ruby/gc.c

7715 строки
196 KiB
C
Исходник Обычный вид История

/**********************************************************************
gc.c -
$Author$
created at: Tue Oct 5 09:44:46 JST 1993
* encoding.c: provide basic features for M17N. * parse.y: encoding aware parsing. * parse.y (pragma_encoding): encoding specification pragma. * parse.y (rb_intern3): encoding specified symbols. * string.c (rb_str_length): length based on characters. for older behavior, bytesize method added. * string.c (rb_str_index_m): index based on characters. rindex as well. * string.c (succ_char): encoding aware succeeding string. * string.c (rb_str_reverse): reverse based on characters. * string.c (rb_str_inspect): encoding aware string description. * string.c (rb_str_upcase_bang): encoding aware case conversion. downcase, capitalize, swapcase as well. * string.c (rb_str_tr_bang): tr based on characters. delete, squeeze, tr_s, count as well. * string.c (rb_str_split_m): split based on characters. * string.c (rb_str_each_line): encoding aware each_line. * string.c (rb_str_each_char): added. iteration based on characters. * string.c (rb_str_strip_bang): encoding aware whitespace stripping. lstrip, rstrip as well. * string.c (rb_str_justify): encoding aware justifying (ljust, rjust, center). * string.c (str_encoding): get encoding attribute from a string. * re.c (rb_reg_initialize): encoding aware regular expression * sprintf.c (rb_str_format): formatting (i.e. length count) based on characters. * io.c (rb_io_getc): getc to return one-character string. for older behavior, getbyte method added. * ext/stringio/stringio.c (strio_getc): ditto. * io.c (rb_io_ungetc): allow pushing arbitrary string at the current reading point. * ext/stringio/stringio.c (strio_ungetc): ditto. * ext/strscan/strscan.c: encoding support. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@13261 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2007-08-25 07:29:39 +04:00
Copyright (C) 1993-2007 Yukihiro Matsumoto
Copyright (C) 2000 Network Applied Communication Laboratory, Inc.
Copyright (C) 2000 Information-technology Promotion Agency, Japan
**********************************************************************/
#include "ruby/ruby.h"
#include "ruby/st.h"
#include "ruby/re.h"
#include "ruby/io.h"
#include "ruby/thread.h"
#include "ruby/util.h"
#include "ruby/debug.h"
#include "eval_intern.h"
#include "vm_core.h"
#include "internal.h"
#include "gc.h"
#include "constant.h"
#include "ruby_atomic.h"
* probes.d: add DTrace probe declarations. [ruby-core:27448] * array.c (empty_ary_alloc, ary_new): added array create DTrace probe. * compile.c (rb_insns_name): allowing DTrace probes to access instruction sequence name. * Makefile.in: translate probes.d file to appropriate header file. * common.mk: declare dependencies on the DTrace header. * configure.in: add a test for existence of DTrace. * eval.c (setup_exception): add a probe for when an exception is raised. * gc.c: Add DTrace probes for mark begin and end, and sweep begin and end. * hash.c (empty_hash_alloc): Add a probe for hash allocation. * insns.def: Add probes for function entry and return. * internal.h: function declaration for compile.c change. * load.c (rb_f_load): add probes for `load` entry and exit, require entry and exit, and wrapping search_required for load path search. * object.c (rb_obj_alloc): added a probe for general object creation. * parse.y (yycompile0): added a probe around parse and compile phase. * string.c (empty_str_alloc, str_new): DTrace probes for string allocation. * test/dtrace/*: tests for DTrace probes. * vm.c (vm_invoke_proc): add probes for function return on exception raise, hash create, and instruction sequence execution. * vm_core.h: add probe declarations for function entry and exit. * vm_dump.c: add probes header file. * vm_eval.c (vm_call0_cfunc, vm_call0_cfunc_with_frame): add probe on function entry and return. * vm_exec.c: expose instruction number to instruction name function. * vm_insnshelper.c: add function entry and exit probes for cfunc methods. * vm_insnhelper.h: vm usage information is always collected, so uncomment the functions. 12 19:14:50 2012 Akinori MUSHA <knu@iDaemons.org> * configure.in (isinf, isnan): isinf() and isnan() are macros on DragonFly which cannot be found by AC_REPLACE_FUNCS(). This workaround enforces the fact that they exist on DragonFly. 12 15:59:38 2012 Shugo Maeda <shugo@ruby-lang.org> * vm_core.h (rb_call_info_t::refinements), compile.c (new_callinfo), vm_insnhelper.c (vm_search_method): revert r37616 because it's too slow. [ruby-dev:46477] * test/ruby/test_refinement.rb (test_inline_method_cache): skip the test until the bug is fixed efficiently. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37631 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-11-13 01:52:12 +04:00
#include "probes.h"
#include <stdio.h>
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#include <stdarg.h>
#include <setjmp.h>
#include <sys/types.h>
#include <assert.h>
#ifndef __has_feature
# define __has_feature(x) 0
#endif
#ifndef HAVE_MALLOC_USABLE_SIZE
# ifdef _WIN32
# define HAVE_MALLOC_USABLE_SIZE
# define malloc_usable_size(a) _msize(a)
# elif defined HAVE_MALLOC_SIZE
# define HAVE_MALLOC_USABLE_SIZE
# define malloc_usable_size(a) malloc_size(a)
# endif
#endif
#ifdef HAVE_MALLOC_USABLE_SIZE
# ifdef HAVE_MALLOC_H
# include <malloc.h>
# elif defined(HAVE_MALLOC_NP_H)
# include <malloc_np.h>
# elif defined(HAVE_MALLOC_MALLOC_H)
# include <malloc/malloc.h>
# endif
#endif
#if /* is ASAN enabled? */ \
__has_feature(address_sanitizer) /* Clang */ || \
defined(__SANITIZE_ADDRESS__) /* GCC 4.8.x */
#define ATTRIBUTE_NO_ADDRESS_SAFETY_ANALYSIS \
__attribute__((no_address_safety_analysis)) \
__attribute__((noinline))
#else
#define ATTRIBUTE_NO_ADDRESS_SAFETY_ANALYSIS
#endif
#ifdef HAVE_SYS_TIME_H
#include <sys/time.h>
#endif
#ifdef HAVE_SYS_RESOURCE_H
#include <sys/resource.h>
#endif
#if defined(__native_client__) && defined(NACL_NEWLIB)
# include "nacl/resource.h"
# undef HAVE_POSIX_MEMALIGN
# undef HAVE_MEMALIGN
#endif
#if defined _WIN32 || defined __CYGWIN__
#include <windows.h>
#elif defined(HAVE_POSIX_MEMALIGN)
#elif defined(HAVE_MEMALIGN)
#include <malloc.h>
#endif
#define rb_setjmp(env) RUBY_SETJMP(env)
#define rb_jmp_buf rb_jmpbuf_t
#if defined(HAVE_RB_GC_GUARDED_PTR_VAL) && HAVE_RB_GC_GUARDED_PTR_VAL
/* trick the compiler into thinking a external signal handler uses this */
volatile VALUE rb_gc_guarded_val;
volatile VALUE *
rb_gc_guarded_ptr_val(volatile VALUE *ptr, VALUE val)
{
rb_gc_guarded_val = val;
return ptr;
}
#endif
#ifndef GC_HEAP_FREE_SLOTS
#define GC_HEAP_FREE_SLOTS 4096
#endif
#ifndef GC_HEAP_INIT_SLOTS
#define GC_HEAP_INIT_SLOTS 10000
#endif
#ifndef GC_HEAP_GROWTH_FACTOR
#define GC_HEAP_GROWTH_FACTOR 1.8
#endif
#ifndef GC_HEAP_GROWTH_MAX_SLOTS
#define GC_HEAP_GROWTH_MAX_SLOTS 0 /* 0 is disable */
#endif
#ifndef GC_HEAP_OLDOBJECT_LIMIT_FACTOR
#define GC_HEAP_OLDOBJECT_LIMIT_FACTOR 2.0
#endif
#ifndef GC_HEAP_FREE_SLOTS_MIN_RATIO
#define GC_HEAP_FREE_SLOTS_MIN_RATIO 0.3
#endif
#ifndef GC_HEAP_FREE_SLOTS_MAX_RATIO
#define GC_HEAP_FREE_SLOTS_MAX_RATIO 0.8
#endif
#ifndef GC_MALLOC_LIMIT_MIN
#define GC_MALLOC_LIMIT_MIN (16 * 1024 * 1024 /* 16MB */)
#endif
#ifndef GC_MALLOC_LIMIT_MAX
#define GC_MALLOC_LIMIT_MAX (32 * 1024 * 1024 /* 32MB */)
#endif
#ifndef GC_MALLOC_LIMIT_GROWTH_FACTOR
#define GC_MALLOC_LIMIT_GROWTH_FACTOR 1.4
#endif
#ifndef GC_OLDMALLOC_LIMIT_MIN
#define GC_OLDMALLOC_LIMIT_MIN (16 * 1024 * 1024 /* 16MB */)
#endif
#ifndef GC_OLDMALLOC_LIMIT_GROWTH_FACTOR
#define GC_OLDMALLOC_LIMIT_GROWTH_FACTOR 1.2
#endif
#ifndef GC_OLDMALLOC_LIMIT_MAX
#define GC_OLDMALLOC_LIMIT_MAX (128 * 1024 * 1024 /* 128MB */)
#endif
typedef struct {
size_t heap_init_slots;
size_t heap_free_slots;
double growth_factor;
size_t growth_max_slots;
double oldobject_limit_factor;
size_t malloc_limit_min;
size_t malloc_limit_max;
double malloc_limit_growth_factor;
size_t oldmalloc_limit_min;
size_t oldmalloc_limit_max;
double oldmalloc_limit_growth_factor;
#if defined(ENABLE_VM_OBJSPACE) && ENABLE_VM_OBJSPACE
VALUE gc_stress;
#endif
} ruby_gc_params_t;
static ruby_gc_params_t gc_params = {
GC_HEAP_INIT_SLOTS,
GC_HEAP_FREE_SLOTS,
GC_HEAP_GROWTH_FACTOR,
GC_HEAP_GROWTH_MAX_SLOTS,
GC_HEAP_OLDOBJECT_LIMIT_FACTOR,
GC_MALLOC_LIMIT_MIN,
GC_MALLOC_LIMIT_MAX,
GC_MALLOC_LIMIT_GROWTH_FACTOR,
GC_OLDMALLOC_LIMIT_MIN,
GC_OLDMALLOC_LIMIT_MAX,
GC_OLDMALLOC_LIMIT_GROWTH_FACTOR,
#if defined(ENABLE_VM_OBJSPACE) && ENABLE_VM_OBJSPACE
FALSE,
#endif
};
/* GC_DEBUG:
* enable to embed GC debugging information.
*/
#ifndef GC_DEBUG
#define GC_DEBUG 0
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
/* RGENGC_DEBUG:
* 1: basic information
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
* 2: remember set operation
* 3: mark
* 4:
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
* 5: sweep
*/
#ifndef RGENGC_DEBUG
#define RGENGC_DEBUG 0
#endif
/* RGENGC_CHECK_MODE
* 0: disable all assertions
* 1: enable assertions (to debug RGenGC)
* 2: enable generational bits check (for debugging)
* 3: enable livness check
* 4: show all references
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
*/
#ifndef RGENGC_CHECK_MODE
#define RGENGC_CHECK_MODE 0
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif
/* RGENGC_PROFILE
* 0: disable RGenGC profiling
* 1: enable profiling for basic information
* 2: enable profiling for each types
*/
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#ifndef RGENGC_PROFILE
#define RGENGC_PROFILE 0
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif
/* RGENGC_AGE2_PROMOTION
* Enable/disable age2 promotion strategy
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
* 0: Infant gen -> Old gen
* 1: Infant gen -> Young -> Old gen
*/
#ifndef RGENGC_AGE2_PROMOTION
#define RGENGC_AGE2_PROMOTION 1
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
#endif
/* RGENGC_ESTIMATE_OLDMALLOC
* Enable/disable to estimate increase size of malloc'ed size by old objects.
* If estimation exceeds threshold, then will invoke full GC.
* 0: disable estimation.
* 1: enable estimation.
*/
#ifndef RGENGC_ESTIMATE_OLDMALLOC
#define RGENGC_ESTIMATE_OLDMALLOC 1
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#else /* USE_RGENGC */
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#define RGENGC_DEBUG 0
#define RGENGC_CHECK_MODE 0
#define RGENGC_PROFILE 0
#define RGENGC_AGE2_PROMOTION 0
#define RGENGC_ESTIMATE_OLDMALLOC 0
#endif /* USE_RGENGC */
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#ifndef GC_PROFILE_MORE_DETAIL
#define GC_PROFILE_MORE_DETAIL 0
#endif
#ifndef GC_PROFILE_DETAIL_MEMORY
#define GC_PROFILE_DETAIL_MEMORY 0
#endif
#ifndef GC_ENABLE_LAZY_SWEEP
#define GC_ENABLE_LAZY_SWEEP 1
#endif
#ifndef CALC_EXACT_MALLOC_SIZE
#define CALC_EXACT_MALLOC_SIZE 0
#endif
#if defined(HAVE_MALLOC_USABLE_SIZE) || CALC_EXACT_MALLOC_SIZE > 0
#ifndef MALLOC_ALLOCATED_SIZE
#define MALLOC_ALLOCATED_SIZE 0
#endif
#else
#define MALLOC_ALLOCATED_SIZE 0
#endif
#ifndef MALLOC_ALLOCATED_SIZE_CHECK
#define MALLOC_ALLOCATED_SIZE_CHECK 0
#endif
typedef enum {
GPR_FLAG_NONE = 0x000,
/* major reason */
GPR_FLAG_MAJOR_BY_NOFREE = 0x001,
GPR_FLAG_MAJOR_BY_OLDGEN = 0x002,
GPR_FLAG_MAJOR_BY_SHADY = 0x004,
GPR_FLAG_MAJOR_BY_RESCAN = 0x008,
GPR_FLAG_MAJOR_BY_STRESS = 0x010,
#if RGENGC_ESTIMATE_OLDMALLOC
GPR_FLAG_MAJOR_BY_OLDMALLOC = 0x020,
#endif
GPR_FLAG_MAJOR_MASK = 0x0ff,
/* gc reason */
GPR_FLAG_NEWOBJ = 0x100,
GPR_FLAG_MALLOC = 0x200,
GPR_FLAG_METHOD = 0x400,
GPR_FLAG_CAPI = 0x800,
GPR_FLAG_STRESS = 0x1000,
/* others */
GPR_FLAG_IMMEDIATE_SWEEP = 0x2000,
GPR_FLAG_HAVE_FINALIZE = 0x4000
} gc_profile_record_flag;
typedef struct gc_profile_record {
int flags;
double gc_time;
double gc_invoke_time;
size_t heap_total_objects;
size_t heap_use_size;
size_t heap_total_size;
#if GC_PROFILE_MORE_DETAIL
double gc_mark_time;
double gc_sweep_time;
2009-05-13 18:08:26 +04:00
size_t heap_use_pages;
size_t heap_live_objects;
size_t heap_free_objects;
2009-05-13 18:08:26 +04:00
size_t allocate_increase;
size_t allocate_limit;
double prepare_time;
size_t removing_objects;
size_t empty_objects;
#if GC_PROFILE_DETAIL_MEMORY
long maxrss;
long minflt;
long majflt;
#endif
#endif
#if MALLOC_ALLOCATED_SIZE
size_t allocated_size;
#endif
#if RGENGC_PROFILE > 0
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
size_t old_objects;
size_t remembered_normal_objects;
size_t remembered_shady_objects;
#endif
} gc_profile_record;
#if defined(_MSC_VER) || defined(__BORLANDC__) || defined(__CYGWIN__)
#pragma pack(push, 1) /* magic for reducing sizeof(RVALUE): 24 -> 20 */
#endif
typedef struct RVALUE {
union {
struct {
VALUE flags; /* always 0 for freed obj */
struct RVALUE *next;
} free;
struct RBasic basic;
struct RObject object;
struct RClass klass;
struct RFloat flonum;
struct RString string;
struct RArray array;
struct RRegexp regexp;
struct RHash hash;
struct RData data;
struct RTypedData typeddata;
struct RStruct rstruct;
struct RBignum bignum;
struct RFile file;
struct RNode node;
struct RMatch match;
struct RRational rational;
struct RComplex complex;
* parse.y: support Symbol GC. [ruby-trunk Feature #9634] See this ticket about Symbol GC. * include/ruby/ruby.h: Declare few functions. * rb_sym2id: almost same as old SYM2ID but support dynamic symbols. * rb_id2sym: almost same as old ID2SYM but support dynamic symbols. * rb_sym2str: almost same as `rb_id2str(SYM2ID(sym))` but not pin down a dynamic symbol. Declare a new struct. * struct RSymbol: represents a dynamic symbol as object in Ruby's heaps. Add few macros. * STATIC_SYM_P: check a static symbol. * DYNAMIC_SYM_P: check a dynamic symbol. * RSYMBOL: cast to RSymbol * gc.c: declare RSymbol. support T_SYMBOL. * internal.h: Declare few functions. * rb_gc_free_dsymbol: free up a dynamic symbol. GC call this function at a sweep phase. * rb_str_dynamic_intern: convert a string to a dynamic symbol. * rb_check_id_without_pindown: not pinning function. * rb_sym2id_without_pindown: ditto. * rb_check_id_cstr_without_pindown: ditto. * string.c (Init_String): String#intern and String#to_sym use rb_str_dynamic_intern. * template/id.h.tmpl: use LSB of ID as a flag for determining a static symbol, so we shift left other ruby_id_types. * string.c: use rb_sym2str instead `rb_id2str(SYM2ID(sym))` to avoid pinning. * load.c: use xx_without_pindown function at creating temporary ID to avoid pinning. * object.c: ditto. * sprintf.c: ditto. * struct.c: ditto. * thread.c: ditto. * variable.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@45426 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-03-26 08:57:47 +04:00
struct RSymbol symbol;
struct {
struct RBasic basic;
VALUE v1;
VALUE v2;
VALUE v3;
} values;
} as;
#if GC_DEBUG
const char *file;
int line;
#endif
} RVALUE;
#if defined(_MSC_VER) || defined(__BORLANDC__) || defined(__CYGWIN__)
#pragma pack(pop)
#endif
typedef uintptr_t bits_t;
enum {
BITS_SIZE = sizeof(bits_t),
BITS_BITLENGTH = ( BITS_SIZE * CHAR_BIT )
};
struct heap_page_header {
struct heap_page *page;
};
struct heap_page_body {
struct heap_page_header header;
/* char gap[]; */
/* RVALUE values[]; */
};
struct gc_list {
VALUE *varptr;
struct gc_list *next;
};
#define STACK_CHUNK_SIZE 500
typedef struct stack_chunk {
VALUE data[STACK_CHUNK_SIZE];
struct stack_chunk *next;
} stack_chunk_t;
typedef struct mark_stack {
stack_chunk_t *chunk;
stack_chunk_t *cache;
size_t index;
size_t limit;
size_t cache_size;
size_t unused_cache_size;
} mark_stack_t;
typedef struct rb_heap_struct {
struct heap_page *pages;
struct heap_page *free_pages;
struct heap_page *using_page;
struct heap_page *sweep_pages;
RVALUE *freelist;
size_t page_length; /* total page count in a heap */
size_t total_slots; /* total slot count (page_length * HEAP_OBJ_LIMIT) */
} rb_heap_t;
typedef struct rb_objspace {
struct {
size_t limit;
size_t increase;
#if MALLOC_ALLOCATED_SIZE
size_t allocated_size;
size_t allocations;
#endif
} malloc_params;
rb_heap_t eden_heap;
rb_heap_t tomb_heap; /* heap for zombies and ghosts */
struct {
struct heap_page **sorted;
size_t used;
size_t length;
RVALUE *range[2];
size_t limit;
size_t increment;
size_t swept_slots;
size_t min_free_slots;
size_t max_free_slots;
/* final */
size_t final_slots;
RVALUE *deferred_final;
} heap_pages;
struct {
int dont_gc;
int dont_lazy_sweep;
int during_gc;
rb_atomic_t finalizing;
} flags;
st_table *finalizer_table;
mark_stack_t mark_stack;
struct {
int run;
gc_profile_record *records;
gc_profile_record *current_record;
size_t next_index;
size_t size;
#if GC_PROFILE_MORE_DETAIL
double prepare_time;
#endif
double invoke_time;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
size_t minor_gc_count;
size_t major_gc_count;
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
#if RGENGC_PROFILE > 0
size_t generated_normal_object_count;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
size_t generated_shady_object_count;
size_t shade_operation_count;
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
size_t promote_infant_count;
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
size_t promote_young_count;
#endif
size_t remembered_normal_object_count;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
size_t remembered_shady_object_count;
#if RGENGC_PROFILE >= 2
size_t generated_normal_object_count_types[RUBY_T_MASK];
size_t generated_shady_object_count_types[RUBY_T_MASK];
size_t shade_operation_count_types[RUBY_T_MASK];
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
size_t promote_infant_types[RUBY_T_MASK];
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
size_t promote_young_types[RUBY_T_MASK];
#endif
size_t remembered_normal_object_count_types[RUBY_T_MASK];
size_t remembered_shady_object_count_types[RUBY_T_MASK];
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif /* RGENGC_PROFILE */
#endif /* USE_RGENGC */
/* temporary profiling space */
double gc_sweep_start_time;
size_t total_allocated_object_num_at_gc_start;
size_t heap_used_at_gc_start;
/* basic statistics */
size_t count;
size_t total_allocated_object_num;
size_t total_freed_object_num;
int latest_gc_info;
} profile;
struct gc_list *global_list;
rb_event_flag_t hook_events; /* this place may be affinity with memory cache */
VALUE gc_stress;
struct mark_func_data_struct {
void *data;
void (*mark_func)(VALUE v, void *data);
} *mark_func_data;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
struct {
int during_minor_gc;
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
int parent_object_is_old;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
int need_major_gc;
size_t remembered_shady_object_count;
size_t remembered_shady_object_limit;
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
size_t old_object_count;
size_t old_object_limit;
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
size_t young_object_count;
#endif
#if RGENGC_ESTIMATE_OLDMALLOC
size_t oldmalloc_increase;
size_t oldmalloc_increase_limit;
#endif
#if RGENGC_CHECK_MODE >= 2
struct st_table *allrefs_table;
size_t error_count;
#endif
} rgengc;
#endif /* USE_RGENGC */
} rb_objspace_t;
#ifndef HEAP_ALIGN_LOG
/* default tiny heap size: 16KB */
#define HEAP_ALIGN_LOG 14
#endif
#define CEILDIV(i, mod) (((i) + (mod) - 1)/(mod))
enum {
HEAP_ALIGN = (1UL << HEAP_ALIGN_LOG),
HEAP_ALIGN_MASK = (~(~0UL << HEAP_ALIGN_LOG)),
REQUIRED_SIZE_BY_MALLOC = (sizeof(size_t) * 5),
HEAP_SIZE = (HEAP_ALIGN - REQUIRED_SIZE_BY_MALLOC),
HEAP_OBJ_LIMIT = (unsigned int)((HEAP_SIZE - sizeof(struct heap_page_header))/sizeof(struct RVALUE)),
HEAP_BITMAP_LIMIT = CEILDIV(CEILDIV(HEAP_SIZE, sizeof(struct RVALUE)), BITS_BITLENGTH),
HEAP_BITMAP_SIZE = ( BITS_SIZE * HEAP_BITMAP_LIMIT),
HEAP_BITMAP_PLANES = USE_RGENGC ? 3 : 1 /* RGENGC: mark bits, rememberset bits and oldgen bits */
};
struct heap_page {
struct heap_page_body *body;
RVALUE *freelist;
RVALUE *start;
size_t final_slots;
size_t limit;
struct heap_page *next;
struct heap_page *prev;
struct heap_page *free_next;
rb_heap_t *heap;
int before_sweep;
bits_t mark_bits[HEAP_BITMAP_LIMIT];
#if USE_RGENGC
bits_t rememberset_bits[HEAP_BITMAP_LIMIT];
bits_t oldgen_bits[HEAP_BITMAP_LIMIT];
#endif
};
#define GET_PAGE_BODY(x) ((struct heap_page_body *)((bits_t)(x) & ~(HEAP_ALIGN_MASK)))
#define GET_PAGE_HEADER(x) (&GET_PAGE_BODY(x)->header)
#define GET_HEAP_PAGE(x) (GET_PAGE_HEADER(x)->page)
#define GET_HEAP_MARK_BITS(x) (&GET_HEAP_PAGE(x)->mark_bits[0])
#define GET_HEAP_REMEMBERSET_BITS(x) (&GET_HEAP_PAGE(x)->rememberset_bits[0])
#define GET_HEAP_OLDGEN_BITS(x) (&GET_HEAP_PAGE(x)->oldgen_bits[0])
#define NUM_IN_PAGE(p) (((bits_t)(p) & HEAP_ALIGN_MASK)/sizeof(RVALUE))
#define BITMAP_INDEX(p) (NUM_IN_PAGE(p) / BITS_BITLENGTH )
#define BITMAP_OFFSET(p) (NUM_IN_PAGE(p) & (BITS_BITLENGTH-1))
#define BITMAP_BIT(p) ((bits_t)1 << BITMAP_OFFSET(p))
/* Bitmap Operations */
#define MARKED_IN_BITMAP(bits, p) ((bits)[BITMAP_INDEX(p)] & BITMAP_BIT(p))
#define MARK_IN_BITMAP(bits, p) ((bits)[BITMAP_INDEX(p)] = (bits)[BITMAP_INDEX(p)] | BITMAP_BIT(p))
#define CLEAR_IN_BITMAP(bits, p) ((bits)[BITMAP_INDEX(p)] = (bits)[BITMAP_INDEX(p)] & ~BITMAP_BIT(p))
/* Aliases */
#if defined(ENABLE_VM_OBJSPACE) && ENABLE_VM_OBJSPACE
#define rb_objspace (*GET_VM()->objspace)
#define ruby_initial_gc_stress gc_params.gc_stress
VALUE *ruby_initial_gc_stress_ptr = &ruby_initial_gc_stress;
#else
static rb_objspace_t rb_objspace = {{GC_MALLOC_LIMIT_MIN}};
VALUE *ruby_initial_gc_stress_ptr = &rb_objspace.gc_stress;
#endif
#define malloc_limit objspace->malloc_params.limit
#define malloc_increase objspace->malloc_params.increase
#define malloc_allocated_size objspace->malloc_params.allocated_size
#define heap_pages_sorted objspace->heap_pages.sorted
#define heap_pages_used objspace->heap_pages.used
#define heap_pages_length objspace->heap_pages.length
#define heap_pages_lomem objspace->heap_pages.range[0]
#define heap_pages_himem objspace->heap_pages.range[1]
#define heap_pages_swept_slots objspace->heap_pages.swept_slots
#define heap_pages_increment objspace->heap_pages.increment
#define heap_pages_min_free_slots objspace->heap_pages.min_free_slots
#define heap_pages_max_free_slots objspace->heap_pages.max_free_slots
#define heap_pages_final_slots objspace->heap_pages.final_slots
#define heap_pages_deferred_final objspace->heap_pages.deferred_final
#define heap_eden (&objspace->eden_heap)
#define heap_tomb (&objspace->tomb_heap)
#define dont_gc objspace->flags.dont_gc
#define during_gc objspace->flags.during_gc
#define finalizing objspace->flags.finalizing
#define finalizer_table objspace->finalizer_table
#define global_List objspace->global_list
#define ruby_gc_stress objspace->gc_stress
#define monitor_level objspace->rgengc.monitor_level
#define monitored_object_table objspace->rgengc.monitored_object_table
#define is_lazy_sweeping(heap) ((heap)->sweep_pages != 0)
#if SIZEOF_LONG == SIZEOF_VOIDP
# define nonspecial_obj_id(obj) (VALUE)((SIGNED_VALUE)(obj)|FIXNUM_FLAG)
# define obj_id_to_ref(objid) ((objid) ^ FIXNUM_FLAG) /* unset FIXNUM_FLAG */
#elif SIZEOF_LONG_LONG == SIZEOF_VOIDP
# define nonspecial_obj_id(obj) LL2NUM((SIGNED_VALUE)(obj) / 2)
# define obj_id_to_ref(objid) (FIXNUM_P(objid) ? \
((objid) ^ FIXNUM_FLAG) : (NUM2PTR(objid) << 1))
#else
# error not supported
#endif
#define RANY(o) ((RVALUE*)(o))
#define nomem_error GET_VM()->special_exceptions[ruby_error_nomemory]
int ruby_gc_debug_indent = 0;
VALUE rb_mGC;
int ruby_disable_gc_stress = 0;
void rb_gcdebug_print_obj_condition(VALUE obj);
static void rb_objspace_call_finalizer(rb_objspace_t *objspace);
static VALUE define_final0(VALUE obj, VALUE block);
static void negative_size_allocation_error(const char *);
static void *aligned_malloc(size_t, size_t);
static void aligned_free(void *);
static void init_mark_stack(mark_stack_t *stack);
static VALUE lazy_sweep_enable(void);
static int ready_to_gc(rb_objspace_t *objspace);
static int heap_ready_to_gc(rb_objspace_t *objspace, rb_heap_t *heap);
static int garbage_collect(rb_objspace_t *, int full_mark, int immediate_sweep, int reason);
static int garbage_collect_body(rb_objspace_t *, int full_mark, int immediate_sweep, int reason);
static int gc_heap_lazy_sweep(rb_objspace_t *objspace, rb_heap_t *heap);
static void gc_rest_sweep(rb_objspace_t *objspace);
static void gc_heap_rest_sweep(rb_objspace_t *objspace, rb_heap_t *heap);
static void gc_mark_stacked_objects(rb_objspace_t *);
static void gc_mark(rb_objspace_t *objspace, VALUE ptr);
static void gc_mark_maybe(rb_objspace_t *objspace, VALUE ptr);
static void gc_mark_children(rb_objspace_t *objspace, VALUE ptr);
static size_t obj_memsize_of(VALUE obj, int use_tdata);
static double getrusage_time(void);
static inline void gc_prof_setup_new_record(rb_objspace_t *objspace, int reason);
static inline void gc_prof_timer_start(rb_objspace_t *);
static inline void gc_prof_timer_stop(rb_objspace_t *);
static inline void gc_prof_mark_timer_start(rb_objspace_t *);
static inline void gc_prof_mark_timer_stop(rb_objspace_t *);
static inline void gc_prof_sweep_timer_start(rb_objspace_t *);
static inline void gc_prof_sweep_timer_stop(rb_objspace_t *);
static inline void gc_prof_set_malloc_info(rb_objspace_t *);
static inline void gc_prof_set_heap_info(rb_objspace_t *);
#define gc_prof_record(objspace) (objspace)->profile.current_record
#define gc_prof_enabled(objspace) ((objspace)->profile.run && (objspace)->profile.current_record)
#ifdef HAVE_VA_ARGS_MACRO
# define rgengc_report(level, objspace, fmt, ...) \
if ((level) > RGENGC_DEBUG) {} else rgengc_report_body(level, objspace, fmt, ##__VA_ARGS__)
#else
# define rgengc_report if (!(RGENGC_DEBUG)) {} else rgengc_report_body
#endif
PRINTF_ARGS(static void rgengc_report_body(int level, rb_objspace_t *objspace, const char *fmt, ...), 3, 4);
static const char * type_name(int type, VALUE obj);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
static const char *obj_type_name(VALUE obj);
#if USE_RGENGC
static int rgengc_remembered(rb_objspace_t *objspace, VALUE obj);
static int rgengc_remember(rb_objspace_t *objspace, VALUE obj);
static void rgengc_mark_and_rememberset_clear(rb_objspace_t *objspace, rb_heap_t *heap);
static void rgengc_rememberset_mark(rb_objspace_t *objspace, rb_heap_t *heap);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#define FL_TEST2(x,f) ((RGENGC_CHECK_MODE && SPECIAL_CONST_P(x)) ? (rb_bug("FL_TEST2: SPECIAL_CONST"), 0) : FL_TEST_RAW((x),(f)) != 0)
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#define FL_SET2(x,f) do {if (RGENGC_CHECK_MODE && SPECIAL_CONST_P(x)) rb_bug("FL_SET2: SPECIAL_CONST"); RBASIC(x)->flags |= (f);} while (0)
#define FL_UNSET2(x,f) do {if (RGENGC_CHECK_MODE && SPECIAL_CONST_P(x)) rb_bug("FL_UNSET2: SPECIAL_CONST"); RBASIC(x)->flags &= ~(f);} while (0)
#define RVALUE_WB_PROTECTED_RAW(obj) FL_TEST2((obj), FL_WB_PROTECTED)
#define RVALUE_WB_PROTECTED(obj) RVALUE_WB_PROTECTED_RAW(check_gen_consistency((VALUE)obj))
#define RVALUE_OLDGEN_BITMAP(obj) MARKED_IN_BITMAP(GET_HEAP_OLDGEN_BITS(obj), (obj))
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
static inline int is_pointer_to_heap(rb_objspace_t *objspace, void *ptr);
static inline int gc_marked(rb_objspace_t *objspace, VALUE ptr);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
static inline VALUE
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency(VALUE obj)
{
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (RGENGC_CHECK_MODE > 0) {
int old_flag = RVALUE_OLDGEN_BITMAP(obj) != 0;
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
int promoted_flag = FL_TEST2(obj, FL_PROMOTED);
rb_objspace_t *objspace = &rb_objspace;
obj_memsize_of((VALUE)obj, FALSE);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (!is_pointer_to_heap(objspace, (void *)obj)) {
rb_bug("check_gen_consistency: %p (%s) is not Ruby object.", (void *)obj, obj_type_name(obj));
}
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (promoted_flag) {
if (!RVALUE_WB_PROTECTED_RAW(obj)) {
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
const char *type = old_flag ? "old" : "young";
rb_bug("check_gen_consistency: %p (%s) is not WB protected, but %s object.", (void *)obj, obj_type_name(obj), type);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
}
#if !RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (!old_flag) {
rb_bug("check_gen_consistency: %p (%s) is not infant, but is not old (on 2gen).", (void *)obj, obj_type_name(obj));
}
#endif
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (old_flag && objspace->rgengc.during_minor_gc && !gc_marked(objspace, obj)) {
rb_bug("check_gen_consistency: %p (%s) is old, but is not marked while minor marking.", (void *)obj, obj_type_name(obj));
}
}
else {
if (old_flag) {
rb_bug("check_gen_consistency: %p (%s) is not infant, but is old.", (void *)obj, obj_type_name(obj));
}
}
}
return obj;
}
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
static inline VALUE
RVALUE_INFANT_P(VALUE obj)
{
check_gen_consistency(obj);
return !FL_TEST2(obj, FL_PROMOTED);
}
static inline VALUE
RVALUE_OLD_BITMAP_P(VALUE obj)
{
check_gen_consistency(obj);
return (RVALUE_OLDGEN_BITMAP(obj) != 0);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
}
static inline VALUE
RVALUE_OLD_P(VALUE obj)
{
check_gen_consistency(obj);
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
return FL_TEST2(obj, FL_PROMOTED) && RVALUE_OLD_BITMAP_P(obj);
#else
return FL_TEST2(obj, FL_PROMOTED);
#endif
}
static inline VALUE
RVALUE_PROMOTED_P(VALUE obj)
{
check_gen_consistency(obj);
return FL_TEST2(obj, FL_PROMOTED);
}
static inline void
RVALUE_PROMOTE_INFANT(rb_objspace_t *objspace, VALUE obj)
{
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency(obj);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (RGENGC_CHECK_MODE && !RVALUE_INFANT_P(obj)) rb_bug("RVALUE_PROMOTE_INFANT: %p (%s) is not infant object.", (void *)obj, obj_type_name(obj));
FL_SET2(obj, FL_PROMOTED);
#if RGENGC_AGE2_PROMOTION
/* infant -> young */
objspace->rgengc.young_object_count++;
#else
/* infant -> old */
objspace->rgengc.old_object_count++;
#endif
#if !RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
MARK_IN_BITMAP(GET_HEAP_OLDGEN_BITS(obj), obj);
#endif
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency(obj);
#if RGENGC_PROFILE >= 1
{
rb_objspace_t *objspace = &rb_objspace;
objspace->profile.promote_infant_count++;
#if RGENGC_PROFILE >= 2
objspace->profile.promote_infant_types[BUILTIN_TYPE(obj)]++;
#endif
}
#endif
}
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
/*
* Age1 promotion: Infant (0) -> Old (1 and later).
