[Bug 217558] New: In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[Bug 217558] New: In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

            Bug ID: 217558
           Summary: In KVM guest with VF of X710 NIC passthrough, the mac
                    address of VF is inconsistent with it in host
           Product: Virtualization
           Version: unspecified
          Hardware: Intel
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P3
         Component: kvm
          Assignee: virtualization_kvm@kernel-bugs.osdl.org
          Reporter: farrah.chen@intel.com
        Regression: No

Environment:

kernel: https://git.kernel.org/pub/scm/virt/kvm/kvm.git branch: next commit:
39428f6e kernel version:6.4.0-rc2
Qemu: https://gitlab.com/qemu-project/qemu.git branch: master commit: 7efd6542
Host OS: CentOS stream 9
Guest OS: CentOS stream 9
CPU:
Architecture:            x86_64
Vendor ID:               GenuineIntel
  BIOS Vendor ID:        Intel(R) Corporation
  Model name:            Intel(R) Xeon(R) Platinum

Bug detail description: 

Create VF of X710 NIC on host, create VM with VF passthrough, there is a
probability that the mac of VF in guest is inconsistent with it in host, the VF
in guest gets a random mac. 

Reproduce steps: 

1.Create two VFs of NIC X710:

[root@spr-2s2 ~]# lspci -k -s 98:00.0
98:00.0 Ethernet controller: Intel Corporation Ethernet Controller X710 for
10GbE SFP+ (rev 01)
        Subsystem: Intel Corporation Ethernet Converged Network Adapter X710-2
        Kernel driver in use: i40e
        Kernel modules: i40e
[root@spr-2s2 ~]# echo 2 > /sys/bus/pci/devices/0000:98:00.0/sriov_numvfs

2. Check mac and driver of the 2 VFs

[root@spr-2s2 xf]# ip address

14: ens28f0v0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP
group default qlen 1000
    link/ether 32:40:f7:6a:dc:8a brd ff:ff:ff:ff:ff:ff
    altname enp152s0f0v0
15: ens28f0v1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP
group default qlen 1000
    link/ether 6e:bd:a8:ee:83:c4 brd ff:ff:ff:ff:ff:ff
    altname enp152s0f0v1

[root@spr-2s2 xf]# ethtool -i ens28f0v0
driver: iavf
version: 6.4.0-rc2
bus-info: 0000:98:02.0
[root@spr-2s2 xf]# ethtool -i ens28f0v1
driver: iavf
version: 6.4.0-rc2
bus-info: 0000:98:02.1

[root@spr-2s2 xf]# lspci -k -s 98:02.0
98:02.0 Ethernet controller: Intel Corporation Ethernet Virtual Function 700
Series (rev 01)
        Subsystem: Intel Corporation Device 0000
        Kernel driver in use: iavf
        Kernel modules: iavf
[root@spr-2s2 xf]# lspci -k -s 98:02.1
98:02.1 Ethernet controller: Intel Corporation Ethernet Virtual Function 700
Series (rev 01)
        Subsystem: Intel Corporation Device 0000
        Kernel driver in use: iavf
        Kernel modules: iavf

3. Unbind the VFs from iavf driver and bind them to vfio-pci

[root@spr-2s2 xf]# echo 0000:98:02.0 >
/sys/bus/pci/devices/0000\:98\:02.0/driver/unbind
[root@spr-2s2 xf]# echo 0000:98:02.1 >
/sys/bus/pci/devices/0000\:98\:02.1/driver/unbind

[root@spr-2s2 xf]# modprobe vfio-pci
[root@spr-2s2 xf]# lspci -n -s 98:02.0
98:02.0 0200: 8086:154c (rev 01)
[root@spr-2s2 xf]# echo 8086 154c > /sys/bus/pci/drivers/vfio-pci/new_id
[root@spr-2s2 xf]# lspci -k -s 98:02.0
98:02.0 Ethernet controller: Intel Corporation Ethernet Virtual Function 700
Series (rev 01)
        Subsystem: Intel Corporation Device 0000
        Kernel driver in use: vfio-pci
        Kernel modules: iavf
[root@spr-2s2 xf]# lspci -k -s 98:02.1
98:02.1 Ethernet controller: Intel Corporation Ethernet Virtual Function 700
Series (rev 01)
        Subsystem: Intel Corporation Device 0000
        Kernel driver in use: vfio-pci
        Kernel modules: iavf

