ARROW-4297: [C++] Fix build error with MinGW-w64 32-bit

appveyor.yml

@@ -36,14 +36,6 @@ cache:
 
 matrix:
  fast_finish: true
- allow_failures:
- # Can't build with 32-bit MinGW for now.
- # See https://issues.apache.org/jira/browse/ARROW-4297
- - JOB: "MinGW32"
- MINGW_PACKAGE_PREFIX: mingw-w64-i686
- MINGW_PREFIX: c:\msys64\mingw32
- MSYSTEM: MINGW32
- USE_CLCACHE: false
 
 environment:
  global:

ci/appveyor-cpp-build-mingw.bat

@@ -44,8 +44,11 @@ cmake ^
  -DARROW_PYTHON=ON ^
  -DPythonInterp_FIND_VERSION=ON ^
  -DPythonInterp_FIND_VERSION_MAJOR=3 ^
+ -DARROW_BUILD_TESTS=ON ^
+ -DARROW_PYTHON=OFF ^
  .. || exit /B
 make -j4 || exit /B
+ctest --output-on-failure -j2 || exit /B
 make install || exit /B
 popd
 

ci/appveyor-cpp-setup-mingw.bat

@@ -25,6 +25,7 @@ pacman -S --noconfirm ^
  "%MINGW_PACKAGE_PREFIX%-cmake" ^
  "%MINGW_PACKAGE_PREFIX%-flatbuffers" ^
  "%MINGW_PACKAGE_PREFIX%-gcc" ^
+ "%MINGW_PACKAGE_PREFIX%-gflags" ^
  "%MINGW_PACKAGE_PREFIX%-gobject-introspection" ^
  "%MINGW_PACKAGE_PREFIX%-gtk-doc" ^
  "%MINGW_PACKAGE_PREFIX%-lz4" ^

cpp/src/arrow/CMakeLists.txt

@@ -301,13 +301,24 @@ arrow_add_pkg_config("arrow")
 #
 
 add_arrow_test(allocator-test)
-add_arrow_test(array-test
- SOURCES
- array-test.cc
- array-binary-test.cc
- array-dict-test.cc
- array-list-test.cc
- array-struct-test.cc)
+
+if(WIN32)
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+else()
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-dict-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+endif()
+
 add_arrow_test(buffer-test)
 
 if(ARROW_IPC)

cpp/src/arrow/compute/kernels/cast-test.cc

@@ -424,6 +424,7 @@ TEST_F(TestCast, FloatingPointToInt) {
  options);
 }
 
+#if ARROW_BITNESS >= 64
 TEST_F(TestCast, IntToFloatingPoint) {
  auto options = CastOptions::Safe();
 
@@ -438,6 +439,7 @@ TEST_F(TestCast, IntToFloatingPoint) {
  UnsafeVectorCast<double, int64_t>(v1),
  options);
 }
+#endif
 
 TEST_F(TestCast, TimestampToTimestamp) {
  CastOptions options;

cpp/src/arrow/util/hash-util.h

@@ -81,10 +81,13 @@ class HashUtil {
  const uint8_t* p = reinterpret_cast<const uint8_t*>(data);
  const uint8_t* end = p + nbytes;
 
+#if ARROW_BITNESS >= 64
  while (p <= end - 8) {
  hash = HW_crc32_u64(hash, *reinterpret_cast<const uint64_t*>(p));
  p += 8;
  }
+#endif
+
  while (p <= end - 4) {
  hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
  p += 4;
@@ -113,6 +116,7 @@ class HashUtil {
  uint32_t h1 = static_cast<uint32_t>(hash >> 32);
  uint32_t h2 = static_cast<uint32_t>(hash);
 
+#if ARROW_BITNESS >= 64
  while (nbytes >= 16) {
  h1 = HW_crc32_u64(h1, *reinterpret_cast<const uint64_t*>(p));
  h2 = HW_crc32_u64(h2, *reinterpret_cast<const uint64_t*>(p + 8));
@@ -125,6 +129,15 @@ class HashUtil {
  nbytes -= 8;
  p += 8;
  }
+#else
+ while (nbytes >= 8) {
+ h1 = HW_crc32_u32(h1, *reinterpret_cast<const uint32_t*>(p));
+ h2 = HW_crc32_u32(h2, *reinterpret_cast<const uint32_t*>(p + 4));
+ nbytes -= 8;
+ p += 8;
+ }
+#endif
+
  if (nbytes >= 4) {
  h1 = HW_crc32_u16(h1, *reinterpret_cast<const uint16_t*>(p));
  h2 = HW_crc32_u16(h2, *reinterpret_cast<const uint16_t*>(p + 2));
@@ -151,58 +164,6 @@ class HashUtil {
  return (static_cast<uint64_t>(h1) << 32) + h2;
  }
 
- /// CrcHash() specialized for 1-byte data
- static inline uint32_t CrcHash1(const void* v, uint32_t hash) {
- const uint8_t* s = reinterpret_cast<const uint8_t*>(v);
- hash = HW_crc32_u8(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 2-byte data
- static inline uint32_t CrcHash2(const void* v, uint32_t hash) {
- const uint16_t* s = reinterpret_cast<const uint16_t*>(v);
- hash = HW_crc32_u16(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 4-byte data
- static inline uint32_t CrcHash4(const void* v, uint32_t hash) {
- const uint32_t* p = reinterpret_cast<const uint32_t*>(v);
- hash = HW_crc32_u32(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 8-byte data
- static inline uint32_t CrcHash8(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 12-byte data
- static inline uint32_t CrcHash12(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 16-byte data
- static inline uint32_t CrcHash16(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
  static const uint64_t MURMUR_PRIME = 0xc6a4a7935bd1e995;
  static const int MURMUR_R = 47;
 

cpp/src/arrow/util/macros.h

@@ -18,6 +18,8 @@
 #ifndef ARROW_UTIL_MACROS_H
 #define ARROW_UTIL_MACROS_H
 
+#include <cstdint>
+
 #define ARROW_STRINGIFY(x) #x
 #define ARROW_CONCAT(x, y) x##y
 
@@ -123,6 +125,17 @@
 #define ARROW_DISABLE_UBSAN(feature)
 #endif
 
+// ----------------------------------------------------------------------
+// Machine information
+
+#if INTPTR_MAX == INT64_MAX
+#define ARROW_BITNESS 64
+#elif INTPTR_MAX == INT32_MAX
+#define ARROW_BITNESS 32
+#else
+#error Unexpected INTPTR_MAX
+#endif
+
 // ----------------------------------------------------------------------
 // From googletest
 // (also in parquet-cpp)

cpp/src/arrow/util/sse-util.h

@@ -110,7 +110,11 @@ static inline uint32_t SSE4_crc32_u32(uint32_t crc, uint32_t v) {
 }
 
 static inline uint32_t SSE4_crc32_u64(uint32_t crc, uint64_t v) {
+#if ARROW_BITNESS == 32
+ return 0;
+#else
  return static_cast<uint32_t>(_mm_crc32_u64(crc, v));
+#endif
 }
 
 #else // without SSE 4.2.

