diff options
Diffstat (limited to 'tools/rb')
| -rw-r--r-- | tools/rb/README | 7 | ||||
| -rwxr-xr-x | tools/rb/filter-log.pl | 44 | ||||
| -rwxr-xr-x | tools/rb/find-comptr-leakers.pl | 114 | ||||
| -rwxr-xr-x | tools/rb/find_leakers.py | 100 | ||||
| -rwxr-xr-x | tools/rb/fix_linux_stack.py | 317 | ||||
| -rwxr-xr-x | tools/rb/fix_macosx_stack.py | 133 | ||||
| -rwxr-xr-x | tools/rb/fix_stack_using_bpsyms.py | 163 | ||||
| -rwxr-xr-x | tools/rb/make-tree.pl | 303 |
8 files changed, 1181 insertions, 0 deletions
diff --git a/tools/rb/README b/tools/rb/README new file mode 100644 index 000000000..c9b5c282c --- /dev/null +++ b/tools/rb/README @@ -0,0 +1,7 @@ +This is the Refcount Balancer. See +https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Refcount_tracing_and_balancing +for documentation. + +Previous CVS history for the perl scripts is available at: +http://www.mozilla.org/webtools/bonsai/cvslog.cgi?file=mozilla-org/html/performance/find-leakers.pl&rev=&root=/cvsroot/ +http://www.mozilla.org/webtools/bonsai/cvslog.cgi?file=mozilla-org/html/performance/make-tree.pl&rev=&root=/cvsroot/ diff --git a/tools/rb/filter-log.pl b/tools/rb/filter-log.pl new file mode 100755 index 000000000..4a1f66741 --- /dev/null +++ b/tools/rb/filter-log.pl @@ -0,0 +1,44 @@ +#!/usr/bin/perl -w +# +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +# Filter a refcount log to show only the entries for a single object. +# Useful when manually examining refcount logs containing multiple +# objects. + +use 5.004; +use strict; +use Getopt::Long; + +GetOptions("object=s"); + +$::opt_object || + die qq{ +usage: filter-log-for.pl < logfile + --object <obj> The address of the object to examine (required) +}; + +warn "object $::opt_object\n"; + +LINE: while (<>) { + next LINE if (! /^</); + my $line = $_; + my @fields = split(/ /, $_); + + my $class = shift(@fields); + my $obj = shift(@fields); + next LINE unless ($obj eq $::opt_object); + my $sno = shift(@fields); + my $op = shift(@fields); + my $cnt = shift(@fields); + + print $line; + + # The lines in the stack trace + CALLSITE: while (<>) { + print; + last CALLSITE if (/^$/); + } +} diff --git a/tools/rb/find-comptr-leakers.pl b/tools/rb/find-comptr-leakers.pl new file mode 100755 index 000000000..925119935 --- /dev/null +++ b/tools/rb/find-comptr-leakers.pl @@ -0,0 +1,114 @@ +#!/usr/bin/perl -w +# +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +# Script loosely based on Chris Waterson's find-leakers.pl and make-tree.pl + +use 5.004; +use strict; +use Getopt::Long; + +# GetOption will create $opt_object, so ignore the +# warning that gets spit out about those vbls. +GetOptions("object=s", "list", "help"); + +# use $::opt_help twice to eliminate warning... +($::opt_help) && ($::opt_help) && die qq{ +usage: find-comptr-leakers.pl < logfile + --object <obj> Examine only object <obj> + --list Only list leaked objects + --help This message :-) +}; + +if ($::opt_object) { + warn "Examining only object $::opt_object (THIS IS BROKEN)\n"; +} else { + warn "Examining all objects\n"; +} + +my %allocs = ( ); +my %counter; +my $id = 0; + +my $accumulating = 0; +my $savedata = 0; +my $class; +my $obj; +my $sno; +my $op; +my $cnt; +my $ptr; +my $strace; + +sub save_data { + # save the data + if ($op eq 'nsCOMPtrAddRef') { + push @{ $allocs{$sno}->{$ptr} }, [ +1, $strace ]; + } + elsif ($op eq 'nsCOMPtrRelease') { + push @{ $allocs{$sno}->{$ptr} }, [ -1, $strace ]; + my $sum = 0; + my @ptrallocs = @{ $allocs{$sno}->{$ptr} }; + foreach my $alloc (@ptrallocs) { + $sum += @$alloc[0]; + } + if ( $sum == 0 ) { + delete($allocs{$sno}{$ptr}); + } + } +} + +LINE: while (<>) { + if (/^</) { + chop; # avoid \n in $ptr + my @fields = split(/ /, $_); + + ($class, $obj, $sno, $op, $cnt, $ptr) = @fields; + + $strace = ""; + + if ($::opt_list) { + save_data(); + } elsif (!($::opt_object) || ($::opt_object eq $obj)) { + $accumulating = 1; + } + } elsif ( $accumulating == 1 ) { + if ( /^$/ ) { + # if line is empty + $accumulating = 0; + save_data(); + } else { + $strace = $strace . $_; + } + } +} +if ( $accumulating == 1) { + save_data(); +} + +foreach my $serial (keys(%allocs)) { + foreach my $comptr (keys( %{$allocs{$serial}} )) { + my $sum = 0; + my @ptrallocs = @{ $allocs{$serial}->{$comptr} }; + foreach my $alloc (@ptrallocs) { + $sum += @$alloc[0]; + } + print "Object ", $serial, " held by ", $comptr, " is ", $sum, " out of balance.\n"; + unless ($::opt_list) { + print "\n"; + foreach my $alloc (@ptrallocs) { + if (@$alloc[0] == +1) { + print "Put into nsCOMPtr at:\n"; + } elsif (@$alloc[0] == -1) { + print "Released from nsCOMPtr at:\n"; + } + print @$alloc[1]; # the stack trace + print "\n"; + } + print "\n\n"; + } + } +} + diff --git a/tools/rb/find_leakers.py b/tools/rb/find_leakers.py new file mode 100755 index 000000000..4405d7a17 --- /dev/null +++ b/tools/rb/find_leakers.py @@ -0,0 +1,100 @@ +#!