chrisj at rtems.org
Sat May 28 04:41:41 UTC 2005
Joel Sherrill <joel at OARcorp.com> wrote:
> If you want to do stack usage in the capture engine, then I would
> suggest splitting the stack checker code into two pieces. Stack
> filling and usage helpers and the checking extension itself.
Yes I agree. A common pattern for each approach. Having the score's
thread create fill the stack will enforce a common approach.
> The stack checking code is fairly complicated and the amouunt of
> guard checked varies based upon the CPU's stack frame and caching.
> For example, the SPARC and i960 have relatively large call frame
> caches before they write to memory.
Yes and why I want to have a common set of code.
> Shouldn't really require much more than some mechanical code'
> and file splitting.
Ok. I have no problem with the capture engine borrowing code from else
where in RTEMS, so what if I place the stack fill code into the score,
and split the current stackchk code into parts usable by the capture
More information about the users