shdr parsing in hackyelf, linkmap parsing bug fixes

This commit is contained in:
PoroCYon 2020-10-16 04:13:44 +02:00
parent 6ad8eed5b2
commit e37a0d8e50
2 changed files with 213 additions and 9 deletions

View File

@ -8,6 +8,7 @@
from struct import unpack
from typing import *
ELFCLASS32 = 1
ELFCLASS64 = 2
@ -24,6 +25,30 @@ DT_NEEDED = 1
DT_STRTAB = 5
DT_SYMTAB = 6
SHT_NULL = 0
SHT_PROGBITS = 1
SHT_SYMTAB = 2
SHT_STRTAB = 3
SHT_DYNSYM = 11
STB_LOCAL = 0
STB_GLOBAL = 1
STB_WEAK = 2
STT_NOTYPE = 0
STT_OBJECT = 1
STT_FUNC = 2
STT_SECTION= 3
STT_FILE = 4
STT_COMMON = 5
STT_TLS = 6
STT_GNU_IFUNC = 10
STV_DEFAULT = 0
STV_INTERNAL = 1
STV_HIDDEN = 2
STV_PROTECTED = 3
class Phdr(NamedTuple):
ptype: int
off : int
@ -38,6 +63,27 @@ class Dyn(NamedTuple):
tag: int
val: int
class Shdr(NamedTuple):
name: Union[int, str]
type: int
flags: int
addr: int
offset: int
size: int
link: int
info: int
addralign: int
entsize: int
class Sym(NamedTuple):
name: str
value: int
size: int
type: int
binding: int
visibility: int
shndx: int
class ELF(NamedTuple):
data : bytes
ident : bytes
@ -46,8 +92,18 @@ class ELF(NamedTuple):
entry : int
phdrs : Sequence[Phdr]
dyn : Sequence[Dyn]
shdrs : Sequence[Shdr]
symtab: Sequence[Sym]
dynsym: Sequence[Sym]
is32bit: bool
def readstr(data: bytes, off: int) -> str:
strb = bytearray()
while data[off] != 0 and off < len(data):
strb.append(data[off])
off = off + 1
return strb.decode('utf-8')
# yeah, there's some code duplication here
# idgaf
@ -74,6 +130,46 @@ def parse_dyn32(data: bytes, dynp: Phdr) -> Dyn:
return ds
def parse_shdr32(data: bytes, shoff: int, shentsz: int, shnum: int,
shstrndx: int) -> Sequence[Shdr]:
if shnum*shentsz+shoff > len(data) or shentsz==0 or shnum==0 or shoff==0:
print("snum*shentsz+shoff",shnum*shentsz+shoff)
print("len(data)",len(data))
print("shentsz",shentsz)
print("shnum",shnum)
print("shoff",shoff)
return []
ss = []
for off in range(shoff, shoff+shentsz*shnum, shentsz):
noff, typ, flags, addr, off, size, link, info, align, entsz = \
unpack('<IIIIIIIIII', data[off:off+10*4])
s = Shdr(noff, typ, flags, addr, off, size, link, info, align, entsz)
ss.append(s)
if shstrndx < shnum:
shstr = ss[shstrndx]
for i in range(len(ss)):
sname = readstr(data, shstr.offset + ss[i].name) \
if ss[i].name < shstr.size else None
ss[i] = Shdr(sname, ss[i].type, ss[i].flags, ss[i].addr,
ss[i].offset, ss[i].size, ss[i].link, ss[i].info,
ss[i].addralign, ss[i].entsize)
return ss
def parse_sym32(data: bytes, sym: Shdr, strt: Shdr) -> Sequence[Sym]:
ss = []
for off in range(sym.offset, sym.offset+sym.size, sym.entsize):
noff, val, sz, info, other, shndx = \
unpack('<IIIBBH', data[off:off+3*4+2+2])
sn = readstr(data, strt.offset + noff) \
if noff < strt.size else None
s = Sym(sn, val, sz, (info & 15), (info >> 4), other, shndx)
ss.append(s)
return sorted(ss, key=lambda x:x.value)
def parse_32(data: bytes) -> ELF:
ident = data[:16]
eclass = data[4]
@ -81,8 +177,12 @@ def parse_32(data: bytes) -> ELF:
entry = unpack('<I', data[24:24+4])[0]
phoff = unpack('<I', data[28:28+4])[0]
shoff = unpack('<I', data[32:32+4])[0]
phentsz = unpack('<H', data[42:42+2])[0]
phnum = unpack('<H', data[44:44+2])[0]
shentsz = unpack('<H', data[46:46+2])[0]
shnum = unpack('<H', data[48:48+2])[0]
shstrndx= unpack('<H', data[50:50+2])[0]
phdrs = parse_phdr32(data, phoff, phentsz, phnum)
dyn = None
@ -92,12 +192,37 @@ def parse_32(data: bytes) -> ELF:
dyn = parse_dyn32(data, p)
break
return ELF(data, ident, eclass, mach, entry, phdrs, dyn, True)
shdrs = parse_shdr32(data, shoff, shentsz, shnum, shstrndx)
#print("shdrs",shdrs)
symtabsh = [s for s in shdrs if s.type == SHT_SYMTAB and s.name == ".symtab"]
strtabsh = [s for s in shdrs if s.type == SHT_STRTAB and s.name == ".strtab"]
dynsymsh = [s for s in shdrs if s.type == SHT_SYMTAB and s.name == ".dynsym"]
