[PATCH 1/5] f2fs: do retry operations with cond_resched

[PATCH 1/5] f2fs: do retry operations with cond_resched

[Jaegeuk Kim]

This patch revists retrial paths in f2fs.
The basic idea is to use cond_resched instead of retrying from the very early
stage.

Suggested-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/f2fs.h    |  7 +++++++
 fs/f2fs/gc.c      |  5 ++---
 fs/f2fs/node.c    | 41 +++++++++--------------------------------
 fs/f2fs/segment.c |  5 ++---
 4 files changed, 20 insertions(+), 38 deletions(-)

diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index c873140..c787fe3 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -1021,6 +1021,13 @@ retry:
 	return entry;
 }
 
+static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
+				unsigned long index, void *item)
+{
+	while (radix_tree_insert(root, index, item))
+		cond_resched();
+}
+
 #define RAW_IS_INODE(p)	((p)->footer.nid == (p)->footer.ino)
 
 static inline bool IS_INODE(struct page *page)
diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
index 2c58c58..eec0933 100644
--- a/fs/f2fs/gc.c
+++ b/fs/f2fs/gc.c
@@ -356,12 +356,11 @@ static void add_gc_inode(struct gc_inode_list *gc_list, struct inode *inode)
 		iput(inode);
 		return;
 	}
-retry:
 	new_ie = f2fs_kmem_cache_alloc(winode_slab, GFP_NOFS);
 	new_ie->inode = inode;
-
+retry:
 	if (radix_tree_insert(&gc_list->iroot, inode->i_ino, new_ie)) {
-		kmem_cache_free(winode_slab, new_ie);
+		cond_resched();
 		goto retry;
 	}
 	list_add_tail(&new_ie->list, &gc_list->ilist);
diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index 8de4f55..f83326c 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -147,7 +147,7 @@ static void __set_nat_cache_dirty(struct f2fs_nm_info *nm_i,
 
 	if (get_nat_flag(ne, IS_DIRTY))
 		return;
-retry:
+
 	head = radix_tree_lookup(&nm_i->nat_set_root, set);
 	if (!head) {
 		head = f2fs_kmem_cache_alloc(nat_entry_set_slab, GFP_ATOMIC);
@@ -156,11 +156,7 @@ retry:
 		INIT_LIST_HEAD(&head->set_list);
 		head->set = set;
 		head->entry_cnt = 0;
-
-		if (radix_tree_insert(&nm_i->nat_set_root, set, head)) {
-			kmem_cache_free(nat_entry_set_slab, head);
-			goto retry;
-		}
+		f2fs_radix_tree_insert(&nm_i->nat_set_root, set, head);
 	}
 	list_move_tail(&ne->list, &head->entry_list);
 	nm_i->dirty_nat_cnt++;
@@ -238,13 +234,8 @@ static struct nat_entry *grab_nat_entry(struct f2fs_nm_info *nm_i, nid_t nid)
 {
 	struct nat_entry *new;
 
