir);
 	si->utilization = utilization(sbi);
@@ -226,6 +230,8 @@ static int stat_show(struct seq_file *s, void *v)
 		seq_printf(s, "Other: %u)\n  - Data: %u\n",
 			   si->valid_node_count - si->valid_inode_count,
 			   si->valid_count - si->valid_node_count);
+		seq_printf(s, "  - Inline_xattr Inode: %u\n",
+			   si->inline_xattr);
 		seq_printf(s, "  - Inline_data Inode: %u\n",
 			   si->inline_inode);
 		seq_printf(s, "  - Inline_dentry Inode: %u\n",
@@ -276,10 +282,16 @@ static int stat_show(struct seq_file *s, void *v)
 				si->bg_data_blks);
 		seq_printf(s, "  - node blocks : %d (%d)\n", si->node_blks,
 				si->bg_node_blks);
-		seq_printf(s, "\nExtent Hit Ratio: %d / %d\n",
+		seq_puts(s, "\nExtent Cache:\n");
-			   si->hit_ext, si->total_ext);
+		seq_printf(s, "  - Hit Count: L1-1:%d L1-2:%d L2:%d\n",
-		seq_printf(s, "\nExtent Tree Count: %d\n", si->ext_tree);
+				si->hit_largest, si->hit_cached,
-		seq_printf(s, "\nExtent Node Count: %d\n", si->ext_node);
+				si->hit_rbtree);
+		seq_printf(s, "  - Hit Ratio: %d%% (%d / %d)\n",
+				!si->total_ext ? 0 :
+				(si->hit_total * 100) / si->total_ext,
+				si->hit_total, si->total_ext);
+		seq_printf(s, "  - Inner Struct Count: tree: %d, node: %d\n",
+				si->ext_tree, si->ext_node);
 		seq_puts(s, "\nBalancing F2FS Async:\n");
 		seq_printf(s, "  - inmem: %4d, wb: %4d\n",
 			   si->inmem_pages, si->wb_pages);
@@ -366,6 +378,12 @@ int f2fs_build_stats(struct f2fs_sb_info *sbi)
 	si->sbi = sbi;
 	sbi->stat_info = si;
 
+	atomic_set(&sbi->total_hit_ext, 0);
+	atomic_set(&sbi->read_hit_rbtree, 0);
+	atomic_set(&sbi->read_hit_largest, 0);
+	atomic_set(&sbi->read_hit_cached, 0);
+
+	atomic_set(&sbi->inline_xattr, 0);
 	atomic_set(&sbi->inline_inode, 0);
 	atomic_set(&sbi->inline_dir, 0);
 	atomic_set(&sbi->inplace_count, 0);

fs/f2fs/dir.c

@@ -718,8 +718,8 @@ void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
 	if (inode)
 		f2fs_drop_nlink(dir, inode, NULL);
 
-	if (bit_pos == NR_DENTRY_IN_BLOCK) {
+	if (bit_pos == NR_DENTRY_IN_BLOCK &&
-		truncate_hole(dir, page->index, page->index + 1);
+			!truncate_hole(dir, page->index, page->index + 1)) {
 		clear_page_dirty_for_io(page);
 		ClearPagePrivate(page);
 		ClearPageUptodate(page);

