X-Git-Url: https://code.wpia.club/?a=blobdiff_plain;f=lib%2Fopenssl%2Fcrypto%2Fsha%2Fasm%2Fsha512-ppc.pl;fp=lib%2Fopenssl%2Fcrypto%2Fsha%2Fasm%2Fsha512-ppc.pl;h=fe95b0150907f3309350c12d57d0efab536eda80;hb=02ed66432c92de70694700164f986190aad3cbc5;hp=6b44a68e599e5fcc59876e9be9af73cb1f54094d;hpb=89016837dcbf2775cd15dc8cbaba00dc6379f86e;p=cassiopeia.git

diff --git a/lib/openssl/crypto/sha/asm/sha512-ppc.pl b/lib/openssl/crypto/sha/asm/sha512-ppc.pl
index 6b44a68..fe95b01 100755
--- a/lib/openssl/crypto/sha/asm/sha512-ppc.pl
+++ b/lib/openssl/crypto/sha/asm/sha512-ppc.pl
@@ -1,7 +1,14 @@
-#!/usr/bin/env perl
+#! /usr/bin/env perl
+# Copyright 2006-2016 The OpenSSL Project Authors. All Rights Reserved.
+#
+# Licensed under the OpenSSL license (the "License").  You may not use
+# this file except in compliance with the License.  You can obtain a copy
+# in the file LICENSE in the source distribution or at
+# https://www.openssl.org/source/license.html
+
 
 # ====================================================================
-# Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
+# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL
 # project. The module is, however, dual licensed under OpenSSL and
 # CRYPTOGAMS licenses depending on where you obtain it. For further
 # details see http://www.openssl.org/~appro/cryptogams/.
@@ -9,8 +16,7 @@
 
 # I let hardware handle unaligned input, except on page boundaries
 # (see below for details). Otherwise straightforward implementation
-# with X vector in register bank. The module is big-endian [which is
-# not big deal as there're no little-endian targets left around].
+# with X vector in register bank.
 
 #			sha256		|	sha512
 # 			-m64	-m32	|	-m64	-m32
@@ -56,6 +62,8 @@ if ($flavour =~ /64/) {
 	$PUSH="stw";
 } else { die "nonsense $flavour"; }
 
+$LITTLE_ENDIAN = ($flavour=~/le$/) ? $SIZE_T : 0;
+
 $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
 ( $xlate="${dir}ppc-xlate.pl" and -f $xlate ) or
 ( $xlate="${dir}../../perlasm/ppc-xlate.pl" and -f $xlate) or
@@ -64,7 +72,7 @@ die "can't locate ppc-xlate.pl";
 open STDOUT,"| $^X $xlate $flavour $output" || die "can't call $xlate: $!";
 
 if ($output =~ /512/) {
-	$func="sha512_block_data_order";
+	$func="sha512_block_ppc";
 	$SZ=8;
 	@Sigma0=(28,34,39);
 	@Sigma1=(14,18,41);
@@ -76,7 +84,7 @@ if ($output =~ /512/) {
 	$ROR="rotrdi";
 	$SHR="srdi";
 } else {
-	$func="sha256_block_data_order";
+	$func="sha256_block_ppc";
 	$SZ=4;
 	@Sigma0=( 2,13,22);
 	@Sigma1=( 6,11,25);
@@ -110,7 +118,7 @@ $B  ="r9";
 $C  ="r10";
 $D  ="r11";
 $E  ="r12";
-$F  ="r13";	$F="r2" if ($SIZE_T==8);# reassigned to exempt TLS pointer
+$F  =$t1;	$t1 = "r0";	# stay away from "r13";
 $G  ="r14";
 $H  ="r15";
 
@@ -118,24 +126,23 @@ $H  ="r15";
 @X=("r16","r17","r18","r19","r20","r21","r22","r23",
     "r24","r25","r26","r27","r28","r29","r30","r31");
 
-$inp="r31";	# reassigned $inp! aliases with @X[15]
+$inp="r31" if($SZ==4 || $SIZE_T==8);	# reassigned $inp! aliases with @X[15]
 
 sub ROUND_00_15 {
 my ($i,$a,$b,$c,$d,$e,$f,$g,$h)=@_;
 $code.=<<___;
-	$LD	$T,`$i*$SZ`($Tbl)
 	$ROR	$a0,$e,$Sigma1[0]
 	$ROR	$a1,$e,$Sigma1[1]
 	and	$t0,$f,$e
-	andc	$t1,$g,$e
-	add	$T,$T,$h
 	xor	$a0,$a0,$a1
+	add	$h,$h,$t1
+	andc	$t1,$g,$e
 	$ROR	$a1,$a1,`$Sigma1[2]-$Sigma1[1]`
 	or	$t0,$t0,$t1		; Ch(e,f,g)
-	add	$T,$T,@X[$i]
+	add	$h,$h,@X[$i%16]
 	xor	$a0,$a0,$a1		; Sigma1(e)
-	add	$T,$T,$t0
-	add	$T,$T,$a0
+	add	$h,$h,$t0
+	add	$h,$h,$a0
 
