Add a benchmark to compare performance of
1) uprobe;
2) user program w/o args;
3) user program w/ args;
4) user program w/ args on random cpu.
Sample output:
./test_progs -t uprobe_vs_user_prog -v
test_uprobe_vs_user_prog:PASS:uprobe_vs_user_prog__open_and_load 0 nsec
test_uprobe_vs_user_prog:PASS:get_base_addr 0 nsec
test_uprobe_vs_user_prog:PASS:attach_uprobe 0 nsec
run_perf_test:PASS:uprobe 0 nsec
Each uprobe uses 1419 nanoseconds
run_perf_test:PASS:user_prog_no_args 0 nsec
Each user_prog_no_args uses 313 nanoseconds
run_perf_test:PASS:user_prog_with_args 0 nsec
Each user_prog_with_args uses 335 nanoseconds
run_perf_test:PASS:user_prog_with_args_on_cpu 0 nsec
Each user_prog_with_args_on_cpu uses 2821 nanoseconds
Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED
Signed-off-by: Song Liu <songliubrav...@fb.com>
---
.../bpf/prog_tests/uprobe_vs_user_prog.c | 101 ++++++++++++++++++
.../selftests/bpf/progs/uprobe_vs_user_prog.c | 21 ++++
2 files changed, 122 insertions(+)
create mode 100644 tools/testing/selftests/bpf/prog_tests/uprobe_vs_user_prog.c
create mode 100644 tools/testing/selftests/bpf/progs/uprobe_vs_user_prog.c
diff --git a/tools/testing/selftests/bpf/prog_tests/uprobe_vs_user_prog.c
b/tools/testing/selftests/bpf/prog_tests/uprobe_vs_user_prog.c
new file mode 100644
index 0000000000000..dadd7b56e69ec
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/uprobe_vs_user_prog.c
@@ -0,0 +1,101 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <test_progs.h>
+#include "uprobe_vs_user_prog.skel.h"
+
+#define REPEAT_CNT 10000ULL
+
+static int duration;
+
+static noinline void uprobe_target(void)
+{
+ asm ("");
+}
+
+struct bpf_prog_test_run_attr attr;
+
+static void call_user_prog(void)
+{
+ bpf_prog_test_run_xattr(&attr);
+}
+
+static int numcpu;
+
+static void call_user_prog_on_cpu(void)
+{
+ static int cpu = 0;
+
+ attr.cpu_plus = cpu + 1;
+ bpf_prog_test_run_xattr(&attr);
+ cpu = (cpu + 1) % numcpu;
+}
+
+typedef void (__run_func)(void);
+
+static void run_perf_test(struct uprobe_vs_user_prog *skel,
+ __run_func func, const char *name)
+{
+ __u64 start_time, total_time;
+ int i;
+
+ skel->bss->sum = 0;
+
+ start_time = time_get_ns();
+ for (i = 0; i < REPEAT_CNT; i++)
+ func();
+ total_time = time_get_ns() - start_time;
+
+ CHECK(skel->bss->sum != REPEAT_CNT, name,
+ "missed %llu times\n", REPEAT_CNT - skel->bss->sum);
+ printf("Each %s uses %llu nanoseconds\n", name, total_time /
REPEAT_CNT);
+}
+
+void test_uprobe_vs_user_prog(void)
+{
+ struct bpf_user_prog_args args = {};
+ struct uprobe_vs_user_prog *skel;
+ struct bpf_link *uprobe_link;
+ size_t uprobe_offset;
+ ssize_t base_addr;
+
+ skel = uprobe_vs_user_prog__open_and_load();
+
+ if (CHECK(!skel, "uprobe_vs_user_prog__open_and_load",
+ "skeleton open_and_laod failed\n"))
+ return;
+
+ base_addr = get_base_addr();
+ if (CHECK(base_addr < 0, "get_base_addr",
+ "failed to find base addr: %zd", base_addr))
+ return;
+ uprobe_offset = (size_t)&uprobe_target - base_addr;
+ uprobe_link = bpf_program__attach_uprobe(skel->progs.handle_uprobe,
+ false /* retprobe */,
+ 0 /* self pid */,
+ "/proc/self/exe",
+ uprobe_offset);
+
+ if (CHECK(IS_ERR(uprobe_link), "attach_uprobe",
+ "err %ld\n", PTR_ERR(uprobe_link)))
+ goto cleanup;
+ skel->links.handle_uprobe = uprobe_link;
+
+ run_perf_test(skel, uprobe_target, "uprobe");
+
+ attr.prog_fd = bpf_program__fd(skel->progs.user_prog);
+ run_perf_test(skel, call_user_prog, "user_prog_no_args");
+
+ attr.data_size_in = sizeof(args);
+ attr.data_in = &args;
+ run_perf_test(skel, call_user_prog, "user_prog_with_args");
+
+ numcpu = libbpf_num_possible_cpus();
+
+ if (numcpu <= 0)
+ goto cleanup;
+
+ run_perf_test(skel, call_user_prog_on_cpu,
+ "user_prog_with_args_on_cpu");
+
+cleanup:
+ uprobe_vs_user_prog__destroy(skel);
+}
diff --git a/tools/testing/selftests/bpf/progs/uprobe_vs_user_prog.c
b/tools/testing/selftests/bpf/progs/uprobe_vs_user_prog.c
new file mode 100644
index 0000000000000..8b327b7cee30d
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/uprobe_vs_user_prog.c
@@ -0,0 +1,21 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2017 Facebook
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+volatile __u64 sum = 0;
+
+SEC("uprobe/func")
+int handle_uprobe(struct pt_regs *ctx)
+{
+ sum++;
+ return 0;
+}
+
+SEC("user")
+int user_prog(struct pt_regs *ctx)
+{
+ sum++;
+ return 0;
+}
--
2.24.1
