hw_breakpoint userland interface

[Roland McGrath <roland at redhat dot com>]

There are indeed many things to be figured out about a userland interface
for hw_breakpoint.  It's certainly quite possible that something entirely
separate from ptrace is the right approach for that.  I haven't given it
any thought, really.  It should be worked out with Jan and other GDB folks
as well as whoever in the kernel sphere cares about hw_breakpoint stuff.
It might be a ptrace-based interface, or it might be a separate interface
that interacts with ptrace (perhaps just generating some new PTRACE_EVENT_*
type for its stops).

If the hw_breakpoint stuff is well-integrated with perf event stuff, then
perhaps a good approach involves a more general-purpose interlock between
ptrace and perf event stuff.  That is, a way for GDB to make use of all
that stuff in a somewhat clean fashion for its case, that being of
unprivileged all-userland control that ties into ptrace stops somehow.


Thanks,
Roland