Re: [v4 6/6] mm/memory_hotplug: optimize memory hotplug

[Michal Hocko]

On Thu 15-02-18 11:59:20, Pavel Tatashin wrote:
[...]
> @@ -260,21 +260,12 @@ static int __meminit __add_section(int nid, unsigned long phys_start_pfn,
>  		return ret;
>  
>  	/*
> -	 * Make all the pages reserved so that nobody will stumble over half
> -	 * initialized state.
> -	 * FIXME: We also have to associate it with a node because page_to_nid
> -	 * relies on having page with the proper node.
> +	 * The first page in every section holds node id, this is because we
> +	 * will need it in online_pages().
>  	 */
> -	for (i = 0; i < PAGES_PER_SECTION; i++) {
> -		unsigned long pfn = phys_start_pfn + i;
> -		struct page *page;
> -		if (!pfn_valid(pfn))
> -			continue;
> -
> -		page = pfn_to_page(pfn);
> -		set_page_node(page, nid);
> -		SetPageReserved(page);
> -	}
> +	page = pfn_to_page(phys_start_pfn);
> +	mm_zero_struct_page(page);
> +	set_page_node(page, nid);

I really dislike this part. It is just too subtle assumption. We can
safely store the node id into memory_block and push it down the way to
online. Or am I missing something?

Btw. the rest of the series seem good as is so I would go with it and
keep this last patch aparat and make sure to do it properly rather than
add more hacks.
  
>  	if (!want_memblock)
>  		return 0;
-- 
Michal Hocko
SUSE Labs