[libc][math][c++23] Add bf16{add,sub}{,f,l,f128} math functions #152774

krishna2803 · 2025-08-08T18:21:07Z

This PR adds implements following basic math functions for BFloat16 type along with the tests:

bf16add
bf16addf
bf16addl
bf16addf128
bf16sub
bf16subf
bf16subl
bf16subf128

Signed-off-by: Krishna Pandey <[email protected]>

llvmbot · 2025-08-08T18:21:43Z

@llvm/pr-subscribers-libc

Author: Krishna Pandey (krishna2803)

Changes

This PR adds implements following basic math functions for BFloat16 type along with the tests:

bf16add
bf16addf
bf16addl
bf16addf128
bf16sub
bf16subf
bf16subl
bf16subf128

Patch is 39.71 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/152774.diff

39 Files Affected:

(modified) libc/config/linux/x86_64/entrypoints.txt (+14)
(modified) libc/src/math/CMakeLists.txt (+10)
(added) libc/src/math/bf16add.h (+21)
(added) libc/src/math/bf16addf.h (+21)
(added) libc/src/math/bf16addf128.h (+21)
(added) libc/src/math/bf16addl.h (+21)
(added) libc/src/math/bf16sub.h (+21)
(added) libc/src/math/bf16subf.h (+21)
(added) libc/src/math/bf16subf128.h (+21)
(added) libc/src/math/bf16subl.h (+21)
(modified) libc/src/math/generic/CMakeLists.txt (+113)
(added) libc/src/math/generic/bf16add.cpp (+22)
(added) libc/src/math/generic/bf16addf.cpp (+22)
(added) libc/src/math/generic/bf16addf128.cpp (+22)
(added) libc/src/math/generic/bf16addl.cpp (+22)
(added) libc/src/math/generic/bf16sub.cpp (+22)
(added) libc/src/math/generic/bf16subf.cpp (+22)
(added) libc/src/math/generic/bf16subf128.cpp (+22)
(added) libc/src/math/generic/bf16subl.cpp (+22)
(modified) libc/test/UnitTest/FPMatcher.h (+1-1)
(modified) libc/test/src/math/CMakeLists.txt (+112)
(added) libc/test/src/math/bf16add_test.cpp (+14)
(added) libc/test/src/math/bf16addf128_test.cpp (+14)
(added) libc/test/src/math/bf16addf_test.cpp (+14)
(added) libc/test/src/math/bf16addl_test.cpp (+14)
(added) libc/test/src/math/bf16sub_test.cpp (+14)
(added) libc/test/src/math/bf16subf128_test.cpp (+14)
(added) libc/test/src/math/bf16subf_test.cpp (+14)
(added) libc/test/src/math/bf16subl_test.cpp (+14)
(modified) libc/test/src/math/smoke/CMakeLists.txt (+128)
(added) libc/test/src/math/smoke/bf16add_test.cpp (+14)
(added) libc/test/src/math/smoke/bf16addf128_test.cpp (+14)
(added) libc/test/src/math/smoke/bf16addf_test.cpp (+14)
(added) libc/test/src/math/smoke/bf16addl_test.cpp (+14)
(added) libc/test/src/math/smoke/bf16sub_test.cpp (+14)
(added) libc/test/src/math/smoke/bf16subf128_test.cpp (+14)
(added) libc/test/src/math/smoke/bf16subf_test.cpp (+14)
(added) libc/test/src/math/smoke/bf16subl_test.cpp (+14)
(modified) libc/utils/MPFRWrapper/MPFRUtils.cpp (+33)

diff --git a/libc/config/linux/x86_64/entrypoints.txt b/libc/config/linux/x86_64/entrypoints.txt
index 066dc21497691..0a2b24b420ac2 100644
--- a/libc/config/linux/x86_64/entrypoints.txt
+++ b/libc/config/linux/x86_64/entrypoints.txt
@@ -896,6 +896,12 @@ endif()
 list(APPEND TARGET_LIBM_ENTRYPOINTS
   # bfloat16 entrypoints
   libc.src.math.ceilbf16
+  libc.src.math.bf16add
+  libc.src.math.bf16addf
+  libc.src.math.bf16addl
+  libc.src.math.bf16sub
+  libc.src.math.bf16subf
+  libc.src.math.bf16subl
   libc.src.math.fabsbf16
   libc.src.math.floorbf16
   libc.src.math.roundbf16
@@ -903,6 +909,14 @@ list(APPEND TARGET_LIBM_ENTRYPOINTS
   libc.src.math.truncbf16
 )
 
