void syms_add_global_csl(expression_t * exp)
{
if (!exp)
return;
if (exp->operator == ',') {
syms_add_global_csl(exp->left);
syms_add_global_csl(exp->right);
}
else {
syms_add_global(exp->name);
}
}
static asm86_t *gnu_get_statement(void)
/* Get a pseudo op or machine instruction with arguments. */
{
token_t *t= get_token(0);
token_t *tn;
asm86_t *a;
mnemonic_t *m;
int n;
int prefix_seen;
int deref;
assert(t->type == T_WORD);
a= new_asm86();
/* Process instruction prefixes. */
for (prefix_seen= 0;; prefix_seen= 1) {
if (strcmp(t->name, "rep") == 0
|| strcmp(t->name, "repe") == 0
|| strcmp(t->name, "repne") == 0
|| strcmp(t->name, "repz") == 0
|| strcmp(t->name, "repnz") == 0
) {
if (a->rep != ONCE) {
parse_err(1, t,
"can't have more than one rep\n");
}
switch (t->name[3]) {
case 0: a->rep= REP; break;
case 'e':
case 'z': a->rep= REPE; break;
case 'n': a->rep= REPNE; break;
}
} else
if (!prefix_seen) {
/* No prefix here, get out! */
break;
} else {
/* No more prefixes, next must be an instruction. */
if (t->type != T_WORD
|| (m= search_mnem(t->name)) == nil
|| m->optype == PSEUDO
) {
parse_err(1, t,
"machine instruction expected after instruction prefix\n");
del_asm86(a);
return nil;
}
break;
}
/* Skip the prefix and extra newlines. */
do {
skip_token(1);
} while ((t= get_token(0))->symbol == ';');
}
/* All the readahead being done upsets the line counter. */
a->line= t->line;
/* Read a machine instruction or pseudo op. */
if ((m= search_mnem(t->name)) == nil) {
/* we assume that unknown stuff is part of unresolved macro */
a->opcode = UNKNOWN;
if (zap_unknown(a)) {
parse_err(1, t, "unknown instruction '%s'\n", t->name);
del_asm86(a);
return nil;
}
return a;
}
a->opcode= m->opcode;
a->optype= m->optype;
a->oaz= 0;
if (a->optype == OWORD) {
a->oaz|= OPZ;
a->optype= WORD;
}
else if (a->optype == JUMP16) {
a->oaz|= OPZ;
a->optype= JUMP;
}
switch (a->opcode) {
case IN:
case OUT:
case INT:
deref= 0;
break;
default:
deref= (a->optype >= BYTE);
}
n= 1;
if (get_token(1)->type != T_COMMENT && get_token(1)->symbol != ';'
&& (a->args= gnu_get_oplist(a, &n, deref)) == nil) {
del_asm86(a);
return nil;
}
tn = get_token(n);
if (tn->type == T_COMMENT) {
a->raw_string = malloc(tn->len + 1);
if (!a->raw_string)
return NULL;
strcpy(a->raw_string, tn->name);
} else
if (get_token(n)->symbol != ';') {
parse_err(1, t, "garbage at end of instruction\n");
del_asm86(a);
return nil;
}
if (!is_pseudo(a->opcode)) {
/* GNU operand order is the other way around. */
expression_t *e, *t;
e= a->args;
while (e != nil && e->operator == ',') {
t= e->right; e->right= e->left; e->left= t;
e= e->left;
}
}
switch (a->opcode) {
case DOT_ALIGN:
/* Delete two argument .align, because ACK can't do it.
*/
if (a->args == nil || a->args->operator != 'W') {
del_asm86(a);
return nil;
}
if (a->args != nil && a->args->operator == 'W'
&& isanumber(a->args->name)
) {
unsigned n;
char num[sizeof(int) * CHAR_BIT / 3 + 1];
n= strtoul(a->args->name, nil, 0);
sprintf(num, "%u", n);
deallocate(a->args->name);
a->args->name= copystr(num);
}
break;
case DOT_DEFINE:
case DOT_EXTERN:
syms_add_global_csl(a->args);
break;
case DOT_COMM:
syms_add_global(a->args->left->name);
break;
case DOT_LCOMM:
syms_add(a->args->left->name);
break;
case JMPF:
case CALLF:
/*FALL THROUGH*/
case JMP:
case CALL:
break;
default:;
}
skip_token(n+1);
return a;
}
static asm86_t *ack_get_statement(void)
/* Get a pseudo op or machine instruction with arguments. */
{
token_t *t= get_token(0);
token_t *tn;
asm86_t *a;
mnemonic_t *m;
int n;
int prefix_seen;
int oaz_prefix;
int deref;
assert(t->type == T_WORD);
if (strcmp(t->name, ".sect") == 0) {
/* .sect .text etc. Accept only four segment names. */
skip_token(1);
t= get_token(0);
if (t->type != T_WORD || (
strcmp(t->name, ".text") != 0
&& strcmp(t->name, ".rom") != 0
&& strcmp(t->name, ".data") != 0
&& strcmp(t->name, ".bss") != 0
&& strcmp(t->name, ".end") != 0
)) {
parse_err(1, t, "weird section name to .sect\n");
return nil;
}
}
a= new_asm86();
/* Process instruction prefixes. */
oaz_prefix= 0;
for (prefix_seen= 0;; prefix_seen= 1) {
