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 */
};
enum op_op {
OP_NOT,
OP_CDQ,
OP_STOS,
+ OP_MOVS,
OP_RET,
OP_ADD,
OP_SUB,
OP_SHL,
OP_SHR,
OP_SAR,
+ OP_ROL,
+ OP_ROR,
OP_ADC,
OP_SBB,
OP_INC,
int offset;
};
-#define MAX_OPS 1024
+enum ida_func_attr {
+ IDAFA_BP_FRAME = (1 << 0),
+ IDAFA_LIB_FUNC = (1 << 1),
+ IDAFA_STATIC = (1 << 2),
+ IDAFA_NORETURN = (1 << 3),
+ IDAFA_THUNK = (1 << 4),
+ IDAFA_FPD = (1 << 5),
+};
+
+#define MAX_OPS 4096
static struct parsed_op ops[MAX_OPS];
static struct parsed_equ *g_eqs;
static char g_func[256];
static char g_comment[256];
static int g_bp_frame;
-static int g_bp_stack;
+static int g_sp_frame;
+static int g_stack_fsz;
#define ferr(op_, fmt, ...) do { \
printf("error:%s:#%ld: '%s': " fmt, g_func, (op_) - ops, \
dump_op(op_), ##__VA_ARGS__); \
return c;
}
+static const char *parse_stack_el(const char *name)
+{
+ const char *p, *s;
+ char *endp = NULL;
+ char buf[32];
+ long val;
+ int len;
+
+ if (!strncmp(name, "ebp+", 4)
+ && !('0' <= name[4] && name[4] <= '9'))
+ {
+ return name + 4;
+ }
+ if (strncmp(name, "esp+", 4) != 0)
+ return NULL;
+
+ p = strchr(name + 4, '+');
+ if (p) {
+ // must be a number after esp+, already converted to 0x..
+ s = name + 4;
+ if (!('0' <= *s && *s <= '9')) {
+ aerr("%s nan?\n", __func__);
+ return NULL;
+ }
+ if (s[0] == '0' && s[1] == 'x')
+ s += 2;
+ len = p - s;
+ if (len < sizeof(buf) - 1) {
+ strncpy(buf, s, len);
+ buf[len] = 0;
+ val = strtol(buf, &endp, 16);
+ if (val == 0 || *endp != 0) {
+ aerr("%s num parse fail for '%s'\n", __func__, buf);
+ return NULL;
+ }
+ }
+ p++;
+ }
+ else
+ p = name + 4;
+
+ if ('0' <= *p && *p <= '9')
+ return NULL;
+
+ return p;
+}
+
static int guess_lmod_from_name(struct parsed_opr *opr)
{
if (!strncmp(opr->name, "dword_", 6)) {
static int parse_operand(struct parsed_opr *opr,
int *regmask, int *regmask_indirect,
- char words[16][256], int wordc, int w, unsigned int op_flags)
+ char words[16][256], int wordc, int w, unsigned int op_flags)
{
enum opr_lenmod tmplmod;
int ret, len;
long number;
+ int wordc_in;
int i;
- if (w >= wordc)
- aerr("parse_operand w %d, wordc %d\n", w, wordc);
+ if (w >= wordc)
+ aerr("parse_operand w %d, wordc %d\n", w, wordc);
- opr->reg = xUNSPEC;
+ opr->reg = xUNSPEC;
- for (i = w; i < wordc; i++) {
- len = strlen(words[i]);
- if (words[i][len - 1] == ',') {
- words[i][len - 1] = 0;
- wordc = i + 1;
- break;
- }
- }
+ for (i = w; i < wordc; i++) {
+ len = strlen(words[i]);
+ if (words[i][len - 1] == ',') {
+ words[i][len - 1] = 0;
+ wordc = i + 1;
+ break;
+ }
+ }
- if (op_flags & OPF_JMP) {
- const char *label;
-
- if (wordc - w == 3 && IS(words[w + 1], "ptr"))
- label = words[w + 2];
- else if (wordc - w == 2 && IS(words[w], "short"))
- label = words[w + 1];
- else if (wordc - w == 1)
- label = words[w];
- else
- aerr("jump parse error");
-
- opr->type = OPT_LABEL;
- strcpy(opr->name, label);
- return wordc;
- }
+ wordc_in = wordc - w;
- if (wordc - w >= 3) {
- if (IS(words[w + 1], "ptr")) {
- if (IS(words[w], "dword"))
- opr->lmod = OPLM_DWORD;
- else if (IS(words[w], "word"))
- opr->lmod = OPLM_WORD;
- else if (IS(words[w], "byte"))
- opr->lmod = OPLM_BYTE;
- else
- aerr("type parsing failed\n");
- w += 2;
- }
- }
+ if ((op_flags & OPF_JMP) && wordc_in > 0
+ && !('0' <= words[w][0] && words[w][0] <= '9'))
+ {
+ const char *label = NULL;
+
+ if (wordc_in == 3 && !strncmp(words[w], "near", 4)
+ && IS(words[w + 1], "ptr"))
+ label = words[w + 2];
+ else if (wordc_in == 2 && IS(words[w], "short"))
+ label = words[w + 1];
+ else if (wordc_in == 1
+ && strchr(words[w], '[') == NULL
+ && parse_reg(&tmplmod, words[w]) < 0)
+ label = words[w];
+
+ if (label != NULL) {
