memory footprint

[Joel Sherrill]

Wulf Hofbauer wrote:
> 
> On Tue, Nov 26, 2002 at 02:32:26PM -0600, Joel Sherrill wrote:
> > You're on a mvme147, right?  If I recall everything correctly,
> > that avoids conflicting with memory space used by Mbug.
> 
> MVME147S, to be precise. The executable there is linked for
> base address 0x7000 which is out of the 147bug region anyway
> (IIRC, 147bug uses only memory below 0x4000, and the initial
> stack seems to grow down from 0x5fff). In any case, it's a waste
> - if you put the code in ROM and copy it to RAM on startup, the
> ROM space is still wasted. That data space could just as well
> live in .bss and, if absolute memory addresses were required for
> architectural reasons (which is not the case, AFAIK), it would be
> easy enough to provide them either via a linker script or simply
> by .equ.

I co-wrote the original mvme135 BSP and the linker script loads
it a few K lower and leaves a gap.  Maybe the author didn't adjust
the fudge out when they copied it.  I am betting it can be removed 
if you are careful.  Just watch the vectors during the startup
transition to RTEMS and exit path back to the bug.  

It is also possible that some memory is being used as a starting
stack which might be possible to avoid since the Bug gives an
application stack.  That should be enough to start RTEMS on.

> - Wulf

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel at OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
Support Available                (256) 722-9985