Linux Mainline

[eloranta]

OK, got it to work. Compiled directly from ayufan's mainline repo. Copied the vmlinux to Image (in boot), used dracut to make the initrd image (in /boot) and copied the new kernel rock64 dtb to /boot. It really helped to have the serial console when getting this going.

[Osiander]

Hello together,

what are the correct exports for cross-compiling RK3328 mainline kernel?

Are this correct:

Code:

export ARCH=arm64
export CROSS_COMPILE=aarch64-linux-gnu-

Thank you very much

[maal]

1 GbE problem has been acknowledged by mainline kernel maintainers (no solution for now unfortunately):

https://git.kernel.org/pub/scm/linux/ker...3a367c83b3

arm64: dts: rockchip: limit rk3328-rock64 gmac speed to 100MBit for now

[thewonderer]

1 GbE problem has been acknowledged by mainline kernel maintainers (no solution for now unfortunately):

https://git.kernel.org/pub/scm/linux/ker...3a367c83b3

arm64: dts: rockchip: limit rk3328-rock64 gmac speed to 100MBit for now

[maal]

What problems are occuring with the GbE?  Drop outs?  Does the rock64 recover automatically or is it just a slow down in performance?

https://forum.pine64.org/showthread.php?...1#pid32501

It is not just a slow down, Ethernet is unusable at 1 Gbit/s speed in the mainline kernel.
I did not try the newest kernels (4.15-rc).

[ClaudiaF]

What problems are occuring with the GbE?  Drop outs?  Does the rock64 recover automatically or is it just a slow down in performance?

https://forum.pine64.org/showthread.php?...1#pid32501

It is not just a slow down, Ethernet is unusable at 1 Gbit/s speed in the mainline kernel.
I did not try the newest kernels (4.15-rc).

Do you know that you can work around it by using ethtool?
ethtool -K eth0 rx off tx off
The rx is not really needed but can't hurt anyway.

another trick might be to set the mtu size to something smaller than 1500 like 1492.
Both worked for me.

Cheers
Claudia

[maal]

Unfortunately this does not work for me.

Code:

[root@cep03 ~]#  uname -a
Linux cep03 4.15.0-rc6 #2 SMP PREEMPT Sat Jan 6 18:45:37 UTC 2018 aarch64 aarch64 aarch64 GNU/Linux
[root@cep03 ~]# ip link show dev eth0
3: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DEFAULT group default qlen 1000
   link/ether 66:ad:ab:e3:95:a8 brd ff:ff:ff:ff:ff:ff
[root@cep03 ~]# ethtool -S eth0 | grep err
    mmc_tx_underflow_error: 0
    mmc_tx_deferred: 0
    mmc_tx_carrier_error: 0
    mmc_rx_crc_error: 4
    mmc_rx_align_error: 0
    mmc_rx_run_error: 0
    mmc_rx_jabber_error: 0
    mmc_rx_length_error: 0
    mmc_rx_watchdog_error: 0
    mmc_rx_ipv4_hderr: 0
    mmc_rx_ipv4_hderr_octets: 0
    mmc_rx_ipv6_hderr_octets: 0
    mmc_rx_ipv6_hderr: 0
    mmc_rx_udp_err: 4
    mmc_rx_tcp_err: 0
    mmc_rx_icmp_err: 0
    mmc_rx_udp_err_octets: 224
    mmc_rx_tcp_err_octets: 0
    mmc_rx_icmp_err_octets: 0
    tx_deferred: 0
    tx_payload_error: 0
    tx_ip_header_error: 0
    overflow_error: 0
    ipc_csum_error: 0
    rx_crc_errors: 0
    fatal_bus_error_irq: 0
    phy_eee_wakeup_error_n: 0
    ip_hdr_err: 0
    ip_payload_err: 0
[root@cep03 ~]# ethtool -k eth0 | grep checksumming
Cannot get device udp-fragmentation-offload settings: Operation not supported
rx-checksumming: off
tx-checksumming: off
[root@cep03 ~]# ethtool -K eth0 rx off tx off                                                                                                                                      
Cannot get device udp-fragmentation-offload settings: Operation not supported
Cannot get device udp-fragmentation-offload settings: Operation not supported
[root@cep03 ~]# ethtool -k eth0 | grep checksumming
Cannot get device udp-fragmentation-offload settings: Operation not supported
rx-checksumming: off
tx-checksumming: off
[root@cep03 ~]# iperf3 -c 192.168.11.100 -R
Connecting to host 192.168.11.100, port 5201
Reverse mode, remote host 192.168.11.100 is sending
[  4] local 192.168.11.103 port 51274 connected to 192.168.11.100 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec  22.6 KBytes   185 Kbits/sec                  
[  4]   1.00-2.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   2.00-3.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   3.00-4.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   4.00-5.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   5.00-6.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   6.00-7.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   7.00-8.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   8.00-9.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   9.00-10.00  sec  0.00 Bytes  0.00 bits/sec                  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth       Retr
[  4]   0.00-10.00  sec   109 KBytes  89.2 Kbits/sec   18             sender
[  4]   0.00-10.00  sec  22.6 KBytes  18.5 Kbits/sec                  receiver

