ruby/vm_eval.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

2617 строки
74 KiB
C
Исходник Обычный вид История

/**********************************************************************
vm_eval.c -
$Author$
created at: Sat May 24 16:02:32 JST 2008
Copyright (C) 1993-2007 Yukihiro Matsumoto
Copyright (C) 2000 Network Applied Communication Laboratory, Inc.
Copyright (C) 2000 Information-technology Promotion Agency, Japan
**********************************************************************/
#include "internal/thread.h"
struct local_var_list {
VALUE tbl;
};
static inline VALUE method_missing(rb_execution_context_t *ec, VALUE obj, ID id, int argc, const VALUE *argv, enum method_missing_reason call_status, int kw_splat);
static inline VALUE vm_yield_with_cref(rb_execution_context_t *ec, int argc, const VALUE *argv, int kw_splat, const rb_cref_t *cref, int is_lambda);
static inline VALUE vm_yield(rb_execution_context_t *ec, int argc, const VALUE *argv, int kw_splat);
static inline VALUE vm_yield_with_block(rb_execution_context_t *ec, int argc, const VALUE *argv, VALUE block_handler, int kw_splat);
static inline VALUE vm_yield_force_blockarg(rb_execution_context_t *ec, VALUE args);
VALUE vm_exec(rb_execution_context_t *ec, bool jit_enable_p);
static void vm_set_eval_stack(rb_execution_context_t * th, const rb_iseq_t *iseq, const rb_cref_t *cref, const struct rb_block *base_block);
static int vm_collect_local_variables_in_heap(const VALUE *dfp, const struct local_var_list *vars);
static VALUE rb_eUncaughtThrow;
static ID id_result, id_tag, id_value;
#define id_mesg idMesg
typedef enum call_type {
CALL_PUBLIC,
CALL_FCALL,
CALL_VCALL,
CALL_PUBLIC_KW,
CALL_FCALL_KW,
CALL_TYPE_MAX
} call_type;
static VALUE send_internal(int argc, const VALUE *argv, VALUE recv, call_type scope);
static VALUE vm_call0_body(rb_execution_context_t* ec, struct rb_calling_info *calling, const VALUE *argv);
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
#ifndef MJIT_HEADER
MJIT_FUNC_EXPORTED VALUE
rb_vm_call0(rb_execution_context_t *ec, VALUE recv, ID id, int argc, const VALUE *argv, const rb_callable_method_entry_t *cme, int kw_splat)
{
struct rb_calling_info calling = {
.ci = &VM_CI_ON_STACK(id, kw_splat ? VM_CALL_KW_SPLAT : 0, argc, NULL),
.cc = &VM_CC_ON_STACK(Qfalse, vm_call_general, {{ 0 }}, cme),
.block_handler = vm_passed_block_handler(ec),
.recv = recv,
.argc = argc,
.kw_splat = kw_splat,
};
return vm_call0_body(ec, &calling, argv);
}
MJIT_FUNC_EXPORTED VALUE
rb_vm_call_with_refinements(rb_execution_context_t *ec, VALUE recv, ID id, int argc, const VALUE *argv, int kw_splat)
{
const rb_callable_method_entry_t *me =
rb_callable_method_entry_with_refinements(CLASS_OF(recv), id, NULL);
if (me) {
return rb_vm_call0(ec, recv, id, argc, argv, me, kw_splat);
}
else {
/* fallback to funcall (e.g. method_missing) */
return rb_funcallv(recv, id, argc, argv);
}
}
static inline VALUE
vm_call0_cc(rb_execution_context_t *ec, VALUE recv, ID id, int argc, const VALUE *argv, const struct rb_callcache *cc, int kw_splat)
{
struct rb_calling_info calling = {
.ci = &VM_CI_ON_STACK(id, kw_splat ? VM_CALL_KW_SPLAT : 0, argc, NULL),
.cc = cc,
.block_handler = vm_passed_block_handler(ec),
.recv = recv,
.argc = argc,
.kw_splat = kw_splat,
};
return vm_call0_body(ec, &calling, argv);
}
static VALUE
vm_call0_cme(rb_execution_context_t *ec, struct rb_calling_info *calling, const VALUE *argv, const rb_callable_method_entry_t *cme)
{
calling->cc = &VM_CC_ON_STACK(Qfalse, vm_call_general, {{ 0 }}, cme);
return vm_call0_body(ec, calling, argv);
}
static VALUE
vm_call0_super(rb_execution_context_t *ec, struct rb_calling_info *calling, const VALUE *argv, VALUE klass, enum method_missing_reason ex)
{
ID mid = vm_ci_mid(calling->ci);
klass = RCLASS_SUPER(klass);
if (klass) {
const rb_callable_method_entry_t *cme = rb_callable_method_entry(klass, mid);
if (cme) {
RUBY_VM_CHECK_INTS(ec);
return vm_call0_cme(ec, calling, argv, cme);
}
}
vm_passed_block_handler_set(ec, calling->block_handler);
return method_missing(ec, calling->recv, mid, calling->argc, argv, ex, calling->kw_splat);
}
static VALUE
vm_call0_cfunc_with_frame(rb_execution_context_t* ec, struct rb_calling_info *calling, const VALUE *argv)
{
const struct rb_callinfo *ci = calling->ci;
VALUE val;
const rb_callable_method_entry_t *me = vm_cc_cme(calling->cc);
const rb_method_cfunc_t *cfunc = UNALIGNED_MEMBER_PTR(me->def, body.cfunc);
int len = cfunc->argc;
VALUE recv = calling->recv;
int argc = calling->argc;
ID mid = vm_ci_mid(ci);
VALUE block_handler = calling->block_handler;
int frame_flags = VM_FRAME_MAGIC_CFUNC | VM_FRAME_FLAG_CFRAME | VM_ENV_FLAG_LOCAL;
if (calling->kw_splat) {
if (argc > 0 && RB_TYPE_P(argv[argc-1], T_HASH) && RHASH_EMPTY_P(argv[argc-1])) {
argc--;
}
else {
frame_flags |= VM_FRAME_FLAG_CFRAME_KW;
}
}
RUBY_DTRACE_CMETHOD_ENTRY_HOOK(ec, me->owner, me->def->original_id);
EXEC_EVENT_HOOK(ec, RUBY_EVENT_C_CALL, recv, me->def->original_id, mid, me->owner, Qnil);
{
rb_control_frame_t *reg_cfp = ec->cfp;
vm_push_frame(ec, 0, frame_flags, recv,
block_handler, (VALUE)me,
0, reg_cfp->sp, 0, 0);
if (len >= 0) rb_check_arity(argc, len, len);
val = (*cfunc->invoker)(recv, argc, argv, cfunc->func);
CHECK_CFP_CONSISTENCY("vm_call0_cfunc_with_frame");
rb_vm_pop_frame(ec);
}
EXEC_EVENT_HOOK(ec, RUBY_EVENT_C_RETURN, recv, me->def->original_id, mid, me->owner, val);
RUBY_DTRACE_CMETHOD_RETURN_HOOK(ec, me->owner, me->def->original_id);
return val;
}
static VALUE
vm_call0_cfunc(rb_execution_context_t *ec, struct rb_calling_info *calling, const VALUE *argv)
{
return vm_call0_cfunc_with_frame(ec, calling, argv);
}
static void
vm_call_check_arity(struct rb_calling_info *calling, int argc, const VALUE *argv)
{
if (calling->kw_splat &&
calling->argc > 0 &&
RB_TYPE_P(argv[calling->argc-1], T_HASH) &&
RHASH_EMPTY_P(argv[calling->argc-1])) {
calling->argc--;
}
rb_check_arity(calling->argc, argc, argc);
}
/* `ci' should point temporal value (on stack value) */
static VALUE
vm_call0_body(rb_execution_context_t *ec, struct rb_calling_info *calling, const VALUE *argv)
{
const struct rb_callinfo *ci = calling->ci;
const struct rb_callcache *cc = calling->cc;
VALUE ret;
retry:
switch (vm_cc_cme(cc)->def->type) {
case VM_METHOD_TYPE_ISEQ:
{
rb_control_frame_t *reg_cfp = ec->cfp;
int i;
CHECK_VM_STACK_OVERFLOW(reg_cfp, calling->argc + 1);
vm_check_canary(ec, reg_cfp->sp);
*reg_cfp->sp++ = calling->recv;
for (i = 0; i < calling->argc; i++) {
*reg_cfp->sp++ = argv[i];
}
vm_call_iseq_setup(ec, reg_cfp, calling);
VM_ENV_FLAGS_SET(ec->cfp->ep, VM_FRAME_FLAG_FINISH);
return vm_exec(ec, true); /* CHECK_INTS in this function */
}
case VM_METHOD_TYPE_NOTIMPLEMENTED:
case VM_METHOD_TYPE_CFUNC:
ret = vm_call0_cfunc(ec, calling, argv);
goto success;
case VM_METHOD_TYPE_ATTRSET:
vm_call_check_arity(calling, 1, argv);
2021-09-18 10:15:24 +03:00
VM_CALL_METHOD_ATTR(ret,
rb_ivar_set(calling->recv, vm_cc_cme(cc)->def->body.attr.id, argv[0]),
(void)0);
goto success;
case VM_METHOD_TYPE_IVAR:
vm_call_check_arity(calling, 0, argv);
2021-09-18 10:15:24 +03:00
VM_CALL_METHOD_ATTR(ret,
rb_attr_get(calling->recv, vm_cc_cme(cc)->def->body.attr.id),
(void)0);
goto success;
case VM_METHOD_TYPE_BMETHOD:
ret = vm_call_bmethod_body(ec, calling, argv);
goto success;
case VM_METHOD_TYPE_ZSUPER:
{
VALUE klass = RCLASS_ORIGIN(vm_cc_cme(cc)->defined_class);
return vm_call0_super(ec, calling, argv, klass, MISSING_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
case VM_METHOD_TYPE_REFINED:
{
const rb_callable_method_entry_t *cme = vm_cc_cme(cc);
* 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 (cme->def->body.refined.orig_me) {
const rb_callable_method_entry_t *orig_cme = refined_method_callable_without_refinement(cme);
return vm_call0_cme(ec, calling, argv, orig_cme);
}
VALUE klass = cme->defined_class;
return vm_call0_super(ec, calling, argv, klass, 0);
}
case VM_METHOD_TYPE_ALIAS:
{
const rb_callable_method_entry_t *cme = vm_cc_cme(cc);
const rb_callable_method_entry_t *orig_cme = aliased_callable_method_entry(cme);
if (cme == orig_cme) rb_bug("same!!");
if (vm_cc_markable(cc)) {
return vm_call0_cme(ec, calling, argv, orig_cme);
}
else {
*((const rb_callable_method_entry_t **)&cc->cme_) = orig_cme;
goto retry;
}
}
case VM_METHOD_TYPE_MISSING:
{
vm_passed_block_handler_set(ec, calling->block_handler);
return method_missing(ec, calling->recv, vm_ci_mid(ci), calling->argc,
