| 160ab8efc1d814a335e3b31d8ba1344ebd135053 |
17-Jan-2017 |
Mathieu Chartier <mathieuc@google.com> |
Do test_getStackTraces last for test 129
Doing this check last leaves time for the heap task daemon to call
TaskProcessor::RunAllTasks. This should hopefully fix the flakiness
of test 129 during startup.
Test: test-art-host-run-test
Bug: 34342857
Change-Id: I85f8d78083092df802728c59a4c86f63be8d949a
ain.java
|
| 91011afe8ed113b02878a4186fc103eaaef3c102 |
13-Jan-2017 |
Nicolas Geoffray <ngeoffray@google.com> |
Make 129-ThreadGetId more robust.
Make sure we get the heap task daemon, to avoid a race with the runtime
creating the stack trace, and then updating the peer.
test: 129-ThreadGetId
bug: 28261069
Change-Id: I739ab6cd0180e2be07b7cecac6ad8a905a7c9cd0
ain.java
|
| 4201cf014cfe00c145edc0b32bf30b1ceaf1495f |
12-Jan-2017 |
Mathieu Chartier <mathieuc@google.com> |
Avoid suspending heap task thread for getting stack traces
Instead of suspending the heap task thread, GetThreadStack (called by
VMStack_fillStackTraceElements and VMStack_getThreadStackTrace) will
return an empty thread stack. This fixes possible deadlocks caused by
suspending the GC thread and doing allocations for the stack trace.
Bug: 28261069
Test: test-art-host
Change-Id: I45a0b8ac94a99d6bbcfcdc2b41afadf941ec0138
ain.java
|
| 905f5914fec4540156309c4de9bec3591d11bedc |
12-Dec-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add regression test for deadlock caused by Thread.getAllStackTraces
Also added stress test for Thread.getAllStackTraces. The test calls
Thread.getAllStackTraces from multiple threads to stress race
conditions which can cause deadlocks.
Bug: 18661622
Change-Id: Ideee7073fdcce784414808d08eca9ea310240073
ain.java
|