"Martin M. Hunt" wrote:
> 
> The clear command improperly detects overlays and fails
> to clear breakpoints if overlays are not disabled.
> 
> Tested with linux-x-mips (overlays enabled) and linux x86 native.

Martin, you really made me think with this one.  Sorry it took so long.
I had to go back eleven years in the code base to understand what this
code was trying to do -- which made me realize that it's painfully
obsolete.  It has two inner loops with identical control conditions
(except that they've gotten out of sync), just because they didn't
have ALL_BREAKPOINTS_SAFE when this code was written.

So I rewrote the whole damn function.   ;-)

If you'd like to check the attached, to make sure that it 
preserves the intent of your change?  I take it your
intentions were:

  1) Make sure that the two inner loops have the same
     control conditions.  Instead of that, I combined them.
     Having two was just bad, it allowed them to get 
     out of sync in the first place.
  2) Allowing an overlay breakpoint to be cleared
     if overlay debugging is disabled.  I modified 
     that part of your change slightly.