[REGRESSION] v6.2-rc1 on Turris Omnia: DSA switch ports stuck in NO-CARRIER state

[Klaus Kudielka]

Hello,



I'm facing a regression with v6.2-rc1 on Turris Omnia (Armada 385).
All ports of the Marvell 88E6176 switch remain in NO-CARRIER state,
even if devices are connected.

Links are negotiated properly by the hardware, but the kernel doesn't
get it. End result: No network connectivity via the switch.

The usual (working) behaviour at the end of the boot process is (dmesg):
[    4.328259] mv88e6085 f1072004.mdio-mii:10 lan1: Link is Up - 1Gbps/Full - flow control rx/tx
[    4.328284] IPv6: ADDRCONF(NETDEV_CHANGE): lan1: link becomes ready
[    4.328344] br0: port 4(lan1) entered blocking state
[    4.328353] br0: port 4(lan1) entered forwarding state
[    4.329331] IPv6: ADDRCONF(NETDEV_CHANGE): br0: link becomes ready
But this just DOESN'T happen with v6.2-rc1.



I bisected this issue (v6.1..v6.2-rc1), and the offending commit is:

c4de4667f15d ARM: dts: armada-38x: Fix compatible string for gpios

Reverting this single commit (on top of v6.2-rc1) "fixes" the issue
for me, and I have network connectivity again.



As further diagnosis, here the relevant content of /proc/interrupts:

*** GOOD (v6.2-rc1 with c4de4667f15d reverted) ***
 43:          3          0  f1018140.gpio  13 Level     mv88e6xxx-f1072004.mdio-mii:10
 64:          0          0  mv88e6xxx-g1   3 Edge      mv88e6xxx-f1072004.mdio-mii:10-g1-atu-prob
 66:          0          0  mv88e6xxx-g1   5 Edge      mv88e6xxx-f1072004.mdio-mii:10-g1-vtu-prob
 68:          0          3  mv88e6xxx-g1   7 Edge      mv88e6xxx-f1072004.mdio-mii:10-g2
 70:          0          0  mv88e6xxx-g2   0 Edge      mv88e6xxx-1:00
 71:          0          1  mv88e6xxx-g2   1 Edge      mv88e6xxx-1:01
 72:          0          0  mv88e6xxx-g2   2 Edge      mv88e6xxx-1:02
 73:          0          0  mv88e6xxx-g2   3 Edge      mv88e6xxx-1:03
 74:          0          2  mv88e6xxx-g2   4 Edge      mv88e6xxx-1:04
 85:          0          0  mv88e6xxx-g2  15 Edge      mv88e6xxx-f1072004.mdio-mii:10-watchdog

*** BAD (v6.2-rc1) ***
 43:          0          0  f1018140.gpio  13 Level     mv88e6xxx-f1072004.mdio-mii:10
 64:          0          0  mv88e6xxx-g1   3 Edge      mv88e6xxx-f1072004.mdio-mii:10-g1-atu-prob
 66:          0          0  mv88e6xxx-g1   5 Edge      mv88e6xxx-f1072004.mdio-mii:10-g1-vtu-prob
 68:          0          0  mv88e6xxx-g1   7 Edge      mv88e6xxx-f1072004.mdio-mii:10-g2
 70:          0          0  mv88e6xxx-g2   0 Edge      mv88e6xxx-1:00
 71:          0          0  mv88e6xxx-g2   1 Edge      mv88e6xxx-1:01
 72:          0          0  mv88e6xxx-g2   2 Edge      mv88e6xxx-1:02
 73:          0          0  mv88e6xxx-g2   3 Edge      mv88e6xxx-1:03
 74:          0          0  mv88e6xxx-g2   4 Edge      mv88e6xxx-1:04
 85:          0          0  mv88e6xxx-g2  15 Edge      mv88e6xxx-f1072004.mdio-mii:10-watchdog

(The mv88e6xxx driver doesn't seem to receive interrupts anymore).



Best regards, Klaus

#regzbot introduced: c4de4667f15d