ARROW-17790: [C++][Gandiva] Adapt to LLVM opaque pointer

.github/workflows/cpp.yml

@@ -56,7 +56,7 @@ jobs:
     name: ${{ matrix.title }}
     runs-on: ubuntu-latest
     if: ${{ !contains(github.event.pull_request.title, 'WIP') }}
-    timeout-minutes: 60
+    timeout-minutes: 75
     strategy:
       fail-fast: false
       matrix:
@@ -122,7 +122,7 @@ jobs:
     name: AMD64 macOS 11 C++
     runs-on: macos-latest
     if: ${{ !contains(github.event.pull_request.title, 'WIP') }}
-    timeout-minutes: 60
+    timeout-minutes: 75
     strategy:
       fail-fast: false
     env:
@@ -179,7 +179,11 @@ jobs:
           key: cpp-ccache-macos-${{ hashFiles('cpp/**') }}
           restore-keys: cpp-ccache-macos-
       - name: Build
-        run: ci/scripts/cpp_build.sh $(pwd) $(pwd)/build
+        # use brew version of clang, to be consistent with LLVM lib, see ARROW-17790.
+        run: |
+          export CC=$(brew --prefix llvm)/bin/clang
+          export CXX=$(brew --prefix llvm)/bin/clang++
+          ci/scripts/cpp_build.sh $(pwd) $(pwd)/build
       - name: Test
         shell: bash
         run: |

.github/workflows/python.yml

@@ -160,7 +160,10 @@ jobs:
             -r python/requirements-test.txt
       - name: Build
         shell: bash
+        # use brew version of clang, to be consistent with LLVM lib, see ARROW-17790.
         run: |
+          export CC=$(brew --prefix llvm)/bin/clang
+          export CXX=$(brew --prefix llvm)/bin/clang++
           export PYTHON=python3
           ci/scripts/cpp_build.sh $(pwd) $(pwd)/build
           ci/scripts/python_build.sh $(pwd) $(pwd)/build

.github/workflows/ruby.yml

@@ -164,7 +164,10 @@ jobs:
           key: ruby-ccache-macos-${{ hashFiles('cpp/**') }}
           restore-keys: ruby-ccache-macos-
       - name: Build C++
+        # use brew version of clang, to be consistent with LLVM lib, see ARROW-17790.
         run: |
+          export CC=$(brew --prefix llvm)/bin/clang
+          export CXX=$(brew --prefix llvm)/bin/clang++
           ci/scripts/cpp_build.sh $(pwd) $(pwd)/build
       - name: Build GLib
         run: |

cpp/src/arrow/array/array_test.cc

@@ -2831,8 +2831,6 @@ class DecimalTest : public ::testing::TestWithParam<int> {
     auto type = std::make_shared<TYPE>(precision, 4);
     auto builder = std::make_shared<DecimalBuilder>(type);
 
-    size_t null_count = 0;
-
     const size_t size = draw.size();
 
     ARROW_EXPECT_OK(builder->Reserve(size));
@@ -2842,7 +2840,6 @@ class DecimalTest : public ::testing::TestWithParam<int> {
         ARROW_EXPECT_OK(builder->Append(draw[i]));
       } else {
         ARROW_EXPECT_OK(builder->AppendNull());
-        ++null_count;
       }
     }
 

cpp/src/arrow/filesystem/gcsfs_test.cc

@@ -17,6 +17,7 @@
 
 #include <algorithm>  // Missing include in boost/process
 
+#define BOOST_NO_CXX98_FUNCTION_BASE  // ARROW-17805
 // This boost/asio/io_context.hpp include is needless for no MinGW
 // build.
 //

cpp/src/arrow/filesystem/s3_test_util.cc

@@ -34,6 +34,7 @@
 #ifdef __MINGW32__
 #include <boost/asio/io_context.hpp>
 #endif
+#define BOOST_NO_CXX98_FUNCTION_BASE  // ARROW-17805
 // We need BOOST_USE_WINDOWS_H definition with MinGW when we use
 // boost/process.hpp. See BOOST_USE_WINDOWS_H=1 in
 // cpp/cmake_modules/ThirdpartyToolchain.cmake for details.

cpp/src/arrow/flight/test_util.cc

@@ -30,11 +30,12 @@
 // We need Windows fixes before including Boost
 #include "arrow/util/windows_compatibility.h"
 
+#include <gtest/gtest.h>
 #include <boost/filesystem.hpp>
+#define BOOST_NO_CXX98_FUNCTION_BASE  // ARROW-17805
 // We need BOOST_USE_WINDOWS_H definition with MinGW when we use
 // boost/process.hpp. See BOOST_USE_WINDOWS_H=1 in
 // cpp/cmake_modules/ThirdpartyToolchain.cmake for details.
-#include <gtest/gtest.h>
 #include <boost/process.hpp>
 
