[v3,bpf-next,2/2] selftests/bpf: Add selftest for bits iter

Commit Message

Yafang Shao March 5, 2024, 6:43 a.m. UTC

Add selftests for the newly added bits iter.
- bits_iter_success
  - percpu data extracted from the percpu struct should be expected
  - RCU lock is not required
  - It is fine without calling bpf_iter_cpumask_next()
  - It can work as expected when invalid arguments are passed

- bits_iter_failure
  - bpf_iter_bits_destroy() is required after calling
    bpf_iter_bits_new()
  - bpf_iter_bits_destroy() can only destroy an initialized iter
  - bpf_iter_bits_next() must use an initialized iter

This test case can't work correctly on s390x for unknonw reason, thus it
is added to DENYLIST.s390x.

Signed-off-by: Yafang Shao <[e-mail]>
---
 tools/testing/selftests/bpf/DENYLIST.s390x    |   3 +-
 .../selftests/bpf/prog_tests/bits_iter.c      | 137 ++++++++++++++++++
 .../bpf/progs/test_bits_iter_failure.c        |  54 +++++++
 .../bpf/progs/test_bits_iter_success.c        | 122 ++++++++++++++++
 4 files changed, 315 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/bits_iter.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_bits_iter_failure.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_bits_iter_success.c

Comments

Andrii Nakryiko March 5, 2024, 10:05 p.m. UTC | #1

On Mon, Mar 4, 2024 at 10:44 PM Yafang Shao <[e-mail]> wrote:
>
> Add selftests for the newly added bits iter.
> - bits_iter_success
>   - percpu data extracted from the percpu struct should be expected
>   - RCU lock is not required
>   - It is fine without calling bpf_iter_cpumask_next()
>   - It can work as expected when invalid arguments are passed
>
> - bits_iter_failure
>   - bpf_iter_bits_destroy() is required after calling
>     bpf_iter_bits_new()
>   - bpf_iter_bits_destroy() can only destroy an initialized iter
>   - bpf_iter_bits_next() must use an initialized iter
>
> This test case can't work correctly on s390x for unknonw reason, thus it
> is added to DENYLIST.s390x.

That's an unusual way of handling "doesn't work for unknown reason" :)
This might be an endianness issue I pointed out in patch #1.

pw-bot: cr

>
> Signed-off-by: Yafang Shao <[e-mail]>
> ---
>  tools/testing/selftests/bpf/DENYLIST.s390x    |   3 +-
>  .../selftests/bpf/prog_tests/bits_iter.c      | 137 ++++++++++++++++++
>  .../bpf/progs/test_bits_iter_failure.c        |  54 +++++++
>  .../bpf/progs/test_bits_iter_success.c        | 122 ++++++++++++++++
>  4 files changed, 315 insertions(+), 1 deletion(-)
>  create mode 100644 tools/testing/selftests/bpf/prog_tests/bits_iter.c
>  create mode 100644 tools/testing/selftests/bpf/progs/test_bits_iter_failure.c
>  create mode 100644 tools/testing/selftests/bpf/progs/test_bits_iter_success.c
>
> diff --git a/tools/testing/selftests/bpf/DENYLIST.s390x b/tools/testing/selftests/bpf/DENYLIST.s390x
> index 1a63996c0304..0cd6d2bf1ff4 100644
> --- a/tools/testing/selftests/bpf/DENYLIST.s390x
> +++ b/tools/testing/selftests/bpf/DENYLIST.s390x
> @@ -1,5 +1,6 @@
>  # TEMPORARY
>  # Alphabetical order
> -exceptions                              # JIT does not support calling kfunc bpf_throw                                (exceptions)
> +bits_iter                                # cpumask iter can't work as expected                                         (?)
> +exceptions                               # JIT does not support calling kfunc bpf_throw                                (exceptions)
>  get_stack_raw_tp                         # user_stack corrupted user stack                                             (no backchain userspace)
>  stacktrace_build_id                      # compare_map_keys stackid_hmap vs. stackmap err -2 errno 2                   (?)

[...]

