port to Sun Studio 12u1 Sun C++ 5.10 SunOS_i386 128229-02 2009/09/21

change makefile to compile for both i386 and x86_64 on Darwin/Mac OS X

GNUmakefile

@@ -31,7 +31,9 @@ GAS217_OR_LATER = $(shell echo "" | $(AS) -v 2>&1 | $(EGREP) -c "GNU assembler v
 ISMINGW = $(shell $(CXX) --version 2>&1 | $(EGREP) -c "mingw")
 
 ifneq ($(GCC42_OR_LATER),0)
-ifneq ($(UNAME),Darwin)
+ifeq ($(UNAME),Darwin)
+CXXFLAGS += -arch x86_64 -arch i386
+else
 CXXFLAGS += -march=native -mtune=native
 endif
 endif
@@ -86,6 +88,9 @@ LDLIBS += -lnsl -lsocket
 ifeq ($(CXX),CC)	# override flags for CC (Solaris native C++ compiler)
 CXXFLAGS = -DNDEBUG -O -g0 -native -template=no%extdef -m$(shell isainfo -b)
 LDFLAGS =
+AR = CC
+ARFLAGS = -xar -o
+RANLIB = true
 ifeq ($(ISX86),1)
 # SSE2 intrinsics should work in Sun Studio 12, but we're not using SSE2 intrinsics anymore
 # CXXFLAGS += -xarch=sse2 -D__SSE2__

idea.cpp

@@ -129,7 +129,7 @@ static inline IDEA::Word AddInv(IDEA::Word x)
 void IDEA::Base::DeKey()
 {
 	FixedSizeSecBlock<IDEA::Word, 6*ROUNDS+4> tempkey;
-	unsigned int i;
+	size_t i;
 
 	for (i=0; i<ROUNDS; i++)
 	{

integer.cpp

@@ -115,7 +115,7 @@ static word AtomicInverseModPower2(word A)
 	#elif defined(__DECCXX)
 		#define MultiplyWordsLoHi(p0, p1, a, b)		p0 = a*b; p1 = asm("umulh %a0, %a1, %v0", a, b);
 	#elif defined(__x86_64__)
-		#ifdef __SUNPRO_CC
+		#if defined(__SUNPRO_CC) && __SUNPRO_CC < 0x5100
 			// Sun Studio's gcc-style inline assembly is heavily bugged as of version 5.9 Patch 124864-09 2008/12/16, but this one works
 			#define MultiplyWordsLoHi(p0, p1, a, b)		asm ("mulq %3" : "=a"(p0), "=d"(p1) : "a"(a), "r"(b) : "cc");
 		#else

stdcpp.h

@@ -14,6 +14,11 @@
 #include <map>
 #include <vector>
 
+#ifdef __SUNPRO_CC
+// workaround needed on Sun Studio 12u1 Sun C++ 5.10 SunOS_i386 128229-02 2009/09/21
+#include <vector.cc>
+#endif
+
 // for alloca
 #ifdef __sun
 #include <alloca.h>

tea.cpp

@@ -62,8 +62,14 @@ void XTEA::Enc::ProcessAndXorBlock(const byte *inBlock, const byte *xorBlock, by
 	word32 y, z;
 	Block::Get(inBlock)(y)(z);
 
+#ifdef __SUNPRO_CC
+	// workaround needed on Sun Studio 12u1 Sun C++ 5.10 SunOS_i386 128229-02 2009/09/21
+	size_t sum = 0;
+	while ((sum&0xffffffff) != m_limit)
+#else
 	word32 sum = 0;
 	while (sum != m_limit)
+#endif
 	{   
 		y += (z<<4 ^ z>>5) + z ^ sum + m_k[sum&3];
 		sum += DELTA;
@@ -78,8 +84,14 @@ void XTEA::Dec::ProcessAndXorBlock(const byte *inBlock, const byte *xorBlock, by
 	word32 y, z;
 	Block::Get(inBlock)(y)(z);
 
