Tejun Heo 61debc251c sched_ext: Use per-CPU DSQs instead of per-node global DSQs in bypass mode ... Bypass mode routes tasks through fallback dispatch queues. Originally a single global DSQ, b7b3b2dbae ("sched_ext: Split the global DSQ per NUMA node") changed this to per-node DSQs to resolve NUMA-related livelocks. Dan Schatzberg found per-node DSQs can still livelock when many threads are pinned to different small CPU subsets: each CPU must scan many incompatible tasks to find runnable ones, causing severe contention with high CPU counts. Switch to per-CPU bypass DSQs. Each task queues on its current CPU. Default idle CPU selection and direct dispatch handle most cases well. This introduces a failure mode when tasks concentrate on one CPU in over-saturated systems. If the BPF scheduler severely skews placement before triggering bypass, that CPU's queue may be too long to drain, causing RCU stalls. A load balancer in a future patch will address this. The bypass DSQ is separate from local DSQ to enable load balancing: local DSQs use rq locks, preventing efficient scanning and transfer across CPUs, especially problematic when systems are already contended. v2: Clarified why bypass DSQ is separate from local DSQ (Andrea Righi). Reported-by: Dan Schatzberg <schatzberg.dan@gmail.com> Reviewed-by: Dan Schatzberg <schatzberg.dan@gmail.com> Reviewed-by: Andrea Righi <arighi@nvidia.com> Reviewed-by: Emil Tsalapatis <emil@etsalapatis.com> Signed-off-by: Tejun Heo <tj@kernel.org>		2025-11-12 06:43:44 -10:00
..
affinity.h
autogroup.h
clock.h
cond_resched.h
coredump.h
cpufreq.h
cputime.h
deadline.h
debug.h
ext.h	sched_ext: Use per-CPU DSQs instead of per-node global DSQs in bypass mode	2025-11-12 06:43:44 -10:00
hotplug.h
idle.h
init.h
isolation.h
jobctl.h
loadavg.h
mm.h
nohz.h
numa_balancing.h
posix-timers.h
prio.h
rseq_api.h
rt.h
sd_flags.h
signal.h
smt.h
stat.h
sysctl.h
task_flags.h
task_stack.h
task.h	Patch series in this pull request:	2025-10-02 18:44:54 -07:00
thread_info_api.h
topology.h
types.h
user.h
vhost_task.h
wake_q.h
xacct.h