Article updated
at 10 Jun 2021
The Toradex Yocto BSP reference images currently provide two utilities for network configuration: connman and systemd-networkd. By default, the Ethernet port is managed by connman, and the USB RNDIS interface by systemd-networkd. Depending on the application, it might make sense to use one of those utilities over the other.
By default, connman will try to configure any available network interfaces and tries to keep only (the best) connection connected at any time. The design is targeted towards interactive, wireless devices such as tablets, etc.
The systemd provided utility systemd-networkd, on the other hand, emphasizes headless server systems with static configuration.
Below are instructions on how to use connman and/or systemd-networkd, or how to replace connman with NetworkManager on Yocto.
TorizonCore uses NetworkManager. Even though this article has some information about it, it's meant for people using the Yocto BSP reference images, not people using TorizonCore.
If you are using TorizonCore, go directly to Networking with TorizonCore.
Our latest V2.x images use connman as the overall connection management framework. The GUI management tool can be started with connman-properties or by starting the taskbar applet connman-applet.
The command-line interface tool connmanctl can be used in an interactive shell mode or as a single command using arguments.
root@apalis-t30:~# connmanctl services
*AO Wired { ethernet_00142d269dfd_cable }
root@apalis-t30:~# connmanctl services ethernet_00142d269dfd_cable
[ ethernet_00142d269dfd_cable ]
Type = ethernet
Security = [ ]
State = online
Favorite = True
Immutable = False
AutoConnect = True
Name = Wired
Ethernet = { Method=auto Interface=enp7s0 Address=00:14:2D:26:9D:FD MTU=1500 }
IPv4 = { Method=dhcp Address=192.168.10.162 Netmask=255.255.255.0 Gateway=192.168.10.1 }
IPv4.Configuration = { Method=dhcp }
IPv6 = { }
IPv6.Configuration = { Method=auto Privacy=disabled }
Nameservers = [ 8.8.8.8 ]
Nameservers.Configuration = [ ]
Timeservers = [ 192.168.10.1 ]
Timeservers.Configuration = [ ]
Domains = [ colibri.net ]
Domains.Configuration = [ ]
Proxy = { URL=http://wpad.colibri.net/wpad.dat Method=auto }
Proxy.Configuration = { }
Provider = { }
Using the interactive mode to configure a static Ethernet address
root@colibri-t30:~# connmanctl
connmanctl> config ethernet_00142d259a48_cable --ipv4 manual 192.168.10.2 255.255.255.0 192.168.10.1
connmanctl> exit
Using the interactive mode to configure the nameservers (DNS)
root@colibri-t30:~# connmanctl config ethernet_00142d259a48_cable --nameservers 8.8.8.8
Using the interactive mode to get an IP automatically via DHCP
root@colibri-t30:~# connmanctl
connmanctl> config ethernet_00142d259a48_cable --ipv4 dhcp
connmanctl> exit
Connman can be used for tethering which allows sharing a connection of a device for another locally connected device. The Toradex module running Connman would be the device sharing the connection.
Tethering puts some requirements on the kernel. The README in the Connman source tree holds more information on these requirements.
Note: Since image V2.4 systemd-networkd has control over the USB gadget Ethernet interface (RNDIS). Remove the rndis.network file under /etc/systemd/networkd and make sure usb/rndis is not part of NetworkInterfaceBlacklist in /etc/connman/main.conf.
To enable tethering on the USB gadget interface make sure the USB gadget driver is loaded/enabled first (see USB Device Mode (Linux)). Then Connman can be used to enable tethering mode:
# connmanctl
connmanctl> tether gadget on
connmanctl> [ 194.725436] device usb0 entered promiscuous mode
[ 194.737525] tether: port 1(usb0) entered forwarding state
[ 194.743077] tether: port 1(usb0) entered forwarding state
Enabled tethering for gadget
To preconfigure Connman to automatically start Gadget interfaces upon start-up two configuration files are needed:
The main.conf under /etc/connman/main.conf
[General]
PersistentTetheringMode=true
TetheringTechnologies=gadget
The service settings file under /var/lib/connman/settings
[global]
OfflineMode=false
[Wired]
Enable=true
Tethering=false
[Gadget]
Enable=true
Tethering=true
For other configuration options, the Python interface documentation has some useful configuration examples: https://pythonhosted.org/pyconnman/config.html
By default, ConnMan is configured to only have one connected technology. This means it will connect only to one interface after boot, even if there are multiply available and configured to auto-connect. But sometimes one wants to have two or more interfaces to be auto connected, e.g. ethernet and WiFi. To achieve this, one can change the ConnMan configuration file at /etc/connman/main.conf, e.g. to enable auto-connect for both ethernet and WiFi, change the following entries accordingly in the configuration file:
/etc/connman/main.conf
PreferredTechnologies=wifi,ethernet
SingleConnectedTechnology=false
Check the man-page of the ConnMan configuration file for more details.
