vmCompile() checks that all PC destinations are in range. STRING and CLASS opcodes read the argument length from their string, ignoring the explicit length given in the program.

minproto.leg

@@ -1,6 +1,6 @@
 # minproto.leg -- minimal prototype langauge for semantic experiments
 #
-# last edited: 2024-06-20 02:30:46 by piumarta on m1mbp
+# last edited: 2024-06-20 03:14:11 by piumarta on m1mbp
 
 %{
     ;
@@ -4852,7 +4852,8 @@ vmInsn *vmCompile(oop grammar, oop symbol)
     oop  *prog  = _get(program, Object,indexed);
     int   plen  = _get(program, Object,isize);
     if (plen % 5 != 0) valueError("__match__", "program length is not a multiple of five", program);
-    vmInsn *code = calloc(plen / 5, sizeof(*code));
+    int   clen  = plen / 5;
+    vmInsn *code = calloc(clen, sizeof(*code));
     vmCachePut(grammar, symbol, code);
     oop     env  = nil;
     int     ppc  = 0;
@@ -4863,11 +4864,14 @@ vmInsn *vmCompile(oop grammar, oop symbol)
         int arglen = integerValue(prog[ppc++], "__match__");
         int ok     = integerValue(prog[ppc++], "__match__");
         int ko     = integerValue(prog[ppc++], "__match__");
+        if (ok < 0 || ok >= clen) valueError("__match__", "OK destination out of range", program);
+        if (ko < 0 || ko >= clen) valueError("__match__", "KO destination out of range", program);
         code[cpc]  = (vmInsn){ .arg.obj = arg, .op = op, .arglen = arglen, .ok = ok, .ko = ko };
         switch (op) {
             case CLASS:
             case STRING: {
-                code[cpc].arg.str = stringValue(code[cpc].arg.obj, "__match__");
+                code[cpc].arglen  = stringLength(code[cpc].arg.obj, "__match__");
+                code[cpc].arg.str = stringValue (code[cpc].arg.obj, "__match__");
                 break;
             }
             case RULE: {
@@ -4875,15 +4879,15 @@ vmInsn *vmCompile(oop grammar, oop symbol)
                 break;
             }
             case CALL: {
-                valueError("__match__", "program contains CALL opcode", program);
+                valueError("__match__", "program contains explicit CALL opcode", program);
                 break;
             }
             case ACTION: {
-                if (!isClosure(code[cpc].arg.obj)) valueError("__match__", "ACTION argument is not a function", program);
+                if (!isClosure(code[cpc].arg.obj)) valueError("__match__", "ACTION argument must be a closure", program);
                 break;
             }
             case SET: {
-                if (!is(Symbol, code[cpc].arg.obj)) valueError("__match__", "SET argument is not a symbol", program);
+                if (!is(Symbol, code[cpc].arg.obj)) valueError("__match__", "SET argument must be a symbol", program);
                 if (nil == env) env = new(pObject);
                 Object_put(env, code[cpc].arg.obj, nil);
                 break;
@@ -4894,10 +4898,6 @@ vmInsn *vmCompile(oop grammar, oop symbol)
         }
         ++cpc;
     }
-#if 0
-    if (!cpc || code[0].op != ENTER) valueError("__match__", "program does not begin with ENTER", program);
-    code[0].arg.obj = env; // clone this to make a list of local variables for the rule
-#endif
 
     if (opt_d) {
         printf("---- BEGIN ");  println(symbol, 0);
@@ -5098,7 +5098,7 @@ int vmRun(oop grammar, oop symbol, char *text, int start, int length)
                 continue;
             }
             default: {
-                fatal("this cannot happen");
+                fatal("this cannot happen %d", i->op);
                 break;
             }
         }