Re: [patch] mm: vmscan fix mapping use after free

[Mel Gorman <mel@csn.ul.ie>]

On Tue, Jun 15, 2010 at 03:43:44PM +1000, Nick Piggin wrote:
> On Tue, Jun 15, 2010 at 03:12:42PM +1000, Nick Piggin wrote:
> > > Once that page is unlocked, we can't touch *mapping - its inode can be
> > > concurrently reclaimed.  Although I guess the technique in
> > > handle_write_error() can be reused.
> > 
> > Nasty. That guy needs to be using lock_page_nosync().
> --
> 
> Need lock_page_nosync here because we have no reference to the mapping when
> taking the page lock.
> 
> Signed-off-by: Nick Piggin <npiggin@suse.de>
> 

Thanks, I've picked this up and merged it into the series and removed
the "hand grenade" comment.

> ---
>  mm/vmscan.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> Index: linux-2.6/mm/vmscan.c
> ===================================================================
> --- linux-2.6.orig/mm/vmscan.c
> +++ linux-2.6/mm/vmscan.c
> @@ -296,7 +296,7 @@ static int may_write_to_queue(struct bac
>  static void handle_write_error(struct address_space *mapping,
>  				struct page *page, int error)
>  {
> -	lock_page(page);
> +	lock_page_nosync(page);
>  	if (page_mapping(page) == mapping)
>  		mapping_set_error(mapping, error);
>  	unlock_page(page);
> 

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab