RFR (S) 8218266: G1 crash in AccessInternal::PostRuntimeDispatch

[zgu at redhat.com]

> > problem. Therefore, I would like to withdraw my review.
> 
> Okay, I can try to explain it.  All of the dictionary entries point
> to a 
> linked list of protection domains that were used to load the
> class.   
> The elements in this list point directly to Hashtable entries in the 
> ProtectionDomainTable.  The protection domains can be GC'd before
> the 
> dictionary entry is unloaded, so that the elements in the linked
> list 
> need to be cleaned out.  I chose cleaning to happen right before the 
> protection domain cache table is cleaned up so that the entries are
> not 
> dangling pointers.  Since the table is cleaned out concurrently now,
> I 
> have to do the cleanup there so I know it's happened.
> 
Thanks for educating me.

I were not sure if the crash was due to concurrent access, e.g.
DictionaryEntry could *leak* outside of Dictionary, then
ProtectedDomain can be accessed without protection. It looks like that
was not the case, but still should be avoided. How about places these
methods under private? it built.

http://cr.openjdk.java.net/~zgu/JDK-8218266-review/webrev.00/src/hotspo
t/share/classfile/dictionary.hpp.udiff.html


I did found an instance of unprotected access -
DictionaryEntry::contains_protection_domain() may be unprotected when
it walks protection domain list.

I added "assert_locked_or_safepoint(SystemDictionary_lock);" inside the
method:

http://cr.openjdk.java.net/~zgu/JDK-8218266-review/webrev.00/src/hotspo
t/share/classfile/dictionary.cpp.udiff.html

the assertion failed.

Thanks,

-Zhengyu



> Thanks,
> Coleen
> > 
> > Sorry.
> > 
> > -Zhengyu
> > 
> > 
> > > Coleen
> > > > Thanks,
> > > > 
> > > > -Zhengyu
> > > > 
> > > > > Thanks,
> > > > > Coleen
> 
>