From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Martin K. Petersen" Subject: Re: [PATCH] sd: Optimal I/O size is in bytes, not sectors Date: Thu, 12 May 2016 22:32:23 -0400 Message-ID: References: <1453305683-22424-1-git-send-email-martin.petersen@oracle.com> <20160512040433.GA20240@ad.usersys.redhat.com> Mime-Version: 1.0 Content-Type: text/plain Return-path: Received: from userp1040.oracle.com ([156.151.31.81]:16435 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751882AbcEMCcd (ORCPT ); Thu, 12 May 2016 22:32:33 -0400 In-Reply-To: <20160512040433.GA20240@ad.usersys.redhat.com> (Fam Zheng's message of "Thu, 12 May 2016 12:04:33 +0800") Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Fam Zheng Cc: "Martin K. Petersen" , linux-scsi@vger.kernel.org, borntraeger@de.ibm.com >>>>> "Fam" == Fam Zheng writes: Fam, Fam> This looks wrong to me, maybe I'm missing the obvious? Here sdkp-> Fam> opt_xfer_blocks is in block size unit, and rw_max is in byte sdkp-> Fam> unit. Oh, joy. It's the commit that keeps on giving! Back when the I/O topology was added we were set on eradicating block layer sectors from the stack and count everything in bytes. So the new fields were added as bytes but for various reasons the existing values were never transitioned. And now we have an error prone interface. I think that for 4.8 I'll fix this up properly. I'll post a fix for 4.7 and older tomorrow... -- Martin K. Petersen Oracle Linux Engineering