appveyor.yml

@@ -36,14 +36,6 @@ cache:
 
 matrix:
  fast_finish: true
- allow_failures:
- # Can't build with 32-bit MinGW for now.
- # See https://issues.apache.org/jira/browse/ARROW-4297
- - JOB: "MinGW32"
- MINGW_PACKAGE_PREFIX: mingw-w64-i686
- MINGW_PREFIX: c:\msys64\mingw32
- MSYSTEM: MINGW32
- USE_CLCACHE: false
 
 environment:
  global:

ci/appveyor-cpp-build-mingw.bat

@@ -44,8 +44,11 @@ cmake ^
  -DARROW_PYTHON=ON ^
  -DPythonInterp_FIND_VERSION=ON ^
  -DPythonInterp_FIND_VERSION_MAJOR=3 ^
+ -DARROW_BUILD_TESTS=ON ^
+ -DARROW_PYTHON=OFF ^
  .. || exit /B
 make -j4 || exit /B
+ctest --output-on-failure -j2 || exit /B
 make install || exit /B
 popd
 

ci/appveyor-cpp-setup-mingw.bat

@@ -25,6 +25,7 @@ pacman -S --noconfirm ^
  "%MINGW_PACKAGE_PREFIX%-cmake" ^
  "%MINGW_PACKAGE_PREFIX%-flatbuffers" ^
  "%MINGW_PACKAGE_PREFIX%-gcc" ^
+ "%MINGW_PACKAGE_PREFIX%-gflags" ^
  "%MINGW_PACKAGE_PREFIX%-gobject-introspection" ^
  "%MINGW_PACKAGE_PREFIX%-gtk-doc" ^
  "%MINGW_PACKAGE_PREFIX%-lz4" ^

cpp/src/arrow/CMakeLists.txt

@@ -301,13 +301,24 @@ arrow_add_pkg_config("arrow")
 #
 
 add_arrow_test(allocator-test)
-add_arrow_test(array-test
- SOURCES
- array-test.cc
- array-binary-test.cc
- array-dict-test.cc
- array-list-test.cc
- array-struct-test.cc)
+
+if(WIN32)
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+else()
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-dict-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+endif()
+
 add_arrow_test(buffer-test)
 
 if(ARROW_IPC)

cpp/src/arrow/compute/kernels/cast-test.cc

@@ -424,6 +424,7 @@ TEST_F(TestCast, FloatingPointToInt) {
  options);
 }
 
+#if ARROW_BITNESS >= 64
 TEST_F(TestCast, IntToFloatingPoint) {
  auto options = CastOptions::Safe();
 
@@ -438,6 +439,7 @@ TEST_F(TestCast, IntToFloatingPoint) {
  UnsafeVectorCast<double, int64_t>(v1),
  options);
 }
+#endif
 
 TEST_F(TestCast, TimestampToTimestamp) {
  CastOptions options;

cpp/src/arrow/util/hash-util.h

@@ -81,10 +81,13 @@ class HashUtil {
  const uint8_t* p = reinterpret_cast<const uint8_t*>(data);
  const uint8_t* end = p + nbytes;
 
+#if ARROW_BITNESS >= 64
  while (p <= end - 8) {
  hash = HW_crc32_u64(hash, *reinterpret_cast<const uint64_t*>(p));
  p += 8;
  }
+#endif
+
  while (p <= end - 4) {
  hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
  p += 4;
@@ -113,6 +116,7 @@ class HashUtil {
  uint32_t h1 = static_cast<uint32_t>(hash >> 32);
  uint32_t h2 = static_cast<uint32_t>(hash);
 
+#if ARROW_BITNESS >= 64
  while (nbytes >= 16) {
  h1 = HW_crc32_u64(h1, *reinterpret_cast<const uint64_t*>(p));
  h2 = HW_crc32_u64(h2, *reinterpret_cast<const uint64_t*>(p + 8));
@@ -125,6 +129,15 @@ class HashUtil {
  nbytes -= 8;
  p += 8;
  }
+#else
+ while (nbytes >= 8) {
+ h1 = HW_crc32_u32(h1, *reinterpret_cast<const uint32_t*>(p));
+ h2 = HW_crc32_u32(h2, *reinterpret_cast<const uint32_t*>(p + 4));
+ nbytes -= 8;
+ p += 8;
+ }
+#endif
+
  if (nbytes >= 4) {
  h1 = HW_crc32_u16(h1, *reinterpret_cast<const uint16_t*>(p));
  h2 = HW_crc32_u16(h2, *reinterpret_cast<const uint16_t*>(p + 2));
@@ -151,58 +164,6 @@ class HashUtil {
  return (static_cast<uint64_t>(h1) << 32) + h2;
  }
 
- /// CrcHash() specialized for 1-byte data
- static inline uint32_t CrcHash1(const void* v, uint32_t hash) {
- const uint8_t* s = reinterpret_cast<const uint8_t*>(v);
- hash = HW_crc32_u8(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 2-byte data
- static inline uint32_t CrcHash2(const void* v, uint32_t hash) {
- const uint16_t* s = reinterpret_cast<const uint16_t*>(v);
- hash = HW_crc32_u16(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 4-byte data
- static inline uint32_t CrcHash4(const void* v, uint32_t hash) {
- const uint32_t* p = reinterpret_cast<const uint32_t*>(v);
- hash = HW_crc32_u32(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 8-byte data
- static inline uint32_t CrcHash8(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 12-byte data
- static inline uint32_t CrcHash12(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 16-byte data
- static inline uint32_t CrcHash16(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
  static const uint64_t MURMUR_PRIME = 0xc6a4a7935bd1e995;
  static const int MURMUR_R = 47;
 

cpp/src/arrow/util/macros.h

@@ -18,6 +18,8 @@
 #ifndef ARROW_UTIL_MACROS_H
 #define ARROW_UTIL_MACROS_H
 
+#include <cstdint>
+
 #define ARROW_STRINGIFY(x) #x
 #define ARROW_CONCAT(x, y) x##y
 
@@ -123,6 +125,17 @@
 #define ARROW_DISABLE_UBSAN(feature)
 #endif
 
+// ----------------------------------------------------------------------
+// Machine information
+
+#if INTPTR_MAX == INT64_MAX
+#define ARROW_BITNESS 64
+#elif INTPTR_MAX == INT32_MAX
+#define ARROW_BITNESS 32
+#else
+#error Unexpected INTPTR_MAX
+#endif
+
 // ----------------------------------------------------------------------
 // From googletest
 // (also in parquet-cpp)

cpp/src/arrow/util/sse-util.h

@@ -110,7 +110,11 @@ static inline uint32_t SSE4_crc32_u32(uint32_t crc, uint32_t v) {
 }
 
 static inline uint32_t SSE4_crc32_u64(uint32_t crc, uint64_t v) {
+#if ARROW_BITNESS == 32
+ return 0;
+#else
  return static_cast<uint32_t>(_mm_crc32_u64(crc, v));
+#endif
 }
 
 #else // without SSE 4.2.

