PowerPC assembly pack: add POWER8 support.OpenSSL-fips-2_0_13

Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
(cherry picked from commit 4577871ca393275ac0436b2b08f1a75661ced314)

21 files changed, 5096 insertions, 36 deletions

diff --git a/Configure b/Configure
index a89fc9a367..4fff98b031 100755
--- a/Configure
+++ b/Configure
@@ -139,8 +139,8 @@ my $armv4_asm="armcap.o armv4cpuid.o:bn_asm.o armv4-mont.o armv4-gf2m.o::aes_cbc
 my $aarch64_asm="armcap.o arm64cpuid.o mem_clr.o:::aes_core.o aes_cbc.o aesv8-armx.o:::sha1-armv8.o sha256-armv8.o sha512-armv8.o:::::::ghashv8-armx.o:";
 my $parisc11_asm="pariscid.o:bn_asm.o parisc-mont.o::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o::32";
 my $parisc20_asm="pariscid.o:pa-risc2W.o parisc-mont.o::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o::64";
-my $ppc32_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o::aes_core.o aes_cbc.o aes-ppc.o:::sha1-ppc.o sha256-ppc.o::::::::";
-my $ppc64_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o::aes_core.o aes_cbc.o aes-ppc.o:::sha1-ppc.o sha256-ppc.o sha512-ppc.o::::::::";
+my $ppc32_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o::aes_core.o aes_cbc.o aes-ppc.o aesp8-ppc.o:::sha1-ppc.o sha256-ppc.o sha512-ppc.o sha256p8-ppc.o sha512p8-ppc.o:::::::ghashp8-ppc.o:";
+my $ppc64_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o::aes_core.o aes_cbc.o aes-ppc.o aesp8-ppc.o:::sha1-ppc.o sha256-ppc.o sha512-ppc.o sha256p8-ppc.o sha512p8-ppc.o:::::::ghashp8-ppc.o:";
 my $no_asm=":::::::::::::::void";
 
 # As for $BSDthreads. Idea is to maintain "collective" set of flags,
diff --git a/crypto/aes/Makefile b/crypto/aes/Makefile
index 1d9e82aad6..34760d174a 100644
--- a/crypto/aes/Makefile
+++ b/crypto/aes/Makefile
@@ -71,6 +71,8 @@ aes-sparcv9.s: asm/aes-sparcv9.pl
 
 aes-ppc.s:	asm/aes-ppc.pl
 	$(PERL) asm/aes-ppc.pl $(PERLASM_SCHEME) $@
+aesp8-ppc.s:	asm/aesp8-ppc.pl
+	$(PERL) asm/aesp8-ppc.pl $(PERLASM_SCHEME) $@
 
 aes-parisc.s:	asm/aes-parisc.pl
 	$(PERL) asm/aes-parisc.pl $(PERLASM_SCHEME) $@
diff --git a/crypto/aes/asm/aes-ppc.pl b/crypto/aes/asm/aes-ppc.pl
index 7c52cbe5f9..58a98232d1 100644
--- a/crypto/aes/asm/aes-ppc.pl
+++ b/crypto/aes/asm/aes-ppc.pl
@@ -548,7 +548,7 @@ Lenc_loop:
 	xor	$s2,$t2,$acc14
 	xor	$s3,$t3,$acc15
 	addi	$key,$key,16
-	bdnz-	Lenc_loop
+	bdnz	Lenc_loop
 
 	addi	$Tbl2,$Tbl0,2048
 	nop
@@ -982,7 +982,7 @@ Ldec_loop:
 	xor	$s2,$t2,$acc14
 	xor	$s3,$t3,$acc15
 	addi	$key,$key,16
-	bdnz-	Ldec_loop
+	bdnz	Ldec_loop
 
 	addi	$Tbl2,$Tbl0,2048
 	nop
diff --git a/crypto/aes/asm/aesp8-ppc.pl b/crypto/aes/asm/aesp8-ppc.pl
new file mode 100755
index 0000000000..7ef189d249
--- /dev/null
+++ b/crypto/aes/asm/aesp8-ppc.pl
@@ -0,0 +1,3726 @@
+#!/usr/bin/env perl
+#
+# ====================================================================
+# 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/.
+# ====================================================================
+#
+# This module implements support for AES instructions as per PowerISA
+# specification version 2.07, first implemented by POWER8 processor.
+# The module is endian-agnostic in sense that it supports both big-
+# and little-endian cases. Data alignment in parallelizable modes is
+# handled with VSX loads and stores, which implies MSR.VSX flag being
+# set. It should also be noted that ISA specification doesn't prohibit
+# alignment exceptions for these instructions on page boundaries.
+# Initially alignment was handled in pure AltiVec/VMX way [when data
+# is aligned programmatically, which in turn guarantees exception-
+# free execution], but it turned to hamper performance when vcipher
+# instructions are interleaved. It's reckoned that eventual
+# misalignment penalties at page boundaries are in average lower
+# than additional overhead in pure AltiVec approach.
+#
+# May 2016
+#
+# Add XTS subroutine, 9x on little- and 12x improvement on big-endian
+# systems were measured.
+#
+######################################################################
+# Current large-block performance in cycles per byte processed with
+# 128-bit key (less is better).
