blob: 478fc049102d6bf540790a6a0ff1d608058914ed [file] [log] [blame] [view]
andybonse6a8f2bd2015-08-31 22:46:011# Tips for debugging on Linux
andybons3322f762015-08-24 21:37:092
andybonsad92aa32015-08-31 02:27:443This page is for Chromium-specific debugging tips; learning how to run gdb is
4out of scope.
andybons3322f762015-08-24 21:37:095
andybonsad92aa32015-08-31 02:27:446[TOC]
andybons3322f762015-08-24 21:37:097
8## Symbolized stack trace
9
andybonsad92aa32015-08-31 02:27:4410The sandbox can interfere with the internal symbolizer. Use `--no-sandbox` (but
11keep this temporary) or an external symbolizer (see
12`tools/valgrind/asan/asan_symbolize.py`).
andybons3322f762015-08-24 21:37:0913
andybonsad92aa32015-08-31 02:27:4414Generally, do not use `--no-sandbox` on waterfall bots, sandbox testing is
15needed. Talk to security@chromium.org.
andybons3322f762015-08-24 21:37:0916
17## GDB
andybonsad92aa32015-08-31 02:27:4418
nodira6074d4c2015-09-01 04:26:4519*** promo
20GDB-7.7 is required in order to debug Chrome on Linux.
21***
andybons3322f762015-08-24 21:37:0922
23Any prior version will fail to resolve symbols or segfault.
24
25### Basic browser process debugging
26
andybonsad92aa32015-08-31 02:27:4427 gdb -tui -ex=r --args out/Debug/chrome --disable-seccomp-sandbox \
28 http://google.com
andybons3322f762015-08-24 21:37:0929
30### Allowing attaching to foreign processes
andybonsad92aa32015-08-31 02:27:4431
32On distributions that use the
33[Yama LSM](https://www.kernel.org/doc/Documentation/security/Yama.txt) (that
34includes Ubuntu and Chrome OS), process A can attach to process B only if A is
35an ancestor of B.
andybons3322f762015-08-24 21:37:0936
37You will probably want to disable this feature by using
andybonsad92aa32015-08-31 02:27:4438
39 echo 0 | sudo tee /proc/sys/kernel/yama/ptrace_scope
andybons3322f762015-08-24 21:37:0940
41If you don't you'll get an error message such as "Could not attach to process".
42
andybonsad92aa32015-08-31 02:27:4443Note that you'll also probably want to use `--no-sandbox`, as explained below.
andybons3322f762015-08-24 21:37:0944
45### Multiprocess Tricks
andybonsad92aa32015-08-31 02:27:4446
andybons3322f762015-08-24 21:37:0947#### Getting renderer subprocesses into gdb
andybonsad92aa32015-08-31 02:27:4448
49Since Chromium itself spawns the renderers, it can be tricky to grab a
50particular with gdb. This command does the trick:
51
andybons3322f762015-08-24 21:37:0952```
53chrome --no-sandbox --renderer-cmd-prefix='xterm -title renderer -e gdb --args'
54```
andybonsad92aa32015-08-31 02:27:4455
56The `--no-sandbox` flag is needed because otherwise the seccomp sandbox will
57kill the renderer process on startup, or the setuid sandbox will prevent xterm's
58execution. The "xterm" is necessary or gdb will run in the current terminal,
59which can get particularly confusing since it's running in the background, and
60if you're also running the main process in gdb, won't work at all (the two
61instances will fight over the terminal). To auto-start the renderers in the
62debugger, send the "run" command to the debugger:
63
nodira6074d4c2015-09-01 04:26:4564 chrome --no-sandbox --renderer-cmd-prefix='xterm -title renderer -e gdb \
65 -ex run --args
andybonsad92aa32015-08-31 02:27:4466
andybons3322f762015-08-24 21:37:0967If you're using Emacs and `M-x gdb`, you can do
andybons3322f762015-08-24 21:37:0968
andybonsad92aa32015-08-31 02:27:4469 chrome "--renderer-cmd-prefix=gdb --args"
andybons3322f762015-08-24 21:37:0970
nodira6074d4c2015-09-01 04:26:4571*** note
andybonsad92aa32015-08-31 02:27:4472Note: using the `--renderer-cmd-prefix` option bypasses the zygote launcher, so
73the renderers won't be sandboxed. It is generally not an issue, except when you
74are trying to debug interactions with the sandbox. If that's what you are doing,
75you will need to attach your debugger to a running renderer process (see below).
nodira6074d4c2015-09-01 04:26:4576***
andybons3322f762015-08-24 21:37:0977
andybonsad92aa32015-08-31 02:27:4478You may also want to pass `--disable-hang-monitor` to suppress the hang monitor,
79which is rather annoying.
