RFR: 8315566: [JVMCI] deadlock in JVMCI startup when bad option specified

Tobias Hartmann thartmann at openjdk.org
Mon Sep 4 05:18:37 UTC 2023


On Sat, 2 Sep 2023 11:00:00 GMT, Doug Simon <dnsimon at openjdk.org> wrote:

> This PR changes `HotSpotJVMCIRuntime.vmEventListeners` to be initialized in the `HotSpotJVMCIRuntime` constructor. This avoids a deadlock that can occur when this list is lazily initialized. Lazily initializing this list has no measurable performance benefit, especially in the context of libgraal.

Looks reasonable to me.

-------------

Marked as reviewed by thartmann (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/15549#pullrequestreview-1608772165


More information about the hotspot-compiler-dev mailing list