view arm7dis/thumbdis.c @ 389:623316d1ece7

compal/iMelody: capturing initial observations
author Mychaela Falconia <falcon@freecalypso.org>
date Thu, 31 Mar 2022 20:56:56 +0000
parents c883e60df239
children
line wrap: on
line source

/*
 * Lean and mean ARM7TDMI disassembler
 * Written by Spacefalcon the Outlaw
 */

#include <sys/types.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <strings.h>

extern char *binfilename;
extern u_char *filemap;
extern unsigned disasm_len, base_vma;

extern unsigned get_u16(), get_u32();

extern char *regnames[16], *condition_decode[16], *shift_types[4];

static void
format_1_2(word)
	unsigned word;
{
	unsigned op, imm;

	op = (word >> 11) & 3;
	if (op != 3) {
		/* format 1 */
		imm = (word >> 6) & 0x1F;
		if (op != 0 && imm == 0)
			imm = 32;
		printf("%s\t%s, %s, #%u\n", shift_types[op], regnames[word&7],
			regnames[(word>>3)&7], imm);
		return;
	}
	/* format 2 */
	printf("%s\t%s, %s, ", word&0x200 ? "sub" : "add", regnames[word&7],
		regnames[(word>>3)&7]);
	if (word & 0x400)
		printf("#%u\n", (word >> 6) & 7);
	else
		printf("%s\n", regnames[(word >> 6) & 7]);
}

static void
format_3(word)
	unsigned word;
{
	static char *opctab[4] = {"mov", "cmp", "add", "sub"};
	unsigned imm;

	imm = word & 0xFF;
	printf("%s\t%s, #%u", opctab[(word>>11)&3], regnames[(word>>8)&7], imm);
	if (imm > 9)
		printf("\t; 0x%x", imm);
	putchar('\n');
}

static void
format_4(word)
	unsigned word;
{
	static char *opc[16] = {"and", "eor", "lsl", "lsr",
				"asr", "adc", "sbc", "ror",
				"tst", "neg", "cmp", "cmn",
				"orr", "mul", "bic", "mvn"};

	printf("%s\t%s, %s\n", opc[(word>>6)&0xF], regnames[word&7],
		regnames[(word>>3)&7]);
}

static void
format_5_bx(word)
	unsigned word;
{
	if (word & 0x80)
		printf("<invalid: blx instead of bx>\n");
	else
		printf("bx\t%s\n", regnames[(word>>3)&0xF]);
}

static void
format_5_hiops(word)
	unsigned word;
{
	static char *opctab[3] = {"add", "cmp", "mov"};
	int reg1, reg2, op;

	if (word & 0xC0) {
		reg1 = word & 7;
		if (word & 0x80)
			reg1 += 8;
		reg2 = (word >> 3) & 0xF;
		op = (word >> 8) & 3;
		if (op == 2 && reg1 == reg2 && reg1 != 15)
			printf("nop\t\t\t(mov %s, %s)\n",
				regnames[reg1], regnames[reg2]);
		else
			printf("%s\t%s, %s\n", opctab[op],
				regnames[reg1], regnames[reg2]);
	} else
		printf("<invalid: hi-reg format with both low regs>\n");
}

static void
format_5(word)
	unsigned word;
{
	if ((word & 0x300) == 0x300)
		format_5_bx(word);
	else
		format_5_hiops(word);
}

static void
format_6(off, word)
	unsigned off, word;
{
	unsigned loff, litoff;

	loff = (word & 0xFF) << 2;
	off &= ~3;
	off += 4;
	litoff = off + loff;
	if (litoff+4 <= disasm_len)
		printf("ldr\t%s, =0x%x\t; via 0x%x\n", regnames[(word>>8)&7],
			get_u32(filemap + litoff), base_vma + litoff);
	else
		printf("ldr\t%s, [pc, #%u]\t(0x%x)\n", regnames[(word>>8)&7],
			loff, base_vma + litoff);
}

static void
format_7(word)
	unsigned word;
{
	printf("%s%s\t%s, [%s, %s]\n", word&0x800 ? "ldr" : "str",
		word&0x400 ? "b" : "", regnames[word&7],
		regnames[(word>>3)&7], regnames[(word>>6)&7]);
}

static void
format_8(word)
	unsigned word;
{
	static char *opc[4] = {"strh", "ldrsb", "ldrh", "ldrsh"};

	printf("%s\t%s, [%s, %s]\n", opc[(word>>10)&3], regnames[word&7],
		regnames[(word>>3)&7], regnames[(word>>6)&7]);
}

static void
format_9(word)
	unsigned word;
{
	unsigned loff;

	loff = (word >> 6) & 0x1F;
	if (!(word & 0x1000))
		loff <<= 2;
	printf("%s%s\t%s, [%s, #%u]", word&0x800 ? "ldr" : "str",
		word&0x1000 ? "b" : "", regnames[word&7],
		regnames[(word>>3)&7], loff);
	if (loff >= 10)
		printf("\t; 0x%x", loff);
	putchar('\n');
}

static void
format_10(word)
	unsigned word;
{
	unsigned loff;

	loff = (word >> 6) & 0x1F;
		loff <<= 1;
	printf("%sh\t%s, [%s, #%u]", word&0x800 ? "ldr" : "str",
		regnames[word&7], regnames[(word>>3)&7], loff);
	if (loff >= 10)
		printf("\t; 0x%x", loff);
	putchar('\n');
}