argv, MISSING_NOENTRY, calling->kw_splat);
}
case VM_METHOD_TYPE_OPTIMIZED:
switch (vm_cc_cme(cc)->def->body.optimized.type) {
case OPTIMIZED_METHOD_TYPE_SEND:
ret = send_internal(calling->argc, argv, calling->recv, calling->kw_splat ? CALL_FCALL_KW : CALL_FCALL);
goto success;
case OPTIMIZED_METHOD_TYPE_CALL:
{
rb_proc_t *proc;
GetProcPtr(calling->recv, proc);
ret = rb_vm_invoke_proc(ec, proc, calling->argc, argv, calling->kw_splat, calling->block_handler);
goto success;
}
case OPTIMIZED_METHOD_TYPE_STRUCT_AREF:
vm_call_check_arity(calling, 0, argv);
ret = vm_call_opt_struct_aref0(ec, calling);
goto success;
case OPTIMIZED_METHOD_TYPE_STRUCT_ASET:
vm_call_check_arity(calling, 1, argv);
ret = vm_call_opt_struct_aset0(ec, calling, argv[0]);
goto success;
default:
rb_bug("vm_call0: unsupported optimized method type (%d)", vm_cc_cme(cc)->def->body.optimized.type);
}
break;
case VM_METHOD_TYPE_UNDEF:
break;
}
rb_bug("vm_call0: unsupported method type (%d)", vm_cc_cme(cc)->def->type);
return Qundef;
success:
RUBY_VM_CHECK_INTS(ec);
return ret;
}
MJIT_FUNC_EXPORTED VALUE
rb_vm_call_kw(rb_execution_context_t *ec, VALUE recv, VALUE id, int argc, const VALUE *argv, const rb_callable_method_entry_t *me, int kw_splat)
{
return rb_vm_call0(ec, recv, id, argc, argv, me, kw_splat);
}
static inline VALUE
vm_call_super(rb_execution_context_t *ec, int argc, const VALUE *argv, int kw_splat)
{
VALUE recv = ec->cfp->self;
VALUE klass;
ID id;
rb_control_frame_t *cfp = ec->cfp;
* 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 *me = rb_vm_frame_method_entry(cfp);
if (VM_FRAME_RUBYFRAME_P(cfp)) {
rb_bug("vm_call_super: should not be reached");
}
* 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
klass = RCLASS_ORIGIN(me->defined_class);
klass = RCLASS_SUPER(klass);
* 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
id = me->def->original_id;
me = rb_callable_method_entry(klass, id);
if (!me) {
return method_missing(ec, recv, id, argc, argv, MISSING_SUPER, kw_splat);
}
return rb_vm_call_kw(ec, recv, id, argc, argv, me, kw_splat);
}
VALUE
rb_call_super_kw(int argc, const VALUE *argv, int kw_splat)
{
rb_execution_context_t *ec = GET_EC();
PASS_PASSED_BLOCK_HANDLER_EC(ec);
return vm_call_super(ec, argc, argv, kw_splat);
}
VALUE
rb_call_super(int argc, const VALUE *argv)
{
2021-09-21 16:59:35 +03:00
return rb_call_super_kw(argc, argv, RB_NO_KEYWORDS);
}
VALUE
rb_current_receiver(void)
{
const rb_execution_context_t *ec = GET_EC();
rb_control_frame_t *cfp;
if (!ec || !(cfp = ec->cfp)) {
rb_raise(rb_eRuntimeError, "no self, no life");
}
return cfp->self;
}
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
#endif /* #ifndef MJIT_HEADER */
static inline void
stack_check(rb_execution_context_t *ec)
{
if (!rb_ec_raised_p(ec, RAISED_STACKOVERFLOW) &&
rb_ec_stack_check(ec)) {
rb_ec_raised_set(ec, RAISED_STACKOVERFLOW);
rb_ec_stack_overflow(ec, FALSE);
}
}
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
#ifndef MJIT_HEADER
void
rb_check_stack_overflow(void)
{
#ifndef RB_THREAD_LOCAL_SPECIFIER
if (!ruby_current_ec_key) return;
#endif
rb_execution_context_t *ec = GET_EC();
if (ec) stack_check(ec);
}
NORETURN(static void uncallable_object(VALUE recv, 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
static inline const rb_callable_method_entry_t *rb_search_method_entry(VALUE recv, ID mid);
static inline enum method_missing_reason rb_method_call_status(rb_execution_context_t *ec, const rb_callable_method_entry_t *me, call_type scope, VALUE self);
static const struct rb_callcache *
cc_new(VALUE klass, ID mid, int argc, const rb_callable_method_entry_t *cme)
{
const struct rb_callcache *cc = NULL;
RB_VM_LOCK_ENTER();
{
struct rb_class_cc_entries *ccs;
struct rb_id_table *cc_tbl = RCLASS_CC_TBL(klass);
VALUE ccs_data;
if (rb_id_table_lookup(cc_tbl, mid, &ccs_data)) {
// ok
ccs = (struct rb_class_cc_entries *)ccs_data;
}
else {
ccs = vm_ccs_create(klass, cme);
rb_id_table_insert(cc_tbl, mid, (VALUE)ccs);
}
for (int i=0; i<ccs->len; i++) {
cc = ccs->entries[i].cc;
if (vm_cc_cme(cc) == cme) {
break;
}
cc = NULL;
}
if (cc == NULL) {
const struct rb_callinfo *ci = vm_ci_new(mid, 0, argc, NULL); // TODO: proper ci
cc = vm_cc_new(klass, cme, vm_call_general);
METHOD_ENTRY_CACHED_SET((struct rb_callable_method_entry_struct *)cme);
vm_ccs_push(klass, ccs, ci, cc);
}
}
RB_VM_LOCK_LEAVE();
return cc;
}
static VALUE
gccct_hash(VALUE klass, ID mid)
{
return (klass >> 3) ^ (VALUE)mid;
}
NOINLINE(static const struct rb_callcache *gccct_method_search_slowpath(rb_vm_t *vm, VALUE klass, ID mid, int argc, unsigned int index));
static const struct rb_callcache *
gccct_method_search_slowpath(rb_vm_t *vm, VALUE klass, ID mid, int argc, unsigned int index)
{
const rb_callable_method_entry_t *cme = rb_callable_method_entry(klass, mid);
const struct rb_callcache *cc;
if (cme != NULL) {
cc = cc_new(klass, mid, argc, cme);
}
else {
cc = NULL;
}
return vm->global_cc_cache_table[index] = cc;
}
static inline const struct rb_callcache *
gccct_method_search(rb_execution_context_t *ec, VALUE recv, ID mid, int argc)
{
VALUE klass;
if (!SPECIAL_CONST_P(recv)) {
klass = RBASIC_CLASS(recv);
if (UNLIKELY(!klass)) uncallable_object(recv, mid);
}
else {
klass = CLASS_OF(recv);
}
// search global method cache
unsigned int index = (unsigned int)(gccct_hash(klass, mid) % VM_GLOBAL_CC_CACHE_TABLE_SIZE);
rb_vm_t *vm = rb_ec_vm_ptr(ec);
const struct rb_callcache *cc = vm->global_cc_cache_table[index];
if (LIKELY(cc)) {
if (LIKELY(vm_cc_class_check(cc, klass))) {
const rb_callable_method_entry_t *cme = vm_cc_cme(cc);
if (LIKELY(!METHOD_ENTRY_INVALIDATED(cme) &&
cme->called_id == mid)) {
`Primitive.mandatory_only?` for fast path Compare with the C methods, A built-in methods written in Ruby is slower if only mandatory parameters are given because it needs to check the argumens and fill default values for optional and keyword parameters (C methods can check the number of parameters with `argc`, so there are no overhead). Passing mandatory arguments are common (optional arguments are exceptional, in many cases) so it is important to provide the fast path for such common cases. `Primitive.mandatory_only?` is a special builtin function used with `if` expression like that: ```ruby def self.at(time, subsec = false, unit = :microsecond, in: nil) if Primitive.mandatory_only? Primitive.time_s_at1(time) else Primitive.time_s_at(time, subsec, unit, Primitive.arg!(:in)) end end ``` and it makes two ISeq, ``` def self.at(time, subsec = false, unit = :microsecond, in: nil) Primitive.time_s_at(time, subsec, unit, Primitive.arg!(:in)) end def self.at(time) Primitive.time_s_at1(time) end ``` and (2) is pointed by (1). Note that `Primitive.mandatory_only?` should be used only in a condition of an `if` statement and the `if` statement should be equal to the methdo body (you can not put any expression before and after the `if` statement). A method entry with `mandatory_only?` (`Time.at` on the above case) is marked as `iseq_overload`. When the method will be dispatch only with mandatory arguments (`Time.at(0)` for example), make another method entry with ISeq (2) as mandatory only method entry and it will be cached in an inline method cache. The idea is similar discussed in https://bugs.ruby-lang.org/issues/16254 but it only checks mandatory parameters or more, because many cases only mandatory parameters are given. If we find other cases (optional or keyword parameters are used frequently and it hurts performance), we can extend the feature.
2021-11-12 20:12:20 +03:00
VM_ASSERT(vm_cc_check_cme(cc, rb_callable_method_entry(klass, mid)));
RB_DEBUG_COUNTER_INC(gccct_hit);
return cc;
}
}
}
else {
RB_DEBUG_COUNTER_INC(gccct_null);
}
RB_DEBUG_COUNTER_INC(gccct_miss);
return gccct_method_search_slowpath(vm, klass, mid, argc, index);
}
/*!
* \internal
* calls the specified method.
*
* This function is called by functions in rb_call* family.
* \param ec current execution context
* \param recv receiver of the method
* \param mid an ID that represents the name of the method
* \param argc the number of method arguments
* \param argv a pointer to an array of method arguments
* \param scope
* \param self self in the caller. Qundef means no self is considered and
* protected methods cannot be called
*
* \note \a self is used in order to controlling access to protected methods.
