Add `xsk_rr` an AF_XDP benchmark to measure latency
Subsystems (6)
RFC PATCH v2 1/1 Add `xsk_rr` an AF_XDP benchmark to measure latency
Note: This is a benchmarking tool that is used for experiments in the upcoming Napi threaded busypoll series. Not intended to be merged.
xsk_rr is a benchmarking tool to measure latency using AF_XDP between two nodes. The benchmark can be run with different arguments to simulate traffic:
- Payload Size
- Packet frequency (1/period)
- Application Processing (delay)
- Busy poll in separate core
- Payload verification
- Open Loop sampling
Server:
chrt -f 50 taskset -c 3-5 tools/testing/selftests/bpf/xsk_rr -o 0 \ -B <bytes> -i eth0 -4 -D <IP-dest> -S <IP-src> -M <MAC-dest> \ -m <MAC-src> -p <PORT> -h -v -t
Client:
chrt -f 50 taskset -c 3-5 tools/testing/selftests/bpf/xsk_rr -o 0 \ -B <bytes> -i eth0 -4 -S <IP-src> -D <IP-dest> -m <MAC-src> \ -M <MAC-dst> -p <PORT> -P <send-period-usecs> -d <recv-delay-usecs> \ -T -l <sample capture length in seconds> -v -t
Sample Output:
min: 13069 max: 881340 avg: 13322 empty: 98
count: 6249 p5: 13500
count: 61508 p50: 14000
count: 114428 p95: 14700
count: 119153 p99: 14800
rate: 12499
outstanding packets: 3
---
v2:
- Rebased
- Using needs wakeup
- Added PREFER_BUSY_POLL
--- tools/testing/selftests/bpf/Makefile | 9 +- .../selftests/bpf/progs/xsk_rr_progs.c | 60 ++ tools/testing/selftests/bpf/xsk_rr.c | 901 ++++++++++++++++++ 3 files changed, 969 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/bpf/progs/xsk_rr_progs.c create mode 100644 tools/testing/selftests/bpf/xsk_rr.c
tools/testing/selftests/bpf/Makefile +8 -1
@@ -135,7 +135,8 @@ TEST_GEN_PROGS_EXTENDED = \ xdp_hw_metadata \ xdp_synproxy \ xdping \ - xskxceiver + xskxceiver \ + xsk_rr TEST_GEN_FILES += liburandom_read.so urandom_read sign-file uprobe_multi
@@ -516,6 +517,7 @@ test_subskeleton.skel.h-deps := test_subskeleton_lib2.bpf.o test_subskeleton_lib test_subskeleton_lib.skel.h-deps := test_subskeleton_lib2.bpf.o test_subskeleton_lib.bpf.o test_usdt.skel.h-deps := test_usdt.bpf.o test_usdt_multispec.bpf.o xsk_xdp_progs.skel.h-deps := xsk_xdp_progs.bpf.o +xsk_rr_progs.skel.h-deps := xsk_rr_progs.bpf.o xdp_hw_metadata.skel.h-deps := xdp_hw_metadata.bpf.o xdp_features.skel.h-deps := xdp_features.bpf.o
@@ -781,6 +783,11 @@ $(OUTPUT)/xskxceiver: $(EXTRA_SRC) xskxceiver.c xskxceiver.h $(OUTPUT)/network_h $(call msg,BINARY,,$@) $(Q)$(CC) $(CFLAGS) $(filter %.a %.o %.c,$^) $(LDLIBS) -o $@ +$(OUTPUT)/xsk_rr: $(EXTRA_SRC) xsk_rr.c $(OUTPUT)/xsk.o $(OUTPUT)/xsk_rr_progs.skel.h $(BPFOBJ) | $(OUTPUT) + $(call msg,BINARY,,$@) + $(Q)$(CC) $(CFLAGS) $(filter %.a %.o %.c,$^) $(LDLIBS) -o $@ + + $(OUTPUT)/xdp_hw_metadata: xdp_hw_metadata.c $(OUTPUT)/network_helpers.o $(OUTPUT)/xsk.o $(OUTPUT)/xdp_hw_metadata.skel.h | $(OUTPUT) $(call msg,BINARY,,$@) $(Q)$(CC) $(CFLAGS) $(filter %.a %.o %.c,$^) $(LDLIBS) -o $@
tools/testing/selftests/bpf/progs/xsk_rr_progs.c new file +60 -0
60 lines not on this page · raw message
tools/testing/selftests/bpf/xsk_rr.c new file +901 -0
901 lines not on this page · raw message
base-commit: c3199adbe4ffffc7b6536715e0290d1919a45cd9
Signature · 1 line
2.51.0.338.gd7d06c2dae-goog
Quoted 35 lines · Samiullah Khawaja
Note: This is a benchmarking tool that is used for experiments in the upcoming Napi threaded busypoll series. Not intended to be merged.
xsk_rris a benchmarking tool to measure latency usingAF_XDPbetween two nodes. The benchmark can be run with different arguments to simulate traffic:- Payload Size
- Packet frequency (1/period)
- Application Processing (delay)
- Busy poll in separate core
- Payload verification
- Open Loop samplingServer:
chrt -f 50 taskset -c 3-5 tools/testing/selftests/bpf/xsk_rr -o 0 \ -B <bytes> -i eth0 -4 -D <IP-dest> -S <IP-src> -M <MAC-dest> \ -m <MAC-src> -p <PORT> -h -v -tClient:
chrt -f 50 taskset -c 3-5 tools/testing/selftests/bpf/xsk_rr -o 0 \ -B <bytes> -i eth0 -4 -S <IP-src> -D <IP-dest> -m <MAC-src> \ -M <MAC-dst> -p <PORT> -P <send-period-usecs> -d <recv-delay-usecs> \ -T -l <sample capture length in seconds> -v -tSample Output:
min: 13069 max: 881340 avg: 13322 empty: 98
count: 6249 p5: 13500
count: 61508 p50: 14000
count: 114428 p95: 14700
count: 119153 p99: 14800
rate: 12499
outstanding packets: 3
Quoted 8 lines
Signed-off-by Samiullah Khawaja <[email protected]>---
v2:
- Rebased
- Using needs wakeup
- AddedPREFER_BUSY_POLL
I cannot find SO_PREFER_BUSY_POLL anywhere in the code?
Thanks,
Martin