Update documentation

This testing occurs on Windows XP. We are still rockin it

Doxyfile

@@ -41,7 +41,7 @@ PROJECT_NAME           = Crypto++
 # could be handy for archiving the generated documentation or if some version
 # control system is used.
 
-PROJECT_NUMBER         = 8.1
+PROJECT_NUMBER         = 8.3
 
 # Using the PROJECT_BRIEF tag one can provide an optional one line description
 # for a project that appears at the top of each page and should give viewer a

Filelist.txt

@@ -66,6 +66,16 @@ channels.h
 cmac.cpp
 cmac.h
 config.h
+config_align.h
+config_asm.h
+config_cpu.h
+config_cxx.h
+config_dll.h
+config_int.h
+config_misc.h
+config_ns.h
+config_os.h
+config_ver.h
 cpu.cpp
 cpu.h
 crc.cpp
@@ -269,6 +279,7 @@ rdrand.s
 rdrand.asm
 rdrand.cpp
 rdrand.h
+rdseed.asm
 rdtables.cpp
 regtest1.cpp
 regtest2.cpp
@@ -304,6 +315,12 @@ serpentp.h
 sha.cpp
 sha_simd.cpp
 sha.h
+sha1_armv4.h
+sha1_armv4.S
+sha256_armv4.h
+sha256_armv4.S
+sha512_armv4.h
+sha512_armv4.S
 sha3.cpp
 sha3.h
 shacal2.cpp
@@ -421,6 +438,8 @@ TestData/dlie2048.dat
 TestData/dsa1024.dat
 TestData/dsa1024b.dat
 TestData/dsa512.dat
+TestData/ecies_p160.dat
+TestData/ecies_t163.dat
 TestData/ed25519.dat
 TestData/ed25519_ver.dat
 TestData/ed25519v0.dat
@@ -542,6 +561,7 @@ TestVectors/shacal2.txt
 TestVectors/simeck.txt
 TestVectors/simon.txt
 TestVectors/siphash.txt
+TestVectors/skipjack.txt
 TestVectors/sm3.txt
 TestVectors/sm4.txt
 TestVectors/sosemanuk.txt
@@ -561,9 +581,10 @@ TestPrograms/test_arm_asimd.cxx
 TestPrograms/test_arm_crc.cxx
 TestPrograms/test_arm_neon.cxx
 TestPrograms/test_arm_pmull.cxx
-TestPrograms/test_arm_sha.cxx
+TestPrograms/test_arm_sha1.cxx
-TestPrograms/test_arm_sha3.cxx
+TestPrograms/test_arm_sha256.cxx
 TestPrograms/test_arm_sha512.cxx
+TestPrograms/test_arm_sha3.cxx
 TestPrograms/test_arm_sm3.cxx
 TestPrograms/test_arm_sm4.cxx
 TestPrograms/test_cxx.cxx
@@ -583,6 +604,8 @@ TestPrograms/test_x86_avx2.cxx
 TestPrograms/test_x86_avx512.cxx
 TestPrograms/test_x86_clmul.cxx
 TestPrograms/test_x86_cpuid.cxx
+TestPrograms/test_x86_rdrand.cxx
+TestPrograms/test_x86_rdseed.cxx
 TestPrograms/test_x86_sha.cxx
 TestPrograms/test_x86_sse2.cxx
 TestPrograms/test_x86_sse3.cxx

GNUmakefile

@@ -10,12 +10,16 @@ SHELL = /bin/sh
 
 # If needed
 TMPDIR ?= /tmp
-# Used for ARMv7 and NEON.
-FP_ABI ?= hard
 # Used for feature tests
 TOUT ?= a.out
 TOUT := $(strip $(TOUT))
 
+# Allow override for the cryptest.exe recipe. Change to
+# ./libcryptopp.so or ./libcryptopp.dylib to suit your
+# taste. https://github.com/weidai11/cryptopp/issues/866
+LINK_LIBRARY ?= libcryptopp.a
+LINK_LIBRARY_PATH ?= ./
+
 # Command and arguments
 AR ?= ar
 ARFLAGS ?= -cr # ar needs the dash on OpenBSD
@@ -25,6 +29,7 @@ CP ?= cp
 MV ?= mv
 RM ?= rm -f
 GREP ?= grep
+SED ?= sed
 CHMOD ?= chmod
 MKDIR ?= mkdir -p
 
@@ -35,11 +40,17 @@ LDCONF ?= /sbin/ldconfig -n
 ifneq ($(wildcard /usr/xpg4/bin/grep),)
   GREP := /usr/xpg4/bin/grep
 endif
+ifneq ($(wildcard /usr/xpg4/bin/sed),)
+  SED := /usr/xpg4/bin/sed
+endif
 
-# Attempt to determine target machine, fallback to "this" machine.
+# Yet another Clang hack. I think the LLVM devs are making the shit up
-#   The target machine is the one the package runs on. Most people
+# as they go. Also see https://github.com/weidai11/cryptopp/issues/831.
-#   call this the "target", but not Autotools.
+MACHINEX := $(shell $(CXX) $(CXXFLAGS) -dumpmachine 2>/dev/null)
-HOSTX := $(shell $(CXX) $(CXXFLAGS) -dumpmachine 2>/dev/null | cut -f 1 -d '-')
+ifeq ($(MACHINEX),armv8l-unknown-linux-gnueabihf)
+  MACHINEX := armv7l-unknown-linux-gnueabihf
+endif
+HOSTX := $(shell echo $(MACHINEX) | cut -f 1 -d '-')
 ifeq ($(HOSTX),)
   HOSTX := $(shell uname -m 2>/dev/null)
 endif
@@ -53,8 +64,6 @@ IS_SPARC64 := $(shell echo "$(HOSTX)" | $(GREP) -i -c -E 'sun|sparc64')
 IS_ARM32 := $(shell echo "$(HOSTX)" | $(GREP) -i -c -E 'arm|armhf|arm7l|eabihf')
 IS_ARMV8 := $(shell echo "$(HOSTX)" | $(GREP) -i -c -E 'aarch32|aarch64|arm64|armv8')
 
-IS_NEON := $(shell $(CXX) $(CXXFLAGS) -dumpmachine 2>/dev/null | $(GREP) -i -c -E 'armv7|armhf|arm7l|eabihf|armv8|aarch32|aarch64')
-
 # Attempt to determine platform
 SYSTEMX := $(shell $(CXX) $(CXXFLAGS) -dumpmachine 2>/dev/null)
 ifeq ($(SYSTEMX),)
@@ -116,9 +125,10 @@ else ifeq ($(findstring distclean,$(MAKECMDGOALS)),trim)
   DETECT_FEATURES := 0
 endif
 
-# Strip out -Wall, -Wextra and friends for feature testing
+# Strip out -Wall, -Wextra and friends for feature testing. FORTIFY_SOURCE is removed
+# because it requires -O1 or higher, but we use -O0 to tame the optimizer.
 ifeq ($(DETECT_FEATURES),1)
-  TCXXFLAGS := $(filter-out -Wall -Wextra -Werror% -Wunused -Wconversion -Wp%, $(CXXFLAGS))
+  TCXXFLAGS := $(filter-out -D_FORTIFY_SOURCE=% -Wall -Wextra -Werror% -Wunused -Wconversion -Wp%, $(CXXFLAGS))
   ifneq ($(strip $(TCXXFLAGS)),)
     $(info Using testing flags: $(TCXXFLAGS))
   endif
@@ -161,6 +171,11 @@ else
   ZOPT = -O0
 endif
 
+# Fix CXX on Cygwin 1.1.4
+ifeq ($(CXX),gcc)
+CXX := g++
+endif
+
 # On ARM we may compile aes_armv4.S though the CC compiler
 ifeq ($(GCC_COMPILER),1)
   CC=gcc
@@ -187,19 +202,11 @@ ifeq ($(INCLUDEDIR),)
 INCLUDEDIR := $(PREFIX)/include
 endif
 
-# Fix CXX on Cygwin 1.1.4
-ifeq ($(CXX),gcc)
-CXX := g++
-endif
-
 # We honor ARFLAGS, but the "v" option used by default causes a noisy make
 ifeq ($(ARFLAGS),rv)
 ARFLAGS = r
 endif
 
-# Clang integrated assembler will be used with -Wa,-q
-CLANG_INTEGRATED_ASSEMBLER ?= 0
-
 # Original MinGW targets Win2k by default, but lacks proper Win2k support
 # if target Windows version is not specified, use Windows XP instead
 ifeq ($(IS_MINGW),1)
@@ -273,9 +280,7 @@ ifeq ($(DETECT_FEATURES),1)
     KECCAK_FLAG = $(SSSE3_FLAG)
     LEA_FLAG = $(SSSE3_FLAG)
     SIMECK_FLAG = $(SSSE3_FLAG)
-    SIMON64_FLAG = $(SSSE3_FLAG)
     SIMON128_FLAG = $(SSSE3_FLAG)
-    SPECK64_FLAG = $(SSSE3_FLAG)
     SPECK128_FLAG = $(SSSE3_FLAG)
     SUN_LDFLAGS += $(SSSE3_FLAG)
   else
@@ -439,37 +444,37 @@ endif
 #####            ARM A-32, Aach64 and NEON            #####
 ###########################################################
 
-ifneq ($(IS_ARM32)$(IS_ARMV8)$(IS_NEON),000)
+ifneq ($(IS_ARM32)$(IS_ARMV8),00)
 ifeq ($(DETECT_FEATURES),1)
 
-ifeq ($(IS_ARM32)$(IS_NEON),11)
+ifneq ($(IS_ARM32),0)
 
   TPROG = TestPrograms/test_arm_neon.cxx
-  TOPT = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+  TOPT = -march=armv7-a -mfpu=neon
   HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
   ifeq ($(strip $(HAVE_OPT)),0)
-    NEON_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    NEON_FLAG = -march=armv7-a -mfpu=neon
-    ARIA_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    ARIA_FLAG = -march=armv7-a -mfpu=neon
-    AES_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    AES_FLAG = -march=armv7-a -mfpu=neon
-    CRC_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    CRC_FLAG = -march=armv7-a -mfpu=neon
-    GCM_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    GCM_FLAG = -march=armv7-a -mfpu=neon
-    BLAKE2B_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    BLAKE2B_FLAG = -march=armv7-a -mfpu=neon
-    BLAKE2S_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    BLAKE2S_FLAG = -march=armv7-a -mfpu=neon
-    CHACHA_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    CHACHA_FLAG = -march=armv7-a -mfpu=neon
-    CHAM_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    CHAM_FLAG = -march=armv7-a -mfpu=neon
-    LEA_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    LEA_FLAG = -march=armv7-a -mfpu=neon
-    SHA_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    SHA_FLAG = -march=armv7-a -mfpu=neon
-    SIMECK_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    SIMECK_FLAG = -march=armv7-a -mfpu=neon
-    SIMON64_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    SIMON64_FLAG = -march=armv7-a -mfpu=neon
-    SIMON128_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    SIMON128_FLAG = -march=armv7-a -mfpu=neon
-    SPECK64_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    SPECK64_FLAG = -march=armv7-a -mfpu=neon
-    SPECK128_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    SPECK128_FLAG = -march=armv7-a -mfpu=neon
-    SM4_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    SM4_FLAG = -march=armv7-a -mfpu=neon
   else
     CXXFLAGS += -DCRYPTOPP_DISABLE_ASM
   endif
 
-# IS_NEON
+# IS_ARM32
 endif
 
 ifeq ($(IS_ARMV8),1)
@@ -485,7 +490,7 @@ ifeq ($(IS_ARMV8),1)
 
   TPROG = TestPrograms/test_arm_asimd.cxx
   TOPT = -march=armv8-a
-  HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+  HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
   ifeq ($(strip $(HAVE_OPT)),0)
     ASIMD_FLAG = -march=armv8-a
     ARIA_FLAG = -march=armv8-a
@@ -508,7 +513,7 @@ ifeq ($(IS_ARMV8),1)
   ifneq ($(ASIMD_FLAG),)
     TPROG = TestPrograms/test_arm_crc.cxx
     TOPT = -march=armv8-a+crc
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifeq ($(strip $(HAVE_OPT)),0)
       CRC_FLAG = -march=armv8-a+crc
     else
@@ -517,7 +522,7 @@ ifeq ($(IS_ARMV8),1)
 
     TPROG = TestPrograms/test_arm_aes.cxx
     TOPT = -march=armv8-a+crypto
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifeq ($(strip $(HAVE_OPT)),0)
       AES_FLAG = -march=armv8-a+crypto
     else
@@ -526,7 +531,7 @@ ifeq ($(IS_ARMV8),1)
 
     TPROG = TestPrograms/test_arm_pmull.cxx
     TOPT = -march=armv8-a+crypto
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifeq ($(strip $(HAVE_OPT)),0)
       GCM_FLAG = -march=armv8-a+crypto
       GF2N_FLAG = -march=armv8-a+crypto
@@ -534,9 +539,9 @@ ifeq ($(IS_ARMV8),1)
       CXXFLAGS += -DCRYPTOPP_ARM_PMULL_AVAILABLE=0
     endif
 
-    TPROG = TestPrograms/test_arm_sha.cxx
+    TPROG = TestPrograms/test_arm_sha1.cxx
     TOPT = -march=armv8-a+crypto
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifeq ($(strip $(HAVE_OPT)),0)
       SHA_FLAG = -march=armv8-a+crypto
     else
@@ -545,7 +550,7 @@ ifeq ($(IS_ARMV8),1)
 
     TPROG = TestPrograms/test_arm_sm3.cxx
     TOPT = -march=armv8.4-a+crypto
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifeq ($(strip $(HAVE_OPT)),0)
       SM3_FLAG = -march=armv8.4-a+crypto
       SM4_FLAG = -march=armv8.4-a+crypto
@@ -553,7 +558,7 @@ ifeq ($(IS_ARMV8),1)
 
     TPROG = TestPrograms/test_arm_sha3.cxx
     TOPT = -march=armv8.4-a+crypto
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifeq ($(strip $(HAVE_OPT)),0)
       SHA3_FLAG = -march=armv8.4-a+crypto
     endif
@@ -567,7 +572,7 @@ endif
 # DETECT_FEATURES
 endif
 
-# IS_ARM32, IS_ARMV8, IS_NEON
+# IS_ARM32, IS_ARMV8
 endif
 
 ###########################################################
@@ -622,13 +627,21 @@ ifeq ($(DETECT_FEATURES),1)
   TOPT = $(POWER8_FLAG)
   HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
   ifeq ($(strip $(HAVE_OPT)),0)
+    AES_FLAG = $(POWER8_FLAG)
+    ARIA_FLAG = $(POWER8_FLAG)
     BLAKE2B_FLAG = $(POWER8_FLAG)
+    BLAKE2S_FLAG = $(POWER8_FLAG)
+    CHACHA_FLAG = $(POWER8_FLAG)
+    CHAM_FLAG = $(POWER8_FLAG)
     CRC_FLAG = $(POWER8_FLAG)
     GCM_FLAG = $(POWER8_FLAG)
     GF2N_FLAG = $(POWER8_FLAG)
-    AES_FLAG = $(POWER8_FLAG)
+    LEA_FLAG = $(POWER8_FLAG)
     SHA_FLAG = $(POWER8_FLAG)
     SHACAL2_FLAG = $(POWER8_FLAG)
+    SIMECK_FLAG = $(POWER8_FLAG)
+    SIMON64_FLAG = $(POWER8_FLAG)
+    SPECK64_FLAG = $(POWER8_FLAG)
     SIMON128_FLAG = $(POWER8_FLAG)
     SPECK128_FLAG = $(POWER8_FLAG)
   else
@@ -641,16 +654,7 @@ ifeq ($(DETECT_FEATURES),1)
   TPROG = TestPrograms/test_ppc_power7.cxx
   TOPT = $(POWER7_FLAG)
   HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
-  ifeq ($(strip $(HAVE_OPT)),0)
+  ifneq ($(strip $(HAVE_OPT)),0)
-    ARIA_FLAG = $(POWER7_FLAG)
-    BLAKE2S_FLAG = $(POWER7_FLAG)
-    CHACHA_FLAG = $(POWER7_FLAG)
-    CHAM_FLAG = $(POWER7_FLAG)
-    LEA_FLAG = $(POWER7_FLAG)
-    SIMECK_FLAG = $(POWER7_FLAG)
-    SIMON64_FLAG = $(POWER7_FLAG)
-    SPECK64_FLAG = $(POWER7_FLAG)
-  else
     POWER7_FLAG =
   endif
 
@@ -693,18 +697,12 @@ ifeq ($(DETECT_FEATURES),1)
   #####################################################################
   # Fixups for algorithms that can drop to a lower ISA, if needed
 
-  # Drop to Power7 if Power8 is not available.
+  # Drop to Power4 if Power8 not available
   ifeq ($(POWER8_FLAG),)
-    ifneq ($(POWER7_FLAG),)
-      GCM_FLAG = $(POWER7_FLAG)
-    endif
-  endif
-
-  # Drop to Power4 if Power7 not available
-  ifeq ($(POWER7_FLAG),)
     ifneq ($(ALTIVEC_FLAG),)
       BLAKE2S_FLAG = $(ALTIVEC_FLAG)
       CHACHA_FLAG = $(ALTIVEC_FLAG)
+      GCM_FLAG = $(ALTIVEC_FLAG)
       SIMON64_FLAG = $(ALTIVEC_FLAG)
       SPECK64_FLAG = $(ALTIVEC_FLAG)
     endif
@@ -869,6 +867,15 @@ ifeq ($(findstring native,$(MAKECMDGOALS)),native)
     NATIVE_OPT = -march=native
   endif # NATIVE_OPT
 
+  # And tune
+  ifeq ($(NATIVE_OPT),)
+    TOPT = -mtune=native
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    ifeq ($(strip $(HAVE_OPT)),0)
+      NATIVE_OPT = -mtune=native
+    endif # NATIVE_OPT
+  endif
+
   # Try SunCC next
   ifeq ($(NATIVE_OPT),)
     TOPT = -native
@@ -956,9 +963,9 @@ ifneq ($(filter valgrind,$(MAKECMDGOALS)),)
   CXXFLAGS := $(CXXFLAGS:-g%=-g3)
   CXXFLAGS := $(CXXFLAGS:-O%=-O1)
   CXXFLAGS := $(CXXFLAGS:-xO%=-xO1)
-  ifeq ($(findstring -DCRYPTOPP_VALGRIND,$(CXXFLAGS)),)
+  ifeq ($(findstring -DCRYPTOPP_COVERAGE,$(CXXFLAGS)),)
-    CXXFLAGS += -DCRYPTOPP_VALGRIND
+    CXXFLAGS += -DCRYPTOPP_COVERAGE
-  endif # -DCRYPTOPP_VALGRIND
+  endif # -DCRYPTOPP_COVERAGE
 endif # Valgrind
 
 # Debug testing on GNU systems. Triggered by -DDEBUG.
@@ -1004,7 +1011,7 @@ ifeq ($(findstring lean,$(MAKECMDGOALS)),lean)
 endif # Dead code stripping
 
 # For Shared Objects, Diff, Dist/Zip rules
-LIB_VER := $(shell $(GREP) "define CRYPTOPP_VERSION" config.h | cut -d" " -f 3)
+LIB_VER := $(shell $(GREP) "define CRYPTOPP_VERSION" config_ver.h | cut -d" " -f 3)
 LIB_MAJOR := $(shell echo $(LIB_VER) | cut -c 1)
 LIB_MINOR := $(shell echo $(LIB_VER) | cut -c 2)
 LIB_PATCH := $(shell echo $(LIB_VER) | cut -c 3)
@@ -1061,12 +1068,18 @@ ifneq ($(IS_MINGW),0)
 INCL += resource.h
 endif
 
-# Cryptogams AES for ARMv4 and above. We couple to ARMv7.
+# Cryptogams source files. We couple to ARMv7.
-# Avoid iOS. It cannot consume the assembly.
+# Limit to Linux. The source files target the GNU assembler.
-ifeq ($(IS_ARM32),1)
+# Also see https://www.cryptopp.com/wiki/Cryptogams.
-  CRYPTOGAMS_AES_FLAG = -march=armv7-a
+ifeq ($(IS_ARM32)$(IS_LINUX),11)
-  CRYPTOGAMS_AES_FLAG += -Wa,--noexecstack
+  ifeq ($(CLANG_COMPILER),1)
-  SRCS += aes_armv4.S
+    CRYPTOGAMS_ARMV7_FLAG = -march=armv7-a -Wa,--noexecstack
+    CRYPTOGAMS_ARMV7_THUMB_FLAG = -march=armv7-a -mthumb -Wa,--noexecstack
+  else
+    CRYPTOGAMS_ARMV7_FLAG = -march=armv7-a -Wa,--noexecstack
+    CRYPTOGAMS_ARMV7_THUMB_FLAG = -march=armv7-a -Wa,--noexecstack
+  endif
+  SRCS += aes_armv4.S sha1_armv4.S sha256_armv4.S sha512_armv4.S
 endif
 
 # List cryptlib.cpp first, then cpu.cpp, then integer.cpp to tame C++ static initialization problems.
@@ -1281,10 +1294,14 @@ remove uninstall:
 	-$(RM) -r $(DESTDIR)$(INCLUDEDIR)/cryptopp
 	-$(RM) $(DESTDIR)$(LIBDIR)/libcryptopp.a
 	-$(RM) $(DESTDIR)$(BINDIR)/cryptest.exe
-	@-$(RM) $(DESTDIR)$(LIBDIR)/libcryptopp.dylib
+ifneq ($(wildcard $(DESTDIR)$(LIBDIR)/libcryptopp.dylib),)
+	-$(RM) $(DESTDIR)$(LIBDIR)/libcryptopp.dylib
+endif
+ifneq ($(wildcard $(DESTDIR)$(LIBDIR)/libcryptopp.so),)
+	-$(RM) $(DESTDIR)$(LIBDIR)/libcryptopp.so
+endif
 	@-$(RM) $(DESTDIR)$(LIBDIR)/libcryptopp.so$(SOLIB_VERSION_SUFFIX)
 	@-$(RM) $(DESTDIR)$(LIBDIR)/libcryptopp.so$(SOLIB_COMPAT_SUFFIX)
-	@-$(RM) $(DESTDIR)$(LIBDIR)/libcryptopp.so
 	@-$(RM) $(DESTDIR)$(LIBDIR)/pkgconfig/libcryptopp.pc
 	@-$(RM) -r $(DESTDIR)$(DATADIR)/cryptopp
 
@@ -1313,8 +1330,8 @@ endif
 libcryptopp.dylib: $(LIBOBJS)
 	$(CXX) -dynamiclib -o $@ $(strip $(CXXFLAGS)) -install_name "$@" -current_version "$(LIB_MAJOR).$(LIB_MINOR).$(LIB_PATCH)" -compatibility_version "$(LIB_MAJOR).$(LIB_MINOR)" -headerpad_max_install_names $(LDFLAGS) $(LIBOBJS)
 
-cryptest.exe:libcryptopp.a $(TESTOBJS)
+cryptest.exe: $(LINK_LIBRARY) $(TESTOBJS)
-	$(CXX) -o $@ $(strip $(CXXFLAGS)) $(TESTOBJS) ./libcryptopp.a $(LDFLAGS) $(LDLIBS)
+	$(CXX) -o $@ $(strip $(CXXFLAGS)) $(TESTOBJS) $(LINK_LIBRARY_PATH)$(LINK_LIBRARY) $(LDFLAGS) $(LDLIBS)
 
 # Makes it faster to test changes
 nolib: $(OBJS)
@@ -1349,7 +1366,7 @@ libcryptopp.pc:
 	@echo '' >> libcryptopp.pc
 	@echo 'Name: Crypto++' >> libcryptopp.pc
 	@echo 'Description: Crypto++ cryptographic library' >> libcryptopp.pc
-	@echo 'Version: 8.1' >> libcryptopp.pc
+	@echo 'Version: 8.3' >> libcryptopp.pc
 	@echo 'URL: https://cryptopp.com/' >> libcryptopp.pc
 	@echo '' >> libcryptopp.pc
 	@echo 'Cflags: -I$${includedir}' >> libcryptopp.pc
@@ -1366,14 +1383,14 @@ endif
 .PHONY: trim
 trim:
 ifneq ($(IS_DARWIN),0)
-	sed -i '' -e's/[[:space:]]*$$//' *.supp *.txt *.sh .*.yml *.h *.cpp *.asm *.s *.S
+	$(SED) -i '' -e's/[[:space:]]*$$//' *.supp *.txt *.sh .*.yml *.h *.cpp *.asm *.s *.S
-	sed -i '' -e's/[[:space:]]*$$//' *.sln *.vcxproj *.filters GNUmakefile GNUmakefile-cross
+	$(SED) -i '' -e's/[[:space:]]*$$//' *.sln *.vcxproj *.filters GNUmakefile GNUmakefile-cross
-	sed -i '' -e's/[[:space:]]*$$//' TestData/*.dat TestVectors/*.txt TestPrograms/*.cxx TestScripts/*.*
+	$(SED) -i '' -e's/[[:space:]]*$$//' TestData/*.dat TestVectors/*.txt TestPrograms/*.cxx TestScripts/*.*
 	make convert
 else
-	sed -i -e's/[[:space:]]*$$//' *.supp *.txt *.sh .*.yml *.h *.cpp *.asm *.s *.S
+	$(SED) -i -e's/[[:space:]]*$$//' *.supp *.txt *.sh .*.yml *.h *.cpp *.asm *.s *.S
-	sed -i -e's/[[:space:]]*$$//' *.sln *.vcxproj *.filters GNUmakefile GNUmakefile-cross
+	$(SED) -i -e's/[[:space:]]*$$//' *.sln *.vcxproj *.filters GNUmakefile GNUmakefile-cross
-	sed -i -e's/[[:space:]]*$$//' TestData/*.dat TestVectors/*.txt TestPrograms/*.cxx TestScripts/*.*
+	$(SED) -i -e's/[[:space:]]*$$//' TestData/*.dat TestVectors/*.txt TestPrograms/*.cxx TestScripts/*.*
 	make convert
 endif
 
@@ -1428,9 +1445,9 @@ ifeq ($(wildcard GNUmakefile.deps),GNUmakefile.deps)
 -include GNUmakefile.deps
 endif # Dependencies
 
-# Cryptogams ARM asm implementation.
+# Cryptogams ARM asm implementation. AES needs -mthumb for Clang
 aes_armv4.o : aes_armv4.S
-	$(CC) $(strip $(CXXFLAGS) $(CRYPTOGAMS_AES_FLAG) -mfloat-abi=$(FP_ABI) -c) $<
+	$(CXX) $(strip $(CXXFLAGS) $(CRYPTOGAMS_ARMV7_THUMB_FLAG) -c) $<
 
 # SSSE3 or NEON available
 aria_simd.o : aria_simd.cpp
@@ -1512,6 +1529,18 @@ rijndael_simd.o : rijndael_simd.cpp
 sha_simd.o : sha_simd.cpp
 	$(CXX) $(strip $(CXXFLAGS) $(SHA_FLAG) -c) $<
 
+# Cryptogams SHA1 asm implementation.
+sha1_armv4.o : sha1_armv4.S
+	$(CXX) $(strip $(CXXFLAGS) $(CRYPTOGAMS_ARMV7_FLAG) -c) $<
+
+# Cryptogams SHA256 asm implementation.
+sha256_armv4.o : sha256_armv4.S
+	$(CXX) $(strip $(CXXFLAGS) $(CRYPTOGAMS_ARMV7_FLAG) -c) $<
+
+# Cryptogams SHA512 asm implementation.
+sha512_armv4.o : sha512_armv4.S
+	$(CXX) $(strip $(CXXFLAGS) $(CRYPTOGAMS_ARMV7_FLAG) -c) $<
+
 sha3_simd.o : sha3_simd.cpp
 	$(CXX) $(strip $(CXXFLAGS) $(SHA3_FLAG) -c) $<
 

GNUmakefile-cross

@@ -11,6 +11,12 @@ FP_ABI ?= hard
 TOUT ?= a.out
 TOUT := $(strip $(TOUT))
 
+# Allow override for the cryptest.exe recipe. Change to
+# ./libcryptopp.so or ./libcryptopp.dylib to suit your
+# taste. https://github.com/weidai11/cryptopp/issues/866
+LINK_LIBRARY ?= ./libcryptopp.a
+LINK_LIBRARY_PATH ?= ./
+
 # Default CXXFLAGS if none were provided
 CXXFLAGS ?= -DNDEBUG -g2 -O3 -fPIC -pipe
 
@@ -21,7 +27,8 @@ CP ?= cp
 MV ?= mv
 CHMOD ?= chmod
 MKDIR ?= mkdir -p
-EGREP ?= egrep
+GREP ?= grep
+SED ?= sed
 
 LN ?= ln -sf
 LDCONF ?= /sbin/ldconfig -n
@@ -31,6 +38,19 @@ IS_ANDROID ?= 0
 IS_ARM_EMBEDDED ?= 0
 IS_NEON ?= 0
 
+# Yet another Clang hack. I think the LLVM devs are making the shit up
+# as they go. Also see https://github.com/weidai11/cryptopp/issues/831.
+MACHINEX := $(shell $(CXX) $(CXXFLAGS) -dumpmachine 2>/dev/null)
+ifeq ($(MACHINEX),armv8l-unknown-linux-gnueabihf)
+  MACHINEX := armv7l-unknown-linux-gnueabihf
+endif
+HOSTX := $(shell echo $(MACHINEX) | cut -f 1 -d '-')
+ifeq ($(HOSTX),)
+  HOSTX := $(shell uname -m 2>/dev/null)
+endif
+
+IS_LINUX := $(shell echo $(MACHINEX) | $(GREP) -i -c "Linux")
+
 # Can be used by Android and Embeeded cross-compiles. Disable by default because
 #   Android and embedded users typically don't run this configuration.
 HAS_SOLIB_VERSION ?= 0
@@ -136,8 +156,8 @@ endif
 
 # Wait until CXXFLAGS have been set by setenv scripts.
 
-GCC_COMPILER := $(shell $(CXX) --version 2>/dev/null | $(EGREP) -v -E 'llvm|clang' | $(EGREP) -i -c -E '(gcc|g\+\+)')
+GCC_COMPILER := $(shell $(CXX) --version 2>/dev/null | $(GREP) -v -E 'llvm|clang' | $(GREP) -i -c -E '(gcc|g\+\+)')
-CLANG_COMPILER := $(shell $(CXX) --version 2>/dev/null | $(EGREP) -i -c -E 'llvm|clang')
+CLANG_COMPILER := $(shell $(CXX) --version 2>/dev/null | $(GREP) -i -c -E 'llvm|clang')
 
 HOSTX := $(shell $(CXX) $(CXXFLAGS) -dumpmachine 2>/dev/null | cut -f 1 -d '-')
 ifeq ($(HOSTX),)
@@ -149,15 +169,15 @@ endif
 # -arch arm64 yields x86_64 instead of aarch64 or arm64.
 
 ifeq ($(CLANG_COMPILER),1)
-  IS_X86 := $(shell echo $(CXXFLAGS) | $(EGREP) -v 64 | $(EGREP) -i -c -E 'i.86')
+  IS_X86 := $(shell echo $(CXXFLAGS) | $(GREP) -v 64 | $(GREP) -i -c -E 'i.86')
-  IS_X64 := $(shell echo $(CXXFLAGS) | $(EGREP) -i -c -E 'x86_64|amd64')
+  IS_X64 := $(shell echo $(CXXFLAGS) | $(GREP) -i -c -E 'x86_64|amd64')
-  IS_ARM32 := $(shell echo $(CXXFLAGS) | $(EGREP) -v 64 | $(EGREP) -i -c -E 'arm|armhf|arm7l|eabihf')
+  IS_ARM32 := $(shell echo $(CXXFLAGS) | $(GREP) -v 64 | $(GREP) -i -c -E 'arm|armhf|arm7l|eabihf')
-  IS_ARMV8 := $(shell echo $(CXXFLAGS) | $(EGREP) -i -c -E 'aarch32|aarch64|arm64|armv8')
+  IS_ARMV8 := $(shell echo $(CXXFLAGS) | $(GREP) -i -c -E 'aarch32|aarch64|arm64|armv8')
 else
-  IS_X86 := $(shell echo $(HOSTX) | $(EGREP) -v 64 | $(EGREP) -i -c -E 'i.86')
+  IS_X86 := $(shell echo $(HOSTX) | $(GREP) -v 64 | $(GREP) -i -c -E 'i.86')
-  IS_X64 := $(shell echo $(HOSTX) | $(EGREP) -i -c -E 'x86_64|amd64')
+  IS_X64 := $(shell echo $(HOSTX) | $(GREP) -i -c -E 'x86_64|amd64')
-  IS_ARM32 := $(shell echo $(HOSTX) | $(EGREP) -v 64 | $(EGREP) -i -c -E 'arm|armhf|arm7l|eabihf')
+  IS_ARM32 := $(shell echo $(HOSTX) | $(GREP) -v 64 | $(GREP) -i -c -E 'arm|armhf|arm7l|eabihf')
-  IS_ARMV8 := $(shell echo $(HOSTX) | $(EGREP) -i -c -E 'aarch32|aarch64|arm64|armv8')
+  IS_ARMV8 := $(shell echo $(HOSTX) | $(GREP) -i -c -E 'aarch32|aarch64|arm64|armv8')
 endif
 
 $(info Here's what we found... IS_X86: $(IS_X86), IS_X64: $(IS_X64), IS_ARM32: $(IS_ARM32), IS_ARMV8: $(IS_ARMV8))
@@ -178,9 +198,10 @@ else ifeq ($(findstring distclean,$(MAKECMDGOALS)),trim)
   DETECT_FEATURES := 0
 endif
 
-# Strip out -Wall, -Wextra and friends for feature testing
+# Strip out -Wall, -Wextra and friends for feature testing. FORTIFY_SOURCE is removed
+# because it requires -O1 or higher, but we use -O0 to tame the optimizer.
 ifeq ($(DETECT_FEATURES),1)
-  TCXXFLAGS := $(filter-out -Wall -Wextra -Werror% -Wunused -Wconversion -Wp%, $(CXXFLAGS))
+  TCXXFLAGS := $(filter-out -D_FORTIFY_SOURCE=% -Wall -Wextra -Werror% -Wunused -Wconversion -Wp%, $(CXXFLAGS))
   ifneq ($(strip $(TCXXFLAGS)),)
     $(info Using testing flags: $(TCXXFLAGS))
   endif
@@ -227,9 +248,7 @@ ifeq ($(DETECT_FEATURES),1)
     CHAM_FLAG = $(SSSE3_FLAG)
     LEA_FLAG = $(SSSE3_FLAG)
     SIMECK_FLAG = $(SSSE3_FLAG)
-    SIMON64_FLAG = $(SSSE3_FLAG)
     SIMON128_FLAG = $(SSSE3_FLAG)
-    SPECK64_FLAG = $(SSSE3_FLAG)
     SPECK128_FLAG = $(SSSE3_FLAG)
   else
     SSSE3_FLAG =
@@ -384,10 +403,10 @@ ifeq ($(IS_ARM32),1)
     SM4_FLAG = $(NEON_FLAG)
   else
     NEON_FLAG =
-    CXXFLAGS += -DCRYPTOPP_DISABLE_NEON
+    CXXFLAGS += -DCRYPTOPP_DISABLE_ASM
   endif
 
-# IS_NEON
+# IS_ARM32
 endif
 
 ifeq ($(IS_ARMV8),1)
@@ -417,7 +436,7 @@ ifeq ($(IS_ARMV8),1)
 
   TPROG = TestPrograms/test_arm_asimd.cxx
   TOPT = $(ASIMD_FLAG)
-  HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+  HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
   ifeq ($(strip $(HAVE_OPT)),0)
     ARIA_FLAG = $(ASIMD_FLAG)
     BLAKE2B_FLAG = $(ASIMD_FLAG)
@@ -439,7 +458,7 @@ ifeq ($(IS_ARMV8),1)
 
     TPROG = TestPrograms/test_arm_crc.cxx
     TOPT = $(CRC_FLAG)
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifneq ($(strip $(HAVE_OPT)),0)
       CRC_FLAG =
       CXXFLAGS += -DCRYPTOPP_ARM_CRC32_AVAILABLE=0
@@ -447,7 +466,7 @@ ifeq ($(IS_ARMV8),1)
 
     TPROG = TestPrograms/test_arm_aes.cxx
     TOPT = $(AES_FLAG)
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifneq ($(strip $(HAVE_OPT)),0)
       AES_FLAG =
       CXXFLAGS += -DCRYPTOPP_ARM_AES_AVAILABLE=0
@@ -455,15 +474,15 @@ ifeq ($(IS_ARMV8),1)
 
     TPROG = TestPrograms/test_arm_pmull.cxx
     TOPT = $(PMULL_FLAG)
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifneq ($(strip $(HAVE_OPT)),0)
       PMULL_FLAG =
       CXXFLAGS += -DCRYPTOPP_ARM_PMULL_AVAILABLE=0
     endif
 
-    TPROG = TestPrograms/test_arm_sha.cxx
+    TPROG = TestPrograms/test_arm_sha1.cxx
     TOPT = $(SHA_FLAG)
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifneq ($(strip $(HAVE_OPT)),0)
       SHA_FLAG =
       CXXFLAGS += -DCRYPTOPP_ARM_SHA_AVAILABLE=0
@@ -471,7 +490,7 @@ ifeq ($(IS_ARMV8),1)
 
     TPROG = TestPrograms/test_arm_sm3.cxx
     TOPT = -march=armv8.4-a+crypto
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifeq ($(strip $(HAVE_OPT)),0)
       SM3_FLAG = -march=armv8.4-a+crypto
       SM4_FLAG = -march=armv8.4-a+crypto
@@ -479,7 +498,7 @@ ifeq ($(IS_ARMV8),1)
 
     TPROG = TestPrograms/test_arm_sha3.cxx
     TOPT = -march=armv8.4-a+crypto
-    HAVE_OPT = $(shell $(CXX) $(CXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
+    HAVE_OPT = $(shell $(CXX) $(TCXXFLAGS) $(ACLE_FLAG) $(ZOPT) $(TOPT) $(TPROG) -o $(TOUT) 2>&1 | tr ' ' '\n' | wc -l)
     ifeq ($(strip $(HAVE_OPT)),0)
       SHA3_FLAG = -march=armv8.4-a+crypto
       SHA512_FLAG = -march=armv8.4-a+crypto
@@ -531,7 +550,7 @@ endif # Asan
 # LD gold linker testing. Triggered by 'LD=ld.gold'.
 ifeq ($(findstring ld.gold,$(LD)),ld.gold)
   ifeq ($(findstring -fuse-ld=gold,$(CXXFLAGS)),)
-    ELF_FORMAT := $(shell file `which ld.gold` 2>&1 | cut -d":" -f 2 | $(EGREP) -i -c "elf")
+    ELF_FORMAT := $(shell file `which ld.gold` 2>&1 | cut -d":" -f 2 | $(GREP) -i -c "elf")
     ifneq ($(ELF_FORMAT),0)
       LDFLAGS += -fuse-ld=gold
     endif # ELF/ELF64
@@ -544,15 +563,15 @@ ifneq ($(filter valgrind,$(MAKECMDGOALS)),)
   CXXFLAGS := $(CXXFLAGS:-g%=-g3)
   CXXFLAGS := $(CXXFLAGS:-O%=-O1)
   CXXFLAGS := $(CXXFLAGS:-xO%=-xO1)
-  ifeq ($(findstring -DCRYPTOPP_VALGRIND,$(CXXFLAGS)),)
+  ifeq ($(findstring -DCRYPTOPP_COVERAGE,$(CXXFLAGS)),)
-    CXXFLAGS += -DCRYPTOPP_VALGRIND
+    CXXFLAGS += -DCRYPTOPP_COVERAGE
-  endif # -DCRYPTOPP_VALGRIND
+  endif # -DCRYPTOPP_COVERAGE
 endif # Valgrind
 
 # Debug testing on GNU systems. Triggered by -DDEBUG.
 #   Newlib test due to http://sourceware.org/bugzilla/show_bug.cgi?id=20268
 ifneq ($(filter -DDEBUG -DDEBUG=1,$(CXXFLAGS)),)
-  USING_GLIBCXX := $(shell $(CXX) $(CXXFLAGS) -E pch.cpp 2>&1 | $(EGREP) -i -c "__GLIBCXX__")
+  USING_GLIBCXX := $(shell $(CXX) $(CXXFLAGS) -E pch.cpp 2>&1 | $(GREP) -i -c "__GLIBCXX__")
   ifneq ($(USING_GLIBCXX),0)
     ifeq ($(HAS_NEWLIB),0)
       ifeq ($(findstring -D_GLIBCXX_DEBUG,$(CXXFLAGS)),)
@@ -590,14 +609,18 @@ SRCS := cryptlib.cpp cpu.cpp integer.cpp $(filter-out cryptlib.cpp cpu.cpp integ
 # For Makefile.am; resource.h is Windows
 INCL := $(filter-out resource.h,$(sort $(wildcard *.h)))
 
-# Cryptogams AES for ARMv4 and above. We couple to ARMv7.
+# Cryptogams source files. We couple to ARMv7.
-# Avoid iOS. It cannot consume the assembly.
+# Limit to Linux. The source files target the GNU assembler.
-ifeq ($(IS_ARM32),1)
+# Also see https://www.cryptopp.com/wiki/Cryptogams.
-  ifneq ($(IS_IOS),1)
+ifeq ($(IS_ARM32)$(IS_LINUX),11)
-    CRYPTOGAMS_AES_FLAG = -march=armv7-a
+  ifeq ($(CLANG_COMPILER),1)
-    CRYPTOGAMS_AES_FLAG += -Wa,--noexecstack
+    CRYPTOGAMS_ARMV7_FLAG = -march=armv7-a -Wa,--noexecstack
-    SRCS += aes_armv4.S
+    CRYPTOGAMS_ARMV7_THUMB_FLAG = -march=armv7-a -mthumb -Wa,--noexecstack
+  else
+    CRYPTOGAMS_ARMV7_FLAG = -march=armv7-a -Wa,--noexecstack
+    CRYPTOGAMS_ARMV7_THUMB_FLAG = -march=armv7-a -Wa,--noexecstack
   endif
+  SRCS += aes_armv4.S sha1_armv4.S sha256_armv4.S sha512_armv4.S
 endif
 
 # List cryptlib.cpp first, then cpu.cpp, then integer.cpp to tame C++ static initialization problems.
@@ -613,7 +636,7 @@ TESTOBJS := $(TESTSRCS:.cpp=.o)
 LIBOBJS := $(filter-out $(TESTOBJS),$(OBJS))
 
 # For Shared Objects, Diff, Dist/Zip rules
-LIB_VER := $(shell $(EGREP) "define CRYPTOPP_VERSION" config.h | cut -d" " -f 3)
+LIB_VER := $(shell $(GREP) "define CRYPTOPP_VERSION" config_ver.h | cut -d" " -f 3)
 LIB_MAJOR := $(shell echo $(LIB_VER) | cut -c 1)
 LIB_MINOR := $(shell echo $(LIB_VER) | cut -c 2)
 LIB_PATCH := $(shell echo $(LIB_VER) | cut -c 3)
@@ -770,8 +793,8 @@ endif
 libcryptopp.dylib: $(LIBOBJS) $(AOSP_CPU_OBJ)
 	$(CXX) -dynamiclib -o $@ $(strip $(CXXFLAGS)) -install_name "$@" -current_version "$(LIB_MAJOR).$(LIB_MINOR).$(LIB_PATCH)" -compatibility_version "$(LIB_MAJOR).$(LIB_MINOR)" -headerpad_max_install_names $(LDFLAGS) $(LIBOBJS) $(AOSP_CPU_OBJ)
 
-cryptest.exe: libcryptopp.a $(TESTOBJS)
+cryptest.exe: $(LINK_LIBRARY) $(TESTOBJS)
-	$(CXX) -o $@ $(strip $(CXXFLAGS)) $(TESTOBJS) ./libcryptopp.a $(LDFLAGS) $(LDLIBS)
+	$(CXX) -o $@ $(strip $(CXXFLAGS)) $(TESTOBJS) $(LINK_LIBRARY_PATH)$(LINK_LIBRARY) $(LDFLAGS) $(LDLIBS)
 
 # Used to generate list of source files for Autotools, CMakeList and Android.mk
 .PHONY: sources
@@ -800,9 +823,9 @@ ifeq ($(wildcard GNUmakefile.deps),GNUmakefile.deps)
 -include GNUmakefile.deps
 endif # Dependencies
 
-# Cryptogams ARM asm implementation.
+# Cryptogams ARM asm implementation. AES needs -mthumb for Clang
 aes_armv4.o : aes_armv4.S
-	$(CXX) $(strip $(CXXFLAGS) -fpermissive $(CRYPTOGAMS_AES_FLAG) -c) $<
+	$(CXX) $(strip $(CXXFLAGS) $(CRYPTOGAMS_ARMV7_THUMB_FLAG) -c) $<
 
 cpu-features.o: cpu-features.h cpu-features.c
 	$(CXX) $(strip $(CXXFLAGS) -fpermissive -c) cpu-features.c
@@ -867,6 +890,18 @@ rijndael_simd.o : rijndael_simd.cpp
 sha_simd.o : sha_simd.cpp
 	$(CXX) $(strip $(CXXFLAGS) $(SHA_FLAG) -c) $<
 
+# Cryptogams SHA1 asm implementation.
+sha1_armv4.o : sha1_armv4.S
+	$(CXX) $(strip $(CXXFLAGS) $(CRYPTOGAMS_ARMV7_FLAG) -c) $<
+
+# Cryptogams SHA256 asm implementation.
+sha256_armv4.o : sha256_armv4.S
+	$(CXX) $(strip $(CXXFLAGS) $(CRYPTOGAMS_ARMV7_FLAG) -c) $<
+
+# Cryptogams SHA512 asm implementation.
+sha512_armv4.o : sha512_armv4.S
+	$(CXX) $(strip $(CXXFLAGS) $(CRYPTOGAMS_ARMV7_FLAG) -c) $<
+
 # SSE4.2/SHA-NI or ARMv8a available
 shacal2_simd.o : shacal2_simd.cpp
 	$(CXX) $(strip $(CXXFLAGS) $(SHA_FLAG) -c) $<

History.txt

@@ -455,10 +455,10 @@ last several releases.
 8.1.0 - February 22, 2019
       - minor release, no recompile of programs required
       - expanded community input and support
-        56 unique contributors as of this release
+        * 56 unique contributors as of this release
       - fix OS X PowerPC builds with Clang
       - add Microsoft ARM64 support
-      - fix iPhone Simulator build due to missign symbols
+      - fix iPhone Simulator build due to missing symbols
       - add CRYPTOPP_BUGGY_SIMD_LOAD_AND_STORE
       - add carryless multiplies for NIST b233 and k233 curves
       - fix OpenMP build due to use of OpenMP 4 with down-level compilers
@@ -467,3 +467,16 @@ last several releases.
       - add SHAKE-128 and SHAKE-256
       - fix AVX2 build due to _mm256_broadcastsi128_si256
       - add IETF ChaCha, XChaCha, ChaChaPoly1305 and XChaChaPoly1305
+
+8.2.0 - April 28, 2019
+      - minor release, no recompile of programs required
+      - expanded community input and support
+        * 56 unique contributors as of this release
+      - use PowerPC unaligned loads and stores with Power8
+      - add SKIPJACK test vectors
+      - fix SHAKE-128 and SHAKE-256 compile
+      - removed IS_NEON from Makefile
+      - fix Aarch64 build on Fedora 29
+      - fix missing GF2NT_233_Multiply_Reduce_CLMUL in FIPS DLL
+      - add missing BLAKE2 constructors
+      - fix missing BlockSize() in BLAKE2 classes

License.txt

@@ -29,8 +29,9 @@ Daniel J. Bernstein, Jack Lloyd - chacha.cpp, chacha_simd.cpp, chacha_avx.cpp
 Andrew Moon - ed25519, x25519, donna_32.cpp, donna_64.cpp, donna_sse.cpp
 
 The Crypto++ Library uses portions of Andy Polyakov's CRYPTOGAMS for Poly1305
-scalar multiplication and aes_armv4.S. CRYPTOGAMS is dual licensed with a
+scalar multiplication, aes_armv4.S, sha1_armv4.S and sha256_armv4.S. CRYPTOGAMS
-permissive BSD-style license. The CRYPTOGAMS license is reproduced below.
+is dual licensed with a permissive BSD-style license. The CRYPTOGAMS license is
+reproduced below.
 
 The Crypto++ Library uses portions of Jack Lloyd's Botan for ChaCha SSE2 and
 AVX. Botan placed the code in public domain for Crypto++ to use.

Readme.txt

@@ -1,5 +1,5 @@
 Crypto++: free C++ Class Library of Cryptographic Schemes
-Version 8.0 - DEC/28/2018
+Version 8.3 - TBD
 
 Crypto++ Library is a free C++ class library of cryptographic schemes.
 Currently the library contains the following algorithms:
@@ -205,23 +205,29 @@ library in your programs to help avoid unwanted redirections.
 
 *** Side Channel Attacks ***
 
-Crypto++ attempts to resist side channel attacks using various remediations. We
+Crypto++ attempts to resist side channel attacks using various remediations.
-believe the library is mostly hardened but the remdiations may be incomplete. The
+The remdiations are applied as a best effort but are probably incomplete. They
-first line of defense uses hardware instructions when possible for block ciphers,
+are incomplete due to cpu speculation bugs like Spectre, Meltdown, Foreshadow.
-hashes and other primitives. Hardware acceleration remediates many timing attacks.
+The attacks target both cpu caches and internal buffers. Intel generally refers
-The library also uses cache-aware algoirthms and access patterns to minimize leakage.
+to internal buffer attacks as "Microarchitectural Data Sampling" (MDS).
 
-Some of the public key algorithms have branches and some of the branches depend on
+The library uses hardware instructions when possible for block ciphers, hashes
-data that can be private or secret. The branching occurs in some field operations
+and other operations. The hardware acceleration remediates some timing
-like exponentiation over integers and elliptic curves. The branching has been
+attacks. The library also uses cache-aware algoirthms and access patterns
-minimized but not completely eliminated.
+to minimize leakage cache evictions.
 
-Crypto++ does not enagage Specter remediations at this time. The GCC options for
+Some of the public key algorithms have branches and some of the branches depend
-Specter are -mfunction-return=thunk and -mindirect-branch=thunk, and the library
+on data that can be private or secret. The branching occurs in some field
-uses them during testing. If you want the Specter workarounds then add the GCC
+operations like exponentiation over integers and elliptic curves. The branching
-options to your CXXFLAGS when building the library.
+has been minimized but not completely eliminated.
 
-If you suspect or find an information leak then please report it.
+Crypto++ does not enagage Specter remediations at this time. The GCC options
+for Specter are -mfunction-return=thunk and -mindirect-branch=thunk, and the
+library uses them during testing. If you want the Specter workarounds then add
+the GCC options to your CXXFLAGS when building the library.
+
+To help resist attacks you should disable hyperthreading on cpus. If you
+suspect or find an information leak then please report it.
 
 *** Documentation and Support ***
 
@@ -290,8 +296,36 @@ documentation is one of the highest returns on investment.
 The items in this section comprise the most recent history. Please see History.txt
 for the record back to Crypto++ 1.0.
 
