On Thu, Apr 25, 2024 at 7:06 AM Andrea Righi andrea.righi@canonical.com wrote:
Add a testcase for the ring_buffer__consume_n() API.
The test produces multiple samples in a ring buffer, using a sys_getpid() fentry prog, and consumes them from user-space in batches, rather than consuming all of them greedily, like ring_buffer__consume() does.
Link: https://lore.kernel.org/lkml/CAEf4BzaR4zqUpDmj44KNLdpJ=Tpa97GrvzuzVNO5nM6b7o... Signed-off-by: Andrea Righi andrea.righi@canonical.com
tools/testing/selftests/bpf/Makefile | 2 +- .../selftests/bpf/prog_tests/ringbuf.c | 64 +++++++++++++++++++ .../selftests/bpf/progs/test_ringbuf_n.c | 47 ++++++++++++++ 3 files changed, 112 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/bpf/progs/test_ringbuf_n.c
ChangeLog v2 -> v3:
- move skel_n inside ringbuf_n_subtest()
ChangeLog v1 -> v2:
- replace CHECK() with ASSERT_EQ()
- fix skel -> skel_n
- drop unused "seq" field from struct sample
[...]
/* Produce N_TOT_SAMPLES samples in the ring buffer by calling getpid() */skel_n->bss->value = SAMPLE_VALUE;for (i = 0; i < N_TOT_SAMPLES; i++)syscall(__NR_getpgid);/* Consume all samples from the ring buffer in batches of N_SAMPLES */for (i = 0; i < N_TOT_SAMPLES; i += err) {err = ring_buffer__consume_n(ringbuf, N_SAMPLES);ASSERT_EQ(err, N_SAMPLES, "rb_consume");
if something goes wrong and err is < 0, we might end up with a very long loop. I changed this to:
if (!ASSERT_EQ(...)) goto cleanup_ringbuf;
to avoid this problem
}+cleanup_ringbuf:
ring_buffer__free(ringbuf);+cleanup:
test_ringbuf_n_lskel__destroy(skel_n);+}
[...]