x86 assembler pack: add bn_GF2m_mul_2x2 implementations (see x86-gf2m.pl for

details and performance data).

Configure

@@ -123,7 +123,7 @@ my $tlib="-lnsl -lsocket";
 my $bits1="THIRTY_TWO_BIT ";
 my $bits2="SIXTY_FOUR_BIT ";
 
-my $x86_asm="x86cpuid.o:bn-586.o co-586.o x86-mont.o:des-586.o crypt586.o:aes-586.o aesni-x86.o:bf-586.o:md5-586.o:sha1-586.o sha256-586.o sha512-586.o:cast-586.o:rc4-586.o:rmd-586.o:rc5-586.o:wp_block.o wp-mmx.o:cmll-x86.o:ghash-x86.o";
+my $x86_asm="x86cpuid.o:bn-586.o co-586.o x86-mont.o x86-gf2m.o:des-586.o crypt586.o:aes-586.o aesni-x86.o:bf-586.o:md5-586.o:sha1-586.o sha256-586.o sha512-586.o:cast-586.o:rc4-586.o:rmd-586.o:rc5-586.o:wp_block.o wp-mmx.o:cmll-x86.o:ghash-x86.o";
 
 my $x86_elf_asm="$x86_asm:elf";
 

TABLE

@@ -233,7 +233,7 @@ $sys_id       =
 $lflags       = 
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -265,7 +265,7 @@ $sys_id       =
 $lflags       = 
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -329,7 +329,7 @@ $sys_id       = CYGWIN32
 $lflags       = 
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -393,7 +393,7 @@ $sys_id       = MSDOS
 $lflags       = -L/dev/env/WATT_ROOT/lib -lwatt
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -553,7 +553,7 @@ $sys_id       =
 $lflags       = -lsocket
 $bn_ops       = DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -745,7 +745,7 @@ $sys_id       = WIN32
 $lflags       = 
 $bn_ops       = BN_LLONG RC4_INDEX EXPORT_VAR_AS_FN RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1097,7 +1097,7 @@ $sys_id       = BEOS
 $lflags       = -lbe -lbind -lsocket
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1129,7 +1129,7 @@ $sys_id       = BEOS
 $lflags       = -lbe -lnet
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1161,7 +1161,7 @@ $sys_id       =
 $lflags       = -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1289,7 +1289,7 @@ $sys_id       = MACOSX
 $lflags       = -Wl,-search_paths_first%
 $bn_ops       = BN_LLONG RC4_INT RC4_CHUNK DES_UNROLL BF_PTR
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1449,7 +1449,7 @@ $sys_id       =
 $lflags       = 
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1513,7 +1513,7 @@ $sys_id       = WIN32
 $lflags       = 
 $bn_ops       = BN_LLONG RC4_INDEX EXPORT_VAR_AS_FN RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1801,7 +1801,7 @@ $sys_id       =
 $lflags       = 
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1833,7 +1833,7 @@ $sys_id       = MACOSX
 $lflags       = -Wl,-search_paths_first%
 $bn_ops       = BN_LLONG RC4_INT RC4_CHUNK DES_UNROLL BF_PTR
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1961,7 +1961,7 @@ $sys_id       =
 $lflags       = -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -1993,7 +1993,7 @@ $sys_id       =
 $lflags       = -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -2089,7 +2089,7 @@ $sys_id       =
 $lflags       = -lefence -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -2121,7 +2121,7 @@ $sys_id       =
 $lflags       = -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -2249,7 +2249,7 @@ $sys_id       =
 $lflags       = -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -2281,7 +2281,7 @@ $sys_id       =
 $lflags       = -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -2345,7 +2345,7 @@ $sys_id       =
 $lflags       = 
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -2537,7 +2537,7 @@ $sys_id       =
 $lflags       = -rdynamic -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -2729,7 +2729,7 @@ $sys_id       =
 $lflags       = -lnsl -lsocket
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -3305,7 +3305,7 @@ $sys_id       =
 $lflags       = -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -3657,7 +3657,7 @@ $sys_id       =
 $lflags       = 
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -3721,7 +3721,7 @@ $sys_id       =
 $lflags       = -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -3817,7 +3817,7 @@ $sys_id       =
 $lflags       = -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -4201,7 +4201,7 @@ $sys_id       = MINGW32
 $lflags       = -lws2_32 -lgdi32 -lcrypt32
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT EXPORT_VAR_AS_FN
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -4809,7 +4809,7 @@ $sys_id       =
 $lflags       = -lsocket -lnsl
 $bn_ops       = DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -4841,7 +4841,7 @@ $sys_id       =
 $lflags       = -lsocket -lnsl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -5097,7 +5097,7 @@ $sys_id       =
 $lflags       = -lsocket -lnsl -ldl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -5545,7 +5545,7 @@ $sys_id       =
 $lflags       = -lsocket -lnsl
 $bn_ops       = BN_LLONG MD2_CHAR RC4_INDEX DES_PTR DES_RISC1 DES_UNROLL
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -5577,7 +5577,7 @@ $sys_id       =
 $lflags       = -lsocket -lnsl
 $bn_ops       = BN_LLONG DES_PTR DES_RISC1 DES_UNROLL RC4_INDEX MD2_INT
 $cpuid_obj    = x86cpuid.o
-$bn_obj       = bn-586.o co-586.o x86-mont.o
+$bn_obj       = bn-586.o co-586.o x86-mont.o x86-gf2m.o
 $des_obj      = des-586.o crypt586.o
 $aes_obj      = aes-586.o aesni-x86.o
 $bf_obj       = bf-586.o
@@ -5696,6 +5696,38 @@ $ranlib       =
 $arflags      = 
 $multilib     = 
 
+*** vxworks-ppc60x
+$cc           = ccppc
+$cflags       = -D_REENTRANT -mrtp -mhard-float -mstrict-align -fno-implicit-fp -DPPC32_fp60x -O2 -fstrength-reduce -fno-builtin -fno-strict-aliasing -Wall -DCPU=PPC32 -DTOOL_FAMILY=gnu -DTOOL=gnu -I$(WIND_BASE)/target/usr/h -I$(WIND_BASE)/target/usr/h/wrn/coreip
+$unistd       = 
+$thread_cflag = 
+$sys_id       = VXWORKS
+$lflags       = -Wl,--defsym,__wrs_rtp_base=0xe0000000 -L $(WIND_BASE)/target/usr/lib/ppc/PPC32/common
+$bn_ops       = 
+$cpuid_obj    = 
+$bn_obj       = 
+$des_obj      = 
+$aes_obj      = 
+$bf_obj       = 
+$md5_obj      = 
+$sha1_obj     = 
+$cast_obj     = 
+$rc4_obj      = 
+$rmd160_obj   = 
+$rc5_obj      = 
+$wp_obj       = 
+$cmll_obj     = 
+$modes_obj    = 
+$perlasm_scheme = 
+$dso_scheme   = 
+$shared_target= 
+$shared_cflag = 
+$shared_ldflag = 
+$shared_extension = 
+$ranlib       = 
+$arflags      = 
+$multilib     = 
+
 *** vxworks-ppc750
 $cc           = ccppc
 $cflags       = -ansi -nostdinc -DPPC750 -D_REENTRANT -fvolatile -fno-builtin -fno-for-scope -fsigned-char -Wall -msoft-float -mlongcall -DCPU=PPC604 -I$(WIND_BASE)/target/h $(DEBUG_FLAG)
@@ -5791,3 +5823,35 @@ $shared_extension =
 $ranlib       = 
 $arflags      = 
 $multilib     = 
+
+*** vxworks-ppcgen
+$cc           = ccppc
+$cflags       = -D_REENTRANT -mrtp -msoft-float -mstrict-align -O1 -fno-builtin -fno-strict-aliasing -Wall -DCPU=PPC32 -DTOOL_FAMILY=gnu -DTOOL=gnu -I$(WIND_BASE)/target/usr/h -I$(WIND_BASE)/target/usr/h/wrn/coreip
+$unistd       = 
+$thread_cflag = 
+$sys_id       = VXWORKS
+$lflags       = -Wl,--defsym,__wrs_rtp_base=0xe0000000 -L $(WIND_BASE)/target/usr/lib/ppc/PPC32/sfcommon
+$bn_ops       = 
+$cpuid_obj    = 
+$bn_obj       = 
+$des_obj      = 
+$aes_obj      = 
+$bf_obj       = 
+$md5_obj      = 
+$sha1_obj     = 
+$cast_obj     = 
+$rc4_obj      = 
+$rmd160_obj   = 
+$rc5_obj      = 
+$wp_obj       = 
+$cmll_obj     = 
+$modes_obj    = 
+$perlasm_scheme = 
+$dso_scheme   = 
+$shared_target= 
+$shared_cflag = 
+$shared_ldflag = 
+$shared_extension = 
+$ranlib       = 
+$arflags      = 
+$multilib     = 

