/*
 * Copyright (c) 2000-2006 Apple Computer, Inc. All rights reserved.
 *
 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
 * 
 * This file contains Original Code and/or Modifications of Original Code
 * as defined in and that are subject to the Apple Public Source License
 * Version 2.0 (the 'License'). You may not use this file except in
 * compliance with the License. The rights granted to you under the License
 * may not be used to create, or enable the creation or redistribution of,
 * unlawful or unlicensed copies of an Apple operating system, or to
 * circumvent, violate, or enable the circumvention or violation of, any
 * terms of an Apple operating system software license agreement.
 * 
 * Please obtain a copy of the License at
 * http://www.opensource.apple.com/apsl/ and read it before using this file.
 * 
 * The Original Code and all software distributed under the License are
 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
 * Please see the License for the specific language governing rights and
 * limitations under the License.
 * 
 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
 */

#include <libkern/OSAtomic.h>

enum {
	false	= 0,
	true	= 1
};

#ifndef NULL
#define NULL ((void *)0)
#endif

/*
 * atomic operations
 *	these are _the_ atomic operations, currently cast atop CompareAndSwap,
 *	which is implemented in assembler.	if we are worried about the cost of
 *	this layering (we shouldn't be), then all this stuff could be
 *	implemented in assembler, as it is in MacOS8/9
 *	(derived from SuperMario/NativeLibs/IO/DriverServices/Synchronization.s,
 *	which I wrote for NuKernel in a previous life with a different last name...)
 *
 * native Boolean	CompareAndSwap(UInt32 oldValue, UInt32 newValue, UInt32 * oldValuePtr);
 *
 * We've since implemented a few more of these -- OSAddAtomic, OSDequeueAtomic,
 * OSEnqueueAtomic etc -- in assembler, either for speed or correctness.  See also the
 * commpage atomic operations, and the platform specific versions.
 * Like standards, there are a lot of atomic ops to choose from!
 */

#if defined(__i386__) || defined(__x86_64__)
/* Implemented in assembly for i386 and x86_64 */
#else
#error Unsupported arch
#endif

#undef OSIncrementAtomic
SInt32	OSIncrementAtomic(volatile SInt32 * value)
{
	return OSAddAtomic(1, value);
}

#undef OSDecrementAtomic
SInt32	OSDecrementAtomic(volatile SInt32 * value)
{
	return OSAddAtomic(-1, value);
}

static UInt32	OSBitwiseAtomic(UInt32 and_mask, UInt32 or_mask, UInt32 xor_mask, volatile UInt32 * value)
{
	UInt32	oldValue;
	UInt32	newValue;
	
	do {
		oldValue = *value;
		newValue = ((oldValue & and_mask) | or_mask) ^ xor_mask;
	} while (! OSCompareAndSwap(oldValue, newValue, value));
	
	return oldValue;
}

#undef OSBitAndAtomic
UInt32	OSBitAndAtomic(UInt32 mask, volatile UInt32 * value)
{
	return OSBitwiseAtomic(mask, 0, 0, value);
}

#undef OSBitOrAtomic
UInt32	OSBitOrAtomic(UInt32 mask, volatile UInt32 * value)
{
	return OSBitwiseAtomic((UInt32) -1, mask, 0, value);
}

#undef OSBitXorAtomic
UInt32	OSBitXorAtomic(UInt32 mask, volatile UInt32 * value)
{
	return OSBitwiseAtomic((UInt32) -1, 0, mask, value);
}

#if defined(__i386__) || defined(__x86_64__)
static Boolean OSCompareAndSwap8(UInt8 oldValue8, UInt8 newValue8, volatile UInt8 * value8)
{
	UInt32				mask		= 0x000000ff;
	UInt32				alignment	= (UInt32)((unsigned long) value8) & (sizeof(UInt32) - 1);
	UInt32				shiftValues = (24 << 24) | (16 << 16) | (8 << 8);
	int					shift		= (UInt32) *(((UInt8 *) &shiftValues) + alignment);
	volatile UInt32 *	value32		= (volatile UInt32 *) ((uintptr_t)value8 - alignment);
	UInt32				oldValue;
	UInt32				newValue;

	mask <<= shift;

	oldValue = *value32;
	oldValue = (oldValue & ~mask) | (oldValue8 << shift);
	newValue = (oldValue & ~mask) | (newValue8 << shift);

	return OSCompareAndSwap(oldValue, newValue, value32);
}
#endif

static Boolean	OSTestAndSetClear(UInt32 bit, Boolean wantSet, volatile UInt8 * startAddress)
{
	UInt8		mask = 1;
	UInt8		oldValue;
	UInt8		wantValue;
	
	startAddress += (bit / 8);
	mask <<= (7 - (bit % 8));
	wantValue = wantSet ? mask : 0;
	
	do {
		oldValue = *startAddress;
		if ((oldValue & mask) == wantValue) {
			break;
		}
	} while (! OSCompareAndSwap8(oldValue, (oldValue & ~mask) | wantValue, startAddress));
	
	return (oldValue & mask) == wantValue;
}

Boolean OSTestAndSet(UInt32 bit, volatile UInt8 * startAddress)
{
	return OSTestAndSetClear(bit, true, startAddress);
}

