Re: iPAQ214 Resume tracing/boot support and mmutrace unknown instruction.

Kevin O'Connor wrote:
> Yes - the hooking of the resume vector seems to work on wince5
> phones. (It works on both my Apache and Hermes.) I know of at least
> one other phone (Magician I think) running wince4 that the current
> system did not work.
>
>
This is winCE 5.2 (Mobile 6 classic). I think the code must be specific
to the machine though, since the suspend resume code does a lot of
fiddling with pxa3xx specific things.
>
> It would be great if you could send a patch over. Maybe Philipp can
> compare it to what he had on Magician.
>
>
Ok, I'll send you something some time next week.
> You should put a copy of objdump into the same directory as
> haretconsole - it will automatically disassemble for you then.
>
>
My copy of objdump for arm runs inside scratchbox, so that might be a
pain but does sounds useful.
> Yes. I never implemented ldm (load multiple) or stm (store multiple),
> because it is a bit of a pain and I never ran into anything that
> actually needed it. It is definitely possible to emulate it.
>
>
Yes, I've now done so now. I've got the emulation working for all
multi-word store and loads. I just have the actual trace logging to sort
out. I'll send you the patch for that as well in a few days. It took a
while to sort out what happens with all the different pre/post
inc/decrementing cases, though turns out to be quite simple.

The only problem I can think of is that it will fail (get things wrong
and not notice) if the fault occurred at a memory address half way
through the operation. It would require the operation to be taking place
over the edge of the region haret has remapped. I'm not even sure if
this is possible though.

Oliver
Received on Fri Apr 11 2008 - 05:09:08 EDT