From 0d6526f82c3cdefcca47f73f5fc08dc6f335eac6 Mon Sep 17 00:00:00 2001 From: Tim Chen Date: Mon, 21 Sep 2026 17:37:22 -0700 Subject: sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling bug alb_break_llc() decides whether to break LLC preference during active load balance. It does so by testing that every runnable fair task on the source rq prefers its LLC: env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable But the two counters cover different sets. nr_pref_llc_running is updated in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued, so it follows queued tasks. h_nr_runnable is updated in set_delayed()/ clear_delayed() and drops delay-dequeued tasks. So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks, alb_break_llc() returns false, and active balance is free to pull a task off its preferred LLC. Active balance only moves runnable tasks, and this is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB skips the per-task test in can_migrate_task(). The runnable set is the one we want. Fix it on the counter side. A task should be counted in nr_pref_llc_running exactly while it is both queued on its preferred LLC (pref_llc_queued) and runnable (!sched_delayed). Define that membership once in task_pref_llc_runnable(), and adjust the counter only through pref_llc_running_inc()/pref_llc_running_dec() from the four sites that change either input: account_llc_enqueue(), account_llc_dequeue(), set_delayed() and clear_delayed(). Gating every update on the same predicate keeps the delay, wake and dequeue paths from double-counting or underflowing; see the comments at those sites for the ordering. nr_llc_running and sd->llc_counts are not touched and stay on queued semantics. Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC") Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/ Reported-by: Zhan Xusheng Suggested-by: Chen Yu Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Kayra Cizmeci Cc: # v7.2.x Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com --- Documentation/ABI/testing/debugfs-scmi | 70 ++++++++++++++++++++++++++++++++++ 1 file changed, 70 insertions(+) create mode 100644 Documentation/ABI/testing/debugfs-scmi (limited to 'Documentation/ABI/testing/debugfs-scmi') diff --git a/Documentation/ABI/testing/debugfs-scmi b/Documentation/ABI/testing/debugfs-scmi new file mode 100644 index 000000000..ee7179ab2 --- /dev/null +++ b/Documentation/ABI/testing/debugfs-scmi @@ -0,0 +1,70 @@ +What: /sys/kernel/debug/scmi//instance_name +Date: March 2023 +KernelVersion: 6.3 +Contact: cristian.marussi@arm.com +Description: The name of the underlying SCMI instance described by + all the debugfs accessors rooted at /sys/kernel/debug/scmi/, + expressed as the full name of the top DT SCMI node under which + this SCMI instance is rooted. +Users: Debugging, any userspace test suite + +What: /sys/kernel/debug/scmi//atomic_threshold_us +Date: March 2023 +KernelVersion: 6.3 +Contact: cristian.marussi@arm.com +Description: An optional time value, expressed in microseconds, representing, + on this SCMI instance , the threshold above which any SCMI + command, advertised to have an higher-than-threshold execution + latency, should not be considered for atomic mode of operation, + even if requested. +Users: Debugging, any userspace test suite + +What: /sys/kernel/debug/scmi//transport/type +Date: March 2023 +KernelVersion: 6.3 +Contact: cristian.marussi@arm.com +Description: A string representing the type of transport configured for this + SCMI instance . +Users: Debugging, any userspace test suite + +What: /sys/kernel/debug/scmi//transport/is_atomic +Date: March 2023 +KernelVersion: 6.3 +Contact: cristian.marussi@arm.com +Description: A boolean stating if the transport configured on the underlying + SCMI instance is capable of atomic mode of operation. +Users: Debugging, any userspace test suite + +What: /sys/kernel/debug/scmi//transport/max_rx_timeout_ms +Date: March 2023 +KernelVersion: 6.3 +Contact: cristian.marussi@arm.com +Description: Timeout in milliseconds allowed for SCMI synchronous replies + for the currently configured SCMI transport for instance . +Users: Debugging, any userspace test suite + +What: /sys/kernel/debug/scmi//transport/max_msg_size +Date: March 2023 +KernelVersion: 6.3 +Contact: cristian.marussi@arm.com +Description: Max message size of allowed SCMI messages for the currently + configured SCMI transport for instance . +Users: Debugging, any userspace test suite + +What: /sys/kernel/debug/scmi//transport/tx_max_msg +Date: March 2023 +KernelVersion: 6.3 +Contact: cristian.marussi@arm.com +Description: Max number of concurrently allowed in-flight SCMI messages for + the currently configured SCMI transport for instance on the + TX channels. +Users: Debugging, any userspace test suite + +What: /sys/kernel/debug/scmi//transport/rx_max_msg +Date: March 2023 +KernelVersion: 6.3 +Contact: cristian.marussi@arm.com +Description: Max number of concurrently allowed in-flight SCMI messages for + the currently configured SCMI transport for instance on the + RX channels. +Users: Debugging, any userspace test suite -- cgit v1.3.1