Yafang Shao March 6, 2024, 2:24 a.m. UTC | #2

On Wed, Mar 6, 2024 at 6:05 AM Andrii Nakryiko
<[e-mail]> wrote:
>
> On Mon, Mar 4, 2024 at 10:44 PM Yafang Shao <[e-mail]> wrote:
> >
> > Add selftests for the newly added bits iter.
> > - bits_iter_success
> >   - percpu data extracted from the percpu struct should be expected
> >   - RCU lock is not required
> >   - It is fine without calling bpf_iter_cpumask_next()
> >   - It can work as expected when invalid arguments are passed
> >
> > - bits_iter_failure
> >   - bpf_iter_bits_destroy() is required after calling
> >     bpf_iter_bits_new()
> >   - bpf_iter_bits_destroy() can only destroy an initialized iter
> >   - bpf_iter_bits_next() must use an initialized iter
> >
> > This test case can't work correctly on s390x for unknonw reason, thus it
> > is added to DENYLIST.s390x.
>
> That's an unusual way of handling "doesn't work for unknown reason" :)
> This might be an endianness issue I pointed out in patch #1.
>
> pw-bot: cr

Thanks for your reminder.
will check it.

Patch

diff --git a/tools/testing/selftests/bpf/DENYLIST.s390x b/tools/testing/selftests/bpf/DENYLIST.s390x
index 1a63996c0304..0cd6d2bf1ff4 100644
--- a/tools/testing/selftests/bpf/DENYLIST.s390x
+++ b/tools/testing/selftests/bpf/DENYLIST.s390x
@@ -1,5 +1,6 @@ 
 # TEMPORARY
 # Alphabetical order
-exceptions				 # JIT does not support calling kfunc bpf_throw				       (exceptions)
+bits_iter                                # cpumask iter can't work as expected                                         (?)
+exceptions                               # JIT does not support calling kfunc bpf_throw                                (exceptions)
 get_stack_raw_tp                         # user_stack corrupted user stack                                             (no backchain userspace)
 stacktrace_build_id                      # compare_map_keys stackid_hmap vs. stackmap err -2 errno 2                   (?)
diff --git a/tools/testing/selftests/bpf/prog_tests/bits_iter.c b/tools/testing/selftests/bpf/prog_tests/bits_iter.c
new file mode 100644
index 000000000000..ff4f921c91c9
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/bits_iter.c
@@ -0,0 +1,137 @@ 
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2024 Yafang Shao <[e-mail]> */
+
+#define _GNU_SOURCE
+#include <sched.h>
+
+#include <test_progs.h>
+#include "test_bits_iter_success.skel.h"
+#include "test_bits_iter_failure.skel.h"
+#include "cgroup_helpers.h"
+
+static const char * const positive_testcases[] = {
+	"cpumask_memalloc",
+	"cpumask_copy",
+};
+
+static const char * const negative_testcases[] = {
+	"null_pointer",
+	"zero_bit",
+	"no_mem",
+};
+
+static int read_percpu_data(struct bpf_link *link)
+{
+	int iter_fd, len;
+	char buf[128];
+	size_t left;
+	char *p;
+
+	iter_fd = bpf_iter_create(bpf_link__fd(link));
+	if (!ASSERT_GE(iter_fd, 0, "iter_fd"))
+		return -1;
+
+	memset(buf, 0, sizeof(buf));
+	left = ARRAY_SIZE(buf);
+	p = buf;
+	while ((len = read(iter_fd, p, left)) > 0) {
+		p += len;
+		left -= len;
+	}
+
+	close(iter_fd);
+	return 0;
+}
+
+static void verify_iter_success(const char *prog_name, bool negative)
+{
+	DECLARE_LIBBPF_OPTS(bpf_iter_attach_opts, opts);
+	struct test_bits_iter_success *skel;
+	union bpf_iter_link_info linfo;
+	int cgrp_fd, err, i, nr_cpus;
+	struct bpf_program *prog;
+	struct bpf_link *link;
