Re: remote watchpoint support

[Mark Salter <msalter at redhat dot com>]

>>>>> Steven Johnson writes:

> Mark Salter wrote:
>> 
>> Ok. Here's a patch that allows a remote target to pass watchpoint
>> information back to gdb using the T packet. It also fixes the T
>> packet handling to ignore other optional info which may not be
>> recognized. This changes the behavior to match the documentation.
>> 

> I was just looking in this area, and noticed that the 'T' packet seems to
> ignore the signal number it is sent! Isn't this a bug? 

> All the other responses ('S','N','X') update status->value.sig to be the value
> of the signal, 'T' doesn't. Is this valid?

If you look closely, you will see that the 'T' case falls through to
the 'S' case where the signal value is read and reported. This itself
raises a question, though. There is some Cisco-specific support in
the 'S' case where a check is made to see if the signal is followed
by a 'p' or a 'k'. Is this same check supposed to be done for a
T packet as well (it currently is)?

--Mark