Refactor rel32 searching process for x64 to make it more similar to x86.

courgette/disassembler_win32_x64.cc

Index: courgette/disassembler_win32_x64.cc
diff --git a/courgette/disassembler_win32_x64.cc b/courgette/disassembler_win32_x64.cc
index c4d89e452e62875fe402af773f60110710691236..d4ed28cd24b8080c311214ab5cffebe7c16365bf 100644
--- a/courgette/disassembler_win32_x64.cc
+++ b/courgette/disassembler_win32_x64.cc
@@ -10,6 +10,7 @@
 #include "base/numerics/safe_conversions.h"
 #include "courgette/assembly_program.h"
 #include "courgette/courgette.h"
+#include "courgette/rel32_finder_x64.h"
 
 #if COURGETTE_HISTOGRAM_TARGETS
 #include <iostream>
@@ -43,7 +44,6 @@ void DisassemblerWin32X64::ParseRel32RelocsFromSection(const Section* section) {
 
   FileOffset start_file_offset = section->file_offset_of_raw_data;
   FileOffset end_file_offset = start_file_offset + section->size_of_raw_data;
-  RVA relocs_start_rva = base_relocation_table().address_;
 
   const uint8_t* start_pointer = FileOffsetToPointer(start_file_offset);
   const uint8_t* end_pointer = FileOffsetToPointer(end_file_offset);
@@ -51,92 +51,17 @@ void DisassemblerWin32X64::ParseRel32RelocsFromSection(const Section* section) {
   RVA start_rva = FileOffsetToRVA(start_file_offset);
   RVA end_rva = start_rva + section->virtual_size;
 
-  // Quick way to convert from Pointer to RVA within a single Section is to
-  // subtract |pointer_to_rva|.
-  const uint8_t* const adjust_pointer_to_rva = start_pointer - start_rva;
-
-  std::vector<RVA>::iterator abs32_pos = abs32_locations_.begin();
-
-  // Find the rel32 relocations.
-  const uint8_t* p = start_pointer;
-  while (p < end_pointer) {
-    RVA current_rva = static_cast<RVA>(p - adjust_pointer_to_rva);
-    if (current_rva == relocs_start_rva) {
-      uint32_t relocs_size = base_relocation_table().size_;
-      if (relocs_size) {
-        p += relocs_size;
-        continue;
-      }
-    }
-
-    // Heuristic discovery of rel32 locations in instruction stream: are the
-    // next few bytes the start of an instruction containing a rel32
-    // addressing mode?
-    const uint8_t* rel32 = nullptr;
-    bool is_rip_relative = false;
-
-    if (p + 5 <= end_pointer) {
-      if (*p == 0xE8 || *p == 0xE9)  // jmp rel32 and call rel32
-        rel32 = p + 1;
-    }
-    if (p + 6 <= end_pointer) {
-      if (*p == 0x0F && (*(p + 1) & 0xF0) == 0x80) {  // Jcc long form
-        if (p[1] != 0x8A && p[1] != 0x8B)  // JPE/JPO unlikely
-          rel32 = p + 2;
-      } else if (*p == 0xFF && (*(p + 1) == 0x15 || *(p + 1) == 0x25)) {
-        // rip relative call/jmp
-        rel32 = p + 2;
-        is_rip_relative = true;
-      }
-    }
-    if (p + 7 <= end_pointer) {
-      if ((*p & 0xFB) == 0x48 && *(p + 1) == 0x8D &&
-          (*(p + 2) & 0xC7) == 0x05) {
-        // rip relative lea
-        rel32 = p + 3;
-        is_rip_relative = true;
-      } else if ((*p & 0xFB) == 0x48 && *(p + 1) == 0x8B &&
-                 (*(p + 2) & 0xC7) == 0x05) {
-        // rip relative mov
-        rel32 = p + 3;
-        is_rip_relative = true;
-      }
-    }
-
-    if (rel32) {
-      RVA rel32_rva = static_cast<RVA>(rel32 - adjust_pointer_to_rva);
-
-      // Is there an abs32 reloc overlapping the candidate?
-      while (abs32_pos != abs32_locations_.end() && *abs32_pos < rel32_rva - 3)
-        ++abs32_pos;
-      // Now: (*abs32_pos > rel32_rva - 4) i.e. the lowest addressed 4-byte
-      // region that could overlap rel32_rva.
-      if (abs32_pos != abs32_locations_.end()) {
-        if (*abs32_pos < rel32_rva + 4) {
-          // Beginning of abs32 reloc is before end of rel32 reloc so they
-          // overlap. Skip four bytes past the abs32 reloc.
-          p += (*abs32_pos + 4) - current_rva;
-          continue;
-        }
-      }
-
-      // + 4 since offset is relative to start of next instruction.
-      RVA target_rva = rel32_rva + 4 + Read32LittleEndian(rel32);
-      // To be valid, rel32 target must be within image, and within this
-      // section.
-      if (target_rva < size_of_image_ &&  // Subsumes rva != kUnassignedRVA.
-          (is_rip_relative ||
-           (start_rva <= target_rva && target_rva < end_rva))) {
-        rel32_locations_.push_back(rel32_rva);
+  Rel32FinderX64 finder(
+      base_relocation_table().address_,
+      base_relocation_table().address_ + base_relocation_table().size_,
+      size_of_image_);
+  finder.Find(start_pointer, end_pointer, start_rva, end_rva, abs32_locations_);
+  finder.SwapRel32Locations(&rel32_locations_);
+
 #if COURGETTE_HISTOGRAM_TARGETS
-        ++rel32_target_rvas_[target_rva];
+  DCHECK(rel32_target_rvas_.empty());
+  finder.SwapRel32TargetRVAs(&rel32_target_rvas_);
 #endif
-        p = rel32 + 4;
-        continue;
-      }
-    }
-    p += 1;
-  }
 }
 
 }  // namespace courgette