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: <55de7884-a564-72a4-fe83-173a80aac6ad@benettiengineering.com>
Date:   Wed, 19 Oct 2022 18:29:07 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.3.3
Subject: Re: [PATCH v2 2/2] ARM: mm: convert empty_zero_page to array for
 consistency
Content-Language: en-US
To:     "Russell King (Oracle)" <linux@armlinux.org.uk>
Cc:     linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        Anshuman Khandual <anshuman.khandual@arm.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Kefeng Wang <wangkefeng.wang@huawei.com>,
        Arnd Bergmann <arnd@arndb.de>, Will Deacon <will@kernel.org>
References: <20221018222503.90118-1-giulio.benetti@benettiengineering.com>
 <20221018222503.90118-2-giulio.benetti@benettiengineering.com>
 <Y1ANW9ZRRPDJxlmc@shell.armlinux.org.uk>
From:   Giulio Benetti <giulio.benetti@benettiengineering.com>
In-Reply-To: <Y1ANW9ZRRPDJxlmc@shell.armlinux.org.uk>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

Hello Russell,

On 19/10/22 16:44, Russell King (Oracle) wrote:
> On Wed, Oct 19, 2022 at 12:25:03AM +0200, Giulio Benetti wrote:
>> ARM architecture is the only one to have empty_zero_page to be a
>> struct page pointer, while in all other implementations empty_zero_page is
>> a data pointer or directly an array(the zero page itself). So let's convert
>> empty_zero_page to an array for consistency and to avoid an early
>> allocation+dcache flush. Being the array in .bss it will be cleared earlier
>> in a more linear way(and a bit faster) way.
>>
>> Suggested-by: Arnd Bergmann <arnd@arndb.de>
>> Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
> 
> I'm completely against this approach. It introduces inefficiencies in
> paths we don't need, and also means that the zero page is at a fixed
> location relative to the kernel, neither of which I like in the
> slightest.

I haven't considered those details, I'm pretty new in this topic.
I was thinking with a no-mmu approach in my mind, that's why the
.bss approach. And also the exposure of the entire array to the other
subsystem is not a good idea.

Thank you for pointing me

Best regads
-- 
Giulio Benetti
CEO/CTO@Benetti Engineering sas