Replace all uses of magic numbers when operating on OPENSSL_ia32_P[] by

meaningful constants in a private header file, so that reviewers can actually
get a chance to figure out what the code is attempting to do without knowing
all cpuid bits.

While there, turn it from an array of two 32-bit ints into a properly aligned
64-bit int.

Use of OPENSSL_ia32_P is now restricted to the assembler parts. C code will
now always use OPENSSL_cpu_caps() and check for the proper bits in the
whole 64-bit word it returns.

i386 tests and ok jsing@

4 files changed, 16 insertions, 13 deletions

diff --git a/lib/libcrypto/bn/asm/bn-586.pl b/lib/libcrypto/bn/asm/bn-586.pl
index 332ef3e91d6..c4e2baa6c5a 100644
--- a/lib/libcrypto/bn/asm/bn-586.pl
+++ b/lib/libcrypto/bn/asm/bn-586.pl
@@ -25,7 +25,7 @@ sub bn_mul_add_words
 	{
 	local($name)=@_;
 
-	&function_begin_B($name,$sse2?"EXTRN\t_OPENSSL_ia32cap_P:DWORD":"");
+	&function_begin_B($name,"");
 
 	$r="eax";
 	$a="edx";
@@ -33,7 +33,7 @@ sub bn_mul_add_words
 
 	if ($sse2) {
 		&picmeup("eax","OPENSSL_ia32cap_P");
-		&bt(&DWP(0,"eax"),26);
+		&bt(&DWP(0,"eax"),"\$IA32CAP_BIT0_SSE2");
 		&jnc(&label("maw_non_sse2"));
 
 		&mov($r,&wparam(0));
@@ -211,7 +211,7 @@ sub bn_mul_words
 	{
 	local($name)=@_;
 
-	&function_begin_B($name,$sse2?"EXTRN\t_OPENSSL_ia32cap_P:DWORD":"");
+	&function_begin_B($name,"");
 
 	$r="eax";
 	$a="edx";
@@ -219,7 +219,7 @@ sub bn_mul_words
 
 	if ($sse2) {
 		&picmeup("eax","OPENSSL_ia32cap_P");
-		&bt(&DWP(0,"eax"),26);
+		&bt(&DWP(0,"eax"),"\$IA32CAP_BIT0_SSE2");
 		&jnc(&label("mw_non_sse2"));
 
 		&mov($r,&wparam(0));
@@ -322,7 +322,7 @@ sub bn_sqr_words
 	{
 	local($name)=@_;
 
-	&function_begin_B($name,$sse2?"EXTRN\t_OPENSSL_ia32cap_P:DWORD":"");
+	&function_begin_B($name,"");
 
 	$r="eax";
 	$a="edx";
@@ -330,7 +330,7 @@ sub bn_sqr_words
 
 	if ($sse2) {
 		&picmeup("eax","OPENSSL_ia32cap_P");
-		&bt(&DWP(0,"eax"),26);
+		&bt(&DWP(0,"eax"),"\$IA32CAP_BIT0_SSE2");
 		&jnc(&label("sqr_non_sse2"));
 
 		&mov($r,&wparam(0));
diff --git a/lib/libcrypto/bn/asm/x86-gf2m.pl b/lib/libcrypto/bn/asm/x86-gf2m.pl
index 808a1e59691..97d91362602 100644
--- a/lib/libcrypto/bn/asm/x86-gf2m.pl
+++ b/lib/libcrypto/bn/asm/x86-gf2m.pl
@@ -203,12 +203,12 @@ if (!$x86only) {
 	&picmeup("edx","OPENSSL_ia32cap_P");
 	&mov	("eax",&DWP(0,"edx"));
 	&mov	("edx",&DWP(4,"edx"));
-	&test	("eax",1<<23);		# check MMX bit
+	&test	("eax","\$IA32CAP_MASK0_MMX");	# check MMX bit
 	&jz	(&label("ialu"));
 if ($sse2) {
-	&test	("eax",1<<24);		# check FXSR bit
+	&test	("eax","\$IA32CAP_MASK0_FXSR");	# check FXSR bit
 	&jz	(&label("mmx"));
-	&test	("edx",1<<1);		# check PCLMULQDQ bit
+	&test	("edx","\$IA32CAP_MASK1_PCLMUL");	# check PCLMULQDQ bit
 	&jz	(&label("mmx"));
 
 	&movups		("xmm0",&QWP(8,"esp"));
diff --git a/lib/libcrypto/bn/asm/x86-mont.pl b/lib/libcrypto/bn/asm/x86-mont.pl
index 5cd3cd2ed50..a0bdd5787e5 100755
--- a/lib/libcrypto/bn/asm/x86-mont.pl
+++ b/lib/libcrypto/bn/asm/x86-mont.pl
@@ -114,7 +114,7 @@ $temp="mm6";
 $mask="mm7";
 
 	&picmeup("eax","OPENSSL_ia32cap_P");
-	&bt	(&DWP(0,"eax"),26);
+	&bt	(&DWP(0,"eax"),"\$IA32CAP_BIT0_SSE2");
 	&jnc	(&label("non_sse2"));
 
 	&mov	("eax",-1);
@@ -527,8 +527,10 @@ $sbit=$num;
 	&jle	(&label("sqradd"));
 
 	&mov	($carry,"edx");
-	&lea	("edx",&DWP(0,$sbit,"edx",2));
+	&add	("edx","edx");
 	&shr	($carry,31);
+	&add	("edx",$sbit);
+	&adc	($carry,0);
 &set_label("sqrlast");
 	&mov	($word,$_n0);
 	&mov	($inp,$_np);
diff --git a/lib/libcrypto/bn/asm/x86_64-gf2m.pl b/lib/libcrypto/bn/asm/x86_64-gf2m.pl
index 8e45c7479b3..3ecb425dad2 100644
--- a/lib/libcrypto/bn/asm/x86_64-gf2m.pl
+++ b/lib/libcrypto/bn/asm/x86_64-gf2m.pl
@@ -163,12 +163,13 @@ ___
 
 $code.=<<___;
 .extern	OPENSSL_ia32cap_P
+.hidden	OPENSSL_ia32cap_P
 .globl	bn_GF2m_mul_2x2
 .type	bn_GF2m_mul_2x2,\@abi-omnipotent
 .align	16
 bn_GF2m_mul_2x2:
-	mov	OPENSSL_ia32cap_P(%rip),%rax
-	bt	\$33,%rax
+	mov	OPENSSL_ia32cap_P+4(%rip),%eax
+	bt	\$IA32CAP_BIT1_PCLMUL,%eax
 	jnc	.Lvanilla_mul_2x2
 
 	movd		$a1,%xmm0