+ opr->type = OPT_LABEL;
+ strcpy(opr->name, label);
+ return wordc;
+ }
+ }
- if (wordc - w == 2 && IS(words[w], "offset")) {
- opr->type = OPT_OFFSET;
- strcpy(opr->name, words[w + 1]);
- return wordc;
- }
+ if (wordc_in >= 3) {
+ if (IS(words[w + 1], "ptr")) {
+ if (IS(words[w], "dword"))
+ opr->lmod = OPLM_DWORD;
+ else if (IS(words[w], "word"))
+ opr->lmod = OPLM_WORD;
+ else if (IS(words[w], "byte"))
+ opr->lmod = OPLM_BYTE;
+ else
+ aerr("type parsing failed\n");
+ w += 2;
+ wordc_in = wordc - w;
+ }
+ }
- if (wordc - w != 1)
+ if (wordc_in == 2 && IS(words[w], "offset")) {
+ opr->type = OPT_OFFSET;
+ strcpy(opr->name, words[w + 1]);
+ return wordc;
+ }
+
+ if (wordc_in != 1)
aerr("parse_operand 1 word expected\n");
strcpy(opr->name, words[w]);
aerr("[] parse failure\n");
parse_indmode(opr->name, regmask_indirect, 1);
- if (opr->lmod == OPLM_UNSPEC && !strncmp(opr->name, "ebp+", 4)
- && !('0' <= opr->name[4] && opr->name[4] <= '9'))
- {
+ if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name)) {
// might be an equ
- struct parsed_equ *eq = equ_find(NULL, opr->name + 4);
+ struct parsed_equ *eq =
+ equ_find(NULL, parse_stack_el(opr->name));
if (eq)
opr->lmod = eq->lmod;
}
{ "stosb",OP_STOS, 0, 0, OPF_DATA },
{ "stosw",OP_STOS, 0, 0, OPF_DATA },
{ "stosd",OP_STOS, 0, 0, OPF_DATA },
+ { "movsb",OP_MOVS, 0, 0, OPF_DATA },
+ { "movsw",OP_MOVS, 0, 0, OPF_DATA },
+ { "movsd",OP_MOVS, 0, 0, OPF_DATA },
{ "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 },
{ "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
{ "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
{ "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
+ { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
+ { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
{ "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
{ "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
{ "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
else if (IS(words[op_w], "stosd"))
lmod = OPLM_DWORD;
op->operand_cnt = 3;
- setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_dst);
- setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_dst);
+ setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
+ setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
+ op->regmask_dst = op->regmask_src;
setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src);
break;
+ case OP_MOVS:
+ if (op->operand_cnt != 0)
+ break;
+ if (IS(words[op_w], "movsb"))
+ lmod = OPLM_BYTE;
+ else if (IS(words[op_w], "movsw"))
+ lmod = OPLM_WORD;
+ else if (IS(words[op_w], "movsd"))
+ lmod = OPLM_DWORD;
+ op->operand_cnt = 3;
+ setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
+ setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
+ setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
+ op->regmask_dst = op->regmask_src;
+ break;
+
case OP_IMUL:
if (op->operand_cnt != 1)
break;
case OP_SHL:
case OP_SHR:
case OP_SAR:
+ case OP_ROL:
+ case OP_ROR:
if (op->operand[1].lmod == OPLM_UNSPEC)
op->operand[1].lmod = OPLM_BYTE;
break;
return out;
}
+static const char *lmod_cast_u(struct parsed_op *po,
+ enum opr_lenmod lmod)
+{
+ switch (lmod) {
+ case OPLM_DWORD:
+ return "";
+ 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_ptr(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_s(struct parsed_op *po,
+ enum opr_lenmod lmod)
+{
+ switch (lmod) {
+ case OPLM_DWORD:
+ return "(s32)";
+ case OPLM_WORD:
+ return "(s16)";
+ case OPLM_BYTE:
+ return "(s8)";
+ default:
+ ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
+ return "(_invalid_)";
+ }
+}
+
+static const char *lmod_cast(struct parsed_op *po,
+ enum opr_lenmod lmod, int is_signed)
+{
+ return is_signed ?
