File _patchinfo of Package patchinfo.42378

<patchinfo incident="42378">
  <issue tracker="bnc" id="1198323">Migration of google-guest-configs ignores custom modification in gce-blacklist.conf</issue>
  <issue tracker="bnc" id="1256906">L3: google-guest-configs update to 20260112.00</issue>
  <packager>glaubitz</packager>
  <rating>important</rating>
  <category>recommended</category>
  <summary>Recommended update for google-guest-configs</summary>
  <description>This update for google-guest-configs fixes the following issues:

- Update to version 20260116.00 (bsc#1256906)
    * set_multiqueue: Only set XPS on "multinic accelerator platforms"
- Update to version 20260112.00
    * Make c4x a "multinic accelerator platform"
    * set_multiqueue xps: stop assuming 2 numa nodes
    * Add IDPF irq setting; improve a4x-max performance
    * Allow test injection of the root directory and metadata server endpoint
    * add nic naming support for connextx VF in baremetal
    * bugfix for idpf only rename got skipped.
    * add a4x-max to google_set_multiqueue is_multinic_accelerator_platform
    * remove unnecessary link up and down
    * fix inconsistent NIC index between smart NICs and GPU NICs.
- Mark %{_modprobedir}/gce-blacklist.conf as %config(noreplace) (bsc#1198323)
- Update to version 20251014.00
- Update to version 20250913.00
    * Swap guest-config rule from checking the build VM OS to taking in a variable for target version
- from version 20250826.00
    * Moved tx/rx IRQ logging after assignment
    * Fix core assignment in set_irq_range
    * Correct IRQ tx/rx affinity core assignment
- Update to version 20250807.00
    * Avoid duplicate entries for the metadata server in /etc/hosts
- Update to version 20250709.00
    * Add comments in scripts to document the behavior in google hostname setting.
    * Always use primary NIC IP for NetworkManager dispatcher hook.
- from version 20250626.00
    * Fix spelling error: "explicilty" to "explicitly"
- Update to version 20250605.00
    * Added comment to the bitmap conversion functions
    * Remove IRQ affinity overwrite to XPS affinity
    * Update XPS affinity to assign the remaining unassigned CPUs
      to the last queue when populating the last queue
    * Fix set_xps_affinity to correctly parse cpus array
    * Update XPS CPU assignment logic
    * Update CPU assignment algorithm in XPS affinity
    * Remove commented code
    * Update XPS affinity vCPU distribution algorithm s.t. the vCPUs assigned
      to a queue are on the same core - fixed IRQ affinity on NUMA1 not using
      the correct bind_cores_index
    * Fixed NUMA comparison error in set_xps_affinity
    * Update XPS affinity setup to be NUMA aware and support 64 bit CPU mask calculation
- from version 20250604.00
    * Bug fix: bind_cores_begin to bind_cores_index
    * Name smart NICs in lexicographic order
- Run %postun to modify %{_sysconfdir}/sysconfig/network/ifcfg-eth0
  during uninstall only to avoid removal of POST_UP_SCRIPT on upgrade
- Update to version 20250516.00
    * Remove unused fset
    * Remove unused lines
    * Update google_set_multiqueue to unpack IRQ ranges before core assignment
- Update to version 20250501.00
    * Configure local domain as route only domain to support cloud dns local
      domain but avoid adding it to the search path.
- from version 20250409.00
    * Change RDMA test condition to ensure renaming race conditions can be detected.
- from version 20250328.00
    * Revert "Include systemd-networkd hook in Ubuntu packaging"
- from version 20250326.00
    * Update google_set_multiqueue to check pnic_ids
- from version 20250221.00
    * Make google_set_multiqueue aware A4X is multinic_accelerator_platform
- from version 20250207.00
    * Update google_set_multiqueue to adapt A4 platform
    * Merge branch 'GoogleCloudPlatform:master' into master
    * Fix IS_A3_PLATFORM syntax
    * Correct IS_A3_PLATFORM to save is_a3_platform results
    * Remove excess empty line.
    * Store is_a3_platform results into a global variable to avoid redundant curl calls
    * Skip tx affinity binding on non-gvnic interfaces only on A3 platforms.
    * Update comments for get_vcpu_ranges_on_accelerator_platform to reflect the expected vcpu ranges
    * rename get_vcpu_ranges to get_vcpu_ranges_on_accelerator_platform
    * Avoid IRQ binding on vCPU 0
    * Fix returned value for get_vcpu_ranges
    * Update get_vcpu_ranges to read from sys file instead of hardcoded value
    * Update google_set_multiqueue to set vCPU ranges based on platform
    * Merge branch 'GoogleCloudPlatform:master' into master
    * Add comment for handling IRQ binding on non-gvnic devices
    * Update is_gvnic to include gvnic driver checks
    * revert removed echo lines
    * Update google_set_multiqueue to skip set_irq if nic is not a gvnic device.
    * Update google_set_multiqueue to enable on A3Ultra family
- from version 20250124.00
    * Fix missing files. This is a no-op.
    * Also force virtio_scsi
- from version 20250116.00
    * Add GPL-2 to licensing information
- from version 20250107.00
    * Restore IDPF devices for renaming rules
- from version 20241213.00
    * Remove Pat from owners file
  </description>
</patchinfo>