Rock64: General ethernet and USB flakiness

[marcosscriven]

I got the Rock64 specifically for router-like functionality.

 

In addition to the native ethernet port, I've added two USB3 RTL8153 Gigabit adapters: http://www.realtek.com.tw/products/productsView.aspx?Langid=1&PFid=56&Level=5&Conn=4&ProdID=326

 

I've tried multiple 0.5.x and 0.6.x Ayufan images, and also Stretch armbian here (which also uses the Ayufan kernel).

 

I've tried both the 4.4 kernel all these distros come with, and up to 4.17 as well.

 

In all cases I see terrible and unreliable USB3/USB2 and ethernet performance.

 

In the Armbian Sketch image I see 'eth1' go up and down contstantly:

 

Jun 17 09:48:44 localhost avahi-daemon[712]: Joining mDNS multicast group on interface eth1.IPv6 with address fe80::e824:33dd:2309:841c.
Jun 17 09:48:44 localhost avahi-daemon[712]: New relevant interface eth1.IPv6 for mDNS.
Jun 17 09:48:44 localhost avahi-daemon[712]: Registering new address record for fe80::e824:33dd:2309:841c on eth1.*.
Jun 17 09:48:45 localhost kernel: [  249.804881] Rockchip integrated EPHY stmmac-1:00: rockchip_integrated_phy_fixup_100M: reset phy
Jun 17 09:48:47 localhost kernel: [  251.863022] rk_gmac-dwmac ff550000.ethernet eth1: Link is Down
Jun 17 09:48:48 localhost kernel: [  252.865169] rk_gmac-dwmac ff550000.ethernet eth1: Link is Up - 100Mbps/Half - flow con8:48 localhost NetworkManager[708]: <info>  [1529228928.6344] deort 67 interval 5
Jun 17 09:48:56 localhost systemd[1]: Starte00: rockchip_integrated_phy_fixup_100M: reset phy
Jun 17 09:49:04 localhost kernel: [  268.951741] rk_gmac-dwmac ff550000.ethernet eth1: Link is Down
Jun 17 09:49:05 localhost kernel: [  269.953899] rk_gmac-dwmac ff550000.ethernet eth1: Link is Up - 100Mbps/Half - flow control off
Jun 17 09:49:05 localhost NetworkManager[708]: <info>  [1529228945.7255] device (eth1): link connected
Jun 17 09:49:07 localhost dhclient[3157]: DHCPDISCOVER on eth1 to 255.255.255.255 port 67 interval 21
Jun 17 09:49:19 localhost kernel: [  283.986689] Rockchip integrated EPHY stmmac-1:00: rockchip_integrated_phy_fixup_100M: reset phy
Jun 17 09:49:21 localhost kernel: [  286.046267] rk_gmac-dwmac ff550000.ethernet eth1: Link is Down
Jun 17 09:49:22 localhost kernel: [  287.048219] rk_gmac-dwmac ff550000.ethernet eth1: Link is Up - 100Mbps/Half - flow control off
Jun 17 09:49:22 localhost NetworkManager[708]: <info>  [1529228962.8166] device (eth1): link connected
Jun 17 09:49:27 localhost NetworkManager[708]: <warn>  [1529228967.8545] dhcp4 (eth1): request timed out
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.8553] dhcp4 (eth1): state changed unknown -> timeout
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.8724] dhcp4 (eth1): canceled DHCP transaction, DHCP client pid 3157
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.8728] dhcp4 (eth1): state changed timeout -> done
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.8792] device (eth1): state change: ip-config -> failed (reason 'ip-config-unavailable') [70 120 5]
Jun 17 09:49:27 localhost NetworkManager[708]: <warn>  [1529228967.8834] device (eth1): Activation: failed for connection 'Wired connection 2'
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.8915] device (eth1): state change: failed -> disconnected (reason 'none') [120 30 0]
Jun 17 09:49:27 localhost avahi-daemon[712]: Withdrawing address record for fe80::e824:33dd:2309:841c on eth1.
Jun 17 09:49:27 localhost avahi-daemon[712]: Leaving mDNS multicast group on interface eth1.IPv6 with address fe80::e824:33dd:2309:841c.
Jun 17 09:49:27 localhost avahi-daemon[712]: Interface eth1.IPv6 no longer relevant for mDNS.
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.9290] policy: auto-activating connection 'Wired connection 2'
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.9487] device (eth1): Activation: starting connection 'Wired connection 2' (2f03db29-a92b-3782-86d7-64b613f8e9d0)
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.9530] device (eth1): state change: disconnected -> prepare (reason 'none') [30 40 0]
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.9630] device (eth1): state change: prepare -> config (reason 'none') [40 50 0]
Jun 17 09:49:27 localhost NetworkManager[708]: <info>  [1529228967.9829] device (eth1): state change: config -> ip-config (reason 'none') [50 70 0]
Jun 17 09:49:28 localhost NetworkManager[708]: <info>  [1529228967.9949] dhcp4 (eth1): activation: beginning transaction (timeout in 45 seconds)
Jun 17 09:49:28 localhost NetworkManager[708]: <info>  [1529228968.0254] dhcp4 (eth1): dhclient started with pid 3425
Jun 17 09:49:28 localhost dhclient[3425]: DHCPDISCOVER on eth1 to 255.255.255.255 port 67 interval 7

I've read now multiple reports of these issues. See:

 

* https://github.com/ayufan-rock64/linux-build/issues/112#issuecomment-397817650

* https://forum.pine64.org/showthread.php?tid=5557

 

What I can't find is how to resolve this? It seems Rock64 don't actually support this board properly, and just rely on Ayufan.

[marcosscriven]

Further, I can fix the errors above with 'ifconfig eth1 down' - I'm not sure why that interface is even there, it seems to be an IPV6 version of the eth0 interface.

 

Anyway, with that noise out of the way, here's what I see when I try to use both USB3 ethernet interfaces:

 

un 16 20:04:04 localhost kernel: [  273.782909] NETDEV WATCHDOG: enx00e04c68d929 (r8152): transmit queue 0 timed out
Jun 16 20:04:04 localhost kernel: [  273.783169] ------------[ cut here ]------------
Jun 16 20:04:04 localhost kernel: [  273.783183] WARNING: at net/sched/sch_generic.c:306
Jun 16 20:04:04 localhost kernel: [  273.783191] Modules linked in: tun r8152 poly_hash_ce ip_tables x_tables
Jun 16 20:04:04 localhost kernel: [  273.783223]
Jun 16 20:04:04 localhost kernel: [  273.783247] CPU: 2 PID: 2927 Comm: unattended-upgr Not tainted 4.4.124-rk3328 #24
Jun 16 20:04:04 localhost kernel: [  273.783257] Hardware name: Pine64 Rock64 (DT)
Jun 16 20:04:04 localhost kernel: [  273.783271] task: ffffffc0e46b4380 task.stack: ffffffc0e4500000
Jun 16 20:04:04 localhost kernel: [  273.783304] PC is at dev_watchdog+0x1e4/0x234
Jun 16 20:04:04 localhost kernel: [  273.783315] LR is at dev_watchdog+0x1e4/0x234

With an intervening stack dump, and ending in:

Jun 16 20:04:06 localhost kernel: [  275.908967] usb 5-1.2: reset SuperSpeed USB device number 4 using xhci-hcd

At which point one of the USB 3 ethernet adapters dies.