 	$ROR	$a0,$a,$Sigma0[0]
 	$ROR	$a1,$a,$Sigma0[1]
@@ -146,9 +153,14 @@ $code.=<<___;
 	xor	$t0,$t0,$t1
 	and	$t1,$b,$c
 	xor	$a0,$a0,$a1		; Sigma0(a)
-	add	$d,$d,$T
+	add	$d,$d,$h
 	xor	$t0,$t0,$t1		; Maj(a,b,c)
-	add	$h,$T,$a0
+___
+$code.=<<___ if ($i<15);
+	$LD	$t1,`($i+1)*$SZ`($Tbl)
+___
+$code.=<<___;
+	add	$h,$h,$a0
 	add	$h,$h,$t0
 
 ___
@@ -169,10 +181,11 @@ $code.=<<___;
 	add	@X[$i],@X[$i],@X[($i+9)%16]
 	xor	$a0,$a0,$a1		; sigma0(X[(i+1)&0x0f])
 	xor	$t0,$t0,$t1		; sigma1(X[(i+14)&0x0f])
+	$LD	$t1,`$i*$SZ`($Tbl)
 	add	@X[$i],@X[$i],$a0
 	add	@X[$i],@X[$i],$t0
 ___
-&ROUND_00_15($i,$a,$b,$c,$d,$e,$f,$g,$h);
+&ROUND_00_15($i+16,$a,$b,$c,$d,$e,$f,$g,$h);
 }
 
 $code=<<___;
@@ -188,8 +201,6 @@ $func:
 
 	$PUSH	$ctx,`$FRAME-$SIZE_T*22`($sp)
 
-	$PUSH	$toc,`$FRAME-$SIZE_T*20`($sp)
-	$PUSH	r13,`$FRAME-$SIZE_T*19`($sp)
 	$PUSH	r14,`$FRAME-$SIZE_T*18`($sp)
 	$PUSH	r15,`$FRAME-$SIZE_T*17`($sp)
 	$PUSH	r16,`$FRAME-$SIZE_T*16`($sp)
@@ -209,7 +220,10 @@ $func:
 	$PUSH	r30,`$FRAME-$SIZE_T*2`($sp)
 	$PUSH	r31,`$FRAME-$SIZE_T*1`($sp)
 	$PUSH	r0,`$FRAME+$LRSAVE`($sp)
+___
 
+if ($SZ==4 || $SIZE_T==8) {
+$code.=<<___;
 	$LD	$A,`0*$SZ`($ctx)
 	mr	$inp,r4				; incarnate $inp
 	$LD	$B,`1*$SZ`($ctx)
@@ -219,7 +233,16 @@ $func:
 	$LD	$F,`5*$SZ`($ctx)
 	$LD	$G,`6*$SZ`($ctx)
 	$LD	$H,`7*$SZ`($ctx)
+___
+} else {
+  for ($i=16;$i<32;$i++) {
+    $code.=<<___;
+	lwz	r$i,`$LITTLE_ENDIAN^(4*($i-16))`($ctx)
+___
+  }
+}
 
+$code.=<<___;
 	bl	LPICmeup
 LPICedup:
 	andi.	r0,$inp,3
@@ -243,7 +266,7 @@ Lunaligned:
 	andi.	$t1,$t1,`4096-16*$SZ`	; distance to closest page boundary
 	beq	Lcross_page
 	$UCMP	$num,$t1
-	ble-	Laligned		; didn't cross the page boundary
+	ble	Laligned		; didn't cross the page boundary
 	subfc	$num,$t1,$num
 	add	$t1,$inp,$t1
 	$PUSH	$num,`$FRAME-$SIZE_T*25`($sp)	; save real remaining num
@@ -255,6 +278,9 @@ Lunaligned:
 Lcross_page:
 	li	$t1,`16*$SZ/4`
 	mtctr	$t1
+___
+if ($SZ==4 || $SIZE_T==8) {
+$code.=<<___;
 	addi	r20,$sp,$LOCALS			; aligned spot below the frame
 Lmemcpy:
 	lbz	r16,0($inp)
@@ -268,7 +294,26 @@ Lmemcpy:
 	stb	r19,3(r20)
 	addi	r20,r20,4
 	bdnz	Lmemcpy
+___
+} else {
+$code.=<<___;
+	addi	r12,$sp,$LOCALS			; aligned spot below the frame
+Lmemcpy:
+	lbz	r8,0($inp)
+	lbz	r9,1($inp)
+	lbz	r10,2($inp)
+	lbz	r11,3($inp)
+	addi	$inp,$inp,4
+	stb	r8,0(r12)
+	stb	r9,1(r12)
+	stb	r10,2(r12)
+	stb	r11,3(r12)
+	addi	r12,r12,4
+	bdnz	Lmemcpy
+___
+}
 
