On Tue, 21 Oct 2008 17:57:35 +0900, KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:
> On Tue, 21 Oct 2008 16:35:09 +0800
> Li Zefan <lizf@cn.fujitsu.com> wrote:
> 
> > KAMEZAWA Hiroyuki wrote:
> > > On Tue, 21 Oct 2008 15:21:07 +0800
> > > Li Zefan <lizf@cn.fujitsu.com> wrote:
> > >> dmesg is attached.
> > >>
> > > Thanks....I think I caught some. (added Mel Gorman to CC:)
> > > 
> > > NODE_DATA(nid)->spanned_pages just means sum of zone->spanned_pages in node.
> > > 
> > > So, If there is a hole between zone, node->spanned_pages doesn't mean
> > > length of node's memmap....(then, some hole can be skipped.)
> > > 
> > > OMG....Could you try this ? 
> > > 
> > 
> > No luck, the same bug still exists. :(
> > 
> This is a little fixed one..
> 
I can reproduce a similar problem(hang on boot) on 2.6.27-git9,
but this patch doesn't help either on my environment...

I attach a console log(I've not seen NULL pointer dereference yet).


Daisuke Nishimura.

> please..
> -Kame
> ==
> NODE_DATA(nid)->node_spanned_pages doesn't means width of node's memory.
> 
> alloc_node_page_cgroup() misunderstand it. This patch tries to use
> the same algorithm as alloc_node_mem_map() for allocating page_cgroup()
> for node.
> 
> Changelog:
>  - fixed range of initialization loop.
> 
> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> 
>  mm/page_cgroup.c |   19 +++++++++++++++----
>  1 file changed, 15 insertions(+), 4 deletions(-)
> 
> Index: linux-2.6.27/mm/page_cgroup.c
> ===================================================================
> --- linux-2.6.27.orig/mm/page_cgroup.c
> +++ linux-2.6.27/mm/page_cgroup.c
> @@ -9,6 +9,8 @@
>  static void __meminit
>  __init_page_cgroup(struct page_cgroup *pc, unsigned long pfn)
>  {
> +	if (!pfn_valid(pfn))
> +		return;
>  	pc->flags = 0;
>  	pc->mem_cgroup = NULL;
>  	pc->page = pfn_to_page(pfn);
> @@ -41,10 +43,18 @@ static int __init alloc_node_page_cgroup
>  {
>  	struct page_cgroup *base, *pc;
>  	unsigned long table_size;
> -	unsigned long start_pfn, nr_pages, index;
> +	unsigned long start, end, start_pfn, nr_pages, index;
>  
> +	/*
> +	 * Instead of allocating page_cgroup for [start, end)
> +	 * We allocate page_cgroup to the same size of mem_map.
> +	 * See page_alloc.c::alloc_node_mem_map()
> +	 */
>  	start_pfn = NODE_DATA(nid)->node_start_pfn;
> -	nr_pages = NODE_DATA(nid)->node_spanned_pages;
> +	start = start_pfn & ~(MAX_ORDER_NR_PAGES - 1);
> +	end = start_pfn	+ NODE_DATA(nid)->node_spanned_pages;
> +	end = ALIGN(end, MAX_ORDER_NR_PAGES);
> +	nr_pages = end - start;
>  
>  	table_size = sizeof(struct page_cgroup) * nr_pages;
>  
> @@ -52,11 +62,12 @@ static int __init alloc_node_page_cgroup
>  			table_size, PAGE_SIZE, __pa(MAX_DMA_ADDRESS));
>  	if (!base)
>  		return -ENOMEM;
> +
>  	for (index = 0; index < nr_pages; index++) {
>  		pc = base + index;
> -		__init_page_cgroup(pc, start_pfn + index);
> +		__init_page_cgroup(pc, start + index);
>  	}
> -	NODE_DATA(nid)->node_page_cgroup = base;
> +	NODE_DATA(nid)->node_page_cgroup = base + start_pfn - start;
>  	total_usage += table_size;
>  	return 0;
>  }
>