Re: [PATCH] selftests/bpf: Add ring_buffer__consume_n test.

From: Andrea Righi
Date: Thu Apr 25 2024 - 02:19:25 EST


On Sun, Apr 21, 2024 at 10:11:33PM +0200, Jiri Olsa wrote:
..
> > static struct test_ringbuf_map_key_lskel *skel_map_key;
> > +static struct test_ringbuf_n_lskel *skel_n;
>
> seems like there's no need for this to be static variable

Can you elaborate more? I think we want these pointers to be static to
limit the scope to this file, no?

>
> > static struct test_ringbuf_lskel *skel;
> > static struct ring_buffer *ringbuf;
> >
> > @@ -326,6 +328,67 @@ static void ringbuf_subtest(void)
> > test_ringbuf_lskel__destroy(skel);
> > }
> >
> > +/*
> > + * Test ring_buffer__consume_n() by producing N_TOT_SAMPLES samples in the ring
> > + * buffer, via getpid(), and consuming them in chunks of N_SAMPLES.
> > + */
> > +#define N_TOT_SAMPLES 32
> > +#define N_SAMPLES 4
> > +
> > +/* Sample value to verify the callback validity */
> > +#define SAMPLE_VALUE 42L
> > +
> > +static int process_n_sample(void *ctx, void *data, size_t len)
> > +{
> > + struct sample *s = data;
> > +
> > + CHECK(s->value != SAMPLE_VALUE,
> > + "sample_value", "exp %ld, got %ld\n", SAMPLE_VALUE, s->value);
>
> I think we should use ASSERT macros instead in the new code

Good catch, I'll change this to an ASSERT_EQ().

>
> > +
> > + return 0;
> > +}
> > +
> > +static void ringbuf_n_subtest(void)
> > +{
> > + int err, i;
> > +
> > + skel_n = test_ringbuf_n_lskel__open();
> > + if (!ASSERT_OK_PTR(skel_n, "test_ringbuf_n_lskel__open"))
> > + return;
> > +
> > + skel_n->maps.ringbuf.max_entries = getpagesize();
> > + skel_n->bss->pid = getpid();
> > +
> > + err = test_ringbuf_n_lskel__load(skel_n);
> > + if (!ASSERT_OK(err, "test_ringbuf_n_lskel__load"))
> > + goto cleanup;
> > +
> > + ringbuf = ring_buffer__new(skel_n->maps.ringbuf.map_fd,
> > + process_n_sample, NULL, NULL);
> > + if (!ASSERT_OK_PTR(ringbuf, "ring_buffer__new"))
> > + goto cleanup;
> > +
> > + err = test_ringbuf_n_lskel__attach(skel_n);
> > + if (!ASSERT_OK(err, "test_ringbuf_n_lskel__attach"))
> > + goto cleanup_ringbuf;
> > +
> > + /* Produce N_TOT_SAMPLES samples in the ring buffer by calling getpid() */
> > + skel->bss->value = SAMPLE_VALUE;
>
> skel_n ?

Absolutely... I'm suprised that it works actually, I guess pure luck
(unluck) to reuse the old pointer and have value mapped to the same
location. Anyway, I'll fix this.

>
> > + 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");
> > + }
> > +
>
> SNIP
>
> > diff --git a/tools/testing/selftests/bpf/progs/test_ringbuf_n.c b/tools/testing/selftests/bpf/progs/test_ringbuf_n.c
> > new file mode 100644
> > index 000000000000..b98b5bb20699
> > --- /dev/null
> > +++ b/tools/testing/selftests/bpf/progs/test_ringbuf_n.c
> > @@ -0,0 +1,52 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +// Copyright (c) 2024 Andrea Righi <andrea.righi@xxxxxxxxxxxxx>
> > +
> > +#include <linux/bpf.h>
> > +#include <sched.h>
> > +#include <unistd.h>
> > +#include <bpf/bpf_helpers.h>
> > +#include "bpf_misc.h"
> > +
> > +char _license[] SEC("license") = "GPL";
> > +
> > +#define TASK_COMM_LEN 16
> > +
> > +struct sample {
> > + int pid;
> > + int seq;
>
> seq does not seem to be checked, is it needed?

seq is not used at all, I can definitely drop it.

Thanks for the review! I'll send a v2.

-Andrea