Cannot modify sriov_numvfs when pod is running on another vf

[Caio Bruchert]

The existing tc "test_create_sriov_when_pod_running_on_a_vf_intf_on_other_sriov" tries to modify the amount of vf and it does not increase
Running command :

system host-if-modify controller-1 enp191s0f1 --ifname sriov425 --ifclass pci-sriov --num-vfs 64 --vf-driver netdevice'

Steps to Reproduce

Choose an interface and a controller: in this case controller-0 and enp181s0f1

Setup steps with controller-1 as active controller:

system host-lock controller-0
system host-if-add controller-0 test_intf vf sriov0 --imtu 1500 --vf-driver vfio --ifclass pci-sriov --num-vfs 1 --max-tx-rate 36000
system datanetwork-add --mtu=1500 testnetwork vlan
system interface-datanetwork-assign controller-0 test_intf testnetwork
system host-port-show controller-0 enp181s0f1 # Take note of number of sriov_totalvfs (64) and sriov_numvfs (0) pick sriov_totalvfs:

Test steps:

system host-port-show controller-0 enp181s0f1
system host-if-modify controller-0 enp181s0f1 --ifname sriov425 --ifclass pci-sriov --num-vfs 64 --vf-driver netdevice
system host-unlock controller-0 at this poit you will get the message:
   "Expecting number of interface sriov_numvfs=64. Please wait a few minutes for inventory update and retry host-unlock"
After some minutes it could be checked using: system host-port-show controller-0 enp181s0f1
    and it remains in sriov_numvfs=0
Expected Behavior

The sriov_numvfs of the interface chosen by tc increment to the desired value

Actual Behavior

After command execution, it seems to do nothing

Reproducibility

<Reproducible>

System Configuration

A duplex system, IPv4.

Timestamp/Logs

[2022-04-19 21:10:09,440] 135 INFO MainThread utils.wait_for_expected_exit_code:: expected exit code: 0, actual exit code: 1, output is 'Expecting number of interface sriov_numvfs=64. Please wait a few minutes for inventory update and retry host-unlock.'

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/starlingx/config/+/844324

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/starlingx/config/+/844324
Committed: https://opendev.org/starlingx/config/commit/1288b7e99b8a3a4d80b593be159c68dc0b7409b6
Submitter: "Zuul (22348)"
Branch: master

commit 1288b7e99b8a3a4d80b593be159c68dc0b7409b6
Author: Caio Bruchert <email address hidden>
Date: Wed Jun 1 13:06:48 2022 -0400

    Fix N3000 onboard device PCI address mismatch

    In a non-simplex system, when an N3000 FPGA is reset, its onboard
    devices can have their PCI addresses changed after the database is
    populated.
    When this happens some commands can try to use the outdated address
    and fail (usually after checking the PCI address in
    /sys/bus/pci/devices/)
    This fix detects PCI address changes on 0d58 devices and updates the
    database accordingly.

    Test Plan:
        PASS: New unit tests using simplex and non-simplex simulation
        PASS: Configure pci-sriov interfaces on onboard N3000 devices
        PASS: Unlock and check Puppet logs for errors
        PASS: Upgrade AIO-DX

    Closes-Bug: 1976540

    Change-Id: I57ef19011b8f4d9a98ada4db66b5c7cfcb12de21
    Signed-off-by: Caio Bruchert <email address hidden>