squashfs-tools: added 3.1r2 (normal and lzma) - closes #1466

5 files changed, 2776 insertions, 0 deletions

diff --git a/packages/squashfs-tools/files/002-squashfs_lzma.patch b/packages/squashfs-tools/files/002-squashfs_lzma.patch
new file mode 100644
index 0000000000..a166e84971
--- /dev/null
+++ b/packages/squashfs-tools/files/002-squashfs_lzma.patch
@@ -0,0 +1,886 @@
+diff -Nur linux-2.6.16/fs/squashfs/inode.c linux-2.6.16-owrt/fs/squashfs/inode.c
+--- linux-2.6.16/fs/squashfs/inode.c	2006-03-21 10:55:59.000000000 +0100
++++ linux-2.6.16-owrt/fs/squashfs/inode.c	2006-03-21 12:24:37.000000000 +0100
+@@ -4,6 +4,9 @@
+  * Copyright (c) 2002, 2003, 2004, 2005, 2006
+  * Phillip Lougher <phillip@lougher.org.uk>
+  *
++ * LZMA decompressor support added by Oleg I. Vdovikin
++ * Copyright (c) 2005 Oleg I.Vdovikin <oleg@cs.msu.su>
++ *
+  * This program is free software; you can redistribute it and/or
+  * modify it under the terms of the GNU General Public License
+  * as published by the Free Software Foundation; either version 2,
+@@ -21,6 +24,7 @@
+  * inode.c
+  */
+ 
++#define SQUASHFS_LZMA
+ #include <linux/types.h>
+ #include <linux/squashfs_fs.h>
+ #include <linux/module.h>
+@@ -44,6 +48,19 @@
+ 
+ #include "squashfs.h"
+ 
++#ifdef SQUASHFS_LZMA
++#include "LzmaDecode.h"
++
++/* default LZMA settings, should be in sync with mksquashfs */
++#define LZMA_LC 3
++#define LZMA_LP 0
++#define LZMA_PB 2
++
++#define LZMA_WORKSPACE_SIZE ((LZMA_BASE_SIZE + \
++      (LZMA_LIT_SIZE << (LZMA_LC + LZMA_LP))) * sizeof(CProb))
++
++#endif
++
+ static void squashfs_put_super(struct super_block *);
+ static int squashfs_statfs(struct super_block *, struct kstatfs *);
+ static int squashfs_symlink_readpage(struct file *file, struct page *page);
+@@ -64,7 +81,11 @@
+ 				const char *, void *);
+ 
+ 
++#ifdef SQUASHFS_LZMA
++static unsigned char lzma_workspace[LZMA_WORKSPACE_SIZE];
++#else
+ static z_stream stream;
++#endif
+ 
+ static struct file_system_type squashfs_fs_type = {
+ 	.owner = THIS_MODULE,
+@@ -249,6 +270,15 @@
+ 	if (compressed) {
+ 		int zlib_err;
+ 
++#ifdef SQUASHFS_LZMA
++		if ((zlib_err = LzmaDecode(lzma_workspace, 
++			LZMA_WORKSPACE_SIZE, LZMA_LC, LZMA_LP, LZMA_PB, 
++			c_buffer, c_byte, buffer, msblk->read_size, &bytes)) != LZMA_RESULT_OK)
++		{
++			ERROR("lzma returned unexpected result 0x%x\n", zlib_err);
++			bytes = 0;
++		}
++#else
+ 		stream.next_in = c_buffer;
+ 		stream.avail_in = c_byte;
+ 		stream.next_out = buffer;
+@@ -263,6 +293,7 @@
+ 			bytes = 0;
+ 		} else
+ 			bytes = stream.total_out;
++#endif
+ 		
+ 		up(&msblk->read_data_mutex);
+ 	}
+@@ -2046,15 +2077,19 @@
+ 	printk(KERN_INFO "squashfs: version 3.0 (2006/03/15) "
+ 		"Phillip Lougher\n");
+ 
++#ifndef SQUASHFS_LZMA
+ 	if (!(stream.workspace = vmalloc(zlib_inflate_workspacesize()))) {
+ 		ERROR("Failed to allocate zlib workspace\n");
+ 		destroy_inodecache();
+ 		err = -ENOMEM;
+ 		goto out;
+ 	}
++#endif
+ 
+ 	if ((err = register_filesystem(&squashfs_fs_type))) {
++#ifndef SQUASHFS_LZMA
+ 		vfree(stream.workspace);
++#endif
+ 		destroy_inodecache();
+ 	}
+ 
+@@ -2065,7 +2100,9 @@
+ 
+ static void __exit exit_squashfs_fs(void)
+ {
++#ifndef SQUASHFS_LZMA
+ 	vfree(stream.workspace);
++#endif
+ 	unregister_filesystem(&squashfs_fs_type);
+ 	destroy_inodecache();
+ }
+diff -Nur linux-2.6.16/fs/squashfs/LzmaDecode.c linux-2.6.16-owrt/fs/squashfs/LzmaDecode.c
+--- linux-2.6.16/fs/squashfs/LzmaDecode.c	1970-01-01 01:00:00.000000000 +0100
++++ linux-2.6.16-owrt/fs/squashfs/LzmaDecode.c	2006-03-21 10:56:57.000000000 +0100
+@@ -0,0 +1,663 @@
++/*
++  LzmaDecode.c
++  LZMA Decoder
++  
++  LZMA SDK 4.05 Copyright (c) 1999-2004 Igor Pavlov (2004-08-25)
++  http://www.7-zip.org/
++
++  LZMA SDK is licensed under two licenses:
++  1) GNU Lesser General Public License (GNU LGPL)
++  2) Common Public License (CPL)
++  It means that you can select one of these two licenses and 
++  follow rules of that license.
++
++  SPECIAL EXCEPTION:
++  Igor Pavlov, as the author of this code, expressly permits you to 
++  statically or dynamically link your code (or bind by name) to the 
++  interfaces of this file without subjecting your linked code to the 
++  terms of the CPL or GNU LGPL. Any modifications or additions 
++  to this file, however, are subject to the LGPL or CPL terms.
++*/
++
++#include "LzmaDecode.h"
++
++#ifndef Byte
++#define Byte unsigned char
++#endif
++
++#define kNumTopBits 24
++#define kTopValue ((UInt32)1 << kNumTopBits)
++
++#define kNumBitModelTotalBits 11
++#define kBitModelTotal (1 << kNumBitModelTotalBits)
++#define kNumMoveBits 5
++
++typedef struct _CRangeDecoder
++{
++  Byte *Buffer;
++  Byte *BufferLim;
++  UInt32 Range;
++  UInt32 Code;
++  #ifdef _LZMA_IN_CB
++  ILzmaInCallback *InCallback;
++  int Result;
++  #endif
++  int ExtraBytes;
++} CRangeDecoder;
++
++Byte RangeDecoderReadByte(CRangeDecoder *rd)
++{
++  if (rd->Buffer == rd->BufferLim)
++  {
++    #ifdef _LZMA_IN_CB
++    UInt32 size;
++    rd->Result = rd->InCallback->Read(rd->InCallback, &rd->Buffer, &size);
++    rd->BufferLim = rd->Buffer + size;
++    if (size == 0)
++    #endif
++    {
++      rd->ExtraBytes = 1;
++      return 0xFF;
++    }
++  }
++  return (*rd->Buffer++);
++}
++
++/* #define ReadByte (*rd->Buffer++) */
++#define ReadByte (RangeDecoderReadByte(rd))
++
++void RangeDecoderInit(CRangeDecoder *rd,
++  #ifdef _LZMA_IN_CB
++    ILzmaInCallback *inCallback
++  #else
++    Byte *stream, UInt32 bufferSize
++  #endif
++    )
++{
++  int i;
++  #ifdef _LZMA_IN_CB
++  rd->InCallback = inCallback;
++  rd->Buffer = rd->BufferLim = 0;
++  #else
++  rd->Buffer = stream;
++  rd->BufferLim = stream + bufferSize;
++  #endif
++  rd->ExtraBytes = 0;
++  rd->Code = 0;
++  rd->Range = (0xFFFFFFFF);
++  for(i = 0; i < 5; i++)
++    rd->Code = (rd->Code << 8) | ReadByte;
++}
++
++#define RC_INIT_VAR UInt32 range = rd->Range; UInt32 code = rd->Code;        
++#define RC_FLUSH_VAR rd->Range = range; rd->Code = code;
++#define RC_NORMALIZE if (range < kTopValue) { range <<= 8; code = (code << 8) | ReadByte; }
++
++UInt32 RangeDecoderDecodeDirectBits(CRangeDecoder *rd, int numTotalBits)
++{
++  RC_INIT_VAR
++  UInt32 result = 0;
++  int i;
++  for (i = numTotalBits; i > 0; i--)
++  {
++    /* UInt32 t; */
++    range >>= 1;
++
++    result <<= 1;
++    if (code >= range)
++    {
++      code -= range;
++      result |= 1;
++    }
++    /*
++    t = (code - range) >> 31;
++    t &= 1;
++    code -= range & (t - 1);
++    result = (result + result) | (1 - t);
++    */
++    RC_NORMALIZE
++  }
++  RC_FLUSH_VAR
++  return result;
++}
++
++int RangeDecoderBitDecode(CProb *prob, CRangeDecoder *rd)
++{
++  UInt32 bound = (rd->Range >> kNumBitModelTotalBits) * *prob;
++  if (rd->Code < bound)
++  {
++    rd->Range = bound;
++    *prob += (kBitModelTotal - *prob) >> kNumMoveBits;
++    if (rd->Range < kTopValue)
++    {
++      rd->Code = (rd->Code << 8) | ReadByte;
++      rd->Range <<= 8;
++    }
++    return 0;
++  }
++  else
++  {
++    rd->Range -= bound;
++    rd->Code -= bound;
++    *prob -= (*prob) >> kNumMoveBits;
++    if (rd->Range < kTopValue)
++    {
++      rd->Code = (rd->Code << 8) | ReadByte;
++      rd->Range <<= 8;
++    }
++    return 1;
++  }
++}
++
++#define RC_GET_BIT2(prob, mi, A0, A1) \
++  UInt32 bound = (range >> kNumBitModelTotalBits) * *prob; \
++  if (code < bound) \
++    { A0; range = bound; *prob += (kBitModelTotal - *prob) >> kNumMoveBits; mi <<= 1; } \
++  else \
++    { A1; range -= bound; code -= bound; *prob -= (*prob) >> kNumMoveBits; mi = (mi + mi) + 1; } \
++  RC_NORMALIZE
++
++#define RC_GET_BIT(prob, mi) RC_GET_BIT2(prob, mi, ; , ;)               
++
++int RangeDecoderBitTreeDecode(CProb *probs, int numLevels, CRangeDecoder *rd)
++{
++  int mi = 1;
++  int i;
++  #ifdef _LZMA_LOC_OPT
++  RC_INIT_VAR
++  #endif
++  for(i = numLevels; i > 0; i--)
++  {
++    #ifdef _LZMA_LOC_OPT
++    CProb *prob = probs + mi;
++    RC_GET_BIT(prob, mi)
++    #else
++    mi = (mi + mi) + RangeDecoderBitDecode(probs + mi, rd);
++    #endif
++  }
++  #ifdef _LZMA_LOC_OPT
++  RC_FLUSH_VAR
++  #endif
++  return mi - (1 << numLevels);
++}
++
++int RangeDecoderReverseBitTreeDecode(CProb *probs, int numLevels, CRangeDecoder *rd)
++{
++  int mi = 1;
++  int i;
++  int symbol = 0;
++  #ifdef _LZMA_LOC_OPT
++  RC_INIT_VAR
++  #endif
++  for(i = 0; i < numLevels; i++)
++  {
++    #ifdef _LZMA_LOC_OPT
++    CProb *prob = probs + mi;
++    RC_GET_BIT2(prob, mi, ; , symbol |= (1 << i))
++    #else
++    int bit = RangeDecoderBitDecode(probs + mi, rd);
++    mi = mi + mi + bit;
++    symbol |= (bit << i);
++    #endif
++  }
++  #ifdef _LZMA_LOC_OPT
++  RC_FLUSH_VAR
++  #endif
++  return symbol;
++}
++
++Byte LzmaLiteralDecode(CProb *probs, CRangeDecoder *rd)
++{ 
++  int symbol = 1;
++  #ifdef _LZMA_LOC_OPT
++  RC_INIT_VAR
++  #endif
++  do
++  {
++    #ifdef _LZMA_LOC_OPT
++    CProb *prob = probs + symbol;
++    RC_GET_BIT(prob, symbol)
++    #else
++    symbol = (symbol + symbol) | RangeDecoderBitDecode(probs + symbol, rd);
++    #endif
++  }
++  while (symbol < 0x100);
++  #ifdef _LZMA_LOC_OPT
++  RC_FLUSH_VAR
++  #endif
++  return symbol;
++}
++
++Byte LzmaLiteralDecodeMatch(CProb *probs, CRangeDecoder *rd, Byte matchByte)
++{ 
++  int symbol = 1;
++  #ifdef _LZMA_LOC_OPT
++  RC_INIT_VAR
++  #endif
++  do
++  {
++    int bit;
++    int matchBit = (matchByte >> 7) & 1;
++    matchByte <<= 1;
++    #ifdef _LZMA_LOC_OPT
++    {
++      CProb *prob = probs + ((1 + matchBit) << 8) + symbol;
++      RC_GET_BIT2(prob, symbol, bit = 0, bit = 1)
++    }
++    #else
++    bit = RangeDecoderBitDecode(probs + ((1 + matchBit) << 8) + symbol, rd);
++    symbol = (symbol << 1) | bit;
++    #endif
++    if (matchBit != bit)
++    {
++      while (symbol < 0x100)
++      {
++        #ifdef _LZMA_LOC_OPT
++        CProb *prob = probs + symbol;
++        RC_GET_BIT(prob, symbol)
++        #else
++        symbol = (symbol + symbol) | RangeDecoderBitDecode(probs + symbol, rd);
++        #endif
++      }
++      break;
++    }
++  }
++  while (symbol < 0x100);
++  #ifdef _LZMA_LOC_OPT
++  RC_FLUSH_VAR
++  #endif
++  return symbol;
++}
++
++#define kNumPosBitsMax 4
++#define kNumPosStatesMax (1 << kNumPosBitsMax)
++
++#define kLenNumLowBits 3
++#define kLenNumLowSymbols (1 << kLenNumLowBits)
++#define kLenNumMidBits 3
++#define kLenNumMidSymbols (1 << kLenNumMidBits)
++#define kLenNumHighBits 8
++#define kLenNumHighSymbols (1 << kLenNumHighBits)
++
++#define LenChoice 0
++#define LenChoice2 (LenChoice + 1)
++#define LenLow (LenChoice2 + 1)
++#define LenMid (LenLow + (kNumPosStatesMax << kLenNumLowBits))
++#define LenHigh (LenMid + (kNumPosStatesMax << kLenNumMidBits))
++#define kNumLenProbs (LenHigh + kLenNumHighSymbols) 
++
++int LzmaLenDecode(CProb *p, CRangeDecoder *rd, int posState)
++{
++  if(RangeDecoderBitDecode(p + LenChoice, rd) == 0)
++    return RangeDecoderBitTreeDecode(p + LenLow +
++        (posState << kLenNumLowBits), kLenNumLowBits, rd);
++  if(RangeDecoderBitDecode(p + LenChoice2, rd) == 0)
++    return kLenNumLowSymbols + RangeDecoderBitTreeDecode(p + LenMid +
++        (posState << kLenNumMidBits), kLenNumMidBits, rd);
++  return kLenNumLowSymbols + kLenNumMidSymbols + 
++      RangeDecoderBitTreeDecode(p + LenHigh, kLenNumHighBits, rd);
++}
++
++#define kNumStates 12