+ lmod_cast_s(po, lmod) :
+ lmod_cast_u(po, lmod);
+}
+
+static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
+{
+ switch (lmod) {
+ case OPLM_DWORD:
+ return 4;
+ case OPLM_WORD:
+ return 2;
+ case OPLM_BYTE:
+ return 1;
+ default:
+ ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
+ return 0;
+ }
+}
+
static const char *opr_name(struct parsed_op *po, int opr_num)
{
if (opr_num >= po->operand_cnt)
return &g_eqs[i];
}
-static void bg_frame_access(struct parsed_op *po, enum opr_lenmod lmod,
- char *buf, size_t buf_size, const char *bp_arg,
- int is_src, int is_lea)
+static void stack_frame_access(struct parsed_op *po,
+ enum opr_lenmod lmod, char *buf, size_t buf_size,
+ const char *name, int is_src, int is_lea)
{
const char *prefix = "";
struct parsed_equ *eq;
int i, arg_i, arg_s;
+ const char *bp_arg;
+ int stack_ra = 0;
int sf_ofs;
+ bp_arg = parse_stack_el(name);
snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
-
eq = equ_find(po, bp_arg);
+ if (eq == NULL)
+ ferr(po, "detected but missing eq\n");
+
+ if (!strncmp(name, "ebp", 3))
+ stack_ra = 4;
+
+ if (stack_ra <= eq->offset && eq->offset < stack_ra + 4)
+ ferr(po, "reference to ra? %d %d\n", eq->offset, stack_ra);
- if (eq->offset >= 0) {
- arg_i = eq->offset / 4 - 2;
+ if (eq->offset > stack_ra) {
+ arg_i = (eq->offset - stack_ra - 4) / 4;
if (arg_i < 0 || arg_i >= g_func_pp.argc_stack)
- ferr(po, "offset %d doesn't map to any arg\n", eq->offset);
+ ferr(po, "offset %d (%s) doesn't map to any arg\n",
+ eq->offset, bp_arg);
for (i = arg_s = 0; i < g_func_pp.argc; i++) {
if (g_func_pp.arg[i].reg != NULL)
snprintf(buf, buf_size, "%sa%d", is_src ? "(u32)" : "", i + 1);
}
else {
- if (g_bp_stack == 0)
- ferr(po, "bp_stack access after it was not detected\n");
+ if (g_stack_fsz == 0)
+ ferr(po, "stack var access without stackframe\n");
- sf_ofs = g_bp_stack + eq->offset;
+ sf_ofs = g_stack_fsz + eq->offset;
if (sf_ofs < 0)
- ferr(po, "bp_stack offset %d/%d\n", eq->offset, g_bp_stack);
+ ferr(po, "bp_stack offset %d/%d\n", eq->offset, g_stack_fsz);
if (is_lea)
prefix = "(u32)&";
static char *out_src_opr(char *buf, size_t buf_size,
struct parsed_op *po, struct parsed_opr *popr, int is_lea)
{
- const char *cast = "";
char tmp1[256], tmp2[256];
char expr[256];
int ret;
break;
case OPT_REGMEM:
- if (g_bp_frame && !strncmp(popr->name, "ebp+", 4)
- && !('0' <= popr->name[4] && popr->name[4] <= '9'))
- {
- bg_frame_access(po, popr->lmod, buf, buf_size,
- popr->name + 4, 1, is_lea);
+ if (parse_stack_el(popr->name)) {
+ stack_frame_access(po, popr->lmod, buf, buf_size,
+ popr->name, 1, is_lea);
break;
}
break;
}
- switch (popr->lmod) {
- case OPLM_DWORD:
- cast = "*(u32 *)";
- break;
- case OPLM_WORD:
- cast = "*(u16 *)";
- break;
- case OPLM_BYTE:
- cast = "*(u8 *)";
- break;
- default:
- ferr(po, "invalid lmod: %d\n", popr->lmod);
- }
- snprintf(buf, buf_size, "%s(%s)", cast, expr);
+ snprintf(buf, buf_size, "%s(%s)",
+ lmod_cast_u_ptr(po, popr->lmod), expr);
break;
case OPT_LABEL:
break;
case OPT_REGMEM:
- if (g_bp_frame && !strncmp(popr->name, "ebp+", 4)
- && !('0' <= popr->name[4] && popr->name[4] <= '9'))
- {
- bg_frame_access(po, popr->lmod, buf, buf_size,
- popr->name + 4, 0, 0);
+ if (parse_stack_el(popr->name)) {
+ stack_frame_access(po, popr->lmod, buf, buf_size,
+ popr->name, 0, 0);
break;
}
return buf;
}
-static const char *lmod_cast_u(struct parsed_op *po,
- enum opr_lenmod lmod)
-{
- switch (lmod) {
- case OPLM_DWORD:
- return "";
- 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_s(struct parsed_op *po,
- enum opr_lenmod lmod)
-{
- switch (lmod) {
- case OPLM_DWORD:
- return "(s32)";
- case OPLM_WORD:
- return "(s16)";
- case OPLM_BYTE:
- return "(s8)";
- default:
- ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
- return "(_invalid_)";
- }
-}
-
-static const char *lmod_cast(struct parsed_op *po,
- enum opr_lenmod lmod, int is_signed)
-{
- return is_signed ?