+$code.=<<___;
 	$PUSH	$inp,`$FRAME-$SIZE_T*26`($sp)	; save real inp
 	addi	$t1,$sp,`$LOCALS+16*$SZ`	; fictitious end pointer
 	addi	$inp,$sp,$LOCALS		; fictitious inp pointer
@@ -279,12 +324,10 @@ Lmemcpy:
 	$POP	$inp,`$FRAME-$SIZE_T*26`($sp)	; restore real inp
 	$POP	$num,`$FRAME-$SIZE_T*25`($sp)	; restore real num
 	addic.	$num,$num,`-16*$SZ`		; num--
-	bne-	Lunaligned
+	bne	Lunaligned
 
 Ldone:
 	$POP	r0,`$FRAME+$LRSAVE`($sp)
-	$POP	$toc,`$FRAME-$SIZE_T*20`($sp)
-	$POP	r13,`$FRAME-$SIZE_T*19`($sp)
 	$POP	r14,`$FRAME-$SIZE_T*18`($sp)
 	$POP	r15,`$FRAME-$SIZE_T*17`($sp)
 	$POP	r16,`$FRAME-$SIZE_T*16`($sp)
@@ -309,27 +352,48 @@ Ldone:
 	.long	0
 	.byte	0,12,4,1,0x80,18,3,0
 	.long	0
+___
 
+if ($SZ==4 || $SIZE_T==8) {
+$code.=<<___;
 .align	4
 Lsha2_block_private:
+	$LD	$t1,0($Tbl)
 ___
 for($i=0;$i<16;$i++) {
-$code.=<<___ if ($SZ==4);
+$code.=<<___ if ($SZ==4 && !$LITTLE_ENDIAN);
 	lwz	@X[$i],`$i*$SZ`($inp)
 ___
+$code.=<<___ if ($SZ==4 && $LITTLE_ENDIAN);
+	lwz	$a0,`$i*$SZ`($inp)
+	rotlwi	@X[$i],$a0,8
+	rlwimi	@X[$i],$a0,24,0,7
+	rlwimi	@X[$i],$a0,24,16,23
+___
 # 64-bit loads are split to 2x32-bit ones, as CPU can't handle
 # unaligned 64-bit loads, only 32-bit ones...
-$code.=<<___ if ($SZ==8);
+$code.=<<___ if ($SZ==8 && !$LITTLE_ENDIAN);
 	lwz	$t0,`$i*$SZ`($inp)
 	lwz	@X[$i],`$i*$SZ+4`($inp)
 	insrdi	@X[$i],$t0,32,0
+___
+$code.=<<___ if ($SZ==8 && $LITTLE_ENDIAN);
+	lwz	$a0,`$i*$SZ`($inp)
+	 lwz	$a1,`$i*$SZ+4`($inp)
+	rotlwi	$t0,$a0,8
+	 rotlwi	@X[$i],$a1,8
+	rlwimi	$t0,$a0,24,0,7
+	 rlwimi	@X[$i],$a1,24,0,7
+	rlwimi	$t0,$a0,24,16,23
+	 rlwimi	@X[$i],$a1,24,16,23
+	insrdi	@X[$i],$t0,32,0
 ___
 	&ROUND_00_15($i,@V);
 	unshift(@V,pop(@V));
 }
 $code.=<<___;
-	li	$T,`$rounds/16-1`
-	mtctr	$T
+	li	$t0,`$rounds/16-1`
+	mtctr	$t0
 .align	4
 Lrounds:
 	addi	$Tbl,$Tbl,`16*$SZ`
@@ -339,7 +403,7 @@ for(;$i<32;$i++) {
 	unshift(@V,pop(@V));
 }
 $code.=<<___;
-	bdnz-	Lrounds
+	bdnz	Lrounds
 
