Add ARMv8.4 cpu feature detection support (GH weidai11#685) (weidai11…

…#687)

This PR adds ARMv8.4 cpu feature detection support. Previously we only needed ARMv8.1 and things were much easier. For example, ARMv8.1 `__ARM_FEATURE_CRYPTO` meant PMULL, AES, SHA-1 and SHA-256 were available. ARMv8.4 `__ARM_FEATURE_CRYPTO` means PMULL, AES, SHA-1, SHA-256, SHA-512, SHA-3, SM3 and SM4 are  available. 

We still use the same pattern as before. We make something available based on compiler version and/or preprocessor macros. But this time around we had to tighten things up a bit to ensure ARMv8.4 did not cross-pollinate down into ARMv8.1.

ARMv8.4 is largely untested at the moment. There is no hardware in the field and CI lacks QEMU with the relevant patches/support. We will probably have to revisit some of this stuff in the future.

Since this update applies to ARM gadgets we took the time to expand Android and iOS testing on Travis. Travis now tests more platforms, and includes Autotools and CMake builds, too.

.travis.yml

@@ -51,8 +51,16 @@ matrix:
   include:
     - os: linux
       env:
-        - BUILD_MODE=android
-        - PLATFORM=armeabi
+        - BUILD_MODE=autotools
+    - os: linux
+      env:
+        - BUILD_MODE=cmake
+    - os: osx
+      env:
+        - BUILD_MODE=autotools
+    - os: osx
+      env:
+        - BUILD_MODE=cmake
     - os: linux
       env:
         - BUILD_MODE=android
@@ -85,6 +93,10 @@ matrix:
       env:
         - BUILD_MODE=ios
         - PLATFORM=iPhoneOS
+    - os: osx
+      env:
+        - BUILD_MODE=ios
+        - PLATFORM=Arm64
     - os: osx
       env:
         - BUILD_MODE=ios
@@ -103,21 +115,43 @@ matrix:
       env:
         - BUILD_MODE=ios
         - PLATFORM=WatchOS
+    - os: linux
+      env:
+        - BUILD_MODE=android
+        - PLATFORM=armeabi
+    - os: linux
+      env:
+        - BUILD_MODE=android
+        - PLATFORM=mipsel
+    - os: linux
+      env:
+        - BUILD_MODE=android
+        - PLATFORM=mipsel64
 
 before_install:
   - |
     if [[ "$BUILD_MODE" == "android" ]]; then
+        # https://github.com/travis-ci/travis-ci/issues/9037
+        sudo apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv 0C49F3730359A145
         TestScripts/install-android.sh
     fi
 
 script:
   - |
     if [[ "$BUILD_MODE" == "ios" ]]; then
         cp ./TestScripts/setenv-ios.sh .
-        TestScripts/cryptest-ios.sh
+        cp ./TestScripts/cryptest-ios.sh .
+        ./cryptest-ios.sh
     elif [[ "$BUILD_MODE" == "android" ]]; then
-        cp ./TestScripts/setenv-android.sh .
-        TestScripts/cryptest-android.sh
+        cp ./TestScripts/setenv-android-gcc.sh .
+        cp ./TestScripts/cryptest-android.sh .
+        ./cryptest-android.sh
+    elif [[ "$BUILD_MODE" == "autotools" ]]; then
+        cp ./TestScripts/cryptest-autotools.sh .
+        ./cryptest-autotools.sh
+    elif [[ "$BUILD_MODE" == "cmake" ]]; then
+        cp ./TestScripts/cryptest-cmake.sh .
+        ./cryptest-cmake.sh
     elif [[ "$BUILD_MODE" == "debug" ]]; then
         CXXFLAGS="-DDEBUG -g2 -O1" make -j "$BUILD_JOBS"
         ./cryptest.exe v

GNUmakefile

@@ -385,12 +385,14 @@ ifeq ($(IS_NEON),1)
   HAVE_NEON = $(shell echo | $(CXX) -x c++ $(CXXFLAGS) -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon -dM -E - 2>/dev/null | $(GREP) -i -c -E '\<__ARM_NEON\>')
   ifeq ($(HAVE_NEON),1)
     NEON_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    ARIA_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
     AES_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    CRC_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
     GCM_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
-    ARIA_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
     BLAKE2_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
     CHAM_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
     LEA_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
+    SHA_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
     SIMECK_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
     SIMON_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon
     SPECK_FLAG = -march=armv7-a -mfloat-abi=$(FP_ABI) -mfpu=neon