+#
+#		CBC en-/decrypt	CTR	XTS
+# POWER8[le]	3.96/0.72	0.74	1.1
+# POWER8[be]	3.75/0.65	0.66	1.0
+
+$flavour = shift;
+
+if ($flavour =~ /64/) {
+	$SIZE_T	=8;
+	$LRSAVE	=2*$SIZE_T;
+	$STU	="stdu";
+	$POP	="ld";
+	$PUSH	="std";
+	$UCMP	="cmpld";
+	$SHL	="sldi";
+} elsif ($flavour =~ /32/) {
+	$SIZE_T	=4;
+	$LRSAVE	=$SIZE_T;
+	$STU	="stwu";
+	$POP	="lwz";
+	$PUSH	="stw";
+	$UCMP	="cmplw";
+	$SHL	="slwi";
+} 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
+die "can't locate ppc-xlate.pl";
+
+open STDOUT,"| $^X $xlate $flavour ".shift || die "can't call $xlate: $!";
+
+$FRAME=8*$SIZE_T;
+$prefix="aes_p8";
+
+$sp="r1";
+$vrsave="r12";
+
+#########################################################################
+{{{	# Key setup procedures						#
+my ($inp,$bits,$out,$ptr,$cnt,$rounds)=map("r$_",(3..8));
+my ($zero,$in0,$in1,$key,$rcon,$mask,$tmp)=map("v$_",(0..6));
+my ($stage,$outperm,$outmask,$outhead,$outtail)=map("v$_",(7..11));
+
+$code.=<<___;
+.machine	"any"
+
+.text
+
+.align	7
+rcon:
+.long	0x01000000, 0x01000000, 0x01000000, 0x01000000	?rev
+.long	0x1b000000, 0x1b000000, 0x1b000000, 0x1b000000	?rev
+.long	0x0d0e0f0c, 0x0d0e0f0c, 0x0d0e0f0c, 0x0d0e0f0c	?rev
+.long	0,0,0,0						?asis
+Lconsts:
+	mflr	r0
+	bcl	20,31,\$+4
+	mflr	$ptr	 #vvvvv "distance between . and rcon
+	addi	$ptr,$ptr,-0x48
+	mtlr	r0
+	blr
+	.long	0
+	.byte	0,12,0x14,0,0,0,0,0
+.asciz	"AES for PowerISA 2.07, CRYPTOGAMS by <appro\@openssl.org>"
+
+.globl	.${prefix}_set_encrypt_key
+.align	5
+.${prefix}_set_encrypt_key:
+Lset_encrypt_key:
+	mflr		r11
+	$PUSH		r11,$LRSAVE($sp)
+
+	li		$ptr,-1
+	${UCMP}i	$inp,0
+	beq-		Lenc_key_abort		# if ($inp==0) return -1;
+	${UCMP}i	$out,0
+	beq-		Lenc_key_abort		# if ($out==0) return -1;
+	li		$ptr,-2
+	cmpwi		$bits,128
+	blt-		Lenc_key_abort
+	cmpwi		$bits,256
+	bgt-		Lenc_key_abort
+	andi.		r0,$bits,0x3f
+	bne-		Lenc_key_abort
+
+	lis		r0,0xfff0
+	mfspr		$vrsave,256
+	mtspr		256,r0
+
+	bl		Lconsts
+	mtlr		r11
+
+	neg		r9,$inp
+	lvx		$in0,0,$inp
+	addi		$inp,$inp,15		# 15 is not typo
+	lvsr		$key,0,r9		# borrow $key
+	li		r8,0x20
+	cmpwi		$bits,192
+	lvx		$in1,0,$inp
+	le?vspltisb	$mask,0x0f		# borrow $mask
+	lvx		$rcon,0,$ptr
+	le?vxor		$key,$key,$mask		# adjust for byte swap
+	lvx		$mask,r8,$ptr
+	addi		$ptr,$ptr,0x10
+	vperm		$in0,$in0,$in1,$key	# align [and byte swap in LE]
+	li		$cnt,8
+	vxor		$zero,$zero,$zero
+	mtctr		$cnt
+
+	?lvsr		$outperm,0,$out
+	vspltisb	$outmask,-1
+	lvx		$outhead,0,$out
+	?vperm		$outmask,$zero,$outmask,$outperm
+
+	blt		Loop128
+	addi		$inp,$inp,8
+	beq		L192
+	addi		$inp,$inp,8
+	b		L256
+
+.align	4
+Loop128:
+	vperm		$key,$in0,$in0,$mask	# rotate-n-splat
+	vsldoi		$tmp,$zero,$in0,12	# >>32
+	 vperm		$outtail,$in0,$in0,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	vcipherlast	$key,$key,$rcon
+	 stvx		$stage,0,$out
+	 addi		$out,$out,16
+
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	 vadduwm	$rcon,$rcon,$rcon
+	vxor		$in0,$in0,$key
+	bdnz		Loop128
+
+	lvx		$rcon,0,$ptr		# last two round keys
+
+	vperm		$key,$in0,$in0,$mask	# rotate-n-splat
+	vsldoi		$tmp,$zero,$in0,12	# >>32
