[PATCH v4 bpf-next 0/2] bpf: Fix fill_link_info and add selftest

[PATCH v4 bpf-next 0/2] bpf: Fix fill_link_info and add selftest

[Yafang Shao]

Patch #1: Fix an error in fill_link_info reported by Dan
Patch #2: Add selftest for #1

v3->v4:
  - Comments from Yonghong
    - re-word the kptr comment
    - Avoid unnecessary switch statement 
    - No need to check the return value of bpf_link__fd()
    - offset is always 0
    - Return directly when test_fill_link_info__open_and_load() fails
    - Cleanup skel when load_kallsyms_refresh() fails

v2->v3:
  - Comments from Jiri
    - Verify wrong arguments as soon as possible
    - Use CONFIG_X86_KERNEL_IBT
    - No need to make the test serial
    - Add test case for kprobe_multi

v1->v2:
  - Fix BPF CI failure due to the enabled ENDBDR

Yafang Shao (2):
  bpf: Fix uninitialized symbol in bpf_perf_link_fill_kprobe()
  selftests/bpf: Add selftest for fill_link_info

 kernel/bpf/syscall.c                               |   4 +-
 tools/testing/selftests/bpf/DENYLIST.aarch64       |   3 +
 .../selftests/bpf/prog_tests/fill_link_info.c      | 337 +++++++++++++++++++++
 .../selftests/bpf/progs/test_fill_link_info.c      |  42 +++
 4 files changed, 384 insertions(+), 2 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/fill_link_info.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_fill_link_info.c

-- 
1.8.3.1

[PATCH v4 bpf-next 1/2] bpf: Fix uninitialized symbol in bpf_perf_link_fill_kprobe()

[Yafang Shao]

The patch 1b715e1b0ec5: "bpf: Support ->fill_link_info for
perf_event" from Jul 9, 2023, leads to the following Smatch static
checker warning:

    kernel/bpf/syscall.c:3416 bpf_perf_link_fill_kprobe()
    error: uninitialized symbol 'type'.

That can happens when uname is NULL. So fix it by verifying the uname
when we really need to fill it.

Fixes: 1b715e1b0ec5 ("bpf: Support ->fill_link_info for perf_event")
Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Closes: https://lore.kernel.org/bpf/85697a7e-f897-4f74-8b43-82721bebc462@kili.mountain/
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Yonghong Song <yonghong.song@linux.dev>
---
 kernel/bpf/syscall.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 7f4e8c3..166390f 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -3378,14 +3378,14 @@ static int bpf_perf_link_fill_common(const struct perf_event *event,
 
 	if (!ulen ^ !uname)
 		return -EINVAL;
-	if (!uname)
-		return 0;
 
 	err = bpf_get_perf_event_info(event, &prog_id, fd_type, &buf,
 				      probe_offset, probe_addr);
 	if (err)
 		return err;
 
+	if (!uname)
+		return 0;
 	if (buf) {
 		len = strlen(buf);
 		err = bpf_copy_to_user(uname, buf, ulen, len);
-- 
1.8.3.1

[PATCH v4 bpf-next 2/2] selftests/bpf: Add selftest for fill_link_info

[Yafang Shao]

Add selftest for the fill_link_info of uprobe, kprobe and tracepoint.
The result:

  $ tools/testing/selftests/bpf/test_progs --name=fill_link_info
  #79/1    fill_link_info/kprobe_link_info:OK
  #79/2    fill_link_info/kretprobe_link_info:OK
  #79/3    fill_link_info/kprobe_fill_invalid_user_buff:OK
  #79/4    fill_link_info/tracepoint_link_info:OK
  #79/5    fill_link_info/uprobe_link_info:OK
  #79/6    fill_link_info/uretprobe_link_info:OK
  #79/7    fill_link_info/kprobe_multi_link_info:OK
  #79/8    fill_link_info/kretprobe_multi_link_info:OK
  #79/9    fill_link_info/kprobe_multi_ubuff:OK
  #79      fill_link_info:OK
  Summary: 1/9 PASSED, 0 SKIPPED, 0 FAILED

The test case for kprobe_multi won't be run on aarch64, as it is not
supported.

Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
---
 tools/testing/selftests/bpf/DENYLIST.aarch64       |   3 +
 .../selftests/bpf/prog_tests/fill_link_info.c      | 337 +++++++++++++++++++++
 .../selftests/bpf/progs/test_fill_link_info.c      |  42 +++
 3 files changed, 382 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/fill_link_info.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_fill_link_info.c

