bindexplib: supporting llvm bitcode formats using llvm-nm

11 files changed, 144 insertions, 12 deletions

diff --git a/Source/bindexplib.cxx b/Source/bindexplib.cxx
index a107294..7e5db26 100644
--- a/Source/bindexplib.cxx
+++ b/Source/bindexplib.cxx
@@ -64,9 +64,12 @@
  */
 #include "bindexplib.h"
 
+#include "cmSystemTools.h"
 #include "cmsys/Encoding.hxx"
 #include "cmsys/FStream.hxx"
 #include <iostream>
+#include <sstream>
+#include <vector>
 #include <windows.h>
 
 #ifndef IMAGE_FILE_MACHINE_ARM
@@ -301,7 +304,63 @@ private:
   bool IsI386;
 };
 
-bool DumpFile(const char* filename, std::set<std::string>& symbols,
+bool DumpFileWithLlvmNm(std::string const& nmPath, const char* filename,
+                        std::set<std::string>& symbols,
+                        std::set<std::string>& dataSymbols)
+{
+  std::string output;
+  // break up command line into a vector
+  std::vector<std::string> command;
+  command.push_back(nmPath);
+  command.push_back("--no-weak");
+  command.push_back("--defined-only");
+  command.push_back("--format=posix");
+  command.push_back(filename);
+
+  // run the command
+  int exit_code = 0;
+  cmSystemTools::RunSingleCommand(command, &output, &output, &exit_code, "",
+                                  cmSystemTools::OUTPUT_NONE);
+
+  if (exit_code != 0) {
+    fprintf(stderr, "llvm-nm returned an error: %s\n", output.c_str());
+    return false;
+  }
+
+  std::istringstream ss(output);
+  std::string line;
+  while (std::getline(ss, line)) {
+    if (line.empty()) { // last line
+      continue;
+    }
+    size_t sym_end = line.find(" ");
+    if (sym_end == std::string::npos) {
+      fprintf(stderr, "Couldn't parse llvm-nm output line: %s\n",
+              line.c_str());
+      return false;
+    }
+    if (line.size() < sym_end + 1) {
+      fprintf(stderr, "Couldn't parse llvm-nm output line: %s\n",
+              line.c_str());
+      return false;
+    }
+    const std::string sym = line.substr(0, sym_end);
+    const char sym_type = line[sym_end + 1];
+    switch (sym_type) {
+      case 'D':
+        dataSymbols.insert(sym);
+        break;
+      case 'T':
+        symbols.insert(sym);
+        break;
+    }
+  }
+
+  return true;
+}
+
+bool DumpFile(std::string const& nmPath, const char* filename,
+              std::set<std::string>& symbols,
               std::set<std::string>& dataSymbols)
 {
   HANDLE hFile;
@@ -356,16 +415,26 @@ bool DumpFile(const char* filename, std::set<std::string>& symbols,
         (imageHeader->Machine == IMAGE_FILE_MACHINE_I386));
       symbolDumper.DumpObjFile();
     } else {
-      // check for /bigobj format
+      // check for /bigobj and llvm LTO format
       cmANON_OBJECT_HEADER_BIGOBJ* h =
         (cmANON_OBJECT_HEADER_BIGOBJ*)lpFileBase;
       if (h->Sig1 == 0x0 && h->Sig2 == 0xffff) {
+        // bigobj
         DumpSymbols<cmANON_OBJECT_HEADER_BIGOBJ, cmIMAGE_SYMBOL_EX>
           symbolDumper((cmANON_OBJECT_HEADER_BIGOBJ*)lpFileBase, symbols,
                        dataSymbols, (h->Machine == IMAGE_FILE_MACHINE_I386));
         symbolDumper.DumpObjFile();
+      } else if (
+        // BCexCODE - llvm bitcode
+        (h->Sig1 == 0x4342 && h->Sig2 == 0xDEC0) ||
+        // 0x0B17C0DE - llvm bitcode BC wrapper
+        (h->Sig1 == 0x0B17 && h->Sig2 == 0xC0DE)) {
+
+        return DumpFileWithLlvmNm(nmPath, filename, symbols, dataSymbols);
+
       } else {
-        printf("unrecognized file format in '%s'\n", filename);
+        printf("unrecognized file format in '%s, %u'\n", filename,
+               imageHeader->Machine);
         return false;
       }
     }
@@ -378,7 +447,7 @@ bool DumpFile(const char* filename, std::set<std::string>& symbols,
 
 bool bindexplib::AddObjectFile(const char* filename)
 {
-  return DumpFile(filename, this->Symbols, this->DataSymbols);
+  return DumpFile(NmPath, filename, this->Symbols, this->DataSymbols);
 }
 
 bool bindexplib::AddDefinitionFile(const char* filename)
@@ -419,3 +488,8 @@ void bindexplib::WriteFile(FILE* file)
     fprintf(file, "\t%s\n", s.c_str());
   }
 }
+
+void bindexplib::SetNmPath(std::string const& nm)
+{
+  NmPath = nm;
+}
diff --git a/Source/bindexplib.h b/Source/bindexplib.h
index 3e22ac7..6c066c5 100644
--- a/Source/bindexplib.h
+++ b/Source/bindexplib.h
@@ -12,13 +12,16 @@
 class bindexplib
 {
 public:
-  bindexplib() {}
+  bindexplib() { NmPath = "nm"; }
   bool AddDefinitionFile(const char* filename);
   bool AddObjectFile(const char* filename);
   void WriteFile(FILE* file);
 
+  void SetNmPath(std::string const& nm);
+
 private:
   std::set<std::string> Symbols;
   std::set<std::string> DataSymbols;
+  std::string NmPath;
 };
 #endif
diff --git a/Source/cmMakefileTargetGenerator.cxx b/Source/cmMakefileTargetGenerator.cxx
index f99fe4e..f84ec76 100644
--- a/Source/cmMakefileTargetGenerator.cxx
+++ b/Source/cmMakefileTargetGenerator.cxx
@@ -15,6 +15,7 @@
 #include "cmGeneratorExpression.h"
 #include "cmGeneratorTarget.h"
 #include "cmGlobalUnixMakefileGenerator3.h"
+#include "cmLocalCommonGenerator.h"
 #include "cmLocalUnixMakefileGenerator3.h"
 #include "cmMakefile.h"
 #include "cmMakefileExecutableTargetGenerator.h"
@@ -1738,6 +1739,12 @@ void cmMakefileTargetGenerator::GenDefFile(
     this->LocalGenerator->MaybeConvertToRelativePath(
       this->LocalGenerator->GetCurrentBinaryDirectory(), objlist_file),
     cmOutputConverter::SHELL);
+  const char* nm_executable = this->Makefile->GetDefinition("CMAKE_NM");
+  if (nm_executable && *nm_executable) {
+    cmd += " --nm=";
+    cmd += this->LocalCommonGenerator->ConvertToOutputFormat(
+      nm_executable, cmOutputConverter::SHELL);
+  }
   real_link_commands.insert(real_link_commands.begin(), cmd);
   // create a list of obj files for the -E __create_def to read
   cmGeneratedFileStream fout(objlist_file);
diff --git a/Source/cmNinjaNormalTargetGenerator.cxx b/Source/cmNinjaNormalTargetGenerator.cxx
index 17a5527..df4d45d 100644
--- a/Source/cmNinjaNormalTargetGenerator.cxx
+++ b/Source/cmNinjaNormalTargetGenerator.cxx
@@ -20,6 +20,7 @@
 #include "cmGlobalNinjaGenerator.h"
 #include "cmLinkLineComputer.h"
 #include "cmLinkLineDeviceComputer.h"
+#include "cmLocalCommonGenerator.h"
 #include "cmLocalGenerator.h"
 #include "cmLocalNinjaGenerator.h"
 #include "cmMakefile.h"
@@ -974,6 +975,13 @@ void cmNinjaNormalTargetGenerator::WriteLinkStatement()
     std::string obj_list_file = mdi->DefFile + ".objs";
     cmd += this->GetLocalGenerator()->ConvertToOutputFormat(
       obj_list_file, cmOutputConverter::SHELL);
+
+    const char* nm_executable = GetMakefile()->GetDefinition("CMAKE_NM");
+    if (nm_executable && *nm_executable) {
+      cmd += " --nm=";
+      cmd += this->LocalCommonGenerator->ConvertToOutputFormat(
+        nm_executable, cmOutputConverter::SHELL);
+    }
     preLinkCmdLines.push_back(std::move(cmd));
 
     // create a list of obj files for the -E __create_def to read
diff --git a/Source/cmcmd.cxx b/Source/cmcmd.cxx
index 2be8bae..08cad7d 100644
--- a/Source/cmcmd.cxx
+++ b/Source/cmcmd.cxx
@@ -558,8 +558,8 @@ int cmcmd::ExecuteCMakeCommand(std::vector<std::string> const& args)
 #if defined(_WIN32) && !defined(CMAKE_BOOTSTRAP)
     else if (args[1] == "__create_def") {
       if (args.size() < 4) {
-        std::cerr
-          << "__create_def Usage: -E __create_def outfile.def objlistfile\n";
+        std::cerr << "__create_def Usage: -E __create_def outfile.def "
+                     "objlistfile [-nm=nm-path]\n";
         return 1;
       }
       FILE* fout = cmsys::SystemTools::Fopen(args[2].c_str(), "w+");
@@ -576,6 +576,15 @@ int cmcmd::ExecuteCMakeCommand(std::vector<std::string> const& args)
       }
       std::string file;
       bindexplib deffile;
+      if (args.size() >= 5) {
+        auto a = args[4];
+        if (cmHasLiteralPrefix(a, "--nm=")) {
+          deffile.SetNmPath(a.substr(5));
+          std::cerr << a.substr(5) << "\n";
+        } else {
+          std::cerr << "unknown argument: " << a << "\n";
+        }
+      }
       while (cmSystemTools::GetLineFromStream(fin, file)) {
         std::string const& ext = cmSystemTools::GetFilenameLastExtension(file);
         if (cmSystemTools::LowerCase(ext) == ".def") {
diff --git a/Tests/Module/CheckIPOSupported-C/CMakeLists.txt b/Tests/Module/CheckIPOSupported-C/CMakeLists.txt
index 4a41a98..c5cd03e 100644
--- a/Tests/Module/CheckIPOSupported-C/CMakeLists.txt
+++ b/Tests/Module/CheckIPOSupported-C/CMakeLists.txt
@@ -13,8 +13,18 @@ elseif(CMake_TEST_IPO_WORKS_C)
 endif()
 
 add_library(foo foo.c)
+if(NOT CYGWIN AND (NOT WIN32 OR "x${CMAKE_C_COMPILER_ID}" STREQUAL "xClang"))
+  add_library(bar SHARED bar.c)
+  if(WIN32)
+    # Bindexplib for clang supports LTO objects
+    set_target_properties(bar PROPERTIES WINDOWS_EXPORT_ALL_SYMBOLS ON)
+  endif()
+else()
+  # TODO: bindexplib doesn't support exporting IPO symbols with other compilers on Windows
+  add_library(bar STATIC bar.c)
+endif()
 add_executable(CheckIPOSupported-C main.c)
-target_link_libraries(CheckIPOSupported-C PUBLIC foo)
+target_link_libraries(CheckIPOSupported-C PUBLIC foo bar)
 
 enable_testing()
 add_test(NAME CheckIPOSupported-C COMMAND CheckIPOSupported-C)
diff --git a/Tests/Module/CheckIPOSupported-C/bar.c b/Tests/Module/CheckIPOSupported-C/bar.c
new file mode 100644
index 0000000..680f213
--- /dev/null
+++ b/Tests/Module/CheckIPOSupported-C/bar.c
@@ -0,0 +1,4 @@
+int bar()
+{
+  return 0x42;
+}
diff --git a/Tests/Module/CheckIPOSupported-C/main.c b/Tests/Module/CheckIPOSupported-C/main.c
index 99204ab..28ab26f 100644
--- a/Tests/Module/CheckIPOSupported-C/main.c
+++ b/Tests/Module/CheckIPOSupported-C/main.c
@@ -1,8 +1,9 @@
 int foo();
+int bar();
 
 int main()
 {
-  if (foo() == 0) {
+  if (foo() != bar()) {
     return 1;
   }
   return 0;
diff --git a/Tests/Module/CheckIPOSupported-CXX/CMakeLists.txt b/Tests/Module/CheckIPOSupported-CXX/CMakeLists.txt
index 1bb2b84..237bf1d 100644
--- a/Tests/Module/CheckIPOSupported-CXX/CMakeLists.txt
+++ b/Tests/Module/CheckIPOSupported-CXX/CMakeLists.txt
@@ -12,9 +12,20 @@ elseif(CMake_TEST_IPO_WORKS_CXX)
   message(FATAL_ERROR "IPO expected to work, but the check failed:\n  ${ipo_output}")
 endif()
 
-add_library(foo foo.cpp)
+
+add_library(foo STATIC foo.cpp)
+if(NOT CYGWIN AND (NOT WIN32 OR "x${CMAKE_CXX_COMPILER_ID}" STREQUAL "Clang"))
+  add_library(bar SHARED bar.cpp)
+  if(WIN32)
+    # Bindexplib for clang supports LTO objects
+    set_target_properties(bar PROPERTIES WINDOWS_EXPORT_ALL_SYMBOLS ON)
+  endif()
+else()
+  # TODO: bindexplib doesn't support exporting IPO symbols with other compilers on Windows
+  add_library(bar STATIC bar.cpp)
+endif()
 add_executable(CheckIPOSupported-CXX main.cpp)
-target_link_libraries(CheckIPOSupported-CXX PUBLIC foo)
+target_link_libraries(CheckIPOSupported-CXX PUBLIC foo bar)
 
 enable_testing()
 add_test(NAME CheckIPOSupported-CXX COMMAND CheckIPOSupported-CXX)
diff --git a/Tests/Module/CheckIPOSupported-CXX/bar.cpp b/Tests/Module/CheckIPOSupported-CXX/bar.cpp
new file mode 100644
index 0000000..680f213
--- /dev/null
+++ b/Tests/Module/CheckIPOSupported-CXX/bar.cpp
@@ -0,0 +1,4 @@
+int bar()
+{
+  return 0x42;
+}
diff --git a/Tests/Module/CheckIPOSupported-CXX/main.cpp b/Tests/Module/CheckIPOSupported-CXX/main.cpp
index 99204ab..28ab26f 100644
--- a/Tests/Module/CheckIPOSupported-CXX/main.cpp
+++ b/Tests/Module/CheckIPOSupported-CXX/main.cpp
@@ -1,8 +1,9 @@
 int foo();
+int bar();
 
 int main()
 {
-  if (foo() == 0) {
+  if (foo() != bar()) {
     return 1;
   }
   return 0;