Merge pull request #5032 from RoryO/add-objspace-memsize-of

Add ObjectSpace.memsize_of

1 files changed, 246 insertions, 0 deletions

diff --git a/mrbgems/mruby-objectspace/src/mruby_objectspace.c b/mrbgems/mruby-objectspace/src/mruby_objectspace.c
index b89fb0580..7892c6a1b 100644
--- a/mrbgems/mruby-objectspace/src/mruby_objectspace.c
+++ b/mrbgems/mruby-objectspace/src/mruby_objectspace.c
@@ -2,6 +2,14 @@
 #include <mruby/gc.h>
 #include <mruby/hash.h>
 #include <mruby/class.h>
+#include <mruby/object.h>
+#include <mruby/numeric.h>
+#include <mruby/string.h>
+#include <mruby/array.h>
+#include <mruby/variable.h>
+#include <mruby/proc.h>
+#include <mruby/value.h>
+#include <mruby/range.h>
 
 struct os_count_struct {
   mrb_int total;
@@ -168,12 +176,250 @@ os_each_object(mrb_state *mrb, mrb_value self)
   return mrb_fixnum_value(d.count);
 }
 
+static void os_memsize_of_object(mrb_state*,mrb_value,mrb_value,mrb_int*);
+
+struct os_memsize_cb_data {
+  mrb_int *t;
+  mrb_value recurse;
+};
+
+static int
+os_memsize_ivar_cb(mrb_state *mrb, mrb_sym _name, mrb_value obj, void *data)
+{
+  struct os_memsize_cb_data *cb_data = (struct os_memsize_cb_data *)(data);
+  os_memsize_of_object(mrb, obj, cb_data->recurse, cb_data->t);
+  return 0;
+}
+
+static void
+os_memsize_of_ivars(mrb_state* mrb, mrb_value obj, mrb_value recurse, mrb_int *t)
+{
+  (*t) += mrb_obj_iv_tbl_memsize(mrb, obj);
+  if(!mrb_nil_p(recurse)) {
+    struct os_memsize_cb_data cb_data = {t, recurse};
+    mrb_iv_foreach(mrb, obj, os_memsize_ivar_cb, &cb_data);
+  }
+}
+
+static void
+os_memsize_of_irep(mrb_state* state, struct mrb_irep *irep, mrb_int* t)
+{
+  mrb_int i;
+  (*t) += (irep->slen * sizeof(mrb_sym)) +
+          (irep->plen * sizeof(mrb_code)) +
+          (irep->ilen * sizeof(mrb_code));
+
+  for(i = 0; i < irep->rlen; i++) {
+    os_memsize_of_irep(state, irep->reps[i], t);
+  }
+}
+
+static void
+os_memsize_of_method(mrb_state* mrb, mrb_value method_obj, mrb_int* t)
+{
+  mrb_value proc_value = mrb_obj_iv_get(mrb, mrb_obj_ptr(method_obj),
+                                        mrb_intern_lit(mrb, "_proc"));
+  struct RProc *proc = mrb_proc_ptr(proc_value);
+
+  (*t) += sizeof(struct RProc);
+  if(!MRB_PROC_CFUNC_P(proc)) os_memsize_of_irep(mrb, proc->body.irep, t);
+}
+
+static void
+os_memsize_of_methods(mrb_state* mrb, mrb_value obj, mrb_int* t)
+{
+  mrb_value method_list;
+  mrb_int i;
+  if(!mrb_respond_to(mrb, obj, mrb_intern_lit(mrb, "instance_methods"))) return;
+  method_list = mrb_funcall(mrb, obj, "instance_methods", 1, mrb_false_value());
+  for(i = 0; i < RARRAY_LEN(method_list); i++) {
+    mrb_value method = mrb_funcall(mrb, obj, "instance_method", 1,
+                                   mrb_ary_ref(mrb, method_list, i));
+    os_memsize_of_method(mrb, method, t);
+  }
+}
+
+static void
+os_memsize_of_object(mrb_state* mrb, mrb_value obj, mrb_value recurse, mrb_int* t)
+{
+  if(!mrb_nil_p(recurse)) {
+    const mrb_value obj_id = mrb_fixnum_value(mrb_obj_id(obj));
