Windows: Add support for unicode command lines

[majnemer]

Summary

The MSVCRT deliberately sends main() code-page specific characters.
This isn't too useful to LLVM as we end up converting the arguments to
UTF-16 and subsequently attempt to use the result as, for example, a
file name. Instead, we need to have the ability to access the Unicode
command line and transform it to UTF-8.

This has the distinct advantage over using the MSVC-specific wmain()
function as our entry point because:

• It doesn't work on cygwin.
• It only work on MinGW with caveats and only then on certain versions.
• We get to keep our entry point as main(). :)

N.B. This patch includes fixes to other parts of lib/Support/Windows
s.t. we would be able to take advantage of getting the Unicode paths.
E.G. clang spawning clang -cc1 would want to give it Unicode arguments.

[rnk]

lib/Support/Windows/Process.inc

178–181

It seems silly to return the new args twice, once by return value and once by out param. Maybe it would be better to return an error_code?

191

Rather than doing this dance with StringOffsets to keep stable pointers into a std::vector, you could use a caller-allocated BumpPtrAllocator to allocate the strings. That way you get stable strings by default.

lib/Support/Windows/Program.inc

51

I'd rather not loop with goto. There are a couple easy ways to rewrite this with a loop construct.

lib/Support/Windows/Windows.h

27

You should IWYU. This header doesn't use either of these, does it?

[majnemer]

lib/Support/Windows/Process.inc

178–181

The intent is not to provide the caller with meaningful state in the last two parameters, they may not be used by an implementation; for example, the POSIX implementation doesn't use the second and third arguments at all, it just returns the first parameter.

I suppose I could use ErrorOr<> instead of Optional<>.

191

This can be done.

lib/Support/Windows/Program.inc

51

What's wrong with the goto? I don't see an issue. Would the other ways result in more or less code?

[majnemer]

lib/Support/Windows/Process.inc

191

Does the following interface seem reasonable to you?

error_code
GetArgumentVector(SmallVectorImpl<const char *> &NewArgs,
                  ArrayRef<const char *> ArgsFromMain,
                  SpecificBumpPtrAllocator<char> &ArgAllocator);

[ruiu]

include/llvm/Support/Process.h

175

In Unix this function does nothing and does not use any argument. In Windows the first argument (OriginalMainArgs) is ignored. It feels like they are really different, and they don't not necessarily have to have the same interface.

I'd probably rather make this a function defined only when LLVM_ON_WIN32 is defined, and guard the call of this function in LLVM's main() with LLVM_ON_WIN32.

lib/Support/Windows/Program.inc

51

This can easily be re-written with for(;;) { ...; if (len <= buffer.capacity()) break; ...}. It seems that's more normal way to write this kind of logic than goto.

[rnk]

include/llvm/Support/Process.h

175

I'd rather avoid ifdefs in main.

lib/Support/Windows/Process.inc

191

Looks good.

If we wanted to get even more minimal, we could use a MutableArrayRef<> to modify the original argv array in place! =D If the arg counts don't match we'd return an error and leave the array untouched.

lib/Support/Windows/Program.inc

51

I was thinking:

DWORD len = progNameUnicode.capacity();
do {
  progNameUnicode.resize(len);  // should be a nop on the first iteration.
  len = SearchPathW(...);
} while (len > buffer.capacity());

[ruiu]

include/llvm/Support/Process.h

175

I thought the same thing first, but found that we already have an #ifdef _WIN32 in main.

[majnemer]

Address review comments.

[rnk]

LGTM

[ygao]

In lib/Support/Windows/Signals.inc,
+ HMODULE hLib = ::LoadLibrary(TEXT("Dbghelp.dll"));

Will LoadLibrary accept unicode file names, or do you have to use LoadLibraryW instead?

autoconf/configure.ac

 dnl=== SECTION 5: Check for libraries
 dnl===
 dnl===-----------------------------------------------------------------------===
 
