Message-ID: <83b3f3ad-19a2-3737-77dd-3c3f9db82194@redhat.com>
Date:   Wed, 8 Mar 2023 16:57:55 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.8.0
Content-Language: en-US
To:     Peter Xu <peterx@redhat.com>
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        Muhammad Usama Anjum <usama.anjum@collabora.com>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Axel Rasmussen <axelrasmussen@google.com>,
        Mike Rapoport <rppt@linux.vnet.ibm.com>,
        Nadav Amit <nadav.amit@gmail.com>,
        Paul Gofman <pgofman@codeweavers.com>,
        Andrew Morton <akpm@linux-foundation.org>
References: <20230306213925.617814-1-peterx@redhat.com>
 <20230306213925.617814-2-peterx@redhat.com>
 <94be7b9f-c33e-c6dc-4132-6cb78f7c0624@redhat.com> <ZAiqPU0RX3JfDFRP@x1n>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat
Subject: Re: [PATCH v3 1/2] mm/uffd: UFFD_FEATURE_WP_UNPOPULATED
In-Reply-To: <ZAiqPU0RX3JfDFRP@x1n>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

>>
>>> With WP_UNPOPUATED, application like QEMU can avoid pre-read faults all the
>>> memory before wr-protect during taking a live snapshot.  Quotting from
>>> Muhammad's test result here [3] based on a simple program [4]:
>>>
>>>     (1) With huge page disabled
>>>     echo madvise > /sys/kernel/mm/transparent_hugepage/enabled
>>>     ./uffd_wp_perf
>>>     Test DEFAULT: 4
>>>     Test PRE-READ: 1111453 (pre-fault 1101011)
>>>     Test MADVISE: 278276 (pre-fault 266378)
>>>     Test WP-UNPOPULATE: 11712
>>>
>>>     (2) With Huge page enabled
>>>     echo always > /sys/kernel/mm/transparent_hugepage/enabled
>>>     ./uffd_wp_perf
>>>     Test DEFAULT: 4
>>>     Test PRE-READ: 22521 (pre-fault 22348)
>>>     Test MADVISE: 4909 (pre-fault 4743)
>>>     Test WP-UNPOPULATE: 14448
>>>
>>> There'll be a great perf boost for no-thp case, while for thp enabled with
>>> extreme case of all-thp-zero WP_UNPOPULATED can be slower than MADVISE, but
>>> that's low possibility in reality, also the overhead was not reduced but
>>> postponed until a follow up write on any huge zero thp, so potentitially it
>>
>> s/potentitially/potentially/
>>
>>> is faster by making the follow up writes slower.
>>
>> What I realized, interrestingly not only the writes, but also the reads. In
> 
> Curious why reading a zeropage would be a problem?

Oh, my thinking was that with markers you postpone placing the shared 
zeropage. So the next read access will require a pagefault to map the 
shared zeropage. Your v1 would have performed best in that case I guess.

-- 
Thanks,

David / dhildenb