+	 vperm		$outtail,$in0,$in0,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	vcipherlast	$key,$key,$rcon
+	 stvx		$stage,0,$out
+	 addi		$out,$out,16
+
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	 vadduwm	$rcon,$rcon,$rcon
+	vxor		$in0,$in0,$key
+
+	vperm		$key,$in0,$in0,$mask	# rotate-n-splat
+	vsldoi		$tmp,$zero,$in0,12	# >>32
+	 vperm		$outtail,$in0,$in0,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	vcipherlast	$key,$key,$rcon
+	 stvx		$stage,0,$out
+	 addi		$out,$out,16
+
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	vxor		$in0,$in0,$key
+	 vperm		$outtail,$in0,$in0,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	 stvx		$stage,0,$out
+
+	addi		$inp,$out,15		# 15 is not typo
+	addi		$out,$out,0x50
+
+	li		$rounds,10
+	b		Ldone
+
+.align	4
+L192:
+	lvx		$tmp,0,$inp
+	li		$cnt,4
+	 vperm		$outtail,$in0,$in0,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	 stvx		$stage,0,$out
+	 addi		$out,$out,16
+	vperm		$in1,$in1,$tmp,$key	# align [and byte swap in LE]
+	vspltisb	$key,8			# borrow $key
+	mtctr		$cnt
+	vsububm		$mask,$mask,$key	# adjust the mask
+
+Loop192:
+	vperm		$key,$in1,$in1,$mask	# roate-n-splat
+	vsldoi		$tmp,$zero,$in0,12	# >>32
+	vcipherlast	$key,$key,$rcon
+
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+
+	 vsldoi		$stage,$zero,$in1,8
+	vspltw		$tmp,$in0,3
+	vxor		$tmp,$tmp,$in1
+	vsldoi		$in1,$zero,$in1,12	# >>32
+	 vadduwm	$rcon,$rcon,$rcon
+	vxor		$in1,$in1,$tmp
+	vxor		$in0,$in0,$key
+	vxor		$in1,$in1,$key
+	 vsldoi		$stage,$stage,$in0,8
+
+	vperm		$key,$in1,$in1,$mask	# rotate-n-splat
+	vsldoi		$tmp,$zero,$in0,12	# >>32
+	 vperm		$outtail,$stage,$stage,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	vcipherlast	$key,$key,$rcon
+	 stvx		$stage,0,$out
+	 addi		$out,$out,16
+
+	 vsldoi		$stage,$in0,$in1,8
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	 vperm		$outtail,$stage,$stage,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	 stvx		$stage,0,$out
+	 addi		$out,$out,16
+
+	vspltw		$tmp,$in0,3
+	vxor		$tmp,$tmp,$in1
+	vsldoi		$in1,$zero,$in1,12	# >>32
+	 vadduwm	$rcon,$rcon,$rcon
+	vxor		$in1,$in1,$tmp
+	vxor		$in0,$in0,$key
+	vxor		$in1,$in1,$key
+	 vperm		$outtail,$in0,$in0,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	 stvx		$stage,0,$out
+	 addi		$inp,$out,15		# 15 is not typo
+	 addi		$out,$out,16
+	bdnz		Loop192
+
+	li		$rounds,12
+	addi		$out,$out,0x20
+	b		Ldone
+
+.align	4
+L256:
+	lvx		$tmp,0,$inp
+	li		$cnt,7
+	li		$rounds,14
+	 vperm		$outtail,$in0,$in0,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	 stvx		$stage,0,$out
+	 addi		$out,$out,16
+	vperm		$in1,$in1,$tmp,$key	# align [and byte swap in LE]
+	mtctr		$cnt
+
+Loop256:
+	vperm		$key,$in1,$in1,$mask	# rotate-n-splat
+	vsldoi		$tmp,$zero,$in0,12	# >>32
+	 vperm		$outtail,$in1,$in1,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	vcipherlast	$key,$key,$rcon
+	 stvx		$stage,0,$out
+	 addi		$out,$out,16
+
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in0,$in0,$tmp
+	 vadduwm	$rcon,$rcon,$rcon
+	vxor		$in0,$in0,$key
+	 vperm		$outtail,$in0,$in0,$outperm	# rotate
+	 vsel		$stage,$outhead,$outtail,$outmask
+	 vmr		$outhead,$outtail
