Merge main -> VMS_64BITBRANCH_VMS_64BIT

13 files changed, 3654 insertions, 0 deletions

diff --git a/crypto/camellia/.cvsignore b/crypto/camellia/.cvsignore
new file mode 100644
index 0000000000..b7f68a8535
--- /dev/null
+++ b/crypto/camellia/.cvsignore
@@ -0,0 +1,3 @@
+lib
+Makefile.save
+cmll-*.s
diff --git a/crypto/camellia/Makefile b/crypto/camellia/Makefile
new file mode 100644
index 0000000000..ff5fe4a01d
--- /dev/null
+++ b/crypto/camellia/Makefile
@@ -0,0 +1,103 @@
+#
+# crypto/camellia/Makefile
+#
+
+DIR= camellia
+TOP=	../..
+CC=	cc
+CPP=	$(CC) -E
+INCLUDES=
+CFLAG=-g
+MAKEFILE=	Makefile
+AR=		ar r
+
+CMLL_ENC= camellia.o cmll_misc.o cmll_cbc.o
+
+CFLAGS= $(INCLUDES) $(CFLAG)
+ASFLAGS= $(INCLUDES) $(ASFLAG)
+AFLAGS= $(ASFLAGS)
+
+GENERAL=Makefile
+#TEST=camelliatest.c
+APPS=
+
+LIB=$(TOP)/libcrypto.a
+LIBSRC=camellia.c cmll_misc.c cmll_ecb.c cmll_cbc.c cmll_ofb.c \
+	   cmll_cfb.c cmll_ctr.c 
+
+LIBOBJ= cmll_ecb.o cmll_ofb.o cmll_cfb.o cmll_ctr.o $(CMLL_ENC)
+
+SRC= $(LIBSRC)
+
+EXHEADER= camellia.h
+HEADER= cmll_locl.h $(EXHEADER)
+
+ALL=    $(GENERAL) $(SRC) $(HEADER)
+
+top:
+	(cd ../..; $(MAKE) DIRS=crypto SDIRS=$(DIR) sub_all)
+
+all:	lib
+
+lib:	$(LIBOBJ)
+	$(AR) $(LIB) $(LIBOBJ)
+	$(RANLIB) $(LIB) || echo Never mind.
+	@touch lib
+
+cmll-x86.s:	asm/cmll-x86.pl ../perlasm/x86asm.pl
+	$(PERL) asm/cmll-x86.pl $(PERLASM_SCHEME) $(CFLAGS) $(PROCESSOR) > $@
+cmll-x86_64.s:  asm/cmll-x86_64.pl
+	$(PERL) asm/cmll-x86_64.pl $(PERLASM_SCHEME) > $@
+
+files:
+	$(PERL) $(TOP)/util/files.pl Makefile >> $(TOP)/MINFO
+
+links:
+	@$(PERL) $(TOP)/util/mklink.pl ../../include/openssl $(EXHEADER)
+	@$(PERL) $(TOP)/util/mklink.pl ../../test $(TEST)
+	@$(PERL) $(TOP)/util/mklink.pl ../../apps $(APPS)
+
+install:
+	@[ -n "$(INSTALLTOP)" ] # should be set by top Makefile...
+	@headerlist="$(EXHEADER)"; for i in $$headerlist ; \
+	do  \
+	(cp $$i $(INSTALL_PREFIX)$(INSTALLTOP)/include/openssl/$$i; \
+	chmod 644 $(INSTALL_PREFIX)$(INSTALLTOP)/include/openssl/$$i ); \
+	done;
+
+tags:
+	ctags $(SRC)
+
+tests:
+
+lint:
+	lint -DLINT $(INCLUDES) $(SRC)>fluff
+
+depend:
+	@[ -n "$(MAKEDEPEND)" ] # should be set by upper Makefile...
+	$(MAKEDEPEND) -- $(CFLAG) $(INCLUDES) $(DEPFLAG) -- $(PROGS) $(LIBSRC)
+
+dclean:
+	$(PERL) -pe 'if (/^# DO NOT DELETE THIS LINE/) {print; exit(0);}' $(MAKEFILE) >Makefile.new
+	mv -f Makefile.new $(MAKEFILE)
+
+clean:
+	rm -f *.s *.o *.obj lib tags core .pure .nfs* *.old *.bak fluff
+
+# DO NOT DELETE THIS LINE -- make depend depends on it.
+
+camellia.o: ../../include/openssl/opensslconf.h camellia.c camellia.h
+camellia.o: cmll_locl.h
+cmll_cbc.o: ../../include/openssl/camellia.h ../../include/openssl/modes.h
+cmll_cbc.o: ../../include/openssl/opensslconf.h cmll_cbc.c
+cmll_cfb.o: ../../include/openssl/camellia.h ../../include/openssl/modes.h
+cmll_cfb.o: ../../include/openssl/opensslconf.h cmll_cfb.c
+cmll_ctr.o: ../../include/openssl/camellia.h ../../include/openssl/modes.h
+cmll_ctr.o: ../../include/openssl/opensslconf.h cmll_ctr.c
+cmll_ecb.o: ../../include/openssl/camellia.h
+cmll_ecb.o: ../../include/openssl/opensslconf.h cmll_ecb.c cmll_locl.h
+cmll_misc.o: ../../include/openssl/camellia.h
+cmll_misc.o: ../../include/openssl/opensslconf.h
+cmll_misc.o: ../../include/openssl/opensslv.h cmll_locl.h cmll_misc.c
+cmll_ofb.o: ../../include/openssl/camellia.h ../../include/openssl/modes.h
+cmll_ofb.o: ../../include/openssl/opensslconf.h cmll_ofb.c
diff --git a/crypto/camellia/asm/cmll-x86.pl b/crypto/camellia/asm/cmll-x86.pl
new file mode 100644
index 0000000000..027302ac86
--- /dev/null
+++ b/crypto/camellia/asm/cmll-x86.pl
@@ -0,0 +1,1138 @@
+#!/usr/bin/env perl
+
+# ====================================================================
+# Copyright (c) 2008 Andy Polyakov <appro@openssl.org>
+#
+# This module may be used under the terms of either the GNU General
+# Public License version 2 or later, the GNU Lesser General Public
+# License version 2.1 or later, the Mozilla Public License version
+# 1.1 or the BSD License. The exact terms of either license are
+# distributed along with this module. For further details see
+# http://www.openssl.org/~appro/camellia/.
+# ====================================================================
+
+# Performance in cycles per processed byte (less is better) in
+# 'openssl speed ...' benchmark:
+#
+#			AMD K8	Core2	PIII	P4
+# -evp camellia-128-ecb	21.5	22.8	27.0	28.9
+# + over gcc 3.4.6	+90/11% +70/10%	+53/4%	+160/64%
+# + over icc 8.0	+48/19% +21/15%	+21/17%	+55/37%
+#
+# camellia-128-cbc	17.3	21.1	23.9	25.9
+#
+# 128-bit key setup	196	280	256	240	cycles/key
+# + over gcc 3.4.6	+30/0%	+17/11%	+11/0%	+63/40%
+# + over icc 8.0	+18/3%	+10/0%	+10/3%	+21/10%
+#
+# Pairs of numbers in "+" rows represent performance improvement over
+# compiler generated position-independent code, PIC, and non-PIC
+# respectively. PIC results are of greater relevance, as this module
+# is position-independent, i.e. suitable for a shared library or PIE.
+# Position independence "costs" one register, which is why compilers
+# are so close with non-PIC results, they have an extra register to
+# spare. CBC results are better than ECB ones thanks to "zero-copy"
+# private _x86_* interface, and are ~30-40% better than with compiler
+# generated cmll_cbc.o, and reach ~80-90% of x86_64 performance on
+# same CPU (where applicable).
+
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+push(@INC,"${dir}","${dir}../../perlasm");
+require "x86asm.pl";
+
+$OPENSSL=1;
+
+&asm_init($ARGV[0],"cmll-586.pl",$ARGV[$#ARGV] eq "386");
+
+@T=("eax","ebx","ecx","edx");
+$idx="esi";
+$key="edi";
+$Tbl="ebp";
+
+# stack frame layout in _x86_Camellia_* routines, frame is allocated
+# by caller
+$__ra=&DWP(0,"esp");	# return address
+$__s0=&DWP(4,"esp");	# s0 backing store
+$__s1=&DWP(8,"esp");	# s1 backing store
+$__s2=&DWP(12,"esp");	# s2 backing store
+$__s3=&DWP(16,"esp");	# s3 backing store
+$__end=&DWP(20,"esp");	# pointer to end/start of key schedule
+
+# stack frame layout in Camellia_[en|crypt] routines, which differs from
+# above by 4 and overlaps by pointer to end/start of key schedule
+$_end=&DWP(16,"esp");
+$_esp=&DWP(20,"esp");
+
+# const unsigned int Camellia_SBOX[4][256];
+# Well, sort of... Camellia_SBOX[0][] is interleaved with [1][],
+# and [2][] - with [3][]. This is done to optimize code size.
+$SBOX1_1110=0;		# Camellia_SBOX[0]
+$SBOX4_4404=4;		# Camellia_SBOX[1]
+$SBOX2_0222=2048;	# Camellia_SBOX[2]
+$SBOX3_3033=2052;	# Camellia_SBOX[3]
+&static_label("Camellia_SIGMA");
+&static_label("Camellia_SBOX");
+
+sub Camellia_Feistel {
+my $i=@_[0];
+my $seed=defined(@_[1])?@_[1]:0;
+my $scale=$seed<0?-8:8;
+my $frame=defined(@_[2])?@_[2]:0;
+my $j=($i&1)*2;
+my $t0=@T[($j)%4],$t1=@T[($j+1)%4],$t2=@T[($j+2)%4],$t3=@T[($j+3)%4];
+
+	&xor	($t0,$idx);				# t0^=key[0]
+	&xor	($t1,&DWP($seed+$i*$scale+4,$key));	# t1^=key[1]
+	&movz	($idx,&HB($t0));			# (t0>>8)&0xff
+	&mov	($t3,&DWP($SBOX3_3033,$Tbl,$idx,8));	# t3=SBOX3_3033[0]
+	&movz	($idx,&LB($t0));			# (t0>>0)&0xff
+	&xor	($t3,&DWP($SBOX4_4404,$Tbl,$idx,8));	# t3^=SBOX4_4404[0]
+	&shr	($t0,16);
+	&movz	($idx,&LB($t1));			# (t1>>0)&0xff
+	&mov	($t2,&DWP($SBOX1_1110,$Tbl,$idx,8));	# t2=SBOX1_1110[1]
+	&movz	($idx,&HB($t0));			# (t0>>24)&0xff
+	&xor	($t3,&DWP($SBOX1_1110,$Tbl,$idx,8));	# t3^=SBOX1_1110[0]
+	&movz	($idx,&HB($t1));			# (t1>>8)&0xff
+	&xor	($t2,&DWP($SBOX4_4404,$Tbl,$idx,8));	# t2^=SBOX4_4404[1]
+	&shr	($t1,16);
+	&movz	($t0,&LB($t0));				# (t0>>16)&0xff
+	&xor	($t3,&DWP($SBOX2_0222,$Tbl,$t0,8));	# t3^=SBOX2_0222[0]
+	&movz	($idx,&HB($t1));			# (t1>>24)&0xff
+	&mov	($t0,&DWP($frame+4*(($j+3)%4),"esp"));	# prefetch "s3"
+	&xor	($t2,$t3);				# t2^=t3
+	&rotr	($t3,8);				# t3=RightRotate(t3,8)
+	&xor	($t2,&DWP($SBOX2_0222,$Tbl,$idx,8));	# t2^=SBOX2_0222[1]
+	&movz	($idx,&LB($t1));			# (t1>>16)&0xff
+	&mov	($t1,&DWP($frame+4*(($j+2)%4),"esp"));	# prefetch "s2"
+	&xor	($t3,$t0);				# t3^=s3
+	&xor	($t2,&DWP($SBOX3_3033,$Tbl,$idx,8));	# t2^=SBOX3_3033[1]
+	&mov	($idx,&DWP($seed+($i+1)*$scale,$key));	# prefetch key[i+1]
+	&xor	($t3,$t2);				# t3^=t2
+	&mov	(&DWP($frame+4*(($j+3)%4),"esp"),$t3);	# s3=t3
+	&xor	($t2,$t1);				# t2^=s2
+	&mov	(&DWP($frame+4*(($j+2)%4),"esp"),$t2);	# s2=t2
+}
+
+# void Camellia_EncryptBlock_Rounds(
+#		int grandRounds,
+#		const Byte plaintext[],
+#		const KEY_TABLE_TYPE keyTable,
+#		Byte ciphertext[])
+&function_begin("Camellia_EncryptBlock_Rounds");
+	&mov	("eax",&wparam(0));	# load grandRounds
+	&mov	($idx,&wparam(1));	# load plaintext pointer
+	&mov	($key,&wparam(2));	# load key schedule pointer
+
+	&mov	("ebx","esp");
+	&sub	("esp",7*4);		# place for s[0-3],keyEnd,esp and ra
+	&and	("esp",-64);
+
+	# place stack frame just "above mod 1024" the key schedule
+	# this ensures that cache associativity of 2 suffices
+	&lea	("ecx",&DWP(-64-63,$key));
+	&sub	("ecx","esp");
+	&neg	("ecx");
+	&and	("ecx",0x3C0);	# modulo 1024, but aligned to cache-line
+	&sub	("esp","ecx");
+	&add	("esp",4);	# 4 is reserved for callee's return address
+
+	&shl	("eax",6);
+	&lea	("eax",&DWP(0,$key,"eax"));
+	&mov	($_esp,"ebx");	# save %esp
+	&mov	($_end,"eax");	# save keyEnd
+
+	&call	(&label("pic_point"));
+	&set_label("pic_point");
+	&blindpop($Tbl);
+	&lea	($Tbl,&DWP(&label("Camellia_SBOX")."-".&label("pic_point"),$Tbl));
+
+	&mov	(@T[0],&DWP(0,$idx));	# load plaintext
+	&mov	(@T[1],&DWP(4,$idx));
+	&mov	(@T[2],&DWP(8,$idx));
+	&bswap	(@T[0]);
+	&mov	(@T[3],&DWP(12,$idx));
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+
+	&call	("_x86_Camellia_encrypt");
+
+	&mov	("esp",$_esp);
+	&bswap	(@T[0]);
+	&mov	($idx,&wparam(3));	# load ciphertext pointer
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+	&mov	(&DWP(0,$idx),@T[0]);	# write ciphertext
+	&mov	(&DWP(4,$idx),@T[1]);
+	&mov	(&DWP(8,$idx),@T[2]);
+	&mov	(&DWP(12,$idx),@T[3]);
+&function_end("Camellia_EncryptBlock_Rounds");
+# V1.x API
+&function_begin_B("Camellia_EncryptBlock");
+	&mov	("eax",128);
+	&sub	("eax",&wparam(0));	# load keyBitLength
+	&mov	("eax",3);
+	&adc	("eax",0);		# keyBitLength==128?3:4
+	&mov	(&wparam(0),"eax");
+	&jmp	(&label("Camellia_EncryptBlock_Rounds"));
+&function_end_B("Camellia_EncryptBlock");
+
+if ($OPENSSL) {
+# void Camellia_encrypt(
+#		const unsigned char *in,
+#		unsigned char *out,
+#		const CAMELLIA_KEY *key)
+&function_begin("Camellia_encrypt");
+	&mov	($idx,&wparam(0));	# load plaintext pointer
+	&mov	($key,&wparam(2));	# load key schedule pointer
+
+	&mov	("ebx","esp");
+	&sub	("esp",7*4);		# place for s[0-3],keyEnd,esp and ra
+	&and	("esp",-64);
+	&mov	("eax",&DWP(272,$key));	# load grandRounds counter
+
+	# place stack frame just "above mod 1024" the key schedule
+	# this ensures that cache associativity of 2 suffices
+	&lea	("ecx",&DWP(-64-63,$key));
+	&sub	("ecx","esp");
+	&neg	("ecx");
+	&and	("ecx",0x3C0);	# modulo 1024, but aligned to cache-line
+	&sub	("esp","ecx");
+	&add	("esp",4);	# 4 is reserved for callee's return address
+
+	&shl	("eax",6);
+	&lea	("eax",&DWP(0,$key,"eax"));
+	&mov	($_esp,"ebx");	# save %esp
+	&mov	($_end,"eax");	# save keyEnd
+
+	&call	(&label("pic_point"));
+	&set_label("pic_point");
+	&blindpop($Tbl);
+	&lea	($Tbl,&DWP(&label("Camellia_SBOX")."-".&label("pic_point"),$Tbl));
+
+	&mov	(@T[0],&DWP(0,$idx));	# load plaintext
+	&mov	(@T[1],&DWP(4,$idx));
+	&mov	(@T[2],&DWP(8,$idx));
+	&bswap	(@T[0]);
+	&mov	(@T[3],&DWP(12,$idx));
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+
+	&call	("_x86_Camellia_encrypt");
+
+	&mov	("esp",$_esp);
+	&bswap	(@T[0]);
+	&mov	($idx,&wparam(1));	# load ciphertext pointer
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+	&mov	(&DWP(0,$idx),@T[0]);	# write ciphertext
+	&mov	(&DWP(4,$idx),@T[1]);
+	&mov	(&DWP(8,$idx),@T[2]);
+	&mov	(&DWP(12,$idx),@T[3]);
+&function_end("Camellia_encrypt");
+}
+
+&function_begin_B("_x86_Camellia_encrypt");
+	&xor	(@T[0],&DWP(0,$key));	# ^=key[0-3]
+	&xor	(@T[1],&DWP(4,$key));
+	&xor	(@T[2],&DWP(8,$key));
+	&xor	(@T[3],&DWP(12,$key));
+	&mov	($idx,&DWP(16,$key));	# prefetch key[4]
+
+	&mov	($__s0,@T[0]);		# save s[0-3]
+	&mov	($__s1,@T[1]);
+	&mov	($__s2,@T[2]);
+	&mov	($__s3,@T[3]);
+
+&set_label("loop",16);
+	for ($i=0;$i<6;$i++) { Camellia_Feistel($i,16,4); }
+
+	&add	($key,16*4);
+	&cmp	($key,$__end);
+	&je	(&label("done"));
+
+	# @T[0-1] are preloaded, $idx is preloaded with key[0]
+	&and	($idx,@T[0]);
+	 &mov	 (@T[3],$__s3);
+	&rotl	($idx,1);
+	 &mov	 (@T[2],@T[3]);
+	&xor	(@T[1],$idx);
+	 &or	 (@T[2],&DWP(12,$key));
+	&mov	($__s1,@T[1]);		# s1^=LeftRotate(s0&key[0],1);
+	 &xor	 (@T[2],$__s2);
+
+	&mov	($idx,&DWP(4,$key));
+	 &mov	 ($__s2,@T[2]);		# s2^=s3|key[3];
+	&or	($idx,@T[1]);
+	 &and	 (@T[2],&DWP(8,$key));
+	&xor	(@T[0],$idx);
+	 &rotl	 (@T[2],1);
+	&mov	($__s0,@T[0]);		# s0^=s1|key[1];
+	 &xor	 (@T[3],@T[2]);
+	&mov	($idx,&DWP(16,$key));		# prefetch key[4]
+	 &mov	 ($__s3,@T[3]);		# s3^=LeftRotate(s2&key[2],1);
+	&jmp	(&label("loop"));
+
+&set_label("done",8);
+	&mov	(@T[2],@T[0]);		# SwapHalf
+	&mov	(@T[3],@T[1]);
+	&mov	(@T[0],$__s2);
+	&mov	(@T[1],$__s3);
+	&xor	(@T[0],$idx);		# $idx is preloaded with key[0]
+	&xor	(@T[1],&DWP(4,$key));
+	&xor	(@T[2],&DWP(8,$key));
+	&xor	(@T[3],&DWP(12,$key));
+	&ret	();
+&function_end_B("_x86_Camellia_encrypt");
+
