RFR(S) JDK-8022284: Hide internal data structure in PhaseCFG
Niclas Adlertz
niclas.adlertz at oracle.com
Tue Aug 6 17:53:48 PDT 2013
Sorry about that.
http://cr.openjdk.java.net/~adlertz/JDK-8022284/webrev02/
Kind Regards,
Niclas Adlertz
On 7 aug 2013, at 02:20, Vladimir Kozlov <Vladimir.Kozlov at oracle.com> wrote:
> Niclas,
>
> You missed cfg check change in block.cpp at line 314.
> You did not fixed pinch check in output.cpp, lines 2645 and 2715.
>
> We also put () around such checks: "((cfg != NULL) &&"
>
> Thanks,
> Vladimir
>
> On 8/6/13 4:58 PM, Niclas Adlertz wrote:
>> Thank you Roland and Vladimir.
>>
>> Here's the new webrev based on the changes you suggested Roland:
>> http://cr.openjdk.java.net/~adlertz/JDK-8022284/webrev01/
>>
>> Kind Regards,
>> Niclas Adlertz
>>
>>
>> On 7 aug 2013, at 01:31, Roland Westrelin <roland.westrelin at oracle.com> wrote:
>>
>>>
>>> Hi Niclas,
>>>
>>>> A review would be much appreciated.
>>>>
>>>> JBS: https://jbs.oracle.com/bugs/browse/JDK-8022284
>>>> WEBREV: http://cr.openjdk.java.net/~adlertz/JDK-8022284/webrev00/
>>>
>>> Looks good.
>>>
>>> While you're cleaning up this code:
>>> In block.cpp:
>>> 331 if (cfg && bh->is_Loop() && !head()->is_Root()) {
>>>
>>> should be cfg != NULL
>>>
>>> In lcm.cpp, !best should be best == NULL.
>>>
>>> In output.cpp, same thing with !pinch.
>>>
>>> Also, in vmStructs.cpp, no need for _block_arena.
>>>
>>> Roland.
>>
More information about the hotspot-compiler-dev
mailing list