basic DWARF4 support

src/bintools/dwarf/enums.py

@@ -150,8 +150,13 @@
     0x66: 'elemental',
     0x67: 'pure',
     0x68: 'recursive',
-
+    # DWARF 4 values
+    0x69: 'signature',
+    0x6a: 'main_subprogram',
+    0x6b: 'data_bit_offset',
+    0x6c: 'const_expr',
     0x6d: 'enum_class',
+    0x6e: 'linkage_name',
 
     0x2000: 'lo_user',
     0x2007: 'MIPS_linkage_name',
@@ -416,16 +421,17 @@ class DW_OP(object):
     stack_value = 0x9f
     lo_user = 0xe0
 
+    # https://fedorahosted.org/elfutils/wiki/DwarfExtensions
     GNU_uninit = 0xf0
     GNU_encoded_addr = 0xf1
     GNU_implicit_pointer = 0xf2
-    GNU_entry_value = 0xf3
-    GNU_const_type = 0xf4
-    GNU_regval_type = 0xf5
-    GNU_deref_type = 0xf6
-    GNU_convert = 0xf7
+    GNU_entry_value = 0xf3  # http://www.dwarfstd.org/ShowIssue.php?issue=100909.1
+    GNU_const_type = 0xf4 # http://www.dwarfstd.org/doc/040408.1.html
+    GNU_regval_type = 0xf5 # http://www.dwarfstd.org/doc/040408.1.html
+    GNU_deref_type = 0xf6 # http://www.dwarfstd.org/doc/040408.1.html
+    GNU_convert = 0xf7 # http://www.dwarfstd.org/doc/040408.1.html
     GNU_reinterpret = 0xf9
-    GNU_parameter_ref = 0xfa
+    GNU_parameter_ref = 0xfa # https://gcc.gnu.org/ml/gcc-patches/2011-06/msg00649.html
     GNU_addr_index = 0xfb
     GNU_const_index = 0xfc
 

src/bintools/dwarf/expressions.py

@@ -42,8 +42,8 @@ def __str__(self):
     DW_OP.const4s: ('sdata4', None),
     DW_OP.const8u: ('data8', None),
     DW_OP.const8s: ('sdata8', None),
-    DW_OP.constu : ('read_udata', None),
-    DW_OP.consts : ('read_sdata', None),
+    DW_OP.constu : ('udata', None),
+    DW_OP.consts : ('sdata', None),
 
     # Register Based Addressing
     DW_OP.fbreg : ('sdata', None),
@@ -89,6 +89,12 @@ def __str__(self):
 
     # GNU
     DW_OP.GNU_implicit_pointer: ('addr', 'sdata'),
+    DW_OP.GNU_entry_value: ('exprloc', None),
+    DW_OP.GNU_const_type: ('udata', 'block1'),
+    DW_OP.GNU_regval_type: ('udata', 'udata'),
+    DW_OP.GNU_deref_type: ('data1', 'udata'),
+    DW_OP.GNU_convert: ('udata', None),
+    DW_OP.GNU_parameter_ref: ('data4', None),
 }
 
 
@@ -106,8 +112,9 @@ def __init__(self, dwarf, length):
 
             opcode = dwarf.u08()
             if opcode not in DW_OP:
-                raise ParseError("Unknown DW_OP code: %d" % opcode)
-
+                previnst = ','.join(str(x) for x in self.instructions)
+                raise ParseError("Unknown DW_OP code: %d (after %s, offset 0x%x)" % (opcode, previnst, dwarf.io.tell()))
+
             operand_1 = operand_2 = None
             if opcode in DW_OP_OPERANDS:
                 type_1, type_2 = DW_OP_OPERANDS[opcode]

src/bintools/dwarf/info.py

@@ -101,7 +101,7 @@ def __init__(self, dwarf, overall_offset):
         length = dwarf.u32()
         stop = dwarf.io.tell() + length
 
-        ver = dwarf.check_version(handled=[2, 3])
+        ver = dwarf.check_version(handled=[2, 3, 4])
 
         abbrev_offset = dwarf.u32()
         self.pointer_size = dwarf.u08()

src/bintools/dwarf/stream.py

@@ -23,14 +23,16 @@ def __init__(self, addr_size=4):
         elif addr_size == 4:
             self.read_addr = self.u32
             self.max_addr = 0xFFFFFFFF
-
         elif addr_size == 8:
             self.read_addr = self.u64
             self.max_addr = 0xFFFFFFFFFFFFFFFF
+
         if self.bits == ELFCLASS.ELFCLASS32:
             self.CIE_ID = 0xFFFFFFFF
+            self.read_sec_offset = self.u32
         elif self.bits == ELFCLASS.ELFCLASS64:
             self.CIE_ID = 0xFFFFFFFFFFFFFFFF
+            self.read_sec_offset = self.u64
 
         # Read methods aliases
         self.read_data1 = self.read_ref1 = self.u08
@@ -104,6 +106,9 @@ def read_strp(self):
     def read_flag(self):
         return (self.u08() != 0)
 
+    def read_flag_present(self): # the attribute is implicitly indicated as present
+        return True
+
     def read_indirect(self):
         return self.read_form(self.ULEB128())
 
@@ -135,6 +140,8 @@ def read_expr_block(self, form):
     def read_expr(self):
         return Expression(self, self.u16())
 
+    def read_exprloc(self):
+        return Expression(self, self.ULEB128())
 
 class SectionLoader(object):
     def __init__(self, dwarf, section_name, Entry):

src/dwarfhelpers.py

@@ -13,7 +13,7 @@ def expect_ref(attr):
     return attr.value
 
 def expect_flag(attr):
-    assert(attr.form in ['flag'])
+    assert(attr.form in ['flag','flag_present'])
     return attr.value
 
 def expect_addr(attr):