ftrace: Synchronize the initialization of ftrace_opsgrafted

There's some internal state that ftrace_ops needs to have set, but since
it can be declared outside of the ftrace.c code, it calls
ftrace_ops_init() on the ops in every global function. The issue is that
if two tasks call it on the same ops at the same time it is possible to
have the initialization of one corrupt the initialization of the other
call.

Create a ops_mutex to use to synchronize every initialization of the
ftrace_ops. The mutex is taken within checking the ftrace_ops flag that
states it was initializied but the flag is checked again after the mutex
has been taken. Checking first outside the mutex allows it to shortcut
having to take the mutex. But then the check needs to be done again after
the mute is taken in case of races.

Cc: stable@vger.kernel.org
Link: https://patch.msgid.link/20260902095501.6b59af20@gandalf.local.home
Fixes: f04f24fb7e48d ("ftrace, kprobes: Fix a deadlock on ftrace_regex_lock")
Reported-by: sashiko-bot@kernel.org
Close: https://lore.kernel.org/all/20260829025528.49A831F000E9@smtp.kernel.org/
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

1 files changed, 139 insertions, 0 deletions

diff --git a/tools/perf/bench/find-bit-bench.c b/tools/perf/bench/find-bit-bench.c
new file mode 100644
index 000000000..e697c2095
--- /dev/null
+++ b/tools/perf/bench/find-bit-bench.c
@@ -0,0 +1,139 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Benchmark find_next_bit and related bit operations.
+ *
+ * Copyright 2020 Google LLC.
+ */
+#include <stdlib.h>
+#include "bench.h"
+#include "../util/stat.h"
+#include <linux/bitmap.h>
+#include <linux/bitops.h>
+#include <linux/time64.h>
+#include <subcmd/parse-options.h>
+
+static unsigned int outer_iterations = 5;
+static unsigned int inner_iterations = 100000;
+
+static const struct option options[] = {
+	OPT_UINTEGER('i', "outer-iterations", &outer_iterations,
+		"Number of outer iterations used"),
+	OPT_UINTEGER('j', "inner-iterations", &inner_iterations,
+		"Number of inner iterations used"),
+	OPT_END()
+};
+
+static const char *const bench_usage[] = {
+	"perf bench mem find_bit <options>",
+	NULL
+};
+
+static unsigned int accumulator;
+static unsigned int use_of_val;
+
+static noinline void workload(int val)
+{
+	use_of_val += val;
+	accumulator++;
+}
+
+#if defined(__i386__) || defined(__x86_64__)
+static bool asm_test_bit(long nr, const unsigned long *addr)
+{
+	bool oldbit;
+
+	asm volatile("bt %2,%1"
+		     : "=@ccc" (oldbit)
+		     : "m" (*(unsigned long *)addr), "Ir" (nr) : "memory");
+
+	return oldbit;
+}
+#else
+#define asm_test_bit test_bit
+#endif
+
+static int do_for_each_set_bit(unsigned int num_bits)
+{
+	unsigned long *to_test = bitmap_zalloc(num_bits);
+	struct timeval start, end, diff;
+	u64 runtime_us;
+	struct stats fb_time_stats, tb_time_stats;
+	double time_average, time_stddev;
+	unsigned int bit, i, j;
+	unsigned int set_bits, skip;
+
+	init_stats(&fb_time_stats);
+	init_stats(&tb_time_stats);
+
+	for (set_bits = 1; set_bits <= num_bits; set_bits <<= 1) {
+		bitmap_zero(to_test, num_bits);
+		skip = num_bits / set_bits;
+		for (i = 0; i < num_bits; i += skip)
+			__set_bit(i, to_test);
+
+		for (i = 0; i < outer_iterations; i++) {
+#ifndef NDEBUG
+			unsigned int old = accumulator;
+#endif
+
+			gettimeofday(&start, NULL);
+			for (j = 0; j < inner_iterations; j++) {
+				for_each_set_bit(bit, to_test, num_bits)
+					workload(bit);
+			}
+			gettimeofday(&end, NULL);
+			assert(old + (inner_iterations * set_bits) == accumulator);
+			timersub(&end, &start, &diff);
+			runtime_us = diff.tv_sec * USEC_PER_SEC + diff.tv_usec;
+			update_stats(&fb_time_stats, runtime_us);
+
+#ifndef NDEBUG
+			old = accumulator;
+#endif
+			gettimeofday(&start, NULL);
+			for (j = 0; j < inner_iterations; j++) {
+				for (bit = 0; bit < num_bits; bit++) {
+					if (asm_test_bit(bit, to_test))
+						workload(bit);
+				}
+			}
+			gettimeofday(&end, NULL);
+			assert(old + (inner_iterations * set_bits) == accumulator);
+			timersub(&end, &start, &diff);
+			runtime_us = diff.tv_sec * USEC_PER_SEC + diff.tv_usec;
+			update_stats(&tb_time_stats, runtime_us);
+		}
+
+		printf("%d operations %d bits set of %d bits\n",
+			inner_iterations, set_bits, num_bits);
+		time_average = avg_stats(&fb_time_stats);
+		time_stddev = stddev_stats(&fb_time_stats);
+		printf("  Average for_each_set_bit took: %.3f usec (+- %.3f usec)\n",
+			time_average, time_stddev);
+		time_average = avg_stats(&tb_time_stats);
+		time_stddev = stddev_stats(&tb_time_stats);
+		printf("  Average test_bit loop took:    %.3f usec (+- %.3f usec)\n",
+			time_average, time_stddev);
+
+		if (use_of_val == accumulator)  /* Try to avoid compiler tricks. */
+			printf("\n");
+	}
+	bitmap_free(to_test);
+	return 0;
+}
+
+int bench_mem_find_bit(int argc, const char **argv)
+{
+	int err = 0, i;
+
+	argc = parse_options(argc, argv, options, bench_usage, 0);
+	if (argc) {
+		usage_with_options(bench_usage, options);
+		exit(EXIT_FAILURE);
+	}
+
+	for (i = 1; i <= 2048; i <<= 1)
+		do_for_each_set_bit(i);
+
+	return err;
+}