From: ard.biesheuvel@linaro.org (Ard Biesheuvel)
To: linux-arm-kernel@lists.infradead.org
Subject: [RFC PATCH] types.h: use GCC supplied typedefs if appropriate
Date: Thu, 8 Aug 2013 13:06:50 +0200 [thread overview]
Message-ID: <1375960010-4214-1-git-send-email-ard.biesheuvel@linaro.org> (raw)
GCC supplies a set of builtin defines that are meant to be used in the typedefs
for types such as uint8_t, uint16_t etc. In fact, this is exactly what the
stdint.h header does (of which GCC supplies its own version for freestanding
builds). So in stdint.h, the types are defined as
typedef __UINT16_TYPE__ uint16_t
typedef __UINT32_TYPE__ uint32_t
However, types.h in the kernel contains its own type definitions for these
stdint.h types, and these do not depend on the GCC builtins.
In the ARM world, both bare metal and glibc targeted versions of GCC are
supported for building the kernel, and unfortunately, these do not agree on the
definition of __UINT32_TYPE__ (likewise for __INT32_TYPE__ and __UINTPTR_TYPE__)
- bare metal uses 'long unsigned int'
- glibc GCC uses 'unsigned int'
The result of this is that, while it is perfectly feasible in principle to
support code that includes 'stdint.h' by compiling with -ffreestanding, (such as
code using NEON intrinsics, whose header 'arm_neon.h' includes 'stdint.h'), in
practice this breaks because we may end up with conflicting type definitions for
uint32_t (and uintptr_t) depending on whether you are using bare metal GCC or
glibc GCC.
Arguably, this is a GCC issue because a) it does not pick up on the fact that
'typedef unsigned int uint32_t' and 'typedef long unsigned int uint32_t' are not
in fact conflicting or b) it maintains this trivial difference between bare
metal and glibc targeted build configs.
However, even if I am aware that stdint.h support or matters related to it may
be controversial subjects, fixing it in the kernel is not /that/ obtrusive, and
solves matters for older GCCs as well, hence this RFC patch.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
---
include/linux/types.h | 55 +++++++++++++++++++++++++++++++++++++++------------
1 file changed, 42 insertions(+), 13 deletions(-)
diff --git a/include/linux/types.h b/include/linux/types.h
index 4d118ba..40c5925 100644
--- a/include/linux/types.h
+++ b/include/linux/types.h
@@ -33,7 +33,11 @@ typedef __kernel_gid32_t gid_t;
typedef __kernel_uid16_t uid16_t;
typedef __kernel_gid16_t gid16_t;
-typedef unsigned long uintptr_t;
+#ifndef __UINTPTR_TYPE__
+#define __UINTPTR_TYPE__ unsigned long
+#endif
+
+typedef __UINTPTR_TYPE__ uintptr_t;
#ifdef CONFIG_UID16
/* This is defined by include/asm-{arch}/posix_types.h */
@@ -91,26 +95,51 @@ typedef unsigned short ushort;
typedef unsigned int uint;
typedef unsigned long ulong;
+#ifndef __UINT8_TYPE__
+#define __UINT8_TYPE__ __u8
+#endif
+#ifndef __INT8_TYPE__
+#define __INT8_TYPE__ __s8
+#endif
+#ifndef __UINT16_TYPE__
+#define __UINT16_TYPE__ __u16
+#endif
+#ifndef __INT16_TYPE__
+#define __INT16_TYPE__ __s16
+#endif
+#ifndef __UINT32_TYPE__
+#define __UINT32_TYPE__ __u32
+#endif
+#ifndef __INT32_TYPE__
+#define __INT32_TYPE__ __s32
+#endif
+#ifndef __UINT64_TYPE__
+#define __UINT64_TYPE__ __u64
+#endif
+#ifndef __INT64_TYPE__
+#define __INT64_TYPE__ __s64
+#endif
+
#ifndef __BIT_TYPES_DEFINED__
#define __BIT_TYPES_DEFINED__
-typedef __u8 u_int8_t;
-typedef __s8 int8_t;
-typedef __u16 u_int16_t;
-typedef __s16 int16_t;
-typedef __u32 u_int32_t;
-typedef __s32 int32_t;
+typedef __UINT8_TYPE__ u_int8_t;
+typedef __INT8_TYPE__ int8_t;
+typedef __UINT16_TYPE__ u_int16_t;
+typedef __INT16_TYPE__ int16_t;
+typedef __UINT32_TYPE__ u_int32_t;
+typedef __INT32_TYPE__ int32_t;
#endif /* !(__BIT_TYPES_DEFINED__) */
-typedef __u8 uint8_t;
-typedef __u16 uint16_t;
-typedef __u32 uint32_t;
+typedef __UINT8_TYPE__ uint8_t;
+typedef __UINT16_TYPE__ uint16_t;
+typedef __UINT32_TYPE__ uint32_t;
#if defined(__GNUC__)
-typedef __u64 uint64_t;
-typedef __u64 u_int64_t;
-typedef __s64 int64_t;
+typedef __UINT64_TYPE__ uint64_t;
+typedef __UINT64_TYPE__ u_int64_t;
+typedef __INT64_TYPE__ int64_t;
#endif
/* this is a special 64bit data type that is 8-byte aligned */
--
1.8.1.2
next reply other threads:[~2013-08-08 11:06 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-08-08 11:06 Ard Biesheuvel [this message]
2013-08-08 17:43 ` [RFC PATCH] types.h: use GCC supplied typedefs if appropriate Dave Martin
2013-08-09 6:39 ` Ard Biesheuvel
2013-08-09 14:03 ` Dave Martin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1375960010-4214-1-git-send-email-ard.biesheuvel@linaro.org \
--to=ard.biesheuvel@linaro.org \
--cc=linux-arm-kernel@lists.infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).