Re: [PATCH 5/5] vfs: change nondirectory i_mutex ordering to fix quota deadlock

[Jan Kara <jack@suse.cz>]

On Wed 25-04-12 11:28:58, J. Bruce Fields wrote:
> On Wed, Apr 25, 2012 at 11:22:09AM -0400, bfields@fieldses.org wrote:
> > diff --git a/fs/inode.c b/fs/inode.c
> > index 487c924..13d23b6 100644
> > --- a/fs/inode.c
> > +++ b/fs/inode.c
> > @@ -961,6 +961,17 @@ void unlock_new_inode(struct inode *inode)
> >  }
> >  EXPORT_SYMBOL(unlock_new_inode);
> >  
> > +/*
> > + * We order !IS_NOQUOTA files before ISNOQUOTA files, and by pointer
> > + * within each category.
> > + */
> > +static bool nondir_mutex_ordered(struct inode *inode1, struct inode *inode2)
> > +{
> > +	if (IS_NOQUOTA(inode1) == IS_NOQUOTA(inode2))
> > +		return inode1 < inode2;
> > +	return IS_NOQUOTA(inode2);
> > +}
> 
> This seems kind of awful.  Is it what you were thinking of originally,
> Al, and could we live with it?
  Yeah, it's pretty ugly and also racy. I'm just now testing patches that
would get rid of I_MUTEX_QUOTA usage for filesystems (except GFS2) and
quota code. GFS2 could be certainly dealt with as well (at least by
introducing a new GFS2 internal lock) so this ugly code can go away.

								Honza