fix: arguments now hold their state correctly throughout macro parsing

2 files changed, 41 insertions, 15 deletions

diff --git a/sys/cmd/cc/cc.c b/sys/cmd/cc/cc.c
index 74040eb..d7bb12c 100644
--- a/sys/cmd/cc/cc.c
+++ b/sys/cmd/cc/cc.c
@@ -249,6 +249,7 @@ init(void)
     memset(C.iostk, 0, sizeof(C.iostk));
 
     C.lxr     = (Lexer){ 0 };
+    C.lxr.b   = C.lxr.buf;
 }
 
 error
diff --git a/sys/cmd/cc/pp.c b/sys/cmd/cc/pp.c
index 264f11a..6ab4ecd 100644
--- a/sys/cmd/cc/pp.c
+++ b/sys/cmd/cc/pp.c
@@ -3,6 +3,8 @@
 // -----------------------------------------------------------------------
 // helper functions
 
+// NOTE: The iterator of lexer lx->b IS NOT reset.
+//       Its the caller's responsibility.
 static
 string
 ident(Lexer *lx)
@@ -16,17 +18,18 @@ ident(Lexer *lx)
         return "";
     }
 
-    for (s = lx->buf;;) {
-        *s++ = b;
+    s = lx->b;
+    for (;;) {
+        *lx->b++ = b;
         b = getbyte(lx);
         if (isalnum(b) || b == '_' || b >= RuneSelf)
             continue;
         ungetbyte(lx);
         break;
     }
-    *s = '\0';
+    *lx->b++ = '\0';
 
-    return lx->buf;
+    return s;
 }
 
 static
@@ -424,6 +427,7 @@ ppund(Lexer *lx)
 
     s = ident(lx);
     intern(&s);
+    lx->b = lx->buf;
 
     printf("FORGETTING %s\n", s);
     err = forget(&lx->sym, s);
@@ -442,7 +446,7 @@ ppdef(Lexer *lx)
     int    b;
     Sym   *sym;
     int    i, j, f, n, dot;
-    string s, a, base, buf, args[PPnarg];
+    string s, a, base, end, buf, args[PPnarg];
 
     s = ident(lx);
     if (!s) {
@@ -450,6 +454,7 @@ ppdef(Lexer *lx)
         goto Bad;
     }
     intern(&s);
+    lx->b = lx->buf;
 
     sym = lookup(&lx->sym, s);
     if (sym)
@@ -463,6 +468,7 @@ ppdef(Lexer *lx)
         if (b != ')') {
             ungetbyte(lx);
             for (;;) {
+                // NOTE: This is a pointer to the lx->buffer.
                 a = identdots(lx, &dot);
                 if (a == nil) {
                     errorat(lx->pos, "macro syntax error: improper argument");
@@ -492,24 +498,34 @@ ppdef(Lexer *lx)
             b = getnsbyte(lx);
 
     base = lx->b;
+    end  = lx->buf + arrlen(lx->buf);
+    if (base >= end) {
+        errorat(lx->pos, "out of macro buffer space!");
+        goto Bad;
+    }
     buf  = str·makef("%c%c", n, PPbeg);
     for (;;) {
         if (isalpha(b) || b == '_') {
+            lx->b    = base;
             *lx->b++ = b;
 
             b = getbyte(lx);
             while (isalnum(b) || b == '_') {
                 *lx->b++ = b;
+                if (lx->b >= end) {
+                    errorat(lx->pos, "out of macro buffer space!");
+                    goto Bad;
+                }
                 b = getbyte(lx);
             }
             *lx->b++ = '\0';
 
