Re: [PATCH bpf-next 1/2] bpf: Allow access to const void pointer arguments in tracing programs

[Kafai Wan]

On Mon, Apr 14, 2025 at 6:35 PM Jiri Olsa <olsajiri@xxxxxxxxx> wrote:
>
> On Sun, Apr 13, 2025 at 01:06:25AM +0800, KaFai Wan wrote:
> > Adding support to access arguments with const void pointer arguments
> > in tracing programs.
> >
> > Currently we allow tracing programs to access void pointers. If we try to
> > access argument which is pointer to const void like 2nd argument in kfree,
> > verifier will fail to load the program with;
> >
> > 0: R1=ctx() R10=fp0
> > ; asm volatile ("r2 = *(u64 *)(r1 + 8); ");
> > 0: (79) r2 = *(u64 *)(r1 +8)
> > func 'kfree' arg1 type UNKNOWN is not a struct
> >
> > Adding is_void_ptr to generic void  pointer check.
> >
> > Cc: Leon Hwang <leon.hwang@xxxxxxxxx>
> > Signed-off-by: KaFai Wan <kafai.wan@xxxxxxxxxxx>
> > ---
> >  kernel/bpf/btf.c | 10 +++++++++-
> >  1 file changed, 9 insertions(+), 1 deletion(-)
> >
> > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
> > index 16ba36f34dfa..e11d3afd0562 100644
> > --- a/kernel/bpf/btf.c
> > +++ b/kernel/bpf/btf.c
> > @@ -6383,6 +6383,14 @@ struct btf *bpf_prog_get_target_btf(const struct bpf_prog *prog)
> >               return prog->aux->attach_btf;
> >  }
> >
> > +static bool is_void_ptr(struct btf *btf, const struct btf_type *t)
> > +{
> > +     /* skip modifiers */
> > +     t = btf_type_skip_modifiers(btf, t->type, NULL);
> > +
> > +     return t->type == 0;
>
> I think you can use btf_type_is_void in here
>

Yes, I will use btf_type_is_void.

> > +}
> > +
> >  static bool is_int_ptr(struct btf *btf, const struct btf_type *t)
> >  {
> >       /* skip modifiers */
> > @@ -6776,7 +6784,7 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
> >               }
> >       }
> >
> > -     if (t->type == 0)
> > +     if (is_void_ptr(btf, t))
>
> lgtm,
>
> nit, the is_void_ptr name suggest there's also ptr check in the helper function,
> which is not the case. I understand it follows is_int_ptr name, but perhaps we
> could rename both helpers to is_void and is_int ... feel free to ignore ;-)

you are right, I will rename it. but I'm not sure if it's possible to merge
these two functions into one like is_void_or_int, they both skip modifiers.

>
> jirka
>
>
> >               /* This is a pointer to void.
> >                * It is the same as scalar from the verifier safety pov.
> >                * No further pointer walking is allowed.
> > --
> > 2.43.0
> >

thanks,
kafai