RFR(S): 8037970: make PrintMethodData a diagnostic options
Igor Veresov
igor.veresov at oracle.com
Fri Mar 28 19:05:57 UTC 2014
Looks good to me.
igor
On Mar 28, 2014, at 9:43 AM, Vladimir Kozlov <vladimir.kozlov at oracle.com> wrote:
> Ask Igor to look on it.
>
> Vladimir K
>
> On 3/28/14 8:53 AM, Roland Westrelin wrote:
>> Can I push this with a single review?
>>
>> Roland.
>>
>> On Mar 21, 2014, at 10:05 AM, Roland Westrelin <roland.westrelin at oracle.com> wrote:
>>
>>> Thanks for the review, Vladimir.
>>>
>>> Roland.
>>>
>>>
>>> On Mar 21, 2014, at 2:16 AM, Vladimir Kozlov <vladimir.kozlov at oracle.com> wrote:
>>>
>>>> Looks good.
>>>>
>>>> Thanks,
>>>> Vladimir
>>>>
>>>> On 3/20/14 3:23 PM, Roland Westrelin wrote:
>>>>>> Looks reasonable. Why you removed +compiled_invocation_count() in compare_methods()?
>>>>>
>>>>> Thanks for reviewing this. I removed compiled_invocation_count() in compare_methods() by mistake. Here is a new webrev:
>>>>>
>>>>> http://cr.openjdk.java.net/~roland/8037970/webrev.01/
>>>>>
>>>>> Roland.
>>>>>
>>>>>
>>>>>>
>>>>>> Thanks,
>>>>>> Vladimir
>>>>>>
>>>>>> On 3/20/14 9:20 AM, Roland Westrelin wrote:
>>>>>>> I’d like to make PrintMethodData a diagnostic option. When investigating nashorn performance, I found it useful to have the result of profiling.
>>>>>>>
>>>>>>> This change also:
>>>>>>>
>>>>>>> - changes the format of PrintMethodData to use:
>>>>>>>
>>>>>>> java.lang.invoke.DirectMethodHandle::internalMemberName
>>>>>>>
>>>>>>> instead of
>>>>>>>
>>>>>>> java/lang/invoke/DirectMethodHandle.internalMemberName
>>>>>>>
>>>>>>> for method names. Very useful for lambda forms: java.lang.invoke.LambdaForm$BMH/853119666::reinvoke
>>>>>>>
>>>>>>> - dumps data for all methods including lambda forms
>>>>>>>
>>>>>>> In debug builds, the print compiler oracle command can also cause profile data to be dumped for a particular method. In a product build, I restricted it to PrintMethodData only so when don’t iterate over all methods every time the JVM exits.
>>>>>>>
>>>>>>> http://cr.openjdk.java.net/~roland/8037970/webrev.00/
>>>>>>>
>>>>>>> Roland.
>>>>>>>
>>>>>
>>>
>>
More information about the hotspot-compiler-dev
mailing list