[clang] [llvm] APFloat: Fix maxnum and minnum with sNaN (PR #112854)

[YunQiang Su via llvm-commits]

================
@@ -582,7 +582,46 @@ TEST(APFloatTest, MinNum) {
   APFloat zp(0.0);
   APFloat zn(-0.0);
   EXPECT_EQ(-0.0, minnum(zp, zn).convertToDouble());
-  EXPECT_EQ(-0.0, minnum(zn, zp).convertToDouble());
+
+  APInt intPayload_89ab(64, 0x89ab);
----------------
wzssyqa wrote:

It is not easy to do so, as lit test need to utilize something like `expandFMINNUMFMAXNUM` to run the test.

Anyway, I think that it should be another patch. Maybe we can call them
   better_runtime_apfloat_fmax_test.ll
;)

https://github.com/llvm/llvm-project/pull/112854