On linux/x86_64, my test results are as follows:

SinkChannelTest fails intermittently (<1/10 runs) with:

  java.net.BindException: The address is not available
  at org.apache.harmony.luni.platform.OSNetworkSystem.bind(OSNetworkSystem.java)
  at org.apache.harmony.luni.net.PlainSocketImpl.bind(PlainSocketImpl.java:178)
  at java.net.ServerSocket.bind(ServerSocket.java:406)
  at org.apache.harmony.nio.internal.ServerSocketChannelImpl$ServerSocketAdapter.bind(ServerSocketChannelImpl.java:210)
  at java.net.ServerSocket.bind(ServerSocket.java:359)
  at org.apache.harmony.nio.internal.ServerSocketChannelImpl$ServerSocketAdapter.bind(ServerSocketChannelImpl.java:280)
  at org.apache.harmony.nio.tests.java.nio.channels.SinkChannelTest.test_socketChannel_read_close(SinkChannelTest.java:497)

I suspect this is just an unstable test.

PipedInputStreamTest.test_readException() fails reliably with the wrong
exception message "Pipe broken" not "Write end dead".  I've modified the
test so that throws the exception on failure which gives more helpful
output:

  java.io.IOException: Pipe broken
  at java.io.PipedInputStream.read(PipedInputStream.java:180)
  at org.apache.harmony.luni.tests.java.io.PipedInputStreamTest
     .test_readException(PipedInputStreamTest.java:90)

I've seen both of these problems on earlier milestones and don't see
them as blockers.

I also see a number of timeouts in awt/swing but these are not
regressions.

Regards,
-Mark.