diff options
author | Andy Polyakov <appro@openssl.org> | 2011-09-07 00:45:36 +0400 |
---|---|---|
committer | Andy Polyakov <appro@openssl.org> | 2011-09-07 00:45:36 +0400 |
commit | ed28aef8b455be436f252dfceac49a958a92e53b (patch) | |
tree | cd34319f2128612e24d78d06571cdc4c24eab614 | |
parent | 0486cce653b62d26a8ca37ac12f69f1a6b998844 (diff) |
Padlock engine: make it independent of inline assembler.
-rwxr-xr-x | Configure | 54 | ||||
-rw-r--r-- | Makefile.org | 4 | ||||
-rw-r--r-- | TABLE | 183 | ||||
-rw-r--r-- | crypto/perlasm/x86masm.pl | 7 | ||||
-rw-r--r-- | crypto/perlasm/x86nasm.pl | 7 | ||||
-rw-r--r-- | engines/Makefile | 13 | ||||
-rw-r--r-- | engines/asm/e_padlock-x86.pl | 424 | ||||
-rw-r--r-- | engines/asm/e_padlock-x86_86.pl | 327 | ||||
-rw-r--r-- | engines/e_padlock.c | 969 |
9 files changed, 1173 insertions, 815 deletions
@@ -123,24 +123,24 @@ 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 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_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:e_padlock-x86.o"; my $x86_elf_asm="$x86_asm:elf"; -my $x86_64_asm="x86_64cpuid.o:x86_64-gcc.o x86_64-mont.o x86_64-mont5.o x86_64-gf2m.o modexp512-x86_64.o::aes-x86_64.o aesni-x86_64.o aesni-sha1-x86_64.o::md5-x86_64.o:sha1-x86_64.o sha256-x86_64.o sha512-x86_64.o::rc4-x86_64.o rc4-md5-x86_64.o:::wp-x86_64.o:cmll-x86_64.o cmll_misc.o:ghash-x86_64.o"; -my $ia64_asm="ia64cpuid.o:bn-ia64.o ia64-mont.o::aes_core.o aes_cbc.o aes-ia64.o::md5-ia64.o:sha1-ia64.o sha256-ia64.o sha512-ia64.o::rc4-ia64.o rc4_skey.o:::::ghash-ia64.o:void"; -my $sparcv9_asm="sparcv9cap.o sparccpuid.o:bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o:des_enc-sparc.o fcrypt_b.o:aes_core.o aes_cbc.o aes-sparcv9.o:::sha1-sparcv9.o sha256-sparcv9.o sha512-sparcv9.o:::::::ghash-sparcv9.o:void"; -my $sparcv8_asm=":sparcv8.o:des_enc-sparc.o fcrypt_b.o::::::::::::void"; -my $alpha_asm="alphacpuid.o:bn_asm.o alpha-mont.o:::::sha1-alpha.o:::::::ghash-alpha.o:void"; -my $mips32_asm=":bn-mips.o::aes_cbc.o aes-mips.o:::sha1-mips.o sha256-mips.o:::::::"; -my $mips64_asm=":bn-mips.o mips-mont.o::aes_cbc.o aes-mips.o:::sha1-mips.o sha256-mips.o sha512-mips.o:::::::"; -my $s390x_asm="s390xcap.o s390xcpuid.o:bn-s390x.o s390x-mont.o s390x-gf2m.o::aes_ctr.o aes-s390x.o:::sha1-s390x.o sha256-s390x.o sha512-s390x.o::rc4-s390x.o:::::ghash-s390x.o"; -my $armv4_asm="armcap.o armv4cpuid.o:bn_asm.o armv4-mont.o armv4-gf2m.o::aes_cbc.o aes-armv4.o:::sha1-armv4-large.o sha256-armv4.o sha512-armv4.o:::::::ghash-armv4.o:void"; -my $parisc11_asm="pariscid.o:bn_asm.o parisc-mont.o::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o:32"; -my $parisc20_asm="pariscid.o:pa-risc2W.o parisc-mont.o::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o:64"; -my $ppc32_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o::aes_core.o aes_cbc.o aes-ppc.o:::sha1-ppc.o sha256-ppc.o:::::::"; -my $ppc64_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o::aes_core.o aes_cbc.o aes-ppc.o:::sha1-ppc.o sha256-ppc.o sha512-ppc.o:::::::"; -my $no_asm="::::::::::::::void"; +my $x86_64_asm="x86_64cpuid.o:x86_64-gcc.o x86_64-mont.o x86_64-mont5.o x86_64-gf2m.o modexp512-x86_64.o::aes-x86_64.o aesni-x86_64.o aesni-sha1-x86_64.o::md5-x86_64.o:sha1-x86_64.o sha256-x86_64.o sha512-x86_64.o::rc4-x86_64.o rc4-md5-x86_64.o:::wp-x86_64.o:cmll-x86_64.o cmll_misc.o:ghash-x86_64.o:e_padlock-x86_64.o"; +my $ia64_asm="ia64cpuid.o:bn-ia64.o ia64-mont.o::aes_core.o aes_cbc.o aes-ia64.o::md5-ia64.o:sha1-ia64.o sha256-ia64.o sha512-ia64.o::rc4-ia64.o rc4_skey.o:::::ghash-ia64.o::void"; +my $sparcv9_asm="sparcv9cap.o sparccpuid.o:bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o:des_enc-sparc.o fcrypt_b.o:aes_core.o aes_cbc.o aes-sparcv9.o:::sha1-sparcv9.o sha256-sparcv9.o sha512-sparcv9.o:::::::ghash-sparcv9.o::void"; +my $sparcv8_asm=":sparcv8.o:des_enc-sparc.o fcrypt_b.o:::::::::::::void"; +my $alpha_asm="alphacpuid.o:bn_asm.o alpha-mont.o:::::sha1-alpha.o:::::::ghash-alpha.o::void"; +my $mips32_asm=":bn-mips.o::aes_cbc.o aes-mips.o:::sha1-mips.o sha256-mips.o::::::::"; +my $mips64_asm=":bn-mips.o mips-mont.o::aes_cbc.o aes-mips.o:::sha1-mips.o sha256-mips.o sha512-mips.o::::::::"; +my $s390x_asm="s390xcap.o s390xcpuid.o:bn-s390x.o s390x-mont.o s390x-gf2m.o::aes_ctr.o aes-s390x.o:::sha1-s390x.o sha256-s390x.o sha512-s390x.o::rc4-s390x.o:::::ghash-s390x.o:"; +my $armv4_asm="armcap.o armv4cpuid.o:bn_asm.o armv4-mont.o armv4-gf2m.o::aes_cbc.o aes-armv4.o:::sha1-armv4-large.o sha256-armv4.o sha512-armv4.o:::::::ghash-armv4.o::void"; +my $parisc11_asm="pariscid.o:bn_asm.o parisc-mont.o::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o::32"; +my $parisc20_asm="pariscid.o:pa-risc2W.o parisc-mont.o::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o::64"; +my $ppc32_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o::aes_core.o aes_cbc.o aes-ppc.o:::sha1-ppc.o sha256-ppc.o::::::::"; +my $ppc64_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o::aes_core.o aes_cbc.o aes-ppc.o:::sha1-ppc.o sha256-ppc.o sha512-ppc.o::::::::"; +my $no_asm=":::::::::::::::void"; # As for $BSDthreads. Idea is to maintain "collective" set of flags, # which would cover all BSD flavors. -pthread applies to them all, @@ -151,7 +151,7 @@ my $no_asm="::::::::::::::void"; # seems to be sufficient? my $BSDthreads="-pthread -D_THREAD_SAFE -D_REENTRANT"; -#config-string $cc : $cflags : $unistd : $thread_cflag : $sys_id : $lflags : $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 : $dso_scheme : $shared_target : $shared_cflag : $shared_ldflag : $shared_extension : $ranlib : $arflags : $multilib +#config-string $cc : $cflags : $unistd : $thread_cflag : $sys_id : $lflags : $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 : $engines_obj : $dso_scheme : $shared_target : $shared_cflag : $shared_ldflag : $shared_extension : $ranlib : $arflags : $multilib my %table=( # File 'TABLE' (created by 'make TABLE') contains the data from this list, @@ -188,7 +188,7 @@ my %table=( "debug-linux-ppro","gcc:-DBN_DEBUG -DREF_CHECK -DCONF_DEBUG -DBN_CTX_DEBUG -DCRYPTO_MDEBUG -DL_ENDIAN -DTERMIO -g -mcpu=pentiumpro -Wall::-D_REENTRANT::-ldl:BN_LLONG ${x86_gcc_des} ${x86_gcc_opts}:${x86_elf_asm}:dlfcn", "debug-linux-elf","gcc:-DBN_DEBUG -DREF_CHECK -DCONF_DEBUG -DBN_CTX_DEBUG -DCRYPTO_MDEBUG -DL_ENDIAN -DTERMIO -g -march=i486 -Wall::-D_REENTRANT::-lefence -ldl:BN_LLONG ${x86_gcc_des} ${x86_gcc_opts}:${x86_elf_asm}:dlfcn:linux-shared:-fPIC::.so.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", "debug-linux-elf-noefence","gcc:-DBN_DEBUG -DREF_CHECK -DCONF_DEBUG -DBN_CTX_DEBUG -DCRYPTO_MDEBUG -DL_ENDIAN -DTERMIO -g -march=i486 -Wall::-D_REENTRANT::-ldl:BN_LLONG ${x86_gcc_des} ${x86_gcc_opts}:${x86_elf_asm}:dlfcn:linux-shared:-fPIC::.so.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", -"debug-linux-ia32-aes", "gcc:-DAES_EXPERIMENTAL -DL_ENDIAN -DTERMIO -O3 -fomit-frame-pointer -Wall::-D_REENTRANT::-ldl:BN_LLONG ${x86_gcc_des} ${x86_gcc_opts}:x86cpuid.o:bn-586.o co-586.o x86-mont.o:des-586.o crypt586.o:aes_x86core.o aes_cbc.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::ghash-x86.o:elf:dlfcn:linux-shared:-fPIC::.so.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", +"debug-linux-ia32-aes", "gcc:-DAES_EXPERIMENTAL -DL_ENDIAN -DTERMIO -O3 -fomit-frame-pointer -Wall::-D_REENTRANT::-ldl:BN_LLONG ${x86_gcc_des} ${x86_gcc_opts}:x86cpuid.o:bn-586.o co-586.o x86-mont.o:des-586.o crypt586.o:aes_x86core.o aes_cbc.o aesni-x86.o:bf-586.o:md5-586.o:sha1-586.o sha256-586.o sha512-586.o:cast-586.o:rc4-586.o:rmd-586.o:rc5-586.o:wp_block.o wp-mmx.o::ghash-x86.o:e_padlock-x86.o:elf:dlfcn:linux-shared:-fPIC::.so.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", "debug-linux-generic32","gcc:-DBN_DEBUG -DREF_CHECK -DCONF_DEBUG -DCRYPTO_MDEBUG -DTERMIO -g -Wall::-D_REENTRANT::-ldl:BN_LLONG RC4_CHAR RC4_CHUNK DES_INT DES_UNROLL BF_PTR:${no_asm}:dlfcn:linux-shared:-fPIC::.so.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", "debug-linux-generic64","gcc:-DBN_DEBUG -DREF_CHECK -DCONF_DEBUG -DCRYPTO_MDEBUG -DTERMIO -g -Wall::-D_REENTRANT::-ldl:SIXTY_FOUR_BIT_LONG RC4_CHAR RC4_CHUNK DES_INT DES_UNROLL BF_PTR:${no_asm}:dlfcn:linux-shared:-fPIC::.so.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", "debug-linux-x86_64","gcc:-DBN_DEBUG -DREF_CHECK -DCONF_DEBUG -DCRYPTO_MDEBUG -m64 -DL_ENDIAN -DTERMIO -g -Wall::-D_REENTRANT::-ldl:SIXTY_FOUR_BIT_LONG RC4_CHUNK DES_INT DES_UNROLL:${x86_64_asm}:elf:dlfcn:linux-shared:-fPIC:-m64:.so.\$(SHLIB_MAJOR).\$(SHLIB_MINOR):::64", @@ -294,8 +294,8 @@ my %table=( "hpux-parisc-cc-o4","cc:-Ae +O4 +ESlit -z -DB_ENDIAN -DBN_DIV2W -DMD32_XARRAY::-D_REENTRANT::-ldld:BN_LLONG DES_PTR DES_UNROLL DES_RISC1:${no_asm}:dl:hpux-shared:+Z:-b:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", "hpux-parisc-gcc","gcc:-O3 -DB_ENDIAN -DBN_DIV2W::-D_REENTRANT::-Wl,+s -ldld:BN_LLONG DES_PTR DES_UNROLL DES_RISC1:${no_asm}:dl:hpux-shared:-fPIC:-shared:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", "hpux-parisc1_1-gcc","gcc:-O3 -DB_ENDIAN -DBN_DIV2W::-D_REENTRANT::-Wl,+s -ldld:BN_LLONG DES_PTR DES_UNROLL DES_RISC1:${parisc11_asm}:dl:hpux-shared:-fPIC:-shared:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", -"hpux-parisc2-gcc","gcc:-march=2.0 -O3 -DB_ENDIAN -D_REENTRANT::::-Wl,+s -ldld:SIXTY_FOUR_BIT RC4_CHAR RC4_CHUNK DES_PTR DES_UNROLL DES_RISC1::pa-risc2.o:::::::::::::void:dl:hpux-shared:-fPIC:-shared:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", -"hpux64-parisc2-gcc","gcc:-O3 -DB_ENDIAN -D_REENTRANT::::-ldl:SIXTY_FOUR_BIT_LONG MD2_CHAR RC4_INDEX RC4_CHAR DES_UNROLL DES_RISC1 DES_INT::pa-risc2W.o:::::::::::::void:dlfcn:hpux-shared:-fpic:-shared:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR):::/pa20_64", +"hpux-parisc2-gcc","gcc:-march=2.0 -O3 -DB_ENDIAN -D_REENTRANT::::-Wl,+s -ldld:SIXTY_FOUR_BIT RC4_CHAR RC4_CHUNK DES_PTR DES_UNROLL DES_RISC1::pa-risc2.o::::::::::::::void:dl:hpux-shared:-fPIC:-shared:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", +"hpux64-parisc2-gcc","gcc:-O3 -DB_ENDIAN -D_REENTRANT::::-ldl:SIXTY_FOUR_BIT_LONG MD2_CHAR RC4_INDEX RC4_CHAR DES_UNROLL DES_RISC1 DES_INT::pa-risc2W.o::::::::::::::void:dlfcn:hpux-shared:-fpic:-shared:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR):::/pa20_64", # More attempts at unified 10.X and 11.X targets for HP C compiler. # @@ -303,7 +303,7 @@ my %table=( # Kevin Steves <ks@hp.se> "hpux-parisc-cc","cc:+O3 +Optrs_strongly_typed -Ae +ESlit -DB_ENDIAN -DBN_DIV2W -DMD32_XARRAY::-D_REENTRANT::-Wl,+s -ldld:MD2_CHAR RC4_INDEX RC4_CHAR DES_UNROLL DES_RISC1 DES_INT:${no_asm}:dl:hpux-shared:+Z:-b:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", "hpux-parisc1_1-cc","cc:+DA1.1 +O3 +Optrs_strongly_typed -Ae +ESlit -DB_ENDIAN -DMD32_XARRAY::-D_REENTRANT::-Wl,+s -ldld:MD2_CHAR RC4_INDEX RC4_CHAR DES_UNROLL DES_RISC1 DES_INT:${parisc11_asm}:dl:hpux-shared:+Z:-b:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR):::/pa1.1", -"hpux-parisc2-cc","cc:+DA2.0 +DS2.0 +O3 +Optrs_strongly_typed -Ae +ESlit -DB_ENDIAN -DMD32_XARRAY -D_REENTRANT::::-Wl,+s -ldld:SIXTY_FOUR_BIT MD2_CHAR RC4_INDEX RC4_CHAR DES_UNROLL DES_RISC1 DES_INT::pa-risc2.o:::::::::::::void:dl:hpux-shared:+Z:-b:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", +"hpux-parisc2-cc","cc:+DA2.0 +DS2.0 +O3 +Optrs_strongly_typed -Ae +ESlit -DB_ENDIAN -DMD32_XARRAY -D_REENTRANT::::-Wl,+s -ldld:SIXTY_FOUR_BIT MD2_CHAR RC4_INDEX RC4_CHAR DES_UNROLL DES_RISC1 DES_INT::pa-risc2.o::::::::::::::void:dl:hpux-shared:+Z:-b:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR)", "hpux64-parisc2-cc","cc:+DD64 +O3 +Optrs_strongly_typed -Ae +ESlit -DB_ENDIAN -DMD32_XARRAY -D_REENTRANT::::-ldl:SIXTY_FOUR_BIT_LONG MD2_CHAR RC4_INDEX RC4_CHAR DES_UNROLL DES_RISC1 DES_INT:${parisc20_asm}:dlfcn:hpux-shared:+Z:+DD64 -b:.sl.