Ulrich Weigand wrote:
> > Though that may be better taken care of by the kernel-level monitor...
> > So okay, we could make this an ioctl call. But then, I guess
> > you'd need to provide a similar system for breakpoints and
> > such... the problem is that a breakpoint interface would
> > be unusable by remote GDB, without weird hacks.
>
> GDB uses remote memory access to write 'int 3' instructions?
Yes. It has the remote stub set the trap flag though...
rather inconsistant, coming to think of it ;)
> Well, either we just do it that way and accept that this might
> give interactions with certain guests, or else we'll have to
> analyse those writes and convert them to monitor breakpoint
> calls.
That may be a good idea. Though it makes it hard to differentiate
between writes to code and explicit writes to data... we need to
think about that.
Idea: in the monitor, we may use an invalid instruction for
breakpoints, in stead of int3. This will trigger emulate()
which will identify it as a breakpoint for the host debugger.
Ramon
