static const char *reserved_names[] = {
"name",
+ "type",
"offset",
+ "aam",
+ "text",
+ "size",
+ "c",
};
static int is_name_reserved(const char *name)
nonlocal_bt[nonlocal_bt_cnt++] = ea;
}
-static void do_def_line(char *buf, size_t buf_size, const char *line)
+// is instruction a (un)conditional jump (not call)?
+static int is_insn_jmp(uint16 itype)
{
- char *endp = NULL;
- ea_t ea, *ea_ret;
+ return itype == NN_jmp || (NN_ja <= itype && itype <= NN_jz);
+}
+
+static void do_def_line(char *buf, size_t buf_size, const char *line,
+ ea_t ea)
+{
+ ea_t *ea_ret;
+ char *p;
int len;
tag_remove(line, buf, buf_size); // remove color codes
}
memmove(buf, buf + 9, len - 9 + 1); // rm address
- if (IS_START(buf, "loc_")) {
- ea = strtoul(buf + 4, &endp, 16);
- if (ea != 0 && *endp == ':') {
- ea_ret = (ea_t *)bsearch(&ea, nonlocal_bt, nonlocal_bt_cnt,
- sizeof(nonlocal_bt[0]), nonlocal_bt_cmp);
- if (ea_ret != 0) {
- if (endp[1] != ' ')
- msg("no trailing blank in '%s'\n", buf);
- else
- endp[1] = ':';
- }
+ p = buf;
+ while (*p && *p != ' ' && *p != ':')
+ p++;
+ if (*p == ':') {
+ ea_ret = (ea_t *)bsearch(&ea, nonlocal_bt, nonlocal_bt_cnt,
+ sizeof(nonlocal_bt[0]), nonlocal_bt_cmp);
+ if (ea_ret != 0) {
+ if (p[1] != ' ')
+ msg("no trailing blank in '%s'\n", buf);
+ else
+ p[1] = ':';
}
}
}
FILE *fout = NULL;
int fout_line = 0;
char buf[MAXSTR];
- int drop_large = 0;
+ char buf2[MAXSTR];
+ const char *name;
struc_t *frame;
func_t *func;
ea_t ui_ea_block = 0, ea_size;
ea_t tmp_ea, target_ea;
ea_t ea;
+ flags_t ea_flags;
+ uval_t idx;
int i, o, m, n;
int ret;
char *p;
nonlocal_bt_cnt = 0;
+ // get rid of structs, masm doesn't understand them
+ idx = get_first_struc_idx();
+ while (idx != BADNODE) {
+ tid_t tid = get_struc_by_idx(idx);
+ struc_t *struc = get_struc(tid);
+ get_struc_name(tid, buf, sizeof(buf));
+ msg("removing struct '%s'\n", buf);
+ //del_struc_members(struc, 0, get_max_offset(struc));
+ del_struc(struc);
+
+ idx = get_first_struc_idx();
+ }
+
// 1st pass: walk through all funcs
func = get_func(inf.minEA);
while (func != NULL)
if (IS_START(buf, "arg_") || IS_START(buf, "var_"))
continue;
- if (is_name_reserved(buf)) {
+ // check for dupe names
+ int m1, dupe = 0;
+ for (m1 = 0; m1 < m; m1++) {
+ get_member_name(frame->members[m1].id, buf2, sizeof(buf2));
+ if (stricmp(buf, buf2) == 0)
+ dupe = 1;
+ }
+
+ if (is_name_reserved(buf) || dupe) {
msg("%x: renaming '%s'\n", ea, buf);
qstrncat(buf, "_", sizeof(buf));
