test.dart: Print stack traces of all threads when a test times out

test.dart: Print stack traces of all threads when a test times out

This helps diagnosing tests timing out when running test.py.

BUG=
R=whesse@google.com

Committed: https://github.com/dart-lang/sdk/commit/30f16437b1fcdcf81748890b486d449136808a4c

[Florian Schneider, 2016-09-07 18:18:30]

fschneider@google.com changed reviewers:
+ whesse@google.com

[Florian Schneider, 2016-09-07 18:18:31]

Would something like this work?

[rmacnak, 2016-09-07 19:33:47]

rmacnak@google.com changed reviewers:
+ rmacnak@google.com

[rmacnak, 2016-09-07 19:33:48]

https://codereview.chromium.org/2320473004/diff/1/tools/testing/dart/test_run...
File tools/testing/dart/test_runner.dart (right):

https://codereview.chromium.org/2320473004/diff/1/tools/testing/dart/test_run...
tools/testing/dart/test_runner.dart:1857: io.Process.run('gdb',
Only attempt to use gdb on Linux.

[siva, 2016-09-07 19:55:19]

asiva@google.com changed reviewers:
+ asiva@google.com

[siva, 2016-09-07 19:55:20]

https://codereview.chromium.org/2320473004/diff/1/tools/testing/dart/test_run...
File tools/testing/dart/test_runner.dart (right):

https://codereview.chromium.org/2320473004/diff/1/tools/testing/dart/test_run...
tools/testing/dart/test_runner.dart:1864: })
Do we also need a catchError handler here to print a diagnostic message if gdb
fails to start.

We also have to be sure that the bots have the process attach capability set
(see /etc/sysctl.d/10-ptrace.conf).

[Florian Schneider, 2016-09-07 21:02:58]

https://codereview.chromium.org/2320473004/diff/1/tools/testing/dart/test_run...
File tools/testing/dart/test_runner.dart (right):

https://codereview.chromium.org/2320473004/diff/1/tools/testing/dart/test_run...
tools/testing/dart/test_runner.dart:1857: io.Process.run('gdb',
On 2016/09/07 19:33:47, rmacnak wrote:
> Only attempt to use gdb on Linux.

Done.

https://codereview.chromium.org/2320473004/diff/1/tools/testing/dart/test_run...
tools/testing/dart/test_runner.dart:1864: })
On 2016/09/07 19:55:20, siva wrote:
> Do we also need a catchError handler here to print a diagnostic message if gdb
> fails to start.
> 

Yes. Added printing. Also, I should only invoke gdb if it is available on the
system. Is there a standard way in Dart to find if gdb is installed on a system?

> We also have to be sure that the bots have the process attach capability set
> (see /etc/sysctl.d/10-ptrace.conf).

[Bill Hesse, 2016-09-08 14:58:01]

lgtm

https://codereview.chromium.org/2320473004/diff/1/tools/testing/dart/test_run...
File tools/testing/dart/test_runner.dart (right):

https://codereview.chromium.org/2320473004/diff/1/tools/testing/dart/test_run...
tools/testing/dart/test_runner.dart:1864: })
On 2016/09/07 21:02:58, Florian Schneider wrote:
> On 2016/09/07 19:55:20, siva wrote:
> > Do we also need a catchError handler here to print a diagnostic message if
gdb
> > fails to start.
> > 
> 
> Yes. Added printing. Also, I should only invoke gdb if it is available on the
> system. Is there a standard way in Dart to find if gdb is installed on a
system?
> 
> > We also have to be sure that the bots have the process attach capability set
> > (see /etc/sysctl.d/10-ptrace.conf).
> 

I would just run "which gdb" in a subprocess, and check the return code.

[siva, 2016-09-08 16:42:41]

/usr/bin/eu-stack also seems to produce a stack trace
for a process, something like

eu-stack -p 1466

produces

PID 1466 - process
TID 1466:
#0  0x00007fa639c17404 pthread_cond_wait@@GLIBC_2.3.2
#1  0x000000000076b326 dart::Monitor::WaitMicros(long)
#2  0x000000000058e07a Dart_RunLoop
#3  0x0000000000566bde dart::bin::RunMainIsolate(char const*,
dart::bin::CommandLineOptions*)
#4  0x0000000000560398 dart::bin::main(int, char**)
#5  0x0000000000560549 main
#6  0x00007fa638c3bf45 __libc_start_main
#7  0x00000000005650cb _start
TID 1467:
#0  0x00007fa638d14a13 epoll_wait
#1  0x000000000059e995 dart::bin::EventHandlerImplementation::Poll(unsigned
long)
#2  0x0000000000599c39 dart::bin::ThreadStart(void*)
#3  0x00007fa639c13184 start_thread
#4  0x00007fa638d1437d __clone
TID 1468:
#0  0x00007fa639c177be pthread_cond_timedwait@@GLIBC_2.3.2
#1  0x000000000076b405 dart::Monitor::WaitMicros(long)
#2  0x000000000083b6ec dart::ThreadInterrupter::ThreadMain(unsigned long)
#3  0x000000000076a826 dart::ThreadStart(void*)
#4  0x00007fa639c13184 start_thread
#5  0x00007fa638d1437d __clone
TID 1469:
#0  0x00007fa639c17404 pthread_cond_wait@@GLIBC_2.3.2
#1  0x000000000076b326 dart::Monitor::WaitMicros(long)
#2  0x00000000005e158a dart::BackgroundCompiler::Run()
#3  0x000000000083c5ac dart::ThreadPool::Worker::Loop()
#4  0x000000000083c76f dart::ThreadPool::Worker::Main(unsigned long)
#5  0x000000000076a826 dart::ThreadStart(void*)
#6  0x00007fa639c13184 start_thread
#7  0x00007fa638d1437d __clone
TID 1470:
#0  0x00007fa638d0572d write
#1  0x00007fa638c92f03 _IO_file_write@@GLIBC_2.2.5
#2  0x00007fa638c943dc _IO_do_write@@GLIBC_2.2.5
#3  0x00007fa638c947b3 _IO_file_overflow@@GLIBC_2.2.5
#4  0x00007fa638c8b4e9 fputc
#5  0x00000000005677dd
dart::bin::Builtin_Builtin_PrintString(_Dart_NativeArguments*)
#6  0x00000000006ffea9
dart::NativeEntry::NativeCallWrapperNoStackCheck(_Dart_NativeArguments*, void
(*)(_Dart_NativeArguments*))
#7  0x00007fa639e956af

