changed line program entries to contain commands and arguments, AND state. this allows more faithful decoding in readelf
diff --git a/elftools/dwarf/lineprogram.py b/elftools/dwarf/lineprogram.py
index 2dc4b3f..5db7204 100644
--- a/elftools/dwarf/lineprogram.py
+++ b/elftools/dwarf/lineprogram.py
@@ -7,11 +7,38 @@
# This code is in the public domain
#-------------------------------------------------------------------------------
import copy
+from collections import namedtuple
from ..common.utils import struct_parse
from .constants import *
+# LineProgramEntry - an entry in the line program.
+# A line program is a sequence of encoded entries. Some of these entries add a
+# new LineState (mapping between line and address), and some don't.
+#
+# command:
+# The command/opcode - always numeric. For standard commands - it's the opcode
+# that can be matched with one of the DW_LNS_* constants. For extended commands
+# it's the extended opcode that can be matched with one of the DW_LNE_*
+# constants. For special commands, it's the opcode itself.
+#
+# args:
+# A list of decoded arguments of the command.
+#
+# is_extended:
+# Since extended commands are encoded by a zero followed by an extended
+# opcode, and these extended opcodes overlap with other opcodes, this
+# flag is needed to mark that the command has an extended opcode.
+#
+# state:
+# For commands that add a new state, it's the relevant LineState object.
+# For commands that don't add a new state, it's None.
+#
+LineProgramEntry = namedtuple(
+ 'LineProgramEntry', 'command is_extended args state')
+
+
class LineState(object):
""" Represents a line program state (or a "row" in the matrix
describing debug location information for addresses).
@@ -70,15 +97,15 @@
self.structs = structs
self.program_start_offset = program_start_offset
self.program_end_offset = program_end_offset
+ self._decoded_entries = None
- self._line_table = None
-
- def get_line_table(self):
- """ Get the decoded line table for this line program
+ def get_entries(self):
+ """ Get the decoded entries for this line program. Return a list of
+ LineProgramEntry objects.
"""
- if self._line_table is None:
- self._line_table = self._decode_line_program()
- return self._line_table
+ if self._decoded_entries is None:
+ self._decoded_entries = self._decode_line_program()
+ return self._decoded_entries
#------ PRIVATE ------#
@@ -88,18 +115,22 @@
return self.header[name]
def _decode_line_program(self):
- linetable = []
+ entries = []
state = LineState(self.header['default_is_stmt'])
- def add_state_to_table():
- # Used by instructions that have to add the current state to the
- # line table. After adding, some state registers have to be
- # cleared.
- linetable.append(copy.copy(state))
+ def add_entry_new_state(cmd, args, is_extended=False):
+ # Add an entry that sets a new state.
+ # After adding, clear some state registers.
+ entries.append(LineProgramEntry(
+ cmd, is_extended, args, copy.copy(state)))
state.basic_block = False
state.prologue_end = False
state.epilogue_begin = False
+ def add_entry_old_state(cmd, args, is_extended=False):
+ # Add an entry that doesn't visibly set a new state
+ entries.append(LineProgramEntry(cmd, is_extended, args, None))
+
offset = self.program_start_offset
while offset < self.program_end_offset:
opcode = struct_parse(
@@ -115,11 +146,13 @@
if opcode >= self.header['opcode_base']:
# Special opcode (follow the recipe in 6.2.5.1)
adjusted_opcode = opcode - self['opcode_base']
- state.address += ((adjusted_opcode / self['line_range']) *
+ address_addend = ((adjusted_opcode / self['line_range']) *
self['minimum_instruction_length'])
- state.line += (self['line_base'] +
- adjusted_opcode % self['line_range'])
- add_state_to_table()
+ state.address += address_addend
+ line_addend = (self['line_base'] +
+ adjusted_opcode % self['line_range'])
+ state.line += line_addend
+ add_entry_new_state(opcode, [line_addend, address_addend])
elif opcode == 0:
