* string.c (rb_str_sum): string may be altered. [ruby-dev:24381]

* eval.c (rb_f_eval): defer pointer retrieval to prevent unsafe
  sourcefile string modification.  [ruby-dev:24373]

* io.c (io_read): block string buffer modification during
  rb_io_fread() by freezing it temporarily. [ruby-dev:24366]

* io.c (rb_io_s_popen): mode argument may be altered.
  [ruby-dev:24375]

* file.c (rb_file_s_basename): ext argument may be altered.
  [ruby-dev:24377]

* enum.c (enum_sort_by): use NODE instead of 2 element arrays.
  [ruby-dev:24378]

* string.c (rb_str_chomp_bang): StringValue() may change the
  receiver.  [ruby-dev:24371]


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@6976 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,27 @@
+Sat Oct  2 00:42:20 2004  Yukihiro Matsumoto  <matz@ruby-lang.org>
+
+	* string.c (rb_str_sum): string may be altered.  [ruby-dev:24381]
+
+Fri Oct  1 11:40:14 2004  Yukihiro Matsumoto  <matz@ruby-lang.org>
+
+	* eval.c (rb_f_eval): defer pointer retrieval to prevent unsafe
+	  sourcefile string modification.  [ruby-dev:24373]
+
+	* io.c (io_read): block string buffer modification during
+	  rb_io_fread() by freezing it temporarily. [ruby-dev:24366]
+
+	* io.c (rb_io_s_popen): mode argument may be altered.
+	  [ruby-dev:24375]
+
+	* file.c (rb_file_s_basename): ext argument may be altered.
+	  [ruby-dev:24377]
+
+	* enum.c (enum_sort_by): use NODE instead of 2 element arrays.
+	  [ruby-dev:24378]
+
+	* string.c (rb_str_chomp_bang): StringValue() may change the
+	  receiver.  [ruby-dev:24371]
+
 Fri Oct  1 11:25:20 2004  Hidetoshi NAGAI  <nagai@ai.kyutech.ac.jp>
 
 	* ext/tk/lib/tk/grid.rb: revive TkGrid.grid

enum.c

@@ -389,26 +389,26 @@ static VALUE
 sort_by_i(i, ary)
     VALUE i, ary;
 {
-    VALUE v, e;
+    VALUE v;
+    NODE *memo;
 
     v = rb_yield(i);
-    e = rb_assoc_new(v, i);
+    memo = rb_node_newnode(NODE_MEMO, v, i, 0);
-    OBJ_FREEZE(e);
+    rb_ary_push(ary, (VALUE)memo);
-    rb_ary_push(ary, e);
     return Qnil;
 }
 
 static VALUE
 sort_by_cmp(values, ary)
-    VALUE values;
+    VALUE values, ary;
 {
-    VALUE a = RARRAY(values)->ptr[0];
+    NODE *a = (NODE*)RARRAY(values)->ptr[0];
-    VALUE b = RARRAY(values)->ptr[1];
+    NODE *b = (NODE*)RARRAY(values)->ptr[1];
 
-    /* pedantic check; they must be arrays */
+    /* pedantic check; they must be memo nodes */
-    Check_Type(a, T_ARRAY);
+    Check_Type(a, T_NODE);
-    Check_Type(b, T_ARRAY);
+    Check_Type(b, T_NODE);
-    return rb_funcall(RARRAY(a)->ptr[0], id_cmp, 1, RARRAY(b)->ptr[0]);
+    return rb_funcall(a->u1.value, id_cmp, 1, b->u1.value);
 }
 
 /*
@@ -498,8 +498,7 @@ enum_sort_by(obj)
 	rb_iterate(rb_ary_sort_bang, ary, sort_by_cmp, ary);
     }
     for (i=0; i<RARRAY(ary)->len; i++) {
-	VALUE e = RARRAY(ary)->ptr[i];
+	RARRAY(ary)->ptr[i] = RNODE(RARRAY(ary)->ptr[i])->u2.value;
-	RARRAY(ary)->ptr[i] = RARRAY(e)->ptr[1];
     }
     return ary;
 }

eval.c

@@ -6187,12 +6187,13 @@ rb_f_eval(argc, argv, self)
 	SafeStringValue(src);
     }
     if (argc >= 3) {
-	file = StringValuePtr(vfile);
+	StringValue(vfile);
     }
     if (argc >= 4) {
 	line = NUM2INT(vline);
     }
 
