progdb.py

   1 #
   2 # This module deals with various "program databases" as required
   3 # for analysis passes. Among them: function database (funcdb.yaml),
   4 # databases of structures, etc.
   5 #
   6 import os.path
   7 import copy
   8 import logging
   9
  10 import yaml
  11 import yamlutils
  12
  13 import utils
  14 from utils import repr_stable
  15 import core
  16 from core import is_addr, is_value, is_expr, EXPR
  17
  18
  19 _log = logging.getLogger(__name__)
  20
  21 FUNC_DB = {}
  22 FUNC_DB_BY_ADDR = {}
  23 struct_types = {}
  24 struct_instances = {}
  25
  26
  27 def set_funcdb(db):
  28     global FUNC_DB, FUNC_DB_BY_ADDR
  29     # index by name in addition to by addr
  30     for addr, props in list(db.items()):
  31         FUNC_DB[props["label"]] = props
  32     FUNC_DB_BY_ADDR = db
  33
  34
  35 REG_PROPS = [
  36     "callsites_live_out", "modifieds", "preserveds", "reach_exit", "reach_exit_maybe",
  37     "params", "estimated_params", "returns",
  38 ]
  39
  40
  41 def reglist2set(regs):
  42     return set(core.REG(x) for x in regs)
  43
  44
  45 def preprocess_funcdb(FUNC_DB):
  46     for addr, props in FUNC_DB.items():
  47         for prop in REG_PROPS:
  48             if prop in props:
  49                 props[prop] = reglist2set(props[prop])
  50
  51         if "calls_live_out" in props:
  52             new = [(x, core.ADDR(y), reglist2set(z)) for x, y, z in props["calls_live_out"]]
  53             props["calls_live_out"] = new
  54
  55
  56 def postprocess_funcdb(FUNC_DB):
  57     for addr, props in FUNC_DB.items():
  58         for prop in REG_PROPS:
  59             if prop in props:
  60                 props[prop] = sorted([x.name for x in props[prop]], key=utils.natural_sort_key)
  61
  62
  63 def load_funcdb(*fnames):
  64     FUNC_DB = {}
  65     for fname in fnames:
  66         with open(fname) as f:
  67             db = yaml.safe_load(f)
  68             FUNC_DB.update(db)
  69
  70     preprocess_funcdb(FUNC_DB)
  71     set_funcdb(FUNC_DB)
  72
  73
  74 def save_funcdb(fname, backup=True):
  75     db = copy.deepcopy(FUNC_DB_BY_ADDR)
  76     postprocess_funcdb(db)
  77     if backup and os.path.exists(fname):
  78         os.rename(fname, fname + ".bak")
  79
  80     with open(fname, "w") as f:
  81         yaml.dump(db, f)
  82
  83
  84 def check_invariants(cfg):
  85     if "reach_exit" in cfg.props:
  86         reach = cfg.props["reach_exit"]
  87         reach_maybe = cfg.props.get("reach_exit_maybe", set())
  88         assert reach_maybe.issubset(reach), "%s: maybe: %s, sure: %s" % (cfg.props["name"],
  89             repr_stable(reach_maybe), repr_stable(reach))
  90
  91
  92 def update_funcdb(cfg):
  93     "Aggregate data from each CFG processed into a function DB."
  94     if "addr" not in cfg.props:
  95         return
  96
  97     check_invariants(cfg)
  98
  99     func_props = FUNC_DB_BY_ADDR.setdefault(cfg.props["addr"], {})
 100     func_props["label"] = cfg.props["name"]
 101
 102     PROPS = (
 103         "params", "estimated_params", "params_why", "modifieds", "preserveds",
 104         "reach_exit", "reach_exit_maybe", "calls_live_out", "noreturn",
 105         "has_infloops"
 106     )
 107
 108     for prop in PROPS:
 109         if prop in cfg.props:
 110             func_props[prop] = cfg.props[prop]
 111
 112     for prop in ("calls", "calls_indir", "func_refs", "mmio_refs"):
 113         if prop in cfg.props:
 114             def ext_repr(x):
 115                 if is_addr(x):
 116                     return x.addr
 117                 if is_value(x):
 118                     return hex(x.val)
 119                 if is_expr(x):
 120                     if x.op == "+" and len(x.args) == 2:
 121                         if is_value(x.args[1]):
 122                             x = EXPR("+", [x.args[1], x.args[0]])
 123                     return str(x)
 124                 return str(x)
 125             func_props[prop] = sorted([ext_repr(x) for x in cfg.props[prop]])
 126
 127 #
 128 # Updated funcs tracking
 129 #
 130
 131 UPDATED_FUNCS = set()
 132
 133 def clear_updated():
 134     UPDATED_FUNCS.clear()
 135
 136 def mark_updated(func):
 137     UPDATED_FUNCS.add(func)
 138
 139 def update_cfg_prop(cfg, prop, new_val):
 140     if cfg.props.get(prop) != new_val:
 141         mark_updated(cfg.props["name"])
 142         _log.info("%s: %s updated from %s to %s" % (cfg.props["name"], prop,
 143             utils.repr_stable(cfg.props.get(prop)), utils.repr_stable(new_val)))
 144     cfg.props[prop] = new_val
 145
 146 #
 147 # Symtab functions
 148 #
 149
 150 SYMTAB = {}
 151
 152 def load_symtab(fname):
 153     with open(fname) as f:
 154         for l in f:
 155             l = l.strip()
 156             addr, sym = l.split()
 157             SYMTAB[sym] = int(addr, 16)
 158
 159 # Returns real memory (not symbolic) address, for bindata module.
 160 def lookup_sym(sym):
 161     return SYMTAB[sym]
 162
 163 #
 164 # Struct database functions
 165 #
 166
 167 def set_struct_types(data):
 168     global struct_types
 169     struct_types = data
 170
 171
 172 def set_struct_instances(data):
 173     global struct_instances
 174     struct_instances = data
 175
 176
 177 def get_struct_types():
 178     global struct_types
 179     return struct_types
 180
 181
 182 def get_struct_instances():
 183     global struct_instances
 184     return struct_instances