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 S1751258AbbCNQT3 (ORCPT <rfc822;w@1wt.eu>);
	Sat, 14 Mar 2015 12:19:29 -0400
Received: from mail-we0-f178.google.com ([74.125.82.178]:35866 "EHLO
	mail-we0-f178.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1750767AbbCNQT1 (ORCPT
	<rfc822;linux-kernel@vger.kernel.org>);
	Sat, 14 Mar 2015 12:19:27 -0400
Message-ID: <55045F84.70605@gmail.com>
Date: Sat, 14 Mar 2015 18:19:16 +0200
From: Alex Dowad <alexinbeijing@gmail.com>
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.4.0
MIME-Version: 1.0
To: josh@joshtriplett.org
CC: linux-kernel@vger.kernel.org, Andrew Morton <akpm@linux-foundation.org>,
        Oleg Nesterov <oleg@redhat.com>, Peter Zijlstra <peterz@infradead.org>,
        "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
        Rik van Riel <riel@redhat.com>,
        Vladimir Davydov <vdavydov@parallels.com>
Subject: Re: [PATCH 01/32] do_fork(): Rename 'stack_size' argument to reflect
 actual use
References: <1426269888-25600-1-git-send-email-alexinbeijing@gmail.com> <20150313230354.GA11522@cloud>
In-Reply-To: <20150313230354.GA11522@cloud>
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org


On 14/03/15 01:04, josh@joshtriplett.org wrote:
> On Fri, Mar 13, 2015 at 08:04:16PM +0200, Alex Dowad wrote:
>> The 'stack_size' argument is never used to pass a stack size. It's only used when
>> forking a kernel thread, in which case it is an argument which should be passed
>> to the 'main' function which the kernel thread executes. Hence, rename it to
>> 'kthread_arg'.
> That's not the only use of stack_size.  Take a look at the clone2 system
> call (very minimally documented in the clone manpage) and the
> implementation of copy_thread on ia64, which does use stack_size in the
> non-kthread path.
Thanks for pointing that out. I searched for all uses with cscope but 
missed sys_clone2 (which is implemented in asm). Won't make that mistake 
again...

I've just been searching for history on sys_clone2() but have come up 
empty. "git blame" isn't helping, either... the current code dates back 
before the start of the git history.

So out of curiosity, if you are willing to explain more: why does clone2 
only exist on IA-64? Is there some characteristic of the architecture 
that makes being able to specify the size of the user-mode stack 
especially valuable, as compared to other archs? Is it used much (such 
as being called from the C library)?

Thanks for your feedback,
Alex Dowad