80
81You can also use `--renderer-startup-dialog` and attach to the process in order
82to debug the renderer code. Go to
xiaoyin.l1003c0b2016-12-06 02:51:1783https://www.chromium.org/blink/getting-started-with-blink-debugging for more
andybonsad92aa32015-08-31 02:27:4484information on how this can be done.
andybons3322f762015-08-24 21:37:0985
86#### Choosing which renderers to debug
andybons3322f762015-08-24 21:37:0987
andybonsad92aa32015-08-31 02:27:4488If you are starting multiple renderers then the above means that multiple gdb's
89start and fight over the console. Instead, you can set the prefix to point to
90this shell script:
91
92```sh
andybons3322f762015-08-24 21:37:0993#!/bin/sh
94
95echo "**** Child $$ starting: y to debug"
96read input
97if [ "$input" = "y" ] ; then
98 gdb --args $*
99else
100 $*
101fi
102```
103
104#### Selective breakpoints
andybonsad92aa32015-08-31 02:27:44105
106When debugging both the browser and renderer process, you might want to have
107separate set of breakpoints to hit. You can use gdb's command files to
108accomplish this by putting breakpoints in separate files and instructing gdb to
109load them.
andybons3322f762015-08-24 21:37:09110
111```
andybonsad92aa32015-08-31 02:27:44112gdb -x ~/debug/browser --args chrome --no-sandbox --disable-hang-monitor \
113 --renderer-cmd-prefix='xterm -title renderer -e gdb -x ~/debug/renderer \
114 --args '
andybons3322f762015-08-24 21:37:09115```
116
andybonsad92aa32015-08-31 02:27:44117Also, instead of running gdb, you can use the script above, which let's you
118select which renderer process to debug. Note: you might need to use the full
119path to the script and avoid `$HOME` or `~/.`
andybons3322f762015-08-24 21:37:09120
121#### Connecting to a running renderer
122
andybonsad92aa32015-08-31 02:27:44123Usually `ps aux | grep chrome` will not give very helpful output. Try
124`pstree -p | grep chrome` to get something like
andybons3322f762015-08-24 21:37:09125
126```
127 | |-bash(21969)---chrome(672)-+-chrome(694)
128 | | |-chrome(695)---chrome(696)-+-{chrome}(697)
129 | | | \-{chrome}(709)
130 | | |-{chrome}(675)
131 | | |-{chrome}(678)
132 | | |-{chrome}(679)
133 | | |-{chrome}(680)
134 | | |-{chrome}(681)
135 | | |-{chrome}(682)
136 | | |-{chrome}(684)
137 | | |-{chrome}(685)
138 | | |-{chrome}(705)
139 | | \-{chrome}(717)
140```
141
andybonsad92aa32015-08-31 02:27:44142Most of those are threads. In this case the browser process would be 672 and the
143(sole) renderer process is 696. You can use `gdb -p 696` to attach.
144Alternatively, you might find out the process ID from Chrome's built-in Task
145Manager (under the Tools menu). Right-click on the Task Manager, and enable
146"Process ID" in the list of columns.
andybons3322f762015-08-24 21:37:09147
andybonsad92aa32015-08-31 02:27:44148Note: by default, sandboxed processes can't be attached by a debugger. To be
149able to do so, you will need to pass the `--allow-sandbox-debugging` option.
andybons3322f762015-08-24 21:37:09150
andybonsad92aa32015-08-31 02:27:44151If the problem only occurs with the seccomp sandbox enabled (and the previous
152tricks don't help), you could try enabling core-dumps (see the **Core files**
153section). That would allow you to get a backtrace and see some local variables,
154though you won't be able to step through the running program.
