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 # Copied from solenv/gbuild/extensions/pre_MergedLibsList.mk
  17 # TODO there has to be a way to run gmake and get it to dump this list for me
  18 merged_libs = { \
  19     "avmedia" \
  20     ,"basctl" \
  21     ,"basprov" \
  22     ,"basegfx" \
  23     ,"canvasfactory" \
  24     ,"canvastools" \
  25     ,"comphelper" \
  26     ,"configmgr" \
  27     ,"cppcanvas" \
  28     ,"crashreport)" \
  29     ,"dbtools" \
  30     ,"deployment" \
  31     ,"deploymentmisc" \
  32     ,"desktopbe1)" \
  33     ,"desktop_detector)" \
  34     ,"drawinglayer" \
  35     ,"editeng" \
  36     ,"filterconfig" \
  37     ,"fsstorage" \
  38     ,"fwk" \
  39     ,"helplinker)" \
  40     ,"i18npool" \
  41     ,"i18nutil" \
  42     ,"lng" \
  43     ,"localebe1" \
  44     ,"msfilter" \
  45     ,"mtfrenderer" \
  46     ,"opencl" \
  47     ,"package2" \
  48     ,"sax" \
  49     ,"sb" \
  50     ,"simplecanvas" \
  51     ,"sfx" \
  52     ,"sofficeapp" \
  53     ,"sot" \
  54     ,"spl" \
  55     ,"stringresource" \
  56     ,"svl" \
  57     ,"svt" \
  58     ,"svx" \
  59     ,"svxcore" \
  60     ,"tk" \
  61     ,"tl" \
  62     ,"ucb1" \
  63     ,"ucbhelper" \
  64     ,"ucpexpand1" \
  65     ,"ucpfile1" \
  66     ,"unoxml" \
  67     ,"utl" \
  68     ,"uui" \
  69     ,"vcl" \
  70     ,"xmlscript" \
  71     ,"xo" \
  72     ,"xstor" }
  73
  74 # look for symbols exported by libmerged
  75 subprocess_nm = subprocess.Popen("nm -D instdir/program/libmergedlo.so", stdout=subprocess.PIPE, shell=True)
  76 with subprocess_nm.stdout as txt:
  77     # We are looking for lines something like:
  78     # 0000000000036ed0 T flash_component_getFactory
  79     line_regex = re.compile(r'^[0-9a-fA-F]+ T ')
  80     for line in txt:
  81         line = line.strip().decode("utf-8")
  82         if line_regex.match(line):
  83             exported_symbols.add(line.split(" ")[2])
  84 subprocess_nm.terminate()
  85
  86 # look for symbols imported from libmerged
  87 subprocess_find = subprocess.Popen("(find instdir/program/ -type f; ls ./workdir/LinkTarget/CppunitTest/*.so) | xargs grep -l mergedlo",
  88         stdout=subprocess.PIPE, shell=True)
  89 with subprocess_find.stdout as txt:
  90     for line in txt:
  91         sharedlib = line.strip().decode("utf-8")
  92         s = sharedlib[sharedlib.find("/lib") + 4 : len(sharedlib) - 3]
  93         if s in merged_libs: continue
  94         # look for imported symbols
  95         subprocess_objdump = subprocess.Popen("objdump -T " + sharedlib, stdout=subprocess.PIPE, shell=True)
  96         with subprocess_objdump.stdout as txt2:
  97             # ignore some header bumpf
  98             txt2.readline()
  99             txt2.readline()
 100             txt2.readline()
 101             txt2.readline()
 102             # We are looking for lines something like (noting that one of them uses spaces, and the other tabs)
 103             # 0000000000000000      DF *UND*  0000000000000000     _ZN16FilterConfigItem10WriteInt32ERKN3rtl8OUStringEi
 104             for line2 in txt2:
 105                 line2 = line2.strip().decode("utf-8")
 106                 if line2.find("*UND*") == -1: continue
 107                 tokens = line2.split(" ")
 108                 sym = tokens[len(tokens)-1].strip()
 109                 imported_symbols.add(sym)
 110         subprocess_objdump.terminate()
 111 subprocess_find.terminate()
 112
 113 intersec_symbols = exported_symbols.intersection(imported_symbols)
 114 print("no symbols exported from libmerged   = " + str(len(exported_symbols)))
 115 print("no symbols that can be made internal = " + str(len(intersec_symbols)))
 116
 117 # Now look for classes where none of the class symbols are imported,
 118 # i.e. we can mark the whole class as hidden
 119
 120 def extract_class(sym):
 121     filtered_sym = subprocess.check_output(["c++filt", sym]).strip().decode("utf-8")
 122     if filtered_sym.startswith("vtable for "):
 123         classname = filtered_sym[11:]
 124         return classname
 125     if filtered_sym.startswith("non-virtual thunk to "):
 126         filtered_sym = filtered_sym[21:]
 127     elif filtered_sym.startswith("virtual thunk to "):
 128         filtered_sym = filtered_sym[17:]
 129     i = filtered_sym.find("(")
 130     if i != -1:
 131         i = filtered_sym.rfind("::", 0, i)
 132         if i != -1:
 133             classname = filtered_sym[:i]
 134             return classname
 135     return ""
 136
 137 pool = multiprocessing.Pool(multiprocessing.cpu_count())
 138 classes_with_exported_symbols = set(pool.map(extract_class, list(exported_symbols)))
 139 classes_with_imported_symbols = set(pool.map(extract_class, list(imported_symbols)))
 140
 141 # Some stuff is particular to Windows, so won't be found by a Linux analysis, so remove
 142 # those classes.
 143 can_be_private_classes = classes_with_exported_symbols - classes_with_imported_symbols;
 144 can_be_private_classes.discard("SpinField")
 145
 146 with open("bin/find-mergedlib-can-be-private.classes.results", "wt") as f:
 147     for sym in sorted(can_be_private_classes):
 148         if sym.startswith("std::") or sym.startswith("void std::"): continue
 149         f.write(sym + "\n")