\$(SHLIB_MAJOR).\$(SHLIB_MINOR):::/pa20_64", # HP/UX IA-64 targets @@ -512,9 +512,9 @@ my %table=( # Visual C targets # # Win64 targets, WIN64I denotes IA-64 and WIN64A - AMD64 -"VC-WIN64I","cl:-W3 -Gs0 -Gy -nologo -DOPENSSL_SYSNAME_WIN32 -DWIN32_LEAN_AND_MEAN -DL_ENDIAN -DUNICODE -D_UNICODE -D_CRT_SECURE_NO_DEPRECATE:::WIN64I::SIXTY_FOUR_BIT RC4_CHUNK_LL DES_INT EXPORT_VAR_AS_FN:ia64cpuid.o:ia64.o ia64-mont.o::aes_core.o aes_cbc.o aes-ia64.o::md5-ia64.o:sha1-ia64.o sha256-ia64.o sha512-ia64.o:::::::ghash-ia64.o:ias:win32", +"VC-WIN64I","cl:-W3 -Gs0 -Gy -nologo -DOPENSSL_SYSNAME_WIN32 -DWIN32_LEAN_AND_MEAN -DL_ENDIAN -DUNICODE -D_UNICODE -D_CRT_SECURE_NO_DEPRECATE:::WIN64I::SIXTY_FOUR_BIT RC4_CHUNK_LL DES_INT EXPORT_VAR_AS_FN:ia64cpuid.o:ia64.o ia64-mont.o::aes_core.o aes_cbc.o aes-ia64.o::md5-ia64.o:sha1-ia64.o sha256-ia64.o sha512-ia64.o:::::::ghash-ia64.o::ias:win32", "VC-WIN64A","cl:-W3 -Gs0 -Gy -nologo -DOPENSSL_SYSNAME_WIN32 -DWIN32_LEAN_AND_MEAN -DL_ENDIAN -DUNICODE -D_UNICODE -D_CRT_SECURE_NO_DEPRECATE:::WIN64A::SIXTY_FOUR_BIT RC4_CHUNK_LL DES_INT EXPORT_VAR_AS_FN:".eval{my $asm=$x86_64_asm;$asm=~s/x86_64-gcc\.o/bn_asm.o/;$asm}.":auto:win32", -"debug-VC-WIN64I","cl:-W3 -Gs0 -Gy -Zi -nologo -DOPENSSL_SYSNAME_WIN32 -DWIN32_LEAN_AND_MEAN -DL_ENDIAN -DUNICODE -D_UNICODE -D_CRT_SECURE_NO_DEPRECATE:::WIN64I::SIXTY_FOUR_BIT RC4_CHUNK_LL DES_INT EXPORT_VAR_AS_FN:ia64cpuid.o:ia64.o::aes_core.o aes_cbc.o aes-ia64.o::md5-ia64.o:sha1-ia64.o sha256-ia64.o sha512-ia64.o:::::::ghash-ia64.o:ias:win32", +"debug-VC-WIN64I","cl:-W3 -Gs0 -Gy -Zi -nologo -DOPENSSL_SYSNAME_WIN32 -DWIN32_LEAN_AND_MEAN -DL_ENDIAN -DUNICODE -D_UNICODE -D_CRT_SECURE_NO_DEPRECATE:::WIN64I::SIXTY_FOUR_BIT RC4_CHUNK_LL DES_INT EXPORT_VAR_AS_FN:ia64cpuid.o:ia64.o::aes_core.o aes_cbc.o aes-ia64.o::md5-ia64.o:sha1-ia64.o sha256-ia64.o sha512-ia64.o:::::::ghash-ia64.o::ias:win32", "debug-VC-WIN64A","cl:-W3 -Gs0 -Gy -Zi -nologo -DOPENSSL_SYSNAME_WIN32 -DWIN32_LEAN_AND_MEAN -DL_ENDIAN -DUNICODE -D_UNICODE -D_CRT_SECURE_NO_DEPRECATE:::WIN64A::SIXTY_FOUR_BIT RC4_CHUNK_LL DES_INT EXPORT_VAR_AS_FN:".eval{my $asm=$x86_64_asm;$asm=~s/x86_64-gcc\.o/bn_asm.o/;$asm}.":auto:win32", # x86 Win32 target defaults to ANSI API, if you want UNICODE, complement # 'perl Configure VC-WIN32' with '-DUNICODE -D_UNICODE' @@ -637,6 +637,7 @@ my $idx_rc5_obj = $idx++; my $idx_wp_obj = $idx++; my $idx_cmll_obj = $idx++; my $idx_modes_obj = $idx++; +my $idx_engines_obj = $idx++; my $idx_perlasm_scheme = $idx++; my $idx_dso_scheme = $idx++; my $idx_shared_target = $idx++; @@ -1212,6 +1213,7 @@ my $rc5_obj = $fields[$idx_rc5_obj]; my $wp_obj = $fields[$idx_wp_obj]; my $cmll_obj = $fields[$idx_cmll_obj]; my $modes_obj = $fields[$idx_modes_obj]; +my $engines_obj = $fields[$idx_engines_obj]; my $perlasm_scheme = $fields[$idx_perlasm_scheme]; my $dso_scheme = $fields[$idx_dso_scheme]; my $shared_target = $fields[$idx_shared_target]; @@ -1372,7 +1374,7 @@ if ($no_asm) { $cpuid_obj=$bn_obj= $des_obj=$aes_obj=$bf_obj=$cast_obj=$rc4_obj=$rc5_obj=$cmll_obj= - $modes_obj=$sha1_obj=$md5_obj=$rmd160_obj=$wp_obj=""; + $modes_obj=$sha1_obj=$md5_obj=$rmd160_obj=$wp_obj=$engines_obj=""; $cflags=~s/\-D[BL]_ENDIAN// if ($fips); $thread_cflags=~s/\-D[BL]_ENDIAN// if ($fips); } @@ -1701,6 +1703,7 @@ while (<IN>) s/^WP_ASM_OBJ=.*$/WP_ASM_OBJ= $wp_obj/; s/^CMLL_ENC=.*$/CMLL_ENC= $cmll_obj/; s/^MODES_ASM_OBJ.=*$/MODES_ASM_OBJ= $modes_obj/; + s/^ENGINES_ASM_OBJ.=*$/ENGINES_ASM_OBJ= $engines_obj/; s/^PERLASM_SCHEME=.*$/PERLASM_SCHEME= $perlasm_scheme/; s/^PROCESSOR=.*/PROCESSOR= $processor/; s/^ARFLAGS=.*/ARFLAGS= $arflags/; @@ -1773,6 +1776,7 @@ print "SHA1_OBJ_ASM =$sha1_obj\n"; print "RMD160_OBJ_ASM=$rmd160_obj\n"; print "CMLL_ENC =$cmll_obj\n"; print "MODES_OBJ =$modes_obj\n"; +print "ENGINES_OBJ =$engines_obj\n"; print "PROCESSOR =$processor\n"; print "RANLIB =$ranlib\n"; print "ARFLAGS =$arflags\n"; @@ -2192,7 +2196,8 @@ sub print_table_entry (my $cc,my $cflags,my $unistd,my $thread_cflag,my $sys_id,my $lflags, my $bn_ops,my $cpuid_obj,my $bn_obj,my $des_obj,my $aes_obj, my $bf_obj, my $md5_obj,my $sha1_obj,my $cast_obj,my $rc4_obj,my $rmd160_obj, - my $rc5_obj,my $wp_obj,my $cmll_obj,my $modes_obj,my $perlasm_scheme,my $dso_scheme,my $shared_target,my $shared_cflag, + my $rc5_obj,my $wp_obj,my $cmll_obj,my $modes_obj, my $engines_obj, + my $perlasm_scheme,my $dso_scheme,my $shared_target,my $shared_cflag, my $shared_ldflag,my $shared_extension,my $ranlib,my $arflags,my $multilib)= split(/\s*:\s*/,$table{$target} . ":" x 30 , -1); @@ -2220,6 +2225,7 @@ sub print_table_entry \$wp_obj = $wp_obj \$cmll_obj = $cmll_obj \$modes_obj = $modes_obj +\$engines_obj = $engines_obj \$perlasm_scheme = $perlasm_scheme \$dso_scheme = $dso_scheme \$shared_target= $shared_target diff --git a/Makefile.org b/Makefile.org index 720965b012..a81dc51a52 100644 --- a/Makefile.org +++ b/Makefile.org @@ -100,6 +100,7 @@ RMD160_ASM_OBJ= WP_ASM_OBJ= CMLL_ENC= MODES_ASM_OBJ= +ENGINES_ASM_OBJ= PERLASM_SCHEME= # KRB5 stuff @@ -232,6 +233,7 @@ BUILDENV= PLATFORM='$(PLATFORM)' PROCESSOR='$(PROCESSOR)' \ RMD160_ASM_OBJ='$(RMD160_ASM_OBJ)' \ WP_ASM_OBJ='$(WP_ASM_OBJ)' \ MODES_ASM_OBJ='$(MODES_ASM_OBJ)' \ + ENGINES_ASM_OBJ='$(ENGINES_ASM_OBJ)' \ PERLASM_SCHEME='$(PERLASM_SCHEME)' \ FIPSLIBDIR='${FIPSLIBDIR}' \ FIPSCANLIB="$${FIPSCANLIB:-$(FIPSCANLIB)}" \ @@ -375,7 +377,7 @@ build_crypto: build_ssl: @dir=ssl; target=all; $(BUILD_ONE_CMD) build_engines: - @dir=engines; target=all; $(BUILD_ONE_CMD) + @dir=engines; target=all; AS='$(CC) -c'; export AS; $(BUILD_ONE_CMD) build_apps: @dir=apps; target=all; $(BUILD_ONE_CMD) build_tests: @@ -22,6 +22,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = win32 $shared_target= @@ -54,6 +55,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -86,6 +88,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= bsd-gcc-shared @@ -118,6 +121,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= bsd-gcc-shared @@ -150,6 +154,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= bsd-gcc-shared @@ -182,6 +187,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-sparcv9.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= bsd-gcc-shared @@ -214,6 +220,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= bsd-gcc-shared @@ -246,6 +253,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = a.out $dso_scheme = dlfcn $shared_target= bsd-shared @@ -278,6 +286,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= bsd-shared @@ -310,6 +319,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= bsd-gcc-shared @@ -342,6 +352,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = coff $dso_scheme = dlfcn $shared_target= cygwin-shared @@ -374,6 +385,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = win32 $shared_target= @@ -406,6 +418,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = a.out $dso_scheme = $shared_target= @@ -438,6 +451,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -470,6 +484,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -502,6 +517,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -534,6 +550,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= bsd-gcc-shared @@ -566,6 +583,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= bsd-gcc-shared @@ -598,6 +616,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= reliantunix-shared @@ -630,6 +649,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -662,6 +682,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -694,6 +715,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = win32 $shared_target= @@ -726,6 +748,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = win32 $shared_target= @@ -758,6 +781,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = win32n $dso_scheme = win32 $shared_target= @@ -790,6 +814,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = auto $dso_scheme = win32 $shared_target= @@ -822,6 +847,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = ias $dso_scheme = win32 $shared_target= @@ -854,6 +880,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = aix32 $dso_scheme = dlfcn $shared_target= aix-shared @@ -886,6 +913,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = aix32 $dso_scheme = dlfcn $shared_target= aix-shared @@ -918,6 +946,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -950,6 +979,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = aix64 $dso_scheme = dlfcn $shared_target= aix-shared @@ -982,6 +1012,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = aix64 $dso_scheme = dlfcn $shared_target= aix-shared @@ -1014,6 +1045,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -1046,6 +1078,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-armv4.