+    if(mrb_hash_key_p(mrb, recurse, obj_id)) return;
+    mrb_hash_set(mrb, recurse, obj_id, mrb_true_value());
+  }
+
+  switch(obj.tt) {
+    case MRB_TT_STRING:
+      (*t) += RSTRING_LEN(obj);
+      break;
+    case MRB_TT_CLASS:
+    case MRB_TT_MODULE:
+    case MRB_TT_EXCEPTION:
+    case MRB_TT_SCLASS:
+    case MRB_TT_ICLASS:
+    case MRB_TT_OBJECT: {
+      (*t) += mrb_objspace_page_slot_size();
+      os_memsize_of_ivars(mrb, obj, recurse, t);
+      if(mrb_obj_is_kind_of(mrb, obj, mrb_class_get(mrb, "UnboundMethod"))) {
+        os_memsize_of_method(mrb, obj, t);
+      }
+      else {
+        os_memsize_of_methods(mrb, obj, t);
+      }
+      break;
+    }
+    case MRB_TT_HASH: {
+      (*t) += mrb_objspace_page_slot_size() +
+              mrb_os_memsize_of_hash_table(obj);
+      if(!mrb_nil_p(recurse)) {
+        os_memsize_of_object(mrb, mrb_hash_keys(mrb, obj), recurse, t);
+        os_memsize_of_object(mrb, mrb_hash_values(mrb, obj), recurse, t);
+      }
+      break;
+    }
+    case MRB_TT_ARRAY: {
+      mrb_int len, i;
+      len = RARRAY_LEN(obj);
+      /* Arrays that do not fit within an RArray perform a heap allocation
+      *  storing an array of pointers to the original objects*/
+      (*t) += mrb_objspace_page_slot_size();
+      if(len > MRB_ARY_EMBED_LEN_MAX) (*t) += sizeof(mrb_value *) * len;
+
+      if(!mrb_nil_p(recurse)) {
+        for(i = 0; i < len; i++) {
+          os_memsize_of_object(mrb, ARY_PTR(mrb_ary_ptr(obj))[i], recurse, t);
+        }
+      }
+      break;
+    }
+    case MRB_TT_PROC: {
+      struct RProc* proc = mrb_proc_ptr(obj);
+      (*t) += mrb_objspace_page_slot_size();
+      (*t) += MRB_ENV_LEN(proc->e.env) * sizeof(mrb_value);
+      if(!MRB_PROC_CFUNC_P(proc)) os_memsize_of_irep(mrb, proc->body.irep, t);
+      break;
+    }
+    case MRB_TT_DATA:
+      (*t) += mrb_objspace_page_slot_size();
+      if(mrb_respond_to(mrb, obj, mrb_intern_lit(mrb, "memsize"))) {
+        (*t) += mrb_fixnum(mrb_funcall(mrb, obj, "memsize", 0));
+      }
+      break;
+    #ifndef MRB_WITHOUT_FLOAT
+    case MRB_TT_FLOAT:
+      #ifdef MRB_WORD_BOXING
+      (*t) += mrb_objspace_page_slot_size() +
+              sizeof(struct RFloat);
+      #endif
+      break;
+    #endif
+    case MRB_TT_RANGE:
+    #ifndef MRB_RANGE_EMBED
+      (*t) += mrb_objspace_page_slot_size() +
+              sizeof(struct mrb_range_edges);
+    #endif
+      break;
+    case MRB_TT_FIBER: {
+      struct RFiber* f = (struct RFiber *)mrb_ptr(obj);
+      mrb_callinfo *ci_p = f->cxt->cibase;
+      ptrdiff_t stack_size = f->cxt->stend - f->cxt->stbase;
+      ptrdiff_t ci_size = f->cxt->ciend - f->cxt->cibase;
+      mrb_int i = 0;
+
+      while(ci_p < f->cxt->ciend) {
+        if(ci_p->proc) os_memsize_of_irep(mrb, ci_p->proc->body.irep, t);
+        ci_p++;
+      }
+
+      if(f->cxt->esize) {
+        for(i = 0; i <= f->cxt->esize; i++) {
+          os_memsize_of_irep(mrb, f->cxt->ensure[i]->body.irep, t);
+        }
+      }
+
+      (*t) += mrb_objspace_page_slot_size() +
