Merge tag 'for-f2fs-4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs

Pull f2fs updates from Jaegeuk Kim:
 "Most part of the patches include enhancing the stability and
  performance of in-memory extent caches feature.

  In addition, it introduces several new features and configurable
  points:
   - F2FS_GOING_DOWN_METAFLUSH ioctl to test power failures
   - F2FS_IOC_WRITE_CHECKPOINT ioctl to trigger checkpoint by users
   - background_gc=sync mount option to do gc synchronously
   - periodic checkpoints
   - sysfs entry to control readahead blocks for free nids

  And the following bug fixes have been merged.
   - fix SSA corruption by collapse/insert_range
   - correct a couple of gc behaviors
   - fix the results of f2fs_map_blocks
   - fix error case handling of volatile/atomic writes"

* tag 'for-f2fs-4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs: (54 commits)
  f2fs: fix to skip shrinking extent nodes
  f2fs: fix error path of ->symlink
  f2fs: fix to clear GCed flag for atomic written page
  f2fs: don't need to submit bio on error case
  f2fs: fix leakage of inmemory atomic pages
  f2fs: refactor __find_rev_next_{zero}_bit
  f2fs: support fiemap for inline_data
  f2fs: flush dirty data for bmap
  f2fs: relocate the tracepoint for background_gc
  f2fs crypto: fix racing of accessing encrypted page among
  f2fs: export ra_nid_pages to sysfs
  f2fs: readahead for free nids building
  f2fs: support lower priority asynchronous readahead in ra_meta_pages
  f2fs: don't tag REQ_META for temporary non-meta pages
  f2fs: add a tracepoint for f2fs_read_data_pages
  f2fs: set GFP_NOFS for grab_cache_page
  f2fs: fix SSA updates resulting in corruption
  Revert "f2fs: do not skip dentry block writes"
  f2fs: add F2FS_GOING_DOWN_METAFLUSH to test power-failure
  f2fs: merge meta writes as many possible
  ...

Documentation/ABI/testing/sysfs-fs-f2fs

@@ -80,3 +80,15 @@ Date:		February 2015
 Contact:	"Jaegeuk Kim" <jaegeuk@kernel.org>
 Description:
 		 Controls the trimming rate in batch mode.
+
+What:		/sys/fs/f2fs/<disk>/cp_interval
+Date:		October 2015
+Contact:	"Jaegeuk Kim" <jaegeuk@kernel.org>
+Description:
+		 Controls the checkpoint timing.
+
+What:		/sys/fs/f2fs/<disk>/ra_nid_pages
+Date:		October 2015
+Contact:	"Chao Yu" <chao2.yu@samsung.com>
+Description:
+		 Controls the count of nid pages to be readaheaded.

Documentation/filesystems/f2fs.txt

@@ -102,7 +102,8 @@ background_gc=%s       Turn on/off cleaning operations, namely garbage
                        collection, triggered in background when I/O subsystem is
                        idle. If background_gc=on, it will turn on the garbage
                        collection and if background_gc=off, garbage collection
-                       will be truned off.
+                       will be truned off. If background_gc=sync, it will turn
+                       on synchronous garbage collection running in background.
                        Default value for this option is on. So garbage
                        collection is on by default.
 disable_roll_forward   Disable the roll-forward recovery routine

fs/f2fs/checkpoint.c

@@ -47,7 +47,8 @@ struct page *grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index)
 /*
  * We guarantee no failure on the returned page.
  */
-struct page *get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index)
+static struct page *__get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index,
+							bool is_meta)
 {
 	struct address_space *mapping = META_MAPPING(sbi);
 	struct page *page;
@@ -58,6 +59,9 @@ struct page *get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index)
 		.blk_addr = index,
 		.encrypted_page = NULL,
 	};
+
+	if (unlikely(!is_meta))
+		fio.rw &= ~REQ_META;
 repeat:
 	page = grab_cache_page(mapping, index);
 	if (!page) {
@@ -91,6 +95,17 @@ struct page *get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index)
 	return page;
 }
 
+struct page *get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index)
+{
+	return __get_meta_page(sbi, index, true);
+}
+
+/* for POR only */
+struct page *get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index)
+{
+	return __get_meta_page(sbi, index, false);
+}
+
 bool is_valid_blkaddr(struct f2fs_sb_info *sbi, block_t blkaddr, int type)
 {
 	switch (type) {
@@ -125,7 +140,8 @@ bool is_valid_blkaddr(struct f2fs_sb_info *sbi, block_t blkaddr, int type)
 /*
  * Readahead CP/NAT/SIT/SSA pages
  */
-int ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, int type)
+int ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
+							int type, bool sync)
 {
 	block_t prev_blk_addr = 0;
 	struct page *page;
@@ -133,10 +149,13 @@ int ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, int type
 	struct f2fs_io_info fio = {
 		.sbi = sbi,
 		.type = META,
-		.rw = READ_SYNC | REQ_META | REQ_PRIO,
+		.rw = sync ? (READ_SYNC | REQ_META | REQ_PRIO) : READA,
 		.encrypted_page = NULL,
 	};
 
+	if (unlikely(type == META_POR))
+		fio.rw &= ~REQ_META;
+
 	for (; nrpages-- > 0; blkno++) {
 
 		if (!is_valid_blkaddr(sbi, blkno, type))
@@ -196,7 +215,7 @@ void ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index)
 	f2fs_put_page(page, 0);
 
 	if (readahead)
-		ra_meta_pages(sbi, index, MAX_BIO_BLOCKS(sbi), META_POR);
+		ra_meta_pages(sbi, index, MAX_BIO_BLOCKS(sbi), META_POR, true);
 }
 
 static int f2fs_write_meta_page(struct page *page,
@@ -257,7 +276,7 @@ long sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
 						long nr_to_write)
 {
 	struct address_space *mapping = META_MAPPING(sbi);
-	pgoff_t index = 0, end = LONG_MAX;
+	pgoff_t index = 0, end = LONG_MAX, prev = LONG_MAX;
 	struct pagevec pvec;
 	long nwritten = 0;
 	struct writeback_control wbc = {
@@ -277,6 +296,13 @@ long sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
 		for (i = 0; i < nr_pages; i++) {
 			struct page *page = pvec.pages[i];
 
+			if (prev == LONG_MAX)
+				prev = page->index - 1;
+			if (nr_to_write != LONG_MAX && page->index != prev + 1) {
+				pagevec_release(&pvec);
+				goto stop;
+			}
+
 			lock_page(page);
 
 			if (unlikely(page->mapping != mapping)) {
@@ -297,13 +323,14 @@ long sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
 				break;
 			}
 			nwritten++;
+			prev = page->index;
 			if (unlikely(nwritten >= nr_to_write))
 				break;
 		}
 		pagevec_release(&pvec);
 		cond_resched();
 	}
-
+stop:
 	if (nwritten)
 		f2fs_submit_merged_bio(sbi, type, WRITE);
 
@@ -495,7 +522,7 @@ int recover_orphan_inodes(struct f2fs_sb_info *sbi)
 	start_blk = __start_cp_addr(sbi) + 1 + __cp_payload(sbi);
 	orphan_blocks = __start_sum_addr(sbi) - 1 - __cp_payload(sbi);
 
