xfs
[Top] [All Lists]

Re: [PATCH 2/7] xfs: split inode flushing from xfs_sync_inodes_ag

To: Eric Sandeen <sandeen@xxxxxxxxxxx>
Subject: Re: [PATCH 2/7] xfs: split inode flushing from xfs_sync_inodes_ag
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Wed, 27 May 2009 06:58:41 -0400
Cc: Christoph Hellwig <hch@xxxxxxxxxxxxx>, xfs@xxxxxxxxxxx
In-reply-to: <4A1C54FB.2080705@xxxxxxxxxxx>
References: <20090514171233.942489000@xxxxxxxxxxxxxxxxxxxxxx> <20090514171558.298098000@xxxxxxxxxxxxxxxxxxxxxx> <4A1C54FB.2080705@xxxxxxxxxxx>
User-agent: Mutt/1.5.18 (2008-05-17)
On Tue, May 26, 2009 at 03:45:47PM -0500, Eric Sandeen wrote:
> What happened to the XFS_IFLUSH_DELWRI case?
> 
> You mentioned "keep the current behavior exactly" but this seems like a
> change, no?

Yeah, this got lost when playing with variations of the patch.  Correct
version below:

Subject: xfs: split inode flushing from xfs_sync_inodes_ag
From: Christoph Hellwig <hch@xxxxxx>


In many cases we only want to sync inode metadata. Split out the inode
flushing into a separate helper to prepare factoring the inode sync code.

Based on a patch from Dave Chinner, but redone to keep the current behaviour
exactly and leave changes to the flushing logic to another patch.


Signed-off-by: Christoph Hellwig <hch@xxxxxx>

Index: xfs/fs/xfs/linux-2.6/xfs_sync.c
===================================================================
--- xfs.orig/fs/xfs/linux-2.6/xfs_sync.c        2009-05-27 12:08:49.137850883 
+0200
+++ xfs/fs/xfs/linux-2.6/xfs_sync.c     2009-05-27 12:11:09.530815659 +0200
@@ -77,6 +77,35 @@ xfs_sync_inode_data(
        return error;
 }
 
+STATIC int
+xfs_sync_inode_attr(
+       struct xfs_inode        *ip,
+       int                     flags)
+{
+       int                     error = 0;
+
+       xfs_ilock(ip, XFS_ILOCK_SHARED);
+       if (xfs_inode_clean(ip))
+               goto out_unlock;
+       if (!xfs_iflock_nowait(ip)) {
+               if (!(flags & SYNC_WAIT))
+                       goto out_unlock;
+               xfs_iflock(ip);
+       }
+
+       if (xfs_inode_clean(ip)) {
+               xfs_ifunlock(ip);
+               goto out_unlock;
+       }
+
+       error = xfs_iflush(ip, (flags & SYNC_WAIT) ?
+                          XFS_IFLUSH_SYNC : XFS_IFLUSH_DELWRI);
+
+ out_unlock:
+       xfs_iunlock(ip, XFS_ILOCK_SHARED);
+       return error;
+}
+
 /*
  * Sync all the inodes in the given AG according to the
  * direction given by the flags.
@@ -96,7 +125,6 @@ xfs_sync_inodes_ag(
        do {
                struct inode    *inode;
                xfs_inode_t     *ip = NULL;
-               int             lock_flags = XFS_ILOCK_SHARED;
 
                /*
                 * use a gang lookup to find the next inode in the tree
@@ -155,22 +183,10 @@ xfs_sync_inodes_ag(
                if (flags & SYNC_DELWRI)
                        error = xfs_sync_inode_data(ip, flags);
 
-               xfs_ilock(ip, XFS_ILOCK_SHARED);
-               if ((flags & SYNC_ATTR) && !xfs_inode_clean(ip)) {
-                       if (flags & SYNC_WAIT) {
-                               xfs_iflock(ip);
-                               if (!xfs_inode_clean(ip))
-                                       error = xfs_iflush(ip, XFS_IFLUSH_SYNC);
-                               else
-                                       xfs_ifunlock(ip);
-                       } else if (xfs_iflock_nowait(ip)) {
-                               if (!xfs_inode_clean(ip))
-                                       error = xfs_iflush(ip, 
XFS_IFLUSH_DELWRI);
-                               else
-                                       xfs_ifunlock(ip);
-                       }
-               }
-               xfs_iput(ip, lock_flags);
+               if (flags & SYNC_ATTR)
+                       error = xfs_sync_inode_attr(ip, flags);
+
+               IRELE(ip);
 
                if (error)
                        last_error = error;

<Prev in Thread] Current Thread [Next in Thread>