 AC_CHECK_LIB(m,sin)
 if test "$llvm_cv_os_type" = "MingW" ; then
   AC_CHECK_LIB(imagehlp, main)
   AC_CHECK_LIB(psapi, main)
+  AC_CHECK_LIB(shell32, main)
 fi
 
 dnl dlopen() is required for plugin support.
 AC_SEARCH_LIBS(dlopen,dl,AC_DEFINE([HAVE_DLOPEN],[1],
                [Define if dlopen() is available on this platform.]),
                AC_MSG_WARN([dlopen() not found - disabling plugin support]))
 
 dnl Search for the clock_gettime() function. Note that we rely on the POSIX

cmake/config-ix.cmake

else ()
   if (EXISTS ${CMAKE_SOURCE_DIR}/lib/Target/${LLVM_NATIVE_ARCH}/Disassembler/CMakeLists.txt)
     set(LLVM_NATIVE_DISASSEMBLER LLVMInitialize${LLVM_NATIVE_ARCH}Disassembler)
   endif ()
 endif ()
 
 if( MINGW )
   set(HAVE_LIBIMAGEHLP 1)
   set(HAVE_LIBPSAPI 1)
+  set(HAVE_LIBSHELL32 1)
   # TODO: Check existence of libraries.
   #   include(CheckLibraryExists)
   #   CHECK_LIBRARY_EXISTS(imagehlp ??? . HAVE_LIBIMAGEHLP)
 endif( MINGW )
 
 if (NOT HAVE_STRTOLL)
   # Use _strtoi64 if strtoll is not available.
   check_symbol_exists(_strtoi64 stdlib.h have_strtoi64)

cmake/modules/LLVM-Config.cmake

 function(get_system_libs return_var)
   # Returns in `return_var' a list of system libraries used by LLVM.
   if( NOT MSVC )
     if( MINGW )
-      set(system_libs ${system_libs} imagehlp psapi)
+      set(system_libs ${system_libs} imagehlp psapi shell32)
     elseif( CMAKE_HOST_UNIX )
       if( HAVE_LIBRT )
         set(system_libs ${system_libs} rt)
       endif()
       if( HAVE_LIBDL )
         set(system_libs ${system_libs} ${CMAKE_DL_LIBS})
       endif()
       if(LLVM_ENABLE_TERMINFO)

cmake/modules/TableGen.cmake

if( ${${project}_TABLEGEN} STREQUAL "${target}" )
       add_custom_target(${project}NativeTableGen DEPENDS ${${project}_TABLEGEN_EXE})
       add_dependencies(${project}NativeTableGen ConfigureNativeTableGen)
 
       add_dependencies(${target} ${project}NativeTableGen)
     endif()
   endif()
 
   if( MINGW )
-    target_link_libraries(${target} imagehlp psapi)
+    target_link_libraries(${target} imagehlp psapi shell32)
     if(CMAKE_SIZEOF_VOID_P MATCHES "8")
       set_target_properties(${target} PROPERTIES LINK_FLAGS -Wl,--stack,16777216)
     endif(CMAKE_SIZEOF_VOID_P MATCHES "8")
   endif( MINGW )
   if( LLVM_ENABLE_THREADS AND HAVE_LIBPTHREAD AND NOT BEOS )
     target_link_libraries(${target} pthread)
   endif()
 
   if (${project} STREQUAL LLVM AND NOT LLVM_INSTALL_TOOLCHAIN_ONLY)
     install(TARGETS ${target} RUNTIME DESTINATION bin)
   endif()
 endmacro()

configure

if test $ac_cv_lib_psapi_main = yes; then
   cat >>confdefs.h <<_ACEOF
 #define HAVE_LIBPSAPI 1
 _ACEOF
 
   LIBS="-lpsapi $LIBS"
 
 fi
 
+
+{ echo "$as_me:$LINENO: checking for main in -lshell32" >&5
+echo $ECHO_N "checking for main in -lshell32... $ECHO_C" >&6; }
+if test "${ac_cv_lib_shell32_main+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lshell32  $LIBS"
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+return main ();
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } &&
+   { ac_try='test -z "$ac_c_werror_flag" || test ! -s conftest.err'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; } &&
+   { ac_try='test -s conftest$ac_exeext'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  ac_cv_lib_shell32_main=yes
+else
+  echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_cv_lib_shell32_main=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext \
+      conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ echo "$as_me:$LINENO: result: $ac_cv_lib_shell32_main" >&5
+echo "${ECHO_T}$ac_cv_lib_shell32_main" >&6; }
+if test $ac_cv_lib_shell32_main = yes; then
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_LIBSHELL32 1
+_ACEOF
+
+  LIBS="-lshell32 $LIBS"
+
+fi
+
 fi
 