If you want to use NetworkManager, remove connman and add NetworkManager on a custom OpenEmbedded build. See an example below:
local.conf
# Remove connman
IMAGE_INSTALL_remove = " connman"
IMAGE_INSTALL_remove = " connman-client"
IMAGE_INSTALL_remove = " connman-gnome"
IMAGE_INSTALL_remove = " connman-plugin-wifi"
IMAGE_INSTALL_remove = " connman-plugin-ethernet"
IMAGE_INSTALL_remove = " connman-plugin-loopback"
# Add NetworkManager
IMAGE_INSTALL_append = " networkmanager"
IMAGE_INSTALL_append = " modemmanager"
Check out the OpenEmbedded (core) article to learn how to build and customize a sample Linux image with OpenEmbedded.
If you plan to use systemd-networkd as the primary network configuration utility, you should remove connman first (when using a console-only image, you won't need to remove connman-gnome). You can use the package manager to remove the package from pre-built images by invoking
opkg remove connman-gnome connman
To remove connman when building an image using OpenEmbedded remove CONMANPKGS from IMAGE_INSTALL in the image recipe recipes/images/angstrom-lxde-image.bb in the meta-toradex layer (see OpenEmbedded (core) how to build a complete image).
Warning: if you are using TorizonCore, go directly to Networking with TorizonCore - Static Network Configuration.
Create a file named 'wired.network' under /etc/systemd/network/
root@colibri-vf:~# vi /etc/systemd/network/wired.network
For dynamic IP connection fill in the file with the following content:
[Match]
Name=eth0
[Network]
DHCP=ipv4
For static IP connection:
[Match]
Name=eth0
[Network]
Address=192.168.0.1/24
Gateway=192.168.0.254
DNS=8.8.8.8
Using systemd-networkd one can easily setup a network bridge between two Ethernet interfaces:
/etc/systemd/network/eth.network
[Match]
Name=eth*
[Network]
Bridge=br0
/etc/systemd/network/bridged.network
[Match]
Name=br0
[Network]
DHCP=ipv4
This example configures a dynamic IP configuration on the ethernet device (untagged) and also creates a new Ethernet interface associated with VLAN 100 (tagged). The example uses a dynamic IP configuration on VLAN 100 too.
/etc/systemd/network/wired.network
[Match]
Name=eth0
[Network]
DHCP=ipv4
VLAN=eth0.100
/etc/systemd/network/eth0.100.netdev
[NetDev]
Name=eth0.100
Kind=vlan
[VLAN]
Id=100
/etc/systemd/network/vlan100.network
[Match]
Name=eth0.100
[Network]
DHCP=v4
The integration should be done similarly to the existing rndis.network file, see bbappend and files subdirectory of our systemd configuration:
First place the above file 'wired.network' inside systemd folder:
~/oe-core/stuff/meta-toradex/recipes-core/systemd/systemd
Than edit the bbappend file adding the 'wired.network' file like bellow:
vi ~/oe-core/stuff/meta-toradex/recipes-core/systemd/systemd_%.bbappend
FILESEXTRAPATHS_prepend := "${THISDIR}/systemd:"
SRC_URI += " \
file://wired.network \
file://rndis.network \
file://0001-systemd-udevd.service.in-set-MountFlags-shared.patch \
"
PACKAGECONFIG_append = " networkd"
do_install_append() {
# The network files need to be in /usr/lib/systemd, not ${systemd_unitdir}...
install -d ${D}${prefix}/lib/systemd/network/
install -d ${D}/etc/systemd/network/
install -m 0644 ${WORKDIR}/rndis.network ${D}${prefix}/lib/systemd/network/
install -m 0644 ${WORKDIR}/wired.network ${D}/etc/systemd/network/
}
FILES_${PN} += " \
${nonarch_base_libdir}/systemd/network \
"
For testing and debugging purposes, one can use iproute2 utilities to manually set up a static IP and BusyBox's udhcpc utility to explicitly request an IP address using DHCP. Note that such a configuration will not be persistent across reboots.
