Networking and Ethernet devices A short network guide...

This page collects a lot of useful information about how to configure Ethernet interfaces under Linux, how to test the VLAN support, how to use the ethtool support etc.
Although, this page collects generic network information for Linux, the targets used as example are the MAC/GMAC embedded devices.

Enable/Disable the interface

After booting the kernel, it's possible to manually enable/disable the interface by usig the following command:

• ifconfig eth0 down/up

How to set the MAC Address

• Bring down the interface by ifconfig eth0 down.
• Assign a MAC address by ifconfig hw ether eth0 <new_MAC_address>.
• Bring up the interface by ifconfig eth0 up.

How to modify the MTU

• ifconfig eth0 down
• ifconfig eth0 mtu <new_value>
• ifconfig eth0 up

VLAN support

VLAN is an implementation of the 802.1Q VLAN protocol for Linux. It allows one to have Virtual LANs on a single Ethernet cable.

To test it, follow the steps belows:

• Make sure that the kernel has the VLAN support
• Create the VLAN interface by adding vlan ID. To do that you have to use the vconfig command as shown below: vconfig add eth0 <vlan ID> where vlan ID is any integer number. Vlan ID on both the machines should be same
• Assign IP address to the vlan interface using ifconfig eth0,2 <IP address> up
• Ping the remote machine
• So, the VLAN interface should be created successfully. This can be verified by using ifconfig -a command.
• Ping should be successful without any error.

Hw notes

Latest MAC/GMAC cores support IEEE 802.1Q VLAN tag detection for reception frames without vlan filtering.
The MAC core compares the TIPD field (bytes 13,14 within the frame) and, if they match with the value in the VLAN Tag register, the length of the frame is incremented from 1518 to 1522 octets.

Ethtool Support

Display or change ethernet card settings

How to get general driver information

It's possible to get the interface information related to link such as ports, speed, duplex, auto negotiation (ANE) etc. are displayed correctly by using:

root@:~# ethtool eth0
Settings for eth0:
        Supported ports: [ TP MII ]
        Supported link modes:   10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
        Supports auto-negotiation: Yes
        Advertised link modes:  10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
        Advertised auto-negotiation: Yes
        Speed: 100Mb/s
        Duplex: Full
        Port: MII
        PHYAD: 14
        Transceiver: external
        Auto-negotiation: on
        Supports Wake-on: d
        Wake-on: d
        Current message level: 0x0000003f (63)
        Link detected: yes

How to change the debug level

ethtool -s eth0 msglvl 0xffff

Where 0xffff is the maximum debug level for these drivers.

Note: the driver debug level has been written following the NETIF linux standard message level.
Please see the Documentation/networking/netif-msg.tx file included in the Linux Kernel source tree.

Dump registers

Starting from the STLinux-2.3, Ethtool has been patched in order to dump the mac 10/100/1000 registers.

root@:~# ethtool -d eth0
MAC 10/100 Registers
control reg  0x1010210C
addr HI 0x00004718
addr LO 0x12E18000
multicast hash HI 0x00000000
multicast hash LO 0x80000000
MII addr 0x00007440
MII data 00000380
flow control 0x02000002
VLAN1 tag 0x0000FFFF
VLAN2 tag 0x0000FFFF
mac wakeup frame 0x00000000
mac wakeup crtl 0x00000000

DMA Registers
CSR0  0x00002000
CSR1  0x00000000
CSR2  0x00000000
CSR3  0x42C06000
CSR4  0x433D9000
CSR5  0x00660004
CSR6  0x0040A002
CSR7  0x00018041
CSR8  0x00000000
DMA cur tx buf addr 0x00000000
DMA cur rx buf addr 0x00000000

Get extra statistics

root@:~# ethtool -S eth0
NIC statistics:
     tx_underflow: 0
     tx_carrier: 0
     tx_losscarrier: 0
     tx_heartbeat: 0
     tx_deferred: 0
     tx_vlan: 0
     tx_jabber: 0
     tx_frame_flushed: 0
     rx_desc: 0
     rx_partial: 0
     rx_runt: 0
     rx_toolong: 0
     rx_collision: 0
     rx_crc: 0
     rx_lenght: 0
     rx_mii: 0
     rx_multicast: 0
     rx_overflow: 0
     rx_watchdog: 0
     rx_filter: 0
     rx_dropped: 0
     rx_bytes: 0
     tx_bytes: 0
     tx_irq_n: 0
     rx_irq_n: 0
     tx_undeflow_irq: 0
     tx_threshold: 32
     tx_process_stopped_irq: 0
     tx_jabber_irq: 0
     rx_overflow_irq: 0
     rx_buf_unav_irq: 0
     rx_process_stopped_irq: 0
     rx_watchdog_irq: 0
     tx_early_irq: 0
     fatal_bus_error_irq: 0
     rx_poll_n: 0

Flow Control

Latest MAC and GMAC Embedded devices allow flow control in Full Duplex by using the PAUSE operation and control frame transmission. They also support FC in Half Duplex using back pressure.

See the driver's command line options for details about how to change the pause value and turn-on flow-control.

It's possible to change the pause parameters by using ethtool as shown below:

ethtool -A ethX [autoneg on|off] [rx on|off] [tx on|off]

Offload parameters

Get offload information

# ethtool -k eth0
Offload parameters for eth0:
rx-checksumming: on
tx-checksumming: on
scatter-gather: on
tcp segmentation offload: off
udp fragmentation offload: off
generic segmentation offload: off

Turn-on scatter-gather (enabled by default for the stmmac driver)

ethtool -K eth0 sg on

Turn-on rx/tx checksumming

ethtool -K eth0 rx on

ethtool -K eth0 tx on

Multicast

Multicast packets are used to send the same data to multiple network devices at the same time, with the same packet stream.
One common situation in which it is used is when it is necessary to distribute real time audio and video.

Configuring the Kernel

Kernel configuration requires CONFIG_IP_MULTICAST.

Enable/disable all-multicast mode with ifconfig

• Disable promisc by running ifconfig eth0 -promisc
• Enable the allmulti mode run the command ifconfig eth0 allmulti
• Ping some multicast IP address from remote machine: ping -I eth0 <multicast_ip_addr>
• On the target enabled with allmulti mode check for RX packets count
• Do allmulti off by running ifconfig eth0 -allmulti on target.

As expected results: if allmulti is enabled, ifconfig eth0 output should show ALLMULTI in second line

Testing multicast

Make sure the kernel will not ignore all ICMP ECHO and TIMESTAMP requests sent to it via broadcast/multicast.
So executing on the target the following command:

target# echo 0 > /proc/sys/net/ipv4/icmp_echo_ignore_broadcasts

Sets all of the class D (multicast) IP routes to go via "eth0":

target#  route add -net 224.0.0.0 netmask 240.0.0.0 dev eth0

Join a multicast group:

target# ./mjoin eth0 224.1.0.37

Get here the mjoin C source.

Ping multicast ip address:

host# ping -r -I eth0 224.1.0.37

To display multicast group membership information on the target:

target# netstat -g
IPv6/IPv4 Group Memberships
Interface       RefCnt Group
--------------- ------ ---------------------
lo              1      224.0.0.1
eth0            1      224.1.0.37
eth0            1      224.0.0.1