List Info

Thread: changes to bit-twiddling macros & __arraycount()




changes to bit-twiddling macros & __arraycount()
user name
2006-08-13 01:04:17
On Mon, Aug 07, 2006 at 09:19:53PM +0200, Matthias Drochner
wrote:
> 
> dyoungpobox.com said:
> > Per previous discussions, I am moving the
bit-twiddling macros [1] and
> > __arraycount from lib/libkern/libkern.h to
sys/cdefs.h
> 
> Sorry, I missed that.
> 
> > I have
> > added a __-prefix to SHIFTIN & SHIFTOUT
> 
> What's the objective? I could imagine that you want to
use it in userlevel
> programs too, but then these would be highly
unportable. So an extra
> header and names without "__" prefix would
be more reasonable.

I want them in userland for the convenience of programs for
control &
diagnostics of device drivers.  Portability is not key.

> And: It seems to me that the u_int32_t limitation is
not technically
> necessary. One could probably use "unsigned
long" and get the longest
> integer directly usable by the platform. Drawback would
be that
> people with 64-bit platforms could write non-partable
code. Gain would
> be that the macros could be used more effectively on
64-bit.
> And: The "(__n) == 32)" should better be a
">=". ">=8*sizeof(long)"
> if you follow my suggestion above.

I have heard that criticism at least once before.  I have
tried to make
the macros widthless.  See attached.

Dave

-- 
David Young             OJC Technologies
dyoungojctech.com      Urbana, IL * (217) 278-3933
/*	$NetBSD: cdefs.h,v 1.60 2005/12/11 12:25:20 christos Exp
$	*/

/*
 * Copyright (c) 1991, 1993
 *	The Regents of the University of California.  All rights
reserved.
 *
 * This code is derived from software contributed to
Berkeley by
 * Berkeley Software Design, Inc.
 *
 * Redistribution and use in source and binary forms, with
or without
 * modification, are permitted provided that the following
conditions
 * are met:
 * 1. Redistributions of source code must retain the above
copyright
 *    notice, this list of conditions and the following
disclaimer.
 * 2. Redistributions in binary form must reproduce the
above copyright
 *    notice, this list of conditions and the following
disclaimer in the
 *    documentation and/or other materials provided with the
distribution.
 * 3. Neither the name of the University nor the names of
its contributors
 *    may be used to endorse or promote products derived
from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS
``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR
CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	(#)cdefs.h	8.8 (Berkeley) 1/9/95
 */

#ifndef	_SYS_CDEFS_H_
#define	_SYS_CDEFS_H_

/*
 * Macro to test if we're using a GNU C compiler of a
specific vintage
 * or later, for e.g. features that appeared in a particular
version
 * of GNU C.  Usage:
 *
 *	#if __GNUC_PREREQ__(major, minor)
 *	...cool feature...
 *	#else
 *	...delete feature...
 *	#endif
 */
#ifdef __GNUC__
#define	__GNUC_PREREQ__(x, y)						\
	((__GNUC__ == (x) && __GNUC_MINOR__ >= (y))
||			\
	 (__GNUC__ > (x)))
#else
#define	__GNUC_PREREQ__(x, y)	0
#endif

#include <machine/cdefs.h>
#ifdef __ELF__
#include <sys/cdefs_elf.h>
#else
#include <sys/cdefs_aout.h>
#endif

#if defined(__cplusplus)
#define	__BEGIN_DECLS		extern "C" {
#define	__END_DECLS		}
#define	__static_cast(x,y)	static_cast<x>(y)
#else
#define	__BEGIN_DECLS
#define	__END_DECLS
#define	__static_cast(x,y)	(x)y
#endif

/*
 * The __CONCAT macro is used to concatenate parts of symbol
names, e.g.
 * with "#define OLD(foo) __CONCAT(old,foo)",
OLD(foo) produces oldfoo.
 * The __CONCAT macro is a bit tricky -- make sure you
don't put spaces
 * in between its arguments.  __CONCAT can also concatenate
double-quoted
 * strings produced by the __STRING macro, but this only
works with ANSI C.
 */

#define	___STRING(x)	__STRING(x)
#define	___CONCAT(x,y)	__CONCAT(x,y)

#if __STDC__ || defined(__cplusplus)
#define	__P(protos)	protos		/* full-blown ANSI C */
#define	__CONCAT(x,y)	x ## y
#define	__STRING(x)	#x

