Message-ID: <267812a3-6576-33ce-4d41-054f6ecb812d@redhat.com>
Date:   Mon, 13 Mar 2023 17:04:27 +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: [PATCH v6 0/6] ksm: support tracking KSM-placed zero-pages
Content-Language: en-US
To:     Claudio Imbrenda <imbrenda@linux.ibm.com>
Cc:     yang.yang29@zte.com.cn, akpm@linux-foundation.org,
        jiang.xuexin@zte.com.cn, linux-kernel@vger.kernel.org,
        linux-mm@kvack.org, ran.xiaokai@zte.com.cn, xu.xin.sc@gmail.com,
        xu.xin16@zte.com.cn, Hugh Dickins <hughd@google.com>
References: <202302100915227721315@zte.com.cn>
 <9d7a8be3-ee9e-3492-841b-a0af9952ef36@redhat.com>
 <20230313151014.5e17fc19@p-imbrenda>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat
In-Reply-To: <20230313151014.5e17fc19@p-imbrenda>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

>> If we'd really want to identify whether a zeropage was deduplciated by
>> KSM, we could try storing that information inside the PTE instead of
> 
> this is interesting, but needs caution, for the reason you mention below
> 
>> inside the RMAP. Then, we could directly adjust the counter when zapping
>> the shared zeropage or during MADV_DONTNEED/when unmerging.
>>
>> Eventually, we could simply say that
>> * !pte_dirty(): zeropage placed during fault
>> * pte_dirty(): zeropage placed by KSM
>>
>> Then it would also be easy to adjust counters and unmerge. We'd limit
>> this handling to known-working architectures initially (spec64 still has

^ I meant sparc64 here. We can (and should) have a testcase that 
deduplicates the shared zeropage using KSM and makes sure that writes 
properly lead to a write fault.

>> the issue that pte_mkdirty() will set a pte writable ... and my patch to
>> fix that was not merged yet). We'd have to double-check all
>> pte_mkdirty/pte_mkclean() callsites.
> 
> this will be... interesting

IIRC, most code that touches the dirty bit makes sure that it operates 
on a proper struct page (via vm_normal_folio()). 
madvise_free_pte_range() is one such user. But we have to double-check.

-- 
Thanks,

David / dhildenb