From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: To: Jens Axboe Cc: Christoph Hellwig , Junichi Nomura , "linux-block\@vger.kernel.org" , linux-scsi , Chaitanya Kulkarni , "Martin K. Petersen" Subject: Re: [REGRESSION v4.10-rc1] blkdev_issue_zeroout() returns -EREMOTEIO on the first call for SCSI device that doesn't support WRITE SAME From: "Martin K. Petersen" References: <20170203161239.GA3880@lst.de> <1e819f0d-ecdc-e54a-bd3d-17de2f71c8a7@kernel.dk> Date: Fri, 03 Feb 2017 17:45:25 -0500 In-Reply-To: <1e819f0d-ecdc-e54a-bd3d-17de2f71c8a7@kernel.dk> (Jens Axboe's message of "Fri, 3 Feb 2017 09:14:11 -0700") Message-ID: MIME-Version: 1.0 Content-Type: text/plain List-ID: >>>>> "Jens" == Jens Axboe writes: >> I think we should fix sd.c to only send WRITE SAME if either of the >> variants are explicitly listed as supported through REPORT SUPPORTED >> OPERATION CODES, or maybe through a whitelist if there are important >> enough devices. Jens> Yep I hate it too. But the reason it's assumed on is that there is essentially no heuristic that works. Just like we assume that READ always works. Out of the ~200 devices I have access to in the lab: - 100% of the SAS/FC disk drives and SSDs support WRITE SAME - Only 2 out of about 50 different drive models support RSOC - About half of the arrays support WRITE SAME(10/16) - None of the arrays I have support RSOC So even if we were to entertain using RSOC for "enterprise" transport classes (which I concur would be nice for other reasons), it wouldn't solve the WRITE SAME problem... -- Martin K. Petersen Oracle Linux Engineering From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Martin K. Petersen" Subject: Re: [REGRESSION v4.10-rc1] blkdev_issue_zeroout() returns -EREMOTEIO on the first call for SCSI device that doesn't support WRITE SAME Date: Fri, 03 Feb 2017 17:45:25 -0500 Message-ID: References: <20170203161239.GA3880@lst.de> <1e819f0d-ecdc-e54a-bd3d-17de2f71c8a7@kernel.dk> Mime-Version: 1.0 Content-Type: text/plain Return-path: Received: from userp1040.oracle.com ([156.151.31.81]:32303 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752527AbdBCWpj (ORCPT ); Fri, 3 Feb 2017 17:45:39 -0500 In-Reply-To: <1e819f0d-ecdc-e54a-bd3d-17de2f71c8a7@kernel.dk> (Jens Axboe's message of "Fri, 3 Feb 2017 09:14:11 -0700") Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Jens Axboe Cc: Christoph Hellwig , Junichi Nomura , "linux-block@vger.kernel.org" , linux-scsi , Chaitanya Kulkarni , "Martin K. Petersen" >>>>> "Jens" == Jens Axboe writes: >> I think we should fix sd.c to only send WRITE SAME if either of the >> variants are explicitly listed as supported through REPORT SUPPORTED >> OPERATION CODES, or maybe through a whitelist if there are important >> enough devices. Jens> Yep I hate it too. But the reason it's assumed on is that there is essentially no heuristic that works. Just like we assume that READ always works. Out of the ~200 devices I have access to in the lab: - 100% of the SAS/FC disk drives and SSDs support WRITE SAME - Only 2 out of about 50 different drive models support RSOC - About half of the arrays support WRITE SAME(10/16) - None of the arrays I have support RSOC So even if we were to entertain using RSOC for "enterprise" transport classes (which I concur would be nice for other reasons), it wouldn't solve the WRITE SAME problem... -- Martin K. Petersen Oracle Linux Engineering