From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755529AbbBPLXv (ORCPT ); Mon, 16 Feb 2015 06:23:51 -0500 Received: from svenfoo.org ([82.94.215.22]:60311 "EHLO mail.zonque.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755250AbbBPLXt (ORCPT ); Mon, 16 Feb 2015 06:23:49 -0500 Message-ID: <54E1D342.7090005@zonque.org> Date: Mon, 16 Feb 2015 12:23:46 +0100 From: Daniel Mack User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.4.0 MIME-Version: 1.0 To: Vasily Khoruzhick , Robert Jarzmik CC: Vinod Koul , Arnd Bergmann , Olof Johansson , Daniel Mack , Haojian Zhuang , dmaengine@vger.kernel.org, "linux-kernel@vger.kernel.org" , arm-linux Subject: Re: [PATCH RFC 2/2] ARM: pxa: transition to dmaengine phase 1 References: <1423954053-21760-1-git-send-email-robert.jarzmik@free.fr> <1423954053-21760-3-git-send-email-robert.jarzmik@free.fr> In-Reply-To: Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 02/16/2015 12:14 PM, Vasily Khoruzhick wrote: > What about dropping old PXA DMA code completely? Daniel Mack did port > for most of PXA drivers to dma engine, > I've rebased his patches against 3.17 several months ago and fixed > oopses in pxamci and asoc drivers, but I didn't resubmit whole series > due to lack of time. Yes, I did the same locally. > My 3.17 tree is at [1], I've tested it on pxa270 machine (Zipit Z2), > and everything works fine so far. I guess it won't be too much work to > rebase it against linux-3.20. Rebasing is easy, but there's more to that. We should at least try to convert all drivers over to dmaengine, otherwise they will just stop working. The biggest problem that I currently see is the pxa camera driver, which also uses an advanced way of descriptor hot-linking at runtime. Last time I checked, this wasn't possible to achieve with the dma-slave API. Hence, this driver might need to be completely rewritten, which only someone with such hardware at hand can work on. Maybe the cyclic DMA support in mmp-dma already suffices for that. Robert, weren't you already working on that? Thanks, Daniel