RFR (XS) JDK-8214275: CondyRepeatFailedResolution asserts "Dynamic constant has no fixed basic type"

[coleen.phillimore at oracle.com]

btw. This change looks trivial.  Not the analysis...
Thanks,
Coleen

On 12/4/18 11:39 AM, Lois Foltan wrote:
> Thanks Coleen for the review!
> Lois
>
> On 12/4/2018 10:48 AM, coleen.phillimore at oracle.com wrote:
>> Looks good!
>> Coleen
>>
>> On 12/4/18 7:23 AM, Lois Foltan wrote:
>>> Please review this small fix to handle a failed dynamic constant 
>>> resolution tag within GenerateOopMap::do_ldc().  This issue only 
>>> happens for a debug build.  The purpose of the test is to repeatedly 
>>> try to resolve a CONSTANT_Dynamic that has failed resolution with a 
>>> LinkageError in order to ensure that the bootstrap method for the 
>>> condy is never reexecuted after the first failed attempt. See JVMS 
>>> section 5.4.3.  Within the ASSERT conditional of 
>>> GenerateOopMap::do_ldc(), it must check for a CONSTANT_Dynamic tag 
>>> as well as a CONSTANT_DynamicInError tag.  If either tag is present, 
>>> constantTag::basic_type() should not be called.
>>>
>>> open webrev at 
>>> http://cr.openjdk.java.net/~lfoltan/bug_jdk8214275.1/webrev/
>>> bug link at https://bugs.openjdk.java.net/browse/JDK-8214275
>>>
>>> Testing: hs-tier1-3, jdk-tier1-3 (complete), hs-tier4-8 (in progress 
>>> for linux-x64-debug)
>>>                CondyRepeatFailedResolution.java test repeat 100x on 
>>> solaris-sparcv9-debug
>>>
>>> Thanks,
>>> Lois
>>
>