Boolean OSTestAndClear(UInt32 bit, volatile UInt8 * startAddress)
{
	return OSTestAndSetClear(bit, false, startAddress);
}

/*
 * silly unaligned versions
 */

SInt8	OSIncrementAtomic8(volatile SInt8 * value)
{
	return OSAddAtomic8(1, value);
}

SInt8	OSDecrementAtomic8(volatile SInt8 * value)
{
	return OSAddAtomic8(-1, value);
}

#if defined(__i386__) || defined(__x86_64__)
SInt8	OSAddAtomic8(SInt32 amount, volatile SInt8 * value)
{
	SInt8	oldValue;
	SInt8	newValue;
	
	do {
		oldValue = *value;
		newValue = oldValue + amount;
	} while (! OSCompareAndSwap8((UInt8) oldValue, (UInt8) newValue, (volatile UInt8 *) value));
	
	return oldValue;
}
#endif

static UInt8	OSBitwiseAtomic8(UInt32 and_mask, UInt32 or_mask, UInt32 xor_mask, volatile UInt8 * value)
{
	UInt8	oldValue;
	UInt8	newValue;
	
	do {
		oldValue = *value;
		newValue = ((oldValue & and_mask) | or_mask) ^ xor_mask;
	} while (! OSCompareAndSwap8(oldValue, newValue, value));
	
	return oldValue;
}

UInt8	OSBitAndAtomic8(UInt32 mask, volatile UInt8 * value)
{
	return OSBitwiseAtomic8(mask, 0, 0, value);
}

UInt8	OSBitOrAtomic8(UInt32 mask, volatile UInt8 * value)
{
	return OSBitwiseAtomic8((UInt32) -1, mask, 0, value);
}

UInt8	OSBitXorAtomic8(UInt32 mask, volatile UInt8 * value)
{
	return OSBitwiseAtomic8((UInt32) -1, 0, mask, value);
}

#if defined(__i386__) || defined(__x86_64__)
static Boolean OSCompareAndSwap16(UInt16 oldValue16, UInt16 newValue16, volatile UInt16 * value16)
{
	UInt32				mask		= 0x0000ffff;
	UInt32				alignment	= (UInt32)((unsigned long) value16) & (sizeof(UInt32) - 1);
	UInt32				shiftValues = (16 << 24) | (16 << 16);
	UInt32				shift		= (UInt32) *(((UInt8 *) &shiftValues) + alignment);
	volatile UInt32 *	value32		= (volatile UInt32 *) (((unsigned long) value16) - alignment);
	UInt32				oldValue;
	UInt32				newValue;

	mask <<= shift;

	oldValue = *value32;
	oldValue = (oldValue & ~mask) | (oldValue16 << shift);
	newValue = (oldValue & ~mask) | (newValue16 << shift);

	return OSCompareAndSwap(oldValue, newValue, value32);
}
#endif

SInt16	OSIncrementAtomic16(volatile SInt16 * value)
{
	return OSAddAtomic16(1, value);
}

SInt16	OSDecrementAtomic16(volatile SInt16 * value)
{
	return OSAddAtomic16(-1, value);
}

#if defined(__i386__) || defined(__x86_64__)
SInt16	OSAddAtomic16(SInt32 amount, volatile SInt16 * value)
{
	SInt16	oldValue;
	SInt16	newValue;
	
	do {
		oldValue = *value;
		newValue = oldValue + amount;
	} while (! OSCompareAndSwap16((UInt16) oldValue, (UInt16) newValue, (volatile UInt16 *) value));
	
	return oldValue;
}
#endif

static UInt16	OSBitwiseAtomic16(UInt32 and_mask, UInt32 or_mask, UInt32 xor_mask, volatile UInt16 * value)
{
	UInt16	oldValue;
	UInt16	newValue;
	
	do {
		oldValue = *value;
		newValue = ((oldValue & and_mask) | or_mask) ^ xor_mask;
	} while (! OSCompareAndSwap16(oldValue, newValue, value));
	
	return oldValue;
}

UInt16	OSBitAndAtomic16(UInt32 mask, volatile UInt16 * value)
{
	return OSBitwiseAtomic16(mask, 0, 0, value);
}

UInt16	OSBitOrAtomic16(UInt32 mask, volatile UInt16 * value)
{
	return OSBitwiseAtomic16((UInt32) -1, mask, 0, value);
}

UInt16	OSBitXorAtomic16(UInt32 mask, volatile UInt16 * value)
{
	return OSBitwiseAtomic16((UInt32) -1, 0, mask, value);
}