24 bit integer support, forms strx3, addrx3 (#553)

* 24 bit integer support, forms strx3, addrx3

* More DWARFv1 constants

* More DWARFv1 constants

* LLVM lineprog header content type codes

* No global declarations

elftools/common/construct_utils.py

@@ -6,9 +6,10 @@
 # Eli Bendersky ([email protected])
 # This code is in the public domain
 #-------------------------------------------------------------------------------
+from struct import Struct
 from ..construct import (
     Subconstruct, ConstructError, ArrayError, Adapter, Field, RepeatUntil,
-    Rename, SizeofError, Construct
+    Rename, SizeofError, Construct, StaticField
     )
 
 
@@ -110,3 +111,30 @@ def _build(self, obj, stream, context):
         context[self.name] = stream.tell()
     def _sizeof(self, context):
         return 0
+
+_UBInt24_packer = Struct(">BH")
+_ULInt24_packer = Struct("<HB")
+
+class UBInt24(StaticField):
+    """unsigned, big endian 24-bit integer"""
+    def __init__(self, name):
+        StaticField.__init__(self, name, 3)
+
+    def _parse(self, stream, context):
+        (h, l) = _UBInt24_packer.unpack(StaticField._parse(self, stream, context))
+        return l | (h << 16)
+
+    def _build(self, obj, stream, context):
+        StaticField._build(self, _UBInt24_packer.pack(obj >> 16, obj & 0xFFFF), stream, context)
+
+class ULInt24(StaticField):
+    """unsigned, little endian 24-bit integer"""
+    def __init__(self, name):
+        StaticField.__init__(self, name, 3)
+
+    def _parse(self, stream, context):
+        (l, h) = _ULInt24_packer.unpack(StaticField._parse(self, stream, context))
+        return l | (h << 16)
+
+    def _build(self, obj, stream, context):
+        StaticField._build(self, _ULInt24_packer.pack(obj & 0xFFFF, obj >> 16), stream, context)

elftools/dwarf/constants.py

@@ -170,6 +170,8 @@
 DW_LNCT_size = 0x04
 DW_LNCT_MD5 = 0x05
 DW_LNCT_lo_user = 0x2000
+DW_LNCT_LLVM_source = 0x2001
+DW_LNCT_LLVM_is_MD5 = 0x2002
 DW_LNCT_hi_user = 0x3fff
 
 # Call frame instructions

elftools/dwarf/enums.py

@@ -143,6 +143,7 @@
     DW_AT_const_value               = 0x1c,
     DW_AT_containing_type           = 0x1d,
     DW_AT_default_value             = 0x1e,
+    DW_AT_friends                   = 0x1f,
     DW_AT_inline                    = 0x20,
     DW_AT_is_optional               = 0x21,
     DW_AT_lower_bound               = 0x22,
@@ -152,7 +153,9 @@
     DW_AT_protected                 = 0x26,
     DW_AT_prototyped                = 0x27,
     DW_AT_public                    = 0x28,
+    DW_AT_pure_virtual              = 0x29,
     DW_AT_return_addr               = 0x2a,
+    # In DWARFv1, DW_AT_specification was at 0x2b, moved to 0x47 in v2
     DW_AT_start_scope               = 0x2c,
     DW_AT_bit_stride                = 0x2e,
     DW_AT_stride_size               = 0x2e,
@@ -413,6 +416,8 @@
     DW_LNCT_size             = 0x4,
     DW_LNCT_MD5              = 0x5,
     DW_LNCT_lo_user          = 0x2000,
+    DW_LNCT_LLVM_source      = 0x2001,
+    DW_LNCT_LLVM_is_MD5      = 0x2002,
     DW_LNCT_hi_user          = 0x3fff
 )
 

elftools/dwarf/structs.py

@@ -16,7 +16,7 @@
     String, Switch, Value
     )
 from ..common.construct_utils import (RepeatUntilExcluding, ULEB128, SLEB128,
-    StreamOffset)
+    StreamOffset, ULInt24, UBInt24)
 from .enums import *
 
 
@@ -121,6 +121,7 @@ def _create_structs(self):
         if self.little_endian:
             self.Dwarf_uint8 = ULInt8
             self.Dwarf_uint16 = ULInt16
+            self.Dwarf_uint24 = ULInt24
             self.Dwarf_uint32 = ULInt32
             self.Dwarf_uint64 = ULInt64
             self.Dwarf_offset = ULInt32 if self.dwarf_format == 32 else ULInt64
@@ -134,6 +135,7 @@ def _create_structs(self):
         else:
             self.Dwarf_uint8 = UBInt8
             self.Dwarf_uint16 = UBInt16
+            self.Dwarf_uint24 = UBInt24
             self.Dwarf_uint32 = UBInt32
             self.Dwarf_uint64 = UBInt64
             self.Dwarf_offset = UBInt32 if self.dwarf_format == 32 else UBInt64
@@ -253,7 +255,7 @@ def _create_dw_form(self):
             DW_FORM_addrx=self.Dwarf_uleb128(''),
             DW_FORM_addrx1=self.Dwarf_uint8(''),
             DW_FORM_addrx2=self.Dwarf_uint16(''),
-            # DW_FORM_addrx3=self.Dwarf_uint24(''),  # TODO
+            DW_FORM_addrx3=self.Dwarf_uint24(''),
             DW_FORM_addrx4=self.Dwarf_uint32(''),
 
             DW_FORM_block1=self._make_block_struct(self.Dwarf_uint8),
@@ -276,7 +278,7 @@ def _create_dw_form(self):
             DW_FORM_line_strp=self.Dwarf_offset(''),
             DW_FORM_strx1=self.Dwarf_uint8(''),
             DW_FORM_strx2=self.Dwarf_uint16(''),
-            # DW_FORM_strx3=self.Dwarf_uint24(''),  # TODO
+            DW_FORM_strx3=self.Dwarf_uint24(''),
             DW_FORM_strx4=self.Dwarf_uint64(''),
             DW_FORM_flag=self.Dwarf_uint8(''),
 

test/test_int24.py

@@ -0,0 +1,30 @@
+#------------------------------------------------------------------------------
+# elftools tests
+#
+# Seva Alekseyev ([email protected])
+# This code is in the public domain
+#------------------------------------------------------------------------------
+import unittest
+import random
+from io import BytesIO
+
+from elftools.common.construct_utils import ULInt24, UBInt24
+from elftools.common.utils import struct_parse
+
+class TestInt24(unittest.TestCase):
+    def test_main(self):
+        # Testing parsing and building, both LE and BE
+        b = random.randbytes(3)
+
+        n = struct_parse(UBInt24(''), BytesIO(b))
+        self.assertEqual(n, (b[0] << 16) | (b[1] << 8) | b[2])
+        s = UBInt24('').build(n)
+        self.assertEqual(s, b)
+
+        n = struct_parse(ULInt24(''), BytesIO(b))
+        self.assertEqual(n, b[0] | (b[1] << 8) | (b[2] << 16))
+        s = ULInt24('').build(n)
+        self.assertEqual(s, b)
+
+if __name__ == '__main__':
+    unittest.main()