Received-SPF: softfail (google.com: domain of transitioning linux-kernel-owner@vger.kernel.org does not designate 23.128.96.19 as permitted sender) client-ip=23.128.96.19;
Message-ID: <d29d5b05-ad7e-d459-be58-f183f32fe8a9@redhat.com>
Date:   Wed, 30 Mar 2022 15:16:17 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.6.2
Subject: Re: [PATCH v2 2/2] mm: fix contiguous memmap assumptions about
 alloc/free pages
Content-Language: en-US
To:     Chen Wandun <chenwandun@huawei.com>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, akpm@linux-foundation.org,
        willy@infradead.org
References: <20220330102534.1053240-1-chenwandun@huawei.com>
 <20220330102534.1053240-3-chenwandun@huawei.com>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat
In-Reply-To: <20220330102534.1053240-3-chenwandun@huawei.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 30.03.22 12:25, Chen Wandun wrote:
> It isn't true for only SPARSEMEM configs to assume that a compound page
> has virtually contiguous page structs, so use nth_page to iterate each
> page.

I really don't see how that is currently the case. The buddy deals with
order < MAX_ORDER and we know that these always fall into a single
memory section.

IOW, this patch here would result in overhead where it's not required to
have that overhead.

What am I missing and which scenario are we fixing?

> 
> Signed-off-by: Chen Wandun <chenwandun@huawei.com>
> ---
>  mm/page_alloc.c | 16 +++++++++-------
>  1 file changed, 9 insertions(+), 7 deletions(-)
> 
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 855211dea13e..758d8f069b32 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -721,7 +721,7 @@ static void prep_compound_head(struct page *page, unsigned int order)
>  
>  static void prep_compound_tail(struct page *head, int tail_idx)
>  {
> -	struct page *p = head + tail_idx;
> +	struct page *p = nth_page(head, tail_idx);
>  
>  	p->mapping = TAIL_MAPPING;
>  	set_compound_head(p, head);
> @@ -1199,10 +1199,10 @@ static inline int check_free_page(struct page *page)
>  	return 1;
>  }
>  
> -static int free_tail_pages_check(struct page *head_page, struct page *page)
> +static int free_tail_pages_check(struct page *head_page, int index)
>  {
> +	struct page *page = nth_page(head_page, index);
>  	int ret = 1;
> -
>  	/*
>  	 * We rely page->lru.next never has bit 0 set, unless the page
>  	 * is PageTail(). Let's make sure that's true even for poisoned ->lru.
> @@ -1213,7 +1213,7 @@ static int free_tail_pages_check(struct page *head_page, struct page *page)
>  		ret = 0;
>  		goto out;
>  	}
> -	switch (page - head_page) {
> +	switch (index) {
>  	case 1:
>  		/* the first tail page: ->mapping may be compound_mapcount() */
>  		if (unlikely(compound_mapcount(page))) {
> @@ -1322,6 +1322,7 @@ static __always_inline bool free_pages_prepare(struct page *page,
>  	if (unlikely(order)) {
>  		bool compound = PageCompound(page);
>  		int i;
> +		struct page *tail_page;
>  
>  		VM_BUG_ON_PAGE(compound && compound_order(page) != order, page);
>  
> @@ -1330,13 +1331,14 @@ static __always_inline bool free_pages_prepare(struct page *page,
>  			ClearPageHasHWPoisoned(page);
>  		}
>  		for (i = 1; i < (1 << order); i++) {
> +			tail_page = nth_page(page, i);
>  			if (compound)
> -				bad += free_tail_pages_check(page, page + i);
> -			if (unlikely(check_free_page(page + i))) {
> +				bad += free_tail_pages_check(page, i);
> +			if (unlikely(check_free_page(tail_page))) {
>  				bad++;
>  				continue;
>  			}
> -			(page + i)->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
> +			tail_page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
>  		}
>  	}
>  	if (PageMappingFlags(page))


-- 
Thanks,

David / dhildenb