[PATCH v4 0/4] sched: Simplify CONFIG_SCHED_SMT ifdef usage

[Shrikanth Hegde <sshegde@linux.ibm.com>]

Semantics
=========
- For CONFIG_SCHED_SMT=y:
    No functional change.
- For CONFIG_SCHED_SMT=n:
    - cpu_smt_mask(cpu) becomes cpumask_of(cpu), effectively making it
      per CPU with no siblings.
    - sched_smt_present remains defined, but never becomes active
      Since cpumask_weight(cpumask_of(cpu)) == 1

Performance impact
==================
- CONFIG_SCHED_SMT=y:
    No change in generated code.
- CONFIG_SCHED_SMT=n:
    - Small increase in text size (~0.01%) due to removal of compile-time
      stubs. Most paths remain effectively dead due to static keys.
    - Fast paths are protected using IS_ENABLED(CONFIG_SCHED_SMT).

With that, cpu_smt_mask() can be used unconditionally and reduces
CONFIG_SCHED_SMT ifdef usage, improving readability and maintainability.
This leaves the remaining use of CONFIG_SCHED_SMT mainly for topology
handling bits.

Also, remove the explicit use of sched_smt_present. Make use of
sched_smt_active. It differs slightly for update_idle_core, which
used with unlikely flag earlier. Now it will be likely. Should be ok,
since most systems have SMT active today.

Since v3[3]:
- Minor re-order of sched_smt_active to bail out early for !smt
  (Valentin Schneider)
- Collected tags. Thanks again.

Since v2[2]:
- Changed patch 3 significantly to use sched_smt_active check for
  fastpath. (K Prateek Nayak)
- Collected the tags.
  Thanks to Phil Auld, Valentin Schneider, Tejun Heo, K Prateek Nayak.
- Since patch 3 has changed by quite a bit, i have not applied the tags.
  Please review it again. Sorry for the inconvenience.
- Added patch 4 to unify checking on SMT active using
  sched_smt_active().

Since v1[1]:
- Dropped changes for stop_core_cpuslocked. intel_ifs is the only
  user and it gets compiled with SCHED_SMT always. Added comments
  around it why ifdefs are still being kept.
- Decided not to add sched_smt_active checks for sched_core_cpu_deactivate
  and sched_core_cpu_starting, they bail out if weight(smt_mask) == 1.
  Hence core_lock won't be held for long. It is not fastpath either.
- Thanks to Valentin Schneider and Tejun Heo for checking out the series
  and the feedback.

[1]: https://lore.kernel.org/all/20260506110052.9974-1-sshegde@linux.ibm.com/
[2]: https://lore.kernel.org/all/20260512152125.308280-1-sshegde@linux.ibm.com/
[3]: https://lore.kernel.org/all/20260513133934.380347-1-sshegde@linux.ibm.com/

Shrikanth Hegde (4):
  topology: Introduce cpu_smt_mask for CONFIG_SCHED_SMT=n
  sched: Simplify ifdeffery around cpu_smt_mask
  sched/fair: Add sched_smt_active check for fastpaths
  sched: Unify SMT active check via sched_smt_active()

 include/linux/sched/smt.h |  4 ----
 include/linux/topology.h  | 15 ++++++++++++-
 kernel/sched/core.c       |  6 -----
 kernel/sched/core_sched.c |  2 +-
 kernel/sched/ext_idle.c   |  6 -----
 kernel/sched/fair.c       | 46 +++++----------------------------------
 kernel/sched/sched.h      |  8 +------
 kernel/sched/topology.c   |  2 --
 kernel/stop_machine.c     |  5 +++++
 kernel/workqueue.c        |  4 ----
 10 files changed, 26 insertions(+), 72 deletions(-)

-- 
2.47.3