RFR (S): 8250628: ZGC: `fixup_partial_loads` was removed, but still are referenced

[Kim Barrett]

> On Aug 2, 2020, at 6:22 AM, Albert Yang <albert.m.yang at oracle.com> wrote:
> 
> I have further tested with tier1 and vmTestbase_nsk_jvmti (with the patch or explicitly setting them `true`), and find no failure. Any other tests that might be relevant?

I think that’s good enough.  Thanks.

> On 2020-08-01 09:14, Kim Barrett wrote:
>>> On Jul 29, 2020, at 8:19 AM, Albert Yang <albert.m.yang at oracle.com> wrote:
>>> 
>>> Hi,
>>> 
>>> https://bugs.openjdk.java.net/browse/JDK-8250628
>>> 
>>> http://cr.openjdk.java.net/~lkorinth/albert/8250628/0/
>>> 
>>> Case 2&3 can be removed, since all roots will go through the load barrier before a safepoint. To be safe, I have tried setting them to true via `FLAG_SET_DEFAULT(_, true);`, and no failures are observed for hotspot-gc.
>>> 
>>> Tested: hotspot-gc
>>> 
>>> /Albert
>> The change looks okay, but I'm not sure about the testing.
>> 
>> VerifyBeforeIteration is (only) used by jvmti heap iteration
>> functions, and I don't think the hotspot_gc test group includes any
>> jvmti tests.
>> 
> -- 
> /Albert