[libc][math] Refactor sinf implementation to header-only in src/__support/math folder. by nico · Pull Request #177963 · llvm/llvm-project

nico · 2026-01-26T14:09:10Z

in preparation for:

https://discourse.llvm.org/t/rfc-make-clang-builtin-math-functions-constexpr-with-llvm-libc-to-support-c-23-constexpr-math-functions/86450

I used #177583 to create the initial version of this. (It got a bit confused by the #if/#else/#endif for sinf, but easy to fix up manually.)

…port/math folder. Part of llvm#147386 in preparation for: https://discourse.llvm.org/t/rfc-make-clang-builtin-math-functions-constexpr-with-llvm-libc-to-support-c-23-constexpr-math-functions/86450

llvmbot · 2026-01-26T14:11:55Z

@llvm/pr-subscribers-libc

Author: Nico Weber (nico)

Changes

Part of #147386

in preparation for:

https://discourse.llvm.org/t/rfc-make-clang-builtin-math-functions-constexpr-with-llvm-libc-to-support-c-23-constexpr-math-functions/86450

I used #177583 to create the initial version of this. (It got a bit confused by the #if/#else/#endif for sinf, but easy to fix up manually.)

Full diff: https://github.com/llvm/llvm-project/pull/177963.diff

9 Files Affected:

(modified) libc/shared/math.h (+1)
(added) libc/shared/math/sinf.h (+23)
(modified) libc/src/__support/math/CMakeLists.txt (+16)
(added) libc/src/__support/math/sinf.h (+194)
(modified) libc/src/math/generic/CMakeLists.txt (+1-9)
(modified) libc/src/math/generic/sinf.cpp (+2-168)
(modified) libc/test/shared/CMakeLists.txt (+1)
(modified) libc/test/shared/shared_math_test.cpp (+1)
(modified) utils/bazel/llvm-project-overlay/libc/BUILD.bazel (+15-7)

diff --git a/libc/shared/math.h b/libc/shared/math.h
index 5126e46c9772e..9e76e710ecb01 100644
--- a/libc/shared/math.h
+++ b/libc/shared/math.h
@@ -84,6 +84,7 @@
 #include "math/rsqrtf.h"
 #include "math/rsqrtf16.h"
 #include "math/sin.h"
+#include "math/sinf.h"
 #include "math/tan.h"
 #include "math/tanf.h"
 
