From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bryan Wu Subject: Re: [PATCH 1/3] Blackfin ethernet driver: on chip ethernet MAC controller driver Date: Sun, 15 Jul 2007 17:52:30 +0800 Message-ID: <1184493150.3140.65.camel@roc-laptop> References: <1184438993.3140.23.camel@roc-laptop> <200707142138.36619.mb@bu3sch.de> <1184491016.3140.52.camel@roc-laptop> <200707151136.05223.mb@bu3sch.de> Reply-To: bryan.wu@analog.com Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: bryan.wu@analog.com, Jeff Garzik , LKML , netdev@vger.kernel.org, Andrew Morton To: Michael Buesch Return-path: Received: from nwd2mail11.analog.com ([137.71.25.57]:22773 "EHLO nwd2mail11.analog.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756367AbXGOJwj (ORCPT ); Sun, 15 Jul 2007 05:52:39 -0400 In-Reply-To: <200707151136.05223.mb@bu3sch.de> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Sun, 2007-07-15 at 11:36 +0200, Michael Buesch wrote: > On Sunday 15 July 2007 11:16:56 Bryan Wu wrote: > > In current Blackfin DMA allocation/free, the return value from > > bfin_mac_alloc() is used as the dma_handle, here it is the > > tx_desc/rx_desc. > > The "dma_handle" is useless in the following code. > > I think a comment has to be added, at least, as it's very confusing. No problem, it should be. > > > This is some magic code for PHY ID. In the future, we will rewrite some > > code based on kernel phy abstraction layer to support more phy device. > > Ok, nice idea. > > > > Unwind the allocations above, if registering fails. > > > > > > > In fact, it is safe. Because if registering fails, bf537mac_probe will > > return none zero to bfin_mac_probe which will do free_netdev. > > Hm, weren't there some DMA allocations, too? Are they free'd properly? > Oh, the DMA allocation happens in bf537mac_open not in bf537mac_probe. so when registering fails, only free_netdev should be ok. I sent out a new driver patch for #2 try, please give us some comments Thanks - Bryan Wu