1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850
|
# -*- coding: utf-8 -*-
# GDB will automatically load this module when you attach to the binary moar.
# but first you'll have to tell gdb that it's okay to load it. gdb will instruct
# you on how to do that.
# If it doesn't, you may need to copy or symlink this script right next to the
# moar binary in install/bin.
#
# cd /path/to/install/bin
# ln -s path/to/moarvm/tools/moar-gdb.py
# If you're developing/extending/changing this script, or if you're getting
# python exception messages, this command will be very helpful:
#
# set python print-stack full
# This script contains a few helpers to make debugging MoarVM a bit more pleasant.
#
# So far, there's:
#
# - A semi-functional pretty-printer for MVMString and MVMString*
# - A non-working pretty-printer for MVMObject in general
# - A command "moar-heap" that walks the nursery and gen2 and displays
# statistics about the REPRs found in them, as well as a display of
# the fragmentation of the gen2 pages.
# - A command "diff-moar-heap" that diffs (so far only) the two last
# snapshots of the nursery, or whatever snapshot number you supply
# as the argument.
# Here's the TODO list:
#
# - Figure out if the string pretty-printer is hosed wrt. ropes or if
# it's something wrong with MaarVM's ropes in general.
# - Implement diffing for the gen2 in some sensible manner
# - The backtrace should also display a backtrace of the interpreter
# state. That's relatively easy, as you can just dump_backtrace(tc).
# - Give the object prety printer a children method that figures
# stuff out about attributes of a P6opaque, or CStruct.
# - Let VMArray and MVMHash be displayed with the right display_hint
# and also give them values for the children method
# - Pretty print P6bigint as their value
# - Pretty print P6int and P6num as their value
# Here's some wishlist items
#
# - Offer an HTML rendering of the stats, since gdb insists on printing
# a pager header right in between our pretty gen2 graphs most of the time
import gdb
from collections import defaultdict
from itertools import chain
import math
import random
#import blessings
import sys
import traceback # debugging
# These are the flags from MVMString's body.flags
str_t_info = {0: 'blob_32',
1: 'blob_ascii',
2: 'blob_8',
3: 'strands'}
# How big to make the histograms and such
PRETTY_WIDTH=50
# This must be kept in sync with your MoarVM binary, otherwise we'll end up
# reading bogus data from the gen2 pages.
MVM_GEN2_PAGE_ITEMS = 256
MVM_GEN2_BIN_BITS = 3
MVM_GEN2_BINS = 32
# This ought to give the same results as the equivalent code in gen2.
def bucket_index_to_size(idx):
return (idx + 1) << MVM_GEN2_BIN_BITS
# This is the size the gen2 pictures should have, just so we don't have to
# calculate the same sqrt of a constant over and over again.
MVM_GEN2_PAGE_CUBE_SIZE = int(math.sqrt(MVM_GEN2_PAGE_ITEMS))
# If you'd like more precision for the REPR histogram in the gen2, and have a
# bit of extra patience, turn this up.
EXTRA_SAMPLES = 2
# This corresponds to the defines in MVMArray.h for MVMArrayREPRData.slot_type.
array_storage_types = [
'obj',
'str',
'i64', 'i32', 'i16', 'i8',
'n64', 'n32',
'u64', 'u32', 'u16', 'u8'
]
# These are used to display the hilbert curves extra-prettily.
halfblocks = u"█▀▄ ░▒▓"
def shade_block(u, d):
if u == d == True:
return halfblocks[0]
elif u == True and d == False:
return halfblocks[1]
elif u == False and d == True:
return halfblocks[2]
elif u == d == False:
return halfblocks[3]
elif u == d == None:
return halfblocks[4]
elif u == None and d == False or u == False and d == None:
return halfblocks[5]
elif u == None and d == True or u == True and d == None:
return halfblocks[6]
