[llvm] [APFloat] Add exp function for APFloat::IEEESsingle using expf implementation from LLVM libc. (PR #143959)

[via llvm-commits]

================
@@ -8356,4 +8356,23 @@ TEST(APFloatTest, hasSignBitInMSB) {
   EXPECT_FALSE(APFloat::hasSignBitInMSB(APFloat::Float8E8M0FNU()));
 }
 
+#ifdef LLVM_INTEGRATE_LIBC
+TEST(APFloatTest, expf) {
+  EXPECT_EQ(
+      1.0f,
+      llvm::exp(APFloat(0.0f), APFloat::rmNearestTiesToEven).convertToFloat());
----------------
lntue wrote:

Yes, or those odd cases.  But in here we have float and normal values, and I hope people don't alias `float` with x87 80-bit fp type <finger crossed>

https://github.com/llvm/llvm-project/pull/143959