* Age2 promotion: Infant (0) -> Young (1) -> Old (2 and later).
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
*/
static inline VALUE
RVALUE_YOUNG_P(VALUE obj)
{
check_gen_consistency(obj);
return FL_TEST2(obj, FL_PROMOTED) && (RVALUE_OLDGEN_BITMAP(obj) == 0);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
}
static inline void
RVALUE_PROMOTE_YOUNG(rb_objspace_t *objspace, VALUE obj)
{
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency(obj);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (RGENGC_CHECK_MODE && !RVALUE_YOUNG_P(obj)) rb_bug("RVALUE_PROMOTE_YOUNG: %p (%s) is not young object.", (void *)obj, obj_type_name(obj));
MARK_IN_BITMAP(GET_HEAP_OLDGEN_BITS(obj), obj);
objspace->rgengc.young_object_count--;
objspace->rgengc.old_object_count++;
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency(obj);
#if RGENGC_PROFILE >= 1
{
rb_objspace_t *objspace = &rb_objspace;
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
objspace->profile.promote_young_count++;
#if RGENGC_PROFILE >= 2
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
objspace->profile.promote_young_types[BUILTIN_TYPE(obj)]++;
#endif
}
#endif
}
static inline void
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
RVALUE_DEMOTE_FROM_YOUNG(VALUE obj)
{
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (RGENGC_CHECK_MODE && !RVALUE_YOUNG_P(obj))
rb_bug("RVALUE_DEMOTE_FROM_YOUNG: %p (%s) is not young object.", (void *)obj, obj_type_name(obj));
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency(obj);
FL_UNSET2(obj, FL_PROMOTED);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency(obj);
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
static inline void
RVALUE_DEMOTE_FROM_OLD(VALUE obj)
{
if (RGENGC_CHECK_MODE && !RVALUE_OLD_P(obj))
rb_bug("RVALUE_DEMOTE_FROM_OLD: %p (%s) is not old object.", (void *)obj, obj_type_name(obj));
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency(obj);
FL_UNSET2(obj, FL_PROMOTED);
CLEAR_IN_BITMAP(GET_HEAP_OLDGEN_BITS(obj), obj);
check_gen_consistency(obj);
}
#endif /* USE_RGENGC */
/*
--------------------------- ObjectSpace -----------------------------
*/
#if defined(ENABLE_VM_OBJSPACE) && ENABLE_VM_OBJSPACE
rb_objspace_t *
rb_objspace_alloc(void)
{
rb_objspace_t *objspace = malloc(sizeof(rb_objspace_t));
memset(objspace, 0, sizeof(*objspace));
ruby_gc_stress = ruby_initial_gc_stress;
malloc_limit = gc_params.malloc_limit_min;
return objspace;
}
#endif
#if defined(ENABLE_VM_OBJSPACE) && ENABLE_VM_OBJSPACE
static void free_stack_chunks(mark_stack_t *);
static void heap_page_free(rb_objspace_t *objspace, struct heap_page *page);
void
rb_objspace_free(rb_objspace_t *objspace)
{
gc_rest_sweep(objspace);
if (objspace->profile.records) {
free(objspace->profile.records);
objspace->profile.records = 0;
}
if (global_List) {
struct gc_list *list, *next;
for (list = global_List; list; list = next) {
next = list->next;
xfree(list);
}
}
if (heap_pages_sorted) {
size_t i;
for (i = 0; i < heap_pages_used; ++i) {
heap_page_free(objspace, heap_pages_sorted[i]);
}
free(heap_pages_sorted);
heap_pages_used = 0;
heap_pages_length = 0;
heap_pages_lomem = 0;
heap_pages_himem = 0;
objspace->eden_heap.page_length = 0;
objspace->eden_heap.total_slots = 0;
objspace->eden_heap.pages = NULL;
}
free_stack_chunks(&objspace->mark_stack);
free(objspace);
}
#endif
static void
heap_pages_expand_sorted(rb_objspace_t *objspace)
{
size_t next_length = heap_pages_increment;
next_length += heap_eden->page_length;
next_length += heap_tomb->page_length;
if (next_length > heap_pages_length) {
struct heap_page **sorted;
size_t size = next_length * sizeof(struct heap_page *);
rgengc_report(3, objspace, "heap_pages_expand_sorted: next_length: %d, size: %d\n", (int)next_length, (int)size);
if (heap_pages_length > 0) {
sorted = (struct heap_page **)realloc(heap_pages_sorted, size);
if (sorted) heap_pages_sorted = sorted;
}
else {
sorted = heap_pages_sorted = (struct heap_page **)malloc(size);
}
if (sorted == 0) {
during_gc = 0;
rb_memerror();
}
heap_pages_length = next_length;
}
}
static inline void
heap_page_add_freeobj(rb_objspace_t *objspace, struct heap_page *page, VALUE obj)
{
RVALUE *p = (RVALUE *)obj;
p->as.free.flags = 0;
p->as.free.next = page->freelist;
page->freelist = p;
rgengc_report(3, objspace, "heap_page_add_freeobj: %p (%s) is added to freelist\n", p, obj_type_name(obj));
}
static inline void
heap_add_freepage(rb_objspace_t *objspace, rb_heap_t *heap, struct heap_page *page)
{
if (page->freelist) {
page->free_next = heap->free_pages;
heap->free_pages = page;
}
}
static void
heap_unlink_page(rb_objspace_t *objspace, rb_heap_t *heap, struct heap_page *page)
{
if (page->prev) page->prev->next = page->next;
if (page->next) page->next->prev = page->prev;
if (heap->pages == page) heap->pages = page->next;
page->prev = NULL;
page->next = NULL;
page->heap = NULL;
heap->page_length--;
heap->total_slots -= page->limit;
}
static void
heap_page_free(rb_objspace_t *objspace, struct heap_page *page)
{
heap_pages_used--;
aligned_free(page->body);
free(page);
}
static void
heap_pages_free_unused_pages(rb_objspace_t *objspace)
{
size_t i, j;
if (heap_tomb->pages) {
for (i = j = 1; j < heap_pages_used; i++) {
struct heap_page *page = heap_pages_sorted[i];
if (page->heap == heap_tomb && page->final_slots == 0) {
if (heap_pages_swept_slots - page->limit > heap_pages_max_free_slots) {
if (0) fprintf(stderr, "heap_pages_free_unused_pages: %d free page %p, heap_pages_swept_slots: %d, heap_pages_max_free_slots: %d\n",
(int)i, page, (int)heap_pages_swept_slots, (int)heap_pages_max_free_slots);
heap_pages_swept_slots -= page->limit;
heap_unlink_page(objspace, heap_tomb, page);
heap_page_free(objspace, page);
continue;
}
else if (i == j) {
return; /* no need to check rest pages */
}
}
if (i != j) {
heap_pages_sorted[j] = page;
}
j++;
}
assert(j == heap_pages_used);
}
}
static struct heap_page *
heap_page_allocate(rb_objspace_t *objspace)
{
RVALUE *start, *end, *p;
struct heap_page *page;
struct heap_page_body *page_body = 0;
size_t hi, lo, mid;
size_t limit = HEAP_OBJ_LIMIT;
/* assign heap_page body (contains heap_page_header and RVALUEs) */
page_body = (struct heap_page_body *)aligned_malloc(HEAP_ALIGN, HEAP_SIZE);
if (page_body == 0) {
during_gc = 0;
rb_memerror();
}
/* assign heap_page entry */
page = (struct heap_page *)malloc(sizeof(struct heap_page));
if (page == 0) {
aligned_free(page_body);
during_gc = 0;
rb_memerror();
}
MEMZERO((void*)page, struct heap_page, 1);
page->body = page_body;
/* setup heap_pages_sorted */
lo = 0;
hi = heap_pages_used;
while (lo < hi) {
struct heap_page *mid_page;
mid = (lo + hi) / 2;
mid_page = heap_pages_sorted[mid];
if (mid_page->body < page_body) {
lo = mid + 1;
}
else if (mid_page->body > page_body) {
hi = mid;
}
else {
rb_bug("same heap page is allocated: %p at %"PRIuVALUE, (void *)page_body, (VALUE)mid);
}
}
if (hi < heap_pages_used) {
MEMMOVE(&heap_pages_sorted[hi+1], &heap_pages_sorted[hi], struct heap_page_header*, heap_pages_used - hi);
}
heap_pages_sorted[hi] = page;
heap_pages_used++;
assert(heap_pages_used <= heap_pages_length);
/* adjust obj_limit (object number available in this page) */
start = (RVALUE*)((VALUE)page_body + sizeof(struct heap_page_header));
if ((VALUE)start % sizeof(RVALUE) != 0) {
int delta = (int)(sizeof(RVALUE) - ((VALUE)start % sizeof(RVALUE)));
start = (RVALUE*)((VALUE)start + delta);
limit = (HEAP_SIZE - (size_t)((VALUE)start - (VALUE)page_body))/sizeof(RVALUE);
}
end = start + limit;
if (heap_pages_lomem == 0 || heap_pages_lomem > start) heap_pages_lomem = start;
if (heap_pages_himem < end) heap_pages_himem = end;
page->start = start;
page->limit = limit;
page_body->header.page = page;
for (p = start; p != end; p++) {
rgengc_report(3, objspace, "assign_heap_page: %p is added to freelist\n", p);
heap_page_add_freeobj(objspace, page, (VALUE)p);
}
return page;
}
static struct heap_page *
heap_page_resurrect(rb_objspace_t *objspace)
{
struct heap_page *page;
if ((page = heap_tomb->pages) != NULL) {
heap_unlink_page(objspace, heap_tomb, page);
return page;
}
return NULL;
}
static struct heap_page *
heap_page_create(rb_objspace_t *objspace)
{
struct heap_page *page = heap_page_resurrect(objspace);
const char *method = "recycle";
if (page == NULL) {
page = heap_page_allocate(objspace);
method = "allocate";
}
if (0) fprintf(stderr, "heap_page_create: %s - %p, heap_pages_used: %d, heap_pages_used: %d, tomb->page_length: %d\n",
method, page, (int)heap_pages_length, (int)heap_pages_used, (int)heap_tomb->page_length);
return page;
}
static void
heap_add_page(rb_objspace_t *objspace, rb_heap_t *heap, struct heap_page *page)
{
page->heap = heap;
page->next = heap->pages;
if (heap->pages) heap->pages->prev = page;
heap->pages = page;
heap->page_length++;
heap->total_slots += page->limit;
}
static void
heap_assign_page(rb_objspace_t *objspace, rb_heap_t *heap)
{
struct heap_page *page = heap_page_create(objspace);
heap_add_page(objspace, heap, page);
heap_add_freepage(objspace, heap, page);
}
static void
heap_add_pages(rb_objspace_t *objspace, rb_heap_t *heap, size_t add)
{
size_t i;
heap_pages_increment = add;
heap_pages_expand_sorted(objspace);
for (i = 0; i < add; i++) {
heap_assign_page(objspace, heap);
}
heap_pages_increment = 0;
}
static size_t
heap_extend_pages(rb_objspace_t *objspace)
{
size_t used = heap_pages_used - heap_tomb->page_length;
size_t next_used_limit = (size_t)(used * gc_params.growth_factor);
if (gc_params.growth_max_slots > 0) {
size_t max_used_limit = (size_t)(used + gc_params.growth_max_slots/HEAP_OBJ_LIMIT);
if (next_used_limit > max_used_limit) next_used_limit = max_used_limit;
}
return next_used_limit - used;
}
static void
heap_set_increment(rb_objspace_t *objspace, size_t additional_pages)
{
size_t used = heap_eden->page_length;
size_t next_used_limit = used + additional_pages;
if (next_used_limit == heap_pages_used) next_used_limit++;
heap_pages_increment = next_used_limit - used;
heap_pages_expand_sorted(objspace);
}
static int
heap_increment(rb_objspace_t *objspace, rb_heap_t *heap)
{
rgengc_report(5, objspace, "heap_increment: heap_pages_length: %d, heap_pages_inc: %d, heap->page_length: %d\n",
(int)heap_pages_length, (int)heap_pages_increment, (int)heap->page_length);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
if (heap_pages_increment > 0) {
heap_pages_increment--;
heap_assign_page(objspace, heap);
return TRUE;
}
return FALSE;
}
static struct heap_page *
heap_prepare_freepage(rb_objspace_t *objspace, rb_heap_t *heap)
{
if (!GC_ENABLE_LAZY_SWEEP && objspace->flags.dont_lazy_sweep) {
if (heap_increment(objspace, heap) == 0 &&
garbage_collect(objspace, FALSE, TRUE, GPR_FLAG_NEWOBJ) == 0) {
goto err;
}
goto ok;
}
if (!heap_ready_to_gc(objspace, heap)) return heap->free_pages;
during_gc++;
if ((is_lazy_sweeping(heap) && gc_heap_lazy_sweep(objspace, heap)) || heap_increment(objspace, heap)) {
goto ok;
}
#if GC_PROFILE_MORE_DETAIL
objspace->profile.prepare_time = 0;
#endif
if (garbage_collect_body(objspace, 0, 0, GPR_FLAG_NEWOBJ) == 0) {
err:
during_gc = 0;
rb_memerror();
}
ok:
during_gc = 0;
return heap->free_pages;
}
static RVALUE *
heap_get_freeobj_from_next_freepage(rb_objspace_t *objspace, rb_heap_t *heap)
{
struct heap_page *page;
RVALUE *p;
page = heap->free_pages;
while (page == NULL) {
page = heap_prepare_freepage(objspace, heap);
}
heap->free_pages = page->free_next;
heap->using_page = page;
p = page->freelist;
page->freelist = NULL;
return p;
}
static inline VALUE
heap_get_freeobj(rb_objspace_t *objspace, rb_heap_t *heap)
{
RVALUE *p = heap->freelist;
while (1) {
if (p) {
heap->freelist = p->as.free.next;
return (VALUE)p;
}
else {
p = heap_get_freeobj_from_next_freepage(objspace, heap);
}
}
}
void
rb_objspace_set_event_hook(const rb_event_flag_t event)
{
rb_objspace_t *objspace = &rb_objspace;
objspace->hook_events = event & RUBY_INTERNAL_EVENT_OBJSPACE_MASK;
}
static void
gc_event_hook_body(rb_objspace_t *objspace, const rb_event_flag_t event, VALUE data)
{
rb_thread_t *th = GET_THREAD();
EXEC_EVENT_HOOK(th, event, th->cfp->self, 0, 0, data);
}
#define gc_event_hook(objspace, event, data) do { \
if (UNLIKELY((objspace)->hook_events & (event))) { \
gc_event_hook_body((objspace), (event), (data)); \
} \
} while (0)
static VALUE
newobj_of(VALUE klass, VALUE flags, VALUE v1, VALUE v2, VALUE v3)
{
rb_objspace_t *objspace = &rb_objspace;
VALUE obj;
if (UNLIKELY(during_gc)) {
dont_gc = 1;
during_gc = 0;
rb_bug("object allocation during garbage collection phase");
}
if (UNLIKELY(ruby_gc_stress && !ruby_disable_gc_stress)) {
if (!garbage_collect(objspace, FALSE, FALSE, GPR_FLAG_NEWOBJ)) {
during_gc = 0;
rb_memerror();
}
}
obj = heap_get_freeobj(objspace, heap_eden);
/* OBJSETUP */
RBASIC(obj)->flags = flags;
RBASIC_SET_CLASS_RAW(obj, klass);
* safe.c (rb_set_safe_level, safe_setter): raise an ArgumentError when $SAFE is set to 4. $SAFE=4 is now obsolete. [ruby-core:55222] [Feature #8468] * object.c (rb_obj_untrusted, rb_obj_untrust, rb_obj_trust): Kernel#untrusted?, untrust, and trust are now deprecated. Their behavior is same as tainted?, taint, and untaint, respectively. * include/ruby/ruby.h (OBJ_UNTRUSTED, OBJ_UNTRUST): OBJ_UNTRUSTED() and OBJ_UNTRUST() are aliases of OBJ_TAINTED() and OBJ_TAINT(), respectively. * array.c, class.c, debug.c, dir.c, encoding.c, error.c, eval.c, ext/curses/curses.c, ext/dbm/dbm.c, ext/dl/cfunc.c, ext/dl/cptr.c, ext/dl/dl.c, ext/etc/etc.c, ext/fiddle/fiddle.c, ext/fiddle/pointer.c, ext/gdbm/gdbm.c, ext/readline/readline.c, ext/sdbm/init.c, ext/socket/ancdata.c, ext/socket/basicsocket.c, ext/socket/socket.c, ext/socket/udpsocket.c, ext/stringio/stringio.c, ext/syslog/syslog.c, ext/tk/tcltklib.c, ext/win32ole/win32ole.c, file.c, gc.c, hash.c, io.c, iseq.c, load.c, marshal.c, object.c, proc.c, process.c, random.c, re.c, safe.c, string.c, thread.c, transcode.c, variable.c, vm_insnhelper.c, vm_method.c, vm_trace.c: remove code for $SAFE=4. * test/dl/test_dl2.rb, test/erb/test_erb.rb, test/readline/test_readline.rb, test/readline/test_readline_history.rb, test/ruby/test_alias.rb, test/ruby/test_array.rb, test/ruby/test_dir.rb, test/ruby/test_encoding.rb, test/ruby/test_env.rb, test/ruby/test_eval.rb, test/ruby/test_exception.rb, test/ruby/test_file_exhaustive.rb, test/ruby/test_hash.rb, test/ruby/test_io.rb, test/ruby/test_method.rb, test/ruby/test_module.rb, test/ruby/test_object.rb, test/ruby/test_pack.rb, test/ruby/test_rand.rb, test/ruby/test_regexp.rb, test/ruby/test_settracefunc.rb, test/ruby/test_struct.rb, test/ruby/test_thread.rb, test/ruby/test_time.rb: remove tests for $SAFE=4. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@41259 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-06-12 18:20:51 +04:00
if (rb_safe_level() >= 3) FL_SET((obj), FL_TAINT);
RANY(obj)->as.values.v1 = v1;
RANY(obj)->as.values.v2 = v2;
RANY(obj)->as.values.v3 = v3;
#if GC_DEBUG
RANY(obj)->file = rb_sourcefile();
RANY(obj)->line = rb_sourceline();
assert(!SPECIAL_CONST_P(obj)); /* check alignment */
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if RGENGC_PROFILE
if (flags & FL_WB_PROTECTED) {
objspace->profile.generated_normal_object_count++;
#if RGENGC_PROFILE >= 2
objspace->profile.generated_normal_object_count_types[BUILTIN_TYPE(obj)]++;
#endif
}
else {
objspace->profile.generated_shady_object_count++;
#if RGENGC_PROFILE >= 2
objspace->profile.generated_shady_object_count_types[BUILTIN_TYPE(obj)]++;
#endif
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif
rgengc_report(5, objspace, "newobj: %p (%s)\n", (void *)obj, obj_type_name(obj));
#if USE_RGENGC && RGENGC_CHECK_MODE
if (RVALUE_PROMOTED_P(obj)) rb_bug("newobj: %p (%s) is promoted.", (void *)obj, obj_type_name(obj));
if (rgengc_remembered(objspace, (VALUE)obj)) rb_bug("newobj: %p (%s) is remembered.", (void *)obj, obj_type_name(obj));
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif
objspace->profile.total_allocated_object_num++;
gc_event_hook(objspace, RUBY_INTERNAL_EVENT_NEWOBJ, obj);
return obj;
}
VALUE
rb_newobj(void)
{
return newobj_of(0, T_NONE, 0, 0, 0);
}
VALUE
rb_newobj_of(VALUE klass, VALUE flags)
{
return newobj_of(klass, flags, 0, 0, 0);
}
NODE*
rb_node_newnode(enum node_type type, VALUE a0, VALUE a1, VALUE a2)
{
VALUE flags = (RGENGC_WB_PROTECTED_NODE_CREF && type == NODE_CREF ? FL_WB_PROTECTED : 0);
NODE *n = (NODE *)newobj_of(0, T_NODE | flags, a0, a1, a2);
nd_set_type(n, type);
return n;
}
VALUE
rb_data_object_alloc(VALUE klass, void *datap, RUBY_DATA_FUNC dmark, RUBY_DATA_FUNC dfree)
{
if (klass) Check_Type(klass, T_CLASS);
return newobj_of(klass, T_DATA, (VALUE)dmark, (VALUE)dfree, (VALUE)datap);
}
VALUE
rb_data_typed_object_alloc(VALUE klass, void *datap, const rb_data_type_t *type)
{
if (klass) Check_Type(klass, T_CLASS);
return newobj_of(klass, T_DATA | (type->flags & ~T_MASK), (VALUE)type, (VALUE)1, (VALUE)datap);
}
size_t
rb_objspace_data_type_memsize(VALUE obj)
{
if (RTYPEDDATA_P(obj) && RTYPEDDATA_TYPE(obj)->function.dsize) {
return RTYPEDDATA_TYPE(obj)->function.dsize(RTYPEDDATA_DATA(obj));
}
else {
return 0;
}
}
const char *
rb_objspace_data_type_name(VALUE obj)
{
if (RTYPEDDATA_P(obj)) {
return RTYPEDDATA_TYPE(obj)->wrap_struct_name;
}
else {
return 0;
}
}
static inline int
is_pointer_to_heap(rb_objspace_t *objspace, void *ptr)
{
register RVALUE *p = RANY(ptr);
register struct heap_page *page;
register size_t hi, lo, mid;
if (p < heap_pages_lomem || p > heap_pages_himem) return FALSE;
if ((VALUE)p % sizeof(RVALUE) != 0) return FALSE;
/* check if p looks like a pointer using bsearch*/
lo = 0;
hi = heap_pages_used;
while (lo < hi) {
mid = (lo + hi) / 2;
page = heap_pages_sorted[mid];
if (page->start <= p) {
if (p < page->start + page->limit) {
return TRUE;
}
lo = mid + 1;
}
else {
hi = mid;
}
}
return FALSE;
}
static int
free_method_entry_i(st_data_t key, st_data_t value, st_data_t data)
{
rb_method_entry_t *me = (rb_method_entry_t *)value;
if (!me->mark) {
rb_free_method_entry(me);
}
return ST_CONTINUE;
}
static void
rb_free_m_tbl(st_table *tbl)
{
st_foreach(tbl, free_method_entry_i, 0);
st_free_table(tbl);
}
void
rb_free_m_tbl_wrapper(struct method_table_wrapper *wrapper)
{
if (wrapper->tbl) {
rb_free_m_tbl(wrapper->tbl);
}
xfree(wrapper);
}
static int
free_const_entry_i(st_data_t key, st_data_t value, st_data_t data)
{
rb_const_entry_t *ce = (rb_const_entry_t *)value;
xfree(ce);
return ST_CONTINUE;
}
void
rb_free_const_table(st_table *tbl)
{
st_foreach(tbl, free_const_entry_i, 0);
st_free_table(tbl);
}
static inline void
make_deferred(rb_objspace_t *objspace,RVALUE *p)
{
p->as.basic.flags = T_ZOMBIE;
p->as.free.next = heap_pages_deferred_final;
heap_pages_deferred_final = p;
}
static inline void
make_io_deferred(rb_objspace_t *objspace,RVALUE *p)
{
rb_io_t *fptr = p->as.file.fptr;
make_deferred(objspace, p);
p->as.data.dfree = (void (*)(void*))rb_io_fptr_finalize;
p->as.data.data = fptr;
}
static int
obj_free(rb_objspace_t *objspace, VALUE obj)
{
gc_event_hook(objspace, RUBY_INTERNAL_EVENT_FREEOBJ, obj);
switch (BUILTIN_TYPE(obj)) {
case T_NIL:
case T_FIXNUM:
case T_TRUE:
case T_FALSE:
rb_bug("obj_free() called for broken object");
break;
}
if (FL_TEST(obj, FL_EXIVAR)) {
rb_free_generic_ivar((VALUE)obj);
FL_UNSET(obj, FL_EXIVAR);
}
#if USE_RGENGC
#if RGENGC_AGE2_PROMOTION
if (RVALUE_YOUNG_P(obj)) {
objspace->rgengc.young_object_count--;
}
#endif
if (MARKED_IN_BITMAP(GET_HEAP_OLDGEN_BITS(obj),obj)) {
CLEAR_IN_BITMAP(GET_HEAP_OLDGEN_BITS(obj),obj);
}
#endif
switch (BUILTIN_TYPE(obj)) {
case T_OBJECT:
if (!(RANY(obj)->as.basic.flags & ROBJECT_EMBED) &&
RANY(obj)->as.object.as.heap.ivptr) {
xfree(RANY(obj)->as.object.as.heap.ivptr);
}
break;
case T_MODULE:
case T_CLASS:
if (RCLASS_M_TBL_WRAPPER(obj)) {
rb_free_m_tbl_wrapper(RCLASS_M_TBL_WRAPPER(obj));
}
if (RCLASS_IV_TBL(obj)) {
st_free_table(RCLASS_IV_TBL(obj));
}
if (RCLASS_CONST_TBL(obj)) {
rb_free_const_table(RCLASS_CONST_TBL(obj));
}
if (RCLASS_IV_INDEX_TBL(obj)) {
st_free_table(RCLASS_IV_INDEX_TBL(obj));
}
if (RCLASS_EXT(obj)->subclasses) {
if (BUILTIN_TYPE(obj) == T_MODULE) {
rb_class_detach_module_subclasses(obj);
}
else {
rb_class_detach_subclasses(obj);
}
RCLASS_EXT(obj)->subclasses = NULL;
}
rb_class_remove_from_module_subclasses(obj);
rb_class_remove_from_super_subclasses(obj);
if (RANY(obj)->as.klass.ptr)
xfree(RANY(obj)->as.klass.ptr);
RANY(obj)->as.klass.ptr = NULL;
break;
case T_STRING:
rb_str_free(obj);
break;
case T_ARRAY:
rb_ary_free(obj);
break;
case T_HASH:
if (RANY(obj)->as.hash.ntbl) {
st_free_table(RANY(obj)->as.hash.ntbl);
}
break;
case T_REGEXP:
if (RANY(obj)->as.regexp.ptr) {
onig_free(RANY(obj)->as.regexp.ptr);
}
break;
case T_DATA:
if (DATA_PTR(obj)) {
int free_immediately = FALSE;
if (RTYPEDDATA_P(obj)) {
free_immediately = (RANY(obj)->as.typeddata.type->flags & RUBY_TYPED_FREE_IMMEDIATELY) != 0;
RDATA(obj)->dfree = RANY(obj)->as.typeddata.type->function.dfree;
if (0 && free_immediately == 0) /* to expose non-free-immediate T_DATA */
fprintf(stderr, "not immediate -> %s\n", RANY(obj)->as.typeddata.type->wrap_struct_name);
}
if (RANY(obj)->as.data.dfree == RUBY_DEFAULT_FREE) {
xfree(DATA_PTR(obj));
}
else if (RANY(obj)->as.data.dfree) {
if (free_immediately) {
(RDATA(obj)->dfree)(DATA_PTR(obj));
}
else {
make_deferred(objspace, RANY(obj));
return 1;
}
}
}
break;
case T_MATCH:
if (RANY(obj)->as.match.rmatch) {
struct rmatch *rm = RANY(obj)->as.match.rmatch;
onig_region_free(&rm->regs, 0);
if (rm->char_offset)
xfree(rm->char_offset);
xfree(rm);
}
break;
case T_FILE:
if (RANY(obj)->as.file.fptr) {
make_io_deferred(objspace, RANY(obj));
return 1;
}
break;
case T_RATIONAL:
case T_COMPLEX:
break;
case T_ICLASS:
/* iClass shares table with the module */
if (RCLASS_EXT(obj)->subclasses) {
rb_class_detach_subclasses(obj);
RCLASS_EXT(obj)->subclasses = NULL;
}
rb_class_remove_from_module_subclasses(obj);
rb_class_remove_from_super_subclasses(obj);
xfree(RANY(obj)->as.klass.ptr);
RANY(obj)->as.klass.ptr = NULL;
break;
case T_FLOAT:
break;
case T_BIGNUM:
if (!(RBASIC(obj)->flags & BIGNUM_EMBED_FLAG) && BIGNUM_DIGITS(obj)) {
xfree(BIGNUM_DIGITS(obj));
}
break;
case T_NODE:
switch (nd_type(obj)) {
case NODE_SCOPE:
if (RANY(obj)->as.node.u1.tbl) {
xfree(RANY(obj)->as.node.u1.tbl);
}
break;
case NODE_ARGS:
if (RANY(obj)->as.node.u3.args) {
xfree(RANY(obj)->as.node.u3.args);
}
break;
case NODE_ALLOCA:
xfree(RANY(obj)->as.node.u1.node);
break;
}
break; /* no need to free iv_tbl */
case T_STRUCT:
if ((RBASIC(obj)->flags & RSTRUCT_EMBED_LEN_MASK) == 0 &&
RANY(obj)->as.rstruct.as.heap.ptr) {
xfree((void *)RANY(obj)->as.rstruct.as.heap.ptr);
}
break;
* parse.y: support Symbol GC. [ruby-trunk Feature #9634] See this ticket about Symbol GC. * include/ruby/ruby.h: Declare few functions. * rb_sym2id: almost same as old SYM2ID but support dynamic symbols. * rb_id2sym: almost same as old ID2SYM but support dynamic symbols. * rb_sym2str: almost same as `rb_id2str(SYM2ID(sym))` but not pin down a dynamic symbol. Declare a new struct. * struct RSymbol: represents a dynamic symbol as object in Ruby's heaps. Add few macros. * STATIC_SYM_P: check a static symbol. * DYNAMIC_SYM_P: check a dynamic symbol. * RSYMBOL: cast to RSymbol * gc.c: declare RSymbol. support T_SYMBOL. * internal.h: Declare few functions. * rb_gc_free_dsymbol: free up a dynamic symbol. GC call this function at a sweep phase. * rb_str_dynamic_intern: convert a string to a dynamic symbol. * rb_check_id_without_pindown: not pinning function. * rb_sym2id_without_pindown: ditto. * rb_check_id_cstr_without_pindown: ditto. * string.c (Init_String): String#intern and String#to_sym use rb_str_dynamic_intern. * template/id.h.tmpl: use LSB of ID as a flag for determining a static symbol, so we shift left other ruby_id_types. * string.c: use rb_sym2str instead `rb_id2str(SYM2ID(sym))` to avoid pinning. * load.c: use xx_without_pindown function at creating temporary ID to avoid pinning. * object.c: ditto. * sprintf.c: ditto. * struct.c: ditto. * thread.c: ditto. * variable.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@45426 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-03-26 08:57:47 +04:00
case T_SYMBOL:
{
rb_gc_free_dsymbol(obj);
}
break;
default:
rb_bug("gc_sweep(): unknown data type 0x%x(%p) 0x%"PRIxVALUE,
BUILTIN_TYPE(obj), (void*)obj, RBASIC(obj)->flags);
}
return 0;
}
void
Init_heap(void)
{
rb_objspace_t *objspace = &rb_objspace;
#if RGENGC_ESTIMATE_OLDMALLOC
objspace->rgengc.oldmalloc_increase_limit = gc_params.oldmalloc_limit_min;
#endif
heap_add_pages(objspace, heap_eden, gc_params.heap_init_slots / HEAP_OBJ_LIMIT);
init_mark_stack(&objspace->mark_stack);
#ifdef USE_SIGALTSTACK
{
/* altstack of another threads are allocated in another place */
rb_thread_t *th = GET_THREAD();
void *tmp = th->altstack;
th->altstack = malloc(rb_sigaltstack_size());
free(tmp); /* free previously allocated area */
}
#endif
objspace->profile.invoke_time = getrusage_time();
finalizer_table = st_init_numtable();
}
typedef int each_obj_callback(void *, void *, size_t, void *);
struct each_obj_args {
each_obj_callback *callback;
void *data;
};
static VALUE
objspace_each_objects(VALUE arg)
{
size_t i;
struct heap_page_body *last_body = 0;
struct heap_page *page;
RVALUE *pstart, *pend;
rb_objspace_t *objspace = &rb_objspace;
struct each_obj_args *args = (struct each_obj_args *)arg;
i = 0;
while (i < heap_pages_used) {
while (0 < i && last_body < heap_pages_sorted[i-1]->body) i--;
while (i < heap_pages_used && heap_pages_sorted[i]->body <= last_body) i++;
if (heap_pages_used <= i) break;
page = heap_pages_sorted[i];
last_body = page->body;
pstart = page->start;
pend = pstart + page->limit;
if ((*args->callback)(pstart, pend, sizeof(RVALUE), args->data)) {
break;
}
}
return Qnil;
}
/*
* rb_objspace_each_objects() is special C API to walk through
* Ruby object space. This C API is too difficult to use it.
* To be frank, you should not use it. Or you need to read the
* source code of this function and understand what this function does.
*
* 'callback' will be called several times (the number of heap page,
* at current implementation) with:
* vstart: a pointer to the first living object of the heap_page.
* vend: a pointer to next to the valid heap_page area.
* stride: a distance to next VALUE.
*
* If callback() returns non-zero, the iteration will be stopped.
*
* This is a sample callback code to iterate liveness objects:
*
* int
* sample_callback(void *vstart, void *vend, int stride, void *data) {
* VALUE v = (VALUE)vstart;
* for (; v != (VALUE)vend; v += stride) {
* if (RBASIC(v)->flags) { // liveness check
* // do something with live object 'v'
* }
* return 0; // continue to iteration
* }
*
* Note: 'vstart' is not a top of heap_page. This point the first
* living object to grasp at least one object to avoid GC issue.
* This means that you can not walk through all Ruby object page
* including freed object page.
*
* Note: On this implementation, 'stride' is same as sizeof(RVALUE).
* However, there are possibilities to pass variable values with
* 'stride' with some reasons. You must use stride instead of
* use some constant value in the iteration.
