Re: [PATCH 15/49] writeback: move backing_dev_info->wb_lock and ->worklist into bdi_writeback

[Tejun Heo <tj@kernel.org>]

Hello, Jan.

On Mon, Apr 20, 2015 at 05:32:24PM +0200, Jan Kara wrote:
> > @@ -454,9 +451,9 @@ EXPORT_SYMBOL(bdi_init);
> >  
> >  void bdi_destroy(struct backing_dev_info *bdi)
> >  {
> > -	bdi_wb_shutdown(bdi);
> > -
> > -	WARN_ON(!list_empty(&bdi->work_list));
> > +	/* make sure nobody finds us on the bdi_list anymore */
> > +	bdi_remove_from_list(bdi);
> > +	wb_shutdown(&bdi->wb);
> >  
> >  	if (bdi->dev) {
> >  		bdi_debug_unregister(bdi);
>   But if someone ends up calling bdi_destroy() on unregistered bdi,
> bdi_remove_from_list() will be corrupting memory, won't it? And if I

bdi_init() does INIT_LIST_HEAD() on it, so it should be fine, no?

> remember right there were some corner cases where this really happened.
> Previously we were careful and checked WB_registered. I guess we could
> check for !list_empty(&bdi->bdi_list) and also reinit bdi_list in
> bdi_remove_from_list() after synchronize_rcu_expedited().

But we can't call bdi_destroy() more than once no matter what.  We'd
be doing double frees.

Thanks.

-- 
tejun

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>