 	$POP	$ctx,`$FRAME-$SIZE_T*22`($sp)
 	$POP	$inp,`$FRAME-$SIZE_T*23`($sp)	; inp pointer
@@ -377,7 +441,282 @@ $code.=<<___;
 	blr
 	.long	0
 	.byte	0,12,0x14,0,0,0,0,0
+.size	$func,.-$func
+___
+} else {
+########################################################################
+# SHA512 for PPC32, X vector is off-loaded to stack...
+#
+#			|	sha512
+#			|	-m32
+# ----------------------+-----------------------
+# PPC74x0,gcc-4.0.1	|	+48%
+# POWER6,gcc-4.4.6	|	+124%(*)
+# POWER7,gcc-4.4.6	|	+79%(*)
+# e300,gcc-4.1.0	|	+167%
+#
+# (*)	~1/3 of -m64 result [and ~20% better than -m32 code generated
+#	by xlc-12.1]
+
+my $XOFF=$LOCALS;
+
+my @V=map("r$_",(16..31));	# A..H
+
+my ($s0,$s1,$t0,$t1,$t2,$t3,$a0,$a1,$a2,$a3)=map("r$_",(0,5,6,8..12,14,15));
+my ($x0,$x1)=("r3","r4");	# zaps $ctx and $inp
+
+sub ROUND_00_15_ppc32 {
+my ($i,	$ahi,$alo,$bhi,$blo,$chi,$clo,$dhi,$dlo,
+	$ehi,$elo,$fhi,$flo,$ghi,$glo,$hhi,$hlo)=@_;
+
+$code.=<<___;
+	lwz	$t2,`$SZ*($i%16)+($LITTLE_ENDIAN^4)`($Tbl)
+	 xor	$a0,$flo,$glo
+	lwz	$t3,`$SZ*($i%16)+($LITTLE_ENDIAN^0)`($Tbl)
+	 xor	$a1,$fhi,$ghi
+	addc	$hlo,$hlo,$t0			; h+=x[i]
+	stw	$t0,`$XOFF+0+$SZ*($i%16)`($sp)	; save x[i]
+
+	srwi	$s0,$elo,$Sigma1[0]
+	srwi	$s1,$ehi,$Sigma1[0]
+	 and	$a0,$a0,$elo
+	adde	$hhi,$hhi,$t1
+	 and	$a1,$a1,$ehi
+	stw	$t1,`$XOFF+4+$SZ*($i%16)`($sp)
+	srwi	$t0,$elo,$Sigma1[1]
+	srwi	$t1,$ehi,$Sigma1[1]
+	 addc	$hlo,$hlo,$t2			; h+=K512[i]
+	insrwi	$s0,$ehi,$Sigma1[0],0
+	insrwi	$s1,$elo,$Sigma1[0],0
+	 xor	$a0,$a0,$glo			; Ch(e,f,g)
+	 adde	$hhi,$hhi,$t3
+	 xor	$a1,$a1,$ghi
+	insrwi	$t0,$ehi,$Sigma1[1],0
+	insrwi	$t1,$elo,$Sigma1[1],0
+	 addc	$hlo,$hlo,$a0			; h+=Ch(e,f,g)
+	srwi	$t2,$ehi,$Sigma1[2]-32
+	srwi	$t3,$elo,$Sigma1[2]-32
+	xor	$s0,$s0,$t0
+	xor	$s1,$s1,$t1
+	insrwi	$t2,$elo,$Sigma1[2]-32,0
+	insrwi	$t3,$ehi,$Sigma1[2]-32,0
+	 xor	$a0,$alo,$blo			; a^b, b^c in next round
+	 adde	$hhi,$hhi,$a1
+	 xor	$a1,$ahi,$bhi
+	xor	$s0,$s0,$t2			; Sigma1(e)
+	xor	$s1,$s1,$t3
+
+	srwi	$t0,$alo,$Sigma0[0]
+	 and	$a2,$a2,$a0
+	 addc	$hlo,$hlo,$s0			; h+=Sigma1(e)
+	 and	$a3,$a3,$a1
+	srwi	$t1,$ahi,$Sigma0[0]
+	srwi	$s0,$ahi,$Sigma0[1]-32
+	 adde	$hhi,$hhi,$s1
+	srwi	$s1,$alo,$Sigma0[1]-32
+	insrwi	$t0,$ahi,$Sigma0[0],0
+	insrwi	$t1,$alo,$Sigma0[0],0
+	 xor	$a2,$a2,$blo			; Maj(a,b,c)
+	 addc	$dlo,$dlo,$hlo			; d+=h
+	 xor	$a3,$a3,$bhi
+	insrwi	$s0,$alo,$Sigma0[1]-32,0
+	insrwi	$s1,$ahi,$Sigma0[1]-32,0
+	 adde	$dhi,$dhi,$hhi
+	srwi	$t2,$ahi,$Sigma0[2]-32
+	srwi	$t3,$alo,$Sigma0[2]-32
+	xor	$s0,$s0,$t0
+	 addc	$hlo,$hlo,$a2			; h+=Maj(a,b,c)
