Code review request: CR 7141259 Native stack is missing in hs_err

[Karen Kinnear]

Zhengyu,

The code looks like it will work.

I would appreciate a comment that we never cleanup the error handler decoder since it
is only used on crashing.
Just to be sure I understand - there are currently no callers of Decoder::shutdown(), i.e.
no users of the shared decoder are in the system yet.
Longer term, with additional cases I would recommend managing the scope of the decoder
instance outside the decoder code itself.

thanks,
Karen

On Feb 6, 2012, at 9:30 AM, Zhengyu Gu wrote:

> David,
> 
> I total agree that it is a bit of hack. The problem is that, decoder actually hides behind os::dll_address_to_function_name(), which makes it less flexible in design. Unless we are willing to mark it as multi-thread unsafe, and manage decoder instance and lock outside the scope. Native memory tracking has such flexibility, I think that the real issue is how to redesign dll_address_to_function_name() API.
> 
> 
> Thanks,
> 
> -Zhengyu
> 
> On 2/6/2012 3:14 AM, David Holmes wrote:
>> Hi Zhengyu,
>> 
>> On 4/02/2012 8:12 AM, Zhengyu Gu wrote:
>>> Thanks for reviewing. I think I made wrong assumption that the thread is
>>> the last running thread.
>>> 
>>> 803 // This should be the only running thread, decoder can safely run in
>>> 804 // single threaded mode without acquiring lock
>>> 805 Decoder::set_single_threaded(true);
>>> 806
>>> 
>>> To avoid locking in error handler, the only solution I can think of, is
>>> to have its own private decoder.
>>> 
>>> Please see revised webrev:
>>> http://cr.openjdk.java.net/~zgu/7141259/webrev.01/
>> 
>> This seems functionally ok. It is a bit strange to have a static API built on the assumption of a singleton instance now having to special case "is this the VM error thread" and use more than once instance, but I think changing that requires revisiting the whole Decoder framework.
>> 
>> David
>> -----
>> 
>>>> Later when there are multiple users of the Decoder I don't know how
>>>> you will fix this because those other users can run concurrently with
>>>> report_and_die(). That said I'm unclear what it is about the Decoder
>>>> that requires locking - what state does it maintain? Is there a reason
>>>> not to create a Decoder only when needed?
>>>> 
>>> Decoder is created on demand, but would prefer to be shared, since it
>>> can consume noticeable amount memory on Unix platform, by loading Elf
>>> files' symbol tables and string tables into memory (not impact on
>>> Windows). The lock is needed to protect a linked list structure in Elf
>>> decoder, or current file position.
>>> 
>>> Thanks,
>>> 
>>> -Zhengyu
>>>> ---
>>>> 
>>>> In decoder.cpp you can still use MutexLockerEx and as it will take a
>>>> null mutex and just act as a no-op eg:
>>>> 
>>>> MutexLockerEx locker( _single_thread ? NULL : _decoder_lock, true);
>>>> 
>>>> Cheers,
>>>> David
>>>> -----
>>>> 
>>>>> 
>>>>> Please review
>>>>> Webrev: http://cr.openjdk.java.net/~zgu/7141259/webrev.00/
>>>>> 
>>>>> 
>>>>> Thanks,
>>>>> 
>>>>> -Zhengyu
>>>>>