[libc-commits] [libc] e5e93f6 - [libc] Return a float NaN for log1pf instead of double NaN.

[Tue Ly via libc-commits]

Author: Tue Ly
Date: 2022-02-07T21:07:09-05:00
New Revision: e5e93f60ee0249d7aa8fe1723f8531e795adc479

URL: https://github.com/llvm/llvm-project/commit/e5e93f60ee0249d7aa8fe1723f8531e795adc479
DIFF: https://github.com/llvm/llvm-project/commit/e5e93f60ee0249d7aa8fe1723f8531e795adc479.diff

LOG: [libc] Return a float NaN for log1pf instead of double NaN.

Added: 
    

Modified: 
    libc/src/math/generic/log1pf.cpp

Removed: 
    


################################################################################
diff  --git a/libc/src/math/generic/log1pf.cpp b/libc/src/math/generic/log1pf.cpp
index 25e5271266ead..b9494a4e02080 100644
--- a/libc/src/math/generic/log1pf.cpp
+++ b/libc/src/math/generic/log1pf.cpp
@@ -44,7 +44,8 @@ INLINE_FMA static inline float log(double x) {
 
   if (xbits.uintval() > FPBits::MAX_NORMAL) {
     if (xbits.get_sign() && !xbits.is_nan()) {
-      return FPBits::build_nan(1 << (fputil::MantissaWidth<float>::VALUE - 1));
+      return fputil::FPBits<float>::build_nan(
+          1 << (fputil::MantissaWidth<float>::VALUE - 1));
     }
     return static_cast<float>(x);
   }