Bulk zeroing reduction didn't work with G1, because arraycopy would call pre-barriers on uninitialized oops. The solution is to have version of arraycopy stubs that don't have pre-barriers. Also refactored arraycopy stubs generation on SPARC to be more readable and reduced the number of stubs necessary in some cases.
Reviewed-by: jrose, kvn, never
Introduced a new write_ref_array(HeapWords* start, size_t count) method that does the requisite MemRegion range calculation so (some of the) clients of the erstwhile write_ref_array(MemRegion mr) do not need to worry. This removed all external uses of array_size(), which was also simplified and made private. Asserts were added to catch other possible issues. Further, less essential, fixes stemming from this investigation are deferred to CR 6904516 (to follow shortly in hs17).
Reviewed-by: kvn, coleenp, jmasa