ret = set_member_name(frame, frame->members[m].soff, buf);
func = get_next_func(ea);
}
- // 2nd pass over whole .text segment
+ // 2nd pass over whole .text and .(ro)data segments
for (ea = inf.minEA; ea != BADADDR; ea = next_head(ea, inf.maxEA))
{
segment_t *seg = getseg(ea);
- if (!seg || seg->type != SEG_CODE)
+ if (!seg)
+ break;
+ if (seg->type == SEG_XTRN)
+ continue;
+ if (seg->type != SEG_CODE && seg->type != SEG_DATA)
break;
- flags_t ea_flags = get_flags_novalue(ea);
+ ea_flags = get_flags_novalue(ea);
func = get_func(ea);
if (isCode(ea_flags))
{
continue;
}
+ // masm doesn't understand IDA's float/xmm types
+ if (cmd.itype == NN_fld || cmd.itype == NN_fst
+ || cmd.itype == NN_movapd || cmd.itype == NN_movlpd)
+ {
+ for (o = 0; o < UA_MAXOP; o++) {
+ if (cmd.Operands[o].type == o_void)
+ break;
+
+ if (cmd.Operands[o].type == o_mem) {
+ tmp_ea = cmd.Operands[o].addr;
+ flags_t tmp_ea_flags = get_flags_novalue(tmp_ea);
+ if (!isUnknown(tmp_ea_flags)) {
+ buf[0] = 0;
+ get_name(ea, tmp_ea, buf, sizeof(buf));
+ msg("%x: undefining %x '%s'\n", ea, tmp_ea, buf);
+ do_unknown(tmp_ea, DOUNK_EXPAND);
+ }
+ }
+ }
+ }
+ // detect code alignment
+ else if (cmd.itype == NN_lea) {
+ if (cmd.Operands[0].reg == cmd.Operands[1].reg
+ && cmd.Operands[1].type == o_displ
+ && cmd.Operands[1].addr == 0)
+ {
+ tmp_ea = next_head(ea, inf.maxEA);
+ if ((tmp_ea & 0x03) == 0) {
+ n = calc_max_align(tmp_ea);
+ if (n > 4) // masm doesn't like more..
+ n = 4;
+ msg("%x: align %d\n", ea, 1 << n);
+ do_unknown(ea, DOUNK_SIMPLE);
+ doAlign(ea, tmp_ea - ea, n);
+ }
+ }
+ }
+
// find non-local branches
- if ((cmd.itype == NN_jmp || insn_jcc())
- && cmd.Operands[0].type == o_near)
+ if (is_insn_jmp(cmd.itype) && cmd.Operands[0].type == o_near)
{
target_ea = cmd.Operands[0].addr;
if (func == NULL)
}
}
else { // not code
+ int do_undef = 0;
+ ea_size = get_item_size(ea);
+
if (func == NULL && isOff0(ea_flags)) {
- ea_size = get_item_size(ea);
for (tmp_ea = 0; tmp_ea < ea_size; tmp_ea += 4)
nonlocal_add(get_long(ea + tmp_ea));
}
+
+ // IDA vs masm float/mmx/xmm type incompatibility
+ if (isDouble(ea_flags) || isTbyt(ea_flags)
+ || isPackReal(ea_flags))
+ {
+ do_undef = 1;
+ }
+ else if (isOwrd(ea_flags)) {
+ buf[0] = 0;
+ get_name(BADADDR, ea, buf, sizeof(buf));
+ if (IS_START(buf, "xmm"))
+ do_undef = 1;
+ }
+ // masm doesn't understand IDA's unicode
+ else if (isASCII(ea_flags) && ea_size >= 4
+ && (get_long(ea) & 0xff00ff00) == 0) // lame..
