From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: [RFC] How to implement linux_block commands in scsi midlayer Date: Mon, 19 Feb 2007 10:07:46 -0600 Message-ID: <1171901266.3385.15.camel@mulgrave.il.steeleye.com> References: <87abzcxur6.fsf@denkblock.local> <45D7D24B.4080802@torque.net> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from hancock.steeleye.com ([71.30.118.248]:36037 "EHLO hancock.sc.steeleye.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S932375AbXBSQHv (ORCPT ); Mon, 19 Feb 2007 11:07:51 -0500 In-Reply-To: <45D7D24B.4080802@torque.net> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: dougg@torque.net Cc: linux-scsi@vger.kernel.org, Jens Axboe On Sat, 2007-02-17 at 23:12 -0500, Douglas Gilbert wrote: > IOW don't do it! > > Better would be to flag the request for internal use. This would be my opinion too. We're trying to move away from encapsulated commands wherever we can (although we're definitely not there yet). The idea is that the block layer would be able to transport native commands to underlying devices. However, we were also thinking of adding control packets that could be interpreted by the processing ULD (as park essentially is, since it's a state in the disk management model). However, realistically, you don't ever send a command to park a disk ... this won't be beneficial during most times of operation ... you really want expose the power states and management so the OS can use them as part of its model ... Tejun already has the beginnings of a patch to do this. James