/usr/bin/python +# +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +# This script processes a `refcount' log, and finds out if any object leaked. +# It simply goes through the log, finds `AddRef' or `Ctor' lines, and then +# sees if they `Release' or `Dtor'. If not, it reports them as leaks. +# Please see README file in the same directory. + + +import sys + +def print_output(allocation, obj_to_class): + '''Formats and prints output.''' + items = [] + for obj, count, in allocation.iteritems(): + # Adding items to a list, so we can sort them. + items.append((obj, count)) + # Sorting by count. + items.sort(key=lambda item: item[1]) + + for obj, count, in items: + print "{obj} ({count}) @ {class_name}".format(obj=obj, + count=count, + class_name=obj_to_class[obj]) + +def process_log(log_lines): + '''Process through the log lines, and print out the result. + + @param log_lines: List of strings. + ''' + allocation = {} + class_count = {} + obj_to_class = {} + + for log_line in log_lines: + if not log_line.startswith('<'): + continue + + (class_name, + obj, + ignore, + operation, + count,) = log_line.strip('\r\n').split(' ')[:5] + + # for AddRef/Release `count' is the refcount, + # for Ctor/Dtor it's the size. + + if ((operation == 'AddRef' and count == '1') or + operation == 'Ctor'): + # Examples: + # <nsStringBuffer> 0x01AFD3B8 1 AddRef 1 + # <PStreamNotifyParent> 0x08880BD0 8 Ctor (20) + class_count[class_name] = class_count.setdefault(class_name, 0) + 1 + allocation[obj] = class_count[class_name] + obj_to_class[obj] = class_name + + elif ((operation == 'Release' and count == '0') or + operation == 'Dtor'): + # Examples: + # <nsStringBuffer> 0x01AFD3B8 1 Release 0 + # <PStreamNotifyParent> 0x08880BD0 8 Dtor (20) + if obj not in allocation: + print "An object was released that wasn't allocated!", + print obj, "@", class_name + else: + allocation.pop(obj) + obj_to_class.pop(obj) + + # Printing out the result. + print_output(allocation, obj_to_class) + + +def print_usage(): + print + print "Usage: find-leakers.py [log-file]" + print + print "If `log-file' provided, it will read that as the input log." + print "Else, it will read the stdin as the input log." + print + +def main(): + '''Main method of the script.''' + if len(sys.argv) == 1: + # Reading log from stdin. + process_log(sys.stdin.readlines()) + elif len(sys.argv) == 2: + # Reading log from file. + with open(sys.argv[1], 'r') as log_file: + log_lines = log_file.readlines() + process_log(log_lines) + else: + print 'ERROR: Invalid number of arguments' + print_usage() + +if __name__ == '__main__': + main() + diff --git a/tools/rb/fix_linux_stack.py b/tools/rb/fix_linux_stack.py new file mode 100755 index 000000000..bdc8a15dc --- /dev/null +++ b/tools/rb/fix_linux_stack.py @@ -0,0 +1,317 @@ +#!/usr/bin/python +# vim:sw=4:ts=4:et: +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +# This script uses addr2line (part of binutils) to post-process the entries +# produced by NS_FormatCodeAddress(), which on Linux often lack a function +# name, a file name and a line number. + +import subprocess +import sys +import re +import os +import pty +import termios +from StringIO import StringIO + +class unbufferedLineConverter: + """ + Wrap a child process that responds to each line of input with one line of + output. Uses pty to trick the child into providing unbuffered output. + """ + def __init__(self, command, args = []): + pid, fd = pty.fork() + if pid == 0: + # We're the child. Transfer control to command. + os.execvp(command, [command] + args) + else: + # Disable echoing. + attr = termios.tcgetattr(fd) + attr[3] = attr[3] & ~termios.ECHO + termios.tcsetattr(fd, termios.TCSANOW, attr) + # Set up a file()-like interface to the child process + self.r = os.fdopen(fd, "r", 1) + self.w = os.fdopen(os.dup(fd), "w", 1) + def convert(self, line): + self.w.write(line + "\n") + return (self.r.readline().rstrip("\r\n"), self.r.readline().rstrip("\r\n")) + @staticmethod + def test(): + assert unbufferedLineConverter("rev").convert("123") == "321" + assert unbufferedLineConverter("cut", ["-c3"]).convert("abcde") == "c" + print "Pass" + +objdump_section_re = re.compile("^ [0-9a-f]* ([0-9a-f ]{8}) ([0-9a-f ]{8}) ([0-9a-f ]{8}) ([0-9a-f ]{8}).