def iterateList(self, mappings):
''' iterate forward, then backward, until null or duplicate '''
done = [0]
obj = self
#print 'going forward '
for fieldname in [forward, backward]:
link = getattr(obj, fieldname)
addr = utils.getaddress(link)
log.debug('iterateList got a <%s>/0x%x' %
(link.__class__.__name__, addr))
nb = 0
while addr not in done:
#print '%x %s'%(addr, addr in done)
done.append(addr)
memoryMap = utils.is_valid_address_value(
addr, mappings, structType)
if memoryMap == False:
raise ValueError(
'the link of this linked list has a bad value')
st = memoryMap.readStruct(addr, structType)
model.keepRef(st, structType, addr)
log.debug("keepRefx2 %s.%s @%x" %
(structType, fieldname, addr))
yield addr
# next
link = getattr(st, fieldname)
addr = utils.getaddress(link)
#print 'going backward after %x'%(addr)
raise StopIteration
def iterateList(self, mappings):
''' iterate forward, then backward, until null or duplicate '''
done = [0]
obj = self
#print 'going forward '
for fieldname in [forward, backward]:
link = getattr(obj, fieldname)
addr = utils.getaddress(link)
log.debug('iterateList got a <%s>/0x%x'%(link.__class__.__name__,addr))
nb=0
while addr not in done:
#print '%x %s'%(addr, addr in done)
done.append(addr)
memoryMap = utils.is_valid_address_value( addr, mappings, structType)
if memoryMap == False:
raise ValueError('the link of this linked list has a bad value')
st = memoryMap.readStruct( addr, structType)
model.keepRef(st, structType, addr)
log.debug("keepRefx2 %s.%s @%x"%(structType, fieldname, addr ))
yield addr
# next
link = getattr(st, fieldname)
addr = utils.getaddress(link)
#print 'going backward after %x'%(addr)
raise StopIteration
def test_keepRef(self):
self.assertNotEqual( self.mappings, None )
for fname, ftype in self.heap_obj.getFields():
attr = getattr(self.heap_obj, fname)
if isCStringPointer(ftype):
# ignore that - attr_addr = getaddress(attr.ptr)
continue
elif isPointerType(ftype):
attr_addr = getaddress(attr)
else:
continue
if attr_addr == 0:
continue
self.assertTrue( utils.is_valid_address_value(attr_addr, self.mappings), '%s: 0x%x is not valid'%(fname, attr_addr))
# the book should register struct type, not pointer to strut type
attr_type = model.get_subtype(ftype)
# look in the books
saved = model.getRefByAddr( attr_addr )
_class, _addr, _obj = saved[0]
self.assertEquals( attr_addr, _addr)
self.assertEquals( attr_type, _class, '%s != %s' %(type(ftype), type(_class)))
self.assertTrue( model.hasRef( model.get_subtype(ftype), attr_addr))
return
def _HEAP_getFreeLists(self, mappings):
''' Understanding_the_LFH.pdf page 18 ++
We iterate on _HEAP.FreeLists to get ALL free blocks.
@returns freeblock_addr : the address of the _HEAP_ENTRY (chunk header)
size : the size of the free chunk + _HEAP_ENTRY header size, in blocks.
'''
res = list()
sentinel = self._orig_address_ + 0xc4 # utils.offsetof(_HEAP, 'FreeLists')
# FIXME - should be
## for freeblock in self.iterateListField( mappings, 'FreeLists'):
for freeblock_addr in self.FreeLists._iterateList(mappings):
if freeblock_addr == sentinel:
continue
m = mappings.getMmapForAddr(freeblock_addr)
freeblock = m.readStruct(freeblock_addr, _LIST_ENTRY)
blink_value = utils.getaddress(freeblock.BLink)
if (blink_value & 1): # points to _HEAP_BUCKET +1
log.warning('This freeblock BLink point to _HEAP_BUCKET at %x' %
(blink_value))
# its then a HEAP_ENTRY..
#chunk_header = m.readStruct( freeblock_addr - 2*Config.WORDSIZE, _HEAP_ENTRY)
chunk_header = m.readStruct(freeblock_addr - 2 * Config.WORDSIZE,
N11_HEAP_ENTRY3DOT_13DOT_2E) # Union stuff
if self.EncodeFlagMask:
log.debug('EncodeFlagMask is set on the HEAP. decoding is needed.')
