From mboxrd@z Thu Jan 1 00:00:00 1970 From: Petr Cvek Subject: Re: [PATCH 3/4] mmc: pxamci: Disable DATA_TRAN_DONE interrupt sooner Date: Fri, 21 Apr 2017 01:37:04 +0200 Message-ID: References: <04139c1a-cee8-f64c-4b0e-2e3c472169a3@tul.cz> <87k26g6whu.fsf@belgarion.home> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-2 Content-Transfer-Encoding: 7bit Return-path: Received: from bubo.tul.cz ([147.230.16.1]:33744 "EHLO bubo.tul.cz" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S947895AbdDTXgb (ORCPT ); Thu, 20 Apr 2017 19:36:31 -0400 In-Reply-To: <87k26g6whu.fsf@belgarion.home> Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: Robert Jarzmik Cc: ulf.hansson@linaro.org, linux-mmc@vger.kernel.org, linux-arm-kernel@lists.infradead.org Dne 19.4.2017 v 21:14 Robert Jarzmik napsal(a): > Petr Cvek writes: > >> Disable the DATA_TRAN_DONE interrupt as soon as possible in the handler. > Yeah, but why, please explain. > If that's only to "reduce" the race occurrence, then I'd rather have this patch > dropped. Otherwise if there is a compelling reason let's see ... > I created it during researching where put the IRQ enable in the callback and then I left it that way the data done interrupt will be disabled if there was an error at the beginning of the pxamci_data_done(). I dropped the patch and re-tested and it works (in the case the tests at the start of the pxamci_data_done() fail there will be probably some irq hell though :-D). > Cheers. > From mboxrd@z Thu Jan 1 00:00:00 1970 From: petr.cvek@tul.cz (Petr Cvek) Date: Fri, 21 Apr 2017 01:37:04 +0200 Subject: [PATCH 3/4] mmc: pxamci: Disable DATA_TRAN_DONE interrupt sooner In-Reply-To: <87k26g6whu.fsf@belgarion.home> References: <04139c1a-cee8-f64c-4b0e-2e3c472169a3@tul.cz> <87k26g6whu.fsf@belgarion.home> Message-ID: To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Dne 19.4.2017 v 21:14 Robert Jarzmik napsal(a): > Petr Cvek writes: > >> Disable the DATA_TRAN_DONE interrupt as soon as possible in the handler. > Yeah, but why, please explain. > If that's only to "reduce" the race occurrence, then I'd rather have this patch > dropped. Otherwise if there is a compelling reason let's see ... > I created it during researching where put the IRQ enable in the callback and then I left it that way the data done interrupt will be disabled if there was an error at the beginning of the pxamci_data_done(). I dropped the patch and re-tested and it works (in the case the tests at the start of the pxamci_data_done() fail there will be probably some irq hell though :-D). > Cheers. >