Improved constant arithmetic: had several bugs

lang/cem/cpp.ansi/LLlex.c

@@ -259,7 +259,10 @@ garbage:
 	}
 	case STNUM:				/* a numeric constant	*/
 	{			/* it may only be an integer constant */
-		register int base = 10, val = 0, vch;
+		register int base = 10, vch;
+		register arith val = 0;
+		int ovfl = 0;
+		arith ubound = ~(1<<(sizeof(arith)*8-1))/(base/2);
 
 		/* Since the preprocessor only knows integers and has
 		 * nothing to do with ellipsis we just return when the
@@ -279,11 +282,35 @@ garbage:
 
 		}
 		while ((vch = val_in_base(ch, base)) >= 0) {
-			val = val * base + vch;		/* overflow? nah */
+			if (val < 0 || val > ubound) ovfl++;
+			val *= base;
+			if (val < 0 && val + vch >= 0) ovfl++;
+			val += vch;
 			ch = GetChar();
 		}
-		while (ch == 'l' || ch == 'L' || ch == 'u' || ch == 'U')
+		ptok->tk_unsigned = 0;
+		if (ch == 'u' || ch == 'U') {
+			ptok->tk_unsigned = 1;
 			ch = GetChar();
+			if (ch == 'l' || ch == 'L') {
+				ch = GetChar();
+			}
+		}
+		else if (ch == 'l' || ch == 'L') {
+			ch = GetChar();
+			if (ch == 'u' || ch == 'U') {
+				ptok->tk_unsigned = 1;
+				ch = GetChar();
+			}
+		}
+		if (ovfl) {
+			warning("overflow in constant");
+			ptok->tk_unsigned = 1;
+		}
+		else if (val < 0) {
+			/* give warning??? */
+			ptok->tk_unsigned = 1;
+		}
 		UnGetChar();
 		ptok->tk_val = val;
 		return ptok->tk_symb = INTEGER;

lang/cem/cpp.ansi/LLlex.h

@@ -15,7 +15,10 @@ struct token	{
 	int tok_symb;		/* the token itself */
 	union	{
 		char *tok_str;
-		arith tok_val;		/* for INTEGER */
+		struct {
+			int tok_unsigned;
+			arith tok_val;		/* for INTEGER */
+		} tok_int;
 	} tok_data;
 };
 
@@ -23,7 +26,8 @@ struct token	{
 
 #define tk_symb	tok_symb
 #define tk_str	tok_data.tok_str
-#define tk_val	tok_data.tok_val
+#define tk_val	tok_data.tok_int.tok_val
+#define tk_unsigned tok_data.tok_int.tok_unsigned
 
 extern struct token dot;
 

lang/cem/cpp.ansi/ch3bin.c

@@ -8,22 +8,89 @@
 #include	"Lpars.h"
 #include	"arith.h"
 
