Fixup inter-bank movq/movd operations, emit bytes for pclmulqdq again.

Fixes builds gcc + Apple's assembler, working on reenabling builds with older
OpenBSD releases.

based on OpenSSL commit:
 https://git.openssl.org/?p=openssl.git;a=commitdiff;h=902b30df193afc3417a96ba72a81ed390bd50de3

ok miod@

6 files changed, 54 insertions, 26 deletions

diff --git a/src/lib/libcrypto/bn/asm/x86_64-gf2m.pl b/src/lib/libcrypto/bn/asm/x86_64-gf2m.pl
index 226c66c35e..8e45c7479b 100644
--- a/src/lib/libcrypto/bn/asm/x86_64-gf2m.pl
+++ b/src/lib/libcrypto/bn/asm/x86_64-gf2m.pl
@@ -143,12 +143,12 @@ $code.=<<___;
         mov     (%rsp,$i1,8),$t1
         mov     $t1,$t0
         shl     \$`8*$n-4`,$t1
-        movq    $R,$i0
+        movd    $R,$i0
         shr     \$`64-(8*$n-4)`,$t0
         xor     $t1,$lo
         psrldq  \$8,$R
         xor     $t0,$hi
-        movq    $R,$i1
+        movd    $R,$i1
         xor     $i0,$lo
         xor     $i1,$hi
 
@@ -171,15 +171,15 @@ bn_GF2m_mul_2x2:
         bt      \$33,%rax
         jnc     .Lvanilla_mul_2x2
 
-        movq            $a1,%xmm0
-        movq            $b1,%xmm1
-        movq            $a0,%xmm2
+        movd            $a1,%xmm0
+        movd            $b1,%xmm1
+        movd            $a0,%xmm2
 ___
 $code.=<<___ if ($win64);
         movq            40(%rsp),%xmm3
 ___
 $code.=<<___ if (!$win64);
-        movq            $b0,%xmm3
+        movd            $b0,%xmm3
 ___
 $code.=<<___;
         movdqa          %xmm0,%xmm4
diff --git a/src/lib/libcrypto/bn/asm/x86_64-mont5.pl b/src/lib/libcrypto/bn/asm/x86_64-mont5.pl
index 9c88884d42..81e5c53728 100755
--- a/src/lib/libcrypto/bn/asm/x86_64-mont5.pl
+++ b/src/lib/libcrypto/bn/asm/x86_64-mont5.pl
@@ -120,7 +120,7 @@ $code.=<<___;
         lea     $STRIDE($bp),$bp
         por     %xmm3,%xmm0
 
-        movq    %xmm0,$m0               # m0=bp[0]
+        movd    %xmm0,$m0               # m0=bp[0]
 
         mov     ($n0),$n0               # pull n0[0] value
         mov     ($ap),%rax
@@ -183,7 +183,7 @@ $code.=<<___;
         cmp     $num,$j
         jl      .L1st
 
-        movq    %xmm0,$m0               # bp[1]
+        movd    %xmm0,$m0               # bp[1]
 
         add     %rax,$hi1
         mov     ($ap),%rax              # ap[0]
@@ -266,7 +266,7 @@ $code.=<<___;
         cmp     $num,$j
         jl      .Linner
 
-        movq    %xmm0,$m0               # bp[i+1]
+        movd    %xmm0,$m0               # bp[i+1]
 
         add     %rax,$hi1
         mov     ($ap),%rax              # ap[0]
@@ -403,7 +403,7 @@ $code.=<<___;
         lea     $STRIDE($bp),$bp
         por     %xmm3,%xmm0
 
-        movq    %xmm0,$m0               # m0=bp[0]
+        movd    %xmm0,$m0               # m0=bp[0]
         mov     ($n0),$n0               # pull n0[0] value
         mov     ($ap),%rax
 
@@ -550,7 +550,7 @@ $code.=<<___;
         mov     $N[1],-16(%rsp,$j,8)    # tp[j-1]
         mov     %rdx,$N[0]
 
