Small Device C Compiler suite

On Mon, 29 Jul 2002 pagliar@... wrote:

> Hi all,
> it seems sdcc is doing wrong bank selections during sfr bit operation:
> I assume this is the right syntax... if not, please suggest.
> 
> sdcc version is:
> SDCC : mcs51/gbz80/z80/avr/ds390/pic14/TININative/xa51 2.3.2 (Jul 25 2002)
> (CYGWIN)

That's another bug with bank selecting. I just have gotten three bug 
reports with respect to selecting banks. (There are actually two bugs in 
the code you sent; the other is that the bank fix-up occurs after the 
return).

Looks like I'll be working on bank selection stuff again...

Thanks for the report,
Scott

On Mon, 29 Jul 2002, Scott Dattalo wrote:

> That's another bug with bank selecting. I just have gotten three bug
> reports with respect to selecting banks. (There are actually two bugs in
> the code you sent; the other is that the bank fix-up occurs after the
> return).

I've got the latest CVS as of noon PST on the 29th, and I don't see any
bank selection going on at all.  My C code is very basic:

#define _16F877
#include "p16f877.h"
void main(void) {
  PORTD = 0;
  TRISD = 0;
  PORTD = 0x0f;
}

which produces..:   (sdcc -mpic14 -p16f877)

. . .
_main   ;Function start
; 2 exit points
;#CSRC  test.c 5
;  PORTD = 0;
        CLRF    _PORTD
;#CSRC  test.c 6
;  TRISD = 0;
        CLRF    _TRISD
;#CSRC  test.c 7
;  PORTD = 0x0f;
        MOVLW   0x0f
        MOVWF   _PORTD
        RETURN
; exit point of _main
        end

There's no bank selection at all going on.  Am I missing something in the
code or commandline?

      Erik Walthinsen <omega@...> - System Administrator
        __
       /  \                GStreamer - The only way to stream!
      |    | M E G A        ***** http://gstreamer.net/ *****
      _\  /_

On Mon, 29 Jul 2002, Erik Walthinsen wrote:

> On Mon, 29 Jul 2002, Scott Dattalo wrote:
> 
> > That's another bug with bank selecting. I just have gotten three bug
> > reports with respect to selecting banks. (There are actually two bugs in
> > the code you sent; the other is that the bank fix-up occurs after the
> > return).
> 
> I've got the latest CVS as of noon PST on the 29th, and I don't see any
> bank selection going on at all.  My C code is very basic:
> 
> #define _16F877
> #include "p16f877.h"
> void main(void) {
>   PORTD = 0;
>   TRISD = 0;
>   PORTD = 0x0f;
> }
> 
> which produces..:   (sdcc -mpic14 -p16f877)


I tried this with the copy in CVS (~5:30 A.M.  PST ) and didn't see this. 
I don't recall seeing this bug with older versions. Please grab the latest 
from CVS and see if the problem still exists. If it does, then you're 
include file needs to be updated. You can try this one if you wish:

http://www.dattalo.com/p16f877.h

BTW, the include file has several SDCC specific pragmas that define the 
processor memory map.

Scott

PS. and yes, I live in the PST time zone :)

On Tue, 30 Jul 2002, Scott Dattalo wrote:

> http://www.dattalo.com/p16f877.h
That solved it.  I'll see if I can re-generate the header from the .inc
file, and if it still does the same thing, I'll send a copy to you.  I
have MPLAB 5.50, if that's relevant.

> PS. and yes, I live in the PST time zone :)
How convenient ;-)

      Erik Walthinsen <omega@...> - System Administrator
        __
       /  \                GStreamer - The only way to stream!
      |    | M E G A        ***** http://gstreamer.net/ *****
      _\  /_