Need help. when porting RTEMS to L4

[xi yang]

On Thu, Mar 20, 2008 at 1:06 AM, Joel Sherrill
<joel.sherrill at oarcorp.com> wrote:
> xi yang wrote:
>  > Hi all:
>  > I am porting RTEMS on L4 micro kernel . I just
>  > do it like what UNIX bsp does . But there are
>  > some problems
>  > I develop BSP like this
>  > * cpukit/score/cpu/l4  (Like unix)
>  > * c/src/lib/libbsp/l4/okl4 (like UNIX/posix)
>  >  I copy files form unix bsp and change Makefile.am
>  > and other files .
>  > * touch okl4.cfg
>  >   RTEMS_CPU=l4
>  >   RTMES_CPU_FAMILY=arm
>  >   RTEMS_CPU_MODEL=pxa255
>  >
>  >   RTEMS_BSP_FAMILY=okl4
>  >
>  The configure infrastructure uses RTEMS_CPU to select
>  directories at the score/cpu and c/src/lib/libcpu and
>  c/src/lib/libbsp level.
>
> > * Add score/cpu/l4/Makefile to cpukit/configure.ac
>  >
>  >
>  So you accounted for this one.  D
>
> > then ../rtems/configure --enable-rtemsbsp="okl4"
>  > This cause the toolchain to the host toochain ,
>  > maybe it can compile some files ,then
>  >
>  Right. This configures the target to be your native machine (e.g. posix
>  port)
>  with a BSP that does not exist under libbsp/unix.
>
>
>  > Make
>  > but It doesn't compile any files , just say
>  > make all , then ../rtems/configure --target=arm-rtems --enable-rtemsbsp="okl4"
>  > Make
>  > It doesnot compile any files, just say make all,
>  >
>  >
>  That is looking for okl4 in libbsp/arm.
So, if I want to do this way , I must have the l4-rtems- toolchain ?
>
> > What else should I modify the build system to let it compile ?
>  > anyone can help me ?
>  >
>  >
>  Your configure points are confusing me.  You are treating L4 as a
>  CPU architecture from RTEMS perspective.  What really has to
>  change?
>
>  We have to know the answer to the above question to
>  know where to distinguish between real hardware and
>  an L4 version of the same hardware.
>
>  My gut feeling is that L4 should be treated as a specific CPU
>  model within a CPU family because you still want to compile to
>  ARM but probably have a somewhat perverted view of the
>  interrupts and maybe the hardware, right?

Yeah, I want to compile it to ARM now .
>
>  This would mean you should add okl4 as a BSP and possibly
>  some ARM-L4 CPU model.  Say it is pxa255 for real, then make
>  it pxa255l4.  Then you can treat it as a real raw pxa255 when
>  possible and still ahve the ability to detect it as an L4 virtualized
>  pxa255 when it makes something different.

Yeah, that will add some #if #endif things to source code ^-^.
Thanks, joel .
>
>  It may even be that the same BSP can work with one variant for
>  real hardware and another for L4 virtualized with a few conditionals
>  to account for the CPU model differences.  This is done by a few
>  BSPs including the gen68360, mbx8xx, and motorola_shared.
>
>  --joel
>  > Regards
>  > _______________________________________________
>  > rtems-users mailing list
>  > rtems-users at rtems.com
>  > http://rtems.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
>
>
>