From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [PATCH] bnx2fc: fix incorrect DMA memory mapping in bnx2fc_map_sg() Date: Fri, 22 Aug 2014 14:11:06 +0200 Message-ID: <20140822121106.GA14110@lst.de> References: <1407140413-4561-1-git-send-email-mlombard@redhat.com> <1408556122.10706.20.camel@lbirv-waie-lx2.broadcom.com> <53F7091C.1080400@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from verein.lst.de ([213.95.11.211]:36117 "EHLO newverein.lst.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753273AbaHVMLK (ORCPT ); Fri, 22 Aug 2014 08:11:10 -0400 Content-Disposition: inline In-Reply-To: Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Chad Dupuis Cc: Maurizio Lombardi , Eddie Wai , QLogic-Storage-Upstream@qlogic.com, linux-scsi@vger.kernel.org, JBottomley@parallels.com, hch@lst.de On Fri, Aug 22, 2014 at 08:08:33AM -0400, Chad Dupuis wrote: > Eddie, Maurizio, > > Since it looks like there can be a difference in the pdev would it make > sense instead to convert the unmap function to use the bare DMA API so > we're consistent in our use of hba->pcidev->dev? Maybe something like > this: This looks like a better plan to me. Can you also document why you are using plain dma_map/unmap_sg and a different dma device in the code?