GNUmakefile-cross

@@ -32,10 +32,11 @@ endif
 
 IS_i686 := $(shell echo "$HOSTX" | $(EGREP) -v 64 | $(EGREP) -i -c 'i.86')
 IS_x86_64 := $(shell echo "$HOSTX" | $(EGREP) -i -c 'x86_64|amd64')
-IS_ARM := $(shell echo "$HOSTX" | $(EGREP) -i -c 'arm')
+IS_ARM32 := $(shell echo "$(HOSTX)" | $(EGREP) -i -c -E 'arm|armhf|arm7l|eabihf')
 IS_ARMv8 := $(shell echo "$HOSTX" | $(EGREP) -i -c 'aarch32|aarch64')
 
-CLANG_COMPILER := $(shell $(CXX) --version 2>&1 | $(EGREP) -i -c "clang")
+GCC_COMPILER := $(shell $(CXX) --version 2>/dev/null | $(EGREP) -v -E '(llvm|clang)' | $(EGREP) -i -c -E '(gcc|g\+\+)')
+CLANG_COMPILER := $(shell $(CXX) --version 2>/dev/null | $(EGREP) -i -c -E '(llvm|clang)')
 
 IS_IOS ?= 0
 IS_ANDROID ?= 0
@@ -321,16 +322,25 @@ endif
 
 # List cryptlib.cpp first, then cpu.cpp, then integer.cpp to tame C++ static initialization problems.
 SRCS := cryptlib.cpp cpu.cpp integer.cpp $(filter-out cryptlib.cpp cpu.cpp integer.cpp pch.cpp simple.cpp winpipes.cpp cryptlib_bds.cpp,$(sort $(wildcard *.cpp)))
-
 # 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.
+# Disable Thumb via -marm due to unaligned byte buffers.
+ifeq ($(IS_ARM32),1)
+CRYPTOGAMS_AES_ARCH = -march=armv7-a -marm
+SRCS += aes-armv4.S
+endif
+
 # List cryptlib.cpp first, then cpu.cpp, then integer.cpp to tame C++ static initialization problems.
 OBJS := $(SRCS:.cpp=.o)
+OBJS := $(OBJS:.S=.o)
 
 # List test.cpp first to tame C++ static initialization problems.
-TESTSRCS := adhoc.cpp test.cpp bench1.cpp bench2.cpp validat0.cpp validat1.cpp validat2.cpp validat3.cpp validat4.cpp datatest.cpp regtest1.cpp regtest2.cpp regtest3.cpp fipsalgt.cpp dlltest.cpp
+TESTSRCS := adhoc.cpp test.cpp bench1.cpp bench2.cpp validat0.cpp validat1.cpp validat2.cpp validat3.cpp validat4.cpp datatest.cpp regtest1.cpp regtest2.cpp regtest3.cpp dlltest.cpp fipsalgt.cpp
 TESTINCL := bench.h factory.h validate.h
+
+# Test objects
 TESTOBJS := $(TESTSRCS:.cpp=.o)
 LIBOBJS := $(filter-out $(TESTOBJS),$(OBJS))
 
@@ -494,6 +504,10 @@ endif # Dependencies
 cpu-features.o: cpu-features.h cpu-features.c
 	$(CXX) $(strip $(CXXFLAGS) -fpermissive -c) cpu-features.c
 
+# Cryptogams ARM asm implementation. CRYPTOGAMS_AES_ARCH includes -marm.
+aes-armv4.o : aes-armv4.S
+	$(CXX) $(strip $(CXXFLAGS) $(CRYPTOGAMS_AES_ARCH) -c) $<
+
 # SSE4.2 or NEON available
 aria-simd.o : aria-simd.cpp
 	$(CXX) $(strip $(CXXFLAGS) $(ARIA_FLAG) -c) $<

Readme.txt

@@ -407,3 +407,4 @@ June 2015 - Changing of the guard. Wei Dai turned the library over to the
         still provides guidance when we have questions.
 
 Originally written by Wei Dai, maintained by the Crypto++ Project
+

TestScripts/cryptest-android.sh

@@ -17,6 +17,7 @@ else
 	PLATFORMS=(${PLATFORM})
 fi
 RUNTIMES=(gnu-static gnu-shared stlport-static stlport-shared) #llvm-static llvm-shared
+
 for platform in ${PLATFORMS[@]}
 do
 	for runtime in ${RUNTIMES[@]}
@@ -30,14 +31,15 @@ do
 		# Test if we can set the environment for the platform
 		./setenv-android-gcc.sh "$platform" "$runtime"
 
-		if [ "$?" -eq "0" ]; then
+		if [ "$?" -eq "0" ];
+		then
 			echo
 			echo "Building for $platform using $runtime..."
 			echo
 