iperf Done.
[root@cep03 ~]# ethtool -S eth0 | grep err | grep -v ": 0"
    mmc_rx_crc_error: 33
    mmc_rx_udp_err: 4
    mmc_rx_tcp_err: 29
    mmc_rx_udp_err_octets: 224
    mmc_rx_tcp_err_octets: 42920
[root@cep03 ~]# ip link set eth0 down
[root@cep03 ~]# ip link set mtu 1200 dev eth0
[root@cep03 ~]# ip link set eth0 up
[  136.029978] RTL8211F Gigabit Ethernet stmmac-0:00: attached PHY driver [RTL8211F Gigabit Ethernet] (mii_bus:phy_addr=stmmac-0:00, irq=POLL)
[  136.044509] rk_gmac-dwmac ff540000.ethernet eth0: RX IPC Checksum Offload disabled
[  136.045250] rk_gmac-dwmac ff540000.ethernet eth0: PTP not supported by HW
[root@cep03 ~]# [  139.110570] rk_gmac-dwmac ff540000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx

[root@cep03 ~]# ip link show dev eth0
3: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1200 qdisc mq state UP mode DEFAULT group default qlen 1000
   link/ether 66:ad:ab:e3:95:a8 brd ff:ff:ff:ff:ff:ff
[root@cep03 ~]# iperf3 -c 192.168.11.100 -R
Connecting to host 192.168.11.100, port 5201
Reverse mode, remote host 192.168.11.100 is sending
[  4] local 192.168.11.103 port 51278 connected to 192.168.11.100 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec  14.6 KBytes   119 Kbits/sec                  
[  4]   1.00-2.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   2.00-3.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   3.00-4.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   4.00-5.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   5.00-6.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   6.00-7.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   7.00-8.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   8.00-9.00   sec  0.00 Bytes  0.00 bits/sec                  
[  4]   9.00-10.00  sec  0.00 Bytes  0.00 bits/sec                  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth       Retr
[  4]   0.00-10.00  sec  42.6 KBytes  34.9 Kbits/sec   29             sender
[  4]   0.00-10.00  sec  14.6 KBytes  11.9 Kbits/sec                  receiver

iperf Done.
[root@cep03 ~]# ethtool -S eth0 | grep err | grep -v ": 0"
    mmc_rx_crc_error: 33
    mmc_rx_tcp_err: 32
    mmc_rx_icmp_err: 1
    mmc_rx_tcp_err_octets: 38060
    mmc_rx_icmp_err_octets: 72
[root@cep03 ~]#

Would you please be so kind and show your results?

[thewonderer]

What problems are occuring with the GbE?  Drop outs?  Does the rock64 recover automatically or is it just a slow down in performance?

https://forum.pine64.org/showthread.php?...1#pid32501

It is not just a slow down, Ethernet is unusable at 1 Gbit/s speed in the mainline kernel.
I did not try the newest kernels (4.15-rc).

[dkryder]

i would tend to think you would want to raise mtu for the GbE and not lower it. i mean why send a smaller payload? i have not had to tweak my ethernet because i don't have a fast connection so i have no advice to give but it just seems incrrect to lower mtu. maybe 15 years ago but with GbE also?

[maal]

This is pretty serious issue with the NIC.  Are there any Linux builds that don\t have this issue? 0.69 or until the Kernel gets fixed up, this is an issue for all users?  Hopefully a workaround can be sorted out.

The issue we are talking about occurs with mainline kernel only.
Almost all linux distros for Rock64 come with kernel based on Rockchip sources - no issues there, GbE works fine (sometimes tweaking required).
Users of those distros are not affected in any way.

i would tend to think you would want to raise mtu for the GbE and not lower it.

I've tried many different values of mtu (up to max allowed by hardware 3582). None of these worked.