[PATCH 3.15 1/2] f2fs: adjust free mem size to flush dentry blocks

[PATCH 3.15 1/2] f2fs: adjust free mem size to flush dentry blocks

[Jaegeuk Kim]

From: Jaegeuk Kim <jaegeuk.kim@samsung.com>

commit 6fb03f3a40805a412c9b285010ffdc2e7563f81b upstream.

If so many dirty dentry blocks are cached, not reached to the flush condition,
we should fall into livelock in balance_dirty_pages.
So, let's consider the mem size for the condition.

Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>

Conflicts:
	fs/f2fs/node.h
---
 fs/f2fs/data.c |  3 ++-
 fs/f2fs/f2fs.h |  1 +
 fs/f2fs/node.c | 44 ++++++++++++++++++++++++++------------------
 fs/f2fs/node.h |  5 +++--
 4 files changed, 32 insertions(+), 21 deletions(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index 45abd60..716645e 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -867,7 +867,8 @@ static int f2fs_write_data_pages(struct address_space *mapping,
 		return 0;
 
 	if (S_ISDIR(inode->i_mode) && wbc->sync_mode == WB_SYNC_NONE &&
-			get_dirty_dents(inode) < nr_pages_to_skip(sbi, DATA))
+			get_dirty_dents(inode) < nr_pages_to_skip(sbi, DATA) &&
+			available_free_memory(sbi, DIRTY_DENTS))
 		goto skip_write;
 
 	diff = nr_pages_to_write(sbi, DATA, wbc);
diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 2ecac83..656523c 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -1140,6 +1140,7 @@ f2fs_hash_t f2fs_dentry_hash(const char *, size_t);
 struct dnode_of_data;
 struct node_info;
 
+bool available_free_memory(struct f2fs_sb_info *, int);
 int is_checkpointed_node(struct f2fs_sb_info *, nid_t);
 bool fsync_mark_done(struct f2fs_sb_info *, nid_t);
 void get_node_info(struct f2fs_sb_info *, nid_t, struct node_info *);
diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index a161e95..a9c0495 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -26,20 +26,26 @@
 static struct kmem_cache *nat_entry_slab;
 static struct kmem_cache *free_nid_slab;
 
-static inline bool available_free_memory(struct f2fs_nm_info *nm_i, int type)
+bool available_free_memory(struct f2fs_sb_info *sbi, int type)
 {
+	struct f2fs_nm_info *nm_i = NM_I(sbi);
 	struct sysinfo val;
 	unsigned long mem_size = 0;
+	bool res = false;
 
 	si_meminfo(&val);
-	if (type == FREE_NIDS)
-		mem_size = nm_i->fcnt * sizeof(struct free_nid);
-	else if (type == NAT_ENTRIES)
-		mem_size += nm_i->nat_cnt * sizeof(struct nat_entry);
-	mem_size >>= 12;
-
-	/* give 50:50 memory for free nids and nat caches respectively */
-	return (mem_size < ((val.totalram * nm_i->ram_thresh) >> 11));
+	/* give 25%, 25%, 50% memory for each components respectively */
+	if (type == FREE_NIDS) {
+		mem_size = (nm_i->fcnt * sizeof(struct free_nid)) >> 12;
+		res = mem_size < ((val.totalram * nm_i->ram_thresh / 100) >> 2);
+	} else if (type == NAT_ENTRIES) {
+		mem_size = (nm_i->nat_cnt * sizeof(struct nat_entry)) >> 12;
+		res = mem_size < ((val.totalram * nm_i->ram_thresh / 100) >> 2);
+	} else if (type == DIRTY_DENTS) {
+		mem_size = get_pages(sbi, F2FS_DIRTY_DENTS);
+		res = mem_size < ((val.totalram * nm_i->ram_thresh / 100) >> 1);
+	}
+	return res;
 }
 
 static void clear_node_page_dirty(struct page *page)
@@ -243,7 +249,7 @@ int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink)
 {
 	struct f2fs_nm_info *nm_i = NM_I(sbi);
 
-	if (available_free_memory(nm_i, NAT_ENTRIES))
+	if (available_free_memory(sbi, NAT_ENTRIES))
 		return 0;
 
 	write_lock(&nm_i->nat_tree_lock);
@@ -1315,13 +1321,14 @@ static void __del_from_free_nid_list(struct f2fs_nm_info *nm_i,
 	radix_tree_delete(&nm_i->free_nid_root, i->nid);
 }
 
-static int add_free_nid(struct f2fs_nm_info *nm_i, nid_t nid, bool build)
+static int add_free_nid(struct f2fs_sb_info *sbi, nid_t nid, bool build)
 {
+	struct f2fs_nm_info *nm_i = NM_I(sbi);
 	struct free_nid *i;
 	struct nat_entry *ne;
 	bool allocated = false;
 