-	ra_meta_pages(sbi, start_blk, orphan_blocks, META_CP);
+	ra_meta_pages(sbi, start_blk, orphan_blocks, META_CP, true);
 
 	for (i = 0; i < orphan_blocks; i++) {
 		struct page *page = get_meta_page(sbi, start_blk + i);
@@ -1000,6 +1027,11 @@ static void do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
 
 	start_blk = __start_cp_addr(sbi);
 
+	/* need to wait for end_io results */
+	wait_on_all_pages_writeback(sbi);
+	if (unlikely(f2fs_cp_error(sbi)))
+		return;
+
 	/* write out checkpoint buffer at block 0 */
 	update_meta_page(sbi, ckpt, start_blk++);
 
@@ -1109,6 +1141,9 @@ void write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
 	if (cpc->reason == CP_RECOVERY)
 		f2fs_msg(sbi->sb, KERN_NOTICE,
 			"checkpoint: version = %llx", ckpt_ver);
+
+	/* do checkpoint periodically */
+	sbi->cp_expires = round_jiffies_up(jiffies + HZ * sbi->cp_interval);
 out:
 	mutex_unlock(&sbi->cp_mutex);
 	trace_f2fs_write_checkpoint(sbi->sb, cpc->reason, "finish checkpoint");

fs/f2fs/debug.c

@@ -33,11 +33,11 @@ static void update_general_status(struct f2fs_sb_info *sbi)
 	int i;
 
 	/* validation check of the segment numbers */
-	si->hit_largest = atomic_read(&sbi->read_hit_largest);
-	si->hit_cached = atomic_read(&sbi->read_hit_cached);
-	si->hit_rbtree = atomic_read(&sbi->read_hit_rbtree);
+	si->hit_largest = atomic64_read(&sbi->read_hit_largest);
+	si->hit_cached = atomic64_read(&sbi->read_hit_cached);
+	si->hit_rbtree = atomic64_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->total_ext = atomic64_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);
@@ -118,7 +118,7 @@ static void update_sit_info(struct f2fs_sb_info *sbi)
 		}
 	}
 	dist = div_u64(MAIN_SECS(sbi) * hblks_per_sec * hblks_per_sec, 100);
-	si->bimodal = div_u64(bimodal, dist);
+	si->bimodal = div64_u64(bimodal, dist);
 	if (si->dirty_count)
 		si->avg_vblocks = div_u64(total_vblocks, ndirty);
 	else
@@ -198,9 +198,9 @@ static void update_mem_info(struct f2fs_sb_info *sbi)
 
 	si->page_mem = 0;
 	npages = NODE_MAPPING(sbi)->nrpages;
-	si->page_mem += npages << PAGE_CACHE_SHIFT;
+	si->page_mem += (unsigned long long)npages << PAGE_CACHE_SHIFT;
 	npages = META_MAPPING(sbi)->nrpages;
-	si->page_mem += npages << PAGE_CACHE_SHIFT;
+	si->page_mem += (unsigned long long)npages << PAGE_CACHE_SHIFT;
 }
 
 static int stat_show(struct seq_file *s, void *v)
@@ -283,12 +283,12 @@ static int stat_show(struct seq_file *s, void *v)
 		seq_printf(s, "  - node blocks : %d (%d)\n", si->node_blks,
 				si->bg_node_blks);
 		seq_puts(s, "\nExtent Cache:\n");
-		seq_printf(s, "  - Hit Count: L1-1:%d L1-2:%d L2:%d\n",
+		seq_printf(s, "  - Hit Count: L1-1:%llu L1-2:%llu L2:%llu\n",
 				si->hit_largest, si->hit_cached,
 				si->hit_rbtree);
-		seq_printf(s, "  - Hit Ratio: %d%% (%d / %d)\n",
+		seq_printf(s, "  - Hit Ratio: %llu%% (%llu / %llu)\n",
 				!si->total_ext ? 0 :
-				(si->hit_total * 100) / si->total_ext,
+				div64_u64(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);
@@ -333,13 +333,13 @@ static int stat_show(struct seq_file *s, void *v)
 
 		/* memory footprint */
 		update_mem_info(si->sbi);
-		seq_printf(s, "\nMemory: %u KB\n",
+		seq_printf(s, "\nMemory: %llu KB\n",
 			(si->base_mem + si->cache_mem + si->page_mem) >> 10);
-		seq_printf(s, "  - static: %u KB\n",
+		seq_printf(s, "  - static: %llu KB\n",
 				si->base_mem >> 10);
-		seq_printf(s, "  - cached: %u KB\n",
+		seq_printf(s, "  - cached: %llu KB\n",
 				si->cache_mem >> 10);
-		seq_printf(s, "  - paged : %u KB\n",
+		seq_printf(s, "  - paged : %llu KB\n",
 				si->page_mem >> 10);
 	}
 	mutex_unlock(&f2fs_stat_mutex);
@@ -378,10 +378,10 @@ 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);
+	atomic64_set(&sbi->total_hit_ext, 0);
+	atomic64_set(&sbi->read_hit_rbtree, 0);
+	atomic64_set(&sbi->read_hit_largest, 0);
+	atomic64_set(&sbi->read_hit_cached, 0);
 
 	atomic_set(&sbi->inline_xattr, 0);
 	atomic_set(&sbi->inline_inode, 0);

fs/f2fs/dir.c

@@ -258,7 +258,7 @@ struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p)
 	if (f2fs_has_inline_dentry(dir))
 		return f2fs_parent_inline_dir(dir, p);
 
-	page = get_lock_data_page(dir, 0);
+	page = get_lock_data_page(dir, 0, false);
 	if (IS_ERR(page))
 		return NULL;
 
@@ -740,7 +740,7 @@ bool f2fs_empty_dir(struct inode *dir)
 		return f2fs_empty_inline_dir(dir);
 
 	for (bidx = 0; bidx < nblock; bidx++) {
-		dentry_page = get_lock_data_page(dir, bidx);
+		dentry_page = get_lock_data_page(dir, bidx, false);
 		if (IS_ERR(dentry_page)) {
 			if (PTR_ERR(dentry_page) == -ENOENT)
 				continue;
@@ -787,7 +787,6 @@ bool f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
 		else
 			d_type = DT_UNKNOWN;
 
-		/* encrypted case */
 		de_name.name = d->filename[bit_pos];
 		de_name.len = le16_to_cpu(de->name_len);
 
@@ -795,12 +794,20 @@ bool f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
 			int save_len = fstr->len;
 			int ret;
 
+			de_name.name = kmalloc(de_name.len, GFP_NOFS);
+			if (!de_name.name)
+				return false;
+
+			memcpy(de_name.name, d->filename[bit_pos], de_name.len);
+
 			ret = f2fs_fname_disk_to_usr(d->inode, &de->hash_code,
 							&de_name, fstr);
-			de_name = *fstr;
-			fstr->len = save_len;
+			kfree(de_name.name);
 			if (ret < 0)
 				return true;
+
+			de_name = *fstr;
+			fstr->len = save_len;
 		}
 
 		if (!dir_emit(ctx, de_name.name, de_name.len,
@@ -847,7 +854,7 @@ static int f2fs_readdir(struct file *file, struct dir_context *ctx)
 				min(npages - n, (pgoff_t)MAX_DIR_RA_PAGES));
 
 	for (; n < npages; n++) {
-		dentry_page = get_lock_data_page(inode, n);
+		dentry_page = get_lock_data_page(inode, n, false);
 		if (IS_ERR(dentry_page))
 			continue;
 

fs/f2fs/extent_cache.c

@@ -155,11 +155,12 @@ static unsigned int __free_extent_tree(struct f2fs_sb_info *sbi,
 	return count - et->count;
 }
 
