HOWTO for the linux packet generator¶
Enable CONFIG_NET_PKTGEN to compile and build pktgen either in-kernel or as a module. A module is preferred; modprobe pktgen if needed. Once running, pktgen creates a thread for each CPU with affinity to that CPU. Monitoring and controlling is done via /proc. It is easiest to select a suitable sample script and configure that.
On a dual CPU:
ps aux | grep pkt root 129 0.3 0.0 0 0 ? SW 2003 523:20 [kpktgend_0] root 130 0.3 0.0 0 0 ? SW 2003 509:50 [kpktgend_1]
For monitoring and control pktgen creates:
/proc/net/pktgen/pgctrl /proc/net/pktgen/kpktgend_X /proc/net/pktgen/ethX
Tuning NIC for max performance¶
The default NIC settings are (likely) not tuned for pktgen's artificial overload type of benchmarking, as this could hurt the normal use-case.
Specifically increasing the TX ring buffer in the NIC:
# ethtool -G ethX tx 1024
A larger TX ring can improve pktgen's performance, while it can hurt in the general case, 1) because the TX ring buffer might get larger than the CPU's L1/L2 cache, 2) because it allows more queueing in the NIC HW layer (which is bad for bufferbloat).
One should hesitate to conclude that packets/descriptors in the HW TX ring cause delay. Drivers usually delay cleaning up the ring-buffers for various performance reasons, and packets stalling the TX ring might just be waiting for cleanup.
This cleanup issue is specifically the case for the driver ixgbe (Intel 82599 chip). This driver (ixgbe) combines TX+RX ring cleanups, and the cleanup interval is affected by the ethtool --coalesce setting of parameter "rx-usecs".
For ixgbe use e.g. "30" resulting in approx 33K interrupts/sec (1/30*10^6):
# ethtool -C ethX rx-usecs 30
Pktgen creates a thread for each CPU with affinity to that CPU. Which is controlled through procfile /proc/net/pktgen/kpktgend_X.
Running: Stopped: eth4@0 Result: OK: add_device=eth4@0
Most important are the devices assigned to the thread.
The two basic thread commands are:
add_device DEVICE@NAME -- adds a single device
rem_device_all -- remove all associated devices
When adding a device to a thread, a corresponding procfile is created which is used for configuring this device. Thus, device names need to be unique.
To support adding the same device to multiple threads, which is useful with multi queue NICs, the device naming scheme is extended with "@": device@something
The part after "@" can be anything, but it is custom to use the thread number.
The Params section holds configured information. The Current section holds running statistics. The Result is printed after a run or after interruption. Example:
/proc/net/pktgen/eth4@0 Params: count 100000 min_pkt_size: 60 max_pkt_size: 60 frags: 0 delay: 0 clone_skb: 64 ifname: eth4@0 flows: 0 flowlen: 0 queue_map_min: 0 queue_map_max: 0 dst_min: 192.168.81.2 dst_max: src_min: src_max: src_mac: 90:e2:ba:0a:56:b4 dst_mac: 00:1b:21:3c:9d:f8 udp_src_min: 9 udp_src_max: 109 udp_dst_min: 9 udp_dst_max: 9 src_mac_count: 0 dst_mac_count: 0 Flags: UDPSRC_RND NO_TIMESTAMP QUEUE_MAP_CPU Current: pkts-sofar: 100000 errors: 0 started: 623913381008us stopped: 623913396439us idle: 25us seq_num: 100001 cur_dst_mac_offset: 0 cur_src_mac_offset: 0 cur_saddr: 192.168.8.3 cur_daddr: 192.168.81.2 cur_udp_dst: 9 cur_udp_src: 42 cur_queue_map: 0 flows: 0 Result: OK: 15430(c15405+d25) usec, 100000 (60byte,0frags) 6480562pps 3110Mb/sec (3110669760bps) errors: 0
This is done via the /proc interface, and most easily done via pgset as defined in the sample scripts. You need to specify PGDEV environment variable to use functions from sample scripts, i.e.:
export PGDEV=/proc/net/pktgen/eth4@0 source samples/pktgen/functions.sh
