[mlir][GPUToNVVM] Fix memref function args/results (#96392)

The `gpu.func` op lowering accounts for memref arguments/results (both
"normal" and bare-pointer supported), but the `gpu.return` op lowering
did not. The lowering produced invalid IR that did not verify.

This commit uses the same lowering strategy as for `func.return` in the
`gpu.return` lowering. (The C++ implementation is copied. We may want to
share some code between `func` and `gpu` lowerings in the future.)

Author: matthias-springer

mlir/lib/Conversion/GPUCommon/GPUOpsLowering.cpp

@@ -684,6 +684,62 @@ LogicalResult GPUDynamicSharedMemoryOpLowering::matchAndRewrite(
   return success();
 }
 
+LogicalResult GPUReturnOpLowering::matchAndRewrite(
+    gpu::ReturnOp op, OpAdaptor adaptor,
+    ConversionPatternRewriter &rewriter) const {
+  Location loc = op.getLoc();
+  unsigned numArguments = op.getNumOperands();
+  SmallVector<Value, 4> updatedOperands;
+
+  bool useBarePtrCallConv = getTypeConverter()->getOptions().useBarePtrCallConv;
+  if (useBarePtrCallConv) {
+    // For the bare-ptr calling convention, extract the aligned pointer to
+    // be returned from the memref descriptor.
+    for (auto it : llvm::zip(op->getOperands(), adaptor.getOperands())) {
+      Type oldTy = std::get<0>(it).getType();
+      Value newOperand = std::get<1>(it);
+      if (isa<MemRefType>(oldTy) && getTypeConverter()->canConvertToBarePtr(
+                                        cast<BaseMemRefType>(oldTy))) {
+        MemRefDescriptor memrefDesc(newOperand);
+        newOperand = memrefDesc.allocatedPtr(rewriter, loc);
+      } else if (isa<UnrankedMemRefType>(oldTy)) {
+        // Unranked memref is not supported in the bare pointer calling
+        // convention.
+        return failure();
+      }
+      updatedOperands.push_back(newOperand);
+    }
+  } else {
+    updatedOperands = llvm::to_vector<4>(adaptor.getOperands());
+    (void)copyUnrankedDescriptors(rewriter, loc, op.getOperands().getTypes(),
+                                  updatedOperands,
+                                  /*toDynamic=*/true);
+  }
+
+  // If ReturnOp has 0 or 1 operand, create it and return immediately.
+  if (numArguments <= 1) {
+    rewriter.replaceOpWithNewOp<LLVM::ReturnOp>(
+        op, TypeRange(), updatedOperands, op->getAttrs());
+    return success();
+  }
+
+  // Otherwise, we need to pack the arguments into an LLVM struct type before
+  // returning.
+  auto packedType = getTypeConverter()->packFunctionResults(
+      op.getOperandTypes(), useBarePtrCallConv);
+  if (!packedType) {
+    return rewriter.notifyMatchFailure(op, "could not convert result types");
+  }
+
+  Value packed = rewriter.create<LLVM::UndefOp>(loc, packedType);
+  for (auto [idx, operand] : llvm::enumerate(updatedOperands)) {
+    packed = rewriter.create<LLVM::InsertValueOp>(loc, packed, operand, idx);
+  }
+  rewriter.replaceOpWithNewOp<LLVM::ReturnOp>(op, TypeRange(), packed,
+                                              op->getAttrs());
+  return success();
+}
+
 void mlir::populateGpuMemorySpaceAttributeConversions(
     TypeConverter &typeConverter, const MemorySpaceMapping &mapping) {
   typeConverter.addTypeAttributeConversion(

mlir/lib/Conversion/GPUCommon/GPUOpsLowering.h

@@ -112,10 +112,7 @@ struct GPUReturnOpLowering : public ConvertOpToLLVMPattern<gpu::ReturnOp> {
 
   LogicalResult
   matchAndRewrite(gpu::ReturnOp op, OpAdaptor adaptor,
-                  ConversionPatternRewriter &rewriter) const override {
-    rewriter.replaceOpWithNewOp<LLVM::ReturnOp>(op, adaptor.getOperands());
-    return success();
-  }
+                  ConversionPatternRewriter &rewriter) const override;
 };
 
 namespace impl {

mlir/test/Conversion/GPUToNVVM/gpu-to-nvvm.mlir

@@ -1,4 +1,5 @@
 // RUN: mlir-opt %s -convert-gpu-to-nvvm='has-redux=1' -split-input-file | FileCheck %s
+// RUN: mlir-opt %s -convert-gpu-to-nvvm='has-redux=1 use-bare-ptr-memref-call-conv=1' -split-input-file | FileCheck %s --check-prefix=CHECK-BARE
 // RUN: mlir-opt %s -transform-interpreter | FileCheck %s
 
 gpu.module @test_module_0 {
@@ -670,7 +671,7 @@ gpu.module @test_module_32 {
   }
 }
 
-gpu.module @gpumodule {
+gpu.module @test_module_33 {
 // CHECK-LABEL: func @kernel_with_block_size()
 // CHECK: attributes {gpu.kernel, gpu.known_block_size = array<i32: 128, 1, 1>, nvvm.kernel, nvvm.maxntid = array<i32: 128, 1, 1>}
   gpu.func @kernel_with_block_size() kernel attributes {known_block_size = array<i32: 128, 1, 1>} {
@@ -679,6 +680,28 @@ gpu.module @gpumodule {
 }
 
 
+gpu.module @test_module_34 {
+  // CHECK-LABEL: llvm.func @memref_signature(
+  //  CHECK-SAME:     %{{.*}}: !llvm.ptr, %{{.*}}: !llvm.ptr, %{{.*}}: i64, %{{.*}}: i64, %{{.*}}: i64, %{{.*}}: f32) -> !llvm.struct<(struct<(ptr, ptr, i64, array<1 x i64>, array<1 x i64>)>, f32)>
+  //       CHECK:   llvm.mlir.undef
+  //       CHECK:   llvm.insertvalue
+  //       CHECK:   llvm.insertvalue
+  //       CHECK:   llvm.insertvalue
+  //       CHECK:   llvm.insertvalue
+  //       CHECK:   llvm.insertvalue
+  //       CHECK:   llvm.mlir.undef
+  //       CHECK:   llvm.insertvalue
+  //       CHECK:   llvm.insertvalue
+  //       CHECK:   llvm.return
+
+  // CHECK-BARE-LABEL: llvm.func @memref_signature(
+  //  CHECK-BARE-SAME:     %{{.*}}: !llvm.ptr, %{{.*}}: f32) -> !llvm.struct<(ptr, f32)>
+  gpu.func @memref_signature(%m: memref<2xf32>, %f: f32) -> (memref<2xf32>, f32) {
+    gpu.return %m, %f : memref<2xf32>, f32
+  }
+}
+
+
 module attributes {transform.with_named_sequence} {
   transform.named_sequence @__transform_main(%toplevel_module: !transform.any_op {transform.readonly}) {
     %gpu_module = transform.structured.match ops{["gpu.module"]} in %toplevel_module
@@ -701,9 +724,7 @@ module attributes {transform.with_named_sequence} {
     } with type_converter {
       transform.apply_conversion_patterns.memref.memref_to_llvm_type_converter
         {index_bitwidth = 64,
-        use_bare_ptr = true,
-        use_bare_ptr_memref_call_conv = true,
-        use_opaque_pointers = true}
+        use_bare_ptr_call_conv = false}
     } {
       legal_dialects = ["llvm", "memref", "nvvm", "test"],
       legal_ops = ["func.func", "gpu.module", "gpu.module_end", "gpu.yield"],