Address expressions are now fully algebraic.

mspdebug.man

@@ -296,17 +296,21 @@ renamed.
 .SH ADDRESS EXPRESSIONS
 Any command which accepts a memory address, length or register value
 as an argument may be given an address expression. An address
-expression consists of a sequence of one or more address values
+expression consists of an algebraic combination of values.
-separated by the operators \fB+\fR or \fB\-\fR.
 
 An address value may be either a decimal value, a hexadecimal value
 preceeded by the prefix \fB0x\fR, or a symbol name.
 
+The operators recognised are the usual algebraic operators: \fB+\fR, \fB-\fR,
+\fB*\fR, \fB/\fR, \fB%\fR, \fB(\fR and \fB)\fR. Operator precedence is the
+same as in C-like languages, and the \fB-\fR operator may be used as a
+unary negation operator.
+
 The following are all valid examples of address expressions:
 
-.B 64
+.B 2+2
 .br
-.B 0xffe0
+.B table_start + (elem_size + elem_pad)*4
 .br
 .B main+0x3f
 .br

util.c

@@ -27,6 +27,7 @@
 #include <unistd.h>
 #include <errno.h>
 #include <signal.h>
+#include <assert.h>
 
 #ifdef USE_READLINE
 #include <readline/readline.h>
@@ -665,82 +666,269 @@ int ctrlc_check(void)
 	return ctrlc_flag;
 }
 
-static char token_buf[64];
-static int token_len;
-static int token_mult;
-static int token_sum;
-
 static token_func_t token_func;
 
-static int token_add(void)
-{
-	int i;
-	int value;
-
-	if (!token_len)
-		return 0;
-
-	token_buf[token_len] = 0;
-	token_len = 0;
-
-	/* Is it a decimal? */
-	i = 0;
-	while (token_buf[i] && isdigit(token_buf[i]))
-		i++;
-	if (!token_buf[i]) {
-		token_sum += token_mult * atoi(token_buf);
-		return 0;
-	}
-
-	/* Is it hex? */
-	if (token_buf[0] == '0' && tolower(token_buf[1]) == 'x') {
-		token_sum += token_mult * strtol(token_buf + 2, NULL, 16);
-		return 0;
-	}
-
-	/* Look up the name in the symbol table */
-	if (token_func && !token_func(token_buf, &value)) {
-		token_sum += token_mult * value;
-		return 0;
-	}
-
-	fprintf(stderr, "unknown token: %s\n", token_buf);
-	return -1;
-}
-
 void set_token_func(token_func_t func)
 {
 	token_func = func;
 }
 
-int addr_exp(const char *text, int *addr)
+struct addr_exp_state {
-{
+	int     last_operator;
-	token_len = 0;
+	int     data_stack[32];
-	token_mult = 1;
+	int     data_stack_size;
-	token_sum = 0;
+	int     op_stack[32];
+	int     op_stack_size;
+};
 
-	while (*text) {
+static int addr_exp_data(struct addr_exp_state *s, const char *text)
-		if (isalnum(*text) || *text == '_' || *text == '$' ||
+{
-		    *text == '.' || *text == ':') {
+	int value;
-			if (token_len + 1 < sizeof(token_buf))
+
-				token_buf[token_len++] = *text;
+	if (!s->last_operator || s->last_operator == ')') {
-		} else {
+		fprintf(stderr, "syntax error at token %s\n", text);
-			if (token_add() < 0)
+		return -1;
+	}
+
+	/* Hex value */
+	if (*text == '0' && text[1] == 'x')
+		value = strtoul(text + 2, NULL, 16);
+	else if (isdigit(*text))
+		value = atoi(text);
+	else if (!token_func || token_func(text, &value) < 0) {
+		fprintf(stderr, "can't parse token: %s\n", text);
+		return -1;
+	}
+
+	if (s->data_stack_size + 1 > ARRAY_LEN(s->data_stack)) {
+		fprintf(stderr, "data stack overflow at token %s\n", text);
+		return -1;
+	}
+
+	s->data_stack[s->data_stack_size++] = value;
+	s->last_operator = 0;
+	return 0;
+}
+
+static int addr_exp_pop(struct addr_exp_state *s)
+{
+	char op = s->op_stack[--s->op_stack_size];
+	int data1 = s->data_stack[--s->data_stack_size];
+	int data2 = 0;
+
+	int result = 0;
+
+	if (op != 'N')
+		data2 = s->data_stack[--s->data_stack_size];
+
+	assert (s->op_stack_size >= 0);
+	assert (s->data_stack_size >= 0);
+
+	switch (op) {
+	case '+':
+		result = data2 + data1;
+		break;
+
+	case '-':
+		result = data2 - data1;
+		break;
+
+	case '*':
+		result = data2 * data1;
+		break;
+
+	case '/':
+		if (!data1)
+			goto divzero;
+		result = data2 / data1;
+		break;
+
+	case '%':
+		if (!data1)
+			goto divzero;
+		result = data2 % data1;
+		break;
+
+	case 'N':
+		result = -data1;
+		break;
+	}
+
+	s->data_stack[s->data_stack_size++] = result;
+	return 0;
+
+ divzero:
+	fprintf(stderr, "divide by zero\n");
+	return -1;
+}
+
+static int can_push(struct addr_exp_state *s, char op)
+{
+	char top;
+
+	if (!s->op_stack_size || op == '(')
+		return 1;
+
+	top = s->op_stack[s->op_stack_size - 1];
+
+	if (top == '(')
+		return 1;
+
+	switch (op) {
+	case 'N':
+		return 1;
+
+	case '*':
+	case '%':
+	case '/':
+		return top == '+' || top == '-';
+
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int addr_exp_op(struct addr_exp_state *s, char op)
+{
+	if (op == '(') {
+		if (!s->last_operator || s->last_operator == ')')
+			goto syntax_error;
+	} else if (op == '-') {
+		if (s->last_operator && s->last_operator != ')')
+			op = 'N';
+	} else {
+		if (s->last_operator && s->last_operator != ')')
+			goto syntax_error;
+	}
+
+	if (op == ')') {
+		/* ) collapses the stack to the last matching ( */
+		while (s->op_stack_size &&
+		       s->op_stack[s->op_stack_size - 1] != '(')
+			if (addr_exp_pop(s) < 0)
 				return -1;
-			if (*text == '+')
+
-				token_mult = 1;
+		if (!s->op_stack_size) {
-			if (*text == '-')
+			fprintf(stderr, "parenthesis mismatch: )\n");
-				token_mult = -1;
+			return -1;
 		}
 