+ {
+ do_undef = 1;
+ }
+ // masm doesn't understand large aligns
+ else if (isAlign(ea_flags) && ea_size > 0x10) {
+ msg("%x: undefining align %d\n", ea, ea_size);
+ do_unknown(ea, DOUNK_EXPAND);
+ }
+
+ if (do_undef) {
+ buf[0] = 0;
+ get_name(BADADDR, ea, buf, sizeof(buf));
+ msg("%x: undefining '%s'\n", ea, buf);
+ do_unknown(ea, DOUNK_EXPAND);
+ }
+ }
+ }
+
+ // check namelist for reserved names
+ n = get_nlist_size();
+ for (i = 0; i < n; i++) {
+ ea = get_nlist_ea(i);
+ name = get_nlist_name(i);
+ if (name == NULL) {
+ msg("%x: null name?\n", ea);
+ continue;
+ }
+
+ // rename vars with '?@' (funcs are ok)
+ int change_qat = 0;
+ ea_flags = get_flags_novalue(ea);
+ if (!isCode(ea_flags) && strpbrk(name, "?@"))
+ change_qat = 1;
+
+ if (change_qat || is_name_reserved(name)) {
+ msg("%x: renaming name '%s'\n", ea, name);
+ qsnprintf(buf, sizeof(buf), "%s_g", name);
+
+ if (change_qat) {
+ for (p = buf; *p != 0; p++) {
+ if (*p == '?' || *p == '@') {
+ qsnprintf(buf2, sizeof(buf2), "%02x", (unsigned char)*p);
+ memmove(p + 1, p, strlen(p) + 1);
+ memcpy(p, buf2, 2);
+ }
+ }
+ }
+
+ set_name(ea, buf);
}
}
ln.set_place(&pl);
n = ln.get_linecnt();
for (i = 0; i < n - 1; i++) {
- do_def_line(buf, sizeof(buf), ln.down());
+ do_def_line(buf, sizeof(buf), ln.down(), ea);
if (strstr(buf, "include"))
continue;
memcpy(p, ".xmm", 4);
fout_line++;
qfprintf(fout, "%s\n", buf);
+ continue;
+ }
+ p = strstr(buf, ".model");
+ if (p != NULL) {
+ qstrncpy(p, "include imports.inc", sizeof(buf) - (p - buf));
+ fout_line++;
+ qfprintf(fout, "\n%s\n", buf);
+ i++;
+ break;
}
}
+ pl.lnnum = i;
for (;;)
{
- drop_large = 0;
+ int drop_large = 0, do_rva = 0, set_scale = 0, jmp_near = 0;
+ int word_imm = 0, dword_imm = 0, do_pushf = 0;
if ((ea >> 14) != ui_ea_block) {
ui_ea_block = ea >> 14;
}
segment_t *seg = getseg(ea);
- if (!seg || seg->type != SEG_CODE)
+ if (!seg || (seg->type != SEG_CODE && seg->type != SEG_DATA))
goto pass;
- if (!decode_insn(ea))
- goto pass;
-
- // note: decode_insn() picks up things like dd, size is then weird
- //cmd_size = cmd.size;
- for (o = 0; o < UA_MAXOP; o++) {
- if (cmd.Operands[o].type == o_void)
- break;
+ ea_flags = get_flags_novalue(ea);
+ if (isCode(ea_flags))
+ {
+ if (!decode_insn(ea))
+ goto pass;
- if (cmd.Operands[o].type == o_mem
- && cmd.Operands[o].specval_shorts.high == 0x21) // correct?
+ if (is_insn_jmp(cmd.itype) && cmd.Operands[0].type == o_near
+ && cmd.Operands[0].dtyp == dt_dword)
{
- drop_large = 1;
+ jmp_near = 1;
}
-#if 0
- if (cmd.Operands[o].type == o_displ && cmd.Operands[o].reg == 5) {
- member_t *m;
-
- m = get_stkvar(cmd.Operands[o], cmd.Operands[o].addr, NULL);
- if (m == NULL) {
- msg("%x: no stkvar for offs %x\n",
- ea, cmd.Operands[o].addr);
- goto out;
+ else if ((cmd.itype == NN_pushf || cmd.itype == NN_popf)
+ && natop())
+ {
+ do_pushf = 1;
+ }
+
+ for (o = 0; o < UA_MAXOP; o++) {
+ const op_t &opr = cmd.Operands[o];
+ if (opr.type == o_void)
+ break;
+
+ // correct?
+ if (opr.type == o_mem && opr.specval_shorts.high == 0x21)
+ drop_large = 1;
+ if (opr.hasSIB && x86_scale(opr) == 0
+ && x86_index(opr) != INDEX_NONE)
+ {
+ set_scale = 1;
}
- if (get_struc_name(m->id, buf, sizeof(buf)) <= 0) {
- msg("%x: stkvar with offs %x has no name?\n",
- ea, cmd.Operands[o].addr);
- goto out;
+ // annoying alignment variant..