pg_ctrl start starts injection. pg_ctrl stop aborts injection. Also, ^C aborts generator. pgset "clone_skb 1" sets the number of copies of the same packet pgset "clone_skb 0" use single SKB for all transmits pgset "burst 8" uses xmit_more API to queue 8 copies of the same packet and update HW tx queue tail pointer once. "burst 1" is the default pgset "pkt_size 9014" sets packet size to 9014 pgset "frags 5" packet will consist of 5 fragments pgset "count 200000" sets number of packets to send, set to zero for continuous sends until explicitly stopped. pgset "delay 5000" adds delay to hard_start_xmit(). nanoseconds pgset "dst 10.0.0.1" sets IP destination address (BEWARE! This generator is very aggressive!) pgset "dst_min 10.0.0.1" Same as dst pgset "dst_max 10.0.0.254" Set the maximum destination IP. pgset "src_min 10.0.0.1" Set the minimum (or only) source IP. pgset "src_max 10.0.0.254" Set the maximum source IP. pgset "dst6 fec0::1" IPV6 destination address pgset "src6 fec0::2" IPV6 source address pgset "dstmac 00:00:00:00:00:00" sets MAC destination address pgset "srcmac 00:00:00:00:00:00" sets MAC source address pgset "queue_map_min 0" Sets the min value of tx queue interval pgset "queue_map_max 7" Sets the max value of tx queue interval, for multiqueue devices To select queue 1 of a given device, use queue_map_min=1 and queue_map_max=1 pgset "src_mac_count 1" Sets the number of MACs we'll range through. The 'minimum' MAC is what you set with srcmac. pgset "dst_mac_count 1" Sets the number of MACs we'll range through. The 'minimum' MAC is what you set with dstmac. pgset "flag [name]" Set a flag to determine behaviour. Current flags are: IPSRC_RND # IP source is random (between min/max) IPDST_RND # IP destination is random UDPSRC_RND, UDPDST_RND, MACSRC_RND, MACDST_RND TXSIZE_RND, IPV6, MPLS_RND, VID_RND, SVID_RND FLOW_SEQ, QUEUE_MAP_RND # queue map random QUEUE_MAP_CPU # queue map mirrors smp_processor_id() UDPCSUM, IPSEC # IPsec encapsulation (needs CONFIG_XFRM) NODE_ALLOC # node specific memory allocation NO_TIMESTAMP # disable timestamping pgset 'flag ![name]' Clear a flag to determine behaviour. Note that you might need to use single quote in interactive mode, so that your shell wouldn't expand the specified flag as a history command. pgset "spi [SPI_VALUE]" Set specific SA used to transform packet. pgset "udp_src_min 9" set UDP source port min, If < udp_src_max, then cycle through the port range. pgset "udp_src_max 9" set UDP source port max. pgset "udp_dst_min 9" set UDP destination port min, If < udp_dst_max, then cycle through the port range. pgset "udp_dst_max 9" set UDP destination port max. pgset "mpls 0001000a,0002000a,0000000a" set MPLS labels (in this example outer label=16,middle label=32, inner label=0 (IPv4 NULL)) Note that there must be no spaces between the arguments. Leading zeros are required. Do not set the bottom of stack bit, that's done automatically. If you do set the bottom of stack bit, that indicates that you want to randomly generate that address and the flag MPLS_RND will be turned on. You can have any mix of random and fixed labels in the label stack. pgset "mpls 0" turn off mpls (or any invalid argument works too!) pgset "vlan_id 77" set VLAN ID 0-4095 pgset "vlan_p 3" set priority bit 0-7 (default 0) pgset "vlan_cfi 0" set canonical format identifier 0-1 (default 0) pgset "svlan_id 22" set SVLAN ID 0-4095 pgset "svlan_p 3" set priority bit 0-7 (default 0) pgset "svlan_cfi 0" set canonical format identifier 0-1 (default 0) pgset "vlan_id 9999" > 4095 remove vlan and svlan tags pgset "svlan 9999" > 4095 remove svlan tag pgset "tos XX" set former IPv4 TOS field (e.g. "tos 28" for AF11 no ECN, default 00) pgset "traffic_class XX" set former IPv6 TRAFFIC CLASS (e.g. "traffic_class B8" for EF no ECN, default 00) pgset "rate 300M" set rate to 300 Mb/s pgset "ratep 1000000" set rate to 1Mpps pgset "xmit_mode netif_receive" RX inject into stack netif_receive_skb() Works with "burst" but not with "clone_skb". Default xmit_mode is "start_xmit".
