Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220318021314.3225240-1-davidgow@google.com> <20220318021314.3225240-2-davidgow@google.com>
In-Reply-To: <20220318021314.3225240-2-davidgow@google.com>
From:   Daniel Latypov <dlatypov@google.com>
Date:   Wed, 4 May 2022 15:35:15 -0500
Message-ID: <CAGS_qxqm1ys1qUz__4uXWOgs=34M5MB3QMnhg40FHtRhatF3+g@mail.gmail.com>
Subject: Re: [RFC PATCH 1/2] kunit: Expose 'static stub' API to redirect functions
To:     David Gow <davidgow@google.com>
Cc:     Brendan Higgins <brendanhiggins@google.com>,
        Kees Cook <keescook@chromium.org>,
        Shuah Khan <skhan@linuxfoundation.org>,
        Steven Rostedt <rostedt@goodmis.org>,
        kunit-dev@googlegroups.com, linux-kselftest@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Thu, Mar 17, 2022 at 9:13 PM David Gow <davidgow@google.com> wrote:
> +#define kunit_activate_static_stub(test, real_fn_addr, replacement_addr)=
 do {  \
> +       typecheck(typeof(real_fn_addr), replacement_addr);               =
       \

We can't call this macro in the same scope for functions w/ different
signatures.

E.g. if we add this func to the example test
  static void other_func(void) {}
then trying to call kunit_activate_static_stub() on it in the same
test case, we get

./include/linux/typecheck.h:10:14: error: conflicting types for
=E2=80=98__dummy=E2=80=99; have =E2=80=98void(void)=E2=80=99
   10 | ({      type __dummy; \
      |              ^~~~~~~
./include/kunit/static_stub.h:99:9: note: in expansion of macro =E2=80=98ty=
pecheck=E2=80=99
   99 |         typecheck(typeof(real_fn_addr), replacement_addr);
                 \
      |         ^~~~~~~~~
lib/kunit/example-test.c:64:9: note: in expansion of macro
=E2=80=98kunit_activate_static_stub=E2=80=99
   64 |         kunit_activate_static_stub(test, other_func, other_func);
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~~
./include/linux/typecheck.h:10:14: note: previous declaration of
=E2=80=98__dummy=E2=80=99 with type =E2=80=98int(int)=E2=80=99
   10 | ({      type __dummy; \
      |              ^~~~~~~
./include/kunit/static_stub.h:99:9: note: in expansion of macro =E2=80=98ty=
pecheck=E2=80=99
   99 |         typecheck(typeof(real_fn_addr), replacement_addr);
                 \
      |         ^~~~~~~~~
lib/kunit/example-test.c:62:9: note: in expansion of macro
=E2=80=98kunit_activate_static_stub=E2=80=99
   62 |         kunit_activate_static_stub(test, add_one, subtract_one);
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~~

Afaict, the problem is that GCC thinks we're declaring a *function*
called __dummy, not a variable.
So it bleeds across the scope boundary of do-while unlike normal variables.

There's the typecheck_fn macro, but it doesn't work either.