*/
static inline VALUE
rb_call0(rb_execution_context_t *ec,
VALUE recv, ID mid, int argc, const VALUE *argv,
call_type call_scope, VALUE self)
{
enum method_missing_reason call_status;
call_type scope = call_scope;
int kw_splat = RB_NO_KEYWORDS;
switch (scope) {
case CALL_PUBLIC_KW:
scope = CALL_PUBLIC;
kw_splat = 1;
break;
case CALL_FCALL_KW:
scope = CALL_FCALL;
kw_splat = 1;
break;
default:
break;
}
const struct rb_callcache *cc = gccct_method_search(ec, recv, mid, argc);
if (scope == CALL_PUBLIC) {
RB_DEBUG_COUNTER_INC(call0_public);
const rb_callable_method_entry_t *cc_cme = cc ? vm_cc_cme(cc) : NULL;
const rb_callable_method_entry_t *cme = callable_method_entry_refeinements0(CLASS_OF(recv), mid, NULL, true, cc_cme);
call_status = rb_method_call_status(ec, cme, scope, self);
if (UNLIKELY(call_status != MISSING_NONE)) {
return method_missing(ec, recv, mid, argc, argv, call_status, kw_splat);
}
else if (UNLIKELY(cc_cme != cme)) { // refinement is solved
stack_check(ec);
return rb_vm_call_kw(ec, recv, mid, argc, argv, cme, kw_splat);
}
}
else {
RB_DEBUG_COUNTER_INC(call0_other);
call_status = rb_method_call_status(ec, cc ? vm_cc_cme(cc) : NULL, scope, self);
if (UNLIKELY(call_status != MISSING_NONE)) {
return method_missing(ec, recv, mid, argc, argv, call_status, kw_splat);
}
}
stack_check(ec);
return vm_call0_cc(ec, recv, mid, argc, argv, cc, kw_splat);
}
struct rescue_funcall_args {
VALUE defined_class;
VALUE recv;
ID mid;
rb_execution_context_t *ec;
const rb_callable_method_entry_t *cme;
unsigned int respond: 1;
unsigned int respond_to_missing: 1;
int argc;
const VALUE *argv;
int kw_splat;
};
static VALUE
check_funcall_exec(VALUE v)
{
struct rescue_funcall_args *args = (void *)v;
return call_method_entry(args->ec, args->defined_class,
args->recv, idMethodMissing,
args->cme, args->argc, args->argv, args->kw_splat);
}
static VALUE
check_funcall_failed(VALUE v, VALUE e)
{
struct rescue_funcall_args *args = (void *)v;
int ret = args->respond;
if (!ret) {
switch (method_boundp(args->defined_class, args->mid,
BOUND_PRIVATE|BOUND_RESPONDS)) {
case 2:
ret = TRUE;
break;
case 0:
ret = args->respond_to_missing;
break;
default:
ret = FALSE;
break;
}
}
if (ret) {
rb_exc_raise(e);
}
return Qundef;
}
static int
check_funcall_respond_to(rb_execution_context_t *ec, VALUE klass, VALUE recv, ID mid)
{
return vm_respond_to(ec, klass, recv, mid, TRUE);
}
static int
check_funcall_callable(rb_execution_context_t *ec, const rb_callable_method_entry_t *me)
{
return rb_method_call_status(ec, me, CALL_FCALL, ec->cfp->self) == MISSING_NONE;
}
static VALUE
check_funcall_missing(rb_execution_context_t *ec, VALUE klass, VALUE recv, ID mid, int argc, const VALUE *argv, int respond, VALUE def, int kw_splat)
{
struct rescue_funcall_args args;
const rb_callable_method_entry_t *cme;
VALUE ret = Qundef;
ret = basic_obj_respond_to_missing(ec, klass, recv,
ID2SYM(mid), Qtrue);
if (!RTEST(ret)) return def;
args.respond = respond > 0;
2022-11-15 07:24:08 +03:00
args.respond_to_missing = !UNDEF_P(ret);
ret = def;
cme = callable_method_entry(klass, idMethodMissing, &args.defined_class);
if (cme && !METHOD_ENTRY_BASIC(cme)) {
VALUE argbuf, *new_args = ALLOCV_N(VALUE, argbuf, argc+1);
new_args[0] = ID2SYM(mid);
#ifdef __GLIBC__
if (!argv) {
static const VALUE buf = Qfalse;
VM_ASSERT(argc == 0);
argv = &buf;
}
#endif
MEMCPY(new_args+1, argv, VALUE, argc);
ec->method_missing_reason = MISSING_NOENTRY;
args.ec = ec;
args.recv = recv;
args.cme = cme;
args.mid = mid;
args.argc = argc + 1;
args.argv = new_args;
args.kw_splat = kw_splat;
ret = rb_rescue2(check_funcall_exec, (VALUE)&args,
check_funcall_failed, (VALUE)&args,
rb_eNoMethodError, (VALUE)0);
ALLOCV_END(argbuf);
}
return ret;
}
static VALUE rb_check_funcall_default_kw(VALUE recv, ID mid, int argc, const VALUE *argv, VALUE def, int kw_splat);
VALUE
rb_check_funcall_kw(VALUE recv, ID mid, int argc, const VALUE *argv, int kw_splat)
{
return rb_check_funcall_default_kw(recv, mid, argc, argv, Qundef, kw_splat);
}
VALUE
rb_check_funcall(VALUE recv, ID mid, int argc, const VALUE *argv)
{
return rb_check_funcall_default_kw(recv, mid, argc, argv, Qundef, RB_NO_KEYWORDS);
}
static VALUE
rb_check_funcall_default_kw(VALUE recv, ID mid, int argc, const VALUE *argv, VALUE def, int kw_splat)
{
VM_ASSERT(ruby_thread_has_gvl_p());
VALUE klass = CLASS_OF(recv);
* 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 *me;
rb_execution_context_t *ec = GET_EC();
int respond = check_funcall_respond_to(ec, klass, recv, mid);
if (!respond)
return 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
me = rb_search_method_entry(recv, mid);
if (!check_funcall_callable(ec, me)) {
VALUE ret = check_funcall_missing(ec, klass, recv, mid, argc, argv,
respond, def, kw_splat);
2022-11-15 07:24:08 +03:00
if (UNDEF_P(ret)) ret = def;
return ret;
}
stack_check(ec);
return rb_vm_call_kw(ec, recv, mid, argc, argv, me, kw_splat);
}
VALUE
rb_check_funcall_default(VALUE recv, ID mid, int argc, const VALUE *argv, VALUE def)
{
return rb_check_funcall_default_kw(recv, mid, argc, argv, def, RB_NO_KEYWORDS);
}
VALUE
rb_check_funcall_with_hook_kw(VALUE recv, ID mid, int argc, const VALUE *argv,
rb_check_funcall_hook *hook, VALUE arg, int kw_splat)
{
VALUE klass = CLASS_OF(recv);
* 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 *me;
rb_execution_context_t *ec = GET_EC();
int respond = check_funcall_respond_to(ec, klass, recv, mid);
if (!respond) {
(*hook)(FALSE, recv, mid, argc, argv, arg);
return Qundef;
}
* 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
me = rb_search_method_entry(recv, mid);
if (!check_funcall_callable(ec, me)) {
VALUE ret = check_funcall_missing(ec, klass, recv, mid, argc, argv,
respond, Qundef, kw_splat);
2022-11-15 07:24:08 +03:00
(*hook)(!UNDEF_P(ret), recv, mid, argc, argv, arg);
return ret;
}
stack_check(ec);
(*hook)(TRUE, recv, mid, argc, argv, arg);
return rb_vm_call_kw(ec, recv, mid, argc, argv, me, kw_splat);
}
VALUE
rb_check_funcall_with_hook(VALUE recv, ID mid, int argc, const VALUE *argv,
rb_check_funcall_hook *hook, VALUE arg)
{
return rb_check_funcall_with_hook_kw(recv, mid, argc, argv, hook, arg, RB_NO_KEYWORDS);
}
const char *
rb_type_str(enum ruby_value_type type)
{
#define type_case(t) t: return #t
switch (type) {
case type_case(T_NONE);
case type_case(T_OBJECT);
case type_case(T_CLASS);
case type_case(T_MODULE);
case type_case(T_FLOAT);
case type_case(T_STRING);
case type_case(T_REGEXP);
case type_case(T_ARRAY);
case type_case(T_HASH);
case type_case(T_STRUCT);
case type_case(T_BIGNUM);
case type_case(T_FILE);
case type_case(T_DATA);
case type_case(T_MATCH);
case type_case(T_COMPLEX);
case type_case(T_RATIONAL);
case type_case(T_NIL);
case type_case(T_TRUE);
case type_case(T_FALSE);
case type_case(T_SYMBOL);
case type_case(T_FIXNUM);
case type_case(T_IMEMO);
case type_case(T_UNDEF);
case type_case(T_NODE);
case type_case(T_ICLASS);
case type_case(T_ZOMBIE);
case type_case(T_MOVED);
case T_MASK: break;
}
#undef type_case
return NULL;
}
static void
uncallable_object(VALUE recv, ID mid)
{
VALUE flags;
int type;
const char *typestr;
VALUE mname = rb_id2str(mid);
if (SPECIAL_CONST_P(recv)) {
rb_raise(rb_eNotImpError,
"method `%"PRIsVALUE"' called on unexpected immediate object (%p)",
mname, (void *)recv);
}
else if ((flags = RBASIC(recv)->flags) == 0) {
rb_raise(rb_eNotImpError,
"method `%"PRIsVALUE"' called on terminated object (%p)",
mname, (void *)recv);
}
else if (!(typestr = rb_type_str(type = BUILTIN_TYPE(recv)))) {
rb_raise(rb_eNotImpError,
"method `%"PRIsVALUE"' called on broken T_?""?""?(0x%02x) object"
" (%p flags=0x%"PRIxVALUE")",
mname, type, (void *)recv, flags);
}
else if (T_OBJECT <= type && type < T_NIL) {
rb_raise(rb_eNotImpError,
"method `%"PRIsVALUE"' called on hidden %s object"
" (%p flags=0x%"PRIxVALUE")",
mname, typestr, (void *)recv, flags);
}
else {
rb_raise(rb_eNotImpError,
"method `%"PRIsVALUE"' called on unexpected %s object"
" (%p flags=0x%"PRIxVALUE")",
mname, typestr, (void *)recv, flags);
}
}
* 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 inline const rb_callable_method_entry_t *
rb_search_method_entry(VALUE recv, ID mid)
{
VALUE klass = CLASS_OF(recv);
if (!klass) uncallable_object(recv, 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 rb_callable_method_entry(klass, mid);
}
static inline enum method_missing_reason
rb_method_call_status(rb_execution_context_t *ec, const rb_callable_method_entry_t *me, call_type scope, VALUE self)
{
2021-01-29 06:54:43 +03:00
if (UNLIKELY(UNDEFINED_METHOD_ENTRY_P(me))) {
goto undefined;
}
2021-01-29 06:54:43 +03:00
else if (UNLIKELY(me->def->type == VM_METHOD_TYPE_REFINED)) {
me = rb_resolve_refined_method_callable(Qnil, me);
if (UNDEFINED_METHOD_ENTRY_P(me)) goto undefined;
}
* 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
2021-01-29 06:54:43 +03:00
rb_method_visibility_t visi = METHOD_ENTRY_VISI(me);
2021-01-29 06:54:43 +03:00
/* receiver specified form for private method */
if (UNLIKELY(visi != METHOD_VISI_PUBLIC)) {
if (me->def->original_id == idMethodMissing) {
return MISSING_NONE;
}
else if (visi == METHOD_VISI_PRIVATE &&
scope == CALL_PUBLIC) {
return MISSING_PRIVATE;
}
/* self must be kind of a specified form for protected method */
else if (visi == METHOD_VISI_PROTECTED &&
scope == CALL_PUBLIC) {
2021-01-29 06:54:43 +03:00
VALUE defined_class = me->owner;
if (RB_TYPE_P(defined_class, T_ICLASS)) {
defined_class = RBASIC(defined_class)->klass;
}
2022-11-15 07:24:08 +03:00
if (UNDEF_P(self) || !rb_obj_is_kind_of(self, defined_class)) {
2021-01-29 06:54:43 +03:00
return MISSING_PROTECTED;
}
}
}
return MISSING_NONE;
2021-01-29 06:54:43 +03:00
undefined:
return scope == CALL_VCALL ? MISSING_VCALL : MISSING_NOENTRY;
}
/*!
* \internal
* calls the specified method.
*
* This function is called by functions in rb_call* family.
* \param recv receiver
* \param mid an ID that represents the name of the method
* \param argc the number of method arguments
* \param argv a pointer to an array of method arguments
* \param scope
*/
static inline VALUE
rb_call(VALUE recv, ID mid, int argc, const VALUE *argv, call_type scope)
{
rb_execution_context_t *ec = GET_EC();
return rb_call0(ec, recv, mid, argc, argv, scope, ec->cfp->self);
}
NORETURN(static void raise_method_missing(rb_execution_context_t *ec, int argc, const VALUE *argv,
VALUE obj, enum method_missing_reason call_status));
/*
* call-seq:
* obj.method_missing(symbol [, *args] ) -> result
*
* Invoked by Ruby when <i>obj</i> is sent a message it cannot handle.
* <i>symbol</i> is the symbol for the method called, and <i>args</i>
* are any arguments that were passed to it. By default, the interpreter
* raises an error when this method is called. However, it is possible
* to override the method to provide more dynamic behavior.
* If it is decided that a particular method should not be handled, then
* <i>super</i> should be called, so that ancestors can pick up the
* missing method.
* The example below creates
* a class <code>Roman</code>, which responds to methods with names
* consisting of roman numerals, returning the corresponding integer
* values.
*
* class Roman
* def roman_to_int(str)
* # ...
