The latest merge pulled in the ParallelArray work and this caused a large number of jit-test failures. For now I disabled the ParallelArray tests on the BC branch (see intrinsic_ParallelTestsShouldPass). To fix in this bug: (1) Some tests don't like us changing the use count when type information changes.. (2) Baseline has to support callsite cloning. (3) The hack in intrinsic_ParallelTestsShouldPass has to be removed.
- Enables the parallel array tests again. - Only calls resetUseCount if baseline is enabled (the call does not exist on m-c so it's good to minimize the number of changes with --no-baseline). - Handle callsite clones in DoCallFallback. - Don't attach a stub if the script requires a callsite clone. It's a lot of complexity just for ParallelArray, and I'm not sure if it matters with baseline since we will enter Ion for hot code. Furthermore, callsite cloning is a temporary stopgap until we fix bug 804676, hopefully that will be done soon after baseline lands.
Attachment #727686 - Flags: review?(kvijayan)
Attachment #727686 - Flags: review?(kvijayan) → review+
Status: ASSIGNED → RESOLVED
Last Resolved: 6 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.