-	new = kmem_cache_alloc(nat_entry_slab, GFP_ATOMIC);
-	if (!new)
-		return NULL;
-	if (radix_tree_insert(&nm_i->nat_root, nid, new)) {
-		kmem_cache_free(nat_entry_slab, new);
-		return NULL;
-	}
+	new = f2fs_kmem_cache_alloc(nat_entry_slab, GFP_ATOMIC);
+	f2fs_radix_tree_insert(&nm_i->nat_root, nid, new);
 	memset(new, 0, sizeof(struct nat_entry));
 	nat_set_nid(new, nid);
 	nat_reset_flag(new);
@@ -257,15 +248,11 @@ static void cache_nat_entry(struct f2fs_nm_info *nm_i, nid_t nid,
 						struct f2fs_nat_entry *ne)
 {
 	struct nat_entry *e;
-retry:
+
 	down_write(&nm_i->nat_tree_lock);
 	e = __lookup_nat_cache(nm_i, nid);
 	if (!e) {
 		e = grab_nat_entry(nm_i, nid);
-		if (!e) {
-			up_write(&nm_i->nat_tree_lock);
-			goto retry;
-		}
 		node_info_from_raw_nat(&e->ni, ne);
 	}
 	up_write(&nm_i->nat_tree_lock);
@@ -276,15 +263,11 @@ static void set_node_addr(struct f2fs_sb_info *sbi, struct node_info *ni,
 {
 	struct f2fs_nm_info *nm_i = NM_I(sbi);
 	struct nat_entry *e;
-retry:
+
 	down_write(&nm_i->nat_tree_lock);
 	e = __lookup_nat_cache(nm_i, ni->nid);
 	if (!e) {
 		e = grab_nat_entry(nm_i, ni->nid);
-		if (!e) {
-			up_write(&nm_i->nat_tree_lock);
-			goto retry;
-		}
 		e->ni = *ni;
 		f2fs_bug_on(sbi, ni->blk_addr == NEW_ADDR);
 	} else if (new_blkaddr == NEW_ADDR) {
@@ -1833,19 +1816,13 @@ static void remove_nats_in_journal(struct f2fs_sb_info *sbi)
 		nid_t nid = le32_to_cpu(nid_in_journal(sum, i));
 
 		raw_ne = nat_in_journal(sum, i);
-retry:
+
 		down_write(&nm_i->nat_tree_lock);
 		ne = __lookup_nat_cache(nm_i, nid);
-		if (ne)
-			goto found;
-
-		ne = grab_nat_entry(nm_i, nid);
 		if (!ne) {
-			up_write(&nm_i->nat_tree_lock);
-			goto retry;
+			ne = grab_nat_entry(nm_i, nid);
+			node_info_from_raw_nat(&ne->ni, &raw_ne);
 		}
-		node_info_from_raw_nat(&ne->ni, &raw_ne);
-found:
 		__set_nat_cache_dirty(nm_i, ne);
 		up_write(&nm_i->nat_tree_lock);
 	}
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 9a33e34..c79d67e 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -179,13 +179,13 @@ void register_inmem_page(struct inode *inode, struct page *page)
 	struct f2fs_inode_info *fi = F2FS_I(inode);
 	struct inmem_pages *new;
 	int err;
-retry:
+
 	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);
@@ -195,7 +195,6 @@ retry:
 		return;
 	} else if (err) {
 		mutex_unlock(&fi->inmem_lock);
-		kmem_cache_free(inmem_entry_slab, new);
 		goto retry;
 	}
 	get_page(page);
-- 
2.1.1


------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk

[PATCH 2/5] f2fs: count inline_xx in do_read_inode

[Jaegeuk Kim]

In do_read_inode, if we failed __recover_inline_status, the inode has inline
flag without increasing its count.
Later, f2fs_evict_inode will decrease the count, which causes -1.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/inode.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c
index 9fe110e..196cc78 100644
--- a/fs/f2fs/inode.c
+++ b/fs/f2fs/inode.c
@@ -148,6 +148,10 @@ static int do_read_inode(struct inode *inode)
 	__get_inode_rdev(inode, ri);
 
 	f2fs_put_page(node_page, 1);
+
+	stat_inc_inline_inode(inode);
+	stat_inc_inline_dir(inode);
+
 	return err;
 }
 
@@ -199,8 +203,6 @@ make_now:
 		goto bad_inode;
 	}
 	unlock_new_inode(inode);
-	stat_inc_inline_inode(inode);
-	stat_inc_inline_dir(inode);
 	trace_f2fs_iget(inode);
 	return inode;
 
-- 
2.1.1


------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk

[PATCH 3/5] f2fs: set page private for inmemory pages for truncation

[Jaegeuk Kim]

The inmemory pages should be handled by invalidate_page since it needs to be
released int the truncation path.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/segment.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index c79d67e..3ebcf96 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -180,6 +180,8 @@ void register_inmem_page(struct inode *inode, struct page *page)
 	struct inmem_pages *new;
 	int err;
 
