[foreign-memaccess+abi] RFR: Add support for high-level functions to copy to and from Java arrays [v2]

[Lee Rhodes]

On Mon, 21 Jun 2021 20:42:40 GMT, Maurizio Cimadamore <mcimadamore at openjdk.org> wrote:

>> test/jdk/java/foreign/TestMemoryCopy.java line 49:
>> 
>>> 47:  * These tests exercise the MemoryCopy copyFromArray(...) and copyToArray(...).
>>> 48:  * To make these tests more challenging the segment is a view of the given array,
>>> 49:  * which makes the copy operations overlapping self-copies.  Thus, this checks the claim:
>> 
>> I'm not sure I understand this claim; it seems to me that the test sets up an array view segment, but then copies the contents into a new array (using toArray) as src/dst to do the actual copying? i.e. the source and destination seem disjoint.
>
> This comment was in the original patch - I'll fix or drop it.

I think what you are seeing the extra copy is in the creation of the `truthSegment`, where an extra copy had to be used because there is no API (yet), to help the code determine which direction the copy should be performed in a self-copy scenario.  If the call to the MemoryCopy API does not perform the self-copy correctly, comparing its result with the `truthSegment` will definitely reveal an error.

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

PR: https://git.openjdk.java.net/panama-foreign/pull/555