diff --git a/tools/testing/selftests/bpf/DENYLIST.aarch64 b/tools/testing/selftests/bpf/DENYLIST.aarch64
index 3b61e8b..b2f46b6 100644
--- a/tools/testing/selftests/bpf/DENYLIST.aarch64
+++ b/tools/testing/selftests/bpf/DENYLIST.aarch64
@@ -12,3 +12,6 @@ kprobe_multi_test/skel_api                       # libbpf: failed to load BPF sk
 module_attach                                    # prog 'kprobe_multi': failed to auto-attach: -95
 fentry_test/fentry_many_args                     # fentry_many_args:FAIL:fentry_many_args_attach unexpected error: -524
 fexit_test/fexit_many_args                       # fexit_many_args:FAIL:fexit_many_args_attach unexpected error: -524
+fill_link_info/kprobe_multi_link_info            # bpf_program__attach_kprobe_multi_opts unexpected error: -95
+fill_link_info/kretprobe_multi_link_info         # bpf_program__attach_kprobe_multi_opts unexpected error: -95
+fill_link_info/kprobe_multi_ubuff                # bpf_program__attach_kprobe_multi_opts unexpected error: -95
diff --git a/tools/testing/selftests/bpf/prog_tests/fill_link_info.c b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
new file mode 100644
index 0000000..001a8b5
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
@@ -0,0 +1,337 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (C) 2023 Yafang Shao <laoar.shao@gmail.com> */
+
+#include <string.h>
+#include <linux/bpf.h>
+#include <linux/limits.h>
+#include <test_progs.h>
+#include "trace_helpers.h"
+#include "test_fill_link_info.skel.h"
+
+#define TP_CAT "sched"
+#define TP_NAME "sched_switch"
+#define KPROBE_FUNC "tcp_rcv_established"
+#define UPROBE_FILE "/proc/self/exe"
+#define KMULTI_CNT (4)
+
+/* uprobe attach point */
+static noinline void uprobe_func(void)
+{
+	asm volatile ("");
+}
+
+static int verify_perf_link_info(int fd, enum bpf_perf_event_type type, long addr,
+				 ssize_t offset, ssize_t entry_offset)
+{
+	struct bpf_link_info info;
+	__u32 len = sizeof(info);
+	char buf[PATH_MAX];
+	int err = 0;
+
+	memset(&info, 0, sizeof(info));
+	buf[0] = '\0';
+
+again:
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	if (!ASSERT_OK(err, "get_link_info"))
+		return -1;
+
+	if (!ASSERT_EQ(info.type, BPF_LINK_TYPE_PERF_EVENT, "link_type"))
+		return -1;
+	if (!ASSERT_EQ(info.perf_event.type, type, "perf_type_match"))
+		return -1;
+
+	switch (info.perf_event.type) {
+	case BPF_PERF_EVENT_KPROBE:
+	case BPF_PERF_EVENT_KRETPROBE:
+		ASSERT_EQ(info.perf_event.kprobe.offset, offset, "kprobe_offset");
+
+		/* In case kernel.kptr_restrict is not permitted or MAX_SYMS is reached */
+		if (addr)
+			ASSERT_EQ(info.perf_event.kprobe.addr, addr + entry_offset,
+				  "kprobe_addr");
+
+		if (!info.perf_event.kprobe.func_name) {
+			ASSERT_EQ(info.perf_event.kprobe.name_len, 0, "name_len");
+			info.perf_event.kprobe.func_name = ptr_to_u64(&buf);
+			info.perf_event.kprobe.name_len = sizeof(buf);
+			goto again;
+		}
+
+		err = strncmp(u64_to_ptr(info.perf_event.kprobe.func_name), KPROBE_FUNC,
+			      strlen(KPROBE_FUNC));
+		ASSERT_EQ(err, 0, "cmp_kprobe_func_name");
+		break;
+	case BPF_PERF_EVENT_TRACEPOINT:
+		if (!info.perf_event.tracepoint.tp_name) {
+			ASSERT_EQ(info.perf_event.tracepoint.name_len, 0, "name_len");
+			info.perf_event.tracepoint.tp_name = ptr_to_u64(&buf);
+			info.perf_event.tracepoint.name_len = sizeof(buf);
+			goto again;
+		}
+
+		err = strncmp(u64_to_ptr(info.perf_event.tracepoint.tp_name), TP_NAME,
+			      strlen(TP_NAME));
+		ASSERT_EQ(err, 0, "cmp_tp_name");
+		break;
+	case BPF_PERF_EVENT_UPROBE:
+	case BPF_PERF_EVENT_URETPROBE:
+		ASSERT_EQ(info.perf_event.uprobe.offset, offset, "uprobe_offset");
+
+		if (!info.perf_event.uprobe.file_name) {
+			ASSERT_EQ(info.perf_event.uprobe.name_len, 0, "name_len");
+			info.perf_event.uprobe.file_name = ptr_to_u64(&buf);
+			info.perf_event.uprobe.name_len = sizeof(buf);
+			goto again;
+		}
+
+		err = strncmp(u64_to_ptr(info.perf_event.uprobe.file_name), UPROBE_FILE,
+			      strlen(UPROBE_FILE));
+			ASSERT_EQ(err, 0, "cmp_file_name");
+		break;
+	default:
+		break;
+	}
+	return err;
+}
+
+static void kprobe_fill_invalid_user_buffer(int fd)
+{
+	struct bpf_link_info info;
+	__u32 len = sizeof(info);
+	int err;
+
+	memset(&info, 0, sizeof(info));
+
+	info.perf_event.kprobe.func_name = 0x1; /* invalid address */
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	ASSERT_EQ(err, -EINVAL, "invalid_buff_and_len");
+
+	info.perf_event.kprobe.name_len = 64;
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	ASSERT_EQ(err, -EFAULT, "invalid_buff");
+
+	info.perf_event.kprobe.func_name = 0;
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	ASSERT_EQ(err, -EINVAL, "invalid_len");
+
+	ASSERT_EQ(info.perf_event.kprobe.addr, 0, "func_addr");
+	ASSERT_EQ(info.perf_event.kprobe.offset, 0, "func_offset");
+	ASSERT_EQ(info.perf_event.type, 0, "type");
+}
+
+static void test_kprobe_fill_link_info(struct test_fill_link_info *skel,
+				       enum bpf_perf_event_type type,
+				       bool retprobe, bool invalid)
+{
+	DECLARE_LIBBPF_OPTS(bpf_kprobe_opts, opts,
+		.attach_mode = PROBE_ATTACH_MODE_LINK,
+		.retprobe = retprobe,
+	);
+	ssize_t offset = 0, entry_offset = 0;
+	int link_fd, err;
+	long addr;
+
+	skel->links.kprobe_run = bpf_program__attach_kprobe_opts(skel->progs.kprobe_run,
+								 KPROBE_FUNC, &opts);
+	if (!ASSERT_OK_PTR(skel->links.kprobe_run, "attach_kprobe"))
+		return;
+
+	link_fd = bpf_link__fd(skel->links.kprobe_run);
+	addr = ksym_get_addr(KPROBE_FUNC);
+	if (!invalid) {
+		/* See also arch_adjust_kprobe_addr(). */
+		if (skel->kconfig->CONFIG_X86_KERNEL_IBT)
+			entry_offset = 4;
+		err = verify_perf_link_info(link_fd, type, addr, offset, entry_offset);
+		ASSERT_OK(err, "verify_perf_link_info");
+	} else {
+		kprobe_fill_invalid_user_buffer(link_fd);
+	}
+	bpf_link__detach(skel->links.kprobe_run);
+}
+
+static void test_tp_fill_link_info(struct test_fill_link_info *skel)
+{
+	int link_fd, err;
+
+	skel->links.tp_run = bpf_program__attach_tracepoint(skel->progs.tp_run, TP_CAT, TP_NAME);
+	if (!ASSERT_OK_PTR(skel->links.tp_run, "attach_tp"))
+		return;
+
+	link_fd = bpf_link__fd(skel->links.tp_run);
+	err = verify_perf_link_info(link_fd, BPF_PERF_EVENT_TRACEPOINT, 0, 0, 0);
+	ASSERT_OK(err, "verify_perf_link_info");
+	bpf_link__detach(skel->links.tp_run);
+}
+
+static void test_uprobe_fill_link_info(struct test_fill_link_info *skel,
+				       enum bpf_perf_event_type type, ssize_t offset,
+				       bool retprobe)
+{
+	int link_fd, err;
+
+	skel->links.uprobe_run = bpf_program__attach_uprobe(skel->progs.uprobe_run, retprobe,
+							    0, /* self pid */
+							    UPROBE_FILE, offset);
+	if (!ASSERT_OK_PTR(skel->links.uprobe_run, "attach_uprobe"))
+		return;
+
+	link_fd = bpf_link__fd(skel->links.uprobe_run);
+	err = verify_perf_link_info(link_fd, type, 0, offset, 0);
+	ASSERT_OK(err, "verify_perf_link_info");
+	bpf_link__detach(skel->links.uprobe_run);
+}
+
+static int verify_kmulti_link_info(int fd, const __u64 *addrs, bool retprobe)
+{
+	__u64 kmulti_addrs[KMULTI_CNT];
+	struct bpf_link_info info;
+	__u32 len = sizeof(info);
+	int flags, i, err = 0;
+
+	memset(&info, 0, sizeof(info));
+
+again:
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	if (!ASSERT_OK(err, "get_link_info"))
+		return -1;
+
+	if (!ASSERT_EQ(info.type, BPF_LINK_TYPE_KPROBE_MULTI, "kmulti_type"))
+		return -1;
+
+	ASSERT_EQ(info.kprobe_multi.count, KMULTI_CNT, "func_cnt");
+	flags = info.kprobe_multi.flags & BPF_F_KPROBE_MULTI_RETURN;
+	if (!retprobe)
+		ASSERT_EQ(flags, 0, "kmulti_flags");
+	else
+		ASSERT_NEQ(flags, 0, "kretmulti_flags");
+
+	if (!info.kprobe_multi.addrs) {
+		info.kprobe_multi.addrs = ptr_to_u64(kmulti_addrs);
+		goto again;
+	}
+	for (i = 0; i < KMULTI_CNT; i++)
+		ASSERT_EQ(kmulti_addrs[i], addrs[i], "kmulti_addrs");
+	return 0;
+}
+
+static void verify_kmulti_user_buffer(int fd, const __u64 *addrs)
+{
+	__u64 kmulti_addrs[KMULTI_CNT];
+	struct bpf_link_info info;
+	__u32 len = sizeof(info);
+	int err, i;
+
+	memset(&info, 0, sizeof(info));
+
+	info.kprobe_multi.count = KMULTI_CNT;
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	ASSERT_EQ(err, -EINVAL, "no_addr");
+
+	info.kprobe_multi.addrs = ptr_to_u64(kmulti_addrs);
+	info.kprobe_multi.count = 0;
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	ASSERT_EQ(err, -EINVAL, "no_cnt");
+
+	for (i = 0; i < KMULTI_CNT; i++)
+		kmulti_addrs[i] = 0;
+	info.kprobe_multi.count = KMULTI_CNT - 1;
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	ASSERT_EQ(err, -ENOSPC, "smaller_cnt");
+	for (i = 0; i < KMULTI_CNT - 1; i++)
+		ASSERT_EQ(kmulti_addrs[i], addrs[i], "kmulti_addrs");
+	ASSERT_EQ(kmulti_addrs[i], 0, "kmulti_addrs");
+
+	for (i = 0; i < KMULTI_CNT; i++)
+		kmulti_addrs[i] = 0;
+	info.kprobe_multi.count = KMULTI_CNT + 1;
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	ASSERT_EQ(err, 0, "bigger_cnt");
+	for (i = 0; i < KMULTI_CNT; i++)
+		ASSERT_EQ(kmulti_addrs[i], addrs[i], "kmulti_addrs");
+
+	info.kprobe_multi.count = KMULTI_CNT;
+	info.kprobe_multi.addrs = 0x1; /* invalid addr */
+	err = bpf_link_get_info_by_fd(fd, &info, &len);
+	ASSERT_EQ(err, -EFAULT, "invalid_buff");
+}
+
+static int symbols_cmp_r(const void *a, const void *b)
+{
+	const char **str_a = (const char **) a;
+	const char **str_b = (const char **) b;
+
+	return strcmp(*str_a, *str_b);
+}
+
+static void test_kprobe_multi_fill_link_info(struct test_fill_link_info *skel,
+					     bool retprobe, bool buffer)
+{
+	LIBBPF_OPTS(bpf_kprobe_multi_opts, opts);
+	const char *syms[KMULTI_CNT] = {
+		"schedule_timeout_interruptible",
+		"schedule_timeout_uninterruptible",
+		"schedule_timeout_idle",
+		"schedule_timeout_killable",
+	};
+	__u64 addrs[KMULTI_CNT];
+	int link_fd, i, err = 0;
+
+	qsort(syms, KMULTI_CNT, sizeof(syms[0]), symbols_cmp_r);
+	opts.syms = syms;
+	opts.cnt = KMULTI_CNT;
+	opts.retprobe = retprobe;
+	skel->links.kmulti_run = bpf_program__attach_kprobe_multi_opts(skel->progs.kmulti_run,
+								       NULL, &opts);
+	if (!ASSERT_OK_PTR(skel->links.kmulti_run, "attach_kprobe_multi"))
+		return;
+
+	link_fd = bpf_link__fd(skel->links.kmulti_run);
+	for (i = 0; i < KMULTI_CNT; i++)
+		addrs[i] = ksym_get_addr(syms[i]);
+
+	if (!buffer)
+		err = verify_kmulti_link_info(link_fd, addrs, retprobe);
+	else
+		verify_kmulti_user_buffer(link_fd, addrs);
+	ASSERT_OK(err, "verify_kmulti_link_info");
+	bpf_link__detach(skel->links.kmulti_run);
+}
+
+void test_fill_link_info(void)
+{
+	struct test_fill_link_info *skel;
+	ssize_t offset;
+
+	skel = test_fill_link_info__open_and_load();
+	if (!ASSERT_OK_PTR(skel, "skel_open"))
+		return;
+
+	/* load kallsyms to compare the addr */
+	if (!ASSERT_OK(load_kallsyms_refresh(), "load_kallsyms_refresh"))
+		goto cleanup;
+	if (test__start_subtest("kprobe_link_info"))
+		test_kprobe_fill_link_info(skel, BPF_PERF_EVENT_KPROBE, false, false);
+	if (test__start_subtest("kretprobe_link_info"))
+		test_kprobe_fill_link_info(skel, BPF_PERF_EVENT_KRETPROBE, true, false);
+	if (test__start_subtest("kprobe_fill_invalid_user_buff"))
+		test_kprobe_fill_link_info(skel, BPF_PERF_EVENT_KPROBE, false, true);
+	if (test__start_subtest("tracepoint_link_info"))
+		test_tp_fill_link_info(skel);
+
+	offset = get_uprobe_offset(&uprobe_func);
+	if (test__start_subtest("uprobe_link_info"))
+		test_uprobe_fill_link_info(skel, BPF_PERF_EVENT_UPROBE, offset, false);
+	if (test__start_subtest("uretprobe_link_info"))
+		test_uprobe_fill_link_info(skel, BPF_PERF_EVENT_URETPROBE, offset, true);
+	if (test__start_subtest("kprobe_multi_link_info"))
+		test_kprobe_multi_fill_link_info(skel, false, false);
+	if (test__start_subtest("kretprobe_multi_link_info"))
+		test_kprobe_multi_fill_link_info(skel, true, false);
+	if (test__start_subtest("kprobe_multi_ubuff"))
+		test_kprobe_multi_fill_link_info(skel, true, true);
+
+cleanup:
+	test_fill_link_info__destroy(skel);
+}
diff --git a/tools/testing/selftests/bpf/progs/test_fill_link_info.c b/tools/testing/selftests/bpf/progs/test_fill_link_info.c
new file mode 100644
index 0000000..564f402
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/test_fill_link_info.c
@@ -0,0 +1,42 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (C) 2023 Yafang Shao <laoar.shao@gmail.com> */
+
+#include "vmlinux.h"
+#include <bpf/bpf_tracing.h>
+#include <stdbool.h>
+
+extern bool CONFIG_X86_KERNEL_IBT __kconfig __weak;
+
+/* This function is here to have CONFIG_X86_KERNEL_IBT
+ * used and added to object BTF.
+ */
+int unused(void)
+{
+	return CONFIG_X86_KERNEL_IBT ? 0 : 1;
+}
+
+SEC("kprobe")
+int BPF_PROG(kprobe_run)
+{
+	return 0;
+}
+
+SEC("uprobe")
+int BPF_PROG(uprobe_run)
+{
+	return 0;
+}
+
+SEC("tracepoint")
+int BPF_PROG(tp_run)
+{
+	return 0;
+}
+
+SEC("kprobe.multi")
+int BPF_PROG(kmulti_run)
+{
+	return 0;
+}
+
+char _license[] SEC("license") = "GPL";
-- 
1.8.3.1

