diff options
| author | 2023-12-26 15:43:25 -0800 | |
|---|---|---|
| committer | 2023-12-26 15:43:25 -0800 | |
| commit | 225de25638f0e7bfa4d4beb2c82f4d60b971c5af (patch) | |
| tree | dd3888dedf50462bf711b8f8530ee883ba156b7d /compiler/jit/jit_logger.cc | |
| parent | c411c29cbf80bb48d63cc5bd8fbb69ae97a95902 (diff) | |
Always follow DecrementSuspendCount by broadcast
In one case we were not doing so; in another case we make it more
obvious that we are.
This appears to have been a long-standing issue, which may have been
hidden by the previous complex transitioning_to_runnable logic, and
was then exposed by a recent CL that deleted it:
https://android-review.googlesource.com/c/platform/art/+/2883289
Unfortunately, this looks quite random, and has low occurrence
probability. I don't think there is a real way to test for this
without somehow controlling scheduling.
Bug: 317036326
Bug: 297973401
Test: Host run tests, TreeHugger
Change-Id: I8f9b55d2edd878072f8486dc7e564ea901be5ec8
Diffstat (limited to 'compiler/jit/jit_logger.cc')
0 files changed, 0 insertions, 0 deletions