RFR(10)(XS): JDK-8171365: nsk/jvmti/scenarios/events/EM04/em04t001: many errors for missed events

[Chris Plummer]

Hi Vladimir,

Tobias already looked at the changes and they have been pushed. If Dean 
finds any AOT related concerns, we'll need to address them with a 
separate CR. However, I don't think there will be an issue with AOT, at 
least not for JVMTI's use (and it is the only user of this API), since 
it is ignoring all nmethods (it just want's blobs that don't represent 
java methods).

thanks,

Chris

On 6/7/17 3:17 PM, Vladimir Kozlov wrote:
> Thank you Chris for finding the issue.
>
> I would like Dean to look on it since he did this changes.
> One concern is change to FOR_ALL_HEAPS() will also scan AOT heap. We 
> need to figure-out to how to scan all blobs in regular CodeCache 
> excluding only AOT.
>
> Thanks,
> Vladimir
>
> On 6/5/17 11:19 PM, Tobias Hartmann wrote:
>> Hi Chris,
>>
>> On 06.06.2017 04:48, Chris Plummer wrote:
>>> I'll wait for the compiler team to chime in. I've already asked them 
>>> to have a look (this issue not withstanding) because it is their code.
>>
>> Your change looks good to me and I'm fine with the simple fix without 
>> a closure.
>>
>> This is actually a regression introduced by the AOT integration 
>> (JDK-8171008):
>> http://hg.openjdk.java.net/jdk9/jdk9/hotspot/rev/777aaa19c4b1#l116.123
>>
>> Before, CodeCache::blobs_do() used to iterate over all code heaps.
>>
>> Best regards,
>> Tobias
>>