Unexpected trap(0 x4) ... fp disabled

[Aitor Viana]

What optimization options are you using? Some options (off the top of my
head -O3) sometimes use floating point registers when passing parameters. If
that's so, use -O2

--
Aitor


On Wed, Jun 23, 2010 at 7:36 PM, Eric Norum <wenorum at lbl.gov> wrote:

> 1) Floating point exceptions on many architectures can be fairly imprecise.
>   Could the offending code be back a line or two?
> 2) There's always the possibility that a runaway pointer or bad array index
> has resulting in your code being mangled.   Can you use a debugger to check
> that the instructions which you expect to be around that address are indeed
> still there?
>
> On Jun 23, 2010, at 10:07 AM, Gardner, Michael T wrote:
>
> I am running a test program on a leon3 using rtems 4.10.  When the program
> executes a standard “read()” call I see the following error message and
> RTEMS and the program terminate abnormally:
> Unexpected trap (0x 4) at address
> 0x40006710
> fp disabled
>
> The offending source line of code is (all arguments are valid):
> *bytes_read = read(fd, the_packet, max_read_size);
>
> I have tried compiling my source code both with and without the
> –msoft-float compiler flag but it crashes each time.  There are some
> external libraries that are not compiled with the –msoft-float option that
> are linked into the code.  Could this be contributing to the problem?  I am
> using the POSIX API with RTEMS and therefore believe that all of my threads
> are floating point threads. (ref: RTEMS Wiki – Floating Point Support).
>
> I have looked at previous posts and the RTEMS Wiki for answers regarding
> this issue, but am still somewhat confused about what I need to do (I’m an
> RTEMS newbie).  In the Wiki there is a section entitled “Escaping Disaster”
> in which several options are provided.  Our application needs to use the
> hardware FPU and the second option seems to be the option that is most
> applicable.  Does anyone have experience with implementing this option?
> What lessons learned can be passed on to make this as painless as possible?
>
> Michael T. Gardner
> Sandia National Labs
> mtgardn at sandia.gov
> 505-844-1299
>
>
>  --
> Eric Norum
> wenorum at lbl.gov
>
>
> _______________________________________________
> rtems-users mailing list
> rtems-users at rtems.org
> http://www.rtems.org/mailman/listinfo/rtems-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20100623/90decd79/attachment.html>