+    if (!NIL_P(vfile)) file = RSTRING(vfile)->ptr;
     if (NIL_P(scope) && ruby_frame->prev) {
 	struct FRAME *prev;
 	VALUE val;

file.c

@@ -1729,8 +1729,7 @@ rb_file_s_lchmod(argc, argv)
 {
     VALUE vmode;
     VALUE rest;
-    int mode;
+    long mode, n;
-    long n;
 
     rb_secure(2);
     rb_scan_args(argc, argv, "1*", &vmode, &rest);
@@ -2654,11 +2653,11 @@ rb_file_s_basename(argc, argv)
     VALUE *argv;
 {
     VALUE fname, fext, basename;
-    char *name, *p, *ext = NULL;
+    char *name, *p;
     int f;
 
     if (rb_scan_args(argc, argv, "11", &fname, &fext) == 2) {
-	ext = StringValueCStr(fext);
+	StringValue(fext);
     }
     StringValue(fname);
     if (RSTRING(fname)->len == 0 || !*(name = RSTRING(fname)->ptr))
@@ -2674,7 +2673,7 @@ rb_file_s_basename(argc, argv)
 #endif
     }
     else if (!(p = strrdirsep(name))) {
-	if (NIL_P(fext) || !(f = rmext(name, ext))) {
+	if (NIL_P(fext) || !(f = rmext(name, StringValueCStr(fext)))) {
 	    f = chompdirsep(name) - name;
 	    if (f == RSTRING(fname)->len) return fname;
 	}
@@ -2682,7 +2681,7 @@ rb_file_s_basename(argc, argv)
     }
     else {
 	while (isdirsep(*p)) p++; /* skip last / */
-	if (NIL_P(fext) || !(f = rmext(p, ext))) {
+	if (NIL_P(fext) || !(f = rmext(p, StringValueCStr(fext)))) {
 	    f = chompdirsep(p) - p;
 	}
     }

gc.c

@@ -92,7 +92,7 @@ static unsigned long malloc_increase = 0;
 static unsigned long malloc_limit = GC_MALLOC_LIMIT;
 static void run_final();
 static VALUE nomem_error;
-static void gc_internal();
+static void garbage_collect();
 
 void
 rb_memerror()
@@ -120,11 +120,11 @@ ruby_xmalloc(size)
     malloc_increase += size;
 
     if (malloc_increase > malloc_limit) {
-	gc_internal();
+	garbage_collect();
     }
     RUBY_CRITICAL(mem = malloc(size));
     if (!mem) {
-	gc_internal();
+	garbage_collect();
 	RUBY_CRITICAL(mem = malloc(size));
 	if (!mem) {
 	    rb_memerror();
@@ -161,7 +161,7 @@ ruby_xrealloc(ptr, size)
     malloc_increase += size;
     RUBY_CRITICAL(mem = realloc(ptr, size));
     if (!mem) {
-	gc_internal();
+	garbage_collect();
 	RUBY_CRITICAL(mem = realloc(ptr, size));
 	if (!mem) {
 	    rb_memerror();
@@ -381,7 +381,7 @@ rb_newobj()
 {
     VALUE obj;
 
-    if (!freelist) gc_internal();
+    if (!freelist) garbage_collect();
 
     obj = (VALUE)freelist;
     freelist = freelist->as.free.next;
@@ -1288,7 +1288,7 @@ int rb_setjmp (rb_jmp_buf);
 #endif /* __GNUC__ */
 
 static void
-gc_internal()
+garbage_collect()
 {
     struct gc_list *list;
     struct FRAME * volatile frame; /* gcc 2.7.2.3 -O2 bug??  */
@@ -1403,7 +1403,7 @@ gc_internal()
 void
 rb_gc()
 {
-    gc_internal();
+    garbage_collect();
     rb_gc_finalize_deferred();
 }
 

io.c

@@ -1194,11 +1194,13 @@ io_read(argc, argv, io)
 	StringValue(str);
 	rb_str_modify(str);
 	rb_str_resize(str,len);
+	FL_SET(str, FL_FREEZE);
     }
     if (len == 0) return str;
 
     READ_CHECK(fptr->f);
     n = rb_io_fread(RSTRING(str)->ptr, len, fptr->f);
+    FL_UNSET(str, FL_FREEZE);
     if (n == 0) {
 	rb_str_resize(str,0);
 	if (!fptr->f) return Qnil;
@@ -2972,7 +2974,9 @@ rb_io_s_popen(argc, argv, klass)
 	mode = rb_io_modenum_mode(FIX2INT(pmode), mbuf);
     }
     else {
-	mode = StringValuePtr(pmode);
+	strncpy(mbuf, StringValuePtr(pmode), sizeof(mbuf) - 1);
+	mbuf[sizeof(mbuf) - 1] = 0;
+	mode = mbuf;
     }
     tmp = rb_check_array_type(pname);
     if (!NIL_P(tmp)) {

