ruby/vm_method.c

2472 строки
72 KiB
C
Исходник Обычный вид История

/*
* This file is included by vm.c
*/
#include "id_table.h"
#define METHOD_DEBUG 0
static int vm_redefinition_check_flag(VALUE klass);
static void rb_vm_check_redefinition_opt_method(const rb_method_entry_t *me, VALUE klass);
#define object_id idObject_id
#define added idMethod_added
#define singleton_added idSingleton_method_added
#define removed idMethod_removed
#define singleton_removed idSingleton_method_removed
#define undefined idMethod_undefined
#define singleton_undefined idSingleton_method_undefined
#define attached id__attached__
#define ruby_running (GET_VM()->running)
/* int ruby_running = 0; */
static enum rb_id_table_iterator_result
vm_ccs_dump_i(ID mid, VALUE val, void *data)
{
const struct rb_class_cc_entries *ccs = (struct rb_class_cc_entries *)val;
fprintf(stderr, " | %s (%d) ", rb_id2name(mid), ccs->len);
rp(ccs->cme);
for (int i=0; i<ccs->len; i++) {
fprintf(stderr, " | [%d] ", i); vm_ci_dump(ccs->entries[i].ci);
rp_m( " | ", ccs->entries[i].cc);
}
return ID_TABLE_CONTINUE;
}
static void
vm_ccs_dump(VALUE klass, ID target_mid)
{
struct rb_id_table *cc_tbl = RCLASS_CC_TBL(klass);
if (cc_tbl) {
const struct rb_class_cc_entries *ccs;
if (target_mid) {
if (rb_id_table_lookup(cc_tbl, target_mid, (VALUE *)&ccs)) {
fprintf(stderr, " [CCTB] %p\n", (void *)cc_tbl);
vm_ccs_dump_i(target_mid, (VALUE)ccs, NULL);
}
}
else {
fprintf(stderr, " [CCTB] %p\n", (void *)cc_tbl);
rb_id_table_foreach(cc_tbl, vm_ccs_dump_i, (void *)target_mid);
}
}
}
static enum rb_id_table_iterator_result
vm_cme_dump_i(ID mid, VALUE val, void *data)
{
ID target_mid = (ID)data;
if (target_mid == 0 || mid == target_mid) {
rp_m(" > ", val);
}
return ID_TABLE_CONTINUE;
}
static VALUE
vm_mtbl_dump(VALUE klass, ID target_mid)
{
fprintf(stderr, "# vm_mtbl\n");
while (klass) {
rp_m(" -> ", klass);
rb_method_entry_t *me;
if (RCLASS_M_TBL(klass)) {
if (target_mid != 0) {
if (rb_id_table_lookup(RCLASS_M_TBL(klass), target_mid, (VALUE *)&me)) {
rp_m(" [MTBL] ", me);
}
}
else {
fprintf(stderr, " ## RCLASS_M_TBL (%p)\n", (void *)RCLASS_M_TBL(klass));
rb_id_table_foreach(RCLASS_M_TBL(klass), vm_cme_dump_i, NULL);
}
}
else {
fprintf(stderr, " MTBL: NULL\n");
}
if (RCLASS_CALLABLE_M_TBL(klass)) {
if (target_mid != 0) {
if (rb_id_table_lookup(RCLASS_CALLABLE_M_TBL(klass), target_mid, (VALUE *)&me)) {
rp_m(" [CM**] ", me);
}
}
else {
fprintf(stderr, " ## RCLASS_CALLABLE_M_TBL\n");
rb_id_table_foreach(RCLASS_CALLABLE_M_TBL(klass), vm_cme_dump_i, NULL);
}
}
if (RCLASS_CC_TBL(klass)) {
vm_ccs_dump(klass, target_mid);
}
klass = RCLASS_SUPER(klass);
}
return Qnil;
}
void
rb_vm_mtbl_dump(const char *msg, VALUE klass, ID target_mid)
{
fprintf(stderr, "[%s] ", msg);
vm_mtbl_dump(klass, target_mid);
}
static inline void
vm_me_invalidate_cache(rb_callable_method_entry_t *cme)
{
VM_ASSERT(IMEMO_TYPE_P(cme, imemo_ment));
VM_ASSERT(callable_method_entry_p(cme));
METHOD_ENTRY_INVALIDATED_SET(cme);
RB_DEBUG_COUNTER_INC(cc_cme_invalidate);
}
void
rb_clear_constant_cache(void)
{
INC_GLOBAL_CONSTANT_STATE();
}
static rb_method_entry_t *rb_method_entry_alloc(ID called_id, VALUE owner, VALUE defined_class, const rb_method_definition_t *def);
const rb_method_entry_t * rb_method_entry_clone(const rb_method_entry_t *src_me);
static const rb_callable_method_entry_t *copmplemented_callable_method_entry(VALUE klass, ID id);
static void
clear_method_cache_by_id_in_class(VALUE klass, ID mid)
{
VM_ASSERT(RB_TYPE_P(klass, T_CLASS) || RB_TYPE_P(klass, T_ICLASS));
if (LIKELY(RCLASS_EXT(klass)->subclasses == NULL)) {
// no subclasses
// check only current class
struct rb_id_table *cc_tbl = RCLASS_CC_TBL(klass);
struct rb_class_cc_entries *ccs;
// invalidate CCs
if (cc_tbl && rb_id_table_lookup(cc_tbl, mid, (VALUE *)&ccs)) {
rb_vm_ccs_free(ccs);
rb_id_table_delete(cc_tbl, mid);
RB_DEBUG_COUNTER_INC(cc_invalidate_leaf_ccs);
}
// remove from callable_m_tbl, if exists
struct rb_id_table *cm_tbl;
if ((cm_tbl = RCLASS_CALLABLE_M_TBL(klass)) != NULL) {
rb_id_table_delete(cm_tbl, mid);
RB_DEBUG_COUNTER_INC(cc_invalidate_leaf_callable);
}
RB_DEBUG_COUNTER_INC(cc_invalidate_leaf);
}
else {
const rb_callable_method_entry_t *cme = copmplemented_callable_method_entry(klass, mid);
if (cme) {
// invalidate cme if found to invalidate the inline method cache.
if (METHOD_ENTRY_CACHED(cme)) {
if (METHOD_ENTRY_COMPLEMENTED(cme)) {
// do nothing
}
else {
// invalidate cc by invalidating cc->cme
VALUE owner = cme->owner;
VM_ASSERT(BUILTIN_TYPE(owner) == T_CLASS);
rb_callable_method_entry_t *new_cme =
(rb_callable_method_entry_t *)rb_method_entry_clone((const rb_method_entry_t *)cme);
struct rb_id_table *mtbl = RCLASS_M_TBL(RCLASS_ORIGIN(owner));
rb_id_table_insert(mtbl, mid, (VALUE)new_cme);
RB_OBJ_WRITTEN(owner, cme, new_cme);
}
vm_me_invalidate_cache((rb_callable_method_entry_t *)cme);
RB_DEBUG_COUNTER_INC(cc_invalidate_tree_cme);
}
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
// invalidate complement tbl
if (METHOD_ENTRY_COMPLEMENTED(cme)) {
VALUE defined_class = cme->defined_class;
struct rb_id_table *cm_tbl = RCLASS_CALLABLE_M_TBL(defined_class);
VM_ASSERT(cm_tbl != NULL);
int r = rb_id_table_delete(cm_tbl, mid);
VM_ASSERT(r == TRUE); (void)r;
RB_DEBUG_COUNTER_INC(cc_invalidate_tree_callable);
}
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
RB_DEBUG_COUNTER_INC(cc_invalidate_tree);
}
}
}
static void
clear_iclass_method_cache_by_id(VALUE iclass, VALUE d)
{
VM_ASSERT(RB_TYPE_P(iclass, T_ICLASS));
ID mid = (ID)d;
clear_method_cache_by_id_in_class(iclass, mid);
}
static void
clear_iclass_method_cache_by_id_for_refinements(VALUE klass, VALUE d)
{
if (RB_TYPE_P(klass, T_ICLASS)) {
ID mid = (ID)d;
clear_method_cache_by_id_in_class(klass, mid);
}
}
void
rb_clear_method_cache(VALUE klass_or_module, ID mid)
{
if (RB_TYPE_P(klass_or_module, T_MODULE)) {
VALUE module = klass_or_module; // alias
if (FL_TEST(module, RMODULE_IS_REFINEMENT)) {
VALUE refined_class = rb_refinement_module_get_refined_class(module);
rb_clear_method_cache(refined_class, mid);
rb_class_foreach_subclass(refined_class, clear_iclass_method_cache_by_id_for_refinements, mid);
}
rb_class_foreach_subclass(module, clear_iclass_method_cache_by_id, mid);
}
else {
clear_method_cache_by_id_in_class(klass_or_module, mid);
}
}
// gc.c
void rb_cc_table_free(VALUE klass);
static int
invalidate_all_cc(void *vstart, void *vend, size_t stride, void *data)
{
VALUE v = (VALUE)vstart;
for (; v != (VALUE)vend; v += stride) {
if (RBASIC(v)->flags) { // liveness check
if (RB_TYPE_P(v, T_CLASS) ||
RB_TYPE_P(v, T_ICLASS)) {
if (RCLASS_CC_TBL(v)) {
rb_cc_table_free(v);
}
RCLASS_CC_TBL(v) = NULL;
}
}
}
return 0; // continue to iteration
}
void
rb_clear_method_cache_all(void)
{
rb_objspace_each_objects(invalidate_all_cc, NULL);
}
VALUE
drop-in type check for rb_define_method The rb_define_method function takes a pointer to ANYARGS-ed functions, which in fact varies 18 different prototypes. We still need to preserve ANYARGS for storages but why not check the consistencies if possible. Q&As: Q: Where did the magic number "18" came from in the description above? A: Count the case branch of vm_method.c:call_cfunc_invoker_func(). Note also that the 18 branches has lasted for at least 25 years. See also 200e0ee2fd3c1c006c528874a88f684447215524. Q: What is this __weakref__ thing? A: That is a kind of function overloading mechanism that GCC provides. In this case for instance rb_define_method0 is an alias of rb_define_method, with a strong type. Q: What is this __transparent_union__ thing? A: That is another kind of function overloading mechanism that GCC provides. In this case the attributed function pointer is either VALUE(*)(int,VALUE*,VALUE) or VALUE(*)(int,const VALUE*,VALUE). This is better than void* or ANYARGS because we can reject all other possibilities than the two. Q: What does this rb_define_method macro mean? A: It selects appropriate alias of the rb_define_method function, depending on the arity. Q: Why the prototype change of rb_f_notimplement? A: Function pointer to rb_f_notimplement is special cased in vm_method.c:rb_add_method_cfunc(). That should be handled by the __builtin_choose_expr chain inside of rb_define_method macro expansion. In order to do so, comparison like (func == rb_f_notimplement) is inappropriate for __builtin_choose_expr's expression (which must be a compile-time integer constant but the address of rb_f_notimplement is not fixed until the linker). So instead we are using __builtin_types_compatible_p, and in doing so we need to distinguish rb_f_notimplement from others, by type.
2019-08-28 11:11:23 +03:00
rb_f_notimplement(int argc, const VALUE *argv, VALUE obj, VALUE marker)
{
rb_notimplement();
* encoding.c (rb_enc_codepoint_len): Use UNREACHABLE to avoid "control reaches end of non-void function" warnings. [ruby-trunk - Bug #6066] * re.c (name_to_backref_number): ditto. * object.c (rb_Float): ditto. * io.c (io_readpartial): ditto. * io.c (io_read_nonblock): ditto. * pack.c (rb_uv_to_utf8): ditto. * proc.c (rb_method_entry_arity): ditto. * vm_method.c (rb_f_notimplement): ditto. * struct.c (rb_struct_aset_id): ditto. * class.c (rb_scan_args): ditto. * process.c (rlimit_resource_type): ditto. * process.c (rlimit_resource_value): ditto. * process.c (p_uid_switch): ditto. * process.c (p_gid_switch): ditto. * ext/digest/digest.c (rb_digest_instance_update): ditto. * ext/digest/digest.c (rb_digest_instance_finish): ditto. * ext/digest/digest.c (rb_digest_instance_reset): ditto. * ext/digest/digest.c (rb_digest_instance_block_length): ditto. * ext/bigdecimal/bigdecimal.c (BigDecimalCmp): ditto. * ext/dl/handle.c (rb_dlhandle_close): ditto. * ext/tk/tcltklib.c (pending_exception_check0): ditto. * ext/tk/tcltklib.c (pending_exception_check1): ditto. * ext/tk/tcltklib.c (ip_cancel_eval_core): ditto. * ext/tk/tcltklib.c (lib_get_reltype_name): ditto. * ext/tk/tcltklib.c (create_dummy_encoding_for_tk_core): ditto. * ext/tk/tkutil/tkutil.c (tk_hash_kv): ditto. * ext/openssl/ossl_ssl.c (ossl_ssl_session_reused): ditto. * ext/openssl/ossl_pkey_ec.c (ossl_ec_key_dsa_verify_asn1): ditto. * ext/openssl/ossl_pkey_ec.c (ossl_ec_point_is_at_infinit): ditto. * ext/openssl/ossl_pkey_ec.c (ossl_ec_point_is_on_curve): ditto. * ext/fiddle/conversions.c (generic_to_value): ditto. * ext/socket/raddrinfo.c (rsock_io_socket_addrinfo): ditto. * ext/socket/socket.c (sock_s_getnameinfo): ditto. * ext/ripper/eventids2.c (ripper_token2eventid): ditto. * cont.c (return_fiber): ditto. * dmydln.c (dln_load): ditto. * vm_insnhelper.c (vm_search_normal_superclass): ditto. * bignum.c (big_fdiv): ditto. * marshal.c (r_symlink): ditto. * marshal.c (r_symbol): ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@35321 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-04-14 03:45:37 +04:00
UNREACHABLE_RETURN(Qnil);
}
static void
rb_define_notimplement_method_id(VALUE mod, ID id, rb_method_visibility_t visi)
{
rb_add_method(mod, id, VM_METHOD_TYPE_NOTIMPLEMENTED, (void *)1, visi);
}
void
rb_add_method_cfunc(VALUE klass, ID mid, VALUE (*func)(ANYARGS), int argc, rb_method_visibility_t visi)
{
if (argc < -2 || 15 < argc) rb_raise(rb_eArgError, "arity out of range: %d for -2..15", argc);
if (func != rb_f_notimplement) {
rb_method_cfunc_t opt;
opt.func = func;
opt.argc = argc;
rb_add_method(klass, mid, VM_METHOD_TYPE_CFUNC, &opt, visi);
}
else {
rb_define_notimplement_method_id(klass, mid, visi);
}
}
static void
rb_method_definition_release(rb_method_definition_t *def, int complemented)
{
if (def != NULL) {
const int alias_count = def->alias_count;
const int complemented_count = def->complemented_count;
VM_ASSERT(alias_count >= 0);
VM_ASSERT(complemented_count >= 0);
if (alias_count + complemented_count == 0) {
if (METHOD_DEBUG) fprintf(stderr, "-%p-%s:%d,%d (remove)\n", (void *)def,
Support targetting TracePoint [Feature #15289] * vm_trace.c (rb_tracepoint_enable_for_target): support targetting TracePoint. [Feature #15289] Tragetting TracePoint is only enabled on specified method, proc and so on, example: `tp.enable(target: code)`. `code` should be consisted of InstructionSeuqnece (iseq) (RubyVM::InstructionSeuqnece.of(code) should not return nil) If code is a tree of iseq, TracePoint is enabled on all of iseqs in a tree. Enabled tragetting TracePoints can not enabled again with and without target. * vm_core.h (rb_iseq_t): introduce `rb_iseq_t::local_hooks` to store local hooks. `rb_iseq_t::aux::trace_events` is renamed to `global_trace_events` to contrast with `local_hooks`. * vm_core.h (rb_hook_list_t): add `rb_hook_list_t::running` to represent how many Threads/Fibers are used this list. If this field is 0, nobody using this hooks and we can delete it. This is why we can remove code from cont.c. * vm_core.h (rb_vm_t): because of above change, we can eliminate `rb_vm_t::trace_running` field. Also renamed from `rb_vm_t::event_hooks` to `global_hooks`. * vm_core.h, vm.c (ruby_vm_event_enabled_global_flags): renamed from `ruby_vm_event_enabled_flags. * vm_core.h, vm.c (ruby_vm_event_local_num): added to count enabled targetting TracePoints. * vm_core.h, vm_trace.c (rb_exec_event_hooks): accepts hook list. * vm_core.h (rb_vm_global_hooks): added for convinience. * method.h (rb_method_bmethod_t): added to maintain Proc and `rb_hook_list_t` for bmethod (defined by define_method). * prelude.rb (TracePoint#enable): extracet a keyword parameter (because it is easy than writing in C). It calls `TracePoint#__enable` internal method written in C. * vm_insnhelper.c (vm_trace): check also iseq->local_hooks. * vm.c (invoke_bmethod): check def->body.bmethod.hooks. * vm.c (hook_before_rewind): check iseq->local_hooks and def->body.bmethod.hooks before rewind by exception. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@66003 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-11-26 21:16:39 +03:00
rb_id2name(def->original_id), alias_count, complemented_count);
VM_ASSERT(def->type == VM_METHOD_TYPE_BMETHOD ? def->body.bmethod.hooks == NULL : TRUE);
xfree(def);
}
else {
if (complemented) {
VM_ASSERT(def->complemented_count > 0);
def->complemented_count--;
}
else if (def->alias_count > 0) {
def->alias_count--;
}
if (METHOD_DEBUG) fprintf(stderr, "-%p-%s:%d->%d,%d->%d (dec)\n", (void *)def, rb_id2name(def->original_id),
alias_count, def->alias_count, complemented_count, def->complemented_count);
}
}
* 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
}
void
rb_free_method_entry(const rb_method_entry_t *me)
{
rb_method_definition_release(me->def, METHOD_ENTRY_COMPLEMENTED(me));
}
static inline rb_method_entry_t *search_method(VALUE klass, ID id, VALUE *defined_class_ptr);
mjit_compile.c: merge initial JIT compiler which has been developed by Takashi Kokubun <takashikkbn@gmail> as YARV-MJIT. Many of its bugs are fixed by wanabe <s.wanabe@gmail.com>. This JIT compiler is designed to be a safe migration path to introduce JIT compiler to MRI. So this commit does not include any bytecode changes or dynamic instruction modifications, which are done in original MJIT. This commit even strips off some aggressive optimizations from YARV-MJIT, and thus it's slower than YARV-MJIT too. But it's still fairly faster than Ruby 2.5 in some benchmarks (attached below). Note that this JIT compiler passes `make test`, `make test-all`, `make test-spec` without JIT, and even with JIT. Not only it's perfectly safe with JIT disabled because it does not replace VM instructions unlike MJIT, but also with JIT enabled it stably runs Ruby applications including Rails applications. I'm expecting this version as just "initial" JIT compiler. I have many optimization ideas which are skipped for initial merging, and you may easily replace this JIT compiler with a faster one by just replacing mjit_compile.c. `mjit_compile` interface is designed for the purpose. common.mk: update dependencies for mjit_compile.c. internal.h: declare `rb_vm_insn_addr2insn` for MJIT. vm.c: exclude some definitions if `-DMJIT_HEADER` is provided to compiler. This avoids to include some functions which take a long time to compile, e.g. vm_exec_core. Some of the purpose is achieved in transform_mjit_header.rb (see `IGNORED_FUNCTIONS`) but others are manually resolved for now. Load mjit_helper.h for MJIT header. mjit_helper.h: New. This is a file used only by JIT-ed code. I'll refactor `mjit_call_cfunc` later. vm_eval.c: add some #ifdef switches to skip compiling some functions like Init_vm_eval. win32/mkexports.rb: export thread/ec functions, which are used by MJIT. include/ruby/defines.h: add MJIT_FUNC_EXPORTED macro alis to clarify that a function is exported only for MJIT. array.c: export a function used by MJIT. bignum.c: ditto. class.c: ditto. compile.c: ditto. error.c: ditto. gc.c: ditto. hash.c: ditto. iseq.c: ditto. numeric.c: ditto. object.c: ditto. proc.c: ditto. re.c: ditto. st.c: ditto. string.c: ditto. thread.c: ditto. variable.c: ditto. vm_backtrace.c: ditto. vm_insnhelper.c: ditto. vm_method.c: ditto. I would like to improve maintainability of function exports, but I believe this way is acceptable as initial merging if we clarify the new exports are for MJIT (so that we can use them as TODO list to fix) and add unit tests to detect unresolved symbols. I'll add unit tests of JIT compilations in succeeding commits. Author: Takashi Kokubun <takashikkbn@gmail.com> Contributor: wanabe <s.wanabe@gmail.com> Part of [Feature #14235] --- * Known issues * Code generated by gcc is faster than clang. The benchmark may be worse in macOS. Following benchmark result is provided by gcc w/ Linux. * Performance is decreased when Google Chrome is running * JIT can work on MinGW, but it doesn't improve performance at least in short running benchmark. * Currently it doesn't perform well with Rails. We'll try to fix this before release. --- * Benchmark reslts Benchmarked with: Intel 4.0GHz i7-4790K with 16GB memory under x86-64 Ubuntu 8 Cores - 2.0.0-p0: Ruby 2.0.0-p0 - r62186: Ruby trunk (early 2.6.0), before MJIT changes - JIT off: On this commit, but without `--jit` option - JIT on: On this commit, and with `--jit` option ** Optcarrot fps Benchmark: https://github.com/mame/optcarrot | |2.0.0-p0 |r62186 |JIT off |JIT on | |:--------|:--------|:--------|:--------|:--------| |fps |37.32 |51.46 |51.31 |58.88 | |vs 2.0.0 |1.00x |1.38x |1.37x |1.58x | ** MJIT benchmarks Benchmark: https://github.com/benchmark-driver/mjit-benchmarks (Original: https://github.com/vnmakarov/ruby/tree/rtl_mjit_branch/MJIT-benchmarks) | |2.0.0-p0 |r62186 |JIT off |JIT on | |:----------|:--------|:--------|:--------|:--------| |aread |1.00 |1.09 |1.07 |2.19 | |aref |1.00 |1.13 |1.11 |2.22 | |aset |1.00 |1.50 |1.45 |2.64 | |awrite |1.00 |1.17 |1.13 |2.20 | |call |1.00 |1.29 |1.26 |2.02 | |const2 |1.00 |1.10 |1.10 |2.19 | |const |1.00 |1.11 |1.10 |2.19 | |fannk |1.00 |1.04 |1.02 |1.00 | |fib |1.00 |1.32 |1.31 |1.84 | |ivread |1.00 |1.13 |1.12 |2.43 | |ivwrite |1.00 |1.23 |1.21 |2.40 | |mandelbrot |1.00 |1.13 |1.16 |1.28 | |meteor |1.00 |2.97 |2.92 |3.17 | |nbody |1.00 |1.17 |1.15 |1.49 | |nest-ntimes|1.00 |1.22 |1.20 |1.39 | |nest-while |1.00 |1.10 |1.10 |1.37 | |norm |1.00 |1.18 |1.16 |1.24 | |nsvb |1.00 |1.16 |1.16 |1.17 | |red-black |1.00 |1.02 |0.99 |1.12 | |sieve |1.00 |1.30 |1.28 |1.62 | |trees |1.00 |1.14 |1.13 |1.19 | |while |1.00 |1.12 |1.11 |2.41 | ** Discourse's script/bench.rb Benchmark: https://github.com/discourse/discourse/blob/v1.8.7/script/bench.rb NOTE: Rails performance was somehow a little degraded with JIT for now. We should fix this. (At least I know opt_aref is performing badly in JIT and I have an idea to fix it. Please wait for the fix.) *** JIT off Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 17 75: 18 90: 22 99: 29 home_admin: 50: 21 75: 21 90: 27 99: 40 topic_admin: 50: 17 75: 18 90: 22 99: 32 categories: 50: 35 75: 41 90: 43 99: 77 home: 50: 39 75: 46 90: 49 99: 95 topic: 50: 46 75: 52 90: 56 99: 101 *** JIT on Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 19 75: 21 90: 25 99: 33 home_admin: 50: 24 75: 26 90: 30 99: 35 topic_admin: 50: 19 75: 20 90: 25 99: 30 categories: 50: 40 75: 44 90: 48 99: 76 home: 50: 42 75: 48 90: 51 99: 89 topic: 50: 49 75: 55 90: 58 99: 99 git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@62197 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-02-04 14:22:28 +03:00
extern int rb_method_definition_eq(const rb_method_definition_t *d1, const rb_method_definition_t *d2);
static VALUE
(*call_cfunc_invoker_func(int argc))(VALUE recv, int argc, const VALUE *, VALUE (*func)(ANYARGS))
{
switch (argc) {
case -2: return &call_cfunc_m2;
case -1: return &call_cfunc_m1;
case 0: return &call_cfunc_0;
case 1: return &call_cfunc_1;
case 2: return &call_cfunc_2;
case 3: return &call_cfunc_3;
case 4: return &call_cfunc_4;
case 5: return &call_cfunc_5;
case 6: return &call_cfunc_6;
case 7: return &call_cfunc_7;
case 8: return &call_cfunc_8;
case 9: return &call_cfunc_9;
case 10: return &call_cfunc_10;
case 11: return &call_cfunc_11;
case 12: return &call_cfunc_12;
case 13: return &call_cfunc_13;
case 14: return &call_cfunc_14;
case 15: return &call_cfunc_15;
default:
rb_bug("call_cfunc_func: unsupported length: %d", argc);
}
}
static void
setup_method_cfunc_struct(rb_method_cfunc_t *cfunc, VALUE (*func)(), int argc)
{
cfunc->func = func;
cfunc->argc = argc;
cfunc->invoker = call_cfunc_invoker_func(argc);
}
MJIT_FUNC_EXPORTED void
rb_method_definition_set(const rb_method_entry_t *me, rb_method_definition_t *def, void *opts)
{
*(rb_method_definition_t **)&me->def = def;
if (opts != NULL) {
switch (def->type) {
case VM_METHOD_TYPE_ISEQ:
{
rb_method_iseq_t *iseq_body = (rb_method_iseq_t *)opts;
rb_cref_t *method_cref, *cref = iseq_body->cref;
/* setup iseq first (before invoking GC) */
RB_OBJ_WRITE(me, &def->body.iseq.iseqptr, iseq_body->iseqptr);
if (0) vm_cref_dump("rb_method_definition_create", cref);
if (cref) {
method_cref = cref;
}
else {
method_cref = vm_cref_new_toplevel(GET_EC()); /* TODO: can we reuse? */
}
RB_OBJ_WRITE(me, &def->body.iseq.cref, method_cref);
return;
}
case VM_METHOD_TYPE_CFUNC:
{
rb_method_cfunc_t *cfunc = (rb_method_cfunc_t *)opts;
setup_method_cfunc_struct(UNALIGNED_MEMBER_PTR(def, body.cfunc), cfunc->func, cfunc->argc);
return;
}
case VM_METHOD_TYPE_ATTRSET:
case VM_METHOD_TYPE_IVAR:
{
const rb_execution_context_t *ec = GET_EC();
rb_control_frame_t *cfp;
int line;
def->body.attr.id = (ID)(VALUE)opts;
cfp = rb_vm_get_ruby_level_next_cfp(ec, ec->cfp);
if (cfp && (line = rb_vm_get_sourceline(cfp))) {
VALUE location = rb_ary_new3(2, rb_iseq_path(cfp->iseq), INT2FIX(line));
RB_OBJ_WRITE(me, &def->body.attr.location, rb_ary_freeze(location));
}
else {
VM_ASSERT(def->body.attr.location == 0);
}
return;
}
case VM_METHOD_TYPE_BMETHOD:
RB_OBJ_WRITE(me, &def->body.bmethod.proc, (VALUE)opts);
return;
case VM_METHOD_TYPE_NOTIMPLEMENTED:
setup_method_cfunc_struct(UNALIGNED_MEMBER_PTR(def, body.cfunc), rb_f_notimplement, -1);
return;
case VM_METHOD_TYPE_OPTIMIZED:
def->body.optimize_type = (enum method_optimized_type)(intptr_t)opts;
return;
case VM_METHOD_TYPE_REFINED:
{
const rb_method_refined_t *refined = (rb_method_refined_t *)opts;
RB_OBJ_WRITE(me, &def->body.refined.orig_me, refined->orig_me);
RB_OBJ_WRITE(me, &def->body.refined.owner, refined->owner);
return;
}
case VM_METHOD_TYPE_ALIAS:
RB_OBJ_WRITE(me, &def->body.alias.original_me, (rb_method_entry_t *)opts);
return;
case VM_METHOD_TYPE_ZSUPER:
case VM_METHOD_TYPE_UNDEF:
case VM_METHOD_TYPE_MISSING:
return;
}
}
}
static void
method_definition_reset(const rb_method_entry_t *me)
{
rb_method_definition_t *def = me->def;
switch(def->type) {
case VM_METHOD_TYPE_ISEQ:
2015-07-22 01:52:59 +03:00
RB_OBJ_WRITTEN(me, Qundef, def->body.iseq.iseqptr);
RB_OBJ_WRITTEN(me, Qundef, def->body.iseq.cref);
break;
case VM_METHOD_TYPE_ATTRSET:
case VM_METHOD_TYPE_IVAR:
RB_OBJ_WRITTEN(me, Qundef, def->body.attr.location);
break;
case VM_METHOD_TYPE_BMETHOD:
RB_OBJ_WRITTEN(me, Qundef, def->body.bmethod.proc);
Support targetting TracePoint [Feature #15289] * vm_trace.c (rb_tracepoint_enable_for_target): support targetting TracePoint. [Feature #15289] Tragetting TracePoint is only enabled on specified method, proc and so on, example: `tp.enable(target: code)`. `code` should be consisted of InstructionSeuqnece (iseq) (RubyVM::InstructionSeuqnece.of(code) should not return nil) If code is a tree of iseq, TracePoint is enabled on all of iseqs in a tree. Enabled tragetting TracePoints can not enabled again with and without target. * vm_core.h (rb_iseq_t): introduce `rb_iseq_t::local_hooks` to store local hooks. `rb_iseq_t::aux::trace_events` is renamed to `global_trace_events` to contrast with `local_hooks`. * vm_core.h (rb_hook_list_t): add `rb_hook_list_t::running` to represent how many Threads/Fibers are used this list. If this field is 0, nobody using this hooks and we can delete it. This is why we can remove code from cont.c. * vm_core.h (rb_vm_t): because of above change, we can eliminate `rb_vm_t::trace_running` field. Also renamed from `rb_vm_t::event_hooks` to `global_hooks`. * vm_core.h, vm.c (ruby_vm_event_enabled_global_flags): renamed from `ruby_vm_event_enabled_flags. * vm_core.h, vm.c (ruby_vm_event_local_num): added to count enabled targetting TracePoints. * vm_core.h, vm_trace.c (rb_exec_event_hooks): accepts hook list. * vm_core.h (rb_vm_global_hooks): added for convinience. * method.h (rb_method_bmethod_t): added to maintain Proc and `rb_hook_list_t` for bmethod (defined by define_method). * prelude.rb (TracePoint#enable): extracet a keyword parameter (because it is easy than writing in C). It calls `TracePoint#__enable` internal method written in C. * vm_insnhelper.c (vm_trace): check also iseq->local_hooks. * vm.c (invoke_bmethod): check def->body.bmethod.hooks. * vm.c (hook_before_rewind): check iseq->local_hooks and def->body.bmethod.hooks before rewind by exception. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@66003 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-11-26 21:16:39 +03:00
