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: <fc866097-f529-158e-8f24-5d42b11d28b1@redhat.com>
Date:   Tue, 7 Jun 2022 14:55:05 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.9.0
Subject: Re: [PATCH] mm: add a new emergency page migratetype.
Content-Language: en-US
To:     Huanpeng Xin <xinhuanpeng9@gmail.com>, akpm@linux-foundation.org
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        xinhuanpeng <xinhuanpeng@xiaomi.com>
References: <20220606032709.11800-1-xinhuanpeng9@gmail.com>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat
In-Reply-To: <20220606032709.11800-1-xinhuanpeng9@gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 06.06.22 05:27, Huanpeng Xin wrote:
> From: xinhuanpeng <xinhuanpeng@xiaomi.com>
> 
> add a new page migratetype reserved for
> non-costly non-NOWARN page allocation failure.

Sorry to say, but this patch description is not expressive enough. I
have absolutely no clue what you are trying to achieve and why we should
care.

Especially, why do we care about locally grouping these allocations
(that's what pageblock flags are for after all)?

Your Kconfig option is also not particularly user friendly to read either:

"This enables the migration type MIGRATE_EMERGENCY,which reserves
 a small amount of memory for non-costly non-NOWARN page allocation
 failure."

Usually we reserve memory via different mechanisms, like atomic
reserves? Why can't something like that be used.

On first sight, defining a new pageblock migratype feels wrong to me.
But then, I have no clue what you are actually trying to achieve.

-- 
Thanks,

David / dhildenb