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/admin-guide/sysctl/debug.rst | 52 ++++++++++++++++++++++++++++++ 1 file changed, 52 insertions(+) create mode 100644 Documentation/admin-guide/sysctl/debug.rst (limited to 'Documentation/admin-guide/sysctl/debug.rst') diff --git a/Documentation/admin-guide/sysctl/debug.rst b/Documentation/admin-guide/sysctl/debug.rst new file mode 100644 index 000000000..506bd5e48 --- /dev/null +++ b/Documentation/admin-guide/sysctl/debug.rst @@ -0,0 +1,52 @@ +================ +/proc/sys/debug/ +================ + +These files show up in ``/proc/sys/debug/``, depending on the +kernel configuration: + +.. contents:: :local: + +exception-trace +=============== + +This flag controls whether the kernel prints information about unhandled +signals (like segmentation faults) to the kernel log (``dmesg``). + +- ``0``: Unhandled signals are not traced. +- ``1``: Information about unhandled signals is printed. + +The default value is ``1`` on most architectures (like x86, MIPS, RISC-V), +but it is ``0`` on **arm64**. + +The actual information printed and the context provided varies +significantly depending on the CPU architecture. For example: + +- On **x86**, it typically prints the instruction pointer (IP), error + code, and address that caused a page fault. +- On **PowerPC**, it may print the next instruction pointer (NIP), + link register (LR), and other relevant registers. + +When enabled, this feature is often rate-limited to prevent the kernel +log from being flooded during a crash loop. + +kprobes-optimization +==================== + +This flag enables or disables the optimization of Kprobes on certain +architectures (like x86). + +- ``0``: Kprobes optimization is turned off. +- ``1``: Kprobes optimization is turned on (default). + +For more details on Kprobes and its optimization, please refer to +Documentation/trace/kprobes.rst. + +Copyright (c) 2026, Shubham Chakraborty + +For general info and legal blurb, please look in +Documentation/admin-guide/sysctl/index.rst. + +.. See scripts/check-sysctl-docs to keep this up to date: +.. scripts/check-sysctl-docs -vtable="debug" \ +.. $(git grep -l register_sysctl_) -- cgit v1.3.1