};
// datap:
-// OP_CALL - parser proto hint (str)
+// on start: function/data type hint (sctproto)
+// after analysis:
// (OPF_CC) - points to one of (OPF_FLAGS) that affects cc op
// OP_PUSH - points to OP_POP in complex push/pop graph
// OP_POP - points to OP_PUSH in simple push/pop pair
IDAFA_FPD = (1 << 5),
};
+enum sct_func_attr {
+ SCTFA_CLEAR_SF = (1 << 0), // clear stack frame
+ SCTFA_CLEAR_REGS = (1 << 1), // clear registers (mask)
+};
+
enum x87_const {
X87_CONST_1 = 1,
X87_CONST_2T,
static int g_stack_frame_used;
static int g_stack_fsz;
static int g_ida_func_attr;
+static int g_sct_func_attr;
+static int g_stack_clear_start; // in dwords
+static int g_stack_clear_len;
+static int g_regmask_init;
static int g_skip_func;
static int g_allow_regfunc;
static int g_quiet_pp;
};
#define mxAX (1 << xAX)
+#define mxCX (1 << xCX)
#define mxDX (1 << xDX)
#define mxST0 (1 << xST0)
#define mxST1 (1 << xST1)
{
buf[0] = 0;
- if (!opr->is_ptr)
+ if (!opr->is_ptr || strchr(opr->name, '['))
return buf;
if (opr->pp == NULL || opr->pp->type.name == NULL
|| opr->pp->is_fptr)
static int get_pp_arg_regmask_dst(const struct parsed_proto *pp)
{
+ int regmask = 0;
+ int i, reg;
+
+ if (pp->has_retreg) {
+ for (i = 0; i < pp->argc; i++) {
+ if (pp->arg[i].type.is_retreg) {
+ reg = char_array_i(regs_r32,
+ ARRAY_SIZE(regs_r32), pp->arg[i].reg);
+ ferr_assert(ops, reg >= 0);
+ regmask |= 1 << reg;
+ }
+ }
+ }
+
if (strstr(pp->ret_type.name, "int64"))
- return (1 << xAX) | (1 << xDX);
+ return regmask | (1 << xAX) | (1 << xDX);
if (IS(pp->ret_type.name, "float")
|| IS(pp->ret_type.name, "double"))
{
- return mxST0;
+ return regmask | mxST0;
}
if (strcasecmp(pp->ret_type.name, "void") == 0)
- return 0;
+ return regmask;
- return mxAX;
+ return regmask | mxAX;
}
static void resolve_branches_parse_calls(int opcnt)
po->bt_i = -1;
po->btj = NULL;
+ if (po->datap != NULL) {
+ pp = calloc(1, sizeof(*pp));
+ my_assert_not(pp, NULL);
+
+ ret = parse_protostr(po->datap, pp);
+ if (ret < 0)
+ ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
+ free(po->datap);
+ po->datap = NULL;
+ po->pp = pp;
+ }
+
if (po->op == OP_CALL) {
pp = NULL;
- if (po->operand[0].type == OPT_LABEL) {
+ if (po->pp != NULL)
+ pp = po->pp;
+ else if (po->operand[0].type == OPT_LABEL)
+ {
tmpname = opr_name(po, 0);
if (IS_START(tmpname, "loc_"))
ferr(po, "call to loc_*\n");
my_assert_not(pp, NULL);
}
}
- else if (po->datap != NULL) {
- pp = calloc(1, sizeof(*pp));
- my_assert_not(pp, NULL);
-
- ret = parse_protostr(po->datap, pp);
- if (ret < 0)
- ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
- free(po->datap);
- po->datap = NULL;
- }
if (pp != NULL) {
if (pp->is_fptr)
}
}
-static const struct parsed_proto *resolve_icall(int i, int opcnt,
- int *pp_i, int *multi_src)
-{
- const struct parsed_proto *pp = NULL;
- int search_advice = 0;
-
- *multi_src = 0;
- *pp_i = -1;
-
- switch (ops[i].operand[0].type) {
- case OPT_REGMEM:
- case OPT_LABEL:
- case OPT_OFFSET:
- pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
- if (!search_advice)
- break;
- // fallthrough
- default:
- scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
- pp_i, multi_src);
- break;
- }
-
- return pp;
-}
-
// find an instruction that changed opr before i op
// *op_i must be set to -1 by the caller
