Re: [RFC][PATCH] slub: Check for page NULL before doing the node_matchcheck

[Christoph Lameter]

On Thu, 17 Jan 2013, Steven Rostedt wrote:

> >  	c->tid = next_tid(c->tid);
> > -	c->page = NULL;
> >  	c->freelist = NULL;
> > +	c->page = NULL;
>
> I'm assuming that this is to deal with the same CPU being able to touch
> the code?
>
> If so, it requires "barrier()". If this can affect other CPUs, then we
> need a smp_wmb() here, and smp_rmb() where it matters.

This is dealing with the same cpu being interrupted. Some of these
segments are in interrupt disable sections so they are not affected.

The above is a section where interrupts are enabled so it needs the
barriers.

> > @@ -2227,8 +2227,8 @@ redo:
> >  	if (unlikely(!node_match(page, node))) {
> >  		stat(s, ALLOC_NODE_MISMATCH);
> >  		deactivate_slab(s, page, c->freelist);
> > -		c->page = NULL;
> >  		c->freelist = NULL;
> > +		c->page = NULL;
> >  		goto new_slab;
> >  	}
> >

Interrupts are disabled so we do not need to change anything here.


> > @@ -2239,8 +2239,8 @@ redo:
> >  	 */
> >  	if (unlikely(!pfmemalloc_match(page, gfpflags))) {
> >  		deactivate_slab(s, page, c->freelist);
> > -		c->page = NULL;
> >  		c->freelist = NULL;
> > +		c->page = NULL;
> >  		goto new_slab;
> >  	}
> >

Ditto which leaves us with:

Index: linux/mm/slub.c
===================================================================
--- linux.orig/mm/slub.c	2013-01-17 15:49:57.417491975 -0600
+++ linux/mm/slub.c	2013-01-17 15:50:49.010287150 -0600
@@ -1993,8 +1993,9 @@ static inline void flush_slab(struct kme
 	deactivate_slab(s, c->page, c->freelist);

 	c->tid = next_tid(c->tid);
-	c->page = NULL;
 	c->freelist = NULL;
+	barrier();
+	c->page = NULL;
 }

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