Make dump_syms output an INFO CODE_ID line that includes the code file and code identifier. (Currently disabled to give Breakpad users time to update their processor code.)

R=mark at http://breakpad.appspot.com/180001/show

git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@710 4c0a9323-5329-0410-9bdc-e9ce6186880e

4 files changed, 204 insertions, 14 deletions

diff --git a/src/common/windows/pdb_source_line_writer.cc b/src/common/windows/pdb_source_line_writer.cc
index 48a35bed..9969530b 100644
--- a/src/common/windows/pdb_source_line_writer.cc
+++ b/src/common/windows/pdb_source_line_writer.cc
@@ -28,8 +28,8 @@
 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 #include <atlbase.h>
-#include <DbgHelp.h>
 #include <dia2.h>
+#include <ImageHlp.h>
 #include <stdio.h>
 
 #include "common/windows/string_utils-inl.h"
@@ -45,6 +45,24 @@
 
 namespace google_breakpad {
 
+using std::vector;
+
+// A helper class to scope a PLOADED_IMAGE.
+class AutoImage {
+ public:
+  explicit AutoImage(PLOADED_IMAGE img) : img_(img) {}
+  ~AutoImage() {
+    if (img_)
+      ImageUnload(img_);
+  }
+
+  operator PLOADED_IMAGE() { return img_; }
+  PLOADED_IMAGE operator->() { return img_; }
+
+ private:
+  PLOADED_IMAGE img_;
+};
+
 PDBSourceLineWriter::PDBSourceLineWriter() : output_(NULL) {
 }
 
@@ -78,6 +96,7 @@ bool PDBSourceLineWriter::Open(const wstring &file, FileFormat format) {
         fprintf(stderr, "loadDataForExe failed\n");
         return false;
       }
+      code_file_ = file;
       break;
     case ANY_FILE:
       if (FAILED(data_source->loadDataFromPdb(file.c_str()))) {
@@ -85,6 +104,7 @@ bool PDBSourceLineWriter::Open(const wstring &file, FileFormat format) {
           fprintf(stderr, "loadDataForPdb and loadDataFromExe failed\n");
           return false;
         }
+	code_file_ = file;
       }
       break;
     default:
@@ -498,6 +518,18 @@ bool PDBSourceLineWriter::PrintPDBInfo() {
   return true;
 }
 
+bool PDBSourceLineWriter::PrintPEInfo() {
+  PEModuleInfo info;
+  if (!GetPEInfo(&info)) {
+    return false;
+  }
+
+  fprintf(output_, "INFO CODE_ID %ws %ws\n",
+	  info.code_identifier.c_str(),
+	  info.code_file.c_str());
+  return true;
+}
+
 // wcstol_positive_strict is sort of like wcstol, but much stricter.  string
 // should be a buffer pointing to a null-terminated string containing only
 // decimal digits.  If the entire string can be converted to an integer
@@ -535,6 +567,35 @@ static bool wcstol_positive_strict(wchar_t *string, int *result) {
   return true;
 }
 
+bool PDBSourceLineWriter::FindPEFile() {
+  CComPtr<IDiaSymbol> global;
+  if (FAILED(session_->get_globalScope(&global))) {
+    fprintf(stderr, "get_globalScope failed\n");
+    return false;
+  }
+
+  CComBSTR symbols_file;
+  if (SUCCEEDED(global->get_symbolsFileName(&symbols_file))) {
+    wstring file(symbols_file);
+    
+    // Look for an EXE or DLL file.
+    const wchar_t *extensions[] = { L"exe", L"dll" };
+    for (int i = 0; i < sizeof(extensions) / sizeof(extensions[0]); i++) {
+      size_t dot_pos = file.find_last_of(L".");
+      if (dot_pos != wstring::npos) {
+	file.replace(dot_pos + 1, wstring::npos, extensions[i]);
+	// Check if this file exists.
+	if (GetFileAttributesW(file.c_str()) != INVALID_FILE_ATTRIBUTES) {
+	  code_file_ = file;
+	  return true;
+	}
+      }
+    }
+  }
+
+  return false;
+}
+
 // static
 bool PDBSourceLineWriter::GetSymbolFunctionName(IDiaSymbol *function,
                                                 BSTR *name,
@@ -737,10 +798,15 @@ next_child:
 bool PDBSourceLineWriter::WriteMap(FILE *map_file) {
   output_ = map_file;
 
-  bool ret = PrintPDBInfo() &&
-             PrintSourceFiles() && 
-             PrintFunctions() &&
-             PrintFrameData();
+  bool ret = PrintPDBInfo();
+  // TODO(ted): This is currently disabled to allow Breakpad users to update
+  // processing infrastructure.
+  // This is not a critical piece of the symbol file.
+  // PrintPEInfo();
+  ret = ret &&
+    PrintSourceFiles() && 
+    PrintFunctions() &&
+    PrintFrameData();
 
   output_ = NULL;
   return ret;
@@ -846,6 +912,54 @@ bool PDBSourceLineWriter::GetModuleInfo(PDBModuleInfo *info) {
   return true;
 }
 
