Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756208Ab3FTJTk (ORCPT ); Thu, 20 Jun 2013 05:19:40 -0400 Received: from bear.ext.ti.com ([192.94.94.41]:47008 "EHLO bear.ext.ti.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752359Ab3FTJTi (ORCPT ); Thu, 20 Jun 2013 05:19:38 -0400 Message-ID: <51C2C909.7060001@ti.com> Date: Thu, 20 Jun 2013 14:49:05 +0530 From: Sekhar Nori User-Agent: Mozilla/5.0 (Windows NT 5.1; rv:17.0) Gecko/20130509 Thunderbird/17.0.6 MIME-Version: 1.0 To: Philip Avinash CC: , , , , , , Subject: Re: [PATCH v2 7/7] ARM: davinci: Start using gpiolib API inplace of inline functions References: <1371202532-14628-1-git-send-email-avinashphilip@ti.com> <1371202532-14628-8-git-send-email-avinashphilip@ti.com> In-Reply-To: <1371202532-14628-8-git-send-email-avinashphilip@ti.com> Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1573 Lines: 44 On 6/14/2013 3:05 PM, Philip Avinash wrote: > Remove NEED_MACH_GPIO_H config select option for ARCH_DAVINCI to start > use gpiolib interface for davinci platforms. However with this software > latencies for gpio_get/set APIs will affect. Latency has increased by 18 > microsecond with gpiolib API as compared with inline API's. > > Software latency is calculated on da850 EVM for gpio_get_value API by > taking the printk timing for API execution with interrupts disabled. > Experiment has done for inline and gpiolib API interface. > > inline gpio API with interrupt disabled > [ 29.734337] before gpio_get > [ 29.736847] after gpio_get > > Time difference 0.00251 > > gpio library with interrupt disabled > [ 272.876763] before gpio_get > [ 272.879291] after gpio_get > > Time difference 0.002528 > Latency increased by (0.002528 - 0.00251) = 18 microsecond. > > Also being here > - Moved following definitions from mach folder to include directory > struct davinci_gpio_controller > Macro GPIO(x) > inline function __gpio_mask > - Removed GPIO_TYPE_DAVINCI enum definition as GPIO Davinci is converted > to Linux device driver model. This is bit out-of-place here. Why not do it along with removal of its last usage in .c files? Rest of the patch looks good to me. Thanks, Sekhar -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/