Re: [PATCH 01/10] x86/mm/numa: Remove uninitialized_var() usage

From: Kees Cook <keescook@chromium.org>
To: Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>,
	linux-kernel <linux-kernel@vger.kernel.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Alexander Potapenko <glider@google.com>,
	Joe Perches <joe@perches.com>, Andy Whitcroft <apw@canonical.com>,
	"maintainer:X86 ARCHITECTURE (32-BIT AND 64-BIT)"
	<x86@kernel.org>,
	drbd-dev@lists.linbit.com, linux-block@vger.kernel.org,
	b43-dev@lists.infradead.org,
	Network Development <netdev@vger.kernel.org>,
	linux-wireless <linux-wireless@vger.kernel.org>,
	linux-ide@vger.kernel.org, linux-clk@vger.kernel.org,
	linux-spi@vger.kernel.org, Linux-MM <linux-mm@kvack.org>,
	clang-built-linux <clang-built-linux@googlegroups.com>
Subject: Re: [PATCH 01/10] x86/mm/numa: Remove uninitialized_var() usage
Date: Thu, 4 Jun 2020 07:56:40 -0700	[thread overview]
Message-ID: <202006040745.525ECD1@keescook> (raw)
In-Reply-To: <CANiq72kLqvriYmMkdD3yU+xJwbn-68Eiu-fTNtC+Lb+1ZRM75g@mail.gmail.com>

On Thu, Jun 04, 2020 at 01:41:07PM +0200, Miguel Ojeda wrote:
> On Thu, Jun 4, 2020 at 9:58 AM Thomas Gleixner <tglx@linutronix.de> wrote:
> >
> > but if we ever lose the 1 then the above will silently compile the code
> > within the IS_ENABLED() section out.
> 
> Yeah, I believe `IS_ENABLED()` is only meant for Kconfig symbols, not
> macro defs in general. A better option would be `__is_defined()` which
> works for defined-to-nothing too.

Er? That's not what it looked like to me:

#define IS_BUILTIN(option) __is_defined(option)
#define IS_ENABLED(option) __or(IS_BUILTIN(option), IS_MODULE(option))

But just to be sure, I just tested in with a real build:

[    3.242160] IS_ENABLED(TEST_UNDEF) false
[    3.242691] __is_defined(TEST_UNDEF) false
[    3.243240] IS_ENABLED(TEST_VALUE_EMPTY) false
[    3.243794] __is_defined(TEST_VALUE_EMPTY) false
[    3.244353] IS_ENABLED(TEST_VALUE_1) true
[    3.244848] __is_defined(TEST_VALUE_1) true

and nope, it only works with a defined value present.

diff --git a/init/main.c b/init/main.c
index 03371976d387..378a9e54b6dc 100644
--- a/init/main.c
+++ b/init/main.c
@@ -1406,6 +1406,34 @@ static int __ref kernel_init(void *unused)
 	 */
 	pti_finalize();
 
+#undef TEST_UNDEF
+	if (IS_ENABLED(TEST_UNDEF))
+		pr_info("IS_ENABLED(TEST_UNDEF) true\n");
+	else
+		pr_info("IS_ENABLED(TEST_UNDEF) false\n");
+	if (__is_defined(TEST_UNDEF))
+		pr_info("__is_defined(TEST_UNDEF) true\n");
+	else
+		pr_info("__is_defined(TEST_UNDEF) false\n");
+#define TEST_VALUE_EMPTY
+	if (IS_ENABLED(TEST_VALUE_EMPTY))
+		pr_info("IS_ENABLED(TEST_VALUE_EMPTY) true\n");
+	else
+		pr_info("IS_ENABLED(TEST_VALUE_EMPTY) false\n");
+	if (__is_defined(TEST_VALUE_EMPTY))
+		pr_info("__is_defined(TEST_VALUE_EMPTY) true\n");
+	else
+		pr_info("__is_defined(TEST_VALUE_EMPTY) false\n");
+#define TEST_VALUE_1 1
+	if (IS_ENABLED(TEST_VALUE_1))
+		pr_info("IS_ENABLED(TEST_VALUE_1) true\n");
+	else
+		pr_info("IS_ENABLED(TEST_VALUE_1) false\n");
+	if (__is_defined(TEST_VALUE_1))
+		pr_info("__is_defined(TEST_VALUE_1) true\n");
+	else
+		pr_info("__is_defined(TEST_VALUE_1) false\n");
+
 	system_state = SYSTEM_RUNNING;
 	numa_default_policy();
 

which means a few other __is_defined() users are not correct too...

-- 
Kees Cook