chunk_header = _HEAP_ENTRY_decode(chunk_header, self)
#log.debug('chunk_header: %s'%(chunk_header.toString()))
res.append(
(freeblock_addr, chunk_header.Size)) # size = header + freespace
return res
def test_keepRef(self):
self.assertNotEqual(self.mappings, None)
for fname, ftype in self.heap_obj.getFields():
attr = getattr(self.heap_obj, fname)
if isCStringPointer(ftype):
# ignore that - attr_addr = getaddress(attr.ptr)
continue
elif isPointerType(ftype):
attr_addr = getaddress(attr)
else:
continue
if attr_addr == 0:
continue
self.assertTrue(
utils.is_valid_address_value(attr_addr, self.mappings),
'%s: 0x%x is not valid' % (fname, attr_addr))
# the book should register struct type, not pointer to strut type
attr_type = model.get_subtype(ftype)
# look in the books
saved = model.getRefByAddr(attr_addr)
_class, _addr, _obj = saved[0]
self.assertEquals(attr_addr, _addr)
self.assertEquals(attr_type, _class,
'%s != %s' % (type(ftype), type(_class)))
self.assertTrue(model.hasRef(model.get_subtype(ftype), attr_addr))
return
def _HEAP_getFreeLists(self, mappings):
""" Understanding_the_LFH.pdf page 18 ++
We iterate on _HEAP.FreeLists to get ALL free blocks.
@returns freeblock_addr : the address of the _HEAP_ENTRY (chunk header)
size : the size of the free chunk + _HEAP_ENTRY header size, in blocks.
"""
res = list()
sentinel = self._orig_address_ + 0xC4 # utils.offsetof(_HEAP, 'FreeLists')
# FIXME - should be
## for freeblock in self.iterateListField( mappings, 'FreeLists'):
for freeblock_addr in self.FreeLists._iterateList(mappings):
if freeblock_addr == sentinel:
continue
m = mappings.getMmapForAddr(freeblock_addr)
freeblock = m.readStruct(freeblock_addr, _LIST_ENTRY)
blink_value = utils.getaddress(freeblock.BLink)
if blink_value & 1: # points to _HEAP_BUCKET +1
log.warning("This freeblock BLink point to _HEAP_BUCKET at %x" % (blink_value))
# its then a HEAP_ENTRY..
# chunk_header = m.readStruct( freeblock_addr - 2*Config.WORDSIZE, _HEAP_ENTRY)
chunk_header = m.readStruct(freeblock_addr - 2 * Config.WORDSIZE, N11_HEAP_ENTRY3DOT_13DOT_2E) # Union stuff
if self.EncodeFlagMask:
log.debug("EncodeFlagMask is set on the HEAP. decoding is needed.")
chunk_header = _HEAP_ENTRY_decode(chunk_header, self)
# log.debug('chunk_header: %s'%(chunk_header.toString()))
res.append((freeblock_addr, chunk_header.Size)) # size = header + freespace
return res
def _HEAP_getFreeLists_by_blocksindex(self, mappings):
""" Understanding_the_LFH.pdf page 21
Not Implemented yet
"""
freeList = []
# 128 blocks
start = ctypes.addressof(self.BlocksIndex)
bi_addr = utils.getaddress(self.BlocksIndex)
# enumerate BlocksIndex recursively on ExtendedLookup param
while bi_addr != 0:
log.debug("BLocksIndex is at %x" % (bi_addr))
m = mappings.getMmapForAddr(bi_addr)
bi = m.readStruct(bi_addr, _HEAP_LIST_LOOKUP)
"""
('ExtendedLookup', POINTER(_HEAP_LIST_LOOKUP)),
('ArraySize', __uint32_t),
('ExtraItem', __uint32_t),
('ItemCount', __uint32_t),
('OutOfRangeItems', __uint32_t),
('BaseIndex', __uint32_t),
('ListHead', POINTER(_LIST_ENTRY)),
('ListsInUseUlong', POINTER(__uint32_t)),
('ListHints', POINTER(POINTER(_LIST_ENTRY))),
"""
log.debug("ArraySize is %d" % (bi.ArraySize))
log.debug("BlocksIndex: %s" % (bi.toString()))
hints_addr = utils.getaddress(bi.ListHints)
log.debug("ListHints is pointing to %x" % (hints_addr))
