llvm.mc.Disassembler.decode now yields the byte sequence which generated the instruction

added custom method to extra.h to return a PyBytes object
from the bytes read from MemoryObject.readBytes.

also moved the generic methods from StringRefMemoryObject
up to MemoryObject

llvm/mc/__init__.py

@@ -126,9 +126,11 @@ class Disassembler(object):
     def decode(self, bs, base_addr):
         '''
         decodes some the bytes in @bs into instructions and yields
-        each instructionas it is decoded. @base_addr is the base address
+        each instruction as it is decoded. @base_addr is the base address
         where the instruction bytes are from (not an offset into
-        @bs)
+        @bs). yields instructions in the form of (addr, data, inst) where
+        addr is an integer, data is a bytearray and inst is an instance of
+        llvm.mc.Instr
         '''
 
         if not isinstance(bs, bytes):
@@ -146,15 +148,27 @@ class Disassembler(object):
             addr = code.getBase() + idx
             status, size = self.mdasm.getInstruction(inst, code, addr)
 
-            if status == MCDisassembler.DecodeStatus.Fail:
-                yield (addr, None)
-            elif status == MCDisassembler.DecodeStatus.SoftFail:
-                yield (addr, self.bad_instr(inst))
-            else:
-                yield (addr, self.instr(inst))
-
             if size < 1:
-               idx += (align - (idx % align))
-            else:
-               idx += size
+                size = (align - (idx % align))
 
+            amt_left = code.getExtent() - idx
+            if amt_left >= size:
+                data = code.readBytes(addr, size)
+            elif amt_left < 1:
+                break 
+            else:
+                data = code.readBytes(addr, amt_left)
+
+            if sys.version_info.major < 3:
+                data = bytearray(map(lambda b: ord(b), data))
+            else:
+                data = bytearray(data)
+
+            if status == MCDisassembler.DecodeStatus.Fail:
+                yield (addr, data, None)
+            elif status == MCDisassembler.DecodeStatus.SoftFail:
+                yield (addr, data, self.bad_instr(inst))
+            else:
+                yield (addr, data, self.instr(inst))
+
+            idx += size

llvmpy/include/llvm_binding/extra.h

@@ -955,6 +955,36 @@ PyObject* TargetRegistry_targets_list()
                   "llvm::Target", "llvm::Target");
 }
 
+static
+PyObject* MemoryObject_readBytes(const llvm::MemoryObject *mobj,
+                                 uint64_t addr,
+                                 uint64_t size
+                                 )
+{
+    int status;
+    uint8_t *bytes;
+    PyObject* po;
+
+    if(size < 1)
+        goto fail;
+
+    bytes = new uint8_t[size];
+    if(bytes == NULL)
+        goto fail;
+
+    status = mobj->readBytes(addr, size, bytes);
+    if(status != 0) {
+        delete bytes;
+        goto fail;
+    }
+
+    po = PyBytes_FromStringAndSize((const char *) bytes, size);
+    delete bytes;
+
+    return po;
+fail:
+    Py_RETURN_NONE;
+}
 
 #if LLVM_VERSION_MAJOR >= 3 and LLVM_VERSION_MINOR >= 4
 static
@@ -972,6 +1002,7 @@ PyObject* MCDisassembler_getInstruction(llvm::MCDisassembler *disasm,
                                     llvm::nulls(), llvm::nulls());
     return Py_BuildValue("(i,i)", int(status), size);
 }
+
 #endif /* llvm >= 3.4 */
 
 static

llvmpy/src/Support/StringRefMemoryObject.py

@@ -8,7 +8,23 @@ if LLVM_VERSION >= (3, 4):
 
 @MemoryObject
 class MemoryObject:
-    pass
+    _include_ = "llvm/Support/MemoryObject.h"
+
+    getBase = Method(cast(Uint64, int))
+    getExtent = Method(cast(Uint64, int))
+
+    readBytes = CustomMethod('MemoryObject_readBytes',
+                             PyObjectPtr,
+                             cast(int, Uint64), #address
+                             cast(int, Uint64)  #size 
+                             )
+    @CustomPythonMethod
+    def readAll(self):
+        result = self.readBytes(self.getBase(), self.getExtent())
+        if not result:
+            raise Exception("expected readBytes to be successful!")
+        return result
+
 
 if LLVM_VERSION >= (3, 4):
     @StringRefMemoryObject
@@ -16,6 +32,3 @@ if LLVM_VERSION >= (3, 4):
         _include_ = "llvm/Support/StringRefMemoryObject.h"
 
         new = Constructor(cast(bytes, StringRef), cast(int, Uint64))
-
-        getBase = Method(cast(Uint64, int))
-        getExtent = Method(cast(Uint64, int))

test/example-disassemble.py

@@ -10,7 +10,8 @@ if llvm.version >= (3, 4):
 
     def print_instructions(dasm, bs):
         print("print instructions")
-        for (addr, inst) in dasm.decode(bs, 0x4000):
+        for (addr, data, inst) in dasm.decode(bs, 0x4000):
+
             if inst is None:
                 print("\t0x%x => (bad)" % (addr))
             else:
@@ -22,7 +23,7 @@ if llvm.version >= (3, 4):
 
                 print("\t\tflags = 0x%x, tsflags = 0x%x" % (inst.flags, inst.ts_flags))
                 for line in str(inst).split("\n"):
-                    print("\t%s" % (line))
+                    print("\t\t%-24s %s" % ("".join(map(lambda b: "%02x" % b, data))+":", line.strip()))
 
 
     x86 = TargetMachine.x86()

test/testall.py

@@ -659,7 +659,7 @@ def do_llvm_mc():
     tm = TargetMachine.x86()
     dasm = mc.Disassembler(tm)
 
-    for (offset, instr) in dasm.decode("c3", 0):
+    for (offset, data, instr) in dasm.decode("c3", 0):
         pass
 
 def main():