-	if (!available_free_memory(nm_i, FREE_NIDS))
+	if (!available_free_memory(sbi, FREE_NIDS))
 		return -1;
 
 	/* 0 nid should not be used */
@@ -1374,9 +1381,10 @@ static void remove_free_nid(struct f2fs_nm_info *nm_i, nid_t nid)
 		kmem_cache_free(free_nid_slab, i);
 }
 
-static void scan_nat_page(struct f2fs_nm_info *nm_i,
+static void scan_nat_page(struct f2fs_sb_info *sbi,
 			struct page *nat_page, nid_t start_nid)
 {
+	struct f2fs_nm_info *nm_i = NM_I(sbi);
 	struct f2fs_nat_block *nat_blk = page_address(nat_page);
 	block_t blk_addr;
 	int i;
@@ -1391,7 +1399,7 @@ static void scan_nat_page(struct f2fs_nm_info *nm_i,
 		blk_addr = le32_to_cpu(nat_blk->entries[i].block_addr);
 		f2fs_bug_on(blk_addr == NEW_ADDR);
 		if (blk_addr == NULL_ADDR) {
-			if (add_free_nid(nm_i, start_nid, true) < 0)
+			if (add_free_nid(sbi, start_nid, true) < 0)
 				break;
 		}
 	}
@@ -1415,7 +1423,7 @@ static void build_free_nids(struct f2fs_sb_info *sbi)
 	while (1) {
 		struct page *page = get_current_nat_page(sbi, nid);
 
-		scan_nat_page(nm_i, page, nid);
+		scan_nat_page(sbi, page, nid);
 		f2fs_put_page(page, 1);
 
 		nid += (NAT_ENTRY_PER_BLOCK - (nid % NAT_ENTRY_PER_BLOCK));
@@ -1435,7 +1443,7 @@ static void build_free_nids(struct f2fs_sb_info *sbi)
 		block_t addr = le32_to_cpu(nat_in_journal(sum, i).block_addr);
 		nid = le32_to_cpu(nid_in_journal(sum, i));
 		if (addr == NULL_ADDR)
-			add_free_nid(nm_i, nid, true);
+			add_free_nid(sbi, nid, true);
 		else
 			remove_free_nid(nm_i, nid);
 	}
@@ -1512,7 +1520,7 @@ void alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid)
 	spin_lock(&nm_i->free_nid_list_lock);
 	i = __lookup_free_nid_list(nm_i, nid);
 	f2fs_bug_on(!i || i->state != NID_ALLOC);
-	if (!available_free_memory(nm_i, FREE_NIDS)) {
+	if (!available_free_memory(sbi, FREE_NIDS)) {
 		__del_from_free_nid_list(nm_i, i);
 		need_free = true;
 	} else {
@@ -1843,7 +1851,7 @@ flush_now:
 		}
 
 		if (nat_get_blkaddr(ne) == NULL_ADDR &&
-				add_free_nid(NM_I(sbi), nid, false) <= 0) {
+				add_free_nid(sbi, nid, false) <= 0) {
 			write_lock(&nm_i->nat_tree_lock);
 			__del_from_nat_cache(nm_i, ne);
 			write_unlock(&nm_i->nat_tree_lock);
diff --git a/fs/f2fs/node.h b/fs/f2fs/node.h
index 5decc1a..b517077 100644
--- a/fs/f2fs/node.h
+++ b/fs/f2fs/node.h
@@ -75,9 +75,10 @@ static inline void node_info_from_raw_nat(struct node_info *ni,
 	ni->version = raw_ne->version;
 }
 
-enum nid_type {
+enum mem_type {
 	FREE_NIDS,	/* indicates the free nid list */
-	NAT_ENTRIES	/* indicates the cached nat entry */
+	NAT_ENTRIES,	/* indicates the cached nat entry */
+	DIRTY_DENTS	/* indicates dirty dentry pages */
 };
 
 /*
-- 
2.0.1.472.g6f92e5f

[PATCH 3.15 2/2] f2fs: check bdi->dirty_exceeded when trying to skip data writes

[Jaegeuk Kim]

commit dd6b9bf5883c3ca9c17bac80ccd8615fe5a452a3 upstream.

If we don't check the current backing device status, balance_dirty_pages can
fall into infinite pausing routine.

This can be occurred when a lot of directories make a small number of dirty
dentry pages including files.