/* give up to check all in a list */
if (def->body.bmethod.hooks) rb_gc_writebarrier_remember((VALUE)me);
break;
case VM_METHOD_TYPE_REFINED:
RB_OBJ_WRITTEN(me, Qundef, def->body.refined.orig_me);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
RB_OBJ_WRITTEN(me, Qundef, def->body.refined.owner);
break;
case VM_METHOD_TYPE_ALIAS:
RB_OBJ_WRITTEN(me, Qundef, def->body.alias.original_me);
break;
case VM_METHOD_TYPE_CFUNC:
case VM_METHOD_TYPE_ZSUPER:
case VM_METHOD_TYPE_MISSING:
case VM_METHOD_TYPE_OPTIMIZED:
case VM_METHOD_TYPE_UNDEF:
case VM_METHOD_TYPE_NOTIMPLEMENTED:
break;
}
}
MJIT_FUNC_EXPORTED rb_method_definition_t *
rb_method_definition_create(rb_method_type_t type, ID mid)
{
rb_method_definition_t *def;
def = ZALLOC(rb_method_definition_t);
def->type = type;
def->original_id = mid;
static uintptr_t method_serial = 1;
def->method_serial = method_serial++;
return def;
}
static rb_method_definition_t *
method_definition_addref(rb_method_definition_t *def)
{
def->alias_count++;
if (METHOD_DEBUG) fprintf(stderr, "+%p-%s:%d\n", (void *)def, rb_id2name(def->original_id), def->alias_count);
return def;
}
static rb_method_definition_t *
method_definition_addref_complement(rb_method_definition_t *def)
{
def->complemented_count++;
if (METHOD_DEBUG) fprintf(stderr, "+%p-%s:%d\n", (void *)def, rb_id2name(def->original_id), def->complemented_count);
return def;
}
static rb_method_entry_t *
rb_method_entry_alloc(ID called_id, VALUE owner, VALUE defined_class, const rb_method_definition_t *def)
{
rb_method_entry_t *me = (rb_method_entry_t *)rb_imemo_new(imemo_ment, (VALUE)def, (VALUE)called_id, owner, defined_class);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
return me;
}
static VALUE
filter_defined_class(VALUE klass)
{
switch (BUILTIN_TYPE(klass)) {
case T_CLASS:
return klass;
case T_MODULE:
return 0;
case T_ICLASS:
break;
}
rb_bug("filter_defined_class: %s", rb_obj_info(klass));
}
rb_method_entry_t *
rb_method_entry_create(ID called_id, VALUE klass, rb_method_visibility_t visi, const rb_method_definition_t *def)
{
rb_method_entry_t *me = rb_method_entry_alloc(called_id, klass, filter_defined_class(klass), def);
METHOD_ENTRY_FLAGS_SET(me, visi, ruby_running ? FALSE : TRUE);
if (def != NULL) method_definition_reset(me);
return me;
}
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
const rb_method_entry_t *
rb_method_entry_clone(const rb_method_entry_t *src_me)
{
rb_method_entry_t *me = rb_method_entry_alloc(src_me->called_id, src_me->owner, src_me->defined_class,
method_definition_addref(src_me->def));
if (METHOD_ENTRY_COMPLEMENTED(src_me)) {
method_definition_addref_complement(src_me->def);
}
METHOD_ENTRY_FLAGS_COPY(me, src_me);
return me;
}
mjit_compile.c: merge initial JIT compiler which has been developed by Takashi Kokubun <takashikkbn@gmail> as YARV-MJIT. Many of its bugs are fixed by wanabe <s.wanabe@gmail.com>. This JIT compiler is designed to be a safe migration path to introduce JIT compiler to MRI. So this commit does not include any bytecode changes or dynamic instruction modifications, which are done in original MJIT. This commit even strips off some aggressive optimizations from YARV-MJIT, and thus it's slower than YARV-MJIT too. But it's still fairly faster than Ruby 2.5 in some benchmarks (attached below). Note that this JIT compiler passes `make test`, `make test-all`, `make test-spec` without JIT, and even with JIT. Not only it's perfectly safe with JIT disabled because it does not replace VM instructions unlike MJIT, but also with JIT enabled it stably runs Ruby applications including Rails applications. I'm expecting this version as just "initial" JIT compiler. I have many optimization ideas which are skipped for initial merging, and you may easily replace this JIT compiler with a faster one by just replacing mjit_compile.c. `mjit_compile` interface is designed for the purpose. common.mk: update dependencies for mjit_compile.c. internal.h: declare `rb_vm_insn_addr2insn` for MJIT. vm.c: exclude some definitions if `-DMJIT_HEADER` is provided to compiler. This avoids to include some functions which take a long time to compile, e.g. vm_exec_core. Some of the purpose is achieved in transform_mjit_header.rb (see `IGNORED_FUNCTIONS`) but others are manually resolved for now. Load mjit_helper.h for MJIT header. mjit_helper.h: New. This is a file used only by JIT-ed code. I'll refactor `mjit_call_cfunc` later. vm_eval.c: add some #ifdef switches to skip compiling some functions like Init_vm_eval. win32/mkexports.rb: export thread/ec functions, which are used by MJIT. include/ruby/defines.h: add MJIT_FUNC_EXPORTED macro alis to clarify that a function is exported only for MJIT. array.c: export a function used by MJIT. bignum.c: ditto. class.c: ditto. compile.c: ditto. error.c: ditto. gc.c: ditto. hash.c: ditto. iseq.c: ditto. numeric.c: ditto. object.c: ditto. proc.c: ditto. re.c: ditto. st.c: ditto. string.c: ditto. thread.c: ditto. variable.c: ditto. vm_backtrace.c: ditto. vm_insnhelper.c: ditto. vm_method.c: ditto. I would like to improve maintainability of function exports, but I believe this way is acceptable as initial merging if we clarify the new exports are for MJIT (so that we can use them as TODO list to fix) and add unit tests to detect unresolved symbols. I'll add unit tests of JIT compilations in succeeding commits. Author: Takashi Kokubun <takashikkbn@gmail.com> Contributor: wanabe <s.wanabe@gmail.com> Part of [Feature #14235] --- * Known issues * Code generated by gcc is faster than clang. The benchmark may be worse in macOS. Following benchmark result is provided by gcc w/ Linux. * Performance is decreased when Google Chrome is running * JIT can work on MinGW, but it doesn't improve performance at least in short running benchmark. * Currently it doesn't perform well with Rails. We'll try to fix this before release. --- * Benchmark reslts Benchmarked with: Intel 4.0GHz i7-4790K with 16GB memory under x86-64 Ubuntu 8 Cores - 2.0.0-p0: Ruby 2.0.0-p0 - r62186: Ruby trunk (early 2.6.0), before MJIT changes - JIT off: On this commit, but without `--jit` option - JIT on: On this commit, and with `--jit` option ** Optcarrot fps Benchmark: https://github.com/mame/optcarrot | |2.0.0-p0 |r62186 |JIT off |JIT on | |:--------|:--------|:--------|:--------|:--------| |fps |37.32 |51.46 |51.31 |58.88 | |vs 2.0.0 |1.00x |1.38x |1.37x |1.58x | ** MJIT benchmarks Benchmark: https://github.com/benchmark-driver/mjit-benchmarks (Original: https://github.com/vnmakarov/ruby/tree/rtl_mjit_branch/MJIT-benchmarks) | |2.0.0-p0 |r62186 |JIT off |JIT on | |:----------|:--------|:--------|:--------|:--------| |aread |1.00 |1.09 |1.07 |2.19 | |aref |1.00 |1.13 |1.11 |2.22 | |aset |1.00 |1.50 |1.45 |2.64 | |awrite |1.00 |1.17 |1.13 |2.20 | |call |1.00 |1.29 |1.26 |2.02 | |const2 |1.00 |1.10 |1.10 |2.19 | |const |1.00 |1.11 |1.10 |2.19 | |fannk |1.00 |1.04 |1.02 |1.00 | |fib |1.00 |1.32 |1.31 |1.84 | |ivread |1.00 |1.13 |1.12 |2.43 | |ivwrite |1.00 |1.23 |1.21 |2.40 | |mandelbrot |1.00 |1.13 |1.16 |1.28 | |meteor |1.00 |2.97 |2.92 |3.17 | |nbody |1.00 |1.17 |1.15 |1.49 | |nest-ntimes|1.00 |1.22 |1.20 |1.39 | |nest-while |1.00 |1.10 |1.10 |1.37 | |norm |1.00 |1.18 |1.16 |1.24 | |nsvb |1.00 |1.16 |1.16 |1.17 | |red-black |1.00 |1.02 |0.99 |1.12 | |sieve |1.00 |1.30 |1.28 |1.62 | |trees |1.00 |1.14 |1.13 |1.19 | |while |1.00 |1.12 |1.11 |2.41 | ** Discourse's script/bench.rb Benchmark: https://github.com/discourse/discourse/blob/v1.8.7/script/bench.rb NOTE: Rails performance was somehow a little degraded with JIT for now. We should fix this. (At least I know opt_aref is performing badly in JIT and I have an idea to fix it. Please wait for the fix.) *** JIT off Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 17 75: 18 90: 22 99: 29 home_admin: 50: 21 75: 21 90: 27 99: 40 topic_admin: 50: 17 75: 18 90: 22 99: 32 categories: 50: 35 75: 41 90: 43 99: 77 home: 50: 39 75: 46 90: 49 99: 95 topic: 50: 46 75: 52 90: 56 99: 101 *** JIT on Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 19 75: 21 90: 25 99: 33 home_admin: 50: 24 75: 26 90: 30 99: 35 topic_admin: 50: 19 75: 20 90: 25 99: 30 categories: 50: 40 75: 44 90: 48 99: 76 home: 50: 42 75: 48 90: 51 99: 89 topic: 50: 49 75: 55 90: 58 99: 99 git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@62197 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-02-04 14:22:28 +03:00
MJIT_FUNC_EXPORTED const rb_callable_method_entry_t *
rb_method_entry_complement_defined_class(const rb_method_entry_t *src_me, ID called_id, VALUE defined_class)
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
{
rb_method_definition_t *def = src_me->def;
rb_method_entry_t *me;
struct {
const struct rb_method_entry_struct *orig_me;
VALUE owner;
} refined = {0};
if (!src_me->defined_class &&
def->type == VM_METHOD_TYPE_REFINED &&
def->body.refined.orig_me) {
const rb_method_entry_t *orig_me =
rb_method_entry_clone(def->body.refined.orig_me);
RB_OBJ_WRITE((VALUE)orig_me, &orig_me->defined_class, defined_class);
refined.orig_me = orig_me;
refined.owner = orig_me->owner;
def = NULL;
}
else {
def = method_definition_addref_complement(def);
}
me = rb_method_entry_alloc(called_id, src_me->owner, defined_class, def);
METHOD_ENTRY_FLAGS_COPY(me, src_me);
METHOD_ENTRY_COMPLEMENTED_SET(me);
if (!def) {
def = rb_method_definition_create(VM_METHOD_TYPE_REFINED, called_id);
rb_method_definition_set(me, def, &refined);
}
VM_ASSERT(RB_TYPE_P(me->owner, T_MODULE));
return (rb_callable_method_entry_t *)me;
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
}
void
rb_method_entry_copy(rb_method_entry_t *dst, const rb_method_entry_t *src)
{
*(rb_method_definition_t **)&dst->def = method_definition_addref(src->def);
method_definition_reset(dst);
dst->called_id = src->called_id;
RB_OBJ_WRITE((VALUE)dst, &dst->owner, src->owner);
RB_OBJ_WRITE((VALUE)dst, &dst->defined_class, src->defined_class);
METHOD_ENTRY_FLAGS_COPY(dst, src);
}
static void
make_method_entry_refined(VALUE owner, rb_method_entry_t *me)
* 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
{
if (me->def->type == VM_METHOD_TYPE_REFINED) {
return;
}
else {
struct {
struct rb_method_entry_struct *orig_me;
VALUE owner;
} refined;
rb_method_definition_t *def;
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
rb_vm_check_redefinition_opt_method(me, me->owner);
refined.orig_me =
rb_method_entry_alloc(me->called_id, me->owner,
me->defined_class ?
me->defined_class : owner,
method_definition_addref(me->def));
METHOD_ENTRY_FLAGS_COPY(refined.orig_me, me);
refined.owner = owner;
def = rb_method_definition_create(VM_METHOD_TYPE_REFINED, me->called_id);
rb_method_definition_set(me, def, (void *)&refined);
METHOD_ENTRY_VISI_SET(me, METHOD_VISI_PUBLIC);
}
* 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
}
static inline rb_method_entry_t *
lookup_method_table(VALUE klass, ID id)
{
st_data_t body;
struct rb_id_table *m_tbl = RCLASS_M_TBL(klass);
if (rb_id_table_lookup(m_tbl, id, &body)) {
return (rb_method_entry_t *) body;
}
else {
return 0;
}
}
* 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
void
rb_add_refined_method_entry(VALUE refined_class, ID mid)
{
rb_method_entry_t *me = lookup_method_table(refined_class, mid);
* 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
if (me) {
make_method_entry_refined(refined_class, me);
rb_clear_method_cache(refined_class, mid);
* 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
}
else {
rb_add_method(refined_class, mid, VM_METHOD_TYPE_REFINED, 0, METHOD_VISI_PUBLIC);
* 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
}
}
static void
check_override_opt_method_i(VALUE klass, VALUE arg)
{
ID mid = (ID)arg;
const rb_method_entry_t *me, *newme;
if (vm_redefinition_check_flag(klass)) {
me = lookup_method_table(RCLASS_ORIGIN(klass), mid);
if (me) {
newme = rb_method_entry(klass, mid);
if (newme != me) rb_vm_check_redefinition_opt_method(me, me->owner);
}
}
rb_class_foreach_subclass(klass, check_override_opt_method_i, (VALUE)mid);
}
static void
check_override_opt_method(VALUE klass, VALUE mid)
{
if (rb_vm_check_optimizable_mid(mid)) {
check_override_opt_method_i(klass, mid);
}
}
/*
* klass->method_table[mid] = method_entry(defined_class, visi, def)
*
* If def is given (!= NULL), then just use it and ignore original_id and otps.
* If not given, then make a new def with original_id and opts.
