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, 307 insertions, 0 deletions

diff --git a/drivers/acpi/acpica/utbuffer.c b/drivers/acpi/acpica/utbuffer.c
new file mode 100644
index 000000000..4193ceb93
--- /dev/null
+++ b/drivers/acpi/acpica/utbuffer.c
@@ -0,0 +1,307 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/******************************************************************************
+ *
+ * Module Name: utbuffer - Buffer dump routines
+ *
+ * Copyright (C) 2000 - 2026, Intel Corp.
+ *
+ *****************************************************************************/
+
+#include <acpi/acpi.h>
+#include "accommon.h"
+
+#define _COMPONENT          ACPI_UTILITIES
+ACPI_MODULE_NAME("utbuffer")
+
+/*******************************************************************************
+ *
+ * FUNCTION:    acpi_ut_dump_buffer
+ *
+ * PARAMETERS:  buffer              - Buffer to dump
+ *              count               - Amount to dump, in bytes
+ *              display             - BYTE, WORD, DWORD, or QWORD display:
+ *                                      DB_BYTE_DISPLAY
+ *                                      DB_WORD_DISPLAY
+ *                                      DB_DWORD_DISPLAY
+ *                                      DB_QWORD_DISPLAY
+ *              base_offset         - Beginning buffer offset (display only)
+ *
+ * RETURN:      None
+ *
+ * DESCRIPTION: Generic dump buffer in both hex and ascii.
+ *
+ ******************************************************************************/
+void acpi_ut_dump_buffer(u8 *buffer, u32 count, u32 display, u32 base_offset)
+{
+	u32 i = 0;
+	u32 j;
+	u32 temp32;
+	u8 buf_char;
+	u32 display_data_only = display & DB_DISPLAY_DATA_ONLY;
+
+	display &= ~DB_DISPLAY_DATA_ONLY;
+	if (!buffer) {
+		acpi_os_printf("Null Buffer Pointer in DumpBuffer!\n");
+		return;
+	}
+
+	if ((count < 4) || (count & 0x01)) {
+		display = DB_BYTE_DISPLAY;
+	}
+
+	/* Nasty little dump buffer routine! */
+
+	while (i < count) {
+
+		/* Print current offset */
+
+		if (!display_data_only) {
+			acpi_os_printf("%8.4X: ", (base_offset + i));
+		}
+
+		/* Print 16 hex chars */
+
+		for (j = 0; j < 16;) {
+			if (i + j >= count) {
+
+				/* Dump fill spaces */
+
+				acpi_os_printf("%*s", ((display * 2) + 1), " ");
+				j += display;
+				continue;
+			}
+
+			switch (display) {
+			case DB_BYTE_DISPLAY:
+			default:	/* Default is BYTE display */
+
+				acpi_os_printf("%02X ",
+					       buffer[(acpi_size)i + j]);
+				break;
+
+			case DB_WORD_DISPLAY:
+
+				ACPI_MOVE_16_TO_32(&temp32,
+						   &buffer[(acpi_size)i + j]);
+				acpi_os_printf("%04X ", temp32);
+				break;
+
+			case DB_DWORD_DISPLAY:
+
+				ACPI_MOVE_32_TO_32(&temp32,
+						   &buffer[(acpi_size)i + j]);
+				acpi_os_printf("%08X ", temp32);
+				break;
+
+			case DB_QWORD_DISPLAY:
+
+				ACPI_MOVE_32_TO_32(&temp32,
+						   &buffer[(acpi_size)i + j]);
+				acpi_os_printf("%08X", temp32);
+
+				ACPI_MOVE_32_TO_32(&temp32,
+						   &buffer[(acpi_size)i + j +
+							   4]);
+				acpi_os_printf("%08X ", temp32);
+				break;
+			}
+
+			j += display;
+		}
+
+		/*
+		 * Print the ASCII equivalent characters but watch out for the bad
+		 * unprintable ones (printable chars are 0x20 through 0x7E)
+		 */
+		if (!display_data_only) {
+			acpi_os_printf(" ");
+			for (j = 0; j < 16; j++) {
+				if (i + j >= count) {
+					acpi_os_printf("\n");
+					return;
+				}
+
+				/*
+				 * Add comment characters so rest of line is ignored when
+				 * compiled
+				 */
+				if (j == 0) {
+					acpi_os_printf("// ");
+				}
+
+				buf_char = buffer[(acpi_size)i + j];
+				if (isprint(buf_char)) {
+					acpi_os_printf("%c", buf_char);
+				} else {
+					acpi_os_printf(".");
+				}
+			}
+
+			/* Done with that line. */
+
+			acpi_os_printf("\n");
+		}
+		i += 16;
+	}
+
+	return;
+}
+
+/*******************************************************************************
+ *
+ * FUNCTION:    acpi_ut_debug_dump_buffer
+ *
+ * PARAMETERS:  buffer              - Buffer to dump
+ *              count               - Amount to dump, in bytes
+ *              display             - BYTE, WORD, DWORD, or QWORD display:
+ *                                      DB_BYTE_DISPLAY
+ *                                      DB_WORD_DISPLAY
+ *                                      DB_DWORD_DISPLAY
+ *                                      DB_QWORD_DISPLAY
+ *              component_ID        - Caller's component ID
+ *
+ * RETURN:      None
+ *
+ * DESCRIPTION: Generic dump buffer in both hex and ascii.
