[flang-commits] [flang] [mlir] [mlir][math] Replace roundeven call by nearbyint call when target does not have C23 features (PR #88687)

[Corentin Ferry via flang-commits]

================
@@ -185,8 +186,15 @@ void mlir::populateMathToLibmConversionPatterns(RewritePatternSet &patterns) {
   populatePatternsForOp<math::Log10Op>(patterns, ctx, "log10f", "log10");
   populatePatternsForOp<math::Log1pOp>(patterns, ctx, "log1pf", "log1p");
   populatePatternsForOp<math::PowFOp>(patterns, ctx, "powf", "pow");
-  populatePatternsForOp<math::RoundEvenOp>(patterns, ctx, "roundevenf",
-                                           "roundeven");
+  if (options.allowC23Features)
+    populatePatternsForOp<math::RoundEvenOp>(patterns, ctx, "roundevenf",
+                                             "roundeven");
+  else if (options.roundingModeIsDefault)
+    populatePatternsForOp<math::RoundEvenOp>(patterns, ctx, "nearbyintf",
+                                             "nearbyint");
+  // Roundeven: using nearbyint (pre-C23) for roundeven requires the
----------------
cferry-AMD wrote:

Thanks @frederik-h. I have touched `flang` code at some point in an attempt to fix a compile-time bug that went unnoticed on my end, but went backwards and found out I could fix it at the MathToLibm end. So this has nothing to do with Flang... sorry for the noise.

My use case is indeed a C compiler that does not support (yet) roundeven.

https://github.com/llvm/llvm-project/pull/88687