more annoying, nitpicky pedantry as i continue my tour of the kernel
source. what is the rationale behind using the gcc keyword alias
"__signed__" in many of the architecture-specific types.h files?
the general form of many of those files is:
===================
#ifndef __ASSEMBLY__
/*
* __xx is ok: it doesn't pollute the POSIX namespace. Use these in the
* header files exported to user space
*/
typedef __signed__ char __s8; <-- see?
typedef unsigned char __u8;
typedef __signed__ short __s16; <-- here, too, and so on.
typedef unsigned short __u16;
...
#ifdef __KERNEL__
#ifndef __ASSEMBLY__
typedef signed char s8; <-- but now, things change
typedef unsigned char u8;
typedef signed short s16;
typedef unsigned short u16;
...
===================
so the keyword alias "__signed__" is used early on in nearly every
types.h file but, if __KERNEL__ is defined, the file falls back to
just using "signed". (the use of "unsigned" is, of course, consistent
throughout.)
so, first, what is the rationale behind the use of "__signed__" in
this context? and is there any reason that most of that content can't
be centralized in a single arch-independent file and simply
#include'd? for the most part, that general content looks identical
regardless of the architecture. (in some cases, the files are
*absolutely* identical, such as for arm and arm26, and one would think
that a simple #ifdef or two could handle differing word sizes of the
architecture.)
however, having said all that, there are some puzzling exceptions to
the above pattern. asm-ia64/types.h typedefs the second set based on
the *earlier* typedefs:
=============================
...
typedef __signed__ int __s32;
typedef unsigned int __u32;
typedef __signed__ long __s64;
typedef unsigned long __u64;
/*
* These aren't exported outside the kernel to avoid name space
clashes
*/
# ifdef __KERNEL__
typedef __s8 s8; <-- like that
typedef __u8 u8;
typedef __s16 s16;
typedef __u16 u16;
...
=============================
then there's asm-mips/sh/types.h, which *continues* to use
"__signed__" after everyone else switches to merely "signed":
=============================
...
#ifdef __KERNEL__
#define BITS_PER_LONG 32
#ifndef __ASSEMBLY__
typedef __signed__ char s8; <-- still using that gcc keyword alias
typedef unsigned char u8;
typedef __signed__ short s16;
typedef unsigned short u16;
...
=============================
and, finally, there's include/asm-mips/types.h, which curiously
introduces "__signed":
=============================
...
#ifdef __KERNEL__
#define BITS_PER_LONG _MIPS_SZLONG
#ifndef __ASSEMBLY__
typedef __signed char s8; <-- ?????
typedef unsigned char u8;
typedef __signed short s16;
typedef unsigned short u16;
...
=============================
is there any reason why a lot of this can't be standardized? it
certainly looks like a lot of unnecessary duplication, interspersed
with puzzling anomalies just to keep life interesting. :-)
rday
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