andybons3322f762015-08-24 21:37:09155
andybonsad92aa32015-08-31 02:27:44156Note: If you're interested in debugging LinuxSandboxIPC process, you can attach
157to 694 in the above diagram. The LinuxSandboxIPC process has the same command
158line flag as the browser process so that it's easy to identify it if you run
159`pstree -pa`.
andybons3322f762015-08-24 21:37:09160
161#### Getting GPU subprocesses into gdb
andybons3322f762015-08-24 21:37:09162
andybonsad92aa32015-08-31 02:27:44163Use `--gpu-launcher` flag instead of `--renderer-cmd-prefix` in the instructions
164for renderer above.
165
166#### Getting `browser_tests` launched browsers into gdb
167
168Use environment variable `BROWSER_WRAPPER` instead of `--renderer-cmd-prefix`
169switch in the instructions above.
andybons3322f762015-08-24 21:37:09170
171Example:
andybonsad92aa32015-08-31 02:27:44172
173```shell
174BROWSER_WRAPPER='xterm -title renderer -e gdb --eval-command=run \
175 --eval-command=quit --args' out/Debug/browser_tests --gtest_filter=Print
176```
andybons3322f762015-08-24 21:37:09177
178#### Plugin Processes
andybons3322f762015-08-24 21:37:09179
andybonsad92aa32015-08-31 02:27:44180Same strategies as renderers above, but the flag is called `--plugin-launcher`:
181
182 chrome --plugin-launcher='xterm -e gdb --args'
183
nodira6074d4c2015-09-01 04:26:45184*** note
185Note: For now, this does not currently apply to PPAPI plugins because they
186currently run in the renderer process.
187***
andybons3322f762015-08-24 21:37:09188
189#### Single-Process mode
andybons3322f762015-08-24 21:37:09190
andybonsad92aa32015-08-31 02:27:44191Depending on whether it's relevant to the problem, it's often easier to just run
192in "single process" mode where the renderer threads are in-process. Then you can
193just run gdb on the main process.
andybons3322f762015-08-24 21:37:09194
andybonsad92aa32015-08-31 02:27:44195 gdb --args chrome --single-process
196
197Currently, the `--disable-gpu` flag is also required, as there are known crashes
198that occur under TextureImageTransportSurface without it. The crash described in
xiaoyin.l1003c0b2016-12-06 02:51:17199https://crbug.com/361689 can also sometimes occur, but that crash can be
andybonsad92aa32015-08-31 02:27:44200continued from without harm.
201
202Note that for technical reasons plugins cannot be in-process, so
203`--single-process` only puts the renderers in the browser process. The flag is
204still useful for debugging plugins (since it's only two processes instead of
205three) but you'll still need to use `--plugin-launcher` or another approach.
andybons3322f762015-08-24 21:37:09206
207### Printing Chromium types
andybons3322f762015-08-24 21:37:09208
Tom Andersonf06ac382019-04-10 03:49:38209gdb 7 lets us use Python to write pretty-printers for Chromium types. See
210[gdbinit](https://chromium.googlesource.com/chromium/src/+/master/docs/gdbinit.md)
211to enable pretty-printing of Chromium types. This will import Blink
212pretty-printers as well.
Kenichi Ishibashie17b8d9f2018-04-26 03:32:46213
andybonsad92aa32015-08-31 02:27:44214Pretty printers for std types shouldn't be necessary in gdb 7, but they're
215provided here in case you're using an older gdb. Put the following into
216`~/.gdbinit`:
217
andybons3322f762015-08-24 21:37:09218```
219# Print a C++ string.
220define ps
221 print $arg0.c_str()
222end
223
224# Print a C++ wstring or wchar_t*.
225define pws
226 printf "\""
227 set $c = (wchar_t*)$arg0
228 while ( *$c )
229 if ( *$c > 0x7f )
230 printf "[%x]", *$c
231 else
232 printf "%c", *$c
233 end
234 set $c++
235 end
236 printf "\"\n"
237end
238```
239
240[More STL GDB macros](http://www.yolinux.com/TUTORIALS/src/dbinit_stl_views-1.01.txt)
241
242### Graphical Debugging Aid for Chromium Views
243
244The following link describes a tool that can be used on Linux, Windows and Mac under GDB.
