Re: [ECOS] eCos Code at address 0x000? (MPC8270)

[Gary Thomas <gary at mlbassoc dot com>]

On Tue, 2004-05-18 at 12:53, Andrew Lunn wrote:
> > Interesting.  Does that suggest that someone may be calling 
> > a NULL function pointer?  Where is the code for this stub? 
> > 
> > We've seen some crashes at address 0x4.  Dumping that area, 
> > it appeared that the values around 0 had been written to 
> > (trampled), because they no longer matched the power up 
> > values.
> > 
> > Later, I manually "trampled" the area around 0 with a debug 
> > command, and the target crashed shortly afterwards. 
> 
> I don't know anything about the PPC.......
> 
> On the ARM, the exception vectors are in the first few words of
> memory. One classic problem i've had is dereferensing a null pointer
> to a structure and stomping over the ISR vector. This does not cause
> an imeadiate crash, rather it crashes on the next interrupt... This
> makes it very had to track down what exactly is stomping over the
> vector. You might has a similar problem?

PowerPC vectors are very different from ARM - they are 0x100 bytes
each as opposed to 0x4 (one instruction).  That said, as I pointed
out, nothing should be touching the memory near 0x0 in any case, so
I think that Tim has some sleuthing to do...

-- 
Gary Thomas <gary@mlbassoc.com>
MLB Associates


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss