RFR (S): 8003195 - AbstractAssembler should not store code pointers but use the CodeSection directly

[Christian Thalinger]

Looks good.  An integration job is in the queue.

-- Chris

On Nov 30, 2012, at 5:07 AM, Bharadwaj Yadavalli <bharadwaj.yadavalli at oracle.com> wrote:

> Updated webrev. Please review.
> 
> http://cr.openjdk.java.net/~bharadwaj/8003195/webrev/
> 
> Re-ran JPRT successfully.
> 
> Thanks,
> 
> Bharadwaj
> 
> On 11/29/2012 2:48 PM, Vladimir Kozlov wrote:
>> On Nov 29, 2012, at 11:48 AM, Christian Thalinger wrote:
>> 
>>> On Nov 29, 2012, at 11:33 AM, Vladimir Kozlov<vladimir.kozlov at oracle.com>  wrote:
>>> 
>>>> I don't like explicit casting of address to int64 or int32. I would prefer original code moved to codeBuffer as other functions.
>>> You're saying adding:
>>> 
>>> +   void emit_address( address  x) { *((address*)  end()) = x; set_end(end() + sizeof(address)); }
>>> 
>>> to CodeSection?
>> Yes.
>> 
>>>> An other good cleanup would be remove emit_word() and emit_long() which are very confusing sizewise. But it is for next round.
>>> Yes.  That's the plan for my other upcoming changes.  This is the first step in that direction.
>> OK.
>> 
>> Vladimir
>> 
>>> -- Chris
>>> 
>>>> Thanks,
>>>> Vladimir
>>>> 
>>>> 
>>>> On Nov 29, 2012, at 8:07 AM, Bharadwaj Yadavalli wrote:
>>>> 
>>>>> Please review
>>>>> http://cr.openjdk.java.net/~bharadwaj/8003195/webrev/
>>>>> 
>>>>> This change is a result of applying the patch from Christian Thalinger attached in the bug report with a couple of minor tweaks and are for x86 architecture only.
>>>>> 
>>>>> Thanks,
>>>>> 
>>>>> Bharadwaj
>>>>>