From mboxrd@z Thu Jan  1 00:00:00 1970
From: Wang Chen <wangchen@cn.fujitsu.com>
Subject: Re: v3 [PATCH net-next 4/7] ipv6: Check return of dev_set_allmulti
Date: Mon, 07 Jul 2008 22:45:52 +0800
Message-ID: <48722C20.10801@cn.fujitsu.com>
References: <48718014.4030009@cn.fujitsu.com> <4871814A.2070707@cn.fujitsu.com> <4871FBFC.2050408@trash.net> <48721780.9060103@cn.fujitsu.com> <48721984.8060202@trash.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-15
Content-Transfer-Encoding: 7bit
Cc: "David S. Miller" <davem@davemloft.net>,
	NETDEV <netdev@vger.kernel.org>,
	YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
To: Patrick McHardy <kaber@trash.net>
Return-path: <netdev-owner@vger.kernel.org>
Received: from cn.fujitsu.com ([222.73.24.84]:57669 "EHLO song.cn.fujitsu.com"
	rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP
	id S1750804AbYGGOuq (ORCPT <rfc822;netdev@vger.kernel.org>);
	Mon, 7 Jul 2008 10:50:46 -0400
In-Reply-To: <48721984.8060202@trash.net>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

Patrick McHardy said the following on 2008-7-7 21:26:
> Wang Chen wrote:
>> Patrick McHardy said the following on 2008-7-7 19:20:
>>>> @@ -627,13 +633,14 @@ static int mif6_add(struct mif6ctl *vifc, int
>>>> mrtsock)
>>>>          if (!dev)
>>>>              return -EADDRNOTAVAIL;
>>>>          dev_put(dev);
>>>> +        err = dev_set_allmulti(dev, 1);
>>>> +        if (err)
>>>> +            return err;
>>> The use of dev after putting it looks wrong, but thats already
>>> present before your patch. So ACK for your patch, but we need
>>> a fix on top.
>>>
>>
>> Wait a moment :)
>> ---
>>     case 0:
>>         dev = dev_get_by_index(&init_net, vifc->mif6c_pifi);
>>         if (!dev)
>>             return -EADDRNOTAVAIL;
>>         dev_put(dev);
>> ---
>> dev_get_by_index() holds the dev, so I think dev_put() just for hold/put
>> balance.
> 
> Sure, but its used after dropping the reference again.

If so, my patch wrong.
Because, 

>               dev = ip6mr_reg_vif();
>               if (!dev)
>                       return -ENOBUFS;

here, needs dev_hold(dev) also.
so, I think I should fix it in this patch too, not a fix on top.
How do you think, Patrick?

> +               err = dev_set_allmulti(dev, 1);
> +               if (err) {
> +                       unregister_netdevice(dev);
> +                       return err;
> +               }
>               break;