linux-kernel - Re: [PATCH] mm: remove redundant compound

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: <808194f6-e5d0-59e4-ffd5-2c77a907232c@suse.cz>
Date:   Fri, 13 Aug 2021 16:02:33 +0200
From:   Vlastimil Babka <vbabka@...e.cz>
To:     Muchun Song <songmuchun@...edance.com>, akpm@...ux-foundation.org,
        willy@...radead.org, william.kucharski@...cle.com,
        kirill.shutemov@...ux.intel.com, dhowells@...hat.com,
        hannes@...xchg.org
Cc:     linux-kernel@...r.kernel.org, linux-mm@...ck.org
Subject: Re: [PATCH] mm: remove redundant compound_head() calling

On 8/11/21 12:14 PM, Muchun Song wrote:
> There is a READ_ONCE() in the macro of compound_head(), which will
> prevent compiler from optimizing the code when there are more than
> once calling of it in a function. Remove the redundant calling of
> compound_head() from page_to_index() and page_add_file_rmap() for
> better code generation.
> 
> Signed-off-by: Muchun Song <songmuchun@...edance.com>

Seems to be compatible with folio/for-next and not made redundant by that (yet?
didn't check the branches planned for future versions), so OK. But long-term I'd
expect these optimizations to be obsoleted by the folio work.

> ---
>  include/linux/pagemap.h | 7 +++----
>  mm/rmap.c               | 6 ++++--
>  2 files changed, 7 insertions(+), 6 deletions(-)
> 
> diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
> index 79ec90e97e94..03b9a957ef10 100644
> --- a/include/linux/pagemap.h
> +++ b/include/linux/pagemap.h
> @@ -608,18 +608,17 @@ static inline struct page *read_mapping_page(struct address_space *mapping,
>   */
>  static inline pgoff_t page_to_index(struct page *page)
>  {
> -	pgoff_t pgoff;
> +	struct page *head;
>  
>  	if (likely(!PageTransTail(page)))
>  		return page->index;
>  
> +	head = compound_head(page);
>  	/*
>  	 *  We don't initialize ->index for tail pages: calculate based on
>  	 *  head page
>  	 */
> -	pgoff = compound_head(page)->index;
> -	pgoff += page - compound_head(page);
> -	return pgoff;
> +	return head->index + page - head;
>  }
>  
>  extern pgoff_t hugetlb_basepage_index(struct page *page);
> diff --git a/mm/rmap.c b/mm/rmap.c
> index 232494888628..2e216835f07c 100644
> --- a/mm/rmap.c
> +++ b/mm/rmap.c
> @@ -1231,11 +1231,13 @@ void page_add_file_rmap(struct page *page, bool compound)
>  						nr_pages);
>  	} else {
>  		if (PageTransCompound(page) && page_mapping(page)) {
> +			struct page *head = compound_head(page);
> +
>  			VM_WARN_ON_ONCE(!PageLocked(page));
>  
> -			SetPageDoubleMap(compound_head(page));
> +			SetPageDoubleMap(head);
>  			if (PageMlocked(page))
> -				clear_page_mlock(compound_head(page));
> +				clear_page_mlock(head);
>  		}
>  		if (!atomic_inc_and_test(&page->_mapcount))
>  			goto out;
>