Re: [PATCH v4 bpf-next 2/2] selftests/bpf: Add selftest for fill_link_info

[Yonghong Song]

On 8/4/23 3:57 AM, Yafang Shao wrote:
> Add selftest for the fill_link_info of uprobe, kprobe and tracepoint.
> The result:
> 
>    $ tools/testing/selftests/bpf/test_progs --name=fill_link_info
>    #79/1    fill_link_info/kprobe_link_info:OK
>    #79/2    fill_link_info/kretprobe_link_info:OK
>    #79/3    fill_link_info/kprobe_fill_invalid_user_buff:OK
>    #79/4    fill_link_info/tracepoint_link_info:OK
>    #79/5    fill_link_info/uprobe_link_info:OK
>    #79/6    fill_link_info/uretprobe_link_info:OK
>    #79/7    fill_link_info/kprobe_multi_link_info:OK
>    #79/8    fill_link_info/kretprobe_multi_link_info:OK
>    #79/9    fill_link_info/kprobe_multi_ubuff:OK
>    #79      fill_link_info:OK
>    Summary: 1/9 PASSED, 0 SKIPPED, 0 FAILED
> 
> The test case for kprobe_multi won't be run on aarch64, as it is not
> supported.
> 
> Signed-off-by: Yafang Shao <laoar.shao@gmail.com>

Ack with a few nits below.

