Re: [PATCH 5/5] selftests/bpf: make cfi_stubs globals const

[Caleb Sander Mateos]

On Wed, Dec 31, 2025 at 10:04 AM <bot+bpf-ci@xxxxxxxxxx> wrote:
>
> > diff --git a/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c b/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c
> > index 90c4b1a51de6..5e460b1dbdb6 100644
> > --- a/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c
> > +++ b/tools/testing/selftests/bpf/test_kmods/bpf_testmod.c
>
> [ ... ]
>
> > @@ -1275,7 +1275,7 @@ bpf_testmod_ops__test_return_ref_kptr(int dummy, struct task_struct *task__ref,
> >       return NULL;
> >  }
> >
> > -static struct bpf_testmod_ops __bpf_testmod_ops = {
> > +static const struct bpf_testmod_ops __bpf_testmod_ops = {
> >       .test_1 = bpf_testmod_test_1,
> >       .test_2 = bpf_testmod_test_2,
>
> Is it safe to make __bpf_testmod_ops const here? In bpf_testmod_init(),
> this struct is modified at runtime:
>
>     tramp = (void **)&__bpf_testmod_ops.tramp_1;
>     while (tramp <= (void **)&__bpf_testmod_ops.tramp_40)
>         *tramp++ = bpf_testmod_tramp;
>
> Writing to a const-qualified object is undefined behavior and may cause a
> protection fault when the compiler places this in read-only memory. Would
> the module fail to load on systems where .rodata is actually read-only?

Yup, that's indeed the bug caught by KASAN. Missed this mutation at
init time, I'll leave __bpf_testmod_ops as mutable.

Thanks,
Caleb

>
>
> ---
> AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
> See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md
>
> CI run summary: https://github.com/kernel-patches/bpf/actions/runs/20624206229