fs/f2fs/extent_cache.c

@@ -0,0 +1,791 @@
+/*
+ * f2fs extent cache support
+ *
+ * Copyright (c) 2015 Motorola Mobility
+ * Copyright (c) 2015 Samsung Electronics
+ * Authors: Jaegeuk Kim <jaegeuk@kernel.org>
+ *          Chao Yu <chao2.yu@samsung.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/fs.h>
+#include <linux/f2fs_fs.h>
+
+#include "f2fs.h"
+#include "node.h"
+#include <trace/events/f2fs.h>
+
+static struct kmem_cache *extent_tree_slab;
+static struct kmem_cache *extent_node_slab;
+
+static struct extent_node *__attach_extent_node(struct f2fs_sb_info *sbi,
+				struct extent_tree *et, struct extent_info *ei,
+				struct rb_node *parent, struct rb_node **p)
+{
+	struct extent_node *en;
+
+	en = kmem_cache_alloc(extent_node_slab, GFP_ATOMIC);
+	if (!en)
+		return NULL;
+
+	en->ei = *ei;
+	INIT_LIST_HEAD(&en->list);
+
+	rb_link_node(&en->rb_node, parent, p);
+	rb_insert_color(&en->rb_node, &et->root);
+	et->count++;
+	atomic_inc(&sbi->total_ext_node);
+	return en;
+}
+
+static void __detach_extent_node(struct f2fs_sb_info *sbi,
+				struct extent_tree *et, struct extent_node *en)
+{
+	rb_erase(&en->rb_node, &et->root);
+	et->count--;
+	atomic_dec(&sbi->total_ext_node);
+
+	if (et->cached_en == en)
+		et->cached_en = NULL;
+}
+
+static struct extent_tree *__grab_extent_tree(struct inode *inode)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	struct extent_tree *et;
+	nid_t ino = inode->i_ino;
+
+	down_write(&sbi->extent_tree_lock);
+	et = radix_tree_lookup(&sbi->extent_tree_root, ino);
+	if (!et) {
+		et = f2fs_kmem_cache_alloc(extent_tree_slab, GFP_NOFS);
+		f2fs_radix_tree_insert(&sbi->extent_tree_root, ino, et);
+		memset(et, 0, sizeof(struct extent_tree));
+		et->ino = ino;
+		et->root = RB_ROOT;
+		et->cached_en = NULL;
+		rwlock_init(&et->lock);
+		atomic_set(&et->refcount, 0);
+		et->count = 0;
+		sbi->total_ext_tree++;
+	}
+	atomic_inc(&et->refcount);
+	up_write(&sbi->extent_tree_lock);
+
+	/* never died until evict_inode */
+	F2FS_I(inode)->extent_tree = et;
+
+	return et;
+}
+
+static struct extent_node *__lookup_extent_tree(struct f2fs_sb_info *sbi,
+				struct extent_tree *et, unsigned int fofs)
+{
+	struct rb_node *node = et->root.rb_node;
+	struct extent_node *en = et->cached_en;
+
+	if (en) {
+		struct extent_info *cei = &en->ei;
+
+		if (cei->fofs <= fofs && cei->fofs + cei->len > fofs) {
+			stat_inc_cached_node_hit(sbi);
+			return en;
+		}
+	}
+
+	while (node) {
+		en = rb_entry(node, struct extent_node, rb_node);
+
+		if (fofs < en->ei.fofs) {
+			node = node->rb_left;
+		} else if (fofs >= en->ei.fofs + en->ei.len) {
+			node = node->rb_right;
+		} else {
+			stat_inc_rbtree_node_hit(sbi);
+			return en;
+		}
+	}
+	return NULL;
+}
+
+static struct extent_node *__init_extent_tree(struct f2fs_sb_info *sbi,
+				struct extent_tree *et, struct extent_info *ei)
+{
+	struct rb_node **p = &et->root.rb_node;
+	struct extent_node *en;
+
+	en = __attach_extent_node(sbi, et, ei, NULL, p);
+	if (!en)
+		return NULL;
+
+	et->largest = en->ei;
+	et->cached_en = en;
+	return en;
+}
+
+static unsigned int __free_extent_tree(struct f2fs_sb_info *sbi,
+					struct extent_tree *et, bool free_all)
+{
+	struct rb_node *node, *next;
+	struct extent_node *en;
+	unsigned int count = et->count;
+
+	node = rb_first(&et->root);
+	while (node) {
+		next = rb_next(node);
+		en = rb_entry(node, struct extent_node, rb_node);
+
+		if (free_all) {
+			spin_lock(&sbi->extent_lock);
+			if (!list_empty(&en->list))
+				list_del_init(&en->list);
+			spin_unlock(&sbi->extent_lock);
+		}
+
+		if (free_all || list_empty(&en->list)) {
+			__detach_extent_node(sbi, et, en);
+			kmem_cache_free(extent_node_slab, en);
+		}
+		node = next;
+	}
+
+	return count - et->count;
+}
+
+static void __drop_largest_extent(struct inode *inode, pgoff_t fofs)
+{
+	struct extent_info *largest = &F2FS_I(inode)->extent_tree->largest;
+
+	if (largest->fofs <= fofs && largest->fofs + largest->len > fofs)
+		largest->len = 0;
+}
+
+void f2fs_drop_largest_extent(struct inode *inode, pgoff_t fofs)
+{
+	if (!f2fs_may_extent_tree(inode))
+		return;
+
+	__drop_largest_extent(inode, fofs);
+}
+
+void f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	struct extent_tree *et;
+	struct extent_node *en;
+	struct extent_info ei;
+
+	if (!f2fs_may_extent_tree(inode))
+		return;
+
+	et = __grab_extent_tree(inode);
+
+	if (!i_ext || le32_to_cpu(i_ext->len) < F2FS_MIN_EXTENT_LEN)
+		return;
+
+	set_extent_info(&ei, le32_to_cpu(i_ext->fofs),
+		le32_to_cpu(i_ext->blk), le32_to_cpu(i_ext->len));
+
+	write_lock(&et->lock);
+	if (et->count)
+		goto out;
+
+	en = __init_extent_tree(sbi, et, &ei);
+	if (en) {
+		spin_lock(&sbi->extent_lock);
+		list_add_tail(&en->list, &sbi->extent_list);
+		spin_unlock(&sbi->extent_lock);
+	}
+out:
+	write_unlock(&et->lock);
+}
+
+static bool f2fs_lookup_extent_tree(struct inode *inode, pgoff_t pgofs,
+							struct extent_info *ei)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	struct extent_tree *et = F2FS_I(inode)->extent_tree;
+	struct extent_node *en;
+	bool ret = false;
+
+	f2fs_bug_on(sbi, !et);
+
+	trace_f2fs_lookup_extent_tree_start(inode, pgofs);
+
+	read_lock(&et->lock);
+
+	if (et->largest.fofs <= pgofs &&
+			et->largest.fofs + et->largest.len > pgofs) {
+		*ei = et->largest;
+		ret = true;
+		stat_inc_largest_node_hit(sbi);
+		goto out;
+	}
+
+	en = __lookup_extent_tree(sbi, et, pgofs);
+	if (en) {
+		*ei = en->ei;
+		spin_lock(&sbi->extent_lock);
+		if (!list_empty(&en->list))
+			list_move_tail(&en->list, &sbi->extent_list);
+		et->cached_en = en;
+		spin_unlock(&sbi->extent_lock);
+		ret = true;
+	}
+out:
+	stat_inc_total_hit(sbi);
+	read_unlock(&et->lock);
+
+	trace_f2fs_lookup_extent_tree_end(inode, pgofs, ei);
+	return ret;
+}
+
+
+/*
+ * lookup extent at @fofs, if hit, return the extent
+ * if not, return NULL and
+ * @prev_ex: extent before fofs
+ * @next_ex: extent after fofs
+ * @insert_p: insert point for new extent at fofs
+ * in order to simpfy the insertion after.
+ * tree must stay unchanged between lookup and insertion.
+ */
+static struct extent_node *__lookup_extent_tree_ret(struct extent_tree *et,
+				unsigned int fofs,
+				struct extent_node **prev_ex,
+				struct extent_node **next_ex,
+				struct rb_node ***insert_p,
+				struct rb_node **insert_parent)
+{
+	struct rb_node **pnode = &et->root.rb_node;
+	struct rb_node *parent = NULL, *tmp_node;
+	struct extent_node *en = et->cached_en;
+
+	*insert_p = NULL;
+	*insert_parent = NULL;
+	*prev_ex = NULL;
+	*next_ex = NULL;
+
+	if (RB_EMPTY_ROOT(&et->root))
+		return NULL;
+
+	if (en) {
+		struct extent_info *cei = &en->ei;
+
+		if (cei->fofs <= fofs && cei->fofs + cei->len > fofs)
+			goto lookup_neighbors;
+	}
+
+	while (*pnode) {
+		parent = *pnode;
+		en = rb_entry(*pnode, struct extent_node, rb_node);
+
+		if (fofs < en->ei.fofs)
+			pnode = &(*pnode)->rb_left;
+		else if (fofs >= en->ei.fofs + en->ei.len)
+			pnode = &(*pnode)->rb_right;
+		else
+			goto lookup_neighbors;
+	}
+
+	*insert_p = pnode;
+	*insert_parent = parent;
+
+	en = rb_entry(parent, struct extent_node, rb_node);
+	tmp_node = parent;
+	if (parent && fofs > en->ei.fofs)
+		tmp_node = rb_next(parent);
+	*next_ex = tmp_node ?
+		rb_entry(tmp_node, struct extent_node, rb_node) : NULL;
+
+	tmp_node = parent;
+	if (parent && fofs < en->ei.fofs)
+		tmp_node = rb_prev(parent);
+	*prev_ex = tmp_node ?
+		rb_entry(tmp_node, struct extent_node, rb_node) : NULL;
+	return NULL;
+
+lookup_neighbors:
+	if (fofs == en->ei.fofs) {
+		/* lookup prev node for merging backward later */
+		tmp_node = rb_prev(&en->rb_node);
+		*prev_ex = tmp_node ?
+			rb_entry(tmp_node, struct extent_node, rb_node) : NULL;
+	}
+	if (fofs == en->ei.fofs + en->ei.len - 1) {
+		/* lookup next node for merging frontward later */
+		tmp_node = rb_next(&en->rb_node);
+		*next_ex = tmp_node ?
+			rb_entry(tmp_node, struct extent_node, rb_node) : NULL;
+	}
+	return en;
+}
+
+static struct extent_node *__try_merge_extent_node(struct f2fs_sb_info *sbi,
+				struct extent_tree *et, struct extent_info *ei,
+				struct extent_node **den,
+				struct extent_node *prev_ex,
+				struct extent_node *next_ex)
+{
+	struct extent_node *en = NULL;
+
+	if (prev_ex && __is_back_mergeable(ei, &prev_ex->ei)) {
+		prev_ex->ei.len += ei->len;
+		ei = &prev_ex->ei;
+		en = prev_ex;
+	}
+
+	if (next_ex && __is_front_mergeable(ei, &next_ex->ei)) {
+		if (en) {
+			__detach_extent_node(sbi, et, prev_ex);
+			*den = prev_ex;
+		}
+		next_ex->ei.fofs = ei->fofs;
+		next_ex->ei.blk = ei->blk;
+		next_ex->ei.len += ei->len;
+		en = next_ex;
+	}
+
+	if (en) {
+		if (en->ei.len > et->largest.len)
+			et->largest = en->ei;
+		et->cached_en = en;
+	}
+	return en;
+}
+
+static struct extent_node *__insert_extent_tree(struct f2fs_sb_info *sbi,
+				struct extent_tree *et, struct extent_info *ei,
+				struct rb_node **insert_p,
+				struct rb_node *insert_parent)
+{
+	struct rb_node **p = &et->root.rb_node;
+	struct rb_node *parent = NULL;
+	struct extent_node *en = NULL;
+
+	if (insert_p && insert_parent) {
+		parent = insert_parent;
+		p = insert_p;
+		goto do_insert;
+	}
+
+	while (*p) {
+		parent = *p;
+		en = rb_entry(parent, struct extent_node, rb_node);
+
+		if (ei->fofs < en->ei.fofs)
+			p = &(*p)->rb_left;
+		else if (ei->fofs >= en->ei.fofs + en->ei.len)
+			p = &(*p)->rb_right;
+		else
+			f2fs_bug_on(sbi, 1);
+	}
+do_insert:
+	en = __attach_extent_node(sbi, et, ei, parent, p);
+	if (!en)
+		return NULL;
+
+	if (en->ei.len > et->largest.len)
+		et->largest = en->ei;
+	et->cached_en = en;
+	return en;
+}
+
+unsigned int f2fs_update_extent_tree_range(struct inode *inode,
+				pgoff_t fofs, block_t blkaddr, unsigned int len)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	struct extent_tree *et = F2FS_I(inode)->extent_tree;
+	struct extent_node *en = NULL, *en1 = NULL, *en2 = NULL, *en3 = NULL;
+	struct extent_node *prev_en = NULL, *next_en = NULL;
+	struct extent_info ei, dei, prev;
+	struct rb_node **insert_p = NULL, *insert_parent = NULL;
+	unsigned int end = fofs + len;
+	unsigned int pos = (unsigned int)fofs;
+
+	if (!et)
+		return false;
+
+	write_lock(&et->lock);
+
+	if (is_inode_flag_set(F2FS_I(inode), FI_NO_EXTENT)) {
+		write_unlock(&et->lock);
+		return false;
+	}
+
+	prev = et->largest;
+	dei.len = 0;
+
+	/* we do not guarantee that the largest extent is cached all the time */
+	__drop_largest_extent(inode, fofs);
+
+	/* 1. lookup first extent node in range [fofs, fofs + len - 1] */
+	en = __lookup_extent_tree_ret(et, fofs, &prev_en, &next_en,
+					&insert_p, &insert_parent);
+	if (!en) {
+		if (next_en) {
+			en = next_en;
+			f2fs_bug_on(sbi, en->ei.fofs <= pos);
+			pos = en->ei.fofs;
+		} else {
+			/*
+			 * skip searching in the tree since there is no
+			 * larger extent node in the cache.
+			 */
+			goto update_extent;
+		}
+	}
+
+	/* 2. invlidate all extent nodes in range [fofs, fofs + len - 1] */
+	while (en) {
+		struct rb_node *node;
+
+		if (pos >= end)
+			break;
+
+		dei = en->ei;
+		en1 = en2 = NULL;
+
+		node = rb_next(&en->rb_node);
+
+		/*
+		 * 2.1 there are four cases when we invalidate blkaddr in extent
+		 * node, |V: valid address, X: will be invalidated|
+		 */
+		/* case#1, invalidate right part of extent node |VVVVVXXXXX| */
+		if (pos > dei.fofs && end >= dei.fofs + dei.len) {
+			en->ei.len = pos - dei.fofs;
+
+			if (en->ei.len < F2FS_MIN_EXTENT_LEN) {
+				__detach_extent_node(sbi, et, en);
+				insert_p = NULL;
+				insert_parent = NULL;
+				goto update;
+			}
+
+			if (__is_extent_same(&dei, &et->largest))
+				et->largest = en->ei;
+			goto next;
+		}
+
+		/* case#2, invalidate left part of extent node |XXXXXVVVVV| */
+		if (pos <= dei.fofs && end < dei.fofs + dei.len) {
+			en->ei.fofs = end;
+			en->ei.blk += end - dei.fofs;
+			en->ei.len -= end - dei.fofs;
+
+			if (en->ei.len < F2FS_MIN_EXTENT_LEN) {
+				__detach_extent_node(sbi, et, en);
+				insert_p = NULL;
+				insert_parent = NULL;
+				goto update;
+			}
+
+			if (__is_extent_same(&dei, &et->largest))
+				et->largest = en->ei;
+			goto next;
+		}
+
+		__detach_extent_node(sbi, et, en);
+
+		/*
+		 * if we remove node in rb-tree, our parent node pointer may
+		 * point the wrong place, discard them.
+		 */
+		insert_p = NULL;
+		insert_parent = NULL;
+
+		/* case#3, invalidate entire extent node |XXXXXXXXXX| */
+		if (pos <= dei.fofs && end >= dei.fofs + dei.len) {
+			if (__is_extent_same(&dei, &et->largest))
+				et->largest.len = 0;
+			goto update;
+		}
+
+		/*
+		 * case#4, invalidate data in the middle of extent node
+		 * |VVVXXXXVVV|
+		 */
+		if (dei.len > F2FS_MIN_EXTENT_LEN) {
+			unsigned int endofs;
+
+			/*  insert left part of split extent into cache */
+			if (pos - dei.fofs >= F2FS_MIN_EXTENT_LEN) {
+				set_extent_info(&ei, dei.fofs, dei.blk,
+							pos - dei.fofs);
+				en1 = __insert_extent_tree(sbi, et, &ei,
+								NULL, NULL);
+			}
+
+			/* insert right part of split extent into cache */
+			endofs = dei.fofs + dei.len;
+			if (endofs - end >= F2FS_MIN_EXTENT_LEN) {
+				set_extent_info(&ei, end,
+						end - dei.fofs + dei.blk,
+						endofs - end);
+				en2 = __insert_extent_tree(sbi, et, &ei,
+								NULL, NULL);
+			}
+		}
+update:
+		/* 2.2 update in global extent list */
+		spin_lock(&sbi->extent_lock);
+		if (en && !list_empty(&en->list))
+			list_del(&en->list);
+		if (en1)
+			list_add_tail(&en1->list, &sbi->extent_list);
+		if (en2)
+			list_add_tail(&en2->list, &sbi->extent_list);
+		spin_unlock(&sbi->extent_lock);
+
+		/* 2.3 release extent node */
+		if (en)
+			kmem_cache_free(extent_node_slab, en);
+next:
+		en = node ? rb_entry(node, struct extent_node, rb_node) : NULL;
+		next_en = en;
+		if (en)
+			pos = en->ei.fofs;
+	}
+
+update_extent:
+	/* 3. update extent in extent cache */
+	if (blkaddr) {
+		struct extent_node *den = NULL;
+
+		set_extent_info(&ei, fofs, blkaddr, len);
+		en3 = __try_merge_extent_node(sbi, et, &ei, &den,
+							prev_en, next_en);
+		if (!en3)
+			en3 = __insert_extent_tree(sbi, et, &ei,
+						insert_p, insert_parent);
+
+		/* give up extent_cache, if split and small updates happen */
+		if (dei.len >= 1 &&
+				prev.len < F2FS_MIN_EXTENT_LEN &&
+				et->largest.len < F2FS_MIN_EXTENT_LEN) {
+			et->largest.len = 0;
+			set_inode_flag(F2FS_I(inode), FI_NO_EXTENT);
+		}
+
+		spin_lock(&sbi->extent_lock);
+		if (en3) {
+			if (list_empty(&en3->list))
+				list_add_tail(&en3->list, &sbi->extent_list);
+			else
+				list_move_tail(&en3->list, &sbi->extent_list);
+		}
+		if (den && !list_empty(&den->list))
+			list_del(&den->list);
+		spin_unlock(&sbi->extent_lock);
+
+		if (den)
+			kmem_cache_free(extent_node_slab, den);
+	}
+
+	if (is_inode_flag_set(F2FS_I(inode), FI_NO_EXTENT))
+		__free_extent_tree(sbi, et, true);
+
+	write_unlock(&et->lock);
+
+	return !__is_extent_same(&prev, &et->largest);
+}
+
+unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink)
+{
+	struct extent_tree *treevec[EXT_TREE_VEC_SIZE];
+	struct extent_node *en, *tmp;
+	unsigned long ino = F2FS_ROOT_INO(sbi);
+	struct radix_tree_root *root = &sbi->extent_tree_root;
+	unsigned int found;
+	unsigned int node_cnt = 0, tree_cnt = 0;
+	int remained;
+
+	if (!test_opt(sbi, EXTENT_CACHE))
+		return 0;
+
+	if (!down_write_trylock(&sbi->extent_tree_lock))
+		goto out;
+
+	/* 1. remove unreferenced extent tree */
+	while ((found = radix_tree_gang_lookup(root,
+				(void **)treevec, ino, EXT_TREE_VEC_SIZE))) {
+		unsigned i;
+
+		ino = treevec[found - 1]->ino + 1;
+		for (i = 0; i < found; i++) {
+			struct extent_tree *et = treevec[i];
+
+			if (!atomic_read(&et->refcount)) {
+				write_lock(&et->lock);
+				node_cnt += __free_extent_tree(sbi, et, true);
+				write_unlock(&et->lock);
+
+				radix_tree_delete(root, et->ino);
+				kmem_cache_free(extent_tree_slab, et);
+				sbi->total_ext_tree--;
+				tree_cnt++;
+
+				if (node_cnt + tree_cnt >= nr_shrink)
+					goto unlock_out;
+			}
+		}
+	}
+	up_write(&sbi->extent_tree_lock);
+
+	/* 2. remove LRU extent entries */
+	if (!down_write_trylock(&sbi->extent_tree_lock))
+		goto out;
+
+	remained = nr_shrink - (node_cnt + tree_cnt);
+
+	spin_lock(&sbi->extent_lock);
+	list_for_each_entry_safe(en, tmp, &sbi->extent_list, list) {
+		if (!remained--)
+			break;
+		list_del_init(&en->list);
+	}
+	spin_unlock(&sbi->extent_lock);
+
+	while ((found = radix_tree_gang_lookup(root,
+				(void **)treevec, ino, EXT_TREE_VEC_SIZE))) {
+		unsigned i;
+
+		ino = treevec[found - 1]->ino + 1;
+		for (i = 0; i < found; i++) {
+			struct extent_tree *et = treevec[i];
+
+			write_lock(&et->lock);
+			node_cnt += __free_extent_tree(sbi, et, false);
+			write_unlock(&et->lock);
+
+			if (node_cnt + tree_cnt >= nr_shrink)
+				break;
+		}
+	}
+unlock_out:
+	up_write(&sbi->extent_tree_lock);
+out:
+	trace_f2fs_shrink_extent_tree(sbi, node_cnt, tree_cnt);
+
+	return node_cnt + tree_cnt;
+}
+
+unsigned int f2fs_destroy_extent_node(struct inode *inode)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	struct extent_tree *et = F2FS_I(inode)->extent_tree;
+	unsigned int node_cnt = 0;
+
+	if (!et)
+		return 0;
+
+	write_lock(&et->lock);
+	node_cnt = __free_extent_tree(sbi, et, true);
+	write_unlock(&et->lock);
+
+	return node_cnt;
+}
+
+void f2fs_destroy_extent_tree(struct inode *inode)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	struct extent_tree *et = F2FS_I(inode)->extent_tree;
+	unsigned int node_cnt = 0;
+
+	if (!et)
+		return;
+
+	if (inode->i_nlink && !is_bad_inode(inode) && et->count) {
+		atomic_dec(&et->refcount);
+		return;
+	}
+
+	/* free all extent info belong to this extent tree */
+	node_cnt = f2fs_destroy_extent_node(inode);
+
+	/* delete extent tree entry in radix tree */
+	down_write(&sbi->extent_tree_lock);
+	atomic_dec(&et->refcount);
+	f2fs_bug_on(sbi, atomic_read(&et->refcount) || et->count);
+	radix_tree_delete(&sbi->extent_tree_root, inode->i_ino);
+	kmem_cache_free(extent_tree_slab, et);
+	sbi->total_ext_tree--;
+	up_write(&sbi->extent_tree_lock);
+
+	F2FS_I(inode)->extent_tree = NULL;
+
+	trace_f2fs_destroy_extent_tree(inode, node_cnt);
+}
+
+bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
+					struct extent_info *ei)
+{
+	if (!f2fs_may_extent_tree(inode))
+		return false;
+
+	return f2fs_lookup_extent_tree(inode, pgofs, ei);
+}
+
+void f2fs_update_extent_cache(struct dnode_of_data *dn)
+{
+	struct f2fs_inode_info *fi = F2FS_I(dn->inode);
+	pgoff_t fofs;
+
+	if (!f2fs_may_extent_tree(dn->inode))
+		return;
+
+	f2fs_bug_on(F2FS_I_SB(dn->inode), dn->data_blkaddr == NEW_ADDR);
+
+
+	fofs = start_bidx_of_node(ofs_of_node(dn->node_page), fi) +
+							dn->ofs_in_node;
+
+	if (f2fs_update_extent_tree_range(dn->inode, fofs, dn->data_blkaddr, 1))
+		sync_inode_page(dn);
+}
+
+void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
+				pgoff_t fofs, block_t blkaddr, unsigned int len)
+
+{
+	if (!f2fs_may_extent_tree(dn->inode))
+		return;
+
+	if (f2fs_update_extent_tree_range(dn->inode, fofs, blkaddr, len))
+		sync_inode_page(dn);
+}
+
+void init_extent_cache_info(struct f2fs_sb_info *sbi)
+{
+	INIT_RADIX_TREE(&sbi->extent_tree_root, GFP_NOIO);
+	init_rwsem(&sbi->extent_tree_lock);
+	INIT_LIST_HEAD(&sbi->extent_list);
+	spin_lock_init(&sbi->extent_lock);
+	sbi->total_ext_tree = 0;
+	atomic_set(&sbi->total_ext_node, 0);
+}
+
+int __init create_extent_cache(void)
+{
+	extent_tree_slab = f2fs_kmem_cache_create("f2fs_extent_tree",
+			sizeof(struct extent_tree));
+	if (!extent_tree_slab)
+		return -ENOMEM;
+	extent_node_slab = f2fs_kmem_cache_create("f2fs_extent_node",
+			sizeof(struct extent_node));
+	if (!extent_node_slab) {
+		kmem_cache_destroy(extent_tree_slab);
+		return -ENOMEM;
+	}
+	return 0;
+}
+
+void destroy_extent_cache(void)
+{
+	kmem_cache_destroy(extent_node_slab);
+	kmem_cache_destroy(extent_tree_slab);
+}

fs/f2fs/f2fs.h

@@ -19,6 +19,7 @@
 #include <linux/magic.h>
 #include <linux/kobject.h>
 #include <linux/sched.h>
+#include <linux/bio.h>
 
 #ifdef CONFIG_F2FS_CHECK_FS
 #define f2fs_bug_on(sbi, condition)	BUG_ON(condition)
@@ -228,6 +229,7 @@ static inline bool __has_cursum_space(struct f2fs_summary_block *sum, int size,
 #define F2FS_IOC_START_VOLATILE_WRITE	_IO(F2FS_IOCTL_MAGIC, 3)
 #define F2FS_IOC_RELEASE_VOLATILE_WRITE	_IO(F2FS_IOCTL_MAGIC, 4)
 #define F2FS_IOC_ABORT_VOLATILE_WRITE	_IO(F2FS_IOCTL_MAGIC, 5)
+#define F2FS_IOC_GARBAGE_COLLECT	_IO(F2FS_IOCTL_MAGIC, 6)
 
 #define F2FS_IOC_SET_ENCRYPTION_POLICY					\
 		_IOR('f', 19, struct f2fs_encryption_policy)
@@ -320,7 +322,7 @@ enum {
 					 */
 };
 
-#define F2FS_LINK_MAX		32000	/* maximum link count per file */
+#define F2FS_LINK_MAX	0xffffffff	/* maximum link count per file */
 
