From mboxrd@z Thu Jan 1 00:00:00 1970 From: Grant Grundler Subject: Re: [PATCH] net: tulip: turn compile-time warning into dev_warn() Date: Thu, 19 Nov 2015 13:57:30 -0800 Message-ID: References: <9720627.53btSdPcQU@wuerfel> <20151119122610.GF22786@arm.com> <564E313B.6000801@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Cc: Will Deacon , Arnd Bergmann , "open list:TULIP NETWORK DRI..." , David Miller , ard.biesheuvel@linaro.org, Grant Grundler To: Florian Fainelli Return-path: Received: from mail-wm0-f44.google.com ([74.125.82.44]:37540 "EHLO mail-wm0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758977AbbKSV5b (ORCPT ); Thu, 19 Nov 2015 16:57:31 -0500 Received: by wmww144 with SMTP id w144so135363431wmw.0 for ; Thu, 19 Nov 2015 13:57:30 -0800 (PST) In-Reply-To: <564E313B.6000801@gmail.com> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, Nov 19, 2015 at 12:29 PM, Florian Fainelli wrote: > On 19/11/15 04:26, Will Deacon wrote: ... >> /me waits for on-soc tulip integration. > > FWIW, this already happened, the ADMtek/Infineon ADM8668 actually > integrated a Tulip chip. I have not submitted these patches below from > the OpenWrt tree because the chip is barely used nowadays, and it was > only mostly popular with the Linksys WRTU54G. > > The patches could be made less intrusive if we did convert the pci_dma* > calls into regular DMA-API calls, which they are nowadays, oh well! I agree. IIRC, there was no DMA-API when this driver was written. James Bottomley added DMA API later and there was no need to convert since Tulip devices were _only_ PCI at the time. > https://dev.openwrt.org/browser/trunk/target/linux/adm8668/patches-3.18/004-tulip_pci_split.patch In general this would be a reasonable patch to submit here with some caveats: 1) convert to DMA API (first patch) 2) add CONFIG_PCI code (second patch) to handle the remaining discovery and PCI Config space bits. Some additional minor refactoring of the code could convert this into a "multi-bus driver" if there is any system that could incorporate both a platform device and a PCI device. I expect the conversion to DMA API to be straight forward as the next patch shows: > https://dev.openwrt.org/browser/trunk/target/linux/adm8668/patches-3.18/005-tulip_platform.patch Split this patch into two parts: convert to DMA-API (first patch) and platform device support (third patch). Should be a "no brainer" to accept. Lastly, net/ethernet/dec/tulip driver is up for adoption. I've just been extremely lazy about updating the MAINTAINERS entry but will submit that shortly (apologies to Arndt for the bounced email - I know it's a bit disconcerting to see that.) I'm happy to continue review tulip code changes anyway. cheers, grant > -- > Florian