OPF_JMP = (1 << 3), /* branches, ret and call */
OPF_CC = (1 << 4), /* uses flags */
OPF_TAIL = (1 << 5), /* ret or tail call */
- OPF_REP = (1 << 6), /* prefixed by rep */
- OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */
+ OPF_RSAVE = (1 << 6), /* push/pop is local reg save/load */
+ OPF_REP = (1 << 7), /* prefixed by rep */
+ OPF_REPZ = (1 << 8), /* rep is repe/repz */
+ OPF_REPNZ = (1 << 9), /* rep is repne/repnz */
};
enum op_op {
OP_CDQ,
OP_STOS,
OP_MOVS,
+ OP_CMPS,
OP_RET,
OP_ADD,
OP_SUB,
struct parsed_opr {
enum opr_type type;
enum opr_lenmod lmod;
- unsigned int is_ptr:1; // pointer in C
+ unsigned int is_ptr:1; // pointer in C
+ unsigned int is_array:1; // array in C
int reg;
unsigned int val;
char name[256];
int pfomask; // flagop: parsed_flag_op that can't be delayed
int argmask; // push: args that are altered before call
int cc_scratch; // scratch storage during analysis
- int bt_i; // branch target (for branches)
- struct parsed_op *lrl; // label reference list entry
+ int bt_i; // branch target for branches
+ struct parsed_data *btj;// branch targets for jumptables
void *datap;
};
int offset;
};
+struct parsed_data {
+ char label[256];
+ enum opr_type type;
+ enum opr_lenmod lmod;
+ int count;
+ int count_alloc;
+ struct {
+ union {
+ char *label;
+ unsigned int val;
+ } u;
+ int bt_i;
+ } *d;
+};
+
+struct label_ref {
+ int i;
+ struct label_ref *next;
+};
+
enum ida_func_attr {
IDAFA_BP_FRAME = (1 << 0),
IDAFA_LIB_FUNC = (1 << 1),
static struct parsed_equ *g_eqs;
static int g_eqcnt;
static char g_labels[MAX_OPS][32];
-static struct parsed_op *g_label_refs[MAX_OPS];
+static struct label_ref g_label_refs[MAX_OPS];
static struct parsed_proto g_func_pp;
+static struct parsed_data *g_func_pd;
+static int g_func_pd_cnt;
static char g_func[256];
static char g_comment[256];
static int g_bp_frame;
static int g_sp_frame;
static int g_stack_fsz;
+static int g_ida_func_attr;
#define ferr(op_, fmt, ...) do { \
- printf("error:%s:#%ld: '%s': " fmt, g_func, (op_) - ops, \
+ printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
dump_op(op_), ##__VA_ARGS__); \
fcloseall(); \
exit(1); \
return -1;
}
-static void printf_number(char *buf, size_t buf_size, long number)
+static void printf_number(char *buf, size_t buf_size,
+ unsigned long number)
{
// output in C-friendly form
snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
return -1;
}
-static long parse_number(const char *number)
+static unsigned long parse_number(const char *number)
{
int len = strlen(number);
const char *p = number;
char *endp = NULL;
+ unsigned long ret;
int neg = 0;
int bad;
- long ret;
if (*p == '-') {
neg = 1;
if (len > 1 && *p == '0')
p++;
if (number[len - 1] == 'h') {
- ret = strtol(p, &endp, 16);
+ ret = strtoul(p, &endp, 16);
bad = (*endp != 'h');
}
else {
- ret = strtol(p, &endp, 10);
+ ret = strtoul(p, &endp, 10);
bad = (*endp != 0);
}
if (bad)
return c;
}
-static const char *parse_stack_el(const char *name)
+static int is_reg_in_str(const char *s)
{
- const char *p, *s;
+ int i;
+
+ if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
+ return 0;
+
+ for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
+ if (!strncmp(s, regs_r32[i], 3))
