Re: [RFC] killing boilerplate checks in ->link/->mkdir/->rename
From: Artem Bityutskiy
Date: Tue Feb 07 2012 - 02:19:48 EST
On Mon, 2012-02-06 at 17:11 +0000, Al Viro wrote:
> On Mon, Feb 06, 2012 at 07:05:55PM +0200, Artem Bityutskiy wrote:
> > On Mon, 2012-02-06 at 13:56 +0000, Al Viro wrote:
> > > On Mon, Feb 06, 2012 at 10:50:44AM +0200, Artem Bityutskiy wrote:
> > >
> > > > WRT 'sen_nlink()' - I can use it instead of 'drop_nlink()'/'inc_nlink()'
> > > > of course. But I do not really see why is this better. E.g.,
> > > > 'drop_nlink()' additionally gives me ' WARN_ON()' in case of 'i_nlink'
> > > > wrapping.
> > >
> > > So does inc_nlink() when you are asking to get from nlink=0 to nlink=1.
> > > I.e. on failure exit in your unlink()...
> > Indeed! I'll switch to 'clear_nlink(inode)' and 'set_nlink(inode, 1)'
> > for those inodes which are being unliked.
> Huh? I thought ubifs allowed link(2)... IOW, i_nlink could've been greater
> than 1 when you called ubifs_unlink().
You are right, I'll take this correctly when I change the code rather
than writing an e-mail. Thanks! I think I'll submit a patch this week.
Description: This is a digitally signed message part