 #define MAX_DIR_RA_PAGES	4	/* maximum ra pages of dir */
 
@@ -349,6 +351,7 @@ struct extent_tree {
 	nid_t ino;			/* inode number */
 	struct rb_root root;		/* root of extent info rb-tree */
 	struct extent_node *cached_en;	/* recently accessed extent node */
+	struct extent_info largest;	/* largested extent info */
 	rwlock_t lock;			/* protect extent info rb-tree */
 	atomic_t refcount;		/* reference count of rb-tree */
 	unsigned int count;		/* # of extent node in rb-tree*/
@@ -372,6 +375,12 @@ struct f2fs_map_blocks {
 	unsigned int m_flags;
 };
 
+/* for flag in get_data_block */
+#define F2FS_GET_BLOCK_READ		0
+#define F2FS_GET_BLOCK_DIO		1
+#define F2FS_GET_BLOCK_FIEMAP		2
+#define F2FS_GET_BLOCK_BMAP		3
+
 /*
  * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
  */
@@ -420,14 +429,13 @@ struct f2fs_inode_info {
 	unsigned int clevel;		/* maximum level of given file name */
 	nid_t i_xattr_nid;		/* node id that contains xattrs */
 	unsigned long long xattr_ver;	/* cp version of xattr modification */
-	struct extent_info ext;		/* in-memory extent cache entry */
-	rwlock_t ext_lock;		/* rwlock for single extent cache */
 	struct inode_entry *dirty_dir;	/* the pointer of dirty dir */
 
-	struct radix_tree_root inmem_root;	/* radix tree for inmem pages */
 	struct list_head inmem_pages;	/* inmemory pages managed by f2fs */
 	struct mutex inmem_lock;	/* lock for inmemory pages */
 
+	struct extent_tree *extent_tree;	/* cached extent_tree entry */
+
 #ifdef CONFIG_F2FS_FS_ENCRYPTION
 	/* Encryption params */
 	struct f2fs_crypt_info *i_crypt_info;
@@ -779,7 +787,11 @@ struct f2fs_sb_info {
 	unsigned int segment_count[2];		/* # of allocated segments */
 	unsigned int block_count[2];		/* # of allocated blocks */
 	atomic_t inplace_count;		/* # of inplace update */
-	int total_hit_ext, read_hit_ext;	/* extent cache hit ratio */
+	atomic_t total_hit_ext;			/* # of lookup extent cache */
+	atomic_t read_hit_rbtree;		/* # of hit rbtree extent node */
+	atomic_t read_hit_largest;		/* # of hit largest extent node */
+	atomic_t read_hit_cached;		/* # of hit cached extent node */
+	atomic_t inline_xattr;			/* # of inline_xattr inodes */
 	atomic_t inline_inode;			/* # of inline_data inodes */
 	atomic_t inline_dir;			/* # of inline_dentry inodes */
 	int bg_gc;				/* background gc calls */
@@ -791,6 +803,11 @@ struct f2fs_sb_info {
 	/* For sysfs suppport */
 	struct kobject s_kobj;
 	struct completion s_kobj_unregister;
+
+	/* For shrinker support */
+	struct list_head s_list;
+	struct mutex umount_mutex;
+	unsigned int shrinker_run_no;
 };
 
 /*
@@ -1039,7 +1056,8 @@ static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
 
 static inline void inode_dec_dirty_pages(struct inode *inode)
 {
-	if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode))
+	if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
+			!S_ISLNK(inode->i_mode))
 		return;
 
 	atomic_dec(&F2FS_I(inode)->dirty_pages);
@@ -1234,16 +1252,24 @@ static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
 						gfp_t flags)
 {
 	void *entry;
-retry:
-	entry = kmem_cache_alloc(cachep, flags);
-	if (!entry) {
-		cond_resched();
-		goto retry;
-	}
 
+	entry = kmem_cache_alloc(cachep, flags);
+	if (!entry)
+		entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
 	return entry;
 }
 
+static inline struct bio *f2fs_bio_alloc(int npages)
+{
+	struct bio *bio;
+
+	/* No failure on bio allocation */
+	bio = bio_alloc(GFP_NOIO, npages);
+	if (!bio)
+		bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages);
+	return bio;
+}
+
 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
 				unsigned long index, void *item)
 {
@@ -1342,6 +1368,7 @@ enum {
 	FI_INC_LINK,		/* need to increment i_nlink */
 	FI_ACL_MODE,		/* indicate acl mode */
 	FI_NO_ALLOC,		/* should not allocate any blocks */
+	FI_FREE_NID,		/* free allocated nide */
 	FI_UPDATE_DIR,		/* should update inode block for consistency */
 	FI_DELAY_IPUT,		/* used for the recovery */
 	FI_NO_EXTENT,		/* not to use the extent cache */
@@ -1541,6 +1568,17 @@ static inline bool is_dot_dotdot(const struct qstr *str)
 	return false;
 }
 
+static inline bool f2fs_may_extent_tree(struct inode *inode)
+{
+	mode_t mode = inode->i_mode;
+
+	if (!test_opt(F2FS_I_SB(inode), EXTENT_CACHE) ||
+			is_inode_flag_set(F2FS_I(inode), FI_NO_EXTENT))
+		return false;
+
+	return S_ISREG(mode);
+}
+
 #define get_inode_mode(i) \
 	((is_inode_flag_set(F2FS_I(i), FI_ACL_MODE)) ? \
 	 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
@@ -1557,7 +1595,7 @@ static inline bool is_dot_dotdot(const struct qstr *str)
 int f2fs_sync_file(struct file *, loff_t, loff_t, int);
 void truncate_data_blocks(struct dnode_of_data *);
 int truncate_blocks(struct inode *, u64, bool);
-void f2fs_truncate(struct inode *);
+int f2fs_truncate(struct inode *, bool);
 int f2fs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
 int f2fs_setattr(struct dentry *, struct iattr *);
 int truncate_hole(struct inode *, pgoff_t, pgoff_t);
@@ -1649,7 +1687,7 @@ int get_dnode_of_data(struct dnode_of_data *, pgoff_t, int);
 int truncate_inode_blocks(struct inode *, pgoff_t);
 int truncate_xattr_node(struct inode *, struct page *);
 int wait_on_node_pages_writeback(struct f2fs_sb_info *, nid_t);
-void remove_inode_page(struct inode *);
+int remove_inode_page(struct inode *);
 struct page *new_inode_page(struct inode *);
 struct page *new_node_page(struct dnode_of_data *, unsigned int, struct page *);
 void ra_node_page(struct f2fs_sb_info *, nid_t);
@@ -1660,6 +1698,7 @@ int sync_node_pages(struct f2fs_sb_info *, nid_t, struct writeback_control *);
 bool alloc_nid(struct f2fs_sb_info *, nid_t *);
 void alloc_nid_done(struct f2fs_sb_info *, nid_t);
 void alloc_nid_failed(struct f2fs_sb_info *, nid_t);
+int try_to_free_nids(struct f2fs_sb_info *, int);
 void recover_inline_xattr(struct inode *, struct page *);
 void recover_xattr_data(struct inode *, struct page *, block_t);
 int recover_inode_page(struct f2fs_sb_info *, struct page *);
@@ -1675,7 +1714,7 @@ void destroy_node_manager_caches(void);
  * segment.c
  */
 void register_inmem_page(struct inode *, struct page *);
-void commit_inmem_pages(struct inode *, bool);
+int commit_inmem_pages(struct inode *, bool);
 void f2fs_balance_fs(struct f2fs_sb_info *);
 void f2fs_balance_fs_bg(struct f2fs_sb_info *);
 int f2fs_issue_flush(struct f2fs_sb_info *);
@@ -1685,7 +1724,7 @@ void invalidate_blocks(struct f2fs_sb_info *, block_t);
 void refresh_sit_entry(struct f2fs_sb_info *, block_t, block_t);
 void clear_prefree_segments(struct f2fs_sb_info *, struct cp_control *);
 void release_discard_addrs(struct f2fs_sb_info *);
-void discard_next_dnode(struct f2fs_sb_info *, block_t);
+bool discard_next_dnode(struct f2fs_sb_info *, block_t);
 int npages_for_summary_flush(struct f2fs_sb_info *, bool);
 void allocate_new_segments(struct f2fs_sb_info *);
 int f2fs_trim_fs(struct f2fs_sb_info *, struct fstrim_range *);
@@ -1727,7 +1766,7 @@ int acquire_orphan_inode(struct f2fs_sb_info *);
 void release_orphan_inode(struct f2fs_sb_info *);
 void add_orphan_inode(struct f2fs_sb_info *, nid_t);
 void remove_orphan_inode(struct f2fs_sb_info *, nid_t);
-void recover_orphan_inodes(struct f2fs_sb_info *);
+int recover_orphan_inodes(struct f2fs_sb_info *);
 int get_valid_checkpoint(struct f2fs_sb_info *);
 void update_dirty_page(struct inode *, struct page *);
 void add_dirty_dir_inode(struct inode *);
@@ -1746,21 +1785,14 @@ int f2fs_submit_page_bio(struct f2fs_io_info *);
 void f2fs_submit_page_mbio(struct f2fs_io_info *);
 void set_data_blkaddr(struct dnode_of_data *);
 int reserve_new_block(struct dnode_of_data *);
+int f2fs_get_block(struct dnode_of_data *, pgoff_t);
 int f2fs_reserve_block(struct dnode_of_data *, pgoff_t);
-void f2fs_shrink_extent_tree(struct f2fs_sb_info *, int);
-void f2fs_destroy_extent_tree(struct inode *);
-void f2fs_init_extent_cache(struct inode *, struct f2fs_extent *);
-void f2fs_update_extent_cache(struct dnode_of_data *);
-void f2fs_preserve_extent_tree(struct inode *);
 struct page *get_read_data_page(struct inode *, pgoff_t, int);
 struct page *find_data_page(struct inode *, pgoff_t);
 struct page *get_lock_data_page(struct inode *, pgoff_t);
 struct page *get_new_data_page(struct inode *, struct page *, pgoff_t, bool);
 int do_write_data_page(struct f2fs_io_info *);
 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *, u64, u64);
-void init_extent_cache_info(struct f2fs_sb_info *);
-int __init create_extent_cache(void);
-void destroy_extent_cache(void);
 void f2fs_invalidate_page(struct page *, unsigned int, unsigned int);
 int f2fs_release_page(struct page *, gfp_t);
 
@@ -1788,11 +1820,13 @@ struct f2fs_stat_info {
 	struct f2fs_sb_info *sbi;
 	int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
 	int main_area_segs, main_area_sections, main_area_zones;
-	int hit_ext, total_ext, ext_tree, ext_node;
+	int hit_largest, hit_cached, hit_rbtree, hit_total, total_ext;
+	int ext_tree, ext_node;
 	int ndirty_node, ndirty_dent, ndirty_dirs, ndirty_meta;
 	int nats, dirty_nats, sits, dirty_sits, fnids;
 	int total_count, utilization;
-	int bg_gc, inline_inode, inline_dir, inmem_pages, wb_pages;
+	int bg_gc, inmem_pages, wb_pages;
+	int inline_xattr, inline_inode, inline_dir;
 	unsigned int valid_count, valid_node_count, valid_inode_count;
 	unsigned int bimodal, avg_vblocks;
 	int util_free, util_valid, util_invalid;
@@ -1823,8 +1857,20 @@ static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
 #define stat_inc_bggc_count(sbi)	((sbi)->bg_gc++)
 #define stat_inc_dirty_dir(sbi)		((sbi)->n_dirty_dirs++)
 #define stat_dec_dirty_dir(sbi)		((sbi)->n_dirty_dirs--)
-#define stat_inc_total_hit(sb)		((F2FS_SB(sb))->total_hit_ext++)
+#define stat_inc_total_hit(sbi)		(atomic_inc(&(sbi)->total_hit_ext))
-#define stat_inc_read_hit(sb)		((F2FS_SB(sb))->read_hit_ext++)
+#define stat_inc_rbtree_node_hit(sbi)	(atomic_inc(&(sbi)->read_hit_rbtree))
+#define stat_inc_largest_node_hit(sbi)	(atomic_inc(&(sbi)->read_hit_largest))
+#define stat_inc_cached_node_hit(sbi)	(atomic_inc(&(sbi)->read_hit_cached))
+#define stat_inc_inline_xattr(inode)					\
+	do {								\
+		if (f2fs_has_inline_xattr(inode))			\
+			(atomic_inc(&F2FS_I_SB(inode)->inline_xattr));	\
+	} while (0)
+#define stat_dec_inline_xattr(inode)					\
+	do {								\
+		if (f2fs_has_inline_xattr(inode))			\
+			(atomic_dec(&F2FS_I_SB(inode)->inline_xattr));	\
+	} while (0)
 #define stat_inc_inline_inode(inode)					\
 	do {								\
 		if (f2fs_has_inline_data(inode))			\
@@ -1894,7 +1940,11 @@ void f2fs_destroy_root_stats(void);
 #define stat_inc_dirty_dir(sbi)
 #define stat_dec_dirty_dir(sbi)
 #define stat_inc_total_hit(sb)
-#define stat_inc_read_hit(sb)
+#define stat_inc_rbtree_node_hit(sb)
+#define stat_inc_largest_node_hit(sbi)
+#define stat_inc_cached_node_hit(sbi)
+#define stat_inc_inline_xattr(inode)
+#define stat_dec_inline_xattr(inode)
 #define stat_inc_inline_inode(inode)
 #define stat_dec_inline_inode(inode)
 #define stat_inc_inline_dir(inode)
@@ -1949,6 +1999,30 @@ bool f2fs_empty_inline_dir(struct inode *);
 int f2fs_read_inline_dir(struct file *, struct dir_context *,
 						struct f2fs_str *);
 
+/*
+ * shrinker.c
+ */
+unsigned long f2fs_shrink_count(struct shrinker *, struct shrink_control *);
+unsigned long f2fs_shrink_scan(struct shrinker *, struct shrink_control *);
+void f2fs_join_shrinker(struct f2fs_sb_info *);
+void f2fs_leave_shrinker(struct f2fs_sb_info *);
+
+/*
+ * extent_cache.c
+ */
+unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *, int);
+void f2fs_drop_largest_extent(struct inode *, pgoff_t);
+void f2fs_init_extent_tree(struct inode *, struct f2fs_extent *);
+unsigned int f2fs_destroy_extent_node(struct inode *);
+void f2fs_destroy_extent_tree(struct inode *);
+bool f2fs_lookup_extent_cache(struct inode *, pgoff_t, struct extent_info *);
+void f2fs_update_extent_cache(struct dnode_of_data *);
+void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
+						pgoff_t, block_t, unsigned int);
+void init_extent_cache_info(struct f2fs_sb_info *);
+int __init create_extent_cache(void);
+void destroy_extent_cache(void);
+
 /*
  * crypto support
  */

fs/f2fs/file.c

@@ -27,6 +27,7 @@
 #include "segment.h"
 #include "xattr.h"
 #include "acl.h"
+#include "gc.h"
 #include "trace.h"
 #include <trace/events/f2fs.h>
 
@@ -85,6 +86,8 @@ static int f2fs_vm_page_mkwrite(struct vm_area_struct *vma,
 mapped:
 	/* fill the page */
 	f2fs_wait_on_page_writeback(page, DATA);
+	/* if gced page is attached, don't write to cold segment */
+	clear_cold_data(page);
 out:
 	sb_end_pagefault(inode->i_sb);
 	return block_page_mkwrite_return(err);
@@ -203,8 +206,8 @@ int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	}
 