[Florian Schneider, 2016-09-08 17:14:42]

On 2016/09/08 16:42:41, siva wrote:
> /usr/bin/eu-stack also seems to produce a stack trace
> for a process, something like
> 
> eu-stack -p 1466
> 
> produces
> 
> PID 1466 - process
> TID 1466:
> #0  0x00007fa639c17404 mailto:pthread_cond_wait@@GLIBC_2.3.2
> #1  0x000000000076b326 dart::Monitor::WaitMicros(long)
> #2  0x000000000058e07a Dart_RunLoop
> #3  0x0000000000566bde dart::bin::RunMainIsolate(char const*,
> dart::bin::CommandLineOptions*)
> #4  0x0000000000560398 dart::bin::main(int, char**)
> #5  0x0000000000560549 main
> #6  0x00007fa638c3bf45 __libc_start_main
> #7  0x00000000005650cb _start
> TID 1467:
> #0  0x00007fa638d14a13 epoll_wait
> #1  0x000000000059e995 dart::bin::EventHandlerImplementation::Poll(unsigned
> long)
> #2  0x0000000000599c39 dart::bin::ThreadStart(void*)
> #3  0x00007fa639c13184 start_thread
> #4  0x00007fa638d1437d __clone
> TID 1468:
> #0  0x00007fa639c177be mailto:pthread_cond_timedwait@@GLIBC_2.3.2
> #1  0x000000000076b405 dart::Monitor::WaitMicros(long)
> #2  0x000000000083b6ec dart::ThreadInterrupter::ThreadMain(unsigned long)
> #3  0x000000000076a826 dart::ThreadStart(void*)
> #4  0x00007fa639c13184 start_thread
> #5  0x00007fa638d1437d __clone
> TID 1469:
> #0  0x00007fa639c17404 mailto:pthread_cond_wait@@GLIBC_2.3.2
> #1  0x000000000076b326 dart::Monitor::WaitMicros(long)
> #2  0x00000000005e158a dart::BackgroundCompiler::Run()
> #3  0x000000000083c5ac dart::ThreadPool::Worker::Loop()
> #4  0x000000000083c76f dart::ThreadPool::Worker::Main(unsigned long)
> #5  0x000000000076a826 dart::ThreadStart(void*)
> #6  0x00007fa639c13184 start_thread
> #7  0x00007fa638d1437d __clone
> TID 1470:
> #0  0x00007fa638d0572d write
> #1  0x00007fa638c92f03 mailto:_IO_file_write@@GLIBC_2.2.5
> #2  0x00007fa638c943dc mailto:_IO_do_write@@GLIBC_2.2.5
> #3  0x00007fa638c947b3 mailto:_IO_file_overflow@@GLIBC_2.2.5
> #4  0x00007fa638c8b4e9 fputc
> #5  0x00000000005677dd
> dart::bin::Builtin_Builtin_PrintString(_Dart_NativeArguments*)
> #6  0x00000000006ffea9
> dart::NativeEntry::NativeCallWrapperNoStackCheck(_Dart_NativeArguments*, void
> (*)(_Dart_NativeArguments*))
> #7  0x00007fa639e956af

Thanks. I assume it needs the same permission as gdb, right?

[Florian Schneider, 2016-09-08 17:49:19]

Description was changed from

==========
test.dart: Print stack traces of all threads when a test times out

This helps diagnosing tests timing out when running test.py.

BUG=
==========

to

==========
test.dart: Print stack traces of all threads when a test times out

This helps diagnosing tests timing out when running test.py.

BUG=
R=whesse@google.com

Committed:
https://github.com/dart-lang/sdk/commit/30f16437b1fcdcf81748890b486d449136808a4c
==========

[Florian Schneider, 2016-09-08 17:49:21]

Committed patchset #3 (id:40001) manually as
30f16437b1fcdcf81748890b486d449136808a4c (presubmit successful).