+# void Camellia_DecryptBlock_Rounds(
+#		int grandRounds,
+#		const Byte ciphertext[],
+#		const KEY_TABLE_TYPE keyTable,
+#		Byte plaintext[])
+&function_begin("Camellia_DecryptBlock_Rounds");
+	&mov	("eax",&wparam(0));	# load grandRounds
+	&mov	($idx,&wparam(1));	# load ciphertext pointer
+	&mov	($key,&wparam(2));	# load key schedule pointer
+
+	&mov	("ebx","esp");
+	&sub	("esp",7*4);		# place for s[0-3],keyEnd,esp and ra
+	&and	("esp",-64);
+
+	# place stack frame just "above mod 1024" the key schedule
+	# this ensures that cache associativity of 2 suffices
+	&lea	("ecx",&DWP(-64-63,$key));
+	&sub	("ecx","esp");
+	&neg	("ecx");
+	&and	("ecx",0x3C0);	# modulo 1024, but aligned to cache-line
+	&sub	("esp","ecx");
+	&add	("esp",4);	# 4 is reserved for callee's return address
+
+	&shl	("eax",6);
+	&mov	(&DWP(4*4,"esp"),$key);	# save keyStart
+	&lea	($key,&DWP(0,$key,"eax"));
+	&mov	(&DWP(5*4,"esp"),"ebx");# save %esp
+
+	&call	(&label("pic_point"));
+	&set_label("pic_point");
+	&blindpop($Tbl);
+	&lea	($Tbl,&DWP(&label("Camellia_SBOX")."-".&label("pic_point"),$Tbl));
+
+	&mov	(@T[0],&DWP(0,$idx));	# load ciphertext
+	&mov	(@T[1],&DWP(4,$idx));
+	&mov	(@T[2],&DWP(8,$idx));
+	&bswap	(@T[0]);
+	&mov	(@T[3],&DWP(12,$idx));
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+
+	&call	("_x86_Camellia_decrypt");
+
+	&mov	("esp",&DWP(5*4,"esp"));
+	&bswap	(@T[0]);
+	&mov	($idx,&wparam(3));	# load plaintext pointer
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+	&mov	(&DWP(0,$idx),@T[0]);	# write plaintext
+	&mov	(&DWP(4,$idx),@T[1]);
+	&mov	(&DWP(8,$idx),@T[2]);
+	&mov	(&DWP(12,$idx),@T[3]);
+&function_end("Camellia_DecryptBlock_Rounds");
+# V1.x API
+&function_begin_B("Camellia_DecryptBlock");
+	&mov	("eax",128);
+	&sub	("eax",&wparam(0));	# load keyBitLength
+	&mov	("eax",3);
+	&adc	("eax",0);		# keyBitLength==128?3:4
+	&mov	(&wparam(0),"eax");
+	&jmp	(&label("Camellia_DecryptBlock_Rounds"));
+&function_end_B("Camellia_DecryptBlock");
+
+if ($OPENSSL) {
+# void Camellia_decrypt(
+#		const unsigned char *in,
+#		unsigned char *out,
+#		const CAMELLIA_KEY *key)
+&function_begin("Camellia_decrypt");
+	&mov	($idx,&wparam(0));	# load ciphertext pointer
+	&mov	($key,&wparam(2));	# load key schedule pointer
+
+	&mov	("ebx","esp");
+	&sub	("esp",7*4);		# place for s[0-3],keyEnd,esp and ra
+	&and	("esp",-64);
+	&mov	("eax",&DWP(272,$key));	# load grandRounds counter
+
+	# place stack frame just "above mod 1024" the key schedule
+	# this ensures that cache associativity of 2 suffices
+	&lea	("ecx",&DWP(-64-63,$key));
+	&sub	("ecx","esp");
+	&neg	("ecx");
+	&and	("ecx",0x3C0);	# modulo 1024, but aligned to cache-line
+	&sub	("esp","ecx");
+	&add	("esp",4);	# 4 is reserved for callee's return address
+
+	&shl	("eax",6);
+	&mov	(&DWP(4*4,"esp"),$key);	# save keyStart
+	&lea	($key,&DWP(0,$key,"eax"));
+	&mov	(&DWP(5*4,"esp"),"ebx");# save %esp
+
+	&call	(&label("pic_point"));
+	&set_label("pic_point");
+	&blindpop($Tbl);
+	&lea	($Tbl,&DWP(&label("Camellia_SBOX")."-".&label("pic_point"),$Tbl));
+
+	&mov	(@T[0],&DWP(0,$idx));	# load ciphertext
+	&mov	(@T[1],&DWP(4,$idx));
+	&mov	(@T[2],&DWP(8,$idx));
+	&bswap	(@T[0]);
+	&mov	(@T[3],&DWP(12,$idx));
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+
+	&call	("_x86_Camellia_decrypt");
+
+	&mov	("esp",&DWP(5*4,"esp"));
+	&bswap	(@T[0]);
+	&mov	($idx,&wparam(1));	# load plaintext pointer
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+	&mov	(&DWP(0,$idx),@T[0]);	# write plaintext
+	&mov	(&DWP(4,$idx),@T[1]);
+	&mov	(&DWP(8,$idx),@T[2]);
+	&mov	(&DWP(12,$idx),@T[3]);
+&function_end("Camellia_decrypt");
+}
+
+&function_begin_B("_x86_Camellia_decrypt");
+	&xor	(@T[0],&DWP(0,$key));	# ^=key[0-3]
+	&xor	(@T[1],&DWP(4,$key));
+	&xor	(@T[2],&DWP(8,$key));
+	&xor	(@T[3],&DWP(12,$key));
+	&mov	($idx,&DWP(-8,$key));	# prefetch key[-2]
+
+	&mov	($__s0,@T[0]);		# save s[0-3]
+	&mov	($__s1,@T[1]);
+	&mov	($__s2,@T[2]);
+	&mov	($__s3,@T[3]);
+
+&set_label("loop",16);
+	for ($i=0;$i<6;$i++) { Camellia_Feistel($i,-8,4); }
+
+	&sub	($key,16*4);
+	&cmp	($key,$__end);
+	&je	(&label("done"));
+
+	# @T[0-1] are preloaded, $idx is preloaded with key[2]
+	&and	($idx,@T[0]);
+	 &mov	 (@T[3],$__s3);
+	&rotl	($idx,1);
+	 &mov	 (@T[2],@T[3]);
+	&xor	(@T[1],$idx);
+	 &or	 (@T[2],&DWP(4,$key));
+	&mov	($__s1,@T[1]);		# s1^=LeftRotate(s0&key[0],1);
+	 &xor	 (@T[2],$__s2);
+
+	&mov	($idx,&DWP(12,$key));
+	 &mov	 ($__s2,@T[2]);		# s2^=s3|key[3];
+	&or	($idx,@T[1]);
+	 &and	 (@T[2],&DWP(0,$key));
+	&xor	(@T[0],$idx);
+	 &rotl	 (@T[2],1);
+	&mov	($__s0,@T[0]);		# s0^=s1|key[1];
+	 &xor	 (@T[3],@T[2]);
+	&mov	($idx,&DWP(-8,$key));	# prefetch key[4]
+	 &mov	 ($__s3,@T[3]);		# s3^=LeftRotate(s2&key[2],1);
+	&jmp	(&label("loop"));
+
+&set_label("done",8);
+	&mov	(@T[2],@T[0]);		# SwapHalf
+	&mov	(@T[3],@T[1]);
+	&mov	(@T[0],$__s2);
+	&mov	(@T[1],$__s3);
+	&xor	(@T[2],$idx);		# $idx is preloaded with key[2]
+	&xor	(@T[3],&DWP(12,$key));
+	&xor	(@T[0],&DWP(0,$key));
+	&xor	(@T[1],&DWP(4,$key));
+	&ret	();
+&function_end_B("_x86_Camellia_decrypt");
+
+# shld is very slow on Intel P4 family. Even on AMD it limits
+# instruction decode rate [because it's VectorPath] and consequently
+# performance. PIII, PM and Core[2] seem to be the only ones which
+# execute this code ~7% faster...
+sub __rotl128 {
+  my ($i0,$i1,$i2,$i3,$rot,$rnd,@T)=@_;
+
+    $rnd *= 2;
+    if ($rot) {
+	&mov	($idx,$i0);
+	&shld	($i0,$i1,$rot);
+	&shld	($i1,$i2,$rot);
+	&shld	($i2,$i3,$rot);
+	&shld	($i3,$idx,$rot);
+    }
+    &mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i0 eq @T[0]);
+    &mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i1 eq @T[0]);
+    &mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i2 eq @T[0]);
+    &mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i3 eq @T[0]);
+}
+
+# ... Implementing 128-bit rotate without shld gives >3x performance
+# improvement on P4, only ~7% degradation on other Intel CPUs and
+# not worse performance on AMD. This is therefore preferred.
+sub _rotl128 {
+  my ($i0,$i1,$i2,$i3,$rot,$rnd,@T)=@_;
+
+    $rnd *= 2;
+    if ($rot) {
+	&mov	($Tbl,$i0);
+	&shl	($i0,$rot);
+	&mov	($idx,$i1);
+	&shr	($idx,32-$rot);
+	&shl	($i1,$rot);
+	&or	($i0,$idx);
+	&mov	($idx,$i2);
+	&shl	($i2,$rot);
+	&mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i0 eq @T[0]);
+	&shr	($idx,32-$rot);
+	&or	($i1,$idx);
+	&shr	($Tbl,32-$rot);
+	&mov	($idx,$i3);
+	&shr	($idx,32-$rot);
+	&mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i1 eq @T[0]);
+	&shl	($i3,$rot);
+	&or	($i2,$idx);
+	&or	($i3,$Tbl);
+	&mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i2 eq @T[0]);
+	&mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i3 eq @T[0]);
+    } else {
+	&mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i0 eq @T[0]);
+	&mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i1 eq @T[0]);
+	&mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i2 eq @T[0]);
+	&mov	(&DWP(-128+4*$rnd++,$key),shift(@T))	if ($i3 eq @T[0]);
+    }
+}
+
+sub _saveround {
+my ($rnd,$key,@T)=@_;
+my $bias=int(@T[0])?shift(@T):0;
+
+	&mov	(&DWP($bias+$rnd*8+0,$key),@T[0]);
+	&mov	(&DWP($bias+$rnd*8+4,$key),@T[1])	if ($#T>=1);
+	&mov	(&DWP($bias+$rnd*8+8,$key),@T[2])	if ($#T>=2);
+	&mov	(&DWP($bias+$rnd*8+12,$key),@T[3])	if ($#T>=3);
+}
+
+sub _loadround {
+my ($rnd,$key,@T)=@_;
+my $bias=int(@T[0])?shift(@T):0;
+
+	&mov	(@T[0],&DWP($bias+$rnd*8+0,$key));
+	&mov	(@T[1],&DWP($bias+$rnd*8+4,$key))	if ($#T>=1);
+	&mov	(@T[2],&DWP($bias+$rnd*8+8,$key))	if ($#T>=2);
+	&mov	(@T[3],&DWP($bias+$rnd*8+12,$key))	if ($#T>=3);
+}
+
+# void Camellia_Ekeygen(
+#		const int keyBitLength,
+#		const Byte *rawKey,
+#		KEY_TABLE_TYPE keyTable)
+&function_begin("Camellia_Ekeygen");
+{ my $step=0;
+
+	&stack_push(4);				# place for s[0-3]
+
+	&mov	($Tbl,&wparam(0));		# load arguments
+	&mov	($idx,&wparam(1));
+	&mov	($key,&wparam(2));
+
+	&mov	(@T[0],&DWP(0,$idx));		# load 0-127 bits
+	&mov	(@T[1],&DWP(4,$idx));
+	&mov	(@T[2],&DWP(8,$idx));
+	&mov	(@T[3],&DWP(12,$idx));
+
+	&bswap	(@T[0]);
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+
+	&_saveround	(0,$key,@T);		# KL<<<0
+
+	&cmp	($Tbl,128);
+	&je	(&label("1st128"));
+
+	&mov	(@T[0],&DWP(16,$idx));		# load 128-191 bits
+	&mov	(@T[1],&DWP(20,$idx));
+	&cmp	($Tbl,192);
+	&je	(&label("1st192"));
+	&mov	(@T[2],&DWP(24,$idx));		# load 192-255 bits
+	&mov	(@T[3],&DWP(28,$idx));
+	&jmp	(&label("1st256"));
+&set_label("1st192",4);
+	&mov	(@T[2],@T[0]);
+	&mov	(@T[3],@T[1]);
+	&not	(@T[2]);
+	&not	(@T[3]);
+&set_label("1st256",4);
+	&bswap	(@T[0]);
+	&bswap	(@T[1]);
+	&bswap	(@T[2]);
+	&bswap	(@T[3]);
+
+	&_saveround	(4,$key,@T);		# temporary storage for KR!
+
+	&xor	(@T[0],&DWP(0*8+0,$key));	# KR^KL
+	&xor	(@T[1],&DWP(0*8+4,$key));
+	&xor	(@T[2],&DWP(1*8+0,$key));
+	&xor	(@T[3],&DWP(1*8+4,$key));
+
+&set_label("1st128",4);
+	&call	(&label("pic_point"));
+	&set_label("pic_point");
+	&blindpop($Tbl);
+	&lea	($Tbl,&DWP(&label("Camellia_SBOX")."-".&label("pic_point"),$Tbl));
+	&lea	($key,&DWP(&label("Camellia_SIGMA")."-".&label("Camellia_SBOX"),$Tbl));
+
+	&mov	($idx,&DWP($step*8,$key));	# prefetch SIGMA[0]
+	&mov	(&swtmp(0),@T[0]);		# save s[0-3]
+	&mov	(&swtmp(1),@T[1]);
+	&mov	(&swtmp(2),@T[2]);
+	&mov	(&swtmp(3),@T[3]);
+	&Camellia_Feistel($step++);
+	&Camellia_Feistel($step++);
+	&mov	(@T[2],&swtmp(2));
+	&mov	(@T[3],&swtmp(3));
+
+	&mov	($idx,&wparam(2));
+	&xor	(@T[0],&DWP(0*8+0,$idx));	# ^KL
+	&xor	(@T[1],&DWP(0*8+4,$idx));
+	&xor	(@T[2],&DWP(1*8+0,$idx));
+	&xor	(@T[3],&DWP(1*8+4,$idx));
+
+	&mov	($idx,&DWP($step*8,$key));	# prefetch SIGMA[4]
+	&mov	(&swtmp(0),@T[0]);		# save s[0-3]
+	&mov	(&swtmp(1),@T[1]);
+	&mov	(&swtmp(2),@T[2]);
+	&mov	(&swtmp(3),@T[3]);
+	&Camellia_Feistel($step++);
+	&Camellia_Feistel($step++);
+	&mov	(@T[2],&swtmp(2));
+	&mov	(@T[3],&swtmp(3));
+
+	&mov	($idx,&wparam(0));
+	&cmp	($idx,128);
+	&jne	(&label("2nd256"));
+
+	&mov	($key,&wparam(2));
+	&lea	($key,&DWP(128,$key));		# size optimization
+
+	####### process KA
+	&_saveround	(2,$key,-128,@T);	# KA<<<0
+	&_rotl128	(@T,15,6,@T);		# KA<<<15
+	&_rotl128	(@T,15,8,@T);		# KA<<<(15+15=30)
+	&_rotl128	(@T,15,12,@T[0],@T[1]);	# KA<<<(30+15=45)
+	&_rotl128	(@T,15,14,@T);		# KA<<<(45+15=60)
+	push		(@T,shift(@T));		# rotl128(@T,32);
+	&_rotl128	(@T,2,20,@T);		# KA<<<(60+32+2=94)
+	&_rotl128	(@T,17,24,@T);		# KA<<<(94+17=111)
+
+	####### process KL
+	&_loadround	(0,$key,-128,@T);	# load KL
+	&_rotl128	(@T,15,4,@T);		# KL<<<15
+	&_rotl128	(@T,30,10,@T);		# KL<<<(15+30=45)
+	&_rotl128	(@T,15,13,@T[2],@T[3]);	# KL<<<(45+15=60)
+	&_rotl128	(@T,17,16,@T);		# KL<<<(60+17=77)
+	&_rotl128	(@T,17,18,@T);		# KL<<<(77+17=94)
+	&_rotl128	(@T,17,22,@T);		# KL<<<(94+17=111)
+
+	while (@T[0] ne "eax")			# restore order
+	{   unshift	(@T,pop(@T));   }
+
+	&mov	("eax",3);			# 3 grandRounds
+	&jmp	(&label("done"));
+
+&set_label("2nd256",16);
+	&mov	($idx,&wparam(2));
+	&_saveround	(6,$idx,@T);		# temporary storage for KA!
+
+	&xor	(@T[0],&DWP(4*8+0,$idx));	# KA^KR
+	&xor	(@T[1],&DWP(4*8+4,$idx));
+	&xor	(@T[2],&DWP(5*8+0,$idx));
+	&xor	(@T[3],&DWP(5*8+4,$idx));
+
+	&mov	($idx,&DWP($step*8,$key));	# prefetch SIGMA[8]
+	&mov	(&swtmp(0),@T[0]);		# save s[0-3]
+	&mov	(&swtmp(1),@T[1]);
+	&mov	(&swtmp(2),@T[2]);
+	&mov	(&swtmp(3),@T[3]);
+	&Camellia_Feistel($step++);
+	&Camellia_Feistel($step++);
+	&mov	(@T[2],&swtmp(2));
+	&mov	(@T[3],&swtmp(3));
+
+	&mov	($key,&wparam(2));
+	&lea	($key,&DWP(128,$key));		# size optimization
+
+	####### process KB
+	&_saveround	(2,$key,-128,@T);	# KB<<<0
+	&_rotl128	(@T,30,10,@T);		# KB<<<30
+	&_rotl128	(@T,30,20,@T);		# KB<<<(30+30=60)
+	push		(@T,shift(@T));		# rotl128(@T,32);
+	&_rotl128	(@T,19,32,@T);		# KB<<<(60+32+19=111)
+
+	####### process KR
+	&_loadround	(4,$key,-128,@T);	# load KR
+	&_rotl128	(@T,15,4,@T);		# KR<<<15
+	&_rotl128	(@T,15,8,@T);		# KR<<<(15+15=30)
+	&_rotl128	(@T,30,18,@T);		# KR<<<(30+30=60)
+	push		(@T,shift(@T));		# rotl128(@T,32);
+	&_rotl128	(@T,2,26,@T);		# KR<<<(60+32+2=94)
+
+	####### process KA
+	&_loadround	(6,$key,-128,@T);	# load KA
+	&_rotl128	(@T,15,6,@T);		# KA<<<15
+	&_rotl128	(@T,30,14,@T);		# KA<<<(15+30=45)
+	push		(@T,shift(@T));		# rotl128(@T,32);
+	&_rotl128	(@T,0,24,@T);		# KA<<<(45+32+0=77)
+	&_rotl128	(@T,17,28,@T);		# KA<<<(77+17=94)
+
+	####### process KL
+	&_loadround	(0,$key,-128,@T);	# load KL
+	push		(@T,shift(@T));		# rotl128(@T,32);
+	&_rotl128	(@T,13,12,@T);		# KL<<<(32+13=45)
+	&_rotl128	(@T,15,16,@T);		# KL<<<(45+15=60)
+	&_rotl128	(@T,17,22,@T);		# KL<<<(60+17=77)
+	push		(@T,shift(@T));		# rotl128(@T,32);
+	&_rotl128	(@T,2,30,@T);		# KL<<<(77+32+2=111)
+
+	while (@T[0] ne "eax")			# restore order
+	{   unshift	(@T,pop(@T));   }
+
+	&mov	("eax",4);			# 4 grandRounds
+&set_label("done");
+	&lea	("edx",&DWP(272-128,$key));	# end of key schedule
+	&stack_pop(4);
+}
+&function_end("Camellia_Ekeygen");
+
+if ($OPENSSL) {
+# int Camellia_set_key (
+#		const unsigned char *userKey,
+#		int bits,
+#		CAMELLIA_KEY *key)
+&function_begin_B("Camellia_set_key");
+	&push	("ebx");
+	&mov	("ecx",&wparam(0));	# pull arguments
+	&mov	("ebx",&wparam(1));
+	&mov	("edx",&wparam(2));
+
+	&mov	("eax",-1);
+	&test	("ecx","ecx");
+	&jz	(&label("done"));	# userKey==NULL?
+	&test	("edx","edx");
+	&jz	(&label("done"));	# key==NULL?