+ if (opr.type == o_imm && opr.dtyp == dt_dword
+ && (opr.value < 0x80 || opr.value > 0xffffff80)
+ && cmd.size >= opr.offb + 4)
+ {
+ if (get_long(ea + opr.offb) == opr.value)
+ dword_imm = 1;
+ }
+ else if (opr.type == o_imm && opr.dtyp == dt_word
+ && (opr.value < 0x80 || opr.value > 0xff80)
+ && cmd.size >= opr.offb + 2)
+ {
+ if (get_word(ea + opr.offb) == (ushort)opr.value)
+ word_imm = 1;
}
- msg("%x: name '%s'\n", ea, buf);
}
-#endif
+ }
+ else { // not code
+ if (isOff0(ea_flags))
+ do_rva = 1;
}
pass:
- do_def_line(buf, sizeof(buf), ln.down());
- if (drop_large) {
- p = strstr(buf, "large ");
- if (p != NULL)
- memmove(p, p + 6, strlen(p + 6) + 1);
- }
+ n = ln.get_linecnt();
+ for (i = pl.lnnum; i < n; i++) {
+ do_def_line(buf, sizeof(buf), ln.down(), ea);
+
+ char *fw;
+ for (fw = buf; *fw != 0 && *fw == ' '; )
+ fw++;
+
+ // patches..
+ if (drop_large) {
+ p = strstr(fw, "large ");
+ if (p != NULL)
+ memmove(p, p + 6, strlen(p + 6) + 1);
+ }
+ while (do_rva) {
+ p = strstr(fw, " rva ");
+ if (p == NULL)
+ break;
+ memmove(p + 4 + 3, p + 4, strlen(p + 4) + 1);
+ memcpy(p + 1, "offset", 6);
+ }
+ if (set_scale) {
+ p = strchr(fw, '[');
+ if (p != NULL)
+ p = strchr(p, '+');
+ if (p != NULL && p[1] == 'e') {
+ p += 4;
+ // scale is 1, must specify it explicitly so that
+ // masm chooses the right scaled reg
+ memmove(p + 2, p, strlen(p) + 1);
+ memcpy(p, "*1", 2);
+ }
+ }
+ else if (jmp_near) {
+ p = strchr(fw, 'j');
+ while (p && *p != ' ')
+ p++;
+ while (p && *p == ' ')
+ p++;
+ if (p != NULL) {
+ memmove(p + 9, p, strlen(p) + 1);
+ memcpy(p, "near ptr ", 9);
+ }
+ }
+ if (word_imm) {
+ p = strstr(fw, ", ");
+ if (p != NULL && '0' <= p[2] && p[2] <= '9') {
+ p += 2;
+ memmove(p + 9, p, strlen(p) + 1);
+ memcpy(p, "word ptr ", 9);
+ }
+ }
+ else if (dword_imm) {
+ p = strstr(fw, ", ");
+ if (p != NULL && '0' <= p[2] && p[2] <= '9') {
+ p += 2;
+ memmove(p + 10, p, strlen(p) + 1);
+ memcpy(p, "dword ptr ", 10);
+ }
+ }
+ else if (do_pushf) {
+ p = strstr(fw, "pushf");
+ if (p == NULL)
+ p = strstr(fw, "popf");
+ if (p != NULL) {
+ p = strchr(p, 'f') + 1;
+ memmove(p + 1, p, strlen(p) + 1);
+ *p = 'd';
+ }
+ }
- fout_line++;
- qfprintf(fout, "%s\n", buf);
+ if (fw[0] == 'a' && IS_START(fw, "assume cs")) {
+ // "assume cs" causes problems with ext syms
+ memmove(fw + 1, fw, strlen(fw) + 1);
+ *fw = ';';
+ }
+ else if (fw[0] == 'e' && IS_START(fw, "end") && fw[3] == ' ') {
+ fout_line++;
+ qfprintf(fout, "include public.inc\n\n");
+
+ // kill entry point
+ fw[3] = 0;
+ }
+
+ fout_line++;
+ qfprintf(fout, "%s\n", buf);
+ }
// note: next_head skips some undefined stuff
ea = next_not_tail(ea); // correct?
pl.ea = ea;
pl.lnnum = 0;
ln.set_place(&pl);
- n = ln.get_linecnt();
- for (i = 0; i < n - 1; i++)
- {
- fout_line++;
- do_def_line(buf, sizeof(buf), ln.down());
- qfprintf(fout, "%s\n", buf);
- }
}
if (fout != NULL)