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -1078,6 +1111,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1110,6 +1144,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = beos $shared_target= beos-shared @@ -1142,6 +1177,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = beos $shared_target= beos-shared @@ -1174,6 +1210,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= bsd-gcc-shared @@ -1206,6 +1243,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1238,6 +1276,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1270,6 +1309,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1302,6 +1342,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = macosx $dso_scheme = dlfcn $shared_target= darwin-shared @@ -1334,6 +1375,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = osx32 $dso_scheme = dlfcn $shared_target= darwin-shared @@ -1366,6 +1408,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = osx64 $dso_scheme = dlfcn $shared_target= darwin-shared @@ -1398,6 +1441,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = macosx $dso_scheme = dlfcn $shared_target= darwin-shared @@ -1430,6 +1474,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1462,6 +1507,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= bsd-shared @@ -1494,6 +1540,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= cygwin-shared @@ -1526,6 +1573,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = win32n $dso_scheme = win32 $shared_target= @@ -1558,6 +1606,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = auto $dso_scheme = win32 $shared_target= @@ -1590,6 +1639,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = ias $dso_scheme = win32 $shared_target= @@ -1622,6 +1672,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1654,6 +1705,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1686,6 +1738,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1718,6 +1771,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1750,6 +1804,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1782,6 +1837,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -1814,6 +1870,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = $shared_target= @@ -1846,6 +1903,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = macosx $dso_scheme = dlfcn $shared_target= darwin-shared @@ -1878,6 +1936,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = osx32 $dso_scheme = dlfcn $shared_target= darwin-shared @@ -1910,6 +1969,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -1942,6 +2002,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -1974,6 +2035,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -2006,6 +2068,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -2038,6 +2101,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -2070,6 +2134,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -2102,6 +2167,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -2134,6 +2200,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -2166,6 +2233,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -2198,6 +2266,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -2230,6 +2299,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -2262,6 +2332,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= @@ -2294,6 +2365,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= @@ -2326,6 +2398,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -2358,6 +2431,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = $shared_target= @@ -2390,6 +2464,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -2422,6 +2497,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -2454,6 +2530,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-sparcv9.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -2486,6 +2563,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-sparcv9.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -2518,6 +2596,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -2550,6 +2629,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -2582,6 +2662,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -2614,6 +2695,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = win32 $shared_target= cygwin-shared @@ -2646,6 +2728,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = $shared_target= @@ -2678,6 +2761,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -2710,6 +2794,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -2742,6 +2827,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = $shared_target= @@ -2774,6 +2860,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -2806,6 +2893,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -2838,6 +2926,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dl $shared_target= hpux-shared @@ -2870,6 +2959,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dl $shared_target= hpux-shared @@ -2902,6 +2992,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= hpux-shared @@ -2934,6 +3025,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= hpux-shared @@ -2966,6 +3058,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dl $shared_target= hpux-shared @@ -2998,6 +3091,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dl $shared_target= hpux-shared @@ -3030,6 +3124,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dl $shared_target= hpux-shared @@ -3062,6 +3157,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-parisc.o +$engines_obj = $perlasm_scheme = 32 $dso_scheme = dl $shared_target= hpux-shared @@ -3094,6 +3190,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-parisc.o +$engines_obj = $perlasm_scheme = 32 $dso_scheme = dl $shared_target= hpux-shared @@ -3126,6 +3223,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dl $shared_target= hpux-shared @@ -3158,6 +3256,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dl $shared_target= hpux-shared @@ -3190,6 +3289,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= hpux-shared @@ -3222,6 +3322,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= hpux-shared @@ -3254,6 +3355,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-parisc.o +$engines_obj = $perlasm_scheme = 64 $dso_scheme = dlfcn $shared_target= hpux-shared @@ -3286,6 +3388,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= hpux-shared @@ -3318,6 +3421,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -3350,6 +3454,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = o32 $dso_scheme = dlfcn $shared_target= irix-shared @@ -3382,6 +3487,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = o32 $dso_scheme = dlfcn $shared_target= irix-shared @@ -3414,6 +3520,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = n32 $dso_scheme = dlfcn $shared_target= irix-shared @@ -3446,6 +3553,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = n32 $dso_scheme = dlfcn $shared_target= irix-shared @@ -3478,6 +3586,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = 64 $dso_scheme = dlfcn $shared_target= irix-shared @@ -3510,6 +3619,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = 64 $dso_scheme = dlfcn $shared_target= irix-shared @@ -3542,6 +3652,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-alpha.o +$engines_obj = $perlasm_scheme = void $dso_scheme = $shared_target= @@ -3574,6 +3685,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-alpha.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -3606,6 +3718,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-alpha.o +$engines_obj = $perlasm_scheme = void $dso_scheme = $shared_target= @@ -3638,6 +3751,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-alpha.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -3670,6 +3784,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = a.out $dso_scheme = $shared_target= @@ -3702,6 +3817,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-armv4.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -3734,6 +3850,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -3766,6 +3883,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -3798,6 +3916,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -3830,6 +3949,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -3862,6 +3982,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -3894,6 +4015,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -3926,6 +4048,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-ia64.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -3958,6 +4081,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = linux32 $dso_scheme = dlfcn $shared_target= linux-shared @@ -3990,6 +4114,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = linux64 $dso_scheme = dlfcn $shared_target= linux-shared @@ -4022,6 +4147,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -4054,6 +4180,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-sparcv9.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -4086,6 +4213,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= linux-shared @@ -4118,6 +4246,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-s390x.o +$engines_obj = $perlasm_scheme = 31 $dso_scheme = dlfcn $shared_target= linux-shared @@ -4150,6 +4279,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-s390x.o +$engines_obj = $perlasm_scheme = 64 $dso_scheme = dlfcn $shared_target= linux-shared @@ -4182,6 +4312,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-sparcv9.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= linux-shared @@ -4214,6 +4345,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = coff $dso_scheme = win32 $shared_target= cygwin-shared @@ -4246,6 +4378,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = mingw64 $dso_scheme = win32 $shared_target= cygwin-shared @@ -4278,6 +4411,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4310,6 +4444,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4342,6 +4477,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4374,6 +4510,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4406,6 +4543,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4438,6 +4576,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4470,6 +4609,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4502,6 +4642,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4534,6 +4675,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4566,6 +4708,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4598,6 +4741,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4630,6 +4774,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4662,6 +4807,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-alpha.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= alpha-osf1-shared @@ -4694,6 +4840,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-alpha.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= alpha-osf1-shared @@ -4726,6 +4873,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4758,6 +4906,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -4790,6 +4939,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = $shared_target= @@ -4822,6 +4972,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= svr3-shared @@ -4854,6 +5005,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= svr3-shared @@ -4886,6 +5038,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -4918,6 +5071,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -4950,6 +5104,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -4982,6 +5137,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -5014,6 +5170,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-sparcv9.