From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Zubair Lutfullah :" Subject: Re: [PATCH 1/4] input: ti_am335x_tsc: correct step mask update after IRQ Date: Fri, 16 Aug 2013 22:42:24 +0100 Message-ID: <20130816214224.GE2636@gmail.com> References: <1376412499-21007-1-git-send-email-zubair.lutfullah@gmail.com> <1376412499-21007-2-git-send-email-zubair.lutfullah@gmail.com> <20130816085350.GA26895@linutronix.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20130816085350.GA26895@linutronix.de> Sender: linux-kernel-owner@vger.kernel.org To: Sebastian Andrzej Siewior Cc: Zubair Lutfullah , jic23@cam.ac.uk, dmitry.torokhov@gmail.com, sameo@linux.intel.com, lee.jones@linaro.org, linux-iio@vger.kernel.org, linux-input@vger.kernel.org, linux-kernel@vger.kernel.org, gregkh@linuxfoundation.org, Russ.Dill@ti.com List-Id: linux-input@vger.kernel.org On Fri, Aug 16, 2013 at 10:53:50AM +0200, Sebastian Andrzej Siewior wrote: > * Zubair Lutfullah | 2013-08-13 17:48:16 [+0100]: > > >@@ -316,7 +318,7 @@ static irqreturn_t titsc_irq(int irq, void *dev) > > > > if (irqclr) { > > titsc_writel(ts_dev, REG_IRQSTATUS, irqclr); > >- am335x_tsc_se_update(ts_dev->mfd_tscadc); > >+ am335x_tsc_se_set(ts_dev->mfd_tscadc, ts_dev->step_mask); > > return IRQ_HANDLED; > > } > > return IRQ_NONE; > > titsc_step_config() computes the mask once since it does not change. It > is then assigned via am335x_tsc_se_set() to ->reg_se_cache() and later > always udpated via am335x_tsc_se_update(). This should ensure that ADC's > and TSC's bits are in sync and clear each other out. > Now you call am335x_tsc_se_set() in every irq which adds the TSC's mask > to ->reg_se_cache but why? It was never removed. > > Sebastian The problem is when ADC/TSC are used together. reg_se_cache would get updated with masks in the se_set function in MFD core. >>From TSC driver, the TSC steps would be set in the reg_se_cache variable. >>From ADC driver, the ADC steps would be set in the reg_se_cache variable. But the ADC masks weren't being cleared from the reg_se_cache variable anywhere. After a while of using ADC/TSC together, the reg_se_cache variable would have FFFF always. Resulting in redundant sampling and data in ADC FIFO0. MFD has received fixes to update the reg_se_cache upon every call to the se_set functions. Thus, the step_mask must be set every time to ensure that it is updated correctly every time. Hope that clears the confusion. This TSC/ADC sharing can be pretty confusing. Thanks Zubair