extlookup_addr = utils.getaddress(bi.ExtendedLookup)
log.debug("ExtendedLookup is pointing to %x" % (extlookup_addr))
if extlookup_addr == 0:
""" all chunks of size greater than or equal to BlocksIndex->ArraySize - 1 will
be stored in ascending order in FreeList[ArraySize-BaseIndex – 1] """
log.debug(
"Free chunks >= %d stored at FreeList[ArraySize(%d)-BaseIndex(%d) – 1]"
% (bi.ArraySize - 1, bi.ArraySize, bi.BaseIndex)
)
# raise NotImplementedError()
log.debug("-" * 80)
bi_addr = extlookup_addr
#
raise NotImplementedError("NOT FINISHED")
raise StopIteration
def _HEAP_getFreeLists_by_blocksindex(self, mappings):
''' Understanding_the_LFH.pdf page 21
Not Implemented yet
'''
freeList = []
# 128 blocks
start = ctypes.addressof(self.BlocksIndex)
bi_addr = utils.getaddress(self.BlocksIndex)
# enumerate BlocksIndex recursively on ExtendedLookup param
while bi_addr != 0:
log.debug('BLocksIndex is at %x' % (bi_addr))
m = mappings.getMmapForAddr(bi_addr)
bi = m.readStruct(bi_addr, _HEAP_LIST_LOOKUP)
'''
('ExtendedLookup', POINTER(_HEAP_LIST_LOOKUP)),
('ArraySize', __uint32_t),
('ExtraItem', __uint32_t),
('ItemCount', __uint32_t),
('OutOfRangeItems', __uint32_t),
('BaseIndex', __uint32_t),
('ListHead', POINTER(_LIST_ENTRY)),
('ListsInUseUlong', POINTER(__uint32_t)),
('ListHints', POINTER(POINTER(_LIST_ENTRY))),
'''
log.debug('ArraySize is %d' % (bi.ArraySize))
log.debug('BlocksIndex: %s' % (bi.toString()))
hints_addr = utils.getaddress(bi.ListHints)
log.debug('ListHints is pointing to %x' % (hints_addr))
extlookup_addr = utils.getaddress(bi.ExtendedLookup)
log.debug('ExtendedLookup is pointing to %x' % (extlookup_addr))
if extlookup_addr == 0:
''' all chunks of size greater than or equal to BlocksIndex->ArraySize - 1 will
be stored in ascending order in FreeList[ArraySize-BaseIndex – 1] '''
log.debug(
'Free chunks >= %d stored at FreeList[ArraySize(%d)-BaseIndex(%d) – 1]'
% (bi.ArraySize - 1, bi.ArraySize, bi.BaseIndex))
#raise NotImplementedError()
log.debug('-' * 80)
bi_addr = extlookup_addr
#
raise NotImplementedError('NOT FINISHED')
raise StopIteration
def _HEAP_SUBSEGMENT_get_freeblocks(self):
"""
Use AggregateExchg.Depth and NextFreeoffset to fetch the head, then traverse the links
"""
userblocks_addr = utils.getaddress(self.UserBlocks)
if not bool(userblocks_addr):
return []
if self.AggregateExchg.FreeEntryOffset == 0x2:
log.debug(" * FirstFreeOffset==0x2 Depth==%d" % (self.AggregateExchg.Depth))
# self.AggregateExchg.Depth the size of UserBlock divided by the HeapBucket size
# self.AggregateExchg.FreeEntryOffset starts at 0x2 (blocks), which means 0x10 bytes after UserBlocks
# see Understanding LFH page 14
# nextoffset of user data is at current + offset*8 + len(HEAP_ENTRY)
freeblocks = [
(userblocks_addr + (self.AggregateExchg.FreeEntryOffset * 8) + self.BlockSize * 8 * i, self.BlockSize * 8)
for i in range(self.AggregateExchg.Depth)
]
return freeblocks
def _HEAP_SUBSEGMENT_get_freeblocks(self):
'''
Use AggregateExchg.Depth and NextFreeoffset to fetch the head, then traverse the links
'''
userblocks_addr = utils.getaddress(self.UserBlocks)
if not bool(userblocks_addr):
return []
if self.AggregateExchg.FreeEntryOffset == 0x2:
log.debug(' * FirstFreeOffset==0x2 Depth==%d' %
(self.AggregateExchg.Depth))
