@@ -1,8 +1,16 @@
-TEST_GEN_PROGS := seccomp_bpf
-CFLAGS += -Wl,-no-as-needed -Wall
-LDFLAGS += -lpthread
+all:
include ../lib.mk
-$(TEST_GEN_PROGS): seccomp_bpf.c ../kselftest_harness.h
- $(CC) $(CFLAGS) $(LDFLAGS) $< -o $@
+.PHONY: all clean
+
+BINARIES := seccomp_bpf seccomp_benchmark
+CFLAGS += -Wl,-no-as-needed -Wall
+
+seccomp_bpf: seccomp_bpf.c ../kselftest_harness.h
+ $(CC) $(CFLAGS) $(LDFLAGS) -lpthread $< -o $@
+
+TEST_PROGS += $(BINARIES)
+EXTRA_CLEAN := $(BINARIES)
+
+all: $(BINARIES)
new file mode 100644
@@ -0,0 +1,99 @@
+/*
+ * Strictly speaking, this is not a test. But it can report during test
+ * runs so relative performace can be measured.
+ */
+#define _GNU_SOURCE
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+#include <unistd.h>
+#include <linux/filter.h>
+#include <linux/seccomp.h>
+#include <sys/prctl.h>
+#include <sys/syscall.h>
+#include <sys/types.h>
+
+#define ARRAY_SIZE(a) (sizeof(a) / sizeof(a[0]))
+
+unsigned long long timing(clockid_t clk_id, unsigned long long samples)
+{
+ pid_t pid, ret;
+ unsigned long long i;
+ struct timespec start, finish;
+
+ pid = getpid();
+ assert(clock_gettime(clk_id, &start) == 0);
+ for (i = 0; i < samples; i++) {
+ ret = syscall(__NR_getpid);
+ assert(pid == ret);
+ }
+ assert(clock_gettime(clk_id, &finish) == 0);
+
+ i = finish.tv_sec - start.tv_sec;
+ i *= 1000000000;
+ i += finish.tv_nsec - start.tv_nsec;
+
+ printf("%lu.%09lu - %lu.%09lu = %llu\n",
+ finish.tv_sec, finish.tv_nsec,
+ start.tv_sec, start.tv_nsec,
+ i);
+
+ return i;
+}
+
+unsigned long long calibrate(void)
+{
+ unsigned long long i;
+
+ printf("Calibrating reasonable sample size...\n");
+
+ for (i = 5; ; i++) {
+ unsigned long long samples = 1 << i;
+
+ /* Find something that takes more than 5 seconds to run. */
+ if (timing(CLOCK_REALTIME, samples) / 1000000000ULL > 5)
+ return samples;
+ }
+}
+
+int main(int argc, char *argv[])
+{
+ struct sock_filter filter[] = {
+ BPF_STMT(BPF_RET|BPF_K, SECCOMP_RET_ALLOW),
+ };
+ struct sock_fprog prog = {
+ .len = (unsigned short)ARRAY_SIZE(filter),
+ .filter = filter,
+ };
+ long ret;
+ unsigned long long samples;
+ unsigned long long native, filtered;
+
+ if (argc > 1)
+ samples = strtoull(argv[1], NULL, 0);
+ else
+ samples = calibrate();
+
+ printf("Benchmarking %llu samples...\n", samples);
+
+ native = timing(CLOCK_PROCESS_CPUTIME_ID, samples) / samples;
+ printf("getpid native: %llu ns\n", native);
+
+ ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
+ assert(ret == 0);
+
+ ret = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, &prog);
+ assert(ret == 0);
+
+ filtered = timing(CLOCK_PROCESS_CPUTIME_ID, samples) / samples;
+ printf("getpid RET_ALLOW: %llu ns\n", filtered);
+
+ printf("Estimated seccomp overhead per syscall: %llu ns\n",
+ filtered - native);
+
+ if (filtered == native)
+ printf("Trying running again with more samples.\n");
+
+ return 0;
+}
This attempts to produce a comparison between native getpid() and a RET_ALLOW-filtered getpid(), to measure the overhead cost of using seccomp(). Signed-off-by: Kees Cook <keescook@chromium.org> --- As with the other, I'd like this to go via the seccomp tree. Getting an Ack would be great. :) Thanks! --- tools/testing/selftests/seccomp/Makefile | 18 ++-- .../testing/selftests/seccomp/seccomp_benchmark.c | 99 ++++++++++++++++++++++ 2 files changed, 112 insertions(+), 5 deletions(-) create mode 100644 tools/testing/selftests/seccomp/seccomp_benchmark.c