[jdk23u] RFR: 8337753: Target class of upcall stub may be unloaded
Jorn Vernee
jvernee at openjdk.org
Thu Oct 24 14:46:09 UTC 2024
On Mon, 14 Oct 2024 12:34:18 GMT, Jorn Vernee <jvernee at openjdk.org> wrote:
> Backport of: https://github.com/openjdk/jdk/pull/20479
>
> There were some trivial conflicts in `stubGenerator_aarch64.cpp` and `stubGenerator_riscv.cpp`. Some code still there in jdk23, already removed in later versions.
After discussion, decided to not fix this in jdk23u. The fix is not exactly low-risk, and the scenario in which it occurs is fairly rare: only when many threads are racing to create FFM upcall stubs targeting the same Java method (really the same LambdaForm) at the same time, and we've only been able to reproduce on Linux/aarch64 so far. Given the current release timeline for jdk23u, if this fix does end up causing a regression, we might not have enough time to address it for jdk23u GA.
-------------
PR Comment: https://git.openjdk.org/jdk23u/pull/163#issuecomment-2435493030
More information about the jdk-updates-dev
mailing list