 	/* if the inode is dirty, let's recover all the time */
-	if (!datasync && is_inode_flag_set(fi, FI_DIRTY_INODE)) {
+	if (!datasync) {
-		update_inode_page(inode);
+		f2fs_write_inode(inode, NULL);
 		goto go_write;
 	}
 
@@ -442,9 +445,9 @@ static int f2fs_file_open(struct inode *inode, struct file *filp)
 
 int truncate_data_blocks_range(struct dnode_of_data *dn, int count)
 {
-	int nr_free = 0, ofs = dn->ofs_in_node;
 	struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
 	struct f2fs_node *raw_node;
+	int nr_free = 0, ofs = dn->ofs_in_node, len = count;
 	__le32 *addr;
 
 	raw_node = F2FS_NODE(dn->node_page);
@@ -457,14 +460,22 @@ int truncate_data_blocks_range(struct dnode_of_data *dn, int count)
 
 		dn->data_blkaddr = NULL_ADDR;
 		set_data_blkaddr(dn);
-		f2fs_update_extent_cache(dn);
 		invalidate_blocks(sbi, blkaddr);
 		if (dn->ofs_in_node == 0 && IS_INODE(dn->node_page))
 			clear_inode_flag(F2FS_I(dn->inode),
 						FI_FIRST_BLOCK_WRITTEN);
 		nr_free++;
 	}
+
 	if (nr_free) {
+		pgoff_t fofs;
+		/*
+		 * once we invalidate valid blkaddr in range [ofs, ofs + count],
+		 * we will invalidate all blkaddr in the whole range.
+		 */
+		fofs = start_bidx_of_node(ofs_of_node(dn->node_page),
+						F2FS_I(dn->inode)) + ofs;
+		f2fs_update_extent_cache_range(dn, fofs, 0, len);
 		dec_valid_block_count(sbi, dn->inode, nr_free);
 		set_page_dirty(dn->node_page);
 		sync_inode_page(dn);
@@ -576,24 +587,30 @@ int truncate_blocks(struct inode *inode, u64 from, bool lock)
 	return err;
 }
 
-void f2fs_truncate(struct inode *inode)
+int f2fs_truncate(struct inode *inode, bool lock)
 {
+	int err;
+
 	if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
 				S_ISLNK(inode->i_mode)))
-		return;
+		return 0;
 
 	trace_f2fs_truncate(inode);
 
 	/* we should check inline_data size */
 	if (f2fs_has_inline_data(inode) && !f2fs_may_inline_data(inode)) {
-		if (f2fs_convert_inline_inode(inode))
+		err = f2fs_convert_inline_inode(inode);
-			return;
+		if (err)
+			return err;
 	}
 
-	if (!truncate_blocks(inode, i_size_read(inode), true)) {
+	err = truncate_blocks(inode, i_size_read(inode), lock);
-		inode->i_mtime = inode->i_ctime = CURRENT_TIME;
+	if (err)
-		mark_inode_dirty(inode);
+		return err;
-	}
+
+	inode->i_mtime = inode->i_ctime = CURRENT_TIME;
+	mark_inode_dirty(inode);
+	return 0;
 }
 
 int f2fs_getattr(struct vfsmount *mnt,
@@ -653,7 +670,9 @@ int f2fs_setattr(struct dentry *dentry, struct iattr *attr)
 
 		if (attr->ia_size <= i_size_read(inode)) {
 			truncate_setsize(inode, attr->ia_size);
-			f2fs_truncate(inode);
+			err = f2fs_truncate(inode, true);
+			if (err)
+				return err;
 			f2fs_balance_fs(F2FS_I_SB(inode));
 		} else {
 			/*
@@ -692,14 +711,14 @@ const struct inode_operations f2fs_file_inode_operations = {
 	.fiemap		= f2fs_fiemap,
 };
 
-static void fill_zero(struct inode *inode, pgoff_t index,
+static int fill_zero(struct inode *inode, pgoff_t index,
 					loff_t start, loff_t len)
 {
 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
 	struct page *page;
 
 	if (!len)
-		return;
+		return 0;
 
 	f2fs_balance_fs(sbi);
 
@@ -707,12 +726,14 @@ static void fill_zero(struct inode *inode, pgoff_t index,
 	page = get_new_data_page(inode, NULL, index, false);
 	f2fs_unlock_op(sbi);
 
-	if (!IS_ERR(page)) {
+	if (IS_ERR(page))
-		f2fs_wait_on_page_writeback(page, DATA);
+		return PTR_ERR(page);
-		zero_user(page, start, len);
+
-		set_page_dirty(page);
+	f2fs_wait_on_page_writeback(page, DATA);
-		f2fs_put_page(page, 1);
+	zero_user(page, start, len);
-	}
+	set_page_dirty(page);
+	f2fs_put_page(page, 1);
+	return 0;
 }
 
 int truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end)
@@ -760,14 +781,22 @@ static int punch_hole(struct inode *inode, loff_t offset, loff_t len)
 	off_end = (offset + len) & (PAGE_CACHE_SIZE - 1);
 
 	if (pg_start == pg_end) {
-		fill_zero(inode, pg_start, off_start,
+		ret = fill_zero(inode, pg_start, off_start,
 						off_end - off_start);
+		if (ret)
+			return ret;
 	} else {
-		if (off_start)
+		if (off_start) {
-			fill_zero(inode, pg_start++, off_start,
+			ret = fill_zero(inode, pg_start++, off_start,
-					PAGE_CACHE_SIZE - off_start);
+						PAGE_CACHE_SIZE - off_start);
-		if (off_end)
+			if (ret)
-			fill_zero(inode, pg_end, 0, off_end);
+				return ret;
+		}
+		if (off_end) {
+			ret = fill_zero(inode, pg_end, 0, off_end);
+			if (ret)
+				return ret;
+		}
 
 		if (pg_start < pg_end) {
 			struct address_space *mapping = inode->i_mapping;
@@ -797,11 +826,11 @@ static int f2fs_do_collapse(struct inode *inode, pgoff_t start, pgoff_t end)
 	pgoff_t nrpages = (i_size_read(inode) + PAGE_SIZE - 1) / PAGE_SIZE;
 	int ret = 0;
 
-	f2fs_lock_op(sbi);
-
 	for (; end < nrpages; start++, end++) {
 		block_t new_addr, old_addr;
 
+		f2fs_lock_op(sbi);
+
 		set_new_dnode(&dn, inode, NULL, NULL, 0);
 		ret = get_dnode_of_data(&dn, end, LOOKUP_NODE_RA);
 		if (ret && ret != -ENOENT) {
@@ -817,13 +846,16 @@ static int f2fs_do_collapse(struct inode *inode, pgoff_t start, pgoff_t end)
 		if (new_addr == NULL_ADDR) {
 			set_new_dnode(&dn, inode, NULL, NULL, 0);
 			ret = get_dnode_of_data(&dn, start, LOOKUP_NODE_RA);
-			if (ret && ret != -ENOENT)
+			if (ret && ret != -ENOENT) {
 				goto out;
-			else if (ret == -ENOENT)
+			} else if (ret == -ENOENT) {
+				f2fs_unlock_op(sbi);
 				continue;
+			}
 
 			if (dn.data_blkaddr == NULL_ADDR) {
 				f2fs_put_dnode(&dn);
+				f2fs_unlock_op(sbi);
 				continue;
 			} else {
 				truncate_data_blocks_range(&dn, 1);
@@ -862,8 +894,9 @@ static int f2fs_do_collapse(struct inode *inode, pgoff_t start, pgoff_t end)
 
 			f2fs_put_dnode(&dn);
 		}
+		f2fs_unlock_op(sbi);
 	}
-	ret = 0;
+	return 0;
 out:
 	f2fs_unlock_op(sbi);
 	return ret;
@@ -885,6 +918,14 @@ static int f2fs_collapse_range(struct inode *inode, loff_t offset, loff_t len)
 	if (offset & (F2FS_BLKSIZE - 1) || len & (F2FS_BLKSIZE - 1))
 		return -EINVAL;
 
+	f2fs_balance_fs(F2FS_I_SB(inode));
+
+	if (f2fs_has_inline_data(inode)) {
+		ret = f2fs_convert_inline_inode(inode);
+		if (ret)
+			return ret;
+	}
+
 	pg_start = offset >> PAGE_CACHE_SHIFT;
 	pg_end = (offset + len) >> PAGE_CACHE_SHIFT;
 
@@ -946,14 +987,21 @@ static int f2fs_zero_range(struct inode *inode, loff_t offset, loff_t len,
 	off_end = (offset + len) & (PAGE_CACHE_SIZE - 1);
 
 	if (pg_start == pg_end) {
-		fill_zero(inode, pg_start, off_start, off_end - off_start);
+		ret = fill_zero(inode, pg_start, off_start,
+						off_end - off_start);
+		if (ret)
+			return ret;
+
 		if (offset + len > new_size)
 			new_size = offset + len;
 		new_size = max_t(loff_t, new_size, offset + len);
 	} else {
 		if (off_start) {
-			fill_zero(inode, pg_start++, off_start,
+			ret = fill_zero(inode, pg_start++, off_start,
-					PAGE_CACHE_SIZE - off_start);
+						PAGE_CACHE_SIZE - off_start);
+			if (ret)
+				return ret;
+
 			new_size = max_t(loff_t, new_size,
 						pg_start << PAGE_CACHE_SHIFT);
 		}
@@ -995,7 +1043,10 @@ static int f2fs_zero_range(struct inode *inode, loff_t offset, loff_t len,
 		}
 
 		if (off_end) {
-			fill_zero(inode, pg_end, 0, off_end);
+			ret = fill_zero(inode, pg_end, 0, off_end);
+			if (ret)
+				goto out;
+
 			new_size = max_t(loff_t, new_size, offset + len);
 		}
 	}
@@ -1033,6 +1084,12 @@ static int f2fs_insert_range(struct inode *inode, loff_t offset, loff_t len)
 
 	f2fs_balance_fs(sbi);
 
+	if (f2fs_has_inline_data(inode)) {
+		ret = f2fs_convert_inline_inode(inode);
+		if (ret)
+			return ret;
+	}
+
 	ret = truncate_blocks(inode, i_size_read(inode), true);
 	if (ret)
 		return ret;
@@ -1302,6 +1359,7 @@ static int f2fs_ioc_getversion(struct file *filp, unsigned long arg)
 static int f2fs_ioc_start_atomic_write(struct file *filp)
 {
 	struct inode *inode = file_inode(filp);
+	int ret;
 
 	if (!inode_owner_or_capable(inode))
 		return -EACCES;
@@ -1311,9 +1369,12 @@ static int f2fs_ioc_start_atomic_write(struct file *filp)
 	if (f2fs_is_atomic_file(inode))
 		return 0;
 
-	set_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE);
+	ret = f2fs_convert_inline_inode(inode);
+	if (ret)
+		return ret;
 
-	return f2fs_convert_inline_inode(inode);
+	set_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE);
+	return 0;
 }
 
 static int f2fs_ioc_commit_atomic_write(struct file *filp)
@@ -1333,10 +1394,13 @@ static int f2fs_ioc_commit_atomic_write(struct file *filp)
 
 	if (f2fs_is_atomic_file(inode)) {
 		clear_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE);
-		commit_inmem_pages(inode, false);
+		ret = commit_inmem_pages(inode, false);
+		if (ret)
+			goto err_out;
 	}
 
-	ret = f2fs_sync_file(filp, 0, LONG_MAX, 0);
+	ret = f2fs_sync_file(filp, 0, LLONG_MAX, 0);
+err_out:
 	mnt_drop_write_file(filp);
 	return ret;
 }
@@ -1344,6 +1408,7 @@ static int f2fs_ioc_commit_atomic_write(struct file *filp)
 static int f2fs_ioc_start_volatile_write(struct file *filp)
 {
 	struct inode *inode = file_inode(filp);
+	int ret;
 
 	if (!inode_owner_or_capable(inode))
 		return -EACCES;
@@ -1351,9 +1416,12 @@ static int f2fs_ioc_start_volatile_write(struct file *filp)
 	if (f2fs_is_volatile_file(inode))
 		return 0;
 
-	set_inode_flag(F2FS_I(inode), FI_VOLATILE_FILE);
+	ret = f2fs_convert_inline_inode(inode);
+	if (ret)
+		return ret;
 
-	return f2fs_convert_inline_inode(inode);
+	set_inode_flag(F2FS_I(inode), FI_VOLATILE_FILE);
+	return 0;
 }
 
 static int f2fs_ioc_release_volatile_write(struct file *filp)
@@ -1389,7 +1457,7 @@ static int f2fs_ioc_abort_volatile_write(struct file *filp)
 
 	if (f2fs_is_atomic_file(inode)) {
 		clear_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE);
-		commit_inmem_pages(inode, false);
+		commit_inmem_pages(inode, true);
 	}
 
 	if (f2fs_is_volatile_file(inode))
@@ -1544,6 +1612,35 @@ static int f2fs_ioc_get_encryption_pwsalt(struct file *filp, unsigned long arg)
 	return 0;
 }
 
+static int f2fs_ioc_gc(struct file *filp, unsigned long arg)
+{
+	struct inode *inode = file_inode(filp);
+	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	__u32 i, count;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	if (get_user(count, (__u32 __user *)arg))
+		return -EFAULT;
+
+	if (!count || count > F2FS_BATCH_GC_MAX_NUM)
+		return -EINVAL;
+
+	for (i = 0; i < count; i++) {
+		if (!mutex_trylock(&sbi->gc_mutex))
+			break;
+
+		if (f2fs_gc(sbi))
+			break;
+	}
+
+	if (put_user(i, (__u32 __user *)arg))
+		return -EFAULT;
+
+	return 0;
+}
+
 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 {
 	switch (cmd) {
@@ -1573,6 +1670,8 @@ long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 		return f2fs_ioc_get_encryption_policy(filp, arg);
 	case F2FS_IOC_GET_ENCRYPTION_PWSALT:
 		return f2fs_ioc_get_encryption_pwsalt(filp, arg);
+	case F2FS_IOC_GARBAGE_COLLECT:
+		return f2fs_ioc_gc(filp, arg);
 	default:
 		return -ENOTTY;
 	}

fs/f2fs/gc.c

@@ -391,23 +391,27 @@ static int check_valid_map(struct f2fs_sb_info *sbi,
  * On validity, copy that node with cold status, otherwise (invalid node)
  * ignore that.
  */
-static void gc_node_segment(struct f2fs_sb_info *sbi,
+static int gc_node_segment(struct f2fs_sb_info *sbi,
 		struct f2fs_summary *sum, unsigned int segno, int gc_type)
 {
 	bool initial = true;
 	struct f2fs_summary *entry;
+	block_t start_addr;
 	int off;
 
+	start_addr = START_BLOCK(sbi, segno);
+
 next_step:
 	entry = sum;
 
 	for (off = 0; off < sbi->blocks_per_seg; off++, entry++) {
 		nid_t nid = le32_to_cpu(entry->nid);
 		struct page *node_page;
+		struct node_info ni;
 
 		/* stop BG_GC if there is not enough free sections. */
 		if (gc_type == BG_GC && has_not_enough_free_secs(sbi, 0))
-			return;
+			return 0;
 
 		if (check_valid_map(sbi, segno, off) == 0)
 			continue;
@@ -426,6 +430,12 @@ static void gc_node_segment(struct f2fs_sb_info *sbi,
 			continue;
 		}
 