diff --git a/libc/shared/math/sinf.h b/libc/shared/math/sinf.h
new file mode 100644
index 0000000000000..5157b1e2d7cdc
--- /dev/null
+++ b/libc/shared/math/sinf.h
@@ -0,0 +1,23 @@
+//===-- Shared sinf function ------------------------------------*- 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_SHARED_MATH_SINF_H
+#define LLVM_LIBC_SHARED_MATH_SINF_H
+
+#include "shared/libc_common.h"
+#include "src/__support/math/sinf.h"
+
+namespace LIBC_NAMESPACE_DECL {
+namespace shared {
+
+using math::sinf;
+
+} // namespace shared
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SHARED_MATH_SINF_H
diff --git a/libc/src/__support/math/CMakeLists.txt b/libc/src/__support/math/CMakeLists.txt
index d18ebc03ba7b8..9fe1eb967d5e0 100644
--- a/libc/src/__support/math/CMakeLists.txt
+++ b/libc/src/__support/math/CMakeLists.txt
@@ -1286,3 +1286,19 @@ add_header_library(
     libc.src.__support.FPUtil.polyeval
     libc.src.__support.macros.optimization
 )
+
+add_header_library(
+  sinf
+  HDRS
+    sinf.h
+  DEPENDS
+    .range_reduction
+    .sincosf_utils
+    libc.src.__support.FPUtil.basic_operations
+    libc.src.__support.FPUtil.fenv_impl
+    libc.src.__support.FPUtil.fma
+    libc.src.__support.FPUtil.fp_bits
+    libc.src.__support.FPUtil.polyeval
+    libc.src.__support.FPUtil.rounding_mode
+    libc.src.__support.macros.optimization
+)
diff --git a/libc/src/__support/math/sinf.h b/libc/src/__support/math/sinf.h
new file mode 100644
index 0000000000000..b07ebba61bcd9
--- /dev/null
+++ b/libc/src/__support/math/sinf.h
@@ -0,0 +1,194 @@
+//===-- Single-precision sin function -------------------------------------===//
+//
+// 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 LIBC_SRC___SUPPORT_MATH_SINF_H
+#define LIBC_SRC___SUPPORT_MATH_SINF_H
+
+#include "src/__support/FPUtil/BasicOperations.h"
+#include "src/__support/FPUtil/FEnvImpl.h"
+#include "src/__support/FPUtil/FPBits.h"
+#include "src/__support/FPUtil/PolyEval.h"
+#include "src/__support/FPUtil/multiply_add.h"
+#include "src/__support/FPUtil/rounding_mode.h"
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/optimization.h"            // LIBC_UNLIKELY
+#include "src/__support/macros/properties/cpu_features.h" // LIBC_TARGET_CPU_HAS_FMA
+
+#if defined(LIBC_MATH_HAS_SKIP_ACCURATE_PASS) &&                               \
+    defined(LIBC_MATH_HAS_INTERMEDIATE_COMP_IN_FLOAT) &&                       \
+    defined(LIBC_TARGET_CPU_HAS_FMA_FLOAT)
+
+#include "sincosf_float_eval.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+namespace math {
+
+LIBC_INLINE static float sinf(float x) {
+  return math::sincosf_float_eval::sincosf_eval</*IS_SIN*/ true>(x);
+}
+
+} // namespace math
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#else // !LIBC_MATH_HAS_INTERMEDIATE_COMP_IN_FLOAT
+
+#include "src/__support/math/sincosf_utils.h"
+
+#ifdef LIBC_TARGET_CPU_HAS_FMA_DOUBLE
+#include "src/__support/math/range_reduction_fma.h"
+#else // !LIBC_TARGET_CPU_HAS_FMA_DOUBLE
+#include "src/__support/math/range_reduction.h"
+#endif // LIBC_TARGET_CPU_HAS_FMA_DOUBLE
+
+namespace LIBC_NAMESPACE_DECL {
+
+namespace math {
+
+LIBC_INLINE static float sinf(float x) {
+  using FPBits = typename fputil::FPBits<float>;
+  FPBits xbits(x);
+
+  uint32_t x_u = xbits.uintval();
+  uint32_t x_abs = x_u & 0x7fff'ffffU;
+  double xd = static_cast<double>(x);
+
+  // Range reduction:
+  // For |x| > pi/32, we perform range reduction as follows:
+  // Find k and y such that:
+  //   x = (k + y) * pi/32
+  //   k is an integer
+  //   |y| < 0.5
+  // For small range (|x| < 2^45 when FMA instructions are available, 2^22
+  // otherwise), this is done by performing:
+  //   k = round(x * 32/pi)
+  //   y = x * 32/pi - k
+  // For large range, we will omit all the higher parts of 32/pi such that the
+  // least significant bits of their full products with x are larger than 63,
+  // since sin((k + y + 64*i) * pi/32) = sin(x + i * 2pi) = sin(x).
+  //
+  // When FMA instructions are not available, we store the digits of 32/pi in
+  // chunks of 28-bit precision.  This will make sure that the products:
+  //   x * THIRTYTWO_OVER_PI_28[i] are all exact.
+  // When FMA instructions are available, we simply store the digits of 32/pi in
+  // chunks of doubles (53-bit of precision).
+  // So when multiplying by the largest values of single precision, the
+  // resulting output should be correct up to 2^(-208 + 128) ~ 2^-80.  By the
+  // worst-case analysis of range reduction, |y| >= 2^-38, so this should give
+  // us more than 40 bits of accuracy. For the worst-case estimation of range
+  // reduction, see for instances:
+  //   Elementary Functions by J-M. Muller, Chapter 11,
+  //   Handbook of Floating-Point Arithmetic by J-M. Muller et. al.,
+  //   Chapter 10.2.
+  //
