chip:1043. LSDK:2108. kernel:5.10
my board use RGMIIx2 .QSGMII x1.
dts:
Using iperf3 to test maximum rate----
board: iperf3 -s -p xxx
PC: iperf3 -c xx -b 1000M -p xxx
use fmc tool: fmc -c /etc/fmc/config/private/ls1043ardb/RR_FQPP_1455/config.xml -p /etc/fmc/config/private/ls1043ardb/RR_FQPP_1455/policy_ipv4.xml -a
serdes config:0x4558
the test result is
RGMII + QSGMII RX speed = 3Gpbs; TX speed = 1Gbps.
I'm not sure why the speed is evenly divided(3Gbps) across multiple network ports.
Thank you for your reply.
I have tested the P parameter before.It's also problematic.
Do I need to make any special modifications to the device tree I have only modified the MDIO section now.
Hi yiping
I found that it is not a rate bottleneck, but rather a block in establishing data frames for iperf3:
When testing two network ports at 2Gbps, 3th port iperf3 just stop here
I will try the following tests: ethtool -K xxx gro off.There's no problem with retesting.
The problem occurred in kernel version:5.10. and kernel v4.12 looks good.
May I ask if this is a GRO bug?
