Re: [PATCH V4 4/6] mm: reclaim MADV_FREE pages

From	Johannes Weiner <hannes@cmpxchg.org>
Newsgroups	linux.kernel
Subject	Re: [PATCH V4 4/6] mm: reclaim MADV_FREE pages
Date	2017-02-23 17:20 +0100
Message-ID	<te4hA-3BE-3@gated-at.bofh.it> (permalink)
References	<tdKiR-62X-7@gated-at.bofh.it> <tdKiT-62X-49@gated-at.bofh.it>
Organization	linux.* mail to news gateway

Show all headers | View raw

On Wed, Feb 22, 2017 at 10:50:42AM -0800, Shaohua Li wrote:
> @@ -1424,6 +1424,12 @@ static int try_to_unmap_one(struct page *page, struct vm_area_struct *vma,
>  				dec_mm_counter(mm, MM_ANONPAGES);
>  				rp->lazyfreed++;
>  				goto discard;
> +			} else if (!PageSwapBacked(page)) {
> +				/* dirty MADV_FREE page */
> +				set_pte_at(mm, address, pvmw.pte, pteval);
> +				ret = SWAP_DIRTY;
> +				page_vma_mapped_walk_done(&pvmw);
> +				break;
>  			}
>  
>  			if (swap_duplicate(entry) < 0) {
> @@ -1525,8 +1531,8 @@ int try_to_unmap(struct page *page, enum ttu_flags flags)
>  
>  	if (ret != SWAP_MLOCK && !page_mapcount(page)) {
>  		ret = SWAP_SUCCESS;
> -		if (rp.lazyfreed && !PageDirty(page))
> -			ret = SWAP_LZFREE;
> +		if (rp.lazyfreed && PageDirty(page))
> +			ret = SWAP_DIRTY;

Can this actually happen? If the page is dirty, ret should already be
SWAP_DIRTY, right? How would a dirty page get fully unmapped?

It seems to me rp.lazyfreed can be removed entirely now that we don't
have to identify the lazyfree case anymore. The failure case is much
easier to identify - all it takes is a single pte to be dirty.

> @@ -1118,8 +1120,10 @@ static unsigned long shrink_page_list(struct list_head *page_list,
>  		/*
>  		 * Anonymous process memory has backing store?
>  		 * Try to allocate it some swap space here.
> +		 * Lazyfree page could be freed directly
>  		 */
> -		if (PageAnon(page) && !PageSwapCache(page)) {
> +		if (PageAnon(page) && !PageSwapCache(page) &&
> +		    PageSwapBacked(page)) {

Nit: I'd do PageAnon(page) && PageSwapBacked(page) && !PageSwapCache()
since anon && swapbacked together describe the page type and swapcache
the state. Plus, anon && swapbacked go together everywhere else.

Otherwise, looks very straight-forward!

Thanks

Thread

[PATCH V4 0/6] mm: fix some MADV_FREE issues Shaohua Li <shli@fb.com> - 2017-02-22 20:00 +0100
  [PATCH V4 3/6] mm: move MADV_FREE pages into LRU_INACTIVE_FILE list Shaohua Li <shli@fb.com> - 2017-02-22 20:00 +0100
    Re: [PATCH V4 3/6] mm: move MADV_FREE pages into LRU_INACTIVE_FILE  list Johannes Weiner <hannes@cmpxchg.org> - 2017-02-23 17:10 +0100
      Re: [PATCH V4 3/6] mm: move MADV_FREE pages into LRU_INACTIVE_FILE  list Johannes Weiner <hannes@cmpxchg.org> - 2017-02-23 20:00 +0100
    Re: [PATCH V4 3/6] mm: move MADV_FREE pages into LRU_INACTIVE_FILE  list Minchan Kim <minchan@kernel.org> - 2017-02-24 03:00 +0100
      Re: [PATCH V4 3/6] mm: move MADV_FREE pages into LRU_INACTIVE_FILE  list Minchan Kim <minchan@kernel.org> - 2017-02-25 00:40 +0100
  [PATCH V4 5/6] mm: enable MADV_FREE for swapless system Shaohua Li <shli@fb.com> - 2017-02-22 20:00 +0100
  [PATCH V4 2/6] mm: don't assume anonymous pages have SwapBacked flag Shaohua Li <shli@fb.com> - 2017-02-22 20:00 +0100
  [PATCH V4 4/6] mm: reclaim MADV_FREE pages Shaohua Li <shli@fb.com> - 2017-02-22 20:00 +0100
    Re: [PATCH V4 4/6] mm: reclaim MADV_FREE pages Johannes Weiner <hannes@cmpxchg.org> - 2017-02-23 17:20 +0100
    Re: [PATCH V4 4/6] mm: reclaim MADV_FREE pages Minchan Kim <minchan@kernel.org> - 2017-02-24 03:20 +0100
      Re: [PATCH V4 4/6] mm: reclaim MADV_FREE pages Johannes Weiner <hannes@cmpxchg.org> - 2017-02-24 16:50 +0100
        Re: [PATCH V4 4/6] mm: reclaim MADV_FREE pages Minchan Kim <minchan@kernel.org> - 2017-02-25 00:30 +0100

csiph-web