crypto/bn/Makefile

@@ -66,6 +66,8 @@ co-586.s:	asm/co-586.pl ../perlasm/x86asm.pl
 	$(PERL) asm/co-586.pl $(PERLASM_SCHEME) $(CFLAGS) $(PROCESSOR) > $@
 x86-mont.s:	asm/x86-mont.pl ../perlasm/x86asm.pl
 	$(PERL) asm/x86-mont.pl $(PERLASM_SCHEME) $(CFLAGS) $(PROCESSOR) > $@
+x86-gf2m.s:	asm/x86-gf2m.pl ../perlasm/x86asm.pl
+	$(PERL) asm/x86-gf2m.pl $(PERLASM_SCHEME) $(CFLAGS) $(PROCESSOR) > $@
 
 sparcv8.o:	asm/sparcv8.S
 	$(CC) $(CFLAGS) -c asm/sparcv8.S

crypto/bn/asm/x86-gf2m.pl

+#!/usr/bin/env perl
+#
+# ====================================================================
+# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL
+# project. The module is, however, dual licensed under OpenSSL and
+# CRYPTOGAMS licenses depending on where you obtain it. For further
+# details see http://www.openssl.org/~appro/cryptogams/.
+# ====================================================================
+#
+# May 2011
+#
+# The module implements bn_GF2m_mul_2x2 polynomial multiplication used
+# in bn_gf2m.c. It's kind of low-hanging mechanical port from C for
+# the time being... Except that it has three code paths: pure integer
+# code suitable for any x86 CPU, MMX code suitable for PIII and later
+# and PCLMULQDQ suitable for Westmere and later. Improvement varies
+# from one benchmark and -arch to another. Below are interval values
+# for 163- and 571-bit ECDH benchmarks relative to compiler-generated
+# code:
+#
+# PIII		16%-30%
+# P4		12%-12%
+# Opteron	18%-40%
+# Core2		19%-44%
+# Atom		38%-64%
+# Westmere	53%-121%(PCLMULQDQ)/20%-32%(MMX)
+# Sandy Bridge	72%-127%(PCLMULQDQ)/27%-23%(MMX)
+#
+# Note that above improvement coefficients are not coefficients for
+# bn_GF2m_mul_2x2 itself. For example 120% ECDH improvement is result
+# of bn_GF2m_mul_2x2 being >4x faster. As it gets faster, benchmark
+# is more and more dominated by other subroutines, most notably by
+# BN_GF2m_mod[_mul]_arr...
+
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+push(@INC,"${dir}","${dir}../../perlasm");
+require "x86asm.pl";
+
+&asm_init($ARGV[0],$0,$x86only = $ARGV[$#ARGV] eq "386");
+
+$sse2=0;
+for (@ARGV) { $sse2=1 if (/-DOPENSSL_IA32_SSE2/); }