 { echo "$as_me:$LINENO: checking for library containing dlopen" >&5
 echo $ECHO_N "checking for library containing dlopen... $ECHO_C" >&6; }
 if test "${ac_cv_search_dlopen+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
   ac_func_search_save_LIBS=$LIBS

include/llvm/Config/config.h.cmake

 #undef HAVE_LIBM
 
 /* Define to 1 if you have the `psapi' library (-lpsapi). */
 #cmakedefine HAVE_LIBPSAPI ${HAVE_LIBPSAPI}
 
 /* Define to 1 if you have the `pthread' library (-lpthread). */
 #cmakedefine HAVE_LIBPTHREAD ${HAVE_LIBPTHREAD}
 
+/* Define to 1 if you have the `shell32' library (-lshell32). */
+#cmakedefine HAVE_LIBSHELL32 ${HAVE_LIBSHELL32}
+
 /* Define to 1 if you have the `udis86' library (-ludis86). */
 #undef HAVE_LIBUDIS86
 
 /* Define to 1 if you have the 'z' library (-lz). */
 #cmakedefine HAVE_LIBZ ${HAVE_LIBZ}
 
 /* Define to 1 if you have the <limits.h> header file. */
 #cmakedefine HAVE_LIMITS_H ${HAVE_LIMITS_H}

include/llvm/Config/config.h.in

 #undef HAVE_LIBM
 
 /* Define to 1 if you have the `psapi' library (-lpsapi). */
 #undef HAVE_LIBPSAPI
 
 /* Define to 1 if you have the `pthread' library (-lpthread). */
 #undef HAVE_LIBPTHREAD
 
+/* Define to 1 if you have the `shell32' library (-lshell32). */
+#undef HAVE_LIBSHELL32
+
 /* Define to 1 if you have the `udis86' library (-ludis86). */
 #undef HAVE_LIBUDIS86
 
 /* Define to 1 if you have the `z' library (-lz). */
 #undef HAVE_LIBZ
 
 /* Define if you can use -rdynamic. */
 #undef HAVE_LINK_EXPORT_DYNAMIC

include/llvm/Support/Process.h

 /// current process. The goal is to migrate users of this API over to the new
 /// interfaces.
 ///
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_PROCESS_H
 #define LLVM_SUPPORT_PROCESS_H
 
+#include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/Optional.h"
+#include "llvm/ADT/OwningPtr.h"
 #include "llvm/Config/llvm-config.h"
 #include "llvm/Support/DataTypes.h"
 #include "llvm/Support/TimeValue.h"
 
 namespace llvm {
 class StringRef;
 
 namespace sys {
public:
   /// occur when a program fails.
   /// @brief Prevent core file generation.
   static void PreventCoreFiles();
 
   // This function returns the environment variable \arg name's value as a UTF-8
   // string. \arg Name is assumed to be in UTF-8 encoding too.
   static Optional<std::string> GetEnv(StringRef name);
 
+  /// This function, given arguments passed to main, returns arguments suitable
+  /// for program execution.
+  static Optional<ArrayRef<const char *> >
+  GetArgumentVector(ArrayRef<const char *> OriginalMainArgs,
+                    OwningArrayPtr<char *> &NewArgVectorPointers,
+                    std::vector<char> &NewArgVectorStrings);
+
   /// This function determines if the standard input is connected directly
   /// to a user's input (keyboard probably), rather than coming from a file
   /// or pipe.
   static bool StandardInIsUserInput();
 
   /// This function determines if the standard output is connected to a
   /// "tty" or "console" window. That is, the output would be displayed to
   /// the user rather than being put on a pipe or stored in a file.

lib/Support/Unix/Process.inc

 Optional<std::string> Process::GetEnv(StringRef Name) {
   std::string NameStr = Name.str();
   const char *Val = ::getenv(NameStr.c_str());
   if (!Val)
     return None;
   return std::string(Val);
 }
 