+	xor	$s1,$s1,$t1
+	insrwi	$t2,$alo,$Sigma0[2]-32,0
+	insrwi	$t3,$ahi,$Sigma0[2]-32,0
+	 adde	$hhi,$hhi,$a3
 ___
+$code.=<<___ if ($i>=15);
+	lwz	$t0,`$XOFF+0+$SZ*(($i+2)%16)`($sp)
+	lwz	$t1,`$XOFF+4+$SZ*(($i+2)%16)`($sp)
+___
+$code.=<<___ if ($i<15 && !$LITTLE_ENDIAN);
+	lwz	$t1,`$SZ*($i+1)+0`($inp)
+	lwz	$t0,`$SZ*($i+1)+4`($inp)
+___
+$code.=<<___ if ($i<15 && $LITTLE_ENDIAN);
+	lwz	$a2,`$SZ*($i+1)+0`($inp)
+	 lwz	$a3,`$SZ*($i+1)+4`($inp)
+	rotlwi	$t1,$a2,8
+	 rotlwi	$t0,$a3,8
+	rlwimi	$t1,$a2,24,0,7
+	 rlwimi	$t0,$a3,24,0,7
+	rlwimi	$t1,$a2,24,16,23
+	 rlwimi	$t0,$a3,24,16,23
+___
+$code.=<<___;
+	xor	$s0,$s0,$t2			; Sigma0(a)
+	xor	$s1,$s1,$t3
+	addc	$hlo,$hlo,$s0			; h+=Sigma0(a)
+	adde	$hhi,$hhi,$s1
+___
+$code.=<<___ if ($i==15);
+	lwz	$x0,`$XOFF+0+$SZ*(($i+1)%16)`($sp)
+	lwz	$x1,`$XOFF+4+$SZ*(($i+1)%16)`($sp)
+___
+}
+sub ROUND_16_xx_ppc32 {
+my ($i,	$ahi,$alo,$bhi,$blo,$chi,$clo,$dhi,$dlo,
+	$ehi,$elo,$fhi,$flo,$ghi,$glo,$hhi,$hlo)=@_;
+
+$code.=<<___;
+	srwi	$s0,$t0,$sigma0[0]
+	srwi	$s1,$t1,$sigma0[0]
+	srwi	$t2,$t0,$sigma0[1]
+	srwi	$t3,$t1,$sigma0[1]
+	insrwi	$s0,$t1,$sigma0[0],0
+	insrwi	$s1,$t0,$sigma0[0],0
+	srwi	$a0,$t0,$sigma0[2]
+	insrwi	$t2,$t1,$sigma0[1],0
+	insrwi	$t3,$t0,$sigma0[1],0
+	insrwi	$a0,$t1,$sigma0[2],0
+	xor	$s0,$s0,$t2
+	 lwz	$t2,`$XOFF+0+$SZ*(($i+14)%16)`($sp)
+	srwi	$a1,$t1,$sigma0[2]
+	xor	$s1,$s1,$t3
+	 lwz	$t3,`$XOFF+4+$SZ*(($i+14)%16)`($sp)
+	xor	$a0,$a0,$s0
+	 srwi	$s0,$t2,$sigma1[0]
+	xor	$a1,$a1,$s1
+	 srwi	$s1,$t3,$sigma1[0]
+	addc	$x0,$x0,$a0			; x[i]+=sigma0(x[i+1])
+	 srwi	$a0,$t3,$sigma1[1]-32
+	insrwi	$s0,$t3,$sigma1[0],0
+	insrwi	$s1,$t2,$sigma1[0],0
+	adde	$x1,$x1,$a1
+	 srwi	$a1,$t2,$sigma1[1]-32
+
+	insrwi	$a0,$t2,$sigma1[1]-32,0
+	srwi	$t2,$t2,$sigma1[2]
+	insrwi	$a1,$t3,$sigma1[1]-32,0
+	insrwi	$t2,$t3,$sigma1[2],0
+	xor	$s0,$s0,$a0
+	 lwz	$a0,`$XOFF+0+$SZ*(($i+9)%16)`($sp)
+	srwi	$t3,$t3,$sigma1[2]
+	xor	$s1,$s1,$a1
+	 lwz	$a1,`$XOFF+4+$SZ*(($i+9)%16)`($sp)
+	xor	$s0,$s0,$t2
+	 addc	$x0,$x0,$a0			; x[i]+=x[i+9]
+	xor	$s1,$s1,$t3
+	 adde	$x1,$x1,$a1
+	addc	$x0,$x0,$s0			; x[i]+=sigma1(x[i+14])
+	adde	$x1,$x1,$s1
+___
+	($t0,$t1,$x0,$x1) = ($x0,$x1,$t0,$t1);
+	&ROUND_00_15_ppc32(@_);
+}
+
+$code.=<<___;
+.align	4
+Lsha2_block_private:
+___
+$code.=<<___ if (!$LITTLE_ENDIAN);
+	lwz	$t1,0($inp)
+	xor	$a2,@V[3],@V[5]		; B^C, magic seed
+	lwz	$t0,4($inp)
+	xor	$a3,@V[2],@V[4]
+___
+$code.=<<___ if ($LITTLE_ENDIAN);
+	lwz	$a1,0($inp)
+	xor	$a2,@V[3],@V[5]		; B^C, magic seed
+	lwz	$a0,4($inp)
+	xor	$a3,@V[2],@V[4]