-8.0.0 - December 28, 2018
+8.2.0 - April 28, 2019
+      - minor release, no recompile of programs required
+      - expanded community input and support
+        * 56 unique contributors as of this release
+      - use PowerPC unaligned loads and stores with Power8
+      - add SKIPJACK test vectors
+      - fix SHAKE-128 and SHAKE-256 compile
+      - removed IS_NEON from Makefile
+      - fix Aarch64 build on Fedora 29
+      - fix missing GF2NT_233_Multiply_Reduce_CLMUL in FIPS DLL
+      - add missing BLAKE2 constructors
+      - fix missing BlockSize() in BLAKE2 classes
 
+8.1.0 - February 22, 2019
+      - minor release, no recompile of programs required
+      - expanded community input and support
+        * 56 unique contributors as of this release
+      - fix OS X PowerPC builds with Clang
+      - add Microsoft ARM64 support
+      - fix iPhone Simulator build due to missing symbols
+      - add CRYPTOPP_BUGGY_SIMD_LOAD_AND_STORE
+      - add carryless multiplies for NIST b233 and k233 curves
+      - fix OpenMP build due to use of OpenMP 4 with down-level compilers
+      - add SignStream and VerifyStream for ed25519 and large files
+      - fix missing AlgorithmProvider in PanamaHash
+      - add SHAKE-128 and SHAKE-256
+      - fix AVX2 build due to _mm256_broadcastsi128_si256
+      - add IETF ChaCha, XChaCha, ChaChaPoly1305 and XChaChaPoly1305
+
+8.0.0 - December 28, 2018
       - major release, recompile of programs required
       - expanded community input and support
          * 54 unique contributors as of this release

TestData/ecies_p160.dat

@@ -0,0 +1 @@
+3081C80201003081A406072A8648CE3D0201308198020101302006072A8648CE3D0101021500FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF7FFFFFFF302C0414FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF7FFFFFFC04141C97BEFC54BD7A8B65ACF89F81D4D4ADC565FA450429044A96B5688EF573284664698968C38BB913CBFC8223A628553168947D59DCC912042351377AC5FB3202150100000000000000000001F4C8F927AED3CA752257020101041C301A02010104150023A68821ABB99DBB8429ED2320D61A8EA4C6D81B

TestData/ecies_t163.dat

@@ -0,0 +1 @@
+3081D10201003081AD06072A8648CE3D02013081A1020101302506072A8648CE3D0102301A020200A306092A8648CE3D010203033009020103020106020107302E041507B6882CAAEFA84F9554FF8428BD88E246D2782AE204150713612DCDDCB40AAB946BDA29CA91F73AF958AFD9042B040369979697AB43897789566789567F787A7876A65400435EDB42EFAFB2989D51FEFCE3C80988F41FF883021503FFFFFFFFFFFFFFFFFFFF48AAB689C29CA710279B020102041C301A02010104150003693AB4D83EE8B544548BE7647AEA0EA64E8211

TestData/elgc1024.dat

@@ -1 +1 @@
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

TestPrograms/test_arm_sha1.cxx

@@ -0,0 +1,18 @@
+#include <arm_neon.h>
+#include <stdint.h>
+
+// test_acle.h determines if this is available. Then,
+// -DCRYPTOPP_ARM_ACLE_AVAILABLE=0 is added to CXXFLAGS
+// if the ACLE header is not available.
+#if (CRYPTOPP_ARM_ACLE_AVAILABLE)
+#  include <arm_acle.h>
+#endif
+
+int main(int argc, char* argv[])
+{
+	uint32x4_t y = {0};
+	y=vsha1cq_u32(y,0,y);
+	y=vsha1mq_u32(y,1,y);
+	y=vsha1pq_u32(y,2,y);
+	return 0;
+}

TestPrograms/test_arm_sha256.cxx

@@ -11,9 +11,6 @@
 int main(int argc, char* argv[])
 {
 	uint32x4_t y = {0};
-	y=vsha1cq_u32(y,0,y);
-	y=vsha1mq_u32(y,1,y);
-	y=vsha1pq_u32(y,2,y);
 	y=vsha256hq_u32(y, y, y);
 	y=vsha256h2q_u32(y, y, y);
 	y=vsha256su1q_u32(y, y, y);

TestPrograms/test_x86_avx2.cxx

@@ -1,7 +1,9 @@
 #include <immintrin.h>
 int main(int argc, char* argv[])
 {
-	__m256i x = _mm256_setzero_si256();
+	// _mm256_broadcastsi128_si256 due to Clang
-	x=_mm256_add_epi64 (x,x);
+	__m128i x = _mm_setzero_si128 ();
+	__m256i y = _mm256_broadcastsi128_si256 (x);
+	y = _mm256_add_epi64 (y,y);
 	return 0;
 }

TestPrograms/test_x86_rdrand.cxx

@@ -0,0 +1,7 @@
+#include <immintrin.h>
+int main(int argc, char* argv[])
+{
+    unsigned int x=0;
+    (void)_rdrand32_step (&x);
+    return x == 0 ? 0 : 0;
+}

TestPrograms/test_x86_rdseed.cxx

@@ -0,0 +1,7 @@
+#include <immintrin.h>
+int main(int argc, char* argv[])
+{
+    unsigned int x=0;
+    (void)_rdseed32_step (&x);
+    return x == 0 ? 0 : 0;
+}

TestScripts/change-version.sh

@@ -4,11 +4,14 @@
 # building the docs. Before running the script, copy it to the root
 # directory. After running this script, you can 'make docs'
 
-sed 's|Library 8.1 API|Library 8.0 API|g' cryptlib.h > cryptlib.h.new
+sed 's|Library 8.3 API|Library 8.2 API|g' cryptlib.h > cryptlib.h.new
 mv cryptlib.h.new cryptlib.h
 
-sed 's|= 8.1|= 8.0|g' Doxyfile > Doxyfile.new
+sed 's|= 8.3|= 8.2|g' Doxyfile > Doxyfile.new
 mv Doxyfile.new Doxyfile
 
-sed 's|CRYPTOPP_VERSION 810|CRYPTOPP_VERSION 800|g' config.h > config.h.new
+sed 's|CRYPTOPP_MINOR 3|CRYPTOPP_MINOR 2|g' config_ver.h > config_ver.h.new
-mv config.h.new config.h
+mv config_ver.h.new config_ver.h
+
+sed 's|CRYPTOPP_VERSION 830|CRYPTOPP_VERSION 820|g' config_ver.h > config_ver.h.new
+mv config_ver.h.new config_ver.h

TestScripts/cryptest-autotools.sh

@@ -6,17 +6,13 @@ function cleanup {
 }
 trap cleanup EXIT
 
+#############################################################################
+
 GREP=grep
 SED=sed
 AWK=awk
 MAKE=make
 
-# Fixup ancient Bash
-# https://unix.stackexchange.com/q/468579/56041
-if [[ -z "$BASH_SOURCE" ]]; then
-	BASH_SOURCE="$0"
-fi
-
 # Fixup, Solaris and friends
 if [[ (-d /usr/xpg4/bin) ]]; then
 	SED=/usr/xpg4/bin/sed
@@ -34,7 +30,6 @@ if [[ "$IS_DARWIN" -ne 0 ]]; then
 	export LC_ALL=C
 fi
 
-# Fixup for Solaris and BSDs
 # Fixup for Solaris and BSDs
 if [[ ! -z $(command -v gmake) ]]; then
 	MAKE=gmake
@@ -53,24 +48,30 @@ elif [[ ! -z $(command -v glibtool) ]]; then
 	LIBTOOLIZE=$(command -v glibtool)
 fi
 
-# Fecth the three required files
+#############################################################################
-if ! wget --no-check-certificate 'https://raw.githubusercontent.com/noloader/cryptopp-autotools/master/Makefile.am' -O Makefile.am; then
-	echo "Makefile.am download failed"
-	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
-fi
 
-if ! wget --no-check-certificate 'https://raw.githubusercontent.com/noloader/cryptopp-autotools/master/configure.ac' -O configure.ac; then
+echo "Downloading configure.ac"
+if ! wget -O configure.ac -q --no-check-certificate 'https://raw.githubusercontent.com/noloader/cryptopp-autotools/master/configure.ac'; then
 	echo "configure.ac download failed"
 	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
 fi
 
-if ! wget --no-check-certificate 'https://raw.githubusercontent.com/noloader/cryptopp-autotools/master/libcryptopp.pc.in' -O libcryptopp.pc.in; then
+echo "Downloading Makefile.am"
+if ! wget -O Makefile.am -q --no-check-certificate 'https://raw.githubusercontent.com/noloader/cryptopp-autotools/master/Makefile.am'; then
+	echo "Makefile.am download failed"
+	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
+fi
+
+echo "Downloading libcryptopp.pc.in"
+if ! wget -O libcryptopp.pc.in -q --no-check-certificate 'https://raw.githubusercontent.com/noloader/cryptopp-autotools/master/libcryptopp.pc.in'; then
 	echo "libcryptopp.pc.in download failed"
 	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
 fi
 
 mkdir -p m4/
 
+#############################################################################
+
 if [[ -z $(command -v autoupdate) ]]; then
 	echo "Cannot find autoupdate. Things may fail."
 fi
@@ -79,56 +80,71 @@ if [[ -z "$LIBTOOLIZE" ]]; then
 	echo "Cannot find libtoolize. Things may fail."
 fi
 
+if [[ -z $(command -v automake) ]]; then
+	echo "Cannot find automake. Things may fail."
+fi
+
 if [[ -z $(command -v autoreconf) ]]; then
 	echo "Cannot find autoreconf. Things may fail."
 fi
 
 echo "Running autoupdate"
-if ! autoupdate 2>/dev/null; then
+if ! autoupdate &>/dev/null; then
 	echo "autoupdate failed."
 	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
 fi
 
 echo "Running libtoolize"
-if ! "$LIBTOOLIZE" 2>/dev/null; then
+if ! "$LIBTOOLIZE" --force --install &>/dev/null; then
-	echo "libtoolize failed."
+	echo "libtoolize failed... skipping."
-	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
+	# [[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
 fi
 
 # Run autoreconf twice on failure. Also see
 # https://github.com/tracebox/tracebox/issues/57
 echo "Running autoreconf"
-if ! autoreconf 2>/dev/null; then
+if ! autoreconf --force --install &>/dev/null; then
 	echo "autoreconf failed, running again."
-	if ! autoreconf -fi; then
+	if ! autoreconf --force --install; then
 		echo "autoreconf failed, again."
 		[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
 	fi
 fi
 
-# Sparc need +w
+#############################################################################
-if [[ -e config.sub ]]; then
-	chmod +w config.sub
-fi
-if [[ -e config.guess ]]; then
-	chmod +w config.guess
-fi
 
 # Update config.sub config.guess. GNU recommends using the latest for all projects.
 echo "Updating config.sub"
-wget --no-check-certificate 'https://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub' -O config.sub
+wget -O config.sub.new -q --no-check-certificate 'https://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub'
 
-if [[ -e config.sub ]]; then
+# Solaris removes +w, can't overwrite
+chmod +w config.sub
+mv config.sub.new config.sub
 chmod +x config.sub
+
+if [[ "$IS_DARWIN" -ne 0 ]] && [[ -n $(command -v xattr) ]]; then
+	echo "Removing config.sub quarantine"
+	xattr -d "com.apple.quarantine" config.sub &>/dev/null
 fi
 
 echo "Updating config.guess"
-wget --no-check-certificate 'https://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess' -O config.guess
+wget -O config.guess.new -q --no-check-certificate 'https://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess'
 
-if [[ -e config.guess ]]; then
+# Solaris removes +w, can't overwrite
+chmod +w config.guess
+mv config.guess.new config.guess
 chmod +x config.guess
+
+if [[ "$IS_DARWIN" -ne 0 ]] && [[ -n $(command -v xattr) ]]; then
+	echo "Removing config.guess quarantine"
+	xattr -d "com.apple.quarantine" config.guess &>/dev/null
 fi
 
+#############################################################################
+
+echo "Running configure"
+echo ""
+
 if ! ./configure; then
 	echo "configure failed."
 	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
@@ -136,6 +152,8 @@ fi
 
 "$MAKE" clean 2>/dev/null
 
+#############################################################################
+
 if ! "$MAKE" -j2 -f Makefile; then
 	echo "make failed."
 	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1

TestScripts/cryptest-symbols.sh

@@ -18,7 +18,7 @@ fi
 
 PWD_DIR=$(pwd)
 function cleanup {
-    rm -f adhoc.cpp *.a *.o *.so *.dylib GNUmakefile-symbols
+    rm -f adhoc.cpp *.a *.o *.so *.dylib
     cd "$PWD_DIR"
 }
 trap cleanup EXIT
@@ -26,7 +26,7 @@ trap cleanup EXIT
 ############################################
 # Tags to test
 
-OLD_VERSION_TAG=CRYPTOPP_8_0_0
+OLD_VERSION_TAG=CRYPTOPP_8_2_0
 NEW_VERSION_TAG=master
 
 ############################################
@@ -40,7 +40,7 @@ if [[ ! (-z "$DIRTY") ]]; then
 	read -p "Type 'Y' to proceed or 'N' to exit. Proceed? " -n 1 -r
 	echo # (optional) move to a new line
 	if [[ !($REPLY =~ ^[Yy]$) ]]; then
-		[[ "$0" = "$BASH_SOURCE" ]] && exit 0 || return 0
+		exit 0
 	fi
 else
 	echo
@@ -222,29 +222,13 @@ fi
 
 "$MAKE" distclean &>/dev/null
 
-rm -f GNUmakefile-symbols
-
 git checkout master -f &>/dev/null
-cp GNUmakefile GNUmakefile-symbols
 
 git checkout "$OLD_VERSION_TAG" -f &>/dev/null
 
 if [[ "$?" -ne "0" ]]; then
 	echo "Failed to checkout $OLD_VERSION_TAG"
-	[[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+	exit 1
-fi
-
-echo
-echo "****************************************************************"
-echo "Patching makefile for dynamic linking by cryptest.exe"
-echo "****************************************************************"
-
-if [[ "$IS_DARWIN" -ne "0" ]]; then
-	"$SED" "$SED_OPTS" -e 's|libcryptopp.a $(TESTOBJS)|libcryptopp.dylib $(TESTOBJS)|g' GNUmakefile-symbols
-	"$SED" "$SED_OPTS" -e 's|$(TESTOBJS) ./libcryptopp.a |$(TESTOBJS) ./libcryptopp.dylib |g' GNUmakefile-symbols
-else
-	"$SED" "$SED_OPTS" -e 's|libcryptopp.a $(TESTOBJS)|libcryptopp.so $(TESTOBJS)|g' GNUmakefile-symbols
-	"$SED" "$SED_OPTS" -e 's|$(TESTOBJS) ./libcryptopp.a |$(TESTOBJS) ./libcryptopp.so |g' GNUmakefile-symbols
 fi
 
 echo
@@ -253,17 +237,17 @@ echo "Building dynamic library for $OLD_VERSION_TAG"
 echo "****************************************************************"
 echo
 
-"$MAKE" "${MAKEARGS[@]}" -f GNUmakefile-symbols dynamic
-
 if [[ "$IS_DARWIN" -ne "0" ]]; then
-	LIBNAME=libcryptopp.dylib
+	LINK_LIBRARY=libcryptopp.dylib
 else
-	LIBNAME=libcryptopp.so
+	LINK_LIBRARY=libcryptopp.so
 fi
 
-if [[ ! -f "$LIBNAME" ]]; then
+LINK_LIBRARY="$LINK_LIBRARY" "$MAKE" "${MAKEARGS[@]}" -f GNUmakefile dynamic
+
+if [[ ! -f "$LINK_LIBRARY" ]]; then
 	echo "Failed to make $OLD_VERSION_TAG library"
-	[[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+	exit 1
 fi
 
 echo
@@ -272,11 +256,11 @@ echo "Building cryptest.exe for $OLD_VERSION_TAG"
 echo "****************************************************************"
 echo
 
-"$MAKE" "${MAKEARGS[@]}" -f GNUmakefile-symbols cryptest.exe
+"$MAKE" "${MAKEARGS[@]}" -f GNUmakefile cryptest.exe
 
 if [[ ! -f "cryptest.exe" ]]; then
 	echo "Failed to make cryptest.exe"
-	[[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+	exit 1
 fi
 
 echo
@@ -304,7 +288,7 @@ git checkout "$NEW_VERSION_TAG" -f &>/dev/null
 
 if [[ "$?" -ne "0" ]]; then
 	echo "Failed to checkout $OLD_VERSION_TAG"
-	[[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+	exit 1
 fi
 
 echo
@@ -313,11 +297,11 @@ echo "Building dynamic library for $NEW_VERSION_TAG"
 echo "****************************************************************"
 echo
 
-"$MAKE" "${MAKEARGS[@]}" -f GNUmakefile-symbols dynamic
+LINK_LIBRARY="$LINK_LIBRARY" "$MAKE" "${MAKEARGS[@]}" -f GNUmakefile dynamic
 
-if [[ ! -f "$LIBNAME" ]]; then
+if [[ ! -f "$LINK_LIBRARY" ]]; then
 	echo "Failed to make $NEW_VERSION_TAG library"
-	[[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+	exit 1
 fi
 
 echo
@@ -338,7 +322,7 @@ git checkout master -f &>/dev/null
 
 if [[ "$?" -ne "0" ]]; then
 	echo "Failed to checkout Master"
-	[[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+	exit 1
 fi
 
-[[ "$0" = "$BASH_SOURCE" ]] && exit 0 || return 0
+exit 0

TestScripts/governor.sh

@@ -1,25 +1,25 @@
 #!/usr/bin/env bash
 
 # This scripts queries and modifies CPU scaling frequencies to produce more
-# accurate benchmark results. To move from a low energy state C-state to a
+# accurate benchmark results. To move from a low energy state to a higher
-# higher one, run 'governor.sh performance'. To move back to a low power state
+# one, run 'governor.sh performance'. To move back to a low power state
-# run 'governor.sh powersave' or reboot. The script is based on code by
+# run 'governor.sh powersave' or 'governor.sh ondemand' or reboot. The script
-# Andy Polyakov, http://www.openssl.org/~appro/cryptogams/.
+# based on code by Andy Polyakov, http://www.openssl.org/~appro/cryptogams/.
 
 # Fixup ancient Bash
 # https://unix.stackexchange.com/q/468579/56041
-if [[ -z "$BASH_SOURCE" ]]; then
+if [[ -z "${BASH_SOURCE[0]}" ]]; then
 	BASH_SOURCE="$0"
 fi
 
 if [[ "$EUID" -ne 0 ]]; then
     echo "This script must be run as root"
-    [[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+    [[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
 fi
 
 if [ "x$1" = "x" ]; then
     echo "usage: $0 on[demand]|pe[rformance]|po[wersave]|us[erspace]?"
-    [[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+    [[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
 fi
 
 # "on demand" may result in a "invalid write argument" or similar
@@ -27,40 +27,41 @@ case $1 in
     on*|de*)    governor="ondemand";;
     po*|pw*)    governor="powersave";;
     pe*)        governor="performance";;
+    co*)        governor="conservative";;
     us*)        governor="userspace";;
     \?)         ;;
     *)          echo "$1: unrecognized governor";;
 esac
 
 if [ -z "$governor" ]; then
-	[[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
 fi
 
 cpus=$(ls /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor 2>/dev/null)
 
 if [ -z "$cpus" ]; then
 	echo "Failed to read CPU system device tree"
-	[[ "$0" = "$BASH_SOURCE" ]] && exit 1 || return 1
+	[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 1 || return 1
 fi
 
 echo "Current CPU governor scaling settings:"
 count=0
 for cpu in $cpus; do
-	echo "  CPU $count:" $(cat "$cpu")
+	echo "  CPU $count: $(cat "$cpu")"
 	((count++))
 done
 
 if [ "x$governor" != "x" ]; then
     for cpu in $cpus; do
-        echo $governor > $cpu
+        echo "$governor" > "$cpu"
     done
 fi
 
 echo "New CPU governor scaling settings:"
 count=0
 for cpu in $cpus; do
-	echo "  CPU $count:" $(cat "$cpu")
+	echo "  CPU $count: $(cat "$cpu")"
 	((count++))
 done
 
-[[ "$0" = "$BASH_SOURCE" ]] && exit 0 || return 0
+[[ "$0" = "${BASH_SOURCE[0]}" ]] && exit 0 || return 0

TestScripts/setenv-android-gcc.sh

@@ -18,6 +18,12 @@
 
 # set -eu
 
+# Sanity check
+if [ "$0" = "${BASH_SOURCE[0]}" ]; then
+    echo "Please source this setenv script"
+    exit 0
+fi
+
 unset IS_CROSS_COMPILE
 
 unset IS_IOS
@@ -39,7 +45,7 @@ unset CPP CC CXX LD AS AR RANLIB STRIP
 # Similar to a "make clean"
 if [ x"${1-}" = "xunset" ]; then
     echo "Unsetting script variables. PATH may remain tainted"
-	[ "$0" = "$BASH_SOURCE" ] && exit 0 || return 0
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 0 || return 0
 fi
 
 # Set AOSP_TOOLCHAIN_SUFFIX to your preference of tools and STL library.
@@ -71,7 +77,7 @@ if [ -z "${AOSP_API-}" ]; then
 else
     echo "WARNING: Using AOSP_API has been deprecated. Please use AOSP_API_VERSION instead."
     echo "If you set for example AOSP_API=android-23 then now instead set AOSP_API_VERSION=23"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 #####################################################################
@@ -98,7 +104,7 @@ fi
 # Error checking
 if [ ! -d "$ANDROID_NDK_ROOT/toolchains" ]; then
     echo "ERROR: ANDROID_NDK_ROOT is not a valid path. Please set it."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 #####################################################################
@@ -127,7 +133,7 @@ case "$THE_ARCH" in
     ;;
   hard|armv7a-hard|armeabi-v7a-hard)
     echo hard, armv7a-hard and armeabi-v7a-hard are not supported, as android uses softfloats
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
     #TOOLCHAIN_ARCH="arm-linux-androideabi"
     #TOOLCHAIN_NAME="arm-linux-androideabi"
     #AOSP_ABI="armeabi-v7a"
@@ -178,7 +184,7 @@ case "$THE_ARCH" in
     ;;
   *)
     echo "ERROR: Unknown architecture $1"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
     ;;
 esac
 
@@ -218,48 +224,48 @@ done
 # Error checking
 if [ -z "$AOSP_TOOLCHAIN_PATH" ] || [ ! -d "$AOSP_TOOLCHAIN_PATH" ]; then
     echo "ERROR: AOSP_TOOLCHAIN_PATH is not valid. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$CPP" ]; then
     echo "ERROR: Failed to find Android cpp. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$CC" ]; then
     echo "ERROR: Failed to find Android gcc. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$CXX" ]; then
     echo "ERROR: Failed to find Android g++. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$RANLIB" ]; then
     echo "ERROR: Failed to find Android ranlib. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$AR" ]; then
     echo "ERROR: Failed to find Android ar. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$AS" ]; then
     echo "ERROR: Failed to find Android as. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$LD" ]; then
     echo "ERROR: Failed to find Android ld. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Only modify/export PATH if AOSP_TOOLCHAIN_PATH good
@@ -278,10 +284,10 @@ fi
 # Error checking
 if [ ! -d "$ANDROID_NDK_ROOT/platforms/$AOSP_API" ]; then
     echo "ERROR: AOSP_API is not valid. Does the NDK support the API? Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 elif [ ! -d "$ANDROID_NDK_ROOT/platforms/$AOSP_API/$AOSP_ARCH" ]; then
     echo "ERROR: AOSP_ARCH is not valid. Does the NDK support the architecture? Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Android SYSROOT. It will be used on the command line with --sysroot
@@ -328,7 +334,7 @@ case "$THE_STL" in
     echo WARNING: llvm is still in experimental state and migth not work as expected
     if [ ! -d "$LLVM_INCLUDE_DIR" ]; then
         echo "ERROR: Unable to locate include LLVM directory at $LLVM_INCLUDE_DIR -- has it moved since NDK r16beta1?"
-		[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+        [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
     fi
     AOSP_STL_INC="$LLVM_INCLUDE_DIR"
     AOSP_STL_LIB="$ANDROID_NDK_ROOT/sources/cxx-stl/llvm-libc++/libs/$AOSP_ABI/libc++_static.a"
@@ -337,26 +343,26 @@ case "$THE_STL" in
     echo WARNING: llvm is still in experimental state and migth not work as expected
     if [ ! -d "$LLVM_INCLUDE_DIR" ]; then
         echo "ERROR: Unable to locate LLVM include directory at $LLVM_INCLUDE_DIR -- has it moved since NDK r16beta1?"
-		[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+        [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
     fi
     AOSP_STL_INC="$LLVM_INCLUDE_DIR"
     AOSP_STL_LIB="$ANDROID_NDK_ROOT/sources/cxx-stl/llvm-libc++/libs/$AOSP_ABI/libc++_shared.so"
     ;;
   *)
     echo "ERROR: Unknown STL library $2"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 esac
 
 # Error checking
 if [ ! -d "$AOSP_STL_INC" ] || [ ! -e "$AOSP_STL_INC/memory" ]; then
     echo "ERROR: AOSP_STL_INC is not valid. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_STL_LIB" ]; then
     echo "ERROR: AOSP_STL_LIB is not valid. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 export AOSP_STL_INC
@@ -372,13 +378,13 @@ fi
 
 if [[ ! -e "$ANDROID_NDK_ROOT/sources/android/cpufeatures/cpu-features.h" ]]; then
     echo "ERROR: Unable to locate cpu-features.h"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 cp "$ANDROID_NDK_ROOT/sources/android/cpufeatures/cpu-features.h" .
 
 if [[ ! -e "$ANDROID_NDK_ROOT/sources/android/cpufeatures/cpu-features.c" ]]; then
     echo "ERROR: Unable to locate cpu-features.c"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 cp "$ANDROID_NDK_ROOT/sources/android/cpufeatures/cpu-features.c" .
 
@@ -446,4 +452,4 @@ echo "shared object using 'HAS_SOLIB_VERSION=1 make -f GNUmakefile-cross'"
 echo "*******************************************************************************"
 echo
 
-[ "$0" = "$BASH_SOURCE" ] && exit 0 || return 0
+[ "$0" = "${BASH_SOURCE[0]}" ] && exit 0 || return 0

TestScripts/setenv-android-old.sh

@@ -20,6 +20,12 @@
 
 # set -eu
 
+# Sanity check
+if [ "$0" = "${BASH_SOURCE[0]}" ]; then
+    echo "Please source this setenv script"
+    exit 0
+fi
+
 unset IS_CROSS_COMPILE
 
 unset IS_IOS
@@ -39,7 +45,7 @@ unset CPP CC CXX LD AS AR RANLIB STRIP
 # Similar to a "make clean"
 if [ x"${1-}" = "xunset" ]; then
     echo "Unsetting script variables. PATH may remain tainted"
-	[ "$0" = "$BASH_SOURCE" ] && exit 0 || return 0
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 0 || return 0
 fi
 
 # Set AOSP_TOOLCHAIN_SUFFIX to your preference of tools and STL library.
@@ -90,7 +96,7 @@ fi
 # Error checking
 if [ ! -d "$ANDROID_NDK_ROOT/toolchains" ]; then
     echo "ERROR: ANDROID_NDK_ROOT is not a valid path. Please set it."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 #####################################################################
@@ -168,7 +174,7 @@ case "$THE_ARCH" in
     ;;
   *)
     echo "ERROR: Unknown architecture $1"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
     ;;
 esac
 
@@ -204,48 +210,48 @@ done
 # Error checking
 if [ -z "$AOSP_TOOLCHAIN_PATH" ] || [ ! -d "$AOSP_TOOLCHAIN_PATH" ]; then
     echo "ERROR: AOSP_TOOLCHAIN_PATH is not valid. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$CPP" ]; then
     echo "ERROR: Failed to find Android cpp. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$CC" ]; then
     echo "ERROR: Failed to find Android gcc. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$CXX" ]; then
     echo "ERROR: Failed to find Android g++. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$RANLIB" ]; then
     echo "ERROR: Failed to find Android ranlib. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$AR" ]; then
     echo "ERROR: Failed to find Android ar. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$AS" ]; then
     echo "ERROR: Failed to find Android as. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_TOOLCHAIN_PATH/$LD" ]; then
     echo "ERROR: Failed to find Android ld. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Only modify/export PATH if AOSP_TOOLCHAIN_PATH good
@@ -264,10 +270,10 @@ fi
 # Error checking
 if [ ! -d "$ANDROID_NDK_ROOT/platforms/$AOSP_API" ]; then
     echo "ERROR: AOSP_API is not valid. Does the NDK support the API? Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 elif [ ! -d "$ANDROID_NDK_ROOT/platforms/$AOSP_API/$AOSP_ARCH" ]; then
     echo "ERROR: AOSP_ARCH is not valid. Does the NDK support the architecture? Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Android SYSROOT. It will be used on the command line with --sysroot
@@ -312,7 +318,7 @@ case "$THE_STL" in
   llvm-static)
       if [ ! -d "$LLVM_INCLUDE_DIR" ]; then
         echo "ERROR: Unable to locate include LLVM directory at $LLVM_INCLUDE_DIR -- has it moved since NDK r16beta1?"
-		[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+        [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
     fi
     AOSP_STL_INC="$LLVM_INCLUDE_DIR"
     AOSP_STL_LIB="$ANDROID_NDK_ROOT/sources/cxx-stl/llvm-libc++/libs/$AOSP_ABI/libc++_static.a"
@@ -320,26 +326,26 @@ case "$THE_STL" in
   llvm|llvm-shared)
       if [ ! -d "$LLVM_INCLUDE_DIR" ]; then
         echo "ERROR: Unable to locate LLVM include directory at $LLVM_INCLUDE_DIR -- has it moved since NDK r16beta1?"
-		[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+        [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
     fi
     AOSP_STL_INC="$LLVM_INCLUDE_DIR"
     AOSP_STL_LIB="$ANDROID_NDK_ROOT/sources/cxx-stl/llvm-libc++/libs/$AOSP_ABI/libc++_shared.so"
     ;;
   *)
     echo "ERROR: Unknown STL library $2"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 esac
 
 # Error checking
 if [ ! -d "$AOSP_STL_INC" ] || [ ! -e "$AOSP_STL_INC/memory" ]; then
     echo "ERROR: AOSP_STL_INC is not valid. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Error checking
 if [ ! -e "$AOSP_STL_LIB" ]; then
     echo "ERROR: AOSP_STL_LIB is not valid. Please edit this script."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 export AOSP_STL_INC
@@ -355,13 +361,13 @@ fi
 
 if [[ ! -e "$ANDROID_NDK_ROOT/sources/android/cpufeatures/cpu-features.h" ]]; then
     echo "ERROR: Unable to locate cpu-features.h"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 cp "$ANDROID_NDK_ROOT/sources/android/cpufeatures/cpu-features.h" .
 
 if [[ ! -e "$ANDROID_NDK_ROOT/sources/android/cpufeatures/cpu-features.c" ]]; then
     echo "ERROR: Unable to locate cpu-features.c"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 cp "$ANDROID_NDK_ROOT/sources/android/cpufeatures/cpu-features.c" .
 
@@ -427,4 +433,4 @@ echo "shared object using 'HAS_SOLIB_VERSION=1 make -f GNUmakefile-cross'"
 echo "*******************************************************************************"
 echo
 
-[ "$0" = "$BASH_SOURCE" ] && exit 0 || return 0
+[ "$0" = "${BASH_SOURCE[0]}" ] && exit 0 || return 0

TestScripts/setenv-embedded.sh

@@ -15,6 +15,12 @@
 
 # set -eu
 
+# Sanity check
+if [ "$0" = "${BASH_SOURCE[0]}" ]; then
+    echo "Please source this setenv script"
+    exit 0
+fi
+
 # Unset old options
 
 unset IS_CROSS_COMPILE
@@ -29,7 +35,7 @@ fi
 
 if [ ! -d "$ARM_EMBEDDED_TOOLCHAIN" ]; then
     echo "ARM_EMBEDDED_TOOLCHAIN is not valid"
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Fedora
@@ -50,37 +56,37 @@ export RANLIB="$ARM_EMBEDDED_TOOLCHAIN/$TOOL_PREFIX-ranlib"
 # Test a few of the tools
 if [ ! -e "$CPP" ]; then
   echo "ERROR: CPP is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 if [ ! -e "$CC" ]; then
   echo "ERROR: CC is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 if [ ! -e "$CXX" ]; then
   echo "ERROR: CXX is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 if [ ! -e "$AR" ]; then
   echo "ERROR: AR is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 if [ ! -e "$AS" ]; then
   echo "ERROR: AS is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 if [ ! -e "$RANLIB" ]; then
   echo "ERROR: RANLIB is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 if [ ! -e "$LD" ]; then
   echo "ERROR: LD is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # The Crypto++ Makefile uses these to disable host settings like
@@ -94,7 +100,7 @@ fi
 
 if [ ! -d "$ARM_EMBEDDED_SYSROOT" ]; then
   echo "ERROR: ARM_EMBEDDED_SYSROOT is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Fix C++ header paths for Ubuntu
@@ -104,12 +110,12 @@ ARM_EMBEDDED_CXX_HEADERS="$ARM_EMBEDDED_SYSROOT/include/c++/$ARM_EMBEDDED_TOOLCH
 
 if [ ! -d "$ARM_EMBEDDED_CXX_HEADERS" ]; then
   echo "ERROR: ARM_EMBEDDED_CXX_HEADERS is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 if [ ! -d "$ARM_EMBEDDED_CXX_HEADERS/arm-linux-gnueabi" ]; then
   echo "ERROR: ARM_EMBEDDED_CXX_HEADERS is not valid"
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Finally, the flags...
@@ -142,4 +148,4 @@ echo "shared object using 'HAS_SOLIB_VERSION=1 make -f GNUmakefile-cross'"
 echo "*******************************************************************************"
 echo
 
-[ "$0" = "$BASH_SOURCE" ] && exit 0 || return 0
+[ "$0" = "${BASH_SOURCE[0]}" ] && exit 0 || return 0

TestScripts/setenv-ios.sh

@@ -13,6 +13,12 @@
 
 # set -eu
 
+# Sanity check
+if [ "$0" = "${BASH_SOURCE[0]}" ]; then
+    echo "Please source this setenv script"
+    exit 0
+fi
+
 #########################################
 #####       Clear old options       #####
 #########################################
@@ -149,7 +155,7 @@ fi
 
 if [ ! -d "$XCODE_DEVELOPER" ]; then
   echo "ERROR: unable to find XCODE_DEVELOPER directory."
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Default toolchain location
@@ -157,7 +163,7 @@ XCODE_TOOLCHAIN="$XCODE_DEVELOPER/usr/bin"
 
 if [ ! -d "$XCODE_TOOLCHAIN" ]; then
   echo "ERROR: unable to find XCODE_TOOLCHAIN directory."
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # XCODE_DEVELOPER_TOP is the top of the development tools tree
@@ -165,7 +171,7 @@ XCODE_DEVELOPER_TOP="$XCODE_DEVELOPER/Platforms/$APPLE_SDK.platform/Developer"
 
 if [ ! -d "$XCODE_DEVELOPER_TOP" ]; then
   echo "ERROR: unable to find XCODE_DEVELOPER_TOP directory."
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # IOS_TOOLCHAIN is the location of the actual compiler tools.
@@ -177,7 +183,7 @@ fi
 
 if [ -z "$IOS_TOOLCHAIN" ] || [ ! -d "$IOS_TOOLCHAIN" ]; then
   echo "ERROR: unable to find Xcode cross-compiler tools."
-  [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+  [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 #
@@ -196,7 +202,7 @@ done
 # Error checking
 if [ -z "$XCODE_SDK" ]; then
     echo "ERROR: unable to find a SDK."
-    [ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # https://github.com/weidai11/cryptopp/issues/635
@@ -237,7 +243,7 @@ fi
 # https://stackoverflow.com/q/24841283/608639
 # -tvos_simulator_version_min does not work though it is in LLVM sources.
 if [ "$APPLE_SDK" == "AppleTVSimulator" ]; then
-  IOS_FLAGS="$IOS_FLAGS -tvos_simulator_version_min -DCRYPTOPP_DISABLE_ASM"
+  IOS_FLAGS="$IOS_FLAGS -DCRYPTOPP_DISABLE_ASM"
 fi
 
 # Simulator uses i386 or x86_64, Device uses ARMv5, ARMv6, ARMv7, ARMv7s or ARMv8
@@ -285,7 +291,7 @@ if [ ! -z "$IOS_TOOLCHAIN" ] && [ ! -z "$XCODE_TOOLCHAIN" ]; then
     fi
 else
     echo "ERROR: unable to set new PATH."
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 ########################################
@@ -307,7 +313,7 @@ do
 done
 
 if [ "$FOUND_ALL" -eq "0" ]; then
-	[ "$0" = "$BASH_SOURCE" ] && exit 1 || return 1
+    [ "$0" = "${BASH_SOURCE[0]}" ] && exit 1 || return 1
 fi
 
 # Exports added for Autotools. GNUmakefile-cross does not use them.
@@ -329,4 +335,4 @@ echo "shared object using 'HAS_SOLIB_VERSION=1 make -f GNUmakefile-cross'"
 echo "*******************************************************************************"
 echo
 
-[ "$0" = "$BASH_SOURCE" ] && exit 0 || return 0
+[ "$0" = "${BASH_SOURCE[0]}" ] && exit 0 || return 0

TestVectors/all.txt

@@ -49,6 +49,7 @@ Test: TestVectors/sha.txt
 Test: TestVectors/simeck.txt
 Test: TestVectors/simon.txt
 Test: TestVectors/siphash.txt
+Test: TestVectors/skipjack.txt
 Test: TestVectors/sm3.txt
 Test: TestVectors/sm4.txt
 Test: TestVectors/sosemanuk.txt

adv_simd.h

@@ -355,9 +355,7 @@ inline size_t AdvancedProcessBlocks128_6x1_NEON(F1 func1, F6 func6,
     CRYPTOPP_ASSERT(length >= 16);
 
     const unsigned int w_one[] = {0, 0<<24, 0, 1<<24};
-    const unsigned int w_two[] = {0, 2<<24, 0, 2<<24};
     const uint32x4_t s_one = vld1q_u32(w_one);
-    const uint32x4_t s_two = vld1q_u32(w_two);
 
     const size_t blockSize = 16;
     // const size_t neonBlockSize = 16;
@@ -511,9 +509,7 @@ inline size_t AdvancedProcessBlocks128_4x1_NEON(F1 func1, F4 func4,
     CRYPTOPP_ASSERT(length >= 16);
 
     const unsigned int w_one[] = {0, 0<<24, 0, 1<<24};
-    const unsigned int w_two[] = {0, 2<<24, 0, 2<<24};
     const uint32x4_t s_one = vld1q_u32(w_one);
-    const uint32x4_t s_two = vld1q_u32(w_two);
 
     const size_t blockSize = 16;
     // const size_t neonBlockSize = 16;
@@ -647,9 +643,7 @@ inline size_t AdvancedProcessBlocks128_6x2_NEON(F2 func2, F6 func6,
     CRYPTOPP_ASSERT(length >= 16);
 
     const unsigned int w_one[] = {0, 0<<24, 0, 1<<24};
-    const unsigned int w_two[] = {0, 2<<24, 0, 2<<24};
     const uint32x4_t s_one = vld1q_u32(w_one);
-    const uint32x4_t s_two = vld1q_u32(w_two);
 
     const size_t blockSize = 16;
     // const size_t neonBlockSize = 16;

aes_armv4.S

@@ -7,31 +7,6 @@
 @ details see http://www.openssl.org/~appro/cryptogams/.
 @ ====================================================================
 
-@ AES for ARMv4
-
-@ January 2007.
-@
-@ Code uses single 1K S-box and is >2 times faster than code generated
-@ by gcc-3.4.1. This is thanks to unique feature of ARMv4 ISA, which
-@ allows to merge logical or arithmetic operation with shift or rotate
-@ in one instruction and emit combined result every cycle. The module
-@ is endian-neutral. The performance is ~42 cycles/byte for 128-bit
-@ key [on single-issue Xscale PXA250 core].
-
-@ May 2007.
-@
-@ AES_set_[en|de]crypt_key is added.
-
-@ July 2010.
-@
-@ Rescheduling for dual-issue pipeline resulted in 12% improvement on
-@ Cortex A8 core and ~25 cycles per byte processed with 128-bit key.
-
-@ February 2011.
-@
-@ Profiler-assisted and platform-specific optimization resulted in 16%
-@ improvement on Cortex A8 core and ~21.5 cycles per byte.
-
 @ JW, JUL 2018: Begin defines from taken from arm_arch.h
 @               The defines were included through the header.
 
@@ -92,7 +67,6 @@
 @ JW, JUL 2018: End defines from taken from arm_arch.h
 @               Back to original Cryptogams code
 
-.text
 #if defined(__thumb2__) && !defined(__APPLE__)
 .syntax	unified
 .thumb
@@ -101,6 +75,8 @@
 #undef __thumb2__
 #endif
 
+.text
+
 .type	AES_Te,%object
 .align	5
 AES_Te:
@@ -1237,6 +1213,3 @@ _armv4_AES_decrypt:
 	sub	r10,r10,#1024
 	ldr	pc,[sp],#4		@ pop and return
 .size	_armv4_AES_decrypt,.-_armv4_AES_decrypt
-.byte	65,69,83,32,102,111,114,32,65,82,77,118,52,44,32,67,82,89,80,84,79,71,65,77,83,32,98,121,32,60,97,112,112,114,111,64,111,112,101,110,115,115,108,46,111,114,103,62,0
-.align	2
-.align	2

algparam.h

@@ -320,10 +320,12 @@ public:
 			if (m_throwIfNotUsed && !m_used)
 				throw ParameterNotUsed(m_name);
 		}
-#if !defined(CRYPTOPP_CXX17_EXCEPTIONS) && !defined(CRYPTOPP_UNCAUGHT_EXCEPTION_AVAILABLE)
+#if !defined(CRYPTOPP_UNCAUGHT_EXCEPTION_AVAILABLE)
+# if !defined(CRYPTOPP_CXX17_EXCEPTIONS)
 		catch(const Exception&)
 		{
 		}
+# endif
 #endif
 	}
 

aria_simd.cpp

@@ -24,7 +24,7 @@
 # include <arm_acle.h>
 #endif
 
-// Clang __m128i casts, http://bugs.llvm.org/show_bug.cgi?id=20670
+// Clang intrinsic casts, http://bugs.llvm.org/show_bug.cgi?id=20670
 #define M128_CAST(x) ((__m128i *)(void *)(x))
 #define CONST_M128_CAST(x) ((const __m128i *)(const void *)(x))
 

asn.cpp

@@ -394,14 +394,20 @@ void EncodedObjectFilter::Put(const byte *inString, size_t length)
 	}
 }
 
+BERGeneralDecoder::BERGeneralDecoder(BufferedTransformation &inQueue)
+	: m_inQueue(inQueue), m_length(0), m_finished(false)
+{
+	Init(DefaultTag);
+}
+
 BERGeneralDecoder::BERGeneralDecoder(BufferedTransformation &inQueue, byte asnTag)
-	: m_inQueue(inQueue), m_finished(false)
+	: m_inQueue(inQueue), m_length(0), m_finished(false)
 {
 	Init(asnTag);
 }
 
 BERGeneralDecoder::BERGeneralDecoder(BERGeneralDecoder &inQueue, byte asnTag)
-	: m_inQueue(inQueue), m_finished(false)
+	: m_inQueue(inQueue), m_length(0), m_finished(false)
 {
 	Init(asnTag);
 }
@@ -501,13 +507,18 @@ lword BERGeneralDecoder::ReduceLength(lword delta)
 	return delta;
 }
 
+DERGeneralEncoder::DERGeneralEncoder(BufferedTransformation &outQueue)
+	: m_outQueue(outQueue), m_asnTag(DefaultTag), m_finished(false)
+{
+}
+
 DERGeneralEncoder::DERGeneralEncoder(BufferedTransformation &outQueue, byte asnTag)
-	: ByteQueue(), m_outQueue(outQueue), m_asnTag(asnTag), m_finished(false)
+	: m_outQueue(outQueue), m_asnTag(asnTag), m_finished(false)
 {
 }
 
 DERGeneralEncoder::DERGeneralEncoder(DERGeneralEncoder &outQueue, byte asnTag)
-	: ByteQueue(), m_outQueue(outQueue), m_asnTag(asnTag), m_finished(false)
+	: m_outQueue(outQueue), m_asnTag(asnTag), m_finished(false)
 {
 }
 

asn.h

@@ -258,21 +258,99 @@ private:
 class CRYPTOPP_DLL BERGeneralDecoder : public Store
 {
 public:
+	/// \brief Default ASN.1 tag
+	enum {DefaultTag = SEQUENCE | CONSTRUCTED};
+
 	virtual ~BERGeneralDecoder();
 
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \details BERGeneralDecoder uses DefaultTag
+	explicit BERGeneralDecoder(BufferedTransformation &inQueue);
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \param asnTag ASN.1 tag
 	explicit BERGeneralDecoder(BufferedTransformation &inQueue, byte asnTag);
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \param asnTag ASN.1 tag
 	explicit BERGeneralDecoder(BERGeneralDecoder &inQueue, byte asnTag);
 
-	bool IsDefiniteLength() const {return m_definiteLength;}
+	/// \brief Determine length encoding
-	lword RemainingLength() const {CRYPTOPP_ASSERT(m_definiteLength); return m_length;}
+	/// \returns true if the ASN.1 object is definite length encoded, false otherwise
+	bool IsDefiniteLength() const {
+		return m_definiteLength;
+	}
+
+	/// \brief Determine remaining length
+	/// \returns number of octets that remain to be consumed
+	/// \details RemainingLength() is only valid if IsDefiniteLength()
+	///  returns true.
+	lword RemainingLength() const {
+		CRYPTOPP_ASSERT(m_definiteLength);
+		return IsDefiniteLength() ? m_length : 0;
+	}
+
+	/// \brief Determine end of stream
+	/// \returns true if all octets have been consumed, false otherwise
 	bool EndReached() const;
+
+	/// \brief Determine next octet
+	/// \returns next octet in the stream
+	/// \details PeekByte does not consume the octet.
+	/// \throws BERDecodeError if there are no octets remaining
 	byte PeekByte() const;
+
+	/// \brief Determine next octet
+	/// \details CheckByte reads the next byte in the stream and verifies
+	///  the octet matches b.
+	/// \throws BERDecodeError if the next octet is not b
 	void CheckByte(byte b);
 
+	/// \brief Transfer bytes to another BufferedTransformation
+	/// \param target the destination BufferedTransformation
+	/// \param transferBytes the number of bytes to transfer
+	/// \param channel the channel on which the transfer should occur
+	/// \param blocking specifies whether the object should block when
+	///  processing input
+	/// \return the number of bytes that remain in the transfer block
+	///  (i.e., bytes not transferred)
+	/// \details TransferTo2() removes bytes and moves
+	///  them to the destination. Transfer begins at the index position
+	///  in the current stream, and not from an absolute position in the
+	///  stream.
+	/// \details transferBytes is an \a IN and \a OUT parameter. When
+	///  the call is made, transferBytes is the requested size of the
+	///  transfer. When the call returns, transferBytes is the number
+	///  of bytes that were transferred.
 	size_t TransferTo2(BufferedTransformation &target, lword &transferBytes, const std::string &channel=DEFAULT_CHANNEL, bool blocking=true);
+
+	/// \brief Copy bytes to another BufferedTransformation
+	/// \param target the destination BufferedTransformation
+	/// \param begin the 0-based index of the first byte to copy in
+	///  the stream
+	/// \param end the 0-based index of the last byte to copy in
+	///  the stream
+	/// \param channel the channel on which the transfer should occur
+	/// \param blocking specifies whether the object should block when
+	///  processing input
+	/// \return the number of bytes that remain in the copy block
+	///  (i.e., bytes not copied)
+	/// \details CopyRangeTo2 copies bytes to the
+	///  destination. The bytes are not removed from this object. Copying
+	///  begins at the index position in the current stream, and not from
+	///  an absolute position in the stream.
+	/// \details begin is an \a IN and \a OUT parameter. When the call is
+	///  made, begin is the starting position of the copy. When the call
+	///  returns, begin is the position of the first byte that was \a not
+	///  copied (which may be different than end). begin can be used for
+	///  subsequent calls to CopyRangeTo2().
 	size_t CopyRangeTo2(BufferedTransformation &target, lword &begin, lword end=LWORD_MAX, const std::string &channel=DEFAULT_CHANNEL, bool blocking=true) const;
 
-	// call this to denote end of sequence
+	/// \brief Signals the end of messages to the object
+	/// \details Call this to denote end of sequence
 	void MessageEnd();
 
 protected:
@@ -291,12 +369,28 @@ private:
 class CRYPTOPP_DLL DERGeneralEncoder : public ByteQueue
 {
 public:
+	/// \brief Default ASN.1 tag
+	enum {DefaultTag = SEQUENCE | CONSTRUCTED};
+
 	virtual ~DERGeneralEncoder();
 
-	explicit DERGeneralEncoder(BufferedTransformation &outQueue, byte asnTag = SEQUENCE | CONSTRUCTED);
+	/// \brief Construct an ASN.1 encoder
-	explicit DERGeneralEncoder(DERGeneralEncoder &outQueue, byte asnTag = SEQUENCE | CONSTRUCTED);
+	/// \param outQueue output byte queue
+	/// \details DERGeneralEncoder uses DefaultTag
+	explicit DERGeneralEncoder(BufferedTransformation &outQueue);
 
-	// call this to denote end of sequence
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \param asnTag ASN.1 tag
+	explicit DERGeneralEncoder(BufferedTransformation &outQueue, byte asnTag);
+
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \param asnTag ASN.1 tag
+	explicit DERGeneralEncoder(DERGeneralEncoder &outQueue, byte asnTag);
+
+	/// \brief Signals the end of messages to the object
+	/// \details Call this to denote end of sequence
 	void MessageEnd();
 
 private:
@@ -309,9 +403,31 @@ private:
 class CRYPTOPP_DLL BERSequenceDecoder : public BERGeneralDecoder
 {
 public:
-	explicit BERSequenceDecoder(BufferedTransformation &inQueue, byte asnTag = SEQUENCE | CONSTRUCTED)
+	/// \brief Default ASN.1 tag
+	enum {DefaultTag = SEQUENCE | CONSTRUCTED};
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \details BERSequenceDecoder uses DefaultTag
+	explicit BERSequenceDecoder(BufferedTransformation &inQueue)
+		: BERGeneralDecoder(inQueue, DefaultTag) {}
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \param asnTag ASN.1 tag
+	explicit BERSequenceDecoder(BufferedTransformation &inQueue, byte asnTag)
 		: BERGeneralDecoder(inQueue, asnTag) {}
-	explicit BERSequenceDecoder(BERSequenceDecoder &inQueue, byte asnTag = SEQUENCE | CONSTRUCTED)
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \details BERSequenceDecoder uses DefaultTag
+	explicit BERSequenceDecoder(BERSequenceDecoder &inQueue)
+		: BERGeneralDecoder(inQueue, DefaultTag) {}
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \param asnTag ASN.1 tag
+	explicit BERSequenceDecoder(BERSequenceDecoder &inQueue, byte asnTag)
 		: BERGeneralDecoder(inQueue, asnTag) {}
 };
 
@@ -319,9 +435,31 @@ public:
 class CRYPTOPP_DLL DERSequenceEncoder : public DERGeneralEncoder
 {
 public:
-	explicit DERSequenceEncoder(BufferedTransformation &outQueue, byte asnTag = SEQUENCE | CONSTRUCTED)
+	/// \brief Default ASN.1 tag
+	enum {DefaultTag = SEQUENCE | CONSTRUCTED};
+
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \details DERSequenceEncoder uses DefaultTag
+	explicit DERSequenceEncoder(BufferedTransformation &outQueue)
+		: DERGeneralEncoder(outQueue, DefaultTag) {}
+
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \param asnTag ASN.1 tag
+	explicit DERSequenceEncoder(BufferedTransformation &outQueue, byte asnTag)
 		: DERGeneralEncoder(outQueue, asnTag) {}
-	explicit DERSequenceEncoder(DERSequenceEncoder &outQueue, byte asnTag = SEQUENCE | CONSTRUCTED)
+
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \details DERSequenceEncoder uses DefaultTag
+	explicit DERSequenceEncoder(DERSequenceEncoder &outQueue)
+		: DERGeneralEncoder(outQueue, DefaultTag) {}
+
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \param asnTag ASN.1 tag
+	explicit DERSequenceEncoder(DERSequenceEncoder &outQueue, byte asnTag)
 		: DERGeneralEncoder(outQueue, asnTag) {}
 };
 
@@ -329,9 +467,31 @@ public:
 class CRYPTOPP_DLL BERSetDecoder : public BERGeneralDecoder
 {
 public:
-	explicit BERSetDecoder(BufferedTransformation &inQueue, byte asnTag = SET | CONSTRUCTED)
+	/// \brief Default ASN.1 tag
+	enum {DefaultTag = SET | CONSTRUCTED};
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \details BERSetDecoder uses DefaultTag
+	explicit BERSetDecoder(BufferedTransformation &inQueue)
+		: BERGeneralDecoder(inQueue, DefaultTag) {}
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \param asnTag ASN.1 tag
+	explicit BERSetDecoder(BufferedTransformation &inQueue, byte asnTag)
 		: BERGeneralDecoder(inQueue, asnTag) {}
-	explicit BERSetDecoder(BERSetDecoder &inQueue, byte asnTag = SET | CONSTRUCTED)
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \details BERSetDecoder uses DefaultTag
+	explicit BERSetDecoder(BERSetDecoder &inQueue)
+		: BERGeneralDecoder(inQueue, DefaultTag) {}
+
+	/// \brief Construct an ASN.1 decoder
+	/// \param inQueue input byte queue
+	/// \param asnTag ASN.1 tag
+	explicit BERSetDecoder(BERSetDecoder &inQueue, byte asnTag)
 		: BERGeneralDecoder(inQueue, asnTag) {}
 };
 
@@ -339,9 +499,31 @@ public:
 class CRYPTOPP_DLL DERSetEncoder : public DERGeneralEncoder
 {
 public:
-	explicit DERSetEncoder(BufferedTransformation &outQueue, byte asnTag = SET | CONSTRUCTED)
+	/// \brief Default ASN.1 tag
+	enum {DefaultTag = SET | CONSTRUCTED};
+
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \details DERSetEncoder uses DefaultTag
+	explicit DERSetEncoder(BufferedTransformation &outQueue)
+		: DERGeneralEncoder(outQueue, DefaultTag) {}
+
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \param asnTag ASN.1 tag
+	explicit DERSetEncoder(BufferedTransformation &outQueue, byte asnTag)
 		: DERGeneralEncoder(outQueue, asnTag) {}
-	explicit DERSetEncoder(DERSetEncoder &outQueue, byte asnTag = SET | CONSTRUCTED)
+
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \details DERSetEncoder uses DefaultTag
+	explicit DERSetEncoder(DERSetEncoder &outQueue)
+		: DERGeneralEncoder(outQueue, DefaultTag) {}
+
+	/// \brief Construct an ASN.1 encoder
+	/// \param outQueue output byte queue
+	/// \param asnTag ASN.1 tag
+	explicit DERSetEncoder(DERSetEncoder &outQueue, byte asnTag)
 		: DERGeneralEncoder(outQueue, asnTag) {}
 };
 

basecode.h

@@ -36,7 +36,9 @@ public:
 		, m_outputBlockSize(0), m_bytePos(0), m_bitPos(0)
 	{
 		Detach(attachment);
-		IsolatedInitialize(MakeParameters(Name::EncodingLookupArray(), alphabet)
+		BaseN_Encoder::IsolatedInitialize(
+			MakeParameters
+				(Name::EncodingLookupArray(), alphabet)
 				(Name::Log2Base(), log2base)
 				(Name::Pad(), padding != -1)
 				(Name::PaddingByte(), byte(padding)));
@@ -78,7 +80,10 @@ public:
 		, m_outputBlockSize(0), m_bytePos(0), m_bitPos(0)
 	{
 		Detach(attachment);
-		IsolatedInitialize(MakeParameters(Name::DecodingLookupArray(), lookup)(Name::Log2Base(), log2base));
+		BaseN_Decoder::IsolatedInitialize(
+			MakeParameters
+				(Name::DecodingLookupArray(), lookup)
+				(Name::Log2Base(), log2base));
 	}
 
 	void IsolatedInitialize(const NameValuePairs &parameters);
@@ -121,7 +126,9 @@ public:
 		: m_groupSize(0), m_counter(0)
 	{
 		Detach(attachment);
-		IsolatedInitialize(MakeParameters(Name::GroupSize(), groupSize)
+		Grouper::IsolatedInitialize(
+			MakeParameters
+				(Name::GroupSize(), groupSize)
 				(Name::Separator(), ConstByteArrayParameter(separator))
 				(Name::Terminator(), ConstByteArrayParameter(terminator)));
 	}

bench.h

@@ -16,14 +16,48 @@ NAMESPACE_BEGIN(Test)
 
 // More granular control over benchmarks
 enum TestClass {
-	UnkeyedRNG=(1<<0),UnkeyedHash=(1<<1),UnkeyedOther=(1<<2),
+	/// \brief Random number generators
-	SharedKeyMAC=(1<<3),SharedKeyStream=(1<<4),SharedKeyBlock=(1<<5),SharedKeyOther=(1<<6),
+	UnkeyedRNG=(1<<0),
-	PublicKeyAgreement=(1<<7),PublicKeyEncryption=(1<<8),PublicKeySignature=(1<<9),PublicKeyOther=(1<<10),
+	/// \brief Message digests
+	UnkeyedHash=(1<<1),
+	/// \brief Other unkeyed algorithms
+	UnkeyedOther=(1<<2),
+
+	/// \brief Message authentication codes
+	SharedKeyMAC=(1<<3),
+	/// \brief Stream ciphers
+	SharedKeyStream=(1<<4),
+	/// \brief Block ciphers ciphers
+	SharedKeyBlock=(1<<5),
+	/// \brief Other shared key algorithms
+	SharedKeyOther=(1<<6),
+
+	/// \brief Key agreement algorithms over integers
+	PublicKeyAgreement=(1<<7),
+	/// \brief Encryption algorithms over integers
+	PublicKeyEncryption=(1<<8),
+	/// \brief Signature algorithms over integers
+	PublicKeySignature=(1<<9),
+	/// \brief Other public key algorithms over integers
+	PublicKeyOther=(1<<10),
+
+	/// \brief Key agreement algorithms over EC
+	PublicKeyAgreementEC=(1<<11),
+	/// \brief Encryption algorithms over EC
+	PublicKeyEncryptionEC=(1<<12),
+	/// \brief Signature algorithms over EC
+	PublicKeySignatureEC=(1<<13),
+	/// \brief Other public key algorithms over EC
+	PublicKeyOtherEC=(1<<14),
+
 	Unkeyed=UnkeyedRNG|UnkeyedHash|UnkeyedOther,
 	SharedKey=SharedKeyMAC|SharedKeyStream|SharedKeyBlock|SharedKeyOther,
 	PublicKey=PublicKeyAgreement|PublicKeyEncryption|PublicKeySignature|PublicKeyOther,
-	All=Unkeyed|SharedKey|PublicKey,
+	PublicKeyEC=PublicKeyAgreementEC|PublicKeyEncryptionEC|PublicKeySignatureEC|PublicKeyOtherEC,
-	TestFirst=(0), TestLast=(1<<11)
+
+	All=Unkeyed|SharedKey|PublicKey|PublicKeyEC,
+
+	TestFirst=(0), TestLast=(1<<15)
 };
 
 extern const double CLOCK_TICKS_PER_SECOND;
@@ -45,8 +79,10 @@ void Benchmark(Test::TestClass suites, double t, double hertz);
 void Benchmark1(double t, double hertz);
 // Shared key systems
 void Benchmark2(double t, double hertz);
-// Public key systems
+// Public key systems over integers
 void Benchmark3(double t, double hertz);
+// Public key systems over elliptic curves
+void Benchmark4(double t, double hertz);
 
 // These are defined in bench1.cpp
 extern void OutputResultKeying(double iterations, double timeTaken);

bench1.cpp

@@ -351,6 +351,8 @@ void BenchmarkWithCommand(int argc, const char* const argv[])
 
 	if (command == "b")  // All benchmarks
 		Benchmark(Test::All, runningTime, cpuFreq);
+	else if (command == "b4")  // Public key algorithms over EC
+		Test::Benchmark(Test::PublicKeyEC, runningTime, cpuFreq);
 	else if (command == "b3")  // Public key algorithms
 		Test::Benchmark(Test::PublicKey, runningTime, cpuFreq);
 	else if (command == "b2")  // Shared key algorithms
@@ -392,6 +394,13 @@ void Benchmark(Test::TestClass suites, double t, double hertz)
 		Benchmark3(t, hertz);
 	}
 
+	// Public key algorithms over EC
+	if (suites & Test::PublicKeyEC)
+	{
+		std::cout << "\n<BR>";
+		Benchmark4(t, hertz);
+	}
+
 	g_testEnd = ::time(NULLPTR);
 
 	std::ostringstream oss;

bench2.cpp

@@ -12,6 +12,12 @@
 #include "smartptr.h"
 #include "stdcpp.h"
 
+#include "vmac.h"
+#include "hmac.h"
+#include "ttmac.h"
+#include "cmac.h"
+#include "dmac.h"
+
 #if CRYPTOPP_MSC_VERSION
 # pragma warning(disable: 4355)
 #endif

bench3.cpp

@@ -365,21 +365,29 @@ void Benchmark3(double t, double hertz)
 		BenchMarkKeyAgreement<LUC_DH>("TestData/lucd1024.dat", "LUCDIF 1024", t);
 		BenchMarkKeyAgreement<MQV>("TestData/mqv1024.dat", "MQV 1024", t);
 		BenchMarkKeyAgreement<MQV>("TestData/mqv2048.dat", "MQV 2048", t);
-
-#if 0
-		BenchMarkKeyAgreement<ECHMQV160>("TestData/hmqv160.dat", "HMQV P-160", t);
-		BenchMarkKeyAgreement<ECHMQV256>("TestData/hmqv256.dat", "HMQV P-256", t);
-		BenchMarkKeyAgreement<ECHMQV384>("TestData/hmqv384.dat", "HMQV P-384", t);
-		BenchMarkKeyAgreement<ECHMQV512>("TestData/hmqv512.dat", "HMQV P-512", t);
-
-		BenchMarkKeyAgreement<ECFHMQV160>("TestData/fhmqv160.dat", "FHMQV P-160", t);
-		BenchMarkKeyAgreement<ECFHMQV256>("TestData/fhmqv256.dat", "FHMQV P-256", t);
-		BenchMarkKeyAgreement<ECFHMQV384>("TestData/fhmqv384.dat", "FHMQV P-384", t);
-		BenchMarkKeyAgreement<ECFHMQV512>("TestData/fhmqv512.dat", "FHMQV P-512", t);
-#endif
 	}
 
-	std::cout << "\n<TBODY style=\"background: yellow;\">";
+	std::cout << "\n</TABLE>" << std::endl;
+}
+
+void Benchmark4(double t, double hertz)
+{
+	g_allocatedTime = t;
+	g_hertz = hertz;
+
+	const char *mco;
+	if (g_hertz > 1.0f)
+		mco = "<TH>Megacycles/Operation";
+	else
+		mco = "";
+
+	std::cout << "\n<TABLE>";
+	std::cout << "\n<COLGROUP><COL style=\"text-align: left;\"><COL style=";
+	std::cout << "\"text-align: right;\"><COL style=\"text-align: right;\">";
+	std::cout << "\n<THEAD style=\"background: #F0F0F0\">";
+	std::cout << "\n<TR><TH>Operation<TH>Milliseconds/Operation" << mco;
+
+	std::cout << "\n<TBODY style=\"background: white;\">";
 	{
 		ed25519::Signer sign(Test::GlobalRNG());
 		ed25519::Verifier verify(sign);
@@ -391,7 +399,27 @@ void Benchmark3(double t, double hertz)
 		BenchMarkAgreement("x25519", agree, t);
 	}
 
-	std::cout << "\n<TBODY style=\"background: white;\">";
+#if 0
+	std::cout << "\n<TBODY style=\"background: yellow;\">";
+	{
+		BenchMarkKeyAgreement<ECMQV160>("TestData/mqv160.dat", "MQV P-160", t);
+		BenchMarkKeyAgreement<ECMQV256>("TestData/mqv256.dat", "MQV P-256", t);
+		BenchMarkKeyAgreement<ECMQV384>("TestData/mqv384.dat", "MQV P-384", t);
+		BenchMarkKeyAgreement<ECMQV512>("TestData/mqv512.dat", "MQV P-521", t);
+
+		BenchMarkKeyAgreement<ECHMQV160>("TestData/hmqv160.dat", "HMQV P-160", t);
+		BenchMarkKeyAgreement<ECHMQV256>("TestData/hmqv256.dat", "HMQV P-256", t);
+		BenchMarkKeyAgreement<ECHMQV384>("TestData/hmqv384.dat", "HMQV P-384", t);
+		BenchMarkKeyAgreement<ECHMQV512>("TestData/hmqv512.dat", "HMQV P-521", t);
+
+		BenchMarkKeyAgreement<ECFHMQV160>("TestData/fhmqv160.dat", "FHMQV P-160", t);
+		BenchMarkKeyAgreement<ECFHMQV256>("TestData/fhmqv256.dat", "FHMQV P-256", t);
+		BenchMarkKeyAgreement<ECFHMQV384>("TestData/fhmqv384.dat", "FHMQV P-384", t);
+		BenchMarkKeyAgreement<ECFHMQV512>("TestData/fhmqv512.dat", "FHMQV P-521", t);
+	}
+#endif
+
+	std::cout << "\n<TBODY style=\"background: yellow;\">";
 	{
 		ECIES<ECP>::Decryptor cpriv(Test::GlobalRNG(), ASN1::secp256k1());
 		ECIES<ECP>::Encryptor cpub(cpriv);
@@ -418,7 +446,7 @@ void Benchmark3(double t, double hertz)
 		BenchMarkAgreement("ECMQVC over GF(p) 256", ecmqvc, t);
 	}
 
-	std::cout << "\n<TBODY style=\"background: yellow;\">";
+	std::cout << "\n<TBODY style=\"background: white;\">";
 	{
 		ECIES<EC2N>::Decryptor cpriv(Test::GlobalRNG(), ASN1::sect233r1());
 		ECIES<EC2N>::Encryptor cpub(cpriv);

blake2.cpp

@@ -38,24 +38,34 @@
 // https://github.com/weidai11/cryptopp/issues/743
 #if defined(__xlC__) && (__xlC__ < 0x0d01)
 # define CRYPTOPP_DISABLE_ALTIVEC 1
-# define CRYPTOPP_POWER7_ALTIVEC 1
 # undef CRYPTOPP_POWER7_AVAILABLE
+# undef CRYPTOPP_POWER8_AVAILABLE
 # undef CRYPTOPP_ALTIVEC_AVAILABLE
 #endif
 
+// Can't use GetAlignmentOf<word64>() because of C++11 and constexpr
+// Can use 'const unsigned int' because of MSVC 2013
+#if (CRYPTOPP_BOOL_X86 || CRYPTOPP_BOOL_X32 || CRYPTOPP_BOOL_X64)
+# define ALIGN_SPEC32 16
+# define ALIGN_SPEC64 16
+#else
+# define ALIGN_SPEC32 4
+# define ALIGN_SPEC64 8
+#endif
+
 NAMESPACE_BEGIN(CryptoPP)
 
 // Export the tables to the SIMD files
 extern const word32 BLAKE2S_IV[8];
 extern const word64 BLAKE2B_IV[8];
 
-CRYPTOPP_ALIGN_DATA(16)
+CRYPTOPP_ALIGN_DATA(ALIGN_SPEC32)
 const word32 BLAKE2S_IV[8] = {
     0x6A09E667UL, 0xBB67AE85UL, 0x3C6EF372UL, 0xA54FF53AUL,
     0x510E527FUL, 0x9B05688CUL, 0x1F83D9ABUL, 0x5BE0CD19UL
 };
 
-CRYPTOPP_ALIGN_DATA(16)
+CRYPTOPP_ALIGN_DATA(ALIGN_SPEC64)
 const word64 BLAKE2B_IV[8] = {
     W64LIT(0x6a09e667f3bcc908), W64LIT(0xbb67ae8584caa73b),
     W64LIT(0x3c6ef372fe94f82b), W64LIT(0xa54ff53a5f1d36f1),
@@ -72,7 +82,7 @@ using CryptoPP::word32;
 using CryptoPP::word64;
 using CryptoPP::rotrConstant;
 
-CRYPTOPP_ALIGN_DATA(16)
+CRYPTOPP_ALIGN_DATA(ALIGN_SPEC32)
 const byte BLAKE2S_SIGMA[10][16] = {
     {  0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15 },
     { 14, 10,  4,  8,  9, 15, 13,  6,  1, 12,  0,  2, 11,  7,  5,  3 },
@@ -86,7 +96,7 @@ const byte BLAKE2S_SIGMA[10][16] = {
     { 10,  2,  8,  4,  7,  6,  1,  5, 15, 11,  9, 14,  3, 12, 13 , 0 },
 };
 
-CRYPTOPP_ALIGN_DATA(16)
+CRYPTOPP_ALIGN_DATA(ALIGN_SPEC32)
 const byte BLAKE2B_SIGMA[12][16] = {
     {  0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15 },
     { 14, 10,  4,  8,  9, 15, 13,  6,  1, 12,  0,  2, 11,  7,  5,  3 },
@@ -171,8 +181,8 @@ extern void BLAKE2_Compress32_NEON(const byte* input, BLAKE2s_State& state);
 extern void BLAKE2_Compress64_NEON(const byte* input, BLAKE2b_State& state);
 #endif
 
-#if CRYPTOPP_POWER7_AVAILABLE
+#if CRYPTOPP_POWER8_AVAILABLE
-extern void BLAKE2_Compress32_POWER7(const byte* input, BLAKE2s_State& state);
+extern void BLAKE2_Compress32_POWER8(const byte* input, BLAKE2s_State& state);
 #elif CRYPTOPP_ALTIVEC_AVAILABLE
 extern void BLAKE2_Compress32_ALTIVEC(const byte* input, BLAKE2s_State& state);
 #endif
@@ -233,8 +243,8 @@ unsigned int BLAKE2s::OptimalDataAlignment() const
         return 4;
     else
 #endif
-#if (CRYPTOPP_POWER7_AVAILABLE)
+#if (CRYPTOPP_POWER8_AVAILABLE)
-    if (HasPower7())
+    if (HasPower8())
         return 16;
     else
 #elif (CRYPTOPP_ALTIVEC_AVAILABLE)
@@ -257,9 +267,9 @@ std::string BLAKE2s::AlgorithmProvider() const
         return "NEON";
     else
 #endif
-#if (CRYPTOPP_POWER7_AVAILABLE)
+#if (CRYPTOPP_POWER8_AVAILABLE)
-    if (HasPower7())
+    if (HasPower8())
-        return "Power7";
+        return "Power8";
     else
 #elif (CRYPTOPP_ALTIVEC_AVAILABLE)
     if (HasAltivec())
@@ -343,6 +353,26 @@ BLAKE2b::BLAKE2b(bool treeMode, unsigned int digestSize)
         (Name::TreeMode(), treeMode));
 }
 
+BLAKE2s::BLAKE2s(unsigned int digestSize)
+    : m_digestSize(digestSize), m_keyLength(0), m_treeMode(false)
+{
+    CRYPTOPP_ASSERT(digestSize <= DIGESTSIZE);
+
+    UncheckedSetKey(NULLPTR, 0, MakeParameters
+        (Name::DigestSize(), (int)digestSize)
+        (Name::TreeMode(), false));
+}
+
+BLAKE2b::BLAKE2b(unsigned int digestSize)
+    : m_digestSize(digestSize), m_keyLength(0), m_treeMode(false)
+{
+    CRYPTOPP_ASSERT(digestSize <= DIGESTSIZE);
+
+    UncheckedSetKey(NULLPTR, 0, MakeParameters
+        (Name::DigestSize(), (int)digestSize)
+        (Name::TreeMode(), false));
+}
+
 BLAKE2s::BLAKE2s(const byte *key, size_t keyLength, const byte* salt, size_t saltLength,
     const byte* personalization, size_t personalizationLength, bool treeMode, unsigned int digestSize)
     : m_digestSize(digestSize), m_keyLength(static_cast<unsigned int>(keyLength)), m_treeMode(treeMode)
@@ -670,10 +700,10 @@ void BLAKE2s::Compress(const byte *input)
         return BLAKE2_Compress32_NEON(input, m_state);
     }
 #endif
-#if CRYPTOPP_POWER7_AVAILABLE
+#if CRYPTOPP_POWER8_AVAILABLE
-    if(HasPower7())
+    if(HasPower8())
     {
-        return BLAKE2_Compress32_POWER7(input, m_state);
+        return BLAKE2_Compress32_POWER8(input, m_state);
     }
 #elif CRYPTOPP_ALTIVEC_AVAILABLE
     if(HasAltivec())

blake2.h

@@ -256,17 +256,25 @@ public:
     /// \brief Construct a BLAKE2s hash
     /// \param digestSize the digest size, in bytes
     /// \param treeMode flag indicating tree mode
+    /// \since Crypto++ 5.6.4
     BLAKE2s(bool treeMode=false, unsigned int digestSize = DIGESTSIZE);
 
+    /// \brief Construct a BLAKE2s hash
+    /// \param digestSize the digest size, in bytes
+    /// \details treeMode flag is set to false
+    /// \since Crypto++ 8.2
+    BLAKE2s(unsigned int digestSize);
+
     /// \brief Construct a BLAKE2s hash
     /// \param key a byte array used to key the cipher
     /// \param keyLength the size of the byte array
     /// \param salt a byte array used as salt
     /// \param saltLength the size of the byte array
-    /// \param personalization a byte array used as prsonalization string
+    /// \param personalization a byte array used as personalization string
     /// \param personalizationLength the size of the byte array
     /// \param treeMode flag indicating tree mode
     /// \param digestSize the digest size, in bytes
+    /// \since Crypto++ 5.6.4
     BLAKE2s(const byte *key, size_t keyLength, const byte* salt = NULLPTR, size_t saltLength = 0,
         const byte* personalization = NULLPTR, size_t personalizationLength = 0,
         bool treeMode=false, unsigned int digestSize = DIGESTSIZE);
@@ -278,6 +286,7 @@ public:
     /// Message Authentication Code (MAC)</A>. For example, "BLAKE2b-512" and "BLAKE2s-256".
     std::string AlgorithmName() const {return std::string(BLAKE2s_Info::StaticAlgorithmName()) + "-" + IntToString(DigestSize()*8);}
 
+    unsigned int BlockSize() const {return BLOCKSIZE;}
     unsigned int DigestSize() const {return m_digestSize;}
     unsigned int OptimalDataAlignment() const;
 
@@ -354,17 +363,25 @@ public:
     /// \brief Construct a BLAKE2b hash
     /// \param digestSize the digest size, in bytes
     /// \param treeMode flag indicating tree mode
+    /// \since Crypto++ 5.6.4
     BLAKE2b(bool treeMode=false, unsigned int digestSize = DIGESTSIZE);
 
+    /// \brief Construct a BLAKE2s hash
+    /// \param digestSize the digest size, in bytes
+    /// \details treeMode flag is set to false
+    /// \since Crypto++ 8.2
+    BLAKE2b(unsigned int digestSize);
+
     /// \brief Construct a BLAKE2b hash
     /// \param key a byte array used to key the cipher
     /// \param keyLength the size of the byte array
     /// \param salt a byte array used as salt
     /// \param saltLength the size of the byte array
-    /// \param personalization a byte array used as prsonalization string
+    /// \param personalization a byte array used as personalization string
     /// \param personalizationLength the size of the byte array
     /// \param treeMode flag indicating tree mode
     /// \param digestSize the digest size, in bytes
+    /// \since Crypto++ 5.6.4
     BLAKE2b(const byte *key, size_t keyLength, const byte* salt = NULLPTR, size_t saltLength = 0,
         const byte* personalization = NULLPTR, size_t personalizationLength = 0,
         bool treeMode=false, unsigned int digestSize = DIGESTSIZE);
@@ -376,6 +393,7 @@ public:
     /// Message Authentication Code (MAC)</A>. For example, "BLAKE2b-512" and "BLAKE2s-256".
     std::string AlgorithmName() const {return std::string(BLAKE2b_Info::StaticAlgorithmName()) + "-" + IntToString(DigestSize()*8);}
 
+    unsigned int BlockSize() const {return BLOCKSIZE;}
     unsigned int DigestSize() const {return m_digestSize;}
     unsigned int OptimalDataAlignment() const;
 

blake2b_simd.cpp

@@ -28,10 +28,17 @@
 // https://github.com/weidai11/cryptopp/issues/743
 #if defined(__xlC__) && (__xlC__ < 0x0d01)
 # define CRYPTOPP_DISABLE_ALTIVEC 1
-# undef CRYPTOPP_POWER7_AVAILABLE
+# undef CRYPTOPP_POWER8_AVAILABLE
 # undef CRYPTOPP_ALTIVEC_AVAILABLE
 #endif
 
+#if defined(__XOP__)
+# include <ammintrin.h>
+# if defined(__GNUC__)
+#  include <x86intrin.h>
+# endif
+#endif
+
 #if (CRYPTOPP_SSE41_AVAILABLE)
 # include <emmintrin.h>
 # include <tmmintrin.h>

blake2s_simd.cpp

@@ -8,9 +8,9 @@
 //    appropriate instructions sets in some build configurations.
 