+if(LIBC_TYPES_HAS_FLOAT128)
+  list(APPEND TARGET_LIBM_ENTRYPOINTS
+    # math.h C++23 mixed bfloat16 and _Float128 entrypoints
+    libc.src.math.bf16addf128
+    libc.src.math.bf16subf128
+  )
+endif()
+
 if(LIBC_COMPILER_HAS_FIXED_POINT)
   list(APPEND TARGET_LIBM_ENTRYPOINTS
     # stdfix.h _Fract and _Accum entrypoints
diff --git a/libc/src/math/CMakeLists.txt b/libc/src/math/CMakeLists.txt
index c3840d3c4aa89..660c3681717c8 100644
--- a/libc/src/math/CMakeLists.txt
+++ b/libc/src/math/CMakeLists.txt
@@ -563,3 +563,13 @@ add_math_entrypoint_object(ufromfpxf)
 add_math_entrypoint_object(ufromfpxl)
 add_math_entrypoint_object(ufromfpxf16)
 add_math_entrypoint_object(ufromfpxf128)
+
+add_math_entrypoint_object(bf16add)
+add_math_entrypoint_object(bf16addf)
+add_math_entrypoint_object(bf16addl)
+add_math_entrypoint_object(bf16addf128)
+
+add_math_entrypoint_object(bf16sub)
+add_math_entrypoint_object(bf16subf)
+add_math_entrypoint_object(bf16subl)
+add_math_entrypoint_object(bf16subf128)
diff --git a/libc/src/math/bf16add.h b/libc/src/math/bf16add.h
new file mode 100644
index 0000000000000..a29970eb334fa
--- /dev/null
+++ b/libc/src/math/bf16add.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for bf16add -----------------------*- 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_MATH_BF16ADD_H
+#define LLVM_LIBC_SRC_MATH_BF16ADD_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+bfloat16 bf16add(double x, double y);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_BF16ADD_H
diff --git a/libc/src/math/bf16addf.h b/libc/src/math/bf16addf.h
new file mode 100644
index 0000000000000..80a5e2a7640df
--- /dev/null
+++ b/libc/src/math/bf16addf.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for bf16addf ----------------------*- 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_MATH_BF16ADDF_H
+#define LLVM_LIBC_SRC_MATH_BF16ADDF_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+bfloat16 bf16addf(float x, float y);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_BF16ADDF_H
diff --git a/libc/src/math/bf16addf128.h b/libc/src/math/bf16addf128.h
new file mode 100644
index 0000000000000..3c2f3a15eb39c
--- /dev/null
+++ b/libc/src/math/bf16addf128.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for bf16addf128 -------------------*- 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_MATH_BF16ADDF128_H
+#define LLVM_LIBC_SRC_MATH_BF16ADDF128_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+bfloat16 bf16addf128(float128 x, float128 y);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_BF16ADDF128_H
diff --git a/libc/src/math/bf16addl.h b/libc/src/math/bf16addl.h
new file mode 100644
index 0000000000000..a9e7d68660728
--- /dev/null
+++ b/libc/src/math/bf16addl.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for bf16addl ----------------------*- 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_MATH_BF16ADDL_H
+#define LLVM_LIBC_SRC_MATH_BF16ADDL_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+bfloat16 bf16addl(long double x, long double y);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_BF16ADDL_H
diff --git a/libc/src/math/bf16sub.h b/libc/src/math/bf16sub.h
new file mode 100644
index 0000000000000..8108e9146859f
--- /dev/null
+++ b/libc/src/math/bf16sub.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for bf16sub -----------------------*- 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_MATH_BF16SUB_H
+#define LLVM_LIBC_SRC_MATH_BF16SUB_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+bfloat16 bf16sub(double x, double y);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_BF16SUB_H
diff --git a/libc/src/math/bf16subf.h b/libc/src/math/bf16subf.h
new file mode 100644
index 0000000000000..1bd79bfe119ce
--- /dev/null
+++ b/libc/src/math/bf16subf.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for bf16subf ----------------------*- 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_MATH_BF16SUBF_H
+#define LLVM_LIBC_SRC_MATH_BF16SUBF_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+bfloat16 bf16subf(float x, float y);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_BF16SUBF_H
diff --git a/libc/src/math/bf16subf128.h b/libc/src/math/bf16subf128.h
new file mode 100644
index 0000000000000..19590e8c67508
--- /dev/null
