From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Martin K. Petersen" Subject: Re: [PATCH 2/5] block: add function to issue compare and write Date: Fri, 17 Oct 2014 19:38:37 -0400 Message-ID: References: <1413437835-13778-1-git-send-email-michaelc@cs.wisc.edu> <1413437835-13778-3-git-send-email-michaelc@cs.wisc.edu> <20141017095555.GA360@infradead.org> Mime-Version: 1.0 Content-Type: text/plain Return-path: In-Reply-To: <20141017095555.GA360@infradead.org> (Christoph Hellwig's message of "Fri, 17 Oct 2014 02:55:55 -0700") Sender: target-devel-owner@vger.kernel.org To: Christoph Hellwig Cc: michaelc@cs.wisc.edu, linux-scsi@vger.kernel.org, target-devel@vger.kernel.org, ceph-devel@vger.kernel.org, axboe@kernel.dk, linux-kernel@vger.kernel.org List-Id: linux-scsi@vger.kernel.org >>>>> "Christoph" == Christoph Hellwig writes: Christoph> We already have req->cmd_type which actually is a fairly good Christoph> description of what we get except for REQ_TYPE_FS, which is a Christoph> horrible overload using req->cmd_flags. Christoph> Given that you're just one of many currently ongoing patches Christoph> to add more flags here I think you need to byte the bullet Christoph> and fix this up by replacing REQ_TYPE_FS with: Christoph> REQ_TYPE_WRITE REQ_TYPE_READ REQ_TYPE_FLUSH REQ_TYPE_DISCARD Christoph> REQ_TYPE_WRITE_SAME REQ_TYPE_CMP_AND_WRITE The problem with this is that, as it stands, a bio has no type. And it would suck if we couldn't keep bio rw and request flags in sync. I wonder if it would make more sense to move the remaining rq types to cmd_flags after I'm done with the 64-bit conversion? -- Martin K. Petersen Oracle Linux Engineering