// Copyright (c) 2002-2009 Nokia Corporation and/or its subsidiary(-ies).

// All rights reserved.

// This component and the accompanying materials are made available

// under the terms of the License "Eclipse Public License v1.0"

// which accompanies this distribution, and is available

// at the URL "http://www.eclipse.org/legal/epl-v10.html".

//

// Initial Contributors:

// Nokia Corporation - initial contribution.

//

// Contributors:

//

// Description:

// e32\euser\epoc\arm\uc_huffman.cia

// 

//

#include "e32huffman.h"

#include <cpudefs.h>

#ifdef __HUFFMAN_MACHINE_CODED__

// NB The EABI layout of TBitInput is different from GCC. Its vtbl is at 

// offset 0 rather than offset 16

EXPORT_C __NAKED__ TUint TBitInput::ReadL()

/** Read a single bit from the input

	Return the next bit in the input stream. This will call UnderflowL() if

	there are no more bits available.

	@return The next bit in the stream

	@leave "UnderflowL()" It the bit stream is exhausted more UnderflowL is called

		to get more data

*/

	{

	asm(".code 32");

#ifdef __EABI__

	asm("ldmib r0!, {r2,r3}");

#else

	asm("ldmia r0!, {r2,r3}");

#endif

	// r0 = this+8

	asm("subs	r2, r2, #1");

	asm("movpl	r12, r3, asl #1");

#ifdef __EABI__

	asm("stmplda r0, {r2,r12}");

#else

	asm("stmpldb r0, {r2,r12}");

#endif

	asm("movpl	r0, r3, lsr #31");

	__JUMP(pl,lr);

	asm("mov r12, #0");

	asm("b __readmorebits");	// r2==-1, r12==0 -> so all set

	}

EXPORT_C __NAKED__ TUint TBitInput::ReadL(TInt aSize)

/** Read a multi-bit value from the input

	Return the next few bits as an unsigned integer. The last bit read is

	the least significant bit of the returned value, and the value is

	zero extended to return a 32-bit result.

	A read of zero bits will always reaturn zero.

	This will call UnderflowL() if there are not enough bits available.

	@param "TInt aSize" The number of bits to read

	@return The bits read from the stream

	@leave "UnderflowL()" It the bit stream is exhausted more UnderflowL is called

		to get more data

*/

	{

	asm(".code 32");

#ifdef __EABI__

	asm("   ldmib r0!, {r2,r3}");

#else

	asm("   ldmia r0!, {r2,r3}");

#endif

	asm("	subs r2, r2, r1");

	asm("	movpl r12, r3, asl r1");

#ifdef __EABI__

	asm("   stmplda r0, {r2,r12}");

#else

	asm("   stmpldb r0, {r2,r12}");

#endif

	asm("	rsb r1, r1, #32");

	asm("	movpl r0, r3, lsr r1");

	__JUMP(pl,lr);

	// r0=this+8, r1=32-#reqd, r2=count-#reqd (<0), r3=bits

	asm("   sub r1, r1, r2");		// r1 = 32-iCount

	asm("	mov r3, r3, lsr r1");	// scrub existing bits

	asm("	rsb r1, r2, #0");		// r1 = #reqd-iCount

	asm("	mov r12, r3, lsl r1");	// r12 has bits in final position

	asm("__readmorebits:");

	//r0=this+8

	asm("	stmfd sp!, {r12,lr}");

#ifdef __EABI__

	asm("	ldmib r0!, {r1,r14}");	// r1=remain, r14=ptr

#else

	asm("	ldmia r0!, {r1,r14}");	// r1=remain, r14=ptr

#endif

	//r0=this+16

	asm("	cmp r1, #0");

	asm("	ble	__readunderflow");

	asm("	ldr	r12, [r14], #4");

	asm("	subs r3, r1, #32");

#ifdef __EABI__

	asm("	stmda r0!, {r3,r14}");

#else

	asm("	stmdb r0!, {r3,r14}");