+	rotlwi	$t1,$a1,8
+	 rotlwi	$t0,$a0,8
+	rlwimi	$t1,$a1,24,0,7
+	 rlwimi	$t0,$a0,24,0,7
+	rlwimi	$t1,$a1,24,16,23
+	 rlwimi	$t0,$a0,24,16,23
+___
+for($i=0;$i<16;$i++) {
+	&ROUND_00_15_ppc32($i,@V);
+	unshift(@V,pop(@V));	unshift(@V,pop(@V));
+	($a0,$a1,$a2,$a3) = ($a2,$a3,$a0,$a1);
+}
+$code.=<<___;
+	li	$a0,`$rounds/16-1`
+	mtctr	$a0
+.align	4
+Lrounds:
+	addi	$Tbl,$Tbl,`16*$SZ`
+___
+for(;$i<32;$i++) {
+	&ROUND_16_xx_ppc32($i,@V);
+	unshift(@V,pop(@V));	unshift(@V,pop(@V));
+	($a0,$a1,$a2,$a3) = ($a2,$a3,$a0,$a1);
+}
+$code.=<<___;
+	bdnz	Lrounds
+
+	$POP	$ctx,`$FRAME-$SIZE_T*22`($sp)
+	$POP	$inp,`$FRAME-$SIZE_T*23`($sp)	; inp pointer
+	$POP	$num,`$FRAME-$SIZE_T*24`($sp)	; end pointer
+	subi	$Tbl,$Tbl,`($rounds-16)*$SZ`	; rewind Tbl
+
+	lwz	$t0,`$LITTLE_ENDIAN^0`($ctx)
+	lwz	$t1,`$LITTLE_ENDIAN^4`($ctx)
+	lwz	$t2,`$LITTLE_ENDIAN^8`($ctx)
+	lwz	$t3,`$LITTLE_ENDIAN^12`($ctx)
+	lwz	$a0,`$LITTLE_ENDIAN^16`($ctx)
+	lwz	$a1,`$LITTLE_ENDIAN^20`($ctx)
+	lwz	$a2,`$LITTLE_ENDIAN^24`($ctx)
+	addc	@V[1],@V[1],$t1
+	lwz	$a3,`$LITTLE_ENDIAN^28`($ctx)
+	adde	@V[0],@V[0],$t0
+	lwz	$t0,`$LITTLE_ENDIAN^32`($ctx)
+	addc	@V[3],@V[3],$t3
+	lwz	$t1,`$LITTLE_ENDIAN^36`($ctx)
+	adde	@V[2],@V[2],$t2
+	lwz	$t2,`$LITTLE_ENDIAN^40`($ctx)
+	addc	@V[5],@V[5],$a1
+	lwz	$t3,`$LITTLE_ENDIAN^44`($ctx)
+	adde	@V[4],@V[4],$a0
+	lwz	$a0,`$LITTLE_ENDIAN^48`($ctx)
+	addc	@V[7],@V[7],$a3
+	lwz	$a1,`$LITTLE_ENDIAN^52`($ctx)
+	adde	@V[6],@V[6],$a2
+	lwz	$a2,`$LITTLE_ENDIAN^56`($ctx)
+	addc	@V[9],@V[9],$t1
+	lwz	$a3,`$LITTLE_ENDIAN^60`($ctx)
+	adde	@V[8],@V[8],$t0
+	stw	@V[0],`$LITTLE_ENDIAN^0`($ctx)
+	stw	@V[1],`$LITTLE_ENDIAN^4`($ctx)
+	addc	@V[11],@V[11],$t3
+	stw	@V[2],`$LITTLE_ENDIAN^8`($ctx)
+	stw	@V[3],`$LITTLE_ENDIAN^12`($ctx)
+	adde	@V[10],@V[10],$t2
+	stw	@V[4],`$LITTLE_ENDIAN^16`($ctx)
+	stw	@V[5],`$LITTLE_ENDIAN^20`($ctx)
+	addc	@V[13],@V[13],$a1
+	stw	@V[6],`$LITTLE_ENDIAN^24`($ctx)
+	stw	@V[7],`$LITTLE_ENDIAN^28`($ctx)
+	adde	@V[12],@V[12],$a0
+	stw	@V[8],`$LITTLE_ENDIAN^32`($ctx)
+	stw	@V[9],`$LITTLE_ENDIAN^36`($ctx)
+	addc	@V[15],@V[15],$a3
+	stw	@V[10],`$LITTLE_ENDIAN^40`($ctx)
+	stw	@V[11],`$LITTLE_ENDIAN^44`($ctx)
+	adde	@V[14],@V[14],$a2
+	stw	@V[12],`$LITTLE_ENDIAN^48`($ctx)
+	stw	@V[13],`$LITTLE_ENDIAN^52`($ctx)
+	stw	@V[14],`$LITTLE_ENDIAN^56`($ctx)
+	stw	@V[15],`$LITTLE_ENDIAN^60`($ctx)
+
+	addi	$inp,$inp,`16*$SZ`		; advance inp
+	$PUSH	$inp,`$FRAME-$SIZE_T*23`($sp)
+	$UCMP	$inp,$num
+	bne	Lsha2_block_private
+	blr
+	.long	0
+	.byte	0,12,0x14,0,0,0,0,0
+.size	$func,.-$func
+___
+}
 