# self.AggregateExchg.Depth the size of UserBlock divided by the HeapBucket size
# self.AggregateExchg.FreeEntryOffset starts at 0x2 (blocks), which means 0x10 bytes after UserBlocks
# see Understanding LFH page 14
# nextoffset of user data is at current + offset*8 + len(HEAP_ENTRY)
freeblocks = [
(userblocks_addr + (self.AggregateExchg.FreeEntryOffset * 8) +
self.BlockSize * 8 * i, self.BlockSize * 8)
for i in range(self.AggregateExchg.Depth)
]
return freeblocks
def _HEAP_SEGMENT_loadMember(self, attr, attrname, attrtype, mappings,
maxDepth):
#log.debug('_loadMember attrname : %s'%(attrname))
if attrname == 'LastValidEntry':
# isPointerType code.
_attrType = model.get_subtype(attrtype)
attr_obj_address = utils.getaddress(attr)
####
memoryMap = utils.is_valid_address(attr, mappings, _attrType)
if (not memoryMap):
log.debug("LastValidEntry out of mapping - 0x%lx - ignore " %
(attr_obj_address))
return True
from haystack.model import getRef, keepRef, delRef # TODO CLEAN
ref = getRef(_attrType, attr_obj_address)
if ref:
#log.debug("%s %s loading from references cache %s/0x%lx"%(attrname,attr,_attrType,attr_obj_address ))
#DO NOT CHANGE STUFF SOUPID attr.contents = ref
return True
#log.debug("%s %s loading from 0x%lx (is_valid_address: %s)"%(attrname,attr,attr_obj_address, memoryMap ))
##### Read the struct in memory and make a copy to play with.
### ERRROR attr.contents=_attrType.from_buffer_copy(memoryMap.readStruct(attr_obj_address, _attrType ))
contents = memoryMap.readStruct(attr_obj_address, _attrType)
# save that validated and loaded ref and original addr so we dont need to recopy it later
keepRef(contents, _attrType, attr_obj_address)
#log.debug("%s %s loaded memcopy from 0x%lx to 0x%lx"%(attrname, attr, attr_obj_address, (utils.getaddress(attr)) ))
# recursive validation checks on new struct
if not bool(attr):
log.warning('Member %s is null after copy: %s' % (attrname, attr))
return True
# go and load the pointed struct members recursively
if not contents.loadMembers(mappings, maxDepth):
log.debug('member %s was not loaded' % (attrname))
#invalidate the cache ref.
delRef(_attrType, attr_obj_address)
return False
return True
else:
return super(_HEAP_SEGMENT, self)._loadMember(attr, attrname, attrtype,
mappings, maxDepth)
def _HEAP_SUBSEGMENT_get_userblocks(self):
"""
AggregateExchg contains info on userblocks, number left, depth
"""
userblocks_addr = utils.getaddress(self.UserBlocks)
if not bool(userblocks_addr):
log.debug("Userblocks is null")
return []
# its basically an array of self.BlockCount blocks of self.BlockSie*8 bytes.
log.debug("fetching %d blocks of %d bytes" % (self.BlockCount, self.BlockSize * 8))
# UserBlocks points to _HEAP_USERDATA_HEADER. Real user data blocks will starts after sizeof( _HEAP_USERDATA_HEADER ) = 0x10
# each chunk starts with a 8 byte header + n user-writeable data
# user writable chunk starts with 2 bytes for next offset
# basically, first committed block is first.
# ( page 38 )
userblocks = [(userblocks_addr + 0x10 + self.BlockSize * 8 * i, self.BlockSize * 8) for i in range(self.BlockCount)]
#
## we need to substract non allocated blocks
# self.AggregateExchg.Depth counts how many blocks are remaining free
# if self.AggregateExchg.FreeEntryOffset == 0x2, there a are no commited blocks
return userblocks
def _HEAP_SEGMENT_loadMember(self, attr, attrname, attrtype, mappings, maxDepth):
# log.debug('_loadMember attrname : %s'%(attrname))
if attrname == "LastValidEntry":
