[lworld] RFR: 8378273: [lworld] Use acmp map to detect oops for VarHandle [v3]

Paul Hübner phubner at openjdk.org
Mon Feb 23 08:47:43 UTC 2026


On Sun, 22 Feb 2026 05:38:28 GMT, Chen Liang <liach at openjdk.org> wrote:

>> src/java.base/share/classes/jdk/internal/value/ValueClass.java line 58:
>> 
>>> 56: 
>>> 57:     /// {@return whether the flat layout for fields of this type contains references}
>>> 58:     /// Returns true if there is no flat layout for fields of the given type.
>> 
>> Confirming, will return true for an identityClass with only primitive fields?
>> The `hasOops` name might need to be more descriptive of the cases for an arbitrary class.
>
> No, if there's a field of an identityClass type in another class, that field is always laid out as a reference.

There might be an opportunity to strengthen the language even more:

{@return whether the flat layout for fields of this type contains references}
    /// Returns true if there is no flat layout for fields of the given type.

-->

{@return whether fields of the given class, when using a flat layout, contain oops}
    /// Returns true if the given class there is no flat layout.

A possible more descriptive name of the method could be `flatFieldHasOops`.

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

PR Review Comment: https://git.openjdk.org/valhalla/pull/2143#discussion_r2839656347


More information about the valhalla-dev mailing list