 // The BLAKE2b and BLAKE2s numbers are consistent with the BLAKE2 team's
-// numbers. However, we have an Altivec/POWER7 implementation of BLAKE2s,
+// numbers. However, we have an Altivec/POWER8 implementation of BLAKE2s,
 // and a POWER8 implementation of BLAKE2b (BLAKE2 is missing them). The
-// Altivec/POWER7 code is about 2x faster than C++ when using GCC 5.0 or
+// Altivec/POWER8 code is about 2x faster than C++ when using GCC 5.0 or
 // above. The POWER8 code is about 2.5x faster than C++ when using GCC 5.0
 // or above. If you use GCC 4.0 (PowerMac) or GCC 4.8 (GCC Compile Farm)
 // then the PowerPC code will be slower than C++. Be sure to use GCC 5.0
@@ -38,11 +38,17 @@
 // https://github.com/weidai11/cryptopp/issues/743
 #if defined(__xlC__) && (__xlC__ < 0x0d01)
 # define CRYPTOPP_DISABLE_ALTIVEC 1
-# define CRYPTOPP_POWER7_ALTIVEC 1
+# undef CRYPTOPP_POWER8_AVAILABLE
-# undef CRYPTOPP_POWER7_AVAILABLE
 # undef CRYPTOPP_ALTIVEC_AVAILABLE
 #endif
 
+#if defined(__XOP__)
+# include <ammintrin.h>
+# if defined(__GNUC__)
+#  include <x86intrin.h>
+# endif
+#endif
+
 #if (CRYPTOPP_SSE41_AVAILABLE)
 # include <emmintrin.h>
 # include <tmmintrin.h>
@@ -692,7 +698,7 @@ void BLAKE2_Compress32_NEON(const byte* input, BLAKE2s_State& state)
 }
 #endif  // CRYPTOPP_ARM_NEON_AVAILABLE
 
-#if (CRYPTOPP_POWER7_AVAILABLE || CRYPTOPP_ALTIVEC_AVAILABLE)
+#if (CRYPTOPP_POWER8_AVAILABLE || CRYPTOPP_ALTIVEC_AVAILABLE)
 
 inline uint32x4_p VecLoad32(const void* p)
 {
@@ -838,7 +844,7 @@ inline uint32x4_p VectorSet32(const uint32x4_p a, const uint32x4_p b,
     const uint32x4_p t0 = VectorSet32<W,X>(a, b);
     const uint32x4_p t1 = VectorSet32<Y,Z>(c, d);
 
-    // Power7 follows SSE2's implementation, and this is _mm_set_epi32.
+    // PowerPC follows SSE2's implementation, and this is _mm_set_epi32.
     const uint8x16_p mask = {20,21,22,23, 16,17,18,19, 4,5,6,7, 0,1,2,3};
     return VecPermute(t0, t1, mask);
 }
@@ -863,10 +869,10 @@ uint32x4_p VectorSet32<3,1,3,1>(const uint32x4_p a, const uint32x4_p b,
     return VecPermute(a, c, mask);
 }
 
-// BLAKE2_Compress32_CORE will use either POWER7 or ALTIVEC,
+// BLAKE2_Compress32_CORE will use either POWER8 or ALTIVEC,
 // depending on the flags used to compile this source file. The
 // abstractions are handled in VecLoad, VecStore and friends. In
-// the future we may to provide both POWER7 or ALTIVEC at the same
+// the future we may provide both POWER8 or ALTIVEC at the same
 // time to better support distros.
 void BLAKE2_Compress32_CORE(const byte* input, BLAKE2s_State& state)
 {
@@ -1015,11 +1021,11 @@ void BLAKE2_Compress32_CORE(const byte* input, BLAKE2s_State& state)
     VecStore32LE(state.h()+0, VecXor(ff0, VecXor(row1, row3)));
     VecStore32LE(state.h()+4, VecXor(ff1, VecXor(row2, row4)));
 }
-#endif  // CRYPTOPP_POWER7_AVAILABLE || CRYPTOPP_ALTIVEC_AVAILABLE
+#endif  // CRYPTOPP_POWER8_AVAILABLE || CRYPTOPP_ALTIVEC_AVAILABLE
 
-#if (CRYPTOPP_POWER7_AVAILABLE)
+#if (CRYPTOPP_POWER8_AVAILABLE)
 
-void BLAKE2_Compress32_POWER7(const byte* input, BLAKE2s_State& state)
+void BLAKE2_Compress32_POWER8(const byte* input, BLAKE2s_State& state)
 {
     BLAKE2_Compress32_CORE(input, state);
 }

chacha.cpp

@@ -28,8 +28,8 @@ extern void ChaCha_OperateKeystream_AVX2(const word32 *state, const byte* input,
 extern void ChaCha_OperateKeystream_SSE2(const word32 *state, const byte* input, byte *output, unsigned int rounds);
 #endif
 
-#if (CRYPTOPP_POWER7_AVAILABLE)
+#if (CRYPTOPP_POWER8_AVAILABLE)
-extern void ChaCha_OperateKeystream_POWER7(const word32 *state, const byte* input, byte *output, unsigned int rounds);
+extern void ChaCha_OperateKeystream_POWER8(const word32 *state, const byte* input, byte *output, unsigned int rounds);
 #elif (CRYPTOPP_ALTIVEC_AVAILABLE)
 extern void ChaCha_OperateKeystream_ALTIVEC(const word32 *state, const byte* input, byte *output, unsigned int rounds);
 #endif
@@ -153,13 +153,13 @@ void ChaCha_OperateKeystream(KeystreamOperation operation,
         }
 #endif
 
-#if (CRYPTOPP_POWER7_AVAILABLE)
+#if (CRYPTOPP_POWER8_AVAILABLE)
-        if (HasPower7())
+        if (HasPower8())
         {
             while (iterationCount >= 4 && MultiBlockSafe(state[12], 4))
             {
                 const bool xorInput = (operation & INPUT_NULL) != INPUT_NULL;
-                ChaCha_OperateKeystream_POWER7(state, xorInput ? input : NULLPTR, output, rounds);
+                ChaCha_OperateKeystream_POWER8(state, xorInput ? input : NULLPTR, output, rounds);
 
                 // MultiBlockSafe avoids overflow on the counter words
                 state[12] += 4;
@@ -267,9 +267,9 @@ std::string ChaCha_AlgorithmProvider()
         return "NEON";
     else
 #endif
-#if (CRYPTOPP_POWER7_AVAILABLE)
+#if (CRYPTOPP_POWER8_AVAILABLE)
-    if (HasPower7())
+    if (HasPower8())
-        return "Power7";
+        return "Power8";
     else
 #elif (CRYPTOPP_ALTIVEC_AVAILABLE)
     if (HasAltivec())

chacha_avx.cpp

@@ -91,14 +91,14 @@ NAMESPACE_BEGIN(CryptoPP)
 
 void ChaCha_OperateKeystream_AVX2(const word32 *state, const byte* input, byte *output, unsigned int rounds)
 {
-    MAYBE_CONST __m128i* state_mm = (MAYBE_CONST __m128i*)(state);
+    const __m256i state0 = _mm256_broadcastsi128_si256(
-    MAYBE_CONST __m256i* input_mm = (MAYBE_CONST __m256i*)(input);
+        _mm_loadu_si128(reinterpret_cast<const __m128i*>(state+0*4)));
-    __m256i* output_mm = reinterpret_cast<__m256i*>(output);
+    const __m256i state1 = _mm256_broadcastsi128_si256(
-
+        _mm_loadu_si128(reinterpret_cast<const __m128i*>(state+1*4)));
-    const __m256i state0 = _mm256_broadcastsi128_si256(_mm_loadu_si128(state_mm + 0));
+    const __m256i state2 = _mm256_broadcastsi128_si256(
-    const __m256i state1 = _mm256_broadcastsi128_si256(_mm_loadu_si128(state_mm + 1));
+        _mm_loadu_si128(reinterpret_cast<const __m128i*>(state+2*4)));
-    const __m256i state2 = _mm256_broadcastsi128_si256(_mm_loadu_si128(state_mm + 2));
+    const __m256i state3 = _mm256_broadcastsi128_si256(
-    const __m256i state3 = _mm256_broadcastsi128_si256(_mm_loadu_si128(state_mm + 3));
+        _mm_loadu_si128(reinterpret_cast<const __m128i*>(state+3*4)));
 
     const __m256i CTR0 = _mm256_set_epi32(0, 0, 0, 0, 0, 0, 0, 4);
     const __m256i CTR1 = _mm256_set_epi32(0, 0, 0, 1, 0, 0, 0, 5);
@@ -304,80 +304,112 @@ void ChaCha_OperateKeystream_AVX2(const word32 *state, const byte* input, byte *
     X3_3 = _mm256_add_epi32(X3_3, state3);
     X3_3 = _mm256_add_epi64(X3_3, CTR3);
 
-    if (input_mm)
+    if (input)
     {
-        _mm256_storeu_si256(output_mm + 0, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 0),
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+0*32),
-            _mm256_permute2x128_si256(X0_0, X0_1, 1 + (3 << 4))));
+            _mm256_xor_si256(_mm256_permute2x128_si256(X0_0, X0_1, 1 + (3 << 4)),
-        _mm256_storeu_si256(output_mm + 1, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 1),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+0*32))));
-            _mm256_permute2x128_si256(X0_2, X0_3, 1 + (3 << 4))));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+1*32),
-        _mm256_storeu_si256(output_mm + 2, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 2),
+            _mm256_xor_si256(_mm256_permute2x128_si256(X0_2, X0_3, 1 + (3 << 4)),
-            _mm256_permute2x128_si256(X1_0, X1_1, 1 + (3 << 4))));
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+1*32))));
-        _mm256_storeu_si256(output_mm + 3, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 3),
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+2*32),
-            _mm256_permute2x128_si256(X1_2, X1_3, 1 + (3 << 4))));
+            _mm256_xor_si256(_mm256_permute2x128_si256(X1_0, X1_1, 1 + (3 << 4)),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+2*32))));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+3*32),
+            _mm256_xor_si256(_mm256_permute2x128_si256(X1_2, X1_3, 1 + (3 << 4)),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+3*32))));
     }
     else
     {
-        _mm256_storeu_si256(output_mm + 0, _mm256_permute2x128_si256(X0_0, X0_1, 1 + (3 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+0*32),
-        _mm256_storeu_si256(output_mm + 1, _mm256_permute2x128_si256(X0_2, X0_3, 1 + (3 << 4)));
+            _mm256_permute2x128_si256(X0_0, X0_1, 1 + (3 << 4)));
-        _mm256_storeu_si256(output_mm + 2, _mm256_permute2x128_si256(X1_0, X1_1, 1 + (3 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+1*32),
-        _mm256_storeu_si256(output_mm + 3, _mm256_permute2x128_si256(X1_2, X1_3, 1 + (3 << 4)));
+            _mm256_permute2x128_si256(X0_2, X0_3, 1 + (3 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+2*32),
+            _mm256_permute2x128_si256(X1_0, X1_1, 1 + (3 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+3*32),
+            _mm256_permute2x128_si256(X1_2, X1_3, 1 + (3 << 4)));
     }
 
-    if (input_mm)
+    if (input)
     {
-        _mm256_storeu_si256(output_mm + 4, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 4),
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+4*32),
-            _mm256_permute2x128_si256(X2_0, X2_1, 1 + (3 << 4))));
+            _mm256_xor_si256(_mm256_permute2x128_si256(X2_0, X2_1, 1 + (3 << 4)),
-        _mm256_storeu_si256(output_mm + 5, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 5),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+4*32))));
-            _mm256_permute2x128_si256(X2_2, X2_3, 1 + (3 << 4))));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+5*32),
-        _mm256_storeu_si256(output_mm + 6, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 6),
+            _mm256_xor_si256(_mm256_permute2x128_si256(X2_2, X2_3, 1 + (3 << 4)),
-            _mm256_permute2x128_si256(X3_0, X3_1, 1 + (3 << 4))));
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+5*32))));
-        _mm256_storeu_si256(output_mm + 7, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 7),
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+6*32),
-            _mm256_permute2x128_si256(X3_2, X3_3, 1 + (3 << 4))));
+            _mm256_xor_si256(_mm256_permute2x128_si256(X3_0, X3_1, 1 + (3 << 4)),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+6*32))));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+7*32),
+            _mm256_xor_si256(_mm256_permute2x128_si256(X3_2, X3_3, 1 + (3 << 4)),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+7*32))));
     }
     else
     {
-        _mm256_storeu_si256(output_mm + 4, _mm256_permute2x128_si256(X2_0, X2_1, 1 + (3 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+4*32),
-        _mm256_storeu_si256(output_mm + 5, _mm256_permute2x128_si256(X2_2, X2_3, 1 + (3 << 4)));
+            _mm256_permute2x128_si256(X2_0, X2_1, 1 + (3 << 4)));
-        _mm256_storeu_si256(output_mm + 6, _mm256_permute2x128_si256(X3_0, X3_1, 1 + (3 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+5*32),
-        _mm256_storeu_si256(output_mm + 7, _mm256_permute2x128_si256(X3_2, X3_3, 1 + (3 << 4)));
+            _mm256_permute2x128_si256(X2_2, X2_3, 1 + (3 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+6*32),
+            _mm256_permute2x128_si256(X3_0, X3_1, 1 + (3 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+7*32),
+            _mm256_permute2x128_si256(X3_2, X3_3, 1 + (3 << 4)));
     }
 
-    if (input_mm)
+    if (input)
     {
-        _mm256_storeu_si256(output_mm +  8, _mm256_xor_si256(_mm256_loadu_si256(input_mm +  8),
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+ 8*32),
-            _mm256_permute2x128_si256(X0_0, X0_1, 0 + (2 << 4))));
+            _mm256_xor_si256(_mm256_permute2x128_si256(X0_0, X0_1, 0 + (2 << 4)),
-        _mm256_storeu_si256(output_mm +  9, _mm256_xor_si256(_mm256_loadu_si256(input_mm +  9),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+8*32))));
-            _mm256_permute2x128_si256(X0_2, X0_3, 0 + (2 << 4))));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+ 9*32),
-        _mm256_storeu_si256(output_mm + 10, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 10),
+            _mm256_xor_si256(_mm256_permute2x128_si256(X0_2, X0_3, 0 + (2 << 4)),
-            _mm256_permute2x128_si256(X1_0, X1_1, 0 + (2 << 4))));
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+9*32))));
-        _mm256_storeu_si256(output_mm + 11, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 11),
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+10*32),
-            _mm256_permute2x128_si256(X1_2, X1_3, 0 + (2 << 4))));
+            _mm256_xor_si256(_mm256_permute2x128_si256(X1_0, X1_1, 0 + (2 << 4)),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+10*32))));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+11*32),
+            _mm256_xor_si256(_mm256_permute2x128_si256(X1_2, X1_3, 0 + (2 << 4)),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+11*32))));
     }
     else
     {
-        _mm256_storeu_si256(output_mm +  8, _mm256_permute2x128_si256(X0_0, X0_1, 0 + (2 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+ 8*32),
-        _mm256_storeu_si256(output_mm +  9, _mm256_permute2x128_si256(X0_2, X0_3, 0 + (2 << 4)));
+            _mm256_permute2x128_si256(X0_0, X0_1, 0 + (2 << 4)));
-        _mm256_storeu_si256(output_mm + 10, _mm256_permute2x128_si256(X1_0, X1_1, 0 + (2 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+ 9*32),
-        _mm256_storeu_si256(output_mm + 11, _mm256_permute2x128_si256(X1_2, X1_3, 0 + (2 << 4)));
+            _mm256_permute2x128_si256(X0_2, X0_3, 0 + (2 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+10*32),
+            _mm256_permute2x128_si256(X1_0, X1_1, 0 + (2 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+11*32),
+            _mm256_permute2x128_si256(X1_2, X1_3, 0 + (2 << 4)));
     }
 
-    if (input_mm)
+    if (input)
     {
-        _mm256_storeu_si256(output_mm + 12, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 12),
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+12*32),
-            _mm256_permute2x128_si256(X2_0, X2_1, 0 + (2 << 4))));
+            _mm256_xor_si256(_mm256_permute2x128_si256(X2_0, X2_1, 0 + (2 << 4)),
-        _mm256_storeu_si256(output_mm + 13, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 13),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+12*32))));
-            _mm256_permute2x128_si256(X2_2, X2_3, 0 + (2 << 4))));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+13*32),
-        _mm256_storeu_si256(output_mm + 14, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 14),
+            _mm256_xor_si256(_mm256_permute2x128_si256(X2_2, X2_3, 0 + (2 << 4)),
-            _mm256_permute2x128_si256(X3_0, X3_1, 0 + (2 << 4))));
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+13*32))));
-        _mm256_storeu_si256(output_mm + 15, _mm256_xor_si256(_mm256_loadu_si256(input_mm + 15),
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+14*32),
-            _mm256_permute2x128_si256(X3_2, X3_3, 0 + (2 << 4))));
+            _mm256_xor_si256(_mm256_permute2x128_si256(X3_0, X3_1, 0 + (2 << 4)),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+14*32))));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+15*32),
+            _mm256_xor_si256(_mm256_permute2x128_si256(X3_2, X3_3, 0 + (2 << 4)),
+            _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input+15*32))));
     }
     else
     {
-        _mm256_storeu_si256(output_mm + 12, _mm256_permute2x128_si256(X2_0, X2_1, 0 + (2 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+12*32),
-        _mm256_storeu_si256(output_mm + 13, _mm256_permute2x128_si256(X2_2, X2_3, 0 + (2 << 4)));
+            _mm256_permute2x128_si256(X2_0, X2_1, 0 + (2 << 4)));
-        _mm256_storeu_si256(output_mm + 14, _mm256_permute2x128_si256(X3_0, X3_1, 0 + (2 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+13*32),
-        _mm256_storeu_si256(output_mm + 15, _mm256_permute2x128_si256(X3_2, X3_3, 0 + (2 << 4)));
+            _mm256_permute2x128_si256(X2_2, X2_3, 0 + (2 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+14*32),
+            _mm256_permute2x128_si256(X3_0, X3_1, 0 + (2 << 4)));
+        _mm256_storeu_si256(reinterpret_cast<__m256i*>(output+15*32),
+            _mm256_permute2x128_si256(X3_2, X3_3, 0 + (2 << 4)));
     }
 
     // https://software.intel.com/en-us/articles/avoiding-avx-sse-transition-penalties

chacha_simd.cpp

@@ -46,6 +46,9 @@
 
 #if defined(__XOP__)
 # include <ammintrin.h>
+# if defined(__GNUC__)
+#  include <x86intrin.h>
+# endif
 #endif
 
 // C1189: error: This header is specific to ARM targets
@@ -209,7 +212,7 @@ inline __m128i RotateLeft<16>(const __m128i val)
 
 #if (CRYPTOPP_ALTIVEC_AVAILABLE)
 
-// ChaCha_OperateKeystream_POWER7 is optimized for POWER7. However, Altivec
+// ChaCha_OperateKeystream_POWER8 is optimized for POWER7. However, Altivec
 // is supported by using vec_ld and vec_st, and using a composite VecAdd
 // that supports 64-bit element adds. vec_ld and vec_st add significant
 // overhead when memory is not aligned. Despite the drawbacks Altivec
@@ -565,14 +568,10 @@ void ChaCha_OperateKeystream_NEON(const word32 *state, const byte* input, byte *
 
 void ChaCha_OperateKeystream_SSE2(const word32 *state, const byte* input, byte *output, unsigned int rounds)
 {
-    const __m128i* state_mm = reinterpret_cast<const __m128i*>(state);
+    const __m128i state0 = _mm_load_si128(reinterpret_cast<const __m128i*>(state+0*4));
-    const __m128i* input_mm = reinterpret_cast<const __m128i*>(input);
+    const __m128i state1 = _mm_load_si128(reinterpret_cast<const __m128i*>(state+1*4));
-    __m128i* output_mm = reinterpret_cast<__m128i*>(output);
+    const __m128i state2 = _mm_load_si128(reinterpret_cast<const __m128i*>(state+2*4));
-
+    const __m128i state3 = _mm_load_si128(reinterpret_cast<const __m128i*>(state+3*4));
-    const __m128i state0 = _mm_load_si128(state_mm + 0);
-    const __m128i state1 = _mm_load_si128(state_mm + 1);
-    const __m128i state2 = _mm_load_si128(state_mm + 2);
-    const __m128i state3 = _mm_load_si128(state_mm + 3);
 
     __m128i r0_0 = state0;
     __m128i r0_1 = state1;
@@ -772,62 +771,62 @@ void ChaCha_OperateKeystream_SSE2(const word32 *state, const byte* input, byte *
     r3_3 = _mm_add_epi32(r3_3, state3);
     r3_3 = _mm_add_epi64(r3_3, _mm_set_epi32(0, 0, 0, 3));
 
-    if (input_mm)
+    if (input)
     {
-        r0_0 = _mm_xor_si128(_mm_loadu_si128(input_mm + 0), r0_0);
+        r0_0 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+0*16)), r0_0);
-        r0_1 = _mm_xor_si128(_mm_loadu_si128(input_mm + 1), r0_1);
+        r0_1 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+1*16)), r0_1);
-        r0_2 = _mm_xor_si128(_mm_loadu_si128(input_mm + 2), r0_2);
+        r0_2 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+2*16)), r0_2);
-        r0_3 = _mm_xor_si128(_mm_loadu_si128(input_mm + 3), r0_3);
+        r0_3 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+3*16)), r0_3);
     }
 
-    _mm_storeu_si128(output_mm + 0, r0_0);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+0*16), r0_0);
-    _mm_storeu_si128(output_mm + 1, r0_1);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+1*16), r0_1);
-    _mm_storeu_si128(output_mm + 2, r0_2);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+2*16), r0_2);
-    _mm_storeu_si128(output_mm + 3, r0_3);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+3*16), r0_3);
 
-    if (input_mm)
+    if (input)
     {
-        r1_0 = _mm_xor_si128(_mm_loadu_si128(input_mm + 4), r1_0);
+        r1_0 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+4*16)), r1_0);
-        r1_1 = _mm_xor_si128(_mm_loadu_si128(input_mm + 5), r1_1);
+        r1_1 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+5*16)), r1_1);
-        r1_2 = _mm_xor_si128(_mm_loadu_si128(input_mm + 6), r1_2);
+        r1_2 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+6*16)), r1_2);
-        r1_3 = _mm_xor_si128(_mm_loadu_si128(input_mm + 7), r1_3);
+        r1_3 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+7*16)), r1_3);
     }
 
-    _mm_storeu_si128(output_mm + 4, r1_0);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+4*16), r1_0);
-    _mm_storeu_si128(output_mm + 5, r1_1);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+5*16), r1_1);
-    _mm_storeu_si128(output_mm + 6, r1_2);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+6*16), r1_2);
-    _mm_storeu_si128(output_mm + 7, r1_3);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+7*16), r1_3);
 
-    if (input_mm)
+    if (input)
     {
-        r2_0 = _mm_xor_si128(_mm_loadu_si128(input_mm + 8), r2_0);
+        r2_0 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+ 8*16)), r2_0);
-        r2_1 = _mm_xor_si128(_mm_loadu_si128(input_mm + 9), r2_1);
+        r2_1 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+ 9*16)), r2_1);
-        r2_2 = _mm_xor_si128(_mm_loadu_si128(input_mm + 10), r2_2);
+        r2_2 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+10*16)), r2_2);
-        r2_3 = _mm_xor_si128(_mm_loadu_si128(input_mm + 11), r2_3);
+        r2_3 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+11*16)), r2_3);
     }
 
-    _mm_storeu_si128(output_mm + 8, r2_0);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+ 8*16), r2_0);
-    _mm_storeu_si128(output_mm + 9, r2_1);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+ 9*16), r2_1);
-    _mm_storeu_si128(output_mm + 10, r2_2);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+10*16), r2_2);
-    _mm_storeu_si128(output_mm + 11, r2_3);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+11*16), r2_3);
 
-    if (input_mm)
+    if (input)
     {
-        r3_0 = _mm_xor_si128(_mm_loadu_si128(input_mm + 12), r3_0);
+        r3_0 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+12*16)), r3_0);
-        r3_1 = _mm_xor_si128(_mm_loadu_si128(input_mm + 13), r3_1);
+        r3_1 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+13*16)), r3_1);
-        r3_2 = _mm_xor_si128(_mm_loadu_si128(input_mm + 14), r3_2);
+        r3_2 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+14*16)), r3_2);
-        r3_3 = _mm_xor_si128(_mm_loadu_si128(input_mm + 15), r3_3);
+        r3_3 = _mm_xor_si128(_mm_loadu_si128(reinterpret_cast<const __m128i*>(input+15*16)), r3_3);
     }
 
-    _mm_storeu_si128(output_mm + 12, r3_0);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+12*16), r3_0);
-    _mm_storeu_si128(output_mm + 13, r3_1);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+13*16), r3_1);
-    _mm_storeu_si128(output_mm + 14, r3_2);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+14*16), r3_2);
-    _mm_storeu_si128(output_mm + 15, r3_3);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(output+15*16), r3_3);
 }
 
 #endif  // CRYPTOPP_SSE2_INTRIN_AVAILABLE
 
-#if (CRYPTOPP_POWER7_AVAILABLE || CRYPTOPP_ALTIVEC_AVAILABLE)
+#if (CRYPTOPP_POWER8_AVAILABLE || CRYPTOPP_ALTIVEC_AVAILABLE)
 
 // ChaCha_OperateKeystream_CORE will use either POWER7 or ALTIVEC,
 // depending on the flags used to compile this source file. The
@@ -1096,11 +1095,11 @@ inline void ChaCha_OperateKeystream_CORE(const word32 *state, const byte* input,
     VecStore32LE(output + 15*16, r3_3);
 }
 
-#endif  // CRYPTOPP_POWER7_AVAILABLE || CRYPTOPP_ALTIVEC_AVAILABLE
+#endif  // CRYPTOPP_POWER8_AVAILABLE || CRYPTOPP_ALTIVEC_AVAILABLE
 
-#if (CRYPTOPP_POWER7_AVAILABLE)
+#if (CRYPTOPP_POWER8_AVAILABLE)
 
-void ChaCha_OperateKeystream_POWER7(const word32 *state, const byte* input, byte *output, unsigned int rounds)
+void ChaCha_OperateKeystream_POWER8(const word32 *state, const byte* input, byte *output, unsigned int rounds)
 {
     ChaCha_OperateKeystream_CORE(state, input, output, rounds);
 }

cham_simd.cpp

@@ -24,6 +24,9 @@
 
 #if defined(__XOP__)
 # include <ammintrin.h>
+# if defined(__GNUC__)
+#  include <x86intrin.h>
+# endif
 #endif
 
 #if defined(__AVX512F__)
@@ -31,6 +34,10 @@
 # include <immintrin.h>
 #endif
 
+// Clang intrinsic casts, http://bugs.llvm.org/show_bug.cgi?id=20670
+#define DOUBLE_CAST(x) ((double*)(void*)(x))
+#define CONST_DOUBLE_CAST(x) ((const double*)(const void*)(x))
+
 // Squash MS LNK4221 and libtool warnings
 extern const char CHAM_SIMD_FNAME[] = __FILE__;
 
@@ -355,7 +362,7 @@ inline void CHAM64_Enc_Block(__m128i &block0,
     for (int i=0; i<static_cast<int>(rounds); i+=4)
     {
         __m128i k, kr, t1, t2, t3, t4;
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i+0) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i+0) & MASK])));
 
         // Shuffle out key
         kr = _mm_shuffle_epi8(k, _mm_set_epi8(1,0,1,0, 1,0,1,0, 1,0,1,0, 1,0,1,0));
@@ -428,7 +435,7 @@ inline void CHAM64_Dec_Block(__m128i &block0,
     for (int i = static_cast<int>(rounds)-1; i >= 0; i-=4)
     {
         __m128i k, kr, t1, t2, t3, t4;
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i-3) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i-3) & MASK])));
 
         // Shuffle out key
         kr = _mm_shuffle_epi8(k, _mm_set_epi8(7,6,7,6, 7,6,7,6, 7,6,7,6, 7,6,7,6));
@@ -505,7 +512,7 @@ inline void CHAM64_Enc_2_Blocks(__m128i &block0,
     for (int i=0; i<static_cast<int>(rounds); i+=4)
     {
         __m128i k, kr, t1, t2, t3, t4;
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[i & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i+0) & MASK])));
 
         // Shuffle out key
         kr = _mm_shuffle_epi8(k, _mm_set_epi8(1,0,1,0, 1,0,1,0, 1,0,1,0, 1,0,1,0));
@@ -579,7 +586,7 @@ inline void CHAM64_Dec_2_Blocks(__m128i &block0,
     for (int i = static_cast<int>(rounds)-1; i >= 0; i-=4)
     {
         __m128i k, kr, t1, t2, t3, t4;
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i-3) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i-3) & MASK])));
 
         // Shuffle out key
         kr = _mm_shuffle_epi8(k, _mm_set_epi8(7,6,7,6, 7,6,7,6, 7,6,7,6, 7,6,7,6));
@@ -814,7 +821,7 @@ inline void CHAM128_Enc_Block(__m128i &block0,
     for (int i=0; i<static_cast<int>(rounds); i+=4)
     {
         __m128i k, k1, k2, t1, t2;
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i+0) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i+0) & MASK])));
 
         // Shuffle out two subkeys
         k1 = _mm_shuffle_epi8(k, _mm_set_epi8(3,2,1,0, 3,2,1,0, 3,2,1,0, 3,2,1,0));
@@ -831,8 +838,7 @@ inline void CHAM128_Enc_Block(__m128i &block0,
         b = RotateLeft32<1>(_mm_add_epi32(t1, t2));
 
         counter = _mm_add_epi32(counter, increment);
-
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i+2) & MASK])));
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i+2) & MASK])));
 
         // Shuffle out two subkeys
         k1 = _mm_shuffle_epi8(k, _mm_set_epi8(3,2,1,0, 3,2,1,0, 3,2,1,0, 3,2,1,0));
@@ -874,7 +880,7 @@ inline void CHAM128_Dec_Block(__m128i &block0,
     for (int i = static_cast<int>(rounds)-1; i >= 0; i-=4)
     {
         __m128i k, k1, k2, t1, t2;
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i-1) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i-1) & MASK])));
 
         // Shuffle out two subkeys
         k1 = _mm_shuffle_epi8(k, _mm_set_epi8(7,6,5,4, 7,6,5,4, 7,6,5,4, 7,6,5,4));
@@ -893,7 +899,7 @@ inline void CHAM128_Dec_Block(__m128i &block0,
         c = _mm_xor_si128(_mm_sub_epi32(t1, t2), counter);
 
         counter = _mm_sub_epi32(counter, decrement);
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i-3) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i-3) & MASK])));
 
         // Shuffle out two subkeys
         k1 = _mm_shuffle_epi8(k, _mm_set_epi8(7,6,5,4, 7,6,5,4, 7,6,5,4, 7,6,5,4));
@@ -937,7 +943,7 @@ inline void CHAM128_Enc_4_Blocks(__m128i &block0, __m128i &block1,
     for (int i=0; i<static_cast<int>(rounds); i+=4)
     {
         __m128i k, k1, k2, t1, t2;
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i+0) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i+0) & MASK])));
 
         // Shuffle out two subkeys
         k1 = _mm_shuffle_epi8(k, _mm_set_epi8(3,2,1,0, 3,2,1,0, 3,2,1,0, 3,2,1,0));
@@ -954,7 +960,7 @@ inline void CHAM128_Enc_4_Blocks(__m128i &block0, __m128i &block1,
         b = RotateLeft32<1>(_mm_add_epi32(t1, t2));
 
         counter = _mm_add_epi32(counter, increment);
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i+2) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i+2) & MASK])));
 
         // Shuffle out two subkeys
         k1 = _mm_shuffle_epi8(k, _mm_set_epi8(3,2,1,0, 3,2,1,0, 3,2,1,0, 3,2,1,0));
@@ -999,7 +1005,7 @@ inline void CHAM128_Dec_4_Blocks(__m128i &block0, __m128i &block1,
     for (int i = static_cast<int>(rounds)-1; i >= 0; i-=4)
     {
         __m128i k, k1, k2, t1, t2;
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i-1) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i-1) & MASK])));
 
         // Shuffle out two subkeys
         k1 = _mm_shuffle_epi8(k, _mm_set_epi8(7,6,5,4, 7,6,5,4, 7,6,5,4, 7,6,5,4));
@@ -1018,7 +1024,7 @@ inline void CHAM128_Dec_4_Blocks(__m128i &block0, __m128i &block1,
         c = _mm_xor_si128(_mm_sub_epi32(t1, t2), counter);
 
         counter = _mm_sub_epi32(counter, decrement);
-        k = _mm_castpd_si128(_mm_load_sd((const double*)(&subkeys[(i-3) & MASK])));
+        k = _mm_castpd_si128(_mm_load_sd(CONST_DOUBLE_CAST(&subkeys[(i-3) & MASK])));
 
         // Shuffle out two subkeys
         k1 = _mm_shuffle_epi8(k, _mm_set_epi8(7,6,5,4, 7,6,5,4, 7,6,5,4, 7,6,5,4));

config_align.h

@@ -0,0 +1,61 @@
+// config_align.h - written and placed in public domain by Jeffrey Walton
+//                  the bits that make up this source file are from the
+//                  library's monolithic config.h.
+
+/// \file config_align.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+/// \since Crypto++ 8.3
+
+#ifndef CRYPTOPP_CONFIG_ALIGN_H
+#define CRYPTOPP_CONFIG_ALIGN_H
+
+#include "config_cpu.h"
+#include "config_cxx.h"
+#include "config_ver.h"
+
+// Nearly all Intel's and AMD's have SSE. Enable it independent of SSE ASM and intrinscs
+#if (CRYPTOPP_BOOL_X86 || CRYPTOPP_BOOL_X32 || CRYPTOPP_BOOL_X64 || CRYPTOPP_BOOL_PPC32 || CRYPTOPP_BOOL_PPC64) && !defined(CRYPTOPP_DISABLE_ASM)
+	#define CRYPTOPP_BOOL_ALIGN16 1
+#else
+	#define CRYPTOPP_BOOL_ALIGN16 0
+#endif
+
+// How to allocate 16-byte aligned memory (for SSE2)
+// posix_memalign see https://forum.kde.org/viewtopic.php?p=66274
+#if defined(_MSC_VER)
+	#define CRYPTOPP_MM_MALLOC_AVAILABLE
+#elif defined(__linux__) || defined(__sun__) || defined(__CYGWIN__)
+	#define CRYPTOPP_MEMALIGN_AVAILABLE
+#elif defined(__APPLE__) || defined(__NetBSD__) || defined(__OpenBSD__) || defined(__DragonFly__)
+	#define CRYPTOPP_MALLOC_ALIGNMENT_IS_16
+#elif (defined(_GNU_SOURCE) || ((_XOPEN_SOURCE + 0) >= 600)) && (_POSIX_ADVISORY_INFO > 0)
+	#define CRYPTOPP_POSIX_MEMALIGN_AVAILABLE
+#else
+	#define CRYPTOPP_NO_ALIGNED_ALLOC
+#endif
+
+// Sun Studio Express 3 (December 2006) provides GCC-style attributes.
+// IBM XL C/C++ alignment modifier per Optimization Guide, pp. 19-20.
+// __IBM_ATTRIBUTES per XLC 12.1 AIX Compiler Manual, p. 473.
+// CRYPTOPP_ALIGN_DATA may not be reliable on AIX.
+#ifndef CRYPTOPP_ALIGN_DATA
+	#if defined(CRYPTOPP_CXX11_ALIGNAS)
+		#define CRYPTOPP_ALIGN_DATA(x) alignas(x)
+	#elif defined(_MSC_VER)
+		#define CRYPTOPP_ALIGN_DATA(x) __declspec(align(x))
+	#elif defined(__GNUC__) || defined(__clang__) || (__SUNPRO_CC >= 0x5100)
+		#define CRYPTOPP_ALIGN_DATA(x) __attribute__((aligned(x)))
+	#elif defined(__xlc__) || defined(__xlC__)
+		#define CRYPTOPP_ALIGN_DATA(x) __attribute__((aligned(x)))
+	#else
+		#define CRYPTOPP_ALIGN_DATA(x)
+	#endif
+#endif
+
+#endif  // CRYPTOPP_CONFIG_ALIGN_H

config_asm.h

@@ -0,0 +1,418 @@
+// config_asm.h - written and placed in public domain by Jeffrey Walton
+//                the bits that make up this source file are from the
+//                library's monolithic config.h.
+
+/// \file config_asm.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+/// \since Crypto++ 8.3
+
+#ifndef CRYPTOPP_CONFIG_ASM_H
+#define CRYPTOPP_CONFIG_ASM_H
+
+#include "config_os.h"
+#include "config_cpu.h"
+#include "config_ver.h"
+
+// Define this to disable ASM, intrinsics and built-ins. The library will be
+// compiled using C++ only. The library code will not include SSE2 (and
+// above), NEON, Aarch32, Aarch64, or Altivec (and above). Note the compiler
+// may use higher ISAs depending on compiler options, but the library will not
+// explictly use the ISAs. When disabling ASM, it is best to do it from
+// config.h to ensure the library and all programs share the setting.
+// #define CRYPTOPP_DISABLE_ASM 1
+
+// https://github.com/weidai11/cryptopp/issues/719
+#if defined(__native_client__) && !defined(CRYPTOPP_DISABLE_ASM)
+# define CRYPTOPP_DISABLE_ASM 1
+#endif
+
+// Some Clang and SunCC cannot handle mixed asm with positional arguments,
+// where the body is Intel style with no prefix and the templates are
+// AT&T style. Define this if the Makefile misdetects the configuration.
+// Also see https://bugs.llvm.org/show_bug.cgi?id=39895 .
+// #define CRYPTOPP_DISABLE_MIXED_ASM 1
+
+// CRYPTOPP_ALLOW_UNALIGNED_DATA_ACCESS is no longer honored. It
+// was removed at https://github.com/weidai11/cryptopp/issues/682
+// #define CRYPTOPP_ALLOW_UNALIGNED_DATA_ACCESS 1
+
+// It is OK to remove the hard stop below, but you are on your own.
+// After building the library be sure to run self tests described
+// https://www.cryptopp.com/wiki/Release_Process#Self_Tests
+// Some relevant bug reports can be found at:
+// * Clang: http://github.com/weidai11/cryptopp/issues/147
+// * Native Client: https://github.com/weidai11/cryptopp/issues/719
+#if (defined(_MSC_VER) && defined(__clang__))
+# error: "Unsupported configuration"
+#endif
+
+// You may need to force include a C++ header on Android when using STLPort to ensure
+// _STLPORT_VERSION is defined: CXXFLAGS="-DNDEBUG -g2 -O2 -std=c++11 -include iosfwd"
+// TODO: Figure out C++17 and lack of std::uncaught_exception
+#if (defined(_MSC_VER) && _MSC_VER <= 1300) || defined(__MWERKS__) || (defined(_STLPORT_VERSION) && ((_STLPORT_VERSION < 0x450) || defined(_STLP_NO_UNCAUGHT_EXCEPT_SUPPORT)))
+#define CRYPTOPP_DISABLE_UNCAUGHT_EXCEPTION
+#endif
+
+// ***************** IA32 CPU features ********************
+
+#if (CRYPTOPP_BOOL_X86 || CRYPTOPP_BOOL_X32 || CRYPTOPP_BOOL_X64)
+
+// Apple Clang prior to 5.0 cannot handle SSE2
+#if !defined(CRYPTOPP_DISABLE_ASM) && defined(CRYPTOPP_APPLE_CLANG_VERSION) && (CRYPTOPP_APPLE_CLANG_VERSION < 50000)
+# define CRYPTOPP_DISABLE_ASM 1
+#endif
+
+// Sun Studio 12.1 provides GCC inline assembly
+// http://blogs.oracle.com/x86be/entry/gcc_style_asm_inlining_support
+#if !defined(CRYPTOPP_DISABLE_ASM) && defined(__SUNPRO_CC) && (__SUNPRO_CC < 0x5100)
+# define CRYPTOPP_DISABLE_ASM 1
+#endif
+
+#if !defined(CRYPTOPP_DISABLE_ASM) && ((defined(_MSC_VER) && defined(_M_IX86)) || (defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))))
+	// C++Builder 2010 does not allow "call label" where label is defined within inline assembly
+	#define CRYPTOPP_X86_ASM_AVAILABLE 1
+
+	#if !defined(CRYPTOPP_DISABLE_SSE2) && (defined(_MSC_VER) || CRYPTOPP_GCC_VERSION >= 30300 || defined(__SSE2__))
+		#define CRYPTOPP_SSE2_ASM_AVAILABLE 1
+	#endif
+
+	#if !defined(CRYPTOPP_DISABLE_SSSE3) && (_MSC_VER >= 1500 || CRYPTOPP_GCC_VERSION >= 40300 || defined(__SSSE3__))
+		#define CRYPTOPP_SSSE3_ASM_AVAILABLE 1
+	#endif
+#endif
+
+#if !defined(CRYPTOPP_DISABLE_ASM) && defined(_MSC_VER) && defined(_M_X64)
+	#define CRYPTOPP_X64_MASM_AVAILABLE 1
+#endif
+
+#if !defined(CRYPTOPP_DISABLE_ASM) && defined(__GNUC__) && defined(__x86_64__)
+	#define CRYPTOPP_X64_ASM_AVAILABLE 1
+#endif
+
+// 32-bit SunCC does not enable SSE2 by default.
+#if !defined(CRYPTOPP_DISABLE_ASM) && !defined(CRYPTOPP_DISABLE_SSE2) && (defined(_MSC_VER) || CRYPTOPP_GCC_VERSION >= 30300 || defined(__SSE2__) || (__SUNPRO_CC >= 0x5100))
+	#define CRYPTOPP_SSE2_INTRIN_AVAILABLE 1
+#endif
+
+#if !defined(CRYPTOPP_DISABLE_ASM) && !defined(CRYPTOPP_DISABLE_SSSE3)
+# if defined(__SSSE3__) || (_MSC_VER >= 1500) || \
+	(CRYPTOPP_GCC_VERSION >= 40300) || (__INTEL_COMPILER >= 1000) || (__SUNPRO_CC >= 0x5110) || \
+	(CRYPTOPP_LLVM_CLANG_VERSION >= 20300) || (CRYPTOPP_APPLE_CLANG_VERSION >= 40000)
+	#define CRYPTOPP_SSSE3_AVAILABLE 1
+# endif
+#endif
+
+// Intrinsics availible in GCC 4.3 (http://gcc.gnu.org/gcc-4.3/changes.html) and
+// MSVC 2008 (http://msdn.microsoft.com/en-us/library/bb892950%28v=vs.90%29.aspx)
+// SunCC could generate SSE4 at 12.1, but the intrinsics are missing until 12.4.
+#if !defined(CRYPTOPP_DISABLE_SSE4) && defined(CRYPTOPP_SSSE3_AVAILABLE) && \
+	(defined(__SSE4_1__) || (CRYPTOPP_MSC_VERSION >= 1500) || \
+	(CRYPTOPP_GCC_VERSION >= 40300) || (__INTEL_COMPILER >= 1000) || (__SUNPRO_CC >= 0x5110) || \
+	(CRYPTOPP_LLVM_CLANG_VERSION >= 20300) || (CRYPTOPP_APPLE_CLANG_VERSION >= 40000))
+	#define CRYPTOPP_SSE41_AVAILABLE 1
+#endif
+
+#if !defined(CRYPTOPP_DISABLE_SSE4) && defined(CRYPTOPP_SSSE3_AVAILABLE) && \
+	(defined(__SSE4_2__) || (CRYPTOPP_MSC_VERSION >= 1500) || (__SUNPRO_CC >= 0x5110) || \
+	(CRYPTOPP_GCC_VERSION >= 40300) || (__INTEL_COMPILER >= 1000) || \
+	(CRYPTOPP_LLVM_CLANG_VERSION >= 20300) || (CRYPTOPP_APPLE_CLANG_VERSION >= 40000))
+	#define CRYPTOPP_SSE42_AVAILABLE 1
+#endif
+
+// Couple to CRYPTOPP_DISABLE_AESNI, but use CRYPTOPP_CLMUL_AVAILABLE so we can selectively
+//  disable for misbehaving platofrms and compilers, like Solaris or some Clang.
+#if defined(CRYPTOPP_DISABLE_AESNI)
+	#define CRYPTOPP_DISABLE_CLMUL 1
+#endif
+
+// Requires Sun Studio 12.3 (SunCC 0x5120) in theory.
+#if !defined(CRYPTOPP_DISABLE_ASM) && !defined(CRYPTOPP_DISABLE_CLMUL) && defined(CRYPTOPP_SSE42_AVAILABLE) && \
+	(defined(__PCLMUL__) || (_MSC_FULL_VER >= 150030729) || (__SUNPRO_CC >= 0x5120) || \
+	(CRYPTOPP_GCC_VERSION >= 40300) || (__INTEL_COMPILER >= 1110) || \
+	(CRYPTOPP_LLVM_CLANG_VERSION >= 30200) || (CRYPTOPP_APPLE_CLANG_VERSION >= 40300))
+	#define CRYPTOPP_CLMUL_AVAILABLE 1
+#endif
+
+// Requires Sun Studio 12.3 (SunCC 0x5120)
+#if !defined(CRYPTOPP_DISABLE_ASM) && !defined(CRYPTOPP_DISABLE_AESNI) && defined(CRYPTOPP_SSE42_AVAILABLE) && \
+	(defined(__AES__) || (_MSC_FULL_VER >= 150030729) || (__SUNPRO_CC >= 0x5120) || \
+	(CRYPTOPP_GCC_VERSION >= 40300) || (__INTEL_COMPILER >= 1110) || \
+	(CRYPTOPP_LLVM_CLANG_VERSION >= 30200) || (CRYPTOPP_APPLE_CLANG_VERSION >= 40300))
+	#define CRYPTOPP_AESNI_AVAILABLE 1
+#endif
+
+// Requires Binutils 2.24
+#if !defined(CRYPTOPP_DISABLE_AVX) && defined(CRYPTOPP_SSE42_AVAILABLE) && \
+	(defined(__AVX2__) || (CRYPTOPP_MSC_VERSION >= 1800) || (__SUNPRO_CC >= 0x5130) || \
+	(CRYPTOPP_GCC_VERSION >= 40700) || (__INTEL_COMPILER >= 1400) || \
+	(CRYPTOPP_LLVM_CLANG_VERSION >= 30100) || (CRYPTOPP_APPLE_CLANG_VERSION >= 40600))
+#define CRYPTOPP_AVX_AVAILABLE 1
+#endif
+
+// Requires Binutils 2.24
+#if !defined(CRYPTOPP_DISABLE_AVX2) && defined(CRYPTOPP_AVX_AVAILABLE) && \
+	(defined(__AVX2__) || (CRYPTOPP_MSC_VERSION >= 1800) || (__SUNPRO_CC >= 0x5130) || \
+	(CRYPTOPP_GCC_VERSION >= 40900) || (__INTEL_COMPILER >= 1400) || \
+	(CRYPTOPP_LLVM_CLANG_VERSION >= 30100) || (CRYPTOPP_APPLE_CLANG_VERSION >= 40600))
+#define CRYPTOPP_AVX2_AVAILABLE 1
+#endif
+
+// Guessing at SHA for SunCC. Its not in Sun Studio 12.6. Also see
+// http://stackoverflow.com/questions/45872180/which-xarch-for-sha-extensions-on-solaris
+#if !defined(CRYPTOPP_DISABLE_ASM) && !defined(CRYPTOPP_DISABLE_SHANI) && defined(CRYPTOPP_SSE42_AVAILABLE) && \
+	(defined(__SHA__) || (CRYPTOPP_MSC_VERSION >= 1900) || (__SUNPRO_CC >= 0x5160) || \
+	(CRYPTOPP_GCC_VERSION >= 40900) || (__INTEL_COMPILER >= 1300) || \
+	(CRYPTOPP_LLVM_CLANG_VERSION >= 30400) || (CRYPTOPP_APPLE_CLANG_VERSION >= 50100))
+	#define CRYPTOPP_SHANI_AVAILABLE 1
+#endif
+
+// Fixup Android and SSE, Crypto. It may be enabled based on compiler version.
+#if (defined(__ANDROID__) || defined(ANDROID))
+# if (CRYPTOPP_BOOL_X86)
+#  undef CRYPTOPP_SSE41_AVAILABLE
+#  undef CRYPTOPP_SSE42_AVAILABLE
+#  undef CRYPTOPP_CLMUL_AVAILABLE
+#  undef CRYPTOPP_AESNI_AVAILABLE
+#  undef CRYPTOPP_SHANI_AVAILABLE
+# endif
+# if (CRYPTOPP_BOOL_X64)
+#  undef CRYPTOPP_CLMUL_AVAILABLE
+#  undef CRYPTOPP_AESNI_AVAILABLE
+#  undef CRYPTOPP_SHANI_AVAILABLE
+# endif
+#endif
+
+// Fixup for SunCC 12.1-12.4. Bad code generation in AES_Encrypt and friends.
+#if defined(__SUNPRO_CC) && (__SUNPRO_CC <= 0x5130)
+# undef CRYPTOPP_AESNI_AVAILABLE
+#endif
+
+// Fixup for SunCC 12.1-12.6. Compiler crash on GCM_Reduce_CLMUL and friends.
+// http://github.com/weidai11/cryptopp/issues/226
+#if defined(__SUNPRO_CC) && (__SUNPRO_CC <= 0x5150)
+# undef CRYPTOPP_CLMUL_AVAILABLE
+#endif
+
+#endif  // X86, X32, X64
+
+// ***************** ARM CPU features ********************
+
+#if (CRYPTOPP_BOOL_ARM32 || CRYPTOPP_BOOL_ARMV8)
+
+// We don't have an ARM big endian test rig. Disable
+// ARM-BE ASM and instrinsics until we can test it.
+#if (CRYPTOPP_BIG_ENDIAN)
+# define CRYPTOPP_DISABLE_ASM 1
+#endif
+
+// Requires ACLE 1.0. -mfpu=neon or above must be present
+// Requires GCC 4.3, Clang 2.8 or Visual Studio 2012
+// Do not use APPLE_CLANG_VERSION; use __ARM_FEATURE_XXX instead.
+#if !defined(CRYPTOPP_ARM_NEON_AVAILABLE) && !defined(CRYPTOPP_DISABLE_ASM)
+# if defined(__arm__) || defined(__ARM_NEON) || defined(__ARM_FEATURE_NEON) || defined(_M_ARM)
+#  if (CRYPTOPP_GCC_VERSION >= 40300) || (CRYPTOPP_LLVM_CLANG_VERSION >= 20800) || \
+      (CRYPTOPP_MSC_VERSION >= 1700)
+#   define CRYPTOPP_ARM_NEON_AVAILABLE 1
+#  endif  // Compilers
+# endif  // Platforms
+#endif
+
+// ARMv8 and ASIMD. -march=armv8-a or above must be present
+// Requires GCC 4.8, Clang 3.3 or Visual Studio 2017
+// Do not use APPLE_CLANG_VERSION; use __ARM_FEATURE_XXX instead.
+#if !defined(CRYPTOPP_ARM_ASIMD_AVAILABLE) && !defined(CRYPTOPP_DISABLE_ASM)
+# if defined(__aarch32__) || defined(__aarch64__) || defined(_M_ARM64)
+#  if defined(__ARM_NEON) || defined(__ARM_FEATURE_NEON) || defined(__ARM_FEATURE_ASIMD) || \
+      (CRYPTOPP_GCC_VERSION >= 40800) || (CRYPTOPP_LLVM_CLANG_VERSION >= 30300) || \
+      (CRYPTOPP_MSC_VERSION >= 1916)
+#   define CRYPTOPP_ARM_NEON_AVAILABLE 1
+#   define CRYPTOPP_ARM_ASIMD_AVAILABLE 1
+#  endif  // Compilers
+# endif  // Platforms
+#endif
+
+// ARMv8 and ASIMD. -march=armv8-a+crc or above must be present
+// Requires GCC 4.8, Clang 3.3 or Visual Studio 2017
+// Do not use APPLE_CLANG_VERSION; use __ARM_FEATURE_XXX instead.
+#if !defined(CRYPTOPP_ARM_CRC32_AVAILABLE) && !defined(CRYPTOPP_DISABLE_ASM)
+# if defined(__aarch32__) || defined(__aarch64__) || defined(_M_ARM64)
+#  if defined(__ARM_FEATURE_CRC32) || (CRYPTOPP_GCC_VERSION >= 40800) || \
+      (CRYPTOPP_LLVM_CLANG_VERSION >= 30300) || (CRYPTOPP_MSC_VERSION >= 1916)
+#   define CRYPTOPP_ARM_CRC32_AVAILABLE 1
+#  endif  // Compilers
+# endif  // Platforms
+#endif
+
+// ARMv8 and ASIMD. -march=armv8-a+crypto or above must be present
+// Requires GCC 4.8, Clang 3.3 or Visual Studio 2017
+// Do not use APPLE_CLANG_VERSION; use __ARM_FEATURE_XXX instead.
+#if !defined(CRYPTOPP_ARM_PMULL_AVAILABLE) && !defined(CRYPTOPP_DISABLE_ASM)
+# if defined(__aarch32__) || defined(__aarch64__) || defined(_M_ARM64)
+#  if defined(__ARM_FEATURE_CRYPTO) || (CRYPTOPP_GCC_VERSION >= 40800) || \
+      (CRYPTOPP_LLVM_CLANG_VERSION >= 30300) || (CRYPTOPP_MSC_VERSION >= 1916)
+#   define CRYPTOPP_ARM_PMULL_AVAILABLE 1
+#  endif  // Compilers
+# endif  // Platforms
+#endif
+
+// ARMv8 and AES. -march=armv8-a+crypto or above must be present
+// Requires GCC 4.8, Clang 3.3 or Visual Studio 2017
+// Do not use APPLE_CLANG_VERSION; use __ARM_FEATURE_XXX instead.
+#if !defined(CRYPTOPP_ARM_AES_AVAILABLE) && !defined(CRYPTOPP_DISABLE_ASM)
+# if defined(__aarch32__) || defined(__aarch64__) || defined(_M_ARM64)
+#  if defined(__ARM_FEATURE_CRYPTO) || (CRYPTOPP_GCC_VERSION >= 40800) || \
+      (CRYPTOPP_LLVM_CLANG_VERSION >= 30300) || (CRYPTOPP_MSC_VERSION >= 1916)
+#   define CRYPTOPP_ARM_AES_AVAILABLE 1
+#  endif  // Compilers
+# endif  // Platforms
+#endif
+
+// ARMv8 and SHA-1, SHA-256. -march=armv8-a+crypto or above must be present
+// Requires GCC 4.8, Clang 3.3 or Visual Studio 2017
+// Do not use APPLE_CLANG_VERSION; use __ARM_FEATURE_XXX instead.
+#if !defined(CRYPTOPP_ARM_SHA_AVAILABLE) && !defined(CRYPTOPP_DISABLE_ASM)
+# if defined(__aarch32__) || defined(__aarch64__) || defined(_M_ARM64)
+#  if defined(__ARM_FEATURE_CRYPTO) || (CRYPTOPP_GCC_VERSION >= 40800) || \
+      (CRYPTOPP_LLVM_CLANG_VERSION >= 30300) || (CRYPTOPP_MSC_VERSION >= 1916)
+#   define CRYPTOPP_ARM_SHA1_AVAILABLE 1
+#   define CRYPTOPP_ARM_SHA2_AVAILABLE 1
+#  endif  // Compilers
+# endif  // Platforms
+#endif
+
+// ARMv8 and SHA-512, SHA-3. -march=armv8.4-a+crypto or above must be present
+// Requires GCC 8.0, Clang ??? or Visual Studio 20??
+// Do not use APPLE_CLANG_VERSION; use __ARM_FEATURE_XXX instead.
+#if !defined(CRYPTOPP_ARM_SHA3_AVAILABLE) && !defined(CRYPTOPP_DISABLE_ASM)
+# if defined(__aarch32__) || defined(__aarch64__) || defined(_M_ARM64)
+#  if defined(__ARM_FEATURE_SHA3) || (CRYPTOPP_GCC_VERSION >= 80000)
+#   define CRYPTOPP_ARM_SHA512_AVAILABLE 1
+#   define CRYPTOPP_ARM_SHA3_AVAILABLE 1
+#  endif  // Compilers
+# endif  // Platforms
+#endif
+
+// ARMv8 and SM3, SM4. -march=armv8.4-a+crypto or above must be present
+// Requires GCC 8.0, Clang ??? or Visual Studio 20??
+// Do not use APPLE_CLANG_VERSION; use __ARM_FEATURE_XXX instead.
+#if !defined(CRYPTOPP_ARM_SM3_AVAILABLE) && !defined(CRYPTOPP_DISABLE_ASM)
+# if defined(__aarch32__) || defined(__aarch64__) || defined(_M_ARM64)
+#  if defined(__ARM_FEATURE_SM3) || (CRYPTOPP_GCC_VERSION >= 80000)
+#   define CRYPTOPP_ARM_SM3_AVAILABLE 1
+#   define CRYPTOPP_ARM_SM4_AVAILABLE 1
+#  endif  // Compilers
+# endif  // Platforms
+#endif
+
+// Limit the <arm_acle.h> include.
+#if !defined(CRYPTOPP_ARM_ACLE_AVAILABLE)
+# if defined(__aarch32__) || defined(__aarch64__) || (__ARM_ARCH >= 8) || defined(__ARM_ACLE)
+#  if !defined(__ANDROID__) && !defined(ANDROID) && !defined(__APPLE__)
+#   define CRYPTOPP_ARM_ACLE_AVAILABLE 1
+#  endif
+# endif
+#endif
+
+// Fixup Apple Clang and PMULL. Apple defines __ARM_FEATURE_CRYPTO for Xcode 6
+// but does not provide PMULL. TODO: determine when PMULL is available.
+#if defined(CRYPTOPP_APPLE_CLANG_VERSION) && (CRYPTOPP_APPLE_CLANG_VERSION < 70000)
+# undef CRYPTOPP_ARM_PMULL_AVAILABLE
+#endif
+
+// Fixup Android and CRC32. It may be enabled based on compiler version.
+#if (defined(__ANDROID__) || defined(ANDROID)) && !defined(__ARM_FEATURE_CRC32)
+# undef CRYPTOPP_ARM_CRC32_AVAILABLE
+#endif
+
+// Fixup Android and Crypto. It may be enabled based on compiler version.
+#if (defined(__ANDROID__) || defined(ANDROID)) && !defined(__ARM_FEATURE_CRYPTO)
+# undef CRYPTOPP_ARM_PMULL_AVAILABLE
+# undef CRYPTOPP_ARM_AES_AVAILABLE
+# undef CRYPTOPP_ARM_SHA1_AVAILABLE
+# undef CRYPTOPP_ARM_SHA2_AVAILABLE
+#endif
+
+// Cryptogams offers an ARM asm implementations for AES and SHA. Crypto++ does
+// not provide an asm implementation. The Cryptogams AES implementation is
+// about 50% faster than C/C++, and SHA implementation is about 30% faster
+// than C/C++. Define this to use the Cryptogams AES and SHA implementations
+// on GNU Linux systems. When defined, Crypto++ will use aes_armv4.S,
+// sha1_armv4.S and sha256_armv4.S. https://www.cryptopp.com/wiki/Cryptogams.
+#if !defined(CRYPTOPP_DISABLE_ASM) && defined(__arm__) && defined(__linux__)
+# if defined(__GNUC__) || defined(__clang__)
+#  define CRYPTOGAMS_ARM_AES      1
+#  define CRYPTOGAMS_ARM_SHA1     1
+#  define CRYPTOGAMS_ARM_SHA256   1
+#  define CRYPTOGAMS_ARM_SHA512   1
+# endif
+#endif
+
+#endif  // ARM32, ARM64
+
+// ***************** AltiVec and Power8 ********************
+
+#if (CRYPTOPP_BOOL_PPC32 || CRYPTOPP_BOOL_PPC64)
+
+#if defined(CRYPTOPP_DISABLE_ALTIVEC) || defined(CRYPTOPP_DISABLE_ASM)
+# undef CRYPTOPP_DISABLE_ALTIVEC
+# undef CRYPTOPP_DISABLE_POWER7
+# undef CRYPTOPP_DISABLE_POWER8
+# undef CRYPTOPP_DISABLE_POWER9
+# define CRYPTOPP_DISABLE_ALTIVEC 1
+# define CRYPTOPP_DISABLE_POWER7 1
+# define CRYPTOPP_DISABLE_POWER8 1
+# define CRYPTOPP_DISABLE_POWER9 1
+#endif
+
+// An old Apple G5 with GCC 4.01 has AltiVec, but its only Power4 or so.
+#if !defined(CRYPTOPP_ALTIVEC_AVAILABLE) && !defined(CRYPTOPP_DISABLE_ALTIVEC)
+# if defined(_ARCH_PWR4) || defined(__ALTIVEC__) || \
+	(CRYPTOPP_XLC_VERSION >= 100000) || (CRYPTOPP_GCC_VERSION >= 40001) || \
+    (CRYPTOPP_LLVM_CLANG_VERSION >= 20900)
+#  define CRYPTOPP_ALTIVEC_AVAILABLE 1
+# endif
+#endif
+
+// We need Power7 for unaligned loads and stores
+#if !defined(CRYPTOPP_POWER7_AVAILABLE) && !defined(CRYPTOPP_DISABLE_POWER7) && defined(CRYPTOPP_ALTIVEC_AVAILABLE)
+# if defined(_ARCH_PWR7) || (CRYPTOPP_XLC_VERSION >= 100000) || \
+    (CRYPTOPP_GCC_VERSION >= 40100) || (CRYPTOPP_LLVM_CLANG_VERSION >= 30100)
+#  define CRYPTOPP_POWER7_AVAILABLE 1
+# endif
+#endif
+
+// We need Power8 for in-core crypto and 64-bit vector types
+#if !defined(CRYPTOPP_POWER8_AVAILABLE) && !defined(CRYPTOPP_DISABLE_POWER8) && defined(CRYPTOPP_POWER7_AVAILABLE)
+# if defined(_ARCH_PWR8) || (CRYPTOPP_XLC_VERSION >= 130000) || \
+    (CRYPTOPP_GCC_VERSION >= 40800) || (CRYPTOPP_LLVM_CLANG_VERSION >= 70000)
+#  define CRYPTOPP_POWER8_AVAILABLE 1
+# endif
+#endif
+
+// Power9 for random numbers
+#if !defined(CRYPTOPP_POWER9_AVAILABLE) && !defined(CRYPTOPP_DISABLE_POWER9) && defined(CRYPTOPP_POWER8_AVAILABLE)
+# if defined(_ARCH_PWR9) || (CRYPTOPP_XLC_VERSION >= 130200) || \
+    (CRYPTOPP_GCC_VERSION >= 70000) || (CRYPTOPP_LLVM_CLANG_VERSION >= 80000)
+#  define CRYPTOPP_POWER9_AVAILABLE 1
+# endif
+#endif
+
+#if !defined(CRYPTOPP_POWER8_AES_AVAILABLE) && !defined(CRYPTOPP_DISABLE_POWER8_AES) && defined(CRYPTOPP_POWER8_AVAILABLE)
+# if defined(__CRYPTO__) || defined(_ARCH_PWR8) || (CRYPTOPP_XLC_VERSION >= 130000) || \
+    (CRYPTOPP_GCC_VERSION >= 40800) || (CRYPTOPP_LLVM_CLANG_VERSION >= 70000)
+//#  define CRYPTOPP_POWER8_CRC_AVAILABLE 1
+#  define CRYPTOPP_POWER8_AES_AVAILABLE 1
+#  define CRYPTOPP_POWER8_VMULL_AVAILABLE 1
+#  define CRYPTOPP_POWER8_SHA_AVAILABLE 1
+# endif
+#endif
+
+#endif  // PPC32, PPC64
+
+#endif  // CRYPTOPP_CONFIG_ASM_H

config_cpu.h

@@ -0,0 +1,103 @@
+// config_cpu.h - written and placed in public domain by Jeffrey Walton
+//                the bits that make up this source file are from the
+//                library's monolithic config.h.
+
+/// \file config_cpu.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+///  <A HREF="https://sourceforge.net/p/predef/wiki/Architectures/">Sourceforge
+///  Pre-defined Compiler Macros</A>
+/// \since Crypto++ 8.3
+
+#ifndef CRYPTOPP_CONFIG_CPU_H
+#define CRYPTOPP_CONFIG_CPU_H
+
+#include "config_ver.h"
+
+#if (defined(__ILP32__) || defined(_ILP32)) && defined(__x86_64__)
+	#define CRYPTOPP_BOOL_X32 1
+#elif (defined(_M_X64) || defined(__x86_64__))
+	#define CRYPTOPP_BOOL_X64 1
+#elif (defined(_M_IX86) || defined(__i386__) || defined(__i386) || defined(_X86_) || defined(__I86__) || defined(__INTEL__))
+	#define CRYPTOPP_BOOL_X86 1
+#endif
+
+// Microsoft added ARM64 define December 2017.
+#if defined(__arm64__) || defined(__aarch32__) || defined(__aarch64__) || defined(_M_ARM64)
+	#define CRYPTOPP_BOOL_ARMV8 1
+#endif
+#if defined(__arm64__) || defined(__aarch64__) || defined(_M_ARM64)
+	#define CRYPTOPP_BOOL_ARM64 1
+#elif defined(__arm__) || defined(_M_ARM)
+	#define CRYPTOPP_BOOL_ARM32 1
+#endif
+
+// And PowerPC.
+#if defined(__ppc64__) || defined(__powerpc64__) || defined(__PPC64__) || defined(_ARCH_PPC64)
+	#define CRYPTOPP_BOOL_PPC64 1
+#elif defined(__powerpc__) || defined(__ppc__) || defined(__PPC__) || defined(_ARCH_PPC)
+	#define CRYPTOPP_BOOL_PPC32 1
+#endif
+
+// And MIPS. TODO: finish these defines
+#if defined(__mips64__)
+	#define CRYPTOPP_BOOL_MIPS64 1
+#elif defined(__mips__)
+	#define CRYPTOPP_BOOL_MIPS32 1
+#endif
+
+// And SPARC.
+#if defined(__sparc64__) || defined(__sparc64) || defined(__sparcv9) || defined(__sparc_v9__)
+	#define CRYPTOPP_BOOL_SPARC64 1
+#elif defined(__sparc__) || defined(__sparc) || defined(__sparcv8) || defined(__sparc_v8__)
+	#define CRYPTOPP_BOOL_SPARC32 1
+#endif
+
+// This should be a lower bound on the L1 cache line size.
+// It's used for defense against timing attacks.
+#ifndef CRYPTOPP_L1_CACHE_LINE_SIZE
+	#if defined(CRYPTOPP_BOOL_X32) || defined(CRYPTOPP_BOOL_X64) || defined(CRYPTOPP_BOOL_ARMV8) || \
+	    defined(CRYPTOPP_BOOL_PPC64) || defined(CRYPTOPP_BOOL_MIPS64) || defined(CRYPTOPP_BOOL_SPARC64)
+		#define CRYPTOPP_L1_CACHE_LINE_SIZE 64
+	#else
+		// L1 cache line size is 32 on Pentium III and earlier
+		#define CRYPTOPP_L1_CACHE_LINE_SIZE 32
+	#endif
+#endif
+
+// The section attribute attempts to initialize CPU flags to avoid Valgrind findings above -O1
+#if ((defined(__MACH__) && defined(__APPLE__)) && ((CRYPTOPP_LLVM_CLANG_VERSION >= 30600) || (CRYPTOPP_APPLE_CLANG_VERSION >= 70100) || (CRYPTOPP_GCC_VERSION >= 40300)))
+	#define CRYPTOPP_SECTION_INIT __attribute__((section ("__DATA,__data")))
+#elif (defined(__ELF__) && (CRYPTOPP_GCC_VERSION >= 40300))
+	#define CRYPTOPP_SECTION_INIT __attribute__((section ("nocommon")))
+#elif defined(__ELF__) && (defined(__xlC__) || defined(__ibmxl__))
+	#define CRYPTOPP_SECTION_INIT __attribute__((section ("nocommon")))
+#else
+	#define CRYPTOPP_SECTION_INIT
+#endif
+
+// How to disable CPU feature probing. We determine machine
+// capabilities by performing an os/platform *query* first,
+// like getauxv(). If the *query* fails, we move onto a
+// cpu *probe*. The cpu *probe* tries to exeute an instruction
+// and then catches a SIGILL on Linux or the exception
+// EXCEPTION_ILLEGAL_INSTRUCTION on Windows. Some OSes
+// fail to hangle a SIGILL gracefully, like Apple OSes. Apple
+// machines corrupt memory and variables around the probe.
+#if defined(__APPLE__)
+	#define CRYPTOPP_NO_CPU_FEATURE_PROBES 1
+#endif
+
+// Flavor of inline assembly language
+#if defined(_MSC_VER) || defined(__BORLANDC__)
+	#define CRYPTOPP_MS_STYLE_INLINE_ASSEMBLY 1
+#else
+	#define CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY 1
+#endif
+
+#endif

config_cxx.h

@@ -0,0 +1,233 @@
+// config_cxx.h - written and placed in public domain by Jeffrey Walton
+//                the bits that make up this source file are from the
+//                library's monolithic config.h.
+
+/// \file config_cxx.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+/// \since Crypto++ 8.3
+
+// Visual Studio began at VS2010, http://msdn.microsoft.com/en-us/library/hh567368%28v=vs.110%29.aspx
+//   and https://docs.microsoft.com/en-us/cpp/visual-cpp-language-conformance
+// Intel, http://software.intel.com/en-us/articles/c0x-features-supported-by-intel-c-compiler
+// GCC, http://gcc.gnu.org/projects/cxx0x.html
+// Clang, http://clang.llvm.org/cxx_status.html
+
+#ifndef CRYPTOPP_CONFIG_CXX_H
+#define CRYPTOPP_CONFIG_CXX_H
+
+#include "config_os.h"
+#include "config_cpu.h"
+#include "config_ver.h"
+
+// Ancient Crypto++ define, dating back to C++98 and C++03.
+#ifndef CRYPTOPP_DISABLE_UNCAUGHT_EXCEPTION
+# define CRYPTOPP_UNCAUGHT_EXCEPTION_AVAILABLE 1
+#endif
+
+// Compatibility with non-clang compilers.
+#ifndef __has_feature
+# define __has_feature(x) 0
+#endif
+
+// Define CRYPTOPP_NO_CXX11 to avoid C++11 related features shown at the
+// end of this file. Some compilers and standard C++ headers advertise C++11
+// but they are really just C++03 with some additional C++11 headers and
+// non-conforming classes. You might also consider `-std=c++03` or
+// `-std=gnu++03`, but they are required options when building the library
+// and all programs. CRYPTOPP_NO_CXX11 is probably easier to manage but it may
+// cause -Wterminate warnings under GCC. MSVC++ has a similar warning.
+// Also see https://github.com/weidai11/cryptopp/issues/529
+// #define CRYPTOPP_NO_CXX11 1
+
+// Define CRYPTOPP_NO_CXX17 to avoid C++17 related features shown at the end of
+// this file. At the moment it should only affect std::uncaught_exceptions.
+// #define CRYPTOPP_NO_CXX17 1
+
+// C++11 macro version, https://stackoverflow.com/q/7223991/608639
+#if !defined(CRYPTOPP_NO_CXX11)
+#  if ((_MSC_VER >= 1600) || (__cplusplus >= 201103L)) && !defined(_STLPORT_VERSION)
+#    define CRYPTOPP_CXX11 1
+#  endif
+#endif
+
+// Hack ahead. Apple's standard library does not have C++'s unique_ptr in C++11.
+// We can't test for unique_ptr directly because some of the non-Apple Clangs
+// on OS X fail the same way. However, modern standard libraries have
+// <forward_list>, so we test for it instead. Thanks to Jonathan Wakely for
+// devising the clever test for modern/ancient versions. TODO: test under
+// Xcode 3, where g++ is really g++.
+#if defined(__APPLE__) && defined(__clang__)
+#  if !(defined(__has_include) && __has_include(<forward_list>))
+#    undef CRYPTOPP_CXX11
+#  endif
+#endif
+
+// C++14 macro version, https://stackoverflow.com/q/26089319/608639
+#if defined(CRYPTOPP_CXX11) && !defined(CRYPTOPP_NO_CXX14)
+#  if ((_MSC_VER >= 1900) || (__cplusplus >= 201402L)) && !defined(_STLPORT_VERSION)
+#    define CRYPTOPP_CXX14 1
+#  endif
+#endif
+
+// C++17 macro version, https://stackoverflow.com/q/38456127/608639
+#if defined(CRYPTOPP_CXX14) && !defined(CRYPTOPP_NO_CXX17)
+#  if ((_MSC_VER >= 1900) || (__cplusplus >= 201703L)) && !defined(_STLPORT_VERSION)
+#    define CRYPTOPP_CXX17 1
+#  endif
+#endif
+
+// ***************** C++11 and above ********************
+
+#if defined(CRYPTOPP_CXX11)
+
+// atomics: MS at VS2012 (17.00); GCC at 4.4; Clang at 3.1/3.2; Intel 13.0; SunCC 5.14.
+#if (CRYPTOPP_MSC_VERSION >= 1700) || __has_feature(cxx_atomic) || \
+	(__INTEL_COMPILER >= 1300) || (CRYPTOPP_GCC_VERSION >= 40400) || (__SUNPRO_CC >= 0x5140)
+# define CRYPTOPP_CXX11_ATOMICS 1
+#endif // atomics
+
+// synchronization: MS at VS2012 (17.00); GCC at 4.4; Clang at 3.3; Xcode 5.0; Intel 12.0; SunCC 5.13.
+// TODO: verify Clang and Intel versions; find __has_feature(x) extension for Clang
+#if (CRYPTOPP_MSC_VERSION >= 1700) || (CRYPTOPP_LLVM_CLANG_VERSION >= 30300) || \
+	(CRYPTOPP_APPLE_CLANG_VERSION >= 50000) || (__INTEL_COMPILER >= 1200) || \
+	(CRYPTOPP_GCC_VERSION >= 40400) || (__SUNPRO_CC >= 0x5130)
+// Hack ahead. New GCC compilers like GCC 6 on AIX 7.0 or earlier as well as original MinGW
+// don't have the synchronization gear. However, Wakely's test used for Apple does not work
+// on the GCC/AIX combination. Another twist is we need other stuff from C++11,
+// like no-except destructors. Dumping preprocessors shows the following may
+// apply: http://stackoverflow.com/q/14191566/608639.
+# include <cstddef>
+# if !defined(__GLIBCXX__) || defined(_GLIBCXX_HAS_GTHREADS)
+#  define CRYPTOPP_CXX11_SYNCHRONIZATION 1
+# endif
+#endif // synchronization
+
+// Dynamic Initialization and Destruction with Concurrency ("Magic Statics")
+// MS at VS2015 with Vista (19.00); GCC at 4.3; LLVM Clang at 2.9; Apple Clang at 4.0; Intel 11.1; SunCC 5.13.
+// Microsoft's implementation only works for Vista and above, so its further
+// limited. http://connect.microsoft.com/VisualStudio/feedback/details/1789709
+#if (__cpp_threadsafe_static_init >= 200806) || \
+	(CRYPTOPP_MSC_VERSION >= 1900) && ((WINVER >= 0x0600) || (_WIN32_WINNT >= 0x0600)) || \
+	(CRYPTOPP_LLVM_CLANG_VERSION >= 20900) || (CRYPTOPP_APPLE_CLANG_VERSION >= 40000)  || \
+	(__INTEL_COMPILER >= 1110) || (CRYPTOPP_GCC_VERSION >= 40300) || (__SUNPRO_CC >= 0x5130)
+# define CRYPTOPP_CXX11_DYNAMIC_INIT 1
+#endif // Dynamic Initialization compilers
+
+// deleted functions: MS at VS2013 (18.00); GCC at 4.3; Clang at 2.9; Intel 12.1; SunCC 5.13.
+#if (CRYPTOPP_MSC_VERSION >= 1800) || (CRYPTOPP_LLVM_CLANG_VERSION >= 20900) || \
+	(CRYPTOPP_APPLE_CLANG_VERSION >= 40000) || (__INTEL_COMPILER >= 1210) || \
+	(CRYPTOPP_GCC_VERSION >= 40300) || (__SUNPRO_CC >= 0x5130)
+# define CRYPTOPP_CXX11_DELETED_FUNCTIONS 1
+#endif // deleted functions
+
+// alignof/alignas: MS at VS2015 (19.00); GCC at 4.8; Clang at 3.0; Intel 15.0; SunCC 5.13.
+#if (CRYPTOPP_MSC_VERSION >= 1900) || __has_feature(cxx_alignas) || \
+	(__INTEL_COMPILER >= 1500) || (CRYPTOPP_GCC_VERSION >= 40800) || (__SUNPRO_CC >= 0x5130)
+#  define CRYPTOPP_CXX11_ALIGNAS 1
+#endif // alignas
+
+// alignof: MS at VS2015 (19.00); GCC at 4.5; Clang at 2.9; Intel 15.0; SunCC 5.13.
+#if (CRYPTOPP_MSC_VERSION >= 1900) || __has_feature(cxx_alignof) || \
+	(__INTEL_COMPILER >= 1500) || (CRYPTOPP_GCC_VERSION >= 40500) || (__SUNPRO_CC >= 0x5130)
+#  define CRYPTOPP_CXX11_ALIGNOF 1
+#endif // alignof
+
+// lambdas: MS at VS2012 (17.00); GCC at 4.9; Clang at 3.3; Intel 12.0; SunCC 5.14.
+#if (CRYPTOPP_MSC_VERSION >= 1700) || __has_feature(cxx_lambdas) || \
+	(__INTEL_COMPILER >= 1200) || (CRYPTOPP_GCC_VERSION >= 40900) || (__SUNPRO_CC >= 0x5140)
+#  define CRYPTOPP_CXX11_LAMBDA 1
+#endif // lambdas
+
+// noexcept: MS at VS2015 (19.00); GCC at 4.6; Clang at 3.0; Intel 14.0; SunCC 5.13.
+#if (CRYPTOPP_MSC_VERSION >= 1900) || __has_feature(cxx_noexcept) || \
+	(__INTEL_COMPILER >= 1400) || (CRYPTOPP_GCC_VERSION >= 40600) || (__SUNPRO_CC >= 0x5130)
+# define CRYPTOPP_CXX11_NOEXCEPT 1
+#endif // noexcept compilers
+
+// variadic templates: MS at VS2013 (18.00); GCC at 4.3; Clang at 2.9; Intel 12.1; SunCC 5.13.
+#if (__cpp_variadic_templates >= 200704) || __has_feature(cxx_variadic_templates) || \
+	(CRYPTOPP_MSC_VERSION >= 1800) || (__INTEL_COMPILER >= 1210) || \
+	(CRYPTOPP_GCC_VERSION >= 40300) || (__SUNPRO_CC >= 0x5130)
+# define CRYPTOPP_CXX11_VARIADIC_TEMPLATES 1
+#endif // variadic templates
+
+// constexpr: MS at VS2015 (19.00); GCC at 4.6; Clang at 3.1; Intel 16.0; SunCC 5.13.
+// Intel has mis-supported the feature since at least ICPC 13.00
+#if (__cpp_constexpr >= 200704) || __has_feature(cxx_constexpr) || \
+	(CRYPTOPP_MSC_VERSION >= 1900) || (__INTEL_COMPILER >= 1600) || \
+	(CRYPTOPP_GCC_VERSION >= 40600) || (__SUNPRO_CC >= 0x5130)
+# define CRYPTOPP_CXX11_CONSTEXPR 1
+#endif // constexpr compilers
+
+// strong typed enums: MS at VS2012 (17.00); GCC at 4.4; Clang at 3.3; Intel 14.0; SunCC 5.12.
+// Mircorosft and Intel had partial support earlier, but we require full support.
+#if (CRYPTOPP_MSC_VERSION >= 1700) || __has_feature(cxx_strong_enums) || \
+	(__INTEL_COMPILER >= 1400) || (CRYPTOPP_GCC_VERSION >= 40400) || (__SUNPRO_CC >= 0x5120)
+# define CRYPTOPP_CXX11_ENUM 1
+#endif // constexpr compilers
+
+// nullptr_t: MS at VS2010 (16.00); GCC at 4.6; Clang at 3.3; Intel 10.0; SunCC 5.13.
+#if (CRYPTOPP_MSC_VERSION >= 1600) || __has_feature(cxx_nullptr) || \
+	(__INTEL_COMPILER >= 1000) || (CRYPTOPP_GCC_VERSION >= 40600) || \
+    (__SUNPRO_CC >= 0x5130) || defined(__IBMCPP_NULLPTR)
+# define CRYPTOPP_CXX11_NULLPTR 1
+#endif // nullptr_t compilers
+
+#endif // CRYPTOPP_CXX11
+
+// ***************** C++14 and above ********************
+
+#if defined(CRYPTOPP_CXX14)
+
+// Extended static_assert with one argument
+// Microsoft cannot handle the single argument static_assert as of VS2019 (cl.exe 19.00)
+#if (__cpp_static_assert >= 201411)
+# define CRYPTOPP_CXX14_STATIC_ASSERT 1
+#endif // static_assert
+
+#endif
+
+// ***************** C++17 and above ********************
+
+// C++17 is available
+#if defined(CRYPTOPP_CXX17)
+
+// C++17 uncaught_exceptions: MS at VS2015 (19.00); GCC at 6.0; Clang at 3.5; Intel 18.0.
+// Clang and __EXCEPTIONS see http://releases.llvm.org/3.6.0/tools/clang/docs/ReleaseNotes.html
+#if defined(__clang__)
+# if __EXCEPTIONS && __has_feature(cxx_exceptions)
+#  if __cpp_lib_uncaught_exceptions
+#   define CRYPTOPP_CXX17_EXCEPTIONS 1
+#  endif
+# endif
+#elif (CRYPTOPP_MSC_VERSION >= 1900) || (__INTEL_COMPILER >= 1800) || \
+      (CRYPTOPP_GCC_VERSION >= 60000) || (__cpp_lib_uncaught_exceptions)
+# define CRYPTOPP_CXX17_EXCEPTIONS 1
+#endif // uncaught_exceptions compilers
+
+#endif  // CRYPTOPP_CXX17
+
+// ***************** C++ fixups ********************
+
+#if defined(CRYPTOPP_CXX11_NOEXCEPT)
+#  define CRYPTOPP_THROW noexcept(false)
+#  define CRYPTOPP_NO_THROW noexcept(true)
+#else
+#  define CRYPTOPP_THROW
+#  define CRYPTOPP_NO_THROW
+#endif // CRYPTOPP_CXX11_NOEXCEPT
+
+// Hack... C++11 nullptr_t type safety and analysis
+#if defined(CRYPTOPP_CXX11_NULLPTR) && !defined(NULLPTR)
+# define NULLPTR nullptr
+#elif !defined(NULLPTR)
+# define NULLPTR NULL
+#endif // CRYPTOPP_CXX11_NULLPTR
+
+#endif  // CRYPTOPP_CONFIG_CXX_H

config_dll.h

@@ -0,0 +1,77 @@
+// config_dll.h - written and placed in public domain by Jeffrey Walton
+//                the bits that make up this source file are from the
+//                library's monolithic config.h.
+
+/// \file config_dll.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+/// \since Crypto++ 8.3
+
+#ifndef CRYPTOPP_CONFIG_DLL_H
+#define CRYPTOPP_CONFIG_DLL_H
+
+#include "config_os.h"
+
+#if !defined(CRYPTOPP_DOXYGEN_PROCESSING)
+
+#if defined(CRYPTOPP_WIN32_AVAILABLE)
+
+#ifdef CRYPTOPP_EXPORTS
+#  define CRYPTOPP_IS_DLL
+#  define CRYPTOPP_DLL __declspec(dllexport)
+#elif defined(CRYPTOPP_IMPORTS)
+#  define CRYPTOPP_IS_DLL
+#  define CRYPTOPP_DLL __declspec(dllimport)
+#else
+#  define CRYPTOPP_DLL
+#endif
+
+// C++ makes const internal linkage
+#define CRYPTOPP_TABLE extern
+#define CRYPTOPP_API __cdecl
+
+#else	// not CRYPTOPP_WIN32_AVAILABLE
+
+// C++ makes const internal linkage
+#define CRYPTOPP_TABLE extern
+#define CRYPTOPP_DLL
+#define CRYPTOPP_API
+
+#endif	// CRYPTOPP_WIN32_AVAILABLE
+
+#if defined(__MWERKS__)
+#  define CRYPTOPP_EXTERN_DLL_TEMPLATE_CLASS extern class CRYPTOPP_DLL
+#elif defined(__BORLANDC__) || defined(__SUNPRO_CC)
+#  define CRYPTOPP_EXTERN_DLL_TEMPLATE_CLASS template class CRYPTOPP_DLL
+#else
+#  define CRYPTOPP_EXTERN_DLL_TEMPLATE_CLASS extern template class CRYPTOPP_DLL
+#endif
+
+#if defined(CRYPTOPP_MANUALLY_INSTANTIATE_TEMPLATES) && !defined(CRYPTOPP_IMPORTS)
+#  define CRYPTOPP_DLL_TEMPLATE_CLASS template class CRYPTOPP_DLL
+#else
+#  define CRYPTOPP_DLL_TEMPLATE_CLASS CRYPTOPP_EXTERN_DLL_TEMPLATE_CLASS
+#endif
+
+#if defined(__MWERKS__)
+#  define CRYPTOPP_EXTERN_STATIC_TEMPLATE_CLASS extern class
+#elif defined(__BORLANDC__) || defined(__SUNPRO_CC)
+#  define CRYPTOPP_EXTERN_STATIC_TEMPLATE_CLASS template class
+#else
+#  define CRYPTOPP_EXTERN_STATIC_TEMPLATE_CLASS extern template class
+#endif
+
+#if defined(CRYPTOPP_MANUALLY_INSTANTIATE_TEMPLATES) && !defined(CRYPTOPP_EXPORTS)
+#  define CRYPTOPP_STATIC_TEMPLATE_CLASS template class
+#else
+#  define CRYPTOPP_STATIC_TEMPLATE_CLASS CRYPTOPP_EXTERN_STATIC_TEMPLATE_CLASS
+#endif
+
+#endif  // CRYPTOPP_DOXYGEN_PROCESSING
+
+#endif  // CRYPTOPP_CONFIG_DLL_H

config_int.h

@@ -0,0 +1,103 @@
+// config_int.h - written and placed in public domain by Jeffrey Walton
+//                the bits that make up this source file are from the
+//                library's monolithic config.h.
+
+/// \file config_int.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+/// \since Crypto++ 8.3
+
+#ifndef CRYPTOPP_CONFIG_INT_H
+#define CRYPTOPP_CONFIG_INT_H
+
+#include "config_ns.h"
+#include "config_ver.h"
+
+// Originally in global namespace to avoid ambiguity with other byte typedefs.
+// Moved to Crypto++ namespace due to C++17, std::byte and potential compile
+// problems. Also see http://www.cryptopp.com/wiki/std::byte and
+// http://github.com/weidai11/cryptopp/issues/442.
+// typedef unsigned char byte;
+#define CRYPTOPP_NO_GLOBAL_BYTE 1
+
+NAMESPACE_BEGIN(CryptoPP)
+
+// Signed words added at Issue 609 for early versions of and Visual Studio and
+// the NaCl gear. Also see https://github.com/weidai11/cryptopp/issues/609.
+
+typedef unsigned char byte;
+typedef unsigned short word16;
+typedef unsigned int word32;
+
+typedef signed char sbyte;
+typedef signed short sword16;
+typedef signed int sword32;
+
+#if defined(_MSC_VER) || defined(__BORLANDC__)
+	typedef signed __int64 sword64;
+	typedef unsigned __int64 word64;
+	#define SW64LIT(x) x##i64
+	#define W64LIT(x) x##ui64
+#elif (_LP64 || __LP64__)
+	typedef signed long sword64;
+	typedef unsigned long word64;
+	#define SW64LIT(x) x##L
+	#define W64LIT(x) x##UL
+#else
+	typedef signed long long sword64;
+	typedef unsigned long long word64;
+	#define SW64LIT(x) x##LL
+	#define W64LIT(x) x##ULL
+#endif
+
+// define large word type, used for file offsets and such
+typedef word64 lword;
+const lword LWORD_MAX = W64LIT(0xffffffffffffffff);
+
+// define hword, word, and dword. these are used for multiprecision integer arithmetic
+// Intel compiler won't have _umul128 until version 10.0. See http://softwarecommunity.intel.com/isn/Community/en-US/forums/thread/30231625.aspx
+#if (defined(_MSC_VER) && (!defined(__INTEL_COMPILER) || __INTEL_COMPILER >= 1000) && (defined(_M_X64) || defined(_M_IA64))) || (defined(__DECCXX) && defined(__alpha__)) || (defined(__INTEL_COMPILER) && defined(__x86_64__)) || (defined(__SUNPRO_CC) && defined(__x86_64__))
+	typedef word32 hword;
+	typedef word64 word;
+#else
+	#define CRYPTOPP_NATIVE_DWORD_AVAILABLE 1
+	#if defined(__alpha__) || defined(__ia64__) || defined(_ARCH_PPC64) || defined(__x86_64__) || defined(__mips64) || defined(__sparc64__)
+		#if ((CRYPTOPP_GCC_VERSION >= 30400) || (CRYPTOPP_LLVM_CLANG_VERSION >= 30000) || (CRYPTOPP_APPLE_CLANG_VERSION >= 40300)) && (__SIZEOF_INT128__ >= 16)
+			// GCC 4.0.1 on MacOS X is missing __umodti3 and __udivti3
+			// GCC 4.8.3 and bad uint128_t ops on PPC64/POWER7 (Issue 421)
+			// mode(TI) division broken on amd64 with GCC earlier than GCC 3.4
+			typedef word32 hword;
+			typedef word64 word;
+			typedef __uint128_t dword;
+			typedef __uint128_t word128;
+			#define CRYPTOPP_WORD128_AVAILABLE 1
+		#else
+			// if we're here, it means we're on a 64-bit CPU but we don't have a way to obtain 128-bit multiplication results
+			typedef word16 hword;
+			typedef word32 word;
+			typedef word64 dword;
+		#endif
+	#else
+		// being here means the native register size is probably 32 bits or less
+		#define CRYPTOPP_BOOL_SLOW_WORD64 1
+		typedef word16 hword;
+		typedef word32 word;
+		typedef word64 dword;
+	#endif
+#endif
+
+#ifndef CRYPTOPP_BOOL_SLOW_WORD64
+# define CRYPTOPP_BOOL_SLOW_WORD64 0
+#endif
+
+const unsigned int WORD_SIZE = sizeof(word);
+const unsigned int WORD_BITS = WORD_SIZE * 8;
+
+NAMESPACE_END
+
+#endif  // CRYPTOPP_CONFIG_INT_H

config_misc.h

@@ -0,0 +1,180 @@
+// config_misc.h - written and placed in public domain by Jeffrey Walton
+//                 the bits that make up this source file are from the
+//                 library's monolithic config.h.
+
+/// \file config_misc.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+/// \since Crypto++ 8.3
+
+#ifndef CRYPTOPP_CONFIG_MISC_H
+#define CRYPTOPP_CONFIG_MISC_H
+
+#include "config_asm.h"
+#include "config_cxx.h"
+#include "config_os.h"
+#include "config_ver.h"
+
+// Define this if running on a big-endian CPU
+// big endian will be assumed if CRYPTOPP_LITTLE_ENDIAN is not non-0
+#if !defined(CRYPTOPP_LITTLE_ENDIAN) && !defined(CRYPTOPP_BIG_ENDIAN) && (defined(__BIG_ENDIAN__) || (defined(__s390__) || defined(__s390x__) || defined(__zarch__)) || (defined(__m68k__) || defined(__MC68K__)) || defined(__sparc) || defined(__sparc__) || defined(__hppa__) || defined(__MIPSEB__) || defined(__ARMEB__) || (defined(__MWERKS__) && !defined(__INTEL__)))
+#	define CRYPTOPP_BIG_ENDIAN 1
+#endif
+
+// Define this if running on a little-endian CPU
+// big endian will be assumed if CRYPTOPP_LITTLE_ENDIAN is not non-0
+#if !defined(CRYPTOPP_BIG_ENDIAN) && !defined(CRYPTOPP_LITTLE_ENDIAN)
+#	define CRYPTOPP_LITTLE_ENDIAN 1
+#endif
+
+// Define this if you want to set a prefix for TestData/ and TestVectors/
+// Be sure to add the trailing slash since its simple concatenation.
+// After https://github.com/weidai11/cryptopp/issues/760 the library
+// should find the test vectors and data without much effort. It
+// will search in "./" and "$ORIGIN/../share/cryptopp" automatically.
+#ifndef CRYPTOPP_DATA_DIR
+# define CRYPTOPP_DATA_DIR ""
+#endif
+
+// Define this to disable the test suite from searching for test
+// vectors and data in "./" and "$ORIGIN/../share/cryptopp". The
+// library will still search in CRYPTOPP_DATA_DIR, regardless.
+// Some distros may want to disable this feature. Also see
+// https://github.com/weidai11/cryptopp/issues/760
+// #ifndef CRYPTOPP_DISABLE_DATA_DIR_SEARCH
+// # define CRYPTOPP_DISABLE_DATA_DIR_SEARCH
+// #endif
+
+// Define this if you want or need the library's memcpy_s and memmove_s.
+// See http://github.com/weidai11/cryptopp/issues/28.
+// #if !defined(CRYPTOPP_WANT_SECURE_LIB)
+// # define CRYPTOPP_WANT_SECURE_LIB
+// #endif
+
+// Define this if ARMv8 shifts are slow. ARM Cortex-A53 and Cortex-A57 shift
+// operation perform poorly, so NEON and ASIMD code that relies on shifts
+// or rotates often performs worse than C/C++ code. Also see
+// http://github.com/weidai11/cryptopp/issues/367.
+#define CRYPTOPP_SLOW_ARMV8_SHIFT 1
+
+// CRYPTOPP_DEBUG enables the library's CRYPTOPP_ASSERT. CRYPTOPP_ASSERT
+// raises a SIGTRAP (Unix) or calls DebugBreak() (Windows). CRYPTOPP_ASSERT
+// is only in effect when CRYPTOPP_DEBUG, DEBUG or _DEBUG is defined. Unlike
+// Posix assert, CRYPTOPP_ASSERT is not affected by NDEBUG (or failure to
+// define it).
+// Also see http://github.com/weidai11/cryptopp/issues/277, CVE-2016-7420
+#if (defined(DEBUG) || defined(_DEBUG)) && !defined(CRYPTOPP_DEBUG)
+# define CRYPTOPP_DEBUG 1
+#endif
+
+// File system code to use when creating GZIP archive.
+// http://www.gzip.org/format.txt
+#if !defined(GZIP_OS_CODE)
+# if defined(__macintosh__)
+#  define GZIP_OS_CODE 7
+# elif defined(__unix__) || defined(__linux__)
+#  define GZIP_OS_CODE 3
+# else
+#  define GZIP_OS_CODE 0
+# endif
+#endif
+
+// Try this if your CPU has 256K internal cache or a slow multiply instruction
+// and you want a (possibly) faster IDEA implementation using log tables
+// #define IDEA_LARGECACHE
+
+// Define this if, for the linear congruential RNG, you want to use
+// the original constants as specified in S.K. Park and K.W. Miller's
+// CACM paper.
+// #define LCRNG_ORIGINAL_NUMBERS
+
+// Define this if you want Integer's operator<< to honor std::showbase (and
+// std::noshowbase). If defined, Integer will use a suffix of 'b', 'o', 'h'
+// or '.' (the last for decimal) when std::showbase is in effect. If
+// std::noshowbase is set, then the suffix is not added to the Integer. If
+// not defined, existing behavior is preserved and Integer will use a suffix
+// of 'b', 'o', 'h' or '.' (the last for decimal).
+// #define CRYPTOPP_USE_STD_SHOWBASE
+
+// Define this if you want to decouple AlgorithmParameters and Integer
+// The decoupling should make it easier for the linker to remove Integer
+// related code for those who do not need Integer, and avoid a potential
+// race during AssignIntToInteger pointer initialization. Also
+// see http://github.com/weidai11/cryptopp/issues/389.
+// #define CRYPTOPP_NO_ASSIGN_TO_INTEGER
+
+// Need GCC 4.6/Clang 1.7/Apple Clang 2.0 or above due to "GCC diagnostic {push|pop}"
+#if (CRYPTOPP_GCC_VERSION >= 40600) || (CRYPTOPP_LLVM_CLANG_VERSION >= 10700) || \
+    (CRYPTOPP_APPLE_CLANG_VERSION >= 20000)
+	#define CRYPTOPP_GCC_DIAGNOSTIC_AVAILABLE 1
+#endif
+
+// Portable way to suppress warnings.
+// Moved from misc.h due to circular depenedencies.
+#ifndef CRYPTOPP_UNUSED
+	#define CRYPTOPP_UNUSED(x) ((void)(x))
+#endif
+
+// how to disable inlining
+#if defined(_MSC_VER)
+#	define CRYPTOPP_NOINLINE_DOTDOTDOT
+#	define CRYPTOPP_NOINLINE __declspec(noinline)
+#elif defined(__xlc__) || defined(__xlC__) || defined(__ibmxl__)
+#	define CRYPTOPP_NOINLINE_DOTDOTDOT ...
+#	define CRYPTOPP_NOINLINE __attribute__((noinline))
+#elif defined(__GNUC__)
+#	define CRYPTOPP_NOINLINE_DOTDOTDOT
+#	define CRYPTOPP_NOINLINE __attribute__((noinline))
+#else
+#	define CRYPTOPP_NOINLINE_DOTDOTDOT ...
+#	define CRYPTOPP_NOINLINE
+#endif
+
+// http://stackoverflow.com/a/13867690/608639
+#if defined(CRYPTOPP_CXX11_CONSTEXPR)
+#  define CRYPTOPP_STATIC_CONSTEXPR static constexpr
+#  define CRYPTOPP_CONSTEXPR constexpr
+#else
+#  define CRYPTOPP_STATIC_CONSTEXPR static
+#  define CRYPTOPP_CONSTEXPR
+#endif // CRYPTOPP_CXX11_CONSTEXPR
+
+#if defined(CRYPTOPP_DOXYGEN_PROCESSING)
+# define CRYPTOPP_CONSTANT(x) static const int x;
+#elif defined(CRYPTOPP_CXX11_ENUM)
+# define CRYPTOPP_CONSTANT(x) enum : int { x };
+#elif defined(CRYPTOPP_CXX11_CONSTEXPR)
+# define CRYPTOPP_CONSTANT(x) constexpr static int x;
+#else
+# define CRYPTOPP_CONSTANT(x) static const int x;
+#endif
+
+// Warnings
+#ifdef _MSC_VER
+	// 4127: conditional expression is constant
+	// 4512: assignment operator not generated
+	// 4661: no suitable definition provided for explicit template instantiation request
+	// 4910: '__declspec(dllexport)' and 'extern' are incompatible on an explicit instantiation
+#	pragma warning(disable: 4127 4512 4661 4910)
+	// Security related, possible defects
+	// http://blogs.msdn.com/b/vcblog/archive/2010/12/14/off-by-default-compiler-warnings-in-visual-c.aspx
+#	pragma warning(once: 4191 4242 4263 4264 4266 4302 4826 4905 4906 4928)
+#endif
+
+#ifdef __BORLANDC__
+// 8037: non-const function called for const object. needed to work around BCB2006 bug
+#	pragma warn -8037
+#endif
+
+// [GCC Bug 53431] "C++ preprocessor ignores #pragma GCC diagnostic". Clang honors it.
+#if CRYPTOPP_GCC_DIAGNOSTIC_AVAILABLE
+# pragma GCC diagnostic ignored "-Wunknown-pragmas"
+# pragma GCC diagnostic ignored "-Wunused-function"
+#endif
+
+#endif  // CRYPTOPP_CONFIG_MISC_H

config_ns.h

@@ -0,0 +1,68 @@
+// config_ns.h - written and placed in public domain by Jeffrey Walton
+//               the bits that make up this source file are from the
+//               library's monolithic config.h.
+
+/// \file config_ns.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+/// \since Crypto++ 8.3
+
+#ifndef CRYPTOPP_CONFIG_NAMESPACE_H
+#define CRYPTOPP_CONFIG_NAMESPACE_H
+
+// namespace support is now required
+#ifdef NO_NAMESPACE
+# error namespace support is now required
+#endif
+
+#ifdef CRYPTOPP_DOXYGEN_PROCESSING
+
+/// \namespace CryptoPP
+/// \brief Crypto++ library namespace
+/// \details Nearly all classes are located in the CryptoPP namespace. Within
+/// the namespace, there are two additional namespaces.
+///   <ul>
+///     <li>Name - namespace for names used with NameValuePairs and documented
+///         in argnames.h
+///     <li>NaCl - namespace for NaCl test functions like crypto_box,
+///         crypto_box_open, crypto_sign, and crypto_sign_open
+///     <li>Donna - namespace for curve25519 library operations. The name was
+///         selected due to use of Langley and Moon's curve25519-donna.
+///     <li>Test - namespace for testing and benchmarks classes
+///     <li>Weak - namespace for weak and wounded algorithms, like ARC4, MD5
+///         and Pananma
+///   </ul>
+namespace CryptoPP { }
+
+// Bring in the symbols found in the weak namespace; and fold Weak1 into Weak
+#define CRYPTOPP_ENABLE_NAMESPACE_WEAK 1
+#define Weak1 Weak
+// Avoid putting "CryptoPP::" in front of everything in Doxygen output
+#define CryptoPP
+#define NAMESPACE_BEGIN(x)
+#define NAMESPACE_END
+// Get Doxygen to generate better documentation for these typedefs
+#define DOCUMENTED_TYPEDEF(x, y) class y : public x {};
+// Make "protected" "private" so the functions and members are not documented
+#define protected private
+
+#else
+// Not Doxygen
+#define NAMESPACE_BEGIN(x) namespace x {
+#define NAMESPACE_END }
+#define DOCUMENTED_TYPEDEF(x, y) typedef x y;
+
+#endif  // CRYPTOPP_DOXYGEN_PROCESSING
+
+#define ANONYMOUS_NAMESPACE_BEGIN namespace {
+#define ANONYMOUS_NAMESPACE_END }
+#define USING_NAMESPACE(x) using namespace x;
+#define DOCUMENTED_NAMESPACE_BEGIN(x) namespace x {
+#define DOCUMENTED_NAMESPACE_END }
+
+#endif  // CRYPTOPP_CONFIG_NAMESPACE_H

config_os.h

@@ -0,0 +1,152 @@
+// config_os.h - written and placed in public domain by Jeffrey Walton
+//               the bits that make up this source file are from the
+//               library's monolithic config.h.
+
+/// \file config_os.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+/// \since Crypto++ 8.3
+
+#ifndef CRYPTOPP_CONFIG_OS_H
+#define CRYPTOPP_CONFIG_OS_H
+
+#include "config_ver.h"
+
+// Windows platform
+#if defined(_WIN32) || defined(_WIN64) || defined(__CYGWIN__)
+#define CRYPTOPP_WIN32_AVAILABLE
+#endif
+
+// Unix and Linux platforms
+#if defined(__unix__) || defined(__MACH__) || defined(__NetBSD__) || defined(__sun)
+#define CRYPTOPP_UNIX_AVAILABLE
+#endif
+
+// BSD platforms
+#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || defined(__DragonFly__)
+#define CRYPTOPP_BSD_AVAILABLE
+#endif
+
+// Microsoft compilers
+#if defined(_MSC_VER) || defined(__fastcall)
+	#define CRYPTOPP_FASTCALL __fastcall
+#else
+	#define CRYPTOPP_FASTCALL
+#endif
+
+// Microsoft compilers
+#if defined(_MSC_VER)
+	#define CRYPTOPP_NO_VTABLE __declspec(novtable)
+#else
+	#define CRYPTOPP_NO_VTABLE
+#endif
+
+// Define this if you want to disable all OS-dependent features,
+// such as sockets and OS-provided random number generators
+// #define NO_OS_DEPENDENCE
+
+// Define this to use features provided by Microsoft's CryptoAPI.
+// Currently the only feature used is Windows random number generation.
+// This macro will be ignored if NO_OS_DEPENDENCE is defined.
+// #define USE_MS_CRYPTOAPI
+
+// Define this to use features provided by Microsoft's CryptoNG API.
+// CryptoNG API is available in Vista and above and its cross platform,
+// including desktop apps and store apps. Currently the only feature
+// used is Windows random number generation.
+// This macro will be ignored if NO_OS_DEPENDENCE is defined.
+// #define USE_MS_CNGAPI
+
+// If the user did not make a choice, then select CryptoNG if
+// targeting Windows 8 or above.
+#if !defined(USE_MS_CRYPTOAPI) && !defined(USE_MS_CNGAPI)
+# if !defined(_USING_V110_SDK71_) && ((WINVER >= 0x0602 /*_WIN32_WINNT_WIN8*/) || \
+     (_WIN32_WINNT >= 0x0602 /*_WIN32_WINNT_WIN8*/))
+#  define USE_MS_CNGAPI
+# else
+#  define USE_MS_CRYPTOAPI
+# endif
+#endif
+
+// Begin OS features, like init priorities and random numbers
+#ifndef NO_OS_DEPENDENCE
+
+// CRYPTOPP_INIT_PRIORITY attempts to manage initialization of C++ static objects.
+// Under GCC, the library uses init_priority attribute in the range
+// [CRYPTOPP_INIT_PRIORITY, CRYPTOPP_INIT_PRIORITY+100]. Under Windows,
+// CRYPTOPP_INIT_PRIORITY enlists "#pragma init_seg(lib)". The platforms
+// with gaps are Apple and Sun because they require linker scripts. Apple and
+// Sun will use the library's Singletons to initialize and acquire resources.
+// Also see http://cryptopp.com/wiki/Static_Initialization_Order_Fiasco
+#ifndef CRYPTOPP_INIT_PRIORITY
+# define CRYPTOPP_INIT_PRIORITY 250
+#endif
+
+// CRYPTOPP_USER_PRIORITY is for other libraries and user code that is using Crypto++
+// and managing C++ static object creation. It is guaranteed not to conflict with
+// values used by (or would be used by) the Crypto++ library.
+#ifndef CRYPTOPP_USER_PRIORITY
+# define CRYPTOPP_USER_PRIORITY (CRYPTOPP_INIT_PRIORITY+101)
+#endif
+
+// Most platforms allow us to specify when to create C++ objects. Apple and Sun do not.
+#if (CRYPTOPP_INIT_PRIORITY > 0) && !(defined(NO_OS_DEPENDENCE) || defined(__APPLE__) || defined(__sun__))
+# if (CRYPTOPP_GCC_VERSION >= 30000) || (CRYPTOPP_LLVM_CLANG_VERSION >= 20900) || (_INTEL_COMPILER >= 800)
+#  define HAVE_GCC_INIT_PRIORITY 1
+# elif (CRYPTOPP_MSC_VERSION >= 1310)
+#  define HAVE_MSC_INIT_PRIORITY 1
+# elif defined(__xlc__) || defined(__xlC__) || defined(__ibmxl__)
+#  define HAVE_XLC_INIT_PRIORITY 1
+# endif
+#endif  // CRYPTOPP_INIT_PRIORITY, NO_OS_DEPENDENCE, Apple, Sun
+
+#if defined(CRYPTOPP_WIN32_AVAILABLE) || defined(CRYPTOPP_UNIX_AVAILABLE)
+#	define HIGHRES_TIMER_AVAILABLE
+#endif
+
+#ifdef CRYPTOPP_WIN32_AVAILABLE
+# if !defined(WINAPI_FAMILY)
+#	define THREAD_TIMER_AVAILABLE
+# elif defined(WINAPI_FAMILY)
+#   if (WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP))
+#	  define THREAD_TIMER_AVAILABLE
+#  endif
+# endif
+#endif
+
+#if defined(CRYPTOPP_UNIX_AVAILABLE) || defined(CRYPTOPP_DOXYGEN_PROCESSING)
+#	define NONBLOCKING_RNG_AVAILABLE
+#	define BLOCKING_RNG_AVAILABLE
+#	define OS_RNG_AVAILABLE
+#endif
+
+// Cygwin/Newlib requires _XOPEN_SOURCE=600
+#if defined(CRYPTOPP_UNIX_AVAILABLE)
+# define UNIX_SIGNALS_AVAILABLE 1
+#endif
+
+#ifdef CRYPTOPP_WIN32_AVAILABLE
+# if !defined(WINAPI_FAMILY)
+#	define NONBLOCKING_RNG_AVAILABLE
+#	define OS_RNG_AVAILABLE
+# elif defined(WINAPI_FAMILY)
+#   if (WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP))
+#	  define NONBLOCKING_RNG_AVAILABLE
+#	  define OS_RNG_AVAILABLE
+#   elif !(WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP))
+#     if ((WINVER >= 0x0A00 /*_WIN32_WINNT_WIN10*/) || (_WIN32_WINNT >= 0x0A00 /*_WIN32_WINNT_WIN10*/))
+#	    define NONBLOCKING_RNG_AVAILABLE
+#	    define OS_RNG_AVAILABLE
+#     endif
+#   endif
+# endif
+#endif
+
+#endif	// NO_OS_DEPENDENCE
+
+#endif  // CRYPTOPP_CONFIG_OS_H

config_ver.h

@@ -0,0 +1,46 @@
+// config_ver.h - written and placed in public domain by Jeffrey Walton
+//                the bits that make up this source file are from the
+//                library's monolithic config.h.
+
+/// \file config_ver.h
+/// \brief Library configuration file
+/// \details <tt>config.h</tt> was split into components in May 2019 to better
+///  integrate with Autoconf and its feature tests. The splitting occured so
+///  users could continue to include <tt>config.h</tt> while allowing Autoconf
+///  to write new <tt>config_asm.h</tt> and new <tt>config_cxx.h</tt> using
+///  its feature tests.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/835">Issue 835</A>
+/// \since Crypto++ 8.3
+
+#ifndef CRYPTOPP_CONFIG_VERSION_H
+#define CRYPTOPP_CONFIG_VERSION_H
+
+// Library version macro. Since this macro is in a header, it reflects
+// the version of the library the headers came from. It is not
+// necessarily the version of the library built as a shared object if
+// versions are inadvertently mixed and matched.
+#define CRYPTOPP_MAJOR 8
+#define CRYPTOPP_MINOR 3
+#define CRYPTOPP_REVISION 0
+#define CRYPTOPP_VERSION 830
+
+#ifdef __GNUC__
+# define CRYPTOPP_GCC_VERSION (__GNUC__ * 10000 + __GNUC_MINOR__ * 100 + __GNUC_PATCHLEVEL__)
+#endif
+
+#if defined(__xlc__) || defined(__xlC__)
+# define CRYPTOPP_XLC_VERSION ((__xlC__ / 256) * 10000 + (__xlC__ % 256) * 100)
+#endif
+
+// Apple and LLVM's Clang. Apple Clang version 7.0 roughly equals LLVM Clang version 3.7
+#if defined(__clang__) && defined(__apple_build_version__)
+# define CRYPTOPP_APPLE_CLANG_VERSION (__clang_major__ * 10000 + __clang_minor__ * 100 + __clang_patchlevel__)
+#elif defined(__clang__)
+# define CRYPTOPP_LLVM_CLANG_VERSION  (__clang_major__ * 10000 + __clang_minor__ * 100 + __clang_patchlevel__)
+#endif
+
+#ifdef _MSC_VER
+# define CRYPTOPP_MSC_VERSION (_MSC_VER)
+#endif
+
+#endif  // CRYPTOPP_CONFIG_VERSION_H

cpu.cpp

@@ -58,7 +58,7 @@ unsigned long int getauxval(unsigned long int) { return 0; }
 
 // Visual Studio 2008 and below is missing _xgetbv. See x64dll.asm for the body.
 #if defined(_MSC_VER) && _MSC_VER <= 1500 && defined(_M_X64)
-extern "C" unsigned long long __fastcall ExtendedControlRegister(unsigned int);
+extern "C" unsigned long long __fastcall XGETBV64(unsigned int);
 #endif
 