+Optional<ArrayRef<const char *> >
+Process::GetArgumentVector(ArrayRef<const char *> OriginalArgVector,
+                           OwningArrayPtr<char *> &, std::vector<char> &) {
+  return OriginalArgVector;
+}
+
 bool Process::StandardInIsUserInput() {
   return FileDescriptorIsDisplayed(STDIN_FILENO);
 }
 
 bool Process::StandardOutIsDisplayed() {
   return FileDescriptorIsDisplayed(STDOUT_FILENO);
 }
 

lib/Support/Windows/DynamicLibrary.inc

if (!filename) {
     if (OpenedHandles == 0)
       OpenedHandles = new DenseSet<HMODULE>();
 
     EnumerateLoadedModules(GetCurrentProcess(), ELM_Callback, 0);
     // Dummy library that represents "search all handles".
     // This is mostly to ensure that the return value still shows up as "valid".
     return DynamicLibrary(&OpenedHandles);
   }
-  
-  HMODULE a_handle = LoadLibrary(filename);
+  SmallVector<wchar_t, 128> filenameUnicode;
+  if (error_code ec = windows::UTF8ToUTF16(filename, filenameUnicode)) {
+    SetLastError(ec.value());
+    MakeErrMsg(errMsg, std::string(filename) + ": Can't convert to UTF-16: ");
+    return DynamicLibrary();
+  }
+  
+  HMODULE a_handle = LoadLibraryW(filenameUnicode.data());
 
   if (a_handle == 0) {
     MakeErrMsg(errMsg, std::string(filename) + ": Can't open : ");
     return DynamicLibrary();
   }
 
   if (OpenedHandles == 0)
     OpenedHandles = new DenseSet<HMODULE>();

lib/Support/Windows/Path.inc

retry_random_path:
   random_path_utf16.set_size(0);
   for (SmallVectorImpl<wchar_t>::const_iterator i = model_utf16.begin(),
                                                 e = model_utf16.end();
                                                 i != e; ++i) {
     if (*i == L'%') {
       BYTE val = 0;
       if (!::CryptGenRandom(CryptoProvider, 1, &val))
           return windows_error(::GetLastError());
-      random_path_utf16.push_back("0123456789abcdef"[val & 15]);
+      random_path_utf16.push_back(L"0123456789abcdef"[val & 15]);
     }
     else
       random_path_utf16.push_back(*i);
   }
   // Make random_path_utf16 null terminated.
   random_path_utf16.push_back(0);
   random_path_utf16.pop_back();
 
retry_cur_dir:
 
   // If there's insufficient space, the len returned is larger than the len
   // given.
   if (len > cur_path.capacity()) {
     cur_path.reserve(len);
     goto retry_cur_dir;
   }
 
+  // On success, GetCurrentDirectoryW returns the number of characters not
+  // including the null-terminator.
   cur_path.set_size(len);
   return UTF16ToUTF8(cur_path.begin(), cur_path.size(), result);
 }
 
 error_code create_directory(const Twine &path, bool &existed) {
   SmallString<128> path_storage;
   SmallVector<wchar_t, 128> path_utf16;
 

lib/Support/Windows/Process.inc

 //
 //===----------------------------------------------------------------------===//
 
 #include "Windows.h"
 #include <direct.h>
 #include <io.h>
 #include <malloc.h>
 #include <psapi.h>
+#include <Shellapi.h>
 
 #ifdef __MINGW32__
  #if (HAVE_LIBPSAPI != 1)
   #error "libpsapi.a should be present"
  #endif
+ #if (HAVE_LIBSHELL32 != 1)
+  #error "libshell32.a should be present"
+ #endif
 #else
- #pragma comment(lib, "psapi.lib")
+#pragma comment(lib, "Shell32.lib")
 #endif
 
 //===----------------------------------------------------------------------===//
 //=== WARNING: Implementation here must contain only Win32 specific code
 //===          and must not be UNIX code
 //===----------------------------------------------------------------------===//
 
 #ifdef __MINGW32__
Optional<std::string> Process::GetEnv(StringRef Name) {
 
   // Convert the result from UTF-16 to UTF-8.
   SmallVector<char, 128> Res;
   if (error_code ec = windows::UTF16ToUTF8(&Buf[0], Size, Res))
     return None;
   return std::string(&Res[0]);
 }
 