#endif

	// r0=this+8

	asm("	addmi r1, r2, r1");

	asm("	addpl r1, r2, #32");		// r1 = bits left in

	asm("	eor	r3, r12, r12, ror #16");

	asm("	bic	r3, r3, #16711680");

	asm("	mov	r3, r3, lsr #8");

	asm("	eor	r3, r3, r12, ror #8");		// r3=bits

	// r0=this+8, r1=count-#reqd, r2=-#reqd, r3=bits, sp[0]=accum, sp[4]=return-pc

	asm("	cmp r1, #0");

	asm("	bmi __readgoroundagain");

	asm("__readhavebits:");

	asm("	rsb r12, r2, #0");

	asm("	mov r12, r3, asl r12");

#ifdef __EABI__

	asm("	stmda r0, {r1,r12}");

#else

	asm("	stmdb r0, {r1,r12}");

#endif

	asm("	ldmfd sp!, {r12,lr}");

	asm("	add r2, r2, #32");

	asm("	orr	r0, r12, r3, lsr r2");

	__JUMP(,lr);

	asm("__readunderflow:");

#ifdef __EABI__

	asm("	ldr	r3, [r0, #-16]!");	// restore r0 to this

#else

	asm("	ldr	r3, [r0], #-16");	// restore r0 to this

#endif

	asm("	stmfd sp!, {r0,r2}");

	// at this point the stack has grown by 16 bytes from the CFA

	__EH_FRAME_ADDRESS(sp,16)

	// The stack looks like this

	// CFA-4  -> lr

	// CFA-8  -> r12

	// CFA-12 -> r2

	// CFA-16 -> r0

	// So we need to record lr @ CFA-4

	// Nothing else needs to be restored since there's no barrier here

	__EH_FRAME_SAVE1(lr,-4)

	asm("	adr lr, 1f ");

#ifdef __EABI__

	asm("	ldr	ip, [r3, #0]");

#else

	asm("	ldr	ip, [r3, #8]");

#endif

	__JUMP(,ip);

	asm("1: ");

	asm("	ldmfd sp!, {r0,r2}");

#ifdef __EABI__

	asm("	ldmib r0!, {r1,r3}");	// r1=count, r3=bits

#else

	asm("	ldmia r0!, {r1,r3}");	// r1=count, r3=bits

#endif

	asm("	adds r1, r1, r2");		// adjust count for bits to extract

	asm("	bpl __readhavebits");

	asm("__readgoroundagain:");

	// r0=this+8, r1=count-#reqd (<0), r2=-#reqd, r3=bits, sp[0]=accum, sp[4]=return-pc

	asm("   sub r12, r1, r2");		// r12 = iCount

	asm("	rsb r12, r12, #32");	// r12= 32-iCount

	asm("	mov r3, r3, lsr r12");	// scrub existing bits

	asm("	ldmfd sp!, {r12,lr}");

	asm("	mov r2, r1");			// r2 = iCount-#reqd

	asm("	rsb r1, r1, #0");		// r1 = #reqd-iCount

	asm("	orr r12, r12, r3, lsl r1");	// r12 has bits in final position

	asm("	b __readmorebits");

	}

EXPORT_C __NAKED__ TUint TBitInput::HuffmanL(const TUint32* aTree)

/** Read and decode a Huffman Code

	Interpret the next bits in the input as a Huffman code in the specified

	decoding. The decoding tree should be the output from Huffman::Decoding().

	@param "const TUint32* aTree" The huffman decoding tree

	@return The symbol that was decoded

	@leave "UnderflowL()" It the bit stream is exhausted more UnderflowL is called

		to get more data

*/

	{

	asm(".code 32");

	asm("__huffman:");

#ifdef __EABI__

	asm("   ldmib r0!, {r2,r3}");

#else

	asm("   ldmia r0!, {r2,r3}");

