Remove WIN32, WIN64 and MINGW32 tentacles.

Also check for _LP64 rather than __arch64__ (the former being more reliable
than __LP64__ or __arch64__) to tell 64-bit int platforms apart from 32-bit
int platforms.

Loosely based upon a diff from Martijn van Duren on tech@

14 files changed, 62 insertions, 89 deletions

diff --git a/lib/libssl/src/crypto/aes/aes_x86core.c b/lib/libssl/src/crypto/aes/aes_x86core.c
index 295ea22bb46..8b3b29e28c6 100644
--- a/lib/libssl/src/crypto/aes/aes_x86core.c
+++ b/lib/libssl/src/crypto/aes/aes_x86core.c
@@ -79,10 +79,7 @@ prefetch256(const void *table)
 #undef GETU32
 #define GETU32(p) (*((u32*)(p)))
 
-#if (defined(_WIN32) || defined(_WIN64)) && !defined(__MINGW32__)
-typedef unsigned __int64 u64;
-#define U64(C)	C##UI64
-#elif defined(__arch64__)
+#if defined(_LP64)
 typedef unsigned long u64;
 #define U64(C)	C##UL
 #else
diff --git a/lib/libssl/src/crypto/bn/asm/x86_64-gcc.c b/lib/libssl/src/crypto/bn/asm/x86_64-gcc.c
index b97b394661a..6a7fa4aea95 100644
--- a/lib/libssl/src/crypto/bn/asm/x86_64-gcc.c
+++ b/lib/libssl/src/crypto/bn/asm/x86_64-gcc.c
@@ -1,3 +1,7 @@
+#include "../bn_lcl.h"
+#if !(defined(__GNUC__) && __GNUC__>=2)
+# include "../bn_asm.c"	/* kind of dirty hack for Sun Studio */
+#else
 /*
  * x86_64 BIGNUM accelerator version 0.1, December 2002.
  *
@@ -13,24 +17,50 @@
  * A. Well, that's because this code is basically a quick-n-dirty
  *    proof-of-concept hack. As you can see it's implemented with
  *    inline assembler, which means that you're bound to GCC and that
- *    there must be a room for fine-tuning.
+ *    there might be enough room for further improvement.
  *
  * Q. Why inline assembler?
- * A. x86_64 features own ABI I'm not familiar with. Which is why
- *    I decided to let the compiler take care of subroutine
- *    prologue/epilogue as well as register allocation.
+ * A. x86_64 features own ABI which I'm not familiar with. This is
+ *    why I decided to let the compiler take care of subroutine
+ *    prologue/epilogue as well as register allocation. For reference.
+ *    Win64 implements different ABI for AMD64, different from Linux.
  *
  * Q. How much faster does it get?
- * A. Unfortunately people sitting on x86_64 hardware are prohibited
- *    to disclose the performance numbers, so they (SuSE labs to be
- *    specific) wouldn't tell me. However! Very similar coding technique
- *    (reaching out for 128-bit result from 64x64-bit multiplication)
- *    results in >3 times performance improvement on MIPS and I see no
- *    reason why gain on x86_64 would be so much different:-)
+ * A. 'apps/openssl speed rsa dsa' output with no-asm:
+ *
+ *	                  sign    verify    sign/s verify/s
+ *	rsa  512 bits   0.0006s   0.0001s   1683.8  18456.2
+ *	rsa 1024 bits   0.0028s   0.0002s    356.0   6407.0
+ *	rsa 2048 bits   0.0172s   0.0005s     58.0   1957.8
+ *	rsa 4096 bits   0.1155s   0.0018s      8.7    555.6
+ *	                  sign    verify    sign/s verify/s
+ *	dsa  512 bits   0.0005s   0.0006s   2100.8   1768.3
+ *	dsa 1024 bits   0.0014s   0.0018s    692.3    559.2
+ *	dsa 2048 bits   0.0049s   0.0061s    204.7    165.0
+ *
+ *    'apps/openssl speed rsa dsa' output with this module:
+ *
+ *	                  sign    verify    sign/s verify/s
+ *	rsa  512 bits   0.0004s   0.0000s   2767.1  33297.9
+ *	rsa 1024 bits   0.0012s   0.0001s    867.4  14674.7
+ *	rsa 2048 bits   0.0061s   0.0002s    164.0   5270.0
+ *	rsa 4096 bits   0.0384s   0.0006s     26.1   1650.8
+ *	                  sign    verify    sign/s verify/s
+ *	dsa  512 bits   0.0002s   0.0003s   4442.2   3786.3
+ *	dsa 1024 bits   0.0005s   0.0007s   1835.1   1497.4
+ *	dsa 2048 bits   0.0016s   0.0020s    620.4    504.6
+ *
+ *    For the reference. IA-32 assembler implementation performs
+ *    very much like 64-bit code compiled with no-asm on the same
+ *    machine.
  */
 