++
++#define kStartPosModelIndex 4
++#define kEndPosModelIndex 14
++#define kNumFullDistances (1 << (kEndPosModelIndex >> 1))
++
++#define kNumPosSlotBits 6
++#define kNumLenToPosStates 4
++
++#define kNumAlignBits 4
++#define kAlignTableSize (1 << kNumAlignBits)
++
++#define kMatchMinLen 2
++
++#define IsMatch 0
++#define IsRep (IsMatch + (kNumStates << kNumPosBitsMax))
++#define IsRepG0 (IsRep + kNumStates)
++#define IsRepG1 (IsRepG0 + kNumStates)
++#define IsRepG2 (IsRepG1 + kNumStates)
++#define IsRep0Long (IsRepG2 + kNumStates)
++#define PosSlot (IsRep0Long + (kNumStates << kNumPosBitsMax))
++#define SpecPos (PosSlot + (kNumLenToPosStates << kNumPosSlotBits))
++#define Align (SpecPos + kNumFullDistances - kEndPosModelIndex)
++#define LenCoder (Align + kAlignTableSize)
++#define RepLenCoder (LenCoder + kNumLenProbs)
++#define Literal (RepLenCoder + kNumLenProbs)
++
++#if Literal != LZMA_BASE_SIZE
++StopCompilingDueBUG
++#endif
++
++#ifdef _LZMA_OUT_READ
++
++typedef struct _LzmaVarState
++{
++  CRangeDecoder RangeDecoder;
++  Byte *Dictionary;
++  UInt32 DictionarySize;
++  UInt32 DictionaryPos;
++  UInt32 GlobalPos;
++  UInt32 Reps[4];
++  int lc;
++  int lp;
++  int pb;
++  int State;
++  int PreviousIsMatch;
++  int RemainLen;
++} LzmaVarState;
++
++int LzmaDecoderInit(
++    unsigned char *buffer, UInt32 bufferSize,
++    int lc, int lp, int pb,
++    unsigned char *dictionary, UInt32 dictionarySize,
++    #ifdef _LZMA_IN_CB
++    ILzmaInCallback *inCallback
++    #else
++    unsigned char *inStream, UInt32 inSize
++    #endif
++    )
++{
++  LzmaVarState *vs = (LzmaVarState *)buffer;
++  CProb *p = (CProb *)(buffer + sizeof(LzmaVarState));
++  UInt32 numProbs = Literal + ((UInt32)LZMA_LIT_SIZE << (lc + lp));
++  UInt32 i;
++  if (bufferSize < numProbs * sizeof(CProb) + sizeof(LzmaVarState))
++    return LZMA_RESULT_NOT_ENOUGH_MEM;
++  vs->Dictionary = dictionary;
++  vs->DictionarySize = dictionarySize;
++  vs->DictionaryPos = 0;
++  vs->GlobalPos = 0;
++  vs->Reps[0] = vs->Reps[1] = vs->Reps[2] = vs->Reps[3] = 1;
++  vs->lc = lc;
++  vs->lp = lp;
++  vs->pb = pb;
++  vs->State = 0;
++  vs->PreviousIsMatch = 0;
++  vs->RemainLen = 0;
++  dictionary[dictionarySize - 1] = 0;
++  for (i = 0; i < numProbs; i++)
++    p[i] = kBitModelTotal >> 1; 
++  RangeDecoderInit(&vs->RangeDecoder, 
++      #ifdef _LZMA_IN_CB
++      inCallback
++      #else
++      inStream, inSize
++      #endif
++  );
++  return LZMA_RESULT_OK;
++}
++
++int LzmaDecode(unsigned char *buffer, 
++    unsigned char *outStream, UInt32 outSize,
++    UInt32 *outSizeProcessed)
++{
++  LzmaVarState *vs = (LzmaVarState *)buffer;
++  CProb *p = (CProb *)(buffer + sizeof(LzmaVarState));
++  CRangeDecoder rd = vs->RangeDecoder;
++  int state = vs->State;
++  int previousIsMatch = vs->PreviousIsMatch;
++  Byte previousByte;
++  UInt32 rep0 = vs->Reps[0], rep1 = vs->Reps[1], rep2 = vs->Reps[2], rep3 = vs->Reps[3];
++  UInt32 nowPos = 0;
++  UInt32 posStateMask = (1 << (vs->pb)) - 1;
++  UInt32 literalPosMask = (1 << (vs->lp)) - 1;
++  int lc = vs->lc;
++  int len = vs->RemainLen;
++  UInt32 globalPos = vs->GlobalPos;
++
++  Byte *dictionary = vs->Dictionary;
++  UInt32 dictionarySize = vs->DictionarySize;
++  UInt32 dictionaryPos = vs->DictionaryPos;
++
++  if (len == -1)
++  {
++    *outSizeProcessed = 0;
++    return LZMA_RESULT_OK;
++  }
++
++  while(len > 0 && nowPos < outSize)
++  {
++    UInt32 pos = dictionaryPos - rep0;
++    if (pos >= dictionarySize)
++      pos += dictionarySize;
++    outStream[nowPos++] = dictionary[dictionaryPos] = dictionary[pos];
++    if (++dictionaryPos == dictionarySize)
++      dictionaryPos = 0;
++    len--;
++  }
++  if (dictionaryPos == 0)
++    previousByte = dictionary[dictionarySize - 1];
++  else
++    previousByte = dictionary[dictionaryPos - 1];
++#else
++
++int LzmaDecode(
++    Byte *buffer, UInt32 bufferSize,
++    int lc, int lp, int pb,
++    #ifdef _LZMA_IN_CB
++    ILzmaInCallback *inCallback,
++    #else
++    unsigned char *inStream, UInt32 inSize,
++    #endif
++    unsigned char *outStream, UInt32 outSize,
++    UInt32 *outSizeProcessed)
++{
++  UInt32 numProbs = Literal + ((UInt32)LZMA_LIT_SIZE << (lc + lp));
++  CProb *p = (CProb *)buffer;
++  CRangeDecoder rd;
++  UInt32 i;
++  int state = 0;
++  int previousIsMatch = 0;
++  Byte previousByte = 0;
++  UInt32 rep0 = 1, rep1 = 1, rep2 = 1, rep3 = 1;
++  UInt32 nowPos = 0;
++  UInt32 posStateMask = (1 << pb) - 1;
++  UInt32 literalPosMask = (1 << lp) - 1;
++  int len = 0;
++  if (bufferSize < numProbs * sizeof(CProb))
++    return LZMA_RESULT_NOT_ENOUGH_MEM;
++  for (i = 0; i < numProbs; i++)
++    p[i] = kBitModelTotal >> 1; 
++  RangeDecoderInit(&rd, 
++      #ifdef _LZMA_IN_CB
++      inCallback
++      #else
++      inStream, inSize
++      #endif
++      );
++#endif
++
++  *outSizeProcessed = 0;
++  while(nowPos < outSize)
++  {
++    int posState = (int)(
++        (nowPos 
++        #ifdef _LZMA_OUT_READ
++        + globalPos
++        #endif
++        )
++        & posStateMask);
++    #ifdef _LZMA_IN_CB
++    if (rd.Result != LZMA_RESULT_OK)
++      return rd.Result;
++    #endif
++    if (rd.ExtraBytes != 0)
++      return LZMA_RESULT_DATA_ERROR;
++    if (RangeDecoderBitDecode(p + IsMatch + (state << kNumPosBitsMax) + posState, &rd) == 0)
++    {
++      CProb *probs = p + Literal + (LZMA_LIT_SIZE * 
++        (((
++        (nowPos 
++        #ifdef _LZMA_OUT_READ
++        + globalPos
++        #endif
++        )
++        & literalPosMask) << lc) + (previousByte >> (8 - lc))));
++
++      if (state < 4) state = 0;
++      else if (state < 10) state -= 3;
++      else state -= 6;
++      if (previousIsMatch)
++      {
++        Byte matchByte;
++        #ifdef _LZMA_OUT_READ
++        UInt32 pos = dictionaryPos - rep0;
++        if (pos >= dictionarySize)
++          pos += dictionarySize;
++        matchByte = dictionary[pos];
++        #else
++        matchByte = outStream[nowPos - rep0];
++        #endif
++        previousByte = LzmaLiteralDecodeMatch(probs, &rd, matchByte);
++        previousIsMatch = 0;
++      }
++      else
++        previousByte = LzmaLiteralDecode(probs, &rd);
++      outStream[nowPos++] = previousByte;
++      #ifdef _LZMA_OUT_READ
++      dictionary[dictionaryPos] = previousByte;
++      if (++dictionaryPos == dictionarySize)
++        dictionaryPos = 0;
++      #endif
++    }
++    else             
++    {
++      previousIsMatch = 1;
++      if (RangeDecoderBitDecode(p + IsRep + state, &rd) == 1)
++      {
++        if (RangeDecoderBitDecode(p + IsRepG0 + state, &rd) == 0)
++        {
++          if (RangeDecoderBitDecode(p + IsRep0Long + (state << kNumPosBitsMax) + posState, &rd) == 0)
++          {
++            #ifdef _LZMA_OUT_READ
++            UInt32 pos;
++            #endif
++            if (
++               (nowPos 
++                #ifdef _LZMA_OUT_READ
++                + globalPos
++                #endif
++               )
++               == 0)
++              return LZMA_RESULT_DATA_ERROR;
++            state = state < 7 ? 9 : 11;
++            #ifdef _LZMA_OUT_READ
++            pos = dictionaryPos - rep0;
++            if (pos >= dictionarySize)
++              pos += dictionarySize;
++            previousByte = dictionary[pos];
++            dictionary[dictionaryPos] = previousByte;
++            if (++dictionaryPos == dictionarySize)
++              dictionaryPos = 0;
++            #else
++            previousByte = outStream[nowPos - rep0];
++            #endif
++            outStream[nowPos++] = previousByte;
++            continue;
++          }
++        }
++        else
++        {
++          UInt32 distance;
++          if(RangeDecoderBitDecode(p + IsRepG1 + state, &rd) == 0)
++            distance = rep1;
++          else 
++          {
++            if(RangeDecoderBitDecode(p + IsRepG2 + state, &rd) == 0)
++              distance = rep2;
++            else
++            {
++              distance = rep3;
++              rep3 = rep2;
++            }
++            rep2 = rep1;
++          }
++          rep1 = rep0;
++          rep0 = distance;
++        }
++        len = LzmaLenDecode(p + RepLenCoder, &rd, posState);
++        state = state < 7 ? 8 : 11;
++      }
++      else
++      {
++        int posSlot;
++        rep3 = rep2;
++        rep2 = rep1;
++        rep1 = rep0;
++        state = state < 7 ? 7 : 10;
++        len = LzmaLenDecode(p + LenCoder, &rd, posState);
++        posSlot = RangeDecoderBitTreeDecode(p + PosSlot +
++            ((len < kNumLenToPosStates ? len : kNumLenToPosStates - 1) << 
++            kNumPosSlotBits), kNumPosSlotBits, &rd);
++        if (posSlot >= kStartPosModelIndex)
++        {
++          int numDirectBits = ((posSlot >> 1) - 1);
++          rep0 = ((2 | ((UInt32)posSlot & 1)) << numDirectBits);
++          if (posSlot < kEndPosModelIndex)
++          {
++            rep0 += RangeDecoderReverseBitTreeDecode(
++                p + SpecPos + rep0 - posSlot - 1, numDirectBits, &rd);
++          }
++          else
++          {
++            rep0 += RangeDecoderDecodeDirectBits(&rd, 
++                numDirectBits - kNumAlignBits) << kNumAlignBits;
++            rep0 += RangeDecoderReverseBitTreeDecode(p + Align, kNumAlignBits, &rd);
++          }
++        }
++        else
++          rep0 = posSlot;
++        rep0++;
++      }
++      if (rep0 == (UInt32)(0))
++      {
++        /* it's for stream version */
++        len = -1;
++        break;
++      }
++      if (rep0 > nowPos 
++        #ifdef _LZMA_OUT_READ
++        + globalPos
++        #endif
++        )
++      {
++        return LZMA_RESULT_DATA_ERROR;
++      }
++      len += kMatchMinLen;
++      do
++      {
++        #ifdef _LZMA_OUT_READ
++        UInt32 pos = dictionaryPos - rep0;
++        if (pos >= dictionarySize)
++          pos += dictionarySize;
++        previousByte = dictionary[pos];
++        dictionary[dictionaryPos] = previousByte;
++        if (++dictionaryPos == dictionarySize)
++          dictionaryPos = 0;
++        #else
++        previousByte = outStream[nowPos - rep0];
++        #endif
++        outStream[nowPos++] = previousByte;
++        len--;
++      }
++      while(len > 0 && nowPos < outSize);
++    }
++  }
++
++  #ifdef _LZMA_OUT_READ
++  vs->RangeDecoder = rd;
++  vs->DictionaryPos = dictionaryPos;
++  vs->GlobalPos = globalPos + nowPos;
++  vs->Reps[0] = rep0;
++  vs->Reps[1] = rep1;
++  vs->Reps[2] = rep2;
++  vs->Reps[3] = rep3;
++  vs->State = state;
++  vs->PreviousIsMatch = previousIsMatch;
++  vs->RemainLen = len;
++  #endif
++
++  *outSizeProcessed = nowPos;
++  return LZMA_RESULT_OK;
++}
+diff -Nur linux-2.6.16/fs/squashfs/LzmaDecode.h linux-2.6.16-owrt/fs/squashfs/LzmaDecode.h
+--- linux-2.6.16/fs/squashfs/LzmaDecode.h	1970-01-01 01:00:00.000000000 +0100
++++ linux-2.6.16-owrt/fs/squashfs/LzmaDecode.h	2006-03-21 10:56:57.000000000 +0100
+@@ -0,0 +1,100 @@
++/* 
++  LzmaDecode.h
++  LZMA Decoder interface
++
++  LZMA SDK 4.05 Copyright (c) 1999-2004 Igor Pavlov (2004-08-25)
++  http://www.7-zip.org/
++
++  LZMA SDK is licensed under two licenses:
++  1) GNU Lesser General Public License (GNU LGPL)
++  2) Common Public License (CPL)
++  It means that you can select one of these two licenses and 
++  follow rules of that license.
