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, 146 insertions, 0 deletions

diff --git a/tools/perf/util/copyfile.c b/tools/perf/util/copyfile.c
new file mode 100644
index 000000000..47e03de7c
--- /dev/null
+++ b/tools/perf/util/copyfile.c
@@ -0,0 +1,146 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "util/copyfile.h"
+#include "util/namespaces.h"
+#include <internal/lib.h>
+#include <sys/mman.h>
+#include <sys/stat.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+static int slow_copyfile(const char *from, const char *to, struct nsinfo *nsi)
+{
+	int err = -1;
+	char *line = NULL;
+	size_t n;
+	FILE *from_fp, *to_fp;
+	struct nscookie nsc;
+
+	nsinfo__mountns_enter(nsi, &nsc);
+	from_fp = fopen(from, "r");
+	nsinfo__mountns_exit(&nsc);
+	if (from_fp == NULL)
+		goto out;
+
+	to_fp = fopen(to, "w");
+	if (to_fp == NULL)
+		goto out_fclose_from;
+
+	while (getline(&line, &n, from_fp) > 0)
+		if (fputs(line, to_fp) == EOF)
+			goto out_fclose_to;
+	err = 0;
+out_fclose_to:
+	fclose(to_fp);
+	free(line);
+out_fclose_from:
+	fclose(from_fp);
+out:
+	return err;
+}
+
+int copyfile_offset(int ifd, loff_t off_in, int ofd, loff_t off_out, u64 size)
+{
+	void *ptr;
+	loff_t pgoff;
+
+	pgoff = off_in & ~(page_size - 1);
+	off_in -= pgoff;
+
+	ptr = mmap(NULL, off_in + size, PROT_READ, MAP_PRIVATE, ifd, pgoff);
+	if (ptr == MAP_FAILED)
+		return -1;
+
+	while (size) {
+		ssize_t ret = pwrite(ofd, ptr + off_in, size, off_out);
+		if (ret < 0 && errno == EINTR)
+			continue;
+		if (ret <= 0)
+			break;
+
+		size -= ret;
+		off_in += ret;
+		off_out += ret;
+	}
+	munmap(ptr, off_in + size);
+
+	return size ? -1 : 0;
+}
+
+static int copyfile_mode_ns(const char *from, const char *to, mode_t mode,
+			    struct nsinfo *nsi)
+{
+	int fromfd, tofd;
+	struct stat st;
+	int err;
+	char *tmp = NULL, *ptr = NULL;
+	struct nscookie nsc;
+
+	nsinfo__mountns_enter(nsi, &nsc);
+	err = stat(from, &st);
+	nsinfo__mountns_exit(&nsc);
+	if (err)
+		goto out;
+	err = -1;
+
+	/* extra 'x' at the end is to reserve space for '.' */
+	if (asprintf(&tmp, "%s.XXXXXXx", to) < 0) {
+		tmp = NULL;
+		goto out;
+	}
+	ptr = strrchr(tmp, '/');
+	if (!ptr)
+		goto out;
+	ptr = memmove(ptr + 1, ptr, strlen(ptr) - 1);
+	*ptr = '.';
+
+	tofd = mkstemp(tmp);
+	if (tofd < 0)
+		goto out;
+
+	if (st.st_size == 0) { /* /proc? do it slowly... */
+		err = slow_copyfile(from, tmp, nsi);
+		if (!err && fchmod(tofd, mode))
+			err = -1;
+		goto out_close_to;
+	}
+
+	if (fchmod(tofd, mode))
+		goto out_close_to;
+
+	nsinfo__mountns_enter(nsi, &nsc);
+	fromfd = open(from, O_RDONLY);
+	nsinfo__mountns_exit(&nsc);
+	if (fromfd < 0)
+		goto out_close_to;
+
+	err = copyfile_offset(fromfd, 0, tofd, 0, st.st_size);
+
+	close(fromfd);
+out_close_to:
+	close(tofd);
+	if (!err)
+		err = link(tmp, to);
+	unlink(tmp);
+out:
+	free(tmp);
+	return err;
+}
+
+int copyfile_ns(const char *from, const char *to, struct nsinfo *nsi)
+{
+	return copyfile_mode_ns(from, to, 0755, nsi);
+}
+
+int copyfile_mode(const char *from, const char *to, mode_t mode)
+{
+	return copyfile_mode_ns(from, to, mode, NULL);
+}
+
+int copyfile(const char *from, const char *to)
+{
+	return copyfile_mode(from, to, 0755);
+}