*") +def elf_section(file, section): + """ + Return the requested ELF section of the file as a str, representing + a sequence of bytes. + """ + # We can read the .gnu_debuglink section using either of: + # objdump -s --section=.gnu_debuglink $file + # readelf -x .gnu_debuglink $file + # Since readelf prints things backwards on little-endian platforms + # for some versions only (backwards on Fedora Core 6, forwards on + # Fedora 7), use objdump. + objdump = subprocess.Popen(['objdump', '-s', '--section=' + section, file], + stdout=subprocess.PIPE, + # redirect stderr so errors don't get printed + stderr=subprocess.PIPE) + (objdump_stdout, objdump_stderr) = objdump.communicate() + if objdump.returncode != 0: + return None + result = "" + # Turn hexadecimal dump into the bytes it represents + for line in StringIO(objdump_stdout).readlines(): + m = objdump_section_re.match(line) + if m: + for gnum in [0, 1, 2, 3]: + word = m.groups()[gnum] + if word != " ": + for idx in [0, 2, 4, 6]: + result += chr(int(word[idx:idx+2], 16)) + return result + +# FIXME: Hard-coded to gdb defaults (works on Fedora and Ubuntu). +global_debug_dir = '/usr/lib/debug'; + +endian_re = re.compile("\s*Data:\s+.*(little|big) endian.*$") + +# Table of 256 values, per documentation of .gnu_debuglink sections. +gnu_debuglink_crc32_table = [ + 0x00000000, 0x77073096, 0xee0e612c, 0x990951ba, 0x076dc419, + 0x706af48f, 0xe963a535, 0x9e6495a3, 0x0edb8832, 0x79dcb8a4, + 0xe0d5e91e, 0x97d2d988, 0x09b64c2b, 0x7eb17cbd, 0xe7b82d07, + 0x90bf1d91, 0x1db71064, 0x6ab020f2, 0xf3b97148, 0x84be41de, + 0x1adad47d, 0x6ddde4eb, 0xf4d4b551, 0x83d385c7, 0x136c9856, + 0x646ba8c0, 0xfd62f97a, 0x8a65c9ec, 0x14015c4f, 0x63066cd9, + 0xfa0f3d63, 0x8d080df5, 0x3b6e20c8, 0x4c69105e, 0xd56041e4, + 0xa2677172, 0x3c03e4d1, 0x4b04d447, 0xd20d85fd, 0xa50ab56b, + 0x35b5a8fa, 0x42b2986c, 0xdbbbc9d6, 0xacbcf940, 0x32d86ce3, + 0x45df5c75, 0xdcd60dcf, 0xabd13d59, 0x26d930ac, 0x51de003a, + 0xc8d75180, 0xbfd06116, 0x21b4f4b5, 0x56b3c423, 0xcfba9599, + 0xb8bda50f, 0x2802b89e, 0x5f058808, 0xc60cd9b2, 0xb10be924, + 0x2f6f7c87, 0x58684c11, 0xc1611dab, 0xb6662d3d, 0x76dc4190, + 0x01db7106, 0x98d220bc, 0xefd5102a, 0x71b18589, 0x06b6b51f, + 0x9fbfe4a5, 0xe8b8d433, 0x7807c9a2, 0x0f00f934, 0x9609a88e, + 0xe10e9818, 0x7f6a0dbb, 0x086d3d2d, 0x91646c97, 0xe6635c01, + 0x6b6b51f4, 0x1c6c6162, 0x856530d8, 0xf262004e, 0x6c0695ed, + 0x1b01a57b, 0x8208f4c1, 0xf50fc457, 0x65b0d9c6, 0x12b7e950, + 0x8bbeb8ea, 0xfcb9887c, 0x62dd1ddf, 0x15da2d49, 0x8cd37cf3, + 0xfbd44c65, 0x4db26158, 0x3ab551ce, 0xa3bc0074, 0xd4bb30e2, + 0x4adfa541, 0x3dd895d7, 0xa4d1c46d, 0xd3d6f4fb, 0x4369e96a, + 0x346ed9fc, 0xad678846, 0xda60b8d0, 0x44042d73, 0x33031de5, + 0xaa0a4c5f, 0xdd0d7cc9, 0x5005713c, 0x270241aa, 0xbe0b1010, + 0xc90c2086, 0x5768b525, 0x206f85b3, 0xb966d409, 0xce61e49f, + 0x5edef90e, 0x29d9c998, 0xb0d09822, 0xc7d7a8b4, 0x59b33d17, + 0x2eb40d81, 0xb7bd5c3b, 0xc0ba6cad, 0xedb88320, 0x9abfb3b6, + 0x03b6e20c, 0x74b1d29a, 0xead54739, 0x9dd277af, 0x04db2615, + 0x73dc1683, 0xe3630b12, 0x94643b84, 0x0d6d6a3e, 0x7a6a5aa8, + 0xe40ecf0b, 0x9309ff9d, 0x0a00ae27, 0x7d079eb1, 0xf00f9344, + 0x8708a3d2, 0x1e01f268, 0x6906c2fe, 0xf762575d, 0x806567cb, + 0x196c3671, 0x6e6b06e7, 0xfed41b76, 0x89d32be0, 0x10da7a5a, + 0x67dd4acc, 0xf9b9df6f, 0x8ebeeff9, 0x17b7be43, 0x60b08ed5, + 0xd6d6a3e8, 0xa1d1937e, 0x38d8c2c4, 0x4fdff252, 0xd1bb67f1, + 0xa6bc5767, 0x3fb506dd, 0x48b2364b, 0xd80d2bda, 0xaf0a1b4c, + 0x36034af6, 0x41047a60, 0xdf60efc3, 0xa867df55, 0x316e8eef, + 0x4669be79, 0xcb61b38c, 0xbc66831a, 0x256fd2a0, 0x5268e236, + 0xcc0c7795, 0xbb0b4703, 0x220216b9, 0x5505262f, 0xc5ba3bbe, + 0xb2bd0b28, 0x2bb45a92, 0x5cb36a04, 0xc2d7ffa7, 0xb5d0cf31, + 0x2cd99e8b, 0x5bdeae1d, 0x9b64c2b0, 0xec63f226, 0x756aa39c, + 0x026d930a, 0x9c0906a9, 0xeb0e363f, 0x72076785, 0x05005713, + 0x95bf4a82, 0xe2b87a14, 0x7bb12bae, 0x0cb61b38, 0x92d28e9b, + 0xe5d5be0d, 0x7cdcefb7, 0x0bdbdf21, 0x86d3d2d4, 0xf1d4e242, + 0x68ddb3f8, 0x1fda836e, 0x81be16cd, 0xf6b9265b, 0x6fb077e1, + 0x18b74777, 0x88085ae6, 0xff0f6a70, 0x66063bca, 0x11010b5c, + 0x8f659eff, 0xf862ae69, 0x616bffd3, 0x166ccf45, 0xa00ae278, + 0xd70dd2ee, 0x4e048354, 0x3903b3c2, 0xa7672661, 0xd06016f7, + 0x4969474d, 0x3e6e77db, 0xaed16a4a, 0xd9d65adc, 0x40df0b66, + 0x37d83bf0, 0xa9bcae53, 0xdebb9ec5, 0x47b2cf7f, 0x30b5ffe9, + 0xbdbdf21c, 0xcabac28a, 0x53b39330, 0x24b4a3a6, 0xbad03605, + 0xcdd70693, 0x54de5729, 0x23d967bf, 0xb3667a2e, 0xc4614ab8, + 0x5d681b02, 0x2a6f2b94, 0xb40bbe37, 0xc30c8ea1, 0x5a05df1b, + 0x2d02ef8d +] + +def gnu_debuglink_crc32(stream): + # Note that python treats bitwise operators as though integers have + # an infinite number of bits (and thus such that negative integers + # 1-pad out to infinity). + crc = 0xffffffff + while True: + # Choose to read in 4096 byte chunks. + bytes = stream.read(4096) + if len(bytes) == 0: + break + for byte in bytes: + crc = gnu_debuglink_crc32_table[(crc ^ ord(byte)) & 0xff] ^ (crc >> 8) + return ~crc & 0xffffffff + +def separate_debug_file_for(file): + """ + Finds a separated file with the debug sections