linux-kernel - Re: [PATCH 6/7] mm/gup: Accelerate thp gup even for "pages != NULL"

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <d8c76484-1030-44a3-b148-7e69fa84243a@lucifer.local>
Date:   Sat, 17 Jun 2023 21:27:22 +0100
From:   Lorenzo Stoakes <lstoakes@...il.com>
To:     Peter Xu <peterx@...hat.com>
Cc:     linux-kernel@...r.kernel.org, linux-mm@...ck.org,
        Matthew Wilcox <willy@...radead.org>,
        Andrea Arcangeli <aarcange@...hat.com>,
        John Hubbard <jhubbard@...dia.com>,
        Mike Rapoport <rppt@...nel.org>,
        David Hildenbrand <david@...hat.com>,
        Vlastimil Babka <vbabka@...e.cz>,
        "Kirill A . Shutemov" <kirill@...temov.name>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Mike Kravetz <mike.kravetz@...cle.com>,
        James Houghton <jthoughton@...gle.com>,
        Hugh Dickins <hughd@...gle.com>
Subject: Re: [PATCH 6/7] mm/gup: Accelerate thp gup even for "pages != NULL"

On Tue, Jun 13, 2023 at 05:53:45PM -0400, Peter Xu wrote:
> The acceleration of THP was done with ctx.page_mask, however it'll be
> ignored if **pages is non-NULL.
>
> The old optimization was introduced in 2013 in 240aadeedc4a ("mm:
> accelerate mm_populate() treatment of THP pages").  It didn't explain why
> we can't optimize the **pages non-NULL case.  It's possible that at that
> time the major goal was for mm_populate() which should be enough back then.
>
> Optimize thp for all cases, by properly looping over each subpage, doing
> cache flushes, and boost refcounts / pincounts where needed in one go.
>
> This can be verified using gup_test below:
>
>   # chrt -f 1 ./gup_test -m 512 -t -L -n 1024 -r 10
>
> Before:    13992.50 ( +-8.75%)
> After:       378.50 (+-69.62%)
>
> Signed-off-by: Peter Xu <peterx@...hat.com>
> ---
>  mm/gup.c | 36 +++++++++++++++++++++++++++++-------
>  1 file changed, 29 insertions(+), 7 deletions(-)
>
> diff --git a/mm/gup.c b/mm/gup.c
> index a2d1b3c4b104..cdabc8ea783b 100644
> --- a/mm/gup.c
> +++ b/mm/gup.c
> @@ -1210,16 +1210,38 @@ static long __get_user_pages(struct mm_struct *mm,
>  			goto out;
>  		}
>  next_page:
> -		if (pages) {
> -			pages[i] = page;
> -			flush_anon_page(vma, page, start);
> -			flush_dcache_page(page);
> -			ctx.page_mask = 0;
> -		}
> -
>  		page_increm = 1 + (~(start >> PAGE_SHIFT) & ctx.page_mask);
>  		if (page_increm > nr_pages)
>  			page_increm = nr_pages;
> +
> +		if (pages) {
> +			struct page *subpage;
> +			unsigned int j;
> +
> +			/*
> +			 * This must be a large folio (and doesn't need to
> +			 * be the whole folio; it can be part of it), do
> +			 * the refcount work for all the subpages too.
> +			 * Since we already hold refcount on the head page,
> +			 * it should never fail.
> +			 *
> +			 * NOTE: here the page may not be the head page
> +			 * e.g. when start addr is not thp-size aligned.
> +			 */
> +			if (page_increm > 1)
> +				WARN_ON_ONCE(
> +				    try_grab_folio(compound_head(page),
> +						   page_increm - 1,
> +						   foll_flags) == NULL);

I'm not sure this should be warning but otherwise ignoring this returning
NULL?  This feels like a case that could come up in realtiy,
e.g. folio_ref_try_add_rcu() fails, or !folio_is_longterm_pinnable().

Side-note: I _hate_ the semantics of GUP such that try_grab_folio()
(invoked, other than for huge page cases, by the GUP-fast logic) will
explicitly fail if neither FOLL_GET or FOLL_PIN are specified,
differentiating it from try_grab_page() in this respect.

This is a side-note and not relevant here, as all callers to
__get_user_pages() either explicitly set FOLL_GET if not set by user (in
__get_user_pages_locked()) or don't set pages (e.g. in
faultin_vma_page_range())

> +
> +			for (j = 0; j < page_increm; j++) {
> +				subpage = nth_page(page, j);
> +				pages[i+j] = subpage;
> +				flush_anon_page(vma, subpage, start + j * PAGE_SIZE);
> +				flush_dcache_page(subpage);
> +			}
> +		}
> +
>  		i += page_increm;
>  		start += page_increm * PAGE_SIZE;
>  		nr_pages -= page_increm;
> --
> 2.40.1
>