RFR: 8368303: AlwaysAtomicAccesses is excessively strict

[Andrew Haley]

On Mon, 22 Sep 2025 21:53:29 GMT, David Holmes <dholmes at openjdk.org> wrote:

> This flag also affects C2 - or did when it was introduced. 

The C2 handling is not too strict, it's fine.

> Seems to me this flag is not really serving any useful purpose though - the JMM updates it was supposed to support evaluation of never eventuated.

It still has some uses. Quoth @shipilev, "For example, for 64-bit longs on 32-bit platforms, and/or more recently, in Valhalla. I think the flag should stay at least until Valhalla sticks with a particular access atomicity story."

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

PR Comment: https://git.openjdk.org/jdk/pull/27432#issuecomment-3323362710