[9] RFR (S) File leak in MemNotifyThread::start() in hotspot.src.os.linux.vm.os_linux.cpp

Yumin Qi yumin.qi at oracle.com
Fri Oct 31 16:13:24 UTC 2014


Looks good to me. I am not a "R"eviewer by the way.

Thanks
Yumin

On 10/31/2014 12:23 AM, Chris Plummer wrote:
> I could still use one more reviewer for this change.
>
> thanks,
>
> Chris
>
> On 10/27/14 11:33 PM, Chris Plummer wrote:
>> Hi,
>>
>> Please review this change to fix a leaking fd.
>>
>> Summary: Fixed by removing all code related to LowMemoryProtection, 
>> which removed offending code.
>>
>> Webrev: http://cr.openjdk.java.net/~cjplummer/8043770/webrev.00/
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8043770
>>
>> Tested with jprt "-testset hotspot"
>>
>> Details: The offending code is there to support the 
>> -XX:+LowMemoryProtection flag, which relies on linux kernel support 
>> for /dev/mem_notify. Although the leak was easy to fix, I ran into an 
>> issue testing it. Apparently we have no devices that support 
>> /dev/mem_notify, and SQE is not testing this feature. It also turns 
>> out that in general /dev/mem_notify support has not been adopted, so 
>> LowMemoryProtection is of limited or no value, and in fact is a 
>> liability since we are not testing it. The decision was made to 
>> remove the support instead, thus also fixing the fd leak. This only 
>> affects builds with JAVASE_EMBEDDED enabled, so will not impact any 
>> JDK release.
>>
>> thanks,
>>
>> Chris
>



More information about the hotspot-runtime-dev mailing list