XS RFR: 8009731: loader constraint violation error message
Karen Kinnear
karen.kinnear at oracle.com
Wed Mar 27 05:11:55 PDT 2013
Thank you Dan, David, Coleen and Yumin for your quick reviews.
I updated the bug to show the before and after.
thanks,
Karen
On Mar 26, 2013, at 9:05 PM, Daniel D. Daugherty wrote:
> > http://cr.openjdk.java.net/~acorn/8009731.2/webrev/
>
> src/share/vm/classfile/systemDictionary.cpp
> src/share/vm/classfile/systemDictionary.hpp
> src/share/vm/interpreter/linkResolver.cpp
> src/share/vm/oops/klassVtable.cpp
> No comments on any of the above.
>
> Could you add a note to the bug showing the original test output
> versus the revised test output? There's a lot of commentary in
> the bug report and a clear statement of "what it was" versus
> "what it is" will help...
>
> Dan
>
>
> On 3/26/13 6:45 PM, Karen Kinnear wrote:
>> Thank you Coleen and Yumin and Dan. Good catches.
>>
>> I did Dan's suggested changes and included systemDictionary.hpp in the following webrev:
>>
>> http://cr.openjdk.java.net/~acorn/8009731.2/webrev/
>>
>> thanks,
>> Karen
>>
>> On Mar 26, 2013, at 6:56 PM, Daniel D. Daugherty wrote:
>>
>>> On 3/26/13 4:33 PM, Karen Kinnear wrote:
>>>> Thank you Coleen and David.
>>>>
>>>> http://cr.openjdk.java.net/~acorn/8009731.1/webrev/
>>> Thumbs up.
>>>
>>> src/share/vm/classfile/systemDictionary.cpp
>>> 2201 Symbol* s = sig_strm.as_symbol(CHECK_NULL);
>>> 2202 Symbol* sig = s;
>>> I know you didn't change these lines, but:
>>>
>>> Local 's' isn't used except to set 'sig'. Is there a reason
>>> for the duplicate variable?
>>>
>>> Feel free to ignore the above if you're pressed for time.
>>>
>>> src/share/vm/interpreter/linkResolver.cpp
>>> No comments.
>>>
>>> src/share/vm/oops/klassVtable.cpp
>>> No comments.
>>>
>>> Dan
>>>
>>>
>>>
>>>> Problem was introduced with 6990754, changeset 2059.
>>>>
>>>> I like Coleen's solution better - so we don't run the risk of someone "fixing" the apparent need for a ResourceMark
>>>> in the future. So I changed this to use a Symbol*.
>>>>
>>>> vm.quick.testlist rerun in progress.
>>>>
>>>> thanks,
>>>> Karen
>>>>
>>>> On Mar 25, 2013, at 8:47 PM, Coleen Phillimore wrote:
>>>>
>>>>> On 3/25/2013 8:10 PM, David Holmes wrote:
>>>>>> Hi Karen,
>>>>>>
>>>>>> Looks good to me.
>>>>>>
>>>>>> On 26/03/2013 7:08 AM, Karen Kinnear wrote:
>>>>>>> Webrev: http://cr.openjdk.java.net/~acorn/8009731/webrev/
>>>>>> I can't quite determine why the extra ResourceMark caused the observed symptoms. I would have expected something more dramatic if we returned a string that was already released - are we simply overwriting it with a later class name (hence the wrong name in the message)? Do we also know when this regression was introduced?
>>>>> So that's why I said "Looks ok" rather than "Looks good." I think that function should return a Symbol* because it's not obvious to the casual observer that the as_C_string() string is resource allocated and the ResourceMark destructor will reclaim the memory. I think the error message might have been either garbled or pointing to a different string. It could have been like that for a while and nobody noticed.
>>>>>
>>>>> Coleen
>>>>>
>>>>>>> Bug: https://jbs.oracle.com/bugs/browse/JDK-8009731
>>>>>> This is not visible external to Oracle.
>>>>>>
>>>>>> Thanks,
>>>>>> David
>>>>>>
>>>>>>> Fix loader constraint violation error message.
>>>>>>>
>>>>>>> tests:
>>>>>>> bug report test
>>>>>>> vm.quick.testlist in parallel
>>>>>>>
>
More information about the hotspot-runtime-dev
mailing list