+        sizeof(struct RFiber) +
+        sizeof(struct mrb_context) +
+        sizeof(struct RProc *) * f->cxt->esize +
+        sizeof(uint16_t *) * f->cxt->rsize +
+        stack_size +
+        ci_size;
+      break;
+    }
+    case MRB_TT_ISTRUCT:
+      (*t) += mrb_objspace_page_slot_size();
+      break;
+    /*  zero heap size types.
+     *  immediate VM stack values, contained within mrb_state, or on C stack */
+    case MRB_TT_TRUE:
+    case MRB_TT_FALSE:
+    case MRB_TT_FIXNUM:
+    case MRB_TT_BREAK:
+    case MRB_TT_CPTR:
+    case MRB_TT_SYMBOL:
+    case MRB_TT_FREE:
+    case MRB_TT_UNDEF:
+    case MRB_TT_ENV:
+    /* never used, silences compiler warning
+     * not having a default: clause lets the compiler tell us when there is a new
+     * TT not accounted for */
+    case MRB_TT_MAXDEFINE:
+      break;
+  }
+}
+
+/*
+ *  call-seq:
+ *    ObjectSpace.memsize_of(obj, recurse: false) -> Numeric
+ *
+ *  Returns the amount of heap memory allocated for object in size_t units.
+ *
+ *  The return value depends on the definition of size_t on that platform,
+ *  therefore the value is not comparable across platform types.
+ *
+ *  Immediate values such as integers, booleans, symbols and unboxed float numbers
+ *  return 0. Additionally special objects which are small enough to fit inside an
+ *  object pointer, termed embedded objects, will return the size of the object pointer.
+ *  Strings and arrays below a compile-time defined size may be embedded.
+ *
+ *  Setting recurse: true descends into instance variables, array members,
+ *  hash keys and hash values recursively, calculating the child objects and adding to
+ *  the final sum. It avoids infinite recursion and over counting objects by
+ *  internally tracking discovered object ids.
+ *
+ *  MRB_TT_DATA objects aren't calculated beyond their original page slot. However,
+ *  if the object implements a memsize method it will call that method and add the
+ *  return value to the total. This provides an opportunity for C based data structures
+ *  to report their memory usage.
+ *
+ */
+
+static mrb_value
+os_memsize_of(mrb_state *mrb, mrb_value self)
+{
+  mrb_int total;
+  mrb_value obj;
+  mrb_value recurse;
+  const char *kw_names[1] = { "recurse" };
+  mrb_value kw_values[1];
+  const mrb_kwargs kwargs = { 1, kw_values, kw_names, 0, NULL };
+
+  mrb_get_args(mrb, "o:", &obj, &kwargs);
+  recurse = mrb_obj_eq(mrb, kw_values[0], mrb_true_value())? mrb_hash_new(mrb) : mrb_nil_value();
+
+  total = 0;
+  os_memsize_of_object(mrb, obj, recurse, &total);
+
+  return mrb_fixnum_value(total);
+}
+
 void
 mrb_mruby_objectspace_gem_init(mrb_state *mrb)
 {
   struct RClass *os = mrb_define_module(mrb, "ObjectSpace");
   mrb_define_class_method(mrb, os, "count_objects", os_count_objects, MRB_ARGS_OPT(1));
   mrb_define_class_method(mrb, os, "each_object", os_each_object, MRB_ARGS_OPT(1));
+  mrb_define_class_method(mrb, os, "memsize_of", os_memsize_of, MRB_ARGS_REQ(1)|MRB_ARGS_OPT(1));
 }
 
 void