 #include "arrow/array.h"

cpp/src/gandiva/decimal_ir.cc

@@ -96,8 +96,9 @@ void DecimalIR::InitializeIntrinsics() {
 // CPP:  return kScaleMultipliers[scale]
 llvm::Value* DecimalIR::GetScaleMultiplier(llvm::Value* scale) {
   auto const_array = module()->getGlobalVariable(kScaleMultipliersName);
-  auto ptr = CreateGEP(ir_builder(), const_array, {types()->i32_constant(0), scale});
-  return CreateLoad(ir_builder(), ptr);
+  auto ptr = ir_builder()->CreateGEP(const_array->getValueType(), const_array,
+                                     {types()->i32_constant(0), scale});
+  return ir_builder()->CreateLoad(types()->i128_type(), ptr);
 }
 
 // CPP:  x <= y ? y : x
@@ -248,8 +249,8 @@ llvm::Value* DecimalIR::AddLarge(const ValueFull& x, const ValueFull& y,
   ir_builder()->CreateCall(module()->getFunction("add_large_decimal128_decimal128"),
                            args);
 
-  auto out_high = CreateLoad(ir_builder(), out_high_ptr);
-  auto out_low = CreateLoad(ir_builder(), out_low_ptr);
+  auto out_high = ir_builder()->CreateLoad(types()->i64_type(), out_high_ptr);
+  auto out_low = ir_builder()->CreateLoad(types()->i64_type(), out_low_ptr);
   auto sum = ValueSplit(out_high, out_low).AsInt128(this);
   ADD_TRACE_128("AddLarge : sum", sum);
   return sum;
@@ -445,8 +446,8 @@ llvm::Value* DecimalIR::CallDecimalFunction(const std::string& function_name,
     // Make call to pre-compiled IR function.
     ir_builder()->CreateCall(module()->getFunction(function_name), dis_assembled_args);
 
-    auto out_high = CreateLoad(ir_builder(), out_high_ptr);
-    auto out_low = CreateLoad(ir_builder(), out_low_ptr);
+    auto out_high = ir_builder()->CreateLoad(i64, out_high_ptr);
+    auto out_low = ir_builder()->CreateLoad(i64, out_low_ptr);
     result = ValueSplit(out_high, out_low).AsInt128(this);
   } else {
     DCHECK_NE(return_type, types()->void_type());

cpp/src/gandiva/engine.cc

@@ -234,11 +234,11 @@ Status Engine::LoadPreCompiledIR() {
                   Status::CodeGenError("Could not load module from IR: ",
                                        buffer_or_error.getError().message()));
 
-  std::unique_ptr<llvm::MemoryBuffer> buffer = move(buffer_or_error.get());
+  std::unique_ptr<llvm::MemoryBuffer> buffer = std::move(buffer_or_error.get());
 
   /// Parse the IR module.
   llvm::Expected<std::unique_ptr<llvm::Module>> module_or_error =
-      llvm::getOwningLazyBitcodeModule(move(buffer), *context());
+      llvm::getOwningLazyBitcodeModule(std::move(buffer), *context());
   if (!module_or_error) {
     // NOTE: llvm::handleAllErrors() fails linking with RTTI-disabled LLVM builds
     // (ARROW-5148)
@@ -247,14 +247,14 @@ Status Engine::LoadPreCompiledIR() {
     stream << module_or_error.takeError();
     return Status::CodeGenError(stream.str());
   }
-  std::unique_ptr<llvm::Module> ir_module = move(module_or_error.get());
+  std::unique_ptr<llvm::Module> ir_module = std::move(module_or_error.get());
 
   // set dataLayout
   SetDataLayout(ir_module.get());
 
   ARROW_RETURN_IF(llvm::verifyModule(*ir_module, &llvm::errs()),
                   Status::CodeGenError("verify of IR Module failed"));
-  ARROW_RETURN_IF(llvm::Linker::linkModules(*module_, move(ir_module)),
+  ARROW_RETURN_IF(llvm::Linker::linkModules(*module_, std::move(ir_module)),
                   Status::CodeGenError("failed to link IR Modules"));
 
   return Status::OK();

cpp/src/gandiva/engine_llvm_test.cc

@@ -80,8 +80,9 @@ class TestEngine : public ::testing::Test {
     loop_var->addIncoming(loop_update, loop_body);
 
     // get the current value
-    llvm::Value* offset = CreateGEP(builder, arg_elements, loop_var, "offset");
-    llvm::Value* current_value = CreateLoad(builder, offset, "value");
+    llvm::Value* offset =
+        builder->CreateGEP(types->i64_type(), arg_elements, loop_var, "offset");
+    llvm::Value* current_value = builder->CreateLoad(types->i64_type(), offset, "value");
 
     // setup sum PHI
     llvm::Value* sum_update = builder->CreateAdd(sum, current_value, "sum+ith");