Replace libdisasm with capstone

src/processor/disassembler_x86.h

Index: src/processor/disassembler_x86.h
diff --git a/src/processor/disassembler_x86.h b/src/processor/disassembler_x86.h
index 71069410721740b610b65b1a762ae73384249155..09aa359e2e224596f3049dd2eadc6959ed961340 100644
--- a/src/processor/disassembler_x86.h
+++ b/src/processor/disassembler_x86.h
@@ -28,7 +28,7 @@
 
 // disassembler_x86.h: Basic x86 bytecode disassembler
 //
-// Provides a simple disassembler which wraps libdisasm. This allows simple
+// Provides a simple disassembler which wraps capstone. This allows simple
 // tests to be run against bytecode to test for various properties.
 //
 // Author: Cris Neckar
@@ -41,12 +41,12 @@
 
 #include "google_breakpad/common/breakpad_types.h"
 
-namespace libdis {
-#include "third_party/libdisasm/libdis.h"
-}
+#include "capstone.h"
+
 
 namespace google_breakpad {
 
+// Flags for current disassembler state.
 enum {
   DISX86_NONE =                 0x0,
   DISX86_BAD_BRANCH_TARGET =    0x1,
@@ -71,26 +71,35 @@ class DisassemblerX86 {
     // including any registers marked as bad through setBadRead()
     // or setBadWrite(). This method can be called in a loop to
     // disassemble until the end of a region.
-    uint32_t NextInstruction();
+    //
+    // Returns the size of the instruction in bytes, or zero if
+    // disassembly failed.
+    size_t NextInstruction();
 
     // Indicates whether the current disassembled instruction was valid.
     bool currentInstructionValid() { return instr_valid_; }
 
-    // Returns the current instruction as defined in libdis.h,
+    // Returns the current instruction as defined in capstone.h,
     // or NULL if the current instruction is not valid.
-    const libdis::x86_insn_t* currentInstruction() {
-      return instr_valid_ ? &current_instr_ : NULL;
+    const cs_insn* currentInstruction() {
+      return instr_valid_ ? current_instr_ : nullptr;
     }
 
-    // Returns the type of the current instruction as defined in libdis.h.
-    libdis::x86_insn_group currentInstructionGroup() {
-      return current_instr_.group;
+    // Returns true if the current instruction is in group.
+    bool currentInstructionIsGroup(x86_insn_group group) {
+      return instr_valid_ ?
+          cs_insn_group(handle_, current_instr_, group) : false;
     }
 
+    // Returns true if the current instruction is a block data instruction.
+    bool currentInstructionIsBlockData();
+
+
     // Indicates whether a return instruction has been encountered.
     bool endOfBlock() { return end_of_block_; }
 
-    // The flags set so far for the disassembly.
+    // The flags set so far for the disassembly, from the set defined in
+    // the anonymous enum above.
     uint16_t flags() { return flags_; }
 
     // This sets an indicator that the register used to determine
@@ -102,19 +111,24 @@ class DisassemblerX86 {
     bool setBadWrite();
 
   protected:
-    const uint8_t *bytecode_;
-    uint32_t size_;
-    uint32_t virtual_address_;
-    uint32_t current_byte_offset_;
-    uint32_t current_inst_offset_;
+    // Memory containing instructions to disassemble.
+    // Owned by caller.
+    const uint8_t* bytecode_;
+    // Size of remaining bytecode.
+    size_t size_;
+    // Virtual address of current instruction.
+    uint64_t virtual_address_;
+
+    // capstone library handle
+    csh handle_;
 
     bool instr_valid_;
-    libdis::x86_insn_t current_instr_;
+    cs_insn* current_instr_;
 
     // TODO(cdn): Maybe also track an expression's index register.
     // ex: mov eax, [ebx + ecx]; ebx is base, ecx is index.
     bool register_valid_;
-    libdis::x86_reg_t bad_register_;
+    x86_reg bad_register_;
 
     bool pushed_bad_value_;
     bool end_of_block_;