* end
*
* def method_missing(symbol, *args)
* str = symbol.id2name
* begin
* roman_to_int(str)
* rescue
* super(symbol, *args)
* end
* end
* end
*
* r = Roman.new
* r.iv #=> 4
* r.xxiii #=> 23
* r.mm #=> 2000
* r.foo #=> NoMethodError
*/
static VALUE
rb_method_missing(int argc, const VALUE *argv, VALUE obj)
{
rb_execution_context_t *ec = GET_EC();
raise_method_missing(ec, argc, argv, obj, ec->method_missing_reason);
UNREACHABLE_RETURN(Qnil);
}
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 VALUE
rb_make_no_method_exception(VALUE exc, VALUE format, VALUE obj,
int argc, const VALUE *argv, int priv)
{
VALUE name = argv[0];
if (!format) {
format = rb_fstring_lit("undefined method `%s' for %s%s%s");
}
if (exc == rb_eNoMethodError) {
VALUE args = rb_ary_new4(argc - 1, argv + 1);
return rb_nomethod_err_new(format, obj, name, args, priv);
}
else {
return rb_name_err_new(format, obj, name);
}
}
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
#endif /* #ifndef MJIT_HEADER */
static void
raise_method_missing(rb_execution_context_t *ec, int argc, const VALUE *argv, VALUE obj,
enum method_missing_reason last_call_status)
{
VALUE exc = rb_eNoMethodError;
VALUE format = 0;
if (UNLIKELY(argc == 0)) {
rb_raise(rb_eArgError, "no method name given");
}
else if (UNLIKELY(!SYMBOL_P(argv[0]))) {
const VALUE e = rb_eArgError; /* TODO: TypeError? */
rb_raise(e, "method name must be a Symbol but %"PRIsVALUE" is given",
rb_obj_class(argv[0]));
}
stack_check(ec);
if (last_call_status & MISSING_PRIVATE) {
format = rb_fstring_lit("private method `%s' called for %s%s%s");
}
else if (last_call_status & MISSING_PROTECTED) {
format = rb_fstring_lit("protected method `%s' called for %s%s%s");
}
else if (last_call_status & MISSING_VCALL) {
format = rb_fstring_lit("undefined local variable or method `%s' for %s%s%s");
exc = rb_eNameError;
}
else if (last_call_status & MISSING_SUPER) {
format = rb_fstring_lit("super: no superclass method `%s' for %s%s%s");
}
{
exc = rb_make_no_method_exception(exc, format, obj, argc, argv,
last_call_status & (MISSING_FCALL|MISSING_VCALL));
if (!(last_call_status & MISSING_MISSING)) {
rb_vm_pop_cfunc_frame();
}
rb_exc_raise(exc);
}
}
static void
vm_raise_method_missing(rb_execution_context_t *ec, int argc, const VALUE *argv,
VALUE obj, int call_status)
{
vm_passed_block_handler_set(ec, VM_BLOCK_HANDLER_NONE);
raise_method_missing(ec, argc, argv, obj, call_status | MISSING_MISSING);
}
static inline VALUE
method_missing(rb_execution_context_t *ec, VALUE obj, ID id, int argc, const VALUE *argv, enum method_missing_reason call_status, int kw_splat)
{
VALUE *nargv, result, work, klass;
VALUE block_handler = vm_passed_block_handler(ec);
const rb_callable_method_entry_t *me;
ec->method_missing_reason = call_status;
if (id == idMethodMissing) {
goto missing;
}
nargv = ALLOCV_N(VALUE, work, argc + 1);
nargv[0] = ID2SYM(id);
#ifdef __GLIBC__
if (!argv) {
static const VALUE buf = Qfalse;
VM_ASSERT(argc == 0);
argv = &buf;
}
#endif
MEMCPY(nargv + 1, argv, VALUE, argc);
++argc;
argv = nargv;
klass = CLASS_OF(obj);
if (!klass) goto missing;
me = rb_callable_method_entry(klass, idMethodMissing);
if (!me || METHOD_ENTRY_BASIC(me)) goto missing;
vm_passed_block_handler_set(ec, block_handler);
result = rb_vm_call_kw(ec, obj, idMethodMissing, argc, argv, me, kw_splat);
if (work) ALLOCV_END(work);
return result;
missing:
raise_method_missing(ec, argc, argv, obj, call_status | MISSING_MISSING);
UNREACHABLE_RETURN(Qundef);
}
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
#ifndef MJIT_HEADER
static inline VALUE
rb_funcallv_scope(VALUE recv, ID mid, int argc, const VALUE *argv, call_type scope)
{
rb_execution_context_t *ec = GET_EC();
const struct rb_callcache *cc = gccct_method_search(ec, recv, mid, argc);
VALUE self = ec->cfp->self;
if (LIKELY(cc) &&
LIKELY(rb_method_call_status(ec, vm_cc_cme(cc), scope, self) == MISSING_NONE)) {
// fastpath
return vm_call0_cc(ec, recv, mid, argc, argv, cc, false);
}
else {
return rb_call0(ec, recv, mid, argc, argv, scope, self);
}
}
#ifdef rb_funcallv
#undef rb_funcallv
#endif
VALUE
rb_funcallv(VALUE recv, ID mid, int argc, const VALUE *argv)
{
VM_ASSERT(ruby_thread_has_gvl_p());
return rb_funcallv_scope(recv, mid, argc, argv, CALL_FCALL);
}
VALUE
rb_funcallv_kw(VALUE recv, ID mid, int argc, const VALUE *argv, int kw_splat)
{
VM_ASSERT(ruby_thread_has_gvl_p());
return rb_call(recv, mid, argc, argv, kw_splat ? CALL_FCALL_KW : CALL_FCALL);
}
VALUE
rb_apply(VALUE recv, ID mid, VALUE args)
{
int argc;
VALUE *argv, ret;
argc = RARRAY_LENINT(args);
if (argc >= 0x100) {
args = rb_ary_subseq(args, 0, argc);
* include/ruby/ruby.h: constify RBasic::klass and add RBASIC_CLASS(obj) macro which returns a class of `obj'. This change is a part of RGENGC branch [ruby-trunk - Feature #8339]. * object.c: add new function rb_obj_reveal(). This function reveal interal (hidden) object by rb_obj_hide(). Note that do not change class before and after hiding. Only permitted example is: klass = RBASIC_CLASS(obj); rb_obj_hide(obj); .... rb_obj_reveal(obj, klass); TODO: API design. rb_obj_reveal() should be replaced with others. TODO: modify constified variables using cast may be harmful for compiler's analysis and optimizaton. Any idea to prohibt inserting RBasic::klass directly? If rename RBasic::klass and force to use RBASIC_CLASS(obj), then all codes such as `RBASIC(obj)->klass' will be compilation error. Is it acceptable? (We have similar experience at Ruby 1.9, for example "RARRAY(ary)->ptr" to "RARRAY_PTR(ary)". * internal.h: add some macros. * RBASIC_CLEAR_CLASS(obj) clear RBasic::klass to make it internal object. * RBASIC_SET_CLASS(obj, cls) set RBasic::klass. * RBASIC_SET_CLASS_RAW(obj, cls) same as RBASIC_SET_CLASS without write barrier (planned). * RCLASS_SET_SUPER(a, b) set super class of a. * array.c, class.c, compile.c, encoding.c, enum.c, error.c, eval.c, file.c, gc.c, hash.c, io.c, iseq.c, marshal.c, object.c, parse.y, proc.c, process.c, random.c, ruby.c, sprintf.c, string.c, thread.c, transcode.c, vm.c, vm_eval.c, win32/file.c: Use above macros and functions to access RBasic::klass. * ext/coverage/coverage.c, ext/readline/readline.c, ext/socket/ancdata.c, ext/socket/init.c, * ext/zlib/zlib.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@40691 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-05-13 14:49:11 +04:00
RBASIC_CLEAR_CLASS(args);
OBJ_FREEZE(args);
ret = rb_call(recv, mid, argc, RARRAY_CONST_PTR(args), CALL_FCALL);
RB_GC_GUARD(args);
return ret;
}
argv = ALLOCA_N(VALUE, argc);
MEMCPY(argv, RARRAY_CONST_PTR_TRANSIENT(args), VALUE, argc);
return rb_funcallv(recv, mid, argc, argv);
}
#ifdef rb_funcall
#undef rb_funcall
#endif
VALUE
rb_funcall(VALUE recv, ID mid, int n, ...)
{
VALUE *argv;
va_list ar;
if (n > 0) {
long i;
va_start(ar, n);
argv = ALLOCA_N(VALUE, n);
for (i = 0; i < n; i++) {
argv[i] = va_arg(ar, VALUE);
}
va_end(ar);
}
else {
argv = 0;
}
return rb_funcallv(recv, mid, n, argv);
}
/*!
* Calls a method only if it is the basic method of `ancestor`
* otherwise returns Qundef;
* \param recv receiver of the method
* \param mid an ID that represents the name of the method
* \param ancestor the Class that defined the basic method
* \param argc the number of arguments
* \param argv pointer to an array of method arguments
* \param kw_splat bool
*/
VALUE
rb_check_funcall_basic_kw(VALUE recv, ID mid, VALUE ancestor, int argc, const VALUE *argv, int kw_splat)
{
const rb_callable_method_entry_t *cme;
rb_execution_context_t *ec;
VALUE klass = CLASS_OF(recv);
if (!klass) return Qundef; /* hidden object */
cme = rb_callable_method_entry(klass, mid);
if (cme && METHOD_ENTRY_BASIC(cme) && RBASIC_CLASS(cme->defined_class) == ancestor) {
ec = GET_EC();
return rb_vm_call0(ec, recv, mid, argc, argv, cme, kw_splat);
}
return Qundef;
}
VALUE
rb_funcallv_public(VALUE recv, ID mid, int argc, const VALUE *argv)
{
return rb_funcallv_scope(recv, mid, argc, argv, CALL_PUBLIC);
}
VALUE
rb_funcallv_public_kw(VALUE recv, ID mid, int argc, const VALUE *argv, int kw_splat)
{
return rb_call(recv, mid, argc, argv, kw_splat ? CALL_PUBLIC_KW : CALL_PUBLIC);
}
VALUE
rb_funcall_passing_block(VALUE recv, ID mid, int argc, const VALUE *argv)
{
PASS_PASSED_BLOCK_HANDLER();
return rb_funcallv_public(recv, mid, argc, argv);
}
VALUE
rb_funcall_passing_block_kw(VALUE recv, ID mid, int argc, const VALUE *argv, int kw_splat)
{
PASS_PASSED_BLOCK_HANDLER();
return rb_call(recv, mid, argc, argv, kw_splat ? CALL_PUBLIC_KW : CALL_PUBLIC);
}
VALUE
rb_funcall_with_block(VALUE recv, ID mid, int argc, const VALUE *argv, VALUE passed_procval)
{
if (!NIL_P(passed_procval)) {
vm_passed_block_handler_set(GET_EC(), passed_procval);
}
return rb_funcallv_public(recv, mid, argc, argv);
}
VALUE
rb_funcall_with_block_kw(VALUE recv, ID mid, int argc, const VALUE *argv, VALUE passed_procval, int kw_splat)
{
if (!NIL_P(passed_procval)) {
vm_passed_block_handler_set(GET_EC(), passed_procval);
}
return rb_call(recv, mid, argc, argv, kw_splat ? CALL_PUBLIC_KW : CALL_PUBLIC);
}
static VALUE *
current_vm_stack_arg(const rb_execution_context_t *ec, const VALUE *argv)
{
rb_control_frame_t *prev_cfp = RUBY_VM_PREVIOUS_CONTROL_FRAME(ec->cfp);
if (RUBY_VM_CONTROL_FRAME_STACK_OVERFLOW_P(ec, prev_cfp)) return NULL;
if (prev_cfp->sp + 1 != argv) return NULL;
return prev_cfp->sp + 1;
}
static VALUE
send_internal(int argc, const VALUE *argv, VALUE recv, call_type scope)
{
ID id;
VALUE vid;
VALUE self;
VALUE ret, vargv = 0;
rb_execution_context_t *ec = GET_EC();
int public = scope == CALL_PUBLIC || scope == CALL_PUBLIC_KW;
if (public) {
self = Qundef;
}
else {
self = RUBY_VM_PREVIOUS_CONTROL_FRAME(ec->cfp)->self;
}
if (argc == 0) {
rb_raise(rb_eArgError, "no method name given");
}
vid = *argv;
id = rb_check_id(&vid);
if (!id) {
if (rb_method_basic_definition_p(CLASS_OF(recv), idMethodMissing)) {
VALUE exc = rb_make_no_method_exception(rb_eNoMethodError, 0,
recv, argc, argv,
!public);
rb_exc_raise(exc);
}
if (!SYMBOL_P(*argv)) {
VALUE *tmp_argv = current_vm_stack_arg(ec, argv);
vid = rb_str_intern(vid);
if (tmp_argv) {
tmp_argv[0] = vid;
}
else if (argc > 1) {
tmp_argv = ALLOCV_N(VALUE, vargv, argc);
tmp_argv[0] = vid;
MEMCPY(tmp_argv+1, argv+1, VALUE, argc-1);
argv = tmp_argv;
}
else {
argv = &vid;
}
}
id = idMethodMissing;
ec->method_missing_reason = MISSING_NOENTRY;
}
else {
argv++; argc--;
}
PASS_PASSED_BLOCK_HANDLER_EC(ec);
ret = rb_call0(ec, recv, id, argc, argv, scope, self);
ALLOCV_END(vargv);
return ret;
}
static VALUE
send_internal_kw(int argc, const VALUE *argv, VALUE recv, call_type scope)
{
if (rb_keyword_given_p()) {
switch (scope) {
case CALL_PUBLIC:
scope = CALL_PUBLIC_KW;
break;
case CALL_FCALL:
scope = CALL_FCALL_KW;
break;
default:
break;
}
}
return send_internal(argc, argv, recv, scope);
}
/*
* call-seq:
* foo.send(symbol [, args...]) -> obj
* foo.__send__(symbol [, args...]) -> obj
* foo.send(string [, args...]) -> obj
* foo.__send__(string [, args...]) -> obj
*
* Invokes the method identified by _symbol_, passing it any
2020-11-05 14:51:17 +03:00
* arguments specified.