 ANONYMOUS_NAMESPACE_BEGIN
@@ -109,6 +109,8 @@ bool IsAppleMachineARMv8(unsigned int device, unsigned int version)
 
 bool IsAppleMachineARMv84(unsigned int device, unsigned int version)
 {
+    CRYPTOPP_UNUSED(device);
+    CRYPTOPP_UNUSED(version);
 	return false;
 }
 #endif  // __APPLE__
@@ -274,10 +276,14 @@ static inline bool IsIntel(const word32 output[4])
 
 static inline bool IsAMD(const word32 output[4])
 {
-	// This is the "AuthenticAMD" string. Some early K5's can return "AMDisbetter!"
+	// This is the "AuthenticAMD" string.
-	return (output[1] /*EBX*/ == 0x68747541) &&
+	return ((output[1] /*EBX*/ == 0x68747541) &&
 		(output[2] /*ECX*/ == 0x444D4163) &&
-		(output[3] /*EDX*/ == 0x69746E65);
+		(output[3] /*EDX*/ == 0x69746E65)) ||
+		// Some early K5's can return "AMDisbetter!"
+		((output[1] /*EBX*/ == 0x69444d41) &&
+		(output[2] /*ECX*/ == 0x74656273) &&
+		(output[3] /*EDX*/ == 0x21726574));
 }
 
 static inline bool IsHygon(const word32 output[4])
@@ -290,10 +296,14 @@ static inline bool IsHygon(const word32 output[4])
 
 static inline bool IsVIA(const word32 output[4])
 {
-	// This is the "CentaurHauls" string. Some non-PadLock's can return "VIA VIA VIA "
+	// This is the "CentaurHauls" string.
-	return (output[1] /*EBX*/ == 0x746e6543) &&
+	return ((output[1] /*EBX*/ == 0x746e6543) &&
 		(output[2] /*ECX*/ == 0x736c7561) &&
-		(output[3] /*EDX*/ == 0x48727561);
+		(output[3] /*EDX*/ == 0x48727561)) ||
+		// Some non-PadLock's return "VIA VIA VIA "
+		((output[1] /*EBX*/ == 0x32414956) &&
+		(output[2] /*ECX*/ == 0x32414956) &&
+		(output[3] /*EDX*/ == 0x32414956));
 }
 
 void DetectX86Features()
@@ -341,8 +351,8 @@ void DetectX86Features()
 		word64 xcr0 = a | static_cast<word64>(d) << 32;
 		g_hasAVX = (xcr0 & YMM_FLAG) == YMM_FLAG;
 
-// Visual Studio 2008 and below lack xgetbv
+// Visual Studio 2010 and below lack xgetbv
-#elif defined(_MSC_VER) && _MSC_VER <= 1500 && defined(_M_IX86)
+#elif defined(_MSC_VER) && _MSC_VER <= 1600 && defined(_M_IX86)
 		word32 a=0, d=0;
 		__asm {
 			push eax
@@ -363,7 +373,7 @@ void DetectX86Features()
 
 // Visual Studio 2008 and below lack xgetbv
 #elif defined(_MSC_VER) && _MSC_VER <= 1500 && defined(_M_X64)
-		word64 xcr0 = ExtendedControlRegister(0);
+		word64 xcr0 = XGETBV64(0);
 		g_hasAVX = (xcr0 & YMM_FLAG) == YMM_FLAG;
 
 // Downlevel SunCC
@@ -425,6 +435,7 @@ void DetectX86Features()
 	}
 	else if (IsVIA(cpuid0))
 	{
+		// Two bits: available and enabled
 		CRYPTOPP_CONSTANT( RNG_FLAGS = (0x3 << 2))
 		CRYPTOPP_CONSTANT( ACE_FLAGS = (0x3 << 6))
 		CRYPTOPP_CONSTANT(ACE2_FLAGS = (0x3 << 8))
@@ -432,9 +443,10 @@ void DetectX86Features()
 		CRYPTOPP_CONSTANT( PMM_FLAGS = (0x3 << 12))
 
 		CpuId(0xC0000000, 0, cpuid2);
-		if (cpuid2[0] >= 0xC0000001)
+		word32 extendedFeatures = cpuid2[0];
+
+		if (extendedFeatures >= 0xC0000001)
 		{
-			// Extended features available
 			CpuId(0xC0000001, 0, cpuid2);
 			g_hasPadlockRNG  = (cpuid2[3] /*EDX*/ & RNG_FLAGS) == RNG_FLAGS;
 			g_hasPadlockACE  = (cpuid2[3] /*EDX*/ & ACE_FLAGS) == ACE_FLAGS;
@@ -442,7 +454,21 @@ void DetectX86Features()
 			g_hasPadlockPHE  = (cpuid2[3] /*EDX*/ & PHE_FLAGS) == PHE_FLAGS;
 			g_hasPadlockPMM  = (cpuid2[3] /*EDX*/ & PMM_FLAGS) == PMM_FLAGS;
 		}
+
+		if (extendedFeatures >= 0xC0000005)
+		{
+			CpuId(0xC0000005, 0, cpuid2);
+			g_cacheLineSize = GETBYTE(cpuid2[2] /*ECX*/, 0);
 		}
+	}
+
+#if defined(_SC_LEVEL1_DCACHE_LINESIZE)
+	// Glibc does not implement on some platforms. The runtime returns 0 instead of error.
+	// https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/posix/sysconf.c
+	int cacheLineSize = sysconf(_SC_LEVEL1_DCACHE_LINESIZE);
+	if (g_cacheLineSize == 0 && cacheLineSize > 0)
+		g_cacheLineSize = cacheLineSize;
+#endif
 
 	if (g_cacheLineSize == 0)
 		g_cacheLineSize = CRYPTOPP_L1_CACHE_LINE_SIZE;
@@ -474,7 +500,7 @@ word32 CRYPTOPP_SECTION_INIT g_cacheLineSize = CRYPTOPP_L1_CACHE_LINE_SIZE;
 //   then *Probe* the cpu executing an instruction and an observe a SIGILL if unsupported.
 // The probes are in source files where compilation options like -march=armv8-a+crc make
 //   intrinsics available. They are expensive when compared to a standard OS feature query.
-//   Always perform the feature quesry first. For Linux see
+//   Always perform the feature query first. For Linux see
 //   http://sourceware.org/ml/libc-help/2017-08/msg00012.html
 // Avoid probes on Apple platforms because Apple's signal handling for SIGILLs appears broken.
 //   We are trying to figure out a way to feature test without probes. Also see
@@ -486,13 +512,14 @@ extern bool CPU_ProbeNEON();
 extern bool CPU_ProbeCRC32();
 extern bool CPU_ProbeAES();
 extern bool CPU_ProbeSHA1();
-extern bool CPU_ProbeSHA2();
+extern bool CPU_ProbeSHA256();
 extern bool CPU_ProbeSHA512();
 extern bool CPU_ProbeSHA3();
 extern bool CPU_ProbeSM3();
 extern bool CPU_ProbeSM4();
 extern bool CPU_ProbePMULL();
 
+// https://github.com/torvalds/linux/blob/master/arch/arm/include/uapi/asm/hwcap.h
 // https://github.com/torvalds/linux/blob/master/arch/arm64/include/uapi/asm/hwcap.h
 #ifndef HWCAP_ARMv7
 # define HWCAP_ARMv7 (1 << 29)
@@ -500,8 +527,8 @@ extern bool CPU_ProbePMULL();
 #ifndef HWCAP_ASIMD
 # define HWCAP_ASIMD (1 << 1)
 #endif
-#ifndef HWCAP_ARM_NEON
+#ifndef HWCAP_NEON
-# define HWCAP_ARM_NEON 4096
+# define HWCAP_NEON (1 << 12)
 #endif
 #ifndef HWCAP_CRC32
 # define HWCAP_CRC32 (1 << 7)
@@ -548,15 +575,13 @@ extern bool CPU_ProbePMULL();
 
 inline bool CPU_QueryARMv7()
 {
-#if defined(__aarch32__) || defined(__aarch64__)
+#if defined(__ANDROID__) && defined(__arm__)
-	// ARMv7 or above
-	return true;
-#elif defined(__ANDROID__) && defined(__arm__)
 	if (((android_getCpuFamily() & ANDROID_CPU_FAMILY_ARM) != 0) &&
 		((android_getCpuFeatures() & ANDROID_CPU_ARM_FEATURE_ARMv7) != 0))
 		return true;
 #elif defined(__linux__) && defined(__arm__)
-	if ((getauxval(AT_HWCAP) & HWCAP_ARMv7) != 0)
+	if ((getauxval(AT_HWCAP) & HWCAP_ARMv7) != 0 ||
+	    (getauxval(AT_HWCAP) & HWCAP_NEON) != 0)
 		return true;
 #elif defined(__APPLE__) && defined(__arm__)
 	// Apple hardware is ARMv7 or above.
@@ -582,7 +607,7 @@ inline bool CPU_QueryNEON()
 	if ((getauxval(AT_HWCAP2) & HWCAP2_ASIMD) != 0)
 		return true;
 #elif defined(__linux__) && defined(__arm__)
-	if ((getauxval(AT_HWCAP) & HWCAP_ARM_NEON) != 0)
+	if ((getauxval(AT_HWCAP) & HWCAP_NEON) != 0)
 		return true;
 #elif defined(__APPLE__) && defined(__aarch64__)
 	// Core feature set for Aarch32 and Aarch64.
@@ -685,7 +710,7 @@ inline bool CPU_QuerySHA1()
 	return false;
 }
 
-inline bool CPU_QuerySHA2()
+inline bool CPU_QuerySHA256()
 {
 #if defined(__ANDROID__) && defined(__aarch64__)
 	if (((android_getCpuFamily() & ANDROID_CPU_FAMILY_ARM64) != 0) &&
@@ -819,13 +844,13 @@ void DetectArmFeatures()
 	g_hasPMULL = CPU_QueryPMULL() || CPU_ProbePMULL();
 	g_hasAES  = CPU_QueryAES() || CPU_ProbeAES();
 	g_hasSHA1 = CPU_QuerySHA1() || CPU_ProbeSHA1();
-	g_hasSHA2 = CPU_QuerySHA2() || CPU_ProbeSHA2();
+	g_hasSHA2 = CPU_QuerySHA256() || CPU_ProbeSHA256();
 	g_hasSHA512 = CPU_QuerySHA512(); // || CPU_ProbeSHA512();
 	g_hasSHA3 = CPU_QuerySHA3(); // || CPU_ProbeSHA3();
 	g_hasSM3 = CPU_QuerySM3(); // || CPU_ProbeSM3();
 	g_hasSM4 = CPU_QuerySM4(); // || CPU_ProbeSM4();
 
-#if defined(__linux__) && defined(_SC_LEVEL1_DCACHE_LINESIZE)
+#if defined(_SC_LEVEL1_DCACHE_LINESIZE)
 	// Glibc does not implement on some platforms. The runtime returns 0 instead of error.
 	// https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/posix/sysconf.c
 	int cacheLineSize = sysconf(_SC_LEVEL1_DCACHE_LINESIZE);
@@ -833,6 +858,9 @@ void DetectArmFeatures()
 		g_cacheLineSize = cacheLineSize;
 #endif
 
+	if (g_cacheLineSize == 0)
+		g_cacheLineSize = CRYPTOPP_L1_CACHE_LINE_SIZE;
+
 	*const_cast<volatile bool*>(&g_ArmDetectionDone) = true;
 }
 
@@ -862,24 +890,6 @@ extern bool CPU_ProbeSHA256();
 extern bool CPU_ProbeSHA512();
 extern bool CPU_ProbeDARN();
 
-// Linux define values from 64-Bit ELF V2 ABI Specification.
-// http://openpowerfoundation.org/wp-content/uploads/resources/leabi/content/ch_preface.html
-#ifndef PPC_FEATURE_HAS_ALTIVEC
-# define PPC_FEATURE_HAS_ALTIVEC  0x10000000
-#endif
-#ifndef PPC_FEATURE_ARCH_2_06
-# define PPC_FEATURE_ARCH_2_06    0x00000100
-#endif
-#ifndef PPC_FEATURE2_ARCH_2_07
-# define PPC_FEATURE2_ARCH_2_07   0x80000000
-#endif
-#ifndef PPC_FEATURE2_ARCH_3_00
-# define PPC_FEATURE2_ARCH_3_00   0x00800000
-#endif
-#ifndef PPC_FEATURE2_VEC_CRYPTO
-# define PPC_FEATURE2_VEC_CRYPTO  0x02000000
-#endif
-
 // AIX defines. We used to just call __power_7_andup()
 // and friends but at Power9, too many compilers were
 // missing __power_9_andup(). Instead we switched to
@@ -901,7 +911,7 @@ extern bool CPU_ProbeDARN();
 // was available much earlier for other vendors.
 inline bool CPU_QueryAltivec()
 {
-#if defined(__linux__)
+#if defined(__linux__) && defined(PPC_FEATURE_HAS_ALTIVEC)
 	if ((getauxval(AT_HWCAP) & PPC_FEATURE_HAS_ALTIVEC) != 0)
 		return true;
 #elif defined(_AIX)
@@ -918,7 +928,7 @@ inline bool CPU_QueryAltivec()
 inline bool CPU_QueryPower7()
 {
 	// Power7 and ISA 2.06
-#if defined(__linux__)
+#if defined(__linux__) && defined(PPC_FEATURE_ARCH_2_06)
 	if ((getauxval(AT_HWCAP) & PPC_FEATURE_ARCH_2_06) != 0)
 		return true;
 #elif defined(_AIX)
@@ -931,7 +941,7 @@ inline bool CPU_QueryPower7()
 inline bool CPU_QueryPower8()
 {
 	// Power8 and ISA 2.07 provide in-core crypto.
-#if defined(__linux__)
+#if defined(__linux__) && defined(PPC_FEATURE2_ARCH_2_07)
 	if ((getauxval(AT_HWCAP2) & PPC_FEATURE2_ARCH_2_07) != 0)
 		return true;
 #elif defined(_AIX)
@@ -944,7 +954,7 @@ inline bool CPU_QueryPower8()
 inline bool CPU_QueryPower9()
 {
 	// Power9 and ISA 3.0.
-#if defined(__linux__)
+#if defined(__linux__) && defined(PPC_FEATURE2_ARCH_3_00)
 	if ((getauxval(AT_HWCAP2) & PPC_FEATURE2_ARCH_3_00) != 0)
 		return true;
 #elif defined(_AIX)
@@ -958,7 +968,7 @@ inline bool CPU_QueryAES()
 {
 	// Power8 and ISA 2.07 provide in-core crypto. Glibc
 	// 2.24 or higher is required for PPC_FEATURE2_VEC_CRYPTO.
-#if defined(__linux__)
+#if defined(__linux__) && defined(PPC_FEATURE2_VEC_CRYPTO)
 	if ((getauxval(AT_HWCAP2) & PPC_FEATURE2_VEC_CRYPTO) != 0)
 		return true;
 #elif defined(_AIX)
@@ -972,7 +982,7 @@ inline bool CPU_QueryPMULL()
 {
 	// Power8 and ISA 2.07 provide in-core crypto. Glibc
 	// 2.24 or higher is required for PPC_FEATURE2_VEC_CRYPTO.
-#if defined(__linux__)
+#if defined(__linux__) && defined(PPC_FEATURE2_VEC_CRYPTO)
 	if ((getauxval(AT_HWCAP2) & PPC_FEATURE2_VEC_CRYPTO) != 0)
 		return true;
 #elif defined(_AIX)
@@ -986,7 +996,7 @@ inline bool CPU_QuerySHA256()
 {
 	// Power8 and ISA 2.07 provide in-core crypto. Glibc
 	// 2.24 or higher is required for PPC_FEATURE2_VEC_CRYPTO.
-#if defined(__linux__)
+#if defined(__linux__) && defined(PPC_FEATURE2_VEC_CRYPTO)
 	if ((getauxval(AT_HWCAP2) & PPC_FEATURE2_VEC_CRYPTO) != 0)
 		return true;
 #elif defined(_AIX)
@@ -999,7 +1009,7 @@ inline bool CPU_QuerySHA512()
 {
 	// Power8 and ISA 2.07 provide in-core crypto. Glibc
 	// 2.24 or higher is required for PPC_FEATURE2_VEC_CRYPTO.
-#if defined(__linux__)
+#if defined(__linux__) && defined(PPC_FEATURE2_VEC_CRYPTO)
 	if ((getauxval(AT_HWCAP2) & PPC_FEATURE2_VEC_CRYPTO) != 0)
 		return true;
 #elif defined(_AIX)
@@ -1013,7 +1023,7 @@ inline bool CPU_QuerySHA512()
 inline bool CPU_QueryDARN()
 {
 	// Power9 and ISA 3.0 provide DARN.
-#if defined(__linux__)
+#if defined(__linux__) && defined(PPC_FEATURE2_ARCH_3_00)
 	if ((getauxval(AT_HWCAP2) & PPC_FEATURE2_ARCH_3_00) != 0)
 		return true;
 #elif defined(_AIX)
@@ -1042,7 +1052,7 @@ void DetectPowerpcFeatures()
 	int cacheLineSize = getsystemcfg(SC_L1C_DLS);
 	if (cacheLineSize > 0)
 		g_cacheLineSize = cacheLineSize;
-#elif defined(__linux__) && defined(_SC_LEVEL1_DCACHE_LINESIZE)
+#elif defined(_SC_LEVEL1_DCACHE_LINESIZE)
 	// Glibc does not implement on some platforms. The runtime returns 0 instead of error.
 	// https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/posix/sysconf.c
 	int cacheLineSize = sysconf(_SC_LEVEL1_DCACHE_LINESIZE);
@@ -1050,6 +1060,9 @@ void DetectPowerpcFeatures()
 		g_cacheLineSize = cacheLineSize;
 #endif
 
+	if (g_cacheLineSize == 0)
+		g_cacheLineSize = CRYPTOPP_L1_CACHE_LINE_SIZE;
+
 	*const_cast<volatile bool*>(&g_PowerpcDetectionDone) = true;
 }
 

cryptdll.vcxproj

@@ -212,6 +212,7 @@
     <ClCompile Include="gcm.cpp" />
     <ClCompile Include="gcm_simd.cpp" />
     <ClCompile Include="gf2n.cpp" />
+    <ClCompile Include="gf2n_simd.cpp" />
     <ClCompile Include="gfpcrypt.cpp" />
     <ClCompile Include="hex.cpp" />
     <ClCompile Include="hmac.cpp" />
@@ -255,6 +256,16 @@
     <ClInclude Include="channels.h" />
     <ClInclude Include="cmac.h" />
     <ClInclude Include="config.h" />
+    <ClInclude Include="config_align.h" />
+    <ClInclude Include="config_asm.h" />
+    <ClInclude Include="config_cpu.h" />
+    <ClInclude Include="config_cxx.h" />
+    <ClInclude Include="config_dll.h" />
+    <ClInclude Include="config_int.h" />
+    <ClInclude Include="config_misc.h" />
+    <ClInclude Include="config_ns.h" />
+    <ClInclude Include="config_os.h" />
+    <ClInclude Include="config_ver.h" />
     <ClInclude Include="cpu.h" />
     <ClInclude Include="cryptlib.h" />
     <ClInclude Include="des.h" />

cryptdll.vcxproj.filters

@@ -98,6 +98,9 @@
     <ClCompile Include="gf2n.cpp">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="gf2n_simd.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
     <ClCompile Include="gfpcrypt.cpp">
       <Filter>Source Files</Filter>
     </ClCompile>
@@ -229,6 +232,36 @@
     <ClInclude Include="config.h">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="config_align.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_asm.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_cpu.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_cxx.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_dll.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_int.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_misc.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_ns.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_os.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_ver.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
     <ClInclude Include="cpu.h">
       <Filter>Header Files</Filter>
     </ClInclude>

cryptest.nmake

@@ -69,21 +69,22 @@ LIB_SRCS = \
     gcm_simd.cpp gf256.cpp gf2_32.cpp gf2n.cpp gf2n_simd.cpp gfpcrypt.cpp \
     gost.cpp gzip.cpp hc128.cpp hc256.cpp hex.cpp hight.cpp hmac.cpp \
     hrtimer.cpp ida.cpp idea.cpp iterhash.cpp kalyna.cpp kalynatab.cpp \
-    keccak.cpp keccakc.cpp lea.cpp lea_simd.cpp luc.cpp mars.cpp marss.cpp \
+    keccak.cpp keccak_core.cpp keccak_simd.cpp lea.cpp lea_simd.cpp luc.cpp \
-    md2.cpp md4.cpp md5.cpp misc.cpp modes.cpp mqueue.cpp mqv.cpp \
+    mars.cpp marss.cpp md2.cpp md4.cpp md5.cpp misc.cpp modes.cpp mqueue.cpp \
-    nbtheory.cpp neon_simd.cpp oaep.cpp osrng.cpp padlkrng.cpp panama.cpp \
+    mqv.cpp nbtheory.cpp neon_simd.cpp oaep.cpp osrng.cpp padlkrng.cpp \
-    pkcspad.cpp poly1305.cpp polynomi.cpp ppc_power7.cpp ppc_power8.cpp \
+    panama.cpp pkcspad.cpp poly1305.cpp polynomi.cpp ppc_power7.cpp \
-    ppc_power9.cpp ppc_simd.cpp pssr.cpp pubkey.cpp queue.cpp rabbit.cpp \
+    ppc_power8.cpp ppc_power9.cpp ppc_simd.cpp pssr.cpp pubkey.cpp queue.cpp \
-    rabin.cpp randpool.cpp rc2.cpp rc5.cpp rc6.cpp rdrand.cpp rdtables.cpp \
+    rabbit.cpp rabin.cpp randpool.cpp rc2.cpp rc5.cpp rc6.cpp rdrand.cpp \
-    rijndael.cpp rijndael_simd.cpp ripemd.cpp rng.cpp rsa.cpp rw.cpp \
+    rdtables.cpp rijndael.cpp rijndael_simd.cpp ripemd.cpp rng.cpp rsa.cpp \
-    safer.cpp salsa.cpp scrypt.cpp seal.cpp seed.cpp serpent.cpp sha.cpp \
+    rw.cpp safer.cpp salsa.cpp scrypt.cpp seal.cpp seed.cpp serpent.cpp \
-    sha3.cpp sha_simd.cpp shacal2.cpp shacal2_simd.cpp shark.cpp sharkbox.cpp \
+    sha.cpp sha3.cpp sha_simd.cpp shacal2.cpp shacal2_simd.cpp shake.cpp \
-    simeck.cpp simeck_simd.cpp simon.cpp simon128_simd.cpp simon64_simd.cpp \
+    shark.cpp sharkbox.cpp simeck.cpp simeck_simd.cpp simon.cpp \
-    skipjack.cpp sm3.cpp sm4.cpp sm4_simd.cpp sosemanuk.cpp speck.cpp \
+    simon128_simd.cpp simon64_simd.cpp skipjack.cpp sm3.cpp sm4.cpp \
-    speck128_simd.cpp speck64_simd.cpp square.cpp squaretb.cpp sse_simd.cpp \
+    sm4_simd.cpp sosemanuk.cpp speck.cpp speck128_simd.cpp speck64_simd.cpp \
-    strciphr.cpp tea.cpp tftables.cpp threefish.cpp tiger.cpp tigertab.cpp \
+    square.cpp squaretb.cpp sse_simd.cpp strciphr.cpp tea.cpp tftables.cpp \
-    ttmac.cpp tweetnacl.cpp twofish.cpp vmac.cpp wake.cpp whrlpool.cpp \
+    threefish.cpp tiger.cpp tigertab.cpp ttmac.cpp tweetnacl.cpp twofish.cpp \
-    xed25519.cpp xtr.cpp xtrcrypt.cpp zdeflate.cpp zinflate.cpp zlib.cpp
+    vmac.cpp wake.cpp whrlpool.cpp xed25519.cpp xtr.cpp xtrcrypt.cpp \
+    zdeflate.cpp zinflate.cpp zlib.cpp
 
 LIB_OBJS = \
     cryptlib.obj cpu.obj integer.obj 3way.obj adler32.obj algebra.obj \
@@ -99,24 +100,25 @@ LIB_OBJS = \
     gcm_simd.obj gf256.obj gf2_32.obj gf2n.obj gf2n_simd.obj gfpcrypt.obj \
     gost.obj gzip.obj hc128.obj hc256.obj hex.obj hight.obj hmac.obj \
     hrtimer.obj ida.obj idea.obj iterhash.obj kalyna.obj kalynatab.obj \
-    keccak.obj keccakc.obj lea.obj lea_simd.obj luc.obj mars.obj marss.obj \
+    keccak.obj keccak_core.obj keccak_simd.obj lea.obj lea_simd.obj luc.obj \
-    md2.obj md4.obj md5.obj misc.obj modes.obj mqueue.obj mqv.obj \
+    mars.obj marss.obj md2.obj md4.obj md5.obj misc.obj modes.obj mqueue.obj \
-    nbtheory.obj neon_simd.obj oaep.obj osrng.obj padlkrng.obj panama.obj \
+    mqv.obj nbtheory.obj neon_simd.obj oaep.obj osrng.obj padlkrng.obj \
-    pkcspad.obj poly1305.obj polynomi.obj ppc_power7.obj ppc_power8.obj \
+    panama.obj pkcspad.obj poly1305.obj polynomi.obj ppc_power7.obj \
-    ppc_power9.obj ppc_simd.obj pssr.obj pubkey.obj queue.obj rabbit.obj \
+    ppc_power8.obj ppc_power9.obj ppc_simd.obj pssr.obj pubkey.obj queue.obj \
-    rabin.obj randpool.obj rc2.obj rc5.obj rc6.obj rdrand.obj rdtables.obj \
+    rabbit.obj rabin.obj randpool.obj rc2.obj rc5.obj rc6.obj rdrand.obj \
-    rijndael.obj rijndael_simd.obj ripemd.obj rng.obj rsa.obj rw.obj \
+    rdtables.obj rijndael.obj rijndael_simd.obj ripemd.obj rng.obj rsa.obj \
-    safer.obj salsa.obj scrypt.obj seal.obj seed.obj serpent.obj sha.obj \
+    rw.obj safer.obj salsa.obj scrypt.obj seal.obj seed.obj serpent.obj \
-    sha3.obj sha_simd.obj shacal2.obj shacal2_simd.obj shark.obj sharkbox.obj \
+    sha.obj sha3.obj sha_simd.obj shacal2.obj shacal2_simd.obj shake.obj \
-    simeck.obj simeck_simd.obj simon.obj simon128_simd.obj simon64_simd.obj \
+    shark.obj sharkbox.obj simeck.obj simeck_simd.obj simon.obj \
-    skipjack.obj sm3.obj sm4.obj sm4_simd.obj sosemanuk.obj speck.obj \
+    simon128_simd.obj simon64_simd.obj skipjack.obj sm3.obj sm4.obj \
-    speck128_simd.obj speck64_simd.obj square.obj squaretb.obj sse_simd.obj \
+    sm4_simd.obj sosemanuk.obj speck.obj speck128_simd.obj speck64_simd.obj \
-    strciphr.obj tea.obj tftables.obj threefish.obj tiger.obj tigertab.obj \
+    square.obj squaretb.obj sse_simd.obj strciphr.obj tea.obj tftables.obj \
-    ttmac.obj tweetnacl.obj twofish.obj vmac.obj wake.obj whrlpool.obj \
+    threefish.obj tiger.obj tigertab.obj ttmac.obj tweetnacl.obj twofish.obj \
-    xed25519.obj xtr.obj xtrcrypt.obj zdeflate.obj zinflate.obj zlib.obj
+    vmac.obj wake.obj whrlpool.obj xed25519.obj xtr.obj xtrcrypt.obj \
+    zdeflate.obj zinflate.obj zlib.obj
 
 ASM_OBJS = \
-    rdrand-x86.obj rdrand-x64.obj x64masm.obj x64dll.obj
+    rdrand-x86.obj rdrand-x64.obj rdseed-x86.obj rdseed-x64.obj x64masm.obj x64dll.obj
 
 TEST_SRCS = \
     test.cpp bench1.cpp bench2.cpp bench3.cpp datatest.cpp \
@@ -151,12 +153,14 @@ LDFLAGS = /nologo /SUBSYSTEM:CONSOLE /DEBUG
 ARFLAGS = /nologo
 LDLIBS =
 
-# Debug build.
+# Compiler debug build.
 # CXXFLAGS = $(CXXFLAGS) /DDEBUG /D_DEBUG /Oi /Oy- /Od /MTd
-# Release build. Add /OPT:REF to linker
+# Compiler release build.
 CXXFLAGS = $(CXXFLAGS) /DNDEBUG /D_NDEBUG /Oi /Oy /O2 /MT
-# Linker flags.
+# Linker debug build.
-LDFLAGS = $(LDFLAGS) /OPT:REF
+# LDFLAGS = $(LDFLAGS) /DEBUG
+# Linker release build.
+LDFLAGS = $(LDFLAGS) /DEBUG /OPT:REF
 
 # Attempt to detect when <sdkddkver.h> and <winapifamily.h> are available
 #   http://stackoverflow.com/q/40577415 ?
@@ -192,10 +196,12 @@ PLATFORM = x64
 # CXXFLAGS = $(CXXFLAGS) /DWINAPI_FAMILY=WINAPI_FAMILY_APP
 AS = ml.exe
 ASFLAGS = /nologo /D_M_X86 /W3 /Cx /Zi /safeseh
-LIB_SRCS = $(LIB_SRCS) rdrand.cpp rdrand.asm
+LIB_SRCS = $(LIB_SRCS) rdrand.cpp rdrand.asm rdseed.asm
-LIB_OBJS = $(LIB_OBJS) rdrand-x86.obj
+LIB_OBJS = $(LIB_OBJS) rdrand-x86.obj rdseed-x86.obj
 LDFLAGS = $(LDFLAGS) /MACHINE:X86
 LDLIBS = $(LDLIBS) kernel32.lib
+RDRAND_OBJ = rdrand-x86.obj
+RDSEED_OBJ = rdseed-x86.obj
 !ENDIF
 
 # May need $(VCINSTALLDIR)\bin\amd64\ml64.exe
@@ -205,10 +211,12 @@ LDLIBS = $(LDLIBS) kernel32.lib
 # CXXFLAGS = $(CXXFLAGS) /DWINAPI_FAMILY=WINAPI_FAMILY_APP
 AS = ml64.exe
 ASFLAGS = /nologo /D_M_X64 /W3 /Cx /Zi
-LIB_SRCS = $(LIB_SRCS) rdrand.cpp rdrand.asm
+LIB_SRCS = $(LIB_SRCS) rdrand.cpp rdrand.asm rdseed.asm
-LIB_OBJS = $(LIB_OBJS) rdrand-x64.obj x64masm.obj x64dll.obj
+LIB_OBJS = $(LIB_OBJS) rdrand-x64.obj rdseed-x64.obj x64masm.obj x64dll.obj
 LDFLAGS = $(LDFLAGS) /MACHINE:X64
 LDLIBS = $(LDLIBS) kernel32.lib
+RDRAND_OBJ = rdrand-x64.obj
+RDSEED_OBJ = rdseed-x64.obj
 !ENDIF
 
 !IF "$(PLATFORM)" == "ARM" || "$(PLATFORM)" == "arm" || "$(PLATFORM)" == "ARM64" || "$(PLATFORM)" == "arm64"
@@ -256,13 +264,13 @@ clean ::
 	$(RM) /F /Q pch.obj $(LIB_OBJS) $(ASM_OBJS) $(TEST_OBJS) *.pdb 2>nul
 
 distclean :: clean
-!IF EXIST ($(USERNAME).sdf)
+!IF EXIST ("$(USERNAME).sdf")
-	attrib -r -a -s -h $(USERNAME).sdf 2>nul
+	attrib -r -a -s -h "$(USERNAME).sdf" 2>nul
-	$(RM) /F /Q $(USERNAME).sdf 2>nul
+	$(RM) /F /Q "$(USERNAME).sdf" 2>nul
 !ENDIF
-!IF EXIST ($(USERNAME).suo)
+!IF EXIST ("$(USERNAME).suo")
-	attrib -r -a -s -h $(USERNAME).suo 2>nul
+	attrib -r -a -s -h $(USERNAME).suo" 2>nul
-	$(RM) /F /Q $(USERNAME).suo 2>nul
+	$(RM) /F /Q $(USERNAME).suo" 2>nul
 !ENDIF
 !IF EXIST (Win32\)
 	$(RMDIR) Win32\ /q /s 2>nul
@@ -279,21 +287,25 @@ pch.pch: pch.h pch.cpp
 	$(CXX) $(CXXFLAGS) /Yc"pch.h" /Fp"pch.pch" /c pch.cpp
 
 # No precompiled headers
-iterhash.obj:
+iterhash.obj: iterhash.h iterhash.cpp
 	$(CXX) $(CXXFLAGS) /Y- /c iterhash.cpp
-dll.obj:
+dll.obj: dll.h dll.cpp
 	$(CXX) $(CXXFLAGS) /Y- /c dll.cpp
-rdrand.obj:
+rdrand.obj: rdrand.h rdrand.cpp
 	$(CXX) $(CXXFLAGS) /c rdrand.cpp
 
 # Built for x86/x64
-rdrand-x86.obj:
+rdrand-x86.obj: rdrand.asm
 	$(AS) $(ASFLAGS) /Fo rdrand-x86.obj /c rdrand.asm
-rdrand-x64.obj:
+rdrand-x64.obj: rdrand.asm
 	$(AS) $(ASFLAGS) /Fo rdrand-x64.obj /c rdrand.asm
-x64masm.obj:
+rdseed-x86.obj: rdseed.asm
+	$(AS) $(ASFLAGS) /Fo rdseed-x86.obj /c rdseed.asm
+rdseed-x64.obj: rdseed.asm
+	$(AS) $(ASFLAGS) /Fo rdseed-x64.obj /c rdseed.asm
+x64masm.obj: x64masm.asm
 	$(AS) $(ASFLAGS) /Fo x64masm.obj /c x64masm.asm
-x64dll.obj:
+x64dll.obj: x64dll.asm
 	$(AS) $(ASFLAGS) /Fo x64dll.obj /c x64dll.asm
 
 # You may need to delete this on early versions of Visual Studio.

cryptest.vcxproj

@@ -195,7 +195,10 @@
   <!-- The rule copies cryptest.exe to the project   -->
   <!-- root directory so it can be executed in place -->
   <Target Name="CopyCryptestToRoot">
-    <Exec Command="copy $(Platform)\Output\$(Configuration)\cryptest.exe $(SolutionDir)" />
+    <Exec Command="copy &quot;$(Platform)\Output\$(Configuration)\cryptest.exe&quot; &quot;$(SolutionDir)&quot;" />
+  </Target>
+  <Target Name="CopyToRoot">
+    <Exec Command="copy &quot;$(Platform)\Output\$(Configuration)\cryptest.exe&quot; &quot;$(SolutionDir)&quot;" />
   </Target>
   <!-- Source Files -->
   <ItemGroup>

cryptlib.cpp

@@ -16,7 +16,6 @@
 #ifndef CRYPTOPP_IMPORTS
 
 #include "cryptlib.h"
-#include "misc.h"
 #include "filters.h"
 #include "algparam.h"
 #include "fips140.h"
@@ -26,6 +25,7 @@
 #include "secblock.h"
 #include "smartptr.h"
 #include "stdcpp.h"
+#include "misc.h"
 
 NAMESPACE_BEGIN(CryptoPP)
 
@@ -333,20 +333,20 @@ void RandomNumberGenerator::GenerateIntoBufferedTransformation(BufferedTransform
 	}
 }
 
-size_t KeyDerivationFunction::MinDerivedLength() const
+size_t KeyDerivationFunction::MinDerivedKeyLength() const
 {
 	return 0;
 }
 
-size_t KeyDerivationFunction::MaxDerivedLength() const
+size_t KeyDerivationFunction::MaxDerivedKeyLength() const
 {
 	return static_cast<size_t>(-1);
 }
 
-void KeyDerivationFunction::ThrowIfInvalidDerivedLength(size_t length) const
+void KeyDerivationFunction::ThrowIfInvalidDerivedKeyLength(size_t length) const
 {
 	if (!IsValidDerivedLength(length))
-		throw InvalidDerivedLength(GetAlgorithm().AlgorithmName(), length);
+		throw InvalidDerivedKeyLength(GetAlgorithm().AlgorithmName(), length);
 }
 
 void KeyDerivationFunction::SetParameters(const NameValuePairs& params) {
@@ -405,8 +405,9 @@ RandomNumberGenerator & NullRNG()
 
 bool HashTransformation::TruncatedVerify(const byte *digest, size_t digestLength)
 {
+	// Allocate at least 1 for calculated to avoid triggering diagnostics
 	ThrowIfInvalidTruncatedSize(digestLength);
-	SecByteBlock calculated(digestLength);
+	SecByteBlock calculated(digestLength ? digestLength : 1);
 	TruncatedFinal(calculated, digestLength);
 	return VerifyBufsEqual(calculated, digest, digestLength);
 }
@@ -742,6 +743,12 @@ size_t BufferedTransformation::ChannelPutWord32(const std::string &channel, word
 	return ChannelPut(channel, m_buf, 4, blocking);
 }
 
+size_t BufferedTransformation::ChannelPutWord64(const std::string &channel, word64 value, ByteOrder order, bool blocking)
+{
+	PutWord(false, order, m_buf, value);
+	return ChannelPut(channel, m_buf, 8, blocking);
+}
+
 size_t BufferedTransformation::PutWord16(word16 value, ByteOrder order, bool blocking)
 {
 	return ChannelPutWord16(DEFAULT_CHANNEL, value, order, blocking);
@@ -752,22 +759,20 @@ size_t BufferedTransformation::PutWord32(word32 value, ByteOrder order, bool blo
 	return ChannelPutWord32(DEFAULT_CHANNEL, value, order, blocking);
 }
 
-// Issue 340
+size_t BufferedTransformation::PutWord64(word64 value, ByteOrder order, bool blocking)
-#if CRYPTOPP_GCC_DIAGNOSTIC_AVAILABLE
+{
-# pragma GCC diagnostic push
+	return ChannelPutWord64(DEFAULT_CHANNEL, value, order, blocking);
-# pragma GCC diagnostic ignored "-Wconversion"
+}
-# pragma GCC diagnostic ignored "-Wsign-conversion"
-#endif
 
 size_t BufferedTransformation::PeekWord16(word16 &value, ByteOrder order) const
 {
 	byte buf[2] = {0, 0};
 	size_t len = Peek(buf, 2);
 
-	if (order)
+	if (order == BIG_ENDIAN_ORDER)
-		value = (buf[0] << 8) | buf[1];
+		value = ((word16)buf[0] << 8) | (word16)buf[1];
 	else
-		value = (buf[1] << 8) | buf[0];
+		value = ((word16)buf[1] << 8) | (word16)buf[0];
 
 	return len;
 }
@@ -777,18 +782,32 @@ size_t BufferedTransformation::PeekWord32(word32 &value, ByteOrder order) const
 	byte buf[4] = {0, 0, 0, 0};
 	size_t len = Peek(buf, 4);
 
-	if (order)
+	if (order == BIG_ENDIAN_ORDER)
-		value = (buf[0] << 24) | (buf[1] << 16) | (buf[2] << 8) | buf [3];
+		value = ((word32)buf[0] << 24) | ((word32)buf[1] << 16) |
+		        ((word32)buf[2] << 8)  |  (word32)buf[3];
 	else
-		value = (buf[3] << 24) | (buf[2] << 16) | (buf[1] << 8) | buf [0];
+		value = ((word32)buf[3] << 24) | ((word32)buf[2] << 16) |
+		        ((word32)buf[1] << 8)  |  (word32)buf[0];
 
 	return len;
 }
 
-// Issue 340
+size_t BufferedTransformation::PeekWord64(word64 &value, ByteOrder order) const
-#if CRYPTOPP_GCC_DIAGNOSTIC_AVAILABLE
+{
-# pragma GCC diagnostic pop
+	byte buf[8] = {0, 0, 0, 0, 0, 0, 0, 0};
-#endif
+	size_t len = Peek(buf, 8);
+
+	if (order == BIG_ENDIAN_ORDER)
+		value = ((word64)buf[0] << 56) | ((word64)buf[1] << 48) | ((word64)buf[2] << 40) |
+		        ((word64)buf[3] << 32) | ((word64)buf[4] << 24) | ((word64)buf[5] << 16) |
+		        ((word64)buf[6] << 8)  |  (word64)buf[7];
+	else
+		value = ((word64)buf[7] << 56) | ((word64)buf[6] << 48) | ((word64)buf[5] << 40) |
+		        ((word64)buf[4] << 32) | ((word64)buf[3] << 24) | ((word64)buf[2] << 16) |
+		        ((word64)buf[1] << 8)  |  (word64)buf[0];
+
+	return len;
+}
 
 size_t BufferedTransformation::GetWord16(word16 &value, ByteOrder order)
 {
@@ -800,6 +819,11 @@ size_t BufferedTransformation::GetWord32(word32 &value, ByteOrder order)
 	return (size_t)Skip(PeekWord32(value, order));
 }
 
+size_t BufferedTransformation::GetWord64(word64 &value, ByteOrder order)
+{
+	return (size_t)Skip(PeekWord64(value, order));
+}
+
 void BufferedTransformation::Attach(BufferedTransformation *newAttachment)
 {
 	if (AttachedTransformation() && AttachedTransformation()->Attachable())

cryptlib.h

@@ -3,7 +3,7 @@
 /// \file cryptlib.h
 /// \brief Abstract base classes that provide a uniform interface to this library.
 
-/*!	\mainpage Crypto++ Library 8.1 API Reference
+/*!	\mainpage Crypto++ Library 8.3 API Reference
 <dl>
 <dt>Abstract Base Classes<dd>
 	cryptlib.h
@@ -202,6 +202,9 @@ private:
 class CRYPTOPP_DLL InvalidArgument : public Exception
 {
 public:
+	/// \brief Construct an InvalidArgument
+	/// \param s the message for the exception
+	/// \details The member function <tt>what()</tt> returns <tt>s</tt>.
 	explicit InvalidArgument(const std::string &s) : Exception(INVALID_ARGUMENT, s) {}
 };
 
@@ -209,6 +212,9 @@ public:
 class CRYPTOPP_DLL InvalidDataFormat : public Exception
 {
 public:
+	/// \brief Construct an InvalidDataFormat
+	/// \param s the message for the exception
+	/// \details The member function <tt>what()</tt> returns <tt>s</tt>.
 	explicit InvalidDataFormat(const std::string &s) : Exception(INVALID_DATA_FORMAT, s) {}
 };
 
@@ -216,6 +222,9 @@ public:
 class CRYPTOPP_DLL InvalidCiphertext : public InvalidDataFormat
 {
 public:
+	/// \brief Construct an InvalidCiphertext
+	/// \param s the message for the exception
+	/// \details The member function <tt>what()</tt> returns <tt>s</tt>.
 	explicit InvalidCiphertext(const std::string &s) : InvalidDataFormat(s) {}
 };
 
@@ -223,6 +232,9 @@ public:
 class CRYPTOPP_DLL NotImplemented : public Exception
 {
 public:
+	/// \brief Construct an NotImplemented
+	/// \param s the message for the exception
+	/// \details The member function <tt>what()</tt> returns <tt>s</tt>.
 	explicit NotImplemented(const std::string &s) : Exception(NOT_IMPLEMENTED, s) {}
 };
 
@@ -230,6 +242,9 @@ public:
 class CRYPTOPP_DLL CannotFlush : public Exception
 {
 public:
+	/// \brief Construct an CannotFlush
+	/// \param s the message for the exception
+	/// \details The member function <tt>what()</tt> returns <tt>s</tt>.
 	explicit CannotFlush(const std::string &s) : Exception(CANNOT_FLUSH, s) {}
 };
 
@@ -238,6 +253,13 @@ class CRYPTOPP_DLL OS_Error : public Exception
 {
 public:
 	virtual ~OS_Error() throw() {}
+
+	/// \brief Construct an OS_Error
+	/// \param errorType the error type
+	/// \param s the message for the exception
+	/// \param operation the operation for the exception
+	/// \param errorCode the error code
+	/// \details The member function <tt>what()</tt> returns <tt>s</tt>.
 	OS_Error(ErrorType errorType, const std::string &s, const std::string& operation, int errorCode)
 		: Exception(errorType, s), m_operation(operation), m_errorCode(errorCode) {}
 
@@ -255,7 +277,8 @@ protected:
 struct CRYPTOPP_DLL DecodingResult
 {
 	/// \brief Constructs a DecodingResult
-	/// \details isValidCoding is initialized to false and messageLength is initialized to 0.
+	/// \details isValidCoding is initialized to false and messageLength is
+	///  initialized to 0.
 	explicit DecodingResult() : isValidCoding(false), messageLength(0) {}
 	/// \brief Constructs a DecodingResult
 	/// \param len the message length
@@ -264,11 +287,13 @@ struct CRYPTOPP_DLL DecodingResult
 
 	/// \brief Compare two DecodingResult
 	/// \param rhs the other DecodingResult
-	/// \return true if both isValidCoding and messageLength are equal, false otherwise
+	/// \return true if either isValidCoding or messageLength is \a not equal,
+	///  false otherwise
 	bool operator==(const DecodingResult &rhs) const {return isValidCoding == rhs.isValidCoding && messageLength == rhs.messageLength;}
 	/// \brief Compare two DecodingResult
 	/// \param rhs the other DecodingResult
-	/// \return true if either isValidCoding or messageLength is \a not equal, false otherwise
+	/// \return true if either isValidCoding or messageLength is \a not equal,
+	///  false otherwise
 	/// \details Returns <tt>!operator==(rhs)</tt>.
 	bool operator!=(const DecodingResult &rhs) const {return !operator==(rhs);}
 
@@ -279,24 +304,28 @@ struct CRYPTOPP_DLL DecodingResult
 };
 
 /// \brief Interface for retrieving values given their names
-/// \details This class is used to safely pass a variable number of arbitrarily typed arguments to functions
+/// \details This class is used to safely pass a variable number of arbitrarily
-///   and to read values from keys and crypto parameters.
+///  typed arguments to functions and to read values from keys and crypto parameters.
-/// \details To obtain an object that implements NameValuePairs for the purpose of parameter
+/// \details To obtain an object that implements NameValuePairs for the purpose of
-///   passing, use the MakeParameters() function.
+///  parameter passing, use the MakeParameters() function.
-/// \details To get a value from NameValuePairs, you need to know the name and the type of the value.
+/// \details To get a value from NameValuePairs, you need to know the name and the
-///   Call GetValueNames() on a NameValuePairs object to obtain a list of value names that it supports.
+///  type of the value. Call GetValueNames() on a NameValuePairs object to obtain a
-///   then look at the Name namespace documentation to see what the type of each value is, or
+///  list of value names that it supports. then look at the Name namespace
-///   alternatively, call GetIntValue() with the value name, and if the type is not int, a
+///  documentation to see what the type of each value is, or alternatively, call
-///   ValueTypeMismatch exception will be thrown and you can get the actual type from the exception object.
+///  GetIntValue() with the value name, and if the type is not int, a
+///  ValueTypeMismatch exception will be thrown and you can get the actual type from
+///  the exception object.
 /// \sa NullNameValuePairs, g_nullNameValuePairs,
-///   <A HREF="http://www.cryptopp.com/wiki/NameValuePairs">NameValuePairs</A> on the Crypto++ wiki
+///  <A HREF="http://www.cryptopp.com/wiki/NameValuePairs">NameValuePairs</A> on the
+///  Crypto++ wiki
 class NameValuePairs
 {
 public:
 	virtual ~NameValuePairs() {}
 
 	/// \brief Thrown when an unexpected type is encountered
-	/// \details Exception thrown when trying to retrieve a value using a different type than expected
+	/// \details Exception thrown when trying to retrieve a value using a different
+	///  type than expected
 	class CRYPTOPP_DLL ValueTypeMismatch : public InvalidArgument
 	{
 	public:
@@ -1477,11 +1506,11 @@ public:
 
 	/// \brief Determine minimum number of bytes
 	/// \returns Minimum number of bytes which can be derived
-	virtual size_t MinDerivedLength() const;
+	virtual size_t MinDerivedKeyLength() const;
 
 	/// \brief Determine maximum number of bytes
 	/// \returns Maximum number of bytes which can be derived
-	virtual size_t MaxDerivedLength() const;
+	virtual size_t MaxDerivedKeyLength() const;
 
 	/// \brief Returns a valid key length for the derivation function
 	/// \param keylength the size of the derived key, in bytes
@@ -1503,7 +1532,7 @@ public:
 	/// \param secretLen the size of the secret buffer, in bytes
 	/// \param params additional initialization parameters to configure this object
 	/// \returns the number of iterations performed
-	/// \throws InvalidDerivedLength if <tt>derivedLen</tt> is invalid for the scheme
+	/// \throws InvalidDerivedKeyLength if <tt>derivedLen</tt> is invalid for the scheme
 	/// \details DeriveKey() provides a standard interface to derive a key from
 	///   a secret seed and other parameters. Each class that derives from KeyDerivationFunction
 	///   provides an overload that accepts most parameters used by the derivation function.
@@ -1525,7 +1554,7 @@ protected:
 	/// \brief Validates the derived key length
 	/// \param length the size of the derived key material, in bytes
 	/// \throws InvalidKeyLength if the key length is invalid
-	void ThrowIfInvalidDerivedLength(size_t length) const;
+	void ThrowIfInvalidDerivedKeyLength(size_t length) const;
 };
 
 /// \brief Interface for password based key derivation functions
@@ -1614,8 +1643,8 @@ public:
 		/// \brief Input a byte for processing
 		/// \param inByte the 8-bit byte (octet) to be processed.
 		/// \param blocking specifies whether the object should block when processing input.
-		/// \return the number of bytes that remain in the block (i.e., bytes not processed). 0 indicates all
+		/// \return the number of bytes that remain in the block (i.e., bytes not processed).
-		///   bytes were processed.
+		///   0 indicates all bytes were processed.
 		/// \details <tt>Put(byte)</tt> calls <tt>Put(byte*, size_t)</tt>.
 		size_t Put(byte inByte, bool blocking=true)
 			{return Put(&inByte, 1, blocking);}
@@ -1624,8 +1653,8 @@ public:
 		/// \param inString the byte buffer to process
 		/// \param length the size of the string, in bytes
 		/// \param blocking specifies whether the object should block when processing input
-		/// \return the number of bytes that remain in the block (i.e., bytes not processed). 0 indicates all
+		/// \return the number of bytes that remain in the block (i.e., bytes not processed).
-		///   bytes were processed.
+		///   0 indicates all bytes were processed.
 		/// \details Internally, Put() calls Put2().
 		size_t Put(const byte *inString, size_t length, bool blocking=true)
 			{return Put2(inString, length, 0, blocking);}
@@ -1634,18 +1663,26 @@ public:
 		/// \param value the 16-bit value to be processed
 		/// \param order the ByteOrder of the value to be processed.
 		/// \param blocking specifies whether the object should block when processing input
-		/// \return the number of bytes that remain in the block (i.e., bytes not processed). 0 indicates all
+		/// \return the number of bytes that remain in the block (i.e., bytes not processed).
-		///   bytes were processed.
+		///   0 indicates all bytes were processed.
 		size_t PutWord16(word16 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);
 
 		/// Input a 32-bit word for processing.
 		/// \param value the 32-bit value to be processed.
 		/// \param order the ByteOrder of the value to be processed.
 		/// \param blocking specifies whether the object should block when processing input.
-		/// \return the number of bytes that remain in the block (i.e., bytes not processed). 0 indicates all
+		/// \return the number of bytes that remain in the block (i.e., bytes not processed).
-		///   bytes were processed.
+		///   0 indicates all bytes were processed.
 		size_t PutWord32(word32 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);
 
+		/// Input a 64-bit word for processing.
+		/// \param value the 64-bit value to be processed.
+		/// \param order the ByteOrder of the value to be processed.
+		/// \param blocking specifies whether the object should block when processing input.
+		/// \return the number of bytes that remain in the block (i.e., bytes not processed).
+		///   0 indicates all bytes were processed.
+		size_t PutWord64(word64 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);
+
 		/// \brief Request space which can be written into by the caller
 		/// \param size the requested size of the buffer
 		/// \return byte pointer to the space to input data
@@ -1781,18 +1818,23 @@ public:
 
 		/// \brief Flush buffered input and/or output, with signal propagation
 		/// \param hardFlush is used to indicate whether all data should be flushed
-		/// \param propagation the number of attached transformations the Flush() signal should be passed
+		/// \param propagation the number of attached transformations the Flush()
-		/// \param blocking specifies whether the object should block when processing input
+		///  signal should be passed
-		/// \details propagation count includes this object. Setting propagation to <tt>1</tt> means this
+		/// \param blocking specifies whether the object should block when processing
-		///   object only. Setting propagation to <tt>-1</tt> means unlimited propagation.
+		///  input
-		/// \note Hard flushes must be used with care. It means try to process and output everything, even if
+		/// \details propagation count includes this object. Setting propagation to
-		///   there may not be enough data to complete the action. For example, hard flushing a HexDecoder
+		///  <tt>1</tt> means this object only. Setting propagation to <tt>-1</tt>
-		///   would cause an error if you do it after inputing an odd number of hex encoded characters.
+		///  means unlimited propagation.
-		/// \note For some types of filters, like  ZlibDecompressor, hard flushes can only
+		/// \note Hard flushes must be used with care. It means try to process and
-		///   be done at "synchronization points". These synchronization points are positions in the data
+		///  output everything, even if there may not be enough data to complete the
-		///   stream that are created by hard flushes on the corresponding reverse filters, in this
+		///  action. For example, hard flushing a HexDecoder would cause an error if
-		///   example ZlibCompressor. This is useful when zlib compressed data is moved across a
+		///  you do it after inputing an odd number of hex encoded characters.
-		///   network in packets and compression state is preserved across packets, as in the SSH2 protocol.
+		/// \note For some types of filters, like  ZlibDecompressor, hard flushes can
+		///  only be done at "synchronization points". These synchronization points
+		///  are positions in the data stream that are created by hard flushes on the
+		///  corresponding reverse filters, in this example ZlibCompressor. This is
+		///  useful when zlib compressed data is moved across a network in packets
+		///  and compression state is preserved across packets, as in the SSH2 protocol.
 		virtual bool Flush(bool hardFlush, int propagation=-1, bool blocking=true);
 
 		/// \brief Marks the end of a series of messages, with signal propagation
@@ -1856,7 +1898,7 @@ public:
 		/// \param peekMax the number of bytes to Peek
 		/// \return the number of bytes read during the call.
 		/// \details Peek does not remove bytes from the object. Use the return value of
-		///     Get() to detect short reads.
+		///     Peek() to detect short reads.
 		virtual size_t Peek(byte *outString, size_t peekMax) const;
 
 		/// \brief Retrieve a 16-bit word
@@ -1870,15 +1912,23 @@ public:
 		/// \param value the 32-bit value to be retrieved
 		/// \param order the ByteOrder of the value to be processed.
 		/// \return the number of bytes consumed during the call.
-		/// \details Use the return value of GetWord16() to detect short reads.
+		/// \details Use the return value of GetWord32() to detect short reads.
 		size_t GetWord32(word32 &value, ByteOrder order=BIG_ENDIAN_ORDER);
 
+		/// \brief Retrieve a 64-bit word
+		/// \param value the 64-bit value to be retrieved
+		/// \param order the ByteOrder of the value to be processed.
+		/// \return the number of bytes consumed during the call.
+		/// \details Use the return value of GetWord64() to detect short reads.
+		/// \since Crypto++ 8.3
+		size_t GetWord64(word64 &value, ByteOrder order=BIG_ENDIAN_ORDER);
+
 		/// \brief Peek a 16-bit word
 		/// \param value the 16-bit value to be retrieved
 		/// \param order the ByteOrder of the value to be processed.
 		/// \return the number of bytes consumed during the call.
 		/// \details Peek does not consume bytes in the stream. Use the return value
-		///    of GetWord16() to detect short reads.
+		///    of PeekWord16() to detect short reads.
 		size_t PeekWord16(word16 &value, ByteOrder order=BIG_ENDIAN_ORDER) const;
 
 		/// \brief Peek a 32-bit word
@@ -1886,9 +1936,18 @@ public:
 		/// \param order the ByteOrder of the value to be processed.
 		/// \return the number of bytes consumed during the call.
 		/// \details Peek does not consume bytes in the stream. Use the return value
-		///    of GetWord16() to detect short reads.
+		///    of PeekWord32() to detect short reads.
 		size_t PeekWord32(word32 &value, ByteOrder order=BIG_ENDIAN_ORDER) const;
 
+		/// \brief Peek a 64-bit word
+		/// \param value the 64-bit value to be retrieved
+		/// \param order the ByteOrder of the value to be processed.
+		/// \return the number of bytes consumed during the call.
+		/// \details Peek does not consume bytes in the stream. Use the return value
+		///    of PeekWord64() to detect short reads.
+		/// \since Crypto++ 8.3
+		size_t PeekWord64(word64 &value, ByteOrder order=BIG_ENDIAN_ORDER) const;
+
 		/// move transferMax bytes of the buffered output to target as input
 
 		/// \brief Transfer bytes from this object to another BufferedTransformation
@@ -2035,7 +2094,7 @@ public:
 		/// \param channel the channel on which the transfer should occur
 		/// \param blocking specifies whether the object should block when processing input
 		/// \return the number of bytes that remain in the transfer block (i.e., bytes not transferred)
-		/// \details TransferTo() removes bytes from this object and moves them to the destination.
+		/// \details TransferTo2() removes bytes from this object and moves them to the destination.
 		///   Transfer begins at the index position in the current stream, and not from an absolute
 		///   position in the stream.
 		/// \details byteCount is an \a IN and \a OUT parameter. When the call is made,
@@ -2144,6 +2203,15 @@ public:
 		///   number of bytes that were not processed.
 		size_t ChannelPutWord32(const std::string &channel, word32 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);
 
+		/// \brief Input a 64-bit word for processing on a channel.
+		/// \param channel the channel to process the data.
+		/// \param value the 64-bit value to be processed.
+		/// \param order the ByteOrder of the value to be processed.
+		/// \param blocking specifies whether the object should block when processing input.
+		/// \return 0 indicates all bytes were processed during the call. Non-0 indicates the
+		///   number of bytes that were not processed.
+		size_t ChannelPutWord64(const std::string &channel, word64 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);
+
 		/// \brief Signal the end of a message
 		/// \param channel the channel to process the data.
 		/// \param propagation the number of attached transformations the ChannelMessageEnd() signal should be passed
@@ -2272,7 +2340,10 @@ protected:
 		{return propagation != 0 ? propagation - 1 : 0;}
 
 private:
-	byte m_buf[4];	// for ChannelPutWord16 and ChannelPutWord32, to ensure buffer isn't deallocated before non-blocking operation completes
+	// for ChannelPutWord16, ChannelPutWord32 and ChannelPutWord64,
+	// to ensure the buffer isn't deallocated before non-blocking
+	// operation completes
+	byte m_buf[8];
 };
 
 /// \brief An input discarding BufferedTransformation

cryptlib.vcxproj

@@ -373,6 +373,16 @@
       <Command>ml64.exe /c /nologo /D_M_X64 /W3 /Cx /Zi /Fo"$(IntDir)rdrand-x64.obj" "%(FullPath)"</Command>
       <Outputs>$(IntDir)\rdrand-x64.obj;%(Outputs)</Outputs>
     </CustomBuild>
+    <CustomBuild Condition="'$(Platform)'=='Win32'" Include="rdseed.asm">
+      <Message>Building and assembling rdseed.asm</Message>
+      <Command>ml.exe /c /nologo /D_M_X86 /W3 /Cx /Zi /safeseh /Fo"$(IntDir)rdseed-x86.obj" "%(FullPath)"</Command>
+      <Outputs>$(IntDir)\rdseed-x86.obj;%(Outputs)</Outputs>
+    </CustomBuild>
+    <CustomBuild Condition="'$(Platform)'=='x64'" Include="rdseed.asm">
+      <Message>Building and assembling rdseed.asm</Message>
+      <Command>ml64.exe /c /nologo /D_M_X64 /W3 /Cx /Zi /Fo"$(IntDir)rdseed-x64.obj" "%(FullPath)"</Command>
+      <Outputs>$(IntDir)\rdseed-x64.obj;%(Outputs)</Outputs>
+    </CustomBuild>
     <CustomBuild Condition="'$(Platform)'=='x64' AND ('$(Configuration)'=='Debug' Or '$(Configuration)'=='Release')" Include="x64dll.asm">
       <Message>Building and assembling x64dll.asm</Message>
       <Command>ml64.exe /c /nologo /D_M_X64 /W3 /Zi /Fo"$(IntDir)x64dll.obj" "%(FullPath)"</Command>
@@ -412,6 +422,16 @@
     <ClInclude Include="channels.h" />
     <ClInclude Include="cmac.h" />
     <ClInclude Include="config.h" />
+    <ClInclude Include="config_align.h" />
+    <ClInclude Include="config_asm.h" />
+    <ClInclude Include="config_cpu.h" />
+    <ClInclude Include="config_cxx.h" />
+    <ClInclude Include="config_dll.h" />
+    <ClInclude Include="config_int.h" />
+    <ClInclude Include="config_misc.h" />
+    <ClInclude Include="config_ns.h" />
+    <ClInclude Include="config_os.h" />
+    <ClInclude Include="config_ver.h" />
     <ClInclude Include="cpu.h" />
     <ClInclude Include="crc.h" />
     <ClInclude Include="cryptlib.h" />

cryptlib.vcxproj.filters

@@ -615,6 +615,36 @@
     <ClInclude Include="config.h">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="config_align.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_asm.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_cpu.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_cxx.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_dll.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_int.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_misc.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_ns.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_os.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="config_ver.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
     <ClInclude Include="cpu.h">
       <Filter>Header Files</Filter>
     </ClInclude>
@@ -1034,6 +1064,9 @@
     <CustomBuild Include="rdrand.asm">
       <Filter>Source Files</Filter>
     </CustomBuild>
+    <CustomBuild Include="rdseed.asm">
+      <Filter>Source Files</Filter>
+    </CustomBuild>
     <CustomBuild Include="x64dll.asm">
       <Filter>Source Files</Filter>
     </CustomBuild>

cryptopp.rc

@@ -27,8 +27,8 @@ LANGUAGE LANG_ENGLISH, SUBLANG_ENGLISH_US
 //
 
 VS_VERSION_INFO VERSIONINFO
- FILEVERSION 8,1,0,0
+ FILEVERSION 8,3,0,0
- PRODUCTVERSION 8,1,0,0
+ PRODUCTVERSION 8,3,0,0
  FILEFLAGSMASK 0x3fL
 #ifdef _DEBUG
  FILEFLAGS 0x1L
@@ -46,13 +46,13 @@ BEGIN
             VALUE "Comments", "Free crypto library, more information available at www.cryptopp.com"
             VALUE "CompanyName", "Wei Dai"
             VALUE "FileDescription", "Crypto++® Library DLL"
-            VALUE "FileVersion", "8, 1, 0, 0"
+            VALUE "FileVersion", "8, 3, 0, 0"
             VALUE "InternalName", "cryptopp"
-            VALUE "LegalCopyright", "Copyright© 1995-2018 by Wei Dai"
+            VALUE "LegalCopyright", "Copyright© 1995-2019 by Wei Dai"
             VALUE "LegalTrademarks", "Crypto++®"
             VALUE "OriginalFilename", "cryptopp.dll"
             VALUE "ProductName", "Crypto++® Library"
-            VALUE "ProductVersion", "8, 1, 0, 0"
+            VALUE "ProductVersion", "8, 3, 0, 0"
         END
     END
     BLOCK "VarFileInfo"

darn.h

@@ -2,11 +2,12 @@
 //          DARN requires POWER9/ISA 3.0.
 