Make sure to disable the networking management daemons to avoid interference:
systemctl stop connman.service
systemctl stop systemd-networkd.socket
systemctl stop systemd-networkd.service
Get a list of interfaces using ip link:
# ip link
1: lo: mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: eth0: mtu 1500 qdisc pfifo_fast state DOWN mode DEFAULT group default qlen 1000
link/ether 00:14:2d:59:ca:81 brd ff:ff:ff:ff:ff:ff
3: sit0@NONE: mtu 1480 qdisc noop state DOWN mode DEFAULT group default
link/sit 0.0.0.0 brd 0.0.0.0
4: can0: mtu 16 qdisc noop state DOWN mode DEFAULT group default qlen 10
link/can
Enable the network interface (e.g. eth0):
# ip link set $interface up
Assign a static IP address:
# ip addr add $IP_address/$subnet_mask broadcast $broadcast_address dev $interface
Then add your gateway IP address:
# ip route add default via $default_gateway
For example:
# ip link set eth0 up
$ ip addr add 192.168.10.2/24 broadcast 192.168.10.255 dev eth0
$ ip route add default via 192.168.10.1
To undo these configurations (e.g. before switching to a dynamic IP), first remove any assigned IP address:
# ip addr flush dev $interface
Then remove any assigned gateway:
# ip route flush dev $interface
And finally disable the interface:
# ip link set $interface down
The utility udhcpc can be used to get a one time lease for an IP address from the DHCP server serving the local network:
# udhcpc -i eth0
udhcpc (v1.23.2) started
...
Sending discover...
Sending select for 192.168.10.108...
Lease of 192.168.10.108 obtained, lease time 86400
...
The Apalis iMX6 modules provide a Gigabit Ethernet network connection using an internal Ethernet controller communicating via the RGMII interface.
[ 1.447124] fec 2188000.ethernet eth0: registered PHC device 0
[ 6.725715] fec 2188000.ethernet eth0: Freescale FEC PHY driver [Micrel KSZ9031 Gigabit PHY] (mii_bus:phy_addr=2188000.ethernet:07, irq=70)
[ 6.744126] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 300.899121] fec 2188000.ethernet eth0: Link is Up - 1Gbps/Full - flow control off
[ 300.907002] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
The IEEE 1588 main clock required to provide accurate timestamps is routed through an external SoC Pad GPIO16 (SPDIF1_IN). This is a hardware limitation, for details refer to the Apalis iMX6 Datasheet Chapter 5.3 Ethernet. The following device tree changes make sure the time stamping clock is properly routed to the IEEE 1588 block:
--- a/arch/arm/boot/dts/imx6qdl-apalis.dtsi
+++ b/arch/arm/boot/dts/imx6qdl-apalis.dtsi
@@ -688,6 +688,7 @@
MX6QDL_PAD_RGMII_RD2__RGMII_RD2 0x1b0b0
MX6QDL_PAD_RGMII_RD3__RGMII_RD3 0x1b0b0
MX6QDL_PAD_RGMII_RX_CTL__RGMII_RX_CTL 0x1b0b0
+ MX6QDL_PAD_GPIO_16__ENET_REF_CLK ((1<<30) | 0x1b0b0)
>;
};
pinctrl_enet_ctrl: enet_ctrlgrp {
@@ -871,7 +872,6 @@
spdif {
pinctrl_spdif: spdifgrp {
fsl,pins = <
- MX6QDL_PAD_GPIO_16__SPDIF_IN 0x1b0b0
MX6QDL_PAD_GPIO_17__SPDIF_OUT 0x1b0b0
>;
};
The Apalis iMX8 / iMX8X modules provide a Gigabit Ethernet network connection using an internal Ethernet controller communicating via RGMII interface.
root@apalis-imx8:~# dmesg | grep fec
[ 2.753007] fec 5b040000.ethernet: 5b040000.ethernet supply phy not found, using dummy regulator
[ 2.783775] libphy: fec_enet_mii_bus: probed
[ 2.791087] fec 5b040000.ethernet eth0: registered PHC device 0
[ 13.962747] fec 5b040000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx
Disconnect and reconnect the network cable:
[ 42.626377] fec 5b040000.ethernet eth0: Link is Down
[ 42.644164] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 44.674417] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 44.682705] fec 5b040000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx
The Apalis TK1 and T30 modules provide an Ethernet network connection through an Intel I210-AT Gigabit Ethernet Controller Chip
[ 5.541133] igb 0000:01:00.0: added PHC on eth0
[ 5.547812] igb 0000:01:00.0: Intel(R) Gigabit Ethernet Network Connection
[ 5.556847] igb 0000:01:00.0: eth0: (PCIe:2.5GT/s:Width x1)
[ 5.564655] igb 0000:01:00.0 eth0: MAC: 00:14:2d:2b:13:23
[ 5.572149] igb 0000:01:00.0: eth0: PBA No: Unknown
[ 5.579107] igb 0000:01:00.0: LRO is disabled
[ 5.585458] igb 0000:01:00.0: Using MSI-X interrupts. 1 rx queue(s), 1 tx queue(s)
[ 11.123624] IPv6: ADDRCONF(NETDEV_UP): rndis0: link is not ready
[ 13.817577] igb 0000:01:00.0 enp1s0: igb: enp1s0 NIC Link is Up 100 Mbps Full Duplex, Flow Control: RX/TX
[ 13.838881] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready
The Colibri T20/T30 modules provide an Ethernet network connection through an AX88772B chip connected to the second EHCI USB 2.0 controller.