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -5046,6 +5203,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-sparcv9.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -5078,6 +5236,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -5110,6 +5269,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= solaris-shared @@ -5142,6 +5302,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-sparcv9.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -5174,6 +5335,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-sparcv9.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= solaris-shared @@ -5206,6 +5368,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= solaris-shared @@ -5238,6 +5401,7 @@ $rc5_obj = $wp_obj = wp-x86_64.o $cmll_obj = cmll-x86_64.o cmll_misc.o $modes_obj = ghash-x86_64.o +$engines_obj = e_padlock-x86_64.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= solaris-shared @@ -5270,6 +5434,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = $shared_target= @@ -5302,6 +5467,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5334,6 +5500,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = ghash-alpha.o +$engines_obj = $perlasm_scheme = void $dso_scheme = dlfcn $shared_target= alpha-osf1-shared @@ -5366,6 +5533,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = $shared_target= linux-shared @@ -5398,6 +5566,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = $shared_target= linux-shared @@ -5430,6 +5599,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5462,6 +5632,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5494,6 +5665,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5526,6 +5698,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5558,6 +5731,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= svr5-shared @@ -5590,6 +5764,7 @@ $rc5_obj = rc5-586.o $wp_obj = wp_block.o wp-mmx.o $cmll_obj = cmll-x86.o $modes_obj = ghash-x86.o +$engines_obj = e_padlock-x86.o $perlasm_scheme = elf $dso_scheme = dlfcn $shared_target= gnu-shared @@ -5622,6 +5797,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = $shared_target= @@ -5654,6 +5830,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = void $dso_scheme = $shared_target= @@ -5686,6 +5863,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5718,6 +5896,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5750,6 +5929,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5782,6 +5962,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5814,6 +5995,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= @@ -5846,6 +6028,7 @@ $rc5_obj = $wp_obj = $cmll_obj = $modes_obj = +$engines_obj = $perlasm_scheme = $dso_scheme = $shared_target= diff --git a/crypto/perlasm/x86masm.pl b/crypto/perlasm/x86masm.pl index 6154b079b9..ee446de5c1 100644 --- a/crypto/perlasm/x86masm.pl +++ b/crypto/perlasm/x86masm.pl @@ -186,4 +186,11 @@ ___ sub ::dataseg { push(@out,"$segment\tENDS\n_DATA\tSEGMENT\n"); $segment="_DATA"; } +sub ::safeseh +{ my $nm=shift; + push(@out,"IF \@Version GE 710\n"); + push(@out,".SAFESEH ".&::LABEL($nm,$nmdecor.$nm)."\n"); + push(@out,"ENDIF\n"); +} + 1; diff --git a/crypto/perlasm/x86nasm.pl b/crypto/perlasm/x86nasm.pl index 23a84ba563..ca2511c9eb 100644 --- a/crypto/perlasm/x86nasm.pl +++ b/crypto/perlasm/x86nasm.pl @@ -167,4 +167,11 @@ sub ::dataseg else { push(@out,"section\t.data align=4\n"); } } +sub ::safeseh +{ my $nm=shift; + push(@out,"%if __NASM_VERSION_ID__ >= 0x02030000\n"); + push(@out,"safeseh ".&::LABEL($nm,$nmdecor.$nm)."\n"); + push(@out,"%endif\n"); +} + 1; diff --git a/engines/Makefile b/engines/Makefile index 65c4991135..190d4a1cdd 100644 --- a/engines/Makefile +++ b/engines/Makefile @@ -16,10 +16,13 @@ RECURSIVE_MAKE= [ -z "$(ENGDIRS)" ] || for i in $(ENGDIRS) ; do \ $(MAKE) -e TOP=../.. DIR=$$i $$target ) || exit 1; \ done; +ENGINES_ASM_OBJ= + PEX_LIBS= EX_LIBS= CFLAGS= $(INCLUDES) $(CFLAG) +ASFLAGS= $(INCLUDES) $(ASFLAG) GENERAL=Makefile engines.com install.com engine_vector.mar TEST= @@ -49,7 +52,8 @@ LIBOBJ= e_4758cca.o \ e_sureware.o \ e_ubsec.o \ e_padlock.o \ - e_capi.o + e_capi.o \ + $(ENGINES_ASM_OBJ) SRC= $(LIBSRC) @@ -77,7 +81,7 @@ lib: $(LIBOBJ) set -e; \ for l in $(LIBNAMES); do \ $(MAKE) -f ../Makefile.shared -e \ - LIBNAME=$$l LIBEXTRAS=e_$$l.o \ + LIBNAME=$$l LIBEXTRAS="e_$$l*.o" \ LIBDEPS='-L.. -lcrypto $(EX_LIBS)' \ link_o.$(SHLIB_TARGET); \ done; \ @@ -87,6 +91,11 @@ lib: $(LIBOBJ) fi; \ touch lib +e_padlock-x86.s: asm/e_padlock-x86.pl + $(PERL) asm/e_padlock-x86.pl $(PERLASM_SCHEME) $(CFLAGS) $(PROCESSOR) > $@ +e_padlock-x86_64.s: asm/e_padlock-x86_64.pl + $(PERL) asm/e_padlock-x86_64.pl $(PERLASM_SCHEME) > $@ + subdirs: echo $(EDIRS) @target=all; $(RECURSIVE_MAKE) diff --git a/engines/asm/e_padlock-x86.pl b/engines/asm/e_padlock-x86.pl new file mode 100644 index 0000000000..4a00d9d687 --- /dev/null +++ b/engines/asm/e_padlock-x86.pl @@ -0,0 +1,424 @@ +#!/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/. +# ==================================================================== + +# September 2011 +# +# Assembler helpers for Padlock engine. Compared to original engine +# version relying on inline assembler and compiled with gcc 3.4.6 it +# was measured to provide ~100% improvement on misaligned data in ECB +# mode and ~75% in CBC mode. For aligned data improvement can be +# observed for short inputs only, e.g. 45% for 64-byte messages in +# ECB mode, 20% in CBC. Difference in performance for aligned vs. +# misaligned data depends on misalignment and is either ~1.8x or +# ~2.9x. These are approximately same factors as for hardware support, +# so there is little reason to rely on the latter. It might actually +# hurt performance in mixture of aligned and misaligned buffers, +# because a) if you choose to flip 'align' flag on per-buffer basis, +# then you'd have to reload key context; b) if you choose to set +# 'align' flag permanently, it limits performance for aligned data +# to ~1/2. All results were collected on 1.5GHz C7. + +$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; +push(@INC,"${dir}","${dir}../../crypto/perlasm"); +require "x86asm.pl"; + +&asm_init($ARGV[0],$0); + +$PADLOCK_CHUNK=512; # Must be a power of 2 larger than 16 + +$ctx="edx"; +$out="edi"; +$inp="esi"; +$len="ecx"; +$chunk="ebx"; + +&function_begin_B("padlock_capability"); + &push ("ebx"); + &pushf (); + &pop ("eax"); + &mov ("ecx","eax"); + &xor ("eax",1<<21); + &push ("eax"); + &popf (); + &pushf (); + &pop ("eax"); + &xor ("ecx","eax"); + &xor ("eax","eax"); + &bt ("ecx",21); + &jnc (&label("noluck")); + &cpuid (); + &xor ("eax","eax"); + &cmp ("ebx","0x".unpack("H*",'tneC')); + &jne (&label("noluck")); + &cmp ("edx","0x".unpack("H*",'Hrua')); + &jne (&label("noluck")); + &cmp ("ecx","0x".unpack("H*",'slua')); + &jne (&label("noluck")); + &mov ("eax",0xC0000000); + &cpuid (); + &mov ("edx","eax"); + &xor ("eax","eax"); + &cmp ("edx",0xC0000001); + &jb (&label("noluck")); + &mov ("eax",1); + &cpuid (); + &or ("eax",0x0f); + &xor ("ebx","ebx"); + &and ("eax",0x0fff); + &cmp ("eax",0x06ff); # check for Nano + &sete ("bl"); + &mov ("eax",0xC0000001); + &push ("ebx"); + &cpuid (); + &pop ("ebx"); + &mov ("eax","edx"); + &shl ("ebx",4); # bit#4 denotes Nano + &and ("eax",0xffffffef); + &or ("eax","ebx") +&set_label("noluck"); + &pop ("ebx"); + &ret (); +&function_end_B("padlock_capability") + +&function_begin_B("padlock_key_bswap"); + &mov ("edx",&wparam(0)); + &mov ("ecx",&DWP(240,"edx")); +&set_label("bswap_loop"); + &mov ("eax",&DWP(0,"edx")); + &bswap ("eax"); + &mov (&DWP(0,"edx"),"eax"); + &lea ("edx",&DWP(4,"edx")); + &sub ("ecx",1); + &jnz (&label("bswap_loop")); + &ret (); +&function_end_B("padlock_key_bswap"); + +# This is heuristic key context tracing. At first one +# believes that one should use atomic swap instructions, +# but it's not actually necessary. Point is that if +# padlock_saved_context was changed by another thread +# after we've read it and before we compare it with ctx, +# our key *shall* be reloaded upon thread context switch +# and we are therefore set in either case... +&static_label("padlock_saved_context"); + +&function_begin_B("padlock_verify_context"); + &mov ($ctx,&wparam(0)); + &lea ("eax",&DWP("padlock_saved_context-".&label("verify_pic_point"))); + &pushf (); + &call ("_padlock_verify_ctx"); +&set_label("verify_pic_point"); + &lea ("esp",&DWP(4,"esp")); + &ret (); +&function_end_B("padlock_verify_context"); + +&function_begin_B("_padlock_verify_ctx"); + &add ("eax",&DWP(0,"esp")); # &padlock_saved_context + &bt (&DWP(4,"esp"),30); # eflags + &jnc (&label("verified")); + &cmp ($ctx,&DWP(0,"eax")); + &je (&label("verified")); + &pushf (); + &popf (); +&set_label("verified"); + &mov (&DWP(0,"eax"),$ctx); + &ret (); +&function_end_B("_padlock_verify_ctx"); + +&function_begin_B("padlock_reload_key"); + &pushf (); + &popf (); + &ret (); +&function_end_B("padlock_reload_key"); + +&function_begin_B("padlock_aes_block"); + &push ("edi"); + &push ("esi"); + &push ("ebx"); + &mov ($out,&wparam(0)); # must be 16-byte aligned + &mov ($inp,&wparam(1)); # must be 16-byte aligned + &mov ($ctx,&wparam(2)); + &mov ($len,1); + &lea ("ebx",&DWP(32,$ctx)); # key + &lea ($ctx,&DWP(16,$ctx)); # control word + &data_byte(0xf3,0x0f,0xa7,0xc8); # rep xcryptecb + &pop ("ebx"); + &pop ("esi"); + &pop ("edi"); + &ret (); +&function_end_B("padlock_aes_block"); + +sub generate_mode { +my ($mode,$opcode) = @_; +# int padlock_$mode_encrypt(void *out, const void *inp, +# struct padlock_cipher_data *ctx, size_t len); +&function_begin("padlock_${mode}_encrypt"); + &mov ($out,&wparam(0)); + &mov ($inp,&wparam(1)); + &mov ($ctx,&wparam(2)); + &mov ($len,&wparam(3)); + &test ($ctx,15); + &jnz (&label("${mode}_abort")); + &test ($len,15); + &jnz (&label("${mode}_abort")); + &lea ("eax",&DWP("padlock_saved_context-".&label("${mode}_pic_point"))); + &pushf (); + &cld (); + &call ("_padlock_verify_ctx"); +&set_label("${mode}_pic_point"); + &lea ($ctx,&DWP(16,$ctx)); # control word + &xor ("eax","eax"); + if ($mode eq "ctr16") { + &movdqa ("xmm0",&QWP(-16,$ctx));# load iv + } else { + &xor ("ebx","ebx"); + &test (&DWP(0,$ctx),1<<5); # align bit in control word + &jnz (&label("${mode}_aligned")); + &test ($out,0x0f); + &setz ("al"); # !out_misaligned + &test ($inp,0x0f); + &setz ("bl"); # !inp_misaligned + &test ("eax","ebx"); + &jnz (&label("${mode}_aligned")); + &neg ("eax"); + } + &mov ($chunk,$PADLOCK_CHUNK); + ¬ ("eax"); # out_misaligned?