appveyor.yml

@@ -36,14 +36,6 @@ cache:
 
 matrix:
  fast_finish: true
- allow_failures:
- # Can't build with 32-bit MinGW for now.
- # See https://issues.apache.org/jira/browse/ARROW-4297
- - JOB: "MinGW32"
- MINGW_PACKAGE_PREFIX: mingw-w64-i686
- MINGW_PREFIX: c:\msys64\mingw32
- MSYSTEM: MINGW32
- USE_CLCACHE: false
 
 environment:
  global:

ci/appveyor-cpp-build-mingw.bat

@@ -44,8 +44,11 @@ cmake ^
  -DARROW_PYTHON=ON ^
  -DPythonInterp_FIND_VERSION=ON ^
  -DPythonInterp_FIND_VERSION_MAJOR=3 ^
+ -DARROW_BUILD_TESTS=ON ^
+ -DARROW_PYTHON=OFF ^
  .. || exit /B
 make -j4 || exit /B
+ctest --output-on-failure -j2 || exit /B
 make install || exit /B
 popd
 

ci/appveyor-cpp-setup-mingw.bat

@@ -25,6 +25,7 @@ pacman -S --noconfirm ^
  "%MINGW_PACKAGE_PREFIX%-cmake" ^
  "%MINGW_PACKAGE_PREFIX%-flatbuffers" ^
  "%MINGW_PACKAGE_PREFIX%-gcc" ^
+ "%MINGW_PACKAGE_PREFIX%-gflags" ^
  "%MINGW_PACKAGE_PREFIX%-gobject-introspection" ^
  "%MINGW_PACKAGE_PREFIX%-gtk-doc" ^
  "%MINGW_PACKAGE_PREFIX%-lz4" ^

cpp/src/arrow/CMakeLists.txt

@@ -301,13 +301,24 @@ arrow_add_pkg_config("arrow")
 #
 
 add_arrow_test(allocator-test)
-add_arrow_test(array-test
- SOURCES
- array-test.cc
- array-binary-test.cc
- array-dict-test.cc
- array-list-test.cc
- array-struct-test.cc)
+
+if(WIN32)
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+else()
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-dict-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+endif()
+
 add_arrow_test(buffer-test)
 
 if(ARROW_IPC)

cpp/src/arrow/compute/kernels/cast-test.cc

@@ -424,6 +424,7 @@ TEST_F(TestCast, FloatingPointToInt) {
  options);
 }
 
+#if ARROW_BITNESS >= 64
 TEST_F(TestCast, IntToFloatingPoint) {
  auto options = CastOptions::Safe();
 
@@ -438,6 +439,7 @@ TEST_F(TestCast, IntToFloatingPoint) {
  UnsafeVectorCast<double, int64_t>(v1),
  options);
 }
+#endif
 
 TEST_F(TestCast, TimestampToTimestamp) {
  CastOptions options;

cpp/src/arrow/util/hash-util.h

@@ -81,10 +81,13 @@ class HashUtil {
  const uint8_t* p = reinterpret_cast<const uint8_t*>(data);
  const uint8_t* end = p + nbytes;
 
+#if ARROW_BITNESS >= 64
  while (p <= end - 8) {
  hash = HW_crc32_u64(hash, *reinterpret_cast<const uint64_t*>(p));
  p += 8;
  }
+#endif
+
  while (p <= end - 4) {
  hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
  p += 4;
@@ -113,6 +116,7 @@ class HashUtil {
  uint32_t h1 = static_cast<uint32_t>(hash >> 32);
  uint32_t h2 = static_cast<uint32_t>(hash);
 
+#if ARROW_BITNESS >= 64
  while (nbytes >= 16) {
  h1 = HW_crc32_u64(h1, *reinterpret_cast<const uint64_t*>(p));
  h2 = HW_crc32_u64(h2, *reinterpret_cast<const uint64_t*>(p + 8));
@@ -125,6 +129,15 @@ class HashUtil {
  nbytes -= 8;
  p += 8;
  }
+#else
+ while (nbytes >= 8) {
+ h1 = HW_crc32_u32(h1, *reinterpret_cast<const uint32_t*>(p));
+ h2 = HW_crc32_u32(h2, *reinterpret_cast<const uint32_t*>(p + 4));
+ nbytes -= 8;
+ p += 8;
+ }
+#endif
+
  if (nbytes >= 4) {
  h1 = HW_crc32_u16(h1, *reinterpret_cast<const uint16_t*>(p));
  h2 = HW_crc32_u16(h2, *reinterpret_cast<const uint16_t*>(p + 2));
@@ -151,58 +164,6 @@ class HashUtil {
  return (static_cast<uint64_t>(h1) << 32) + h2;
  }
 
- /// CrcHash() specialized for 1-byte data
- static inline uint32_t CrcHash1(const void* v, uint32_t hash) {
- const uint8_t* s = reinterpret_cast<const uint8_t*>(v);
- hash = HW_crc32_u8(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 2-byte data
- static inline uint32_t CrcHash2(const void* v, uint32_t hash) {
- const uint16_t* s = reinterpret_cast<const uint16_t*>(v);
- hash = HW_crc32_u16(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 4-byte data
- static inline uint32_t CrcHash4(const void* v, uint32_t hash) {
- const uint32_t* p = reinterpret_cast<const uint32_t*>(v);
- hash = HW_crc32_u32(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 8-byte data
- static inline uint32_t CrcHash8(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 12-byte data
- static inline uint32_t CrcHash12(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 16-byte data
- static inline uint32_t CrcHash16(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
  static const uint64_t MURMUR_PRIME = 0xc6a4a7935bd1e995;
  static const int MURMUR_R = 47;
 

cpp/src/arrow/util/macros.h

@@ -18,6 +18,8 @@
 #ifndef ARROW_UTIL_MACROS_H
 #define ARROW_UTIL_MACROS_H
 
+#include <cstdint>
+
 #define ARROW_STRINGIFY(x) #x
 #define ARROW_CONCAT(x, y) x##y
 
@@ -123,6 +125,17 @@
 #define ARROW_DISABLE_UBSAN(feature)
 #endif
 
+// ----------------------------------------------------------------------
+// Machine information
+
+#if INTPTR_MAX == INT64_MAX
+#define ARROW_BITNESS 64
+#elif INTPTR_MAX == INT32_MAX
+#define ARROW_BITNESS 32
+#else
+#error Unexpected INTPTR_MAX
+#endif
+
 // ----------------------------------------------------------------------
 // From googletest
 // (also in parquet-cpp)

cpp/src/arrow/util/sse-util.h

@@ -110,7 +110,11 @@ static inline uint32_t SSE4_crc32_u32(uint32_t crc, uint32_t v) {
 }
 
 static inline uint32_t SSE4_crc32_u64(uint32_t crc, uint64_t v) {
+#if ARROW_BITNESS == 32
+ return 0;
+#else
  return static_cast<uint32_t>(_mm_crc32_u64(crc, v));
+#endif
 }
 
 #else // without SSE 4.2.

