RFR: 8069367: assert(_nextMarkBitMap->isMarked((HeapWord*) obj)) failed

[Kim Barrett]

Another round.

I've updated my local repo from the team repo, merging with the
changes from 8076265: Simplify deal_with_reference.

I've backed out the G1EagerReclaimHumongousPreSnapshotTypeArrays
experimental configuration option, per Bengt's suggestion that it will
likely be more trouble than it's worth.

Because of the latter, the test to determine whether a humongous
region should be a candidate for eager reclamation has been
simplified, since at present there is no need to check the allocation
time of the object.  But I've left the commentary there about
different handling of objects containing references, and noting the
special handling of typeArray objects in this respect.

Also had to move some new inline function definitions from
g1CollectedHeap.hpp to g1CollectedHeap.inline.hpp, to account for
header cleanups since my repo was last updated.

CR:
https://bugs.openjdk.java.net/browse/JDK-8069367 

Webrev:
http://cr.openjdk.java.net/~kbarrett/8069367/webrev.04/

Incremental webrev:
http://cr.openjdk.java.net/~kbarrett/8069367/webrev.04.incr/
The incremental webrev might be a little misleading, since the base
for it is not the version from the previous round of review, but
rather that version plus the merge for 8076265.

Testing:
Hand testing
JPRT
Aurora Ad-hoc GC Nightly using G1
local RefWorkload using G1