+
+	&mov	("eax",-2);
+	&cmp	("ebx",256);
+	&je	(&label("arg_ok"));	# bits==256?
+	&cmp	("ebx",192);
+	&je	(&label("arg_ok"));	# bits==192?
+	&cmp	("ebx",128);
+	&jne	(&label("done"));	# bits!=128?
+&set_label("arg_ok",4);
+
+	&push	("edx");		# push arguments
+	&push	("ecx");
+	&push	("ebx");
+	&call	("Camellia_Ekeygen");
+	&stack_pop(3);
+
+	# eax holds grandRounds and edx points at where to put it
+	&mov	(&DWP(0,"edx"),"eax");
+	&xor	("eax","eax");
+&set_label("done",4);
+	&pop	("ebx");
+	&ret	();
+&function_end_B("Camellia_set_key");
+}
+
+@SBOX=(
+112,130, 44,236,179, 39,192,229,228,133, 87, 53,234, 12,174, 65,
+ 35,239,107,147, 69, 25,165, 33,237, 14, 79, 78, 29,101,146,189,
+134,184,175,143,124,235, 31,206, 62, 48,220, 95, 94,197, 11, 26,
+166,225, 57,202,213, 71, 93, 61,217,  1, 90,214, 81, 86,108, 77,
+139, 13,154,102,251,204,176, 45,116, 18, 43, 32,240,177,132,153,
+223, 76,203,194, 52,126,118,  5,109,183,169, 49,209, 23,  4,215,
+ 20, 88, 58, 97,222, 27, 17, 28, 50, 15,156, 22, 83, 24,242, 34,
+254, 68,207,178,195,181,122,145, 36,  8,232,168, 96,252,105, 80,
+170,208,160,125,161,137, 98,151, 84, 91, 30,149,224,255,100,210,
+ 16,196,  0, 72,163,247,117,219,138,  3,230,218,  9, 63,221,148,
+135, 92,131,  2,205, 74,144, 51,115,103,246,243,157,127,191,226,
+ 82,155,216, 38,200, 55,198, 59,129,150,111, 75, 19,190, 99, 46,
+233,121,167,140,159,110,188,142, 41,245,249,182, 47,253,180, 89,
+120,152,  6,106,231, 70,113,186,212, 37,171, 66,136,162,141,250,
+114,  7,185, 85,248,238,172, 10, 54, 73, 42,104, 60, 56,241,164,
+ 64, 40,211,123,187,201, 67,193, 21,227,173,244,119,199,128,158);
+
+sub S1110 { my $i=shift; $i=@SBOX[$i]; return $i<<24|$i<<16|$i<<8; }
+sub S4404 { my $i=shift; $i=($i<<1|$i>>7)&0xff; $i=@SBOX[$i]; return $i<<24|$i<<16|$i; }	
+sub S0222 { my $i=shift; $i=@SBOX[$i]; $i=($i<<1|$i>>7)&0xff; return $i<<16|$i<<8|$i; }	
+sub S3033 { my $i=shift; $i=@SBOX[$i]; $i=($i>>1|$i<<7)&0xff; return $i<<24|$i<<8|$i; }	
+
+&set_label("Camellia_SIGMA",64);
+&data_word(
+    0xa09e667f, 0x3bcc908b, 0xb67ae858, 0x4caa73b2,
+    0xc6ef372f, 0xe94f82be, 0x54ff53a5, 0xf1d36f1c,
+    0x10e527fa, 0xde682d1d, 0xb05688c2, 0xb3e6c1fd,
+    0,          0,          0,          0);
+&set_label("Camellia_SBOX",64);
+# tables are interleaved, remember?
+for ($i=0;$i<256;$i++) { &data_word(&S1110($i),&S4404($i)); }
+for ($i=0;$i<256;$i++) { &data_word(&S0222($i),&S3033($i)); }
+
+# void Camellia_cbc_encrypt (const void char *inp, unsigned char *out,
+#			size_t length, const CAMELLIA_KEY *key,
+#			unsigned char *ivp,const int enc);
+{
+# stack frame layout
+#             -4(%esp)		# return address	 0(%esp)
+#              0(%esp)		# s0			 4(%esp)
+#              4(%esp)		# s1			 8(%esp)
+#              8(%esp)		# s2			12(%esp)
+#             12(%esp)		# s3			16(%esp)
+#             16(%esp)		# end of key schedule	20(%esp)
+#             20(%esp)		# %esp backup
+my $_inp=&DWP(24,"esp");	#copy of wparam(0)
+my $_out=&DWP(28,"esp");	#copy of wparam(1)
+my $_len=&DWP(32,"esp");	#copy of wparam(2)
+my $_key=&DWP(36,"esp");	#copy of wparam(3)
+my $_ivp=&DWP(40,"esp");	#copy of wparam(4)
+my $ivec=&DWP(44,"esp");	#ivec[16]
+my $_tmp=&DWP(44,"esp");	#volatile variable [yes, aliases with ivec]
+my ($s0,$s1,$s2,$s3) = @T;
+
+&function_begin("Camellia_cbc_encrypt");
+	&mov	($s2 eq "ecx"? $s2 : "",&wparam(2));	# load len
+	&cmp	($s2,0);
+	&je	(&label("enc_out"));
+
+	&pushf	();
+	&cld	();
+
+	&mov	($s0,&wparam(0));	# load inp
+	&mov	($s1,&wparam(1));	# load out
+	#&mov	($s2,&wparam(2));	# load len
+	&mov	($s3,&wparam(3));	# load key
+	&mov	($Tbl,&wparam(4));	# load ivp
+
+	# allocate aligned stack frame...
+	&lea	($idx,&DWP(-64,"esp"));
+	&and	($idx,-64);
+
+	# place stack frame just "above mod 1024" the key schedule
+	# this ensures that cache associativity of 2 suffices
+	&lea	($key,&DWP(-64-63,$s3));
+	&sub	($key,$idx);
+	&neg	($key);
+	&and	($key,0x3C0);	# modulo 1024, but aligned to cache-line
+	&sub	($idx,$key);
+
+	&mov	($key,&wparam(5));	# load enc
+
+	&exch	("esp",$idx);
+	&add	("esp",4);		# reserve for return address!
+	&mov	($_esp,$idx);		# save %esp
+
+	&mov	($_inp,$s0);		# save copy of inp
+	&mov	($_out,$s1);		# save copy of out
+	&mov	($_len,$s2);		# save copy of len
+	&mov	($_key,$s3);		# save copy of key
+	&mov	($_ivp,$Tbl);		# save copy of ivp
+
+	&call   (&label("pic_point"));	# make it PIC!
+	&set_label("pic_point");
+	&blindpop($Tbl);
+	&lea    ($Tbl,&DWP(&label("Camellia_SBOX")."-".&label("pic_point"),$Tbl));
+
+	&mov	($idx,32);
+	&set_label("prefetch_sbox",4);
+		&mov	($s0,&DWP(0,$Tbl));
+		&mov	($s1,&DWP(32,$Tbl));
+		&mov	($s2,&DWP(64,$Tbl));
+		&mov	($s3,&DWP(96,$Tbl));
+		&lea	($Tbl,&DWP(128,$Tbl));
+		&dec	($idx);
+	&jnz	(&label("prefetch_sbox"));
+	&mov	($s0,$_key);
+	&sub	($Tbl,4096);
+	&mov	($idx,$_inp);
+	&mov	($s3,&DWP(272,$s0));		# load grandRounds
+
+	&cmp	($key,0);
+	&je	(&label("DECRYPT"));
+
+	&mov	($s2,$_len);
+	&mov	($key,$_ivp);
+	&shl	($s3,6);
+	&lea	($s3,&DWP(0,$s0,$s3));
+	&mov	($_end,$s3);
+
+	&test	($s2,0xFFFFFFF0);
+	&jz	(&label("enc_tail"));		# short input...
+
+	&mov	($s0,&DWP(0,$key));		# load iv
+	&mov	($s1,&DWP(4,$key));
+
+	&set_label("enc_loop",4);
+		&mov	($s2,&DWP(8,$key));
+		&mov	($s3,&DWP(12,$key));
+
+		&xor	($s0,&DWP(0,$idx));	# xor input data
+		&xor	($s1,&DWP(4,$idx));
+		&xor	($s2,&DWP(8,$idx));
+		&bswap	($s0);
+		&xor	($s3,&DWP(12,$idx));
+		&bswap	($s1);
+		&mov	($key,$_key);		# load key
+		&bswap	($s2);
+		&bswap	($s3);
+
+		&call	("_x86_Camellia_encrypt");
+
+		&mov	($idx,$_inp);		# load inp
+		&mov	($key,$_out);		# load out
+
+		&bswap	($s0);
+		&bswap	($s1);
+		&bswap	($s2);
+		&mov	(&DWP(0,$key),$s0);	# save output data
+		&bswap	($s3);
+		&mov	(&DWP(4,$key),$s1);
+		&mov	(&DWP(8,$key),$s2);
+		&mov	(&DWP(12,$key),$s3);
+
+		&mov	($s2,$_len);		# load len
+
+		&lea	($idx,&DWP(16,$idx));
+		&mov	($_inp,$idx);		# save inp
+
+		&lea	($s3,&DWP(16,$key));
+		&mov	($_out,$s3);		# save out
+
+		&sub	($s2,16);
+		&test	($s2,0xFFFFFFF0);
+		&mov	($_len,$s2);		# save len
+	&jnz	(&label("enc_loop"));
+	&test	($s2,15);
+	&jnz	(&label("enc_tail"));
+	&mov	($idx,$_ivp);		# load ivp
+	&mov	($s2,&DWP(8,$key));	# restore last dwords
+	&mov	($s3,&DWP(12,$key));
+	&mov	(&DWP(0,$idx),$s0);	# save ivec
+	&mov	(&DWP(4,$idx),$s1);
+	&mov	(&DWP(8,$idx),$s2);
+	&mov	(&DWP(12,$idx),$s3);
+
+	&mov	("esp",$_esp);
+	&popf	();
+    &set_label("enc_out");
+	&function_end_A();
+	&pushf	();			# kludge, never executed
+
+    &set_label("enc_tail",4);
+	&mov	($s0,$key eq "edi" ? $key : "");
+	&mov	($key,$_out);			# load out
+	&push	($s0);				# push ivp
+	&mov	($s1,16);
+	&sub	($s1,$s2);
+	&cmp	($key,$idx);			# compare with inp
+	&je	(&label("enc_in_place"));
+	&align	(4);
+	&data_word(0xA4F3F689);	# rep movsb	# copy input
+	&jmp	(&label("enc_skip_in_place"));
+    &set_label("enc_in_place");
+	&lea	($key,&DWP(0,$key,$s2));
+    &set_label("enc_skip_in_place");
+	&mov	($s2,$s1);
+	&xor	($s0,$s0);
+	&align	(4);
+	&data_word(0xAAF3F689);	# rep stosb	# zero tail
+	&pop	($key);				# pop ivp
+
+	&mov	($idx,$_out);			# output as input
+	&mov	($s0,&DWP(0,$key));
+	&mov	($s1,&DWP(4,$key));
+	&mov	($_len,16);			# len=16
+	&jmp	(&label("enc_loop"));		# one more spin...
+
+#----------------------------- DECRYPT -----------------------------#
+&set_label("DECRYPT",16);
+	&shl	($s3,6);
+	&lea	($s3,&DWP(0,$s0,$s3));
+	&mov	($_end,$s0);
+	&mov	($_key,$s3);
+
+	&cmp	($idx,$_out);
+	&je	(&label("dec_in_place"));	# in-place processing...
+
+	&mov	($key,$_ivp);			# load ivp
+	&mov	($_tmp,$key);
+
+	&set_label("dec_loop",4);
+		&mov	($s0,&DWP(0,$idx));	# read input
+		&mov	($s1,&DWP(4,$idx));
+		&mov	($s2,&DWP(8,$idx));
+		&bswap	($s0);
+		&mov	($s3,&DWP(12,$idx));
+		&bswap	($s1);
+		&mov	($key,$_key);		# load key
+		&bswap	($s2);
+		&bswap	($s3);
+
+		&call	("_x86_Camellia_decrypt");
+
+		&mov	($key,$_tmp);		# load ivp
+		&mov	($idx,$_len);		# load len
+
+		&bswap	($s0);
+		&bswap	($s1);
+		&bswap	($s2);
+		&xor	($s0,&DWP(0,$key));	# xor iv
+		&bswap	($s3);
+		&xor	($s1,&DWP(4,$key));
+		&xor	($s2,&DWP(8,$key));
+		&xor	($s3,&DWP(12,$key));
+
+		&sub	($idx,16);
+		&jc	(&label("dec_partial"));
+		&mov	($_len,$idx);		# save len
+		&mov	($idx,$_inp);		# load inp
+		&mov	($key,$_out);		# load out
+
+		&mov	(&DWP(0,$key),$s0);	# write output
+		&mov	(&DWP(4,$key),$s1);
+		&mov	(&DWP(8,$key),$s2);
+		&mov	(&DWP(12,$key),$s3);
+
+		&mov	($_tmp,$idx);		# save ivp
+		&lea	($idx,&DWP(16,$idx));
+		&mov	($_inp,$idx);		# save inp
+
+		&lea	($key,&DWP(16,$key));
+		&mov	($_out,$key);		# save out
+
+	&jnz	(&label("dec_loop"));
+	&mov	($key,$_tmp);		# load temp ivp
+    &set_label("dec_end");
+	&mov	($idx,$_ivp);		# load user ivp
+	&mov	($s0,&DWP(0,$key));	# load iv
+	&mov	($s1,&DWP(4,$key));
+	&mov	($s2,&DWP(8,$key));
+	&mov	($s3,&DWP(12,$key));
+	&mov	(&DWP(0,$idx),$s0);	# copy back to user
+	&mov	(&DWP(4,$idx),$s1);
+	&mov	(&DWP(8,$idx),$s2);
+	&mov	(&DWP(12,$idx),$s3);
+	&jmp	(&label("dec_out"));
+
+    &set_label("dec_partial",4);
+	&lea	($key,$ivec);
+	&mov	(&DWP(0,$key),$s0);	# dump output to stack
+	&mov	(&DWP(4,$key),$s1);
+	&mov	(&DWP(8,$key),$s2);
+	&mov	(&DWP(12,$key),$s3);
+	&lea	($s2 eq "ecx" ? $s2 : "",&DWP(16,$idx));
+	&mov	($idx eq "esi" ? $idx : "",$key);
+	&mov	($key eq "edi" ? $key : "",$_out);	# load out
+	&data_word(0xA4F3F689);	# rep movsb		# copy output
+	&mov	($key,$_inp);				# use inp as temp ivp
+	&jmp	(&label("dec_end"));
+
+    &set_label("dec_in_place",4);
+	&set_label("dec_in_place_loop");
+		&lea	($key,$ivec);
+		&mov	($s0,&DWP(0,$idx));	# read input
+		&mov	($s1,&DWP(4,$idx));
+		&mov	($s2,&DWP(8,$idx));
+		&mov	($s3,&DWP(12,$idx));
+
+		&mov	(&DWP(0,$key),$s0);	# copy to temp
+		&mov	(&DWP(4,$key),$s1);
+		&mov	(&DWP(8,$key),$s2);
+		&bswap	($s0);
+		&mov	(&DWP(12,$key),$s3);
+		&bswap	($s1);
+		&mov	($key,$_key);		# load key
+		&bswap	($s2);
+		&bswap	($s3);
+
+		&call	("_x86_Camellia_decrypt");
+
+		&mov	($key,$_ivp);		# load ivp
+		&mov	($idx,$_out);		# load out
+
+		&bswap	($s0);
+		&bswap	($s1);
+		&bswap	($s2);
+		&xor	($s0,&DWP(0,$key));	# xor iv
+		&bswap	($s3);
+		&xor	($s1,&DWP(4,$key));
+		&xor	($s2,&DWP(8,$key));
+		&xor	($s3,&DWP(12,$key));
+
+		&mov	(&DWP(0,$idx),$s0);	# write output
+		&mov	(&DWP(4,$idx),$s1);
+		&mov	(&DWP(8,$idx),$s2);
+		&mov	(&DWP(12,$idx),$s3);
+
+		&lea	($idx,&DWP(16,$idx));
+		&mov	($_out,$idx);		# save out
+
+		&lea	($idx,$ivec);
+		&mov	($s0,&DWP(0,$idx));	# read temp
+		&mov	($s1,&DWP(4,$idx));
+		&mov	($s2,&DWP(8,$idx));
+		&mov	($s3,&DWP(12,$idx));
+
+		&mov	(&DWP(0,$key),$s0);	# copy iv
+		&mov	(&DWP(4,$key),$s1);
+		&mov	(&DWP(8,$key),$s2);
+		&mov	(&DWP(12,$key),$s3);
+
+		&mov	($idx,$_inp);		# load inp
+
+		&lea	($idx,&DWP(16,$idx));
+		&mov	($_inp,$idx);		# save inp
+
+		&mov	($s2,$_len);		# load len
+		&sub	($s2,16);
+		&jc	(&label("dec_in_place_partial"));
+		&mov	($_len,$s2);		# save len
+	&jnz	(&label("dec_in_place_loop"));
+	&jmp	(&label("dec_out"));
+
+    &set_label("dec_in_place_partial",4);
+	# one can argue if this is actually required...
+	&mov	($key eq "edi" ? $key : "",$_out);
+	&lea	($idx eq "esi" ? $idx : "",$ivec);
+	&lea	($key,&DWP(0,$key,$s2));
+	&lea	($idx,&DWP(16,$idx,$s2));
+	&neg	($s2 eq "ecx" ? $s2 : "");
+	&data_word(0xA4F3F689);	# rep movsb	# restore tail
+
+    &set_label("dec_out",4);
+    &mov	("esp",$_esp);
+    &popf	();
+&function_end("Camellia_cbc_encrypt");
+}
+
+&asciz("Camellia for x86 by <appro\@openssl.org>");
+
+&asm_finish();
diff --git a/crypto/camellia/asm/cmll-x86_64.pl b/crypto/camellia/asm/cmll-x86_64.pl
new file mode 100644
index 0000000000..76955e4726
--- /dev/null
+++ b/crypto/camellia/asm/cmll-x86_64.pl
@@ -0,0 +1,1080 @@
+#!/usr/bin/env perl
+
+# ====================================================================
+# Copyright (c) 2008 Andy Polyakov <appro@openssl.org>
+#
+# This module may be used under the terms of either the GNU General
+# Public License version 2 or later, the GNU Lesser General Public
+# License version 2.1 or later, the Mozilla Public License version
+# 1.1 or the BSD License. The exact terms of either license are
+# distributed along with this module. For further details see
+# http://www.openssl.org/~appro/camellia/.
+# ====================================================================
+
+# Performance in cycles per processed byte (less is better) in
+# 'openssl speed ...' benchmark:
+#
+#			AMD64	Core2	EM64T
+# -evp camellia-128-ecb	16.7	21.0	22.7
+# + over gcc 3.4.6	+25%	+5%	0%
+#
+# camellia-128-cbc	15.7	20.4	21.1
+#
+# 128-bit key setup	128	216	205	cycles/key
+# + over gcc 3.4.6	+54%	+39%	+15%
+#
+# Numbers in "+" rows represent performance improvement over compiler
+# generated code. Key setup timings are impressive on AMD and Core2
+# thanks to 64-bit operations being covertly deployed. Improvement on
+# EM64T, pre-Core2 Intel x86_64 CPU, is not as impressive, because it
+# apparently emulates some of 64-bit operations in [32-bit] microcode.
+
+$flavour = shift;
+$output  = shift;
+if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
+
+$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);
+
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or
+( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or
+die "can't locate x86_64-xlate.pl";
+
+open STDOUT,"| $^X $xlate $flavour $output";
+
+sub hi() { my $r=shift; $r =~ s/%[er]([a-d])x/%\1h/;    $r; }
+sub lo() { my $r=shift; $r =~ s/%[er]([a-d])x/%\1l/;
+                        $r =~ s/%[er]([sd]i)/%\1l/;
+                        $r =~ s/%(r[0-9]+)[d]?/%\1b/;   $r; }
+
+$t0="%eax";$t1="%ebx";$t2="%ecx";$t3="%edx";
+@S=("%r8d","%r9d","%r10d","%r11d");
+$i0="%esi";
+$i1="%edi";
+$Tbl="%rbp";	# size optimization
+$inp="%r12";
+$out="%r13";
+$key="%r14";
+$keyend="%r15";
+$arg0d=$win64?"%ecx":"%edi";
+
+# const unsigned int Camellia_SBOX[4][256];
+# Well, sort of... Camellia_SBOX[0][] is interleaved with [1][],
+# and [2][] - with [3][]. This is done to minimize code size.