+  // Once k and y are computed, we then deduce the answer by the sine of sum
+  // formula:
+  //   sin(x) = sin((k + y)*pi/32)
+  //          = sin(y*pi/32) * cos(k*pi/32) + cos(y*pi/32) * sin(k*pi/32)
+  // The values of sin(k*pi/32) and cos(k*pi/32) for k = 0..31 are precomputed
+  // and stored using a vector of 32 doubles. Sin(y*pi/32) and cos(y*pi/32) are
+  // computed using degree-7 and degree-6 minimax polynomials generated by
+  // Sollya respectively.
+
+  // |x| <= pi/16
+  if (LIBC_UNLIKELY(x_abs <= 0x3e49'0fdbU)) {
+
+    // |x| < 0x1.d12ed2p-12f
+    if (LIBC_UNLIKELY(x_abs < 0x39e8'9769U)) {
+      if (LIBC_UNLIKELY(x_abs == 0U)) {
+        // For signed zeros.
+        return x;
+      }
+      // When |x| < 2^-12, the relative error of the approximation sin(x) ~ x
+      // is:
+      //   |sin(x) - x| / |sin(x)| < |x^3| / (6|x|)
+      //                           = x^2 / 6
+      //                           < 2^-25
+      //                           < epsilon(1)/2.
+      // So the correctly rounded values of sin(x) are:
+      //   = x - sign(x)*eps(x) if rounding mode = FE_TOWARDZERO,
+      //                        or (rounding mode = FE_UPWARD and x is
+      //                        negative),
+      //   = x otherwise.
+      // To simplify the rounding decision and make it more efficient, we use
+      //   fma(x, -2^-25, x) instead.
+      // An exhaustive test shows that this formula work correctly for all
+      // rounding modes up to |x| < 0x1.c555dep-11f.
+      // Note: to use the formula x - 2^-25*x to decide the correct rounding, we
+      // do need fma(x, -2^-25, x) to prevent underflow caused by -2^-25*x when
+      // |x| < 2^-125. For targets without FMA instructions, we simply use
+      // double for intermediate results as it is more efficient than using an
+      // emulated version of FMA.
+#if defined(LIBC_TARGET_CPU_HAS_FMA_FLOAT)
+      return fputil::multiply_add(x, -0x1.0p-25f, x);
+#else
+      return static_cast<float>(fputil::multiply_add(xd, -0x1.0p-25, xd));
+#endif // LIBC_TARGET_CPU_HAS_FMA_FLOAT
+    }
+
+    // |x| < pi/16.
+    double xsq = xd * xd;
+
+    // Degree-9 polynomial approximation:
+    //   sin(x) ~ x + a_3 x^3 + a_5 x^5 + a_7 x^7 + a_9 x^9
+    //          = x (1 + a_3 x^2 + ... + a_9 x^8)
+    //          = x * P(x^2)
+    // generated by Sollya with the following commands:
+    // > display = hexadecimal;
+    // > Q = fpminimax(sin(x)/x, [|0, 2, 4, 6, 8|], [|1, D...|], [0, pi/16]);
+    double result =
+        fputil::polyeval(xsq, 1.0, -0x1.55555555554c6p-3, 0x1.1111111085e65p-7,
+                         -0x1.a019f70fb4d4fp-13, 0x1.718d179815e74p-19);
+    return static_cast<float>(xd * result);
+  }
+
+#ifndef LIBC_MATH_HAS_SKIP_ACCURATE_PASS
+  if (LIBC_UNLIKELY(x_abs == 0x4619'9998U)) { // x = 0x1.33333p13
+    float r = -0x1.63f4bap-2f;
+    int rounding = fputil::quick_get_round();
+    if ((rounding == FE_DOWNWARD && xbits.is_pos()) ||
+        (rounding == FE_UPWARD && xbits.is_neg()))
+      r = -0x1.63f4bcp-2f;
+    return xbits.is_neg() ? -r : r;
+  }
+#endif // !LIBC_MATH_HAS_SKIP_ACCURATE_PASS
+
+  if (LIBC_UNLIKELY(x_abs >= 0x7f80'0000U)) {
+    if (xbits.is_signaling_nan()) {
+      fputil::raise_except_if_required(FE_INVALID);
+      return FPBits::quiet_nan().get_val();
+    }
+
+    if (x_abs == 0x7f80'0000U) {
+      fputil::set_errno_if_required(EDOM);
+      fputil::raise_except_if_required(FE_INVALID);
+    }
+    return x + FPBits::quiet_nan().get_val();
+  }
+
+  // Combine the results with the sine of sum formula:
+  //   sin(x) = sin((k + y)*pi/32)
+  //          = sin(y*pi/32) * cos(k*pi/32) + cos(y*pi/32) * sin(k*pi/32)
+  //          = sin_y * cos_k + (1 + cosm1_y) * sin_k
+  //          = sin_y * cos_k + (cosm1_y * sin_k + sin_k)
+  double sin_k, cos_k, sin_y, cosm1_y;
+
+  sincosf_eval(xd, x_abs, sin_k, cos_k, sin_y, cosm1_y);
+
+  return static_cast<float>(fputil::multiply_add(
+      sin_y, cos_k, fputil::multiply_add(cosm1_y, sin_k, sin_k)));
+}
+
+} // namespace math
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LIBC_MATH_HAS_INTERMEDIATE_COMP_IN_FLOAT
+
+#endif // LIBC_SRC___SUPPORT_MATH_SINF_H
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index 3addfab6c6603..d07ef01567d6f 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -372,16 +372,8 @@ add_entrypoint_object(
   HDRS
     ../sinf.h
   DEPENDS
-    libc.src.__support.math.range_reduction
-    libc.src.__support.math.sincosf_utils
+    libc.src.__support.math.sinf
     libc.src.errno.errno
-    libc.src.__support.FPUtil.basic_operations
-    libc.src.__support.FPUtil.fenv_impl
-    libc.src.__support.FPUtil.fp_bits
-    libc.src.__support.FPUtil.fma
-    libc.src.__support.FPUtil.polyeval
-    libc.src.__support.FPUtil.rounding_mode
-    libc.src.__support.macros.optimization
 )
 
