Gitiles
Code Review Sign In
review.blissroms.org / platform_external_skia / refs/heads/universe / . / tools / bloaty_treemap.py
blob: f1c92e726763820bd20eb5d4f83b442b73fe6a49 [file] [log] [blame]
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]1#!/usr/bin/env python3
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]2
3# Copyright 2021 Google LLC
4#
5# Use of this source code is governed by a BSD-style license that can be
6# found in the LICENSE file.
7
8
9# This script is written to process the output from bloaty, read via stdin
10# The easiest way to use the script:
11#
12# bloaty <path_to_binary> -d compileunits,symbols -n 0 --tsv | bloaty_treemap.py > bloaty.html
13#
14# Open the resulting .html file in your browser.
15
16# TODO: Deal with symbols vs. fullsymbols, even both?
17# TODO: Support aggregation by scope, rather than file (split C++ identifiers on '::')
18# TODO: Deal with duplicate symbols better. These are actually good targets for optimization.
19# They are sometimes static functions in headers (so they appear in multiple .o files),
20# There are also symbols that appear multiple times due to inlining (eg, kNoCropRect).
21# TODO: Figure out why some symbols are misattributed. Eg, Swizzle::Convert and ::Make are tied
22# to the header by nm, and then to one caller (at random) by bloaty. They're not inlined,
23# though. Unless LTO is doing something wacky here? Scope-aggregation may be the answer?
24# Ultimately, this seems like an issue with bloaty and/or debug information itself.
25
26import os
27import sys
28
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]29parent_map = {}
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]30
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]31# For a given filepath "foo/bar/baz.cpp", `add_path` outputs rows to the data table
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]32# establishing the node hierarchy, and ensures that each line is emitted exactly once:
33#
34# ['foo/bar/baz.cpp', 'foo/bar', 0],
35# ['foo/bar', 'foo', 0],
36# ['foo', 'ROOT', 0],
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]37def add_path(path):
38 if not path in parent_map:
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]39 head = os.path.split(path)[0]
40 if not head:
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]41 parent_map[path] = "ROOT"
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]42 else:
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]43 add_path(head)
44 parent_map[path] = head
Leandro Lovisolo5a7d91c2021-12-01 16:03:27 -0500[diff] [blame]45
46 # We add a suffix to paths to eliminate the chances of a path name colliding with a symbol
47 # name. This is important because google.visualization.TreeMap requires node names to be
48 # unique, and a file such as test/foo/bar.cpp would create a node named "test", which could
49 # collide with a symbol named "test" defined in a C++ file.
50 #
51 # Assumptions made:
52 # - No C++ symbol ends with " (Path)".
53 # - No C++ symbol is named "ROOT".
54 parent = parent_map[path]
55 if parent != "ROOT": parent = "%s (Path)" % parent
56 print("['%s (Path)', '%s', 0]," % (path, parent))
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]57
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]58def main():
59 # HTML/script header, plus the first two (fixed) rows of the data table
60 print("""
61 <html>
62 <head>
63 <script type="text/javascript" src="https://www.gstatic.com/charts/loader.js"></script>
64 <script type="text/javascript">
65 google.charts.load('current', {'packages':['treemap']});
66 google.charts.setOnLoadCallback(drawChart);
67 function drawChart() {
68 const data = google.visualization.arrayToDataTable([
69 ['Name', 'Parent', 'Size'],
70 ['ROOT', null, 0],""")
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]71
Leandro Lovisolo5a7d91c2021-12-01 16:03:27 -0500[diff] [blame]72 symbol_frequencies = {}
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]73
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]74 # Skip header row
75 # TODO: In the future, we could use this to automatically detect the source columns
76 next(sys.stdin)
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]77
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]78 for line in sys.stdin:
79 vals = line.rstrip().split("\t")
80 if len(vals) != 4:
81 print("ERROR: Failed to match line\n" + line)
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]82 sys.exit(1)
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]83 (filepath, symbol, vmsize, filesize) = vals
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]84
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]85 # Skip any entry where the filepath or symbol starts with '['
86 # These tend to be section meta-data and debug information
87 if filepath.startswith("[") or symbol.startswith("["):
88 continue
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]89
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]90 # Strip the leading ../../ from paths
91 while filepath.startswith("../"):
John Stilese7830652021-09-13 12:00:35 -0400[diff] [blame]92 filepath = filepath[3:];
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]93
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]94 # Files in third_party sometimes have absolute paths. Strip those:
95 if filepath.startswith("/"):
96 rel_path_start = filepath.find("third_party")
97 if rel_path_start >= 0:
98 filepath = filepath[rel_path_start:]
99 else:
100 print("ERROR: Unexpected absolute path:\n" + filepath)
101 sys.exit(1)
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]102
Brian Osmanf57f12d2021-07-19 14:24:11 -0400[diff] [blame]103 # Symbols involving C++ lambdas can contain single quotes
104 symbol = symbol.replace("'", "\\'")
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]105
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]106 # Ensure that we've added intermediate nodes for all portions of this file path
107 add_path(filepath)
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]108
Leandro Lovisolo5a7d91c2021-12-01 16:03:27 -0500[diff] [blame]109 # Ensure that our final symbol name is unique (a repeated "foo" symbol becomes "foo_1",
110 # "foo_2", etc.)
111 if symbol not in symbol_frequencies:
112 symbol_frequencies[symbol] = 1
113 else:
114 freq = symbol_frequencies[symbol]
115 symbol_frequencies[symbol] = freq + 1
116 symbol += "_" + str(freq)
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]117
118 # Append another row for our sanitized data
Leandro Lovisolo5a7d91c2021-12-01 16:03:27 -0500[diff] [blame]119 print("['%s', '%s (Path)', %d]," % (symbol, filepath, int(filesize)))
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]120
121 # HTML/script footer
122 print(""" ]);
123 tree = new google.visualization.TreeMap(document.getElementById('chart_div'));
124 tree.draw(data, {
125 generateTooltip: showTooltip
126 });
127
128 function showTooltip(row, size, value) {
129 const escapedLabel = data.getValue(row, 0)
130 .replace('&', '&amp;')
131 .replace('<', '&lt;')
132 .replace('>', '&gt;')
133 return `<div style="background:#fd9; padding:10px; border-style:solid">
134 <span style="font-family:Courier"> ${escapedLabel} <br>
135 Size: ${size} </div>`;
136 }
Brian Osman060dd702021-07-15 13:14:51 -0400[diff] [blame]137 }
Brian Osman3c1c4c02021-07-19 09:41:47 -0400[diff] [blame]138 </script>
139 </head>
140 <body>
141 <div id="chart_div" style="width: 100%; height: 100%;"></div>
142 </body>
143 </html>""")
144
145if __name__ == "__main__":
146 main()