#define	__const		const		/* define reserved names to standard
*/
#define	__signed	signed
#define	__volatile	volatile
#if defined(__cplusplus)
#define	__inline	inline		/* convert to C++ keyword */
#else
#if !defined(__GNUC__) && !defined(__lint__)
#define	__inline			/* delete GCC keyword */
#endif /* !__GNUC__  && !__lint__ */
#endif /* !__cplusplus */

#else	/* !(__STDC__ || __cplusplus) */
#define	__P(protos)	()		/* traditional C preprocessor */
#define	__CONCAT(x,y)	x/**/y
#define	__STRING(x)	"x"

#ifndef __GNUC__
#define	__const				/* delete pseudo-ANSI C keywords */
#define	__inline
#define	__signed
#define	__volatile
#endif	/* !__GNUC__ */

/*
 * In non-ANSI C environments, new programs will want
ANSI-only C keywords
 * deleted from the program and old programs will want them
left alone.
 * Programs using the ANSI C keywords const, inline etc. as
normal
 * identifiers should define -DNO_ANSI_KEYWORDS.
 */
#ifndef	NO_ANSI_KEYWORDS
#define	const		__const		/* convert ANSI C keywords */
#define	inline		__inline
#define	signed		__signed
#define	volatile	__volatile
#endif /* !NO_ANSI_KEYWORDS */
#endif	/* !(__STDC__ || __cplusplus) */

/*
 * Used for internal auditing of the NetBSD source tree.
 */
#ifdef __AUDIT__
#define	__aconst	__const
#else
#define	__aconst
#endif

/*
 * The following macro is used to remove const cast-away
warnings
 * from gcc -Wcast-qual; it should be used with caution
because it
 * can hide valid errors; in particular most valid uses are
in
 * situations where the API requires it, not to cast away
string
 * constants. We don't use *intptr_t on purpose here and we
are
 * explicit about unsigned long so that we don't have
additional
 * dependencies.
 */
#define __UNCONST(a)	((void *)(unsigned long)(const void
*)(a))

/*
 * The following macro is used to remove the volatile
cast-away warnings
 * from gcc -Wcast-qual; as above it should be used with
caution
 * because it can hide valid errors or warnings.  Valid uses
include
 * making it possible to pass a volatile pointer to
memset().
 * For the same reasons as above, we use unsigned long and
not intptr_t.
 */
#define __UNVOLATILE(a)	((void *)(unsigned long)(volatile
void *)(a))

/*
 * GCC2 provides __extension__ to suppress warnings for
various GNU C
 * language extensions under "-ansi -pedantic".
 */
#if !__GNUC_PREREQ__(2, 0)
#define	__extension__		/* delete __extension__ if non-gcc or
gcc1 */
#endif

/*
 * GCC1 and some versions of GCC2 declare dead
(non-returning) and
 * pure (no side effects) functions using
"volatile" and "const";
 * unfortunately, these then cause warnings under
"-ansi -pedantic".
 * GCC2 uses a new, peculiar __attribute__((attrs)) style. 
All of
 * these work for GNU C++ (modulo a slight glitch in the C++
grammar
 * in the distribution version of 2.5.5).
 */
#if !__GNUC_PREREQ__(2, 5)
#define	__attribute__(x)	/* delete __attribute__ if non-gcc
or gcc1 */
#if defined(__GNUC__) && !defined(__STRICT_ANSI__)
#define	__dead		__volatile
#define	__pure		__const
#endif
#endif

/* Delete pseudo-keywords wherever they are not available or
needed. */
#ifndef __dead
#define	__dead
#define	__pure
#endif

#if __GNUC_PREREQ__(2, 7)
#define	__unused	__attribute__((__unused__))
#else
#define	__unused	/* delete */
#endif

#if __GNUC_PREREQ__(3, 1)
#define	__used		__attribute__((__used__))
#else
#define	__used		/* delete */
#endif

#if __GNUC_PREREQ__(2, 7)
#define	__packed	__attribute__((__packed__))
#define	__aligned(x)	__attribute__((__aligned__(x)))
#define	__section(x)	__attribute__((__section__(x)))
#elif defined(__lint__)
#define	__packed	/* delete */
#define	__aligned(x)	/* delete */
#define	__section(x)	/* delete */
#else
#define	__packed	error: no __packed for this compiler
#define	__aligned(x)	error: no __aligned for this compiler
#define	__section(x)	error: no __section for this compiler
#endif