-ch3bin(pval, oper, val)
+#define arith_sign (1 << (sizeof(arith)*8-1))
+
+ch3bin(pval, pis_uns, oper, val, is_uns)
 	register arith *pval, val;
-	int oper;
+	int oper, is_uns, *pis_uns;
 {
+	if (is_uns) *pis_uns = 1;
 	switch (oper)	{
-	case '%':
-		if (val == 0)
-			error("%% by 0");
-		else
-			*pval = *pval % val;
-		break;
 	case '/':
-		if (val == 0)
+		if (val == 0) {
 			error("/ by 0");
-		else
+			break;
+		}
+		if (*pis_uns) {
+#ifdef UNSIGNED_ARITH
+			*pval /= (UNSIGNED_ARITH) val;
+#else
+			/*	this is more of a problem than you might
+				think on C compilers which do not have
+				unsigned arith (== long (probably)).
+			*/
+			if (val & arith_sign)	{/* val > max_arith */
+				*pval = ! (*pval >= 0 || *pval < val);
+				/*	this is the unsigned test
+					*pval < val for val > max_arith
+				*/
+			}
+			else	{		/* val <= max_arith */
+				arith half, bit, hdiv, hrem, rem;
+
+				half = (*pval >> 1) & ~arith_sign;
+				bit = *pval & 01;
+				/*	now *pval == 2 * half + bit
+					and half <= max_arith
+					and bit <= max_arith
+				*/
+				hdiv = half / val;
+				hrem = half % val;
+				rem = 2 * hrem + bit;
+				*pval = 2 * hdiv + (rem < 0 || rem >= val);
+				/*	that is the unsigned compare
+					rem >= val for val <= max_arith
+				*/
+			}
+#endif
+		}
+		else {
+			*pval = *pval % val;
+		}
+		break;
+	case '%':
+		if (val == 0) {
+			error("%% by 0");
+			break;
+		}
+		if (*pis_uns) {
+#ifdef UNSIGNED_ARITH
+			*pval %= (UNSIGNED_ARITH) val;
+#else
+			if (val & arith_sign)	{/* val > max_arith */
+				*pval = (*pval >= 0 || *pval < val) ? *pval : *pval - val;
+				/*	this is the unsigned test
+					*pval < val for val > max_arith
+				*/
+			}
+			else	{		/* val <= max_arith */
+				arith half, bit, hrem, rem;
+
+				half = (*pval >> 1) & ~arith_sign;
+				bit = *pval & 01;
+				/*	now *pval == 2 * half + bit
+					and half <= max_arith
+					and bit <= max_arith
+				*/
+				hrem = half % val;
+				rem = 2 * hrem + bit;
+				*pval = (rem < 0 || rem >= val) ? rem - val : rem;
+			}
+#endif
+		}
+		else {
 			*pval = *pval / val;
+		}
 		break;
 	case '*':
 		*pval = *pval * val;
@@ -38,19 +105,44 @@ ch3bin(pval, oper, val)
 		*pval = *pval << val;
 		break;
 	case RIGHT:
-		*pval = *pval >> val;
+		if (val == 0) break;
+		if (*pis_uns) {
+			*pval = (*pval >> 1) & ~arith_sign;
+			*pval = *pval >> (val - 1);
+		}
+		else *pval = *pval >> val;
 		break;
 	case '<':
-		*pval = (*pval < val);
+		{	arith tmp = *pval; *pval = val; val = tmp; }
-		break;
+		/* fall through */
 	case '>':
-		*pval = (*pval > val);
+		if (*pis_uns) {
+#ifdef UNSIGNED_ARITH
+			*pval = (UNSIGNED_ARITH) *pval > (UNSIGNED_ARITH) val;
+#else
+			*pval = (*pval & arith_sign ?
+				(val & arith_sign ? *pval > val : 1) :
+				(val & arith_sign ? 0 : *pval > val)
+			);
+#endif
+		}
+		else	*pval = (*pval > val);
 		break;
 	case LESSEQ:
-		*pval = (*pval <= val);
+		{	arith tmp = *pval; *pval = val; val = tmp; }
-		break;
+		/* fall through */
 	case GREATEREQ:
-		*pval = (*pval >= val);
+		if (*pis_uns) {
+#ifdef UNSIGNED_ARITH
+			*pval = (UNSIGNED_ARITH) *pval >= (UNSIGNED_ARITH) val;
+#else
+			*pval = (*pval & arith_sign ?
+				(val & arith_sign ? *pval >= val : 1) :
+				(val & arith_sign ? 0 : *pval >= val)
+			);
+#endif
+		}
+		else	*pval = (*pval >= val);
 		break;
 	case EQUAL:
 		*pval = (*pval == val);
@@ -74,6 +166,7 @@ ch3bin(pval, oper, val)
 		*pval = (*pval || val);
 		break;
 	case ',':
+		*pis_uns = is_uns;
 		*pval = val;
 		break;
 	}

lang/cem/cpp.ansi/ch3mon.c

@@ -8,8 +8,10 @@
 #include	"Lpars.h"
 #include	"arith.h"
 
-ch3mon(oper, pval)
+/*ARGSUSED2*/
+ch3mon(oper, pval, puns)
 	register arith *pval;
+	int *puns;
 {
 	switch (oper)	{
 	case '~':

lang/cem/cpp.ansi/domacro.c

@@ -665,7 +665,7 @@ get_text(formals, length)
 
 	c = GetChar();
 
