Request for reviews (XS): 6902000: use ShouldNotReachHere() for btos/ctos/stos in TemplateInterpreterGenerator::set_short_entry_points

[Coleen Phillimore]

If there's any complication, it's not worth doing then.
The simple fix looks fine.
Coleen

John Rose wrote:
> IIRC, one place where that distinction matters is on return from native methods, where the EAX/O0 value needs conditioning before it pushes onto the JVM stack.  But of course BasicType is the real story at that point.
>
> -- John
>
> On Nov 19, 2009, at 4:19 AM, Christian Thalinger wrote:
>
>   
>> On Wed, 2009-11-18 at 14:37 -0500, Coleen Phillimore - Sun Microsystems
>> wrote:
>>     
>>> While you are there, I wonder if you can't just get rid of the btos,
>>> ctos, and stos entries in EntryPoint and save some footprint too.  I
>>> am not sure if it's worth it: 2 tables (normal+safepoint) * 256
>>> entries * 8 tos states * sizeof (address).  4096 pointers vs. 2560
>>> pointers.  Not counting the safepoint and other tables of EntryPoints.
>>>       
>> Hmm, sounds like a good idea if it is safe to remove these.  I'm not
>> very familiar with the interpreter but I can give it a try.
>>
>> -- Christian
>>
>>     
>
>