From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Warren Subject: Re: [PATCH 04/11] ARM: set arch_gettimeoffset directly Date: Fri, 09 Nov 2012 14:07:26 -0700 Message-ID: <509D708E.6080703@wwwdotorg.org> References: <1352408516-21988-1-git-send-email-swarren@wwwdotorg.org> <1352408516-21988-6-git-send-email-swarren@wwwdotorg.org> <509C3AE7.7030703@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: Received: from avon.wwwdotorg.org ([70.85.31.133]:60520 "EHLO avon.wwwdotorg.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753443Ab2KIVHb (ORCPT ); Fri, 9 Nov 2012 16:07:31 -0500 In-Reply-To: <509C3AE7.7030703@gmail.com> Sender: linux-arch-owner@vger.kernel.org List-ID: To: Ryan Mallon Cc: Russell King , Olof Johansson , Arnd Bergmann , John Stultz , Thomas Gleixner , linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, linux-arch@vger.kernel.org, Stephen Warren , Andrew Victor , Nicolas Ferre , Jean-Christophe Plagniol-Villard , Hartley Sweeten , Ben Dooks , Kukjin Kim On 11/08/2012 04:06 PM, Ryan Mallon wrote: > On 09/11/12 08:01, Stephen Warren wrote: >> remove ARM's struct sys_timer .offset function pointer, and instead >> directly set the arch_gettimeoffset function pointer when the timer >> driver is initialized. This requires multiplying all function results >> by 1000, since the removed arm_gettimeoffset() did this. Also, >> s/unsigned long/u32/ just to make the function prototypes exactly >> match that of arch_gettimeoffset. >> +static u32 ep93xx_gettimeoffset(void) >> +{ >> + int offset; >> + >> + offset = __raw_readl(EP93XX_TIMER4_VALUE_LOW) - last_jiffy_time; >> + >> + /* Calculate (1000000 / 983040) * offset. */ > > This comment is now incorrect, it should say: > > /* Calculate (1000000000 / 983040) * offset */ > > or perhaps to better explain what is being done: > > /* > * Timer 4 is based on a 983.04 kHz reference clock, > * so dividing by 983040 gives a milli-second value. > * Refactor the calculation to avoid overflow. > */ > >> + return (offset + (53 * offset / 3072)) * 1000; Thanks. I expanded on that slightly and went for: /* * Timer 4 is based on a 983.04 kHz reference clock, * so dividing by 983040 gives the fraction of a second, * so dividing by 0.983040 converts to uS. * Refactor the calculation to avoid overflow. * Finally, multiply by 1000 to give nS. */