245
andybonsad92aa32015-08-31 02:27:44246[graphical_debugging_aid_chromium_views](graphical_debugging_aid_chromium_views.md)
andybons3322f762015-08-24 21:37:09247
248### Faster startup
249
andybonsad92aa32015-08-31 02:27:44250Use the `gdb-add-index` script (e.g.
251`build/gdb-add-index out/Debug/browser_tests`)
andybons3322f762015-08-24 21:37:09252
andybonsad92aa32015-08-31 02:27:44253Only makes sense if you run the binary multiple times or maybe if you use the
nodira6074d4c2015-09-01 04:26:45254component build since most `.so` files won't require reindexing on a rebuild.
andybons3322f762015-08-24 21:37:09255
andybonsad92aa32015-08-31 02:27:44256See
257https://groups.google.com/a/chromium.org/forum/#!searchin/chromium-dev/gdb-add-index/chromium-dev/ELRuj1BDCL4/5Ki4LGx41CcJ
258for more info.
andybons3322f762015-08-24 21:37:09259
andybons8c02a1f2015-09-04 17:02:32260You can improve GDB load time significantly at the cost of link time by
brettw20d800c2016-04-12 00:10:49261splitting symbols from the object files. In GN, set `use_debug_fission=false` in
262your "gn args".
andybons3322f762015-08-24 21:37:09263
Tom Anderson71df8872018-06-21 19:02:25264### Source level debug with -fdebug-compilation-dir
Takuto Ikuta3ae0e03b2018-05-18 06:10:40265
Nico Weber8940a782018-10-22 23:28:28266When `strip_absolute_paths_from_debug_symbols` is enabled (which is the
Tom Andersonf06ac382019-04-10 03:49:38267default), gdb may not be able to find debug files, making source-level debugging
268impossible. See
269[gdbinit](https://chromium.googlesource.com/chromium/src/+/master/docs/gdbinit.md)
270to configure gdb to be able to find debug files.
Takuto Ikuta3ae0e03b2018-05-18 06:10:40271
andybons3322f762015-08-24 21:37:09272## Core files
andybons3322f762015-08-24 21:37:09273
andybonsad92aa32015-08-31 02:27:44274`ulimit -c unlimited` should cause all Chrome processes (run from that shell) to
275dump cores, with the possible exception of some sandboxed processes.
andybons3322f762015-08-24 21:37:09276
andybonsad92aa32015-08-31 02:27:44277Some sandboxed subprocesses might not dump cores unless you pass the
278`--allow-sandbox-debugging` flag.
279
280If the problem is a freeze rather than a crash, you may be able to trigger a
281core-dump by sending SIGABRT to the relevant process:
282
283 kill -6 [process id]
andybons3322f762015-08-24 21:37:09284
285## Breakpad minidump files
286
andybonsad92aa32015-08-31 02:27:44287See [linux_minidump_to_core.md](linux_minidump_to_core.md)
andybons3322f762015-08-24 21:37:09288
289## Running Tests
andybonsad92aa32015-08-31 02:27:44290
291Many of our tests bring up windows on screen. This can be annoying (they steal
292your focus) and hard to debug (they receive extra events as you mouse over them).
293Instead, use `Xvfb` or `Xephyr` to run a nested X session to debug them, as
Marijn Kruisselbrinkeebac46b2019-05-28 19:33:16294outlined on [testing/web_tests_linux.md](testing/web_tests_linux.md).
andybons3322f762015-08-24 21:37:09295
296### Browser tests
andybonsad92aa32015-08-31 02:27:44297
298By default the `browser_tests` forks a new browser for each test. To debug the
299browser side of a single test, use a command like
300
andybons3322f762015-08-24 21:37:09301```
302gdb --args out/Debug/browser_tests --single_process --gtest_filter=MyTestName
303```
andybonsad92aa32015-08-31 02:27:44304
305**note the underscore in `single_process`** -- this makes the test harness and
306browser process share the outermost process.
andybons3322f762015-08-24 21:37:09307
308
309To debug a renderer process in this case, use the tips above about renderers.
