From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1754373AbbIIQ2z (ORCPT <rfc822;w@1wt.eu>);
	Wed, 9 Sep 2015 12:28:55 -0400
Received: from mx1.redhat.com ([209.132.183.28]:39795 "EHLO mx1.redhat.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1752843AbbIIQ2x (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
	Wed, 9 Sep 2015 12:28:53 -0400
Date: Wed, 9 Sep 2015 18:26:05 +0200
From: Oleg Nesterov <oleg@redhat.com>
To: Chen Gang <xili_gchen_5257@hotmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
        "kirill.shutemov@linux.intel.com" <kirill.shutemov@linux.intel.com>,
        "riel@redhat.com" <riel@redhat.com>, Michal Hocko <mhocko@suse.cz>,
        "sasha.levin@oracle.com" <sasha.levin@oracle.com>,
        "pfeiner@google.com" <pfeiner@google.com>,
        "aarcange@redhat.com" <aarcange@redhat.com>,
        "vishnu.ps@samsung.com" <vishnu.ps@samsung.com>,
        Linux Memory <linux-mm@kvack.org>,
        kernel mailing list <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] mm/mmap.c: Remove useless statement "vma = NULL" in
	find_vma()
Message-ID: <20150909162605.GA4373@redhat.com>
References: <COL130-W64A6555222F8CEDA513171B9560@phx.gbl> <COL130-W6916929C85FB1943CC1B11B9530@phx.gbl>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <COL130-W6916929C85FB1943CC1B11B9530@phx.gbl>
User-Agent: Mutt/1.5.18 (2008-05-17)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 09/08, Chen Gang wrote:
>
> I also want to consult: the comments of find_vma() says:

Sorry, I don't understand the question ;)

> "Look up the first VMA which satisfies addr < vm_end, ..."
>
> Is it OK?

Why not?

> (why not "vm_start <= addr < vm_end"),

Because this some callers actually want to find the 1st vma which
satisfies addr < vm_end? For example, shift_arg_pages().

OTOH, I think that another helper,

	find_vma_xxx(mm, addr)
	{
		vma = find_vma(...)
		if (vma && vma->vm_start > addr)
			vma = NULL;
		return vma;
	}

makes sense. It can have a lot of users.

> need we let "vma = tmp"
> in "if (tmp->vm_start <= addr)"? -- it looks the comments is not match
> the implementation, precisely (maybe not 1st VMA).

This contradicts with above... I mean, it is not clear what exactly do
you blame, semantics or implementation.

The implementation looks correct. Why do you think it can be not 1st vma?

Oleg.