Re: [Sdcc-user] crashing a microcontroller - killing 8051 - hanging a controller !!

[karthik b. g. <sdc...@ya...>]

ya i'm basically an Electronics Engg . Thanx for
your tips . Xpecting the same from you.
regards,
karthik bala guru


--- Shehryar Shaheen <she...@ul...> wrote:
> Dear Friend
>                    I think you never paid attention
> during your
>  computer architecture course.
> 
> It's just the way a stack works in every processor.
> 
> Every call you make to a function the processor
>  pushes the current program counter location on to
> the
> stack if the stack is say 2 locations deep then
> greater than
>  1  nested function call from within a function will
> cause
>  a stack over flow and the processor will never be
> able to
>  return to the point from where it made the call.
> 
> Function calls from within functions are dependant
> on
>  stack depth. Depending on processor architecture if
> the
>  nested call exceeds the stack depth the last
> location on the
>  stack is either over written or the last pushed
> address will
>  be lost.
> 
> In any case stack over flow as well as under flows (
> under flow is when you
>  'pop' more than what has been 'pushed')  to be
> avoided and such information
> is
>  always given in microcontroller data sheets.
> 
> Same is with nested interrupts, nested interrupts
> are also dependant on
> stack depth.
> 
> These are considerations that you need to take in
> account  when writing code
> specially
>  for smaller microcontrollers.
> 
> Best Regards
> 
> Shehryar
> 
> 
> ----- Original Message -----
> From: "karthik bala guru" <sdc...@ya...>
> To: <sdc...@li...>
> Sent: Sunday, July 20, 2003 3:40 PM
> Subject: RE: [Sdcc-user] crashing a microcontroller
> - killing 8051 - hanging
> a controller !!
> 
> 
> > Dear friends,
> > I made an interesting observation. The stack
> > pointer moves on each function call and the
> controller
> > restarts/hangs when the stack limit is reached.
> > This occurs especially when a function calls
> > a function and that function inturn calls another
> > function and the flows goes on.
> > So, I think it is better to avoid calling
> > a function from within a function . Even if we
> > do that, we have restrict it to a limit.
> >
> > MONITOR THE STACK .....
> >
> > karthik bala guru
> > li...@ya...
> >
> >
> >
> > --- Simon Hosie <Sim...@co...>
> wrote:
> > > karthik bala:
> > > > fine guess. I am into some of the undocumented
> > > methods tooo. I am
> > > > waiting for such postings from this group . At
> > > present i am mainly
> > > > concentrating on the s/w segment. I would get
> deep
> > > into the h/w
> > > > concepts dealing with this too.
> > >
> > > Andy Green:
> > > > It will be a bit of a short experiment --
> there is
> > > only one undefined
> > > > opcode in 8051, 0xa5.
> > >
> > > There are some other undefined instructions,
> too.
> > > For example, I recall an Atmel datasheet saying
> > > something like 'mov A, ACC' was undefined.
> > >
> > > Remember, though, that hacks like this are more
> > > often about finding things that the designers
> didn't
> > > think of, not about the things the designers
> > > wouldn't specify.


__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free, easy-to-use web site design software
http://sitebuilder.yahoo.com