design Real-Time Process in RTEMS

[rtems skyeye]

hi:
  the RTP prototype will be tesed in X86 CPU.
  I already let paging (1phy--1 virtual) work correctly. (just use the
Functions already in RTEMS)
  I also try  the process  CPL0--> CPL3 then syscall from CPL3-->CPL0 and
made it work.
  Now, I am trying the user-level thread(in CPL3) can cooperate with
kernel-level thread(in CPL0) without paging.
  Next,  I will try the user-level thread(in CPL3) can cooperate with
kernel-level thread(in CPL0) with  1-1 paging.

  I spent my spare time on these works, so the progress is slowly.

  Anyone have some advice?

On 9/3/07, rtems skyeye <skyeye.rtems at gmail.com> wrote:
>
> hi,
>   Good idea!
>   I think we can first realize the initial VM support in RTEMS.
>   because there  are already some page support funcs in rtems X86 port.
>   then we can try syscall,....
>
>
>
> ---------- Forwarded message ----------
> From: 阎淼 <yanmiaobest at gmail.com>
> Date: Sep 3, 2007 11:21 AM
> Subject: Re: design Real-Time Process in RTEMS
> To: rtems skyeye <skyeye.rtems at gmail.com>
>
> Hi:
>
>    I have been thinking about it too.
>    here is what I think:
>
>    1.  page routine should be created to support page manipulations(
> page table operation, TBL, page allocate, set page state ...)
>    2.  identify all system call routines
>    3.  create entry points and exit points for system call
>    4.  adjust context switch ralated routine
>    5.  add privilege transition code
>
>    best regards!
>
> 2007/8/31, rtems skyeye <skyeye.rtems at gmail.com>:
> > hi,
> >   VxWorks 6.x have the Real-Time Process (RTP) extension.
> >   I think we also can implement the RTP in RTEMS.
> >   The Idea is :
> >   1 RTP contains 1~n user-level rtems-task, and process the
> > virtual memory related things.
> >   2 the mapping relation of virtual memory :phyiscal memory is 1:1
> >   3 the user-level task in RTP can access kernel function
> > by syscall
> >   4 the user-level tasks and the kernel-level tasks are in the same
> > scheduling strategy.
> >
> >   the advantage of RTP is:
> >    1 the crash of user-level tasks in RTP can not destroy user-level
> > tasks in other RTP and the kernel.
> >    2 compatible with the old task model.
> >    3 the context switch  is faster than that of general OS (such as
> linux)
> >
> >    anyone have  some advices on RTP in RTEMS?
> >
> > Regards
> >
> >   skyeye.rtems
> >
> >
> > _______________________________________________
> > 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/20070917/02352129/attachment-0001.html>