RFR(XXS) 8199741: NMT: Tag safepoint polling pages

Zhengyu Gu zgu at redhat.com
Mon Mar 19 17:35:30 UTC 2018


Thanks, Aleksey.

-Zhengyu

On 03/19/2018 01:33 PM, Aleksey Shipilev wrote:
> If you need a second reviewer, the patch looks good.
> 
> -Aleksey
> 
> On 03/19/2018 06:31 PM, Zhengyu Gu wrote:
>> Hi,
>>
>> Do I need second reviewer for this trivial change?
>>
>> BTW, Submit-hs test came back clean.
>>
>> Mach5 mach5-one-zgu-JDK-8199741-20180319-1538-14990: Builds PASSED. Testing SUCCESSFUL.
>>
>> Thanks,
>>
>> -Zhengyu
>>
>> On 03/16/2018 11:50 AM, coleen.phillimore at oracle.com wrote:
>>>
>>> Yay, a quick code review.  Is there a better mt tag than mtInternal though?  There are too many
>>> mtInternal allocations.  Can you add mtSafepoint or mtSynchronize (or better name?) and we could
>>> have an RFE to make some of these into the same:
>>>
>>> share/runtime/semaphore.hpp:class Semaphore : public CHeapObj<mtInternal> {
>>> share/runtime/mutex.hpp:class Monitor : public CHeapObj<mtInternal> {
>>> share/runtime/objectMonitor.cpp:  return AllocateHeap(size, mtInternal);
>>> share/runtime/objectMonitor.cpp:  char * knobs = (char *) os::malloc(sz + 2, mtInternal);
>>> share/runtime/monitorChunk.cpp:  _monitors           = NEW_C_HEAP_ARRAY(BasicObjectLock,
>>> number_on_monitors, mtInternal);
>>>
>>> Or even mtOS so we can move some of the os allocations to this category?
>>>
>>> Why is nothing simple?! :)
>>>
>>> Coleen
>>>
>>> On 3/16/18 11:34 AM, Zhengyu Gu wrote:
>>>> A trivial fix to add NMT tag to Safepoint poll pages.
>>>>
>>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8199741
>>>> Webrev: http://cr.openjdk.java.net/~zgu/8199741/webrev.00/
>>>>
>>>> Test:
>>>>    Eyeball NMT output on Linux x64.
>>>>
>>>> Thanks,
>>>>
>>>> -Zhengyu
>>>
> 
> 


More information about the hotspot-runtime-dev mailing list