Re: [PATCH bpf-next v2 5/5] selftests/bpf: verify zext_dst annotations for various instructions

On Sat, 2026-08-01 at 01:33 +0000, [email protected] wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issu=
e(s) to consider:
> - [Medium] Missing bpf_testmod dependency check in test_verifier_zext, ca=
using spurious test failures when the module is not loaded.
> --
>=20
> --- Patch [5]: [PATCH bpf-next v2 5/5] selftests/bpf: verify zext_dst ann=
otations for various instructions ---
> commit 9e656f5e6b04a69526327df6f70d8c362d02d3d4
> Author: Eduard Zingerman <[email protected]>
>=20
> selftests/bpf: verify zext_dst annotations for various instructions
>=20
> This commit adds a suite of selftests to verify the verifier's zero exten=
sion
> behavior across various instructions and state pruning scenarios. It cove=
rs
> cases like 32-bit and 64-bit operations, atomic instructions, kfunc param=
eters,
> and CFG edge cases.
>=20
> > diff --git a/tools/testing/selftests/bpf/prog_tests/verifier.c b/tools/=
testing/selftests/bpf/prog_tests/verifier.c
> > index be97f6887f0e7..fba562c819691 100644
> > --- a/tools/testing/selftests/bpf/prog_tests/verifier.c
> > +++ b/tools/testing/selftests/bpf/prog_tests/verifier.c
> [ ... ]
> > @@ -277,6 +278,7 @@ void test_irq(void)			=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0 { RUN(irq); }
> > =C2=A0void test_verifier_mtu(void)		=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 { RU=
N(verifier_mtu); }
> > =C2=A0void test_verifier_jit_inline(void)=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 { RUN(verifier_jit_i=
nline); }
> > =C2=A0void test_verifier_ctx_ptr_param(void)=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0 { RUN(verifier_ctx_ptr_param); }
> > +void test_verifier_zext(void)=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 { RUN_TESTS(verif=
ier_zext); }
>=20
> [Severity: Medium]
> Since verifier_zext.c calls bpf_kfunc_call_test4 (an extern kfunc from
> bpf_testmod), will this test fail spuriously if bpf_testmod is not loaded=
?
>=20
> Should this check env.has_testmod and call test__skip() before RUN_TESTS
> to gracefully skip the test when the module is unavailable?

Idk, bpf_kfunc_call_test4() is used by the test progs/kfunc_call_test.c
prog_tests/kfunc_call.c, which itself does not check for env.has_testmod.
I don't think we enforce this rigorously, nor that we should.