RFR (S) : 8026941 : [TESTBUG] java.lang.ClassNotFoundException: java.lang.invoke.InvokeGeneric

Igor Ignatyev igor.ignatyev at oracle.com
Thu Dec 12 12:32:04 PST 2013


Vladimirs, thank you both.

Igor

On 12/13/2013 12:14 AM, Vladimir Kozlov wrote:
> Good.
>
> Vladimir K
>
> On 12/12/13 11:42 AM, Igor Ignatyev wrote:
>> I have removed all package-private and removed classes from classNames[].
>>
>> For deadlock w/ all classes, I have filed a new bug -- JDK-8030063.
>>
>> updated webrev: http://cr.openjdk.java.net/~iignatyev/8026941/webrev.01/
>>
>> Igor
>>
>> On 12/12/2013 10:03 PM, igor.ignatyev at oracle.com wrote:
>>> Ok, I will just filter out hard-coded list and add the version of test
>>> with enumeration on all classes as a proof to a new bug about deadlock.
>>>
>>> -- Igor
>>>
>>> ----- Reply message -----
>>> From: "Vladimir Ivanov" <vladimir.x.ivanov at oracle.com>
>>> To: "Igor Ignatyev" <igor.ignatyev at oracle.com>,
>>> "hotspot-compiler-dev at openjdk.java.net compiler"
>>> <hotspot-compiler-dev at openjdk.java.net>
>>> Subject: RFR (S) : 8026941 : [TESTBUG] java.lang.ClassNotFoundException:
>>> java.lang.invoke.InvokeGeneric
>>> Date: Thu, Dec 12, 2013 21:12
>>>
>>> Igor,
>>>
>>> Why don't you simply filter out all package-private & private classes
>>> out of classNames array, but leave the list hard-coded in the test? All
>>> public classes are part of public API, so the list shouldn't change.
>>>
>>> Best regards,
>>> Vladimir Ivanov
>>>
>>> On 12/12/13 8:24 PM, Igor Ignatyev wrote:
>>>> Hi all,
>>>>
>>>> Problem:
>>>> Test compiler/jsr292/ConcurrentClassLoadingTest has a hard-coded
>>>> list of
>>>> classes to load, after JDK-8024761 'j.l.i.InvokeGeneric' doesn't exist,
>>>> so test fails w/ CNFE
>>>>
>>>> Fix:
>>>> To testlibrary, added class to print all public classes --
>>>> ClassesPrinter
>>>> In ConcurrentClassLoadingTest, hard-coded list was replaced by the list
>>>> of public classes from j.l.i which exist in bootCP.
>>>>
>>>> At first, I tried to use all classes from j.l.i, not only public,
>>>> but it
>>>> can lead to deadlock. After discussion w/ Vladimir I, we decided that
>>>> since it isn't related to original bug (8022595) we should limited
>>>> classes to public only. I will file a separate product bug about that
>>>> deadlock.
>>>>
>>>> webrev:http://cr.openjdk.java.net/~iignatyev/8026941/webrev.00/
>>>> jbs:https://bugs.openjdk.java.net/browse/JDK-8026941
>>>> testing: jprt
>>>>
>>>> Thanks,
>>>> Igor
>>>


More information about the hotspot-compiler-dev mailing list