Re: [PATCH 1/2] split inode_wb_list_lock into bdi_writeback.list_lock

[Wu Fengguang <fengguang.wu@intel.com>]

On Wed, Apr 27, 2011 at 05:42:09AM +0800, Andrew Morton wrote:
> On Tue, 26 Apr 2011 22:42:19 +0800
> Wu Fengguang <fengguang.wu@intel.com> wrote:
> 
> > @@ -55,13 +55,16 @@ EXPORT_SYMBOL(I_BDEV);
> >  static void bdev_inode_switch_bdi(struct inode *inode,
> >  			struct backing_dev_info *dst)
> >  {
> > -	spin_lock(&inode_wb_list_lock);
> > +	struct backing_dev_info *old = inode->i_data.backing_dev_info;
> > +
> > +	bdi_lock_two(&old->wb, &dst->wb);
> >  	spin_lock(&inode->i_lock);
> >  	inode->i_data.backing_dev_info = dst;
> >  	if (inode->i_state & I_DIRTY)
> >  		list_move(&inode->i_wb_list, &dst->wb.b_dirty);
> >  	spin_unlock(&inode->i_lock);
> > -	spin_unlock(&inode_wb_list_lock);
> > +	spin_unlock(&old->wb.list_lock);
> > +	spin_unlock(&dst->wb.list_lock);
> >  }
> 
> Has this patch been well tested under lockdep?

Yes, it runs OK on concurrent dd and dd+tar workloads over all major
filesystems, including NFS. Lockdep is always enabled in my kernels,
no warnings are found in dmesg.

Thanks,
Fengguang