appveyor.yml

@@ -36,14 +36,6 @@ cache:
 
 matrix:
  fast_finish: true
- allow_failures:
- # Can't build with 32-bit MinGW for now.
- # See https://issues.apache.org/jira/browse/ARROW-4297
- - JOB: "MinGW32"
- MINGW_PACKAGE_PREFIX: mingw-w64-i686
- MINGW_PREFIX: c:\msys64\mingw32
- MSYSTEM: MINGW32
- USE_CLCACHE: false
 
 environment:
  global:

ci/appveyor-cpp-build-mingw.bat

@@ -44,8 +44,11 @@ cmake ^
  -DARROW_PYTHON=ON ^
  -DPythonInterp_FIND_VERSION=ON ^
  -DPythonInterp_FIND_VERSION_MAJOR=3 ^
+ -DARROW_BUILD_TESTS=ON ^
+ -DARROW_PYTHON=OFF ^
  .. || exit /B
 make -j4 || exit /B
+ctest --output-on-failure -j2 || exit /B
 make install || exit /B
 popd
 

ci/appveyor-cpp-setup-mingw.bat

@@ -25,6 +25,7 @@ pacman -S --noconfirm ^
  "%MINGW_PACKAGE_PREFIX%-cmake" ^
  "%MINGW_PACKAGE_PREFIX%-flatbuffers" ^
  "%MINGW_PACKAGE_PREFIX%-gcc" ^
+ "%MINGW_PACKAGE_PREFIX%-gflags" ^
  "%MINGW_PACKAGE_PREFIX%-gobject-introspection" ^
  "%MINGW_PACKAGE_PREFIX%-gtk-doc" ^
  "%MINGW_PACKAGE_PREFIX%-lz4" ^

cpp/src/arrow/CMakeLists.txt

@@ -301,13 +301,24 @@ arrow_add_pkg_config("arrow")
 #
 
 add_arrow_test(allocator-test)
-add_arrow_test(array-test
- SOURCES
- array-test.cc
- array-binary-test.cc
- array-dict-test.cc
- array-list-test.cc
- array-struct-test.cc)
+
+if(WIN32)
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+else()
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-dict-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+endif()
+
 add_arrow_test(buffer-test)
 
 if(ARROW_IPC)

cpp/src/arrow/compute/kernels/cast-test.cc

@@ -424,6 +424,7 @@ TEST_F(TestCast, FloatingPointToInt) {
  options);
 }
 
+#if ARROW_BITNESS >= 64
 TEST_F(TestCast, IntToFloatingPoint) {
  auto options = CastOptions::Safe();
 
@@ -438,6 +439,7 @@ TEST_F(TestCast, IntToFloatingPoint) {
  UnsafeVectorCast<double, int64_t>(v1),
  options);
 }
+#endif
 
 TEST_F(TestCast, TimestampToTimestamp) {
  CastOptions options;

cpp/src/arrow/util/hash-util.h

@@ -81,10 +81,13 @@ class HashUtil {
  const uint8_t* p = reinterpret_cast<const uint8_t*>(data);
  const uint8_t* end = p + nbytes;
 
+#if ARROW_BITNESS >= 64
  while (p <= end - 8) {
  hash = HW_crc32_u64(hash, *reinterpret_cast<const uint64_t*>(p));
  p += 8;
  }
+#endif
+
  while (p <= end - 4) {
  hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
  p += 4;
@@ -113,6 +116,7 @@ class HashUtil {
  uint32_t h1 = static_cast<uint32_t>(hash >> 32);
  uint32_t h2 = static_cast<uint32_t>(hash);
 
+#if ARROW_BITNESS >= 64
  while (nbytes >= 16) {
  h1 = HW_crc32_u64(h1, *reinterpret_cast<const uint64_t*>(p));
  h2 = HW_crc32_u64(h2, *reinterpret_cast<const uint64_t*>(p + 8));
@@ -125,6 +129,15 @@ class HashUtil {
  nbytes -= 8;
  p += 8;
  }
+#else
+ while (nbytes >= 8) {
+ h1 = HW_crc32_u32(h1, *reinterpret_cast<const uint32_t*>(p));
+ h2 = HW_crc32_u32(h2, *reinterpret_cast<const uint32_t*>(p + 4));
+ nbytes -= 8;
+ p += 8;
+ }
+#endif
+
  if (nbytes >= 4) {
  h1 = HW_crc32_u16(h1, *reinterpret_cast<const uint16_t*>(p));
  h2 = HW_crc32_u16(h2, *reinterpret_cast<const uint16_t*>(p + 2));
@@ -151,58 +164,6 @@ class HashUtil {
  return (static_cast<uint64_t>(h1) << 32) + h2;
  }
 
- /// CrcHash() specialized for 1-byte data
- static inline uint32_t CrcHash1(const void* v, uint32_t hash) {
- const uint8_t* s = reinterpret_cast<const uint8_t*>(v);
- hash = HW_crc32_u8(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 2-byte data
- static inline uint32_t CrcHash2(const void* v, uint32_t hash) {
- const uint16_t* s = reinterpret_cast<const uint16_t*>(v);
- hash = HW_crc32_u16(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 4-byte data
- static inline uint32_t CrcHash4(const void* v, uint32_t hash) {
- const uint32_t* p = reinterpret_cast<const uint32_t*>(v);
- hash = HW_crc32_u32(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 8-byte data
- static inline uint32_t CrcHash8(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 12-byte data
- static inline uint32_t CrcHash12(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 16-byte data
- static inline uint32_t CrcHash16(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
  static const uint64_t MURMUR_PRIME = 0xc6a4a7935bd1e995;
  static const int MURMUR_R = 47;
 

cpp/src/arrow/util/macros.h

@@ -18,6 +18,8 @@
 #ifndef ARROW_UTIL_MACROS_H
 #define ARROW_UTIL_MACROS_H
 
+#include <cstdint>
+
 #define ARROW_STRINGIFY(x) #x
 #define ARROW_CONCAT(x, y) x##y
 
@@ -123,6 +125,17 @@
 #define ARROW_DISABLE_UBSAN(feature)
 #endif
 
+// ----------------------------------------------------------------------
+// Machine information
+
+#if INTPTR_MAX == INT64_MAX
+#define ARROW_BITNESS 64
+#elif INTPTR_MAX == INT32_MAX
+#define ARROW_BITNESS 32
+#else
+#error Unexpected INTPTR_MAX
+#endif
+
 // ----------------------------------------------------------------------
 // From googletest
 // (also in parquet-cpp)

cpp/src/arrow/util/sse-util.h

@@ -110,7 +110,11 @@ static inline uint32_t SSE4_crc32_u32(uint32_t crc, uint32_t v) {
 }
 
 static inline uint32_t SSE4_crc32_u64(uint32_t crc, uint64_t v) {
+#if ARROW_BITNESS == 32
+ return 0;
+#else
  return static_cast<uint32_t>(_mm_crc32_u64(crc, v));
+#endif
 }
 
 #else // without SSE 4.2.