+Optional<ArrayRef<const char *> >
+Process::GetArgumentVector(ArrayRef<const char *>,
+                           OwningArrayPtr<char *> &NewArgVectorPointers,
+                           std::vector<char> &NewArgVectorStrings) {
+  int NewArgCount;
+  error_code ec;
+
+  wchar_t **UnicodeCommandLine =
+      CommandLineToArgvW(GetCommandLineW(), &NewArgCount);
+  if (!UnicodeCommandLine)
+    return None;
+
+  SmallVector<size_t, 256> StringOffsets;
+  StringOffsets.reserve(NewArgCount);
+
+  for (int i = 0; i < NewArgCount; ++i) {
+    SmallVector<char, 128> NewArgString;
+    ec = windows::UTF16ToUTF8(UnicodeCommandLine[i],
+                              wcslen(UnicodeCommandLine[i]), NewArgString);
+    if (ec)
+      break;
+
+    StringOffsets.push_back(NewArgVectorStrings.size());
+    NewArgVectorStrings.insert(NewArgVectorStrings.end(), NewArgString.begin(),
+                               NewArgString.end());
+    NewArgVectorStrings.push_back(0);
+  }
+  LocalFree(UnicodeCommandLine);
+  if (ec)
+    return None;
+
+  NewArgVectorPointers.reset(new char *[NewArgCount + 1]);
+  for (int i = 0; i < NewArgCount; ++i)
+    NewArgVectorPointers[i] = NewArgVectorStrings.data() + StringOffsets[i];
+  NewArgVectorPointers[NewArgCount] = 0;
+
+  return ArrayRef<const char *>(NewArgVectorPointers.get(), NewArgCount);
+}
+
 bool Process::StandardInIsUserInput() {
   return FileDescriptorIsDisplayed(0);
 }
 
 bool Process::StandardOutIsDisplayed() {
   return FileDescriptorIsDisplayed(1);
 }
 

lib/Support/Windows/Program.inc

std::string sys::FindProgramByName(const std::string &progName) {
   std::string temp = progName;
   // Return paths with slashes verbatim.
   if (progName.find('\\') != std::string::npos ||
       progName.find('/') != std::string::npos)
     return temp;
 
   // At this point, the file name is valid and does not contain slashes.
   // Let Windows search for it.
-  std::string buffer;
-  buffer.resize(MAX_PATH);
-  char *dummy = NULL;
-  DWORD len = SearchPath(NULL, progName.c_str(), ".exe", MAX_PATH,
-                         &buffer[0], &dummy);
+  SmallVector<wchar_t, 128> buffer;
+  SmallVector<wchar_t, 128> progNameUnicode;
+  wchar_t *dummy = NULL;
+  if (windows::UTF8ToUTF16(progName, progNameUnicode))
+    return "";
+
+retry_search_path:
+  DWORD len = SearchPathW(NULL, progNameUnicode.data(), L".exe",
+                          buffer.capacity(), buffer.data(), &dummy);
 
   // See if it wasn't found.
   if (len == 0)
     return "";
 
-  // See if we got the entire path.
-  if (len < MAX_PATH)
-    return buffer;
-
   // Buffer was too small; grow and retry.
-  while (true) {
-    buffer.resize(len+1);
-    DWORD len2 = SearchPath(NULL, progName.c_str(), ".exe", len+1, &buffer[0], &dummy);
-
-    // It is unlikely the search failed, but it's always possible some file
-    // was added or removed since the last search, so be paranoid...
-    if (len2 == 0)
-      return "";
-    else if (len2 <= len)
-      return buffer;
+  if (len > buffer.capacity()) {
+    buffer.reserve(len);
+    goto retry_search_path;
+  }
+
+  buffer.set_size(len);
+  SmallVector<char, 128> result;
+  if (windows::UTF16ToUTF8(buffer.begin(), buffer.size(), result))
 
-    len = len2;
-  }
+  return std::string(result.data(), result.size());
 }
 
 static HANDLE RedirectIO(const StringRef *path, int fd, std::string* ErrMsg) {
   HANDLE h;
   if (path == 0) {
-    DuplicateHandle(GetCurrentProcess(), (HANDLE)_get_osfhandle(fd),
+    if (!DuplicateHandle(GetCurrentProcess(), (HANDLE)_get_osfhandle(fd),
                          GetCurrentProcess(), &h,
-                    0, TRUE, DUPLICATE_SAME_ACCESS);
+                         0, TRUE, DUPLICATE_SAME_ACCESS))
+      return INVALID_HANDLE_VALUE;
     return h;
   }
 
   std::string fname;
   if (path->empty())
     fname = "NUL";
   else
     fname = *path;
 
