[Courgette] Using LabelManager to reduce Courgette-apply peak RAM by 25%.

courgette/disassembler_win32_x86.cc

Index: courgette/disassembler_win32_x86.cc
diff --git a/courgette/disassembler_win32_x86.cc b/courgette/disassembler_win32_x86.cc
index 974e864033d433bd5443e5ccd70d55036efe06d0..c24aed5074b36729c08058b9619b0f74aff61652 100644
--- a/courgette/disassembler_win32_x86.cc
+++ b/courgette/disassembler_win32_x86.cc
@@ -252,6 +252,9 @@ bool DisassemblerWin32X86::Disassemble(AssemblyProgram* target) {
 
   ParseRel32RelocsFromSections();
 
+  PrecomputeLabels(target);
+  RemoveUnusedRel32Locations(target);
+
   if (!ParseFile(target))
     return false;
 
@@ -362,6 +365,25 @@ std::string DisassemblerWin32X86::SectionName(const Section* section) {
   return name;
 }
 
+RvaVisitor* DisassemblerWin32X86::CreateAbs32TargetRvaVisitor() {
+  return new RvaVisitor_Abs32(abs32_locations_, *this);
+}
+
+RvaVisitor* DisassemblerWin32X86::CreateRel32TargetRvaVisitor() {
+  return new RvaVisitor_Rel32(rel32_locations_, *this);
+}
+
+void DisassemblerWin32X86::RemoveUnusedRel32Locations(

[Will Harris, 2016/05/04 18:02:26]

hmm I wonder if a future refactor could introduce DisassemblerWin32 to sit
between DisassemblerWin32X86/DisassemblerWin32X64 and Disassembler since parts
of this code is now looking very similar between the two.

[huangs, 2016/05/04 19:33:59]

Yup!  That's planned in go/courgette-ideas-2016 under "Deduping x86 and x64
Parsing Code".  

+    AssemblyProgram* program) {
+  auto cond = [this, program](RVA rva) -> bool {
+    RVA target_rva = rva + 4 + Read32LittleEndian(RVAToPointer(rva));

[Will Harris, 2016/05/04 18:02:26]

can you explain this +4?

[huangs, 2016/05/04 19:33:59]

Added comment. Also updating rel32_finder_win32_x86.cc for consistency.

+    return program->FindRel32Label(target_rva) == nullptr;
+  };
+  rel32_locations_.erase(
+      std::remove_if(rel32_locations_.begin(), rel32_locations_.end(), cond),
+      rel32_locations_.end());
+}
+
 CheckBool DisassemblerWin32X86::ParseFile(AssemblyProgram* program) {
   // Walk all the bytes in the file, whether or not in a section.
   FileOffset file_offset = 0;
@@ -543,7 +565,9 @@ CheckBool DisassemblerWin32X86::ParseFileRegion(const Section* section,
       DCHECK_NE(kNoRVA, target_rva);
       // TODO(sra): target could be Label+offset.  It is not clear how to guess
       // which it might be.  We assume offset==0.
-      if (!program->EmitAbs32(program->FindOrMakeAbs32Label(target_rva)))
+      Label* label = program->FindAbs32Label(target_rva);
+      DCHECK(label);
+      if (!program->EmitAbs32(label))
         return false;
       p += 4;
       continue;
@@ -554,7 +578,9 @@ CheckBool DisassemblerWin32X86::ParseFileRegion(const Section* section,
 
     if (rel32_pos != rel32_locations_.end() && *rel32_pos == current_rva) {
       RVA target_rva = current_rva + 4 + Read32LittleEndian(p);
-      if (!program->EmitRel32(program->FindOrMakeRel32Label(target_rva)))
+      Label* label = program->FindRel32Label(target_rva);
+      DCHECK(label);
+      if (!program->EmitRel32(label))
         return false;
       p += 4;
       continue;