RFC: powerpc-multilibs: remove -mcpu=601 and -Dmpc8260 multilibs

[Joel Sherrill]

On 02/21/2011 11:35 AM, Till Straumann wrote:
> I don't know who made that (IMHO a bit unfortunate) change.
> The BSP should IMO use -D__ppc_generic -msoft-float.
What is should use and what it always has used are
apparently two different things.  It has had that same
compiler flags since it was added in 2004.

http://www.rtems.org/viewvc/rtems/make/custom/mvme2100.cfg?hideattic=0&revision=1.2&view=markup
> All BSPs that use a classic PPC should get away with
> -D__ppc_generic with or w/o -msoft-float, depending
> on whether the CPU has or doesn't have a FPU.
> If the CPU has AltiVec then -maltivec, -mabi=altivec
> should be added. I.e., for classic PPCs only three
> multilibs are really needed:
>
> generic PPC instructions with FPU
> generic PPC instructions with FPU and ALtiVec
> generic PPC instructions w/o FPU and w/o AltiVec
>
> For booke-type CPUs another (or multiple other) multilib(s)
> (than the generic PPC one) is/are necessary.
>
I won't argue with you on this from a functional
viewpoint.  Whether RTEMS itself can be reduced
to this set I don't know.  I know the libcpu code
has to distinguish in places between a 60x and 750.
I have no idea about cpukit/score/ though.
> T.
>
>
> On 02/21/2011 06:24 PM, Joel Sherrill wrote:
>> On 02/21/2011 11:21 AM, Till Straumann wrote:
>>> FYI: users of the mvme2100 BSP should make sure
>>> the -Dmpc8260 multilib is no longer needed.
>> The mvme2100 is an mpc8240. It uses these flags:
>>
>> CPU_CFLAGS = -mcpu=603e -Dppc603e
>>
>> So it should not be impacted.
>>
>> --joel
>>> T.
>>>
>>>
>>> On 02/21/2011 05:24 PM, Ralf Corsepius wrote:
>>>> On 02/18/2011 03:13 PM, Ralf Corsepius wrote:
>>>>> Hi,
>>>>>
>>>>> When looking into Sebastian's proposal to change the powerpc
>>>>> multilibs[1], I believe to have identified the
>>>>>
>>>>> * -mcpu=601 [2]
>>>>> and
>>>>> * -Dmpc8260 [3]
>>>>>
>>>>> multilibs as unused.
>>>>>
>>>>> Am I missing something or does this apply? The latter would mean these
>>>>> multilibs can be removed.
>>>> I removed these 2 multilibs in rtems4.11's latest toolchain update.
>>>>
>>>> Please speak up ASAP, should I have broken something.
>>>>
>>>> Shouldn't I hear something by the end of this week, I'll likely also
>>>> remove these multilibs from the rtems4.10 toolchains, as well.
>>>>
>>>> Ralf
>>>> _______________________________________________
>>>> rtems-users mailing list
>>>> rtems-users at rtems.org
>>>> http://www.rtems.org/mailman/listinfo/rtems-users
>>> _______________________________________________
>>> rtems-users mailing list
>>> rtems-users at rtems.org
>>> http://www.rtems.org/mailman/listinfo/rtems-users
>>


-- 
Joel Sherrill, Ph.D.             Director of Research&  Development
joel.sherrill at OARcorp.com        On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
    Support Available             (256) 722-9985