+	 stvx		$stage,0,$out
+	 addi		$inp,$out,15		# 15 is not typo
+	 addi		$out,$out,16
+	bdz		Ldone
+
+	vspltw		$key,$in0,3		# just splat
+	vsldoi		$tmp,$zero,$in1,12	# >>32
+	vsbox		$key,$key
+
+	vxor		$in1,$in1,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in1,$in1,$tmp
+	vsldoi		$tmp,$zero,$tmp,12	# >>32
+	vxor		$in1,$in1,$tmp
+
+	vxor		$in1,$in1,$key
+	b		Loop256
+
+.align	4
+Ldone:
+	lvx		$in1,0,$inp		# redundant in aligned case
+	vsel		$in1,$outhead,$in1,$outmask
+	stvx		$in1,0,$inp
+	li		$ptr,0
+	mtspr		256,$vrsave
+	stw		$rounds,0($out)
+
+Lenc_key_abort:
+	mr		r3,$ptr
+	blr
+	.long		0
+	.byte		0,12,0x14,1,0,0,3,0
+	.long		0
+.size	.${prefix}_set_encrypt_key,.-.${prefix}_set_encrypt_key
+
+.globl	.${prefix}_set_decrypt_key
+.align	5
+.${prefix}_set_decrypt_key:
+	$STU		$sp,-$FRAME($sp)
+	mflr		r10
+	$PUSH		r10,$FRAME+$LRSAVE($sp)
+	bl		Lset_encrypt_key
+	mtlr		r10
+
+	cmpwi		r3,0
+	bne-		Ldec_key_abort
+
+	slwi		$cnt,$rounds,4
+	subi		$inp,$out,240		# first round key
+	srwi		$rounds,$rounds,1
+	add		$out,$inp,$cnt		# last round key
+	mtctr		$rounds
+
+Ldeckey:
+	lwz		r0, 0($inp)
+	lwz		r6, 4($inp)
+	lwz		r7, 8($inp)
+	lwz		r8, 12($inp)
+	addi		$inp,$inp,16
+	lwz		r9, 0($out)
+	lwz		r10,4($out)
+	lwz		r11,8($out)
+	lwz		r12,12($out)
+	stw		r0, 0($out)
+	stw		r6, 4($out)
+	stw		r7, 8($out)
+	stw		r8, 12($out)
+	subi		$out,$out,16
+	stw		r9, -16($inp)
+	stw		r10,-12($inp)
+	stw		r11,-8($inp)
+	stw		r12,-4($inp)
+	bdnz		Ldeckey
+
+	xor		r3,r3,r3		# return value
+Ldec_key_abort:
+	addi		$sp,$sp,$FRAME
+	blr
+	.long		0
+	.byte		0,12,4,1,0x80,0,3,0
+	.long		0
+.size	.${prefix}_set_decrypt_key,.-.${prefix}_set_decrypt_key
+___
+}}}
+#########################################################################
+{{{	# Single block en- and decrypt procedures			#
+sub gen_block () {
+my $dir = shift;
+my $n   = $dir eq "de" ? "n" : "";
+my ($inp,$out,$key,$rounds,$idx)=map("r$_",(3..7));
+
+$code.=<<___;
+.globl	.${prefix}_${dir}crypt
+.align	5
+.${prefix}_${dir}crypt:
+	lwz		$rounds,240($key)
+	lis		r0,0xfc00
+	mfspr		$vrsave,256
+	li		$idx,15			# 15 is not typo
+	mtspr		256,r0
+
+	lvx		v0,0,$inp
+	neg		r11,$out
+	lvx		v1,$idx,$inp
+	lvsl		v2,0,$inp		# inpperm
+	le?vspltisb	v4,0x0f
+	?lvsl		v3,0,r11		# outperm
+	le?vxor		v2,v2,v4
+	li		$idx,16
+	vperm		v0,v0,v1,v2		# align [and byte swap in LE]
+	lvx		v1,0,$key
+	?lvsl		v5,0,$key		# keyperm
+	srwi		$rounds,$rounds,1
+	lvx		v2,$idx,$key
+	addi		$idx,$idx,16
+	subi		$rounds,$rounds,1
+	?vperm		v1,v1,v2,v5		# align round key
+
+	vxor		v0,v0,v1
+	lvx		v1,$idx,$key
+	addi		$idx,$idx,16
+	mtctr		$rounds
+
+Loop_${dir}c:
+	?vperm		v2,v2,v1,v5
+	v${n}cipher	v0,v0,v2
+	lvx		v2,$idx,$key
+	addi		$idx,$idx,16
+	?vperm		v1,v1,v2,v5
+	v${n}cipher	v0,v0,v1
+	lvx		v1,$idx,$key
+	addi		$idx,$idx,16
+	bdnz		Loop_${dir}c
+
+	?vperm		v2,v2,v1,v5
+	v${n}cipher	v0,v0,v2
+	lvx		v2,$idx,$key
+	?vperm		v1,v1,v2,v5
+	v${n}cipherlast	v0,v0,v1
+
+	vspltisb	v2,-1
+	vxor		v1,v1,v1
+	li		$idx,15			# 15 is not typo
+	?vperm		v2,v1,v2,v3		# outmask
+	le?vxor		v3,v3,v4
+	lvx		v1,0,$out		# outhead
+	vperm		v0,v0,v0,v3		# rotate [and byte swap in LE]
