scripts/mono-heapviz

   1 #!/usr/bin/env python
   2
   3 # Generate a heap visualization for SGen from the heap dump written by
   4 # mono if the MONO_GC_DEBUG is set to something like
   5 # "heap-dump=/path/to/file".  This script accepts the file as stdin
   6 # and generates HTML and PNG files.
   7
   8 from __future__ import print_function
   9 import sys, os
  10 import Image, ImageDraw
  11 from xml.sax import ContentHandler, saxutils, make_parser
  12 from xml.sax.handler import feature_namespaces
  13 from optparse import OptionParser
  14
  15 chunk_size = 1024                       # number of bytes in a chunk
  16 chunk_pixel_size = 2                    # a chunk is a square with this side length
  17 large_sections = False
  18
  19 def mark_chunk (img_draw, i, color, section_width):
  20     row = i / section_width
  21     col = i % section_width
  22     pixel_col = col * chunk_pixel_size
  23     pixel_row = row * chunk_pixel_size
  24     img_draw.rectangle ([(pixel_col, pixel_row), (pixel_col + chunk_pixel_size - 1, pixel_row + chunk_pixel_size - 1)], fill = color)
  25
  26 class Range:
  27     pass
  28
  29 class OccupiedRange (Range):
  30     def __init__ (self, offset, size):
  31         self.offset = offset
  32         self.size = size
  33
  34     def mark (self, img_draw, color, section_width):
  35         start = self.offset / chunk_size
  36         end = (self.offset + self.size - 1) / chunk_size
  37         for i in range (start, end + 1):
  38             mark_chunk (img_draw, i, color, section_width)
  39
  40 class ObjectRange (OccupiedRange):
  41     def __init__ (self, klass, offset, size):
  42         OccupiedRange.__init__ (self, offset, size)
  43         self.klass = klass;
  44
  45 class SectionHandler:
  46     def __init__ (self, width):
  47         self.width = width
  48         self.ranges = []
  49         self.size = 0
  50         self.used = 0
  51
  52     def add_object (self, klass, offset, size):
  53         self.ranges.append (ObjectRange (klass, offset, size))
  54         self.used += size
  55
  56     def add_occupied (self, offset, size):
  57         self.ranges.append (OccupiedRange (offset, size))
  58         self.used += size
  59
  60     def draw (self):
  61         height = (((self.size / chunk_size) + (self.width - 1)) / self.width) * chunk_pixel_size
  62
  63         color_background = (255, 255, 255)
  64         color_free = (0, 0, 0)
  65         color_occupied = (0, 255, 0)
  66
  67         img = Image.new ('RGB', (self.width * chunk_pixel_size, height), color_free)
  68         img_draw = ImageDraw.Draw (img)
  69         #FIXME: remove filling after end of heap
  70
  71         for r in self.ranges:
  72             r.mark (img_draw, color_occupied, self.width)
  73
  74         return img
  75
  76     def emit (self, collection_file, collection_kind, collection_num, section_num):
  77         print ('<h2>%s</h2>' % self.header (), file = collection_file)
  78         print ('<p>Size %d kB - ' % (self.size / 1024), file = collection_file)
  79         print ('used %d kB</p>' % (self.used / 1024), file = collection_file)
  80
  81         filename = '%s_%d_%d.png' % (collection_kind, collection_num, section_num)
  82         print ('<p><img src="%s"></img></p>' % filename, file = collection_file)
  83         img = self.draw ()
  84         img.save (filename)
  85
  86 class SmallSectionHandler (SectionHandler):
  87     def __init__ (self):
  88         SectionHandler.__init__ (self, -1)
  89         self.offset = 0
  90
  91     def start_section (self, kind, size):
  92         assert kind == 'old'
  93         if self.width <= 0:
  94             self.width = (size + chunk_size - 1) / chunk_size
  95         else:
  96             assert self.width == (size + chunk_size - 1) / chunk_size
  97         self.size += size
  98
  99     def add_object (self, klass, offset, size):
 100         SectionHandler.add_object (self, klass, self.offset + offset, size)
 101
 102     def add_occupied (self, offset, size):
 103         SectionHandler.add_occupied (self, self.offset + offset, size)
 104
 105     def end_section (self):
 106         self.offset += self.width * chunk_size
 107
 108     def header (self):
 109         return 'old sections'
 110
 111 class LargeSectionHandler (SectionHandler):
 112     def __init__ (self):
 113         SectionHandler.__init__ (self, 500)
 114
 115     def start_section (self, kind, size):
 116         self.kind = kind
 117         self.ranges = []
 118         self.size = size
 119         self.used = 0
 120
 121     def end_section (self):
 122         pass
 123
 124     def header (self):
 125         return self.kind + ' section'
 126
 127 class DocHandler (saxutils.DefaultHandler):
 128     def start (self):
 129         self.collection_index = 0
 130         self.index_file = open ('index.html', 'w')
 131         print ('<html><body>', file = self.index_file)
 132
 133     def end (self):
 134         print ('</body></html>', file = self.index_file)
 135         self.index_file.close ()
 136
 137     def startElement (self, name, attrs):
 138         if name == 'collection':
 139             self.collection_kind = attrs.get('type', None)
 140             self.collection_num = int(attrs.get('num', None))
 141             reason = attrs.get('reason', None)
 142             if reason:
 143                 reason = ' (%s)' % reason
 144             else:
 145                 reason = ''
 146             self.section_num = 0
 147             filename = 'collection_%d.html' % self.collection_index
 148             print ('<a href="%s">%s%s collection %d</a>' % (filename, self.collection_kind, reason, self.collection_num), file = self.index_file)
 149             self.collection_file = open (filename, 'w')
 150             print ('<html><body>', file = self.collection_file)