+++ b/libc/src/math/bf16subf128.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for bf16subf128 -------------------*- 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_MATH_BF16SUBF128_H
+#define LLVM_LIBC_SRC_MATH_BF16SUBF128_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+bfloat16 bf16subf128(float128 x, float128 y);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_BF16SUBF128_H
diff --git a/libc/src/math/bf16subl.h b/libc/src/math/bf16subl.h
new file mode 100644
index 0000000000000..13b2093a92ffa
--- /dev/null
+++ b/libc/src/math/bf16subl.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for bf16subl ----------------------*- 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_MATH_BF16SUBL_H
+#define LLVM_LIBC_SRC_MATH_BF16SUBL_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+bfloat16 bf16subl(long double x, long double y);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_BF16SUBL_H
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index 893606609dfc3..5aeacc85d4d5e 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -4911,3 +4911,116 @@ add_header_library(
     libc.src.__support.math.expf16_utils
     libc.src.__support.math.exp10_float16_constants
 )
+
+add_entrypoint_object(
+  bf16add
+  SRCS
+    bf16add.cpp
+  HDRS
+    ../bf16add.h
+  DEPENDS
+    libc.src.__support.common
+    libc.src.__support.FPUtil.bfloat16
+    libc.src.__support.FPUtil.generic.add_sub
+    libc.src.__support.macros.config
+    libc.src.__support.macros.properties.types
+)
+
+add_entrypoint_object(
+  bf16addf
+  SRCS
+    bf16addf.cpp
+  HDRS
+    ../bf16addf.h
+  DEPENDS
+    libc.src.__support.common
+    libc.src.__support.FPUtil.bfloat16
+    libc.src.__support.FPUtil.generic.add_sub
+    libc.src.__support.macros.config
+    libc.src.__support.macros.properties.types
+)
+
+add_entrypoint_object(
+  bf16addl
+  SRCS
+    bf16addl.cpp
+  HDRS
+    ../bf16addl.h
+  DEPENDS
+    libc.src.__support.common
+    libc.src.__support.FPUtil.bfloat16
+    libc.src.__support.FPUtil.generic.add_sub
+    libc.src.__support.macros.config
+    libc.src.__support.macros.properties.types
+)
+
+add_entrypoint_object(
+  bf16addf128
+  SRCS
+    bf16addf128.cpp
+  HDRS
+    ../bf16addf128.h
+  DEPENDS
+    libc.src.__support.common
+    libc.src.__support.FPUtil.bfloat16
+    libc.src.__support.FPUtil.generic.add_sub
+    libc.src.__support.macros.config
+    libc.src.__support.macros.properties.types
+)
+
+
+add_entrypoint_object(
+  bf16sub
+  SRCS
+    bf16sub.cpp
+  HDRS
+    ../bf16sub.h
+  DEPENDS
+    libc.src.__support.common
+    libc.src.__support.FPUtil.bfloat16
+    libc.src.__support.FPUtil.generic.add_sub
+    libc.src.__support.macros.config
+    libc.src.__support.macros.properties.types
+)
+
+add_entrypoint_object(
+  bf16subf
+  SRCS
+    bf16subf.cpp
+  HDRS
+    ../bf16subf.h
+  DEPENDS
+    libc.src.__support.common
+    libc.src.__support.FPUtil.bfloat16
+    libc.src.__support.FPUtil.generic.add_sub
+    libc.src.__support.macros.config
+    libc.src.__support.macros.properties.types
+)
+
+add_entrypoint_object(
+  bf16subl
+  SRCS
+    bf16subl.cpp
+  HDRS
+    ../bf16subl.h
+  DEPENDS
+    libc.src.__support.common
+    libc.src.__support.FPUtil.bfloat16
+    libc.src.__support.FPUtil.generic.add_sub
+    libc.src.__support.macros.config
+    libc.src.__support.macros.properties.types
+)
+
+add_entrypoint_object(
+  bf16subf128
+  SRCS
+    bf16subf128.cpp
+  HDRS
+    ../bf16subf128.h
+  DEPENDS
+    libc.src.__support.common
+    libc.src.__support.FPUtil.bfloat16
+    libc.src.__support.FPUtil.generic.add_sub
+    libc.src.__support.macros.config
+    libc.src.__support.macros.properties.types
+)
diff --git a/libc/src/math/generic/bf16add.cpp b/libc/src/math/generic/bf16add.cpp
new file mode 100644
index 0000000000000..686848576f7a5
--- /dev/null
+++ b/libc/src/math/generic/bf16add.cpp
@@ -0,0 +1,22 @@
+//===-- Implementation of bf16add 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
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/bf16add.h"
+
+#include "src/__support/FPUtil/bfloat16.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(bfloat16, bf16add, (double x, double y)) {
+  return fputil::generic::add<bfloat16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/math/generic/bf16addf.cpp b/libc/src/math/generic/bf16addf.cpp
new file mode 100644
index 0000000000000..cd57a5ffad53d
--- /dev/null