-        movq    %xmm0,$m0               # bp[1]
+        movd    %xmm0,$m0               # bp[1]
 
         xor     $N[1],$N[1]
         add     $A[0],$N[0]
@@ -718,7 +718,7 @@ $code.=<<___;
         mov     $N[0],-24(%rsp,$j,8)    # tp[j-1]
         mov     %rdx,$N[0]
 
-        movq    %xmm0,$m0               # bp[i+1]
+        movd    %xmm0,$m0               # bp[i+1]
         mov     $N[1],-16(%rsp,$j,8)    # tp[j-1]
 
         xor     $N[1],$N[1]
diff --git a/src/lib/libcrypto/perlasm/x86_64-xlate.pl b/src/lib/libcrypto/perlasm/x86_64-xlate.pl
index ed1f3ed6ab..82992f41e5 100755
--- a/src/lib/libcrypto/perlasm/x86_64-xlate.pl
+++ b/src/lib/libcrypto/perlasm/x86_64-xlate.pl
@@ -121,7 +121,7 @@ my %globals;
                 $self->{sz} = "";
             } elsif ($self->{op} =~ /^v/) { # VEX
                 $self->{sz} = "";
-            } elsif ($self->{op} =~ /movq/ && $line =~ /%xmm/) {
+            } elsif ($self->{op} =~ /mov[dq]/ && $line =~ /%xmm/) {
                 $self->{sz} = "";
             } elsif ($self->{op} =~ /([a-z]{3,})([qlwb])$/) {
                 $self->{op} = $1;
@@ -698,6 +698,20 @@ my $pinsrd = sub {
     }
 };
 
+my $pclmulqdq = sub {
+        if (shift =~ /\$([x0-9a-f]+),\s*%xmm([0-9]+),\s*%xmm([0-9]+)/) {
+                my @opcode=(0x66);
+                rex(\@opcode,$3,$2);
+                push @opcode,0x0f,0x3a,0x44;
+                push @opcode,0xc0|($2&7)|(($3&7)<<3);           # ModR/M
+                my $c=$1;
+                push @opcode,$c=~/^0/?oct($c):$c;
+                @opcode;
+        } else {
+                ();
+        }
+};
+
 if ($nasm) {
     print <<___;
 default rel
diff --git a/src/lib/libssl/src/crypto/bn/asm/x86_64-gf2m.pl b/src/lib/libssl/src/crypto/bn/asm/x86_64-gf2m.pl
index 226c66c35e..8e45c7479b 100644
--- a/src/lib/libssl/src/crypto/bn/asm/x86_64-gf2m.pl
+++ b/src/lib/libssl/src/crypto/bn/asm/x86_64-gf2m.pl
@@ -143,12 +143,12 @@ $code.=<<___;
         mov     (%rsp,$i1,8),$t1
         mov     $t1,$t0
         shl     \$`8*$n-4`,$t1
-        movq    $R,$i0
+        movd    $R,$i0
         shr     \$`64-(8*$n-4)`,$t0
         xor     $t1,$lo
         psrldq  \$8,$R
         xor     $t0,$hi
-        movq    $R,$i1
+        movd    $R,$i1
         xor     $i0,$lo
         xor     $i1,$hi
 
@@ -171,15 +171,15 @@ bn_GF2m_mul_2x2:
         bt      \$33,%rax
         jnc     .Lvanilla_mul_2x2
 
-        movq            $a1,%xmm0
-        movq            $b1,%xmm1
-        movq            $a0,%xmm2
+        movd            $a1,%xmm0
+        movd            $b1,%xmm1
+        movd            $a0,%xmm2
 ___
 $code.=<<___ if ($win64);
         movq            40(%rsp),%xmm3
 ___
 $code.=<<___ if (!$win64);
-        movq            $b0,%xmm3
+        movd            $b0,%xmm3
 ___
 $code.=<<___;
         movdqa          %xmm0,%xmm4
diff --git a/src/lib/libssl/src/crypto/bn/asm/x86_64-mont5.pl b/src/lib/libssl/src/crypto/bn/asm/x86_64-mont5.pl
index 9c88884d42..81e5c53728 100755
--- a/src/lib/libssl/src/crypto/bn/asm/x86_64-mont5.pl
+++ b/src/lib/libssl/src/crypto/bn/asm/x86_64-mont5.pl
@@ -120,7 +120,7 @@ $code.=<<___;
         lea     $STRIDE($bp),$bp
         por     %xmm3,%xmm0
 