 151             print ('<p><a href="collection_%d.html">Prev</a> <a href="collection_%d.html">Next</a> <a href="index.html">Index</a></p>' % (self.collection_index - 1, self.collection_index + 1), file = self.collection_file)
 152             print ('<h1>%s collection %d</h1>' % (self.collection_kind, self.collection_num), file = self.collection_file)
 153             self.usage = {}
 154             self.los_usage = {}
 155             self.pinned_usage = {}
 156             self.in_los = False
 157             self.in_pinned = False
 158             self.heap_used = 0
 159             self.heap_size = 0
 160             self.los_size = 0
 161             if large_sections:
 162                 self.section_handler = LargeSectionHandler ()
 163             else:
 164                 self.section_handler = self.small_section_handler = SmallSectionHandler ()
 165         elif name == 'pinned':
 166             kind = attrs.get('type', None)
 167             bytes = int(attrs.get('bytes', None))
 168             print ('Pinned from %s: %d kB<br>' % (kind, bytes / 1024), file = self.collection_file)
 169         elif name == 'section':
 170             kind = attrs.get('type', None)
 171             size = int(attrs.get('size', None))
 172
 173             self.heap_size += size
 174
 175             if not large_sections:
 176                 if kind == 'nursery':
 177                     self.section_handler = LargeSectionHandler ()
 178                 else:
 179                     self.section_handler = self.small_section_handler
 180
 181             self.section_handler.start_section (kind, size)
 182         elif name == 'object':
 183             klass = attrs.get('class', None)
 184             size = int(attrs.get('size', None))
 185
 186             if self.in_los:
 187                 usage_dict = self.los_usage
 188                 self.los_size += size
 189             elif self.in_pinned:
 190                 location = attrs.get('location', None)
 191                 if location not in self.pinned_usage:
 192                     self.pinned_usage[location] = {}
 193                 usage_dict = self.pinned_usage[location]
 194             else:
 195                 usage_dict = self.usage
 196                 offset = int(attrs.get('offset', None))
 197
 198                 self.section_handler.add_object (klass, offset, size)
 199                 self.heap_used += size
 200             if not (klass in usage_dict):
 201                 usage_dict [klass] = (0, 0)
 202             usage = usage_dict [klass]
 203             usage_dict [klass] = (usage [0] + 1, usage [1] + size)
 204         elif name == 'occupied':
 205             offset = int(attrs.get('offset', None))
 206             size = int(attrs.get('size', None))
 207
 208             self.section_handler.add_occupied (offset, size)
 209             self.heap_used += size
 210         elif name == 'los':
 211             self.in_los = True
 212         elif name == 'pinned-objects':
 213             self.in_pinned = True
 214
 215     def dump_usage (self, usage_dict, limit):
 216         klasses = sorted (usage_dict.keys (), lambda x, y: usage_dict [y][1] - usage_dict [x][1])
 217         if limit:
 218             klasses = klasses [0:limit]
 219         for klass in klasses:
 220             usage = usage_dict [klass]
 221             if usage [1] < 100000:
 222                 print ('%s %d bytes' % (klass, usage [1]), file = self.collection_file)
 223             else:
 224                 print ('%s %d kB' % (klass, usage [1] / 1024), file = self.collection_file)
 225             print (' (%d)<br>' % usage [0], file = self.collection_file)
 226
 227     def endElement (self, name):
 228         if name == 'section':
 229             self.section_handler.end_section ()
 230
 231             if large_sections or self.section_handler != self.small_section_handler:
 232                 self.section_handler.emit (self.collection_file, self.collection_kind, self.collection_num, self.section_num)
 233                 self.section_num += 1
 234         elif name == 'collection':
 235             if not large_sections:
 236                 self.small_section_handler.emit (self.collection_file, self.collection_kind, self.collection_num, self.section_num)
 237
 238             self.dump_usage (self.usage, 10)
 239             print ('<h3>LOS</h3>', file = self.collection_file)
 240             self.dump_usage (self.los_usage, None)
 241             print ('<h3>Pinned</h3>', file = self.collection_file)
 242             for location in sorted (self.pinned_usage.keys ()):
 243                 print ('<h4>%s</h4>' % location, file = self.collection_file)
 244                 self.dump_usage (self.pinned_usage[location], None)
 245             print ('</body></html>', file = self.collection_file)
 246             print (' - %d kB / %d kB (%d%%) - %d kB LOS</a><br>' % (self.heap_used / 1024, self.heap_size / 1024, int(100.0 * self.heap_used / self.heap_size), self.los_size / 1024), file = self.index_file)
 247             self.collection_file.close ()
 248             self.collection_index += 1
 249         elif name == 'los':
 250             self.in_los = False
 251         elif name == 'pinned-objects':
 252             self.in_pinned = False
 253
 254 def main ():
 255     usage = "usage: %prog [options]"
 256     parser = OptionParser (usage)
 257     parser.add_option ("-l", "--large-sections", action = "store_true", dest = "large_sections")
 258     parser.add_option ("-s", "--small-sections", action = "store_false", dest = "large_sections")
 259     (options, args) = parser.parse_args ()
 260     if options.large_sections:
 261         large_sections = True
 262
 263     dh = DocHandler ()
 264     parser = make_parser ()
 265     parser.setFeature (feature_namespaces, 0)
 266     parser.setContentHandler (dh)
 267     dh.start ()
 268     parser.parse (sys.stdin)
 269     dh.end ()
 270
 271 if __name__ == "__main__":
 272     main ()