/*
 * C99 defines the restrict type qualifier keyword, which
was made available
 * in GCC 2.92.
 */
#if __STDC_VERSION__ >= 199901L
#define	__restrict	restrict
#else
#if !__GNUC_PREREQ__(2, 92)
#define	__restrict	/* delete __restrict when not supported
*/
#endif
#endif

/*
 * C99 defines __func__ predefined identifier, which was
made available
 * in GCC 2.95.
 */
#if !(__STDC_VERSION__ >= 199901L)
#if __GNUC_PREREQ__(2, 6)
#define	__func__	__PRETTY_FUNCTION__
#elif __GNUC_PREREQ__(2, 4)
#define	__func__	__FUNCTION__
#else
#define	__func__	""
#endif
#endif /* !(__STDC_VERSION__ >= 199901L) */

#if defined(_KERNEL)
#if defined(NO_KERNEL_RCSIDS)
#undef __KERNEL_RCSID
#define	__KERNEL_RCSID(_n, _s)		/* nothing */
#endif /* NO_KERNEL_RCSIDS */
#endif /* _KERNEL */

#if !defined(_STANDALONE) && !defined(_KERNEL)
#ifdef __GNUC__
#define	__RENAME(x)	___RENAME(x)
#else
#ifdef __lint__
#define	__RENAME(x)	__symbolrename(x)
#else
#error "No function renaming possible"
#endif /* __lint__ */
#endif /* __GNUC__ */
#else /* _STANDALONE || _KERNEL */
#define	__RENAME(x)	no renaming in kernel or standalone
environment
#endif

/*
 * A barrier to stop the optimizer from moving code or
assume live
 * register values. This is gcc specific, the version is
more or less
 * arbitrary, might work with older compilers.
 */
#if __GNUC_PREREQ__(2, 95)
#define	__insn_barrier()	__asm
__volatile("":::"memory")
#else
#define	__insn_barrier()	/* */
#endif

/*
 * GNU C version 2.96 adds explicit branch prediction so
that
 * the CPU back-end can hint the processor and also so that
 * code blocks can be reordered such that the predicted path
 * sees a more linear flow, thus improving cache behavior,
etc.
 *
 * The following two macros provide us with a way to use
this
 * compiler feature.  Use __predict_true() if you expect the
expression
 * to evaluate to true, and __predict_false() if you expect
the
 * expression to evaluate to false.
 *
 * A few notes about usage:
 *
 *	* Generally, __predict_false() error condition checks
(unless
 *	  you have some _strong_ reason to do otherwise, in which
case
 *	  document it), and/or __predict_true() `no-error'
condition
 *	  checks, assuming you want to optimize for the no-error
case.
 *
 *	* Other than that, if you don't know the likelihood of a
test
 *	  succeeding from empirical or other `hard' evidence,
don't
 *	  make predictions.
 *
 *	* These are meant to be used in places that are run `a
lot'.
 *	  It is wasteful to make predictions in code that is run
 *	  seldomly (e.g. at subsystem initialization time) as the
 *	  basic block reordering that this affects can often
generate
 *	  larger code.
 */
#if __GNUC_PREREQ__(2, 96)
#define	__predict_true(exp)	__builtin_expect((exp) != 0, 1)
#define	__predict_false(exp)	__builtin_expect((exp) != 0, 0)
#else
#define	__predict_true(exp)	(exp)
#define	__predict_false(exp)	(exp)
#endif

