From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tony Lindgren Subject: Re: linux-omap3-2.6.39: spurious irq 95, during heavy SD card load Date: Thu, 3 Nov 2011 12:37:32 -0700 Message-ID: <20111103193732.GY31337@atomide.com> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from mho-03-ewr.mailhop.org ([204.13.248.66]:17082 "EHLO mho-01-ewr.mailhop.org" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1756479Ab1KCThe (ORCPT ); Thu, 3 Nov 2011 15:37:34 -0400 Content-Disposition: inline In-Reply-To: Sender: linux-omap-owner@vger.kernel.org List-Id: linux-omap@vger.kernel.org To: Manfred Cc: linux-omap@vger.kernel.org * Manfred [111103 02:15]: > Dear users/developers, > > I am trying to install the xenomai real-time linux on a gumstix overo fire with > an omap3530. > After some starting problems, I succeed to patch the 2.6.33 kernel and run some > latency tests > with normal worst case latencies ~60us. However, the overall worst case latency > is very bad (5ms), > because i occasionally get spurious interrupt errors like this: > Spurious irq 95: 0xffffffdf, please flush posted write for irq 86 > Spurious irq 95: 0xffffffdf, please flush posted write for irq 86 ... > Or is there a specific patch to resolve this? > I was not able to reproduce this on the NAND-flash, > but it is also to small to make large continous data transfers. What the warnings mean is that the ack for irq86 did not reach the hardware before that interrupt got unmasked again. To fix this, find the irq handler for irq86, and add a readback of the interrupt status register or revision register to the interrupt handler right after the write acking the interrupt. To see some examples, please do: $ grep -ri "flush posted write" drivers/ Regards, Tony