[libvirt PATCH v2 0/5] qemu: support stable NIC device naming

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



With this series applied it is possible to get stable NIC device naming
in combination with QEMU >= 6.0.0 (release imminent), by setting the
ACPI device index against NICs.

    <acpi index="NNN"/>

The index is an unique integer that can be assigned to any PCI/PCIe
device. systemd uses this value for its stable NIC naming scheme.
If not present, then it falls back to an SMBIOS device index, but
QEMU doesn't support that concept.

Note that although this is only known to be useful for NICs, the
ACPI feature is conceptually available for any PCI device, and so
libvirt doesn't restrict the usage to only NICs.

Tested with current QEMU git master, and Fedora 30 guest. It works
with i440fx, but fails with Q35. The latter problem is reported to
qemu-devel and awaiting response.

<interface type="network">
      <mac address="52:54:00:c3:d3:ef" />
      <source network="default" portid="c2361e1b-3ebf-484e-813e-9ee8e2b760c7" bridge="virbr0" />
      <target dev="vnet6" />
      <model type="virtio" />
      <alias name="net0" />
      <acpi index="103" />
      <address type="pci" domain="0x0000" bus="0x00" slot="0x03" function="0x0" />
    </interface>
<interface type="network">
      <mac address="52:54:00:98:22:a6" />
      <source network="default" portid="e6cdb3da-ca24-42fa-973e-8cdb92fcad3f" bridge="virbr0" />
      <target dev="vnet7" />
      <model type="virtio" />
      <alias name="net1" />
      <acpi index="113" />
      <address type="pci" domain="0x0000" bus="0x00" slot="0x0a" function="0x0" />
    </interface>
<interface type="network">
      <mac address="52:54:00:42:00:bc" />
      <source network="default" portid="10efeab4-f59c-4ef7-90a3-32b3895080b7" bridge="virbr0" />
      <target dev="vnet8" />
      <model type="virtio" />
      <alias name="net2" />
      <acpi index="123" />
      <address type="pci" domain="0x0000" bus="0x00" slot="0x0b" function="0x0" />
    </interface>
<interface type="network">
      <mac address="52:54:00:36:8e:5a" />
      <source network="default" portid="fb016189-d764-41b2-b256-f5783ade3726" bridge="virbr0" />
      <target dev="vnet9" />
      <model type="virtio" />
      <alias name="net3" />
      <acpi index="133" />
      <address type="pci" domain="0x0000" bus="0x00" slot="0x0c" function="0x0" />
    </interface>

 Name       MAC address          Protocol     Address
-------------------------------------------------------------------------------
 vnet6      52:54:00:c3:d3:ef    ipv4         192.168.122.61/24
 vnet7      52:54:00:98:22:a6    ipv4         192.168.122.69/24
 vnet8      52:54:00:42:00:bc    ipv4         192.168.122.17/24
 vnet9      52:54:00:36:8e:5a    ipv4         192.168.122.13/24

$ ip -4 addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
2: eno103: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
    inet 192.168.122.61/24 brd 192.168.122.255 scope global dynamic noprefixroute eno103
       valid_lft 2987sec preferred_lft 2987sec
3: eno113: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
    inet 192.168.122.69/24 brd 192.168.122.255 scope global dynamic noprefixroute eno113
       valid_lft 2987sec preferred_lft 2987sec
4: eno123: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
    inet 192.168.122.17/24 brd 192.168.122.255 scope global dynamic noprefixroute eno123
       valid_lft 2987sec preferred_lft 2987sec
5: eno133: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
    inet 192.168.122.13/24 brd 192.168.122.255 scope global dynamic noprefixroute eno133
       valid_lft 2988sec preferred_lft 2988sec

Changed in v2:

 - Fix typos
 - Document limits / uniqueness
 - Drop caps refresh
 - Add xml2xml test

Daniel P. Berrangé (5):
  conf: add support for <acpi index='NNN'/> for PCI devices
  qemu: fix indentation off-by-1
  qemu: use a switch when building device addresses
  qemu: probe for "acpi-index" property
  qemu: wire up command line support for ACPI index

 docs/formatdomain.rst                         |  7 ++
 docs/schemas/domaincommon.rng                 | 73 +++++++++++++++++++
 src/conf/device_conf.h                        |  3 +
 src/conf/domain_conf.c                        | 19 +++++
 src/qemu/qemu_capabilities.c                  |  6 ++
 src/qemu/qemu_capabilities.h                  |  1 +
 src/qemu/qemu_command.c                       | 42 +++++++++--
 src/qemu/qemu_validate.c                      | 35 +++++++++
 .../caps_6.0.0.x86_64.xml                     |  1 +
 .../devices-acpi-index.x86_64-latest.args     | 57 +++++++++++++++
 tests/qemuxml2argvdata/devices-acpi-index.xml | 62 ++++++++++++++++
 tests/qemuxml2argvtest.c                      |  2 +
 .../devices-acpi-index.x86_64-latest.xml      | 73 +++++++++++++++++++
 tests/qemuxml2xmltest.c                       |  2 +
 14 files changed, 376 insertions(+), 7 deletions(-)
 create mode 100644 tests/qemuxml2argvdata/devices-acpi-index.x86_64-latest.args
 create mode 100644 tests/qemuxml2argvdata/devices-acpi-index.xml
 create mode 100644 tests/qemuxml2xmloutdata/devices-acpi-index.x86_64-latest.xml

-- 
2.30.2





[Index of Archives]     [Virt Tools]     [Libvirt Users]     [Lib OS Info]     [Fedora Users]     [Fedora Desktop]     [Fedora SELinux]     [Big List of Linux Books]     [Yosemite News]     [KDE Users]     [Fedora Tools]

  Powered by Linux