*/
void
rb_objspace_each_objects(each_obj_callback *callback, void *data)
{
struct each_obj_args args;
rb_objspace_t *objspace = &rb_objspace;
int prev_dont_lazy_sweep = objspace->flags.dont_lazy_sweep;
gc_rest_sweep(objspace);
objspace->flags.dont_lazy_sweep = TRUE;
args.callback = callback;
args.data = data;
if (prev_dont_lazy_sweep) {
objspace_each_objects((VALUE)&args);
}
else {
rb_ensure(objspace_each_objects, (VALUE)&args, lazy_sweep_enable, Qnil);
}
}
void
rb_objspace_each_objects_without_setup(each_obj_callback *callback, void *data)
{
struct each_obj_args args;
args.callback = callback;
args.data = data;
objspace_each_objects((VALUE)&args);
}
struct os_each_struct {
size_t num;
VALUE of;
};
static int
internal_object_p(VALUE obj)
{
RVALUE *p = (RVALUE *)obj;
if (p->as.basic.flags) {
switch (BUILTIN_TYPE(p)) {
case T_NONE:
case T_ICLASS:
case T_NODE:
case T_ZOMBIE:
break;
case T_CLASS:
if (FL_TEST(p, FL_SINGLETON))
break;
default:
if (!p->as.basic.klass) break;
return 0;
}
}
return 1;
}
int
rb_objspace_internal_object_p(VALUE obj)
{
return internal_object_p(obj);
}
static int
os_obj_of_i(void *vstart, void *vend, size_t stride, void *data)
{
struct os_each_struct *oes = (struct os_each_struct *)data;
RVALUE *p = (RVALUE *)vstart, *pend = (RVALUE *)vend;
for (; p != pend; p++) {
volatile VALUE v = (VALUE)p;
if (!internal_object_p(v)) {
if (!oes->of || rb_obj_is_kind_of(v, oes->of)) {
rb_yield(v);
oes->num++;
}
}
}
* eval_load.c (Init_load): delay allocating an array for rb_load_path to avoid GC problem in very early stage. (RUBY_GC_STRESS causes GC in such stage.) * variable.c (rb_gc_mark_global_tbl): rb_global_tbl may be 0 in very early stage. * thread.c (thread_cleanup_func) [IA64]: clear register stack position. (thread_start_func_2) [IA64]: record the beginning of register stack using extra argument. (rb_gc_save_machine_context) [IA64]: record the end of register stack. * gc.c [IA64] (SET_STACK_END): record the end of register stack. (garbage_collect) [IA64]: use recorded register stack area for GC marking. (yarv_machine_stack_mark) [IA64]: GC mark from the register stack area. * yarvcore.c [IA64] (rb_gc_register_stack_start): defined. (Init_VM): store th->self on stack to fix GC problem. (Init_yarv) [IA64]: initialize the beginning of register stack. * yarvcore.h (struct rb_thread_struct) [IA64]: new members for register stack area. * thread_pthread.ci (thread_start_func_1) [IA64]: call thread_start_func_2 with the end of register stack. * cont.c (struct rb_context_struct) [IA64]: new members for register stack area. (cont_mark) [IA64]: GC mark from register stack area. (cont_free) [IA64]: free saved register stack. (cont_save_machine_stack) [IA64]: record the position and contents of the register stack. (cont_capture): store cont->self on stack to fix GC problem. (cont_restore_1) [IA64]: restore the register stack. [IA64] (register_stack_extend): new function. (cont_restore_0) [IA64]: call register_stack_extend instead of cont_restore_1. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@12537 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2007-06-14 12:35:20 +04:00
return 0;
}
static VALUE
os_obj_of(VALUE of)
{
struct os_each_struct oes;
oes.num = 0;
oes.of = of;
rb_objspace_each_objects(os_obj_of_i, &oes);
return SIZET2NUM(oes.num);
}
/*
* call-seq:
* ObjectSpace.each_object([module]) {|obj| ... } -> fixnum
* ObjectSpace.each_object([module]) -> an_enumerator
*
* Calls the block once for each living, nonimmediate object in this
* Ruby process. If <i>module</i> is specified, calls the block
* for only those classes or modules that match (or are a subclass of)
* <i>module</i>. Returns the number of objects found. Immediate
* objects (<code>Fixnum</code>s, <code>Symbol</code>s
* <code>true</code>, <code>false</code>, and <code>nil</code>) are
* never returned. In the example below, <code>each_object</code>
* returns both the numbers we defined and several constants defined in
* the <code>Math</code> module.
*
* If no block is given, an enumerator is returned instead.
*
* a = 102.7
* b = 95 # Won't be returned
* c = 12345678987654321
* count = ObjectSpace.each_object(Numeric) {|x| p x }
* puts "Total count: #{count}"
*
* <em>produces:</em>
*
* 12345678987654321
* 102.7
* 2.71828182845905
* 3.14159265358979
* 2.22044604925031e-16
* 1.7976931348623157e+308
* 2.2250738585072e-308
* Total count: 7
*
*/
static VALUE
os_each_obj(int argc, VALUE *argv, VALUE os)
{
VALUE of;
if (argc == 0) {
of = 0;
}
else {
rb_scan_args(argc, argv, "01", &of);
}
RETURN_ENUMERATOR(os, 1, &of);
return os_obj_of(of);
}
/*
* call-seq:
* ObjectSpace.undefine_finalizer(obj)
*
* Removes all finalizers for <i>obj</i>.
*
*/
static VALUE
undefine_final(VALUE os, VALUE obj)
{
return rb_undefine_finalizer(obj);
}
VALUE
rb_undefine_finalizer(VALUE obj)
{
rb_objspace_t *objspace = &rb_objspace;
st_data_t data = obj;
rb_check_frozen(obj);
st_delete(finalizer_table, &data, 0);
FL_UNSET(obj, FL_FINALIZE);
return obj;
}
static void
should_be_callable(VALUE block)
{
if (!rb_obj_respond_to(block, rb_intern("call"), TRUE)) {
rb_raise(rb_eArgError, "wrong type argument %s (should be callable)",
rb_obj_classname(block));
}
}
static void
should_be_finalizable(VALUE obj)
{
rb_check_frozen(obj);
if (!FL_ABLE(obj)) {
rb_raise(rb_eArgError, "cannot define finalizer for %s",
rb_obj_classname(obj));
}
}
/*
* call-seq:
* ObjectSpace.define_finalizer(obj, aProc=proc())
*
* Adds <i>aProc</i> as a finalizer, to be called after <i>obj</i>
* was destroyed.
*
*/
static VALUE
define_final(int argc, VALUE *argv, VALUE os)
{
VALUE obj, block;
rb_scan_args(argc, argv, "11", &obj, &block);
should_be_finalizable(obj);
if (argc == 1) {
block = rb_block_proc();
}
else {
should_be_callable(block);
}
* probes.d: add DTrace probe declarations. [ruby-core:27448] * array.c (empty_ary_alloc, ary_new): added array create DTrace probe. * compile.c (rb_insns_name): allowing DTrace probes to access instruction sequence name. * Makefile.in: translate probes.d file to appropriate header file. * common.mk: declare dependencies on the DTrace header. * configure.in: add a test for existence of DTrace. * eval.c (setup_exception): add a probe for when an exception is raised. * gc.c: Add DTrace probes for mark begin and end, and sweep begin and end. * hash.c (empty_hash_alloc): Add a probe for hash allocation. * insns.def: Add probes for function entry and return. * internal.h: function declaration for compile.c change. * load.c (rb_f_load): add probes for `load` entry and exit, require entry and exit, and wrapping search_required for load path search. * object.c (rb_obj_alloc): added a probe for general object creation. * parse.y (yycompile0): added a probe around parse and compile phase. * string.c (empty_str_alloc, str_new): DTrace probes for string allocation. * test/dtrace/*: tests for DTrace probes. * vm.c (vm_invoke_proc): add probes for function return on exception raise, hash create, and instruction sequence execution. * vm_core.h: add probe declarations for function entry and exit. * vm_dump.c: add probes header file. * vm_eval.c (vm_call0_cfunc, vm_call0_cfunc_with_frame): add probe on function entry and return. * vm_exec.c: expose instruction number to instruction name function. * vm_insnshelper.c: add function entry and exit probes for cfunc methods. * vm_insnhelper.h: vm usage information is always collected, so uncomment the functions. 12 19:14:50 2012 Akinori MUSHA <knu@iDaemons.org> * configure.in (isinf, isnan): isinf() and isnan() are macros on DragonFly which cannot be found by AC_REPLACE_FUNCS(). This workaround enforces the fact that they exist on DragonFly. 12 15:59:38 2012 Shugo Maeda <shugo@ruby-lang.org> * vm_core.h (rb_call_info_t::refinements), compile.c (new_callinfo), vm_insnhelper.c (vm_search_method): revert r37616 because it's too slow. [ruby-dev:46477] * test/ruby/test_refinement.rb (test_inline_method_cache): skip the test until the bug is fixed efficiently. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37631 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-11-13 01:52:12 +04:00
return define_final0(obj, block);
}
static VALUE
define_final0(VALUE obj, VALUE block)
{
rb_objspace_t *objspace = &rb_objspace;
VALUE table;
st_data_t data;
RBASIC(obj)->flags |= FL_FINALIZE;
block = rb_ary_new3(2, INT2FIX(rb_safe_level()), block);
OBJ_FREEZE(block);
if (st_lookup(finalizer_table, obj, &data)) {
table = (VALUE)data;
rb_ary_push(table, block);
}
else {
table = rb_ary_new3(1, block);
* include/ruby/ruby.h: constify RBasic::klass and add RBASIC_CLASS(obj) macro which returns a class of `obj'. This change is a part of RGENGC branch [ruby-trunk - Feature #8339]. * object.c: add new function rb_obj_reveal(). This function reveal interal (hidden) object by rb_obj_hide(). Note that do not change class before and after hiding. Only permitted example is: klass = RBASIC_CLASS(obj); rb_obj_hide(obj); .... rb_obj_reveal(obj, klass); TODO: API design. rb_obj_reveal() should be replaced with others. TODO: modify constified variables using cast may be harmful for compiler's analysis and optimizaton. Any idea to prohibt inserting RBasic::klass directly? If rename RBasic::klass and force to use RBASIC_CLASS(obj), then all codes such as `RBASIC(obj)->klass' will be compilation error. Is it acceptable? (We have similar experience at Ruby 1.9, for example "RARRAY(ary)->ptr" to "RARRAY_PTR(ary)". * internal.h: add some macros. * RBASIC_CLEAR_CLASS(obj) clear RBasic::klass to make it internal object. * RBASIC_SET_CLASS(obj, cls) set RBasic::klass. * RBASIC_SET_CLASS_RAW(obj, cls) same as RBASIC_SET_CLASS without write barrier (planned). * RCLASS_SET_SUPER(a, b) set super class of a. * array.c, class.c, compile.c, encoding.c, enum.c, error.c, eval.c, file.c, gc.c, hash.c, io.c, iseq.c, marshal.c, object.c, parse.y, proc.c, process.c, random.c, ruby.c, sprintf.c, string.c, thread.c, transcode.c, vm.c, vm_eval.c, win32/file.c: Use above macros and functions to access RBasic::klass. * ext/coverage/coverage.c, ext/readline/readline.c, ext/socket/ancdata.c, ext/socket/init.c, * ext/zlib/zlib.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40691 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 14:49:11 +04:00
RBASIC_CLEAR_CLASS(table);
st_add_direct(finalizer_table, obj, table);
}
return block;
}
VALUE
rb_define_finalizer(VALUE obj, VALUE block)
{
should_be_finalizable(obj);
should_be_callable(block);
return define_final0(obj, block);
}
void
rb_gc_copy_finalizer(VALUE dest, VALUE obj)
{
rb_objspace_t *objspace = &rb_objspace;
VALUE table;
st_data_t data;
if (!FL_TEST(obj, FL_FINALIZE)) return;
if (st_lookup(finalizer_table, obj, &data)) {
table = (VALUE)data;
st_insert(finalizer_table, dest, table);
}
FL_SET(dest, FL_FINALIZE);
}
static VALUE
run_single_final(VALUE arg)
{
VALUE *args = (VALUE *)arg;
rb_eval_cmd(args[0], args[1], (int)args[2]);
return Qnil;
}
static void
run_finalizer(rb_objspace_t *objspace, VALUE obj, VALUE table)
{
long i;
int status;
VALUE args[3];
VALUE objid = nonspecial_obj_id(obj);
if (RARRAY_LEN(table) > 0) {
args[1] = rb_obj_freeze(rb_ary_new3(1, objid));
}
else {
args[1] = 0;
}
args[2] = (VALUE)rb_safe_level();
for (i=0; i<RARRAY_LEN(table); i++) {
VALUE final = RARRAY_AREF(table, i);
args[0] = RARRAY_AREF(final, 1);
args[2] = FIX2INT(RARRAY_AREF(final, 0));
status = 0;
rb_protect(run_single_final, (VALUE)args, &status);
if (status)
rb_set_errinfo(Qnil);
}
}
static void
run_final(rb_objspace_t *objspace, VALUE obj)
{
RUBY_DATA_FUNC free_func = 0;
st_data_t key, table;
heap_pages_final_slots--;
* include/ruby/ruby.h: constify RBasic::klass and add RBASIC_CLASS(obj) macro which returns a class of `obj'. This change is a part of RGENGC branch [ruby-trunk - Feature #8339]. * object.c: add new function rb_obj_reveal(). This function reveal interal (hidden) object by rb_obj_hide(). Note that do not change class before and after hiding. Only permitted example is: klass = RBASIC_CLASS(obj); rb_obj_hide(obj); .... rb_obj_reveal(obj, klass); TODO: API design. rb_obj_reveal() should be replaced with others. TODO: modify constified variables using cast may be harmful for compiler's analysis and optimizaton. Any idea to prohibt inserting RBasic::klass directly? If rename RBasic::klass and force to use RBASIC_CLASS(obj), then all codes such as `RBASIC(obj)->klass' will be compilation error. Is it acceptable? (We have similar experience at Ruby 1.9, for example "RARRAY(ary)->ptr" to "RARRAY_PTR(ary)". * internal.h: add some macros. * RBASIC_CLEAR_CLASS(obj) clear RBasic::klass to make it internal object. * RBASIC_SET_CLASS(obj, cls) set RBasic::klass. * RBASIC_SET_CLASS_RAW(obj, cls) same as RBASIC_SET_CLASS without write barrier (planned). * RCLASS_SET_SUPER(a, b) set super class of a. * array.c, class.c, compile.c, encoding.c, enum.c, error.c, eval.c, file.c, gc.c, hash.c, io.c, iseq.c, marshal.c, object.c, parse.y, proc.c, process.c, random.c, ruby.c, sprintf.c, string.c, thread.c, transcode.c, vm.c, vm_eval.c, win32/file.c: Use above macros and functions to access RBasic::klass. * ext/coverage/coverage.c, ext/readline/readline.c, ext/socket/ancdata.c, ext/socket/init.c, * ext/zlib/zlib.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40691 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 14:49:11 +04:00
RBASIC_CLEAR_CLASS(obj);
if (RTYPEDDATA_P(obj)) {
free_func = RTYPEDDATA_TYPE(obj)->function.dfree;
}
else {
free_func = RDATA(obj)->dfree;
}
if (free_func) {
(*free_func)(DATA_PTR(obj));
}
key = (st_data_t)obj;
if (st_delete(finalizer_table, &key, &table)) {
run_finalizer(objspace, obj, (VALUE)table);
}
}
static void
finalize_list(rb_objspace_t *objspace, RVALUE *p)
{
while (p) {
RVALUE *tmp = p->as.free.next;
struct heap_page *page = GET_HEAP_PAGE(p);
run_final(objspace, (VALUE)p);
objspace->profile.total_freed_object_num++;
page->final_slots--;
heap_page_add_freeobj(objspace, GET_HEAP_PAGE(p), (VALUE)p);
heap_pages_swept_slots++;
p = tmp;
}
}
static void
finalize_deferred(rb_objspace_t *objspace)
{
RVALUE *p;
while ((p = ATOMIC_PTR_EXCHANGE(heap_pages_deferred_final, 0)) != 0) {
finalize_list(objspace, p);
}
}
static void
gc_finalize_deferred(void *dmy)
{
rb_objspace_t *objspace = &rb_objspace;
if (ATOMIC_EXCHANGE(finalizing, 1)) return;
finalize_deferred(objspace);
ATOMIC_SET(finalizing, 0);
}
/* TODO: to keep compatibility, maybe unused. */
void
rb_gc_finalize_deferred(void)
{
gc_finalize_deferred(0);
}
static void
gc_finalize_deferred_register(void)
{
if (rb_postponed_job_register_one(0, gc_finalize_deferred, 0) == 0) {
rb_bug("gc_finalize_deferred_register: can't register finalizer.");
}
}
struct force_finalize_list {
VALUE obj;
VALUE table;
struct force_finalize_list *next;
};
static int
force_chain_object(st_data_t key, st_data_t val, st_data_t arg)
{
struct force_finalize_list **prev = (struct force_finalize_list **)arg;
struct force_finalize_list *curr = ALLOC(struct force_finalize_list);
curr->obj = key;
curr->table = val;
curr->next = *prev;
*prev = curr;
return ST_CONTINUE;
}
void
rb_gc_call_finalizer_at_exit(void)
{
rb_objspace_call_finalizer(&rb_objspace);
}
static void
rb_objspace_call_finalizer(rb_objspace_t *objspace)
{
RVALUE *p, *pend;
size_t i;
gc_rest_sweep(objspace);
if (ATOMIC_EXCHANGE(finalizing, 1)) return;
/* run finalizers */
finalize_deferred(objspace);
assert(heap_pages_deferred_final == 0);
/* force to run finalizer */
while (finalizer_table->num_entries) {
struct force_finalize_list *list = 0;
st_foreach(finalizer_table, force_chain_object, (st_data_t)&list);
while (list) {
struct force_finalize_list *curr = list;
st_data_t obj = (st_data_t)curr->obj;
run_finalizer(objspace, curr->obj, curr->table);
st_delete(finalizer_table, &obj, 0);
list = curr->next;
xfree(curr);
}
}
/* finalizers are part of garbage collection */
during_gc++;
/* run data object's finalizers */
for (i = 0; i < heap_pages_used; i++) {
p = heap_pages_sorted[i]->start; pend = p + heap_pages_sorted[i]->limit;
while (p < pend) {
switch (BUILTIN_TYPE(p)) {
case T_DATA:
if (!DATA_PTR(p) || !RANY(p)->as.data.dfree) break;
if (rb_obj_is_thread((VALUE)p)) break;
if (rb_obj_is_mutex((VALUE)p)) break;
if (rb_obj_is_fiber((VALUE)p)) break;
p->as.free.flags = 0;
if (RTYPEDDATA_P(p)) {
RDATA(p)->dfree = RANY(p)->as.typeddata.type->function.dfree;
}
if (RANY(p)->as.data.dfree == (RUBY_DATA_FUNC)-1) {
xfree(DATA_PTR(p));
}
else if (RANY(p)->as.data.dfree) {
make_deferred(objspace, RANY(p));
}
break;
case T_FILE:
if (RANY(p)->as.file.fptr) {
make_io_deferred(objspace, RANY(p));
}
break;
}
p++;
}
}
during_gc = 0;
if (heap_pages_deferred_final) {
finalize_list(objspace, heap_pages_deferred_final);
}
st_free_table(finalizer_table);
finalizer_table = 0;
ATOMIC_SET(finalizing, 0);
}
static inline int
is_id_value(rb_objspace_t *objspace, VALUE ptr)
{
if (!is_pointer_to_heap(objspace, (void *)ptr)) return FALSE;
if (BUILTIN_TYPE(ptr) > T_FIXNUM) return FALSE;
if (BUILTIN_TYPE(ptr) == T_ICLASS) return FALSE;
return TRUE;
}
static inline int
heap_is_swept_object(rb_objspace_t *objspace, rb_heap_t *heap, VALUE ptr)
{
struct heap_page *page = GET_HEAP_PAGE(ptr);
return page->before_sweep ? FALSE : TRUE;
}
static inline int
is_swept_object(rb_objspace_t *objspace, VALUE ptr)
{
if (heap_is_swept_object(objspace, heap_eden, ptr)) {
return TRUE;
}
else {
return FALSE;
}
}
static inline int
is_dead_object(rb_objspace_t *objspace, VALUE ptr)
{
if (!is_lazy_sweeping(heap_eden) || MARKED_IN_BITMAP(GET_HEAP_MARK_BITS(ptr), ptr)) return FALSE;
if (!is_swept_object(objspace, ptr)) return TRUE;
return FALSE;
}
static inline int
is_live_object(rb_objspace_t *objspace, VALUE ptr)
{
switch (BUILTIN_TYPE(ptr)) {
case 0: case T_ZOMBIE:
return FALSE;
}
if (is_dead_object(objspace, ptr)) return FALSE;
return TRUE;
}
static inline int
is_markable_object(rb_objspace_t *objspace, VALUE obj)
{
if (rb_special_const_p(obj)) return 0; /* special const is not markable */
if (RGENGC_CHECK_MODE) {
if (!is_pointer_to_heap(objspace, (void *)obj)) rb_bug("is_markable_object: %p is not pointer to heap", (void *)obj);
if (BUILTIN_TYPE(obj) == T_NONE) rb_bug("is_markable_object: %p is T_NONE", (void *)obj);
if (BUILTIN_TYPE(obj) == T_ZOMBIE) rb_bug("is_markable_object: %p is T_ZOMBIE", (void *)obj);
}
return 1;
}
int
rb_objspace_markable_object_p(VALUE obj)
{
rb_objspace_t *objspace = &rb_objspace;
return is_markable_object(objspace, obj) && is_live_object(objspace, obj);
}
/*
* call-seq:
* ObjectSpace._id2ref(object_id) -> an_object
*
* Converts an object id to a reference to the object. May not be
* called on an object id passed as a parameter to a finalizer.
*
* s = "I am a string" #=> "I am a string"
* r = ObjectSpace._id2ref(s.object_id) #=> "I am a string"
* r == s #=> true
*
*/
static VALUE
id2ref(VALUE obj, VALUE objid)
{
#if SIZEOF_LONG == SIZEOF_VOIDP
#define NUM2PTR(x) NUM2ULONG(x)
#elif SIZEOF_LONG_LONG == SIZEOF_VOIDP
#define NUM2PTR(x) NUM2ULL(x)
#endif
rb_objspace_t *objspace = &rb_objspace;
VALUE ptr;
void *p0;
ptr = NUM2PTR(objid);
p0 = (void *)ptr;
if (ptr == Qtrue) return Qtrue;
if (ptr == Qfalse) return Qfalse;
if (ptr == Qnil) return Qnil;
if (FIXNUM_P(ptr)) return (VALUE)ptr;
if (FLONUM_P(ptr)) return (VALUE)ptr;
ptr = obj_id_to_ref(objid);
if ((ptr % sizeof(RVALUE)) == (4 << 2)) {
ID symid = ptr / sizeof(RVALUE);
if (rb_id2name(symid) == 0)
rb_raise(rb_eRangeError, "%p is not symbol id value", p0);
return ID2SYM(symid);
}
if (!is_id_value(objspace, ptr)) {
rb_raise(rb_eRangeError, "%p is not id value", p0);
}
if (!is_live_object(objspace, ptr)) {
rb_raise(rb_eRangeError, "%p is recycled object", p0);
}
return (VALUE)ptr;
}
/*
* Document-method: __id__
* Document-method: object_id
*
* call-seq:
* obj.__id__ -> integer
* obj.object_id -> integer
*
* Returns an integer identifier for +obj+.
*
* The same number will be returned on all calls to +object_id+ for a given object,
* and no two active objects will share an id.
*
* Note that some objects of builtin classes are reused for optimization.
* This is the case for immediate values and frozen string litterals.
* Immediate values are not passed by reference but are passed by value:
* +nil+, +true+, +false+, Fixnums, Symbols. Some Floats may be immediates too.
*
* Object.new.object_id == Object.new.object_id # => false
* (21 * 2).object_id == (21 * 2).object_id # => true
* "hello".object_id == "hello".object_id # => false
* "hi".freeze.object_id == "hi".freeze.object_id # => true
*/
VALUE
rb_obj_id(VALUE obj)
{
/*
* 32-bit VALUE space
* MSB ------------------------ LSB
* false 00000000000000000000000000000000
* true 00000000000000000000000000000010
* nil 00000000000000000000000000000100
* undef 00000000000000000000000000000110
* symbol ssssssssssssssssssssssss00001110
* object oooooooooooooooooooooooooooooo00 = 0 (mod sizeof(RVALUE))
* fixnum fffffffffffffffffffffffffffffff1
*
* object_id space
* LSB
* false 00000000000000000000000000000000
* true 00000000000000000000000000000010
* nil 00000000000000000000000000000100
* undef 00000000000000000000000000000110
* symbol 000SSSSSSSSSSSSSSSSSSSSSSSSSSS0 S...S % A = 4 (S...S = s...s * A + 4)
* object oooooooooooooooooooooooooooooo0 o...o % A = 0
* fixnum fffffffffffffffffffffffffffffff1 bignum if required
*
* where A = sizeof(RVALUE)/4
*
* sizeof(RVALUE) is
* 20 if 32-bit, double is 4-byte aligned
* 24 if 32-bit, double is 8-byte aligned
* 40 if 64-bit
*/
* parse.y: support Symbol GC. [ruby-trunk Feature #9634] See this ticket about Symbol GC. * include/ruby/ruby.h: Declare few functions. * rb_sym2id: almost same as old SYM2ID but support dynamic symbols. * rb_id2sym: almost same as old ID2SYM but support dynamic symbols. * rb_sym2str: almost same as `rb_id2str(SYM2ID(sym))` but not pin down a dynamic symbol. Declare a new struct. * struct RSymbol: represents a dynamic symbol as object in Ruby's heaps. Add few macros. * STATIC_SYM_P: check a static symbol. * DYNAMIC_SYM_P: check a dynamic symbol. * RSYMBOL: cast to RSymbol * gc.c: declare RSymbol. support T_SYMBOL. * internal.h: Declare few functions. * rb_gc_free_dsymbol: free up a dynamic symbol. GC call this function at a sweep phase. * rb_str_dynamic_intern: convert a string to a dynamic symbol. * rb_check_id_without_pindown: not pinning function. * rb_sym2id_without_pindown: ditto. * rb_check_id_cstr_without_pindown: ditto. * string.c (Init_String): String#intern and String#to_sym use rb_str_dynamic_intern. * template/id.h.tmpl: use LSB of ID as a flag for determining a static symbol, so we shift left other ruby_id_types. * string.c: use rb_sym2str instead `rb_id2str(SYM2ID(sym))` to avoid pinning. * load.c: use xx_without_pindown function at creating temporary ID to avoid pinning. * object.c: ditto. * sprintf.c: ditto. * struct.c: ditto. * thread.c: ditto. * variable.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@45426 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-03-26 08:57:47 +04:00
if (STATIC_SYM_P(obj)) {
return (SYM2ID(obj) * sizeof(RVALUE) + (4 << 2)) | FIXNUM_FLAG;
}
else if (FLONUM_P(obj)) {
#if SIZEOF_LONG == SIZEOF_VOIDP
return LONG2NUM((SIGNED_VALUE)obj);
#else
return LL2NUM((SIGNED_VALUE)obj);
#endif
}
else if (SPECIAL_CONST_P(obj)) {
return LONG2NUM((SIGNED_VALUE)obj);
}
return nonspecial_obj_id(obj);
}
size_t rb_str_memsize(VALUE);
size_t rb_ary_memsize(VALUE);
size_t rb_io_memsize(const rb_io_t *);
size_t rb_generic_ivar_memsize(VALUE);
#include "regint.h"
static size_t
obj_memsize_of(VALUE obj, int use_tdata)
{
size_t size = 0;
if (SPECIAL_CONST_P(obj)) {
return 0;
}
if (FL_TEST(obj, FL_EXIVAR)) {
size += rb_generic_ivar_memsize(obj);
}
switch (BUILTIN_TYPE(obj)) {
case T_OBJECT:
if (!(RBASIC(obj)->flags & ROBJECT_EMBED) &&
ROBJECT(obj)->as.heap.ivptr) {
size += ROBJECT(obj)->as.heap.numiv * sizeof(VALUE);
}
break;
case T_MODULE:
case T_CLASS:
if (RCLASS_M_TBL_WRAPPER(obj)) {
size += sizeof(struct method_table_wrapper);
}
if (RCLASS_M_TBL(obj)) {
size += st_memsize(RCLASS_M_TBL(obj));
}
if (RCLASS_EXT(obj)) {
if (RCLASS_IV_TBL(obj)) {
size += st_memsize(RCLASS_IV_TBL(obj));
}
if (RCLASS_IV_INDEX_TBL(obj)) {
size += st_memsize(RCLASS_IV_INDEX_TBL(obj));
}
if (RCLASS(obj)->ptr->iv_tbl) {
size += st_memsize(RCLASS(obj)->ptr->iv_tbl);
}
if (RCLASS(obj)->ptr->const_tbl) {
size += st_memsize(RCLASS(obj)->ptr->const_tbl);
}
size += sizeof(rb_classext_t);
}
break;
case T_STRING:
size += rb_str_memsize(obj);
break;
case T_ARRAY:
size += rb_ary_memsize(obj);
break;
case T_HASH:
if (RHASH(obj)->ntbl) {
size += st_memsize(RHASH(obj)->ntbl);
}
break;
case T_REGEXP:
if (RREGEXP(obj)->ptr) {
size += onig_memsize(RREGEXP(obj)->ptr);
}
break;
case T_DATA:
if (use_tdata) size += rb_objspace_data_type_memsize(obj);
break;
case T_MATCH:
if (RMATCH(obj)->rmatch) {
struct rmatch *rm = RMATCH(obj)->rmatch;
size += onig_region_memsize(&rm->regs);
size += sizeof(struct rmatch_offset) * rm->char_offset_num_allocated;
size += sizeof(struct rmatch);
}
break;
case T_FILE:
if (RFILE(obj)->fptr) {
size += rb_io_memsize(RFILE(obj)->fptr);
}
break;
case T_RATIONAL:
case T_COMPLEX:
break;
case T_ICLASS:
/* iClass shares table with the module */
break;
case T_FLOAT:
* parse.y: support Symbol GC. [ruby-trunk Feature #9634] See this ticket about Symbol GC. * include/ruby/ruby.h: Declare few functions. * rb_sym2id: almost same as old SYM2ID but support dynamic symbols. * rb_id2sym: almost same as old ID2SYM but support dynamic symbols. * rb_sym2str: almost same as `rb_id2str(SYM2ID(sym))` but not pin down a dynamic symbol. Declare a new struct. * struct RSymbol: represents a dynamic symbol as object in Ruby's heaps. Add few macros. * STATIC_SYM_P: check a static symbol. * DYNAMIC_SYM_P: check a dynamic symbol. * RSYMBOL: cast to RSymbol * gc.c: declare RSymbol. support T_SYMBOL. * internal.h: Declare few functions. * rb_gc_free_dsymbol: free up a dynamic symbol. GC call this function at a sweep phase. * rb_str_dynamic_intern: convert a string to a dynamic symbol. * rb_check_id_without_pindown: not pinning function. * rb_sym2id_without_pindown: ditto. * rb_check_id_cstr_without_pindown: ditto. * string.c (Init_String): String#intern and String#to_sym use rb_str_dynamic_intern. * template/id.h.tmpl: use LSB of ID as a flag for determining a static symbol, so we shift left other ruby_id_types. * string.c: use rb_sym2str instead `rb_id2str(SYM2ID(sym))` to avoid pinning. * load.c: use xx_without_pindown function at creating temporary ID to avoid pinning. * object.c: ditto. * sprintf.c: ditto. * struct.c: ditto. * thread.c: ditto. * variable.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@45426 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-03-26 08:57:47 +04:00
case T_SYMBOL:
break;
case T_BIGNUM:
if (!(RBASIC(obj)->flags & BIGNUM_EMBED_FLAG) && BIGNUM_DIGITS(obj)) {
size += BIGNUM_LEN(obj) * sizeof(BDIGIT);
}
break;
case T_NODE:
switch (nd_type(obj)) {
case NODE_SCOPE:
if (RNODE(obj)->u1.tbl) {
/* TODO: xfree(RANY(obj)->as.node.u1.tbl); */
}
break;
case NODE_ALLOCA:
/* TODO: xfree(RANY(obj)->as.node.u1.node); */
;
}
break; /* no need to free iv_tbl */
case T_STRUCT:
if ((RBASIC(obj)->flags & RSTRUCT_EMBED_LEN_MASK) == 0 &&
RSTRUCT(obj)->as.heap.ptr) {
size += sizeof(VALUE) * RSTRUCT_LEN(obj);
}
break;
case T_ZOMBIE:
break;
default:
rb_bug("objspace/memsize_of(): unknown data type 0x%x(%p)",
BUILTIN_TYPE(obj), (void*)obj);
}
return size;
}
size_t
rb_obj_memsize_of(VALUE obj)
{
return obj_memsize_of(obj, TRUE);
}
static int
set_zero(st_data_t key, st_data_t val, st_data_t arg)
{
VALUE k = (VALUE)key;
VALUE hash = (VALUE)arg;
rb_hash_aset(hash, k, INT2FIX(0));
return ST_CONTINUE;
}
/*
* call-seq:
* ObjectSpace.count_objects([result_hash]) -> hash
*
* Counts objects for each type.
*
* It returns a hash, such as:
* {
* :TOTAL=>10000,
* :FREE=>3011,
* :T_OBJECT=>6,
* :T_CLASS=>404,
* # ...
* }
*
* The contents of the returned hash are implementation specific.
* It may be changed in future.
*
* If the optional argument +result_hash+ is given,
* it is overwritten and returned. This is intended to avoid probe effect.
*
* This method is only expected to work on C Ruby.