+$SBOX1_1110=0;		# Camellia_SBOX[0]
+$SBOX4_4404=4;		# Camellia_SBOX[1]
+$SBOX2_0222=2048;	# Camellia_SBOX[2]
+$SBOX3_3033=2052;	# Camellia_SBOX[3]
+
+sub Camellia_Feistel {
+my $i=@_[0];
+my $seed=defined(@_[1])?@_[1]:0;
+my $scale=$seed<0?-8:8;
+my $j=($i&1)*2;
+my $s0=@S[($j)%4],$s1=@S[($j+1)%4],$s2=@S[($j+2)%4],$s3=@S[($j+3)%4];
+
+$code.=<<___;
+	xor	$s0,$t0				# t0^=key[0]
+	xor	$s1,$t1				# t1^=key[1]
+	movz	`&hi("$t0")`,$i0		# (t0>>8)&0xff
+	movz	`&lo("$t1")`,$i1		# (t1>>0)&0xff
+	mov	$SBOX3_3033($Tbl,$i0,8),$t3	# t3=SBOX3_3033[0]
+	mov	$SBOX1_1110($Tbl,$i1,8),$t2	# t2=SBOX1_1110[1]
+	movz	`&lo("$t0")`,$i0		# (t0>>0)&0xff
+	shr	\$16,$t0
+	movz	`&hi("$t1")`,$i1		# (t1>>8)&0xff
+	xor	$SBOX4_4404($Tbl,$i0,8),$t3	# t3^=SBOX4_4404[0]
+	shr	\$16,$t1
+	xor	$SBOX4_4404($Tbl,$i1,8),$t2	# t2^=SBOX4_4404[1]
+	movz	`&hi("$t0")`,$i0		# (t0>>24)&0xff
+	movz	`&lo("$t1")`,$i1		# (t1>>16)&0xff
+	xor	$SBOX1_1110($Tbl,$i0,8),$t3	# t3^=SBOX1_1110[0]
+	xor	$SBOX3_3033($Tbl,$i1,8),$t2	# t2^=SBOX3_3033[1]
+	movz	`&lo("$t0")`,$i0		# (t0>>16)&0xff
+	movz	`&hi("$t1")`,$i1		# (t1>>24)&0xff
+	xor	$SBOX2_0222($Tbl,$i0,8),$t3	# t3^=SBOX2_0222[0]
+	xor	$SBOX2_0222($Tbl,$i1,8),$t2	# t2^=SBOX2_0222[1]
+	mov	`$seed+($i+1)*$scale`($key),$t1	# prefetch key[i+1]
+	mov	`$seed+($i+1)*$scale+4`($key),$t0
+	xor	$t3,$t2				# t2^=t3
+	ror	\$8,$t3				# t3=RightRotate(t3,8)
+	xor	$t2,$s2
+	xor	$t2,$s3
+	xor	$t3,$s3
+___
+}
+
+# void Camellia_EncryptBlock_Rounds(
+#		int grandRounds,
+#		const Byte plaintext[],
+#		const KEY_TABLE_TYPE keyTable,
+#		Byte ciphertext[])
+$code=<<___;
+.text
+
+# V1.x API
+.globl	Camellia_EncryptBlock
+.type	Camellia_EncryptBlock,\@abi-omnipotent
+.align	16
+Camellia_EncryptBlock:
+	movl	\$128,%eax
+	subl	$arg0d,%eax
+	movl	\$3,$arg0d
+	adcl	\$0,$arg0d	# keyBitLength==128?3:4
+	jmp	.Lenc_rounds
+.size	Camellia_EncryptBlock,.-Camellia_EncryptBlock
+# V2
+.globl	Camellia_EncryptBlock_Rounds
+.type	Camellia_EncryptBlock_Rounds,\@function,4
+.align	16
+.Lenc_rounds:
+Camellia_EncryptBlock_Rounds:
+	push	%rbx
+	push	%rbp
+	push	%r13
+	push	%r14
+	push	%r15
+.Lenc_prologue:
+
+	#mov	%rsi,$inp		# put away arguments
+	mov	%rcx,$out
+	mov	%rdx,$key
+
+	shl	\$6,%edi		# process grandRounds
+	lea	.LCamellia_SBOX(%rip),$Tbl
+	lea	($key,%rdi),$keyend
+
+	mov	0(%rsi),@S[0]		# load plaintext
+	mov	4(%rsi),@S[1]
+	mov	8(%rsi),@S[2]
+	bswap	@S[0]
+	mov	12(%rsi),@S[3]
+	bswap	@S[1]
+	bswap	@S[2]
+	bswap	@S[3]
+
+	call	_x86_64_Camellia_encrypt
+
+	bswap	@S[0]
+	bswap	@S[1]
+	bswap	@S[2]
+	mov	@S[0],0($out)
+	bswap	@S[3]
+	mov	@S[1],4($out)
+	mov	@S[2],8($out)
+	mov	@S[3],12($out)
+
+	mov	0(%rsp),%r15
+	mov	8(%rsp),%r14
+	mov	16(%rsp),%r13
+	mov	24(%rsp),%rbp
+	mov	32(%rsp),%rbx
+	lea	40(%rsp),%rsp
+.Lenc_epilogue:
+	ret
+.size	Camellia_EncryptBlock_Rounds,.-Camellia_EncryptBlock_Rounds
+
+.type	_x86_64_Camellia_encrypt,\@abi-omnipotent
+.align	16
+_x86_64_Camellia_encrypt:
+	xor	0($key),@S[1]
+	xor	4($key),@S[0]		# ^=key[0-3]
+	xor	8($key),@S[3]
+	xor	12($key),@S[2]
+.align	16
+.Leloop:
+	mov	16($key),$t1		# prefetch key[4-5]
+	mov	20($key),$t0
+
+___
+	for ($i=0;$i<6;$i++) { Camellia_Feistel($i,16); }
+$code.=<<___;
+	lea	16*4($key),$key
+	cmp	$keyend,$key
+	mov	8($key),$t3		# prefetch key[2-3]
+	mov	12($key),$t2
+	je	.Ledone
+
+	and	@S[0],$t0
+	or	@S[3],$t3
+	rol	\$1,$t0
+	xor	$t3,@S[2]		# s2^=s3|key[3];
+	xor	$t0,@S[1]		# s1^=LeftRotate(s0&key[0],1);
+	and	@S[2],$t2
+	or	@S[1],$t1
+	rol	\$1,$t2
+	xor	$t1,@S[0]		# s0^=s1|key[1];
+	xor	$t2,@S[3]		# s3^=LeftRotate(s2&key[2],1);
+	jmp	.Leloop
+
+.align	16
+.Ledone:
+	xor	@S[2],$t0		# SwapHalf
+	xor	@S[3],$t1
+	xor	@S[0],$t2
+	xor	@S[1],$t3
+
+	mov	$t0,@S[0]
+	mov	$t1,@S[1]
+	mov	$t2,@S[2]
+	mov	$t3,@S[3]
+
+	.byte	0xf3,0xc3		# rep ret
+.size	_x86_64_Camellia_encrypt,.-_x86_64_Camellia_encrypt
+
+# V1.x API
+.globl	Camellia_DecryptBlock
+.type	Camellia_DecryptBlock,\@abi-omnipotent
+.align	16
+Camellia_DecryptBlock:
+	movl	\$128,%eax
+	subl	$arg0d,%eax
+	movl	\$3,$arg0d
+	adcl	\$0,$arg0d	# keyBitLength==128?3:4
+	jmp	.Ldec_rounds
+.size	Camellia_DecryptBlock,.-Camellia_DecryptBlock
+# V2
+.globl	Camellia_DecryptBlock_Rounds
+.type	Camellia_DecryptBlock_Rounds,\@function,4
+.align	16
+.Ldec_rounds:
+Camellia_DecryptBlock_Rounds:
+	push	%rbx
+	push	%rbp
+	push	%r13
+	push	%r14
+	push	%r15
+.Ldec_prologue:
+
+	#mov	%rsi,$inp		# put away arguments
+	mov	%rcx,$out
+	mov	%rdx,$keyend
+
+	shl	\$6,%edi		# process grandRounds
+	lea	.LCamellia_SBOX(%rip),$Tbl
+	lea	($keyend,%rdi),$key
+
+	mov	0(%rsi),@S[0]		# load plaintext
+	mov	4(%rsi),@S[1]
+	mov	8(%rsi),@S[2]
+	bswap	@S[0]
+	mov	12(%rsi),@S[3]
+	bswap	@S[1]
+	bswap	@S[2]
+	bswap	@S[3]
+
+	call	_x86_64_Camellia_decrypt
+
+	bswap	@S[0]
+	bswap	@S[1]
+	bswap	@S[2]
+	mov	@S[0],0($out)
+	bswap	@S[3]
+	mov	@S[1],4($out)
+	mov	@S[2],8($out)
+	mov	@S[3],12($out)
+
+	mov	0(%rsp),%r15
+	mov	8(%rsp),%r14
+	mov	16(%rsp),%r13
+	mov	24(%rsp),%rbp
+	mov	32(%rsp),%rbx
+	lea	40(%rsp),%rsp
+.Ldec_epilogue:
+	ret
+.size	Camellia_DecryptBlock_Rounds,.-Camellia_DecryptBlock_Rounds
+
+.type	_x86_64_Camellia_decrypt,\@abi-omnipotent
+.align	16
+_x86_64_Camellia_decrypt:
+	xor	0($key),@S[1]
+	xor	4($key),@S[0]		# ^=key[0-3]
+	xor	8($key),@S[3]
+	xor	12($key),@S[2]
+.align	16
+.Ldloop:
+	mov	-8($key),$t1		# prefetch key[4-5]
+	mov	-4($key),$t0
+
+___
+	for ($i=0;$i<6;$i++) { Camellia_Feistel($i,-8); }
+$code.=<<___;
+	lea	-16*4($key),$key
+	cmp	$keyend,$key
+	mov	0($key),$t3		# prefetch key[2-3]
+	mov	4($key),$t2
+	je	.Lddone
+
+	and	@S[0],$t0
+	or	@S[3],$t3
+	rol	\$1,$t0
+	xor	$t3,@S[2]		# s2^=s3|key[3];
+	xor	$t0,@S[1]		# s1^=LeftRotate(s0&key[0],1);
+	and	@S[2],$t2
+	or	@S[1],$t1
+	rol	\$1,$t2
+	xor	$t1,@S[0]		# s0^=s1|key[1];
+	xor	$t2,@S[3]		# s3^=LeftRotate(s2&key[2],1);
+
+	jmp	.Ldloop
+
+.align	16
+.Lddone:
+	xor	@S[2],$t2
+	xor	@S[3],$t3
+	xor	@S[0],$t0
+	xor	@S[1],$t1
+
+	mov	$t2,@S[0]		# SwapHalf
+	mov	$t3,@S[1]
+	mov	$t0,@S[2]
+	mov	$t1,@S[3]
+
+	.byte	0xf3,0xc3		# rep ret
+.size	_x86_64_Camellia_decrypt,.-_x86_64_Camellia_decrypt
+___
+
+sub _saveround {
+my ($rnd,$key,@T)=@_;
+my $bias=int(@T[0])?shift(@T):0;
+
+    if ($#T==3) {
+	$code.=<<___;
+	mov	@T[1],`$bias+$rnd*8+0`($key)
+	mov	@T[0],`$bias+$rnd*8+4`($key)
+	mov	@T[3],`$bias+$rnd*8+8`($key)
+	mov	@T[2],`$bias+$rnd*8+12`($key)
+___
+    } else {
+	$code.="	mov	@T[0],`$bias+$rnd*8+0`($key)\n";
+	$code.="	mov	@T[1],`$bias+$rnd*8+8`($key)\n"	if ($#T>=1);
+    }
+}
+
+sub _loadround {
+my ($rnd,$key,@T)=@_;
+my $bias=int(@T[0])?shift(@T):0;
+
+$code.="	mov	`$bias+$rnd*8+0`($key),@T[0]\n";
+$code.="	mov	`$bias+$rnd*8+8`($key),@T[1]\n"	if ($#T>=1);
+}
+
+# shld is very slow on Intel EM64T family. Even on AMD it limits
+# instruction decode rate [because it's VectorPath] and consequently
+# performance...
+sub __rotl128 {
+my ($i0,$i1,$rot)=@_;
+
+    if ($rot) {
+	$code.=<<___;
+	mov	$i0,%r11
+	shld	\$$rot,$i1,$i0
+	shld	\$$rot,%r11,$i1
+___
+    }
+}
+
+# ... Implementing 128-bit rotate without shld gives 80% better
+# performance EM64T, +15% on AMD64 and only ~7% degradation on
+# Core2. This is therefore preferred.
+sub _rotl128 {
+my ($i0,$i1,$rot)=@_;
+
+    if ($rot) {
+	$code.=<<___;
+	mov	$i0,%r11
+	shl	\$$rot,$i0
+	mov	$i1,%r9
+	shr	\$`64-$rot`,%r9
+	shr	\$`64-$rot`,%r11
+	or	%r9,$i0
+	shl	\$$rot,$i1
+	or	%r11,$i1
+___
+    }
+}
+
+{ my $step=0;
+
+$code.=<<___;
+.globl	Camellia_Ekeygen
+.type	Camellia_Ekeygen,\@function,3
+.align	16
+Camellia_Ekeygen:
+	push	%rbx
+	push	%rbp
+	push	%r13
+	push	%r14
+	push	%r15
+.Lkey_prologue:
+
+	mov	%rdi,$keyend		# put away arguments, keyBitLength
+	mov	%rdx,$out		# keyTable
+
+	mov	0(%rsi),@S[0]		# load 0-127 bits
+	mov	4(%rsi),@S[1]
+	mov	8(%rsi),@S[2]
+	mov	12(%rsi),@S[3]
+
+	bswap	@S[0]
+	bswap	@S[1]
+	bswap	@S[2]
+	bswap	@S[3]
+___
+	&_saveround	(0,$out,@S);	# KL<<<0
+$code.=<<___;
+	cmp	\$128,$keyend		# check keyBitLength
+	je	.L1st128
+
+	mov	16(%rsi),@S[0]		# load 128-191 bits
+	mov	20(%rsi),@S[1]
+	cmp	\$192,$keyend
+	je	.L1st192
+	mov	24(%rsi),@S[2]		# load 192-255 bits
+	mov	28(%rsi),@S[3]
+	jmp	.L1st256
+.L1st192:
+	mov	@S[0],@S[2]
+	mov	@S[1],@S[3]
+	not	@S[2]
+	not	@S[3]
+.L1st256:
+	bswap	@S[0]
+	bswap	@S[1]
+	bswap	@S[2]
+	bswap	@S[3]
+___
+	&_saveround	(4,$out,@S);	# temp storage for KR!
+$code.=<<___;
+	xor	0($out),@S[1]		# KR^KL
+	xor	4($out),@S[0]
+	xor	8($out),@S[3]
+	xor	12($out),@S[2]
+
+.L1st128:
+	lea	.LCamellia_SIGMA(%rip),$key
+	lea	.LCamellia_SBOX(%rip),$Tbl
+
+	mov	0($key),$t1
+	mov	4($key),$t0
+___
+	&Camellia_Feistel($step++);
+	&Camellia_Feistel($step++);
+$code.=<<___;
+	xor	0($out),@S[1]		# ^KL
+	xor	4($out),@S[0]
+	xor	8($out),@S[3]
+	xor	12($out),@S[2]
+___
+	&Camellia_Feistel($step++);
+	&Camellia_Feistel($step++);
+$code.=<<___;
+	cmp	\$128,$keyend
+	jne	.L2nd256
+
+	lea	128($out),$out		# size optimization
+	shl	\$32,%r8		# @S[0]||
+	shl	\$32,%r10		# @S[2]||
+	or	%r9,%r8			# ||@S[1]
+	or	%r11,%r10		# ||@S[3]
+___
+	&_loadround	(0,$out,-128,"%rax","%rbx");	# KL
+	&_saveround	(2,$out,-128,"%r8","%r10");	# KA<<<0
+	&_rotl128	("%rax","%rbx",15);
+	&_saveround	(4,$out,-128,"%rax","%rbx");	# KL<<<15
+	&_rotl128	("%r8","%r10",15);
+	&_saveround	(6,$out,-128,"%r8","%r10");	# KA<<<15
+	&_rotl128	("%r8","%r10",15);		# 15+15=30
+	&_saveround	(8,$out,-128,"%r8","%r10");	# KA<<<30
+	&_rotl128	("%rax","%rbx",30);		# 15+30=45
+	&_saveround	(10,$out,-128,"%rax","%rbx");	# KL<<<45
+	&_rotl128	("%r8","%r10",15);		# 30+15=45
+	&_saveround	(12,$out,-128,"%r8");		# KA<<<45
+	&_rotl128	("%rax","%rbx",15);		# 45+15=60
+	&_saveround	(13,$out,-128,"%rbx");		# KL<<<60
+	&_rotl128	("%r8","%r10",15);		# 45+15=60
+	&_saveround	(14,$out,-128,"%r8","%r10");	# KA<<<60
+	&_rotl128	("%rax","%rbx",17);		# 60+17=77
+	&_saveround	(16,$out,-128,"%rax","%rbx");	# KL<<<77
+	&_rotl128	("%rax","%rbx",17);		# 77+17=94
+	&_saveround	(18,$out,-128,"%rax","%rbx");	# KL<<<94
+	&_rotl128	("%r8","%r10",34);		# 60+34=94
+	&_saveround	(20,$out,-128,"%r8","%r10");	# KA<<<94
+	&_rotl128	("%rax","%rbx",17);		# 94+17=111
+	&_saveround	(22,$out,-128,"%rax","%rbx");	# KL<<<111
+	&_rotl128	("%r8","%r10",17);		# 94+17=111
+	&_saveround	(24,$out,-128,"%r8","%r10");	# KA<<<111
+$code.=<<___;
+	mov	\$3,%eax
+	jmp	.Ldone
+.align	16
+.L2nd256:
+___
+	&_saveround	(6,$out,@S);	# temp storage for KA!
