harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gregory Shimansky <gshiman...@apache.org>
Subject Re: [general] Board report due
Date Thu, 13 Dec 2007 13:37:36 GMT
Tim Ellison said the following on 13.12.2007 13:51:
> Gregory Shimansky wrote:
>> Tim Ellison said the following on 12.12.2007 21:05:
>>> Gregory Shimansky wrote:
>>>> On 12 декабря 2007 Pavel Ozhdikhin wrote:
>>>>> I've added several comments about new JIT features/optimizations.
>>>>>
>>>>> I think the following item should be removed from the list:
>>>>>
>>>>>  - 'full speed debug' : debugging with JIT enabled
>>>>>
>>>>> It seems Gregory's comment was misinterpreted - he probably meant
>>>>> that the
>>>>> JIT-mode JVMTI debugging now works on x86_64. AFAIK it worked fine
>>>>> on IA32
>>>>> in M3. Gregory, could you please confirm?
>>>> Yes, what I meant is that now it is possible to debug applications in
>>>> default (JIT) mode on x86_64. Previously in M3 this mode worked only
>>>> on x86, on x86_64 it was necessary to use -Xint to use interpreter.
>>> I'm sorry, that was my fault.  I misunderstood.  So what does 'debug
>>> applications in default JIT mode' mean?
>> It means that debugged application's bytecode is compiled, not interpreted.
> 
> So that's what I mean by 'full speed debugging' -- usually written in
> quotes since it typically isn't really 'full speed'.  Certain aggressive
> JIT optimizations will have to be turned off to permit debugging.
> 
> Are we talking about the same thing?

Yes, with an exception that no optimizations are done when a program is 
debugged. But execution is still much faster than when the code is 
interpreted.

> e.g. http://java.sun.com/j2se/1.4.2/docs/guide/jpda/enhancements.html#fsd

Sun doesn't describe whether they do any compiling optimization in this 
mode, maybe they do the same thing and call it FSD :)

-- 
Gregory

Mime
View raw message