310
Kent Tamura59ffb022018-11-27 05:30:56311### Web tests
andybonsad92aa32015-08-31 02:27:44312
Marijn Kruisselbrinkeebac46b2019-05-28 19:33:16313See [testing/web_tests_linux.md](testing/web_tests_linux.md) for some tips. In particular,
Kent Tamura59ffb022018-11-27 05:30:56314note that it's possible to debug a web test via `ssh`ing to a Linux box; you
andybonsad92aa32015-08-31 02:27:44315don't need anything on screen if you use `Xvfb`.
andybons3322f762015-08-24 21:37:09316
317### UI tests
andybons3322f762015-08-24 21:37:09318
andybonsad92aa32015-08-31 02:27:44319UI tests are run in forked browsers. Unlike browser tests, you cannot do any
320single process tricks here to debug the browser. See below about
321`BROWSER_WRAPPER`.
322
323To pass flags to the browser, use a command line like
324`--extra-chrome-flags="--foo --bar"`.
andybons3322f762015-08-24 21:37:09325
326### Timeouts
andybonsad92aa32015-08-31 02:27:44327
328UI tests have a confusing array of timeouts in place. (Pawel is working on
329reducing the number of timeouts.) To disable them while you debug, set the
330timeout flags to a large value:
331
332* `--test-timeout=100000000`
333* `--ui-test-action-timeout=100000000`
334* `--ui-test-terminate-timeout=100000000`
andybons3322f762015-08-24 21:37:09335
336### To replicate Window Manager setup on the bots
andybonsad92aa32015-08-31 02:27:44337
338Chromium try bots and main waterfall's bots run tests under Xvfb&openbox
339combination. Xvfb is an X11 server that redirects the graphical output to the
340memory, and openbox is a simple window manager that is running on top of Xvfb.
341The behavior of openbox is markedly different when it comes to focus management
342and other window tasks, so test that runs fine locally may fail or be flaky on
343try bots. To run the tests on a local machine as on a bot, follow these steps:
andybons3322f762015-08-24 21:37:09344
345Make sure you have openbox:
andybonsad92aa32015-08-31 02:27:44346
347 apt-get install openbox
348
andybons3322f762015-08-24 21:37:09349Start Xvfb and openbox on a particular display:
andybonsad92aa32015-08-31 02:27:44350
351 Xvfb :6.0 -screen 0 1280x1024x24 & DISPLAY=:6.0 openbox &
352
andybons3322f762015-08-24 21:37:09353Run your tests with graphics output redirected to that display:
andybonsad92aa32015-08-31 02:27:44354
355 DISPLAY=:6.0 out/Debug/browser_tests --gtest_filter="MyBrowserTest.MyActivateWindowTest"
356
andybons3322f762015-08-24 21:37:09357You can look at a snapshot of the output by:
andybonsad92aa32015-08-31 02:27:44358
359 xwd -display :6.0 -root | xwud
andybons3322f762015-08-24 21:37:09360
361Alternatively, you can use testing/xvfb.py to set up your environment for you:
andybons3322f762015-08-24 21:37:09362
thomasanderson3d074282016-12-06 18:21:12363 testing/xvfb.py out/Debug/browser_tests \
andybonsad92aa32015-08-31 02:27:44364 --gtest_filter="MyBrowserTest.MyActivateWindowTest"
andybons3322f762015-08-24 21:37:09365
nodira6074d4c2015-09-01 04:26:45366### BROWSER_WRAPPER
andybonsad92aa32015-08-31 02:27:44367
368You can also get the browser under a debugger by setting the `BROWSER_WRAPPER`
369environment variable. (You can use this for `browser_tests` too, but see above
370for discussion of a simpler way.)