+		get_node_info(sbi, nid, &ni);
+		if (ni.blk_addr != start_addr + off) {
+			f2fs_put_page(node_page, 1);
+			continue;
+		}
+
 		/* set page dirty and write it */
 		if (gc_type == FG_GC) {
 			f2fs_wait_on_page_writeback(node_page, NODE);
@@ -451,13 +461,11 @@ static void gc_node_segment(struct f2fs_sb_info *sbi,
 		};
 		sync_node_pages(sbi, 0, &wbc);
 
-		/*
+		/* return 1 only if FG_GC succefully reclaimed one */
-		 * In the case of FG_GC, it'd be better to reclaim this victim
+		if (get_valid_blocks(sbi, segno, 1) == 0)
-		 * completely.
+			return 1;
-		 */
-		if (get_valid_blocks(sbi, segno, 1) != 0)
-			goto next_step;
 	}
+	return 0;
 }
 
 /*
@@ -487,7 +495,7 @@ block_t start_bidx_of_node(unsigned int node_ofs, struct f2fs_inode_info *fi)
 	return bidx * ADDRS_PER_BLOCK + ADDRS_PER_INODE(fi);
 }
 
-static int check_dnode(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
+static bool is_alive(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
 		struct node_info *dni, block_t blkaddr, unsigned int *nofs)
 {
 	struct page *node_page;
@@ -500,13 +508,13 @@ static int check_dnode(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
 
 	node_page = get_node_page(sbi, nid);
 	if (IS_ERR(node_page))
-		return 0;
+		return false;
 
 	get_node_info(sbi, nid, dni);
 
 	if (sum->version != dni->version) {
 		f2fs_put_page(node_page, 1);
-		return 0;
+		return false;
 	}
 
 	*nofs = ofs_of_node(node_page);
@@ -514,8 +522,8 @@ static int check_dnode(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
 	f2fs_put_page(node_page, 1);
 
 	if (source_blkaddr != blkaddr)
-		return 0;
+		return false;
-	return 1;
+	return true;
 }
 
 static void move_encrypted_block(struct inode *inode, block_t bidx)
@@ -552,7 +560,10 @@ static void move_encrypted_block(struct inode *inode, block_t bidx)
 	fio.page = page;
 	fio.blk_addr = dn.data_blkaddr;
 
-	fio.encrypted_page = grab_cache_page(META_MAPPING(fio.sbi), fio.blk_addr);
+	fio.encrypted_page = pagecache_get_page(META_MAPPING(fio.sbi),
+					fio.blk_addr,
+					FGP_LOCK|FGP_CREAT,
+					GFP_NOFS);
 	if (!fio.encrypted_page)
 		goto put_out;
 
@@ -636,7 +647,7 @@ static void move_data_page(struct inode *inode, block_t bidx, int gc_type)
  * If the parent node is not valid or the data block address is different,
  * the victim data block is ignored.
  */
-static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
+static int gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
 		struct gc_inode_list *gc_list, unsigned int segno, int gc_type)
 {
 	struct super_block *sb = sbi->sb;
@@ -659,7 +670,7 @@ static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
 
 		/* stop BG_GC if there is not enough free sections. */
 		if (gc_type == BG_GC && has_not_enough_free_secs(sbi, 0))
-			return;
+			return 0;
 
 		if (check_valid_map(sbi, segno, off) == 0)
 			continue;
@@ -670,7 +681,7 @@ static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
 		}
 
 		/* Get an inode by ino with checking validity */
-		if (check_dnode(sbi, entry, &dni, start_addr + off, &nofs) == 0)
+		if (!is_alive(sbi, entry, &dni, start_addr + off, &nofs))
 			continue;
 
 		if (phase == 1) {
@@ -724,15 +735,11 @@ static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
 	if (gc_type == FG_GC) {
 		f2fs_submit_merged_bio(sbi, DATA, WRITE);
 
-		/*
+		/* return 1 only if FG_GC succefully reclaimed one */
-		 * In the case of FG_GC, it'd be better to reclaim this victim
+		if (get_valid_blocks(sbi, segno, 1) == 0)
-		 * completely.
+			return 1;
-		 */
-		if (get_valid_blocks(sbi, segno, 1) != 0) {
-			phase = 2;
-			goto next_step;
-		}
 	}
+	return 0;
 }
 
 static int __get_victim(struct f2fs_sb_info *sbi, unsigned int *victim,
@@ -748,12 +755,13 @@ static int __get_victim(struct f2fs_sb_info *sbi, unsigned int *victim,
 	return ret;
 }
 
-static void do_garbage_collect(struct f2fs_sb_info *sbi, unsigned int segno,
+static int do_garbage_collect(struct f2fs_sb_info *sbi, unsigned int segno,
 				struct gc_inode_list *gc_list, int gc_type)
 {
 	struct page *sum_page;
 	struct f2fs_summary_block *sum;
 	struct blk_plug plug;
+	int nfree = 0;
 
 	/* read segment summary of victim */
 	sum_page = get_sum_page(sbi, segno);
@@ -773,10 +781,11 @@ static void do_garbage_collect(struct f2fs_sb_info *sbi, unsigned int segno,
 
 	switch (GET_SUM_TYPE((&sum->footer))) {
 	case SUM_TYPE_NODE:
-		gc_node_segment(sbi, sum->entries, segno, gc_type);
+		nfree = gc_node_segment(sbi, sum->entries, segno, gc_type);
 		break;
 	case SUM_TYPE_DATA:
-		gc_data_segment(sbi, sum->entries, gc_list, segno, gc_type);
+		nfree = gc_data_segment(sbi, sum->entries, gc_list,
+							segno, gc_type);
 		break;
 	}
 	blk_finish_plug(&plug);
@@ -785,11 +794,13 @@ static void do_garbage_collect(struct f2fs_sb_info *sbi, unsigned int segno,
 	stat_inc_call_count(sbi->stat_info);
 
 	f2fs_put_page(sum_page, 0);
+	return nfree;
 }
 
 int f2fs_gc(struct f2fs_sb_info *sbi)
 {
-	unsigned int segno, i;
+	unsigned int segno = NULL_SEGNO;
+	unsigned int i;
 	int gc_type = BG_GC;
 	int nfree = 0;
 	int ret = -1;
@@ -808,10 +819,11 @@ int f2fs_gc(struct f2fs_sb_info *sbi)
 
 	if (gc_type == BG_GC && has_not_enough_free_secs(sbi, nfree)) {
 		gc_type = FG_GC;
-		write_checkpoint(sbi, &cpc);
+		if (__get_victim(sbi, &segno, gc_type) || prefree_segments(sbi))
+			write_checkpoint(sbi, &cpc);
 	}
 
-	if (!__get_victim(sbi, &segno, gc_type))
+	if (segno == NULL_SEGNO && !__get_victim(sbi, &segno, gc_type))
 		goto stop;
 	ret = 0;
 
@@ -821,13 +833,10 @@ int f2fs_gc(struct f2fs_sb_info *sbi)
 								META_SSA);
 
 	for (i = 0; i < sbi->segs_per_sec; i++)
-		do_garbage_collect(sbi, segno + i, &gc_list, gc_type);
+		nfree += do_garbage_collect(sbi, segno + i, &gc_list, gc_type);
 
-	if (gc_type == FG_GC) {
+	if (gc_type == FG_GC)
 		sbi->cur_victim_sec = NULL_SEGNO;
-		nfree++;
-		WARN_ON(get_valid_blocks(sbi, segno, sbi->segs_per_sec));
-	}
 
 	if (has_not_enough_free_secs(sbi, nfree))
 		goto gc_more;

fs/f2fs/gc.h

@@ -19,6 +19,12 @@
 #define LIMIT_INVALID_BLOCK	40 /* percentage over total user space */
 #define LIMIT_FREE_BLOCK	40 /* percentage over invalid + free space */
 
+/*
+ * with this macro, we can control the max time we do garbage collection,
+ * when user triggers batch mode gc by ioctl.
+ */
+#define F2FS_BATCH_GC_MAX_NUM		16
+
 /* Search max. number of dirty segments to select a victim segment */
 #define DEF_MAX_VICTIM_SEARCH 4096 /* covers 8GB */
 

fs/f2fs/inline.c

@@ -360,6 +360,10 @@ int make_empty_inline_dir(struct inode *inode, struct inode *parent,
 	return 0;
 }
 
+/*
+ * NOTE: ipage is grabbed by caller, but if any error occurs, we should
+ * release ipage in this function.
+ */
 static int f2fs_convert_inline_dir(struct inode *dir, struct page *ipage,
 				struct f2fs_inline_dentry *inline_dentry)
 {
@@ -369,8 +373,10 @@ static int f2fs_convert_inline_dir(struct inode *dir, struct page *ipage,
 	int err;
 
 	page = grab_cache_page(dir->i_mapping, 0);
-	if (!page)
+	if (!page) {
+		f2fs_put_page(ipage, 1);
 		return -ENOMEM;
+	}
 
 	set_new_dnode(&dn, dir, ipage, NULL, 0);
 	err = f2fs_reserve_block(&dn, 0);
@@ -378,13 +384,21 @@ static int f2fs_convert_inline_dir(struct inode *dir, struct page *ipage,
 		goto out;
 
 	f2fs_wait_on_page_writeback(page, DATA);
-	zero_user_segment(page, 0, PAGE_CACHE_SIZE);
+	zero_user_segment(page, MAX_INLINE_DATA, PAGE_CACHE_SIZE);
 
 	dentry_blk = kmap_atomic(page);
 
 	/* copy data from inline dentry block to new dentry block */
 	memcpy(dentry_blk->dentry_bitmap, inline_dentry->dentry_bitmap,
 					INLINE_DENTRY_BITMAP_SIZE);
+	memset(dentry_blk->dentry_bitmap + INLINE_DENTRY_BITMAP_SIZE, 0,
+			SIZE_OF_DENTRY_BITMAP - INLINE_DENTRY_BITMAP_SIZE);
+	/*
+	 * we do not need to zero out remainder part of dentry and filename
+	 * field, since we have used bitmap for marking the usage status of
+	 * them, besides, we can also ignore copying/zeroing reserved space
+	 * of dentry block, because them haven't been used so far.
+	 */
 	memcpy(dentry_blk->dentry, inline_dentry->dentry,
 			sizeof(struct f2fs_dir_entry) * NR_INLINE_DENTRY);
 	memcpy(dentry_blk->filename, inline_dentry->filename,
@@ -434,8 +448,9 @@ int f2fs_add_inline_entry(struct inode *dir, const struct qstr *name,
 						slots, NR_INLINE_DENTRY);
 	if (bit_pos >= NR_INLINE_DENTRY) {
 		err = f2fs_convert_inline_dir(dir, ipage, dentry_blk);
-		if (!err)
+		if (err)
-			err = -EAGAIN;
+			return err;
+		err = -EAGAIN;
 		goto out;
 	}
 

fs/f2fs/inode.c

@@ -12,7 +12,6 @@
 #include <linux/f2fs_fs.h>
 #include <linux/buffer_head.h>
 #include <linux/writeback.h>
-#include <linux/bitops.h>
 
 #include "f2fs.h"
 #include "node.h"
@@ -34,8 +33,8 @@ void f2fs_set_inode_flags(struct inode *inode)
 		new_fl |= S_NOATIME;
 	if (flags & FS_DIRSYNC_FL)
 		new_fl |= S_DIRSYNC;
-	set_mask_bits(&inode->i_flags,
+	inode_set_flags(inode, new_fl,
-			S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC, new_fl);
+			S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
 }
 
 static void __get_inode_rdev(struct inode *inode, struct f2fs_inode *ri)
@@ -139,7 +138,7 @@ static int do_read_inode(struct inode *inode)
 	fi->i_pino = le32_to_cpu(ri->i_pino);
 	fi->i_dir_level = ri->i_dir_level;
 
-	f2fs_init_extent_cache(inode, &ri->i_ext);
+	f2fs_init_extent_tree(inode, &ri->i_ext);
 
 	get_inline_info(fi, ri);
 
@@ -155,6 +154,7 @@ static int do_read_inode(struct inode *inode)
 
 	f2fs_put_page(node_page, 1);
 
+	stat_inc_inline_xattr(inode);
 	stat_inc_inline_inode(inode);
 	stat_inc_inline_dir(inode);
 
@@ -237,10 +237,11 @@ void update_inode(struct inode *inode, struct page *node_page)
 	ri->i_size = cpu_to_le64(i_size_read(inode));
 	ri->i_blocks = cpu_to_le64(inode->i_blocks);
 
-	read_lock(&F2FS_I(inode)->ext_lock);
+	if (F2FS_I(inode)->extent_tree)
-	set_raw_extent(&F2FS_I(inode)->ext, &ri->i_ext);
+		set_raw_extent(&F2FS_I(inode)->extent_tree->largest,
-	read_unlock(&F2FS_I(inode)->ext_lock);
+							&ri->i_ext);
-
+	else
+		memset(&ri->i_ext, 0, sizeof(ri->i_ext));
 	set_raw_inline(F2FS_I(inode), ri);
 
 	ri->i_atime = cpu_to_le64(inode->i_atime.tv_sec);
@@ -314,7 +315,9 @@ int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc)
 void f2fs_evict_inode(struct inode *inode)
 {
 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
-	nid_t xnid = F2FS_I(inode)->i_xattr_nid;
+	struct f2fs_inode_info *fi = F2FS_I(inode);
+	nid_t xnid = fi->i_xattr_nid;
+	int err = 0;
 
 	/* some remained atomic pages should discarded */
 	if (f2fs_is_atomic_file(inode))
@@ -330,41 +333,62 @@ void f2fs_evict_inode(struct inode *inode)
 	f2fs_bug_on(sbi, get_dirty_pages(inode));
 	remove_dirty_dir_inode(inode);
 
+	f2fs_destroy_extent_tree(inode);
+
 	if (inode->i_nlink || is_bad_inode(inode))
 		goto no_delete;
 
 	sb_start_intwrite(inode->i_sb);
-	set_inode_flag(F2FS_I(inode), FI_NO_ALLOC);
+	set_inode_flag(fi, FI_NO_ALLOC);
 	i_size_write(inode, 0);
 
 	if (F2FS_HAS_BLOCKS(inode))
-		f2fs_truncate(inode);
+		err = f2fs_truncate(inode, true);
 
-	f2fs_lock_op(sbi);
+	if (!err) {
-	remove_inode_page(inode);
+		f2fs_lock_op(sbi);
-	f2fs_unlock_op(sbi);
+		err = remove_inode_page(inode);
+		f2fs_unlock_op(sbi);
+	}
 
 	sb_end_intwrite(inode->i_sb);
 no_delete:
+	stat_dec_inline_xattr(inode);
 	stat_dec_inline_dir(inode);
 	stat_dec_inline_inode(inode);
 