4.Create guest with these 2 VFs passthrough

qemu-system-x86_64 -accel kvm -m 4096 -cpu host -drive
file=centos9.qcow2,if=none,id=virtio-disk0 -device
virtio-blk-pci,drive=virtio-disk0,bootindex=0 -smp 4 -device
vfio-pci,host=98:02.0 -net none -device vfio-pci,host=98:02.1 -net none
-daemonize -vnc :5

5. Check the mac and ip address of the these 2 VFs in guest

[root@localhost ~]# ip ad
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group
default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: ens4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group
default qlen 1000
    link/ether c2:4e:37:99:fc:ff brd ff:ff:ff:ff:ff:ff
    altname enp0s4
    inet 192.168.111.27/20 brd 192.168.111.255 scope global dynamic
noprefixroute ens4
       valid_lft 278sec preferred_lft 278sec
    inet6 fe80::ac6c:bd39:6513:d29/64 scope link noprefixroute
       valid_lft forever preferred_lft forever
3: ens5: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group
default qlen 1000
    link/ether 6e:bd:a8:ee:83:c4 brd ff:ff:ff:ff:ff:ff
    altname enp0s5
    inet 192.168.111.26/20 brd 192.168.111.255 scope global dynamic
noprefixroute ens5
       valid_lft 276sec preferred_lft 276sec
    inet6 fe80::7a32:aa2c:c2ee:8d5d/64 scope link noprefixroute
       valid_lft forever preferred_lft forever

The mac of VF1 in host is 32:40:f7:6a:dc:8a, but it in guest is
c2:4e:37:99:fc:ff, VF2 gets the same mac 6e:bd:a8:ee:83:c4

The guest dmesg about iavf:

[    3.516623] iavf: Intel(R) Ethernet Adaptive Virtual Function Network Driver
[    3.517302] Copyright (c) 2013 - 2018 Intel Corporation.
[    3.588858] ppdev: user-space parallel port driver
[    3.590703] iavf 0000:00:04.0: Invalid MAC address 00:00:00:00:00:00, using
random
[    3.592382] iavf 0000:00:04.0: Multiqueue Enabled: Queue pair count = 4
[    3.593697] iavf 0000:00:05.0: Multiqueue Enabled: Queue pair count = 4
[    3.594086] iavf 0000:00:04.0: MAC address: c2:4e:37:99:fc:ff
[    3.595297] iavf 0000:00:04.0: GRO is enabled
[    3.595932] iavf 0000:00:05.0: MAC address: 6e:bd:a8:ee:83:c4
[    3.596503] iavf 0000:00:05.0: GRO is enabled
[    3.598333] iavf 0000:00:04.0 ens4: renamed from eth0
[    3.606666] XFS (vda2): Mounting V5 Filesystem
8736b23e-ddde-4cca-9166-2623d2e57e5a
[    3.606987] iavf 0000:00:05.0 ens5: renamed from eth1
[    3.614474] XFS (vda2): Ending clean mount
[    3.692511] RPC: Registered named UNIX socket transport module.
[    3.693055] RPC: Registered udp transport module.
[    3.693665] RPC: Registered tcp transport module.
[    3.694243] RPC: Registered tcp NFSv4.1 backchannel transport module.
[    4.125995] iavf 0000:00:04.0 ens4: NIC Link is Up Speed is 10 Gbps Full
Duplex
[    4.126628] IPv6: ADDRCONF(NETDEV_CHANGE): ens4: link becomes ready
[    4.158025] iavf 0000:00:05.0 ens5: NIC Link is Up Speed is 10 Gbps Full
Duplex
[    4.158648] IPv6: ADDRCONF(NETDEV_CHANGE): ens5: link becomes ready

Currect Result:

There is a probability that the mac of VF in guest is inconsistent with it in
host

Expected Result

When passthrough NIC VF to guest, the VF in guest should always get the same
mac as it in host.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

