+/*
+ * ia32rtools
+ * (C) notaz, 2013,2014
+ *
+ * This work is licensed under the terms of 3-clause BSD license.
+ * See COPYING file in the top-level directory.
+ */
+
#define NO_OBSOLETE_FUNCS
#include <ida.hpp>
#include <idp.hpp>
#include <name.hpp>
#include <frame.hpp>
#include <struct.hpp>
+#include <offset.hpp>
#include <auto.hpp>
#include <intel.hpp>
#define IS_START(w, y) !strncmp(w, y, strlen(y))
#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
+static char **name_cache;
+static size_t name_cache_size;
+
// non-local branch targets
static ea_t *nonlocal_bt;
static int nonlocal_bt_alloc;
//--------------------------------------------------------------------------
static void idaapi term(void)
{
+ size_t i;
+
if (nonlocal_bt != NULL) {
free(nonlocal_bt);
nonlocal_bt = NULL;
}
nonlocal_bt_alloc = 0;
+
+ if (name_cache != NULL) {
+ for (i = 0; i < name_cache_size; i++)
+ free(name_cache[i]);
+ free(name_cache);
+ name_cache = NULL;
+ }
+ name_cache_size = 0;
}
//--------------------------------------------------------------------------
"type",
"offset",
"aam",
+ "text",
+ "size",
+ "c",
+ "align",
};
static int is_name_reserved(const char *name)
return itype == NN_jmp || (NN_ja <= itype && itype <= NN_jz);
}
-static void do_def_line(char *buf, size_t buf_size, const char *line)
+static void do_def_line(char *buf, size_t buf_size, const char *line,
+ ea_t ea)
{
- char *endp = NULL;
- ea_t ea, *ea_ret;
+ 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] = ':';
+ }
+ }
+}
+
+static int name_cache_cmp(const void *p1, const void *p2)
+{
+ // masm ignores case, so do we
+ return stricmp(*(char * const *)p1, *(char * const *)p2);
+}
+
+static void rebuild_name_cache(void)
+{
+ size_t i, newsize;
+ void *tmp;
+
+ // build a sorted name cache
+ newsize = get_nlist_size();
+ if (newsize > name_cache_size) {
+ tmp = realloc(name_cache, newsize * sizeof(name_cache[0]));
+ if (tmp == NULL) {
+ msg("OOM for name cache\n");
+ return;
}
+ name_cache = (char **)tmp;
+ }
+ for (i = 0; i < name_cache_size; i++)
+ free(name_cache[i]);
+ for (i = 0; i < newsize; i++)
+ name_cache[i] = strdup(get_nlist_name(i));
+
+ name_cache_size = newsize;
+ qsort(name_cache, name_cache_size, sizeof(name_cache[0]),
+ name_cache_cmp);
+}
+
+static void my_rename(ea_t ea, char *name)
+{
+ char buf[256];
+ char *p, **pp;
+ int n = 0;
+
+ qsnprintf(buf, sizeof(buf), "%s", name);
+ do {
+ p = buf;
+ pp = (char **)bsearch(&p, name_cache, name_cache_size,
+ sizeof(name_cache[0]), name_cache_cmp);
+ if (pp == NULL)
+ break;
+
+ qsnprintf(buf, sizeof(buf), "%s_g%d", name, n);
+ n++;
+ }
+ while (n < 100);
+
+ if (n == 100)
+ msg("rename failure? '%s'\n", name);
+
+ do_name_anyway(ea, buf);
+ rebuild_name_cache();
+}
+
+static void make_align(ea_t ea)
+{
+ ea_t tmp_ea;
+ int n;
+
+ 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);
}
}
{
// isEnabled(ea) // address belongs to disassembly
// ea_t ea = get_screen_ea();
+ // extern foo;
// foo = DecodeInstruction(ScreenEA());
FILE *fout = NULL;
int fout_line = 0;
uval_t idx;
int i, o, m, n;
int ret;
+ char **pp;
char *p;
nonlocal_bt_cnt = 0;
idx = get_first_struc_idx();
}
+ rebuild_name_cache();
+
// 1st pass: walk through all funcs
- func = get_func(inf.minEA);
+ ea = inf.minEA;
+ func = get_func(ea);
while (func != NULL)
{
func_tail_iterator_t fti(func);
}
}
- tmp_ea = get_name_ea(ea, buf);
- if (tmp_ea == 0 || tmp_ea == ~0)
+ p = buf;
+ pp = (char **)bsearch(&p, name_cache, name_cache_size,
+ sizeof(name_cache[0]), name_cache_cmp);
+ if (pp == NULL)
continue;
- msg("%x: from %x: renaming '%s'\n", tmp_ea, ea, buf);
- qstrncat(buf, "_g", sizeof(buf));
- set_name(tmp_ea, buf);
+ tmp_ea = get_name_ea(BADADDR, *pp);
+ msg("%x: renaming '%s' because of '%s' at %x\n",
+ tmp_ea, *pp, buf, ea);
+ my_rename(tmp_ea, *pp);
}
}
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)) {
+ // ..but base float is ok..