# Precalculate which index into the gen2 page goes at which coordinates in
# our super pretty hilbert curve.
def generate_hilbert(amount):
hilbert_coords = []
# adapted from the english wikipedia article on the Hilbert Curve
n = int(math.sqrt(amount))
def rot(s, x, y, rx, ry):
if ry == 0:
if rx == 1:
x = s - 1 - x
y = s - 1 - y
return (y, x)
return (x, y)
def xy2d(x, y):
rx = 0
ry = 0
d = 0
s = int(n / 2)
while s > 0:
rx = (x & s) > 0
ry = (y & s) > 0
d += s * s * ((3 * rx) ^ ry)
(x, y) = rot(s, x, y, rx, ry)
s = s // 2
return d
for y in range(n):
hilbert_coords.append([])
for x in range(n):
hilbert_coords[-1].append(xy2d(x, y))
return hilbert_coords
hilbert_coords = generate_hilbert(MVM_GEN2_PAGE_ITEMS)
# Sizes are easier to read if they have .s in them.
def prettify_size(num):
rest = str(num)
result = ""
while len(rest) > 3:
result = rest[-3:] + "." + result
rest = rest[:-3]
if len(rest) >= 1:
result = rest + "." + result
return result[:-1]
class MVMStringPPrinter(object):
"""Whenever gdb encounters an MVMString or an MVMString*, this class gets
instantiated and its to_string method tries its best to print out the
actual contents of the MVMString's storage."""
def __init__(self, val, pointer = False):
self.val = val
self.pointer = pointer
def stringify(self):
stringtyp = str_t_info[int(self.val['body']['storage_type']) & 0b11]
if stringtyp in ("blob_32", "blob_ascii", "blob_8"):
zero_reached = False
data = self.val['body']['storage'][stringtyp]
i = 0
pieces = []
graphs = int(self.val['body']['num_graphs'])
# XXX are the strings actually null-terminated, or do we have to
# XXX check the graphs attribute?
for i in range(graphs):
pdata = int((data + i).dereference())
try:
# ugh, unicode woes ...
pieces.append(chr(pdata))
except:
pieces.append("\\x%x" % pdata)
return "".join(pieces)
elif stringtyp == "strands":
# XXX here be dragons and/or wrong code
# XXX This is still true now
# i = 0
# pieces = []
# data = self.val['body']['storage']['strands']
# end_reached = False
# previous_index = 0
# previous_string = None
# while not end_reached:
# strand_data = (data + i).dereference()
# if strand_data['blob_string'] == 0:
# end_reached = True
# pieces.append(previous_string[1:-1])
# else:
# the_string = strand_data['blob_string'].dereference()
# if previous_string is not None:
# pieces.append(
# str(previous_string)[1:-1][
# int(strand_data['start']) :
# int(strand_data['end']) - previous_index]
# )
# previous_string = str(the_string)
# previous_index = int(strand_data['end'])
# i = i + 1
# return "r(" + ")(".join(pieces) + ")"
return None
else:
return "string of type " + stringtyp
def to_string(self):
result = self.stringify()
if result:
if self.pointer:
return "pointer to '" + self.stringify() + "'"
else:
return "'" + self.stringify() + "'"
else:
return None
# currently nonfunctional
class MVMObjectPPrinter(object):
def __init__(self, val, pointer = False):
self.val = val
self.pointer = pointer
def stringify(self):
if self.pointer:
as_mvmobject = self.val.cast("MVMObject *").dereference()
else:
as_mvmobject = self.val.cast("MVMObject")
_repr = as_mvmobject['st']['REPR']
reprname = _repr['name'].string()
debugname = as_mvmobject['st']['debug_name']
return str(self.val.type.name) + " (" + debugname + ") of repr " + reprname
def to_string(self):
if self.pointer:
return "pointer to " + self.stringify()
else:
return self.stringify()
def show_histogram(hist, sort="value", multiply=False):
"""In the context of this function, a histogram is a hash from an object
that is to be counted to the number the object was found.
sort takes "value" or "key" and gives you the ability to either sort
by "order of buckets" or by "frequency of occurence".