-1:0 + &lea ("ebp",&DWP(-24,"esp")); + &cmp ($len,$chunk); + &cmovc ($chunk,$len); # chunk=len>PADLOCK_CHUNK?PADLOCK_CHUNK:len + &and ("eax",$chunk); # out_misaligned?chunk:0 + &mov ($chunk,$len); + &neg ("eax"); + &and ($chunk,$PADLOCK_CHUNK-1); # chunk=len%PADLOCK_CHUNK + &lea ("esp",&DWP(0,"eax","ebp")); # alloca + &and ("esp",-16); + &jmp (&label("${mode}_loop")); + +&set_label("${mode}_loop",16); + &mov (&DWP(0,"ebp"),$out); # save parameters + &mov (&DWP(4,"ebp"),$inp); + &mov (&DWP(8,"ebp"),$len); + &mov ($len,$chunk); + &mov (&DWP(12,"ebp"),$chunk); # chunk + if ($mode eq "ctr16") { + &pextrw ("ecx","xmm0",7); # borrow $len + &mov ($inp,1); + &xor ($out,$out); + &xchg ("ch","cl"); +&set_label("${mode}_prepare"); + &movdqa (&QWP(0,"esp",$out),"xmm0"); + &lea ("eax",&DWP(0,"ecx",$inp)); + &xchg ("ah","al"); + &lea ($out,&DWP(16,$out)); + &pinsrw ("xmm0","eax",7); + &lea ($inp,&DWP(1,$inp)); + &cmp ($out,$chunk); + &jb (&label("${mode}_prepare")); + + &lea ($inp,&DWP(0,"esp")); + &lea ($out,&DWP(0,"esp")); + &mov ($len,$chunk); + } else { + &test ($out,0x0f); # out_misaligned + &cmovnz ($out,"esp"); + &test ($inp,0x0f); # inp_misaligned + &jz (&label("${mode}_inp_aligned")); + &shr ($len,2); + &data_byte(0xf3,0xa5); # rep movsl + &sub ($out,$chunk); + &mov ($len,$chunk); + &mov ($inp,$out); +&set_label("${mode}_inp_aligned"); + } + &lea ("eax",&DWP(-16,$ctx)); # ivp + &lea ("ebx",&DWP(16,$ctx)); # key + &shr ($len,4); # len/=AES_BLOCK_SIZE + &data_byte(0xf3,0x0f,0xa7,$opcode); # rep xcrypt* + if ($mode !~ /ecb|ctr/) { + &movdqa ("xmm0",&QWP(0,"eax")); + &movdqa (&DWP(-16,$ctx),"xmm0"); # copy [or refresh] iv + } + &mov ($out,&DWP(0,"ebp")); # restore parameters + &mov ($chunk,&DWP(12,"ebp")); + if ($mode eq "ctr16") { + &mov ($inp,&DWP(4,"ebp")); + &xor ($len,$len); +&set_label("${mode}_xor"); + &movdqu ("xmm1",&QWP(0,$inp,$len)); + &lea ($len,&DWP(16,$len)); + &pxor ("xmm1",&QWP(-16,"esp",$len)); + &movdqu (&QWP(-16,$out,$len),"xmm1"); + &cmp ($len,$chunk); + &jb (&label("${mode}_xor")); + } else { + &test ($out,0x0f); + &jz (&label("${mode}_out_aligned")); + &mov ($len,$chunk); + &shr ($len,2); + &lea ($inp,&DWP(0,"esp")); + &data_byte(0xf3,0xa5); # rep movsl + &sub ($out,$chunk); +&set_label("${mode}_out_aligned"); + &mov ($inp,&DWP(4,"ebp")); + } + &mov ($len,&DWP(8,"ebp")); + &add ($out,$chunk); + &add ($inp,$chunk); + &sub ($len,$chunk); + &mov ($chunk,$PADLOCK_CHUNK); + &jnz (&label("${mode}_loop")); + if ($mode eq "ctr16") { + &movdqa (&QWP(-16,$ctx),"xmm0"); # write out iv + &pxor ("xmm0","xmm0"); + &pxor ("xmm1","xmm1"); + } else { + &test ($out,0x0f); # out_misaligned + &jz (&label("${mode}_done")); + } + &mov ($len,"ebp"); + &mov ($out,"esp"); + &sub ($len,"esp"); + &xor ("eax","eax"); + &shr ($len,2); + &data_byte(0xf3,0xab); # rep stosl +&set_label("${mode}_done"); + &lea ("esp",&DWP(24,"ebp")); + if ($mode ne "ctr16") { + &jmp (&label("${mode}_exit")); + +&set_label("${mode}_aligned",16); + &lea ("eax",&DWP(-16,$ctx)); # ivp + &lea ("ebx",&DWP(16,$ctx)); # key + &shr ($len,4); # len/=AES_BLOCK_SIZE + &data_byte(0xf3,0x0f,0xa7,$opcode); # rep xcrypt* + if ($mode ne "ecb") { + &movdqa ("xmm0",&QWP(0,"eax")); + &movdqa (&DWP(-16,$ctx),"xmm0"); # copy [or refresh] iv + } +&set_label("${mode}_exit"); } + &mov ("eax",1); + &lea ("esp",&DWP(4,"esp")); # popf +&set_label("${mode}_abort"); +&function_end("padlock_${mode}_encrypt"); +} + +&generate_mode("ecb",0xc8); +&generate_mode("cbc",0xd0); +&generate_mode("cfb",0xe0); +&generate_mode("ofb",0xe8); +&generate_mode("ctr16",0xc8); # yes, it implements own ctr with ecb opcode, + # because hardware ctr was introduced later + # and even has errata on certain CPU stepping. + # own implementation *always* works... + +&function_begin_B("padlock_xstore"); + &push ("edi"); + &mov ("edi",&wparam(0)); + &mov ("edx",&wparam(1)); + &data_byte(0x0f,0xa7,0xc0); # xstore + &pop ("edi"); + &ret (); +&function_end_B("padlock_xstore"); + +&function_begin_B("_win32_segv_handler"); + &mov ("eax",1); # ExceptionContinueSearch + &mov ("edx",&wparam(0)); # *ExceptionRecord + &mov ("ecx",&wparam(2)); # *ContextRecord + &cmp (&DWP(0,"edx"),0xC0000005) # ExceptionRecord->ExceptionCode == STATUS_ACCESS_VIOLATION + &jne (&label("ret")); + &add (&DWP(184,"ecx"),4); # skip over rep sha* + &mov ("eax",0); # ExceptionContinueExecution +&set_label("ret"); + &ret (); +&function_end_B("_win32_segv_handler"); +&safeseh("_win32_segv_handler") if ($::win32); + +&function_begin_B("padlock_sha1_oneshot"); + &push ("edi"); + &push ("esi"); + &xor ("eax","eax"); + if ($::win32 or $::coff) { + &push (&::islabel("_win32_segv_handler")); + &data_byte(0x64,0xff,0x30); # push %fs:(%eax) + &data_byte(0x64,0x89,0x20); # mov %esp,%fs:(%eax) + } + &mov ("edi",&wparam(0)); + &mov ("esi",&wparam(1)); + &mov ("ecx",&wparam(2)); + &data_byte(0xf3,0x0f,0xa6,0xc8); # rep xsha1 + if ($::win32 or $::coff) { + &data_byte(0x64,0x8f,0x05,0,0,0,0); # pop %fs:0 + &lea ("esp",&DWP(4,"esp")); + } + &pop ("esi"); + &pop ("edi"); + &ret (); +&function_end_B("padlock_sha1_oneshot"); + +&function_begin_B("padlock_sha1"); + &push ("edi"); + &push ("esi"); + &mov ("eax",-1); + &mov ("edi",&wparam(0)); + &mov ("esi",&wparam(1)); + &mov ("ecx",&wparam(2)); + &data_byte(0xf3,0x0f,0xa6,0xc8); # rep xsha1 + &pop ("esi"); + &pop ("edi"); + &ret (); +&function_end_B("padlock_sha1"); + +&function_begin_B("padlock_sha256_oneshot"); + &push ("edi"); + &push ("esi"); + &xor ("eax","eax"); + if ($::win32 or $::coff) { + &push (&::islabel("_win32_segv_handler")); + &data_byte(0x64,0xff,0x30); # push %fs:(%eax) + &data_byte(0x64,0x89,0x20); # mov %esp,%fs:(%eax) + } + &mov ("edi",&wparam(0)); + &mov ("esi",&wparam(1)); + &mov ("ecx",&wparam(2)); + &data_byte(0xf3,0x0f,0xa6,0xd0); # rep xsha256 + if ($::win32 or $::coff) { + &data_byte(0x64,0x8f,0x05,0,0,0,0); # pop %fs:0 + &lea ("esp",&DWP(4,"esp")); + } + &pop ("esi"); + &pop ("edi"); + &ret (); +&function_end_B("padlock_sha256_oneshot"); + +&function_begin_B("padlock_sha256"); + &push ("edi"); + &push ("esi"); + &mov ("eax",-1); + &mov ("edi",&wparam(0)); + &mov ("esi",&wparam(1)); + &mov ("ecx",&wparam(2)); + &data_byte(0xf3,0x0f,0xa6,0xd0); # rep xsha256 + &pop ("esi"); + &pop ("edi"); + &ret (); +&function_end_B("padlock_sha256"); + +&asciz ("VIA Padlock x86 module, CRYPTOGAMS by <appro\@openssl.org>"); +&align (16); + +&dataseg(); +# Essentially this variable belongs in thread local storage. +# Having this variable global on the other hand can only cause +# few bogus key reloads [if any at all on signle-CPU system], +# so we accept the panalty... +&set_label("padlock_saved_context",4); +&data_word(0); + +&asm_finish(); diff --git a/engines/asm/e_padlock-x86_86.pl b/engines/asm/e_padlock-x86_86.pl new file mode 100644 index 0000000000..56cabc0f15 --- /dev/null +++ b/engines/asm/e_padlock-x86_86.pl @@ -0,0 +1,327 @@ +#!/usr/bin/env perl + +# ==================================================================== +# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL +# project. The module is, however, dual licensed under OpenSSL and +# CRYPTOGAMS licenses depending on where you obtain it. For further +# details see http://www.openssl.org/~appro/cryptogams/. +# ==================================================================== + +# September 2011 +# +# Assembler helpers for Padlock engine. + +$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}../../crypto/perlasm/x86_64-xlate.pl" and -f $xlate) or +die "can't locate x86_64-xlate.pl"; + +open STDOUT,"| $^X $xlate $flavour $output"; + +$code=".text\n"; + +$PADLOCK_CHUNK=512; # Must be a power of 2 larger than 16 + +$ctx="%rdx"; +$out="%rdi"; +$inp="%rsi"; +$len="%rcx"; +$chunk="%rbx"; + +($arg1,$arg2,$arg3,$arg4)=$win64?("%rcx","%rdx","%r8", "%r9") : # Win64 order + ("%rdi","%rsi","%rdx","%rcx"); # Unix order + +$code.=<<___; +.globl padlock_capability +.type padlock_capability,\@abi-omnipotent +.align 16 +padlock_capability: + mov %rbx,%r8 + xor %eax,%eax + cpuid + xor %eax,%eax + cmp \$`"0x".unpack("H*",'tneC')`,%ebx + jne .Lnoluck + cmp \$`"0x".unpack("H*",'Hrua')`,%edx + jne .Lnoluck + cmp \$`"0x".unpack("H*",'slua')`,%ecx + jne .Lnoluck + mov \$0xC0000000,%eax + cpuid + mov %eax,%edx + xor %eax,%eax + cmp \$0xC0000001,%edx + jb .Lnoluck + mov \$0xC0000001,%eax + cpuid + mov %edx,%eax + and \$0xffffffef,%eax + or \$0x10,%eax # set Nano bit#4 +.Lnoluck: + mov %r8,%rbx + ret +.size padlock_capability,.-padlock_capability + +.globl padlock_key_bswap +.type padlock_key_bswap,\@abi-omnipotent,0 +.align 16 +padlock_key_bswap: + mov 240($arg1),%edx +.Lbswap_loop: + mov ($arg1),%eax + bswap %eax + mov %eax,($arg1) + lea 4($arg1),$arg1 + sub \$1,%edx + jnz .Lbswap_loop + ret +.size padlock_key_bswap,.-padlock_key_bswap + +.globl padlock_verify_context +.type padlock_verify_context,\@abi-omnipotent +.align 16 +padlock_verify_context: + mov $arg1,$ctx + pushf + lea .Lpadlock_saved_context(%rip),%rax + call _padlock_verify_ctx + lea 8(%rsp),%rsp + ret +.size padlock_verify_context,.-padlock_verify_context + +.type _padlock_verify_ctx,\@abi-omnipotent +.align 16 +_padlock_verify_ctx: + mov 8(%rsp),%r8 + bt \$30,%r8 + jnc .Lverified + cmp (%rax),$ctx + je .Lverified + pushf + popf +.Lverified: + mov $ctx,(%rax) + ret +.size _padlock_verify_ctx,.-_padlock_verify_ctx + +.globl padlock_reload_key +.type padlock_reload_key,\@abi-omnipotent +.align 16 +padlock_reload_key: + pushf + popf + ret +.size padlock_reload_key,.-padlock_reload_key + +.globl padlock_aes_block +.type padlock_aes_block,\@function,3 +.align 16 +padlock_aes_block: + mov %rbx,%r8 + mov \$1,$len + lea 32($ctx),%rbx # key + lea 16($ctx),$ctx # control word + .byte 0xf3,0x0f,0xa7,0xc8 # rep xcryptecb + mov %r8,%rbx + ret +.size padlock_aes_block,.-padlock_aes_block + +.globl padlock_xstore +.type padlock_xstore,\@function,2 +.align 16 +padlock_xstore: + mov %esi,%edx + .byte 0x0f,0xa7,0xc0 # xstore + ret +.size padlock_xstore,.-padlock_xstore + +.globl padlock_sha1_oneshot +.type padlock_sha1_oneshot,\@function,3 +.align 16 +padlock_sha1_oneshot: + xor %rax,%rax + mov %rdx,%rcx + .byte 0xf3,0x0f,0xa6,0xc8 # rep xsha1 + ret +.size padlock_sha1_oneshot,.-padlock_sha1_oneshot + +.globl padlock_sha1 +.type padlock_sha1,\@function,3 +.align 16 +padlock_sha1: + mov \$-1,%rax + mov %rdx,%rcx + .byte 0xf3,0x0f,0xa6,0xc8 # rep xsha1 + ret +.size padlock_sha1,.