[ 4.193401] usb 1-1: new high speed USB device using tegra-ehci and address 2
[ 4.443757] usb 1-1: New USB device found, idVendor=0b95, idProduct=772b
[ 4.450451] usb 1-1: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[ 4.457589] usb 1-1: Product: AX88772B
[ 4.461332] usb 1-1: Manufacturer: ASIX Elec. Corp.
[ 4.466219] usb 1-1: SerialNumber: 000001
...
[ 5.314474] ASIX USB Ethernet Adapter:v4.1.0 09:33:56 Jul 4 2011
[ 5.314479] <6> http://www.asix.com.tw
...
[ 5.337205] eth%d: status ep1in, 8 bytes period 11
[ 5.342226] eth0: register 'asix' at usb-tegra-ehci.1-1, ASIX AX88772B USB 2.0 Ethernet, 00:0e:c6:87:72:01
...
[ 5.674034] Netfilter messages via NETLINK v0.30.
[ 5.678773] nf_conntrack version 0.5.0 (2227 buckets, 8908 max)
[ 5.684918] ctnetlink v0.93: registering with nfnetlink.
[ 5.690323] xt_time: kernel timezone is -0000
[ 5.694915] ip_tables: (C) 2000-2006 Netfilter Core Team
[ 5.700294] arp_tables: (C) 2002 David S. Miller
[ 5.704950] TCP cubic registered
[ 5.708173] NET: Registered protocol family 17
[ 5.712623] NET: Registered protocol family 15
...
[ 9.958314] eth0: rxqlen 0 --> 5
[ 10.032823] eth0: ax88772b - Link status is: 0
The Colibri VF50/VF61 modules provide an Ethernet network connection using an internal Ethernet controller communicating via the RMII interface.
[ 20.417573] fec 400d1000.ethernet eth0: NXP/Freescale FEC PHY driver [Micrel KSZ8041] (mii_bus:phy_addr=400d1000.etherne:00, irq=-1)
[ 20.569406] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 160.418064] fec 400d1000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx
[ 160.426085] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Vybrid has two on-chip Ethernet controllers available via two independent RMII interfaces. One controller is connected to the on-module PHY which is used according to the standard Colibri Ethernet pin-out. To use the second on-chip Ethernet controller an external PHY (on the carrier board) needs to be connected to the RMII interface. Refer to the Colibri Vybrid datasheet.
Please use the link below to download the reference schematics and CAE data for RMII based Ethernet:
Note: The RMII pin-out of the second Ethernet controller is not compatible with other modules in the Colibri family.
To enable software support for the second Ethernet controller, a custom device tree is required. Our Kernel tree contains a custom device tree for the dual Ethernet Colibri Carrier Board:
Device trees for dual Ethernet Board
To make sure U-Boot selects the device tree automatically, you can make use of the fdt_board variable:
setenv fdt_board dual-eth
saveenv
For more details on how to customize the device tree refer to the article Device Tree Customization.
Note: The second instance of the internal Ethernet controller (FEC1) is the default/on-module Ethernet controller on Colibri VFxx while the first instance of the internal Ethernet controller (FEC0) is the additional Ethernet controller. However, currently, the native order of the FEC instances is used to assign the interface names (eth0 for FEC0, eth1 for FEC1). The device tree alias is used by U-Boot to assign the primary MAC address (ethaddr) to the device behind the alias ethernet0 and the secondary MAC address (eth1addr) to the device behind the alias ethernet1. However, the Linux kernel does not name interfaces according to those aliases. A method to work around this limitation is to use a udev rule file which renames the network interfaces (/etc/udev/rules.d/20-ethernet.rules):
/etc/udev/rules.d/20-ethernet.rules
# Ethernet network device rules file
#
# Do not assign ethX names, since they might be already in use.
# Use udevadm to get possible selectors.
# udevadm info -a -p /sys/class/net/eth0
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", KERNELS=="400d0000.ethernet", NAME="fec1"
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", KERNELS=="400d1000.ethernet", NAME="fec0"
The NXP Vybrid SoC provides a hardware Ethernet Switch controller which can control the two network interfaces (MACs). This allows Ethernet frames to pass through without any intervention by the CPU.