++
++  SPECIAL EXCEPTION:
++  Igor Pavlov, as the author of this code, expressly permits you to 
++  statically or dynamically link your code (or bind by name) to the 
++  interfaces of this file without subjecting your linked code to the 
++  terms of the CPL or GNU LGPL. Any modifications or additions 
++  to this file, however, are subject to the LGPL or CPL terms.
++*/
++
++#ifndef __LZMADECODE_H
++#define __LZMADECODE_H
++
++/* #define _LZMA_IN_CB */
++/* Use callback for input data */
++
++/* #define _LZMA_OUT_READ */
++/* Use read function for output data */
++
++/* #define _LZMA_PROB32 */
++/* It can increase speed on some 32-bit CPUs, 
++   but memory usage will be doubled in that case */
++
++/* #define _LZMA_LOC_OPT */
++/* Enable local speed optimizations inside code */
++
++#ifndef UInt32
++#ifdef _LZMA_UINT32_IS_ULONG
++#define UInt32 unsigned long
++#else
++#define UInt32 unsigned int
++#endif
++#endif
++
++#ifdef _LZMA_PROB32
++#define CProb UInt32
++#else
++#define CProb unsigned short
++#endif
++
++#define LZMA_RESULT_OK 0
++#define LZMA_RESULT_DATA_ERROR 1
++#define LZMA_RESULT_NOT_ENOUGH_MEM 2
++
++#ifdef _LZMA_IN_CB
++typedef struct _ILzmaInCallback
++{
++  int (*Read)(void *object, unsigned char **buffer, UInt32 *bufferSize);
++} ILzmaInCallback;
++#endif
++
++#define LZMA_BASE_SIZE 1846
++#define LZMA_LIT_SIZE 768
++
++/* 
++bufferSize = (LZMA_BASE_SIZE + (LZMA_LIT_SIZE << (lc + lp)))* sizeof(CProb)
++bufferSize += 100 in case of _LZMA_OUT_READ
++by default CProb is unsigned short, 
++but if specify _LZMA_PROB_32, CProb will be UInt32(unsigned int)
++*/
++
++#ifdef _LZMA_OUT_READ
++int LzmaDecoderInit(
++    unsigned char *buffer, UInt32 bufferSize,
++    int lc, int lp, int pb,
++    unsigned char *dictionary, UInt32 dictionarySize,
++  #ifdef _LZMA_IN_CB
++    ILzmaInCallback *inCallback
++  #else
++    unsigned char *inStream, UInt32 inSize
++  #endif
++);
++#endif
++
++int LzmaDecode(
++    unsigned char *buffer, 
++  #ifndef _LZMA_OUT_READ
++    UInt32 bufferSize,
++    int lc, int lp, int pb,
++  #ifdef _LZMA_IN_CB
++    ILzmaInCallback *inCallback,
++  #else
++    unsigned char *inStream, UInt32 inSize,
++  #endif
++  #endif
++    unsigned char *outStream, UInt32 outSize,
++    UInt32 *outSizeProcessed);
++
++#endif
+diff -Nur linux-2.6.16/fs/squashfs/Makefile linux-2.6.16-owrt/fs/squashfs/Makefile
+--- linux-2.6.16/fs/squashfs/Makefile	2006-03-21 10:55:59.000000000 +0100
++++ linux-2.6.16-owrt/fs/squashfs/Makefile	2006-03-21 10:57:08.000000000 +0100
+@@ -5,3 +5,4 @@
+ obj-$(CONFIG_SQUASHFS) += squashfs.o
+ squashfs-y += inode.o
+ squashfs-y += squashfs2_0.o
++squashfs-y += LzmaDecode.o
diff --git a/packages/squashfs-tools/files/Makefile-3.1.patch b/packages/squashfs-tools/files/Makefile-3.1.patch
new file mode 100644
index 0000000000..1b7c49878f
--- /dev/null
+++ b/packages/squashfs-tools/files/Makefile-3.1.patch
@@ -0,0 +1,34 @@
+Modifies the Makefile to work with OpenEmbedded.
+
+I copied the CFLAGS_R approach from the existing 2.0 squashfs-tools, 
+without actually knowing why this approach was taken. It works though.
+
+Leon Woestenberg <leonw@mailcan.com>
+
+--- squashfs-tools/Makefile	2006-10-10 13:59:05.000000000 +0200
++++ squashfs-tools/Makefile.oe	2006-10-10 14:06:01.000000000 +0200
+@@ -1,17 +1,22 @@
+ INCLUDEDIR = .
++CFLAGS_R = -I$(INCLUDEDIR) -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE
++LIBS = -lz -lpthread
+ 
+ CFLAGS := -I$(INCLUDEDIR) -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -O2
+ 
+ all: mksquashfs unsquashfs
+ 
+ mksquashfs: mksquashfs.o read_fs.o sort.o
+-	$(CC) mksquashfs.o read_fs.o sort.o -lz -lpthread -o $@
++	$(CC) $(LDFLAGS) mksquashfs.o read_fs.o sort.o $(LIBS) -o $@
+ 
+ mksquashfs.o: mksquashfs.c squashfs_fs.h mksquashfs.h global.h sort.h
++	$(CC) $(CFLAGS_R) $(CFLAGS) -c -o mksquashfs.o mksquashfs.c
+ 
+ read_fs.o: read_fs.c squashfs_fs.h read_fs.h global.h
+-
++	$(CC) $(CFLAGS_R) $(CFLAGS) -c -o read_fs.o read_fs.c
++	
+ sort.o: sort.c squashfs_fs.h global.h sort.h
++	$(CC) $(CFLAGS_R) $(CFLAGS) -c -o sort.o sort.c
+ 
+ unsquashfs: unsquashfs.o
+ 	$(CC) unsquashfs.o -lz -o $@
diff --git a/packages/squashfs-tools/files/Makefile.patch b/packages/squashfs-tools/files/Makefile.patch
new file mode 100644
index 0000000000..beabea88f9
--- /dev/null
+++ b/packages/squashfs-tools/files/Makefile.patch
@@ -0,0 +1,31 @@
+Index: squashfs-tools/Makefile
+===================================================================
+--- squashfs-tools.orig/Makefile	2006-08-21 01:42:45.000000000 +0200
++++ squashfs-tools/Makefile	2006-09-21 11:42:10.000000000 +0200
+@@ -1,19 +1,24 @@
+ INCLUDEDIR = .
+ 
+ CFLAGS := -I$(INCLUDEDIR) -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -O2
++CFLAGS_R = -I$(INCLUDEDIR) -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE
++LIBS = -lz
+ 
+ all: mksquashfs unsquashfs
+ 
+ mksquashfs: mksquashfs.o read_fs.o sort.o
+-	$(CC) mksquashfs.o read_fs.o sort.o -lz -lpthread -o $@
++	$(CC) $(LDFLAGS) mksquashfs.o read_fs.o sort.o -lz -lpthread -o $@
+ 
+ mksquashfs.o: mksquashfs.c squashfs_fs.h mksquashfs.h global.h sort.h
++ 	$(CC) $(CFLAGS_R) $(CFLAGS) -c -o mksquashfs.o mksquashfs.c
+ 
+ read_fs.o: read_fs.c squashfs_fs.h read_fs.h global.h
++ 	$(CC) $(CFLAGS_R) $(CFLAGS) -c -o read_fs.o read_fs.c
+ 
+ sort.o: sort.c squashfs_fs.h global.h sort.h
++ 	$(CC) $(CFLAGS_R) $(CFLAGS) -c -o sort.o sort.c
+ 
+ unsquashfs: unsquashfs.o
+-	$(CC) unsquashfs.o -lz -o $@
++	$(CC) $(LDFLAGS) unsquashfs.o -lz -o $@
+ 
+ unsquashfs.o: unsquashfs.c squashfs_fs.h read_fs.h global.h
diff --git a/packages/squashfs-tools/files/squashfs3.1r2-tools-lzma_Makefile.patch b/packages/squashfs-tools/files/squashfs3.1r2-tools-lzma_Makefile.patch
new file mode 100644
index 0000000000..eeb578ab12
--- /dev/null
+++ b/packages/squashfs-tools/files/squashfs3.1r2-tools-lzma_Makefile.patch
@@ -0,0 +1,40 @@
+Modifies the Makefile so that it builds {mk,un}squashfs-lzma instead
+of {mk,un}squashfs. This requires the common (non-lzma) patch called
+Makefile-3.1 to be applied first.
+
+Leon Woestenberg <leonw@mailcan.com>
+
+--- squashfs-tools/Makefile	2006-10-10 16:07:43.000000000 +0200
++++ squashfs-lzma-tools/Makefile	2006-10-10 15:44:57.000000000 +0200
+@@ -1,13 +1,15 @@
+ INCLUDEDIR = .
+ CFLAGS_R = -I$(INCLUDEDIR) -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE
+-LIBS = -lz -lpthread
++#LZMAPATH = ./lzma/C/7zip/Compress/LZMA_Lib
++LIBS = -llzma -lz -lpthread
+ 
+ CFLAGS := -I$(INCLUDEDIR) -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -O2
+ 
+ all: mksquashfs unsquashfs
+ 
+-mksquashfs: mksquashfs.o read_fs.o sort.o
+-	$(CC) $(LDFLAGS) mksquashfs.o read_fs.o sort.o $(LIBS) -o $@
++mksquashfs-lzma: mksquashfs.o read_fs.o sort.o
++#	make -C $(LZMAPATH)
++	$(CXX) $(LDFLAGS) mksquashfs.o read_fs.o sort.o $(LIBS) -o $@
+ 
+ mksquashfs.o: mksquashfs.c squashfs_fs.h mksquashfs.h global.h sort.h
+ 	$(CC) $(CFLAGS_R) $(CFLAGS) -c -o mksquashfs.o mksquashfs.c
+@@ -18,7 +20,10 @@
+ sort.o: sort.c squashfs_fs.h global.h sort.h
+ 	$(CC) $(CFLAGS_R) $(CFLAGS) -c -o sort.o sort.c
+ 
+-unsquashfs: unsquashfs.o
+-	$(CC) unsquashfs.o -lz -o $@
++unsquashfs-lzma: unsquashfs.o
++#	make -C $(LZMAPATH)
++	$(CXX) unsquashfs.o $(LIBS) -o $@
+ 
+ unsquashfs.o: unsquashfs.c squashfs_fs.h read_fs.h global.h
++
++clean:
diff --git a/packages/squashfs-tools/files/squashfs3.1r2-tools-replace_zlib_with_lzma.patch b/packages/squashfs-tools/files/squashfs3.1r2-tools-replace_zlib_with_lzma.patch
new file mode 100644
index 0000000000..2947de0858
--- /dev/null
+++ b/packages/squashfs-tools/files/squashfs3.1r2-tools-replace_zlib_with_lzma.patch
@@ -0,0 +1,1785 @@
+Modify the source code to use the lzma {,de}compression routines.
+
+Leon Woestenberg <leonw@mailcan.com>
+
+diff -Nur squashfs-tools/mksquashfs.c squashfs-lzma-tools/mksquashfs.c
+--- squashfs-tools/mksquashfs.c	2006-08-31 00:07:37.000000000 +0200
++++ squashfs-lzma-tools/mksquashfs.c	2006-08-22 09:35:14.000000000 +0200
+@@ -40,16 +40,13 @@
+ #include <signal.h>
+ #include <setjmp.h>
+ #include <sys/mman.h>
+-#include <pthread.h>
+ 
+ #ifndef linux
+ #define __BYTE_ORDER BYTE_ORDER
+ #define __BIG_ENDIAN BIG_ENDIAN
+ #define __LITTLE_ENDIAN LITTLE_ENDIAN
+-#include <sys/sysctl.h>
+ #else
+ #include <endian.h>
+-#include <sys/sysinfo.h>
+ #endif
+ 
+ #include <squashfs_fs.h>
+@@ -149,7 +146,7 @@
+ struct inode_info *inode_info[INODE_HASH_SIZE];
+ 
+ /* hash tables used to do fast duplicate searches in duplicate check */
+-struct file_info *dupl[65536];
++struct file_info *dupl[65536], *frag_dups[65536];
+ int dup_files = 0;
+ 
+ /* list of exclude dirs/files */
+@@ -165,7 +162,7 @@
+ 
+ /* fragment block data structures */
+ int fragments = 0;
+-struct file_buffer *fragment_data = NULL;
++char fragment_data[SQUASHFS_FILE_SIZE];
+ int fragment_size = 0;
+ struct fragment {
+ 	unsigned int		index;
+@@ -174,7 +171,6 @@
+ };
+ #define FRAG_SIZE 32768
+ squashfs_fragment_entry *fragment_table = NULL;
+-int fragments_outstanding = 0;
+ 
+ 
+ /* current inode number for directories and non directories */
+@@ -198,15 +194,13 @@
+ 
+ /* in memory file info */
+ struct file_info {
+-	long long		file_size;
+ 	long long		bytes;
+ 	unsigned short		checksum;
+-	unsigned short		fragment_checksum;
+ 	long long		start;
+ 	unsigned int		*block_list;
+ 	struct file_info	*next;
+ 	struct fragment		*fragment;
+-	char			checksum_flag;
++	unsigned short		fragment_checksum;
+ };
+ 
+ /* count of how many times SIGINT or SIGQUIT has been sent */
+@@ -225,7 +219,6 @@
+ 	sdir_count, sfifo_count, ssock_count, sdup_files;
+ int sfragments;
+ int restore = 0;
+-int threads;
+ 
+ /* flag whether destination file is a block device */
+ int block_device = 0;
+@@ -236,67 +229,14 @@
+ /* save destination file name for deleting on error */
+ char *destination_file = NULL;
+ 
+-/* data allocator status struct.  Allocators are used to keep
+-  track of memory buffers passed between different threads */
+-struct allocator {
+-	int		max_buffers;
+-	int		count;
+-	int		buffer_size;
+-	pthread_mutex_t	mutex;
+-	pthread_cond_t	wait;
+-};
+-
+-/* struct describing a memory buffer passed between threads */
+-struct file_buffer {
+-	struct allocator	*allocator;
+-	void			(*release)(int);
+-	int			release_data;
+-	long long		block;
+-	int			size;
+-	int			c_byte;
+-	unsigned int		block_order;
+-	int			fragment;
+-	int			error;
+-	struct file_buffer	*next;
+-	char			data[0];
+-};
+-
+-/* struct describing queues used to pass data between threads */
+-struct queue {
+-	int			size;
+-	int			readp;
+-	int			writep;
+-	pthread_mutex_t		mutex;
+-	pthread_cond_t		empty;
+-	pthread_cond_t		full;
+-	void			**data;
+-};
+-
+-/* describes the list of blocks in a file which is a possible
+-   duplicate.  For each block, it indicates whether the block is
+-   in memory or on disk */
+-struct buffer_list {
+-	long long start;
+-	int size;
+-	struct file_buffer *read_buffer;
++/* structure to used to pass in a pointer or an integer
++ * to duplicate buffer read helper functions.