* When the method is identified by a string, the string is converted
* to a symbol.
*
* BasicObject implements +__send__+, Kernel implements +send+.
2020-11-05 14:51:17 +03:00
* <code>__send__</code> is safer than +send+
* when _obj_ has the same method name like <code>Socket</code>.
* See also <code>public_send</code>.
*
* class Klass
* def hello(*args)
* "Hello " + args.join(' ')
* end
* end
* k = Klass.new
* k.send :hello, "gentle", "readers" #=> "Hello gentle readers"
*/
VALUE
rb_f_send(int argc, VALUE *argv, VALUE recv)
{
return send_internal_kw(argc, argv, recv, CALL_FCALL);
}
/*
* call-seq:
* obj.public_send(symbol [, args...]) -> obj
* obj.public_send(string [, args...]) -> obj
*
* Invokes the method identified by _symbol_, passing it any
* arguments specified. Unlike send, public_send calls public
* methods only.
* When the method is identified by a string, the string is converted
* to a symbol.
*
* 1.public_send(:puts, "hello") # causes NoMethodError
*/
static VALUE
rb_f_public_send(int argc, VALUE *argv, VALUE recv)
{
return send_internal_kw(argc, argv, recv, CALL_PUBLIC);
}
/* yield */
static inline VALUE
rb_yield_0_kw(int argc, const VALUE * argv, int kw_splat)
{
return vm_yield(GET_EC(), argc, argv, kw_splat);
}
static inline VALUE
rb_yield_0(int argc, const VALUE * argv)
{
return vm_yield(GET_EC(), argc, argv, RB_NO_KEYWORDS);
}
VALUE
rb_yield_1(VALUE val)
{
return rb_yield_0(1, &val);
}
VALUE
rb_yield(VALUE val)
{
2022-11-15 07:24:08 +03:00
if (UNDEF_P(val)) {
return rb_yield_0(0, NULL);
}
else {
return rb_yield_0(1, &val);
}
}
#undef rb_yield_values
VALUE
rb_yield_values(int n, ...)
{
if (n == 0) {
return rb_yield_0(0, 0);
}
else {
int i;
VALUE *argv;
va_list args;
argv = ALLOCA_N(VALUE, n);
va_start(args, n);
for (i=0; i<n; i++) {
argv[i] = va_arg(args, VALUE);
}
va_end(args);
return rb_yield_0(n, argv);
}
}
VALUE
rb_yield_values2(int argc, const VALUE *argv)
{
return rb_yield_0(argc, argv);
}
VALUE
rb_yield_values_kw(int argc, const VALUE *argv, int kw_splat)
{
return rb_yield_0_kw(argc, argv, kw_splat);
}
VALUE
rb_yield_splat(VALUE values)
{
VALUE tmp = rb_check_array_type(values);
VALUE v;
if (NIL_P(tmp)) {
rb_raise(rb_eArgError, "not an array");
}
v = rb_yield_0(RARRAY_LENINT(tmp), RARRAY_CONST_PTR(tmp));
RB_GC_GUARD(tmp);
return v;
}
VALUE
rb_yield_splat_kw(VALUE values, int kw_splat)
{
VALUE tmp = rb_check_array_type(values);
VALUE v;
if (NIL_P(tmp)) {
rb_raise(rb_eArgError, "not an array");
}
v = rb_yield_0_kw(RARRAY_LENINT(tmp), RARRAY_CONST_PTR(tmp), kw_splat);
RB_GC_GUARD(tmp);
return v;
}
VALUE
rb_yield_force_blockarg(VALUE values)
{
return vm_yield_force_blockarg(GET_EC(), values);
}
VALUE
rb_yield_block(RB_BLOCK_CALL_FUNC_ARGLIST(val, arg))
{
return vm_yield_with_block(GET_EC(), argc, argv,
NIL_P(blockarg) ? VM_BLOCK_HANDLER_NONE : blockarg,
rb_keyword_given_p());
}
static VALUE
loop_i(VALUE _)
{
for (;;) {
rb_yield_0(0, 0);
}
return Qnil;
}
static VALUE
loop_stop(VALUE dummy, VALUE exc)
{
return rb_attr_get(exc, id_result);
}
static VALUE
rb_f_loop_size(VALUE self, VALUE args, VALUE eobj)
{
return DBL2NUM(HUGE_VAL);
}
/*
* call-seq:
* loop { block }
* loop -> an_enumerator
*
* Repeatedly executes the block.
*
* If no block is given, an enumerator is returned instead.
*
* loop do
* print "Input: "
* line = gets
2022-12-09 00:14:08 +03:00
* break if !line or line =~ /^q/i
* # ...
* end
*
* StopIteration raised in the block breaks the loop. In this case,
* loop returns the "result" value stored in the exception.
*
* enum = Enumerator.new { |y|
* y << "one"
* y << "two"
* :ok
* }
*
* result = loop {
* puts enum.next
* } #=> :ok
*/
static VALUE
rb_f_loop(VALUE self)
{
RETURN_SIZED_ENUMERATOR(self, 0, 0, rb_f_loop_size);
return rb_rescue2(loop_i, (VALUE)0, loop_stop, (VALUE)0, rb_eStopIteration, (VALUE)0);
}
#if VMDEBUG
static const char *
vm_frametype_name(const rb_control_frame_t *cfp);
#endif
static VALUE
rb_iterate0(VALUE (* it_proc) (VALUE), VALUE data1,
const struct vm_ifunc *const ifunc,
rb_execution_context_t *ec)
{
enum ruby_tag_type state;
volatile VALUE retval = Qnil;
rb_control_frame_t *const cfp = ec->cfp;
EC_PUSH_TAG(ec);
state = EC_EXEC_TAG();
if (state == 0) {
iter_retry:
{
VALUE block_handler;
2022-07-21 19:23:58 +03:00
if (ifunc) {
struct rb_captured_block *captured = VM_CFP_TO_CAPTURED_BLOCK(cfp);
captured->code.ifunc = ifunc;
block_handler = VM_BH_FROM_IFUNC_BLOCK(captured);
}
else {
block_handler = VM_CF_BLOCK_HANDLER(cfp);
}
vm_passed_block_handler_set(ec, block_handler);
}
retval = (*it_proc) (data1);
}
else if (state == TAG_BREAK || state == TAG_RETRY) {
const struct vm_throw_data *const err = (struct vm_throw_data *)ec->errinfo;
const rb_control_frame_t *const escape_cfp = THROW_DATA_CATCH_FRAME(err);
if (cfp == escape_cfp) {
rb_vm_rewind_cfp(ec, cfp);
state = 0;
ec->tag->state = TAG_NONE;
ec->errinfo = Qnil;
if (state == TAG_RETRY) goto iter_retry;
retval = THROW_DATA_VAL(err);
}
else if (0) {
SDR(); fprintf(stderr, "%p, %p\n", (void *)cfp, (void *)escape_cfp);
}
}
EC_POP_TAG();
if (state) {
EC_JUMP_TAG(ec, state);
}
return retval;
}
static VALUE
rb_iterate_internal(VALUE (* it_proc)(VALUE), VALUE data1,
rb_block_call_func_t bl_proc, VALUE data2)
{
return rb_iterate0(it_proc, data1,
bl_proc ? rb_vm_ifunc_proc_new(bl_proc, (void *)data2) : 0,
GET_EC());
}
VALUE
rb_iterate(VALUE (* it_proc)(VALUE), VALUE data1,
rb_block_call_func_t bl_proc, VALUE data2)
{
return rb_iterate_internal(it_proc, data1, bl_proc, data2);
}
struct iter_method_arg {
VALUE obj;
ID mid;
int argc;
const VALUE *argv;
int kw_splat;
};
static VALUE
iterate_method(VALUE obj)
{
const struct iter_method_arg * arg =
(struct iter_method_arg *) obj;
return rb_call(arg->obj, arg->mid, arg->argc, arg->argv, arg->kw_splat ? CALL_FCALL_KW : CALL_FCALL);
}
VALUE rb_block_call_kw(VALUE obj, ID mid, int argc, const VALUE * argv, rb_block_call_func_t bl_proc, VALUE data2, int kw_splat);
VALUE
rb_block_call(VALUE obj, ID mid, int argc, const VALUE * argv,
rb_block_call_func_t bl_proc, VALUE data2)
{
return rb_block_call_kw(obj, mid, argc, argv, bl_proc, data2, RB_NO_KEYWORDS);
}
VALUE
rb_block_call_kw(VALUE obj, ID mid, int argc, const VALUE * argv,
rb_block_call_func_t bl_proc, VALUE data2, int kw_splat)
{
struct iter_method_arg arg;
arg.obj = obj;
arg.mid = mid;
arg.argc = argc;
arg.argv = argv;
arg.kw_splat = kw_splat;
return rb_iterate_internal(iterate_method, (VALUE)&arg, bl_proc, data2);
}
VALUE
rb_lambda_call(VALUE obj, ID mid, int argc, const VALUE *argv,
rb_block_call_func_t bl_proc, int min_argc, int max_argc,
VALUE data2)
{
struct iter_method_arg arg;
struct vm_ifunc *block;
if (!bl_proc) rb_raise(rb_eArgError, "NULL lambda function");
arg.obj = obj;
arg.mid = mid;
arg.argc = argc;
arg.argv = argv;
arg.kw_splat = 0;
block = rb_vm_ifunc_new(bl_proc, (void *)data2, min_argc, max_argc);
return rb_iterate0(iterate_method, (VALUE)&arg, block, GET_EC());
}
static VALUE
iterate_check_method(VALUE obj)
{
const struct iter_method_arg * arg =
(struct iter_method_arg *) obj;
return rb_check_funcall(arg->obj, arg->mid, arg->argc, arg->argv);
}
VALUE
rb_check_block_call(VALUE obj, ID mid, int argc, const VALUE *argv,
rb_block_call_func_t bl_proc, VALUE data2)
{
struct iter_method_arg arg;
arg.obj = obj;
arg.mid = mid;
arg.argc = argc;
arg.argv = argv;
arg.kw_splat = 0;
return rb_iterate_internal(iterate_check_method, (VALUE)&arg, bl_proc, data2);
}
VALUE
rb_each(VALUE obj)
{
return rb_call(obj, idEach, 0, 0, CALL_FCALL);
}
void rb_parser_warn_location(VALUE, int);
static VALUE eval_default_path;
static const rb_iseq_t *
eval_make_iseq(VALUE src, VALUE fname, int line, const rb_binding_t *bind,
const struct rb_block *base_block)
{
const VALUE parser = rb_parser_new();
const rb_iseq_t *const parent = vm_block_iseq(base_block);
rb_iseq_t *iseq = NULL;
rb_ast_t *ast;
int isolated_depth = 0;
// Conditionally enable coverage depending on the current mode:
VALUE coverage_enabled = RBOOL(rb_get_coverage_mode() & COVERAGE_TARGET_EVAL);
2022-09-17 11:19:57 +03:00
{
int depth = 1;
const VALUE *ep = vm_block_ep(base_block);
while (1) {
if (VM_ENV_FLAGS(ep, VM_ENV_FLAG_ISOLATED)) {
isolated_depth = depth;
break;
}
else if (VM_ENV_LOCAL_P(ep)) {
break;
}
ep = VM_ENV_PREV_EP(ep);
depth++;
}
}
if (!fname) {
fname = rb_source_location(&line);
}
2022-11-15 07:24:08 +03:00
if (!UNDEF_P(fname)) {
if (!NIL_P(fname)) fname = rb_fstring(fname);
}
else {
fname = rb_fstring_lit("(eval)");
if (!eval_default_path) {
eval_default_path = rb_fstring_lit("(eval)");
rb_gc_register_mark_object(eval_default_path);
}
fname = eval_default_path;
2022-09-17 11:19:57 +03:00
coverage_enabled = Qfalse;
}
rb_parser_set_context(parser, parent, FALSE);
ast = rb_parser_compile_string_path(parser, fname, src, line);
if (ast->body.root) {
2022-09-17 11:19:57 +03:00
if (ast->body.compile_option == Qnil) {
ast->body.compile_option = rb_obj_hide(rb_ident_hash_new());
}
rb_hash_aset(ast->body.compile_option, rb_sym_intern_ascii_cstr("coverage_enabled"), coverage_enabled);
iseq = rb_iseq_new_eval(&ast->body,
ISEQ_BODY(parent)->location.label,