+
+&external_label("OPENSSL_ia32cap_P") if ($sse2);
+
+$a="eax";
+$b="ebx";
+($a1,$a2,$a4)=("ecx","edx","ebp");
+
+$R="mm0";
+@T=("mm1","mm2");
+($A,$B,$B30,$B31)=("mm2","mm3","mm4","mm5");
+@i=("esi","edi");
+
+					if (!$x86only) {
+&function_begin_B("_mul_1x1_mmx");
+	&sub	("esp",32+4);
+	 &mov	($a1,$a);
+	 &lea	($a2,&DWP(0,$a,$a));
+	 &and	($a1,0x3fffffff);
+	 &lea	($a4,&DWP(0,$a2,$a2));
+	 &mov	(&DWP(0*4,"esp"),0);
+	 &and	($a2,0x7fffffff);
+	&movd	($A,$a);
+	&movd	($B,$b);
+	 &mov	(&DWP(1*4,"esp"),$a1);	# a1
+	 &xor	($a1,$a2);		# a1^a2
+	&pxor	($B31,$B31);
+	&pxor	($B30,$B30);
+	 &mov	(&DWP(2*4,"esp"),$a2);	# a2
+	 &xor	($a2,$a4);		# a2^a4
+	 &mov	(&DWP(3*4,"esp"),$a1);	# a1^a2
+	&pcmpgtd($B31,$A);		# broadcast 31st bit
+	&paddd	($A,$A);		# $A<<=1
+	 &xor	($a1,$a2);		# a1^a4=a1^a2^a2^a4
+	 &mov	(&DWP(4*4,"esp"),$a4);	# a4
+	 &xor	($a4,$a2);		# a2=a4^a2^a4
+	&pand	($B31,$B);
+	&pcmpgtd($B30,$A);		# broadcast 30th bit
+	 &mov	(&DWP(5*4,"esp"),$a1);	# a1^a4
+	 &xor	($a4,$a1);		# a1^a2^a4
+	&psllq	($B31,31);
+	&pand	($B30,$B);
+	 &mov	(&DWP(6*4,"esp"),$a2);	# a2^a4
+	&mov	(@i[0],0x7);
+	 &mov	(&DWP(7*4,"esp"),$a4);	# a1^a2^a4
+	 &mov	($a4,@i[0]);
+	&and	(@i[0],$b);
+	&shr	($b,3);
+	&mov	(@i[1],$a4);
+	&psllq	($B30,30);
+	&and	(@i[1],$b);
+	&shr	($b,3);
+	&movd	($R,&DWP(0,"esp",@i[0],4));
+	&mov	(@i[0],$a4);
+	&and	(@i[0],$b);
+	&shr	($b,3);
+	for($n=1;$n<9;$n++) {
+		&movd	(@T[1],&DWP(0,"esp",@i[1],4));
+		&mov	(@i[1],$a4);
+		&psllq	(@T[1],3*$n);
+		&and	(@i[1],$b);
+		&shr	($b,3);
+		&pxor	($R,@T[1]);
+
+		push(@i,shift(@i)); push(@T,shift(@T));
+	}
+	&movd	(@T[1],&DWP(0,"esp",@i[1],4));
+	&pxor	($R,$B30);
+	&psllq	(@T[1],3*$n++);
+	&pxor	($R,@T[1]);
+
+	&movd	(@T[0],&DWP(0,"esp",@i[0],4));
+	&pxor	($R,$B31);
+	&psllq	(@T[0],3*$n);
+	&add	("esp",32+4);
+	&pxor	($R,@T[0]);
+	&ret	();
+&function_end_B("_mul_1x1_mmx");
+					}
+
+($lo,$hi)=("eax","edx");
+@T=("ecx","ebp");
+
+&function_begin_B("_mul_1x1_ialu");
+	&sub	("esp",32+4);
+	 &mov	($a1,$a);
+	 &lea	($a2,&DWP(0,$a,$a));
+	 &lea	($a4,&DWP(0,"",$a,4));
+	 &and	($a1,0x3fffffff);
+	&lea	(@i[1],&DWP(0,$lo,$lo));
+	&sar	($lo,31);		# broadcast 31st bit
+	 &mov	(&DWP(0*4,"esp"),0);