+ return 1;
+
+ return 0;
+}
+
+static const char *parse_stack_el(const char *name, char *extra_reg)
+{
+ const char *p, *p2, *s;
char *endp = NULL;
char buf[32];
long val;
int len;
- if (IS_START(name, "ebp+")
- && !('0' <= name[4] && name[4] <= '9'))
- {
- return name + 4;
+ p = name;
+ if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
+ p += 4;
+ if (extra_reg != NULL) {
+ strncpy(extra_reg, name, 3);
+ extra_reg[4] = 0;
+ }
+ }
+
+ if (IS_START(p, "ebp+")) {
+ p += 4;
+
+ p2 = strchr(p, '+');
+ if (p2 != NULL && is_reg_in_str(p)) {
+ if (extra_reg != NULL) {
+ strncpy(extra_reg, p, p2 - p);
+ extra_reg[p2 - p] = 0;
+ }
+ p = p2 + 1;
+ }
+
+ if (!('0' <= *p && *p <= '9'))
+ return p;
+
+ return NULL;
}
+
if (!IS_START(name, "esp+"))
return NULL;
return 0;
}
+static enum opr_type lmod_from_directive(const char *d)
+{
+ if (IS(d, "dd"))
+ return OPLM_DWORD;
+ else if (IS(d, "dw"))
+ return OPLM_WORD;
+ else if (IS(d, "db"))
+ return OPLM_BYTE;
+
+ aerr("unhandled directive: '%s'\n", d);
+ return OPLM_UNSPEC;
+}
+
static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
int *regmask)
{
{
struct parsed_proto pp;
enum opr_lenmod tmplmod;
+ unsigned long number;
int ret, len;
- long number;
int wordc_in;
char *tmp;
int i;
aerr("[] parse failure\n");
parse_indmode(opr->name, regmask_indirect, 1);
- if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name)) {
+ if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) {
// might be an equ
struct parsed_equ *eq =
- equ_find(NULL, parse_stack_el(opr->name), &i);
+ equ_find(NULL, parse_stack_el(opr->name, NULL), &i);
if (eq)
opr->lmod = eq->lmod;
}
}
else if (opr->lmod == OPLM_UNSPEC)
guess_lmod_from_c_type(opr, pp.ret_type);
+ opr->is_array = pp.is_array;
}
proto_release(&pp);
unsigned int flags;
} pref_table[] = {
{ "rep", OPF_REP },
+ { "repe", OPF_REP|OPF_REPZ },
+ { "repz", OPF_REP|OPF_REPZ },
+ { "repne", OPF_REP|OPF_REPNZ },
+ { "repnz", OPF_REP|OPF_REPNZ },
};
static const struct {
{ "movsb",OP_MOVS, 0, 0, OPF_DATA },
{ "movsw",OP_MOVS, 0, 0, OPF_DATA },
{ "movsd",OP_MOVS, 0, 0, OPF_DATA },
+ { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
+ { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
+ { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
{ "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
{ "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
{ "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
break;
case OP_MOVS:
+ case OP_CMPS:
if (op->operand_cnt != 0)
break;
- if (IS(words[op_w], "movsb"))
+ if (words[op_w][4] == 'b')
lmod = OPLM_BYTE;
- else if (IS(words[op_w], "movsw"))
+ else if (words[op_w][4] == 'w')
lmod = OPLM_WORD;
- else if (IS(words[op_w], "movsd"))
+ else if (words[op_w][4] == 'd')
lmod = OPLM_DWORD;
op->operand_cnt = 3;
setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
op->operand[1].lmod = OPLM_BYTE;
break;
+ case OP_PUSH:
+ if (op->operand[0].lmod == OPLM_UNSPEC
+ && (op->operand[0].type == OPT_CONST
+ || op->operand[0].type == OPT_OFFSET
+ || op->operand[0].type == OPT_LABEL))
+ op->operand[0].lmod = OPLM_DWORD;
+ break;
+
default:
break;
}
char *p = out;
int i;
+ if (po == NULL)
+ return "???";
+