# isPointerType code.
_attrType = model.get_subtype(attrtype)
attr_obj_address = utils.getaddress(attr)
####
memoryMap = utils.is_valid_address(attr, mappings, _attrType)
if not memoryMap:
log.debug("LastValidEntry out of mapping - 0x%lx - ignore " % (attr_obj_address))
return True
from haystack.model import getRef, keepRef, delRef # TODO CLEAN
ref = getRef(_attrType, attr_obj_address)
if ref:
# log.debug("%s %s loading from references cache %s/0x%lx"%(attrname,attr,_attrType,attr_obj_address ))
# DO NOT CHANGE STUFF SOUPID attr.contents = ref
return True
# log.debug("%s %s loading from 0x%lx (is_valid_address: %s)"%(attrname,attr,attr_obj_address, memoryMap ))
##### Read the struct in memory and make a copy to play with.
### ERRROR attr.contents=_attrType.from_buffer_copy(memoryMap.readStruct(attr_obj_address, _attrType ))
contents = memoryMap.readStruct(attr_obj_address, _attrType)
# save that validated and loaded ref and original addr so we dont need to recopy it later
keepRef(contents, _attrType, attr_obj_address)
# log.debug("%s %s loaded memcopy from 0x%lx to 0x%lx"%(attrname, attr, attr_obj_address, (utils.getaddress(attr)) ))
# recursive validation checks on new struct
if not bool(attr):
log.warning("Member %s is null after copy: %s" % (attrname, attr))
return True
# go and load the pointed struct members recursively
if not contents.loadMembers(mappings, maxDepth):
log.debug("member %s was not loaded" % (attrname))
# invalidate the cache ref.
delRef(_attrType, attr_obj_address)
return False
return True
else:
return super(_HEAP_SEGMENT, self)._loadMember(attr, attrname, attrtype, mappings, maxDepth)
def _HEAP_SUBSEGMENT_get_userblocks(self):
'''
AggregateExchg contains info on userblocks, number left, depth
'''
userblocks_addr = utils.getaddress(self.UserBlocks)
if not bool(userblocks_addr):
log.debug('Userblocks is null')
return []
# its basically an array of self.BlockCount blocks of self.BlockSie*8 bytes.
log.debug('fetching %d blocks of %d bytes' %
(self.BlockCount, self.BlockSize * 8))
# UserBlocks points to _HEAP_USERDATA_HEADER. Real user data blocks will starts after sizeof( _HEAP_USERDATA_HEADER ) = 0x10
# each chunk starts with a 8 byte header + n user-writeable data
# user writable chunk starts with 2 bytes for next offset
# basically, first committed block is first.
# ( page 38 )
userblocks = [(userblocks_addr + 0x10 + self.BlockSize * 8 * i,
self.BlockSize * 8) for i in range(self.BlockCount)]
#
## we need to substract non allocated blocks
# self.AggregateExchg.Depth counts how many blocks are remaining free
# if self.AggregateExchg.FreeEntryOffset == 0x2, there a are no commited blocks
return userblocks
def _HEAP_getFrontendChunks(self, mappings):
""" windows xp ?
the list of chunks from the frontend are deleted from the segment chunk list.
Functionnaly, (page 28) LFH_HEAP should be fetched by HEAP_BUCKET calcul
"""
res = list()
all_free = list()
all_committed = list()
log.debug("_HEAP_getFrontendChunks")
if self.FrontEndHeapType == 1: # windows XP per default
ptr = self.FrontEndHeap
## TODO delete this ptr from the heap-segment entries chunks
for x in range(128):
log.debug("finding lookaside %d at @%x" % (x, ptr))
m = mappings.getMmapForAddr(ptr)
st = m.readStruct(ptr, _HEAP_LOOKASIDE)
# load members on self.FrontEndHeap car c'est un void *
for free in st.iterateList("ListHead"): # single link list.