/*
 * Macros for manipulating "link sets".  Link
sets are arrays of pointers
 * to objects, which are gathered up by the linker.
 *
 * Object format-specific code has provided us with the
following macros:
 *
 *	__link_set_add_text(set, sym)
 *		Add a reference to the .text symbol `sym' to `set'.
 *
 *	__link_set_add_rodata(set, sym)
 *		Add a reference to the .rodata symbol `sym' to `set'.
 *
 *	__link_set_add_data(set, sym)
 *		Add a reference to the .data symbol `sym' to `set'.
 *
 *	__link_set_add_bss(set, sym)
 *		Add a reference to the .bss symbol `sym' to `set'.
 *
 *	__link_set_decl(set, ptype)
 *		Provide an extern declaration of the set `set', which
 *		contains an array of the pointer type `ptype'.  This
 *		macro must be used by any code which wishes to reference
 *		the elements of a link set.
 *
 *	__link_set_start(set)
 *		This points to the first slot in the link set.
 *
 *	__link_set_end(set)
 *		This points to the (non-existent) slot after the last
 *		entry in the link set.
 *
 *	__link_set_count(set)
 *		Count the number of entries in link set `set'.
 *
 * In addition, we provide the following macros for
accessing link sets:
 *
 *	__link_set_foreach(pvar, set)
 *		Iterate over the link set `set'.  Because a link set is
 *		an array of pointers, pvar must be declared as
"type **pvar",
 *		and the actual entry accessed as "*pvar".
 *
 *	__link_set_entry(set, idx)
 *		Access the link set entry at index `idx' from set
`set'.
 */
#define	__link_set_foreach(pvar, set)					\
	for (pvar = __link_set_start(set); pvar <
__link_set_end(set); pvar++)

#define	__link_set_entry(set,
idx)	(__link_set_begin(set)[idx])

/*
 * Return the number of elements in a statically-allocated
array,
 * __x.
 */
#define	__arraycount(__x)	(sizeof(__x) / sizeof(__x[0]))

/* __BIT(n): nth bit, where __BIT(0) == 0x1. */
#define	__BIT(__n)	\
	(((__n) >= NBBY * sizeof(uintmax_t)) ? 0 : ((uintmax_t)1
<< (__n)))

/* __BITS(m, n): bits m through n, m < n. */
#define	__BITS(__m, __n)	\
	((__BIT(MAX((__m), (__n)) + 1) - 1) ^ (__BIT(MIN((__m),
(__n))) - 1))

/* find least significant bit that is set */
#define	__LOWEST_SET_BIT(__mask) ((((__mask) - 1) &
(__mask)) ^ (__mask))

#define	__PRIxBIT	PRIxMAX
#define	__PRIxBITS	__PRIxBIT

#define	__SHIFTOUT(__x, __mask)	\
	((typeof(__x))(((__x) & (__mask)) /
__LOWEST_SET_BIT(__mask)))
#define	__SHIFTIN(__x, __mask) ((typeof(__x))((__x) *
__LOWEST_SET_BIT(__mask)))
#define	__SHIFTOUT_MASK(__mask) __SHIFTOUT((__mask),
(__mask))

#endif /* !_SYS_CDEFS_H_ */
changes to bit-twiddling macros & __arraycount()
user name
2006-08-13 12:27:13
On Sat, Aug 12, 2006 at 08:04:17PM -0500, David Young wrote:
> I want them in userland for the convenience of programs
for control &
> diagnostics of device drivers.  Portability is not key.

I personally think we should get rid of the bit macros
completely - they
decrease code readability and lead to abuse.

As an especially bad example: we have code in our tree that
does:

   switch (ISSET(t->c_cflag, CSIZE)) {
    case CS5:
      ..
    case CS6:
      ..
    ..
   }

Martin
changes to bit-twiddling macros & __arraycount()
user name
2006-08-13 15:37:00
On Sun, Aug 13, 2006 at 02:27:13PM +0200, Martin Husemann
wrote:
> On Sat, Aug 12, 2006 at 08:04:17PM -0500, David Young
wrote:
> > I want them in userland for the convenience of
programs for control &
> > diagnostics of device drivers.  Portability is not
key.
> 
> I personally think we should get rid of the bit macros
completely - they
> decrease code readability and lead to abuse.
> 
> As an especially bad example: we have code in our tree
that does:
> 
>    switch (ISSET(t->c_cflag, CSIZE)) {
>     case CS5:
>       ..
>     case CS6:
>       ..
>     ..

Very well, but you are talking about different bit macros
than mine.
I never intended for mine to be used with ISSET and family,
and that
is not the way they are used.  Moreover, my macros ease
reading and
writing code.  Using __BITS()/__SHIFTIN()/__SHIFTOUT() for
register
definitions eases comparisons between source code and
standards documents
or datasheets, obviates the need for separate xxx_SHIFT
#defines,
and saves a developer the trouble of "manually"
shifting bits into a
hexadecimal mask.

Dave

-- 
David Young             OJC Technologies
dyoungojctech.com      Urbana, IL * (217) 278-3933
[1-3]

about | contact  Other archives ( Real Estate discussion Medical topics )