[flang-commits] [flang] [flang][debug] Add support for fixed size arrays. (PR #92568)

[Abid Qadeer via flang-commits]

================
@@ -37,6 +37,44 @@ static mlir::LLVM::DITypeAttr genPlaceholderType(mlir::MLIRContext *context) {
                       llvm::dwarf::DW_ATE_signed);
 }
 
+mlir::LLVM::DITypeAttr DebugTypeGenerator::convertSequenceType(
+    fir::SequenceType seqTy, mlir::LLVM::DIFileAttr fileAttr,
+    mlir::LLVM::DIScopeAttr scope, mlir::Location loc) {
+
+  mlir::MLIRContext *context = module.getContext();
+  // FIXME: Only fixed sizes arrays handled at the moment.
+  if (seqTy.hasDynamicExtents())
+    return genPlaceholderType(context);
+
+  llvm::SmallVector<mlir::LLVM::DINodeAttr> elements;
+  auto elemTy = convertType(seqTy.getEleTy(), fileAttr, scope, loc);
+
+  for (auto dim : seqTy.getShape()) {
+    auto intTy = mlir::IntegerType::get(context, 64);
+    // FIXME: Only supporting lower bound of 1 at the moment. The
+    // 'SequenceType' has information about the shape but not the shift. In
+    // cases, where the conversion originated during the processing of
+    // 'DeclareOp', it may be possible to pass on this information. But the
+    // type conversion should ideally be based on what information present in
+    // the type class so that it works from everywhere (e.g. when it is part
+    // of a module or a derived type.)
----------------
abidh wrote:

I can see an `fir.declare` where it is used inside `FuncOp` but not at the point of declaration in `GlobalOp`. Also if we use the info in `fir.declare`, we will have to pass it through all the type generation functions. It will be much cleaner if it could be added to `SequenceType` instead.

https://github.com/llvm/llvm-project/pull/92568