Re: [RFC PATCH bpf-next 2/2] selftests/bpf: Add bpf link iter test

[Yonghong Song <yhs@fb.com>]

On 4/22/22 11:22 AM, Dmitrii Dolgov wrote:
> Add a simple test for bpf link iterator
> 
> Signed-off-by: Dmitrii Dolgov <9erthalion6@gmail.com>
> ---
>   .../selftests/bpf/prog_tests/bpf_iter.c        | 15 +++++++++++++++
>   .../selftests/bpf/progs/bpf_iter_bpf_link.c    | 18 ++++++++++++++++++
>   2 files changed, 33 insertions(+)
>   create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_bpf_link.c
> 
> diff --git a/tools/testing/selftests/bpf/prog_tests/bpf_iter.c b/tools/testing/selftests/bpf/prog_tests/bpf_iter.c
> index 2c403ddc8076..e14a7a6d925c 100644
> --- a/tools/testing/selftests/bpf/prog_tests/bpf_iter.c
> +++ b/tools/testing/selftests/bpf/prog_tests/bpf_iter.c
> @@ -26,6 +26,7 @@
>   #include "bpf_iter_bpf_sk_storage_map.skel.h"
>   #include "bpf_iter_test_kern5.skel.h"
>   #include "bpf_iter_test_kern6.skel.h"
> +#include "bpf_iter_bpf_link.skel.h"
>   
>   static int duration;
>   
> @@ -1172,6 +1173,20 @@ static void test_buf_neg_offset(void)
>   		bpf_iter_test_kern6__destroy(skel);
>   }
>   
> +static void test_link_iter(void)

This function is used. Please add a proper subtest for this
in function test_bpf_iter().

> +{
> +	struct bpf_iter_bpf_link *skel;
> +
> +	skel = bpf_iter_bpf_link__open_and_load();
> +	if (CHECK(skel, "bpf_iter_bpf_link__open_and_load",
> +		  "skeleton open_and_load unexpected success\n"))
> +		return;

This is not correct. You should have CHECK(!skel, ...) to return
only if skel is NULL. The error message "skeleton open_and_load 
unexpected success\n" is not correct either. Probably a copy-paste
error.

Also, since you are working on this file, probably convert all
CHECK's in this file to ASSERT_*() macros as patch #2. Then
this patch itself can be patch #3 using ASSERT_*() as well.

> +
> +	do_dummy_read(skel->progs.dump_bpf_link);
> +
> +	bpf_iter_bpf_link__destroy(skel);
> +}
> +
>   #define CMP_BUFFER_SIZE 1024
>   static char task_vma_output[CMP_BUFFER_SIZE];
>   static char proc_maps_output[CMP_BUFFER_SIZE];
> diff --git a/tools/testing/selftests/bpf/progs/bpf_iter_bpf_link.c b/tools/testing/selftests/bpf/progs/bpf_iter_bpf_link.c
> new file mode 100644
> index 000000000000..a5041fa1cda9
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/progs/bpf_iter_bpf_link.c
> @@ -0,0 +1,18 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/* Copyright (c) 2020 Facebook */

copyright issue.

> +#include "bpf_iter.h"
> +#include <bpf/bpf_helpers.h>
> +
> +char _license[] SEC("license") = "GPL";
> +
> +SEC("iter/bpf_link")
> +int dump_bpf_link(struct bpf_iter__bpf_link *ctx)

Please put bpf_iter__bpf_link definition in bpf_iter.h so
the test can work with an old version of vmlinux.h.

> +{
> +	struct seq_file *seq = ctx->meta->seq;
> +	struct bpf_link *link = ctx->link;
> +	int link_id;

The 'link' pointer could be NULL as in previous patch
we have:

+	.ctx_arg_info		= {
+		{ offsetof(struct bpf_iter__bpf_link, link),
+		  PTR_TO_BTF_ID_OR_NULL },
+	},

So you need to add a check below.

	if (!link)
		return 0;

> +
> +	link_id = link->id;
> +	bpf_seq_write(seq, &link_id, sizeof(link_id));
> +	return 0;
> +}