-padlock_sha1 + +.globl padlock_sha256_oneshot +.type padlock_sha256_oneshot,\@function,3 +.align 16 +padlock_sha256_oneshot: + xor %rax,%rax + mov %rdx,%rcx + .byte 0xf3,0x0f,0xa6,0xd0 # rep xsha256 + ret +.size padlock_sha256_oneshot,.-padlock_sha256_oneshot + +.globl padlock_sha256 +.type padlock_sha256,\@function,3 +.align 16 +padlock_sha256: + mov \$-1,%rax + mov %rdx,%rcx + .byte 0xf3,0x0f,0xa6,0xd0 # rep xsha256 + ret +.size padlock_sha256,.-padlock_sha256 +___ + +sub generate_mode { +my ($mode,$opcode) = @_; +# int padlock_$mode_encrypt(void *out, const void *inp, +# struct padlock_cipher_data *ctx, size_t len); +$code.=<<___; +.globl padlock_${mode}_encrypt +.type padlock_${mode}_encrypt,\@function,4 +.align 16 +padlock_${mode}_encrypt: + push %rbp + push %rbx + + xor %eax,%eax + test \$15,$ctx + jnz .L${mode}_abort + test \$15,$len + jnz .L${mode}_abort + lea .Lpadlock_saved_context(%rip),%rax + pushf + cld + call _padlock_verify_ctx + lea 16($ctx),$ctx # control word + xor %eax,%eax + xor %ebx,%ebx + test \$`1<<5`,($ctx) # align bit in control word + test \$0x0f,$out + setz %al # !out_misaligned + test \$0x0f,$inp + setz %bl # !inp_misaligned + test %ebx,%eax + jnz .L${mode}_aligned + neg %rax + mov \$$PADLOCK_CHUNK,$chunk + not %rax # out_misaligned?-1:0 + lea (%rsp),%rbp + cmp $chunk,$len + cmovc $len,$chunk # chunk=len>PADLOCK_CHUNK?PADLOCK_CHUNK:len + and $chunk,%rax # out_misaligned?chunk:0 + mov $len,$chunk + neg %rax + and \$$PADLOCK_CHUNK-1,$chunk # chunk%=PADLOCK_CHUNK + lea (%rax,%rbp),%rsp + jmp .L${mode}_loop +.align 16 +.L${mode}_loop: + mov $out,%r8 # save parameters + mov $inp,%r9 + mov $len,%r10 + mov $chunk,$len + mov $chunk,%r11 + test \$0x0f,$out # out_misaligned + cmovnz %rsp,$out + test \$0x0f,$inp # inp_misaligned + jz .L${mode}_inp_aligned + shr \$3,$len + .byte 0xf3,0x48,0xa5 # rep movsq + sub $chunk,$out + mov $chunk,$len + mov $out,$inp +.L${mode}_inp_aligned: + lea -16($ctx),%rax # ivp + lea 16($ctx),%rbx # key + shr \$4,$len + .byte 0xf3,0x0f,0xa7,$opcode # rep xcrypt* +___ +$code.=<<___ if ($mode !~ /ecb|ctr/); + movdqa (%rax),%xmm0 + movdqa %xmm0,-16($ctx) # copy [or refresh] iv +___ +$code.=<<___; + mov %r8,$out # restore paramters + mov %r11,$chunk + test \$0x0f,$out + jz .L${mode}_out_aligned + mov $chunk,$len + shr \$3,$len + lea (%rsp),$inp + .byte 0xf3,0x48,0xa5 # rep movsq + sub $chunk,$out +.L${mode}_out_aligned: + mov %r9,$inp + mov %r10,$len + add $chunk,$out + add $chunk,$inp + sub $chunk,$len + mov \$$PADLOCK_CHUNK,$chunk + jnz .L${mode}_loop + + test \$0x0f,$out + jz .L${mode}_done + + mov %rbp,$len + mov %rsp,$out + sub %rsp,$len + xor %rax,%rax + shr \$3,$len + .byte 0xf3,0x48,0xab # rep stosq +.L${mode}_done: + lea (%rbp),%rsp + jmp .L${mode}_exit + +.align 16 +.L${mode}_aligned: + lea -16($ctx),%rax # ivp + lea 16($ctx),%rbx # key + shr \$4,$len # len/=AES_BLOCK_SIZE + .byte 0xf3,0x0f,0xa7,$opcode # rep xcrypt* +___ +$code.=<<___ if ($mode !~ /ecb|ctr/); + movdqa (%rax),%xmm0 + movdqa %xmm0,-16($ctx) # copy [or refresh] iv +___ +$code.=<<___; +.L${mode}_exit: + mov \$1,%eax + lea 8(%rsp),%rsp +.L${mode}_abort: + pop %rbx + pop %rbp + ret +.size padlock_${mode}_encrypt,.-padlock_${mode}_encrypt +___ +} + +&generate_mode("ecb",0xc8); +&generate_mode("cbc",0xd0); +&generate_mode("cfb",0xe0); +&generate_mode("ofb",0xe8); +&generate_mode("ctr16",0xd8); + +$code.=<<___; +.asciz "VIA Padlock x86_64 module, CRYPTOGAMS by <appro\@openssl.org>" +.align 16 +.data +.align 8 +.Lpadlock_saved_context: + .quad 0 +___ +$code =~ s/\`([^\`]*)\`/eval($1)/gem; + +print $code; + +close STDOUT; diff --git a/engines/e_padlock.c b/engines/e_padlock.c index 271253dada..7f78135165 100644 --- a/engines/e_padlock.c +++ b/engines/e_padlock.c @@ -95,17 +95,14 @@ /* VIA PadLock AES is available *ONLY* on some x86 CPUs. Not only that it doesn't exist elsewhere, but it - even can't be compiled on other platforms! + even can't be compiled on other platforms! */ - In addition, because of the heavy use of inline assembler, - compiler choice is limited to GCC and Microsoft C. */ #undef COMPILE_HW_PADLOCK -#if !defined(I386_ONLY) && !defined(OPENSSL_NO_INLINE_ASM) -# if (defined(__GNUC__) && __GNUC__>=2 && \ - (defined(__i386__) || defined(__i386) || \ - defined(__x86_64__) || defined(__x86_64)) \ - ) || \ - (defined(_MSC_VER) && defined(_M_IX86)) +#if !defined(I386_ONLY) && !defined(OPENSSL_NO_ASM) +# if defined(__i386__) || defined(__i386) || \ + defined(__x86_64__) || defined(__x86_64) || \ + defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64) || \ + defined(__INTEL__) # define COMPILE_HW_PADLOCK # ifdef OPENSSL_NO_DYNAMIC_ENGINE static ENGINE *ENGINE_padlock (void); @@ -130,19 +127,6 @@ void ENGINE_load_padlock (void) #endif #ifdef COMPILE_HW_PADLOCK -/* We do these includes here to avoid header problems on platforms that - do not have the VIA padlock anyway... */ -#include <stdlib.h> -#ifdef _WIN32 -# include <malloc.h> -# ifndef alloca -# define alloca _alloca -# endif -#elif defined(__GNUC__) -# ifndef alloca -# define alloca(s) __builtin_alloca((s)) -# endif -#endif /* Function for ENGINE detection and control */ static int padlock_available(void); @@ -163,9 +147,6 @@ static char padlock_name[100]; /* Available features */ static int padlock_use_ace = 0; /* Advanced Cryptography Engine */ static int padlock_use_rng = 0; /* Random Number Generator */ -#ifndef OPENSSL_NO_AES -static int padlock_aes_align_required = 1; -#endif /* ===== Engine "management" functions ===== */ @@ -284,202 +265,37 @@ struct padlock_cipher_data } cword; /* Control word */ AES_KEY ks; /* Encryption key */ }; - -/* - * Essentially this variable belongs in thread local storage. - * Having this variable global on the other hand can only cause - * few bogus key reloads [if any at all on single-CPU system], - * so we accept the penatly... - */ -static volatile struct padlock_cipher_data *padlock_saved_context; -#endif - -/* - * ======================================================= - * Inline assembler section(s). - * ======================================================= - * Order of arguments is chosen to facilitate Windows port - * using __fastcall calling convention. If you wish to add - * more routines, keep in mind that first __fastcall - * argument is passed in %ecx and second - in %edx. - * ======================================================= - */ -#if defined(__GNUC__) && __GNUC__>=2 -#if defined(__i386__) || defined(__i386) -/* - * As for excessive "push %ebx"/"pop %ebx" found all over. - * When generating position-independent code GCC won't let - * us use "b" in assembler templates nor even respect "ebx" - * in "clobber description." Therefore the trouble... - */ - -/* Helper function - check if a CPUID instruction - is available on this CPU */ -static int -padlock_insn_cpuid_available(void) -{ - int result = -1; - - /* We're checking if the bit #21 of EFLAGS - can be toggled. If yes = CPUID is available. */ - asm volatile ( - "pushf\n" - "popl %%eax\n" - "xorl $0x200000, %%eax\n" - "movl %%eax, %%ecx\n" - "andl $0x200000, %%ecx\n" - "pushl %%eax\n" - "popf\n" - "pushf\n" - "popl %%eax\n" - "andl $0x200000, %%eax\n" - "xorl %%eax, %%ecx\n" - "movl %%ecx, %0\n" - : "=r" (result) : : "eax", "ecx"); - - return (result == 0); -} - -/* Load supported features of the CPU to see if - the PadLock is available. */ -static int -padlock_available(void) -{ - char vendor_string[16]; - unsigned int eax, edx; - - /* First check if the CPUID instruction is available at all... */ - if (! padlock_insn_cpuid_available()) - return 0; - - /* Are we running on the Centaur (VIA) CPU? */ - eax = 0x00000000; - vendor_string[12] = 0; - asm volatile ( - "pushl %%ebx\n" - "cpuid\n" - "movl %%ebx,(%%edi)\n" - "movl %%edx,4(%%edi)\n" - "movl %%ecx,8(%%edi)\n" - "popl %%ebx" - : "+a"(eax) : "D"(vendor_string) : "ecx", "edx"); - if (strcmp(vendor_string, "CentaurHauls") != 0) - return 0; - - /* Check for Centaur Extended Feature Flags presence */ - eax = 0xC0000000; - asm volatile ("pushl %%ebx; cpuid; popl %%ebx" - : "+a"(eax) : : "ecx", "edx"); - if (eax < 0xC0000001) - return 0; - - /* Read the Centaur Extended Feature Flags */ - eax = 0xC0000001; - asm volatile ("pushl %%ebx; cpuid; popl %%ebx" - : "+a"(eax), "=d"(edx) : : "ecx"); - - /* Fill up some flags */ - padlock_use_ace = ((edx & (0x3<<6)) == (0x3<<6)); - padlock_use_rng = ((edx & (0x3<<2)) == (0x3<<2)); - - return padlock_use_ace + padlock_use_rng; -} - -/* Force key reload from memory to the CPU microcode. - Loading EFLAGS from the stack clears EFLAGS[30] - which does the trick. */ -static inline void -padlock_reload_key(void) -{ - asm volatile ("pushfl; popfl"); -} - -#ifndef OPENSSL_NO_AES -/* - * This is heuristic key context tracing. At first one - * believes that one should use atomic swap instructions, - * but it's not actually necessary. Point is that if - * padlock_saved_context was changed by another thread - * after we've read it and before we compare it with cdata, - * our key *shall* be reloaded upon thread context switch - * and we are therefore set in either case... - */ -static inline void -padlock_verify_context(struct padlock_cipher_data *cdata) -{ - asm volatile ( - "pushfl\n" -" btl $30,(%%esp)\n" -" jnc 1f\n" -" cmpl %2,%1\n" -" je 1f\n" -" popfl\n" -" subl $4,%%esp\n" -"1: addl $4,%%esp\n" -" movl %2,%0" - :"+m"(padlock_saved_context) - : "r"(padlock_saved_context), "r"(cdata) : "cc"); -} - -/* Template for padlock_xcrypt_* modes */ -/* BIG FAT WARNING: - * The offsets used with 'leal' instructions - * describe items of the 'padlock_cipher_data' - * structure. - */ -#define PADLOCK_XCRYPT_ASM(name,rep_xcrypt) \ -static inline void *name(size_t cnt, \ - struct padlock_cipher_data *cdata, \ - void *out, const void *inp) \ -{ void *iv; \ - asm volatile ( "pushl %%ebx\n" \ - " leal 16(%0),%%edx\n" \ - " leal 32(%0),%%ebx\n" \ - rep_xcrypt "\n" \ - " popl %%ebx" \ - : "=a"(iv), "=c"(cnt), "=D"(out), "=S"(inp) \ - : "0"(cdata), "1"(cnt), "2"(out), "3"(inp) \ - : "edx", "cc", "memory"); \ - return iv; \ -} #endif -#elif defined(__x86_64__) || defined(__x86_64) +/* Interface to assembler module */ +unsigned int padlock_capability(); +void padlock_key_bswap(AES_KEY *key); +void padlock_verify_context(struct padlock_cipher_data *ctx); +void padlock_reload_key(); +void padlock_aes_block(void *out, const void *inp, + struct padlock_cipher_data *ctx); +int padlock_ecb_encrypt(void *out, const void *inp, + struct padlock_cipher_data *ctx, size_t len); +int padlock_cbc_encrypt(void *out, const void *inp, + struct padlock_cipher_data *ctx, size_t