RFR [10]: 8185362: Replace use of AtomicReferenceFieldUpdater from BufferedInputStream with Unsafe

[Martin Buchholz]

Thanks.  Approved.

I missed that buf was protected and so the javadoc here actually matters.
Having proper javadoc here for the meaning of null buf (including for
subclasses) "should" be done, but out of scope for this change.

On Mon, Aug 21, 2017 at 8:09 AM, Claes Redestad <claes.redestad at oracle.com>
wrote:

> Hi Martin,
>
>
> On 08/21/2017 04:33 PM, Martin Buchholz wrote:
>
>> This comment still seems applicable. Can we move it into the docstring
>> for buf?
>>
>> We use nullness
>> -     * of buf[] as primary indicator that this stream is closed. (The
>> -     * "in" field is also nulled out on close.)
>>
>>
> this is a reasonable request, the catch is that the field is protected and
> thus visible
> in javadoc, so it needs to be updated as a CSR(?). Perhaps we can move it
> to a regular
> comment associated with the field to avoid this implementation detail
> showing up in
> javadocs?
>
> http://cr.openjdk.java.net/~redestad/8185362/jdk.02/
>
> Thanks!
>
> /Claes
>