+		s->op_stack_size--;
+	} else {
+		while (!can_push(s, op))
+			if (addr_exp_pop(s) < 0)
+				return -1;
+
+		if (s->op_stack_size + 1 > ARRAY_LEN(s->op_stack)) {
+			fprintf(stderr, "operator stack overflow: %c\n", op);
+			return -1;
+		}
+
+		s->op_stack[s->op_stack_size++] = op;
+	}
+
+	s->last_operator = op;
+	return 0;
+
+ syntax_error:
+	fprintf(stderr, "syntax error at operator %c\n", op);
+	return -1;
+}
+
+static int addr_exp_finish(struct addr_exp_state *s, int *ret)
+{
+	if (s->last_operator && s->last_operator != ')') {
+		fprintf(stderr, "syntax error at end of expression\n");
+		return -1;
+	}
+
+	while (s->op_stack_size) {
+		if (s->op_stack[s->op_stack_size - 1] == '(') {
+			fprintf(stderr, "parenthesis mismatch: (\n");
+			return -1;
+		}
+
+		if (addr_exp_pop(s) < 0)
+			return -1;
+	}
+
+	if (s->data_stack_size != 1) {
+		fprintf(stderr, "no data: stack size is %d\n",
+			s->data_stack_size);
+		return -1;
+	}
+
+	if (ret)
+		*ret = s->data_stack[0];
+
+	return 0;
+}
+
+int addr_exp(const char *text, int *addr)
+{
+	const char *text_save = text;
+	int last_cc = 1;
+	char token_buf[64];
+	int token_len = 0;
+	struct addr_exp_state s = {0};
+
+	s.last_operator = '(';
+
+	for (;;) {
+		int cc;
+
+		/* Figure out what class this character is */
+		if (*text == '+' || *text == '-' ||
+		    *text == '*' || *text == '/' ||
+		    *text == '%' || *text == '(' ||
+		    *text == ')')
+			cc = 1;
+		else if (!*text || isspace(*text))
+			cc = 2;
+		else if (isalnum(*text) || *text == '.' || *text == '_' ||
+			 *text == '$' || *text == ':')
+			cc = 3;
+		else {
+			fprintf(stderr, "illegal character in expression: %c\n",
+				*text);
+			return -1;
+		}
+
+		/* Accumulate and process token text */
+		if (cc == 3) {
+			if (token_len + 1 < sizeof(token_buf))
+				token_buf[token_len++] = *text;
+		} else if (token_len) {
+			token_buf[token_len] = 0;
+			token_len = 0;
+
+			if (addr_exp_data(&s, token_buf) < 0)
+				goto fail;
+		}
+
+		/* Process operators */
+		if (cc == 1) {
+			if (addr_exp_op(&s, *text) < 0)
+				goto fail;
+		}
+
+		if (!*text)
+			break;
+
+		last_cc = cc;
 		text++;
 	}
 
-	if (token_add() < 0)
+	if (addr_exp_finish(&s, addr) < 0)
-		return -1;
+		goto fail;
 
-	*addr = token_sum & 0xffff;
 	return 0;
+
+ fail:
+	fprintf(stderr, "bad address expression: %s\n", text_save);
+	return -1;
 }
 
 static int modify_flags;