-// *entry is set to 1 if one source is determined to be the caller
+// *is_caller is set to 1 if one source is determined to be g_func arg
// returns 1 if found, *op_i is then set to origin
+// returns -1 if multiple origins are found
static int resolve_origin(int i, const struct parsed_opr *opr,
int magic, int *op_i, int *is_caller)
{
return -1;
}
+static const struct parsed_proto *resolve_icall(int i, int opcnt,
+ int *pp_i, int *multi_src)
+{
+ const struct parsed_proto *pp = NULL;
+ int search_advice = 0;
+ int offset = -1;
+ char name[256];
+ char s_reg[4];
+ int reg, len;
+ int ret;
+
+ *multi_src = 0;
+ *pp_i = -1;
+
+ switch (ops[i].operand[0].type) {
+ case OPT_REGMEM:
+ // try to resolve struct member calls
+ ret = sscanf(ops[i].operand[0].name, "%3s+%x%n",
+ s_reg, &offset, &len);
+ if (ret == 2 && len == strlen(ops[i].operand[0].name))
+ {
+ reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s_reg);
+ if (reg >= 0) {
+ struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, reg);
+ int j = -1;
+ ret = resolve_origin(i, &opr, i + opcnt * 19, &j, NULL);
+ if (ret != 1)
+ break;
+ if (ops[j].op == OP_MOV && ops[j].operand[1].type == OPT_REGMEM
+ && ops[j].operand[0].lmod == OPLM_DWORD
+ && ops[j].pp == NULL) // no hint
+ {
+ // allow one simple dereference (directx)
+ reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32),
+ ops[j].operand[1].name);
+ if (reg < 0)
+ break;
+ struct parsed_opr opr2 = OPR_INIT(OPT_REG, OPLM_DWORD, reg);
+ int k = -1;
+ ret = resolve_origin(j, &opr2, j + opcnt * 19, &k, NULL);
+ if (ret != 1)
+ break;
+ j = k;
+ }
+ if (ops[j].op != OP_MOV)
+ break;
+ if (ops[j].operand[0].lmod != OPLM_DWORD)
+ break;
+ if (ops[j].pp != NULL) {
+ // type hint in asm
+ pp = ops[j].pp;
+ }
+ else if (ops[j].operand[1].type == OPT_REGMEM) {
+ // allow 'hello[ecx]' - assume array of same type items
+ ret = sscanf(ops[j].operand[1].name, "%[^[][e%2s]",
+ name, s_reg);
+ if (ret != 2)
+ break;
+ pp = proto_parse(g_fhdr, name, g_quiet_pp);
+ }
+ else if (ops[j].operand[1].type == OPT_LABEL)
+ pp = proto_parse(g_fhdr, ops[j].operand[1].name, g_quiet_pp);
+ else
+ break;
+ if (pp == NULL)
+ break;
+ if (pp->is_func || pp->is_fptr || !pp->type.is_struct) {
+ pp = NULL;
+ break;
+ }
+ pp = proto_lookup_struct(g_fhdr, pp->type.name, offset);
+ }
+ break;
+ }
+ // fallthrough
+ case OPT_LABEL:
+ case OPT_OFFSET:
+ pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
+ if (!search_advice)
+ break;
+ // fallthrough
+ default:
+ scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
+ pp_i, multi_src);
+ break;
+ }
+
+ return pp;
+}
+
static struct parsed_proto *process_call_early(int i, int opcnt,
int *adj_i)
{
if (g_bp_frame && !(po->flags & OPF_EBP_S))
regmask_new &= ~(1 << xBP);
- if (po->op == OP_CALL) {
- // allow fastcall calls from anywhere, calee may be also sitting
- // in some fastcall table even when it's not using reg args
- if (regmask_new & po->regmask_src & (1 << xCX)) {
- *regmask_init |= (1 << xCX);
- regmask_now |= (1 << xCX);
- regmask_new &= ~(1 << xCX);
- }
- if (regmask_new & po->regmask_src & (1 << xDX)) {
- *regmask_init |= (1 << xDX);
- regmask_now |= (1 << xDX);
- regmask_new &= ~(1 << xDX);
- }
- }
-
if (regmask_new != 0)
