From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Subject: Re: [PATCH 15/16] block: split scsi_request out of struct request To: Christoph Hellwig , Bart Van Assche References: <1485185361-29786-1-git-send-email-hch@lst.de> <1485185361-29786-16-git-send-email-hch@lst.de> <455a58ef-905c-abf9-ae59-ca98f003f31e@sandisk.com> <20170126193734.GA26037@lst.de> CC: Mike Snitzer , , , From: Jens Axboe Message-ID: Date: Thu, 26 Jan 2017 12:42:32 -0700 MIME-Version: 1.0 In-Reply-To: <20170126193734.GA26037@lst.de> Content-Type: text/plain; charset="windows-1252" Sender: linux-scsi-owner@vger.kernel.org List-ID: On 01/26/2017 12:37 PM, Christoph Hellwig wrote: > On Thu, Jan 26, 2017 at 11:12:51AM -0800, Bart Van Assche wrote: >> Where does the '* 3' come from? I think that deserves a comment. >> Additionally, this patch introduces a new warning when building with W=1: > > It's a magic factor copied from the old code :( > > That beeing said I really wonder if we should keep the tracing for > the scsi_requests fields at all. We have tracing in SCSI, and poking > into these SCSI fields from block code is painful and just screams > layering violations. > > Jens, any opinion on just removing the printout of the SCSI CDB > for blktrace? Kill it with fire, I don't think there's much value to having it there to begin with. -- Jens Axboe