-static void __drop_largest_extent(struct inode *inode, pgoff_t fofs)
+static void __drop_largest_extent(struct inode *inode,
+					pgoff_t fofs, unsigned int len)
 {
 	struct extent_info *largest = &F2FS_I(inode)->extent_tree->largest;
 
-	if (largest->fofs <= fofs && largest->fofs + largest->len > fofs)
+	if (fofs < largest->fofs + largest->len && fofs + len > largest->fofs)
 		largest->len = 0;
 }
 
@@ -168,7 +169,7 @@ void f2fs_drop_largest_extent(struct inode *inode, pgoff_t fofs)
 	if (!f2fs_may_extent_tree(inode))
 		return;
 
-	__drop_largest_extent(inode, fofs);
+	__drop_largest_extent(inode, fofs, 1);
 }
 
 void f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext)
@@ -350,8 +351,7 @@ static struct extent_node *__try_merge_extent_node(struct f2fs_sb_info *sbi,
 	}
 
 	if (en) {
-		if (en->ei.len > et->largest.len)
-			et->largest = en->ei;
+		__try_update_largest_extent(et, en);
 		et->cached_en = en;
 	}
 	return en;
@@ -388,18 +388,17 @@ static struct extent_node *__insert_extent_tree(struct f2fs_sb_info *sbi,
 	if (!en)
 		return NULL;
 
-	if (en->ei.len > et->largest.len)
-		et->largest = en->ei;
+	__try_update_largest_extent(et, en);
 	et->cached_en = en;
 	return en;
 }
 
-unsigned int f2fs_update_extent_tree_range(struct inode *inode,
+static 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 *en = NULL, *en1 = 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;
@@ -409,6 +408,8 @@ unsigned int f2fs_update_extent_tree_range(struct inode *inode,
 	if (!et)
 		return false;
 
+	trace_f2fs_update_extent_tree_range(inode, fofs, blkaddr, len);
+
 	write_lock(&et->lock);
 
 	if (is_inode_flag_set(F2FS_I(inode), FI_NO_EXTENT)) {
@@ -419,148 +420,99 @@ unsigned int f2fs_update_extent_tree_range(struct inode *inode,
 	prev = et->largest;
 	dei.len = 0;
 
-	/* we do not guarantee that the largest extent is cached all the time */
-	__drop_largest_extent(inode, fofs);
+	/*
+	 * drop largest extent before lookup, in case it's already
+	 * been shrunk from extent tree
+	 */
+	__drop_largest_extent(inode, fofs, len);
 
 	/* 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) {
+	if (!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;
+	while (en && en->ei.fofs < end) {
+		unsigned int org_end;
+		int parts = 0;	/* # of parts current extent split into */
 
-		if (pos >= end)
-			break;
+		next_en = en1 = NULL;
 
 		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;
-			}
+		org_end = dei.fofs + dei.len;
+		f2fs_bug_on(sbi, pos >= org_end);
 
-			if (__is_extent_same(&dei, &et->largest))
-				et->largest = en->ei;
-			goto next;
+		if (pos > dei.fofs &&	pos - dei.fofs >= F2FS_MIN_EXTENT_LEN) {
+			en->ei.len = pos - en->ei.fofs;
+			prev_en = en;
+			parts = 1;
 		}
 
-		/* case#2, invalidate left part of extent node |XXXXXVVVVV| */
-		if (pos <= dei.fofs && end < dei.fofs + dei.len) {
+		if (end < org_end && org_end - end >= F2FS_MIN_EXTENT_LEN) {
+			if (parts) {
+				set_extent_info(&ei, end,
+						end - dei.fofs + dei.blk,
+						org_end - end);
+				en1 = __insert_extent_tree(sbi, et, &ei,
+							NULL, NULL);
+				next_en = en1;
+			} else {
 				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;
+				next_en = en;
+			}
+			parts++;
 		}
 
-			if (__is_extent_same(&dei, &et->largest))
-				et->largest = en->ei;
-			goto next;
+		if (!next_en) {
+			struct rb_node *node = rb_next(&en->rb_node);
+
+			next_en = node ?
+				rb_entry(node, struct extent_node, rb_node)
+				: NULL;
 		}
 
+		if (parts)
+			__try_update_largest_extent(et, en);
+		else
 			__detach_extent_node(sbi, et, en);
 
 		/*
-		 * if we remove node in rb-tree, our parent node pointer may
-		 * point the wrong place, discard them.
+		 * if original extent is split into zero or two parts, extent
+		 * tree has been altered by deletion or insertion, therefore
+		 * invalidate pointers regard to tree.
 		 */
+		if (parts != 1) {
 			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 */
+		/* update in global extent list */
 		spin_lock(&sbi->extent_lock);
-		if (en && !list_empty(&en->list))
+		if (!parts && !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)
+		/* release extent node */
+		if (!parts)
 			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;
+
+		en = next_en;
 	}
 
-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,
+		en1 = __try_merge_extent_node(sbi, et, &ei, &den,
 							prev_en, next_en);
-		if (!en3)
-			en3 = __insert_extent_tree(sbi, et, &ei,
+		if (!en1)
+			en1 = __insert_extent_tree(sbi, et, &ei,
 						insert_p, insert_parent);
 
 		/* give up extent_cache, if split and small updates happen */
@@ -572,11 +524,11 @@ unsigned int f2fs_update_extent_tree_range(struct inode *inode,
 		}
 
 		spin_lock(&sbi->extent_lock);
-		if (en3) {
-			if (list_empty(&en3->list))
-				list_add_tail(&en3->list, &sbi->extent_list);
+		if (en1) {
+			if (list_empty(&en1->list))
+				list_add_tail(&en1->list, &sbi->extent_list);
 			else
-				list_move_tail(&en3->list, &sbi->extent_list);
+				list_move_tail(&en1->list, &sbi->extent_list);
 		}
 		if (den && !list_empty(&den->list))
 			list_del(&den->list);
@@ -650,6 +602,11 @@ unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink)
 	}
 	spin_unlock(&sbi->extent_lock);
 
+	/*
+	 * reset ino for searching victims from beginning of global extent tree.
+	 */
+	ino = F2FS_ROOT_INO(sbi);
+
 	while ((found = radix_tree_gang_lookup(root,
 				(void **)treevec, ino, EXT_TREE_VEC_SIZE))) {
 		unsigned i;
@@ -663,7 +620,7 @@ unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink)
 			write_unlock(&et->lock);
 
 			if (node_cnt + tree_cnt >= nr_shrink)
-				break;
+				goto unlock_out;
 		}
 	}
 unlock_out:

fs/f2fs/f2fs.h

@@ -19,6 +19,7 @@
 #include <linux/magic.h>
 #include <linux/kobject.h>
 #include <linux/sched.h>
+#include <linux/vmalloc.h>
 #include <linux/bio.h>
 
 #ifdef CONFIG_F2FS_CHECK_FS
@@ -52,6 +53,7 @@
 #define F2FS_MOUNT_NOBARRIER		0x00000800
 #define F2FS_MOUNT_FASTBOOT		0x00001000
 #define F2FS_MOUNT_EXTENT_CACHE		0x00002000
+#define F2FS_MOUNT_FORCE_FG_GC		0x00004000
 
