From mboxrd@z Thu Jan 1 00:00:00 1970 From: Aaron Lu Subject: Re: [PATCH 1/2] don't wait on disk to start on resume Date: Mon, 04 Feb 2013 22:27:44 +0800 Message-ID: <510FC560.6020002@intel.com> References: <510E025A.9000103@intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: "dbasehore ." Cc: Alan Stern , James Bottomley , Jeff Garzik , linux-ide@vger.kernel.org, linux-kernel , SCSI development list , Linux-pm mailing list List-Id: linux-pm@vger.kernel.org On 02/04/2013 08:07 AM, dbasehore . wrote: > On the topic that we do a fast return for both scsi and ata. Now I > don't remember everything about this (and correct me if I'm wrong) > since I figured this out a few months ago. > > There are some dependencies that scsi has on the resume path of ata. I > think it's that before we can send the command to spin up the disk, we > need to wait for the ata host controller to come up. As Aaron Lu > pointed out, it takes seconds for the ata port to resume. On the hand, I just did some more recording, the result is: host controller takes 1ms or less to resume; port reset takes the most time, almost the same as the whole port resume callback, 2-4 seconds; sd resume callback takes 16ms. So for ata disks, the most time consuming part is in port's reset routine, which is executed in port's resume callback. -Aaron