if (strcmp(t->name, "o16") == 0) {
if (use16()) {
parse_err(1, t, "o16 in an 8086 section\n");
}
oaz_prefix|= OPZ;
} else
if (strcmp(t->name, "o32") == 0) {
if (use32()) {
parse_err(1, t, "o32 in an 80386 section\n");
}
oaz_prefix|= OPZ;
} else
if (strcmp(t->name, "a16") == 0) {
if (use16()) {
parse_err(1, t, "a16 in an 8086 section\n");
}
oaz_prefix|= ADZ;
} else
if (strcmp(t->name, "a32") == 0) {
if (use32()) {
parse_err(1, t, "a32 in an 80386 section\n");
}
oaz_prefix|= ADZ;
} else
if (strcmp(t->name, "rep") == 0
|| strcmp(t->name, "repe") == 0
|| strcmp(t->name, "repne") == 0
|| strcmp(t->name, "repz") == 0
|| strcmp(t->name, "repnz") == 0
) {
if (a->rep != ONCE) {
parse_err(1, t,
"can't have more than one rep\n");
}
switch (t->name[3]) {
case 0: a->rep= REP; break;
case 'e':
case 'z': a->rep= REPE; break;
case 'n': a->rep= REPNE; break;
}
} else
if (strchr("cdefgs", t->name[0]) != nil
&& strcmp(t->name+1, "seg") == 0) {
if (a->seg != DEFSEG) {
parse_err(1, t,
"can't have more than one segment prefix\n");
}
switch (t->name[0]) {
case 'c': a->seg= CSEG; break;
case 'd': a->seg= DSEG; break;
case 'e': a->seg= ESEG; break;
case 'f': a->seg= FSEG; break;
case 'g': a->seg= GSEG; break;
case 's': a->seg= SSEG; break;
}
} else
if (!prefix_seen) {
/* No prefix here, get out! */
break;
} else {
/* No more prefixes, next must be an instruction. */
if (t->type != T_WORD
|| (m= search_mnem(t->name)) == nil
|| m->optype == PSEUDO
) {
parse_err(1, t,
"machine instruction expected after instruction prefix\n");
del_asm86(a);
return nil;
}
if (oaz_prefix != 0 && m->optype != JUMP
&& m->optype != WORD) {
parse_err(1, t,
"'%s' can't have an operand size prefix\n", m->name);
}
break;
}
/* Skip the prefix and extra newlines. */
do {
skip_token(1);
} while ((t= get_token(0))->symbol == ';');
}
/* All the readahead being done upsets the line counter. */
a->line= t->line;
/* Read a machine instruction or pseudo op. */
if ((m= search_mnem(t->name)) == nil) {
/* we assume that unknown stuff is part of unresolved macro */
a->opcode = UNKNOWN;
if (zap_unknown(a)) {
parse_err(1, t, "unknown instruction '%s'\n", t->name);
del_asm86(a);
return nil;
}
return a;
}
a->opcode= m->opcode;
a->optype= m->optype;
a->oaz = oaz_prefix;
switch (a->opcode) {
case IN:
case OUT:
case INT:
deref= 0;
break;
default:
deref= (a->optype >= BYTE);
}
n= 1;
if (get_token(1)->type != T_COMMENT && get_token(1)->symbol != ';'
&& (a->args= ack_get_oplist(a, &n, deref)) == nil) {
del_asm86(a);
return nil;
}
tn = get_token(n);
if (tn->type == T_COMMENT) {
a->raw_string = malloc(tn->len + 1);
if (!a->raw_string)
return NULL;
strcpy(a->raw_string, tn->name);
} else
if (tn->symbol != ';') {
parse_err(1, t, "garbage at end of instruction\n");
del_asm86(a);
return nil;
}
switch (a->opcode) {
case DOT_ALIGN:
/* Restrict .align to have a single numeric argument, some
* assemblers think of the argument as a power of two, so
* we need to be able to change the value.
*/
if (a->args == nil || a->args->operator != 'W'
|| !isanumber(a->args->name)) {
parse_warn(1, t,
".align is restricted to one numeric argument. Unpreprocessed file?\n");
}
break;
case DOT_DEFINE:
case DOT_EXTERN:
syms_add_global_csl(a->args);
break;
case DOT_COMM:
syms_add_global(a->args->left->name);
break;
case DOT_LCOMM:
syms_add(a->args->left->name);
break;
case JMPF:
case CALLF:
/* NCC jmpf off,seg -> ACK jmpf seg:off */
if (dialect == NCC && a->args != nil
&& a->args->operator == ',') {
expression_t *t;
t= a->args->left;
a->args->left= a->args->right;
a->args->right= t;
break;
}
/*FALL THROUGH*/
case JMP:
case CALL:
/* NCC jmp @(reg) -> ACK jmp (reg) */
if (dialect == NCC && a->args != nil && (
(a->args->operator == '('
&& a->args->middle != nil
&& a->args->middle->operator == 'O')
|| (a->args->operator == 'O'
&& a->args->left == nil
&& a->args->middle != nil
&& a->args->right == nil)
)) {
expression_t *t;
t= a->args;
a->args= a->args->middle;
t->middle= nil;
del_expr(t);
if (a->args->operator == 'B') a->args->operator= 'W';
}
break;
default:;
}
skip_token(n+1);
return a;
}