for a binary. Such + files are commonly installed by debug packages on linux distros. + Rules for finding them are documented in: + https://sourceware.org/gdb/current/onlinedocs/gdb/Separate-Debug-Files.html + """ + def have_debug_file(debugfile): + return os.path.isfile(debugfile) + + endian = None + readelf = subprocess.Popen(['readelf', '-h', file], + stdout=subprocess.PIPE) + for line in readelf.stdout.readlines(): + m = endian_re.match(line) + if m: + endian = m.groups()[0] + break + readelf.terminate() + if endian is None: + sys.stderr.write("Could not determine endianness of " + file + "\n") + return None + + def word32(s): + if type(s) != str or len(s) != 4: + raise StandardError("expected 4 byte string input") + s = list(s) + if endian == "big": + s.reverse() + return sum(map(lambda idx: ord(s[idx]) * (256 ** idx), range(0, 4))) + + buildid = elf_section(file, ".note.gnu.build-id"); + if buildid is not None: + # The build ID is an ELF note section, so it begins with a + # name size (4), a description size (size of contents), a + # type (3), and the name "GNU\0". + note_header = buildid[0:16] + buildid = buildid[16:] + if word32(note_header[0:4]) != 4 or \ + word32(note_header[4:8]) != len(buildid) or \ + word32(note_header[8:12]) != 3 or \ + note_header[12:16] != "GNU\0": + sys.stderr.write("malformed .note.gnu.build_id in " + file + "\n") + else: + buildid = "".join(map(lambda ch: "%02X" % ord(ch), buildid)).lower() + f = os.path.join(global_debug_dir, ".build-id", buildid[0:2], buildid[2:] + ".debug") + if have_debug_file(f): + return f + + debuglink = elf_section(file, ".gnu_debuglink"); + if debuglink is not None: + # The debuglink section contains a string, ending with a + # null-terminator and then 0 to three bytes of padding to fill the + # current 32-bit unit. (This padding is usually null bytes, but + # I've seen null-null-H, on Ubuntu x86_64.) This is followed by + # a 4-byte CRC. + debuglink_name = debuglink[:-4] + null_idx = debuglink_name.find("\0") + if null_idx == -1 or null_idx + 4 < len(debuglink_name): + sys.stderr.write("Malformed .gnu_debuglink in " + file + "\n") + return None + debuglink_name = debuglink_name[0:null_idx] + + debuglink_crc = word32(debuglink[-4:]) + + dirname = os.path.dirname(file) + possible_files = [ + os.path.join(dirname, debuglink_name), + os.path.join(dirname, ".debug", debuglink_name), + os.path.join(global_debug_dir, dirname.lstrip("/"), debuglink_name) + ] + for f in possible_files: + if have_debug_file(f): + fio = open(f, mode="r") + file_crc = gnu_debuglink_crc32(fio) + fio.close() + if file_crc == debuglink_crc: + return f + return None + +elf_type_re = re.compile("^\s*Type:\s+(\S+)") +elf_text_section_re = re.compile("^\s*\[\s*\d+\]\s+\.text\s+\w+\s+(\w+)\s+(\w+)\s+") + +def address_adjustment_for(file): + """ + Return the address adjustment to use for a file. + + addr2line wants offsets relative to the base address for shared + libraries, but it wants addresses including the base address offset + for executables. This returns the appropriate address adjustment to + add to an offset within file. See bug 230336. + """ + readelf = subprocess.Popen(['readelf', '-h', file], + stdout=subprocess.PIPE) + elftype = None + for line in readelf.stdout.readlines(): + m = elf_type_re.match(line) + if m: + elftype = m.groups()[0] + break + readelf.terminate() + + if elftype != "EXEC": + # If we're not dealing with an executable, return 0. + return 0 + + adjustment = 0 + readelf = subprocess.Popen(['readelf', '-S', file], + stdout=subprocess.PIPE) + for line in readelf.stdout.readlines(): + m = elf_text_section_re.match(line) + if m: + # Subtract the .text section's offset within the + # file from its base address. + adjustment = int(m.groups()[0], 16) - int(m.groups()[1], 16); + break + readelf.terminate() + return adjustment + +addr2lines = {} +def addressToSymbol(file, address): + converter = None + address_adjustment = None + cache = None + if not file in addr2lines: + debug_file = separate_debug_file_for(file) or file + converter = unbufferedLineConverter('/usr/bin/addr2line', ['-C', '-f', '-e', debug_file]) + address_adjustment = address_adjustment_for(file) + cache = {} + addr2lines[file] = (converter, address_adjustment, cache) + else: + (converter, address_adjustment, cache) = addr2lines[file] + if address in cache: + return cache[address] + result = converter.convert(hex(int(address, 16) + address_adjustment)) + cache[address] = result + return result + +# Matches lines produced by NS_FormatCodeAddress(). +line_re = re.compile("^(.*#\d+: )(.+)\[(.+) \+(0x[0-9A-Fa-f]+)\](.