Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <2cb95372-f7dd-61f9-9d2a-461390fb0907@redhat.com>
Date:   Thu, 11 Aug 2022 18:24:59 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.11.0
Subject: Re: [PATCH v2 2/2] mm/hugetlb: support write-faults in shared
 mappings
Content-Language: en-US
To:     Peter Xu <peterx@redhat.com>
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        stable@vger.kernel.org
References: <20220811103435.188481-1-david@redhat.com>
 <20220811103435.188481-3-david@redhat.com> <YvULP8DtFFhJYNO4@xz-m1.local>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat
In-Reply-To: <YvULP8DtFFhJYNO4@xz-m1.local>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 11.08.22 15:59, Peter Xu wrote:
> On Thu, Aug 11, 2022 at 12:34:35PM +0200, David Hildenbrand wrote:
>> Reason is that uffd-wp doesn't clear the uffd-wp PTE bit when
>> unregistering and consequently keeps the PTE writeprotected. Reason for
>> this is to avoid the additional overhead when unregistering. Note
>> that this is the case also for !hugetlb and that we will end up with
>> writable PTEs that still have the uffd-wp PTE bit set once we return
>> from hugetlb_wp(). I'm not touching the uffd-wp PTE bit for now, because it
>> seems to be a generic thing -- wp_page_reuse() also doesn't clear it.
> 
> This may justify that lazy reset of ptes may not really be a good idea,
> including anonymous.  I'm indeed not aware of any app that do frequent
> reg/unreg at least.

Yeah. QEMU snapshots come to mind, but I guess the reg/unreg overhead is
the smallest issue.


> 
> I'll prepare a patch to change it from uffd side too.
> 
> Thanks again for finding this problem.

YW!

-- 
Thanks,

David / dhildenb