 add_entrypoint_object(
diff --git a/libc/src/math/generic/sinf.cpp b/libc/src/math/generic/sinf.cpp
index c362628fb106c..648d57b82488c 100644
--- a/libc/src/math/generic/sinf.cpp
+++ b/libc/src/math/generic/sinf.cpp
@@ -7,176 +7,10 @@
 //===----------------------------------------------------------------------===//
 
 #include "src/math/sinf.h"
-#include "src/__support/FPUtil/BasicOperations.h"
-#include "src/__support/FPUtil/FEnvImpl.h"
-#include "src/__support/FPUtil/FPBits.h"
-#include "src/__support/FPUtil/PolyEval.h"
-#include "src/__support/FPUtil/multiply_add.h"
-#include "src/__support/FPUtil/rounding_mode.h"
-#include "src/__support/common.h"
-#include "src/__support/macros/config.h"
-#include "src/__support/macros/optimization.h"            // LIBC_UNLIKELY
-#include "src/__support/macros/properties/cpu_features.h" // LIBC_TARGET_CPU_HAS_FMA
-
-#if defined(LIBC_MATH_HAS_SKIP_ACCURATE_PASS) &&                               \
-    defined(LIBC_MATH_HAS_INTERMEDIATE_COMP_IN_FLOAT) &&                       \
-    defined(LIBC_TARGET_CPU_HAS_FMA_FLOAT)
-
-#include "src/__support/math/sincosf_float_eval.h"
+#include "src/__support/math/sinf.h"
 