when giving multiply a value, you'll get a display of key * value on
the right of the histogram, useful for a "size of object" to
"count of objects" histogram so you'll get a sum of the taken space
The histogram will not include values that are less than 2, because
that may sometimes lead to "long tail" trouble and buttloads of
pages of output to scroll through."""
if len(hist) == 0:
print("(empty histogram)")
return
if sort == "value":
items = sorted(list(hist.items()), key = lambda vals: -vals[1])
elif sort == "key":
items = sorted(list(hist.items()), key = lambda vals: vals[0])
else:
print("sorting mode", sort, "not implemented")
maximum = max(hist.values())
keymax = min(max([len(str(key)) for key in hist.keys()]), 30)
lines_so_far = 0
group = -1
num_in_group = 0
for key, val in items:
if lines_so_far < 50:
try:
str(key)
except TypeError:
key = repr(key)
if val < 2:
continue
appendix = prettify_size(int(key) * int(val)).rjust(10) if multiply else ""
print(str(key).ljust(keymax + 1), ("[" + "=" * int((float(hist[key]) / maximum) * PRETTY_WIDTH)).ljust(PRETTY_WIDTH + 1), str(val).ljust(len(str(maximum)) + 2), appendix)
else:
if val == group:
num_in_group += 1
else:
if num_in_group > 1:
print(num_in_group, " x ", group)
group = val
num_in_group = 1
lines_so_far += 1
print()
def diff_histogram(hist_before, hist_after, sort="value", multiply=False):
"""Works almost exactly like show_histogram, but takes two histograms that
should have matching keys and displays the difference both in graphical
form and as numbers on the side."""
max_hist = defaultdict(int)
min_hist = defaultdict(int)
zip_hist = {}
max_val = 0
max_key = 0
longest_key = ""
for k,v in chain(hist_before.items(), hist_after.items()):
max_hist[k] = max(max_hist[k], v)
min_hist[k] = min(max_hist[k], v)
max_val = max(max_val, v)
max_key = max(max_key, k)
longest_key = str(k) if len(str(k)) > len(longest_key) else longest_key
for k in max_hist.keys():
zip_hist[k] = (hist_before[k], hist_after[k])
if sort == "value":
items = sorted(list(zip_hist.items()), key = lambda vals: -max(vals[1][0], vals[1][1]))
elif sort == "key":
items = sorted(list(zip_hist.items()), key = lambda vals: vals[0])
else:
print("sorting mode", sort, "not implemented")
for key, (val1, val2) in items:
lv, rv = min(val1, val2), max(val1, val2)
lc, rc = ("+", "-") if val1 > val2 else ("-", "+")
bars = "[" \
+ lc * int((float(lv) / max_val) * PRETTY_WIDTH) \
+ rc * int((float(rv - lv) / max_val) * PRETTY_WIDTH)
values = str(val1).ljust(len(str(max_val)) + 1) \
+ " -> " \
+ str(val2).ljust(len(str(max_val)) + 1)
appendix = prettify_size(int(key) * int(val1)).rjust(10) \
+ " -> " \
+ prettify_size(int(key) * int(val2)).rjust(10) \
if multiply else ""
print(str(key).ljust(len(longest_key) + 2), bars.ljust(PRETTY_WIDTH), values, appendix)
class CommonHeapData(object):
"""This base class holds a bunch of histograms and stuff that are
interesting regardless of wether we are looking at nursery objects
or gen2 objects."""
number_objects = None
number_stables = None
number_typeobs = None
size_histogram = None
repr_histogram = None
opaq_histogram = None
arrstr_hist = None
arrusg_hist = None
string_histogram = None
generation = None
def __init__(self, generation):
self.generation = generation
self.size_histogram = defaultdict(int)
self.repr_histogram = defaultdict(int)
self.opaq_histogram = defaultdict(int)
self.arrstr_hist = defaultdict(int)
self.arrusg_hist = defaultdict(int)
self.string_histogram = defaultdict(int)
self.number_objects = 0
self.number_stables = 0
self.number_typeobs = 0
def analyze_single_object(self, cursor):
"""Given a pointer into the nursery or gen2 that points at the
beginning of a MVMObject of any sort, run a statistical analysis
of what the object is and some extra info depending on its REPR.