*)$") + +def fixSymbols(line): + result = line_re.match(line) + if result is not None: + (before, fn, file, address, after) = result.groups() + + if os.path.exists(file) and os.path.isfile(file): + (name, fileline) = addressToSymbol(file, address) + + # If addr2line gave us something useless, keep what we had before. + if name == "??": + name = fn + if fileline == "??:0" or fileline == "??:?": + fileline = file + + nl = '\n' if line[-1] == '\n' else '' + return "%s%s (%s)%s%s" % (before, name, fileline, after, nl) + else: + sys.stderr.write("Warning: File \"" + file + "\" does not exist.\n") + return line + else: + return line + +if __name__ == "__main__": + for line in sys.stdin: + sys.stdout.write(fixSymbols(line)) diff --git a/tools/rb/fix_macosx_stack.py b/tools/rb/fix_macosx_stack.py new file mode 100755 index 000000000..7d076d9b6 --- /dev/null +++ b/tools/rb/fix_macosx_stack.py @@ -0,0 +1,133 @@ +#!/usr/bin/python +# vim:sw=4:ts=4:et: +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +# This script uses |atos| to post-process the entries produced by +# NS_FormatCodeAddress(), which on Mac often lack a file name and a line +# number. + +import subprocess +import sys +import re +import os +import pty +import termios + +class unbufferedLineConverter: + """ + Wrap a child process that responds to each line of input with one line of + output. Uses pty to trick the child into providing unbuffered output. + """ + def __init__(self, command, args = []): + pid, fd = pty.fork() + if pid == 0: + # We're the child. Transfer control to command. + os.execvp(command, [command] + args) + else: + # Disable echoing. + attr = termios.tcgetattr(fd) + attr[3] = attr[3] & ~termios.ECHO + termios.tcsetattr(fd, termios.TCSANOW, attr) + # Set up a file()-like interface to the child process + self.r = os.fdopen(fd, "r", 1) + self.w = os.fdopen(os.dup(fd), "w", 1) + def convert(self, line): + self.w.write(line + "\n") + return self.r.readline().rstrip("\r\n") + @staticmethod + def test(): + assert unbufferedLineConverter("rev").convert("123") == "321" + assert unbufferedLineConverter("cut", ["-c3"]).convert("abcde") == "c" + print "Pass" + +def separate_debug_file_for(file): + return None + +address_adjustments = {} +def address_adjustment(file): + if not file in address_adjustments: + result = None + otool = subprocess.Popen(["otool", "-l", file], stdout=subprocess.PIPE) + while True: + line = otool.stdout.readline() + if line == "": + break + if line == " segname __TEXT\n": + line = otool.stdout.readline() + if not line.startswith(" vmaddr "): + raise StandardError("unexpected otool output") + result = int(line[10:], 16) + break + otool.stdout.close() + + if result is None: + raise StandardError("unexpected otool output") + + address_adjustments[file] = result + + return address_adjustments[file] + +atoses = {} +def addressToSymbol(file, address): + converter = None + if not file in atoses: + debug_file = separate_debug_file_for(file) or file + converter = unbufferedLineConverter('/usr/bin/xcrun', ['atos', '-arch', 'x86_64', '-o', debug_file]) + atoses[file] = converter + else: + converter = atoses[file] + return converter.convert("0x%X" % address) + +cxxfilt_proc = None +def cxxfilt(sym): + if cxxfilt_proc is None: + # --no-strip-underscores because atos already stripped the underscore + globals()["cxxfilt_proc"] = subprocess.Popen(['c++filt', + '--no-strip-underscores', + '--format', 'gnu-v3'], + stdin=subprocess.PIPE, + stdout=subprocess.PIPE) + cxxfilt_proc.stdin.write(sym + "\n") + return cxxfilt_proc.stdout.readline().rstrip("\n") + +# Matches lines produced by NS_FormatCodeAddress(). +line_re = re.compile("^(.*#\d+: )(.+)\[(.+) \+(0x[0-9A-Fa-f]+)\](.*)$") +atos_name_re = re.compile("^(.+) \(in ([^)]+)\) \((.+)\)$") + +def fixSymbols(line): + result = line_re.match(line) + if result is not None: + (before, fn, file, address, after) = result.groups() + address = int(address, 16) + + if os.path.exists(file) and os.path.isfile(file): + address += address_adjustment(file) + info = addressToSymbol(file, address) + + # atos output seems to have three forms: + # address + # address (in foo.dylib) + # symbol (in foo.dylib) (file:line) + name_result = atos_name_re.match(info) + if name_result is not None: + # Print the first two forms as-is, and transform the third + (name, library, fileline) = name_result.groups() + # atos demangles, but occasionally it fails. cxxfilt can mop + # up the remaining cases(!), which will begin with '_Z'. + if (name.startswith("_Z")): + name = cxxfilt(name) + info = "%s (%s, in %s)" % (name, fileline, library) + + nl = '\n' if line[-1] == '\n' else '' + return before + info + after + nl + else: + sys.stderr.write("Warning: File \"" + file + "\" does not exist.