RFR (S): [MVT] Ensure JVMS spec rules for creation and loading of DVT are followed

[David Simms]

Yeah, MVT stuff only after resolve fails: I tried not to break 
"VanillaNothingToDoWithMVT$Value" use case, so the class will be found 
first if it does exist. Otherwise, yeah, "has_vcc_annotation()", just 
for getting the correct exceptions from the appropriate places, helps 
with debugging.


On 12/09/17 12:07, Maurizio Cimadamore wrote:
> Looks very good!
>
> In instanceKlass.hpp am I right that you just reordered the flags so 
> that MVT related stuff occurs last? That is, the only change there is 
> the added _extra_has_vcc_annotation - everything else is just 
> reshuffling?
>
> Maurizio
>
>
> On 12/09/17 10:38, David Simms wrote:
>> Greetings,
>>
>> Here's a patch containing further testing, and fix for MVT loading of 
>> VM derived value type class, even before touching its Value Capable 
>> Class. Throws correct exception according to draft spec (ICCE when 
>> VCC rules broken etc, 
>> http://cr.openjdk.java.net/~dlsmith/values.html#values-5.3)
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8184919
>>
>> Webrev: http://cr.openjdk.java.net/~dsimms/valhalla/8184919/webrev0/
>>
>>
>> Cheers
>>
>> /David Simms
>>
>