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, 87 insertions, 0 deletions

diff --git a/arch/powerpc/boot/simpleboot.c b/arch/powerpc/boot/simpleboot.c
new file mode 100644
index 000000000..27591df41
--- /dev/null
+++ b/arch/powerpc/boot/simpleboot.c
@@ -0,0 +1,87 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * The simple platform -- for booting when firmware doesn't supply a device
+ *                        tree or any platform configuration information.
+ *                        All data is extracted from an embedded device tree
+ *                        blob.
+ *
+ * Authors: Scott Wood <scottwood@freescale.com>
+ *          Grant Likely <grant.likely@secretlab.ca>
+ *
+ * Copyright (c) 2007 Freescale Semiconductor, Inc.
+ * Copyright (c) 2008 Secret Lab Technologies Ltd.
+ */
+
+#include "ops.h"
+#include "types.h"
+#include "io.h"
+#include "stdio.h"
+#include <libfdt.h>
+
+BSS_STACK(4*1024);
+
+extern int platform_specific_init(void) __attribute__((weak));
+
+void platform_init(unsigned long r3, unsigned long r4, unsigned long r5,
+		   unsigned long r6, unsigned long r7)
+{
+	const u32 *na, *ns, *reg, *timebase;
+	u64 memsize64;
+	int node, size, i;
+
+	/* Make sure FDT blob is sane */
+	if (fdt_check_header(_dtb_start) != 0)
+		fatal("Invalid device tree blob\n");
+
+	/* Find the #address-cells and #size-cells properties */
+	node = fdt_path_offset(_dtb_start, "/");
+	if (node < 0)
+		fatal("Cannot find root node\n");
+	na = fdt_getprop(_dtb_start, node, "#address-cells", &size);
+	if (!na || (size != 4))
+		fatal("Cannot find #address-cells property");
+	ns = fdt_getprop(_dtb_start, node, "#size-cells", &size);
+	if (!ns || (size != 4))
+		fatal("Cannot find #size-cells property");
+
+	/* Find the memory range */
+	node = fdt_node_offset_by_prop_value(_dtb_start, -1, "device_type",
+					     "memory", sizeof("memory"));
+	if (node < 0)
+		fatal("Cannot find memory node\n");
+	reg = fdt_getprop(_dtb_start, node, "reg", &size);
+	if (size < (*na+*ns) * sizeof(u32))
+		fatal("cannot get memory range\n");
+
+	/* Only interested in memory based at 0 */
+	for (i = 0; i < *na; i++)
+		if (*reg++ != 0)
+			fatal("Memory range is not based at address 0\n");
+
+	/* get the memsize and truncate it to under 4G on 32 bit machines */
+	memsize64 = 0;
+	for (i = 0; i < *ns; i++)
+		memsize64 = (memsize64 << 32) | *reg++;
+	if (sizeof(void *) == 4 && memsize64 >= 0x100000000ULL)
+		memsize64 = 0xffffffff;
+
+	/* finally, setup the timebase */
+	node = fdt_node_offset_by_prop_value(_dtb_start, -1, "device_type",
+					     "cpu", sizeof("cpu"));
+	if (node < 0)
+		fatal("Cannot find cpu node\n");
+	timebase = fdt_getprop(_dtb_start, node, "timebase-frequency", &size);
+	if (timebase && (size == 4))
+		timebase_period_ns = 1000000000 / *timebase;
+
+	/* Now we have the memory size; initialize the heap */
+	simple_alloc_init(_end, memsize64 - (unsigned long)_end, 32, 64);
+
+	/* prepare the device tree and find the console */
+	fdt_init(_dtb_start);
+
+	if (platform_specific_init)
+		platform_specific_init();
+
+	serial_console_init();
+}