From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Garzik Subject: Re: [RFC/PATCH] Deferred disk spinup during system resume Date: Thu, 13 Jan 2011 12:20:00 -0500 Message-ID: <4D2F3440.30701@pobox.com> References: <1294795457-9006-1-git-send-email-maksim.rayskiy@gmail.com> <20110112112142.GA9610@mtj.dyndns.org> <4D2DF45B.5010101@pobox.com> <20110113153707.GB12596@htj.dyndns.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from mail-vw0-f46.google.com ([209.85.212.46]:34142 "EHLO mail-vw0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756855Ab1AMRUE (ORCPT ); Thu, 13 Jan 2011 12:20:04 -0500 Received: by vws16 with SMTP id 16so681918vws.19 for ; Thu, 13 Jan 2011 09:20:04 -0800 (PST) In-Reply-To: <20110113153707.GB12596@htj.dyndns.org> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Tejun Heo Cc: maksim.rayskiy@gmail.com, linux-scsi@vger.kernel.org On 01/13/2011 10:37 AM, Tejun Heo wrote: > Hey, Jeff. > > On Wed, Jan 12, 2011 at 01:35:07PM -0500, Jeff Garzik wrote: >> The previous patch breaks READ VERIFY translation for all cases >> -except- this one. > > Ooh, right; then, can we just fix that? > >> The bottom line is that this patch simply wants to trigger an ATA >> command, and return immediately, discarding the command results. >> I'm not even sure a "run this command in background, and discard >> results" facility requires the EH. > > It doens't necessarily require EH but the dependency is rather > delicate because we still ride on SCSI EH. For example, we don't have > any provision for running qc's without the associated scmd. If the > command fails, libata EH rides on SCSI EH and the interaction relies > on the association between qc's and scmd's. We may be able to add > privison for this case but we already have working mechanism in EH for > internal commands so it's just easiser that way. In this case we -do- have an associated scmd, otherwise he would not be able to trigger the READ VERIFY translation code. So it seems a question of (a) who generates READ VERIFY, and (b) can we have them issue the command without waiting for the results? Which is something I would have thought SG_IO/bsg already provided... It remains unclear to me why EH must be updated, simply to execute a non-data ATA command without waiting. Jeff