 namespace LIBC_NAMESPACE_DECL {
 
-LLVM_LIBC_FUNCTION(float, sinf, (float x)) {
-  return math::sincosf_float_eval::sincosf_eval</*IS_SIN*/ true>(x);
-}
-
-} // namespace LIBC_NAMESPACE_DECL
-
-#else // !LIBC_MATH_HAS_INTERMEDIATE_COMP_IN_FLOAT
-
-#include "src/__support/math/sincosf_utils.h"
-
-#ifdef LIBC_TARGET_CPU_HAS_FMA_DOUBLE
-#include "src/__support/math/range_reduction_fma.h"
-#else // !LIBC_TARGET_CPU_HAS_FMA_DOUBLE
-#include "src/__support/math/range_reduction.h"
-#endif // LIBC_TARGET_CPU_HAS_FMA_DOUBLE
-
-namespace LIBC_NAMESPACE_DECL {
-
-LLVM_LIBC_FUNCTION(float, sinf, (float x)) {
-  using FPBits = typename fputil::FPBits<float>;
-  FPBits xbits(x);
-
-  uint32_t x_u = xbits.uintval();
-  uint32_t x_abs = x_u & 0x7fff'ffffU;
-  double xd = static_cast<double>(x);
-
-  // Range reduction:
-  // For |x| > pi/32, we perform range reduction as follows:
-  // Find k and y such that:
-  //   x = (k + y) * pi/32
-  //   k is an integer
-  //   |y| < 0.5
-  // For small range (|x| < 2^45 when FMA instructions are available, 2^22
-  // otherwise), this is done by performing:
-  //   k = round(x * 32/pi)
-  //   y = x * 32/pi - k
-  // For large range, we will omit all the higher parts of 32/pi such that the
-  // least significant bits of their full products with x are larger than 63,
-  // since sin((k + y + 64*i) * pi/32) = sin(x + i * 2pi) = sin(x).
-  //
-  // When FMA instructions are not available, we store the digits of 32/pi in
-  // chunks of 28-bit precision.  This will make sure that the products:
-  //   x * THIRTYTWO_OVER_PI_28[i] are all exact.
-  // When FMA instructions are available, we simply store the digits of 32/pi in
-  // chunks of doubles (53-bit of precision).
-  // So when multiplying by the largest values of single precision, the
-  // resulting output should be correct up to 2^(-208 + 128) ~ 2^-80.  By the
-  // worst-case analysis of range reduction, |y| >= 2^-38, so this should give
-  // us more than 40 bits of accuracy. For the worst-case estimation of range
-  // reduction, see for instances:
-  //   Elementary Functions by J-M. Muller, Chapter 11,
-  //   Handbook of Floating-Point Arithmetic by J-M. Muller et. al.,
-  //   Chapter 10.2.
-  //
-  // Once k and y are computed, we then deduce the answer by the sine of sum
-  // formula:
-  //   sin(x) = sin((k + y)*pi/32)
-  //          = sin(y*pi/32) * cos(k*pi/32) + cos(y*pi/32) * sin(k*pi/32)
-  // The values of sin(k*pi/32) and cos(k*pi/32) for k = 0..31 are precomputed
-  // and stored using a vector of 32 doubles. Sin(y*pi/32) and cos(y*pi/32) are
-  // computed using degree-7 and degree-6 minimax polynomials generated by
-  // Sollya respectively.
-
-  // |x| <= pi/16
-  if (LIBC_UNLIKELY(x_abs <= 0x3e49'0fdbU)) {
-
-    // |x| < 0x1.d12ed2p-12f
-    if (LIBC_UNLIKELY(x_abs < 0x39e8'9769U)) {
-      if (LIBC_UNLIKELY(x_abs == 0U)) {
-        // For signed zeros.
-        return x;
-      }
-      // When |x| < 2^-12, the relative error of the approximation sin(x) ~ x
-      // is:
-      //   |sin(x) - x| / |sin(x)| < |x^3| / (6|x|)
-      //                           = x^2 / 6
-      //                           < 2^-25
