[PATCH bpf-next] selftests/bpf: make sure signal interrupts BPF_PROG_TEST_RUN

[PATCH bpf-next] selftests/bpf: make sure signal interrupts BPF_PROG_TEST_RUN

[Stanislav Fomichev]

Simple test that I used to reproduce the issue in the previous commit:
Do BPF_PROG_TEST_RUN with max iterations, each program is 4096 simple
move instructions. File alarm in 0.1 second and check that
bpf_prog_test_run is interrupted (i.e. test doesn't hang).

Note: reposting this for bpf-next to avoid linux-next conflict. In this
version I test both BPF_PROG_TYPE_SOCKET_FILTER (which uses generic
bpf_test_run implementation) and BPF_PROG_TYPE_FLOW_DISSECTOR (which has
it own loop with preempt handling in bpf_prog_test_run_flow_dissector).

Signed-off-by: Stanislav Fomichev <sdf@google.com>
---
 tools/testing/selftests/bpf/test_progs.c | 44 ++++++++++++++++++++++++
 1 file changed, 44 insertions(+)

diff --git a/tools/testing/selftests/bpf/test_progs.c b/tools/testing/selftests/bpf/test_progs.c
index c52bd90fbb34..c59d2e015d16 100644
--- a/tools/testing/selftests/bpf/test_progs.c
+++ b/tools/testing/selftests/bpf/test_progs.c
@@ -12,6 +12,7 @@
 #include <stdlib.h>
 #include <stdarg.h>
 #include <time.h>
+#include <signal.h>
 
 #include <linux/types.h>
 typedef __u16 __sum16;
@@ -28,6 +29,7 @@ typedef __u16 __sum16;
 #include <sys/ioctl.h>
 #include <sys/wait.h>
 #include <sys/types.h>
+#include <sys/time.h>
 #include <fcntl.h>
 #include <pthread.h>
 #include <linux/bpf.h>
@@ -2108,6 +2110,46 @@ static void test_map_lock(void)
 	bpf_object__close(obj);
 }
 
+static void sigalrm_handler(int s) {}
+static struct sigaction sigalrm_action = {
+	.sa_handler = sigalrm_handler,
+};
+
+static void test_signal_pending(enum bpf_prog_type prog_type)
+{
+	struct bpf_insn prog[4096];
+	struct itimerval timeo = {
+		.it_value.tv_usec = 100000, /* 100ms */
+	};
+	__u32 duration, retval;
+	int prog_fd;
+	int err;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(prog); i++)
+		prog[i] = BPF_ALU64_IMM(BPF_MOV, BPF_REG_0, 0);
+	prog[ARRAY_SIZE(prog) - 1] = BPF_EXIT_INSN();
+
+	prog_fd = bpf_load_program(prog_type, prog, ARRAY_SIZE(prog),
+				   "GPL", 0, NULL, 0);
+	CHECK(prog_fd < 0, "test-run", "errno %d\n", errno);
+
+	err = sigaction(SIGALRM, &sigalrm_action, NULL);
+	CHECK(err, "test-run-signal-sigaction", "errno %d\n", errno);
+
+	err = setitimer(ITIMER_REAL, &timeo, NULL);
+	CHECK(err, "test-run-signal-timer", "errno %d\n", errno);
+
+	err = bpf_prog_test_run(prog_fd, 0xffffffff, &pkt_v4, sizeof(pkt_v4),
+				NULL, NULL, &retval, &duration);
+	CHECK(duration > 500000000, /* 500ms */
+	      "test-run-signal-duration",
+	      "duration %dns > 500ms\n",
+	      duration);
+
+	signal(SIGALRM, SIG_DFL);
+}
+
 int main(void)
 {
 	srand(time(NULL));
@@ -2138,6 +2180,8 @@ int main(void)
 	test_flow_dissector();
 	test_spinlock();
 	test_map_lock();
+	test_signal_pending(BPF_PROG_TYPE_SOCKET_FILTER);
+	test_signal_pending(BPF_PROG_TYPE_FLOW_DISSECTOR);
 
 	printf("Summary: %d PASSED, %d FAILED\n", pass_cnt, error_cnt);
 	return error_cnt ? EXIT_FAILURE : EXIT_SUCCESS;
-- 
2.21.0.rc0.258.g878e2cd30e-goog

Re: [PATCH bpf-next] selftests/bpf: make sure signal interrupts BPF_PROG_TEST_RUN

[Daniel Borkmann]

On 02/21/2019 08:11 PM, Stanislav Fomichev wrote:
> Simple test that I used to reproduce the issue in the previous commit:
> Do BPF_PROG_TEST_RUN with max iterations, each program is 4096 simple
> move instructions. File alarm in 0.1 second and check that
> bpf_prog_test_run is interrupted (i.e. test doesn't hang).
> 
> Note: reposting this for bpf-next to avoid linux-next conflict. In this
> version I test both BPF_PROG_TYPE_SOCKET_FILTER (which uses generic
> bpf_test_run implementation) and BPF_PROG_TYPE_FLOW_DISSECTOR (which has
> it own loop with preempt handling in bpf_prog_test_run_flow_dissector).
> 
> Signed-off-by: Stanislav Fomichev <sdf@google.com>

Applied, thanks!