[PATCH v5 1/2] libbpf: kprobe.multi: cross filter using available_filter_functions and kallsyms

[PATCH v5 1/2] libbpf: kprobe.multi: cross filter using available_filter_functions and kallsyms

[Jackie Liu]

From: Jackie Liu <liuyun01@kylinos.cn>

When using regular expression matching with "kprobe multi", it scans all
the functions under "/proc/kallsyms" that can be matched. However, not all
of them can be traced by kprobe.multi. If any one of the functions fails
to be traced, it will result in the failure of all functions. The best
approach is to filter out the functions that cannot be traced to ensure
proper tracking of the functions.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202307030355.TdXOHklM-lkp@intel.com/
Suggested-by: Jiri Olsa <jolsa@kernel.org>
Suggested-by: Andrii Nakryiko <andrii.nakryiko@gmail.com>
Signed-off-by: Jackie Liu <liuyun01@kylinos.cn>
--- 
 v4->v5: simplified code

 tools/lib/bpf/libbpf.c | 106 +++++++++++++++++++++++++++++++++++++----
 1 file changed, 96 insertions(+), 10 deletions(-)

diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 214f828ece6b..327b4a9e9f0d 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -10224,6 +10224,12 @@ static const char *tracefs_uprobe_events(void)
 	return use_debugfs() ? DEBUGFS"/uprobe_events" : TRACEFS"/uprobe_events";
 }
 
+static const char *tracefs_available_filter_functions(void)
+{
+	return use_debugfs() ? DEBUGFS"/available_filter_functions" :
+			       TRACEFS"/available_filter_functions";
+}
+
 static void gen_kprobe_legacy_event_name(char *buf, size_t buf_sz,
 					 const char *kfunc_name, size_t offset)
 {
@@ -10539,14 +10545,26 @@ struct kprobe_multi_resolve {
 	size_t cnt;
 };
 
-static int
-resolve_kprobe_multi_cb(unsigned long long sym_addr, char sym_type,
-			const char *sym_name, void *ctx)
+static int avail_compare_function(const void *a, const void *b)
+{
+	return strcmp(*(const char **)a, *(const char **)b);
+}
+
+struct avail_kallsyms_data {
+	const char **syms;
+	size_t cnt;
+	struct kprobe_multi_resolve *res;
+};
+
+static int avail_kallsyms_cb(unsigned long long sym_addr, char sym_type,
+			     const char *sym_name, void *ctx)
 {
-	struct kprobe_multi_resolve *res = ctx;
+	struct avail_kallsyms_data *data = ctx;
+	struct kprobe_multi_resolve *res = data->res;
 	int err;
 
-	if (!glob_match(sym_name, res->pattern))
+	if (!bsearch(&sym_name, data->syms, data->cnt, sizeof(void *),
+		     avail_compare_function))
 		return 0;
 
 	err = libbpf_ensure_mem((void **) &res->addrs, &res->cap, sizeof(unsigned long),
@@ -10558,6 +10576,78 @@ resolve_kprobe_multi_cb(unsigned long long sym_addr, char sym_type,
 	return 0;
 }
 
+static int libbpf_available_kallsyms_parse(struct kprobe_multi_resolve *res)
+{
+	struct avail_kallsyms_data data;
+	char sym_name[500];
+	const char *available_functions_file = tracefs_available_filter_functions();
+	FILE *f;
+	int err = 0, ret, i;
+	const char **syms;
+	size_t cap = 0, cnt = 0;
+
+	f = fopen(available_functions_file, "r");
+	if (!f) {
+		err = -errno;
+		pr_warn("failed to open %s\n", available_functions_file);
+		return err;
+	}
+
+	while (true) {
+		char *name;
+
+		ret = fscanf(f, "%499s%*[^\n]\n", sym_name);
+		if (ret == EOF && feof(f))
+			break;
+
+		if (ret != 1) {
+			pr_warn("failed to read available function file entry: %d\n",
+				ret);
+			err = -EINVAL;
+			goto cleanup;
+		}
+
+		if (!glob_match(sym_name, res->pattern))
+			continue;
+
+		err = libbpf_ensure_mem((void **)&syms, &cap, sizeof(void *),
+					cnt + 1);
+		if (err)
+			goto cleanup;
+
+		name = strdup(sym_name);
+		if (!name) {
+			err = -errno;
+			goto cleanup;
+		}
+
+		syms[cnt++] = name;
+	}
+	fclose(f);
+
+	/* not found entry, return direct */
+	if (!cnt)
+		return -ENOENT;
+
+	/* sort available functions */
+	qsort(syms, cnt, sizeof(void *), avail_compare_function);
+
+	data.syms = syms;
+	data.res = res;
+	data.cnt = cnt;
+	libbpf_kallsyms_parse(avail_kallsyms_cb, &data);
+
+	if (!res->cnt)
+		err = -ENOENT;
+
+cleanup:
+	for (i = 0; i < cnt; i++)
+		free((char *)syms[i]);
+	free(syms);
+
+	return err;
+}
+
 struct bpf_link *
 bpf_program__attach_kprobe_multi_opts(const struct bpf_program *prog,
 				      const char *pattern,
@@ -10594,13 +10684,9 @@ bpf_program__attach_kprobe_multi_opts(const struct bpf_program *prog,
 		return libbpf_err_ptr(-EINVAL);
 
