From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sanjeev Jorapur Subject: Re: [PATCH 3/3] NetXen: 64-bit memory fixes, driver cleanup Date: Thu, 23 Nov 2006 08:32:52 -0800 Message-ID: <1164299572.4505.9.camel@nbsanjeev> References: <200611171536.kAHFa3os005563@dut39.unminc.com> <20061117084049.555e14d0@localhost.localdomain> <1163845689.4822.6.camel@nbsanjeev> <20061120104553.434b6ec5@freekitty> Reply-To: sanjeev@netxen.com Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: "Linsys Contractor Amit S. Kale" , netdev , brazilnut , jeff@garzik.org, netxenproj@linsyssoft.com, rob@netxen.com, wendyx Return-path: Received: from adsl-63-194-91-143.dsl.snfc21.pacbell.net ([63.194.91.143]:5029 "EHLO nbsanjeev.unminc.com") by vger.kernel.org with ESMTP id S1757411AbWKWQh2 (ORCPT ); Thu, 23 Nov 2006 11:37:28 -0500 To: Stephen Hemminger In-Reply-To: <20061120104553.434b6ec5@freekitty> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Mon, 2006-11-20 at 10:45 -0800, Stephen Hemminger wrote: > On Sat, 18 Nov 2006 15:58:09 +0530 > Sanjeev Jorapur wrote: > > > On Fri, 2006-11-17 at 08:40 -0800, Stephen Hemminger wrote: > > > On Fri, 17 Nov 2006 07:36:03 -0800 > > > "Linsys Contractor Amit S. Kale" wrote: > > > > > > > NetXen: 1G/10G Ethernet Driver updates > > > > - These fixes take care of driver on machines with >4G memory > > > > - Driver cleanup > > > > > > > > Signed-off-by: Amit S. Kale > > > > > > > > netxen_nic.h | 41 ++++++---- > > > > netxen_nic_ethtool.c | 19 ++-- > > > > netxen_nic_hw.c | 10 +- > > > > netxen_nic_hw.h | 4 > > > > netxen_nic_init.c | 51 +++++++++++- > > > > netxen_nic_isr.c | 3 > > > > netxen_nic_main.c | 204 +++++++++++++++++++++++++++++++++++++++++++++++--- > > > > netxen_nic_phan_reg.h | 10 +- > > > > 8 files changed, 293 insertions(+), 49 deletions(-) > > > > > > > > > > > > > If you can't DMA from high memory, then don't set NETIF_F_HIGHDMA, why do > > > you need explicit bounce buffers. If you can't DMA from unaligned address, > > > the write a small routine to copy the skb to a new one. > > > > > > The hardware supports DMA into 35 bit addresses. The intent is to > > enable DMA into addresses upto 32G. > > > > You should then set the same value for pci_set_dma_mask, because then the IOMMU > can help. See both b44 or tg3 drivers, they have to deal with odd size masks. > I don't think you have to do all the bounce buffer work in the driver. > We had tried something like this earlier, but found that on some platforms (Opteron, IA64), we got kernel panics when the kernel ran out of translation entries. We will re-try this and let you know. Sanjeev.