+ int is_flt = isDwrd(tmp_ea_flags) || isFloat(tmp_ea_flags);
+ if (!is_flt && !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);
}
}
}
- // detect code alignment
else if (cmd.itype == NN_lea) {
+ // detect code alignment
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);
+ // lea eax, [eax+0]
+ make_align(ea);
+ }
+ else if (!isDefArg1(ea_flags)
+ && cmd.Operands[1].type == o_mem // why o_mem?
+ && cmd.Operands[1].dtyp == dt_dword)
+ {
+ if (inf.minEA <= cmd.Operands[1].addr
+ && cmd.Operands[1].addr < inf.maxEA)
+ {
+ // lea to segments, like ds:58D6A8h[edx*8]
+ msg("%x: lea offset to %x\n", ea, cmd.Operands[1].addr);
+ op_offset(ea, 1, REF_OFF32);
+ }
+ else
+ {
+ // ds:0[eax*8] -> [eax*8+0]
+ msg("%x: dropping ds: for %x\n", ea, cmd.Operands[1].addr);
+ op_hex(ea, 1);
}
}
}
+ else if (cmd.itype == NN_mov && cmd.segpref == 0x1e // 2e?
+ && cmd.Operands[0].type == o_reg
+ && cmd.Operands[1].type == o_reg
+ && cmd.Operands[0].dtyp == cmd.Operands[1].dtyp
+ && cmd.Operands[0].reg == cmd.Operands[1].reg)
+ {
+ // db 2Eh; mov eax, eax
+ make_align(ea);
+ }
// find non-local branches
if (is_insn_jmp(cmd.itype) && cmd.Operands[0].type == o_near)
continue;
}
- if (is_name_reserved(name)) {
+ // 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);
- set_name(ea, buf);
+ qsnprintf(buf, sizeof(buf), "%s", 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);
+ }
+ }
+ }
+
+ my_rename(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 (;;)
{
- int drop_large = 0, drop_rva = 0, set_scale = 0, jmp_near = 0;
- int word_imm = 0, dword_imm = 0, do_pushf = 0;
+ int drop_large = 0, do_rva = 0, set_scale = 0, jmp_near = 0;
+ int word_imm = 0, dword_imm = 0, do_pushf = 0, do_nops = 0;
if ((ea >> 14) != ui_ea_block) {
ui_ea_block = ea >> 14;
if (get_word(ea + opr.offb) == (ushort)opr.value)
word_imm = 1;
}
+ else if (opr.type == o_displ && opr.addr == 0
+ && opr.offb != 0 && opr.hasSIB && opr.sib == 0x24)
+ {
+ // uses [esp+0] with 0 encoded into op
+ do_nops++;
+ }
}
}
else { // not code
if (isOff0(ea_flags))
- drop_rva = 1;
+ do_rva = 1;
}
pass:
n = ln.get_linecnt();
for (i = pl.lnnum; i < n; i++) {
- do_def_line(buf, sizeof(buf), ln.down());
+ do_def_line(buf, sizeof(buf), ln.down(), ea);
+
+ char *fw;
+ for (fw = buf; *fw != 0 && *fw == ' '; )
+ fw++;
// patches..
if (drop_large) {
- p = strstr(buf, "large ");
+ p = strstr(fw, "large ");
if (p != NULL)
memmove(p, p + 6, strlen(p + 6) + 1);
}
- while (drop_rva) {
- p = strstr(buf, " rva ");
+ while (do_rva) {
+ p = strstr(fw, " rva ");
if (p == NULL)
break;
- memmove(p, p + 4, strlen(p + 4) + 1);
+ memmove(p + 4 + 3, p + 4, strlen(p + 4) + 1);
+ memcpy(p + 1, "offset", 6);
}
if (set_scale) {
- p = strchr(buf, '[');
+ p = strchr(fw, '[');
if (p != NULL)
p = strchr(p, '+');
if (p != NULL && p[1] == 'e') {
}
}
else if (jmp_near) {
- p = strchr(buf, 'j');
+ p = NULL;
+ if (fw != buf && fw[0] == 'j')
+ p = fw;
while (p && *p != ' ')
p++;
while (p && *p == ' ')
if (p != NULL) {
memmove(p + 9, p, strlen(p) + 1);
memcpy(p, "near ptr ", 9);
+ jmp_near = 0;
}
}
if (word_imm) {
- p = strstr(buf, ", ");
+ p = strstr(fw, ", ");
if (p != NULL && '0' <= p[2] && p[2] <= '9') {
p += 2;
memmove(p + 9, p, strlen(p) + 1);
}
}
else if (dword_imm) {
- p = strstr(buf, ", ");
+ p = strstr(fw, ", ");
if (p != NULL && '0' <= p[2] && p[2] <= '9') {
p += 2;
memmove(p + 10, p, strlen(p) + 1);
}
}
else if (do_pushf) {
- p = strstr(buf, "pushf");
+ p = strstr(fw, "pushf");
if (p == NULL)
- p = strstr(buf, "popf");
+ p = strstr(fw, "popf");
if (p != NULL) {
p = strchr(p, 'f') + 1;
memmove(p + 1, p, strlen(p) + 1);
}
}
+ 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);
}
+ while (do_nops-- > 0)
+ qfprintf(fout, " nop ; adj\n");
+
// note: next_head skips some undefined stuff
ea = next_not_tail(ea); // correct?
if (ea == BADADDR)