Re: [PATCH bpf-next v12 2/3] bpftool: Add recursive map dumping

[Quentin Monnet <qmo@kernel.org>]

2026-11-11 16:19 UTC+0900 ~ Tianyi Chen <hi@tychen.cc>
> Dumping an array_of_maps or a hash_of_maps shows the IDs of the inner
> maps but not their contents. Add a recursive keyword to map dump that
> also dumps each inner map in the entry that references it. JSON gains
> an "inner_map" object next to "inner_map_id". Plain output separates
> inner maps with blank lines and indents their raw entries and element
> counts. Output without the keyword is unchanged.
> 
> The kernel permits only one map-in-map level. Look up outer entries
> individually, then open, dump and close each referenced inner map
> immediately. This uses one extra map FD; inner dumps may still use batch
> lookups. Shared inner maps are dumped at every referring entry. If an
> inner map has disappeared, report the error in that entry and continue.
> 
> Link: https://github.com/libbpf/bpftool/issues/58
> Suggested-by: Alexei Starovoitov <ast@kernel.org>
> Assisted-by: LLM
> Signed-off-by: Tianyi Chen <hi@tychen.cc>
> ---
>  .../bpf/bpftool/Documentation/bpftool-map.rst |  10 +-
>  tools/bpf/bpftool/bash-completion/bpftool     |   3 +
>  tools/bpf/bpftool/map.c                       | 123 +++++++++++++++---
>  3 files changed, 113 insertions(+), 23 deletions(-)
> 

[...]

> diff --git a/tools/bpf/bpftool/map.c b/tools/bpf/bpftool/map.c
> index 9e4dc3881060..79d33af257b6 100644
> --- a/tools/bpf/bpftool/map.c
> +++ b/tools/bpf/bpftool/map.c

[...]

> @@ -1003,6 +1036,43 @@ map_dump(int fd, struct bpf_map_info *info, json_writer_t *wtr,
>  	return err;
>  }
>  
> +/* The kernel has one level of map-in-map: inner maps have no inner maps. */
> +static void dump_inner_map(__u32 id)
> +{
> +	LIBBPF_OPTS(bpf_get_fd_by_id_opts, opts, .open_flags = BPF_F_RDONLY);
> +	struct bpf_map_info info = {};
> +	json_writer_t *btf_wtr = NULL;
> +	__u32 len = sizeof(info);
> +	int fd;
> +
> +	fd = bpf_map_get_fd_by_id_opts(id, &opts);
> +	if (fd < 0) {
> +		p_err("can't get map by id (%u): %s", id, strerror(errno));
> +		return;
> +	}
> +	if (bpf_map_get_info_by_fd(fd, &info, &len)) {
> +		p_err("can't get map info: %s", strerror(errno));
> +		goto out;
> +	}
> +
> +	if (!json_output && info.btf_value_type_id)
> +		btf_wtr = get_btf_writer();


This will not indent inner map entries for inner maps with BTF info,
will it? But I guess indenting them would require changing the JSON
writer, probably not worth it. It's probably OK if we have blank lines
to delimit the different maps in that case.

Acked-by: Quentin Monnet <qmo@kernel.org>

Thank you!

The series will need to be rebased once the batch lookups are merged, so
that it can go through CI.