[02/15] ocfs2: free inode when i_count becomes zero

Commit Message

Andrew Morton Dec. 15, 2014, 10:50 p.m. UTC

From: Xue jiufei <[e-mail]>
Subject: ocfs2: free inode when i_count becomes zero

Disk inode deletion may be heavily delayed when one node unlink a file
after the same dentry is freed on another node(say N1) because of memory
shrink but inode is left in memory.  This inode can only be freed while N1
doing the orphan scan work.

However, N1 may skip orphan scan for several times because other nodes may
do the work earlier.  In our tests, it may take 1 hour on 4 nodes cluster
and this will cause bad user experience.  So we think the inode should be
freed when i_count becomes zero to avoid such circumstances.

[[e-mail]: coding-style fixes]
Signed-off-by: joyce.xue <[e-mail]>
Cc: Mark Fasheh <[e-mail]>
Cc: Joel Becker <[e-mail]>
Signed-off-by: Andrew Morton <[e-mail]>
---

 fs/ocfs2/inode.c |   10 +---------
 1 file changed, 1 insertion(+), 9 deletions(-)

Comments

Xue jiufei Dec. 16, 2014, 1:37 a.m. UTC | #1

Hi, Andrew,
This patch may lead to data loss so please remove it from mm tree please.
Here is the situation: 
When i_count becomes zero but there still exists dirty pages in i_mapping,
the dirty pages would be freed without flushing the data.

To avoid this problem, we should flush dirty page before dropping
the inode, but I don't think it it a good idea to flush page in
function ocfs2_drop_inode().

So now there is no better way to solve this problem.

Thanks,
Xuejiufei

On 2014/12/16 6:50, [e-mail] wrote:
> From: Xue jiufei <[e-mail]>
> Subject: ocfs2: free inode when i_count becomes zero
> 
> Disk inode deletion may be heavily delayed when one node unlink a file
> after the same dentry is freed on another node(say N1) because of memory
> shrink but inode is left in memory.  This inode can only be freed while N1
> doing the orphan scan work.
> 
> However, N1 may skip orphan scan for several times because other nodes may
> do the work earlier.  In our tests, it may take 1 hour on 4 nodes cluster
> and this will cause bad user experience.  So we think the inode should be
> freed when i_count becomes zero to avoid such circumstances.
> 
> [[e-mail]: coding-style fixes]
> Signed-off-by: joyce.xue <[e-mail]>
> Cc: Mark Fasheh <[e-mail]>
> Cc: Joel Becker <[e-mail]>
> Signed-off-by: Andrew Morton <[e-mail]>
> ---
> 
>  fs/ocfs2/inode.c |   10 +---------
>  1 file changed, 1 insertion(+), 9 deletions(-)
> 
> diff -puN fs/ocfs2/inode.c~ocfs2-free-inode-when-i_count-becomes-zero fs/ocfs2/inode.c
> --- a/fs/ocfs2/inode.c~ocfs2-free-inode-when-i_count-becomes-zero
> +++ a/fs/ocfs2/inode.c
> @@ -1191,17 +1191,9 @@ void ocfs2_evict_inode(struct inode *ino
>  int ocfs2_drop_inode(struct inode *inode)
>  {
>  	struct ocfs2_inode_info *oi = OCFS2_I(inode);
> -	int res;
> -
>  	trace_ocfs2_drop_inode((unsigned long long)oi->ip_blkno,
>  				inode->i_nlink, oi->ip_flags);
> -
> -	if (oi->ip_flags & OCFS2_INODE_MAYBE_ORPHANED)
> -		res = 1;
> -	else
> -		res = generic_drop_inode(inode);
> -
> -	return res;
> +	return 1;
>  }
>  
>  /*
> _
> .
>

Patch

diff -puN fs/ocfs2/inode.c~ocfs2-free-inode-when-i_count-becomes-zero fs/ocfs2/inode.c
--- a/fs/ocfs2/inode.c~ocfs2-free-inode-when-i_count-becomes-zero
+++ a/fs/ocfs2/inode.c
@@ -1191,17 +1191,9 @@  void ocfs2_evict_inode(struct inode *ino
 int ocfs2_drop_inode(struct inode *inode)
 {
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
-	int res;
-
 	trace_ocfs2_drop_inode((unsigned long long)oi->ip_blkno,
 				inode->i_nlink, oi->ip_flags);
-
-	if (oi->ip_flags & OCFS2_INODE_MAYBE_ORPHANED)
-		res = 1;
-	else
-		res = generic_drop_inode(inode);
-
-	return res;
+	return 1;
 }
 
 /*