Re: [PATCH] Use CORE_ADDR as return type from x86_dr_low_get_addr

[Simon Marchi via Gdb-patches <gdb-patches@sourceware.org>]

On 2021-09-03 12:32 p.m., Tom Tromey wrote:
> On a Windows build locally, watchpoints started failing.  I tracked
> this down to x86_dr_low_get_addr returning an 'unsigned long'... in
> this particular build, this is a 32-bit type, but the inferior is a
> 64-bit program.
> 
> This patch fixes the problem by changing the return type.  No other
> change is really required, because this matches the function pointer
> in struct x86_dr_low_type.
> 
> However, I found a spot in windows-nat.c that is a bit clearer if it
> uses CORE_ADDR as well, so I've also made this change.
> ---
>  gdb/nat/x86-dregs.c | 2 +-
>  gdb/windows-nat.c   | 4 ++--
>  2 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/gdb/nat/x86-dregs.c b/gdb/nat/x86-dregs.c
> index cf8e517eb0d..24751647005 100644
> --- a/gdb/nat/x86-dregs.c
> +++ b/gdb/nat/x86-dregs.c
> @@ -52,7 +52,7 @@ x86_dr_low_set_addr (struct x86_debug_reg_state *new_state, int i)
>  
>  /* Return the inferior's debug register REGNUM.  */
>  
> -static unsigned long
> +static CORE_ADDR
>  x86_dr_low_get_addr (int i)
>  {

This bit looks good to me.

>    return x86_dr_low.get_addr (i);
> diff --git a/gdb/windows-nat.c b/gdb/windows-nat.c
> index 0c2d55ef67b..b45de3cab7a 100644
> --- a/gdb/windows-nat.c
> +++ b/gdb/windows-nat.c
> @@ -122,7 +122,7 @@ enum
>  	| CONTEXT_SEGMENTS | CONTEXT_DEBUG_REGISTERS \
>  	| CONTEXT_EXTENDED_REGISTERS
>  
> -static uintptr_t dr[8];
> +static CORE_ADDR dr[8];

I don't really mind the change, but I don't know if that's really any
more clear, because not all DR registers contain addresses, which is
what using CORE_ADDR suggests.  I guess the point of using uintptr_t is
that they are 32 bits when GDB is built as 32-bit (and it can't debug
64-bit programs then) and 64 bits when GDB is built as 64-bit (and the
top 32 bits are unused when debugging a 32-bit program).  What prompted
you to change this part?

Simon