Chen, Fan (farrah.chen@intel.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Bisected commit-id|                            |6.0.0-rc7
                 CC|                            |farrah.chen@intel.com,
                   |                            |xudong.hao@intel.com
     Kernel Version|                            |6.4.0-rc2
         Regression|No                          |Yes

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #1 from Chen, Fan (farrah.chen@intel.com) ---
This issue is not found from this commit, it has been going on for some days,
we found kernel 6.0.0-rc7 is good, but not sure when it started from.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

Bagas Sanjaya (bagasdotme@gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bagasdotme@gmail.com

--- Comment #2 from Bagas Sanjaya (bagasdotme@gmail.com) ---
(In reply to Chen, Fan from comment #1)
> This issue is not found from this commit, it has been going on for some
> days, we found kernel 6.0.0-rc7 is good, but not sure when it started from.

Can you please then perform bisection between 6.0-rc7 and 6.4-rc2?

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #3 from Chen, Fan (farrah.chen@intel.com) ---
We bisect and found the first bad commit, I don't understand if this commit is
intend to do so? Make the mac of VF in VM different from it in host? We think
they use the same mac is better for users, with the same mac, the mac of VF is
known for us before creating VM, then we can get IP address of the VF interface
in VM from mac without using vnc or other UI or serial port. How can we make
the mac of VF keep the same in host and VM now?

commit ceb29474bbbc377e11be3a70589a0005305e4fc3
Author: Sylwester Dziedziuch <sylwesterx.dziedziuch@intel.com>
Date:   Thu Mar 30 10:00:22 2023 -0700

    i40e: Add support for VF to specify its primary MAC address

    Currently in the i40e driver there is no implementation of different
    MAC address handling depending on whether it is a legacy or primary.
    Introduce new checks for VF to be able to specify its primary MAC
    address based on the VIRTCHNL_ETHER_ADDR_PRIMARY type.

    Primary MAC address are treated differently compared to legacy
    ones in a scenario where:
    1. If a unicast MAC is being added and it's specified as
    VIRTCHNL_ETHER_ADDR_PRIMARY, then replace the current
    default_lan_addr.addr.
    2. If a unicast MAC is being deleted and it's type
    is specified as VIRTCHNL_ETHER_ADDR_PRIMARY, then zero the
    hw_lan_addr.addr.

Actually, we also tried to use "ip link set ens28f0 vf 0 mac <mac>" to fix a
mac for VF, it works in host, but we create VM with this VF passthrough, it has
a high probability failed to assign the mac to VF in vm, and the interface
failed yo get IP either, error dmesg in VM: 
[    3.037955] iavf 0000:00:04.0: Invalid MAC address 00:00:00:00:00:00, using
random
[    3.039523] iavf 0000:00:04.0: Multiqueue Enabled: Queue pair count = 4
[    3.040466] iavf 0000:00:04.0: MAC address: 3a:38:ca:66:f9:65
[    3.040980] iavf 0000:00:04.0: GRO is enabled
[    3.042240] iavf 0000:00:05.0: Multiqueue Enabled: Queue pair count = 4
[    3.043232] iavf 0000:00:05.0: MAC address: 36:91:96:9d:5d:05
[    3.043770] iavf 0000:00:05.0: GRO is enabled
[    3.044401] iavf 0000:00:04.0 ens4: renamed from eth0
[    3.049199] iavf 0000:00:05.0 ens5: renamed from eth1
[    3.072576] ppdev: user-space parallel port driver
[    3.094077] XFS (vda2): Mounting V5 Filesystem
8736b23e-ddde-4cca-9166-2623d2e57e5a
[    3.102245] XFS (vda2): Ending clean mount
[    3.550242] iavf 0000:00:04.0: Failed to add MAC filter, error IAVF_ERR_NVM
[    3.617176] iavf 0000:00:04.0 ens4: NIC Link is Up Speed is 10 Gbps Full
Duplex
[    3.618173] IPv6: ADDRCONF(NETDEV_CHANGE): ens4: link becomes ready
[    3.628189] iavf 0000:00:05.0 ens5: NIC Link is Up Speed is 10 Gbps Full
Duplex
[    3.629039] IPv6: ADDRCONF(NETDEV_CHANGE): ens5: link becomes ready

Fix this issue will be very helpful for us.

