Re: [RFC] Insight build problem: was [06/15] Per-breakpoint architecture support

["Ulrich Weigand" <uweigand@de.ibm.com>]

Pierre Muller wrote:

>    This seems harmless as the function is not in any header, but
> this function is used in gdbtk sources.
>    I don't really know if gdbtk failures qualify for patches
> as such, but it seems to me that it would be difficult to 
> change the gdbtk-bp.c code to avoid using this function.

Ah, I must have thought the missing static was just an oversight.
(And the overlooked the change when putting together the ChangeLog
entry, sorry.)

>   Is this failure a sufficient reason to accept this patch 
> or should we go another route (maybe add set_raw_breakpoint
> to breakpoint.h header ?).

It's not particularly nice, but I'm OK with leaving it the way
it was for now ...   Longer term we should find out why the
exported interfaces like set_breakpoint are not suitable for gdbtk,
and extend them if necessary.

> @@ -4396,8 +4397,9 @@ static void free_bp_location (struct bp_
>  
>  /* Helper to set_raw_breakpoint below.  Creates a breakpoint
>     that has type BPTYPE and has no locations as yet.  */
> +/* This function is used in gdbtk sources and thus can not be made static.
> */
>  
> -static struct breakpoint *
> +struct breakpoint *
>  set_raw_breakpoint_without_location (struct gdbarch *gdbarch,
>  				     enum bptype bptype)
>  {

This chunk seems unnecessary; set_raw_breakpoint_without_location
was static before my patch as well.

The rest is OK.

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com