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

   1 #!/usr/bin/python2
   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 sys
  10 import re
  11 import multiprocessing
  12
  13 exported_symbols = set()
  14 imported_symbols = set()
  15
  16
  17 # Copied from solenv/gbuild/extensions/pre_MergedLibsList.mk
  18 # TODO there has to be a way to run gmake and get it to dump this list for me
  19 merged_libs = { \
  20     "avmedia" \
  21     ,"basctl" \
  22     ,"basprov" \
  23     ,"basegfx" \
  24     ,"canvasfactory" \
  25     ,"canvastools" \
  26     ,"comphelper" \
  27     ,"configmgr" \
  28     ,"cppcanvas" \
  29     ,"crashreport)" \
  30     ,"dbtools" \
  31     ,"deployment" \
  32     ,"deploymentmisc" \
  33     ,"desktopbe1)" \
  34     ,"desktop_detector)" \
  35     ,"drawinglayer" \
  36     ,"editeng" \
  37     ,"expwrap" \
  38     ,"filterconfig" \
  39     ,"fsstorage" \
  40     ,"fwk" \
  41     ,"helplinker)" \
  42     ,"i18npool" \
  43     ,"i18nutil" \
  44     ,"lng" \
  45     ,"localebe1" \
  46     ,"msfilter" \
  47     ,"mtfrenderer" \
  48     ,"opencl" \
  49     ,"package2" \
  50     ,"sax" \
  51     ,"sb" \
  52     ,"simplecanvas" \
  53     ,"sfx" \
  54     ,"sofficeapp" \
  55     ,"sot" \
  56     ,"spl" \
  57     ,"stringresource" \
  58     ,"svl" \
  59     ,"svt" \
  60     ,"svx" \
  61     ,"svxcore" \
  62     ,"tk" \
  63     ,"tl" \
  64     ,"ucb1" \
  65     ,"ucbhelper" \
  66     ,"ucpexpand1" \
  67     ,"ucpfile1" \
  68     ,"unoxml" \
  69     ,"utl" \
  70     ,"uui" \
  71     ,"vcl" \
  72     ,"xmlscript" \
  73     ,"xo" \
  74     ,"xstor" }
  75
  76 # look for symbols exported by libmerged
  77 subprocess_nm = subprocess.Popen("nm -D instdir/program/libmergedlo.so", stdout=subprocess.PIPE, shell=True)
  78 with subprocess_nm.stdout as txt:
  79     # We are looking for lines something like:
  80     # 0000000000036ed0 T flash_component_getFactory
  81     line_regex = re.compile(r'^[0-9a-fA-F]+ T ')
  82     for line in txt:
  83         line = line.strip()
  84         if line_regex.match(line):
  85             exported_symbols.add(line.split(" ")[2])
  86 subprocess_nm.terminate()
  87
  88 # look for symbols imported from libmerged
  89 subprocess_find = subprocess.Popen("(find instdir/program/ -type f; ls ./workdir/LinkTarget/CppunitTest/*.so) | xargs grep -l mergedlo",
  90         stdout=subprocess.PIPE, shell=True)
  91 with subprocess_find.stdout as txt:
  92     for line in txt:
  93         sharedlib = line.strip()
  94         s = sharedlib[sharedlib.find("/lib") + 4 : len(sharedlib) - 3]
  95         if s in merged_libs: continue
  96         # look for imported symbols
  97         subprocess_objdump = subprocess.Popen("objdump -T " + sharedlib, stdout=subprocess.PIPE, shell=True)
  98         with subprocess_objdump.stdout as txt2:
  99             # ignore some header bumpf
 100             txt2.readline()
 101             txt2.readline()
 102             txt2.readline()
 103             txt2.readline()
 104             # We are looking for lines something like (noting that one of them uses spaces, and the other tabs)
 105             # 0000000000000000      DF *UND*  0000000000000000     _ZN16FilterConfigItem10WriteInt32ERKN3rtl8OUStringEi
 106             for line2 in txt2:
 107                 line2 = line2.strip()
 108                 if line2.find("*UND*") == -1: continue
 109                 tokens = line2.split(" ")
 110                 sym = tokens[len(tokens)-1].strip()
 111                 imported_symbols.add(sym)
 112         subprocess_objdump.terminate()
 113 subprocess_find.terminate()
 114
 115 intersec_symbols = exported_symbols.intersection(imported_symbols)
 116 print("no symbols exported from libmerged   = " + str(len(exported_symbols)))
 117 print("no symbols that can be made internal = " + str(len(intersec_symbols)))
 118
 119 # Now look for classes where none of the class symbols are imported,
 120 # i.e. we can mark the whole class as hidden
 121
 122 def extract_class(sym):
 123     filtered_sym = subprocess.check_output(["c++filt", sym]).strip()
 124     if filtered_sym.startswith("vtable for "):
 125         classname = filtered_sym[11:]
 126         return classname
 127     if filtered_sym.startswith("non-virtual thunk to "):
 128         filtered_sym = filtered_sym[21:]
 129     elif filtered_sym.startswith("virtual thunk to "):
 130         filtered_sym = filtered_sym[17:]
 131     i = filtered_sym.find("(")
 132     if i != -1:
 133         i = filtered_sym.rfind("::", 0, i)
 134         if i != -1:
 135             classname = filtered_sym[:i]
 136             return classname
 137     return ""
 138
 139 pool = multiprocessing.Pool(multiprocessing.cpu_count())
 140 classes_with_exported_symbols = set(pool.map(extract_class, list(exported_symbols)))
 141 classes_with_imported_symbols = set(pool.map(extract_class, list(imported_symbols)))
 142
 143 # Some stuff is particular to Windows, so won't be found by a Linux analysis, so remove
 144 # those classes.
 145 can_be_private_classes = classes_with_exported_symbols - classes_with_imported_symbols;
 146 can_be_private_classes.discard("SpinField")
 147
 148 with open("bin/find-mergedlib-can-be-private.classes.results", "wt") as f:
 149     for sym in sorted(can_be_private_classes):
 150         if sym.startswith("std::") or sym.startswith("void std::"): continue
 151         f.write(sym + "\n")