To make this scheme work well with the nursery analysis, it returns
the size of the object analysed."""
stooge = cursor.cast(gdb.lookup_type("MVMObjectStooge").pointer())
size = int(stooge['common']['header']['size'])
flags = int(stooge['common']['header']['flags'])
is_typeobj = flags & 1
is_stable = flags & 2
STable = stooge['common']['st'].dereference()
if not is_stable:
REPR = STable["REPR"]
REPRname = REPR["name"].string()
try:
debugname = STable['debug_name'].string()
except gdb.MemoryError:
debugname = "n/a"
if is_typeobj:
self.number_typeobs += 1
else:
self.number_objects += 1
else:
REPR = None
REPRname = "STable"
debugname = "n/a"
self.number_stables += 1
self.size_histogram[int(size)] += 1
if debugname != "n/a":
self.repr_histogram[debugname] += 1
else:
self.repr_histogram[REPRname] += 1
if REPRname == "P6opaque":
self.opaq_histogram[int(size)] += 1
elif REPRname == "VMArray":
slot_type = int(STable['REPR_data'].cast(gdb.lookup_type("MVMArrayREPRData").pointer())['slot_type'])
self.arrstr_hist[array_storage_types[slot_type]] += 1
array_body = cursor.cast(gdb.lookup_type("MVMArray").pointer())['body']
if array_body['ssize'] == 0:
usage_perc = "N/A"
else:
usage_perc = (int(array_body['elems'] * 10) / int(array_body['ssize'])) * 10
if usage_perc < 0 or usage_perc > 100:
usage_perc = "inv"
self.arrusg_hist[usage_perc] += 1
elif REPRname == "MVMString":
try:
casted = cursor.cast(gdb.lookup_type('MVMString').pointer())
stringresult = MVMStringPPrinter(casted).stringify()
if stringresult is not None:
self.string_histogram[stringresult] += 1
else:
self.string_histogram["mvmstr@" + hex(int(cursor.address.cast(gdb.lookup_type("int"))))] += 1
except gdb.MemoryError as e:
print(e)
print(e.traceback())
print(cursor.cast(gdb.lookup_type('MVMString').pointer()))
pass
return size
class NurseryData(CommonHeapData):
"""The Nursery Data contains the current position where we allocate as
well as the beginning and end of the given nursery."""
allocation_offs = None
start_addr = None
end_addr = None
def __init__(self, generation, start_addr, end_addr, allocation_offs):
super(NurseryData, self).__init__(generation)
self.start_addr = gdb.Value(start_addr)
self.end_addr = gdb.Value(end_addr)
self.allocation_offs = allocation_offs
def analyze(self, tc):
cursor = gdb.Value(self.start_addr)
info_step = int(self.allocation_offs - cursor) // 50
next_info = cursor + info_step
print("_" * 50)
while cursor < self.allocation_offs:
try:
size = self.analyze_single_object(cursor)
except:
print("while trying to analyze single object:");
traceback.print_exc()
print(stooge)
print(stooge.__repr__())
cursor += size
if cursor > next_info:
next_info += info_step
sys.stdout.write("-")
sys.stdout.flush()
print()
def summarize(self):
print("nursery state:")
sizes = (int(self.allocation_offs - self.start_addr), int(self.end_addr - self.allocation_offs))
relsizes = [1.0 * size / (float(int(self.end_addr - self.start_addr))) for size in sizes]
print("[" + "=" * int(relsizes[0] * 20) + " " * int(relsizes[1] * 20) + "] ", int(relsizes[0] * 100),"%")
print(self.number_objects, "objects;", self.number_typeobs, " type objects;", self.number_stables, " STables")
print("sizes of objects/stables:")
show_histogram(self.size_histogram, "key", True)
print("sizes of P6opaques only:")
show_histogram(self.opaq_histogram, "key", True)
print("debugnames:")
show_histogram(self.repr_histogram)
print("VMArray storage types:")
show_histogram(self.arrstr_hist)
print("VMArray usage percentages:")
show_histogram(self.arrusg_hist, "key")
print("strings:")
show_histogram(self.string_histogram)
def diff(self, other):
print("nursery state --DIFF--:")
print("sizes of objects/stables:")
diff_histogram(self.size_histogram, other.size_histogram, "key", True)
print("sizes of P6opaques only:")
diff_histogram(self.opaq_histogram, other.opaq_histogram, "key", True)
print("debugnames:")
diff_histogram(self.repr_histogram, other.repr_histogram)
print("VMArray storage types:")
diff_histogram(self.arrstr_hist, other.arrstr_hist)
print("VMArray usage percentages:")
diff_histogram(self.arrusg_hist, other.arrusg_hist, "key")
class Gen2Data(CommonHeapData):
"""One Gen2Data instance gets created per size class.
