Re: [PATCH 2/2] Make is_mem_section_removable more conformable with offlining code v3

[Wu Fengguang <fengguang.wu@intel.com>]

On Fri, Sep 03, 2010 at 07:42:13PM +0800, Michal Hocko wrote:

> +/*
> + * A free or LRU pages block are removable
> + * Do not use MIGRATE_MOVABLE because it can be insufficient and
> + * other MIGRATE types are tricky.
> + * Do not hold zone->lock as this is used from user space by the
> + * sysfs interface.
> + */
> +bool is_page_removable(struct page *page)
> +{
> +	int page_block = 1 << pageblock_order;
> +
> +	/* All pages from the MOVABLE zone are movable */
> +	if (zone_idx(page_zone(page)) == ZONE_MOVABLE)
> +		return true;
> +
> +	while (page_block > 0) {
> +		int order = 0;
> +
> +		if (pfn_valid_within(page_to_pfn(page))) {
> +			if (!page_count(page) && PageBuddy(page)) {

PageBuddy() is true only for the head page and false for all tail
pages. So when is_page_removable() is given a random 4k page
(get_any_page() will exactly do that), the above test is not enough.

It's recommended to reuse is_free_buddy_page(). (Need to do some
cleanup work first: remove the "#ifdef CONFIG_MEMORY_FAILURE" and
abstract out an __is_free_buddy_page() that takes no lock.)

> @@ -5277,14 +5277,11 @@ int set_migratetype_isolate(struct page *page)
>  	struct memory_isolate_notify arg;
>  	int notifier_ret;
>  	int ret = -EBUSY;
> -	int zone_idx;
>  
>  	zone = page_zone(page);
> -	zone_idx = zone_idx(zone);
>  
>  	spin_lock_irqsave(&zone->lock, flags);
> -	if (get_pageblock_migratetype(page) == MIGRATE_MOVABLE ||
> -	    zone_idx == ZONE_MOVABLE) {
> +	if (is_page_removable(page)) {
>  		ret = 0;
>  		goto out;

The above check only applies to the first page in the page block.
The following "if (!page_count(curr_page) || PageLRU(curr_page))"
check in the same function should be converted too (and that's another
reason to use __is_free_buddy_page(): it will be tested for every 4k
pages, including both the head and tail pages).

Thanks,
Fengguang