- lmod_cast_s(po, lmod) :
- lmod_cast_u(po, lmod);
-}
-
-static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
-{
- switch (lmod) {
- case OPLM_DWORD:
- return 4;
- case OPLM_WORD:
- return 2;
- case OPLM_BYTE:
- return 1;
- default:
- ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
- return 0;
- }
-}
-
static enum parsed_flag_op split_cond(struct parsed_op *po,
enum op_op op, int *is_inv)
{
}
}
+static void set_flag_no_dup(struct parsed_op *po, enum op_flags flag,
+ enum op_flags flag_check)
+{
+ if (po->flags & flag)
+ ferr(po, "flag %x already set\n", flag);
+ if (po->flags & flag_check)
+ ferr(po, "flag_check %x already set\n", flag_check);
+
+ po->flags |= flag;
+}
+
static int scan_for_pop(int i, int opcnt, const char *reg,
- int magic, int do_patch)
+ int magic, int depth, int *maxdepth, int do_flags)
{
struct parsed_op *po;
int ret = 0;
if (po->flags & OPF_TAIL)
return -1; // deadend
- if (po->flags & OPF_RMD)
+ if ((po->flags & OPF_RMD)
+ || (po->op == OP_PUSH && po->argmask)) // arg push
continue;
if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
}
if (po->flags & OPF_CC) {
- ret |= scan_for_pop(po->bt_i, opcnt, reg, magic, do_patch);
+ ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
+ depth, maxdepth, do_flags);
if (ret < 0)
return ret; // dead end
}
continue;
}
- if (po->op == OP_POP && po->operand[0].type == OPT_REG
+ if ((po->op == OP_POP || po->op == OP_PUSH)
+ && po->operand[0].type == OPT_REG
&& IS(po->operand[0].name, reg))
{
- if (do_patch)
- po->flags |= OPF_RMD;
- return 1;
+ if (po->op == OP_PUSH) {
+ depth++;
+ if (depth > *maxdepth)
+ *maxdepth = depth;
+ if (do_flags)
+ set_flag_no_dup(po, OPF_RSAVE, OPF_RMD);
+ }
+ else if (depth == 0) {
+ if (do_flags)
+ set_flag_no_dup(po, OPF_RMD, OPF_RSAVE);
+ return 1;
+ }
+ else {
+ depth--;
+ if (depth < 0) // should not happen
+ ferr(po, "fail with depth\n");
+ if (do_flags)
+ set_flag_no_dup(po, OPF_RSAVE, OPF_RMD);
+ }
}
}
}
// scan for pop starting from 'ret' op (all paths)
-static int scan_for_pop_ret(int i, int opcnt, const char *reg, int do_patch)
+static int scan_for_pop_ret(int i, int opcnt, const char *reg,
+ int flag_set)
{
int found = 0;
int j;
&& IS(ops[j].operand[0].name, reg))
{
found = 1;
- if (do_patch)
- ops[j].flags |= OPF_RMD;
+ ops[j].flags |= flag_set;
break;
}
return -1;
}
+// scan for positive, constant esp adjust
+static int scan_for_esp_adjust(int i, int opcnt, int *adj)
+{
+ for (; i < opcnt; i++) {
+ if (ops[i].op == OP_ADD && ops[i].operand[0].reg == xSP) {
+ if (ops[i].operand[1].type != OPT_CONST)
+ ferr(&ops[i], "non-const esp adjust?\n");
+ *adj = ops[i].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)
+ return -1;
+ if (g_labels[i][0] != 0)
+ return -1;
+ }
+
+ return -1;
+}
+
static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
{
struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
int cmp_result_vars = 0;
int need_mul_var = 0;
int had_decl = 0;
+ int regmask_save = 0;
int regmask_arg = 0;
int regmask = 0;
int pfomask = 0;
+ int depth = 0;
int no_output;
int dummy;
int arg;
int reg;
int ret;
- g_bp_frame = g_bp_stack = 0;
+ g_bp_frame = g_sp_frame = g_stack_fsz = 0;
ret = proto_parse(fhdr, funcn, &g_func_pp);
if (ret)
fprintf(fout, ")\n{\n");
// pass1:
- // - handle ebp frame, remove ops related to it
+ // - handle ebp/esp frame, remove ops related to it
if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
&& ops[1].op == OP_MOV
&& IS(opr_name(&ops[1], 0), "ebp")
ops[1].flags |= OPF_RMD;
if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
- g_bp_stack = opr_const(&ops[2], 1);
+ g_stack_fsz = opr_const(&ops[2], 1);
ops[2].flags |= OPF_RMD;
}
else {
// another way msvc builds stack frame..
