Re: [Sdcc-user] Pic led.c compile problem

[Terry P. <te...@d2...>]

On Wed, Jul 24, 2002 at 06:17:54AM -0700, Scott Dattalo wrote:
> On Wed, 24 Jul 2002, Terry Porter wrote:
> 
> > I've got a problem compiling, and was wondering
> > if anyone could please give me a pointer  ?
> > 
> > It's been a while since I did any PIC programming, and I'm rusty, especially
> > with the Trisb command.
> >  
> > The version is the latest SDCC from CVS and has Scotts most recent (ISR)
> > fix. gpsim-0.20.14, and gputils-0.10.3.
> > 
> > I must say that Gpsim has come so far since last I looked, and it
> > really rocks!
> 
> Thanks Terry. gpsim, btw for others, is the gnupic simulator. Most of the 
> "it rocks" part is due to Ralf Forsberg enhancements in the gui.

Scott is too humble, for a start I can now compile without readline
probs, and gpsim is the coolest without the GUI, otherwise how could we
poor PIC designers simulate our unworthy code :)

> 
> <snip led.c>
> 
> > 
> > This is Scott's standard example file, and its unmodified.
> > 
> > I get the following error :-
> > 
> > 
> > 
> > sh-2.05a$ make sim
> > sdcc -S -mpic14 led.c
> > printIvalChar
> > No registers saved on this pass
> > No registers saved on this pass
> > gpasm -c led.asm
> > led.asm:500:Message [302] Register in operand not in bank 0.
> > Ensure bank bits are correct.
> 
> This is no bug. It's a nuisance warning from gpasm (which in this case is 
> immitating exactly the nuisance warning you get from MPASM).
> 
> I personally ignore these. However, there is a command line switch in 
> gpasm that will allow you to warning messages. You can individually select 
> which message to ignore. I don't recall the command line option off 
> hand... 

Warnings don't bother me, as I get a lot when compiling my C code for
x86.

> 
>  > 
> > ........................... led.asm............................
> > ;  TRISB = 0;
> > 	CLRF	_TRISB                    << line 500
> > 	BCF	_STATUS,5
> 
> And there's probably a "BSF _STATUS,5" a few lines up.

So there is:-

_main	;Function start
; 2 exit points
	BSF	_STATUS,5
;#CSRC	led.c 64
;  TRISB = 0;
	CLRF	_TRISB
	BCF	_STATUS,5

> The BSF/BCF are 
> switching the register banks. Take a look at the 'f877 memory map in the 
> data sheet. You'll see that TRISB and PORTB have the same lower 7-bits in 
> their addresses. The upper two bits are selected by RP0 and RP1 in the 
> status register. You (or SDCC in this case) need to set these upper bits 
> appropriately so that the proper register is accessed! What gpasm is 
> telling you in the warning, is that the address of _TRISB (which is 0x86) 
> is greater than the 7-bits allotted in the CLRF instruction.

I thought that was the case, however gpsim wouldnt run, and like a moron
I assumed that the code wasnt compiled. Sometimes when I don't get
something I *really* don't get something.

I've just run the code in gpsim manually without using your "led"
example makefile and its working perfectly!

Thanks again Scott, I must say the SDCC PIC support has a rock solid
feel to it.

Does its support the PIC16C/F84 at all ?

Although the chip is depreciated, I've a couple here and would like to
try some test code on my r&d setup.

-- 
              Kind Regards from Terry 
My Desktop is powered by GNU/LinuX, Sorcerer kernel 2.4.17  
Free Micro burner: http://w3w.arafuraconnect.com.au/~tp/burn.html          
 ** Linux Registration Number: 103931,  http://counter.li.org **