Our BSP V2.6 Beta 2 and later provide the driver as a configuration option CONFIG_FSL_L2_SWITCH. Apply the following device tree changes to enable the driver:
diff --git a/arch/arm/boot/dts/vf-colibri-dual-eth.dtsi b/arch/arm/boot/dts/vf-colibri-dual-eth.dtsi
index 8fed730..4cc2421 100644
--- a/arch/arm/boot/dts/vf-colibri-dual-eth.dtsi
+++ b/arch/arm/boot/dts/vf-colibri-dual-eth.dtsi
@@ -14,20 +14,15 @@
};
aliases {
- ethernet0 = &fec0;
- ethernet1 = &fec1;
+ ethernet0 = &esw;
};
};
-&fec0 {
- phy-mode = "rmii";
- pinctrl-names = "default";
- pinctrl-0 = <&pinctrl_fec0>;
- status = "okay";
-};
-
-&fec1 {
- status = "okay";
+&esw {
+ pinctrl-names = "default";
+ pinctrl-0 = <&pinctrl_fec0 &pinctrl_fec1>;
+ status = "okay";
};
The boot log should show the assignment of the two PHYs:
[ 0.856500] libphy: fsl l2 switch MII Bus: probed
[ 0.861269] fsl l2 switch MII Bus mdiobus(0) register ok.
[ 0.867796] eth-switch 400e8000.l2-switch eth0: eth0: Ethernet switch 00:14:2d:49:ca:81
Note: With the L2 switch driver the Ethernet physical layer transceiver (PHY's) is no longer connected to the Linux Ethernet port. This means that the Linux port will always appear to be UP, even when both Ethernet ports are disconnected. Earlier versions of the driver did connect only the PHY of port 0 to the Ethernet interface, which added a dependency between PHY0 and the Ethernet interface (see net: fsl_l2_switch: drop registering phy in switch mode).
The Colibri iMX6, iMX6ULL, iMX7, and iMX8X modules provide an Ethernet network connection through Microchip/Micrel KSZ8041NL Ethernet PHY
[ 0.669871] fec 30be0000.ethernet: Linked as a consumer to regulator.12
[ 0.681626] libphy: fec_enet_mii_bus: probed
[ 0.682447] fec 30be0000.ethernet eth0: registered PHC device 0
...
[ 5.140230] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 5.152342] Micrel KSZ8041 30be0000.ethernet-1:00: attached PHY driver [Micrel KSZ8041] (mii_bus:phy_addr=30be0000.ethernet-1:00, irq=POLL)
[ 5.152698] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
...
[ 7.273061] fec 30be0000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx
According to the Verdin Family Specification, one Gigabit Ethernet port is Always Compatible. It is a 10/100/1000 Mbit media-dependent interface, with the Ethernet PHY on the module.
Another gigabit media independent interface (RGMII) may be available on Reserved pins. To use it, you must place an Ethernet PHY on your carrier board. Our open-source Verdin Development Board can be used as a reference design.
Learn more about the pin-compatibility concepts of Always Compatible, Reserved and Module-specific in the Verdin Family Specification.
Since the i.MX 8M Mini SoC features only a single Ethernet MAC, only the Always Compatible Verdin Ethernet is available on a module. The standard SKU versions of the Verdin iMX8M Mini feature an on-module Microchip KSZ9131 Gigabit Ethernet PHY and therefore provide the media-dependent interface.
Modules with RGMII output are available as build-to-order assembly variants.
root@verdin-imx8mm:~# dmesg | grep fec
[ 3.621159] libphy: fec_enet_mii_bus: probed
[ 3.626634] fec 30be0000.ethernet eth0: registered PHC device 0
[ 11.267886] fec 30be0000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx
Disconnect and reconnect the network cable:
[ 220.094536] fec 30be0000.ethernet eth0: Link is Down
[ 220.113357] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 225.941052] fec 30be0000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx
[ 225.948853] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Early samples of the Verdin iMX8M Mini modules such as V1.0A and V1.0B are using the KSZ9031 PHY which works on BSP 3.0, however, adding the correct skew settings are recommended. Refer to Example of Device Tree customization to add support for KSZ9131 for correct settings.
