Vector Packet Processing

VPP (part of Fast Data - Input/Output project) is an extensible userspace framework providing networking functionality. It is built around the concept of packet processing graph (see What is VPP?).

Detailed instructions for simplified steps 1-3 below, can be found on VPP Quick Start Guide.

SPDK supports VPP version 19.01.1.

1. Building VPP (optional)

Please skip this step if using already built packages.

Clone and checkout VPP

git clone https://gerrit.fd.io/r/vpp && cd vpp
git checkout v19.01.1
git cherry-pick 97dcf5bd26ca6de580943f5d39681f0144782c3d
git cherry-pick f5dc9fbf814865b31b52b20f5bf959e9ff818b25

NOTE: Cherry-picks are required for better integration with SPDK. They are already merged to VPP 19.04.

NOTE: We have noticed that VPP tries to close connections to the non existing, already closed applications, after timeout. It causes intermittent VPP application segfaults when few instances of VPP clients connects and disconnects several times. The following workaround for this issue helps to create more stable environment for VPP v19.01.1. This issue should be solved in the next release of VPP.

git apply test/common/config/patch/vpp/workaround-dont-notify-transport-closing.patch

Install VPP build dependencies

make install-dep

Build and create .rpm packages

make pkg-rpm

Alternatively, build and create .deb packages

make bootstrap && make pkg-deb

Packages can be found in vpp/build-root/ directory.

For more in depth instructions please see Building section in VPP documentation

2. Installing VPP

Packages can be installed from a distribution repository or built in previous step. Minimal set of packages consists of vpp, vpp-lib and vpp-devel.

Note: Please remove or modify /etc/sysctl.d/80-vpp.conf file with appropriate values dependent on number of hugepages that will be used on system.

3. Running VPP

VPP takes over any network interfaces that were bound to userspace driver, for details please see DPDK guide on Binding and Unbinding Network Ports to/from the Kernel Modules.

VPP is installed as service and disabled by default. To start VPP with default config:

sudo systemctl start vpp

Alternatively, use vpp binary directly

sudo vpp unix {cli-listen /run/vpp/cli.sock} session { evt_qs_memfd_seg } socksvr { socket-name /run/vpp-api.sock }

4. Configure VPP

VPP can be configured using a VPP startup file and the vppctl command; By default, the VPP startup file is /etc/vpp/startup.conf, however, you can pass any file with the -c vpp command argument.

Startup configuration

Some key values from iSCSI point of view includes:

CPU section (cpu):

  • main-core <lcore> – logical CPU core used for main thread.
  • corelist-workers <lcore list> – logical CPU cores where worker threads are running.

DPDK section (dpdk):

  • num-rx-queues <num> – number of receive queues.
  • num-tx-queues <num> – number of transmit queues.
  • dev <PCI address> – whitelisted device.
  • num-mbufs – numbers of allocated buffers. For the most of our scenarios this parameter requires to be increased over default value.

Session section (session):

  • evt_qs_memfd_seg – uses a memfd segment for event queues. This is required for SPDK.

Socket server session (socksvr):

  • socket-name <path> – configure API socket filename (curently SPDK uses default path /run/vpp-api.sock).

Plugins section (plugins):

  • plugin <plugin name> { [enable|disable] } – enable or disable VPP plugin.

Example:

unix {
nodaemon
cli-listen /run/vpp/cli.sock
}
cpu {
main-core 1
}
dpdk {
num-mbufs 128000
}
session {
evt_qs_memfd_seg
}
socksvr {
socket-name /run/vpp-api.sock
}
plugins {
plugin default { disable }
plugin dpdk_plugin.so { enable }
}

vppctl command tool

The vppctl command tool allows users to control VPP at runtime via a command prompt

sudo vppctl

Or, by sending single command directly. For example to display interfaces within VPP:

sudo vppctl show interface

Useful commands:

  • show interface – show interfaces settings, state and some basic statistics.
  • show interface address – show interfaces state and assigned addresses.
  • set interface ip address <VPP interface> <Address> – set interfaces IP address.
  • set interface state <VPP interface> [up|down] – bring interface up or down.
  • show errors – show error counts.

Example: Configure two interfaces to be available via VPP

We want to configure two DPDK ports with PCI addresses 0000:09:00.1 and 0000:0b:00.1 to be used as portals 10.0.0.1/24 and 10.10.0.1/24.

In the VPP startup file (e.g. /etc/vpp/startup.conf), whitelist the interfaces by specifying PCI addresses in section dpdk:

dev 0000:09:00.1
dev 0000:0b:00.1

Bind PCI NICs to UIO driver (igb_uio or uio_pci_generic).

Restart vpp and use vppctl tool to verify interfaces.

$ vppctl show interface
Name Idx State MTU (L3/IP4/IP6/MPLS) Counter Count
FortyGigabitEthernet9/0/1 1 down 9000/0/0/0
FortyGigabitEthernetb/0/1 2 down 9000/0/0/0

Set appropriate addresses and bring interfaces up:

$ vppctl set interface ip address FortyGigabitEthernet9/0/1 10.0.0.1/24
$ vppctl set interface state FortyGigabitEthernet9/0/1 up
$ vppctl set interface ip address FortyGigabitEthernetb/0/1 10.10.0.1/24
$ vppctl set interface state FortyGigabitEthernetb/0/1 up

Verify configuration:

$ vppctl show interface address
FortyGigabitEthernet9/0/1 (up):
L3 10.0.0.1/24
FortyGigabitEthernetb/0/1 (up):
L3 10.10.0.1/24

Now, both interfaces are ready to use. To verify conectivity you can ping 10.0.0.1 and 10.10.0.1 addresses from another machine.

Example: Tap interfaces on single host

For functional test purposes a virtual tap interface can be created, so no additional network hardware is required. This will allow network communication between SPDK iSCSI target using VPP end of tap and kernel iSCSI initiator using the kernel part of tap. A single host is used in this scenario.

Create tap interface via VPP

vppctl tap connect tap0
vppctl set interface state tapcli-0 up
vppctl set interface ip address tapcli-0 10.0.0.1/24
vppctl show int addr

Assign address on kernel interface

sudo ip addr add 10.0.0.2/24 dev tap0
sudo ip link set tap0 up

To verify connectivity

ping 10.0.0.1

5. Building SPDK with VPP

Support for VPP can be built into SPDK by using configuration option.

configure --with-vpp

Alternatively, directory with built libraries can be pointed at and will be used for compilation instead of installed packages.

configure --with-vpp=/path/to/vpp/repo/build-root/install-vpp-native/vpp

6. Running SPDK with VPP

VPP application has to be started before SPDK application, in order to enable usage of network interfaces. For example, if you use SPDK iSCSI target or NVMe-oF target, after the initialization finishes, interfaces configured within VPP will be available to be configured as portal addresses.

Moreover, you do not need to specifiy which TCP sock implementation (e.g., posix, VPP) to be used through configuration file or RPC call. Since SPDK program automatically determines the protocol according to the configured portal addresses info. For example, you can specify a Listen address in NVMe-oF subsystem configuration such as "Listen TCP 10.0.0.1:4420". SPDK programs automatically uses different implemenation to listen this provided portal info via posix or vpp implemenation(if compiled in SPDK program), and only one implementation can successfully listen on the provided portal.