Thus, every instance corresponds to an exact size of object.
The class also handles the free list that is chained through the
gen2, so that fragmentation can be determined and stale objects not
sampled for analysis."""
size_bucket = None
length_freelist = None
bucket_size = None
g2sc = None
page_addrs = None
pagebuckets = None
empty = False
cur_page = 0
repr_histogram = None
size_histogram = None
def sizes(self):
# XXX this ought to return a tuple with the sizes we
# accept in this bucket
return bucket_index_to_size(self.size_bucket), bucket_index_to_size(self.size_bucket + 1) - 1
def __init__(self, generation, gen2sizeclass, size_bucket):
super(Gen2Data, self).__init__(generation)
self.size_bucket = size_bucket
self.g2sc = gen2sizeclass
self.bucket_size = bucket_index_to_size(self.size_bucket)
self.repr_histogram = defaultdict(int)
self.size_histogram = defaultdict(int)
def analyze(self, tc):
if int(self.g2sc['pages'].cast(gdb.lookup_type("int"))) == 0:
self.empty = True
return
pagebuckets = [[True for i in range(MVM_GEN2_PAGE_ITEMS)] for i in range(int(self.g2sc['num_pages']))]
page_addrs = []
self.page_addrs = page_addrs
self.pagebuckets = pagebuckets
self.cur_page = int(self.g2sc['cur_page'])
# we need to make sure we don't accidentally run into free_list'd slots
# that's why we just collect addresses up front and sample them later on.
sample_stooges = []
page_cursor = self.g2sc['pages']
for page_idx in range(int(self.g2sc['num_pages'])):
# collect the page addresses so that we can match arbitrary
# pointers to page index/bucket index pairs later on.
page_addrs.append(page_cursor.dereference())
if page_idx == self.cur_page:
# XXX cur_page is going to be removed from MoarVM, as it's only
# XXX needed for this exact code here.
# if the page we're looking at is the "current" page, we look
# at the alloc_pos to find out where allocated objects stop.
alloc_bucket_idx = int(int(self.g2sc['alloc_pos'] - page_cursor.dereference()) // self.bucket_size)
pagebuckets[page_idx][alloc_bucket_idx:] = [False] * (MVM_GEN2_PAGE_ITEMS - alloc_bucket_idx)
elif page_idx > self.cur_page:
# if we're past the page we're currently allocating in, the
# pages are empty by definition
alloc_bucket_idx = 0
pagebuckets[page_idx] = [False] * MVM_GEN2_PAGE_ITEMS
else:
# otherwise, the whole page is potentially allocated objects.
alloc_bucket_idx = MVM_GEN2_PAGE_ITEMS
# sample a few objects for later analysis (after free list checking)
samplecount = int(min(MVM_GEN2_PAGE_CUBE_SIZE * EXTRA_SAMPLES, alloc_bucket_idx // 4))
samples = sorted(random.sample(range(0, alloc_bucket_idx), samplecount))
for idx in samples:
stooge = (page_cursor.dereference() + (idx * self.bucket_size)).cast(gdb.lookup_type("MVMObjectStooge").pointer())
sample_stooges.append((stooge, page_idx, idx))
page_cursor += 1
# now punch holes in our page_buckets
free_cursor = self.g2sc['free_list']
def address_to_page_and_bucket(addr):
for idx, base in enumerate(page_addrs):
end = base + self.bucket_size * MVM_GEN2_PAGE_ITEMS
if base <= addr < end:
return idx, int((addr - base) // self.bucket_size)
self.length_freelist = 0
while free_cursor.cast(gdb.lookup_type("int")) != 0:
if free_cursor.dereference().cast(gdb.lookup_type("int")) != 0:
result = address_to_page_and_bucket(free_cursor.dereference())
if result:
page, bucket = result
pagebuckets[page][bucket] = False
self.length_freelist += 1
free_cursor = free_cursor.dereference().cast(gdb.lookup_type("char").pointer().pointer())
print("")
#doubles = defaultdict(int)
# now we can actually sample our objects
for stooge, page, idx in sample_stooges:
if pagebuckets[page][idx] != True:
continue
try:
size = self.analyze_single_object(stooge)
except Exception as e:
print("while trying to analyze single object:");
traceback.print_exc()
print(stooge)
print(stooge.__repr__())
#if len(doubles) > 10:
#show_histogram(doubles)
def summarize(self):
print("size bucket:", self.bucket_size)
if self.empty:
print("(unallocated)")
return
cols_per_block = int(math.sqrt(MVM_GEN2_PAGE_ITEMS))
lines_per_block = cols_per_block // 2
outlines = [[] for i in range(lines_per_block + 1)]
break_step = PRETTY_WIDTH // (lines_per_block + 1)
next_break = break_step
pgct = 0
fullpages = 0
drawn = False
for pgnum, page in enumerate(self.pagebuckets):
if pgnum > self.cur_page:
break
if not all(page) and any(page):
# hacky "take two at a time"
for outline, (line, nextline) in enumerate(zip(*[iter(hilbert_coords)] * 2)):
for idx, (upper, lower) in enumerate(zip(line, nextline)):
upper, lower = page[upper], page[lower]
outlines[-lines_per_block + outline].append(shade_block(upper, lower))
outlines[-lines_per_block - 1].append(str(str(pgnum + 1) + "st pg").center(cols_per_block) + " ")
pgct += 1
drawn = True
else:
fullpages += 1
drawn = False
if pgct > next_break:
outlines.extend([[] for i in range(lines_per_block + 2)])
next_break += break_step
elif pgnum != self.cur_page and drawn:
for line_num in range(lines_per_block):
outlines[-line_num - 1].append(" ")
print((u"\n".join(map(lambda l: u"".join(l), outlines))))
if fullpages > 0:
print("(and", fullpages, "completely filled pages)",)
if self.cur_page < len(self.pagebuckets):
print("(and", (len(self.pagebuckets) - self.cur_page + 1), "empty pages)",)
if self.length_freelist > 0:
print("(freelist with", self.length_freelist, "entries)",)
print("")
# does the allocator/copier set the size of the object to the exact bucket size
# automatically?
if len(self.size_histogram) > 1:
print("sizes of objects/stables:")
try:
show_histogram(self.size_histogram, "key", True)
except Exception as e:
print("while trying to show the size histogram...")
print(e)
print(e.traceback())
if len(self.repr_histogram) >= 1:
print("debugnames:")
try:
show_histogram(self.repr_histogram)
except Exception as e:
print("while trying to show the repr histogram...")
