[BOLT] Enable standalone build (#97130)

Continue from #87196 as author did not have much time, I have taken over
working on this PR. We would like to have this so it'll be easier to
package for Nix.

Can be tested by copying cmake, bolt, third-party, and llvm directories
out into their own directory with this PR applied and then build bolt.

---------

Co-authored-by: pca006132 <john.lck40@gmail.com>

bolt/CMakeLists.txt

@@ -1,6 +1,17 @@
+cmake_minimum_required(VERSION 3.20.0)
+
 set(LLVM_SUBPROJECT_TITLE "BOLT")
 
-include(ExternalProject)
+if(NOT DEFINED LLVM_COMMON_CMAKE_UTILS)
+  set(LLVM_COMMON_CMAKE_UTILS ${CMAKE_CURRENT_SOURCE_DIR}/../cmake)
+endif()
+include(${LLVM_COMMON_CMAKE_UTILS}/Modules/CMakePolicy.cmake
+  NO_POLICY_SCOPE)
+
+if(CMAKE_SOURCE_DIR STREQUAL CMAKE_CURRENT_SOURCE_DIR)
+  project(bolt)
+  set(BOLT_BUILT_STANDALONE TRUE)
+endif()
 
 set(BOLT_SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR})
 set(BOLT_BINARY_DIR ${CMAKE_CURRENT_BINARY_DIR})
@@ -9,6 +20,42 @@ set(CMAKE_CXX_STANDARD 17)
 # Add path for custom modules.
 list(INSERT CMAKE_MODULE_PATH 0 "${BOLT_SOURCE_DIR}/cmake/modules")
 
+include(GNUInstallDirs)
+
+# standalone build, copied from clang
+if(BOLT_BUILT_STANDALONE)
+  set(CMAKE_CXX_STANDARD 17 CACHE STRING "C++ standard to conform to")
+  set(CMAKE_CXX_STANDARD_REQUIRED YES)
+  set(CMAKE_CXX_EXTENSIONS NO)
+
+  if(NOT MSVC_IDE)
+    set(LLVM_ENABLE_ASSERTIONS ${ENABLE_ASSERTIONS}
+      CACHE BOOL "Enable assertions")
+    # Assertions should follow llvm-config's.
+    mark_as_advanced(LLVM_ENABLE_ASSERTIONS)
+  endif()
+
+  find_package(LLVM REQUIRED HINTS "${LLVM_CMAKE_DIR}")
+  list(APPEND CMAKE_MODULE_PATH "${LLVM_DIR}")
+
+  set(LLVM_MAIN_SRC_DIR "${CMAKE_CURRENT_SOURCE_DIR}/../llvm" CACHE PATH "Path to LLVM source tree")
+  find_program(LLVM_TABLEGEN_EXE "llvm-tblgen" ${LLVM_TOOLS_BINARY_DIR}
+    NO_DEFAULT_PATH)
+
+  # They are used as destination of target generators.
+  set(LLVM_RUNTIME_OUTPUT_INTDIR ${CMAKE_BINARY_DIR}/${CMAKE_CFG_INTDIR}/bin)
+  set(LLVM_LIBRARY_OUTPUT_INTDIR ${CMAKE_BINARY_DIR}/${CMAKE_CFG_INTDIR}/lib${LLVM_LIBDIR_SUFFIX})
+
+  include(AddLLVM)
+  include(TableGen)
+  include_directories(${LLVM_INCLUDE_DIRS})
+  link_directories("${LLVM_LIBRARY_DIR}")
+
+  set( CMAKE_RUNTIME_OUTPUT_DIRECTORY "${CMAKE_INSTALL_BINDIR}" )
+  set( CMAKE_LIBRARY_OUTPUT_DIRECTORY "${CMAKE_INSTALL_LIBDIR}/${LLVM_LIBDIR_SUFFIX}" )
+  set( CMAKE_ARCHIVE_OUTPUT_DIRECTORY "${CMAKE_INSTALL_LIBDIR}/${LLVM_LIBDIR_SUFFIX}")
+endif() # standalone
+
 # Determine default set of targets to build -- the intersection of
 # those BOLT supports and those LLVM is targeting.
 set(BOLT_TARGETS_TO_BUILD_all "AArch64;X86;RISCV")
@@ -94,6 +141,8 @@ if (BOLT_ENABLE_RUNTIME)
   if(CMAKE_SYSROOT)
     list(APPEND extra_args -DCMAKE_SYSROOT=${CMAKE_SYSROOT})
   endif()
+
+  include(ExternalProject)
   ExternalProject_Add(bolt_rt
     SOURCE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/runtime"
     STAMP_DIR ${CMAKE_CURRENT_BINARY_DIR}/bolt_rt-stamps
@@ -104,6 +153,7 @@ if (BOLT_ENABLE_RUNTIME)
                -DCMAKE_MAKE_PROGRAM=${CMAKE_MAKE_PROGRAM}
                -DLLVM_LIBDIR_SUFFIX=${LLVM_LIBDIR_SUFFIX}
                -DLLVM_LIBRARY_DIR=${LLVM_LIBRARY_DIR}
+               -DBOLT_BUILT_STANDALONE=${BOLT_BUILT_STANDALONE}
                ${extra_args}
     INSTALL_COMMAND ""
     BUILD_ALWAYS True
@@ -113,6 +163,8 @@ if (BOLT_ENABLE_RUNTIME)
   add_llvm_install_targets(install-bolt_rt
     DEPENDS bolt_rt bolt
     COMPONENT bolt)
+  set(LIBBOLT_RT_INSTR "${CMAKE_CURRENT_BINARY_DIR}/bolt_rt-bins/lib/libbolt_rt_instr.a")
+  set(LIBBOLT_RT_HUGIFY "${CMAKE_CURRENT_BINARY_DIR}/bolt_rt-bins/lib/libbolt_rt_hugify.a")
 endif()
 