+	vsel		v1,v1,v0,v2
+	lvx		v4,$idx,$out
+	stvx		v1,0,$out
+	vsel		v0,v0,v4,v2
+	stvx		v0,$idx,$out
+
+	mtspr		256,$vrsave
+	blr
+	.long		0
+	.byte		0,12,0x14,0,0,0,3,0
+	.long		0
+.size	.${prefix}_${dir}crypt,.-.${prefix}_${dir}crypt
+___
+}
+&gen_block("en");
+&gen_block("de");
+}}}
+#########################################################################
+{{{	# CBC en- and decrypt procedures				#
+my ($inp,$out,$len,$key,$ivp,$enc,$rounds,$idx)=map("r$_",(3..10));
+my ($rndkey0,$rndkey1,$inout,$tmp)=		map("v$_",(0..3));
+my ($ivec,$inptail,$inpperm,$outhead,$outperm,$outmask,$keyperm)=
+						map("v$_",(4..10));
+$code.=<<___;
+.globl	.${prefix}_cbc_encrypt
+.align	5
+.${prefix}_cbc_encrypt:
+	${UCMP}i	$len,16
+	bltlr-
+
+	cmpwi		$enc,0			# test direction
+	lis		r0,0xffe0
+	mfspr		$vrsave,256
+	mtspr		256,r0
+
+	li		$idx,15
+	vxor		$rndkey0,$rndkey0,$rndkey0
+	le?vspltisb	$tmp,0x0f
+
+	lvx		$ivec,0,$ivp		# load [unaligned] iv
+	lvsl		$inpperm,0,$ivp
+	lvx		$inptail,$idx,$ivp
+	le?vxor		$inpperm,$inpperm,$tmp
+	vperm		$ivec,$ivec,$inptail,$inpperm
+
+	neg		r11,$inp
+	?lvsl		$keyperm,0,$key		# prepare for unaligned key
+	lwz		$rounds,240($key)
+
+	lvsr		$inpperm,0,r11		# prepare for unaligned load
+	lvx		$inptail,0,$inp
+	addi		$inp,$inp,15		# 15 is not typo
+	le?vxor		$inpperm,$inpperm,$tmp
+
+	?lvsr		$outperm,0,$out		# prepare for unaligned store
+	vspltisb	$outmask,-1
+	lvx		$outhead,0,$out
+	?vperm		$outmask,$rndkey0,$outmask,$outperm
+	le?vxor		$outperm,$outperm,$tmp
+
+	srwi		$rounds,$rounds,1
+	li		$idx,16
+	subi		$rounds,$rounds,1
+	beq		Lcbc_dec
+
+Lcbc_enc:
+	vmr		$inout,$inptail
+	lvx		$inptail,0,$inp
+	addi		$inp,$inp,16
+	mtctr		$rounds
+	subi		$len,$len,16		# len-=16
+
+	lvx		$rndkey0,0,$key
+	 vperm		$inout,$inout,$inptail,$inpperm
+	lvx		$rndkey1,$idx,$key
+	addi		$idx,$idx,16
+	?vperm		$rndkey0,$rndkey0,$rndkey1,$keyperm
+	vxor		$inout,$inout,$rndkey0
+	lvx		$rndkey0,$idx,$key
+	addi		$idx,$idx,16
+	vxor		$inout,$inout,$ivec
+
+Loop_cbc_enc:
+	?vperm		$rndkey1,$rndkey1,$rndkey0,$keyperm
+	vcipher		$inout,$inout,$rndkey1
+	lvx		$rndkey1,$idx,$key
+	addi		$idx,$idx,16
+	?vperm		$rndkey0,$rndkey0,$rndkey1,$keyperm
+	vcipher		$inout,$inout,$rndkey0
+	lvx		$rndkey0,$idx,$key
+	addi		$idx,$idx,16
+	bdnz		Loop_cbc_enc
+
+	?vperm		$rndkey1,$rndkey1,$rndkey0,$keyperm
+	vcipher		$inout,$inout,$rndkey1
+	lvx		$rndkey1,$idx,$key
+	li		$idx,16
+	?vperm		$rndkey0,$rndkey0,$rndkey1,$keyperm
+	vcipherlast	$ivec,$inout,$rndkey0
+	${UCMP}i	$len,16
+
+	vperm		$tmp,$ivec,$ivec,$outperm
+	vsel		$inout,$outhead,$tmp,$outmask
+	vmr		$outhead,$tmp
+	stvx		$inout,0,$out
+	addi		$out,$out,16
+	bge		Lcbc_enc
+
+	b		Lcbc_done
+
+.align	4
+Lcbc_dec:
+	${UCMP}i	$len,128
+	bge		_aesp8_cbc_decrypt8x
+	vmr		$tmp,$inptail
+	lvx		$inptail,0,$inp
+	addi		$inp,$inp,16
+	mtctr		$rounds
+	subi		$len,$len,16		# len-=16
+
+	lvx		$rndkey0,0,$key
+	 vperm		$tmp,$tmp,$inptail,$inpperm
+	lvx		$rndkey1,$idx,$key
+	addi		$idx,$idx,16
+	?vperm		$rndkey0,$rndkey0,$rndkey1,$keyperm
+	vxor		$inout,$tmp,$rndkey0
+	lvx		$rndkey0,$idx,$key
+	addi		$idx,$idx,16
+
+Loop_cbc_dec:
+	?vperm		$rndkey1,$rndkey1,$rndkey0,$keyperm
+	vncipher	$inout,$inout,$rndkey1
+	lvx		$rndkey1,$idx,$key
+	addi		$idx,$idx,16
+	?vperm		$rndkey0,$rndkey0,$rndkey1,$keyperm
+	vncipher	$inout,$inout,$rndkey0
+	lvx		$rndkey0,$idx,$key
+	addi		$idx,$idx,16
+	bdnz		Loop_cbc_dec
+
+	?vperm		$rndkey1,$rndkey1,$rndkey0,$keyperm
+	vncipher	$inout,$inout,$rndkey1
+	lvx		$rndkey1,$idx,$key
+	li		$idx,16
+	?vperm		$rndkey0,$rndkey0,$rndkey1,$keyperm
+	vncipherlast	$inout,$inout,$rndkey0
+	${UCMP}i	$len,16
+
+	vxor		$inout,$inout,$ivec
+	vmr		$ivec,$tmp
+	vperm		$tmp,$inout,$inout,$outperm
+	vsel		$inout,$outhead,$tmp,$outmask
+	vmr		$outhead,$tmp
+	stvx		$inout,0,$out
+	addi		$out,$out,16
+	bge		Lcbc_dec
+
+Lcbc_done:
+	addi		$out,$out,-1
+	lvx		$inout,0,$out		# redundant in aligned case
+	vsel		$inout,$outhead,$inout,$outmask
+	stvx		$inout,0,$out
+
+	neg		$enc,$ivp		# write [unaligned] iv
+	li		$idx,15			# 15 is not typo
+	vxor		$rndkey0,$rndkey0,$rndkey0
+	vspltisb	$outmask,-1
+	le?vspltisb	$tmp,0x0f
+	?lvsl		$outperm,0,$enc
+	?vperm		$outmask,$rndkey0,$outmask,$outperm
+	le?vxor		$outperm,$outperm,$tmp
+	lvx		$outhead,0,$ivp
+	vperm		$ivec,$ivec,$ivec,$outperm
+	vsel		$inout,$outhead,$ivec,$outmask
+	lvx		$inptail,$idx,$ivp
+	stvx		$inout,0,$ivp
+	vsel		$inout,$ivec,$inptail,$outmask
+	stvx		$inout,$idx,$ivp
+
+	mtspr		256,$vrsave
+	blr
+	.long		0
+	.byte		0,12,0x14,0,0,0,6,0
+	.long		0
+___
+#########################################################################
+{{	# Optimized CBC decrypt procedure				#
+my $key_="r11";
+my ($x00,$x10,$x20,$x30,$x40,$x50,$x60,$x70)=map("r$_",(0,8,26..31));
+    $x00=0 if ($flavour =~ /osx/);
+my ($in0, $in1, $in2, $in3, $in4, $in5, $in6, $in7 )=map("v$_",(0..3,10..13));
+my ($out0,$out1,$out2,$out3,$out4,$out5,$out6,$out7)=map("v$_",(14..21));
+my $rndkey0="v23";	# v24-v25 rotating buffer for first found keys
+			# v26-v31 last 6 round keys
+my ($tmp,$keyperm)=($in3,$in4);	# aliases with "caller", redundant assignment
+
+$code.=<<___;
+.align	5
+_aesp8_cbc_decrypt8x:
+	$STU		$sp,-`($FRAME+21*16+6*$SIZE_T)`($sp)
+	li		r10,`$FRAME+8*16+15`
+	li		r11,`$FRAME+8*16+31`
+	stvx		v20,r10,$sp		# ABI says so
+	addi		r10,r10,32
+	stvx		v21,r11,$sp
+	addi		r11,r11,32
+	stvx		v22,r10,$sp
+	addi		r10,r10,32
+	stvx		v23,r11,$sp
+	addi		r11,r11,32
+	stvx		v24,r10,$sp
+	addi		r10,r10,32
+	stvx		v25,r11,$sp
+	addi		r11,r11,32
+	stvx		v26,r10,$sp
+	addi		r10,r10,32
+	stvx		v27,r11,$sp
+	addi		r11,r11,32
+	stvx		v28,r10,$sp
+	addi		r10,r10,32
+	stvx		v29,r11,$sp
+	addi		r11,r11,32
+	stvx		v30,r10,$sp
+	stvx		v31,r11,$sp
+	li		r0,-1
+	stw		$vrsave,`$FRAME+21*16-4`($sp)	# save vrsave
+	li		$x10,0x10
+	$PUSH		r26,`$FRAME+21*16+0*$SIZE_T`($sp)
+	li		$x20,0x20
+	$PUSH		r27,`$FRAME+21*16+1*$SIZE_T`($sp)
+	li		$x30,0x30
+	$PUSH		r28,`$FRAME+21*16+2*$SIZE_T`($sp)
+	li		$x40,0x40
+	$PUSH		r29,`$FRAME+21*16+3*$SIZE_T`($sp)
+	li		$x50,0x50
+	$PUSH		r30,`$FRAME+21*16+4*$SIZE_T`($sp)
+	li		$x60,0x60
+	$PUSH		r31,`$FRAME+21*16+5*$SIZE_T`($sp)
+	li		$x70,0x70
+	mtspr		256,r0
+
+	subi		$rounds,$rounds,3	# -4 in total
+	subi		$len,$len,128		# bias
+
+	lvx		$rndkey0,$x00,$key	# load key schedule
