Skip to contentSkip to conversation

sched_ext: Changes for v6.19

GIT PULL Merged Tejun Heo 2 participants 2 messages Dec 1, 2025 – Dec 3, 2025 Scheduler - sched_ext

lore
Share & tools
lore

Tejun Heo

author reviewer
#0 lore raw

The following changes since commit 81e839803f9355084bc78f9e0c350c4e2853c880:

Merge tag 'cgroup-for-6.19' into pr-base-for-6.19 (2025-12-01 13:20:56 -1000)

are available in the Git repository at:

https://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext.git tags/sched_ext-for-6.19

for you to fetch changes up to 1dd6c84f1c544e552848a8968599220bd464e338:

sched_ext: Fix incorrect sched_class settings for per-cpu migration tasks (2025-12-01 10:58:49 -1000)

Please note that sched_ext-for-6.19 is based on v6.18 + tip/sched-core-2025-12-01
+ cgroup/for-6.19.

When merging sched_ext-for-6.19 into the above base, there is a trivial merge conflict in kernel/sched/ext.c around pick_task_scx(). Both sched_ext and tip sched/core independently pulled in a8ad873113d3 ("sched_ext: defer queue_balance_callback() until after ops.dispatch") resolving conflicts slightly differently (whitespace only). 67fa319f5fff ("sched_ext: Allow forcibly picking an scx task") then added further changes in the same area. Resolve by taking the sched_ext side.

----------------------------------------------------------------
sched_ext: Changes for v6.19

- Improve recovery from misbehaving BPF schedulers. When a scheduler puts many tasks with varying affinity restrictions on a shared DSQ, CPUs scanning through tasks they cannot run can overwhelm the system, causing lockups. Bypass mode now uses per-CPU DSQs with a load balancer to avoid this, and hooks into the hardlockup detector to attempt recovery. Add scx_cpu0 example scheduler to demonstrate this scenario.

- Add lockless peek operation for DSQs to reduce lock contention for schedulers that need to query queue state during load balancing.

- Allow scx_bpf_reenqueue_local() to be called from anywhere in preparation for deprecating cpu_acquire/release() callbacks in favor of generic BPF hooks.

- Prepare for hierarchical scheduler support: add scx_bpf_task_set_slice() and scx_bpf_task_set_dsq_vtime() kfuncs, make scx_bpf_dsq_insert*() return bool, and wrap kfunc args in structs for future aux__prog parameter.

- Implement cgroup_set_idle() callback to notify BPF schedulers when a cgroup's idle state changes.

- Fix migration tasks being incorrectly downgraded from stop_sched_class to rt_sched_class across sched_ext enable/disable. Applied late as the fix is low risk and the bug subtle but needs stable backporting.

- Various fixes and cleanups including cgroup exit ordering, SCX_KICK_WAIT reliability, and backward compatibility improvements.

----------------------------------------------------------------
Andrea Righi (4):
sched_ext: Exit early on hotplug events during attach
sched_ext: Allow forcibly picking an scx task
sched_ext: Fix scx_bpf_dsq_peek() with FIFO DSQs
sched_ext: Update comments replacing breather with aborting mechanism

Rong Tao (1):
sched_ext: tools: Removing duplicate targets during non-cross compilation

Ryan Newton (2):
sched_ext: Add lockless peek operation for DSQs
sched_ext: Add a selftest for scx_bpf_dsq_peek

