[flang-commits] [flang] [flang][cuda] Add fir.cuda_allocate operation (PR #88586)

[Slava Zakharin via flang-commits]

================
@@ -3190,4 +3190,36 @@ def fir_CUDADataTransferOp : fir_Op<"cuda_data_transfer", []> {
   }];
 }
 
+def fir_CUDAAllocateOp : fir_Op<"cuda_allocate", [AttrSizedOperandSegments,
+    MemoryEffects<[MemAlloc<DefaultResource>]>]> {
----------------
vzakhari wrote:

Should it also have a write effect on the box reference operand?

https://github.com/llvm/llvm-project/pull/88586