appveyor.yml

@@ -36,14 +36,6 @@ cache:
 
 matrix:
  fast_finish: true
- allow_failures:
- # Can't build with 32-bit MinGW for now.
- # See https://issues.apache.org/jira/browse/ARROW-4297
- - JOB: "MinGW32"
- MINGW_PACKAGE_PREFIX: mingw-w64-i686
- MINGW_PREFIX: c:\msys64\mingw32
- MSYSTEM: MINGW32
- USE_CLCACHE: false
 
 environment:
  global:

ci/appveyor-cpp-build-mingw.bat

@@ -44,8 +44,11 @@ cmake ^
  -DARROW_PYTHON=ON ^
  -DPythonInterp_FIND_VERSION=ON ^
  -DPythonInterp_FIND_VERSION_MAJOR=3 ^
+ -DARROW_BUILD_TESTS=ON ^
+ -DARROW_PYTHON=OFF ^
  .. || exit /B
 make -j4 || exit /B
+ctest --output-on-failure -j2 || exit /B
 make install || exit /B
 popd
 

ci/appveyor-cpp-setup-mingw.bat

@@ -25,6 +25,7 @@ pacman -S --noconfirm ^
  "%MINGW_PACKAGE_PREFIX%-cmake" ^
  "%MINGW_PACKAGE_PREFIX%-flatbuffers" ^
  "%MINGW_PACKAGE_PREFIX%-gcc" ^
+ "%MINGW_PACKAGE_PREFIX%-gflags" ^
  "%MINGW_PACKAGE_PREFIX%-gobject-introspection" ^
  "%MINGW_PACKAGE_PREFIX%-gtk-doc" ^
  "%MINGW_PACKAGE_PREFIX%-lz4" ^

cpp/src/arrow/CMakeLists.txt

@@ -301,13 +301,24 @@ arrow_add_pkg_config("arrow")
 #
 
 add_arrow_test(allocator-test)
-add_arrow_test(array-test
- SOURCES
- array-test.cc
- array-binary-test.cc
- array-dict-test.cc
- array-list-test.cc
- array-struct-test.cc)
+
+if(WIN32)
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+else()
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-dict-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+endif()
+
 add_arrow_test(buffer-test)
 
 if(ARROW_IPC)

cpp/src/arrow/compute/kernels/cast-test.cc

@@ -424,6 +424,7 @@ TEST_F(TestCast, FloatingPointToInt) {
  options);
 }
 
+#if ARROW_BITNESS >= 64
 TEST_F(TestCast, IntToFloatingPoint) {
  auto options = CastOptions::Safe();
 
@@ -438,6 +439,7 @@ TEST_F(TestCast, IntToFloatingPoint) {
  UnsafeVectorCast<double, int64_t>(v1),
  options);
 }
+#endif
 
 TEST_F(TestCast, TimestampToTimestamp) {
  CastOptions options;

cpp/src/arrow/util/hash-util.h

@@ -81,10 +81,13 @@ class HashUtil {
  const uint8_t* p = reinterpret_cast<const uint8_t*>(data);
  const uint8_t* end = p + nbytes;
 
+#if ARROW_BITNESS >= 64
  while (p <= end - 8) {
  hash = HW_crc32_u64(hash, *reinterpret_cast<const uint64_t*>(p));
  p += 8;
  }
+#endif
+
  while (p <= end - 4) {
  hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
  p += 4;
@@ -113,6 +116,7 @@ class HashUtil {
  uint32_t h1 = static_cast<uint32_t>(hash >> 32);
  uint32_t h2 = static_cast<uint32_t>(hash);
 
+#if ARROW_BITNESS >= 64
  while (nbytes >= 16) {
  h1 = HW_crc32_u64(h1, *reinterpret_cast<const uint64_t*>(p));
  h2 = HW_crc32_u64(h2, *reinterpret_cast<const uint64_t*>(p + 8));
@@ -125,6 +129,15 @@ class HashUtil {
  nbytes -= 8;
  p += 8;
  }
+#else
+ while (nbytes >= 8) {
+ h1 = HW_crc32_u32(h1, *reinterpret_cast<const uint32_t*>(p));
+ h2 = HW_crc32_u32(h2, *reinterpret_cast<const uint32_t*>(p + 4));
+ nbytes -= 8;
+ p += 8;
+ }
+#endif
+
  if (nbytes >= 4) {
  h1 = HW_crc32_u16(h1, *reinterpret_cast<const uint16_t*>(p));
  h2 = HW_crc32_u16(h2, *reinterpret_cast<const uint16_t*>(p + 2));
@@ -151,58 +164,6 @@ class HashUtil {
  return (static_cast<uint64_t>(h1) << 32) + h2;
  }
 
- /// CrcHash() specialized for 1-byte data
- static inline uint32_t CrcHash1(const void* v, uint32_t hash) {
- const uint8_t* s = reinterpret_cast<const uint8_t*>(v);
- hash = HW_crc32_u8(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 2-byte data
- static inline uint32_t CrcHash2(const void* v, uint32_t hash) {
- const uint16_t* s = reinterpret_cast<const uint16_t*>(v);
- hash = HW_crc32_u16(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 4-byte data
- static inline uint32_t CrcHash4(const void* v, uint32_t hash) {
- const uint32_t* p = reinterpret_cast<const uint32_t*>(v);
- hash = HW_crc32_u32(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 8-byte data
- static inline uint32_t CrcHash8(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 12-byte data
- static inline uint32_t CrcHash12(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 16-byte data
- static inline uint32_t CrcHash16(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
  static const uint64_t MURMUR_PRIME = 0xc6a4a7935bd1e995;
  static const int MURMUR_R = 47;
 

cpp/src/arrow/util/macros.h

@@ -18,6 +18,8 @@
 #ifndef ARROW_UTIL_MACROS_H
 #define ARROW_UTIL_MACROS_H
 
+#include <cstdint>
+
 #define ARROW_STRINGIFY(x) #x
 #define ARROW_CONCAT(x, y) x##y
 
@@ -123,6 +125,17 @@
 #define ARROW_DISABLE_UBSAN(feature)
 #endif
 
+// ----------------------------------------------------------------------
+// Machine information
+
+#if INTPTR_MAX == INT64_MAX
+#define ARROW_BITNESS 64
+#elif INTPTR_MAX == INT32_MAX
+#define ARROW_BITNESS 32
+#else
+#error Unexpected INTPTR_MAX
+#endif
+
 // ----------------------------------------------------------------------
 // From googletest
 // (also in parquet-cpp)

cpp/src/arrow/util/sse-util.h

@@ -110,7 +110,11 @@ static inline uint32_t SSE4_crc32_u32(uint32_t crc, uint32_t v) {
 }
 
 static inline uint32_t SSE4_crc32_u64(uint32_t crc, uint64_t v) {
+#if ARROW_BITNESS == 32
+ return 0;
+#else
  return static_cast<uint32_t>(_mm_crc32_u64(crc, v));
+#endif
 }
 
 #else // without SSE 4.2.

