Bug#1031839: network devices do not work after kernel upgrade
Hi Thorsten,
On 2/25/23 11:56, Thorsten Alteholz wrote:
> The I225-V are working fine, the other four make trouble.
right, but those are copper interfaces.
> I am using transceiver modules AXS85-192-M3 from 10Gtek.
It looks like they are not flashable (like flexoptix and others), so I
presume these are "non-Intel"-branded.
> The allow_unsupported_sfp=1 does not make a difference. Shouldn't there
> be at least a syslog message if an unsupported sfp is detected?
Just to be extra sure, you've added:
ixgbe.allow_unsupported_sfp=1
to your kernel cmdline, right?
I've checked with an up2date bookworm test-server and kernel 6.1.12-1..
when inserting a Intel-branded flexoptix SFP, I'll get this message:
Feb 27 14:59:06 xxx kernel: ixgbe 0000:5e:00.0 ens2f0: detected SFP+: 5
whereas with a Arista-branded one, I'll get the same:
Feb 27 15:00:30 xxx kernel: ixgbe 0000:5e:00.0 ens2f0: detected SFP+: 5
Just to document it.. this is dmesg from after rebooting the machine,
there's no allow_unsupported_sfp set at all, and there's an
Intel-branded SFP in one slot (ens2f0), an Arista-branded one in the
other (ens2f1):
[ 3.178235] ixgbe: Intel(R) 10 Gigabit PCI Express Network Driver
[ 3.178385] ixgbe: Copyright (c) 1999-2016 Intel Corporation.
[ 3.927189] ixgbe 0000:5e:00.0: Multiqueue Enabled: Rx Queue count =
63, Tx Queue count = 63 XDP Queue count = 0
[ 3.927726] ixgbe 0000:5e:00.0: 32.000 Gb/s available PCIe bandwidth
(5.0 GT/s PCIe x8 link)
[ 3.928058] ixgbe 0000:5e:00.0: MAC: 2, PHY: 19, SFP+: 5, PBA No:
0210FF-0FF
[ 3.928320] ixgbe 0000:5e:00.0: 0c:c4:7a:8f:48:f6
[ 3.940393] ixgbe 0000:5e:00.0: Intel(R) 10 Gigabit Network Connection
[ 4.121469] ixgbe 0000:5e:00.1: Multiqueue Enabled: Rx Queue count =
63, Tx Queue count = 63 XDP Queue count = 0
[ 4.122320] ixgbe 0000:5e:00.1: 32.000 Gb/s available PCIe bandwidth
(5.0 GT/s PCIe x8 link)
[ 4.122938] ixgbe 0000:5e:00.1: MAC: 2, PHY: 20, SFP+: 6, PBA No:
0210FF-0FF
[ 4.123483] ixgbe 0000:5e:00.1: 0c:c4:7a:8f:48:f7
[ 4.134330] ixgbe 0000:5e:00.1: Intel(R) 10 Gigabit Network Connection
[ 4.315894] ixgbe 0000:5e:00.0 ens2f0: renamed from eth2
[ 4.411504] ixgbe 0000:5e:00.1 ens2f1: renamed from eth3
[ 8.043579] ixgbe 0000:5e:00.0: registered PHC device on ens2f0
[ 8.227099] ixgbe 0000:5e:00.0 ens2f0: detected SFP+: 5
it's not obvious from the messages that one SFP is working and the other
one is not.
the only difference I can see is that the one with the Intel-branded SFP
has this line:
[ 8.043579] ixgbe 0000:5e:00.0: registered PHC device on ens2f0
there's also no difference wrt/ the debug level (I've testet with printk
set to 7 and 8, no additional messages are shown).
Hope that helps - my guess would be to try and verify with an Intel or
Intel-flashed SFP to rule out.
Hope that helps.
Regards,
Daniel
Reply to: