Re: [PATCH 07/15] writeback: split inode_wb_list_lock into bdi_writeback.list_lock

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

On Wed, Jun 08, 2011 at 07:03:19AM +0800, Andrew Morton wrote:
> On Wed, 08 Jun 2011 05:32:43 +0800
> Wu Fengguang <fengguang.wu@intel.com> wrote:
> 
> >  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;
> > +
> > +	if (unlikely(dst == old))		/* deadlock avoidance */
> > +		return;
> 
> Why does this occur?

That's a fix from Hugh Dickins:

        Yesterday's mmotm hangs at startup, and with lockdep it reports:
        BUG: spinlock recursion on CPU#1, blkid/284 - with bdi_lock_two()
        called from bdev_inode_switch_bdi() in the backtrace.  It appears
        that this function is sometimes called with new the same as old.

The problem becomes clear when looking at bdi_lock_two(), which will
immediately deadlock itself if called with (wb1 == wb2):

  void bdi_lock_two(struct bdi_writeback *wb1, struct bdi_writeback *wb2)
  {
          if (wb1 < wb2) {
                  spin_lock(&wb1->list_lock);                                                                         
                  spin_lock_nested(&wb2->list_lock, 1);
          } else {
                  spin_lock(&wb2->list_lock);
                  spin_lock_nested(&wb1->list_lock, 1);
          }
  }

Thanks,
Fengguang