RFR(M): 8054889 - Additional Compiler Diagnostic commands

Vladimir Kozlov vladimir.kozlov at oracle.com
Fri Sep 5 22:09:21 UTC 2014


Looks good to me.

Thanks,
Vladimir

On 9/5/14 7:08 AM, Nils Eliasson wrote:
> Hi Albert,
>
> On 2014-09-05 15:27, Albert wrote:
>> Hi Nils,
>>
>> I looked at your changes. Here is some feedback:
>>
>> - Would it make sense to add an assert in ' CompileQueue::print()'
>> that checks that we are either at a safepoint or that the
>> MethodCompileQueue_lock is held?
>>   ( assert_locked_or_safepoint(CodeCache_lock);). This seems to make
>> the implementation more robust.
>
> Yes - certainly. Fixed.
>
>>
>> - CodeCache::print_codelist() iterates over all 'alive' nmethods.
>> However, the description says:
>>   'Print all compiled methods in code cache.' I think either the
>> description or the implementation should be adapted to keep things
>> consistent.
>
> Good. Yes it make more sense to print all nmethods.
>
> I also found that I had forgotten to add a file
> (MethodIdentifierParser.java).
>
> New webrev here:
> http://cr.openjdk.java.net/~neliasso/8054889/webrev.04
>
> Thank you!
> //Nils
>
>>
>> Best,
>> Albert
>>
>>
>> Best,
>> Albert
>> On 09/05/2014 01:12 PM, Nils Eliasson wrote:
>>> Hi,
>>>
>>> This fix introduces three new diagnostic commands and corresponding
>>> tests. The commands are Compiler.queue for printing the contents of
>>> the compiler queue, Compiler.codelist for printing all nmethods in
>>> the codecache and Compiler.codecache for printing an informational
>>> code cache summary.
>>>
>>> Description here:
>>> https://bugs.openjdk.java.net/browse/JDK-8054889
>>>
>>> Webrev:
>>> http://cr.openjdk.java.net/~neliasso/8054889/webrev.02/
>>>
>>> Thanks,
>>> Nils Eliasson
>>
>


More information about the hotspot-compiler-dev mailing list