[llvm] annotate remaining Support and ADT library interfaces for DLL export #145354

andrurogerz · 2025-06-23T16:05:34Z

Purpose

This patch is one in a series of code-mods that annotate LLVM’s public interface for export. This patch annotates the remaining LLVM Support and ADT interfaces that were missed in, or modified since, previous patches. The annotations currently have no meaningful impact on the LLVM build; however, they are a prerequisite to support an LLVM Windows DLL (shared library) build.

Background

This effort is tracked in #109483. Additional context is provided in this discourse, and documentation for LLVM_ABI and related annotations is found in the LLVM repo here.

Overview

The bulk of these changes were generated automatically using the Interface Definition Scanner (IDS) tool, followed formatting with git clang-format.

The following manual adjustments were also applied after running IDS:

Annotate the private method IEEEFloat::addOrSubtractSignificand with LLVM_ABI because it is used by a unit test friend class.
Annotate several friend function declarations with LLVM_ABI_FRIEND since the corresponding function declarations are annotated with LLVM_ABI (required by MSVC).

Validation

Local builds and tests to validate cross-platform compatibility. This included llvm, clang, and lldb on the following configurations:

Windows with MSVC
Windows with Clang
Linux with GCC
Linux with Clang
Darwin with Clang

llvmbot · 2025-06-23T20:58:45Z

@llvm/pr-subscribers-llvm-support

@llvm/pr-subscribers-llvm-adt

Author: Andrew Rogers (andrurogerz)

Changes

Purpose

This patch is one in a series of code-mods that annotate LLVM’s public interface for export. This patch annotates the remaining LLVM Support and ADT interfaces that were missed in, or modified since, previous patches. The annotations currently have no meaningful impact on the LLVM build; however, they are a prerequisite to support an LLVM Windows DLL (shared library) build.

Background

This effort is tracked in #109483. Additional context is provided in this discourse, and documentation for LLVM_ABI and related annotations is found in the LLVM repo here.

Overview

The bulk of these changes were generated automatically using the Interface Definition Scanner (IDS) tool, followed formatting with git clang-format.

The following manual adjustments were also applied after running IDS:

Annotate the private method IEEEFloat::addOrSubtractSignificand with LLVM_ABI because it is used by a unit test friend class.
Annotate several friend function declarations with LLVM_ABI_FRIEND since the corresponding function declarations are annotated with LLVM_ABI (required by MSVC).

Validation

Local builds and tests to validate cross-platform compatibility. This included llvm, clang, and lldb on the following configurations:

Windows with MSVC
Windows with Clang
Linux with GCC
Linux with Clang
Darwin with Clang

Full diff: https://github.com/llvm/llvm-project/pull/145354.diff

8 Files Affected:

(modified) llvm/include/llvm/ADT/APFloat.h (+3-1)
(modified) llvm/include/llvm/ADT/APInt.h (+1-1)
(modified) llvm/include/llvm/ADT/DynamicAPInt.h (+1-1)
(modified) llvm/include/llvm/Support/ConvertUTF.h (+3-3)
(modified) llvm/include/llvm/Support/KnownFPClass.h (+7-5)
(modified) llvm/include/llvm/Support/ProgramStack.h (+4-3)
(modified) llvm/include/llvm/Support/TextEncoding.h (+5-3)
(modified) llvm/include/llvm/Support/Timer.h (+1-1)

diff --git a/llvm/include/llvm/ADT/APFloat.h b/llvm/include/llvm/ADT/APFloat.h
index 13df838da3dad..a3ff0184d666d 100644
--- a/llvm/include/llvm/ADT/APFloat.h
+++ b/llvm/include/llvm/ADT/APFloat.h
@@ -673,7 +673,9 @@ class IEEEFloat final {
 
   integerPart addSignificand(const IEEEFloat &);
   integerPart subtractSignificand(const IEEEFloat &, integerPart);
-  lostFraction addOrSubtractSignificand(const IEEEFloat &, bool subtract);
+  // Exported for IEEEFloatUnitTestHelper.
+  LLVM_ABI lostFraction addOrSubtractSignificand(const IEEEFloat &,
+                                                 bool subtract);
   lostFraction multiplySignificand(const IEEEFloat &, IEEEFloat,
                                    bool ignoreAddend = false);
   lostFraction multiplySignificand(const IEEEFloat&);
diff --git a/llvm/include/llvm/ADT/APInt.h b/llvm/include/llvm/ADT/APInt.h
index fe48c60466e96..164e9e52970d8 100644
--- a/llvm/include/llvm/ADT/APInt.h
+++ b/llvm/include/llvm/ADT/APInt.h
@@ -561,7 +561,7 @@ class [[nodiscard]] APInt {
   }
 
   /// Overload to compute a hash_code for an APInt value.
-  friend hash_code hash_value(const APInt &Arg);
+  LLVM_ABI_FRIEND friend hash_code hash_value(const APInt &Arg);
 
   /// This function returns a pointer to the internal storage of the APInt.
   /// This is useful for writing out the APInt in binary form without any
diff --git a/llvm/include/llvm/ADT/DynamicAPInt.h b/llvm/include/llvm/ADT/DynamicAPInt.h
index 48c5c6eac9013..373c0e6247969 100644
--- a/llvm/include/llvm/ADT/DynamicAPInt.h
+++ b/llvm/include/llvm/ADT/DynamicAPInt.h
@@ -212,7 +212,7 @@ class DynamicAPInt {
   friend DynamicAPInt operator/(int64_t A, const DynamicAPInt &B);
   friend DynamicAPInt operator%(int64_t A, const DynamicAPInt &B);
 
-  friend hash_code hash_value(const DynamicAPInt &x); // NOLINT
+  LLVM_ABI_FRIEND friend hash_code hash_value(const DynamicAPInt &x); // NOLINT
 
   LLVM_ABI void static_assert_layout(); // NOLINT
 
diff --git a/llvm/include/llvm/Support/ConvertUTF.h b/llvm/include/llvm/Support/ConvertUTF.h
index 3bb238e7df2ed..bb1723518a490 100644
--- a/llvm/include/llvm/Support/ConvertUTF.h
+++ b/llvm/include/llvm/Support/ConvertUTF.h
@@ -346,9 +346,9 @@ LLVM_ABI bool convertUTF32ToUTF8String(ArrayRef<UTF32> Src, std::string &Out);
 LLVM_ABI bool convertUTF8ToUTF16String(StringRef SrcUTF8,
                                        SmallVectorImpl<UTF16> &DstUTF16);
 
-bool IsSingleCodeUnitUTF8Codepoint(unsigned);
-bool IsSingleCodeUnitUTF16Codepoint(unsigned);
-bool IsSingleCodeUnitUTF32Codepoint(unsigned);
+LLVM_ABI bool IsSingleCodeUnitUTF8Codepoint(unsigned);
+LLVM_ABI bool IsSingleCodeUnitUTF16Codepoint(unsigned);
+LLVM_ABI bool IsSingleCodeUnitUTF32Codepoint(unsigned);
 
 #if defined(_WIN32)
 namespace sys {
diff --git a/llvm/include/llvm/Support/KnownFPClass.h b/llvm/include/llvm/Support/KnownFPClass.h
index 9ebdf260e0ec7..b3c18bcf6b34b 100644
--- a/llvm/include/llvm/Support/KnownFPClass.h
+++ b/llvm/include/llvm/Support/KnownFPClass.h
@@ -15,6 +15,7 @@
 #define LLVM_SUPPORT_KNOWNFPCLASS_H
 
 #include "llvm/ADT/FloatingPointMode.h"
+#include "llvm/Support/Compiler.h"
 #include <optional>
 
 namespace llvm {
@@ -78,13 +79,13 @@ struct KnownFPClass {
   /// Return true if it's know this can never be interpreted as a zero. This
   /// extends isKnownNeverZero to cover the case where the assumed
   /// floating-point mode for the function interprets denormals as zero.
-  bool isKnownNeverLogicalZero(DenormalMode Mode) const;
+  LLVM_ABI bool isKnownNeverLogicalZero(DenormalMode Mode) const;
 
   /// Return true if it's know this can never be interpreted as a negative zero.
-  bool isKnownNeverLogicalNegZero(DenormalMode Mode) const;
+  LLVM_ABI bool isKnownNeverLogicalNegZero(DenormalMode Mode) const;
 
   /// Return true if it's know this can never be interpreted as a positive zero.
-  bool isKnownNeverLogicalPosZero(DenormalMode Mode) const;
+  LLVM_ABI bool isKnownNeverLogicalPosZero(DenormalMode Mode) const;
 
   static constexpr FPClassTest OrderedLessThanZeroMask =
       fcNegSubnormal | fcNegNormal | fcNegInf;
@@ -209,7 +210,7 @@ struct KnownFPClass {
   ///
   /// This assumes a copy-like operation and will replace any currently known
   /// information.
-  void propagateDenormal(const KnownFPClass &Src, DenormalMode Mode);
+  LLVM_ABI void propagateDenormal(const KnownFPClass &Src, DenormalMode Mode);
 
   /// Report known classes if \p Src is evaluated through a potentially
   /// canonicalizing operation. We can assume signaling nans will not be
@@ -217,7 +218,8 @@ struct KnownFPClass {
   ///
   /// This assumes a copy-like operation and will replace any currently known
   /// information.
-  void propagateCanonicalizingSrc(const KnownFPClass &Src, DenormalMode Mode);
+  LLVM_ABI void propagateCanonicalizingSrc(const KnownFPClass &Src,
+                                           DenormalMode Mode);
 
   void resetAll() { *this = KnownFPClass(); }
 };
diff --git a/llvm/include/llvm/Support/ProgramStack.h b/llvm/include/llvm/Support/ProgramStack.h
index 55964c9779209..0dd8235b90c06 100644
--- a/llvm/include/llvm/Support/ProgramStack.h
+++ b/llvm/include/llvm/Support/ProgramStack.h
@@ -10,6 +10,7 @@
 #define LLVM_SUPPORT_PROGRAMSTACK_H
 
 #include "llvm/ADT/STLFunctionalExtras.h"
+#include "llvm/Support/Compiler.h"
 
 // LLVM_HAS_SPLIT_STACKS is exposed in the header because CrashRecoveryContext
 // needs to know if it's running on another thread or not.
@@ -28,12 +29,12 @@ namespace llvm {
 ///
 /// The value is not guaranteed to point to anything specific. It can be used to
 /// estimate how much stack space has been used since the previous call.
-uintptr_t getStackPointer();
+LLVM_ABI uintptr_t getStackPointer();
 
 /// \returns the default stack size for this platform.
 ///
 /// Based on \p RLIMIT_STACK or the equivalent.
-unsigned getDefaultStackSize();
+LLVM_ABI unsigned getDefaultStackSize();
 
 /// Runs Fn on a new stack of at least the given size.
 ///
@@ -42,7 +43,7 @@ unsigned getDefaultStackSize();
 ///
 /// The preferred implementation is split stacks on platforms that have a good
 /// debugging experience for them. On other platforms a new thread is used.
-void runOnNewStack(unsigned StackSize, function_ref<void()> Fn);
+LLVM_ABI void runOnNewStack(unsigned StackSize, function_ref<void()> Fn);
 
 template <typename R, typename... Ts>
 std::enable_if_t<!std::is_same_v<R, void>, R>
diff --git a/llvm/include/llvm/Support/TextEncoding.h b/llvm/include/llvm/Support/TextEncoding.h
index e204b95dd2dd7..8a304910aa5dd 100644
--- a/llvm/include/llvm/Support/TextEncoding.h
+++ b/llvm/include/llvm/Support/TextEncoding.h
@@ -18,6 +18,7 @@
 #include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Config/config.h"
+#include "llvm/Support/Compiler.h"
 #include "llvm/Support/ErrorOr.h"
 
 #include <string>
@@ -92,8 +93,8 @@ class TextEncodingConverter {
   /// \param[in] From the source character encoding
   /// \param[in] To the target character encoding
   /// \return a TextEncodingConverter instance or an error code
-  static ErrorOr<TextEncodingConverter> create(TextEncoding From,
-                                               TextEncoding To);
+  LLVM_ABI static ErrorOr<TextEncodingConverter> create(TextEncoding From,
+                                                        TextEncoding To);
 
   /// Creates a TextEncodingConverter instance.
   /// Returns std::errc::invalid_argument in case the requested conversion is
@@ -101,7 +102,8 @@ class TextEncodingConverter {
   /// \param[in] From name of the source character encoding
   /// \param[in] To name of the target character encoding
   /// \return a TextEncodingConverter instance or an error code
-  static ErrorOr<TextEncodingConverter> create(StringRef From, StringRef To);
+  LLVM_ABI static ErrorOr<TextEncodingConverter> create(StringRef From,
+                                                        StringRef To);
 
   TextEncodingConverter(const TextEncodingConverter &) = delete;
   TextEncodingConverter &operator=(const TextEncodingConverter &) = delete;
diff --git a/llvm/include/llvm/Support/Timer.h b/llvm/include/llvm/Support/Timer.h
index 36890c75a65e3..c303ff06f3de9 100644
--- a/llvm/include/llvm/Support/Timer.h
+++ b/llvm/include/llvm/Support/Timer.h
@@ -258,7 +258,7 @@ class TimerGroup {
 
 private:
   friend class Timer;
-  friend void PrintStatisticsJSON(raw_ostream &OS);
+  LLVM_ABI_FRIEND friend void PrintStatisticsJSON(raw_ostream &OS);
   void addTimer(Timer &T);
   void removeTimer(Timer &T);
   void prepareToPrintList(bool reset_time = false);

… interfaces

andrurogerz · 2025-06-24T18:57:31Z

@compnerd, @vgvassilev here's a quick one if you don't mind, thanks!

… export (llvm#145354) ## Purpose This patch is one in a series of code-mods that annotate LLVM’s public interface for export. This patch annotates the remaining LLVM Support and ADT interfaces that were missed in, or modified since, previous patches. The annotations currently have no meaningful impact on the LLVM build; however, they are a prerequisite to support an LLVM Windows DLL (shared library) build. ## Background This effort is tracked in llvm#109483. Additional context is provided in [this discourse](https://discourse.llvm.org/t/psa-annotating-llvm-public-interface/85307), and documentation for `LLVM_ABI` and related annotations is found in the LLVM repo [here](https://github.com/llvm/llvm-project/blob/main/llvm/docs/InterfaceExportAnnotations.rst). ## Overview The bulk of these changes were generated automatically using the [Interface Definition Scanner (IDS)](https://github.com/compnerd/ids) tool, followed formatting with `git clang-format`. The following manual adjustments were also applied after running IDS: - Annotate the private method `IEEEFloat::addOrSubtractSignificand` with `LLVM_ABI` because it is used by a unit test friend class. - Annotate several `friend` function declarations with `LLVM_ABI_FRIEND` since the corresponding function declarations are annotated with `LLVM_ABI` (required by MSVC). ## Validation Local builds and tests to validate cross-platform compatibility. This included llvm, clang, and lldb on the following configurations: - Windows with MSVC - Windows with Clang - Linux with GCC - Linux with Clang - Darwin with Clang

andrurogerz marked this pull request as ready for review June 23, 2025 20:58

llvmbot added llvm:support llvm:adt labels Jun 23, 2025

andrurogerz added 3 commits June 24, 2025 09:41

[llvm] annotate remaining Support library interfaces for DLL export

0982dc3

[llvm] manual fix-ups to IDS codemod of remaining Support and ADT lib…

1fb42db

… interfaces

[llvm] clang-format changes

0637881

andrurogerz force-pushed the llvmdll-misc-1 branch from c4d9943 to 0637881 Compare June 24, 2025 17:01

andrurogerz mentioned this pull request Jun 24, 2025

[llvm] annotate remaining LLVM interfaces for DLL export #144746

Closed

compnerd approved these changes Jun 24, 2025

View reviewed changes

compnerd merged commit 3574560 into llvm:main Jun 24, 2025
7 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[llvm] annotate remaining Support and ADT library interfaces for DLL export #145354

[llvm] annotate remaining Support and ADT library interfaces for DLL export #145354

Uh oh!

andrurogerz commented Jun 23, 2025 •

edited

Loading

Uh oh!

llvmbot commented Jun 23, 2025 •

edited

Loading

Purpose

Background

Overview

Validation

Uh oh!

andrurogerz commented Jun 24, 2025

Uh oh!

Uh oh!

Uh oh!

[llvm] annotate remaining Support and ADT library interfaces for DLL export #145354

[llvm] annotate remaining Support and ADT library interfaces for DLL export #145354

Uh oh!

Conversation

andrurogerz commented Jun 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Purpose

Background

Overview

Validation

Uh oh!

llvmbot commented Jun 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Purpose

Background

Overview

Validation

Uh oh!

andrurogerz commented Jun 24, 2025

Uh oh!

Uh oh!

Uh oh!

andrurogerz commented Jun 23, 2025 •

edited

Loading

llvmbot commented Jun 23, 2025 •

edited

Loading