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, 80 insertions, 0 deletions

diff --git a/arch/powerpc/boot/hack-coff.c b/arch/powerpc/boot/hack-coff.c
new file mode 100644
index 000000000..6bf0c9430
--- /dev/null
+++ b/arch/powerpc/boot/hack-coff.c
@@ -0,0 +1,80 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * hack-coff.c - hack the header of an xcoff file to fill in
+ * a few fields needed by the Open Firmware xcoff loader on
+ * Power Macs but not initialized by objcopy.
+ *
+ * Copyright (C) Paul Mackerras 1997.
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <string.h>
+#include "rs6000.h"
+
+#define AOUT_MAGIC	0x010b
+
+#define get_16be(x)	((((unsigned char *)(x))[0] << 8) \
+			 + ((unsigned char *)(x))[1])
+#define put_16be(x, v)	(((unsigned char *)(x))[0] = (v) >> 8, \
+			 ((unsigned char *)(x))[1] = (v) & 0xff)
+#define get_32be(x)	((((unsigned char *)(x))[0] << 24) \
+			 + (((unsigned char *)(x))[1] << 16) \
+			 + (((unsigned char *)(x))[2] << 8) \
+			 + ((unsigned char *)(x))[3])
+
+int
+main(int ac, char **av)
+{
+    int fd;
+    int i, nsect;
+    int aoutsz;
+    struct external_filehdr fhdr;
+    struct aouthdr aout;
+    struct external_scnhdr shdr;
+
+    if (ac != 2) {
+	fprintf(stderr, "Usage: hack-coff coff-file\n");
+	exit(1);
+    }
+    if ((fd = open(av[1], 2)) == -1) {
+	perror(av[2]);
+	exit(1);
+    }
+    if (read(fd, &fhdr, sizeof(fhdr)) != sizeof(fhdr))
+	goto readerr;
+    i = get_16be(fhdr.f_magic);
+    if (i != U802TOCMAGIC && i != U802WRMAGIC && i != U802ROMAGIC) {
+	fprintf(stderr, "%s: not an xcoff file\n", av[1]);
+	exit(1);
+    }
+    aoutsz = get_16be(fhdr.f_opthdr);
+    if (read(fd, &aout, aoutsz) != aoutsz)
+	goto readerr;
+    nsect = get_16be(fhdr.f_nscns);
+    for (i = 0; i < nsect; ++i) {
+	if (read(fd, &shdr, sizeof(shdr)) != sizeof(shdr))
+	    goto readerr;
+	if (strcmp(shdr.s_name, ".text") == 0) {
+	    put_16be(aout.o_snentry, i+1);
+	    put_16be(aout.o_sntext, i+1);
+	} else if (strcmp(shdr.s_name, ".data") == 0) {
+	    put_16be(aout.o_sndata, i+1);
+	} else if (strcmp(shdr.s_name, ".bss") == 0) {
+	    put_16be(aout.o_snbss, i+1);
+	}
+    }
+    put_16be(aout.magic, AOUT_MAGIC);
+    if (lseek(fd, (long) sizeof(struct external_filehdr), 0) == -1
+	|| write(fd, &aout, aoutsz) != aoutsz) {
+	fprintf(stderr, "%s: write error\n", av[1]);
+	exit(1);
+    }
+    close(fd);
+    exit(0);
+
+readerr:
+    fprintf(stderr, "%s: read error or file too short\n", av[1]);
+    exit(1);
+}