关于virtio_net网卡命名的小问题

最近看了一个小问题，涉及到一致性网络设备命名(Consistent Network Device Naming)，在此记录一下。

系统是 4.18.0-240.el8.x86_64，centos 8.3

系统上的网卡如下：

1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
6: eno1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether b4:05:5d:d5:1e:1d brd ff:ff:ff:ff:ff:ff
    inet *.*.*.*/24 brd 10.20.25.255 scope global noprefixroute eno1
       valid_lft forever preferred_lft forever
    inet6 fe80::b605:5dff:fed5:1e1d/64 scope link
       valid_lft forever preferred_lft forever
7: eno2: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN group default qlen 1000
    link/ether b4:05:5d:d5:1e:1e brd ff:ff:ff:ff:ff:ff
8: eno3: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN group default qlen 1000
    link/ether b4:05:5d:d5:1e:1f brd ff:ff:ff:ff:ff:ff
9: eno4: <BROADCAST,MULTICAST> mtu 1500 qdisc mq state DOWN group default qlen 1000
    link/ether b4:05:5d:d5:1e:20 brd ff:ff:ff:ff:ff:ff
18: ens47: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 52:54:00:0e:26:13 brd ff:ff:ff:ff:ff:ff
19: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 52:54:00:6b:c6:c7 brd ff:ff:ff:ff:ff:ff
    inet 192.188.2.10/24 brd 192.188.2.255 scope global noprefixroute eth1
       valid_lft forever preferred_lft forever
    inet6 fe80::5054:ff:fe6b:c6c7/64 scope link noprefixroute
       valid_lft forever preferred_lft forever
20: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 52:54:00:09:b3:32 brd ff:ff:ff:ff:ff:ff
    inet 192.188.3.10/24 brd 192.188.3.255 scope global noprefixroute eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::5054:ff:fe09:b332/64 scope link noprefixroute
       valid_lft forever preferred_lft forever
21: eth2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 52:54:00:95:de:fd brd ff:ff:ff:ff:ff:ff
    inet 192.188.4.10/24 brd 192.188.4.255 scope global noprefixroute eth2
       valid_lft forever preferred_lft forever
    inet6 fe80::5054:ff:fe95:defd/64 scope link noprefixroute
       valid_lft forever preferred_lft forever

测试小伙伴查看网卡，发现同样是virtio_net设备，为啥有的是正常的eth0，有的是ens47。

[root]# ethtool -i ens47
**driver: virtio_net**
version: 1.0.0
firmware-version:
expansion-rom-version:
bus-info: 0000:c1:00.0
supports-statistics: yes
supports-test: no
supports-eeprom-access: no
supports-register-dump: no
supports-priv-flags: no
[root]# ethtool -i eth0
**driver: virtio_net**
version: 1.0.0
firmware-version:
expansion-rom-version:
bus-info: 0000:c3:00.0
supports-statistics: yes
supports-test: no
supports-eeprom-access: no
supports-register-dump: no
supports-priv-flags: no

virtio_net 驱动中关于网卡name的调用流程是：

static int virtnet_probe(struct virtio_device *vdev)
{
....
	/* Allocate ourselves a network device with room for our info */
	dev = alloc_etherdev_mq(sizeof(struct virtnet_info), max_queue_pairs);//caq:里面会设置name的格式eth%d
....
        err = register_netdev(dev);//caq:满足一个net_dev的语义
}

具体来看下 alloc_etherdev_mq 的行为：

struct net_device *alloc_etherdev_mqs(int sizeof_priv, unsigned int txqs,
				      unsigned int rxqs)
{
	return alloc_netdev_mqs(sizeof_priv, "eth%d", NET_NAME_UNKNOWN,
				ether_setup, txqs, rxqs);//caq:命名为eth，同时传参设置 name_assign_type 为 NET_NAME_UNKNOWN
}
/* interface name assignment types (sysfs name_assign_type attribute) */
#define NET_NAME_UNKNOWN	0	/* unknown origin (not exposed to userspace) *///caq:请注意，提示了不暴露给用户态
#define NET_NAME_ENUM		1	/* enumerated by kernel */
#define NET_NAME_PREDICTABLE	2	/* predictably named by the kernel */
#define NET_NAME_USER		3	/* provided by user-space */
#define NET_NAME_RENAMED	4	/* renamed by user-space */