+$code.=<<___;
+	xor	`4*8+0`($out),@S[1]	# KA^KR
+	xor	`4*8+4`($out),@S[0]
+	xor	`5*8+0`($out),@S[3]
+	xor	`5*8+4`($out),@S[2]
+___
+	&Camellia_Feistel($step++);
+	&Camellia_Feistel($step++);
+
+	&_loadround	(0,$out,"%rax","%rbx");	# KL
+	&_loadround	(4,$out,"%rcx","%rdx");	# KR
+	&_loadround	(6,$out,"%r14","%r15");	# KA
+$code.=<<___;
+	lea	128($out),$out		# size optimization
+	shl	\$32,%r8		# @S[0]||
+	shl	\$32,%r10		# @S[2]||
+	or	%r9,%r8			# ||@S[1]
+	or	%r11,%r10		# ||@S[3]
+___
+	&_saveround	(2,$out,-128,"%r8","%r10");	# KB<<<0
+	&_rotl128	("%rcx","%rdx",15);
+	&_saveround	(4,$out,-128,"%rcx","%rdx");	# KR<<<15
+	&_rotl128	("%r14","%r15",15);
+	&_saveround	(6,$out,-128,"%r14","%r15");	# KA<<<15
+	&_rotl128	("%rcx","%rdx",15);		# 15+15=30
+	&_saveround	(8,$out,-128,"%rcx","%rdx");	# KR<<<30
+	&_rotl128	("%r8","%r10",30);
+	&_saveround	(10,$out,-128,"%r8","%r10");	# KB<<<30
+	&_rotl128	("%rax","%rbx",45);
+	&_saveround	(12,$out,-128,"%rax","%rbx");	# KL<<<45
+	&_rotl128	("%r14","%r15",30);		# 15+30=45
+	&_saveround	(14,$out,-128,"%r14","%r15");	# KA<<<45
+	&_rotl128	("%rax","%rbx",15);		# 45+15=60
+	&_saveround	(16,$out,-128,"%rax","%rbx");	# KL<<<60
+	&_rotl128	("%rcx","%rdx",30);		# 30+30=60
+	&_saveround	(18,$out,-128,"%rcx","%rdx");	# KR<<<60
+	&_rotl128	("%r8","%r10",30);		# 30+30=60
+	&_saveround	(20,$out,-128,"%r8","%r10");	# KB<<<60
+	&_rotl128	("%rax","%rbx",17);		# 60+17=77
+	&_saveround	(22,$out,-128,"%rax","%rbx");	# KL<<<77
+	&_rotl128	("%r14","%r15",32);		# 45+32=77
+	&_saveround	(24,$out,-128,"%r14","%r15");	# KA<<<77
+	&_rotl128	("%rcx","%rdx",34);		# 60+34=94
+	&_saveround	(26,$out,-128,"%rcx","%rdx");	# KR<<<94
+	&_rotl128	("%r14","%r15",17);		# 77+17=94
+	&_saveround	(28,$out,-128,"%r14","%r15");	# KA<<<77
+	&_rotl128	("%rax","%rbx",34);		# 77+34=111
+	&_saveround	(30,$out,-128,"%rax","%rbx");	# KL<<<111
+	&_rotl128	("%r8","%r10",51);		# 60+51=111
+	&_saveround	(32,$out,-128,"%r8","%r10");	# KB<<<111
+$code.=<<___;
+	mov	\$4,%eax
+.Ldone:
+	mov	0(%rsp),%r15
+	mov	8(%rsp),%r14
+	mov	16(%rsp),%r13
+	mov	24(%rsp),%rbp
+	mov	32(%rsp),%rbx
+	lea	40(%rsp),%rsp
+.Lkey_epilogue:
+	ret
+.size	Camellia_Ekeygen,.-Camellia_Ekeygen
+___
+}
+
+@SBOX=(
+112,130, 44,236,179, 39,192,229,228,133, 87, 53,234, 12,174, 65,
+ 35,239,107,147, 69, 25,165, 33,237, 14, 79, 78, 29,101,146,189,
+134,184,175,143,124,235, 31,206, 62, 48,220, 95, 94,197, 11, 26,
+166,225, 57,202,213, 71, 93, 61,217,  1, 90,214, 81, 86,108, 77,
+139, 13,154,102,251,204,176, 45,116, 18, 43, 32,240,177,132,153,
+223, 76,203,194, 52,126,118,  5,109,183,169, 49,209, 23,  4,215,
+ 20, 88, 58, 97,222, 27, 17, 28, 50, 15,156, 22, 83, 24,242, 34,
+254, 68,207,178,195,181,122,145, 36,  8,232,168, 96,252,105, 80,
+170,208,160,125,161,137, 98,151, 84, 91, 30,149,224,255,100,210,
+ 16,196,  0, 72,163,247,117,219,138,  3,230,218,  9, 63,221,148,
+135, 92,131,  2,205, 74,144, 51,115,103,246,243,157,127,191,226,
+ 82,155,216, 38,200, 55,198, 59,129,150,111, 75, 19,190, 99, 46,
+233,121,167,140,159,110,188,142, 41,245,249,182, 47,253,180, 89,
+120,152,  6,106,231, 70,113,186,212, 37,171, 66,136,162,141,250,
+114,  7,185, 85,248,238,172, 10, 54, 73, 42,104, 60, 56,241,164,
+ 64, 40,211,123,187,201, 67,193, 21,227,173,244,119,199,128,158);
+
+sub S1110 { my $i=shift; $i=@SBOX[$i]; $i=$i<<24|$i<<16|$i<<8; sprintf("0x%08x",$i); }
+sub S4404 { my $i=shift; $i=($i<<1|$i>>7)&0xff; $i=@SBOX[$i]; $i=$i<<24|$i<<16|$i; sprintf("0x%08x",$i); }
+sub S0222 { my $i=shift; $i=@SBOX[$i]; $i=($i<<1|$i>>7)&0xff; $i=$i<<16|$i<<8|$i; sprintf("0x%08x",$i); }
+sub S3033 { my $i=shift; $i=@SBOX[$i]; $i=($i>>1|$i<<7)&0xff; $i=$i<<24|$i<<8|$i; sprintf("0x%08x",$i); }
+
+$code.=<<___;
+.align	64
+.LCamellia_SIGMA:
+.long	0x3bcc908b, 0xa09e667f, 0x4caa73b2, 0xb67ae858
+.long	0xe94f82be, 0xc6ef372f, 0xf1d36f1c, 0x54ff53a5
+.long	0xde682d1d, 0x10e527fa, 0xb3e6c1fd, 0xb05688c2
+.long	0,          0,          0,          0
+.LCamellia_SBOX:
+___
+# tables are interleaved, remember?
+sub data_word { $code.=".long\t".join(',',@_)."\n"; }
+for ($i=0;$i<256;$i++) { &data_word(&S1110($i),&S4404($i)); }
+for ($i=0;$i<256;$i++) { &data_word(&S0222($i),&S3033($i)); }
+
+# void Camellia_cbc_encrypt (const void char *inp, unsigned char *out,
+#			size_t length, const CAMELLIA_KEY *key,
+#			unsigned char *ivp,const int enc);
+{
+$_key="0(%rsp)";
+$_end="8(%rsp)";	# inp+len&~15
+$_res="16(%rsp)";	# len&15
+$ivec="24(%rsp)";
+$_ivp="40(%rsp)";
+$_rsp="48(%rsp)";
+
+$code.=<<___;
+.globl	Camellia_cbc_encrypt
+.type	Camellia_cbc_encrypt,\@function,6
+.align	16
+Camellia_cbc_encrypt:
+	cmp	\$0,%rdx
+	je	.Lcbc_abort
+	push	%rbx
+	push	%rbp
+	push	%r12
+	push	%r13
+	push	%r14
+	push	%r15
+.Lcbc_prologue:
+
+	mov	%rsp,%rbp
+	sub	\$64,%rsp
+	and	\$-64,%rsp
+
+	# place stack frame just "above mod 1024" the key schedule,
+	# this ensures that cache associativity suffices
+	lea	-64-63(%rcx),%r10
+	sub	%rsp,%r10
+	neg	%r10
+	and	\$0x3C0,%r10
+	sub	%r10,%rsp
+	#add	\$8,%rsp		# 8 is reserved for callee's ra
+
+	mov	%rdi,$inp		# inp argument
+	mov	%rsi,$out		# out argument
+	mov	%r8,%rbx		# ivp argument
+	mov	%rcx,$key		# key argument
+	mov	272(%rcx),${keyend}d	# grandRounds
+
+	mov	%r8,$_ivp
+	mov	%rbp,$_rsp
+
+.Lcbc_body:
+	lea	.LCamellia_SBOX(%rip),$Tbl
+
+	mov	\$32,%ecx
+.align	4
+.Lcbc_prefetch_sbox:
+	mov	0($Tbl),%rax
+	mov	32($Tbl),%rsi
+	mov	64($Tbl),%rdi
+	mov	96($Tbl),%r11
+	lea	128($Tbl),$Tbl
+	loop	.Lcbc_prefetch_sbox
+	sub	\$4096,$Tbl
+	shl	\$6,$keyend
+	mov	%rdx,%rcx		# len argument
+	lea	($key,$keyend),$keyend
+
+	cmp	\$0,%r9d		# enc argument
+	je	.LCBC_DECRYPT
+
+	and	\$-16,%rdx
+	and	\$15,%rcx		# length residue
+	lea	($inp,%rdx),%rdx
+	mov	$key,$_key
+	mov	%rdx,$_end
+	mov	%rcx,$_res
+
+	cmp	$inp,%rdx
+	mov	0(%rbx),@S[0]		# load IV
+	mov	4(%rbx),@S[1]
+	mov	8(%rbx),@S[2]
+	mov	12(%rbx),@S[3]
+	je	.Lcbc_enc_tail
+	jmp	.Lcbc_eloop
+
+.align	16
+.Lcbc_eloop:
+	xor	0($inp),@S[0]
+	xor	4($inp),@S[1]
+	xor	8($inp),@S[2]
+	bswap	@S[0]
+	xor	12($inp),@S[3]
+	bswap	@S[1]
+	bswap	@S[2]
+	bswap	@S[3]
+
+	call	_x86_64_Camellia_encrypt
+
+	mov	$_key,$key		# "rewind" the key
+	bswap	@S[0]
+	mov	$_end,%rdx
+	bswap	@S[1]
+	mov	$_res,%rcx
+	bswap	@S[2]
+	mov	@S[0],0($out)
+	bswap	@S[3]
+	mov	@S[1],4($out)
+	mov	@S[2],8($out)
+	lea	16($inp),$inp
+	mov	@S[3],12($out)
+	cmp	%rdx,$inp
+	lea	16($out),$out
+	jne	.Lcbc_eloop
+
+	cmp	\$0,%rcx
+	jne	.Lcbc_enc_tail
+
+	mov	$_ivp,$out
+	mov	@S[0],0($out)		# write out IV residue
+	mov	@S[1],4($out)
+	mov	@S[2],8($out)
+	mov	@S[3],12($out)
+	jmp	.Lcbc_done
+
+.align	16
+.Lcbc_enc_tail:
+	xor	%rax,%rax
+	mov	%rax,0+$ivec
+	mov	%rax,8+$ivec
+	mov	%rax,$_res
+
+.Lcbc_enc_pushf:
+	pushfq
+	cld
+	mov	$inp,%rsi
+	lea	8+$ivec,%rdi
+	.long	0x9066A4F3		# rep movsb
+	popfq
+.Lcbc_enc_popf:
+
+	lea	$ivec,$inp
+	lea	16+$ivec,%rax
+	mov	%rax,$_end
+	jmp	.Lcbc_eloop		# one more time
+
+.align	16
+.LCBC_DECRYPT:
+	xchg	$key,$keyend
+	add	\$15,%rdx
+	and	\$15,%rcx		# length residue
+	and	\$-16,%rdx
+	mov	$key,$_key
+	lea	($inp,%rdx),%rdx
+	mov	%rdx,$_end
+	mov	%rcx,$_res
+
+	mov	(%rbx),%rax		# load IV
+	mov	8(%rbx),%rbx
+	jmp	.Lcbc_dloop
+.align	16
+.Lcbc_dloop:
+	mov	0($inp),@S[0]
+	mov	4($inp),@S[1]
+	mov	8($inp),@S[2]
+	bswap	@S[0]
+	mov	12($inp),@S[3]
+	bswap	@S[1]
+	mov	%rax,0+$ivec		# save IV to temporary storage
+	bswap	@S[2]
+	mov	%rbx,8+$ivec
+	bswap	@S[3]
+
+	call	_x86_64_Camellia_decrypt
+
+	mov	$_key,$key		# "rewind" the key
+	mov	$_end,%rdx
+	mov	$_res,%rcx
+
+	bswap	@S[0]
+	mov	($inp),%rax		# load IV for next iteration
+	bswap	@S[1]
+	mov	8($inp),%rbx
+	bswap	@S[2]
+	xor	0+$ivec,@S[0]
+	bswap	@S[3]
+	xor	4+$ivec,@S[1]
+	xor	8+$ivec,@S[2]
+	lea	16($inp),$inp
+	xor	12+$ivec,@S[3]
+	cmp	%rdx,$inp
+	je	.Lcbc_ddone
+
+	mov	@S[0],0($out)
+	mov	@S[1],4($out)
+	mov	@S[2],8($out)
+	mov	@S[3],12($out)
+
+	lea	16($out),$out
+	jmp	.Lcbc_dloop
+
+.align	16
+.Lcbc_ddone:
+	mov	$_ivp,%rdx
+	cmp	\$0,%rcx
+	jne	.Lcbc_dec_tail
+
+	mov	@S[0],0($out)
+	mov	@S[1],4($out)
+	mov	@S[2],8($out)
+	mov	@S[3],12($out)
+
+	mov	%rax,(%rdx)		# write out IV residue
+	mov	%rbx,8(%rdx)
+	jmp	.Lcbc_done
+.align	16
+.Lcbc_dec_tail:
+	mov	@S[0],0+$ivec
+	mov	@S[1],4+$ivec
+	mov	@S[2],8+$ivec
+	mov	@S[3],12+$ivec
+
+.Lcbc_dec_pushf:
+	pushfq
+	cld
+	lea	8+$ivec,%rsi
+	lea	($out),%rdi
+	.long	0x9066A4F3		# rep movsb
+	popfq
+.Lcbc_dec_popf:
+
+	mov	%rax,(%rdx)		# write out IV residue
+	mov	%rbx,8(%rdx)
+	jmp	.Lcbc_done
+
+.align	16
+.Lcbc_done:
+	mov	$_rsp,%rcx
+	mov	0(%rcx),%r15
+	mov	8(%rcx),%r14
+	mov	16(%rcx),%r13
+	mov	24(%rcx),%r12
+	mov	32(%rcx),%rbp
+	mov	40(%rcx),%rbx
+	lea	48(%rcx),%rsp
+.Lcbc_abort:
+	ret
+.size	Camellia_cbc_encrypt,.-Camellia_cbc_encrypt
+
+.asciz	"Camellia for x86_64 by <appro\@openssl.org>"
+___
+}
+
+# EXCEPTION_DISPOSITION handler (EXCEPTION_RECORD *rec,ULONG64 frame,
+#		CONTEXT *context,DISPATCHER_CONTEXT *disp)
+if ($win64) {
+$rec="%rcx";
+$frame="%rdx";
+$context="%r8";
+$disp="%r9";
+
+$code.=<<___;
+.extern	__imp_RtlVirtualUnwind
+.type	common_se_handler,\@abi-omnipotent
+.align	16
+common_se_handler:
+	push	%rsi
+	push	%rdi
+	push	%rbx
+	push	%rbp
+	push	%r12
+	push	%r13
+	push	%r14
+	push	%r15
+	pushfq
+	lea	-64(%rsp),%rsp
+
+	mov	120($context),%rax	# pull context->Rax
+	mov	248($context),%rbx	# pull context->Rip
+
+	mov	8($disp),%rsi		# disp->ImageBase
+	mov	56($disp),%r11		# disp->HandlerData
+
+	mov	0(%r11),%r10d		# HandlerData[0]
+	lea	(%rsi,%r10),%r10	# prologue label
+	cmp	%r10,%rbx		# context->Rip<prologue label
+	jb	.Lin_prologue
+
+	mov	152($context),%rax	# pull context->Rsp
+
+	mov	4(%r11),%r10d		# HandlerData[1]
+	lea	(%rsi,%r10),%r10	# epilogue label
+	cmp	%r10,%rbx		# context->Rip>=epilogue label
+	jae	.Lin_prologue
+
+	lea	40(%rax),%rax
+	mov	-8(%rax),%rbx
+	mov	-16(%rax),%rbp
+	mov	-24(%rax),%r13
+	mov	-32(%rax),%r14
+	mov	-40(%rax),%r15
+	mov	%rbx,144($context)	# restore context->Rbx
+	mov	%rbp,160($context)	# restore context->Rbp
+	mov	%r13,224($context)	# restore context->R13
+	mov	%r14,232($context)	# restore context->R14
+	mov	%r15,240($context)	# restore context->R15
+
+.Lin_prologue:
+	mov	8(%rax),%rdi
+	mov	16(%rax),%rsi
+	mov	%rax,152($context)	# restore context->Rsp
+	mov	%rsi,168($context)	# restore context->Rsi
+	mov	%rdi,176($context)	# restore context->Rdi
+
+	jmp	.Lcommon_seh_exit
+.size	common_se_handler,.-common_se_handler
+
+.type	cbc_se_handler,\@abi-omnipotent
+.align	16
+cbc_se_handler:
+	push	%rsi
+	push	%rdi
+	push	%rbx
+	push	%rbp
+	push	%r12
+	push	%r13
+	push	%r14
+	push	%r15
+	pushfq
+	lea	-64(%rsp),%rsp
+
+	mov	120($context),%rax	# pull context->Rax
+	mov	248($context),%rbx	# pull context->Rip
+
+	lea	.Lcbc_prologue(%rip),%r10
+	cmp	%r10,%rbx		# context->Rip<.Lcbc_prologue
+	jb	.Lin_cbc_prologue
+
+	lea	.Lcbc_body(%rip),%r10
+	cmp	%r10,%rbx		# context->Rip<.Lcbc_body
+	jb	.Lin_cbc_frame_setup
+
+	mov	152($context),%rax	# pull context->Rsp
+
+	lea	.Lcbc_abort(%rip),%r10
+	cmp	%r10,%rbx		# context->Rip>=.Lcbc_abort
+	jae	.Lin_cbc_prologue
+
+	# handle pushf/popf in Camellia_cbc_encrypt
+	lea	.Lcbc_enc_pushf(%rip),%r10
+	cmp	%r10,%rbx		# context->Rip<=.Lcbc_enc_pushf
+	jbe	.Lin_cbc_no_flag
+	lea	8(%rax),%rax
+	lea	.Lcbc_enc_popf(%rip),%r10
+	cmp	%r10,%rbx		# context->Rip<.Lcbc_enc_popf
+	jb	.Lin_cbc_no_flag
+	lea	-8(%rax),%rax
+	lea	.Lcbc_dec_pushf(%rip),%r10
+	cmp	%r10,%rbx		# context->Rip<=.Lcbc_dec_pushf
+	jbe	.Lin_cbc_no_flag
+	lea	8(%rax),%rax
+	lea	.Lcbc_dec_popf(%rip),%r10
+	cmp	%r10,%rbx		# context->Rip<.Lcbc_dec_popf
+	jb	.Lin_cbc_no_flag
+	lea	-8(%rax),%rax
+
+.Lin_cbc_no_flag:
+	mov	48(%rax),%rax		# $_rsp
+	lea	48(%rax),%rax
+
+.Lin_cbc_frame_setup:
+	mov	-8(%rax),%rbx
+	mov	-16(%rax),%rbp
+	mov	-24(%rax),%r12
+	mov	-32(%rax),%r13
+	mov	-40(%rax),%r14
+	mov	-48(%rax),%r15
+	mov	%rbx,144($context)	# restore context->Rbx
+	mov	%rbp,160($context)	# restore context->Rbp
+	mov	%r12,216($context)	# restore context->R12
+	mov	%r13,224($context)	# restore context->R13
+	mov	%r14,232($context)	# restore context->R14
+	mov	%r15,240($context)	# restore context->R15
+
+.Lin_cbc_prologue:
+	mov	8(%rax),%rdi
+	mov	16(%rax),%rsi
+	mov	%rax,152($context)	# restore context->Rsp
+	mov	%rsi,168($context)	# restore context->Rsi
+	mov	%rdi,176($context)	# restore context->Rdi
+
+.align	4
+.Lcommon_seh_exit:
+
+	mov	40($disp),%rdi		# disp->ContextRecord
+	mov	$context,%rsi		# context
+	mov	\$`1232/8`,%ecx		# sizeof(CONTEXT)
+	.long	0xa548f3fc		# cld; rep movsq
+
+	mov	$disp,%rsi
+	xor	%rcx,%rcx		# arg1, UNW_FLAG_NHANDLER
+	mov	8(%rsi),%rdx		# arg2, disp->ImageBase
+	mov	0(%rsi),%r8		# arg3, disp->ControlPc
+	mov	16(%rsi),%r9		# arg4, disp->FunctionEntry
+	mov	40(%rsi),%r10		# disp->ContextRecord
+	lea	56(%rsi),%r11		# &disp->HandlerData
+	lea	24(%rsi),%r12		# &disp->EstablisherFrame
+	mov	%r10,32(%rsp)		# arg5
+	mov	%r11,40(%rsp)		# arg6
+	mov	%r12,48(%rsp)		# arg7
+	mov	%rcx,56(%rsp)		# arg8, (NULL)
+	call	*__imp_RtlVirtualUnwind(%rip)
+
+	mov	\$1,%eax		# ExceptionContinueSearch
+	lea	64(%rsp),%rsp
+	popfq
+	pop	%r15
+	pop	%r14
+	pop	%r13
+	pop	%r12
+	pop	%rbp
+	pop	%rbx
+	pop	%rdi
+	pop	%rsi
+	ret
+.size	cbc_se_handler,.-cbc_se_handler
+
+.section	.pdata
+.align	4
+	.rva	.LSEH_begin_Camellia_EncryptBlock_Rounds
+	.rva	.LSEH_end_Camellia_EncryptBlock_Rounds
+	.rva	.LSEH_info_Camellia_EncryptBlock_Rounds
+
+	.rva	.LSEH_begin_Camellia_DecryptBlock_Rounds
+	.rva	.LSEH_end_Camellia_DecryptBlock_Rounds
+	.rva	.LSEH_info_Camellia_DecryptBlock_Rounds
+
+	.rva	.LSEH_begin_Camellia_Ekeygen
+	.rva	.LSEH_end_Camellia_Ekeygen
+	.rva	.LSEH_info_Camellia_Ekeygen
+
+	.rva	.LSEH_begin_Camellia_cbc_encrypt
+	.rva	.LSEH_end_Camellia_cbc_encrypt
+	.rva	.LSEH_info_Camellia_cbc_encrypt
+
+.section	.xdata
+.align	8
+.LSEH_info_Camellia_EncryptBlock_Rounds:
+	.byte	9,0,0,0
+	.rva	common_se_handler
+	.rva	.Lenc_prologue,.Lenc_epilogue	# HandlerData[]
+.LSEH_info_Camellia_DecryptBlock_Rounds:
+	.byte	9,0,0,0
+	.rva	common_se_handler
+	.rva	.Ldec_prologue,.Ldec_epilogue	# HandlerData[]
+.LSEH_info_Camellia_Ekeygen:
+	.byte	9,0,0,0
+	.rva	common_se_handler
+	.rva	.Lkey_prologue,.Lkey_epilogue	# HandlerData[]
+.LSEH_info_Camellia_cbc_encrypt:
+	.byte	9,0,0,0
+	.rva	cbc_se_handler
+___
+}
+
+$code =~ s/\`([^\`]*)\`/eval $1/gem;
+print $code;
+close STDOUT;
diff --git a/crypto/camellia/camellia.c b/crypto/camellia/camellia.c
new file mode 100644
index 0000000000..75fc8991c0
--- /dev/null
+++ b/crypto/camellia/camellia.c
@@ -0,0 +1,582 @@
+/* crypto/camellia/camellia.c -*- mode:C; c-file-style: "eay" -*- */
+/* ====================================================================
+ * Copyright 2006 NTT (Nippon Telegraph and Telephone Corporation) . 