# Extended opcode: start with a zero byte, followed by
# instruction size and the instruction itself.
@@ -130,26 +163,30 @@
if ex_opcode == DW_LNE_end_sequence:
state.end_sequence = True
- add_state_to_table()
+ add_entry_new_state(ex_opcode, [], is_extended=True)
# reset state
state = LineState(self.header['default_is_stmt'])
elif ex_opcode == DW_LNE_set_address:
operand = struct_parse(self.structs.Dwarf_target_addr(''),
self.stream)
state.address = operand
+ add_entry_old_state(ex_opcode, [operand], is_extended=True)
elif ex_opcode == DW_LNE_define_file:
operand = struct_parse(
self.structs.Dwarf_lineprog_file_entry, self.stream)
self['file_entry'].append(operand)
+ add_entry_old_state(ex_opcode, [operand], is_extended=True)
else: # 0 < opcode < opcode_base
# Standard opcode
if opcode == DW_LNS_copy:
- add_state_to_table()
+ add_entry_new_state(opcode, [])
elif opcode == DW_LNS_advance_pc:
operand = struct_parse(self.structs.Dwarf_uleb128(''),
self.stream)
- state.address += (
+ address_addend = (
operand * self.header['minimum_instruction_length'])
+ state.address += address_addend
+ add_entry_old_state(opcode, [address_addend])
elif opcode == DW_LNS_advance_line:
operand = struct_parse(self.structs.Dwarf_sleb128(''),
self.stream)
@@ -158,33 +195,43 @@
operand = struct_parse(self.structs.Dwarf_sleb128(''),
self.stream)
state.file = operand
+ add_entry_old_state(opcode, [operand])
elif opcode == DW_LNS_set_column:
operand = struct_parse(self.structs.Dwarf_uleb128(''),
self.stream)
state.column = operand
+ add_entry_old_state(opcode, [operand])
elif opcode == DW_LNS_negate_stmt:
state.is_stmt = not state.is_stmt
+ add_entry_old_state(opcode, [operand])
elif opcode == DW_LNS_set_basic_block:
state.basic_block = True
+ add_entry_old_state(opcode, [operand])
elif opcode == DW_LNS_const_add_pc:
adjusted_opcode = 255 - self['opcode_base']
- state.address += ((adjusted_opcode / self['line_range']) *
+ address_addend = ((adjusted_opcode / self['line_range']) *
self['minimum_instruction_length'])
+ state.address += address_addend
+ add_entry_old_state(opcode, [address_addend])
elif opcode == DW_LNS_fixed_advance_pc:
operand = struct_parse(self.structs.Dwarf_uint16(''),
self.stream)
state.address += operand
+ add_entry_old_state(opcode, [operand])
elif opcode == DW_LNS_set_prologue_end:
state.prologue_end = True
+ add_entry_old_state(opcode, [])
elif opcode == DW_LNS_set_epilogue_begin:
state.epilogue_begin = True
+ add_entry_old_state(opcode, [])
elif opcode == DW_LNS_set_isa:
operand = struct_parse(self.structs.Dwarf_uleb128(''),
self.stream)
state.isa = operand
+ add_entry_old_state(opcode, [operand])
else:
dwarf_assert(False, 'Invalid standard line program opcode: %s' % (
opcode,))
offset = self.stream.tell()
- return linetable
+ return entries
diff --git a/scripts/readelf.py b/scripts/readelf.py
index e95dac3..c11a197 100755
--- a/scripts/readelf.py
+++ b/scripts/readelf.py
@@ -36,6 +36,8 @@
)
from elftools.dwarf.dwarfinfo import DWARFInfo
from elftools.dwarf.descriptions import describe_attr_value
+from elftools.dwarf.constants import (
+ DW_LNS_copy, DW_LNS_set_file, DW_LNE_define_file)
class ReadElf(object):
@@ -544,7 +546,7 @@
""" Dump the (decoded) line programs from .debug_line
The programs are dumped in the order of the CUs they belong to.
"""
- self._emitline('Decoded dump of debug contents of section .debug_line:')
+ self._emitline('Decoded dump of debug contents of section .debug_line:\n')
for cu in self._dwarfinfo.iter_CUs():
lineprogram = self._dwarfinfo.line_program_for_CU(cu)
@@ -560,16 +562,38 @@
self._emitline('CU: %s:' % cu_filename)
self._emitline('File name Line number Starting address')
- # readelf doesn't print the state after end_sequence instructions.
- # I think it's a bug but to be compatible I don't print them too.
- for state in lineprogram.get_line_table():
- if not state.end_sequence:
+ # Print each state's file, line and address information. For some
+ # instructions other output is needed to be compatible with
+ # readelf.