Acked-by: Yonghong Song <yonghong.song@linux.dev>

> ---
>   tools/testing/selftests/bpf/DENYLIST.aarch64       |   3 +
>   .../selftests/bpf/prog_tests/fill_link_info.c      | 337 +++++++++++++++++++++
>   .../selftests/bpf/progs/test_fill_link_info.c      |  42 +++
>   3 files changed, 382 insertions(+)
>   create mode 100644 tools/testing/selftests/bpf/prog_tests/fill_link_info.c
>   create mode 100644 tools/testing/selftests/bpf/progs/test_fill_link_info.c
> 
> diff --git a/tools/testing/selftests/bpf/DENYLIST.aarch64 b/tools/testing/selftests/bpf/DENYLIST.aarch64
> index 3b61e8b..b2f46b6 100644
> --- a/tools/testing/selftests/bpf/DENYLIST.aarch64
> +++ b/tools/testing/selftests/bpf/DENYLIST.aarch64
> @@ -12,3 +12,6 @@ kprobe_multi_test/skel_api                       # libbpf: failed to load BPF sk
>   module_attach                                    # prog 'kprobe_multi': failed to auto-attach: -95
>   fentry_test/fentry_many_args                     # fentry_many_args:FAIL:fentry_many_args_attach unexpected error: -524
>   fexit_test/fexit_many_args                       # fexit_many_args:FAIL:fexit_many_args_attach unexpected error: -524
> +fill_link_info/kprobe_multi_link_info            # bpf_program__attach_kprobe_multi_opts unexpected error: -95
> +fill_link_info/kretprobe_multi_link_info         # bpf_program__attach_kprobe_multi_opts unexpected error: -95
> +fill_link_info/kprobe_multi_ubuff                # bpf_program__attach_kprobe_multi_opts unexpected error: -95
> diff --git a/tools/testing/selftests/bpf/prog_tests/fill_link_info.c b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
> new file mode 100644
> index 0000000..001a8b5
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
> @@ -0,0 +1,337 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/* Copyright (C) 2023 Yafang Shao <laoar.shao@gmail.com> */
> +
> +#include <string.h>
> +#include <linux/bpf.h>
> +#include <linux/limits.h>
> +#include <test_progs.h>
> +#include "trace_helpers.h"
> +#include "test_fill_link_info.skel.h"
> +
> +#define TP_CAT "sched"
> +#define TP_NAME "sched_switch"
> +#define KPROBE_FUNC "tcp_rcv_established"
> +#define UPROBE_FILE "/proc/self/exe"
> +#define KMULTI_CNT (4)
> +
> +/* uprobe attach point */
> +static noinline void uprobe_func(void)
> +{
> +	asm volatile ("");
> +}
> +
> +static int verify_perf_link_info(int fd, enum bpf_perf_event_type type, long addr,
> +				 ssize_t offset, ssize_t entry_offset)
> +{
> +	struct bpf_link_info info;
> +	__u32 len = sizeof(info);
> +	char buf[PATH_MAX];
> +	int err = 0;
> +
> +	memset(&info, 0, sizeof(info));
> +	buf[0] = '\0';
> +
> +again:
> +	err = bpf_link_get_info_by_fd(fd, &info, &len);
> +	if (!ASSERT_OK(err, "get_link_info"))
> +		return -1;
> +
> +	if (!ASSERT_EQ(info.type, BPF_LINK_TYPE_PERF_EVENT, "link_type"))
> +		return -1;
> +	if (!ASSERT_EQ(info.perf_event.type, type, "perf_type_match"))
> +		return -1;
> +
> +	switch (info.perf_event.type) {
> +	case BPF_PERF_EVENT_KPROBE:
> +	case BPF_PERF_EVENT_KRETPROBE:
> +		ASSERT_EQ(info.perf_event.kprobe.offset, offset, "kprobe_offset");
> +
> +		/* In case kernel.kptr_restrict is not permitted or MAX_SYMS is reached */
> +		if (addr)
> +			ASSERT_EQ(info.perf_event.kprobe.addr, addr + entry_offset,
> +				  "kprobe_addr");
> +
> +		if (!info.perf_event.kprobe.func_name) {
> +			ASSERT_EQ(info.perf_event.kprobe.name_len, 0, "name_len");
> +			info.perf_event.kprobe.func_name = ptr_to_u64(&buf);
> +			info.perf_event.kprobe.name_len = sizeof(buf);
> +			goto again;
> +		}
> +
> +		err = strncmp(u64_to_ptr(info.perf_event.kprobe.func_name), KPROBE_FUNC,
> +			      strlen(KPROBE_FUNC));
> +		ASSERT_EQ(err, 0, "cmp_kprobe_func_name");
> +		break;
> +	case BPF_PERF_EVENT_TRACEPOINT:
> +		if (!info.perf_event.tracepoint.tp_name) {
> +			ASSERT_EQ(info.perf_event.tracepoint.name_len, 0, "name_len");
> +			info.perf_event.tracepoint.tp_name = ptr_to_u64(&buf);
> +			info.perf_event.tracepoint.name_len = sizeof(buf);
> +			goto again;
> +		}
> +
> +		err = strncmp(u64_to_ptr(info.perf_event.tracepoint.tp_name), TP_NAME,
> +			      strlen(TP_NAME));
> +		ASSERT_EQ(err, 0, "cmp_tp_name");
> +		break;
> +	case BPF_PERF_EVENT_UPROBE:
> +	case BPF_PERF_EVENT_URETPROBE:
> +		ASSERT_EQ(info.perf_event.uprobe.offset, offset, "uprobe_offset");
> +
> +		if (!info.perf_event.uprobe.file_name) {
> +			ASSERT_EQ(info.perf_event.uprobe.name_len, 0, "name_len");
> +			info.perf_event.uprobe.file_name = ptr_to_u64(&buf);
> +			info.perf_event.uprobe.name_len = sizeof(buf);
> +			goto again;
> +		}
> +
> +		err = strncmp(u64_to_ptr(info.perf_event.uprobe.file_name), UPROBE_FILE,
> +			      strlen(UPROBE_FILE));
> +			ASSERT_EQ(err, 0, "cmp_file_name");
> +		break;
> +	default:

Is the 'default' case possible? Probably not, right? Then
let us add
		err = -1;
to indicate this path is not possible.

> +		break;
> +	}
> +	return err;
> +}
> +
[...]
> +
> +static void test_kprobe_fill_link_info(struct test_fill_link_info *skel,
> +				       enum bpf_perf_event_type type,
> +				       bool retprobe, bool invalid)
> +{
> +	DECLARE_LIBBPF_OPTS(bpf_kprobe_opts, opts,
> +		.attach_mode = PROBE_ATTACH_MODE_LINK,
> +		.retprobe = retprobe,
> +	);
> +	ssize_t offset = 0, entry_offset = 0;

Remove 'offset = 0'.