-        movq    %xmm0,$m0               # m0=bp[0]
+        movd    %xmm0,$m0               # m0=bp[0]
 
         mov     ($n0),$n0               # pull n0[0] value
         mov     ($ap),%rax
@@ -183,7 +183,7 @@ $code.=<<___;
         cmp     $num,$j
         jl      .L1st
 
-        movq    %xmm0,$m0               # bp[1]
+        movd    %xmm0,$m0               # bp[1]
 
         add     %rax,$hi1
         mov     ($ap),%rax              # ap[0]
@@ -266,7 +266,7 @@ $code.=<<___;
         cmp     $num,$j
         jl      .Linner
 
-        movq    %xmm0,$m0               # bp[i+1]
+        movd    %xmm0,$m0               # bp[i+1]
 
         add     %rax,$hi1
         mov     ($ap),%rax              # ap[0]
@@ -403,7 +403,7 @@ $code.=<<___;
         lea     $STRIDE($bp),$bp
         por     %xmm3,%xmm0
 
-        movq    %xmm0,$m0               # m0=bp[0]
+        movd    %xmm0,$m0               # m0=bp[0]
         mov     ($n0),$n0               # pull n0[0] value
         mov     ($ap),%rax
 
@@ -550,7 +550,7 @@ $code.=<<___;
         mov     $N[1],-16(%rsp,$j,8)    # tp[j-1]
         mov     %rdx,$N[0]
 
-        movq    %xmm0,$m0               # bp[1]
+        movd    %xmm0,$m0               # bp[1]
 
         xor     $N[1],$N[1]
         add     $A[0],$N[0]
@@ -718,7 +718,7 @@ $code.=<<___;
         mov     $N[0],-24(%rsp,$j,8)    # tp[j-1]
         mov     %rdx,$N[0]
 
-        movq    %xmm0,$m0               # bp[i+1]
+        movd    %xmm0,$m0               # bp[i+1]
         mov     $N[1],-16(%rsp,$j,8)    # tp[j-1]
 
         xor     $N[1],$N[1]
diff --git a/src/lib/libssl/src/crypto/perlasm/x86_64-xlate.pl b/src/lib/libssl/src/crypto/perlasm/x86_64-xlate.pl
index ed1f3ed6ab..82992f41e5 100755
--- a/src/lib/libssl/src/crypto/perlasm/x86_64-xlate.pl
+++ b/src/lib/libssl/src/crypto/perlasm/x86_64-xlate.pl
@@ -121,7 +121,7 @@ my %globals;
                 $self->{sz} = "";
             } elsif ($self->{op} =~ /^v/) { # VEX
                 $self->{sz} = "";
-            } elsif ($self->{op} =~ /movq/ && $line =~ /%xmm/) {
+            } elsif ($self->{op} =~ /mov[dq]/ && $line =~ /%xmm/) {
                 $self->{sz} = "";
             } elsif ($self->{op} =~ /([a-z]{3,})([qlwb])$/) {
                 $self->{op} = $1;
@@ -698,6 +698,20 @@ my $pinsrd = sub {
     }
 };
 
+my $pclmulqdq = sub {
+        if (shift =~ /\$([x0-9a-f]+),\s*%xmm([0-9]+),\s*%xmm([0-9]+)/) {
+                my @opcode=(0x66);
+                rex(\@opcode,$3,$2);
+                push @opcode,0x0f,0x3a,0x44;
+                push @opcode,0xc0|($2&7)|(($3&7)<<3);           # ModR/M
+                my $c=$1;
+                push @opcode,$c=~/^0/?oct($c):$c;
+                @opcode;
+        } else {
+                ();
+        }
+};
+
 if ($nasm) {
     print <<___;
 default rel