Received-SPF: pass (google.com: domain of linux-kernel+bounces-104291-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Message-ID: <b8b3d0d9-7764-4fc4-b848-c02dd06fccc6@arm.com>
Date: Fri, 15 Mar 2024 10:55:14 +0000
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH v4 6/6] mm: madvise: Avoid split during MADV_PAGEOUT and
 MADV_COLD
Content-Language: en-GB
To: David Hildenbrand <david@redhat.com>,
 Andrew Morton <akpm@linux-foundation.org>,
 Matthew Wilcox <willy@infradead.org>, Huang Ying <ying.huang@intel.com>,
 Gao Xiang <xiang@kernel.org>, Yu Zhao <yuzhao@google.com>,
 Yang Shi <shy828301@gmail.com>, Michal Hocko <mhocko@suse.com>,
 Kefeng Wang <wangkefeng.wang@huawei.com>, Barry Song <21cnbao@gmail.com>,
 Chris Li <chrisl@kernel.org>
Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org
References: <20240311150058.1122862-1-ryan.roberts@arm.com>
 <20240311150058.1122862-7-ryan.roberts@arm.com>
 <ffeee7da-e625-40dc-8da8-b70e4e6ef935@redhat.com>
From: Ryan Roberts <ryan.roberts@arm.com>
In-Reply-To: <ffeee7da-e625-40dc-8da8-b70e4e6ef935@redhat.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

On 15/03/2024 10:35, David Hildenbrand wrote:
>> -        if (!pageout && pte_young(ptent)) {
>> -            ptent = ptep_get_and_clear_full(mm, addr, pte,
>> -                            tlb->fullmm);
>> -            ptent = pte_mkold(ptent);
>> -            set_pte_at(mm, addr, pte, ptent);
>> -            tlb_remove_tlb_entry(tlb, pte, addr);
>> +        if (!pageout) {
>> +            for (; nr != 0; nr--, pte++, addr += PAGE_SIZE) {
>> +                if (ptep_test_and_clear_young(vma, addr, pte))
>> +                    tlb_remove_tlb_entry(tlb, pte, addr);
>> +            }
>>           }
> 
> 
> The following might turn out a bit nicer: Make folio_pte_batch() collect
> "any_young", then doing something like we do with "any_writable" in the fork()
> case:
> 
> ...
>     nr = folio_pte_batch(folio, addr, pte, ptent, max_nr,
>                  fpb_flags, NULL, any_young);
>     if (any_young)
>         pte_mkyoung(ptent)
> ...
> 
> if (!pageout && pte_young(ptent)) {
>     mkold_full_ptes(mm, addr, pte, nr, tlb->fullmm);
>     tlb_remove_tlb_entries(tlb, pte, nr, addr);
> }
> 

I thought about that but decided that it would be better to only TLBI the actual
entries that were young. Although looking at tlb_remove_tlb_entry() I see that
it just maintains a range between the lowest and highest address, so this won't
actually make any difference.

So, yes, this will be a nice improvement, and also prevent the O(n^2) pte reads
for the contpte case. I'll change in the next version.