dynstrsh = [s for s in shdrs if s.type == SHT_STRTAB and s.name == ".dynstr"]
#print("symtab",symtabsh)
#print("strtab",strtabsh)
assert len(symtabsh) < 2
assert len(strtabsh) < 2
assert len(dynsymsh) < 2
assert len(dynstrsh) < 2
symtab, dynsym = None, None
if len(symtabsh) and len(strtabsh):
symtab = parse_sym32(data, symtabsh[0], strtabsh[0]) \
if len(shdrs) > 0 else []
if len(dynsymsh) and len(dynstrsh):
dynsym = parse_sym32(data, symtabsh[0], strtabsh[0]) \
if len(shdrs) > 0 else []
return ELF(data, ident, eclass, mach, entry, phdrs, dyn, shdrs,
symtab, dynsym, True)
def parse_phdr64(data: bytes, phoff:int, phentsz:int, phnum:int) -> Sequence[Phdr]:
ps = []
for off in range(phoff, phoff+phentsz*phnum, phentsz):
# TODO
# TODO # what is TODO exactly??
ptype, flags, off, vaddr, paddr, filesz, memsz, align = \
unpack('<IIQQQQQQ', data[off:off+2*4+6*8])
p = Phdr(ptype, off, vaddr, paddr, filesz, memsz, flags, align)
@ -118,6 +243,41 @@ def parse_dyn64(data: bytes, dynp: Phdr) -> Dyn:
return ds
def parse_shdr64(data: bytes, shoff: int, shentsz: int, shnum: int,
shstrndx: int) -> Sequence[Shdr]:
if shnum*shentsz+shoff >= len(data) or shentsz==0 or shnum==0 or shoff==0:
return []
ss = []
for off in range(shoff, shoff+shentsz*shnum, shentsz):
noff, typ, flags, addr, off, size, link, info, align, entsz = \
unpack('<IIQQQQIIQQ', data[off:off+4*4+6*8])
s = Shdr(noff, typ, flags, addr, off, size, link, info, align, entsz)
ss.append(s)
if shstrndx < shnum:
shstr = ss[shstrndx]
for i in range(len(ss)):
sname = readstr(data, shstr.offset + ss[i].name) \
if ss[i].name < shstr.size else None
ss[i] = Shdr(sname, ss[i].type, ss[i].flags, ss[i].addr,
ss[i].offset, ss[i].size, ss[i].link, ss[i].info,
ss[i].addralign, ss[i].entsize)
return ss
def parse_sym64(data: bytes, sym: Shdr, strt: Shdr) -> Sequence[Sym]:
ss = []
for off in range(sym.offset, sym.offset+sym.size, sym.entsize):
noff, info, other, shndx, value, sz = \
unpack('<IBBHQQ', data[off:off+4+2+2+8*2])
sn = readstr(data, strt.offset + noff) \
if noff < strt.size else None
s = Sym(sn, val, sz, (info & 15), (info >> 4), other, shndx)
ss.append(s)
return sorted(ss, key=lambda x:x.value)
def parse_64(data: bytes) -> ELF:
ident = data[:16]
eclass = data[4]
@ -125,8 +285,12 @@ def parse_64(data: bytes) -> ELF:
entry = unpack('<Q', data[24:24+8])[0]
phoff = unpack('<Q', data[32:32+8])[0]
shoff = unpack('<Q', data[40:40+8])[0]
phentsz = unpack('<H', data[54:54+2])[0]
phnum = unpack('<H', data[56:56+2])[0]
shentsz = unpack('<H', data[58:58+2])[0]
shnum = unpack('<H', data[60:60+2])[0]
shstrndx= unpack('<H', data[62:62+2])[0]
phdrs = parse_phdr64(data, phoff, phentsz, phnum)
dyn = None
@ -136,7 +300,28 @@ def parse_64(data: bytes) -> ELF:
dyn = parse_dyn64(data, p)
break
return ELF(data, ident, eclass, mach, entry, phdrs, dyn, False)
shdrs = parse_shdr64(data, shoff, shentsz, shnum, shstrndx)
symtabsh = [s for s in shdrs if s.type == SHT_SYMTAB and s.name == ".symtab"]
strtabsh = [s for s in shdrs if s.type == SHT_STRTAB and s.name == ".strtab"]
dynsymsh = [s for s in shdrs if s.type == SHT_SYMTAB and s.name == ".dynsym"]
dynstrsh = [s for s in shdrs if s.type == SHT_STRTAB and s.name == ".dynstr"]
assert len(symtabsh) < 2
assert len(strtabsh) < 2
assert len(dynsymsh) < 2
assert len(dynstrsh) < 2
symtab, dynsym = None, None
if len(symtabsh) and len(strtabsh):
symtab = parse_sym64(data, symtabsh[0], strtabsh[0]) \
if len(shdrs) > 0 else []
if len(dynsymsh) and len(dynstrsh):
dynsym = parse_sym64(data, symtabsh[0], strtabsh[0]) \
if len(shdrs) > 0 else []
return ELF(data, ident, eclass, mach, entry, phdrs, dyn, shdrs,
symtab, dynsym, False)
def parse(data: bytes) -> ELF:
assert data[:4] == b'\x7FELF', "Not a valid ELF file" # good enough

