'MPASM bug ?'
|Hi all ...
I'm just starting out with the PIC chip so if this is a silly question,
scratch it up to the learning curve :)
I'm writing a small routine to test the serial line connection (232c
using a max232 and the usart on the 16c74a) but Mpasm keeps giving me
the 'incorrect bank' error message (#302). Here's the code chunk -
anyone know if this is my code or mpasm ?
; Serial line test program
#define ClkFreq 20000000
#define baud(X) ((10*ClkFreq/(64*X))+5)/10 -1
#define TXSTA_INIT 0xA0
#define RCSTA_INIT 0x90
RESET call Setup_Async_Mode
movwf SPBRG ;MPASM gives error #302 here ...
movwf TXSTA ;... and here
PollTxmt btfss PIR1,4
Mark Dennehy, B.A., B.A.I. Email : tcd.iemdennehy
Computer Vision and Robotics Research Group,
Computer Science Dept., Trinity College Dublin
|Mark Dennehy <MITVMA.MIT.EDU> wrote: PICLIST
> I'm just starting out with the PIC chip so if this is a silly
> question, scratch it up to the learning curve :)
It's not a silly question, Mark... It gets asked all the time.
Strictly speaking, the error's in your code... Although this
section of your program will assemble correctly and work as you
Here's what's happening:
SPBRG and TXSTA are, as you know, located on register-page 1;
the PIC16C74A.INC file equates them to 0x99 and 0x98,
The numbers 0x98 and 0x99 are 8 bits wide, but if you look at
your PIC16C74 data shet, you'll see that the MOVWF instruction
only has room for a SEVEN-bit register number.
This, of course, is why you need to set the RP0 bit before
accessing registers on page 1, and clear it before accessing
registers on page 0; RP0 holds the eighth bit of the register
When you write "MOVWF SPBRG" (or "MOVWF 0x99"), MPASM notices
that you're trying to force an eight-bit register number into a
space only large enough to hold seven bits, so it builds the
instruction using only the low seven bits of the register number
(0x19, in this case) and generates Warning #302 to inform you.
Your code will assemble and work fine, since you really only WANT
the "MOVWF" insruction to hold the low seven bits of the register
number, but if you really want to do it RIGHT, you should change
your "MOVWF SPBRG" and "MOVWF TXSTA" instructions to:
MOVWF SPBRG^080H ; The "^080H" masks off the high bit
; of the register number.
MOVWF TXSTA^080H ; The "^080H" masks off the high bit
; of the register number.
Of course, you can also just turn off the warnings by putting an
"ERRORLEVEL -302" line near the start of your program.
=== Andrew Warren - ix.netcom.comfastfwd
=== Fast Forward Engineering - Vista, California
=== For PICLIST help (including "unsubscribe" instructions),
=== put the single word "help" in the body of a message and
=== send it to: mitvma.mit.edulistserv
Check you .lst file. It probably says something like:
Message: Register in operand not in bank 0. Ensure that bank bits
My program is full of these things and it gets to be a little
annoying. If you check the code generated, you will probably see that it
made the right numbers. It just spits out this warning to highlight
places where you could make a mistake (I saw in your listing that you did
If you get your serial I/O working, I would be VERY interested in
seeing your code. I wrote an interrupt based routine that does not
receive. For a complete listing of my code, see the archive. You can
search by name (Fred Thompson), or by date (Wed 13 Aug, 1997).
|Andrew Warren wrote:
> Mark Dennehy <MITVMA.MIT.EDU> wrote: PICLIST
> > I'm just starting out with the PIC chip so if this is a silly
> > question, scratch it up to the learning curve :)
> It's not a silly question, Mark... It gets asked all the time.
Thanks, Andy, for clarifying this one.... again. :-)
On this same topic, there is are two reasons for defining SFR's and
variables with values greater than 7 bits.
1. You can use variable names while debugging MPLAB in the Modify
window or in the Execute Opcode dialog.
2. You can use the new MPASM directives described below.
In the latest beta version of MPASM (v1.99.xx) there are new
directives, one called BANKSEL. Originally designed for use with the
Linker and relocatable code only, it also now works for
non-relocatable output code from MPASM.
This allows you to do the bank switching to access an SFR or
RAM variable without actually having to toggle the specific bank
bits (there's also BANKISEL for indirect addressing, and PAGESEL
for ROM paging).
When making relocatable code the addresses of RAM variable may not be
known until the Linker is invoked, there needs to be a way to set up
the proper RAM bank page. This is done like this:
The Linker will generate the proper bit-banging bank instructions for
processor currently set with the MPASM P directive... So if you move
your code from a processor that has two RAM banks to one that has
four, or even if you move it from a 16C54 to a 16C84, the proper
banking code will be generated.
If you are not generating relocatable code (using the /o option with
MPASM), then MPASM will expand the BANKSEL directive to the proper
instructions. You may want to use
to suppress those MPASM messages that are distracting, or as Andy also
suggested, AND the operand to strip off the extra bits.
At 12:14 PM 8/20/97 +0100, you wrote:
Message 302 is not an error message. It is a warning. If you have
correctly set RP0, ignore it. Or you can disable it by putting:
in the beginning of your source file.
Lynn Richardson - Design Eng.|WA0ZNL |Progress Instruments, Inc.
DC - 1GHz, RX, TX 100W, PLL |WA0ZNL.AMPR.ORG |807 NW Commerce Drive
ASM 6805, 8051, Z8, PIC |220.127.116.11 |Lee's Summit, MO 64086
C |proginst.com|P(816)524-4442 F 246-4556 lrich
More... (looser matching)
- Last day of these posts
- In 1997
, 1998 only
- New search...