-	repl->r_ptr = repl->r_text = Malloc(repl->r_size = ITEXTSIZE);
+	repl->r_ptr = repl->r_text = Malloc((unsigned)(repl->r_size = ITEXTSIZE));
 	*repl->r_ptr = '\0';
 	while ((c != EOI) && (class(c) != STNL)) {
 		if (BLANK(c)) {

lang/cem/cpp.ansi/error.c

@@ -23,12 +23,12 @@ err_hdr(s)
 	char *s;
 {
 	if (FileName) {
-		fprint(ERROUT, "\"%s\", line %d: %s", FileName, LineNumber, s);
+		fprint(ERROUT, "\"%s\", line %d: %s", FileName, (int)LineNumber, s);
 	}
 	else	fprint(ERROUT, s);
 }
 
-/*VARARGS1*/
+/*VARARGS*/
 error(va_alist)
 	va_dcl
 {
@@ -44,7 +44,7 @@ error(va_alist)
 	va_end(ap);
 }
 
-/*VARARGS1*/
+/*VARARGS*/
 warning(va_alist)
 	va_dcl
 {
@@ -59,7 +59,7 @@ warning(va_alist)
 	va_end(ap);
 }
 
-/*VARARGS1*/
+/*VARARGS*/
 strict(va_alist)
 	va_dcl
 {
@@ -74,7 +74,7 @@ strict(va_alist)
 	va_end(ap);
 }
 
-/*VARARGS1*/
+/*VARARGS*/
 crash(va_alist)
 	va_dcl
 {
@@ -90,7 +90,7 @@ crash(va_alist)
 	sys_stop(S_ABORT);
 }
 
-/*VARARGS1*/
+/*VARARGS*/
 fatal(va_alist)
 	va_dcl
 {

lang/cem/cpp.ansi/expression.g

@@ -15,70 +15,75 @@ extern arith ifval;
 }
 
 if_expression
+{ int is_unsigned = 0; }
 :
-	constant_expression(&ifval)
+	constant_expression(&ifval, &is_unsigned)
 ;
 
 /* 7.1 */
-primary(arith *pval;)
+primary(arith *pval; int *is_uns;)
 :
-	constant(pval)
+	constant(pval, is_uns)
 |
-	'(' expression(pval) ')'
+	'(' expression(pval, is_uns) ')'
 ;
 
-unary(arith *pval;)
+unary(arith *pval; int *is_uns;)
 	{int oper;}
 :
 	unop(&oper)
-	unary(pval)
+	unary(pval, is_uns)
-	{ ch3mon(oper, pval); }
+	{ ch3mon(oper, pval, is_uns); }
 |
-	primary(pval)
+	primary(pval, is_uns)
 ;
 
-binary_expression(int maxrank; arith *pval;)
+binary_expression(int maxrank; arith *pval; int *is_uns;)
-	{int oper; arith val1;}
+	{int oper; arith val1; int u;}
 :
-	unary(pval)
+	unary(pval, is_uns)
 	[%while (rank_of(DOT) <= maxrank)
 		binop(&oper)
-		binary_expression(rank_of(oper)-1, &val1)
+		binary_expression(rank_of(oper)-1, &val1, &u)
 		{
-			ch3bin(pval, oper, val1);
+			ch3bin(pval, is_uns, oper, val1, u);
 		}
 	]*
 ;
 
 /* 7.13 */
-conditional_expression(arith *pval;)
+conditional_expression(arith *pval; int *is_uns)
-	{arith val1 = 0, val2 = 0;}
+	{arith val1 = 0, val2 = 0; int u;}
 :
 	/* allow all binary operators */
-	binary_expression(rank_of('?') - 1, pval)
+	binary_expression(rank_of('?') - 1, pval, is_uns)
 	[	'?'
-		expression(&val1)
+		expression(&val1, is_uns)
 		':'
-		assignment_expression(&val2)
+		assignment_expression(&val2, &u)
-		{ *pval = (*pval ? val1 : val2); }
+		{ if (*pval) *pval = val1;
+		  else { *pval = val2; *is_uns = u; }
+		}
 	]?
 ;
 
 /* 7.14 */
-assignment_expression(arith *pval;)
+assignment_expression(arith *pval; int *is_uns)
 :
-	conditional_expression(pval)
+	conditional_expression(pval, is_uns)
 ;
 