*
*/
static VALUE
count_objects(int argc, VALUE *argv, VALUE os)
{
rb_objspace_t *objspace = &rb_objspace;
size_t counts[T_MASK+1];
size_t freed = 0;
size_t total = 0;
size_t i;
VALUE hash;
if (rb_scan_args(argc, argv, "01", &hash) == 1) {
if (!RB_TYPE_P(hash, T_HASH))
rb_raise(rb_eTypeError, "non-hash given");
}
for (i = 0; i <= T_MASK; i++) {
counts[i] = 0;
}
for (i = 0; i < heap_pages_used; i++) {
struct heap_page *page = heap_pages_sorted[i];
RVALUE *p, *pend;
p = page->start; pend = p + page->limit;
for (;p < pend; p++) {
if (p->as.basic.flags) {
counts[BUILTIN_TYPE(p)]++;
}
else {
freed++;
}
}
total += page->limit;
}
if (hash == Qnil) {
hash = rb_hash_new();
}
else if (!RHASH_EMPTY_P(hash)) {
st_foreach(RHASH_TBL_RAW(hash), set_zero, hash);
}
rb_hash_aset(hash, ID2SYM(rb_intern("TOTAL")), SIZET2NUM(total));
rb_hash_aset(hash, ID2SYM(rb_intern("FREE")), SIZET2NUM(freed));
for (i = 0; i <= T_MASK; i++) {
VALUE type;
switch (i) {
#define COUNT_TYPE(t) case (t): type = ID2SYM(rb_intern(#t)); break;
COUNT_TYPE(T_NONE);
COUNT_TYPE(T_OBJECT);
COUNT_TYPE(T_CLASS);
COUNT_TYPE(T_MODULE);
COUNT_TYPE(T_FLOAT);
COUNT_TYPE(T_STRING);
COUNT_TYPE(T_REGEXP);
COUNT_TYPE(T_ARRAY);
COUNT_TYPE(T_HASH);
COUNT_TYPE(T_STRUCT);
COUNT_TYPE(T_BIGNUM);
COUNT_TYPE(T_FILE);
COUNT_TYPE(T_DATA);
COUNT_TYPE(T_MATCH);
COUNT_TYPE(T_COMPLEX);
COUNT_TYPE(T_RATIONAL);
COUNT_TYPE(T_NIL);
COUNT_TYPE(T_TRUE);
COUNT_TYPE(T_FALSE);
COUNT_TYPE(T_SYMBOL);
COUNT_TYPE(T_FIXNUM);
COUNT_TYPE(T_UNDEF);
COUNT_TYPE(T_NODE);
COUNT_TYPE(T_ICLASS);
COUNT_TYPE(T_ZOMBIE);
#undef COUNT_TYPE
default: type = INT2NUM(i); break;
}
if (counts[i])
rb_hash_aset(hash, type, SIZET2NUM(counts[i]));
}
return hash;
}
/*
------------------------ Garbage Collection ------------------------
*/
/* Sweeping */
static VALUE
lazy_sweep_enable(void)
{
rb_objspace_t *objspace = &rb_objspace;
objspace->flags.dont_lazy_sweep = FALSE;
return Qnil;
}
static size_t
objspace_live_slot(rb_objspace_t *objspace)
{
return objspace->profile.total_allocated_object_num - objspace->profile.total_freed_object_num;
}
static size_t
objspace_total_slot(rb_objspace_t *objspace)
{
return heap_eden->total_slots + heap_tomb->total_slots;
}
static size_t
objspace_free_slot(rb_objspace_t *objspace)
{
return objspace_total_slot(objspace) - (objspace_live_slot(objspace) - heap_pages_final_slots);
}
static void
gc_setup_mark_bits(struct heap_page *page)
{
#if USE_RGENGC
/* copy oldgen bitmap to mark bitmap */
memcpy(&page->mark_bits[0], &page->oldgen_bits[0], HEAP_BITMAP_SIZE);
#else
/* clear mark bitmap */
memset(&page->mark_bits[0], 0, HEAP_BITMAP_SIZE);
#endif
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
static inline void
gc_page_sweep(rb_objspace_t *objspace, rb_heap_t *heap, struct heap_page *sweep_page)
{
int i;
size_t empty_slots = 0, freed_slots = 0, final_slots = 0;
RVALUE *p, *pend,*offset;
bits_t *bits, bitset;
rgengc_report(1, objspace, "page_sweep: start.\n");
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
sweep_page->before_sweep = 0;
p = sweep_page->start; pend = p + sweep_page->limit;
offset = p - NUM_IN_PAGE(p);
bits = sweep_page->mark_bits;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
/* create guard : fill 1 out-of-range */
bits[BITMAP_INDEX(p)] |= BITMAP_BIT(p)-1;
bits[BITMAP_INDEX(pend)] |= ~(BITMAP_BIT(pend) - 1);
for (i=0; i < HEAP_BITMAP_LIMIT; i++) {
bitset = ~bits[i];
if (bitset) {
p = offset + i * BITS_BITLENGTH;
do {
if ((bitset & 1) && BUILTIN_TYPE(p) != T_ZOMBIE) {
if (p->as.basic.flags) {
rgengc_report(3, objspace, "page_sweep: free %p (%s)\n", p, obj_type_name((VALUE)p));
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC && RGENGC_CHECK_MODE
if (objspace->rgengc.during_minor_gc && RVALUE_OLD_P((VALUE)p)) rb_bug("page_sweep: %p (%s) is old while minor GC.", p, obj_type_name((VALUE)p));
if (rgengc_remembered(objspace, (VALUE)p)) rb_bug("page_sweep: %p (%s) is remembered.", p, obj_type_name((VALUE)p));
#endif
if (obj_free(objspace, (VALUE)p)) {
final_slots++;
}
else if (FL_TEST(p, FL_FINALIZE)) {
RDATA(p)->dfree = 0;
make_deferred(objspace,p);
final_slots++;
}
else {
(void)VALGRIND_MAKE_MEM_UNDEFINED((void*)p, sizeof(RVALUE));
heap_page_add_freeobj(objspace, sweep_page, (VALUE)p);
rgengc_report(3, objspace, "page_sweep: %p (%s) is added to freelist\n", p, obj_type_name((VALUE)p));
freed_slots++;
}
}
else {
empty_slots++;
}
}
p++;
bitset >>= 1;
} while (bitset);
}
}
gc_setup_mark_bits(sweep_page);
#if GC_PROFILE_MORE_DETAIL
if (gc_prof_enabled(objspace)) {
gc_profile_record *record = gc_prof_record(objspace);
record->removing_objects += final_slots + freed_slots;
record->empty_objects += empty_slots;
}
#endif
if (final_slots + freed_slots + empty_slots == sweep_page->limit) {
/* there are no living objects -> move this page to tomb heap */
heap_unlink_page(objspace, heap, sweep_page);
heap_add_page(objspace, heap_tomb, sweep_page);
}
else {
if (freed_slots + empty_slots > 0) {
heap_add_freepage(objspace, heap, sweep_page);
}
else {
sweep_page->free_next = NULL;
}
}
heap_pages_swept_slots += freed_slots + empty_slots;
objspace->profile.total_freed_object_num += freed_slots;
heap_pages_final_slots += final_slots;
sweep_page->final_slots = final_slots;
if (0) fprintf(stderr, "gc_page_sweep(%d): freed?: %d, limt: %d, freed_slots: %d, empty_slots: %d, final_slots: %d\n",
(int)rb_gc_count(),
final_slots + freed_slots + empty_slots == sweep_page->limit,
(int)sweep_page->limit, (int)freed_slots, (int)empty_slots, (int)final_slots);
if (heap_pages_deferred_final && !finalizing) {
rb_thread_t *th = GET_THREAD();
if (th) {
gc_finalize_deferred_register();
}
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
rgengc_report(1, objspace, "page_sweep: end.\n");
}
/* allocate additional minimum page to work */
static void
gc_heap_prepare_minimum_pages(rb_objspace_t *objspace, rb_heap_t *heap)
{
if (!heap->free_pages) {
/* there is no free after page_sweep() */
heap_set_increment(objspace, 1);
if (!heap_increment(objspace, heap)) { /* can't allocate additional free objects */
during_gc = 0;
rb_memerror();
}
}
}
static void
gc_before_heap_sweep(rb_objspace_t *objspace, rb_heap_t *heap)
{
heap->sweep_pages = heap->pages;
heap->free_pages = NULL;
if (heap->using_page) {
RVALUE **p = &heap->using_page->freelist;
while (*p) {
p = &(*p)->as.free.next;
}
*p = heap->freelist;
heap->using_page = NULL;
}
heap->freelist = NULL;
}
#if defined(__GNUC__) && __GNUC__ == 4 && __GNUC_MINOR__ == 4
__attribute__((noinline))
#endif
static void
gc_before_sweep(rb_objspace_t *objspace)
{
rb_heap_t *heap;
size_t total_limit_slot;
rgengc_report(1, objspace, "gc_before_sweep\n");
/* sweep unlinked method entries */
if (GET_VM()->unlinked_method_entry_list) {
rb_sweep_method_entry(GET_VM());
}
/* sometimes heap_pages_increment is not 0 */
heap_pages_swept_slots = heap_pages_increment * HEAP_OBJ_LIMIT;
total_limit_slot = objspace_total_slot(objspace);
heap_pages_min_free_slots = (size_t)(total_limit_slot * GC_HEAP_FREE_SLOTS_MIN_RATIO);
if (heap_pages_min_free_slots < gc_params.heap_free_slots) {
heap_pages_min_free_slots = gc_params.heap_free_slots;
}
heap_pages_max_free_slots = (size_t)(total_limit_slot * GC_HEAP_FREE_SLOTS_MAX_RATIO);
if (heap_pages_max_free_slots < gc_params.heap_init_slots) {
heap_pages_max_free_slots = gc_params.heap_init_slots;
}
if (0) fprintf(stderr, "heap_pages_min_free_slots: %d, heap_pages_max_free_slots: %d\n",
(int)heap_pages_min_free_slots, (int)heap_pages_max_free_slots);
heap = heap_eden;
gc_before_heap_sweep(objspace, heap);
gc_prof_set_malloc_info(objspace);
/* reset malloc info */
if (0) fprintf(stderr, "%d\t%d\t%d\n", (int)rb_gc_count(), (int)malloc_increase, (int)malloc_limit);
{
size_t inc = ATOMIC_SIZE_EXCHANGE(malloc_increase, 0);
size_t old_limit = malloc_limit;
if (inc > malloc_limit) {
malloc_limit = (size_t)(inc * gc_params.malloc_limit_growth_factor);
if (gc_params.malloc_limit_max > 0 && /* ignore max-check if 0 */
malloc_limit > gc_params.malloc_limit_max) {
malloc_limit = gc_params.malloc_limit_max;
}
}
else {
malloc_limit = (size_t)(malloc_limit * 0.98); /* magic number */
if (malloc_limit < gc_params.malloc_limit_min) {
malloc_limit = gc_params.malloc_limit_min;
}
}
if (0) {
if (old_limit != malloc_limit) {
fprintf(stderr, "[%"PRIuSIZE"] malloc_limit: %"PRIuSIZE" -> %"PRIuSIZE"\n",
rb_gc_count(), old_limit, malloc_limit);
}
else {
fprintf(stderr, "[%"PRIuSIZE"] malloc_limit: not changed (%"PRIuSIZE")\n",
rb_gc_count(), malloc_limit);
}
}
}
/* reset oldmalloc info */
#if RGENGC_ESTIMATE_OLDMALLOC
if (objspace->rgengc.during_minor_gc) {
if (objspace->rgengc.oldmalloc_increase > objspace->rgengc.oldmalloc_increase_limit) {
objspace->rgengc.need_major_gc = GPR_FLAG_MAJOR_BY_OLDMALLOC;;
objspace->rgengc.oldmalloc_increase_limit =
(size_t)(objspace->rgengc.oldmalloc_increase_limit * gc_params.oldmalloc_limit_growth_factor);
if (objspace->rgengc.oldmalloc_increase_limit > gc_params.oldmalloc_limit_max) {
objspace->rgengc.oldmalloc_increase_limit = gc_params.oldmalloc_limit_max;
}
}
if (0) fprintf(stderr, "%d\t%d\t%u\t%u\t%d\n",
(int)rb_gc_count(),
(int)objspace->rgengc.need_major_gc,
(unsigned int)objspace->rgengc.oldmalloc_increase,
(unsigned int)objspace->rgengc.oldmalloc_increase_limit,
(unsigned int)gc_params.oldmalloc_limit_max);
}
else {
/* major GC */
objspace->rgengc.oldmalloc_increase = 0;
if ((objspace->profile.latest_gc_info & GPR_FLAG_MAJOR_BY_OLDMALLOC) == 0) {
objspace->rgengc.oldmalloc_increase_limit =
(size_t)(objspace->rgengc.oldmalloc_increase_limit / ((gc_params.oldmalloc_limit_growth_factor - 1)/10 + 1));
if (objspace->rgengc.oldmalloc_increase_limit < gc_params.oldmalloc_limit_min) {
objspace->rgengc.oldmalloc_increase_limit = gc_params.oldmalloc_limit_min;
}
}
}
#endif
}
static void
gc_after_sweep(rb_objspace_t *objspace)
{
rb_heap_t *heap = heap_eden;
rgengc_report(1, objspace, "after_gc_sweep: heap->total_slots: %d, heap->swept_slots: %d, min_free_slots: %d\n",
(int)heap->total_slots, (int)heap_pages_swept_slots, (int)heap_pages_min_free_slots);
if (heap_pages_swept_slots < heap_pages_min_free_slots) {
heap_set_increment(objspace, heap_extend_pages(objspace));
heap_increment(objspace, heap);
#if USE_RGENGC
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (objspace->rgengc.remembered_shady_object_count + objspace->rgengc.old_object_count > (heap_pages_length * HEAP_OBJ_LIMIT) / 2) {
/* if [old]+[remembered shady] > [all object count]/2, then do major GC */
objspace->rgengc.need_major_gc = GPR_FLAG_MAJOR_BY_RESCAN;
}
#endif
}
gc_prof_set_heap_info(objspace);
heap_pages_free_unused_pages(objspace);
/* if heap_pages has unused pages, then assign them to increment */
if (heap_pages_increment < heap_tomb->page_length) {
heap_pages_increment = heap_tomb->page_length;
}
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
#if RGENGC_PROFILE > 0
if (0) {
fprintf(stderr, "%d\t%d\t%d\t%d\t%d\t%d\t%d\n",
(int)rb_gc_count(),
(int)objspace->profile.major_gc_count,
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
(int)objspace->profile.minor_gc_count,
(int)objspace->profile.promote_infant_count,
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
(int)objspace->profile.promote_young_count,
#else
0,
#endif
(int)objspace->profile.remembered_normal_object_count,
(int)objspace->rgengc.remembered_shady_object_count);
}
#endif
gc_event_hook(objspace, RUBY_INTERNAL_EVENT_GC_END_SWEEP, 0);
}
static int
gc_heap_lazy_sweep(rb_objspace_t *objspace, rb_heap_t *heap)
{
struct heap_page *page = heap->sweep_pages, *next;
int result = FALSE;
if (page == NULL) return FALSE;
#if GC_ENABLE_LAZY_SWEEP
gc_prof_sweep_timer_start(objspace);
#endif
while (page) {
heap->sweep_pages = next = page->next;
gc_page_sweep(objspace, heap, page);
if (!next) gc_after_sweep(objspace);
if (heap->free_pages) {
result = TRUE;
break;
}
page = next;
}
#if GC_ENABLE_LAZY_SWEEP
gc_prof_sweep_timer_stop(objspace);
#endif
return result;
}
static void
gc_heap_rest_sweep(rb_objspace_t *objspace, rb_heap_t *heap)
{
if (is_lazy_sweeping(heap)) {
during_gc++;
while (is_lazy_sweeping(heap)) {
gc_heap_lazy_sweep(objspace, heap);
}
during_gc = 0;
}
}
static void
gc_rest_sweep(rb_objspace_t *objspace)
{
rb_heap_t *heap = heap_eden; /* lazy sweep only for eden */
gc_heap_rest_sweep(objspace, heap);
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
static void
gc_sweep(rb_objspace_t *objspace, int immediate_sweep)
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
{
if (immediate_sweep) {
#if !GC_ENABLE_LAZY_SWEEP
gc_prof_sweep_timer_start(objspace);
#endif
gc_before_sweep(objspace);
gc_heap_rest_sweep(objspace, heap_eden);
#if !GC_ENABLE_LAZY_SWEEP
gc_prof_sweep_timer_stop(objspace);
#endif
}
else {
struct heap_page *page;
gc_before_sweep(objspace);
page = heap_eden->sweep_pages;
while (page) {
page->before_sweep = 1;
page = page->next;
}
gc_heap_lazy_sweep(objspace, heap_eden);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
}
/* Marking - Marking stack */
static void push_mark_stack(mark_stack_t *, VALUE);
static int pop_mark_stack(mark_stack_t *, VALUE *);
static void shrink_stack_chunk_cache(mark_stack_t *stack);
static stack_chunk_t *
stack_chunk_alloc(void)
{
stack_chunk_t *res;
res = malloc(sizeof(stack_chunk_t));
if (!res)
rb_memerror();
return res;
}
static inline int
is_mark_stack_empty(mark_stack_t *stack)
{
return stack->chunk == NULL;
}
static void
add_stack_chunk_cache(mark_stack_t *stack, stack_chunk_t *chunk)
{
chunk->next = stack->cache;
stack->cache = chunk;
stack->cache_size++;
}
static void
shrink_stack_chunk_cache(mark_stack_t *stack)
{
stack_chunk_t *chunk;
if (stack->unused_cache_size > (stack->cache_size/2)) {
chunk = stack->cache;
stack->cache = stack->cache->next;
stack->cache_size--;
free(chunk);
}
stack->unused_cache_size = stack->cache_size;
}
static void
push_mark_stack_chunk(mark_stack_t *stack)
{
stack_chunk_t *next;
assert(stack->index == stack->limit);
if (stack->cache_size > 0) {
next = stack->cache;
stack->cache = stack->cache->next;
stack->cache_size--;
if (stack->unused_cache_size > stack->cache_size)
stack->unused_cache_size = stack->cache_size;
}
else {
next = stack_chunk_alloc();
}
next->next = stack->chunk;
stack->chunk = next;
stack->index = 0;
}
static void
pop_mark_stack_chunk(mark_stack_t *stack)
{
stack_chunk_t *prev;
prev = stack->chunk->next;
assert(stack->index == 0);
add_stack_chunk_cache(stack, stack->chunk);
stack->chunk = prev;
stack->index = stack->limit;
}
#if defined(ENABLE_VM_OBJSPACE) && ENABLE_VM_OBJSPACE
static void
free_stack_chunks(mark_stack_t *stack)
{
stack_chunk_t *chunk = stack->chunk;
stack_chunk_t *next = NULL;
while (chunk != NULL) {
next = chunk->next;
free(chunk);
chunk = next;
}
}
#endif
static void
push_mark_stack(mark_stack_t *stack, VALUE data)
{
if (stack->index == stack->limit) {
push_mark_stack_chunk(stack);
}
stack->chunk->data[stack->index++] = data;
}
static int
pop_mark_stack(mark_stack_t *stack, VALUE *data)
{
if (is_mark_stack_empty(stack)) {
return FALSE;
}
if (stack->index == 1) {
*data = stack->chunk->data[--stack->index];
pop_mark_stack_chunk(stack);
}
else {
*data = stack->chunk->data[--stack->index];
}
return TRUE;
}
static void
init_mark_stack(mark_stack_t *stack)
{
int i;
if (0) push_mark_stack_chunk(stack);
stack->index = stack->limit = STACK_CHUNK_SIZE;
for (i=0; i < 4; i++) {
add_stack_chunk_cache(stack, stack_chunk_alloc());
}
stack->unused_cache_size = stack->cache_size;
}
/* Marking */
#ifdef __ia64
#define SET_STACK_END (SET_MACHINE_STACK_END(&th->machine.stack_end), th->machine.register_stack_end = rb_ia64_bsp())
#else
#define SET_STACK_END SET_MACHINE_STACK_END(&th->machine.stack_end)
#endif
#define STACK_START (th->machine.stack_start)
#define STACK_END (th->machine.stack_end)
#define STACK_LEVEL_MAX (th->machine.stack_maxsize/sizeof(VALUE))
#if STACK_GROW_DIRECTION < 0
# define STACK_LENGTH (size_t)(STACK_START - STACK_END)
#elif STACK_GROW_DIRECTION > 0
# define STACK_LENGTH (size_t)(STACK_END - STACK_START + 1)
#else
# define STACK_LENGTH ((STACK_END < STACK_START) ? (size_t)(STACK_START - STACK_END) \
: (size_t)(STACK_END - STACK_START + 1))
#endif
#if !STACK_GROW_DIRECTION
int ruby_stack_grow_direction;
int
ruby_get_stack_grow_direction(volatile VALUE *addr)
{
VALUE *end;
SET_MACHINE_STACK_END(&end);
if (end > addr) return ruby_stack_grow_direction = 1;
return ruby_stack_grow_direction = -1;
}
#endif
size_t
ruby_stack_length(VALUE **p)
{
rb_thread_t *th = GET_THREAD();
SET_STACK_END;
if (p) *p = STACK_UPPER(STACK_END, STACK_START, STACK_END);
return STACK_LENGTH;
}
#if !(defined(POSIX_SIGNAL) && defined(SIGSEGV) && defined(HAVE_SIGALTSTACK))
static int
stack_check(int water_mark)
{
int ret;
rb_thread_t *th = GET_THREAD();
SET_STACK_END;
ret = STACK_LENGTH > STACK_LEVEL_MAX - water_mark;
#ifdef __ia64
if (!ret) {
ret = (VALUE*)rb_ia64_bsp() - th->machine.register_stack_start >
th->machine.register_stack_maxsize/sizeof(VALUE) - water_mark;
}
#endif
return ret;
}
#endif
#define STACKFRAME_FOR_CALL_CFUNC 512
int
ruby_stack_check(void)
{
#if defined(POSIX_SIGNAL) && defined(SIGSEGV) && defined(HAVE_SIGALTSTACK)
return 0;
#else
return stack_check(STACKFRAME_FOR_CALL_CFUNC);
#endif
}
ATTRIBUTE_NO_ADDRESS_SAFETY_ANALYSIS
static void
mark_locations_array(rb_objspace_t *objspace, register VALUE *x, register long n)
{
VALUE v;
while (n--) {
v = *x;
gc_mark_maybe(objspace, v);
x++;
}
}
static void
gc_mark_locations(rb_objspace_t *objspace, VALUE *start, VALUE *end)
{
long n;
if (end <= start) return;
n = end - start;
mark_locations_array(objspace, start, n);
}
void
rb_gc_mark_locations(VALUE *start, VALUE *end)
{
gc_mark_locations(&rb_objspace, start, end);
}
#define rb_gc_mark_locations(start, end) gc_mark_locations(objspace, (start), (end))
struct mark_tbl_arg {
rb_objspace_t *objspace;
};
static int
mark_entry(st_data_t key, st_data_t value, st_data_t data)
{
struct mark_tbl_arg *arg = (void*)data;
gc_mark(arg->objspace, (VALUE)value);
return ST_CONTINUE;
}
static void
mark_tbl(rb_objspace_t *objspace, st_table *tbl)
{
struct mark_tbl_arg arg;
if (!tbl || tbl->num_entries == 0) return;
arg.objspace = objspace;
st_foreach(tbl, mark_entry, (st_data_t)&arg);
}
static int
mark_key(st_data_t key, st_data_t value, st_data_t data)
{
struct mark_tbl_arg *arg = (void*)data;
gc_mark(arg->objspace, (VALUE)key);
return ST_CONTINUE;
}
static void
mark_set(rb_objspace_t *objspace, st_table *tbl)
{
struct mark_tbl_arg arg;
if (!tbl) return;
arg.objspace = objspace;
st_foreach(tbl, mark_key, (st_data_t)&arg);
}
void
rb_mark_set(st_table *tbl)
{
mark_set(&rb_objspace, tbl);
}
static int
mark_keyvalue(st_data_t key, st_data_t value, st_data_t data)
{
struct mark_tbl_arg *arg = (void*)data;
gc_mark(arg->objspace, (VALUE)key);
gc_mark(arg->objspace, (VALUE)value);
return ST_CONTINUE;
}
static void
mark_hash(rb_objspace_t *objspace, st_table *tbl)
{
struct mark_tbl_arg arg;
if (!tbl) return;
arg.objspace = objspace;
st_foreach(tbl, mark_keyvalue, (st_data_t)&arg);
}
void
rb_mark_hash(st_table *tbl)
{
mark_hash(&rb_objspace, tbl);
}
static void
mark_method_entry(rb_objspace_t *objspace, const rb_method_entry_t *me)
{
const rb_method_definition_t *def = me->def;
gc_mark(objspace, me->klass);
* revised r37993 to avoid SEGV/ILL in tests. In r37993, a method entry with VM_METHOD_TYPE_REFINED holds only the original method definition, so ci->me is set to a method entry allocated in the stack, and it causes SEGV/ILL. In this commit, a method entry with VM_METHOD_TYPE_REFINED holds the whole original method entry. Furthermore, rb_thread_mark() is changed to mark cfp->klass to avoid GC for iclasses created by copy_refinement_iclass(). * vm_method.c (rb_method_entry_make): add a method entry with VM_METHOD_TYPE_REFINED to the class refined by the refinement if the target module is a refinement. When a method entry with VM_METHOD_TYPE_UNDEF is invoked by vm_call_method(), a method with the same name is searched in refinements. If such a method is found, the method is invoked. Otherwise, the original method in the refined class (rb_method_definition_t::body.orig_me) is invoked. This change is made to simplify the normal method lookup and to improve the performance of normal method calls. * vm_method.c (EXPR1, search_method, rb_method_entry), vm_eval.c (rb_call0, rb_search_method_entry): do not use refinements for method lookup. * vm_insnhelper.c (vm_call_method): search methods in refinements if ci->me is VM_METHOD_TYPE_REFINED. If the method is called by super (i.e., ci->call == vm_call_super_method), skip the same method entry as the current method to avoid infinite call of the same method. * class.c (include_modules_at): add a refined method entry for each method defined in a module included in a refinement. * class.c (rb_prepend_module): set an empty table to RCLASS_M_TBL(klass) to add refined method entries, because refinements should have priority over prepended modules. * proc.c (mnew): use rb_method_entry_with_refinements() to get a refined method. * vm.c (rb_thread_mark): mark cfp->klass for iclasses created by copy_refinement_iclass(). * vm.c (Init_VM), cont.c (fiber_init): initialize th->cfp->klass. * test/ruby/test_refinement.rb (test_inline_method_cache): do not skip the test because it should pass successfully. * test/ruby/test_refinement.rb (test_redefine_refined_method): new test for the case a refined method is redefined. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@38236 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-12-06 17:08:41 +04:00
again:
if (!def) return;
switch (def->type) {
case VM_METHOD_TYPE_ISEQ:
gc_mark(objspace, def->body.iseq->self);
break;
case VM_METHOD_TYPE_BMETHOD:
gc_mark(objspace, def->body.proc);
break;
case VM_METHOD_TYPE_ATTRSET:
case VM_METHOD_TYPE_IVAR:
gc_mark(objspace, def->body.attr.location);
break;
* revised r37993 to avoid SEGV/ILL in tests. In r37993, a method entry with VM_METHOD_TYPE_REFINED holds only the original method definition, so ci->me is set to a method entry allocated in the stack, and it causes SEGV/ILL. In this commit, a method entry with VM_METHOD_TYPE_REFINED holds the whole original method entry. Furthermore, rb_thread_mark() is changed to mark cfp->klass to avoid GC for iclasses created by copy_refinement_iclass(). * vm_method.c (rb_method_entry_make): add a method entry with VM_METHOD_TYPE_REFINED to the class refined by the refinement if the target module is a refinement. When a method entry with VM_METHOD_TYPE_UNDEF is invoked by vm_call_method(), a method with the same name is searched in refinements. If such a method is found, the method is invoked. Otherwise, the original method in the refined class (rb_method_definition_t::body.orig_me) is invoked. This change is made to simplify the normal method lookup and to improve the performance of normal method calls. * vm_method.c (EXPR1, search_method, rb_method_entry), vm_eval.c (rb_call0, rb_search_method_entry): do not use refinements for method lookup. * vm_insnhelper.c (vm_call_method): search methods in refinements if ci->me is VM_METHOD_TYPE_REFINED. If the method is called by super (i.e., ci->call == vm_call_super_method), skip the same method entry as the current method to avoid infinite call of the same method. * class.c (include_modules_at): add a refined method entry for each method defined in a module included in a refinement. * class.c (rb_prepend_module): set an empty table to RCLASS_M_TBL(klass) to add refined method entries, because refinements should have priority over prepended modules. * proc.c (mnew): use rb_method_entry_with_refinements() to get a refined method. * vm.c (rb_thread_mark): mark cfp->klass for iclasses created by copy_refinement_iclass(). * vm.c (Init_VM), cont.c (fiber_init): initialize th->cfp->klass. * test/ruby/test_refinement.rb (test_inline_method_cache): do not skip the test because it should pass successfully. * test/ruby/test_refinement.rb (test_redefine_refined_method): new test for the case a refined method is redefined. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@38236 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-12-06 17:08:41 +04:00
case VM_METHOD_TYPE_REFINED:
if (def->body.orig_me) {
def = def->body.orig_me->def;
goto again;
}
break;
default:
break; /* ignore */
}
}
void
rb_mark_method_entry(const rb_method_entry_t *me)
{
mark_method_entry(&rb_objspace, me);
}
static int
mark_method_entry_i(st_data_t key, st_data_t value, st_data_t data)
{
const rb_method_entry_t *me = (const rb_method_entry_t *)value;
struct mark_tbl_arg *arg = (void*)data;
mark_method_entry(arg->objspace, me);
return ST_CONTINUE;
}
static void
mark_m_tbl_wrapper(rb_objspace_t *objspace, struct method_table_wrapper *wrapper)
{
struct mark_tbl_arg arg;
if (!wrapper || !wrapper->tbl) return;
if (LIKELY(objspace->mark_func_data == 0)) {
/* prevent multiple marking during same GC cycle,
* since m_tbl is shared between several T_ICLASS */
size_t serial = rb_gc_count();
if (wrapper->serial == serial) return;
wrapper->serial = serial;
}
arg.objspace = objspace;
st_foreach(wrapper->tbl, mark_method_entry_i, (st_data_t)&arg);
}
static int
mark_const_entry_i(st_data_t key, st_data_t value, st_data_t data)
{
const rb_const_entry_t *ce = (const rb_const_entry_t *)value;
struct mark_tbl_arg *arg = (void*)data;
gc_mark(arg->objspace, ce->value);
gc_mark(arg->objspace, ce->file);
return ST_CONTINUE;
}
static void
mark_const_tbl(rb_objspace_t *objspace, st_table *tbl)
{
struct mark_tbl_arg arg;
if (!tbl) return;
arg.objspace = objspace;
st_foreach(tbl, mark_const_entry_i, (st_data_t)&arg);
}
#if STACK_GROW_DIRECTION < 0
#define GET_STACK_BOUNDS(start, end, appendix) ((start) = STACK_END, (end) = STACK_START)
#elif STACK_GROW_DIRECTION > 0
#define GET_STACK_BOUNDS(start, end, appendix) ((start) = STACK_START, (end) = STACK_END+(appendix))
#else
#define GET_STACK_BOUNDS(start, end, appendix) \
((STACK_END < STACK_START) ? \
((start) = STACK_END, (end) = STACK_START) : ((start) = STACK_START, (end) = STACK_END+(appendix)))
#endif
static void
mark_current_machine_context(rb_objspace_t *objspace, rb_thread_t *th)
{
union {
rb_jmp_buf j;
VALUE v[sizeof(rb_jmp_buf) / sizeof(VALUE)];
} save_regs_gc_mark;
VALUE *stack_start, *stack_end;
FLUSH_REGISTER_WINDOWS;
/* This assumes that all registers are saved into the jmp_buf (and stack) */
rb_setjmp(save_regs_gc_mark.j);
/* SET_STACK_END must be called in this function because
* the stack frame of this function may contain
* callee save registers and they should be marked. */
SET_STACK_END;
GET_STACK_BOUNDS(stack_start, stack_end, 1);
mark_locations_array(objspace, save_regs_gc_mark.v, numberof(save_regs_gc_mark.v));
rb_gc_mark_locations(stack_start, stack_end);
#ifdef __ia64
rb_gc_mark_locations(th->machine.register_stack_start, th->machine.register_stack_end);
#endif
#if defined(__mc68000__)
mark_locations_array(objspace, (VALUE*)((char*)STACK_END + 2),
(STACK_START - STACK_END));
#endif
}
void
rb_gc_mark_machine_stack(rb_thread_t *th)
{
rb_objspace_t *objspace = &rb_objspace;
VALUE *stack_start, *stack_end;
GET_STACK_BOUNDS(stack_start, stack_end, 0);
rb_gc_mark_locations(stack_start, stack_end);
#ifdef __ia64
rb_gc_mark_locations(th->machine.register_stack_start, th->machine.register_stack_end);
#endif
}
void
rb_mark_tbl(st_table *tbl)
{
mark_tbl(&rb_objspace, tbl);
}
static void
gc_mark_maybe(rb_objspace_t *objspace, VALUE obj)
{
(void)VALGRIND_MAKE_MEM_DEFINED(&obj, sizeof(obj));
if (is_pointer_to_heap(objspace, (void *)obj)) {
int type = BUILTIN_TYPE(obj);
if (type != T_ZOMBIE && type != T_NONE) {
gc_mark(objspace, obj);
}
}
}
void
rb_gc_mark_maybe(VALUE obj)
{
gc_mark_maybe(&rb_objspace, obj);
}
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
static inline int
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
gc_marked(rb_objspace_t *objspace, VALUE ptr)
{
register bits_t *bits = GET_HEAP_MARK_BITS(ptr);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
if (MARKED_IN_BITMAP(bits, ptr)) return 1;
return 0;
}
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
static inline int
gc_mark_ptr(rb_objspace_t *objspace, VALUE ptr)
{
register bits_t *bits = GET_HEAP_MARK_BITS(ptr);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
if (gc_marked(objspace, ptr)) return 0;
MARK_IN_BITMAP(bits, ptr);
return 1;
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
static void
rgengc_check_relation(rb_objspace_t *objspace, VALUE obj)
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
{
#if USE_RGENGC
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (objspace->rgengc.parent_object_is_old) {
if (!RVALUE_WB_PROTECTED(obj)) {
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (rgengc_remember(objspace, obj)) {
objspace->rgengc.remembered_shady_object_count++;
}
}
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
else {
if (gc_marked(objspace, obj)) {
if (!RVALUE_OLD_P(obj)) {
/* An object pointed from an OLD object should be OLD. */
rgengc_remember(objspace, obj);
}
}
else {
if (RVALUE_INFANT_P(obj)) {
RVALUE_PROMOTE_INFANT(objspace, obj);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
}
}
}
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
#endif
}
static void
gc_mark(rb_objspace_t *objspace, VALUE ptr)
{
if (!is_markable_object(objspace, ptr)) return;
if (LIKELY(objspace->mark_func_data == 0)) {
rgengc_check_relation(objspace, ptr);
if (!gc_mark_ptr(objspace, ptr)) return; /* already marked */
push_mark_stack(&objspace->mark_stack, ptr);
}
else {
objspace->mark_func_data->mark_func(ptr, objspace->mark_func_data->data);
}
}
void
rb_gc_mark(VALUE ptr)
{
gc_mark(&rb_objspace, ptr);
}
/* CAUTION: THIS FUNCTION ENABLE *ONLY BEFORE* SWEEPING.
* This function is only for GC_END_MARK timing.
