Code review request, 8013809 deadlock in SSLSocketImpl between between write and close

[Brad Wetmore]

On 7/25/2013 2:11 AM, Xuelei Fan wrote:
> Hi Brad,
>
> Are you available to review this fix?
>
> Webrev: http://cr.openjdk.java.net/~xuelei/8013809/webrev.00/
>
> No new regression test, hard to reproduce the issue.

Your immediate fix looks good, however, IIRC, the reason for having 
getConnectionState() was to provide synchronized access (and syncing on 
the SSLSocketImpl object) to the connectionState variable before the 
Java "volatile" semantics were cleaned up in JDK 1.5.

Now that you've made this change, does it make sense to get rid of 
getConnectionState(). What do you think?

I hate race conditions!  ;)

Brad