## TODO delete this free from the heap-segment entries chunks
log.debug("free")
res.append(free) # ???
pass
ptr += ctypes.sizeof(_HEAP_LOOKASIDE)
elif self.FrontEndHeapType == 2: # win7 per default
ptr = self.FrontEndHeap
log.debug("finding frontend at @%x" % (ptr))
m = mappings.getMmapForAddr(ptr)
st = m.readStruct(ptr, _LFH_HEAP)
# LFH is a big chunk allocated by the backend allocator, called subsegment
# but rechopped as small chunks of a heapbin.
# Active subsegment hold that big chunk.
#
#
# load members on self.FrontEndHeap car c'est un void *
if not st.loadMembers(mappings, 10):
log.error("Error on loading frontend")
raise model.NotValid("Frontend load at @%x is not valid" % (ptr))
# log.debug(st.LocalData[0].toString())
#
for sinfo in st.LocalData[0].SegmentInfo: #### 128 _HEAP_LOCAL_SEGMENT_INFO
# TODO , what about ActiveSubsegment ?
for items_ptr in sinfo.CachedItems: # 16 caches items max
items_addr = utils.getaddress(items_ptr)
if not bool(items_addr):
# log.debug('NULL pointer items')
continue
m = mappings.getMmapForAddr(items_addr)
subsegment = m.readStruct(items_addr, _HEAP_SUBSEGMENT)
# log.debug(subsegment)
## TODO current subsegment.SFreeListEntry is on error at some depth.
## bad pointer value on the second subsegment
chunks = subsegment.get_userblocks()
free = subsegment.get_freeblocks()
committed = set(chunks) - set(free)
all_free.extend(free)
all_committed.extend(committed)
log.debug(
"subseg: 0x%0.8x, commit: %d chunks free: %d chunks" % (items_addr, len(committed), len(free))
)
else:
# print 'FrontEndHeapType == %d'%(self.FrontEndHeapType)
# raise StopIteration
pass
return all_committed, all_free
def _HEAP_getSegmentList(self, mappings):
""" list all heap entries attached to one Heap structure.
Heap.SegmentList points to the _LIST_ENTRY of Heap.Segment. +0x10
Iteration on SegmentList should return Segment.
Offset is needed and provided through _listHead_.
FIXME: offset could/should be automated by declaring child _LIST_ENTRY fieldname (SegmentListEntry)
listmodel.py:164
BUT iteration should stop on sentinel values.
Internal algoritm of iterateListField will not use the root LIST_Entry address as a valid object.
It will use it as a sentinel.
If there is another sentinel... bad luck....
Heap.Segment.Entry is a valid Heap_entry, addr+size*8 should point to FirstValidEntry.
"""
allocated = list()
free = list()
# log.debug('%s'%(self))
for segment in self.iterateListField(mappings, "SegmentList"):
# for segment_addr in self.SegmentList._iterateList( mappings):
segment_addr = segment._orig_addr_
first_addr = utils.getaddress(segment.FirstEntry)
last_addr = utils.getaddress(segment.LastValidEntry)
log.debug(
"Heap.Segment: 0x%0.8x FirstEntry: 0x%0.8x LastValidEntry: 0x%0.8x" % (segment_addr, first_addr, last_addr)
)
skiplist = dict()
for ucr in segment.iterateListField(mappings, "UCRSegmentList"):
ucr_addr = ucr._orig_addr_
if ucr_addr is None:
log.error("None in _orig_addr_")
seg_addr = utils.getaddress(ucr.SegmentEntry.FLink)
if ucr.Address is None:
log.error("None in ucr.Address")
try:
log.debug(
"Heap.Segment.UCRSegmentList: 0x%0.8x addr: 0x%0.8x size: 0x%0.5x"
% (ucr_addr, ucr.Address, ucr.Size * 8)
)
except TypeError, e:
import code
code.interact(local=locals())