*/
static rb_method_entry_t *
rb_method_entry_make(VALUE klass, ID mid, VALUE defined_class, rb_method_visibility_t visi,
rb_method_type_t type, rb_method_definition_t *def, ID original_id, void *opts)
{
rb_method_entry_t *me;
struct rb_id_table *mtbl;
st_data_t data;
* 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
int make_refined = 0;
if (NIL_P(klass)) {
klass = rb_cObject;
}
if (!FL_TEST(klass, FL_SINGLETON) &&
type != VM_METHOD_TYPE_NOTIMPLEMENTED &&
type != VM_METHOD_TYPE_ZSUPER) {
switch (mid) {
case idInitialize:
case idInitialize_copy:
case idInitialize_clone:
case idInitialize_dup:
case idRespond_to_missing:
visi = METHOD_VISI_PRIVATE;
}
}
rb_class_modify_check(klass);
if (FL_TEST(klass, RMODULE_IS_REFINEMENT)) {
VALUE refined_class = rb_refinement_module_get_refined_class(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
rb_add_refined_method_entry(refined_class, mid);
}
if (type == VM_METHOD_TYPE_REFINED) {
rb_method_entry_t *old_me = lookup_method_table(RCLASS_ORIGIN(klass), mid);
* 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
if (old_me) rb_vm_check_redefinition_opt_method(old_me, klass);
}
else {
klass = RCLASS_ORIGIN(klass);
}
mtbl = RCLASS_M_TBL(klass);
/* check re-definition */
if (rb_id_table_lookup(mtbl, mid, &data)) {
rb_method_entry_t *old_me = (rb_method_entry_t *)data;
rb_method_definition_t *old_def = old_me->def;
if (rb_method_definition_eq(old_def, def)) return old_me;
rb_vm_check_redefinition_opt_method(old_me, klass);
if (old_def->type == VM_METHOD_TYPE_REFINED) make_refined = 1;
if (RTEST(ruby_verbose) &&
type != VM_METHOD_TYPE_UNDEF &&
(old_def->alias_count == 0) &&
!make_refined &&
old_def->type != VM_METHOD_TYPE_UNDEF &&
old_def->type != VM_METHOD_TYPE_ZSUPER &&
old_def->type != VM_METHOD_TYPE_ALIAS) {
const rb_iseq_t *iseq = 0;
rb_warning("method redefined; discarding old %"PRIsVALUE, rb_id2str(mid));
switch (old_def->type) {
case VM_METHOD_TYPE_ISEQ:
iseq = def_iseq_ptr(old_def);
break;
case VM_METHOD_TYPE_BMETHOD:
iseq = rb_proc_get_iseq(old_def->body.bmethod.proc, 0);
break;
default:
break;
}
if (iseq) {
rb_compile_warning(RSTRING_PTR(rb_iseq_path(iseq)),
FIX2INT(iseq->body->location.first_lineno),
"previous definition of %"PRIsVALUE" was here",
rb_id2str(old_def->original_id));
}
}
}
/* create method entry */
me = rb_method_entry_create(mid, defined_class, visi, NULL);
if (def == NULL) def = rb_method_definition_create(type, original_id);
rb_method_definition_set(me, def, opts);
rb_clear_method_cache(klass, mid);
/* check mid */
if (klass == rb_cObject) {
switch (mid) {
case idInitialize:
case idRespond_to_missing:
case idMethodMissing:
case idRespond_to:
rb_warn("redefining Object#%s may cause infinite loop", rb_id2name(mid));
}
}
/* check mid */
if (mid == object_id || mid == id__send__) {
if (type == VM_METHOD_TYPE_ISEQ && search_method(klass, mid, 0)) {
rb_warn("redefining `%s' may cause serious problems", rb_id2name(mid));
}
}
* 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
if (make_refined) {
make_method_entry_refined(klass, me);
* 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
}
rb_id_table_insert(mtbl, mid, (VALUE)me);
RB_OBJ_WRITTEN(klass, Qundef, (VALUE)me);
VM_ASSERT(me->def != NULL);
/* check optimized method override by a prepended module */
if (RB_TYPE_P(klass, T_MODULE)) {
check_override_opt_method(klass, (VALUE)mid);
}
return me;
}
#define CALL_METHOD_HOOK(klass, hook, mid) do { \
const VALUE arg = ID2SYM(mid); \
VALUE recv_class = (klass); \
ID hook_id = (hook); \
if (FL_TEST((klass), FL_SINGLETON)) { \
recv_class = rb_ivar_get((klass), attached); \
hook_id = singleton_##hook; \
} \
rb_funcallv(recv_class, hook_id, 1, &arg); \
} while (0)
static void
method_added(VALUE klass, ID mid)
{
if (ruby_running) {
CALL_METHOD_HOOK(klass, added, mid);
}
}
void
rb_add_method(VALUE klass, ID mid, rb_method_type_t type, void *opts, rb_method_visibility_t visi)
{
rb_method_entry_make(klass, mid, klass, visi, type, NULL, mid, opts);
* 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
if (type != VM_METHOD_TYPE_UNDEF && type != VM_METHOD_TYPE_REFINED) {
method_added(klass, mid);
}
}
MJIT_FUNC_EXPORTED void
2015-07-22 01:52:59 +03:00
rb_add_method_iseq(VALUE klass, ID mid, const rb_iseq_t *iseq, rb_cref_t *cref, rb_method_visibility_t visi)
{
struct { /* should be same fields with rb_method_iseq_struct */
const rb_iseq_t *iseqptr;
rb_cref_t *cref;
} iseq_body;
iseq_body.iseqptr = iseq;
iseq_body.cref = cref;
rb_add_method(klass, mid, VM_METHOD_TYPE_ISEQ, &iseq_body, visi);
}
static rb_method_entry_t *
method_entry_set(VALUE klass, ID mid, const rb_method_entry_t *me,
rb_method_visibility_t visi, VALUE defined_class)
{
rb_method_entry_t *newme = rb_method_entry_make(klass, mid, defined_class, visi,
me->def->type, method_definition_addref(me->def), 0, NULL);
method_added(klass, mid);
return newme;
}
rb_method_entry_t *
rb_method_entry_set(VALUE klass, ID mid, const rb_method_entry_t *me, rb_method_visibility_t visi)
{
return method_entry_set(klass, mid, me, visi, klass);
}
#define UNDEF_ALLOC_FUNC ((rb_alloc_func_t)-1)
void
rb_define_alloc_func(VALUE klass, VALUE (*func)(VALUE))
{
Check_Type(klass, T_CLASS);
RCLASS_EXT(klass)->allocator = func;
}
void
rb_undef_alloc_func(VALUE klass)
{
rb_define_alloc_func(klass, UNDEF_ALLOC_FUNC);
}
rb_alloc_func_t
rb_get_alloc_func(VALUE klass)
{
Check_Type(klass, T_CLASS);
for (; klass; klass = RCLASS_SUPER(klass)) {
rb_alloc_func_t allocator = RCLASS_EXT(klass)->allocator;
if (allocator == UNDEF_ALLOC_FUNC) break;
if (allocator) return allocator;
}
return 0;
}
const rb_method_entry_t *
rb_method_entry_at(VALUE klass, ID id)
{
return lookup_method_table(klass, id);
}
static inline rb_method_entry_t*
* 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
search_method(VALUE klass, ID id, VALUE *defined_class_ptr)
{
rb_method_entry_t *me = NULL;
RB_DEBUG_COUNTER_INC(mc_search);
for (; klass; klass = RCLASS_SUPER(klass)) {
RB_DEBUG_COUNTER_INC(mc_search_super);
if ((me = lookup_method_table(klass, id)) != 0) {
break;
}
}
if (defined_class_ptr) *defined_class_ptr = klass;
if (me == NULL) RB_DEBUG_COUNTER_INC(mc_search_notfound);
VM_ASSERT(me == NULL || !METHOD_ENTRY_INVALIDATED(me));
* 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
return me;
}
static rb_method_entry_t *
search_method_protect(VALUE klass, ID id, VALUE *defined_class_ptr)
{
rb_method_entry_t *me = search_method(klass, id, defined_class_ptr);
if (!UNDEFINED_METHOD_ENTRY_P(me)) {
return me;
}
else {
return NULL;
}
}
MJIT_FUNC_EXPORTED const rb_method_entry_t *
rb_method_entry(VALUE klass, ID id)
{
return search_method(klass, id, NULL);
}
static inline const rb_callable_method_entry_t *
prepare_callable_method_entry(VALUE defined_class, ID id, const rb_method_entry_t * const me, int create)
{
struct rb_id_table *mtbl;
const rb_callable_method_entry_t *cme;
if (me) {
if (me->defined_class == 0) {
RB_DEBUG_COUNTER_INC(mc_cme_complement);
VM_ASSERT(RB_TYPE_P(defined_class, T_ICLASS) || RB_TYPE_P(defined_class, T_MODULE));
VM_ASSERT(me->defined_class == 0);
mtbl = RCLASS_CALLABLE_M_TBL(defined_class);
if (mtbl && rb_id_table_lookup(mtbl, id, (VALUE *)&cme)) {
RB_DEBUG_COUNTER_INC(mc_cme_complement_hit);
VM_ASSERT(callable_method_entry_p(cme));
VM_ASSERT(!METHOD_ENTRY_INVALIDATED(cme));
}
else if (create) {
if (!mtbl) {
mtbl = RCLASS_EXT(defined_class)->callable_m_tbl = rb_id_table_create(0);
}
cme = rb_method_entry_complement_defined_class(me, me->called_id, defined_class);
rb_id_table_insert(mtbl, id, (VALUE)cme);
VM_ASSERT(callable_method_entry_p(cme));
}
else {
return NULL;
}
}
else {
cme = (const rb_callable_method_entry_t *)me;
VM_ASSERT(callable_method_entry_p(cme));
VM_ASSERT(!METHOD_ENTRY_INVALIDATED(cme));
}
return cme;
}
else {
return NULL;
}
}
static const rb_callable_method_entry_t *
copmplemented_callable_method_entry(VALUE klass, ID id)
{
VALUE defined_class;
rb_method_entry_t *me = search_method_protect(klass, id, &defined_class);
return prepare_callable_method_entry(defined_class, id, me, FALSE);
}
static const rb_callable_method_entry_t *
cached_callable_method_entry(VALUE klass, ID mid)
{
struct rb_id_table *cc_tbl = RCLASS_CC_TBL(klass);
struct rb_class_cc_entries *ccs;
if (cc_tbl && rb_id_table_lookup(cc_tbl, mid, (VALUE *)&ccs)) {
VM_ASSERT(vm_ccs_p(ccs));
if (LIKELY(!METHOD_ENTRY_INVALIDATED(ccs->cme))) {
VM_ASSERT(ccs->cme->called_id == mid);
RB_DEBUG_COUNTER_INC(ccs_found);
return ccs->cme;
}
else {
rb_vm_ccs_free(ccs);
rb_id_table_delete(cc_tbl, mid);
}
}
return NULL;
}
static void
cache_callable_method_entry(VALUE klass, ID mid, const rb_callable_method_entry_t *cme)
{
struct rb_id_table *cc_tbl = RCLASS_CC_TBL(klass);
struct rb_class_cc_entries *ccs;
if (!cc_tbl) {
cc_tbl = RCLASS_CC_TBL(klass) = rb_id_table_create(2);
}
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
if (rb_id_table_lookup(cc_tbl, mid, (VALUE *)&ccs)) {
VM_ASSERT(ccs->cme == cme);
}
else {
ccs = vm_ccs_create(klass, cme);
rb_id_table_insert(cc_tbl, mid, (VALUE)ccs);
}
* 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
}
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
static const rb_callable_method_entry_t *
callable_method_entry(VALUE klass, ID mid, VALUE *defined_class_ptr)
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
{
VM_ASSERT(RB_TYPE_P(klass, T_CLASS) || RB_TYPE_P(klass, T_ICLASS));
const rb_callable_method_entry_t *cme = cached_callable_method_entry(klass, mid);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
if (cme) {
if (defined_class_ptr != NULL) *defined_class_ptr = cme->defined_class;
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
}
else {
VALUE defined_class;
rb_method_entry_t *me = search_method_protect(klass, mid, &defined_class);
if (defined_class_ptr) *defined_class_ptr = defined_class;
cme = prepare_callable_method_entry(defined_class, mid, me, TRUE);
if (cme) cache_callable_method_entry(klass, mid, cme);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
}
return cme;
}
mjit_compile.c: merge initial JIT compiler which has been developed by Takashi Kokubun <takashikkbn@gmail> as YARV-MJIT. Many of its bugs are fixed by wanabe <s.wanabe@gmail.com>. This JIT compiler is designed to be a safe migration path to introduce JIT compiler to MRI. So this commit does not include any bytecode changes or dynamic instruction modifications, which are done in original MJIT. This commit even strips off some aggressive optimizations from YARV-MJIT, and thus it's slower than YARV-MJIT too. But it's still fairly faster than Ruby 2.5 in some benchmarks (attached below). Note that this JIT compiler passes `make test`, `make test-all`, `make test-spec` without JIT, and even with JIT. Not only it's perfectly safe with JIT disabled because it does not replace VM instructions unlike MJIT, but also with JIT enabled it stably runs Ruby applications including Rails applications. I'm expecting this version as just "initial" JIT compiler. I have many optimization ideas which are skipped for initial merging, and you may easily replace this JIT compiler with a faster one by just replacing mjit_compile.c. `mjit_compile` interface is designed for the purpose. common.mk: update dependencies for mjit_compile.c. internal.h: declare `rb_vm_insn_addr2insn` for MJIT. vm.c: exclude some definitions if `-DMJIT_HEADER` is provided to compiler. This avoids to include some functions which take a long time to compile, e.g. vm_exec_core. Some of the purpose is achieved in transform_mjit_header.rb (see `IGNORED_FUNCTIONS`) but others are manually resolved for now. Load mjit_helper.h for MJIT header. mjit_helper.h: New. This is a file used only by JIT-ed code. I'll refactor `mjit_call_cfunc` later. vm_eval.c: add some #ifdef switches to skip compiling some functions like Init_vm_eval. win32/mkexports.rb: export thread/ec functions, which are used by MJIT. include/ruby/defines.h: add MJIT_FUNC_EXPORTED macro alis to clarify that a function is exported only for MJIT. array.c: export a function used by MJIT. bignum.c: ditto. class.c: ditto. compile.c: ditto. error.c: ditto. gc.c: ditto. hash.c: ditto. iseq.c: ditto. numeric.c: ditto. object.c: ditto. proc.c: ditto. re.c: ditto. st.c: ditto. string.c: ditto. thread.c: ditto. variable.c: ditto. vm_backtrace.c: ditto. vm_insnhelper.c: ditto. vm_method.c: ditto. I would like to improve maintainability of function exports, but I believe this way is acceptable as initial merging if we clarify the new exports are for MJIT (so that we can use them as TODO list to fix) and add unit tests to detect unresolved symbols. I'll add unit tests of JIT compilations in succeeding commits. Author: Takashi Kokubun <takashikkbn@gmail.com> Contributor: wanabe <s.wanabe@gmail.com> Part of [Feature #14235] --- * Known issues * Code generated by gcc is faster than clang. The benchmark may be worse in macOS. Following benchmark result is provided by gcc w/ Linux. * Performance is decreased when Google Chrome is running * JIT can work on MinGW, but it doesn't improve performance at least in short running benchmark. * Currently it doesn't perform well with Rails. We'll try to fix this before release. --- * Benchmark reslts Benchmarked with: Intel 4.0GHz i7-4790K with 16GB memory under x86-64 Ubuntu 8 Cores - 2.0.0-p0: Ruby 2.0.0-p0 - r62186: Ruby trunk (early 2.6.0), before MJIT changes - JIT off: On this commit, but without `--jit` option - JIT on: On this commit, and with `--jit` option ** Optcarrot fps Benchmark: https://github.com/mame/optcarrot | |2.0.0-p0 |r62186 |JIT off |JIT on | |:--------|:--------|:--------|:--------|:--------| |fps |37.32 |51.46 |51.31 |58.88 | |vs 2.0.0 |1.00x |1.38x |1.37x |1.58x | ** MJIT benchmarks Benchmark: https://github.com/benchmark-driver/mjit-benchmarks (Original: https://github.com/vnmakarov/ruby/tree/rtl_mjit_branch/MJIT-benchmarks) | |2.0.0-p0 |r62186 |JIT off |JIT on | |:----------|:--------|:--------|:--------|:--------| |aread |1.00 |1.09 |1.07 |2.19 | |aref |1.00 |1.13 |1.11 |2.22 | |aset |1.00 |1.50 |1.45 |2.64 | |awrite |1.00 |1.17 |1.13 |2.20 | |call |1.00 |1.29 |1.26 |2.02 | |const2 |1.00 |1.10 |1.10 |2.19 | |const |1.00 |1.11 |1.10 |2.19 | |fannk |1.00 |1.04 |1.02 |1.00 | |fib |1.00 |1.32 |1.31 |1.84 | |ivread |1.00 |1.13 |1.12 |2.43 | |ivwrite |1.00 |1.23 |1.21 |2.40 | |mandelbrot |1.00 |1.13 |1.16 |1.28 | |meteor |1.00 |2.97 |2.92 |3.17 | |nbody |1.00 |1.17 |1.15 |1.49 | |nest-ntimes|1.00 |1.22 |1.20 |1.39 | |nest-while |1.00 |1.10 |1.10 |1.37 | |norm |1.00 |1.18 |1.16 |1.24 | |nsvb |1.00 |1.16 |1.16 |1.17 | |red-black |1.00 |1.02 |0.99 |1.12 | |sieve |1.00 |1.30 |1.28 |1.62 | |trees |1.00 |1.14 |1.13 |1.19 | |while |1.00 |1.12 |1.11 |2.41 | ** Discourse's script/bench.rb Benchmark: https://github.com/discourse/discourse/blob/v1.8.7/script/bench.rb NOTE: Rails performance was somehow a little degraded with JIT for now. We should fix this. (At least I know opt_aref is performing badly in JIT and I have an idea to fix it. Please wait for the fix.) *** JIT off Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 17 75: 18 90: 22 99: 29 home_admin: 50: 21 75: 21 90: 27 99: 40 topic_admin: 50: 17 75: 18 90: 22 99: 32 categories: 50: 35 75: 41 90: 43 99: 77 home: 50: 39 75: 46 90: 49 99: 95 topic: 50: 46 75: 52 90: 56 99: 101 *** JIT on Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 19 75: 21 90: 25 99: 33 home_admin: 50: 24 75: 26 90: 30 99: 35 topic_admin: 50: 19 75: 20 90: 25 99: 30 categories: 50: 40 75: 44 90: 48 99: 76 home: 50: 42 75: 48 90: 51 99: 89 topic: 50: 49 75: 55 90: 58 99: 99 git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@62197 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-02-04 14:22:28 +03:00
MJIT_FUNC_EXPORTED const rb_callable_method_entry_t *
rb_callable_method_entry(VALUE klass, ID mid)
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
{
return callable_method_entry(klass, mid, NULL);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
}
static const rb_method_entry_t *resolve_refined_method(VALUE refinements, const rb_method_entry_t *me, VALUE *defined_class_ptr);
static const rb_method_entry_t *
method_entry_resolve_refinement(VALUE klass, ID id, int with_refinement, VALUE *defined_class_ptr)
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
{
const rb_method_entry_t *me = search_method_protect(klass, id, defined_class_ptr);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
if (me) {
if (me->def->type == VM_METHOD_TYPE_REFINED) {
if (with_refinement) {