2022-09-25 10:45:28 +03:00
fname, Qnil, line,
parent, isolated_depth);
}
rb_ast_dispose(ast);
if (iseq != NULL) {
if (0 && iseq) { /* for debug */
VALUE disasm = rb_iseq_disasm(iseq);
printf("%s\n", StringValuePtr(disasm));
}
rb_exec_event_hook_script_compiled(GET_EC(), iseq, src);
}
return iseq;
}
static VALUE
eval_string_with_cref(VALUE self, VALUE src, rb_cref_t *cref, VALUE file, int line)
{
rb_execution_context_t *ec = GET_EC();
struct rb_block block;
const rb_iseq_t *iseq;
rb_control_frame_t *cfp = rb_vm_get_ruby_level_next_cfp(ec, ec->cfp);
if (!cfp) {
rb_raise(rb_eRuntimeError, "Can't eval on top of Fiber or Thread");
}
block.as.captured = *VM_CFP_TO_CAPTURED_BLOCK(cfp);
block.as.captured.self = self;
block.as.captured.code.iseq = cfp->iseq;
block.type = block_type_iseq;
iseq = eval_make_iseq(src, file, line, NULL, &block);
if (!iseq) {
rb_exc_raise(ec->errinfo);
}
/* TODO: what the code checking? */
if (!cref && block.as.captured.code.val) {
rb_cref_t *orig_cref = vm_get_cref(vm_block_ep(&block));
cref = vm_cref_dup(orig_cref);
}
vm_set_eval_stack(ec, iseq, cref, &block);
/* kick */
return vm_exec(ec, true);
}
static VALUE
eval_string_with_scope(VALUE scope, VALUE src, VALUE file, int line)
{
rb_execution_context_t *ec = GET_EC();
rb_binding_t *bind = Check_TypedStruct(scope, &ruby_binding_data_type);
const rb_iseq_t *iseq = eval_make_iseq(src, file, line, bind, &bind->block);
if (!iseq) {
rb_exc_raise(ec->errinfo);
}
vm_set_eval_stack(ec, iseq, NULL, &bind->block);
/* save new env */
if (ISEQ_BODY(iseq)->local_table_size > 0) {
vm_bind_update_env(scope, bind, vm_make_env_object(ec, ec->cfp));
}
/* kick */
return vm_exec(ec, true);
}
/*
* call-seq:
* eval(string [, binding [, filename [,lineno]]]) -> obj
*
* Evaluates the Ruby expression(s) in <em>string</em>. If
* <em>binding</em> is given, which must be a Binding object, the
* evaluation is performed in its context. If the optional
* <em>filename</em> and <em>lineno</em> parameters are present, they
* will be used when reporting syntax errors.
*
* def get_binding(str)
* return binding
* end
* str = "hello"
* eval "str + ' Fred'" #=> "hello Fred"
* eval "str + ' Fred'", get_binding("bye") #=> "bye Fred"
*/
VALUE
rb_f_eval(int argc, const VALUE *argv, VALUE self)
{
VALUE src, scope, vfile, vline;
VALUE file = Qundef;
int line = 1;
rb_scan_args(argc, argv, "13", &src, &scope, &vfile, &vline);
SafeStringValue(src);
if (argc >= 3) {
StringValue(vfile);
}
if (argc >= 4) {
line = NUM2INT(vline);
}
if (!NIL_P(vfile))
file = vfile;
if (NIL_P(scope))
return eval_string_with_cref(self, src, NULL, file, line);
else
return eval_string_with_scope(scope, src, file, line);
}
/** @note This function name is not stable. */
VALUE
ruby_eval_string_from_file(const char *str, const char *filename)
{
VALUE file = filename ? rb_str_new_cstr(filename) : 0;
rb_execution_context_t *ec = GET_EC();
rb_control_frame_t *cfp = ec ? rb_vm_get_ruby_level_next_cfp(ec, ec->cfp) : NULL;
VALUE self = cfp ? cfp->self : rb_vm_top_self();
return eval_string_with_cref(self, rb_str_new2(str), NULL, file, 1);
}
VALUE
rb_eval_string(const char *str)
{
return ruby_eval_string_from_file(str, "eval");
}
static VALUE
eval_string_protect(VALUE str)
{
return rb_eval_string((char *)str);
}
VALUE
rb_eval_string_protect(const char *str, int *pstate)
{
return rb_protect(eval_string_protect, (VALUE)str, pstate);
}
struct eval_string_wrap_arg {
VALUE top_self;
VALUE klass;
const char *str;
};
static VALUE
eval_string_wrap_protect(VALUE data)
{
const struct eval_string_wrap_arg *const arg = (struct eval_string_wrap_arg*)data;
rb_cref_t *cref = rb_vm_cref_new_toplevel();
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
cref->klass_or_self = arg->klass;
return eval_string_with_cref(arg->top_self, rb_str_new_cstr(arg->str), cref, rb_str_new_cstr("eval"), 1);
}
VALUE
rb_eval_string_wrap(const char *str, int *pstate)
{
int state;
rb_thread_t *th = GET_THREAD();
VALUE self = th->top_self;
VALUE wrapper = th->top_wrapper;
VALUE val;
struct eval_string_wrap_arg data;
th->top_wrapper = rb_module_new();
th->top_self = rb_obj_clone(rb_vm_top_self());
rb_extend_object(th->top_self, th->top_wrapper);
data.top_self = th->top_self;
data.klass = th->top_wrapper;
data.str = str;
val = rb_protect(eval_string_wrap_protect, (VALUE)&data, &state);
th->top_self = self;
th->top_wrapper = wrapper;
if (pstate) {
*pstate = state;
}
else if (state != TAG_NONE) {
EC_JUMP_TAG(th->ec, state);
}
return val;
}
VALUE
rb_eval_cmd_kw(VALUE cmd, VALUE arg, int kw_splat)
{
enum ruby_tag_type state;
volatile VALUE val = Qnil; /* OK */
rb_execution_context_t * volatile ec = GET_EC();
EC_PUSH_TAG(ec);
if ((state = EC_EXEC_TAG()) == TAG_NONE) {
if (!RB_TYPE_P(cmd, T_STRING)) {
val = rb_funcallv_kw(cmd, idCall, RARRAY_LENINT(arg),
RARRAY_CONST_PTR(arg), kw_splat);
}
else {
val = eval_string_with_cref(rb_vm_top_self(), cmd, NULL, 0, 0);
}
}
EC_POP_TAG();
if (state) EC_JUMP_TAG(ec, state);
return val;
}
/* block eval under the class/module context */
static VALUE
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
yield_under(VALUE self, int singleton, int argc, const VALUE *argv, int kw_splat)
{
rb_execution_context_t *ec = GET_EC();
rb_control_frame_t *cfp = ec->cfp;
VALUE block_handler = VM_CF_BLOCK_HANDLER(cfp);
VALUE new_block_handler = 0;
const struct rb_captured_block *captured = NULL;
struct rb_captured_block new_captured;
const VALUE *ep = NULL;
rb_cref_t *cref;
int is_lambda = FALSE;
if (block_handler != VM_BLOCK_HANDLER_NONE) {
again:
switch (vm_block_handler_type(block_handler)) {
case block_handler_type_iseq:
captured = VM_BH_TO_CAPT_BLOCK(block_handler);
new_captured = *captured;
new_block_handler = VM_BH_FROM_ISEQ_BLOCK(&new_captured);
break;
case block_handler_type_ifunc:
captured = VM_BH_TO_CAPT_BLOCK(block_handler);
new_captured = *captured;
new_block_handler = VM_BH_FROM_IFUNC_BLOCK(&new_captured);
break;
case block_handler_type_proc:
is_lambda = rb_proc_lambda_p(block_handler) != Qfalse;
block_handler = vm_proc_to_block_handler(VM_BH_TO_PROC(block_handler));
goto again;
case block_handler_type_symbol:
return rb_sym_proc_call(SYM2ID(VM_BH_TO_SYMBOL(block_handler)),
argc, argv, kw_splat,
VM_BLOCK_HANDLER_NONE);
}
new_captured.self = self;
ep = captured->ep;
VM_FORCE_WRITE_SPECIAL_CONST(&VM_CF_LEP(ec->cfp)[VM_ENV_DATA_INDEX_SPECVAL], new_block_handler);
}
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
VM_ASSERT(singleton || RB_TYPE_P(self, T_MODULE) || RB_TYPE_P(self, T_CLASS));
cref = vm_cref_push(ec, self, ep, TRUE, singleton);
return vm_yield_with_cref(ec, argc, argv, kw_splat, cref, is_lambda);
}
VALUE
rb_yield_refine_block(VALUE refinement, VALUE refinements)
{
rb_execution_context_t *ec = GET_EC();
VALUE block_handler = VM_CF_BLOCK_HANDLER(ec->cfp);
if (vm_block_handler_type(block_handler) != block_handler_type_iseq) {
rb_bug("rb_yield_refine_block: an iseq block is required");
}
else {
const struct rb_captured_block *captured = VM_BH_TO_ISEQ_BLOCK(block_handler);
struct rb_captured_block new_captured = *captured;
VALUE new_block_handler = VM_BH_FROM_ISEQ_BLOCK(&new_captured);
const VALUE *ep = captured->ep;
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
rb_cref_t *cref = vm_cref_push(ec, refinement, ep, TRUE, FALSE);
CREF_REFINEMENTS_SET(cref, refinements);
VM_FORCE_WRITE_SPECIAL_CONST(&VM_CF_LEP(ec->cfp)[VM_ENV_DATA_INDEX_SPECVAL], new_block_handler);
new_captured.self = refinement;
return vm_yield_with_cref(ec, 0, NULL, RB_NO_KEYWORDS, cref, FALSE);
}
}
/* string eval under the class/module context */
static VALUE
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
eval_under(VALUE self, int singleton, VALUE src, VALUE file, int line)
{
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
rb_cref_t *cref = vm_cref_push(GET_EC(), self, NULL, FALSE, singleton);
SafeStringValue(src);
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return eval_string_with_cref(self, src, cref, file, line);
}
static VALUE
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
specific_eval(int argc, const VALUE *argv, VALUE self, int singleton, int kw_splat)
{
if (rb_block_given_p()) {
rb_check_arity(argc, 0, 0);
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return yield_under(self, singleton, 1, &self, kw_splat);
}
else {
VALUE file = Qundef;
int line = 1;
VALUE code;
2022-07-21 19:23:58 +03:00
rb_check_arity(argc, 1, 3);
code = argv[0];
SafeStringValue(code);
if (argc > 2)
line = NUM2INT(argv[2]);
if (argc > 1) {
file = argv[1];
if (!NIL_P(file)) StringValue(file);
}
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return eval_under(self, singleton, code, file, line);
}
}
/*
* call-seq:
* obj.instance_eval(string [, filename [, lineno]] ) -> obj
* obj.instance_eval {|obj| block } -> obj
*
* Evaluates a string containing Ruby source code, or the given block,
* within the context of the receiver (_obj_). In order to set the
* context, the variable +self+ is set to _obj_ while
* the code is executing, giving the code access to _obj_'s
* instance variables and private methods.