static void
format_11(word)
	unsigned word;
{
	unsigned loff;

	loff = (word & 0xFF) << 2;
	printf("%s\t%s, [sp, #%u]", word&0x800 ? "ldr" : "str",
		regnames[(word>>8)&7], loff);
	if (loff >= 10)
		printf("\t; 0x%x", loff);
	putchar('\n');
}

static void
format_12(off, word)
	unsigned off, word;
{
	unsigned loff;

	loff = (word & 0xFF) << 2;
	printf("add\t%s, %s, #%u", regnames[(word>>8)&7],
		word&0x800 ? "sp" : "pc", loff);
	if (loff >= 10)
		printf("\t; 0x%x", loff);
	putchar('\n');
}

static void
format_13(word)
	unsigned word;
{
	unsigned loff;

	if ((word & 0xFF00) != 0xB000) {
		printf("<invalid format 13>\n");
		return;
	}
	loff = (word & 0x7F) << 2;
	printf("%s\tsp, #%u", word&0x80 ? "sub" : "add", loff);
	if (loff >= 10)
		printf("\t; 0x%x", loff);
	putchar('\n');
}

static void
format_14(word)
	unsigned word;
{
	int r, flag;

	if ((word & 0xF600) != 0xB400) {
		printf("<invalid format 14>\n");
		return;
	}
	printf("%s\t{", word&0x800 ? "pop" : "push");
	flag = 0;
	for (r = 0; r < 9; r++)
		if (word & (1 << r)) {
			if (flag)
				fputs(", ", stdout);
			if (r == 8)
				fputs(word&0x800 ? "pc" : "lr", stdout);
			else
				fputs(regnames[r], stdout);
			flag = 1;
		}
	putchar('}');
	putchar('\n');
}

static void
format_15(word)
	unsigned word;
{
	int r, flag;

	printf("%sia\t%s!, {", word&0x800 ? "ldm" : "stm",
		regnames[(word>>8)&7]);
	flag = 0;
	for (r = 0; r < 8; r++)
		if (word & (1 << r)) {
			if (flag)
				fputs(", ", stdout);
			fputs(regnames[r], stdout);
			flag = 1;
		}
	putchar('}');
	putchar('\n');
}

static void
format_16_17(off, word)
	unsigned off, word;
{
	unsigned cond;
	unsigned dest;

	cond = (word >> 8) & 0xF;
	switch (cond) {
	case 0xE:
		printf("<invalid: bal>\n");
		return;
	case 0xF:
		printf("swi\t0x%x\n", word & 0xFF);
		return;
	}
	dest = (word & 0xFF) << 1;
	if (dest & 0x00000100)
		dest |= 0xFFFFFE00;
	dest += base_vma + off + 4;
	printf("b%s\t0x%x\n", condition_decode[cond], dest);
}

static void
format_18(off, word)
	unsigned off, word;
{
	unsigned dest;

	if (word & 0x800) {
		printf("<invalid format 18>\n");
		return;
	}
	dest = (word & 0x7FF) << 1;
	if (dest & 0x00000800)
		dest |= 0xFFFFF000;
	dest += base_vma + off + 4;
	printf("b\t0x%x\n", dest);
}

void
thumb_disasm_line(off)
	unsigned off;
{
	unsigned word;

	word = get_u16(filemap + off);
	printf("%8x:\t%04x\t\t", base_vma + off, word);
	switch (word >> 12) {
	case 0:
	case 1:
		format_1_2(word);
		return;
	case 2:
	case 3:
		format_3(word);
		return;
	case 4:
		if (word & 0x800)
			format_6(off, word);
		else if (word & 0x400)
			format_5(word);
		else
			format_4(word);
		return;
	case 5:
		if (word & 0x200)
			format_8(word);
		else
			format_7(word);
		return;
	case 6:
	case 7:
		format_9(word);
		return;
	case 8:
		format_10(word);
		return;
	case 9:
		format_11(word);
		return;
	case 0xA:
		format_12(off, word);
		return;
	case 0xB:
		if (word & 0x400)
			format_14(word);
		else
			format_13(word);
		return;
	case 0xC:
		format_15(word);
		return;
	case 0xD:
		format_16_17(off, word);
		return;
	case 0xE:
		format_18(off, word);
		return;
	case 0xF:
		printf("<half-bl>\n");
		return;
	}
}

thumb_check_bl(off)
	unsigned off;
{
	unsigned ins1, ins2;
	unsigned dest;

	if (off + 4 > disasm_len)
		return(0);
	ins1 = get_u16(filemap + off);
	if ((ins1 & 0xF800) != 0xF000)
		return(0);
	ins2 = get_u16(filemap + off + 2);
	if ((ins2 & 0xF800) != 0xF800)
		return(0);
	/* match */
	dest = ((ins1 & 0x7FF) << 12) | ((ins2 & 0x7FF) << 1);
	if (dest & 0x00400000)
		dest |= 0xFF800000;
	dest += base_vma + off + 4;
	printf("%8x:\t%04x %04x\tbl\t0x%x\n", base_vma + off, ins1, ins2, dest);
	return(1);
}

main(argc, argv)
	char **argv;
{
	unsigned off;

	common_init(argc, argv, 2);
	for (off = 0; off < disasm_len; ) {
		if (thumb_check_bl(off))
			off += 4;
		else {
			thumb_disasm_line(off);
			off += 2;
		}
	}
	exit(0);
}