sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling buggrafted

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 <zhanxusheng@xiaomi.com>
Suggested-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Kayra Cizmeci <kayracizmeci@gmail.com>
Cc: <stable@kernel.org> # v7.2.x
Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com

1 files changed, 48 insertions, 0 deletions

diff --git a/arch/xtensa/Kconfig.debug b/arch/xtensa/Kconfig.debug
new file mode 100644
index 000000000..e84172a77
--- /dev/null
+++ b/arch/xtensa/Kconfig.debug
@@ -0,0 +1,48 @@
+# SPDX-License-Identifier: GPL-2.0
+
+config DEBUG_TLB_SANITY
+	bool "Debug TLB sanity"
+	depends on DEBUG_KERNEL && MMU
+	help
+	  Enable this to turn on TLB sanity check on each entry to userspace.
+	  This check can spot missing TLB invalidation/wrong PTE permissions/
+	  premature page freeing.
+
+	  If unsure, say N.
+
+config LD_NO_RELAX
+	bool "Disable linker relaxation"
+	default y
+	help
+	  Enable this function to disable link-time optimizations.
+	  The default linker behavior is to combine identical literal
+	  values to reduce code size and remove unnecessary overhead from
+	  assembler-generated 'longcall' sequences.
+	  Enabling this option improves the link time but increases the
+	  code size, and possibly execution time.
+
+config S32C1I_SELFTEST
+	bool "Perform S32C1I instruction self-test at boot"
+	default y
+	help
+	  Enable this option to test S32C1I instruction behavior at boot.
+	  Correct operation of this instruction requires some cooperation from hardware
+	  external to the processor (such as bus bridge, bus fabric, or memory controller).
+	  It is easy to make wrong hardware configuration, this test should catch it early.
+
+	  Say 'N' on stable hardware.
+
+config PRINT_STACK_DEPTH
+	int "Stack depth to print" if DEBUG_KERNEL
+	default 64
+	help
+	  This option allows you to set the stack depth that the kernel
+	  prints in stack traces.
+
+config PRINT_USER_CODE_ON_UNHANDLED_EXCEPTION
+	bool "Dump user code around unhandled exception address"
+	help
+	  Enable this option to display user code around PC of the unhandled
+	  exception (starting at address aligned on 16 byte boundary).
+	  This may simplify finding faulting code in the absence of other
+	  debug facilities.