process.c

@@ -1207,8 +1207,8 @@ rb_check_argv(argc, argv)
 	    rb_raise(rb_eArgError, "wrong first argument");
 	}
 	prog = RARRAY(tmp)->ptr[0];
-	SafeStringValue(prog);
 	argv[0] = RARRAY(tmp)->ptr[1];
+	SafeStringValue(prog);
     }
     for (i = 0; i < argc; i++) {
 	SafeStringValue(argv[i]);

re.c

@@ -1722,8 +1722,6 @@ rb_reg_initialize_m(argc, argv, self)
 	len = RREGEXP(argv[0])->len;
     }
     else {
-	s = StringValuePtr(argv[0]);
-	len = RSTRING(argv[0])->len;
 	if (argc >= 2) {
 	    if (FIXNUM_P(argv[1])) flags = FIX2INT(argv[1]);
 	    else if (RTEST(argv[1])) flags = RE_OPTION_IGNORECASE;
@@ -1749,6 +1747,8 @@ rb_reg_initialize_m(argc, argv, self)
 		break;
 	    }
 	}
+	s = StringValuePtr(argv[0]);
+	len = RSTRING(argv[0])->len;
     }
     rb_reg_initialize(self, s, len, flags);
     return self;

string.c

@@ -3139,7 +3139,7 @@ tr_setup_table(str, table, init)
 	buf[c & 0xff] = !cflag;
     }
     for (i=0; i<256; i++) {
-	table[i] = table[i]&&buf[i];
+	table[i] = table[i] && buf[i];
     }
 }
 
@@ -3598,6 +3598,17 @@ rb_f_split(argc, argv)
 }
 
 
+static inline void
+str_mod_check(s, p, len)
+    VALUE s;
+    char *p;
+    long len;
+{
+    if (RSTRING(s)->ptr != p || RSTRING(s)->len != len) {
+	rb_raise(rb_eRuntimeError, "string modified");
+    }
+}
+
 /*
  *  call-seq:
  *     str.each(separator=$/) {|substr| block }        => str
@@ -3672,8 +3683,7 @@ rb_str_each_line(argc, argv, str)
 	    line = rb_str_new5(str, s, p - s);
 	    OBJ_INFECT(line, str);
 	    rb_yield(line);
-	    if (RSTRING(str)->ptr != ptr || RSTRING(str)->len != len)
+	    str_mod_check(str, ptr, len);
-		rb_raise(rb_eArgError, "string modified");
 	    s = p;
 	}
     }
@@ -3843,11 +3853,13 @@ rb_str_chomp_bang(argc, argv, str)
 {
     VALUE rs;
     int newline;
-    char *p = RSTRING(str)->ptr;
+    char *p;
-    long len = RSTRING(str)->len, rslen;
+    long len, rslen;
 
     if (rb_scan_args(argc, argv, "01", &rs) == 0) {
+	len = RSTRING(str)->len;
 	if (len == 0) return Qnil;
+	p = RSTRING(str)->ptr;
 	rs = rb_rs;
 	if (rs == rb_default_rs) {
 	  smart_chomp:
@@ -3871,9 +3883,10 @@ rb_str_chomp_bang(argc, argv, str)
 	}
     }
     if (NIL_P(rs)) return Qnil;
-    if (len == 0) return Qnil;
-
     StringValue(rs);
+    len = RSTRING(str)->len;
+    if (len == 0) return Qnil;
+    p = RSTRING(str)->ptr;
     rslen = RSTRING(rs)->len;
     if (rslen == 0) {
 	while (len>0 && p[len-1] == '\n') {
@@ -4393,13 +4406,16 @@ rb_str_sum(argc, argv, str)
     VALUE vbits;
     int bits;
     char *p, *pend;
+    long len;
 
     if (rb_scan_args(argc, argv, "01", &vbits) == 0) {
 	bits = 16;
     }
     else bits = NUM2INT(vbits);
 
-    p = RSTRING(str)->ptr; pend = p + RSTRING(str)->len;
+    p = RSTRING(str)->ptr;
+    len = RSTRING(str)->len;
+    pend = p + len;
     if (bits > sizeof(long)*CHAR_BIT) {
 	VALUE res = INT2FIX(0);
 	VALUE mod;
@@ -4408,6 +4424,7 @@ rb_str_sum(argc, argv, str)
 	mod = rb_funcall(mod, '-', 1, INT2FIX(1));
 
 	while (p < pend) {
+	    str_mod_check(str, p, len);
 	    res = rb_funcall(res, '+', 1, INT2FIX((unsigned int)*p));
 	    p++;
 	}
@@ -4422,6 +4439,7 @@ rb_str_sum(argc, argv, str)
 	    mod = -1;
 	}
 	while (p < pend) {
+	    str_mod_check(str, p, len);
 	    res += (unsigned int)*p;
 	    p++;
 	}