++ */
++struct duplicate_buffer_handle {
++	char	*ptr;
++	long long	start;
+ };
+ 
+-struct allocator *reader_buffer, *writer_buffer, *fragment_buffer;
+-struct queue *to_reader, *from_reader, *to_writer, *from_writer, *from_deflate, *to_frag;
+-pthread_t *thread, *deflator_thread, *frag_deflator_thread;
+-pthread_mutex_t	fragment_mutex;
+-pthread_cond_t fragment_waiting;
+-pthread_mutex_t	pos_mutex;
+-
+-/* user options that control parallelisation */
+-int processors = -1;
+-/* default size of output buffer in Mbytes */
+-#define WRITER_BUFFER_DEFAULT 512
+-/* default size of input buffer in Mbytes */
+-#define READER_BUFFER_DEFAULT 64
+-int writer_buffer_size;
+-int reader_buffer_size;
+-
+ void add_old_root_entry(char *name, squashfs_inode inode, int inode_number, int type);
+ extern int read_super(int fd, squashfs_super_block *sBlk, int *be, char *source);
+ extern long long read_filesystem(char *root_name, int fd, squashfs_super_block *sBlk, char **cinode_table,
+@@ -309,166 +249,18 @@
+ 		unsigned int *inode_dir_inode_number, unsigned int *inode_dir_parent_inode,
+ 		void (push_directory_entry)(char *, squashfs_inode, int, int),
+ 		squashfs_fragment_entry **fragment_table);
+-extern int read_sort_file(char *filename, int source, char *source_path[]);
+-extern void sort_files_and_write(struct dir_info *dir);
+-struct file_info *duplicate(long long file_size, long long bytes, unsigned int **block_list, long long *start, struct fragment **fragment, struct file_buffer *file_buffer, struct buffer_list *buffer_list, int blocks, unsigned short checksum, unsigned short fragment_checksum, int checksum_flag);
++int get_sorted_inode(squashfs_inode *inode, struct stat *buf);
++int read_sort_file(char *filename, int source, char *source_path[]);
++void sort_files_and_write(struct dir_info *dir);
++struct file_info *duplicate(char *(get_next_file_block)(struct duplicate_buffer_handle *, unsigned int), struct duplicate_buffer_handle *file_start, long long bytes, unsigned int **block_list, long long *start, int blocks, struct fragment **fragment, char *frag_data, int frag_bytes);
+ struct dir_info *dir_scan1(char *, int (_readdir)(char *, char *, struct dir_info *));
+-struct file_info *add_non_dup(long long file_size, long long bytes, unsigned int *block_list, long long start, struct fragment *fragment, unsigned short checksum, unsigned short fragment_checksum, int checksum_flag);
+-extern void generate_file_priorities(struct dir_info *dir, int priority, struct stat *buf);
+-extern struct priority_entry *priority_list[65536];
+-
+-
+-struct allocator *alloc_init(int buffer_size, int max_buffers)
+-{
+-	struct allocator *allocator = malloc(sizeof(struct allocator));
+-
+-	if(allocator == NULL)
+-		return NULL;
+-
+-	allocator->max_buffers = max_buffers;
+-	allocator->buffer_size = buffer_size;
+-	allocator->count = 0;
+-	pthread_mutex_init(&allocator->mutex, NULL);
+-	pthread_cond_init(&allocator->wait, NULL);
+-
+-	return allocator;
+-}
+-
+-
+-struct file_buffer *alloc_get(struct allocator *allocator)
+-{
+-	struct file_buffer *file_buffer;
+-	
+-	pthread_mutex_lock(&allocator->mutex);
+-
+-	while(allocator->count == allocator->max_buffers)
+-		pthread_cond_wait(&allocator->wait, &allocator->mutex);
+-
+-	if((file_buffer = malloc(sizeof(struct file_buffer) + allocator->buffer_size)) == NULL)
+-		goto failed;
+-
+-	file_buffer->release = NULL;
+-	file_buffer->allocator = allocator;
+-	allocator->count ++;
+-
+-failed:
+-	pthread_mutex_unlock(&allocator->mutex);
+-	return file_buffer;
+-}
+-
+-
+-struct file_buffer *alloc_get_2(struct allocator *allocator, void (*release)(int), int release_data)
+-{
+-	struct file_buffer *file_buffer = alloc_get(allocator);
+-
+-	if(file_buffer) {
+-		file_buffer->release = release;
+-		file_buffer->release_data = release_data;
+-	}
+-	return file_buffer;
+-}
+-
+-
+-void alloc_free(struct file_buffer *file_buffer)
+-{
+-	struct allocator *allocator;
+-
+-	if(file_buffer == NULL)
+-		return;
+-
+-	allocator = file_buffer->allocator;
+-
+-	if(file_buffer->release)
+-		file_buffer->release(file_buffer->release_data);
+-
+-	pthread_mutex_lock(&allocator->mutex);
+-	free(file_buffer);
+-	if(allocator->count == 0)
+-		ERROR("alloc_free: freeing buffer for empty allocator!\n");
+-	else
+-		allocator->count --;
+-	pthread_cond_signal(&allocator->wait);
+-	pthread_mutex_unlock(&allocator->mutex);
+-}
+-
+-
+-struct queue *queue_init(int size)
+-{
+-	struct queue *queue = malloc(sizeof(struct queue));
+-
+-	if(queue == NULL)
+-		return NULL;
+-
+-	if((queue->data = malloc(sizeof(void *) * (size + 1))) == NULL) {
+-		free(queue);
+-		return NULL;
+-	}
+-
+-	queue->size = size + 1;
+-	queue->readp = queue->writep = 0;
+-	pthread_mutex_init(&queue->mutex, NULL);
+-	pthread_cond_init(&queue->empty, NULL);
+-	pthread_cond_init(&queue->full, NULL);
+-
+-	return queue;
+-}
+-
+-
+-void queue_put(struct queue *queue, void *data)
+-{
+-	int nextp;
+-
+-	pthread_mutex_lock(&queue->mutex);
+-
+-	while((nextp = (queue->writep + 1) % queue->size) == queue->readp)
+-		pthread_cond_wait(&queue->full, &queue->mutex);
+-
+-	queue->data[queue->writep] = data;
+-	queue->writep = nextp;
+-	pthread_cond_signal(&queue->empty);
+-	pthread_mutex_unlock(&queue->mutex);
+-}
+-
+-
+-void *queue_get(struct queue *queue)
+-{
+-	void *data;
+-	pthread_mutex_lock(&queue->mutex);
+-
+-	while(queue->readp == queue->writep)
+-		pthread_cond_wait(&queue->empty, &queue->mutex);
+-
+-	data = queue->data[queue->readp];
+-	queue->readp = (queue->readp + 1) % queue->size;
+-	pthread_cond_signal(&queue->full);
+-	pthread_mutex_unlock(&queue->mutex);
+-
+-	return data;
+-}
+-
+ 
+ #define MKINODE(A)	((squashfs_inode)(((squashfs_inode) inode_bytes << 16) + (((char *)A) - data_cache)))
+ 
+ 
+-inline void waitforthread(int i)
+-{
+-	TRACE("Waiting for thread %d\n", i);
+-	while(thread[i] != 0)
+-		sched_yield();
+-}
+-
+-
+ void restorefs()
+ {
+-	int i;
+-
+ 	ERROR("Exiting - restoring original filesystem!\n\n");
+-
+-	for(i = 0; i < 2 + processors * 2; i++)
+-		pthread_kill(thread[i], SIGUSR1);
+-	for(i = 0; i < 2 + processors * 2; i++)
+-		waitforthread(i);
+-	TRACE("All threads in signal handler\n");
+ 	bytes = sbytes;
+ 	memcpy(data_cache, sdata_cache, cache_bytes = scache_bytes);
+ 	memcpy(directory_data_cache, sdirectory_data_cache, directory_cache_bytes = sdirectory_cache_bytes);
+@@ -495,13 +287,12 @@
+ 
+ void sighandler()
+ {
+-	if(++interrupted > 2)
+-		return;
+-	if(interrupted == 2)
++	if(interrupted == 1)
+ 		restorefs();
+ 	else {
+ 		ERROR("Interrupting will restore original filesystem!\n");
+ 		ERROR("Interrupt again to quit\n");
++		interrupted ++;
+ 	}
+ }
+ 
+@@ -512,81 +303,22 @@
+ }
+ 
+ 
+-void sigusr1_handler()
+-{
+-	int i;
+-	sigset_t sigmask;
+-	pthread_t thread_id = pthread_self();
+-
+-	for(i = 0; i < (2 + processors * 2) && thread[i] != thread_id; i++);
+-	thread[i] = (pthread_t) 0;
+-
+-	TRACE("Thread %d(%p) in sigusr1_handler\n", i, thread_id);
+-
+-	sigemptyset(&sigmask);
+-	sigaddset(&sigmask, SIGINT);
+-	sigaddset(&sigmask, SIGQUIT);
+-	sigaddset(&sigmask, SIGUSR1);
+-	while(1) {
+-		sigsuspend(&sigmask);
+-		TRACE("After wait in sigusr1_handler :(\n");
+-	}
+-}
+-
+-
+-unsigned int mangle2(z_stream **strm, char *d, char *s, int size, int block_size, int uncompressed, int data_block)
++unsigned int mangle(char *d, char *s, int size, int block_size, int uncompressed, int data_block)
+ {
+-	unsigned long c_byte;
++	unsigned long c_byte = block_size << 1;
+ 	unsigned int res;
+-	z_stream *stream = *strm;
+-
+-	if(uncompressed)
+-		goto notcompressed;
+-
+-	if(stream == NULL) {
+-		if((stream = *strm = malloc(sizeof(z_stream))) == NULL)
+-			BAD_ERROR("mangle::compress failed, not enough memory\n");
+-
+-		stream->zalloc = Z_NULL;
+-		stream->zfree = Z_NULL;
+-		stream->opaque = 0;
+ 
+-		if((res = deflateInit(stream, 9)) != Z_OK) {
+-			if(res == Z_MEM_ERROR)
+-				BAD_ERROR("zlib::compress failed, not enough memory\n");
+-			else if(res == Z_STREAM_ERROR)
+-				BAD_ERROR("zlib::compress failed, not a valid compression level\n");
+-			else if(res == Z_VERSION_ERROR)
+-				BAD_ERROR("zlib::compress failed, incorrect zlib version\n");
+-			else
+-				BAD_ERROR("zlib::compress failed, unknown error %d\n", res);
+-		}
+-	} else if((res = deflateReset(stream)) != Z_OK) {
+-		if(res == Z_STREAM_ERROR)
+-			BAD_ERROR("zlib::compress failed, stream state inconsistent\n");
+-		else
+-			BAD_ERROR("zlib::compress failed, unknown error %d\n", res);
+-	}
+-
+-	stream->next_in = (unsigned char *) s;
+-	stream->avail_in = size;
+-	stream->next_out = (unsigned char *) d;
+-	stream->avail_out = block_size;
+-
+-	res = deflate(stream, Z_FINISH);
+-	if(res != Z_STREAM_END && res != Z_OK) {
+-		if(res == Z_STREAM_ERROR)
+-			BAD_ERROR("zlib::compress failed, stream state inconsistent\n");
++	if(!uncompressed && (res = compress2((unsigned char *) d, &c_byte, (unsigned char *) s, size, 9)) != Z_OK) {
++		if(res == Z_MEM_ERROR)
++			BAD_ERROR("zlib::compress failed, not enough memory\n");
+ 		else if(res == Z_BUF_ERROR)
+-			BAD_ERROR("zlib::compress failed, no progress possible\n");
++			BAD_ERROR("zlib::compress failed, not enough room in output buffer\n");
+ 		else
+ 			BAD_ERROR("zlib::compress failed, unknown error %d\n", res);
++		return 0;
+ 	}
+ 
+-	c_byte = stream->total_out;
+-
+-	if(res != Z_STREAM_END || c_byte >= size) {
+-notcompressed:
++	if(uncompressed || c_byte >= size) {
+ 		memcpy(d, s, size);
+ 		return size | (data_block ? SQUASHFS_COMPRESSED_BIT_BLOCK : SQUASHFS_COMPRESSED_BIT);
+ 	}
+@@ -595,14 +327,6 @@
+ }
+ 
+ 
+-unsigned int mangle(char *d, char *s, int size, int block_size, int uncompressed, int data_block)
+-{
+-	static z_stream *stream = NULL;
+-
+-	return mangle2(&stream, d, s, size, block_size, uncompressed, data_block);
+-}
+-
+-
+ squashfs_base_inode_header *get_inode(int req_size)
+ {
+ 	int data_space;
+@@ -655,7 +379,6 @@
+ {
+ 	off_t off = byte;
+ 
+-	pthread_mutex_lock(&pos_mutex);
+ 	if(lseek(fd, off, SEEK_SET) == -1) {
+ 		perror("Lseek on destination failed");
+ 		EXIT_MKSQUASHFS();
+@@ -665,7 +388,6 @@
+ 		perror("Read on destination failed");
+ 		EXIT_MKSQUASHFS();
+ 	}
+-	pthread_mutex_unlock(&pos_mutex);
+ }
+ 
+ 
+@@ -673,9 +395,6 @@
+ {
+ 	off_t off = byte;
+ 
+-	if(interrupted < 2)
+-		pthread_mutex_lock(&pos_mutex);
+-
+ 	if(lseek(fd, off, SEEK_SET) == -1) {
+ 		perror("Lseek on destination failed");
+ 		EXIT_MKSQUASHFS();
+@@ -685,9 +404,6 @@
+ 		perror("Write on destination failed");
+ 		EXIT_MKSQUASHFS();
+ 	}
+-	
+-	if(interrupted < 2)
+-		pthread_mutex_unlock(&pos_mutex);
+ }
+ 
+ 
+@@ -1173,23 +889,10 @@
+ }
+ 
+ 
+-char *get_fragment(char *buffer, struct fragment *fragment, int *cached_fragment)
++char *get_fragment(char *buffer, struct fragment *fragment)
+ {
+-	squashfs_fragment_entry *disk_fragment;
+-	int size;
+-
+-	if(fragment->index == *cached_fragment || fragment->index == SQUASHFS_INVALID_FRAG)