*/
int
rb_objspace_marked_object_p(VALUE obj)
{
rb_objspace_t *objspace = &rb_objspace;
return gc_marked(objspace, obj) ? TRUE : FALSE;
}
/* resurrect non-marked `obj' if obj is before swept */
void
rb_gc_resurrect(VALUE obj)
{
rb_objspace_t *objspace = &rb_objspace;
if (is_lazy_sweeping(heap_eden) &&
!gc_marked(objspace, obj) &&
!is_swept_object(objspace, obj)) {
gc_mark_ptr(objspace, obj);
}
}
static void
gc_mark_children(rb_objspace_t *objspace, VALUE ptr)
{
register RVALUE *obj = RANY(ptr);
goto marking; /* skip */
again:
if (LIKELY(objspace->mark_func_data == 0)) {
obj = RANY(ptr);
if (!is_markable_object(objspace, ptr)) return;
rgengc_check_relation(objspace, ptr);
if (!gc_mark_ptr(objspace, ptr)) return; /* already marked */
}
else {
gc_mark(objspace, ptr);
return;
}
marking:
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency((VALUE)obj);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (LIKELY(objspace->mark_func_data == 0)) {
/* minor/major common */
if (RVALUE_WB_PROTECTED(obj)) {
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (RVALUE_INFANT_P((VALUE)obj)) {
/* infant -> young */
RVALUE_PROMOTE_INFANT(objspace, (VALUE)obj);
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
objspace->rgengc.parent_object_is_old = FALSE;
#else
objspace->rgengc.parent_object_is_old = TRUE;
#endif
rgengc_report(3, objspace, "gc_mark_children: promote infant -> young %p (%s).\n", (void *)obj, obj_type_name((VALUE)obj));
}
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
else {
objspace->rgengc.parent_object_is_old = TRUE;
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (RVALUE_YOUNG_P((VALUE)obj)) {
/* young -> old */
RVALUE_PROMOTE_YOUNG(objspace, (VALUE)obj);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
rgengc_report(3, objspace, "gc_mark_children: promote young -> old %p (%s).\n", (void *)obj, obj_type_name((VALUE)obj));
}
else {
#endif
if (!objspace->rgengc.during_minor_gc) {
/* major/full GC */
objspace->rgengc.old_object_count++;
}
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
}
#endif
}
}
else {
rgengc_report(3, objspace, "gc_mark_children: do not promote non-WB-protected %p (%s).\n", (void *)obj, obj_type_name((VALUE)obj));
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
objspace->rgengc.parent_object_is_old = FALSE;
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency((VALUE)obj);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif /* USE_RGENGC */
if (FL_TEST(obj, FL_EXIVAR)) {
rb_mark_generic_ivar(ptr);
}
switch (BUILTIN_TYPE(obj)) {
case T_NIL:
case T_FIXNUM:
rb_bug("rb_gc_mark() called for broken object");
break;
case T_NODE:
switch (nd_type(obj)) {
case NODE_IF: /* 1,2,3 */
case NODE_FOR:
case NODE_ITER:
case NODE_WHEN:
case NODE_MASGN:
case NODE_RESCUE:
case NODE_RESBODY:
case NODE_CLASS:
case NODE_BLOCK_PASS:
gc_mark(objspace, (VALUE)obj->as.node.u2.node);
/* fall through */
case NODE_BLOCK: /* 1,3 */
case NODE_ARRAY:
case NODE_DSTR:
case NODE_DXSTR:
case NODE_DREGX:
case NODE_DREGX_ONCE:
case NODE_ENSURE:
case NODE_CALL:
case NODE_DEFS:
case NODE_OP_ASGN1:
gc_mark(objspace, (VALUE)obj->as.node.u1.node);
/* fall through */
case NODE_SUPER: /* 3 */
case NODE_FCALL:
case NODE_DEFN:
case NODE_ARGS_AUX:
ptr = (VALUE)obj->as.node.u3.node;
goto again;
case NODE_WHILE: /* 1,2 */
case NODE_UNTIL:
case NODE_AND:
case NODE_OR:
case NODE_CASE:
case NODE_SCLASS:
case NODE_DOT2:
case NODE_DOT3:
case NODE_FLIP2:
case NODE_FLIP3:
case NODE_MATCH2:
case NODE_MATCH3:
case NODE_OP_ASGN_OR:
case NODE_OP_ASGN_AND:
case NODE_MODULE:
case NODE_ALIAS:
case NODE_VALIAS:
case NODE_ARGSCAT:
gc_mark(objspace, (VALUE)obj->as.node.u1.node);
/* fall through */
case NODE_GASGN: /* 2 */
case NODE_LASGN:
case NODE_DASGN:
case NODE_DASGN_CURR:
case NODE_IASGN:
case NODE_IASGN2:
case NODE_CVASGN:
case NODE_COLON3:
case NODE_OPT_N:
case NODE_EVSTR:
case NODE_UNDEF:
case NODE_POSTEXE:
ptr = (VALUE)obj->as.node.u2.node;
goto again;
case NODE_HASH: /* 1 */
case NODE_LIT:
case NODE_STR:
case NODE_XSTR:
case NODE_DEFINED:
case NODE_MATCH:
case NODE_RETURN:
case NODE_BREAK:
case NODE_NEXT:
case NODE_YIELD:
case NODE_COLON2:
case NODE_SPLAT:
case NODE_TO_ARY:
ptr = (VALUE)obj->as.node.u1.node;
goto again;
case NODE_SCOPE: /* 2,3 */
case NODE_CDECL:
case NODE_OPT_ARG:
gc_mark(objspace, (VALUE)obj->as.node.u3.node);
ptr = (VALUE)obj->as.node.u2.node;
goto again;
case NODE_ARGS: /* custom */
{
struct rb_args_info *args = obj->as.node.u3.args;
if (args) {
if (args->pre_init) gc_mark(objspace, (VALUE)args->pre_init);
if (args->post_init) gc_mark(objspace, (VALUE)args->post_init);
if (args->opt_args) gc_mark(objspace, (VALUE)args->opt_args);
if (args->kw_args) gc_mark(objspace, (VALUE)args->kw_args);
if (args->kw_rest_arg) gc_mark(objspace, (VALUE)args->kw_rest_arg);
}
}
ptr = (VALUE)obj->as.node.u2.node;
goto again;
case NODE_ZARRAY: /* - */
case NODE_ZSUPER:
case NODE_VCALL:
case NODE_GVAR:
case NODE_LVAR:
case NODE_DVAR:
case NODE_IVAR:
case NODE_CVAR:
case NODE_NTH_REF:
case NODE_BACK_REF:
case NODE_REDO:
case NODE_RETRY:
case NODE_SELF:
case NODE_NIL:
case NODE_TRUE:
case NODE_FALSE:
case NODE_ERRINFO:
case NODE_BLOCK_ARG:
break;
case NODE_ALLOCA:
mark_locations_array(objspace,
(VALUE*)obj->as.node.u1.value,
obj->as.node.u3.cnt);
gc_mark(objspace, (VALUE)obj->as.node.u2.node);
break;
case NODE_CREF:
gc_mark(objspace, obj->as.node.nd_refinements);
gc_mark(objspace, (VALUE)obj->as.node.nd_clss);
ptr = (VALUE)obj->as.node.nd_next;
goto again;
default: /* unlisted NODE */
gc_mark_maybe(objspace, (VALUE)obj->as.node.u1.node);
gc_mark_maybe(objspace, (VALUE)obj->as.node.u2.node);
gc_mark_maybe(objspace, (VALUE)obj->as.node.u3.node);
}
return; /* no need to mark class. */
}
gc_mark(objspace, obj->as.basic.klass);
switch (BUILTIN_TYPE(obj)) {
case T_ICLASS:
case T_CLASS:
case T_MODULE:
mark_m_tbl_wrapper(objspace, RCLASS_M_TBL_WRAPPER(obj));
if (!RCLASS_EXT(obj)) break;
mark_tbl(objspace, RCLASS_IV_TBL(obj));
mark_const_tbl(objspace, RCLASS_CONST_TBL(obj));
* include/ruby/ruby.h: constify RBasic::klass and add RBASIC_CLASS(obj) macro which returns a class of `obj'. This change is a part of RGENGC branch [ruby-trunk - Feature #8339]. * object.c: add new function rb_obj_reveal(). This function reveal interal (hidden) object by rb_obj_hide(). Note that do not change class before and after hiding. Only permitted example is: klass = RBASIC_CLASS(obj); rb_obj_hide(obj); .... rb_obj_reveal(obj, klass); TODO: API design. rb_obj_reveal() should be replaced with others. TODO: modify constified variables using cast may be harmful for compiler's analysis and optimizaton. Any idea to prohibt inserting RBasic::klass directly? If rename RBasic::klass and force to use RBASIC_CLASS(obj), then all codes such as `RBASIC(obj)->klass' will be compilation error. Is it acceptable? (We have similar experience at Ruby 1.9, for example "RARRAY(ary)->ptr" to "RARRAY_PTR(ary)". * internal.h: add some macros. * RBASIC_CLEAR_CLASS(obj) clear RBasic::klass to make it internal object. * RBASIC_SET_CLASS(obj, cls) set RBasic::klass. * RBASIC_SET_CLASS_RAW(obj, cls) same as RBASIC_SET_CLASS without write barrier (planned). * RCLASS_SET_SUPER(a, b) set super class of a. * array.c, class.c, compile.c, encoding.c, enum.c, error.c, eval.c, file.c, gc.c, hash.c, io.c, iseq.c, marshal.c, object.c, parse.y, proc.c, process.c, random.c, ruby.c, sprintf.c, string.c, thread.c, transcode.c, vm.c, vm_eval.c, win32/file.c: Use above macros and functions to access RBasic::klass. * ext/coverage/coverage.c, ext/readline/readline.c, ext/socket/ancdata.c, ext/socket/init.c, * ext/zlib/zlib.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40691 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 14:49:11 +04:00
ptr = RCLASS_SUPER((VALUE)obj);
goto again;
case T_ARRAY:
if (FL_TEST(obj, ELTS_SHARED)) {
ptr = obj->as.array.as.heap.aux.shared;
goto again;
}
else {
long i, len = RARRAY_LEN(obj);
const VALUE *ptr = RARRAY_CONST_PTR(obj);
for (i=0; i < len; i++) {
gc_mark(objspace, *ptr++);
}
}
break;
case T_HASH:
mark_hash(objspace, obj->as.hash.ntbl);
ptr = obj->as.hash.ifnone;
goto again;
case T_STRING:
if (STR_SHARED_P(obj)) {
ptr = obj->as.string.as.heap.aux.shared;
goto again;
}
break;
case T_DATA:
if (RTYPEDDATA_P(obj)) {
RUBY_DATA_FUNC mark_func = obj->as.typeddata.type->function.dmark;
if (mark_func) (*mark_func)(DATA_PTR(obj));
}
else {
if (obj->as.data.dmark) (*obj->as.data.dmark)(DATA_PTR(obj));
}
break;
case T_OBJECT:
{
long i, len = ROBJECT_NUMIV(obj);
VALUE *ptr = ROBJECT_IVPTR(obj);
for (i = 0; i < len; i++) {
gc_mark(objspace, *ptr++);
}
}
break;
case T_FILE:
if (obj->as.file.fptr) {
gc_mark(objspace, obj->as.file.fptr->pathv);
gc_mark(objspace, obj->as.file.fptr->tied_io_for_writing);
gc_mark(objspace, obj->as.file.fptr->writeconv_asciicompat);
gc_mark(objspace, obj->as.file.fptr->writeconv_pre_ecopts);
gc_mark(objspace, obj->as.file.fptr->encs.ecopts);
gc_mark(objspace, obj->as.file.fptr->write_lock);
}
break;
case T_REGEXP:
ptr = obj->as.regexp.src;
goto again;
case T_FLOAT:
case T_BIGNUM:
* parse.y: support Symbol GC. [ruby-trunk Feature #9634] See this ticket about Symbol GC. * include/ruby/ruby.h: Declare few functions. * rb_sym2id: almost same as old SYM2ID but support dynamic symbols. * rb_id2sym: almost same as old ID2SYM but support dynamic symbols. * rb_sym2str: almost same as `rb_id2str(SYM2ID(sym))` but not pin down a dynamic symbol. Declare a new struct. * struct RSymbol: represents a dynamic symbol as object in Ruby's heaps. Add few macros. * STATIC_SYM_P: check a static symbol. * DYNAMIC_SYM_P: check a dynamic symbol. * RSYMBOL: cast to RSymbol * gc.c: declare RSymbol. support T_SYMBOL. * internal.h: Declare few functions. * rb_gc_free_dsymbol: free up a dynamic symbol. GC call this function at a sweep phase. * rb_str_dynamic_intern: convert a string to a dynamic symbol. * rb_check_id_without_pindown: not pinning function. * rb_sym2id_without_pindown: ditto. * rb_check_id_cstr_without_pindown: ditto. * string.c (Init_String): String#intern and String#to_sym use rb_str_dynamic_intern. * template/id.h.tmpl: use LSB of ID as a flag for determining a static symbol, so we shift left other ruby_id_types. * string.c: use rb_sym2str instead `rb_id2str(SYM2ID(sym))` to avoid pinning. * load.c: use xx_without_pindown function at creating temporary ID to avoid pinning. * object.c: ditto. * sprintf.c: ditto. * struct.c: ditto. * thread.c: ditto. * variable.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@45426 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-03-26 08:57:47 +04:00
case T_SYMBOL:
break;
case T_MATCH:
gc_mark(objspace, obj->as.match.regexp);
if (obj->as.match.str) {
ptr = obj->as.match.str;
goto again;
}
break;
case T_RATIONAL:
gc_mark(objspace, obj->as.rational.num);
ptr = obj->as.rational.den;
goto again;
case T_COMPLEX:
gc_mark(objspace, obj->as.complex.real);
ptr = obj->as.complex.imag;
goto again;
case T_STRUCT:
{
long len = RSTRUCT_LEN(obj);
const VALUE *ptr = RSTRUCT_CONST_PTR(obj);
while (len--) {
gc_mark(objspace, *ptr++);
}
}
break;
default:
#if GC_DEBUG
rb_gcdebug_print_obj_condition((VALUE)obj);
#endif
if (BUILTIN_TYPE(obj) == T_NONE) rb_bug("rb_gc_mark(): %p is T_NONE", (void *)obj);
if (BUILTIN_TYPE(obj) == T_ZOMBIE) rb_bug("rb_gc_mark(): %p is T_ZOMBIE", (void *)obj);
rb_bug("rb_gc_mark(): unknown data type 0x%x(%p) %s",
BUILTIN_TYPE(obj), (void *)obj,
is_pointer_to_heap(objspace, obj) ? "corrupted object" : "non object");
}
}
static void
gc_mark_stacked_objects(rb_objspace_t *objspace)
{
mark_stack_t *mstack = &objspace->mark_stack;
VALUE obj = 0;
if (!mstack->index) return;
while (pop_mark_stack(mstack, &obj)) {
if (RGENGC_CHECK_MODE > 0 && !gc_marked(objspace, obj)) {
rb_bug("gc_mark_stacked_objects: %p (%s) is infant, but not marked.", (void *)obj, obj_type_name(obj));
}
gc_mark_children(objspace, obj);
}
shrink_stack_chunk_cache(mstack);
}
#ifndef RGENGC_PRINT_TICK
#define RGENGC_PRINT_TICK 0
#endif
/* the following code is only for internal tuning. */
/* Source code to use RDTSC is quoted and modified from
* http://www.mcs.anl.gov/~kazutomo/rdtsc.html
* written by Kazutomo Yoshii <kazutomo@mcs.anl.gov>
*/
#if RGENGC_PRINT_TICK
#if defined(__GNUC__) && defined(__i386__)
typedef unsigned long long tick_t;
static inline tick_t
tick(void)
{
unsigned long long int x;
__asm__ __volatile__ ("rdtsc" : "=A" (x));
return x;
}
#elif defined(__GNUC__) && defined(__x86_64__)
typedef unsigned long long tick_t;
static __inline__ tick_t
tick(void)
{
unsigned long hi, lo;
__asm__ __volatile__ ("rdtsc" : "=a"(lo), "=d"(hi));
return ((unsigned long long)lo)|( ((unsigned long long)hi)<<32);
}
#elif defined(_WIN32) && defined(_MSC_VER)
#include <intrin.h>
typedef unsigned __int64 tick_t;
static inline tick_t
tick(void)
{
return __rdtsc();
}
#else /* use clock */
typedef clock_t tick_t;
static inline tick_t
tick(void)
{
return clock();
}
#endif
#define MAX_TICKS 0x100
static tick_t mark_ticks[MAX_TICKS];
static const char *mark_ticks_categories[MAX_TICKS];
static void
show_mark_ticks(void)
{
int i;
fprintf(stderr, "mark ticks result:\n");
for (i=0; i<MAX_TICKS; i++) {
const char *category = mark_ticks_categories[i];
if (category) {
fprintf(stderr, "%s\t%8lu\n", category, (unsigned long)mark_ticks[i]);
}
else {
break;
}
}
}
#endif /* RGENGC_PRINT_TICK */
static void
gc_mark_roots(rb_objspace_t *objspace, int full_mark, const char **categoryp)
{
struct gc_list *list;
rb_thread_t *th = GET_THREAD();
if (categoryp) *categoryp = "xxx";
#if RGENGC_PRINT_TICK
tick_t start_tick = tick();
int tick_count = 0;
const char *prev_category = 0;
if (mark_ticks_categories[0] == 0) {
atexit(show_mark_ticks);
}
#endif
#if RGENGC_PRINT_TICK
#define MARK_CHECKPOINT_PRINT_TICK(category) do { \
if (prev_category) { \
tick_t t = tick(); \
mark_ticks[tick_count] = t - start_tick; \
mark_ticks_categories[tick_count] = prev_category; \
tick_count++; \
} \
prev_category = category; \
start_tick = tick(); \
} while (0)
#else /* RGENGC_PRINT_TICK */
#define MARK_CHECKPOINT_PRINT_TICK(category)
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif
#define MARK_CHECKPOINT(category) do { \
if (categoryp) *categoryp = category; \
MARK_CHECKPOINT_PRINT_TICK(category); \
} while (0)
MARK_CHECKPOINT("vm");
SET_STACK_END;
th->vm->self ? rb_gc_mark(th->vm->self) : rb_vm_mark(th->vm);
MARK_CHECKPOINT("finalizers");
mark_tbl(objspace, finalizer_table);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
MARK_CHECKPOINT("machine_context");
mark_current_machine_context(objspace, th);
MARK_CHECKPOINT("symbols");
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
#if USE_RGENGC
objspace->rgengc.parent_object_is_old = TRUE;
rb_gc_mark_symbols(full_mark);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
objspace->rgengc.parent_object_is_old = FALSE;
#else
rb_gc_mark_symbols(full_mark);
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
MARK_CHECKPOINT("encodings");
rb_gc_mark_encodings();
/* mark protected global variables */
MARK_CHECKPOINT("global_list");
for (list = global_List; list; list = list->next) {
rb_gc_mark_maybe(*list->varptr);
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
MARK_CHECKPOINT("end_proc");
rb_mark_end_proc();
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
MARK_CHECKPOINT("global_tbl");
rb_gc_mark_global_tbl();
/* mark generic instance variables for special constants */
MARK_CHECKPOINT("generic_ivars");
rb_mark_generic_ivar_tbl();
MARK_CHECKPOINT("parser");
rb_gc_mark_parser();
MARK_CHECKPOINT("live_method_entries");
rb_gc_mark_unlinked_live_method_entries(th->vm);
MARK_CHECKPOINT("finish");
#undef MARK_CHECKPOINT
}
static void
gc_marks_body(rb_objspace_t *objspace, int full_mark)
{
/* start marking */
rgengc_report(1, objspace, "gc_marks_body: start (%s)\n", full_mark ? "full" : "minor");
#if USE_RGENGC
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
objspace->rgengc.parent_object_is_old = FALSE;
objspace->rgengc.during_minor_gc = full_mark ? FALSE : TRUE;
if (objspace->rgengc.during_minor_gc) {
objspace->profile.minor_gc_count++;
rgengc_rememberset_mark(objspace, heap_eden);
}
else {
objspace->profile.major_gc_count++;
#if RGENGC_AGE2_PROMOTION
/* all old objects change to young objects */
objspace->rgengc.young_object_count += objspace->rgengc.old_object_count;
#endif
objspace->rgengc.remembered_shady_object_count = 0;
objspace->rgengc.old_object_count = 0;
rgengc_mark_and_rememberset_clear(objspace, heap_eden);
}
#endif
gc_mark_roots(objspace, full_mark, 0);
gc_mark_stacked_objects(objspace);
gc_event_hook(objspace, RUBY_INTERNAL_EVENT_GC_END_MARK, 0);
rgengc_report(1, objspace, "gc_marks_body: end (%s)\n", full_mark ? "full" : "minor");
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
#if RGENGC_CHECK_MODE >= 3
#define MAKE_ROOTSIG(obj) (((VALUE)(obj) << 1) | 0x01)
#define IS_ROOTSIG(obj) ((VALUE)(obj) & 0x01)
#define GET_ROOTSIG(obj) ((const char *)((VALUE)(obj) >> 1))
struct reflist {
VALUE *list;
int pos;
int size;
};
static struct reflist *
reflist_create(VALUE obj)
{
struct reflist *refs = xmalloc(sizeof(struct reflist));
refs->size = 1;
refs->list = ALLOC_N(VALUE, refs->size);
refs->list[0] = obj;
refs->pos = 1;
return refs;
}
static void
reflist_destruct(struct reflist *refs)
{
xfree(refs->list);
xfree(refs);
}
static void
reflist_add(struct reflist *refs, VALUE obj)
{
if (refs->pos == refs->size) {
refs->size *= 2;
SIZED_REALLOC_N(refs->list, VALUE, refs->size, refs->size/2);
}
refs->list[refs->pos++] = obj;
}
static void
reflist_dump(struct reflist *refs)
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
{
int i;
for (i=0; i<refs->pos; i++) {
VALUE obj = refs->list[i];
if (IS_ROOTSIG(obj)) { /* root */
fprintf(stderr, "<root@%s>", GET_ROOTSIG(obj));
}
else {
fprintf(stderr, "<%p@%s>", (void *)obj, obj_type_name(obj));
}
if (i+1 < refs->pos) fprintf(stderr, ", ");
}
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if RGENGC_CHECK_MODE >= 3
static int
reflist_refered_from_machine_context(struct reflist *refs)
{
int i;
for (i=0; i<refs->pos; i++) {
VALUE obj = refs->list[i];
if (IS_ROOTSIG(obj) && strcmp(GET_ROOTSIG(obj), "machine_context") == 0) return 1;
}
return 0;
}
#endif
struct allrefs {
rb_objspace_t *objspace;
/* a -> obj1
* b -> obj1
* c -> obj1
* c -> obj2
* d -> obj3
* #=> {obj1 => [a, b, c], obj2 => [c, d]}
*/
struct st_table *references;
const char *category;
VALUE root_obj;
};
static int
allrefs_add(struct allrefs *data, VALUE obj)
{
struct reflist *refs;
if (st_lookup(data->references, obj, (st_data_t *)&refs)) {
reflist_add(refs, data->root_obj);
return 0;
}
else {
refs = reflist_create(data->root_obj);
st_insert(data->references, obj, (st_data_t)refs);
return 1;
}
}
static void
allrefs_i(VALUE obj, void *ptr)
{
struct allrefs *data = (struct allrefs *)ptr;
if (allrefs_add(data, obj)) {
push_mark_stack(&data->objspace->mark_stack, obj);
}
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
static void
allrefs_roots_i(VALUE obj, void *ptr)
{
struct allrefs *data = (struct allrefs *)ptr;
if (strlen(data->category) == 0) rb_bug("!!!");
data->root_obj = MAKE_ROOTSIG(data->category);
if (allrefs_add(data, obj)) {
push_mark_stack(&data->objspace->mark_stack, obj);
}
}
static st_table *
objspace_allrefs(rb_objspace_t *objspace)
{
struct allrefs data;
struct mark_func_data_struct mfd;
VALUE obj;
data.objspace = objspace;
data.references = st_init_numtable();
mfd.mark_func = allrefs_roots_i;
mfd.data = &data;
/* traverse root objects */
objspace->mark_func_data = &mfd;
gc_mark_roots(objspace, TRUE, &data.category);
objspace->mark_func_data = 0;
/* traverse rest objects reachable from root objects */
while (pop_mark_stack(&objspace->mark_stack, &obj)) {
rb_objspace_reachable_objects_from(data.root_obj = obj, allrefs_i, &data);
}
shrink_stack_chunk_cache(&objspace->mark_stack);
return data.references;
}
static int
objspaec_allrefs_destruct_i(st_data_t key, st_data_t value, void *ptr)
{
struct reflist *refs = (struct reflist *)value;
reflist_destruct(refs);
return ST_CONTINUE;
}
static void
objspace_allrefs_destruct(struct st_table *refs)
{
st_foreach(refs, objspaec_allrefs_destruct_i, 0);
st_free_table(refs);
}
#if RGENGC_CHECK_MODE >= 4
static int
allrefs_dump_i(st_data_t k, st_data_t v, st_data_t ptr)
{
VALUE obj = (VALUE)k;
struct reflist *refs = (struct reflist *)v;
fprintf(stderr, "[allrefs_dump_i] %p (%s%s%s%s) <- ",
(void *)obj, obj_type_name(obj),
RVALUE_OLD_P(obj) ? "[O]" : "[Y]",
RVALUE_WB_PROTECTED(obj) ? "[W]" : "",
MARKED_IN_BITMAP(GET_HEAP_REMEMBERSET_BITS(obj), obj) ? "[R]" : "");
reflist_dump(refs);
fprintf(stderr, "\n");
return ST_CONTINUE;
}
static void
allrefs_dump(rb_objspace_t *objspace)
{
fprintf(stderr, "[all refs] (size: %d)\n", (int)objspace->rgengc.allrefs_table->num_entries);
st_foreach(objspace->rgengc.allrefs_table, allrefs_dump_i, 0);
}
#endif
#if RGENGC_CHECK_MODE >= 3
static int
gc_check_after_marks_i(st_data_t k, st_data_t v, void *ptr)
{
VALUE obj = k;
struct reflist *refs = (struct reflist *)v;
rb_objspace_t *objspace = (rb_objspace_t *)ptr;
/* object should be marked or oldgen */
if (!MARKED_IN_BITMAP(GET_HEAP_MARK_BITS(obj), obj)) {
fprintf(stderr, "gc_check_after_marks_i: %p (%s) is not marked and not oldgen.\n", (void *)obj, obj_type_name(obj));
fprintf(stderr, "gc_check_after_marks_i: %p is referred from ", (void *)obj);
reflist_dump(refs);
if (reflist_refered_from_machine_context(refs)) {
fprintf(stderr, " (marked from machine stack).\n");
/* marked from machine context can be false positive */
}
else {
objspace->rgengc.error_count++;
fprintf(stderr, "\n");
}
}
return ST_CONTINUE;
}
#endif
static void
gc_marks_check(rb_objspace_t *objspace, int (*checker_func)(ANYARGS), const char *checker_name)
{
size_t saved_malloc_increase = objspace->malloc_params.increase;
#if RGENGC_ESTIMATE_OLDMALLOC
size_t saved_oldmalloc_increase = objspace->rgengc.oldmalloc_increase;
#endif
VALUE already_disabled = rb_gc_disable();
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
objspace->rgengc.allrefs_table = objspace_allrefs(objspace);
if (checker_func) {
st_foreach(objspace->rgengc.allrefs_table, checker_func, (st_data_t)objspace);
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
if (objspace->rgengc.error_count > 0) {
#if RGENGC_CHECK_MODE >= 4
allrefs_dump(objspace);
#endif
if (checker_name) rb_bug("%s: GC has problem.", checker_name);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
objspace_allrefs_destruct(objspace->rgengc.allrefs_table);
objspace->rgengc.allrefs_table = 0;
if (already_disabled == Qfalse) rb_gc_enable();
objspace->malloc_params.increase = saved_malloc_increase;
#if RGENGC_ESTIMATE_OLDMALLOC
objspace->rgengc.oldmalloc_increase = saved_oldmalloc_increase;
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
#endif /* RGENGC_CHECK_MODE >= 2 */
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
struct verify_internal_consistency_struct {
rb_objspace_t *objspace;
int err_count;
VALUE parent;
};
static void
verify_internal_consistency_reachable_i(VALUE child, void *ptr)
{
struct verify_internal_consistency_struct *data = (struct verify_internal_consistency_struct *)ptr;
assert(RVALUE_OLD_P(data->parent));
if (!RVALUE_OLD_P(child)) {
if (!MARKED_IN_BITMAP(GET_HEAP_PAGE(data->parent)->rememberset_bits, data->parent) &&
!MARKED_IN_BITMAP(GET_HEAP_PAGE(child)->rememberset_bits, child)) {
fprintf(stderr, "verify_internal_consistency_reachable_i: WB miss %p (%s) -> %p (%s)\n",
(void *)data->parent, obj_type_name(data->parent),
(void *)child, obj_type_name(child));
data->err_count++;
}
}
}
static int
verify_internal_consistency_i(void *page_start, void *page_end, size_t stride, void *ptr)
{
struct verify_internal_consistency_struct *data = (struct verify_internal_consistency_struct *)ptr;
VALUE v;
for (v = (VALUE)page_start; v != (VALUE)page_end; v += stride) {
if (is_live_object(data->objspace, v)) {
if (RVALUE_OLD_P(v)) {
data->parent = v;
/* reachable objects from an oldgen object should be old or (young with remember) */
rb_objspace_reachable_objects_from(v, verify_internal_consistency_reachable_i, (void *)data);
}
}
}
return 0;
}
#endif /* USE_RGENGC */
/*
* call-seq:
* GC.verify_internal_consistency -> nil
*
* Verify internal consistency.
*
* This method is implementation specific.
* Now this method checks generational consistency
* if RGenGC is supported.
*/
static VALUE
gc_verify_internal_consistency(VALUE self)
{
#if USE_RGENGC
struct verify_internal_consistency_struct data;
data.objspace = &rb_objspace;
data.err_count = 0;
{
struct each_obj_args eo_args;
eo_args.callback = verify_internal_consistency_i;
eo_args.data = (void *)&data;
objspace_each_objects((VALUE)&eo_args);
}
if (data.err_count != 0) {
#if RGENGC_CHECK_MODE >= 4
rb_objspace_t *objspace = &rb_objspace;
objspace->rgengc.error_count = data.err_count;
gc_marks_check(objspace, NULL, NULL);
allrefs_dump(objspace);
#endif
rb_bug("gc_verify_internal_consistency: found internal consistency.");
}
#endif
return Qnil;
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
static void
gc_marks(rb_objspace_t *objspace, int full_mark)
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
{
struct mark_func_data_struct *prev_mark_func_data;
gc_prof_mark_timer_start(objspace);
{
/* setup marking */
prev_mark_func_data = objspace->mark_func_data;
objspace->mark_func_data = 0;
#if USE_RGENGC
#if RGENGC_CHECK_MODE >= 2
gc_verify_internal_consistency(Qnil);
#endif
if (full_mark == TRUE) { /* major/full GC */
gc_marks_body(objspace, TRUE);
{
/* See the comment about RUBY_GC_HEAP_OLDOBJECT_LIMIT_FACTOR */
const double r = gc_params.oldobject_limit_factor;
objspace->rgengc.remembered_shady_object_limit = (size_t)(objspace->rgengc.remembered_shady_object_count * r);
objspace->rgengc.old_object_limit = (size_t)(objspace->rgengc.old_object_count * r);
}
}
else { /* minor GC */
gc_marks_body(objspace, FALSE);
}
#if RGENGC_PROFILE > 0
if (gc_prof_record(objspace)) {
gc_profile_record *record = gc_prof_record(objspace);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
record->old_objects = objspace->rgengc.old_object_count;
}
#endif
#if RGENGC_CHECK_MODE >= 3
gc_marks_check(objspace, gc_check_after_marks_i, "after_marks");
#endif
#else /* USE_RGENGC */
gc_marks_body(objspace, TRUE);
#endif
objspace->mark_func_data = prev_mark_func_data;
}
gc_prof_mark_timer_stop(objspace);
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
/* RGENGC */
static void
rgengc_report_body(int level, rb_objspace_t *objspace, const char *fmt, ...)