-      //                           < epsilon(1)/2.
-      // So the correctly rounded values of sin(x) are:
-      //   = x - sign(x)*eps(x) if rounding mode = FE_TOWARDZERO,
-      //                        or (rounding mode = FE_UPWARD and x is
-      //                        negative),
-      //   = x otherwise.
-      // To simplify the rounding decision and make it more efficient, we use
-      //   fma(x, -2^-25, x) instead.
-      // An exhaustive test shows that this formula work correctly for all
-      // rounding modes up to |x| < 0x1.c555dep-11f.
-      // Note: to use the formula x - 2^-25*x to decide the correct rounding, we
-      // do need fma(x, -2^-25, x) to prevent underflow caused by -2^-25*x when
-      // |x| < 2^-125. For targets without FMA instructions, we simply use
-      // double for intermediate results as it is more efficient than using an
-      // emulated version of FMA.
-#if defined(LIBC_TARGET_CPU_HAS_FMA_FLOAT)
-      return fputil::multiply_add(x, -0x1.0p-25f, x);
-#else
-      return static_cast<float>(fputil::multiply_add(xd, -0x1.0p-25, xd));
-#endif // LIBC_TARGET_CPU_HAS_FMA_FLOAT
-    }
-
-    // |x| < pi/16.
-    double xsq = xd * xd;
-
-    // Degree-9 polynomial approximation:
-    //   sin(x) ~ x + a_3 x^3 + a_5 x^5 + a_7 x^7 + a_9 x^9
-    //          = x (1 + a_3 x^2 + ... + a_9 x^8)
-    //          = x * P(x^2)
-    // generated by Sollya with the following commands:
-    // > display = hexadecimal;
-    // > Q = fpminimax(sin(x)/x, [|0, 2, 4, 6, 8|], [|1, D...|], [0, pi/16]);
-    double result =
-        fputil::polyeval(xsq, 1.0, -0x1.55555555554c6p-3, 0x1.1111111085e65p-7,
-                         -0x1.a019f70fb4d4fp-13, 0x1.718d179815e74p-19);
-    return static_cast<float>(xd * result);
-  }
-
-#ifndef LIBC_MATH_HAS_SKIP_ACCURATE_PASS
-  if (LIBC_UNLIKELY(x_abs == 0x4619'9998U)) { // x = 0x1.33333p13
-    float r = -0x1.63f4bap-2f;
-    int rounding = fputil::quick_get_round();
-    if ((rounding == FE_DOWNWARD && xbits.is_pos()) ||
-        (rounding == FE_UPWARD && xbits.is_neg()))
-      r = -0x1.63f4bcp-2f;
-    return xbits.is_neg() ? -r : r;
-  }
-#endif // !LIBC_MATH_HAS_SKIP_ACCURATE_PASS
-
-  if (LIBC_UNLIKELY(x_abs >= 0x7f80'0000U)) {
-    if (xbits.is_signaling_nan()) {
-      fputil::raise_except_if_required(FE_INVALID);
-      return FPBits::quiet_nan().get_val();
-    }
-
-    if (x_abs == 0x7f80'0000U) {
-      fputil::set_errno_if_required(EDOM);
-      fputil::raise_except_if_required(FE_INVALID);
-    }
-    return x + FPBits::quiet_nan().get_val();
-  }
-
-  // Combine the results with the sine of sum formula:
-  //   sin(x) = sin((k + y)*pi/32)
-  //          = sin(y*pi/32) * cos(k*pi/32) + cos(y*pi/32) * sin(k*pi/32)
-  //          = sin_y * cos_k + (1 + cosm1_y) * sin_k
-  //          = sin_y * cos_k + (cosm1_y * sin_k + sin_k)
-  double sin_k, cos_k, sin_y, cosm1_y;
-
-  sincosf_eval(xd, x_abs, sin_k, cos_k, sin_y, cosm1_y);
-
-  return static_cast<float>(fputil::multiply_add(
-      sin_y, cos_k, fputil::multiply_add(cosm1_y, sin_k, sin_k)));
-}
+LLVM_LIBC_FUNCTION(float, sinf, (float x)) { return math::sinf(x); }
 