The two Ethernet MAC that the i.MX 8M Plus SoC provides are slightly different from the i.MX 8M Mini. Both Ethernet controllers support Energy Efficient Ethernet (EEE), Ethernet AVB, and IEEE 1588. Only one Ethernet controller features additional Time-Sensitive Networking (TSN). This controller is used for the Always Compatible on-module Gigabit Ethernet PHY, the Microchip KSZ9131.
root@verdin-imx8mp:~# dmesg | grep imx-dwmac
[ 1.302554] imx-dwmac 30bf0000.ethernet: IRQ eth_lpi not found
[ 1.308468] imx-dwmac 30bf0000.ethernet: no reset control found
[ 3.284332] imx-dwmac 30bf0000.ethernet: IRQ eth_lpi not found
[ 3.290382] imx-dwmac 30bf0000.ethernet: no reset control found
[ 3.308229] imx-dwmac 30bf0000.ethernet: User ID: 0x10, Synopsys ID: 0x51
[ 3.315157] imx-dwmac 30bf0000.ethernet: DWMAC4/5
[ 3.320046] imx-dwmac 30bf0000.ethernet: DMA HW capability register supported
[ 3.327315] imx-dwmac 30bf0000.ethernet: RX Checksum Offload Engine supported
[ 3.334582] imx-dwmac 30bf0000.ethernet: TX Checksum insertion supported
[ 3.341410] imx-dwmac 30bf0000.ethernet: Wake-Up On Lan supported
[ 3.347650] imx-dwmac 30bf0000.ethernet: Enable RX Mitigation via HW Watchdog Timer
[ 3.355452] imx-dwmac 30bf0000.ethernet: Enabled Flow TC (entries=8)
[ 3.361930] imx-dwmac 30bf0000.ethernet: Enabling HW TC (entries=256, max_off=256)
[ 5.947505] imx-dwmac 30bf0000.ethernet eth0: PHY [stmmac-1:07] driver [Microchip KSZ9131 Gigabit PHY]
[ 5.986512] imx-dwmac 30bf0000.ethernet eth0: No Safety Features support found
[ 5.986532] imx-dwmac 30bf0000.ethernet eth0: IEEE 1588-2008 Advanced Timestamp supported
[ 5.990368] imx-dwmac 30bf0000.ethernet eth0: registered PTP clock
[ 6.040561] imx-dwmac 30bf0000.ethernet eth0: configuring for phy/rgmii-id link mode
Connect, disconnect and reconnect the network cable:
[ 2335.143057] imx-dwmac 30bf0000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx
[ 2335.151363] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 2348.452145] imx-dwmac 30bf0000.ethernet eth0: Link is Down
[ 2366.886638] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 2366.893330] imx-dwmac 30bf0000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx
The second Ethernet controller (without TSN) is available as an RGMII interface on Reserved pins of the module edge connector pins.
root@verdin-imx8mp:~# dmesg | grep fec
[ 1.224395] libphy: fec_enet_mii_bus: probed
[ 1.229590] fec 30be0000.ethernet eth1: registered PHC device 0
[ 5.428439] fec 30be0000.ethernet eth1: Link is Up - 100Mbps/Full - flow control rx/tx
Connect, disconnect and reconnect the network cable:
[ 2191.837002] IPv6: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready
[ 2191.843574] fec 30be0000.ethernet eth1: Link is Up - 1Gbps/Full - flow control rx/tx
[ 2198.461504] fec 30be0000.ethernet eth1: Link is Down
[ 2205.465840] IPv6: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready
[ 2205.469285] fec 30be0000.ethernet eth1: Link is Up - 1Gbps/Full - flow control rx/tx
The Gigabit Ethernet MAC in the i.MX8M Plus SoC integrates an accurate IEEE 1588 compliant timer for clock synchronization for distributed control nodes used in industrial automation applications. The interface features external IEEE 1588 synchronization pins on alternate functions. The Ethernet interface supports Audio Video Bridging (AVB) and Time-Sensitive Networking (TSN).
The Colibri SoMs provide a sustained data rate of around 95 Mbits/sec for a 100Base-T connection while the Apalis and Verdin SoMs provide a sustained data rate of around 700-900 Mbits/sec for a 1000Base-T connection. The CPU utilization usually stays well below 25 % on one single core.
Note that the i.MX6 bandwidth is limited to around 400 Mbits/sec, see NXP's Errata ERR004512.
Note that the CPU's DVFS in combination with TCP's flow control may negatively impact throughput.