 #define clear_opt(sbi, option)	(sbi->mount_opt.opt &= ~F2FS_MOUNT_##option)
 #define set_opt(sbi, option)	(sbi->mount_opt.opt |= F2FS_MOUNT_##option)
@@ -122,6 +124,7 @@ enum {
 		(SM_I(sbi)->trim_sections * (sbi)->segs_per_sec)
 #define BATCHED_TRIM_BLOCKS(sbi)	\
 		(BATCHED_TRIM_SEGMENTS(sbi) << (sbi)->log_blocks_per_seg)
+#define DEF_CP_INTERVAL			60	/* 60 secs */
 
 struct cp_control {
 	int reason;
@@ -230,6 +233,7 @@ static inline bool __has_cursum_space(struct f2fs_summary_block *sum, int size,
 #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_WRITE_CHECKPOINT	_IO(F2FS_IOCTL_MAGIC, 7)
 
 #define F2FS_IOC_SET_ENCRYPTION_POLICY					\
 		_IOR('f', 19, struct f2fs_encryption_policy)
@@ -246,6 +250,7 @@ static inline bool __has_cursum_space(struct f2fs_summary_block *sum, int size,
 #define F2FS_GOING_DOWN_FULLSYNC	0x0	/* going down with full sync */
 #define F2FS_GOING_DOWN_METASYNC	0x1	/* going down with metadata */
 #define F2FS_GOING_DOWN_NOSYNC		0x2	/* going down */
+#define F2FS_GOING_DOWN_METAFLUSH	0x3	/* going down with meta flush */
 
 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 /*
@@ -492,12 +497,20 @@ static inline bool __is_front_mergeable(struct extent_info *cur,
 	return __is_extent_mergeable(cur, front);
 }
 
+static inline void __try_update_largest_extent(struct extent_tree *et,
+						struct extent_node *en)
+{
+	if (en->ei.len > et->largest.len)
+		et->largest = en->ei;
+}
+
 struct f2fs_nm_info {
 	block_t nat_blkaddr;		/* base disk address of NAT */
 	nid_t max_nid;			/* maximum possible node ids */
 	nid_t available_nids;		/* maximum available node ids */
 	nid_t next_scan_nid;		/* the next nid to be scanned */
 	unsigned int ram_thresh;	/* control the memory footprint */
+	unsigned int ra_nid_pages;	/* # of nid pages to be readaheaded */
 
 	/* NAT cache management */
 	struct radix_tree_root nat_root;/* root of the nat entry cache */
@@ -724,6 +737,7 @@ struct f2fs_sb_info {
 	struct rw_semaphore node_write;		/* locking node writes */
 	struct mutex writepages;		/* mutex for writepages() */
 	wait_queue_head_t cp_wait;
+	long cp_expires, cp_interval;		/* next expected periodic cp */
 
 	struct inode_management im[MAX_INO_ENTRY];      /* manage inode cache */
 
@@ -787,10 +801,10 @@ 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 */
-	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 */
+	atomic64_t total_hit_ext;		/* # of lookup extent cache */
+	atomic64_t read_hit_rbtree;		/* # of hit rbtree extent node */
+	atomic64_t read_hit_largest;		/* # of hit largest extent node */
+	atomic64_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 */
@@ -1220,6 +1234,24 @@ static inline unsigned int valid_inode_count(struct f2fs_sb_info *sbi)
 	return sbi->total_valid_inode_count;
 }
 
+static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
+						pgoff_t index, bool for_write)
+{
+	if (!for_write)
+		return grab_cache_page(mapping, index);
+	return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
+}
+
+static inline void f2fs_copy_page(struct page *src, struct page *dst)
+{
+	char *src_kaddr = kmap(src);
+	char *dst_kaddr = kmap(dst);
+
+	memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
+	kunmap(dst);
+	kunmap(src);
+}
+
 static inline void f2fs_put_page(struct page *page, int unlock)
 {
 	if (!page)
@@ -1579,6 +1611,26 @@ static inline bool f2fs_may_extent_tree(struct inode *inode)
 	return S_ISREG(mode);
 }
 
+static inline void *f2fs_kvmalloc(size_t size, gfp_t flags)
+{
+	void *ret;
+
+	ret = kmalloc(size, flags | __GFP_NOWARN);
+	if (!ret)
+		ret = __vmalloc(size, flags, PAGE_KERNEL);
+	return ret;
+}
+
+static inline void *f2fs_kvzalloc(size_t size, gfp_t flags)
+{
+	void *ret;
+
+	ret = kzalloc(size, flags | __GFP_NOWARN);
+	if (!ret)
+		ret = __vmalloc(size, flags | __GFP_ZERO, PAGE_KERNEL);
+	return ret;
+}
+
 #define get_inode_mode(i) \
 	((is_inode_flag_set(F2FS_I(i), FI_ACL_MODE)) ? \
 	 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
@@ -1721,6 +1773,7 @@ int f2fs_issue_flush(struct f2fs_sb_info *);
 int create_flush_cmd_control(struct f2fs_sb_info *);
 void destroy_flush_cmd_control(struct f2fs_sb_info *);
 void invalidate_blocks(struct f2fs_sb_info *, block_t);
+bool is_checkpointed_data(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 *);
@@ -1739,6 +1792,7 @@ void f2fs_replace_block(struct f2fs_sb_info *, struct dnode_of_data *,
 void allocate_data_block(struct f2fs_sb_info *, struct page *,
 		block_t, block_t *, struct f2fs_summary *, int);
 void f2fs_wait_on_page_writeback(struct page *, enum page_type);
+void f2fs_wait_on_encrypted_page_writeback(struct f2fs_sb_info *, block_t);
 void write_data_summaries(struct f2fs_sb_info *, block_t);
 void write_node_summaries(struct f2fs_sb_info *, block_t);
 int lookup_journal_in_cursum(struct f2fs_summary_block *,
@@ -1754,8 +1808,9 @@ void destroy_segment_manager_caches(void);
  */
 struct page *grab_meta_page(struct f2fs_sb_info *, pgoff_t);
 struct page *get_meta_page(struct f2fs_sb_info *, pgoff_t);
+struct page *get_tmp_page(struct f2fs_sb_info *, pgoff_t);
 bool is_valid_blkaddr(struct f2fs_sb_info *, block_t, int);
-int ra_meta_pages(struct f2fs_sb_info *, block_t, int, int);
+int ra_meta_pages(struct f2fs_sb_info *, block_t, int, int, bool);
 void ra_meta_pages_cond(struct f2fs_sb_info *, pgoff_t);
 long sync_meta_pages(struct f2fs_sb_info *, enum page_type, long);
 void add_dirty_inode(struct f2fs_sb_info *, nid_t, int type);
@@ -1787,9 +1842,9 @@ 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);
-struct page *get_read_data_page(struct inode *, pgoff_t, int);
+struct page *get_read_data_page(struct inode *, pgoff_t, int, bool);
 struct page *find_data_page(struct inode *, pgoff_t);
-struct page *get_lock_data_page(struct inode *, pgoff_t);
+struct page *get_lock_data_page(struct inode *, pgoff_t, bool);
 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);
@@ -1802,7 +1857,7 @@ int f2fs_release_page(struct page *, gfp_t);
 int start_gc_thread(struct f2fs_sb_info *);
 void stop_gc_thread(struct f2fs_sb_info *);
 block_t start_bidx_of_node(unsigned int, struct f2fs_inode_info *);
-int f2fs_gc(struct f2fs_sb_info *);
+int f2fs_gc(struct f2fs_sb_info *, bool);
 void build_gc_manager(struct f2fs_sb_info *);
 
 /*
@@ -1820,7 +1875,8 @@ 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_largest, hit_cached, hit_rbtree, hit_total, total_ext;
+	unsigned long long hit_largest, hit_cached, hit_rbtree;
+	unsigned long long 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;
@@ -1844,7 +1900,7 @@ struct f2fs_stat_info {
 	unsigned int segment_count[2];
 	unsigned int block_count[2];
 	unsigned int inplace_count;
-	unsigned base_mem, cache_mem, page_mem;
+	unsigned long long base_mem, cache_mem, page_mem;
 };
 
 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
@@ -1857,10 +1913,10 @@ 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(sbi)		(atomic_inc(&(sbi)->total_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_total_hit(sbi)		(atomic64_inc(&(sbi)->total_hit_ext))
+#define stat_inc_rbtree_node_hit(sbi)	(atomic64_inc(&(sbi)->read_hit_rbtree))
+#define stat_inc_largest_node_hit(sbi)	(atomic64_inc(&(sbi)->read_hit_largest))
+#define stat_inc_cached_node_hit(sbi)	(atomic64_inc(&(sbi)->read_hit_cached))
 #define stat_inc_inline_xattr(inode)					\
 	do {								\
 		if (f2fs_has_inline_xattr(inode))			\
@@ -1998,6 +2054,8 @@ void f2fs_delete_inline_entry(struct f2fs_dir_entry *, struct page *,
 bool f2fs_empty_inline_dir(struct inode *);
 int f2fs_read_inline_dir(struct file *, struct dir_context *,
 						struct f2fs_str *);
+int f2fs_inline_data_fiemap(struct inode *,
+		struct fiemap_extent_info *, __u64, __u64);
 
 /*
  * shrinker.c

fs/f2fs/gc.c

@@ -78,9 +78,12 @@ static int gc_thread_func(void *data)
 		stat_inc_bggc_count(sbi);
 
 		/* if return value is not zero, no victim was selected */
-		if (f2fs_gc(sbi))
+		if (f2fs_gc(sbi, test_opt(sbi, FORCE_FG_GC)))
 			wait_ms = gc_th->no_gc_sleep_time;
 
+		trace_f2fs_background_gc(sbi->sb, wait_ms,
+				prefree_segments(sbi), free_segments(sbi));
+
 		/* balancing f2fs's metadata periodically */
 		f2fs_balance_fs_bg(sbi);
 
@@ -257,6 +260,7 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
 	struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
 	struct victim_sel_policy p;
 	unsigned int secno, max_cost;
+	unsigned int last_segment = MAIN_SEGS(sbi);
 	int nsearched = 0;
 
 	mutex_lock(&dirty_i->seglist_lock);
@@ -267,6 +271,9 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
 	p.min_segno = NULL_SEGNO;
 	p.min_cost = max_cost = get_max_cost(sbi, &p);
 
+	if (p.max_search == 0)
+		goto out;
+
 	if (p.alloc_mode == LFS && gc_type == FG_GC) {
 		p.min_segno = check_bg_victims(sbi);
 		if (p.min_segno != NULL_SEGNO)
@@ -277,9 +284,10 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
 		unsigned long cost;
 		unsigned int segno;
 
-		segno = find_next_bit(p.dirty_segmap, MAIN_SEGS(sbi), p.offset);
-		if (segno >= MAIN_SEGS(sbi)) {
+		segno = find_next_bit(p.dirty_segmap, last_segment, p.offset);
+		if (segno >= last_segment) {
 			if (sbi->last_victim[p.gc_mode]) {
+				last_segment = sbi->last_victim[p.gc_mode];
 				sbi->last_victim[p.gc_mode] = 0;
 				p.offset = 0;
 				continue;
@@ -327,6 +335,7 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
 				sbi->cur_victim_sec,
 				prefree_segments(sbi), free_segments(sbi));
 	}
+out:
 	mutex_unlock(&dirty_i->seglist_lock);
 
 	return (p.min_segno == NULL_SEGNO) ? 0 : 1;
@@ -541,7 +550,7 @@ static void move_encrypted_block(struct inode *inode, block_t bidx)
 	int err;
 
 	/* do not read out */
-	page = grab_cache_page(inode->i_mapping, bidx);
+	page = f2fs_grab_cache_page(inode->i_mapping, bidx, false);
 	if (!page)
 		return;
 
@@ -550,8 +559,16 @@ static void move_encrypted_block(struct inode *inode, block_t bidx)
 	if (err)
 		goto out;
 
-	if (unlikely(dn.data_blkaddr == NULL_ADDR))
+	if (unlikely(dn.data_blkaddr == NULL_ADDR)) {
+		ClearPageUptodate(page);
 		goto put_out;
+	}
+
+	/*
+	 * don't cache encrypted data into meta inode until previous dirty
+	 * data were writebacked to avoid racing between GC and flush.
+	 */
+	f2fs_wait_on_page_writeback(page, DATA);
 
 	get_node_info(fio.sbi, dn.nid, &ni);
 	set_summary(&sum, dn.nid, dn.ofs_in_node, ni.version);
@@ -580,7 +597,7 @@ static void move_encrypted_block(struct inode *inode, block_t bidx)
 		goto put_page_out;
 
 	set_page_dirty(fio.encrypted_page);
-	f2fs_wait_on_page_writeback(fio.encrypted_page, META);
+	f2fs_wait_on_page_writeback(fio.encrypted_page, DATA);
 	if (clear_page_dirty_for_io(fio.encrypted_page))
 		dec_page_count(fio.sbi, F2FS_DIRTY_META);
 
@@ -611,7 +628,7 @@ static void move_data_page(struct inode *inode, block_t bidx, int gc_type)
 {
 	struct page *page;
 
-	page = get_lock_data_page(inode, bidx);
+	page = get_lock_data_page(inode, bidx, true);
 	if (IS_ERR(page))
 		return;
 
@@ -705,7 +722,7 @@ static int gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
 
 			start_bidx = start_bidx_of_node(nofs, F2FS_I(inode));
 			data_page = get_read_data_page(inode,
-					start_bidx + ofs_in_node, READA);
+					start_bidx + ofs_in_node, READA, true);
 			if (IS_ERR(data_page)) {
 				iput(inode);
 				continue;
@@ -797,13 +814,12 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi, unsigned int segno,
 	return nfree;
 }
 
-int f2fs_gc(struct f2fs_sb_info *sbi)
+int f2fs_gc(struct f2fs_sb_info *sbi, bool sync)
 {
-	unsigned int segno = NULL_SEGNO;
-	unsigned int i;
-	int gc_type = BG_GC;
-	int nfree = 0;
-	int ret = -1;
+	unsigned int segno, i;
+	int gc_type = sync ? FG_GC : BG_GC;
+	int sec_freed = 0;
+	int ret = -EINVAL;
 	struct cp_control cpc;
 	struct gc_inode_list gc_list = {
 		.ilist = LIST_HEAD_INIT(gc_list.ilist),
@@ -812,12 +828,14 @@ int f2fs_gc(struct f2fs_sb_info *sbi)
 
 	cpc.reason = __get_cp_reason(sbi);
 gc_more:
+	segno = NULL_SEGNO;
+
 	if (unlikely(!(sbi->sb->s_flags & MS_ACTIVE)))
 		goto stop;
 	if (unlikely(f2fs_cp_error(sbi)))
 		goto stop;
 
