From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from bombadil.infradead.org ([65.50.211.133]:37753 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933334AbdC3MGl (ORCPT ); Thu, 30 Mar 2017 08:06:41 -0400 Date: Thu, 30 Mar 2017 05:06:39 -0700 From: Christoph Hellwig To: Chuck Lever Cc: linux-rdma@vger.kernel.org, linux-nfs@vger.kernel.org Subject: Re: [PATCH v2 01/13] svcrdma: Move send_wr to svc_rdma_op_ctxt Message-ID: <20170330120639.GA31144@infradead.org> References: <20170327132329.5585.98465.stgit@klimt.1015granger.net> <20170327134802.5585.70150.stgit@klimt.1015granger.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <20170327134802.5585.70150.stgit@klimt.1015granger.net> Sender: linux-nfs-owner@vger.kernel.org List-ID: > + svc_rdma_build_send_wr(ctxt, sge_no); > + send_wr = &ctxt->send_wr; > + if (inv_rkey) { > + send_wr->opcode = IB_WR_SEND_WITH_INV; > + send_wr->ex.invalidate_rkey = inv_rkey; > + } Seems a bit odd to overwrite the work that svc_rdma_build_send_wr in the caller. I would propbably add this work to svc_rdma_build_send_wr. Also any reason svc_rdma_build_send_wr can't call svc_rdma_send directly (although it would probably need a new name then)?