RFR(T) 8221643: Tighten up assert(_keep_alive >= 0) in CLD::inc_keep_alive

[Zhengyu Gu]

Thanks, Coleen.

-Zhengyu

On 3/29/19 9:50 AM, coleen.phillimore at oracle.com wrote:
> 
> Looks good and trivial.
> Thanks!
> Coleen
> 
> On 3/29/19 8:38 AM, Zhengyu Gu wrote:
>> I would like to restore the assertion to its original form, cause the 
>> current form suggests the possibility of resurrection not kept alive CLD.
>>
>> The current form was a workaround introduced by JDK-8162553, when null 
>> class loader data had keep_alive value of 0. Now, the value is 
>> initialized to 1 by CLD constructor, so the workaround is not needed.
>>
>>
>> diff -r f0fec71d2fff -r d023b009df71 
>> src/hotspot/share/classfile/classLoaderData.cpp
>> --- a/src/hotspot/share/classfile/classLoaderData.cpp   Thu Mar 28 
>> 19:43:59 2019 +0100
>> +++ b/src/hotspot/share/classfile/classLoaderData.cpp   Thu Mar 28 
>> 16:11:59 2019 -0400
>> @@ -288,7 +288,7 @@
>>  // it is being defined, therefore _keep_alive is not volatile or atomic.
>>  void ClassLoaderData::inc_keep_alive() {
>>    if (is_unsafe_anonymous()) {
>> -    assert(_keep_alive >= 0, "Invalid keep alive increment count");
>> +    assert(_keep_alive > 0, "Invalid keep alive increment count");
>>      _keep_alive++;
>>    }
>>  }
>>
>>
>> Test:
>>   hotspot_runtime on Linux x86 fastdebug
>>   Sumbit tests.
>>
>>
>> Thanks,
>>
>> -Zhengyu
>