On Sunday 18 April 2010, [email protected] wrote:
> >> feature allows breaks upon an access across a range of addresses.  However,
> >> theOpenOCD wp command does not support this
> >
> >In what sense "not support"?  I hae certainly configured and used OpenOCD
> >watchpoints with of 64 KByte size, thus, "a range of addresses".  (And ISTR
> >larger ranges, where hardware allows.)
> 
> Does the size argument to the wp command refer to the size of a block of
> memory?


Yes ... as you assumed, but that's not always a trivial case.





 I assumed it meant the size of a single access (byte, half-word, or 
> word). Ditto for the size argument to the bp command (2 for thumb mode, 4 for
> arm mode).

And for 8-byte accesses, floating point, LDM/STM???


restrictions to 1/2/4 bytes would be  needlessly limited.




> Indeed, the xscale implementation of wp returns an error if not 1, 
> 2, or 4. And the same code subsequently ignores that argument, so if size
> refers to a memory range, xscale doesn't support it.

Fixable bug.  See how e.g. various Cortex chips handle it, EICE,
and ARMv6-m (DPM support), etc....  Size not ignored .... but it
is properly validated.

DPM hardware   (on ARMv6 and ARMv7 is most flexible there.)


> If my interpretation of 
> the argument is incorrect, perhaps the correct remedy is to use the mask
> capability of the xscale debug hardware to implement the command correctly.


I'd do that, yes.



> I'll take a look at the implementation in other targets. It would be good to
> patch the documentation to clarify the meaning.


ISTR noticing doc issues, yes.



> Thanks,
> Mike
> 
> ----
> 
> _______________________________________________
> Openocd-development mailing list
> [email protected]
> https://lists.berlios.de/mailman/listinfo/openocd-development
> 
> 


_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development

Reply via email to