+	SetPagePrivate(page);
+
 	new = f2fs_kmem_cache_alloc(inmem_entry_slab, GFP_NOFS);
 
 	/* add atomic page indices to the list */
-- 
2.1.1


------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk

[PATCH 4/5] f2fs: release inmemory pages when the file was closed

[Jaegeuk Kim]

If file is closed, let's drop inmemory pages.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/file.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
index 146e58a..b6f3fbf 100644
--- a/fs/f2fs/file.c
+++ b/fs/f2fs/file.c
@@ -901,6 +901,14 @@ static int f2fs_ioc_start_atomic_write(struct file *filp)
 	return f2fs_convert_inline_inode(inode);
 }
 
+static int f2fs_release_file(struct inode *inode, struct file *filp)
+{
+	/* some remained atomic pages should discarded */
+	if (f2fs_is_atomic_file(inode) || f2fs_is_volatile_file(inode))
+		commit_inmem_pages(inode, true);
+	return 0;
+}
+
 static int f2fs_ioc_commit_atomic_write(struct file *filp)
 {
 	struct inode *inode = file_inode(filp);
@@ -1010,6 +1018,7 @@ const struct file_operations f2fs_file_operations = {
 	.read_iter	= generic_file_read_iter,
 	.write_iter	= generic_file_write_iter,
 	.open		= generic_file_open,
+	.release	= f2fs_release_file,
 	.mmap		= f2fs_file_mmap,
 	.fsync		= f2fs_sync_file,
 	.fallocate	= f2fs_fallocate,
-- 
2.1.1


------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk

[PATCH 5/5] f2fs: count the number of inmemory pages

[Jaegeuk Kim]

This patch adds counting # of inmemory pages in the page cache.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/debug.c   | 3 +++
 fs/f2fs/f2fs.h    | 3 ++-
 fs/f2fs/segment.c | 3 +++
 3 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/debug.c b/fs/f2fs/debug.c
index 40b679c..4e2e39c 100644
--- a/fs/f2fs/debug.c
+++ b/fs/f2fs/debug.c
@@ -39,6 +39,7 @@ static void update_general_status(struct f2fs_sb_info *sbi)
 	si->ndirty_dent = get_pages(sbi, F2FS_DIRTY_DENTS);
 	si->ndirty_dirs = sbi->n_dirty_dirs;
 	si->ndirty_meta = get_pages(sbi, F2FS_DIRTY_META);
+	si->inmem_pages = get_pages(sbi, F2FS_INMEM_PAGES);
 	si->total_count = (int)sbi->user_block_count / sbi->blocks_per_seg;
 	si->rsvd_segs = reserved_segments(sbi);
 	si->overp_segs = overprovision_segments(sbi);
@@ -249,6 +250,8 @@ static int stat_show(struct seq_file *s, void *v)
 		seq_printf(s, "\nExtent Hit Ratio: %d / %d\n",
 			   si->hit_ext, si->total_ext);
 		seq_puts(s, "\nBalancing F2FS Async:\n");
+		seq_printf(s, "  - inmem: %4d\n",
+			   si->inmem_pages);
 		seq_printf(s, "  - nodes: %4d in %4d\n",
 			   si->ndirty_node, si->node_pages);
 		seq_printf(s, "  - dents: %4d in dirs:%4d\n",
diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index c787fe3..f60b817 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -462,6 +462,7 @@ enum count_type {
 	F2FS_DIRTY_DENTS,
 	F2FS_DIRTY_NODES,
 	F2FS_DIRTY_META,
+	F2FS_INMEM_PAGES,
 	NR_COUNT_TYPE,
 };
 