+	 &and	($a2,0x7fffffff);
+	 &mov	(&DWP(1*4,"esp"),$a1);	# a1
+	 &xor	($a1,$a2);		# a1^a2
+	 &mov	(&DWP(2*4,"esp"),$a2);	# a2
+	 &xor	($a2,$a4);		# a2^a4
+	 &mov	(&DWP(3*4,"esp"),$a1);	# a1^a2
+	 &xor	($a1,$a2);		# a1^a4=a1^a2^a2^a4
+	 &mov	(&DWP(4*4,"esp"),$a4);	# a4
+	 &xor	($a4,$a2);		# a2=a4^a2^a4
+	 &mov	(&DWP(5*4,"esp"),$a1);	# a1^a4
+	 &xor	($a4,$a1);		# a1^a2^a4
+	&sar	(@i[1],31);		# broardcast 30th bit
+	&and	($lo,$b);
+	 &mov	(&DWP(6*4,"esp"),$a2);	# a2^a4
+	&and	(@i[1],$b);
+	 &mov	(&DWP(7*4,"esp"),$a4);	# a1^a2^a4
+	&mov	($hi,$lo);
+	&shl	($lo,31);
+	&mov	(@T[0],@i[1]);
+	&shr	($hi,1);
+
+	 &mov	(@i[0],0x7);
+	&shl	(@i[1],30);
+	 &and	(@i[0],$b);
+	&shr	(@T[0],2);
+	&xor	($lo,@i[1]);
+
+	&shr	($b,3);
+	&mov	(@i[1],0x7);		# 5-byte instruction!?
+	&and	(@i[1],$b);
+	&shr	($b,3);
+	 &xor	($hi,@T[0]);
+	&xor	($lo,&DWP(0,"esp",@i[0],4));
+	&mov	(@i[0],0x7);
+	&and	(@i[0],$b);
+	&shr	($b,3);
+	for($n=1;$n<9;$n++) {
+		&mov	(@T[1],&DWP(0,"esp",@i[1],4));
+		&mov	(@i[1],0x7);
+		&mov	(@T[0],@T[1]);
+		&shl	(@T[1],3*$n);
+		&and	(@i[1],$b);
+		&shr	(@T[0],32-3*$n);
+		&xor	($lo,@T[1]);
+		&shr	($b,3);
+		&xor	($hi,@T[0]);
+
+		push(@i,shift(@i)); push(@T,shift(@T));
+	}
+	&mov	(@T[1],&DWP(0,"esp",@i[1],4));
+	&mov	(@T[0],@T[1]);
+	&shl	(@T[1],3*$n);
+	&mov	(@i[1],&DWP(0,"esp",@i[0],4));
+	&shr	(@T[0],32-3*$n);	$n++;
+	&mov	(@i[0],@i[1]);
+	&xor	($lo,@T[1]);
+	&shl	(@i[1],3*$n);
+	&xor	($hi,@T[0]);
+	&shr	(@i[0],32-3*$n);
+	&xor	($lo,@i[1]);
+	&xor	($hi,@i[0]);
+
+	&add	("esp",32+4);
+	&ret	();
+&function_end_B("_mul_1x1_ialu");
+
+sub pclmulqdq
+{ my($dst,$src,$imm)=@_;
+    if ("$dst:$src" =~ /xmm([0-7]):xmm([0-7])/)
+    {   &data_byte(0x66,0x0f,0x3a,0x44,0xc0|($1<<3)|$2,$imm);   }
+}
+
+# void bn_GF2m_mul_2x2(BN_ULONG *r, BN_ULONG a1, BN_ULONG a0, BN_ULONG b1, BN_ULONG b0);
+&function_begin_B("bn_GF2m_mul_2x2");
+if (!$x86only) {
+	&picmeup("edx","OPENSSL_ia32cap_P");
+	&mov	("eax",&DWP(0,"edx"));
+	&mov	("edx",&DWP(4,"edx"));
+	&test	("eax",1<<23);		# check MMX bit
+	&jz	(&label("ialu"));
+if ($sse2) {
+	&test	("eax",1<<24);		# check FXSR bit
+	&jz	(&label("mmx"));
+	&test	("edx",1<<1);		# check PCLMULQDQ bit