len); +int padlock_cfb_encrypt(void *out, const void *inp, + struct padlock_cipher_data *ctx, size_t len); +int padlock_ofb_encrypt(void *out, const void *inp, + struct padlock_cipher_data *ctx, size_t len); +int padlock_ctr32_encrypt(void *out, const void *inp, + struct padlock_cipher_data *ctx, size_t len); +int padlock_xstore(void *out,int edx); +void padlock_sha1_oneshot(void *ctx,const void *inp,size_t len); +void padlock_sha1(void *ctx,const void *inp,size_t len); +void padlock_sha256_oneshot(void *ctx,const void *inp,size_t len); +void padlock_sha256(void *ctx,const void *inp,size_t len); /* Load supported features of the CPU to see if the PadLock is available. */ static int padlock_available(void) { - char vendor_string[16]; - unsigned int eax, edx; - size_t scratch; - - /* Are we running on the Centaur (VIA) CPU? */ - eax = 0x00000000; - vendor_string[12] = 0; - asm volatile ( - "movq %%rbx,%1\n" - "cpuid\n" - "movl %%ebx,(%2)\n" - "movl %%edx,4(%2)\n" - "movl %%ecx,8(%2)\n" - "movq %1,%%rbx" - : "+a"(eax), "=&r"(scratch) : "r"(vendor_string) : "rcx", "rdx"); - if (strcmp(vendor_string, "CentaurHauls") != 0) - return 0; - - /* Check for Centaur Extended Feature Flags presence */ - eax = 0xC0000000; - asm volatile ("movq %%rbx,%1; cpuid; movq %1,%%rbx" - : "+a"(eax), "=&r"(scratch) : : "rcx", "rdx"); - if (eax < 0xC0000001) - return 0; - - /* Read the Centaur Extended Feature Flags */ - eax = 0xC0000001; - asm volatile ("movq %%rbx,%2; cpuid; movq %2,%%rbx" - : "+a"(eax), "=d"(edx), "=&r"(scratch) : : "rcx"); + unsigned int edx = padlock_capability(); /* Fill up some flags */ padlock_use_ace = ((edx & (0x3<<6)) == (0x3<<6)); @@ -488,257 +304,6 @@ padlock_available(void) return padlock_use_ace + padlock_use_rng; } -/* Force key reload from memory to the CPU microcode. - Loading EFLAGS from the stack clears EFLAGS[30] - which does the trick. */ -static inline void -padlock_reload_key(void) -{ - asm volatile ("pushfq; popfq"); -} - -#ifndef OPENSSL_NO_AES -/* - * This is heuristic key context tracing. At first one - * believes that one should use atomic swap instructions, - * but it's not actually necessary. Point is that if - * padlock_saved_context was changed by another thread - * after we've read it and before we compare it with cdata, - * our key *shall* be reloaded upon thread context switch - * and we are therefore set in either case... - */ -static inline void -padlock_verify_context(struct padlock_cipher_data *cdata) -{ - asm volatile ( - "pushfq\n" -" btl $30,(%%rsp)\n" -" jnc 1f\n" -" cmpq %2,%1\n" -" je 1f\n" -" popfq\n" -" subq $8,%%rsp\n" -"1: addq $8,%%rsp\n" -" movq %2,%0" - :"+m"(padlock_saved_context) - : "r"(padlock_saved_context), "r"(cdata) : "cc"); -} - -/* Template for padlock_xcrypt_* modes */ -/* BIG FAT WARNING: - * The offsets used with 'leal' instructions - * describe items of the 'padlock_cipher_data' - * structure. - */ -#define PADLOCK_XCRYPT_ASM(name,rep_xcrypt) \ -static inline void *name(size_t cnt, \ - struct padlock_cipher_data *cdata, \ - void *out, const void *inp) \ -{ void *iv; \ - size_t scratch; \ - asm volatile ( "movq %%rbx,%4\n" \ - " leaq 16(%0),%%rdx\n" \ - " leaq 32(%0),%%rbx\n" \ - rep_xcrypt "\n" \ - " movq %4,%%rbx" \ - : "=a"(iv), "=c"(cnt), "=D"(out), "=S"(inp), "=&r"(scratch) \ - : "0"(cdata), "1"(cnt), "2"(out), "3"(inp) \ - : "rdx", "cc", "memory"); \ - return iv; \ -} -#endif - -#endif /* cpu */ - -#ifndef OPENSSL_NO_AES -/* Generate all functions with appropriate opcodes */ -PADLOCK_XCRYPT_ASM(padlock_xcrypt_ecb, ".byte 0xf3,0x0f,0xa7,0xc8") /* rep xcryptecb */ -PADLOCK_XCRYPT_ASM(padlock_xcrypt_cbc, ".byte 0xf3,0x0f,0xa7,0xd0") /* rep xcryptcbc */ -PADLOCK_XCRYPT_ASM(padlock_xcrypt_cfb, ".byte 0xf3,0x0f,0xa7,0xe0") /* rep xcryptcfb */ -PADLOCK_XCRYPT_ASM(padlock_xcrypt_ofb, ".byte 0xf3,0x0f,0xa7,0xe8") /* rep xcryptofb */ - -/* Our own htonl()/ntohl() */ -static inline void -padlock_bswapl(AES_KEY *ks) -{ - size_t i = sizeof(ks->rd_key)/sizeof(ks->rd_key[0]); - unsigned int *key = ks->rd_key; - - while (i--) { - asm volatile ("bswapl %0" : "+r"(*key)); - key++; - } -} -#endif - -/* The RNG call itself */ -static inline unsigned int -padlock_xstore(void *addr, unsigned int edx_in) -{ - unsigned int eax_out; - - asm volatile (".byte 0x0f,0xa7,0xc0" /* xstore */ - : "=a"(eax_out),"=m"(*(unsigned *)addr) - : "D"(addr), "d" (edx_in) - ); - - return eax_out; -} - -/* Why not inline 'rep movsd'? I failed to find information on what - * value in Direction Flag one can expect and consequently have to - * apply "better-safe-than-sorry" approach and assume "undefined." - * I could explicitly clear it and restore the original value upon - * return from padlock_aes_cipher, but it's presumably too much - * trouble for too little gain... - * - * In case you wonder 'rep xcrypt*' instructions above are *not* - * affected by the Direction Flag and pointers advance toward - * larger addresses unconditionally. - */ -static inline unsigned char * -padlock_memcpy(void *dst,const void *src,size_t n) -{ - size_t *d=dst; - const size_t *s=src; - - n /= sizeof(*d); - do { *d++ = *s++; } while (--n); - - return dst; -} - -#elif defined(_MSC_VER) -/* - * Unlike GCC these are real functions. In order to minimize impact - * on performance we adhere to __fastcall calling convention in - * order to get two first arguments passed through %ecx and %edx. - * Which kind of suits very well, as instructions in question use - * both %ecx and %edx as input:-) - */ -#define REP_XCRYPT(code) \ - _asm _emit 0xf3 \ - _asm _emit 0x0f _asm _emit 0xa7 \ - _asm _emit code - -/* BIG FAT WARNING: - * The offsets used with 'lea' instructions - * describe items of the 'padlock_cipher_data' - * structure. - */ -#define PADLOCK_XCRYPT_ASM(name,code) \ -static void * __fastcall \ - name (size_t cnt, void *cdata, \ - void *outp, const void *inp) \ -{ _asm mov eax,edx \ - _asm lea edx,[eax+16] \ - _asm lea ebx,[eax+32] \ - _asm mov edi,outp \ - _asm mov esi,inp \ - REP_XCRYPT(code) \ -} - -PADLOCK_XCRYPT_ASM(padlock_xcrypt_ecb,0xc8) -PADLOCK_XCRYPT_ASM(padlock_xcrypt_cbc,0xd0) -PADLOCK_XCRYPT_ASM(padlock_xcrypt_cfb,0xe0) -PADLOCK_XCRYPT_ASM(padlock_xcrypt_ofb,0xe8) - -static int __fastcall -padlock_xstore(void *outp,unsigned int code) -{ _asm mov edi,ecx - _asm _emit 0x0f _asm _emit 0xa7 _asm _emit 0xc0 -} - -static void __fastcall -padlock_reload_key(void) -{ _asm pushfd _asm popfd } - -static void __fastcall -padlock_verify_context(void *cdata) -{ _asm { - pushfd - bt DWORD PTR[esp],30 - jnc skip - cmp ecx,padlock_saved_context - je skip - popfd - sub esp,4 - skip: add esp,4 - mov padlock_saved_context,ecx - } -} - -static int -padlock_available(void) -{ _asm { - pushfd - pop eax - mov ecx,eax - xor eax,1<<21 - push eax - popfd - pushfd - pop eax - xor eax,ecx - bt eax,21 - jnc noluck - mov eax,0 - cpuid - xor eax,eax - cmp ebx,'tneC' - jne noluck - cmp edx,'Hrua' - jne noluck - cmp ecx,'slua' - jne noluck - mov eax,0xC0000000 - cpuid - mov edx,eax - xor eax,eax - cmp edx,0xC0000001 - jb noluck - mov eax,0xC0000001 - cpuid - xor eax,eax - bt edx,6 - jnc skip_a - bt edx,7 - jnc skip_a - mov padlock_use_ace,1 - inc eax - skip_a: bt edx,2 - jnc skip_r - bt edx,3 - jnc skip_r - mov padlock_use_rng,1 - inc eax - skip_r: - noluck: - } -} - -static void __fastcall -padlock_bswapl(void *key) -{ _asm { - pushfd - cld - mov esi,ecx - mov edi,ecx - mov ecx,60 - up: lodsd - bswap eax - stosd - loop up - popfd - } -} - -/* MS actually specifies status of Direction Flag and compiler even - * manages to compile following as 'rep movsd' all by itself... - */ -#define padlock_memcpy(o,i,n) ((unsigned char *)memcpy((o),(i),(n)&~3U)) -#endif - /* ===== AES encryption/decryption ===== */ #ifndef OPENSSL_NO_AES @@ -789,14 +354,157 @@ static int padlock_cipher_nids_num = (sizeof(padlock_cipher_nids)/ /* Function prototypes ... */ static int padlock_aes_init_key(EVP_CIPHER_CTX *ctx, const unsigned char *key, const unsigned char *iv, int enc); -static int padlock_aes_cipher(EVP_CIPHER_CTX *ctx, unsigned char *out, - const unsigned char *in, size_t nbytes); #define NEAREST_ALIGNED(ptr) ( (unsigned char *)(ptr) + \ ( (0x10 - ((size_t)(ptr) & 0x0F)) & 0x0F ) ) #define ALIGNED_CIPHER_DATA(ctx) ((struct padlock_cipher_data *)\ NEAREST_ALIGNED(ctx->cipher_data)) +static int +padlock_ecb_cipher(EVP_CIPHER_CTX *ctx, unsigned char *out_arg, + const unsigned char *in_arg, size_t nbytes) +{ + return padlock_ecb_encrypt(out_arg,in_arg, + ALIGNED_CIPHER_DATA(ctx),nbytes); +} +static int +padlock_cbc_cipher(EVP_CIPHER_CTX *ctx, unsigned char *out_arg, + const unsigned char *in_arg, size_t nbytes) +{ + struct padlock_cipher_data *cdata = ALIGNED_CIPHER_DATA(ctx); + int ret; + + memcpy(cdata->iv, ctx->iv, AES_BLOCK_SIZE); + if ((ret = padlock_cbc_encrypt(out_arg,in_arg,cdata,nbytes))) + memcpy(ctx->iv, cdata->iv, AES_BLOCK_SIZE); + return ret; +} + +static int +padlock_cfb_cipher(EVP_CIPHER_CTX *ctx, unsigned char *out_arg, + const unsigned char *in_arg, size_t nbytes) +{ + struct padlock_cipher_data *cdata = ALIGNED_CIPHER_DATA(ctx); + size_t chunk; + + if ((chunk = ctx->num)) { /* borrow chunk variable */ + unsigned char *ivp=ctx->iv; + + if (chunk >= AES_BLOCK_SIZE) + return 0; /* bogus value */ + + if (ctx->encrypt) + while (chunk<AES_BLOCK_SIZE && nbytes!=0) { + ivp[chunk] = *(out_arg++) = *(in_arg++) ^ ivp[chunk]; + chunk++, nbytes--; + } + else while (chunk<AES_BLOCK_SIZE && nbytes!=0) { + unsigned char c = *(in_arg++); + *(out_arg++) = c ^ ivp[chunk]; + ivp[chunk++] = c, nbytes--; + } + + ctx->num = chunk%AES_BLOCK_SIZE; + } + + if (nbytes == 0) + return 1; + + memcpy (cdata->iv, ctx->iv, AES_BLOCK_SIZE); + + if ((chunk = nbytes & ~(AES_BLOCK_SIZE-1))) { + if (!padlock_cfb_encrypt(out_arg,in_arg,cdata,chunk)) + return 0; + nbytes -= chunk; + } + + if (nbytes) { + unsigned char *ivp = cdata->iv; + + out_arg += chunk; + in_arg += chunk; + ctx->num = nbytes; + if (cdata->cword.b.encdec) { + cdata->cword.b.encdec=0; + padlock_reload_key(); + padlock_aes_block(ivp,ivp,cdata); + cdata->cword.b.encdec=1; + padlock_reload_key(); + while(nbytes) { + unsigned char c = *(in_arg++); + *(out_arg++) = c ^ *ivp; + *(ivp++) = c, nbytes--; + } + } + else { padlock_reload_key(); + padlock_aes_block(ivp,ivp,cdata); + padlock_reload_key(); + while (nbytes) { + *ivp = *(out_arg++) = *(in_arg++) ^ *ivp; + ivp++, nbytes--; + } + } + } + + memcpy(ctx->iv, cdata->iv, AES_BLOCK_SIZE); + + return 1; +} + +static int +padlock_ofb_cipher(EVP_CIPHER_CTX *ctx, unsigned char *out_arg, + const unsigned char *in_arg, size_t nbytes) +{ + struct padlock_cipher_data *cdata = ALIGNED_CIPHER_DATA(ctx); + size_t chunk; + + /* ctx->num is maintained in byte-oriented modes, + such as CFB and OFB... */ + if ((chunk = ctx->num)) { /* borrow chunk variable */ + unsigned char *ivp=ctx->iv; + + if (chunk >= AES_BLOCK_SIZE) + return 0; /* bogus value */ + + while (chunk<AES_BLOCK_SIZE && nbytes!