const rb_cref_t *cref = rb_vm_cref();
VALUE refinements = cref ? CREF_REFINEMENTS(cref) : Qnil;
me = resolve_refined_method(refinements, me, defined_class_ptr);
}
else {
me = resolve_refined_method(Qnil, me, defined_class_ptr);
}
if (UNDEFINED_METHOD_ENTRY_P(me)) me = NULL;
}
}
return me;
}
const rb_method_entry_t *
rb_method_entry_with_refinements(VALUE klass, ID id, VALUE *defined_class_ptr)
{
return method_entry_resolve_refinement(klass, id, TRUE, defined_class_ptr);
}
mjit_compile.c: merge initial JIT compiler which has been developed by Takashi Kokubun <takashikkbn@gmail> as YARV-MJIT. Many of its bugs are fixed by wanabe <s.wanabe@gmail.com>. This JIT compiler is designed to be a safe migration path to introduce JIT compiler to MRI. So this commit does not include any bytecode changes or dynamic instruction modifications, which are done in original MJIT. This commit even strips off some aggressive optimizations from YARV-MJIT, and thus it's slower than YARV-MJIT too. But it's still fairly faster than Ruby 2.5 in some benchmarks (attached below). Note that this JIT compiler passes `make test`, `make test-all`, `make test-spec` without JIT, and even with JIT. Not only it's perfectly safe with JIT disabled because it does not replace VM instructions unlike MJIT, but also with JIT enabled it stably runs Ruby applications including Rails applications. I'm expecting this version as just "initial" JIT compiler. I have many optimization ideas which are skipped for initial merging, and you may easily replace this JIT compiler with a faster one by just replacing mjit_compile.c. `mjit_compile` interface is designed for the purpose. common.mk: update dependencies for mjit_compile.c. internal.h: declare `rb_vm_insn_addr2insn` for MJIT. vm.c: exclude some definitions if `-DMJIT_HEADER` is provided to compiler. This avoids to include some functions which take a long time to compile, e.g. vm_exec_core. Some of the purpose is achieved in transform_mjit_header.rb (see `IGNORED_FUNCTIONS`) but others are manually resolved for now. Load mjit_helper.h for MJIT header. mjit_helper.h: New. This is a file used only by JIT-ed code. I'll refactor `mjit_call_cfunc` later. vm_eval.c: add some #ifdef switches to skip compiling some functions like Init_vm_eval. win32/mkexports.rb: export thread/ec functions, which are used by MJIT. include/ruby/defines.h: add MJIT_FUNC_EXPORTED macro alis to clarify that a function is exported only for MJIT. array.c: export a function used by MJIT. bignum.c: ditto. class.c: ditto. compile.c: ditto. error.c: ditto. gc.c: ditto. hash.c: ditto. iseq.c: ditto. numeric.c: ditto. object.c: ditto. proc.c: ditto. re.c: ditto. st.c: ditto. string.c: ditto. thread.c: ditto. variable.c: ditto. vm_backtrace.c: ditto. vm_insnhelper.c: ditto. vm_method.c: ditto. I would like to improve maintainability of function exports, but I believe this way is acceptable as initial merging if we clarify the new exports are for MJIT (so that we can use them as TODO list to fix) and add unit tests to detect unresolved symbols. I'll add unit tests of JIT compilations in succeeding commits. Author: Takashi Kokubun <takashikkbn@gmail.com> Contributor: wanabe <s.wanabe@gmail.com> Part of [Feature #14235] --- * Known issues * Code generated by gcc is faster than clang. The benchmark may be worse in macOS. Following benchmark result is provided by gcc w/ Linux. * Performance is decreased when Google Chrome is running * JIT can work on MinGW, but it doesn't improve performance at least in short running benchmark. * Currently it doesn't perform well with Rails. We'll try to fix this before release. --- * Benchmark reslts Benchmarked with: Intel 4.0GHz i7-4790K with 16GB memory under x86-64 Ubuntu 8 Cores - 2.0.0-p0: Ruby 2.0.0-p0 - r62186: Ruby trunk (early 2.6.0), before MJIT changes - JIT off: On this commit, but without `--jit` option - JIT on: On this commit, and with `--jit` option ** Optcarrot fps Benchmark: https://github.com/mame/optcarrot | |2.0.0-p0 |r62186 |JIT off |JIT on | |:--------|:--------|:--------|:--------|:--------| |fps |37.32 |51.46 |51.31 |58.88 | |vs 2.0.0 |1.00x |1.38x |1.37x |1.58x | ** MJIT benchmarks Benchmark: https://github.com/benchmark-driver/mjit-benchmarks (Original: https://github.com/vnmakarov/ruby/tree/rtl_mjit_branch/MJIT-benchmarks) | |2.0.0-p0 |r62186 |JIT off |JIT on | |:----------|:--------|:--------|:--------|:--------| |aread |1.00 |1.09 |1.07 |2.19 | |aref |1.00 |1.13 |1.11 |2.22 | |aset |1.00 |1.50 |1.45 |2.64 | |awrite |1.00 |1.17 |1.13 |2.20 | |call |1.00 |1.29 |1.26 |2.02 | |const2 |1.00 |1.10 |1.10 |2.19 | |const |1.00 |1.11 |1.10 |2.19 | |fannk |1.00 |1.04 |1.02 |1.00 | |fib |1.00 |1.32 |1.31 |1.84 | |ivread |1.00 |1.13 |1.12 |2.43 | |ivwrite |1.00 |1.23 |1.21 |2.40 | |mandelbrot |1.00 |1.13 |1.16 |1.28 | |meteor |1.00 |2.97 |2.92 |3.17 | |nbody |1.00 |1.17 |1.15 |1.49 | |nest-ntimes|1.00 |1.22 |1.20 |1.39 | |nest-while |1.00 |1.10 |1.10 |1.37 | |norm |1.00 |1.18 |1.16 |1.24 | |nsvb |1.00 |1.16 |1.16 |1.17 | |red-black |1.00 |1.02 |0.99 |1.12 | |sieve |1.00 |1.30 |1.28 |1.62 | |trees |1.00 |1.14 |1.13 |1.19 | |while |1.00 |1.12 |1.11 |2.41 | ** Discourse's script/bench.rb Benchmark: https://github.com/discourse/discourse/blob/v1.8.7/script/bench.rb NOTE: Rails performance was somehow a little degraded with JIT for now. We should fix this. (At least I know opt_aref is performing badly in JIT and I have an idea to fix it. Please wait for the fix.) *** JIT off Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 17 75: 18 90: 22 99: 29 home_admin: 50: 21 75: 21 90: 27 99: 40 topic_admin: 50: 17 75: 18 90: 22 99: 32 categories: 50: 35 75: 41 90: 43 99: 77 home: 50: 39 75: 46 90: 49 99: 95 topic: 50: 46 75: 52 90: 56 99: 101 *** JIT on Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 19 75: 21 90: 25 99: 33 home_admin: 50: 24 75: 26 90: 30 99: 35 topic_admin: 50: 19 75: 20 90: 25 99: 30 categories: 50: 40 75: 44 90: 48 99: 76 home: 50: 42 75: 48 90: 51 99: 89 topic: 50: 49 75: 55 90: 58 99: 99 git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@62197 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-02-04 14:22:28 +03:00
MJIT_FUNC_EXPORTED const rb_callable_method_entry_t *
rb_callable_method_entry_with_refinements(VALUE klass, ID id, VALUE *defined_class_ptr)
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
{
const rb_callable_method_entry_t *cme = callable_method_entry(klass, id, defined_class_ptr);
if (cme == NULL || cme->def->type != VM_METHOD_TYPE_REFINED) {
return cme;
}
else {
VALUE defined_class, *dcp = defined_class_ptr ? defined_class_ptr : &defined_class;
const rb_method_entry_t *me = method_entry_resolve_refinement(klass, id, TRUE, dcp);
return prepare_callable_method_entry(*dcp, id, me, TRUE);
}
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
}
const rb_method_entry_t *
rb_method_entry_without_refinements(VALUE klass, ID id, VALUE *defined_class_ptr)
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
{
return method_entry_resolve_refinement(klass, id, FALSE, defined_class_ptr);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
}
mjit_compile.c: merge initial JIT compiler which has been developed by Takashi Kokubun <takashikkbn@gmail> as YARV-MJIT. Many of its bugs are fixed by wanabe <s.wanabe@gmail.com>. This JIT compiler is designed to be a safe migration path to introduce JIT compiler to MRI. So this commit does not include any bytecode changes or dynamic instruction modifications, which are done in original MJIT. This commit even strips off some aggressive optimizations from YARV-MJIT, and thus it's slower than YARV-MJIT too. But it's still fairly faster than Ruby 2.5 in some benchmarks (attached below). Note that this JIT compiler passes `make test`, `make test-all`, `make test-spec` without JIT, and even with JIT. Not only it's perfectly safe with JIT disabled because it does not replace VM instructions unlike MJIT, but also with JIT enabled it stably runs Ruby applications including Rails applications. I'm expecting this version as just "initial" JIT compiler. I have many optimization ideas which are skipped for initial merging, and you may easily replace this JIT compiler with a faster one by just replacing mjit_compile.c. `mjit_compile` interface is designed for the purpose. common.mk: update dependencies for mjit_compile.c. internal.h: declare `rb_vm_insn_addr2insn` for MJIT. vm.c: exclude some definitions if `-DMJIT_HEADER` is provided to compiler. This avoids to include some functions which take a long time to compile, e.g. vm_exec_core. Some of the purpose is achieved in transform_mjit_header.rb (see `IGNORED_FUNCTIONS`) but others are manually resolved for now. Load mjit_helper.h for MJIT header. mjit_helper.h: New. This is a file used only by JIT-ed code. I'll refactor `mjit_call_cfunc` later. vm_eval.c: add some #ifdef switches to skip compiling some functions like Init_vm_eval. win32/mkexports.rb: export thread/ec functions, which are used by MJIT. include/ruby/defines.h: add MJIT_FUNC_EXPORTED macro alis to clarify that a function is exported only for MJIT. array.c: export a function used by MJIT. bignum.c: ditto. class.c: ditto. compile.c: ditto. error.c: ditto. gc.c: ditto. hash.c: ditto. iseq.c: ditto. numeric.c: ditto. object.c: ditto. proc.c: ditto. re.c: ditto. st.c: ditto. string.c: ditto. thread.c: ditto. variable.c: ditto. vm_backtrace.c: ditto. vm_insnhelper.c: ditto. vm_method.c: ditto. I would like to improve maintainability of function exports, but I believe this way is acceptable as initial merging if we clarify the new exports are for MJIT (so that we can use them as TODO list to fix) and add unit tests to detect unresolved symbols. I'll add unit tests of JIT compilations in succeeding commits. Author: Takashi Kokubun <takashikkbn@gmail.com> Contributor: wanabe <s.wanabe@gmail.com> Part of [Feature #14235] --- * Known issues * Code generated by gcc is faster than clang. The benchmark may be worse in macOS. Following benchmark result is provided by gcc w/ Linux. * Performance is decreased when Google Chrome is running * JIT can work on MinGW, but it doesn't improve performance at least in short running benchmark. * Currently it doesn't perform well with Rails. We'll try to fix this before release. --- * Benchmark reslts Benchmarked with: Intel 4.0GHz i7-4790K with 16GB memory under x86-64 Ubuntu 8 Cores - 2.0.0-p0: Ruby 2.0.0-p0 - r62186: Ruby trunk (early 2.6.0), before MJIT changes - JIT off: On this commit, but without `--jit` option - JIT on: On this commit, and with `--jit` option ** Optcarrot fps Benchmark: https://github.com/mame/optcarrot | |2.0.0-p0 |r62186 |JIT off |JIT on | |:--------|:--------|:--------|:--------|:--------| |fps |37.32 |51.46 |51.31 |58.88 | |vs 2.0.0 |1.00x |1.38x |1.37x |1.58x | ** MJIT benchmarks Benchmark: https://github.com/benchmark-driver/mjit-benchmarks (Original: https://github.com/vnmakarov/ruby/tree/rtl_mjit_branch/MJIT-benchmarks) | |2.0.0-p0 |r62186 |JIT off |JIT on | |:----------|:--------|:--------|:--------|:--------| |aread |1.00 |1.09 |1.07 |2.19 | |aref |1.00 |1.13 |1.11 |2.22 | |aset |1.00 |1.50 |1.45 |2.64 | |awrite |1.00 |1.17 |1.13 |2.20 | |call |1.00 |1.29 |1.26 |2.02 | |const2 |1.00 |1.10 |1.10 |2.19 | |const |1.00 |1.11 |1.10 |2.19 | |fannk |1.00 |1.04 |1.02 |1.00 | |fib |1.00 |1.32 |1.31 |1.84 | |ivread |1.00 |1.13 |1.12 |2.43 | |ivwrite |1.00 |1.23 |1.21 |2.40 | |mandelbrot |1.00 |1.13 |1.16 |1.28 | |meteor |1.00 |2.97 |2.92 |3.17 | |nbody |1.00 |1.17 |1.15 |1.49 | |nest-ntimes|1.00 |1.22 |1.20 |1.39 | |nest-while |1.00 |1.10 |1.10 |1.37 | |norm |1.00 |1.18 |1.16 |1.24 | |nsvb |1.00 |1.16 |1.16 |1.17 | |red-black |1.00 |1.02 |0.99 |1.12 | |sieve |1.00 |1.30 |1.28 |1.62 | |trees |1.00 |1.14 |1.13 |1.19 | |while |1.00 |1.12 |1.11 |2.41 | ** Discourse's script/bench.rb Benchmark: https://github.com/discourse/discourse/blob/v1.8.7/script/bench.rb NOTE: Rails performance was somehow a little degraded with JIT for now. We should fix this. (At least I know opt_aref is performing badly in JIT and I have an idea to fix it. Please wait for the fix.) *** JIT off Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 17 75: 18 90: 22 99: 29 home_admin: 50: 21 75: 21 90: 27 99: 40 topic_admin: 50: 17 75: 18 90: 22 99: 32 categories: 50: 35 75: 41 90: 43 99: 77 home: 50: 39 75: 46 90: 49 99: 95 topic: 50: 46 75: 52 90: 56 99: 101 *** JIT on Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 19 75: 21 90: 25 99: 33 home_admin: 50: 24 75: 26 90: 30 99: 35 topic_admin: 50: 19 75: 20 90: 25 99: 30 categories: 50: 40 75: 44 90: 48 99: 76 home: 50: 42 75: 48 90: 51 99: 89 topic: 50: 49 75: 55 90: 58 99: 99 git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@62197 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-02-04 14:22:28 +03:00
MJIT_FUNC_EXPORTED const rb_callable_method_entry_t *
rb_callable_method_entry_without_refinements(VALUE klass, ID id, VALUE *defined_class_ptr)
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
{
VALUE defined_class, *dcp = defined_class_ptr ? defined_class_ptr : &defined_class;
const rb_method_entry_t *me = method_entry_resolve_refinement(klass, id, FALSE, dcp);
return prepare_callable_method_entry(*dcp, id, me, TRUE);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
}
static const rb_method_entry_t *
resolve_refined_method(VALUE refinements, const rb_method_entry_t *me, VALUE *defined_class_ptr)
* 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
{
while (me && me->def->type == VM_METHOD_TYPE_REFINED) {
* 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
VALUE refinement;
const rb_method_entry_t *tmp_me;
VALUE super;
* 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
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
refinement = find_refinement(refinements, me->owner);
if (!NIL_P(refinement)) {
tmp_me = search_method_protect(refinement, me->called_id, defined_class_ptr);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
if (tmp_me && tmp_me->def->type != VM_METHOD_TYPE_REFINED) {
return tmp_me;
}
* 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
}
tmp_me = me->def->body.refined.orig_me;
if (tmp_me) {
if (defined_class_ptr) *defined_class_ptr = tmp_me->defined_class;
return tmp_me;
}
super = RCLASS_SUPER(me->owner);
if (!super) {
return 0;
}
me = search_method_protect(super, me->called_id, defined_class_ptr);
* 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
}
return me;
}
const rb_method_entry_t *
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me)
{
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
return resolve_refined_method(refinements, me, NULL);
}
MJIT_FUNC_EXPORTED
const rb_callable_method_entry_t *
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me)
{
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
VALUE defined_class = me->defined_class;
const rb_method_entry_t *resolved_me = resolve_refined_method(refinements, (const rb_method_entry_t *)me, &defined_class);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
if (resolved_me && resolved_me->defined_class == 0) {
return rb_method_entry_complement_defined_class(resolved_me, me->called_id, defined_class);
}
else {
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
return (const rb_callable_method_entry_t *)resolved_me;
}
}
static void
remove_method(VALUE klass, ID mid)
{
VALUE data;
rb_method_entry_t *me = 0;
VALUE self = klass;
klass = RCLASS_ORIGIN(klass);
rb_class_modify_check(klass);
if (mid == object_id || mid == id__send__ || mid == idInitialize) {
rb_warn("removing `%s' may cause serious problems", rb_id2name(mid));
}
if (!rb_id_table_lookup(RCLASS_M_TBL(klass), mid, &data) ||
!(me = (rb_method_entry_t *)data) ||
(!me->def || me->def->type == VM_METHOD_TYPE_UNDEF) ||
UNDEFINED_REFINED_METHOD_P(me->def)) {
rb_name_err_raise("method `%1$s' not defined in %2$s",
klass, ID2SYM(mid));
}
rb_clear_method_cache(klass, mid);
rb_id_table_delete(RCLASS_M_TBL(klass), mid);
rb_vm_check_redefinition_opt_method(me, klass);
if (me->def->type == VM_METHOD_TYPE_REFINED) {
rb_add_refined_method_entry(klass, mid);
}
CALL_METHOD_HOOK(self, removed, mid);
}
void
rb_remove_method_id(VALUE klass, ID mid)
{
remove_method(klass, mid);
}
void
rb_remove_method(VALUE klass, const char *name)
{
remove_method(klass, rb_intern(name));
}
/*
* call-seq:
* remove_method(symbol) -> self
* remove_method(string) -> self
*
* Removes the method identified by _symbol_ from the current
* class. For an example, see Module#undef_method.
