[flang-commits] [flang] [mlir] [flang][OpenMP] Support for "atomic compare capture" (PR #202315)

[via flang-commits]

================
@@ -4815,13 +4852,150 @@ convertOmpAtomicCapture(omp::AtomicCaptureOp atomicCaptureOp,
   if (failed(checkImplementationStatus(*atomicCaptureOp)))
     return failure();
 
+  omp::AtomicUpdateOp atomicUpdateOp = atomicCaptureOp.getAtomicUpdateOp();
+  omp::AtomicWriteOp atomicWriteOp = atomicCaptureOp.getAtomicWriteOp();
+  omp::AtomicCompareOp atomicCompareOp = atomicCaptureOp.getAtomicCompareOp();
+
+  // If the capture contains an atomic.compare, delegate to
+  // createAtomicCompare with the capture variable (V) set.
+  if (atomicCompareOp) {
+    omp::AtomicReadOp atomicReadOp = atomicCaptureOp.getAtomicReadOp();
+    assert(atomicReadOp && "expected atomic.read in capture+compare");
+
+    Region &region = atomicCompareOp.getRegion();
+    Block &block = region.front();
+
+    llvm::Type *llvmXElementType =
+        moduleTranslation.convertType(block.getArgument(0).getType());
+    llvm::Value *llvmX = moduleTranslation.lookupValue(atomicCompareOp.getX());
+    llvm::Value *llvmV = moduleTranslation.lookupValue(atomicReadOp.getV());
----------------
SunilKuravinakop wrote:

moduleTranslation.convertType() returns 
   1) typeTranslator.translateType(type) at line 2402 in file mlir/lib/Target/LLVMIR/ModuleTranslation.cpp. 
    2) TypeToLLVMIRTranslatorImpl::translateType() at line 31 in file mlir/lib/Target/LLVMIR/TypeToLLVM.cpp never returns NULL.. If the type is not on
    
    Because of this, there is no need to check for NULL. Justify why a check is needed here.


https://github.com/llvm/llvm-project/pull/202315