+	&jz	(&label("mmx"));
+
+	&movups		("xmm0",&QWP(8,"esp"));
+	&shufps		("xmm0","xmm0",0b10110001);
+	&pclmulqdq	("xmm0","xmm0",1);
+	&mov		("eax",&DWP(4,"esp"));
+	&movups		(&QWP(0,"eax"),"xmm0");
+	&ret	();
+
+&set_label("mmx",16);
+}
+	&push	("ebp");
+	&push	("ebx");
+	&push	("esi");
+	&push	("edi");
+	&mov	($a,&wparam(1));
+	&mov	($b,&wparam(3));
+	&call	("_mul_1x1_mmx");	# a1b1
+	&movq	("mm7",$R);
+
+	&mov	($a,&wparam(2));
+	&mov	($b,&wparam(4));
+	&call	("_mul_1x1_mmx");	# a0b0
+	&movq	("mm6",$R);
+
+	&mov	($a,&wparam(1));
+	&mov	($b,&wparam(3));
+	&xor	($a,&wparam(2));
+	&xor	($b,&wparam(4));
+	&call	("_mul_1x1_mmx");	# (a0+a1)(b0+b1)
+	&pxor	($R,"mm7");
+	&mov	($a,&wparam(0));
+	&pxor	($R,"mm6");		# (a0+a1)(b0+b1)-a1b1-a0b0
+
+	&movq	($A,$R);
+	&psllq	($R,32);
+	&pop	("edi");
+	&psrlq	($A,32);
+	&pop	("esi");
+	&pxor	($R,"mm6");
+	&pop	("ebx");
+	&pxor	($A,"mm7");
+	&movq	(&QWP(0,$a),$R);
+	&pop	("ebp");
+	&movq	(&QWP(8,$a),$A);
+	&emms	();
+	&ret	();
+&set_label("ialu",16);
+}
+	&push	("ebp");
+	&push	("ebx");
+	&push	("esi");
+	&push	("edi");
+	&stack_push(4+1);
+
+	&mov	($a,&wparam(1));
+	&mov	($b,&wparam(3));
+	&call	("_mul_1x1_ialu");	# a1b1
+	&mov	(&DWP(8,"esp"),$lo);
+	&mov	(&DWP(12,"esp"),$hi);
+
+	&mov	($a,&wparam(2));
+	&mov	($b,&wparam(4));
+	&call	("_mul_1x1_ialu");	# a0b0
+	&mov	(&DWP(0,"esp"),$lo);
+	&mov	(&DWP(4,"esp"),$hi);
+
+	&mov	($a,&wparam(1));
+	&mov	($b,&wparam(3));
+	&xor	($a,&wparam(2));
+	&xor	($b,&wparam(4));
+	&call	("_mul_1x1_ialu");	# (a0+a1)(b0+b1)
+
+	&mov	("ebp",&wparam(0));
+		 @r=("ebx","ecx","edi","esi");
+	&mov	(@r[0],&DWP(0,"esp"));
+	&mov	(@r[1],&DWP(4,"esp"));
+	&mov	(@r[2],&DWP(8,"esp"));
+	&mov	(@r[3],&DWP(12,"esp"));
+
+	&xor	($lo,$hi);
+	&xor	($hi,@r[1]);
+	&xor	($lo,@r[0]);
+	&mov	(&DWP(0,"ebp"),@r[0]);
+	&xor	($hi,@r[2]);
+	&mov	(&DWP(12,"ebp"),@r[3]);
+	&xor	($lo,@r[3]);
+	&stack_pop(4+1);
+	&xor	($hi,@r[3]);
+	&pop	("edi");
+	&xor	($lo,$hi);
+	&pop	("esi");
+	&mov	(&DWP(8,"ebp"),$hi);
+	&pop	("ebx");
+	&mov	(&DWP(4,"ebp"),$lo);
+	&pop	("ebp");
+	&ret	();
+&function_end_B("bn_GF2m_mul_2x2");
+
+&asciz	("GF2m Multiplication for x86, CRYPTOGAMS by <appro\@openssl.org>");
+
+&asm_finish();