* String arguments are converted to symbols.
*/
static VALUE
rb_mod_remove_method(int argc, VALUE *argv, VALUE mod)
{
int i;
for (i = 0; i < argc; i++) {
VALUE v = argv[i];
ID id = rb_check_id(&v);
if (!id) {
rb_name_err_raise("method `%1$s' not defined in %2$s",
mod, v);
}
remove_method(mod, id);
}
return mod;
}
static void
rb_export_method(VALUE klass, ID name, rb_method_visibility_t visi)
{
rb_method_entry_t *me;
VALUE defined_class;
VALUE origin_class = RCLASS_ORIGIN(klass);
me = search_method(origin_class, name, &defined_class);
if (!me && RB_TYPE_P(klass, T_MODULE)) {
* 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
me = search_method(rb_cObject, name, &defined_class);
}
if (UNDEFINED_METHOD_ENTRY_P(me) ||
UNDEFINED_REFINED_METHOD_P(me->def)) {
rb_print_undef(klass, name, METHOD_VISI_UNDEF);
}
if (METHOD_ENTRY_VISI(me) != visi) {
rb_vm_check_redefinition_opt_method(me, klass);
if (klass == defined_class || origin_class == defined_class) {
METHOD_ENTRY_VISI_SET(me, visi);
if (me->def->type == VM_METHOD_TYPE_REFINED && me->def->body.refined.orig_me) {
METHOD_ENTRY_VISI_SET((rb_method_entry_t *)me->def->body.refined.orig_me, visi);
* 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
}
rb_clear_method_cache(klass, name);
}
else {
rb_add_method(klass, name, VM_METHOD_TYPE_ZSUPER, 0, visi);
}
}
}
#define BOUND_PRIVATE 0x01
#define BOUND_RESPONDS 0x02
int
rb_method_boundp(VALUE klass, ID id, int ex)
{
const rb_method_entry_t *me;
if (ex & BOUND_RESPONDS) {
me = method_entry_resolve_refinement(klass, id, TRUE, NULL);
}
else {
me = rb_method_entry_without_refinements(klass, id, NULL);
}
if (me != NULL) {
if ((ex & ~BOUND_RESPONDS) &&
((METHOD_ENTRY_VISI(me) == METHOD_VISI_PRIVATE) ||
((ex & BOUND_RESPONDS) && (METHOD_ENTRY_VISI(me) == METHOD_VISI_PROTECTED)))) {
return 0;
}
if (me->def->type == VM_METHOD_TYPE_NOTIMPLEMENTED) {
if (ex & BOUND_RESPONDS) return 2;
return 0;
}
return 1;
}
return 0;
}
static void
vm_cref_set_visibility(rb_method_visibility_t method_visi, int module_func)
{
rb_scope_visibility_t *scope_visi = (rb_scope_visibility_t *)&rb_vm_cref()->scope_visi;
scope_visi->method_visi = method_visi;
scope_visi->module_func = module_func;
}
void
rb_scope_visibility_set(rb_method_visibility_t visi)
{
vm_cref_set_visibility(visi, FALSE);
}
static void
scope_visibility_check(void)
{
/* Check for public/protected/private/module_function called inside a method */
rb_control_frame_t *cfp = rb_current_execution_context()->cfp+1;
if (cfp && cfp->iseq && cfp->iseq->body->type == ISEQ_TYPE_METHOD) {
rb_warn("calling %s without arguments inside a method may not have the intended effect",
rb_id2name(rb_frame_this_func()));
}
}
static void
rb_scope_module_func_set(void)
{
scope_visibility_check();
vm_cref_set_visibility(METHOD_VISI_PRIVATE, TRUE);
}
const rb_cref_t *rb_vm_cref_in_context(VALUE self, VALUE cbase);
void
rb_attr(VALUE klass, ID id, int read, int write, int ex)
{
ID attriv;
rb_method_visibility_t visi;
const rb_execution_context_t *ec = GET_EC();
const rb_cref_t *cref = rb_vm_cref_in_context(klass, klass);
if (!ex || !cref) {
visi = METHOD_VISI_PUBLIC;
}
else {
switch (vm_scope_visibility_get(ec)) {
case METHOD_VISI_PRIVATE:
if (vm_scope_module_func_check(ec)) {
rb_warning("attribute accessor as module_function");
}
visi = METHOD_VISI_PRIVATE;
break;
case METHOD_VISI_PROTECTED:
visi = METHOD_VISI_PROTECTED;
break;
default:
visi = METHOD_VISI_PUBLIC;
break;
}
}
attriv = rb_intern_str(rb_sprintf("@%"PRIsVALUE, rb_id2str(id)));
if (read) {
rb_add_method(klass, id, VM_METHOD_TYPE_IVAR, (void *)attriv, visi);
}
if (write) {
rb_add_method(klass, rb_id_attrset(id), VM_METHOD_TYPE_ATTRSET, (void *)attriv, visi);
}
}
void
rb_undef(VALUE klass, ID id)
{
const rb_method_entry_t *me;
if (NIL_P(klass)) {
rb_raise(rb_eTypeError, "no class to undef method");
}
rb_class_modify_check(klass);
if (id == object_id || id == id__send__ || id == idInitialize) {
rb_warn("undefining `%s' may cause serious problems", rb_id2name(id));
}
* 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
me = search_method(klass, id, 0);
if (me && me->def->type == VM_METHOD_TYPE_REFINED) {
me = rb_resolve_refined_method(Qnil, me);
}
if (UNDEFINED_METHOD_ENTRY_P(me) ||
UNDEFINED_REFINED_METHOD_P(me->def)) {
rb_method_name_error(klass, rb_id2str(id));
}
rb_add_method(klass, id, VM_METHOD_TYPE_UNDEF, 0, METHOD_VISI_PUBLIC);
CALL_METHOD_HOOK(klass, undefined, id);
}
/*
* call-seq:
* undef_method(symbol) -> self
* undef_method(string) -> self
*
* Prevents the current class from responding to calls to the named
* method. Contrast this with <code>remove_method</code>, which deletes
* the method from the particular class; Ruby will still search
* superclasses and mixed-in modules for a possible receiver.
* String arguments are converted to symbols.
*
* class Parent
* def hello
* puts "In parent"
* end
* end
* class Child < Parent
* def hello
* puts "In child"
* end
* end
*
*
* c = Child.new
* c.hello
*
*
* class Child
* remove_method :hello # remove from child, still in parent
* end
* c.hello
*
*
* class Child
* undef_method :hello # prevent any calls to 'hello'
* end
* c.hello
*
* <em>produces:</em>
*
* In child
* In parent
* prog.rb:23: undefined method `hello' for #<Child:0x401b3bb4> (NoMethodError)
*/
static VALUE
rb_mod_undef_method(int argc, VALUE *argv, VALUE mod)
{
int i;
for (i = 0; i < argc; i++) {
VALUE v = argv[i];
ID id = rb_check_id(&v);
if (!id) {
rb_method_name_error(mod, v);
}
rb_undef(mod, id);
}
return mod;
}
static rb_method_visibility_t
check_definition_visibility(VALUE mod, int argc, VALUE *argv)
{
const rb_method_entry_t *me;
VALUE mid, include_super, lookup_mod = mod;
int inc_super;
ID id;
rb_scan_args(argc, argv, "11", &mid, &include_super);
id = rb_check_id(&mid);
if (!id) return METHOD_VISI_UNDEF;
if (argc == 1) {
inc_super = 1;
}
else {
inc_super = RTEST(include_super);
if (!inc_super) {
lookup_mod = RCLASS_ORIGIN(mod);
}
}
me = rb_method_entry_without_refinements(lookup_mod, id, NULL);
if (me) {
if (me->def->type == VM_METHOD_TYPE_NOTIMPLEMENTED) return METHOD_VISI_UNDEF;
if (!inc_super && me->owner != mod) return METHOD_VISI_UNDEF;
return METHOD_ENTRY_VISI(me);
}
return METHOD_VISI_UNDEF;
}
/*
* call-seq:
* mod.method_defined?(symbol, inherit=true) -> true or false
* mod.method_defined?(string, inherit=true) -> true or false
*
* Returns +true+ if the named method is defined by
* _mod_. If _inherit_ is set, the lookup will also search _mod_'s
* ancestors. Public and protected methods are matched.
* String arguments are converted to symbols.
*
* module A
* def method1() end
* def protected_method1() end
* protected :protected_method1
* end
* class B
* def method2() end
* def private_method2() end
* private :private_method2
* end
* class C < B
* include A
* def method3() end
* end
*
* A.method_defined? :method1 #=> true
* C.method_defined? "method1" #=> true
* C.method_defined? "method2" #=> true
* C.method_defined? "method2", true #=> true
* C.method_defined? "method2", false #=> false
* C.method_defined? "method3" #=> true
* C.method_defined? "protected_method1" #=> true
* C.method_defined? "method4" #=> false
* C.method_defined? "private_method2" #=> false
*/
static VALUE
rb_mod_method_defined(int argc, VALUE *argv, VALUE mod)
{
rb_method_visibility_t visi = check_definition_visibility(mod, argc, argv);
return (visi == METHOD_VISI_PUBLIC || visi == METHOD_VISI_PROTECTED) ? Qtrue : Qfalse;
}
static VALUE
check_definition(VALUE mod, int argc, VALUE *argv, rb_method_visibility_t visi)
{
return (check_definition_visibility(mod, argc, argv) == visi) ? Qtrue : Qfalse;
}
/*
* call-seq:
* mod.public_method_defined?(symbol, inherit=true) -> true or false
* mod.public_method_defined?(string, inherit=true) -> true or false
*
* Returns +true+ if the named public method is defined by
* _mod_. If _inherit_ is set, the lookup will also search _mod_'s
* ancestors.
* String arguments are converted to symbols.
*
* module A
* def method1() end
* end
* class B
* protected
* def method2() end
* end
* class C < B
* include A
* def method3() end
* end
*
* A.method_defined? :method1 #=> true
* C.public_method_defined? "method1" #=> true
* C.public_method_defined? "method1", true #=> true
* C.public_method_defined? "method1", false #=> true
* C.public_method_defined? "method2" #=> false
* C.method_defined? "method2" #=> true
*/
static VALUE
rb_mod_public_method_defined(int argc, VALUE *argv, VALUE mod)
{
return check_definition(mod, argc, argv, METHOD_VISI_PUBLIC);
}
/*
* call-seq:
* mod.private_method_defined?(symbol, inherit=true) -> true or false
* mod.private_method_defined?(string, inherit=true) -> true or false
*
* Returns +true+ if the named private method is defined by
* _mod_. If _inherit_ is set, the lookup will also search _mod_'s
* ancestors.
* String arguments are converted to symbols.
*
* module A
* def method1() end
* end
* class B
* private
* def method2() end
* end
* class C < B
* include A
* def method3() end
* end
*
* A.method_defined? :method1 #=> true
* C.private_method_defined? "method1" #=> false
* C.private_method_defined? "method2" #=> true
* C.private_method_defined? "method2", true #=> true
* C.private_method_defined? "method2", false #=> false
* C.method_defined? "method2" #=> false
*/
static VALUE
rb_mod_private_method_defined(int argc, VALUE *argv, VALUE mod)
{
return check_definition(mod, argc, argv, METHOD_VISI_PRIVATE);
}
/*
* call-seq:
* mod.protected_method_defined?(symbol, inherit=true) -> true or false
* mod.protected_method_defined?(string, inherit=true) -> true or false
*
* Returns +true+ if the named protected method is defined
* _mod_. If _inherit_ is set, the lookup will also search _mod_'s
* ancestors.
* String arguments are converted to symbols.
