From mboxrd@z Thu Jan 1 00:00:00 1970 From: Vinod Subject: Re: [PATCH v3 07/18] dmaengine: dma-jz4780: Add support for the JZ4770 SoC Date: Tue, 24 Jul 2018 21:31:40 +0530 Message-ID: <20180724160140.GI3661@vkoul-mobl> References: <20180721110643.19624-1-paul@crapouillou.net> <20180721110643.19624-8-paul@crapouillou.net> <20180724133229.GG3661@vkoul-mobl> <1532444695.2610.1@smtp.crapouillou.net> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Return-path: Content-Disposition: inline In-Reply-To: <1532444695.2610.1@smtp.crapouillou.net> Sender: linux-kernel-owner@vger.kernel.org To: Paul Cercueil Cc: Rob Herring , Mark Rutland , Ralf Baechle , Paul Burton , James Hogan , Zubair Lutfullah Kakakhel , Mathieu Malaterre , Daniel Silsby , dmaengine@vger.kernel.org, devicetree@vger.kernel.org, linux-kernel@vger.kernel.org, linux-mips@linux-mips.org List-Id: devicetree@vger.kernel.org On 24-07-18, 17:04, Paul Cercueil wrote: > Hi Vinod, > > Le mar. 24 juil. 2018 à 15:32, Vinod a écrit : > > On 21-07-18, 13:06, Paul Cercueil wrote: > > > +static inline void jz4780_dma_chan_enable(struct jz4780_dma_dev > > > *jzdma, > > > + unsigned int chn) > > > > right justified and aligned with preceding please. While adding new > > code to a existing driver it is a good idea to conform to existing style > > OK. > > > > +{ > > > + if (jzdma->version == ID_JZ4770) > > > + jz4780_dma_ctrl_writel(jzdma, JZ_DMA_REG_DCKES, BIT(chn)); > > > +} > > > + > > > +static inline void jz4780_dma_chan_disable(struct jz4780_dma_dev > > > *jzdma, > > > + unsigned int chn) > > > +{ > > > + if (jzdma->version == ID_JZ4770) > > > + jz4780_dma_ctrl_writel(jzdma, JZ_DMA_REG_DCKEC, BIT(chn)); > > > > so if another version has this feature we would do: > > if (jzdma->version == ID_JZ4770) || > > if (jzdma->version == ID_JZXXXX)) > > > > and so on.. why not add a value, clk_enable in the description and use > > that. For each controller it is set to true or false > > I agree with what you said in your other answers. > However here I still need to check the "version", because on JZ4725B and > JZ4770+ > the way to start/stop each DMA channel's clock is different, so I can't use > a boolean sure describe the behavior and use that. Versions is not a very scalable way.. -- ~Vinod