i = 2;
while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
- g_bp_stack += 4;
+ g_stack_fsz += 4;
ops[i].flags |= OPF_RMD;
ecx_push++;
i++;
ferr(&ops[i - 1], "'pop ebp' expected\n");
ops[i - 1].flags |= OPF_RMD;
- if (g_bp_stack != 0) {
+ if (g_stack_fsz != 0) {
if (ops[i - 2].op != OP_MOV
|| !IS(opr_name(&ops[i - 2], 0), "esp")
|| !IS(opr_name(&ops[i - 2], 1), "ebp"))
i++;
} while (i < opcnt);
}
+ else {
+ for (i = 0; i < opcnt; i++) {
+ if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
+ break;
+ if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
+ && ops[i].operand[1].type == OPT_CONST)
+ {
+ g_sp_frame = 1;
+ break;
+ }
+ }
+
+ if (g_sp_frame)
+ {
+ g_stack_fsz = ops[i].operand[1].val;
+ ops[i].flags |= OPF_RMD;
+
+ i++;
+ do {
+ for (; i < opcnt; i++)
+ if (ops[i].op == OP_RET)
+ break;
+ if (ops[i - 1].op != OP_ADD
+ || !IS(opr_name(&ops[i - 1], 0), "esp")
+ || ops[i - 1].operand[1].type != OPT_CONST
+ || ops[i - 1].operand[1].val != g_stack_fsz)
+ ferr(&ops[i - 1], "'add esp' expected\n");
+ ops[i - 1].flags |= OPF_RMD;
+
+ i++;
+ } while (i < opcnt);
+ }
+ }
// pass2:
// - resolve all branches
}
// pass3:
- // - find POPs for PUSHes, rm both
- // - scan for all used registers
- // - find flag set ops for their users
// - process calls
- for (i = 0; i < opcnt; i++) {
+ for (i = 0; i < opcnt; i++)
+ {
po = &ops[i];
if (po->flags & OPF_RMD)
continue;
- if (po->op == OP_PUSH && po->operand[0].type == OPT_REG) {
- if (po->operand[0].reg < 0)
- ferr(po, "reg not set for push?\n");
- if (!(regmask & (1 << po->operand[0].reg))) { // reg save
- ret = scan_for_pop(i + 1, opcnt,
- po->operand[0].name, i + opcnt, 0);
- if (ret == 1) {
- po->flags |= OPF_RMD;
- scan_for_pop(i + 1, opcnt, po->operand[0].name,
- i + opcnt * 2, 1);
- continue;
- }
- ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
- if (ret == 0) {
- po->flags |= OPF_RMD;
- scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 1);
- continue;
- }
- }
- }
-
- regmask |= po->regmask_src | po->regmask_dst;
-
- if (po->flags & OPF_CC)
+ if (po->op == OP_CALL)
{
- ret = scan_for_flag_set(i - 1);
- if (ret < 0)
- ferr(po, "unable to trace flag setter\n");
-
- tmp_op = &ops[ret]; // flag setter
- pfo = split_cond(po, po->op, &dummy);
- pfomask = 0;
-
- // to get nicer code, we try to delay test and cmp;
- // if we can't because of operand modification, or if we
- // have math op, make it calculate flags explicitly
- if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) {
- if (scan_for_mod(tmp_op, ret + 1, i) >= 0)
- pfomask = 1 << pfo;
+ pp = calloc(1, sizeof(*pp));
+ my_assert_not(pp, NULL);
+ tmpname = opr_name(po, 0);
+ if (po->operand[0].type != OPT_LABEL)
+ {
+ ret = scan_for_esp_adjust(i + 1, opcnt, &j);
+ if (ret < 0)
+ ferr(po, "non-__cdecl indirect call unhandled yet\n");
+ j /= 4;
+ if (j > ARRAY_SIZE(pp->arg))
+ ferr(po, "esp adjust too large?\n");
+ pp->ret_type = "int";
+ pp->argc = pp->argc_stack = j;
+ for (arg = 0; arg < pp->argc; arg++)
+ pp->arg[arg].type = "int";
}
else {
- if ((pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
- || scan_for_mod_opr0(tmp_op, ret + 1, i) >= 0)
- pfomask = 1 << pfo;
- }
- if (pfomask) {
- tmp_op->pfomask |= pfomask;
- cmp_result_vars |= pfomask;
- po->datap = tmp_op;
+ ret = proto_parse(fhdr, tmpname, pp);
+ if (ret)
+ ferr(po, "proto_parse failed for call '%s'\n", tmpname);
}
- if (po->op == OP_ADC || po->op == OP_SBB)
- cmp_result_vars |= 1 << PFO_C;
- }
- else if (po->op == OP_CALL)
- {
- pp = malloc(sizeof(*pp));
- my_assert_not(pp, NULL);
- tmpname = opr_name(&ops[i], 0);
- ret = proto_parse(fhdr, tmpname, pp);
- if (ret)
- ferr(po, "proto_parse failed for '%s'\n", tmpname);
+ ret = scan_for_esp_adjust(i + 1, opcnt, &j);
+ if (ret >= 0) {
+ if (pp->argc_stack != j / 4)
+ ferr(po, "stack tracking failed: %x %x\n",