 } // namespace LIBC_NAMESPACE_DECL
-#endif // LIBC_MATH_HAS_INTERMEDIATE_COMP_IN_FLOAT
diff --git a/libc/test/shared/CMakeLists.txt b/libc/test/shared/CMakeLists.txt
index dabf5f6b168cc..e3c9fa097f638 100644
--- a/libc/test/shared/CMakeLists.txt
+++ b/libc/test/shared/CMakeLists.txt
@@ -82,4 +82,5 @@ add_fp_unittest(
     libc.src.__support.math.sin
     libc.src.__support.math.tan
     libc.src.__support.math.tanf
+    libc.src.__support.math.sinf
 )
diff --git a/libc/test/shared/shared_math_test.cpp b/libc/test/shared/shared_math_test.cpp
index cd389c41cb7b2..4c107d786d3a7 100644
--- a/libc/test/shared/shared_math_test.cpp
+++ b/libc/test/shared/shared_math_test.cpp
@@ -89,6 +89,7 @@ TEST(LlvmLibcSharedMathTest, AllFloat) {
   EXPECT_EQ(long(0), LIBC_NAMESPACE::shared::llogbf(1.0f));
   EXPECT_FP_EQ(0x0p+0f, LIBC_NAMESPACE::shared::logbf(1.0f));
   EXPECT_FP_EQ(0x1p+0f, LIBC_NAMESPACE::shared::rsqrtf(1.0f));
+  EXPECT_FP_EQ(0.0f, LIBC_NAMESPACE::shared::sinf(0.0f));
   EXPECT_FP_EQ(0.0f, LIBC_NAMESPACE::shared::tanf(0.0f));
 }
 
diff --git a/utils/bazel/llvm-project-overlay/libc/BUILD.bazel b/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
index 346864d2276c0..66e87df60135d 100644
--- a/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
+++ b/utils/bazel/llvm-project-overlay/libc/BUILD.bazel
@@ -3455,6 +3455,20 @@ libc_support_library(
     ],
 )
 
+libc_support_library(
+    name = "__support_math_sinf",
+    hdrs = ["src/__support/math/sinf.h"],
+    deps = [
+        ":__support_fputil_fma",
+        ":__support_fputil_polyeval",
+        ":__support_fputil_rounding_mode",
+        ":__support_macros_optimization",
+        ":__support_macros_properties_cpu_features",
+        ":__support_range_reduction",
+        ":__support_sincosf_utils",
+    ],
+)
+
 libc_support_library(
     name = "__support_math_sinhfcoshf_utils",
     hdrs = ["src/__support/math/sinhfcoshf_utils.h"],
@@ -5058,13 +5072,7 @@ libc_math_function(
 libc_math_function(
     name = "sinf",
     additional_deps = [
-        ":__support_fputil_fma",
-        ":__support_fputil_polyeval",
-        ":__support_fputil_rounding_mode",
-        ":__support_macros_optimization",
-        ":__support_macros_properties_cpu_features",
-        ":__support_range_reduction",
-        ":__support_sincosf_utils",
+        ":__support_math_sinf",
     ],
 )

bassiounix

LGTM with some nits.

libc/src/__support/math/sinf.h

libc/test/shared/CMakeLists.txt

nico · 2026-01-26T20:26:10Z

Thanks, all done! I'll update my script.