+-		return buffer + fragment->offset;
+-
+-	if(fragment_data && fragment->index == fragments)
+-		return fragment_data->data + fragment->offset;
+-
+-	pthread_mutex_lock(&fragment_mutex);
+-	while(fragment_table[fragment->index].pending)
+-		pthread_cond_wait(&fragment_waiting, &fragment_mutex);
+-	pthread_mutex_unlock(&fragment_mutex);
+-	disk_fragment = &fragment_table[fragment->index];
+-	size = SQUASHFS_COMPRESSED_SIZE_BLOCK(disk_fragment->size);
++	squashfs_fragment_entry *disk_fragment = &fragment_table[fragment->index];
++	int size = SQUASHFS_COMPRESSED_SIZE_BLOCK(disk_fragment->size);
+ 
+ 	if(SQUASHFS_COMPRESSED_BLOCK(disk_fragment->size)) {
+ 		int res;
+@@ -1209,74 +912,53 @@
+ 	} else
+ 		read_bytes(fd, disk_fragment->start_block, size, buffer);
+ 
+-	*cached_fragment = fragment->index;
+ 	return buffer + fragment->offset;
+ }
+ 
+ 	
+-void ensure_fragments_flushed()
+-{
+-	pthread_mutex_lock(&fragment_mutex);
+-	while(fragments_outstanding)
+-		pthread_cond_wait(&fragment_waiting, &fragment_mutex);
+-	pthread_mutex_unlock(&fragment_mutex);
+-}
+-
+-
+ void write_fragment()
+ {
++	int compressed_size;
++	char buffer[block_size << 1];
++
+ 	if(fragment_size == 0)
+ 		return;
+ 
+-	pthread_mutex_lock(&fragment_mutex);
+-	if(fragments % FRAG_SIZE == 0) {
+-		if((fragment_table = (squashfs_fragment_entry *) realloc(fragment_table, (fragments + FRAG_SIZE) * sizeof(squashfs_fragment_entry))) == NULL) {
+-			pthread_mutex_unlock(&fragment_mutex);
++	if(fragments % FRAG_SIZE == 0)
++		if((fragment_table = (squashfs_fragment_entry *) realloc(fragment_table, (fragments + FRAG_SIZE) * sizeof(squashfs_fragment_entry))) == NULL)
+ 			BAD_ERROR("Out of memory in fragment table\n");
+-		}
+-	}
+-	fragment_data->size = fragment_size;
+-	fragment_data->block = fragments;
+-	fragment_table[fragments].pending = TRUE;
+-	fragments_outstanding ++;
+-	queue_put(to_frag, fragment_data);
++	fragment_table[fragments].size = mangle(buffer, fragment_data, fragment_size, block_size, noF, 1);
++	fragment_table[fragments].start_block = bytes;
++	compressed_size = SQUASHFS_COMPRESSED_SIZE_BLOCK(fragment_table[fragments].size);
++	write_bytes(fd, bytes, compressed_size, buffer);
++	bytes += compressed_size;
++	total_uncompressed += fragment_size;
++	total_compressed += compressed_size;
++	TRACE("Writing fragment %d, uncompressed size %d, compressed size %d\n",fragments, fragment_size, compressed_size);
+ 	fragments ++;
+ 	fragment_size = 0;
+-	pthread_mutex_unlock(&fragment_mutex);
+-}
+-
+-void frag_release(int block)
+-{
+-	pthread_mutex_lock(&fragment_mutex);
+-	fragment_table[block].pending = FALSE;
+-	pthread_cond_signal(&fragment_waiting);
+-	pthread_mutex_unlock(&fragment_mutex);
+ }
+ 
+ 
+ static struct fragment empty_fragment = {SQUASHFS_INVALID_FRAG, 0, 0};
+-struct fragment *get_and_fill_fragment(struct file_buffer *file_buffer)
++struct fragment *get_and_fill_fragment(char *buff, int size)
+ {
+ 	struct fragment *ffrg;
+-	
+ 
+-	if(file_buffer == NULL || file_buffer->size == 0)
++	if(size == 0)
+ 		return &empty_fragment;
+ 
+-	if(fragment_size + file_buffer->size > block_size)
++	if(fragment_size + size > block_size)
+ 		write_fragment();
+ 
+ 	if((ffrg = (struct fragment *) malloc(sizeof(struct fragment))) == NULL)
+ 		BAD_ERROR("Out of memory in fragment block allocation!\n");
+ 
+-	if(fragment_size == 0)
+-		fragment_data = alloc_get(fragment_buffer);
+-
+ 	ffrg->index = fragments;
+ 	ffrg->offset = fragment_size;
+-	ffrg->size = file_buffer->size;
+-	memcpy(fragment_data->data + fragment_size, file_buffer->data, file_buffer->size);
+-	fragment_size += file_buffer->size;
++	ffrg->size = size;
++	memcpy(fragment_data + fragment_size, buff, size);
++	fragment_size += size;
+ 
+ 	return ffrg;
+ }
+@@ -1333,10 +1015,19 @@
+ }
+ 
+ 
++char *read_from_buffer(struct duplicate_buffer_handle *handle, unsigned int avail_bytes)
++{
++	char *v = handle->ptr;
++	handle->ptr += avail_bytes;	
++	return v;
++}
++
++
+ char read_from_file_buffer[SQUASHFS_FILE_MAX_SIZE];
+-char *read_from_disk(long long start, unsigned int avail_bytes)
++char *read_from_file(struct duplicate_buffer_handle *handle, unsigned int avail_bytes)
+ {
+-	read_bytes(fd, start, avail_bytes, read_from_file_buffer);
++	read_bytes(fd, handle->start, avail_bytes, read_from_file_buffer);
++	handle->start += avail_bytes;
+ 	return read_from_file_buffer;
+ }
+ 
+@@ -1344,205 +1035,99 @@
+ /*
+  * Compute 16 bit BSD checksum over the data
+  */
+-unsigned short get_checksum(char *buff, int bytes, unsigned short chksum)
+-{
+-	unsigned char *b = (unsigned char *) buff;
+-
+-	while(bytes --) {
+-		chksum = (chksum & 1) ? (chksum >> 1) | 0x8000 : chksum >> 1;
+-		chksum += *b++;
+-	}
+-
+-	return chksum;
+-}
+-
+-
+-unsigned short get_checksum_disk(long long start, long long l)
++unsigned short get_checksum(char *(get_next_file_block)(struct duplicate_buffer_handle *, unsigned int), struct duplicate_buffer_handle *handle, long long l)
+ {
+ 	unsigned short chksum = 0;
+-	unsigned int bytes;
++	unsigned int bytes = 0;
++	unsigned char *b;
++	struct duplicate_buffer_handle position = *handle;
+ 
+ 	while(l) {
+ 		bytes = l > SQUASHFS_FILE_MAX_SIZE ? SQUASHFS_FILE_MAX_SIZE : l;
+ 		l -= bytes;
+-		chksum = get_checksum(read_from_disk(start, bytes), bytes, chksum);
+-		start += bytes;
+-	}
+-
+-	return chksum;
+-}
+-
+-
+-unsigned short get_checksum_buffer(struct buffer_list *buffer_list, unsigned int blocks)
+-{
+-	unsigned short chksum = 0;
+-	int block;
+-
+-	for(block = 0; block < blocks; block ++) {
+-		struct buffer_list *b = &buffer_list[block];
+-
+-		if(b->read_buffer)
+-			chksum = get_checksum(b->read_buffer->data, b->read_buffer->size, chksum);
+-		else
+-			chksum = get_checksum(read_from_disk(b->start, b->size), b->size, chksum);
++		b = (unsigned char *) get_next_file_block(&position, bytes);
++		while(bytes--) {
++			chksum = (chksum & 1) ? (chksum >> 1) | 0x8000 : chksum >> 1;
++			chksum += *b++;
++		}
+ 	}
+ 
+ 	return chksum;
+ }
+ 
+ 
+-unsigned short get_checksum_mem(char *buff, int bytes)
+-{
+-	return get_checksum(buff, bytes, 0);
+-}
+-
+-
+-unsigned short get_checksum_mem_buffer(struct file_buffer *file_buffer)
+-{
+-	if(file_buffer == NULL)
+-		return 0;
+-	else
+-		return get_checksum(file_buffer->data, file_buffer->size, 0);
+-}
+-
+-
+ int cached_frag = -1;
+-char fragdata[SQUASHFS_FILE_MAX_SIZE];
+-#define DUP_HASH(a) (a & 0xffff)
+-void add_file(long long start, long long file_size, long long file_bytes, unsigned int *block_listp, int blocks, unsigned int fragment, int offset, int bytes)
++void add_file(long long start, long long file_bytes, unsigned int *block_listp, int blocks, unsigned int fragment, int offset, int bytes)
+ {
+ 	struct fragment *frg;
++	struct file_info *dupl_ptr;
+ 	char *datap;
++	struct duplicate_buffer_handle handle;
+ 	unsigned int *block_list = block_listp;
+-	struct file_info *dupl_ptr = dupl[DUP_HASH(file_size)];
+ 
+-	if(!duplicate_checking || file_size == 0)
++	if(!duplicate_checking)
+ 		return;
+ 
+-	for(; dupl_ptr; dupl_ptr = dupl_ptr->next) {
+-		if(file_size != dupl_ptr->file_size)
+-			continue;
+-		if(blocks != 0 && start != dupl_ptr->start)
+-			continue;
+-		if(fragment != dupl_ptr->fragment->index)
+-			continue;
+-		if(fragment != SQUASHFS_INVALID_FRAG && (offset != dupl_ptr->fragment->offset || bytes != dupl_ptr->fragment->size))
+-			continue;
+-		return;
+-	}
+-
+ 	if((frg = (struct fragment *) malloc(sizeof(struct fragment))) == NULL)
+ 		BAD_ERROR("Out of memory in fragment block allocation!\n");
+ 
+ 	frg->index = fragment;
+ 	frg->offset = offset;
+ 	frg->size = bytes;
+-
+-	add_non_dup(file_size, file_bytes, block_list, start, frg, 0, 0, FALSE);
++	if(fragment == cached_frag || fragment == SQUASHFS_INVALID_FRAG)
++		datap = fragment_data + offset;
++	else
++		datap = get_fragment(fragment_data, frg);
++	handle.start = start;
++	if((dupl_ptr = duplicate(read_from_file, &handle, file_bytes, &block_listp, &start, blocks, &frg, datap, bytes)) != NULL)
++		dupl_ptr->fragment = frg;
++	else
++		free(block_list);
++	cached_frag = fragment;
+ }
+ 
++
+ char cached_fragment[SQUASHFS_FILE_SIZE];
+ int cached_frag1 = -1;
+ 
+-int pre_duplicate(long long file_size)
+-{
+-	struct file_info *dupl_ptr = dupl[DUP_HASH(file_size)];
+-
+-	for(; dupl_ptr; dupl_ptr = dupl_ptr->next)
+-		if(dupl_ptr->file_size == file_size)
+-			return TRUE;
+-
+-	return FALSE;
+-}
+-
+-
+-int pre_duplicate_frag(long long file_size, unsigned short checksum)
++struct file_info *duplicate(char *(get_next_file_block)(struct duplicate_buffer_handle *, unsigned int), struct duplicate_buffer_handle *file_start, long long bytes, unsigned int **block_list, long long *start, int blocks, struct fragment **fragment, char *frag_data, int frag_bytes)
+ {
+-	struct file_info *dupl_ptr = dupl[DUP_HASH(file_size)];
+-
+-	for(; dupl_ptr; dupl_ptr = dupl_ptr->next)
+-		if(dupl_ptr->file_size == file_size) {
+-			if(dupl_ptr->checksum_flag == FALSE) {
+-				dupl_ptr->checksum = get_checksum_disk(dupl_ptr->start, dupl_ptr->bytes);
+-				dupl_ptr->fragment_checksum = get_checksum_mem(get_fragment(cached_fragment, dupl_ptr->fragment, &cached_frag1), file_size);
+-				dupl_ptr->checksum_flag = TRUE;
+-			}
+-			if(dupl_ptr->fragment_checksum == checksum)
+-				return TRUE;
+-		}
+-
+-	return FALSE;
+-}
+-
+-
+-struct file_info *add_non_dup(long long file_size, long long bytes, unsigned int *block_list, long long start, struct fragment *fragment, unsigned short checksum, unsigned short fragment_checksum, int checksum_flag)
+-{
+-	struct file_info *dupl_ptr;
++	unsigned short checksum = get_checksum(get_next_file_block, file_start, bytes);
++	struct duplicate_buffer_handle handle = { frag_data, 0 };
++	unsigned short fragment_checksum = get_checksum(read_from_buffer, &handle, frag_bytes);
++	struct file_info *dupl_ptr = bytes ? dupl[checksum] : frag_dups[fragment_checksum];
+ 
+-	if((dupl_ptr = (struct file_info *) malloc(sizeof(struct file_info))) == NULL) {
+-		BAD_ERROR("Out of memory in dup_files allocation!\n");
+-	}
+-
+-	dupl_ptr->file_size = file_size;
+-	dupl_ptr->bytes = bytes;
+-	dupl_ptr->block_list = block_list;
+-	dupl_ptr->start = start;
+-	dupl_ptr->fragment = fragment;
+-	dupl_ptr->checksum = checksum;
+-	dupl_ptr->fragment_checksum = fragment_checksum;
+-	dupl_ptr->checksum_flag = checksum_flag;
+-	dupl_ptr->next = dupl[DUP_HASH(file_size)];
+-	dupl[DUP_HASH(file_size)] = dupl_ptr;
+-	dup_files ++;
+-
+-	return dupl_ptr;
+-}
+-
+-
+-struct file_info *duplicate(long long file_size, long long bytes, unsigned int **block_list, long long *start, struct fragment **fragment, struct file_buffer *file_buffer, struct buffer_list *buffer_list, int blocks, unsigned short checksum, unsigned short fragment_checksum, int checksum_flag)
+-{
+-	struct file_info *dupl_ptr = dupl[DUP_HASH(file_size)];
+-	int frag_bytes = file_buffer ? file_buffer->size : 0;
+ 
+ 	for(; dupl_ptr; dupl_ptr = dupl_ptr->next)
