RFR (S) 8223657: Remove unused THREAD argument from SymbolTable functions

[David Holmes]

Hi Coleen,

Generally this all looks fine. A couple of follow ups:

src/hotspot/share/classfile/classListParser.cpp

329   TempNewSymbol class_name_symbol = 
SymbolTable::new_symbol(_class_name);
  330   guarantee(!HAS_PENDING_EXCEPTION, "Exception creating a symbol.");

This guarantee seems wrong even before this change, but if no exceptions 
can come from new_symbol then it serves no purpose now.

---

src/hotspot/share/classfile/javaClasses.cpp

572 Symbol* java_lang_String::as_symbol(oop java_string, TRAPS) {

TRAPS is unused now. Further there seems to be no reason to have both 
java_lang_String::as_symbol and java_lang_String::as_symbol_or_null as 
the only distinction appears to be that the former was expected to post 
an exception if it would return NULL. But apparently it is impossible to 
return NULL as we should have aborted the VM on allocation failure - right?

Thanks,
David
-----

On 11/05/2019 1:15 am, coleen.phillimore at oracle.com wrote:
> Summary: also made lookup and lookup_only functions private to 
> SymbolTable.  External callers use new_symbol or probe.
> 
> Ran hs-tier1-3.
> 
> open webrev at http://cr.openjdk.java.net/~coleenp/2019/8223657.01/webrev
> bug link https://bugs.openjdk.java.net/browse/JDK-8223657
> 
> Thanks,
> Coleen
> 
>