RFR: JDK-8282405: Make thread resource areas signal safe [v3]

[Thomas Stuefe]

On Thu, 3 Mar 2022 06:55:56 GMT, David Holmes <dholmes at openjdk.org> wrote:

> Out of curiosity, where does AGCT perform RA allocation? That seems like a bug ... but then AGCT is really a fingers-crossed mechanism anyway so not much point trying to make it too bullet-proof.

I agree that this would be a bug. I try to find out what the call stacks were.

AGCT seems like a best effort mechanism, but async profiler got a lot more popular recently, so it would be good to make it safer. And the RA double buffering mechanism I propose is quite simple, and we need it anyway for error handling.

> 
> A debug mechanism to disallow RA allocation might be interesting ... but might just show we have a lot more RA allocation than we thought. :) Up to you if you want to investigate that further.

I will try, but no promises :) This feels like it could be a whack the mole game.

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

PR: https://git.openjdk.java.net/jdk/pull/7624