371
372 BROWSER_WRAPPER='xterm -e gdb --args' out/Debug/browser_tests
andybons3322f762015-08-24 21:37:09373
qyearsleyc0dc6f42016-12-02 22:13:39374### Replicating try bot Slowness
andybons3322f762015-08-24 21:37:09375
qyearsleyc0dc6f42016-12-02 22:13:39376Try bots are pretty stressed, and can sometimes expose timing issues you can't
andybonsad92aa32015-08-31 02:27:44377normally reproduce locally.
andybons3322f762015-08-24 21:37:09378
andybonsad92aa32015-08-31 02:27:44379You can simulate this by shutting down all but one of the CPUs
380(http://www.cyberciti.biz/faq/debian-rhel-centos-redhat-suse-hotplug-cpu/) and
381running a CPU loading tool (e.g., http://www.devin.com/lookbusy/). Now run your
qyearsleyc0dc6f42016-12-02 22:13:39382test. It will run slowly, but any flakiness found by the try bot should replicate
andybonsad92aa32015-08-31 02:27:44383locally now - and often nearly 100% of the time.
andybons3322f762015-08-24 21:37:09384
385## Logging
andybons3322f762015-08-24 21:37:09386
andybonsad92aa32015-08-31 02:27:44387### Seeing all LOG(foo) messages
388
389Default log level hides `LOG(INFO)`. Run with `--log-level=0` and
390`--enable-logging=stderr` flags.
391
qyearsleyc0dc6f42016-12-02 22:13:39392Newer versions of Chromium with VLOG may need --v=1 too. For more VLOG tips, see
xiaoyin.l1003c0b2016-12-06 02:51:17393[the chromium-dev thread](https://groups.google.com/a/chromium.org/group/chromium-dev/browse_thread/thread/dcd0cd7752b35de6?pli=1).
andybons3322f762015-08-24 21:37:09394
395### Seeing IPC debug messages
andybons3322f762015-08-24 21:37:09396
andybonsad92aa32015-08-31 02:27:44397Run with `CHROME_IPC_LOGGING=1` eg.
398
399 CHROME_IPC_LOGGING=1 out/Debug/chrome
400
401or within gdb:
402
403 set environment CHROME_IPC_LOGGING 1
404
405If some messages show as unknown, check if the list of IPC message headers in
nodira6074d4c2015-09-01 04:26:45406[chrome/common/logging_chrome.cc](/chrome/common/logging_chrome.cc) is
thakis3e861de2016-06-14 14:24:01407up to date. In case this file reference goes out of date, try looking for usage
nodira6074d4c2015-09-01 04:26:45408of macros like `IPC_MESSAGE_LOG_ENABLED` or `IPC_MESSAGE_MACROS_LOG_ENABLED`.
andybons3322f762015-08-24 21:37:09409
andybons3322f762015-08-24 21:37:09410## Profiling
andybonsad92aa32015-08-31 02:27:44411
412See
413https://sites.google.com/a/chromium.org/dev/developers/profiling-chromium-and-webkit
sbc9f033f82015-11-26 00:50:52414and [Linux Profiling](linux_profiling.md).
andybons3322f762015-08-24 21:37:09415
416## i18n
andybons3322f762015-08-24 21:37:09417
andybonsad92aa32015-08-31 02:27:44418We obey your system locale. Try something like:
andybons3322f762015-08-24 21:37:09419
andybonsad92aa32015-08-31 02:27:44420 LANG=ja_JP.UTF-8 out/Debug/chrome
421
422If this doesn't work, make sure that the `LANGUAGE`, `LC_ALL` and `LC_MESSAGE`
423environment variables aren't set -- they have higher priority than LANG in the
424order listed. Alternatively, just do this:
425
426 LANGUAGE=fr out/Debug/chrome
427
428Note that because we use GTK, some locale data comes from the system -- for
429example, file save boxes and whether the current language is considered RTL.
430Without all the language data available, Chrome will use a mixture of your
431system language and the language you run Chrome in.
andybons3322f762015-08-24 21:37:09432
433Here's how to install the Arabic (ar) and Hebrew (he) language packs:
andybonsad92aa32015-08-31 02:27:44434
435 sudo apt-get install language-pack-ar language-pack-he \
436 language-pack-gnome-ar language-pack-gnome-he
437
andybons3322f762015-08-24 21:37:09438Note that the `--lang` flag does **not** work properly for this.
