checkin: found a large bug associated to ident resetting the buffer vector

4 files changed, 59 insertions, 33 deletions

diff --git a/sys/cmd/cc/cc.c b/sys/cmd/cc/cc.c
index e348893..74040eb 100644
--- a/sys/cmd/cc/cc.c
+++ b/sys/cmd/cc/cc.c
@@ -286,10 +286,10 @@ compile(byte *path)
         .line = 1,
         .col  = 1,
     };
+
     while (tok = lex(&C.lxr), tok.kind > Aeof) {
-        puttok(tok);
+        // puttok(tok);
     }
-    freeio(C.lxr.io);
 
     return tok.kind != Anil;
 }
diff --git a/sys/cmd/cc/cc.h b/sys/cmd/cc/cc.h
index dab351a..8a9fd80 100644
--- a/sys/cmd/cc/cc.h
+++ b/sys/cmd/cc/cc.h
@@ -92,6 +92,7 @@ struct Range
 };
 
 void errorat(Pos x, byte *fmt, ...);
+void warnat(Pos x, byte *fmt, ...);
 
 /* pre-processor */
 #define DIRECTIVES                          \
diff --git a/sys/cmd/cc/lex.c b/sys/cmd/cc/lex.c
index fc9dc0f..2a637da 100644
--- a/sys/cmd/cc/lex.c
+++ b/sys/cmd/cc/lex.c
@@ -38,9 +38,12 @@ getnsbyte(Lexer *lx)
     b = getbyte(lx);
     for (;;) {
         if (b == EOF) {
-            popio(lx);
-            b = getbyte(lx);
-            continue;
+            if (lx->io->link) {
+                popio(lx);
+                b = getbyte(lx);
+                continue;
+            } else
+                return b;
         }
         if (b >= RuneSelf || !isspace(b))
             return b;
@@ -293,6 +296,12 @@ Dispatch:
                 lx->pos.line++;
         goto Dispatch;
 
+    case '\\':
+        b = getbyte(lx);
+        if (b != '\n')
+            errorat(lx->pos, "'\\' without a trailing newline");
+        goto GetByte;
+
     Tchar:
     case '\'':
         if (escapechar(lx, '\'', 0, 0, &v)) {
@@ -493,7 +502,10 @@ Dispatch:
             }
             lx->b = s;
             /* reparse number, now with base info */
-            while (b = getbyte(lx), (isdigit(b) || ('a' <= b && b <= 'f') || ('A' <= b && b <= 'F') || b == '_'))
+            while (b = getbyte(lx), (isdigit(b) || 
+                         ('a' <= b && b <= 'f') || 
+                         ('A' <= b && b <= 'F') || 
+                         b == '_'))
                 *lx->b++ = b;
         }
     Rint:
@@ -545,10 +557,12 @@ Dispatch:
                 tok.kind |= Vun;
                 goto Return;
             }
+
             ungetbyte(lx);
-        } else
-            tok.kind |= Vint;
+            goto Return;
+        }
 
+        tok.kind |= Vint;
         ungetbyte(lx);
         goto Return;
 
@@ -624,7 +638,8 @@ Dispatch:
         sym = lookup(&lx->sym, tok.val.s);
         if (sym) {
             io = makeio();
-            io->buf.end += expandmacro(lx, sym, io->b);
+            io->buf.end   += expandmacro(lx, sym, io->b);
+            *io->buf.end++ = EOF;
             pushio(lx, io);
             goto GetByte;
         }
@@ -673,17 +688,14 @@ popio(Lexer *lx)
     Io *prev;
 
     prev = lx->io->link;
+    freeio(lx->io);
+
+    lx->io = prev;
     if (!prev) {
-        panicf("no buffer left");
+        return;
     }
-    /* 
-     * NOTE: should we copy over unget bytes here?
-     *       would modify the buffer...
-     */
-    freeio(lx->io);
 
     lx->pos = prev->store;
-    lx->io  = prev;
 }
 
 // -----------------------------------------------------------------------
@@ -763,5 +775,18 @@ errorat(Pos x, byte *fmt, ...)
     printf("\n");
 
