OpenRISC Assembly Programming

Pure assembly programming

test_asm.S

    .section    .text.startup,"ax",@progbits
    .align  4
.proc   main
    .global _start
    .global main
    .type   main, @function
_start:
main:
    l.sw        -4(r1),r1    # SI store
    l.addi      r1,r1,-4 # addsi3
	l.nop
while_loop:
	l.j			while_loop

compile or1k-elf-as test_asm.S -o test_asm.o or1k-elf-ld test_asm.o -o test_asm.out

Mixed C with assembly

function definition

First make “add.c” for test

    int add(int a, int b)
    {
        return a+b;
    }

make assembly code

    or1k-elf-gcc -O2 -S asm.c

result “add.s”

    .file   "add.c"
    .section .text
    .align  4
.proc   add
    .global add
    .type   add, @function
add:
.LFB0: # ??
    .cfi_startproc
    # Stack pointer backup
    l.sw        -4(r1),r1    # SI store
    .cfi_offset 1, -4
    # Stack pointer move to new postion
    l.addi      r1,r1,-4 # addsi3
    .cfi_def_cfa_offset 4

    # body operation
    # arguments are r3,r4,...
    # return value is r11
    l.add       r11,r3,r4 # addsi3

    # rollback the stack pointer
    l.addi  r1,r1,4
    # return
    l.jr        r9  # return_internal   # delay slot filled
    # restore stack pointer
    l.lwz       r1,-4(r1)    # SI load
    .cfi_endproc
.LFE0: # ??
    .size   add, .-add
    .ident  "GCC: (GNU) 5.3.0"

inline assembly

=r (target register), r (source register), n (number)

uint32_t check_flag(void)
{
    uint32_t flag;
    __asm__ __volatile__("l.mfspr\t\t%0,%1,0": "=r"(flag):"r"(0x0011));
    // l.mfspr   r4, 0x0011
    if(f!=((flag>>9)&0x01))
        return 0;
    else
        return 1;
}

no target register

__asm__ __volatile__("l.mtspr\t\t%0,%1,0": :"r"(0x001E), "r"(a));
// l.mtspr  0x0011, r4

constant numbers as arguments

__asm__ __volatile__("l.inst\t\t%0,%1 %2,%3 %4":"=r"(d_inst):"r"(a),"n"(1),"r"(b),"n"(1));
// l.inst r4, r1 0, r2 1