snprintf(out, sizeof(out), "%s", op_name(po->op));
for (i = 0; i < po->operand_cnt; i++) {
p += strlen(p);
return out;
}
+static const char *lmod_type_u(struct parsed_op *po,
+ enum opr_lenmod lmod)
+{
+ switch (lmod) {
+ case OPLM_DWORD:
+ return "u32";
+ case OPLM_WORD:
+ return "u16";
+ case OPLM_BYTE:
+ return "u8";
+ default:
+ ferr(po, "invalid lmod: %d\n", lmod);
+ return "(_invalid_)";
+ }
+}
+
static const char *lmod_cast_u(struct parsed_op *po,
enum opr_lenmod lmod)
{
const char *name, int is_src, int is_lea)
{
const char *prefix = "";
+ char ofs_reg[16] = { 0, };
struct parsed_equ *eq;
int i, arg_i, arg_s;
const char *bp_arg;
int offset = 0;
int sf_ofs;
- bp_arg = parse_stack_el(name);
+ bp_arg = parse_stack_el(name, ofs_reg);
snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
eq = equ_find(po, bp_arg, &offset);
if (eq == NULL)
if (arg_i < 0 || arg_i >= g_func_pp.argc_stack)
ferr(po, "offset %d (%s) doesn't map to any arg\n",
offset, bp_arg);
+ if (ofs_reg[0] != 0)
+ ferr(po, "offset reg on arg access?\n");
for (i = arg_s = 0; i < g_func_pp.argc; i++) {
if (g_func_pp.arg[i].reg != NULL)
if (is_lea)
ferr(po, "lea to arg?\n");
- snprintf(buf, buf_size, "%sa%d", is_src ? "(u32)" : "", i + 1);
+ switch (lmod)
+ {
+ case OPLM_BYTE:
+ if (is_src && (offset & 3) == 0)
+ snprintf(buf, buf_size, "(u8)a%d", i + 1);
+ else
+ snprintf(buf, buf_size, "BYTE%d(a%d)", offset & 3, i + 1);
+ break;
+
+ case OPLM_WORD:
+ if (offset & 1)
+ ferr(po, "unaligned arg access\n");
+ if (is_src && (offset & 2) == 0)
+ snprintf(buf, buf_size, "(u16)a%d", i + 1);
+ else
+ snprintf(buf, buf_size, "%sWORD(a%d)",
+ (offset & 2) ? "HI" : "LO", i + 1);
+ break;
+
+ case OPLM_DWORD:
+ if (offset & 3)
+ ferr(po, "unaligned arg access\n");
+ snprintf(buf, buf_size, "%sa%d", is_src ? "(u32)" : "", i + 1);
+ break;
+
+ default:
+ ferr(po, "bp_arg bad lmod: %d\n", lmod);
+ }
+
+ // common problem
+ if ((offset & 3)
+ && (strstr(g_func_pp.arg[i].type, "int8")
+ || strstr(g_func_pp.arg[i].type, "int16")))
+ {
+ ferr(po, "bp_arg arg/w offset %d and type '%s'\n",
+ offset, g_func_pp.arg[i].type);
+ }
}
else {
if (g_stack_fsz == 0)
switch (lmod)
{
case OPLM_BYTE:
- snprintf(buf, buf_size, "%ssf.b[%d]", prefix, sf_ofs);
+ snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
+ prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
break;
+
case OPLM_WORD:
+ if ((sf_ofs & 1) || ofs_reg[0] != 0) {
+ // known unaligned or possibly unaligned
+ strcat(g_comment, " unaligned");
+ if (prefix[0] == 0)
+ prefix = "*(u16 *)&";
+ snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
+ prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
+ break;
+ }
snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
break;
+
case OPLM_DWORD:
+ if ((sf_ofs & 3) || ofs_reg[0] != 0) {
+ // known unaligned or possibly unaligned
+ strcat(g_comment, " unaligned");
+ if (prefix[0] == 0)
+ prefix = "*(u32 *)&";
+ snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
+ prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
+ break;
+ }
snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
break;
+
default:
ferr(po, "bp_stack bad lmod: %d\n", lmod);
}
break;
case OPT_REGMEM:
- if (parse_stack_el(popr->name)) {
+ if (parse_stack_el(popr->name, NULL)) {
stack_frame_access(po, popr->lmod, buf, buf_size,
popr->name, 1, is_lea);
break;
if (is_lea)
snprintf(buf, buf_size, "(u32)&%s", popr->name);
else
- snprintf(buf, buf_size, "(u32)%s", popr->name);
+ snprintf(buf, buf_size, "%s%s%s",
+ popr->is_ptr ? "(u32)" : "",
+ popr->name,
+ popr->is_array ? "[0]" : "");
break;
case OPT_OFFSET:
break;
case OPT_REGMEM:
- if (parse_stack_el(popr->name)) {
+ if (parse_stack_el(popr->name, NULL)) {
stack_frame_access(po, popr->lmod, buf, buf_size,
popr->name, 0, 0);
break;
return out_src_opr(buf, buf_size, po, popr, 0);
case OPT_LABEL:
- snprintf(buf, buf_size, "%s", popr->name);
+ snprintf(buf, buf_size, "%s%s", popr->name,
+ popr->is_array ? "[0]" : "");
break;
default:
{
struct parsed_op *po;
int ret = 0;
+ int j;
for (; i < opcnt; i++) {
po = &ops[i];
continue;
if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
+ if (po->btj != NULL) {
+ // jumptable
+ for (j = 0; j < po->btj->count - 1; j++) {
+ ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
+ depth, maxdepth, do_flags);
+ if (ret < 0)
+ return ret; // dead end
+ }
+ // follow last jumptable entry
+ i = po->btj->d[j].bt_i - 1;
+ continue;
+ }
+
if (po->bt_i < 0) {
ferr(po, "dead branch\n");
return -1;
// scan for positive, constant esp adjust
static int scan_for_esp_adjust(int i, int opcnt, int *adj)
{
+ struct parsed_op *po;
+ *adj = 0;
+
for (; i < opcnt; i++) {
- if (ops[i].op == OP_ADD && ops[i].operand[0].reg == xSP) {
- if (ops[i].operand[1].type != OPT_CONST)
+ po = &ops[i];
+
+ if (po->op == OP_ADD && po->operand[0].reg == xSP) {
+ if (po->operand[1].type != OPT_CONST)
ferr(&ops[i], "non-const esp adjust?\n");
- *adj = ops[i].operand[1].val;
+ *adj += po->operand[1].val;
if (*adj & 3)
ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
return i;
}
-
- if ((ops[i].flags & (OPF_JMP|OPF_TAIL))
- || ops[i].op == OP_PUSH || ops[i].op == OP_POP)
+ else if (po->op == OP_PUSH)
+ *adj -= lmod_bytes(po, po->operand[0].lmod);
+ else if (po->op == OP_POP)
+ *adj += lmod_bytes(po, po->operand[0].lmod);
+ else if ((po->flags & (OPF_JMP|OPF_TAIL)) && po->op != OP_CALL)
return -1;
+
if (g_labels[i][0] != 0)
return -1;
}
return -1;
}
+static void add_label_ref(struct label_ref *lr, int op_i)
+{
+ struct label_ref *lr_new;
+
+ if (lr->i == -1) {
+ lr->i = op_i;
+ return;
+ }
+
+ lr_new = calloc(1, sizeof(*lr_new));
+ lr_new->i = op_i;
+ lr->next = lr_new;
+}
+
static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
{
struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
struct parsed_opr *last_arith_dst = NULL;
char buf1[256], buf2[256], buf3[256];
struct parsed_proto *pp, *pp_tmp;
+ struct parsed_data *pd;
const char *tmpname;
enum parsed_flag_op pfo;
int save_arg_vars = 0;
int found = 0;
int depth = 0;
int no_output;
+ int i, j, l;
int dummy;
int arg;
- int i, j;
int reg;
int ret;
g_bp_frame = 1;
ops[0].flags |= OPF_RMD;
ops[1].flags |= OPF_RMD;
+ i = 2;
if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
g_stack_fsz = opr_const(&ops[2], 1);
ops[2].flags |= OPF_RMD;
+ i++;
}
else {
// another way msvc builds stack frame..
ecx_push++;
i++;
}
+ // and another way..
