From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga04.intel.com (mga04.intel.com [192.55.52.120]) by dpdk.org (Postfix) with ESMTP id DAC07E5D for ; Thu, 28 Dec 2017 09:44:14 +0100 (CET) X-Amp-Result: UNKNOWN X-Amp-Original-Verdict: FILE UNKNOWN X-Amp-File-Uploaded: False Received: from fmsmga002.fm.intel.com ([10.253.24.26]) by fmsmga104.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 28 Dec 2017 00:44:13 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.45,470,1508828400"; d="scan'208";a="5422886" Received: from debian-xvivbkq.sh.intel.com ([10.67.104.165]) by fmsmga002.fm.intel.com with ESMTP; 28 Dec 2017 00:44:12 -0800 Date: Thu, 28 Dec 2017 16:43:45 +0800 From: Tiwei Bie To: Ning Li Cc: 'Yuanhan Liu' , 'Maxime Coquelin' , dev@dpdk.org Message-ID: <20171228084345.li5xfnfkbuahp4ur@debian-xvivbkq.sh.intel.com> References: <1513251494-9980-1-git-send-email-muziding001@163.com> <20171218061907.fo3ausw5msfhic6q@debian-xvivbkq> <002401d378b1$46e25960$d4a70c20$@163.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <002401d378b1$46e25960$d4a70c20$@163.com> User-Agent: NeoMutt/20170609 (1.8.3) Subject: Re: [dpdk-dev] [PATCH] net/virtio-user: specify MAC address for tap port X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 28 Dec 2017 08:44:15 -0000 Hi Ning, On Tue, Dec 19, 2017 at 06:08:10PM +0800, Ning Li wrote: > Hi Tiwei, > > > Hi Ning, > > > > On Thu, Dec 14, 2017 at 07:38:14PM +0800, Ning Li wrote: > > > When use virtio_user as exception path, we need to specify a MAC > > > address for the tap port. > > > > Is this a fix? Did you meet any issue? If so, please describe > > the issue and add a fixline. > > Specify the MAC address for tap0 may be a requirement for some applications. > > As described in doc/guides/howto/virtio_user_as_exceptional_path.rst, when > virtio-user with vhost-kernel is used to exchange packet with kernel networking > stack, instead of KNI, application need to set the physical NIC in promiscuous > mode, otherwise, the packet send to tap0 will be discarded by physical NIC. So, > this will be a probleam, when some applications not set NIC in promiscuous mode. > This problem will be easy to solve, If application can specify the MAC address > of the NIC for tap0. > Thanks for the detailed info! >>From my understanding, it's not necessary to set the MAC of the tap via virtio-user. Because the tap is a virtual interface which belongs to kernel, and it's connected to the virtual interface which is emulated by virtio-user. So actually they are different virtual interfaces. If an application wants the virtio-user, tap and any other NIC share the same MAC address, it just needs to set the MAC for each interface individually in the standard way. Any thoughts? Even if we really want to support specifying the MAC via virtio-user for the corresponding tap, the MAC should be passed via another device argument, e.g. tap_mac. Besides, I think you missed my other comments in my last mail. E.g. The patch isn't buildable. Thanks, Tiwei > Thanks, > Ning Li > > > > > > > > > Signed-off-by: Ning Li > > > --- > > > drivers/net/virtio/virtio_user/vhost_kernel.c | 3 ++- > > > drivers/net/virtio/virtio_user/vhost_kernel_tap.c | 8 ++++++++ > > > drivers/net/virtio/virtio_user/vhost_kernel_tap.h | 12 +++++++++++- > > > 3 files changed, 21 insertions(+), 2 deletions(-) > > > > > > diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c > > b/drivers/net/virtio/virtio_user/vhost_kernel.c > > > index 68d28b1..dd24b6b 100644 > > > --- a/drivers/net/virtio/virtio_user/vhost_kernel.c > > > +++ b/drivers/net/virtio/virtio_user/vhost_kernel.c > > > @@ -380,7 +380,8 @@ struct vhost_memory_kernel { > > > else > > > hdr_size = sizeof(struct virtio_net_hdr); > > > > > > - tapfd = vhost_kernel_open_tap(&dev->ifname, hdr_size, req_mq); > > > + tapfd = vhost_kernel_open_tap(&dev->ifname, hdr_size, req_mq, > > > + (char *)dev->mac_addr); > > > if (tapfd < 0) { > > > PMD_DRV_LOG(ERR, "fail to open tap for vhost kernel"); > > > return -1; > > > diff --git a/drivers/net/virtio/virtio_user/vhost_kernel_tap.c > > b/drivers/net/virtio/virtio_user/vhost_kernel_tap.c > > > index 689a5cf..756fde2 100644 > > > --- a/drivers/net/virtio/virtio_user/vhost_kernel_tap.c > > > +++ b/drivers/net/virtio/virtio_user/vhost_kernel_tap.c > > > @@ -123,6 +123,14 @@ > > > > You forgot to add the mac param for vhost_kernel_open_tap(). > > The patch isn't buildable on my machine. > > > > > PMD_DRV_LOG(ERR, "TUNSETOFFLOAD ioctl() failed: %s", > > > strerror(errno)); > > > > > > + memset(&ifr, 0, sizeof(ifr)); > > > + ifr.ifr_hwaddr.sa_family = ARPHRD_ETHER; > > > > ARPHRD_ETHER? Please explain. > > > > > + memcpy(ifr.ifr_hwaddr.sa_data, mac, ETH_ALEN); > > > > You can use ETHER_ADDR_LEN. > > There is no need to define ETH_ALEN. > > > > > + if (ioctl(tapfd, SIOCSIFHWADDR, (void *)&ifr) == -1) { > > > + PMD_DRV_LOG(ERR, "SIOCSIFHWADDR failed: %s", strerror(errno)); > > > + goto error; > > > + } > > > + > > > if (!(*p_ifname)) > > > *p_ifname = strdup(ifr.ifr_name); > > > > > > diff --git a/drivers/net/virtio/virtio_user/vhost_kernel_tap.h > > b/drivers/net/virtio/virtio_user/vhost_kernel_tap.h > > > index eae340c..f59b1ac 100644 > > > --- a/drivers/net/virtio/virtio_user/vhost_kernel_tap.h > > > +++ b/drivers/net/virtio/virtio_user/vhost_kernel_tap.h > > > @@ -64,4 +64,14 @@ > > > /* Constants */ > > > #define PATH_NET_TUN "/dev/net/tun" > > > > > > -int vhost_kernel_open_tap(char **p_ifname, int hdr_size, int req_mq); > > > +/* Socket configuration controls. */ > > > +#define SIOCSIFHWADDR 0x8924 /* set hardware address */ > > > > There is no need to define this. > > > > > + > > > +/* ARP protocol HARDWARE identifiers. */ > > > +#define ARPHRD_ETHER 1 /* Ethernet > > */ > > > + > > > +/* Length of Ethernet address. */ > > > +#define ETH_ALEN 6 > > > + > > > > There is no need to define this. > > > > Thanks, > > Tiwei > > > > > +int vhost_kernel_open_tap(char **p_ifname, int hdr_size, int req_mq, > > > + const char *mac); > > > -- > > > 1.8.3.1 > > > > > > > >