| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
Bug: 12929305
Change-Id: Ic896e164da0f67babdd423d5adb6e84aaa80cab2
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Moved exception delivery into common method DeliverException
- Renamed old DeliverException to QuickDeliverException since it is only
used by quick
- Fixed null checks for arrays returned by GetReference
- Standardized ArrayStoreException error message
- Added additional sleeps to ensure threads stay alive long enough in
051-thread, and that <clinit> is complete for 084-class-init
Change-Id: I9ca306896a4bd10f453150fcf3965d9750fa0cbd
|
|
|
|
| |
Change-Id: I5b7bc76a370238d810d78522bd5531600746700f
|
|
|
|
|
|
|
|
| |
We shouldn't assume we know what order threads will be started in.
Running a debug build, it sometimes takes > 1ms to start the next
thread.
Change-Id: Icb7792f83809ad414cfcdf0828cdc198698deaf1
|
|
|
|
|
|
|
|
|
|
|
|
| |
This changes Thread::Create to clean up the requested stack size before
passing it to pthread_attr_setstacksize.
I modified the test to work around the inherent raciness in the test, which
currently causes us trouble because with all the debug logging, the thread
that initializes System.out takes longer than 1ms, so the first 40 or so
threads all get woken in non-numeric order.
Change-Id: I8047bf00330aaf4210dcb53a4437b1fe0562179e
|
|
These tests are copied straight over. They'll still run, but they're
using the old system.
Change-Id: If494519e52ddf858a9febfc55bdae830468cb3c8
|