On Fri, 2017-02-17 at 22:30 +0100, gmbh@gazeta.pl wrote: > On 17.02.2017 02:57, Ben Hutchings wrote: > > This seems to be a bug in the device tree for the TS419 - it doesn't > > quite match the board code it replaced. Should be easy to fix. > > After applying patch, eth1 is recognized, but it behaves strange. It > looks like it is using a correct MAC during DHCP phase, but after > getting IP from DHCP it clones(?) eth0 MAC. I don't think so. [...] > Here is an relevant output from dmesg: > > [ 0.045321] [Firmware Info]: > /ocp@f1000000/ethernet-controller@72000/ethernet0-port@0: > local-mac-address is not set > [ 0.045412] [Firmware Info]: > /ocp@f1000000/ethernet-controller@76000/ethernet1-port@0: > local-mac-address is not set > [ 60.981436] mv643xx_eth: MV-643xx 10/100/1000 ethernet driver version 1.4 > [ 61.917900] mv643xx_eth_port mv643xx_eth_port.0 eth0: port 0 with MAC > address 00:08:9b:cb:ae:be > [ 62.441157] mv643xx_eth_port mv643xx_eth_port.1 eth1: port 0 with MAC > address 00:08:9b:cb:ae:bf OK, so this worked. [...] > And here is an output from arp on router: > > IP address HW type Flags HW address Mask > Device > 192.168.1.101 0x1 0x2 00:08:9b:cb:ae:be * > br-lan > 192.168.1.100 0x1 0x2 00:08:9b:cb:ae:be * > br-lan > > both IP addresses have same MAC (from eth0). > > I'm not sure what is going here... This is not a driver issue. See: https://serverfault.com/questions/22253/ubuntu-linux-multiple-nics-same-lan-arp-responses-always-go-out-a-single-n (which has two possible solutions). (I do wonder whether having two links to the same VLAN without bonding is at all sensible, anyway.) > BTW. Is this .dtsi file responsible for GPIO configuration? All model-specific stuff is now defined by device trees. The definition for this model is spread across kirkwood-ts419-6282.dts and four other files you can see listed at the top of it. > I've also noticed a strange behavior with qcontrol: > https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=795558 So far as I can remember, device trees do not have a way to request that a GPIO should be automatically exported to userland. That will have to be done by qcontrol. Ben. -- Ben Hutchings Any sufficiently advanced bug is indistinguishable from a feature.
Attachment:
signature.asc
Description: This is a digitally signed message part