From mboxrd@z Thu Jan 1 00:00:00 1970 From: Frans Pop Subject: [PATCH,v2][0/2] sata_mv: harddisk activity led no longer responsive on QNAP TS-109 Date: Wed, 11 Mar 2009 08:13:23 +0100 Message-ID: <200903110813.25650.elendil@planet.nl> Mime-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Return-path: Received: from cpsmtpm-eml105.kpnxchange.com ([195.121.3.9]:64117 "EHLO CPSMTPM-EML105.kpnxchange.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751956AbZCKHN3 (ORCPT ); Wed, 11 Mar 2009 03:13:29 -0400 Content-Disposition: inline Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: linux-arm@vger.kernel.org Cc: Mark Lord , linux-ide@vger.kernel.org, Saeed Bishara , Nicolas Pitre , Lennert Buytenhek On Tuesday 10 March 2009, Mark Lord wrote: > So if you want to rework the first patch so that the blink fix > works for all GEN_IIE chips, then I'll test it here with a few > different drives from different vendors and see how the LEDs behave. Here's version 2 of the patches. Changes: - make the quirk valid for all GenIIe chips and not only for SoC - use a flag to indicate a particular chip needs the quirk - use a hardware op to enable/disable blink mode Open issues: - as I've worked without any specs and only have SoC to test on, please check that mv_iie_enable_led_blink is correct for all GenIIe chips - if mv_iie_enable_led_blink is also valid for GenII chips, the first patch can be simplified a bit (no need for mv6xxx_iie_ops) To test: * load module with ncq_blink_led=0 (default) - check if NCQ is enabled (/sys/block/sda/device/queue_depth > 1) if not -> chip does not support NCQ and is thus not affected? - cause some disk activity; if led stays on without blinking, the chip needs the quirk - disable NCQ (echo 1 >/sys/block/sda/device/queue_depth) - cause some disk activity; led blinking should be very responsive (pre-2.6.26 behavior on my QNAP TS-109) * load module with ncq_blink_led=1 - check if NCQ is enabled (/sys/block/sda/device/queue_depth > 1) - cause some disk activity; led should now blink in a lazy frequency (quirk enabled) - disable NCQ (echo 1 >/sys/block/sda/device/queue_depth) - cause some disk activity; led blinking should be very responsive again (pre-2.6.26 behavior) - re-enable NCQ (echo 31 >/sys/block/sda/device/queue_depth) - cause some disk activity; led blinking should be lazy again Cheers, FJP