+ if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
+ && ops[i].operand[1].type == OPT_CONST
+ && ops[i + 1].op == OP_CALL
+ && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
+ {
+ g_stack_fsz += ops[i].operand[1].val;
+ ops[i].flags |= OPF_RMD;
+ i++;
+ ops[i].flags |= OPF_RMD;
+ i++;
+ }
}
found = 0;
- i = 2;
do {
for (; i < opcnt; i++)
if (ops[i].op == OP_RET)
for (i = 0; i < opcnt; i++) {
po = &ops[i];
po->bt_i = -1;
+ po->btj = NULL;
if ((po->flags & OPF_RMD) || !(po->flags & OPF_JMP)
|| po->op == OP_CALL || po->op == OP_RET)
continue;
- for (j = 0; j < opcnt; j++) {
- if (g_labels[j][0] && IS(po->operand[0].name, g_labels[j])) {
- po->bt_i = j;
- po->lrl = g_label_refs[j];
- g_label_refs[j] = po;
+ if (po->operand[0].type == OPT_REGMEM) {
+ char *p = strchr(po->operand[0].name, '[');
+ if (p == NULL)
+ ferr(po, "unhandled indirect branch\n");
+ ret = p - po->operand[0].name;
+ strncpy(buf1, po->operand[0].name, ret);
+ buf1[ret] = 0;
+
+ for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
+ if (IS(g_func_pd[j].label, buf1)) {
+ pd = &g_func_pd[j];
+ break;
+ }
+ }
+ if (pd == NULL)
+ ferr(po, "label '%s' not parsed?\n", buf1);
+ if (pd->type != OPT_OFFSET)
+ ferr(po, "label '%s' with non-offset data?\n", buf1);
+
+ // find all labels, link
+ for (j = 0; j < pd->count; j++) {
+ for (l = 0; l < opcnt; l++) {
+ if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
+ add_label_ref(&g_label_refs[l], i);
+ pd->d[j].bt_i = l;
+ break;
+ }
+ }
+ }
+
+ po->btj = pd;
+ continue;
+ }
+
+ for (l = 0; l < opcnt; l++) {
+ if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
+ add_label_ref(&g_label_refs[l], i);
+ po->bt_i = l;
break;
}
}
- if (po->bt_i == -1) {
+ if (po->bt_i != -1)
+ continue;
+
+ if (po->operand[0].type == OPT_LABEL) {
// assume tail call
po->op = OP_CALL;
po->flags |= OPF_TAIL;
+ continue;
}
+
+ ferr(po, "unhandled branch\n");
}
// pass3:
ferr(po, "proto_parse failed for call '%s'\n", tmpname);
}
- ret = scan_for_esp_adjust(i + 1, opcnt, &j);
+ // look for and make use of esp adjust
+ ret = -1;
+ if (!pp->is_stdcall)
+ ret = scan_for_esp_adjust(i + 1, opcnt, &j);
if (ret >= 0) {
+ if (pp->is_vararg) {
+ if (j / 4 < pp->argc_stack)
+ ferr(po, "esp adjust is too small: %x < %x\n",
+ j, pp->argc_stack * 4);
+ // modify pp to make it have varargs as normal args
+ arg = pp->argc;
+ pp->argc += j / 4 - pp->argc_stack;
+ for (; arg < pp->argc; arg++) {
+ pp->arg[arg].type = strdup("int");
+ pp->argc_stack++;
+ }
+ if (pp->argc > ARRAY_SIZE(pp->arg))
+ ferr(po, "too many args\n");
+ }
if (pp->argc_stack != j / 4)
ferr(po, "stack tracking failed: %x %x\n",
- pp->argc_stack, j);
+ pp->argc_stack * 4, j);
+
ops[ret].flags |= OPF_RMD;
+ // a bit of a hack, but deals with use of
