Re: [patch/resend/rfa] (1/4) Cleanup static function usage in hppa-tdep.c

[Andrew Cagney <cagney@gnu.org>]

> 2004-04-13  Randolph Chung  <tausq@debian.org>
> 
> 	* hppa-tdep.c (hppa_reg_struct_has_addr, hppa_skip_prologue,
> 	hppa_skip_trampoline_code, hppa_in_solib_call_trampoline,
> 	hppa_in_solib_return_trampoline, hppa_cannot_store_register,
> 	hppa_smash_text_address, hppa_target_read_pc, hppa_target_write_pc):
> 	Remove forward declaration and make static.
> 	(hppa_reg_struct_has_addr): Rename to hppa_use_struct_convention.
> 	(hppa_gdbarch_init): Set use_struct_convention.
> 
> Index: hppa-tdep.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/hppa-tdep.c,v
> retrieving revision 1.143
> diff -u -p -r1.143 hppa-tdep.c
> --- hppa-tdep.c	11 Apr 2004 04:20:51 -0000	1.143
> +++ hppa-tdep.c	14 Apr 2004 05:01:21 -0000
> @@ -121,17 +123,8 @@ static void internalize_unwinds (struct 
>  static void record_text_segment_lowaddr (bfd *, asection *, void *);
>  /* FIXME: brobecker 2002-11-07: We will likely be able to make the
>     following functions static, once we hppa is partially multiarched.  */
> -int hppa_reg_struct_has_addr (int gcc_p, struct type *type);
> -CORE_ADDR hppa_skip_prologue (CORE_ADDR pc);
> -CORE_ADDR hppa_skip_trampoline_code (CORE_ADDR pc);
> -int hppa_in_solib_call_trampoline (CORE_ADDR pc, char *name);
> -int hppa_in_solib_return_trampoline (CORE_ADDR pc, char *name);
>  int hppa_pc_requires_run_before_use (CORE_ADDR pc);
>  int hppa_instruction_nullified (void);
> -int hppa_cannot_store_register (int regnum);
> -CORE_ADDR hppa_smash_text_address (CORE_ADDR addr);
> -CORE_ADDR hppa_target_read_pc (ptid_t ptid);
> -void hppa_target_write_pc (CORE_ADDR v, ptid_t ptid);

Yes, definitly, avoid the duplication (these date back to pre ISO-C days).

Just not this:

> -int
> -hppa_reg_struct_has_addr (int gcc_p, struct type *type)
> +static int
> +hppa_use_struct_convention (int gcc_p, struct type *type)

> +  set_gdbarch_use_struct_convention (gdbarch, hppa_use_struct_convention);
> +

use_struct_convention shouldn't be needed -> "struct_return" provides a 
superset of its functionality.  I suspect that the function 
reg_struct_has_addr can instead simply be removed?

Minus use_struct_convention, it's ok to commit,

Andrew