On Wed, Dec 13, 2023 at 11:51 PM Wentao Zhang
<wentao.zhang@xxxxxxxxxxxxx> wrote:
>
> The function btf_str_by_offset may return NULL when used as an
> input argument for btf_add_str in the context of btf_rewrite_str.
> The added check ensures that both the input string (s) and the
> BTF object (btf) are non-null before proceeding with the function
> logic. If either is null, the function returns an error code
> indicating an invalid argument.
>
> Found by our static analysis tool.
>
> Signed-off-by: Wentao Zhang <wentao.zhang@xxxxxxxxxxxxx>
> ---
> tools/lib/bpf/btf.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c
> index fd2309512978..a6a00bdc7151 100644
> --- a/tools/lib/bpf/btf.c
> +++ b/tools/lib/bpf/btf.c
> @@ -1612,6 +1612,8 @@ int btf__find_str(struct btf *btf, const char *s)
> int btf__add_str(struct btf *btf, const char *s)
> {
> int off;
> + if(!s || !btf)
> + return libbpf_err(-EINVAL);
>
while for some old libbpf code we sometimes did NULL validation,
generally speaking for all new code we assume that passed objects are
non-NULL, unless otherwise explicitly pointed out in API for cases
where NULL is actually meaningful (like no options are specified,
etc).
In this case, both btf and s are expected to be non-NULL by the API
and NULL values for either btf or s are meaningless. So I don't think
we need to check these conditions.
> if (btf->base_btf) {
> off = btf__find_str(btf->base_btf, s);
> --
> 2.35.5
>
>