...due to the feedback on llvm#177963. Patch generated by running rg -l '\bLIBC_SRC' libc/src/__support/math | \ xargs sed -i '' -e \ 's/LIBC_SRC___SUPPORT_MATH_\([A-Z0-9_]*\)/LLVM_LIBC_SRC___SUPPORT_MATH_\1/'

...due to the feedback on #177963. Patch generated by running rg -l '\bLIBC_SRC' libc/src/__support/math | \ xargs sed -i '' -e \ 's/LIBC_SRC___SUPPORT_MATH_\([A-Z0-9_]*\)/LLVM_LIBC_SRC___SUPPORT_MATH_\1/'

...due to the feedback on llvm#177963. Patch generated by running rg -l '\bLIBC_SRC' libc/src/__support/math | \ xargs sed -i '' -e \ 's/LIBC_SRC___SUPPORT_MATH_\([A-Z0-9_]*\)/LLVM_LIBC_SRC___SUPPORT_MATH_\1/'

…port/math folder. (llvm#177963) Part of llvm#147386 in preparation for: https://discourse.llvm.org/t/rfc-make-clang-builtin-math-functions-constexpr-with-llvm-libc-to-support-c-23-constexpr-math-functions/86450

...due to the feedback on llvm#177963. Patch generated by running rg -l '\bLIBC_SRC' libc/src/__support/math | \ xargs sed -i '' -e \ 's/LIBC_SRC___SUPPORT_MATH_\([A-Z0-9_]*\)/LLVM_LIBC_SRC___SUPPORT_MATH_\1/'

…port/math folder. (llvm#177963) Part of llvm#147386 in preparation for: https://discourse.llvm.org/t/rfc-make-clang-builtin-math-functions-constexpr-with-llvm-libc-to-support-c-23-constexpr-math-functions/86450

...due to the feedback on llvm#177963. Patch generated by running rg -l '\bLIBC_SRC' libc/src/__support/math | \ xargs sed -i '' -e \ 's/LIBC_SRC___SUPPORT_MATH_\([A-Z0-9_]*\)/LLVM_LIBC_SRC___SUPPORT_MATH_\1/'

[libc][math] Refactor sinf implementation to header-only in src/__sup…

6247843

…port/math folder. Part of llvm#147386 in preparation for: https://discourse.llvm.org/t/rfc-make-clang-builtin-math-functions-constexpr-with-llvm-libc-to-support-c-23-constexpr-math-functions/86450

nico requested review from aaronmondal, keith and rupprecht as code owners January 26, 2026 14:09

llvmbot added libc bazel "Peripheral" support tier build system: utils/bazel labels Jan 26, 2026

nico requested a review from bassiounix January 26, 2026 14:09

nico changed the title ~~[libc][math] Refactor sinf implementation to header-only in src/__sup…~~ [libc][math] Refactor sinf implementation to header-only in src/__support/math folder. Jan 26, 2026

alphabetize cmake

87f1675

bassiounix mentioned this pull request Jan 26, 2026

[libc][math] Tracking Issue - Refactor Higher Math Functions to Header Only #147386

Open

bassiounix approved these changes Jan 26, 2026

View reviewed changes

libc/src/__support/math/sinf.h Outdated Show resolved Hide resolved

libc/src/__support/math/sinf.h Outdated Show resolved Hide resolved

libc/test/shared/CMakeLists.txt Outdated Show resolved Hide resolved

comments

1f94351

nico mentioned this pull request Jan 26, 2026

[libm] Update include guards in libc/src/__support/math #178055

Merged

nico merged commit 5caf27e into llvm:main Jan 26, 2026
27 of 28 checks passed

nico deleted the libm-sinf branch January 26, 2026 21:41

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[libc][math] Refactor sinf implementation to header-only in src/__support/math folder.#177963

[libc][math] Refactor sinf implementation to header-only in src/__support/math folder.#177963
nico merged 3 commits intollvm:mainfrom
nico:libm-sinf

nico commented Jan 26, 2026 •

edited

Loading

Uh oh!

llvmbot commented Jan 26, 2026

Uh oh!

bassiounix left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

nico commented Jan 26, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

nico commented Jan 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

llvmbot commented Jan 26, 2026

Uh oh!

bassiounix left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

nico commented Jan 26, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

nico commented Jan 26, 2026 •

edited

Loading