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, 163 insertions, 0 deletions

diff --git a/tools/perf/tests/openat-syscall-tp-fields.c b/tools/perf/tests/openat-syscall-tp-fields.c
new file mode 100644
index 000000000..5365889d3
--- /dev/null
+++ b/tools/perf/tests/openat-syscall-tp-fields.c
@@ -0,0 +1,163 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <stdbool.h>
+#include <linux/err.h>
+#include <linux/string.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include "evlist.h"
+#include "evsel.h"
+#include "thread_map.h"
+#include "record.h"
+#include "tests.h"
+#include "debug.h"
+#include "util/mmap.h"
+#include <errno.h>
+#include <perf/mmap.h>
+#include "util/sample.h"
+
+#ifndef O_DIRECTORY
+#define O_DIRECTORY    00200000
+#endif
+#ifndef AT_FDCWD
+#define AT_FDCWD       -100
+#endif
+
+static int test__syscall_openat_tp_fields(struct test_suite *test __maybe_unused,
+					  int subtest __maybe_unused)
+{
+	struct record_opts opts = {
+		.target = {
+			.uses_mmap = true,
+		},
+		.no_buffering = true,
+		.freq	      = 1,
+		.mmap_pages   = 256,
+		.raw_samples  = true,
+	};
+	const char *filename = "/etc/passwd";
+	int flags = O_RDONLY | O_DIRECTORY;
+	struct evlist *evlist = evlist__new();
+	struct evsel *evsel;
+	int ret = TEST_FAIL, err, i, nr_events = 0, nr_polls = 0;
+	char sbuf[STRERR_BUFSIZE];
+
+	if (evlist == NULL) {
+		pr_debug("%s: evlist__new\n", __func__);
+		goto out;
+	}
+
+	evsel = evsel__newtp("syscalls", "sys_enter_openat");
+	if (IS_ERR(evsel)) {
+		pr_debug("%s: evsel__newtp\n", __func__);
+		ret = PTR_ERR(evsel) == -EACCES ? TEST_SKIP : TEST_FAIL;
+		goto out_put_evlist;
+	}
+
+	evlist__add(evlist, evsel);
+
+	err = evlist__create_maps(evlist, &opts.target);
+	if (err < 0) {
+		pr_debug("%s: evlist__create_maps\n", __func__);
+		goto out_put_evlist;
+	}
+
+	evsel__config(evsel, &opts, NULL);
+
+	perf_thread_map__set_pid(evlist__core(evlist)->threads, 0, getpid());
+
+	err = evlist__open(evlist);
+	if (err < 0) {
+		pr_debug("perf_evlist__open: %s\n",
+			 str_error_r(errno, sbuf, sizeof(sbuf)));
+		goto out_put_evlist;
+	}
+
+	err = evlist__do_mmap(evlist, UINT_MAX);
+	if (err < 0) {
+		pr_debug("evlist__mmap: %s\n",
+			 str_error_r(errno, sbuf, sizeof(sbuf)));
+		goto out_put_evlist;
+	}
+
+	evlist__enable(evlist);
+
+	/*
+	 * Generate the event:
+	 */
+	openat(AT_FDCWD, filename, flags);
+
+	while (1) {
+		int before = nr_events;
+
+		for (i = 0; i < evlist__core(evlist)->nr_mmaps; i++) {
+			union perf_event *event;
+			struct mmap *md;
+
+			md = &evlist__mmap(evlist)[i];
+			if (perf_mmap__read_init(&md->core) < 0)
+				continue;
+
+			while ((event = perf_mmap__read_event(&md->core)) != NULL) {
+				const u32 type = event->header.type;
+				int tp_flags;
+				struct perf_sample sample;
+
+				++nr_events;
+
+				if (type != PERF_RECORD_SAMPLE) {
+					perf_mmap__consume(&md->core);
+					continue;
+				}
+
+				perf_sample__init(&sample, /*all=*/false);
+				err = evsel__parse_sample(evsel, event, &sample);
+				if (err) {
+					pr_debug("Can't parse sample, err = %d\n", err);
+					perf_sample__exit(&sample);
+					goto out_put_evlist;
+				}
+
+				tp_flags = perf_sample__intval(&sample, "flags");
+				perf_sample__exit(&sample);
+				/* C library wrapper may set additional flags,
+				   access mode must be unchanged */
+				if ((tp_flags & O_ACCMODE) != (flags & O_ACCMODE) ||
+				    (tp_flags & flags) != flags) {
+					pr_debug("%s: Expected flags=%#x, got %#x\n",
+						 __func__, flags, tp_flags);
+					goto out_put_evlist;
+				}
+
+				goto out_ok;
+			}
+			perf_mmap__read_done(&md->core);
+		}
+
+		if (nr_events == before)
+			evlist__poll(evlist, 10);
+
+		if (++nr_polls > 5) {
+			pr_debug("%s: no events!\n", __func__);
+			goto out_put_evlist;
+		}
+	}
+out_ok:
+	ret = TEST_OK;
+out_put_evlist:
+	evlist__put(evlist);
+out:
+	return ret;
+}
+
+static struct test_case tests__syscall_openat_tp_fields[] = {
+	TEST_CASE_REASON("syscalls:sys_enter_openat event fields",
+			 syscall_openat_tp_fields,
+			 "permissions"),
+	{	.name = NULL, }
+};
+
+struct test_suite suite__syscall_openat_tp_fields = {
+	.desc = "syscalls:sys_enter_openat event fields",
+	.test_cases = tests__syscall_openat_tp_fields,
+};