 			# run in subshell to not keep any env vars
 			(
-				. ./setenv-android-gcc.sh "$platform" "$runtime" > /dev/null 2>&1
+				source ./setenv-android-gcc.sh "$platform" "$runtime" > /dev/null 2>&1
 				make -f GNUmakefile-cross static dynamic cryptest.exe
 				if [ "$?" -eq "0" ]; then
 					echo "$platform:$runtime ==> SUCCESS" >> /tmp/build.log
@@ -48,7 +50,7 @@ do
 			)
 		else
 			echo
-			echo "$platform with $runtime not supported by Android"
+			echo "There were problems testing $platform with $runtime"
 			echo "$platform:$runtime ==> FAILURE" >> /tmp/build.log
 			touch /tmp/build.failed
 		fi

TestScripts/cryptest-ios.sh

@@ -11,7 +11,7 @@
 # ====================================================================
 
 if [ -z "${PLATFORM-}" ]; then
-	PLATFORMS=(iPhoneOS iPhoneSimulator WatchOS WatchSimulator AppleTVOS AppleTVSimulator)
+	PLATFORMS=(iPhoneOS iPhoneSimulator Arm64 WatchOS WatchSimulator AppleTVOS AppleTVSimulator)
 else
 	PLATFORMS=(${PLATFORM})
 fi

TestScripts/setenv-android-gcc.sh

@@ -167,14 +167,14 @@ case "$THE_ARCH" in
 	TOOLCHAIN_NAME="i686-linux-android"
 	AOSP_ABI="x86"
 	AOSP_ARCH="arch-x86"
-	AOSP_FLAGS="-mtune=intel -mssse3 -mfpmath=sse -DCRYPTOPP_DISABLE_SSE4 -funwind-tables -fexceptions -frtti"
+	AOSP_FLAGS="-mtune=intel -mssse3 -mfpmath=sse -funwind-tables -fexceptions -frtti"
 	;;
   x86_64|x64)
 	TOOLCHAIN_ARCH="x86_64"
 	TOOLCHAIN_NAME="x86_64-linux-android"
 	AOSP_ABI="x86_64"
 	AOSP_ARCH="arch-x86_64"
-	AOSP_FLAGS="-march=x86-64 -msse4.2 -mpopcnt -mtune=intel -DCRYPTOPP_DISABLE_CLMUL -DCRYPTOPP_DISABLE_AESNI -DCRYPTOPP_DISABLE_SHANI -funwind-tables -fexceptions -frtti"
+	AOSP_FLAGS="-march=x86-64 -msse4.2 -mpopcnt -mtune=intel -funwind-tables -fexceptions -frtti"
 	;;
   *)
 	echo "ERROR: Unknown architecture $1"

TestScripts/setenv-android-old.sh

@@ -157,14 +157,14 @@ case "$THE_ARCH" in
 	TOOLCHAIN_NAME="i686-linux-android"
 	AOSP_ABI="x86"
 	AOSP_ARCH="arch-x86"
-	AOSP_FLAGS="-mtune=intel -mssse3 -mfpmath=sse -DCRYPTOPP_DISABLE_SSE4 -DCRYPTOPP_DISABLE_CLMUL -DCRYPTOPP_DISABLE_AES -DCRYPTOPP_DISABLE_SHA -funwind-tables -fexceptions -frtti"
+	AOSP_FLAGS="-mtune=intel -mssse3 -mfpmath=sse -funwind-tables -fexceptions -frtti"
 	;;
   x86_64|x64)
 	TOOLCHAIN_ARCH="x86_64"
 	TOOLCHAIN_NAME="x86_64-linux-android"
 	AOSP_ABI="x86_64"
 	AOSP_ARCH="arch-x86_64"
-	AOSP_FLAGS="-march=x86-64 -msse4.2 -mpopcnt -mtune=intel -DCRYPTOPP_DISABLE_CLMUL -DCRYPTOPP_DISABLE_AES -DCRYPTOPP_DISABLE_SHA -funwind-tables -fexceptions -frtti"
+	AOSP_FLAGS="-march=x86-64 -msse4.2 -mpopcnt -mtune=intel -funwind-tables -fexceptions -frtti"
 	;;
   *)
 	echo "ERROR: Unknown architecture $1"

adv-simd.h

@@ -46,6 +46,11 @@
 # include <arm_neon.h>
 #endif
 
+#if defined(CRYPTOPP_ARM_ACLE_AVAILABLE)
+# include <stdint.h>
+# include <arm_acle.h>
+#endif
+
 #if (CRYPTOPP_SSE2_INTRIN_AVAILABLE)
 # include <emmintrin.h>
 # include <pmmintrin.h>