*
* module A
* def method1() end
* end
* class B
* protected
* def method2() end
* end
* class C < B
* include A
* def method3() end
* end
*
* A.method_defined? :method1 #=> true
* C.protected_method_defined? "method1" #=> false
* C.protected_method_defined? "method2" #=> true
* C.protected_method_defined? "method2", true #=> true
* C.protected_method_defined? "method2", false #=> false
* C.method_defined? "method2" #=> true
*/
static VALUE
rb_mod_protected_method_defined(int argc, VALUE *argv, VALUE mod)
{
return check_definition(mod, argc, argv, METHOD_VISI_PROTECTED);
}
int
rb_method_entry_eq(const rb_method_entry_t *m1, const rb_method_entry_t *m2)
{
return rb_method_definition_eq(m1->def, m2->def);
}
static const rb_method_definition_t *
original_method_definition(const rb_method_definition_t *def)
{
again:
if (def) {
switch (def->type) {
case VM_METHOD_TYPE_REFINED:
if (def->body.refined.orig_me) {
def = def->body.refined.orig_me->def;
goto again;
}
break;
case VM_METHOD_TYPE_ALIAS:
def = def->body.alias.original_me->def;
goto again;
default:
break;
}
}
return def;
}
mjit_compile.c: merge initial JIT compiler which has been developed by Takashi Kokubun <takashikkbn@gmail> as YARV-MJIT. Many of its bugs are fixed by wanabe <s.wanabe@gmail.com>. This JIT compiler is designed to be a safe migration path to introduce JIT compiler to MRI. So this commit does not include any bytecode changes or dynamic instruction modifications, which are done in original MJIT. This commit even strips off some aggressive optimizations from YARV-MJIT, and thus it's slower than YARV-MJIT too. But it's still fairly faster than Ruby 2.5 in some benchmarks (attached below). Note that this JIT compiler passes `make test`, `make test-all`, `make test-spec` without JIT, and even with JIT. Not only it's perfectly safe with JIT disabled because it does not replace VM instructions unlike MJIT, but also with JIT enabled it stably runs Ruby applications including Rails applications. I'm expecting this version as just "initial" JIT compiler. I have many optimization ideas which are skipped for initial merging, and you may easily replace this JIT compiler with a faster one by just replacing mjit_compile.c. `mjit_compile` interface is designed for the purpose. common.mk: update dependencies for mjit_compile.c. internal.h: declare `rb_vm_insn_addr2insn` for MJIT. vm.c: exclude some definitions if `-DMJIT_HEADER` is provided to compiler. This avoids to include some functions which take a long time to compile, e.g. vm_exec_core. Some of the purpose is achieved in transform_mjit_header.rb (see `IGNORED_FUNCTIONS`) but others are manually resolved for now. Load mjit_helper.h for MJIT header. mjit_helper.h: New. This is a file used only by JIT-ed code. I'll refactor `mjit_call_cfunc` later. vm_eval.c: add some #ifdef switches to skip compiling some functions like Init_vm_eval. win32/mkexports.rb: export thread/ec functions, which are used by MJIT. include/ruby/defines.h: add MJIT_FUNC_EXPORTED macro alis to clarify that a function is exported only for MJIT. array.c: export a function used by MJIT. bignum.c: ditto. class.c: ditto. compile.c: ditto. error.c: ditto. gc.c: ditto. hash.c: ditto. iseq.c: ditto. numeric.c: ditto. object.c: ditto. proc.c: ditto. re.c: ditto. st.c: ditto. string.c: ditto. thread.c: ditto. variable.c: ditto. vm_backtrace.c: ditto. vm_insnhelper.c: ditto. vm_method.c: ditto. I would like to improve maintainability of function exports, but I believe this way is acceptable as initial merging if we clarify the new exports are for MJIT (so that we can use them as TODO list to fix) and add unit tests to detect unresolved symbols. I'll add unit tests of JIT compilations in succeeding commits. Author: Takashi Kokubun <takashikkbn@gmail.com> Contributor: wanabe <s.wanabe@gmail.com> Part of [Feature #14235] --- * Known issues * Code generated by gcc is faster than clang. The benchmark may be worse in macOS. Following benchmark result is provided by gcc w/ Linux. * Performance is decreased when Google Chrome is running * JIT can work on MinGW, but it doesn't improve performance at least in short running benchmark. * Currently it doesn't perform well with Rails. We'll try to fix this before release. --- * Benchmark reslts Benchmarked with: Intel 4.0GHz i7-4790K with 16GB memory under x86-64 Ubuntu 8 Cores - 2.0.0-p0: Ruby 2.0.0-p0 - r62186: Ruby trunk (early 2.6.0), before MJIT changes - JIT off: On this commit, but without `--jit` option - JIT on: On this commit, and with `--jit` option ** Optcarrot fps Benchmark: https://github.com/mame/optcarrot | |2.0.0-p0 |r62186 |JIT off |JIT on | |:--------|:--------|:--------|:--------|:--------| |fps |37.32 |51.46 |51.31 |58.88 | |vs 2.0.0 |1.00x |1.38x |1.37x |1.58x | ** MJIT benchmarks Benchmark: https://github.com/benchmark-driver/mjit-benchmarks (Original: https://github.com/vnmakarov/ruby/tree/rtl_mjit_branch/MJIT-benchmarks) | |2.0.0-p0 |r62186 |JIT off |JIT on | |:----------|:--------|:--------|:--------|:--------| |aread |1.00 |1.09 |1.07 |2.19 | |aref |1.00 |1.13 |1.11 |2.22 | |aset |1.00 |1.50 |1.45 |2.64 | |awrite |1.00 |1.17 |1.13 |2.20 | |call |1.00 |1.29 |1.26 |2.02 | |const2 |1.00 |1.10 |1.10 |2.19 | |const |1.00 |1.11 |1.10 |2.19 | |fannk |1.00 |1.04 |1.02 |1.00 | |fib |1.00 |1.32 |1.31 |1.84 | |ivread |1.00 |1.13 |1.12 |2.43 | |ivwrite |1.00 |1.23 |1.21 |2.40 | |mandelbrot |1.00 |1.13 |1.16 |1.28 | |meteor |1.00 |2.97 |2.92 |3.17 | |nbody |1.00 |1.17 |1.15 |1.49 | |nest-ntimes|1.00 |1.22 |1.20 |1.39 | |nest-while |1.00 |1.10 |1.10 |1.37 | |norm |1.00 |1.18 |1.16 |1.24 | |nsvb |1.00 |1.16 |1.16 |1.17 | |red-black |1.00 |1.02 |0.99 |1.12 | |sieve |1.00 |1.30 |1.28 |1.62 | |trees |1.00 |1.14 |1.13 |1.19 | |while |1.00 |1.12 |1.11 |2.41 | ** Discourse's script/bench.rb Benchmark: https://github.com/discourse/discourse/blob/v1.8.7/script/bench.rb NOTE: Rails performance was somehow a little degraded with JIT for now. We should fix this. (At least I know opt_aref is performing badly in JIT and I have an idea to fix it. Please wait for the fix.) *** JIT off Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 17 75: 18 90: 22 99: 29 home_admin: 50: 21 75: 21 90: 27 99: 40 topic_admin: 50: 17 75: 18 90: 22 99: 32 categories: 50: 35 75: 41 90: 43 99: 77 home: 50: 39 75: 46 90: 49 99: 95 topic: 50: 46 75: 52 90: 56 99: 101 *** JIT on Your Results: (note for timings- percentile is first, duration is second in millisecs) categories_admin: 50: 19 75: 21 90: 25 99: 33 home_admin: 50: 24 75: 26 90: 30 99: 35 topic_admin: 50: 19 75: 20 90: 25 99: 30 categories: 50: 40 75: 44 90: 48 99: 76 home: 50: 42 75: 48 90: 51 99: 89 topic: 50: 49 75: 55 90: 58 99: 99 git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@62197 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-02-04 14:22:28 +03:00
MJIT_FUNC_EXPORTED int
rb_method_definition_eq(const rb_method_definition_t *d1, const rb_method_definition_t *d2)
{
d1 = original_method_definition(d1);
d2 = original_method_definition(d2);
if (d1 == d2) return 1;
if (!d1 || !d2) return 0;
if (d1->type != d2->type) return 0;
switch (d1->type) {
case VM_METHOD_TYPE_ISEQ:
return d1->body.iseq.iseqptr == d2->body.iseq.iseqptr;
case VM_METHOD_TYPE_CFUNC:
return
d1->body.cfunc.func == d2->body.cfunc.func &&
d1->body.cfunc.argc == d2->body.cfunc.argc;
case VM_METHOD_TYPE_ATTRSET:
case VM_METHOD_TYPE_IVAR:
return d1->body.attr.id == d2->body.attr.id;
case VM_METHOD_TYPE_BMETHOD:
return RTEST(rb_equal(d1->body.bmethod.proc, d2->body.bmethod.proc));
case VM_METHOD_TYPE_MISSING:
return d1->original_id == d2->original_id;
case VM_METHOD_TYPE_ZSUPER:
case VM_METHOD_TYPE_NOTIMPLEMENTED:
case VM_METHOD_TYPE_UNDEF:
return 1;
case VM_METHOD_TYPE_OPTIMIZED:
return d1->body.optimize_type == d2->body.optimize_type;
case VM_METHOD_TYPE_REFINED:
case VM_METHOD_TYPE_ALIAS:
break;
}
rb_bug("rb_method_definition_eq: unsupported type: %d\n", d1->type);
}
static st_index_t
rb_hash_method_definition(st_index_t hash, const rb_method_definition_t *def)
{
hash = rb_hash_uint(hash, def->type);
def = original_method_definition(def);
if (!def) return hash;
switch (def->type) {
case VM_METHOD_TYPE_ISEQ:
return rb_hash_uint(hash, (st_index_t)def->body.iseq.iseqptr);
case VM_METHOD_TYPE_CFUNC:
hash = rb_hash_uint(hash, (st_index_t)def->body.cfunc.func);
return rb_hash_uint(hash, def->body.cfunc.argc);
case VM_METHOD_TYPE_ATTRSET:
case VM_METHOD_TYPE_IVAR:
return rb_hash_uint(hash, def->body.attr.id);
case VM_METHOD_TYPE_BMETHOD:
Support targetting TracePoint [Feature #15289] * vm_trace.c (rb_tracepoint_enable_for_target): support targetting TracePoint. [Feature #15289] Tragetting TracePoint is only enabled on specified method, proc and so on, example: `tp.enable(target: code)`. `code` should be consisted of InstructionSeuqnece (iseq) (RubyVM::InstructionSeuqnece.of(code) should not return nil) If code is a tree of iseq, TracePoint is enabled on all of iseqs in a tree. Enabled tragetting TracePoints can not enabled again with and without target. * vm_core.h (rb_iseq_t): introduce `rb_iseq_t::local_hooks` to store local hooks. `rb_iseq_t::aux::trace_events` is renamed to `global_trace_events` to contrast with `local_hooks`. * vm_core.h (rb_hook_list_t): add `rb_hook_list_t::running` to represent how many Threads/Fibers are used this list. If this field is 0, nobody using this hooks and we can delete it. This is why we can remove code from cont.c. * vm_core.h (rb_vm_t): because of above change, we can eliminate `rb_vm_t::trace_running` field. Also renamed from `rb_vm_t::event_hooks` to `global_hooks`. * vm_core.h, vm.c (ruby_vm_event_enabled_global_flags): renamed from `ruby_vm_event_enabled_flags. * vm_core.h, vm.c (ruby_vm_event_local_num): added to count enabled targetting TracePoints. * vm_core.h, vm_trace.c (rb_exec_event_hooks): accepts hook list. * vm_core.h (rb_vm_global_hooks): added for convinience. * method.h (rb_method_bmethod_t): added to maintain Proc and `rb_hook_list_t` for bmethod (defined by define_method). * prelude.rb (TracePoint#enable): extracet a keyword parameter (because it is easy than writing in C). It calls `TracePoint#__enable` internal method written in C. * vm_insnhelper.c (vm_trace): check also iseq->local_hooks. * vm.c (invoke_bmethod): check def->body.bmethod.hooks. * vm.c (hook_before_rewind): check iseq->local_hooks and def->body.bmethod.hooks before rewind by exception. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@66003 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-11-26 21:16:39 +03:00
return rb_hash_proc(hash, def->body.bmethod.proc);
case VM_METHOD_TYPE_MISSING:
return rb_hash_uint(hash, def->original_id);
case VM_METHOD_TYPE_ZSUPER:
case VM_METHOD_TYPE_NOTIMPLEMENTED:
case VM_METHOD_TYPE_UNDEF:
return hash;
case VM_METHOD_TYPE_OPTIMIZED:
return rb_hash_uint(hash, def->body.optimize_type);
* 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:
case VM_METHOD_TYPE_ALIAS:
break; /* unreachable */
* 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
}
rb_bug("rb_hash_method_definition: unsupported method type (%d)\n", def->type);
}
st_index_t
rb_hash_method_entry(st_index_t hash, const rb_method_entry_t *me)
{
return rb_hash_method_definition(hash, me->def);
}
void
rb_alias(VALUE klass, ID alias_name, ID original_name)
{
const VALUE target_klass = klass;
VALUE defined_class;
const rb_method_entry_t *orig_me;
rb_method_visibility_t visi = METHOD_VISI_UNDEF;
if (NIL_P(klass)) {
rb_raise(rb_eTypeError, "no class to make alias");
}
rb_class_modify_check(klass);
again:
orig_me = search_method(klass, original_name, &defined_class);
if (orig_me && orig_me->def->type == VM_METHOD_TYPE_REFINED) {
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
orig_me = rb_resolve_refined_method(Qnil, orig_me);
}
if (UNDEFINED_METHOD_ENTRY_P(orig_me) ||
UNDEFINED_REFINED_METHOD_P(orig_me->def)) {
if ((!RB_TYPE_P(klass, T_MODULE)) ||
(orig_me = search_method(rb_cObject, original_name, &defined_class),
UNDEFINED_METHOD_ENTRY_P(orig_me))) {
rb_print_undef(klass, original_name, METHOD_VISI_UNDEF);
}
}
if (orig_me->def->type == VM_METHOD_TYPE_ZSUPER) {
klass = RCLASS_SUPER(klass);
original_name = orig_me->def->original_id;
visi = METHOD_ENTRY_VISI(orig_me);
goto again;
}
if (visi == METHOD_VISI_UNDEF) visi = METHOD_ENTRY_VISI(orig_me);
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
if (orig_me->defined_class == 0) {
rb_method_entry_make(target_klass, alias_name, target_klass, visi,
VM_METHOD_TYPE_ALIAS, NULL, orig_me->called_id,
(void *)rb_method_entry_clone(orig_me));
method_added(target_klass, alias_name);
}
else {
rb_method_entry_t *alias_me;
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
alias_me = method_entry_set(target_klass, alias_name, orig_me, visi, orig_me->owner);
RB_OBJ_WRITE(alias_me, &alias_me->owner, target_klass);
RB_OBJ_WRITE(alias_me, &alias_me->defined_class, orig_me->defined_class);
}
}
/*
* call-seq:
* alias_method(new_name, old_name) -> self
*
* Makes <i>new_name</i> a new copy of the method <i>old_name</i>. This can
* be used to retain access to methods that are overridden.
*
* module Mod
* alias_method :orig_exit, :exit
* def exit(code=0)
* puts "Exiting with code #{code}"
* orig_exit(code)
* end
* end
* include Mod
* exit(99)
*
* <em>produces:</em>
*
* Exiting with code 99
*/
static VALUE
rb_mod_alias_method(VALUE mod, VALUE newname, VALUE oldname)
{
ID oldid = rb_check_id(&oldname);
if (!oldid) {
rb_print_undef_str(mod, oldname);
}
rb_alias(mod, rb_to_id(newname), oldid);
return mod;
}
static void
set_method_visibility(VALUE self, int argc, const VALUE *argv, rb_method_visibility_t visi)
{
int i;
rb_check_frozen(self);
if (argc == 0) {
rb_warning("%"PRIsVALUE" with no argument is just ignored",
QUOTE_ID(rb_frame_callee()));
return;
}
for (i = 0; i < argc; i++) {
VALUE v = argv[i];
ID id = rb_check_id(&v);
if (!id) {
rb_print_undef_str(self, v);
}
rb_export_method(self, id, visi);
}
}
static VALUE
set_visibility(int argc, const VALUE *argv, VALUE module, rb_method_visibility_t visi)
{
if (argc == 0) {
scope_visibility_check();
rb_scope_visibility_set(visi);
}
else {
set_method_visibility(module, argc, argv, visi);
}
return module;
}
/*
* call-seq:
* public -> self
* public(symbol, ...) -> self
* public(string, ...) -> self
*
* With no arguments, sets the default visibility for subsequently
* defined methods to public. With arguments, sets the named methods to
* have public visibility.
* String arguments are converted to symbols.
*/
static VALUE
rb_mod_public(int argc, VALUE *argv, VALUE module)
{
return set_visibility(argc, argv, module, METHOD_VISI_PUBLIC);
}
/*
* call-seq:
* protected -> self
* protected(symbol, ...) -> self
* protected(string, ...) -> self
*
* With no arguments, sets the default visibility for subsequently
* defined methods to protected. With arguments, sets the named methods
* to have protected visibility.
* String arguments are converted to symbols.
*
* If a method has protected visibility, it is callable only where
* <code>self</code> of the context is the same as the method.
* (method definition or instance_eval). This behavior is different from
* Java's protected method. Usually <code>private</code> should be used.
*
* Note that a protected method is slow because it can't use inline cache.
*
* To show a private method on RDoc, use <code>:doc:</code> instead of this.
*/
static VALUE
rb_mod_protected(int argc, VALUE *argv, VALUE module)
{
return set_visibility(argc, argv, module, METHOD_VISI_PROTECTED);
}
/*
* call-seq:
* private -> self
* private(symbol, ...) -> self
* private(string, ...) -> self
*
* With no arguments, sets the default visibility for subsequently
* defined methods to private. With arguments, sets the named methods
* to have private visibility.
* String arguments are converted to symbols.
*
* module Mod
* def a() end
* def b() end
* private
* def c() end
* private :a
* end
* Mod.private_instance_methods #=> [:a, :c]
*
* Note that to show a private method on RDoc, use <code>:doc:</code>.
*/
static VALUE
rb_mod_private(int argc, VALUE *argv, VALUE module)
{
return set_visibility(argc, argv, module, METHOD_VISI_PRIVATE);
}
/*
* call-seq:
* ruby2_keywords(method_name, ...) -> self
*
* For the given method names, marks the method as passing keywords through
* a normal argument splat. This should only be called on methods that
* accept an argument splat (<tt>*args</tt>) but not explicit keywords or
* a keyword splat. It marks the method such that if the method is called
* with keyword arguments, the final hash argument is marked with a special
* flag such that if it is the final element of a normal argument splat to
* another method call, and that method call does not include explicit
* keywords or a keyword splat, the final element is interpreted as keywords.
* In other words, keywords will be passed through the method to other
* methods.
*
* This should only be used for methods that delegate keywords to another
* method, and only for backwards compatibility with Ruby versions before
* 2.7.
*
* This method will probably be removed at some point, as it exists only
* for backwards compatibility. As it does not exist in Ruby versions
* before 2.7, check that the module responds to this method before calling
* it. Also, be aware that if this method is removed, the behavior of the
* method will change so that it does not pass through keywords.
*
* module Mod
* def foo(meth, *args, &block)
* send(:"do_#{meth}", *args, &block)
* end
* ruby2_keywords(:foo) if respond_to?(:ruby2_keywords, true)
* end
*/
static VALUE
rb_mod_ruby2_keywords(int argc, VALUE *argv, VALUE module)
{
int i;
VALUE origin_class = RCLASS_ORIGIN(module);
rb_check_arity(argc, 1, UNLIMITED_ARGUMENTS);
rb_check_frozen(module);
for (i = 0; i < argc; i++) {
VALUE v = argv[i];
ID name = rb_check_id(&v);
rb_method_entry_t *me;
VALUE defined_class;
if (!name) {
rb_print_undef_str(module, v);
}
me = search_method(origin_class, name, &defined_class);
if (!me && RB_TYPE_P(module, T_MODULE)) {
me = search_method(rb_cObject, name, &defined_class);
}
if (UNDEFINED_METHOD_ENTRY_P(me) ||
UNDEFINED_REFINED_METHOD_P(me->def)) {
rb_print_undef(module, name, METHOD_VISI_UNDEF);
}
if (module == defined_class || origin_class == defined_class) {
switch (me->def->type) {
case VM_METHOD_TYPE_ISEQ:
if (me->def->body.iseq.iseqptr->body->param.flags.has_rest &&
!me->def->body.iseq.iseqptr->body->param.flags.has_kw &&
!me->def->body.iseq.iseqptr->body->param.flags.has_kwrest) {
me->def->body.iseq.iseqptr->body->param.flags.ruby2_keywords = 1;
rb_clear_method_cache(module, name);
}
else {
rb_warn("Skipping set of ruby2_keywords flag for %s (method accepts keywords or method does not accept argument splat)", rb_id2name(name));
}
break;
case VM_METHOD_TYPE_BMETHOD: {
VALUE procval = me->def->body.bmethod.proc;
if (vm_block_handler_type(procval) == block_handler_type_proc) {
procval = vm_proc_to_block_handler(VM_BH_TO_PROC(procval));
}
if (vm_block_handler_type(procval) == block_handler_type_iseq) {
const struct rb_captured_block *captured = VM_BH_TO_ISEQ_BLOCK(procval);
const rb_iseq_t *iseq = rb_iseq_check(captured->code.iseq);
if (iseq->body->param.flags.has_rest &&
!iseq->body->param.flags.has_kw &&
!iseq->body->param.flags.has_kwrest) {
iseq->body->param.flags.ruby2_keywords = 1;
rb_clear_method_cache(module, name);
}
else {
rb_warn("Skipping set of ruby2_keywords flag for %s (method accepts keywords or method does not accept argument splat)", rb_id2name(name));
}
return Qnil;
}
}
/* fallthrough */
default:
rb_warn("Skipping set of ruby2_keywords flag for %s (method not defined in Ruby)", rb_id2name(name));
break;
}
}
else {
rb_warn("Skipping set of ruby2_keywords flag for %s (can only set in method defining module)", rb_id2name(name));
}
}
return Qnil;
}
/*
* call-seq:
* mod.public_class_method(symbol, ...) -> mod
* mod.public_class_method(string, ...) -> mod
*
* Makes a list of existing class methods public.
