RFR: 8357145: CRC/Inflater/Deflater/Adler32 methods that take a ByteBuffer throw UOE if backed by shared memory segment

[Alan Bateman]

On Wed, 21 May 2025 13:53:15 GMT, Jaikiran Pai <jpai at openjdk.org> wrote:

>> This PR proposes to use ` JavaNioAccess::getBufferAdress` rather than `DirectBuffer::address` in the package `java.util.zip` so that `Buffer` instances backed by `MemorySegment` instances can be used.
>> 
>> This PR passes tier1, tier2, and tier3 tests on multiple platforms and configurations.
>
> src/java.base/share/classes/java/util/zip/Adler32.java line 102:
> 
>> 100:             NIO_ACCESS.acquireSession(buffer);
>> 101:             try {
>> 102:                 adler = updateByteBuffer(adler, NIO_ACCESS.getBufferAddress(buffer), pos, rem);
> 
> Hello Per, from what I can see in the implementation of `NIO_ACCESS.getBufferAddress(...)` it merely returns `buffer.address` https://github.com/openjdk/jdk/blob/master/src/java.base/share/classes/java/nio/Buffer.java#L862. So if I understand this change correctly, what we are proposing here is that we want to avoid the additional checks (checks related to `MemorySegment` that can fail?) that are there in the `DirectBuffer.address()` method implementation? Is that what this change is addressing?

All methods in the API that accept a ByteBuffer and access it as DirectBuffer should have been changed to use the acquireSession/getBufferAddress/releaseSession. Several were missed, I think because we didn't have enough tests to exercise the views.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/25321#discussion_r2100500374