A collection of tutorial scripts and helpers for pktgen is in the samples/pktgen directory. The helper parameters.sh file support easy and consistent parameter parsing across the sample scripts.
Usage example and help:
./pktgen_sample01_simple.sh -i eth4 -m 00:1B:21:3C:9D:F8 -d 192.168.8.2
./pktgen_sample01_simple.sh [-vx] -i ethX -i : ($DEV) output interface/device (required) -s : ($PKT_SIZE) packet size -d : ($DEST_IP) destination IP. CIDR (e.g. 198.18.0.0/15) is also allowed -m : ($DST_MAC) destination MAC-addr -p : ($DST_PORT) destination PORT range (e.g. 433-444) is also allowed -t : ($THREADS) threads to start -f : ($F_THREAD) index of first thread (zero indexed CPU number) -c : ($SKB_CLONE) SKB clones send before alloc new SKB -n : ($COUNT) num messages to send per thread, 0 means indefinitely -b : ($BURST) HW level bursting of SKBs -v : ($VERBOSE) verbose -x : ($DEBUG) debug -6 : ($IP6) IPv6 -w : ($DELAY) Tx Delay value (ns) -a : ($APPEND) Script will not reset generator's state, but will append its config
The global variables being set are also listed. E.g. the required interface/device parameter "-i" sets variable $DEV. Copy the pktgen_sampleXX scripts and modify them to fit your own needs.
Note that when adding devices to a specific CPU it is a good idea to also assign /proc/irq/XX/smp_affinity so that the TX interrupts are bound to the same CPU. This reduces cache bouncing when freeing skbs.
Plus using the device flag QUEUE_MAP_CPU, which maps the SKBs TX queue to the running threads CPU (directly from smp_processor_id()).
Default IPsec transformation with ESP encapsulation plus transport mode can be enabled by simply setting:
pgset "flag IPSEC" pgset "flows 1"
To avoid breaking existing testbed scripts for using AH type and tunnel mode, you can use "pgset spi SPI_VALUE" to specify which transformation mode to employ.
Current commands and configuration options¶
start stop reset
count clone_skb burst debug frags delay src_mac_count dst_mac_count pkt_size min_pkt_size max_pkt_size queue_map_min queue_map_max skb_priority tos (ipv4) traffic_class (ipv6) mpls udp_src_min udp_src_max udp_dst_min udp_dst_max node flag IPSRC_RND IPDST_RND UDPSRC_RND UDPDST_RND MACSRC_RND MACDST_RND TXSIZE_RND IPV6 MPLS_RND VID_RND SVID_RND FLOW_SEQ QUEUE_MAP_RND QUEUE_MAP_CPU UDPCSUM IPSEC NODE_ALLOC NO_TIMESTAMP spi (ipsec) dst_min dst_max src_min src_max dst_mac src_mac clear_counters src6 dst6 dst6_max dst6_min flows flowlen rate ratep xmit_mode <start_xmit|netif_receive> vlan_cfi vlan_id vlan_p svlan_cfi svlan_id svlan_p
Paper from Linux-Kongress in Erlangen 2004. - ftp://robur.slu.se/pub/Linux/net-development/pktgen-testing/pktgen_paper.pdf
Grant Grundler for testing on IA-64 and parisc, Harald Welte, Lennert Buytenhek Stephen Hemminger, Andi Kleen, Dave Miller and many others.
Good luck with the linux net-development.