-	if (gc_type == BG_GC && has_not_enough_free_secs(sbi, nfree)) {
+	if (gc_type == BG_GC && has_not_enough_free_secs(sbi, sec_freed)) {
 		gc_type = FG_GC;
 		if (__get_victim(sbi, &segno, gc_type) || prefree_segments(sbi))
 			write_checkpoint(sbi, &cpc);
@@ -830,23 +848,38 @@ int f2fs_gc(struct f2fs_sb_info *sbi)
 	/* readahead multi ssa blocks those have contiguous address */
 	if (sbi->segs_per_sec > 1)
 		ra_meta_pages(sbi, GET_SUM_BLOCK(sbi, segno), sbi->segs_per_sec,
-								META_SSA);
+							META_SSA, true);
 
-	for (i = 0; i < sbi->segs_per_sec; i++)
-		nfree += do_garbage_collect(sbi, segno + i, &gc_list, gc_type);
+	for (i = 0; i < sbi->segs_per_sec; i++) {
+		/*
+		 * for FG_GC case, halt gcing left segments once failed one
+		 * of segments in selected section to avoid long latency.
+		 */
+		if (!do_garbage_collect(sbi, segno + i, &gc_list, gc_type) &&
+				gc_type == FG_GC)
+			break;
+	}
+
+	if (i == sbi->segs_per_sec && gc_type == FG_GC)
+		sec_freed++;
 
 	if (gc_type == FG_GC)
 		sbi->cur_victim_sec = NULL_SEGNO;
 
-	if (has_not_enough_free_secs(sbi, nfree))
+	if (!sync) {
+		if (has_not_enough_free_secs(sbi, sec_freed))
 			goto gc_more;
 
 		if (gc_type == FG_GC)
 			write_checkpoint(sbi, &cpc);
+	}
 stop:
 	mutex_unlock(&sbi->gc_mutex);
 
 	put_gc_inode(&gc_list);
+
+	if (sync)
+		ret = sec_freed ? 0 : -EAGAIN;
 	return ret;
 }
 

fs/f2fs/gc.h

@@ -19,12 +19,6 @@
 #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

@@ -12,6 +12,7 @@
 #include <linux/f2fs_fs.h>
 
 #include "f2fs.h"
+#include "node.h"
 
 bool f2fs_may_inline_data(struct inode *inode)
 {
@@ -274,12 +275,14 @@ bool recover_inline_data(struct inode *inode, struct page *npage)
 	if (f2fs_has_inline_data(inode)) {
 		ipage = get_node_page(sbi, inode->i_ino);
 		f2fs_bug_on(sbi, IS_ERR(ipage));
-		truncate_inline_inode(ipage, 0);
+		if (!truncate_inline_inode(ipage, 0))
+			return false;
 		f2fs_clear_inline_inode(inode);
 		update_inode(inode, ipage);
 		f2fs_put_page(ipage, 1);
 	} else if (ri && (ri->i_inline & F2FS_INLINE_DATA)) {
-		truncate_blocks(inode, 0, false);
+		if (truncate_blocks(inode, 0, false))
+			return false;
 		goto process_inline;
 	}
 	return false;
@@ -568,3 +571,38 @@ int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
 	f2fs_put_page(ipage, 1);
 	return 0;
 }
+
+int f2fs_inline_data_fiemap(struct inode *inode,
+		struct fiemap_extent_info *fieinfo, __u64 start, __u64 len)
+{
+	__u64 byteaddr, ilen;
+	__u32 flags = FIEMAP_EXTENT_DATA_INLINE | FIEMAP_EXTENT_NOT_ALIGNED |
+		FIEMAP_EXTENT_LAST;
+	struct node_info ni;
+	struct page *ipage;
+	int err = 0;
+
+	ipage = get_node_page(F2FS_I_SB(inode), inode->i_ino);
+	if (IS_ERR(ipage))
+		return PTR_ERR(ipage);
+
+	if (!f2fs_has_inline_data(inode)) {
+		err = -EAGAIN;
+		goto out;
+	}
+
+	ilen = min_t(size_t, MAX_INLINE_DATA, i_size_read(inode));
+	if (start >= ilen)
+		goto out;
+	if (start + len < ilen)
+		ilen = start + len;
+	ilen -= start;
+
+	get_node_info(F2FS_I_SB(inode), inode->i_ino, &ni);
+	byteaddr = (__u64)ni.blk_addr << inode->i_sb->s_blocksize_bits;
+	byteaddr += (char *)inline_data_addr(ipage) - (char *)F2FS_INODE(ipage);
+	err = fiemap_fill_next_extent(fieinfo, start, byteaddr, ilen, flags);
+out:
+	f2fs_put_page(ipage, 1);
+	return err;
+}

fs/f2fs/inode.c

@@ -296,16 +296,12 @@ int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc)
 		return 0;
 
 	/*
-	 * We need to lock here to prevent from producing dirty node pages
+	 * We need to balance fs here to prevent from producing dirty node pages
 	 * during the urgent cleaning time when runing out of free sections.
 	 */
-	f2fs_lock_op(sbi);
 	update_inode_page(inode);
-	f2fs_unlock_op(sbi);
 
-	if (wbc)
 	f2fs_balance_fs(sbi);
-
 	return 0;
 }
 

fs/f2fs/namei.c

@@ -410,11 +410,14 @@ static int f2fs_symlink(struct inode *dir, struct dentry *dentry,
 	 * If the symlink path is stored into inline_data, there is no
 	 * performance regression.
 	 */
-	if (!err)
+	if (!err) {
 		filemap_write_and_wait_range(inode->i_mapping, 0, p_len - 1);
 
 		if (IS_DIRSYNC(dir))
 			f2fs_sync_fs(sbi->sb, 1);
+	} else {
+		f2fs_unlink(dir, dentry);
+	}
 
 	kfree(sd);
 	f2fs_fname_crypto_free_buffer(&disk_link);
@@ -947,8 +950,13 @@ static const char *f2fs_encrypted_follow_link(struct dentry *dentry, void **cook
 
 	/* Symlink is encrypted */
 	sd = (struct f2fs_encrypted_symlink_data *)caddr;
-	cstr.name = sd->encrypted_path;
 	cstr.len = le16_to_cpu(sd->len);
+	cstr.name = kmalloc(cstr.len, GFP_NOFS);
+	if (!cstr.name) {
+		res = -ENOMEM;
+		goto errout;
+	}
+	memcpy(cstr.name, sd->encrypted_path, cstr.len);
 
 	/* this is broken symlink case */
 	if (cstr.name[0] == 0 && cstr.len == 0) {
@@ -970,6 +978,8 @@ static const char *f2fs_encrypted_follow_link(struct dentry *dentry, void **cook
 	if (res < 0)
 		goto errout;
 
+	kfree(cstr.name);
+
 	paddr = pstr.name;
 
 	/* Null-terminate the name */
@@ -979,6 +989,7 @@ static const char *f2fs_encrypted_follow_link(struct dentry *dentry, void **cook
 	page_cache_release(cpage);
 	return *cookie = paddr;
 errout:
+	kfree(cstr.name);
 	f2fs_fname_crypto_free_buffer(&pstr);
 	kunmap(cpage);
 	page_cache_release(cpage);

fs/f2fs/node.c

@@ -1323,23 +1323,24 @@ static int f2fs_write_node_page(struct page *page,
 	nid = nid_of_node(page);
 	f2fs_bug_on(sbi, page->index != nid);
 