+ * ALL RIGHTS RESERVED.
+ *
+ * Intellectual Property information for Camellia:
+ *     http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html
+ *
+ * News Release for Announcement of Camellia open source:
+ *     http://www.ntt.co.jp/news/news06e/0604/060413a.html
+ *
+ * The Camellia Code included herein is developed by
+ * NTT (Nippon Telegraph and Telephone Corporation), and is contributed
+ * to the OpenSSL project.
+ *
+ * The Camellia Code is licensed pursuant to the OpenSSL open source
+ * license provided below.
+ */
+/* ====================================================================
+ * Copyright (c) 2006 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer. 
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ */
+
+/* Algorithm Specification 
+   http://info.isl.ntt.co.jp/crypt/eng/camellia/specifications.html
+*/
+ 
+/*
+ * This release balances code size and performance. In particular key
+ * schedule setup is fully unrolled, because doing so *significantly*
+ * reduces amount of instructions per setup round and code increase is
+ * justifiable. In block functions on the other hand only inner loops
+ * are unrolled, as full unroll gives only nominal performance boost,
+ * while code size grows 4 or 7 times. Also, unlike previous versions
+ * this one "encourages" compiler to keep intermediate variables in
+ * registers, which should give better "all round" results, in other
+ * words reasonable performance even with not so modern compilers.
+ */
+
+#include "camellia.h"
+#include "cmll_locl.h"
+#include <string.h>
+#include <stdlib.h>
+
+/* 32-bit rotations */
+#if !defined(PEDANTIC) && !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM)
+# if defined(_MSC_VER) && (defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64))
+#  define RightRotate(x, s) _lrotr(x, s)
+#  define LeftRotate(x, s)  _lrotl(x, s)
+#  if _MSC_VER >= 1400
+#   define SWAP(x) _byteswap_ulong(x)
+#  else
+#   define SWAP(x) (_lrotl(x, 8) & 0x00ff00ff | _lrotr(x, 8) & 0xff00ff00)
+#  endif
+#  define GETU32(p)   SWAP(*((u32 *)(p)))
+#  define PUTU32(p,v) (*((u32 *)(p)) = SWAP((v)))
+# elif defined(__GNUC__) && __GNUC__>=2
+#  if defined(__i386) || defined(__x86_64)
+#   define RightRotate(x,s) ({u32 ret; asm ("rorl %1,%0":"=r"(ret):"I"(s),"0"(x):"cc"); ret; })
+#   define LeftRotate(x,s)  ({u32 ret; asm ("roll %1,%0":"=r"(ret):"I"(s),"0"(x):"cc"); ret; })
+#   if defined(B_ENDIAN) /* stratus.com does it */
+#    define GETU32(p)   (*(u32 *)(p))
+#    define PUTU32(p,v) (*(u32 *)(p)=(v))
+#   else
+#    define GETU32(p)   ({u32 r=*(const u32 *)(p); asm("bswapl %0":"=r"(r):"0"(r)); r; })
+#    define PUTU32(p,v) ({u32 r=(v); asm("bswapl %0":"=r"(r):"0"(r)); *(u32 *)(p)=r; })
+#   endif
+#  elif defined(_ARCH_PPC) || defined(_ARCH_PPC64) || \
+        defined(__powerpc) || defined(__ppc__) || defined(__powerpc64__)
+#   define LeftRotate(x,s)  ({u32 ret; asm ("rlwinm %0,%1,%2,0,31":"=r"(ret):"r"(x),"I"(s)); ret; })
+#   define RightRotate(x,s) LeftRotate(x,(32-s))
+#  elif defined(__s390x__)
+#   define LeftRotate(x,s)  ({u32 ret; asm ("rll %0,%1,%2":"=r"(ret):"r"(x),"I"(s)); ret; })
+#   define RightRotate(x,s) LeftRotate(x,(32-s))
+#   define GETU32(p)   (*(u32 *)(p))
+#   define PUTU32(p,v) (*(u32 *)(p)=(v))
+#  endif
+# endif
+#endif
+
+#if !defined(RightRotate) && !defined(LeftRotate)
+# define RightRotate(x, s) ( ((x) >> (s)) + ((x) << (32 - s)) )
+# define LeftRotate(x, s)  ( ((x) << (s)) + ((x) >> (32 - s)) )
+#endif
+
+#if !defined(GETU32) && !defined(PUTU32)
+# define GETU32(p)   (((u32)(p)[0] << 24) ^ ((u32)(p)[1] << 16) ^ ((u32)(p)[2] <<  8) ^ ((u32)(p)[3]))
+# define PUTU32(p,v) ((p)[0] = (u8)((v) >> 24), (p)[1] = (u8)((v) >> 16), (p)[2] = (u8)((v) >>  8), (p)[3] = (u8)(v))
+#endif
+
+/* S-box data */
+#define SBOX1_1110 Camellia_SBOX[0]
+#define SBOX4_4404 Camellia_SBOX[1]
+#define SBOX2_0222 Camellia_SBOX[2]
+#define SBOX3_3033 Camellia_SBOX[3]
+static const u32 Camellia_SBOX[][256] = {
+{   0x70707000, 0x82828200, 0x2c2c2c00, 0xececec00, 0xb3b3b300, 0x27272700, 
+    0xc0c0c000, 0xe5e5e500, 0xe4e4e400, 0x85858500, 0x57575700, 0x35353500, 
+    0xeaeaea00, 0x0c0c0c00, 0xaeaeae00, 0x41414100, 0x23232300, 0xefefef00, 
+    0x6b6b6b00, 0x93939300, 0x45454500, 0x19191900, 0xa5a5a500, 0x21212100, 
+    0xededed00, 0x0e0e0e00, 0x4f4f4f00, 0x4e4e4e00, 0x1d1d1d00, 0x65656500, 
+    0x92929200, 0xbdbdbd00, 0x86868600, 0xb8b8b800, 0xafafaf00, 0x8f8f8f00, 
+    0x7c7c7c00, 0xebebeb00, 0x1f1f1f00, 0xcecece00, 0x3e3e3e00, 0x30303000, 
+    0xdcdcdc00, 0x5f5f5f00, 0x5e5e5e00, 0xc5c5c500, 0x0b0b0b00, 0x1a1a1a00, 
+    0xa6a6a600, 0xe1e1e100, 0x39393900, 0xcacaca00, 0xd5d5d500, 0x47474700, 
+    0x5d5d5d00, 0x3d3d3d00, 0xd9d9d900, 0x01010100, 0x5a5a5a00, 0xd6d6d600, 
+    0x51515100, 0x56565600, 0x6c6c6c00, 0x4d4d4d00, 0x8b8b8b00, 0x0d0d0d00, 
+    0x9a9a9a00, 0x66666600, 0xfbfbfb00, 0xcccccc00, 0xb0b0b000, 0x2d2d2d00, 
+    0x74747400, 0x12121200, 0x2b2b2b00, 0x20202000, 0xf0f0f000, 0xb1b1b100, 
+    0x84848400, 0x99999900, 0xdfdfdf00, 0x4c4c4c00, 0xcbcbcb00, 0xc2c2c200, 
+    0x34343400, 0x7e7e7e00, 0x76767600, 0x05050500, 0x6d6d6d00, 0xb7b7b700, 
+    0xa9a9a900, 0x31313100, 0xd1d1d100, 0x17171700, 0x04040400, 0xd7d7d700, 
+    0x14141400, 0x58585800, 0x3a3a3a00, 0x61616100, 0xdedede00, 0x1b1b1b00, 
+    0x11111100, 0x1c1c1c00, 0x32323200, 0x0f0f0f00, 0x9c9c9c00, 0x16161600, 
+    0x53535300, 0x18181800, 0xf2f2f200, 0x22222200, 0xfefefe00, 0x44444400, 
+    0xcfcfcf00, 0xb2b2b200, 0xc3c3c300, 0xb5b5b500, 0x7a7a7a00, 0x91919100, 
+    0x24242400, 0x08080800, 0xe8e8e800, 0xa8a8a800, 0x60606000, 0xfcfcfc00, 
+    0x69696900, 0x50505000, 0xaaaaaa00, 0xd0d0d000, 0xa0a0a000, 0x7d7d7d00, 
+    0xa1a1a100, 0x89898900, 0x62626200, 0x97979700, 0x54545400, 0x5b5b5b00, 
+    0x1e1e1e00, 0x95959500, 0xe0e0e000, 0xffffff00, 0x64646400, 0xd2d2d200, 
+    0x10101000, 0xc4c4c400, 0x00000000, 0x48484800, 0xa3a3a300, 0xf7f7f700, 
+    0x75757500, 0xdbdbdb00, 0x8a8a8a00, 0x03030300, 0xe6e6e600, 0xdadada00, 
+    0x09090900, 0x3f3f3f00, 0xdddddd00, 0x94949400, 0x87878700, 0x5c5c5c00, 
+    0x83838300, 0x02020200, 0xcdcdcd00, 0x4a4a4a00, 0x90909000, 0x33333300, 
+    0x73737300, 0x67676700, 0xf6f6f600, 0xf3f3f300, 0x9d9d9d00, 0x7f7f7f00, 
+    0xbfbfbf00, 0xe2e2e200, 0x52525200, 0x9b9b9b00, 0xd8d8d800, 0x26262600, 
+    0xc8c8c800, 0x37373700, 0xc6c6c600, 0x3b3b3b00, 0x81818100, 0x96969600, 
+    0x6f6f6f00, 0x4b4b4b00, 0x13131300, 0xbebebe00, 0x63636300, 0x2e2e2e00, 
+    0xe9e9e900, 0x79797900, 0xa7a7a700, 0x8c8c8c00, 0x9f9f9f00, 0x6e6e6e00, 
+    0xbcbcbc00, 0x8e8e8e00, 0x29292900, 0xf5f5f500, 0xf9f9f900, 0xb6b6b600, 
+    0x2f2f2f00, 0xfdfdfd00, 0xb4b4b400, 0x59595900, 0x78787800, 0x98989800, 
+    0x06060600, 0x6a6a6a00, 0xe7e7e700, 0x46464600, 0x71717100, 0xbababa00, 
+    0xd4d4d400, 0x25252500, 0xababab00, 0x42424200, 0x88888800, 0xa2a2a200, 
+    0x8d8d8d00, 0xfafafa00, 0x72727200, 0x07070700, 0xb9b9b900, 0x55555500, 
+    0xf8f8f800, 0xeeeeee00, 0xacacac00, 0x0a0a0a00, 0x36363600, 0x49494900, 
+    0x2a2a2a00, 0x68686800, 0x3c3c3c00, 0x38383800, 0xf1f1f100, 0xa4a4a400, 
+    0x40404000, 0x28282800, 0xd3d3d300, 0x7b7b7b00, 0xbbbbbb00, 0xc9c9c900, 
+    0x43434300, 0xc1c1c100, 0x15151500, 0xe3e3e300, 0xadadad00, 0xf4f4f400, 
+    0x77777700, 0xc7c7c700, 0x80808000, 0x9e9e9e00 },
+{   0x70700070, 0x2c2c002c, 0xb3b300b3, 0xc0c000c0, 0xe4e400e4, 0x57570057, 
+    0xeaea00ea, 0xaeae00ae, 0x23230023, 0x6b6b006b, 0x45450045, 0xa5a500a5, 
+    0xeded00ed, 0x4f4f004f, 0x1d1d001d, 0x92920092, 0x86860086, 0xafaf00af, 
+    0x7c7c007c, 0x1f1f001f, 0x3e3e003e, 0xdcdc00dc, 0x5e5e005e, 0x0b0b000b, 
+    0xa6a600a6, 0x39390039, 0xd5d500d5, 0x5d5d005d, 0xd9d900d9, 0x5a5a005a, 
+    0x51510051, 0x6c6c006c, 0x8b8b008b, 0x9a9a009a, 0xfbfb00fb, 0xb0b000b0, 
+    0x74740074, 0x2b2b002b, 0xf0f000f0, 0x84840084, 0xdfdf00df, 0xcbcb00cb, 
+    0x34340034, 0x76760076, 0x6d6d006d, 0xa9a900a9, 0xd1d100d1, 0x04040004, 
+    0x14140014, 0x3a3a003a, 0xdede00de, 0x11110011, 0x32320032, 0x9c9c009c, 
+    0x53530053, 0xf2f200f2, 0xfefe00fe, 0xcfcf00cf, 0xc3c300c3, 0x7a7a007a, 
+    0x24240024, 0xe8e800e8, 0x60600060, 0x69690069, 0xaaaa00aa, 0xa0a000a0, 
+    0xa1a100a1, 0x62620062, 0x54540054, 0x1e1e001e, 0xe0e000e0, 0x64640064, 
+    0x10100010, 0x00000000, 0xa3a300a3, 0x75750075, 0x8a8a008a, 0xe6e600e6, 
+    0x09090009, 0xdddd00dd, 0x87870087, 0x83830083, 0xcdcd00cd, 0x90900090, 
+    0x73730073, 0xf6f600f6, 0x9d9d009d, 0xbfbf00bf, 0x52520052, 0xd8d800d8, 
+    0xc8c800c8, 0xc6c600c6, 0x81810081, 0x6f6f006f, 0x13130013, 0x63630063, 
+    0xe9e900e9, 0xa7a700a7, 0x9f9f009f, 0xbcbc00bc, 0x29290029, 0xf9f900f9, 
+    0x2f2f002f, 0xb4b400b4, 0x78780078, 0x06060006, 0xe7e700e7, 0x71710071, 
+    0xd4d400d4, 0xabab00ab, 0x88880088, 0x8d8d008d, 0x72720072, 0xb9b900b9, 
+    0xf8f800f8, 0xacac00ac, 0x36360036, 0x2a2a002a, 0x3c3c003c, 0xf1f100f1, 
+    0x40400040, 0xd3d300d3, 0xbbbb00bb, 0x43430043, 0x15150015, 0xadad00ad, 
+    0x77770077, 0x80800080, 0x82820082, 0xecec00ec, 0x27270027, 0xe5e500e5, 
+    0x85850085, 0x35350035, 0x0c0c000c, 0x41410041, 0xefef00ef, 0x93930093, 
+    0x19190019, 0x21210021, 0x0e0e000e, 0x4e4e004e, 0x65650065, 0xbdbd00bd, 
+    0xb8b800b8, 0x8f8f008f, 0xebeb00eb, 0xcece00ce, 0x30300030, 0x5f5f005f, 
+    0xc5c500c5, 0x1a1a001a, 0xe1e100e1, 0xcaca00ca, 0x47470047, 0x3d3d003d, 
+    0x01010001, 0xd6d600d6, 0x56560056, 0x4d4d004d, 0x0d0d000d, 0x66660066, 
+    0xcccc00cc, 0x2d2d002d, 0x12120012, 0x20200020, 0xb1b100b1, 0x99990099, 
+    0x4c4c004c, 0xc2c200c2, 0x7e7e007e, 0x05050005, 0xb7b700b7, 0x31310031, 
+    0x17170017, 0xd7d700d7, 0x58580058, 0x61610061, 0x1b1b001b, 0x1c1c001c, 
+    0x0f0f000f, 0x16160016, 0x18180018, 0x22220022, 0x44440044, 0xb2b200b2, 
+    0xb5b500b5, 0x91910091, 0x08080008, 0xa8a800a8, 0xfcfc00fc, 0x50500050, 
+    0xd0d000d0, 0x7d7d007d, 0x89890089, 0x97970097, 0x5b5b005b, 0x95950095, 
+    0xffff00ff, 0xd2d200d2, 0xc4c400c4, 0x48480048, 0xf7f700f7, 0xdbdb00db, 
+    0x03030003, 0xdada00da, 0x3f3f003f, 0x94940094, 0x5c5c005c, 0x02020002, 
+    0x4a4a004a, 0x33330033, 0x67670067, 0xf3f300f3, 0x7f7f007f, 0xe2e200e2, 
+    0x9b9b009b, 0x26260026, 0x37370037, 0x3b3b003b, 0x96960096, 0x4b4b004b, 
+    0xbebe00be, 0x2e2e002e, 0x79790079, 0x8c8c008c, 0x6e6e006e, 0x8e8e008e, 
+    0xf5f500f5, 0xb6b600b6, 0xfdfd00fd, 0x59590059, 0x98980098, 0x6a6a006a, 
+    0x46460046, 0xbaba00ba, 0x25250025, 0x42420042, 0xa2a200a2, 0xfafa00fa, 
+    0x07070007, 0x55550055, 0xeeee00ee, 0x0a0a000a, 0x49490049, 0x68680068, 
+    0x38380038, 0xa4a400a4, 0x28280028, 0x7b7b007b, 0xc9c900c9, 0xc1c100c1, 
+    0xe3e300e3, 0xf4f400f4, 0xc7c700c7, 0x9e9e009e },
+{   0x00e0e0e0, 0x00050505, 0x00585858, 0x00d9d9d9, 0x00676767, 0x004e4e4e, 
+    0x00818181, 0x00cbcbcb, 0x00c9c9c9, 0x000b0b0b, 0x00aeaeae, 0x006a6a6a, 
+    0x00d5d5d5, 0x00181818, 0x005d5d5d, 0x00828282, 0x00464646, 0x00dfdfdf, 
+    0x00d6d6d6, 0x00272727, 0x008a8a8a, 0x00323232, 0x004b4b4b, 0x00424242, 
+    0x00dbdbdb, 0x001c1c1c, 0x009e9e9e, 0x009c9c9c, 0x003a3a3a, 0x00cacaca, 
+    0x00252525, 0x007b7b7b, 0x000d0d0d, 0x00717171, 0x005f5f5f, 0x001f1f1f, 
+    0x00f8f8f8, 0x00d7d7d7, 0x003e3e3e, 0x009d9d9d, 0x007c7c7c, 0x00606060, 
+    0x00b9b9b9, 0x00bebebe, 0x00bcbcbc, 0x008b8b8b, 0x00161616, 0x00343434, 
+    0x004d4d4d, 0x00c3c3c3, 0x00727272, 0x00959595, 0x00ababab, 0x008e8e8e, 
+    0x00bababa, 0x007a7a7a, 0x00b3b3b3, 0x00020202, 0x00b4b4b4, 0x00adadad, 
+    0x00a2a2a2, 0x00acacac, 0x00d8d8d8, 0x009a9a9a, 0x00171717, 0x001a1a1a, 
+    0x00353535, 0x00cccccc, 0x00f7f7f7, 0x00999999, 0x00616161, 0x005a5a5a, 
+    0x00e8e8e8, 0x00242424, 0x00565656, 0x00404040, 0x00e1e1e1, 0x00636363, 
