RFR (XS): 8005116: NPG: Rename -permstat option for jmap in jdk8 to -classstats

Erik Helin erik.helin at oracle.com
Fri Feb 22 17:28:09 UTC 2013


Jon,

On 02/22/2013 03:54 PM, Jon Masamitsu wrote:
>
> On 2/22/2013 1:21 AM, Erik Helin wrote:
>> Jon,
>>
>> thanks for having a look at this change.
>>
>> On 02/21/2013 08:46 PM, Jon Masamitsu wrote:
>>> Early on the flag was -classstats and I understand the
>>> change to -clstats ("sss" hurts the eyes) but it is described
>>> as "class loader" statistics now.  How did we get to that
>>> description?  Meaning "class loader" statistics instead of
>>> "class" statistics?
>>
>> The output from "jmap -permstat" on a Java process that does nothing
>> is the following:
>>
>>> Attaching to process ID 13635, please wait...
>>> Debugger attached successfully.
>>> Server compiler detected.
>>> JVM version is 25.0-b20-fastdebug
>>> finding class loader instances ..done.
>>> computing per loader stat ..done.
>>> please wait.. computing liveness......................done.
>>> class_loader    classes    bytes    parent_loader    alive? type
>>>
>>> <bootstrap>    356    712996      null      live <internal>
>>> 0x00000007d65b10a8    3    10575    0x00000007d659b498 live
>>> sun/misc/Launcher$AppClassLoader at 0x000000077cc38e30
>>> 0x00000007d659b498    0    0      null      live
>>> sun/misc/Launcher$ExtClassLoader at 0x000000077cc2db60
>>>
>>> total = 3    359    723571        N/A        alive=3, dead=0        N/A
>>
>> To me, this seems to be statistics about class loaders and therefore I
>> think "class loader statistics" is a fitting description.
>>
>> What do you think?
>
> It does look like it is meant to be class loader statistics.  I now
> think you're correct in the description.  Thanks for the example
> output.
>
> No other comment on the review.  Looks good.

Thanks!

Erik

> Jon
>
>>
>> Thanks,
>> Erik
>>
>>> Jon
>>>
>>> On 02/21/13 05:06, Erik Helin wrote:
>>>> All,
>>>>
>>>> I've updated the change based on internal feedback and feedback from
>>>> Staffan Larsen.
>>>>
>>>> The bug has been renamed to use -clstats and the help text now uses
>>>> "class loader" instead of "classloader".
>>>>
>>>> Webrev:
>>>> http://cr.openjdk.java.net/~ehelin/8005116/webrev.01/
>>>>
>>>> The first and third change have also been updated.
>>>>
>>>> Thanks,
>>>> Erik
>>>>
>>>> On 02/20/2013 05:02 PM, Erik Helin wrote:
>>>>> Hi all,
>>>>>
>>>>> this change is the second of three changes related to renaming the
>>>>> jmap
>>>>> option permstat to clstats. The change has to be done as three
>>>>> separate
>>>>> changes in order to not break any repositories:
>>>>> 1. Add support for the -clstats option to the SA agent
>>>>> 2. Change the tool jmap to use -clstats instead of -permstat
>>>>> 3. Remove the option -permstat from the SA agent
>>>>>
>>>>> This CR represents the second step, which updates the jmap tool.
>>>>>
>>>>> Step 1 and 3 are also out for review.
>>>>>
>>>>> Webrev:
>>>>> http://cr.openjdk.java.net/~ehelin/8005116/webrev.00/
>>>>>
>>>>> Bug:
>>>>> http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8005116
>>>>>
>>>>> Thanks,
>>>>> Erik
>>>>
>>
>




More information about the hotspot-gc-dev mailing list