Re: [RFA]: threaded watchpoint test

[Daniel Jacobowitz <drow@false.org>]

On Tue, Jul 27, 2004 at 08:12:42PM -0400, Michael Chastain wrote:
> > OK.  Unfortunately I don't think this test is conclusive; unix.exp will
> > always claim to have hardware watchpoints, and on many systems it does
> > not (for instance powerpc-linux).  If that turns out to be true we can
> > introduce gdb_has_hardware_watchpoints or something along those lines.
> 
> Yeah, I'm a bit leery of "set_board_info gdb,no_hardware_watchpoints"
> but it already exists (although not used for that purpose).
> 
> I missed something, though:
> 
>   if [target_info exists gdb,no_hardware_watchpoints] {
>     return 0;
>   }
> 
> Perhaps there should be an UNSUPPORTED in there before the "return 0":
> 
>   unsupported "hardware watchpoints on this target"
> 
> Daniel, what do you think?

Sounds good to me.  We'll probably want to clean up the other bit about
using the board_info too; or... just recognize that we got a software
watchpoint instead of a hardware one?

Maybe not.  It could be a bug to get a software watchpoint when we
expect a hardware one.  The question is whether we want to keep a
separate list of "do we expect hardware watchpoints" in the testsuite.

-- 
Daniel Jacobowitz