   SECURITY_ATTRIBUTES sa;
   sa.nLength = sizeof(sa);
   sa.lpSecurityDescriptor = 0;
   sa.bInheritHandle = TRUE;
 
-  h = CreateFile(fname.c_str(), fd ? GENERIC_WRITE : GENERIC_READ,
+  SmallVector<wchar_t, 128> fnameUnicode;
+  if (windows::UTF8ToUTF16(fname, fnameUnicode))
+    return INVALID_HANDLE_VALUE;
+
+  h = CreateFileW(fnameUnicode.data(), fd ? GENERIC_WRITE : GENERIC_READ,
                   FILE_SHARE_READ, &sa, fd == 0 ? OPEN_EXISTING : CREATE_ALWAYS,
                   FILE_ATTRIBUTE_NORMAL, NULL);
   if (h == INVALID_HANDLE_VALUE) {
     MakeErrMsg(ErrMsg, std::string(fname) + ": Can't open file for " +
         (fd ? "input: " : "output: "));
   }
 
   return h;
 }
 
if (needsQuoting) {
       *p++ = '"';
     }
     *p++ = ' ';
   }
 
   *p = 0;
 
   // The pointer to the environment block for the new process.
-  OwningArrayPtr<char> envblock;
+  std::vector<wchar_t> EnvBlock;
 
   if (envp) {
     // An environment block consists of a null-terminated block of
     // null-terminated strings. Convert the array of environment variables to
     // an environment block by concatenating them.
-
-    // First, determine the length of the environment block.
     len = 0;
-    for (unsigned i = 0; envp[i]; i++)
-      len += strlen(envp[i]) + 1;
-
-    // Now build the environment block.
-    envblock.reset(new char[len+1]);
-    p = envblock.get();
-
-    for (unsigned i = 0; envp[i]; i++) {
-      const char *ev = envp[i];
-      size_t len = strlen(ev) + 1;
-      memcpy(p, ev, len);
-      p += len;
+    for (unsigned i = 0; envp[i]; ++i) {
+      SmallVector<wchar_t, 128> EnvString;
+      if (error_code ec = windows::UTF8ToUTF16(envp[i], EnvString)) {
+        SetLastError(ec.value());
+        MakeErrMsg(ErrMsg, "Unable to convert environment variable to UTF-16");
+        return false;
       }
 
-    *p = 0;
+      EnvBlock.insert(EnvBlock.end(), EnvString.begin(), EnvString.end());
+      EnvBlock.push_back(0);
+    }
+    EnvBlock.push_back(0);
   }
 
   // Create a child process.
-  STARTUPINFO si;
+  STARTUPINFOW si;
   memset(&si, 0, sizeof(si));
   si.cb = sizeof(si);
   si.hStdInput = INVALID_HANDLE_VALUE;
   si.hStdOutput = INVALID_HANDLE_VALUE;
   si.hStdError = INVALID_HANDLE_VALUE;
 
   if (redirects) {
     si.dwFlags = STARTF_USESTDHANDLES;
 
     si.hStdInput = RedirectIO(redirects[0], 0, ErrMsg);
     if (si.hStdInput == INVALID_HANDLE_VALUE) {
       MakeErrMsg(ErrMsg, "can't redirect stdin");
       return false;
     }
     si.hStdOutput = RedirectIO(redirects[1], 1, ErrMsg);
     if (si.hStdOutput == INVALID_HANDLE_VALUE) {
       CloseHandle(si.hStdInput);
       MakeErrMsg(ErrMsg, "can't redirect stdout");
       return false;
     }
     if (redirects[1] && redirects[2] && *(redirects[1]) == *(redirects[2])) {
       // If stdout and stderr should go to the same place, redirect stderr
       // to the handle already open for stdout.
-      DuplicateHandle(GetCurrentProcess(), si.hStdOutput,
+      if (!DuplicateHandle(GetCurrentProcess(), si.hStdOutput,
                            GetCurrentProcess(), &si.hStdError,
-                      0, TRUE, DUPLICATE_SAME_ACCESS);
+                           0, TRUE, DUPLICATE_SAME_ACCESS)) {
+        CloseHandle(si.hStdInput);
+        CloseHandle(si.hStdOutput);
+        MakeErrMsg(ErrMsg, "can't dup stderr to stdout");
+        return false;
+      }
     } else {
       // Just redirect stderr
       si.hStdError = RedirectIO(redirects[2], 2, ErrMsg);
       if (si.hStdError == INVALID_HANDLE_VALUE) {
         CloseHandle(si.hStdInput);
         CloseHandle(si.hStdOutput);
         MakeErrMsg(ErrMsg, "can't redirect stderr");
         return false;
       }
     }
   }
 
