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

Valentin Clement バレンタイン クレメン via flang-commits flang-commits at lists.llvm.org
Mon Apr 15 13:30:30 PDT 2024


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

Done. I added also effect on `errmsg`, `pinned` and `source`. `errmsg`, `pinned` are diagnostic variable that will be written to if present and `source` will be used for source allocation so we need to read from it. 

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


More information about the flang-commits mailing list