Tejun Heo (34):
tools/sched_ext: Strip compatibility macros for cgroup and dispatch APIs
sched_ext: Add scx_bpf_task_set_slice() and scx_bpf_task_set_dsq_vtime()
sched_ext: Wrap kfunc args in struct to prepare for aux__prog
sched_ext: Make scx_bpf_dsq_insert*() return bool
sched_ext/tools: Add compat wrapper for scx_bpf_task_set_slice/dsq_vtime()
sched_ext: Fix scx_bpf_dsq_insert() backward binary compatibility
sched_ext: Don't kick CPUs running higher classes
sched_ext: Fix SCX_KICK_WAIT to work reliably
sched_ext: Rename pnt_seq to kick_sync
sched_ext: Use rhashtable_lookup() instead of rhashtable_lookup_fast()
sched_ext: Add ___compat suffix to scx_bpf_dsq_insert___v2 in compat.bpf.h
sched_ext: Use SCX_TASK_READY test instead of tryget_task_struct() during class switch
sched_ext: Split schedule_deferred() into locked and unlocked variants
sched_ext: Factor out reenq_local() from scx_bpf_reenqueue_local()
sched_ext: Allow scx_bpf_reenqueue_local() to be called from anywhere sched_ext/tools: Restore backward compat with v6.12 kernels
sched_ext: Fix cgroup exit ordering by moving sched_ext_free() to finish_task_switch()
sched_ext: Move __SCX_DSQ_ITER_ALL_FLAGS BUILD_BUG_ON to the right place
sched_ext: Minor cleanups to scx_task_iter
sched_ext: Mark racy bitfields to prevent adding fields that can't tolerate races
sched_ext: Use shorter slice in bypass mode
sched_ext: Refactor do_enqueue_task() local and global DSQ paths
sched_ext: Use per-CPU DSQs instead of per-node global DSQs in bypass mode
sched_ext: Simplify breather mechanism with scx_aborting flag
sched_ext: Exit dispatch and move operations immediately when aborting
sched_ext: Make scx_exit() and scx_vexit() return bool
sched_ext: Refactor lockup handlers into handle_lockup()
sched_ext: Make handle_lockup() propagate scx_verror() result
sched_ext: Hook up hardlockup detector
sched_ext: Add scx_cpu0 example scheduler
sched_ext: Factor out scx_dsq_list_node cursor initialization into INIT_DSQ_LIST_CURSOR
sched_ext: Factor out abbreviated dispatch dequeue into dispatch_dequeue_locked()
sched_ext: Implement load balancer for bypass mode
sched_ext: Pass locked CPU parameter to scx_hardlockup() and add docs

Zqiang (2):
sched_ext: Use kvfree_rcu() to release per-cpu ksyncs object
sched_ext: Fix incorrect sched_class settings for per-cpu migration tasks

zhidao su (1):
sched/ext: Implement cgroup_set_idle() callback

 include/linux/sched/ext.h                        |   27 +-
 include/trace/events/sched_ext.h                 |   39 +
 kernel/fork.c                                    |    1 -
 kernel/sched/core.c                              |    6 +
 kernel/sched/ext.c                               | 1068 ++++++++++++++++------
 kernel/sched/ext_idle.c                          |   43 +-
 kernel/sched/ext_internal.h                      |   29 +-
 kernel/sched/sched.h                             |    4 +-
 kernel/watchdog.c                                |    9 +
 tools/sched_ext/Makefile                         |    4 +-
 tools/sched_ext/include/scx/common.bpf.h         |   15 +-
 tools/sched_ext/include/scx/compat.bpf.h         |  314 ++++--
 tools/sched_ext/include/scx/compat.h             |   14 +
 tools/sched_ext/scx_cpu0.bpf.c                   |   88 ++
 tools/sched_ext/scx_cpu0.c                       |  106 +++
 tools/sched_ext/scx_flatcg.bpf.c                 |   10 +-
 tools/sched_ext/scx_qmap.bpf.c                   |   52 +-
 tools/testing/selftests/sched_ext/Makefile       |    1 +
 tools/testing/selftests/sched_ext/peek_dsq.bpf.c |  251 +++++
 tools/testing/selftests/sched_ext/peek_dsq.c     |  224 +++++
 20 files changed, 1894 insertions(+), 411 deletions(-)
 create mode 100644 tools/sched_ext/scx_cpu0.bpf.c
 create mode 100644 tools/sched_ext/scx_cpu0.c
 create mode 100644 tools/testing/selftests/sched_ext/peek_dsq.bpf.c
 create mode 100644 tools/testing/selftests/sched_ext/peek_dsq.c

Thanks.

Signature · 1 line
tejun
merge bot Merged
#1 lore raw
The pull request you sent on Mon, 01 Dec 2025 13:48:47 -1000:

The pull request you sent on Mon, 01 Dec 2025 13:48:47 -1000:

Quoted 1 line

https://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext.git tags/sched_ext-for-6.19

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/02baaa67d9afc2e56c6e1ac6a1fb1f1dd2be366f

Thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/prtracker.html