-	/* update extent info in inode */
-	if (inode->i_nlink)
-		f2fs_preserve_extent_tree(inode);
-	f2fs_destroy_extent_tree(inode);
-
 	invalidate_mapping_pages(NODE_MAPPING(sbi), inode->i_ino, inode->i_ino);
 	if (xnid)
 		invalidate_mapping_pages(NODE_MAPPING(sbi), xnid, xnid);
-	if (is_inode_flag_set(F2FS_I(inode), FI_APPEND_WRITE))
+	if (is_inode_flag_set(fi, FI_APPEND_WRITE))
 		add_dirty_inode(sbi, inode->i_ino, APPEND_INO);
-	if (is_inode_flag_set(F2FS_I(inode), FI_UPDATE_WRITE))
+	if (is_inode_flag_set(fi, FI_UPDATE_WRITE))
 		add_dirty_inode(sbi, inode->i_ino, UPDATE_INO);
+	if (is_inode_flag_set(fi, FI_FREE_NID)) {
+		if (err && err != -ENOENT)
+			alloc_nid_done(sbi, inode->i_ino);
+		else
+			alloc_nid_failed(sbi, inode->i_ino);
+		clear_inode_flag(fi, FI_FREE_NID);
+	}
+
+	if (err && err != -ENOENT) {
+		if (!exist_written_data(sbi, inode->i_ino, ORPHAN_INO)) {
+			/*
+			 * get here because we failed to release resource
+			 * of inode previously, reminder our user to run fsck
+			 * for fixing.
+			 */
+			set_sbi_flag(sbi, SBI_NEED_FSCK);
+			f2fs_msg(sbi->sb, KERN_WARNING,
+				"inode (ino:%lu) resource leak, run fsck "
+				"to fix this issue!", inode->i_ino);
+		}
+	}
 out_clear:
 #ifdef CONFIG_F2FS_FS_ENCRYPTION
-	if (F2FS_I(inode)->i_crypt_info)
+	if (fi->i_crypt_info)
-		f2fs_free_encryption_info(inode, F2FS_I(inode)->i_crypt_info);
+		f2fs_free_encryption_info(inode, fi->i_crypt_info);
 #endif
 	clear_inode(inode);
 }
@@ -373,6 +397,7 @@ void f2fs_evict_inode(struct inode *inode)
 void handle_failed_inode(struct inode *inode)
 {
 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	int err = 0;
 
 	clear_nlink(inode);
 	make_bad_inode(inode);
@@ -380,13 +405,29 @@ void handle_failed_inode(struct inode *inode)
 
 	i_size_write(inode, 0);
 	if (F2FS_HAS_BLOCKS(inode))
-		f2fs_truncate(inode);
+		err = f2fs_truncate(inode, false);
 
-	remove_inode_page(inode);
+	if (!err)
+		err = remove_inode_page(inode);
 
-	clear_inode_flag(F2FS_I(inode), FI_INLINE_DATA);
+	/*
-	clear_inode_flag(F2FS_I(inode), FI_INLINE_DENTRY);
+	 * if we skip truncate_node in remove_inode_page bacause we failed
-	alloc_nid_failed(sbi, inode->i_ino);
+	 * before, it's better to find another way to release resource of
+	 * this inode (e.g. valid block count, node block or nid). Here we
+	 * choose to add this inode to orphan list, so that we can call iput
+	 * for releasing in orphan recovery flow.
+	 *
+	 * Note: we should add inode to orphan list before f2fs_unlock_op()
+	 * so we can prevent losing this orphan when encoutering checkpoint
+	 * and following suddenly power-off.
+	 */
+	if (err && err != -ENOENT) {
+		err = acquire_orphan_inode(sbi);
+		if (!err)
+			add_orphan_inode(sbi, inode->i_ino);
+	}
+
+	set_inode_flag(F2FS_I(inode), FI_FREE_NID);
 	f2fs_unlock_op(sbi);
 
 	/* iput will drop the inode object */

fs/f2fs/namei.c

@@ -53,7 +53,7 @@ static struct inode *f2fs_new_inode(struct inode *dir, umode_t mode)
 	if (err) {
 		err = -EINVAL;
 		nid_free = true;
-		goto out;
+		goto fail;
 	}
 
 	/* If the directory encrypted, then we should encrypt the inode. */
@@ -65,6 +65,9 @@ static struct inode *f2fs_new_inode(struct inode *dir, umode_t mode)
 	if (f2fs_may_inline_dentry(inode))
 		set_inode_flag(F2FS_I(inode), FI_INLINE_DENTRY);
 
+	f2fs_init_extent_tree(inode, NULL);
+
+	stat_inc_inline_xattr(inode);
 	stat_inc_inline_inode(inode);
 	stat_inc_inline_dir(inode);
 
@@ -72,15 +75,12 @@ static struct inode *f2fs_new_inode(struct inode *dir, umode_t mode)
 	mark_inode_dirty(inode);
 	return inode;
 
-out:
-	clear_nlink(inode);
-	unlock_new_inode(inode);
 fail:
 	trace_f2fs_new_inode(inode, err);
 	make_bad_inode(inode);
-	iput(inode);
 	if (nid_free)
-		alloc_nid_failed(sbi, ino);
+		set_inode_flag(F2FS_I(inode), FI_FREE_NID);
+	iput(inode);
 	return ERR_PTR(err);
 }
 
@@ -89,7 +89,14 @@ static int is_multimedia_file(const unsigned char *s, const char *sub)
 	size_t slen = strlen(s);
 	size_t sublen = strlen(sub);
 
-	if (sublen > slen)
+	/*
+	 * filename format of multimedia file should be defined as:
+	 * "filename + '.' + extension".
+	 */
+	if (slen < sublen + 2)
+		return 0;
+
+	if (s[slen - sublen - 1] != '.')
 		return 0;
 
 	return !strncasecmp(s + slen - sublen, sub, sublen);

fs/f2fs/node.c

@@ -159,7 +159,7 @@ static void __set_nat_cache_dirty(struct f2fs_nm_info *nm_i,
 
 	head = radix_tree_lookup(&nm_i->nat_set_root, set);
 	if (!head) {
-		head = f2fs_kmem_cache_alloc(nat_entry_set_slab, GFP_ATOMIC);
+		head = f2fs_kmem_cache_alloc(nat_entry_set_slab, GFP_NOFS);
 
 		INIT_LIST_HEAD(&head->entry_list);
 		INIT_LIST_HEAD(&head->set_list);
@@ -246,7 +246,7 @@ static struct nat_entry *grab_nat_entry(struct f2fs_nm_info *nm_i, nid_t nid)
 {
 	struct nat_entry *new;
 
-	new = f2fs_kmem_cache_alloc(nat_entry_slab, GFP_ATOMIC);
+	new = f2fs_kmem_cache_alloc(nat_entry_slab, GFP_NOFS);
 	f2fs_radix_tree_insert(&nm_i->nat_root, nid, new);
 	memset(new, 0, sizeof(struct nat_entry));
 	nat_set_nid(new, nid);
@@ -306,6 +306,10 @@ static void set_node_addr(struct f2fs_sb_info *sbi, struct node_info *ni,
 	if (nat_get_blkaddr(e) != NEW_ADDR && new_blkaddr == NULL_ADDR) {
 		unsigned char version = nat_get_version(e);
 		nat_set_version(e, inc_node_version(version));
+
+		/* in order to reuse the nid */
+		if (nm_i->next_scan_nid > ni->nid)
+			nm_i->next_scan_nid = ni->nid;
 	}
 
 	/* change address */
@@ -328,11 +332,11 @@ static void set_node_addr(struct f2fs_sb_info *sbi, struct node_info *ni,
 int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink)
 {
 	struct f2fs_nm_info *nm_i = NM_I(sbi);
+	int nr = nr_shrink;
 
-	if (available_free_memory(sbi, NAT_ENTRIES))
+	if (!down_write_trylock(&nm_i->nat_tree_lock))
 		return 0;
 
-	down_write(&nm_i->nat_tree_lock);
 	while (nr_shrink && !list_empty(&nm_i->nat_entries)) {
 		struct nat_entry *ne;
 		ne = list_first_entry(&nm_i->nat_entries,
@@ -341,7 +345,7 @@ int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink)
 		nr_shrink--;
 	}
 	up_write(&nm_i->nat_tree_lock);
-	return nr_shrink;
+	return nr - nr_shrink;
 }
 
 /*
@@ -898,17 +902,20 @@ int truncate_xattr_node(struct inode *inode, struct page *page)
  * Caller should grab and release a rwsem by calling f2fs_lock_op() and
  * f2fs_unlock_op().
  */
-void remove_inode_page(struct inode *inode)
+int remove_inode_page(struct inode *inode)
 {
 	struct dnode_of_data dn;
+	int err;
 
 	set_new_dnode(&dn, inode, NULL, NULL, inode->i_ino);
-	if (get_dnode_of_data(&dn, 0, LOOKUP_NODE))
+	err = get_dnode_of_data(&dn, 0, LOOKUP_NODE);
-		return;
+	if (err)
+		return err;
 
-	if (truncate_xattr_node(inode, dn.inode_page)) {
+	err = truncate_xattr_node(inode, dn.inode_page);
+	if (err) {
 		f2fs_put_dnode(&dn);
-		return;
+		return err;
 	}
 
 	/* remove potential inline_data blocks */
@@ -922,6 +929,7 @@ void remove_inode_page(struct inode *inode)
 
 	/* will put inode & node pages */
 	truncate_node(&dn);
+	return 0;
 }
 
 struct page *new_inode_page(struct inode *inode)
@@ -991,8 +999,7 @@ struct page *new_node_page(struct dnode_of_data *dn,
 /*
  * Caller should do after getting the following values.
  * 0: f2fs_put_page(page, 0)
- * LOCKED_PAGE: f2fs_put_page(page, 1)
+ * LOCKED_PAGE or error: f2fs_put_page(page, 1)
- * error: nothing
  */
 static int read_node_page(struct page *page, int rw)
 {
@@ -1010,7 +1017,6 @@ static int read_node_page(struct page *page, int rw)
 
 	if (unlikely(ni.blk_addr == NULL_ADDR)) {
 		ClearPageUptodate(page);
-		f2fs_put_page(page, 1);
 		return -ENOENT;
 	}
 
@@ -1041,10 +1047,7 @@ void ra_node_page(struct f2fs_sb_info *sbi, nid_t nid)
 		return;
 
 	err = read_node_page(apage, READA);
-	if (err == 0)
+	f2fs_put_page(apage, err ? 1 : 0);
-		f2fs_put_page(apage, 0);
-	else if (err == LOCKED_PAGE)
-		f2fs_put_page(apage, 1);
 }
 
 struct page *get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid)
@@ -1057,10 +1060,12 @@ struct page *get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid)
 		return ERR_PTR(-ENOMEM);
 
 	err = read_node_page(page, READ_SYNC);
-	if (err < 0)
+	if (err < 0) {
+		f2fs_put_page(page, 1);
 		return ERR_PTR(err);
-	else if (err != LOCKED_PAGE)
+	} else if (err != LOCKED_PAGE) {
 		lock_page(page);
+	}
 
 	if (unlikely(!PageUptodate(page) || nid != nid_of_node(page))) {
 		ClearPageUptodate(page);
@@ -1096,10 +1101,12 @@ struct page *get_node_page_ra(struct page *parent, int start)
 		return ERR_PTR(-ENOMEM);
 
 	err = read_node_page(page, READ_SYNC);
-	if (err < 0)
+	if (err < 0) {
+		f2fs_put_page(page, 1);
 		return ERR_PTR(err);
-	else if (err == LOCKED_PAGE)
+	} else if (err == LOCKED_PAGE) {
 		goto page_hit;
+	}
 
 	blk_start_plug(&plug);
 
@@ -1533,7 +1540,7 @@ static void build_free_nids(struct f2fs_sb_info *sbi)
 		if (unlikely(nid >= nm_i->max_nid))
 			nid = 0;
 
-		if (i++ == FREE_NID_PAGES)
+		if (++i >= FREE_NID_PAGES)
 			break;
 	}
 
@@ -1570,6 +1577,8 @@ bool alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid)
 
 	/* We should not use stale free nids created by build_free_nids */
 	if (nm_i->fcnt && !on_build_free_nids(nm_i)) {
+		struct node_info ni;
+
 		f2fs_bug_on(sbi, list_empty(&nm_i->free_nid_list));
 		list_for_each_entry(i, &nm_i->free_nid_list, list)
 			if (i->state == NID_NEW)
@@ -1580,6 +1589,13 @@ bool alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid)
 		i->state = NID_ALLOC;
 		nm_i->fcnt--;
 		spin_unlock(&nm_i->free_nid_list_lock);
+
+		/* check nid is allocated already */
+		get_node_info(sbi, *nid, &ni);
+		if (ni.blk_addr != NULL_ADDR) {
+			alloc_nid_done(sbi, *nid);
+			goto retry;
+		}
 		return true;
 	}
 	spin_unlock(&nm_i->free_nid_list_lock);
@@ -1636,6 +1652,32 @@ void alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid)
 		kmem_cache_free(free_nid_slab, i);
 }
 
+int try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink)
+{
+	struct f2fs_nm_info *nm_i = NM_I(sbi);
+	struct free_nid *i, *next;
+	int nr = nr_shrink;
+
+	if (!mutex_trylock(&nm_i->build_lock))
+		return 0;
+
+	spin_lock(&nm_i->free_nid_list_lock);
+	list_for_each_entry_safe(i, next, &nm_i->free_nid_list, list) {
+		if (nr_shrink <= 0 || nm_i->fcnt <= NAT_ENTRY_PER_BLOCK)
+			break;
+		if (i->state == NID_ALLOC)
+			continue;
+		__del_from_free_nid_list(nm_i, i);
+		kmem_cache_free(free_nid_slab, i);
+		nm_i->fcnt--;
+		nr_shrink--;
+	}
+	spin_unlock(&nm_i->free_nid_list_lock);
+	mutex_unlock(&nm_i->build_lock);
+
+	return nr - nr_shrink;
+}
+
 void recover_inline_xattr(struct inode *inode, struct page *page)
 {
 	void *src_addr, *dst_addr;

fs/f2fs/recovery.c

@@ -399,14 +399,35 @@ static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
 	f2fs_bug_on(sbi, ni.ino != ino_of_node(page));
 	f2fs_bug_on(sbi, ofs_of_node(dn.node_page) != ofs_of_node(page));
 
-	for (; start < end; start++) {
+	for (; start < end; start++, dn.ofs_in_node++) {
 		block_t src, dest;
 
 		src = datablock_addr(dn.node_page, dn.ofs_in_node);
 		dest = datablock_addr(page, dn.ofs_in_node);
 
-		if (src != dest && dest != NEW_ADDR && dest != NULL_ADDR &&
+		/* skip recovering if dest is the same as src */
-			is_valid_blkaddr(sbi, dest, META_POR)) {
+		if (src == dest)
+			continue;
+
+		/* dest is invalid, just invalidate src block */
+		if (dest == NULL_ADDR) {
+			truncate_data_blocks_range(&dn, 1);
+			continue;
+		}
+
+		/*
+		 * dest is reserved block, invalidate src block
+		 * and then reserve one new block in dnode page.
+		 */
+		if (dest == NEW_ADDR) {
+			truncate_data_blocks_range(&dn, 1);
+			err = reserve_new_block(&dn);
+			f2fs_bug_on(sbi, err);
+			continue;
+		}
+
+		/* dest is valid block, try to recover from src to dest */
+		if (is_valid_blkaddr(sbi, dest, META_POR)) {
 
 			if (src == NULL_ADDR) {
 				err = reserve_new_block(&dn);
@@ -424,7 +445,6 @@ static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
 							ni.version, false);
 			recovered++;
 		}
-		dn.ofs_in_node++;
 	}
 
 	if (IS_INODE(dn.node_page))
@@ -525,14 +545,12 @@ int recover_fsync_data(struct f2fs_sb_info *sbi)
 
 	INIT_LIST_HEAD(&inode_list);
 