> +	int link_fd, err;
> +	long addr;
> +
> +	skel->links.kprobe_run = bpf_program__attach_kprobe_opts(skel->progs.kprobe_run,
> +								 KPROBE_FUNC, &opts);
> +	if (!ASSERT_OK_PTR(skel->links.kprobe_run, "attach_kprobe"))
> +		return;
> +
> +	link_fd = bpf_link__fd(skel->links.kprobe_run);
> +	addr = ksym_get_addr(KPROBE_FUNC);
> +	if (!invalid) {
> +		/* See also arch_adjust_kprobe_addr(). */
> +		if (skel->kconfig->CONFIG_X86_KERNEL_IBT)
> +			entry_offset = 4;
> +		err = verify_perf_link_info(link_fd, type, addr, offset, entry_offset);

Replease 'offset' with '0'.

> +		ASSERT_OK(err, "verify_perf_link_info");
> +	} else {
> +		kprobe_fill_invalid_user_buffer(link_fd);
> +	}
> +	bpf_link__detach(skel->links.kprobe_run);
> +}
> +
[...]
> +
> +static void test_kprobe_multi_fill_link_info(struct test_fill_link_info *skel,
> +					     bool retprobe, bool buffer)
> +{
> +	LIBBPF_OPTS(bpf_kprobe_multi_opts, opts);
> +	const char *syms[KMULTI_CNT] = {
> +		"schedule_timeout_interruptible",
> +		"schedule_timeout_uninterruptible",
> +		"schedule_timeout_idle",
> +		"schedule_timeout_killable",
> +	};
> +	__u64 addrs[KMULTI_CNT];
> +	int link_fd, i, err = 0;

'err = 0' => 'err'.

> +
> +	qsort(syms, KMULTI_CNT, sizeof(syms[0]), symbols_cmp_r);
> +	opts.syms = syms;
> +	opts.cnt = KMULTI_CNT;
> +	opts.retprobe = retprobe;
> +	skel->links.kmulti_run = bpf_program__attach_kprobe_multi_opts(skel->progs.kmulti_run,
> +								       NULL, &opts);
> +	if (!ASSERT_OK_PTR(skel->links.kmulti_run, "attach_kprobe_multi"))
> +		return;
> +
> +	link_fd = bpf_link__fd(skel->links.kmulti_run);
> +	for (i = 0; i < KMULTI_CNT; i++)
> +		addrs[i] = ksym_get_addr(syms[i]);
> +
> +	if (!buffer)
> +		err = verify_kmulti_link_info(link_fd, addrs, retprobe);
> +	else
> +		verify_kmulti_user_buffer(link_fd, addrs);
> +	ASSERT_OK(err, "verify_kmulti_link_info");

if (!buffer) {
	err = verify_kmulti_link_info(link_fd, addrs, retprobe);
	ASSERT_OK(err, "verify_kmulti_link_info");
} else {
	verify_kmulti_user_buffer(link_fd, addrs);
}

> +	bpf_link__detach(skel->links.kmulti_run);
> +}
> +
[...]

Re: [PATCH v4 bpf-next 2/2] selftests/bpf: Add selftest for fill_link_info

[Yafang Shao]

On Sat, Aug 5, 2023 at 12:22 AM Yonghong Song <yonghong.song@linux.dev> wrote:
>
>
>
> On 8/4/23 3:57 AM, Yafang Shao wrote:
> > Add selftest for the fill_link_info of uprobe, kprobe and tracepoint.
> > The result:
> >
> >    $ tools/testing/selftests/bpf/test_progs --name=fill_link_info
> >    #79/1    fill_link_info/kprobe_link_info:OK
> >    #79/2    fill_link_info/kretprobe_link_info:OK
> >    #79/3    fill_link_info/kprobe_fill_invalid_user_buff:OK
> >    #79/4    fill_link_info/tracepoint_link_info:OK
> >    #79/5    fill_link_info/uprobe_link_info:OK
> >    #79/6    fill_link_info/uretprobe_link_info:OK
> >    #79/7    fill_link_info/kprobe_multi_link_info:OK
> >    #79/8    fill_link_info/kretprobe_multi_link_info:OK
> >    #79/9    fill_link_info/kprobe_multi_ubuff:OK
> >    #79      fill_link_info:OK
> >    Summary: 1/9 PASSED, 0 SKIPPED, 0 FAILED
> >
> > The test case for kprobe_multi won't be run on aarch64, as it is not
> > supported.
> >
> > Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
>
> Ack with a few nits below.
>
> Acked-by: Yonghong Song <yonghong.song@linux.dev>
>
> > ---
> >   tools/testing/selftests/bpf/DENYLIST.aarch64       |   3 +
> >   .../selftests/bpf/prog_tests/fill_link_info.c      | 337 +++++++++++++++++++++
> >   .../selftests/bpf/progs/test_fill_link_info.c      |  42 +++
> >   3 files changed, 382 insertions(+)
> >   create mode 100644 tools/testing/selftests/bpf/prog_tests/fill_link_info.c
> >   create mode 100644 tools/testing/selftests/bpf/progs/test_fill_link_info.c
> >
> > diff --git a/tools/testing/selftests/bpf/DENYLIST.aarch64 b/tools/testing/selftests/bpf/DENYLIST.aarch64
> > index 3b61e8b..b2f46b6 100644
> > --- a/tools/testing/selftests/bpf/DENYLIST.aarch64
> > +++ b/tools/testing/selftests/bpf/DENYLIST.aarch64
> > @@ -12,3 +12,6 @@ kprobe_multi_test/skel_api                       # libbpf: failed to load BPF sk
> >   module_attach                                    # prog 'kprobe_multi': failed to auto-attach: -95
> >   fentry_test/fentry_many_args                     # fentry_many_args:FAIL:fentry_many_args_attach unexpected error: -524
> >   fexit_test/fexit_many_args                       # fexit_many_args:FAIL:fexit_many_args_attach unexpected error: -524
> > +fill_link_info/kprobe_multi_link_info            # bpf_program__attach_kprobe_multi_opts unexpected error: -95
> > +fill_link_info/kretprobe_multi_link_info         # bpf_program__attach_kprobe_multi_opts unexpected error: -95
> > +fill_link_info/kprobe_multi_ubuff                # bpf_program__attach_kprobe_multi_opts unexpected error: -95
> > diff --git a/tools/testing/selftests/bpf/prog_tests/fill_link_info.c b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
> > new file mode 100644
> > index 0000000..001a8b5
> > --- /dev/null
> > +++ b/tools/testing/selftests/bpf/prog_tests/fill_link_info.c
> > @@ -0,0 +1,337 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/* Copyright (C) 2023 Yafang Shao <laoar.shao@gmail.com> */
> > +
> > +#include <string.h>
> > +#include <linux/bpf.h>
> > +#include <linux/limits.h>
> > +#include <test_progs.h>
> > +#include "trace_helpers.h"
> > +#include "test_fill_link_info.skel.h"
> > +
> > +#define TP_CAT "sched"
> > +#define TP_NAME "sched_switch"
> > +#define KPROBE_FUNC "tcp_rcv_established"
> > +#define UPROBE_FILE "/proc/self/exe"
> > +#define KMULTI_CNT (4)
> > +
> > +/* uprobe attach point */
> > +static noinline void uprobe_func(void)
> > +{
> > +     asm volatile ("");
> > +}
> > +
> > +static int verify_perf_link_info(int fd, enum bpf_perf_event_type type, long addr,
> > +                              ssize_t offset, ssize_t entry_offset)
> > +{
> > +     struct bpf_link_info info;
> > +     __u32 len = sizeof(info);
> > +     char buf[PATH_MAX];
> > +     int err = 0;
> > +
> > +     memset(&info, 0, sizeof(info));
> > +     buf[0] = '\0';
> > +
> > +again:
> > +     err = bpf_link_get_info_by_fd(fd, &info, &len);
> > +     if (!ASSERT_OK(err, "get_link_info"))
> > +             return -1;
> > +
> > +     if (!ASSERT_EQ(info.type, BPF_LINK_TYPE_PERF_EVENT, "link_type"))
> > +             return -1;
> > +     if (!ASSERT_EQ(info.perf_event.type, type, "perf_type_match"))
> > +             return -1;
> > +
> > +     switch (info.perf_event.type) {
> > +     case BPF_PERF_EVENT_KPROBE:
> > +     case BPF_PERF_EVENT_KRETPROBE:
> > +             ASSERT_EQ(info.perf_event.kprobe.offset, offset, "kprobe_offset");
> > +
> > +             /* In case kernel.kptr_restrict is not permitted or MAX_SYMS is reached */
> > +             if (addr)
> > +                     ASSERT_EQ(info.perf_event.kprobe.addr, addr + entry_offset,
> > +                               "kprobe_addr");
> > +
> > +             if (!info.perf_event.kprobe.func_name) {
> > +                     ASSERT_EQ(info.perf_event.kprobe.name_len, 0, "name_len");
> > +                     info.perf_event.kprobe.func_name = ptr_to_u64(&buf);
> > +                     info.perf_event.kprobe.name_len = sizeof(buf);
> > +                     goto again;
> > +             }
> > +
> > +             err = strncmp(u64_to_ptr(info.perf_event.kprobe.func_name), KPROBE_FUNC,
> > +                           strlen(KPROBE_FUNC));
> > +             ASSERT_EQ(err, 0, "cmp_kprobe_func_name");
> > +             break;
> > +     case BPF_PERF_EVENT_TRACEPOINT:
> > +             if (!info.perf_event.tracepoint.tp_name) {
> > +                     ASSERT_EQ(info.perf_event.tracepoint.name_len, 0, "name_len");
> > +                     info.perf_event.tracepoint.tp_name = ptr_to_u64(&buf);
> > +                     info.perf_event.tracepoint.name_len = sizeof(buf);
> > +                     goto again;
> > +             }
> > +
> > +             err = strncmp(u64_to_ptr(info.perf_event.tracepoint.tp_name), TP_NAME,
> > +                           strlen(TP_NAME));
> > +             ASSERT_EQ(err, 0, "cmp_tp_name");
> > +             break;
> > +     case BPF_PERF_EVENT_UPROBE:
> > +     case BPF_PERF_EVENT_URETPROBE:
> > +             ASSERT_EQ(info.perf_event.uprobe.offset, offset, "uprobe_offset");
> > +
> > +             if (!info.perf_event.uprobe.file_name) {
> > +                     ASSERT_EQ(info.perf_event.uprobe.name_len, 0, "name_len");
> > +                     info.perf_event.uprobe.file_name = ptr_to_u64(&buf);
> > +                     info.perf_event.uprobe.name_len = sizeof(buf);
> > +                     goto again;
> > +             }
> > +
> > +             err = strncmp(u64_to_ptr(info.perf_event.uprobe.file_name), UPROBE_FILE,
> > +                           strlen(UPROBE_FILE));
> > +                     ASSERT_EQ(err, 0, "cmp_file_name");
> > +             break;
> > +     default:
>
> Is the 'default' case possible? Probably not, right? Then
> let us add
>                 err = -1;
> to indicate this path is not possible.