+    0x00090909, 0x00333333, 0x00bfbfbf, 0x00989898, 0x00979797, 0x00858585, 
+    0x00686868, 0x00fcfcfc, 0x00ececec, 0x000a0a0a, 0x00dadada, 0x006f6f6f, 
+    0x00535353, 0x00626262, 0x00a3a3a3, 0x002e2e2e, 0x00080808, 0x00afafaf, 
+    0x00282828, 0x00b0b0b0, 0x00747474, 0x00c2c2c2, 0x00bdbdbd, 0x00363636, 
+    0x00222222, 0x00383838, 0x00646464, 0x001e1e1e, 0x00393939, 0x002c2c2c, 
+    0x00a6a6a6, 0x00303030, 0x00e5e5e5, 0x00444444, 0x00fdfdfd, 0x00888888, 
+    0x009f9f9f, 0x00656565, 0x00878787, 0x006b6b6b, 0x00f4f4f4, 0x00232323, 
+    0x00484848, 0x00101010, 0x00d1d1d1, 0x00515151, 0x00c0c0c0, 0x00f9f9f9, 
+    0x00d2d2d2, 0x00a0a0a0, 0x00555555, 0x00a1a1a1, 0x00414141, 0x00fafafa, 
+    0x00434343, 0x00131313, 0x00c4c4c4, 0x002f2f2f, 0x00a8a8a8, 0x00b6b6b6, 
+    0x003c3c3c, 0x002b2b2b, 0x00c1c1c1, 0x00ffffff, 0x00c8c8c8, 0x00a5a5a5, 
+    0x00202020, 0x00898989, 0x00000000, 0x00909090, 0x00474747, 0x00efefef, 
+    0x00eaeaea, 0x00b7b7b7, 0x00151515, 0x00060606, 0x00cdcdcd, 0x00b5b5b5, 
+    0x00121212, 0x007e7e7e, 0x00bbbbbb, 0x00292929, 0x000f0f0f, 0x00b8b8b8, 
+    0x00070707, 0x00040404, 0x009b9b9b, 0x00949494, 0x00212121, 0x00666666, 
+    0x00e6e6e6, 0x00cecece, 0x00ededed, 0x00e7e7e7, 0x003b3b3b, 0x00fefefe, 
+    0x007f7f7f, 0x00c5c5c5, 0x00a4a4a4, 0x00373737, 0x00b1b1b1, 0x004c4c4c, 
+    0x00919191, 0x006e6e6e, 0x008d8d8d, 0x00767676, 0x00030303, 0x002d2d2d, 
+    0x00dedede, 0x00969696, 0x00262626, 0x007d7d7d, 0x00c6c6c6, 0x005c5c5c, 
+    0x00d3d3d3, 0x00f2f2f2, 0x004f4f4f, 0x00191919, 0x003f3f3f, 0x00dcdcdc, 
+    0x00797979, 0x001d1d1d, 0x00525252, 0x00ebebeb, 0x00f3f3f3, 0x006d6d6d, 
+    0x005e5e5e, 0x00fbfbfb, 0x00696969, 0x00b2b2b2, 0x00f0f0f0, 0x00313131, 
+    0x000c0c0c, 0x00d4d4d4, 0x00cfcfcf, 0x008c8c8c, 0x00e2e2e2, 0x00757575, 
+    0x00a9a9a9, 0x004a4a4a, 0x00575757, 0x00848484, 0x00111111, 0x00454545, 
+    0x001b1b1b, 0x00f5f5f5, 0x00e4e4e4, 0x000e0e0e, 0x00737373, 0x00aaaaaa, 
+    0x00f1f1f1, 0x00dddddd, 0x00595959, 0x00141414, 0x006c6c6c, 0x00929292, 
+    0x00545454, 0x00d0d0d0, 0x00787878, 0x00707070, 0x00e3e3e3, 0x00494949, 
+    0x00808080, 0x00505050, 0x00a7a7a7, 0x00f6f6f6, 0x00777777, 0x00939393, 
+    0x00868686, 0x00838383, 0x002a2a2a, 0x00c7c7c7, 0x005b5b5b, 0x00e9e9e9, 
+    0x00eeeeee, 0x008f8f8f, 0x00010101, 0x003d3d3d },
+{   0x38003838, 0x41004141, 0x16001616, 0x76007676, 0xd900d9d9, 0x93009393, 
+    0x60006060, 0xf200f2f2, 0x72007272, 0xc200c2c2, 0xab00abab, 0x9a009a9a, 
+    0x75007575, 0x06000606, 0x57005757, 0xa000a0a0, 0x91009191, 0xf700f7f7, 
+    0xb500b5b5, 0xc900c9c9, 0xa200a2a2, 0x8c008c8c, 0xd200d2d2, 0x90009090, 
+    0xf600f6f6, 0x07000707, 0xa700a7a7, 0x27002727, 0x8e008e8e, 0xb200b2b2, 
+    0x49004949, 0xde00dede, 0x43004343, 0x5c005c5c, 0xd700d7d7, 0xc700c7c7, 
+    0x3e003e3e, 0xf500f5f5, 0x8f008f8f, 0x67006767, 0x1f001f1f, 0x18001818, 
+    0x6e006e6e, 0xaf00afaf, 0x2f002f2f, 0xe200e2e2, 0x85008585, 0x0d000d0d, 
+    0x53005353, 0xf000f0f0, 0x9c009c9c, 0x65006565, 0xea00eaea, 0xa300a3a3, 
+    0xae00aeae, 0x9e009e9e, 0xec00ecec, 0x80008080, 0x2d002d2d, 0x6b006b6b, 
+    0xa800a8a8, 0x2b002b2b, 0x36003636, 0xa600a6a6, 0xc500c5c5, 0x86008686, 
+    0x4d004d4d, 0x33003333, 0xfd00fdfd, 0x66006666, 0x58005858, 0x96009696, 
+    0x3a003a3a, 0x09000909, 0x95009595, 0x10001010, 0x78007878, 0xd800d8d8, 
+    0x42004242, 0xcc00cccc, 0xef00efef, 0x26002626, 0xe500e5e5, 0x61006161, 
+    0x1a001a1a, 0x3f003f3f, 0x3b003b3b, 0x82008282, 0xb600b6b6, 0xdb00dbdb, 
+    0xd400d4d4, 0x98009898, 0xe800e8e8, 0x8b008b8b, 0x02000202, 0xeb00ebeb, 
+    0x0a000a0a, 0x2c002c2c, 0x1d001d1d, 0xb000b0b0, 0x6f006f6f, 0x8d008d8d, 
+    0x88008888, 0x0e000e0e, 0x19001919, 0x87008787, 0x4e004e4e, 0x0b000b0b, 
+    0xa900a9a9, 0x0c000c0c, 0x79007979, 0x11001111, 0x7f007f7f, 0x22002222, 
+    0xe700e7e7, 0x59005959, 0xe100e1e1, 0xda00dada, 0x3d003d3d, 0xc800c8c8, 
+    0x12001212, 0x04000404, 0x74007474, 0x54005454, 0x30003030, 0x7e007e7e, 
+    0xb400b4b4, 0x28002828, 0x55005555, 0x68006868, 0x50005050, 0xbe00bebe, 
+    0xd000d0d0, 0xc400c4c4, 0x31003131, 0xcb00cbcb, 0x2a002a2a, 0xad00adad, 
+    0x0f000f0f, 0xca00caca, 0x70007070, 0xff00ffff, 0x32003232, 0x69006969, 
+    0x08000808, 0x62006262, 0x00000000, 0x24002424, 0xd100d1d1, 0xfb00fbfb, 
+    0xba00baba, 0xed00eded, 0x45004545, 0x81008181, 0x73007373, 0x6d006d6d, 
+    0x84008484, 0x9f009f9f, 0xee00eeee, 0x4a004a4a, 0xc300c3c3, 0x2e002e2e, 
+    0xc100c1c1, 0x01000101, 0xe600e6e6, 0x25002525, 0x48004848, 0x99009999, 
+    0xb900b9b9, 0xb300b3b3, 0x7b007b7b, 0xf900f9f9, 0xce00cece, 0xbf00bfbf, 
+    0xdf00dfdf, 0x71007171, 0x29002929, 0xcd00cdcd, 0x6c006c6c, 0x13001313, 
+    0x64006464, 0x9b009b9b, 0x63006363, 0x9d009d9d, 0xc000c0c0, 0x4b004b4b, 
+    0xb700b7b7, 0xa500a5a5, 0x89008989, 0x5f005f5f, 0xb100b1b1, 0x17001717, 
+    0xf400f4f4, 0xbc00bcbc, 0xd300d3d3, 0x46004646, 0xcf00cfcf, 0x37003737, 
+    0x5e005e5e, 0x47004747, 0x94009494, 0xfa00fafa, 0xfc00fcfc, 0x5b005b5b, 
+    0x97009797, 0xfe00fefe, 0x5a005a5a, 0xac00acac, 0x3c003c3c, 0x4c004c4c, 
+    0x03000303, 0x35003535, 0xf300f3f3, 0x23002323, 0xb800b8b8, 0x5d005d5d, 
+    0x6a006a6a, 0x92009292, 0xd500d5d5, 0x21002121, 0x44004444, 0x51005151, 
+    0xc600c6c6, 0x7d007d7d, 0x39003939, 0x83008383, 0xdc00dcdc, 0xaa00aaaa, 
+    0x7c007c7c, 0x77007777, 0x56005656, 0x05000505, 0x1b001b1b, 0xa400a4a4, 
+    0x15001515, 0x34003434, 0x1e001e1e, 0x1c001c1c, 0xf800f8f8, 0x52005252, 
+    0x20002020, 0x14001414, 0xe900e9e9, 0xbd00bdbd, 0xdd00dddd, 0xe400e4e4, 
+    0xa100a1a1, 0xe000e0e0, 0x8a008a8a, 0xf100f1f1, 0xd600d6d6, 0x7a007a7a, 
+    0xbb00bbbb, 0xe300e3e3, 0x40004040, 0x4f004f4f }
+};
+
+/* Key generation constants */
+static const u32 SIGMA[] = {
+    0xa09e667f, 0x3bcc908b, 0xb67ae858, 0x4caa73b2, 0xc6ef372f, 0xe94f82be,
+    0x54ff53a5, 0xf1d36f1c, 0x10e527fa, 0xde682d1d, 0xb05688c2, 0xb3e6c1fd
+};
+
+/* The phi algorithm given in C.2.7 of the Camellia spec document. */
+/*
+ * This version does not attempt to minimize amount of temporary
+ * variables, but instead explicitly exposes algorithm's parallelism.
+ * It is therefore most appropriate for platforms with not less than
+ * ~16 registers. For platforms with less registers [well, x86 to be
+ * specific] assembler version should be/is provided anyway...
+ */
+#define Camellia_Feistel(_s0,_s1,_s2,_s3,_key) do {\
+	register u32 _t0,_t1,_t2,_t3;\
+\
+	_t0  = _s0 ^ (_key)[0];\
+	_t3  = SBOX4_4404[_t0&0xff];\
+	_t1  = _s1 ^ (_key)[1];\
+	_t3 ^= SBOX3_3033[(_t0 >> 8)&0xff];\
+	_t2  = SBOX1_1110[_t1&0xff];\
+	_t3 ^= SBOX2_0222[(_t0 >> 16)&0xff];\
+	_t2 ^= SBOX4_4404[(_t1 >> 8)&0xff];\
+	_t3 ^= SBOX1_1110[(_t0 >> 24)];\
+	_t2 ^= _t3;\
+	_t3  = RightRotate(_t3,8);\
+	_t2 ^= SBOX3_3033[(_t1 >> 16)&0xff];\
+	_s3 ^= _t3;\
+	_t2 ^= SBOX2_0222[(_t1 >> 24)];\
+	_s2 ^= _t2; \
+	_s3 ^= _t2;\
+} while(0)
+
+/*
+ * Note that n has to be less than 32. Rotations for larger amount
+ * of bits are achieved by "rotating" order of s-elements and
+ * adjusting n accordingly, e.g. RotLeft128(s1,s2,s3,s0,n-32).
+ */
+#define RotLeft128(_s0,_s1,_s2,_s3,_n) do {\
+	u32 _t0=_s0>>(32-_n);\
+	_s0 = (_s0<<_n) | (_s1>>(32-_n));\
+	_s1 = (_s1<<_n) | (_s2>>(32-_n));\
+	_s2 = (_s2<<_n) | (_s3>>(32-_n));\
+	_s3 = (_s3<<_n) | _t0;\
+} while (0)
+
+int Camellia_Ekeygen(int keyBitLength, const u8 *rawKey, KEY_TABLE_TYPE k)
+	{
+	register u32 s0,s1,s2,s3;
+
+	k[0] = s0 = GETU32(rawKey);
+	k[1] = s1 = GETU32(rawKey+4);
+	k[2] = s2 = GETU32(rawKey+8);
+	k[3] = s3 = GETU32(rawKey+12);
+
+	if (keyBitLength != 128)
+		{
+		k[8] = s0 = GETU32(rawKey+16);
+		k[9] = s1 = GETU32(rawKey+20);
+		if (keyBitLength == 192)
+			{
+			k[10] = s2 = ~s0;
+			k[11] = s3 = ~s1;
+			}
+		else
+			{
+			k[10] = s2 = GETU32(rawKey+24);
+			k[11] = s3 = GETU32(rawKey+28);
+			}
+		s0 ^= k[0], s1 ^= k[1], s2 ^= k[2], s3 ^= k[3];
+		}
+
+	/* Use the Feistel routine to scramble the key material */
+	Camellia_Feistel(s0,s1,s2,s3,SIGMA+0);
+	Camellia_Feistel(s2,s3,s0,s1,SIGMA+2);
+
+	s0 ^= k[0], s1 ^= k[1], s2 ^= k[2], s3 ^= k[3];
+	Camellia_Feistel(s0,s1,s2,s3,SIGMA+4);
+	Camellia_Feistel(s2,s3,s0,s1,SIGMA+6);
+
+	/* Fill the keyTable. Requires many block rotations. */
+	if (keyBitLength == 128)
+		{
+		k[ 4] = s0, k[ 5] = s1, k[ 6] = s2, k[ 7] = s3;
+		RotLeft128(s0,s1,s2,s3,15);	/* KA <<< 15 */
+		k[12] = s0, k[13] = s1, k[14] = s2, k[15] = s3;
+		RotLeft128(s0,s1,s2,s3,15);	/* KA <<< 30 */
+		k[16] = s0, k[17] = s1, k[18] = s2, k[19] = s3;
+		RotLeft128(s0,s1,s2,s3,15);	/* KA <<< 45 */
+		k[24] = s0, k[25] = s1;
+		RotLeft128(s0,s1,s2,s3,15);	/* KA <<< 60 */
+		k[28] = s0, k[29] = s1, k[30] = s2, k[31] = s3;
+		RotLeft128(s1,s2,s3,s0,2);	/* KA <<< 94 */
+		k[40] = s1, k[41] = s2, k[42] = s3, k[43] = s0;
+		RotLeft128(s1,s2,s3,s0,17);	/* KA <<<111 */
+		k[48] = s1, k[49] = s2, k[50] = s3, k[51] = s0;
+
+		s0 = k[ 0], s1 = k[ 1], s2 = k[ 2], s3 = k[ 3];
+		RotLeft128(s0,s1,s2,s3,15);	/* KL <<< 15 */
+		k[ 8] = s0, k[ 9] = s1, k[10] = s2, k[11] = s3;
+		RotLeft128(s0,s1,s2,s3,30);	/* KL <<< 45 */
+		k[20] = s0, k[21] = s1, k[22] = s2, k[23] = s3;
+		RotLeft128(s0,s1,s2,s3,15);	/* KL <<< 60 */
+		k[26] = s2, k[27] = s3;
+		RotLeft128(s0,s1,s2,s3,17);	/* KL <<< 77 */
+		k[32] = s0, k[33] = s1, k[34] = s2, k[35] = s3;
+		RotLeft128(s0,s1,s2,s3,17);	/* KL <<< 94 */
+		k[36] = s0, k[37] = s1, k[38] = s2, k[39] = s3;
+		RotLeft128(s0,s1,s2,s3,17);	/* KL <<<111 */
+		k[44] = s0, k[45] = s1, k[46] = s2, k[47] = s3;
+
+		return 3;	/* grand rounds */
+		}
+	else
+		{
+		k[12] = s0, k[13] = s1, k[14] = s2, k[15] = s3;
+		s0 ^= k[8], s1 ^= k[9], s2 ^=k[10], s3 ^=k[11];
+		Camellia_Feistel(s0,s1,s2,s3,(SIGMA+8));
+		Camellia_Feistel(s2,s3,s0,s1,(SIGMA+10));
+
+		k[ 4] = s0, k[ 5] = s1, k[ 6] = s2, k[ 7] = s3;
+		RotLeft128(s0,s1,s2,s3,30);	/* KB <<< 30 */
+		k[20] = s0, k[21] = s1, k[22] = s2, k[23] = s3;
+		RotLeft128(s0,s1,s2,s3,30);	/* KB <<< 60 */
+		k[40] = s0, k[41] = s1, k[42] = s2, k[43] = s3;
+		RotLeft128(s1,s2,s3,s0,19);	/* KB <<<111 */
+		k[64] = s1, k[65] = s2, k[66] = s3, k[67] = s0;
+
+		s0 = k[ 8], s1 = k[ 9], s2 = k[10], s3 = k[11];
+		RotLeft128(s0,s1,s2,s3,15);	/* KR <<< 15 */
+		k[ 8] = s0, k[ 9] = s1, k[10] = s2, k[11] = s3;
+		RotLeft128(s0,s1,s2,s3,15);	/* KR <<< 30 */
+		k[16] = s0, k[17] = s1, k[18] = s2, k[19] = s3;
+		RotLeft128(s0,s1,s2,s3,30);	/* KR <<< 60 */
+		k[36] = s0, k[37] = s1, k[38] = s2, k[39] = s3;
+		RotLeft128(s1,s2,s3,s0,2);	/* KR <<< 94 */
+		k[52] = s1, k[53] = s2, k[54] = s3, k[55] = s0;
+
+		s0 = k[12], s1 = k[13], s2 = k[14], s3 = k[15];
+		RotLeft128(s0,s1,s2,s3,15);	/* KA <<< 15 */
+		k[12] = s0, k[13] = s1, k[14] = s2, k[15] = s3;
+		RotLeft128(s0,s1,s2,s3,30);	/* KA <<< 45 */
+		k[28] = s0, k[29] = s1, k[30] = s2, k[31] = s3;
+						/* KA <<< 77 */
+		k[48] = s1, k[49] = s2, k[50] = s3, k[51] = s0;
+		RotLeft128(s1,s2,s3,s0,17);	/* KA <<< 94 */
+		k[56] = s1, k[57] = s2, k[58] = s3, k[59] = s0;
+
+		s0 = k[ 0], s1 = k[ 1], s2 = k[ 2], s3 = k[ 3];
+		RotLeft128(s1,s2,s3,s0,13);	/* KL <<< 45 */
+		k[24] = s1, k[25] = s2, k[26] = s3, k[27] = s0;
+		RotLeft128(s1,s2,s3,s0,15);	/* KL <<< 60 */
+		k[32] = s1, k[33] = s2, k[34] = s3, k[35] = s0;
+		RotLeft128(s1,s2,s3,s0,17);	/* KL <<< 77 */
+		k[44] = s1, k[45] = s2, k[46] = s3, k[47] = s0;
+		RotLeft128(s2,s3,s0,s1,2);	/* KL <<<111 */
+		k[60] = s2, k[61] = s3, k[62] = s0, k[63] = s1;
+
+		return 4;	/* grand rounds */
+		}
+	/*
+	 * It is possible to perform certain precalculations, which
+	 * would spare few cycles in block procedure. It's not done,
+	 * because it upsets the performance balance between key
+	 * setup and block procedures, negatively affecting overall
+	 * throughput in applications operating on short messages
+	 * and volatile keys.