appveyor.yml

@@ -36,14 +36,6 @@ cache:
 
 matrix:
  fast_finish: true
- allow_failures:
- # Can't build with 32-bit MinGW for now.
- # See https://issues.apache.org/jira/browse/ARROW-4297
- - JOB: "MinGW32"
- MINGW_PACKAGE_PREFIX: mingw-w64-i686
- MINGW_PREFIX: c:\msys64\mingw32
- MSYSTEM: MINGW32
- USE_CLCACHE: false
 
 environment:
  global:

ci/appveyor-cpp-build-mingw.bat

@@ -44,8 +44,11 @@ cmake ^
  -DARROW_PYTHON=ON ^
  -DPythonInterp_FIND_VERSION=ON ^
  -DPythonInterp_FIND_VERSION_MAJOR=3 ^
+ -DARROW_BUILD_TESTS=ON ^
+ -DARROW_PYTHON=OFF ^
  .. || exit /B
 make -j4 || exit /B
+ctest --output-on-failure -j2 || exit /B
 make install || exit /B
 popd
 

ci/appveyor-cpp-setup-mingw.bat

@@ -25,6 +25,7 @@ pacman -S --noconfirm ^
  "%MINGW_PACKAGE_PREFIX%-cmake" ^
  "%MINGW_PACKAGE_PREFIX%-flatbuffers" ^
  "%MINGW_PACKAGE_PREFIX%-gcc" ^
+ "%MINGW_PACKAGE_PREFIX%-gflags" ^
  "%MINGW_PACKAGE_PREFIX%-gobject-introspection" ^
  "%MINGW_PACKAGE_PREFIX%-gtk-doc" ^
  "%MINGW_PACKAGE_PREFIX%-lz4" ^

cpp/src/arrow/CMakeLists.txt

@@ -301,13 +301,24 @@ arrow_add_pkg_config("arrow")
 #
 
 add_arrow_test(allocator-test)
-add_arrow_test(array-test
- SOURCES
- array-test.cc
- array-binary-test.cc
- array-dict-test.cc
- array-list-test.cc
- array-struct-test.cc)
+
+if(WIN32)
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+else()
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-dict-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+endif()
+
 add_arrow_test(buffer-test)
 
 if(ARROW_IPC)

cpp/src/arrow/compute/kernels/cast-test.cc

@@ -424,6 +424,7 @@ TEST_F(TestCast, FloatingPointToInt) {
  options);
 }
 
+#if ARROW_BITNESS >= 64
 TEST_F(TestCast, IntToFloatingPoint) {
  auto options = CastOptions::Safe();
 
@@ -438,6 +439,7 @@ TEST_F(TestCast, IntToFloatingPoint) {
  UnsafeVectorCast<double, int64_t>(v1),
  options);
 }
+#endif
 
 TEST_F(TestCast, TimestampToTimestamp) {
  CastOptions options;

cpp/src/arrow/util/hash-util.h

@@ -81,10 +81,13 @@ class HashUtil {
  const uint8_t* p = reinterpret_cast<const uint8_t*>(data);
  const uint8_t* end = p + nbytes;
 
+#if ARROW_BITNESS >= 64
  while (p <= end - 8) {
  hash = HW_crc32_u64(hash, *reinterpret_cast<const uint64_t*>(p));
  p += 8;
  }
+#endif
+
  while (p <= end - 4) {
  hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
  p += 4;
@@ -113,6 +116,7 @@ class HashUtil {
  uint32_t h1 = static_cast<uint32_t>(hash >> 32);
  uint32_t h2 = static_cast<uint32_t>(hash);
 
+#if ARROW_BITNESS >= 64
  while (nbytes >= 16) {
  h1 = HW_crc32_u64(h1, *reinterpret_cast<const uint64_t*>(p));
  h2 = HW_crc32_u64(h2, *reinterpret_cast<const uint64_t*>(p + 8));
@@ -125,6 +129,15 @@ class HashUtil {
  nbytes -= 8;
  p += 8;
  }
+#else
+ while (nbytes >= 8) {
+ h1 = HW_crc32_u32(h1, *reinterpret_cast<const uint32_t*>(p));
+ h2 = HW_crc32_u32(h2, *reinterpret_cast<const uint32_t*>(p + 4));
+ nbytes -= 8;
+ p += 8;
+ }
+#endif
+
  if (nbytes >= 4) {
  h1 = HW_crc32_u16(h1, *reinterpret_cast<const uint16_t*>(p));
  h2 = HW_crc32_u16(h2, *reinterpret_cast<const uint16_t*>(p + 2));
@@ -151,58 +164,6 @@ class HashUtil {
  return (static_cast<uint64_t>(h1) << 32) + h2;
  }
 
- /// CrcHash() specialized for 1-byte data
- static inline uint32_t CrcHash1(const void* v, uint32_t hash) {
- const uint8_t* s = reinterpret_cast<const uint8_t*>(v);
- hash = HW_crc32_u8(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 2-byte data
- static inline uint32_t CrcHash2(const void* v, uint32_t hash) {
- const uint16_t* s = reinterpret_cast<const uint16_t*>(v);
- hash = HW_crc32_u16(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 4-byte data
- static inline uint32_t CrcHash4(const void* v, uint32_t hash) {
- const uint32_t* p = reinterpret_cast<const uint32_t*>(v);
- hash = HW_crc32_u32(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 8-byte data
- static inline uint32_t CrcHash8(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 12-byte data
- static inline uint32_t CrcHash12(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 16-byte data
- static inline uint32_t CrcHash16(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
  static const uint64_t MURMUR_PRIME = 0xc6a4a7935bd1e995;
  static const int MURMUR_R = 47;
 

cpp/src/arrow/util/macros.h

@@ -18,6 +18,8 @@
 #ifndef ARROW_UTIL_MACROS_H
 #define ARROW_UTIL_MACROS_H
 
+#include <cstdint>
+
 #define ARROW_STRINGIFY(x) #x
 #define ARROW_CONCAT(x, y) x##y
 
@@ -123,6 +125,17 @@
 #define ARROW_DISABLE_UBSAN(feature)
 #endif
 
+// ----------------------------------------------------------------------
+// Machine information
+
+#if INTPTR_MAX == INT64_MAX
+#define ARROW_BITNESS 64
+#elif INTPTR_MAX == INT32_MAX
+#define ARROW_BITNESS 32
+#else
+#error Unexpected INTPTR_MAX
+#endif
+
 // ----------------------------------------------------------------------
 // From googletest
 // (also in parquet-cpp)

cpp/src/arrow/util/sse-util.h

@@ -110,7 +110,11 @@ static inline uint32_t SSE4_crc32_u32(uint32_t crc, uint32_t v) {
 }
 
 static inline uint32_t SSE4_crc32_u64(uint32_t crc, uint64_t v) {
+#if ARROW_BITNESS == 32
+ return 0;
+#else
  return static_cast<uint32_t>(_mm_crc32_u64(crc, v));
+#endif
 }
 
 #else // without SSE 4.2.