+++ b/libc/src/math/generic/bf16addf.cpp
@@ -0,0 +1,22 @@
+//===-- Implementation of bf16addf 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
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/bf16addf.h"
+
+#include "src/__support/FPUtil/bfloat16.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(bfloat16, bf16addf, (float x, float y)) {
+  return fputil::generic::add<bfloat16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/math/generic/bf16addf128.cpp b/libc/src/math/generic/bf16addf128.cpp
new file mode 100644
index 0000000000000..ff94d3aff1581
--- /dev/null
+++ b/libc/src/math/generic/bf16addf128.cpp
@@ -0,0 +1,22 @@
+//===-- Implementation of bf16addf128 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
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/bf16addf128.h"
+
+#include "src/__support/FPUtil/bfloat16.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(bfloat16, bf16addf128, (float128 x, float128 y)) {
+  return fputil::generic::add<bfloat16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/math/generic/bf16addl.cpp b/libc/src/math/generic/bf16addl.cpp
new file mode 100644
index 0000000000000..7f053c1870d9b
--- /dev/null
+++ b/libc/src/math/generic/bf16addl.cpp
@@ -0,0 +1,22 @@
+//===-- Implementation of bf16addl 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
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/bf16addl.h"
+
+#include "src/__support/FPUtil/bfloat16.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(bfloat16, bf16addl, (long double x, long double y)) {
+  return fputil::generic::add<bfloat16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/math/generic/bf16sub.cpp b/libc/src/math/generic/bf16sub.cpp
new file mode 100644
index 0000000000000..5148f3657abaf
--- /dev/null
+++ b/libc/src/math/generic/bf16sub.cpp
@@ -0,0 +1,22 @@
+//===-- Implementation of bf16sub 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
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/bf16sub.h"
+
+#include "src/__support/FPUtil/bfloat16.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(bfloat16, bf16sub, (double x, double y)) {
+  return fputil::generic::sub<bfloat16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/math/generic/bf16subf.cpp b/libc/src/math/generic/bf16subf.cpp
new file mode 100644
index 0000000000000..dc303329b1dbd
--- /dev/null
+++ b/libc/src/math/generic/bf16subf.cpp
@@ -0,0 +1,22 @@
+//===-- Implementation of bf16subf 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
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/bf16subf.h"
+
+#include "src/__support/FPUtil/bfloat16.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(bfloat16, bf16subf, (float x, float y)) {
+  return fputil::generic::sub<bfloat16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/math/generic/bf16subf128.cpp b/libc/src/math/generic/bf16subf128.cpp
new file mode 100644
index 0000000000000..931c952386597
--- /dev/null
+++ b/libc/src/math/generic/bf16subf128.cpp
@@ -0,0 +1,22 @@
+//===-- Implementation of bf16subf128 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
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/bf16subf128.h"
+
+#include "src/__support/FPUtil/bfloat16.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(bfloat16, bf16subf128, (float128 x, float128 y)) {
+  return fputil::generic::sub<bfloat16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/src/math/generic/bf16subl.cpp b/libc/src/math/generic/bf16subl.cpp
new file mode 100644
index 0000000000000..315675640194f
--- /dev/null
+++ b/libc/src/math/generic/bf16subl.cpp
@@ -0,0 +1,22 @@
+//===-- Implementation of bf16subl 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
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/bf16subl.h"
+
+#include "src/__support/FPUtil/bfloat16.h"
+#include "src/__support/FPUtil/generic/add_sub.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(bfloat16, bf16subl, (long double x, long double y)) {
+  return fputil::generic::sub<bfloat16>(x, y);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/test/UnitTest/FPMatcher.h b/libc/test/UnitTest/FPMatcher.h
index 592cd1b3f37ab..e8d920cb11f54 100644
--- a/libc/test/UnitTest/FPMatcher.h
+++ b/libc/test/UnitTest/FPMatcher.h
@@ -271,7 +271,7 @@ struct ModifyMXCSR {
       expected)                                                                \
       .match(actual)
 