*
* String arguments are converted to symbols.
*/
static VALUE
rb_mod_public_method(int argc, VALUE *argv, VALUE obj)
{
set_method_visibility(rb_singleton_class(obj), argc, argv, METHOD_VISI_PUBLIC);
return obj;
}
/*
* call-seq:
* mod.private_class_method(symbol, ...) -> mod
* mod.private_class_method(string, ...) -> mod
*
* Makes existing class methods private. Often used to hide the default
* constructor <code>new</code>.
*
* String arguments are converted to symbols.
*
* class SimpleSingleton # Not thread safe
* private_class_method :new
* def SimpleSingleton.create(*args, &block)
* @me = new(*args, &block) if ! @me
* @me
* end
* end
*/
static VALUE
rb_mod_private_method(int argc, VALUE *argv, VALUE obj)
{
set_method_visibility(rb_singleton_class(obj), argc, argv, METHOD_VISI_PRIVATE);
return obj;
}
/*
* call-seq:
* public
* public(symbol, ...)
* public(string, ...)
*
* With no arguments, sets the default visibility for subsequently
* defined methods to public. With arguments, sets the named methods to
* have public visibility.
*
* String arguments are converted to symbols.
*/
static VALUE
top_public(int argc, VALUE *argv, VALUE _)
{
return rb_mod_public(argc, argv, rb_cObject);
}
/*
* call-seq:
* private
* private(symbol, ...)
* private(string, ...)
*
* With no arguments, sets the default visibility for subsequently
* defined methods to private. With arguments, sets the named methods to
* have private visibility.
*
* String arguments are converted to symbols.
*/
static VALUE
top_private(int argc, VALUE *argv, VALUE _)
{
return rb_mod_private(argc, argv, rb_cObject);
}
/*
* call-seq:
* ruby2_keywords(method_name, ...) -> self
*
* For the given method names, marks the method as passing keywords through
* a normal argument splat. See Module#ruby2_keywords in detail.
*/
static VALUE
top_ruby2_keywords(int argc, VALUE *argv, VALUE module)
{
return rb_mod_ruby2_keywords(argc, argv, rb_cObject);
}
/*
* call-seq:
* module_function(symbol, ...) -> self
* module_function(string, ...) -> self
*
* Creates module functions for the named methods. These functions may
* be called with the module as a receiver, and also become available
* as instance methods to classes that mix in the module. Module
* functions are copies of the original, and so may be changed
* independently. The instance-method versions are made private. If
* used with no arguments, subsequently defined methods become module
* functions.
* String arguments are converted to symbols.
*
* module Mod
* def one
* "This is one"
* end
* module_function :one
* end
* class Cls
* include Mod
* def call_one
* one
* end
* end
* Mod.one #=> "This is one"
* c = Cls.new
* c.call_one #=> "This is one"
* module Mod
* def one
* "This is the new one"
* end
* end
* Mod.one #=> "This is one"
* c.call_one #=> "This is the new one"
*/
static VALUE
rb_mod_modfunc(int argc, VALUE *argv, VALUE module)
{
int i;
ID id;
const rb_method_entry_t *me;
if (!RB_TYPE_P(module, T_MODULE)) {
rb_raise(rb_eTypeError, "module_function must be called for modules");
}
if (argc == 0) {
rb_scope_module_func_set();
return module;
}
set_method_visibility(module, argc, argv, METHOD_VISI_PRIVATE);
for (i = 0; i < argc; i++) {
VALUE m = module;
id = rb_to_id(argv[i]);
for (;;) {
* 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
me = search_method(m, id, 0);
if (me == 0) {
* 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
me = search_method(rb_cObject, id, 0);
}
if (UNDEFINED_METHOD_ENTRY_P(me)) {
rb_print_undef(module, id, METHOD_VISI_UNDEF);
}
if (me->def->type != VM_METHOD_TYPE_ZSUPER) {
break; /* normal case: need not to follow 'super' link */
}
m = RCLASS_SUPER(m);
if (!m)
break;
}
rb_method_entry_set(rb_singleton_class(module), id, me, METHOD_VISI_PUBLIC);
}
return module;
}
#ifdef __GNUC__
#pragma push_macro("rb_method_basic_definition_p")
#undef rb_method_basic_definition_p
#endif
int
rb_method_basic_definition_p(VALUE klass, ID id)
{
const rb_callable_method_entry_t *cme;
if (!klass) return TRUE; /* hidden object cannot be overridden */
cme = rb_callable_method_entry(klass, id);
return (cme && METHOD_ENTRY_BASIC(cme)) ? TRUE : FALSE;
}
#ifdef __GNUC__
#pragma pop_macro("rb_method_basic_definition_p")
#endif
static VALUE
call_method_entry(rb_execution_context_t *ec, VALUE defined_class, VALUE obj, ID id,
const rb_callable_method_entry_t *cme, int argc, const VALUE *argv, int kw_splat)
{
VALUE passed_block_handler = vm_passed_block_handler(ec);
VALUE result = rb_vm_call_kw(ec, obj, id, argc, argv, cme, kw_splat);
vm_passed_block_handler_set(ec, passed_block_handler);
return result;
}
static VALUE
basic_obj_respond_to_missing(rb_execution_context_t *ec, VALUE klass, VALUE obj,
VALUE mid, VALUE priv)
{
VALUE defined_class, args[2];
const ID rtmid = idRespond_to_missing;
const rb_callable_method_entry_t *const cme = callable_method_entry(klass, rtmid, &defined_class);
if (!cme || METHOD_ENTRY_BASIC(cme)) return Qundef;
args[0] = mid;
args[1] = priv;
return call_method_entry(ec, defined_class, obj, rtmid, cme, 2, args, RB_NO_KEYWORDS);
}
static inline int
basic_obj_respond_to(rb_execution_context_t *ec, VALUE obj, ID id, int pub)
{
VALUE klass = CLASS_OF(obj);
VALUE ret;
switch (rb_method_boundp(klass, id, pub|BOUND_RESPONDS)) {
case 2:
return FALSE;
case 0:
ret = basic_obj_respond_to_missing(ec, klass, obj, ID2SYM(id),
pub ? Qfalse : Qtrue);
return RTEST(ret) && ret != Qundef;
default:
return TRUE;
}
}
static int
vm_respond_to(rb_execution_context_t *ec, VALUE klass, VALUE obj, ID id, int priv)
{
VALUE defined_class;
const ID resid = idRespond_to;
const rb_callable_method_entry_t *const cme = callable_method_entry(klass, resid, &defined_class);
if (!cme) return -1;
if (METHOD_ENTRY_BASIC(cme)) {
return -1;
}
else {
int argc = 1;
VALUE args[2];
VALUE result;
args[0] = ID2SYM(id);
args[1] = Qtrue;
if (priv) {
argc = rb_method_entry_arity((const rb_method_entry_t *)cme);
if (argc > 2) {
rb_raise(rb_eArgError,
"respond_to? must accept 1 or 2 arguments (requires %d)",
argc);
}
if (argc != 1) {
argc = 2;
}
else if (!NIL_P(ruby_verbose)) {
VALUE location = rb_method_entry_location((const rb_method_entry_t *)cme);
rb_warn("%"PRIsVALUE"%c""respond_to?(:%"PRIsVALUE") uses"
" the deprecated method signature, which takes one parameter",
(FL_TEST(klass, FL_SINGLETON) ? obj : klass),
(FL_TEST(klass, FL_SINGLETON) ? '.' : '#'),
QUOTE_ID(id));
if (!NIL_P(location)) {
VALUE path = RARRAY_AREF(location, 0);
VALUE line = RARRAY_AREF(location, 1);
if (!NIL_P(path)) {
rb_compile_warn(RSTRING_PTR(path), NUM2INT(line),
"respond_to? is defined here");
}
}
}
}
result = call_method_entry(ec, defined_class, obj, resid, cme, argc, args, RB_NO_KEYWORDS);
return RTEST(result);
}
}
int
rb_obj_respond_to(VALUE obj, ID id, int priv)
{
rb_execution_context_t *ec = GET_EC();
VALUE klass = CLASS_OF(obj);
int ret = vm_respond_to(ec, klass, obj, id, priv);
if (ret == -1) ret = basic_obj_respond_to(ec, obj, id, !priv);
return ret;
}
int
rb_respond_to(VALUE obj, ID id)
{
return rb_obj_respond_to(obj, id, FALSE);
}
/*
* call-seq:
* obj.respond_to?(symbol, include_all=false) -> true or false
* obj.respond_to?(string, include_all=false) -> true or false
*
* Returns +true+ if _obj_ responds to the given method. Private and
* protected methods are included in the search only if the optional
* second parameter evaluates to +true+.
*
* If the method is not implemented,
* as Process.fork on Windows, File.lchmod on GNU/Linux, etc.,
* false is returned.
*
* If the method is not defined, <code>respond_to_missing?</code>
* method is called and the result is returned.
*
* When the method name parameter is given as a string, the string is
* converted to a symbol.
*/
static VALUE
obj_respond_to(int argc, VALUE *argv, VALUE obj)
{
VALUE mid, priv;
ID id;
rb_execution_context_t *ec = GET_EC();
rb_scan_args(argc, argv, "11", &mid, &priv);
if (!(id = rb_check_id(&mid))) {
VALUE ret = basic_obj_respond_to_missing(ec, CLASS_OF(obj), obj,
rb_to_symbol(mid), priv);
if (ret == Qundef) ret = Qfalse;
return ret;
}
if (basic_obj_respond_to(ec, obj, id, !RTEST(priv)))
return Qtrue;
return Qfalse;
}
/*
* call-seq:
* obj.respond_to_missing?(symbol, include_all) -> true or false
* obj.respond_to_missing?(string, include_all) -> true or false
*
* DO NOT USE THIS DIRECTLY.
*
* Hook method to return whether the _obj_ can respond to _id_ method
* or not.
*
* When the method name parameter is given as a string, the string is
* converted to a symbol.
*
* See #respond_to?, and the example of BasicObject.
*/
static VALUE
obj_respond_to_missing(VALUE obj, VALUE mid, VALUE priv)
{
return Qfalse;
}
void
Init_Method(void)
{
//
}
void
Init_eval_method(void)
{
#undef rb_intern
#define rb_intern(str) rb_intern_const(str)
rb_define_method(rb_mKernel, "respond_to?", obj_respond_to, -1);
rb_define_method(rb_mKernel, "respond_to_missing?", obj_respond_to_missing, 2);
rb_define_method(rb_cModule, "remove_method", rb_mod_remove_method, -1);
rb_define_method(rb_cModule, "undef_method", rb_mod_undef_method, -1);
rb_define_method(rb_cModule, "alias_method", rb_mod_alias_method, 2);
rb_define_private_method(rb_cModule, "public", rb_mod_public, -1);
rb_define_private_method(rb_cModule, "protected", rb_mod_protected, -1);
rb_define_private_method(rb_cModule, "private", rb_mod_private, -1);
rb_define_private_method(rb_cModule, "module_function", rb_mod_modfunc, -1);
rb_define_private_method(rb_cModule, "ruby2_keywords", rb_mod_ruby2_keywords, -1);
rb_define_method(rb_cModule, "method_defined?", rb_mod_method_defined, -1);
rb_define_method(rb_cModule, "public_method_defined?", rb_mod_public_method_defined, -1);
rb_define_method(rb_cModule, "private_method_defined?", rb_mod_private_method_defined, -1);
rb_define_method(rb_cModule, "protected_method_defined?", rb_mod_protected_method_defined, -1);
rb_define_method(rb_cModule, "public_class_method", rb_mod_public_method, -1);
rb_define_method(rb_cModule, "private_class_method", rb_mod_private_method, -1);
rb_define_private_method(rb_singleton_class(rb_vm_top_self()),
"public", top_public, -1);
rb_define_private_method(rb_singleton_class(rb_vm_top_self()),
"private", top_private, -1);
rb_define_private_method(rb_singleton_class(rb_vm_top_self()),
"ruby2_keywords", top_ruby2_keywords, -1);
{
#define REPLICATE_METHOD(klass, id) do { \
* method.h: introduce rb_callable_method_entry_t to remove rb_control_frame_t::klass. [Bug #11278], [Bug #11279] rb_method_entry_t data belong to modules/classes. rb_method_entry_t::owner points defined module or class. module M def foo; end end In this case, owner is M. rb_callable_method_entry_t data belong to only classes. For modules, MRI creates corresponding T_ICLASS internally. rb_callable_method_entry_t can also belong to T_ICLASS. rb_callable_method_entry_t::defined_class points T_CLASS or T_ICLASS. rb_method_entry_t data for classes (not for modules) are also rb_callable_method_entry_t data because it is completely same data. In this case, rb_method_entry_t::owner == rb_method_entry_t::defined_class. For example, there are classes C and D, and incldues M, class C; include M; end class D; include M; end then, two T_ICLASS objects for C's super class and D's super class will be created. When C.new.foo is called, then M#foo is searcheed and rb_callable_method_t data is used by VM to invoke M#foo. rb_method_entry_t data is only one for M#foo. However, rb_callable_method_entry_t data are two (and can be more). It is proportional to the number of including (and prepending) classes (the number of T_ICLASS which point to the module). Now, created rb_callable_method_entry_t are collected when the original module M was modified. We can think it is a cache. We need to select what kind of method entry data is needed. To operate definition, then you need to use rb_method_entry_t. You can access them by the following functions. * rb_method_entry(VALUE klass, ID id); * rb_method_entry_with_refinements(VALUE klass, ID id); * rb_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method(VALUE refinements, const rb_method_entry_t *me); To invoke methods, then you need to use rb_callable_method_entry_t which you can get by the following APIs corresponding to the above listed functions. * rb_callable_method_entry(VALUE klass, ID id); * rb_callable_method_entry_with_refinements(VALUE klass, ID id); * rb_callable_method_entry_without_refinements(VALUE klass, ID id); * rb_resolve_refined_method_callable(VALUE refinements, const rb_callable_method_entry_t *me); VM pushes rb_callable_method_entry_t, so that rb_vm_frame_method_entry() returns rb_callable_method_entry_t. You can check a super class of current method by rb_callable_method_entry_t::defined_class. * method.h: renamed from rb_method_entry_t::klass to rb_method_entry_t::owner. * internal.h: add rb_classext_struct::callable_m_tbl to cache rb_callable_method_entry_t data. We need to consider abotu this field again because it is only active for T_ICLASS. * class.c (method_entry_i): ditto. * class.c (rb_define_attr): rb_method_entry() does not takes defiend_class_ptr. * gc.c (mark_method_entry): mark RCLASS_CALLABLE_M_TBL() for T_ICLASS. * cont.c (fiber_init): rb_control_frame_t::klass is removed. * proc.c: fix `struct METHOD' data structure because rb_callable_method_t has all information. * vm_core.h: remove several fields. * rb_control_frame_t::klass. * rb_block_t::klass. And catch up changes. * eval.c: catch up changes. * gc.c: ditto. * insns.def: ditto. * vm.c: ditto. * vm_args.c: ditto. * vm_backtrace.c: ditto. * vm_dump.c: ditto. * vm_eval.c: ditto. * vm_insnhelper.c: ditto. * vm_method.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51126 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-03 14:24:50 +03:00
const rb_method_entry_t *me = rb_method_entry((klass), (id)); \
rb_method_entry_set((klass), (id), me, METHOD_ENTRY_VISI(me)); \
} while (0)
REPLICATE_METHOD(rb_eException, idMethodMissing);
REPLICATE_METHOD(rb_eException, idRespond_to);
REPLICATE_METHOD(rb_eException, idRespond_to_missing);
}
}