[Top] [All Lists]

Re: [PATCH 3/3] kill vn_revalidate

To: Timothy Shimmin <tes@xxxxxxx>
Subject: Re: [PATCH 3/3] kill vn_revalidate
From: Christoph Hellwig <hch@xxxxxx>
Date: Tue, 15 Jul 2008 14:07:07 +0200
Cc: Christoph Hellwig <hch@xxxxxx>, xfs@xxxxxxxxxxx
In-reply-to: <487C456D.3010509@xxxxxxx>
References: <20080627154602.GC31476@xxxxxx> <20080705172110.GB7177@xxxxxx> <487C456D.3010509@xxxxxxx>
Sender: xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.3.28i
On Tue, Jul 15, 2008 at 04:36:29PM +1000, Timothy Shimmin wrote:
> * in xfs_setattr we now explicitly set for linux inode:
> i_mode, i_uid, i_gid, i_atime, i_mtime, i_ctime
> It looks like we didn't set i_atime in vn_revalidate previously.

That was intentional because the VFS updates this one directly and
we already must only update it from setattr.

> This change looks weird:
> >      */
> > -   iattr.ia_mask = XFS_AT_MODE;
> > +   iattr.ia_valid = ATTR_MODE;
> >     iattr.ia_mode = xfs_vtoi(vp)->i_d.di_mode;
> I didn't think there was an ia_mask field - was this really
> from another patch? Confused.

That's because the previous patch did a botched bhv_vattr to iattr
conversion and I only noticed this when compiling the second patch
and accidentally fixed it there.

I've updated both patches to correct this and attached them.

Attachment: xfs-simplify-xfs_setattr
Description: Text document

Attachment: xfs-kill-vn_revalidate
Description: Text document

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