+ for entry in lineprogram.get_entries():
+ state = entry.state
+ if state is None:
+ # Special handling for commands that don't set a new state
+ if entry.command == DW_LNS_set_file:
+ file_entry = lineprogram['file_entry'][entry.args[0] - 1]
+ if file_entry.dir_index == 0:
+ # current directory
+ self._emitline('\n./%s:[++]' % (
+ file_entry.name))
+ else:
+ self._emitline('\n%s/%s:' % (
+ lineprogram['include_directory'][file_entry.dir_index - 1],
+ file_entry.name))
+ elif entry.command == DW_LNE_define_file:
+ self._emitline('%s:' % (
+ lineprogram['include_directory'][entry.args[0].dir_index]))
+ elif not state.end_sequence:
+ # readelf doesn't print the state after end_sequence
+ # instructions. I think it's a bug but to be compatible
+ # I don't print them too.
self._emitline('%-35s %11d %18s' % (
lineprogram['file_entry'][state.file - 1].name,
state.line,
'0' if state.address == 0 else
self._format_hex(state.address)))
- self._emitline()
+ if entry.command == DW_LNS_copy:
+ # Another readelf oddity...
+ self._emitline()
def _emit(self, s=''):
""" Emit an object to output
diff --git a/tests/test_dwarf_lineprogram.py b/tests/test_dwarf_lineprogram.py
index 50d8af3..56d96c3 100644
--- a/tests/test_dwarf_lineprogram.py
+++ b/tests/test_dwarf_lineprogram.py
@@ -2,8 +2,9 @@
from cStringIO import StringIO
sys.path.extend(['.', '..'])
-from elftools.dwarf.lineprogram import LineProgram, LineState
+from elftools.dwarf.lineprogram import LineProgram, LineState, LineProgramEntry
from elftools.dwarf.structs import DWARFStructs
+from elftools.dwarf.constants import *
class TestLineProgram(unittest.TestCase):
@@ -50,13 +51,21 @@
'\x00\x01\x01')
lp = self._make_program_in_stream(s)
- linetable = lp.get_line_table()
+ linetable = lp.get_entries()
- self.assertLineState(linetable[0], address=0x239, line=3)
- self.assertLineState(linetable[1], address=0x23c, line=5)
- self.assertLineState(linetable[2], address=0x244, line=6)
- self.assertLineState(linetable[3], address=0x24b, line=7, end_sequence=False)
- self.assertLineState(linetable[4], address=0x24d, line=7, end_sequence=True)
+ self.assertEqual(len(linetable), 7)
+ self.assertIs(linetable[0].state, None) # doesn't modify state
+ self.assertEqual(linetable[0].command, DW_LNS_advance_pc)
+ self.assertEqual(linetable[0].args, [0x239])
+ self.assertLineState(linetable[1].state, address=0x239, line=3)
+ self.assertEqual(linetable[1].command, 0xb)
+ self.assertEqual(linetable[1].args, [2, 0])
+ self.assertLineState(linetable[2].state, address=0x23c, line=5)
+ self.assertLineState(linetable[3].state, address=0x244, line=6)
+ self.assertLineState(linetable[4].state, address=0x24b, line=7, end_sequence=False)
+ self.assertEqual(linetable[5].command, DW_LNS_advance_pc)
+ self.assertEqual(linetable[5].args, [2])
+ self.assertLineState(linetable[6].state, address=0x24d, line=7, end_sequence=True)
def test_spec_sample_60(self):
# Sample in figure 60 of DWARFv3
@@ -74,13 +83,17 @@
'\x00\x01\x01')
lp = self._make_program_in_stream(s)
- linetable = lp.get_line_table()
+ linetable = lp.get_entries()
- self.assertLineState(linetable[0], address=0x239, line=3)
- self.assertLineState(linetable[1], address=0x23c, line=5)
- self.assertLineState(linetable[2], address=0x244, line=6)
- self.assertLineState(linetable[3], address=0x24b, line=7, end_sequence=False)
- self.assertLineState(linetable[4], address=0x24d, line=7, end_sequence=True)
+ self.assertEqual(len(linetable), 10)
+ self.assertIs(linetable[0].state, None) # doesn't modify state
+ self.assertEqual(linetable[0].command, DW_LNS_fixed_advance_pc)
+ self.assertEqual(linetable[0].args, [0x239])
+ self.assertLineState(linetable[1].state, address=0x239, line=3)
+ self.assertLineState(linetable[3].state, address=0x23c, line=5)
+ self.assertLineState(linetable[5].state, address=0x244, line=6)
+ self.assertLineState(linetable[7].state, address=0x24b, line=7, end_sequence=False)
+ self.assertLineState(linetable[9].state, address=0x24d, line=7, end_sequence=True)
if __name__ == '__main__':