 # Ugly hack here, because PPC assembler syntax seem to vary too
 # much from platforms to platform...
@@ -395,46 +734,46 @@ LPICmeup:
 	.space	`64-9*4`
 ___
 $code.=<<___ if ($SZ==8);
-	.long	0x428a2f98,0xd728ae22,0x71374491,0x23ef65cd
-	.long	0xb5c0fbcf,0xec4d3b2f,0xe9b5dba5,0x8189dbbc
-	.long	0x3956c25b,0xf348b538,0x59f111f1,0xb605d019
-	.long	0x923f82a4,0xaf194f9b,0xab1c5ed5,0xda6d8118
-	.long	0xd807aa98,0xa3030242,0x12835b01,0x45706fbe
-	.long	0x243185be,0x4ee4b28c,0x550c7dc3,0xd5ffb4e2
-	.long	0x72be5d74,0xf27b896f,0x80deb1fe,0x3b1696b1
-	.long	0x9bdc06a7,0x25c71235,0xc19bf174,0xcf692694
-	.long	0xe49b69c1,0x9ef14ad2,0xefbe4786,0x384f25e3
-	.long	0x0fc19dc6,0x8b8cd5b5,0x240ca1cc,0x77ac9c65
-	.long	0x2de92c6f,0x592b0275,0x4a7484aa,0x6ea6e483
-	.long	0x5cb0a9dc,0xbd41fbd4,0x76f988da,0x831153b5
-	.long	0x983e5152,0xee66dfab,0xa831c66d,0x2db43210
-	.long	0xb00327c8,0x98fb213f,0xbf597fc7,0xbeef0ee4
-	.long	0xc6e00bf3,0x3da88fc2,0xd5a79147,0x930aa725
-	.long	0x06ca6351,0xe003826f,0x14292967,0x0a0e6e70
-	.long	0x27b70a85,0x46d22ffc,0x2e1b2138,0x5c26c926
-	.long	0x4d2c6dfc,0x5ac42aed,0x53380d13,0x9d95b3df
-	.long	0x650a7354,0x8baf63de,0x766a0abb,0x3c77b2a8
-	.long	0x81c2c92e,0x47edaee6,0x92722c85,0x1482353b
-	.long	0xa2bfe8a1,0x4cf10364,0xa81a664b,0xbc423001
-	.long	0xc24b8b70,0xd0f89791,0xc76c51a3,0x0654be30
-	.long	0xd192e819,0xd6ef5218,0xd6990624,0x5565a910
-	.long	0xf40e3585,0x5771202a,0x106aa070,0x32bbd1b8
-	.long	0x19a4c116,0xb8d2d0c8,0x1e376c08,0x5141ab53
-	.long	0x2748774c,0xdf8eeb99,0x34b0bcb5,0xe19b48a8
-	.long	0x391c0cb3,0xc5c95a63,0x4ed8aa4a,0xe3418acb
-	.long	0x5b9cca4f,0x7763e373,0x682e6ff3,0xd6b2b8a3
-	.long	0x748f82ee,0x5defb2fc,0x78a5636f,0x43172f60
-	.long	0x84c87814,0xa1f0ab72,0x8cc70208,0x1a6439ec
-	.long	0x90befffa,0x23631e28,0xa4506ceb,0xde82bde9
-	.long	0xbef9a3f7,0xb2c67915,0xc67178f2,0xe372532b
-	.long	0xca273ece,0xea26619c,0xd186b8c7,0x21c0c207
-	.long	0xeada7dd6,0xcde0eb1e,0xf57d4f7f,0xee6ed178
-	.long	0x06f067aa,0x72176fba,0x0a637dc5,0xa2c898a6
-	.long	0x113f9804,0xbef90dae,0x1b710b35,0x131c471b
-	.long	0x28db77f5,0x23047d84,0x32caab7b,0x40c72493
