On Tue, 23 Jun 2020 23:39:31 +0200 Lorenzo Bianconi <lorenzo@xxxxxxxxxx> wrote: > Introduce XDP_REDIRECT support for eBPF programs attached to cpumap > entries. > This patch has been tested on Marvell ESPRESSObin using a modified > version of xdp_redirect_cpu sample in order to attach a XDP program > to CPUMAP entries to perform a redirect on the mvneta interface. > In particular the following scenario has been tested: > > rq (cpu0) --> mvneta - XDP_REDIRECT (cpu0) --> CPUMAP - XDP_REDIRECT (cpu1) --> mvneta > > $./xdp_redirect_cpu -p xdp_cpu_map0 -d eth0 -c 1 -e xdp_redirect \ > -f xdp_redirect_kern.o -m tx_port -r eth0 > > tx: 285.2 Kpps rx: 285.2 Kpps > > Attacching a simple XDP program on eth0 to perform XDP_TX gives ^^^^^^^^^^ Spelling/typo. > comparable results: > > tx: 288.4 Kpps rx: 288.4 Kpps > > Co-developed-by: Jesper Dangaard Brouer <brouer@xxxxxxxxxx> > Signed-off-by: Jesper Dangaard Brouer <brouer@xxxxxxxxxx> > Signed-off-by: Lorenzo Bianconi <lorenzo@xxxxxxxxxx> Acked-by: Jesper Dangaard Brouer <brouer@xxxxxxxxxx> > --- > include/net/xdp.h | 1 + > include/trace/events/xdp.h | 6 ++++-- > kernel/bpf/cpumap.c | 17 +++++++++++++++-- > 3 files changed, 20 insertions(+), 4 deletions(-) [...] > @@ -276,7 +286,10 @@ static int cpu_map_bpf_prog_run_xdp(struct bpf_cpu_map_entry *rcpu, > } > } > > - rcu_read_unlock(); > + if (stats->redirect) > + xdp_do_flush_map(); > + > + rcu_read_unlock_bh(); /* resched point, may call do_softirq() */ I've tested (on x86) that this extra resched point does not cause sched issues. -- Best regards, Jesper Dangaard Brouer MSc.CS, Principal Kernel Engineer at Red Hat LinkedIn: http://www.linkedin.com/in/brouer
