[libc-commits] [libc] 062d24a - [libc][math] Fix some mis-optimization issue with hypotf16. (#141960)

[via libc-commits]

Author: lntue
Date: 2025-05-29T11:34:50-04:00
New Revision: 062d24a7d6770b1dd7d5a4968bccb28c85495d67

URL: https://github.com/llvm/llvm-project/commit/062d24a7d6770b1dd7d5a4968bccb28c85495d67
DIFF: https://github.com/llvm/llvm-project/commit/062d24a7d6770b1dd7d5a4968bccb28c85495d67.diff

LOG: [libc][math] Fix some mis-optimization issue with hypotf16. (#141960)

Added: 
    

Modified: 
    libc/src/math/generic/hypotf16.cpp

Removed: 
    


################################################################################
diff  --git a/libc/src/math/generic/hypotf16.cpp b/libc/src/math/generic/hypotf16.cpp
index 8f80986204b27..d782c2687cdb6 100644
--- a/libc/src/math/generic/hypotf16.cpp
+++ b/libc/src/math/generic/hypotf16.cpp
@@ -48,10 +48,13 @@ LLVM_LIBC_FUNCTION(float16, hypotf16, (float16 x, float16 y)) {
     return a_bits.get_val();
   }
 
+  // TODO: Investigate why replacing the return line below with:
+  //   return x_bits.get_val() + y_bits.get_val();
+  // fails the hypotf16 smoke tests.
   if (LIBC_UNLIKELY(a_u - b_u >=
                     static_cast<uint16_t>((FPBits::FRACTION_LEN + 2)
                                           << FPBits::FRACTION_LEN)))
-    return x_abs.get_val() + y_abs.get_val();
+    return a_bits.get_val() + b_bits.get_val();
 
   float af = fputil::cast<float>(a_bits.get_val());
   float bf = fputil::cast<float>(b_bits.get_val());