Re: [PATCH] Add nid sanity on alloc_pages_node

[gurudas pai <gurudas.pai@oracle.com>]

> > > > > On Fri, 13 Jul 2007 10:45:07 +0800 Joe Jin <joe.jin@oracle.com> wrote:
> > > > >
> > > > > Something like this?
> > > > >
> > > > > --- a/mm/hugetlb.c~a
> > > > > +++ a/mm/hugetlb.c
> > > > > @@ -105,13 +105,20 @@ static void free_huge_page(struct page *
> > > > >  
> > > > >  static int alloc_fresh_huge_page(void)
> > > > >  {
> > > > > -	static int nid = 0;
> > > > > +	static int prev_nid;
> > > > > +	static DEFINE_SPINLOCK(nid_lock);
> > > > >  	struct page *page;
> > > > > -	page = alloc_pages_node(nid, htlb_alloc_mask|__GFP_COMP|__GFP_NOWARN,
> > > > > -					HUGETLB_PAGE_ORDER);
> > > > > -	nid = next_node(nid, node_online_map);
> > > > > +	int nid;
> > > > > +
> > > > > +	spin_lock(&nid_lock);
> > > > > +	nid = next_node(prev_nid, node_online_map);
> > > > >  	if (nid == MAX_NUMNODES)
> > > > >  		nid = first_node(node_online_map);
> > > > > +	prev_nid = nid;
> > > > > +	spin_unlock(&nid_lock);
> > > > > +
> > > > > +	page = alloc_pages_node(nid, htlb_alloc_mask|__GFP_COMP|__GFP_NOWARN,
> > > > > +					HUGETLB_PAGE_ORDER);
> > > > >  	if (page) {
> > > > >  		set_compound_page_dtor(page, free_huge_page);
> > > > >  		spin_lock(&hugetlb_lock);
> > > > > _
> > I think this will never get pages from node 0 ? Because nid = 
> > next_node(prev_node,node_online_map) and even if prev_node = 0, nid will 
> > become 1.
>
> It'll start out at node 1.  But it will visit the final node (which is less
> than MAX_NUMNODES) and will then advance onto the fist node (which can be
> > = 0).
>
> This code needs a bit of thought and testing for the non-numa case too
> please.  At the least, there might be optimisation opportunities.
I tested on non-numa machine. Your patch works fine.

Thanks,
-Guru


-
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/