[dpdk-dev] [PATCH v2] vhost: add interface name to virtio-net struct

[Ananyev, Konstantin]

Hi Ciara,

> -----Original Message-----
> From: Loftus, Ciara
> Sent: Thursday, December 18, 2014 6:07 PM
> To: dev at dpdk.org
> Cc: thomas.monjalon at 6wind.com; stephen at networkplumber.org; vincent.jardin at 6wind.com; Ananyev, Konstantin; Czesnowicz,
> Przemyslaw; Loftus, Ciara; Anthony Fee
> Subject: [PATCH v2] vhost: add interface name to virtio-net struct
> 
> From: Ciara Loftus <ciara.loftus at intel.com>
> 
> This patch fixes the issue whereby when using userspace vhost ports
> in the context of vSwitching, the name provided to the hypervisor/QEMU
> of the vhost tap device needs to be exposed in the library, in order
> for the vSwitch to be able to direct packets to the correct device.
> This patch introduces an 'ifname' member to the virtio-net structure
> which is populated with the tap device name when QEMU is brought up
> with a vhost device.
> 
> Signed-off-by: Ciara Loftus <ciara.loftus at intel.com>
> Signed-off-by: Anthony Fee <anthonyx.fee at intel.com>
> Acked-by: Huawei Xie <huawei.xie at intel.com>
> ---
>  lib/librte_vhost/rte_virtio_net.h |    3 ++
>  lib/librte_vhost/virtio-net.c     |   48 ++++++++++++++++++++++++++++++++++++-
>  2 files changed, 50 insertions(+), 1 deletions(-)
> 
> diff --git a/lib/librte_vhost/rte_virtio_net.h b/lib/librte_vhost/rte_virtio_net.h
> index 00b1328..0bf07c7 100644
> --- a/lib/librte_vhost/rte_virtio_net.h
> +++ b/lib/librte_vhost/rte_virtio_net.h
> @@ -43,6 +43,8 @@
>  #include <linux/virtio_ring.h>
>  #include <linux/virtio_net.h>
>  #include <sys/eventfd.h>
> +#include <sys/socket.h>
> +#include <linux/if.h>
> 
>  #include <rte_memory.h>
>  #include <rte_mempool.h>
> @@ -96,6 +98,7 @@ struct virtio_net {
>  	uint64_t		features;	/**< Negotiated feature set. */
>  	uint64_t		device_fh;	/**< device identifier. */
>  	uint32_t		flags;		/**< Device flags. Only used to check if device is running on data core. */
> +	char			ifname[IFNAMSIZ];	/**< Name of the tap device. */
>  	void			*priv;		/**< private context */
>  } __rte_cache_aligned;
> 
> diff --git a/lib/librte_vhost/virtio-net.c b/lib/librte_vhost/virtio-net.c
> index 852b6d1..7eae5ee 100644
> --- a/lib/librte_vhost/virtio-net.c
> +++ b/lib/librte_vhost/virtio-net.c
> @@ -43,6 +43,10 @@
>  #include <sys/mman.h>
>  #include <unistd.h>
> 
> +#include <sys/socket.h>
> +#include <linux/if_tun.h>
> +#include <linux/if.h>
> +
>  #include <rte_ethdev.h>
>  #include <rte_log.h>
>  #include <rte_string_fns.h>
> @@ -1000,6 +1004,46 @@ set_vring_kick(struct vhost_device_ctx ctx, struct vhost_vring_file *file)
>  }
> 
>  /*
> + * Function to get the tap device name from the provided file descriptor and
> + * save it in the device structure.
> + */
> +static int
> +get_ifname(struct virtio_net *dev, int tap_fd, int pid)
> +{
> +	struct eventfd_copy fd_tap;
> +	struct ifreq ifr;
> +	uint32_t size, ifr_size;
> +	int ret;
> +
> +	fd_tap.source_fd = eventfd(0, EFD_NONBLOCK | EFD_CLOEXEC);
> +	fd_tap.target_fd = tap_fd;
> +	fd_tap.target_pid = pid;
> +
> +	if (eventfd_copy(dev, &fd_tap))
> +		return -1;
> +
> +	ret = ioctl(fd_tap.source_fd, TUNGETIFF, &ifr);
> +
> +	if (close(fd_tap.source_fd) < 0)
> +		RTE_LOG(ERR, VHOST_CONFIG,
> +			"(%"PRIu64") fd close failed\n",
> +			dev->device_fh);
> +
> +	if (ret >= 0) {
> +		ifr_size = strnlen(ifr.ifr_name, sizeof(ifr.ifr_name));
> +		size = ifr_size > sizeof(dev->ifname)?
> +				sizeof(dev->ifname): ifr_size;

If you saying that  sizeof(ifr.ifr_name) would always be equal to sizeof(dev->ifname),
then why  are you doing the comparison above?
"ifr_size <= sizeof(dev->ifname)" would always be true, right?
>From other side, if  if strlen(ifr.ifr_name)  < sizeof(dev->ifname), you wouldn't copy over terminating 0.
I think it needs to be something like:

size = strnlen(ifr.ifr_name, sizeof(ifr.ifr_name)) + 1;
size = RTE_MIN(size, sizeof(dev->ifname));
strncpy(dev->ifname, ifr.ifr_name, size);
dev->ifname[sizeof(dev->ifname) - 1] = 0;

> +
> +		strncpy(dev->ifname, ifr.ifr_name, size);
> +	} else
> +		RTE_LOG(ERR, VHOST_CONFIG,
> +			"(%"PRIu64") TUNGETIFF ioctl failed\n",
> +			dev->device_fh);
> +
> +	return 0;

Shouldn't we return a failure if ret is negative (ioclt failed)? 
Konstantin

> +}
> +
> +/*
>   * Called from CUSE IOCTL: VHOST_NET_SET_BACKEND
>   * To complete device initialisation when the virtio driver is loaded,
>   * we are provided with a valid fd for a tap device (not used by us).
> @@ -1026,8 +1070,10 @@ set_backend(struct vhost_device_ctx ctx, struct vhost_vring_file *file)
>  	 */
>  	if (!(dev->flags & VIRTIO_DEV_RUNNING)) {
>  		if (((int)dev->virtqueue[VIRTIO_TXQ]->backend != VIRTIO_DEV_STOPPED) &&
> -			((int)dev->virtqueue[VIRTIO_RXQ]->backend != VIRTIO_DEV_STOPPED))
> +			((int)dev->virtqueue[VIRTIO_RXQ]->backend != VIRTIO_DEV_STOPPED)) {
> +			get_ifname(dev, file->fd, ctx.pid);
>  			return notify_ops->new_device(dev);
> +		}
>  	/* Otherwise we remove it. */
>  	} else
>  		if (file->fd == VIRTIO_DEV_STOPPED)
> --
> 1.7.4.1