will add it.

>
> > +             break;
> > +     }
> > +     return err;
> > +}
> > +
> [...]
> > +
> > +static void test_kprobe_fill_link_info(struct test_fill_link_info *skel,
> > +                                    enum bpf_perf_event_type type,
> > +                                    bool retprobe, bool invalid)
> > +{
> > +     DECLARE_LIBBPF_OPTS(bpf_kprobe_opts, opts,
> > +             .attach_mode = PROBE_ATTACH_MODE_LINK,
> > +             .retprobe = retprobe,
> > +     );
> > +     ssize_t offset = 0, entry_offset = 0;
>
> Remove 'offset = 0'.
>
> > +     int link_fd, err;
> > +     long addr;
> > +
> > +     skel->links.kprobe_run = bpf_program__attach_kprobe_opts(skel->progs.kprobe_run,
> > +                                                              KPROBE_FUNC, &opts);
> > +     if (!ASSERT_OK_PTR(skel->links.kprobe_run, "attach_kprobe"))
> > +             return;
> > +
> > +     link_fd = bpf_link__fd(skel->links.kprobe_run);
> > +     addr = ksym_get_addr(KPROBE_FUNC);
> > +     if (!invalid) {
> > +             /* See also arch_adjust_kprobe_addr(). */
> > +             if (skel->kconfig->CONFIG_X86_KERNEL_IBT)
> > +                     entry_offset = 4;
> > +             err = verify_perf_link_info(link_fd, type, addr, offset, entry_offset);
>
> Replease 'offset' with '0'.

will do it.

>
> > +             ASSERT_OK(err, "verify_perf_link_info");
> > +     } else {
> > +             kprobe_fill_invalid_user_buffer(link_fd);
> > +     }
> > +     bpf_link__detach(skel->links.kprobe_run);
> > +}
> > +
> [...]
> > +
> > +static void test_kprobe_multi_fill_link_info(struct test_fill_link_info *skel,
> > +                                          bool retprobe, bool buffer)
> > +{
> > +     LIBBPF_OPTS(bpf_kprobe_multi_opts, opts);
> > +     const char *syms[KMULTI_CNT] = {
> > +             "schedule_timeout_interruptible",
> > +             "schedule_timeout_uninterruptible",
> > +             "schedule_timeout_idle",
> > +             "schedule_timeout_killable",
> > +     };
> > +     __u64 addrs[KMULTI_CNT];
> > +     int link_fd, i, err = 0;
>
> 'err = 0' => 'err'.

will change it.

>
> > +
> > +     qsort(syms, KMULTI_CNT, sizeof(syms[0]), symbols_cmp_r);
> > +     opts.syms = syms;
> > +     opts.cnt = KMULTI_CNT;
> > +     opts.retprobe = retprobe;
> > +     skel->links.kmulti_run = bpf_program__attach_kprobe_multi_opts(skel->progs.kmulti_run,
> > +                                                                    NULL, &opts);
> > +     if (!ASSERT_OK_PTR(skel->links.kmulti_run, "attach_kprobe_multi"))
> > +             return;
> > +
> > +     link_fd = bpf_link__fd(skel->links.kmulti_run);
> > +     for (i = 0; i < KMULTI_CNT; i++)
> > +             addrs[i] = ksym_get_addr(syms[i]);
> > +
> > +     if (!buffer)
> > +             err = verify_kmulti_link_info(link_fd, addrs, retprobe);
> > +     else
> > +             verify_kmulti_user_buffer(link_fd, addrs);
> > +     ASSERT_OK(err, "verify_kmulti_link_info");
>
> if (!buffer) {
>         err = verify_kmulti_link_info(link_fd, addrs, retprobe);
>         ASSERT_OK(err, "verify_kmulti_link_info");
> } else {
>         verify_kmulti_user_buffer(link_fd, addrs);
> }

That is better. Thanks for your review.