+bool PDBSourceLineWriter::GetPEInfo(PEModuleInfo *info) {
+  if (!info) {
+    return false;
+  }
+
+  if (code_file_.empty() && !FindPEFile()) {
+    fprintf(stderr, "Couldn't locate EXE or DLL file.\n");
+    return false;
+  }
+
+  // Convert wchar to native charset because ImageLoad only takes
+  // a PSTR as input.
+  string code_file;
+  if (!WindowsStringUtils::safe_wcstombs(code_file_, &code_file)) {
+    return false;
+  }
+
+  AutoImage img(ImageLoad((PSTR)code_file.c_str(), NULL));
+  if (!img) {
+    fprintf(stderr, "Failed to open PE file: %s\n", code_file.c_str());
+    return false;
+  }
+
+  info->code_file = WindowsStringUtils::GetBaseName(code_file_);
+
+  // The date and time that the file was created by the linker.
+  DWORD TimeDateStamp = img->FileHeader->FileHeader.TimeDateStamp;
+  // The size of the file in bytes, including all headers.
+  DWORD SizeOfImage = 0;
+  PIMAGE_OPTIONAL_HEADER64 opt =
+    &((PIMAGE_NT_HEADERS64)img->FileHeader)->OptionalHeader;
+  if (opt->Magic == IMAGE_NT_OPTIONAL_HDR64_MAGIC) {
+    // 64-bit PE file.
+    SizeOfImage = opt->SizeOfImage;
+  }
+  else {
+    // 32-bit PE file.
+    SizeOfImage = img->FileHeader->OptionalHeader.SizeOfImage;
+  }
+  wchar_t code_identifier[32];
+  swprintf(code_identifier,
+	   sizeof(code_identifier) / sizeof(code_identifier[0]),
+	   L"%08X%X", TimeDateStamp, SizeOfImage);
+  info->code_identifier = code_identifier;
+
+  return true;
+}
+
 bool PDBSourceLineWriter::UsesGUID(bool *uses_guid) {
   if (!uses_guid)
     return false;
diff --git a/src/common/windows/pdb_source_line_writer.h b/src/common/windows/pdb_source_line_writer.h
index 8524d8b0..ef2a27db 100644
--- a/src/common/windows/pdb_source_line_writer.h
+++ b/src/common/windows/pdb_source_line_writer.h
@@ -67,6 +67,21 @@ struct PDBModuleInfo {
   wstring cpu;
 };
 
+// A structure that carries information that identifies a PE file,
+// either an EXE or a DLL.
+struct PEModuleInfo {
+  // The basename of the PE file.
+  wstring code_file;
+
+  // The PE file's code identifier, which consists of its timestamp
+  // and file size concatenated together into a single hex string.
+  // (The fields IMAGE_OPTIONAL_HEADER::SizeOfImage and
+  // IMAGE_FILE_HEADER::TimeDateStamp, as defined in the ImageHlp
+  // documentation.) This is not well documented, if it's documented
+  // at all, but it's what symstore does and what DbgHelp supports.
+  wstring code_identifier;
+};
+
 class PDBSourceLineWriter {
  public:
   enum FileFormat {
@@ -100,6 +115,10 @@ class PDBSourceLineWriter {
   // true on success and false on failure.
   bool GetModuleInfo(PDBModuleInfo *info);
 
+  // Retrieves information about the module's PE file.  Returns
+  // true on success and false on failure.
+  bool GetPEInfo(PEModuleInfo *info);
+
   // Sets uses_guid to true if the opened file uses a new-style CodeView
   // record with a 128-bit GUID, or false if the opened file uses an old-style
   // CodeView record.  When no GUID is available, a 32-bit signature should be
@@ -139,6 +158,11 @@ class PDBSourceLineWriter {
   // its uuid and age.
   bool PrintPDBInfo();
 
+  // Outputs a line identifying the PE file corresponding to the PDB
+  // file that is being dumped, along with its code identifier,
+  // which consists of its timestamp and file size.
+  bool PrintPEInfo();
+
   // Returns true if this filename has already been seen,
   // and an ID is stored for it, or false if it has not.
   bool FileIDIsCached(const wstring &file) {
@@ -170,6 +194,10 @@ class PDBSourceLineWriter {
     return iter->second;
   };
 
+  // Find the PE file corresponding to the loaded PDB file, and
+  // set the code_file_ member. Returns false on failure.
+  bool FindPEFile();
+
   // Returns the function name for a symbol.  If possible, the name is
   // undecorated.  If the symbol's decorated form indicates the size of
   // parameters on the stack, this information is returned in stack_param_size.
@@ -183,6 +211,10 @@ class PDBSourceLineWriter {
   // a failure, returns 0, which is also a valid number of bytes.
   static int GetFunctionStackParamSize(IDiaSymbol *function);
 