+ *
+ ******************************************************************************/
+
+void
+acpi_ut_debug_dump_buffer(u8 *buffer, u32 count, u32 display, u32 component_id)
+{
+
+	/* Only dump the buffer if tracing is enabled */
+
+	if (!((ACPI_LV_TABLES & acpi_dbg_level) &&
+	      (component_id & acpi_dbg_layer))) {
+		return;
+	}
+
+	acpi_ut_dump_buffer(buffer, count, display, 0);
+}
+
+#ifdef ACPI_APPLICATION
+/*******************************************************************************
+ *
+ * FUNCTION:    acpi_ut_dump_buffer_to_file
+ *
+ * PARAMETERS:  file                - File descriptor
+ *              buffer              - Buffer to dump
+ *              count               - Amount to dump, in bytes
+ *              display             - BYTE, WORD, DWORD, or QWORD display:
+ *                                      DB_BYTE_DISPLAY
+ *                                      DB_WORD_DISPLAY
+ *                                      DB_DWORD_DISPLAY
+ *                                      DB_QWORD_DISPLAY
+ *              base_offset         - Beginning buffer offset (display only)
+ *
+ * RETURN:      None
+ *
+ * DESCRIPTION: Generic dump buffer in both hex and ascii to a file.
+ *
+ ******************************************************************************/
+
+void
+acpi_ut_dump_buffer_to_file(ACPI_FILE file,
+			    u8 *buffer, u32 count, u32 display, u32 base_offset)
+{
+	u32 i = 0;
+	u32 j;
+	u32 temp32;
+	u8 buf_char;
+
+	if (!buffer) {
+		fprintf(file, "Null Buffer Pointer in DumpBuffer!\n");
+		return;
+	}
+
+	if ((count < 4) || (count & 0x01)) {
+		display = DB_BYTE_DISPLAY;
+	}
+
+	/* Nasty little dump buffer routine! */
+
+	while (i < count) {
+
+		/* Print current offset */
+
+		fprintf(file, "%8.4X: ", (base_offset + i));
+
+		/* Print 16 hex chars */
+
+		for (j = 0; j < 16;) {
+			if (i + j >= count) {
+
+				/* Dump fill spaces */
+
+				fprintf(file, "%*s", ((display * 2) + 1), " ");
+				j += display;
+				continue;
+			}
+
+			switch (display) {
+			case DB_BYTE_DISPLAY:
+			default:	/* Default is BYTE display */
+
+				fprintf(file, "%02X ",
+					buffer[(acpi_size)i + j]);
+				break;
+
+			case DB_WORD_DISPLAY:
+
+				ACPI_MOVE_16_TO_32(&temp32,
+						   &buffer[(acpi_size)i + j]);
+				fprintf(file, "%04X ", temp32);
+				break;
+
+			case DB_DWORD_DISPLAY:
+
+				ACPI_MOVE_32_TO_32(&temp32,
+						   &buffer[(acpi_size)i + j]);
+				fprintf(file, "%08X ", temp32);
+				break;
+
+			case DB_QWORD_DISPLAY:
+
+				ACPI_MOVE_32_TO_32(&temp32,
+						   &buffer[(acpi_size)i + j]);
+				fprintf(file, "%08X", temp32);
+
+				ACPI_MOVE_32_TO_32(&temp32,
+						   &buffer[(acpi_size)i + j +
+							   4]);
+				fprintf(file, "%08X ", temp32);
+				break;
+			}
+
+			j += display;
+		}
+
+		/*
+		 * Print the ASCII equivalent characters but watch out for the bad
+		 * unprintable ones (printable chars are 0x20 through 0x7E)
+		 */
+		fprintf(file, " ");
+		for (j = 0; j < 16; j++) {
+			if (i + j >= count) {
+				fprintf(file, "\n");
+				return;
+			}
+
+			buf_char = buffer[(acpi_size)i + j];
+			if (isprint(buf_char)) {
+				fprintf(file, "%c", buf_char);
+			} else {
+				fprintf(file, ".");
+			}
+		}
+
+		/* Done with that line. */
+
+		fprintf(file, "\n");
+		i += 16;
+	}
+
+	return;
+}
+#endif