+	 */ 
+	}
+
+void Camellia_EncryptBlock_Rounds(int grandRounds, const u8 plaintext[], 
+		const KEY_TABLE_TYPE keyTable, u8 ciphertext[])
+	{
+	register u32 s0,s1,s2,s3; 
+	const u32 *k = keyTable,*kend = keyTable+grandRounds*16; 
+
+	s0 = GETU32(plaintext)    ^ k[0];
+	s1 = GETU32(plaintext+4)  ^ k[1];
+	s2 = GETU32(plaintext+8)  ^ k[2];
+	s3 = GETU32(plaintext+12) ^ k[3];
+	k += 4;
+
+	while (1)
+		{
+		/* Camellia makes 6 Feistel rounds */
+		Camellia_Feistel(s0,s1,s2,s3,k+0);
+		Camellia_Feistel(s2,s3,s0,s1,k+2);
+		Camellia_Feistel(s0,s1,s2,s3,k+4);
+		Camellia_Feistel(s2,s3,s0,s1,k+6);
+		Camellia_Feistel(s0,s1,s2,s3,k+8);
+		Camellia_Feistel(s2,s3,s0,s1,k+10);
+		k += 12;
+
+		if (k == kend) break;
+
+		/* This is the same function as the diffusion function D
+		 * of the accompanying documentation. See section 3.2
+		 * for properties of the FLlayer function. */
+		s1 ^= LeftRotate(s0 & k[0], 1);
+		s2 ^= s3 | k[3];
+		s0 ^= s1 | k[1];
+		s3 ^= LeftRotate(s2 & k[2], 1);
+		k += 4;
+		}
+
+	s2 ^= k[0], s3 ^= k[1], s0 ^= k[2], s1 ^= k[3];
+
+	PUTU32(ciphertext,   s2);
+	PUTU32(ciphertext+4, s3);
+	PUTU32(ciphertext+8, s0);
+	PUTU32(ciphertext+12,s1);
+	}
+void Camellia_EncryptBlock(int keyBitLength, const u8 plaintext[], 
+		const KEY_TABLE_TYPE keyTable, u8 ciphertext[])
+	{
+	Camellia_EncryptBlock_Rounds(keyBitLength==128?3:4,
+			plaintext,keyTable,ciphertext);
+	}
+
+void Camellia_DecryptBlock_Rounds(int grandRounds, const u8 ciphertext[], 
+		const KEY_TABLE_TYPE keyTable, u8 plaintext[])
+	{
+	u32 s0,s1,s2,s3; 
+	const u32 *k = keyTable+grandRounds*16,*kend = keyTable+4; 
+
+	s0 = GETU32(ciphertext)    ^ k[0];
+	s1 = GETU32(ciphertext+4)  ^ k[1];
+	s2 = GETU32(ciphertext+8)  ^ k[2];
+	s3 = GETU32(ciphertext+12) ^ k[3];
+
+	while (1)
+		{
+		/* Camellia makes 6 Feistel rounds */
+		k -= 12;
+		Camellia_Feistel(s0,s1,s2,s3,k+10);
+		Camellia_Feistel(s2,s3,s0,s1,k+8);
+		Camellia_Feistel(s0,s1,s2,s3,k+6);
+		Camellia_Feistel(s2,s3,s0,s1,k+4);
+		Camellia_Feistel(s0,s1,s2,s3,k+2);
+		Camellia_Feistel(s2,s3,s0,s1,k+0);
+
+		if (k == kend) break;
+
+		/* This is the same function as the diffusion function D
+		 * of the accompanying documentation. See section 3.2
+		 * for properties of the FLlayer function. */
+		k -= 4;
+		s1 ^= LeftRotate(s0 & k[2], 1);
+		s2 ^= s3 | k[1];
+		s0 ^= s1 | k[3];
+		s3 ^= LeftRotate(s2 & k[0], 1);
+		}
+
+	k -= 4;
+	s2 ^= k[0], s3 ^= k[1], s0 ^= k[2], s1 ^= k[3];
+
+	PUTU32(plaintext,   s2);
+	PUTU32(plaintext+4, s3);
+	PUTU32(plaintext+8, s0);
+	PUTU32(plaintext+12,s1);
+	}
+void Camellia_DecryptBlock(int keyBitLength, const u8 plaintext[], 
+		const KEY_TABLE_TYPE keyTable, u8 ciphertext[])
+	{
+	Camellia_DecryptBlock_Rounds(keyBitLength==128?3:4,
+			plaintext,keyTable,ciphertext);
+	}
diff --git a/crypto/camellia/camellia.h b/crypto/camellia/camellia.h
new file mode 100644
index 0000000000..cf0457dd97
--- /dev/null
+++ b/crypto/camellia/camellia.h
@@ -0,0 +1,126 @@
+/* crypto/camellia/camellia.h -*- mode:C; c-file-style: "eay" -*- */
+/* ====================================================================
+ * Copyright (c) 2006 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer. 
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ *
+ */
+
+#ifndef HEADER_CAMELLIA_H
+#define HEADER_CAMELLIA_H
+
+#include <openssl/opensslconf.h>
+
+#ifdef OPENSSL_NO_CAMELLIA
+#error CAMELLIA is disabled.
+#endif
+
+#include <stddef.h>
+
+#define CAMELLIA_ENCRYPT	1
+#define CAMELLIA_DECRYPT	0
+
+/* Because array size can't be a const in C, the following two are macros.
+   Both sizes are in bytes. */
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+/* This should be a hidden type, but EVP requires that the size be known */
+
+#define CAMELLIA_BLOCK_SIZE 16
+#define CAMELLIA_TABLE_BYTE_LEN 272
+#define CAMELLIA_TABLE_WORD_LEN (CAMELLIA_TABLE_BYTE_LEN / 4)
+
+typedef unsigned int KEY_TABLE_TYPE[CAMELLIA_TABLE_WORD_LEN]; /* to match with WORD */
+
+struct camellia_key_st 
+	{
+	union	{
+		double d;	/* ensures 64-bit align */
+		KEY_TABLE_TYPE rd_key;
+		} u;
+	int grand_rounds;
+	};
+typedef struct camellia_key_st CAMELLIA_KEY;
+
+int Camellia_set_key(const unsigned char *userKey, const int bits,
+	CAMELLIA_KEY *key);
+
+void Camellia_encrypt(const unsigned char *in, unsigned char *out,
+	const CAMELLIA_KEY *key);
+void Camellia_decrypt(const unsigned char *in, unsigned char *out,
+	const CAMELLIA_KEY *key);
+
+void Camellia_ecb_encrypt(const unsigned char *in, unsigned char *out,
+	const CAMELLIA_KEY *key, const int enc);
+void Camellia_cbc_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char *ivec, const int enc);
+void Camellia_cfb128_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char *ivec, int *num, const int enc);
+void Camellia_cfb1_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char *ivec, int *num, const int enc);
+void Camellia_cfb8_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char *ivec, int *num, const int enc);
+void Camellia_ofb128_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char *ivec, int *num);
+void Camellia_ctr128_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char ivec[CAMELLIA_BLOCK_SIZE],
+	unsigned char ecount_buf[CAMELLIA_BLOCK_SIZE],
+	unsigned int *num);
+
+#ifdef  __cplusplus
+}
+#endif
+
+#endif /* !HEADER_Camellia_H */
diff --git a/crypto/camellia/cmll_cbc.c b/crypto/camellia/cmll_cbc.c
new file mode 100644
index 0000000000..4c8d455ade
--- /dev/null
+++ b/crypto/camellia/cmll_cbc.c
@@ -0,0 +1,64 @@
+/* crypto/camellia/camellia_cbc.c -*- mode:C; c-file-style: "eay" -*- */
+/* ====================================================================
+ * Copyright (c) 2006 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer. 
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ *
+ */
+
+#include <openssl/camellia.h>
+#include <openssl/modes.h>
+
+void Camellia_cbc_encrypt(const unsigned char *in, unsigned char *out,
+	size_t len, const CAMELLIA_KEY *key,
+	unsigned char *ivec, const int enc) 
+	{
+
+	if (enc)
+		CRYPTO_cbc128_encrypt(in,out,len,key,ivec,(block128_f)Camellia_encrypt);
+	else
+		CRYPTO_cbc128_decrypt(in,out,len,key,ivec,(block128_f)Camellia_decrypt);
+	}
diff --git a/crypto/camellia/cmll_cfb.c b/crypto/camellia/cmll_cfb.c
new file mode 100644
index 0000000000..3d81b51d3f
--- /dev/null
+++ b/crypto/camellia/cmll_cfb.c
@@ -0,0 +1,139 @@
+/* crypto/camellia/camellia_cfb.c -*- mode:C; c-file-style: "eay" -*- */
+/* ====================================================================
+ * Copyright (c) 2006 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer. 
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ *
+ */
+/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
+ * All rights reserved.
+ *
+ * This package is an SSL implementation written
+ * by Eric Young (eay@cryptsoft.com).
+ * The implementation was written so as to conform with Netscapes SSL.
+ * 
+ * This library is free for commercial and non-commercial use as long as
+ * the following conditions are aheared to.  The following conditions
+ * apply to all code found in this distribution, be it the RC4, RSA,
+ * lhash, DES, etc., code; not just the SSL code.  The SSL documentation
+ * included with this distribution is covered by the same copyright terms
+ * except that the holder is Tim Hudson (tjh@cryptsoft.com).
+ * 
+ * Copyright remains Eric Young's, and as such any Copyright notices in
+ * the code are not to be removed.
+ * If this package is used in a product, Eric Young should be given attribution
+ * as the author of the parts of the library used.
+ * This can be in the form of a textual message at program startup or
+ * in documentation (online or textual) provided with the package.
+ * 
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. All advertising materials mentioning features or use of this software
+ *    must display the following acknowledgement:
+ *    "This product includes cryptographic software written by
+ *     Eric Young (eay@cryptsoft.com)"
+ *    The word 'cryptographic' can be left out if the rouines from the library
+ *    being used are not cryptographic related :-).
+ * 4. If you include any Windows specific code (or a derivative thereof) from 
+ *    the apps directory (application code) you must include an acknowledgement:
+ *    "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
+ * 
+ * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ * 
+ * The licence and distribution terms for any publically available version or
+ * derivative of this code cannot be changed.  i.e. this code cannot simply be
+ * copied and put under another distribution licence
+ * [including the GNU Public Licence.]
+ */
+
+#include <openssl/camellia.h>
+#include <openssl/modes.h>
+
+
+/* The input and output encrypted as though 128bit cfb mode is being
+ * used.  The extra state information to record how much of the
+ * 128bit block we have used is contained in *num;
+ */
+
+void Camellia_cfb128_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char *ivec, int *num, const int enc)
+	{
+
+	CRYPTO_cfb128_encrypt(in,out,length,key,ivec,num,enc,(block128_f)Camellia_encrypt);
+	}
+
+/* N.B. This expects the input to be packed, MS bit first */
+void Camellia_cfb1_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char *ivec, int *num, const int enc)
+	{
+	CRYPTO_cfb128_1_encrypt(in,out,length,key,ivec,num,enc,(block128_f)Camellia_encrypt);
+	}
+
+void Camellia_cfb8_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char *ivec, int *num, const int enc)
+	{
+	CRYPTO_cfb128_8_encrypt(in,out,length,key,ivec,num,enc,(block128_f)Camellia_encrypt);
+	}
+
diff --git a/crypto/camellia/cmll_ctr.c b/crypto/camellia/cmll_ctr.c
new file mode 100644
index 0000000000..014e621a34
--- /dev/null
+++ b/crypto/camellia/cmll_ctr.c
@@ -0,0 +1,64 @@
+/* crypto/camellia/camellia_ctr.c -*- mode:C; c-file-style: "eay" -*- */
+/* ====================================================================
+ * Copyright (c) 2006 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer. 
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ *
+ */
+
+#include <openssl/camellia.h>
+#include <openssl/modes.h>
+
+void Camellia_ctr128_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char ivec[CAMELLIA_BLOCK_SIZE],
+	unsigned char ecount_buf[CAMELLIA_BLOCK_SIZE],
+	unsigned int *num) 
+	{
+
+	CRYPTO_ctr128_encrypt(in,out,length,key,ivec,ecount_buf,num,(block128_f)Camellia_encrypt);
+	}
+
diff --git a/crypto/camellia/cmll_ecb.c b/crypto/camellia/cmll_ecb.c
new file mode 100644
index 0000000000..70dc0e5632
--- /dev/null
+++ b/crypto/camellia/cmll_ecb.c
@@ -0,0 +1,74 @@
+/* crypto/camellia/camellia_ecb.c -*- mode:C; c-file-style: "eay" -*- */
+/* ====================================================================
+ * Copyright (c) 2006 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer. 
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ *
+ */
+
+#ifndef CAMELLIA_DEBUG
+# ifndef NDEBUG
+#  define NDEBUG
+# endif
+#endif
+#include <assert.h>
+
+#include <openssl/camellia.h>
+#include "cmll_locl.h"
+
+void Camellia_ecb_encrypt(const unsigned char *in, unsigned char *out,
+	const CAMELLIA_KEY *key, const int enc) 
+	{
+
+	assert(in && out && key);
+	assert((CAMELLIA_ENCRYPT == enc)||(CAMELLIA_DECRYPT == enc));
+
+	if (CAMELLIA_ENCRYPT == enc)
+		Camellia_encrypt(in, out, key);
+	else
+		Camellia_decrypt(in, out, key);
+	}
+
diff --git a/crypto/camellia/cmll_locl.h b/crypto/camellia/cmll_locl.h
new file mode 100644
index 0000000000..4a4d880d16
--- /dev/null
+++ b/crypto/camellia/cmll_locl.h
@@ -0,0 +1,83 @@
+/* crypto/camellia/camellia_locl.h -*- mode:C; c-file-style: "eay" -*- */
+/* ====================================================================
+ * Copyright 2006 NTT (Nippon Telegraph and Telephone Corporation) . 
+ * ALL RIGHTS RESERVED.
+ *
+ * Intellectual Property information for Camellia:
+ *     http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html
+ *
+ * News Release for Announcement of Camellia open source:
+ *     http://www.ntt.co.jp/news/news06e/0604/060413a.html
+ *
+ * The Camellia Code included herein is developed by
+ * NTT (Nippon Telegraph and Telephone Corporation), and is contributed
+ * to the OpenSSL project.
+ *
+ * The Camellia Code is licensed pursuant to the OpenSSL open source
+ * license provided below.
+ */
+/* ====================================================================
+ * Copyright (c) 2006 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer. 
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ */
+
+#ifndef HEADER_CAMELLIA_LOCL_H
+#define HEADER_CAMELLIA_LOCL_H
+
+typedef unsigned int  u32;
+typedef unsigned char u8;
+
+int Camellia_Ekeygen(int keyBitLength, const u8 *rawKey, KEY_TABLE_TYPE keyTable);
+void Camellia_EncryptBlock_Rounds(int grandRounds, const u8 plaintext[], 
+		const KEY_TABLE_TYPE keyTable, u8 ciphertext[]);
+void Camellia_DecryptBlock_Rounds(int grandRounds, const u8 ciphertext[], 
+		const KEY_TABLE_TYPE keyTable, u8 plaintext[]);
+void Camellia_EncryptBlock(int keyBitLength, const u8 plaintext[], 
+		const KEY_TABLE_TYPE keyTable, u8 ciphertext[]);
+void Camellia_DecryptBlock(int keyBitLength, const u8 ciphertext[], 
+		const KEY_TABLE_TYPE keyTable, u8 plaintext[]);
+#endif /* #ifndef HEADER_CAMELLIA_LOCL_H */
diff --git a/crypto/camellia/cmll_misc.c b/crypto/camellia/cmll_misc.c
new file mode 100644
index 0000000000..f44689124b
--- /dev/null
+++ b/crypto/camellia/cmll_misc.c
@@ -0,0 +1,79 @@
+/* crypto/camellia/camellia_misc.c -*- mode:C; c-file-style: "eay" -*- */
+/* ====================================================================
+ * Copyright (c) 2006 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer. 
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ *
+ */
+ 
+#include <openssl/opensslv.h>
+#include <openssl/camellia.h>
+#include "cmll_locl.h"
+
+const char CAMELLIA_version[]="CAMELLIA" OPENSSL_VERSION_PTEXT;
+
+int Camellia_set_key(const unsigned char *userKey, const int bits,
+	CAMELLIA_KEY *key)
+	{
+	if(!userKey || !key)
+		return -1;
+	if(bits != 128 && bits != 192 && bits != 256)
+		return -2;
+	key->grand_rounds = Camellia_Ekeygen(bits , userKey, key->u.rd_key);
+	return 0;
+	}
+
+void Camellia_encrypt(const unsigned char *in, unsigned char *out,
+	const CAMELLIA_KEY *key)
+	{
+	Camellia_EncryptBlock_Rounds(key->grand_rounds, in , key->u.rd_key , out);
+	}
+
+void Camellia_decrypt(const unsigned char *in, unsigned char *out,
+	const CAMELLIA_KEY *key)
+	{
+	Camellia_DecryptBlock_Rounds(key->grand_rounds, in , key->u.rd_key , out);
+	}
diff --git a/crypto/camellia/cmll_ofb.c b/crypto/camellia/cmll_ofb.c
new file mode 100644
index 0000000000..a482befc74
--- /dev/null
+++ b/crypto/camellia/cmll_ofb.c
@@ -0,0 +1,119 @@
+/* crypto/camellia/camellia_ofb.c -*- mode:C; c-file-style: "eay" -*- */
+/* ====================================================================
+ * Copyright (c) 2006 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer. 
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ *
+ */
+/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
+ * All rights reserved.
+ *
+ * This package is an SSL implementation written
+ * by Eric Young (eay@cryptsoft.com).
+ * The implementation was written so as to conform with Netscapes SSL.
+ * 
+ * This library is free for commercial and non-commercial use as long as
+ * the following conditions are aheared to.  The following conditions
+ * apply to all code found in this distribution, be it the RC4, RSA,
+ * lhash, DES, etc., code; not just the SSL code.  The SSL documentation
+ * included with this distribution is covered by the same copyright terms
+ * except that the holder is Tim Hudson (tjh@cryptsoft.com).
+ * 
+ * Copyright remains Eric Young's, and as such any Copyright notices in
+ * the code are not to be removed.
+ * If this package is used in a product, Eric Young should be given attribution
+ * as the author of the parts of the library used.
+ * This can be in the form of a textual message at program startup or
+ * in documentation (online or textual) provided with the package.
+ * 
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. All advertising materials mentioning features or use of this software
+ *    must display the following acknowledgement:
+ *    "This product includes cryptographic software written by
+ *     Eric Young (eay@cryptsoft.com)"
+ *    The word 'cryptographic' can be left out if the rouines from the library
+ *    being used are not cryptographic related :-).
+ * 4. If you include any Windows specific code (or a derivative thereof) from 
+ *    the apps directory (application code) you must include an acknowledgement:
+ *    "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
+ * 
+ * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ * 
+ * The licence and distribution terms for any publically available version or
+ * derivative of this code cannot be changed.  i.e. this code cannot simply be
+ * copied and put under another distribution licence
+ * [including the GNU Public Licence.]
+ */
+
+#include <openssl/camellia.h>
+#include <openssl/modes.h>
+
+/* The input and output encrypted as though 128bit ofb mode is being
+ * used.  The extra state information to record how much of the
+ * 128bit block we have used is contained in *num;
+ */
+void Camellia_ofb128_encrypt(const unsigned char *in, unsigned char *out,
+	size_t length, const CAMELLIA_KEY *key,
+	unsigned char *ivec, int *num) {
+	CRYPTO_ofb128_encrypt(in,out,length,key,ivec,num,(block128_f)Camellia_encrypt);
+}