+	lvx		v30,$x10,$key
+	addi		$key,$key,0x20
+	lvx		v31,$x00,$key
+	?vperm		$rndkey0,$rndkey0,v30,$keyperm
+	addi		$key_,$sp,$FRAME+15
+	mtctr		$rounds
+
+Load_cbc_dec_key:
+	?vperm		v24,v30,v31,$keyperm
+	lvx		v30,$x10,$key
+	addi		$key,$key,0x20
+	stvx		v24,$x00,$key_		# off-load round[1]
+	?vperm		v25,v31,v30,$keyperm
+	lvx		v31,$x00,$key
+	stvx		v25,$x10,$key_		# off-load round[2]
+	addi		$key_,$key_,0x20
+	bdnz		Load_cbc_dec_key
+
+	lvx		v26,$x10,$key
+	?vperm		v24,v30,v31,$keyperm
+	lvx		v27,$x20,$key
+	stvx		v24,$x00,$key_		# off-load round[3]
+	?vperm		v25,v31,v26,$keyperm
+	lvx		v28,$x30,$key
+	stvx		v25,$x10,$key_		# off-load round[4]
+	addi		$key_,$sp,$FRAME+15	# rewind $key_
+	?vperm		v26,v26,v27,$keyperm
+	lvx		v29,$x40,$key
+	?vperm		v27,v27,v28,$keyperm
+	lvx		v30,$x50,$key
+	?vperm		v28,v28,v29,$keyperm
+	lvx		v31,$x60,$key
+	?vperm		v29,v29,v30,$keyperm
+	lvx		$out0,$x70,$key		# borrow $out0
+	?vperm		v30,v30,v31,$keyperm
+	lvx		v24,$x00,$key_		# pre-load round[1]
+	?vperm		v31,v31,$out0,$keyperm
+	lvx		v25,$x10,$key_		# pre-load round[2]
+
+	#lvx		$inptail,0,$inp		# "caller" already did this
+	#addi		$inp,$inp,15		# 15 is not typo
+	subi		$inp,$inp,15		# undo "caller"
+
+	 le?li		$idx,8
+	lvx_u		$in0,$x00,$inp		# load first 8 "words"
+	 le?lvsl	$inpperm,0,$idx
+	 le?vspltisb	$tmp,0x0f
+	lvx_u		$in1,$x10,$inp
+	 le?vxor	$inpperm,$inpperm,$tmp	# transform for lvx_u/stvx_u
+	lvx_u		$in2,$x20,$inp
+	 le?vperm	$in0,$in0,$in0,$inpperm
+	lvx_u		$in3,$x30,$inp
+	 le?vperm	$in1,$in1,$in1,$inpperm
+	lvx_u		$in4,$x40,$inp
+	 le?vperm	$in2,$in2,$in2,$inpperm
+	vxor		$out0,$in0,$rndkey0
+	lvx_u		$in5,$x50,$inp
+	 le?vperm	$in3,$in3,$in3,$inpperm
+	vxor		$out1,$in1,$rndkey0
+	lvx_u		$in6,$x60,$inp
+	 le?vperm	$in4,$in4,$in4,$inpperm
+	vxor		$out2,$in2,$rndkey0
+	lvx_u		$in7,$x70,$inp
+	addi		$inp,$inp,0x80
+	 le?vperm	$in5,$in5,$in5,$inpperm
+	vxor		$out3,$in3,$rndkey0
+	 le?vperm	$in6,$in6,$in6,$inpperm
+	vxor		$out4,$in4,$rndkey0
+	 le?vperm	$in7,$in7,$in7,$inpperm
+	vxor		$out5,$in5,$rndkey0
+	vxor		$out6,$in6,$rndkey0
+	vxor		$out7,$in7,$rndkey0
+
+	mtctr		$rounds
+	b		Loop_cbc_dec8x
+.align	5
+Loop_cbc_dec8x:
+	vncipher	$out0,$out0,v24
+	vncipher	$out1,$out1,v24
+	vncipher	$out2,$out2,v24
+	vncipher	$out3,$out3,v24
+	vncipher	$out4,$out4,v24
+	vncipher	$out5,$out5,v24
+	vncipher	$out6,$out6,v24
+	vncipher	$out7,$out7,v24
+	lvx		v24,$x20,$key_		# round[3]
+	addi		$key_,$key_,0x20
+
+	vncipher	$out0,$out0,v25
+	vncipher	$out1,$out1,v25
+	vncipher	$out2,$out2,v25
+	vncipher	$out3,$out3,v25
+	vncipher	$out4,$out4,v25
+	vncipher	$out5,$out5,v25
+	vncipher	$out6,$out6,v25
+	vncipher	$out7,$out7,v25
+	lvx		v25,$x10,$key_		# round[4]
+	bdnz		Loop_cbc_dec8x
+
+	subic		$len,$len,128		# $len-=128
+	vncipher	$out0,$out0,v24
+	vncipher	$out1,$out1,v24
+	vncipher	$out2,$out2,v24
+	vncipher	$out3,$out3,v24
+	vncipher	$out4,$out4,v24
+	vncipher	$out5,$out5,v24
+	vncipher	$out6,$out6,v24
+	vncipher	$out7,$out7,v24
+
+	subfe.		r0,r0,r0		# borrow?-1:0
+	vncipher	$out0,$out0,v25
+	vncipher	$out1,$out1,v25
+	vncipher	$out2,$out2,v25
+	vncipher	$out3,$out3,v25
