RFR: 8368303: AlwaysAtomicAccesses is excessively strict

[Andrew Haley]

On Tue, 23 Sep 2025 12:04:47 GMT, Albert Mingkun Yang <ayang at openjdk.org> wrote:

> I get the feeling that this "optimization" (using plain store/move for smaller types) should be done at a lower level/later stage, i.e. in respective backend.

That's not possible because this behaviour is being done in shared code, and cannot be overridden by the back ends.

The logic in this PR is sound, I believe: if accesses on this target are multi-copy atomic, then all accesses of a word size or less are single-copy atomic. How could it not be so?

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

PR Comment: https://git.openjdk.org/jdk/pull/27432#issuecomment-3327226522