[lworld] RFR: 8378273: [lworld] Use acmp map to detect oops for VarHandle [v2]

Paul Hübner phubner at openjdk.org
Fri Feb 20 13:26:45 UTC 2026


On Fri, 20 Feb 2026 13:18:47 GMT, Chen Liang <liach at openjdk.org> wrote:

>> Of course. But I interpreted `hasOopsCase` meaning "does this container contain oops" whereby the `[]` is the container and the `int` is the field. So is `hasOopsCase` only called on the element of the array rather than the container?
>
> This really means the layout of a field of the given type has oops when it is inlined. I think this is clear enough.

I disagree. The JavaDoc specifies `{@return whether a flat layout of this class contains references}` . So if we replace "class", with "int[]", we read: "whether a flat layout of int[] contains references". The method returns true, but in the context of the documentation it shouldn't (an int[] is always flat and certainly does not contain oops). 

This is not my area and if you insist/others are fine with it, I'm happy to agree to disagree.

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

PR Review Comment: https://git.openjdk.org/valhalla/pull/2143#discussion_r2833182077


More information about the valhalla-dev mailing list