 // At the moment only GCC 7.0 (and above) seems to support __builtin_darn()
-// and __builtin_darn_32(). Clang 7.0 does not provide them, but it does
+// and __builtin_darn_32(). However, GCC generates incorrect code. Clang 7.0
-// support assembly instructions. XLC is unknown, but there are no hits when
+// does not provide them, but it does support assembly instructions. XLC is
-// searching IBM's site. To cover more platforms we provide GCC inline
+// unknown, but there are no hits when searching IBM's site. To cover more
-// assembly like we do with RDRAND and RDSEED. Platforms that don't support
+// platforms we provide GCC inline assembly like we do with RDRAND and RDSEED.
-// GCC inline assembly or the builtin will fail the compile. Also see
+// Platforms that don't support GCC inline assembly or the builtin will fail
+// to compile. Also see https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91481 and
 // https://gcc.gnu.org/onlinedocs/gcc/Basic-PowerPC-Built-in-Functions-Available-on-ISA-3_002e0.html
 
 /// \file darn.h

datatest.cpp

@@ -8,7 +8,6 @@
 #include "factory.h"
 #include "integer.h"
 #include "filters.h"
-#include "hex.h"
 #include "randpool.h"
 #include "files.h"
 #include "trunhash.h"
@@ -16,6 +15,8 @@
 #include "smartptr.h"
 #include "validate.h"
 #include "stdcpp.h"
+#include "misc.h"
+#include "hex.h"
 #include "trap.h"
 
 #include <iostream>
@@ -208,7 +209,7 @@ void PutDecodedDatumInto(const TestData &data, const char *name, BufferedTransfo
 
 		while (repeat--)
 		{
-			q.Put(reinterpret_cast<const byte*>(&s2[0]), s2.size());
+			q.Put(ConstBytePtr(s2), BytePtrSize(s2));
 			RandomizedTransfer(q, target, false);
 		}
 	}
@@ -265,10 +266,11 @@ public:
 			*reinterpret_cast<int *>(pValue) = atoi(value.c_str());
 		else if (valueType == typeid(word64))
 		{
-			std::string x(value); errno = 0;
+			std::string x(value.empty() ? "0" : value);
 			const char* beg = &x[0];
 			char* end = &x[0] + value.size();
 
+			errno = 0;
 			*reinterpret_cast<word64*>(pValue) = STRTOUL64(beg, &end, 0);
 			if (errno != 0)
 				return false;
@@ -279,7 +281,7 @@ public:
 		{
 			m_temp.clear();
 			PutDecodedDatumInto(m_data, name, StringSink(m_temp).Ref());
-			reinterpret_cast<ConstByteArrayParameter *>(pValue)->Assign(reinterpret_cast<const byte *>(&m_temp[0]), m_temp.size(), false);
+			reinterpret_cast<ConstByteArrayParameter *>(pValue)->Assign(ConstBytePtr(m_temp), BytePtrSize(m_temp), false);
 		}
 		else
 			throw ValueTypeMismatch(name, typeid(std::string), valueType);
@@ -496,8 +498,8 @@ void TestSymmetricCipher(TestData &v, const NameValuePairs &overrideParameters)
 		}
 		else
 		{
-			encryptor->SetKey(reinterpret_cast<const byte*>(&key[0]), key.size(), pairs);
+			encryptor->SetKey(ConstBytePtr(key), BytePtrSize(key), pairs);
-			decryptor->SetKey(reinterpret_cast<const byte*>(&key[0]), key.size(), pairs);
+			decryptor->SetKey(ConstBytePtr(key), BytePtrSize(key), pairs);
 		}
 
 		word64 seek64 = pairs.GetWord64ValueWithDefault("Seek64", 0);
@@ -547,13 +549,13 @@ void TestSymmetricCipher(TestData &v, const NameValuePairs &overrideParameters)
 				encrypted.reserve(10000 * plaintext.size());
 				for (int j=0; j<10000; j++)
 				{
-					cipher->ProcessString(reinterpret_cast<byte*>(&buf[0]), buf.size());
+					cipher->ProcessString(BytePtr(buf), BytePtrSize(buf));
 					encrypted.append(buf.begin(), buf.end());
 				}
 
 				encrypted.erase(0, encrypted.size() - keybuf.size());
-				xorbuf(reinterpret_cast<byte*>(&keybuf[0]), reinterpret_cast<const byte*>(&encrypted[0]), keybuf.size());
+				xorbuf(BytePtr(keybuf), BytePtr(encrypted), BytePtrSize(keybuf));
-				cipher->SetKey(reinterpret_cast<const byte*>(&keybuf[0]), keybuf.size());
+				cipher->SetKey(BytePtr(keybuf), BytePtrSize(keybuf));
 			}
 
 			encrypted.assign(buf.begin(), buf.end());
@@ -641,8 +643,8 @@ void TestAuthenticatedSymmetricCipher(TestData &v, const NameValuePairs &overrid
 		member_ptr<AuthenticatedSymmetricCipher> encryptor, decryptor;
 		encryptor.reset(ObjectFactoryRegistry<AuthenticatedSymmetricCipher, ENCRYPTION>::Registry().CreateObject(name.c_str()));
 		decryptor.reset(ObjectFactoryRegistry<AuthenticatedSymmetricCipher, DECRYPTION>::Registry().CreateObject(name.c_str()));
-		encryptor->SetKey(reinterpret_cast<const byte*>(&key[0]), key.size(), pairs);
+		encryptor->SetKey(ConstBytePtr(key), BytePtrSize(key), pairs);
-		decryptor->SetKey(reinterpret_cast<const byte*>(&key[0]), key.size(), pairs);
+		decryptor->SetKey(ConstBytePtr(key), BytePtrSize(key), pairs);
 
 		// Code coverage
 		(void)encryptor->AlgorithmName();
@@ -736,7 +738,7 @@ void TestDigestOrMAC(TestData &v, bool testDigest)
 		mac.reset(ObjectFactoryRegistry<MessageAuthenticationCode>::Registry().CreateObject(name.c_str()));
 		pHash = mac.get();
 		std::string key = GetDecodedDatum(v, "Key");
-		mac->SetKey(reinterpret_cast<const byte *>(&key[0]), key.size(), pairs);
+		mac->SetKey(ConstBytePtr(key), BytePtrSize(key), pairs);
 
 		// Code coverage
 		(void)mac->AlgorithmName();
@@ -779,8 +781,7 @@ void TestKeyDerivationFunction(TestData &v)
 	kdf.reset(ObjectFactoryRegistry<KeyDerivationFunction>::Registry().CreateObject(name.c_str()));
 
 	std::string calculated; calculated.resize(expected.size());
-	kdf->DeriveKey(reinterpret_cast<byte*>(&calculated[0]), calculated.size(),
+	kdf->DeriveKey(BytePtr(calculated), BytePtrSize(calculated), BytePtr(secret), BytePtrSize(secret), pairs);
-		reinterpret_cast<const byte*>(&secret[0]), secret.size(), pairs);
 
 	if(calculated != expected)
 	{

donna_32.cpp

@@ -37,6 +37,18 @@
 // Squash MS LNK4221 and libtool warnings
 extern const char DONNA32_FNAME[] = __FILE__;
 
+ANONYMOUS_NAMESPACE_BEGIN
+
+// Can't use GetAlignmentOf<word32>() because of C++11 and constexpr
+// Can use 'const unsigned int' because of MSVC 2013
+#if (CRYPTOPP_BOOL_X86 || CRYPTOPP_BOOL_X32 || CRYPTOPP_BOOL_X64)
+# define ALIGN_SPEC 16
+#else
+# define ALIGN_SPEC 4
+#endif
+
+ANONYMOUS_NAMESPACE_END
+
 #if defined(CRYPTOPP_CURVE25519_32BIT)
 
 #include "donna_32.h"
@@ -431,7 +443,7 @@ curve25519_swap_conditional(bignum25519 x, bignum25519 qpx, word32 iswap) {
  */
 void
 curve25519_pow_two5mtwo0_two250mtwo0(bignum25519 b) {
-    ALIGN(16) bignum25519 t0,c;
+    ALIGN(ALIGN_SPEC) bignum25519 t0,c;
 
     /* 2^5  - 2^0 */ /* b */
     /* 2^10 - 2^5 */ curve25519_square_times(t0, b, 5);
@@ -455,7 +467,7 @@ curve25519_pow_two5mtwo0_two250mtwo0(bignum25519 b) {
  */
 void
 curve25519_recip(bignum25519 out, const bignum25519 z) {
-    ALIGN(16) bignum25519 a, t0, b;
+    ALIGN(ALIGN_SPEC) bignum25519 a, t0, b;
 
     /* 2 */ curve25519_square(a, z); /* a = 2 */
     /* 8 */ curve25519_square_times(t0, a, 2);
@@ -913,7 +925,33 @@ curve25519_contract(byte out[32], const bignum25519 in) {
 
 /* out = (flag) ? in : out */
 inline void
-curve25519_move_conditional_bytes(byte out[96], const byte in[96], word32 flag) {
+curve25519_move_conditional_bytes(byte out[96], const byte in[96], word32 flag)
+{
+    // TODO: enable this code path once we can test and benchmark it.
+    // It is about 48 insns shorter, it avoids punning which may be UB,
+    // and it is guaranteed constant time.
+#if defined(__GNUC__) && defined(__i686__) && 0
+    const word32 iter = 96/sizeof(word32);
+    word32* outl = reinterpret_cast<word32*>(out);
+    const word32* inl = reinterpret_cast<const word32*>(in);
+    word32 idx=0, val;
+
+    __asm__ __volatile__ (
+        ".att_syntax                         ;\n"
+        "cmpl     $0, %[flag]                ;\n"  // compare, set ZERO flag
+        "movl     %[iter], %%ecx             ;\n"  // load iteration count
+        "1:                                  ;\n"
+        "  movl     (%[idx],%[out]), %[val]  ;\n"  // val = out[idx]
+        "  cmovnzl  (%[idx],%[in]), %[val]   ;\n"  // copy in[idx] to val if NZ
+        "  movl     %[val], (%[idx],%[out])  ;\n"  // out[idx] = val
+        "  leal     4(%[idx]), %[idx]        ;\n"  // increment index
+        "  loopnz   1b                       ;\n"  // does not affect flags
+        : [out] "+S" (outl), [in] "+D" (inl),
+          [idx] "+b" (idx), [val] "=r" (val)
+        : [flag] "g" (flag), [iter] "I" (iter)
+        : "ecx", "memory", "cc"
+    );
+#else
     const word32 nb = flag - 1, b = ~nb;
     const word32 *inl = (const word32 *)in;
     word32 *outl = (word32 *)out;
@@ -941,6 +979,7 @@ curve25519_move_conditional_bytes(byte out[96], const byte in[96], word32 flag)
     outl[21] = (outl[21] & nb) | (inl[21] & b);
     outl[22] = (outl[22] & nb) | (inl[22] & b);
     outl[23] = (outl[23] & nb) | (inl[23] & b);
+#endif
 }
 
 /* if (iswap) swap(a, b) */
@@ -967,7 +1006,7 @@ curve25519_swap_conditional(bignum25519 a, bignum25519 b, word32 iswap) {
  */
 void
 curve25519_pow_two5mtwo0_two250mtwo0(bignum25519 b) {
-    ALIGN(16) bignum25519 t0,c;
+    ALIGN(ALIGN_SPEC) bignum25519 t0,c;
 
     /* 2^5  - 2^0 */ /* b */
     /* 2^10 - 2^5 */ curve25519_square_times(t0, b, 5);
@@ -991,7 +1030,7 @@ curve25519_pow_two5mtwo0_two250mtwo0(bignum25519 b) {
  */
 void
 curve25519_recip(bignum25519 out, const bignum25519 z) {
-    ALIGN(16) bignum25519 a,t0,b;
+    ALIGN(ALIGN_SPEC) bignum25519 a,t0,b;
 
     /* 2 */ curve25519_square_times(a, z, 1); /* a = 2 */
     /* 8 */ curve25519_square_times(t0, a, 2);
@@ -1009,7 +1048,7 @@ curve25519_recip(bignum25519 out, const bignum25519 z) {
  */
 void
 curve25519_pow_two252m3(bignum25519 two252m3, const bignum25519 z) {
-    ALIGN(16) bignum25519 b,c,t0;
+    ALIGN(ALIGN_SPEC) bignum25519 b,c,t0;
 
     /* 2 */ curve25519_square_times(c, z, 1); /* c = 2 */
     /* 8 */ curve25519_square_times(t0, c, 2); /* t0 = 8 */
@@ -1865,7 +1904,7 @@ ed25519_publickey_CXX(byte publicKey[32], const byte secretKey[32])
     using namespace CryptoPP::Donna::Ed25519;
 
     bignum256modm a;
-    ALIGN(16) ge25519 A;
+    ALIGN(ALIGN_SPEC) ge25519 A;
     hash_512bits extsk;
 
     /* A = aB */
@@ -1889,7 +1928,7 @@ ed25519_sign_CXX(std::istream& stream, const byte sk[32], const byte pk[32], byt
     using namespace CryptoPP::Donna::Ed25519;
 
     bignum256modm r, S, a;
-    ALIGN(16) ge25519 R;
+    ALIGN(ALIGN_SPEC) ge25519 R;
     hash_512bits extsk, hashr, hram;
 
     // Unfortunately we need to read the stream twice. The fisrt time calculates
@@ -1938,7 +1977,7 @@ ed25519_sign_CXX(const byte *m, size_t mlen, const byte sk[32], const byte pk[32
     using namespace CryptoPP::Donna::Ed25519;
 
     bignum256modm r, S, a;
-    ALIGN(16) ge25519 R;
+    ALIGN(ALIGN_SPEC) ge25519 R;
     hash_512bits extsk, hashr, hram;
 
     ed25519_extsk(extsk, sk);
@@ -1990,7 +2029,7 @@ ed25519_sign_open_CXX(std::istream& stream, const byte pk[32], const byte RS[64]
 
     using namespace CryptoPP::Donna::Ed25519;
 
-    ALIGN(16) ge25519 R, A;
+    ALIGN(ALIGN_SPEC) ge25519 R, A;
     hash_512bits hash;
     bignum256modm hram, S;
     byte checkR[32];
@@ -2018,7 +2057,7 @@ ed25519_sign_open_CXX(const byte *m, size_t mlen, const byte pk[32], const byte
 
     using namespace CryptoPP::Donna::Ed25519;
 
-    ALIGN(16) ge25519 R, A;
+    ALIGN(ALIGN_SPEC) ge25519 R, A;
     hash_512bits hash;
     bignum256modm hram, S;
     byte checkR[32];

donna_64.cpp

@@ -37,6 +37,18 @@
 // Squash MS LNK4221 and libtool warnings
 extern const char DONNA64_FNAME[] = __FILE__;
 
+ANONYMOUS_NAMESPACE_BEGIN
+
+// Can't use GetAlignmentOf<word64>() because of C++11 and constexpr
+// Can use 'const unsigned int' because of MSVC 2013
+#if (CRYPTOPP_BOOL_X86 || CRYPTOPP_BOOL_X32 || CRYPTOPP_BOOL_X64)
+# define ALIGN_SPEC 16
+#else
+# define ALIGN_SPEC 8
+#endif
+
+ANONYMOUS_NAMESPACE_END
+
 #if defined(CRYPTOPP_CURVE25519_64BIT)
 
 #include "donna_64.h"
@@ -359,7 +371,7 @@ curve25519_swap_conditional(bignum25519 x, bignum25519 qpx, word64 iswap) {
  */
 void
 curve25519_pow_two5mtwo0_two250mtwo0(bignum25519 b) {
-    ALIGN(16) bignum25519 t0,c;
+    ALIGN(ALIGN_SPEC) bignum25519 t0,c;
 
     /* 2^5  - 2^0 */ /* b */
     /* 2^10 - 2^5 */ curve25519_square_times(t0, b, 5);
@@ -383,7 +395,7 @@ curve25519_pow_two5mtwo0_two250mtwo0(bignum25519 b) {
  */
 void
 curve25519_recip(bignum25519 out, const bignum25519 z) {
-    ALIGN(16) bignum25519 a, t0, b;
+    ALIGN(ALIGN_SPEC) bignum25519 a, t0, b;
 
     /* 2 */ curve25519_square(a, z); /* a = 2 */
     /* 8 */ curve25519_square_times(t0, a, 2);
@@ -735,10 +747,36 @@ curve25519_contract(byte *out, const bignum25519 input) {
 
 /* out = (flag) ? in : out */
 inline void
-curve25519_move_conditional_bytes(byte out[96], const byte in[96], word64 flag) {
+curve25519_move_conditional_bytes(byte out[96], const byte in[96], word64 flag)
+{
+    // TODO: enable this code path once we can test and benchmark it.
+    // It is about 24 insns shorter, it avoids punning which may be UB,
+    // and it is guaranteed constant time.
+#if defined(__GNUC__) && defined(__x86_64__) && 0
+    const word32 iter = 96/sizeof(word64);
+    word64* outq = reinterpret_cast<word64*>(out);
+    const word64* inq = reinterpret_cast<const word64*>(in);
+    word64 idx=0, val;
+
+    __asm__ __volatile__ (
+        ".att_syntax                         ;\n"
+        "cmpq     $0, %[flag]                ;\n"  // compare, set ZERO flag
+        "movq     %[iter], %%rcx             ;\n"  // load iteration count
+        "1:                                  ;\n"
+        "  movq     (%[idx],%[out]), %[val]  ;\n"  // val = out[idx]
+        "  cmovnzq  (%[idx],%[in]), %[val]   ;\n"  // copy in[idx] to val if NZ
+        "  movq     %[val], (%[idx],%[out])  ;\n"  // out[idx] = val
+        "  leaq     8(%[idx]), %[idx]        ;\n"  // increment index
+        "  loopnz   1b                       ;\n"  // does not affect flags
+        : [out] "+S" (outq), [in] "+D" (inq),
+          [idx] "+b" (idx), [val] "=r" (val)
+        : [flag] "g" (flag), [iter] "I" (iter)
+        : "rcx", "memory", "cc"
+    );
+#else
     const word64 nb = flag - 1, b = ~nb;
-    const word64 *inq = (const word64 *)in;
+    const word64 *inq = (const word64 *)(const void*)in;
-    word64 *outq = (word64 *)out;
+    word64 *outq = (word64 *)(void *)out;
     outq[0] = (outq[0] & nb) | (inq[0] & b);
     outq[1] = (outq[1] & nb) | (inq[1] & b);
     outq[2] = (outq[2] & nb) | (inq[2] & b);
@@ -751,6 +789,7 @@ curve25519_move_conditional_bytes(byte out[96], const byte in[96], word64 flag)
     outq[9] = (outq[9] & nb) | (inq[9] & b);
     outq[10] = (outq[10] & nb) | (inq[10] & b);
     outq[11] = (outq[11] & nb) | (inq[11] & b);
+#endif
 }
 
 /* if (iswap) swap(a, b) */
@@ -1065,7 +1104,7 @@ contract256_slidingwindow_modm(signed char r[256], const bignum256modm s, int wi
  */
 void
 curve25519_pow_two5mtwo0_two250mtwo0(bignum25519 b) {
-    ALIGN(16) bignum25519 t0,c;
+    ALIGN(ALIGN_SPEC) bignum25519 t0,c;
 
     /* 2^5  - 2^0 */ /* b */
     /* 2^10 - 2^5 */ curve25519_square_times(t0, b, 5);
@@ -1089,7 +1128,7 @@ curve25519_pow_two5mtwo0_two250mtwo0(bignum25519 b) {
  */
 void
 curve25519_recip(bignum25519 out, const bignum25519 z) {
-    ALIGN(16) bignum25519 a,t0,b;
+    ALIGN(ALIGN_SPEC) bignum25519 a,t0,b;
 
     /* 2 */ curve25519_square_times(a, z, 1); /* a = 2 */
     /* 8 */ curve25519_square_times(t0, a, 2);
@@ -1107,7 +1146,7 @@ curve25519_recip(bignum25519 out, const bignum25519 z) {
  */
 void
 curve25519_pow_two252m3(bignum25519 two252m3, const bignum25519 z) {
-    ALIGN(16) bignum25519 b,c,t0;
+    ALIGN(ALIGN_SPEC) bignum25519 b,c,t0;
 
     /* 2 */ curve25519_square_times(c, z, 1); /* c = 2 */
     /* 8 */ curve25519_square_times(t0, c, 2); /* t0 = 8 */
@@ -1580,7 +1619,7 @@ ed25519_publickey_CXX(byte publicKey[32], const byte secretKey[32])
     using namespace CryptoPP::Donna::Ed25519;
 
     bignum256modm a;
-    ALIGN(16) ge25519 A;
+    ALIGN(ALIGN_SPEC) ge25519 A;
     hash_512bits extsk;
 
     /* A = aB */
@@ -1604,7 +1643,7 @@ ed25519_sign_CXX(std::istream& stream, const byte sk[32], const byte pk[32], byt
     using namespace CryptoPP::Donna::Ed25519;
 
     bignum256modm r, S, a;
-    ALIGN(16) ge25519 R;
+    ALIGN(ALIGN_SPEC) ge25519 R;
     hash_512bits extsk, hashr, hram;
 
     // Unfortunately we need to read the stream twice. The fisrt time calculates
@@ -1652,7 +1691,7 @@ ed25519_sign_CXX(const byte *m, size_t mlen, const byte sk[32], const byte pk[32
     using namespace CryptoPP::Donna::Ed25519;
 
     bignum256modm r, S, a;
-    ALIGN(16) ge25519 R;
+    ALIGN(ALIGN_SPEC) ge25519 R;
     hash_512bits extsk, hashr, hram;
 
     ed25519_extsk(extsk, sk);
@@ -1703,7 +1742,7 @@ ed25519_sign_open_CXX(const byte *m, size_t mlen, const byte pk[32], const byte
 
     using namespace CryptoPP::Donna::Ed25519;
 
-    ALIGN(16) ge25519 R, A;
+    ALIGN(ALIGN_SPEC) ge25519 R, A;
     hash_512bits hash;
     bignum256modm hram, S;
     byte checkR[32];
@@ -1731,7 +1770,7 @@ ed25519_sign_open_CXX(std::istream& stream, const byte pk[32], const byte RS[64]
 
     using namespace CryptoPP::Donna::Ed25519;
 
-    ALIGN(16) ge25519 R, A;
+    ALIGN(ALIGN_SPEC) ge25519 R, A;
     hash_512bits hash;
     bignum256modm hram, S;
     byte checkR[32];

donna_sse.cpp

@@ -1059,13 +1059,13 @@ int curve25519_mult_SSE2(byte sharedKey[32], const byte secretKey[32], const byt
     packed32bignum25519 qx, qz, pqz, pqx;
     packed64bignum25519 nq, sq, sqscalar, prime, primex, primez, nqpq;
     bignum25519mulprecomp preq;
-    size_t i=0, bit=0, lastbit=0;
+    size_t bit=0, lastbit=0;
 
     curve25519_expand(nqpqx, othersKey);
     curve25519_mul_precompute(&preq, nqpqx);
 
     /* do bits 254..3 */
-    for (i = 254, lastbit=0; i >= 3; i--) {
+    for (size_t i = 254, lastbit=0; i >= 3; i--) {
         bit = (e[i/8] >> (i & 7)) & 1;
         curve25519_swap_conditional(nqx, nqpqx, (word32)(bit ^ lastbit));
         curve25519_swap_conditional(nqz, nqpqz, (word32)(bit ^ lastbit));
@@ -1098,7 +1098,7 @@ int curve25519_mult_SSE2(byte sharedKey[32], const byte secretKey[32], const byt
     curve25519_swap_conditional(nqz, nqpqz, (word32)bit);
 
     /* do bits 2..0 */
-    for (i = 0; i < 3; i++) {
+    for (size_t i = 0; i < 3; i++) {
         curve25519_compute_nq(nq, nqx, nqz);
         curve25519_square_packed64(sq, nq); /* sq = nq^2 */
         curve25519_121665_packed64(sqscalar, sq); /* sqscalar = sq * [121666,121665] */

drbg.h

@@ -207,6 +207,9 @@ public:
         size_t nonceLength=0, const byte* personalization=NULLPTR, size_t personalizationLength=0)
         : NIST_DRBG(), m_c(SEEDLENGTH), m_v(SEEDLENGTH), m_reseed(0)
     {
+        std::memset(m_c, 0x00, m_c.size());
+        std::memset(m_v, 0x00, m_v.size());
+
         if (entropy != NULLPTR && entropyLength != 0)
             DRBG_Instantiate(entropy, entropyLength, nonce, nonceLength, personalization, personalizationLength);
     }
@@ -324,6 +327,9 @@ public:
         size_t nonceLength=0, const byte* personalization=NULLPTR, size_t personalizationLength=0)
         : NIST_DRBG(), m_k(HASH::DIGESTSIZE), m_v(HASH::DIGESTSIZE), m_reseed(0)
     {
+        std::memset(m_k, 0x00, m_k.size());
+        std::memset(m_v, 0x00, m_v.size());
+
         if (entropy != NULLPTR && entropyLength != 0)
             DRBG_Instantiate(entropy, entropyLength, nonce, nonceLength, personalization, personalizationLength);
     }

ec2n.cpp

@@ -16,7 +16,8 @@ ANONYMOUS_NAMESPACE_BEGIN
 using CryptoPP::EC2N;
 
 #if defined(HAVE_GCC_INIT_PRIORITY)
-  const EC2N::Point g_identity __attribute__ ((init_priority (CRYPTOPP_INIT_PRIORITY + 50))) = EC2N::Point();
+  #define INIT_ATTRIBUTE __attribute__ ((init_priority (CRYPTOPP_INIT_PRIORITY + 51)))
+  const EC2N::Point g_identity INIT_ATTRIBUTE = EC2N::Point();
 #elif defined(HAVE_MSC_INIT_PRIORITY)
   #pragma warning(disable: 4075)
   #pragma init_seg(".CRT$XCU")
@@ -260,7 +261,7 @@ const EC2N::Point& EC2N::Double(const Point &P) const
 
 // ********************************************************
 
-/*
+#if 0
 EcPrecomputation<EC2N>& EcPrecomputation<EC2N>::operator=(const EcPrecomputation<EC2N> &rhs)
 {
 	m_ec = rhs.m_ec;
@@ -312,7 +313,7 @@ EC2N::Point EcPrecomputation<EC2N>::CascadeExponentiate(const Integer &exponent,
 {
 	return m_ep.CascadeExponentiate(exponent, static_cast<const EcPrecomputation<EC2N> &>(pc2).m_ep, exponent2);
 }
-*/
+#endif
 
 NAMESPACE_END
 

ec2n.h

@@ -3,7 +3,6 @@
 /// \file ec2n.h
 /// \brief Classes for Elliptic Curves over binary fields
 
-
 #ifndef CRYPTOPP_EC2N_H
 #define CRYPTOPP_EC2N_H
 

eccrypto.cpp

@@ -28,6 +28,9 @@
 #include "ec2n.h"
 #include "misc.h"
 
+#include <iostream>
+#include <sstream>
+
 // Squash MS LNK4221 and libtool warnings
 #ifndef CRYPTOPP_MANUALLY_INSTANTIATE_TEMPLATES
 extern const char ECCRYPTO_FNAME[] = __FILE__;
@@ -683,6 +686,13 @@ OID DL_GroupParameters_EC<EC>::GetAlgorithmID() const
 	return ASN1::id_ecPublicKey();
 }
 
+std::ostream& operator<<(std::ostream& os, const DL_GroupParameters_EC<ECP>::Element& obj)
+{
+	std::ostringstream oss;
+	oss << "(" << std::hex << obj.x << ", " << std::hex  << obj.y << ")";
+	return os << oss.str();
+}
+
 // ******************************************************************
 
 template <class EC>

eccrypto.h

@@ -22,6 +22,8 @@
 #include "ecp.h"
 #include "ec2n.h"
 
+#include <iosfwd>
+
 #if CRYPTOPP_MSC_VERSION
 # pragma warning(push)
 # pragma warning(disable: 4231 4275)
@@ -168,6 +170,8 @@ protected:
 	mutable bool m_compress, m_encodeAsOID;		// presentation details
 };
 
+inline std::ostream& operator<<(std::ostream& os, const DL_GroupParameters_EC<ECP>::Element& obj);
+
 /// \brief Elliptic Curve Discrete Log (DL) public key
 /// \tparam EC elliptic curve field
 template <class EC>

ecp.cpp

@@ -15,10 +15,12 @@
 ANONYMOUS_NAMESPACE_BEGIN
 
 using CryptoPP::ECP;
+using CryptoPP::Integer;
 using CryptoPP::ModularArithmetic;
 
 #if defined(HAVE_GCC_INIT_PRIORITY)
-  const ECP::Point g_identity __attribute__ ((init_priority (CRYPTOPP_INIT_PRIORITY + 51))) = ECP::Point();
+  #define INIT_ATTRIBUTE __attribute__ ((init_priority (CRYPTOPP_INIT_PRIORITY + 50)))
+  const ECP::Point g_identity INIT_ATTRIBUTE = ECP::Point();
 #elif defined(HAVE_MSC_INIT_PRIORITY)
   #pragma warning(disable: 4075)
   #pragma init_seg(".CRT$XCU")
@@ -39,6 +41,503 @@ inline ECP::Point FromMontgomery(const ModularArithmetic &mr, const ECP::Point &
 	return P.identity ? P : ECP::Point(mr.ConvertOut(P.x), mr.ConvertOut(P.y));
 }
 
+inline Integer IdentityToInteger(bool val)
+{
+	return val ? Integer::One() : Integer::Zero();
+}
+
+struct ProjectivePoint
+{
+	ProjectivePoint() {}
+	ProjectivePoint(const Integer &x, const Integer &y, const Integer &z)
+		: x(x), y(y), z(z)	{}
+
+	Integer x, y, z;
+};
+
+/// \brief Addition and Double functions
+/// \sa <A HREF="https://eprint.iacr.org/2015/1060.pdf">Complete
+///  addition formulas for prime order elliptic curves</A>
+struct AdditionFunction
+{
+	explicit AdditionFunction(const ECP::Field& field,
+		const ECP::FieldElement &a, const ECP::FieldElement &b, ECP::Point &r);
+
+	// Double(P)
+	ECP::Point operator()(const ECP::Point& P) const;
+	// Add(P, Q)
+	ECP::Point operator()(const ECP::Point& P, const ECP::Point& Q) const;
+
+protected:
+	/// \brief Parameters and representation for Addition
+	/// \details Addition and Doubling will use different algorithms,
+	///  depending on the <tt>A</tt> coefficient and the representation
+	///  (Affine or Montgomery with precomputation).
+	enum Alpha {
+		/// \brief Coefficient A is 0
+		A_0 = 1,
+		/// \brief Coefficient A is -3
+		A_3 = 2,
+		/// \brief Coefficient A is arbitrary
+		A_Star = 4,
+		/// \brief Representation is Montgomery
+		A_Montgomery = 8
+	};
+
+	const ECP::Field& field;
+	const ECP::FieldElement &a, &b;
+	ECP::Point &R;
+
+	Alpha m_alpha;
+};
+
+#define X p.x
+#define Y p.y
+#define Z p.z
+
+#define X1 p.x
+#define Y1 p.y
+#define Z1 p.z
+
+#define X2 q.x
+#define Y2 q.y
+#define Z2 q.z
+
+#define X3 r.x
+#define Y3 r.y
+#define Z3 r.z
+
+AdditionFunction::AdditionFunction(const ECP::Field& field,
+	const ECP::FieldElement &a, const ECP::FieldElement &b, ECP::Point &r)
+	: field(field), a(a), b(b), R(r), m_alpha(static_cast<Alpha>(0))
+{
+	if (field.IsMontgomeryRepresentation())
+	{
+		m_alpha = A_Montgomery;
+	}
+	else
+	{
+		if (a == 0)
+		{
+			m_alpha = A_0;
+		}
+		else if (a == -3 || (a - field.GetModulus()) == -3)
+		{
+			m_alpha = A_3;
+		}
+		else
+		{
+			m_alpha = A_Star;
+		}
+	}
+}
+
+ECP::Point AdditionFunction::operator()(const ECP::Point& P) const
+{
+	if (m_alpha == A_3)
+	{
+		// Gyrations attempt to maintain constant-timeness
+		// We need either (P.x, P.y, 1) or (0, 1, 0).
+		const Integer x = P.x * IdentityToInteger(!P.identity);
+		const Integer y = P.y * IdentityToInteger(!P.identity) + 1 * IdentityToInteger(P.identity);
+		const Integer z = 1 * IdentityToInteger(!P.identity);
+
+		ProjectivePoint p(x, y, z), r;
+
+		ECP::FieldElement t0 = field.Square(X);
+		ECP::FieldElement t1 = field.Square(Y);
+		ECP::FieldElement t2 = field.Square(Z);
+		ECP::FieldElement t3 = field.Multiply(X, Y);
+		t3 = field.Add(t3, t3);
+		Z3 = field.Multiply(X, Z);
+		Z3 = field.Add(Z3, Z3);
+		Y3 = field.Multiply(b, t2);
+		Y3 = field.Subtract(Y3, Z3);
+		X3 = field.Add(Y3, Y3);
+		Y3 = field.Add(X3, Y3);
+		X3 = field.Subtract(t1, Y3);
+		Y3 = field.Add(t1, Y3);
+		Y3 = field.Multiply(X3, Y3);
+		X3 = field.Multiply(X3, t3);
+		t3 = field.Add(t2, t2);
+		t2 = field.Add(t2, t3);
+		Z3 = field.Multiply(b, Z3);
+		Z3 = field.Subtract(Z3, t2);
+		Z3 = field.Subtract(Z3, t0);
+		t3 = field.Add(Z3, Z3);
+		Z3 = field.Add(Z3, t3);
+		t3 = field.Add(t0, t0);
+		t0 = field.Add(t3, t0);
+		t0 = field.Subtract(t0, t2);
+		t0 = field.Multiply(t0, Z3);
+		Y3 = field.Add(Y3, t0);
+		t0 = field.Multiply(Y, Z);
+		t0 = field.Add(t0, t0);
+		Z3 = field.Multiply(t0, Z3);
+		X3 = field.Subtract(X3, Z3);
+		Z3 = field.Multiply(t0, t1);
+		Z3 = field.Add(Z3, Z3);
+		Z3 = field.Add(Z3, Z3);
+
+		const ECP::FieldElement inv = field.MultiplicativeInverse(Z3.IsZero() ? Integer::One() : Z3);
+		X3 = field.Multiply(X3, inv); Y3 = field.Multiply(Y3, inv);
+
+		// More gyrations
+		R.x = X3*Z3.NotZero();
+		R.y = Y3*Z3.NotZero();
+		R.identity = Z3.IsZero();
+
+		return R;
+	}
+	else if (m_alpha == A_0)
+	{
+		// Gyrations attempt to maintain constant-timeness
+		// We need either (P.x, P.y, 1) or (0, 1, 0).
+		const Integer x = P.x * IdentityToInteger(!P.identity);
+		const Integer y = P.y * IdentityToInteger(!P.identity) + 1 * IdentityToInteger(P.identity);
+		const Integer z = 1 * IdentityToInteger(!P.identity);
+
+		ProjectivePoint p(x, y, z), r;
+		const ECP::FieldElement b3 = field.Multiply(b, 3);
+
+		ECP::FieldElement t0 = field.Square(Y);
+		Z3 = field.Add(t0, t0);
+		Z3 = field.Add(Z3, Z3);
+		Z3 = field.Add(Z3, Z3);
+		ECP::FieldElement t1 = field.Add(Y, Z);
+		ECP::FieldElement t2 = field.Square(Z);
+		t2 = field.Multiply(b3, t2);
+		X3 = field.Multiply(t2, Z3);
+		Y3 = field.Add(t0, t2);
+		Z3 = field.Multiply(t1, Z3);
+		t1 = field.Add(t2, t2);
+		t2 = field.Add(t1, t2);
+		t0 = field.Subtract(t0, t2);
+		Y3 = field.Multiply(t0, Y3);
+		Y3 = field.Add(X3, Y3);
+		t1 = field.Multiply(X, Y);
+		X3 = field.Multiply(t0, t1);
+		X3 = field.Add(X3, X3);
+
+		const ECP::FieldElement inv = field.MultiplicativeInverse(Z3.IsZero() ? Integer::One() : Z3);
+		X3 = field.Multiply(X3, inv); Y3 = field.Multiply(Y3, inv);
+
+		// More gyrations
+		R.x = X3*Z3.NotZero();
+		R.y = Y3*Z3.NotZero();
+		R.identity = Z3.IsZero();
+
+		return R;
+	}
+#if 0
+	// Code path disabled at the moment due to https://github.com/weidai11/cryptopp/issues/878
+	else if (m_alpha == A_Star)
+	{
+		// Gyrations attempt to maintain constant-timeness
+		// We need either (P.x, P.y, 1) or (0, 1, 0).
+		const Integer x = P.x * IdentityToInteger(!P.identity);
+		const Integer y = P.y * IdentityToInteger(!P.identity) + 1 * IdentityToInteger(P.identity);
+		const Integer z = 1 * IdentityToInteger(!P.identity);
+
+		ProjectivePoint p(x, y, z), r;
+		const ECP::FieldElement b3 = field.Multiply(b, 3);
+
+		ECP::FieldElement t0 = field.Square(Y);
+		Z3 = field.Add(t0, t0);
+		Z3 = field.Add(Z3, Z3);
+		Z3 = field.Add(Z3, Z3);
+		ECP::FieldElement t1 = field.Add(Y, Z);
+		ECP::FieldElement t2 = field.Square(Z);
+		t2 = field.Multiply(b3, t2);
+		X3 = field.Multiply(t2, Z3);
+		Y3 = field.Add(t0, t2);
+		Z3 = field.Multiply(t1, Z3);
+		t1 = field.Add(t2, t2);
+		t2 = field.Add(t1, t2);
+		t0 = field.Subtract(t0, t2);
+		Y3 = field.Multiply(t0, Y3);
+		Y3 = field.Add(X3, Y3);
+		t1 = field.Multiply(X, Y);
+		X3 = field.Multiply(t0, t1);
+		X3 = field.Add(X3, X3);
+
+		const ECP::FieldElement inv = field.MultiplicativeInverse(Z3.IsZero() ? Integer::One() : Z3);
+		X3 = field.Multiply(X3, inv); Y3 = field.Multiply(Y3, inv);
+
+		// More gyrations
+		R.x = X3*Z3.NotZero();
+		R.y = Y3*Z3.NotZero();
+		R.identity = Z3.IsZero();
+
+		return R;
+	}
+#endif
+	else  // A_Montgomery
+	{
+		// More gyrations
+		bool identity = !!(P.identity + (P.y == field.Identity()));
+
+		ECP::FieldElement t = field.Square(P.x);
+		t = field.Add(field.Add(field.Double(t), t), a);
+		t = field.Divide(t, field.Double(P.y));
+		ECP::FieldElement x = field.Subtract(field.Subtract(field.Square(t), P.x), P.x);
+		R.y = field.Subtract(field.Multiply(t, field.Subtract(P.x, x)), P.y);
+		R.x.swap(x);
+
+		// More gyrations
+		R.x *= IdentityToInteger(!identity);
+		R.y *= IdentityToInteger(!identity);
+		R.identity = identity;
+
+		return R;
+	}
+}
+
+ECP::Point AdditionFunction::operator()(const ECP::Point& P, const ECP::Point& Q) const
+{
+	if (m_alpha == A_3)
+	{
+		// Gyrations attempt to maintain constant-timeness
+		// We need either (P.x, P.y, 1) or (0, 1, 0).
+		const Integer x1 = P.x * IdentityToInteger(!P.identity);
+		const Integer y1 = P.y * IdentityToInteger(!P.identity) + 1 * IdentityToInteger(P.identity);
+		const Integer z1 = 1 * IdentityToInteger(!P.identity);
+
+		const Integer x2 = Q.x * IdentityToInteger(!Q.identity);
+		const Integer y2 = Q.y * IdentityToInteger(!Q.identity) + 1 * IdentityToInteger(Q.identity);
+		const Integer z2 = 1 * IdentityToInteger(!Q.identity);
+
+		ProjectivePoint p(x1, y1, z1), q(x2, y2, z2), r;
+
+		ECP::FieldElement t0 = field.Multiply(X1, X2);
+		ECP::FieldElement t1 = field.Multiply(Y1, Y2);
+		ECP::FieldElement t2 = field.Multiply(Z1, Z2);
+		ECP::FieldElement t3 = field.Add(X1, Y1);
+		ECP::FieldElement t4 = field.Add(X2, Y2);
+		t3 = field.Multiply(t3, t4);
+		t4 = field.Add(t0, t1);
+		t3 = field.Subtract(t3, t4);
+		t4 = field.Add(Y1, Z1);
+		X3 = field.Add(Y2, Z2);
+		t4 = field.Multiply(t4, X3);
+		X3 = field.Add(t1, t2);
+		t4 = field.Subtract(t4, X3);
+		X3 = field.Add(X1, Z1);
+		Y3 = field.Add(X2, Z2);
+		X3 = field.Multiply(X3, Y3);
+		Y3 = field.Add(t0, t2);
+		Y3 = field.Subtract(X3, Y3);
+		Z3 = field.Multiply(b, t2);
+		X3 = field.Subtract(Y3, Z3);
+		Z3 = field.Add(X3, X3);
+		X3 = field.Add(X3, Z3);
+		Z3 = field.Subtract(t1, X3);
+		X3 = field.Add(t1, X3);
+		Y3 = field.Multiply(b, Y3);
+		t1 = field.Add(t2, t2);
+		t2 = field.Add(t1, t2);
+		Y3 = field.Subtract(Y3, t2);
+		Y3 = field.Subtract(Y3, t0);
+		t1 = field.Add(Y3, Y3);
+		Y3 = field.Add(t1, Y3);
+		t1 = field.Add(t0, t0);
+		t0 = field.Add(t1, t0);
+		t0 = field.Subtract(t0, t2);
+		t1 = field.Multiply(t4, Y3);
+		t2 = field.Multiply(t0, Y3);
+		Y3 = field.Multiply(X3, Z3);
+		Y3 = field.Add(Y3, t2);
+		X3 = field.Multiply(t3, X3);
+		X3 = field.Subtract(X3, t1);
+		Z3 = field.Multiply(t4, Z3);
+		t1 = field.Multiply(t3, t0);
+		Z3 = field.Add(Z3, t1);
+
+		const ECP::FieldElement inv = field.MultiplicativeInverse(Z3.IsZero() ? Integer::One() : Z3);
+		X3 = field.Multiply(X3, inv); Y3 = field.Multiply(Y3, inv);
+
+		// More gyrations
+		R.x = X3*Z3.NotZero();
+		R.y = Y3*Z3.NotZero();
+		R.identity = Z3.IsZero();
+
+		return R;
+	}
+	else if (m_alpha == A_0)
+	{
+		// Gyrations attempt to maintain constant-timeness
+		// We need either (P.x, P.y, 1) or (0, 1, 0).
+		const Integer x1 = P.x * IdentityToInteger(!P.identity);
+		const Integer y1 = P.y * IdentityToInteger(!P.identity) + 1 * IdentityToInteger(P.identity);
+		const Integer z1 = 1 * IdentityToInteger(!P.identity);
+
+		const Integer x2 = Q.x * IdentityToInteger(!Q.identity);
+		const Integer y2 = Q.y * IdentityToInteger(!Q.identity) + 1 * IdentityToInteger(Q.identity);
+		const Integer z2 = 1 * IdentityToInteger(!Q.identity);
+
+		ProjectivePoint p(x1, y1, z1), q(x2, y2, z2), r;
+		const ECP::FieldElement b3 = field.Multiply(b, 3);
+
+		ECP::FieldElement t0 = field.Square(Y);
+		Z3 = field.Add(t0, t0);
+		Z3 = field.Add(Z3, Z3);
+		Z3 = field.Add(Z3, Z3);
+		ECP::FieldElement t1 = field.Add(Y, Z);
+		ECP::FieldElement t2 = field.Square(Z);
+		t2 = field.Multiply(b3, t2);
+		X3 = field.Multiply(t2, Z3);
+		Y3 = field.Add(t0, t2);
+		Z3 = field.Multiply(t1, Z3);
+		t1 = field.Add(t2, t2);
+		t2 = field.Add(t1, t2);
+		t0 = field.Subtract(t0, t2);
+		Y3 = field.Multiply(t0, Y3);
+		Y3 = field.Add(X3, Y3);
+		t1 = field.Multiply(X, Y);
+		X3 = field.Multiply(t0, t1);
+		X3 = field.Add(X3, X3);
+
+		const ECP::FieldElement inv = field.MultiplicativeInverse(Z3.IsZero() ? Integer::One() : Z3);
+		X3 = field.Multiply(X3, inv); Y3 = field.Multiply(Y3, inv);
+
+		// More gyrations
+		R.x = X3*Z3.NotZero();
+		R.y = Y3*Z3.NotZero();
+		R.identity = Z3.IsZero();
+
+		return R;
+	}
+#if 0
+	// Code path disabled at the moment due to https://github.com/weidai11/cryptopp/issues/878
+	else if (m_alpha == A_Star)
+	{
+		// Gyrations attempt to maintain constant-timeness
+		// We need either (P.x, P.y, 1) or (0, 1, 0).
+		const Integer x1 = P.x * IdentityToInteger(!P.identity);
+		const Integer y1 = P.y * IdentityToInteger(!P.identity) + 1 * IdentityToInteger(P.identity);
+		const Integer z1 = 1 * IdentityToInteger(!P.identity);
+
+		const Integer x2 = Q.x * IdentityToInteger(!Q.identity);
+		const Integer y2 = Q.y * IdentityToInteger(!Q.identity) + 1 * IdentityToInteger(Q.identity);
+		const Integer z2 = 1 * IdentityToInteger(!Q.identity);
+
+		ProjectivePoint p(x1, y1, z1), q(x2, y2, z2), r;
+		const ECP::FieldElement b3 = field.Multiply(b, 3);
+
+		ECP::FieldElement t0 = field.Multiply(X1, X2);
+		ECP::FieldElement t1 = field.Multiply(Y1, Y2);
+		ECP::FieldElement t2 = field.Multiply(Z1, Z2);
+		ECP::FieldElement t3 = field.Add(X1, Y1);
+		ECP::FieldElement t4 = field.Add(X2, Y2);
+		t3 = field.Multiply(t3, t4);
+		t4 = field.Add(t0, t1);
+		t3 = field.Subtract(t3, t4);
+		t4 = field.Add(X1, Z1);
+		ECP::FieldElement t5 = field.Add(X2, Z2);
+		t4 = field.Multiply(t4, t5);
+		t5 = field.Add(t0, t2);
+		t4 = field.Subtract(t4, t5);
+		t5 = field.Add(Y1, Z1);
+		X3 = field.Add(Y2, Z2);
+		t5 = field.Multiply(t5, X3);
+		X3 = field.Add(t1, t2);
+		t5 = field.Subtract(t5, X3);
+		Z3 = field.Multiply(a, t4);
+		X3 = field.Multiply(b3, t2);
+		Z3 = field.Add(X3, Z3);
+		X3 = field.Subtract(t1, Z3);
+		Z3 = field.Add(t1, Z3);
+		Y3 = field.Multiply(X3, Z3);
+		t1 = field.Add(t0, t0);
+		t1 = field.Add(t1, t0);
+		t2 = field.Multiply(a, t2);
+		t4 = field.Multiply(b3, t4);
+		t1 = field.Add(t1, t2);
+		t2 = field.Subtract(t0, t2);
+		t2 = field.Multiply(a, t2);
+		t4 = field.Add(t4, t2);
+		t0 = field.Multiply(t1, t4);
+		Y3 = field.Add(Y3, t0);
+		t0 = field.Multiply(t5, t4);
+		X3 = field.Multiply(t3, X3);
+		X3 = field.Subtract(X3, t0);
+		t0 = field.Multiply(t3, t1);
+		Z3 = field.Multiply(t5, Z3);
+		Z3 = field.Add(Z3, t0);
+
+		const ECP::FieldElement inv = field.MultiplicativeInverse(Z3.IsZero() ? Integer::One() : Z3);
+		X3 = field.Multiply(X3, inv); Y3 = field.Multiply(Y3, inv);
+
+		// More gyrations
+		R.x = X3*Z3.NotZero();
+		R.y = Y3*Z3.NotZero();
+		R.identity = Z3.IsZero();
+
+		return R;
+	}
+#endif
+	else  // A_Montgomery
+	{
+		// More gyrations
+		bool return_Q = P.identity;
+		bool return_P = Q.identity;
+		bool double_P = field.Equal(P.x, Q.x) && field.Equal(P.y, Q.y);
+		bool identity = field.Equal(P.x, Q.x) && !field.Equal(P.y, Q.y);
+
+		// This code taken from Double(P) for below
+		identity = !!((double_P * (P.identity + (P.y == field.Identity()))) + identity);
+
+		ECP::Point S = R;
+		if (double_P)
+		{
+			// This code taken from Double(P)
+			ECP::FieldElement t = field.Square(P.x);
+			t = field.Add(field.Add(field.Double(t), t), a);
+			t = field.Divide(t, field.Double(P.y));
+			ECP::FieldElement x = field.Subtract(field.Subtract(field.Square(t), P.x), P.x);
+			R.y = field.Subtract(field.Multiply(t, field.Subtract(P.x, x)), P.y);
+			R.x.swap(x);
+		}
+		else
+		{
+			// Original Add(P,Q) code
+			ECP::FieldElement t = field.Subtract(Q.y, P.y);
+			t = field.Divide(t, field.Subtract(Q.x, P.x));
+			ECP::FieldElement x = field.Subtract(field.Subtract(field.Square(t), P.x), Q.x);
+			R.y = field.Subtract(field.Multiply(t, field.Subtract(P.x, x)), P.y);
+			R.x.swap(x);
+		}
+
+		// More gyrations
+		R.x = R.x * IdentityToInteger(!identity);
+		R.y = R.y * IdentityToInteger(!identity);
+		R.identity = identity;
+
+		if (return_Q)
+			return (R = S), Q;
+		else if (return_P)
+			return (R = S), P;
+		else
+			return (S = R), R;
+	}
+}
+
+#undef X
+#undef Y
+#undef Z
+
+#undef X1
+#undef Y1
+#undef Z1
+
+#undef X2
+#undef Y2
+#undef Z2
+
+#undef X3
+#undef Y3
+#undef Z3
+
 ANONYMOUS_NAMESPACE_END
 
 NAMESPACE_BEGIN(CryptoPP)
@@ -243,34 +742,14 @@ const ECP::Point& ECP::Inverse(const Point &P) const
 
 const ECP::Point& ECP::Add(const Point &P, const Point &Q) const
 {
-	if (P.identity) return Q;
+	AdditionFunction add(GetField(), m_a, m_b, m_R);
-	if (Q.identity) return P;
+	return (m_R = add(P, Q));
-	if (GetField().Equal(P.x, Q.x))
-		return GetField().Equal(P.y, Q.y) ? Double(P) : Identity();
-
-	FieldElement t = GetField().Subtract(Q.y, P.y);
-	t = GetField().Divide(t, GetField().Subtract(Q.x, P.x));
-	FieldElement x = GetField().Subtract(GetField().Subtract(GetField().Square(t), P.x), Q.x);
-	m_R.y = GetField().Subtract(GetField().Multiply(t, GetField().Subtract(P.x, x)), P.y);
-
-	m_R.x.swap(x);
-	m_R.identity = false;
-	return m_R;
 }
 
 const ECP::Point& ECP::Double(const Point &P) const
 {
-	if (P.identity || P.y==GetField().Identity()) return Identity();
+	AdditionFunction add(GetField(), m_a, m_b, m_R);
-
+	return (m_R = add(P));
-	FieldElement t = GetField().Square(P.x);
-	t = GetField().Add(GetField().Add(GetField().Double(t), t), m_a);
-	t = GetField().Divide(t, GetField().Double(P.y));
-	FieldElement x = GetField().Subtract(GetField().Subtract(GetField().Square(t), P.x), P.x);
-	m_R.y = GetField().Subtract(GetField().Multiply(t, GetField().Subtract(P.x, x)), P.y);
-
-	m_R.x.swap(x);
-	m_R.identity = false;
-	return m_R;
 }
 
 template <class T, class Iterator> void ParallelInvert(const AbstractRing<T> &ring, Iterator begin, Iterator end)
@@ -310,20 +789,11 @@ template <class T, class Iterator> void ParallelInvert(const AbstractRing<T> &ri
 	}
 }
 
-struct ProjectivePoint
-{
-	ProjectivePoint() {}
-	ProjectivePoint(const Integer &x, const Integer &y, const Integer &z)
-		: x(x), y(y), z(z)	{}
-
-	Integer x,y,z;
-};
-
 class ProjectiveDoubling
 {
 public:
 	ProjectiveDoubling(const ModularArithmetic &m_mr, const Integer &m_a, const Integer &m_b, const ECPPoint &Q)
-		: mr(m_mr), firstDoubling(true), negated(false)
+		: mr(m_mr)
 	{
 		CRYPTOPP_UNUSED(m_b);
 		if (Q.identity)
@@ -360,7 +830,6 @@ public:
 
 	const ModularArithmetic &mr;
 	ProjectivePoint P;
-	bool firstDoubling, negated;
 	Integer sixteenY4, aZ4, twoY, fourY2, S, M;
 };
 

ecp.h

@@ -35,11 +35,15 @@ public:
 	/// \brief Construct an ECP
 	ECP() {}
 
-	/// \brief Copy construct an ECP
+	/// \brief Construct an ECP
 	/// \param ecp the other ECP object
-	/// \param convertToMontgomeryRepresentation flag indicating if the curve should be converted to a MontgomeryRepresentation
+	/// \param convertToMontgomeryRepresentation flag indicating if the curve
+	///  should be converted to a MontgomeryRepresentation.
+	/// \details Prior to Crypto++ 8.3 the default value for
+	///  convertToMontgomeryRepresentation was false. it was changed due to
+	///  two audit tools finding, "Signature-compatible with a copy constructor".
 	/// \sa ModularArithmetic, MontgomeryRepresentation
-	ECP(const ECP &ecp, bool convertToMontgomeryRepresentation = false);
+	ECP(const ECP &ecp, bool convertToMontgomeryRepresentation);
 
 	/// \brief Construct an ECP
 	/// \param modulus the prime modulus
@@ -50,14 +54,22 @@ public:
 
 	/// \brief Construct an ECP from BER encoded parameters
 	/// \param bt BufferedTransformation derived object
-	/// \details This constructor will decode and extract the the fields fieldID and curve of the sequence ECParameters
+	/// \details This constructor will decode and extract the the fields
+	///  fieldID and curve of the sequence ECParameters
 	ECP(BufferedTransformation &bt);
 
-	/// \brief Encode the fields fieldID and curve of the sequence ECParameters
+	/// \brief DER Encode
 	/// \param bt BufferedTransformation derived object
+	/// \details DEREncode encode the fields fieldID and curve of the sequence
+	///  ECParameters
 	void DEREncode(BufferedTransformation &bt) const;
 
+	/// \brief Compare two points
+	/// \param P the first point
+	/// \param Q the second point
+	/// \returns true if equal, false otherwise
 	bool Equal(const Point &P, const Point &Q) const;
+
 	const Point& Identity() const;
 	const Point& Inverse(const Point &P) const;
 	bool InversionIsFast() const {return true;}

elgamal.h

@@ -11,14 +11,17 @@
 #include "integer.h"
 #include "gfpcrypt.h"
 #include "pubkey.h"
-#include "dsa.h"
 #include "misc.h"
+#include "oids.h"
+#include "dsa.h"
+#include "asn.h"
 