\n") + return line + else: + return line + +if __name__ == "__main__": + for line in sys.stdin: + sys.stdout.write(fixSymbols(line)) diff --git a/tools/rb/fix_stack_using_bpsyms.py b/tools/rb/fix_stack_using_bpsyms.py new file mode 100755 index 000000000..5d04cd02a --- /dev/null +++ b/tools/rb/fix_stack_using_bpsyms.py @@ -0,0 +1,163 @@ +#!/usr/bin/env python + +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +# This script uses breakpad symbols to post-process the entries produced by +# NS_FormatCodeAddress(), which on TBPL builds often lack a file name and a +# line number (and on Linux even the symbol is often bad). + +from __future__ import with_statement + +import sys +import os +import re +import subprocess +import bisect + +here = os.path.dirname(__file__) + +def prettyFileName(name): + if name.startswith("../") or name.startswith("..\\"): + # dom_quickstubs.cpp and many .h files show up with relative paths that are useless + # and/or don't correspond to the layout of the source tree. + return os.path.basename(name) + ":" + elif name.startswith("hg:"): + bits = name.split(":") + if len(bits) == 4: + (junk, repo, path, rev) = bits + # We could construct an hgweb URL with /file/ or /annotate/, like this: + # return "http://%s/annotate/%s/%s#l" % (repo, rev, path) + return path + ":" + return name + ":" + +class SymbolFile: + def __init__(self, fn): + addrs = [] # list of addresses, which will be sorted once we're done initializing + funcs = {} # hash: address --> (function name + possible file/line) + files = {} # hash: filenum (string) --> prettified filename ready to have a line number appended + with open(fn) as f: + for line in f: + line = line.rstrip() + # https://chromium.googlesource.com/breakpad/breakpad/+/master/docs/symbol_files.md + if line.startswith("FUNC "): + # FUNC <address> <size> <stack_param_size> <name> + bits = line.split(None, 4) + if len(bits) < 5: + bits.append('unnamed_function') + (junk, rva, size, ss, name) = bits + rva = int(rva,16) + funcs[rva] = name + addrs.append(rva) + lastFuncName = name + elif line.startswith("PUBLIC "): + # PUBLIC <address> <stack_param_size> <name> + (junk, rva, ss, name) = line.split(None, 3) + rva = int(rva,16) + funcs[rva] = name + addrs.append(rva) + elif line.startswith("FILE "): + # FILE <number> <name> + (junk, filenum, name) = line.split(None, 2) + files[filenum] = prettyFileName(name) + elif line[0] in "0123456789abcdef": + # This is one of the "line records" corresponding to the last FUNC record + # <address> <size> <line> <filenum> + (rva, size, line, filenum) = line.split(None) + rva = int(rva,16) + file = files[filenum] + name = lastFuncName + " [" + file + line + "]" + funcs[rva] = name + addrs.append(rva) + # skip everything else + #print "Loaded %d functions from symbol file %s" % (len(funcs), os.path.basename(fn)) + self.addrs = sorted(addrs) + self.funcs = funcs + + def addrToSymbol(self, address): + i = bisect.bisect(self.addrs, address) - 1 + if i > 0: + #offset = address - self.addrs[i] + return self.funcs[self.addrs[i]] + else: + return "" + +def findIdForPath(path): + """Finds the breakpad id for the object file at the given path.""" + # We should always be packaged with a "fileid" executable. + fileid_exe = os.path.join(here, 'fileid') + if not os.path.isfile(fileid_exe): + fileid_exe = fileid_exe + '.exe' + if not os.path.isfile(fileid_exe): + raise Exception("Could not find fileid executable in %s" % here) + + if not os.path.isfile(path): + for suffix in ('.exe', '.dll'): + if os.path.isfile(path + suffix): + path = path + suffix + try: + return subprocess.check_output([fileid_exe, path]).rstrip() + except subprocess.CalledProcessError as e: + raise Exception("Error getting fileid for %s: %s" % + (path, e.output)) + +def guessSymbolFile(full_path, symbolsDir): + """Guess a symbol file based on an object file's basename, ignoring the path and UUID.""" + fn = os.path.basename(full_path) + d1 = os.path.join(symbolsDir, fn) + root, _ = os.path.splitext(fn) + if os.path.exists(os.path.join(symbolsDir, root) + '.pdb'): + d1 = os.path.join(symbolsDir, root) + '.pdb' + fn = root + if not os.path.exists(d1): + return None + uuids = os.listdir(d1) + if len(uuids) == 0: + raise Exception("Missing symbol file for " + fn) + if len(uuids) > 1: + uuid = findIdForPath(full_path) + else: + uuid = uuids[0] + return os.path.join(d1, uuid, fn + ".sym") + +parsedSymbolFiles = {} +def getSymbolFile(file, symbolsDir): + p = None + if not file in parsedSymbolFiles: + symfile = guessSymbolFile(file, symbolsDir) + if symfile: + p = SymbolFile(symfile) + else: + p = None + parsedSymbolFiles[file] = p + else: + p = parsedSymbolFiles[file] + return p + +def addressToSymbol(file, address, symbolsDir): + p = getSymbolFile(file, symbolsDir) + if p: + return p.addrToSymbol(address) + else: + return "" + +# Matches lines produced by NS_FormatCodeAddress(). +line_re = re.compile("^(.*#\d+: )(.+)\[(.+) \+(0x[0-9A-Fa-f]+)\](.*)$") + +def fixSymbols(line, symbolsDir): + result = line_re.match(line) + if result is not None: + (before, fn, file, address, after) = result.groups() + address = int(address, 16) + symbol = addressToSymbol(file, address, symbolsDir) + if not symbol: + symbol = "%s + 0x%x" % (os.path.basename(file), address) + return before + symbol + after + "\n" + else: + return line + +if __name__ == "__main__": + symbolsDir = sys.argv[1] + for line in iter(sys.stdin.readline, ''): + print fixSymbols(line, symbolsDir), diff --git a/tools/rb/make-tree.pl b/tools/rb/make-tree.pl new file mode 100755 index 000000000..04f0d8534 --- /dev/null +++ b/tools/rb/make-tree.pl @@ -0,0 +1,303 @@ +#!/usr/bin/perl -w +# +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +use 5.004; +use strict; +use Getopt::Long; + +$::opt_prune_depth = 0; +$::opt_subtree_size = 0; +$::opt_reverse = 0; + +# GetOption will create $opt_object & $opt_exclude, so ignore the +# warning that gets spit out about those vbls. +GetOptions("object=s", "exclude=s", "comptrs=s", "ignore-balanced", "subtree-size=i", "prune-depth=i", + "collapse-to-method", "collapse-to-class", "old-style", "reverse"); + +$::opt_object || + die qq{ +usage: leak.pl < logfile + --object <obj> The address of the object to examine (required) + --exclude <file> Exclude routines listed in <file> + --comptrs <file> Subtract all the data in the balanced COMPtr log <file> + --ignore-balanced Ignore balanced subtrees + --subtree-size <n> Print subtrees with more than <n> nodes separately + --prune-depth <depth> Prune the tree to <depth> + --collapse-to-method Aggregate data by method + --collapse-to-class Aggregate data by class (subsumes --collapse-to-method) + --reverse Reverse call stacks, showing leaves first + --old-style Old-style formatting +}; + +$::opt_prune_depth = 0 if $::opt_prune_depth < 0; +$::opt_subtree_size = 0 if $::opt_subtree_size < 0; + +warn "object $::opt_object\n"; +warn "ignoring balanced subtrees\n" if $::opt_ignore_balanced; +warn "prune depth $::opt_prune_depth\n" if $::opt_prune_depth; +warn "collapsing to class\n" if $::opt_collapse_to_class; +warn "collapsing to method\n" if $::opt_collapse_to_method && !$::opt_collapse_to_class; +warn "reversing call stacks\n" if $::opt_reverse; + + +# The 'excludes' are functions that, if detected in a particular call +# stack, will cause the _entire_ call stack to be ignored. You might, +# for example, explicitly exclude two functions that have a matching +# AddRef/Release pair. + +my %excludes; + +if ($::opt_exclude) { + open(EXCLUDE, "<".$::opt_exclude) + || die "unable to open $::opt_exclude"; + + while (<EXCLUDE>) { + chomp $_; + warn "excluding $_\n"; + $excludes{$_} = 1; + } +} + +# Each entry in the tree rooted by callGraphRoot contains the following: +# #name# This call's name+offset string +# #refcount# The net reference count of this call +# #label# The label used for this subtree; only defined for labeled nodes +# #children# List of children in alphabetical order +# zero or more children indexed by method name+offset strings. + +my $callGraphRoot; +$callGraphRoot = { '#name#' => '.root', '#refcount#' => 'n/a' }; + +# The 'imbalance' is a gross count of how balanced a particular +# callsite is. It is used to prune away callsites that are detected to +# be balanced; that is, that have matching AddRef/Release() pairs. + +my %imbalance; +$imbalance{'.root'} = 'n/a'; + +# The main read loop. + +sub read_data($$$) { + my ($INFILE, $plus, $minus) = @_; + + LINE: while (<$INFILE>) { + next LINE if (! /^</); + my @fields = split(/ /, $_); + + my $class = shift(@fields); + my $obj = shift(@fields); + my $sno = shift(@fields); + next LINE unless ($obj eq $::opt_object); + + my $op = shift(@fields); + next LINE unless ($op eq $plus || $op eq $minus); + + my $cnt = shift(@fields); + + # Collect the remaining lines to create a stack trace. We need to + # filter out the frame numbers so that frames that differ only in + # their frame number are considered equivalent. However, we need to + # keep a frame number on each line so that the fix*.py scripts can + # parse the output. So we set the frame number to 0 for every frame. + my @stack; + CALLSITE: while (<$INFILE>) { + chomp; + last CALLSITE if (/^$/); + $_ =~ s/#\d+: /#00: /; # replace frame number with 0 + $stack[++$#stack] = $_; + } + + # Reverse the remaining fields to produce the call stack, with the + # oldest frame at the front of the array. + if (! $::opt_reverse) { + @stack = reverse(@stack); + } + + my $call; + + # If any of the functions in the stack are supposed to be excluded, + # march on to the next line. + foreach $call (@stack) { + next LINE if exists($excludes{$call}); + } + + + # Add the callstack as a path through the call graph, updating + # refcounts at each node. + + my $caller = $callGraphRoot; + + foreach $call (@stack) { + + # Chop the method offset if we're 'collapsing to method' or + # 'collapsing to class'. + $call =~ s/\+0x.