在alloc_netdev_mqs中：

struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
		unsigned char name_assign_type,
		void (*setup)(struct net_device *),
		unsigned int txqs, unsigned int rxqs)
{
//caq:关于命名的两行关键代码：
	strcpy(dev->name, name);//caq:拷贝传入的eth%d字符串
	dev->name_assign_type = name_assign_type;//caq:此时为0
}

接下来，查看 register_netdev 的流程，

register_netdev语义中，关于name部分的流程为：

register_netdev--->dev_get_valid_name

int dev_get_valid_name(struct net *net, struct net_device *dev,
		       const char *name)
{
	BUG_ON(!net);

	if (!dev_valid_name(name))
		return -EINVAL;

	if (strchr(name, '%'))//caq:一般走这，比如eth%d
		return dev_alloc_name_ns(net, dev, name);
	else if (__dev_get_by_name(net, name))
		return -EEXIST;
	else if (dev->name != name)
		strlcpy(dev->name, name, IFNAMSIZ);

	return 0;
}

由于我们设置的设备初始name是eth%，所以走 dev_alloc_name_ns 分支，

dev_alloc_name_ns 会进一步调用 __dev_alloc_name：

dev_alloc_name_ns--->__dev_alloc_name

static int __dev_alloc_name(struct net *net, const char *name, char *buf)
{
	int i = 0;
	const char *p;
	const int max_netdevices = 8*PAGE_SIZE;
	unsigned long *inuse;
	struct net_device *d;

	if (!dev_valid_name(name))
		return -EINVAL;

	p = strchr(name, '%');
	if (p) {
		/*
		 * Verify the string as this thing may have come from
		 * the user.  There must be either one "%d" and no other "%"
		 * characters.
		 */
		if (p[1] != 'd' || strchr(p + 2, '%'))
			return -EINVAL;

		/* Use one page as a bit array of possible slots */
		inuse = (unsigned long *) get_zeroed_page(GFP_ATOMIC);
		if (!inuse)
			return -ENOMEM;

		for_each_netdev(net, d) {//caq:在同一个net_namespace中查找
			if (!sscanf(d->name, name, &i))
				continue;
			if (i < 0 || i >= max_netdevices)
				continue;

			/*  avoid cases where sscanf is not exact inverse of printf */
			snprintf(buf, IFNAMSIZ, name, i);
			if (!strncmp(buf, d->name, IFNAMSIZ))
				set_bit(i, inuse);//caq:先遍历一遍，把所有的已被使用的编号设置到inuse中
		}

		i = find_first_zero_bit(inuse, max_netdevices);//caq:然后从inuse中取未使用的第一个id
		free_page((unsigned long) inuse);
	}

	snprintf(buf, IFNAMSIZ, name, i);//caq:如果经历过上面循环，这个时候取的i肯定是没人使用的，如果没有经历循环，则不一定
	if (!__dev_get_by_name(net, buf))//caq:再次校验
		return i;//caq:返回NULL，说明找到的i是有效的

	/* It is possible to run out of possible slots
	 * when the name is long and there isn't enough space left
	 * for the digits, or if all bits are used.
	 */
	return -ENFILE;
}

