From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mark Lord Subject: Re: Hotplug borked after suspend/resume in Linux-3.3 ? Date: Tue, 17 Apr 2012 17:16:48 -0400 Message-ID: <4F8DDDC0.2010609@teksavvy.com> References: <4F8DD4DC.4050503@teksavvy.com> <4F8DD708.8050502@teksavvy.com> <4F8DD939.1050909@pobox.com> <4F8DDB24.9080109@teksavvy.com> <4F8DDC4F.8060103@pobox.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Return-path: Received: from ironport-out.teksavvy.com ([206.248.143.162]:52575 "EHLO ironport-out.teksavvy.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751542Ab2DQVQt (ORCPT ); Tue, 17 Apr 2012 17:16:49 -0400 In-Reply-To: <4F8DDC4F.8060103@pobox.com> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Jeff Garzik Cc: Tejun Heo , IDE/ATA development list On 12-04-17 05:10 PM, Jeff Garzik wrote: > On 04/17/2012 05:05 PM, Mark Lord wrote: >> On 12-04-17 04:57 PM, Jeff Garzik wrote: >>> On 04/17/2012 04:48 PM, Mark Lord wrote: >>>> On 12-04-17 04:38 PM, Mark Lord wrote: >>>>> Okay, so why isn't SATA hotplug working in linux-3.3.2 ? >>>>> I don't know when it stopped working, but it's a blooming pain in the sysadmin. >>>>> >>>>> I need to hotplug a SATA drive into an AHCI port, >>>>> preferably without having to reboot first. >>>>> >>>>> Is there a patch available already for this regression? >>>> >>>> >>>> Answering my own question, this patch appears to address the issue. >>>> I'll test and report back again shortly: >>>> >>>> http://patchwork.ozlabs.org/patch/146326/ >>>> >>>> Odd that it's been sitting in various people's inboxes since 3.3-rc1 >>>> and hasn't been pushed out yet. >>> >>> The vast majority of "hotpluggable" ports are not necessarily covered by this patch. >> >> >> And what majority of ports are currently broken by linux-3.3 ? >> Any idea of what the offending commit may have been, >> so I can test against that and perhaps get it reverted then? >> >> So far, my sample of three systems are all broken with this kernel. >> That's definitely "regression" territory. :) > > Right -- rather than breaking a bunch and fixing a few, we might just need to revert the runtime pm > stuff altogether. > > Commits to look at include > > 966f1212e1ac5fe3ddf04479d21488ddb36a2608 > 33574d68ae41ccbc6686cfabd965c685285c58a0 > e90b1e5a6e04c8892007ff8db20ef6d4fbdb5402 > 9ee4f3933930abf5cc34f8e9d69fe0e08c18f602 > 5ef41082912bdfcb33fa53b8dba2ad17dea2ef90 > 9a6d6a2ddabbd32c07f6a38b659e5f3db319fa5a > Or perhaps there's a simpler solution/patch, maybe a one-liner to change the boot-up default to "no runtime PM until enabled" or something. Probably less risky if the changes are as pervasive as you suggest.