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 17:02:11 -0400 Message-ID: <4F8DDA53.206@teksavvy.com> References: <4F8DD4DC.4050503@teksavvy.com> <4F8DD708.8050502@teksavvy.com> <4F8DD834.5070000@teksavvy.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]:17708 "EHLO ironport-out.teksavvy.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751542Ab2DQVCM (ORCPT ); Tue, 17 Apr 2012 17:02:12 -0400 In-Reply-To: <4F8DD834.5070000@teksavvy.com> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Tejun Heo , Jeff Garzik , IDE/ATA development list On 12-04-17 04:53 PM, Mark Lord wrote: > On 12-04-17 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. > > Ahh.. perhaps because it doesn't fix the problem I'm seeing here. > > Oh well, anyone else got a fix for this regression yet? > Even doing a rescan-scsi-bus doesn't work on the closed down ports. > > Basically, any unoccupied AHCI port is dead to the world > after a simple suspend/resume (RAM) cycle. On one of my machines > I can hot unplug/replug the ExpressCard adapter to recover, > but that doesn't work so well for built-in SATA/eSATA ports. Mmm.. on my little Atom ZBOX systems, the patch linked above "Patchwork [v2] libata: disable runtime pm for hotpluggable port" _does_ appear to fix the issue. Maybe my notebook is suffering from the ExpressCard not being resumed properly, which would be a PCI issue most likely. Anyway, one issue at a time: The "disable runtime pm" patch seems to be the regression fix I'm looking for here so that I can use SATA hotplug after a suspend/resume. Why isn't this upstream as a regression fix yet? Thanks