Re: [PATCH] libbpf: check the btf_type kind to prevent error

[Eduard Zingerman <eddyz87@gmail.com>]

On Tue, 2024-08-06 at 10:38 -0700, Yonghong Song wrote:
> On 8/6/24 3:51 AM, Ma Ke wrote:
> > To prevent potential error return values, it is necessary to check the
> > return value of btf__type_by_id. We can add a kind checking to fix the
> > issue.
> > 
> > Cc: stable@vger.kernel.org
> > Fixes: 430025e5dca5 ("libbpf: Add subskeleton scaffolding")
> > Signed-off-by: Ma Ke <make24@iscas.ac.cn>
> > ---
> >   tools/lib/bpf/libbpf.c | 3 +++
> >   1 file changed, 3 insertions(+)
> > 
> > diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> > index a3be6f8fac09..d1eb45d16054 100644
> > --- a/tools/lib/bpf/libbpf.c
> > +++ b/tools/lib/bpf/libbpf.c
> > @@ -13850,6 +13850,9 @@ int bpf_object__open_subskeleton(struct bpf_object_subskeleton *s)
> >   		var = btf_var_secinfos(map_type);
> >   		for (i = 0; i < len; i++, var++) {
> >   			var_type = btf__type_by_id(btf, var->type);
> > +			if (!var_type)
> > +				return libbpf_err(-ENOENT);
> 
> Could you give a detailed example when this error could be triggered?

I'm curious as well, tools/lib/bpf/btf.c:btf_validate_type() has the following code:

static int btf_validate_type(const struct btf *btf, const struct btf_type *t, __u32 id)
{
	...
	switch (kind) {
	...
	case ...
	case BTF_KIND_VAR:
	case ...
		err = btf_validate_id(btf, t->type, id);
		if (err)
			return err;
		break;

That should catch situations exactly like this one.

> 
> > +
> >   			var_name = btf__name_by_offset(btf, var_type->name_off);
> >   			if (strcmp(var_name, var_skel->name) == 0) {
> >   				*var_skel->addr = map->mmaped + var->offset;