439
Tom Anderson287339e2018-08-22 21:52:02440On non-Debian systems, you need the `gtk30.mo` files. (Please update these docs
andybonsad92aa32015-08-31 02:27:44441with the appropriate instructions if you know what they are.)
andybons3322f762015-08-24 21:37:09442
443## Breakpad
andybonsad92aa32015-08-31 02:27:44444
Daniel Bratellf73f0df2018-09-24 13:52:49445See the last section of [Linux Crash Dumping](linux_crash_dumping.md).
andybons3322f762015-08-24 21:37:09446
447## Drag and Drop
andybonsad92aa32015-08-31 02:27:44448
449If you break in a debugger during a drag, Chrome will have grabbed your mouse
450and keyboard so you won't be able to interact with the debugger! To work around
451this, run via `Xephyr`. Instructions for how to use `Xephyr` are on the
Marijn Kruisselbrinkeebac46b2019-05-28 19:33:16452[Running web tests on Linux](testing/web_tests_linux.md) page.
andybons3322f762015-08-24 21:37:09453
454## Tracking Down Bugs
455
456### Isolating Regressions
andybons3322f762015-08-24 21:37:09457
andybonsad92aa32015-08-31 02:27:44458Old builds are archived here:
xiaoyin.l1003c0b2016-12-06 02:51:17459https://build.chromium.org/buildbot/snapshots/chromium-rel-linux/
nodira6074d4c2015-09-01 04:26:45460(TODO: does not exist).
andybonsad92aa32015-08-31 02:27:44461
462`tools/bisect-builds.py` in the tree automates bisecting through the archived
463builds. Despite a computer science education, I am still amazed how quickly
464binary search will find its target.
andybons3322f762015-08-24 21:37:09465
466### Screen recording for bug reports
andybonsad92aa32015-08-31 02:27:44467
468 sudo apt-get install gtk-recordmydesktop
andybons3322f762015-08-24 21:37:09469
470## Version-specific issues
471
472### Google Chrome
andybonsad92aa32015-08-31 02:27:44473
474Google Chrome binaries don't include symbols. Googlers can read where to get
475symbols from
476[the Google-internal wiki](http://wiki/Main/ChromeOfficialBuildLinux#The_Build_Archive).
andybons3322f762015-08-24 21:37:09477
478### Ubuntu Chromium
andybonsad92aa32015-08-31 02:27:44479
480Since we don't build the Ubuntu packages (Ubuntu does) we can't get useful
481backtraces from them. Direct users to https://wiki.ubuntu.com/Chromium/Debugging
andybons3322f762015-08-24 21:37:09482
483### Fedora's Chromium
andybonsad92aa32015-08-31 02:27:44484
485Like Ubuntu, but direct users to
486https://fedoraproject.org/wiki/TomCallaway/Chromium_Debug
andybons3322f762015-08-24 21:37:09487
488### Xlib
andybonsad92aa32015-08-31 02:27:44489
andybons3322f762015-08-24 21:37:09490If you're trying to track down X errors like:
andybonsad92aa32015-08-31 02:27:44491
andybons3322f762015-08-24 21:37:09492```
493The program 'chrome' received an X Window System error.
494This probably reflects a bug in the program.
495The error was 'BadDrawable (invalid Pixmap or Window parameter)'.
496```
andybonsad92aa32015-08-31 02:27:44497
andybons3322f762015-08-24 21:37:09498Some strategies are:
andybonsad92aa32015-08-31 02:27:44499
500* pass `--sync` on the command line to make all X calls synchronous
501* run chrome via [xtrace](http://xtrace.alioth.debian.org/)
502* turn on IPC debugging (see above section)
andybons3322f762015-08-24 21:37:09503
504### Window Managers
andybons3322f762015-08-24 21:37:09505
andybonsad92aa32015-08-31 02:27:44506To test on various window managers, you can use a nested X server like `Xephyr`.
507Instructions for how to use `Xephyr` are on the
Kent Tamura59ffb022018-11-27 05:30:56508[Running web tests on Linux](web_tests_linux.md) page.
andybonsad92aa32015-08-31 02:27:44509
510If you need to test something with hardware accelerated compositing
511(e.g., compiz), you can use `Xgl` (`sudo apt-get install xserver-xgl`). E.g.:
512
513 Xgl :1 -ac -accel glx:pbuffer -accel xv:pbuffer -screen 1024x768
514
andybons3322f762015-08-24 21:37:09515## Mozilla Tips
andybonsad92aa32015-08-31 02:27:44516
517https://developer.mozilla.org/en/Debugging_Mozilla_on_Linux_FAQ