-	.long	0x3c9ebe0a,0x15c9bebc,0x431d67c4,0x9c100d4c
-	.long	0x4cc5d4be,0xcb3e42b6,0x597f299c,0xfc657e2a
-	.long	0x5fcb6fab,0x3ad6faec,0x6c44198c,0x4a475817
+	.quad	0x428a2f98d728ae22,0x7137449123ef65cd
+	.quad	0xb5c0fbcfec4d3b2f,0xe9b5dba58189dbbc
+	.quad	0x3956c25bf348b538,0x59f111f1b605d019
+	.quad	0x923f82a4af194f9b,0xab1c5ed5da6d8118
+	.quad	0xd807aa98a3030242,0x12835b0145706fbe
+	.quad	0x243185be4ee4b28c,0x550c7dc3d5ffb4e2
+	.quad	0x72be5d74f27b896f,0x80deb1fe3b1696b1
+	.quad	0x9bdc06a725c71235,0xc19bf174cf692694
+	.quad	0xe49b69c19ef14ad2,0xefbe4786384f25e3
+	.quad	0x0fc19dc68b8cd5b5,0x240ca1cc77ac9c65
+	.quad	0x2de92c6f592b0275,0x4a7484aa6ea6e483
+	.quad	0x5cb0a9dcbd41fbd4,0x76f988da831153b5
+	.quad	0x983e5152ee66dfab,0xa831c66d2db43210
+	.quad	0xb00327c898fb213f,0xbf597fc7beef0ee4
+	.quad	0xc6e00bf33da88fc2,0xd5a79147930aa725
+	.quad	0x06ca6351e003826f,0x142929670a0e6e70
+	.quad	0x27b70a8546d22ffc,0x2e1b21385c26c926
+	.quad	0x4d2c6dfc5ac42aed,0x53380d139d95b3df
+	.quad	0x650a73548baf63de,0x766a0abb3c77b2a8
+	.quad	0x81c2c92e47edaee6,0x92722c851482353b
+	.quad	0xa2bfe8a14cf10364,0xa81a664bbc423001
+	.quad	0xc24b8b70d0f89791,0xc76c51a30654be30
+	.quad	0xd192e819d6ef5218,0xd69906245565a910
+	.quad	0xf40e35855771202a,0x106aa07032bbd1b8
+	.quad	0x19a4c116b8d2d0c8,0x1e376c085141ab53
+	.quad	0x2748774cdf8eeb99,0x34b0bcb5e19b48a8
+	.quad	0x391c0cb3c5c95a63,0x4ed8aa4ae3418acb
+	.quad	0x5b9cca4f7763e373,0x682e6ff3d6b2b8a3
+	.quad	0x748f82ee5defb2fc,0x78a5636f43172f60
+	.quad	0x84c87814a1f0ab72,0x8cc702081a6439ec
+	.quad	0x90befffa23631e28,0xa4506cebde82bde9
+	.quad	0xbef9a3f7b2c67915,0xc67178f2e372532b
+	.quad	0xca273eceea26619c,0xd186b8c721c0c207
+	.quad	0xeada7dd6cde0eb1e,0xf57d4f7fee6ed178
+	.quad	0x06f067aa72176fba,0x0a637dc5a2c898a6
+	.quad	0x113f9804bef90dae,0x1b710b35131c471b
+	.quad	0x28db77f523047d84,0x32caab7b40c72493
+	.quad	0x3c9ebe0a15c9bebc,0x431d67c49c100d4c
+	.quad	0x4cc5d4becb3e42b6,0x597f299cfc657e2a
+	.quad	0x5fcb6fab3ad6faec,0x6c44198c4a475817
 ___
 $code.=<<___ if ($SZ==4);
 	.long	0x428a2f98,0x71374491,0xb5c0fbcf,0xe9b5dba5