Re: [PATCH 01/11] mm: page_alloc: handle MIGRATE_ISOLATE infree_pcppages_bulk()

[Mel Gorman]

On Fri, Nov 18, 2011 at 05:43:08PM +0100, Marek Szyprowski wrote:
> From: Michal Nazarewicz <mina86@xxxxxxxxxx>
> 
> If page is on PCP list while pageblock it belongs to gets isolated,
> the page's private still holds the old migrate type.  This means
> that free_pcppages_bulk() will put the page on a freelist of the
> old migrate type instead of MIGRATE_ISOLATE.
> 
> This commit changes that by explicitly checking whether page's
> pageblock's migrate type is MIGRATE_ISOLATE and if it is, overwrites
> page's private data.
> 
> Signed-off-by: Michal Nazarewicz <mina86@xxxxxxxxxx>
> Signed-off-by: Marek Szyprowski <m.szyprowski@xxxxxxxxxxx>
> ---
>  mm/page_alloc.c |   12 ++++++++++++
>  1 files changed, 12 insertions(+), 0 deletions(-)
> 
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 9dd443d..58d1a2e 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -628,6 +628,18 @@ static void free_pcppages_bulk(struct zone *zone, int count,
>  			page = list_entry(list->prev, struct page, lru);
>  			/* must delete as __free_one_page list manipulates */
>  			list_del(&page->lru);
> +
> +			/*
> +			 * When page is isolated in set_migratetype_isolate()
> +			 * function it's page_private is not changed since the
> +			 * function has no way of knowing if it can touch it.
> +			 * This means that when a page is on PCP list, it's
> +			 * page_private no longer matches the desired migrate
> +			 * type.
> +			 */
> +			if (get_pageblock_migratetype(page) == MIGRATE_ISOLATE)
> +				set_page_private(page, MIGRATE_ISOLATE);
> +

How much of a problem is this in practice? It's adding overhead to the
free path for what should be a very rare case which is undesirable. I
know we are already calling get_pageblock_migrate() when freeing
pages but it should be unnecessary to call it again. I'd go as far
to say that it would be preferable to drain the per-CPU lists after
you set pageblocks MIGRATE_ISOLATE. The IPIs also have overhead but it
will be incurred for the rare rather than the common case.

-- 
Mel Gorman
SUSE Labs
--
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/