 #define BN_ULONG unsigned long
 
+#undef mul
+#undef mul_add
+#undef sqr
+
 /*
  * "m"(a), "+m"(r)	is the way to favor DirectPath µ-code;
  * "g"(0)		let the compiler to decide where does it
@@ -72,7 +102,7 @@
 		: "a"(a)		\
 		: "cc");
 
-BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
+BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
 	{
 	BN_ULONG c1=0;
 
@@ -96,7 +126,7 @@ BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
 	return(c1);
 	} 
 
-BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
+BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
 	{
 	BN_ULONG c1=0;
 
@@ -119,7 +149,7 @@ BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
 	return(c1);
 	} 
 
-void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n)
+void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n)
         {
 	if (n <= 0) return;
 
@@ -142,7 +172,7 @@ void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n)
 BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
 {	BN_ULONG ret,waste;
 
-	asm ("divq	%3"
+	asm ("divq	%4"
 		: "=a"(ret),"=d"(waste)
 		: "a"(l),"d"(h),"g"(d)
 		: "cc");
@@ -150,21 +180,21 @@ BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
 	return ret;
 }
 
-BN_ULONG bn_add_words (BN_ULONG *rp, BN_ULONG *ap, BN_ULONG *bp,int n)
-{ BN_ULONG ret,i;
+BN_ULONG bn_add_words (BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,int n)
+{ BN_ULONG ret=0,i=0;
 
 	if (n <= 0) return 0;
 
 	asm (
 	"	subq	%2,%2		\n"
-	".align 16			\n"
+	".p2align 4			\n"
 	"1:	movq	(%4,%2,8),%0	\n"
 	"	adcq	(%5,%2,8),%0	\n"
 	"	movq	%0,(%3,%2,8)	\n"
 	"	leaq	1(%2),%2	\n"
 	"	loop	1b		\n"
 	"	sbbq	%0,%0		\n"
-		: "+a"(ret),"+c"(n),"+r"(i)
+		: "=&a"(ret),"+c"(n),"=&r"(i)
 		: "r"(rp),"r"(ap),"r"(bp)
 		: "cc"
 	);
@@ -173,21 +203,21 @@ BN_ULONG bn_add_words (BN_ULONG *rp, BN_ULONG *ap, BN_ULONG *bp,int n)
 }
 
 #ifndef SIMICS
-BN_ULONG bn_sub_words (BN_ULONG *rp, BN_ULONG *ap, BN_ULONG *bp,int n)
-{ BN_ULONG ret,i;
+BN_ULONG bn_sub_words (BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,int n)
+{ BN_ULONG ret=0,i=0;
 
 	if (n <= 0) return 0;
 