*
* When <code>instance_eval</code> is given a block, _obj_ is also
* passed in as the block's only argument.
*
* When <code>instance_eval</code> is given a +String+, the optional
* second and third parameters supply a filename and starting line number
* that are used when reporting compilation errors.
*
* class KlassWithSecret
* def initialize
* @secret = 99
* end
* private
* def the_secret
* "Ssssh! The secret is #{@secret}."
* end
* end
* k = KlassWithSecret.new
* k.instance_eval { @secret } #=> 99
* k.instance_eval { the_secret } #=> "Ssssh! The secret is 99."
* k.instance_eval {|obj| obj == self } #=> true
*/
static VALUE
rb_obj_instance_eval_internal(int argc, const VALUE *argv, VALUE self)
{
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return specific_eval(argc, argv, self, TRUE, RB_PASS_CALLED_KEYWORDS);
}
VALUE
rb_obj_instance_eval(int argc, const VALUE *argv, VALUE self)
{
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return specific_eval(argc, argv, self, TRUE, RB_NO_KEYWORDS);
}
/*
* call-seq:
* obj.instance_exec(arg...) {|var...| block } -> obj
*
* Executes the given block within the context of the receiver
* (_obj_). In order to set the context, the variable +self+ is set
* to _obj_ while the code is executing, giving the code access to
* _obj_'s instance variables. Arguments are passed as block parameters.
*
* class KlassWithSecret
* def initialize
* @secret = 99
* end
* end
* k = KlassWithSecret.new
* k.instance_exec(5) {|x| @secret+x } #=> 104
*/
static VALUE
rb_obj_instance_exec_internal(int argc, const VALUE *argv, VALUE self)
{
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return yield_under(self, TRUE, argc, argv, RB_PASS_CALLED_KEYWORDS);
}
VALUE
rb_obj_instance_exec(int argc, const VALUE *argv, VALUE self)
{
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return yield_under(self, TRUE, argc, argv, RB_NO_KEYWORDS);
}
/*
* call-seq:
* mod.class_eval(string [, filename [, lineno]]) -> obj
* mod.class_eval {|mod| block } -> obj
* mod.module_eval(string [, filename [, lineno]]) -> obj
* mod.module_eval {|mod| block } -> obj
*
* Evaluates the string or block in the context of _mod_, except that when
* a block is given, constant/class variable lookup is not affected. This
* can be used to add methods to a class. <code>module_eval</code> returns
* the result of evaluating its argument. The optional _filename_ and
* _lineno_ parameters set the text for error messages.
*
* class Thing
* end
* a = %q{def hello() "Hello there!" end}
* Thing.module_eval(a)
* puts Thing.new.hello()
* Thing.module_eval("invalid code", "dummy", 123)
*
* <em>produces:</em>
*
* Hello there!
* dummy:123:in `module_eval': undefined local variable
* or method `code' for Thing:Class
*/
static VALUE
rb_mod_module_eval_internal(int argc, const VALUE *argv, VALUE mod)
{
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return specific_eval(argc, argv, mod, FALSE, RB_PASS_CALLED_KEYWORDS);
}
VALUE
rb_mod_module_eval(int argc, const VALUE *argv, VALUE mod)
{
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return specific_eval(argc, argv, mod, FALSE, RB_NO_KEYWORDS);
}
/*
* call-seq:
* mod.module_exec(arg...) {|var...| block } -> obj
* mod.class_exec(arg...) {|var...| block } -> obj
*
* Evaluates the given block in the context of the class/module.
* The method defined in the block will belong to the receiver.
* Any arguments passed to the method will be passed to the block.
* This can be used if the block needs to access instance variables.
*
* class Thing
* end
* Thing.class_exec{
* def hello() "Hello there!" end
* }
* puts Thing.new.hello()
*
* <em>produces:</em>
*
* Hello there!
*/
static VALUE
rb_mod_module_exec_internal(int argc, const VALUE *argv, VALUE mod)
{
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return yield_under(mod, FALSE, argc, argv, RB_PASS_CALLED_KEYWORDS);
}
VALUE
rb_mod_module_exec(int argc, const VALUE *argv, VALUE mod)
{
Lazily create singletons on instance_{exec,eval} (#5146) * Lazily create singletons on instance_{exec,eval} Previously when instance_exec or instance_eval was called on an object, that object would be given a singleton class so that method definitions inside the block would be added to the object rather than its class. This commit aims to improve performance by delaying the creation of the singleton class unless/until one is needed for method definition. Most of the time instance_eval is used without any method definition. This was implemented by adding a flag to the cref indicating that it represents a singleton of the object rather than a class itself. In this case CREF_CLASS returns the object's existing class, but in cases that we are defining a method (either via definemethod or VM_SPECIAL_OBJECT_CBASE which is used for undef and alias). This also happens to fix what I believe is a bug. Previously instance_eval behaved differently with regards to constant access for true/false/nil than for all other objects. I don't think this was intentional. String::Foo = "foo" "".instance_eval("Foo") # => "foo" Integer::Foo = "foo" 123.instance_eval("Foo") # => "foo" TrueClass::Foo = "foo" true.instance_eval("Foo") # NameError: uninitialized constant Foo This also slightly changes the error message when trying to define a method through instance_eval on an object which can't have a singleton class. Before: $ ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': no class/module to add method (TypeError) After: $ ./ruby -e '123.instance_eval { def foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) IMO this error is a small improvement on the original and better matches the (both old and new) message when definging a method using `def self.` $ ruby -e '123.instance_eval{ def self.foo; end }' -e:1:in `block in <main>': can't define singleton (TypeError) Co-authored-by: Matthew Draper <matthew@trebex.net> * Remove "under" argument from yield_under * Move CREF_SINGLETON_SET into vm_cref_new * Simplify vm_get_const_base * Fix leaf VM_SPECIAL_OBJECT_CONST_BASE Co-authored-by: Matthew Draper <matthew@trebex.net>
2021-12-03 02:53:39 +03:00
return yield_under(mod, FALSE, argc, argv, RB_NO_KEYWORDS);
}
/*
* Document-class: UncaughtThrowError
*
* Raised when +throw+ is called with a _tag_ which does not have
* corresponding +catch+ block.
*
* throw "foo", "bar"
*
* <em>raises the exception:</em>
*
* UncaughtThrowError: uncaught throw "foo"
*/
static VALUE
uncaught_throw_init(int argc, const VALUE *argv, VALUE exc)
{
rb_check_arity(argc, 2, UNLIMITED_ARGUMENTS);
rb_call_super(argc - 2, argv + 2);
rb_ivar_set(exc, id_tag, argv[0]);
rb_ivar_set(exc, id_value, argv[1]);
return exc;
}
/*
* call-seq:
* uncaught_throw.tag -> obj
*
* Return the tag object which was called for.
*/
static VALUE
uncaught_throw_tag(VALUE exc)
{
return rb_ivar_get(exc, id_tag);
}
/*
* call-seq:
* uncaught_throw.value -> obj
*
* Return the return value which was called for.
*/
static VALUE
uncaught_throw_value(VALUE exc)
{
return rb_ivar_get(exc, id_value);
}
/*
* call-seq:
* uncaught_throw.to_s -> string
*
* Returns formatted message with the inspected tag.
*/
static VALUE
uncaught_throw_to_s(VALUE exc)
{
VALUE mesg = rb_attr_get(exc, id_mesg);
VALUE tag = uncaught_throw_tag(exc);
return rb_str_format(1, &tag, mesg);
}
/*
* call-seq:
* throw(tag [, obj])
*
* Transfers control to the end of the active +catch+ block
* waiting for _tag_. Raises +UncaughtThrowError+ if there
* is no +catch+ block for the _tag_. The optional second
* parameter supplies a return value for the +catch+ block,
* which otherwise defaults to +nil+. For examples, see
* Kernel::catch.
*/
static VALUE
rb_f_throw(int argc, VALUE *argv, VALUE _)
{
VALUE tag, value;
rb_scan_args(argc, argv, "11", &tag, &value);
rb_throw_obj(tag, value);
UNREACHABLE_RETURN(Qnil);
}
void
rb_throw_obj(VALUE tag, VALUE value)
{
rb_execution_context_t *ec = GET_EC();
struct rb_vm_tag *tt = ec->tag;
while (tt) {
if (tt->tag == tag) {
tt->retval = value;
break;
}
tt = tt->prev;
}
if (!tt) {
VALUE desc[3];
desc[0] = tag;
desc[1] = value;
desc[2] = rb_str_new_cstr("uncaught throw %p");
rb_exc_raise(rb_class_new_instance(numberof(desc), desc, rb_eUncaughtThrow));
}
ec->errinfo = (VALUE)THROW_DATA_NEW(tag, NULL, TAG_THROW);
EC_JUMP_TAG(ec, TAG_THROW);
}
void
rb_throw(const char *tag, VALUE val)
{
rb_throw_obj(rb_sym_intern_ascii_cstr(tag), val);
}
static VALUE
catch_i(RB_BLOCK_CALL_FUNC_ARGLIST(tag, _))
{
return rb_yield_0(1, &tag);
}
/*
* call-seq:
* catch([tag]) {|tag| block } -> obj
*
* +catch+ executes its block. If +throw+ is not called, the block executes
* normally, and +catch+ returns the value of the last expression evaluated.
*
* catch(1) { 123 } # => 123
*
* If <code>throw(tag2, val)</code> is called, Ruby searches up its stack for
* a +catch+ block whose +tag+ has the same +object_id+ as _tag2_. When found,
* the block stops executing and returns _val_ (or +nil+ if no second argument
* was given to +throw+).
*
* catch(1) { throw(1, 456) } # => 456
* catch(1) { throw(1) } # => nil
*
* When +tag+ is passed as the first argument, +catch+ yields it as the
* parameter of the block.
