MPC555 : wrong assembly instruction with GCC

Sergei Organov osv at javad.ru
Tue Apr 30 14:44:48 UTC 2002


I believe MPC5xx are more like 603 with some peripherals and built-in
memory but without data cache and MMU. Some of them, namely 505/509, do have
instruction cache. Don't know for the rest of 5xx, but 505/509/555 all have
64-bit (double precision) FPU.

BR,
Sergei.

Joel Sherrill <joel.sherrill at oarcorp.com> writes:
> Sergei Organov wrote:
> > 
> > Joel Sherrill <joel.sherrill at OARcorp.com> writes:
> > > Sergei Organov wrote:
> > > >
> > > > It's not gcc. I believe gcc knows nothing about cache.
> > > >
> > > > Take a look at macro call
> > > >
> > > > _CPU_Data_Cache_Block_Flush( slot );
> > > >
> > > > at the end of the routine -- it's definition contains explicit asm statement
> > > > containing 'dcbf' instruction.
> > >
> > > In which case, RTEMS needs to wrap this with a conditional on CPU type
> > > (or better yet move it to libcpu).
> > 
> > Yes, it seems that there are still quite a few issues to be resolved in the
> > "new exception processing" code...
> 
> I think it is more likely that the sheer number of variations
> in the PowerPC family is too blame.  No single person has a 
> complete grasp on what changes or stays the same between
> different PowerPC CPU models.
> 
> With help from a former Motorola Semiconductor person, I have 
> been trying to get a clearer picture of the CPUs.   It turns 
> out there are only a handful of CPU cores which are used 
> again and again.  There is a pattern to the part number which
> indicates the core.  The cache varies widely in spite of that.
> 
> Just to doublecheck my info, does the MPC555 have an FPU?
> Is it essentially a 601 class CPU?  From your perspective
> it may look like a 603 with peripherals.
> 
> -- 
> Joel Sherrill, Ph.D.             Director of Research & Development
> joel at OARcorp.com                 On-Line Applications Research
> Ask me about RTEMS: a free RTOS  Huntsville AL 35805
>    Support Available             (256) 722-9985




More information about the users mailing list