=0) { + *(out_arg++) = *(in_arg++) ^ ivp[chunk]; + chunk++, nbytes--; + } + + ctx->num = chunk%AES_BLOCK_SIZE; + } + + if (nbytes == 0) + return 1; + + memcpy(cdata->iv, ctx->iv, AES_BLOCK_SIZE); + + if ((chunk = nbytes & ~(AES_BLOCK_SIZE-1))) { + if (!padlock_ofb_encrypt(out_arg,in_arg,cdata,chunk)) + return 0; + nbytes -= chunk; + } + + if (nbytes) { + unsigned char *ivp = cdata->iv; + + out_arg += chunk; + in_arg += chunk; + ctx->num = nbytes; + padlock_reload_key(); /* empirically found */ + padlock_aes_block(ivp,ivp,cdata); + padlock_reload_key(); /* empirically found */ + while (nbytes) { + *(out_arg++) = *(in_arg++) ^ *ivp; + ivp++, nbytes--; + } + } + + memcpy(ctx->iv, cdata->iv, AES_BLOCK_SIZE); + + return 1; +} + #define EVP_CIPHER_block_size_ECB AES_BLOCK_SIZE #define EVP_CIPHER_block_size_CBC AES_BLOCK_SIZE #define EVP_CIPHER_block_size_OFB 1 @@ -812,7 +520,7 @@ static const EVP_CIPHER padlock_aes_##ksize##_##lmode = { \ AES_BLOCK_SIZE, \ 0 | EVP_CIPH_##umode##_MODE, \ padlock_aes_init_key, \ - padlock_aes_cipher, \ + padlock_##lmode##_cipher, \ NULL, \ sizeof(struct padlock_cipher_data) + 16, \ EVP_CIPHER_set_asn1_iv, \ @@ -932,15 +640,15 @@ padlock_aes_init_key (EVP_CIPHER_CTX *ctx, const unsigned char *key, and is listed as hardware errata. They most likely will fix it at some point and then a check for stepping would be due here. */ - if (EVP_CIPHER_CTX_mode(ctx) == EVP_CIPH_CFB_MODE || - EVP_CIPHER_CTX_mode(ctx) == EVP_CIPH_OFB_MODE || - enc) - AES_set_encrypt_key(key, key_len, &cdata->ks); - else + if ((EVP_CIPHER_CTX_mode(ctx) == EVP_CIPH_ECB_MODE || + EVP_CIPHER_CTX_mode(ctx) == EVP_CIPH_CBC_MODE) + && !enc) AES_set_decrypt_key(key, key_len, &cdata->ks); + else + AES_set_encrypt_key(key, key_len, &cdata->ks); #ifndef AES_ASM /* OpenSSL C functions use byte-swapped extended key. */ - padlock_bswapl(&cdata->ks); + padlock_key_bswap(&cdata->ks); #endif cdata->cword.b.keygen = 1; break; @@ -960,321 +668,6 @@ padlock_aes_init_key (EVP_CIPHER_CTX *ctx, const unsigned char *key, return 1; } -/* - * Simplified version of padlock_aes_cipher() used when - * 1) both input and output buffers are at aligned addresses. - * or when - * 2) running on a newer CPU that doesn't require aligned buffers. - */ -static int -padlock_aes_cipher_omnivorous(EVP_CIPHER_CTX *ctx, unsigned char *out_arg, - const unsigned char *in_arg, size_t nbytes) -{ - struct padlock_cipher_data *cdata; - void *iv; - - cdata = ALIGNED_CIPHER_DATA(ctx); - padlock_verify_context(cdata); - - switch (EVP_CIPHER_CTX_mode(ctx)) { - case EVP_CIPH_ECB_MODE: - padlock_xcrypt_ecb(nbytes/AES_BLOCK_SIZE, cdata, out_arg, in_arg); - break; - - case EVP_CIPH_CBC_MODE: - memcpy(cdata->iv, ctx->iv, AES_BLOCK_SIZE); - iv = padlock_xcrypt_cbc(nbytes/AES_BLOCK_SIZE, cdata, out_arg, in_arg); - memcpy(ctx->iv, iv, AES_BLOCK_SIZE); - break; - - case EVP_CIPH_CFB_MODE: - memcpy(cdata->iv, ctx->iv, AES_BLOCK_SIZE); - iv = padlock_xcrypt_cfb(nbytes/AES_BLOCK_SIZE, cdata, out_arg, in_arg); - memcpy(ctx->iv, iv, AES_BLOCK_SIZE); - break; - - case EVP_CIPH_OFB_MODE: - memcpy(cdata->iv, ctx->iv, AES_BLOCK_SIZE); - padlock_xcrypt_ofb(nbytes/AES_BLOCK_SIZE, cdata, out_arg, in_arg); - memcpy(ctx->iv, cdata->iv, AES_BLOCK_SIZE); - break; - - default: - return 0; - } - - memset(cdata->iv, 0, AES_BLOCK_SIZE); - - return 1; -} - -#ifndef PADLOCK_CHUNK -# define PADLOCK_CHUNK 512 /* Must be a power of 2 larger than 16 */ -#endif -#if PADLOCK_CHUNK<16 || PADLOCK_CHUNK&(PADLOCK_CHUNK-1) -# error "insane PADLOCK_CHUNK..." -#endif - -/* Re-align the arguments to 16-Bytes boundaries and run the - encryption function itself. This function is not AES-specific. */ -static int -padlock_aes_cipher(EVP_CIPHER_CTX *ctx, unsigned char *out_arg, - const unsigned char *in_arg, size_t nbytes) -{ - struct padlock_cipher_data *cdata; - const void *inp; - unsigned char *out; - void *iv; - int inp_misaligned, out_misaligned, realign_in_loop; - size_t chunk, allocated=0; - - /* ctx->num is maintained in byte-oriented modes, - such as CFB and OFB... */ - if ((chunk = ctx->num)) { /* borrow chunk variable */ - unsigned char *ivp=ctx->iv; - - switch (EVP_CIPHER_CTX_mode(ctx)) { - case EVP_CIPH_CFB_MODE: - if (chunk >= AES_BLOCK_SIZE) - return 0; /* bogus value */ - - if (ctx->encrypt) - while (chunk<AES_BLOCK_SIZE && nbytes!=0) { - ivp[chunk] = *(out_arg++) = *(in_arg++) ^ ivp[chunk]; - chunk++, nbytes--; - } - else while (chunk<AES_BLOCK_SIZE && nbytes!=0) { - unsigned char c = *(in_arg++); - *(out_arg++) = c ^ ivp[chunk]; - ivp[chunk++] = c, nbytes--; - } - - ctx->num = chunk%AES_BLOCK_SIZE; - break; - case EVP_CIPH_OFB_MODE: - if (chunk >= AES_BLOCK_SIZE) - return 0; /* bogus value */ - - while (chunk<AES_BLOCK_SIZE && nbytes!=0) { - *(out_arg++) = *(in_arg++) ^ ivp[chunk]; - chunk++, nbytes--; - } - - ctx->num = chunk%AES_BLOCK_SIZE; - break; - } - } - - if (nbytes == 0) - return 1; -#if 0 - if (nbytes % AES_BLOCK_SIZE) - return 0; /* are we expected to do tail processing? */ -#else - /* nbytes is always multiple of AES_BLOCK_SIZE in ECB and CBC - modes and arbitrary value in byte-oriented modes, such as - CFB and OFB... */ -#endif - - /* VIA promises CPUs that won't require alignment in the future. - For now padlock_aes_align_required is initialized to 1 and - the condition is never met... */ - /* C7 core is capable to manage unaligned input in non-ECB[!] - mode, but performance penalties appear to be approximately - same as for software alignment below or ~3x. They promise to - improve it in the future, but for now we can just as well - pretend that it can only handle aligned input... */ - if (!padlock_aes_align_required && (nbytes%AES_BLOCK_SIZE)==0) - return padlock_aes_cipher_omnivorous(ctx, out_arg, in_arg, nbytes); - - inp_misaligned = (((size_t)in_arg) & 0x0F); - out_misaligned = (((size_t)out_arg) & 0x0F); - - /* Note that even if output is aligned and input not, - * I still prefer to loop instead of copy the whole - * input and then encrypt in one stroke. This is done - * in order to improve L1 cache utilization... */ - realign_in_loop = out_misaligned|inp_misaligned; - - if (!realign_in_loop && (nbytes%AES_BLOCK_SIZE)==0) - return padlock_aes_cipher_omnivorous(ctx, out_arg, in_arg, nbytes); - - /* this takes one "if" out of the loops */ - chunk = nbytes; - chunk %= PADLOCK_CHUNK; - if (chunk==0) chunk = PADLOCK_CHUNK; - - if (out_misaligned) { - /* optmize for small input */ - allocated = (chunk<nbytes?PADLOCK_CHUNK:nbytes); - out = alloca(0x10 + allocated); - out = NEAREST_ALIGNED(out); - } - else - out = out_arg; - - cdata = ALIGNED_CIPHER_DATA(ctx); - padlock_verify_context(cdata); - - switch (EVP_CIPHER_CTX_mode(ctx)) { - case EVP_CIPH_ECB_MODE: - do { - if (inp_misaligned) - inp = padlock_memcpy(out, in_arg, chunk); - else - inp = in_arg; - in_arg += chunk; - - padlock_xcrypt_ecb(chunk/AES_BLOCK_SIZE, cdata, out, inp); - - if (out_misaligned) - out_arg = padlock_memcpy(out_arg, out, chunk) + chunk; - else - out = out_arg+=chunk; - - nbytes -= chunk; - chunk = PADLOCK_CHUNK; - } while (nbytes); - break; - - case EVP_CIPH_CBC_MODE: - memcpy(cdata->iv, ctx->iv, AES_BLOCK_SIZE); - goto cbc_shortcut; - do { - if (iv != cdata->iv) - memcpy(cdata->iv, iv, AES_BLOCK_SIZE); - chunk = PADLOCK_CHUNK; - cbc_shortcut: /* optimize for small input */ - if (inp_misaligned) - inp = padlock_memcpy(out, in_arg, chunk); - else - inp = in_arg; - in_arg += chunk; - - iv = padlock_xcrypt_cbc(chunk/AES_BLOCK_SIZE, cdata, out, inp); - - if (out_misaligned) - out_arg = padlock_memcpy(out_arg, out, chunk) + chunk; - else - out = out_arg+=chunk; - - } while (nbytes -= chunk); - memcpy(ctx->iv, iv, AES_BLOCK_SIZE); - break; - - case EVP_CIPH_CFB_MODE: - memcpy (iv = cdata->iv, ctx->iv, AES_BLOCK_SIZE); - chunk &= ~(AES_BLOCK_SIZE-1); - if (chunk) goto cfb_shortcut; - else goto cfb_skiploop; - do { - if (iv != cdata->iv) - memcpy(cdata->iv, iv, AES_BLOCK_SIZE); - chunk = PADLOCK_CHUNK; - cfb_shortcut: /* optimize for small input */ - if (inp_misaligned) - inp = padlock_memcpy(out, in_arg, chunk); - else - inp = in_arg; - in_arg += chunk; - - iv = padlock_xcrypt_cfb(chunk/AES_BLOCK_SIZE, cdata, out, inp); - - if (out_misaligned) - out_arg = padlock_memcpy(out_arg, out, chunk) + chunk; - else - out = out_arg+=chunk; - - nbytes -= chunk; - } while (nbytes >= AES_BLOCK_SIZE); - - cfb_skiploop: - if (nbytes) { - unsigned char *ivp = cdata->iv; - - if (iv != ivp) { - memcpy(ivp, iv, AES_BLOCK_SIZE); - iv = ivp; - } - ctx->num = nbytes; - if (cdata->cword.b.encdec) { - cdata->cword.b.encdec=0; - padlock_reload_key(); - padlock_xcrypt_ecb(1,cdata,ivp,ivp); - cdata->cword.b.encdec=1; - padlock_reload_key(); - while(nbytes) { - unsigned char c = *(in_arg++); - *(out_arg++) = c ^ *ivp; - *(ivp++) = c, nbytes--; - } - } - else { padlock_reload_key(); - padlock_xcrypt_ecb(1,cdata,ivp,ivp); - padlock_reload_key(); - while (nbytes) { - *ivp = *(out_arg++) = *(in_arg++) ^ *ivp; - ivp++, nbytes--; - } - } - } - - memcpy(ctx->iv, iv, AES_BLOCK_SIZE); - break; - - case EVP_CIPH_OFB_MODE: - memcpy(cdata->iv, ctx->iv, AES_BLOCK_SIZE); - chunk &= ~(AES_BLOCK_SIZE-1); - if (chunk) do { - if (inp_misaligned) - inp = padlock_memcpy(out, in_arg, chunk); - else - inp = in_arg; - in_arg += chunk; - - padlock_xcrypt_ofb(chunk/AES_BLOCK_SIZE, cdata, out, inp); - - if (out_misaligned) - out_arg = padlock_memcpy(out_arg, out, chunk) + chunk; - else - out = out_arg+=chunk; - - nbytes -= chunk; - chunk = PADLOCK_CHUNK; - } while (nbytes >= AES_BLOCK_SIZE); - - if (nbytes) { - unsigned char *ivp = cdata->iv; - - ctx->num = nbytes; - padlock_reload_key(); /* empirically found */ - padlock_xcrypt_ecb(1,cdata,ivp,ivp); - padlock_reload_key(); /* empirically found */ - while (nbytes) { - *(out_arg++) = *(in_arg++) ^ *ivp; - ivp++, nbytes--; - } - } - - memcpy(ctx->iv, cdata->iv, AES_BLOCK_SIZE); - break; - - default: - return 0; - } - - /* Clean the realign buffer if it was used */ - if (out_misaligned) { - volatile unsigned long *p=(void *)out; - size_t n = allocated/sizeof(*p); - while (n--) *p++=0; - } - - memset(cdata->iv, 0, AES_BLOCK_SIZE); - - return 1; -} - #endif /* OPENSSL_NO_AES */ /* ===== Random Number Generator ===== */ |