Reported-by: Brian Chadwick <brianchad@westnet.com.au>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/node.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index a9c0495..5696bde 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -42,6 +42,8 @@ bool available_free_memory(struct f2fs_sb_info *sbi, int type)
 		mem_size = (nm_i->nat_cnt * sizeof(struct nat_entry)) >> 12;
 		res = mem_size < ((val.totalram * nm_i->ram_thresh / 100) >> 2);
 	} else if (type == DIRTY_DENTS) {
+		if (sbi->sb->s_bdi->dirty_exceeded)
+			return false;
 		mem_size = get_pages(sbi, F2FS_DIRTY_DENTS);
 		res = mem_size < ((val.totalram * nm_i->ram_thresh / 100) >> 1);
 	}
-- 
2.0.1.472.g6f92e5f


------------------------------------------------------------------------------
Want fast and easy access to all the code in your enterprise? Index and
search up to 200,000 lines of code with a free copy of Black Duck&#174;
Code Sight&#153; - the same software that powers the world's largest code
search on Ohloh, the Black Duck Open Hub! Try it now.
http://p.sf.net/sfu/bds

Re: [PATCH 3.15 2/2] f2fs: check bdi->dirty_exceeded when trying to skip data writes

[Greg KH]

On Mon, Jul 14, 2014 at 07:52:51AM -0700, Jaegeuk Kim wrote:
> commit dd6b9bf5883c3ca9c17bac80ccd8615fe5a452a3 upstream.

You mean 2743f865543c0c4a5e12fc13edb2bf89a6e9687c, right?

thanks,

greg k-h

> 
> If we don't check the current backing device status, balance_dirty_pages can
> fall into infinite pausing routine.
> 
> This can be occurred when a lot of directories make a small number of dirty
> dentry pages including files.
> 
> Reported-by: Brian Chadwick <brianchad@westnet.com.au>
> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> ---
>  fs/f2fs/node.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
> index a9c0495..5696bde 100644
> --- a/fs/f2fs/node.c
> +++ b/fs/f2fs/node.c
> @@ -42,6 +42,8 @@ bool available_free_memory(struct f2fs_sb_info *sbi, int type)
>  		mem_size = (nm_i->nat_cnt * sizeof(struct nat_entry)) >> 12;
>  		res = mem_size < ((val.totalram * nm_i->ram_thresh / 100) >> 2);
>  	} else if (type == DIRTY_DENTS) {
> +		if (sbi->sb->s_bdi->dirty_exceeded)
> +			return false;
>  		mem_size = get_pages(sbi, F2FS_DIRTY_DENTS);
>  		res = mem_size < ((val.totalram * nm_i->ram_thresh / 100) >> 1);
>  	}
> -- 
> 2.0.1.472.g6f92e5f
> 
> --
> To unsubscribe from this list: send the line "unsubscribe stable" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

------------------------------------------------------------------------------
Want fast and easy access to all the code in your enterprise? Index and
search up to 200,000 lines of code with a free copy of Black Duck
Code Sight - the same software that powers the world's largest code
search on Ohloh, the Black Duck Open Hub! Try it now.
http://p.sf.net/sfu/bds

Re: [PATCH 3.15 2/2] f2fs: check bdi->dirty_exceeded when trying to skip data writes

[Jaegeuk Kim]

On Mon, Jul 14, 2014 at 05:53:36PM -0700, Greg KH wrote:
> On Mon, Jul 14, 2014 at 07:52:51AM -0700, Jaegeuk Kim wrote:
> > commit dd6b9bf5883c3ca9c17bac80ccd8615fe5a452a3 upstream.
> 
> You mean 2743f865543c0c4a5e12fc13edb2bf89a6e9687c, right?

Ah, correct.
Thank you. :)

> 
> thanks,
> 
> greg k-h
> 
> > 
> > If we don't check the current backing device status, balance_dirty_pages can
> > fall into infinite pausing routine.
> > 
> > This can be occurred when a lot of directories make a small number of dirty
> > dentry pages including files.
> > 
> > Reported-by: Brian Chadwick <brianchad@westnet.com.au>
> > Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> > ---
> >  fs/f2fs/node.c | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
> > index a9c0495..5696bde 100644
> > --- a/fs/f2fs/node.c
> > +++ b/fs/f2fs/node.c
> > @@ -42,6 +42,8 @@ bool available_free_memory(struct f2fs_sb_info *sbi, int type)
> >  		mem_size = (nm_i->nat_cnt * sizeof(struct nat_entry)) >> 12;
> >  		res = mem_size < ((val.totalram * nm_i->ram_thresh / 100) >> 2);
> >  	} else if (type == DIRTY_DENTS) {
> > +		if (sbi->sb->s_bdi->dirty_exceeded)
> > +			return false;
> >  		mem_size = get_pages(sbi, F2FS_DIRTY_DENTS);
> >  		res = mem_size < ((val.totalram * nm_i->ram_thresh / 100) >> 1);
> >  	}
> > -- 
> > 2.0.1.472.g6f92e5f
> > 
> > --
> > To unsubscribe from this list: send the line "unsubscribe stable" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html

-- 
Jaegeuk Kim