RFR (XL): 7172640: C2: instrinsic implementations in LibraryCallKit should use argument() instead of pop()

John Rose john.r.rose at oracle.com
Mon Nov 26 22:21:15 PST 2012


Reviewed; ship it.  Some comments:

The cleanups with Bytecodes::has_receiver and arg_size_for_bc are very good too.

I like your renamings:  It feels much more rational and regular.

The for_parse argument to GraphKit::compute_stack_effects is dead now; you may delete it.

I suggest removing "_as" in null_check_receiver_as_before_call; it feels like a "noise word"[1].

We are not in the habit of using "(void)" casts.  I think it would require an astronomically high warning level to benefit from them.

These are all suggestions.  In any case, ship it!

— John

[1] https://wikis.oracle.com/display/HotSpotInternals/StyleGuide#StyleGuide-Naming

On Nov 21, 2012, at 5:02 PM, Christian Thalinger wrote:

> On Nov 21, 2012, at 3:56 PM, John Rose <john.r.rose at oracle.com> wrote:
> 
>> On Nov 20, 2012, at 6:47 PM, Christian Thalinger wrote:
>> 
>>> http://cr.openjdk.java.net/~twisti/7172640
>> 
>> That's a lovely cleanup...
> 
> webrev is updated.




More information about the hotspot-compiler-dev mailing list