   PROCESS_INFORMATION pi;
   memset(&pi, 0, sizeof(pi));
 
   fflush(stdout);
   fflush(stderr);
-  std::string ProgramStr = Program;
-  BOOL rc = CreateProcess(ProgramStr.c_str(), command.get(), NULL, NULL, TRUE,
-                          0, envblock.get(), NULL, &si, &pi);
+
+  SmallVector<wchar_t, 128> ProgramUtf16;
+  if (error_code ec = windows::UTF8ToUTF16(Program, ProgramUtf16)) {
+    SetLastError(ec.value());
+    MakeErrMsg(ErrMsg,
+               std::string("Unable to convert application name to UTF-16"));
+    return false;
+  }
+
+  SmallVector<wchar_t, 128> CommandUtf16;
+  if (error_code ec = windows::UTF8ToUTF16(command.get(), CommandUtf16)) {
+    SetLastError(ec.value());
+    MakeErrMsg(ErrMsg,
+               std::string("Unable to convert command-line to UTF-16"));
+    return false;
+  }
+
+  BOOL rc = CreateProcessW(ProgramUtf16.data(), CommandUtf16.data(), 0, 0, TRUE,
+                           CREATE_UNICODE_ENVIRONMENT,
+                           EnvBlock.empty() ? 0 : EnvBlock.data(), 0, &si, &pi);
   DWORD err = GetLastError();
 
   // Regardless of whether the process got created or not, we are done with
   // the handles we created for it to inherit.
   CloseHandle(si.hStdInput);
   CloseHandle(si.hStdOutput);
   CloseHandle(si.hStdError);
 
   // Now return an error if the process didn't get created.
   if (!rc) {
     SetLastError(err);
     MakeErrMsg(ErrMsg, std::string("Couldn't execute program '") +
-               ProgramStr + "'");
+               Program.str() + "'");
     return false;
   }
 
   PI.Pid = pi.dwProcessId;
   PI.ProcessHandle = pi.hProcess;
 
   // Make sure these get closed no matter what.
   ScopedCommonHandle hThread(pi.hThread);

lib/Support/Windows/Signals.inc

 typedef BOOL (WINAPI *fpSymGetLineFromAddr64)(HANDLE, DWORD64,
                       PDWORD, PIMAGEHLP_LINE64);
 static fpSymGetLineFromAddr64 SymGetLineFromAddr64;
 
 typedef PVOID (WINAPI *fpSymFunctionTableAccess64)(HANDLE, DWORD64);
 static fpSymFunctionTableAccess64 SymFunctionTableAccess64;
 
 static bool load64BitDebugHelp(void) {
-  HMODULE hLib = ::LoadLibrary("Dbghelp.dll");
+  HMODULE hLib = ::LoadLibrary(TEXT("Dbghelp.dll"));
   if (hLib) {
     StackWalk64 = (fpStackWalk64)
                       ::GetProcAddress(hLib, "StackWalk64");
     SymGetModuleBase64 = (fpSymGetModuleBase64)
                       ::GetProcAddress(hLib, "SymGetModuleBase64");
     SymGetSymFromAddr64 = (fpSymGetSymFromAddr64)
                       ::GetProcAddress(hLib, "SymGetSymFromAddr64");
     SymGetLineFromAddr64 = (fpSymGetLineFromAddr64)

lib/Support/Windows/Windows.h

 // mingw-w64 tends to define it as 0x0502 in its headers.
 #undef _WIN32_WINNT
 
 // Require at least Windows XP(5.1) API.
 #define _WIN32_WINNT 0x0501
 #define _WIN32_IE    0x0600 // MinGW at it again.
 #define WIN32_LEAN_AND_MEAN
 
