Re: [alternative-merged] mm-fix-is_pinnable_page-against-on-cma-page.patch removed from -mm tree

[Minchan Kim <minchan@kernel.org>]

On Tue, May 17, 2022 at 12:48:00PM -0700, Andrew Morton wrote:
> 
> The quilt patch titled
>      Subject: mm: fix is_pinnable_page against on cma page
> has been removed from the -mm tree.  Its filename was
>      mm-fix-is_pinnable_page-against-on-cma-page.patch
> 
> This patch was dropped because an alternative patch was merged

Hi Andrew,

I didn't see what an alternative patch was merged.

What I observed from last discussion[1] was that let's use
READ_ONCE in __get_pfnblock_flags_mask and remove READ_ONCE
in is_pinnable_page.

Please point me out if you merged other alternative. Otherwise,
I am happy to post new version based on the comment of [1]

Let me know. 

[1] https://lore.kernel.org/all/b6eef200-43d1-7913-21ed-176b05fcb4fe@nvidia.com/

> 
> ------------------------------------------------------
> From: Minchan Kim <minchan@kernel.org>
> Subject: mm: fix is_pinnable_page against on cma page
> 
> Pages on CMA area could have MIGRATE_ISOLATE as well as MIGRATE_CMA so
> current is_pinnable_page could miss CMA pages which has MIGRATE_ ISOLATE. 
> It ends up pinning CMA pages as longterm at pin_user_pages APIs so CMA
> allocation keep failed until the pin is released.
> 
>      CPU 0                                   CPU 1 - Task B
> 
> cma_alloc
> alloc_contig_range
>                                         pin_user_pages_fast(FOLL_LONGTERM)
> change pageblock as MIGRATE_ISOLATE
>                                         internal_get_user_pages_fast
>                                         lockless_pages_from_mm
>                                         gup_pte_range
>                                         try_grab_folio
>                                         is_pinnable_page
>                                           return true;
>                                         So, pinned the page successfully.
> page migration failure with pinned page
>                                         ..
>                                         .. After 30 sec
>                                         unpin_user_page(page)
> 
> CMA allocation succeeded after 30 sec.
> 
> The CMA allocation path protects the migration type change race using
> zone->lock but what GUP path need to know is just whether the page is on
> CMA area or not rather than exact migration type.  Thus, we don't need
> zone->lock but just checks migration type in either of (MIGRATE_ISOLATE
> and MIGRATE_CMA).
> 
> Adding the MIGRATE_ISOLATE check in is_pinnable_page could cause rejecting
> of pinning pages on MIGRATE_ISOLATE pageblocks even though it's neither
> CMA nor movable zone if the page is temporarily unmovable.  However, such
> a migration failure by unexpected temporal refcount holding is general
> issue, not only come from MIGRATE_ISOLATE and the MIGRATE_ISOLATE is also
> transient state like other temporal elevated refcount problem.
> 
> [akpm@linux-foundation.org: MIGRATE_CMA and MIGRATE_ISOLATE are not bitwise, per Minchan]
> [minchan@kernel.org: restore __mt definition]
> [minchan@kernel.org: changes per Paul and John]
>   Link: https://lkml.kernel.org/r/20220512204143.3961150-1-minchan@kernel.org
> Link: https://lkml.kernel.org/r/20220510211743.95831-1-minchan@kernel.org
> Signed-off-by: Minchan Kim <minchan@kernel.org>
> Cc: David Hildenbrand <david@redhat.com>
> Cc: "Paul E . McKenney" <paulmck@kernel.org>
> Cc: John Hubbard <jhubbard@nvidia.com>
> Cc: John Dias <joaodias@google.com>
> Cc: Stephen Rothwell <sfr@canb.auug.org.au>
> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
> ---
> 
>  include/linux/mm.h |   17 +++++++++++++++--
>  1 file changed, 15 insertions(+), 2 deletions(-)
> 
> --- a/include/linux/mm.h~mm-fix-is_pinnable_page-against-on-cma-page
> +++ a/include/linux/mm.h
> @@ -1625,8 +1625,21 @@ static inline bool page_needs_cow_for_dm
>  #ifdef CONFIG_MIGRATION
>  static inline bool is_pinnable_page(struct page *page)
>  {
> -	return !(is_zone_movable_page(page) || is_migrate_cma_page(page)) ||
> -		is_zero_pfn(page_to_pfn(page));
> +#ifdef CONFIG_CMA
> +	/*
> +	 * Defend against future compiler LTO features, or code refactoring
> +	 * that inlines the above function, by forcing a single read. Because,
> +	 * this routine races with set_pageblock_migratetype(), and we want to
> +	 * avoid reading zero, when actually one or the other flags was set.
> +	 */
> +	int __mt = get_pageblock_migratetype(page);
> +	int mt = __READ_ONCE(__mt);
> +
> +	if (mt == MIGRATE_CMA || mt == MIGRATE_ISOLATE)
> +		return false;
> +#endif
> +
> +	return !(is_zone_movable_page(page) || is_zero_pfn(page_to_pfn(page)));
>  }
>  #else
>  static inline bool is_pinnable_page(struct page *page)
> _
> 
> Patches currently in -mm which might be from minchan@kernel.org are
> 
> mm-dont-be-stuck-to-rmap-lock-on-reclaim-path.patch
>