bin/find-mergedlib-can-be-private.py

   1 #!/usr/bin/python3
   2 #
   3 # Generate a custom linker script/map file for the --enabled-mergedlibs merged library
   4 # which reduces the startup time and enables further optimisations with --enable-lto because 60% or more
   5 # of the symbols become internal only.
   6 #
   7
   8 import subprocess
   9 import re
  10 import multiprocessing
  11
  12 exported_symbols = set()
  13 imported_symbols = set()
  14
  15
  16 # look for symbols exported by libmerged
  17 subprocess_nm = subprocess.Popen("nm -D instdir/program/libmergedlo.so", stdout=subprocess.PIPE, shell=True)
  18 with subprocess_nm.stdout as txt:
  19     # We are looking for lines something like:
  20     # 0000000000036ed0 T flash_component_getFactory
  21     line_regex = re.compile(r'^[0-9a-fA-F]+ T ')
  22     for line in txt:
  23         line = line.strip().decode("utf-8")
  24         if line_regex.match(line):
  25             exported_symbols.add(line.split(" ")[2])
  26 subprocess_nm.terminate()
  27
  28 # look for symbols imported from libmerged
  29 subprocess_find = subprocess.Popen("(find instdir/program/ -type f; ls ./workdir/LinkTarget/CppunitTest/*.so) | xargs grep -l mergedlo",
  30         stdout=subprocess.PIPE, shell=True)
  31 with subprocess_find.stdout as txt:
  32     for line in txt:
  33         sharedlib = line.strip().decode("utf-8")
  34         s = sharedlib[sharedlib.find("/lib") + 4 : len(sharedlib) - 3]
  35         # look for imported symbols
  36         subprocess_objdump = subprocess.Popen("objdump -T " + sharedlib, stdout=subprocess.PIPE, shell=True)
  37         with subprocess_objdump.stdout as txt2:
  38             # ignore some header bumpf
  39             txt2.readline()
  40             txt2.readline()
  41             txt2.readline()
  42             txt2.readline()
  43             # We are looking for lines something like (noting that one of them uses spaces, and the other tabs)
  44             # 0000000000000000      DF *UND*  0000000000000000     _ZN16FilterConfigItem10WriteInt32ERKN3rtl8OUStringEi
  45             for line2 in txt2:
  46                 line2 = line2.strip().decode("utf-8")
  47                 if line2.find("*UND*") == -1: continue
  48                 tokens = line2.split(" ")
  49                 sym = tokens[len(tokens)-1].strip()
  50                 imported_symbols.add(sym)
  51         subprocess_objdump.terminate()
  52 subprocess_find.terminate()
  53
  54 intersec_symbols = exported_symbols.intersection(imported_symbols)
  55 print("no symbols exported from libmerged   = " + str(len(exported_symbols)))
  56 print("no symbols that can be made internal = " + str(len(intersec_symbols)))
  57
  58 # Now look for classes where none of the class symbols are imported,
  59 # i.e. we can mark the whole class as hidden
  60
  61 def extract_class(sym):
  62     filtered_sym = subprocess.check_output(["c++filt", sym]).strip().decode("utf-8")
  63     if filtered_sym.startswith("vtable for "):
  64         classname = filtered_sym[11:]
  65         return classname
  66     if filtered_sym.startswith("non-virtual thunk to "):
  67         filtered_sym = filtered_sym[21:]
  68     elif filtered_sym.startswith("virtual thunk to "):
  69         filtered_sym = filtered_sym[17:]
  70     i = filtered_sym.find("(")
  71     if i != -1:
  72         i = filtered_sym.rfind("::", 0, i)
  73         if i != -1:
  74             classname = filtered_sym[:i]
  75             return classname
  76     return ""
  77
  78 pool = multiprocessing.Pool(multiprocessing.cpu_count())
  79 classes_with_exported_symbols = set(pool.map(extract_class, list(exported_symbols)))
  80 classes_with_imported_symbols = set(pool.map(extract_class, list(imported_symbols)))
  81
  82 # Some stuff is particular to Windows, so won't be found by a Linux analysis, so remove
  83 # those classes.
  84 can_be_private_classes = classes_with_exported_symbols - classes_with_imported_symbols;
  85 can_be_private_classes.discard("SpinField")
  86
  87 with open("bin/find-mergedlib-can-be-private.classes.results", "wt") as f:
  88     for sym in sorted(can_be_private_classes):
  89         if sym.startswith("std::") or sym.startswith("void std::"): continue
  90         f.write(sym + "\n")