+	vncipher	$out4,$out4,v25
+	vncipher	$out5,$out5,v25
+	vncipher	$out6,$out6,v25
+	vncipher	$out7,$out7,v25
+
+	and		r0,r0,$len
+	vncipher	$out0,$out0,v26
+	vncipher	$out1,$out1,v26
+	vncipher	$out2,$out2,v26
+	vncipher	$out3,$out3,v26
+	vncipher	$out4,$out4,v26
+	vncipher	$out5,$out5,v26
+	vncipher	$out6,$out6,v26
+	vncipher	$out7,$out7,v26
+
+	add		$inp,$inp,r0		# $inp is adjusted in such
+						# way that at exit from the
+						# loop inX-in7 are loaded
+						# with last "words"
+	vncipher	$out0,$out0,v27
+	vncipher	$out1,$out1,v27
+	vncipher	$out2,$out2,v27
+	vncipher	$out3,$out3,v27
+	vncipher	$out4,$out4,v27
+	vncipher	$out5,$out5,v27
+	vncipher	$out6,$out6,v27
+	vncipher	$out7,$out7,v27
+
+	addi		$key_,$sp,$FRAME+15	# rewind $key_
+	vncipher	$out0,$out0,v28
+	vncipher	$out1,$out1,v28
+	vncipher	$out2,$out2,v28
+	vncipher	$out3,$out3,v28
+	vncipher	$out4,$out4,v28
+	vncipher	$out5,$out5,v28
+	vncipher	$out6,$out6,v28
+	vncipher	$out7,$out7,v28
+	lvx		v24,$x00,$key_		# re-pre-load round[1]
+
+	vncipher	$out0,$out0,v29
+	vncipher	$out1,$out1,v29
+	vncipher	$out2,$out2,v29
+	vncipher	$out3,$out3,v29
+	vncipher	$out4,$out4,v29
+	vncipher	$out5,$out5,v29
+	vncipher	$out6,$out6,v29
+	vncipher	$out7,$out7,v29
+	lvx		v25,$x10,$key_		# re-pre-load round[2]
+
+	vncipher	$out0,$out0,v30
+	 vxor		$ivec,$ivec,v31		# xor with last round key
+	vncipher	$out1,$out1,v30
+	 vxor		$in0,$in0,v31
+	vncipher	$out2,$out2,v30
+	 vxor		$in1,$in1,v31
+	vncipher	$out3,$out3,v30
+	 vxor		$in2,$in2,v31
+	vncipher	$out4,$out4,v30
+	 vxor		$in3,$in3,v31
+	vncipher	$out5,$out5,v30
+	 vxor		$in4,$in4,v31
+	vncipher	$out6,$out6,v30
+	 vxor		$in5,$in5,v31
+	vncipher	$out7,$out7,v30
+	 vxor		$in6,$in6,v31
+
+	vncipherlast	$out0,$out0,$ivec
+	vncipherlast	$out1,$out1,$in0
+	 lvx_u		$in0,$x00,$inp		# load next input block
+	vncipherlast	$out2,$out2,$in1
+	 lvx_u		$in1,$x10,$inp
+	vncipherlast	$out3,$out3,$in2
+	 le?vperm	$in0,$in0,$in0,$inpperm
+	 lvx_u		$in2,$x20,$inp
+	vncipherlast	$out4,$out4,$in3
+	 le?vperm	$in1,$in1,$in1,$inpperm
+	 lvx_u		$in3,$x30,$inp
+	vncipherlast	$out5,$out5,$in4
+	 le?vperm	$in2,$in2,$in2,$inpperm
+	 lvx_u		$in4,$x40,$inp
+	vncipherlast	$out6,$out6,$in5
+	 le?vperm	$in3,$in3,$in3,$inpperm
+	 lvx_u		$in5,$x50,$inp
+	vncipherlast	$out7,$out7,$in6
+	 le?vperm	$in4,$in4,$in4,$inpperm
+	 lvx_u		$in6,$x60,$inp
+	vmr		$ivec,$in7
+	 le?vperm	$in5,$in5,$in5,$inpperm
+	 lvx_u		$in7,$x70,$inp
+	 addi		$inp,$inp,0x80
+
+	le?vperm	$out0,$out0,$out0,$inpperm
+	le?vperm	$out1,$out1,$out1,$inpperm
+	stvx_u		$out0,$x00,$out
+	 le?vperm	$in6,$in6,$in6,$inpperm
+	 vxor		$out0,$in0,$rndkey0
+	le?vperm	$out2,$out2,$out2,$inpperm
+	stvx_u		$out1,$x10,$out
+	 le?vperm	$in7,$in7,$in7,$inpperm
+	 vxor		$out1,$in1,$rndkey0
+	le?vperm	$out3,$out3,$out3,$inpperm
+	stvx_u		$out2,$x20,$out
+	 vxor		$out2,$in2,$rndkey0
+	le?vperm	$out4,$out4,$out4,$inpperm
+	stvx_u		$out3,$x30,$out
+	 vxor		$out3,$in3,$rndkey0
+	le?vperm	$out5,$out5,$out5,$inpperm
+	stvx_u		$out4,$x40,$out
+	 vxor		$out4,$in4,$rndkey0
+	le?vperm	$out6,$out6,$out6,$inpperm
+	stvx_u		$out5,$x50,$out
+	 vxor		$out5,$in5,$rndkey0</