Re: [PATCH][RESEND] do not redefine userspace's NULL #define

From: Lubos Lunak
Date: Sat Apr 14 2012 - 04:54:31 EST


On Saturday 14 of April 2012, Linus Torvalds wrote:
> On Sat, Apr 14, 2012 at 1:21 AM, Lubos Lunak <l.lunak@xxxxxxx> wrote:
> >  How about the patch? Are there any *actual* problems with it or can it
> > please go in?
>
> If we're going to change that thing, I'd actually prefer to just move
> it inside the #ifdef __KERNEL__ entirely, and get rid of the cplusplus
> case.
>
> IOW, something like this (obviously white-space-mangled) patch:
...
> which protects the kernel NULL definition along with the whole
> offsetof and true/false ones too. And just gets rid of the insane C++
> case entirely.

That's what the original version of my patch did, but Arnd pointed out that
other exported headers might use NULL and thus not be self-contained in
userspace after this change, breaking backwards source compatibility. But I'd
expect such breakages to be rather unlikely (stddef.h is probably pulled in
by pretty much everything), so if you're fine with it, I'm ok with this
solution too.

--
Lubos Lunak
l.lunak@xxxxxxx
--- Begin Message --- GCC's NULL is actually __null, which allows detecting some questionable
NULL usage and warn about it. Moreover each platform/compiler should have
its own stddef.h anyway (which is different from linux/stddef.h).
So there's no good reason to leak kernel's NULL to userspace and
override what the compiler provides.

Signed-off-by: LuboÅ LuÅÃk <l.lunak@xxxxxxx>
---
include/linux/stddef.h | 8 ++------
1 files changed, 2 insertions(+), 6 deletions(-)

diff --git a/include/linux/stddef.h b/include/linux/stddef.h
index 6a40c76..1747b67 100644
--- a/include/linux/stddef.h
+++ b/include/linux/stddef.h
@@ -3,14 +3,10 @@

#include <linux/compiler.h>

+#ifdef __KERNEL__
+
#undef NULL
-#if defined(__cplusplus)
-#define NULL 0
-#else
#define NULL ((void *)0)
-#endif
-
-#ifdef __KERNEL__

enum {
false = 0,
--
1.7.3.4
--
Lubos Lunak
l.lunak@xxxxxxx

--- End Message ---