1.config eth0:
tc qdisc replace dev eth0 parent root handle 100 taprio \
num_tc 3 \
map 0 0 1 2 0 0 0 0 0 0 0 0 0 0 0 0 \
queues 1@0 1@1 1@2 \
base-time 0000 \
sched-entry S 4 400000 \
sched-entry S 2 300000 \
sched-entry S 1 300000 \
flags 2
[ 1793.502127] am65-cpsw-nuss 46000000.ethernet eth0: now is 1642559724801231806
[ 1793.502130] am65-cpsw-nuss 46000000.ethernet eth0: *start is 1642559724802000000
hw start_time = 1642559724802000000 ns
2.iperf priority 0 traffic
eth0 , 1Gbps, iperf -b 10M/50M
3.statistics
By synchronizing the time with the peer AM64XX/TDA4 device, and capture rx hardware timestamp on the peer.
It is expected that in each ms period, the traffic will fall within the 700us~1ms window.
But, the rx hardware timestamp randomly falls between 0~400us or 400~700us in each ms period.
Compared with CLOCK_REALTIME/CLOCK_TAI, all within the expected 700us~1ms window in each ms period.
