import tempfile import os from os import path import sys import re import subprocess from contextlib import contextmanager from collections import defaultdict import colorsys import cxxfilt from elftools.elf.elffile import ELFFile from elftools.elf.enums import ENUM_ST_SHNDX from elftools.elf.descriptions import describe_symbol_type, describe_sh_type import libarchive import matplotlib.cm @contextmanager def chdir(newdir): old_cwd = os.getcwd() try: os.chdir(newdir) yield finally: os.chdir(old_cwd) def trace_source_files(linker, cmdline, trace_sections=[]): with tempfile.TemporaryDirectory() as tempdir: out_path = path.join(tempdir, 'output.elf') output = subprocess.check_output([linker, '-o', out_path, f'-Wl,--cref', *cmdline]) lines = [ line.strip() for line in output.decode().splitlines() ] # FIXME also find isr vector table references defs = {} for line in lines[lines.index('Cross Reference Table')+3:]: try: *left, right = line.split() if left: defs[' '.join(left)] = right except: pass refs = defaultdict(lambda: set()) syms = {} for sym, obj in defs.items(): fn, _, member = re.match('^([^()]+)(\((.+)\))?$', obj).groups() fn = path.abspath(fn) if member: subprocess.check_call(['ar', 'x', '--output', tempdir, fn, member]) fn = path.join(tempdir, member) with open(fn, 'rb') as f: elf = ELFFile(f) symtab = elf.get_section_by_name('.symtab') symtab_demangled = { cxxfilt.demangle(nsym.name).replace(' ', ''): i for i, nsym in enumerate(symtab.iter_symbols()) } def lookup_size(name): name_normalized = name.replace(' ', '') if name_normalized in symtab_demangled: entry = symtab.get_symbol(symtab_demangled[name_normalized]) index = entry['st_shndx'] if index in ENUM_ST_SHNDX: return 'no bits' sec = elf.get_section(index) if describe_sh_type(sec['sh_type']) == 'NOBITS': return 'no bits' return entry['st_size'] else: return None syms[sym] = fn, lookup_size(sym) s = set() sec_map = { sec.name: i for i, sec in enumerate(elf.iter_sections()) } matches = [ i for name, i in sec_map.items() if re.match(f'\.rel\..*\.{sym}', name) ] if matches: sec = elf.get_section(matches[0]) for reloc in sec.iter_relocations(): refsym = symtab.get_symbol(reloc['r_info_sym']) name = refsym.name if refsym.name else elf.get_section(refsym['st_shndx']).name.split('.')[-1] s.add(name) if name not in defs: syms[name] = fn, lookup_size(name) refs[sym] = s for tsec in trace_sections: matches = [ i for name, i in sec_map.items() if name == f'.rel{tsec}' ] s = set() if matches: sec = elf.get_section(matches[0]) for reloc in sec.iter_relocations(): refsym = symtab.get_symbol(reloc['r_info_sym']) s.add(refsym.name) refs[tsec.replace('.', '_')] |= s syms_out = set() with open(out_path, 'rb') as f: elf = ELFFile(f) symtab = elf.get_section_by_name('.symtab') for sym in symtab.iter_symbols(): if describe_symbol_type(sym['st_info']['type']) in ('FUNC', 'OBJECT'): syms_out.add(sym.name) #for sym in defs: # entry = symtab.get_symbol_by_name(sym) # if entry is None: # syms[sym] = defs[sym], None # else: # syms[sym] = defs[sym], entry[0]['st_size'] return syms, refs, syms_out @contextmanager def wrap(leader='', print=print, left='{', right='}'): print(leader, left) yield lambda *args, **kwargs: print(' ', *args, **kwargs) print(right) def mangle(name): return re.sub('[^a-zA-Z0-9_]', '_', name) hexcolor = lambda r, g, b, *_a: f'#{int(r*255):02x}{int(g*255):02x}{int(b*255):02x}' def vhex(val): r,g,b,_a = matplotlib.cm.viridis(1.0-val) fc = hexcolor(r, g, b) h,s,v = colorsys.rgb_to_hsv(r,g,b) cc = '#000000' if v > 0.8 else '#ffffff' return fc, cc if __name__ == '__main__': import argparse parser = argparse.ArgumentParser() parser.add_argument('--trace-sections', type=str, action='append', default=[]) parser.add_argument('--trim-stubs', type=str, action='append', default=[]) parser.add_argument('--highlight-subdirs', type=str, default=None) parser.add_argument('linker_binary') parser.add_argument('linker_args', nargs=argparse.REMAINDER) args = parser.parse_args() trace_sections = args.trace_sections trace_sections_mangled = { sec.replace('.', '_') for sec in trace_sections } syms, refs, syms_out = trace_source_files(args.linker_binary, args.linker_args, trace_sections) for name, (obj, size) in syms.items(): if path.basename(obj) in args.trim_stubs and (not isinstance(size, int) or size <= 8) and not refs.get(name): syms_out.discard(name) clusters = defaultdict(lambda: []) for sym, (obj, size) in syms.items(): if sym in syms_out: clusters[obj].append((sym, size)) obj_size = defaultdict(lambda: 0) for sym, (obj, size) in syms.items(): if isinstance(size, int) and sym in syms_out: obj_size[obj] += size max_size = max([ size for _obj, size in syms.values() if isinstance(size, int) ]) max_osize = max(obj_size.values()) subdir_prefix = path.abspath(args.highlight_subdirs) + '/' if args.highlight_subdirs else '### NO HIGHLIGHT ###' first_comp = lambda le_path: path.dirname(le_path).partition(os.sep)[0] subdir_colors = sorted({ first_comp(obj[len(subdir_prefix):]) for obj in obj_size if obj.startswith(subdir_prefix) }) subdir_colors = { path: hexcolor(*matplotlib.cm.Pastel1(i/len(subdir_colors))) for i, path in enumerate(subdir_colors) } def lookup_highlight(path): if args.highlight_subdirs: if obj.startswith(subdir_prefix): highlight_head = first_comp(path[len(subdir_prefix):]) return subdir_colors[highlight_head], highlight_head else: return '#e0e0e0', None else: return '#ddf7f4', None with wrap('digraph G', print) as lvl1print: print('rankdir=LR;') print('ranksep=1;') print('nodesep=0.2;') print() for i, (obj, syms) in enumerate(clusters.items()): with wrap(f'subgraph cluster_{i}', lvl1print) as lvl2print: print('style = "filled";') highlight_color, highlight_head = lookup_highlight(obj) print(f'bgcolor = "{highlight_color}";') print('pencolor = none;') fc, cc = vhex(obj_size[obj]/max_osize) highlight_subdir_part = f'{highlight_head} / ' if highlight_head else '' lvl2print(f'label = <
' f'{highlight_subdir_part}' f'{path.basename(obj)} ({obj_size[obj]}B)' f'
>;') lvl2print() for sym, size in syms: if sym in syms_out: size_s = f'{size}B' if isinstance(size, int) else size fc, cc = vhex(size/max_size if isinstance(size, int) else 0) lvl2print(f'{mangle(sym)}[label = "{sym} ({size_s})", style="rounded,filled", shape="box", fillcolor="{fc}", fontname="carlito", fontcolor="{cc}" color=none];') lvl1print() for start, ends in refs.items(): for end in ends: if end and (start in syms_out or start in trace_sections_mangled) and end in syms_out: lvl1print(f'{mangle(start)} -> {mangle(end)} [style="bold", color="#333333"];') for sec in trace_sections: lvl1print(f'{sec.replace(".", "_")} [label = "section {sec}", shape="box", style="filled,bold"];')