merge main into amd-staging (#640)

Author: z1-cciauto

.github/CODEOWNERS

@@ -60,6 +60,11 @@
 /mlir/lib/Conversion/*ToROCDL @krzysz00 @kuhar
 /mlir/include/mlir/Dialect/LLVMIR/ROCDL* @krzysz00 @kuhar
 
+# Arith dialect in MLIR.
+/mlir/include/mlir/Dialect/Arith @kuhar
+/mlir/lib/Dialect/Arith @kuhar
+/mlir/lib/Conversion/ArithTo* @kuhar
+
 # XeGPU and XeVM dialects in MLIR.
 /mlir/include/mlir/Dialect/XeGPU @charithaintc @Jianhui-Li
 /mlir/lib/Dialect/XeGPU @charithaintc @Jianhui-Li

clang/lib/Tooling/DependencyScanning/DependencyScannerImpl.cpp

@@ -432,7 +432,7 @@ dependencies::createCompilerInvocation(ArrayRef<std::string> CommandLine,
 }
 
 std::pair<IntrusiveRefCntPtr<llvm::vfs::FileSystem>, std::vector<std::string>>
-dependencies::initVFSForTUBuferScanning(
+dependencies::initVFSForTUBufferScanning(
     IntrusiveRefCntPtr<llvm::vfs::FileSystem> BaseFS,
     ArrayRef<std::string> CommandLine, StringRef WorkingDirectory,
     llvm::MemoryBufferRef TUBuffer) {

clang/lib/Tooling/DependencyScanning/DependencyScannerImpl.h

@@ -113,10 +113,10 @@ createCompilerInvocation(ArrayRef<std::string> CommandLine,
                          DiagnosticsEngine &Diags);
 
 std::pair<IntrusiveRefCntPtr<llvm::vfs::FileSystem>, std::vector<std::string>>
-initVFSForTUBuferScanning(IntrusiveRefCntPtr<llvm::vfs::FileSystem> BaseFS,
-                          ArrayRef<std::string> CommandLine,
-                          StringRef WorkingDirectory,
-                          llvm::MemoryBufferRef TUBuffer);
+initVFSForTUBufferScanning(IntrusiveRefCntPtr<llvm::vfs::FileSystem> BaseFS,
+                           ArrayRef<std::string> CommandLine,
+                           StringRef WorkingDirectory,
+                           llvm::MemoryBufferRef TUBuffer);
 
 std::pair<IntrusiveRefCntPtr<llvm::vfs::OverlayFileSystem>,
           std::vector<std::string>>

clang/lib/Tooling/DependencyScanning/DependencyScanningWorker.cpp

@@ -157,7 +157,7 @@ bool DependencyScanningWorker::computeDependencies(
     DependencyConsumer &Consumer, DependencyActionController &Controller,
     DiagnosticConsumer &DC, std::optional<llvm::MemoryBufferRef> TUBuffer) {
   if (TUBuffer) {
-    auto [FinalFS, FinalCommandLine] = initVFSForTUBuferScanning(
+    auto [FinalFS, FinalCommandLine] = initVFSForTUBufferScanning(
         BaseFS, CommandLine, WorkingDirectory, *TUBuffer);
     return scanDependencies(WorkingDirectory, FinalCommandLine, Consumer,
                             Controller, DC, FinalFS);

compiler-rt/test/asan/TestCases/stack_container_dynamic_lib.cpp

@@ -1,6 +1,6 @@
 // Test to demonstrate compile-time disabling of container-overflow checks
 // in order to handle uninstrumented libraries
-// UNSUPPORTED: target={{.*windows-msvc.*}}
+// UNSUPPORTED: target={{.*windows-.*}}
 
 // Mimic a closed-source library compiled without ASan
 // RUN: %clangxx_asan -fno-sanitize=address -DSHARED_LIB %s %fPIC -shared -o %t-so.so

flang/include/flang/Optimizer/Builder/CUFCommon.h

@@ -39,6 +39,10 @@ int computeElementByteSize(mlir::Location loc, mlir::Type type,
                            fir::KindMapping &kindMap,
                            bool emitErrorOnFailure = true);
 
+mlir::Value computeElementCount(mlir::PatternRewriter &rewriter,
+                                mlir::Location loc, mlir::Value shapeOperand,
+                                mlir::Type seqType, mlir::Type targetType);
+
 } // namespace cuf
 