fnote(po, "uninitialized reg mask: %x\n", regmask_new);
if (po->flags & OPF_TAIL) {
if (regmask_now & (mxST0 | mxST1))
ferr(po, "float regs on tail: %x\n", regmask_now);
- return;
+
+ // there is support for "conditional tailcall", sort of
+ if (!(po->flags & OPF_CC))
+ return;
}
}
}
g_bp_frame = g_sp_frame = g_stack_fsz = 0;
g_stack_frame_used = 0;
+ if (g_sct_func_attr & SCTFA_CLEAR_REGS)
+ regmask_init = g_regmask_init;
g_func_pp = proto_parse(fhdr, funcn, 0);
if (g_func_pp == NULL)
regmask_arg = get_pp_arg_regmask_src(g_func_pp);
regmask_ret = get_pp_arg_regmask_dst(g_func_pp);
- if (g_func_pp->has_retreg) {
- for (arg = 0; arg < g_func_pp->argc; arg++) {
- if (g_func_pp->arg[arg].type.is_retreg) {
- reg = char_array_i(regs_r32,
- ARRAY_SIZE(regs_r32), g_func_pp->arg[arg].reg);
- ferr_assert(ops, reg >= 0);
- regmask_ret |= 1 << reg;
- }
- }
- }
-
// pass1:
// - resolve all branches
// - parse calls with labels
// - find POPs for PUSHes, rm both
// - scan for all used registers
memset(cbits, 0, sizeof(cbits));
- reg_use_pass(0, opcnt, cbits, 0, ®mask,
+ reg_use_pass(0, opcnt, cbits, regmask_init, ®mask,
0, ®mask_save, ®mask_init, regmask_arg);
// pass7:
if (had_decl)
fprintf(fout, "\n");
+ // do stack clear, if needed
+ if (g_sct_func_attr & SCTFA_CLEAR_SF) {
+ fprintf(fout, " ");
+ if (g_stack_clear_len != 0) {
+ if (g_stack_clear_len <= 4) {
+ for (i = 0; i < g_stack_clear_len; i++)
+ fprintf(fout, "sf.d[%d] = ", g_stack_clear_start + i);
+ fprintf(fout, "0;\n");
+ }
+ else {
+ fprintf(fout, "memset(&sf[%d], 0, %d);\n",
+ g_stack_clear_start, g_stack_clear_len * 4);
+ }
+ }
+ else
+ fprintf(fout, "memset(&sf, 0, sizeof(sf));\n");
+ }
+
if (g_func_pp->is_vararg) {
if (g_func_pp->argc_stack == 0)
ferr(ops, "vararg func without stack args?\n");
char *p, namebuf[NAMELEN];
const char *name;
int regmask_dep;
- int argc_stack;
+ int argc_normal;
int j, arg;
for (; count > 0; count--, fp++) {
}
regmask_dep = fp->regmask_dep;
- argc_stack = fp->argc_stack;
+ argc_normal = fp->argc_stack;
fprintf(fout, "%-5s", fp->pp ? fp->pp->ret_type.name :
(fp->has_ret ? "int" : "void"));
- if (regmask_dep && (fp->is_stdcall || argc_stack == 0)
- && (regmask_dep & ~((1 << xCX) | (1 << xDX))) == 0)
+ if (regmask_dep && (fp->is_stdcall || fp->argc_stack > 0)
+ && (regmask_dep & ~mxCX) == 0)
+ {
+ fprintf(fout, "/*__thiscall*/ ");
+ argc_normal++;
+ regmask_dep = 0;
+ }
+ else if (regmask_dep && (fp->is_stdcall || fp->argc_stack == 0)
+ && (regmask_dep & ~(mxCX | mxDX)) == 0)
{
fprintf(fout, " __fastcall ");
- if (!(regmask_dep & (1 << xDX)) && argc_stack == 0)
- argc_stack = 1;
+ if (!(regmask_dep & (1 << xDX)) && fp->argc_stack == 0)
+ argc_normal = 1;
else
- argc_stack += 2;
+ argc_normal += 2;
regmask_dep = 0;
}
else if (regmask_dep && !fp->is_stdcall) {
}
}
- for (j = 0; j < argc_stack; j++) {
+ for (j = 0; j < argc_normal; j++) {
arg++;
if (arg != 1)
fprintf(fout, ", ");
return s + i;
}
-static void scan_variables(FILE *fasm)
+static int cmpstringp(const void *p1, const void *p2)
+{
+ return strcmp(*(char * const *)p1, *(char * const *)p2);
+}
+
+static int is_xref_needed(char *p, char **rlist, int rlist_len)
+{
+ char *p2;
+
+ p = sskip(p);
+ if (strstr(p, "..."))
+ // unable to determine, assume needed
+ return 1;
+
+ if (*p == '.') // .text, .data, ...
