RFR: 8336919: Cleanup and rename tags in placeholders code

[Coleen Phillimore]

On Thu, 25 Jul 2024 12:16:39 GMT, Coleen Phillimore <coleenp at openjdk.org> wrote:

>> src/hotspot/share/classfile/systemDictionary.cpp line 511:
>> 
>>> 509:                                                           class_loader,
>>> 510:                                                           protection_domain,
>>> 511:                                                           false,
>> 
>> Please explain this change!
>
> If you're doing a parallel super load, it could be the super class but not necessarily, it could also be an interface class.  I thought it was an optimization, because if you add the circularity placeholder, and look up the existing super class, it should immediately find it an return.  Now I'm wondering if it's more than an optimization based on the comment in resolve_with_circularity_check.  I'll check this.
> 
> I added the assert above it for the place in static dumping that used this flag as either klass->super or interface, ie where it differentiated based on this flag.

Yes, this looks like an optimization, that we could do unconditionally if we want or not at all.  The comment mentions that we'll get a linkage error if we don't do it but I think it's talking about something else.

      // To support parallel loading: if class is done loading, just return the superclass
      // if the next_name matches class->super()->name() and if the class loaders match.
      // Otherwise, a LinkageError will be thrown later.
      if (klassk != nullptr && is_superclass &&
          ((quicksuperk = klassk->java_super()) != nullptr) &&
           ((quicksuperk->name() == next_name) &&
              (quicksuperk->class_loader() == class_loader()))) {
             return quicksuperk;
       } else {
        // Must check ClassCircularity before checking if superclass is already loaded.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/20279#discussion_r1691620648