-	/* step #1: find fsynced inode numbers */
-	set_sbi_flag(sbi, SBI_POR_DOING);
-
 	/* prevent checkpoint */
 	mutex_lock(&sbi->cp_mutex);
 
 	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
 
+	/* step #1: find fsynced inode numbers */
 	err = find_fsync_dnodes(sbi, &inode_list);
 	if (err)
 		goto out;
@@ -561,11 +579,20 @@ int recover_fsync_data(struct f2fs_sb_info *sbi)
 
 	clear_sbi_flag(sbi, SBI_POR_DOING);
 	if (err) {
-		discard_next_dnode(sbi, blkaddr);
+		bool invalidate = false;
+
+		if (discard_next_dnode(sbi, blkaddr))
+			invalidate = true;
 
 		/* Flush all the NAT/SIT pages */
 		while (get_pages(sbi, F2FS_DIRTY_META))
 			sync_meta_pages(sbi, META, LONG_MAX);
+
+		/* invalidate temporary meta page */
+		if (invalidate)
+			invalidate_mapping_pages(META_MAPPING(sbi),
+							blkaddr, blkaddr);
+
 		set_ckpt_flags(sbi->ckpt, CP_ERROR_FLAG);
 		mutex_unlock(&sbi->cp_mutex);
 	} else if (need_writecp) {

fs/f2fs/segment.c

@@ -197,28 +197,20 @@ void register_inmem_page(struct inode *inode, struct page *page)
 {
 	struct f2fs_inode_info *fi = F2FS_I(inode);
 	struct inmem_pages *new;
-	int err;
 
-	SetPagePrivate(page);
 	f2fs_trace_pid(page);
 
+	set_page_private(page, (unsigned long)ATOMIC_WRITTEN_PAGE);
+	SetPagePrivate(page);
+
 	new = f2fs_kmem_cache_alloc(inmem_entry_slab, GFP_NOFS);
 
 	/* add atomic page indices to the list */
 	new->page = page;
 	INIT_LIST_HEAD(&new->list);
-retry:
+
 	/* increase reference count with clean state */
 	mutex_lock(&fi->inmem_lock);
-	err = radix_tree_insert(&fi->inmem_root, page->index, new);
-	if (err == -EEXIST) {
-		mutex_unlock(&fi->inmem_lock);
-		kmem_cache_free(inmem_entry_slab, new);
-		return;
-	} else if (err) {
-		mutex_unlock(&fi->inmem_lock);
-		goto retry;
-	}
 	get_page(page);
 	list_add_tail(&new->list, &fi->inmem_pages);
 	inc_page_count(F2FS_I_SB(inode), F2FS_INMEM_PAGES);
@@ -227,7 +219,7 @@ void register_inmem_page(struct inode *inode, struct page *page)
 	trace_f2fs_register_inmem_page(page, INMEM);
 }
 
-void commit_inmem_pages(struct inode *inode, bool abort)
+int commit_inmem_pages(struct inode *inode, bool abort)
 {
 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
 	struct f2fs_inode_info *fi = F2FS_I(inode);
@@ -239,6 +231,7 @@ void commit_inmem_pages(struct inode *inode, bool abort)
 		.rw = WRITE_SYNC | REQ_PRIO,
 		.encrypted_page = NULL,
 	};
+	int err = 0;
 
 	/*
 	 * The abort is true only when f2fs_evict_inode is called.
@@ -254,8 +247,8 @@ void commit_inmem_pages(struct inode *inode, bool abort)
 
 	mutex_lock(&fi->inmem_lock);
 	list_for_each_entry_safe(cur, tmp, &fi->inmem_pages, list) {
+		lock_page(cur->page);
 		if (!abort) {
-			lock_page(cur->page);
 			if (cur->page->mapping == inode->i_mapping) {
 				set_page_dirty(cur->page);
 				f2fs_wait_on_page_writeback(cur->page, DATA);
@@ -263,15 +256,20 @@ void commit_inmem_pages(struct inode *inode, bool abort)
 					inode_dec_dirty_pages(inode);
 				trace_f2fs_commit_inmem_page(cur->page, INMEM);
 				fio.page = cur->page;
-				do_write_data_page(&fio);
+				err = do_write_data_page(&fio);
 				submit_bio = true;
+				if (err) {
+					unlock_page(cur->page);
+					break;
+				}
 			}
-			f2fs_put_page(cur->page, 1);
 		} else {
 			trace_f2fs_commit_inmem_page(cur->page, INMEM_DROP);
-			put_page(cur->page);
 		}
-		radix_tree_delete(&fi->inmem_root, cur->page->index);
+		set_page_private(cur->page, 0);
+		ClearPagePrivate(cur->page);
+		f2fs_put_page(cur->page, 1);
+
 		list_del(&cur->list);
 		kmem_cache_free(inmem_entry_slab, cur);
 		dec_page_count(F2FS_I_SB(inode), F2FS_INMEM_PAGES);
@@ -283,6 +281,7 @@ void commit_inmem_pages(struct inode *inode, bool abort)
 		if (submit_bio)
 			f2fs_submit_merged_bio(sbi, DATA, WRITE);
 	}
+	return err;
 }
 
 /*
@@ -304,10 +303,18 @@ void f2fs_balance_fs(struct f2fs_sb_info *sbi)
 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi)
 {
 	/* try to shrink extent cache when there is no enough memory */
-	f2fs_shrink_extent_tree(sbi, EXTENT_CACHE_SHRINK_NUMBER);
+	if (!available_free_memory(sbi, EXTENT_CACHE))
+		f2fs_shrink_extent_tree(sbi, EXTENT_CACHE_SHRINK_NUMBER);
 
-	/* check the # of cached NAT entries and prefree segments */
+	/* check the # of cached NAT entries */
-	if (try_to_free_nats(sbi, NAT_ENTRY_PER_BLOCK) ||
+	if (!available_free_memory(sbi, NAT_ENTRIES))
+		try_to_free_nats(sbi, NAT_ENTRY_PER_BLOCK);
+
+	if (!available_free_memory(sbi, FREE_NIDS))
+		try_to_free_nids(sbi, NAT_ENTRY_PER_BLOCK * FREE_NID_PAGES);
+
+	/* checkpoint is the only way to shrink partial cached entries */
+	if (!available_free_memory(sbi, NAT_ENTRIES) ||
 			excess_prefree_segs(sbi) ||
 			!available_free_memory(sbi, INO_ENTRIES))
 		f2fs_sync_fs(sbi->sb, true);
@@ -323,10 +330,12 @@ static int issue_flush_thread(void *data)
 		return 0;
 
 	if (!llist_empty(&fcc->issue_list)) {
-		struct bio *bio = bio_alloc(GFP_NOIO, 0);
+		struct bio *bio;
 		struct flush_cmd *cmd, *next;
 		int ret;
 
+		bio = f2fs_bio_alloc(0);
+
 		fcc->dispatch_list = llist_del_all(&fcc->issue_list);
 		fcc->dispatch_list = llist_reverse_order(fcc->dispatch_list);
 
@@ -358,8 +367,15 @@ int f2fs_issue_flush(struct f2fs_sb_info *sbi)
 	if (test_opt(sbi, NOBARRIER))
 		return 0;
 
-	if (!test_opt(sbi, FLUSH_MERGE))
+	if (!test_opt(sbi, FLUSH_MERGE)) {
-		return blkdev_issue_flush(sbi->sb->s_bdev, GFP_KERNEL, NULL);
+		struct bio *bio = f2fs_bio_alloc(0);
+		int ret;
+
+		bio->bi_bdev = sbi->sb->s_bdev;
+		ret = submit_bio_wait(WRITE_FLUSH, bio);
+		bio_put(bio);
+		return ret;
+	}
 
 	init_completion(&cmd.wait);
 
@@ -503,7 +519,7 @@ static int f2fs_issue_discard(struct f2fs_sb_info *sbi,
 	return blkdev_issue_discard(sbi->sb->s_bdev, start, len, GFP_NOFS, 0);
 }
 
-void discard_next_dnode(struct f2fs_sb_info *sbi, block_t blkaddr)
+bool discard_next_dnode(struct f2fs_sb_info *sbi, block_t blkaddr)
 {
 	int err = -ENOTSUPP;
 
@@ -513,13 +529,16 @@ void discard_next_dnode(struct f2fs_sb_info *sbi, block_t blkaddr)
 		unsigned int offset = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
 
 		if (f2fs_test_bit(offset, se->discard_map))
-			return;
+			return false;
 
 		err = f2fs_issue_discard(sbi, blkaddr, 1);
 	}
 
-	if (err)
+	if (err) {
 		update_meta_page(sbi, NULL, blkaddr);
+		return true;
+	}
+	return false;
 }
 
 static void __add_discard_entry(struct f2fs_sb_info *sbi,
@@ -1218,7 +1237,8 @@ void allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
 	mutex_lock(&sit_i->sentry_lock);
 
 	/* direct_io'ed data is aligned to the segment for better performance */
-	if (direct_io && curseg->next_blkoff)
+	if (direct_io && curseg->next_blkoff &&
+				!has_not_enough_free_secs(sbi, 0))
 		__allocate_new_segments(sbi, type);
 
 	*new_blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
@@ -1733,7 +1753,7 @@ static struct page *get_next_sit_page(struct f2fs_sb_info *sbi,
 static struct sit_entry_set *grab_sit_entry_set(void)
 {
 	struct sit_entry_set *ses =
-			f2fs_kmem_cache_alloc(sit_entry_set_slab, GFP_ATOMIC);
+			f2fs_kmem_cache_alloc(sit_entry_set_slab, GFP_NOFS);
 
 	ses->entry_cnt = 0;
 	INIT_LIST_HEAD(&ses->set_list);

fs/f2fs/segment.h

@@ -177,6 +177,15 @@ struct segment_allocation {
 	void (*allocate_segment)(struct f2fs_sb_info *, int, bool);
 };
 
+/*
+ * this value is set in page as a private data which indicate that
+ * the page is atomically written, and it is in inmem_pages list.
+ */
+#define ATOMIC_WRITTEN_PAGE		0x0000ffff
+
+#define IS_ATOMIC_WRITTEN_PAGE(page)			\
+		(page_private(page) == (unsigned long)ATOMIC_WRITTEN_PAGE)
+
 struct inmem_pages {
 	struct list_head list;
 	struct page *page;
@@ -555,16 +564,15 @@ static inline unsigned short curseg_blkoff(struct f2fs_sb_info *sbi, int type)
 	return curseg->next_blkoff;
 }
 
-#ifdef CONFIG_F2FS_CHECK_FS
 static inline void check_seg_range(struct f2fs_sb_info *sbi, unsigned int segno)
 {
-	BUG_ON(segno > TOTAL_SEGS(sbi) - 1);
+	f2fs_bug_on(sbi, segno > TOTAL_SEGS(sbi) - 1);
 }
 
 static inline void verify_block_addr(struct f2fs_sb_info *sbi, block_t blk_addr)
 {
-	BUG_ON(blk_addr < SEG0_BLKADDR(sbi));
+	f2fs_bug_on(sbi, blk_addr < SEG0_BLKADDR(sbi)
-	BUG_ON(blk_addr >= MAX_BLKADDR(sbi));
+					|| blk_addr >= MAX_BLKADDR(sbi));
 }
 
 /*
@@ -573,16 +581,11 @@ static inline void verify_block_addr(struct f2fs_sb_info *sbi, block_t blk_addr)
 static inline void check_block_count(struct f2fs_sb_info *sbi,
 		int segno, struct f2fs_sit_entry *raw_sit)
 {
+#ifdef CONFIG_F2FS_CHECK_FS
 	bool is_valid  = test_bit_le(0, raw_sit->valid_map) ? true : false;
 	int valid_blocks = 0;
 	int cur_pos = 0, next_pos;
 
-	/* check segment usage */
-	BUG_ON(GET_SIT_VBLOCKS(raw_sit) > sbi->blocks_per_seg);
-
-	/* check boundary of a given segment number */
-	BUG_ON(segno > TOTAL_SEGS(sbi) - 1);
-
 	/* check bitmap with valid block count */
 	do {
 		if (is_valid) {
@@ -598,35 +601,11 @@ static inline void check_block_count(struct f2fs_sb_info *sbi,
 		is_valid = !is_valid;
 	} while (cur_pos < sbi->blocks_per_seg);
 	BUG_ON(GET_SIT_VBLOCKS(raw_sit) != valid_blocks);
-}
-#else
-static inline void check_seg_range(struct f2fs_sb_info *sbi, unsigned int segno)
-{
-	if (segno > TOTAL_SEGS(sbi) - 1)
-		set_sbi_flag(sbi, SBI_NEED_FSCK);
-}
-
-static inline void verify_block_addr(struct f2fs_sb_info *sbi, block_t blk_addr)
-{
-	if (blk_addr < SEG0_BLKADDR(sbi) || blk_addr >= MAX_BLKADDR(sbi))
-		set_sbi_flag(sbi, SBI_NEED_FSCK);
-}
-
-/*
- * Summary block is always treated as an invalid block
- */
-static inline void check_block_count(struct f2fs_sb_info *sbi,
-		int segno, struct f2fs_sit_entry *raw_sit)
-{
-	/* check segment usage */
-	if (GET_SIT_VBLOCKS(raw_sit) > sbi->blocks_per_seg)
-		set_sbi_flag(sbi, SBI_NEED_FSCK);
-
-	/* check boundary of a given segment number */
-	if (segno > TOTAL_SEGS(sbi) - 1)
-		set_sbi_flag(sbi, SBI_NEED_FSCK);
-}
 #endif
+	/* check segment usage, and check boundary of a given segment number */
+	f2fs_bug_on(sbi, GET_SIT_VBLOCKS(raw_sit) > sbi->blocks_per_seg
+					|| segno > TOTAL_SEGS(sbi) - 1);
+}
 
 static inline pgoff_t current_sit_addr(struct f2fs_sb_info *sbi,
 						unsigned int start)