# log.debug("%s"%(ucr.SegmentEntry)) # TODO - CHECK FOR CONSISTENCY ? more heapdebug than haystack debug
skiplist[ucr.Address] = ucr.Size * 8
# # obviously not usable, first entry sits on heap header.
# chunk_header = segment.Entry
# if self.EncodeFlagMask: #heap.EncodeFlagMask
# log.debug('EncodeFlagMask is set on the HEAP. decoding is needed.')
# chunk_header = _HEAP_ENTRY_decode(chunk_header, self)
# chunk_addr = segment._orig_addr_ + utils.offsetof(_HEAP_SEGMENT, 'Entry')
# log.debug('Heap.Segment.Entry: 0x%0.8x\n%s'%( chunk_addr, chunk_header))
# print segment
from haystack.model import getRef # TODO CLEAN
chunk_addr = first_addr
while chunk_addr < last_addr:
if chunk_addr in skiplist:
size = skiplist[chunk_addr]
log.debug("Skipping 0x%0.8x - skip %0.5x bytes to 0x%0.8x" % (chunk_addr, size, chunk_addr + size))
chunk_addr += size
continue
chunk_header = getRef(_HEAP_ENTRY, chunk_addr)
if chunk_header is None: # force read it
chunk_header = _get_chunk(mappings, self, chunk_addr)
if self.EncodeFlagMask: # heap.EncodeFlagMask
chunk_header = _HEAP_ENTRY_decode(chunk_header, self)
# log.debug('\t\tEntry: 0x%0.8x\n%s'%( chunk_addr, chunk_header))
if (chunk_header.Flags & 1) == 1:
log.debug("Chunk 0x%0.8x is in use size: %0.5x" % (chunk_addr, chunk_header.Size * 8))
allocated.append((chunk_addr, chunk_header.Size * 8))
else:
log.debug("Chunk 0x%0.8x is FREE" % (chunk_addr))
free.append((chunk_addr, chunk_header.Size * 8))
pass
chunk_addr += chunk_header.Size * 8
def _HEAP_getSegmentList(self, mappings):
''' list all heap entries attached to one Heap structure.
Heap.SegmentList points to the _LIST_ENTRY of Heap.Segment. +0x10
Iteration on SegmentList should return Segment.
Offset is needed and provided through _listHead_.
FIXME: offset could/should be automated by declaring child _LIST_ENTRY fieldname (SegmentListEntry)
listmodel.py:164
BUT iteration should stop on sentinel values.
Internal algoritm of iterateListField will not use the root LIST_Entry address as a valid object.
It will use it as a sentinel.
If there is another sentinel... bad luck....
Heap.Segment.Entry is a valid Heap_entry, addr+size*8 should point to FirstValidEntry.
'''
allocated = list()
free = list()
#log.debug('%s'%(self))
for segment in self.iterateListField(mappings, 'SegmentList'):
#for segment_addr in self.SegmentList._iterateList( mappings):
segment_addr = segment._orig_addr_
first_addr = utils.getaddress(segment.FirstEntry)
last_addr = utils.getaddress(segment.LastValidEntry)
log.debug(
'Heap.Segment: 0x%0.8x FirstEntry: 0x%0.8x LastValidEntry: 0x%0.8x'
% (segment_addr, first_addr, last_addr))
skiplist = dict()
for ucr in segment.iterateListField(mappings, 'UCRSegmentList'):
ucr_addr = ucr._orig_addr_
if ucr_addr is None:
log.error('None in _orig_addr_')
seg_addr = utils.getaddress(ucr.SegmentEntry.FLink)
if ucr.Address is None:
log.error('None in ucr.Address')
try:
log.debug(
"Heap.Segment.UCRSegmentList: 0x%0.8x addr: 0x%0.8x size: 0x%0.5x"
% (ucr_addr, ucr.Address, ucr.Size * 8))
except TypeError, e:
import code
code.interact(local=locals())
#log.debug("%s"%(ucr.SegmentEntry)) # TODO - CHECK FOR CONSISTENCY ? more heapdebug than haystack debug