+	if (wbc->for_reclaim) {
+		if (!down_read_trylock(&sbi->node_write))
+			goto redirty_out;
+	} else {
+		down_read(&sbi->node_write);
+	}
+
 	get_node_info(sbi, nid, &ni);
 
 	/* This page is already truncated */
 	if (unlikely(ni.blk_addr == NULL_ADDR)) {
 		ClearPageUptodate(page);
 		dec_page_count(sbi, F2FS_DIRTY_NODES);
+		up_read(&sbi->node_write);
 		unlock_page(page);
 		return 0;
 	}
 
-	if (wbc->for_reclaim) {
-		if (!down_read_trylock(&sbi->node_write))
-			goto redirty_out;
-	} else {
-		down_read(&sbi->node_write);
-	}
-
 	set_page_writeback(page);
 	fio.blk_addr = ni.blk_addr;
 	write_node_page(nid, &fio);
@@ -1528,7 +1529,8 @@ static void build_free_nids(struct f2fs_sb_info *sbi)
 		return;
 
 	/* readahead nat pages to be scanned */
-	ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nid), FREE_NID_PAGES, META_NAT);
+	ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nid), FREE_NID_PAGES,
+							META_NAT, true);
 
 	while (1) {
 		struct page *page = get_current_nat_page(sbi, nid);
@@ -1558,6 +1560,9 @@ static void build_free_nids(struct f2fs_sb_info *sbi)
 			remove_free_nid(nm_i, nid);
 	}
 	mutex_unlock(&curseg->curseg_mutex);
+
+	ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nm_i->next_scan_nid),
+					nm_i->ra_nid_pages, META_NAT, false);
 }
 
 /*
@@ -1803,10 +1808,10 @@ int restore_node_summary(struct f2fs_sb_info *sbi,
 		nrpages = min(last_offset - i, bio_blocks);
 
 		/* readahead node pages */
-		ra_meta_pages(sbi, addr, nrpages, META_POR);
+		ra_meta_pages(sbi, addr, nrpages, META_POR, true);
 
 		for (idx = addr; idx < addr + nrpages; idx++) {
-			struct page *page = get_meta_page(sbi, idx);
+			struct page *page = get_tmp_page(sbi, idx);
 
 			rn = F2FS_NODE(page);
 			sum_entry->nid = rn->footer.nid;
@@ -2000,6 +2005,7 @@ static int init_node_manager(struct f2fs_sb_info *sbi)
 	nm_i->fcnt = 0;
 	nm_i->nat_cnt = 0;
 	nm_i->ram_thresh = DEF_RAM_THRESHOLD;
+	nm_i->ra_nid_pages = DEF_RA_NID_PAGES;
 
 	INIT_RADIX_TREE(&nm_i->free_nid_root, GFP_ATOMIC);
 	INIT_LIST_HEAD(&nm_i->free_nid_list);

fs/f2fs/node.h

@@ -14,9 +14,11 @@
 /* node block offset on the NAT area dedicated to the given start node id */
 #define	NAT_BLOCK_OFFSET(start_nid) (start_nid / NAT_ENTRY_PER_BLOCK)
 
-/* # of pages to perform readahead before building free nids */
+/* # of pages to perform synchronous readahead before building free nids */
 #define FREE_NID_PAGES 4
 
+#define DEF_RA_NID_PAGES	4	/* # of nid pages to be readaheaded */
+
 /* maximum readahead size for node during getting data blocks */
 #define MAX_RA_NODE		128
 

fs/f2fs/recovery.c

@@ -180,7 +180,7 @@ static int find_fsync_dnodes(struct f2fs_sb_info *sbi, struct list_head *head)
 	curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
 	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
 
-	ra_meta_pages(sbi, blkaddr, 1, META_POR);
+	ra_meta_pages(sbi, blkaddr, 1, META_POR, true);
 
 	while (1) {
 		struct fsync_inode_entry *entry;
@@ -188,7 +188,7 @@ static int find_fsync_dnodes(struct f2fs_sb_info *sbi, struct list_head *head)
 		if (!is_valid_blkaddr(sbi, blkaddr, META_POR))
 			return 0;
 
-		page = get_meta_page(sbi, blkaddr);
+		page = get_tmp_page(sbi, blkaddr);
 
 		if (cp_ver != cpver_of_node(page))
 			break;
@@ -383,15 +383,11 @@ static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
 	start = start_bidx_of_node(ofs_of_node(page), fi);
 	end = start + ADDRS_PER_PAGE(page, fi);
 
-	f2fs_lock_op(sbi);
-
 	set_new_dnode(&dn, inode, NULL, NULL, 0);
 
 	err = get_dnode_of_data(&dn, start, ALLOC_NODE);
-	if (err) {
-		f2fs_unlock_op(sbi);
+	if (err)
 		goto out;
-	}
 
 	f2fs_wait_on_page_writeback(dn.node_page, NODE);
 
@@ -456,7 +452,6 @@ static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
 	set_page_dirty(dn.node_page);
 err:
 	f2fs_put_dnode(&dn);
-	f2fs_unlock_op(sbi);
 out:
 	f2fs_msg(sbi->sb, KERN_NOTICE,
 		"recover_data: ino = %lx, recovered = %d blocks, err = %d",
@@ -485,7 +480,7 @@ static int recover_data(struct f2fs_sb_info *sbi,
 
 		ra_meta_pages_cond(sbi, blkaddr);
 
-		page = get_meta_page(sbi, blkaddr);
+		page = get_tmp_page(sbi, blkaddr);
 
 		if (cp_ver != cpver_of_node(page)) {
 			f2fs_put_page(page, 1);
@@ -570,7 +565,7 @@ int recover_fsync_data(struct f2fs_sb_info *sbi)
 
 	/* truncate meta pages to be used by the recovery */
 	truncate_inode_pages_range(META_MAPPING(sbi),
-			MAIN_BLKADDR(sbi) << PAGE_CACHE_SHIFT, -1);
+			(loff_t)MAIN_BLKADDR(sbi) << PAGE_CACHE_SHIFT, -1);
 
 	if (err) {
 		truncate_inode_pages_final(NODE_MAPPING(sbi));

fs/f2fs/segment.h

@@ -137,10 +137,12 @@ enum {
 /*
  * BG_GC means the background cleaning job.
  * FG_GC means the on-demand cleaning job.
+ * FORCE_FG_GC means on-demand cleaning job in background.
  */
 enum {
 	BG_GC = 0,
-	FG_GC
+	FG_GC,
+	FORCE_FG_GC,
 };
 
 /* for a function parameter to select a victim segment */

fs/f2fs/super.c

@@ -213,8 +213,10 @@ F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, ipu_policy, ipu_policy);
 F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_ipu_util, min_ipu_util);
 F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_fsync_blocks, min_fsync_blocks);
 F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ram_thresh, ram_thresh);
+F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ra_nid_pages, ra_nid_pages);
 F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, max_victim_search, max_victim_search);
 F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, dir_level, dir_level);
+F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, cp_interval, cp_interval);
 
 #define ATTR_LIST(name) (&f2fs_attr_##name.attr)
 static struct attribute *f2fs_attrs[] = {
@@ -231,6 +233,8 @@ static struct attribute *f2fs_attrs[] = {
 	ATTR_LIST(max_victim_search),
 	ATTR_LIST(dir_level),
 	ATTR_LIST(ram_thresh),
+	ATTR_LIST(ra_nid_pages),
+	ATTR_LIST(cp_interval),
 	NULL,
 };
 