 /* 7.15 */
-expression(arith *pval;)
+expression(arith *pval; int *is_uns)
-	{arith val1;}
+	{arith val1;
+	 int is_uns1;
+	}
 :
-	assignment_expression(pval)
+	assignment_expression(pval,is_uns)
 	[	','
-		assignment_expression(&val1)
+		assignment_expression(&val1, &is_uns1)
 		{
-			ch3bin(pval, ',', val1);
+			ch3bin(pval, is_uns, ',', val1, is_uns1);
 		}
 	]*
 ;
@@ -119,11 +124,13 @@ binop(int *oper;) :
 	{*oper = DOT;}
 ;
 
-constant(arith *pval;) :
+constant(arith *pval; int *is_uns) :
 	INTEGER
-	{*pval = dot.tk_val;}
+	{*pval = dot.tk_val;
+	 *is_uns = dot.tk_unsigned;
+	}
 ;
 
-constant_expression (arith *pval;) :
+constant_expression (arith *pval; int *is_uns) :
-	assignment_expression(pval)
+	assignment_expression(pval, is_uns)
 ;

lang/cem/cpp.ansi/init.c

@@ -32,6 +32,7 @@ struct mkey	{
 };
 
 char *strcpy();
+char *sprint();
 
 init_pp()
 {

lang/cem/cpp.ansi/input.c

@@ -12,6 +12,7 @@
 #define INP_VAR		finfo
 struct file_info	finfo;
 #include <inp_pkg.body>
+#include <alloc.h>
 
 char *
 getwdir(fn)

lang/cem/cpp.ansi/preprocess.c

@@ -25,6 +25,8 @@ char	bits[128];
 #endif
 extern int InputLevel;
 
+extern char *sprint();
+
 Xflush()
 {
 	sys_write(STDOUT, _obuf, OBUFSIZE);
@@ -46,7 +48,7 @@ static int pragma_nr;
 do_pragma()
 {
 	register int size = ITEXTSIZE;
-	char *cur_line = Malloc(size);
+	char *cur_line = Malloc((unsigned)size);
 	register char *c_ptr = cur_line;
 	register int c = GetChar();
 	register int delim = 0;
@@ -136,7 +138,7 @@ preprocess(fn)
 				register char *p = Xbuf;		\
 				sprint(Xbuf, "%s %d \"%s\"\n",		\
 					LINE_PREFIX,			\
-					LineNumber,			\
+					(int)LineNumber,		\
 					FileName);			\
 				op--;					\
 				while (op >= _obuf			\

lang/cem/cpp.ansi/replace.c

@@ -41,7 +41,7 @@ replace(idf)
 	if (idf->id_macro->mc_flag & NOREPLACE)
 		return 0;
 	repl = new_repl();
-	repl->r_ptr = repl->r_text = Malloc(repl->r_size = LAPBUF);
+	repl->r_ptr = repl->r_text = Malloc((unsigned)(repl->r_size = LAPBUF));
 	repl->r_args = new_args();
 	repl->r_idf = idf;
 	if (!expand_macro(repl, idf))
@@ -206,8 +206,8 @@ expand_defined(repl)
 newarg(args)
 	struct args *args;
 {
-	args->a_expptr = args->a_expbuf = Malloc(args->a_expsize = ARGBUF);
+	args->a_expptr = args->a_expbuf = Malloc((unsigned)(args->a_expsize = ARGBUF));
-	args->a_rawptr = args->a_rawbuf = Malloc(args->a_rawsize = ARGBUF);
+	args->a_rawptr = args->a_rawbuf = Malloc((unsigned)(args->a_rawsize = ARGBUF));
 }
 
 getactuals(repl, idf)
@@ -302,14 +302,23 @@ actual(repl)
 	register int ch = 0;
 	register int level = 0, nostashraw = 0;
 	int lastch;
+	static int Unstacked_missed;
 
 	while (1) {
 		lastch = ch;
 		ch = GetChar();
 
+		if (nostashraw 
+		    && nostashraw >= Unstacked_missed) {
+			nostashraw -= Unstacked_missed;
+			Unstacked_missed = 0;
+		}
 		if (Unstacked) {
 			nostashraw -= Unstacked;
-			if (nostashraw < 0) nostashraw = 0;
+			if (nostashraw < 0) {
+				Unstacked_missed = -nostashraw;
+				nostashraw = 0;
+			}
 			EnableMacros();
 		}
 		if (class(ch) == STIDF || class(ch) == STELL) {