> gcc -I. -E gwlib/octstr.c | grep -C format_width
I get the exact same output as you when I issue this command.
> gcc -I. -E gwlib/octstr.c | grep -C "case 's'"
This is what I get:
break;
case 's':
s = __extension__ (*({ register char * *__ptr; if
((__builtin_classify_type(*( char * *)0) == 8) && sizeof ( char *
)
< 16) { unsigned char __fpr = ( *args )->fpr; if (__fpr < 8)
{ __ptr = (( char * *) (void *) (&(((__va_regsave_t *) (
*args )->reg_save_area)->__fp_save[ __fpr ]))) ; ( *args )->fpr =
__fpr + 1; } else if (sizeof ( char * ) == 8) {
unsigned long __addr = (unsigned long) (( *args )->overflow_arg_area
); __ptr = ( char * *)((__addr + 7) & -8); ( *args
)->overflow_arg_area = (char *)(__ptr + 1); } else {
; } } else if ((__builtin_classify_type(*( char *
*)0) >= 12) || (__builtin_classify_type(*( char * *)0) == 8) ) {
unsigned char __gpr = ( *args )->gpr; if (__gpr < 8) { __ptr = * ((
char * * *) (void *) (&(((__va_regsave_t *) ( *args
)->reg_save_area)->__gp_save[ __gpr ]))) ; ( *args )->gpr = __gpr +
1; } else { char * **__pptr = ( char * **) ((
*args )->overflow_arg_area ); __ptr = * __pptr; ( *args
)->overflow_arg_area = (char *) (__pptr + 1); } } else
{ if (sizeof ( char * ) == 8) { unsigned char __gpr = (
*args )->gpr; if (__gpr < 7) { __gpr += __gpr & 1;
__ptr = (( char * *) (void *) (&(((__va_regsave_t *) ( *args
)->reg_save_area)->__gp_save[ __gpr ]))) ; ( *args )->gpr =
__gpr + 2; } else { unsigned long __addr = (unsigned
long) (( *args )->overflow_arg_area ); __ptr = ( char * *)((__addr +
7) & -8); ( *args )->gpr = 8; ( *args )->overflow_arg_area =
(char *)(__ptr + 1); } } else if (sizeof ( char * )
== 4) { unsigned char __gpr = ( *args )->gpr; if (__gpr <
8) { __ptr = (( char * *) (void *) (&(((__va_regsave_t
*) ( *args )->reg_save_area)->__gp_save[ __gpr ]))) ( *args
)->gpr = __gpr + 1; } else { __ptr = ( char * *) (
*args )->overflow_arg_area ; ( *args )->overflow_arg_area = (char
*)(__ptr + 1); } } else { ; } }
__ptr; })) ;
if (format->has_prec && format->prec < (long) strlen(s))
A lot more than you ;o)
Jacob
-----Original Message-----
From: Richard Braakman [mailto:[EMAIL PROTECTED]]
Sent: 11. december 2001 12:58
To: [EMAIL PROTECTED]
Subject: Re: AW: Kannel on PPC
On Tue, Dec 11, 2001 at 11:47:40AM +0100, Jacob Vennervald Madsen wrote:
> I get these warnings at compile time:
>
> gwlib/octstr.c: In function `octstr_format_valist':
> gwlib/octstr.c:1985: warning: passing arg 3 of `format_width' from
> incompatible pointer type
> gwlib/octstr.c:1986: warning: passing arg 3 of `format_prec' from
> incompatible pointer type
> gwlib/octstr.c:1988: warning: passing arg 4 of `convert' from
> incompatible pointer type
Interesting... arg 3 is the va_list pointer. If va_list were a real
type
(builtin or typedefed), then this could not happen, because
format_width()
takes a "va_list *" and the caller passes the address of a va_list.
But it might be that va_list is a macro on your system, such that
writing
"va_list *" does not work as expected.
I think this command could shed some light on the problem:
gcc -I. -E gwlib/octstr.c | grep -C format_width
(run it from the main kannel directory, i.e. with gwlib as a
subdirectory).
This will expand all preprocessor macros and show the result.
On my system, it gives:
static void format_width(struct format *format, const char **fmt,
va_list *args)
{
--
++fmt;
format_flags(&format, &fmt);
format_width(&format, &fmt, &args);
format_prec(&format, &fmt, &args);
format_type(&format, &fmt);
This shows that va_list is a type of its own, and not a macro. If it
shows
something else on your system, then that would be a clue towards solving
this problem.
Oh, a similar grep might help by showing what va_arg gets expanded to:
gcc -I. -E gwlib/octstr.c | grep -C "case 's'"
shows:
break;
case 's':
s = ( *args = (__gnuc_va_list) ((char *) ( *args ) + (((sizeof
( char * ) + sizeof (int) - 1) / sizeof (int)) * sizeof (int)) ),
*(( char * *) (void *) ((char *) ( *args ) - (((sizeof ( char * ) +
sizeof (int) - 1) / sizeof (int)) * sizeof (int)) ))) ;
(I'm beginning to suspect that the solution might be as simple as
writing
"(*args)" instead of "*args" in the va_arg calls.)
By the way, I'm not really back, I'm just kibitzing and I like puzzling
out interesting C problems :)
Richard Braakman
