Latest i40e/iavf driver does not propagate carrier, preventing bond failover (X710-DA2)

[kayson1]

I have an X710-DA2 with a relatively new firmware, the latest drivers, and Linux kernel 6.17.13 (a proxmox-compiled version). I have several VFs set up on both PFs, and Linux network bonds on top of the VFs. When the physical link goes down, the PF link is brought down, but the VFs stay up. I can see that i40e brings the carrier down, but it doesn't propagate to iavf/the VF's.

(The 0 here is the PF, the 1's are the VF).

> cat /sys/class/net/enp1s0f0*/carrier
0
1
1
1
1

This means that bonds won't failover, VM's with VF's won't see their link go down even though they are, etc.

I can see that the RTNETLINK event fires for the PF:

# ip monitor link
4: enp1s0f0np0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN group default 
link/ether 68:05:ca:9e:a2:70 brd ff:ff:ff:ff:ff:ff

but there are no events for the VF's (nothing on udev either).

Is this expected? If so, what is the recommended way to bring VF links down when the PF is physically disconnected?

Thanks!

NIC Info:

> ethtool -i enp1s0f0np0
driver: i40e
version: 2.28.16
firmware-version: 9.50 0x8000f4ff 1.3682.0
expansion-rom-version: 
bus-info: 0000:01:00.0
supports-statistics: yes
supports-test: yes
supports-eeprom-access: yes
supports-register-dump: yes
supports-priv-flags: yes

> ethtool -i enp1s0f0v0 
driver: iavf
version: 4.13.20
firmware-version: N/A
expansion-rom-version: 
bus-info: 0000:02:02.0
supports-statistics: yes
supports-test: no
supports-eeprom-access: no
supports-register-dump: no
supports-priv-flags: yes

[Fikri_Intel]

Hi kayson1,

Thank you for contacting Intel.

We appreciate you reaching out to us regarding the reported issue. To assist you more effectively and provide an accurate solution, we kindly request the following details:

1- What system model are you currently using? (Brand/Model)

2- Does the X710-DA2 come with a system or purchased the network card separately?

3- Please elaborate more regarding the issue.

4- What troubleshooting has been done?

5- Have any changes been made prior to the issue happening?

Along with that, have you checked the support matrix below:

Intel® Ethernet Controller X710/XXV710/XL710 Feature Support Matrix

Once we have this information, we will be able to guide you better and provide the appropriate support.

Looking forward to your response.

Regards,

Fikri_Intel

[kayson1]

Hi Fikri_Intel,

1. Lenovo ThinkCentre Tiny M920Q
2. Purchased separately
3. There is an issue with the driver where if the network is physically disconnected, the interface associated with the device Physical Function (PF) goes down, but the interfaces associated with the SR-IOV Virtual Functions (VF) do not. This means that guest VMs with a VF NIC won't know their interfaces are down.

4. Updated driver and firmware, tried different interfaces files, udev, etc.

5. No this issue has always existed.