M68k/ColdFire: FPU context initialization
Brett Swimley
brett.swimley at aedbozeman.com
Mon Nov 5 18:22:22 UTC 2007
It is a bit different.
I have a patch that saves and restores the FP context for the MCFv4e
that I keep meaning to submit (based on the -mcfv4e compiler option
found in the newer gcc compilers). I will try to put that together.
I will look at the FPU control register initialization. I don't know if
I did that.
On this topic, would it be possible to add the mcfv4e variant to the
list of m68k targets (t-rtems) and include this in the pre-built RPMs?
I added this variant to my version of the compiler, but I don't know if
I got all the soft-float stuff correct.
Brett
Joel Sherrill wrote:
> Thomas Doerfler wrote:
>
>> Hi,
>>
>> we are currently adding FPU support for a ColdFire v4e CPU derivate.
>> After reading, re-reading (and reading again) the relevant files in
>> cpukit/score/cpu/m68k in the current M68K (with hardware FPU, e.g.
>> MC68040) CPU Context management, I have not found, where/how the FPU
>> context is initialized for a newly created thread.
>>
>> Can somebody please guide me:
>>
>> - Is the FPU context in an arbitrary state when a new thread is started?
>>
>>
> If the thread is not an FP task, you would prefer to disable the FPU.
> If it is FP enabled, then all that you have to initialize is usually
> the control registers.
>
>> - Or: Where is the code which defines the initial state of the FPU in a
>> new thred?
>>
>>
>
> score/cpu/m68k/rtems/score/cpu.h and score/cpu/m68k/cpu_asm.S
>
> Is the Coldfire FPU similar to the 68881 or is it more RISC like?
> Do you have the same type of fsave/restore instructions?
>
>> Any hint appreciated,
>>
>> Thomas Dörfler.
>>
>>
>>
>>
>>
>
> _______________________________________________
> rtems-users mailing list
> rtems-users at rtems.com
> http://rtems.rtems.org/mailman/listinfo/rtems-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20071105/2322bccf/attachment-0001.html>
More information about the users
mailing list