Request for review 8001471: Klass::cast() does nothing now

harold seigel harold.seigel at oracle.com
Mon Nov 12 08:46:56 PST 2012


Hi Coleen,

I incorporated the changes that you suggested below and retested.  
Here's a new webrev containing the changes:

http://cr.openjdk.java.net/~hseigel/bug_8001471_2/ 
<http://cr.openjdk.java.net/%7Ehseigel/bug_8001471_2/>

Please review and send me any comments.

Thanks, Harold

On 11/10/2012 6:29 PM, Coleen Phillimore wrote:
>
> Harold,
>
> http://cr.openjdk.java.net/~hseigel/bug_8001471/src/share/vm/classfile/placeholders.cpp.udiff.html 
>
>
> This looks disturbing.   Can you change the InstanceKlass() member 
> function in placeholders.hpp to be instance_klass() as part of this 
> edit?   It's sort of related to this change (ie from the big permgen 
> elimination renaming) and a small item.
>
> http://cr.openjdk.java.net/~hseigel/bug_8001471/src/share/vm/code/nmethod.cpp.udiff.html 
>
>
> Can you rewrite this to not have the variable ctxk and k, and just 
> have one?  Line 2571 should be deleted in one way or another.
>
> http://cr.openjdk.java.net/~hseigel/bug_8001471/src/share/vm/services/serviceUtil.hpp.sdiff.html 
>
>
> Line 60 can go away here too.    Klass* klass = k; - just use k in the 
> code below.
>
> This is really good.   I love to see this noisy code go away!
> Coleen
>
> On 11/10/2012 5:15 PM, David Holmes wrote:
>> Looks good Harold.
>>
>> David
>>
>> On 9/11/2012 11:52 PM, harold seigel wrote:
>>> Please review the following change to fix bug 8001471.
>>>
>>> Summary: The bug was fixed by removing the unneeded Klass::cast()
>>> operations. All of the casts() were unneeded because they were casting
>>> either objects of type 'Klass *', function results of type 'Klass 
>>> *', or
>>> function results that were subtypes of 'Klass *'.
>>>
>>> Open webrev at http://cr.openjdk.java.net/~hseigel/bug_8001471/
>>> <http://cr.openjdk.java.net/%7Ehseigel/bug_8001471/>
>>>
>>> Bug link at http://bugs.sun.com/view_bug.do?bug_id=800147
>>>
>>> The changes were tested with JPRT, JCK, JTREG, and ute tests.
>>>
>>> Thanks, Harold


More information about the hotspot-runtime-dev mailing list