Message-ID: <7b2a7b7b-0ebc-1f03-5f1b-ac598fc950dc@suse.cz>
Date:   Mon, 20 Mar 2023 16:14:50 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.8.0
Subject: Re: [net PATCH 1/2] mm: Use fixed constant in page_frag_alloc instead
 of size + 1
Content-Language: en-US
From:   Vlastimil Babka <vbabka@suse.cz>
To:     Alexander Duyck <alexander.duyck@gmail.com>,
        netdev@vger.kernel.org, davem@davemloft.net
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org, jannh@google.com
References: <20190215223741.16881.84864.stgit@localhost.localdomain>
 <20190215224412.16881.89296.stgit@localhost.localdomain>
 <d68edefb-4930-a9cf-1150-9bd2a2a9a02f@suse.cz>
In-Reply-To: <d68edefb-4930-a9cf-1150-9bd2a2a9a02f@suse.cz>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 2/17/23 10:30, Vlastimil Babka wrote:
> On 2/15/19 23:44, Alexander Duyck wrote:
>> From: Alexander Duyck <alexander.h.duyck@linux.intel.com>
>> 
>> This patch replaces the size + 1 value introduced with the recent fix for 1
>> byte allocs with a constant value.
>> 
>> The idea here is to reduce code overhead as the previous logic would have
>> to read size into a register, then increment it, and write it back to
>> whatever field was being used. By using a constant we can avoid those
>> memory reads and arithmetic operations in favor of just encoding the
>> maximum value into the operation itself.
>> 
>> Fixes: 2c2ade81741c ("mm: page_alloc: fix ref bias in page_frag_alloc() for 1-byte allocs")
>> Signed-off-by: Alexander Duyck <alexander.h.duyck@linux.intel.com>
>> ---
>>  mm/page_alloc.c |    8 ++++----
>>  1 file changed, 4 insertions(+), 4 deletions(-)
>> 
>> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
>> index ebb35e4d0d90..37ed14ad0b59 100644
>> --- a/mm/page_alloc.c
>> +++ b/mm/page_alloc.c
>> @@ -4857,11 +4857,11 @@ void *page_frag_alloc(struct page_frag_cache *nc,
>>  		/* Even if we own the page, we do not use atomic_set().
>>  		 * This would break get_page_unless_zero() users.
>>  		 */
>> -		page_ref_add(page, size);
>> +		page_ref_add(page, PAGE_FRAG_CACHE_MAX_SIZE);
> 
> But this value can be theoretically too low when PAGE_SIZE >
> PAGE_FRAG_CACHE_MAX_SIZE? Such as on architectures with 64kB page size,
> while PAGE_FRAG_CACHE_MAX_SIZE is 32kB?

Nevermind, PAGE_FRAG_CACHE_MAX_SIZE would be 64kB because

#define PAGE_FRAG_CACHE_MAX_SIZE        __ALIGN_MASK(32768, ~PAGE_MASK)

So all is fine, sorry for the noise.