             for (i = 0; i < n; i++) {
-                if (strcmp(lx->buf, args[i]) == 0) {
+                if (strcmp(base, args[i]) == 0) {
                     goto Args;
                 }
             }
-            str·appendlen(&buf, (lx->b - lx->buf), lx->buf);
+            str·appendlen(&buf, (lx->b - base - 1), base);
             continue;
         Args:
             /* Check for argx ## argy OR argx##argy */
@@ -525,15 +541,19 @@ ppdef(Lexer *lx)
                     goto Arg;
                 }
                 b     = getnsbyte(lx);
-                lx->b = lx->buf;
+                lx->b = base;
                 while (isalnum(b) || b == '_') {
                     *lx->b++ = b;
+                    if (lx->b >= end) {
+                        errorat(lx->pos, "out of macro buffer space!");
+                        goto Bad;
+                    }
                     b = getbyte(lx);
                 }
                 *lx->b = '\0';
 
                 for (j = 0; j < n; j++) {
-                    if (strcmp(lx->buf, args[j]) == 0)
+                    if (strcmp(base, args[j]) == 0)
                         goto CatArgs;
                 }
                 errorat(lx->pos, "macro operator '##' must be terminated by valid variable identifier");
@@ -612,7 +632,7 @@ ppdef(Lexer *lx)
                 goto Bad;
             }
 
-            lx->b = lx->buf;
+            lx->b = base;
             while (isalnum(b) || b == '_') {
                 *lx->b++ = b;
                 b = getbyte(lx);
@@ -620,7 +640,7 @@ ppdef(Lexer *lx)
             *lx->b = '\0';
 
             for (i = 0; i < n; i++) {
-                if (strcmp(lx->buf, args[i]) == 0)
+                if (strcmp(base, args[i]) == 0)
                     goto Str;
             }
             errorat(lx->pos, "macro operator '#' must be followed by a valid variable identifier");
@@ -667,9 +687,6 @@ expandmacro(Lexer *lx, Sym *s, byte *dst)
         strcpy(dst, s->macro + 2);
         return str·len(s->macro)-2;
     }
-    printf("%c\n", s->macro[0]);
-    printf("%c\n", s->macro[1]);
-
     dots  =  *s->macro & PPvar;
     nargs = (*s->macro & ~PPvar) - 1;
 
@@ -854,6 +871,7 @@ Nospace:
     errorat(lx->pos, "out of memory during macro expansion %s", s->name);
 Bad:
     ppend(lx);
+    lx->b = lx->buf;
     errorat(lx->pos, "failed to expand macro %s", s->name);
     return -1;
 }
@@ -925,6 +943,7 @@ ppinc(Lexer *lx)
 
 Bad:
     ungetbyte(lx);
+    lx->b = lx->buf;
     errorat(lx->pos, "failed include");
     ppend(lx);
     return 1;
@@ -942,11 +961,13 @@ ppprag(Lexer *lx)
         errorat(lx->pos, "failed to parse pragma identifier");
         goto Bad;
     }
+    lx->b = lx->buf;
     if (strcmp(s, "once") == 0) {
         pushomit(lx->io->path);
         return 0;
     }
 Bad:
+    lx->b = lx->buf;
     errorat(lx->pos, "unrecognized pragma '%s'", s);
     ppend(lx);
     return 1;
@@ -979,6 +1000,7 @@ ppif(Lexer *lx, int f)
         goto Bad;
     }
     intern(&s);
+    lx->b = lx->buf;
 
     sym = lookup(&lx->sym, s);
     if ((!sym && (f == 3)) || (sym && (f == 2)))
@@ -1004,7 +1026,8 @@ Skip:
         }
         if (!b)
             continue;
-        s = ident(lx);
+        s     = ident(lx);
+        lx->b = lx->buf;
         if (!s) 
             continue;
 
@@ -1032,6 +1055,7 @@ Skip:
     }
 
 Bad:
+    lx->b = lx->buf;
     errorat(lx->pos, "bad syntax in preprocessor conditional directive");
     ppend(lx);
     return 1;
@@ -1088,6 +1112,7 @@ domacro(Lexer *lx)
 
     s = ident(lx);
     intern(&s);
+    lx->b = lx->buf;
     for (n = 0; n < NUM_DIRECTIVES; n++) {
         if ((uintptr)s == (uintptr)directives[n]) {
             goto Do;