+-		if(file_size == dupl_ptr->file_size && bytes == dupl_ptr->bytes && frag_bytes == dupl_ptr->fragment->size) {
+-			char buffer2[SQUASHFS_FILE_MAX_SIZE];
++		if(bytes == dupl_ptr->bytes && frag_bytes == dupl_ptr->fragment->size && fragment_checksum == dupl_ptr->fragment_checksum) {
++			char buffer1[SQUASHFS_FILE_MAX_SIZE];
++			long long dup_bytes = dupl_ptr->bytes;
+ 			long long dup_start = dupl_ptr->start;
++			struct duplicate_buffer_handle position = *file_start;
+ 			char *buffer;
+-			int block;
++			while(dup_bytes) {
++				int avail_bytes = dup_bytes > SQUASHFS_FILE_MAX_SIZE ? SQUASHFS_FILE_MAX_SIZE : dup_bytes;
+ 
+-			if(checksum_flag == FALSE) {
+-				checksum = get_checksum_buffer(buffer_list, blocks);
+-				fragment_checksum = get_checksum_mem_buffer(file_buffer);
+-				checksum_flag = TRUE;
+-			}
+-
+-			if(dupl_ptr->checksum_flag == FALSE) {
+-				dupl_ptr->checksum = get_checksum_disk(dupl_ptr->start, dupl_ptr->bytes);
+-				dupl_ptr->fragment_checksum = get_checksum_mem(get_fragment(cached_fragment, dupl_ptr->fragment, &cached_frag1), frag_bytes);
+-				dupl_ptr->checksum_flag = TRUE;
+-			}
+-
+-			if(checksum != dupl_ptr->checksum || fragment_checksum != dupl_ptr->fragment_checksum)
+-				continue;
+-
+-			for(block = 0; block < blocks; block ++) {
+-				struct buffer_list *b = &buffer_list[block];
+-
+-				if(b->read_buffer)
+-					buffer = b->read_buffer->data;
+-				else
+-					buffer = read_from_disk(b->start, b->size);
+-
+-				read_bytes(fd, dup_start, b->size, buffer2);
+-				if(memcmp(buffer, buffer2, b->size) != 0)
++				buffer = get_next_file_block(&position, avail_bytes);
++				read_bytes(fd, dup_start, avail_bytes, buffer1);
++				if(memcmp(buffer, buffer1, avail_bytes) != 0)
+ 					break;
+-				dup_start += b->size;
++				dup_bytes -= avail_bytes;
++				dup_start += avail_bytes;
+ 			}
+-			if(block == blocks) {
+-				char *fragment_buffer1 = get_fragment(cached_fragment, dupl_ptr->fragment, &cached_frag1);
++			if(dup_bytes == 0) {
++				char *fragment_buffer1;
++				
++				if(dupl_ptr->fragment->index == fragments || dupl_ptr->fragment->index == SQUASHFS_INVALID_FRAG)
++					fragment_buffer1 = fragment_data + dupl_ptr->fragment->offset;
++				else if(dupl_ptr->fragment->index == cached_frag1)
++					fragment_buffer1 = cached_fragment + dupl_ptr->fragment->offset;
++				else {
++					fragment_buffer1 = get_fragment(cached_fragment, dupl_ptr->fragment);
++					cached_frag1 = dupl_ptr->fragment->index;
++				}
+ 
+-				if(frag_bytes == 0 || memcmp(file_buffer->data, fragment_buffer1, frag_bytes) == 0) {
++				if(frag_bytes == 0 || memcmp(frag_data, fragment_buffer1, frag_bytes) == 0) {
+ 					TRACE("Found duplicate file, start 0x%llx, size %lld, checksum 0x%x, fragment %d, size %d, offset %d, checksum 0x%x\n", dupl_ptr->start,
+ 						dupl_ptr->bytes, dupl_ptr->checksum, dupl_ptr->fragment->index, frag_bytes, dupl_ptr->fragment->offset, fragment_checksum);
+ 					*block_list = dupl_ptr->block_list;
+@@ -1554,520 +1139,135 @@
+ 		}
+ 
+ 
+-	return add_non_dup(file_size, bytes, *block_list, *start, *fragment, checksum, fragment_checksum, checksum_flag);
+-}
+-
+-
+-void reader_read_file(struct dir_ent *dir_ent)
+-{
+-	struct stat *buf = &dir_ent->inode->buf;
+-	int count;
+-	int blocks = (buf->st_size + block_size - 1) >> block_log;
+-	int frag_block = !no_fragments && (always_use_fragments ||
+-		(buf->st_size < block_size)) ? buf->st_size >> block_log : -1;
+-	int file;
+-	static int block_order = 0;
+-	struct file_buffer *file_buffer;
+-
+-	if(buf->st_size == 0 || dir_ent->inode->read)
+-		return;
+-
+-	if((file = open(dir_ent->pathname, O_RDONLY)) == -1)
+-		goto read_err;
+-
+-	for(count = 0; count < blocks; count ++) {
+-		file_buffer = alloc_get(reader_buffer);
+-
+-		if((file_buffer->size = read(file, file_buffer->data, block_size)) == -1) {
+-			close(file);
+-			goto read_err2;
+-		}
+-		file_buffer->block = count;
+-		file_buffer->block_order = block_order ++;
+-		file_buffer->error = FALSE;
+-		if(file_buffer->fragment = count == frag_block)
+-			queue_put(from_deflate, file_buffer);
+-		else
+-			queue_put(from_reader, file_buffer);
+-	}
+-
+-	close(file);
+-	dir_ent->inode->read = TRUE;
+-
+-	return;
+-
+-read_err:
+-	file_buffer = alloc_get(reader_buffer);
+-read_err2:
+-	file_buffer->block_order = block_order ++;
+-	file_buffer->error = TRUE;
+-	queue_put(from_deflate, file_buffer);
+-}
+-
+-
+-void reader_scan(struct dir_info *dir) {
+-	int i;
+-
+-	for(i = 0; i < dir->count; i++) {
+-		struct dir_ent *dir_ent = dir->list[i];
+-		struct stat *buf = &dir_ent->inode->buf;
+-		if(dir_ent->data)
+-			continue;
+-
+-		switch(buf->st_mode & S_IFMT) {
+-			case S_IFREG:
+-				reader_read_file(dir_ent);
+-				break;
+-			case S_IFDIR:
+-				reader_scan(dir_ent->dir);
+-				break;
+-		}
+-	}
+-}
+-
+-
+-void *reader(void *arg)
+-{
+-	int oldstate;
+-
+-	pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, &oldstate);
+-	pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, &oldstate);
+-
+-	if(!sorted)
+-		reader_scan(queue_get(to_reader));
+-	else {
+-		int i;
+-		struct priority_entry *entry;
+-
+-		queue_get(to_reader);
+-		for(i = 65535; i >= 0; i--)
+-			for(entry = priority_list[i]; entry; entry = entry->next)
+-				reader_read_file(entry->dir);
+-	}
+-}
+-
+-
+-void *writer(void *arg)
+-{
+-	int write_error = FALSE;
+-	int oldstate;
+-
+-	pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, &oldstate);
+-	pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, &oldstate);
+-
+-	while(1) {
+-		struct file_buffer *file_buffer = queue_get(to_writer);
+-		off_t off;
+-
+-		if(file_buffer == NULL) {
+-			queue_put(from_writer, write_error ? (void *) &write_error : NULL);
+-			continue;
+-		}
+-
+-		off = file_buffer->block;
+-
+-		pthread_mutex_lock(&pos_mutex);
+-
+-		if(!write_error && lseek(fd, off, SEEK_SET) == -1) {
+-			perror("Lseek on destination failed");
+-			write_error = TRUE;
+-		}
+-
+-		if(!write_error && write(fd, file_buffer->data, file_buffer->size) == -1) {
+-			perror("Write on destination failed");
+-			write_error = TRUE;
+-		}
+-		pthread_mutex_unlock(&pos_mutex);
+-
+-		alloc_free(file_buffer);
+-	}
+-}
+-
+-
+-void *deflator(void *arg)
+-{
+-	z_stream *stream = NULL;
+-	int oldstate;
+-
+-	pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, &oldstate);
+-	pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, &oldstate);
+-
+-	while(1) {
+-		struct file_buffer *file_buffer = queue_get(from_reader);
+-		struct file_buffer *write_buffer = alloc_get(writer_buffer);
+-
+-		write_buffer->c_byte = mangle2(&stream, write_buffer->data, file_buffer->data, file_buffer->size, block_size, noD, 1);
+-		write_buffer->block = file_buffer->block;
+-		write_buffer->block_order = file_buffer->block_order;
+-		write_buffer->size = SQUASHFS_COMPRESSED_SIZE_BLOCK(write_buffer->c_byte);
+-		write_buffer->fragment = FALSE;
+-		write_buffer->error = FALSE;
+-		alloc_free(file_buffer);
+-		queue_put(from_deflate, write_buffer);
+-	}
+-}
+-
+-
+-void *frag_deflator(void *arg)
+-{
+-	z_stream *stream = NULL;
+-	int oldstate;
+-
+-	pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, &oldstate);
+-	pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, &oldstate);
+-
+-	while(1) {
+-		int c_byte, compressed_size;
+-		struct file_buffer *file_buffer = queue_get(to_frag);
+-		struct file_buffer *write_buffer = alloc_get_2(writer_buffer, frag_release, file_buffer->block);
+-
+-		c_byte = mangle2(&stream, write_buffer->data, file_buffer->data, file_buffer->size, block_size, noF, 1);
+-		compressed_size = SQUASHFS_COMPRESSED_SIZE_BLOCK(c_byte);
+-		pthread_mutex_lock(&fragment_mutex);
+-		fragment_table[file_buffer->block].size = c_byte;
+-		fragment_table[file_buffer->block].start_block = bytes;
+-		write_buffer->size = compressed_size;
+-		write_buffer->block = bytes;
+-		queue_put(to_writer, write_buffer);
+-		bytes += compressed_size;
+-		total_uncompressed += file_buffer->size;
+-		total_compressed += compressed_size;
+-		TRACE("Writing fragment %d, uncompressed size %d, compressed size %d\n", file_buffer->block, file_buffer->size, compressed_size);
+-		fragments_outstanding --;
+-		pthread_cond_signal(&fragment_waiting);
+-		pthread_mutex_unlock(&fragment_mutex);
+-		alloc_free(file_buffer);
++	if((dupl_ptr = (struct file_info *) malloc(sizeof(struct file_info))) == NULL) {
++		BAD_ERROR("Out of memory in dup_files allocation!\n");
+ 	}
+-}
+-
+-
+-#define HASH_ENTRIES		256
+-#define BLOCK_HASH(a)		(a % HASH_ENTRIES)
+-struct file_buffer		*block_hash[HASH_ENTRIES];
+ 
+-void push_buffer(struct file_buffer *file_buffer)
+-{
+-	int hash = BLOCK_HASH(file_buffer->block_order);
+-
+-	file_buffer->next = block_hash[hash];
+-	block_hash[hash] = file_buffer;
+-}
+-
+-
+-struct file_buffer *get_file_buffer(struct queue *queue)
+-{
+-	static unsigned int block_order = 0;
+-	int hash = BLOCK_HASH(block_order);
+-	struct file_buffer *file_buffer = block_hash[hash], *prev = NULL;
+-
+-	for(;file_buffer; prev = file_buffer, file_buffer = file_buffer->next)
+-		if(file_buffer->block_order == block_order)
+-			break;
++	dupl_ptr->bytes = bytes;
++	dupl_ptr->checksum = checksum;
++	dupl_ptr->start = *start;
++	dupl_ptr->fragment_checksum = fragment_checksum;
++	dupl_ptr->block_list = *block_list;
+ 
+-	if(file_buffer) {
+-		if(prev)
+-			prev->next = file_buffer->next;
+-		else
+-			block_hash[hash] = file_buffer->next;
++	dup_files ++;
++	if(bytes) {
++		dupl_ptr->next = dupl[checksum];
++		dupl[checksum] = dupl_ptr;
+ 	} else {
+-		while(1) {
+-			file_buffer = queue_get(queue);
+-			if(file_buffer->block_order == block_order)
+-				break;
+-			push_buffer(file_buffer);
+-		}
++		dupl_ptr->next = frag_dups[fragment_checksum];
++		frag_dups[fragment_checksum] = dupl_ptr;
+ 	}
+ 
+-	block_order ++;
+-
+-	return file_buffer;
+-}
+-
+-
+-int write_file_empty(squashfs_inode *inode, struct dir_ent *dir_ent, int *duplicate_file)
+-{
+-	file_count ++;
+-	*duplicate_file = FALSE;
+-	return dir_ent->inode->nlink == 1 ?
+-		create_inode(inode, dir_ent, SQUASHFS_FILE_TYPE, 0, 0, 0, NULL, &empty_fragment, NULL) :
+-		create_inode(inode, dir_ent, SQUASHFS_LREG_TYPE, 0, 0, 0, NULL, &empty_fragment, NULL);
+-}
+-
+-
+-int write_file_frag_dup(squashfs_inode *inode, struct dir_ent *dir_ent, int size, int *duplicate_file, struct file_buffer *file_buffer, unsigned short checksum)
+-{
+-	int file;
+-	struct file_info *dupl_ptr;
+-	struct fragment *fragment;
+-	unsigned int *block_listp = NULL;
+-	long long start = 0;
+-
+-	dupl_ptr = duplicate(size, 0, &block_listp, &start, &fragment, file_buffer, NULL, 0, 0, checksum, TRUE);
+-
+-	if(dupl_ptr) {
+-		*duplicate_file = FALSE;
+-		fragment = get_and_fill_fragment(file_buffer);
+-		dupl_ptr->fragment = fragment;
+-	} else
+-		*duplicate_file = TRUE;
+-
+-	alloc_free(file_buffer);
+-
+-	total_bytes += size;
+-	file_count ++;
+-
+-	return dir_ent->inode->nlink == 1 ?
+-		create_inode(inode, dir_ent, SQUASHFS_FILE_TYPE, size, 0, 0, NULL, fragment, NULL) :
+-		create_inode(inode, dir_ent, SQUASHFS_LREG_TYPE, size, 0, 0, NULL, fragment, NULL);
+-}
+-
+-
+-int write_file_frag(squashfs_inode *inode, struct dir_ent *dir_ent, int size, int *duplicate_file)
+-{
+-	struct fragment *fragment;
+-	unsigned short checksum;
+-	struct file_buffer *file_buffer = get_file_buffer(from_deflate);
+-
+-	if(file_buffer->size != size)
+-		printf("bug\n");
+-
+-	if(file_buffer->error) {
+-		alloc_free(file_buffer);
+-		return FALSE;
+-	}
+-
+-	checksum = get_checksum_mem_buffer(file_buffer);
+-
+-	if(pre_duplicate_frag(size, checksum))
+-		return write_file_frag_dup(inode, dir_ent, size, duplicate_file, file_buffer, checksum);
+-		
+-	fragment = get_and_fill_fragment(file_buffer);
+-
+-	alloc_free(file_buffer);
+-
+-	if(duplicate_checking)
+-		add_non_dup(size, 0, NULL, 0, fragment, 0, checksum, TRUE);
+-
+-	total_bytes += size;
+-	file_count ++;
+-
+-	*duplicate_file = FALSE;
+-
+-	return dir_ent->inode->nlink == 1 ?
+-		create_inode(inode, dir_ent, SQUASHFS_FILE_TYPE, size, 0, 0, NULL, fragment, NULL) :
+-		create_inode(inode, dir_ent, SQUASHFS_LREG_TYPE, size, 0, 0, NULL, fragment, NULL);
++	return dupl_ptr;
+ }
+ 
+ 
+-int write_file_blocks(squashfs_inode *inode, struct dir_ent *dir_ent, long long read_size)
++#define MINALLOCBYTES (1024 * 1024)
++int write_file(squashfs_inode *inode, struct dir_ent *dir_ent, long long size, int *duplicate_file)
+ {
+-	int block, status;
++	int block = 0, i, file, whole_file = 1, status;
+ 	unsigned int c_byte, frag_bytes;
+-	long long bbytes, file_bytes, start;
++	long long bbytes, file_bytes = 0, start;
++	char buff[block_size], *c_buffer = NULL, *filename = dir_ent->pathname;
+ 	struct fragment *fragment;
+-	struct file_info *dupl_ptr;
+-	int blocks = (read_size + block_size - 1) >> block_log;
+-	unsigned int *block_list;
+-	struct file_buffer *read_buffer;
+-
+-	if(!no_fragments && always_use_fragments) {
+-		blocks = read_size >> block_log;
+-		frag_bytes = read_size % block_size;
+-	} else
+-		frag_bytes = 0;
++	struct file_info *dupl_ptr = NULL;
++	struct duplicate_buffer_handle handle;
++	long long read_size = (size > SQUASHFS_MAX_FILE_SIZE) ? SQUASHFS_MAX_FILE_SIZE : size;
++	int blocks = (read_size + block_size - 1) >> block_log, allocated_blocks = blocks;
++	unsigned int *block_list, *block_listp;
+ 
+ 	if((block_list = malloc(blocks * sizeof(unsigned int))) == NULL)
+ 		BAD_ERROR("Out of memory allocating block_list\n");
++	block_listp = block_list;
+ 
+-	ensure_fragments_flushed();
+-
+-	file_bytes = 0;
+-	start = bytes;
+-	for(block = 0; block < blocks; block ++) {
+-		read_buffer = get_file_buffer(from_deflate);
+-		if(read_buffer->error)
+-			goto read_err;
+-
+-		block_list[block] = read_buffer->c_byte;
+-		read_buffer->block = bytes;
+-		bytes += read_buffer->size;
+-		file_bytes += read_buffer->size;
+-		queue_put(to_writer, read_buffer);
+-	}
+-
+-	if(frag_bytes != 0) {
+-		read_buffer = get_file_buffer(from_deflate);
+-		if(read_buffer->size != frag_bytes)
+-			printf("bug\n");
+-		if(read_buffer->error)
+-			goto read_err;
+-	} else
+-		read_buffer = NULL;
+-
+-	fragment = get_and_fill_fragment(read_buffer);
+-	alloc_free(read_buffer);
+-
+-	if(duplicate_checking)
+-		add_non_dup(read_size, file_bytes, block_list, start, fragment, 0, 0, FALSE);
+-	file_count ++;
+-	total_bytes += read_size;
+-	if(dir_ent->inode->nlink == 1 && read_size < ((long long) (1<<30) - 1))
+-		status = create_inode(inode, dir_ent, SQUASHFS_FILE_TYPE, read_size, start, blocks, block_list, fragment, NULL);
+-	else
+-		status = create_inode(inode, dir_ent, SQUASHFS_LREG_TYPE, read_size, start, blocks, block_list, fragment, NULL);
+-	if(duplicate_checking == FALSE)
+-		free(block_list);
+-	return status;
+-
+-read_err:
+-	perror("Error in reading file, skipping...");
+-	if(block) {
+-		queue_put(to_writer, NULL);
+-		if(queue_get(from_writer) != 0)
+-			EXIT_MKSQUASHFS();
+-		bytes = start;
+-		if(!block_device)
+-			ftruncate(fd, bytes);
+-	}
+-	free(block_list);
+-	alloc_free(read_buffer);
+-	return FALSE;
+-}
+-
+-
+-int write_file_blocks_dup(squashfs_inode *inode, struct dir_ent *dir_ent, long long read_size, int *duplicate_file)
+-{
+-	int block, status, thresh;
+-	unsigned int c_byte, frag_bytes;
+-	long long bbytes, file_bytes, start;
+-	struct fragment *fragment;
+-	struct file_info *dupl_ptr;
+-	int blocks = (read_size + block_size - 1) >> block_log;
+-	unsigned int *block_list, *block_listp;
+-	struct file_buffer *read_buffer;
+-	struct file_data *file_data;
+-	struct buffer_list *buffer_list;
+-
+-	if(!no_fragments && always_use_fragments) {
+-		blocks = read_size >> block_log;
++	if(!no_fragments && (read_size < block_size || always_use_fragments)) {
++		allocated_blocks = blocks = read_size >> block_log;
+ 		frag_bytes = read_size % block_size;
+ 	} else
+ 		frag_bytes = 0;
+ 
+-	if((block_list = malloc(blocks * sizeof(unsigned int))) == NULL)
+-		BAD_ERROR("Out of memory allocating block_list\n");
+-	block_listp = block_list;
+-
+-	if((buffer_list = malloc(blocks * sizeof(struct buffer_list))) == NULL)
+-		BAD_ERROR("Out of memory allocating file block list\n");
+-
+-	ensure_fragments_flushed();
++	if(size > read_size)
++		ERROR("file %s truncated to %lld bytes\n", filename, SQUASHFS_MAX_FILE_SIZE);
+ 
+-	file_bytes = 0;
+-	start = bytes;
+-	thresh = blocks > (writer_buffer_size - processors) ? blocks - (writer_buffer_size - processors): 0;
+-	for(block = 0; block < blocks; block ++) {
+-		read_buffer = get_file_buffer(from_deflate);
+-		if(read_buffer->error)
+-			goto read_err;
++	total_bytes += read_size;
++	if((file = open(filename, O_RDONLY)) == -1)
++		goto read_err;
+ 
+-		block_list[block] = read_buffer->c_byte;
+-		read_buffer->block = bytes;
+-		bytes += read_buffer->size;
+-		file_bytes += read_buffer->size;
+-
+-		if(block < thresh) {
+-			buffer_list[block].read_buffer = NULL;
+-			queue_put(to_writer, read_buffer);
+-		} else
+-			buffer_list[block].read_buffer = read_buffer;
+-		buffer_list[block].start = read_buffer->block;
+-		buffer_list[block].size = read_buffer->size;
++	do {
++		long long bytes = (((long long) allocated_blocks) + 1) << block_log;
++		if(bytes != ((size_t) bytes) || (c_buffer = (char *) malloc(bytes)) == NULL) {
++			TRACE("Out of memory allocating write_file buffer, allocated_blocks %ld, blocks %d\n", allocated_blocks, blocks);
++			whole_file = 0;
++			if(bytes < MINALLOCBYTES)
++				BAD_ERROR("Out of memory allocating write_file buffer, could not allocate %ld blocks (%d Kbytes)\n", allocated_blocks, allocated_blocks << (block_log - 10));
++			allocated_blocks >>= 1;
++		}
++	} while(!c_buffer);
++
++	for(start = bytes; block < blocks; file_bytes += bbytes) {
++		for(i = 0, bbytes = 0; (i < allocated_blocks) && (block < blocks); i++) {
++			int available_bytes = read_size - (block * block_size) > block_size ? block_size : read_size - (block * block_size);
++			if(read(file, buff, available_bytes) == -1)
++				goto read_err;
++			c_byte = mangle(c_buffer + bbytes, buff, available_bytes, block_size, noD, 1);
++			block_list[block ++] = c_byte;
++			bbytes += SQUASHFS_COMPRESSED_SIZE_BLOCK(c_byte);
++		}
++		if(!whole_file) {
++			write_bytes(fd, bytes, bbytes, c_buffer);
++			bytes += bbytes;
++		}
+ 	}
+ 
+-	if(frag_bytes != 0) {
+-		read_buffer = get_file_buffer(from_deflate);
+-		if(read_buffer->size != frag_bytes)
+-			printf("bug\n");
+-		if(read_buffer->error)
++	if(frag_bytes != 0)
++		if(read(file, buff, frag_bytes) == -1)
+ 			goto read_err;
+-	} else
+-		read_buffer = NULL;
+-
+-	queue_put(to_writer, NULL);
+-	if(queue_get(from_writer) != 0)
+-		EXIT_MKSQUASHFS();
+-
+-	dupl_ptr = duplicate(read_size, file_bytes, &block_listp, &start, &fragment, read_buffer, buffer_list, blocks, 0, 0, FALSE);
+ 
+-	if(dupl_ptr) {
+-		*duplicate_file = FALSE;
+-		for(block = thresh; block < blocks; block ++)
+-			queue_put(to_writer, buffer_list[block].read_buffer);
+-		fragment = get_and_fill_fragment(read_buffer);
+-		dupl_ptr->fragment = fragment;
++	close(file);
++	if(whole_file) {
++		handle.ptr = c_buffer;
++		if(duplicate_checking && (dupl_ptr = duplicate(read_from_buffer, &handle, file_bytes, &block_listp, &start, blocks, &fragment, buff, frag_bytes)) == NULL) {
++			*duplicate_file = TRUE;
++			goto wr_inode;
++		}
++		write_bytes(fd, bytes, file_bytes, c_buffer);
++		bytes += file_bytes;
+ 	} else {
+-		*duplicate_file = TRUE;
+-		for(block = thresh; block < blocks; block ++)
+-			alloc_free(buffer_list[block].read_buffer);
+-		bytes = buffer_list[0].start;
+-		if(thresh && !block_device)
+-			ftruncate(fd, bytes);
++		handle.start = start;
++		if(duplicate_checking && (dupl_ptr = duplicate(read_from_file, &handle, file_bytes, &block_listp, &start, blocks, &fragment, buff, frag_bytes)) == NULL) {
++			bytes = start;
++			if(!block_device)
++				ftruncate(fd, bytes);
++			*duplicate_file = TRUE;
++			goto wr_inode;
++		}
+ 	}
+ 
+-	alloc_free(read_buffer);
+-	free(buffer_list);
+-	file_count ++;
+-	total_bytes += read_size;
++	fragment = get_and_fill_fragment(buff, frag_bytes);
++	if(duplicate_checking)
++		dupl_ptr->fragment = fragment;
++
++	*duplicate_file = FALSE;
+ 
++wr_inode:
++	free(c_buffer);
++	file_count ++;
+ 	if(dir_ent->inode->nlink == 1 && read_size < ((long long) (1<<30) - 1))
+ 		status = create_inode(inode, dir_ent, SQUASHFS_FILE_TYPE, read_size, start, blocks, block_listp, fragment, NULL);
+ 	else
+ 		status = create_inode(inode, dir_ent, SQUASHFS_LREG_TYPE, read_size, start, blocks, block_listp, fragment, NULL);
+-	if(*duplicate_file == TRUE)
++	if(duplicate_checking == FALSE || *duplicate_file == TRUE)
+ 		free(block_list);
+ 	return status;
+ 
+ read_err:
+ 	perror("Error in reading file, skipping...");
+-	if(block && thresh) {
+-		queue_put(to_writer, NULL);
+-		if(queue_get(from_writer) != 0)
+-			EXIT_MKSQUASHFS();
+-		bytes = start;
+-		if(!block_device)
+-			ftruncate(fd, bytes);
+-	}
+-	for(blocks = thresh; blocks < block; blocks ++)
+-		alloc_free(buffer_list[blocks].read_buffer);
+-	free(buffer_list);
++	free(c_buffer);
+ 	free(block_list);
+-	alloc_free(read_buffer);
+ 	return FALSE;
+ }
+ 
+ 
+-int write_file(squashfs_inode *inode, struct dir_ent *dir_ent, long long size, int *duplicate_file)
+-{
+-	long long read_size = (size > SQUASHFS_MAX_FILE_SIZE) ? SQUASHFS_MAX_FILE_SIZE : size;
+-
+-	if(size > read_size)
+-		ERROR("file %s truncated to %lld bytes\n", dir_ent->pathname, SQUASHFS_MAX_FILE_SIZE);
+-
+-	if(read_size == 0)
+-		return write_file_empty(inode, dir_ent, duplicate_file);
+-
+-	if(!no_fragments && (read_size < block_size))
+-		return write_file_frag(inode, dir_ent, read_size, duplicate_file);
+-
+-	if(pre_duplicate(read_size))
+-		return write_file_blocks_dup(inode, dir_ent, read_size, duplicate_file);
+-
+-	*duplicate_file = FALSE;
+-	return write_file_blocks(inode, dir_ent, read_size);
+-}
+-
+-
+ char b_buffer[8192];
+ char *name;
+ char *basename_r();
