Return-path: Received: from cn.fujitsu.com ([222.73.24.84]:55689 "EHLO song.cn.fujitsu.com" rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP id S1752235AbYIFJkj (ORCPT ); Sat, 6 Sep 2008 05:40:39 -0400 Message-ID: <48C24F66.1090803@cn.fujitsu.com> (sfid-20080906_114054_754626_832B15D7) Date: Sat, 06 Sep 2008 17:37:42 +0800 From: Wang Chen MIME-Version: 1.0 To: Dan Williams CC: "John W. Linville" , "David S. Miller" , Jeff Garzik , NETDEV , linux-wireless@vger.kernel.org Subject: Re: [PATCH 14/18] netdevice airo: Convert directly reference of netdev->priv to netdev->ml_priv References: <48C0A219.2030004@cn.fujitsu.com> <48C0A72B.9050407@cn.fujitsu.com> <20080905125036.GA3027@tuxdriver.com> <1220634383.6430.14.camel@localhost.localdomain> In-Reply-To: <1220634383.6430.14.camel@localhost.localdomain> Content-Type: text/plain; charset=UTF-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: Dan Williams said the following on 2008-9-6 1:06: > On Fri, 2008-09-05 at 08:50 -0400, John W. Linville wrote: >> On Fri, Sep 05, 2008 at 11:27:39AM +0800, Wang Chen wrote: >>> We have some reasons to kill netdev->priv: >>> 1. netdev->priv is equal to netdev_priv(). >>> 2. netdev_priv() wraps the calculation of netdev->priv's offset, obviously >>> netdev_priv() is more flexible than netdev->priv. >>> But we cann't kill netdev->priv, because so many drivers reference to it >>> directly. >>> >>> OK, becasue Dave S. Miller said, "every direct netdev->priv usage is a bug", >>> and I want to kill netdev->priv later, I decided to convert all the direct >>> reference of netdev->priv first. >>> >>> Different to readonly reference of netdev->priv, in this driver, netdev->priv >>> was changed. I use netdev->ml_priv to replace netdev->priv. >>> >>> Signed-off-by: Wang Chen >> Thank you for your patch. However, I do not understand why you didn't >> simply replace netdev->priv with netdev_priv()? Can you explain? > > Yeah, that would have been my first choice too... > 1. Why I don't use netdev_priv() to replace netdev->priv here? Because, here > @@ -2665,7 +2666,7 @@ static struct net_device *init_wifidev(struct airo_info *ai, > struct net_device *dev = alloc_netdev(0, "wifi%d", wifi_setup); > if (!dev) > return NULL; > - dev->priv = ethdev->priv; > + dev->ml_priv = ethdev->ml_priv; > @@ -2766,7 +2767,7 @@ static struct net_device *_init_airo_card( unsigned short irq, int port, > return NULL; > } > > - ai = dev->priv; > + ai = dev->ml_priv = netdev_priv(dev); netdev->priv was changed here, but it shouldn't, the memory was allocated when alloc_netdev and netdev->priv should always pointed to that memory. 2. Why I use netdev->ml_priv here to replace netdev->priv? In this driver, netdev->priv are shared by multi wifidevs, that means wifidevs need mid-layer private data, which are all same as their parent netdev. This usage is same as Dave's commit "syncppp: Fix crashes."