I failed to CC the primary author sylwesterx.dziedziuch@intel.com and
mateusz.palczewski@intel.com in bugliza.

Thanks,
Fan

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

Chen, Fan (farrah.chen@intel.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |high

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

RE: [Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[Chen, Farrah]

Add mateusz.palczewski@intel.com and mateusz.palczewski@intel.com

-----Original Message-----
From: bugzilla-daemon@kernel.org <bugzilla-daemon@kernel.org> 
Sent: Wednesday, June 21, 2023 9:26 PM
To: kvm@vger.kernel.org
Subject: [Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #3 from Chen, Fan (farrah.chen@intel.com) --- We bisect and found the first bad commit, I don't understand if this commit is intend to do so? Make the mac of VF in VM different from it in host? We think they use the same mac is better for users, with the same mac, the mac of VF is known for us before creating VM, then we can get IP address of the VF interface in VM from mac without using vnc or other UI or serial port. How can we make the mac of VF keep the same in host and VM now?

commit ceb29474bbbc377e11be3a70589a0005305e4fc3
Author: Sylwester Dziedziuch <sylwesterx.dziedziuch@intel.com>
Date:   Thu Mar 30 10:00:22 2023 -0700

    i40e: Add support for VF to specify its primary MAC address

    Currently in the i40e driver there is no implementation of different
    MAC address handling depending on whether it is a legacy or primary.
    Introduce new checks for VF to be able to specify its primary MAC
    address based on the VIRTCHNL_ETHER_ADDR_PRIMARY type.

    Primary MAC address are treated differently compared to legacy
    ones in a scenario where:
    1. If a unicast MAC is being added and it's specified as
    VIRTCHNL_ETHER_ADDR_PRIMARY, then replace the current
    default_lan_addr.addr.
    2. If a unicast MAC is being deleted and it's type
    is specified as VIRTCHNL_ETHER_ADDR_PRIMARY, then zero the
    hw_lan_addr.addr.

Actually, we also tried to use "ip link set ens28f0 vf 0 mac <mac>" to fix a mac for VF, it works in host, but we create VM with this VF passthrough, it has a high probability failed to assign the mac to VF in vm, and the interface failed yo get IP either, error dmesg in VM: 
[    3.037955] iavf 0000:00:04.0: Invalid MAC address 00:00:00:00:00:00, using
random
[    3.039523] iavf 0000:00:04.0: Multiqueue Enabled: Queue pair count = 4
[    3.040466] iavf 0000:00:04.0: MAC address: 3a:38:ca:66:f9:65
[    3.040980] iavf 0000:00:04.0: GRO is enabled
[    3.042240] iavf 0000:00:05.0: Multiqueue Enabled: Queue pair count = 4
[    3.043232] iavf 0000:00:05.0: MAC address: 36:91:96:9d:5d:05
[    3.043770] iavf 0000:00:05.0: GRO is enabled
[    3.044401] iavf 0000:00:04.0 ens4: renamed from eth0
[    3.049199] iavf 0000:00:05.0 ens5: renamed from eth1
[    3.072576] ppdev: user-space parallel port driver
[    3.094077] XFS (vda2): Mounting V5 Filesystem
8736b23e-ddde-4cca-9166-2623d2e57e5a
[    3.102245] XFS (vda2): Ending clean mount
[    3.550242] iavf 0000:00:04.0: Failed to add MAC filter, error IAVF_ERR_NVM
[    3.617176] iavf 0000:00:04.0 ens4: NIC Link is Up Speed is 10 Gbps Full
Duplex
[    3.618173] IPv6: ADDRCONF(NETDEV_CHANGE): ens4: link becomes ready
[    3.628189] iavf 0000:00:05.0 ens5: NIC Link is Up Speed is 10 Gbps Full
Duplex
[    3.629039] IPv6: ADDRCONF(NETDEV_CHANGE): ens5: link becomes ready

Fix this issue will be very helpful for us.