+  // The filename of the PE file corresponding to the currently-open
+  // pdb file.
+  wstring code_file_;
+
   // The session for the currently-open pdb file.
   CComPtr<IDiaSession> session_;
 
diff --git a/src/common/windows/string_utils-inl.h b/src/common/windows/string_utils-inl.h
index 6f650812..d281aaa1 100644
--- a/src/common/windows/string_utils-inl.h
+++ b/src/common/windows/string_utils-inl.h
@@ -87,6 +87,9 @@ class WindowsStringUtils {
   // without setting wcs.
   static bool safe_mbstowcs(const string &mbs, wstring *wcs);
 
+  // The inverse of safe_mbstowcs.
+  static bool safe_wcstombs(const wstring &wcs, string *mbs);
+
   // Returns the base name of a file, e.g. strips off the path.
   static wstring GetBaseName(const wstring &filename);
 
diff --git a/src/common/windows/string_utils.cc b/src/common/windows/string_utils.cc
index 5d464802..e6ffa082 100644
--- a/src/common/windows/string_utils.cc
+++ b/src/common/windows/string_utils.cc
@@ -28,6 +28,7 @@
 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 #include <cassert>
+#include <vector>
 
 #include "common/windows/string_utils-inl.h"
 
@@ -55,6 +56,7 @@ bool WindowsStringUtils::safe_mbstowcs(const string &mbs, wstring *wcs) {
   if ((err = mbstowcs_s(&wcs_length, NULL, 0, mbs.c_str(), _TRUNCATE)) != 0) {
     return false;
   }
+  assert(wcs_length > 0);
 #else  // _MSC_VER >= 1400
   if ((wcs_length = mbstowcs(NULL, mbs.c_str(), mbs.length())) < 0) {
     return false;
@@ -64,28 +66,67 @@ bool WindowsStringUtils::safe_mbstowcs(const string &mbs, wstring *wcs) {
   ++wcs_length;
 #endif  // _MSC_VER >= 1400
 
-  // TODO(mmentovai): move scoped_ptr into common and use it for wcs_c.
-  wchar_t *wcs_c = new wchar_t[wcs_length];
+  std::vector<wchar_t> wcs_v(wcs_length);
 
   // Now, convert.
 #if _MSC_VER >= 1400  // MSVC 2005/8
-  if ((err = mbstowcs_s(NULL, wcs_c, wcs_length, mbs.c_str(),
+  if ((err = mbstowcs_s(NULL, &wcs_v[0], wcs_length, mbs.c_str(),
                         _TRUNCATE)) != 0) {
-    delete[] wcs_c;
     return false;
   }
 #else  // _MSC_VER >= 1400
-  if (mbstowcs(wcs_c, mbs.c_str(), mbs.length()) < 0) {
-    delete[] wcs_c;
+  if (mbstowcs(&wcs_v[0], mbs.c_str(), mbs.length()) < 0) {
     return false;
   }
 
   // Ensure presence of 0-terminator.
-  wcs_c[wcs_length - 1] = '\0';
+  wcs_v[wcs_length - 1] = '\0';
 #endif  // _MSC_VER >= 1400
 
-  *wcs = wcs_c;
-  delete[] wcs_c;
+  *wcs = &wcs_v[0];
+  return true;
+}
+
+// static
+bool WindowsStringUtils::safe_wcstombs(const wstring &wcs, string *mbs) {
+  assert(mbs);
+
+  // First, determine the length of the destination buffer.
+  size_t mbs_length;
+
+#if _MSC_VER >= 1400  // MSVC 2005/8
+  errno_t err;
+  if ((err = wcstombs_s(&mbs_length, NULL, 0, wcs.c_str(), _TRUNCATE)) != 0) {
+    return false;
+  }
+  assert(mbs_length > 0);
+#else  // _MSC_VER >= 1400
+  if ((mbs_length = wcstombs(NULL, wcs.c_str(), wcs.length())) < 0) {
+    return false;
+  }
+
+  // Leave space for the 0-terminator.
+  ++mbs_length;
+#endif  // _MSC_VER >= 1400
+
+  std::vector<char> mbs_v(mbs_length);
+
+  // Now, convert.
+#if _MSC_VER >= 1400  // MSVC 2005/8
+  if ((err = wcstombs_s(NULL, &mbs_v[0], mbs_length, wcs.c_str(),
+                        _TRUNCATE)) != 0) {
+    return false;
+  }
+#else  // _MSC_VER >= 1400
+  if (wcstombs(&mbs_v[0], wcs.c_str(), wcs.length()) < 0) {
+    return false;
+  }
+
+  // Ensure presence of 0-terminator.
+  mbs_v[mbs_length - 1] = '\0';
+#endif  // _MSC_VER >= 1400
+
+  *mbs = &mbs_v[0];
   return true;
 }