[PATCH] breakpoint: Make sure location types match before swapping

[Keno Fischer]

Bump. It would be great to get this fixed.

On Tue, Mar 31, 2020 at 9:38 PM Keno Fischer <keno@juliacomputing.com> wrote:
>
> This patch fixes PR gdb/25741 "GDB tries to set breakpoint using Z0, but remove it using z1".
> In particular, what occurs in that case is that a hardware breakpoint is hit,
> after which GDB removes it and establishes a single step breakpoint at the
> same location. Afterwards, rather than simply removing this breakpoint and
> re-enabling the hardware breakpoint, GDB simply swaps the activation, without
> informing the server, leading to an inconsistency in GDB's view of the world
> and the server's view of the world. To remidy this situation, this
> patch adds a check that ensures two breakpoint locations have the
> same type before they are considered equal and thus eligible for silent
> swapping.
>
> gdb/ChangeLog:
>         * breakpoint.c (breakpoint_locations_match): Fix PR gdb/25741
>
> Signed-off-by: Keno Fischer <keno@juliacomputing.com>
> ---
>  gdb/breakpoint.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/gdb/breakpoint.c b/gdb/breakpoint.c
> index e49025461b..582dae1946 100644
> --- a/gdb/breakpoint.c
> +++ b/gdb/breakpoint.c
> @@ -6838,7 +6838,7 @@ breakpoint_locations_match (struct bp_location *loc1,
>      /* We compare bp_location.length in order to cover ranged breakpoints.  */
>      return (breakpoint_address_match (loc1->pspace->aspace, loc1->address,
>                                      loc2->pspace->aspace, loc2->address)
> -           && loc1->length == loc2->length);
> +           && loc1->length == loc2->length && loc1->loc_type == loc2->loc_type);
>  }
>
>  static void
> --
> 2.24.0
>