print(e)
print("strings:")
show_histogram(self.string_histogram)
class OverflowData(CommonHeapData):
def analyze(self, tc):
g2a = tc['gen2']
num_overflows = g2a["num_overflows"]
print(num_overflows)
try:
for of_idx in range(num_overflows):
of_obj = g2a["overflows"][of_idx]
self.analyze_single_object(of_obj)
except:
print("error while analyze_single_object or something");
def summarize(self):
print("overflows in the gen2")
print(self.number_objects, "objects;", self.number_typeobs, " type objects;", self.number_stables, " STables")
print("sizes of objects/stables:")
show_histogram(self.size_histogram, "key", True)
print("sizes of P6opaques only:")
show_histogram(self.opaq_histogram, "key", True)
print("debugnames:")
show_histogram(self.repr_histogram)
print("VMArray storage types:")
show_histogram(self.arrstr_hist)
print("VMArray usage percentages:")
show_histogram(self.arrusg_hist, "key")
print("strings:")
show_histogram(self.string_histogram)
class HeapData(object):
run_nursery = None
run_gen2 = None
generation = None
nursery_memory = []
class AnalyzeHeapCommand(gdb.Command):
"""Analyze the nursery and gen2 of MoarVM's garbage collector corresponding
to the current tc, or the tc you pass as the first argument"""
def __init__(self):
super(AnalyzeHeapCommand, self).__init__("moar-heap", gdb.COMMAND_DATA)
def invoke(self, arg, from_tty):
tc = gdb.selected_frame().read_var(arg if arg else "tc")
if not str(tc.type).startswith("MVMThreadContext"):
raise ValueError("Please invoke the heap analyzer command on a MVMThreadContext, usually tc.")
try:
# find out the GC generation we're in (just a number increasing by 1 every time we GC)
instance = tc['instance']
generation = instance['gc_seq_number']
nursery = NurseryData(generation, tc['nursery_tospace'], tc['nursery_alloc_limit'], tc['nursery_alloc'])
nursery.analyze(tc)
nursery_memory.append(nursery)
print("the current generation of the gc is", generation)
sizeclass_data = []
for sizeclass in range(MVM_GEN2_BINS):
g2sc = Gen2Data(generation, tc['gen2']['size_classes'][sizeclass], sizeclass)
sizeclass_data.append(g2sc)
g2sc.analyze(tc)
overflowdata = OverflowData(generation)
overflowdata.analyze(tc)
for g2sc in sizeclass_data:
g2sc.summarize()
nursery.summarize()
overflowdata.summarize()
except KeyboardInterrupt:
print("aborted the analysis.")
class DiffHeapCommand(gdb.Command):
"""Display the difference between two snapshots of the nursery."""
def __init__(self):
super(DiffHeapCommand, self).__init__("diff-moar-heap", gdb.COMMAND_DATA)
def invoke(self, arg, from_tty):
if arg != "":
if " " in arg:
pos1, pos2 = map(int, arg.split(" "))
else:
pos1, pos2 = int(arg), int(arg - 1)
else:
pos1 = -1
pos2 = -2
assert len(nursery_memory) > max(pos1, pos2)
nursery_memory[pos2].diff(nursery_memory[pos1])
def str_lookup_function(val):
if str(val.type) == "MVMString":
return MVMStringPPrinter(val)
elif str(val.type) == "MVMString *":
return MVMStringPPrinter(val, True)
return None
def mvmobject_lookup_function(val):
pointer = str(val.type).endswith("*")
if str(val.type).startswith("MVM"):
try:
val.cast(gdb.lookup_type("MVMObject" + (" *" if pointer else "")))
return MVMObjectPPrinter(val, pointer)
except Exception as e:
print("couldn't cast this:", e)
pass
return None
def register_printers(objfile):
objfile.pretty_printers.append(str_lookup_function)
print("MoarVM string pretty printer registered")
# XXX since this is currently nonfunctional, just ignore it for now
# objfile.pretty_printers.append(mvmobject_lookup_function)
# print("MoarVM Object pretty printer registered")
commands = []
def register_commands(objfile):
commands.append(AnalyzeHeapCommand())
print("moar-heap registered")
commands.append(DiffHeapCommand())
print("diff-moar-heap registered")
# We have to introduce our classes to gdb so that they can be used
if __name__ == "__main__":
the_objfile = gdb.current_objfile()
if the_objfile == None:
the_objfile = gdb.lookup_objfile("libmoar.so")
register_printers(the_objfile)
register_commands(the_objfile)
|