linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] f2fs: optimize alloc_nid_failed()
@ 2013-05-06 11:25 Haicheng Li
  2013-05-06 13:20 ` Haicheng Li
  0 siblings, 1 reply; 2+ messages in thread
From: Haicheng Li @ 2013-05-06 11:25 UTC (permalink / raw)
  To: linux-fsdevel, linux-f2fs-devel, Jaegeuk Kim
  Cc: linux-kernel, Haicheng Li, Haicheng Li

Current alloc_nid_failed() has two issues:
1. unecessarily kmem_cache_free() the free_nid and then realloc a new
   slab cache for it immediately;
2. needs to __lookup_free_nid_list() twice in order to find out the free_nid
   from free_nid_list.

This patch fixes these issues.

Signed-off-by: Haicheng Li <haicheng.li@linux.intel.com>
---
 fs/f2fs/node.c |   17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)

diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index e275218..7ce8e9f 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -1403,8 +1403,21 @@ void alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid)
  */
 void alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid)
 {
-	alloc_nid_done(sbi, nid);
-	add_free_nid(NM_I(sbi), nid);
+	struct f2fs_nm_info *nm_i = NM_I(sbi);
+	struct free_nid *i;
+
+	spin_lock(&nm_i->free_nid_list_lock);
+	i = __lookup_free_nid_list(nid, &nm_i->free_nid_list);
+	if (i) {
+		BUG_ON(i->state != NID_ALLOC);
+		if (nm_i->fcnt > 2 * MAX_FREE_NIDS)
+			__del_from_free_nid_list(i);
+		else {
+			i->state = NID_NEW;
+			nm_i->fcnt++;
+		}
+	}
+	spin_unlock(&nm_i->free_nid_list_lock);
 }
 
 void recover_node_page(struct f2fs_sb_info *sbi, struct page *page,
-- 
1.7.9.5


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] f2fs: optimize alloc_nid_failed()
  2013-05-06 11:25 [PATCH] f2fs: optimize alloc_nid_failed() Haicheng Li
@ 2013-05-06 13:20 ` Haicheng Li
  0 siblings, 0 replies; 2+ messages in thread
From: Haicheng Li @ 2013-05-06 13:20 UTC (permalink / raw)
  To: linux-fsdevel, linux-f2fs-devel, Jaegeuk Kim; +Cc: huajun.li, linux-kernel

Just realized I should develop the patches against Kim's f2fs-dev branch
rather than the f2fs-master branch. (thanks to Huajun for the reminding).

Obviously on f2fs-dev branch, some optimizations of alloc_nid_failed() have
already been done.

Pls. ignore this patch, I'll send out the rebased version of my recent f2fs
patches soon later, which will be based on the f2fs-dev tree.

On Mon, May 06, 2013 at 07:25:06PM +0800, Haicheng Li wrote:
> Current alloc_nid_failed() has two issues:
> 1. unecessarily kmem_cache_free() the free_nid and then realloc a new
>    slab cache for it immediately;
> 2. needs to __lookup_free_nid_list() twice in order to find out the free_nid
>    from free_nid_list.
> 
> This patch fixes these issues.
> 
> Signed-off-by: Haicheng Li <haicheng.li@linux.intel.com>
> ---
>  fs/f2fs/node.c |   17 +++++++++++++++--
>  1 file changed, 15 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
> index e275218..7ce8e9f 100644
> --- a/fs/f2fs/node.c
> +++ b/fs/f2fs/node.c
> @@ -1403,8 +1403,21 @@ void alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid)
>   */
>  void alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid)
>  {
> -	alloc_nid_done(sbi, nid);
> -	add_free_nid(NM_I(sbi), nid);
> +	struct f2fs_nm_info *nm_i = NM_I(sbi);
> +	struct free_nid *i;
> +
> +	spin_lock(&nm_i->free_nid_list_lock);
> +	i = __lookup_free_nid_list(nid, &nm_i->free_nid_list);
> +	if (i) {
> +		BUG_ON(i->state != NID_ALLOC);
> +		if (nm_i->fcnt > 2 * MAX_FREE_NIDS)
> +			__del_from_free_nid_list(i);
> +		else {
> +			i->state = NID_NEW;
> +			nm_i->fcnt++;
> +		}
> +	}
> +	spin_unlock(&nm_i->free_nid_list_lock);
>  }
>  
>  void recover_node_page(struct f2fs_sb_info *sbi, struct page *page,
> -- 
> 1.7.9.5

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2013-05-06 13:20 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2013-05-06 11:25 [PATCH] f2fs: optimize alloc_nid_failed() Haicheng Li
2013-05-06 13:20 ` Haicheng Li

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).