-#define EXPECT_FP_IS_NAN(actual) EXPECT_TRUE((actual) != (actual))
+#define EXPECT_FP_IS_NAN(actual) EXPECT_TRUE(FPBits(actual).is_nan())
 
 #define ASSERT_FP_EQ(expected, actual)                                         \
   ASSERT_THAT(actual, LIBC_NAMESPACE::testing::getMatcher<                     \
diff --git a/libc/test/src/math/CMakeLists.txt b/libc/test/src/math/CMakeLists.txt
index 43cde0d68873e..a74f9fe471963 100644
--- a/libc/test/src/math/CMakeLists.txt
+++ b/libc...
[truncated]

Signed-off-by: Krishna Pandey <[email protected]>

overmighty

Why did you do a327035? libc.src.__support.macros.properties.types is used by the implementation headers.

libc/src/math/generic/bf16addf.cpp

libc/test/UnitTest/FPMatcher.h

libc/utils/MPFRWrapper/MPFRUtils.cpp

Signed-off-by: Krishna Pandey <[email protected]>

This reverts commit a327035.

llvm-ci · 2025-08-08T19:38:42Z

LLVM Buildbot has detected a new failure on builder openmp-offload-amdgpu-runtime-2 running on rocm-worker-hw-02 while building libc at step 6 "test-openmp".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/10/builds/11142

Here is the relevant piece of the build log for the reference

Step 6 (test-openmp) failure: test (failure)
******************** TEST 'libarcher :: races/task-taskwait-nested.c' FAILED ********************
Exit Code: 1

Command Output (stdout):
--
# RUN: at line 13
/home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/./bin/clang -fopenmp  -gdwarf-4 -O1 -fsanitize=thread  -I /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests -I /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/runtime/src -L /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/runtime/src -Wl,-rpath,/home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/runtime/src   /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests/races/task-taskwait-nested.c -o /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/tools/archer/tests/races/Output/task-taskwait-nested.c.tmp -latomic && env TSAN_OPTIONS='ignore_noninstrumented_modules=0:ignore_noninstrumented_modules=1' /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests/deflake.bash /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/tools/archer/tests/races/Output/task-taskwait-nested.c.tmp 2>&1 | tee /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/tools/archer/tests/races/Output/task-taskwait-nested.c.tmp.log | /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/./bin/FileCheck /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests/races/task-taskwait-nested.c
# executed command: /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/./bin/clang -fopenmp -gdwarf-4 -O1 -fsanitize=thread -I /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests -I /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/runtime/src -L /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/runtime/src -Wl,-rpath,/home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/runtime/src /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests/races/task-taskwait-nested.c -o /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/tools/archer/tests/races/Output/task-taskwait-nested.c.tmp -latomic
# note: command had no output on stdout or stderr
# executed command: env TSAN_OPTIONS=ignore_noninstrumented_modules=0:ignore_noninstrumented_modules=1 /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests/deflake.bash /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/tools/archer/tests/races/Output/task-taskwait-nested.c.tmp
# note: command had no output on stdout or stderr
# executed command: tee /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/runtimes/runtimes-bins/openmp/tools/archer/tests/races/Output/task-taskwait-nested.c.tmp.log
# note: command had no output on stdout or stderr
# executed command: /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.build/./bin/FileCheck /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests/races/task-taskwait-nested.c
# .---command stderr------------
# | /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests/races/task-taskwait-nested.c:52:11: error: CHECK: expected string not found in input
# | // CHECK: WARNING: ThreadSanitizer: data race
# |           ^
# | <stdin>:1:1: note: scanning from here
# | DONE
# | ^
# | 
# | Input file: <stdin>
# | Check file: /home/botworker/builds/openmp-offload-amdgpu-runtime-2/llvm.src/openmp/tools/archer/tests/races/task-taskwait-nested.c
# | 
# | -dump-input=help explains the following input dump.
# | 
# | Input was:
# | <<<<<<
# |           1: DONE 
# | check:52     X~~~~ error: no match found
# | >>>>>>
# `-----------------------------
# error: command failed with exit status: 1

--

********************

krishna2803 added 7 commits August 8, 2025 22:36

feat: add bf16add{,f,l,f128} functions

7002552

Signed-off-by: Krishna Pandey <[email protected]>

feat: implement bf16sub{,f,l,f128} functions

797815a

Signed-off-by: Krishna Pandey <[email protected]>

refactor: remove redundant files

73a9b9c

Signed-off-by: Krishna Pandey <[email protected]>

nit: reorder deps and includes

f9bae64

Signed-off-by: Krishna Pandey <[email protected]>

chore: add general bf16{add,sub}{,f,l,f128} tests

33dfdf4

Signed-off-by: Krishna Pandey <[email protected]>

merge: branch 'main' into bf16add

69d924b

fix: tests

6d247ca

Signed-off-by: Krishna Pandey <[email protected]>

llvmbot added the libc label Aug 8, 2025

overmighty requested review from overmighty and lntue August 8, 2025 18:24

nit: remove unused dep

a327035

Signed-off-by: Krishna Pandey <[email protected]>

overmighty reviewed Aug 8, 2025

View reviewed changes

libc/src/math/generic/bf16addf.cpp Outdated Show resolved Hide resolved

libc/test/UnitTest/FPMatcher.h Outdated Show resolved Hide resolved

libc/utils/MPFRWrapper/MPFRUtils.cpp Outdated Show resolved Hide resolved

krishna2803 added 4 commits August 9, 2025 00:07

nit: remove blank line

f781f75

Signed-off-by: Krishna Pandey <[email protected]>

revert: EXPECT_FP_IS_NAN check

c91d831

Signed-off-by: Krishna Pandey <[email protected]>

chore: add matching comment for if condition

9ccf99f

Signed-off-by: Krishna Pandey <[email protected]>

revert: "nit: remove unused dep"

e612973

This reverts commit a327035.

overmighty approved these changes Aug 8, 2025

View reviewed changes

lntue approved these changes Aug 8, 2025

View reviewed changes

lntue merged commit 1ffb995 into llvm:main Aug 8, 2025
19 checks passed

krishna2803 deleted the bf16add branch August 8, 2025 19:36

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[libc][math][c++23] Add bf16{add,sub}{,f,l,f128} math functions #152774

[libc][math][c++23] Add bf16{add,sub}{,f,l,f128} math functions #152774

Uh oh!

krishna2803 commented Aug 8, 2025

Uh oh!

llvmbot commented Aug 8, 2025

Uh oh!

overmighty left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

llvm-ci commented Aug 8, 2025

Uh oh!

Uh oh!

[libc][math][c++23] Add bf16{add,sub}{,f,l,f128} math functions #152774

[libc][math][c++23] Add bf16{add,sub}{,f,l,f128} math functions #152774

Uh oh!

Conversation

krishna2803 commented Aug 8, 2025

Uh oh!

llvmbot commented Aug 8, 2025

Uh oh!

overmighty left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

llvm-ci commented Aug 8, 2025

Uh oh!

Uh oh!