Re: [RFC] Catch exception after stepped over watchpoint.

[Yao Qi <yao at codesourcery dot com>]

On 07/05/2013 10:52 PM, Pedro Alves wrote:
> What's different between the insert_breakpoints calls that makes
> the first one after setting the watchpoints just warn and still the
> program is continued, while this one errors out?

Pedro,
I am not sure I fully understand your question.  In the original code, 
the error is thrown if GDB failed to insert watchpoints in 
insert_breakpoints.  With the patch, we catch the error, just warn, and 
let the GDB continue to process the "watchpoint hit" event.  Your 
question is "what is the difference between the original code and the 
patched code".

If that is what you meant, the difference are
 1) the state on software single step is cleared, so that the internal 
error I mentioned is fixed,
 2) the message on watched value changed is still shown,

(gdb) c
Continuing.
Warning:
Could not insert hardware watchpoint 1.
Could not insert hardware breakpoints:
You may have requested too many hardware breakpoints/watchpoints.

[1]Hardware watchpoint 1: *global_ptr

Old value = <unreadable>
New value = 3 '\003'
func4 () at 
/scratch/yqi/mips-linux/src/gdb-2013.05/gdb/testsuite/gdb.base/watchpoint.c:135
135       buf[0] = 7;

if we stop stepping and return, we can't get the message after [1].

--
Yao (éå)