+ pp->argc_stack, j);
+ ops[ret].flags |= OPF_RMD;
+ }
for (arg = 0; arg < pp->argc; arg++)
if (pp->arg[arg].reg == NULL)
break;
- for (j = i - 1; j >= 0 && arg < pp->argc; j--) {
- if (ops[j].op == OP_CALL) {
+ for (j = i; j >= 0 && arg < pp->argc; )
+ {
+ if (g_labels[j][0] != 0) {
+ if (j > 0 && ((ops[j - 1].flags & OPF_TAIL)
+ || (ops[j - 1].flags & (OPF_JMP|OPF_CC)) == OPF_JMP))
+ {
+ // follow the branch in reverse
+ if (g_label_refs[j] == NULL)
+ 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;
+ continue;
+ }
+ break;
+ }
+ j--;
+
+ if (ops[j].op == OP_CALL)
+ {
pp_tmp = ops[j].datap;
if (pp_tmp == NULL)
ferr(po, "arg collect hit unparsed call\n");
{
break;
}
-
- if (ops[j].op == OP_PUSH) {
+ else if (ops[j].op == OP_PUSH)
+ {
pp->arg[arg].datap = &ops[j];
ret = scan_for_mod(&ops[j], j + 1, i);
if (ret >= 0) {
// mark this push as one that needs operand saving
+ ops[j].flags &= ~OPF_RMD;
ops[j].argmask |= 1 << arg;
save_arg_vars |= 1 << arg;
}
if (pp->arg[arg].reg == NULL)
break;
}
-
- if (g_labels[j][0] != 0) {
- if (j > 0 && ((ops[j - 1].flags & OPF_TAIL)
- || (ops[j - 1].flags & (OPF_JMP|OPF_CC)) == OPF_JMP))
- {
- // follow the branch in reverse
- if (g_label_refs[j] == NULL)
- 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;
- continue;
- }
- break;
- }
}
if (arg < pp->argc)
ferr(po, "arg collect failed for '%s'\n", tmpname);
po->datap = pp;
}
+ }
+
+ // pass4:
+ // - find POPs for PUSHes, rm both
+ // - scan for all used registers
+ // - find flag set ops for their users
+ // - declare indirect functions
+ for (i = 0; i < opcnt; i++) {
+ po = &ops[i];
+ if (po->flags & OPF_RMD)
+ continue;
+
+ if (po->op == OP_PUSH
+ && po->argmask == 0 && !(po->flags & OPF_RSAVE)
+ && po->operand[0].type == OPT_REG)
+ {
+ reg = po->operand[0].reg;
+ if (reg < 0)
+ ferr(po, "reg not set for push?\n");
+
+ depth = 0;
+ ret = scan_for_pop(i + 1, opcnt,
+ po->operand[0].name, i + opcnt, 0, &depth, 0);
+ if (ret == 1) {
+ if (depth > 1)
+ ferr(po, "too much depth: %d\n", depth);
+ if (depth > 0)
+ regmask_save |= 1 << reg;
+
+ po->flags |= OPF_RMD;
+ scan_for_pop(i + 1, opcnt, po->operand[0].name,
+ i + opcnt * 2, 0, &depth, 1);
+ continue;
+ }
+ ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
+ if (ret == 0) {
+ arg = OPF_RMD;
+ if (regmask & (1 << reg)) {
+ if (regmask_save & (1 << reg))
+ ferr(po, "%s already saved?\n", po->operand[0].name);
+ arg = OPF_RSAVE;
+ }
+ po->flags |= arg;
+ scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
+ continue;
+ }
+ }
+
+ regmask |= po->regmask_src | po->regmask_dst;
+
+ if (po->flags & OPF_CC)
+ {
+ ret = scan_for_flag_set(i - 1);
+ if (ret < 0)
+ ferr(po, "unable to trace flag setter\n");
+
+ tmp_op = &ops[ret]; // flag setter
+ pfo = split_cond(po, po->op, &dummy);
+ pfomask = 0;
+
+ // to get nicer code, we try to delay test and cmp;
+ // if we can't because of operand modification, or if we
+ // have math op, make it calculate flags explicitly
+ if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) {
+ if (scan_for_mod(tmp_op, ret + 1, i) >= 0)
+ pfomask = 1 << pfo;
+ }
+ else {
+ if ((pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
+ || scan_for_mod_opr0(tmp_op, ret + 1, i) >= 0)
+ pfomask = 1 << pfo;
+ }
+ if (pfomask) {
+ tmp_op->pfomask |= pfomask;
+ cmp_result_vars |= pfomask;
+ po->datap = tmp_op;
+ }
+
+ if (po->op == OP_ADC || po->op == OP_SBB)
+ cmp_result_vars |= 1 << PFO_C;
+ }
else if (po->op == OP_MUL
|| (po->op == OP_IMUL && po->operand_cnt == 1))
{
need_mul_var = 1;
}
+ else if (po->op == OP_CALL && po->operand[0].type != OPT_LABEL) {
+ pp = po->datap;
+ my_assert_not(pp, NULL);
+ fprintf(fout, " %s (*icall%d)(", pp->ret_type, i);
+ for (j = 0; j < pp->argc; j++) {
+ if (j > 0)
+ fprintf(fout, ", ");
+ fprintf(fout, "%s a%d", pp->arg[j].type, j + 1);
+ }
