* ext/objspace/objspace.c: add a new method ObjectSpace.count_symbols.

[Feature #11158]
* symbol.c (rb_sym_immortal_count): added to count immortal symbols.
* symbol.h: ditto.
* test/objspace/test_objspace.rb: add a test for this method.
* NEWS: describe about this method.



git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51654 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,16 @@
+Fri Aug 21 19:58:48 2015  Koichi Sasada  <ko1@atdot.net>
+
+	* ext/objspace/objspace.c: add a new method ObjectSpace.count_symbols.
+	  [Feature #11158]
+
+	* symbol.c (rb_sym_immortal_count): added to count immortal symbols.
+
+	* symbol.h: ditto.
+
+	* test/objspace/test_objspace.rb: add a test for this method.
+
+	* NEWS: describe about this method.
+
 Fri Aug 21 19:48:17 2015  Nobuyoshi Nakada  <nobu@ruby-lang.org>
 
 	* win32/Makefile.sub ($(LIBRUBY_SO)): needs additional libraries

NEWS

@@ -106,6 +106,7 @@ with all sufficient information, see the ChangeLog file.
     GC overhead
 
 * ObjectSpace (objspace)
+  * ObjectSpace.count_symbols is added.
   * ObjectSpace.count_imemo_objects is added.
   * ObjectSpace.internal_class_of is added.
   * ObjectSpace.internal_super_of is added.

ext/objspace/objspace.c

@@ -18,6 +18,7 @@
 #include <ruby/re.h>
 #include "node.h"
 #include "gc.h"
+#include "symbol.h"
 
 /*
  *  call-seq:
@@ -249,6 +250,84 @@ count_objects_size(int argc, VALUE *argv, VALUE os)
     return hash;
 }
 
+struct dynamic_symbol_counts {
+    size_t mortal;
+    size_t immortal;
+};
+
+static int
+cs_i(void *vstart, void *vend, size_t stride, void *n)
+{
+    struct dynamic_symbol_counts *counts = (struct dynamic_symbol_counts *)n;
+    VALUE v = (VALUE)vstart;
+
+    for (; v != (VALUE)vend; v += stride) {
+	if (RBASIC(v)->flags && BUILTIN_TYPE(v) == T_SYMBOL) {
+	    ID id = RSYMBOL(v)->id;
+	    if ((id & ~ID_SCOPE_MASK) == 0) {
+		counts->mortal++;
+	    }
+	    else {
+		counts->immortal++;
+	    }
+	}
+    }
+
+    return 0;
+}
+
+size_t rb_sym_immortal_count(void);
+
+/*
+ *  call-seq:
+ *     ObjectSpace.count_symbols([result_hash]) -> hash
+ *
+ *  Counts symbols for each Symbol type.
+ *
+ *  This method is only for MRI developers interested in performance and memory
+ *  usage of Ruby programs.
+ *
+ *  If the optional argument, result_hash, is given, it is overwritten and
+ *  returned. This is intended to avoid probe effect.
+ *
+ *  Note:
+ *  The contents of the returned hash is implementation defined.
+ *  It may be changed in future.
+ *
+ *  This method is only expected to work with C Ruby.
+ *
+ *  On this version of MRI, they have 3 types of Symbols (and 1 total counts).
+ *
+ *   * mortal_dynamic_symbol: GC target symbols (collected by GC)
+ *   * immortal_dynamic_symbol: Immortal symbols promoted from dynamic symbols (do not collected by GC)
+ *   * immortal_static_symbol: Immortal symbols (do not collected by GC)
+ *   * immortal_symbol: total immortal symbols (immortal_dynamic_symbol+immortal_static_symbol)
+ */
+
+static VALUE
+count_symbols(int argc, VALUE *argv, VALUE os)
+{
+    struct dynamic_symbol_counts dynamic_counts = {0, 0};
+    VALUE hash = setup_hash(argc, argv);
+
+    size_t immortal_symbols = rb_sym_immortal_count();
+    rb_objspace_each_objects(cs_i, &dynamic_counts);
+
+    if (hash == Qnil) {
+        hash = rb_hash_new();
+    }
+    else if (!RHASH_EMPTY_P(hash)) {
+        st_foreach(RHASH_TBL(hash), set_zero_i, hash);
+    }
+
+    rb_hash_aset(hash, ID2SYM(rb_intern("mortal_dynamic_symbol")),   SIZET2NUM(dynamic_counts.mortal));
+    rb_hash_aset(hash, ID2SYM(rb_intern("immortal_dynamic_symbol")), SIZET2NUM(dynamic_counts.immortal));
+    rb_hash_aset(hash, ID2SYM(rb_intern("immortal_static_symbol")),  SIZET2NUM(immortal_symbols - dynamic_counts.immortal));
+    rb_hash_aset(hash, ID2SYM(rb_intern("immortal_symbol")),         SIZET2NUM(immortal_symbols));
+
+    return hash;
+}
+
 static int
 cn_i(void *vstart, void *vend, size_t stride, void *n)
 {
@@ -887,6 +966,7 @@ Init_objspace(void)
     rb_define_module_function(rb_mObjSpace, "memsize_of_all", memsize_of_all_m, -1);
 
     rb_define_module_function(rb_mObjSpace, "count_objects_size", count_objects_size, -1);
+    rb_define_module_function(rb_mObjSpace, "count_symbols", count_symbols, -1);
     rb_define_module_function(rb_mObjSpace, "count_nodes", count_nodes, -1);
     rb_define_module_function(rb_mObjSpace, "count_tdata_objects", count_tdata_objects, -1);
     rb_define_module_function(rb_mObjSpace, "count_imemo_objects", count_imemo_objects, -1);

symbol.c

@@ -864,6 +864,12 @@ rb_sym_all_symbols(void)
     return ary;
 }
 
+size_t
+rb_sym_immortal_count(void)
+{
+    return (size_t)global_symbols.last_id;
+}
+
 int
 rb_is_const_id(ID id)
 {

symbol.h

@@ -100,4 +100,9 @@ is_global_name_punct(const int c)
 
 ID rb_intern_cstr_without_pindown(const char *, long, rb_encoding *);
 
+RUBY_SYMBOL_EXPORT_BEGIN
+
+size_t rb_sym_immortal_count(void);
+
+RUBY_SYMBOL_EXPORT_END
 #endif

test/objspace/test_objspace.rb

@@ -351,4 +351,15 @@ class TestObjSpace < Test::Unit::TestCase
     }
     assert_operator i, :>, 0
   end
+
+  def test_count_symbols
+    syms = (1..128).map{|i| ("xyzzy#{i}" * 128).to_sym}
+    c = Class.new{define_method(syms[-1]){}}
+
+    h = ObjectSpace.count_symbols
+    assert_operator h[:mortal_dynamic_symbol],   :>=, 128, h.inspect
+    assert_operator h[:immortal_dynamic_symbol], :>=, 1, h.inspect
+    assert_operator h[:immortal_static_symbol],  :>=, Object.methods.size, h.inspect
+    assert_equal h[:immortal_symbol], h[:immortal_dynamic_symbol] + h[:immortal_static_symbol], h.inspect
+  end
 end