以上就是获取ethx之类名字的流程。

了解了以上流程，下面进一步查看，为什么 我们的设备上出现了 测试同学认为的正常的eth0，eth1之类，也出现了所谓异常的ens47。

那是不是内核设置完eth之后，用户就直接使用eth呢？这个在之前说法是成立的，但是从以上的逻辑也可以看出，这个eth0和eth1，没有必然的先后顺序，

之后申请的先后顺序，有时候会出现复位之后eth0和eth1的顺序相反的情况，或者热插拔之后name被别人占用的情况。

所以，centos7之后，由于使用了systemd，systemd和udev引入了一种新的网络设备命名方式：一致网络设备命名（CONSISTENT NETWORK DEVICE NAMING）。

我们来查看一下网卡的udev信息：

[root]# udevadm info /sys/class/net/eth0
P: /devices/pci0000:ae/0000:ae:00.0/0000:af:00.0/0000:b0:12.0/0000:c3:00.0/virtio6/net/eth0
E: DEVPATH=/devices/pci0000:ae/0000:ae:00.0/0000:af:00.0/0000:b0:12.0/0000:c3:00.0/virtio6/net/eth0
E: ID_BUS=pci
E: ID_MODEL_FROM_DATABASE=Virtio network device
E: ID_MODEL_ID=0x1000
E: ID_NET_DRIVER=virtio_net
E: ID_NET_LINK_FILE=/usr/lib/systemd/network/99-default.link
**E: ID_NET_NAME=ens47**
**E: ID_NET_NAME_MAC=enx52540009b332**
**E: ID_NET_NAME_PATH=enp195s0**
**E: ID_NET_NAME_SLOT=ens47**
E: ID_NET_NAMING_SCHEME=rhel-8.0
E: ID_PATH=pci-0000:c3:00.0
E: ID_PATH_TAG=pci-0000_c3_00_0
E: ID_PCI_CLASS_FROM_DATABASE=Network controller
E: ID_PCI_SUBCLASS_FROM_DATABASE=Ethernet controller
E: ID_VENDOR_FROM_DATABASE=Red Hat, Inc.
E: ID_VENDOR_ID=0x1af4
E: IFINDEX=20
E: INTERFACE=eth0
E: SUBSYSTEM=net
E: SYSTEMD_ALIAS=/sys/subsystem/net/devices/ens47
E: TAGS=:systemd:
E: USEC_INITIALIZED=1080595193

[root]# udevadm info /sys/class/net/eth1
P: /devices/pci0000:ae/0000:ae:00.0/0000:af:00.0/0000:b0:11.0/0000:c2:00.0/virtio5/net/eth1
E: DEVPATH=/devices/pci0000:ae/0000:ae:00.0/0000:af:00.0/0000:b0:11.0/0000:c2:00.0/virtio5/net/eth1
E: ID_BUS=pci
E: ID_MODEL_FROM_DATABASE=Virtio network device
E: ID_MODEL_ID=0x1000
E: ID_NET_DRIVER=virtio_net
E: ID_NET_LINK_FILE=/usr/lib/systemd/network/99-default.link
**E: ID_NET_NAME=ens47**
**E: ID_NET_NAME_MAC=enx5254006bc6c7**
**E: ID_NET_NAME_PATH=enp194s0**
**E: ID_NET_NAME_SLOT=ens47**
E: ID_NET_NAMING_SCHEME=rhel-8.0
E: ID_PATH=pci-0000:c2:00.0
E: ID_PATH_TAG=pci-0000_c2_00_0
E: ID_PCI_CLASS_FROM_DATABASE=Network controller
E: ID_PCI_SUBCLASS_FROM_DATABASE=Ethernet controller
E: ID_VENDOR_FROM_DATABASE=Red Hat, Inc.
E: ID_VENDOR_ID=0x1af4
E: IFINDEX=19
E: INTERFACE=eth1
E: SUBSYSTEM=net
E: SYSTEMD_ALIAS=/sys/subsystem/net/devices/ens47
E: TAGS=:systemd:
E: USEC_INITIALIZED=1080405318