+	cpu_set_t set;
+
+	if (setup_cgroup_environment())
+		return;
+
+	/* Utilize the cgroup iter */
+	cgrp_fd = get_root_cgroup();
+	if (!ASSERT_GE(cgrp_fd, 0, "create_cgrp"))
+		goto cleanup;
+
+	skel = test_bits_iter_success__open();
+	if (!ASSERT_OK_PTR(skel, "cpumask_iter_success__open"))
+		goto close_fd;
+
+	skel->bss->pid = getpid();
+
+	err = test_bits_iter_success__load(skel);
+	if (!ASSERT_OK(err, "cpumask_iter_success__load"))
+		goto destroy;
+
+	prog = bpf_object__find_program_by_name(skel->obj, prog_name);
+	if (!ASSERT_OK_PTR(prog, "bpf_object__find_program_by_name"))
+		goto destroy;
+
+	memset(&linfo, 0, sizeof(linfo));
+	linfo.cgroup.cgroup_fd = cgrp_fd;
+	linfo.cgroup.order = BPF_CGROUP_ITER_SELF_ONLY;
+	opts.link_info = &linfo;
+	opts.link_info_len = sizeof(linfo);
+	link = bpf_program__attach_iter(prog, &opts);
+	if (!ASSERT_OK_PTR(link, "bpf_program__attach"))
+		goto destroy;
+
+	if (negative)
+		goto negative;
+
+	CPU_ZERO(&set);
+	nr_cpus = libbpf_num_possible_cpus();
+	/* To guarantee the success of "cpumask_copy" at all times */
+	if (nr_cpus > 16)
+		nr_cpus = 16;
+	for (i = 0; i < nr_cpus; i++)
+		CPU_SET(i, &set);
+	err = sched_setaffinity(skel->bss->pid, sizeof(set), &set);
+	if (!ASSERT_OK(err, "setaffinity_all_cpus"))
+		goto free_link;
+	err = read_percpu_data(link);
+	if (!ASSERT_OK(err, "read_percpu_data"))
+		goto free_link;
+
+negative:
+	ASSERT_OK(skel->bss->err, "not_zero");
+
+free_link:
+	bpf_link__destroy(link);
+destroy:
+	test_bits_iter_success__destroy(skel);
+close_fd:
+	close(cgrp_fd);
+cleanup:
+	cleanup_cgroup_environment();
+}
+
+void test_bits_iter(void)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(positive_testcases); i++) {
+		if (!test__start_subtest(positive_testcases[i]))
+			continue;
+
+		verify_iter_success(positive_testcases[i], false);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(negative_testcases); i++) {
+		if (!test__start_subtest(negative_testcases[i]))
+			continue;
+
+		verify_iter_success(negative_testcases[i], true);
+	}
+
+	RUN_TESTS(test_bits_iter_success);
+	RUN_TESTS(test_bits_iter_failure);
+}
diff --git a/tools/testing/selftests/bpf/progs/test_bits_iter_failure.c b/tools/testing/selftests/bpf/progs/test_bits_iter_failure.c
new file mode 100644
index 000000000000..974d0b7a540e
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/test_bits_iter_failure.c
@@ -0,0 +1,54 @@ 
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (c) 2024 Yafang Shao <[e-mail]> */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+#include "bpf_misc.h"
+#include "task_kfunc_common.h"
+
+char _license[] SEC("license") = "GPL";
+
+int bpf_iter_bits_new(struct bpf_iter_bits *it, const void *unsafe_ptr__ign,
+		      u32 nr_bits) __ksym __weak;
+int *bpf_iter_bits_next(struct bpf_iter_bits *it) __ksym __weak;
+void bpf_iter_bits_destroy(struct bpf_iter_bits *it) __ksym __weak;
+
+SEC("iter.s/cgroup")
+__failure __msg("Unreleased reference id=3 alloc_insn=10")
+int BPF_PROG(no_destroy, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+	struct bpf_iter_bits it;
+	struct task_struct *p;
+
+	p = bpf_task_from_pid(1);
+	if (!p)
+		return 1;
+
+	bpf_iter_bits_new(&it, p->cpus_ptr, 8192);
+
+	bpf_iter_bits_next(&it);
+	bpf_task_release(p);
+	return 0;
+}
+
+SEC("iter/cgroup")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(next_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+	struct bpf_iter_bits *it = NULL;