+@@ -2134,7 +1334,6 @@
+ 		BAD_ERROR("Out of memory in inode hash table entry allocation\n");
+ 
+ 	memcpy(&inode->buf, buf, sizeof(struct stat));
+-	inode->read = FALSE;
+ 	inode->inode = SQUASHFS_INVALID_BLK;
+ 	inode->nlink = 1;
+ 	if((buf->st_mode & S_IFMT) == S_IFDIR)
+@@ -2357,9 +1556,6 @@
+ 		return;
+ 	}
+ 	if(sorted)
+-		generate_file_priorities(dir_info, 0, &dir_info->dir_ent->inode->buf);
+-	queue_put(to_reader, dir_info);
+-	if(sorted)
+ 		sort_files_and_write(dir_info);
+ 	dir_scan2(inode, dir_info);
+ }
+@@ -2495,7 +1691,7 @@
+ 				case SQUASHFS_CHRDEV_TYPE:
+ 					INFO("character device %s inode 0x%llx LINK\n", dir_name, *inode);
+ 					break;
+-				case SQUASHFS_BLKDEV_TYPE:
++				caseSQUASHFS_BLKDEV_TYPE:
+ 					INFO("block device %s inode 0x%llx LINK\n", dir_name, *inode);
+ 					break;
+ 				case SQUASHFS_FIFO_TYPE:
+@@ -2595,76 +1791,8 @@
+ }
+ 
+ 
+-void initialise_threads()
+-{
+-	int i;
+-	sigset_t sigmask, old_mask;
+-
+-	sigemptyset(&sigmask);
+-	sigaddset(&sigmask, SIGINT);
+-	sigaddset(&sigmask, SIGQUIT);
+-	if(sigprocmask(SIG_BLOCK, &sigmask, &old_mask) == -1)
+-		BAD_ERROR("Failed to set signal mask in intialise_threads\n");
+-
+-	signal(SIGUSR1, sigusr1_handler);
+-
+-	if(processors == -1) {
+-#ifndef linux
+-		int mib[2];
+-		size_t len = sizeof(processors);
+-
+-		mib[0] = CTL_HW;
+-#ifdef HW_AVAILCPU
+-		mib[1] = HW_AVAILCPU;
+-#else
+-		mib[1] = HW_NCPU;
+-#endif
+-
+-		if(sysctl(mib, 2, &processors, &len, NULL, 0) == -1) {
+-			ERROR("Failed to get number of available processors.  Defaulting to 1\n");
+-			processors = 1;
+-		}
+-#else
+-		processors = get_nprocs();
+-#endif
+-	}
+-
+-	if((thread = malloc((2 + processors * 2) * sizeof(pthread_t))) == NULL)
+-		BAD_ERROR("Out of memory allocating thread descriptors\n");
+-	deflator_thread = &thread[2];
+-	frag_deflator_thread = &deflator_thread[processors];
+-
+-	to_reader = queue_init(1);
+-	from_reader = queue_init(reader_buffer_size);
+-	to_writer = queue_init(writer_buffer_size);
+-	from_writer = queue_init(1);
+-	from_deflate = queue_init(reader_buffer_size);
+-	to_frag = queue_init(processors * 2);
+-	reader_buffer = alloc_init(SQUASHFS_FILE_MAX_SIZE, reader_buffer_size);
+-	writer_buffer = alloc_init(SQUASHFS_FILE_MAX_SIZE, writer_buffer_size);
+-	fragment_buffer = alloc_init(SQUASHFS_FILE_MAX_SIZE, processors * 2);
+-	pthread_create(&thread[0], NULL, reader, NULL);
+-	pthread_create(&thread[1], NULL, writer, NULL);
+-	pthread_mutex_init(&fragment_mutex, NULL);
+-	pthread_cond_init(&fragment_waiting, NULL);
+-
+-	for(i = 0; i < processors; i++) {
+-		if(pthread_create(&deflator_thread[i], NULL, deflator, NULL) != 0 )
+-			BAD_ERROR("Failed to create thread\n");
+-		if(pthread_create(&frag_deflator_thread[i], NULL, frag_deflator, NULL) != 0)
+-			BAD_ERROR("Failed to create thread\n");
+-	}
+-
+-	printf("Parallel mksquashfs: Using %d processor%s\n", processors,
+-			processors == 1 ? "" : "s");
+-
+-	if(sigprocmask(SIG_SETMASK, &old_mask, NULL) == -1)
+-		BAD_ERROR("Failed to set signal mask in intialise_threads\n");
+-}
+-
+-
+ #define VERSION() \
+-	printf("mksquashfs version 3.1-r2 (2006/08/30)\n");\
++	printf("mksquashfs version 3.0 (2006/03/15)\n");\
+ 	printf("copyright (C) 2006 Phillip Lougher <phillip@lougher.org.uk>\n\n"); \
+     	printf("This program is free software; you can redistribute it and/or\n");\
+ 	printf("modify it under the terms of the GNU General Public License\n");\
+@@ -2682,7 +1810,6 @@
+ 	char *b, *root_name = NULL;
+ 	int be, nopad = FALSE, keep_as_directory = FALSE, orig_be;
+ 	squashfs_inode inode;
+-	int readb_mbytes = READER_BUFFER_DEFAULT, writeb_mbytes = WRITER_BUFFER_DEFAULT;
+ 
+ #if __BYTE_ORDER == __BIG_ENDIAN
+ 	be = TRUE;
+@@ -2701,34 +1828,7 @@
+ 	source_path = argv + 1;
+ 	source = i - 2;
+ 	for(; i < argc; i++) {
+-		if(strcmp(argv[i], "-processors") == 0) {
+-			if((++i == argc) || (processors = strtol(argv[i], &b, 10), *b != '\0')) {
+-				ERROR("%s: -processors missing or invalid processor number\n", argv[0]);
+-				exit(1);
+-			}
+-			if(processors < 1) {
+-				ERROR("%s: -processors should be 1 or larger\n", argv[0]);
+-				exit(1);
+-			}
+-		} else if(strcmp(argv[i], "-read_queue") == 0) {
+-			if((++i == argc) || (readb_mbytes = strtol(argv[i], &b, 10), *b != '\0')) {
+-				ERROR("%s: -read_queue missing or invalid queue size\n", argv[0]);
+-				exit(1);
+-			}
+-			if(readb_mbytes < 1) {
+-				ERROR("%s: -read_queue should be 1 megabyte or larger\n", argv[0]);
+-				exit(1);
+-			}
+-		} else if(strcmp(argv[i], "-write_queue") == 0) {
+-			if((++i == argc) || (writeb_mbytes = strtol(argv[i], &b, 10), *b != '\0')) {
+-				ERROR("%s: -write_queue missing or invalid queue size\n", argv[0]);
+-				exit(1);
+-			}
+-			if(writeb_mbytes < 1) {
+-				ERROR("%s: -write_queue should be 1 megabyte or larger\n", argv[0]);
+-				exit(1);
+-			}
+-		} else if(strcmp(argv[i], "-b") == 0) {
++		if(strcmp(argv[i], "-b") == 0) {
+ 			if((++i == argc) || (block_size = strtol(argv[i], &b, 10), *b !='\0')) {
+ 				ERROR("%s: -b missing or invalid block size\n", argv[0]);
+ 				exit(1);
+@@ -2851,9 +1951,6 @@
+ 			ERROR("-version\t\tprint version, licence and copyright message\n");
+ 			ERROR("-info\t\t\tprint files written to filesystem\n");
+ 			ERROR("-b <block_size>\t\tset data block to <block_size>.  Default %d bytes\n", SQUASHFS_FILE_SIZE);
+-			ERROR("-processors <number>\tUse <number> processors.  By default will use number of\n\t\t\tprocessors available\n");
+-			ERROR("-read-queue <size>\tSet input queue to <size> Mbytes.  Default %d Mbytes\n", READER_BUFFER_DEFAULT);
+-			ERROR("-write-queue <size>\tSet output queue to <size> Mbytes.  Default %d Mbytes\n", WRITER_BUFFER_DEFAULT);
+ 			ERROR("-noI\t\t\tdo not compress inode table\n");
+ 			ERROR("-noD\t\t\tdo not compress data blocks\n");
+ 			ERROR("-noF\t\t\tdo not compress fragment blocks\n");
+@@ -2887,9 +1984,6 @@
+ 		}
+ 	}
+ 
+-	reader_buffer_size = readb_mbytes << (20 - block_log);
+-	writer_buffer_size = writeb_mbytes << (20 - block_log);
+-
+         for(i = 0; i < source; i++)
+                 if(stat(source_path[i], &source_buf) == -1) {
+                         fprintf(stderr, "Cannot stat source directory \"%s\" because %s\n", source_path[i], strerror(errno));
+@@ -2975,8 +2069,6 @@
+ 		else if(strcmp(argv[i], "-b") == 0 || strcmp(argv[i], "-root-becomes") == 0 || strcmp(argv[i], "-ef") == 0)
+ 			i++;
+ 
+-	initialise_threads();
+-
+ 	if(delete) {
+ 		printf("Creating %s %d.%d filesystem on %s, block size %d.\n",
+ 				be ? "big endian" : "little endian", SQUASHFS_MAJOR, SQUASHFS_MINOR, argv[source + 1], block_size);
+@@ -3104,13 +2196,6 @@
+ restore_filesystem:
+ 	write_fragment();
+ 	sBlk.fragments = fragments;
+-	if(interrupted < 2) {
+-		ensure_fragments_flushed();
+-		queue_put(to_writer, NULL);
+-		if(queue_get(from_writer) != 0)
+-			EXIT_MKSQUASHFS();
+-	}
+-
+ 	sBlk.inode_table_start = write_inodes();
+ 	sBlk.directory_table_start = write_directories();
+ 	sBlk.fragment_table_start = write_fragment_table();
+diff -Nur squashfs-tools/read_fs.c squashfs-lzma-tools/read_fs.c
+--- squashfs-tools/read_fs.c	2006-08-21 01:40:35.000000000 +0200
++++ squashfs-lzma-tools/read_fs.c	2006-08-22 09:39:31.000000000 +0200
+@@ -22,7 +22,7 @@
+  */
+ 
+ extern void read_bytes(int, long long, int, char *);
+-extern int add_file(long long, long long, long long, unsigned int *, int, unsigned int, int, int);
++extern int add_file(long long, long long, unsigned int *, int, unsigned int, int, int);
+ 
+ #define TRUE 1
+ #define FALSE 0
+@@ -186,8 +186,7 @@
+ 					+ sBlk->block_size - 1) >> sBlk->block_log : inode.file_size >>
+ 					sBlk->block_log;
+ 				long long file_bytes = 0;
+-				int i;
+-				long long start = inode.start_block;
++				int i, start = inode.start_block;
+ 				unsigned int *block_list;
+ 
+ 				TRACE("scan_inode_table: regular file, file_size %lld, blocks %d\n", inode.file_size, blocks);
+@@ -211,7 +210,7 @@
+ 				for(i = 0; i < blocks; i++)
+ 					file_bytes += SQUASHFS_COMPRESSED_SIZE_BLOCK(block_list[i]);
+ 
+-	                        add_file(start, inode.file_size, file_bytes, block_list, blocks, inode.fragment, inode.offset, frag_bytes);
++	                        add_file(start, file_bytes, block_list, blocks, inode.fragment, inode.offset, frag_bytes);
+ 				cur_ptr += blocks * sizeof(unsigned int);
+ 				break;
+ 			}	
+@@ -220,8 +219,7 @@
+ 				int frag_bytes;
+ 				int blocks;
+ 				long long file_bytes = 0;
+-				int i;
+-				long long start;
++				int i, start;
+ 				unsigned int *block_list;
+ 
+ 				if(swap) {
+@@ -258,7 +256,7 @@
+ 				for(i = 0; i < blocks; i++)
+ 					file_bytes += SQUASHFS_COMPRESSED_SIZE_BLOCK(block_list[i]);
+ 
+-	                        add_file(start, inode.file_size, file_bytes, block_list, blocks, inode.fragment, inode.offset, frag_bytes);
++	                        add_file(start, file_bytes, block_list, blocks, inode.fragment, inode.offset, frag_bytes);
+ 				cur_ptr += blocks * sizeof(unsigned int);
+ 				break;
+ 			}	
+diff -Nur squashfs-tools/README squashfs-lzma-tools/README
+--- squashfs-tools/README	1970-01-01 01:00:00.000000000 +0100
++++ squashfs-lzma-tools/README	2006-08-22 09:36:18.000000000 +0200
+@@ -0,0 +1,2 @@
++This is mksquashfs patched with LZMA support
++and with the patch for 'fragment_table rounding bug'
+\ No newline at end of file
+diff -Nur squashfs-tools/sort.c squashfs-lzma-tools/sort.c
+--- squashfs-tools/sort.c	2006-08-21 01:40:35.000000000 +0200
++++ squashfs-lzma-tools/sort.c	2006-06-25 05:17:43.000000000 +0200
+@@ -71,6 +71,11 @@
+ 
+ struct sort_info *sort_info_list[65536];
+ 
++struct priority_entry {
++	struct dir_ent *dir;
++	struct priority_entry *next;
++};
++
+ struct priority_entry *priority_list[65536];
+ 
+ extern int silent;
+@@ -230,6 +235,8 @@
+ 	squashfs_inode inode;
+ 	int duplicate_file;
+ 
++	generate_file_priorities(dir, 0, &dir->dir_ent->inode->buf);
++
+ 	for(i = 65535; i >= 0; i--)
+ 		for(entry = priority_list[i]; entry; entry = entry->next) {
+ 			TRACE("%d: %s\n", i - 32768, entry->dir->pathname);
+diff -Nur squashfs-tools/sort.h squashfs-lzma-tools/sort.h
+--- squashfs-tools/sort.h	2006-08-21 01:40:35.000000000 +0200
++++ squashfs-lzma-tools/sort.h	2006-06-25 05:17:43.000000000 +0200
+@@ -51,12 +51,6 @@
+ 	squashfs_inode		inode;
+ 	unsigned int		type;
+ 	unsigned int		inode_number;
+-	char			read;
+ 	struct inode_info	*next;
+ };
+-
+-struct priority_entry {
+-	struct dir_ent *dir;
+-	struct priority_entry *next;
+-};
+ #endif
+diff -Nur squashfs-tools/squashfs_fs.h squashfs-lzma-tools/squashfs_fs.h
+--- squashfs-tools/squashfs_fs.h	2006-08-21 02:00:22.000000000 +0200
++++ squashfs-lzma-tools/squashfs_fs.h	2006-06-25 05:17:43.000000000 +0200
+@@ -308,7 +308,7 @@
+ struct squashfs_fragment_entry {
+ 	long long		start_block;
+ 	unsigned int		size;
+-	unsigned int		pending;
++	unsigned int		unused;
+ } __attribute__ ((packed));
+ 
+ extern int squashfs_uncompress_block(void *d, int dstlen, void *s, int srclen);