skiplist[ucr.Address] = ucr.Size * 8
# # obviously not usable, first entry sits on heap header.
#chunk_header = segment.Entry
#if self.EncodeFlagMask: #heap.EncodeFlagMask
# log.debug('EncodeFlagMask is set on the HEAP. decoding is needed.')
# chunk_header = _HEAP_ENTRY_decode(chunk_header, self)
#chunk_addr = segment._orig_addr_ + utils.offsetof(_HEAP_SEGMENT, 'Entry')
#log.debug('Heap.Segment.Entry: 0x%0.8x\n%s'%( chunk_addr, chunk_header))
#print segment
from haystack.model import getRef # TODO CLEAN
chunk_addr = first_addr
while (chunk_addr < last_addr):
if chunk_addr in skiplist:
size = skiplist[chunk_addr]
log.debug('Skipping 0x%0.8x - skip %0.5x bytes to 0x%0.8x' %
(chunk_addr, size, chunk_addr + size))
chunk_addr += size
continue
chunk_header = getRef(_HEAP_ENTRY, chunk_addr)
if chunk_header is None: # force read it
chunk_header = _get_chunk(mappings, self, chunk_addr)
if self.EncodeFlagMask: #heap.EncodeFlagMask
chunk_header = _HEAP_ENTRY_decode(chunk_header, self)
#log.debug('\t\tEntry: 0x%0.8x\n%s'%( chunk_addr, chunk_header))
if ((chunk_header.Flags & 1) == 1):
log.debug('Chunk 0x%0.8x is in use size: %0.5x' %
(chunk_addr, chunk_header.Size * 8))
allocated.append((chunk_addr, chunk_header.Size * 8))
else:
log.debug('Chunk 0x%0.8x is FREE' % (chunk_addr))
free.append((chunk_addr, chunk_header.Size * 8))
pass
chunk_addr += chunk_header.Size * 8
def _HEAP_getFrontendChunks(self, mappings):
''' windows xp ?
the list of chunks from the frontend are deleted from the segment chunk list.
Functionnaly, (page 28) LFH_HEAP should be fetched by HEAP_BUCKET calcul
'''
res = list()
all_free = list()
all_committed = list()
log.debug('_HEAP_getFrontendChunks')
if self.FrontEndHeapType == 1: # windows XP per default
ptr = self.FrontEndHeap
## TODO delete this ptr from the heap-segment entries chunks
for x in range(128):
log.debug('finding lookaside %d at @%x' % (x, ptr))
m = mappings.getMmapForAddr(ptr)
st = m.readStruct(ptr, _HEAP_LOOKASIDE)
# load members on self.FrontEndHeap car c'est un void *
for free in st.iterateList('ListHead'): # single link list.
## TODO delete this free from the heap-segment entries chunks
log.debug('free')
res.append(free) #???
pass
ptr += ctypes.sizeof(_HEAP_LOOKASIDE)
elif self.FrontEndHeapType == 2: # win7 per default
ptr = self.FrontEndHeap
log.debug('finding frontend at @%x' % (ptr))
m = mappings.getMmapForAddr(ptr)
st = m.readStruct(ptr, _LFH_HEAP)
# LFH is a big chunk allocated by the backend allocator, called subsegment
# but rechopped as small chunks of a heapbin.
# Active subsegment hold that big chunk.
#
#
# load members on self.FrontEndHeap car c'est un void *
if not st.loadMembers(mappings, 10):
log.error('Error on loading frontend')
raise model.NotValid('Frontend load at @%x is not valid' % (ptr))
#log.debug(st.LocalData[0].toString())
#
for sinfo in st.LocalData[
0].SegmentInfo: #### 128 _HEAP_LOCAL_SEGMENT_INFO
# TODO , what about ActiveSubsegment ?
for items_ptr in sinfo.CachedItems: # 16 caches items max
items_addr = utils.getaddress(items_ptr)
if not bool(items_addr):
#log.debug('NULL pointer items')
continue
m = mappings.getMmapForAddr(items_addr)
subsegment = m.readStruct(items_addr, _HEAP_SUBSEGMENT)
#log.debug(subsegment)
## TODO current subsegment.SFreeListEntry is on error at some depth.
## bad pointer value on the second subsegment
chunks = subsegment.get_userblocks()
free = subsegment.get_freeblocks()
committed = set(chunks) - set(free)
all_free.extend(free)
all_committed.extend(committed)
log.debug(
'subseg: 0x%0.8x, commit: %d chunks free: %d chunks' %
(items_addr, len(committed), len(free)))
else:
#print 'FrontEndHeapType == %d'%(self.FrontEndHeapType)
#raise StopIteration
pass
return all_committed, all_free
