From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Warren Subject: Re: [PATCH] i2c: i2c-gpio: fix issue when DT node greater than 1 Date: Wed, 19 Sep 2012 09:28:04 -0600 Message-ID: <5059E484.6090708@wwwdotorg.org> References: <1347958949-5598-1-git-send-email-voice.shen@atmel.com> <50588A8F.30203@wwwdotorg.org> <50591C9C.2080809@atmel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <50591C9C.2080809-AIFe0yeh4nAAvxtiuMwx3w@public.gmane.org> Sender: linux-i2c-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Bo Shen Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org, nicolas.ferre-AIFe0yeh4nAAvxtiuMwx3w@public.gmane.org, linux-i2c-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, plagnioj-sclMFOaUSTBWk0Htik3J/w@public.gmane.org, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org, hskinnemoen-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org List-Id: linux-i2c@vger.kernel.org On 09/18/2012 07:15 PM, Bo Shen wrote: > Hi Stephen Warren, > > On 9/18/2012 22:51, Stephen Warren wrote: >> On 09/18/2012 03:02 AM, Bo Shen wrote: >>> When i2c-gpio node number is greater than 1, the second can not >>> sucessfully >>> regiter >>> >>> Using alias method to get the pdev->id, or else the second device >>> will use >>> the same id which will cause fail to register >> >> pdev->id shouldn't be used at all with device tree. Why does >> registration fail without this change? > > I add the debug info, and it give the following error without this patch. > > --<8---------------------------------- > adap->name = i2c-gpio-1 > i2c-gpio i2c.2: using pins 30 (SDA) and 31 (SCL) > adap->name = i2c-gpio-1 > i2c-gpio: probe of i2c.3 failed with error -16 > -->8---------------------------------- > > With this patch, it successfully registered. > --<8---------------------------------- > adap->name = i2c-gpio0 > i2c-gpio i2c.2: using pins 30 (SDA) and 31 (SCL) > adap->name = i2c-gpio1 > i2c-gpio i2c.3: using pins 90 (SDA) and 91 (SCL) > -->8---------------------------------- Yes, that explains why the registration fails, but not why this patch is the correct solution to the problem. The problematic code appears to be: snprintf(adap->name, sizeof(adap->name), "i2c-gpio%d", pdev->id); Instead, I think that should be something more like: adap->name = dev_name(&pdev->dev); or perhaps: if (pdev->dev.of_node) /* named will be based on DT node name */ adap->name = dev_name(&pdev->dev); else snprintf(adap->name, sizeof(adap->name), "i2c-gpio%d", pdev->id); ?