C2 produces redundant (?) assembly for while-loop in certain cases (JDK-8278518)

[dean.long at oracle.com]

Hi, Sergey

On 12/17/21 12:27 AM, Сергей Цыпанов wrote:

> 1) Is my original theory about bounds check wrong and if it indeed is,
> then what is that disappearing assembly part about?
>
> 2) Should this be fixed on JVM or Java side?
> If we choose JVM then the issue needs to be reassigned
> to someone else, because I'm not able
> to elaborate and test the proper fix.

This does look like a JIT issue.  I've changed the bug to hotspot/compiler.

thanks,

dl