[Mlir-commits] [llvm] [mlir] [atomicrmw] fminimumnum/fmaximumnum support (PR #187030)

[Krzysztof Drewniak]

================
@@ -1774,6 +1774,18 @@ Value *SplitPtrStructs::handleMemoryInst(Instruction *I, Value *Arg, Value *Ptr,
           "buffer resources and should've been expanded away");
       break;
     }
+    case AtomicRMWInst::FMaximumNum: {
+      reportFatalUsageError(
+          "atomic floating point fmaximumnum not supported for "
+          "buffer resources and should've been expanded away");
+      break;
+    }
+    case AtomicRMWInst::FMinimumNum: {
+      reportFatalUsageError(
----------------
krzysz00 wrote:

Note to self: know enough about what's going on here to know how true this is long-term

https://github.com/llvm/llvm-project/pull/187030