 	if (pattern) {
-		err = libbpf_kallsyms_parse(resolve_kprobe_multi_cb, &res);
+		err = libbpf_available_kallsyms_parse(&res);
 		if (err)
 			goto error;
-		if (!res.cnt) {
-			err = -ENOENT;
-			goto error;
-		}
 		addrs = res.addrs;
 		cnt = res.cnt;
 	}
-- 
2.25.1

[PATCH v5 2/2] libbpf: kprobe.multi: Filter with available_filter_functions_addrs

[Jackie Liu]

From: Jackie Liu <liuyun01@kylinos.cn>

Now, we provide a new available_filter_functions_addrs interface, which can
help us not need to cross-validate available_filter_functions and kallsyms,
which can effectively improve efficiency. For example, on my device, the
sample program [1] of start time:

$ sudo ./funccount "tcp_*"

before   after
1.2s     1.0s

[1]: https://github.com/JackieLiu1/ketones/tree/master/src/funccount
Signed-off-by: Jackie Liu <liuyun01@kylinos.cn>
---
 tools/lib/bpf/libbpf.c | 62 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 61 insertions(+), 1 deletion(-)

diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 327b4a9e9f0d..03ba17d8dd2f 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -10230,6 +10230,12 @@ static const char *tracefs_available_filter_functions(void)
 			       TRACEFS"/available_filter_functions";
 }
 
+static const char *tracefs_available_filter_functions_addrs(void)
+{
+	return use_debugfs() ? DEBUGFS"/available_filter_functions_addrs" :
+			       TRACEFS"/available_filter_functions_addrs";
+}
+
 static void gen_kprobe_legacy_event_name(char *buf, size_t buf_sz,
 					 const char *kfunc_name, size_t offset)
 {
@@ -10648,6 +10654,57 @@ static int libbpf_available_kallsyms_parse(struct kprobe_multi_resolve *res)
 	return err;
 }
 
+static bool has_available_filter_functions_addrs(void)
+{
+	return access(tracefs_available_filter_functions_addrs(), R_OK) != -1;
+}
+
+static int libbpf_available_kprobes_parse(struct kprobe_multi_resolve *res)
+{
+	char sym_name[500];
+	FILE *f;
+	int ret, err = 0;
+	unsigned long long sym_addr;
+	const char *available_path = tracefs_available_filter_functions_addrs();
+
+	f = fopen(available_path, "r");
+	if (!f) {
+		err = -errno;
+		pr_warn("failed to open %s\n", available_path);
+		return err;
+	}
+
+	while (true) {
+		ret = fscanf(f, "%llx %499s%*[^\n]\n", &sym_addr, sym_name);
+		if (ret == EOF && feof(f))
+			break;
+
+		if (ret != 2) {
+			pr_warn("failed to read available kprobe entry: %d\n",
+				ret);
+			err = -EINVAL;
+			goto cleanup;
+		}
+
+		if (!glob_match(sym_name, res->pattern))
+			continue;
+
+		err = libbpf_ensure_mem((void **) &res->addrs, &res->cap,
+					sizeof(unsigned long), res->cnt + 1);
+		if (err)
+			goto cleanup;
+
+		res->addrs[res->cnt++] = (unsigned long) sym_addr;
+	}
+
+	if (!res->cnt)
+		err = -ENOENT;
+
+cleanup:
+	fclose(f);
+	return err;
+}
+
 struct bpf_link *
 bpf_program__attach_kprobe_multi_opts(const struct bpf_program *prog,
 				      const char *pattern,
@@ -10684,7 +10741,10 @@ bpf_program__attach_kprobe_multi_opts(const struct bpf_program *prog,
 		return libbpf_err_ptr(-EINVAL);
 
 	if (pattern) {
-		err = libbpf_available_kallsyms_parse(&res);
+		if (has_available_filter_functions_addrs())
+			err = libbpf_available_kprobes_parse(&res);
+		else
+			err = libbpf_available_kallsyms_parse(&res);
 		if (err)
 			goto error;
 		addrs = res.addrs;
-- 
2.25.1