[root]# udevadm info /sys/class/net/eth3
Unknown device, --name=, --path=, or absolute path in /dev/ or /sys expected.
[root@192-188-2-10 network-scripts]# udevadm info /sys/class/net/eth2
P: /devices/pci0000:ae/0000:ae:00.0/0000:af:00.0/0000:b0:13.0/0000:c4:00.0/virtio7/net/eth2
E: DEVPATH=/devices/pci0000:ae/0000:ae:00.0/0000:af:00.0/0000:b0:13.0/0000:c4:00.0/virtio7/net/eth2
E: ID_BUS=pci
E: ID_MODEL_FROM_DATABASE=Virtio network device
E: ID_MODEL_ID=0x1000
E: ID_NET_DRIVER=virtio_net
E: ID_NET_LINK_FILE=/usr/lib/systemd/network/99-default.link
**E: ID_NET_NAME=ens47**
**E: ID_NET_NAME_MAC=enx52540095defd**
**E: ID_NET_NAME_PATH=enp196s0**
**E: ID_NET_NAME_SLOT=ens47**
E: ID_NET_NAMING_SCHEME=rhel-8.0
E: ID_PATH=pci-0000:c4:00.0
E: ID_PATH_TAG=pci-0000_c4_00_0
E: ID_PCI_CLASS_FROM_DATABASE=Network controller
E: ID_PCI_SUBCLASS_FROM_DATABASE=Ethernet controller
E: ID_VENDOR_FROM_DATABASE=Red Hat, Inc.
E: ID_VENDOR_ID=0x1af4
E: IFINDEX=21
E: INTERFACE=eth2
E: SUBSYSTEM=net
E: SYSTEMD_ALIAS=/sys/subsystem/net/devices/ens47
E: TAGS=:systemd:
E: USEC_INITIALIZED=1080613166

然后查看对应的日志：

 10:06:08 localhost.localdomain kernel: **virtio_net virtio4 ens47: renamed from eth0**//caq:将eth0重命名为ens47
 10:06:08 localhost.localdomain NetworkManager[1819]: <info>  [1641827168.5899] device (eth0): interface index 18 renamed iface from 'eth0' to 'ens47'//caq:这个是重命名的发起方
 10:06:08 localhost.localdomain systemd-udevd[3465]: could not rename interface '19' from 'eth1' to 'ens47': File exists//caq:eth1根据规则，也需要重命名为ens47，但由于ens47已经被占用，所以命名失败
 10:06:08 localhost.localdomain systemd-udevd[3468]: could not rename interface '20' from 'eth0' to 'ens47': File exists//caq:eth1根据规则，也需要重命名为ens47，但由于ens47已经被占用，所以命名失败
 10:06:08 localhost.localdomain systemd-udevd[3474]: could not rename interface '21' from 'eth2' to 'ens47': File exists//caq:eth1根据规则，也需要重命名为ens47，但由于ens47已经被占用，所以命名失败

而ens47的来源是因为如下的命名规则：

二、Centos7中的命名策略

Scheme 1: 如果从BIOS中能够取到可用的，板载网卡的索引号，则使用这个索引号命名，例如: eno1，如不能则尝试Scheme 2

Scheme 2: 如果从BIOS中能够取到可以用的，网卡所在的PCI-E热插拔插槽的索引号，则使用这个索引号命名，例如: ens1，如不能则尝试Scheme 3

Scheme 3：如果能拿到设备所连接的物理位置信息，则使用这个信息命名，例如:enp2s0，如不能则尝试Scheme 4

Scheme 4 使用网卡的MAC地址来命名，这个方法一般不使用。

Scheme 5：传统的kernel命名方法，例如: eth0，这种命名方法的结果不可预知的，即可能第二块网卡对应eth0，第一块网卡对应eth1。

参考资料：

https://www.cnblogs.com/zyd112/p/8143464.html

https://www.cnblogs.com/yinfutao/p/9634350.html

https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/