Start server on Host:
[user@host ~]$ iperf3 -s
-----------------------------------------------------------
Server listening on 5201
-----------------------------------------------------------
Accepted connection from 192.168.10.215, port 52568
[ 5] local 192.168.10.1 port 5201 connected to 192.168.10.215 port 52570
[ ID] Interval Transfer Bitrate
[ 5] 0.00-1.00 sec 102 MBytes 852 Mbits/sec
[ 5] 1.00-2.00 sec 112 MBytes 939 Mbits/sec
[ 5] 2.00-3.00 sec 112 MBytes 941 Mbits/sec
[ 5] 3.00-4.00 sec 112 MBytes 940 Mbits/sec
[ 5] 4.00-5.00 sec 112 MBytes 941 Mbits/sec
[ 5] 5.00-6.00 sec 112 MBytes 941 Mbits/sec
[ 5] 6.00-7.00 sec 112 MBytes 940 Mbits/sec
[ 5] 7.00-8.00 sec 112 MBytes 941 Mbits/sec
[ 5] 8.00-9.00 sec 112 MBytes 940 Mbits/sec
[ 5] 9.00-10.00 sec 112 MBytes 942 Mbits/sec
[ 5] 10.00-10.04 sec 4.51 MBytes 935 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate
[ 5] 0.00-10.04 sec 1.09 GBytes 932 Mbits/sec receiver
-
Start client on SoM
root@verdin-imx8mm:~# iperf3 -c 192.168.10.1
Connecting to host 192.168.10.1, port 5201
[ 5] local 192.168.10.215 port 52570 connected to 192.168.10.1 port 5201
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 108 MBytes 905 Mbits/sec 0 508 KBytes
[ 5] 1.00-2.00 sec 113 MBytes 947 Mbits/sec 0 508 KBytes
[ 5] 2.00-3.00 sec 112 MBytes 937 Mbits/sec 0 508 KBytes
[ 5] 3.00-4.00 sec 113 MBytes 944 Mbits/sec 0 508 KBytes
[ 5] 4.00-5.00 sec 112 MBytes 939 Mbits/sec 0 508 KBytes
[ 5] 5.00-6.00 sec 112 MBytes 937 Mbits/sec 0 508 KBytes
[ 5] 6.00-7.00 sec 112 MBytes 939 Mbits/sec 0 508 KBytes
[ 5] 7.00-8.00 sec 113 MBytes 944 Mbits/sec 0 508 KBytes
[ 5] 8.00-9.00 sec 112 MBytes 938 Mbits/sec 0 508 KBytes
[ 5] 9.00-10.00 sec 112 MBytes 939 Mbits/sec 0 508 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-10.00 sec 1.09 GBytes 937 Mbits/sec 0 sender
[ 5] 0.00-10.04 sec 1.09 GBytes 932 Mbits/sec receiver
iperf Done.
CPU Utilisation during the test, start top over an SSH connection and observe during bandwidth test:
root@verdin-imx8mm:~# top
...
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
4341 root 20 0 6484 3640 3192 R 6.0 0.2 0:00.40 iperf3
Start server on SoM:
root@verdin-imx8mm:~# iperf3 -s
-----------------------------------------------------------
Server listening on 5201
-----------------------------------------------------------
Accepted connection from 192.168.10.1, port 36318
[ 5] local 192.168.10.215 port 5201 connected to 192.168.10.1 port 36320
[ ID] Interval Transfer Bitrate
[ 5] 0.00-1.00 sec 82.0 MBytes 688 Mbits/sec
[ 5] 1.00-2.00 sec 83.9 MBytes 704 Mbits/sec
[ 5] 2.00-3.00 sec 82.6 MBytes 693 Mbits/sec
[ 5] 3.00-4.00 sec 81.4 MBytes 683 Mbits/sec
[ 5] 4.00-5.00 sec 80.4 MBytes 674 Mbits/sec
[ 5] 5.00-6.00 sec 82.9 MBytes 696 Mbits/sec
[ 5] 6.00-7.00 sec 80.8 MBytes 678 Mbits/sec
[ 5] 7.00-8.00 sec 81.5 MBytes 684 Mbits/sec
[ 5] 8.00-9.00 sec 83.0 MBytes 696 Mbits/sec
[ 5] 9.00-10.00 sec 82.5 MBytes 692 Mbits/sec
[ 5] 10.00-10.00 sec 249 KBytes 536 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate
[ 5] 0.00-10.00 sec 821 MBytes 689 Mbits/sec receiver