fs/f2fs/shrinker.c

@@ -0,0 +1,139 @@
+/*
+ * f2fs shrinker support
+ *   the basic infra was copied from fs/ubifs/shrinker.c
+ *
+ * Copyright (c) 2015 Motorola Mobility
+ * Copyright (c) 2015 Jaegeuk Kim <jaegeuk@kernel.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+#include <linux/fs.h>
+#include <linux/f2fs_fs.h>
+
+#include "f2fs.h"
+
+static LIST_HEAD(f2fs_list);
+static DEFINE_SPINLOCK(f2fs_list_lock);
+static unsigned int shrinker_run_no;
+
+static unsigned long __count_nat_entries(struct f2fs_sb_info *sbi)
+{
+	return NM_I(sbi)->nat_cnt - NM_I(sbi)->dirty_nat_cnt;
+}
+
+static unsigned long __count_free_nids(struct f2fs_sb_info *sbi)
+{
+	if (NM_I(sbi)->fcnt > NAT_ENTRY_PER_BLOCK)
+		return NM_I(sbi)->fcnt - NAT_ENTRY_PER_BLOCK;
+	return 0;
+}
+
+static unsigned long __count_extent_cache(struct f2fs_sb_info *sbi)
+{
+	return sbi->total_ext_tree + atomic_read(&sbi->total_ext_node);
+}
+
+unsigned long f2fs_shrink_count(struct shrinker *shrink,
+				struct shrink_control *sc)
+{
+	struct f2fs_sb_info *sbi;
+	struct list_head *p;
+	unsigned long count = 0;
+
+	spin_lock(&f2fs_list_lock);
+	p = f2fs_list.next;
+	while (p != &f2fs_list) {
+		sbi = list_entry(p, struct f2fs_sb_info, s_list);
+
+		/* stop f2fs_put_super */
+		if (!mutex_trylock(&sbi->umount_mutex)) {
+			p = p->next;
+			continue;
+		}
+		spin_unlock(&f2fs_list_lock);
+
+		/* count extent cache entries */
+		count += __count_extent_cache(sbi);
+
+		/* shrink clean nat cache entries */
+		count += __count_nat_entries(sbi);
+
+		/* count free nids cache entries */
+		count += __count_free_nids(sbi);
+
+		spin_lock(&f2fs_list_lock);
+		p = p->next;
+		mutex_unlock(&sbi->umount_mutex);
+	}
+	spin_unlock(&f2fs_list_lock);
+	return count;
+}
+
+unsigned long f2fs_shrink_scan(struct shrinker *shrink,
+				struct shrink_control *sc)
+{
+	unsigned long nr = sc->nr_to_scan;
+	struct f2fs_sb_info *sbi;
+	struct list_head *p;
+	unsigned int run_no;
+	unsigned long freed = 0;
+
+	spin_lock(&f2fs_list_lock);
+	do {
+		run_no = ++shrinker_run_no;
+	} while (run_no == 0);
+	p = f2fs_list.next;
+	while (p != &f2fs_list) {
+		sbi = list_entry(p, struct f2fs_sb_info, s_list);
+
+		if (sbi->shrinker_run_no == run_no)
+			break;
+
+		/* stop f2fs_put_super */
+		if (!mutex_trylock(&sbi->umount_mutex)) {
+			p = p->next;
+			continue;
+		}
+		spin_unlock(&f2fs_list_lock);
+
+		sbi->shrinker_run_no = run_no;
+
+		/* shrink extent cache entries */
+		freed += f2fs_shrink_extent_tree(sbi, nr >> 1);
+
+		/* shrink clean nat cache entries */
+		if (freed < nr)
+			freed += try_to_free_nats(sbi, nr - freed);
+
+		/* shrink free nids cache entries */
+		if (freed < nr)
+			freed += try_to_free_nids(sbi, nr - freed);
+
+		spin_lock(&f2fs_list_lock);
+		p = p->next;
+		list_move_tail(&sbi->s_list, &f2fs_list);
+		mutex_unlock(&sbi->umount_mutex);
+		if (freed >= nr)
+			break;
+	}
+	spin_unlock(&f2fs_list_lock);
+	return freed;
+}
+
+void f2fs_join_shrinker(struct f2fs_sb_info *sbi)
+{
+	spin_lock(&f2fs_list_lock);
+	list_add_tail(&sbi->s_list, &f2fs_list);
+	spin_unlock(&f2fs_list_lock);
+}
+
+void f2fs_leave_shrinker(struct f2fs_sb_info *sbi)
+{
+	f2fs_shrink_extent_tree(sbi, __count_extent_cache(sbi));
+
+	spin_lock(&f2fs_list_lock);
+	list_del(&sbi->s_list);
+	spin_unlock(&f2fs_list_lock);
+}

fs/f2fs/super.c

@@ -39,6 +39,13 @@ static struct proc_dir_entry *f2fs_proc_root;
 static struct kmem_cache *f2fs_inode_cachep;
 static struct kset *f2fs_kset;
 
+/* f2fs-wide shrinker description */
+static struct shrinker f2fs_shrinker_info = {
+	.scan_objects = f2fs_shrink_scan,
+	.count_objects = f2fs_shrink_count,
+	.seeks = DEFAULT_SEEKS,
+};
+
 enum {
 	Opt_gc_background,
 	Opt_disable_roll_forward,
@@ -58,6 +65,7 @@ enum {
 	Opt_nobarrier,
 	Opt_fastboot,
 	Opt_extent_cache,
+	Opt_noextent_cache,
 	Opt_noinline_data,
 	Opt_err,
 };
@@ -81,6 +89,7 @@ static match_table_t f2fs_tokens = {
 	{Opt_nobarrier, "nobarrier"},
 	{Opt_fastboot, "fastboot"},
 	{Opt_extent_cache, "extent_cache"},
+	{Opt_noextent_cache, "noextent_cache"},
 	{Opt_noinline_data, "noinline_data"},
 	{Opt_err, NULL},
 };
@@ -382,6 +391,9 @@ static int parse_options(struct super_block *sb, char *options)
 		case Opt_extent_cache:
 			set_opt(sbi, EXTENT_CACHE);
 			break;
+		case Opt_noextent_cache:
+			clear_opt(sbi, EXTENT_CACHE);
+			break;
 		case Opt_noinline_data:
 			clear_opt(sbi, INLINE_DATA);
 			break;
@@ -410,9 +422,7 @@ static struct inode *f2fs_alloc_inode(struct super_block *sb)
 	atomic_set(&fi->dirty_pages, 0);
 	fi->i_current_depth = 1;
 	fi->i_advise = 0;
-	rwlock_init(&fi->ext_lock);
 	init_rwsem(&fi->i_sem);
-	INIT_RADIX_TREE(&fi->inmem_root, GFP_NOFS);
 	INIT_LIST_HEAD(&fi->inmem_pages);
 	mutex_init(&fi->inmem_lock);
 
@@ -441,17 +451,22 @@ static int f2fs_drop_inode(struct inode *inode)
 	 */
 	if (!inode_unhashed(inode) && inode->i_state & I_SYNC) {
 		if (!inode->i_nlink && !is_bad_inode(inode)) {
+			/* to avoid evict_inode call simultaneously */
+			atomic_inc(&inode->i_count);
 			spin_unlock(&inode->i_lock);
 
 			/* some remained atomic pages should discarded */
 			if (f2fs_is_atomic_file(inode))
 				commit_inmem_pages(inode, true);
 
+			/* should remain fi->extent_tree for writepage */
+			f2fs_destroy_extent_node(inode);
+
 			sb_start_intwrite(inode->i_sb);
 			i_size_write(inode, 0);
 
 			if (F2FS_HAS_BLOCKS(inode))
-				f2fs_truncate(inode);
+				f2fs_truncate(inode, true);
 
 			sb_end_intwrite(inode->i_sb);
 
@@ -461,6 +476,7 @@ static int f2fs_drop_inode(struct inode *inode)
 					F2FS_I(inode)->i_crypt_info);
 #endif
 			spin_lock(&inode->i_lock);
+			atomic_dec(&inode->i_count);
 		}
 		return 0;
 	}
@@ -498,9 +514,11 @@ static void f2fs_put_super(struct super_block *sb)
 	}
 	kobject_del(&sbi->s_kobj);
 
-	f2fs_destroy_stats(sbi);
 	stop_gc_thread(sbi);
 
+	/* prevent remaining shrinker jobs */
+	mutex_lock(&sbi->umount_mutex);
+
 	/*
 	 * We don't need to do checkpoint when superblock is clean.
 	 * But, the previous checkpoint was not done by umount, it needs to do
@@ -514,6 +532,9 @@ static void f2fs_put_super(struct super_block *sb)
 		write_checkpoint(sbi, &cpc);
 	}
 
+	/* write_checkpoint can update stat informaion */
+	f2fs_destroy_stats(sbi);
+
 	/*
 	 * normally superblock is clean, so we need to release this.
 	 * In addition, EIO will skip do checkpoint, we need this as well.
@@ -521,6 +542,9 @@ static void f2fs_put_super(struct super_block *sb)
 	release_dirty_inode(sbi);
 	release_discard_addrs(sbi);
 
+	f2fs_leave_shrinker(sbi);
+	mutex_unlock(&sbi->umount_mutex);
+
 	iput(sbi->node_inode);
 	iput(sbi->meta_inode);
 
@@ -647,6 +671,8 @@ static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
 		seq_puts(seq, ",fastboot");
 	if (test_opt(sbi, EXTENT_CACHE))
 		seq_puts(seq, ",extent_cache");
+	else
+		seq_puts(seq, ",noextent_cache");
 	seq_printf(seq, ",active_logs=%u", sbi->active_logs);
 
 	return 0;
@@ -667,7 +693,7 @@ static int segment_info_seq_show(struct seq_file *seq, void *offset)
 		struct seg_entry *se = get_seg_entry(sbi, i);
 
 		if ((i % 10) == 0)
-			seq_printf(seq, "%-5d", i);
+			seq_printf(seq, "%-10d", i);
 		seq_printf(seq, "%d|%-3u", se->type,
 					get_valid_blocks(sbi, i, 1));
 		if ((i % 10) == 9 || i == (total_segs - 1))
@@ -699,6 +725,7 @@ static void default_options(struct f2fs_sb_info *sbi)
 
 	set_opt(sbi, BG_GC);
 	set_opt(sbi, INLINE_DATA);
+	set_opt(sbi, EXTENT_CACHE);
 
 #ifdef CONFIG_F2FS_FS_XATTR
 	set_opt(sbi, XATTR_USER);
@@ -970,6 +997,9 @@ static void init_sb_info(struct f2fs_sb_info *sbi)
 
 	sbi->dir_level = DEF_DIR_LEVEL;
 	clear_sbi_flag(sbi, SBI_NEED_FSCK);
+
+	INIT_LIST_HEAD(&sbi->s_list);
+	mutex_init(&sbi->umount_mutex);
 }
 
 /*
@@ -1135,7 +1165,9 @@ static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
 	mutex_init(&sbi->writepages);
 	mutex_init(&sbi->cp_mutex);
 	init_rwsem(&sbi->node_write);
-	clear_sbi_flag(sbi, SBI_POR_DOING);
+
+	/* disallow all the data/node/meta page writes */
+	set_sbi_flag(sbi, SBI_POR_DOING);
 	spin_lock_init(&sbi->stat_lock);
 
 	init_rwsem(&sbi->read_io.io_rwsem);
@@ -1212,8 +1244,12 @@ static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
 		goto free_nm;
 	}
 
+	f2fs_join_shrinker(sbi);
+
 	/* if there are nt orphan nodes free them */
-	recover_orphan_inodes(sbi);
+	err = recover_orphan_inodes(sbi);
+	if (err)
+		goto free_node_inode;
 
 	/* read root inode and dentry */
 	root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
@@ -1275,6 +1311,8 @@ static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
 			goto free_kobj;
 		}
 	}
+	/* recover_fsync_data() cleared this already */
+	clear_sbi_flag(sbi, SBI_POR_DOING);
 
 	/*
 	 * If filesystem is not mounted as read-only then
@@ -1308,7 +1346,10 @@ static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
 	dput(sb->s_root);
 	sb->s_root = NULL;
 free_node_inode:
+	mutex_lock(&sbi->umount_mutex);
+	f2fs_leave_shrinker(sbi);
 	iput(sbi->node_inode);
+	mutex_unlock(&sbi->umount_mutex);
 free_nm:
 	destroy_node_manager(sbi);
 free_sm:
@@ -1404,13 +1445,20 @@ static int __init init_f2fs_fs(void)
 	err = f2fs_init_crypto();
 	if (err)
 		goto free_kset;
-	err = register_filesystem(&f2fs_fs_type);
+
+	err = register_shrinker(&f2fs_shrinker_info);
 	if (err)
 		goto free_crypto;
+
+	err = register_filesystem(&f2fs_fs_type);
+	if (err)
+		goto free_shrinker;
 	f2fs_create_root_stats();
 	f2fs_proc_root = proc_mkdir("fs/f2fs", NULL);
 	return 0;
 
+free_shrinker:
+	unregister_shrinker(&f2fs_shrinker_info);
 free_crypto:
 	f2fs_exit_crypto();
 free_kset:
@@ -1433,6 +1481,7 @@ static void __exit exit_f2fs_fs(void)
 {
 	remove_proc_entry("fs/f2fs", NULL);
 	f2fs_destroy_root_stats();
+	unregister_shrinker(&f2fs_shrinker_info);
 	unregister_filesystem(&f2fs_fs_type);
 	f2fs_exit_crypto();
 	destroy_extent_cache();

fs/f2fs/xattr.c

@@ -499,9 +499,12 @@ static int __f2fs_setxattr(struct inode *inode, int index,
 
 	len = strlen(name);
 
-	if (len > F2FS_NAME_LEN || size > MAX_VALUE_LEN(inode))
+	if (len > F2FS_NAME_LEN)
 		return -ERANGE;
 
+	if (size > MAX_VALUE_LEN(inode))
+		return -E2BIG;
+
 	base_addr = read_all_xattrs(inode, ipage);
 	if (!base_addr)
 		goto exit;

include/linux/f2fs_fs.h

@@ -417,15 +417,25 @@ typedef __le32	f2fs_hash_t;
 
 #define GET_DENTRY_SLOTS(x)	((x + F2FS_SLOT_LEN - 1) >> F2FS_SLOT_LEN_BITS)
 
-/* the number of dentry in a block */
-#define NR_DENTRY_IN_BLOCK	214
-
 /* MAX level for dir lookup */
 #define MAX_DIR_HASH_DEPTH	63
 
 /* MAX buckets in one level of dir */
 #define MAX_DIR_BUCKETS		(1 << ((MAX_DIR_HASH_DEPTH / 2) - 1))
 
+/*
+ * space utilization of regular dentry and inline dentry
+ *		regular dentry			inline dentry
+ * bitmap	1 * 27 = 27			1 * 23 = 23
+ * reserved	1 * 3 = 3			1 * 7 = 7
+ * dentry	11 * 214 = 2354			11 * 182 = 2002
+ * filename	8 * 214 = 1712			8 * 182 = 1456
+ * total	4096				3488
+ *
+ * Note: there are more reserved space in inline dentry than in regular
+ * dentry, when converting inline dentry we should handle this carefully.
+ */
+#define NR_DENTRY_IN_BLOCK	214	/* the number of dentry in a block */
 #define SIZE_OF_DIR_ENTRY	11	/* by byte */
 #define SIZE_OF_DENTRY_BITMAP	((NR_DENTRY_IN_BLOCK + BITS_PER_BYTE - 1) / \
 					BITS_PER_BYTE)

include/trace/events/f2fs.h

@@ -1099,11 +1099,11 @@ TRACE_EVENT(f2fs_lookup_extent_tree_start,
 TRACE_EVENT_CONDITION(f2fs_lookup_extent_tree_end,
 
 	TP_PROTO(struct inode *inode, unsigned int pgofs,
-						struct extent_node *en),
+						struct extent_info *ei),
 
-	TP_ARGS(inode, pgofs, en),
+	TP_ARGS(inode, pgofs, ei),
 
-	TP_CONDITION(en),
+	TP_CONDITION(ei),
 
 	TP_STRUCT__entry(
 		__field(dev_t,	dev)
@@ -1118,9 +1118,9 @@ TRACE_EVENT_CONDITION(f2fs_lookup_extent_tree_end,
 		__entry->dev = inode->i_sb->s_dev;
 		__entry->ino = inode->i_ino;
 		__entry->pgofs = pgofs;
-		__entry->fofs = en->ei.fofs;
+		__entry->fofs = ei->fofs;
-		__entry->blk = en->ei.blk;
+		__entry->blk = ei->blk;
-		__entry->len = en->ei.len;
+		__entry->len = ei->len;
 	),
 
 	TP_printk("dev = (%d,%d), ino = %lu, pgofs = %u, "