Add a new command to summarize memory usage by function addresses in dmprof

The output format is:
vss1;rss1;addr1;symbol1
vss2;rss2;addr2;symbol2
...
The symbol is related to the addr. If the binary file has symbols, it just same as before:function@sourcefile.
Otherwise it will show the pathname the addr belongs to (mainly it's the library).
This new command could be helpful for lcation leaks.

R=kouhei@chromium.org
BUG=427393

Review URL: https://codereview.chromium.org/700773002

Cr-Commit-Position: refs/heads/master@{#302745}

3 files changed, 90 insertions, 0 deletions

diff --git a/tools/deep_memory_profiler/dmprof.py b/tools/deep_memory_profiler/dmprof.py
index 0cb030a..85c1f6d 100644
--- a/tools/deep_memory_profiler/dmprof.py
+++ b/tools/deep_memory_profiler/dmprof.py
@@ -19,6 +19,7 @@ LOGGER = logging.getLogger('dmprof')
 
 def main():
   COMMANDS = {
+    'addr': subcommands.AddrCommand,
     'buckets': subcommands.BucketsCommand,
     'cat': subcommands.CatCommand,
     'csv': subcommands.CSVCommand,
diff --git a/tools/deep_memory_profiler/subcommands/__init__.py b/tools/deep_memory_profiler/subcommands/__init__.py
index 4fb29d0..cc28dfd 100644
--- a/tools/deep_memory_profiler/subcommands/__init__.py
+++ b/tools/deep_memory_profiler/subcommands/__init__.py
@@ -2,6 +2,7 @@
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
+from subcommands.addr import AddrCommand
 from subcommands.buckets import BucketsCommand
 from subcommands.cat import CatCommand
 from subcommands.expand import ExpandCommand
diff --git a/tools/deep_memory_profiler/subcommands/addr.py b/tools/deep_memory_profiler/subcommands/addr.py
new file mode 100644
index 0000000..ee399e8
--- /dev/null
+++ b/tools/deep_memory_profiler/subcommands/addr.py
@@ -0,0 +1,88 @@
+# Copyright 2013 The Chromium Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+import logging
+import sys
+import re
+
+from lib.subcommand import SubCommand
+
+
+LOGGER = logging.getLogger('dmprof')
+
+
+class AddrCommand(SubCommand):
+  def __init__(self):
+    super(AddrCommand, self).__init__('Usage: %prog addr <dump-file>')
+
+  def do(self, sys_argv, out=sys.stdout):
+    _, args = self._parse_args(sys_argv, 1)
+    dump_path = args[1]
+    (bucket_set, dump) = SubCommand.load_basic_files(dump_path, False)
+    AddrCommand._output(dump, bucket_set, out)
+    return 0
+
+  @staticmethod
+  def _output(dump, bucket_set, out):
+    """Prints memory usage by function addresses with resolving symbols.
+
+    Args:
+        bucket_set: A BucketSet object.
+        out: An IO object to output.
+    """
+    sizes = {}
+    library = {}
+
+    _ADDR_PATTERN = re.compile(r'^0x[a-f0-9]+$', re.IGNORECASE)
+    _TCMALLOC_PATTERN = re.compile(r'.*(ProfilerMalloc|MemoryRegionMap::'
+                 '|TypeProfilerMalloc|DoAllocWithArena|SbrkSysAllocator::Alloc'
+                 '|MmapSysAllocator::Alloc|LowLevelAlloc::Alloc'
+                 '|LowLevelAlloc::AllocWithArena).*', re.IGNORECASE)
+
+    for bucket_id, virtual, committed, _, _ in dump.iter_stacktrace:
+      bucket = bucket_set.get(bucket_id)
+      if not bucket:
+        AddrCommand._add_size(bucket_id, "bucket", virtual, committed, sizes)
+        continue
+
+      if _TCMALLOC_PATTERN.match(bucket.symbolized_joined_stackfunction):
+        AddrCommand._add_size("TCMALLOC", "TCMALLOC",
+                              virtual, committed, sizes)
+        continue
+
+      for index, addr in enumerate(bucket.stacktrace):
+        if _ADDR_PATTERN.match(bucket.symbolized_stackfunction[index]):
+          AddrCommand._find_library(addr, dump, library)
+          AddrCommand._add_size(hex(addr), library[addr],
+                                virtual, committed, sizes)
+        else:
+          AddrCommand._add_size(hex(addr),
+                    bucket.symbolized_stackfunction[index]
+                    + "@" + bucket.symbolized_stacksourcefile[index],
+                    virtual, committed, sizes)
+
+    for key in sorted(sizes):
+      out.write('%s;%s;%s;%s\n' % (sizes[key]["vss"], sizes[key]["rss"],
+                    key, sizes[key]["desc"]))
+
+
+  @staticmethod
+  def _add_size(key, desc, vss, rss, sizes):
+    if not key in sizes:
+      sizes[key] = {"desc":desc, "vss":0, "rss":0, "alloc":0, "free": 0}
+    if sizes[key]["desc"] == desc:
+      sizes[key]["vss"] += vss
+      sizes[key]["rss"] += rss
+    else:
+      sys.stderr.write('%s:(%s) or (%s)?\n' % (key, sizes[key]["desc"], desc))
+      sys.exit(1)
+
+  @staticmethod
+  def _find_library(func, dump, library):
+    if not func in library:
+      library[func] = "Unknown"
+      for addr, region in dump.iter_map:
+        if addr[0] >= func and func < addr[1]:
+          library[func] = region[1]['vma']['name']
+          break
\ No newline at end of file