Start client on host:
[user@host ~]$ iperf3 -c 192.168.10.215
Connecting to host 192.168.10.215, port 5201
[ 5] local 192.168.10.1 port 36320 connected to 192.168.10.215 port 5201
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 83.8 MBytes 703 Mbits/sec 0 379 KBytes
[ 5] 1.00-2.00 sec 84.0 MBytes 705 Mbits/sec 0 455 KBytes
[ 5] 2.00-3.00 sec 83.0 MBytes 696 Mbits/sec 0 455 KBytes
[ 5] 3.00-4.00 sec 81.5 MBytes 684 Mbits/sec 0 455 KBytes
[ 5] 4.00-5.00 sec 80.3 MBytes 674 Mbits/sec 0 455 KBytes
[ 5] 5.00-6.00 sec 82.5 MBytes 692 Mbits/sec 0 455 KBytes
[ 5] 6.00-7.00 sec 81.1 MBytes 680 Mbits/sec 0 455 KBytes
[ 5] 7.00-8.00 sec 81.8 MBytes 687 Mbits/sec 0 455 KBytes
[ 5] 8.00-9.00 sec 82.5 MBytes 692 Mbits/sec 0 455 KBytes
[ 5] 9.00-10.00 sec 83.0 MBytes 696 Mbits/sec 0 455 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-10.00 sec 824 MBytes 691 Mbits/sec 0 sender
[ 5] 0.00-10.00 sec 821 MBytes 689 Mbits/sec receiver
CPU Utilisation during the test, start top over an SSH connection and observe during bandwidth test:
root@verdin-imx8mm:~# top
...
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
4036 root 20 0 6484 3616 3160 S 10.3 0.2 0:00.45 iperf3
Start server on SoM
root@verdin-imx8mm:~# iperf3 -s
-----------------------------------------------------------
Server listening on 5201
-----------------------------------------------------------
Accepted connection from ::1, port 53888
[ 5] local ::1 port 5201 connected to ::1 port 53890
[ ID] Interval Transfer Bitrate
[ 5] 0.00-1.00 sec 790 MBytes 6.63 Gbits/sec
[ 5] 1.00-2.00 sec 816 MBytes 6.84 Gbits/sec
[ 5] 2.00-3.00 sec 815 MBytes 6.84 Gbits/sec
[ 5] 3.00-4.00 sec 818 MBytes 6.87 Gbits/sec
[ 5] 4.00-5.00 sec 819 MBytes 6.87 Gbits/sec
[ 5] 5.00-6.00 sec 825 MBytes 6.92 Gbits/sec
[ 5] 6.00-7.00 sec 822 MBytes 6.89 Gbits/sec
[ 5] 7.00-8.00 sec 822 MBytes 6.90 Gbits/sec
[ 5] 8.00-9.00 sec 819 MBytes 6.87 Gbits/sec
[ 5] 9.00-10.00 sec 815 MBytes 6.83 Gbits/sec
[ 5] 10.00-10.04 sec 31.4 MBytes 6.87 Gbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate
[ 5] 0.00-10.04 sec 8.00 GBytes 6.85 Gbits/sec receiver
Start client on SoM, e.g. over ssh:
root@verdin-imx8mm:~# iperf3 -c localhost
Connecting to host localhost, port 5201
[ 5] local ::1 port 53890 connected to ::1 port 5201
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 830 MBytes 6.96 Gbits/sec 0 2.06 MBytes
[ 5] 1.00-2.00 sec 816 MBytes 6.84 Gbits/sec 0 2.06 MBytes
[ 5] 2.00-3.00 sec 815 MBytes 6.84 Gbits/sec 0 2.06 MBytes
[ 5] 3.00-4.00 sec 819 MBytes 6.87 Gbits/sec 0 2.06 MBytes
[ 5] 4.00-5.00 sec 819 MBytes 6.87 Gbits/sec 0 2.06 MBytes
[ 5] 5.00-6.00 sec 825 MBytes 6.92 Gbits/sec 0 2.06 MBytes
[ 5] 6.00-7.00 sec 822 MBytes 6.90 Gbits/sec 0 2.06 MBytes
[ 5] 7.00-8.00 sec 821 MBytes 6.89 Gbits/sec 0 2.06 MBytes
[ 5] 8.00-9.00 sec 819 MBytes 6.87 Gbits/sec 0 2.06 MBytes
[ 5] 9.00-10.00 sec 815 MBytes 6.84 Gbits/sec 0 2.06 MBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-10.00 sec 8.01 GBytes 6.88 Gbits/sec 0 sender
[ 5] 0.00-10.04 sec 8.00 GBytes 6.85 Gbits/sec receiver
iperf Done.