I failed to CC the primary author sylwesterx.dziedziuch@intel.com and mateusz.palczewski@intel.com in bugliza.

Thanks,
Fan

--
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #4 from Chen, Fan (farrah.chen@intel.com) ---
Add mateusz.palczewski@intel.com and mateusz.palczewski@intel.com

-----Original Message-----
From: bugzilla-daemon@kernel.org <bugzilla-daemon@kernel.org> 
Sent: Wednesday, June 21, 2023 9:26 PM
To: kvm@vger.kernel.org
Subject: [Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac
address of VF is inconsistent with it in host

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #3 from Chen, Fan (farrah.chen@intel.com) --- We bisect and found
the first bad commit, I don't understand if this commit is intend to do so?
Make the mac of VF in VM different from it in host? We think they use the same
mac is better for users, with the same mac, the mac of VF is known for us
before creating VM, then we can get IP address of the VF interface in VM from
mac without using vnc or other UI or serial port. How can we make the mac of VF
keep the same in host and VM now?

commit ceb29474bbbc377e11be3a70589a0005305e4fc3
Author: Sylwester Dziedziuch <sylwesterx.dziedziuch@intel.com>
Date:   Thu Mar 30 10:00:22 2023 -0700

    i40e: Add support for VF to specify its primary MAC address

    Currently in the i40e driver there is no implementation of different
    MAC address handling depending on whether it is a legacy or primary.
    Introduce new checks for VF to be able to specify its primary MAC
    address based on the VIRTCHNL_ETHER_ADDR_PRIMARY type.

    Primary MAC address are treated differently compared to legacy
    ones in a scenario where:
    1. If a unicast MAC is being added and it's specified as
    VIRTCHNL_ETHER_ADDR_PRIMARY, then replace the current
    default_lan_addr.addr.
    2. If a unicast MAC is being deleted and it's type
    is specified as VIRTCHNL_ETHER_ADDR_PRIMARY, then zero the
    hw_lan_addr.addr.

Actually, we also tried to use "ip link set ens28f0 vf 0 mac <mac>" to fix a
mac for VF, it works in host, but we create VM with this VF passthrough, it has
a high probability failed to assign the mac to VF in vm, and the interface
failed yo get IP either, error dmesg in VM: 
[    3.037955] iavf 0000:00:04.0: Invalid MAC address 00:00:00:00:00:00, using
random
[    3.039523] iavf 0000:00:04.0: Multiqueue Enabled: Queue pair count = 4
[    3.040466] iavf 0000:00:04.0: MAC address: 3a:38:ca:66:f9:65
[    3.040980] iavf 0000:00:04.0: GRO is enabled
[    3.042240] iavf 0000:00:05.0: Multiqueue Enabled: Queue pair count = 4
[    3.043232] iavf 0000:00:05.0: MAC address: 36:91:96:9d:5d:05
[    3.043770] iavf 0000:00:05.0: GRO is enabled
[    3.044401] iavf 0000:00:04.0 ens4: renamed from eth0
[    3.049199] iavf 0000:00:05.0 ens5: renamed from eth1
[    3.072576] ppdev: user-space parallel port driver
[    3.094077] XFS (vda2): Mounting V5 Filesystem
8736b23e-ddde-4cca-9166-2623d2e57e5a
[    3.102245] XFS (vda2): Ending clean mount
[    3.550242] iavf 0000:00:04.0: Failed to add MAC filter, error IAVF_ERR_NVM
[    3.617176] iavf 0000:00:04.0 ens4: NIC Link is Up Speed is 10 Gbps Full
Duplex
[    3.618173] IPv6: ADDRCONF(NETDEV_CHANGE): ens4: link becomes ready
[    3.628189] iavf 0000:00:05.0 ens5: NIC Link is Up Speed is 10 Gbps Full
Duplex
[    3.629039] IPv6: ADDRCONF(NETDEV_CHANGE): ens5: link becomes ready

Fix this issue will be very helpful for us.

I failed to CC the primary author sylwesterx.dziedziuch@intel.com and
mateusz.palczewski@intel.com in bugliza.

