RTEMS 4.7.99.2 Available -- PowerPC/virtex feedback

[Joel Sherrill]

Chris Caudle wrote:
> On Fri, August 10, 2007 9:19 am, Joel Sherrill wrote:
>   
>>          {"403", PROCESSOR_PPC403,
>>           POWERPC_BASE_MASK | MASK_SOFT_FLOAT | MASK_STRICT_ALIGN},
>>          {"405", PROCESSOR_PPC405,
>>           POWERPC_BASE_MASK | MASK_SOFT_FLOAT | MASK_MULHW | MASK_DLMZB},
>>     
>
>
> Doesn't that imply that the tools for PPC405 should already be using soft
> float?
>   
Correct and the 403 is also soft-float. 

FWIW RTEMS only has a multilib for the 403 right now.  The
405 links with the 403 libraries which means that the 403 has
to be a proper subset of the 405 or you will get incorrect
instructions.

DLMZB -- determine left most zero byte should only
result in improved strlen for 405.
HULMW -- appears to be hardware multiply and accumulate
so it might have some performance impact but that is unknown.


My conclusion is that the 403 is (from gcc's perspective) a proper
subset of the 405.
> Everyone is scurrying around figuring out how to rebuild tools with soft
> float, but it looks like that should be the default already, and so far no
> one has presented any code which confirms that the exception was actually
> caused by using a floating point register.
>
> Shouldn't looking at the instructions to see if that is a feasible
> explanation be the first step?
>   
Yes.

--joel