 NAMESPACE_BEGIN(CryptoPP)
 
 /// \brief ElGamal key agreement and encryption schemes base class
 /// \since Crypto++ 1.0
-class CRYPTOPP_NO_VTABLE ElGamalBase : public DL_KeyAgreementAlgorithm_DH<Integer, NoCofactorMultiplication>,
+class CRYPTOPP_NO_VTABLE ElGamalBase :
+	public DL_KeyAgreementAlgorithm_DH<Integer, NoCofactorMultiplication>,
 	public DL_KeyDerivationAlgorithm<Integer>,
 	public DL_SymmetricEncryptionAlgorithm
 {
@@ -27,7 +30,8 @@ public:
 
 	void Derive(const DL_GroupParameters<Integer> &groupParams, byte *derivedKey, size_t derivedLength, const Integer &agreedElement, const Integer &ephemeralPublicKey, const NameValuePairs &derivationParams) const
 	{
-		CRYPTOPP_UNUSED(groupParams), CRYPTOPP_UNUSED(ephemeralPublicKey), CRYPTOPP_UNUSED(derivationParams);
+		CRYPTOPP_UNUSED(groupParams); CRYPTOPP_UNUSED(ephemeralPublicKey);
+		CRYPTOPP_UNUSED(derivationParams);
 		agreedElement.Encode(derivedKey, derivedLength);
 	}
 
@@ -49,6 +53,8 @@ public:
 	size_t GetMaxSymmetricPlaintextLength(size_t cipherTextLength) const
 	{
 		unsigned int len = GetGroupParameters().GetModulus().ByteCount();
+		CRYPTOPP_ASSERT(len >= 3);
+
 		if (cipherTextLength == len)
 			return STDMIN(255U, len-3);
 		else
@@ -93,9 +99,14 @@ public:
 };
 
 /// \brief ElGamal key agreement and encryption schemes default implementation
+/// \tparam BASE Base class implementation
+/// \tparam SCHEME_OPTIONS Scheme options
+/// \tparam Key ElGamal key classes
 /// \since Crypto++ 1.0
 template <class BASE, class SCHEME_OPTIONS, class KEY>
-class ElGamalObjectImpl : public DL_ObjectImplBase<BASE, SCHEME_OPTIONS, KEY>, public ElGamalBase
+class ElGamalObjectImpl :
+	public DL_ObjectImplBase<BASE, SCHEME_OPTIONS, KEY>,
+	public ElGamalBase
 {
 public:
 	virtual ~ElGamalObjectImpl() {}
@@ -114,30 +125,174 @@ protected:
 	const DL_SymmetricEncryptionAlgorithm & GetSymmetricEncryptionAlgorithm() const {return *this;}
 };
 
+/// \brief ElGamal Public Key adapter
+/// \tparam BASE PublicKey derived class
+/// \details DL_PublicKey_ElGamal provides an override for GetAlgorithmID()
+///  to utilize 1.3.14.7.2.1.1. Prior to DL_PublicKey_ElGamal, the ElGamal
+///  keys [mistakenly] used the OID from DSA due to DL_GroupParmaters_GFP().
+///  If you need to <tt>Load</tt> an ElGamal key with the wrong OID then
+///  see <A HREF="https://www.cryptopp.com/wiki/ElGamal">ElGamal</A> on
+///  the Crypto++ wiki.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/876">Issue 876</A>,
+///  <A HREF="https://github.com/weidai11/cryptopp/issues/567">Issue 567</A>
+/// \since Crypto++ 8.3
+template <class BASE>
+struct DL_PublicKey_ElGamal : public BASE
+{
+	virtual ~DL_PublicKey_ElGamal() {}
+
+	/// \brief Retrieves the OID of the algorithm
+	/// \returns OID of the algorithm
+	/// \details DL_PrivateKey_ElGamal provides an override for GetAlgorithmID()
+	///  to utilize 1.3.14.7.2.1.1. Prior to DL_PrivateKey_ElGamal, the ElGamal
+	///  keys [mistakenly] used the OID from DSA due to DL_GroupParmaters_GFP().
+	///  If you need to <tt>Load</tt> an ElGamal key with the wrong OID then
+	///  see <A HREF="https://www.cryptopp.com/wiki/ElGamal">ElGamal</A> on
+	///  the Crypto++ wiki.
+	/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/876">Issue 876</A>,
+	///  <A HREF="https://github.com/weidai11/cryptopp/issues/567">Issue 567</A>
+	virtual OID GetAlgorithmID() const {
+		return ASN1::elGamal();
+	}
+};
+
+/// \brief ElGamal Private Key adapter
+/// \tparam BASE PrivateKey derived class
+/// \details DL_PrivateKey_ElGamal provides an override for GetAlgorithmID()
+///  to utilize 1.3.14.7.2.1.1. Prior to DL_PrivateKey_ElGamal, the ElGamal
+///  keys [mistakenly] used the OID from DSA due to DL_GroupParmaters_GFP().
+///  If you need to <tt>Load</tt> an ElGamal key with the wrong OID then
+///  see <A HREF="https://www.cryptopp.com/wiki/ElGamal">ElGamal</A> on
+///  the Crypto++ wiki.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/876">Issue 876</A>,
+///  <A HREF="https://github.com/weidai11/cryptopp/issues/567">Issue 567</A>
+/// \since Crypto++ 8.3
+template <class BASE>
+struct DL_PrivateKey_ElGamal : public BASE
+{
+	virtual ~DL_PrivateKey_ElGamal() {}
+
+	/// \brief Retrieves the OID of the algorithm
+	/// \returns OID of the algorithm
+	/// \details DL_PrivateKey_ElGamal provides an override for GetAlgorithmID()
+	///  to utilize 1.3.14.7.2.1.1. Prior to DL_PrivateKey_ElGamal, the ElGamal
+	///  keys [mistakenly] used the OID from DSA due to DL_GroupParmaters_GFP().
+	///  If you need to <tt>Load</tt> an ElGamal key with the wrong OID then
+	///  see <A HREF="https://www.cryptopp.com/wiki/ElGamal">ElGamal</A> on
+	///  the Crypto++ wiki.
+	/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/876">Issue 876</A>,
+	///  <A HREF="https://github.com/weidai11/cryptopp/issues/567">Issue 567</A>
+	virtual OID GetAlgorithmID() const {
+		return ASN1::elGamal();
+	}
+
+	/// \brief Check the key for errors
+	/// \param rng RandomNumberGenerator for objects which use randomized testing
+	/// \param level level of thoroughness
+	/// \return true if the tests succeed, false otherwise
+	/// \details There are four levels of thoroughness:
+	///   <ul>
+	///   <li>0 - using this object won't cause a crash or exception
+	///   <li>1 - this object will probably function, and encrypt, sign, other
+	///           operations correctly
+	///   <li>2 - ensure this object will function correctly, and perform
+	///           reasonable security checks
+	///   <li>3 - perform reasonable security checks, and do checks that may
+	///           take a long time
+	///   </ul>
+	/// \details Level 0 does not require a RandomNumberGenerator. A NullRNG() can
+	///  be used for level 0. Level 1 may not check for weak keys and such.
+	///  Levels 2 and 3 are recommended.
+	bool Validate(RandomNumberGenerator &rng, unsigned int level) const
+	{
+		// Validate() formerly used DL_PrivateKey_GFP implementation through
+		// inheritance. However, it would reject keys from other libraries
+		// like BouncyCastle. The failure was x < q. According to ElGamal's
+		// paper and the HAC, the private key is selected in over [1,p-1],
+		// Later Tsiounis and Yung showed the lower limit as [1,q-1] in
+		// "On the Security of EIGamal Based Encryption". As such, Crypto++
+		// will generate a key in the range [1,q-1], but accept a key
+		// in [1,p-1]. Thanks to JPM for finding the reference. Also see
+		// https://github.com/weidai11/cryptopp/commit/a5a684d92986.
+
+		CRYPTOPP_ASSERT(this->GetAbstractGroupParameters().Validate(rng, level));
+		bool pass = this->GetAbstractGroupParameters().Validate(rng, level);
+
+		const Integer &p = this->GetGroupParameters().GetModulus();
+		const Integer &q = this->GetAbstractGroupParameters().GetSubgroupOrder();
+		const Integer &x = this->GetPrivateExponent();
+
+		// Changed to x < p-1 based on ElGamal's paper and the HAC.
+		CRYPTOPP_ASSERT(x.IsPositive());
+		CRYPTOPP_ASSERT(x < p-1);
+		pass = pass && x.IsPositive() && x < p-1;
+
+		if (level >= 1)
+		{
+			// Minimum security level due to Tsiounis and Yung.
+			CRYPTOPP_ASSERT(Integer::Gcd(x, q) == Integer::One());
+			pass = pass && Integer::Gcd(x, q) == Integer::One();
+		}
+		return pass;
+	}
+};
+
 /// \brief ElGamal key agreement and encryption schemes keys
-/// \details The ElGamalKeys class used DL_PrivateKey_GFP_OldFormat and DL_PublicKey_GFP_OldFormat
+/// \details ElGamalKeys provide the algorithm implementation ElGamal key
-///   for the PrivateKey and PublicKey typedef from about Crypto++ 1.0 through Crypto++ 5.6.5.
+///  agreement and encryption schemes.
-///   At Crypto++ 6.0 the serialization format was cutover to standard PKCS8 and X509 encodings.
+/// \details The ElGamalKeys class used <tt>DL_PrivateKey_GFP_OldFormat</tt>
-/// \sa <A HREF="https://github.com/weidai11/cryptopp/commit/a5a684d92986e8e2">Commit a5a684d92986e8e2</A>
+///  and <tt>DL_PublicKey_GFP_OldFormat</tt> for the <tt>PrivateKey</tt> and
+///  <tt>PublicKey</tt> from about Crypto++ 1.0 through Crypto++ 5.6.5. At
+///  Crypto++ 6.0 the serialization format was cutover to standard PKCS8 and
+///  X509 encodings.
+/// \details The ElGamalKeys class [mistakenly] used the OID for DSA from
+///  about Crypto++ 1.0 through Crypto++ 8.2. At Crypto++ 8.3 the OID was
+///  fixed and now uses ElGamal encryption, which is 1.3.14.7.2.1.1.
+///  If you need to <tt>Load</tt> an ElGamal key with the wrong OID then
+///  see <A HREF="https://www.cryptopp.com/wiki/ElGamal">ElGamal</A> on
+///  the Crypto++ wiki.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/876">Issue 876</A>,
+///  <A HREF="https://github.com/weidai11/cryptopp/issues/567">Issue 567</A>
+/// \since Crypto++ 1.0
 struct ElGamalKeys
 {
+	/// \brief Implements DL_GroupParameters interface
 	typedef DL_CryptoKeys_GFP::GroupParameters GroupParameters;
-	typedef DL_CryptoKeys_GFP::PrivateKey PrivateKey;
+	/// \brief Implements DL_PrivateKey interface
-	typedef DL_CryptoKeys_GFP::PublicKey PublicKey;
+	typedef DL_PrivateKey_ElGamal<DL_CryptoKeys_GFP::PrivateKey> PrivateKey;
+	/// \brief Implements DL_PublicKey interface
+	typedef DL_PublicKey_ElGamal<DL_CryptoKeys_GFP::PublicKey> PublicKey;
 };
 
 /// \brief ElGamal encryption scheme with non-standard padding
+/// \details ElGamal provide the algorithm implementation ElGamal key
+///  agreement and encryption schemes.
+/// \details The ElGamal class [mistakenly] used the OID for DSA from about
+///  Crypto++ 1.0 through Crypto++ 8.2. At Crypto++ 8.3 the OID was fixed
+///  and now uses ElGamal encryption, which is 1.3.14.7.2.1.1.
+///  If you need to <tt>Load</tt> an ElGamal key with the wrong OID then
+///  see <A HREF="https://www.cryptopp.com/wiki/ElGamal">ElGamal</A> on
+///  the Crypto++ wiki.
+/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/876">Issue 876</A>,
+///  <A HREF="https://github.com/weidai11/cryptopp/issues/567">Issue 567</A>
 /// \since Crypto++ 1.0
 struct ElGamal
 {
 	typedef DL_CryptoSchemeOptions<ElGamal, ElGamalKeys, int, int, int> SchemeOptions;
+	typedef SchemeOptions::PrivateKey PrivateKey;
+	typedef SchemeOptions::PublicKey PublicKey;
 
+	/// \brief The algorithm name
+	/// \returns the algorithm name
+	/// \details StaticAlgorithmName returns the algorithm's name as a static
+	///  member function.
 	CRYPTOPP_STATIC_CONSTEXPR const char* StaticAlgorithmName() {return "ElgamalEnc/Crypto++Padding";}
 
+	/// \brief Implements DL_GroupParameters interface
 	typedef SchemeOptions::GroupParameters GroupParameters;
-	/// implements PK_Encryptor interface
+	/// \brief Implements PK_Encryptor interface
 	typedef PK_FinalTemplate<ElGamalObjectImpl<DL_EncryptorBase<Integer>, SchemeOptions, SchemeOptions::PublicKey> > Encryptor;
-	/// implements PK_Decryptor interface
+	/// \brief Implements PK_Encryptor interface
 	typedef PK_FinalTemplate<ElGamalObjectImpl<DL_DecryptorBase<Integer>, SchemeOptions, SchemeOptions::PrivateKey> > Decryptor;
 };
 

fhmqv.h

@@ -30,30 +30,85 @@ public:
 
   virtual ~FHMQV_Domain() {}
 
-  FHMQV_Domain(bool clientRole = true): m_role(clientRole ? RoleClient : RoleServer) {}
+  /// \brief Construct a FHMQV domain
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
+  FHMQV_Domain(bool clientRole = true)
+    : m_role(clientRole ? RoleClient : RoleServer) {}
 
+  /// \brief Construct a FHMQV domain
+  /// \param params group parameters and options
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   FHMQV_Domain(const GroupParameters &params, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer), m_groupParameters(params) {}
 
+  /// \brief Construct a FHMQV domain
+  /// \param bt BufferedTransformation with group parameters and options
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   FHMQV_Domain(BufferedTransformation &bt, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
     {m_groupParameters.BERDecode(bt);}
 
+  /// \brief Construct a FHMQV domain
+  /// \tparam T1 template parameter used as a constructor parameter
+  /// \param v1 first parameter
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details v1 is passed directly to the GROUP_PARAMETERS object.
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   template <class T1>
   FHMQV_Domain(T1 v1, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
     {m_groupParameters.Initialize(v1);}
 
+  /// \brief Construct a FHMQV domain
+  /// \tparam T1 template parameter used as a constructor parameter
+  /// \tparam T2 template parameter used as a constructor parameter
+  /// \param v1 first parameter
+  /// \param v2 second parameter
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details v1 and v2 are passed directly to the GROUP_PARAMETERS object.
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   template <class T1, class T2>
   FHMQV_Domain(T1 v1, T2 v2, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
     {m_groupParameters.Initialize(v1, v2);}
 
+  /// \brief Construct a FHMQV domain
+  /// \tparam T1 template parameter used as a constructor parameter
+  /// \tparam T2 template parameter used as a constructor parameter
+  /// \tparam T3 template parameter used as a constructor parameter
+  /// \param v1 first parameter
+  /// \param v2 second parameter
+  /// \param v3 third parameter
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details v1, v2 and v3 are passed directly to the GROUP_PARAMETERS object.
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   template <class T1, class T2, class T3>
   FHMQV_Domain(T1 v1, T2 v2, T3 v3, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
     {m_groupParameters.Initialize(v1, v2, v3);}
 
+  /// \brief Construct a FHMQV domain
+  /// \tparam T1 template parameter used as a constructor parameter
+  /// \tparam T2 template parameter used as a constructor parameter
+  /// \tparam T3 template parameter used as a constructor parameter
+  /// \tparam T4 template parameter used as a constructor parameter
+  /// \param v1 first parameter
+  /// \param v2 second parameter
+  /// \param v3 third parameter
+  /// \param v4 third parameter
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details v1, v2, v3 and v4 are passed directly to the GROUP_PARAMETERS object.
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   template <class T1, class T2, class T3, class T4>
   FHMQV_Domain(T1 v1, T2 v2, T3 v3, T4 v4, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
@@ -61,28 +116,62 @@ public:
 
 public:
 
+  /// \brief Retrieves the group parameters for this domain
+  /// \return the group parameters for this domain as a const reference
   const GroupParameters & GetGroupParameters() const {return m_groupParameters;}
+
+  /// \brief Retrieves the group parameters for this domain
+  /// \return the group parameters for this domain as a non-const reference
   GroupParameters & AccessGroupParameters() {return m_groupParameters;}
 
+  /// \brief Retrieves the crypto parameters for this domain
+  /// \return the crypto parameters for this domain as a non-const reference
   CryptoParameters & AccessCryptoParameters() {return AccessAbstractGroupParameters();}
 
-  /// return length of agreed value produced
+  /// \brief Provides the size of the agreed value
-  unsigned int AgreedValueLength() const {return GetAbstractGroupParameters().GetEncodedElementSize(false);}
+  /// \return size of agreed value produced in this domain
-  /// return length of static private keys in this domain
+  /// \details The length is calculated using <tt>GetEncodedElementSize(false)</tt>,
-  unsigned int StaticPrivateKeyLength() const {return GetAbstractGroupParameters().GetSubgroupOrder().ByteCount();}
+  ///  which means the element is encoded in a non-reversible format. A
-  /// return length of static public keys in this domain
+  ///  non-reversible format means its a raw byte array, and it lacks presentation
-  unsigned int StaticPublicKeyLength() const{return GetAbstractGroupParameters().GetEncodedElementSize(true);}
+  ///  format like an ASN.1 BIT_STRING or OCTET_STRING.
+  unsigned int AgreedValueLength() const
+    {return GetAbstractGroupParameters().GetEncodedElementSize(false);}
 
-  /// generate static private key
+  /// \brief Provides the size of the static private key
-  /*! \pre size of privateKey == PrivateStaticKeyLength() */
+  /// \return size of static private keys in this domain
+  /// \details The length is calculated using the byte count of the subgroup order.
+  unsigned int StaticPrivateKeyLength() const
+    {return GetAbstractGroupParameters().GetSubgroupOrder().ByteCount();}
+
+  /// \brief Provides the size of the static public key
+  /// \return size of static public keys in this domain
+  /// \details The length is calculated using <tt>GetEncodedElementSize(true)</tt>,
+  ///  which means the element is encoded in a reversible format. A reversible
+  ///  format means it has a presentation format, and its an ANS.1 encoded element
+  ///  or point.
+  unsigned int StaticPublicKeyLength() const
+    {return GetAbstractGroupParameters().GetEncodedElementSize(true);}
+
+  /// \brief Generate static private key in this domain
+  /// \param rng a RandomNumberGenerator derived class
+  /// \param privateKey a byte buffer for the generated private key in this domain
+  /// \details The private key is a random scalar used as an exponent in the range
+  ///  <tt>[1,MaxExponent()]</tt>.
+  /// \pre <tt>COUNTOF(privateKey) == PrivateStaticKeyLength()</tt>
   void GenerateStaticPrivateKey(RandomNumberGenerator &rng, byte *privateKey) const
   {
     Integer x(rng, Integer::One(), GetAbstractGroupParameters().GetMaxExponent());
     x.Encode(privateKey, StaticPrivateKeyLength());
   }
 
-  /// generate static public key
+  /// \brief Generate a static public key from a private key in this domain
-  /*! \pre size of publicKey == PublicStaticKeyLength() */
+  /// \param rng a RandomNumberGenerator derived class
+  /// \param privateKey a byte buffer with the previously generated private key
+  /// \param publicKey a byte buffer for the generated public key in this domain
+  /// \details The public key is an element or point on the curve, and its stored
+  ///  in a revrsible format. A reversible format means it has a presentation
+  ///  format, and its an ANS.1 encoded element or point.
+  /// \pre <tt>COUNTOF(publicKey) == PublicStaticKeyLength()</tt>
   void GenerateStaticPublicKey(RandomNumberGenerator &rng, const byte *privateKey, byte *publicKey) const
   {
     CRYPTOPP_UNUSED(rng);
@@ -92,10 +181,22 @@ public:
     params.EncodeElement(true, y, publicKey);
   }
 
+  /// \brief Provides the size of the ephemeral private key
+  /// \return size of ephemeral private keys in this domain
+  /// \details An ephemeral private key is a private key and public key.
+  ///  The serialized size is different than a static private key.
   unsigned int EphemeralPrivateKeyLength() const {return StaticPrivateKeyLength() + StaticPublicKeyLength();}
+
+  /// \brief Provides the size of the ephemeral public key
+  /// \return size of ephemeral public keys in this domain
+  /// \details An ephemeral public key is a public key.
+  ///  The serialized size is the same as a static public key.
   unsigned int EphemeralPublicKeyLength() const{return StaticPublicKeyLength();}
 
-  /// return length of ephemeral private keys in this domain
+  /// \brief Generate ephemeral private key in this domain
+  /// \param rng a RandomNumberGenerator derived class
+  /// \param privateKey a byte buffer for the generated private key in this domain
+  /// \pre <tt>COUNTOF(privateKey) == EphemeralPrivateKeyLength()</tt>
   void GenerateEphemeralPrivateKey(RandomNumberGenerator &rng, byte *privateKey) const
   {
     const DL_GroupParameters<Element> &params = GetAbstractGroupParameters();
@@ -105,28 +206,46 @@ public:
     params.EncodeElement(true, y, privateKey+StaticPrivateKeyLength());
   }
 
-  /// return length of ephemeral public keys in this domain
+  /// \brief Generate ephemeral public key from a private key in this domain
+  /// \param rng a RandomNumberGenerator derived class
+  /// \param privateKey a byte buffer with the previously generated private key
+  /// \param publicKey a byte buffer for the generated public key in this domain
+  /// \pre <tt>COUNTOF(publicKey) == EphemeralPublicKeyLength()</tt>
   void GenerateEphemeralPublicKey(RandomNumberGenerator &rng, const byte *privateKey, byte *publicKey) const
   {
     CRYPTOPP_UNUSED(rng);
     memcpy(publicKey, privateKey+StaticPrivateKeyLength(), EphemeralPublicKeyLength());
   }
 
-  /// derive agreed value from your private keys and couterparty's public keys, return false in case of failure
+  /// \brief Derive agreed value or shared secret
-  /*! \note The ephemeral public key will always be validated.
+  /// \param agreedValue the shared secret
-  If you have previously validated the static public key, use validateStaticOtherPublicKey=false to save time.
+  /// \param staticPrivateKey your long term private key
-  \pre size of agreedValue == AgreedValueLength()
+  /// \param ephemeralPrivateKey your ephemeral private key
-  \pre length of staticPrivateKey == StaticPrivateKeyLength()
+  /// \param staticOtherPublicKey couterparty's long term public key
-  \pre length of ephemeralPrivateKey == EphemeralPrivateKeyLength()
+  /// \param ephemeralOtherPublicKey couterparty's ephemeral public key
-  \pre length of staticOtherPublicKey == StaticPublicKeyLength()
+  /// \param validateStaticOtherPublicKey flag indicating validation
-  \pre length of ephemeralOtherPublicKey == EphemeralPublicKeyLength()
+  /// \return true upon success, false in case of failure
-  */
+  /// \details Agree() performs the authenticated key agreement. Agree()
+  ///  derives a shared secret from your private keys and couterparty's
+  ///  public keys. Each instance or run of the protocol should use a new
+  ///  ephemeral key pair.
+  /// \details The other's ephemeral public key will always be validated at
+  ///  Level 1 to ensure it is a point on the curve.
+  ///  <tt>validateStaticOtherPublicKey</tt> determines how thoroughly other's
+  ///  static public key is validated. If you have previously validated the
+  ///  couterparty's static public key, then use
+  ///  <tt>validateStaticOtherPublicKey=false</tt> to save time.
+  /// \pre <tt>COUNTOF(agreedValue) == AgreedValueLength()
+  /// \pre <tt>COUNTOF(staticPrivateKey) == StaticPrivateKeyLength()</tt>
+  /// \pre <tt>COUNTOF(ephemeralPrivateKey) == EphemeralPrivateKeyLength()</tt>
+  /// \pre <tt>COUNTOF(staticOtherPublicKey) == StaticPublicKeyLength()</tt>
+  /// \pre <tt>COUNTOF(ephemeralOtherPublicKey) == EphemeralPublicKeyLength()</tt>
   bool Agree(byte *agreedValue,
     const byte *staticPrivateKey, const byte *ephemeralPrivateKey,
     const byte *staticOtherPublicKey, const byte *ephemeralOtherPublicKey,
     bool validateStaticOtherPublicKey=true) const
   {
-    byte *XX = NULLPTR, *YY = NULLPTR, *AA = NULLPTR, *BB = NULLPTR;
+    const byte *XX = NULLPTR, *YY = NULLPTR, *AA = NULLPTR, *BB = NULLPTR;
     size_t xxs = 0, yys = 0, aas = 0, bbs = 0;
 
     // Depending on the role, this will hold either A's or B's static
@@ -144,60 +263,43 @@ public:
         Element B = params.ExponentiateBase(b);
         params.EncodeElement(true, B, tt);
 
-        XX = const_cast<byte*>(ephemeralOtherPublicKey);
+        XX = ephemeralOtherPublicKey;
         xxs = EphemeralPublicKeyLength();
-        YY = const_cast<byte*>(ephemeralPrivateKey) + StaticPrivateKeyLength();
+        YY = ephemeralPrivateKey + StaticPrivateKeyLength();
         yys = EphemeralPublicKeyLength();
-        AA = const_cast<byte*>(staticOtherPublicKey);
+        AA = staticOtherPublicKey;
         aas = StaticPublicKeyLength();
         BB = tt.BytePtr();
         bbs = tt.SizeInBytes();
       }
-      else if(m_role == RoleClient)
+      else
       {
         Integer a(staticPrivateKey, StaticPrivateKeyLength());
         Element A = params.ExponentiateBase(a);
         params.EncodeElement(true, A, tt);
 
-        XX = const_cast<byte*>(ephemeralPrivateKey) + StaticPrivateKeyLength();
+        XX = ephemeralPrivateKey + StaticPrivateKeyLength();
         xxs = EphemeralPublicKeyLength();
-        YY = const_cast<byte*>(ephemeralOtherPublicKey);
+        YY = ephemeralOtherPublicKey;
         yys = EphemeralPublicKeyLength();
         AA = tt.BytePtr();
         aas = tt.SizeInBytes();
-        BB = const_cast<byte*>(staticOtherPublicKey);
+        BB = staticOtherPublicKey;
         bbs = StaticPublicKeyLength();
       }
-      else
-      {
-        CRYPTOPP_ASSERT(0);
-        return false;
-      }
 
-      // DecodeElement calls ValidateElement at level 1. Level 1 only calls
+      Element VV1 = params.DecodeElement(staticOtherPublicKey, validateStaticOtherPublicKey);
-      // VerifyPoint to ensure the element is in G*. If the other's PublicKey is
+      Element VV2 = params.DecodeElement(ephemeralOtherPublicKey, true);
-      // requested to be validated, we manually call ValidateElement at level 3.
-      Element VV1 = params.DecodeElement(staticOtherPublicKey, false);
-      if(!params.ValidateElement(validateStaticOtherPublicKey ? 3 : 1, VV1, NULLPTR))
-        return false;
-
-      // DecodeElement calls ValidateElement at level 1. Level 1 only calls
-      // VerifyPoint to ensure the element is in G*. Crank it up.
-      Element VV2 = params.DecodeElement(ephemeralOtherPublicKey, false);
-      if(!params.ValidateElement(3, VV2, NULLPTR))
-        return false;
 
       const Integer& q = params.GetSubgroupOrder();
       const unsigned int len /*bytes*/ = (((q.BitCount()+1)/2 +7)/8);
-
-      Integer d, e;
       SecByteBlock dd(len), ee(len);
 
       Hash(NULLPTR, XX, xxs, YY, yys, AA, aas, BB, bbs, dd.BytePtr(), dd.SizeInBytes());
-      d.Decode(dd.BytePtr(), dd.SizeInBytes());
+      Integer d(dd.BytePtr(), dd.SizeInBytes());
 
       Hash(NULLPTR, YY, yys, XX, xxs, AA, aas, BB, bbs, ee.BytePtr(), ee.SizeInBytes());
-      e.Decode(ee.BytePtr(), ee.SizeInBytes());
+      Integer e(ee.BytePtr(), ee.SizeInBytes());
 
       Element sigma;
       if(m_role == RoleServer)
@@ -233,6 +335,7 @@ public:
     }
     catch (DL_BadElement &)
     {
+      CRYPTOPP_ASSERT(0);
       return false;
     }
     return true;
@@ -251,9 +354,11 @@ protected:
 
     if(sigma)
     {
-      Integer x = GetAbstractGroupParameters().ConvertElementToInteger(*sigma);
+      //Integer x = GetAbstractGroupParameters().ConvertElementToInteger(*sigma);
-      SecByteBlock sbb(x.MinEncodedSize());
+      //SecByteBlock sbb(x.MinEncodedSize());
-      x.Encode(sbb.BytePtr(), sbb.SizeInBytes());
+      //x.Encode(sbb.BytePtr(), sbb.SizeInBytes());
+      SecByteBlock sbb(GetAbstractGroupParameters().GetEncodedElementSize(false));
+      GetAbstractGroupParameters().EncodeElement(false, *sigma, sbb);
       hash.Update(sbb.BytePtr(), sbb.SizeInBytes());
     }
 
@@ -294,7 +399,7 @@ private:
 /// \details This implementation follows Augustin P. Sarr and Philippe Elbaz–Vincent, and Jean–Claude Bajard's
 ///   <a href="http://eprint.iacr.org/2009/408">A Secure and Efficient Authenticated Diffie-Hellman Protocol</a>.
 ///   Note: this is FHMQV, Protocol 5, from page 11; and not FHMQV-C.
-/// \sa FHMQV, MQV_Domain, HMQV_Domain, AuthenticatedKeyAgreementDomain
+/// \sa FHMQV, MQV_Domain, FHMQV_Domain, AuthenticatedKeyAgreementDomain
 /// \since Crypto++ 5.6.4
 typedef FHMQV_Domain<DL_GroupParameters_GFP_DefaultSafePrime> FHMQV;
 

filters.cpp

@@ -595,7 +595,9 @@ StreamTransformationFilter::StreamTransformationFilter(StreamTransformation &c,
 	m_isSpecial = m_cipher.IsLastBlockSpecial() && m_mandatoryBlockSize > 1;
 	m_reservedBufferSize = STDMAX(2*m_mandatoryBlockSize, m_optimalBufferSize);
 
-	IsolatedInitialize(MakeParameters(Name::BlockPaddingScheme(), padding));
+	FilterWithBufferedInput::IsolatedInitialize(
+		MakeParameters
+			(Name::BlockPaddingScheme(), padding));
 }
 
 StreamTransformationFilter::StreamTransformationFilter(StreamTransformation &c, BufferedTransformation *attachment, BlockPaddingScheme padding, bool authenticated)
@@ -616,7 +618,9 @@ StreamTransformationFilter::StreamTransformationFilter(StreamTransformation &c,
 	m_isSpecial = m_cipher.IsLastBlockSpecial() && m_mandatoryBlockSize > 1;
 	m_reservedBufferSize = STDMAX(2*m_mandatoryBlockSize, m_optimalBufferSize);
 
-	IsolatedInitialize(MakeParameters(Name::BlockPaddingScheme(), padding));
+	FilterWithBufferedInput::IsolatedInitialize(
+		MakeParameters
+			(Name::BlockPaddingScheme(), padding));
 }
 
 size_t StreamTransformationFilter::LastBlockSize(StreamTransformation &c, BlockPaddingScheme padding)
@@ -889,7 +893,10 @@ HashVerificationFilter::HashVerificationFilter(HashTransformation &hm, BufferedT
 	: FilterWithBufferedInput(attachment)
 	, m_hashModule(hm), m_flags(0), m_digestSize(0), m_verified(false)
 {
-	IsolatedInitialize(MakeParameters(Name::HashVerificationFilterFlags(), flags)(Name::TruncatedDigestSize(), truncatedDigestSize));
+	FilterWithBufferedInput::IsolatedInitialize(
+		MakeParameters
+			(Name::HashVerificationFilterFlags(), flags)
+			(Name::TruncatedDigestSize(), truncatedDigestSize));
 }
 
 void HashVerificationFilter::InitializeDerivedAndReturnNewSizes(const NameValuePairs &parameters, size_t &firstSize, size_t &blockSize, size_t &lastSize)
@@ -994,7 +1001,11 @@ AuthenticatedDecryptionFilter::AuthenticatedDecryptionFilter(AuthenticatedSymmet
 	, m_streamFilter(c, new OutputProxy(*this, false), padding, true)
 {
 	CRYPTOPP_ASSERT(!c.IsForwardTransformation() || c.IsSelfInverting());
-	IsolatedInitialize(MakeParameters(Name::BlockPaddingScheme(), padding)(Name::AuthenticatedDecryptionFilterFlags(), flags)(Name::TruncatedDigestSize(), truncatedDigestSize));
+	FilterWithBufferedInput::IsolatedInitialize(
+		MakeParameters
+			(Name::BlockPaddingScheme(), padding)
+			(Name::AuthenticatedDecryptionFilterFlags(), flags)
+			(Name::TruncatedDigestSize(), truncatedDigestSize));
 }
 
 void AuthenticatedDecryptionFilter::InitializeDerivedAndReturnNewSizes(const NameValuePairs &parameters, size_t &firstSize, size_t &blockSize, size_t &lastSize)
@@ -1079,7 +1090,9 @@ SignatureVerificationFilter::SignatureVerificationFilter(const PK_Verifier &veri
 	: FilterWithBufferedInput(attachment)
 	, m_verifier(verifier), m_flags(0), m_verified(0)
 {
-	IsolatedInitialize(MakeParameters(Name::SignatureVerificationFilterFlags(), flags));
+	FilterWithBufferedInput::IsolatedInitialize(
+		MakeParameters
+			(Name::SignatureVerificationFilterFlags(), flags));
 }
 
 void SignatureVerificationFilter::InitializeDerivedAndReturnNewSizes(const NameValuePairs &parameters, size_t &firstSize, size_t &blockSize, size_t &lastSize)

filters.h

@@ -474,7 +474,7 @@ struct BlockPaddingSchemeDef
 		/// \brief 0's padding added to a block
 		/// \since Crypto++ 5.0
 		ZEROS_PADDING,
-		/// \brief PKCS #5 padding added to a block
+		/// \brief PKCS padding added to a block
 		/// \since Crypto++ 5.0
 		PKCS_PADDING,
 		/// \brief 1 and 0's padding added to a block
@@ -1100,6 +1100,7 @@ CRYPTOPP_DLL_TEMPLATE_CLASS StringSinkTemplate<std::string>;
 
 /// \brief Append input to a std::vector<byte> object
 /// \details VectorSink is a typedef for StringSinkTemplate<std::vector<byte> >.
+/// \since Crypto++ 8.0
 DOCUMENTED_TYPEDEF(StringSinkTemplate<std::vector<byte> >, VectorSink)
 CRYPTOPP_DLL_TEMPLATE_CLASS StringSinkTemplate<std::vector<byte> >;
 

fips140.cpp

@@ -47,31 +47,35 @@ bool PowerUpSelfTestInProgressOnThisThread()
 {
 #if CRYPTOPP_ENABLE_COMPLIANCE_WITH_FIPS_140_2
 	return s_inProgress;
-#endif
+#else
 	return false;
+#endif
 }
 
 void SetPowerUpSelfTestInProgressOnThisThread(bool inProgress)
 {
-	CRYPTOPP_UNUSED(inProgress);
 #if CRYPTOPP_ENABLE_COMPLIANCE_WITH_FIPS_140_2
 	s_inProgress = inProgress;
+#else
+	CRYPTOPP_UNUSED(inProgress);
 #endif
 }
 
 void EncryptionPairwiseConsistencyTest_FIPS_140_Only(const PK_Encryptor &encryptor, const PK_Decryptor &decryptor)
 {
-	CRYPTOPP_UNUSED(encryptor), CRYPTOPP_UNUSED(decryptor);
 #if CRYPTOPP_ENABLE_COMPLIANCE_WITH_FIPS_140_2
 	EncryptionPairwiseConsistencyTest(encryptor, decryptor);
+#else
+	CRYPTOPP_UNUSED(encryptor), CRYPTOPP_UNUSED(decryptor);
 #endif
 }
 
 void SignaturePairwiseConsistencyTest_FIPS_140_Only(const PK_Signer &signer, const PK_Verifier &verifier)
 {
-	CRYPTOPP_UNUSED(signer), CRYPTOPP_UNUSED(verifier);
 #if CRYPTOPP_ENABLE_COMPLIANCE_WITH_FIPS_140_2
 	SignaturePairwiseConsistencyTest(signer, verifier);
+#else
+	CRYPTOPP_UNUSED(signer), CRYPTOPP_UNUSED(verifier);
 #endif
 }
 

gcm.cpp

@@ -44,7 +44,7 @@ NAMESPACE_BEGIN(CryptoPP)
 #endif
 #endif  // CRYPTOPP_BOOL_X86 || CRYPTOPP_BOOL_X32 || CRYPTOPP_BOOL_X64
 
-// Clang __m128i casts, http://bugs.llvm.org/show_bug.cgi?id=20670
+// Clang intrinsic casts, http://bugs.llvm.org/show_bug.cgi?id=20670
 #define M128_CAST(x) ((__m128i *)(void *)(x))
 #define CONST_M128_CAST(x) ((const __m128i *)(const void *)(x))
 
@@ -75,8 +75,8 @@ extern void GCM_Xor16_SSE2(byte *a, const byte *b, const byte *c);
 extern void GCM_Xor16_NEON(byte *a, const byte *b, const byte *c);
 #endif
 
-#if CRYPTOPP_POWER7_AVAILABLE
+#if CRYPTOPP_POWER8_AVAILABLE
-extern void GCM_Xor16_POWER7(byte *a, const byte *b, const byte *c);
+extern void GCM_Xor16_POWER8(byte *a, const byte *b, const byte *c);
 #endif
 
 #if CRYPTOPP_CLMUL_AVAILABLE
@@ -213,11 +213,11 @@ void GCM_Base::SetKeyWithoutResync(const byte *userKey, size_t keylength, const
                     for (k=1; k<j; k++)
                         GCM_Xor16_NEON(mulTable+i*256*16+(j+k)*16, mulTable+i*256*16+j*16, mulTable+i*256*16+k*16);
             else
-#elif CRYPTOPP_POWER7_AVAILABLE
+#elif CRYPTOPP_POWER8_AVAILABLE
-            if (HasPower7())
+            if (HasPower8())
                 for (j=2; j<=0x80; j*=2)
                     for (k=1; k<j; k++)
-                        GCM_Xor16_POWER7(mulTable+i*256*16+(j+k)*16, mulTable+i*256*16+j*16, mulTable+i*256*16+k*16);
+                        GCM_Xor16_POWER8(mulTable+i*256*16+(j+k)*16, mulTable+i*256*16+j*16, mulTable+i*256*16+k*16);
             else
 #endif
                 for (j=2; j<=0x80; j*=2)
@@ -277,13 +277,13 @@ void GCM_Base::SetKeyWithoutResync(const byte *userKey, size_t keylength, const
                         GCM_Xor16_NEON(mulTable+1024+i*256+(j+k)*16, mulTable+1024+i*256+j*16, mulTable+1024+i*256+k*16);
                     }
             else
-#elif CRYPTOPP_POWER7_AVAILABLE
+#elif CRYPTOPP_POWER8_AVAILABLE
-            if (HasPower7())
+            if (HasPower8())
                 for (j=2; j<=8; j*=2)
                     for (k=1; k<j; k++)
                     {
-                        GCM_Xor16_POWER7(mulTable+i*256+(j+k)*16, mulTable+i*256+j*16, mulTable+i*256+k*16);
+                        GCM_Xor16_POWER8(mulTable+i*256+(j+k)*16, mulTable+i*256+j*16, mulTable+i*256+k*16);
-                        GCM_Xor16_POWER7(mulTable+1024+i*256+(j+k)*16, mulTable+1024+i*256+j*16, mulTable+1024+i*256+k*16);
+                        GCM_Xor16_POWER8(mulTable+1024+i*256+(j+k)*16, mulTable+1024+i*256+j*16, mulTable+1024+i*256+k*16);
                     }
             else
 #endif
@@ -369,8 +369,8 @@ unsigned int GCM_Base::OptimalDataAlignment() const
         HasSSE2() ? 16 :
 #elif CRYPTOPP_ARM_NEON_AVAILABLE
         HasNEON() ? 4 :
-#elif CRYPTOPP_POWER7_AVAILABLE
+#elif CRYPTOPP_POWER8_AVAILABLE
-        HasPower7() ? 16 :
+        HasPower8() ? 16 :
 #endif
         GetBlockCipher().OptimalDataAlignment();
 }

gcm_simd.cpp

@@ -56,14 +56,10 @@
 # define EXCEPTION_EXECUTE_HANDLER 1
 #endif
 
-// Clang __m128i casts, http://bugs.llvm.org/show_bug.cgi?id=20670
+// Clang intrinsic casts, http://bugs.llvm.org/show_bug.cgi?id=20670
 #define M128_CAST(x) ((__m128i *)(void *)(x))
 #define CONST_M128_CAST(x) ((const __m128i *)(const void *)(x))
 
-// GCC cast warning
-#define UINT64X2_CAST(x) ((uint64x2_t *)(void *)(x))
-#define CONST_UINT64X2_CAST(x) ((const uint64x2_t *)(const void *)(x))
-
 // Squash MS LNK4221 and libtool warnings
 extern const char GCM_SIMD_FNAME[] = __FILE__;
 
@@ -219,10 +215,7 @@ bool CPU_ProbePMULL()
 #if CRYPTOPP_ARM_NEON_AVAILABLE
 void GCM_Xor16_NEON(byte *a, const byte *b, const byte *c)
 {
-    CRYPTOPP_ASSERT(IsAlignedOn(a,GetAlignmentOf<uint64x2_t>()));
+	vst1q_u8(a, veorq_u8(vld1q_u8(b), vld1q_u8(c)));
-    CRYPTOPP_ASSERT(IsAlignedOn(b,GetAlignmentOf<uint64x2_t>()));
-    CRYPTOPP_ASSERT(IsAlignedOn(c,GetAlignmentOf<uint64x2_t>()));
-    *UINT64X2_CAST(a) = veorq_u64(*CONST_UINT64X2_CAST(b), *CONST_UINT64X2_CAST(c));
 }
 #endif  // CRYPTOPP_ARM_NEON_AVAILABLE
 
@@ -569,12 +562,12 @@ void GCM_ReverseHashBufferIfNeeded_CLMUL(byte *hashBuffer)
 
 // ***************************** POWER8 ***************************** //
 
-#if CRYPTOPP_POWER7_AVAILABLE
+#if CRYPTOPP_POWER8_AVAILABLE
-void GCM_Xor16_POWER7(byte *a, const byte *b, const byte *c)
+void GCM_Xor16_POWER8(byte *a, const byte *b, const byte *c)
 {
     VecStore(VecXor(VecLoad(b), VecLoad(c)), a);
 }
-#endif  // CRYPTOPP_POWER7_AVAILABLE
+#endif  // CRYPTOPP_POWER8_AVAILABLE
 
 #if CRYPTOPP_POWER8_VMULL_AVAILABLE
 

gf2n_simd.cpp

@@ -21,6 +21,8 @@
 #include "pch.h"
 #include "config.h"
 
+#ifndef CRYPTOPP_IMPORTS
+
 #include "gf2n.h"
 
 #if (CRYPTOPP_CLMUL_AVAILABLE)
@@ -463,36 +465,33 @@ NAMESPACE_BEGIN(CryptoPP)
 void
 GF2NT_233_Multiply_Reduce_CLMUL(const word* pA, const word* pB, word* pC)
 {
-    const __m128i* pAA = reinterpret_cast<const __m128i*>(pA);
+    enum {S=sizeof(__m128i)/sizeof(word)};
-    const __m128i* pBB = reinterpret_cast<const __m128i*>(pB);
+    __m128i a0 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(pA+0*S));
-    __m128i a0 = _mm_loadu_si128(pAA+0);
+    __m128i a1 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(pA+1*S));
-    __m128i a1 = _mm_loadu_si128(pAA+1);
+    __m128i b0 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(pB+0*S));
-    __m128i b0 = _mm_loadu_si128(pBB+0);
+    __m128i b1 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(pB+1*S));
-    __m128i b1 = _mm_loadu_si128(pBB+1);
 
     __m128i c0, c1, c2, c3;
     F2N_Multiply_256x256_CLMUL(c3, c2, c1, c0, a1, a0, b1, b0);
     GF2NT_233_Reduce_CLMUL(c3, c2, c1, c0);
 
-    __m128i* pCC = reinterpret_cast<__m128i*>(pC);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(pC+0*S), c0);
-    _mm_storeu_si128(pCC+0, c0);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(pC+1*S), c1);
-    _mm_storeu_si128(pCC+1, c1);
 }
 
 void
 GF2NT_233_Square_Reduce_CLMUL(const word* pA, word* pC)
 {
-    const __m128i* pAA = reinterpret_cast<const __m128i*>(pA);
+    enum {S=sizeof(__m128i)/sizeof(word)};
-    __m128i a0 = _mm_loadu_si128(pAA+0);
+    __m128i a0 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(pA+0*S));
-    __m128i a1 = _mm_loadu_si128(pAA+1);
+    __m128i a1 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(pA+1*S));
 
     __m128i c0, c1, c2, c3;
     F2N_Square_256_CLMUL(c3, c2, c1, c0, a1, a0);
     GF2NT_233_Reduce_CLMUL(c3, c2, c1, c0);
 
-    __m128i* pCC = reinterpret_cast<__m128i*>(pC);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(pC+0*S), c0);
-    _mm_storeu_si128(pCC+0, c0);
+    _mm_storeu_si128(reinterpret_cast<__m128i*>(pC+1*S), c1);
-    _mm_storeu_si128(pCC+1, c1);
 }
 
 #elif (CRYPTOPP_ARM_PMULL_AVAILABLE)
@@ -608,3 +607,5 @@ GF2NT_233_Square_Reduce_POWER8(const word* pA, word* pC)
 #endif
 
 NAMESPACE_END
+
+#endif  // CRYPTOPP_IMPORTS

gfpcrypt.cpp

@@ -133,22 +133,27 @@ void DL_SignatureMessageEncodingMethod_NR::ComputeMessageRepresentative(RandomNu
 bool DL_GroupParameters_IntegerBased::ValidateGroup(RandomNumberGenerator &rng, unsigned int level) const
 {
 	const Integer &p = GetModulus(), &q = GetSubgroupOrder();
-
 	bool pass = true;
+
+	CRYPTOPP_ASSERT(p > Integer::One() && p.IsOdd());
 	pass = pass && p > Integer::One() && p.IsOdd();
-	CRYPTOPP_ASSERT(pass);
+
+	CRYPTOPP_ASSERT(q > Integer::One() && q.IsOdd());
 	pass = pass && q > Integer::One() && q.IsOdd();
-	CRYPTOPP_ASSERT(pass);
 
 	if (level >= 1)
 	{
+		CRYPTOPP_ASSERT(GetCofactor() > Integer::One());
+		CRYPTOPP_ASSERT(GetGroupOrder() % q == Integer::Zero());
+
 		pass = pass && GetCofactor() > Integer::One() && GetGroupOrder() % q == Integer::Zero();
-		CRYPTOPP_ASSERT(pass);
 	}
 	if (level >= 2)
 	{
+		CRYPTOPP_ASSERT(VerifyPrime(rng, q, level-2));
+		CRYPTOPP_ASSERT(VerifyPrime(rng, p, level-2));
+
 		pass = pass && VerifyPrime(rng, q, level-2) && VerifyPrime(rng, p, level-2);
-		CRYPTOPP_ASSERT(pass);
 	}
 
 	return pass;
@@ -157,28 +162,28 @@ bool DL_GroupParameters_IntegerBased::ValidateGroup(RandomNumberGenerator &rng,
 bool DL_GroupParameters_IntegerBased::ValidateElement(unsigned int level, const Integer &g, const DL_FixedBasePrecomputation<Integer> *gpc) const
 {
 	const Integer &p = GetModulus(), &q = GetSubgroupOrder();
-
 	bool pass = true;
-	pass = pass && GetFieldType() == 1 ? g.IsPositive() : g.NotNegative();
-	CRYPTOPP_ASSERT(pass);
 
+	CRYPTOPP_ASSERT(GetFieldType() == 1 ? g.IsPositive() : g.NotNegative());
+	pass = pass && GetFieldType() == 1 ? g.IsPositive() : g.NotNegative();
+
+	CRYPTOPP_ASSERT(g < p && !IsIdentity(g));
 	pass = pass && g < p && !IsIdentity(g);
-	CRYPTOPP_ASSERT(pass);
 
 	if (level >= 1)
 	{
 		if (gpc)
 		{
+			CRYPTOPP_ASSERT(gpc->Exponentiate(GetGroupPrecomputation(), Integer::One()) == g);
 			pass = pass && gpc->Exponentiate(GetGroupPrecomputation(), Integer::One()) == g;
-			CRYPTOPP_ASSERT(pass);
 		}
 	}
 	if (level >= 2)
 	{
 		if (GetFieldType() == 2)
 		{
+			CRYPTOPP_ASSERT(Jacobi(g*g-4, p)==-1);
 			pass = pass && Jacobi(g*g-4, p)==-1;
-			CRYPTOPP_ASSERT(pass);
 		}
 
 		// verifying that Lucas((p+1)/2, w, p)==2 is omitted because it's too costly
@@ -188,13 +193,13 @@ bool DL_GroupParameters_IntegerBased::ValidateElement(unsigned int level, const
 		if (fullValidate && pass)
 		{
 			Integer gp = gpc ? gpc->Exponentiate(GetGroupPrecomputation(), q) : ExponentiateElement(g, q);
+			CRYPTOPP_ASSERT(IsIdentity(gp));
 			pass = pass && IsIdentity(gp);
-			CRYPTOPP_ASSERT(pass);
 		}
 		else if (GetFieldType() == 1)
 		{
+			CRYPTOPP_ASSERT(Jacobi(g, p) == 1);
 			pass = pass && Jacobi(g, p) == 1;
-			CRYPTOPP_ASSERT(pass);
 		}
 	}
 

gfpcrypt.h

@@ -69,8 +69,13 @@ public:
     void BERDecode(BufferedTransformation &bt);
     void DEREncode(BufferedTransformation &bt) const;
 
-    // GeneratibleCryptoMaterial interface
+    /// \brief Generate a random key
-    /*! parameters: (ModulusSize, SubgroupOrderSize (optional)) */
+    /// \param rng a RandomNumberGenerator to produce keying material
+    /// \param params additional initialization parameters
+    /// \details Recognised NameValuePairs are ModulusSize and
+    ///  SubgroupOrderSize (optional)
+    /// \throws KeyingErr if a key can't be generated or algorithm parameters
+    ///  are invalid
     void GenerateRandom(RandomNumberGenerator &rng, const NameValuePairs &alg);
     bool GetVoidValue(const char *name, const std::type_info &valueType, void *pValue) const;
     void AssignFrom(const NameValuePairs &source);
@@ -80,26 +85,68 @@ public:
     Integer GetGroupOrder() const {return GetFieldType() == 1 ? GetModulus()-Integer::One() : GetModulus()+Integer::One();}
     bool ValidateGroup(RandomNumberGenerator &rng, unsigned int level) const;
     bool ValidateElement(unsigned int level, const Integer &element, const DL_FixedBasePrecomputation<Integer> *precomp) const;
+
+    /// \brief Determine if subgroup membership check is fast
+    /// \returns true or false
     bool FastSubgroupCheckAvailable() const {return GetCofactor() == 2;}
 
-    // Cygwin i386 crash at -O3; see http://github.com/weidai11/cryptopp/issues/40.
+    /// \brief Encodes the element
+    /// \param reversible flag indicating the encoding format
+    /// \param element reference to the element to encode
+    /// \param encoded destination byte array for the encoded element
+    /// \details EncodeElement() must be implemented in a derived class.
+    /// \pre <tt>COUNTOF(encoded) == GetEncodedElementSize()</tt>
+    /// \sa <A HREF="http://github.com/weidai11/cryptopp/issues/40">Cygwin
+    ///  i386 crash at -O3</A>
     void EncodeElement(bool reversible, const Element &element, byte *encoded) const;
+
+    /// \brief Retrieve the encoded element's size
+    /// \param reversible flag indicating the encoding format
+    /// \return encoded element's size, in bytes
+    /// \details The format of the encoded element varies by the underlying
+    ///  type of the element and the reversible flag.
+    /// \sa GetEncodedElementSize(), EncodeElement(), DecodeElement()
     unsigned int GetEncodedElementSize(bool reversible) const;
 
+    /// \brief Decodes the element
+    /// \param encoded byte array with the encoded element
+    /// \param checkForGroupMembership flag indicating if the element should be validated
+    /// \return Element after decoding
+    /// \details DecodeElement() must be implemented in a derived class.
+    /// \pre <tt>COUNTOF(encoded) == GetEncodedElementSize()</tt>
     Integer DecodeElement(const byte *encoded, bool checkForGroupMembership) const;
+
+    /// \brief Converts an element to an Integer
+    /// \param element the element to convert to an Integer
+    /// \return Element after converting to an Integer
+    /// \details ConvertElementToInteger() must be implemented in a derived class.
     Integer ConvertElementToInteger(const Element &element) const
         {return element;}
-    Integer GetMaxExponent() const;
-    static std::string CRYPTOPP_API StaticAlgorithmNamePrefix() {return "";}
 
+    /// \brief Retrieve the maximum exponent for the group
+    /// \return the maximum exponent for the group
+    Integer GetMaxExponent() const;
+
+    /// \brief Retrieve the OID of the algorithm
+    /// \returns OID of the algorithm
     OID GetAlgorithmID() const;
 
+    /// \brief Retrieve the modulus for the group
+    /// \return the modulus for the group
     virtual const Integer & GetModulus() const =0;
+
+    /// \brief Set group parameters
+    /// \param p the prime modulus
+    /// \param g the group generator
     virtual void SetModulusAndSubgroupGenerator(const Integer &p, const Integer &g) =0;
 
+    /// \brief Set subgroup order
+    /// \param q the subgroup order
     void SetSubgroupOrder(const Integer &q)
         {m_q = q; ParametersChanged();}
 
+    static std::string CRYPTOPP_API StaticAlgorithmNamePrefix() {return "";}
+
 protected:
     Integer ComputeGroupOrder(const Integer &modulus) const
         {return modulus-(GetFieldType() == 1 ? 1 : -1);}
@@ -704,7 +751,7 @@ public:
 
     bool ParameterSupported(const char *name) const {return strcmp(name, Name::EncodingParameters()) == 0;}
     size_t GetSymmetricKeyLength(size_t plaintextLength) const
-        {return plaintextLength + static_cast<size_t>(MAC::DIGESTSIZE);}
+        {return plaintextLength + static_cast<size_t>(MAC::DEFAULT_KEYLENGTH);}
     size_t GetSymmetricCiphertextLength(size_t plaintextLength) const
         {return plaintextLength + static_cast<size_t>(MAC::DIGESTSIZE);}
     size_t GetMaxSymmetricPlaintextLength(size_t ciphertextLength) const

hight.h

@@ -60,7 +60,7 @@ public:
         void ProcessAndXorBlock(const byte *inBlock, const byte *xorBlock, byte *outBlock) const;
     };
 
-    /// \brief Encryption transformation
+    /// \brief Decryption transformation
     /// \details Dec provides implementation for decryption transformation.
     /// \since Crypto++ 8.0
     class CRYPTOPP_NO_VTABLE Dec : public Base

hkdf.h

@@ -39,7 +39,7 @@ public:
 	}
 
 	// KeyDerivationFunction interface
-	size_t MaxDerivedLength() const {
+	size_t MaxDerivedKeyLength() const {
 		return static_cast<size_t>(T::DIGESTSIZE) * 255;
 	}
 
@@ -60,7 +60,7 @@ public:
 	/// \param info the additional input buffer
 	/// \param infoLen the size of the info buffer, in bytes
 	/// \returns the number of iterations performed
-	/// \throws InvalidDerivedLength if <tt>derivedLen</tt> is invalid for the scheme
+	/// \throws InvalidDerivedKeyLength if <tt>derivedLen</tt> is invalid for the scheme
 	/// \details DeriveKey() provides a standard interface to derive a key from
 	///   a seed and other parameters. Each class that derives from KeyDerivationFunction
 	///   provides an overload that accepts most parameters used by the derivation function.
@@ -92,8 +92,8 @@ protected:
 template <class T>
 size_t HKDF<T>::GetValidDerivedLength(size_t keylength) const
 {
-	if (keylength > MaxDerivedLength())
+	if (keylength > MaxDerivedKeyLength())
-		return MaxDerivedLength();
+		return MaxDerivedKeyLength();
 	return keylength;
 }
 
@@ -103,7 +103,7 @@ size_t HKDF<T>::DeriveKey(byte *derived, size_t derivedLen,
 {
 	CRYPTOPP_ASSERT(secret && secretLen);
 	CRYPTOPP_ASSERT(derived && derivedLen);
-	CRYPTOPP_ASSERT(derivedLen <= MaxDerivedLength());
+	CRYPTOPP_ASSERT(derivedLen <= MaxDerivedKeyLength());
 
 	ConstByteArrayParameter p;
 	SecByteBlock salt, info;
@@ -127,9 +127,9 @@ size_t HKDF<T>::DeriveKey(byte *derived, size_t derivedLen, const byte *secret,
 {
 	CRYPTOPP_ASSERT(secret && secretLen);
 	CRYPTOPP_ASSERT(derived && derivedLen);
-	CRYPTOPP_ASSERT(derivedLen <= MaxDerivedLength());
+	CRYPTOPP_ASSERT(derivedLen <= MaxDerivedKeyLength());
 
-	ThrowIfInvalidDerivedLength(derivedLen);
+	ThrowIfInvalidDerivedKeyLength(derivedLen);
 
 	// HKDF business logic. NULL is different than empty.
 	if (salt == NULLPTR)

hmqv.h

@@ -1,5 +1,5 @@
 // hmqv.h - written and placed in the public domain by Uri Blumenthal
-//          Shamelessly based upon Jeffrey Walton's FHMQV and Wei Dai's MQV source files
+//          Shamelessly based upon Wei Dai's MQV source files
 
 #ifndef CRYPTOPP_HMQV_H
 #define CRYPTOPP_HMQV_H
@@ -29,30 +29,85 @@ public:
 
   virtual ~HMQV_Domain() {}
 
-  HMQV_Domain(bool clientRole = true): m_role(clientRole ? RoleClient : RoleServer) {}
+  /// \brief Construct a HMQV domain
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
+  HMQV_Domain(bool clientRole = true)
+    : m_role(clientRole ? RoleClient : RoleServer) {}
 
+  /// \brief Construct a HMQV domain
+  /// \param params group parameters and options
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   HMQV_Domain(const GroupParameters &params, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer), m_groupParameters(params) {}
 
+  /// \brief Construct a HMQV domain
+  /// \param bt BufferedTransformation with group parameters and options
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   HMQV_Domain(BufferedTransformation &bt, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
     {m_groupParameters.BERDecode(bt);}
 
+  /// \brief Construct a HMQV domain
+  /// \tparam T1 template parameter used as a constructor parameter
+  /// \param v1 first parameter
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details v1 is passed directly to the GROUP_PARAMETERS object.
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   template <class T1>
   HMQV_Domain(T1 v1, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
     {m_groupParameters.Initialize(v1);}
 
+  /// \brief Construct a HMQV domain
+  /// \tparam T1 template parameter used as a constructor parameter
+  /// \tparam T2 template parameter used as a constructor parameter
+  /// \param v1 first parameter
+  /// \param v2 second parameter
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details v1 and v2 are passed directly to the GROUP_PARAMETERS object.
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   template <class T1, class T2>
   HMQV_Domain(T1 v1, T2 v2, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
     {m_groupParameters.Initialize(v1, v2);}
 