+ // ref from other data or non-function -> no
+ return 0;
+
+ p2 = strpbrk(p, "+:\r\n\x18");
+ if (p2 != NULL)
+ *p2 = 0;
+ if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
+ // referenced from removed code
+ return 0;
+
+ return 1;
+}
+
+static int xrefs_show_need(FILE *fasm, char *p,
+ char **rlist, int rlist_len)
+{
+ int found_need = 0;
+ char line[256];
+ long pos;
+
+ p = strrchr(p, ';');
+ if (p != NULL && *p == ';' && IS_START(p + 2, "DATA XREF: ")) {
+ p += 13;
+ if (is_xref_needed(p, rlist, rlist_len))
+ return 1;
+ }
+
+ pos = ftell(fasm);
+ while (1)
+ {
+ if (!my_fgets(line, sizeof(line), fasm))
+ break;
+ // non-first line is always indented
+ if (!my_isblank(line[0]))
+ break;
+
+ // should be no content, just comment
+ p = sskip(line);
+ if (*p != ';')
+ break;
+
+ p = strrchr(p, ';');
+ p += 2;
+ // it's printed once, but no harm to check again
+ if (IS_START(p, "DATA XREF: "))
+ p += 11;
+
+ if (is_xref_needed(p, rlist, rlist_len)) {
+ found_need = 1;
+ break;
+ }
+ }
+ fseek(fasm, pos, SEEK_SET);
+ return found_need;
+}
+
+static void scan_variables(FILE *fasm, char **rlist, int rlist_len)
{
struct scanned_var *var;
char line[256] = { 0, };
break;
}
+ // check refs comment(s)
+ if (!xrefs_show_need(fasm, p, rlist, rlist_len))
+ continue;
+
if ((hg_var_cnt & 0xff) == 0) {
hg_vars = realloc(hg_vars, sizeof(hg_vars[0])
* (hg_var_cnt + 0x100));
return strcmp(c1->name, c2->name);
}
-static int cmpstringp(const void *p1, const void *p2)
-{
- return strcmp(*(char * const *)p1, *(char * const *)p2);
-}
-
static void scan_ahead(FILE *fasm)
{
char words[2][256];
}
if (g_header_mode)
- scan_variables(fasm);
+ scan_variables(fasm, rlist, rlist_len);
while (my_fgets(line, sizeof(line), fasm))
{
}
}
}
+ else if (p[2] == 's' && IS_START(p, "; sctattr:"))
+ {
+ static const char *attrs[] = {
+ "clear_sf",
+ "clear_regmask",
+ };
+
+ // parse manual attribute-list comment
+ g_sct_func_attr = 0;
+ p = sskip(p + 10);
+
+ for (; *p != 0; p = sskip(p)) {
+ for (i = 0; i < ARRAY_SIZE(attrs); i++) {
+ if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
+ g_sct_func_attr |= 1 << i;
+ p += strlen(attrs[i]);
+ break;
+ }
+ }
+ if (*p == '=') {
+ j = ret = 0;
+ if (i == 0)
+ // clear_sf=start,len (in dwords)
+ ret = sscanf(p, "=%d,%d%n", &g_stack_clear_start,
+ &g_stack_clear_len, &j);
+ else if (i == 1)
+ // clear_regmask=<mask>
+ ret = sscanf(p, "=%d%n", &g_regmask_init, &j) + 1;
+ if (ret < 2) {
+ anote("unparsed attr value: %s\n", p);
+ break;
+ }
+ p += j;
+ }
+ else if (i == ARRAY_SIZE(attrs)) {
+ anote("unparsed sct attr: %s\n", p);
+ break;
+ }
+ }
+ }
else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
{
p += 30;
pending_endp = 0;
in_func = 0;
g_ida_func_attr = 0;
+ g_sct_func_attr = 0;
+ g_stack_clear_start = 0;
+ g_stack_clear_len = 0;
+ g_regmask_init = 0;
skip_warned = 0;
g_skip_func = 0;
g_func[0] = 0;
parse_op(&ops[pi], words, wordc);
- if (sctproto != NULL) {
- if (ops[pi].op == OP_CALL || ops[pi].op == OP_JMP)
- ops[pi].datap = sctproto;
- sctproto = NULL;
- }
+ ops[pi].datap = sctproto;
+ sctproto = NULL;
pi++;
}
notaz.gp2x.de / ia32rtools.git / blobdiff