     va_end(args);
+    assert(0);
 }
 
+void
+warnat(Pos x, byte *fmt, ...)
+{
+    va_list args;
+    va_start(args, fmt);
+
+    printf("warning:%s:%d:%d: ", os·basename(x.path), x.line, x.col);
+    vprintf(fmt, args);
+    printf("\n");
+
+    va_end(args);
+}
diff --git a/sys/cmd/cc/pp.c b/sys/cmd/cc/pp.c
index 2aa1291..264f11a 100644
--- a/sys/cmd/cc/pp.c
+++ b/sys/cmd/cc/pp.c
@@ -34,7 +34,7 @@ string
 identdots(Lexer *lx, int *dots)
 {
     int    c;
-    string s;
+    byte  *s;
 
     s = ident(lx);
     if (s != nil)
@@ -62,7 +62,7 @@ defmacro(Lexer *lx, string name, string macro)
 {
     Sym *mac;
 
-    printf("DEFINING MACRO %s = %s, ON LINE %d\n", name, macro+2, lx->pos.line);
+    printf("DEFINING MACRO %s ON LINE %d, file %s\n", name, lx->pos.line, os·basename(lx->pos.path));
     mac = define(&lx->sym, name, Smacro);
     mac->macro = macro;
 
@@ -397,6 +397,7 @@ enum
 // -----------------------------------------------------------------------
 // preprocessor functions
 
+/* #endif */
 static
 error
 ppend(Lexer *lx)
@@ -426,10 +427,9 @@ ppund(Lexer *lx)
 
     printf("FORGETTING %s\n", s);
     err = forget(&lx->sym, s);
-    if (err) {
-        errorat(lx->pos, "attempting to undefine unrecognized symbol '%s'", s);
-        return err;
-    }
+    if (err) 
+        warnat(lx->pos, "attempting to undefine unrecognized symbol '%s'", s);
+
     ppend(lx);
     return 0;
 }
@@ -442,7 +442,7 @@ ppdef(Lexer *lx)
     int    b;
     Sym   *sym;
     int    i, j, f, n, dot;
-    string s, a, buf, args[PPnarg];
+    string s, a, base, buf, args[PPnarg];
 
     s = ident(lx);
     if (!s) {
@@ -452,11 +452,8 @@ ppdef(Lexer *lx)
     intern(&s);
 
     sym = lookup(&lx->sym, s);
-    if (sym) {
-        // TODO: should be a warning...
-        errorat(lx->pos, "macro redefined: '%s'", sym->name);
-        // goto Bad;
-    }
+    if (sym)
+        warnat(lx->pos, "macro redefined: '%s'", sym->name);
 
     n   = 0;
     dot = 0;
@@ -494,10 +491,10 @@ ppdef(Lexer *lx)
         if (b != '\n')
             b = getnsbyte(lx);
 
-    buf = str·makef("%c%c", n, PPbeg);
+    base = lx->b;
+    buf  = str·makef("%c%c", n, PPbeg);
     for (;;) {
         if (isalpha(b) || b == '_') {
-            lx->b    = lx->buf;
             *lx->b++ = b;
 
             b = getbyte(lx);
@@ -505,7 +502,7 @@ ppdef(Lexer *lx)
                 *lx->b++ = b;
                 b = getbyte(lx);
             }
-            *lx->b = '\0';
+            *lx->b++ = '\0';
 
             for (i = 0; i < n; i++) {
                 if (strcmp(lx->buf, args[i]) == 0) {
@@ -999,6 +996,10 @@ Skip:
                 lx->pos.line++;
                 b = 1;
             }
+            if (c == EOF) {
+                errorat(lx->pos, "EOF hit while skipping if block. Missing endif");
+                goto Bad;
+            }
             continue;
         }
         if (!b)
@@ -1007,7 +1008,7 @@ Skip:
         if (!s) 
             continue;
 
-        if ((strcmp(s, "elif") == 0) && l == 0)
+        if (l == 0 && (strcmp(s, "elif") == 0))
             ppif(lx, 0);
 
         if (strcmp(s, "endif") == 0) {
@@ -1019,7 +1020,6 @@ Skip:
             return 0;
         }
         if (strcmp(s, "if")    == 0 || 
-            strcmp(s, "elif")  == 0 || 
             strcmp(s, "ifdef") == 0 || 
             strcmp(s, "ifndef") == 0) {
             l++;