Thanks,
Fan

--
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

Patryk Piotrowski (patryk.piotrowski@intel.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |patryk.piotrowski@intel.com

--- Comment #5 from Patryk Piotrowski (patryk.piotrowski@intel.com) ---
Thank you very much for reporting this
AFAIR the patch from Sylwester was response to one of the critical issue
We will take a look at this issue and see what can be done here

Many thanks and regards,
Patryk Piotrowski
Intel Technology Poland

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #6 from Patryk Piotrowski (patryk.piotrowski@intel.com) ---
Hello Fan,
Could you please try to reproduce the issue with the newest net next Kernel?
I'm asking due to reason that the issue is not present there on our setups

Thank you,
Patryk Piotrowski
Intel Technology Poland

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #7 from Chen, Fan (farrah.chen@intel.com) ---
Hello Patryk

We can reproduce this issue on the latest commit in next branch of
https://git.kernel.org/pub/scm/virt/kvm/kvm.git, actually we struggled with
this problem for a long time, no fix found in this repo, so we bisect and found
the first commit.
Do you mean we can try the latest commit in master branch of
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git?

Thanks
Fan

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #8 from Chen, Fan (farrah.chen@intel.com) ---
Hi Patryk

We reproduced this issue with the latest kernel commit
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
a92b7d26 kernel version: 6.4.0_rc7

When you reproduce, did you create multiple VFs and passthrough all of them to
KVM guest? I found that if we only passthrough one VF to guest, the mac is the
same as it in host, but when passthrough 2 or more VFs, sometimes only the
first VF use the same mac as it in host, other VFs' mac are random; sometimes
all the VFs' mac are random.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

Chen, Fan (farrah.chen@intel.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Kernel Version|6.4.0-rc2                   |6.4.0-rc7

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

Chen, Fan (farrah.chen@intel.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|kvm                         |Network
            Version|unspecified                 |2.5
            Product|Virtualization              |Drivers

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

Radoslaw Tyl (radoslawx.tyl@intel.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |radoslawx.tyl@intel.com

--- Comment #9 from Radoslaw Tyl (radoslawx.tyl@intel.com) ---
Created attachment 304564
  --> https://bugzilla.kernel.org/attachment.cgi?id=304564&action=edit
[Patch]Fix inconsistent mac address of VF

Proposal fix inconsistent mac address of VF after remove interface on VM.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #10 from Radoslaw Tyl (radoslawx.tyl@intel.com) ---
Hello Fan,

I reproduced issue on the latest commit in dev-queue branch
https://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue.git/log/drivers/net/ethernet/intel/iavf?h=dev-queue

Could you please try to reproduce the issue with this patch in Attachments
added to above version kernel ?

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #11 from Chen, Fan (farrah.chen@intel.com) ---
Thanks Radoslaw, looks like your patch can fix this issue, I used above kernel
with this patch, tried 5 times to reproduce, can't reproduce, mac of VF is the
same in host and guest, got expected result.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #12 from Patryk Piotrowski (patryk.piotrowski@intel.com) ---
Hello,
The patch is already present on dev queue and it's waiting for "tested by" tag
IWL link:
https://patchwork.ozlabs.org/project/intel-wired-lan/patch/20230807125940.985494-1-radoslawx.tyl@intel.com/

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #13 from Chen, Fan (farrah.chen@intel.com) ---
Hi Patryk,

Do I need to add "tested by"? I cannot find the email.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #14 from Patryk Piotrowski (patryk.piotrowski@intel.com) ---
No, no - it's awaiting for "tested by" tag which is going to be provided by our
VAL team once they verify the patch
Once we will have this tag we will proceed with upstream and apply the patch to
proper trees. 
I will keep you informed

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #15 from Chen, Fan (farrah.chen@intel.com) ---
Hi Patryk,

Got it, thanks, could you help to add me as "Reported by"?

Thanks
Fan

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

--- Comment #16 from Chen, Fan (farrah.chen@intel.com) ---
Fixed with
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c8de44b577eb540e8bfea55afe1d0904bb571b7a

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 217558] In KVM guest with VF of X710 NIC passthrough, the mac address of VF is inconsistent with it in host

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=217558

Chen, Fan (farrah.chen@intel.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |CODE_FIX

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.