 find_program(GNU_LD_EXECUTABLE NAMES ${LLVM_DEFAULT_TARGET_TRIPLE}-ld.bfd ld.bfd DOC "GNU ld")

bolt/include/bolt/RuntimeLibs/RuntimeLibrary.h

@@ -58,7 +58,16 @@ protected:
   uint64_t RuntimeFiniAddress{0};
   uint64_t RuntimeStartAddress{0};
 
-  /// Get the full path to a runtime library specified by \p LibFileName.
+  /// Get the full path to a runtime library specified by \p LibFileName and \p
+  /// ToolPath.
+  static std::string getLibPathByToolPath(StringRef ToolPath,
+                                          StringRef LibFileName);
+
+  /// Get the full path to a runtime library by the install directory.
+  static std::string getLibPathByInstalled(StringRef LibFileName);
+
+  /// Gets the full path to a runtime library based on whether it exists
+  /// in the install libdir or runtime libdir.
   static std::string getLibPath(StringRef ToolPath, StringRef LibFileName);
 
   /// Load a static runtime library specified by \p LibPath.

bolt/lib/CMakeLists.txt

@@ -1,3 +1,5 @@
+add_compile_definitions(CMAKE_INSTALL_FULL_LIBDIR="${CMAKE_INSTALL_FULL_LIBDIR}")
+
 add_subdirectory(Core)
 add_subdirectory(Passes)
 add_subdirectory(Profile)

bolt/lib/RuntimeLibs/HugifyRuntimeLibrary.cpp

@@ -32,9 +32,9 @@ cl::opt<bool>
                     "(which is what --hot-text relies on)."),
            cl::cat(BoltOptCategory));
 
-static cl::opt<std::string> RuntimeHugifyLib(
-    "runtime-hugify-lib",
-    cl::desc("specify file name of the runtime hugify library"),
+static cl::opt<std::string>
+    RuntimeHugifyLib("runtime-hugify-lib",
+                     cl::desc("specify path of the runtime hugify library"),
                      cl::init("libbolt_rt_hugify.a"), cl::cat(BoltOptCategory));
 
 } // namespace opts

bolt/lib/RuntimeLibs/InstrumentationRuntimeLibrary.cpp

@@ -26,7 +26,7 @@ namespace opts {
 
 cl::opt<std::string> RuntimeInstrumentationLib(
     "runtime-instrumentation-lib",
-    cl::desc("specify file name of the runtime instrumentation library"),
+    cl::desc("specify path of the runtime instrumentation library"),
     cl::init("libbolt_rt_instr.a"), cl::cat(BoltOptCategory));
 
 extern cl::opt<bool> InstrumentationFileAppendPID;

bolt/lib/RuntimeLibs/RuntimeLibrary.cpp

@@ -26,7 +26,7 @@ using namespace bolt;
 
 void RuntimeLibrary::anchor() {}
 
-std::string RuntimeLibrary::getLibPath(StringRef ToolPath,
+std::string RuntimeLibrary::getLibPathByToolPath(StringRef ToolPath,
                                                  StringRef LibFileName) {
   StringRef Dir = llvm::sys::path::parent_path(ToolPath);
   SmallString<128> LibPath = llvm::sys::path::parent_path(Dir);
@@ -38,13 +38,36 @@ std::string RuntimeLibrary::getLibPath(StringRef ToolPath,
     llvm::sys::path::append(LibPath, "lib" LLVM_LIBDIR_SUFFIX);
   }
   llvm::sys::path::append(LibPath, LibFileName);
-  if (!llvm::sys::fs::exists(LibPath)) {
-    errs() << "BOLT-ERROR: library not found: " << LibPath << "\n";
-    exit(1);
-  }
   return std::string(LibPath);
 }
 
+std::string RuntimeLibrary::getLibPathByInstalled(StringRef LibFileName) {
+  SmallString<128> LibPath(CMAKE_INSTALL_FULL_LIBDIR);
+  llvm::sys::path::append(LibPath, LibFileName);
+  return std::string(LibPath);
+}
+
+std::string RuntimeLibrary::getLibPath(StringRef ToolPath,
+                                       StringRef LibFileName) {
+  if (llvm::sys::fs::exists(LibFileName)) {
+    return std::string(LibFileName);
+  }
+
+  std::string ByTool = getLibPathByToolPath(ToolPath, LibFileName);
+  if (llvm::sys::fs::exists(ByTool)) {
+    return ByTool;
+  }
+
+  std::string ByInstalled = getLibPathByInstalled(LibFileName);
+  if (llvm::sys::fs::exists(ByInstalled)) {
+    return ByInstalled;
+  }
+
+  errs() << "BOLT-ERROR: library not found: " << ByTool << ", " << ByInstalled
+         << ", or " << LibFileName << "\n";
+  exit(1);
+}
+
 void RuntimeLibrary::loadLibrary(StringRef LibPath, BOLTLinker &Linker,
                                  BOLTLinker::SectionsMapper MapSections) {
   ErrorOr<std::unique_ptr<MemoryBuffer>> MaybeBuf =

bolt/lib/Target/AArch64/CMakeLists.txt

@@ -4,6 +4,18 @@ set(LLVM_LINK_COMPONENTS
   AArch64Desc
   )
 
+if(BOLT_BUILT_STANDALONE)
+  set(LLVM_TARGET_DEFINITIONS ${LLVM_MAIN_SRC_DIR}/lib/Target/AArch64/AArch64.td)
+  list(APPEND LLVM_TABLEGEN_FLAGS -I ${LLVM_MAIN_SRC_DIR}/lib/Target/AArch64)
+  tablegen(LLVM AArch64GenInstrInfo.inc -gen-instr-info)
+  tablegen(LLVM AArch64GenRegisterInfo.inc -gen-register-info)
+  tablegen(LLVM AArch64GenSystemOperands.inc -gen-searchable-tables)
+  tablegen(LLVM AArch64GenSubtargetInfo.inc -gen-subtarget)
+
+  add_public_tablegen_target(AArch64CommonTableGen)
+  include_directories(${CMAKE_CURRENT_BINARY_DIR})
+endif()
+
 add_llvm_library(LLVMBOLTTargetAArch64
   AArch64MCPlusBuilder.cpp
 

bolt/lib/Target/RISCV/CMakeLists.txt

@@ -4,6 +4,19 @@ set(LLVM_LINK_COMPONENTS
   RISCVDesc
   )
 
+if(BOLT_BUILT_STANDALONE)
+  # tablegen, copied from llvm/lib/Target/RISCV/CMakeLists.txt
+  set(LLVM_TARGET_DEFINITIONS ${LLVM_MAIN_SRC_DIR}/lib/Target/RISCV/RISCV.td)
+  list(APPEND LLVM_TABLEGEN_FLAGS -I ${LLVM_MAIN_SRC_DIR}/lib/Target/RISCV)
+  tablegen(LLVM RISCVGenInstrInfo.inc -gen-instr-info)
+  tablegen(LLVM RISCVGenRegisterInfo.inc -gen-register-info)
+  tablegen(LLVM RISCVGenSearchableTables.inc -gen-searchable-tables)
+  tablegen(LLVM RISCVGenSubtargetInfo.inc -gen-subtarget)
+
+  add_public_tablegen_target(RISCVCommonTableGen)
+  include_directories(${CMAKE_CURRENT_BINARY_DIR})
+endif()
+
 add_llvm_library(LLVMBOLTTargetRISCV
   RISCVMCPlusBuilder.cpp
 

bolt/lib/Target/X86/CMakeLists.txt

@@ -5,6 +5,18 @@ set(LLVM_LINK_COMPONENTS
   X86Desc
   )
 
+if(BOLT_BUILT_STANDALONE)
+  set(LLVM_TARGET_DEFINITIONS ${LLVM_MAIN_SRC_DIR}/lib/Target/X86/X86.td)
+  list(APPEND LLVM_TABLEGEN_FLAGS -I ${LLVM_MAIN_SRC_DIR}/lib/Target/X86)
+  tablegen(LLVM X86GenInstrInfo.inc -gen-instr-info -instr-info-expand-mi-operand-info=0)
+  tablegen(LLVM X86GenMnemonicTables.inc -gen-x86-mnemonic-tables -asmwriternum=1)
+  tablegen(LLVM X86GenRegisterInfo.inc -gen-register-info)
+  tablegen(LLVM X86GenSubtargetInfo.inc -gen-subtarget)
+
+  add_public_tablegen_target(X86CommonTableGen)
+  include_directories(${CMAKE_CURRENT_BINARY_DIR})
+endif()
+
 add_llvm_library(LLVMBOLTTargetX86
   X86MCPlusBuilder.cpp
   X86MCSymbolizer.cpp

bolt/lib/Utils/CMakeLists.txt

@@ -1,15 +1,39 @@
+find_first_existing_vc_file("${LLVM_MAIN_SRC_DIR}" llvm_vc)
+find_first_existing_vc_file("${BOLT_SOURCE_DIR}" bolt_vc)
+
+# The VC revision include that we want to generate.
+set(version_inc "${CMAKE_CURRENT_BINARY_DIR}/VCSVersion.inc")
+
+set(generate_vcs_version_script "${LLVM_CMAKE_DIR}/GenerateVersionFromVCS.cmake")
+
+# Create custom target to generate the VC revision include.
+add_custom_command(OUTPUT "${version_inc}"
+  DEPENDS "${llvm_vc}" "${bolt_vc}" "${generate_vcs_version_script}"
+  COMMAND ${CMAKE_COMMAND} "-DNAMES=BOLT"
+                           "-DHEADER_FILE=${version_inc}"
+                           "-DBOLT_SOURCE_DIR=${BOLT_SOURCE_DIR}"
+                           "-DLLVM_VC_REPOSITORY=${llvm_vc_repository}"
+                           "-DLLVM_VC_REVISION=${llvm_vc_revision}"
+                           "-DLLVM_FORCE_VC_REVISION=${LLVM_FORCE_VC_REVISION}"
+                           "-DLLVM_FORCE_VC_REPOSITORY=${LLVM_FORCE_VC_REPOSITORY}"
+                           -P "${generate_vcs_version_script}")
+
+# Mark the generated header as being generated.
+set_source_files_properties("${version_inc}"
+  PROPERTIES GENERATED TRUE
+             HEADER_FILE_ONLY TRUE)
+
+include_directories(${CMAKE_CURRENT_BINARY_DIR})
+
 add_llvm_library(LLVMBOLTUtils
   CommandLineOpts.cpp
   Utils.cpp
-
+  ${version_inc}
   DISABLE_LLVM_LINK_LLVM_DYLIB
 
   LINK_LIBS
   ${LLVM_PTHREAD_LIB}
 
-  DEPENDS
-  llvm_vcsrevision_h
-
   LINK_COMPONENTS
   Support
   )

bolt/lib/Utils/CommandLineOpts.cpp

@@ -11,15 +11,15 @@
 //===----------------------------------------------------------------------===//
 
 #include "bolt/Utils/CommandLineOpts.h"
-#include "llvm/Support/VCSRevision.h"
+#include "VCSVersion.inc"
 
 using namespace llvm;
 
 namespace llvm {
 namespace bolt {
 const char *BoltRevision =
-#ifdef LLVM_REVISION
-    LLVM_REVISION;
+#ifdef BOLT_REVISION
+    BOLT_REVISION;
 #else
     "<unknown>";
 #endif

bolt/runtime/CMakeLists.txt

@@ -16,12 +16,19 @@ add_library(bolt_rt_instr STATIC
   instr.cpp
   ${CMAKE_CURRENT_BINARY_DIR}/config.h
   )
-set_target_properties(bolt_rt_instr PROPERTIES ARCHIVE_OUTPUT_DIRECTORY "${LLVM_LIBRARY_DIR}")
+set_target_properties(bolt_rt_instr PROPERTIES ARCHIVE_OUTPUT_DIRECTORY "${CMAKE_INSTALL_LIBDIR}${LLVM_LIBDIR_SUFFIX}")
 add_library(bolt_rt_hugify STATIC
   hugify.cpp
   ${CMAKE_CURRENT_BINARY_DIR}/config.h
   )
-set_target_properties(bolt_rt_hugify PROPERTIES ARCHIVE_OUTPUT_DIRECTORY "${LLVM_LIBRARY_DIR}")
+set_target_properties(bolt_rt_hugify PROPERTIES ARCHIVE_OUTPUT_DIRECTORY "${CMAKE_INSTALL_LIBDIR}${LLVM_LIBDIR_SUFFIX}")
+
+if(NOT BOLT_BUILT_STANDALONE)
+  add_custom_command(TARGET bolt_rt_instr POST_BUILD
+    COMMAND ${CMAKE_COMMAND} -E copy "${CMAKE_CURRENT_BINARY_DIR}/lib/libbolt_rt_instr.a" "${LLVM_LIBRARY_DIR}")
+  add_custom_command(TARGET bolt_rt_hugify POST_BUILD
+    COMMAND ${CMAKE_COMMAND} -E copy "${CMAKE_CURRENT_BINARY_DIR}/lib/libbolt_rt_hugify.a" "${LLVM_LIBRARY_DIR}")
+endif()
 
 set(BOLT_RT_FLAGS
   -ffreestanding
@@ -46,8 +53,8 @@ target_include_directories(bolt_rt_instr PRIVATE ${CMAKE_CURRENT_BINARY_DIR})
 target_compile_options(bolt_rt_hugify PRIVATE ${BOLT_RT_FLAGS})
 target_include_directories(bolt_rt_hugify PRIVATE ${CMAKE_CURRENT_BINARY_DIR})
 
-install(TARGETS bolt_rt_instr DESTINATION "lib${LLVM_LIBDIR_SUFFIX}")
-install(TARGETS bolt_rt_hugify DESTINATION "lib${LLVM_LIBDIR_SUFFIX}")
+install(TARGETS bolt_rt_instr DESTINATION "${CMAKE_INSTALL_LIBDIR}${LLVM_LIBDIR_SUFFIX}")
+install(TARGETS bolt_rt_hugify DESTINATION "${CMAKE_INSTALL_LIBDIR}${LLVM_LIBDIR_SUFFIX}")
 
 if (CMAKE_CXX_COMPILER_ID MATCHES ".*Clang.*" AND CMAKE_SYSTEM_NAME STREQUAL "Darwin")
   add_library(bolt_rt_instr_osx STATIC
@@ -59,5 +66,10 @@ if (CMAKE_CXX_COMPILER_ID MATCHES ".*Clang.*" AND CMAKE_SYSTEM_NAME STREQUAL "Da
   target_compile_options(bolt_rt_instr_osx PRIVATE
     -target x86_64-apple-darwin19.6.0
     ${BOLT_RT_FLAGS})
-  install(TARGETS bolt_rt_instr_osx DESTINATION "lib${LLVM_LIBDIR_SUFFIX}")
+  install(TARGETS bolt_rt_instr_osx DESTINATION "${CMAKE_INSTALL_LIBDIR}${LLVM_LIBDIR_SUFFIX}")
+
+  if(NOT BOLT_BUILT_STANDALONE)
+    add_custom_command(TARGET bolt_rt_instr_osx POST_BUILD
+      COMMAND ${CMAKE_COMMAND} -E copy "${CMAKE_CURRENT_BINARY_DIR}/lib/libbolt_rt_instr_osx.a" "${LLVM_LIBRARY_DIR}")
+  endif()
 endif()

bolt/test/lit.cfg.py

@@ -92,10 +92,22 @@ link_fdata_cmd = os.path.join(config.test_source_root, "link_fdata.py")
 
 tool_dirs = [config.llvm_tools_dir, config.test_source_root]
 
+llvm_bolt_args = []
+
+if config.libbolt_rt_instr:
+    llvm_bolt_args.append(f"--runtime-instrumentation-lib={config.libbolt_rt_instr}")
+
+if config.libbolt_rt_hugify:
+    llvm_bolt_args.append(f"--runtime-hugify-lib={config.libbolt_rt_hugify}")
+
 tools = [
     ToolSubst("llc", unresolved="fatal"),
     ToolSubst("llvm-dwarfdump", unresolved="fatal"),
-    ToolSubst("llvm-bolt", unresolved="fatal"),
+    ToolSubst(
+        "llvm-bolt",
+        unresolved="fatal",
+        extra_args=llvm_bolt_args,
+    ),
     ToolSubst("llvm-boltdiff", unresolved="fatal"),
     ToolSubst("llvm-bolt-heatmap", unresolved="fatal"),
     ToolSubst("llvm-bat-dump", unresolved="fatal"),

bolt/test/lit.site.cfg.py.in

@@ -19,6 +19,8 @@ config.bolt_clang = "@BOLT_CLANG_EXE@"
 config.bolt_lld = "@BOLT_LLD_EXE@"
 config.targets_to_build = "@BOLT_TARGETS_TO_BUILD@"
 config.gnu_ld = "@GNU_LD_EXECUTABLE@"
+config.libbolt_rt_instr = "@LIBBOLT_RT_INSTR@"
+config.libbolt_rt_hugify = "@LIBBOLT_RT_HUGIFY@"
 
 import lit.llvm
 lit.llvm.initialize(lit_config, config)