contrib/unused_functions.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3 #
   4 # Copyright (c) 2018 Free Software Foundation
   5 # Contributed by Bernhard Reutner-Fischer <aldot@gcc.gnu.org>
   6 # Inspired by bloat-o-meter from busybox.
   7
   8 # This software may be used and distributed according to the terms and
   9 # conditions of the GNU General Public License as published by the Free
  10 # Software Foundation.
  11
  12 # For a set of object-files, determine symbols that are
  13 #  - public but should be static
  14
  15 # Examples:
  16 # unused_functions.py ./gcc/fortran
  17 # unused_functions.py gcc/c  gcc/c-family/ gcc/*-c.o | grep -v "'gt_"
  18 # unused_functions.py gcc/cp gcc/c-family/ gcc/*-c.o | grep -v "'gt_"
  19
  20 import sys, os
  21 from tempfile import mkdtemp
  22 from subprocess import Popen, PIPE
  23
  24 def usage():
  25     sys.stderr.write("usage: %s [-v] [dirs | files] [-- <readelf options>]\n"
  26                         % sys.argv[0])
  27     sys.stderr.write("\t-v\tVerbose output\n");
  28     sys.exit(1)
  29
  30 (odir, sym_args, tmpd, verbose) = (set(), "", None, False)
  31
  32 for i in range(1, len(sys.argv)):
  33     f = sys.argv[i]
  34     if f == '--': # sym_args
  35         sym_args = ' '.join(sys.argv[i + 1:])
  36         break
  37     if f == '-v':
  38         verbose = True
  39         continue
  40     if not os.path.exists(f):
  41         sys.stderr.write("Error: No such file or directory '%s'\n" % f)
  42         usage()
  43     else:
  44         if f.endswith('.a') and tmpd is None:
  45             tmpd = mkdtemp(prefix='unused_fun')
  46         odir.add(f)
  47
  48 def dbg(args):
  49     if not verbose: return
  50     print(args)
  51
  52 def get_symbols(file):
  53     syms = {}
  54     rargs = "readelf -W -s %s %s" % (sym_args, file)
  55     p0 = Popen((a for a in rargs.split(' ') if a.strip() != ''), stdout=PIPE)
  56     p1 = Popen(["c++filt"], stdin=p0.stdout, stdout=PIPE,
  57             universal_newlines=True)
  58     lines = p1.communicate()[0]
  59     for l in lines.split('\n'):
  60         l = l.strip()
  61         if not len(l) or not l[0].isdigit(): continue
  62         larr = l.split()
  63         if len(larr) != 8: continue
  64         num, value, size, typ, bind, vis, ndx, name = larr
  65         if typ == 'SECTION' or typ == 'FILE': continue
  66         # I don't think we have many aliases in gcc, re-instate the addr
  67         # lut otherwise.
  68         if vis != 'DEFAULT': continue
  69         #value = int(value, 16)
  70         #size = int(size, 16) if size.startswith('0x') else int(size)
  71         defined = ndx != 'UND'
  72         globl = bind == 'GLOBAL'
  73         # c++ RID_FUNCTION_NAME dance. FORNOW: Handled as local use
  74         # Is that correct?
  75         if name.endswith('::__FUNCTION__') and typ == 'OBJECT':
  76             name = name[0:(len(name) - len('::__FUNCTION__'))]
  77             if defined: defined = False
  78         if defined and not globl: continue
  79         syms.setdefault(name, {})
  80         syms[name][['use','def'][defined]] = True
  81         syms[name][['local','global'][globl]] = True
  82     # Note: we could filter out e.g. debug_* symbols by looking for
  83     # value in the debug_macro sections.
  84     if p1.returncode != 0:
  85         print("Warning: Reading file '%s' exited with %r|%r"
  86             % (file, p0.returncode, p1.returncode))
  87     p0.kill()
  88     return syms
  89
  90 (oprog, nprog) = ({}, {})
  91
  92 def walker(paths):
  93     def ar_x(archive):
  94         dbg("Archive %s" % path)
  95         f = os.path.abspath(archive)
  96         f = os.path.splitdrive(f)[1]
  97         d = tmpd + os.path.sep + f
  98         d = os.path.normpath(d)
  99         owd = os.getcwd()
 100         try:
 101             os.makedirs(d)
 102             os.chdir(d)
 103             p0 = Popen(["ar", "x", "%s" % os.path.join(owd, archive)],
 104                     stderr=PIPE, universal_newlines=True)
 105             p0.communicate()
 106             if p0.returncode > 0: d = None # assume thin archive
 107         except:
 108             dbg("ar x: Error: %s: %s" % (archive, sys.exc_info()[0]))
 109             os.chdir(owd)
 110             raise
 111         os.chdir(owd)
 112         if d: dbg("Extracted to %s" % (d))
 113         return (archive, d)
 114
 115     def ar_t(archive):
 116         dbg("Thin archive, using existing files:")
 117         try:
 118             p0 = Popen(["ar", "t", "%s" % archive], stdout=PIPE,
 119                     universal_newlines=True)
 120             ret = p0.communicate()[0]
 121             return ret.split('\n')
 122         except:
 123             dbg("ar t: Error: %s: %s" % (archive, sys.exc_info()[0]))
 124             raise
 125
 126     prog = {}
 127     for path in paths:
 128         if os.path.isdir(path):
 129             for r, dirs, files in os.walk(path):
 130                 if files: dbg("Files %s" % ", ".join(files))
 131                 if dirs: dbg("Dirs  %s" % ", ".join(dirs))
 132                 prog.update(walker([os.path.join(r, f) for f in files]))
 133                 prog.update(walker([os.path.join(r, d) for d in dirs]))
 134         else:
 135             if path.endswith('.a'):
 136                 if ar_x(path)[1] is not None: continue # extract worked
 137                 prog.update(walker(ar_t(path)))
 138             if not path.endswith('.o'): continue
 139             dbg("Reading symbols from %s" % (path))
 140             prog[os.path.normpath(path)] = get_symbols(path)
 141     return prog
 142
 143 def resolve(prog):
 144     x = prog.keys()
 145     use = set()
 146     # for each unique pair of different files
 147     for (f, g) in ((f,g) for f in x for g in x if f != g):
 148         refs = set()
 149         # for each defined symbol
 150         for s in (s for s in prog[f] if prog[f][s].get('def') and s in prog[g]):
 151             if prog[g][s].get('use'):
 152                 refs.add(s)
 153         for s in refs:
 154             # Prune externally referenced symbols as speed optimization only
 155             for i in (i for i in x if s in prog[i]): del prog[i][s]
 156         use |= refs
 157     return use
 158
 159 try:
 160     oprog = walker(odir)
 161     if tmpd is not None:
 162         oprog.update(walker([tmpd]))
 163     oused = resolve(oprog)
 164 finally:
 165     try:
 166         p0 = Popen(["rm", "-r", "-f", "%s" % (tmpd)], stderr=PIPE, stdout=PIPE)
 167         p0.communicate()
 168         if p0.returncode != 0: raise "rm '%s' didn't work out" % (tmpd)
 169     except:
 170         from shutil import rmtree
 171         rmtree(tmpd, ignore_errors=True)
 172
 173 for (i,s) in ((i,s) for i in oprog.keys() for s in oprog[i] if oprog[i][s]):
 174     if oprog[i][s].get('def') and not oprog[i][s].get('use'):
 175         print("%s: Symbol '%s' declared extern but never referenced externally"
 176             % (i,s))
 177
 178