*
* catch(1) {|x| x + 2 } # => 3
*
* When no +tag+ is given, +catch+ yields a new unique object (as from
* +Object.new+) as the block parameter. This object can then be used as the
* argument to +throw+, and will match the correct +catch+ block.
*
* catch do |obj_A|
* catch do |obj_B|
* throw(obj_B, 123)
* puts "This puts is not reached"
* end
*
* puts "This puts is displayed"
* 456
* end
*
* # => 456
*
* catch do |obj_A|
* catch do |obj_B|
* throw(obj_A, 123)
* puts "This puts is still not reached"
* end
*
* puts "Now this puts is also not reached"
* 456
* end
*
* # => 123
*/
static VALUE
rb_f_catch(int argc, VALUE *argv, VALUE self)
{
VALUE tag = rb_check_arity(argc, 0, 1) ? argv[0] : rb_obj_alloc(rb_cObject);
return rb_catch_obj(tag, catch_i, 0);
}
VALUE
rb_catch(const char *tag, rb_block_call_func_t func, VALUE data)
{
VALUE vtag = tag ? rb_sym_intern_ascii_cstr(tag) : rb_obj_alloc(rb_cObject);
return rb_catch_obj(vtag, func, data);
}
static VALUE
vm_catch_protect(VALUE tag, rb_block_call_func *func, VALUE data,
enum ruby_tag_type *stateptr, rb_execution_context_t *volatile ec)
{
enum ruby_tag_type state;
VALUE val = Qnil; /* OK */
rb_control_frame_t *volatile saved_cfp = ec->cfp;
EC_PUSH_TAG(ec);
_tag.tag = tag;
if ((state = EC_EXEC_TAG()) == TAG_NONE) {
/* call with argc=1, argv = [tag], block = Qnil to insure compatibility */
val = (*func)(tag, data, 1, (const VALUE *)&tag, Qnil);
}
else if (state == TAG_THROW && THROW_DATA_VAL((struct vm_throw_data *)ec->errinfo) == tag) {
rb_vm_rewind_cfp(ec, saved_cfp);
val = ec->tag->retval;
ec->errinfo = Qnil;
state = 0;
}
EC_POP_TAG();
if (stateptr)
*stateptr = state;
return val;
}
VALUE
rb_catch_protect(VALUE t, rb_block_call_func *func, VALUE data, enum ruby_tag_type *stateptr)
{
return vm_catch_protect(t, func, data, stateptr, GET_EC());
}
VALUE
rb_catch_obj(VALUE t, rb_block_call_func_t func, VALUE data)
{
enum ruby_tag_type state;
rb_execution_context_t *ec = GET_EC();
VALUE val = vm_catch_protect(t, (rb_block_call_func *)func, data, &state, ec);
if (state) EC_JUMP_TAG(ec, state);
return val;
}
static void
local_var_list_init(struct local_var_list *vars)
{
vars->tbl = rb_ident_hash_new();
RBASIC_CLEAR_CLASS(vars->tbl);
}
static VALUE
local_var_list_finish(struct local_var_list *vars)
{
/* TODO: not to depend on the order of st_table */
VALUE ary = rb_hash_keys(vars->tbl);
rb_hash_clear(vars->tbl);
vars->tbl = 0;
return ary;
}
static int
local_var_list_update(st_data_t *key, st_data_t *value, st_data_t arg, int existing)
{
if (existing) return ST_STOP;
*value = (st_data_t)Qtrue; /* INT2FIX(arg) */
return ST_CONTINUE;
}
static void
local_var_list_add(const struct local_var_list *vars, ID lid)
{
if (lid && rb_is_local_id(lid)) {
/* should skip temporary variable */
st_data_t idx = 0; /* tbl->num_entries */
rb_hash_stlike_update(vars->tbl, ID2SYM(lid), local_var_list_update, idx);
}
}
/*
* call-seq:
* local_variables -> array
*
* Returns the names of the current local variables.
*
* fred = 1
* for i in 1..10
* # ...
* end
* local_variables #=> [:fred, :i]
*/
static VALUE
rb_f_local_variables(VALUE _)
{
struct local_var_list vars;
rb_execution_context_t *ec = GET_EC();
rb_control_frame_t *cfp = vm_get_ruby_level_caller_cfp(ec, RUBY_VM_PREVIOUS_CONTROL_FRAME(ec->cfp));
unsigned int i;
local_var_list_init(&vars);
while (cfp) {
if (cfp->iseq) {
for (i = 0; i < ISEQ_BODY(cfp->iseq)->local_table_size; i++) {
local_var_list_add(&vars, ISEQ_BODY(cfp->iseq)->local_table[i]);
}
}
if (!VM_ENV_LOCAL_P(cfp->ep)) {
/* block */
const VALUE *ep = VM_CF_PREV_EP(cfp);
2022-07-21 19:23:58 +03:00
if (vm_collect_local_variables_in_heap(ep, &vars)) {
break;
}
else {
* vm_core.h: remove lfp (local frame pointer) and rename dfp (dynamic frame pointer) to ep (environment pointer). This change make VM `normal' (similar to other interpreters). Before this commit: Each frame has two env pointers lfp and dfp. lfp points local environment which is method/class/toplevel frame. lfp[0] is block pointer. dfp is block local frame. dfp[0] points previous (parent) environment pointer. lfp == dfp when frame is method/class/toplevel. You can get lfp from dfp by traversing previous environment pointers. After this commit: Each frame has only `ep' to point respective enviornoment. If there is parent environment, then ep[0] points parent envioenment (as dfp). If there are no more environment, then ep[0] points block pointer (as lfp). We call such ep as `LEP' (local EP). We add some macros to get LEP and to detect LEP or not. In short, we replace dfp and lfp with ep and LEP. rb_block_t and rb_binding_t member `lfp' and `dfp' are removed and member `ep' is added. rename rb_thread_t's member `local_lfp' and `local_svar' to `root_lep' and `root_svar'. (VM_EP_PREV_EP(ep)): get previous environment pointer. This macro assume that ep is not LEP. (VM_EP_BLOCK_PTR(ep)): get block pointer. This macro assume that ep is LEP. (VM_EP_LEP_P(ep)): detect ep is LEP or not. (VM_ENVVAL_BLOCK_PTR(ptr)): make block pointer. (VM_ENVVAL_BLOCK_PTR_P(v)): detect v is block pointer. (VM_ENVVAL_PREV_EP_PTR(ptr)): make prev environment pointer. (VM_ENVVAL_PREV_EP_PTR_P(v)): detect v is prev env pointer. * vm.c: apply above changes. (VM_EP_LEP(ep)): get LEP. (VM_CF_LEP(cfp)): get LEP of cfp->ep. (VM_CF_PREV_EP(cfp)): utility function VM_EP_PREV_EP(cfp->ep). (VM_CF_BLOCK_PTR(cfp)): utility function VM_EP_BLOCK_PTR(cfp->ep). * vm.c, vm_eval.c, vm_insnhelper.c, vm_insnhelper.h, insns.def: apply above changes. * cont.c: ditto. * eval.c, eval_intern.h: ditto. * proc.c: ditto. * thread.c: ditto. * vm_dump.c: ditto. * vm_exec.h: fix function name (on vm debug mode). git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@36030 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-06-11 07:14:59 +04:00
while (cfp->ep != ep) {
cfp = RUBY_VM_PREVIOUS_CONTROL_FRAME(cfp);
}
}
}
else {
break;
}
}
return local_var_list_finish(&vars);
}
/*
* call-seq:
* block_given? -> true or false
*
* Returns <code>true</code> if <code>yield</code> would execute a
* block in the current context. The <code>iterator?</code> form
* is mildly deprecated.
*
* def try
* if block_given?
* yield
* else
* "no block"
* end
* end
* try #=> "no block"
* try { "hello" } #=> "hello"
* try do "hello" end #=> "hello"
*/
static VALUE
rb_f_block_given_p(VALUE _)
{
rb_execution_context_t *ec = GET_EC();
rb_control_frame_t *cfp = ec->cfp;
cfp = vm_get_ruby_level_caller_cfp(ec, RUBY_VM_PREVIOUS_CONTROL_FRAME(cfp));
2021-08-02 06:06:44 +03:00
return RBOOL(cfp != NULL && VM_CF_BLOCK_HANDLER(cfp) != VM_BLOCK_HANDLER_NONE);
}
/*
* call-seq:
* iterator? -> true or false
*
* Deprecated. Use block_given? instead.
*/
static VALUE
rb_f_iterator_p(VALUE self)
{
rb_warn_deprecated("iterator?", "block_given?");
return rb_f_block_given_p(self);
}
VALUE
rb_current_realfilepath(void)
{
const rb_execution_context_t *ec = GET_EC();
rb_control_frame_t *cfp = ec->cfp;
cfp = vm_get_ruby_level_caller_cfp(ec, RUBY_VM_PREVIOUS_CONTROL_FRAME(cfp));
if (cfp != NULL) {
VALUE path = rb_iseq_realpath(cfp->iseq);
if (RTEST(path)) return path;
// eval context
path = rb_iseq_path(cfp->iseq);
if (path == eval_default_path) {
return Qnil;
}
else {
return path;
}
}
return Qnil;
}
void
Init_vm_eval(void)
{
rb_define_global_function("eval", rb_f_eval, -1);
rb_define_global_function("local_variables", rb_f_local_variables, 0);
rb_define_global_function("iterator?", rb_f_iterator_p, 0);
rb_define_global_function("block_given?", rb_f_block_given_p, 0);
rb_define_global_function("catch", rb_f_catch, -1);
rb_define_global_function("throw", rb_f_throw, -1);
rb_define_global_function("loop", rb_f_loop, 0);
rb_define_method(rb_cBasicObject, "instance_eval", rb_obj_instance_eval_internal, -1);
rb_define_method(rb_cBasicObject, "instance_exec", rb_obj_instance_exec_internal, -1);
rb_define_private_method(rb_cBasicObject, "method_missing", rb_method_missing, -1);
#if 1
rb_add_method(rb_cBasicObject, id__send__,
VM_METHOD_TYPE_OPTIMIZED, (void *)OPTIMIZED_METHOD_TYPE_SEND, METHOD_VISI_PUBLIC);
rb_add_method(rb_mKernel, idSend,
VM_METHOD_TYPE_OPTIMIZED, (void *)OPTIMIZED_METHOD_TYPE_SEND, METHOD_VISI_PUBLIC);
#else
rb_define_method(rb_cBasicObject, "__send__", rb_f_send, -1);
rb_define_method(rb_mKernel, "send", rb_f_send, -1);
#endif
rb_define_method(rb_mKernel, "public_send", rb_f_public_send, -1);
rb_define_method(rb_cModule, "module_exec", rb_mod_module_exec_internal, -1);
rb_define_method(rb_cModule, "class_exec", rb_mod_module_exec_internal, -1);
rb_define_method(rb_cModule, "module_eval", rb_mod_module_eval_internal, -1);
rb_define_method(rb_cModule, "class_eval", rb_mod_module_eval_internal, -1);
rb_eUncaughtThrow = rb_define_class("UncaughtThrowError", rb_eArgError);
rb_define_method(rb_eUncaughtThrow, "initialize", uncaught_throw_init, -1);
rb_define_method(rb_eUncaughtThrow, "tag", uncaught_throw_tag, 0);
rb_define_method(rb_eUncaughtThrow, "value", uncaught_throw_value, 0);
rb_define_method(rb_eUncaughtThrow, "to_s", uncaught_throw_to_s, 0);
id_result = rb_intern_const("result");
id_tag = rb_intern_const("tag");
id_value = rb_intern_const("value");
}
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
#endif /* #ifndef MJIT_HEADER */