xfs
[Top] [All Lists]

Re: [PATCH 0/6] [XFS] introduce a AG inode tree walker

To: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Subject: Re: [PATCH 0/6] [XFS] introduce a AG inode tree walker
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Mon, 16 Mar 2009 20:40:19 +1100
Cc: xfs@xxxxxxxxxxx
In-reply-to: <20090316075338.GA19858@xxxxxxxxxxxxx>
Mail-followup-to: Christoph Hellwig <hch@xxxxxxxxxxxxx>, xfs@xxxxxxxxxxx
References: <1237117603-26071-1-git-send-email-david@xxxxxxxxxxxxx> <20090316075338.GA19858@xxxxxxxxxxxxx>
User-agent: Mutt/1.5.18 (2008-05-17)
On Mon, Mar 16, 2009 at 03:53:38AM -0400, Christoph Hellwig wrote:
> On Sun, Mar 15, 2009 at 10:46:37PM +1100, Dave Chinner wrote:
> > This series splits up the sync and reclaim code into three
> > separate actions. The first is the tree walker, the second is
> > the inode validation and the third is the operation to execute
> > on the inode.
> > 
> > This allows us to somewhat abstract the radix tree away from the
> > act of walking the cached inodes and puts in place mechanisms that
> > can be extended for bulk inode cache lookups.
> > 
> > This also splits the inode writeback into separate data and metadata
> > sync operations and optimises them a little......
> 
> Just did a quick XFSQA run with your whole patch series applied and
> I get this oops in 001:

I haven't seen that - it runs through a UML test run just fine.
I'll see if I can reproduce it...

If you drop the last patch (the iterator patch) does it work
ok?

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx

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