Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Tue, 20 Jul 2021 16:01:27 -0700
From:   Andrew Morton <akpm@linux-foundation.org>
To:     Xiyu Yang <xiyuyang19@fudan.edu.cn>
Cc:     Alistair Popple <apopple@nvidia.com>,
        Yang Shi <shy828301@gmail.com>,
        Shakeel Butt <shakeelb@google.com>,
        Hugh Dickins <hughd@google.com>,
        Miaohe Lin <linmiaohe@huawei.com>,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        yuanxzhang@fudan.edu.cn, Xin Tan <tanxin.ctf@gmail.com>,
        Will Deacon <will.deacon@arm.com>,
        Linus Torvalds <torvalds@linux-foundation.org>
Subject: Re: [PATCH] mm/rmap: Convert from atomic_t to refcount_t on
 anon_vma->refcount
Message-Id: <20210720160127.ac5e76d1e03a374b46f25077@linux-foundation.org>
In-Reply-To: <1626665029-49104-1-git-send-email-xiyuyang19@fudan.edu.cn>
References: <1626665029-49104-1-git-send-email-xiyuyang19@fudan.edu.cn>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Mon, 19 Jul 2021 11:23:35 +0800 Xiyu Yang <xiyuyang19@fudan.edu.cn> wrote:

> refcount_t type and corresponding API can protect refcounters from
> accidental underflow and overflow and further use-after-free situations.

Grumble.

For x86_64 defconfig this takes rmap.o text size from 13226 bytes to
13622.

For x86_64 allmodconfig this takes rmap.o text size from 66576 bytes to
67858.

I didn't check which config option is making the bloat so much worse,
but this really is quite bad.  We bust a gut to make savings which are
1% the size of this!  Is the refcount_t really so much better than a
bare atomic_t that this impact is justified?

Can someone pleeeeeeze take a look at what is happening here and put
the refcount code on a very serious diet?