+  /// \brief Construct a HMQV domain
+  /// \tparam T1 template parameter used as a constructor parameter
+  /// \tparam T2 template parameter used as a constructor parameter
+  /// \tparam T3 template parameter used as a constructor parameter
+  /// \param v1 first parameter
+  /// \param v2 second parameter
+  /// \param v3 third parameter
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details v1, v2 and v3 are passed directly to the GROUP_PARAMETERS object.
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   template <class T1, class T2, class T3>
   HMQV_Domain(T1 v1, T2 v2, T3 v3, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
     {m_groupParameters.Initialize(v1, v2, v3);}
 
+  /// \brief Construct a HMQV domain
+  /// \tparam T1 template parameter used as a constructor parameter
+  /// \tparam T2 template parameter used as a constructor parameter
+  /// \tparam T3 template parameter used as a constructor parameter
+  /// \tparam T4 template parameter used as a constructor parameter
+  /// \param v1 first parameter
+  /// \param v2 second parameter
+  /// \param v3 third parameter
+  /// \param v4 third parameter
+  /// \param clientRole flag indicating initiator or recipient
+  /// \details v1, v2, v3 and v4 are passed directly to the GROUP_PARAMETERS object.
+  /// \details <tt>clientRole = true</tt> indicates initiator, and
+  ///  <tt>clientRole = false</tt> indicates recipient or server.
   template <class T1, class T2, class T3, class T4>
   HMQV_Domain(T1 v1, T2 v2, T3 v3, T4 v4, bool clientRole = true)
     : m_role(clientRole ? RoleClient : RoleServer)
@@ -60,28 +115,62 @@ public:
 
 public:
 
+  /// \brief Retrieves the group parameters for this domain
+  /// \return the group parameters for this domain as a const reference
   const GroupParameters & GetGroupParameters() const {return m_groupParameters;}
+
+  /// \brief Retrieves the group parameters for this domain
+  /// \return the group parameters for this domain as a non-const reference
   GroupParameters & AccessGroupParameters() {return m_groupParameters;}
 
+  /// \brief Retrieves the crypto parameters for this domain
+  /// \return the crypto parameters for this domain as a non-const reference
   CryptoParameters & AccessCryptoParameters() {return AccessAbstractGroupParameters();}
 
-  /// return length of agreed value produced
+  /// \brief Provides the size of the agreed value
-  unsigned int AgreedValueLength() const {return GetAbstractGroupParameters().GetEncodedElementSize(false);}
+  /// \return size of agreed value produced in this domain
-  /// return length of static private keys in this domain
+  /// \details The length is calculated using <tt>GetEncodedElementSize(false)</tt>,
-  unsigned int StaticPrivateKeyLength() const {return GetAbstractGroupParameters().GetSubgroupOrder().ByteCount();}
+  ///  which means the element is encoded in a non-reversible format. A
-  /// return length of static public keys in this domain
+  ///  non-reversible format means its a raw byte array, and it lacks presentation
-  unsigned int StaticPublicKeyLength() const{return GetAbstractGroupParameters().GetEncodedElementSize(true);}
+  ///  format like an ASN.1 BIT_STRING or OCTET_STRING.
+  unsigned int AgreedValueLength() const
+    {return GetAbstractGroupParameters().GetEncodedElementSize(false);}
 
-  /// generate static private key
+  /// \brief Provides the size of the static private key
-  /*! \pre size of privateKey == PrivateStaticKeyLength() */
+  /// \return size of static private keys in this domain
+  /// \details The length is calculated using the byte count of the subgroup order.
+  unsigned int StaticPrivateKeyLength() const
+    {return GetAbstractGroupParameters().GetSubgroupOrder().ByteCount();}
+
+  /// \brief Provides the size of the static public key
+  /// \return size of static public keys in this domain
+  /// \details The length is calculated using <tt>GetEncodedElementSize(true)</tt>,
+  ///  which means the element is encoded in a reversible format. A reversible
+  ///  format means it has a presentation format, and its an ANS.1 encoded element
+  ///  or point.
+  unsigned int StaticPublicKeyLength() const
+    {return GetAbstractGroupParameters().GetEncodedElementSize(true);}
+
+  /// \brief Generate static private key in this domain
+  /// \param rng a RandomNumberGenerator derived class
+  /// \param privateKey a byte buffer for the generated private key in this domain
+  /// \details The private key is a random scalar used as an exponent in the range
+  ///  <tt>[1,MaxExponent()]</tt>.
+  /// \pre <tt>COUNTOF(privateKey) == PrivateStaticKeyLength()</tt>
   void GenerateStaticPrivateKey(RandomNumberGenerator &rng, byte *privateKey) const
   {
     Integer x(rng, Integer::One(), GetAbstractGroupParameters().GetMaxExponent());
     x.Encode(privateKey, StaticPrivateKeyLength());
   }
 
-  /// generate static public key
+  /// \brief Generate a static public key from a private key in this domain
-  /*! \pre size of publicKey == PublicStaticKeyLength() */
+  /// \param rng a RandomNumberGenerator derived class
+  /// \param privateKey a byte buffer with the previously generated private key
+  /// \param publicKey a byte buffer for the generated public key in this domain
+  /// \details The public key is an element or point on the curve, and its stored
+  ///  in a revrsible format. A reversible format means it has a presentation
+  ///  format, and its an ANS.1 encoded element or point.
+  /// \pre <tt>COUNTOF(publicKey) == PublicStaticKeyLength()</tt>
   void GenerateStaticPublicKey(RandomNumberGenerator &rng, const byte *privateKey, byte *publicKey) const
   {
     CRYPTOPP_UNUSED(rng);
@@ -91,10 +180,22 @@ public:
     params.EncodeElement(true, y, publicKey);
   }
 
+  /// \brief Provides the size of the ephemeral private key
+  /// \return size of ephemeral private keys in this domain
+  /// \details An ephemeral private key is a private key and public key.
+  ///  The serialized size is different than a static private key.
   unsigned int EphemeralPrivateKeyLength() const {return StaticPrivateKeyLength() + StaticPublicKeyLength();}
+
+  /// \brief Provides the size of the ephemeral public key
+  /// \return size of ephemeral public keys in this domain
+  /// \details An ephemeral public key is a public key.
+  ///  The serialized size is the same as a static public key.
   unsigned int EphemeralPublicKeyLength() const{return StaticPublicKeyLength();}
 
-  /// return length of ephemeral private keys in this domain
+  /// \brief Generate ephemeral private key in this domain
+  /// \param rng a RandomNumberGenerator derived class
+  /// \param privateKey a byte buffer for the generated private key in this domain
+  /// \pre <tt>COUNTOF(privateKey) == EphemeralPrivateKeyLength()</tt>
   void GenerateEphemeralPrivateKey(RandomNumberGenerator &rng, byte *privateKey) const
   {
     const DL_GroupParameters<Element> &params = GetAbstractGroupParameters();
@@ -104,28 +205,46 @@ public:
     params.EncodeElement(true, y, privateKey+StaticPrivateKeyLength());
   }
 
-  /// return length of ephemeral public keys in this domain
+  /// \brief Generate ephemeral public key from a private key in this domain
+  /// \param rng a RandomNumberGenerator derived class
+  /// \param privateKey a byte buffer with the previously generated private key
+  /// \param publicKey a byte buffer for the generated public key in this domain
+  /// \pre <tt>COUNTOF(publicKey) == EphemeralPublicKeyLength()</tt>
   void GenerateEphemeralPublicKey(RandomNumberGenerator &rng, const byte *privateKey, byte *publicKey) const
   {
     CRYPTOPP_UNUSED(rng);
     memcpy(publicKey, privateKey+StaticPrivateKeyLength(), EphemeralPublicKeyLength());
   }
 
-  /// derive agreed value from your private keys and couterparty's public keys, return false in case of failure
+  /// \brief Derive agreed value or shared secret
-  /*! \note The ephemeral public key will always be validated.
+  /// \param agreedValue the shared secret
-  If you have previously validated the static public key, use validateStaticOtherPublicKey=false to save time.
+  /// \param staticPrivateKey your long term private key
-  \pre size of agreedValue == AgreedValueLength()
+  /// \param ephemeralPrivateKey your ephemeral private key
-  \pre length of staticPrivateKey == StaticPrivateKeyLength()
+  /// \param staticOtherPublicKey couterparty's long term public key
-  \pre length of ephemeralPrivateKey == EphemeralPrivateKeyLength()
+  /// \param ephemeralOtherPublicKey couterparty's ephemeral public key
-  \pre length of staticOtherPublicKey == StaticPublicKeyLength()
+  /// \param validateStaticOtherPublicKey flag indicating validation
-  \pre length of ephemeralOtherPublicKey == EphemeralPublicKeyLength()
+  /// \return true upon success, false in case of failure
-  */
+  /// \details Agree() performs the authenticated key agreement. Agree()
+  ///  derives a shared secret from your private keys and couterparty's
+  ///  public keys. Each instance or run of the protocol should use a new
+  ///  ephemeral key pair.
+  /// \details The other's ephemeral public key will always be validated at
+  ///  Level 1 to ensure it is a point on the curve.
+  ///  <tt>validateStaticOtherPublicKey</tt> determines how thoroughly other's
+  ///  static public key is validated. If you have previously validated the
+  ///  couterparty's static public key, then use
+  ///  <tt>validateStaticOtherPublicKey=false</tt> to save time.
+  /// \pre <tt>COUNTOF(agreedValue) == AgreedValueLength()
+  /// \pre <tt>COUNTOF(staticPrivateKey) == StaticPrivateKeyLength()</tt>
+  /// \pre <tt>COUNTOF(ephemeralPrivateKey) == EphemeralPrivateKeyLength()</tt>
+  /// \pre <tt>COUNTOF(staticOtherPublicKey) == StaticPublicKeyLength()</tt>
+  /// \pre <tt>COUNTOF(ephemeralOtherPublicKey) == EphemeralPublicKeyLength()</tt>
   bool Agree(byte *agreedValue,
     const byte *staticPrivateKey, const byte *ephemeralPrivateKey,
     const byte *staticOtherPublicKey, const byte *ephemeralOtherPublicKey,
     bool validateStaticOtherPublicKey=true) const
   {
-    byte *XX = NULLPTR, *YY = NULLPTR, *AA = NULLPTR, *BB = NULLPTR;
+    const byte *XX = NULLPTR, *YY = NULLPTR, *AA = NULLPTR, *BB = NULLPTR;
     size_t xxs = 0, yys = 0, aas = 0, bbs = 0;
 
     // Depending on the role, this will hold either A's or B's static
@@ -143,63 +262,45 @@ public:
         Element B = params.ExponentiateBase(b);
         params.EncodeElement(true, B, tt);
 
-        XX = const_cast<byte*>(ephemeralOtherPublicKey);
+        XX = ephemeralOtherPublicKey;
         xxs = EphemeralPublicKeyLength();
-        YY = const_cast<byte*>(ephemeralPrivateKey) + StaticPrivateKeyLength();
+        YY = ephemeralPrivateKey + StaticPrivateKeyLength();
         yys = EphemeralPublicKeyLength();
-        AA = const_cast<byte*>(staticOtherPublicKey);
+        AA = staticOtherPublicKey;
         aas = StaticPublicKeyLength();
         BB = tt.BytePtr();
         bbs = tt.SizeInBytes();
       }
-      else if(m_role == RoleClient)
+      else
       {
         Integer a(staticPrivateKey, StaticPrivateKeyLength());
         Element A = params.ExponentiateBase(a);
         params.EncodeElement(true, A, tt);
 
-        XX = const_cast<byte*>(ephemeralPrivateKey) + StaticPrivateKeyLength();
+        XX = ephemeralPrivateKey + StaticPrivateKeyLength();
         xxs = EphemeralPublicKeyLength();
-        YY = const_cast<byte*>(ephemeralOtherPublicKey);
+        YY = ephemeralOtherPublicKey;
         yys = EphemeralPublicKeyLength();
         AA = tt.BytePtr();
         aas = tt.SizeInBytes();
-        BB = const_cast<byte*>(staticOtherPublicKey);
+        BB = staticOtherPublicKey;
         bbs = StaticPublicKeyLength();
       }
-      else
-      {
-        CRYPTOPP_ASSERT(0);
-        return false;
-      }
 
-      // DecodeElement calls ValidateElement at level 1. Level 1 only calls
+      Element VV1 = params.DecodeElement(staticOtherPublicKey, validateStaticOtherPublicKey);
-      // VerifyPoint to ensure the element is in G*. If the other's PublicKey is
+      Element VV2 = params.DecodeElement(ephemeralOtherPublicKey, true);
-      // requested to be validated, we manually call ValidateElement at level 3.
-      Element VV1 = params.DecodeElement(staticOtherPublicKey, false);
-      if(!params.ValidateElement(validateStaticOtherPublicKey ? 3 : 1, VV1, NULLPTR))
-        return false;
 
-      // DecodeElement calls ValidateElement at level 1. Level 1 only calls
-      // VerifyPoint to ensure the element is in G*. Crank it up.
-      Element VV2 = params.DecodeElement(ephemeralOtherPublicKey, false);
-      if(!params.ValidateElement(3, VV2, NULLPTR))
-        return false;
-
-      // const Integer& p = params.GetGroupOrder(); // not used, remove later
       const Integer& q = params.GetSubgroupOrder();
       const unsigned int len /*bytes*/ = (((q.BitCount()+1)/2 +7)/8);
-
-      Integer d, e;
       SecByteBlock dd(len), ee(len);
 
       // Compute $d = \hat{H}(X, \hat{B})$
       Hash(NULLPTR, XX, xxs, BB, bbs, dd.BytePtr(), dd.SizeInBytes());
-      d.Decode(dd.BytePtr(), dd.SizeInBytes());
+      Integer d(dd.BytePtr(), dd.SizeInBytes());
 
       // Compute $e = \hat{H}(Y, \hat{A})$
       Hash(NULLPTR, YY, yys, AA, aas, ee.BytePtr(), ee.SizeInBytes());
-      e.Decode(ee.BytePtr(), ee.SizeInBytes());
+      Integer e(ee.BytePtr(), ee.SizeInBytes());
 
       Element sigma;
       if(m_role == RoleServer)
@@ -226,16 +327,17 @@ public:
         Element B = params.DecodeElement(BB, false);
         Element Y = params.DecodeElement(YY, false);
 
-        Element t1 = params.ExponentiateElement(B, e);
+        Element t3 = params.ExponentiateElement(B, e);
-        Element t2 = m_groupParameters.MultiplyElements(Y, t1);
+        Element t4 = m_groupParameters.MultiplyElements(Y, t3);
 
         // $\sigma_A}=(Y \cdot B^{e})^{s_A}
-        sigma = params.ExponentiateElement(t2, s_A);
+        sigma = params.ExponentiateElement(t4, s_A);
       }
       Hash(&sigma, NULLPTR, 0, NULLPTR, 0, agreedValue, AgreedValueLength());
     }
     catch (DL_BadElement &)
     {
+      CRYPTOPP_ASSERT(0);
       return false;
     }
     return true;
@@ -259,9 +361,11 @@ protected:
       if (e1len != 0 || s1len != 0) {
         CRYPTOPP_ASSERT(0);
       }
-      Integer x = GetAbstractGroupParameters().ConvertElementToInteger(*sigma);
+      //Integer x = GetAbstractGroupParameters().ConvertElementToInteger(*sigma);
-      SecByteBlock sbb(x.MinEncodedSize());
+      //SecByteBlock sbb(x.MinEncodedSize());
-      x.Encode(sbb.BytePtr(), sbb.SizeInBytes());
+      //x.Encode(sbb.BytePtr(), sbb.SizeInBytes());
+      SecByteBlock sbb(GetAbstractGroupParameters().GetEncodedElementSize(false));
+      GetAbstractGroupParameters().EncodeElement(false, *sigma, sbb);
       hash.Update(sbb.BytePtr(), sbb.SizeInBytes());
     } else {
       if (e1len == 0 || s1len == 0) {
@@ -292,8 +396,10 @@ private:
   // The paper uses Initiator and Recipient - make it classical.
   enum KeyAgreementRole { RoleServer = 1, RoleClient };
 
-  DL_GroupParameters<Element> & AccessAbstractGroupParameters() {return m_groupParameters;}
+  DL_GroupParameters<Element> & AccessAbstractGroupParameters()
-  const DL_GroupParameters<Element> & GetAbstractGroupParameters() const{return m_groupParameters;}
+    {return m_groupParameters;}
+  const DL_GroupParameters<Element> & GetAbstractGroupParameters() const
+    {return m_groupParameters;}
 
   GroupParameters m_groupParameters;
   KeyAgreementRole m_role;
@@ -302,7 +408,7 @@ private:
 /// \brief Hashed Menezes-Qu-Vanstone in GF(p)
 /// \details This implementation follows Hugo Krawczyk's <a href="http://eprint.iacr.org/2005/176">HMQV: A High-Performance
 ///   Secure Diffie-Hellman Protocol</a>. Note: this implements HMQV only. HMQV-C with Key Confirmation is not provided.
-/// \sa HMQV, MQV_Domain, FHMQV_Domain, AuthenticatedKeyAgreementDomain
+/// \sa HMQV, HMQV_Domain, FHMQV_Domain, AuthenticatedKeyAgreementDomain
 /// \since Crypto++ 5.6.4
 typedef HMQV_Domain<DL_GroupParameters_GFP_DefaultSafePrime> HMQV;
 

integer.cpp

@@ -18,13 +18,13 @@
 // For Integer::Zero(), Integer::One() and Integer::Two(), we use one of three
 //  strategies. First, if initialization priorities are available then we use
 //  them. Initialization priorities are init_priority() on Linux and init_seg()
-//  on Windows. AIX, OS X and several other platforms lack them. Initialization
+//  on Windows. OS X and several other platforms lack them. Initialization
 //  priorities are platform specific but they are also the most trouble free
 //  with determisitic destruction.
 // Second, if C++11 dynamic initialization is available, then we use it. After
-//  the std::call_once fiasco we dropped the priority dynamic initialization
+//  the std::call_once fiasco we moved to dynamic initialization to avoid
-//  to avoid unknown troubles platforms that are tested less frequently. In
+//  unknown troubles platforms that are tested less frequently. In addition
-//  addition Microsoft platforms mostly do not provide dynamic initialization.
+//  Microsoft platforms mostly do not provide dynamic initialization.
 //  The MSDN docs claim they do but they don't in practice because we need
 //  Visual Studio 2017 and Windows 10 or above.
 // Third, we fall back to Wei's original code of a Singleton. Wei's original
@@ -45,6 +45,12 @@
 //  thousands of times during the life of a program. Each load produces a
 //  memory leak and they can add up quickly. If they library is being used in
 //  Java or .Net then Singleton must be avoided at all costs.
+//
+// The code below has a path cut-in for BMI2 using mulx and adcx instructions.
+//  There was a modest speedup of approximately 0.03 ms in public key Integer
+//  operations. We had to disable BMI2 for the moment because some OS X machines
+//  were advertising BMI/BMI2 support but caused SIGILL's at runtime. Also see
+//  https://github.com/weidai11/cryptopp/issues/850.
 
 #include "pch.h"
 #include "config.h"
@@ -77,7 +83,8 @@
 	#include <c_asm.h>
 #endif
 
-// "Error: The operand ___LKDB cannot be assigned to", http://github.com/weidai11/cryptopp/issues/188
+// "Error: The operand ___LKDB cannot be assigned to",
+// http://github.com/weidai11/cryptopp/issues/188
 #if (__SUNPRO_CC >= 0x5130)
 # define MAYBE_CONST
 # define MAYBE_UNCONST_CAST(x) const_cast<word*>(x)
@@ -211,6 +218,13 @@ static word AtomicInverseModPower2(word A)
 		#if defined(__SUNPRO_CC) && __SUNPRO_CC < 0x5100
 			// Sun Studio's gcc-style inline assembly is heavily bugged as of version 5.9 Patch 124864-09 2008/12/16, but this one works
 			#define MultiplyWordsLoHi(p0, p1, a, b)		asm ("mulq %3" : "=a"(p0), "=d"(p1) : "a"(a), "r"(b) : "cc");
+		#elif defined(__BMI2__) && 0
+			#define MultiplyWordsLoHi(p0, p1, a, b)		asm ("mulxq %3, %0, %1" : "=r"(p0), "=r"(p1) : "d"(a), "r"(b));
+			#define MulAcc(c, d, a, b)		asm ("mulxq %6, %3, %4; addq %3, %0; adcxq %4, %1; adcxq %7, %2;" : "+&r"(c), "+&r"(d##0), "+&r"(d##1), "=&r"(p0), "=&r"(p1) : "d"(a), "r"(b), "r"(W64LIT(0)) : "cc");
+			#define Double3Words(c, d)		asm ("addq %0, %0; adcxq %1, %1; adcxq %2, %2;" : "+r"(c), "+r"(d##0), "+r"(d##1) : : "cc");
+			#define Acc2WordsBy1(a, b)		asm ("addq %2, %0; adcxq %3, %1;" : "+&r"(a##0), "+r"(a##1) : "r"(b), "r"(W64LIT(0)) : "cc");
+			#define Acc2WordsBy2(a, b)		asm ("addq %2, %0; adcxq %3, %1;" : "+r"(a##0), "+r"(a##1) : "r"(b##0), "r"(b##1) : "cc");
+			#define Acc3WordsBy2(c, d, e)	asm ("addq %5, %0; adcxq %6, %1; adcxq %7, %2;" : "+r"(c), "=&r"(e##0), "=&r"(e##1) : "1"(d##0), "2"(d##1), "r"(e##0), "r"(e##1), "r"(W64LIT(0)) : "cc");
 		#else
 			#define MultiplyWordsLoHi(p0, p1, a, b)		asm ("mulq %3" : "=a"(p0), "=d"(p1) : "a"(a), "g"(b) : "cc");
 			#define MulAcc(c, d, a, b)		asm ("mulq %6; addq %3, %0; adcq %4, %1; adcq $0, %2;" : "+r"(c), "+r"(d##0), "+r"(d##1), "=a"(p0), "=d"(p1) : "a"(a), "g"(b) : "cc");
@@ -3537,9 +3551,9 @@ class KDF2_RNG : public RandomNumberGenerator
 {
 public:
 	KDF2_RNG(const byte *seed, size_t seedSize)
-		: m_counter(0), m_counterAndSeed(seedSize + 4)
+		: m_counter(0), m_counterAndSeed(ClampSize(seedSize) + 4)
 	{
-		memcpy(m_counterAndSeed + 4, seed, seedSize);
+		memcpy(m_counterAndSeed + 4, seed, ClampSize(seedSize));
 	}
 
 	void GenerateBlock(byte *output, size_t size)
@@ -3550,6 +3564,15 @@ public:
 		P1363_KDF2<SHA1>::DeriveKey(output, size, m_counterAndSeed, m_counterAndSeed.size(), NULLPTR, 0);
 	}
 
+	// UBsan finding, -Wstringop-overflow
+	inline size_t ClampSize(size_t req) const
+	{
+		// Clamp at 16 MB
+		if (req > 16U*1024*1024)
+			return 16U*1024*1024;
+		return req;
+	}
+
 private:
 	word32 m_counter;
 	SecByteBlock m_counterAndSeed;

integer.h

@@ -306,13 +306,16 @@ public:
 		signed long ConvertToLong() const;
 
 		/// \brief Determines the number of bits required to represent the Integer
-		/// \returns number of significant bits = floor(log2(abs(*this))) + 1
+		/// \returns number of significant bits
+		/// \details BitCount is calculated as <tt>floor(log2(abs(*this))) + 1</tt>.
 		unsigned int BitCount() const;
 		/// \brief Determines the number of bytes required to represent the Integer
-		/// \returns number of significant bytes = ceiling(BitCount()/8)
+		/// \returns number of significant bytes
+		/// \details ByteCount is calculated as <tt>ceiling(BitCount()/8)</tt>.
 		unsigned int ByteCount() const;
 		/// \brief Determines the number of words required to represent the Integer
-		/// \returns number of significant words = ceiling(ByteCount()/sizeof(word))
+		/// \returns number of significant words
+		/// \details WordCount is calculated as <tt>ceiling(ByteCount()/sizeof(word))</tt>.
 		unsigned int WordCount() const;
 
 		/// \brief Provides the i-th bit of the Integer
@@ -354,29 +357,48 @@ public:
 	/// \name MANIPULATORS
 	//@{
 		/// \brief Assignment
+		/// \param t the other Integer
+		/// \returns the result of assignment
 		Integer&  operator=(const Integer& t);
-
 		/// \brief Addition Assignment
+		/// \param t the other Integer
+		/// \returns the result of <tt>*this + t</tt>
 		Integer&  operator+=(const Integer& t);
 		/// \brief Subtraction Assignment
+		/// \param t the other Integer
+		/// \returns the result of <tt>*this - t</tt>
 		Integer&  operator-=(const Integer& t);
 		/// \brief Multiplication Assignment
+		/// \param t the other Integer
+		/// \returns the result of <tt>*this * t</tt>
 		/// \sa a_times_b_mod_c() and a_exp_b_mod_c()
 		Integer&  operator*=(const Integer& t)	{return *this = Times(t);}
 		/// \brief Division Assignment
+		/// \param t the other Integer
+		/// \returns the result of <tt>*this / t</tt>
 		Integer&  operator/=(const Integer& t)	{return *this = DividedBy(t);}
 		/// \brief Remainder Assignment
+		/// \param t the other Integer
+		/// \returns the result of <tt>*this % t</tt>
 		/// \sa a_times_b_mod_c() and a_exp_b_mod_c()
 		Integer&  operator%=(const Integer& t)	{return *this = Modulo(t);}
 		/// \brief Division Assignment
+		/// \param t the other word
+		/// \returns the result of <tt>*this / t</tt>
 		Integer&  operator/=(word t)  {return *this = DividedBy(t);}
 		/// \brief Remainder Assignment
+		/// \param t the other word
+		/// \returns the result of <tt>*this % t</tt>
 		/// \sa a_times_b_mod_c() and a_exp_b_mod_c()
 		Integer&  operator%=(word t)  {return *this = Integer(POSITIVE, 0, Modulo(t));}
 
 		/// \brief Left-shift Assignment
+		/// \param n number of bits to shift
+		/// \returns reference to this Integer
 		Integer&  operator<<=(size_t n);
 		/// \brief Right-shift Assignment
+		/// \param n number of bits to shift
+		/// \returns reference to this Integer
 		Integer&  operator>>=(size_t n);
 
 		/// \brief Bitwise AND Assignment

keccak.h

@@ -94,8 +94,6 @@ private:
 #if !defined(__BORLANDC__)
     // ensure there was no underflow in the math
     CRYPTOPP_COMPILE_ASSERT(BLOCKSIZE < 200);
-    // this is a general expectation by HMAC
-    CRYPTOPP_COMPILE_ASSERT((int)BLOCKSIZE > (int)DIGESTSIZE);
 #endif
 };
 

keccak_simd.cpp

@@ -12,8 +12,8 @@
 // KeccakF1600x2_SSE is ParallelHash128. The SSE2 ParallelHash128
 // implementation was extracted from XKCP using the following command.
 //
-// gcc -I lib/common -I lib/low/KeccakP-1600/Optimized \
+// gcc -I lib/common -I lib/low/KeccakP-1600/Optimized
-//   -I lib/low/KeccakP-1600-times2/SIMD128/SSE2ufull \
+//   -I lib/low/KeccakP-1600-times2/SIMD128/SSE2ufull
 //   lib/low/KeccakP-1600-times2/SIMD128/KeccakP-1600-times2-SIMD128.c -E
 
 #include "pch.h"
@@ -23,7 +23,14 @@
 
 #if (CRYPTOPP_SSSE3_AVAILABLE)
 # include <emmintrin.h>
-# include <immintrin.h>
+# include <tmmintrin.h>
+#endif
+
+#if defined(__XOP__)
+# include <ammintrin.h>
+# if defined(__GNUC__)
+#  include <x86intrin.h>
+# endif
 #endif
 
 // Squash MS LNK4221 and libtool warnings
@@ -39,16 +46,17 @@ extern void KeccakF1600x2_SSE(word64 *state);
 // The F1600 round constants
 extern const word64 KeccakF1600Constants[24];
 
-const word64 rho8[2] = {0x0605040302010007, 0x0E0D0C0B0A09080F};
+CRYPTOPP_ALIGN_DATA(16)
-const word64 rho56[2] = {0x0007060504030201, 0x080F0E0D0C0B0A09};
+const word64
+rho8[2] = {W64LIT(0x0605040302010007), W64LIT(0x0E0D0C0B0A09080F)};
 
-#define V128 __m128i
+CRYPTOPP_ALIGN_DATA(16)
-#define CV128 const __m128i
+const word64
+rho56[2] = {W64LIT(0x0007060504030201), W64LIT(0x080F0E0D0C0B0A09)};
 
-#define CONST128(a)         _mm_load_si128((CV128 *)&(a))
+// Clang intrinsic casts, http://bugs.llvm.org/show_bug.cgi?id=20670
-#define XOREQ128(a, b)      a = _mm_xor_si128((a), (b))
+#define M128_CAST(x) ((__m128i *)(void *)(x))
-#define UNPACKL(a, b)       _mm_unpacklo_epi64((a), (b))
+#define CONST_M128_CAST(x) ((const __m128i *)(const void *)(x))
-#define UNPACKH(a, b)       _mm_unpackhi_epi64((a), (b))
 
 #if defined(__XOP__)
 # define ROL64in128(a, o)    _mm_roti_epi64((a), (o))
@@ -56,8 +64,8 @@ const word64 rho56[2] = {0x0007060504030201, 0x080F0E0D0C0B0A09};
 # define ROL64in128_56(a)    ROL64in128((a), 56)
 #else
 # define ROL64in128(a, o)    _mm_or_si128(_mm_slli_epi64((a), (o)), _mm_srli_epi64(a, 64-(o)))
-# define ROL64in128_8(a)     _mm_shuffle_epi8((a), CONST128(rho8))
+# define ROL64in128_8(a)     _mm_shuffle_epi8((a), _mm_load_si128(CONST_M128_CAST(rho8)))
-# define ROL64in128_56(a)    _mm_shuffle_epi8((a), CONST128(rho56))
+# define ROL64in128_56(a)    _mm_shuffle_epi8((a), _mm_load_si128(CONST_M128_CAST(rho56)))
 #endif
 
 // Damn Visual Studio is missing too many intrinsics...
@@ -74,51 +82,50 @@ inline __m128i SPLAT64(const word64 a)
 // The Keccak ParallelHash128 core function
 void KeccakF1600x2_SSE(word64 *state)
 {
-    V128 *statesAsLanes = (V128 *)state;
+    __m128i Aba, Abe, Abi, Abo, Abu;
+    __m128i Aga, Age, Agi, Ago, Agu;
+    __m128i Aka, Ake, Aki, Ako, Aku;
+    __m128i Ama, Ame, Ami, Amo, Amu;
+    __m128i Asa, Ase, Asi, Aso, Asu;
+    __m128i Bba, Bbe, Bbi, Bbo, Bbu;
+    __m128i Bga, Bge, Bgi, Bgo, Bgu;
+    __m128i Bka, Bke, Bki, Bko, Bku;
+    __m128i Bma, Bme, Bmi, Bmo, Bmu;
+    __m128i Bsa, Bse, Bsi, Bso, Bsu;
+    __m128i Ca, Ce, Ci, Co, Cu;
+    __m128i Da, De, Di, Do, Du;
+    __m128i Eba, Ebe, Ebi, Ebo, Ebu;
+    __m128i Ega, Ege, Egi, Ego, Egu;
+    __m128i Eka, Eke, Eki, Eko, Eku;
+    __m128i Ema, Eme, Emi, Emo, Emu;
+    __m128i Esa, Ese, Esi, Eso, Esu;
 
-    V128 Aba, Abe, Abi, Abo, Abu;
+    __m128i* lanes = reinterpret_cast<__m128i*>(state);
-    V128 Aga, Age, Agi, Ago, Agu;
+    Aba = _mm_loadu_si128(CONST_M128_CAST(lanes+ 0));
-    V128 Aka, Ake, Aki, Ako, Aku;
+    Abe = _mm_loadu_si128(CONST_M128_CAST(lanes+ 1));
-    V128 Ama, Ame, Ami, Amo, Amu;
+    Abi = _mm_loadu_si128(CONST_M128_CAST(lanes+ 2));
-    V128 Asa, Ase, Asi, Aso, Asu;
+    Abo = _mm_loadu_si128(CONST_M128_CAST(lanes+ 3));
-    V128 Bba, Bbe, Bbi, Bbo, Bbu;
+    Abu = _mm_loadu_si128(CONST_M128_CAST(lanes+ 4));
-    V128 Bga, Bge, Bgi, Bgo, Bgu;
+    Aga = _mm_loadu_si128(CONST_M128_CAST(lanes+ 5));
-    V128 Bka, Bke, Bki, Bko, Bku;
+    Age = _mm_loadu_si128(CONST_M128_CAST(lanes+ 6));
-    V128 Bma, Bme, Bmi, Bmo, Bmu;
+    Agi = _mm_loadu_si128(CONST_M128_CAST(lanes+ 7));
-    V128 Bsa, Bse, Bsi, Bso, Bsu;
+    Ago = _mm_loadu_si128(CONST_M128_CAST(lanes+ 8));
-    V128 Ca, Ce, Ci, Co, Cu;
+    Agu = _mm_loadu_si128(CONST_M128_CAST(lanes+ 9));
-    V128 Da, De, Di, Do, Du;
+    Aka = _mm_loadu_si128(CONST_M128_CAST(lanes+10));
-    V128 Eba, Ebe, Ebi, Ebo, Ebu;
+    Ake = _mm_loadu_si128(CONST_M128_CAST(lanes+11));
-    V128 Ega, Ege, Egi, Ego, Egu;
+    Aki = _mm_loadu_si128(CONST_M128_CAST(lanes+12));
-    V128 Eka, Eke, Eki, Eko, Eku;
+    Ako = _mm_loadu_si128(CONST_M128_CAST(lanes+13));
-    V128 Ema, Eme, Emi, Emo, Emu;
+    Aku = _mm_loadu_si128(CONST_M128_CAST(lanes+14));
-    V128 Esa, Ese, Esi, Eso, Esu;
+    Ama = _mm_loadu_si128(CONST_M128_CAST(lanes+15));
-
+    Ame = _mm_loadu_si128(CONST_M128_CAST(lanes+16));
-    Aba = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 0]));
+    Ami = _mm_loadu_si128(CONST_M128_CAST(lanes+17));
-    Abe = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 1]));
+    Amo = _mm_loadu_si128(CONST_M128_CAST(lanes+18));
-    Abi = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 2]));
+    Amu = _mm_loadu_si128(CONST_M128_CAST(lanes+19));
-    Abo = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 3]));
+    Asa = _mm_loadu_si128(CONST_M128_CAST(lanes+20));
-    Abu = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 4]));
+    Ase = _mm_loadu_si128(CONST_M128_CAST(lanes+21));
-    Aga = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 5]));
+    Asi = _mm_loadu_si128(CONST_M128_CAST(lanes+22));
-    Age = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 6]));
+    Aso = _mm_loadu_si128(CONST_M128_CAST(lanes+23));
-    Agi = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 7]));
+    Asu = _mm_loadu_si128(CONST_M128_CAST(lanes+24));
-    Ago = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 8]));
-    Agu = _mm_loadu_si128((CV128 *)&(statesAsLanes[ 9]));
-    Aka = _mm_loadu_si128((CV128 *)&(statesAsLanes[10]));
-    Ake = _mm_loadu_si128((CV128 *)&(statesAsLanes[11]));
-    Aki = _mm_loadu_si128((CV128 *)&(statesAsLanes[12]));
-    Ako = _mm_loadu_si128((CV128 *)&(statesAsLanes[13]));
-    Aku = _mm_loadu_si128((CV128 *)&(statesAsLanes[14]));
-    Ama = _mm_loadu_si128((CV128 *)&(statesAsLanes[15]));
-    Ame = _mm_loadu_si128((CV128 *)&(statesAsLanes[16]));
-    Ami = _mm_loadu_si128((CV128 *)&(statesAsLanes[17]));
-    Amo = _mm_loadu_si128((CV128 *)&(statesAsLanes[18]));
-    Amu = _mm_loadu_si128((CV128 *)&(statesAsLanes[19]));
-    Asa = _mm_loadu_si128((CV128 *)&(statesAsLanes[20]));
-    Ase = _mm_loadu_si128((CV128 *)&(statesAsLanes[21]));
-    Asi = _mm_loadu_si128((CV128 *)&(statesAsLanes[22]));
-    Aso = _mm_loadu_si128((CV128 *)&(statesAsLanes[23]));
-    Asu = _mm_loadu_si128((CV128 *)&(statesAsLanes[24]));
 
     Ca = _mm_xor_si128(Aba, _mm_xor_si128(Aga, _mm_xor_si128(Aka, _mm_xor_si128(Ama, Asa))));
     Ce = _mm_xor_si128(Abe, _mm_xor_si128(Age, _mm_xor_si128(Ake, _mm_xor_si128(Ame, Ase))));
@@ -2646,31 +2653,31 @@ void KeccakF1600x2_SSE(word64 *state)
     Aso = _mm_xor_si128(Bso, _mm_andnot_si128(Bsu, Bsa));
     Asu = _mm_xor_si128(Bsu, _mm_andnot_si128(Bsa, Bse));
 
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 0]), Aba);
+    _mm_storeu_si128(M128_CAST(lanes+ 0), Aba);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 1]), Abe);
+    _mm_storeu_si128(M128_CAST(lanes+ 1), Abe);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 2]), Abi);
+    _mm_storeu_si128(M128_CAST(lanes+ 2), Abi);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 3]), Abo);
+    _mm_storeu_si128(M128_CAST(lanes+ 3), Abo);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 4]), Abu);
+    _mm_storeu_si128(M128_CAST(lanes+ 4), Abu);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 5]), Aga);
+    _mm_storeu_si128(M128_CAST(lanes+ 5), Aga);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 6]), Age);
+    _mm_storeu_si128(M128_CAST(lanes+ 6), Age);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 7]), Agi);
+    _mm_storeu_si128(M128_CAST(lanes+ 7), Agi);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 8]), Ago);
+    _mm_storeu_si128(M128_CAST(lanes+ 8), Ago);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[ 9]), Agu);
+    _mm_storeu_si128(M128_CAST(lanes+ 9), Agu);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[10]), Aka);
+    _mm_storeu_si128(M128_CAST(lanes+10), Aka);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[11]), Ake);
+    _mm_storeu_si128(M128_CAST(lanes+11), Ake);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[12]), Aki);
+    _mm_storeu_si128(M128_CAST(lanes+12), Aki);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[13]), Ako);
+    _mm_storeu_si128(M128_CAST(lanes+13), Ako);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[14]), Aku);
+    _mm_storeu_si128(M128_CAST(lanes+14), Aku);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[15]), Ama);
+    _mm_storeu_si128(M128_CAST(lanes+15), Ama);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[16]), Ame);
+    _mm_storeu_si128(M128_CAST(lanes+16), Ame);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[17]), Ami);
+    _mm_storeu_si128(M128_CAST(lanes+17), Ami);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[18]), Amo);
+    _mm_storeu_si128(M128_CAST(lanes+18), Amo);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[19]), Amu);
+    _mm_storeu_si128(M128_CAST(lanes+19), Amu);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[20]), Asa);
+    _mm_storeu_si128(M128_CAST(lanes+20), Asa);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[21]), Ase);
+    _mm_storeu_si128(M128_CAST(lanes+21), Ase);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[22]), Asi);
+    _mm_storeu_si128(M128_CAST(lanes+22), Asi);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[23]), Aso);
+    _mm_storeu_si128(M128_CAST(lanes+23), Aso);
-    _mm_storeu_si128((V128 *)&(statesAsLanes[24]), Asu);
+    _mm_storeu_si128(M128_CAST(lanes+24), Asu);
 }
 
 #endif

lea.h

@@ -82,7 +82,7 @@ public:
 #endif
     };
 
-    /// \brief Encryption transformation
+    /// \brief Decryption transformation
     /// \details Dec provides implementation for decryption transformation. All key and block
     ///   sizes are supported.
     /// \since Crypto++ 8.0

lea_simd.cpp

@@ -24,6 +24,9 @@
 
 #if defined(__XOP__)
 # include <ammintrin.h>
+# if defined(__GNUC__)
+#  include <x86intrin.h>
+# endif
 #endif
 
 #if defined(__AVX512F__)

misc.cpp

@@ -101,12 +101,12 @@ bool VerifyBufsEqual(const byte *buf, const byte *mask, size_t count)
 {
 	CRYPTOPP_ASSERT(buf != NULLPTR);
 	CRYPTOPP_ASSERT(mask != NULLPTR);
-	CRYPTOPP_ASSERT(count > 0);
+	// CRYPTOPP_ASSERT(count > 0);
 
 	size_t i=0;
 	byte acc8 = 0;
 
-	if (IsAligned<word32>(buf) && IsAligned<word32>(mask))
+	if (IsAligned<word32>(buf) && IsAligned<word32>(mask) && count)
 	{
 		word32 acc32 = 0;
 		if (!CRYPTOPP_BOOL_SLOW_WORD64 && IsAligned<word64>(buf) && IsAligned<word64>(mask))

misc.h

@@ -62,36 +62,66 @@
 #include <stdlib.h>
 #endif
 
-#if defined(__GNUC__) && defined(__linux__)
+#if (defined(__GNUC__) || defined(__clang__)) && defined(__linux__)
-#define CRYPTOPP_BYTESWAP_AVAILABLE
+#define CRYPTOPP_BYTESWAP_AVAILABLE 1
 #include <byteswap.h>
 #endif
 
+// Limit to ARM A-32. Aarch64 is failing self tests.
+#if defined(__arm__) && (defined(__GNUC__) || defined(__clang__)) && (__ARM_ARCH >= 6)
+#define CRYPTOPP_ARM_BYTEREV_AVAILABLE 1
+#endif
+
+// Limit to ARM A-32. Aarch64 is failing self tests.
+#if defined(__arm__) && (defined(__GNUC__) || defined(__clang__)) && (__ARM_ARCH >= 7)
+#define CRYPTOPP_ARM_BITREV_AVAILABLE 1
+#endif
+
 #if defined(__BMI__)
 # include <x86intrin.h>
+# include <immintrin.h>
 #endif  // GCC and BMI
 
+// More LLVM bullshit. Apple Clang 6.0 does not define them.
+// Later version of Clang defines them and results in warnings.
+#if defined(__clang__)
+# ifndef _blsr_u32
+#  define _blsr_u32 __blsr_u32
+# endif
+# ifndef _blsr_u64
+#  define _blsr_u64 __blsr_u64
+# endif
+# ifndef _tzcnt_u32
+#  define _tzcnt_u32 __tzcnt_u32
+# endif
+# ifndef _tzcnt_u64
+#  define _tzcnt_u64 __tzcnt_u64
+# endif
+#endif
+
 #endif  // CRYPTOPP_DOXYGEN_PROCESSING
 
 #if CRYPTOPP_DOXYGEN_PROCESSING
 /// \brief The maximum value of a machine word
-/// \details SIZE_MAX provides the maximum value of a machine word. The value is
+/// \details <tt>SIZE_MAX</tt> provides the maximum value of a machine word. The value
-///   0xffffffff on 32-bit machines, and 0xffffffffffffffff on 64-bit machines.
+///  is <tt>0xffffffff</tt> on 32-bit targets, and <tt>0xffffffffffffffff</tt> on 64-bit
-/// Internally, SIZE_MAX is defined as __SIZE_MAX__ if __SIZE_MAX__ is defined. If not
+///  targets.
-///   defined, then SIZE_T_MAX is tried. If neither __SIZE_MAX__ nor SIZE_T_MAX is
+/// \details If <tt>SIZE_MAX</tt> is not defined, then <tt>__SIZE_MAX__</tt> is used if
-///   is defined, the library uses std::numeric_limits<size_t>::max(). The library
+///  defined. If not defined, then <tt>SIZE_T_MAX</tt> is used if defined. If not defined,
-///   prefers __SIZE_MAX__ because its a constexpr that is optimized well
+///  then the library uses <tt>std::numeric_limits<size_t>::max()</tt>.
-///   by all compilers. std::numeric_limits<size_t>::max() is not a constexpr,
+/// \details The library prefers <tt>__SIZE_MAX__</tt> or <tt>__SIZE_T_MAX__</tt> because
-///   and it is not always optimized well.
+///  they are effectively <tt>constexpr</tt> that is optimized well by all compilers.
+///  <tt>std::numeric_limits<size_t>::max()</tt> is not always a <tt>constexpr</tt>, and
+///  it is not always optimized well.
 #  define SIZE_MAX ...
 #else
 // Its amazing portability problems still plague this simple concept in 2015.
 // http://stackoverflow.com/questions/30472731/which-c-standard-header-defines-size-max
 // Avoid NOMINMAX macro on Windows. http://support.microsoft.com/en-us/kb/143208
 #ifndef SIZE_MAX
-# if defined(__SIZE_MAX__) && (__SIZE_MAX__ > 0)
+# if defined(__SIZE_MAX__)
 #  define SIZE_MAX __SIZE_MAX__
-# elif defined(SIZE_T_MAX) && (SIZE_T_MAX > 0)
+# elif defined(SIZE_T_MAX)
 #  define SIZE_MAX SIZE_T_MAX
 # elif defined(__SIZE_TYPE__)
 #  define SIZE_MAX (~(__SIZE_TYPE__)0)
@@ -112,8 +142,13 @@ class Integer;
 #if CRYPTOPP_DOXYGEN_PROCESSING
 /// \brief Compile time assertion
 /// \param expr the expression to evaluate
-/// \details Asserts the expression expr though a dummy struct.
+/// \details Asserts the expression <tt>expr</tt> during compile. If C++14 and
+///  N3928 are available, then C++14 <tt>static_assert</tt> is used. Otherwise,
+///  a <tt>CompileAssert</tt> structure is used. When the structure is used
+///  a negative-sized array triggers the assert at compile time.
 # define CRYPTOPP_COMPILE_ASSERT(expr) { ... }
+#elif defined(CRYPTOPP_CXX14_STATIC_ASSERT)
+# define CRYPTOPP_COMPILE_ASSERT(expr) static_assert(expr)
 #else // CRYPTOPP_DOXYGEN_PROCESSING
 template <bool b>
 struct CompileAssert
@@ -122,10 +157,13 @@ struct CompileAssert
 };
 
 #define CRYPTOPP_COMPILE_ASSERT(assertion) CRYPTOPP_COMPILE_ASSERT_INSTANCE(assertion, __LINE__)
+#define CRYPTOPP_ASSERT_JOIN(X, Y) CRYPTOPP_DO_ASSERT_JOIN(X, Y)
+#define CRYPTOPP_DO_ASSERT_JOIN(X, Y) X##Y
+
 #if defined(CRYPTOPP_EXPORTS) || defined(CRYPTOPP_IMPORTS)
 # define CRYPTOPP_COMPILE_ASSERT_INSTANCE(assertion, instance)
 #else
-# if defined(__GNUC__)
+# if defined(__GNUC__) || defined(__clang__)
 #  define CRYPTOPP_COMPILE_ASSERT_INSTANCE(assertion, instance) \
        static CompileAssert<(assertion)> \
        CRYPTOPP_ASSERT_JOIN(cryptopp_CRYPTOPP_ASSERT_, instance) __attribute__ ((unused))
@@ -133,10 +171,8 @@ struct CompileAssert
 #  define CRYPTOPP_COMPILE_ASSERT_INSTANCE(assertion, instance) \
        static CompileAssert<(assertion)> \
        CRYPTOPP_ASSERT_JOIN(cryptopp_CRYPTOPP_ASSERT_, instance)
-# endif // __GNUC__
+# endif // GCC or Clang
 #endif
-#define CRYPTOPP_ASSERT_JOIN(X, Y) CRYPTOPP_DO_ASSERT_JOIN(X, Y)
-#define CRYPTOPP_DO_ASSERT_JOIN(X, Y) X##Y
 
 #endif // CRYPTOPP_DOXYGEN_PROCESSING
 
@@ -194,16 +230,21 @@ protected:
 
 /// \brief Ensures an object is not copyable
 /// \details NotCopyable ensures an object is not copyable by making the
-///   copy constructor and assignment operator private. Deleters are not
+///   copy constructor and assignment operator private. Deleters are used
-///   used under C++11.
+///   under C++11.
 /// \sa Clonable class
 class NotCopyable
 {
 public:
 	NotCopyable() {}
+#if CRYPTOPP_CXX11_DELETED_FUNCTIONS
+	NotCopyable(const NotCopyable &) = delete;
+	void operator=(const NotCopyable &) = delete;
+#else
 private:
 	NotCopyable(const NotCopyable &);
 	void operator=(const NotCopyable &);
+#endif
 };
 
 /// \brief An object factory function
@@ -388,6 +429,38 @@ inline size_t PtrByteDiff(const PTR pointer1, const PTR pointer2)
 	return (size_t)(reinterpret_cast<uintptr_t>(pointer1) - reinterpret_cast<uintptr_t>(pointer2));
 }
 
+/// \brief Pointer to the first element of a string
+/// \param str std::string
+/// \details BytePtr returns NULL pointer for an empty string.
+/// \return Pointer to the first element of a string
+inline byte* BytePtr(std::string& str)
+{
+	// Caller wants a writeable pointer
+	CRYPTOPP_ASSERT(str.empty() == false);
+
+	if (str.empty())
+		return NULLPTR;
+	return reinterpret_cast<byte*>(&str[0]);
+}
+
+/// \brief Const pointer to the first element of a string
+/// \param str std::string
+/// \details ConstBytePtr returns non-NULL pointer for an empty string.
+/// \return Pointer to the first element of a string
+inline const byte* ConstBytePtr(const std::string& str)
+{
+	// Use c_str() so a pointer is always available
+	return reinterpret_cast<const byte*>(str.c_str());
+}
+
+/// \brief Size of a string
+/// \param str std::string
+/// \return size of a string
+inline size_t BytePtrSize(const std::string& str)
+{
+	return str.size();
+}
+
 #if (!__STDC_WANT_SECURE_LIB__ && !defined(_MEMORY_S_DEFINED)) || defined(CRYPTOPP_WANT_SECURE_LIB)
 
 /// \brief Bounds checking replacement for memcpy()
@@ -871,11 +944,12 @@ CRYPTOPP_DLL void CRYPTOPP_API xorbuf(byte *output, const byte *input, const byt
 /// \param buf1 the first buffer
 /// \param buf2 the second buffer
 /// \param count the size of the buffers, in bytes
-/// \details The function effectively performs an XOR of the elements in two equally sized
+/// \details VerifyBufsEqual performs an XOR of the elements in two equally sized
-///   buffers and retruns a result based on the XOR operation. The function is near
+///   buffers and retruns a result based on the XOR operation. A count of 0 returns
-///   constant-time because CPU micro-code timings could affect the "constant-ness".
+///   true because two empty buffers are considered equal.
-///   Calling code is responsible for mitigating timing attacks if the buffers are not
+/// \details The function is near constant-time because CPU micro-code timings could
-///   equally sized.
+///   affect the "constant-ness". Calling code is responsible for mitigating timing
+///   attacks if the buffers are not equally sized.
 /// \sa ModPowerOf2
 CRYPTOPP_DLL bool CRYPTOPP_API VerifyBufsEqual(const byte *buf1, const byte *buf2, size_t count);
 
@@ -1193,9 +1267,15 @@ CRYPTOPP_DLL void CRYPTOPP_API CallNewHandler();
 /// \note The function is not constant time because it stops processing when the carry is 0.
 inline void IncrementCounterByOne(byte *inout, unsigned int size)
 {
-	CRYPTOPP_ASSERT(inout != NULLPTR); CRYPTOPP_ASSERT(size < INT_MAX);
+	CRYPTOPP_ASSERT(inout != NULLPTR);
-	for (int i=int(size-1), carry=1; i>=0 && carry; i--)
+
-		carry = !++inout[i];
+	unsigned int carry=1;
+	while (carry && size != 0)
+	{
+		// On carry inout[n] equals 0
+		carry = ! ++inout[size-1];
+		size--;
+	}
 }
 
 /// \brief Performs an addition with carry on a block of bytes
@@ -1207,12 +1287,22 @@ inline void IncrementCounterByOne(byte *inout, unsigned int size)
 /// \details The function is close to near-constant time because it operates on all the bytes in the blocks.
 inline void IncrementCounterByOne(byte *output, const byte *input, unsigned int size)
 {
-	CRYPTOPP_ASSERT(output != NULLPTR); CRYPTOPP_ASSERT(input != NULLPTR); CRYPTOPP_ASSERT(size < INT_MAX);
+	CRYPTOPP_ASSERT(output != NULLPTR);
+	CRYPTOPP_ASSERT(input != NULLPTR);
 
-	int i, carry;
+	unsigned int carry=1;
-	for (i=int(size-1), carry=1; i>=0 && carry; i--)
+	while (carry && size != 0)
-		carry = ((output[i] = input[i]+1) == 0);
+	{
-	memcpy_s(output, size, input, size_t(i)+1);
+		// On carry output[n] equals 0
+		carry = ! (output[size-1] = input[size-1] + 1);
+		size--;
+	}
+
+	while (size != 0)
+	{
+		output[size-1] = input[size-1];
+		size--;
+	}
 }
 
 /// \brief Performs a branchless swap of values a and b if condition c is true
@@ -1936,7 +2026,8 @@ inline unsigned int GetByte(ByteOrder order, T value, unsigned int index)
 
 /// \brief Reverses bytes in a 8-bit value
 /// \param value the 8-bit value to reverse
-/// \note ByteReverse returns the value passed to it since there is nothing to reverse
+/// \note ByteReverse returns the value passed to it since there is nothing to
+///  reverse.
 inline byte ByteReverse(byte value)
 {
 	return value;
@@ -1944,7 +2035,8 @@ inline byte ByteReverse(byte value)
 
 /// \brief Reverses bytes in a 16-bit value
 /// \param value the 16-bit value to reverse
-/// \details ByteReverse calls bswap if available. Otherwise the function performs a 8-bit rotate on the word16
+/// \details ByteReverse calls bswap if available. Otherwise the function
+///  performs a 8-bit rotate on the word16.
 inline word16 ByteReverse(word16 value)
 {
 #if defined(CRYPTOPP_BYTESWAP_AVAILABLE)
@@ -1958,14 +2050,19 @@ inline word16 ByteReverse(word16 value)
 
 /// \brief Reverses bytes in a 32-bit value
 /// \param value the 32-bit value to reverse
-/// \details ByteReverse calls bswap if available. Otherwise the function uses a combination of rotates on the word32
+/// \details ByteReverse calls bswap if available. Otherwise the function uses
+///  a combination of rotates on the word32.
 inline word32 ByteReverse(word32 value)
 {
-#if defined(__GNUC__) && defined(CRYPTOPP_X86_ASM_AVAILABLE)
+#if defined(CRYPTOPP_BYTESWAP_AVAILABLE)
+	return bswap_32(value);
+#elif defined(CRYPTOPP_ARM_BYTEREV_AVAILABLE)
+	word32 rvalue;
+	__asm__ ("rev %0, %1" : "=r" (rvalue) : "r" (value));
+	return rvalue;
+#elif defined(__GNUC__) && defined(CRYPTOPP_X86_ASM_AVAILABLE)
 	__asm__ ("bswap %0" : "=r" (value) : "0" (value));
 	return value;
-#elif defined(CRYPTOPP_BYTESWAP_AVAILABLE)
-	return bswap_32(value);
 #elif defined(__MWERKS__) && TARGET_CPU_PPC
 	return (word32)__lwbrx(&value,0);
 #elif (_MSC_VER >= 1400) || (defined(_MSC_VER) && !defined(_DLL))
@@ -1982,14 +2079,15 @@ inline word32 ByteReverse(word32 value)
 
 /// \brief Reverses bytes in a 64-bit value
 /// \param value the 64-bit value to reverse
-/// \details ByteReverse calls bswap if available. Otherwise the function uses a combination of rotates on the word64
+/// \details ByteReverse calls bswap if available. Otherwise the function uses
+///  a combination of rotates on the word64.
 inline word64 ByteReverse(word64 value)
 {
-#if defined(__GNUC__) && defined(CRYPTOPP_X86_ASM_AVAILABLE) && defined(__x86_64__)
+#if defined(CRYPTOPP_BYTESWAP_AVAILABLE)
+	return bswap_64(value);
+#elif defined(__GNUC__) && defined(CRYPTOPP_X86_ASM_AVAILABLE) && defined(__x86_64__)
 	__asm__ ("bswap %0" : "=r" (value) : "0" (value));
 	return value;
-#elif defined(CRYPTOPP_BYTESWAP_AVAILABLE)
-	return bswap_64(value);
 #elif (_MSC_VER >= 1400) || (defined(_MSC_VER) && !defined(_DLL))
 	return _byteswap_uint64(value);
 #elif CRYPTOPP_BOOL_SLOW_WORD64
@@ -2003,7 +2101,7 @@ inline word64 ByteReverse(word64 value)
 
 /// \brief Reverses bits in a 8-bit value
 /// \param value the 8-bit value to reverse
-/// \details BitReverse performs a combination of shifts on the byte
+/// \details BitReverse performs a combination of shifts on the byte.
 inline byte BitReverse(byte value)
 {
 	value = byte((value & 0xAA) >> 1) | byte((value & 0x55) << 1);
@@ -2013,29 +2111,45 @@ inline byte BitReverse(byte value)
 
 /// \brief Reverses bits in a 16-bit value
 /// \param value the 16-bit value to reverse
-/// \details BitReverse performs a combination of shifts on the word16
+/// \details BitReverse performs a combination of shifts on the word16.
 inline word16 BitReverse(word16 value)
 {
+#if defined(CRYPTOPP_ARM_BITREV_AVAILABLE)
+	// 4 instructions on ARM.
+	word32 rvalue;
+	__asm__ ("rbit %0, %1" : "=r" (rvalue) : "r" (value));
+	return word16(rvalue >> 16);
+#else
+	// 15 instructions on ARM.
 	value = word16((value & 0xAAAA) >> 1) | word16((value & 0x5555) << 1);
 	value = word16((value & 0xCCCC) >> 2) | word16((value & 0x3333) << 2);
 	value = word16((value & 0xF0F0) >> 4) | word16((value & 0x0F0F) << 4);
 	return ByteReverse(value);
+#endif
 }
 
 /// \brief Reverses bits in a 32-bit value
 /// \param value the 32-bit value to reverse
-/// \details BitReverse performs a combination of shifts on the word32
+/// \details BitReverse performs a combination of shifts on the word32.
 inline word32 BitReverse(word32 value)
 {
+#if defined(CRYPTOPP_ARM_BITREV_AVAILABLE)
+	// 2 instructions on ARM.
+	word32 rvalue;
+	__asm__ ("rbit %0, %1" : "=r" (rvalue) : "r" (value));
+	return rvalue;
+#else
+	// 19 instructions on ARM.
 	value = word32((value & 0xAAAAAAAA) >> 1) | word32((value & 0x55555555) << 1);
 	value = word32((value & 0xCCCCCCCC) >> 2) | word32((value & 0x33333333) << 2);
 	value = word32((value & 0xF0F0F0F0) >> 4) | word32((value & 0x0F0F0F0F) << 4);
 	return ByteReverse(value);
+#endif
 }
 
 /// \brief Reverses bits in a 64-bit value
 /// \param value the 64-bit value to reverse
-/// \details BitReverse performs a combination of shifts on the word64
+/// \details BitReverse performs a combination of shifts on the word64.
 inline word64 BitReverse(word64 value)
 {
 #if CRYPTOPP_BOOL_SLOW_WORD64
@@ -2063,9 +2177,11 @@ inline T BitReverse(T value)
 		return (T)BitReverse((word16)value);
 	else if (sizeof(T) == 4)
 		return (T)BitReverse((word32)value);
+	else if (sizeof(T) == 8)
+		return (T)BitReverse((word64)value);
 	else
 	{
-		CRYPTOPP_ASSERT(sizeof(T) == 8);
+		CRYPTOPP_ASSERT(0);
 		return (T)BitReverse((word64)value);
 	}
 }