+ fprintf(fout, ");\n");
+ }
}
// declare stack frame
- if (g_bp_stack)
+ if (g_stack_fsz)
fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
- (g_bp_stack + 3) / 4, (g_bp_stack + 1) / 2, g_bp_stack);
+ (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
// declare arg-registers
for (i = 0; i < g_func_pp.argc; i++) {
}
regmask &= ~regmask_arg;
+ regmask &= ~(1 << xSP);
if (g_bp_frame)
regmask &= ~(1 << xBP);
if (regmask) {
}
}
+ if (regmask_save) {
+ for (reg = 0; reg < 8; reg++) {
+ if (regmask_save & (1 << reg)) {
+ fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
+ had_decl = 1;
+ }
+ }
+ }
+
if (save_arg_vars) {
for (reg = 0; reg < 32; reg++) {
if (save_arg_vars & (1 << reg)) {
// output ops
for (i = 0; i < opcnt; i++)
{
- if (g_labels[i][0] != 0)
+ if (g_labels[i][0] != 0 && g_label_refs[i] != NULL)
fprintf(fout, "\n%s:\n", g_labels[i]);
po = &ops[i];
if (po->flags & OPF_REP) {
fprintf(fout, " for (; ecx != 0; ecx--, edi += %d)\n",
lmod_bytes(po, po->operand[0].lmod));
- fprintf(fout, " *(u32 *)edi = eax;");
+ fprintf(fout, " %sedi = eax;",
+ lmod_cast_u_ptr(po, po->operand[0].lmod));
strcpy(g_comment, "rep stos");
}
else {
- fprintf(fout, " *(u32 *)edi = eax; edi += %d;",
+ fprintf(fout, " %sedi = eax; edi += %d;",
+ lmod_cast_u_ptr(po, po->operand[0].lmod),
lmod_bytes(po, po->operand[0].lmod));
strcpy(g_comment, "stos");
}
break;
+ case OP_MOVS:
+ // assumes DF=0
+ 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,
+ " %sedi = %sesi;", buf1, buf1);
+ strcpy(g_comment, "rep movs");
+ }
+ else {
+ fprintf(fout, " %sedi = %sesi; edi += %d; esi += %d;",
+ buf1, buf1, j, j);
+ strcpy(g_comment, "movs");
+ }
+ break;
+
// arithmetic w/flags
case OP_ADD:
case OP_SUB:
delayed_flag_op = NULL;
break;
+ case OP_SAR:
+ assert_operand_cnt(2);
+ out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
+ fprintf(fout, " %s = %s%s >> %s;", buf1,
+ lmod_cast_s(po, po->operand[0].lmod), buf1,
+ out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0));
+ last_arith_dst = &po->operand[0];
+ delayed_flag_op = NULL;
+ break;
+
+ case OP_ROL:
+ case OP_ROR:
+ assert_operand_cnt(2);
+ out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
+ if (po->operand[1].type == OPT_CONST) {
+ j = po->operand[1].val;
+ j %= lmod_bytes(po, po->operand[0].lmod) * 8;
+ fprintf(fout, po->op == OP_ROL ?
+ " %s = (%s << %d) | (%s >> %d);" :
+ " %s = (%s >> %d) | (%s << %d);",
+ buf1, buf1, j, buf1,
+ lmod_bytes(po, po->operand[0].lmod) * 8 - j);
+ }
+ else
+ ferr(po, "TODO\n");
+ last_arith_dst = &po->operand[0];
+ delayed_flag_op = NULL;
+ break;
+
case OP_XOR:
assert_operand_cnt(2);
propagate_lmod(po, &po->operand[0], &po->operand[1]);
}
goto dualop_arith;
- case OP_SAR:
- assert_operand_cnt(2);
- out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
- fprintf(fout, " %s = %s%s >> %s;", buf1,
- lmod_cast_s(po, po->operand[0].lmod), buf1,
- out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0));
- last_arith_dst = &po->operand[0];
- delayed_flag_op = NULL;
- break;
-
case OP_ADC:
case OP_SBB:
assert_operand_cnt(2);
case OP_INC:
case OP_DEC:
out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
- strcpy(buf2, po->op == OP_INC ? "++" : "--");
- fprintf(fout, " %s%s;", buf1, buf2);
+ if (po->operand[0].type == OPT_REG) {
+ strcpy(buf2, po->op == OP_INC ? "++" : "--");
+ fprintf(fout, " %s%s;", buf1, buf2);
+ }
+ else {
+ strcpy(buf2, po->op == OP_INC ? "+" : "-");
+ fprintf(fout, " %s %s= 1;", buf1, buf2);
+ }
last_arith_dst = &po->operand[0];
delayed_flag_op = NULL;
break;
case OP_CALL:
assert_operand_cnt(1);
- if (po->operand[0].type != OPT_LABEL)
- ferr(po, "unhandled call type\n");
-
pp = po->datap;
if (pp == NULL)
ferr(po, "NULL pp\n");
+ if (po->operand[0].type != OPT_LABEL)
+ fprintf(fout, " icall%d = (void *)%s;\n", i,