 	asm (
 	"	subq	%2,%2		\n"
-	".align 16			\n"
+	".p2align 4			\n"
 	"1:	movq	(%4,%2,8),%0	\n"
 	"	sbbq	(%5,%2,8),%0	\n"
 	"	movq	%0,(%3,%2,8)	\n"
 	"	leaq	1(%2),%2	\n"
 	"	loop	1b		\n"
 	"	sbbq	%0,%0		\n"
-		: "+a"(ret),"+c"(n),"+r"(i)
+		: "=&a"(ret),"+c"(n),"=&r"(i)
 		: "r"(rp),"r"(ap),"r"(bp)
 		: "cc"
 	);
@@ -318,7 +348,6 @@ BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
 
 void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
 	{
-	BN_ULONG bl,bh;
 	BN_ULONG t1,t2;
 	BN_ULONG c1,c2,c3;
 
@@ -423,7 +452,6 @@ void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
 
 void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
 	{
-	BN_ULONG bl,bh;
 	BN_ULONG t1,t2;
 	BN_ULONG c1,c2,c3;
 
@@ -462,9 +490,8 @@ void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
 	r[7]=c2;
 	}
 
-void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
+void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a)
 	{
-	BN_ULONG bl,bh;
 	BN_ULONG t1,t2;
 	BN_ULONG c1,c2,c3;
 
@@ -539,9 +566,8 @@ void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
 	r[15]=c1;
 	}
 
-void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a)
+void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a)
 	{
-	BN_ULONG bl,bh;
 	BN_ULONG t1,t2;
 	BN_ULONG c1,c2,c3;
 
@@ -573,3 +599,4 @@ void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a)
 	r[6]=c1;
 	r[7]=c2;
 	}
+#endif
diff --git a/lib/libssl/src/crypto/bn/bn_exp.c b/lib/libssl/src/crypto/bn/bn_exp.c
index 2047e1cc3f0..22ef643c02c 100644
--- a/lib/libssl/src/crypto/bn/bn_exp.c
+++ b/lib/libssl/src/crypto/bn/bn_exp.c
@@ -114,12 +114,7 @@
 #include "bn_lcl.h"
 
 #include <stdlib.h>
-#ifdef _WIN32
-# include <malloc.h>
-# ifndef alloca
-#  define alloca _alloca
-# endif
-#elif defined(__GNUC__)
+#if defined(__GNUC__)
 # ifndef alloca
 #  define alloca(s) __builtin_alloca((s))
 # endif
diff --git a/lib/libssl/src/crypto/cast/cast_lcl.h b/lib/libssl/src/crypto/cast/cast_lcl.h
index ec14804fbbd..61facd09d8a 100644
--- a/lib/libssl/src/crypto/cast/cast_lcl.h
+++ b/lib/libssl/src/crypto/cast/cast_lcl.h
@@ -56,11 +56,6 @@
  * [including the GNU Public Licence.]
  */
 