-- 
Regards
Yafang

Re: [PATCH v4 bpf-next 2/2] selftests/bpf: Add selftest for fill_link_info

[Jiri Olsa]

On Fri, Aug 04, 2023 at 10:57:32AM +0000, Yafang Shao wrote:

SNIP

> +
> +static void kprobe_fill_invalid_user_buffer(int fd)
> +{
> +	struct bpf_link_info info;
> +	__u32 len = sizeof(info);
> +	int err;
> +
> +	memset(&info, 0, sizeof(info));
> +
> +	info.perf_event.kprobe.func_name = 0x1; /* invalid address */
> +	err = bpf_link_get_info_by_fd(fd, &info, &len);
> +	ASSERT_EQ(err, -EINVAL, "invalid_buff_and_len");
> +
> +	info.perf_event.kprobe.name_len = 64;
> +	err = bpf_link_get_info_by_fd(fd, &info, &len);
> +	ASSERT_EQ(err, -EFAULT, "invalid_buff");
> +
> +	info.perf_event.kprobe.func_name = 0;
> +	err = bpf_link_get_info_by_fd(fd, &info, &len);
> +	ASSERT_EQ(err, -EINVAL, "invalid_len");
> +
> +	ASSERT_EQ(info.perf_event.kprobe.addr, 0, "func_addr");
> +	ASSERT_EQ(info.perf_event.kprobe.offset, 0, "func_offset");
> +	ASSERT_EQ(info.perf_event.type, 0, "type");
> +}
> +
> +static void test_kprobe_fill_link_info(struct test_fill_link_info *skel,
> +				       enum bpf_perf_event_type type,
> +				       bool retprobe, bool invalid)
> +{
> +	DECLARE_LIBBPF_OPTS(bpf_kprobe_opts, opts,
> +		.attach_mode = PROBE_ATTACH_MODE_LINK,
> +		.retprobe = retprobe,

you could got rid of the retprobe argument and just do

		.retprobe = retprobe == BPF_PERF_EVENT_KRETPROBE,


> +	);
> +	ssize_t offset = 0, entry_offset = 0;
> +	int link_fd, err;
> +	long addr;
> +
> +	skel->links.kprobe_run = bpf_program__attach_kprobe_opts(skel->progs.kprobe_run,
> +								 KPROBE_FUNC, &opts);
> +	if (!ASSERT_OK_PTR(skel->links.kprobe_run, "attach_kprobe"))
> +		return;
> +
> +	link_fd = bpf_link__fd(skel->links.kprobe_run);
> +	addr = ksym_get_addr(KPROBE_FUNC);
> +	if (!invalid) {
> +		/* See also arch_adjust_kprobe_addr(). */
> +		if (skel->kconfig->CONFIG_X86_KERNEL_IBT)
> +			entry_offset = 4;
> +		err = verify_perf_link_info(link_fd, type, addr, offset, entry_offset);
> +		ASSERT_OK(err, "verify_perf_link_info");
> +	} else {
> +		kprobe_fill_invalid_user_buffer(link_fd);
> +	}
> +	bpf_link__detach(skel->links.kprobe_run);
> +}
> +
> +static void test_tp_fill_link_info(struct test_fill_link_info *skel)
> +{
> +	int link_fd, err;
> +
> +	skel->links.tp_run = bpf_program__attach_tracepoint(skel->progs.tp_run, TP_CAT, TP_NAME);
> +	if (!ASSERT_OK_PTR(skel->links.tp_run, "attach_tp"))
> +		return;
> +
> +	link_fd = bpf_link__fd(skel->links.tp_run);
> +	err = verify_perf_link_info(link_fd, BPF_PERF_EVENT_TRACEPOINT, 0, 0, 0);
> +	ASSERT_OK(err, "verify_perf_link_info");
> +	bpf_link__detach(skel->links.tp_run);
> +}
> +
> +static void test_uprobe_fill_link_info(struct test_fill_link_info *skel,
> +				       enum bpf_perf_event_type type, ssize_t offset,
> +				       bool retprobe)
> +{
> +	int link_fd, err;
> +
> +	skel->links.uprobe_run = bpf_program__attach_uprobe(skel->progs.uprobe_run, retprobe,
> +							    0, /* self pid */
> +							    UPROBE_FILE, offset);

same here with 'type == BPF_PERF_EVENT_URETPROBE'


> +	if (!ASSERT_OK_PTR(skel->links.uprobe_run, "attach_uprobe"))
> +		return;
> +
> +	link_fd = bpf_link__fd(skel->links.uprobe_run);
> +	err = verify_perf_link_info(link_fd, type, 0, offset, 0);
> +	ASSERT_OK(err, "verify_perf_link_info");
> +	bpf_link__detach(skel->links.uprobe_run);
> +}
> +

SNIP

> +
> +static void test_kprobe_multi_fill_link_info(struct test_fill_link_info *skel,
> +					     bool retprobe, bool buffer)
> +{
> +	LIBBPF_OPTS(bpf_kprobe_multi_opts, opts);
> +	const char *syms[KMULTI_CNT] = {
> +		"schedule_timeout_interruptible",
> +		"schedule_timeout_uninterruptible",
> +		"schedule_timeout_idle",
> +		"schedule_timeout_killable",

nit, might be better to use some of the bpf_fentry_test[1-9] functions,
also for KPROBE_FUNC

> +	};
> +	__u64 addrs[KMULTI_CNT];
> +	int link_fd, i, err = 0;
> +
> +	qsort(syms, KMULTI_CNT, sizeof(syms[0]), symbols_cmp_r);
> +	opts.syms = syms;
> +	opts.cnt = KMULTI_CNT;
> +	opts.retprobe = retprobe;
> +	skel->links.kmulti_run = bpf_program__attach_kprobe_multi_opts(skel->progs.kmulti_run,
> +								       NULL, &opts);
> +	if (!ASSERT_OK_PTR(skel->links.kmulti_run, "attach_kprobe_multi"))
> +		return;
> +
> +	link_fd = bpf_link__fd(skel->links.kmulti_run);
> +	for (i = 0; i < KMULTI_CNT; i++)
> +		addrs[i] = ksym_get_addr(syms[i]);
> +
> +	if (!buffer)
> +		err = verify_kmulti_link_info(link_fd, addrs, retprobe);
> +	else
> +		verify_kmulti_user_buffer(link_fd, addrs);

verify_kmulti_user_buffer is actually what you call 'invalid' in other
tests right? seems better to keep it in here unless I miss something

thanks,
jirka

> +	ASSERT_OK(err, "verify_kmulti_link_info");
> +	bpf_link__detach(skel->links.kmulti_run);
> +}
> +

SNIP

Re: [PATCH v4 bpf-next 1/2] bpf: Fix uninitialized symbol in bpf_perf_link_fill_kprobe()

[Jiri Olsa]

On Fri, Aug 04, 2023 at 10:57:31AM +0000, Yafang Shao wrote:
> The patch 1b715e1b0ec5: "bpf: Support ->fill_link_info for
> perf_event" from Jul 9, 2023, leads to the following Smatch static
> checker warning:
> 
>     kernel/bpf/syscall.c:3416 bpf_perf_link_fill_kprobe()
>     error: uninitialized symbol 'type'.
> 
> That can happens when uname is NULL. So fix it by verifying the uname
> when we really need to fill it.
> 
> Fixes: 1b715e1b0ec5 ("bpf: Support ->fill_link_info for perf_event")
> Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
> Closes: https://lore.kernel.org/bpf/85697a7e-f897-4f74-8b43-82721bebc462@kili.mountain/
> Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> Acked-by: Yonghong Song <yonghong.song@linux.dev>

