On Wed, Mar 07, 2007 at 03:34:00AM -0800, Andrew Morton wrote:
quoted text
> On Wed, 7 Mar 2007 03:20:38 -0800 Andrew Morton <akpm@linux-foundation.org> wrote:
> 
> > 
> > > ===================================================================
> > > --- linux-2.6.orig/mm/memory.c
> > > +++ linux-2.6/mm/memory.c
> > > @@ -1676,6 +1676,17 @@ gotten:
> > >  unlock:
> > >  	pte_unmap_unlock(page_table, ptl);
> > >  	if (dirty_page) {
> > > +		/*
> > > +		 * Yes, Virginia, this is actually required to prevent a race
> > > +		 * with clear_page_dirty_for_io() from clearing the page dirty
> > > +		 * bit after it clear all dirty ptes, but before a racing
> > > +		 * do_wp_page installs a dirty pte.
> > > +		 *
> > > +		 * do_fault is protected similarly by holding the page lock
> > > +		 * after the dirty pte is installed.
> > > +		 */
> > > +		lock_page(dirty_page);
> > > +		unlock_page(dirty_page);
> > >  		set_page_dirty_balance(dirty_page);
> > >  		put_page(dirty_page);
> > 
> > Yes, I think that'll plug it.  A wait_on_page_locked() should suffice.
> 
> Or will it?  Suppose after the unlock_page() a _second_
> clear_page_dirty_for_io() gets run - the same thing happens?
> 
> Extending the lock_page() coverage around the set_page_dirty() would
> prevent that.
> 
> I guess not needed - the second clear_page_dirty_for_io() will have cleaned the
> pte.

Yeah, all we need to do is keep page faults out of that little window
in clear_page_dirty_for_io() where I stuck the comment.

-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/