-#ifdef OPENSSL_SYS_WIN32
-#include <stdlib.h>
-#endif
-
-
 #undef c2l
 #define c2l(c,l)	(l =((unsigned long)(*((c)++)))    , \
 			 l|=((unsigned long)(*((c)++)))<< 8L, \
diff --git a/lib/libssl/src/crypto/des/enc_read.c b/lib/libssl/src/crypto/des/enc_read.c
index 23ad458dcf6..e1ac04c5b39 100644
--- a/lib/libssl/src/crypto/des/enc_read.c
+++ b/lib/libssl/src/crypto/des/enc_read.c
@@ -150,11 +150,7 @@ int DES_enc_read(int fd, void *buf, int len, DES_key_schedule *sched,
 	/* first - get the length */
 	while (net_num < HDRSIZE) 
 		{
-#ifndef OPENSSL_SYS_WIN32
 		i=read(fd,(void *)&(net[net_num]),HDRSIZE-net_num);
-#else
-		i=_read(fd,(void *)&(net[net_num]),HDRSIZE-net_num);
-#endif
 #ifdef EINTR
 		if ((i == -1) && (errno == EINTR)) continue;
 #endif
@@ -176,11 +172,7 @@ int DES_enc_read(int fd, void *buf, int len, DES_key_schedule *sched,
 	net_num=0;
 	while (net_num < rnum)
 		{
-#ifndef OPENSSL_SYS_WIN32
 		i=read(fd,(void *)&(net[net_num]),rnum-net_num);
-#else
-		i=_read(fd,(void *)&(net[net_num]),rnum-net_num);
-#endif
 #ifdef EINTR
 		if ((i == -1) && (errno == EINTR)) continue;
 #endif
diff --git a/lib/libssl/src/crypto/des/enc_writ.c b/lib/libssl/src/crypto/des/enc_writ.c
index 8f6b033c877..18562310eed 100644
--- a/lib/libssl/src/crypto/des/enc_writ.c
+++ b/lib/libssl/src/crypto/des/enc_writ.c
@@ -156,11 +156,7 @@ int DES_enc_write(int fd, const void *_buf, int len,
 		{
 		/* eay 26/08/92 I was not doing writing from where we
 		 * got up to. */
-#ifndef _WIN32
 		i=write(fd,(void *)&(outbuf[j]),outnum-j);
-#else
-		i=_write(fd,(void *)&(outbuf[j]),outnum-j);
-#endif
 		if (i == -1)
 			{
 #ifdef EINTR
diff --git a/lib/libssl/src/crypto/engine/eng_aesni.c b/lib/libssl/src/crypto/engine/eng_aesni.c
index 5fdb33bfded..d547d7f4656 100644
--- a/lib/libssl/src/crypto/engine/eng_aesni.c
+++ b/lib/libssl/src/crypto/engine/eng_aesni.c
@@ -309,11 +309,7 @@ static void aesni_ofb128_encrypt(const unsigned char *in, unsigned char *out,
 }
 /* ===== Engine "management" functions ===== */
 
-#if defined(_WIN32)
-typedef unsigned __int64 IA32CAP;
-#else
 typedef unsigned long long IA32CAP;
-#endif
 
 /* Prepare the ENGINE structure for registration */
 static int
diff --git a/lib/libssl/src/crypto/engine/eng_padlock.c b/lib/libssl/src/crypto/engine/eng_padlock.c
index c27181ba758..5a80b2b16da 100644
--- a/lib/libssl/src/crypto/engine/eng_padlock.c
+++ b/lib/libssl/src/crypto/engine/eng_padlock.c
@@ -129,12 +129,7 @@ void ENGINE_load_padlock (void)
 /* We do these includes here to avoid header problems on platforms that
    do not have the VIA padlock anyway... */
 #include <stdlib.h>
-#ifdef _WIN32
-# include <malloc.h>
-# ifndef alloca
-#  define alloca _alloca
-# endif
-#elif defined(__GNUC__)
+#if defined(__GNUC__)
 # ifndef alloca
 #  define alloca(s) __builtin_alloca(s)
 # endif
diff --git a/lib/libssl/src/crypto/engine/eng_rsax.c b/lib/libssl/src/crypto/engine/eng_rsax.c
index c0f6851601c..1b15b6f1a3f 100644
--- a/lib/libssl/src/crypto/engine/eng_rsax.c
+++ b/lib/libssl/src/crypto/engine/eng_rsax.c
@@ -217,11 +217,7 @@ static int e_rsax_ctrl(ENGINE *e, int cmd, long i, void *p, void (*f)(void))
 
 #ifndef OPENSSL_NO_RSA
 
-#ifdef _WIN32
-typedef unsigned __int64 UINT64;
-#else
 typedef unsigned long long UINT64;
-#endif
 typedef unsigned short UINT16;
 
 /* Table t is interleaved in the following manner:
diff --git a/lib/libssl/src/crypto/modes/modes_lcl.h b/lib/libssl/src/crypto/modes/modes_lcl.h
index 2fc81382734..68c0e355ad9 100644
--- a/lib/libssl/src/crypto/modes/modes_lcl.h
+++ b/lib/libssl/src/crypto/modes/modes_lcl.h
@@ -9,11 +9,7 @@
 #include <machine/endian.h>
 