+#include "llvm/ADT/ArrayRef.h"
+#include "llvm/ADT/OwningPtr.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Config/config.h" // Get build system configuration settings
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/system_error.h"
 #include <windows.h>
 #include <wincrypt.h>
-#include <shlobj.h>
 #include <cassert>
 #include <string>
 #include <vector>
 
 inline bool MakeErrMsg(std::string* ErrMsg, const std::string& prefix) {
   if (!ErrMsg)
     return true;
   char *buffer = NULL;

projects/sample/autoconf/configure.ac

 dnl=== SECTION 5: Check for libraries
 dnl===
 dnl===-----------------------------------------------------------------------===
 
 AC_CHECK_LIB(m,sin)
 if test "$llvm_cv_os_type" = "MingW" ; then
   AC_CHECK_LIB(imagehlp, main)
   AC_CHECK_LIB(psapi, main)
+  AC_CHECK_LIB(shell32, main)
 fi
 
 dnl dlopen() is required for plugin support.
 AC_SEARCH_LIBS(dlopen,dl,AC_DEFINE([HAVE_DLOPEN],[1],
                [Define if dlopen() is available on this platform.]),
                AC_MSG_WARN([dlopen() not found - disabling plugin support]))
 
 dnl Search for the clock_gettime() function. Note that we rely on the POSIX

projects/sample/configure

if test $ac_cv_lib_psapi_main = yes; then
   cat >>confdefs.h <<_ACEOF
 #define HAVE_LIBPSAPI 1
 _ACEOF
 
   LIBS="-lpsapi $LIBS"
 
 fi
 
+
+{ echo "$as_me:$LINENO: checking for main in -lshell32" >&5
+echo $ECHO_N "checking for main in -lshell32... $ECHO_C" >&6; }
+if test "${ac_cv_lib_shell32_main+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lshell32  $LIBS"
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+
+
+int
+main ()
+{
+return main ();
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } &&
+   { ac_try='test -z "$ac_c_werror_flag" || test ! -s conftest.err'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; } &&
+   { ac_try='test -s conftest$ac_exeext'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  ac_cv_lib_shell32_main=yes
+else
+  echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_cv_lib_shell32_main=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext \
+      conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ echo "$as_me:$LINENO: result: $ac_cv_lib_shell32_main" >&5
+echo "${ECHO_T}$ac_cv_lib_shell32_main" >&6; }
+if test $ac_cv_lib_shell32_main = yes; then
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_LIBSHELL32 1
+_ACEOF
+
+  LIBS="-lshell32 $LIBS"
+
+fi
+
 fi
 
 { echo "$as_me:$LINENO: checking for library containing dlopen" >&5
 echo $ECHO_N "checking for library containing dlopen... $ECHO_C" >&6; }
 if test "${ac_cv_search_dlopen+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
   ac_func_search_save_LIBS=$LIBS

utils/FileCheck/CMakeLists.txt

 add_llvm_utility(FileCheck
   FileCheck.cpp
   )
 
 target_link_libraries(FileCheck LLVMSupport)
 if( MINGW )
-  target_link_libraries(FileCheck imagehlp psapi)
+  target_link_libraries(FileCheck imagehlp psapi shell32)
 endif( MINGW )
 if( LLVM_ENABLE_THREADS AND HAVE_LIBPTHREAD )
   target_link_libraries(FileCheck pthread)
 endif()

utils/FileUpdate/CMakeLists.txt

 add_llvm_utility(FileUpdate
   FileUpdate.cpp
   )
 
 target_link_libraries(FileUpdate LLVMSupport)
 if( MINGW )
-  target_link_libraries(FileUpdate imagehlp psapi)
+  target_link_libraries(FileUpdate imagehlp psapi shell32)
 endif( MINGW )
 if( LLVM_ENABLE_THREADS AND HAVE_LIBPTHREAD )
   target_link_libraries(FileUpdate pthread)
 endif()

utils/not/CMakeLists.txt

 add_llvm_utility(not
   not.cpp
   )
 
 target_link_libraries(not LLVMSupport)
 if( MINGW )
-  target_link_libraries(not imagehlp psapi)
+  target_link_libraries(not imagehlp psapi shell32)
 endif( MINGW )
 if( LLVM_ENABLE_THREADS AND HAVE_LIBPTHREAD )
   target_link_libraries(not pthread)
 endif()