#endif

	asm("__decode:");

	asm("	ldr r12, [r1]");

	asm("	subs	r2, r2, #4");

	asm("	bmi __fewbits");

	asm("__fastloop:");

	asm("	movs	r3, r3, asl #1");

	asm("	movcc	r12, r12, asl #16");

	asm("	tst	r12, #65536");

	asm("	bne	__fastdecoded");

	asm("	ldr	r12, [r1, r12, lsr #16]!");

	asm("	sub	r2, r2, #1");

	asm("	movs	r3, r3, asl #1");

	asm("	movcc	r12, r12, asl #16");

	asm("	tst	r12, #65536");

	asm("	bne	__fastdecoded");

	asm("	ldr	r12, [r1, r12, lsr #16]!");

	asm("	sub	r2, r2, #1");

	asm("	movs	r3, r3, asl #1");

	asm("	movcc	r12, r12, asl #16");

	asm("	tst	r12, #65536");

	asm("	bne	__fastdecoded");

	asm("	ldr	r12, [r1, r12, lsr #16]!");

	asm("	sub	r2, r2, #1");

	asm("	movs	r3, r3, asl #1");

	asm("	movcc	r12, r12, asl #16");

	asm("	tst	r12, #65536");

	asm("	bne	__fastdecoded");

	asm("	ldr	r12, [r1, r12, lsr #16]!");

	asm("	subs	r2, r2, #1");

	asm("	bpl __fastloop");

	asm("__fewbits:");

	asm("	adds r2, r2, #3");

	asm("	bmi __nobits");

	asm("__slowloop:");

	asm("	movs	r3, r3, asl #1");

	asm("	movcc	r12, r12, asl #16");

	asm("	tst	r12, #65536");

	asm("	bne	__decoded");

	asm("	ldr	r12, [r1, r12, lsr #16]!");

	asm("	subs	r2, r2, #1");

	asm("	bpl __slowloop");

	asm("__nobits:");

	// r0=this, r1=tree, r2=count (-1), r3=bits (0), r12=node=*tree

	// must preserve r0,r1

#ifdef __EABI__

	asm("	ldmib r0!, {r2,r3}");	// r2=remain, r3=ptr

#else

	asm("	ldmia r0!, {r2,r3}");	// r2=remain, r3=ptr

#endif

	asm("	cmp r2, #0");

	asm("	ble	__underflow");

	asm("	ldr	r12, [r3], #4");

	asm("	subs r2, r2, #32");

#ifdef __EABI__

	asm("	stmda r0!, {r2,r3}");

#else

	asm("	stmdb r0!, {r2,r3}");

#endif

	asm("	addmi r2, r2, #32");

	asm("	movpl r2, #32");

	asm("	eor	r3, r12, r12, ror #16");

	asm("	bic	r3, r3, #16711680");

	asm("	mov	r3, r3, lsr #8");

	asm("	eor	r3, r3, r12, ror #8");

	asm("	b	__decode");

	asm("__underflow:");

#ifdef __EABI__

	asm("	ldr	r3, [r0, #-16]!");	// restore r0 to this

#else

	asm("	ldr	r3, [r0], #-16");	// restore r0 to this

#endif

	asm("	stmfd sp!, {r0,r1,r2,lr}");

	// sp is 16 bytes from CFA

	__EH_FRAME_ADDRESS(sp,16)

	// The stack looks like this:

	// CFA-4  -> lr

	// CFA-8  -> r2

	// CFA-12 -> r1

	// CFA-16 -> r0

	// We saved lr @ CFA-4

	// Nothing else needs to be restored since there's no barrier here

	__EH_FRAME_SAVE1(lr,-4)

	asm("	adr lr, 1f ");

#ifdef __EABI__

	asm("	ldr	ip, [r3, #0]");

#else

	asm("	ldr	ip, [r3, #8]");

#endif

	__JUMP(,ip);

	asm("1: ");

	asm("	ldmfd sp!, {r0,r1,r2,lr}");

	asm("	b __huffman");

	asm("__fastdecoded:");

	asm("	add r2, r2, #3");

	asm("__decoded:");

#ifdef __EABI__

	asm("	stmda r0, {r2, r3}");

#else

	asm("	stmdb r0, {r2, r3}");

#endif

	asm("	mov	r0, r12, lsr #17");

	__JUMP(,lr);

	}

#endif