Acked-by: Jiri Olsa <jolsa@kernel.org>

jirka

> ---
>  kernel/bpf/syscall.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
> index 7f4e8c3..166390f 100644
> --- a/kernel/bpf/syscall.c
> +++ b/kernel/bpf/syscall.c
> @@ -3378,14 +3378,14 @@ static int bpf_perf_link_fill_common(const struct perf_event *event,
>  
>  	if (!ulen ^ !uname)
>  		return -EINVAL;
> -	if (!uname)
> -		return 0;
>  
>  	err = bpf_get_perf_event_info(event, &prog_id, fd_type, &buf,
>  				      probe_offset, probe_addr);
>  	if (err)
>  		return err;
>  
> +	if (!uname)
> +		return 0;
>  	if (buf) {
>  		len = strlen(buf);
>  		err = bpf_copy_to_user(uname, buf, ulen, len);
> -- 
> 1.8.3.1
> 

Re: [PATCH v4 bpf-next 2/2] selftests/bpf: Add selftest for fill_link_info

[Yafang Shao]

On Sun, Aug 6, 2023 at 3:58 AM Jiri Olsa <olsajiri@gmail.com> wrote:
>
> On Fri, Aug 04, 2023 at 10:57:32AM +0000, Yafang Shao wrote:
>
> SNIP
>
> > +
> > +static void kprobe_fill_invalid_user_buffer(int fd)
> > +{
> > +     struct bpf_link_info info;
> > +     __u32 len = sizeof(info);
> > +     int err;
> > +
> > +     memset(&info, 0, sizeof(info));
> > +
> > +     info.perf_event.kprobe.func_name = 0x1; /* invalid address */
> > +     err = bpf_link_get_info_by_fd(fd, &info, &len);
> > +     ASSERT_EQ(err, -EINVAL, "invalid_buff_and_len");
> > +
> > +     info.perf_event.kprobe.name_len = 64;
> > +     err = bpf_link_get_info_by_fd(fd, &info, &len);
> > +     ASSERT_EQ(err, -EFAULT, "invalid_buff");
> > +
> > +     info.perf_event.kprobe.func_name = 0;
> > +     err = bpf_link_get_info_by_fd(fd, &info, &len);
> > +     ASSERT_EQ(err, -EINVAL, "invalid_len");
> > +
> > +     ASSERT_EQ(info.perf_event.kprobe.addr, 0, "func_addr");
> > +     ASSERT_EQ(info.perf_event.kprobe.offset, 0, "func_offset");
> > +     ASSERT_EQ(info.perf_event.type, 0, "type");
> > +}
> > +
> > +static void test_kprobe_fill_link_info(struct test_fill_link_info *skel,
> > +                                    enum bpf_perf_event_type type,
> > +                                    bool retprobe, bool invalid)
> > +{
> > +     DECLARE_LIBBPF_OPTS(bpf_kprobe_opts, opts,
> > +             .attach_mode = PROBE_ATTACH_MODE_LINK,
> > +             .retprobe = retprobe,
>
> you could got rid of the retprobe argument and just do
>
>                 .retprobe = retprobe == BPF_PERF_EVENT_KRETPROBE,

good point. will change it and the others.

>
>
> > +     );
> > +     ssize_t offset = 0, entry_offset = 0;
> > +     int link_fd, err;
> > +     long addr;
> > +
> > +     skel->links.kprobe_run = bpf_program__attach_kprobe_opts(skel->progs.kprobe_run,
> > +                                                              KPROBE_FUNC, &opts);
> > +     if (!ASSERT_OK_PTR(skel->links.kprobe_run, "attach_kprobe"))
> > +             return;
> > +
> > +     link_fd = bpf_link__fd(skel->links.kprobe_run);
> > +     addr = ksym_get_addr(KPROBE_FUNC);
> > +     if (!invalid) {
> > +             /* See also arch_adjust_kprobe_addr(). */
> > +             if (skel->kconfig->CONFIG_X86_KERNEL_IBT)
> > +                     entry_offset = 4;
> > +             err = verify_perf_link_info(link_fd, type, addr, offset, entry_offset);
> > +             ASSERT_OK(err, "verify_perf_link_info");
> > +     } else {
> > +             kprobe_fill_invalid_user_buffer(link_fd);
> > +     }
> > +     bpf_link__detach(skel->links.kprobe_run);
> > +}
> > +
> > +static void test_tp_fill_link_info(struct test_fill_link_info *skel)
> > +{
> > +     int link_fd, err;
> > +
> > +     skel->links.tp_run = bpf_program__attach_tracepoint(skel->progs.tp_run, TP_CAT, TP_NAME);
> > +     if (!ASSERT_OK_PTR(skel->links.tp_run, "attach_tp"))
> > +             return;
> > +
> > +     link_fd = bpf_link__fd(skel->links.tp_run);
> > +     err = verify_perf_link_info(link_fd, BPF_PERF_EVENT_TRACEPOINT, 0, 0, 0);
> > +     ASSERT_OK(err, "verify_perf_link_info");
> > +     bpf_link__detach(skel->links.tp_run);
> > +}
> > +
> > +static void test_uprobe_fill_link_info(struct test_fill_link_info *skel,
> > +                                    enum bpf_perf_event_type type, ssize_t offset,
> > +                                    bool retprobe)
> > +{
> > +     int link_fd, err;
> > +
> > +     skel->links.uprobe_run = bpf_program__attach_uprobe(skel->progs.uprobe_run, retprobe,
> > +                                                         0, /* self pid */
> > +                                                         UPROBE_FILE, offset);
>
> same here with 'type == BPF_PERF_EVENT_URETPROBE'
>
>
> > +     if (!ASSERT_OK_PTR(skel->links.uprobe_run, "attach_uprobe"))
> > +             return;
> > +
> > +     link_fd = bpf_link__fd(skel->links.uprobe_run);
> > +     err = verify_perf_link_info(link_fd, type, 0, offset, 0);
> > +     ASSERT_OK(err, "verify_perf_link_info");
> > +     bpf_link__detach(skel->links.uprobe_run);
> > +}
> > +
>
> SNIP
>
> > +
> > +static void test_kprobe_multi_fill_link_info(struct test_fill_link_info *skel,
> > +                                          bool retprobe, bool buffer)
> > +{
> > +     LIBBPF_OPTS(bpf_kprobe_multi_opts, opts);
> > +     const char *syms[KMULTI_CNT] = {
> > +             "schedule_timeout_interruptible",
> > +             "schedule_timeout_uninterruptible",
> > +             "schedule_timeout_idle",
> > +             "schedule_timeout_killable",
>
> nit, might be better to use some of the bpf_fentry_test[1-9] functions,
> also for KPROBE_FUNC

will use them instead.

>
> > +     };
> > +     __u64 addrs[KMULTI_CNT];
> > +     int link_fd, i, err = 0;
> > +
> > +     qsort(syms, KMULTI_CNT, sizeof(syms[0]), symbols_cmp_r);
> > +     opts.syms = syms;
> > +     opts.cnt = KMULTI_CNT;
> > +     opts.retprobe = retprobe;
> > +     skel->links.kmulti_run = bpf_program__attach_kprobe_multi_opts(skel->progs.kmulti_run,
> > +                                                                    NULL, &opts);
> > +     if (!ASSERT_OK_PTR(skel->links.kmulti_run, "attach_kprobe_multi"))
> > +             return;
> > +
> > +     link_fd = bpf_link__fd(skel->links.kmulti_run);
> > +     for (i = 0; i < KMULTI_CNT; i++)
> > +             addrs[i] = ksym_get_addr(syms[i]);
> > +
> > +     if (!buffer)
> > +             err = verify_kmulti_link_info(link_fd, addrs, retprobe);
> > +     else
> > +             verify_kmulti_user_buffer(link_fd, addrs);
>
> verify_kmulti_user_buffer is actually what you call 'invalid' in other
> tests right? seems better to keep it in here unless I miss something

will rename it.

-- 
Regards
Yafang