+ // single adj for multiple calls
+ ops[ret].operand[1].val -= j;
}
+ else if (pp->is_vararg)
+ ferr(po, "missing esp_adjust for vararg func '%s'\n",
+ pp->name);
// can't call functions with non-__cdecl callbacks yet
for (arg = 0; arg < pp->argc; arg++) {
}
}
+ // collect all stack args
for (arg = 0; arg < pp->argc; arg++)
if (pp->arg[arg].reg == NULL)
break;
|| (ops[j - 1].flags & (OPF_JMP|OPF_CC)) == OPF_JMP))
{
// follow the branch in reverse
- if (g_label_refs[j] == NULL)
+ if (g_label_refs[j].i == -1)
ferr(po, "no refs for '%s'?\n", g_labels[j]);
- if (g_label_refs[j]->lrl != NULL)
- ferr(po, "unhandled multiple fefs to '%s'\n", g_labels[j]);
- j = (g_label_refs[j] - ops) + 1;
+ if (g_label_refs[j].next != NULL)
+ ferr(po, "unhandled multiple refs to '%s'\n", g_labels[j]);
+ j = g_label_refs[j].i + 1;
continue;
}
break;
if (scan_for_mod(tmp_op, ret + 1, i) >= 0)
pfomask = 1 << pfo;
}
+ else if (tmp_op->op == OP_CMPS) {
+ pfomask = 1 << PFO_Z;
+ }
else {
if ((pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
|| scan_for_mod_opr0(tmp_op, ret + 1, i) >= 0)
}
}
+ // output LUTs/jumptables
+ for (i = 0; i < g_func_pd_cnt; i++) {
+ pd = &g_func_pd[i];
+ fprintf(fout, " static const ");
+ if (pd->type == OPT_OFFSET) {
+ fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
+
+ for (j = 0; j < pd->count; j++) {
+ if (j > 0)
+ fprintf(fout, ", ");
+ fprintf(fout, "&&%s", pd->d[j].u.label);
+ }
+ }
+ else {
+ fprintf(fout, "%s %s[] =\n { ",
+ lmod_type_u(ops, pd->lmod), pd->label);
+
+ for (j = 0; j < pd->count; j++) {
+ if (j > 0)
+ fprintf(fout, ", ");
+ fprintf(fout, "%u", pd->d[j].u.val);
+ }
+ }
+ fprintf(fout, " };\n");
+ }
+
// declare stack frame
if (g_stack_fsz)
fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
// output ops
for (i = 0; i < opcnt; i++)
{
- if (g_labels[i][0] != 0 && g_label_refs[i] != NULL)
+ if (g_labels[i][0] != 0 && g_label_refs[i].i != -1)
fprintf(fout, "\n%s:\n", g_labels[i]);
po = &ops[i];
is_delayed = 1;
}
else if (po->datap != NULL) {
- // use preprocessed results
+ // use preprocessed flag calc results
tmp_op = po->datap;
if (!tmp_op || !(tmp_op->pfomask & (1 << pfo)))
ferr(po, "not prepared for pfo %d\n", pfo);
}
break;
+ case OP_CMPS:
+ // assumes DF=0
+ // repe ~ repeat while ZF=1
+ assert_operand_cnt(3);
+ j = lmod_bytes(po, po->operand[0].lmod);
+ strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
+ if (po->flags & OPF_REP) {
+ fprintf(fout,
+ " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n",
+ j, j);
+ fprintf(fout,
+ " if ((cond_z = (%sedi == %sesi)) %s 0)\n",
+ buf1, buf1, (po->flags & OPF_REPZ) ? "==" : "!=");
+ fprintf(fout,
+ " break;");
+ snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
+ (po->flags & OPF_REPZ) ? "e" : "ne");
+ }
