ACK: [SRU][N][PATCH v2 0/1] UBUNTU: SAUCE: Fix cpuset behavior to v1 following the changes from upstream stable patches.

[Edoardo Canepa]

Acked-by: Edoardo Canepa <edoardo.canepa at canonical.com>

On 12/19/25 15:28, Mehmet Basaran wrote:
> BugLink: https://bugs.launchpad.net/bugs/2128722
>
> SRU Justification:
>
> [Impact]
>
> For noble , some of the applied upstream commits are using "CONFIG_CPUSETS_V1"
> to enable/disable some code paths. This changes the behaviour of cpusets and
> currently noble is neither V1 nor V2. Because of this, "cpuset_inherit" and
> "cpuset_hotplug" test cases are failing in ubuntu_ltp_controllers test suite.
> For noble, since this is an LTS release we should continue behave like V1.
>
> cpuset_inherit 37 TFAIL: memory_migrate: Test result - 1 Expected string - "0"
> cpuset_inherit 39 TFAIL: memory_migrate: Test result - 1 Expected string - "0"
> cpuset_inherit 49 TFAIL: sched_load_balance: Test result - 0 Expected string - "1"
>
> cpuset_hotplug 1 TINFO: CPUs are numbered continuously starting at 0 (0-1)
> cpuset_hotplug 1 TINFO: Nodes are numbered continuously starting at 0 (0)
> cpuset_hotplug 1 TFAIL: root group's cpus isn't expected(Result: 0-1, Expect: 0).
> cpuset_hotplug 5 TFAIL: general group's cpus isn't expected(Result: 0-1, Expect: 0).
> cpuset_hotplug 7 TFAIL: test task was still in general group, but its cpus is NULL
> cpuset_hotplug 11 TFAIL: task's cpu present list isn't expected(Result: 0-1, Expect: 0).
>
> [Fix]
>
> For noble, we have to use v1 because there is no support for v2. Added
> "CONFIG_CPUSETS_V1" kernel config option. This was added to upstream via
> the following commit:
>
>      [ Upstream commit 1abab1ba0775036bb67c6c57945c637be644c04f ]
>      cgroup/cpuset: guard cpuset-v1 code under CONFIG_CPUSETS_V1
>
> I have chosen to go with a SAUCE patch since, the commit above depends on
> other commits. And in the future, we can revert this one and apply the
> dependent commits if we want to follow cpusets follow from 6.12 versions so
> that there are less conflicts between upstream stable and noble sources.
>
> The other solution is to remove the use of "CONFIG_CPUSETS_V1" macro from
> sources. However this will also cause noble and upstream stable sources to
> diverge and make it difficult to backport patches.
>
> [Test Plan]
>
> Rerun cpuset_* tests in ubuntu_ltp_controllers test suite. All passed.
>
> [Where problems could occur]
>
> cpusets can behave like neither v1 nor v2 which can lead to incorrect
> allocation of scheduling domains for cgroups.
>
> As long as cpuset code paths are separated correctly for v1 and v2 in the
> upstream stable branches, backported commits should work as well. So,
> essentially, this is a safe fix.
>
>
> Mehmet Basaran (1):
>    UBUNTU: SAUCE: Enable upstream config option "CONFIG_CPUSETS_V1"
>
>   debian.master/config/annotations |  1 +
>   init/Kconfig                     | 13 +++++++++++++
>   2 files changed, 14 insertions(+)
>
> --
> 2.43.0
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_0x20F88172E14F6784.asc
Type: application/pgp-keys
Size: 3167 bytes
Desc: OpenPGP public key
URL: <https://lists.ubuntu.com/archives/kernel-team/attachments/20251219/84652201/attachment-0001.key>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_signature.asc
Type: application/pgp-signature
Size: 840 bytes
Desc: OpenPGP digital signature
URL: <https://lists.ubuntu.com/archives/kernel-team/attachments/20251219/84652201/attachment-0001.sig>