*$//g if ($::opt_collapse_to_method || $::opt_collapse_to_class); + + # Chop the method name if we're 'collapsing to class'. + $call =~ s/::.*$//g if ($::opt_collapse_to_class); + + my $site = $caller->{$call}; + if (!$site) { + # This is the first time we've seen this callsite. Add a + # new entry to the call tree. + + $site = { '#name#' => $call, '#refcount#' => 0 }; + $caller->{$call} = $site; + } + + if ($op eq $plus) { + ++($site->{'#refcount#'}); + ++($imbalance{$call}); + } elsif ($op eq $minus) { + --($site->{'#refcount#'}); + --($imbalance{$call}); + } else { + die "Bad operation $op"; + } + + $caller = $site; + } + } +} + +read_data(*STDIN, "AddRef", "Release"); + +if ($::opt_comptrs) { + warn "Subtracting comptr log ". $::opt_comptrs . "\n"; + open(COMPTRS, "<".$::opt_comptrs) + || die "unable to open $::opt_comptrs"; + + # read backwards to subtract + read_data(*COMPTRS, "nsCOMPtrRelease", "nsCOMPtrAddRef"); +} + +sub num_alpha { + my ($aN, $aS, $bN, $bS); + ($aN, $aS) = ($1, $2) if $a =~ /^(\d+) (.+)$/; + ($bN, $bS) = ($1, $2) if $b =~ /^(\d+) (.+)$/; + return $a cmp $b unless defined $aN && defined $bN; + return $aN <=> $bN unless $aN == $bN; + return $aS cmp $bS; +} + +# Given a subtree and its nesting level, return true if that subtree should be pruned. +# If it shouldn't be pruned, destructively attempt to prune its children. +# Also compute the #children# properties of unpruned nodes. +sub prune($$) { + my ($site, $nest) = @_; + + # If they want us to prune the tree's depth, do so here. + return 1 if ($::opt_prune_depth && $nest >= $::opt_prune_depth); + + # If the subtree is balanced, ignore it. + return 1 if ($::opt_ignore_balanced && !$site->{'#refcount#'}); + + my $name = $site->{'#name#'}; + + # If the symbol isn't imbalanced, then prune here (and warn) + if ($::opt_ignore_balanced && !$imbalance{$name}) { + warn "discarding " . $name . "\n"; +# return 1; + } + + my @children; + foreach my $child (sort num_alpha keys(%$site)) { + if (substr($child, 0, 1) ne '#') { + if (prune($site->{$child}, $nest + 1)) { + delete $site->{$child}; + } else { + push @children, $site->{$child}; + } + } + } + $site->{'#children#'} = \@children; + return 0; +} + + +# Compute the #label# properties of this subtree. +# Return the subtree's number of nodes, not counting nodes reachable +# through a labeled node. +sub createLabels($) { + my ($site) = @_; + my @children = @{$site->{'#children#'}}; + my $nChildren = @children; + my $nDescendants = 0; + + foreach my $child (@children) { + my $childDescendants = createLabels($child); + if ($nChildren > 1 && $childDescendants > $::opt_subtree_size) { + die "Internal error" if defined($child->{'#label#'}); + $child->{'#label#'} = "__label__"; + $childDescendants = 1; + } + $nDescendants += $childDescendants; + } + return $nDescendants + 1; +} + + +my $nextLabel = 0; +my @labeledSubtrees; + +sub list($$$$$) { + my ($site, $nest, $nestStr, $childrenLeft, $root) = @_; + my $label = !$root && $site->{'#label#'}; + + # Assign a unique number to the label. + if ($label) { + die unless $label eq "__label__"; + $label = "__" . ++$nextLabel . "__"; + $site->{'#label#'} = $label; + push @labeledSubtrees, $site; + } + + print $nestStr; + if ($::opt_old_style) { + print $label, " " if $label; + print $site->{'#name#'}, ": bal=", $site->{'#refcount#'}, "\n"; + } else { + my $refcount = $site->{'#refcount#'}; + my $l = 8 - length $refcount; + $l = 1 if $l < 1; + print $refcount, " " x $l; + print $label, " " if $label; + print $site->{'#name#'}, "\n"; + } + + $nestStr .= $childrenLeft && !$::opt_old_style ? "| " : " "; + if (!$label) { + my @children = @{$site->{'#children#'}}; + $childrenLeft = @children; + foreach my $child (@children) { + $childrenLeft--; + list($child, $nest + 1, $nestStr, $childrenLeft); + } + } +} + + +if (!prune($callGraphRoot, 0)) { + createLabels $callGraphRoot if ($::opt_subtree_size); + list $callGraphRoot, 0, "", 0, 1; + while (@labeledSubtrees) { + my $labeledSubtree = shift @labeledSubtrees; + print "\n------------------------------\n", +$labeledSubtree->{'#label#'}, "\n"; + list $labeledSubtree, 0, "", 0, 1; + } + print "\n------------------------------\n" if @labeledSubtrees; +} + +print qq{ +Imbalance +--------- +}; + +foreach my $call (sort num_alpha keys(%imbalance)) { + print $call . " " . $imbalance{$call} . "\n"; +} + |