Microchip PIC32 advice

This chip is supported by the MPLAB IDE and the MPLAB SIMulator.
These comments apply to version 8.73, Windows Version

A suitable mirdef.h file would be 

#define MIRACL 32
#define MR_LITTLE_ENDIAN 
#define mr_utype int                           
#define mr_unsign32 unsigned int
#define MR_IBITS 32    
#define MR_LBITS 32   
#define mr_dltype long long 
#define mr_unsign64 unsigned long long
#define MR_NOASM
#define MAXBASE ((mr_small)1<<(MIRACL-1))


To get started, first build the miracl library from the command line
by executing this batch file

copy mirdef.pic mirdef.h
pic32-gcc -c -O2 -g mrcore.c
pic32-gcc -c -O2 -g mrarth0.c
pic32-gcc -c -O2 -g mrarth1.c
pic32-gcc -c -O2 -g mrarth2.c
pic32-gcc -c -O2 -g mralloc.c
pic32-gcc -c -O2 -g mrsmall.c
pic32-gcc -c -O2 -g mrio1.c
pic32-gcc -c -O2 -g mrio2.c
pic32-gcc -c -O2 -g mrgcd.c
pic32-gcc -c -O2 -g mrjack.c
pic32-gcc -c -O2 -g mrxgcd.c
pic32-gcc -c -O2 -g mrarth3.c
pic32-gcc -c -O2 -g mrbits.c
pic32-gcc -c -O2 -g mrrand.c
pic32-gcc -c -O2 -g mrprime.c
pic32-gcc -c -O2 -g mrcrt.c
pic32-gcc -c -O2 -g mrscrt.c
pic32-gcc -c -O2 -g mrmonty.c
pic32-gcc -c -O2 -g mrpower.c
pic32-gcc -c -O2 -g mrsroot.c
pic32-gcc -c -O2 -g mrcurve.c
pic32-gcc -c -O2 -g mrfast.c
pic32-gcc -c -O2 -g mrlucas.c
pic32-gcc -c -O2 -g mrzzn2.c
pic32-gcc -c -O2 -g mrzzn2b.c
pic32-gcc -c -O2 -g mrzzn3.c
pic32-gcc -c -O2 -g mrshs.c
pic32-gcc -c -O2 -g mrshs256.c
pic32-gcc -c -O2 -g mrshs512.c
pic32-gcc -c -O2 -g mraes.c
pic32-gcc -c -O2 -g mrgcm.c
pic32-gcc -c -O2 -g mrstrong.c
pic32-gcc -c -O2 -g mrbrick.c
pic32-gcc -c -O2 -g mrebrick.c
pic32-gcc -c -O2 -g mrgf2m.c
pic32-gcc -c -O2 -g mrec2m.c
pic32-gcc -c -O2 -g mrecn2.c
rem copy mrmuldv.ccc mrmuldv.c
rem pic32-gcc -c -O2 -g mrmuldv.c

rem
rem Create library 'miracl.lib'
del miracl.a
pic32-ar rc miracl.a mrcore.o mrarth0.o mrarth1.o mrarth2.o mralloc.o mrsmall.o mrgcm.o
pic32-ar r  miracl.a mrio1.o mrio2.o mrjack.o mrgcd.o mrxgcd.o mrarth3.o mrbits.o mrzzn2.o mrzzn3.o
pic32-ar r  miracl.a mrrand.o mrprime.o mrcrt.o mrscrt.o mrmonty.o mrcurve.o mrpower.o mrsroot.o
pic32-ar r  miracl.a mrfast.o mrshs.o mraes.o mrlucas.o mrstrong.o mrbrick.o mrecn2.o
pic32-ar r  miracl.a mrshs256.o mrshs512.o mrebrick.o mrgf2m.o mrec2m.o mrzzn2b.o
rem pic32-ar r  miracl.a mrmuldv.o

del mr*.o

Then fire up the MPLAB IDE, and create a project called demo. 
Choose one of the most powerful members of the PIC32 families to start
with, for example PIC32MX795F512L

Copy pk-demo.c to demo.c and add this to the project. 
Add the Headers miracl.h and mirdef.h
Add the library file miracl.a

Edit demo.c to include at the start

#include <p32xxxx.h> 
#include <stdio.h>  

void _mon_putc (char c) 
{   
	while (U1STAbits.UTXBF);   
	U1TXREG = c; 
} 

At the very start of the main() program add

	U1MODEbits.UARTEN = 0x01;   
	U1STAbits.UTXEN = 0x01;  
 
This magic is required to redirect printf() to UART1

Click on Debugger->Settings and the UART1 IO Tab. Enable UART1 IO
and redirect the output to a window.

Click on Project->Build options->project and the Linker Tab.
Create a heap of 65536 bytes and a stack of 1024 bytes

Build and Run the program. You should see the output in the Output Window, 
SIM Uart1 Tab

The PIC32 uses a MIPS core. This architecture does not support a carry flag
and so Assembly language implementation will not be much faster than 
C generated code.

To run the application ecdhp32.c, add the same "magic" as before
to redirect printf to UART1.

Use as mirdef.h

#define MIRACL 32
#define MR_LITTLE_ENDIAN 
#define mr_utype int                          
#define mr_unsign32 unsigned int
#define MR_IBITS      32    
#define MR_LBITS      32  
#define mr_dltype long long
#define mr_unsign64 unsigned long long
#define MAXBASE ((mr_small)1<<(MIRACL-1))
#define MR_COMBA 5
#define MR_STATIC 5
#define MR_ALWAYS_BINARY
#define MR_STRIPPED_DOWN
#define MR_GENERIC_MT
#define MR_PSEUDO_MERSENNE
#define MR_SPECIAL
#define MR_NOASM

And build the library from the command line

mex 5 c mrcomba
pic32-gcc -c -g mrcore.c
pic32-gcc -c -g mrarth0.c
pic32-gcc -c -g mrarth1.c
pic32-gcc -c -g mrarth2.c
pic32-gcc -c -g mrsmall.c
pic32-gcc -c -g mrio1.c
pic32-gcc -c -g mrjack.c
pic32-gcc -c -g mrbits.c
pic32-gcc -c -g mrxgcd.c
pic32-gcc -c -g mrmonty.c
pic32-gcc -c -g mrsroot.c
pic32-gcc -c -g mrcurve.c
pic32-gcc -c -g mrlucas.c
pic32-gcc -c -g mrebrick.c
pic32-gcc -c -g mrcomba.c

pic32-ar -rc miracl.a mrcore.o mrarth0.o mrarth1.o mrarth2.o mrsmall.o
pic32-ar -r  miracl.a mrio1.o mrjack.o mrxgcd.o 
pic32-ar -r  miracl.a mrmonty.o mrcurve.o 
pic32-ar -r  miracl.a mrebrick.o mrsroot.o mrlucas.o
pic32-ar -r  miracl.a mrbits.o mrcomba.o
del mr*.o

This time no heap is required.

Note that -O2 optimization causes a problem. However introducing
each individual optimization from 

http://gcc.gnu.org/onlinedocs/gcc-3.4.4/gcc/Optimize-Options.html

works fine. 

Note that -O2 optimization works OK with MicroChip's newer MPLAB-X tool, 
which uses a more up-to-date version 4.5 of the GCC compiler.