Assorted bugfixes in object file handling:

- Adds support for sniffing PE/COFF files on win32 (.exe and .dll)
   which are COFF files that have an MS-DOS compatibility stub on
   the front of them.

 - Fixes a bug in the COFFObjectFile's support for the Microsoft COFF
   extension for long symbol names, wherein it was attempting to parse
   the leading '/' in an extended symbol name reference as part of the
   integer offset.

 - Fixes bugs in COFFObjectFile and ELFObjectFile wherein section
   and symbol iterators were being returned with uninitialized bytes;
   the type DataRefImpl is a union between 2 32-bit words (d.a and d.b)
   and a single intptr_t word (p). Only p was being initialized, so in
   32-bit builds the result would be iterators with random upper 32-bit
   words in their DataRefImpls. This caused random failures when
   seeking around in object files.

Patch by Graydon Hoare!

llvm-svn: 128799

llvm/lib/Object/COFFObjectFile.cpp

@@ -91,6 +91,7 @@ extern char coff_coff_section_layout_static_assert
 namespace {
 class COFFObjectFile : public ObjectFile {
 private:
+        uint64_t         HeaderOff;
   const coff_file_header *Header;
   const coff_section     *SectionTable;
   const coff_symbol      *SymbolTable;
@@ -256,7 +257,7 @@ StringRef COFFObjectFile::getSectionName(DataRefImpl Sec) const {
   // Check for string table entry. First byte is '/'.
   if (name[0] == '/') {
     uint32_t Offset;
-    name.getAsInteger(10, Offset);
+    name.substr(1).getAsInteger(10, Offset);
     return StringRef(getString(Offset));
   }
 
@@ -287,9 +288,20 @@ bool COFFObjectFile::isSectionText(DataRefImpl Sec) const {
 
 COFFObjectFile::COFFObjectFile(MemoryBuffer *Object)
   : ObjectFile(Object) {
-  Header = reinterpret_cast<const coff_file_header *>(base);
+
+  HeaderOff = 0;
+
+  if (base[0] == 0x4d && base[1] == 0x5a) {
+    // PE/COFF, seek through MS-DOS compatibility stub and 4-byte
+    // PE signature to find 'normal' COFF header.
+    HeaderOff += *reinterpret_cast<const ulittle32_t *>(base + 0x3c);
+    HeaderOff += 4;
+  }
+
+  Header = reinterpret_cast<const coff_file_header *>(base + HeaderOff);
   SectionTable =
     reinterpret_cast<const coff_section *>( base
+                                          + HeaderOff
                                           + sizeof(coff_file_header)
                                           + Header->SizeOfOptionalHeader);
   SymbolTable =
@@ -303,6 +315,7 @@ COFFObjectFile::COFFObjectFile(MemoryBuffer *Object)
 
 ObjectFile::symbol_iterator COFFObjectFile::begin_symbols() const {
   DataRefImpl ret;
+  memset(&ret, 0, sizeof(DataRefImpl));
   ret.p = reinterpret_cast<intptr_t>(SymbolTable);
   return symbol_iterator(SymbolRef(ret, this));
 }
@@ -310,18 +323,21 @@ ObjectFile::symbol_iterator COFFObjectFile::begin_symbols() const {
 ObjectFile::symbol_iterator COFFObjectFile::end_symbols() const {
   // The symbol table ends where the string table begins.
   DataRefImpl ret;
+  memset(&ret, 0, sizeof(DataRefImpl));
   ret.p = reinterpret_cast<intptr_t>(StringTable);
   return symbol_iterator(SymbolRef(ret, this));
 }
 
 ObjectFile::section_iterator COFFObjectFile::begin_sections() const {
   DataRefImpl ret;
+  memset(&ret, 0, sizeof(DataRefImpl));
   ret.p = reinterpret_cast<intptr_t>(SectionTable);
   return section_iterator(SectionRef(ret, this));
 }
 
 ObjectFile::section_iterator COFFObjectFile::end_sections() const {
   DataRefImpl ret;
+  memset(&ret, 0, sizeof(DataRefImpl));
   ret.p = reinterpret_cast<intptr_t>(SectionTable + Header->NumberOfSections);
   return section_iterator(SectionRef(ret, this));
 }

llvm/lib/Object/ELFObjectFile.cpp

@@ -547,6 +547,7 @@ template<support::endianness target_endianness, bool is64Bits>
 ObjectFile::section_iterator ELFObjectFile<target_endianness, is64Bits>
                                           ::begin_sections() const {
   DataRefImpl ret;
+  memset(&ret, 0, sizeof(DataRefImpl));
   ret.p = reinterpret_cast<intptr_t>(base + Header->e_shoff);
   return section_iterator(SectionRef(ret, this));
 }
@@ -555,6 +556,7 @@ template<support::endianness target_endianness, bool is64Bits>
 ObjectFile::section_iterator ELFObjectFile<target_endianness, is64Bits>
                                           ::end_sections() const {
   DataRefImpl ret;
+  memset(&ret, 0, sizeof(DataRefImpl));
   ret.p = reinterpret_cast<intptr_t>(base
                                      + Header->e_shoff
                                      + (Header->e_shentsize * Header->e_shnum));

llvm/lib/Support/Path.cpp

@@ -15,11 +15,15 @@
 #include "llvm/Support/FileSystem.h"
 #include "llvm/Config/config.h"
 #include "llvm/Support/FileSystem.h"
+#include "llvm/Support/Endian.h"
 #include <cassert>
 #include <cstring>
 #include <ostream>
 using namespace llvm;
 using namespace sys;
+namespace {
+using support::ulittle32_t;
+}
 
 //===----------------------------------------------------------------------===//
 //=== WARNING: Implementation here must contain only TRULY operating system
@@ -129,6 +133,16 @@ sys::IdentifyFileType(const char *magic, unsigned length) {
       if (magic[1] == 0x02)
         return COFF_FileType;
       break;
+
+    case 0x4d: // Possible MS-DOS stub on Windows PE file
+      if (magic[1] == 0x5a) {
+        uint32_t off = *reinterpret_cast<const ulittle32_t *>(magic + 0x3c);
+        // PE/COFF file, either EXE or DLL.
+        if (off < length && memcmp(magic + off, "PE\0\0",4) == 0)
+          return COFF_FileType;
+      }
+      break;
+
     case 0x64: // x86-64 Windows.
       if (magic[1] == char(0x86))
         return COFF_FileType;