AN669 Datasheet, PDF(1/7 Page) Microchip Technology – Embedding Assembly Routines into C Language Using a Floating Point Routine as an Example

English

English German Russian Spanish Italian Polish Chinese Japanese Korean French Portuguese	Language :

AN669 Datasheet, PDF (1/7 Pages) Microchip Technology – Embedding Assembly Routines into C Language Using a Floating Point Routine as an Example

AN669

Embedding Assembly Routines into C Language Using a Floating Point

Routine as an Example

Authors: Rick Evans

Richard Fischer

Microchip Technology, Inc.

INTRODUCTION

With the advent of MPLAB-C, the Microchip C-com-

piler, many PICmicroâ¢ users need to embed existing

assembly language routines and/or Microchip applica-

tion notes into C. This application note explains how to

embed an assembly language program into MPLAB-C,

version 1.10, and the issues therein. For example,

embedding interrupt save and restore must be done

using assembly language. Also, critical timing routines

may require assembly. The 32-bit ï¬oating point multiply

routine from AN575 is used to illustrate this process.

The remaining 32-bit ï¬oating point math routines are

embedded into individual C functions and are included

in the ï¬le accompanying this application note.

PROCEDURE

For this example, weâll use a PIC16C74A with 4K

Program Memory, and 192 bytes of RAM.

Embedding assembly routines

In order to embed an assembly language routine in C

code place the #asm and #endasm directives around

the assembly routine. Furthermore, if this is a subrou-

tine, as is the case with the ï¬oating point multiply, then

embed the assembly code within a C function declara-

tion. The #asm construct is illustrated in Example 1 with

an excerpt from the 32-bit ï¬oating point routine.

EXAMPLE 1: #ASM, #ENDASM

CONSTRUCT

void fpm32(void)

{

#asm

FPM32

MOVF

BTFSS

MOVF

BTFSC

GOTO

AEXP,W

BEXP,W

RES032M

;test for zero

;arguements

M32BNE0

MOVF

XORWF

MOVWF

MOVF

ADDWF

MOVLW

AARGB0,W

BARGB0,W

SIGN

BEXP,W

EXP, F

EXPBIAS-1

;save sign

;in SIGN

;...etc.

#endasm

}

Locating the Routine in Program Memory,

GOTOs and CALLs

There are two 2K word pages of program memory in

the PIC16C74A. Program memory 000h to 7FFh is

page 0, 800h to FFFh is page 1. By making fpm32() a

C function, MPLAB-C initializes the appropriate page

bit in the PCLATH register before the subroutine call is

made. (See data sheet for more on PCLATH).

A potential problem could arise, however, if the new C

function, fpm32(), crosses the page boundary

(7FFh,800h). MPLAB-C does not insert code into the

assembly code to initialize the page bits (remember

MPLAB-C does take care of paging for function calls).

That means it is up to the programmer to either; 1) add

assembly language to initialize PCLATH appropriately,

or 2) move the entire #asm function within a single

page. Option 1 involves more work. The programmer

must ï¬rst compile the C code, then analyze the listing

ï¬le to see if the assembly function crossed a page

boundary. Finally, add the appropriate assembly lan-

guage to initialize PCLATH then re-compile. This solu-

tion is not desirable since every time new C code is

added to or deleted from the program, the routine,

fpm32() can potentially move across the page bound-

ary. Option 2 is the simplest solution - to locate the C

function in a single page.

DS00669A-page 1

▷