 #endif // FORTRAN_OPTIMIZER_TRANSFORMS_CUFCOMMON_H_

flang/lib/Optimizer/Builder/CUFCommon.cpp

@@ -114,3 +114,44 @@ int cuf::computeElementByteSize(mlir::Location loc, mlir::Type type,
     mlir::emitError(loc, "unsupported type");
   return 0;
 }
+
+mlir::Value cuf::computeElementCount(mlir::PatternRewriter &rewriter,
+                                     mlir::Location loc,
+                                     mlir::Value shapeOperand,
+                                     mlir::Type seqType,
+                                     mlir::Type targetType) {
+  if (shapeOperand) {
+    // Dynamic extent - extract from shape operand
+    llvm::SmallVector<mlir::Value> extents;
+    if (auto shapeOp =
+            mlir::dyn_cast<fir::ShapeOp>(shapeOperand.getDefiningOp())) {
+      extents = shapeOp.getExtents();
+    } else if (auto shapeShiftOp = mlir::dyn_cast<fir::ShapeShiftOp>(
+                   shapeOperand.getDefiningOp())) {
+      for (auto i : llvm::enumerate(shapeShiftOp.getPairs()))
+        if (i.index() & 1)
+          extents.push_back(i.value());
+    }
+
+    if (extents.empty())
+      return mlir::Value();
+
+    // Compute total element count by multiplying all dimensions
+    mlir::Value count =
+        fir::ConvertOp::create(rewriter, loc, targetType, extents[0]);
+    for (unsigned i = 1; i < extents.size(); ++i) {
+      auto operand =
+          fir::ConvertOp::create(rewriter, loc, targetType, extents[i]);
+      count = mlir::arith::MulIOp::create(rewriter, loc, count, operand);
+    }
+    return count;
+  } else {
+    // Static extent - use constant array size
+    if (auto seqTy = mlir::dyn_cast_or_null<fir::SequenceType>(seqType)) {
+      mlir::IntegerAttr attr =
+          rewriter.getIntegerAttr(targetType, seqTy.getConstantArraySize());
+      return mlir::arith::ConstantOp::create(rewriter, loc, targetType, attr);
+    }
+  }
+  return mlir::Value();
+}

flang/lib/Optimizer/Transforms/CUFOpConversion.cpp

@@ -651,31 +651,8 @@ struct CUFDataTransferOpConversion
       }
 
       mlir::Type i64Ty = builder.getI64Type();
-      mlir::Value nbElement;
-      if (op.getShape()) {
-        llvm::SmallVector<mlir::Value> extents;
-        if (auto shapeOp =
-                mlir::dyn_cast<fir::ShapeOp>(op.getShape().getDefiningOp())) {
-          extents = shapeOp.getExtents();
-        } else if (auto shapeShiftOp = mlir::dyn_cast<fir::ShapeShiftOp>(
-                       op.getShape().getDefiningOp())) {
-          for (auto i : llvm::enumerate(shapeShiftOp.getPairs()))
-            if (i.index() & 1)
-              extents.push_back(i.value());
-        }
-
-        nbElement = fir::ConvertOp::create(rewriter, loc, i64Ty, extents[0]);
-        for (unsigned i = 1; i < extents.size(); ++i) {
-          auto operand =
-              fir::ConvertOp::create(rewriter, loc, i64Ty, extents[i]);
-          nbElement =
-              mlir::arith::MulIOp::create(rewriter, loc, nbElement, operand);
-        }
-      } else {
-        if (auto seqTy = mlir::dyn_cast_or_null<fir::SequenceType>(dstTy))
-          nbElement = builder.createIntegerConstant(
-              loc, i64Ty, seqTy.getConstantArraySize());
-      }
+      mlir::Value nbElement =
+          cuf::computeElementCount(rewriter, loc, op.getShape(), dstTy, i64Ty);
       unsigned width = 0;
       if (fir::isa_derived(fir::unwrapSequenceType(dstTy))) {
         mlir::Type structTy =

libc/cmake/caches/armv6m-none-eabi.cmake

@@ -2,7 +2,7 @@ set(CMAKE_SYSTEM_PROCESSOR arm CACHE STRING "")
 set(RUNTIMES_TARGET_TRIPLE "armv6m-none-eabi" CACHE STRING "")
 
 foreach(lang C;CXX;ASM)
-    set(CMAKE_${lang}_FLAGS "-march=armv6m -mcpu=cortex-m0plus -mfloat-abi=soft -Wno-atomic-alignment \"-Dvfprintf(stream, format, vlist)=vprintf(format, vlist)\" \"-Dfprintf(stream, format, ...)=printf(format)\" \"-Dfputs(string, stream)=puts(string)\" -D_LIBCPP_PRINT=1" CACHE STRING "")
+    set(CMAKE_${lang}_FLAGS "-march=armv6m -mcpu=cortex-m0plus -mfloat-abi=soft -Wno-atomic-alignment" CACHE STRING "")
 endforeach()
 
 include(${CMAKE_CURRENT_LIST_DIR}/baremetal_common.cmake)

libc/cmake/caches/armv7em-none-eabi.cmake

@@ -2,7 +2,7 @@ set(CMAKE_SYSTEM_PROCESSOR arm CACHE STRING "")
 set(RUNTIMES_TARGET_TRIPLE "armv7em-none-eabi" CACHE STRING "")
 
 foreach(lang C;CXX;ASM)
-    set(CMAKE_${lang}_FLAGS "-march=armv7em -mcpu=cortex-m4 -mfloat-abi=soft -Wno-atomic-alignment \"-Dvfprintf(stream, format, vlist)=vprintf(format, vlist)\" \"-Dfprintf(stream, format, ...)=printf(format)\" \"-Dfputs(string, stream)=puts(string)\" -D_LIBCPP_PRINT=1" CACHE STRING "")
+    set(CMAKE_${lang}_FLAGS "-march=armv7em -mcpu=cortex-m4 -mfloat-abi=soft -Wno-atomic-alignment" CACHE STRING "")
 endforeach()
 
 include(${CMAKE_CURRENT_LIST_DIR}/baremetal_common.cmake)