aarch64 support

3rdparty/libpng/CMakeLists.txt

@@ -14,7 +14,7 @@ ocv_include_directories("${CMAKE_CURRENT_SOURCE_DIR}" ${ZLIB_INCLUDE_DIRS})
 file(GLOB lib_srcs *.c)
 file(GLOB lib_hdrs *.h)
 
-if(NEON AND CMAKE_SIZEOF_VOID_P EQUAL 4)
+if(NEON AND ARM)
   list(APPEND lib_srcs arm/filter_neon.S)
   add_definitions(-DPNG_ARM_NEON)
 endif()

CMakeLists.txt

@@ -225,8 +225,8 @@ OCV_OPTION(ENABLE_POPCNT              "Enable POPCNT instructions"
 OCV_OPTION(ENABLE_AVX                 "Enable AVX instructions"                                  OFF  IF ((MSVC OR CMAKE_COMPILER_IS_GNUCXX) AND (X86 OR X86_64)) )
 OCV_OPTION(ENABLE_AVX2                "Enable AVX2 instructions"                                 OFF  IF ((MSVC OR CMAKE_COMPILER_IS_GNUCXX) AND (X86 OR X86_64)) )
 OCV_OPTION(ENABLE_FMA3                "Enable FMA3 instructions"                                 OFF  IF ((MSVC OR CMAKE_COMPILER_IS_GNUCXX) AND (X86 OR X86_64)) )
-OCV_OPTION(ENABLE_NEON                "Enable NEON instructions"                                 OFF  IF CMAKE_COMPILER_IS_GNUCXX AND (ARM OR IOS) )
-OCV_OPTION(ENABLE_VFPV3               "Enable VFPv3-D32 instructions"                            OFF  IF CMAKE_COMPILER_IS_GNUCXX AND (ARM OR IOS) )
+OCV_OPTION(ENABLE_NEON                "Enable NEON instructions"                                 OFF  IF CMAKE_COMPILER_IS_GNUCXX AND (ARM OR AARCH64 OR IOS) )
+OCV_OPTION(ENABLE_VFPV3               "Enable VFPv3-D32 instructions"                            OFF  IF CMAKE_COMPILER_IS_GNUCXX AND (ARM OR AARCH64 OR IOS) )
 OCV_OPTION(ENABLE_NOISY_WARNINGS      "Show all warnings even if they are too noisy"             OFF )
 OCV_OPTION(OPENCV_WARNINGS_ARE_ERRORS "Treat warnings as errors"                                 OFF )
 OCV_OPTION(ENABLE_WINRT_MODE          "Build with Windows Runtime support"                       OFF  IF WIN32 )

cmake/OpenCVDetectCUDA.cmake

@@ -80,8 +80,13 @@ if(CUDA_FOUND)
 
   if(NOT DEFINED __cuda_arch_bin)
     if(ANDROID)
+      if(ARM)
         set(__cuda_arch_bin "3.2")
         set(__cuda_arch_ptx "")
+      elseif(AARCH64)
+        set(__cuda_arch_bin "5.2")
+        set(__cuda_arch_ptx "")
+      endif()
     else()
       if(${CUDA_VERSION} VERSION_LESS "5.0")
         set(__cuda_arch_bin "1.1 1.2 1.3 2.0 2.1(2.0) 3.0")

cmake/OpenCVDetectCXXCompiler.cmake

@@ -107,8 +107,10 @@ elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "amd64.*|x86_64.*|AMD64.*")
   set(X86_64 1)
 elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "i686.*|i386.*|x86.*|amd64.*|AMD64.*")
   set(X86 1)
-elseif (CMAKE_SYSTEM_PROCESSOR MATCHES "arm.*|ARM.*")
+elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(arm.*|ARM.*)")
   set(ARM 1)
+elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(aarch64.*|AARCH64.*)")
+  set(AARCH64 1)
 endif()
 
 

modules/core/include/opencv2/core/cvdef.h

@@ -191,7 +191,7 @@
 # include "arm_neon.h"
 # define CV_NEON 1
 # define CPU_HAS_NEON_FEATURE (true)
-#elif defined(__ARM_NEON__)
+#elif defined(__ARM_NEON__) || (defined (__ARM_NEON) && defined(__aarch64__))
 #  include <arm_neon.h>
 #  define CV_NEON 1
 #endif

modules/core/src/system.cpp

@@ -319,6 +319,9 @@ struct HWFeatures
         }
 
     #if defined ANDROID || defined __linux__
+    #ifdef __aarch64__
+        f.have[CV_CPU_NEON] = true;
+    #else
         int cpufile = open("/proc/self/auxv", O_RDONLY);
 
         if (cpufile >= 0)
@@ -337,7 +340,8 @@ struct HWFeatures
 
             close(cpufile);
         }
-    #elif (defined __clang__ || defined __APPLE__) && defined __ARM_NEON__
+    #endif
+    #elif (defined __clang__ || defined __APPLE__) && (defined __ARM_NEON__ || (defined __ARM_NEON && defined __aarch64__))
         f.have[CV_CPU_NEON] = true;
     #endif
 

modules/cudalegacy/test/TestHaarCascadeApplication.cpp

@@ -52,7 +52,7 @@ namespace
         ~FpuControl();
 
     private:
-    #if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__)
+    #if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__) && !defined(__aarch64__)
         fpu_control_t fpu_oldcw, fpu_cw;
     #elif defined(_WIN32) && !defined(_WIN64)
         unsigned int fpu_oldcw, fpu_cw;
@@ -61,7 +61,7 @@ namespace
 
     FpuControl::FpuControl()
     {
-    #if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__)
+    #if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__) && !defined(__aarch64__)
         _FPU_GETCW(fpu_oldcw);
         fpu_cw = (fpu_oldcw & ~_FPU_EXTENDED & ~_FPU_DOUBLE & ~_FPU_SINGLE) | _FPU_SINGLE;
         _FPU_SETCW(fpu_cw);
@@ -74,7 +74,7 @@ namespace
 
     FpuControl::~FpuControl()
     {
-    #if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__)
+    #if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__) && !defined(__aarch64__)
         _FPU_SETCW(fpu_oldcw);
     #elif defined(_WIN32) && !defined(_WIN64)
         _controlfp_s(&fpu_cw, fpu_oldcw, _MCW_PC);

modules/cudalegacy/test/test_precomp.hpp

@@ -51,7 +51,7 @@
 #ifndef __OPENCV_TEST_PRECOMP_HPP__
 #define __OPENCV_TEST_PRECOMP_HPP__
 
-#if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__)
+#if defined(__GNUC__) && !defined(__APPLE__) && !defined(__arm__) && !defined(__aarch64__)
     #include <fpu_control.h>
 #endif