 
-#if (defined(_WIN32) || defined(_WIN64)) && !defined(__MINGW32__)
-typedef __int64 i64;
-typedef unsigned __int64 u64;
-#define U64(C) C##UI64
-#elif defined(__arch64__)
+#if defined(_LP64)
 typedef long i64;
 typedef unsigned long u64;
 #define U64(C) C##UL
diff --git a/lib/libssl/src/crypto/sha/sha.h b/lib/libssl/src/crypto/sha/sha.h
index 435352c2080..c0c3c16c08a 100644
--- a/lib/libssl/src/crypto/sha/sha.h
+++ b/lib/libssl/src/crypto/sha/sha.h
@@ -154,10 +154,7 @@ void SHA256_Transform(SHA256_CTX *c, const unsigned char *data);
 #define SHA512_CBLOCK	(SHA_LBLOCK*8)	/* SHA-512 treats input data as a
 					 * contiguous array of 64 bit
 					 * wide big-endian values. */
-#if (defined(_WIN32) || defined(_WIN64)) && !defined(__MINGW32__)
-#define SHA_LONG64 unsigned __int64
-#define U64(C)     C##UI64
-#elif defined(__arch64__)
+#if defined(_LP64)
 #define SHA_LONG64 unsigned long
 #define U64(C)     C##UL
 #else
diff --git a/lib/libssl/src/crypto/srp/srp_lib.c b/lib/libssl/src/crypto/srp/srp_lib.c
index 8cc94f51db8..a3a67eda2e2 100644
--- a/lib/libssl/src/crypto/srp/srp_lib.c
+++ b/lib/libssl/src/crypto/srp/srp_lib.c
@@ -63,9 +63,7 @@
 #include <openssl/evp.h>
 
 #if (BN_BYTES == 8)
-# if (defined(_WIN32) || defined(_WIN64)) && !defined(__MINGW32__)
-#  define bn_pack4(a1,a2,a3,a4) ((a1##UI64<<48)|(a2##UI64<<32)|(a3##UI64<<16)|a4##UI64)
-# elif defined(__arch64__)
+# if defined(_LP64)
 #  define bn_pack4(a1,a2,a3,a4) ((a1##UL<<48)|(a2##UL<<32)|(a3##UL<<16)|a4##UL)
 # else
 #  define bn_pack4(a1,a2,a3,a4) ((a1##ULL<<48)|(a2##ULL<<32)|(a3##ULL<<16)|a4##ULL)
diff --git a/lib/libssl/src/crypto/ts/ts.h b/lib/libssl/src/crypto/ts/ts.h
index 3c5ab727db5..085e062b96c 100644
--- a/lib/libssl/src/crypto/ts/ts.h
+++ b/lib/libssl/src/crypto/ts/ts.h
@@ -89,11 +89,6 @@
 extern "C" {
 #endif
 
-#ifdef WIN32
-/* Under Win32 this is defined in wincrypt.h */
-#undef X509_NAME
-#endif
-
 #include <openssl/x509.h>
 #include <openssl/x509v3.h>
 
diff --git a/lib/libssl/src/crypto/whrlpool/wp_block.c b/lib/libssl/src/crypto/whrlpool/wp_block.c
index 11164e50f56..9c194f968b1 100644
--- a/lib/libssl/src/crypto/whrlpool/wp_block.c
+++ b/lib/libssl/src/crypto/whrlpool/wp_block.c
@@ -40,9 +40,7 @@
 #include <machine/endian.h>
 
 typedef unsigned char		u8;
-#if (defined(_WIN32) || defined(_WIN64)) && !defined(__MINGW32)
-typedef unsigned __int64	u64;
-#elif defined(__arch64__)
+#if defined(_LP64)
 typedef unsigned long		u64;
 #else
 typedef unsigned long long	u64;