Re: [PATCH] mm: vmscan: Add warn on inadvertently reclaiming mapped page

Re: [PATCH] mm: vmscan: Add warn on inadvertently reclaiming mapped page

[Yang Shi]

On 5/26/19 2:23 PM, Hillf Danton wrote:
> In the function isolate_lru_pages(), we check scan_control::may_unmap and set
> isolation mode accordingly in order to not isolate from the lru list any page
> that does not match the isolation mode. For example, we should skip all sill
> mapped pages if isolation mode is set to be ISOLATE_UNMAPPED.
>
> So complain, while scanning the isolated pages, about the very unlikely event
> that we hit a mapped page that we should never have isolated. Note no change
> is added in the current scanning behavior without VM debug configured.
>
> And cut off one line of comment that goes stale.

Looks good to me. Reviewed-by: Yang Shi <yang.shi@linux.alibaba.com>

>
> Cc: Mel Gorman <mgorman@techsingularity.net>
> Cc: Johannes Weiner <hannes@cmpxchg.org>
> Cc: Minchan Kim <minchan@kernel.org>
> Cc: Yang Shi <yang.shi@linux.alibaba.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Signed-off-by: Hillf Danton <hdanton@sina.com>
> ---
>   mm/vmscan.c | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index d9c3e87..799ad9e 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1134,8 +1134,10 @@ static unsigned long shrink_page_list(struct list_head *page_list,
>   		if (unlikely(!page_evictable(page)))
>   			goto activate_locked;
>   
> -		if (!sc->may_unmap && page_mapped(page))
> +		if (!sc->may_unmap && page_mapped(page)) {
> +			VM_WARN_ON(true);
>   			goto keep_locked;
> +		}
>   
>   		/* Double the slab pressure for mapped and swapcache pages */
>   		if ((page_mapped(page) || PageSwapCache(page)) &&
> @@ -1632,7 +1634,6 @@ static __always_inline void update_lru_sizes(struct lruvec *lruvec,
>    * @dst:	The temp list to put pages on to.
>    * @nr_scanned:	The number of pages that were scanned.
>    * @sc:		The scan_control struct for this reclaim session
> - * @mode:	One of the LRU isolation modes
>    * @lru:	LRU list id for isolating
>    *
>    * returns how many pages were moved onto *@dst.
> --

Re: [PATCH] mm: vmscan: Add warn on inadvertently reclaiming mapped page

[Andrew Morton]

On Sun, 26 May 2019 14:23:53 +0800 Hillf Danton <hdanton@sina.com> wrote:

> In the function isolate_lru_pages(), we check scan_control::may_unmap and set
> isolation mode accordingly in order to not isolate from the lru list any page
> that does not match the isolation mode. For example, we should skip all sill
> mapped pages if isolation mode is set to be ISOLATE_UNMAPPED.
> 
> So complain, while scanning the isolated pages, about the very unlikely event
> that we hit a mapped page that we should never have isolated. Note no change
> is added in the current scanning behavior without VM debug configured.

The patch is inoffensive enough, but one wonders what inspired it.  Do
you have reason to believe that this will trigger?

Re: [PATCH] mm: vmscan: Add warn on inadvertently reclaiming mapped page

[Johannes Weiner]

On Tue, May 28, 2019 at 09:22:57PM -0700, Andrew Morton wrote:
> On Sun, 26 May 2019 14:23:53 +0800 Hillf Danton <hdanton@sina.com> wrote:
> 
> > In the function isolate_lru_pages(), we check scan_control::may_unmap and set
> > isolation mode accordingly in order to not isolate from the lru list any page
> > that does not match the isolation mode. For example, we should skip all sill
> > mapped pages if isolation mode is set to be ISOLATE_UNMAPPED.
> > 
> > So complain, while scanning the isolated pages, about the very unlikely event
> > that we hit a mapped page that we should never have isolated. Note no change
> > is added in the current scanning behavior without VM debug configured.
> 
> The patch is inoffensive enough, but one wonders what inspired it.  Do
> you have reason to believe that this will trigger?

I don't think this patch makes sense.

There isn't anything preventing someone from mapping the page between
the LRU isolation and shrink_page_list(). The isolation filter is only
an optimization to reduce unnecessary LRU churn, not a guarantee.