From mboxrd@z Thu Jan 1 00:00:00 1970 From: Lars-Peter Clausen Subject: Re: [PATCH 1/3] i2c: Revert "i2c: xiic: Do not reset controller before every transfer" Date: Wed, 18 Nov 2015 11:01:56 +0100 Message-ID: <564C4C94.6020406@metafoo.de> References: <1447681325-30914-1-git-send-email-lars@metafoo.de> <564AD886.3060108@metafoo.de> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Return-path: Received: from smtp-out-064.synserver.de ([212.40.185.64]:1252 "EHLO smtp-out-188.synserver.de" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S932087AbbKRKCE (ORCPT ); Wed, 18 Nov 2015 05:02:04 -0500 In-Reply-To: Sender: linux-i2c-owner@vger.kernel.org List-Id: linux-i2c@vger.kernel.org To: Shubhrajyoti Datta Cc: Wolfram Sang , Shubhrajyoti Datta , linux-i2c@vger.kernel.org On 11/17/2015 03:28 PM, Shubhrajyoti Datta wrote: > On Tue, Nov 17, 2015 at 1:04 PM, Lars-Peter Clausen wrote: >> On 11/17/2015 06:17 AM, Shubhrajyoti Datta wrote: >>> On Mon, Nov 16, 2015 at 7:12 PM, Lars-Peter Clausen wrote: >>>> Commit d701667bb331 ("i2c: xiic: Do not reset controller before every >>>> transfer") removed the reinitialization of the controller before the start >>>> of each transfer. Apparently this change is not safe to make and the commit >>>> results in random I2C bus failures. >>> >>> Which is the platform and the ip version that you saw the issue. >>> Did you see the issue with read and write as well? >> >> The IP version is the axi-iic v2.0 Revision 8. I've tested this on a few >> platforms, custom ones and standard ones and I could reproduce it on most. >> One of them was on the ZED board. The one where I couldn't reproduce it was >> the ZC706. But that doesn't necessarily mean it doesn't happen there, just >> that it is not triggered by the testcase. > All the boards having the same version of the ip is what I have understood. > > Thanks for the info I will try to reproduce the issue. > >> >> The problem is that it is random corruption, > Of registers? To be honest I don't know if there is corruption during I2C write transfers, but there is definitely corruption during read transactions. - Lars