+ else {
+ fprintf(fout,
+ " cond_z = (%sedi = %sesi); edi += %d; esi += %d;",
+ buf1, buf1, j, j);
+ strcpy(g_comment, "cmps");
+ }
+ pfomask &= ~(1 << PFO_Z);
+ last_arith_dst = NULL;
+ delayed_flag_op = NULL;
+ break;
+
// arithmetic w/flags
case OP_ADD:
case OP_SUB:
}
else
no_output = 1;
+ last_arith_dst = NULL;
delayed_flag_op = po;
break;
case OP_JMP:
assert_operand_cnt(1);
- if (po->operand[0].type != OPT_LABEL)
- ferr(po, "unhandled call type\n");
+ if (po->operand[0].type == OPT_REGMEM) {
+ ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
+ buf1, buf2);
+ if (ret != 2)
+ ferr(po, "parse failure for jmp '%s'\n",
+ po->operand[0].name);
+ fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
+ break;
+ }
+ else if (po->operand[0].type != OPT_LABEL)
+ ferr(po, "unhandled jmp type\n");
fprintf(fout, " goto %s;", po->operand[0].name);
break;
break;
}
+ // some sanity checking
+ if ((po->flags & OPF_REP) && po->op != OP_STOS
+ && po->op != OP_MOVS && po->op != OP_CMPS)
+ ferr(po, "unexpected rep\n");
+ if ((po->flags & (OPF_REPZ|OPF_REPNZ)) && po->op != OP_CMPS)
+ ferr(po, "unexpected repz/repnz\n");
+
if (g_comment[0] != 0) {
fprintf(fout, " // %s", g_comment);
g_comment[0] = 0;
fprintf(fout, "\n");
if (pfomask != 0)
- ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
+ ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
// see is delayed flag stuff is still valid
if (delayed_flag_op != NULL && delayed_flag_op != po) {
// cleanup
for (i = 0; i < opcnt; i++) {
+ struct label_ref *lr, *lr_del;
+
+ lr = g_label_refs[i].next;
+ while (lr != NULL) {
+ lr_del = lr;
+ lr = lr->next;
+ free(lr_del);
+ }
+ g_label_refs[i].i = -1;
+ g_label_refs[i].next = NULL;
+
if (ops[i].op == OP_CALL) {
pp = ops[i].datap;
if (pp) {
int main(int argc, char *argv[])
{
FILE *fout, *fasm, *frlist;
+ struct parsed_data *pd = NULL;
+ int pd_alloc = 0;
+ char **rlist = NULL;
+ int rlist_len = 0;
+ int rlist_alloc = 0;
char line[256];
char words[16][256];
- int ida_func_attr = 0;
+ enum opr_lenmod lmod;
int in_func = 0;
+ int pending_endp = 0;
int skip_func = 0;
int skip_warned = 0;
int eq_alloc;
- char **rlist = NULL;
- int rlist_len = 0;
- int rlist_alloc = 0;
int verbose = 0;
int arg_out;
int arg = 1;
int pi = 0;
- int i, len;
+ int i, j, len;
char *p;
int wordc;
g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
my_assert_not(g_eqs, NULL);
+ for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
+ g_label_refs[i].i = -1;
+ g_label_refs[i].next = NULL;
+ }
+
while (fgets(line, sizeof(line), fasm))
{
+ wordc = 0;
asmln++;
p = sskip(line);
"thunk",
"fpd=",
};
- if (p[2] != 'A' || strncmp(p, "; Attributes:", 13) != 0)
+ if (p[2] == '=' && IS_START(p, "; =============== S U B"))
+ goto do_pending_endp; // eww..
