[libc-commits] [libc] [llvm] [libc][math] Refactor llogb to Header Only (PR #175524)

[Vedant Neve via libc-commits]

================
@@ -0,0 +1,64 @@
+//===-- Implementation header for llogb --------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC___SUPPORT_MATH_LLOGB_H
+#define LLVM_LIBC_SRC___SUPPORT_MATH_LLOGB_H
+
+#include "src/__support/CPP/bit.h"
+#include "src/__support/FPUtil/FPBits.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/optimization.h"  // LIBC_UNLIKELY
+#include <limits.h>
+
+namespace LIBC_NAMESPACE_DECL {
+namespace math {
+
+LIBC_INLINE constexpr long llogb(double x) {
+  using FPBits = fputil::FPBits<double>;
+  FPBits bits(x);
+
+  if (LIBC_UNLIKELY(bits.is_zero())) {
+    return -LONG_MAX - 1; // FP_LLOGB0
+  }
+
+  if (LIBC_UNLIKELY(bits.is_inf())) {
+    return LONG_MAX;
+  }
+
+  if (LIBC_UNLIKELY(bits.is_nan())) {
+#ifdef __FP_LOGBNAN_MIN
+    return -LONG_MAX - 1;
+#else
+    return LONG_MAX; // FP_LLOGBNAN
+#endif
+  }
+
+  int biased_exp = bits.get_biased_exponent();
+
+  // Handle normal numbers
+  if (biased_exp > 0) {
+    return biased_exp - FPBits::EXP_BIAS;
+  }
+
+  // Handle subnormal numbers
+  // Formula: log2(mantissa * 2^-1074) = log2(mantissa) - 1074
+  uint64_t mantissa = bits.get_mantissa();
+
+  // Define constants
+  constexpr int EXP_ADJUSTMENT = FPBits::EXP_BIAS + FPBits::FRACTION_LEN - 1;
+  constexpr int STORAGE_MSB = (sizeof(uint64_t) * CHAR_BIT) - 1;
+
+  // log2(mantissa) = STORAGE_MSB - countl_zero(mantissa)
+  return (STORAGE_MSB - cpp::countl_zero(mantissa)) - EXP_ADJUSTMENT;
+}
----------------
0bVdnt wrote:

Added errno/exception handling and missing dependencies. Skipped range check as double exponent always fits in long

https://github.com/llvm/llvm-project/pull/175524