+
+	bpf_iter_bits_next(it);
+	return 0;
+}
+
+SEC("iter/cgroup")
+__failure __msg("expected an initialized iter_bits as arg #1")
+int BPF_PROG(destroy_uninit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+	struct bpf_iter_bits it = {};
+
+	bpf_iter_bits_destroy(&it);
+	return 0;
+}
diff --git a/tools/testing/selftests/bpf/progs/test_bits_iter_success.c b/tools/testing/selftests/bpf/progs/test_bits_iter_success.c
new file mode 100644
index 000000000000..77081204dec3
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/test_bits_iter_success.c
@@ -0,0 +1,122 @@ 
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (c) 2024 Yafang Shao <[e-mail]> */
+
+#include "vmlinux.h"
+#include <linux/const.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+#include "task_kfunc_common.h"
+
+char _license[] SEC("license") = "GPL";
+
+extern const struct rq runqueues __ksym __weak;
+
+int bpf_iter_bits_new(struct bpf_iter_bits *it, const void *unsafe_ptr__ign,
+		      u32 nr_bits) __ksym __weak;
+int *bpf_iter_bits_next(struct bpf_iter_bits *it) __ksym __weak;
+void bpf_iter_bits_destroy(struct bpf_iter_bits *it) __ksym __weak;
+
+int pid, err;
+
+static int cpumask_iter(struct bpf_iter_meta *meta, struct cgroup *cgrp, u32 nr_cpus)
+{
+	struct task_struct *p;
+	u32 nr_running = 0;
+	struct rq *rq;
+	int *cpu;
+
+	/* epilogue */
+	if (!cgrp)
+		return 0;
+
+	p = bpf_task_from_pid(pid);
+	if (!p)
+		return 1;
+
+	bpf_for_each(bits, cpu, p->cpus_ptr, nr_cpus) {
+		rq = (struct rq *)bpf_per_cpu_ptr(&runqueues, *cpu);
+		/* Every valid CPU should possess a runqueue, even in the event of being offline */
+		if (!rq)
+			break;
+		nr_running += rq->nr_running;
+	}
+	if (nr_running == 0)
+		err = 1;
+	bpf_task_release(p);
+	return 0;
+}
+
+SEC("iter.s/cgroup")
+int BPF_PROG(cpumask_memalloc, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+	return cpumask_iter(meta, cgrp, 128);
+}
+
+SEC("iter.s/cgroup")
+int BPF_PROG(cpumask_copy, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+	return cpumask_iter(meta, cgrp, 16);
+}
+
+SEC("iter.s/cgroup")
+int BPF_PROG(null_pointer, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+	int *cpu;
+
+	bpf_for_each(bits, cpu, NULL, 8192)
+		err++;
+	return 0;
+}
+
+SEC("iter.s/cgroup")
+int BPF_PROG(zero_bit, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+	struct task_struct *p;
+	int *cpu;
+
+	p = bpf_task_from_pid(pid);
+	if (!p)
+		return 1;
+
+	bpf_for_each(bits, cpu, p->cpus_ptr, 0)
+		err++;
+	bpf_task_release(p);
+	return 0;
+}
+
+SEC("iter.s/cgroup")
+int BPF_PROG(no_mem, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+	struct task_struct *p;
+	int *cpu;
+
+	p = bpf_task_from_pid(pid);
+	if (!p)
+		return 1;
+
+	/* The max number of memalloc is 4096, thus it will fail to allocate (8192 * 8) */
+	bpf_for_each(bits, cpu, p->cpus_ptr, 8192 * 8)
+		err++;
+	bpf_task_release(p);
+	return 0;
+}
+
+SEC("iter.s/cgroup")
+int BPF_PROG(no_next, struct bpf_iter_meta *meta, struct cgroup *cgrp)
+{
+	struct bpf_iter_bits it;
+	struct task_struct *p;
+
+	p = bpf_task_from_pid(1);
+	if (!p)
+		return 1;
+
+	bpf_iter_bits_new(&it, p->cpus_ptr, 8192);
+
+	/* It functions properly without invoking bpf_iter_bits_next(). */
+
+	bpf_iter_bits_destroy(&it);
+	bpf_task_release(p);
+	return 0;
+}