+
+ if (p[2] != 'A' || !IS_START(p, "; Attributes:"))
continue;
// parse IDA's attribute-list comment
- ida_func_attr = 0;
+ g_ida_func_attr = 0;
p = sskip(p + 13);
// get rid of random tabs
for (i = 0; p[i] != 0; i++)
for (; *p != 0; p = sskip(p)) {
for (i = 0; i < ARRAY_SIZE(attrs); i++) {
if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
- ida_func_attr |= 1 << i;
+ g_ida_func_attr |= 1 << i;
p += strlen(attrs[i]);
break;
}
continue;
}
+do_pending_endp:
+ // do delayed endp processing to collect switch jumptables
+ if (pending_endp) {
+ if (in_func && !skip_func && wordc >= 2
+ && ((words[0][0] == 'd' && words[0][2] == 0)
+ || (words[1][0] == 'd' && words[1][2] == 0)))
+ {
+ i = 1;
+ if (words[1][0] == 'd' && words[1][2] == 0) {
+ // label
+ if (g_func_pd_cnt >= pd_alloc) {
+ pd_alloc = pd_alloc * 2 + 16;
+ g_func_pd = realloc(g_func_pd,
+ sizeof(g_func_pd[0]) * pd_alloc);
+ my_assert_not(g_func_pd, NULL);
+ }
+ pd = &g_func_pd[g_func_pd_cnt];
+ g_func_pd_cnt++;
+ memset(pd, 0, sizeof(*pd));
+ strcpy(pd->label, words[0]);
+ pd->type = OPT_CONST;
+ pd->lmod = lmod_from_directive(words[1]);
+ i = 2;
+ }
+ else {
+ lmod = lmod_from_directive(words[0]);
+ if (lmod != pd->lmod)
+ aerr("lmod change? %d->%d\n", pd->lmod, lmod);
+ }
+
+ if (pd->count_alloc < pd->count + wordc) {
+ pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
+ pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
+ my_assert_not(pd->d, NULL);
+ }
+ for (; i < wordc; i++) {
+ if (IS(words[i], "offset")) {
+ pd->type = OPT_OFFSET;
+ i++;
+ }
+ p = strchr(words[i], ',');
+ if (p != NULL)
+ *p = 0;
+ if (pd->type == OPT_OFFSET)
+ pd->d[pd->count].u.label = strdup(words[i]);
+ else
+ pd->d[pd->count].u.val = parse_number(words[i]);
+ pd->d[pd->count].bt_i = -1;
+ pd->count++;
+ }
+ continue;
+ }
+
+ if (in_func && !skip_func)
+ gen_func(fout, g_fhdr, g_func, pi);
+
+ pending_endp = 0;
+ in_func = 0;
+ g_ida_func_attr = 0;
+ skip_warned = 0;
+ skip_func = 0;
+ g_func[0] = 0;
+ if (pi != 0) {
+ memset(&ops, 0, pi * sizeof(ops[0]));
+ memset(g_labels, 0, pi * sizeof(g_labels[0]));
+ pi = 0;
+ }
+ g_eqcnt = 0;
+ for (i = 0; i < g_func_pd_cnt; i++) {
+ pd = &g_func_pd[i];
+ if (pd->type == OPT_OFFSET) {
+ for (j = 0; j < pd->count; j++)
+ free(pd->d[j].u.label);
+ }
+ free(pd->d);
+ pd->d = NULL;
+ }
+ g_func_pd_cnt = 0;
+ pd = NULL;
+ if (wordc == 0)
+ continue;
+ }
+
if (IS(words[1], "proc")) {
if (in_func)
aerr("proc '%s' while in_func '%s'?\n",
words[0], g_func);
p = words[0];
- if ((ida_func_attr & IDAFA_THUNK)
+ if ((g_ida_func_attr & IDAFA_THUNK)
|| bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
skip_func = 1;
strcpy(g_func, words[0]);
aerr("endp '%s' while in_func '%s'?\n",
words[0], g_func);
- if (in_func && !skip_func)
- gen_func(fout, g_fhdr, g_func, pi);
-
- in_func = 0;
- skip_warned = 0;
- skip_func = 0;
- g_func[0] = 0;
- if (pi != 0) {
- memset(&ops, 0, pi * sizeof(ops[0]));
- memset(g_labels, 0, pi * sizeof(g_labels[0]));
- memset(g_label_refs, 0, pi * sizeof(g_label_refs[0]));
- pi = 0;
- }
- g_eqcnt = 0;
- ida_func_attr = 0;
+ pending_endp = 1;
continue;
}
continue;
}
- if (IS(words[1], "=")) {
+ if (wordc > 1 && IS(words[1], "=")) {
if (wordc != 5)
aerr("unhandled equ, wc=%d\n", wordc);
if (g_eqcnt >= eq_alloc) {
notaz.gp2x.de / ia32rtools.git / blobdiff