{
if (level <= RGENGC_DEBUG) {
char buf[1024];
FILE *out = stderr;
va_list args;
const char *status = " ";
#if USE_RGENGC
if (during_gc) {
status = objspace->rgengc.during_minor_gc ? "-" : "+";
}
#endif
va_start(args, fmt);
vsnprintf(buf, 1024, fmt, args);
va_end(args);
fprintf(out, "%s|", status);
fputs(buf, out);
}
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
/* bit operations */
static int
rgengc_remembersetbits_get(rb_objspace_t *objspace, VALUE obj)
{
bits_t *bits = GET_HEAP_REMEMBERSET_BITS(obj);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
return MARKED_IN_BITMAP(bits, obj) ? 1 : 0;
}
static int
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
rgengc_remembersetbits_set(rb_objspace_t *objspace, VALUE obj)
{
bits_t *bits = GET_HEAP_REMEMBERSET_BITS(obj);
if (MARKED_IN_BITMAP(bits, obj)) {
return FALSE;
}
else {
MARK_IN_BITMAP(bits, obj);
return TRUE;
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
/* wb, etc */
/* return FALSE if already remembered */
static int
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
rgengc_remember(rb_objspace_t *objspace, VALUE obj)
{
rgengc_report(2, objspace, "rgengc_remember: %p (%s, %s) %s\n", (void *)obj, obj_type_name(obj),
RVALUE_WB_PROTECTED(obj) ? "WB-protected" : "non-WB-protected",
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
rgengc_remembersetbits_get(objspace, obj) ? "was already remembered" : "is remembered now");
#if RGENGC_CHECK_MODE > 0
{
switch (BUILTIN_TYPE(obj)) {
case T_NONE:
case T_ZOMBIE:
rb_bug("rgengc_remember: should not remember %p (%s)",
(void *)obj, obj_type_name(obj));
default:
;
}
}
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
if (RGENGC_PROFILE) {
if (!rgengc_remembered(objspace, obj)) {
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
#if RGENGC_PROFILE > 0
if (RVALUE_WB_PROTECTED(obj)) {
objspace->profile.remembered_normal_object_count++;
#if RGENGC_PROFILE >= 2
objspace->profile.remembered_normal_object_count_types[BUILTIN_TYPE(obj)]++;
#endif
}
else {
objspace->profile.remembered_shady_object_count++;
#if RGENGC_PROFILE >= 2
objspace->profile.remembered_shady_object_count_types[BUILTIN_TYPE(obj)]++;
#endif
}
#endif /* RGENGC_PROFILE > 0 */
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
}
return rgengc_remembersetbits_set(objspace, obj);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
static int
rgengc_remembered(rb_objspace_t *objspace, VALUE obj)
{
int result = rgengc_remembersetbits_get(objspace, obj);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
check_gen_consistency(obj);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
rgengc_report(6, objspace, "gc_remembered: %p (%s) => %d\n", (void *)obj, obj_type_name(obj), result);
return result;
}
static void
rgengc_rememberset_mark(rb_objspace_t *objspace, rb_heap_t *heap)
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
{
size_t j;
RVALUE *p, *offset;
bits_t *bits, bitset;
struct heap_page *page = heap->pages;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if RGENGC_PROFILE > 0
size_t shady_object_count = 0, clear_count = 0;
#endif
while (page) {
p = page->start;
bits = page->rememberset_bits;
offset = p - NUM_IN_PAGE(p);
for (j=0; j < HEAP_BITMAP_LIMIT; j++) {
if (bits[j]) {
p = offset + j * BITS_BITLENGTH;
bitset = bits[j];
do {
if (bitset & 1) {
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
/* mark before RVALUE_PROMOTE_... */
gc_mark_ptr(objspace, (VALUE)p);
if (RVALUE_WB_PROTECTED(p)) {
rgengc_report(2, objspace, "rgengc_rememberset_mark: clear %p (%s)\n", p, obj_type_name((VALUE)p));
#if RGENGC_AGE2_PROMOTION
if (RVALUE_INFANT_P((VALUE)p)) {
RVALUE_PROMOTE_INFANT(objspace, (VALUE)p);
RVALUE_PROMOTE_YOUNG(objspace, (VALUE)p);
}
else if (RVALUE_YOUNG_P((VALUE)p)) {
RVALUE_PROMOTE_YOUNG(objspace, (VALUE)p);
}
#endif
CLEAR_IN_BITMAP(bits, p);
#if RGENGC_PROFILE > 0
clear_count++;
#endif
}
else {
#if RGENGC_PROFILE > 0
shady_object_count++;
#endif
}
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
rgengc_report(2, objspace, "rgengc_rememberset_mark: mark %p (%s)\n", p, obj_type_name((VALUE)p));
gc_mark_children(objspace, (VALUE) p);
}
p++;
bitset >>= 1;
} while (bitset);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
}
page = page->next;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
rgengc_report(2, objspace, "rgengc_rememberset_mark: finished\n");
#if RGENGC_PROFILE > 0
rgengc_report(2, objspace, "rgengc_rememberset_mark: clear_count: %"PRIdSIZE", shady_object_count: %"PRIdSIZE"\n", clear_count, shady_object_count);
if (gc_prof_record(objspace)) {
gc_profile_record *record = gc_prof_record(objspace);
record->remembered_normal_objects = clear_count;
record->remembered_shady_objects = shady_object_count;
}
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
static void
rgengc_mark_and_rememberset_clear(rb_objspace_t *objspace, rb_heap_t *heap)
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
{
struct heap_page *page = heap->pages;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
while (page) {
memset(&page->mark_bits[0], 0, HEAP_BITMAP_SIZE);
memset(&page->rememberset_bits[0], 0, HEAP_BITMAP_SIZE);
page = page->next;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
}
/* RGENGC: APIs */
void
rb_gc_writebarrier(VALUE a, VALUE b)
{
if (RGENGC_CHECK_MODE) {
if (!RVALUE_PROMOTED_P(a)) rb_bug("rb_gc_writebarrier: referer object %p (%s) is not promoted.", (void *)a, obj_type_name(a));
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
}
if (!RVALUE_OLD_P(b) && RVALUE_OLD_BITMAP_P(a)) {
rb_objspace_t *objspace = &rb_objspace;
if (!rgengc_remembered(objspace, a)) {
rgengc_report(2, objspace, "rb_gc_wb: %p (%s) -> %p (%s)\n",
(void *)a, obj_type_name(a), (void *)b, obj_type_name(b));
rgengc_remember(objspace, a);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
}
void
rb_gc_writebarrier_unprotect_promoted(VALUE obj)
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
{
rb_objspace_t *objspace = &rb_objspace;
if (RGENGC_CHECK_MODE) {
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (!RVALUE_PROMOTED_P(obj)) rb_bug("rb_gc_writebarrier_unprotect_promoted: called on non-promoted object");
if (!RVALUE_WB_PROTECTED(obj)) rb_bug("rb_gc_writebarrier_unprotect_promoted: called on shady object");
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
}
rgengc_report(1, objspace, "rb_gc_writebarrier_unprotect_promoted: %p (%s)%s\n", (void *)obj, obj_type_name(obj),
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
rgengc_remembered(objspace, obj) ? " (already remembered)" : "");
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (RVALUE_OLD_P(obj)) {
RVALUE_DEMOTE_FROM_OLD(obj);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
rgengc_remember(objspace, obj);
objspace->rgengc.remembered_shady_object_count++;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if RGENGC_PROFILE
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
objspace->profile.shade_operation_count++;
#if RGENGC_PROFILE >= 2
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
objspace->profile.shade_operation_count_types[BUILTIN_TYPE(obj)]++;
#endif /* RGENGC_PROFILE >= 2 */
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
#endif /* RGENGC_PROFILE */
}
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
else {
RVALUE_DEMOTE_FROM_YOUNG(obj);
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif
}
void
rb_gc_writebarrier_remember_promoted(VALUE obj)
{
rb_objspace_t *objspace = &rb_objspace;
rgengc_remember(objspace, obj);
}
static st_table *rgengc_unprotect_logging_table;
static int
rgengc_unprotect_logging_exit_func_i(st_data_t key, st_data_t val, st_data_t arg)
{
fprintf(stderr, "%s\t%d\n", (char *)key, (int)val);
return ST_CONTINUE;
}
static void
rgengc_unprotect_logging_exit_func(void)
{
st_foreach(rgengc_unprotect_logging_table, rgengc_unprotect_logging_exit_func_i, 0);
}
void
rb_gc_unprotect_logging(void *objptr, const char *filename, int line)
{
VALUE obj = (VALUE)objptr;
if (rgengc_unprotect_logging_table == 0) {
rgengc_unprotect_logging_table = st_init_strtable();
atexit(rgengc_unprotect_logging_exit_func);
}
if (OBJ_WB_PROTECTED(obj)) {
char buff[0x100];
st_data_t cnt = 1;
char *ptr = buff;
snprintf(ptr, 0x100 - 1, "%s|%s:%d", obj_type_name(obj), filename, line);
if (st_lookup(rgengc_unprotect_logging_table, (st_data_t)ptr, &cnt)) {
cnt++;
}
else {
ptr = (char *)malloc(strlen(buff) + 1);
strcpy(ptr, buff);
}
st_insert(rgengc_unprotect_logging_table, (st_data_t)ptr, cnt);
}
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif /* USE_RGENGC */
/* RGENGC analysis information */
VALUE
rb_obj_rgengc_writebarrier_protected_p(VALUE obj)
{
return OBJ_WB_PROTECTED(obj) ? Qtrue : Qfalse;
}
VALUE
rb_obj_rgengc_promoted_p(VALUE obj)
{
return OBJ_PROMOTED(obj) ? Qtrue : Qfalse;
}
size_t
rb_obj_gc_flags(VALUE obj, ID* flags, size_t max)
{
size_t n = 0;
static ID ID_marked;
#if USE_RGENGC
static ID ID_wb_protected, ID_old, ID_remembered;
#if RGENGC_AGE2_PROMOTION
static ID ID_young, ID_infant;
#endif
#endif
if (!ID_marked) {
#define I(s) ID_##s = rb_intern(#s);
I(marked);
#if USE_RGENGC
I(wb_protected);
I(old);
I(remembered);
#if RGENGC_AGE2_PROMOTION
I(young);
I(infant);
#endif
#endif
#undef I
}
#if USE_RGENGC
if (OBJ_WB_PROTECTED(obj) && n<max)
flags[n++] = ID_wb_protected;
if (RVALUE_OLD_P(obj) && n<max)
flags[n++] = ID_old;
#if RGENGC_AGE2_PROMOTION
if (RVALUE_YOUNG_P(obj) && n<max)
flags[n++] = ID_young;
if (RVALUE_INFANT_P(obj) && n<max)
flags[n++] = ID_infant;
#endif
if (MARKED_IN_BITMAP(GET_HEAP_REMEMBERSET_BITS(obj), obj) && n<max)
flags[n++] = ID_remembered;
#endif
if (MARKED_IN_BITMAP(GET_HEAP_MARK_BITS(obj), obj) && n<max)
flags[n++] = ID_marked;
return n;
}
/* GC */
void
rb_gc_force_recycle(VALUE p)
{
rb_objspace_t *objspace = &rb_objspace;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
CLEAR_IN_BITMAP(GET_HEAP_REMEMBERSET_BITS(p), p);
CLEAR_IN_BITMAP(GET_HEAP_OLDGEN_BITS(p), p);
if (!GET_HEAP_PAGE(p)->before_sweep) {
CLEAR_IN_BITMAP(GET_HEAP_MARK_BITS(p), p);
}
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif
objspace->profile.total_freed_object_num++;
heap_page_add_freeobj(objspace, GET_HEAP_PAGE(p), p);
/* Disable counting swept_slots because there are no meaning.
* if (!MARKED_IN_BITMAP(GET_HEAP_MARK_BITS(p), p)) {
* objspace->heap.swept_slots++;
* }
*/
}
void
rb_gc_register_mark_object(VALUE obj)
{
VALUE ary = GET_THREAD()->vm->mark_object_ary;
rb_ary_push(ary, obj);
}
void
rb_gc_register_address(VALUE *addr)
{
rb_objspace_t *objspace = &rb_objspace;
struct gc_list *tmp;
tmp = ALLOC(struct gc_list);
tmp->next = global_List;
tmp->varptr = addr;
global_List = tmp;
}
void
rb_gc_unregister_address(VALUE *addr)
{
rb_objspace_t *objspace = &rb_objspace;
struct gc_list *tmp = global_List;
if (tmp->varptr == addr) {
global_List = tmp->next;
xfree(tmp);
return;
}
while (tmp->next) {
if (tmp->next->varptr == addr) {
struct gc_list *t = tmp->next;
tmp->next = tmp->next->next;
xfree(t);
break;
}
tmp = tmp->next;
}
}
void
rb_global_variable(VALUE *var)
{
rb_gc_register_address(var);
}
#define GC_NOTIFY 0
enum {
gc_stress_no_major,
gc_stress_no_immediate_sweep,
gc_stress_full_mark_after_malloc,
gc_stress_max
};
#define gc_stress_full_mark_after_malloc_p() \
(FIXNUM_P(ruby_gc_stress) && (FIX2LONG(ruby_gc_stress) & (1<<gc_stress_full_mark_after_malloc)))
static int
garbage_collect_body(rb_objspace_t *objspace, int full_mark, int immediate_sweep, int reason)
{
if (ruby_gc_stress && !ruby_disable_gc_stress) {
int flag = FIXNUM_P(ruby_gc_stress) ? FIX2INT(ruby_gc_stress) : 0;
if (flag & (1<<gc_stress_no_major))
reason &= ~GPR_FLAG_MAJOR_MASK;
else
reason |= GPR_FLAG_MAJOR_BY_STRESS;
immediate_sweep = !(flag & (1<<gc_stress_no_immediate_sweep));
}
else {
if (!GC_ENABLE_LAZY_SWEEP || objspace->flags.dont_lazy_sweep) {
immediate_sweep = TRUE;
}
#if USE_RGENGC
if (full_mark) {
reason |= GPR_FLAG_MAJOR_BY_NOFREE;
}
if (objspace->rgengc.need_major_gc) {
reason |= objspace->rgengc.need_major_gc;
objspace->rgengc.need_major_gc = GPR_FLAG_NONE;
}
if (objspace->rgengc.remembered_shady_object_count > objspace->rgengc.remembered_shady_object_limit) {
reason |= GPR_FLAG_MAJOR_BY_SHADY;
}
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
if (objspace->rgengc.old_object_count > objspace->rgengc.old_object_limit) {
reason |= GPR_FLAG_MAJOR_BY_OLDGEN;
}
#endif
}
if (immediate_sweep) reason |= GPR_FLAG_IMMEDIATE_SWEEP;
full_mark = (reason & GPR_FLAG_MAJOR_MASK) ? TRUE : FALSE;
if (GC_NOTIFY) fprintf(stderr, "start garbage_collect(%d, %d, %d)\n", full_mark, immediate_sweep, reason);
objspace->profile.count++;
objspace->profile.latest_gc_info = reason;
gc_event_hook(objspace, RUBY_INTERNAL_EVENT_GC_START, 0 /* TODO: pass minor/immediate flag? */);
objspace->profile.total_allocated_object_num_at_gc_start = objspace->profile.total_allocated_object_num;
objspace->profile.heap_used_at_gc_start = heap_pages_used;
gc_prof_setup_new_record(objspace, reason);
gc_prof_timer_start(objspace);
{
if (during_gc == 0) {
rb_bug("during_gc should not be 0. RUBY_INTERNAL_EVENT_GC_START user should not cause GC in events.");
}
gc_marks(objspace, full_mark);
gc_sweep(objspace, immediate_sweep);
during_gc = 0;
}
gc_prof_timer_stop(objspace);
gc_heap_prepare_minimum_pages(objspace, heap_eden);
if (GC_NOTIFY) fprintf(stderr, "end garbage_collect()\n");
return TRUE;
}
static int
heap_ready_to_gc(rb_objspace_t *objspace, rb_heap_t *heap)
{
if (dont_gc || during_gc) {
if (!heap->freelist && !heap->free_pages) {
if (!heap_increment(objspace, heap)) {
heap_set_increment(objspace, 1);
heap_increment(objspace, heap);
}
}
return FALSE;
}
return TRUE;
}
static int
ready_to_gc(rb_objspace_t *objspace)
{
return heap_ready_to_gc(objspace, heap_eden);
}
static int
garbage_collect(rb_objspace_t *objspace, int full_mark, int immediate_sweep, int reason)
{
if (!heap_pages_used) {
during_gc = 0;
return FALSE;
}
if (!ready_to_gc(objspace)) {
during_gc = 0;
return TRUE;
}
#if GC_PROFILE_MORE_DETAIL
objspace->profile.prepare_time = getrusage_time();
#endif
gc_rest_sweep(objspace);
#if GC_PROFILE_MORE_DETAIL
objspace->profile.prepare_time = getrusage_time() - objspace->profile.prepare_time;
#endif
during_gc++;
return garbage_collect_body(objspace, full_mark, immediate_sweep, reason);
}
struct objspace_and_reason {
rb_objspace_t *objspace;
int reason;
int full_mark;
int immediate_sweep;
};
static void *
gc_with_gvl(void *ptr)
{
struct objspace_and_reason *oar = (struct objspace_and_reason *)ptr;
return (void *)(VALUE)garbage_collect(oar->objspace, oar->full_mark, oar->immediate_sweep, oar->reason);
}
static int
garbage_collect_with_gvl(rb_objspace_t *objspace, int full_mark, int immediate_sweep, int reason)
{
if (dont_gc) return TRUE;
if (ruby_thread_has_gvl_p()) {
return garbage_collect(objspace, full_mark, immediate_sweep, reason);
}
else {
if (ruby_native_thread_p()) {
struct objspace_and_reason oar;
oar.objspace = objspace;
oar.reason = reason;
oar.full_mark = full_mark;
oar.immediate_sweep = immediate_sweep;
return (int)(VALUE)rb_thread_call_with_gvl(gc_with_gvl, (void *)&oar);
}
else {
/* no ruby thread */
fprintf(stderr, "[FATAL] failed to allocate memory\n");
exit(EXIT_FAILURE);
}
}
}
int
rb_garbage_collect(void)
{
return garbage_collect(&rb_objspace, TRUE, TRUE, GPR_FLAG_CAPI);
}
#undef Init_stack
void
Init_stack(volatile VALUE *addr)
{
ruby_init_stack(addr);
}
/*
* call-seq:
* GC.start -> nil
* GC.garbage_collect -> nil
* GC.start(full_mark: true, immediate_sweep: true) -> nil
* GC.garbage_collect(full_mark: true, immediate_sweep: true) -> nil
*
* Initiates garbage collection, unless manually disabled.
*
* This method is defined with keyword arguments that default to true:
*
* def GC.start(full_mark: true, immediate_sweep: true); end
*
* Use full_mark: false to perform a minor GC.
* Use immediate_sweep: false to defer sweeping (use lazy sweep).
*
* Note: These keyword arguments are implementation and version dependent. They
* are not guaranteed to be future-compatible, and may be ignored if the
* underlying implementation does not support them.
*/
static VALUE
gc_start_internal(int argc, VALUE *argv, VALUE self)
{
rb_objspace_t *objspace = &rb_objspace;
int full_mark = TRUE, immediate_sweep = TRUE;
VALUE opt = Qnil;
static ID keyword_ids[2];
rb_scan_args(argc, argv, "0:", &opt);
if (!NIL_P(opt)) {
VALUE kwvals[2];
if (!keyword_ids[0]) {
keyword_ids[0] = rb_intern("full_mark");
keyword_ids[1] = rb_intern("immediate_sweep");
}
rb_get_kwargs(opt, keyword_ids, 0, 2, kwvals);
if (kwvals[0] != Qundef)
full_mark = RTEST(kwvals[0]);
if (kwvals[1] != Qundef)
immediate_sweep = RTEST(kwvals[1]);
}
garbage_collect(objspace, full_mark, immediate_sweep, GPR_FLAG_METHOD);
if (!finalizing) finalize_deferred(objspace);
return Qnil;
}
VALUE
rb_gc_start(void)
{
rb_gc();
return Qnil;
}
void
rb_gc(void)
{
rb_objspace_t *objspace = &rb_objspace;
garbage_collect(objspace, TRUE, TRUE, GPR_FLAG_CAPI);
if (!finalizing) finalize_deferred(objspace);
}
int
rb_during_gc(void)
{
rb_objspace_t *objspace = &rb_objspace;
return during_gc;
}
#if RGENGC_PROFILE >= 2
static void
gc_count_add_each_types(VALUE hash, const char *name, const size_t *types)
{
VALUE result = rb_hash_new();
int i;
for (i=0; i<T_MASK; i++) {
const char *type = type_name(i, 0);
rb_hash_aset(result, ID2SYM(rb_intern(type)), SIZET2NUM(types[i]));
}
rb_hash_aset(hash, ID2SYM(rb_intern(name)), result);
}
#endif
size_t
rb_gc_count(void)
{
return rb_objspace.profile.count;
}
/*
* call-seq:
* GC.count -> Integer
*
* The number of times GC occurred.
*
* It returns the number of times GC occurred since the process started.
*
*/
static VALUE
gc_count(VALUE self)
{
return SIZET2NUM(rb_gc_count());
}
static VALUE
gc_info_decode(int flags, VALUE hash_or_key)
{
static VALUE sym_major_by = Qnil, sym_gc_by, sym_immediate_sweep, sym_have_finalizer;
static VALUE sym_nofree, sym_oldgen, sym_shady, sym_rescan, sym_stress;
#if RGENGC_ESTIMATE_OLDMALLOC
static VALUE sym_oldmalloc;
#endif
static VALUE sym_newobj, sym_malloc, sym_method, sym_capi;
VALUE hash = Qnil, key = Qnil;
VALUE major_by;
if (SYMBOL_P(hash_or_key))
key = hash_or_key;
else if (RB_TYPE_P(hash_or_key, T_HASH))
hash = hash_or_key;
else
rb_raise(rb_eTypeError, "non-hash or symbol given");
if (sym_major_by == Qnil) {
#define S(s) sym_##s = ID2SYM(rb_intern_const(#s))
S(major_by);
S(gc_by);
S(immediate_sweep);
S(have_finalizer);
S(nofree);
S(oldgen);
S(shady);
S(rescan);
S(stress);
#if RGENGC_ESTIMATE_OLDMALLOC
S(oldmalloc);
#endif
S(newobj);
S(malloc);
S(method);
S(capi);
#undef S
}
#define SET(name, attr) \
if (key == sym_##name) \
return (attr); \
else if (hash != Qnil) \
rb_hash_aset(hash, sym_##name, (attr));
major_by =
(flags & GPR_FLAG_MAJOR_BY_OLDGEN) ? sym_oldgen :
(flags & GPR_FLAG_MAJOR_BY_SHADY) ? sym_shady :
(flags & GPR_FLAG_MAJOR_BY_RESCAN) ? sym_rescan :
(flags & GPR_FLAG_MAJOR_BY_STRESS) ? sym_stress :
#if RGENGC_ESTIMATE_OLDMALLOC
(flags & GPR_FLAG_MAJOR_BY_OLDMALLOC) ? sym_oldmalloc :
#endif
(flags & GPR_FLAG_MAJOR_BY_NOFREE) ? sym_nofree :
Qnil;
SET(major_by, major_by);
SET(gc_by,
(flags & GPR_FLAG_NEWOBJ) ? sym_newobj :
(flags & GPR_FLAG_MALLOC) ? sym_malloc :
(flags & GPR_FLAG_METHOD) ? sym_method :
(flags & GPR_FLAG_CAPI) ? sym_capi :
(flags & GPR_FLAG_STRESS) ? sym_stress :
Qnil
);
SET(have_finalizer, (flags & GPR_FLAG_HAVE_FINALIZE) ? Qtrue : Qfalse);
SET(immediate_sweep, (flags & GPR_FLAG_IMMEDIATE_SWEEP) ? Qtrue : Qfalse);
#undef SET
if (!NIL_P(key)) {/* matched key should return above */
rb_raise(rb_eArgError, "unknown key: %"PRIsVALUE, rb_sym2str(key));
}
return hash;
}
VALUE
rb_gc_latest_gc_info(VALUE key)
{
rb_objspace_t *objspace = &rb_objspace;
return gc_info_decode(objspace->profile.latest_gc_info, key);
}
/*
* call-seq:
* GC.latest_gc_info -> {:gc_by=>:newobj}
* GC.latest_gc_info(hash) -> hash
* GC.latest_gc_info(:major_by) -> :malloc
*
* Returns information about the most recent garbage collection.
*/
static VALUE
gc_latest_gc_info(int argc, VALUE *argv, VALUE self)
{
rb_objspace_t *objspace = &rb_objspace;
VALUE arg = Qnil;
if (rb_scan_args(argc, argv, "01", &arg) == 1) {
if (!SYMBOL_P(arg) && !RB_TYPE_P(arg, T_HASH)) {
rb_raise(rb_eTypeError, "non-hash or symbol given");
}
}
if (arg == Qnil)
arg = rb_hash_new();
return gc_info_decode(objspace->profile.latest_gc_info, arg);
}
static VALUE
gc_stat_internal(VALUE hash_or_sym, size_t *out)
{
static VALUE sym_count;
static VALUE sym_heap_used, sym_heap_length, sym_heap_increment;
static VALUE sym_heap_live_slot, sym_heap_free_slot, sym_heap_final_slot, sym_heap_swept_slot;
static VALUE sym_heap_eden_page_length, sym_heap_tomb_page_length;
static VALUE sym_total_allocated_object, sym_total_freed_object;
static VALUE sym_malloc_increase, sym_malloc_limit;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
static VALUE sym_minor_gc_count, sym_major_gc_count;
static VALUE sym_remembered_shady_object, sym_remembered_shady_object_limit;
static VALUE sym_old_object, sym_old_object_limit;
#if RGENGC_AGE2_PROMOTION
static VALUE sym_young_object;
#endif
#if RGENGC_ESTIMATE_OLDMALLOC
static VALUE sym_oldmalloc_increase, sym_oldmalloc_limit;
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if RGENGC_PROFILE
static VALUE sym_generated_normal_object_count, sym_generated_shady_object_count;
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
static VALUE sym_shade_operation_count, sym_promote_infant_count, sym_promote_young_count;
static VALUE sym_remembered_normal_object_count, sym_remembered_shady_object_count;
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#endif /* RGENGC_PROFILE */
#endif /* USE_RGENGC */
rb_objspace_t *objspace = &rb_objspace;
VALUE hash = Qnil, key = Qnil;
if (RB_TYPE_P(hash_or_sym, T_HASH))
hash = hash_or_sym;
else if (SYMBOL_P(hash_or_sym) && out)
key = hash_or_sym;
else
rb_raise(rb_eTypeError, "non-hash or symbol argument");
if (sym_count == 0) {
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#define S(s) sym_##s = ID2SYM(rb_intern_const(#s))
S(count);
S(heap_used);
S(heap_length);
S(heap_increment);
S(heap_live_slot);
S(heap_free_slot);
S(heap_final_slot);
S(heap_swept_slot);
S(heap_eden_page_length);
S(heap_tomb_page_length);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
S(total_allocated_object);
S(total_freed_object);
S(malloc_increase);
S(malloc_limit);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
S(minor_gc_count);
S(major_gc_count);
S(remembered_shady_object);
S(remembered_shady_object_limit);
S(old_object);
S(old_object_limit);
#if RGENGC_AGE2_PROMOTION
S(young_object);
#endif
#if RGENGC_ESTIMATE_OLDMALLOC
S(oldmalloc_increase);
S(oldmalloc_limit);
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if RGENGC_PROFILE
S(generated_normal_object_count);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
S(generated_shady_object_count);
S(shade_operation_count);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
S(promote_infant_count);
S(promote_young_count);
S(remembered_normal_object_count);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
S(remembered_shady_object_count);
#endif /* USE_RGENGC */
#endif /* RGENGC_PROFILE */
#undef S
}
#define SET(name, attr) \
if (key == sym_##name) \
return (*out = attr, Qnil); \
else if (hash != Qnil) \
rb_hash_aset(hash, sym_##name, SIZET2NUM(attr));
SET(count, objspace->profile.count);
/* implementation dependent counters */
SET(heap_used, heap_pages_used);
SET(heap_length, heap_pages_length);
SET(heap_increment, heap_pages_increment);
SET(heap_live_slot, objspace_live_slot(objspace));
SET(heap_free_slot, objspace_free_slot(objspace));
SET(heap_final_slot, heap_pages_final_slots);
SET(heap_swept_slot, heap_pages_swept_slots);
SET(heap_eden_page_length, heap_eden->page_length);
SET(heap_tomb_page_length, heap_tomb->page_length);
SET(total_allocated_object, objspace->profile.total_allocated_object_num);
SET(total_freed_object, objspace->profile.total_freed_object_num);
SET(malloc_increase, malloc_increase);
SET(malloc_limit, malloc_limit);
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if USE_RGENGC
SET(minor_gc_count, objspace->profile.minor_gc_count);
SET(major_gc_count, objspace->profile.major_gc_count);
SET(remembered_shady_object, objspace->rgengc.remembered_shady_object_count);
SET(remembered_shady_object_limit, objspace->rgengc.remembered_shady_object_limit);
SET(old_object, objspace->rgengc.old_object_count);
SET(old_object_limit, objspace->rgengc.old_object_limit);
#if RGENGC_AGE2_PROMOTION
SET(young_object, objspace->rgengc.young_object_count);
#endif
#if RGENGC_ESTIMATE_OLDMALLOC
SET(oldmalloc_increase, objspace->rgengc.oldmalloc_increase);
SET(oldmalloc_limit, objspace->rgengc.oldmalloc_increase_limit);
#endif
* gc.c: support RGENGC. [ruby-trunk - Feature #8339] See this ticet about RGENGC. * gc.c: Add several flags: * RGENGC_DEBUG: if >0, then prints debug information. * RGENGC_CHECK_MODE: if >0, add assertions. * RGENGC_PROFILE: if >0, add profiling features. check GC.stat and GC::Profiler. * include/ruby/ruby.h: disable RGENGC by default (USE_RGENGC == 0). * array.c: add write barriers for T_ARRAY and generate sunny objects. * include/ruby/ruby.h (RARRAY_PTR_USE): added. Use this macro if you want to access raw pointers. If you modify the contents which pointer pointed, then you need to care write barrier. * bignum.c, marshal.c, random.c: generate T_BIGNUM sunny objects. * complex.c, include/ruby/ruby.h: add write barriers for T_COMPLEX and generate sunny objects. * rational.c (nurat_s_new_internal), include/ruby/ruby.h: add write barriers for T_RATIONAL and generate sunny objects. * internal.h: add write barriers for RBasic::klass. * numeric.c (rb_float_new_in_heap): generate sunny T_FLOAT objects. * object.c (rb_class_allocate_instance), range.c: generate sunny T_OBJECT objects. * string.c: add write barriers for T_STRING and generate sunny objects. * variable.c: add write barriers for ivars. * vm_insnhelper.c (vm_setivar): ditto. * include/ruby/ruby.h, debug.c: use two flags FL_WB_PROTECTED and FL_OLDGEN. * node.h (NODE_FL_CREF_PUSHED_BY_EVAL, NODE_FL_CREF_OMOD_SHARED): move flag bits. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40703 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 22:07:47 +04:00
#if RGENGC_PROFILE
SET(generated_normal_object_count, objspace->profile.generated_normal_object_count);
SET(generated_shady_object_count, objspace->profile.generated_shady_object_count);
SET(shade_operation_count, objspace->profile.shade_operation_count);
SET(promote_infant_count, objspace->profile.promote_infant_count);
#if RGENGC_AGE2_PROMOTION
SET(promote_young_count, objspace->profile.promote_young_count);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
#endif
SET(remembered_normal_object_count, objspace->profile.remembered_normal_object_count);
SET(remembered_shady_object_count, objspace->profile.remembered_shady_object_count);
#endif /* RGENGC_PROFILE */
#endif /* USE_RGENGC */
#undef SET
if (!NIL_P(key)) { /* matched key should return above */
rb_raise(rb_eArgError, "unknown key: %"PRIsVALUE, rb_sym2str(key));
}
#if defined(RGENGC_PROFILE) && RGENGC_PROFILE >= 2
if (hash != Qnil) {
gc_count_add_each_types(hash, "generated_normal_object_count_types", objspace->profile.generated_normal_object_count_types);
gc_count_add_each_types(hash, "generated_shady_object_count_types", objspace->profile.generated_shady_object_count_types);
gc_count_add_each_types(hash, "shade_operation_count_types", objspace->profile.shade_operation_count_types);
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
gc_count_add_each_types(hash, "promote_infant_types", objspace->profile.promote_infant_types);
#if RGENGC_AGE2_PROMOTION
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
gc_count_add_each_types(hash, "promote_young_types", objspace->profile.promote_young_types);
#endif
gc_count_add_each_types(hash, "remembered_normal_object_count_types", objspace->profile.remembered_normal_object_count_types);
gc_count_add_each_types(hash, "remembered_shady_object_count_types", objspace->profile.remembered_shady_object_count_types);
}
#endif
return hash;
}
/*
* call-seq:
* GC.stat -> Hash
* GC.stat(hash) -> hash
* GC.stat(:key) -> Numeric
*
* Returns a Hash containing information about the GC.
*
* The hash includes information about internal statistics about GC such as:
*
* {
* :count=>2,
* :heap_used=>9,
* :heap_length=>11,
* :heap_increment=>2,
* :heap_live_slot=>6836,
* :heap_free_slot=>519,
* :heap_final_slot=>0,
* :heap_swept_slot=>818,
* :total_allocated_object=>7674,
* :total_freed_object=>838,
* :malloc_increase=>181034,
* :malloc_limit=>16777216,
* :minor_gc_count=>2,
* :major_gc_count=>0,
* :remembered_shady_object=>55,
* :remembered_shady_object_limit=>0,
* :old_object=>2422,
* :old_object_limit=>0,
* :oldmalloc_increase=>277386,
* :oldmalloc_limit=>16777216
* }
*
* The contents of the hash are implementation specific and may be changed in
* the future.
*
* This method is only expected to work on C Ruby.
*
*/
static VALUE
gc_stat(int argc, VALUE *argv, VALUE self)
{
VALUE arg = Qnil;
if (rb_scan_args(argc, argv, "01", &arg) == 1) {
if (SYMBOL_P(arg)) {
size_t value = 0;
gc_stat_internal(arg, &value);
return SIZET2NUM(value);
}
else if (!RB_TYPE_P(arg, T_HASH)) {
rb_raise(rb_eTypeError, "non-hash or symbol given");
}
}
if (arg == Qnil) {
arg = rb_hash_new();
}
gc_stat_internal(arg, 0);
return arg;
}
size_t
rb_gc_stat(VALUE key)
{
if (SYMBOL_P(key)) {
size_t value = 0;
gc_stat_internal(key, &value);
return value;
}
else {
gc_stat_internal(key, 0);
return 0;
}
}
/*
* call-seq:
* GC.stress -> fixnum, true or false
*
* Returns current status of GC stress mode.
*/
static VALUE
gc_stress_get(VALUE self)
{
rb_objspace_t *objspace = &rb_objspace;
return ruby_gc_stress;
}
/*
* call-seq:
* GC.stress = flag -> flag
*
* Updates the GC stress mode.
*
* When stress mode is enabled, the GC is invoked at every GC opportunity:
* all memory and object allocations.
*
* Enabling stress mode will degrade performance, it is only for debugging.
*
* flag can be true, false, or a fixnum bit-ORed following flags.
* 0x01:: no major GC
* 0x02:: no immediate sweep
* 0x04:: full mark after malloc/calloc/realloc
*/
static VALUE
gc_stress_set(VALUE self, VALUE flag)
{
rb_objspace_t *objspace = &rb_objspace;
rb_secure(2);
ruby_gc_stress = FIXNUM_P(flag) ? flag : (RTEST(flag) ? Qtrue : Qfalse);
return flag;
}
/*
* call-seq:
* GC.enable -> true or false
*
* Enables garbage collection, returning +true+ if garbage
* collection was previously disabled.
*
* GC.disable #=> false
* GC.enable #=> true
* GC.enable #=> false
*
*/
VALUE
rb_gc_enable(void)
{
rb_objspace_t *objspace = &rb_objspace;
int old = dont_gc;
dont_gc = FALSE;
return old ? Qtrue : Qfalse;
}
/*
* call-seq:
* GC.disable -> true or false
*
* Disables garbage collection, returning +true+ if garbage
* collection was already disabled.
*
* GC.disable #=> false
* GC.disable #=> true
*
*/
VALUE
rb_gc_disable(void)
{
rb_objspace_t *objspace = &rb_objspace;
int old = dont_gc;
gc_rest_sweep(objspace);
dont_gc = TRUE;
return old ? Qtrue : Qfalse;
}
static int
get_envparam_size(const char *name, size_t *default_value, size_t lower_bound)
{
char *ptr = getenv(name);
ssize_t val;
if (ptr != NULL && *ptr) {
size_t unit = 0;
char *end;
#if SIZEOF_SIZE_T == SIZEOF_LONG_LONG
val = strtoll(ptr, &end, 0);
#else
val = strtol(ptr, &end, 0);
#endif
switch (*end) {
case 'k': case 'K':
unit = 1024;
++end;
break;
case 'm': case 'M':
unit = 1024*1024;
++end;
break;
case 'g': case 'G':
unit = 1024*1024*1024;
++end;
break;
}
while (*end && isspace((unsigned char)*end)) end++;
if (*end) {
if (RTEST(ruby_verbose)) fprintf(stderr, "invalid string for %s: %s\n", name, ptr);
return 0;
}
if (unit > 0) {
if (val < -(ssize_t)(SIZE_MAX / 2 / unit) || (ssize_t)(SIZE_MAX / 2 / unit) < val) {
if (RTEST(ruby_verbose)) fprintf(stderr, "%s=%s is ignored because it overflows\n", name, ptr);
return 0;
}
val *= unit;
}
if (val > 0 && (size_t)val > lower_bound) {
if (RTEST(ruby_verbose)) {
fprintf(stderr, "%s=%"PRIdSIZE" (default value: %"PRIdSIZE")\n", name, val, *default_value);
}
*default_value = (size_t)val;
return 1;
}
else {
if (RTEST(ruby_verbose)) {
fprintf(stderr, "%s=%"PRIdSIZE" (default value: %"PRIdSIZE") is ignored because it must be greater than %"PRIdSIZE".\n",
name, val, *default_value, lower_bound);
}
return 0;
}
}
return 0;
}
static int
get_envparam_double(const char *name, double *default_value, double lower_bound)
{
char *ptr = getenv(name);
double val;
if (ptr != NULL && *ptr) {
char *end;
val = strtod(ptr, &end);
if (!*ptr || *end) {
if (RTEST(ruby_verbose)) fprintf(stderr, "invalid string for %s: %s\n", name, ptr);
return 0;
}
if (val > lower_bound) {
if (RTEST(ruby_verbose)) fprintf(stderr, "%s=%f (default value: %f)\n", name, val, *default_value);
*default_value = val;
return 1;
}
else {
if (RTEST(ruby_verbose)) fprintf(stderr, "%s=%f (default value: %f) is ignored because it must be greater than %f.\n", name, val, *default_value, lower_bound);
}
}
return 0;
}
static void
gc_set_initial_pages(void)
{
size_t min_pages;
rb_objspace_t *objspace = &rb_objspace;
min_pages = gc_params.heap_init_slots / HEAP_OBJ_LIMIT;
if (min_pages > heap_eden->page_length) {
heap_add_pages(objspace, heap_eden, min_pages - heap_eden->page_length);
}
}
/*
* GC tuning environment variables
*
* * RUBY_GC_HEAP_INIT_SLOTS
* - Initial allocation slots.
* * RUBY_GC_HEAP_FREE_SLOTS
* - Prepare at least this amount of slots after GC.
* - Allocate slots if there are not enough slots.
* * RUBY_GC_HEAP_GROWTH_FACTOR (new from 2.1)
* - Allocate slots by this factor.
* - (next slots number) = (current slots number) * (this factor)
* * RUBY_GC_HEAP_GROWTH_MAX_SLOTS (new from 2.1)
* - Allocation rate is limited to this factor.
* * RUBY_GC_HEAP_OLDOBJECT_LIMIT_FACTOR (new from 2.1.1)
* - Do full GC when the number of old objects is more than R * N
* where R is this factor and
* N is the number of old objects just after last full GC.
