Re: [RFA] regcache.c (register_size): Check for REGISTER_RAW_SIZE_P

[Andrew Cagney <ac131313 at redhat dot com>]

> Hi,
>
> according to my confused mail on the gdb list
> (http://sources.redhat.com/ml/gdb/2003-07/msg00091.html)
>
> I think register_size() in regcache.c needs a patch.  First of all,
> it calls the very same gdb_assert() call twice.  The second reason
> is that REGISTER_RAW_SIZE() is called in this assertion unconditionally.
> This collides with the attempts to get rid of deprecated calls.
>
> A target which already removed the call is so unable to call register_size
> w/o failing in gdbarch_deprecated_register_raw_size().

I've just updated that code adding some comments (and deleting the stray 
assert).

> Therefore I suggest testing for REGISTER_RAW_SIZE_P first.  Does that
> makes sense?

I thought about that when making my change but decided to retain the 
unconditional check.
Until REGISTER_RAW_SIZE is completly eliminated it's value must always 
match that computed by the register cache.  If the architecture didn't 
supply a method then the default needs to match.

Andrew



> 	* regcache.c (register_size): Remove second call to gdb_assert.
> 	Only call assertion if REGISTER_RAW_SIZE is implemented for
> 	this target.
>
> Index: regcache.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/regcache.c,v
> retrieving revision 1.89
> diff -u -p -r1.89 regcache.c
> --- regcache.c	3 Jul 2003 14:34:29 -0000	1.89
> +++ regcache.c	9 Jul 2003 14:01:37 -0000
> @@ -290,8 +290,8 @@ register_size (struct gdbarch *gdbarch, 
>    int size;
>    gdb_assert (regnum >= 0 && regnum < (NUM_REGS + NUM_PSEUDO_REGS));
>    size = descr->sizeof_register[regnum];
> -  gdb_assert (size == REGISTER_RAW_SIZE (regnum)); /* OK */
> -  gdb_assert (size == REGISTER_RAW_SIZE (regnum)); /* OK */
> +  if (REGISTER_RAW_SIZE_P ())
> +    gdb_assert (size == REGISTER_RAW_SIZE (regnum)); /* OK */
>    return size;
>  }