FLOATING POINT with rtems

Luca Cinquepalmi cinquepalmi at planetek.it
Fri Nov 29 12:20:14 UTC 2013


Hi all,
I'm not yet able to use FLOATING POINT coprocessor in my sparc architecture.
I configured my system.h by inserting

#define|CONFIGURE_INIT_TASK_ATTRIBUTES RTEMS_FLOATING_POINT|

but alway the same trap appears:

Unexpected trap(0 x4) ... fp disabled

I don't want to use -msoft-float flag option because I don't use
it when I compile with BCC instead of RCC and it does work without any 
problems.
I don't know why rtems gives me this kind of problems.
Can anyone suggest me something?
Thanks a lot

Luca

Il 19/11/2013 18:24, Luca Cinquepalmi ha scritto:
> Hi all,
> I have the following problem.
>
> I've declared the following structure
>
> struct str {
>     int a;
>     int b;
>     float arr[2]
> }
>
> in my Init task I declared:
>
> struct *str mystr;
> mystr = (str*)calloc(1, sizeof(*mystr)
> mystr -> a = 1;
> mystr -> b = 2;
> mystr -> arr[0] = 1.2;
> mystr -> arr[1] = 2.1;
>
> When i plot my value a trap error occours and no valu
> is plotted
> If my array is of int, all goes good.
> Is there a particular problem for float array in structures
> with RTEMS? thanks.
>
> Luca
>
>
>
> Il 19/11/2013 12:06, Sebastian Huber ha scritto:
>> On 2013-11-19 11:56, Leonard Bise wrote:
>>> Am I correct in thinking that this section of the RTEMS doc :
>>> http://rtems.org/onlinedocs/doc-current/share/rtems/html/c_user/Multiprocessing-Manager.html#Multiprocessing-Manager 
>>>
>>>
>>> Talks about what I want or is that something else entirely? I'm a 
>>> bit loss with
>>> your comment about multiprocessing support in RTEMS Sebastian.
>>
>> RTEMS_MULTIPROCESSING:
>>
>> Independent single-processor RTEMS instances share global objects.  
>> State changes are exchanged by means of message passing.
>>
>> RTEMS_SMP:
>>
>> One symmetric multi-processor RTEMS instance runs on multiple 
>> processors.
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20131129/80e092fe/attachment-0001.html>


More information about the users mailing list