@@ -292,11 +296,16 @@ static int parse_options(struct super_block *sb, char *options)
 
 			if (!name)
 				return -ENOMEM;
-			if (strlen(name) == 2 && !strncmp(name, "on", 2))
+			if (strlen(name) == 2 && !strncmp(name, "on", 2)) {
 				set_opt(sbi, BG_GC);
-			else if (strlen(name) == 3 && !strncmp(name, "off", 3))
+				clear_opt(sbi, FORCE_FG_GC);
+			} else if (strlen(name) == 3 && !strncmp(name, "off", 3)) {
 				clear_opt(sbi, BG_GC);
-			else {
+				clear_opt(sbi, FORCE_FG_GC);
+			} else if (strlen(name) == 4 && !strncmp(name, "sync", 4)) {
+				set_opt(sbi, BG_GC);
+				set_opt(sbi, FORCE_FG_GC);
+			} else {
 				kfree(name);
 				return -EINVAL;
 			}
@@ -631,10 +640,14 @@ static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
 {
 	struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
 
-	if (!f2fs_readonly(sbi->sb) && test_opt(sbi, BG_GC))
-		seq_printf(seq, ",background_gc=%s", "on");
+	if (!f2fs_readonly(sbi->sb) && test_opt(sbi, BG_GC)) {
+		if (test_opt(sbi, FORCE_FG_GC))
+			seq_printf(seq, ",background_gc=%s", "sync");
 		else
+			seq_printf(seq, ",background_gc=%s", "on");
+	} else {
 		seq_printf(seq, ",background_gc=%s", "off");
+	}
 	if (test_opt(sbi, DISABLE_ROLL_FORWARD))
 		seq_puts(seq, ",disable_roll_forward");
 	if (test_opt(sbi, DISCARD))
@@ -742,6 +755,7 @@ static int f2fs_remount(struct super_block *sb, int *flags, char *data)
 	int err, active_logs;
 	bool need_restart_gc = false;
 	bool need_stop_gc = false;
+	bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
 
 	sync_filesystem(sb);
 
@@ -767,6 +781,14 @@ static int f2fs_remount(struct super_block *sb, int *flags, char *data)
 	if (f2fs_readonly(sb) && (*flags & MS_RDONLY))
 		goto skip;
 
+	/* disallow enable/disable extent_cache dynamically */
+	if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
+		err = -EINVAL;
+		f2fs_msg(sbi->sb, KERN_WARNING,
+				"switch extent_cache option is not allowed");
+		goto restore_opts;
+	}
+
 	/*
 	 * We stop the GC thread if FS is mounted as RO
 	 * or if background_gc = off is passed in mount
@@ -996,6 +1018,7 @@ static void init_sb_info(struct f2fs_sb_info *sbi)
 		atomic_set(&sbi->nr_pages[i], 0);
 
 	sbi->dir_level = DEF_DIR_LEVEL;
+	sbi->cp_interval = DEF_CP_INTERVAL;
 	clear_sbi_flag(sbi, SBI_NEED_FSCK);
 
 	INIT_LIST_HEAD(&sbi->s_list);
@@ -1332,6 +1355,8 @@ static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
 		f2fs_commit_super(sbi, true);
 	}
 
+	sbi->cp_expires = round_jiffies_up(jiffies);
+
 	return 0;
 
 free_kobj:

include/trace/events/f2fs.h

@@ -514,6 +514,34 @@ TRACE_EVENT(f2fs_map_blocks,
 		__entry->ret)
 );
 
+TRACE_EVENT(f2fs_background_gc,
+
+	TP_PROTO(struct super_block *sb, long wait_ms,
+			unsigned int prefree, unsigned int free),
+
+	TP_ARGS(sb, wait_ms, prefree, free),
+
+	TP_STRUCT__entry(
+		__field(dev_t,	dev)
+		__field(long,	wait_ms)
+		__field(unsigned int,	prefree)
+		__field(unsigned int,	free)
+	),
+
+	TP_fast_assign(
+		__entry->dev		= sb->s_dev;
+		__entry->wait_ms	= wait_ms;
+		__entry->prefree	= prefree;
+		__entry->free		= free;
+	),
+
+	TP_printk("dev = (%d,%d), wait_ms = %ld, prefree = %u, free = %u",
+		show_dev(__entry),
+		__entry->wait_ms,
+		__entry->prefree,
+		__entry->free)
+);
+
 TRACE_EVENT(f2fs_get_victim,
 
 	TP_PROTO(struct super_block *sb, int type, int gc_type,
@@ -1000,6 +1028,32 @@ TRACE_EVENT(f2fs_writepages,
 		__entry->for_sync)
 );
 
+TRACE_EVENT(f2fs_readpages,
+
+	TP_PROTO(struct inode *inode, struct page *page, unsigned int nrpage),
+
+	TP_ARGS(inode, page, nrpage),
+
+	TP_STRUCT__entry(
+		__field(dev_t,	dev)
+		__field(ino_t,	ino)
+		__field(pgoff_t,	start)
+		__field(unsigned int,	nrpage)
+	),
+
+	TP_fast_assign(
+		__entry->dev	= inode->i_sb->s_dev;
+		__entry->ino	= inode->i_ino;
+		__entry->start	= page->index;
+		__entry->nrpage	= nrpage;
+	),
+
+	TP_printk("dev = (%d,%d), ino = %lu, start = %lu nrpage = %u",
+		show_dev_ino(__entry),
+		(unsigned long)__entry->start,
+		__entry->nrpage)
+);
+
 TRACE_EVENT(f2fs_write_checkpoint,
 
 	TP_PROTO(struct super_block *sb, int reason, char *msg),
@@ -1132,17 +1186,19 @@ TRACE_EVENT_CONDITION(f2fs_lookup_extent_tree_end,
 		__entry->len)
 );
 
-TRACE_EVENT(f2fs_update_extent_tree,
+TRACE_EVENT(f2fs_update_extent_tree_range,
 
-	TP_PROTO(struct inode *inode, unsigned int pgofs, block_t blkaddr),
+	TP_PROTO(struct inode *inode, unsigned int pgofs, block_t blkaddr,
+						unsigned int len),
 
-	TP_ARGS(inode, pgofs, blkaddr),
+	TP_ARGS(inode, pgofs, blkaddr, len),
 
 	TP_STRUCT__entry(
 		__field(dev_t,	dev)
 		__field(ino_t,	ino)
 		__field(unsigned int, pgofs)
 		__field(u32, blk)
+		__field(unsigned int, len)
 	),
 
 	TP_fast_assign(
@@ -1150,12 +1206,15 @@ TRACE_EVENT(f2fs_update_extent_tree,
 		__entry->ino = inode->i_ino;
 		__entry->pgofs = pgofs;
 		__entry->blk = blkaddr;
+		__entry->len = len;
 	),
 
-	TP_printk("dev = (%d,%d), ino = %lu, pgofs = %u, blkaddr = %u",
+	TP_printk("dev = (%d,%d), ino = %lu, pgofs = %u, "
+					"blkaddr = %u, len = %u",
 		show_dev_ino(__entry),
 		__entry->pgofs,
-		__entry->blk)
+		__entry->blk,
+		__entry->len)
 );
 
 TRACE_EVENT(f2fs_shrink_extent_tree,