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, 251 insertions, 0 deletions

diff --git a/tools/testing/vsock/control.c b/tools/testing/vsock/control.c
new file mode 100644
index 000000000..0066e0324
--- /dev/null
+++ b/tools/testing/vsock/control.c
@@ -0,0 +1,251 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Control socket for client/server test execution
+ *
+ * Copyright (C) 2017 Red Hat, Inc.
+ *
+ * Author: Stefan Hajnoczi <stefanha@redhat.com>
+ */
+
+/* The client and server may need to coordinate to avoid race conditions like
+ * the client attempting to connect to a socket that the server is not
+ * listening on yet.  The control socket offers a communications channel for
+ * such coordination tasks.
+ *
+ * If the client calls control_expectln("LISTENING"), then it will block until
+ * the server calls control_writeln("LISTENING").  This provides a simple
+ * mechanism for coordinating between the client and the server.
+ */
+
+#include <errno.h>
+#include <netdb.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/types.h>
+#include <sys/socket.h>
+
+#include "timeout.h"
+#include "control.h"
+#include "util.h"
+
+static int control_fd = -1;
+
+/* Open the control socket, either in server or client mode */
+void control_init(const char *control_host,
+		  const char *control_port,
+		  bool server)
+{
+	struct addrinfo hints = {
+		.ai_socktype = SOCK_STREAM,
+	};
+	struct addrinfo *result = NULL;
+	struct addrinfo *ai;
+	int ret;
+
+	ret = getaddrinfo(control_host, control_port, &hints, &result);
+	if (ret != 0) {
+		fprintf(stderr, "%s\n", gai_strerror(ret));
+		exit(EXIT_FAILURE);
+	}
+
+	for (ai = result; ai; ai = ai->ai_next) {
+		int fd;
+
+		fd = socket(ai->ai_family, ai->ai_socktype, ai->ai_protocol);
+		if (fd < 0)
+			continue;
+
+		if (!server) {
+			if (connect(fd, ai->ai_addr, ai->ai_addrlen) < 0)
+				goto next;
+			control_fd = fd;
+			printf("Control socket connected to %s:%s.\n",
+			       control_host, control_port);
+			break;
+		}
+
+		setsockopt_int_check(fd, SOL_SOCKET, SO_REUSEADDR, 1,
+				     "setsockopt SO_REUSEADDR");
+
+		if (bind(fd, ai->ai_addr, ai->ai_addrlen) < 0)
+			goto next;
+		if (listen(fd, 1) < 0)
+			goto next;
+
+		printf("Control socket listening on %s:%s\n",
+		       control_host, control_port);
+		fflush(stdout);
+
+		control_fd = accept(fd, NULL, 0);
+		close(fd);
+
+		if (control_fd < 0) {
+			perror("accept");
+			exit(EXIT_FAILURE);
+		}
+		printf("Control socket connection accepted...\n");
+		break;
+
+next:
+		close(fd);
+	}
+
+	if (control_fd < 0) {
+		fprintf(stderr, "Control socket initialization failed.  Invalid address %s:%s?\n",
+			control_host, control_port);
+		exit(EXIT_FAILURE);
+	}
+
+	freeaddrinfo(result);
+}
+
+/* Free resources */
+void control_cleanup(void)
+{
+	close(control_fd);
+	control_fd = -1;
+}
+
+/* Write a line to the control socket */
+void control_writeln(const char *str)
+{
+	ssize_t len = strlen(str);
+	ssize_t ret;
+
+	timeout_begin(TIMEOUT);
+
+	do {
+		ret = send(control_fd, str, len, MSG_MORE);
+		timeout_check("send");
+	} while (ret < 0 && errno == EINTR);
+
+	if (ret != len) {
+		perror("send");
+		exit(EXIT_FAILURE);
+	}
+
+	do {
+		ret = send(control_fd, "\n", 1, 0);
+		timeout_check("send");
+	} while (ret < 0 && errno == EINTR);
+
+	if (ret != 1) {
+		perror("send");
+		exit(EXIT_FAILURE);
+	}
+
+	timeout_end();
+}
+
+void control_writeulong(unsigned long value)
+{
+	char str[32];
+
+	if (snprintf(str, sizeof(str), "%lu", value) >= sizeof(str)) {
+		perror("snprintf");
+		exit(EXIT_FAILURE);
+	}
+
+	control_writeln(str);
+}
+
+unsigned long control_readulong(void)
+{
+	unsigned long value;
+	char *str;
+
+	str = control_readln();
+
+	if (!str)
+		exit(EXIT_FAILURE);
+
+	value = strtoul(str, NULL, 10);
+	free(str);
+
+	return value;
+}
+
+/* Return the next line from the control socket (without the trailing newline).
+ *
+ * The program terminates if a timeout occurs.
+ *
+ * The caller must free() the returned string.
+ */
+char *control_readln(void)
+{
+	char *buf = NULL;
+	size_t idx = 0;
+	size_t buflen = 0;
+
+	timeout_begin(TIMEOUT);
+
+	for (;;) {
+		ssize_t ret;
+
+		if (idx >= buflen) {
+			char *new_buf;
+
+			new_buf = realloc(buf, buflen + 80);
+			if (!new_buf) {
+				perror("realloc");
+				exit(EXIT_FAILURE);
+			}
+
+			buf = new_buf;
+			buflen += 80;
+		}
+
+		do {
+			ret = recv(control_fd, &buf[idx], 1, 0);
+			timeout_check("recv");
+		} while (ret < 0 && errno == EINTR);
+
+		if (ret == 0) {
+			fprintf(stderr, "unexpected EOF on control socket\n");
+			exit(EXIT_FAILURE);
+		}
+
+		if (ret != 1) {
+			perror("recv");
+			exit(EXIT_FAILURE);
+		}
+
+		if (buf[idx] == '\n') {
+			buf[idx] = '\0';
+			break;
+		}
+
+		idx++;
+	}
+
+	timeout_end();
+
+	return buf;
+}
+
+/* Wait until a given line is received or a timeout occurs */
+void control_expectln(const char *str)
+{
+	char *line;
+
+	line = control_readln();
+
+	control_cmpln(line, str, true);
+
+	free(line);
+}
+
+bool control_cmpln(char *line, const char *str, bool fail)
+{
+	if (strcmp(str, line) == 0)
+		return true;
+
+	if (fail) {
+		fprintf(stderr, "expected \"%s\" on control socket, got \"%s\"\n",
+			str, line);
+		exit(EXIT_FAILURE);
+	}
+
+	return false;
+}