From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Garzik Subject: Re: Resend: [NETDRV] Merge register_netdev calls Date: Mon, 20 Sep 2004 15:03:51 -0400 Sender: netdev-bounce@oss.sgi.com Message-ID: <414F2997.9030901@pobox.com> References: <20040701111914.GA11120@gondor.apana.org.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev@oss.sgi.com Return-path: To: Herbert Xu In-Reply-To: <20040701111914.GA11120@gondor.apana.org.au> Errors-to: netdev-bounce@oss.sgi.com List-Id: netdev.vger.kernel.org Herbert Xu wrote: > On Fri, Jun 11, 2004 at 12:08:55PM +1000, herbert wrote: > >>In fact it's really making these ISA/MCA probe() functions more >>like the ones we have for PCI. > > > To illustrate this, let's take the first driver touched by 4/x. > In 3c503, the function init_module() essentially does > > for each ioaddr > if (do_el2_probe(ioaddr) == 0) > return 0 > > And do_el2_probe() just calls el2_probe1() which is similar > to your average PCI probe function except that the first thing > it does is to make sure that the device exists at ioaddr. > > This is not that different from PCI where it would look > like > > for each PCI device matching the vendor/product numbers > if (do_el2_probe(device) == 0) > return 0 > > Now before my patch, register_netdev was being called just > after do_el2_probe() returns. My patch simply moves it to > the end of el2_probe1() which is exactly what would happen > if this were a PCI driver. > > I've rediffed it against your net-drivers-2.6 tree. so, there was a lot of churn and I held off on this patch. could you be convinced to rediff and resend (again)? Jeff