possible data race on NGServer.serversocket

[yilongli]

Again, we got this race report when testing the K framework:

Data race on field com.martiansoftware.nailgun.NGServer.serversocket: {{{
    Concurrent read in thread T1 (locks held: {})
 ---->  at com.martiansoftware.nailgun.NGServer.getPort(NGServer.java:384)
        at org.kframework.kserver.KServerFrontEnd.run(KServerFrontEnd.java:74)
        at org.kframework.main.FrontEnd.main(FrontEnd.java:52)
        at org.kframework.main.Main.runApplication(Main.java:109)
        at org.kframework.main.Main.runApplication(Main.java:99)
        at org.kframework.main.Main.main(Main.java:51)
    T1 is the main thread

    Concurrent write in thread T13 (locks held: {})
 ---->  at com.martiansoftware.nailgun.NGServer.run(NGServer.java:413)
    T13 is created by T1
        at org.kframework.kserver.KServerFrontEnd.run(KServerFrontEnd.java:72)
}}}

It says the assignment serversocket = new ServerSocket(port, 0, addr) and the return statement return ((serversocket == null) ? port : serversocket.getLocalPort()); are in a data race. I think it's sufficient to declare serversocket as volatile.

[sbalabanov-zz]

Closing all issues older than 6 months; please reopen if it is still relevant