+ out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0));
+
fprintf(fout, " ");
if (!IS(pp->ret_type, "void")) {
if (po->flags & OPF_TAIL)
fprintf(fout, "(u32)");
}
- fprintf(fout, "%s(", opr_name(po, 0));
+ if (po->operand[0].type != OPT_LABEL)
+ fprintf(fout, "icall%d(", i);
+ else
+ fprintf(fout, "%s(", opr_name(po, 0));
for (arg = 0; arg < pp->argc; arg++) {
if (arg > 0)
fprintf(fout, ", ");
case OP_RET:
if (IS(g_func_pp.ret_type, "void"))
fprintf(fout, " return;");
+ else if (strchr(g_func_pp.ret_type, '*'))
+ fprintf(fout, " return (%s)eax;",
+ g_func_pp.ret_type);
else
fprintf(fout, " return eax;");
break;
case OP_PUSH:
if (po->argmask) {
// special case - saved func arg
+ out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0);
for (j = 0; j < 32; j++) {
- if (po->argmask & (1 << j)) {
- fprintf(fout, " s_a%d = %s;", j + 1,
- out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0));
- }
+ if (po->argmask & (1 << j))
+ fprintf(fout, " s_a%d = %s;", j + 1, buf1);
}
break;
}
- ferr(po, "push encountered\n");
+ else if (po->flags & OPF_RSAVE) {
+ out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0);
+ fprintf(fout, " s_%s = %s;", buf1, buf1);
+ break;
+ }
+ ferr(po, "stray push encountered\n");
break;
case OP_POP:
- ferr(po, "pop encountered\n");
+ if (po->flags & OPF_RSAVE) {
+ out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0);
+ fprintf(fout, " %s = s_%s;", buf1, buf1);
+ break;
+ }
+ ferr(po, "stray pop encountered\n");
break;
case OP_NOP:
proto_release(&g_func_pp);
}
+static void set_label(int i, const char *name)
+{
+ const char *p;
+ int len;
+
+ len = strlen(name);
+ p = strchr(name, ':');
+ if (p != NULL)
+ len = p - name;
+
+ if (len > sizeof(g_labels[0]) - 1)
+ aerr("label '%s' too long: %d\n", name, len);
+ if (g_labels[i][0] != 0)
+ aerr("dupe label '%s'?\n", name);
+ memcpy(g_labels[i], name, len);
+ g_labels[i][len] = 0;
+}
+
// '=' needs special treatment..
static char *next_word_s(char *w, size_t wsize, char *s)
{
FILE *fout, *fasm, *fhdr, *frlist;
char line[256];
char words[16][256];
+ int ida_func_attr = 0;
int in_func = 0;
int skip_func = 0;
int skip_warned = 0;
int arg_out;
int arg = 1;
int pi = 0;
- int len;
+ int i, len;
char *p;
int wordc;
asmln++;
p = sskip(line);
- if (*p == 0 || *p == ';')
+ if (*p == 0)
continue;
+ if (*p == ';') {
+ static const char *attrs[] = {
+ "bp-based frame",
+ "library function",
+ "static",
+ "noreturn",
+ "thunk",
+ "fpd=",
+ };
+ if (p[2] != 'A' || strncmp(p, "; Attributes:", 13) != 0)
+ continue;
+
+ // parse IDA's attribute-list comment
+ ida_func_attr = 0;
+ p = sskip(p + 13);
+ // get rid of random tabs
+ for (i = 0; p[i] != 0; i++)
+ if (p[i] == '\t')
+ p[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;
+ p += strlen(attrs[i]);
+ break;
+ }
+ }
+ if (i == ARRAY_SIZE(attrs)) {
+ anote("unparsed IDA attr: %s\n", p);
+ break;
+ }
+ if (IS(attrs[i], "fpd=")) {
+ p = next_word(words[0], sizeof(words[0]), p);
+ // ignore for now..
+ }
+ }
+ continue;
+ }
+
memset(words, 0, sizeof(words));
for (wordc = 0; wordc < 16; wordc++) {
p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
aerr("proc '%s' while in_func '%s'?\n",
words[0], g_func);
p = words[0];
- if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
+ if ((ida_func_attr & IDAFA_THUNK)
+ || bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
skip_func = 1;
strcpy(g_func, words[0]);
+ set_label(0, words[0]);
in_func = 1;
continue;
}
pi = 0;
}
g_eqcnt = 0;
+ ida_func_attr = 0;
continue;
}
p = strchr(words[0], ':');
if (p != NULL) {
- len = p - words[0];
- if (len > sizeof(g_labels[0]) - 1)
- aerr("label too long: %d\n", len);
- if (g_labels[pi][0] != 0)
- aerr("dupe label?\n");
- memcpy(g_labels[pi], words[0], len);
- g_labels[pi][len] = 0;
+ set_label(pi, words[0]);
continue;
}
notaz.gp2x.de / ia32rtools.git / blobdiff