*
* * obsolete
* * RUBY_FREE_MIN -> RUBY_GC_HEAP_FREE_SLOTS (from 2.1)
* * RUBY_HEAP_MIN_SLOTS -> RUBY_GC_HEAP_INIT_SLOTS (from 2.1)
*
* * RUBY_GC_MALLOC_LIMIT
* * RUBY_GC_MALLOC_LIMIT_MAX (new from 2.1)
* * RUBY_GC_MALLOC_LIMIT_GROWTH_FACTOR (new from 2.1)
*
* * RUBY_GC_OLDMALLOC_LIMIT (new from 2.1)
* * RUBY_GC_OLDMALLOC_LIMIT_MAX (new from 2.1)
* * RUBY_GC_OLDMALLOC_LIMIT_GROWTH_FACTOR (new from 2.1)
*/
void
ruby_gc_set_params(int safe_level)
{
if (safe_level > 0) return;
/* RUBY_GC_HEAP_FREE_SLOTS */
if (get_envparam_size("RUBY_GC_HEAP_FREE_SLOTS", &gc_params.heap_free_slots, 0)) {
/* ok */
}
else if (get_envparam_size("RUBY_FREE_MIN", &gc_params.heap_free_slots, 0)) {
rb_warn("RUBY_FREE_MIN is obsolete. Use RUBY_GC_HEAP_FREE_SLOTS instead.");
}
/* RUBY_GC_HEAP_INIT_SLOTS */
if (get_envparam_size("RUBY_GC_HEAP_INIT_SLOTS", &gc_params.heap_init_slots, 0)) {
gc_set_initial_pages();
}
else if (get_envparam_size("RUBY_HEAP_MIN_SLOTS", &gc_params.heap_init_slots, 0)) {
rb_warn("RUBY_HEAP_MIN_SLOTS is obsolete. Use RUBY_GC_HEAP_INIT_SLOTS instead.");
gc_set_initial_pages();
}
get_envparam_double("RUBY_GC_HEAP_GROWTH_FACTOR", &gc_params.growth_factor, 1.0);
get_envparam_size ("RUBY_GC_HEAP_GROWTH_MAX_SLOTS", &gc_params.growth_max_slots, 0);
get_envparam_double("RUBY_GC_HEAP_OLDOBJECT_LIMIT_FACTOR", &gc_params.oldobject_limit_factor, 0.0);
get_envparam_size ("RUBY_GC_MALLOC_LIMIT", &gc_params.malloc_limit_min, 0);
get_envparam_size ("RUBY_GC_MALLOC_LIMIT_MAX", &gc_params.malloc_limit_max, 0);
get_envparam_double("RUBY_GC_MALLOC_LIMIT_GROWTH_FACTOR", &gc_params.malloc_limit_growth_factor, 1.0);
#if RGENGC_ESTIMATE_OLDMALLOC
if (get_envparam_size("RUBY_GC_OLDMALLOC_LIMIT", &gc_params.oldmalloc_limit_min, 0)) {
rb_objspace_t *objspace = &rb_objspace;
objspace->rgengc.oldmalloc_increase_limit = gc_params.oldmalloc_limit_min;
}
get_envparam_size ("RUBY_GC_OLDMALLOC_LIMIT_MAX", &gc_params.oldmalloc_limit_max, 0);
get_envparam_double("RUBY_GC_OLDMALLOC_LIMIT_GROWTH_FACTOR", &gc_params.oldmalloc_limit_growth_factor, 1.0);
#endif
}
void
rb_objspace_reachable_objects_from(VALUE obj, void (func)(VALUE, void *), void *data)
{
rb_objspace_t *objspace = &rb_objspace;
if (is_markable_object(objspace, obj)) {
struct mark_func_data_struct mfd;
mfd.mark_func = func;
mfd.data = data;
objspace->mark_func_data = &mfd;
gc_mark_children(objspace, obj);
objspace->mark_func_data = 0;
}
}
struct root_objects_data {
const char *category;
void (*func)(const char *category, VALUE, void *);
void *data;
};
static void
root_objects_from(VALUE obj, void *ptr)
{
const struct root_objects_data *data = (struct root_objects_data *)ptr;
(*data->func)(data->category, obj, data->data);
}
void
rb_objspace_reachable_objects_from_root(void (func)(const char *category, VALUE, void *), void *passing_data)
{
rb_objspace_t *objspace = &rb_objspace;
struct root_objects_data data;
struct mark_func_data_struct mfd;
data.func = func;
data.data = passing_data;
mfd.mark_func = root_objects_from;
mfd.data = &data;
objspace->mark_func_data = &mfd;
{
gc_mark_roots(objspace, TRUE, &data.category);
}
objspace->mark_func_data = 0;
}
/*
------------------------ Extended allocator ------------------------
*/
static void objspace_xfree(rb_objspace_t *objspace, void *ptr, size_t size);
static void *
negative_size_allocation_error_with_gvl(void *ptr)
{
rb_raise(rb_eNoMemError, "%s", (const char *)ptr);
return 0; /* should not be reached */
}
static void
negative_size_allocation_error(const char *msg)
{
if (ruby_thread_has_gvl_p()) {
rb_raise(rb_eNoMemError, "%s", msg);
}
else {
if (ruby_native_thread_p()) {
rb_thread_call_with_gvl(negative_size_allocation_error_with_gvl, (void *)msg);
}
else {
fprintf(stderr, "[FATAL] %s\n", msg);
exit(EXIT_FAILURE);
}
}
}
static void *
ruby_memerror_body(void *dummy)
{
rb_memerror();
return 0;
}
static void
ruby_memerror(void)
{
if (ruby_thread_has_gvl_p()) {
rb_memerror();
}
else {
if (ruby_native_thread_p()) {
rb_thread_call_with_gvl(ruby_memerror_body, 0);
}
else {
/* no ruby thread */
fprintf(stderr, "[FATAL] failed to allocate memory\n");
exit(EXIT_FAILURE);
}
}
}
void
rb_memerror(void)
{
rb_thread_t *th = GET_THREAD();
if (!nomem_error ||
rb_thread_raised_p(th, RAISED_NOMEMORY)) {
fprintf(stderr, "[FATAL] failed to allocate memory\n");
exit(EXIT_FAILURE);
}
if (rb_thread_raised_p(th, RAISED_NOMEMORY)) {
rb_thread_raised_clear(th);
GET_THREAD()->errinfo = nomem_error;
JUMP_TAG(TAG_RAISE);
}
rb_thread_raised_set(th, RAISED_NOMEMORY);
rb_exc_raise(nomem_error);
}
static void *
aligned_malloc(size_t alignment, size_t size)
{
void *res;
#if defined __MINGW32__
res = __mingw_aligned_malloc(size, alignment);
#elif defined _WIN32 && !defined __CYGWIN__
void *_aligned_malloc(size_t, size_t);
res = _aligned_malloc(size, alignment);
#elif defined(HAVE_POSIX_MEMALIGN)
if (posix_memalign(&res, alignment, size) == 0) {
return res;
}
else {
return NULL;
}
#elif defined(HAVE_MEMALIGN)
res = memalign(alignment, size);
#else
char* aligned;
res = malloc(alignment + size + sizeof(void*));
aligned = (char*)res + alignment + sizeof(void*);
aligned -= ((VALUE)aligned & (alignment - 1));
((void**)aligned)[-1] = res;
res = (void*)aligned;
#endif
#if defined(_DEBUG) || GC_DEBUG
/* alignment must be a power of 2 */
assert(((alignment - 1) & alignment) == 0);
assert(alignment % sizeof(void*) == 0);
#endif
return res;
}
static void
aligned_free(void *ptr)
{
#if defined __MINGW32__
__mingw_aligned_free(ptr);
#elif defined _WIN32 && !defined __CYGWIN__
_aligned_free(ptr);
#elif defined(HAVE_MEMALIGN) || defined(HAVE_POSIX_MEMALIGN)
free(ptr);
#else
free(((void**)ptr)[-1]);
#endif
}
static inline size_t
objspace_malloc_size(rb_objspace_t *objspace, void *ptr, size_t hint)
{
#ifdef HAVE_MALLOC_USABLE_SIZE
return malloc_usable_size(ptr);
#else
return hint;
#endif
}
enum memop_type {
MEMOP_TYPE_MALLOC = 1,
MEMOP_TYPE_FREE = 2,
MEMOP_TYPE_REALLOC = 3
};
static inline void
atomic_sub_nounderflow(size_t *var, size_t sub)
{
if (sub == 0) return;
while (1) {
size_t val = *var;
if (val < sub) sub = val;
if (ATOMIC_SIZE_CAS(*var, val, val-sub) == val) break;
}
}
static void
objspace_malloc_increase(rb_objspace_t *objspace, void *mem, size_t new_size, size_t old_size, enum memop_type type)
{
if (new_size > old_size) {
ATOMIC_SIZE_ADD(malloc_increase, new_size - old_size);
#if RGENGC_ESTIMATE_OLDMALLOC
ATOMIC_SIZE_ADD(objspace->rgengc.oldmalloc_increase, new_size - old_size);
#endif
}
else {
atomic_sub_nounderflow(&malloc_increase, old_size - new_size);
#if RGENGC_ESTIMATE_OLDMALLOC
atomic_sub_nounderflow(&objspace->rgengc.oldmalloc_increase, old_size - new_size);
#endif
}
if (type != MEMOP_TYPE_FREE &&
ruby_gc_stress && !ruby_disable_gc_stress &&
ruby_native_thread_p()) {
garbage_collect_with_gvl(objspace, gc_stress_full_mark_after_malloc_p(), TRUE, GPR_FLAG_MALLOC);
}
if (type == MEMOP_TYPE_MALLOC) {
retry:
if (malloc_increase > malloc_limit && ruby_native_thread_p()) {
if (ruby_thread_has_gvl_p() && is_lazy_sweeping(heap_eden)) {
gc_rest_sweep(objspace); /* rest_sweep can reduce malloc_increase */
goto retry;
}
garbage_collect_with_gvl(objspace, FALSE, TRUE, GPR_FLAG_MALLOC);
}
}
#if MALLOC_ALLOCATED_SIZE
if (new_size >= old_size) {
ATOMIC_SIZE_ADD(objspace->malloc_params.allocated_size, new_size - old_size);
}
else {
size_t dec_size = old_size - new_size;
size_t allocated_size = objspace->malloc_params.allocated_size;
#if MALLOC_ALLOCATED_SIZE_CHECK
if (allocated_size < dec_size) {
rb_bug("objspace_malloc_increase: underflow malloc_params.allocated_size.");
}
#endif
atomic_sub_nounderflow(&objspace->malloc_params.allocated_size, dec_size);
}
if (0) fprintf(stderr, "incraese - ptr: %p, type: %s, new_size: %d, old_size: %d\n",
mem,
type == MEMOP_TYPE_MALLOC ? "malloc" :
type == MEMOP_TYPE_FREE ? "free " :
type == MEMOP_TYPE_REALLOC ? "realloc": "error",
(int)new_size, (int)old_size);
switch (type) {
case MEMOP_TYPE_MALLOC:
ATOMIC_SIZE_INC(objspace->malloc_params.allocations);
break;
case MEMOP_TYPE_FREE:
{
size_t allocations = objspace->malloc_params.allocations;
if (allocations > 0) {
atomic_sub_nounderflow(&objspace->malloc_params.allocations, 1);
}
#if MALLOC_ALLOCATED_SIZE_CHECK
else {
assert(objspace->malloc_params.allocations > 0);
}
#endif
}
break;
case MEMOP_TYPE_REALLOC: /* ignore */ break;
}
#endif
}
static inline size_t
objspace_malloc_prepare(rb_objspace_t *objspace, size_t size)
{
if ((ssize_t)size < 0) {
negative_size_allocation_error("negative allocation size (or too big)");
}
if (size == 0) size = 1;
#if CALC_EXACT_MALLOC_SIZE
size += sizeof(size_t);
#endif
return size;
}
static inline void *
objspace_malloc_fixup(rb_objspace_t *objspace, void *mem, size_t size)
{
#if CALC_EXACT_MALLOC_SIZE
((size_t *)mem)[0] = size;
mem = (size_t *)mem + 1;
#endif
return mem;
}
#define TRY_WITH_GC(alloc) do { \
if (!(alloc) && \
(!garbage_collect_with_gvl(objspace, 1, 1, GPR_FLAG_MALLOC) || /* full mark && immediate sweep */ \
!(alloc))) { \
ruby_memerror(); \
} \
} while (0)
static void *
objspace_xmalloc(rb_objspace_t *objspace, size_t size)
{
void *mem;
size = objspace_malloc_prepare(objspace, size);
TRY_WITH_GC(mem = malloc(size));
size = objspace_malloc_size(objspace, mem, size);
objspace_malloc_increase(objspace, mem, size, 0, MEMOP_TYPE_MALLOC);
return objspace_malloc_fixup(objspace, mem, size);
}
static void *
objspace_xrealloc(rb_objspace_t *objspace, void *ptr, size_t new_size, size_t old_size)
{
void *mem;
if ((ssize_t)new_size < 0) {
negative_size_allocation_error("negative re-allocation size");
}
if (!ptr) return objspace_xmalloc(objspace, new_size);
/*
* The behavior of realloc(ptr, 0) is implementation defined.
* Therefore we don't use realloc(ptr, 0) for portability reason.
* see http://www.open-std.org/jtc1/sc22/wg14/www/docs/dr_400.htm
*/
if (new_size == 0) {
objspace_xfree(objspace, ptr, old_size);
return 0;
}
#if CALC_EXACT_MALLOC_SIZE
new_size += sizeof(size_t);
ptr = (size_t *)ptr - 1;
old_size = ((size_t *)ptr)[0];
#endif
old_size = objspace_malloc_size(objspace, ptr, old_size);
TRY_WITH_GC(mem = realloc(ptr, new_size));
new_size = objspace_malloc_size(objspace, mem, new_size);
#if CALC_EXACT_MALLOC_SIZE
((size_t *)mem)[0] = new_size;
mem = (size_t *)mem + 1;
#endif
objspace_malloc_increase(objspace, mem, new_size, old_size, MEMOP_TYPE_REALLOC);
return mem;
}
static void
objspace_xfree(rb_objspace_t *objspace, void *ptr, size_t old_size)
{
#if CALC_EXACT_MALLOC_SIZE
ptr = ((size_t *)ptr) - 1;
old_size = ((size_t*)ptr)[0];
#endif
old_size = objspace_malloc_size(objspace, ptr, old_size);
free(ptr);
objspace_malloc_increase(objspace, ptr, 0, old_size, MEMOP_TYPE_FREE);
}
void *
ruby_xmalloc(size_t size)
{
return objspace_xmalloc(&rb_objspace, size);
}
static inline size_t
xmalloc2_size(size_t n, size_t size)
{
size_t len = size * n;
if (n != 0 && size != len / n) {
rb_raise(rb_eArgError, "malloc: possible integer overflow");
}
return len;
}
void *
ruby_xmalloc2(size_t n, size_t size)
{
return objspace_xmalloc(&rb_objspace, xmalloc2_size(n, size));
}
static void *
objspace_xcalloc(rb_objspace_t *objspace, size_t count, size_t elsize)
{
void *mem;
size_t size;
size = xmalloc2_size(count, elsize);
size = objspace_malloc_prepare(objspace, size);
TRY_WITH_GC(mem = calloc(1, size));
size = objspace_malloc_size(objspace, mem, size);
objspace_malloc_increase(objspace, mem, size, 0, MEMOP_TYPE_MALLOC);
return objspace_malloc_fixup(objspace, mem, size);
}
void *
ruby_xcalloc(size_t n, size_t size)
{
return objspace_xcalloc(&rb_objspace, n, size);
}
#ifdef ruby_sized_xrealloc
#undef ruby_sized_xrealloc
#endif
void *
ruby_sized_xrealloc(void *ptr, size_t new_size, size_t old_size)
{
return objspace_xrealloc(&rb_objspace, ptr, new_size, old_size);
}
void *
ruby_xrealloc(void *ptr, size_t new_size)
{
return ruby_sized_xrealloc(ptr, new_size, 0);
}
#ifdef ruby_sized_xrealloc2
#undef ruby_sized_xrealloc2
#endif
void *
ruby_sized_xrealloc2(void *ptr, size_t n, size_t size, size_t old_n)
{
size_t len = size * n;
if (n != 0 && size != len / n) {
rb_raise(rb_eArgError, "realloc: possible integer overflow");
}
return objspace_xrealloc(&rb_objspace, ptr, len, old_n * size);
}
void *
ruby_xrealloc2(void *ptr, size_t n, size_t size)
{
return ruby_sized_xrealloc2(ptr, n, size, 0);
}
#ifdef ruby_sized_xfree
#undef ruby_sized_xfree
#endif
void
ruby_sized_xfree(void *x, size_t size)
{
if (x) {
objspace_xfree(&rb_objspace, x, size);
}
}
void
ruby_xfree(void *x)
{
ruby_sized_xfree(x, 0);
}
/* Mimic ruby_xmalloc, but need not rb_objspace.
* should return pointer suitable for ruby_xfree
*/
void *
ruby_mimmalloc(size_t size)
{
void *mem;
#if CALC_EXACT_MALLOC_SIZE
size += sizeof(size_t);
#endif
mem = malloc(size);
#if CALC_EXACT_MALLOC_SIZE
/* set 0 for consistency of allocated_size/allocations */
((size_t *)mem)[0] = 0;
mem = (size_t *)mem + 1;
#endif
return mem;
}
void
ruby_mimfree(void *ptr)
{
size_t *mem = (size_t *)ptr;
#if CALC_EXACT_MALLOC_SIZE
mem = mem - 1;
#endif
free(mem);
}
#if MALLOC_ALLOCATED_SIZE
/*
* call-seq:
* GC.malloc_allocated_size -> Integer
*
* Returns the size of memory allocated by malloc().
*
* Only available if ruby was built with +CALC_EXACT_MALLOC_SIZE+.
*/
static VALUE
gc_malloc_allocated_size(VALUE self)
{
return UINT2NUM(rb_objspace.malloc_params.allocated_size);
}
/*
* call-seq:
* GC.malloc_allocations -> Integer
*
* Returns the number of malloc() allocations.
*
* Only available if ruby was built with +CALC_EXACT_MALLOC_SIZE+.
*/
static VALUE
gc_malloc_allocations(VALUE self)
{
return UINT2NUM(rb_objspace.malloc_params.allocations);
}
#endif
/*
------------------------------ WeakMap ------------------------------
*/
struct weakmap {
st_table *obj2wmap; /* obj -> [ref,...] */
st_table *wmap2obj; /* ref -> obj */
VALUE final;
};
#define WMAP_DELETE_DEAD_OBJECT_IN_MARK 0
#if WMAP_DELETE_DEAD_OBJECT_IN_MARK
static int
wmap_mark_map(st_data_t key, st_data_t val, st_data_t arg)
{
rb_objspace_t *objspace = (rb_objspace_t *)arg;
VALUE obj = (VALUE)val;
if (!is_live_object(objspace, obj)) return ST_DELETE;
return ST_CONTINUE;
}
#endif
static void
wmap_mark(void *ptr)
{
struct weakmap *w = ptr;
#if WMAP_DELETE_DEAD_OBJECT_IN_MARK
if (w->obj2wmap) st_foreach(w->obj2wmap, wmap_mark_map, (st_data_t)&rb_objspace);
#endif
rb_gc_mark(w->final);
}
static int
wmap_free_map(st_data_t key, st_data_t val, st_data_t arg)
{
VALUE *ptr = (VALUE *)val;
ruby_sized_xfree(ptr, (ptr[0] + 1) * sizeof(VALUE));
return ST_CONTINUE;
}
static void
wmap_free(void *ptr)
{
struct weakmap *w = ptr;
st_foreach(w->obj2wmap, wmap_free_map, 0);
st_free_table(w->obj2wmap);
st_free_table(w->wmap2obj);
}
static int
wmap_memsize_map(st_data_t key, st_data_t val, st_data_t arg)
{
VALUE *ptr = (VALUE *)val;
*(size_t *)arg += (ptr[0] + 1) * sizeof(VALUE);
return ST_CONTINUE;
}
static size_t
wmap_memsize(const void *ptr)
{
size_t size;
const struct weakmap *w = ptr;
if (!w) return 0;
size = sizeof(*w);
size += st_memsize(w->obj2wmap);
size += st_memsize(w->wmap2obj);
st_foreach(w->obj2wmap, wmap_memsize_map, (st_data_t)&size);
return size;
}
static const rb_data_type_t weakmap_type = {
"weakmap",
{
wmap_mark,
wmap_free,
wmap_memsize,
},
NULL, NULL, RUBY_TYPED_FREE_IMMEDIATELY
};
static VALUE
wmap_allocate(VALUE klass)
{
struct weakmap *w;
VALUE obj = TypedData_Make_Struct(klass, struct weakmap, &weakmap_type, w);
w->obj2wmap = st_init_numtable();
w->wmap2obj = st_init_numtable();
w->final = rb_obj_method(obj, ID2SYM(rb_intern("finalize")));
return obj;
}
static int
wmap_final_func(st_data_t *key, st_data_t *value, st_data_t arg, int existing)
{
VALUE wmap, *ptr, size, i, j;
if (!existing) return ST_STOP;
wmap = (VALUE)arg, ptr = (VALUE *)*value;
for (i = j = 1, size = ptr[0]; i <= size; ++i) {
if (ptr[i] != wmap) {
ptr[j++] = ptr[i];
}
}
if (j == 1) {
ruby_sized_xfree(ptr, i * sizeof(VALUE));
return ST_DELETE;
}
if (j < i) {
ptr = ruby_sized_xrealloc2(ptr, j, sizeof(VALUE), i);
ptr[0] = j;
*value = (st_data_t)ptr;
}
return ST_CONTINUE;
}
static VALUE
wmap_finalize(VALUE self, VALUE objid)
{
st_data_t orig, wmap, data;
VALUE obj, *rids, i, size;
struct weakmap *w;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
/* Get reference from object id. */
obj = obj_id_to_ref(objid);
/* obj is original referenced object and/or weak reference. */
orig = (st_data_t)obj;
if (st_delete(w->obj2wmap, &orig, &data)) {
rids = (VALUE *)data;
size = *rids++;
for (i = 0; i < size; ++i) {
wmap = (st_data_t)rids[i];
st_delete(w->wmap2obj, &wmap, NULL);
}
ruby_sized_xfree((VALUE *)data, (size + 1) * sizeof(VALUE));
}
wmap = (st_data_t)obj;
if (st_delete(w->wmap2obj, &wmap, &orig)) {
wmap = (st_data_t)obj;
st_update(w->obj2wmap, orig, wmap_final_func, wmap);
}
return self;
}
struct wmap_iter_arg {
rb_objspace_t *objspace;
VALUE value;
};
static int
wmap_inspect_i(st_data_t key, st_data_t val, st_data_t arg)
{
VALUE str = (VALUE)arg;
VALUE k = (VALUE)key, v = (VALUE)val;
if (RSTRING_PTR(str)[0] == '#') {
rb_str_cat2(str, ", ");
}
else {
rb_str_cat2(str, ": ");
RSTRING_PTR(str)[0] = '#';
}
k = SPECIAL_CONST_P(k) ? rb_inspect(k) : rb_any_to_s(k);
rb_str_append(str, k);
rb_str_cat2(str, " => ");
v = SPECIAL_CONST_P(v) ? rb_inspect(v) : rb_any_to_s(v);
rb_str_append(str, v);
OBJ_INFECT(str, k);
OBJ_INFECT(str, v);
return ST_CONTINUE;
}
static VALUE
wmap_inspect(VALUE self)
{
VALUE str;
VALUE c = rb_class_name(CLASS_OF(self));
struct weakmap *w;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
str = rb_sprintf("-<%"PRIsVALUE":%p", c, (void *)self);
if (w->wmap2obj) {
st_foreach(w->wmap2obj, wmap_inspect_i, str);
}
RSTRING_PTR(str)[0] = '#';
rb_str_cat2(str, ">");
return str;
}
static int
wmap_each_i(st_data_t key, st_data_t val, st_data_t arg)
{
rb_objspace_t *objspace = (rb_objspace_t *)arg;
VALUE obj = (VALUE)val;
if (is_id_value(objspace, obj) && is_live_object(objspace, obj)) {
rb_yield_values(2, (VALUE)key, obj);
}
return ST_CONTINUE;
}
/* Iterates over keys and objects in a weakly referenced object */
static VALUE
wmap_each(VALUE self)
{
struct weakmap *w;
rb_objspace_t *objspace = &rb_objspace;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
st_foreach(w->wmap2obj, wmap_each_i, (st_data_t)objspace);
return self;
}
static int
wmap_each_key_i(st_data_t key, st_data_t val, st_data_t arg)
{
rb_objspace_t *objspace = (rb_objspace_t *)arg;
VALUE obj = (VALUE)val;
if (is_id_value(objspace, obj) && is_live_object(objspace, obj)) {
rb_yield((VALUE)key);
}
return ST_CONTINUE;
}
/* Iterates over keys and objects in a weakly referenced object */
static VALUE
wmap_each_key(VALUE self)
{
struct weakmap *w;
rb_objspace_t *objspace = &rb_objspace;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
st_foreach(w->wmap2obj, wmap_each_key_i, (st_data_t)objspace);
return self;
}
static int
wmap_each_value_i(st_data_t key, st_data_t val, st_data_t arg)
{
rb_objspace_t *objspace = (rb_objspace_t *)arg;
VALUE obj = (VALUE)val;
if (is_id_value(objspace, obj) && is_live_object(objspace, obj)) {
rb_yield(obj);
}
return ST_CONTINUE;
}
/* Iterates over keys and objects in a weakly referenced object */
static VALUE
wmap_each_value(VALUE self)
{
struct weakmap *w;
rb_objspace_t *objspace = &rb_objspace;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
st_foreach(w->wmap2obj, wmap_each_value_i, (st_data_t)objspace);
return self;
}
static int
wmap_keys_i(st_data_t key, st_data_t val, st_data_t arg)
{
struct wmap_iter_arg *argp = (struct wmap_iter_arg *)arg;
rb_objspace_t *objspace = argp->objspace;
VALUE ary = argp->value;
VALUE obj = (VALUE)val;
if (is_id_value(objspace, obj) && is_live_object(objspace, obj)) {
rb_ary_push(ary, (VALUE)key);
}
return ST_CONTINUE;
}
/* Iterates over keys and objects in a weakly referenced object */
static VALUE
wmap_keys(VALUE self)
{
struct weakmap *w;
struct wmap_iter_arg args;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
args.objspace = &rb_objspace;
args.value = rb_ary_new();
st_foreach(w->wmap2obj, wmap_keys_i, (st_data_t)&args);
return args.value;
}
static int
wmap_values_i(st_data_t key, st_data_t val, st_data_t arg)
{
struct wmap_iter_arg *argp = (struct wmap_iter_arg *)arg;
rb_objspace_t *objspace = argp->objspace;
VALUE ary = argp->value;
VALUE obj = (VALUE)val;
if (is_id_value(objspace, obj) && is_live_object(objspace, obj)) {
rb_ary_push(ary, obj);
}
return ST_CONTINUE;
}
/* Iterates over values and objects in a weakly referenced object */
static VALUE
wmap_values(VALUE self)
{
struct weakmap *w;
struct wmap_iter_arg args;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
args.objspace = &rb_objspace;
args.value = rb_ary_new();
st_foreach(w->wmap2obj, wmap_values_i, (st_data_t)&args);
return args.value;
}
static int
wmap_aset_update(st_data_t *key, st_data_t *val, st_data_t arg, int existing)
{
VALUE size, *ptr, *optr;
if (existing) {
size = (ptr = optr = (VALUE *)*val)[0];
++size;
ptr = ruby_sized_xrealloc2(ptr, size + 1, sizeof(VALUE), size);
}
else {
optr = 0;
size = 1;
ptr = ruby_xmalloc2(2, sizeof(VALUE));
}
ptr[0] = size;
ptr[size] = (VALUE)arg;
if (ptr == optr) return ST_STOP;
*val = (st_data_t)ptr;
return ST_CONTINUE;
}
/* Creates a weak reference from the given key to the given value */
static VALUE
wmap_aset(VALUE self, VALUE wmap, VALUE orig)
{
struct weakmap *w;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
should_be_finalizable(orig);
should_be_finalizable(wmap);
define_final0(orig, w->final);
define_final0(wmap, w->final);
st_update(w->obj2wmap, (st_data_t)orig, wmap_aset_update, wmap);
st_insert(w->wmap2obj, (st_data_t)wmap, (st_data_t)orig);
return nonspecial_obj_id(orig);
}
/* Retrieves a weakly referenced object with the given key */
static VALUE
wmap_aref(VALUE self, VALUE wmap)
{
st_data_t data;
VALUE obj;
struct weakmap *w;
rb_objspace_t *objspace = &rb_objspace;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
if (!st_lookup(w->wmap2obj, (st_data_t)wmap, &data)) return Qnil;
obj = (VALUE)data;
if (!is_id_value(objspace, obj)) return Qnil;
if (!is_live_object(objspace, obj)) return Qnil;
return obj;
}
/* Returns +true+ if +key+ is registered */
static VALUE
wmap_has_key(VALUE self, VALUE key)
{
return NIL_P(wmap_aref(self, key)) ? Qfalse : Qtrue;
}
static VALUE
wmap_size(VALUE self)
{
struct weakmap *w;
st_index_t n;
TypedData_Get_Struct(self, struct weakmap, &weakmap_type, w);
n = w->wmap2obj->num_entries;
#if SIZEOF_ST_INDEX_T <= SIZEOF_LONG
return ULONG2NUM(n);
#else
return ULL2NUM(n);
#endif
}
/*
------------------------------ GC profiler ------------------------------
*/
#define GC_PROFILE_RECORD_DEFAULT_SIZE 100
static double
getrusage_time(void)
{
#if defined(HAVE_CLOCK_GETTIME) && defined(CLOCK_PROCESS_CPUTIME_ID)
{
static int try_clock_gettime = 1;
struct timespec ts;
if (try_clock_gettime && clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &ts) == 0) {
return ts.tv_sec + ts.tv_nsec * 1e-9;
}
else {
try_clock_gettime = 0;
}
}
#endif
#ifdef RUSAGE_SELF
{
struct rusage usage;
struct timeval time;
if (getrusage(RUSAGE_SELF, &usage) == 0) {
time = usage.ru_utime;
return time.tv_sec + time.tv_usec * 1e-6;
}
}
#endif
#ifdef _WIN32
{
FILETIME creation_time, exit_time, kernel_time, user_time;
ULARGE_INTEGER ui;
LONG_LONG q;
double t;
if (GetProcessTimes(GetCurrentProcess(),
&creation_time, &exit_time, &kernel_time, &user_time) != 0) {
memcpy(&ui, &user_time, sizeof(FILETIME));
q = ui.QuadPart / 10L;
t = (DWORD)(q % 1000000L) * 1e-6;
q /= 1000000L;
#ifdef __GNUC__
t += q;
#else
t += (double)(DWORD)(q >> 16) * (1 << 16);
t += (DWORD)q & ~(~0 << 16);
#endif
return t;
}
}
#endif
return 0.0;
}
static inline void
gc_prof_setup_new_record(rb_objspace_t *objspace, int reason)
{
if (objspace->profile.run) {
size_t index = objspace->profile.next_index;
gc_profile_record *record;
/* create new record */
objspace->profile.next_index++;
if (!objspace->profile.records) {
objspace->profile.size = GC_PROFILE_RECORD_DEFAULT_SIZE;
objspace->profile.records = malloc(sizeof(gc_profile_record) * objspace->profile.size);
}
if (index >= objspace->profile.size) {
objspace->profile.size += 1000;
objspace->profile.records = realloc(objspace->profile.records, sizeof(gc_profile_record) * objspace->profile.size);
}
if (!objspace->profile.records) {
rb_bug("gc_profile malloc or realloc miss");
}
record = objspace->profile.current_record = &objspace->profile.records[objspace->profile.next_index - 1];
MEMZERO(record, gc_profile_record, 1);
/* setup before-GC parameter */
record->flags = reason | ((ruby_gc_stress && !ruby_disable_gc_stress) ? GPR_FLAG_STRESS : 0);
#if MALLOC_ALLOCATED_SIZE
record->allocated_size = malloc_allocated_size;
#endif
#if GC_PROFILE_DETAIL_MEMORY
#ifdef RUSAGE_SELF
{
struct rusage usage;
if (getrusage(RUSAGE_SELF, &usage) == 0) {
record->maxrss = usage.ru_maxrss;
record->minflt = usage.ru_minflt;
record->majflt = usage.ru_majflt;
}
}
#endif
#endif
}
}
static inline void
gc_prof_timer_start(rb_objspace_t *objspace)
{
if (gc_prof_enabled(objspace)) {
gc_profile_record *record = gc_prof_record(objspace);
#if GC_PROFILE_MORE_DETAIL
record->prepare_time = objspace->profile.prepare_time;
#endif
record->gc_time = 0;
record->gc_invoke_time = getrusage_time();
}
}
static double
elapsed_time_from(double time)
{
double now = getrusage_time();
if (now > time) {
return now - time;
}
else {
return 0;
}
}
static inline void
gc_prof_timer_stop(rb_objspace_t *objspace)
{
if (gc_prof_enabled(objspace)) {
gc_profile_record *record = gc_prof_record(objspace);
record->gc_time = elapsed_time_from(record->gc_invoke_time);
record->gc_invoke_time -= objspace->profile.invoke_time;
}
}
static inline void
gc_prof_mark_timer_start(rb_objspace_t *objspace)
{
* probes.d: add DTrace probe declarations. [ruby-core:27448] * array.c (empty_ary_alloc, ary_new): added array create DTrace probe. * compile.c (rb_insns_name): allowing DTrace probes to access instruction sequence name. * Makefile.in: translate probes.d file to appropriate header file. * common.mk: declare dependencies on the DTrace header. * configure.in: add a test for existence of DTrace. * eval.c (setup_exception): add a probe for when an exception is raised. * gc.c: Add DTrace probes for mark begin and end, and sweep begin and end. * hash.c (empty_hash_alloc): Add a probe for hash allocation. * insns.def: Add probes for function entry and return. * internal.h: function declaration for compile.c change. * load.c (rb_f_load): add probes for `load` entry and exit, require entry and exit, and wrapping search_required for load path search. * object.c (rb_obj_alloc): added a probe for general object creation. * parse.y (yycompile0): added a probe around parse and compile phase. * string.c (empty_str_alloc, str_new): DTrace probes for string allocation. * test/dtrace/*: tests for DTrace probes. * vm.c (vm_invoke_proc): add probes for function return on exception raise, hash create, and instruction sequence execution. * vm_core.h: add probe declarations for function entry and exit. * vm_dump.c: add probes header file. * vm_eval.c (vm_call0_cfunc, vm_call0_cfunc_with_frame): add probe on function entry and return. * vm_exec.c: expose instruction number to instruction name function. * vm_insnshelper.c: add function entry and exit probes for cfunc methods. * vm_insnhelper.h: vm usage information is always collected, so uncomment the functions. 12 19:14:50 2012 Akinori MUSHA <knu@iDaemons.org> * configure.in (isinf, isnan): isinf() and isnan() are macros on DragonFly which cannot be found by AC_REPLACE_FUNCS(). This workaround enforces the fact that they exist on DragonFly. 12 15:59:38 2012 Shugo Maeda <shugo@ruby-lang.org> * vm_core.h (rb_call_info_t::refinements), compile.c (new_callinfo), vm_insnhelper.c (vm_search_method): revert r37616 because it's too slow. [ruby-dev:46477] * test/ruby/test_refinement.rb (test_inline_method_cache): skip the test until the bug is fixed efficiently. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37631 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-11-13 01:52:12 +04:00
if (RUBY_DTRACE_GC_MARK_BEGIN_ENABLED()) {
RUBY_DTRACE_GC_MARK_BEGIN();
}
#if GC_PROFILE_MORE_DETAIL
if (gc_prof_enabled(objspace)) {
gc_prof_record(objspace)->gc_mark_time = getrusage_time();
}
#endif
}
static inline void
gc_prof_mark_timer_stop(rb_objspace_t *objspace)
{
* probes.d: add DTrace probe declarations. [ruby-core:27448] * array.c (empty_ary_alloc, ary_new): added array create DTrace probe. * compile.c (rb_insns_name): allowing DTrace probes to access instruction sequence name. * Makefile.in: translate probes.d file to appropriate header file. * common.mk: declare dependencies on the DTrace header. * configure.in: add a test for existence of DTrace. * eval.c (setup_exception): add a probe for when an exception is raised. * gc.c: Add DTrace probes for mark begin and end, and sweep begin and end. * hash.c (empty_hash_alloc): Add a probe for hash allocation. * insns.def: Add probes for function entry and return. * internal.h: function declaration for compile.c change. * load.c (rb_f_load): add probes for `load` entry and exit, require entry and exit, and wrapping search_required for load path search. * object.c (rb_obj_alloc): added a probe for general object creation. * parse.y (yycompile0): added a probe around parse and compile phase. * string.c (empty_str_alloc, str_new): DTrace probes for string allocation. * test/dtrace/*: tests for DTrace probes. * vm.c (vm_invoke_proc): add probes for function return on exception raise, hash create, and instruction sequence execution. * vm_core.h: add probe declarations for function entry and exit. * vm_dump.c: add probes header file. * vm_eval.c (vm_call0_cfunc, vm_call0_cfunc_with_frame): add probe on function entry and return. * vm_exec.c: expose instruction number to instruction name function. * vm_insnshelper.c: add function entry and exit probes for cfunc methods. * vm_insnhelper.h: vm usage information is always collected, so uncomment the functions. 12 19:14:50 2012 Akinori MUSHA <knu@iDaemons.org> * configure.in (isinf, isnan): isinf() and isnan() are macros on DragonFly which cannot be found by AC_REPLACE_FUNCS(). This workaround enforces the fact that they exist on DragonFly. 12 15:59:38 2012 Shugo Maeda <shugo@ruby-lang.org> * vm_core.h (rb_call_info_t::refinements), compile.c (new_callinfo), vm_insnhelper.c (vm_search_method): revert r37616 because it's too slow. [ruby-dev:46477] * test/ruby/test_refinement.rb (test_inline_method_cache): skip the test until the bug is fixed efficiently. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37631 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-11-13 01:52:12 +04:00
if (RUBY_DTRACE_GC_MARK_END_ENABLED()) {
RUBY_DTRACE_GC_MARK_END();
}
#if GC_PROFILE_MORE_DETAIL
if (gc_prof_enabled(objspace)) {
gc_profile_record *record = gc_prof_record(objspace);
record->gc_mark_time = elapsed_time_from(record->gc_mark_time);
}
#endif
}
static inline void
gc_prof_sweep_timer_start(rb_objspace_t *objspace)
{
* probes.d: add DTrace probe declarations. [ruby-core:27448] * array.c (empty_ary_alloc, ary_new): added array create DTrace probe. * compile.c (rb_insns_name): allowing DTrace probes to access instruction sequence name. * Makefile.in: translate probes.d file to appropriate header file. * common.mk: declare dependencies on the DTrace header. * configure.in: add a test for existence of DTrace. * eval.c (setup_exception): add a probe for when an exception is raised. * gc.c: Add DTrace probes for mark begin and end, and sweep begin and end. * hash.c (empty_hash_alloc): Add a probe for hash allocation. * insns.def: Add probes for function entry and return. * internal.h: function declaration for compile.c change. * load.c (rb_f_load): add probes for `load` entry and exit, require entry and exit, and wrapping search_required for load path search. * object.c (rb_obj_alloc): added a probe for general object creation. * parse.y (yycompile0): added a probe around parse and compile phase. * string.c (empty_str_alloc, str_new): DTrace probes for string allocation. * test/dtrace/*: tests for DTrace probes. * vm.c (vm_invoke_proc): add probes for function return on exception raise, hash create, and instruction sequence execution. * vm_core.h: add probe declarations for function entry and exit. * vm_dump.c: add probes header file. * vm_eval.c (vm_call0_cfunc, vm_call0_cfunc_with_frame): add probe on function entry and return. * vm_exec.c: expose instruction number to instruction name function. * vm_insnshelper.c: add function entry and exit probes for cfunc methods. * vm_insnhelper.h: vm usage information is always collected, so uncomment the functions. 12 19:14:50 2012 Akinori MUSHA <knu@iDaemons.org> * configure.in (isinf, isnan): isinf() and isnan() are macros on DragonFly which cannot be found by AC_REPLACE_FUNCS(). This workaround enforces the fact that they exist on DragonFly. 12 15:59:38 2012 Shugo Maeda <shugo@ruby-lang.org> * vm_core.h (rb_call_info_t::refinements), compile.c (new_callinfo), vm_insnhelper.c (vm_search_method): revert r37616 because it's too slow. [ruby-dev:46477] * test/ruby/test_refinement.rb (test_inline_method_cache): skip the test until the bug is fixed efficiently. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37631 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-11-13 01:52:12 +04:00
if (RUBY_DTRACE_GC_SWEEP_BEGIN_ENABLED()) {
RUBY_DTRACE_GC_SWEEP_BEGIN();
}
if (gc_prof_enabled(objspace)) {
gc_profile_record *record = gc_prof_record(objspace);
if (record->gc_time > 0 || GC_PROFILE_MORE_DETAIL) {
objspace->profile.gc_sweep_start_time = getrusage_time();
}
}
}
static inline void
gc_prof_sweep_timer_stop(rb_objspace_t *objspace)
{
* probes.d: add DTrace probe declarations. [ruby-core:27448] * array.c (empty_ary_alloc, ary_new): added array create DTrace probe. * compile.c (rb_insns_name): allowing DTrace probes to access instruction sequence name. * Makefile.in: translate probes.d file to appropriate header file. * common.mk: declare dependencies on the DTrace header. * configure.in: add a test for existence of DTrace. * eval.c (setup_exception): add a probe for when an exception is raised. * gc.c: Add DTrace probes for mark begin and end, and sweep begin and end. * hash.c (empty_hash_alloc): Add a probe for hash allocation. * insns.def: Add probes for function entry and return. * internal.h: function declaration for compile.c change. * load.c (rb_f_load): add probes for `load` entry and exit, require entry and exit, and wrapping search_required for load path search. * object.c (rb_obj_alloc): added a probe for general object creation. * parse.y (yycompile0): added a probe around parse and compile phase. * string.c (empty_str_alloc, str_new): DTrace probes for string allocation. * test/dtrace/*: tests for DTrace probes. * vm.c (vm_invoke_proc): add probes for function return on exception raise, hash create, and instruction sequence execution. * vm_core.h: add probe declarations for function entry and exit. * vm_dump.c: add probes header file. * vm_eval.c (vm_call0_cfunc, vm_call0_cfunc_with_frame): add probe on function entry and return. * vm_exec.c: expose instruction number to instruction name function. * vm_insnshelper.c: add function entry and exit probes for cfunc methods. * vm_insnhelper.h: vm usage information is always collected, so uncomment the functions. 12 19:14:50 2012 Akinori MUSHA <knu@iDaemons.org> * configure.in (isinf, isnan): isinf() and isnan() are macros on DragonFly which cannot be found by AC_REPLACE_FUNCS(). This workaround enforces the fact that they exist on DragonFly. 12 15:59:38 2012 Shugo Maeda <shugo@ruby-lang.org> * vm_core.h (rb_call_info_t::refinements), compile.c (new_callinfo), vm_insnhelper.c (vm_search_method): revert r37616 because it's too slow. [ruby-dev:46477] * test/ruby/test_refinement.rb (test_inline_method_cache): skip the test until the bug is fixed efficiently. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37631 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-11-13 01:52:12 +04:00
if (RUBY_DTRACE_GC_SWEEP_END_ENABLED()) {
RUBY_DTRACE_GC_SWEEP_END();
}
if (gc_prof_enabled(objspace)) {
double sweep_time;
gc_profile_record *record = gc_prof_record(objspace);
if (record->gc_time > 0) {
sweep_time = elapsed_time_from(objspace->profile.gc_sweep_start_time);
/* need to accumulate GC time for lazy sweep after gc() */
record->gc_time += sweep_time;
}
else if (GC_PROFILE_MORE_DETAIL) {
sweep_time = elapsed_time_from(objspace->profile.gc_sweep_start_time);
}
#if GC_PROFILE_MORE_DETAIL
record->gc_sweep_time += sweep_time;
if (heap_pages_deferred_final) record->flags |= GPR_FLAG_HAVE_FINALIZE;
#endif
if (heap_pages_deferred_final) objspace->profile.latest_gc_info |= GPR_FLAG_HAVE_FINALIZE;
}
}
static inline void
gc_prof_set_malloc_info(rb_objspace_t *objspace)
{
#if GC_PROFILE_MORE_DETAIL
if (gc_prof_enabled(objspace)) {
gc_profile_record *record = gc_prof_record(objspace);
record->allocate_increase = malloc_increase;
record->allocate_limit = malloc_limit;
}
#endif
}
static inline void
gc_prof_set_heap_info(rb_objspace_t *objspace)
{
if (gc_prof_enabled(objspace)) {
gc_profile_record *record = gc_prof_record(objspace);
size_t live = objspace->profile.total_allocated_object_num_at_gc_start - objspace->profile.total_freed_object_num;
size_t total = objspace->profile.heap_used_at_gc_start * HEAP_OBJ_LIMIT;
#if GC_PROFILE_MORE_DETAIL
record->heap_use_pages = objspace->profile.heap_used_at_gc_start;
record->heap_live_objects = live;
record->heap_free_objects = total - live;
#endif
record->heap_total_objects = total;
record->heap_use_size = live * sizeof(RVALUE);
record->heap_total_size = total * sizeof(RVALUE);
}
}
/*
* call-seq:
* GC::Profiler.clear -> nil
*
* Clears the GC profiler data.
