It's great to see that there are still some working on OnBoardC. Being an old
school C programmer I thought about refactoring VFSDos, because it seems to
have many flaws. However, time's limited... :-(
> The inline notation itself is legal (changed that already) however, it's
> the assembly commands themselves which I am worried about: are those legal
> syntax to pass to m68k-palmos-as?
It helps a lot to let the GCC compiler generate assembler source code (option
"-S") and to study that.
>From my experiences writing Phoinix (phoinix.sf.net) I'd recommend to try:
__asm__("link %a6,#0\n\t"
"move.l %a6,-(%a7) ; push a6\n\t"
"jbsr setA6_00(%pc) ; no need of add #4,a7\n\t"
"unlk %a6 ; restore stack\n\t"
"rts\n\t");
and
__asm__("jbsr getA6_00(%pc) ; get value in d0\n\t"
"move.l %d0,%a6 ; restore a6\n\t"
"move.w 4(%a7),%d0 ; return value\n\t"
"unlk %a6 ; restore stack \n\t"
"rts\n\t");
respectively. You may also replace "%a7" by "%sp", or "%a6" by "%fp", if you
like it better. The pseudo-op "jbsr" uses "bsr.s", "bsr.w", or "jsr",
whichever is appropriate.
Anyway, I found the complete source, and it looks better to me to use these
lines, provided that your A5 point to the globals. It saves some calls and
returns.
void Tag(void)
{
__asm__("link %a6,#0 ; set up %a6 for exit()\n\t"
"move.l %a6,[email protected](%a5) ; store %a6 in global memA6\n\t"
"unlk %a6 ; restore stack\n\t"
"rts\n\t");
}
void exit(int code)
{
__asm__("move.l [email protected](%a5),%a6 ; restore %a6 from global memA6\n\t"
"move.w 4(%a7),%d0 ; return value\n\t"
"unlk %a6 ; restore stack\n\t"
"rts\n\t");
}
> I am asking for documentation on SysLib* calls because GCC complains about
> this call:
> err=SysLibClose(StdioLibRef,&usecount); // too many arguments to
> SysLibClose
Well, the header file contains a prototype with the libref only...
HTH, Bodo
--
For information on using the ACCESS Developer Forums, or to unsubscribe, please
see http://www.access-company.com/developers/forums/