appveyor.yml

@@ -36,14 +36,6 @@ cache:
 
 matrix:
  fast_finish: true
- allow_failures:
- # Can't build with 32-bit MinGW for now.
- # See https://issues.apache.org/jira/browse/ARROW-4297
- - JOB: "MinGW32"
- MINGW_PACKAGE_PREFIX: mingw-w64-i686
- MINGW_PREFIX: c:\msys64\mingw32
- MSYSTEM: MINGW32
- USE_CLCACHE: false
 
 environment:
  global:

ci/appveyor-cpp-build-mingw.bat

@@ -44,8 +44,11 @@ cmake ^
  -DARROW_PYTHON=ON ^
  -DPythonInterp_FIND_VERSION=ON ^
  -DPythonInterp_FIND_VERSION_MAJOR=3 ^
+ -DARROW_BUILD_TESTS=ON ^
+ -DARROW_PYTHON=OFF ^
  .. || exit /B
 make -j4 || exit /B
+ctest --output-on-failure -j2 || exit /B
 make install || exit /B
 popd
 

ci/appveyor-cpp-setup-mingw.bat

@@ -25,6 +25,7 @@ pacman -S --noconfirm ^
  "%MINGW_PACKAGE_PREFIX%-cmake" ^
  "%MINGW_PACKAGE_PREFIX%-flatbuffers" ^
  "%MINGW_PACKAGE_PREFIX%-gcc" ^
+ "%MINGW_PACKAGE_PREFIX%-gflags" ^
  "%MINGW_PACKAGE_PREFIX%-gobject-introspection" ^
  "%MINGW_PACKAGE_PREFIX%-gtk-doc" ^
  "%MINGW_PACKAGE_PREFIX%-lz4" ^

cpp/src/arrow/CMakeLists.txt

@@ -301,13 +301,24 @@ arrow_add_pkg_config("arrow")
 #
 
 add_arrow_test(allocator-test)
-add_arrow_test(array-test
- SOURCES
- array-test.cc
- array-binary-test.cc
- array-dict-test.cc
- array-list-test.cc
- array-struct-test.cc)
+
+if(WIN32)
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+else()
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-dict-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+endif()
+
 add_arrow_test(buffer-test)
 
 if(ARROW_IPC)

cpp/src/arrow/compute/kernels/cast-test.cc

@@ -424,6 +424,7 @@ TEST_F(TestCast, FloatingPointToInt) {
  options);
 }
 
+#if ARROW_BITNESS >= 64
 TEST_F(TestCast, IntToFloatingPoint) {
  auto options = CastOptions::Safe();
 
@@ -438,6 +439,7 @@ TEST_F(TestCast, IntToFloatingPoint) {
  UnsafeVectorCast<double, int64_t>(v1),
  options);
 }
+#endif
 
 TEST_F(TestCast, TimestampToTimestamp) {
  CastOptions options;

cpp/src/arrow/util/hash-util.h

@@ -81,10 +81,13 @@ class HashUtil {
  const uint8_t* p = reinterpret_cast<const uint8_t*>(data);
  const uint8_t* end = p + nbytes;
 
+#if ARROW_BITNESS >= 64
  while (p <= end - 8) {
  hash = HW_crc32_u64(hash, *reinterpret_cast<const uint64_t*>(p));
  p += 8;
  }
+#endif
+
  while (p <= end - 4) {
  hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
  p += 4;
@@ -113,6 +116,7 @@ class HashUtil {
  uint32_t h1 = static_cast<uint32_t>(hash >> 32);
  uint32_t h2 = static_cast<uint32_t>(hash);
 
+#if ARROW_BITNESS >= 64
  while (nbytes >= 16) {
  h1 = HW_crc32_u64(h1, *reinterpret_cast<const uint64_t*>(p));
  h2 = HW_crc32_u64(h2, *reinterpret_cast<const uint64_t*>(p + 8));
@@ -125,6 +129,15 @@ class HashUtil {
  nbytes -= 8;
  p += 8;
  }
+#else
+ while (nbytes >= 8) {
+ h1 = HW_crc32_u32(h1, *reinterpret_cast<const uint32_t*>(p));
+ h2 = HW_crc32_u32(h2, *reinterpret_cast<const uint32_t*>(p + 4));
+ nbytes -= 8;
+ p += 8;
+ }
+#endif
+
  if (nbytes >= 4) {
  h1 = HW_crc32_u16(h1, *reinterpret_cast<const uint16_t*>(p));
  h2 = HW_crc32_u16(h2, *reinterpret_cast<const uint16_t*>(p + 2));
@@ -151,58 +164,6 @@ class HashUtil {
  return (static_cast<uint64_t>(h1) << 32) + h2;
  }
 
- /// CrcHash() specialized for 1-byte data
- static inline uint32_t CrcHash1(const void* v, uint32_t hash) {
- const uint8_t* s = reinterpret_cast<const uint8_t*>(v);
- hash = HW_crc32_u8(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 2-byte data
- static inline uint32_t CrcHash2(const void* v, uint32_t hash) {
- const uint16_t* s = reinterpret_cast<const uint16_t*>(v);
- hash = HW_crc32_u16(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 4-byte data
- static inline uint32_t CrcHash4(const void* v, uint32_t hash) {
- const uint32_t* p = reinterpret_cast<const uint32_t*>(v);
- hash = HW_crc32_u32(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 8-byte data
- static inline uint32_t CrcHash8(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 12-byte data
- static inline uint32_t CrcHash12(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 16-byte data
- static inline uint32_t CrcHash16(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
  static const uint64_t MURMUR_PRIME = 0xc6a4a7935bd1e995;
  static const int MURMUR_R = 47;
 

cpp/src/arrow/util/macros.h

@@ -18,6 +18,8 @@
 #ifndef ARROW_UTIL_MACROS_H
 #define ARROW_UTIL_MACROS_H
 
+#include <cstdint>
+
 #define ARROW_STRINGIFY(x) #x
 #define ARROW_CONCAT(x, y) x##y
 
@@ -123,6 +125,17 @@
 #define ARROW_DISABLE_UBSAN(feature)
 #endif
 
+// ----------------------------------------------------------------------
+// Machine information
+
+#if INTPTR_MAX == INT64_MAX
+#define ARROW_BITNESS 64
+#elif INTPTR_MAX == INT32_MAX
+#define ARROW_BITNESS 32
+#else
+#error Unexpected INTPTR_MAX
+#endif
+
 // ----------------------------------------------------------------------
 // From googletest
 // (also in parquet-cpp)

cpp/src/arrow/util/sse-util.h

@@ -110,7 +110,11 @@ static inline uint32_t SSE4_crc32_u32(uint32_t crc, uint32_t v) {
 }
 
 static inline uint32_t SSE4_crc32_u64(uint32_t crc, uint64_t v) {
+#if ARROW_BITNESS == 32
+ return 0;
+#else
  return static_cast<uint32_t>(_mm_crc32_u64(crc, v));
+#endif
 }
 
 #else // without SSE 4.2.

appveyor.yml

@@ -36,14 +36,6 @@ cache:
 
 matrix:
  fast_finish: true
- allow_failures:
- # Can't build with 32-bit MinGW for now.
- # See https://issues.apache.org/jira/browse/ARROW-4297
- - JOB: "MinGW32"
- MINGW_PACKAGE_PREFIX: mingw-w64-i686
- MINGW_PREFIX: c:\msys64\mingw32
- MSYSTEM: MINGW32
- USE_CLCACHE: false
 
 environment:
  global:

ci/appveyor-cpp-build-mingw.bat

@@ -44,8 +44,11 @@ cmake ^
  -DARROW_PYTHON=ON ^
  -DPythonInterp_FIND_VERSION=ON ^
  -DPythonInterp_FIND_VERSION_MAJOR=3 ^
+ -DARROW_BUILD_TESTS=ON ^
+ -DARROW_PYTHON=OFF ^
  .. || exit /B
 make -j4 || exit /B
+ctest --output-on-failure -j2 || exit /B
 make install || exit /B
 popd
 

ci/appveyor-cpp-setup-mingw.bat

@@ -25,6 +25,7 @@ pacman -S --noconfirm ^
  "%MINGW_PACKAGE_PREFIX%-cmake" ^
  "%MINGW_PACKAGE_PREFIX%-flatbuffers" ^
  "%MINGW_PACKAGE_PREFIX%-gcc" ^
+ "%MINGW_PACKAGE_PREFIX%-gflags" ^
  "%MINGW_PACKAGE_PREFIX%-gobject-introspection" ^
  "%MINGW_PACKAGE_PREFIX%-gtk-doc" ^
  "%MINGW_PACKAGE_PREFIX%-lz4" ^

cpp/src/arrow/CMakeLists.txt

@@ -301,13 +301,24 @@ arrow_add_pkg_config("arrow")
 #
 
 add_arrow_test(allocator-test)
-add_arrow_test(array-test
- SOURCES
- array-test.cc
- array-binary-test.cc
- array-dict-test.cc
- array-list-test.cc
- array-struct-test.cc)
+
+if(WIN32)
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+else()
+ add_arrow_test(array-test
+ SOURCES
+ array-test.cc
+ array-binary-test.cc
+ array-dict-test.cc
+ array-list-test.cc
+ array-struct-test.cc)
+endif()
+
 add_arrow_test(buffer-test)
 
 if(ARROW_IPC)

cpp/src/arrow/compute/kernels/cast-test.cc

@@ -424,6 +424,7 @@ TEST_F(TestCast, FloatingPointToInt) {
  options);
 }
 
+#if ARROW_BITNESS >= 64
 TEST_F(TestCast, IntToFloatingPoint) {
  auto options = CastOptions::Safe();
 
@@ -438,6 +439,7 @@ TEST_F(TestCast, IntToFloatingPoint) {
  UnsafeVectorCast<double, int64_t>(v1),
  options);
 }
+#endif
 
 TEST_F(TestCast, TimestampToTimestamp) {
  CastOptions options;

cpp/src/arrow/util/hash-util.h

@@ -81,10 +81,13 @@ class HashUtil {
  const uint8_t* p = reinterpret_cast<const uint8_t*>(data);
  const uint8_t* end = p + nbytes;
 
+#if ARROW_BITNESS >= 64
  while (p <= end - 8) {
  hash = HW_crc32_u64(hash, *reinterpret_cast<const uint64_t*>(p));
  p += 8;
  }
+#endif
+
  while (p <= end - 4) {
  hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
  p += 4;
@@ -113,6 +116,7 @@ class HashUtil {
  uint32_t h1 = static_cast<uint32_t>(hash >> 32);
  uint32_t h2 = static_cast<uint32_t>(hash);
 
+#if ARROW_BITNESS >= 64
  while (nbytes >= 16) {
  h1 = HW_crc32_u64(h1, *reinterpret_cast<const uint64_t*>(p));
  h2 = HW_crc32_u64(h2, *reinterpret_cast<const uint64_t*>(p + 8));
@@ -125,6 +129,15 @@ class HashUtil {
  nbytes -= 8;
  p += 8;
  }
+#else
+ while (nbytes >= 8) {
+ h1 = HW_crc32_u32(h1, *reinterpret_cast<const uint32_t*>(p));
+ h2 = HW_crc32_u32(h2, *reinterpret_cast<const uint32_t*>(p + 4));
+ nbytes -= 8;
+ p += 8;
+ }
+#endif
+
  if (nbytes >= 4) {
  h1 = HW_crc32_u16(h1, *reinterpret_cast<const uint16_t*>(p));
  h2 = HW_crc32_u16(h2, *reinterpret_cast<const uint16_t*>(p + 2));
@@ -151,58 +164,6 @@ class HashUtil {
  return (static_cast<uint64_t>(h1) << 32) + h2;
  }
 
- /// CrcHash() specialized for 1-byte data
- static inline uint32_t CrcHash1(const void* v, uint32_t hash) {
- const uint8_t* s = reinterpret_cast<const uint8_t*>(v);
- hash = HW_crc32_u8(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 2-byte data
- static inline uint32_t CrcHash2(const void* v, uint32_t hash) {
- const uint16_t* s = reinterpret_cast<const uint16_t*>(v);
- hash = HW_crc32_u16(hash, *s);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 4-byte data
- static inline uint32_t CrcHash4(const void* v, uint32_t hash) {
- const uint32_t* p = reinterpret_cast<const uint32_t*>(v);
- hash = HW_crc32_u32(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 8-byte data
- static inline uint32_t CrcHash8(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 12-byte data
- static inline uint32_t CrcHash12(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u32(hash, *reinterpret_cast<const uint32_t*>(p));
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
- /// CrcHash() specialized for 16-byte data
- static inline uint32_t CrcHash16(const void* v, uint32_t hash) {
- const uint64_t* p = reinterpret_cast<const uint64_t*>(v);
- hash = HW_crc32_u64(hash, *p);
- ++p;
- hash = HW_crc32_u64(hash, *p);
- hash = (hash << 16) | (hash >> 16);
- return hash;
- }
-
  static const uint64_t MURMUR_PRIME = 0xc6a4a7935bd1e995;
  static const int MURMUR_R = 47;
 

cpp/src/arrow/util/macros.h

@@ -18,6 +18,8 @@
 #ifndef ARROW_UTIL_MACROS_H
 #define ARROW_UTIL_MACROS_H
 
+#include <cstdint>
+
 #define ARROW_STRINGIFY(x) #x
 #define ARROW_CONCAT(x, y) x##y
 
@@ -123,6 +125,17 @@
 #define ARROW_DISABLE_UBSAN(feature)
 #endif
 
+// ----------------------------------------------------------------------
+// Machine information
+
+#if INTPTR_MAX == INT64_MAX
+#define ARROW_BITNESS 64
+#elif INTPTR_MAX == INT32_MAX
+#define ARROW_BITNESS 32
+#else
+#error Unexpected INTPTR_MAX
+#endif
+
 // ----------------------------------------------------------------------
 // From googletest
 // (also in parquet-cpp)

cpp/src/arrow/util/sse-util.h

@@ -110,7 +110,11 @@ static inline uint32_t SSE4_crc32_u32(uint32_t crc, uint32_t v) {
 }
 
 static inline uint32_t SSE4_crc32_u64(uint32_t crc, uint64_t v) {
+#if ARROW_BITNESS == 32
+ return 0;
+#else
  return static_cast<uint32_t>(_mm_crc32_u64(crc, v));
+#endif
 }
 
 #else // without SSE 4.2.