From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752107AbcHLCIp (ORCPT ); Thu, 11 Aug 2016 22:08:45 -0400 Received: from aserp1040.oracle.com ([141.146.126.69]:36549 "EHLO aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751143AbcHLCIo (ORCPT ); Thu, 11 Aug 2016 22:08:44 -0400 To: Shaun Tancheff Cc: linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org, "James E . J . Bottomley" , "Martin K . Petersen" , Josh Bingaman , Shaun Tancheff Subject: Re: [PATCH] Update WRITE_SAME timeout in sd_setup_discard_cmnd From: "Martin K. Petersen" Organization: Oracle Corporation References: <1470946358-6744-1-git-send-email-shaun@tancheff.com> Date: Thu, 11 Aug 2016 22:08:33 -0400 In-Reply-To: <1470946358-6744-1-git-send-email-shaun@tancheff.com> (Shaun Tancheff's message of "Thu, 11 Aug 2016 15:12:38 -0500") Message-ID: User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/24.5 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain X-Source-IP: userv0022.oracle.com [156.151.31.74] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org >>>>> "Shaun" == Shaun Tancheff writes: Shaun, Shaun> In sd_setup_discard_cmnd() there are a some discard methods that Shaun> fall back to using WRITE_SAME. It appears that those paths using Shaun> WRITE_SAME should also use the SD_WRITE_SAME_TIMEOUT instead of Shaun> the default SD_TIMEOUT. The expectation is that the UNMAP variants update a translation table of some sort in close to constant time. Potentially with some head and tail zeroing on media. And should therefore easily fall within the 30s limit. But you have a point wrt. SD_LBP_ZERO. It was used for one particular type of array that did zero block detection (and thus didn't actually write anything either). I don't think anybody is using this mode of operation anymore since thin provisioning has been formalized in T10 for quite a while. But I'd be OK with a patch that uses SD_WRITE_SAME_TIMEOUT for that particular code path. -- Martin K. Petersen Oracle Linux Engineering