Re: [PATCH v2] mm/vmscan: count layzfree pages and fix nr_isolated_* mismatch

From: Johannes Weiner <hannes@cmpxchg.org>
To: Jaewon Kim <jaewon31.kim@samsung.com>
Cc: minchan@kernel.org, mgorman@suse.de, m.szyprowski@samsung.com,
	mina86@mina86.com, shli@fb.com, akpm@linux-foundation.org,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	jaewon31.kim@gmail.com, ytk.lee@samsung.com
Subject: Re: [PATCH v2] mm/vmscan: count layzfree pages and fix nr_isolated_* mismatch
Date: Wed, 22 Apr 2020 09:07:51 -0400	[thread overview]
Message-ID: <20200422130751.GD358439@cmpxchg.org> (raw)
In-Reply-To: <20200422084815.21913-1-jaewon31.kim@samsung.com>

On Wed, Apr 22, 2020 at 05:48:15PM +0900, Jaewon Kim wrote:
> @@ -1295,11 +1295,15 @@ static unsigned long shrink_page_list(struct list_head *page_list,
>  		 */
>  		if (page_mapped(page)) {
>  			enum ttu_flags flags = ttu_flags | TTU_BATCH_FLUSH;
> +			bool lazyfree = PageAnon(page) && !PageSwapBacked(page);
>  
>  			if (unlikely(PageTransHuge(page)))
>  				flags |= TTU_SPLIT_HUGE_PMD;
> +
>  			if (!try_to_unmap(page, flags)) {
>  				stat->nr_unmap_fail += nr_pages;
> +				if (lazyfree && PageSwapBacked(page))

This looks pretty strange, until you remember that try_to_unmap()
could SetPageSwapbacked again.

This might be more obvious?

			was_swapbacked = PageSwapBacked(page);
			if (!try_to_unmap(page, flags)) {
				stat->nr_unmap_fail += nr_pages;
				if (!was_swapbacked && PageSwapBacked(page))
> +					stat->nr_lazyfree_fail += nr_pages;
>  				goto activate_locked;

Or at least was_lazyfree.

> @@ -1491,8 +1495,8 @@ unsigned long reclaim_clean_pages_from_list(struct zone *zone,
>  		.priority = DEF_PRIORITY,
>  		.may_unmap = 1,
>  	};
> -	struct reclaim_stat dummy_stat;
> -	unsigned long ret;
> +	struct reclaim_stat stat;
> +	unsigned long reclaimed;

nr_reclaimed would be better.

I also prefer keeping dummy_stat, since that's still what it is.