*
*/
static VALUE
gc_profile_clear(void)
{
rb_objspace_t *objspace = &rb_objspace;
if (GC_PROFILE_RECORD_DEFAULT_SIZE * 2 < objspace->profile.size) {
objspace->profile.size = GC_PROFILE_RECORD_DEFAULT_SIZE * 2;
objspace->profile.records = realloc(objspace->profile.records, sizeof(gc_profile_record) * objspace->profile.size);
if (!objspace->profile.records) {
rb_memerror();
}
}
MEMZERO(objspace->profile.records, gc_profile_record, objspace->profile.size);
objspace->profile.next_index = 0;
objspace->profile.current_record = 0;
return Qnil;
}
/*
* call-seq:
* GC::Profiler.raw_data -> [Hash, ...]
*
* Returns an Array of individual raw profile data Hashes ordered
* from earliest to latest by +:GC_INVOKE_TIME+.
*
* For example:
*
* [
* {
* :GC_TIME=>1.3000000000000858e-05,
* :GC_INVOKE_TIME=>0.010634999999999999,
* :HEAP_USE_SIZE=>289640,
* :HEAP_TOTAL_SIZE=>588960,
* :HEAP_TOTAL_OBJECTS=>14724,
* :GC_IS_MARKED=>false
* },
* # ...
* ]
*
* The keys mean:
*
* +:GC_TIME+::
* Time elapsed in seconds for this GC run
* +:GC_INVOKE_TIME+::
* Time elapsed in seconds from startup to when the GC was invoked
* +:HEAP_USE_SIZE+::
* Total bytes of heap used
* +:HEAP_TOTAL_SIZE+::
* Total size of heap in bytes
* +:HEAP_TOTAL_OBJECTS+::
* Total number of objects
* +:GC_IS_MARKED+::
* Returns +true+ if the GC is in mark phase
*
* If ruby was built with +GC_PROFILE_MORE_DETAIL+, you will also have access
* to the following hash keys:
*
* +:GC_MARK_TIME+::
* +:GC_SWEEP_TIME+::
* +:ALLOCATE_INCREASE+::
* +:ALLOCATE_LIMIT+::
* +:HEAP_USE_PAGES+::
* +:HEAP_LIVE_OBJECTS+::
* +:HEAP_FREE_OBJECTS+::
* +:HAVE_FINALIZE+::
*
*/
static VALUE
gc_profile_record_get(void)
{
VALUE prof;
VALUE gc_profile = rb_ary_new();
size_t i;
rb_objspace_t *objspace = (&rb_objspace);
if (!objspace->profile.run) {
return Qnil;
}
for (i =0; i < objspace->profile.next_index; i++) {
gc_profile_record *record = &objspace->profile.records[i];
prof = rb_hash_new();
rb_hash_aset(prof, ID2SYM(rb_intern("GC_FLAGS")), gc_info_decode(record->flags, rb_hash_new()));
rb_hash_aset(prof, ID2SYM(rb_intern("GC_TIME")), DBL2NUM(record->gc_time));
rb_hash_aset(prof, ID2SYM(rb_intern("GC_INVOKE_TIME")), DBL2NUM(record->gc_invoke_time));
rb_hash_aset(prof, ID2SYM(rb_intern("HEAP_USE_SIZE")), SIZET2NUM(record->heap_use_size));
rb_hash_aset(prof, ID2SYM(rb_intern("HEAP_TOTAL_SIZE")), SIZET2NUM(record->heap_total_size));
rb_hash_aset(prof, ID2SYM(rb_intern("HEAP_TOTAL_OBJECTS")), SIZET2NUM(record->heap_total_objects));
rb_hash_aset(prof, ID2SYM(rb_intern("GC_IS_MARKED")), Qtrue);
#if GC_PROFILE_MORE_DETAIL
rb_hash_aset(prof, ID2SYM(rb_intern("GC_MARK_TIME")), DBL2NUM(record->gc_mark_time));
rb_hash_aset(prof, ID2SYM(rb_intern("GC_SWEEP_TIME")), DBL2NUM(record->gc_sweep_time));
rb_hash_aset(prof, ID2SYM(rb_intern("ALLOCATE_INCREASE")), SIZET2NUM(record->allocate_increase));
rb_hash_aset(prof, ID2SYM(rb_intern("ALLOCATE_LIMIT")), SIZET2NUM(record->allocate_limit));
rb_hash_aset(prof, ID2SYM(rb_intern("HEAP_USE_PAGES")), SIZET2NUM(record->heap_use_pages));
rb_hash_aset(prof, ID2SYM(rb_intern("HEAP_LIVE_OBJECTS")), SIZET2NUM(record->heap_live_objects));
rb_hash_aset(prof, ID2SYM(rb_intern("HEAP_FREE_OBJECTS")), SIZET2NUM(record->heap_free_objects));
rb_hash_aset(prof, ID2SYM(rb_intern("REMOVING_OBJECTS")), SIZET2NUM(record->removing_objects));
rb_hash_aset(prof, ID2SYM(rb_intern("EMPTY_OBJECTS")), SIZET2NUM(record->empty_objects));
rb_hash_aset(prof, ID2SYM(rb_intern("HAVE_FINALIZE")), (record->flags & GPR_FLAG_HAVE_FINALIZE) ? Qtrue : Qfalse);
#endif
#if RGENGC_PROFILE > 0
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
rb_hash_aset(prof, ID2SYM(rb_intern("OLD_OBJECTS")), SIZET2NUM(record->old_objects));
rb_hash_aset(prof, ID2SYM(rb_intern("REMEMBED_NORMAL_OBJECTS")), SIZET2NUM(record->remembered_normal_objects));
rb_hash_aset(prof, ID2SYM(rb_intern("REMEMBED_SHADY_OBJECTS")), SIZET2NUM(record->remembered_shady_objects));
#endif
rb_ary_push(gc_profile, prof);
}
return gc_profile;
}
#if GC_PROFILE_MORE_DETAIL
#define MAJOR_REASON_MAX 0x10
static char *
gc_profile_dump_major_reason(int flags, char *buff)
{
int reason = flags & GPR_FLAG_MAJOR_MASK;
int i = 0;
if (reason == GPR_FLAG_NONE) {
buff[0] = '-';
buff[1] = 0;
}
else {
#define C(x, s) \
if (reason & GPR_FLAG_MAJOR_BY_##x) { \
buff[i++] = #x[0]; \
if (i >= MAJOR_REASON_MAX) rb_bug("gc_profile_dump_major_reason: overflow"); \
buff[i] = 0; \
}
C(NOFREE, N);
C(OLDGEN, O);
C(SHADY, S);
C(RESCAN, R);
C(STRESS, T);
#if RGENGC_ESTIMATE_OLDMALLOC
C(OLDMALLOC, M);
#endif
#undef C
}
return buff;
}
#endif
static void
gc_profile_dump_on(VALUE out, VALUE (*append)(VALUE, VALUE))
{
rb_objspace_t *objspace = &rb_objspace;
size_t count = objspace->profile.next_index;
#ifdef MAJOR_REASON_MAX
char reason_str[MAJOR_REASON_MAX];
#endif
if (objspace->profile.run && count /* > 1 */) {
size_t i;
const gc_profile_record *record;
append(out, rb_sprintf("GC %"PRIuSIZE" invokes.\n", objspace->profile.count));
append(out, rb_str_new_cstr("Index Invoke Time(sec) Use Size(byte) Total Size(byte) Total Object GC Time(ms)\n"));
for (i = 0; i < count; i++) {
record = &objspace->profile.records[i];
append(out, rb_sprintf("%5"PRIdSIZE" %19.3f %20"PRIuSIZE" %20"PRIuSIZE" %20"PRIuSIZE" %30.20f\n",
i+1, record->gc_invoke_time, record->heap_use_size,
record->heap_total_size, record->heap_total_objects, record->gc_time*1000));
}
#if GC_PROFILE_MORE_DETAIL
append(out, rb_str_new_cstr("\n\n" \
"More detail.\n" \
"Prepare Time = Previously GC's rest sweep time\n"
"Index Flags Allocate Inc. Allocate Limit"
#if CALC_EXACT_MALLOC_SIZE
" Allocated Size"
#endif
" Use Page Mark Time(ms) Sweep Time(ms) Prepare Time(ms) LivingObj FreeObj RemovedObj EmptyObj"
#if RGENGC_PROFILE
" OldgenObj RemNormObj RemShadObj"
#endif
#if GC_PROFILE_DETAIL_MEMORY
" MaxRSS(KB) MinorFLT MajorFLT"
#endif
"\n"));
for (i = 0; i < count; i++) {
record = &objspace->profile.records[i];
append(out, rb_sprintf("%5"PRIdSIZE" %4s/%c/%6s%c %13"PRIuSIZE" %15"PRIuSIZE
#if CALC_EXACT_MALLOC_SIZE
" %15"PRIuSIZE
#endif
" %9"PRIuSIZE" %17.12f %17.12f %17.12f %10"PRIuSIZE" %10"PRIuSIZE" %10"PRIuSIZE" %10"PRIuSIZE
#if RGENGC_PROFILE
"%10"PRIuSIZE" %10"PRIuSIZE" %10"PRIuSIZE
#endif
#if GC_PROFILE_DETAIL_MEMORY
"%11ld %8ld %8ld"
#endif
"\n",
i+1,
gc_profile_dump_major_reason(record->flags, reason_str),
(record->flags & GPR_FLAG_HAVE_FINALIZE) ? 'F' : '.',
(record->flags & GPR_FLAG_NEWOBJ) ? "NEWOBJ" :
(record->flags & GPR_FLAG_MALLOC) ? "MALLOC" :
(record->flags & GPR_FLAG_METHOD) ? "METHOD" :
(record->flags & GPR_FLAG_CAPI) ? "CAPI__" : "??????",
(record->flags & GPR_FLAG_STRESS) ? '!' : ' ',
record->allocate_increase, record->allocate_limit,
#if CALC_EXACT_MALLOC_SIZE
record->allocated_size,
#endif
record->heap_use_pages,
record->gc_mark_time*1000,
record->gc_sweep_time*1000,
record->prepare_time*1000,
record->heap_live_objects,
record->heap_free_objects,
record->removing_objects,
record->empty_objects
#if RGENGC_PROFILE
,
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
record->old_objects,
record->remembered_normal_objects,
record->remembered_shady_objects
#endif
#if GC_PROFILE_DETAIL_MEMORY
,
record->maxrss / 1024,
record->minflt,
record->majflt
#endif
));
}
#endif
}
}
/*
* call-seq:
* GC::Profiler.result -> String
*
* Returns a profile data report such as:
*
* GC 1 invokes.
* Index Invoke Time(sec) Use Size(byte) Total Size(byte) Total Object GC time(ms)
* 1 0.012 159240 212940 10647 0.00000000000001530000
*/
static VALUE
gc_profile_result(void)
{
VALUE str = rb_str_buf_new(0);
gc_profile_dump_on(str, rb_str_buf_append);
return str;
}
/*
* call-seq:
* GC::Profiler.report
* GC::Profiler.report(io)
*
* Writes the GC::Profiler.result to <tt>$stdout</tt> or the given IO object.
*
*/
static VALUE
gc_profile_report(int argc, VALUE *argv, VALUE self)
{
VALUE out;
if (argc == 0) {
out = rb_stdout;
}
else {
rb_scan_args(argc, argv, "01", &out);
}
gc_profile_dump_on(out, rb_io_write);
return Qnil;
}
/*
* call-seq:
* GC::Profiler.total_time -> float
*
* The total time used for garbage collection in seconds
*/
static VALUE
gc_profile_total_time(VALUE self)
{
double time = 0;
rb_objspace_t *objspace = &rb_objspace;
if (objspace->profile.run && objspace->profile.next_index > 0) {
size_t i;
size_t count = objspace->profile.next_index;
for (i = 0; i < count; i++) {
time += objspace->profile.records[i].gc_time;
}
}
return DBL2NUM(time);
}
/*
* call-seq:
* GC::Profiler.enabled? -> true or false
*
* The current status of GC profile mode.
*/
static VALUE
gc_profile_enable_get(VALUE self)
{
rb_objspace_t *objspace = &rb_objspace;
return objspace->profile.run ? Qtrue : Qfalse;
}
/*
* call-seq:
* GC::Profiler.enable -> nil
*
* Starts the GC profiler.
*
*/
static VALUE
gc_profile_enable(void)
{
rb_objspace_t *objspace = &rb_objspace;
objspace->profile.run = TRUE;
objspace->profile.current_record = 0;
return Qnil;
}
/*
* call-seq:
* GC::Profiler.disable -> nil
*
* Stops the GC profiler.
*
*/
static VALUE
gc_profile_disable(void)
{
rb_objspace_t *objspace = &rb_objspace;
objspace->profile.run = FALSE;
objspace->profile.current_record = 0;
return Qnil;
}
/*
------------------------------ DEBUG ------------------------------
*/
static const char *
type_name(int type, VALUE obj)
{
switch (type) {
#define TYPE_NAME(t) case (t): return #t;
TYPE_NAME(T_NONE);
TYPE_NAME(T_OBJECT);
TYPE_NAME(T_CLASS);
TYPE_NAME(T_MODULE);
TYPE_NAME(T_FLOAT);
TYPE_NAME(T_STRING);
TYPE_NAME(T_REGEXP);
TYPE_NAME(T_ARRAY);
TYPE_NAME(T_HASH);
TYPE_NAME(T_STRUCT);
TYPE_NAME(T_BIGNUM);
TYPE_NAME(T_FILE);
TYPE_NAME(T_MATCH);
TYPE_NAME(T_COMPLEX);
TYPE_NAME(T_RATIONAL);
TYPE_NAME(T_NIL);
TYPE_NAME(T_TRUE);
TYPE_NAME(T_FALSE);
TYPE_NAME(T_SYMBOL);
TYPE_NAME(T_FIXNUM);
TYPE_NAME(T_UNDEF);
TYPE_NAME(T_NODE);
TYPE_NAME(T_ICLASS);
TYPE_NAME(T_ZOMBIE);
case T_DATA:
if (obj && rb_objspace_data_type_name(obj)) {
return rb_objspace_data_type_name(obj);
}
return "T_DATA";
#undef TYPE_NAME
}
return "unknown";
}
static const char *
obj_type_name(VALUE obj)
{
return type_name(TYPE(obj), obj);
}
#if GC_DEBUG
void
rb_gcdebug_print_obj_condition(VALUE obj)
{
rb_objspace_t *objspace = &rb_objspace;
fprintf(stderr, "created at: %s:%d\n", RANY(obj)->file, RANY(obj)->line);
if (is_pointer_to_heap(objspace, (void *)obj)) {
fprintf(stderr, "pointer to heap?: true\n");
}
else {
fprintf(stderr, "pointer to heap?: false\n");
return;
}
fprintf(stderr, "marked? : %s\n", MARKED_IN_BITMAP(GET_HEAP_MARK_BITS(obj), obj) ? "true" : "false");
#if USE_RGENGC
#if RGENGC_AGE2_PROMOTION
fprintf(stderr, "young? : %s\n", RVALUE_YOUNG_P(obj) ? "true" : "false");
* gc.c: add 3gen GC patch, but disabled as default. RGenGC is designed as 2 generational GC, young and old generation. Young objects will be promoted to old objects after one GC. Old objects are not collect until major (full) GC. The issue of this approach is some objects can promoted as old objects accidentally and not freed until major GC. Major GC is not frequently so short-lived but accidentally becoming old objects are not freed. For example, the program "loop{Array.new(1_000_000)}" consumes huge memories because short lived objects (an array which has 1M elements) are promoted while GC and they are not freed before major GC. To solve this problem, generational GC with more generations technique is known. This patch implements three generations gen GC. At first, newly created objects are "Infant" objects. After surviving one GC, "Infant" objects are promoted to "Young" objects. "Young" objects are promoted to "Old" objects after surviving next GC. "Infant" and "Young" objects are collected if it is not marked while minor GC. So that this technique solves this problem. Representation of generations: * Infant: !FL_PROMOTED and !oldgen_bitmap [00] * Young : FL_PROMOTED and !oldgen_bitmap [10] * Old : FL_PROMOTED and oldgen_bitmap [11] The macro "RGENGC_THREEGEN" enables/disables this feature, and turned off as default because there are several problems. (1) Failed sometimes (Heisenbugs). (2) Performance down. Especially on write barriers. We need to detect Young or Old object by oldgen_bitmap. It is slower than checking flags. To evaluate this feature on more applications, I commit this patch. Reports are very welcome. This patch includes some refactoring (renaming names, etc). * include/ruby/ruby.h: catch up 3gen GC. * .gdbinit: fix to show a prompt "[PROMOTED]" for promoted objects. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43530 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-04 22:59:33 +04:00
#endif
fprintf(stderr, "old? : %s\n", RVALUE_OLD_P(obj) ? "true" : "false");
fprintf(stderr, "WB-protected?: %s\n", RVALUE_WB_PROTECTED(obj) ? "true" : "false");
fprintf(stderr, "remembered? : %s\n", MARKED_IN_BITMAP(GET_HEAP_REMEMBERSET_BITS(obj), obj) ? "true" : "false");
#endif
if (is_lazy_sweeping(heap_eden)) {
fprintf(stderr, "lazy sweeping?: true\n");
fprintf(stderr, "swept?: %s\n", is_swept_object(objspace, obj) ? "done" : "not yet");
}
else {
fprintf(stderr, "lazy sweeping?: false\n");
}
}
static VALUE
gcdebug_sential(VALUE obj, VALUE name)
{
fprintf(stderr, "WARNING: object %s(%p) is inadvertently collected\n", (char *)name, (void *)obj);
return Qnil;
}
void
rb_gcdebug_sentinel(VALUE obj, const char *name)
{
rb_define_finalizer(obj, rb_proc_new(gcdebug_sential, (VALUE)name));
}
#endif /* GC_DEBUG */
/*
* Document-module: ObjectSpace
*
* The ObjectSpace module contains a number of routines
* that interact with the garbage collection facility and allow you to
* traverse all living objects with an iterator.
*
* ObjectSpace also provides support for object finalizers, procs that will be
* called when a specific object is about to be destroyed by garbage
* collection.
*
* a = "A"
* b = "B"
*
* ObjectSpace.define_finalizer(a, proc {|id| puts "Finalizer one on #{id}" })
* ObjectSpace.define_finalizer(b, proc {|id| puts "Finalizer two on #{id}" })
*
* _produces:_
*
* Finalizer two on 537763470
* Finalizer one on 537763480
*/
/*
* Document-class: ObjectSpace::WeakMap
*
* An ObjectSpace::WeakMap object holds references to
* any objects, but those objects can get garbage collected.
*
* This class is mostly used internally by WeakRef, please use
* +lib/weakref.rb+ for the public interface.
*/
/* Document-class: GC::Profiler
*
* The GC profiler provides access to information on GC runs including time,
* length and object space size.
*
* Example:
*
* GC::Profiler.enable
*
* require 'rdoc/rdoc'
*
* GC::Profiler.report
*
* GC::Profiler.disable
*
* See also GC.count, GC.malloc_allocated_size and GC.malloc_allocations
*/
/*
* The GC module provides an interface to Ruby's mark and
* sweep garbage collection mechanism.
*
* Some of the underlying methods are also available via the ObjectSpace
* module.
*
* You may obtain information about the operation of the GC through
* GC::Profiler.
*/
void
Init_GC(void)
{
VALUE rb_mObjSpace;
VALUE rb_mProfiler;
VALUE gc_constants;
rb_mGC = rb_define_module("GC");
rb_define_singleton_method(rb_mGC, "start", gc_start_internal, -1);
rb_define_singleton_method(rb_mGC, "enable", rb_gc_enable, 0);
rb_define_singleton_method(rb_mGC, "disable", rb_gc_disable, 0);
rb_define_singleton_method(rb_mGC, "stress", gc_stress_get, 0);
rb_define_singleton_method(rb_mGC, "stress=", gc_stress_set, 1);
rb_define_singleton_method(rb_mGC, "count", gc_count, 0);
rb_define_singleton_method(rb_mGC, "stat", gc_stat, -1);
rb_define_singleton_method(rb_mGC, "latest_gc_info", gc_latest_gc_info, -1);
rb_define_method(rb_mGC, "garbage_collect", gc_start_internal, -1);
gc_constants = rb_hash_new();
rb_hash_aset(gc_constants, ID2SYM(rb_intern("RVALUE_SIZE")), SIZET2NUM(sizeof(RVALUE)));
rb_hash_aset(gc_constants, ID2SYM(rb_intern("HEAP_OBJ_LIMIT")), SIZET2NUM(HEAP_OBJ_LIMIT));
rb_hash_aset(gc_constants, ID2SYM(rb_intern("HEAP_BITMAP_SIZE")), SIZET2NUM(HEAP_BITMAP_SIZE));
rb_hash_aset(gc_constants, ID2SYM(rb_intern("HEAP_BITMAP_PLANES")), SIZET2NUM(HEAP_BITMAP_PLANES));
OBJ_FREEZE(gc_constants);
rb_define_const(rb_mGC, "INTERNAL_CONSTANTS", gc_constants);
rb_mProfiler = rb_define_module_under(rb_mGC, "Profiler");
rb_define_singleton_method(rb_mProfiler, "enabled?", gc_profile_enable_get, 0);
rb_define_singleton_method(rb_mProfiler, "enable", gc_profile_enable, 0);
rb_define_singleton_method(rb_mProfiler, "raw_data", gc_profile_record_get, 0);
rb_define_singleton_method(rb_mProfiler, "disable", gc_profile_disable, 0);
rb_define_singleton_method(rb_mProfiler, "clear", gc_profile_clear, 0);
rb_define_singleton_method(rb_mProfiler, "result", gc_profile_result, 0);
rb_define_singleton_method(rb_mProfiler, "report", gc_profile_report, -1);
rb_define_singleton_method(rb_mProfiler, "total_time", gc_profile_total_time, 0);
rb_mObjSpace = rb_define_module("ObjectSpace");
rb_define_module_function(rb_mObjSpace, "each_object", os_each_obj, -1);
rb_define_module_function(rb_mObjSpace, "garbage_collect", gc_start_internal, -1);
rb_define_module_function(rb_mObjSpace, "define_finalizer", define_final, -1);
rb_define_module_function(rb_mObjSpace, "undefine_finalizer", undefine_final, 1);
rb_define_module_function(rb_mObjSpace, "_id2ref", id2ref, 1);
nomem_error = rb_exc_new3(rb_eNoMemError,
rb_obj_freeze(rb_str_new2("failed to allocate memory")));
OBJ_TAINT(nomem_error);
OBJ_FREEZE(nomem_error);
rb_define_method(rb_cBasicObject, "__id__", rb_obj_id, 0);
rb_define_method(rb_mKernel, "object_id", rb_obj_id, 0);
rb_define_module_function(rb_mObjSpace, "count_objects", count_objects, -1);
{
VALUE rb_cWeakMap = rb_define_class_under(rb_mObjSpace, "WeakMap", rb_cObject);
rb_define_alloc_func(rb_cWeakMap, wmap_allocate);
rb_define_method(rb_cWeakMap, "[]=", wmap_aset, 2);
rb_define_method(rb_cWeakMap, "[]", wmap_aref, 1);
rb_define_method(rb_cWeakMap, "include?", wmap_has_key, 1);
rb_define_method(rb_cWeakMap, "member?", wmap_has_key, 1);
rb_define_method(rb_cWeakMap, "key?", wmap_has_key, 1);
rb_define_method(rb_cWeakMap, "inspect", wmap_inspect, 0);
rb_define_method(rb_cWeakMap, "each", wmap_each, 0);
rb_define_method(rb_cWeakMap, "each_pair", wmap_each, 0);
rb_define_method(rb_cWeakMap, "each_key", wmap_each_key, 0);
rb_define_method(rb_cWeakMap, "each_value", wmap_each_value, 0);
rb_define_method(rb_cWeakMap, "keys", wmap_keys, 0);
rb_define_method(rb_cWeakMap, "values", wmap_values, 0);
rb_define_method(rb_cWeakMap, "size", wmap_size, 0);
rb_define_method(rb_cWeakMap, "length", wmap_size, 0);
rb_define_private_method(rb_cWeakMap, "finalize", wmap_finalize, 1);
rb_include_module(rb_cWeakMap, rb_mEnumerable);
}
/* internal methods */
rb_define_singleton_method(rb_mGC, "verify_internal_consistency", gc_verify_internal_consistency, 0);
#if MALLOC_ALLOCATED_SIZE
rb_define_singleton_method(rb_mGC, "malloc_allocated_size", gc_malloc_allocated_size, 0);
rb_define_singleton_method(rb_mGC, "malloc_allocations", gc_malloc_allocations, 0);
#endif
/* ::GC::OPTS, which shows GC build options */
{
VALUE opts;
rb_define_const(rb_mGC, "OPTS", opts = rb_ary_new());
#define OPT(o) if (o) rb_ary_push(opts, rb_str_new2(#o))
OPT(GC_DEBUG);
OPT(USE_RGENGC);
OPT(RGENGC_DEBUG);
OPT(RGENGC_CHECK_MODE);
OPT(RGENGC_PROFILE);
OPT(RGENGC_AGE2_PROMOTION);
OPT(RGENGC_ESTIMATE_OLDMALLOC);
OPT(GC_PROFILE_MORE_DETAIL);
OPT(GC_ENABLE_LAZY_SWEEP);
OPT(CALC_EXACT_MALLOC_SIZE);
OPT(MALLOC_ALLOCATED_SIZE);
OPT(MALLOC_ALLOCATED_SIZE_CHECK);
OPT(GC_PROFILE_DETAIL_MEMORY);
#undef OPT
}
}