From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner+w=401wt.eu-S1755266AbYJOBhf@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1755266AbYJOBhf (ORCPT <rfc822;w@1wt.eu>);
	Tue, 14 Oct 2008 21:37:35 -0400
Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752056AbYJOBhY
	(ORCPT <rfc822;linux-kernel-outgoing>);
	Tue, 14 Oct 2008 21:37:24 -0400
Received: from cn.fujitsu.com ([222.73.24.84]:63999 "EHLO song.cn.fujitsu.com"
	rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP
	id S1751276AbYJOBhW (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
	Tue, 14 Oct 2008 21:37:22 -0400
Message-ID: <48F548C7.6010609@cn.fujitsu.com>
Date: Wed, 15 Oct 2008 09:35:03 +0800
From: Lai Jiangshan <laijs@cn.fujitsu.com>
User-Agent: Thunderbird 2.0.0.17 (Windows/20080914)
MIME-Version: 1.0
To: Alexey Dobriyan <adobriyan@gmail.com>
CC: Andrew Morton <akpm@linux-foundation.org>, Paul Menage <menage@google.com>,
       Paul Jackson <pj@sgi.com>,
       Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 1/4] seq_file: don't call bitmap_scnprintf_len()
References: <48F1C36D.6080301@cn.fujitsu.com> <20081013063806.GB3596@x200.localdomain> <48F2F620.9020305@cn.fujitsu.com>
In-Reply-To: <48F2F620.9020305@cn.fujitsu.com>
Content-Type: text/plain; charset=UTF-8
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

Lai Jiangshan wrote:
> Alexey Dobriyan wrote:
>> On Sun, Oct 12, 2008 at 05:29:17PM +0800, Lai Jiangshan wrote:
>>> "m->count + len < m->size" is true commonly, so bitmap_scnprintf()
>>> is commonly called. this fix saves a call to bitmap_scnprintf_len().
>> This saves a call if seq buffer is already full which is not a common
>> situation, so you're optimising for rare situations and adding branch
>> for common ones.
> 
> Hi Alexey
> 
> I think this saves a call if seq buffer is _not_ full after called.
> 
> since calling bitmap_scnprintf() is the common situation. so we use
> the return value of bitmap_scnprintf() instead of bitmap_scnprintf_len().
> 
> In old code bitmap_scnprintf_len() and bitmap_scnprintf() must be called
> commonly.
> 
> Thanks Lai.

Hi, Alexey,

Have noticed this comment? I hold my ground, ;-)
And seq_cpumask_list(), seq_nodemask_list() are needed for print mask
safely, Should I send independent patches for them if we can't
make an agreement about this fix.

Lai

> 
>>> --- a/fs/seq_file.c
>>> +++ b/fs/seq_file.c
>>> @@ -452,17 +452,18 @@ int seq_dentry(struct seq_file *m, struct dentry *dentry, char *esc)
>>>  
>>>  int seq_bitmap(struct seq_file *m, unsigned long *bits, unsigned int nr_bits)
>>>  {
>>> -	size_t len = bitmap_scnprintf_len(nr_bits);
>>> -
>>> -	if (m->count + len < m->size) {
>>> -		bitmap_scnprintf(m->buf + m->count, m->size - m->count,
>>> -				 bits, nr_bits);
>>> -		m->count += len;
>>> -		return 0;
>>> +	if (m->count < m->size) {
>>> +		int len = bitmap_scnprintf(m->buf + m->count,
>>> +				m->size - m->count, bits, nr_bits);
>>> +		if (m->count + len < m->size) {
>>> +			m->count += len;
>>> +			return 0;
>>> +		}
>>>  	}
>>>  	m->count = m->size;
>>>  	return -1;
>>>  }
>>> +EXPORT_SYMBOL(seq_bitmap);
>> Modular users, where are they?
>>
>>
>>
> 
>