From aeaafce4644788963d9823067e46b50266af4f05 Mon Sep 17 00:00:00 2001
From: Tom Eccles <tom.eccles@arm.com>
Date: Fri, 31 Jan 2025 09:35:26 +0000
Subject: [mlir][OpenMP][flang] make private variable allocation implicit in
 omp.private (#124019)

The intention of this work is to give MLIR->LLVMIR conversion freedom to
control how the private variable is allocated so that it can be
allocated on the stack in ordinary cases or as part of a structure used
to give closure context for tasks which might outlive the current stack
frame. See RFC:

https://discourse.llvm.org/t/rfc-openmp-supporting-delayed-task-execution-with-firstprivate-variables/83084

For example, a privatizer for an integer used to look like
```mlir
  omp.private {type = private} @x.privatizer : !fir.ref<i32> alloc {
  ^bb0(%arg0: !fir.ref<i32>):
    %0 = ... allocate proper memory for the private clone ...
    omp.yield(%0 : !fir.ref<i32>)
  }
```

After this change, allocation become implicit in the operation:
```mlir
  omp.private {type = private} @x.privatizer : i32
```

For more complex types that require initialization after allocation, an
init region can be used:
``` mlir
  omp.private {type = private} @x.privatizer : !some.type init {
  ^bb0(%arg0: !some.pointer<!some.type>, %arg1: !some.pointer<!some.type>):
    // initialize %arg1, using %arg0 as a mold for allocations
    omp.yield(%arg1 : !some.pointer<!some.type>)
  } dealloc {
    ^bb0(%arg0: !some.pointer<!some.type>):
    ... deallocate memory allocated by the init region ...
    omp.yield
  }
```

This patch lays the groundwork for delayed task execution but is not
enough on its own.

After this patch all gfortran tests which previously passed still pass.
There
are the following changes to the Fujitsu test suite:
- 0380_0009 and 0435_0009 are fixed
- 0688_0041 now fails at runtime. This patch is testing firstprivate
variables with tasks. Previously we got lucky with the undefined
behavior and won the race. After these changes we no longer get lucky.
This patch lays the groundwork for a proper fix for this issue.

In flang the lowering re-uses the existing lowering used for reduction
init and dealloc regions.

In flang, before this patch we hit a TODO with the same wording when
generating the copy region for firstprivate polymorphic variables. After
this patch the box-like fir.class is passed by reference into the copy
region, leading to a different path that didn't hit that old TODO but
the generated code still didn't work so I added a new TODO in
DataSharingProcessor.
---
 flang/lib/Optimizer/CodeGen/CodeGenOpenMP.cpp | 36 +++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

(limited to 'flang/lib/Optimizer/CodeGen/CodeGenOpenMP.cpp')

diff --git a/flang/lib/Optimizer/CodeGen/CodeGenOpenMP.cpp b/flang/lib/Optimizer/CodeGen/CodeGenOpenMP.cpp
index da13ed6..37f1c9f 100644
--- a/flang/lib/Optimizer/CodeGen/CodeGenOpenMP.cpp
+++ b/flang/lib/Optimizer/CodeGen/CodeGenOpenMP.cpp
@@ -90,9 +90,45 @@ struct MapInfoOpConversion
     return mlir::success();
   }
 };
+
+// FIR op specific conversion for PrivateClauseOp that overwrites the default
+// OpenMP Dialect lowering, this allows FIR-aware lowering of types, required
+// for boxes because the OpenMP dialect conversion doesn't know anything about
+// FIR types.
+struct PrivateClauseOpConversion
+    : public OpenMPFIROpConversion<mlir::omp::PrivateClauseOp> {
+  using OpenMPFIROpConversion::OpenMPFIROpConversion;
+
+  llvm::LogicalResult
+  matchAndRewrite(mlir::omp::PrivateClauseOp curOp, OpAdaptor adaptor,
+                  mlir::ConversionPatternRewriter &rewriter) const override {
+    const fir::LLVMTypeConverter &converter = lowerTy();
+    mlir::Type convertedAllocType;
+    if (auto box = mlir::dyn_cast<fir::BaseBoxType>(curOp.getType())) {
+      // In LLVM codegen fir.box<> == fir.ref<fir.box<>> == llvm.ptr
+      // Here we really do want the actual structure
+      if (box.isAssumedRank())
+        TODO(curOp->getLoc(), "Privatize an assumed rank array");
+      unsigned rank = 0;
+      if (auto seqTy = mlir::dyn_cast<fir::SequenceType>(
+              fir::unwrapRefType(box.getEleTy())))
+        rank = seqTy.getShape().size();
+      convertedAllocType = converter.convertBoxTypeAsStruct(box, rank);
+    } else {
+      convertedAllocType = converter.convertType(adaptor.getType());
+    }
+    if (!convertedAllocType)
+      return mlir::failure();
+    rewriter.startOpModification(curOp);
+    curOp.setType(convertedAllocType);
+    rewriter.finalizeOpModification(curOp);
+    return mlir::success();
+  }
+};
 } // namespace
 
 void fir::populateOpenMPFIRToLLVMConversionPatterns(
     const LLVMTypeConverter &converter, mlir::RewritePatternSet &patterns) {
   patterns.add<MapInfoOpConversion>(converter);
+  patterns.add<PrivateClauseOpConversion>(converter);
 }
-- 
cgit v1.1