@@ -1498,7 +1499,7 @@ struct f2fs_stat_info {
 	int ndirty_node, ndirty_dent, ndirty_dirs, ndirty_meta;
 	int nats, sits, fnids;
 	int total_count, utilization;
-	int bg_gc, inline_inode, inline_dir;
+	int bg_gc, inline_inode, inline_dir, inmem_pages;
 	unsigned int valid_count, valid_node_count, valid_inode_count;
 	unsigned int bimodal, avg_vblocks;
 	int util_free, util_valid, util_invalid;
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 3ebcf96..42607a6 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -201,6 +201,7 @@ retry:
 	}
 	get_page(page);
 	list_add_tail(&new->list, &fi->inmem_pages);
+	inc_page_count(F2FS_I_SB(inode), F2FS_INMEM_PAGES);
 	mutex_unlock(&fi->inmem_lock);
 }
 
@@ -216,6 +217,7 @@ void invalidate_inmem_page(struct inode *inode, struct page *page)
 		f2fs_put_page(cur->page, 0);
 		list_del(&cur->list);
 		kmem_cache_free(inmem_entry_slab, cur);
+		dec_page_count(F2FS_I_SB(inode), F2FS_INMEM_PAGES);
 	}
 	mutex_unlock(&fi->inmem_lock);
 }
@@ -257,6 +259,7 @@ void commit_inmem_pages(struct inode *inode, bool abort)
 		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);
 	}
 	if (submit_bio)
 		f2fs_submit_merged_bio(sbi, DATA, WRITE);
-- 
2.1.1


------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk

RE: [f2fs-dev] [PATCH 1/5] f2fs: do retry operations with cond_resched

[Chao Yu]

> -----Original Message-----
> From: Jaegeuk Kim [mailto:jaegeuk@kernel.org]
> Sent: Sunday, December 07, 2014 3:56 AM
> To: linux-kernel@vger.kernel.org; linux-fsdevel@vger.kernel.org;
> linux-f2fs-devel@lists.sourceforge.net
> Cc: Jaegeuk Kim
> Subject: [f2fs-dev] [PATCH 1/5] f2fs: do retry operations with cond_resched
> 
> This patch revists retrial paths in f2fs.
> The basic idea is to use cond_resched instead of retrying from the very early
> stage.
> 
> Suggested-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

Reviewed-by: Chao Yu <chao2.yu@samsung.com>

> ---
>  fs/f2fs/f2fs.h    |  7 +++++++
>  fs/f2fs/gc.c      |  5 ++---
>  fs/f2fs/node.c    | 41 +++++++++--------------------------------
>  fs/f2fs/segment.c |  5 ++---
>  4 files changed, 20 insertions(+), 38 deletions(-)
> 
> diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> index c873140..c787fe3 100644
> --- a/fs/f2fs/f2fs.h
> +++ b/fs/f2fs/f2fs.h
> @@ -1021,6 +1021,13 @@ retry:
>  	return entry;
>  }
> 
> +static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
> +				unsigned long index, void *item)
> +{
> +	while (radix_tree_insert(root, index, item))
> +		cond_resched();
> +}
> +
>  #define RAW_IS_INODE(p)	((p)->footer.nid == (p)->footer.ino)
> 
>  static inline bool IS_INODE(struct page *page)
> diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
> index 2c58c58..eec0933 100644
> --- a/fs/f2fs/gc.c
> +++ b/fs/f2fs/gc.c
> @@ -356,12 +356,11 @@ static void add_gc_inode(struct gc_inode_list *gc_list, struct inode
> *inode)
>  		iput(inode);
>  		return;
>  	}
> -retry:
>  	new_ie = f2fs_kmem_cache_alloc(winode_slab, GFP_NOFS);
>  	new_ie->inode = inode;
> -
> +retry:
>  	if (radix_tree_insert(&gc_list->iroot, inode->i_ino, new_ie)) {
> -		kmem_cache_free(winode_slab, new_ie);
> +		cond_resched();
>  		goto retry;
>  	}

use f2fs_radix_tree_insert instead?

Thanks,
Yu