View File

@ -27,6 +27,17 @@ class XRef(NamedTuple):
name: str
deff: str
reff: Sequence[str]
class ImpObj(NamedTuple):
ar: str
obj: str
class ImpRsnObj(NamedTuple):
sym: str
obj: ImpObj
class ImpRsnSym(NamedTuple):
sym: str
class ArImp(NamedTuple):
impobj: ImpObj
reason: Union[ImpRsnSym, ImpRsnObj]
class LinkMap(NamedTuple):
common : Sequence[CommonSym]
@ -34,11 +45,13 @@ class LinkMap(NamedTuple):
memcfg : Sequence[MemCfg]
mmap : Sequence[MMap]
xref : Sequence[XRef]
arimp : Sequence[ArImp]
def parse_common( ls: Sequence[str]) -> Sequence[CommonSym]: return [] # TODO
def parse_discard(ls: Sequence[str]) -> Sequence[Discard ]: return [] # TODO
def parse_memcfg( ls: Sequence[str]) -> Sequence[MemCfg ]: return [] # TODO
def parse_xref( ls: Sequence[str]) -> Sequence[XRef ]: return [] # TODO
def parse_arimp( ls: Sequence[str]) -> Sequence[ArImp ]: return [] # TODO
def parse_mmap(ls: Sequence[str]) -> Sequence[MMap]:
rrr = []
@ -55,7 +68,9 @@ def parse_mmap(ls: Sequence[str]) -> Sequence[MMap]:
#print(repr(l))
s = l.strip(); w = s.split()
if s.startswith('LOAD ') or s.startswith('OUTPUT('): continue#break
if s.startswith('LOAD ') or s.startswith('OUTPUT(') or \
s.startswith('START GROUP') or s.startswith('END GROUP'):
continue#break
if l[0] != ' ':
bigsect = w[0]
@ -68,7 +83,7 @@ def parse_mmap(ls: Sequence[str]) -> Sequence[MMap]:
continue # addr placed on next line for prettyprinting reasons
#print(repr(l), w[0])
assert w[0].startswith("0x"), "welp, bad symbol addr"
assert w[0].startswith("0x"), "welp, bad symbol addr %s"%w[0]
addr = int(w[0], 16)
@ -90,10 +105,11 @@ def parse(s: str) -> LinkMap:
MEMCFG = 2
MMAP = 3
XREF = 4
ARIMP = 5
curpt = -1
commonl, discardl, memcfgl, mmapl, xrefl = [], [], [], [], []
commonl, discardl, memcfgl, mmapl, xrefl, arimpl = [], [], [], [], [], []
for l in s.split('\n'):
if len(l.strip()) == 0: continue
@ -103,15 +119,18 @@ def parse(s: str) -> LinkMap:
elif ls == "Discarded input sections": curpt = DISCARD
elif ls == "Memory Configuration": curpt = MEMCFG
elif ls == "Linker script and memory map": curpt = MMAP
elif ls == 'Cross Reference Table': curpt = XREF
elif ls == "Cross Reference Table": curpt = XREF
elif ls == "Archive member included to satisfy reference by file (symbol)": curpt = ARIMP
elif curpt == COMMON : commonl.append(l)
elif curpt == DISCARD: discardl.append(l)
elif curpt == MEMCFG : memcfgl.append(l)
elif curpt == MMAP : mmapl.append(l)
elif curpt == XREF : xrefl.append(l)
elif curpt == ARIMP : arimpl.append(l)
else:
assert False, "bad line %s" % ls
return LinkMap(parse_common(commonl), parse_discard(discardl), \
parse_memcfg(memcfgl), parse_mmap(mmapl), parse_xref(xrefl))
return LinkMap(parse_common(commonl), parse_discard(discardl),
parse_memcfg(memcfgl), parse_mmap(mmapl), parse_xref(xrefl),
parse_arimp(arimpl))