+#ifdef __SUNPRO_CC
+	// workaround needed on Sun Studio 12u1 Sun C++ 5.10 SunOS_i386 128229-02 2009/09/21
+	size_t sum = m_limit;
+	while ((sum&0xffffffff) != 0)
+#else
 	word32 sum = m_limit;
 	while (sum != 0)
+#endif
 	{
 		z -= (y<<4 ^ y>>5) + y ^ sum + m_k[sum>>11 & 3];
 		sum -= DELTA;

validat1.cpp

@@ -130,7 +130,10 @@ bool TestSettings()
 
 	cout << "\nTesting Settings...\n\n";
 
-	if (*(word32 *)"\x01\x02\x03\x04" == 0x04030201L)
+	word32 w;
+	memcpy_s(&w, sizeof(w), "\x01\x02\x03\x04", 4);
+
+	if (w == 0x04030201L)
 	{
 #ifdef IS_LITTLE_ENDIAN
 		cout << "passed:  ";
@@ -140,7 +143,7 @@ bool TestSettings()
 #endif
 		cout << "Your machine is little endian.\n";
 	}
-	else if (*(word32 *)"\x01\x02\x03\x04" == 0x01020304L)
+	else if (w == 0x01020304L)
 	{
 #ifndef IS_LITTLE_ENDIAN
 		cout << "passed:  ";

wake.cpp

@@ -16,22 +16,24 @@ void WAKE_TestInstantiations()
 inline word32 WAKE_Base::M(word32 x, word32 y)
 {
 	word32 w = x+y;
-	return (w>>8) ^ t[(byte)w];
+	return (w>>8) ^ t[w & 0xff];
 }
 
 void WAKE_Base::GenKey(word32 k0, word32 k1, word32 k2, word32 k3)
 {
-	long x, z;
-	int p ;
-	static long tt[10]= {
-		0x726a8f3bL,								 // table
-		0xe69a3b5cL,
-		0xd3c71fe5L,
-		0xab3c73d2L,
-		0x4d3a8eb3L,
-		0x0396d6e8L,
-		0x3d4c2f7aL,
-		0x9ee27cf3L, } ;
+	// this code is mostly copied from David Wheeler's paper "A Bulk Data Encryption Algorithm"
+	signed int x, z, p;	
+	// x and z were declared as "long" in Wheeler's paper, which is a signed type. I don't know if that was intentional, but it's too late to change it now. -- Wei 7/4/2010
+	CRYPTOPP_COMPILE_ASSERT(sizeof(x) == 4);
+	static int tt[10]= {
+		0x726a8f3b,								 // table
+		0xe69a3b5c,
+		0xd3c71fe5,
+		0xab3c73d2,
+		0x4d3a8eb3,
+		0x0396d6e8,
+		0x3d4c2f7a,
+		0x9ee27cf3, } ;
 	t[0] = k0;
 	t[1] = k1;
 	t[2] = k2;
@@ -39,16 +41,16 @@ void WAKE_Base::GenKey(word32 k0, word32 k1, word32 k2, word32 k3)
 	for (p=4 ; p<256 ; p++)
 	{
 	  x=t[p-4]+t[p-1] ; 					   // fill t
-	  t[p]= (x>>3) ^ tt[byte(x&7)] ;
+	  t[p]= (x>>3) ^ tt[x&7] ;
 	}
 
 	for (p=0 ; p<23 ; p++)
 		t[p]+=t[p+89] ; 		  // mix first entries
-	x=t[33] ; z=t[59] | 0x01000001L ;
-	z=z&0xff7fffffL ;
+	x=t[33] ; z=t[59] | 0x01000001 ;
+	z=z&0xff7fffff ;
 	for (p=0 ; p<256 ; p++) {		//change top byte to
-	  x=(x&0xff7fffffL)+z ; 		 // a permutation etc
-	  t[p]=(t[p] & 0x00ffffffL) ^ x ; }
+	  x=(x&0xff7fffff)+z ; 		 // a permutation etc
+	  t[p]=(t[p] & 0x00ffffff) ^ x ; }
 
 	t[256]=t[0] ;
 	byte y=byte(x);