RFR: Fix C2 intrinsics barriers

[Roman Kennke]

The latest merge threw off some intrinsics:
- one arraycopy requires write- instead of read-barrier (upstream bug,
but needs to be fixed ASAP, will also fix upstream)
- two other instances seem to require explicit barriers even though
make_unsafe_addr() should already insert them. Possible that the
verifier is too strict here. Re-inserted those explicit barriers with
comment. We should check what's up there

http://cr.openjdk.java.net/~rkennke/fix-instrinsics-barriers/webrev.00/

Testing: specjvm+ -XX:+ShenandoahVerifyOptoBarriers which failed before
now passes

Ok?

Roman