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